Vol. 68 Èasopis Saveza hemijskih inženjera · 2017. 2. 3. · 5 UDK 66:54(05) CODEN HMIDA 8 ISSN 0367–598 X Vol. 68 Èasopis Saveza hemijskih inženjera BEOGRAD, SEPTEMBAR-OKTOBAR

5UDK 66:54(05) CODEN HMIDA 8 ISSN 0367–598 X

Vol. 68 Èasopis Saveza hemijskih inženjera

BEOGRAD, SEPTEMBAR-OKTOBAR 2014. HMIDA 8, 68 (5) 519-651 (2014)

Chemical Industry Химическая промышленность

Časopis Saveza hemijskih inženjera SrbijeJournal of the Association of Chemical Engineers of Serbia

Журнал Союза химических инженеров Сербии

VOL. 68 Beograd, septembar−oktobar 2014 Broj 5 Izdavač Savez hemijskih inženjera Srbije Beograd, Kneza Miloša 9/I Glavni urednik Branko Bugarski Zamenica glavnog i odgovornog urednika Nevenka Bošković-Vragolović Urednici Katarina Jeremić, Ivana Banković-Ilić, Maja Obradović, Dušan Mijin Članovi uredništva Milorad Cakić, Željko Čupić, Željko Grbavčić, Katarina Jeremić, Miodrag Lazić, Slobodan Petrović, Milovan Purenović, Aleksandar Spasić, Dragoslav Stoiljković, Radmila Šećerov-Sokolović, Slobodan Šerbanović, Nikola Nikačević, Svetomir Milojević Članovi uredništva iz inostranstva Dragomir Bukur (SAD), Jiri Hanika (Češka Republika), Valerij Meshalkin (Rusija), Ljubiša Radović (SAD), Constantinos Vayenas (Grčka) Likovno-grafičko rešenje naslovne strane Milan Jovanović Redakcija 11000 Beograd, Kneza Miloša 9/I Tel/fax: 011/3240-018 E-pošta: [email protected] www.ache.org.rs Izlazi dvomesečno, rukopisi se ne vraćaju Za izdavača Tatijana Duduković Sekretar redakcije Slavica Desnica Izdavanje časopisa pomaže Republika Srbija, Ministarstvo prosvete, nauke i tehnološkog razvoja Uplata pretplate i oglasnog prostora vrši se na tekući račun Saveza hemijskih inženjera Srbije, Beograd, broj 205-2172-71, Komercijalna banka a.d., Beograd Kompjuterska priprema Vladimir Panić Štampa Razvojno-istraživački centar grafičkog inženjerstva, Tehnološko-metalurški fakultet, Univerzitet u Beogradu, Karnegijeva 4, 11000 Beograd Indeksiranje Radovi koji se publikuju u časopisu Hemijska Industrija ideksiraju se preko Thompson Reuters Scietific® servisa Science Citation Index - ExpandedTM i Journal Citation Report (JCR), kao i domaćeg SCIndeks servisa Centra za evaluaciju u obrazovanju i nauci

SADRŽAJ Ljubiša J. Bučanović, Mihailo P. Lazarević, Srećko N.Batalov, The

fractional PID controllers tuned by genetic algorithms for expansion turbine in the cryogenic air separation process ...... 519

Zuozhu Wu, Xinqi Qiao, Zhen Huang, Development and validation of a reduced mechanism for methane using a new integral algorithm in a premixed flame ................................................... 529

Xue-Fei Zhou, Oxidation of lignin–carbohydrate complex from bamboo with hydrogen peroxide catalyzed by Co(salen) ........... 541

Dimitrije Ž. Stevanović, Mića B. Jovanović, Marina A. Mihajlović, Jovan M. Jovanović, Željko B. Grbavčić, Analiza simulatora tehnoloških procesa u funkciji projektovanja: Studija slučaja separacije prirodnog gasa .............................................. 547

Sonja M. Vidojković, Antonije E. Onjia, Aleksandar B. Devečerski, Nebojsa N. Grahovac, Aleksandra B. Nastasović Economizer water-wall damages initiated by feedwater impurities ............ 559

Ljiljana M. Djekić, Marija M. Primorac, Formulacija i karakteriza-cija samo-mikroemulgujućih nosača lekovitih supstanci na bazi biokompatibilnih nejonskih surfaktanata .......................... 565

Milovan Vuković, Nada Štrbac, Miroslav Sokić, Vesna Grekulović, Vladimir Cvetkovski, Bioleaching of pollymetallic sulphide concentrate using thermophilic bacteria ................................... 575

Milica M. Petrović, Jelena Z. Mitrović, Miljana D. Radović, Danijela V. Bojić, Miloš M. Kostić, Radomir B. Ljupković, Aleksandar Lj. Bojić, Synthesis of bismuth(III) oxide films based anodes for electrochemical degradation of Reactive blue 19 and Crystal violet ................................................................... 585

Александра Т. Ивановић, Бисерка Т. Трумић, Светлана Љ. Ива-нов, Саша Р. Марјановић, Моделовање утицаја темпе-ратуре и времена хомогенизационог жарења на твр-доћу PdNi5 легуре ..................................................................... 597

Tatjana Nikolin, Mirjana Sevaljević, The examination of the seasonal influence on the efficiency in oil and fats removal through primary treatment from the wastewater of edible oil industry .................................................................................. 605

Branko B. Pejović, Vladan M. Mićić, Mitar D. Perušić, Goran S. Tadić, Ljubica C. Vasiljević, Slavko N. Smiljanić, Predlog za određivanje promene entropije poluidealnog gasa prime-nom srednjih vrednosti temperaturnih funkcija ......................... 615

Tatjana A. Djakov, Ivanka G. Popović, Ljubinka V. Rajaković, Mikro-elektro-mehanički sistemi (MEMS) – Tehnologija za 21. vek ............................................................................................... 629

Maja M. Kuzmanoski, Marija N. Todorović, Mira P. Aničić Urošević, Slavica F. Rajšić, Heavy metal content of soil in urban parks of Belgrade .................................................................................. 643

CONTENTS Ljubiša J. Bučanović, Mihailo P. Lazarević, Srećko N.Batalov, The

fractional PID controllers tuned by genetic algorithms for expansion turbine in the cryogenic air separation process ...... 519

Zuozhu Wu, Xinqi Qiao, Zhen Huang, Development and validation of a reduced mechanism for methane using a new integral algorithm in a premixed flame ................................................... 529

Xue-Fei Zhou, Oxidation of lignin–carbohydrate complex from bamboo with hydrogen peroxide catalyzed by Co(salen) ........... 541

Dimitrije Ž. Stevanović, Mića B. Jovanović, Marina A. Mihajlović, Jovan M. Jovanović, Željko B. Grbavčić, Application of pro-cess simulators in chemical engineering process design –natural gas separation plant case study .................................... 547

Sonja M. Vidojković, Antonije E. Onjia, Aleksandar B. Devečerski, Nebojsa N. Grahovac, Aleksandra B. Nastasović Economizer water-wall damages initiated by feedwater impurities ............ 559

Ljiljana M. Djekić, Marija M. Primorac, Formulation and character-isation of self-microemulsifying drug delivery systems based on biocompatible nonionic surfactants .......................... 565

Milovan Vuković, Nada Štrbac, Miroslav Sokić, Vesna Grekulović, Vladimir Cvetkovski, Bioleaching of pollymetallic sulphide concentrate using thermophilic bacteria ................................... 575

Milica M. Petrović, Jelena Z. Mitrović, Miljana D. Radović, Danijela V. Bojić, Miloš M. Kostić, Radomir B. Ljupković, Aleksandar Lj. Bojić, Synthesis of bismuth(III) oxide films based anodes for electrochemical degradation of Reactive blue 19 and Crystal violet ................................................................... 585

Aleksandra T. Ivanović, Biserka T. Trumić, Svetlana Lj. Ivanov, Saša R. Marjanović, Modeling the effects of temperature and time of homogenization annealing on the hardness of PdNi5 alloy .................................................................................. 597

Tatjana Nikolin, Mirjana Sevaljević, The examination of the seasonal influence on the efficiency in oil and fats removal through primary treatment from the wastewater of edible oil industry .................................................................................. 605

Branko B. Pejović, Vladan M. Mićić, Mitar D. Perušić, Goran S. Tadić, Ljubica C. Vasiljević, Slavko N. Smiljanić, Proposal for determining changes in entropy of semi ideal gas using mean values of temperature functions ....................................... 615

Tatjana A. Djakov, Ivanka G. Popović, Ljubinka V. Rajaković, Micro-electro-mechanical systems (MEMS) – technology for the 21st century ................................................................................. 629

Maja M. Kuzmanoski, Marija N. Todorović, Mira P. Aničić Urošević, Slavica F. Rajšić, Heavy metal content of soil in urban parks of Belgrade.................................................................................. 643

GENERALNI POKROVITELJ

HEMOFARM KONCERN

VRŠAC, Beogradski put bb, tel. 013/821-345, 821-027, 821-129 BEOGRAD, Prote Mateje 70, tel. 011/344-26-63, faks: 344-17-87

E-pošta: [email protected]

IZDAVANJE ČASOPISA POMOGLA JE:

INŽENJERSKA KOMORA SRBIJEBulevar vojvode Mišića 37 11000 Beograd

SUIZDAVAČ I

Tehnološko-metalurški fakultet

Univerziteta u Beogradu, Beograd

Prirodno-matematički fakultet Univerziteta u Novom Sadu, Novi Sad

Hemijski fakutet Univerziteta u Beogradu Beograd

Institut za tehnologiju nuklearnih i drugih

mineralnih sirovina, Beograd

HIP Petrohemija a.d. Pančevo

Tehnološki fakultet Univerziteta u Novom Sadu, Novi Sad

NU Institut za hemiju, tehnologiju i metalurgiju Univerziteta u Beogradu, Beograd

„Nevena Color“ d.o.o.

Leskovac

Tehnološki fakultet Univerziteta u Nišu, Leskovac

DCP Hemigal, Leskovac

519

The fractional PID controllers tuned by genetic algorithms for expansion turbine in the cryogenic air separation process

Ljubiša J. Bučanović1, Mihailo P. Lazarević2, Srećko N.Batalov2 1Department of production of technical gases, Messer Tehnogas, Bor, Serbia 2Faculty of Mechanical Engineering, University of Belgrade, Serbia

Abstract This paper deals with the design of a new algorithm of PID control based on fractionalcalculus (FC) in production of technical gases, i.e. in a cryogenic air separation process. Production of low pressure liquid air was first introduced by P.L. Kapitsa and involved expansion in a gas turbine. For application in the synthesis of the control law, for the inputtemperature and flow of air to the expansion turbine, it is necessary to determine the appropriate differential equations of the cryogenic process of mixing of two gaseousairflows at different temperatures before entrance to the expansion turbine. Thereafter,the model is linearized and decoupled and consequently classical PID and fractional order

β αPI D controllers are taken to assess the quality of the proposed technique. A set ofoptimal parameters of these controllers are achieved through the genetic algorithmoptimization procedure by minimizing a cost function. Our design method focuses onminimizing performance criterion which involves IAE, overshoot, as well as settling time. Atime-domain simulation was used to identify the performance of β αPI D controller with respect to a traditional optimized PID controller.

Keywords: technical gases, cryogenic liquid, fractional PID control, FOPID optimal tuning,genetic algorithms.

SCIENTIFIC PAPER

UDC 621.438:661.9:681.5.015:004

Hem. Ind. 68 (5) 519–528 (2014)

doi: 10.2298/HEMIND130717078B

Available online at the Journal website: http://www.ache.org.rs/HI/

Cryogenic air separation process is used to produce large quantities of purified oxygen, nitrogen or argon for the steel, chemical, food processing, semiconductor and health care industries [1,2]. A cryogenic air separation process is operated at extremely low temperatures (–170 to –190 °C) to separate air com-ponents according to their different boiling tempera-tures. It takes place in air separation units (ASUs) which present cryogenic distillation systems. Due to a high demand for these commodity materials, the ASU has become a crucial technology in many processes, inc-luding the next generation power plants. Cryogenic air separation process is an energy intensive process that consumes a tremendous amount of electrical energy. Therefore, ASU must automatically and rapidly respond to the changing product demands from customers. Pro-cess designers have increasingly utilized mass and energy integration, substituted less efficient unit oper-ations by more efficient ones, accelerated the develop-ment of machinery and equipment used in the process, generated the alternatives to cryogenic production of gaseous products, and set the foundation for imple-mentation of advanced control strategies. As expected,

Correspondence: M. Lazarević, Department of Mechanics, Faculty of Mechanical Engineering, University of Belgrade, Kraljice Marije 16, 11020 Belgrade, Serbia. E-mails: [email protected]; [email protected] Paper received: 17 July, 2013 Paper accepted: 21 October, 2013

there is significant economic interest in reducing the operating costs of ASUs through advanced process control technology. So far, the dominating control prac-tice in ASU processes has been to adapt traditional regulatory controllers to maintain good performance. A number of studies of the process control and optimi-zation of the cryogenic air separation process have been published. For example, in [3] air separation con-trol technologies are discussed and model predictive control is considered as the current state-of-the-art control technology in the air separation industry. Authors in [4,5] investigated the control structure sel-ection and linear model predictive control tuning for an air separation plant.

Here, we focus our research on improvement of implementation of traditional PID controllers. It is known that, due to its functional simplicity and perfor-mance robustness, PID controllers have been widely used in the process industries [6,7]. However, in the recent years, the emergence of effective methods of solving differentiation and integration of non-integer order equations makes fractional-order systems more and more attractive for the control systems commun-ity. Fractional calculus (FC) has existed for over three centuries and the fractional integral-differential oper-ators are generalization of integration and derivation to non-integer order (fractional) operators [8–10]. The increasing number of studies related to the application of fractional controllers in many areas of science and

Lj.J. BUČANOVIĆ et al.: PID CONTROLLERS TUNED BY GENETIC ALGORITHMS FOR EXPANSION TURBINE Hem. ind. 68 (5) 519–528 (2014)

520

engineering is remarkable, where special fractional-order systems are of interest for both modelling and controller design purposes [11–13]. In some of these works, it is verified that the fractional-order controllers can have better disturbance rejection ratios and less sensitivity to plant parameter variations compared to the traditional controllers. According to the develop-ment of fractional order calculus application in the recent years, the researchers are trying to replace the common PID controller use by the use of fractional controller [11,14–16]. Frequency domain approach in using fractional order PID controllers was also studied [17]. So, the fractional-order PIDs (FOPID) are becoming an important research topic since they result into more tuning parameters allowing robust performances to be attained. Nevertheless, it is well known that the addi-tional degrees of freedom, i.e., varying orders of integ-ration and differentiation, are accompanied with some complexity in the synthesis, even if fractional operators allow a compact representation of such high order con-trollers which means that only few parameters need to be adjusted. Further research activities run towards defining new effective tuning techniques for non-inte-ger order controllers. Generally, two tuning methods of

β αPI D controllers are distinguishable, analytic and heuristic [18–20].

In this paper, we propose and elaborate new opti-mal algorithms of fractional PID control based on gene-tic algorithms (GA) in the production of technical gases, i.e., in the cryogenic air separation process. GAs have received much interest in recent years [21–23], the basic operating principles of GA being based on the principles of natural evolution. GA is a stochastic global adaptive search optimization technique based on the mechanisms of natural selection. GA can solve non-linear multi-objective optimization problems, requires little knowledge of the problem itself, and does not require that the search space is differentiable or con-tinuous. GA does not suffer from the basic setback of traditional optimization methods such as getting stuck at local minima. In this regard, a GA is used to find out optimal settings for a fractional β αPI D controller in order to fulfil different design specifications for the closed-loop system, taking advantage of the fractional orders, α and β. We propose time-domain criterion which involves integral absolute error (IAE), overshoot, as well as settling time. This will be done through a fitness function to achieve rise in the performance indices. Both PID and β αPI D controllers, where the gains are optimized by genetic algorithm, are applied to a cryogenic process of mixing of two gaseous air flows at different temperatures before entrance of expansion turbine in the cryogenic air separation process. The quality of the system response is verified by a simul-ation study.

MODEL OF THE CRYOGENIC PROCESS OF MIXING OF TWO GASEOUS AIR FLOWS BEFORE ENTRANCE OF EXPANSION TURBINE

Cryogenics is the science and technology dealing with temperatures less than about 120 K, although this historical summary does not adhere strictly to 120K definition. The techniques used to produce cryogenic temperatures differ in several ways from those dealing with conventional refrigeration. Also, liquid air has been cooled to very low temperatures (cryogenic tem-peratures) so that it has condensed to a pale blue mobile liquid. To protect it from room temperature, it must be kept in a vacuum flask. In practice, these two areas often overlap and the boundary between conven-tional and cryogenic refrigeration is often indistinct. Significant reductions in temperature often have very pronounced effects on the properties of materials and behaviour of the systems. A new way of technical pro-duction of liquid air has been obtained by C. Linde at the end of the nineteenth century [1–3]. Production of low pressure liquid air was first introduced in 1938. by Russian academician P.L. Kapitsa, and includes the pro-duction of liquor air at pressure, p2, of 6–7 bar and its expansion in a gas turbine. So, the expansion turbine in the liquid air production, used for expansion of air from thermodynamic state ( )p pP ,p T to state ( )k kK ,p T , low-ering air temperature from Tp to Tk and the pressure from pp to pk, Fig. 1. Expansion of cold air after the equipment is started, creates a waste heat due to exchange of heat with the environment during this work. The amount of air that expands in the gas tur-bine, does not exceed 25% of the amount of usable air [24]. The air from the compressed state 1, is cooled down to the state 2 by turbo-compressor, Fig. 1b. The compressed air at pressure p2 is led to the reverse heat exchanger where it is cooled to the state 3. Part of the air from the middle of heat exchanger state 3* and part at the state 3, which constitute (kg/kg)em of com-pressed air, are led to the expansion turbine where, after expansion, the state 8 at pressure p1 is obtained. Because of the losses and other non-reversible pro-cesses, the expansion does not follow the adiabatic line to state ad8 , but to state 8, which is shifted to the right. The place for removal of air at state 4 is elected to be at state 8, at the end of expansion, which is in the area near the upper limiting curve (in the TS diagram

= 1.0x ), by 1–3 K above the temperature of saturated steam. The basic devices of the plant are: ТК – turbo-compressor, H – air refrigerator, RR – reverse heat exchanger, ET – expansion turbine, RK – heat exchan-ger, i.e., air condenser, and PV – damping valve.

Liquid air quantity, TVm , can be determined on the basis of the heat balance∗ [25]: ∗ The list of used symbols is given in -Nomenclature.


521

( ) ( )= + − + − −'2 TV 7 TV 10 4 8 do1 (kJ/kg)eh m h m i m h h q (1)

where do (kJ/kg)q is the heat from the environment brought by 1 kg of air, mTV (kg/kg) is the mass of the liquid air and me (kg/kg) is the mass of the air which expands in the expansion turbine. In the ideal case, when ≈do 0q and Δ = − ≈nr 1 10 0T T T , the liquid air mass is:

− − −−= + ≤ +

− − − −10 2 4 8 4 81 2

TV e e' ' ' '10 7 10 7 10 7 10 7

h h h h h hh hm m m

h h h h h h h h (2)

The main advantage of Kapitsa's procedure, com-pared to the other procedures for liquid air production [2], is that even at a low p2 pressure it still does not have to spend excessive work for production of 1 kg of liquid air. Since the turbine is capable of much greater flows than the reciprocating compressor, the proce-dure is adapted for large plants such as are met in prac-tice. For performing expansion of the part of the air after recuperation, in the Factory of technical gasses in Bor, two expansion turbines are built-in, one which is permanently in the process and the other is hot stand-by. The energy obtained in the expansion turbine during the process is used for driving the fan which absorbs atmospheric air, irrespective of the air flow in the gas turbine. The fan compresses the air and in this way prevents an unlimited growth of the number of turns of the turbine; the compressed air is let to the atmosphere which is not rational from the energy point of view. For the purpose of the synthesis of the control of the input air temperature and air flow through the expansion turbine, it is necessary to determine the cor-responding linearized differential equations of the part of the cryogenic process of mixing two streams of gase-ous air of different temperatures at the entrance to the expansion turbine. Figure 2 presents diagram of the process and symbolic-functional scheme with the rele-vant variables nominal value of gaseous air flow at the

entrance to the expansion turbine = 356N N7600 m /hG ,

nominal value of gaseous air temperature at the entrance to the expansion turbine =5 124 KNT , nominal value of temperature of gaseous air in the heat exchanger, =1 153 KNT , nominal value of temperature of air at the end of the cold heat exchanger =3 101 KNT , nominal value of the position of control valve TV946A

=AN 14.7 mmY , nominal value of the position of con-trol valve TV946B =BN 30.2 mmY . Mathematical model-ing of this process is based on the following assump-tions.

Figure 2. Diagram of the cryogenic process (а)and symbolic-functional scheme (b).

A1. Airflow field is homogeneous in the pipeline. Continuity equation for the air flowing through the pipe-line has the following form:

Figure 1. Schematic diagram of the liquid air flow within the plant (a) and TS diagram of the process (b).


522

( ) ( ) ( ) ( )= + −sc12 34 56

dd

M tG t G t G t

t (3)

A2. Air temperature field is homogeneous in the pipeline. Based on this assumption, the heat balance equation is:

( ) ( )( ) ( ) ( )( ) ( ) ( ) ( )

= +

+ − +

512 2

34 4 56 6 o

d

dscM t h t

G t h tt

G t h t G t h t Q (4)

A3. Temperature increment through the pipe wall is negligibly small. Since the pipeline is located in a cold block which is insulated from the environment by perlite material, it can be assumed that ≈o 0Q .

A4. Enthalpies of the air in the given sections can be expressed as functions of appropriate temperatures:

( ) ( ) ( )( ) ( ) ( )( ) ( ) ( )

θ θθ θ

θ θ

= ≈

= ≈

= ≈

2 2 1

4 4 3

6 6 5

;

;p p

p p

p p

h t c t c t

h t c t c t

h t c t c t

(5)

A5. Average air velocity in section 5 is approxi-mately constant, ( ) = = ≈5 5N 5 const.w t w w .

A6. Static characteristics of the control valves are linear.

A7. Control-valve HV912 is opened. A8. Mass flow rate through the control valves dep-

ends only on the linear displacement of control valves:

( ) ( ) ( ) ( )= =12 vA A 34 vB B;G t k Y t G t k Y t (6)

By adopting relative perturbations of the relevant values, the following choice is made:

( ) ( ) ( )Δ = − =56 56 56N 01G t G t G x t

( ) ( ) ( )θ θ θΔ = − =5 5 5N 02t t x t

( ) ( ) ( ) ( )Δ = − = =A A AN A 1 ;Y t Y t Y y t u t

( ) ( ) ( ) ( )Δ = − = =B B BN B 2 ;Y t Y t Y y t u t (7)

( ) ( ) ( )θ θ θΔ = − =1 1 1N 1 ;t t z t

( ) ( ) ( )θ θ θΔ = − =3 3 3N 2t t z t

Therefore, it is possible to obtain a system of dif-ferential equations which presents nonlinear mathe-matical model as follows:

( ) ( ) ( ) ( )( )=o11 o1 A B

d, ,

dx t

f x t y t y tt

(8)

( )

( ) ( ) ( ) ( ) ( ) ( )( )=

=

o2

2 o1 o2 A B 1 2

dd

, , , , ,

x tt

f x t x t y t y t z t z t (9)

A9. Deviations of all values are small enough so that function ( )⋅2f can be replaced by the first two terms of the corresponding Maclaurin series.

After Eqs. (8) and (9) are linearized, one can obtain linear differential equations that describe the cryogenic process of mixing of two gaseous air flows of different temperatures before entrance to the expansion tur-bine, given as the appropriate equations of the state and output as follows:

( ) ( )

( )

− = + −

+ + −

0.2 00 0.2

45.736 28.07 0 0( )

0.174 0.085 0.088 0.112

x t x t

u t z t (10)

( ) ( ) =

o1 00 1

x t x t (11)

or, in condensed form:

( ) ( ) ( ) ( ) ( ) ( )= + + = o,u zx t Ax t B u t B z t x t Cx t (12)

where the corresponding vectors are:

( ) ( ) = A B( )T

u t y t y t , ( ) ( ) = 1 2( )T

z t z t z t ,

and u Z, , ,A B B C are matrices of appropriate dimen-sions. From what has been said above, it is clear that the model represents a MIMO system, where the num-ber of inputs is to equal to that of the outputs, i.e., the system is “square”, therefore, it is possible to apply the decoupling control strategy, whereby each of the inputs can affect only one output. In that way, one may obtain the so-called non-interactive system where the transfer function W(s) of the system is decoupled, i.e., matrix of the system is diagonal and non-singular. To decouple the system, a new input u(t) is introduced by means of a feedback [26]:

( ) ( )= − +c c( )u t K x t F v t (13)

where are ci is i-th row of matrix C and:

( )( )

= ≠

≠= ∀ = − − = ∀

1 21 u 2 u u, ,..., ,det 0

min , 0, 0,1,2,..., 1

1, 0,

mTp p p

m

ji u

ij

i u

N c A B c A B c A B N

j c A Bp j n

n c A B j

(14)

Thus, one can obtain: + + +− −= = 1 21 1 11 1

1 2, ( ... )mp p p Tc c mF N K N c A c A c A (15)

Transfer function W(s) of the original system is:


523

( )−

+ += − =

− + +

1u

45.736 28.070.2 0.2( )

0.174 0.850.2 0.2

s sW s C sI A B

s s

(16a)

and, after applying new control:

( )−= − + 1u c u c( )W s C sI A B K B F (16b)

By taking into account the proposed procedure for c c,F K , it follows:

− −

= = = =

= = =1 2 1 2

1 1u c u c u

[1 0], [0 1], 0, 0,

, ,

c c p p

N B F B K B A (17)

and

( )− = =

1 1 / 0( )

0 1 /s

W s C sIs

(18)

Now, the decoupled system is:

== + +

1 1

2 2 1 2

,0.088 0.112

x vx v z z

(19)

GA-BASED OPTIMAL FRACTIONAL ORDER PID CONTROL

The essence of fractional calculus and FOPID

Fractional calculus (FC) is a mathematical topic with more than 300 years old history, but its application to physics and engineering has been reported only in the recent years. The fractional integral-differential oper-ators are generalization of the integration and deri-vation to non-integer order (fractional) operators. The applications of FC are very wide nowadays: rheology, viscoelasticity, acoustics, optics, chemical physics, ther-modynamics, robotics, control theory of dynamical sys-tems, electrical engineering, and bioengineering [7,8,16]. The main reason for the success of applications of FC is that these new fractional-order models are more accu-rate than integer-order models and that fractional deri-vatives provide an excellent instrument for the des-cription of memory and hereditary properties of vari-ous materials and processes due to the existence of a ”memory” term in their model. There are, today, many different forms of fractional integral operators, ranging from divided-difference types to infinite-sum types, Riemann-Liouville, Grunwald-Letnikov, Caputo’s, Weyl’s and Erdely-Kober, Jumarie’s, etc., definitions of frac-tional derivative [10,28,29]. Three definitions are generally used for the fractional differintegral. The first one is the GL definition, i.e., Grunwald and Letnikov developed an approach to fractional differentiation based on the following definition:

( )( )

( ) ( ) ( )

αα

α

α α→

≤ <∞

Δ=

Δ = − − >

GL 0

0

( ) lim ,

1 , 0

hx h

jh

j

f xD f x

h

f x f x jh hj

(20)

which is the left Grunwald-Letnikov (GL) derivative as a limit of the fractional order backward difference. Similarly, there exists the right one as:

( )( )

( ) ( ) ( )

αα

α

α α

−

→

−≤ <∞

Δ=

Δ = − + <

GL 0

0

( ) lim ,

1 , 0

hx h

jh

j

f xD f x

h

f x f x jh hj

(21)

As indicated above, the previous definition of GL is valid for α > 0 (fractional derivative) and for α < 0 (fractional integral) and, commonly, these two notions are grouped into one single operator called differint-egral. If = − /n t a h is considered, where a is a real constant which expresses a limit value, one may write:

( ) ( )αα

α−

→=

= − −

GL , 0

0

1( ) lim 1

t ah

ja t h

j

D f t f t jhjh

(22)

where t means the integer part of x, and a and t (in sub-script) are the bounds of the operation for α

GL , ( )a tD f t . The left Riemann-Liouville (RL) and the right RL frac-tional integral of the order α for function f(t), for α ∈,a R , can be expressed as follows:

( ) ( ) ( ) ( ) ( )αα α τ τ τα

−−≡ = −Γ 1

RL RL1 d

t

a aa

I f t D f t t f (23)

( ) ( ) ( ) ( )αα α τ τ τα

−−≡ = −Γ 1

RL RL1( ) d

b

b bt

I f t D f t t f (24)

where α α> − < <0, 1n n and Γ(.) is the well-known Euler’s gamma function. Furthermore, the left RL fractional derivative is defined as:

( ) ( ) ( ) ( )αα τ τ τα

− −= −Γ − 1

RL ,1 d

tnn

a t na

dD f t t fn dt

(25)

and the right RL fractional derivative is:

( ) ( )( ) ( ) ( )αα τ τ τ

α− −−

= −Γ − 1

RL ,1

dn bn

nt b n

t

dD f t t fn dt

(26)

where α− ≤ <1n n ; a and b are terminal points of interval [ ],a b , which can also be −∞ ∞, . Also, the Caputo fractional derivatives are defined as follows. The left Caputo fractional derivative is:


524

( ) ( ) ( ) ( ) ( )αα τ τ τα

− −= −Γ − 1

C ,1 d

tn n

a ta

D f t t fn

(27)

and the right Caputo fractional derivative is:

( ) ( )( ) ( ) ( )αα τ τ τ

α− −−

= −Γ − 1 ( )

C ,1

dn b

n nt b

t

D f t t fn

(28)

where τ τ τ=( )( ) ( ) / dn n nf d f and α +− ≤ < ∈1n n . The initial conditions of fractional differential equations with the Caputo derivative have a clear physical mean-ing and the Caputo derivative is extensively used in real applications. The Caputo and Riemann-Liouville formul-ations coincide when the initial conditions are zero [7].

Background of FOPID

Thanks to the widespread industrial use of PID controllers, even a small improvement in PID features, achieved by using β αPI D , could have a relevant imp-act. Recently published results, [12,15,27] indicate that the use of a fractional-order PID controller can improve both the stability and performance robustness of feed-back control systems. In his book [7], Podlubny pro-posed a generalization of the PID controller, namely fractional PID ( β αPI D ), where α and β are the order of integration and derivation, respectively, that can be real numbers. One of the most important advantages of the β αPI D controller is better control of the dyna-mical systems which are described by fractional order mathematical models. Another advantage lies in the fact that the β αPI D controllers are less sensitive to changes of parameters of a controlled system [12]. This is due to the two extra degrees of freedom to better adjust the dynamical properties of a fractional order control system. However, in theory, the β αPI D itself is an infinite dimensional linear filter due to the fractional order in the differentiator or integrator. This also imp-lies that the tuning of the controller can be much more complex. In order to address this problem, different methods for the design of a fractional order PID (FOPID) controller have been proposed in the literature. Some of these techniques are based on an extension of the classical PID control theory. In this paper, a fractional order PID controller ( β αPI D ) is used to control the process of production of technical gases as follows:

u e eα β−= + +p d 0 0( ) ( ) ( ) ( )t i tt K t K D t K D e t (29)

The continuous transfer function of the controller is obtained through Laplace transform of β αPI D :

( )β β α

β α β++ +

= >FOPID( ) , , 0p i dK s K K sG s

s (30)

The controller parameters are: proportional gain Kp, derivative gain Kd, integral gain Ki, as well as non-

integer order of derivative α and integrator β, Eq. (28). Unlike conventional PID controllers, there is no system-atic and rigor design or tuning method for β αPI D controllers. For practical digital realization, the deri-vative part has to be complemented by the first order filter:

α

β

= + + +

FOPID1

( ) 11

dp

di

T sG s K

Ts T sN

(31)

Several methods have been proposed for tuning β αPI D controllers [16,18–20] by many contemporary

researchers. Besides, for the most applications load disturbances are typically low frequency signals and their attenuation is a very important characteristic of a controller. It is shown in [6] that by maximizing integral gain Ki, the effect of load disturbance at the output will be minimum. It is observed that is difficult, for the general adjustment of fractional PID parameters, to satisfy the overall performance at the same time. The design of a fractional PID controller could be treated as a multi-objective optimization problem, which is to compromise the rapidity, stability and accuracy of sys-tem control. Some works use performance indices as the objective functions as follows: integral of the absol-ute value of the error (IAE), mean of the squared error (MSE), integral of time multiplied by absolute error (ITAE), integral of absolute magnitude of the error (IAE) and integral of the squared error (ISE):

( )( )= =

= = =

2

2 2

1( ) d , d ,

( ) d , ( ) d , ( ) d

IAE e t t MSE e t tT

ITAE t e t t ISE e t t ITSE te t t (32)

As a mathematical means for optimization, GA can naturally be applied to the optimal-tuning of fractional order PID controllers.

Optimal tuning of FOPID by using GA

Here, we propose using genetic algorithms (GA) for determining the optimal parameters of fractional order PID controllers, [22]. Recently, GA has been recognized as an effective and efficient technique to solve opti-mization problems [23]. GA is a search technique that manipulates the coding representation of a parameter set to search a near optimal solution through cooper-ation and competition among the potential solutions. This algorithm is highly relevant for the industrial appli-cation, because this algorithm is capable of handling problems with constraints, objectives and dynamic components. GA uses such natural evolution to get the global optimization. Therefore, this paper describes the application of GA to fine-tuning of the parameters for fractional PID controllers. In real coding implement-


525

ation, each chromosome is encoded as a vector of real numbers, of the same lengths as the solution vector. According to control objectives, five parameters Kp, Kd, Ki, α and β of a fractional PID controller are required to be designed in these settings. In this study, a next-to-optimality criterion is introduced which involves, besides steady state error e, i.e., IAE, also the overshoot Po, and settling time Ts:

= + + →0 minsJ P T e dt (33)

Fitness function is designed as:

= + −g max min gf J J J (34)

where Jmax and Jmin are the largest and the smallest values of J, respectively, observed thus far, and Jg is value of the criterion for the current population. All the GA parameters are arranged as follows:

– population size: N = 100; – crossover probability: =c 0.75p ; – mutation probability: ( )=m m0 min 1 l gp p , =m0 0.1p

– initial mutation probability, l = 25 - generation threshold, g, current number of gener-

ation; – generation gap, gr = 0.25. Here, as selection method the remainder stochastic

sampling with replacement is used. In our case, stop-ping conditions for GA are: the GA stops when the max-imum number of generation (2.5N) has been reached or the first 50% of individuals reaches approximately the same value of the fitness function.

SIMULATIONS AND DISCUSSION

To demonstrate the feasibility of the proposed approach to the control of a cryogenic process of mix-ing of two gaseous air flows at different temperatures before entrance of expansion turbine, the system shown in Fig. 1 is used for illustration. Both the fractional PID and conventional PID controllers are designed based on the proposed GA. The Crone approximation of second order [30] was used for the calculation of fractional derivatives and integrals. Here, in order to obtain step response, the simulation model has been developed by using Simulink Library of MATLAB and a special toolbox for non-integer control. In Table 1 the optimal para-meters of the FOPID as well as classical PID controller using GA are presented.

Table 1. The optimal parameters of the fractional PID con-troller and the conventional PID controller based on the proposed GA

Controller Kp Ki Kd β α J PID 1. 15 1 0 1 1 0.81

2. 15 7 0 1 1 19.15 FOPID 1. 13 3 11 0.034 0.073 0.24

2. 14 8 11 0.98 0.069 13.22

In our case, each individual vector has the FOPID parameters (five parameters) where for the purpose of reducing the optimization time, the ranges of FOPID parameters are selected as:

[ ] [ ] [ ]( ] [ ]α β

∈ ∈ ∈

∈ ∈

0,20 , 0,20 , 0,20 ,

0,1 , 0,1p i dK K K

(35)

Figure 3. The step responses of the ( ) 3

1 N( / )i hx t m gas’s air flow at the entrance to the expansion turbine using the optimized FOPID and conventional PID controller.


526

As can be seen in Fig. 3, the step response of ( ) ( )= 3

1 1 N(m /h)iy t x t – gaseous air flow at the entrance to the expansion turbine using the optimized FOPID and conventional PID controllers has a better transient response in the case of using FOPID. Also, we obtained with FOPID controller that overshoot is 0.006% and rise time is 0.184; on the other hand, with classical PID we have 0.43% overshoot and rise time is 0.248. In view of [19], disturbances affect only y2(t) = xi2(t), where the disturbances are = =1 210 K, 10 Kz z . Also, Fig. 4 rep-resents the step response of ( )[ ]=2 2( ) iy t x t K – gas’s air temperature at the entrance to the expansion turbine, obtained by applying the optimized FOPID and con-ventional PID controllers. In a similar way, we obtained that with robust FOPID controller overshoot was 8.19% and rise time was 4.520; on the other hand, with robust classical PID overshoot is 14.25% and rise time is 4.609.

Test of degree of robustness of the proposed FO PID controller

An efficient controller is the one that is still stable, even if a disturbance signal is applied to the plant. Therefore, to establish the effectiveness for a controll-ler, the robustness should be examined.

Particularly, after the optimal values of the FOPID controller have been obtained, the degree of robust-ness of the proposed FOPID controller with optimal values should be tested. The next types of disturbances are applied in turn to test the robustness of the FOPID controller:

– disturbances = = −1 2 10 Kz z . As can be observed in Fig. 5, by applying the pro-

posed disturbances = −1 10 Kz and = −2 10 Kz , the optimized FOPID and conventional PID controllers show significant of degree of robustness, but the step res-

ponse of ( )2ix t has better transient characteristic when using FOPID controller than when using the con-ventional PID controller.

CONCLUSION

The proposed genetic algorithm for the multi-objec-tive optimization design of a fractional PID controller, as well as of a classical PID controller, has been applied to the control of a cryogenic process of mixing of two gaseous air flows at different temperatures before entrance to the expansion turbine. This method allows the optimal design of all major parameters of the fractional PID controller thus enhancing the flexibility and capability of the fractional PID controller. In simul-ations the step responses of these two optimal con-trollers are compared. It was shown that FOPID con-troller improves transient response and provides a better robustness than the conventional PID, particul-arly in disturbance rejection.

Nomenclature

p – pressure, [bar] T,θ – temperature, [K or °C] h – specific enthalpy, [kJ kg–1] t – time, [ ]s x – state variable

ox – output variable u – control variable z – disturbance variable s – complex operator

doq – heat brought from the environment 1 kg air [kJ/kg]

oQ – heat exchanged with the surroundings [kW] scM – mass of gaseous air in the tube [kg]

Figure 4. The step responses of the ( ) 2ix t K – gas’s air temperature at the entrance to the expansion turbine using the optimized FOPID and conventional PID controller.


527

TVm – mass of liquid air em – mass of air which expands in expansion turbine

( )⋅h – specific enthalpy of gaseous air [kJ/kg] ( )⋅G – mass flow rate of gaseous air [ ]

3Nm /h , kg/h

( )Y – position of control valve TV946(.) [mm] ( )y – relative variation of position of the position

control valve ( )946 .TV [ ]mm ( )g t – relative variation of mass flow rate from the

nominal value of gaseous air

3Nm /h .

Acknowledgement

This work is partially supported by the Ministry of Education, Science and Technological Development of Republic of Serbia as Fundamental Research Project 35006.

REFERENCES

[1] R.F.Barron, Cryogenic Systems, Oxford Press, New York, 1985.

[2] W.F.Castle, Air separation and liquefaction: recent developments and prospects for the beginning of the new millennium, Int. J. Refrig. 25 (2002) 158–172.

[3] D.R.Vinson, Air separation control technology, Comput. Chem. Eng. 30 (2006) 1436–1446.

[4] J.O.Trierweiler, S. Engell, A case study for control struc-ture selection: air separation plant, J. Process Cont. 10 (2000) 237–243.

[5] J.O.Trierweiler, L.A. Frarina, RPN tuning strategy for model predictive control, J. Process Cont. 13 (2003) 591–598.

[6] K.J. Astrom, T. Hagglund,PID controller: Theory,design and tuning, Instrument Society of America, Research Triangle Park, NC, 1995.

[7] A. M. Cingara, New simple algebraic root locus method for design of feedback control systems, Hem. Ind. 62 (2008) 269–274.

[8] I. Podlubny, Fractional differential equations, Academic Press, San Diego, CA, 1999.

[9] K.B. Oldham, J. Spanier, The Fractional calculus: theory and applications of differentiation and Integration to arbitrary order, Academic Press, New York, 1974.

[10] H. Ji-Huan, Z.B Li, Converting fractional differential equations into partial differential equations, Therm. Sci., 16 (2012) 331–334.

[11] I. Podlubny, Fractional-order systems and λ μPI D -controllers, IEEE Trans. on Autom. Cont. 44(1) (1999) 208–214.

[12] C.A. Monje, V. Feliu, The fractional-order lead com-pensator, in IEEE Int. Conference on Computational Cybernetics. Vienna, Austria, 2004, pp. 1–6.

[13] A. Oustaloup, J. Sabatier, P. Lanusse, From fractal robustness to CRONE control, Frac. Calc. Appl. Anal. 2 (1999) 1–30.

[14] I. Petras, L. Dorcak, I. Kostial, Control quality enhan-cement by fractional order controllers, Acta Montan. Slovaca 3 (1998) 143–148.

[15] K. Bettou, A.Charef, Control quality enhancement using fractional λ μPI D controller, Int. J. Syst. Sci. 40 (2009) 875–888.

[16] C.A. Monje, Y. Chen, B. Vinagre, V. Feliu, Fractional Order Systems and Controls — Fundamentals and Applications, Springer, Berlin, 2010.

[17] B.M. Vinagre, I. Podlubny, L. Dorcak, V. Feliu, On fractional PID controllers: A frequency domain approach, in Proc. of IFAC Workshop on Dig. Cont. Past, Present and Future of PID Control, Terrasa, Spain, 2000.

[18] C.A. Monje, A.J. Calderon, B.M. Vinagre, Y.Q. Chen, V. Feliu, On fractional λPI controllers: some tuning rules for robustness to plant uncertainties, Nonlinear Dynam. 38 (2004) 369–381.

Figure 5. The step responses of the ( ) 2ix t K – gas’s air temperature at the entrance to the expansion turbine using the optimized FOPID and conventional PID controller under in case of disturbances = = −1 2 10 Kz z .


528

[19] C.A. Monje, B.M. Vinagre,Y.Q. Chen,V. Feliu, P. Lanusse,J. Sabatier, Proposals for fractional

λ μPI D tuning, in 1st IFAC Workshop on Fractional deri-vatives and applications, Bordeaux, France, 2004.

[20] J.Y. Cao, J.Liang , B.G. Cao, Optimization of fractional order PID controllers based on genetic algorithms, in Proceedings of the international conference of machine learning and cybernetics 9 (2005) 5686–5689.

[21] L. Chambers, The Practical Handbook of Genetic Algo-rithms, Applications, Chapman & Hall/CRC, New York, 2001.

[22] R. Haupt, S.E. Haupt, Practical Genetic Algorithms, Wiley-IEEE, New York, 2004.

[23] A.M. Almagrb, T. Hatam, S.B.Glisic, A.M.Orlović, Deter-mination of kinetic parameters for complex trances-terification reaction by standard optimisation methods, Hem.Ind. 68(2) (2014) 149–159.

[24] M. Mitovski, Energetic efficiency of cryogen process, Bakar, Bor, 1994 (in Serbian).

[25] F. Bošnjaković, Basic of heat transfer, Part I, Tehnička knjiga, Zagreb, 1970 (in Serbian).

[26] W. Brogan, Modern Control Theory, QPI, Prentice-Hall, Englewood Cliffs,NY, 1985.

[27] J.O. Sabatier, P. Agrawal, J.A. Tenreiro Machado, Advances in Fractional Calculus, Springer, Berlin, 2007.

[28] A.A. Kilbas, H.M. Srivastava, J.J Trujillo, Theory and Applications of Fractional Differential Equations, Elsevier, Amsterdam, 2006.

[29] G.Jumarie, Modified Riemann-Liouville Derivative and Fractional Taylor Series of Non-Differentiable Functions Further Results, Comput.Math. Appl. 51(10) (2006) 367–1376.

[30] D.Valerio, Fractional robust system control, PhD Thesis, Technical University of Lisboa, Lisbon, 2005.

IZVOD

PID KONTROLERI NECELOBROJNOG REDA PODEŠENI GENETSKIM ALGORITMIMA ZA UPRAVLJANJE EKSPANZIONE TURBINE U PROCESU SEPARACIJE UTEČNJENOG VAZDUHA

Ljubiša J.Bučanović1, Mihailo P. Lazarević2, Srećko N.Batalov2

1Odeljenje za proizvodnju tehničkih gasova, Messer Tehnogas, Bor, Srbija 2Mašinski fakultet, Univerzitet u Beogradu,Srbija

(Naučni rad)

Ovaj rad se bavi realizacijom jednog novog algoritma PID upravljanja zasno-vanog na računu necelobrojnog reda (fractional calculus) u proizvodnji tehničkihgasova, odnosno u procesu separacije utečnjenog vazduha. Proizvodnja uteč-njenog vazduha niskog pritiska je po prvi put bila uvedena od strane Kapice gde seekspanzija odvijala u gasnoj turbini.Za primenu u sintezi upravljanja ulazne tem-perature i protoka vazduha u ekspanzionoj turbini, potrebno je odrediti odgo-varajuće diferencijalne jednačine kriogenog procesa mešanja dva gasa na razli-čitim temperaturama na ulazu u ekspanzionu turbinu. Pri tome, odgovarajućimodel je linearizovan i dekuplovan gde su primenjeni istovremeno klasični PID kaoi β αPI D kontroleri necelobrojnog reda da bi se procenio kvalitet predloženognovog upravljanja datim procesom. Skup optimalnih parametara datih kontrolerase postiže primenom optimizacione procedure bazirane na genetskim algoritmima minimizovanjem odgovarajućeg kriterijuma optimalnosti. Naš metod se fokusira u okviru kriterijuma optimalnosti na smanjenje preskoka, vreme smirenja i mini-mizaciju integralne greske. Simulacije sprovedene u vremenskom domenu poka-zuju bolje performance optimalnog β αPI D kontrolera u odnosu na klasični opti-malni PID kontroler.

Ključne reči: Tehnički gasovi • Utečnjeni gas • PID upravljanje necelobrojnog reda • FOPID optimalno podešavanje •Genetski algoritmi

529

Development and validation of a reduced mechanism for methane using a new integral algorithm in a premixed flame

Zuozhu Wu, Xinqi Qiao, Zhen Huang

Key Laboratory of Power Machinery and Engineering, Ministry of Education, Shanghai Jiao Tong University, Shanghai, China

Abstract A new algorithm based on Computational Singular Perturbation (CSP) is proposed to cons-truct global reduced mechanism. The algorithm introduces species’ concentrations, spe-cies’ net production rates and heat release rates as integral weighting factors to integrate CSP-pointers, including radical pointers and fast reaction pointers, throughout the compu-tational domain. A software package based on the algorithm was developed to make thereduction process more efficient. Input to the algorithm includes: i) the detailed mecha-nism, ii) the numerical solution of the problem for a specific set of operating conditionsand iii) the number of quasi steady state (QSS) species. The proposed algorithm was applied to the reduction of GRI3.0 involving 53 species and 325 reactions leading to thedevelopment of a 15-species reduced mechanism with 10 lumped steps. Then the reducedmechanism was validated in a one-dimensional, unstretched, premixed, laminar steadyflame over a wide range of equivalence ratio, and excellent agreements between resultscalculated with the detailed and the reduced mechanisms can be observed.

Keywords: mechanism reduction, CSP, integral-CSP algorithm, GRI3.0.

SCIENTIFIC PAPER

UDC 519.1:004

Hem. Ind. 68 (5) 529–539 (2014)

doi: 10.2298/HEMIND130422079W


Simulation with chemical kinetic mechanisms plays an increasingly important role in both combustion design and scientific discoveries. However, even for the simplest fossil fuels, combustion simulation with detailed mechanisms still involves hundreds of species and thousands of reactions, which causes large amount of CPU times and low economic efficiency. The simul-ation is further complicated by the existence of highly reactive radicals which induce significant stiffness to the governing equations due to the dramatic diffe-rences in the time scales of the species. Consequently, it is necessary to develop reduced mechanisms with fewer variables and moderated stiffness from the detailed mechanisms, while maintaining the accuracy and comprehensiveness of the detailed mechanisms.

There have been various mechanism reduction methods developed to generate reduced mechanisms. Skeletal reduction techniques, including sensitivity analysis performed by multiplying the rate constant of a reaction by a factor of 2 (both forward and reverse rate constants) [1] or by solving the sensitivity equa-tions, principal component analysis [2], path flux anal-ysis (PFA) [3], directed related graph (DRG) [4–6], directed relation graph method with error propagation (DRGEP) [7], the DRG-aided sensitivity analysis (DRGASA)

Correspondence: Z. Wu, Key Laboratory of Power Machinery and Engineering, Ministry of Education, Shanghai Jiao Tong University, Shanghai 200240, China. E-mail: [email protected] Paper received: 22 April, 2013 Paper accepted: 24 October, 2013

[8], and DRGEP with sensitivity analysis (DRGEPSA) [9] are used to develop skeletal mechanisms by eliminating unimportant species and reactions while introducing acceptable errors which can be controlled by self-defined threshold value. The PFA method analyses the formation and consumption fluxes of each species at multiple reaction path generations and uses the fluxes to identify the important reaction pathways and the associated species. The DRG algorithm maps species to a graph and consequently identifies the species strongly coupled to the major species, thus solving strongly connected components (SCC) group by group successively. The DRG method uses absolute reaction rates, which makes the relation index not conservative (the interaction coefficient or relation index is the ratio of species’ flux). The DRGEP method which employs net reaction rates fails to pick up all of the reaction path when more than one intermediate specie exist and to identify the relation between the species that have both fast production and consumption rate, such as species having catalytic effect [3,10].

Skeletal mechanisms obtained only by removing redundant species and reactions are still stiff and con-tain some important radicals with short time scales which cannot be eliminated. The most acceptable approach to resolve the stiffness is further reducing the skeletal mechanisms by reasonable assumption, such as partial-equilibrium approximation and quasi-steady state approximation (QSSA) [11–17]. The partial-equi-librium approximation assumes that some reactions are partial-equilibrium and the forward and reverse reac-

Z. WU et al.: REDUCED MECHANISM FOR METHANE Hem. ind. 68 (5) 529–539 (2014)

530

tion rates are equal [18], and therefore the net pro-duction rates are approximate zero. The quasi-steady state approximation is based on the assumption that the species are in quasi-steady and their net production rates are zero. The partial-equilibrium approximation and the quasi-steady state approximation are effective ways in reducing the stiffness of reaction systems. However, they are based on the local (in time and space) conditions, which means the QSS species may become non-QSS species in a different condition. Con-sequently the resulting reduced mechanisms got by the partial-equilibrium approximation and the quasi-steady state approximation do not always provide best accur-acy. To resolve such problems, wide range of con-ditions (in time and space) in which the derived reduced mechanisms may be applied have to be taken into account, which will make the reduction process more complicated and limit the range of application of the reduced mechanisms. The computational singular perturbation (CSP) [19–24] method using a program-mable computational algorithm generates time-res-olved simplified models without the need of intuition and experiences. CSP data, such as radical pointer and fast reaction pointer, can be acquired by decoupling and analyzing the Jacobian matrix depending only on the state of a reaction system. The QSS species and fast reactions can be identified by analyzing the CSP data. Reduced mechanisms can be acquired by eliminating the QSS species whose concentrations are calculated by non-QSS species. A reduced mechanism can be called “global” if it is developed and based on global QSS species which do not vary according to time and space.

Lu [25] used the user specified error tolerance to distinguish the quasi steady state (QSS) species from non-QSS species for methane–air mixtures sampled from PSR and auto-ignition, and deduced a 15-steps reduced mechanism. Belcadi [26] derived a 10-steps reduced mechanism from GRI3.0 [30] using the fully automatic algorithm S-STEP [27] for a premixed com-bustion flame. A. Massiasa [19] applied the CSP method to a laminar premixed CH4/air flame and derived a 7-step reduced mechanism from a complex detailed chemical kinetics mechanism consisting of 279 reac-tions and 49 species. The PSR reactor used in Lu [25] is easy to handle, but provide no information about the changes of the state of reaction system (species con-centrations, temperature, etc.) over time and space. Reduced mechanisms deduced in such reactor may not provide satisfied simulation results in transient or in spatial inhomogeneous reactors. Massiasa [19] integ-rated the radical pointers with species’ concentrations and species’ production rates as weighting factors throughout the computational domain. This is reason-able for species with small concentrations or large pro-

duction rates which yield large radical pointers, and will be deemed as QSS species. The integrated radical pointers are better for choosing QSS species, but are prone to neglect some domains where species’ con-centrations are relatively too large or species’ pro-duction rates are too small. Such domains make little contribution to the final integrated radical pointers, thus causing these domains being neglected. Such problem is called “neglecting” problem.

Heat release rate is a critical parameter in eval-uating the intensity of a reaction system. Domains with higher heat release rate mean intensive chemical process and should be placed more emphasis on. Adding heat release rate to weighting factors in the process of integrating radical pointers and fast reaction pointers can reflect the importance of each comput-ational domain and is a good way to relieve the “neg-lecting” problem. In this article, the emphasis is on changing the integrands and weighting factors in integ-rating process to get reasonable QSS species and fast reactions.

The structure of the manuscript is as follows. First, a brief outline of the CSP method shall be introduced. Then, the new integral CSP algorithm shall be pre-sented and shall be applied over GRI3.0 to develop global reduced mechanism. Finally, the obtained global reduced mechanism shall be validated in a one-dimen-sional, unstretched, premixed, laminar steady flame over a wide range of equivalence ratio.

METHODOLOGIES

Outline of the CSP method

A thorough description of the CSP method may be found in [20,21,28]. Here, an overview of the CSP method will be given. A general chemical reaction sys-tem which contains R elementary chemical reactions and N species can be expressed as:

≡ =d( ) ( )

dyg y SF yt

(1)

where y is the ×1N concentration vector of all the species, S the N×R stoichiometric coefficients matrix, F(y) the ×1R species’ production rates vector of the elementary reactions, respectively. By taking the time derivative of Eq. (1), we obtained:

=dd

g Jgt

(2)

where

∂=∂

gJy

(3)


531

is the time-dependent Jacobian matrix. J depends only on the state of reaction system at every time step. By undertaking eigendecomposition of the matrix J, J can be decomposed as:

= ΛJ A B (4)

where A is the matrix of basis vectors and B the inverse matrix of A. If ideal base vector A (eigenvectors) exists, then Λ reduces to a diagonal matrix and its diagonal elements are the eigenvalues of J. Supposing there are M modes being fast modes, and mA , mB being cor-responding ×N M , ×M N fast base matrix, respect-ively. Then the matrix:

≡ mm mQ A B (5)

is called fast projection matrix. The diagonal elements of mQ :

[ ]= diag mD Q (6)

are radical pointers, where D is a N-dimensional vector. A larger diagonal element suggests a better CSP radical candidate, namely QSS specie.

By defining the participation and importance index as [29]:

=

≡ = =

1

, 1,2,..., , 1,2,...,i r

i rr R

i rr

r

b s FP i M r R

b s F (7)

=

≡ = =

1

, 1,2,..., , 1,2,...,i r

i rr R

i rr

r

s FI i M r R

s F (8)

where irP and i

rI are the participation index and importance index, ib the i-th row vector in mB , rs (column vector) the stoichiometric vector of the r-th reaction, rF the reaction rate of the r-th reaction, i

rs the i-th element of the stoichiometric vector of the r-th reaction, we can estimate: i) where the major cancellations occur and ii) the contribution of each step in the production of i-th specie. The fast reactions for each QSS species can be identified by the importance index. The biggest i

rI in the i-th row means the r-th reaction is the fast reaction for the i-th specie.

For simple reaction system, such as PSR, which is homogeneous and steady state, the reduced mecha-nism can be deduced once the QSS species and fast reactions are identified. However, for those inhomo-geneous and transient reaction systems, a QSS specie in one domain may be a non-QSS specie in the other domains. To get global QSS species for constructing global reduced mechanisms, radical pointers in each computational domain need to be integrated to get

overall information. In the following section, such integral algorithm will be introduced.

Introduction of integral-CSP algorithm

Let iD stand for the i-th radical pointer in D. For spatial inhomogeneous reaction systems, an easy way is to integrate radical pointers directly:

≡ 10

1d

Li iI D xL

(9)

For transient reaction systems:

≡ 10

1d

Ti iI D tT

(10)

As indicated by Massias [19], which used the Eq. (9) to directly integrate the radical pointers on a one-dimensional, unstretched, premixed, laminar steady CH4/air flame, the 1

iI method produced poor results. The reactant CH4 was identified as QSS species, which is obviously absurd.

A better integral algorithm is to involve species concentrations and species production rates [19]:

ε ε≡

+ +20 1 max 2

1 1d

iLi ii i

qI D xL X q

(11)


ε ε≡

+ +20 1 max 2

1 1d

iLi ii i

qI D tT X q

(12)

where iD is the i-th radical pointer in D, iX the i-th specie concentration, iq the net production rate of i-th specie, max

iq the maximum production rate for i-th specie throughout the computational domain D, ε1 and ε2 are small positive numbers which are used to avoid numerical problems when iX and max

iq equal zero. ε1 and ε2 must be chosen with care. Their order must be much lower than iX and max

iq so that their influence to 2

iI can be neglected. By adding species’ concentrations and species’ production rates as weight-ing factors in the integral algorithm, the resulting QSS species turn out to be more reasonable [19]. This is consistent with the fact that species with small con-centrations and fast production rates, which result in larger 2

iI , are prone to be deemed as candidates for QSS species.

The method expressed by Eqs. (11) and (12) can give reasonable results. However, to relieve the “neg-lecting” problem, a better algorithm shall be presented next.

Let inR denote integrand for the i-th specie in the n-

th computational domain (representing n-th physical space domain in spatial inhomogeneous reactors or n-th time step in transient reactors):


532

ε ε≡

+ +1 max 2

1 ii i nn n i i

n

qR D

X q (13)

where inD is the original radical pointer, i

nX and inq the

specie concentration and the net production rate of i-th specie in the n-th computational domain, respect-ively. Equation (13) has similar form as Eq. (11) and (12). The percentage of i-th specie of the summation of all radical pointers in the n-th computational domain define as:

≡,sum

ii n

nn

RW

R (14)

where ,sumnR is the summation of all radical pointers in the n-th computational domain. With the definition of

inW , we can define a new algorithm:

≡ 30

1d

Li inI W x

L (15)


≡ 30

1d

Li inI W t

T (16)

By adding heat release rates to Eq. (14), we get:

≡,sum

ii nn n

n

RH h

R (17)

where nh is the heat release rate in the n-th computational domain. By integrating Eq. (17), we obtain:

≡ 40

1d

Li inI H x

L (18)


≡ 40

1d

Ti inI H t

T (19)

In this section, two new radical integral algorithms, 3I and 4I have been proposed. The global QSS species

can be identified once the global radical pointers have been determined. Fast reactions corresponding to each QSS species can be determined using Eq. (8). The global reduced mechanisms can be constructed after the QSS species and fast reactions being found out [13,18–20]. The global reduced mechanisms can be further reduced by using participation index and importance index with user-defined error criterion, which is called “simpli-fication” or “truncation”.

To construct a global reduced mechanism, the number of QSS species M which used to determine the steps of global reduced mechanism(N–M–E), the M QSS species and M fast reactions need to be known. The above method for producing these data will be dis-

cussed next in the context of a laminar premixed CH4/air flame.

Identification of QSS species and fast reactions

To construct global reduced mechanism over a specific problem, the number of QSS species M need to be determined in advance. However no algorithmic method exists for obtaining a prior M. A possible way is to sort the magnitude of integral radical pointers in descending order and then choose the first M species as QSS species, and the remaining N–M species as non-QSS species. Verify the M QSS species until the results is satisfied. If not, change M and iterate the above process. A good human intuition and experience would be helpful for this choosing procedure.

In the following discussion, we shall begin with GRI3.0 which contains 325 reactions, 53 species and 5 elements to develop a global reduced mechanism. A one-dimensional, unstretched, premixed, laminar steady CH4/air flame shall be the problem under study. The equivalence ratio is 1.0 and the pressure is 1 atm. It is presumed that there are 15 non-QSS species, which means the number of QSS species is 38 and 10-steps reduced mechanism. Table 1 shows the twenty species with the smallest integral radical pointer in descending order, calculated with 1

iI , 2iI , 3

iI and 4iI , respectively.

The 1iI method which excludes Ar (inert specie) as non-

-QSS species leads to unrealistic results. The 3iI and 4

iI method produce almost the same results except for the order of the non-QSS species. The 2

iI method contains two different non-QSS species (including CH2CO and C2H2) compared with the 3

iI and 4iI method (including

HO2 and CH2O). In this manuscript, the 4iI method will

be adopted to identify the global non-QSS species. [Table 1] By analyzing with the 4

iI method, the following 15 species are deemed as non-QSS species: HO2, CH2O, H, NO, O, OH, CH4, CO, H2, CO2, H2O, N2O, O2, N2, Ar and the remaining 38 species will be identified as QSS species.

Fast reactions for each QSS species can be found by integrating importance index in Eq. (8):

≡ 01

dLi i

r r nF I h xL

(20)


≡ 01

dTi i

r r nF I h tT

(21)

where irI is the importance index in Eq. (8), nh the heat

release rate in the n-th computational domain. The fast reactions indexes for the 38 QSS species, CN, H2CN, HCNN, C, NH2, NH3, NNH, C2N, NCO, NH, HCN, CH, HCNO, N, C3H7, HNO, HCCOH, HOCN, CH2OH, C2H3, CH2S, HCCO, CH2, CH2CHO, HCO, C3H8, C2H5, HNCO,


533

CH3O, NO2, CH3CHO, C2H2, CH3OH, C2H6, CH2CO, C2H4, CH3 and H2O2 are 220, 237, 241, 49, 201, 278, 204, 171, 224, 192, 250, 125, 274, 180, 319, 212, 82, 273, 169, 173, 97, 176, 148, 304, 168, 312, 286, 249, 119, 187, 301, 295, 95, 158, 114, 25, 10 and 85.The reaction rates of the global reduced mechanism are calculated by those of the detailed mechanism. However, the reac-tion rates of the 38 fast reactions will not be included. Once the global QSS species and fast reactions have been identified, the global reduced mechanisms can be constructed. This is the content of next part.

The Integral CSP (I-CSP) software package and global reduced mechanism

A software package called Integral CSP (I-CSP) was developed to construct the global reduced mechanism. The I-CSP is written in C++ language and uses the interface functions provided by CHEMKIN-PRO to read the reaction solutions generated in Chemkin. The source code can be delivered by corresponding author on request. In this manuscript, first, a one-dimensional, unstretched, premixed, laminar steady CH4/air flame is run in Chemkin. Then, the I-CSP reads the necessary information, such as species’ concentrations, tempera-tures, heat release rates, etc. Next, the I-CSP calculates the radical pointers and fast reaction pointers in each domain and integrates these pointers to get integral radical pointers and integral fast reaction pointers which are used in identifying global QSS species and related fast reactions. Finally, once these 15 species:

H2, H, O, OH, HO2, CH4, CO2, CH2O, NO, N2O, CO, O2, H2O, N2 and Ar (the same order as they appear in the detailed mechanism) are identified as non-QSS species. The following 10-steps global reduced mechanism can be constructed with the help of I-CSP:

2 2 2H O=H +0.5O (S1)

2 20.5H O=H+0.25O (S2)

20.5O =O (S3)

2 20.25O +0.5H O=OH (S4)

2 2 20.75O +0.5H O=HO (S5)

2 4 2CO+2H O=CH +1.5O (S6)

2 2CO+0.5O =CO (S7)

2 2 2H O+CO=CH O+0.5O (S8)

2 20.5O +0.5N =NO (S9)

2 2 20.5O +N =N O (S10) in which Ar is not include for it is an inert specie.

The global reaction rates of the ten steps could be further simplified by using importance index with a user-defined error criterion. Samples of some of the ten non-QSS species are displayed below:

= + + + + +

+ + = − − −− − − −

40 38 40 35 40 34 40 36 40 4338 35 34 36 43

40 33 40 3733 37

d[ ]d

0.8246 0.0872 0.03090.0194 0.0180 0.0099 0.0088

H s F s F s F s F s Ft

s F s F (22)

Table 1. The 20 species with smallest integral radical pointers calculated with 1iI , 2

iI , 3iI and 4

iI

No. 1iI 2

iI 3iI 4

iI 20 NO C2H6 C2H2 C2H6 19 CH2CO C2H4 C2H6 CH2CO 18 CH2CHO HO2 CH2CO C2H4

17 HCO CH2O C2H4 CH3 16 CH2O CH3 CH3 H2O2

15 C2H6 CH2CO HO2 HO2 14 13 12

CH3CO CH3OH

CH3

H NO

C2H2

CH2O NO H

CH2O H

NO 11 CO CH4 O O 10 HO2 O OH OH 9 H2O2 OH CH4 CH4 8 O CO CO CO 7 H H2O H2 H2 6 CO2 N2O H2O CO2 5 H2O O2 CO2 H2O 4 OH CO2 N2 N2O 3 H2 N2 N2O O2 2 CH4 H2 O2 N2 1 O2 Ar Ar Ar


534

= + + + +

+ + = ++ − +

42 85 42 38 42 86 42 4385 38 86 43

42 87 42 487 4

d[ ]d

0.4781 0.3545+0.1536 +0.0079 0.0043 0.0007

OH s F s F s F s Ft

s F s F (23)

= + + +

+ + = −− −

44 52 44 118 44 98452 118 98

44 150 44 160150 160

d[ ]d

0.9761+ 0.01460.0051 0.0029 + 0.0013

CHs F s F s F

ts F s F (24)

where d[ ] / dC t is the derivative of the concentration of specie C over time, i

ks the stoichiometric coefficient of i-th specie in k-th reaction, kF the reaction rate of the k-th reaction. The number below is the integral important index ordered in descending magnitude. Sufficient terms are kept so that the total error of the omitted term is below a user-defined error criterion.

Validation of the global reduced mechanism

The 10-steps global reduced mechanism was vali-dated over a one-dimensional, unstretched, premixed, laminar steady CH4/air flame, which is the same cir-cumstance as the 10-steps reduced mechanisms being conducted. Figures 1–3 show the results of mole frac-tion of three major species CH4, CO2 and CO, and four minor species H, O, OH and NO, and temperature calculated with GRI3.0 and the 10-steps reduced mech-anism at equivalence ratio being 0.6, 1.0 and 1.5 and pressure being 1 atm, respectively. In Fig. 1a, the tem-peratures calculated by GRI3.0 and the 10-steps redu-ced mechanism agree well. Same agreement can be found in the comparison of mole fraction of CH4, CO2 and CO calculated by GRI3.0 and the 10-steps reduced mechanism. However, the 10-steps reduced mecha-

(a)

(b)

Figure 1. Comparison of mole fraction of (a) three major species CH4, CO2 and CO and temperature and (b) four minor species H, O, OH and NO, calculated with GRI3.0 and the 10-steps reduced mechanism at equivalence ratio being 0.6.


535

nism over predicts the mole fraction of CO when the distance is less than 0.4cm. Figure 1b displays the changes of mole fraction of H, O, OH and NO against distance. Small discrepancies between results calcul-ated by GRI3.0 and the 10-steps reduced mechanism can be found in the region of distance being less than 0.4 cm, where the mole fraction of H, O and OH grows rapidly. In the area of distance from 0.4 to 0.7 cm, the mole fraction of the three species decreases. In the following area of distance from 0.7 to 2.0 cm, the mole fraction increases with almost the same slope. Diffe-rent from the aforementioned three species, NO inc-reases once it is created and excellent agreement between results calculated by GRI3.0 and the 10-steps reduced mechanism can be observed.

Figure 2 shows the results calculated at equivalence ratio being 1.0 where the reduced mechanism is con-

ducted. Good agreement between results calculated by GRI3.0 and the 10-steps reduced mechanism can be observed. Figure 3 displays reasonable agreement between results calculated by GRI3.0 and the 10-steps reduced mechanism in the region of distance being over 0.4cm. However, in the region of distance being less than 0.4cm, the 10-steps reduced mechanism over predicts mole fraction of CO2, CO, H, O and OH, and underestimates mole fraction of CH4. Good agreement can be observed for temperature and NO by comparing results calculated by GRI3.0 and the 10-steps reduced mechanism.

Figure 4 displays the final (x→∞) mass fraction of NO and CO2 over a wide range of equivalence ratio (0.4–1.6). Good agreements can be obtained over all

(a)

(b)



536

(a)

(b)


Figure 4. The final mass fraction of NO and CO2 as a function of equivalence ratio, in a one-dimensional, unstretched, premixed, laminar steady CH4/air flame, calculated with GRI3.0 and 10-steps reduced mechanism.


537

equivalence ratios except for the NO while the equi-valence ratio is over 1.5. Figure 5 shows the final tem-perature with the equivalence ratio from 0.6 to 1.6. Satisfied agreements are obtained with the discrepancy less than 1.5%. The temperatures calculated with the 10-steps reduced mechanism are lower than those cal-culated with GRI3.0 while the equivalence ratio is below 1.5. However, the results are opposite while the equivalence ratio is upon 1.5.

It is clear from above validation that the 10-steps reduced mechanism produces high degree of accuracy in predicting concentration profiles of species and tem-perature on premixed flame. Although the global reduced mechanism is deduced on the condition of the equi-valence ratio being 1.0, it is can be used over wide range of equivalence ratio and produces good results.

CONCLUSIONS

A software package named Integral CSP (I-CSP) was developed to construct the global reduced mechanism. I-CSP introduces heat release rates, species’ concentra-tions and specie net production rates as weighting fac-tors to integrate radical pointers and fast reaction pointers throughout the computational domain. The input to I-CSP includes the detailed mechanism and the numerical solution of the problem on a specific set of operating conditions where the reduced mechanism is expected to be valid. The number of QSS species M need to be determined in advance. The output of I-CSP is three files which describe the reduced mechanism and the numerical relations between QSS species and non-QSS species. These files are compatible with Chem-kin, which makes the validation of the reduced mech-anism easier.

The I-CSP method was implemented over a one-dimensional, unstretched, premixed, laminar steady CH4/air flame and conducted a 10-steps global reduced mechanism. The reduced mechanism was validated over a one-dimensional, unstretched, premixed laminar flame with equivalence ratio being 0.6, 1.0 and 1.5. Good agreements can be observed and therefore demonstrate the validity of the reduced mechanism.

One challenge step is determining the number of QSS species M or the number of steps of global reduced mechanism. A feasible way to resolve the pro-blem is first choosing M approximately, then ordering the species with the help of CSP-pointers and checking out whether the species which should appear in the global reduced mechanism are in the N–M non-QSS species, next changing the value of M until reasonable

non-QSS species are found, finally validating the result-ing reduced mechanism until the error is less than user specified error tolerance. The procedure may iterate several times until the satisfied M is obtained. Expe-riences and good human intuition will be helpful in the M choosing procedure.

Acknowledgements

This work was supported by Shanghai Pujiang Pro-gram (11PJD014) and National Natural Science Found-ation of China (NSFC).

REFERENCES

[1] A.S. Tomlin, T. Turanyi, M.J. Pilling, Mathematical tools for the construction, investigation and reduction of combustion mechanisms, Compr. Chem. Kinet. 35 (1997) 293–437.

Figure 5. The final temperature as a function of equivalence ratio, in a one-dimensional, unstretched, premixed, laminar steady CH4/air flame, calculated with GRI3.0 and 10-steps reduced mechanism.


538

[2] S. Vajada, P. Valko, T. Turanyi, Principal component analysis of kinetic models. Int. J. Chem. Kinet. 17 (1985) 55–81.

[3] W. Sun, Z. Chen, X. Gou, Y. Ju, A path flux analysis method for the reduction of detailed chemical kinetic mechanisms, Combust. Flame 157 (2010) 1298–1307.

[4] T.F. Lu, C.K. Law, A directed relation graph method for mechanism reduction, Proc. Combust. Inst. 30 (2005) 1333–1341.

[5] T.F. Lu, C.K. Law, Linear time reduction of large kinetic mechanisms with directed relation graph: n-Heptane and isooctane, Combust. Flame 144 (2006) 24–36.

[6] T.F. Lu, C.K. Law, On the applicability of directed relation graphs to the reduction of reaction mechanisms, Combust. Flame 146 (2006) 472–483.

[7] P. Pepiot, H. Pitsch, An Efficient Error Propagation Based Reduction Method for Large Chemical Kinetic Mecha-nisms, Combust. Flame 154 (2008) 67–81.

[8] X.L. Zheng, T.F. Lu, C.K. Law, Experimental Counterflow Ignition Temperatures and Reaction Mechanisms of 1,3- -Butadiene, Proc. Combust. Inst. 31 (2007) 367–375.

[9] K. Niemeyer, C. Sung, M. Raju, Skeletal Mechanism Generation for Surrogate Fuels using Directed Relation Graph with Error Propagation and Sensitivity Analysis, Combust. Flame 157 (2010) 1760–1770.

[10] T. Ombrello, Y. Ju, A. Fridman, Kinetic ignition enhan-cement of diffusion flames by nonequilibrium magnetic gliding arc plasma, AIAA J. 46 (2008) 2424–2433.

[11] N. Peters, Lecture Notes in Physics, Glowinski et al., Eds., Springer, Berlin, 1985, p. 90.

[12] N. Peters, R.J. Kee, The computation of stretched laminar methane-air diffusion flames using a reduced four-step mechanism, Combust. Flame 68 (1987) 17–29.

[13] J.Y. Chen, A General Procedure for Constructing Reduced Reaction Mechanisms with Given Independent Relations, Combust. Sci. Technol. 57 (1988) 89–94.

[14] M.D. Smooke, Lecture Notes in Physics 384, Springer, Berlin, 1991, p. 1.

[15] Y. Ju, T. Niioka, Reduced kinetic mechanism of ignition for nonpremixed hydrogen/air in a supersonic mixing layer, Combust. Flame 99 (1994) 240–246.

[16] C.J. Sung, C.K. Law, J.Y. Chen, An augmented reduced mechanism for methane oxidation with comprehensive global parametric validation, Proc. Combust. Inst. 27 (1998) 295–304.

[17] T. Løvås, D. Nilsson, F. Mauss, Automatic reduction procedure for chemical mechanisms applied to pre-

mixed methane/air flames, Proc. Combust. Inst. 28 (2000) 1809–1815.

[18] D. Goussis, On the Construction and Use of Reduced Chemical Kinetic Mechanisms Produced on the Basis of Given Algebraic Relations, J. Comput. Physics 128 (1996) 261–273.

[19] A. Massias, D. Diamantis, E. Mastorakos, D. A. Goussis, An algorithm for the construction of global reduced mechanisms with CSP data, Combust. Flame 117 (1999) 685–708.

[20] S.H. Lam, D.A. Coussis, Understanding complex chemical kinetics with computational singular perturbation, Proc. Combust. Inst. 22 (1989) 931–941.

[21] S.H. Lam, D.A. Gousis, The CSP method for simplifying kinetics, Int. J. Chem. Kinet. 26 (1994) 461–486.

[22] M. Valorani, H.N. Najm, D.A. Goussis, CSP analysis of a transient flame-vortex interaction: time scales and manifolds, Combust.Flame 134 (2003) 35–53.

[23] M. Valorani, F. Creta, D.A. Goussis, J.C. Lee, H.N. Najm, An automatic procedure for the simplification of che-mical kinetic mechanisms based on CSP, Combust. Flame 146 (2006) 29–51.

[24] S.H. Lam, in Proceedings of 11th International Con-ference on Numerical Combustion, Granada, Spain, 2006.

[25] T.F. Lu, C.K. Law, A criterion based on computational singular perturbation for the identification of quasi steady state species: A reduced mechanism for methane oxidation with NO chemistry, Combust. Flame 154 (2008) 761–774.

[26] A. Belcadi, M. Assou, E. Affad, E. Chatri, Construction of a reduced mechanism for modelling premixed com-bustion of methane–air, Combust. Theory Modell. 11 (2007) 603–613.

[27] N. Vassilopoulos, Th. Massias, Y. Katsabanis, D.A. Goussis, S-STEP Code, Report for the ESPRIT program, Co. no. 8835, 1996.

[28] D.A. Goussis, S.H. Lam, A study of homogeneous methanol oxidation kinetic using CSP, Proc. Combust. Inst. 24 (1992) 113–120.

[29] S.H. Lam, Using CSP to understand complex chemical kinetics, Combust. Sci. Tech. 89 (1993) 375–417.

[30] G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Goldenberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, Jr., V.V. Lissianski, Z.i Qin, http://www.me.berkeley.edu/gri_mech/ (accessed in Oct, 2014).


539

IZVOD

RAZVOJ I VALIDACIJA REDUKOVANOG MEHANIZMA METANA KORIŠĆENјEM NOVOG INTEGRALNOG ALGORITMA Zuozhu Wu, Xinqi Qiao, Zhen Huang

Key Laboratory of Power Machinery and Engineering, Ministry of Education, Shanghai Jiao Tong University, Shanghai, China

(Naučni rad)

Predložen je novi algoritam baziran na kompjuterskoj singularnoj perturbaciji(KSP) u cilјu konstruisanja globalnog redukovanog mehanizma. Algoritam pred-stavlјa koncentracije vrsta, iznose neto proizvodnje vrsta i iznose oslobađanjatoplote kao integralne faktore merenja radi integrisanja KSP pokazatelјa, uklјu-čujući radikalne pokazatelјe i pokazatelјe brze reakcije, putem kompjuterskogdomena. Razvijen je softverski paket baziran na algoritmu u cilјu efikasnijeg pro-cesa redukcije. Input za algoritam uklјučuje: i) detalјni mahenizam, ii) numeričkorešenje problema za specifični set radnih uslova i iii) broj kvazi-stabilnih stanjavrsta (KSS). Predloženi algoritam je apliciran na redukciju GRI3.0 uklјučujući 53vrste i 325 reakcija koje su vodile ka razvoju 15 vrsta sa redukovanim meha-nizmom i 10 spojenih koraka. Tada je redukovani mehanizam potvrđen u jedno-dimenzionalnom, produženom, laminarnom plamenu shodno širokom spektruekvivalentnih odnosa i odličnih spojeva između pažlјivo izračunatih rezultata iredukovanih mehanizama koji su posmatrani.

Ključne reči: Mehanizam redukcije • KSP• integralni KSP algoritam • GRI3.0

541

Oxidation of lignin–carbohydrate complex from bamboo with hydrogen peroxide catalyzed by Co(salen)

Xue-Fei Zhou

Engineering Research Center for Biomass Resource Utilization and Modification of Sichuan Province of Southwest University of Science and Technology, Mianyang, China Key Laboratory of Eco-Environment-Related Polymer Materials of Education Ministry and Key Laboratory of Polymer Materials of Gansu Province of Northwest Normal University, Lanzhou, China Kunming University of Science and Technology, Kunming, China Hubei Key Laboratory of Pollutant Analysis and Reuse Technique of Hubei Normal University, Huangshi, China State Key Laboratory of Molecular Engineering of Polymers of Fudan University, Shanghai, China State Key Laboratory of New Ceramics and Fine Processing of Tsinghua University, Beijing, China

Abstract The reactivity of salen complexes toward hydrogen peroxide has been long recognized.Co(salen) was tested as catalyst for the aqueous oxidation of a refractory lignin–carbo-hydrate complex (LCC) isolated from sweet bamboo (Dendrocalamus hamiltonii) in the presence of hydrogen peroxide as oxidant. Co(salen) catalyzed the reaction of hydrogenperoxide with LCC. From the spectra analyses, lignin units in LCC were undergoing ring-opening, side chain oxidation, demethoxylation, β-O-4 cleavage with Co(salen) catalytic oxidation. The degradation was also observed in the carbohydrate of LCC. The investi-gation on the refractory LCC degradation catalyzed by Co(salen) may be an importantaspect for environmentally-oriented biomimetic bleaching in pulp and paper industry.

Keywords: LCC; Co(salen); catalytic oxidation; sweet bamboo (Dendrocalamus hamiltonii).

SCIENTIFIC PAPER

UDC 66.094.3.097:582.542

Hem. Ind. 68 (5) 541–546 (2014)

doi: 10.2298/HEMIND1308821080Z


Lignin is a complex chemical compound most com-monly derived from wood, and an integral part of the secondary cell walls of plants and some algae [1,2]. Although lignin is necessary for trees, providing mecha-nical support to bind fibers together, and preventing the cell wall from the attack of microorganisms, lignin is removed for production of most papers. Degradation of lignin (delignification) hence constitutes an important step in paper manufacture [3].

Data from numerous studies indicate that lignins are associated with hemicelluloses forming, in some cases, lignin–carbohydrate complexes (LCC) that are resistant to hydrolysis even under pulping conditions [4–8]. Consequently the search for a selective and envi-ronmentally friendly delignification technology which can selectively oxidize lignin functionalities without harming the cellulose fibers is a current challenge for the paper industry worldwide. Novel processing methods are required to extend lignin selective depolymer-ization. A very attractive attempt is to catalytically enhance the delignification with aid of oxygen and/or hydrogen peroxide in the bleaching cycle, this certainly involves the development of metal complexes. Salen Correspondence: Kunming University of Science and Technology, P.O. Box A302-12, Building No.5, XinyingYuan, NO.50, Huancheng East Road, 650051 Kunming, Yunnan Province, P.R. China. E-mails: [email protected] Paper received: 21 August, 2013 Paper accepted: 31 October, 2013

complexes are an important class of organometallic compounds, which have been used to catalyze a wide variety of reactions with oxidants like oxygen and hyd-rogen peroxide. The synthesis of water or organic sol-vent-soluble salen complexes is simple, easy and low cost. For these reasons, they can be used as catalysts in the field of lignin and wood chemistry. In particular it was demonstrated that they were able to oxidize in high yields lignin model compounds [9]. Bozell et al. have reported that Co(salen) along with dioxygen as the oxidant constitutes the most promising biomimetic degradation of lignin [10]. Similarly, Canevali et al. have studied the oxidation mechanism of apocynol catalyzed by Co(salen) in the homogeneous phase using either chloroform or pyridine as solvent. About 90% conver-sion was obtained after 48 h of reaction [11]. Rajago-palan et al. have studied the catalytic efficacy of Co(salen)(AL) in O2 oxidation reactions in CO2-expanded solvent media and elucidated the axial ligand depen-dence and substrate selectivity [12].

To the best of our knowledge, studies dealing with the use of Co(salen) complex for the oxidation of LCC have not previously been reported. In this work, we report the oxidative degradation of LCC isolated from sweet bamboo (Dendrocalamus hamiltonii) using Co(salen) complex as the catalyst. Several spectro-scopic techniques were used to study this oxidation reaction under experimental condition to provide fun-

X.-F. ZHOU: OXIDATION OF LIGNIN–CARBOHYDRATE COMPLEX CATALYZED BY Co(SALEN) Hem. ind. 68 (5) 541–546 (2014)

542

damental basis with the goal of improving pulping and bleaching operations.

EXPERIMENTAL

Materials

All chemicals are commercially available and were used as received without further purification, unless otherwise noted. Co(salen) was laboratory synthesized, and ethanol, benzene, 1,4-dioxane, acetic acid, pyri-dine, sodium hydroxide, hydrogen peroxide, methylene dichloride, sodium sulphate were purchased from Sinopharm Chemical Reagent Co. Shanghai, China.

Co(salen) synthesis

Synthesis of salen [N,N-bis(salicylaldehyde)ethyl-enediimine acid] and Co(salen) {[N,N-bis(salicylalde-hyde)ethylenediimino]cobalt(II)} was carried out fol-lowing the procedure of published literature (Scheme 1) [13].

Isolation of LCC

The bamboo flour (100 mesh) was pre-extracted with ethanol–benzene (1∶2), vacuum-dried with phos-phorus pentoxide, and then ground using a variable planetary ball-mill for 72 h. The very fine bamboo flour was then extracted with a solvent of dioxane-water (96∶4) at an ambient temperature for 24×3 h. The residues were dissolved in a solvent of acetic acid– –water (1∶1) for 24×3 h and then worked up according to established procedures [14].

Catalytic experiment

The catalyst Co(salen) and axial base pyridine were added and stirred in flask prior to the other reagents. Sodium hydroxide and hydrogen peroxide were then added followed by LCC and at this point pH of the solution was recorded (pH 12.5). The reaction was let to proceed at 90 °C for 5 h and stopped by cooling to

ambient temperature, then filtered with sintered glass filter. The final residue was collected for FTIR and NMR analysis. The filtrate was extracted with methylene dichloride. The organic phase was separated and dried with sodium sulphate, and finally concentrated to 1 mL for reaction product analysis with gas chromatography-mass spectrometry (GC–MS).

FTIR

FTIR spectra of LCC and residual LCC obtained from catalytic experiments were made on a Nicolet 470 FTIR spectrometer in wavelength bands from 4000 to 400 cm–1, using 1 mg of freeze-dried samples and 300 mg of KBr. 1H- and 13C-NMR

The samples (LCC and residual LCC) were dissolved in DMSO-d6 and the spectra were recorded on Bruker DRX 500 apparatus in a 5-mm diameter tube.

GC–MS

The oxidation products dissolved in methylene dichloride were identified by GC–MS with an Agilent Technologies HP 6890/5973 system fitted with a fused silica column (HP-INNOWAX, 30 m×0.25 mm i.d., 0.25 μm film thickness). It was used as the carrier gas (1.0 mL min–1). The mass spectrometer was operated in the electron ionization mode (EI, 50 eV). Compound ident-ification was performed using GC retention times and by Mainlib database.

RESULTS AND DISCUSSION

1H, 13C-NMR and FTIR spectra of the LCC are compared in Figures 1–3.

The signals corresponding to different carbons referred to lignin and carbohydrate in LCC were observed in the spectra in Figures 1 and 2 [14,15]. In the meanwhile, from the increase in carbonyl signals (Figure 1: 13.0/9.5 ppm; Figure 2: 170.0 ppm) and the

NH2

NH2

2

CHO

OH OH HO

N N

2

O O

N N

Co

Co

ethylenediamine salicylaldehyde salenH2

+

+

Co(II)-salen

Scheme 1. Synthesis of Co(II)-salen.


543

Figure 1. 1H-NMR Spectra of untreated and treated LCC.

Figure 2. 13C-NMR Spectra of untreated and treated LCC.

Figure 3. FTIR Spectra of untreated and treated LCC.


544

decrease in benzylic proton (Figure 1: 2.62 ppm) of treated LCC, the lignin units were catalytically oxidized in LCC. The demethoxylation was observed in lignin of LCC according to the weakened signals at 3.80 ppm (Figure 1), 58.5 ppm (Figure 2) in the treated sample. The obvious decrease in signal at 6.7 ppm in Figure 1 implied that lignin of LCC was undergoing ring-opening with Co(salen) catalytic oxidation. On the other hand, it was seen from Figure 1 that the signal of phenolic group at 8.0 ppm was higher in case of treated LCC than untreated LCC. This can be attributed to the demethoxylation and the β-O-4 cleavage in lignin of LCC. This can be confirmed by the lower signal of methoxyl at 3.80 ppm (Figure 1)/58.5 ppm (Figure 2) and β-O-4 at 78.5 ppm (Figure 2) in the treated sample rather than in the untreated sample. These agreed with information obtained from the Co(salen)-catalysed oxi-dation of synthetic lignin and LCC model compound, showing side chain oxidation, demethoxylation, aroma-tic ring opening, and β-O-4 cleavage [16,17].

The degradation was also observed in the carbo-hydrate of LCC. The signals arising from the carbohyd-rate of LCC in 1H- and 13C-NMR spectra, 1.26 ppm (Figure 1) and 100.0/92.0/73.0/70.0 ppm (Figure 2), were shown to be clearly much less intense in the treated sample.

Similar observation on the changes of LCC in the catalytic oxidation was obtained from the signals in FTIR spectra. The band patterns were shown in Figure 3. The bands at 890 and 1733 cm–1 are typical of glu-cans [18]. The spectra in Figure 3 showed bands characteristic of lignin [19]: 1600/1500 cm–1 (aromatic ring), 1225 cm–1 (phenolic OH), 2930/1030 cm–1 (meth-oxy group), and 1120 cm–1 (β-O-4). Band at 1500 cm–1 was useful in demonstrating the presence of lignin in LCC, since it does not overlap with bands from other natural polymers. The decreases in intensities of these bands in treated LCC were evident in the spectra, suggesting that carbohydrate and lignin in LCC has been significantly degraded in the catalytic reaction.

To obtain further information on the nature of LCC degradation, GC–MS analyses of soluble portion were conducted. From the mass spectrum data obtained by GC–MS analysis, LCC-derived products were identified. Figure 4 shows major products characteristic of degrad-ation of lignin in LCC, including 4-hydroxybenzaldehyde (peak 1), 2-methoxy phenol (peak 2), 4-(1-hydroxypro-pyl)-1,2-benzenediol (peak 4), guaiacol (peak 6) and 1,2,3-trimethoxy-5-methylbenzene (peak 8). These compounds arose from p-hydroxyphenyl-(4-hydroxy benzaldehyde), guaiacyl (2-methoxyphenol, 4-(1-hydro-xypropyl)-1,2-benzenediol and guaiacol), and syringyl (1,2,3-trimethoxy-5-methylbenzene) structures, res-pectively. Some products from degradation of carbo-hydrate in LCC (peak 5) can be recognized in Figure 4.

The fragments derived from the LCC were also present (peak 7). The existence of these products suggested that the cleavages between ether, Cβ/Cγ and Cα/Cβ linkages of lignin took place, which was generally due to the well know metal-dioxygen adduct mechanism on the basis of common intermediate phenoxide radical by abstraction of hydrogen atoms [20, 21].

Figure 4. Chromatogram of products isolated from LCC oxi-dation with Co(salen). Peaks: 1 = 4-hydroxybenzaldehyde; 2 = 2-methoxyphenol; 3 = 4-methylcyclohexa-2,5-dien-1-one; 4 = 4-(1-hydroxypropyl)-1,2-benzenediol; 5 = polysaccharide fraction; 6 = guaiacol; 7 = LCC fragments; 8 = 1,2,3-trimethoxy-5-methylbenzene.

CONCLUSION

This study demonstrated that Co(salen) complex is able to work as oxidation catalyst for effective deg-radation of the refractory LCC macromolecule and thus potentially as biomimetic catalyst for the bleaching in pulping industry. Further investigations on LCC model compounds are underway to understand the possible interaction of the Co(salen) complex with the LCC.

Acknowledgements

This work was supported by the Open Project of Key Laboratory of Eco-Environment-Related Polymer Mat-erials of Education Ministry and Key Laboratory of Poly-mer Materials of Gansu Province of Northwest Normal University (KF-13-03), the National Natural Science Foundation of P. R. China (21166011), the Open Project of Engineering Research Center for Biomass Resource Utilization and Modification of Sichuan Province of Southwest University of Science and Technology (12zxsk01), the Open Project of Hubei Key Laboratory of Pollutant Analysis and Reuse Technique of Hubei Normal University (KL2013M10), the Open Project of State Key Laboratory of Molecular Engineering of Polymers of Fudan University (K2013-09) and the Open Project of State Key Laboratory of New Ceramics and Fine Processing of Tsinghua University (KF201402).


545

REFERENCES

[1] S.E. Lebo Jr., J.D. Gargulak, T.J. McNally, Lignin Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley & Sons, New York, 2001.

[2] P.T. Martone, J.M. Estevez, F. Lu, K. Ruel, M.W. Denny, C. Somerville, J. Ralph, Discovery of lignin in seaweed reveals convergent evolution of cell-wall architecture, Current. Biol. 19 (2009) 169–175.

[3] C.V. Stevens, Renewable Bioresources: Scope and Modi-fication for Non Food Application, John Wiley and Sons, England, 2004.

[4] R.M. Rowell, R. Pettersen, J.S. Han, J.S. Rowell, M.A. Tshabalala, Handbook of Wood Chemistry and Wood Composites: Cell Wall Chemistry, CRC Press, Boca Raton, FL, 2005.

[5] M. Lawoko, G. Henriksson, G. Gellerstedt, New method for quantitative preparation of lignin-carbohydrate com-plex from unbleached softwood kraft pulp: lignin-poly-saccharide networks, Holzforschung 57 (2003) 69–74.

[6] T. Nanbu, J. Shimada, M. Kobayashi, K. Hirano, T. Koh, M. Machino, H. Ohno, M. Yamamoto, H. Sakagami, Anti-UV activity of lignin-carbohydrate complex and related compounds, In Vivo 27 (2013) 133–139.

[7] X. Du, G. Gellerstedt, J. Li, Universal fractionation of lignin–carbohydrate complexes (LCCs) from lignocel-lulosic biomass: an example using spruce wood, Plant J. 74 (2013) 328–338.

[8] M. Lawoko, G. Henriksson, G. Gellerstedt, Structural differences between the lignin-carbohydrate complexes present in wood and in chemical pulps, Biomacro-molecules 6 (2005) 3467–3473.

[9] C. Crestini, M. Crucianelli, M. Orlandi, R. Saladino, Oxidative strategies in lignin chemistry: A new environ-mental friendly approach for the functionalisation of lignin and lignocellulosic fibers, Catal. Today 156 (2010) 8–22.

[10] J.J. Bozell, B.R. Hames, D.R. Dimmel, Cobalt-schiff base complex catalyzed oxidation of para-substituted phen-olics. Preparation of benoquinones, J. Org. Chem. 60 (1995) 2398–2404.

[11] C. Canevali, M. Orlandi, L. Pardi, B. Rindone, R. Scotti, J. Sipila, F. Morazzoni, Oxidative degradation of mono-meric and dimeric phenylpropanoids: reactivity and

mechanistic investigation, J. Chem. Soc. Dalton Trans. 15 (2002) 3007–3014.

[12] B. Rajagopalan, H. Cai, D.H. Busch, B. Subramaniam, The catalytic efficacy of Co(salen)(AL) in O2 oxidation reac-tions in CO2-expanded solvent media: Axial ligand dependence and substrate selectivity, Catal. Lett. 123 (2008) 46–50.

[13] Z.C. Liu, F. Liu, Y. Lu, M.X. Xie, Y.Q. Zhang, Studies on characters and synthesis of metal-salen complexes, J. Leshan Teachers College 17 (2002) 30–33 (in Chinese).

[14] K.G. Johnson, R.P. Overend, Lignin-carbohydrate com-plexes from populus deltoids.Ⅰ. Purification and char-acterization, Holzforschung 45 (1991) 469–475.

[15] J.-L. Wen, S.-L. Sun, B.-L. Xue, R.-C. Sun, Recent adv-ances in characterization of lignin polymer by solution-state nuclear magnetic resonance (NMR) methodology, Mater. 6 (2013) 359–391.

[16] X.-F. Zhou, J.-X. Qin, S.-R. Wang, Oxidation of a lignin model compound of benzyl-ether type linkage in water with H2O2 under an oxygen atmosphere catalyzed by Co(Salen), Drewno 54 (2011) 15–25.

[17] X.-F. Zhou, J. Liu, Co(salen)-catalysed oxidation of syn-thetic lignin-like polymer: Co(salen) effects, Hem. Ind. 66 (2012) 685–692.

[18] R. Singh, S. Singh, K.D. Trimukhe, K.V. Pandare, K.B. Bastawade, D.V. Gokhale, A.J. Varma, Lignin-carbohyd-rate complexes from sugarcane bagasse: Preparation, purification, and characterization, Carbohydr. Polym. 62 (2005) 57–66.

[19] H.L. Hergert, Infrared Spectra, in Lignins. Occurrence, Formation, Structure and Reactions, K. V. Sarkanen, C. H. Ludwig, Eds., John Wiley & Sons, New York, 1971, pp. 267–297.

[20] J.M. Mayer, E.A. Mader, J.P. Roth, J.R. Bryant, T. Matsuo, A. Dehestani, B.C. Bales, E.J. Watson, T. Osako, K. Valliant-Saunders, W.H. Lam, D.A. Hrovat, W.T. Bor-den, E.R. Davidson, Stoichiometric oxidations of σ-bonds: Radical and possible non-radical pathways, J. Mol. Catal., A 251 (2006) 24–33.

[21] A. Zombeck, R.S. Drago, B.B. Corden, J.H. Gaul, Acti-vation of molecular oxygen. Mechanistic studies of the oxidation of hindered phenols with cobalt-dioxygen complexes. J. Am. Chem. Soc. 103 (1981) 7580–7585.


546

IZVOD

OKSIDACIJA KOMPLEKSA LIGNIN–UGLJENI HIDRAT IZ BAMBUSA POMOĆU VODONIK-PEROKSIDA KATALIZOVANA Co(SALENOM)

Xue-Fei Zhou

Engineering Research Center for Biomass Resource Utilization and Modification of Sichuan Province of Southwest University of Science and Technology, Mianyang, China Key Laboratory of Eco-Environment-Related Polymer Materials of Education Ministry and Key Laboratory of Polymer Materials of Gansu Province of Northwest Normal University, Lanzhou, China Kunming University of Science and Technology, Kunming, China Hubei Key Laboratory of Pollutant Analysis and Reuse Technique of Hubei Normal University, Huangshi, China State Key Laboratory of Molecular Engineering of Polymers of Fudan University, Shanghai, China State Key Laboratory of New Ceramics and Fine Processing of Tsinghua University, Beijing, China

(Naučni rad)

Reaktivnost salen kompleksa u reakciji sa vodonik-peroksidom je dobro poz-nata. Co(salen) je korišćen kao katalizator u oksidaciji kompleksa lignin–ugljeni hidrat (LCC) izolovanog iz slatkog bambusa (Dendrocalamus hamiltonii) u pri-sustvu vodonik-peroksida kao oksidacionog sredstva. Spektroskopskom analizom(FTIR, NMR i GC–MS) utvrđeno je da lignin u LCC-u podleže otvaranju prstena, reakcijama bočnog niza, demetoksilovanju, β-O-4 razgradnji pri oksidaciji katali-zovanoj Co(salenom). Degradacija je uočena i u ugljeno hidratnom delu LCC-a. Rezultati ispitivanja LCC razgradnje pomoću Co(salena) mogu biti od značaja zaekološki prihvatljivo biomimetičko beljenje u industriji pulpe i papira.

Ključne reči: LCC • Co(salen) • Katalitička oksidacija • Slatki bambus

547

Analiza simulatora tehnoloških procesa u funkciji projektovanja: Studija slučaja separacije prirodnog gasa

Dimitrije Ž. Stevanović1, Mića B. Jovanović2, Marina A. Mihajlović1, Jovan M. Jovanović1, Željko B. Grbavčić2 1 Inovacioni centar Tehnološko–metalurškog fakulteta u Beogradu, Univerzitet u Beogradu, Beograd, Srbija 2 Tehnološko–metalurški fakultet u Beogradu, Univerzitet u Beogradu, Beograd, Srbija

Izvod Softveri za modelovanje i simulaciju tehnoloških procesa u poslednjih nekoliko decenijaimaju značajnu ulogu u razvoju procesne industrije. Korišćenje procesnih simulatora,unašoj zemlji, za potrebe projektovanja tehnoloških procesa nije široko rasprostranjeno, aliteratura je nedovoljna. U ovom radu je dat pregled savremenih procesnih simulatora iprikazane su njihove mogućnosti za projektovanja procesa u zavisnosti od grane hemijskeindustrije. Prikazna je studija slučaja tehnološkog procesa u više procesnih simulatora kojaispituje rezultate, pouzdanost i upotrebljivost ovih programa prilikom rešavanja konkretnih projektantskih zadatka na primeru postrojenja za separaciju prirodnog gasa. Date su upo-redne karakteristike rezulata simuliranja separacije prirodnog gasa za potrebe projekto-vanja procesa, na osnovu simulacija u softverima različitih složenosti i kvaliteta. Procesni simulatori su pokazali svoju upotrebnu vrednost kao značajan oslonac projektantima, iakorezultati koje su pokazali međusobno imaju značajna odstupanja.

Ključne reči: procesni simulatori, projektovanje u procesnoj industriji, simulacija tretmana prirodnog gasa.

STRUČNI RAD

UDK 519.86/.87:66:662.767

Hem. Ind. 68 (5) 547–558 (2014)

doi: 10.2298/HEMIND130424002S

Dostupno na Internetu sa adrese časopisa: http://www.ache.org.rs/HI/

Procesni simulatori (u daljem tekstu PS) spadaju u grupu inženjerskih softvera koji daju podršku pri pro-jektovanju novih procesa i proizvoda, bitnih izmena postojećih tehnoloških procesa, proceni alternativa tehničkih rešenja složenih procesa i identifikacije uzro-ka problema u radu ili optimizaciji postojećih postro-jenja, odnosno pri proceni uticaja na životnu sredinu, itd.

Pojava dinamičkih simulacija omogućila je predvi-đanje i kontrolu realnih procesa u realnom vremenu, a poboljšane mogućnosti softvera našle su primenu i pri-likom projektovanja procesa. Kompletni softverski pa-ket koji u užem smislu odgovara na zahteve projekta-nata i hemijskih inženjera prilikom projektovanja naziva se procesnim simulatorom [1–3].

U svetu je dostupno nekoliko stotina komercijalnih i besplatnih procesnih simulatora, uže ili šire specijali-zacije po različitim oblastima hemijske industrije, a svi su definisani kao CAPE (eng. Computer Aided Process Engineering) softveri.

Može se reći da je u Srbiji nedovoljno zastupljena njihova primena, a softveri se u projektantskim organi-zacijama, razvojnim institutima i industriji retko koriste, dok je odgovarajuća literatura slabo prisutna.

Prepiska: D.Ž. Stevanović, Inovacioni centar Tehnološko–metalurškog fakulteta u Beogradu, Karnegijeva 4, 11000 Beograd, Srbija. E-pošta: [email protected] Rad primljen: 24. april, 2013 Rad prihvaćen: 30. december, 2013

Cilj ovog rada je da se prikažu mogućnosti savre-menih procesnih simulatora, međusobno uporede nji-hove karakteristike i prikažu brojni aspekti njihove pri-mene prilikom projektovanja realnih industrijskih pro-cesa. Rezultati analize bi trebalo da pomognu projek-tantima prilikom izbora odgovarajućeg softvera, u za-visnosti od karakteristika tehnološkog procesa i opre-me. Poseban deo analize, zasnovan na studiji slučaja realnog postrojenja za separaciju prirodnog gasa, pore-di simulacije u različitim aplikacijama, radi ispitivanja njihovih mogućnosti prilikom projektovanja i saglas-nosti rezultata.

PROCESNI SIMULATORI I PROJEKTOVANJE PROCESA

Procesni simulatori se zasnivaju na proračunima masenih i energetskih bilansa, simuliranjem jediničnih tehnoloških operacija, ili kinetike procesa sa hemijskom reakcijom, u odgovarajućim uređajima. Osnove za napred opisane aktivnosti su sledeći elementi softvera: a) baze podataka sa fizičko-hemijskim karakteristikama hemijskih elemenata i jedinjenja, kao i industrijskih smeša koje se koriste u tehnološkim procesima, b) baze jediničnih modela i operacija za pojedinačnu tehno-lošku opremu, c) baze podataka sa termodinamičkim modelima i/ili modelima strujanja fluida i d) baze nu-meričkih metoda.

Ovi programski paketi su, po pravilu, sekvencio-nalno modularani – objektno orijentisani programi, uglavnom zasnovani na jednačinama fenomena pre-

D.Ž. STEVANOVIĆ i sar.: SIMULATOR PROCESA SEPARACIJE PRIRODNOG GASA Hem. ind. 68 (5) 547–558 (2014)

548

nosa. Spadaju u grupu CAD (eng. Computer Aided Design) računarskih alata.

Savremeni simulatori uzimaju u obzir brojne faktore koji utiču na jedan proces poput: detaljne geometrije procesnih jedinica, napredne termodinamike fluida, složenih međuzavisnih operacija, interakcija između komponenti sistema, najbolje dostupne baze jedinje-nja, te izračunavaju energetske gubitke i, u nekim slučajevima, predviđaju ponašanja nedovoljno defini-sanih komponenti ili operacija.

Projektovanje predstavlja institucionalni okvir kroz koji se, ili uz čiju pomoć se odvijaju sve razvojne aktivnosti u procesnoj industriji. Posebno izdvojeno projektovanje procesa može se definisati kao kreativno i stručno uobličavanje tehnoloških procesa i njegovih elemenata u cilju zadovoljena industrijskih potreba. Glavni cilj projektovanja tehnoloških procesa je uve-ćanje željenih i smanjivanje neželjenih reakcija i pro-cesa, dobijanje kvalitetnog proizvoda na siguran i lako ponovljiv način u skladu sa zahtevima investitora, po konkurentnoj ceni i uz poštovanje svih zakonskih ogra-ničenja [4–7].

Upotreba procesnih simulatora počinje već od naj-ranije faze projektovanja, a često i prilikom razvoja samog procesa. Ovim putem omogućeno je rano otkri-vanje slabosti projekta u okviru zadatih ograničenja poput: tehnologije, investicionih troškova, dostupnog prostora, sigurnosti, uticaja na životnu sredinu, proiz-vodnje otpadnih materija, operativnih troškova i troš-kova održavanja. Korišćenje simulacija daje smernice u izradi projekta, kako sledi:

1. Procene opšte izvodljivosti procesa kao celine, kao i procenu njegove fleksibilnosti u radu, pre nego što se pristupi detaljnom tehnološkom projektovanju.

2. Generiše veći broj mogućih tehnoloških alterna-tiva i/ili rešenja, kao valjanu osnovu za dalji izbor pro-jektnata/investitora.

3. Mogućnost preliminarnog definisanja ciljeva procesa pre detaljnog dimenzionisanja opreme.

Još jedna prednost procesnih simulatora ogleda se u njihovoj mogućnosti brzog uočavanja problema u ranim fazama razvoja procesa i davanja odgovora na pitanje „šta ako?“. Na slici 1, prikazani su mogući doprinosi softvera za simulaciju u različitim fazama tehnološkog projektovanja ili prilikom operativnog rada postrojenja [8].

Kako su zaštita životne sredine i sigurnost procesa postali značajni deo prilikom projektovanja procesa bitno je ukazati da pojedini PS projektantima istovre-meno pružaju pomoć pri proceni emisija zagađujućih materija, identifikuju zone opasnosti, a pomoću njih moguće je izvesti poboljšanje energetske efikasnosti ili optimizacije rada tehnoloških postrojenja. Danas su simulacije procesa postale osnova tehnološkog projek-tovanja. Primenom simulatora procesa može se pred-videti ukupno ponašanje, sastav i osobine svih mate-rijalnih tokova procesa, kao i dimenzije, karakteristike svih uređaja i operativni uslovi. Sve navedeno utiče na skraćenje vremena predviđenog za projektovanje i doz-voljava projektantu da brzo testira različite konfigura-cije procesa, pogona, i/ili sistema.

Slika 1. Doprinosi softvera za simulaciju u različitim fazama tehnološkog projektovanja. Figure 1. Software application in different stages of chemical process design.


549

PRINCIPI RADA PROCESNIH SIMULATORA

Matematički modeli u procesnim simulatorima mogu se razlikovati prema kompleksnosti, tako da se može simulirati samo odabrani deo procesne opreme, odabrana operacija u uređaju, odnosno kompletnan tehnološki proces. Projektovanje i analiza procesa i postrojenja često se vrši pomoću makroskopskih, sta-cionarnih modela. Korišćenje matematičkih modela uređaja zahteva unos neophodnih parametara, a izbor i kvalitet modela direktno utiču kako na simulaciju, tako i na rezultate koji se dalje koriste u projektovanju.

Dobri modeli koriste se, u velikoj meri, za donošenje odluka o izvodljivosti operacija prilikom projektovanja, a istovremeno mogu da daju neophodne podatke za rešavanje pojedinih operativnih problema. Razvoj kva-litetnih i pouzdanih matematičkih modela je suština unapređenja modelovanja i simulacije za potrebe pro-jektovanja procesa.

Glavne tehnike dostupne za simulaciju modela pro-cesa svrstane su u tri pristupa: sekvencijalno-modularni – najčešće u upotrebi, najsličniji je klasičnom ručno-vezanom proračunu na osnovu ulaznih prarametara svake tehnološke operacije ili uređaja tehnološkog procesa, zatim simulaciono-modularni pristup sličan prethodnom, sa tim što je više orijentisan na jednačine u cilju linearizacije sistema i na kraju jednačinski-ori-jentisani koji se zasniva se na predstavljanju procesa nizom nelinearnih i drugih jednačina modela, inter-konekcionih jednačina, projektnih uslova, fizičkih i ter-modinamičkih korelacija, koje se rešavaju simultano za sve nepoznate.

U dosadašnjoj praksi razvoja procesa putem ekspe-rimenta, sam postupak može dovesti do optimalnog rešenja suviše sporo za savremeni način proizvodnje i poslovanja. U slučaju primene softvera poklapanja sa eksperimentalnim podacima u referentnim slučajevima kreću se u rasponu od 92–99% u zavisnosti od procesa [9–11], a sa razvojem u ovoj oblasti mogu se očekivati još bolji rezultati.

Vrhunac upotrebe savremeni PS dostižu poveziva-njem sa realnim procesima tokom proizvodnje. Neki od njih imaju mogućnost da prate ključne parametre pro-cesa velikog broja ulaznih promenljivih i automatski, paralelno, kreiraju i dalje za simulaciju koriste nove poboljšane modele. Time se unapred detektuju neže-ljeni efekti i asistira u sigurnom i ekonomičnom vođe-nju procesa. Razvojem preciznih modela na osnovu direktnih podataka iz realnih procesa omogućene su pouzdane prediktivne simulacije proizvodnje.

PREGLED SAVREMENIH PROCESNIH SIMULATORA

Većina savremenih procesnih simulatora danas radi na sličnim principima, a svaki ima određene osobitosti, prednosti ili ograničenja u odnosu na ostale. Generalno gledano svi PS se pored pouzdanosti, mogu uporediti i po više kategorija poput: broja termodinamičkih pake-ta, veličine i pouzdanosti internih baza komponenata, broja modela uređaja, tržišne cene, kvalitetu interfejsa, specijalizaciji za određene grane procesne industrije ili operacije, stepenu poklapanja sa realnim procesom i sl. Svi procesni simulatori mogu se grubo podeliti prema više karakteristika, a neke od podela date su u tabeli 1.

Tabela 1. Tipovi podela procesnih simulatora Table 1. Process simulator types

Redni br. Tip podele Primer 1. Primena: 1. Specijalizovani za pojedine industrijske grane (naftna, petrohemijska, hemijska,

farmaceutska). 2. Specijalizovani za pojedina industrijska postrojenja (reaktori, kolone, razmenjivači toplote, bioreaktori i separtori).

2. Dinamika sistema: 1. Stacionarni, 2. Dinamički, 3. Kombinovani.

3. Operativni sistem: 1. Windows (Aspen Plus), 2. DOS (Kemisimp), 3. iOS (Alph).

4. Baza podataka: 1. Interna (AspenPlus), 2. Eksterna (COCO).

5. Povezanost sa procesom u realnom vremenu:

1. Offline (SuperPro Designer), 2. Online (AspenPlus).

6. Interfejs: 1. Objektno orjentisani, 2. Grafički, 3. Tekstualni, 4. Touch pad.


550

Pored integrisane baze modela, operacija, termodi-namičkih i numeričkih metoda koji se lako uklapaju u modelovanu procesnu šemu, PS koriste pojednostav-ljeni i intuitivni grafički interfejs za komunikaciju sa korisnikom. Uglavnom se u naprednim PS mogu izabrati različiti režimi rada u zavisnosti od namene (računski, projektantski, simulacioni i sl.), što dodatno olakšava primenu softvera. Na slici 2 prikazan je izbor i poveza-nost softverskih opcija radi formiranju modela procesa.

Pored procesnih simulatora u užem smislu grupi-sanih po nameni, za projektovanje procesa mogu se koristiti i rezultati iz dopunskih softvera iz grupe za numeričke analize ili zaštitu životne sredine, kao što je prikazano na slici 3.

Iako se svake godine broj PS za tehnološkog pro-jektovanja uvećava u vodeće spadaju: Aspen Plus, AspenHysys, PRO/II, SuperPro Designer, ChemCad, Design II, VGPro, Unisim, ProSim i drugi softveri nave-deni u tabeli 2.

U slučaju da simulator omogućava direktno prikup-ljanje realnih operativnih podataka iz procesa, gde se ti rezultati koristite za odlučivanje ili predviđanje prome-na u radu, te se kontrolni parametari optimizuju na osnovu ovih rezultata – takve simulacije se nazivaju online. Sa druge strane offline procesne simulacije ne uključuju razmenu podataka sa realnim sistemom, a mogu se koristiti za projektovanje, rešavanje problema i optimizaciju ili modifikaciju postrojenja i sl. Pored klasičnih procesnih simulatora, prilikom projektovanja procesa mogu se koristiti i numerički softveri opštije namene poput softvera Mathlab, Comsol, Polymath ili EES [12–16].

Poslednjih godina dostupne su i napredne verzije PS koje nastaju integrisanjem dva ili više srodnih progra-ma, čime nastaje novi značajno poboljšani softverski proizvod. Tako je, na primer, 2004. godine udruživa-njem kompanije AspenTech i kompanije Fluent Inc.(CFD softver Fluid) nastao novi složeniji softver Aspen HYSYS, koji pored standardnih podataka o materijalnim i ener-getskim karakteristikama procesa, daje detaljne po-datke o dinamici fluida, raspodeli čestica, obliku struj-nica i sl. Primarni cilj integracije je bio da se ukombi-nuju potencijali modelovanja opreme sa širokim mogu-ćnostima modelovanja procesa kako bi se omogućila

brža procena novih inovativnih postrojenja koja uklju-čuju i jedinstvenu opremu (npr. gorivne ćelije, gasifi-katore i složene reaktore). Novi interfejs omogućio je sveobuhvatnu procenu stanja opreme i fluida, tako što su detalji modela mehanike fluida priključeni u zajed-nički procesni dijagram. Ova uspešna integracija, dve jake kompanije u oblasti procesne simulacije, značajno je unapredila softverske mogućnosti pri projektovanju procesa i doprinela daljem razvoju savremenih proces-nih simulatora.

Slika 3. Grupe inženjerskih softvera upotrebljive prilikom tehnološkog projektovanja. Figure 3. Engineering software groups used for chemical process design.

Dinamička simulacija se, pored primene u industriji, može smatrati korisnim nastavnim alatom za širok spektar kurseva iz oblasti hemijskog inženjerstva. Pored ostalog simulacioni softveri omogućavaju demons-traciju i testiranje drugačije vrsta procesa i opreme, upravljačkih algoritama i sigurnosnih procedura pro-cesa. Studije dinamičke simulacije vrše analizu sigur-nosti i bolje fokusiranje ka zahtevnim sigurnosnim stan-dardima (HAZOP/HAZID i sl.), poboljšavaju projekat procesa i daju mogućnost izrade plana upravljanja rizicima [17].

Slika 2. Koraci ka formiranju modela procesa. Figure 2. Creating process model procedure.


551

Tabela 2. Pregled karakterisitika procesnih simulatora i oblasti primene Table 2. Process simulator characteristics and application area

Naziv softvera Karakteristike Oblast primene AspenPlus Najkompletniji komercijalni softver, sa najboljim karakteris-

tikama, velikim brojem gotovih modela, najvećom i najpot-punijom bazom komponeti, koje uključuju i izomere jedi-njenja, elektrolite i sl. Sadrži napredne termodinamičke i fluidne pakete, sa mogućnostima procene ponašanja kom-ponenti. Softver ima mogućnost za dodatna poboljšavanjauslova reakcije, reaktanata i delova opreme. Online opcijeprocesa. Predviđa performanse procesa.Tržišni lider, visoka cena.

Široka oblast primene u različitim sferamarafinerijske, petrohemijske i ostale hemijskeindustrije. Često se koristi za simulacije prili-kom proizvodnje nafte i prirodnog gasa, ter-modinamičkih sistema, za modelovanje goriv-nih ćelija, u proizvodnji vodonika, proizvodnjipare u termoelektranama,za ekonomske pro-cene, optimizacije, usklađivanje proizvodnje izaliha.

Hysys Pouzdan softver, spada u tržišne lidere. Sada je deo Aspen paketa. Simulira u oba režima. Poseduje detaljno razrađene modele destilacionih kolona i dobru bazu termodinamičkihpaketa. Za razliku od ostalih softvera HYSYS po svakoj izmeni simulacije automatski daje nove rezultate.

Dinamika fluida u sistemima, industrija nafte igasa, separacija gasova, rafinerijska prerada islično.

SuperPro Designer Preko 140 gotovih modela i operacija u uređajima, velika baza komponenti. Najbolje rezultate daje za šaržne procese.Proračun karakteristika materijalnih tokova nije detaljan.

Jedan od retkih softvera koji simulira i hemij-ske reaktore. Pouzdan u oblasti biotehnolo-gije, prehrambene industrije, farmacije, zaš-tite životne sredine.

ChemCad Nova poboljšana verzija ovog softvera nudi mogućnosti iostalih savremenih procesnih simulatora. Pored nove pro-širene baze komponenti ovaj softver sada sadrži i savre-meno radno okruženje i veći izbor gotovih modela uređaja.

ChemCad se uspešno koristi za projektovanje isimulaciju velikog broja procesa u oblasti he-mijske industrije od kojih se posebno izdvaja:rafinerisjka proizvodnja i proizvodnja finihhemikalija.

ProSim Komercijalni simulator stacionarnih režima zasnovan nasekvencijalnom proračunu. Integrisan u Autocad koji jepotreban za pokretanje Prosim-a.

Naftna industrija, reaktorski procesi, trofazniprocesi, proizvodnja uglja i gasa, prehrambenaindustrija, zaštita ŽS.

Pro II Radi u stacionarnom stanju. Ovaj komercijalni softver ima veće tržišno učešće u oblastima projektovanja za petrohe-mijsku, naftnu, gasnu i ostalu hemijsku industriju. Od mogućnosti razdvajanje nafte i gasa, pa do reaktivne desti-lacije. PRO/II kombinuje obimne podatke velikog brojahemijskih komponenti iz svoje baze i termodinamičke me-tode za procenu osobina komponenti kao i napredne pri-lagodljive jedinične tehnoloških operacija.

Procesni inženjeri koriste ovaj softver za obav-ljanje obimnih proračuna bilansa energije pot-rebnih za modelovanje većine stabilnih stanjaprocesa u preradi nafte, prirodnog gasa iindustriji polimera.

Design II Nova verzija softvera 11.0 donosi mogućnost simuliranjadinamičkih modela. Jedini poseduje pored klasičnog i teks-tualno bazirani interfejs. Zasniva se na fortranskoj osnovi te se pripremljeni modeli lako mogu prepraviti u posebnomeditoru, što zahteva znanje programskog jezika Fortran.Njegova baza poseduje tipične karakteristike čak 38 vrstasirove nafte.

Pre svega softver je namenjen korišćenju unaftnoj i gasnoj industriji, ima veći izbor mo-dela uređaja za separaciju i rektifikaciju, apogodan je i za određivanje termodinamiketokova.

ProTreat

Savremeni komercijalni simulator, uže specijalizovan za po-jedine procese u oblasti hemijske industrije.

Simuliranje procesa izdvajanja H2S, CO2, mer-kaptana i drugih komponenti iz različitih ga-sova niskog i visokog pritiska absorpcijom utermalno regenerabilnom vodenom rastvorukoji sadrži jedan ili više amina.

K-Spice

Simulator dinamičkih procesa firme Kongsberg. Softvernovije generacije koji integriše njihova tri postojeća soft-vera.

Naftna i gasna industrija.

Coco Simul. Koristi bazu CAPE Open i predstavlja softver nešto slabijih mogućnosti i užeg izbora opreme, ali je pogodan za simu-liranje pojedinih separacionih tehnika. Pogodan za prelimi-narne procene pri projektovanju osnovnih procesnih teh-nika.

Najčešće se koristi za simulacije destilacionihkolona, uređaja za separaciju tečnosti i gasovaili razdvajanje komponenti smeše.


552

Tabela 2. Nastavak Table 2. Continued

Naziv softvera Karakteristike Oblast primene VMGSim Neki od programera softvera HYSYS, nakon prodaje licence

AspenuTech-u, nastavili su razvoj novog softvera u okvirukompanije VMG. Softver je još u razvojnoj fazi.

Rafinerijska i petrohemijska industrija.

Alph Radi na operativnom sistemu iOS, pogodan za korišćenje na tablet računarima. Ima mogućnost simulacije većeg brojajedinica i procesa u okviru hemijske industrije. Sadrži manjibroj modela.

Savremen softver namenjen korišćenju namobilnim hardverima poput iPad-a. Omogu-ćava simulacije direktno na pogonu.

Kemsimp Radi na operativnom sistemu MS DOS. Besplatan je. Skromna primena za statičke simulacije i po-dešavanja procesa.

DWSIM Ovaj softver je „open source“ pisan je u Visual Basicu.Kompatibilan je sa CAPE-open bazom, ima moderan grafičkiinterfejs i napedne termodinamičke pakete. Ima podrška za hemijske reakcije i reaktore. Složeniji za upotrebu.

Proračuni VLE/VLLE jednačine stanja, koefici-jenta aktivnosti i Chao-Seader modele, rigo-rozne destilacione/apsorpcione kolone. Karak-terizacija naftnih frakcija.

Toxchem Trenutno je aktuelna četvrta verzija. Na tržištu oko 10 go-dina. Projektovanje postrojenja za tretman otpadnih voda.

Tretman otpadnih, industrijskih i komunalnihvoda i procena emisija. Proračun emisija iznavedenih sistema i određivanje zona opas-nosti.

Water9 Besplatan softver iz EPA-e koji može da se koristi za pro-cenu emisije u vazduh i vode iz procesa prerade indus-trijskih otpadnih voda, kao i iz ostalih postrojenja. Velikabaza sa lako isparljivim organskim supstancama.

Zaštita životne sredine. Procena uticaja naživotnu sredinu. Projektovanje postrojenja zatretman otpadnih voda.

Einstein Besplatan program za projektovanje i optimizaciju ener-getske mreže pogona, upotrebe toplotne energije. Projek-tovanje energetski efikasnog snabdevanja sistema.

Energetska efikasnosti optimizacija toplotnihtokova. Primenljiv i u malim privrednim sub-jektima.

EMSO Korisnik može da modeluje kompleksne dinamičke ili sta-cionarne procesa. Pored toga, korisnik može da razvije novemodele pomoću „Portraits“ jezika ili da koristi gotove modele iz svoje skromne baze.

Dinamičke simulacije separacionih procesakompleksnih smeša, destilacione kolone,CSTR, PFR i neidealni reaktori, kao i parni sis-temi u petrohemijskoj industriji i termoelek-tranama.

Dymola Besplatan varijanta, open source. Jedini simulacioni alat koji je trenutno na raspolaganju simulacionom jeziku Modelica,koji je zasnovan na objektno orijentisanom modelovanju.

Tek u razvoju, značajan za dinamičke simula-cije procesa.

Gproms Moguća i simulacija gorivnih ćelija. Reaktori, separacija, kristalizacija, biotehno-logija.

Mathlab Iako je MATLAB prvenstveno namenjen za numeričko raču-narstvo i kreiranju grafika, opcioni blokovi koje sadrži omo-gućavaju pristup i grafičkim mogućnostima programiranja, koje mogu biti korišćenje u oblasti hemijskog inženjerstva.Softver široke primene koji može da se koristi kao dopunski alat prilikom projektovanja procesa.

Koristi se za rešavanje, sistema nelinearnihjednačina, složenih jednačina iz termodina-mike ili automatskog upravljanja procesima, au upotrebi je i paket Simulink za modelovanje,simulaciju i analizu multidomenskih dinamič-kih sistema.

EES Eng. Engineering Equation Solver je uglavnom namenjensimulaciji stacionarnih režima rada. Besplatan za edukativne svrhe.

Rešavanje opštih bilansnih jednačina procesa.

ANALIZA KARAKTERISTIKA PROCESNIH SIMULATORA

U ovom poglavlju analizirane su izdvojene karak-teristike odabranih procesnih simulatora, radi lakšeg upoređivanja (cena, broj modela uređaja, broj termo-dinamičlih modela, operativne mogućnosti i veličina baze hemijskih komponenti). Osim nekoliko besplatnih verzija, relativno ograničenog dometa, ovi softveri su

dostupni na tržištu. U tabeli 3 je dat trenutni pregled besplatnih i komercijalnih procesnih simulatora.

U tabeli 4 nalazi se detaljniji opis mogućnosti opera-tivnih karakteristika vodećih procesnih simulatora.

Zajedničko za sve softvere iz ove grupe je da imaju više desetina modela opreme za simulaciju razvrstanih u nekoliko uobičajnih podgrupa: reaktori, razmenjivači toplote, kolone, separatori i ostala oprema. Najveći broj modela ima SuperPro Designer, a najdetaljnija


553

Tabela 3. Pregled besplatnih i komercijalnih procesnih simulatora Table 3. Review of free and comercial process simulators

Naziv softvera Dostupnost Naziv softvera Cena PRO II Komercijalan Toxchem Komercijalan ProSim Komercijalan EcoSim Komercijalan Alph Komercijalan EES Komercijalan AspenPlus Komercijalan Kemisimp Besplatan SuperPro Designer Komercijalan Coco Simulator Besplatan

Hysys Komercijalan DWSIM Besplatan

Tabela 4. Pregled tehničkih mogućnosti i karakteristika procesnih simulatora Table 4. Review of process simulators abillities and characteristics

Kategorija DESIGN II ASPEN Plus HYSYS PRO/ II ProMax Chem Cad SuperPro Designer

Opš

te k

arak

teris

tike

Baze podataka x x x x x x x Termodinamika x x x x x x x Konvergencija recikla x x x x x x x Šaržni režim rada x x Procesiranje gasa x x x x x x x Šaržne simulacije x x x x x Dinamičke simulacije x x x x x x Mreže razmenjivača x x x x x x Elektroliti x x x x Mreže cevovoda x x x

Mod

eli u

ređa

ja

Destilacione kolone x x x x x x x Šaržne kolone x x x x Razmenjivači toplote x x x x x x x Flash separator x x x x x x x Reaktori x x x x x x x Pumpe i kompresori x x x x x x x Tankovi i rezervoari x x x Separatori čvrsto- tečno x x Separator gas – tečno x x x x x x x

podešavanja modela uređaja i operacija imaju softveri AspenPlus i Hysys. U većini programa omogućeno je kreiranje novih ili prepravljenje postojećih elemenata opreme. U tabeli 5 dat je broj modela procesnih jedi-nica (modela uređaja) koje su dostupne u osnovnim verzijama softvera.

Poznato je da više od 75% koda u procesnim simu-latorima posvećeno proceni fizičkih osobina, proračunu

i predviđanju ponašanja hemijskih komponenti, stoga se baze podataka koje sadrže fizičko-hemijske karakte-ristike čistih komponenti i parametre binarnih interak-cija za izračunavanje fazne ravnoteže, intenzivno koris-te i kontinuirano dopunjavaju u savremenim procesim simulatorima [18,19]. U tabeli 6 dat je pregled opšir-nosti baza komponenata odabranih procesnih simu-latora.

Tabela 5. Broj modela tehnoloških operacija Table 5. Number of unit operations

AspenPlus Hysys Unisim SuperPro ChemCAD PRO/II Design II 95 90 80 140 70 86 78

Tabela 6. Broj komponenati koje sadrže baze PS Table 6. Number of components in data base

AspenPlus Hysys Unisim SuperPro ChemCAD PRO/II Design II Od 15–24000 1800 + Aspen baza 1500 600 + 1700 2000 1750 900


554

Korisnik u najvećem broju slučajeva ima mogućnost dodavanja novih komponenti i to delimičnim preprav-kama postojećih ili upisom ključnih fiziko-hemijskih parametara i formiranjem novih komponenata. Važno je napomenuti da pored broja komponenata u bazi podataka, značaj ima i broj fizičkih, hemijskih, bioloških, sigurnosnih ili ekonomskih parametra kojima su defini-sane komponente baze. Takođe, od velikog značaja je pouzdanost baza, a prilikom nekih detaljnih simulacija uočljivu razliku može napraviti i broj značajnih cifara sa kojima su parametri komponente upisani u bazu.

Pored kvalitetnih baza podataka napredni procesni simulatori koriste i detaljne termodinamičke pakete koji se mogu prilagoditi korišćenim fluidima, smešama i drugim sistemima sirovina/reaktanta, te uslovima pro-cesa, radi dobijanja što vernijih rezultata simulacije. Mnoge baze mogu da sadrže i dopunske podatke poput bioloških osobina komponenata ili cena opreme, po-moćnih fluida i sl. Što je veći broj i pouzdanost kom-ponenata u bazi, veća je i univerzalna primena simu-latora, tabela 7.

Na osnovu iskustva i iznetih podataka u prethodnim pasusima po ukupnim performansama i karakteristi-kama prednjači softver AspenPlus, koji danas obuhvata nekoliko desetina specijalizovanih podopcija, iscrpnu bazu komponenata i modela uređaja, brojne termodi-namičke pakete i detaljne opcije podešavanja opreme. Međutim, u zavisnosti od zahteva projekta i neki drugi manje složeni softveri, u nekim slučajevima besplatni, mogu da pruže zadovoljavajuće odgovore na pojedine zahteve projektnih zadataka u oblasti tehnološkog projektovanja.

STUDIJA SLUČAJA SEPARACIJE PRIRODNOG GASA

Kako bi se utvrdila upotrebljivosti i pouzdanost PS prilikom projektovanja realnog procesa napravljena je studija slučaja simuliranja postojećeg procesa u više inženjerskih programa za projektovanje. Na osnovu dostupnih podataka za simulaciju je odabran je jedan relativno jednostavan proces, deo primarnog tretmana prirodnog gasa, separacija. Ovaj proces odvija se u više koraka. Svi podaci odabranog procesa dobijeni su u saradnji sa distributerom prirodnog gasa i zaštićeni su poslovnom tajnom. Ovaj proces je odabran kako bi rezultati simulacije mogli da budu upoređeni sa dos-tupnim realnim parametrima procesa, odakle se više može zaključiti o upotrebljivosti procesnih simulatora prilikom tehnološkog projektovanja. Navedeni proces

simuliran je u 7 različitih PS na osnovu zadatka u nas-tavku, a rezultati su upoređeni sa dostupnim podacima iz referentnog postrojenja.

Zadatak simulacije

Dva materijalna toka prirodnog gasa 1 i 2, defi-nisana u tabeli 8, nakon spajanja u mešaču (MIX-100), tretiraju se u separatoru (InletSep) za gas-tečno. Gasni tok 4 iz separatora se dalje hladi prvo u razmenjivaču toplote (razmenjivač toplote), zatim u električnom hla-dnjaku (Chiller), pre nego što uđe u sledeći separator (LTS) za gas-tečno. Gasna faza iz drugog separatora (tok 9) se koristi za rashlađivanje u razmenjivaču odakle izlazi tok 10 koji sadrži izlazni gas, čija je karakteristika povećan sadržaj metana. Tečna faza, tok 5, iz prvog separatora (InletSep) i drugog separatora (Sep2), tok 8, se spajaju u jedan tok koji se nakon mešanja u mešaču (MIX-200) dalje upućuje u destilacionu kolonu – Depro-panizer, slika 4. Iz kolone izlaze lakša i teža frakcija tokovi 12 i 13.

Tabela 8. Karakteristike ulaznih struja procesa Table 8. Inlet flow data

Materijalni tok 1 (molski udeo)

Materijalni tok 2 (maseni udeo)

N2 0,01 N2 0,02 CO2 0,01 CO2 0,00

Metan 0,60 Metan 0,40 Etan 0,20 Etan 0,20

Propan 0,10 Propan 0,20 i-Butan 0,04 i-Butan 0,10 n-Butan 0,04 n-Butan 0,08

Za shemu toka prikazanu na slici 4 poznati su sledeći radni parametri: sastav, temperatura i protok ulaznih materijalnih tokova, pad pritiska u svim uređajima, zadata temperatura na izlazu iz hladnjaka. Za potrebe projektovanja neophodno je odrediti: sastav svih mate-rijalnih tokova, dimenzionisati opremu, odrediti zahte-ve za grejanje i hlađenje, neophodnu površinu za top-lotnu razmenu, temperaturu i sastav prodajnog gasa nakon razmenjivača.

Na osnovu dostupnih informacija protok materijal-nog toka 1 iznosi 7597,7 kg/h, dok je protok toka 2: 4989,5 kg/h. U oba slučaja: T = 15,5 °C (60 °F), p = 41,4 bar (600 psi). Pad pritiska u svim mešačima je 0 bar, u razmenjivaču po 0,7 bar u omotaču i cevi (broj prolaza fluida u cevi i omotaču: 1), dok je za oba separatora pad

Tabela 7. Broj termodinamičkih paketa u okviru PS Table 7. Number of thermodynamic packages in process simulators

AspenPlus Hysys Unisim SuperPro ChemCAD PRO II Design II 80 30 32 28 20 38 35


555

pritiska po 0,7 bar, kondenzatora i rebojlera 0 bar, kolone 0 bar. Izlazna struja iz električnog hladnjaka je na temperaturi od –17,8 °C. Minimalna razlika teme-pratura u razmnjivaču je 10 °F. Destilaciona kolona ima 10 teorijskih podova. Pritisak u rebojleru je 14,1 bar, a temperatura 93,3 °C. Pritisak u kondenzatoru iznosi 13,8 bar, a temperatura 4,44 °C. Refluksni odnos je 1, a udeo gasne faze koja se vraća je 2. Ukoliko je dostupna, korišćena je Peng-Robinsonova jednačina.

U cilju demostracije mogućnosti softvera u oblasti projektovanja obuhvatiti: dimenzioniranje opreme, procenu efikasnosti procesa i odrediti energetske zah-teve električnih uređaja, proveriti koncepta procesa, i uraditi materijalni i energetski bilans. Sve simulacije se izvode pri istim uslovima, u sledećim procesnim simula-torima: Aspen Hysys (2.9), AspenPlus (7.1), DWSim, Design II (11.8), Super Pro Designer (5.1) i ChemCAD (6.0), Coco Simulator (2.05).

REZULTATI I DIKUSIJA

U ovom poglavlju dati su uporedni rezultati simu-liranog procesa po ključnim parametrima primarne separacije prirodnog gasa iz prethodno definisanog zadatka, a nastali nakon obavljanja više simulacija i provera u navedenim softverima, tabela 9.

Prilikom simulacije u svim navedenim softverima korišćeni su sledeći modeli uređaja: mešač (dvofazni, sa dva ulaza), separator (sa ciljem razdvajanja gasne i tečne faze ili flash bez operacije dogrevanja), razmenjivač toplote (sa cevnim snopom i omotačem,

unakrsni, suprotnostrujni), hladnjak (električni, bez pomoćnog fluida), destilaciona kolona sa podovima (sa parcijalnim kondenzatorom i rebojlerom).

Svi softveri su u svojim bazama podataka imali odgovarajuće ili približne modele ovih uređaja. Najveća prepreka u simulairanju ovog problema bila je u prvom koraku separacije, jer nisu svi simulatori uspeli da, bez dodatnih podešavanja, odvoje gasnu od tečne faze, što je dalje uticalo na razlike u materijalnom i energetskom bilansu. Slučajne greške poput odabira masenog umes-to molskog protoka, imale su značajan uticaj na pre-ciznost rezultata.

Svi korišćeni modeli separatora (flash uređaja), u svim PS, imaju kao osnovnu mogućnost dogrevanja smeše radi boljeg razdvajanja faza ili mogućnost pro-mene pritiska, koja u ovom slučaju nije korišćena. Neki su zasnovani na osnovnim jednačinama ravnoteže para-tečnost dok drugi daju mogućnost korisniku da odabere odgovarajuće jednačine stanja među nekoliko ponuđe-nih opcija. Najdetaljniji model separatora nalazi se u okviru softvera Hysys.

Prilikom simuliranja električnog hladnjaka u svim PS jednostavno se dolazi do podešavanja ovog uređaja, a rezultati se kreću u širem opsegu od 173 do čak 480 kW. Ove vrednosti direktno su zavisne od prethodno dobijenih vrednosti protoka i temperature ulaznog toka, a ti parametri su varirali od slučaja do slučaja.

Potrebno je obratiti pažnju na rezultate koje soft-veri daju prilikom simualcije hladnjaka, jer se negde rezultat odnosi na preporučenu instalisanu snagu apa-rata,negde na realnu snagu ili se daje veći broj jedinica

Slika 4. Procesna šema iz radnog okružnja simulatora Chem Cad. Figure 4. Flow diagram form ChemCad interface.


556

manje snage. Zbog toga je značajano napraviti uvid u teorijske osnove na kojima su izrađeni modeli koriš-ćenih uređaja.

S obzirom na to da nemaju svi softveri rigorozan proračun razmenjivača toplote, koji obuhvata i detaljnu geometriju, izbor materijala i sl., u softverima Hysys i AspenPlus su korišćeni pojednostavljeni modeli ovog uređaja.

Svaki procesni simulator ima drugačiji metod prora-čuna razmenjivača toplote i on se kreće od najjednos-tavnijeg – izborom minimalne razlike temperatura, naj-češće korišćenog prilikom ove simulacije, do detaljnog izbora geometrije, metode, materijala ili rasporeda i broja cevi u omotaču.

Dobijeni rezultati za temperaturu po ključnim para-metrima su u relativno malom odstupanju od obraču-nate srednje vrednosti, a u okviru granica parametara realnog procesa. Najveći broj rezultata poklapa se u opsegu od ±5 °C za traženu temperaturu, ili u opsegu od ±10% u odnosu na sastav toka po odabranoj kom-poneti ili površini za toplotnu razmeru, te bi mogli da budu prihvatljivi u početnim fazama projektovanja pro-cesa.

Očekivano najteži deo za podešavanje modela ure-đaja u svih sedam simulatora, bilo je podešavanje destilacione kolone. Rezultati u svim softverima bili su u okviru realno očekivanog opsega. Očekivani minimal-ni zbirni sadržaj i-butana i n-butana od preko 90% u težoj frakciji iz Depropanizera ostvaren je u 4 od 5 soft-vera, izuzev kod AspenPlus-a (71%).

Obzirom da je primer procesa dat je iz prakse jed-nog domaćeg distributera prirodnog gasa, a prema tim podacima prirodni gas nakon postupka separacije, u zavisnosti od sastava ulaznog gasa, najćešće sadrži preko 98–99 mas.% obe frakcije butana, što je ostva-reno u 3 od 6 prikazanih softvera za simulaciju. Najzah-tevnije je bilo podesiti kolonu u AspenPlus-u i SuperPro Designer-u.

Važno je napomenuti da su noviji softveri lakši za učenje, a kako je pokazano u ovom slučaju, to da li je softver besplatan ili komercijalan ne mora presudno da utiče na njegovu pouzdanost. Međutim softveri u čiji razvoj je uloženo više novca mogu po očekivanju imati bolje razrađene modele uređaja, detaljnije razrađene operacije ili kvalitetnije i obuhvatnije baze podataka.

Besplatni ili jeftiniji PS mogli bi da posluže za obuku i edukaciju studenata, a pri tom pružaju dobre osnove za korišćenje skupljih i naprednijih softvera u ovoj oblasti. U odabranom softveru DWSIM nije bilo moguće izvršiti simulaciju tretmana prirodnog gasa, jer razme-njivač toplote ne konvergira, a ovakav rezultat je oče-kivan obzirom da se radi o novom softrveru koji se još uvek nalazi u fazi razvoja.

Svim procesnim simulatorima, navedenim u okviru ovog poglavlja, se intuitivno rukuje, iako se njihova radna okruženja zasnivaju na različitim načinima unosa podataka.

Većina dostupnih softvera omogućava bar najos-novnije operacije u procesima sinteze i prerade hemij-skih poizvoda, razmene toplote, tretmana otpadnih voda i separacije. Mogućnosti softvera mogu se pobolj-šati dodatnom razradom otvorenih modela i dopunom baza podataka.

Najveći broj PS primenljiv je u industriji nafte i gasa. Ekspanzija industrije nafte i gasa u poslednjih nekoliko decenija uslovila je primarni razvoj softvera u ovom pravcu, te danas većina softvera omogućava precizne simulacije.

U poslednje vreme primetan je trend uže specijali-zacije simulatora te oni sada sadrže podprograme prila-gođene za rigoroznu simulaciju određenih postrojenja ili uređaja (hidrokreking, hidrotriting, bioreaktora, baklji i mreže razmenjivača toplote).

Mane savremenih softvera, mogu biti zatvorenost pojedinih modela za izmene, nedostatak pojedinih ope-racija, nedovoljno precizna podešavanja uređaja, male i

Tabela 9. Prikaz rezultata uporedne simulacije u više različitih softveraTable 9. Results form side by side process simulaton in different softwares

Rezulat Jedinica Viši cenovni rang Srednji cenovni rang Besplatni softveri Srednja vrednostHYSYS AspenPlus SuperPro

Designer ChemCAD Design II Coco

Simulator DWSIM

Temperatura toka 10 °C 9,9 2,9 5,0 10,0 5,6 15,6 – 11,3 Površina za toplotnu razmenu

m2 5,1 6,5 6,9 4,1 37,8 – – 13,1

Snaga hladnjaka kW 315,0 378,3 245,4 481,5 380,0 173,0 – 328,7 Protok na ulazu u kolonu

kg/h 5270,7 5344,2 5589,1 5453,0 5236,2 3802,0 – 5115,8

Sadržaj metana u „izlaznom gasu“

mas.% 59,8 76 60,9 79,9 64,7 62,5 – 67,3

Sadržaj butana u težoj frakciji kolone

mas.% 98,0 71 91,9 98,7 99,1 – – 91,7


557

nekompletne baze komponenti i slično, Iako mnogi od softvera ostavljaju slobodne opcije da korisnik sam pro-gramira ili kreira modele novih uređaja ili tehnoloških operacija, osnovana je pretpostavka da veliki broj ko-risnika u industrijskim uslovima, bez pomoći neke raz-vojno-istraživačke institucije, neće biti u prilici da napravi i razvije kvalitetne modele.

ZAKLJUČAK

U ovom radu dat je pregled više od 20 savremenih procesnih simulatora ili srodnih programa, od kojih su neki detaljnije opisani ili upotrebljeni u studiji slučaja separacije prirodnog gasa. Date su preporuke za nji-hovu primenu u zavisnosi od njihove bliže namene ili mogućnosti po oblastima hemijske industrije, sa poseb-nim osvrtom na značaj koji imaju prilikom projekto-vanja procesa. Procesni simulatori su razdvojeni na procesne simulatore u užem smislu i pomoćne pro-grame prilikom projektovanja. Svi su klasifikovani pre-ma mogućnostima primene i osnovnim parametrima.

Date su uporedne karakteristike rezulata simuliranja separacije prirodnog gasa za potrebe projektovanja procesa, na osnovu simulacija u softverima različitih složenosti i kvaliteta. PS su pokazali svoju upotrebnu vrednost kao značajan oslonac projektantima, iako su rezultati koje su dali pokazali međusobno značajna odstupanja. Ipak, uporedne simulacije separacije pri-rodnog gasa u najskupljim softverima dale su približne rezultate, što potvrđuje da su savremeni procesni simu-latori na visokom nivou pouzdanosti, bez obzira što su zasnovani na različitim modelima, jednačinama, meto-dama operacija, interfejsima ili različitim bazama poda-taka.

Procesni simulatori utiču na skraćivanje predpro-jektnih faza poput istraživanja i razvoja, time dovode do brže komercijalizacije industrijskih ideja. Pored svih navedenih prednosti važno je napomenuti da je za kvalitetno projektovanje procesa i dalje je neophodno dosta vremena, a da su sami simulatori samo oruđe u rukama projektanata, koji ipak sve rezultate moraju potvrditi i na sve druge načine.

Zahvalnica

Istraživanja u ovom radu izvršena su u okviru aktivnosti na projektu TR 34009 koji finansira Minis-tarstvo prosvete, nauke i tehnološkog razvoja Repu-blike Srbije.

LITERATURA

[1] R. Gani, E.N. Pistikopoulos, Property Modelling and Simulation for Product and Process Design, Fluid Phase Equilib. 194–197 (2002) 43–59.

[2] M. Kellner, R. Madachy, D. Raffo, Software Process Modeling and Simulation: Why, What, How, J. Sys. Software 46 (1999) 91–105.

[3] T.B. Thompson, Chemical Industry of the Future: Tech-nology Roadmap for Computational Chemistry, Council for Chemical Research, http://www1.eere.energy.gov/ /manufacturing/resources/chemicals/pdfs/compchemistry_roadmap.pdf, septembar 2012.

[4] M. Jovanović, Osnovi projektovanja I deo: Teorija pro-jektovanja, Tehnološki fakultet, Leskovac, 1994, str. 164.

[5] M. Jovanović, Osnovi tehnološkog projektovanja, SHTS, Beograd, 2004.

[6] M. Jovanović, Z. Popović, Razvoj procesa: Procesna ekonomika sa studijama slučaja hemijske tehnologije, SHTS, Beograd, 2003.

[7] R. Omorjan, Modelovanje i simulacija procesa, Tehno-loški fakultet, Univerzitet u Novom Sadu, http:// //www.tf.uns.ac.rs/omorjan/radovan_omorjan_003_hip/1uvod.pdf , oktobar 2012.

[8] http://www.aspentech.com/solutions/industry_solutions/epc/index.aspx, septembar 2012.

[9] Q. Smejkal, M. Šoóš, Comparison of computer simu-lation of reactive distillation using Aspen Plus and Hysys software, Chem. Eng. Process. 41 (2002) 413–418.

[10] M. Nikooa, N.Manhinpey, Simulation of biomass gasi-fication in fluidized bed reactor using Aspenplus, Bio-mass Bioenerg. 32 (2008) 1245–1254.

[11] M. Tasić, O. Stamenković, V. Veljković, Simulacija meta-nolize suncokretovog ulja u Aspen Plus i Hysys softveru, VIII Simpozijum savremene tehnologije i privredni raz-voj, Zbornik izvoda radova, Leskovac, oktobar 2009, str. 166.

[12] WinSim Co.,Advanced Process Simulation Solutions, www.winsim.com/ourbrochure, novembar 2012.

[13] D. Živković, „Software for the simulation processes in CHP plants“, MF Niš, http://simterm.masfak.ni.ac.rs/ proceedings/13-2007/papers/sessions/5_Matematicko _modeliranje_i_numericka_simulacija/5-4/GZivkovic_ MFNis.pdf, PROCESING'12, Beograd, jun 2012.

[14] J. Sadeq, H.Duarte, R. Serth, Anomalous results from process simulators, Chem. Eng. Educ. 31 (1997) 46–51.

[15] H. Zhang, B. Kitchenham, D. Pfahl, „Software Process Simulation Modeling: Facts, Trends and Directions“, 15th Asia-Pacific Software Engineering Conference, Peking, China, 3–5 december, 2008.

[16] N. Nishida, G. Stephanopoulos, A. Westerberg, A Review of Process SynthesisAIChE , J. 27 (2004) 321–352.

[17] T.M. Komulainen, R. Enemark-Rasmussen, G. Sin, J.P. Fletcher, D. Cameron, Experiences on dynamic simul-ation software in chemical engineering education, Edu-cation for Chemical Engineers 7 (2012) e153–e162.

[18] M. Fermeglia, S. Pricl, and G. Longo,Molecular Modeling and Process Simulation: Real Possibilities and Challen-ges, Chem. Biochem. Eng. Q. 17 (2003) 19–29.

[19] L. Castillo, C.A. Dorao, Decision-making in the oil and gas projects based on game theory: Conceptual process design, Energy Convers. Manage. 66 (2013) 48–55.


558

SUMMARY Application of process simulators in chemical engineering PROCESS design –natural gas separation plant case study Dimitrije Ž. Stevanović1, Mića B. Jovanović2, Marina A. Mihajlović1, Jovan M. Jovanović1, Željko B. Grbavčić2

1Innovation Center, Faculty of Technology and Metallurgy, University of Belgrade, Belgrade, Serbia 2Faculty of Technology and Metallurgy, University of Belgrade, Belgrade, Serbia

(Professional paper)

Software for chemical processes modeling and simulation, in the past fewdecades, plays an important role in the development of chemical-process industry with its growing capabilities and wide range of application. Usage of processsimulators in Serbia for the process design is very limited. This paper gives a briefoverview of the numerous process simulators that are used in the chemical-process industry today. The conceptual design is responsible for most of the investment costs in chemical process industry. Importance of precise design onpreliminary level is obvious. Wrong decisions made at the conceptual level could be carried out throughout the chain in process design to the detailed design procedures and procurement of equipment. Although preliminary design phasecomprises only about 2% of the total cost of the project, it contributes signific-antly to the reduction of cost of the project by more than 30%. Therefore processsimulators play important role in elimination of unnecessary errors in basicprocess design. Here is also shown a case study of parallel process simulated indifferent process simulators which tests the results, the reliability and usefulnessof these programs in solving specific engineering tasks. Comparison of givensimulation results confirm that the modern process simulators are at high level ofconfidence, no matter they are based on different models, equations, methods,operations, interfaces or data bases. Usage of software speeds up the arrival ofoptimized solution during the design and the operational procedures. Thereforesoftware has significant impact on reducing time of pre-project phase such as research, conceptual design, and proving project abilities. Its development leads to the faster commercialization of industrial ideas.

Keywords: Process simulators • Chemical engineering process design • Natural gas treatment plant simulation

559

Economizer water-wall damages initiated by feedwater impurities

Sonja M. Vidojković1, Antonije E. Onjia2, Aleksandar B. Devečerski3, Nebojsa N. Grahovac3, Aleksandra B. Nastasović1 1University of Belgrade, Institute of Chemistry, Technology and Metallurgy, Belgrade, Serbia 2University of Belgrade, Laboratory of Chemical Dynamics and Permanent Education, “Vinča” Institute of Nuclear Sciences, Belgrade, Serbia 3University of Belgrade, Laboratory of Material Science, “Vinča” Institute of Nuclear Sciences, Belgrade, Serbia

Abstract The main causes of efficiency loss in thermal power plants are boiler tube failures thatdiminish unit reliability and availability, and raise the cost of the electric energy. For thatreason, the regular examination of boiler tubes is indispensable measure for prevention offuture malfunctions of power units. The microscopic examination of economizer’s innerwall microstructure, the analysis of chemical composition of deposit using X-ray diffraction (XRD) and scanning electron microscopy/energy dispersive spectroscopy (SEM/EDS) havebeen performed in a subcritical power plant. Stress corrosion cracking, pitting corrosion,destroyed protective magnetite layer, presence of magnetite and hematite in deposit andcorrosive impurities within the cracks have indicated the effect of inadequate quality offeedwater that cannot entirely ensure reliable operation of the boiler. It may be statedthat maintenance of present boiler does not provide its reliable operation. The extensivechemical control of water/steam cycle was recommended.

Keywords: boiler tube, deposition, corrosion, feedwater impurities.

PROFESSIONAL PAPER

UDC 621.311:621.643:620.193:54

Hem. Ind. 68 (5) 559–563 (2014)

doi: 10.2298/HEMIND130715082V


The failure of boiler tubes is one of the foremost problems in most of processing industries involving thermal power plants [1–6]. It causes considerable effi-ciency loss and consequently enormous economic effects [7,8]. Tubes are typically made of carbon or low alloy steel and exposed to stress, aggressive environ-ment and diverse phases of water at high temperature and pressure. This environment has a high potential of corrosion processes development. Thus, majority of boiler tube failures are reported from waterside cor-rosion [9–13].

The review of inspection and maintenance records of 210MW power plant operating at subcritical para-meters revealed numerous shutdowns ensued from boiler tube failure. Furthermore, the number of boiler tube failures and breakdowns on power units has been increasing in the last ten years. Economizer, which is one of the critical components of boiler generally, was the most damaged part. In order to provide evidences for economizer’s inner wall condition and reveal pos-sible causes of its damages, the metallographic exami-nation of metal surface and deposit structure was undertaken. The boiler is coal-fired with natural circul-ation of the boiler water. The boiler tubes were made

Correspondence: S.M. Vidojković, University of Belgrade, Institute of Chemistry, Technology and Metallurgy, Njegoševa 12, 11001 Bel-grade, P.O.B. 473, Serbia. E-mails: [email protected]; [email protected] Paper received: 15 July, 2013 Paper accepted: 29 October, 2013

of steel St.45.8 DIN 17175 and operated 19871 h. All- -volatile treatment (AVT(R)) was employed for feed-water conditioning. This treatment comprises the dos-ing of ammonia, as alkalizing, and hydrazine, as redu-cing agent.

EXPERIMENTAL ANALYSIS

The specimens of the economizer tube were sub-jected to various laboratory examinations and analysis. These included visual inspection, optical microscopic (metallographic) analysis, X-ray diffraction (XRD) and scanning electron microscopy/energy dispersive spec-troscopy (SEM with EDS).

Techniques and instrumentation

The optical microscopy of internal wall was emp-loyed for metallographic examination of tubes. The metallographic examination was carried out along the tube. The examination was made both on the internal surfaces and on cross-sectional surfaces of the tubes. The microstructure analyses were made using light mic-roscope Zeiss Axioplan with diverse enlargements. Spe-cimens were photographed by digital camera. The entire cross-section of the tubes was observed and the thickness of corrosion layer was measured. The measu-rement was carried out at ten points around the tube circumference perpendicularly to the inner surface. The Zeiss-Axio Vision program was used for picture proces-sing and analysis, as well as for measuring corrosion depth of damages in tube material.

S.M. VIDOJKOVIĆ et al.: WATER-WALL DAMAGES INITIATED BY FEEDWATER IMPURITIES Hem. ind. 68 (5) 559–563 (2014)

560

Structural characterization of deposit was made using X-ray diffraction (XRD) technique. Structural anal-ysis of powdered samples was performed by Siemens D-500 powder diffractometer. CuKα radiation was used in conjuction with a CuKβ nickel filter.

Samples and preparation

Economizer specimens were cut out by SiC abrasive disk cooling with water. Scraped deposit was sent to XRD analysis in order to determine its phase compo-sition.

After submergence in the methacrylate, the speci-mens for microstructural examination were honed using SiC papers with fineness from 80 to 1000 and per-manently cooling with water. Polishing was performed by diamond paste fineness of 10 to 5 μm and 2-0 μm. Tube specimens were observed before and after che-mical etching which was performed by 2% nitric acid (HNO3) in ethanol.


Microstructure analysis

Six specimens were analyzed and results indicated that their internal surfaces are covered by scaled, scabrous, bulging and rugged deposit layer which is brown/reddish color. Metal surface was caved, cran-nied and holey. A representative specimen is shown on Figure 1.

Figure 1. Internal surface of economizer.

The microstructures of wall tube cross section for all specimens are presented on Figure 2. Distinctive feature of economizer’s microstructure is presence of longitudinal cracks, approximately 50 μm length in material, which were propagated parallel to the tube surface (specimens 1 and 3). It pointed at the stress corrosion effect. Due to the presence of cracks in the tube wall further exposure to the same environment and strain will lead to the growth of cracks and rupture of tubes. Even very small concentrations of certain highly active chemicals, like chloride, have a huge con-tribution to the uprising the stress corrosion cracking and usually leading to devastating and unexpected fail-ure. Corrosion pits (specimens 2, 4 and 6) that were rec-

(a) (b)

(c) (d)

(e) (f)

Figure 2. Cross-sectional view of specimens: a) specimen 1 with stress corrosion cracking, b) specimen 2 with corrosion pits, c) specimen 3 with stress corrosion, d) specimen 4 with corrosion pits, e) specimen 5 with cracks in oxide layer and f) specimen 6 with corrosion pits.

orded on microphotographs, clearly indicated that pit-ting corrosion took place in all specimens. Pitting cor-rosion could be attributed to the oxygen attack to the metal substrate. Microstructure analysis evinced des-truction of the protective layer (specimen 5) that led to development of nonprotective scale localized on the internal tube surface. The precipitation of magnetite scale can increase the tube metal temperature and cause cracks and other deformations of metal construction [14–16]. Thickness of deposit layer varied between 13 and 160 μm. Whereas analysis revealed excess of chlo-ride in feedwater, one of possible explanations of deve-loping nonprotective magnetite can be based on the mechanism of local formation of acid chlorides. Result of this process is pH drop, due to creating an acidic region with a high Cl– concentration. Hydrolysable chlo-rides cause degradation in these zones of the protec-tive film. The most likely cause of increased concentra-tion of chloride in feedwater is hydrochloric acid (HCl)


561

utilized for regeneration of the ion exchange resins in condensate polisher. Consequently, the chloride con-centration can be diminished by proper regeneration of polishing and water treatment plants and their accu-rate exploitation. Introducing of continuous on-line and daily laboratory monitoring of chloride in demi and feedwater by ion chromatography technique are also important measures for maintaining chloride below specified target values. Aside from that, the attention should be paid to the control and cease of condenser leakage that can be an additional source of chloride in feedwater. On the basis of obtain microphotographs, material microstructure was unmodified and consisted of ferrites and perlites.

Structural and elemental characterization of the deposit

Structural characterization of deposit was made using X-ray diffraction (XRD). The main constituents of deposit from specimens 1, 2, 3, 4, 5 and 6 have con-firmed to be magnetite (Fe3O4) and hematite (Fe2O3) (Figure 3). The deposit layer composition of specimens 5 and 6 consists of magnetite and prevalent amount of hematite. Iron oxides are typical steel oxidation pro-ducts (9,14,17–19). However, the presence of hematite may indicate irregularity in the power cycle operation and high corrosion rate. Moreover, hematite does not provide corrosion protection and has lower electrical and thermal conductivity than magnetite. Hematite also might be carried over to the boiler tubes from other part of the system. Though magnetite is protec-

tive corrosion product, in present case its porous and brittle nature might have an adverse effect on the protection of the boiler tube surface. Porous magnetite rust can enhance the corrosion at the tube surface by acting as an active cathodic site and deploy under-deposited corrosion [20].

The three representative specimens (3, 4 and 5) were considered for the EDS analysis of deposit. Depo-sit seated inside the cavities and craters was undergone analysis. Results for specimen 3 indicated iron-oxide forming elements (Fe 69.15–70.42% and O 27.15– –29.36%) as main components of rust. Other peaks cor-responding to Si, Cu, S, Cr, Mn, Co and Ni were reg-istered with very low intensity. Depth of crater was 130 μm. Deposit of specimen 4 originated from crater depth of 100 μm. Deposit contained O (27.52–31.38%) and Fe (34.91–72.53%). Elements presented in low por-tions were Na, Cl, Mg, Cr, Zn, K and Co, although Si was present at the mass fraction of 19.27%. On specimen 5 the uppermost elements found within crater deep 220 μm were Fe (64.19%) and O (28.81%). Elements pre-sented at minor levels were Si, Cu, S, Mg, Cr, Co, Zn and As. The Cu was present to the extent of 1.16%. The pre-sence of copper found in the brown/reddish deposit-layer, can enforce the localized corrosion of boiler tubes through galvanic corrosion [21]. The source of copper is corrosion and decomposition of condenser tubes expressly in presence of excessive ammonia in the boiler feedwater [21–24].

Figure 3. Difractogram of specimens: a) 1, b) 2, c) 3, d) 4, e) 5 and f) 6.


562

CONCLUSION

Obtained results of microstructure analysis of the boiler tubes lead to the conclusion that impurities pre-sent in feedwater resulted in corrosion pits, stress cor-rosion cracking, as well as destruction of protective magnetite layer. The protective magnetite layer lost its protective function which was followed by magnetite scale formation that can increase the tube metal tem-perature at these localized sites. Magnetite is com-monly protective corrosion product but, in present case, its porous and brittle feature might have an adverse effect on tube surface protection. Structural characterization of the deposit proved that it was com-posed mainly of magnetite and hematite. SEM/EDS have shown that impurities associated with pits and cracking included Fe, O, Cl, Cu, Na and Si. Regular and correct monitoring of corrosive contaminants in water/ /steam cycle is crucial for prevention of corrosion pro-cesses and minimizing deposit formation on water-wall boiler tubes. Consequently, proper water chemistry is paramount in extending the life and increasing reli-ability of the unit.

Acknowledgments

This research was supported by the Ministry of Edu-cation, Science and Technological Development of the Republic of Serbia (Projects Nos. III 43009 and III 45012).

REFERENCES

[1] State-of-Knowledge on deposition, Part I: Parameters Influencing Deposition in Fossil Boilers, TR-1004194, Palo Alto, CA, 2002.

[2] O. Jonas, Monitoring of Steam plants, Mater. Perfor-mance 42 (2003) 38–42.

[3] R.J. Niebo, Implementing a Boiler Tube Failure Mecha-nism Reporting Program, in Proceeding of International Conference on Boiler Tube Failures in Fossil Plants, TR-100493, Electric Power Research Institute, Palo Alto, CA, 1992.

[4] J.P. Dimmer, R.B. Dooley, EPRI’s New BTF Reduction Program, in Proceedings of Third International Confe-rence on Boiler Tube Failures in Fossil Plants, TR-107775, Electric Power Research Institute, Palo Alto, CA, 1998.

[5] E. Thurston, Inspecting for Corrosion Fatigue, Power Eng. 112 (2008) 46–49.

[6] R.B. Dooley, W.P. McNaughton, Don’t let those Boiler Tubes Fail Again, Power Eng. 101 (1997) 56–61.

[7] Cost of Corrosion in the Electric Power Industry (EPRI Report TR-1004662), Palo Alto, CA, 2001.

[8] B.C. Syrett, J.A. Gorman, Cost of Corrosion in the Electric Power Industry – An Update, Mater. Performance 42 (2003) 32–38.

[9] A. Husain, K. Habib, Investigation of tubing failure of super-heater boiler from Kuwait Desalination Electrical Power Plant, Desalination 183 (2005) 203–208.

[10] D. Flynn, The Nalco Water Handbook, 3rd Ed., McGraw-Hill, New York, 2009.

[11] A. Henderson, D. Brazil, Failure analysis of HP feedwater line elbow, Power Plant Chemistry 14 (2012) 76–82.

[12] E. Ring, Start-up, Shut-down and Lay-up Improvements at Earing Power Station, Power Plant Chemistry 14 (2012) 489–507.

[13] M.R. Mozdianfard, E. Behranvand, A field study of foul-ing in CDU preheaters at Esfahan refinery, Appl. Therm. Eng. 50 (2013) 908–917.

[14] W.E. Bornak, Chemistry of iron and its corrosion pro-ducts in boiler system, Corrosion 44 (1988) 153–158.

[15] F.R. Hitchings, P.M. Unterweiser, Eds., On-Load Corros-ion in Tubes of High-Pressure Boilers, ASM Internatio-nal, Materials Park, OH, 1981.

[16] Steam: Its Generation and Use, 38th ed., Babcock and Wilcox, New York, 1972.

[17] Yu.V. Zenkevic, V.E. Sekretar, Formation of Iron Oxide Deposits in Supercritical Steam Generator, Therm. Eng. 11 (1976) 66–69.

[18] N.N. Mankina, A.G. Tkachenko, L.G. Buinovskaya, Methods of Determination of Iron Oxide Deposition on Inner Heat Exchange Surfaces of High-Pressure Boilers, Therm. Eng. 9 (1960) 30–34.

[19] H.M. Shalaby, On the Mechanism of Formation of Hot Spots in Boiler Tubes, Corrosion 62 (2006) 930–941.

[20] M.G. Fontana, Corrosion Engineering, McGraw-Hill Book Company, 1986.

[21] F.N. Kemmer, ed., The NALCO Water Handbook, 2nd Ed., McGraw-Hill, New York, 1988.

[22] Copper Alloy Corrosion in High Purity Feedwater, Elec-tric Power Research Institute, TR-1000456, Palo Alto, CA, 2000.

[23] Corrosion of Cu–Ni–Zn Alloys in Water–Ammonia Power Plant Environments, Electric Power Research Institute, TR-113697, Palo Alto, CA, 1999.

[24] State-of-Knowledge of Copper in Fossil Plant Cycles, Electric power research Institute, TR-108460, Palo Alto, CA, 1997.


563

IZVOD

OŠTEĆENJA UNUTRAŠNJEG ZIDA EKONOMAJZERA IZAZVANA NEČISTOĆAMA U NAPOJNOJ VODI Sonja M. Vidojković1, Antonije E. Onjia2

, Aleksandar B. Devečerski3, Nebojsa N. Grahovac3, Aleksandra B. Nastasović1

1Univerzitet u Beogradu, Institut za hemiju, tehnolgiju i metalurgiju, Beograd, Srbija 2Univerzitet u Beogradu, Laboratorija za hemijsku dinamiku i permanentno obrazovanje, Institut za nuklearne nauke “Vinča”, Beograd, Srbija 3Univerzitet u Beogradu, Laboratorija za materijale, Institut za nuklearne nauke “Vinča”, Beograd, Srbija

(Stručni rad)

Osnovni uzroci smanjenja efikasnosti na termoelektranama su oštećenja ipucanja kotlovskih cevi koja utiču na smanjenje pouzdanosti i raspoloživosti ter-moblokova što dovodi do povećanja cene električne energije. Cevi se najčešće izrađuju od ugljeničnog ili niskolegiranog čelika i izložene su naprezanju, agresiv-nom okruženju i dejstvu različitih faza vode na visokim tempeaturama i pritiscima.Jedan od razloga pucanja cevi i oštećenja konstrukcionog materijala je prisustvo nečistoća u napojnoj vodi koje se akumuliraju ispod naslaga i dovode do razvojakorozionih procesa. Zbog toga, redovno ispitivanje kotlovskih cevi predstavljaneophodnu meru u cilju prevencije zastoja termoblokova. Na termoelektranidokritičnih parametara snage 210 MW zabeležen je porast zastoja izazvanih ošte-ćenjima cevnog sistema koja su u najvećem broju slučajeva bila na ekonomajzeru.Ispitivani kotao je sa prirodnom cirkulacijom, radi na ugalj, a konstrukcionimaterial je čelik St.45.8 prema DIN 17175. Da bi se utvrdilo stanje unutrašnjegzida ekonomajzera i otkrili mogući uzroci oštećenja izvršena su metalografskaispitivanja njegovog unutrašnjeg zida kao i sastava naslaga. Analize su obuhvataleispitivanje mikrostrukture, analizu hemijskog sastava naslaga pomoću rendgenske difrakcije (XRD) i skenirajuće mikroskopske elektronske analize/energetske disper-zione spektroskopije (SEM/EDS). Prsline zbog naponske korozije, tačkasta korozija,razoren zaštitni sloj magnetita, prisustvo hematita u naslagama i korozione nečis-toće nađene u pukotinama cevi ukazale su na dejstvo neadekvatnog kvalitetanapojne vode koja ne obezbeđuje pouzdan rad kotla. Magnetit, koji inače imazaštitnu funkciju, dobio je poroznu strukturu i postao krt i lomljiv, što može imatištetne efekte na cevi kotla. Strukturna karakterizacija naslaga pokazala je da seone sastoje uglavnom od magnetita i hematita, a na osnovu SEM/EDS analizazaključeno je da se u prslinama nalaze Fe, O, Cl, Cu, Na i Si. U cilju prevencije koro-zije i naslaga na kotlovskim cevima predloženo je uvođenje adekvatnog monitor-inga korozionih nečistoća u vodeno/parnom ciklusu. Regularna hemijska kontrolavode je od izuzetnog značaja za produženje radnog veka postojenja i povećanjepouzdanosti termobloka.

Ključne reči: Kotlovske cevi • Naslage •Korozija • Kontaminanti u napojnoj vodi

565

Formulacija i karakterizacija samo-mikroemulgujućih nosača lekovitih supstanci na bazi biokompatibilnih nejonskih surfaktanata

Ljiljana M. Djekić, Marija M. Primorac

Univerzitet u Beogradu, Farmaceutski fakultet, Katedra za farmaceutsku tehnologiju i kozmetologiju, Beograd, Srbija

Izvod Razvoj samo-mikroemulgujućih nosača je značajna savremena strategija za unapređenjeperoralne primene teško rastvorljivih aktivnih supstanci. Cilj rada bio je formulacija i karak-terizacija samo-mikroemulgujućih nosača na bazi smeše biokompatibilnih nejonskih surfak-tanata (PEG-8 kaprilno/kaprinski gliceridi (Labrasol®) i PEG-40 hidrogenizovano ricinusovo ulje (Cremophor® RH40)) za peroralnu primenu ibuprofena i in vitro karakterizacija njihove fizičke stabilnosti i veličine kapi nakon dispergovanja u vodenim medijumima različite pH vrednosti i in vitro profila oslobađanja lekovite supstance iz nosača. Rezultati karakteriza-cije ukazali su na značaj vrste i koncentracije ulja i masenog odnosa upotrebljenih surfak-tanata za sposobnost samo-mikroemulgovanja, kapacitet za solubilizaciju ibuprofena i nje-govu brzinu oslobađanja iz nosača.

Ključne reči: samo-mikroemulgujući nosači lekovitih supstanci; mikroemulzije; Labrasol®; Cremophor® RH40; ibuprofen.

NAUČNI RAD

UDK 619.03:66:544:54

Hem. Ind. 68 (5) 565–573 (2014)

doi: 10.2298/HEMIND130825083D


Samo-mikroemulgujući nosači lekovitih supstanci (eng. Self-microemulsifying drug delivery systems, SMEDDS) su izotropne smeše surfaktanata (HLB* > 12), ulja (≤ 20%) i eventualno, hidrofilnih rastvarača. Aktiv-na supstanca je rastvorena u nosaču. Nakon peroralne primene SMEDDS, u kontaktu sa vodom u gastro-intes-tinalnom traktu (GIT), pri blagoj agitaciji pod uticajem motiliteta želuca i tankog creva, formira se nosač tipa ulje-u-vodi (u/v) mikroemulzija [1]. Mikroemulzije su optički izotropni, transparentni, termodinamički stabilni sistemi. Njihovo formiranje je brzo i odvija se gotovo spontano. Veličina kapi mikroemulzija se kreće u ras-ponu od 10–100 nm [2–4]. Kako je veličina kapi mikro-emulzionog nosača vrlo mala, ukupna površina filma surfaktanta na granici između uljane i vodene faze je izuzetno velika, pa imaju visok kapacitet za solubiliza-ciju teško rastvorljivih aktivnih supstanci i potencijal za poboljšanje njene apsorpcije i ukupne biološke raspolo-živosti. U ovom slučaju apsorpcija lekovite supstance se odigrava gotovo nezavisno od uticaja hrane i endogenih faktora (digestija lipida iz nosača i sekrecija žuči i pan-kreasnog soka), a kod izrazito liposolubilnih lekovitih supstanci moguća je apsorpcija limfnim putem i zaobi-laženje metabolizma u jetri [5–7]. Veliki broj farmako-loški aktivnih supstanci koje su već u upotrebi, kao i one čija se potencijalna primena u farmakoterapiji

Prepiska: Lj. Djekić, Farmaceutski fakultet, Katedra za farmaceutsku tehnologiju i kozmetologiju, Vojvode Stepe 450, 11221 Beograd, Srbija. E-pošta: [email protected] Rad primljen: 25. avgust, 2013 Rad prihvaćen: 8. novembar, 2013 *Hidrofilno–lipofilna ravnoteža (eng. hydrophilic-lipophilic balance, HLB).

istražuje, ima ograničenu rastvorljivost u vodi, od-nosno, biološkim tečnostima. Njihova apsorpcija iz kon-vencionalnih farmaceutskih preparata je uglavnom nekompletna i varijabilna tj., apsorbuje se samo deo primenjene doze, što je nedovoljno za postizanje tera-pijskog efekta, a neresorbovani deo lekovite supstance izaziva neželjene efekte usled neodgovarajuće biodistri-bucije. Ta saznanja su podsticaj za razvoj SMEDDS i drugih strategija za povećanje rastvorljivosti i/ili brzine rastvaranja teško rastvorljivih lekovitih supstanci [8]. Formulacija SMEDDS je komplikovana jer samo smeše određenih farmaceutskih ekscipijenasa u uskom ras-ponu koncentracija mogu u kontaktu sa vodenim medi-jumom da obrazuju u/v mikroemulziju. Lekovita sup-stanca takođe može da utiče na proces mikroemul-govanja i veličinu kapi. Nakon mešanja SMEDDS sa vodenom fazom, hidrofilni sastojci, kao što su etanol i tečni polietilenglikoli, se u njoj rastvaraju, pa može doći do smanjenja njihove koncentracije u međupovršin-skom filmu, destabilizacije formiranog mikroemulzio-nog nosača i precipitacije lekovite supstance. Rizik za precipitaciju se naročito povećava sa porastom koncen-tracije hidrofilnih korastvarača, pa se izbegava njihova upotreba, međutim, time se smanjuje disperzibilnost SMEDDS u vodenom medijumu [1]. Imperativ pri for-mulisanju SMEDDS je da oni formiraju u/v mikroemul-zije i efikasno solubilizuju inkorporiranu terapijsku dozu lekovite supstance, u što širem rasponu koncentracija vodene faze i pH vrednosti. Još uvek nema generalnih smernica koje se odnose na formulisanje ovog tipa nosača, kao ni literaturnih podataka o kinetici osloba-đanja lekovitih supstanci. Takođe, proces formiranja u/v mikroemulzionog nosača in situ nije detaljno raz-jašnjen i predstavlja predmet aktuelnih istraživanja.

Lj.M. DJEKIĆ, M.M. PRIMORAC: SAMO-MIKROEMULGUJUĆI NOSAČI NA BAZI NEJONSKIH SURFAKTANATA Hem. ind. 68 (5) 565–573 (2014)

566

Ibuprofen ((RS)-2-(4-izobutilfenil)propanska kiselina) pripada grupi nesteroidnih antiinflamatornih lekova (NSAIL). Uobičajeno se primenjuje u dozi od 200–400 mg na 4–6 h, kao analgetik i antipiretik, odnosno, 300– –400 mg svakih 6–8 h, u terapiji reumatoidnog artritisa i osteoartritisa. Registrovani su lekovi sa ibuprofenom u obliku tableta, kapsula i oralnih suspenzija [9]. Ibupro-fen deluje kao neselektivni inhibitor enzima ciklooksi-genaza (COX-I i COX-II), odgovornih za sintezu prostag-landina, koji se smatraju medijatorima bola i inflama-cije. Analgetsko, antipiretsko i antinflamatorno dejstvo je posledica dejstva ibuprofena na COX-I, dok dejstvo na COX-II izaziva neželjene reakcije kao što je iritacija GIT. Ibuprofen je praktično nerastvorljiv u vodi [10]. Ispitivanja su pokazala da ova supstanca, zbog slabo kisele prirode (pKa 4,38) [11], ima pH zavisnu rastvor-ljivost, pa se smatra da se bolje rastvara u tankom crevu, gde je pH vrednost viša u odnosu na početne delove GIT [12]. Permeabilnost ibuprofena u kulturi radioaktivno obeleženih Caco-2 ćelija je visoka (vred-nost koeficijenta permeabilnosti (Papp) iznosi 55×10–6 cm/s)) [13], što ukazuje na laku apsorpciju rastvorene supstance iz GIT. Inkorporiranje ibuprofena u nosače tipa SMEDDS je potencijalno koristan pristup da se obezbedi poboljšanje rastvorljivosti i brz početak delo-vanja uz istovremeno izbegavanje visokih lokalnih kon-centracija lekovite supstance, čime se smanjuje rizik za ispoljavanje neželjenih efekata u GIT [14,15]. Moguć-nost upotrebe SMEDDS kao potencijalnih nosača za peroralnu primenu ibuprofena do sada gotovo da nije istraživana. Cilj ovog rada bio je formulacija samo-mik-roemulgujućih nosača na bazi smeše nejonskih surfak-tanata za peroralnu primenu ibuprofena u obliku tvrdih kapsula i in vitro karakterizacija njihove sposobnosti samo-mikroemulgovanja u kiselim i slabo alkalnim vo-denim medijumima, kapaciteta za solubilizaciju inkor-porirane lekovite supstance nakon dispergovanja i in vitro profila oslobađanja lekovite supstance iz nosača.

EKSPERIMENTALNI DEO

Materijal

Kao surfaktanti korišćene su komercijalno dostupne nejonske površinski aktivne supstance PEG-8 kap-rilno/kaprinski gliceridi (Labrasol®, Gattefosse, Fran-cuska) i polioksil-40 hidrogenizovano ricinusovo ulje (Cremophor® RH40, BASF, Nemačka). Labrasol® je označen kao surfaktant, a Cremophor® RH40 kao kosur-faktant. Kao potencijalna uljana faza upotrebljeni su: trigliceridi srednje dužine lanaca (Crodamol® GTCC, Croda, Velika Britanija) i maslinovo ulje (Cropur® Olive, Croda, Velika Britanija). Ibuprofen je dobijen od farma-ceutske kompanije Galenika a.d. (Beograd, Srbija).

Metode

Priprema surfaktant/kosurfaktant/ulje smeša i inkorporiranje ibuprofena

U prvoj fazi istraživanja pripremljene su dve grupe smeša surfaktanta, kosurfaktanta i ulja. Prva grupa (M1-M10) pripremljena je sa srednjelančanim triglice-ridima, a za pripremu druge grupe smeša (O1-O10) ko-rišćeno je maslinovo ulje. Koncentracija ulja kod uzorka M1-M5 i O1-O5 iznosila je 10%, a kod uzoraka M6-M10 i O6-O10 bila je 20%. Uzorci pripremljeni sa istom kon-centracijom ulja razlikovali su se u pogledu masenog odnosa surfaktanta i kosurfaktanta (Km), čija je vred-nost iznosila: 9:1, 7:3, 5:5, 3:7 i 1:9. U tabeli 1 prikazan je sastav pripremljenih smeša surfaktant/kosurfak-tant/ulje.

Tabela 1. Sastav ispitivanih surfaktant/kosurfaktant/ulje smeša (mas.%) Table 1. Composition of the investigated surfactant/cosurfac-tant/oil mixtures (mass%)

Uzorak Surfaktanta Kosurfaktantb Ulje M1 81,0 9,0 10,0c M2 63,0 27,0 10,0c M3 45,0 45,0 10,0c M4 27,0 63,0 10,0c M5 9,0 81,0 10,0c M6 72,0 8,0 20,0c M7 56,0 24,0 20,0c M8 40,0 40,0 20,0c M9 24,0 56,0 20,0c M10 8,0 72,0 20,0c O1 81,0 9,0 10,0d O2 63,0 27,0 10,0d O3 45,0 45,0 10,0d O4 27,0 63,0 10,0d O5 9,0 81,0 10,0d O6 72,0 8,0 20,0d O7 56,0 24,0 20,0d O8 40,0 40,0 20,0 d O9 24,0 56,0 20,0 d O10 8,0 72,0 20,0 d aLabrasol®; bCremophor® RH40; cCrodamol® GTCC; dCropur® Olive

Smeše su pripremljene u staklenim bočicama sa širokim grlom (9 cm×6 cm), tako što su precizno izme-rene potrebne količine sastojaka, koji su zatim izmešani na magnetnoj mešalici, na sobnoj temperaturi. U cilju lakšeg odmeravanja kosurfaktant (Cremophor® RH40) je pre odmeravanja termostatiran tokom 30 min u vodenom kupatilu na temperaturi od 37 °C, tako da mu se konzistencija menja iz polučvrste u tečnu. Priprem-ljenim smešama je dodat ibuprofen, u koncentraciji od


567

10%, uz mešanje na magnetnoj mešalici, na sobnoj tem-peraturi, do potpunog rastvaranja lekovite supstance.

Karakterizacija surfaktant/kosurfaktant/ulje smeša sa ibuprofenom

Karakterizacija surfaktant/kosurfaktant/ulje smeša sa ibuprofenom izvršena je u cilju selekcije potenci-jalnih SMEDDS formulacija i obuhvatila je organolep-tičko ispitivanje uzoraka, ispitivanje sposobnosti disper-govanja u 0,1 M HCl i fosfatnom puferu pH 7,2 (USP). Na osnovu dobijenih rezultata izvršen je izbor poten-cijalnih SMEDDS formulacija kod kojih je određena veli-čina kapi i polidisperzitet nakon dispergovanja i ispitana je in vitro brzina oslobađanja ibuprofena.

Organoleptički pregled Vizuelni pregled pripremljenih surfaktant/kosurfak-

tant/ulje/ibuprofen smeša izvršen je 48 h od izrade, pri čemu su one čuvane na sobnoj temperaturi. Uzorci su posmatrani u cilju detekcije prozirnosti i homogenost ili znakova fizičke nestabilnosti, kao što su zamućenost, raslojavanje i precipitacija lekovite supstance.

Ispitivanje sposobnosti dispergovanja u vodenim medijumima

Sposobnost dispergovanja surfaktant/kosurfaktant/ /ulje smeša sa ibuprofenom ispitivana je u 0,1 M HCl (pH 1,2) i fosfatnom puferu (USP, pH 7,2). U erlenmajer je odmereno 250 ml medijuma i dodat je 1 g ispitivanog uzorka. Uzorci su mućkani tokom 30 min na labora-torijskom šejkeru (IKA KS 260-Basic, Ika, Nemačka), na sobnoj temperaturi, a zatim je izvršen organoleptički pregled u cilju detektovanja zamućenosti, koalescencije kapi uljane faze ili raslojavanja. Kod odabranih homo-genih, bistrih ili opalescentnih uzoraka analizirana je veličina kapi.

Određivanje veličine kapi i polidisperziteta Veličina kapi određivana je tehnikom fotonske kore-

lacione spektroskopije pri čemu je korišćen uređaj Zeta-sizer Nano ZS90 (Malvern Instruments, Velika Britanija), na temperaturi od 20±0,1 °C. Uređaj je opremljen He-Ne laserom koji generiše upadnu koherentnu monohro-matsku svetlost talasne dužine 633 nm. Svetlost rasuta nakon prolaska kroz kivetu sa uzorkom detektuje se pod uglom od 90°. Uređaj je kalibrisan standardnom disperzijom polistirena prečnika 63 nm. Rad uređaja je integrisan sa softverom (Dispersion Technology Soft-ware, DTS) koji operiše algoritmima preko kojih se obezbeđuje optimalno podešavanje optičke konfigura-cije instrumenta za svaki set eksperimentalnih uslova, kontroliše i sprovodi merenje i izvodi analiza podataka i kreira izveštaj o sprovedenom merenju. Dobijeni rezul-tati prikazani su kao prosečna veličina kapi (Z-ave) i indeks polidisperziteta (PdI), za tri uzastopna merenja.

In vitro ispitivanje brzine oslobađanja ibuprofena Ispitivanje brzine oslobađanja ibuprofena sprove-

deno je na uređaju Erweka DT70 (Erweka, Nemačka), uz korišćenje rotirajuće lopatice (USP II aparatura), na temperaturi od 37±0,5 °C. Ispitivane SMEDDS formu-lacije su prethodno napunjene u tvrde želatinske kap-sule veličine 0. U ispitivanju je korišćena količina SMEDDS formulacija koja sadrži 200 mg ibuprofena. Kao medijum je upotrebljen fosfatni pufer pH 7,2 (USP) u količini od 900 ml. Brzina mešanja medijuma lopa-ticom bila je 50 obrt/min. Eksperimentalni uslovi uskla-đeni su sa propisima USP 30-NF 25 [16] za ispitivanje brzine rastvaranja ibuprofena iz oralnih suspenzija i tableta. U definisanim vremenskim intervalima (10, 20, 30, 40, 50 i 60 min) vršeno je uzorkovanje približno 5 ml medijuma i izvršena nadoknada čistim medijumom iste temperature. Koncentracija lekovite supstance u uzor-cima određivana je UV spektrofotometrijski (spektro-fotometar Carry 50, Varian, Namačka) na talasnoj dužini maksimuma apsorpcije (λibuprofen =220 nm). Na isti način ispitana je brzina rastvaranja ibuprofena iz registrovanih lekova u obliku mekih želatinskih kapsula (Rapidol®, PharmaSwiss, Srbija) i obloženih tableta (Brufen®, Galenika, Srbija), sa istom dozom lekovite supstance (200 mg). Dobijeni rezultati predstavljaju srednju vrednost 6 određivanja za svaki uzorak.

REZULTATI I DISKUSIJA

Fizička stabilnost i disperzibilnost surfaktant/kosurfaktant/ulje/ibuprofen smeša

Na tržištu su dostupni različiti farmaceutski eksci-pijensi iz grupe lipida u površinski aktivnih supstanci koji mogu da se upotrebe kao sastojci farmaceutskih preparata za peroralnu primenu, međutim, samo mali broj takvih ekscipijenasa ima fizičko–hemijske osobine koje su pogodne za formulaciju SMEDDS nosača [17]. U ovom istraživanju, sa ciljem da se postigne zadovo-ljavajuća disperzibilnost formulacije, korišćena je kom-binacija dve nejonske površinski aktivne supstance, bez dodavanja hidrofilnih korastvarača. Kao surfaktant upo-trebljen je nejonski tenzid PEG-8 kaprilno/kaprinski gli-ceridi (Labrasol®) koji predstavlja smešu mono-, di- i triglicerida i mono- i di- estara C8-C10 masnih kiselina i polietilenglikola 200-400. To je uljasta tečnost bledo-žute boje. Relativna gustina mu je ∼1,0, a indeks re-frakcije iznosi ∼1,4 na 20 °C. Može da sadrži slobodne (neesterifikovane) makrogole. Ovaj ekscipijens je u skladu sa zahtevima koji su navedeni u monografiji kaprilno/kaprinskih makrogolglicerida (Ph. Eur. 8.0) u pogledu hemijskog sastava, fizičko-hemijskih osobina i sadržaja etilenoksida (<1 ppm) i dioksana (<10 ppm). Uvršten je u FDA IIG listu ekscipijenasa (eng. Food and Drug Administration Inactive Ingredient Guide, FDA


568

IIG)* za upotrebu u oralnim rastvorima i kapsulama. Ima visoku HLB vrednost (∼14) i upotrebljava se kao efikasan solubilizator teško rastvorljivih lekovitih sup-stanci. U dosadašnjim istraživanjima pokazano je da Labrasol® u kombinaciji sa drugim nejonskim surfak-tantima, može da obrazuje biokompatibilne mikroemul-zije, što povećava njihov farmaceutski značaj [18–23]. Kao kosurfaktant korišćen je nejonski tenzid PEG-40 hidrogenizovano ricinusovo ulje (Cremophor® RH40). Dobija se etoksilacijom hidrogenizovanog ricinusovog ulja i dominantno se sastoji do polietoksilovanih hidro-genizovanih triglicerida ricinolne kiseline sa prosečno 40 mol etilenoksida. HLB vrednost ovog tenzida iznosi 14–16. Cremophor® RH40 je u dosadašnjim istraži-vanjima korišćen kao surfaktant ili kosurfaktant za sta-bilizaciju nejonskih mikroemulzija koje su ispitivane kao nosači za peroralnu primenu lekovitih supstanci [24,25].Ulja upotrebljena u ovom radu bili su srednjelančani trigliceridi (Crodamol® GTCC) i maslinovo ulje (Cropur® Olive). Fizičko–hemijske osobine korišćenih ulja navedene su u tabeli 2.

U prvoj fazi istraživanja pripremljene su smeše surfaktant/kosurfaktant/ulje M1–M10, sa srednjelan-čanim trigliceridima, i O1–O10, sa maslinovim uljem, bez lekovite supstance. Na sobnoj temperaturi, smeše su bile homogene, bistre tečnosti ili homogeni, opales-centni ili zamućeni polučvrsti sistemi. Promene u izgle-du i konzistenciji dovedene su u vezu sa smanjenjem Km vrednosti, odnosno sa povećanjem udela kosurfak-tanta. Kosurfaktant Cremophor® RH40, za razliku od

* FDA IIG navodi supstance koje su odobrene za upotrebu kao po-moćne materije u farmaceutskim preparatima na tržištu SAD.

ostalih sastojaka u smeši koji predstavljaju bistre teč-nosti, ima neproziran izgled i polučvrstu konzistenciju na sobnoj temperaturi. Nakon termostatiranja u vode-nom kupatilu na 37 °С, sve smeše su bile homogene, bistre tečnosti. Ibuprofen je rastvoren u smešama u koncentraciji od 10 mas.%. Svi uzorci sa ibuprofenom bili su homogeni na sobnoj temperaturi (kao što je ilus-trovano kod grupe uzoraka M1–M5 pripremljenih sa srednjelančanim trigliceridima, na slici 1a), odnosno, homogene, bistre tečnosti na 37 °С (slika 1b).

Inkorporiranje ibuprofena nije uticalo na izgled i fizičku stabilnost ispitivanih formulacija M1–M10 i O1– –O10, a takođe nisu uočene promene u izgledu ili pre-cipitacija lekovite supstance iz nosača tokom 12 meseci čuvanja na sobnoj temperaturi.

Sposobnost samo-mikroemulgovanja surfaktant/kosurfaktant/ulje formulacija sa ibuprofenom

Ispitan je uticaj vrste i koncentracije ulja i Km vred-nosti na sposobnost dispergovanja i formiranja u/v mik-

roemulzije iz M1–M10 i O1–O10 formulacija sa ibu-profenom u vodenim medijumima. Kriterijumi koji su korišćeni pri selekciji potencijalnih SMEDDS bili su Z-ave ≤ ≤ 100 nm i PdI ≤ 0,250 (monodisperzni uzorci), nakon dispergovanja u kiselom i alkalnom medijumu.

Disperzije formulacija O1–O10 u kiselom medijumu (0,1M HCl (pH 1,2)) imale su opalescentan ili zamućen izgled, pri čemu je kod formulacija O6–O10, koje sadrže 20% maslinovog ulja, uočeno i izdvajanje ulja na povr-šini disperzije. Izgled disperzija formulacija O1–O10 sa maslinovim uljem, u fosfatnom puferu pH 7,2 prikazan je na slici 2.

Tabela 2. Hemijski naziv, izgled, relativna gustina (d), indeks refrakcije (n) i viskozitet (η) korišćenih ulja

Komercijalni naziv i proizvođač Crodamol® GTCC (Croda, Velika Britanija) Cropure® Olive (Croda, Velika Britanija) Hemijski naziv Trigliceridi, srednje dužine lanacaa;

srednje-lančani trigliceridib Maslinovo ulje, rafinisanoa;

maslinovo uljeb Izgled Bistra, bezbojna ili slabo žućkasta tečnost bez mirisa Bistra tečnost karakterističnog mirisa n 1,4500 (t = 20 °C) 1,4662 (t = 15 °C) η / mPa·s 30 (t = 20 °C) 63,28 ( t = 25 °C ) aPh. Eur. 8.0; bUSP 30/NF 25

a) M1 M2 M3 M4 M5 b) M1 M2 M5M4 M3 Slika 1. Izgled formulacija M1–М5 koji su pripremljeni sa srednjelančanim trigliceridima i sadrže 10% ibuprofena: a) na sobnoj temperaturi i b) na 37 °C. Figure 1. The appearance of the formulations M1–М5 prepared with medium chain triglycerides (oil phase) and loaded with ibuprofen (10 mass%): a) at room temperature; b) at 37 °C.


569

Disperzije formulacija O1–O5, koje su pripremljene sa 10% maslinovog ulja, bile su homogene, bistre ili opalescentne (slika 2a), dok su disperzije formulacija O6–O10, gde je koncentracija maslinovog ulja iznosila 20%, imale opalescentan ili zamućen izgled, a uočeno je i izdvajanje uljane faze na površini disperzije (slika 2b). Kod nekih uzoraka iz grupe O6–O10 uočeno je i izdva-janje lekovite supstance u vidu taloga. Uzorci u kojima je uočeno raslojavanje, u kiselom i/ili alkalnom medi-jumu, i/ili precipitacija aktivne supstance isključeni su kao potencijalni SMEDDS. Analiza veličine kapi kod homogenih disperzija pokazala je da je samo u disperziji uzorka O5 (10% maslinovog ulja, Km 1:9) u alkalnom medijumu, prosečna veličina kapi bila manja od 100 nm, međutim, prosečna veličina kapi u kiselom medi-jumu bila je veća od 100 nm i indeks polidisperziteta PdI > 0,250, pa je ovaj uzorak isključen iz daljeg raz-matranja kao potencijalni SMEDDS. Kod ostalih disper-

zija prosečna veličina kapi bila je >> 100 nm i neujed-načena (PdI > 0,250), pa nisu razmatrani u nastavku istraživanja.

Posle dispergovanja formulacija M1–M5, koje su pripremljene sa 10% srednjelančanih triglicerida, u kise-lom i alkalnom medijumu, formirane su transparentne disperzije, osim kod formulacije M1, gde je disperzija bila opalescentna (slika 3).

Vrednosti Z-ave i PdI disperzija dobijenih iz formu-lacija M1–M5, u kiselom i alkalnom medijumu, prika-zane su u tabeli 3.

Prosečna veličina kapi kod disperzija u kiselom medijumu kretala se u rasponu od 20,1–96,0 nm, a ras-podela veličina je bila uska (PdI < 0,250) i ukazivala na monodisperzne uzorke. Dobijeni rezultati pokazali su da su se u kiselom medijumu iz formulacija M1–M5 formirali mikroemulzioni nosači pri svim Km vrednos-tima. U alkalnom medijumu iz uzoraka M2–M4 takođe

a) O1 O2 O2 O3 O4

b) O6 O7 O8 O9 O10

Slika 2. a) Izgled formulacija sa ibuprofenom M1–M5 (sadrže 10% maslinovog ulja) i b) M6–M10 (sadrže 20% maslinovog ulja), nakon dispergovanja u fosfatnom puferu, pH 7,2. Figure 2. a) The appearance of the ibuprofen loaded formulations M1–М5 (prepared with 10% of olive oil) and b) M6–M10 (prepared with 20% of olive oil), dispersed in phosphate buffer (pH 7.2).

a) M1 M2 M3 M4 M5

b) M1 M2 M3 M4 M5

Slika 3. Izgled formulacija sa ibuprofenom M1–M5 (sadrže 10% srednjelančanih triglicerida) nakon dispergovanja u: a) 0,1 M HCl (pH 1,2); b) fosfatnom puferu, pH 7,2. Figure 3. The appearance of the ibuprofen loaded formulations M1–М5 (prepared with 10% of medium chain triglycerides) dispersed in: a) 0.1M HCl (pH 1.2); b) phosphate buffer (pH 7.2).

Tabela 3. Prosečna veličina kapi (Z-ave) i indeks polidisperziteta (PdI) Labrasol®/Cremophor® RH40/Crodamol® GTCC formulacija sa 10 mas.% ibuprofena, nakon dispergovanja u različitim medijumima (1 g/250 ml) Table 3. Average droplet size (Z-ave) and polydispersity index (PdI) of Labrasol®/Cremophor® RH40/Crodamol® GTCC formulations with ibuprofen (10 mass%), after dispersion in different media (1 g/250 ml)

Uzorak Medijum

0,1 M HCl (pH 1,2) Fosfatni pufer, pH 7,2 (USP) Z-ave / nm PdI Z-ave / nm PdI

M1 96,01±1,89 0,248±0,024 911,1±15,6 0,198±0,003 M2 28,60±0,25 0,068±0,008 22,83±0,06 0,075±0,017 M3 20,09±0,15 0,034±0,012 16,80±0,06 0,040±0,017 M4 19,42±0,10 0,233±0,003 14,50±0,03 0,070±0,025 M5 55,76±0,54 0,243±0,007 15,90±0,42 0,188±0,035


570

su formirane disperzije sa kapima čija je prosečna veli-čina bila u rasponu od 22,8–14,5 nm, međutim, kod uzorka M1 (Km 9:1) veličina kapi bila je >> 100 nm (tabela 3). Uočeno je da je pri veličini kapi približno 100 nm i većoj, izgled disperzija bio opalescentan, dok su disperzije sa kapima sitnijim od 100 nm bile trans-parentne. Razlike u izgledu uzoraka su očekivane, bu-dući da se sa smanjenjem veličine kapi smanjuje inten-zitet rasipanja vidljive svetlosti. Kod uzorka М1, kod koga preovladava Labrasol® u smeši sa kosurfaktantom (Km 9:1), kao i kod uzorka М5, gde preovladava Cre-mophor® RH40 u smeši sa surfaktantom (Km 1:9), uočen je veći uticaj pH vrednosti na veličinu kapi disperzija (tabela 3). Iz formulacije M1 u kiseloj sredini se obra-zuje mikroemulzija, dok se u alkalnoj sredini formira emulzija sa kapima prosečne veličine oko 900 nm. Dis-pegovanjem formulacije M5 u kiselom medijumu formi-rana je u/v mikroemulzija sa prosečnom veličinom kapi 55,8 nm, dok je u alkalnom medijumu prosečna veličina kapi bila oko 3 puta manja. Generalno, i u disperzijama uzoraka M2, M3 i M4, veličina kapi u alkalnom medi-jumu je nešto manja u poređenju sa kiselim mediju-mom. Ova zapažanja navela su na pretpostavku da je ibuprofen, zbog jako ograničene rastvorljivosti u kise-lom medijumu uglavnom solubilizovan u međupovr-šinskom filmu, dok u alkalnoj sredini ima veću rastvor-ljivost u vodenom medijumu, pa je samo delimično solubilizovan u kapima mikroemulzionog nosača, zbog čega je njihov hidrodinamički prečnik manji. U disper-zijama formulacija M2, M3 i М4, kod kojih su Km vred-nosti iznosile 7:3, 5:5, odnosno, 3:7, uočeno je formi-ranje mikroemulzija sa sitnim kapima čija je prosečna veličina manja od 30 nm. Može se pretpostaviti da se pri ovim Km vrednostima uspostavlja sinergizam između surfaktanta i kosurfaktanta, na kome se zasniva veća sposobnost samo-mikroemulgovanja srednjelančanih triglicerida, zbog relativno male molarne zapremine ovog ulja, kao i slabiji uticaj pH vrednosti na proces dispergovanja i stabilnost formirane disperzije, u pore-đenju sa uzorcima M1 i M5. Dobijeni rezultati bili su u saglasnosti sa zapažanjem iz studija faznog ponašanja smeša Labrasol®/Cremophor® RH40 sa uljima male molarne zapremine (srednjelančani trigliceridi i izopro-pilmiristat), da se maksimalni sinergizam za solubili-zaciju vodene faze (prečišćena voda) i formiranje v/u mikroemulzija ispoljava pri približno ujednačenom masenom odnosu ova dva tenzida, odnosno, u rasponu Km od 4:6 do 6:4 [20–22].

Disperzije uzoraka pripremljenih sa 20 % srednjelan-čanih triglicerida (M6–M10), bile su opalescentne ili zamućene, u kiselom i alkalnom medijumu. Analizom veličine kapi kod pripremljenih disperzija utvrđeno je da je Z-ave ≥ 100 nm i PdI > 0,250, pa na osnovu toga nisu bili zadovoljeni kriterijumi koji ukazuju na formi-

ranje mikroemulzija, tj. monodisperznih sistema sa veli-činom kapi ispod 100 nm.

Na osnovu dobijenih rezultata utvrđeno je da masli-novo ulje nije bilo pogodno za formulaciju SMEDDS sa ibuprofenom, verovatno zbog toga što se pretežno sastoji od triglicerida masnih kiselina dugog lanca, dominantno oleinske kiseline, koji imaju relativno veli-ku molarnu zapreminu i nisu pogodni za samo-mikro-emulgovanje. U sastavu srednjelančanih triglicerida preovladavaju C8–C12 masne kiseline, pa je njihova molarna zapremina manja u poređenju sa sastojcima maslinovog ulja, a viskozitet niži i lakše podležu procesu formiranja kapi nanometarskih dimenzija (< 100 nm) u prisustvu upotrebljene smeše nejonskih surfaktanata. Za formiranje mikroemulzija bila je značajna i koncen-tracija upotrebljenog ulja, pa je uočeno da je proces bio otežan pri koncentraciji od 20% srednjelančanih trigli-cerida, što je ukazivalo da ukupna koncentracija smeše surfaktanta nije bila dovoljna da omogući proces samo-mikroemulgovanja. Na osnovu rezultata sprovedene karakterizacije u celini, kao nosači SMEDDS tipa ozna-čeni su uzorci M1–M5 koji sadrže 10% srednjelančanih triglicerida kao uljanu fazu.

Kinetika in vitro oslobađanja ibuprofena

Profili oslobađanja ibuprofena iz SMEDDS nosača M1–M5, u obliku tvrdih želatinskih kapsula, obloženih tableta (Brufen®) i mekih kapsula (Rapidol®), prikazani su na slici 4.

Rastvaranje/oslobađanje* ibuprofena iz ispitivanih komercijalnih uzoraka kao i iz uzoraka M1 i M5 bilo je u skladu sa zahtevom američke farmakopeje (USP 30–NF 25), odnosno, posle 60 min bilo je oslobođeno/ras-tvoreno više od 80% supstance. Profili oslobađanja ibuprofena iz uzoraka M1 i M5 su se međusobno razli-kovali. Oslobađanje ibuprofena iz uzorka М1 bilo je kompletno već posle 10 min, dok je oslobađanje iz komercijalnih uzoraka i uzorka M5 bilo sporije, tako da je posle 10 min oslobođeno ∼55 (Brufen®), ∼65 (Rapidol®), odnosno, ∼30 % (M5) supstance. Nasuprot tome, oslobađanje ibuprofena iz uzoraka M2, M3 i M4 bilo je znatno sporije, tako da se za 60 min oslobodilo ∼70 % supstance.

Dobijeni rezultati su dovedeni u vezu sa rezultatima i pretpostavkama koje su postavljene prilikom karak-terizacije sposobnosti samo-mikroemulgovanja nosača. Mada se u kiseloj sredini iz uzorka M1 formira u/v mik-roemulzija, u alkalnom medijumu koji je upotrebljen za ispitivanje brzine rastvaranja/oslobađanja, obrazuje se emulzija koja ima manji kapacitet za solubilizaciju ibu-

* Obložene tablete Brufen® sadrže aktivnu supstancu u čvrstom obliku, pa se tokom ovog ispitivanja odigrava njeno rastvaranje, dok meke kapsule Rapidol® sadrže ibuprofen rastvoren u polietilenglikolu 400, pa se u toku testa prati njeno oslobađanje iz farmaceutskog oblika.


571

profena, pa se on brzo oslobađa i rastvara u medijumu. Sporije oslobađanje iz nosača M2, M3 i M4 dovedeno je u vezu sa formiranjem u/v mikroemulzije sa sitnim kapima i očuvanjem njene stabilnosti i strukture u alkal-nom medijumu, zbog uspostavljanja snažnijeg siner-gizma između surfaktanta i kosurfaktanta u stabilizaciji međupovršinskog filma. Time je verovatno ograničeno oslobađanje molekula lekovite supstance koji su loci-rani unutar ovog filma. Na sličan način se može objas-niti i proces oslobađanja ibuprofena iz uzorka M5, međutim, verovatno zbog slabijeg sinergizma između surfaktanta i kosurfaktanta, pri Km 1:9, oslobađanje supstance iz ovog nosača je slabije ograničeno u od-nosu na nosače M2, M3 i M4. Razlike u profilima oslobađanja ibuprofena iz uzoraka M1 i M5 dovedene su u vezu sa različitom solubilizacionom moći surf-aktanta i kosurfaktanta. Cremophor® RH 40 je vero-vatno „snažniji“ solubilizator u sistemu ulje/voda, u poređenju sa Labrasol®, jer ima duže ugljovodonične lance u hidrofilnom i lipofilnom delu molekula, a time i veću sposobnost da solubilizuje ulje u vodenoj fazi [26,27]. Zbog toga je u disperziji uzorka M5, koji sadrži višestruko više Cremophor® RH 40 (Km 1:9) struktura u/v mikroemulzije očuvana, dok se stabilnost mikro-emulzije narušava nakon dispergovanja uzorka M1, gde dominira Labrasol® (Km 9:1).

ZAKLJUČAK

U radu je izvršena formulacija i in vitro karakteri-zacija SMEDDS koji sadrže Labrasol®, Cremophor® RH

40, Crodamol® GTCC (10%) i ibuprofen (10%), M1–M5. Pripremljeni uzorci imali su zadovoljavajuću fizičku sta-bilnost tokom 12 meseci čuvanja na sobnoj tempera-turi. Ispitivani SMEDDS imali su zadovoljavajući kapa-citet za solubilizaciju inkorporiranog ibuprofena nakon dispergovanja u kiseloj i slabo alkalnoj sredini. Rezultati in vitro karakterizacije pokazali su da sposobnost samo-mikroemulgovanja u kiseloj i alkalnoj sredini, veličina kapi formirane u/v mikroemulzije i brzina oslobađanja solubilizovane lekovite supstance značajno zavise od masenog odnosa surfaktanta i kosurfaktanta (Km). Oslo-bađanje ibuprofena bilo je u skladu sa zahtevom USP 30-NF 25 (najmanje 80% oslobođene supstance nakon 60 min), iz uzoraka М1 (Km 9:1) i M5 (Km 1:9), pri čemu je celokupna količina oslobođena već posle 10 min iz nosača M1, dok je oslobađanje bilo znatno sporije iz nosača M5 (∼30%), kao i iz komercijalnih uzoraka (oblo-žene tablete Brufen® (∼55%) i meke želatinske kapsule Rapidol® (∼65%), koji su ispitani pod istim uslovima. Dobijeni rezultati ukazali su da uzorak M1 predstavlja potencijalni SMEDDS koji u kiseloj sredini želuca može da formira u/v mikroemulziju i obezbedi efikasnu solu-bilizaciju teško rastvorljive supstance, a nakon prelaska u tanko crevo, ona se brzo oslobađa iz formiranog nosača i dostupna je za apsorpciju.

Zahvalnica

Istraživanje je realizovano u okviru projekata III 46010 i TR 34007, koje finansira Ministarstvo prosvete, nauke i tehnološkog razvoja Republike Srbije.

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60

M1

M2

M3

M4

M5

Brufen 200mg

Rapidol

Kol

ičin

a os

lob

ođen

og ib

up

rofe

na

(%)

Vreme (min)

Slika 4. Profil oslobađanja ibuprofena iz SMEDDS formulacija М1–М5 (u obliku tvrdih želatinskih kapsula) i registrovanih obloženih tableta (Brufen®) i mekih želatinskih kapsula (Rapidol®). Figure 4. Ibuprofen release profile from SMEDDS M1-M5 (in hard gelatin capsules) and marketed coated tablets (Brufen®) and soft gelatin capsules (Rapidol®).


572

LITERATURA

[1] C.W. Pouton, Formulation of poorly water-soluble drugs for oral administration: physicochemical and physio-logical issues and the lipid formulation classification system, Eur. J. Pharm. Sci. 29 (2006) 278–287.

[2] M. Kahlweit, Microemulsions, Annu. Rep. Prog. Chem., C 95 (1999) 89–115.

[3] J. Sjöblom, R. Lindbergh, S.E. Friberg, Microemulsions - phase equilibria characterization, structures, appli-cations and chemical reactions, Adv. Colloid Interfac. 65 (1996) 125–287.

[4] R. Strey, Microemulsion microstructure and interfacial curvature, Colloid Polym. Sci. 272 (1994) 1005–1019.

[5] C.J.H Porter, C.W. Pouton, J.F. Cuine, W.N. Charman, Enhancing intestinal drug solubilisation using lipid-based delivery systems, Adv. Drug Delivery Rev. 60 (2008) 673–691.

[6] J.A. Yáñez, S.W.J. Wang, I.W. Knemeyer, M.A. Wirth, K.B. Alton, Intestinal lymphatic transport for drug delivery, Adv. Drug Delivery Rev. 63 (2011) 923–942.

[7] R. Holm, A. Müllertz, H. Mu, Bile salts and their importance for drug absorption, Int. J. Pharm. (2013), doi: 10.1016/j.ijpharm.2013.04.003.

[8] K. Kawakami, Modification of physicochemical charac-teristics of active pharmaceutical ingredients and appli-cation of supersaturatable dosage forms for improving bioavailability of poorly absorbed drugs, Adv. Drug Deli-very Rev. 64 (2012) 480–495.

[9] Nacionalni registar lekova, NRL 2012. Agencija za lekove i medicinska sredstva Srbije, Beograd, 2012.

[10] European Pharmacopoeia, eighth ed., Council of Europe, Strasbourg, 2010.

[11] M. Meloun, S. Bordovská, L. Galla, The thermodynamic dissociation constants of four non-steroidal anti-inflam-matory drugs by the least-squares nonlinear regression of multiwavelength spectrophotometric pH-titration data, J. Pharmaceut. Biomed. 45 (2007) 552–564.

[12] H. Potthast, J.B. Dressman, H.E. Junginger, K.K. Midha, H. Oeser, V.P. Shah, H. Vogelpoel, D.M. Barends, Bio-waiver Monographs for Immediate Release Solid Oral Dosage Forms: Ibuprofen, J. Pharm. Sci. 94 (2005) 2121– –2131.

[13] B.A. Khan, S. Bakhsh, H. Khan, T. Mahmood, A. Rasul, Basics of Self Micro Emulsifying Drug Delivery System, Journal of Pharmacy and Alternative Medicine, 2012 (dostupno na: www.iiste.org).

[14] H. Araya, S. Nagao, M. Tomita, M. Hayashi. The novel formulation design of self-emulsifying drug delivery systems (SEDDS) type O/W microemulsion I: enhancing effects on oral bioavailability of poorly water soluble compounds in rats and beagle dogs, Drug. Metab. Pharmacokinet. 20 (2005) 244–256.

[15] B.B. Subudhi, S. Mandal, Self-Microemulsifying Drug Delivery System: Formulation and Study Intestinal Per-meability of Ibuprofen in Rats, J. Pharm. (2013), Article ID 328769 (http://dx.doi.org/10.1155/2013/328769).

[16] The United States Pharmacopoeia, 30/National Formul-ary 25, Pharmacopoeia Convention Inc., Rockville, MD, 2007.

[17] A. Sprunk, C.J. Strachan, A. Graf, Rational formulation development and in vitro assessment of SMEDDS for oral delivery of poorly water soluble drugs, Eur. J. Pharm. Sci. 46 (2012) 508–515.

[18] L. Djordjevic, M. Primorac, M. Stupar, D. Krajisnik, Char-acterization of caprylocaproyl macrogolglycerides based microemulsion drug delivery vehicles for an amphiphilic drug, Int. J. Pharm. 271 (2004) 11–19.

[19] L. Djordjevic, M. Primorac, M. Stupar, In vitro release od diclofenac diethylamine from caprylocaproyl macrogol-glycerides based microemulsions, Int. J. Pharm. 296 (2005) 73-79.

[20] L. Djekic, M. Primorac, S. Filipic, D. Agbaba, Investigation of surfactant/cosurfactant synergism impact on ibupro-fen solubilization capacity and drug release character-istics of nonionic microemulsions, Int. J. Pharm. 433 (2012) 25–33.

[21] L. Djekic, M. Primorac, J. Jockovic, Phase behaviour, microstructure and ibuprofen solubilization capacity of pseudo-ternary nonionic microemulsions, J. Mol. Liq. 160 (2011) 81–87.

[22] L. Djekic, M. Primorac, The influence of cosurfactants and oils on the formation of pharmaceutical micro-emulsions based on PEG-8 caprylic/capric glycerides, Int. J. Pharm. 352 (2008) 231–239.

[23] M. Milovic, J. Djuris, L. Djekic, D. Vasiljevic, S. Ibric, Characterization and evaluation of solid self-microemul-sifying drug delivery systems with porous carriers as systems for improved carbamazepine release, Int. J. Pharm. 436 (2012) 58–65.

[24] S. Agatonovic-Kustrin, B.D. Glass, M.H. Wisch, R.G. Alany, Prediction of stable microemulsion formulation for the oral delivery of a combination of antitubercular drugs using ANN methodology, Pharm. Res. 20 (2003) 1760–1764.

[25] H. Arya, M. Tomita, M. Hayahashi, The novel formul-ation design of O/W microemulsion for improving the gastrointestinal absorption of poorly water-soluble com-pounds, Int. J. Pharm. 305 (2005) 61–74.

[26] J. Chlebicki, P. Majtyka, Effect of Oxypropylene Chain Length on the Surface Properties of Dialkyl Glycerol Ether Nonionic Surfactants, J. Colloid Interface Sci. 220 (1999) 57–62.

[27] Y. Bayrak, M. Iscan, Studies on the phase behavior of the system non-ionic surfactant/alcohol/alkane/H2O, Colloids Surf. Physicochem. Eng. Aspects 268 (2005) 99– –103.


573

SUMMARY FORMULATION AND CHARACTERISATION OF SELF-MICROEMULSIFYING DRUG DELIVERY SYSTEMS BASED ON BIOCOMPATIBLE NONIONIC SURFACTANTS Ljiljana M. Djekic, Marija M. Primorac

University of Belgrade, Faculty of Pharmacy, Department of Pharmaceutical Technology and Cosmetology, Belgrade, Serbia

(Scientific paper)

Development of self-dispersing drug delivery systems (SMEDDS) is a modernstrategy for oral delivery improvement of poorly soluble drugs. Self-microemul-sifying drug delivery systems (SMEDDS) are isotropic mixtures of oils and hydro-philic surfactants, which form oil-in-water (o/w) microemulsions by dilution inaqueous media (e.g., gastrointestinal fluids). Formulation of SMEDDS carriersrequires consideration of a large number of formulation parameters and theirinfluences on process of self-microemulsifying and releasing of drug. The aim of this work was formulation and characterization of SMEDDS for oral administrationof ibuprofen. In the experimental work, two series of potential SMEDDS wereprepared (M1–M10), using surfactant (Labrasol®, Gattefosse), cosurfactant (PEG--40 hydrogenated castor (Cremophor® RH40), and oil (medium chain triglycerides(Crodamol® GTCC) and olive oil (Cropur® Olive)), at surfactant-to-cosurfactant mass ratios (Km) 9:1, 7:3, 5:5, 3:7, and 1:9, and 10 or 20% of the oil phase. Ibu-profen was dissolved in formulations in concentration of 10%. Characterization of the investigated formulations included evaluation of physical stability, self-micro-emulsification ability in 0.1 M HCl (pH 1.2) and phosphate buffer pH 7.2 (USP) andin vitro drug release. Formation of o/w microemulsions with the average dropletsize (Z-ave) up to 100 nm, was observed in dispersions of formulations preparedwith 10 mass% of medium chain triglycerides, within the entire investigated rangeof the Km values (M1–M5). These formulations were selected as SMEDDS. Resultsof characterization pointed out the importance of the type and concentration ofthe oil as well as the Km value for the self-microemulsing ability, as well as drug release kinetics from the investigated SMEDDS. Ibuprofen release was in accord-ance with the request of USP 30-NF 25 (at least 80% after 60 min) from theformulations M1 (Km 9:1) and M5 (Km 1:9). Furthermore, the ibuprofen release was completed after 10 min from formulation M1, while the release from thecarrier M5 (∼30%) as well as from the commercial tablets Brufen® (∼55%) and soft capsules Rapidol® (∼65%), examined under the same conditions, was significantly slower. The present study revealed that the formulation M1 represents a poten-tial SMEDDS which efficiently dissolves ibuprofen in acidic media, with potentialto minimize the side effects, while on introduction into alkaline intestinal envi-ronment, the drug may rapidly release from the carrier and undergo absorption.

Keywords: Self-microemulsifying drug delivery systems (SMEDDS) • Microemul-sions • Labrasol® • Cremophor® RH40 •Ibuprofen

575

Bioleaching of pollymetallic sulphide concentrate using thermophilic bacteria

Milovan Vuković1, Nada Štrbac1, Miroslav Sokić2, Vesna Grekulović1, Vladimir Cvetkovski3 1University of Belgrade, Technical Faculty in Bor, Bor, Serbia 2Institute for Technology of Nuclear and Other Mineral Raw Materials, Belgrade, Serbia 3Mining and Metallurgy Institute, Bor, Serbia

Abstract An extreme thermophilic, iron-sulphur oxidising bacterial culture was isolated and adaptedto tolerate high metal and solids concentrations at 70 °C. Following isolation and adapt-ation, the culture was used in a batch bioleach test employing a 5-l glass standard magne-tic agitated and aerated reactor, for the bioleaching of a copper-lead-zinc collective concentrate. The culture exhibited stable leach performance over the period of leachoperation and overall copper and zinc extractions higher than 97%. Lead sulphide is trans-formed into lead sulphate remaining in the bioleach residue due to the low solubility in sulphate media. Brine leaching of bioleach residue yields 95% lead extraction.

Keywords: bioleaching, bacteria, chalcopyrite, sphalerite, galena, thermophiles.

SCIENTIFIC PAPER

UDC 502/504:602:66.02/.09:54

Hem. Ind. 68 (5) 575–583 (2014)

doi: 10.2298/HEMIND130905087V


Biohydrometallurgy, biotechnology or bioproces-sing, is a new approach used for extraction of metals; it offers the opportunity to reduce environmental pollu-tion. Namely, biological processes are conducted under mild conditions, usually without the addition of toxic substances. Also, the products of these processes end up in an aqueous solution which is more amenable to containment and treatment than gaseous waste. Bio-technological approach consists of four distinct tech-niques: bioremediation, biosorption, bioaccumulation and bioleaching.

The bioleaching technology involves the extraction of useful elements from ores by bacteria and solution. The whole process consists of six steps: outer diffusion, inner diffusion, leaching reaction, process of precipi-tation and hydrolysis of the element, transportation of microorganisms, heating and cooling of the bed [1]. Bioleaching is used extensively to recover copper and uranium from low grade ores [2–5].

The bacteria most frequently used in bioleaching are of two types: Chemolithotrophic and Heterotrophic. Acidithiobacillus ferrooxidans is a chemolithotrophic bacterium capable of utilizing ferrous iron as the only source of energy for its growth. This bacterium oxidizes Fe(II) and elemental sulphur to Fe(III) and H2SO4, res-pectively, at low acidic conditions as well as high metal ion concentrations. Due to its capacity to oxidize metal sulphides, low concentrations of metals in ores are not a problem for these bacteria because they simply neg-

Correspondence: M. Vuković, Technical Faculty in Bor, Vojske Jugo-slavije 12, 19210 Bor, Serbia. E-mail: [email protected] Paper received: 5 September, 2013 Paper accepted: 26 November, 2013

lect the waste which surrounds the metals, attaining extraction yields of over 90% in some cases [6]. In short, acidolysis is the principle mechanism of the bio-leaching process. Various acids produced by microorg-anism such as citric, oxalic and sulphuric help in the metal dissolution process from ores. Bacterial micro-organisms, in fact, gain energy by breaking down ores into their constituent elements.

The bioleaching process is commercially used to process minerals of copper, nickel, cobalt, zinc, lead and uranium. The applicability of this process is not limited to heap or dump leaching, but has been extended to reactor bioleaching [7,8]. Most secondary copper sulphides can be bioleached successfully at 35– –45 °C, using mesophilic Thiobacillus-Leptospirillum bacterial cultures. However, the bioleaching of chal-copyrite, the major copper-bearing sulphide mineral of commercial interest, is still a major challenge. This is due to relatively slow kinetics and poor extractions, which have mainly been attributed to passivation [9].

The interest in extreme thermophilic bacteria lies in the potential for improving the leach rates of sulphides such as chalcopyrite, pyrite and arsenopyrite. The pot-ential to achieve this is clear from the extensive pub-lished data that has been obtained using shake flask and other laboratory-scale methods, but employing low solids concentrations and relatively mild agitation con-ditions. However, it has been claimed that the potential for using extreme thermophiles, such as Sulfolobus, at temperatures in the range 60–84 °C, may not be real-ized commercially unless their sensitivity to agitation at high solids concentrations can be overcome [10]. Solu-tions suggested by these authors include the selection or isolation of more robust bacteria or the develop-ment of improved reactor designs.

M. VUKOVIĆ et al.: BIOLEACHING OF POLLYMETALLIC SULPHIDE CONCENTRATE Hem. ind. 68 (5) 575–583 (2014)

576

Numerous studies indicate that chalcopyrite leach-ing depends on redox potential, and occurs between 510 to 610 mV at 70 °C. Metal sulphides, such as chalcocite, covellite, sphalerite and galena generally leach far more easily. However, even with these metal sulphides, extended residence times are often required to achieve high metal extractions, using mesophilic Thiobacillus-Leptospirillum bacteria operating at tem-peratures in the range of 35–45 °C.

The following equations describe the direct and indirect mechanism for the bio-oxidation of Cu–Zn–Pb– –Fe sulphides. In the direct mechanism, metal sul-phides can be directly oxidised by extreme thermo-philes or thiobacillus strains to soluble metals sulphates according to Eqs. (1)–(4):

Pyrite: 4FeS2 + 15O2 + 2H2O → 2Fe2(SO4)3 + 2H2SO4 (1) Chalcopyrite: 4CuFeS2 + 17O2 + 2H2SO4 → → 4CuSO4 + 2Fe2(SO4)3 + 2H2O (2) Sphalerite: ZnS + 2O2 → ZnSO4 (3) Galena: PbS + 2O2 → PbSO4 (4)

The direct mechanism means that bacteria attach onto the mineral particle surfaces, where they facilitate mineral dissolution through direct bacterial meta-bolism. In the indirect mechanism, however, the oxi-dation of sulphide minerals is represented by the oxi-dation of sulphide minerals by ferric ions, with bacteria oxidising ferrous ions to ferric ions, and elemental sul-phur to sulphate ions, according to the reactions (5–8):

Chalcopyrite: CuFeS2 + 2Fe2(SO4)3 → CuSO4 + 5FeSO4 + 2S° (5) Sphalerite: ZnS + Fe2(SO4)3 → ZnSO4 + 2FeSO4 + S° (6) Galena: PbS + Fe2(SO4)3 → PbSO4 + 2FeSO4 + S° (7) Pyrite: FeS2 + Fe2(SO4)3 → 3FeSO4 + 2S° (8)

Therefore, the metal dissolution occurs by a cyclic process between reactions (9) and (10), and the for-mation of H+ during the sulphur biooxidation (11) enhances the overall efficiency:

4FeSO4 + 2H2SO4 + O2 → 2Fe2 (SO4)3 + 2H2O (9) (Cu, Zn, Pb, Fe)S + Fe2(SO4)3 → (Cu, Zn, Pb, Fe)SO4 + 2FeSO4 + S° (10) 2S° + 3O2 + 2H2O → 4H+ + 2SO4

2- (11)

Besides these two bioleaching mechanisms, a third mechanism has been proposed, the indirect contact, under which bacterial microorganisms attach onto the mineral particle surfaces, where they increase the ferric ion and acid concentrations in the immediate vicinity of the mineral particles. At present, it is believed that the leaching of all sulphides of the form M2+S2– proceeds according to an indirect mechanisms.

Since the bioleaching reactions imply intensive mass transfer requirements for oxygen and carbon dioxide, there is a need to clearly define the process require-

ments to maximise the reaction kinetics [11]. This may be more critical in terms of significantly reduced solu-bilities of these gases at the operating temperatures required for the extreme thermophiles. Compared to mesophilic and moderately thermophilic bioleaching bacteria, the extreme thermophiles appear to exhibit greater sensitivity to the solids concentration employed; however, this is also influenced by the particle size of the feed solids. Therefore, the aim of this study is to examine the opportunities for furhter increasing rates of oxidation using extreme thermophiles for treatment of the pollymetallic concentrates originating from the Bor mining area (Eastern Serbia), used in intensively agitated laboratory reactors.

EXPERIMENTAL

Microorganisms present in RTB Bor resources

Natural population of bacteria on Bor site Specific organisms that exist in the RTB Bor resources

(copper ores, concentrates, tailings and mine waters) have been identified using molecular tools based on Quantitative Polymerase Chain Reaction (Q-PCR anal-ysis) and quantified in different bioleaching processes [12]. Natural population of microorganisms in RTB Bor resources is presented in Table 1.

The results were then combined with process con-ditions, in order to search and find relations between the appearance of specific organisms and environ-mental conditions/process conditions. The relative pro-portions of the cultures present in the analysed mine samples did not vary much, except in the Bor tailings where fairly high amount of bacteria (4,3×106 per g) and archaea exist. It was shown that the dominant organism was Acidianus sp., with Metallosphaera sp. and Sulfolobus sp. present in lower numbers.

Quantification of microbial population The characterization of the microbial communities

present in RTB Bor mine sulphide minerals has been performed in line with bench-scale bioleach amen-ability testing and integrated piloting in Mintek labora-tory. The microbial populations present in the bioleach reactors were identified and quantified by Mintek, Bioclear and the University of Bangor, using Q-PCR and T-RFLP techniques, respectively [14].

Two sets of samples were analysed; the first was collected after reaching steady-state conditions during open-circuit operation, whereas the second set was taken while the system operated as a fully integrated plant. Similar results were obtained on the two sets of samples and total recorded cell numbers varied between 4×108 and 3×109 cells/ml. Quantification of microbial population is shown in Table 2. Acidianus bri-erleyi dominated the population in all four bioleach


577

reactors, while Metallosphaera sedula was present at much lower levels [6,7].

Relative abundance of microbial populations is pre-sented in Table 3. Although little is known about the physiology of Acidianus brierleyi thermoacidophiles, current research indicates that the relatively poor iron-oxidation capacity of A. brierleyi compared to other thermoacidophilic archaea may make it a more suitable microorganism for the oxidative dissolution of second-ary copper sulphides and chalcopyrite [13,14].

Bacterial culture and nutrients

Isolation An extremely thermophilic, iron–sulphur oxidising

bacterial culture Acidianus brierleyi/Metallosphaera sedula, isolated from a moderate hot mine water of RTB Bor complex was used in the study. The culture was grown in a mineral salts solution (substrate) with the following composition: FeSO4⋅7H2O (15 g/l),

(NH4)2SO4 (3.0 g/l), KCl (0.1 g/l), K2HPO4 (0.5 g/l), MgSO4⋅7H2O (0.5 g/l) and Ca(NO3)2⋅4H2O (0.01 g/l).

Adaptation of culture The mixed culture, prepared from isolates obtained

in enrichment culture was adapted to oxidise poly-metallic copper-lead-zinc sulphide concentrate in a 5 l glass magnetic stirred reactor (220 rpm) at 70 °C. The culture was exposed to increasing concentrations of the concentrate and grew well in the presence of 50 g/l of the concentrate and 2.4 g/l of zinc and 12 g/l of copper in solution over a 40-day period.

Mineral characteristics

The test was performed on a copper–lead–zinc sulphide concentrate, fine-milled in a standard labor-atory ring mill unit to particle sizes of d90 = 10 µm. Modal analysis of the sample presented in Table 4, showed that the major sulphides present were chal-copyrite 69%, galena 10%, sphalerite 7% and pyrite 6%.

Table 1. Natural population of bacteria on Bor/Majdanpek sites

Program PCR-analyse Target-PCR Bacteria/g Bor-tailing

P059P427 Universal; Bacteria 4.3×106 P418P419 Universal; Archaea 3.0×105

Ores, concentrates, mine waters M039M040 Acidianus sp. < 4.6×104 M026P258 Acidimicrobium sp. 1.1×105 P242P243 Acidiphilium sp. < 4.6×104 M038P352 Acidothiobacillus caldus < 2.3×104 P363P364 Acidothiobacillus sp. < 2.3×104 P353P354 Acidothiobacillus thiooxidans < 2.3×104 P361P362 Alicyclobacillus sp. < 8.5×103 P365P366 Ferroplasma sp. < 4.6×104 P071M041 Leptospirillum ferrooxidans < 2.3×104 P251P252 Metallosphaera sp. < 4.6×104 P368P369 Sulfobacillus sp. < 4.7×105 P372P377 Sulfolobus sp. < 4.6×104

Table 2. Quantification of microbial populations using Q-PCR

Sample Acidianus brierleyi cells/ml Metallosphaera sedula cells/ml Open circuit

Reactor 1 2.6×109 1.1×107 Reactor 2 3.6×108 7.5×105 Reactor 3 4.5×109 1.4×107 Reactor 4 8.3×108 1.7×106

Integrated circuit Reactor 1 7.6×108 2.0×105 Reactor 2 2.4×109 8.1×105 Reactor 3 8.1×108 7.3×104 Reactor 4 2.9×109 3.6×105


578

The chemical composition of the Cu/Zn/Pb concentrate is shown in Table 5.

Table 3. Relative abundance of microbial populations in the bioleach reactors; ND = not detected; NA = not analysed

Sample Acidianus brierleyi, % Metallosphaera sedula, %

Q-PCR T-RFLP Q-PCR T-RFLP Open circuit

Reactor 1 99.6 99.8 0.4 2.0 Reactor 2 99.8 ND 0.2 ND Reactor 3 99.7 99.0 0.3 1.0 Reactor 4 99.8 ND 0.2 ND

Integrated circuit Reactor 1 99.9 NA 0.02 NA Reactor 2 99.9 NA 0.03 NA Reactor 3 99.9 NA 0.01 NA

Table 4. Mineralogical analyses of the pollymetalic concentrate

Element Content, mass% CuFeS2 69 PbS 10 ZnS 7 FeS2 6 Gangue 8 Total 100

Particle size analyses

The concentrate was milled to a particle size of d90 = = 10 µm. The particle-size distribution of the received and milled concentrate is illustrated in Fig. 1. The fine-milled concentrate was used as feed material in the test work programme.

Table 5. Chemical analyses of the pollymetalic concentrate

Element Content, mass% Cu 24.2 Pb 8.7 Zn 4.9 Fe 25.6 Ag (g/t) 600 Stot 29 Gangue 7.7 Total 100

The Veliki Krivelj concentrate was first wet-milled in a stirred ball mill in order to achieve a target particle size with a d90 of 10 to 12 µm. The pulp from the mill was filtered to remove excess water, and then the filter cake was dried on a hot drying table. The dried solids were screened through a 600 µm to break up the lumps, formed during the drying process. Finally, the milled dry products were packed and stored.

The particle size distribution in this study corres-ponds to the chosen bioleaching operation mode. While dump and heap leaching utilizes run-of-mine lumps of several inches in size and particles of con-trolled size in the range of half inch, respectively, bio-leaching in stirred test reactor requires particle sizes up to 100 μm or less in order to keep them suspended by mechanical means or by air/lift agitation.

Batch test

A batch bioleach test was performed on the con-centrate with a particle size of d90 = 10 µm, at 5% (w/V) pulp density. The test was performed in a 5 l magnetic stirred reactor (220 rpm) at 70 °C, as used for culture adaptation. Air enriched with CO2 (0.15%) was intro-

Figure 1. Particle-size distribution.


579

duced into the reactor via an air sparger. The test gave an initial cell concentration of about 108 cells/ml. Dis-tilled water was added on a daily basis to compensate for evaporation. The initial pH was set at 2 with sul-phuric acid. The progress of the leachates was followed by daily measurement of pH levels and ORP. The amount of Cu and Zn released during the test were determined by periodic liquor analysis.


Batch test

Batch bioleach adaptation The batch test started with the adaptation of the

culture. Chalcopyrite leaching occurs between 510 to 610 mV. Metal sulphides, such as sphalerite and gal-ena, generally leach far more easily. Pyrite oxidation

would only have been significant once the ORP was higher than 450 mV [15]. The pH of the bioleach test solution was adjusted on the first day to the value of 2 with sulphuric acid. During the bioleach adaptation, pH value was controlled by adding sulphuric acid to main-tain pH 1.3, as illustrated in Fig. 2.

Despite the fact that microorganisms involved in copper bioleaching are also acidophilic, (they are active in the pH range from 1.5 to 3.0), operating pH over 2.0 is not allowed mostly because some chemical reactions can ocurr such as, for instance, the precipitation of jarosites – various types of ferric hydroxides. Also, pH cannot drop down to 1.0–1.5 since the viability of cell could be severely affected.

The operating redox potential (ORP) showed an increase from 486 to 610 mV (reffered to the Ag/AgCl electrode, 0.207 V vs. SHE at 25 °C), as illustrated in Fig. 3. Batch chemical ferric leach tests carried out on the

Figure 2. pH levels measured during the bioleach test.

Figure 3. ORP levels measured during the bioleach test.


580

concentrates under conditions of controlled operating redox potential (ORP), showed that the Cu leach rate is significantly increased at an ORP level of 550 mV compared to an ORP of 430 mV. This indicates that significantly reduced residence times may be possible in bioleach processes, using the extreme thermophiles, if higher ORP levels can be maintained.

Therefore, this increase of ORP may clearly indicate a change in the concentration of Cu, Zn and Fe in solution, as illustrated in Fig. 4. Over the 40-day period of adaptation the final measured Cu concentration was 12 g/l and zinc 2.4 g/l (after first five days of test), while a final Fe concentration was of 3.8 g/l.

Steady state batch bioleach To maintain steady state conditions, a batch bio-

leach test started on day 40. The test comprised a

several steady state tests carried out under the same conditions in a 5 l magnetic stirred reactor. The mag-netic speed in reactor was set at 250 rpm. The feed pulp density was maintained at 5%. The investigated feed was ground to achieve a target particle size with a d90 = 10 µm and the overall unit residence time lasted seven days. The steady state data across the reactor for Cu and Zn extractions at the constant pH 1.3 and ORP 610 mV are shown in Fig. 5. During the bioleach pro-cess, jarosite-type compounds form when sulphides undergo strong oxidation under highly acidic conditions [16].

The jarosite compounds has the theoretical formula MFe3(SO4)2(OH)6, where „M“ is a monovalent cation from the group H3O+, Na+, K+, Ag+, NH4

+ or ½Pb2+. The results indicated that jarosite compounds were

Figure 4. Cu, Zn and Fe extractions obtained during the bioleach test at 70 °C.

Figure 5. Cu, Zn and Fe extractions obtained in steady state bioleach test at 70 °C.


581

precipitated effectively from sulphuric acid solution acid at 70 °C while maintaining pH of about 1.3. After the bioleaching, the amount of Pb decreases from 8.7% contained in concentrate to 7% in the primary bioleach residue due to jarosite precipitation and minor inc-reasing of residue weight of 25%, compared to the sul-phide concentrate processed.

Batch brineleach test Bioleach residue could be treated by means of dif-

ferent techniques to get efficient extraction of valuable metals, namely precious metals and lead. The PLINT process was selected to carry out lead recovery [17]. It was shown that NaCl concentration and solid/liquid ratio (pulp density) are very effective parameters for lead recovery at room temperature, while hydrochloric acid addition in brine leachant causes minor effects on lead recovery [18].

Main chemical reaction involved in the brine leach-ing of lead is according to Eq. (12). Silver follows a similar pathway acording to reaction (13). Several researchers emphasized that the solutions having reactions (14 and 15) due to their solubilities being higher than PbCl2, resulted in an increase in the con-centration of Pb2+ in solution [19]:

PbSO4 + 4NaCl → Na2PbCl4 + Na2SO4 (12) AgFe3(SO4)2(OH)6 + 4NaCl → → NaFe3(SO4)2(OH)6 + Na3AgCl4 (13) PbCl2 + Cl– → PbCl3

– (14) PbCl3

– + Cl– → PbCl42– (15)

For lead recovery, brine leaching tests were done on the secondary leach residue obtained after acid leaching of primary bioleach residue, using 150 g/l H2SO4 at 95 °C in 2 h with pulp density of 200 g/l to

transform all backward lead sulphide and precipitated lead jarosite in lead sulphide.

In this study, six different aqeous NaCl solutions for lead recovery from 50 to 350 g/l were tested at the following constant conditions: solid/liquid ratio 1/8 kg/dm3; reaction duration 20 min; reaction tempera-ture 25 °C. The experimental results are shown in Fig. 6. On the basis of these results, one can accept that the brine concentration is an effective parameter for lead recovery. In short, the higher the NaCl concentration, the higher the Pb extraction degree.

However, it is suggested not to use greater than 350 g/l NaCl concentration in brine leaching because when greater values are used NaCl solution becomes saturated. NaCl solubility in 0.1 l cold water is given as 35.7 g [20,21]. This means that maximum NaCl concen-tration can be chosen as 357 g/l. Over the 20 min period of extraction, the measured lead concentration was 7.3 g/l. Finally, the lead could be precipitated with an alkali to produce pure lead oxide or carbonate con-centrate able to be commercialised.

A summary of base metal extractions based on the analysis of the final residues is given in Table 6. Final extractions of 97% Cu, 97% Zn and 95% Pb were achieved. Ninetynine percent of the sulphides were oxidised, indicating that most of chalcopyrite, sphal-erite and pyrite were leached, while galenite was trans-formed into insoluble lead sulphate. According to the residue analysis, no elemental sulphur was formed, which is consistent with results previously obtained using the extreme thermophilic culture on other Cu–sulphide concentrates [11,13].

Much more attention, therefore, has to be paid to the bioleaching of galena. Namely, in a sulphate system galena is oxidized to insoluble lead sulphate and the

Figure 6. Pb extractions data obtained during the brine bioleach test at 25 °C.


582

creation of this compound does not allow the recovery of lead from bacteria and ferric sulphate leaching via integrated solvent extraction/electrowinning process. It also creates an environmental risk due to the increased solubility of lead sulphate over galena.

Table 6. Summary of the metal extractions

Element Extraction, % Cu 97 Zn 97 Pb 95

An attempt done by da Silva, for instance, revealed that the galena bio-oxidation required considerable acid consumption, because of the occlusion of the produced elemental sulphur by precipitated lead sul-phate [22]. This means that galena oxidation may hin-der the bioleaching of other sulphide minerals.

CONCLUSION

An extremely thermophilic, iron–sulphur oxidising bacterial culture, isolated from a moderate hot mine water of Mining and Smelting Company Bor was used in the study. However, for the subsequent use of the culture in bioleach applications, it was necessary to carry out an adaptation step. This included main-tenance of bacterial oxidative activity at sufficient sul-phide solids concentrations, tolerance to high metal ions concentrations such as copper, zinc and lead, and tolerance to flotation reagents associated with sulphide concentrates.

The bacterial culture was employed in a batch test continuously operated 5 l magnetic stirred reactor over a period of 80 days including the period of isolation and adaptation of the bacterial culture, treating a polly-metalic chalcopyrite–zinc–lead sulphide concentrate. Once established in the reactor, employing standard bioleach conditions, the extreme thermophile bacterial culture exhibited stable leaching performance over the period of batch bioleach operation.

The batch bioleach test confirmed that 97% Cu, 97% Zn and in a brine leaching 95% Pb extractions could be obtained on the chalcopyrite–sphalerite–galena con-centrate at a solids concentration of 5% in an agitated stirred reactor.

A combined bioleach and hydrometallurgical pro-cess has the advantage over the RTB Bor smelting pro-cess, because of lower capital costs, flexibility to treat lower grade copper concentrates, pollymetallic concen-trates, gold bearing concentrates, smelter slags, flota-tion tailings, mine waters and, it is environmentally more acceptable because it does not produce hazard-ous sulphur-containing gaseous waste products and reduces the requirement for imported concentrates.

Acknowledgement

This paper was done within the project No. 34023 by the Ministry of Education, Science and Technological Development of the Republic of Serbia.

REFERENCES

[1] Y. Sheng-hua, W. Ai-xiang, Q. Guan-zhou, Bioleaching of low-grade copper sulphides, Trans. Nonferrous Met. Soc. China 18 (2008) 707–713.

[2] J.A. Brierley, A perspective on developments in biohyd-rometallurgy, Hydrometallurgy 94 (2008) 2–7.

[3] X. Yang. X. Zhang. Y. Fan, H. Li, The leaching of pent-landite by Acidithiobacillus ferrooxidans with a biologi-cal–chemical process, Biochem. Eng. J. 42 (2008) 166– –171.

[4] K.S. Rao, A. Mishra, D. Pradham, G.R. Chaudhary, B.K. Mohapatra, T. Das, L.B. Sukia, B.K. Mishra, Percolation bacterial leaching of low-grade chalcopyrite using acido-philic microorganisms, Korean J. Chem. Eng. 25 (2008) 524–530.

[5] D.B. Nakade, Bioleaching of Copper from Low Grade Ore Bornite Using Halophilic Thiobacillus Ferroxidans, N-11, Res. J. Recent Sci. 2 (2012) 162–166.

[6] S.J. Narayan, S. Sajhana, Bioleaching: A Review, Res. J. Biotech. 4(3) (2009) 72–75.

[7] D.J. Kim, D. Pradhan, G.R. Chaudhury, J-G., Ahn, S.W. Lee, Bioleaching of Complex Sulfides Concentrate and Correlation of Leaching Parameters Using Multivariate Data Analysis Technique, Materials Transactions 50(9) (2009) 2318–2322.

[8] V.B. Cvetkovski, V.T. Conić, M. Vuković, M.V. Cvet-kovska, Mesophilic leaching of copper sulphide sludge, J. Serb. Chem. Soc. 74(2) (2009) 213–221.

[9] M. Vuković, Z.D. Stanković, M. Rajčić-Vujasinović, V. Cvetkovski, Voltammetric Investigations of Anodic Dis-solution of Natural Mineral Chalcopyrite, J. Min. Metall., B 4 (2008) 115–124.

[10] D.A. Clark, P.R. Norris, Oxidation of mineral sulphides by thermophilic microorganisms, Minerals Engineering 11 (1996) 1119–1125.

[11] M. Gericke, A. Pinches, J.V. van Rooyen, Bioleaching of a chalcopyrite concentrate using an extremely thermo-philic culture, Int. J. Mineral Processing 62 (2001) 243– –255.

[12] I. Dinkla, B. Geurkink, Biotechnology for metal bearing materials in Europe” (BioMinE), Priority 3.4, EU Project, Contract Number: NMP2-CT-2005-500329-1, Nov 2005-Oct 2008, Deliverable DII.5, report entitled as 3rd interim report on the use of molecular biology tools for detect-ing, identifying & monitoring bioleaching microbial systems, 2008, p. 15.

[13] S.M. Gericke,Biotechnology for metal bearing materials in Europe” (BioMinE), Priority 3.4, EU Project, Contract Number: NMP2-CT-2005-500329-1, Nov 2005-Oct 2008, Deliverable: DII4, report entitled as Microbial consortia and mineral processing, 2008, p. 73.


583

[14] J. Vilcáez, K. Suto, C. Inoue, Bioleaching of chalcopyrite with thermophiles: Temperature–pH–ORP dependence, Int. J. Miner. Process. 88 (2008) 37–44.

[15] M. Boon, J.J. Heijnen, G.S. Hansford, The mechanism and kinetics of bioleaching sulphide minerals, Miner. Process Extract. Metal. Rev. 19(1–4) (1998) 107–115.

[16] J.E. Dutrizac, S. Kaiman, Synthesis and properties of jar-osite-type compounds, Can. Mineral. 14 (1976) 151– –158.

[17] C. Frías, A. Ibáñez, F. Sánchez, S. Sanguilinda, T. Reuni-das, Biotechnology for metal bearing materials (BioMinE), Priority 3.4, EU Project, Contract Number: NMP2-CT-2005-500329-1, Nov 2005-Oct 2008, BioMinE DIV8B – Part 2, Zn and Zn polymetallics. Report on pilot-ing operation and, pre-feasibility study (Part 2), Actual Submission Date: 2008.

[18] R.A. Aydin, Thesis Submitted to the Graduate Scholl of Natural and Applied Sciences of Middle East Technical University, Recovery of Zinc and lead from Cinkur leach resudies by using hydrometallurgical techniques, Turkey, 2007.

[19] M. Deniz Turan, H. Soner Altundogan, F. Tumen, Reco-very of zinc and lead from zinc plant residue, Hydro-metallurgy 75 (2004) 169–176.

[20] D. Sinadinovic, Ž. Kamberović, A. Šutić, Leaching kinetics of lead from lead (II) sulphate in aqueous calcium chlo-ride and magnesium chloride solutions, Hydrometal-lurgy 47 (1997) 137–147.

[21] C.K. Gupta, T.K. Mukherjee, Hydrometallurgy in Extract-ion Processes (Vol. 1), CRC Press, Boca Raton, FL, 1990.

[22] G. da Silva, Kinetics and mechanism of the bacterial and ferric sulphate oxidation of galena, Hydrometallurgy 75 (2004) 99–110.

IZVOD

BIOLUŽENJE POLIMETALIČNOG SULFIDNOG KONCENTRATA TERMOFILNIM BAKTERIJAMA Milovan Vuković1, Nada Štrbac1, Miroslav Sokić2, Vesna Grekulović1, Vladimir Cvetkovski3

1Univerzitet u Beogradu, Tehnički fakultet u Boru, Bor, Srbija 2Institut za tehnologiju nuklearnih i drugig mineralnih sirovina, Beograd,Srbija 3Institut rudarstva i metalurgije, Bor, Srbija

(Naučni rad)

Biotehnologija kao noviji pristup za ekstrakciju metala nudi mogućnosti zasmanjenje zagađenja životne sredine. U ovom radu se od četiri biotehnološkapostupka – bioremedijacija, biosorpcija, bioakumulacija i bioluženje – sagledavaju različiti aspekti (uključujući i ekološke) bioluženja polimetaličnih sulfidnih koncen-trata koji potiču iz borske rudarske oblasti. Ispitana je mogućnost povećanjabrzine oksidacije sulfida bakra u laboratorijskim uslovima u prisustvu termofilnihbakterija. Ekstremno termofilna i oksidaciona bakterijska kultura, razvijena u pri-sustvu sumpora i gvožđa, izolovana je i adaptirana na visoku koncentraciju jonametala i čestica materijala u rastvoru na temperaturi od 70 °C. Nakon izolacije i adaptacije mikroorganizama, kultura je korišćena u staklenom reaktoru zapreminepet litara za bioluženje (potpomognuto magnetnom agitacijom i aeracijom) poli-metaličnog sulfidnog koncentrata koji je od metala sadržavao bakar, cink i olovo.Eksperimenti sprovedeni u ovoj specifičnoj bakteriološkoj sredini pokazali su da jemoguće, posle postizanja ravnotežnih uslova, postići visoke stepene ekstrakcijebakra i cinka (do 97%) u dužim vremenskim intervalima – do 80 dana. Olovo-sulfid je tokom ovog procesa oksidacijom prešao u olovo-sulfat, te ostao u biolužnom ostatku zbog neznatne rastvorljivosti u sulfatnom rastvoru. U ovom radu je bio-luženje bioostatka sprovedeno po PLINT tehnologiji koja omogućava rastvaranje jedinjenja olova u prisustvu natrijum-hlorida. Rezultati ovih eksperimenata supotvrdili rezultate sličnih istraživanja po kojima je moguće dobiti visoke stepeneekstrakcije olova iz biolužnog ostatka polimetaličnih koncentrata – i do 95%.

Ključne reči: bioluženje • Bakterije •Halkopirit • Svalerit • Galenit • Termofili

585

Synthesis of bismuth(III) oxide films based anodes for electrochemical degradation of Reactive blue 19 and Crystal violet

Milica M. Petrović, Jelena Z. Mitrović, Miljana D. Radović, Danijela V. Bojić, Miloš M. Kostić, Radomir B. Ljupković, Aleksandar Lj. Bojić

Department of Chemistry, Faculty of Science and Mathematics, University of Niš, Niš, Serbia

Abstract The Bi2O3 films-based anodes were synthesized by electrodeposition of Bi on stainless steelsubstrate at constant current density and during different deposition times, followed by calcination, forming Bi2O3. The thickness of the films was determined by two methods: theobservation under the microscope and by calculation from mass difference. The electro-chemical processes at the anodes were ivestigated by linear sweep voltammetry. At the anodes obtained within 2, 5, 10 and 15 min of deposition, two dyes, namely: Reactive blue 19 and Crystal violet, were decolorized by oxidation with •OH radical, generated from H2O2

decomposition at the anodes. Decoloration times of the anodes varied, and the shortestone was achieved with the anode obtained during 5 min of deposition, with the film thick-ness of 2.5±0.3 μm. The optimal H2O2 concentration for the dyes degradation was found tobe 10 mmol dm–3.

Keywords: Bismuth(III) oxide, anodes, synthesis, film thickness, decoloration, hydrogen per-oxide.

SCIENTIFIC PAPER

UDC 544.653.2:547.87–31

Hem. Ind. 68 (5) 585–595 (2014)

doi: 10.2298/HEMIND121001084P


Industry releases huge amounts of more or less colored effluents into the environment. Color itself could be very pernicious to the receiving water sources due to the toxicity towards many aquatic organisms and because colored compounds reduce water trans-parency, which, in turn, affects photosynthetic activity, thus causing severe damage to the ecosystems [1].

The electrochemical processes for wastewater treatment have many advantages, including: environ-mental compatibility, versatility, high energy efficiency and safety, because they operate at mild conditions. For these reasons, they have been largely developed and utilized. Among them, the electrochemical oxi-dation is the most popular electrochemical procedure for removing organic pollutants from wastewaters and it has been widely used for decolorizing and degrading dyes from aqueous solutions. The oxidation of pollut-ants can be done as the direct anodic oxidation, and direct electron transfer to the anode, which yields poor decontamination; or chemical reaction with electro-generated species from water discharge at the anode surface such as “active oxygen”, i.e., hydroxyl radical, •OH, which is considered the responsible species for the electrochemical degradation of organic pollutants [2]. Various materials are used as the anodes: Pt [3], boron-doped diamond [4,5], graphite [6] activated carbon fiber [7] and the electrodes based on metal Correspondence: M. Petrović, Department of Chemistry, Faculty of Science and Mathematics, University of Niš, 18000 Niš, Serbia. E-mail: [email protected] Paper received: 1 October, 2012 Paper accepted: 19 November, 2013

oxides, such as PbO2, RuO2, IrO2, SnO2, SbOx, etc. and their mixtures [8–13]. When hydrogen peroxide is applied to the electrochemical system, the radicals would be electrogenerated with hydrogen peroxide and they would further attack the organic pollutants in the system. In the presence of hydrogen peroxide, both hydroxyl radicals and hydroperoxyl radicals were pro-duced with hydrogen peroxide at the cathode and anode, respectively [14,15]:

H2O2 +e– → •OH + OH– (1)

H2O2 → HO2• + H + +e– (2)

The oxidation of organic dyes by hydroperoxyl radi-cals can be neglected; however, highly reactive hyd-roxyl radicals produced via reaction (1) could react with the organics, resulting in their degradation. In acidic effluents, in the presence of small quantity of Fe2+, H2O2 decomposes to produce •OH and Fe3+ [2]:

H2O2 + Fe2+ → Fe3+ + •OH + OH– (3)

Generated •OH can further decompose the organic molecules.

Anode material is a very important factor deter-mining the extent of decoloration in electrochemical dye degradation processes [2,7,8,12,16–18]. It should possess several important characteristics: an inert sur-face with low adsorption properties which does not provide catalytically active sites for the adsorption of reactants in aqueous media (providing the formation of high concentration of •OH from water discharge), high corrosion stability and high O2 evolution overvoltage.

M.M. PETROVIĆ et al.: SYNTHESIS OF BISMUTH(III) OXIDE BASED ANODES Hem. ind. 68 (5) 585–595 (2014)

586

Boron-doped diamond electrodes possess all these properties and they have the highest color removal efficiency [2,4,18]. However, their application at indus-trial scale is not suitable, mainly due to the difficulties in their preparation and high production cost [2].

Anodes based on metal oxides have high surface area and excellent mechanical and chemical resistance even at high current densities. Various materials based on metal oxides have been used for electrochemical degradation of dyes, showing different color removal efficiency [8–13]. Some semiconductor metal oxide based anodes are used in photoelectrocatalytic pro-cesses [2].

Microcrystalline Bi2O3 can offer large surface area, electrochemical stability and catalysis behavior [20], which makes it an interesting material for electroche-mical oxidation of various organic pollutants. The Bi2O3/Ti electrode was used in oxidative degradation of Acid orange 7 by electrolysis, photocatalytic oxidation and photoelectrocatalytic oxidation processes [21]. In addition, Bi2O3 is relatively low cost and easy to pre-pare.

The electrodeposition is a very convenient method of material synthesis, because it is simple and it offers rigid control of film thickness, uniformity, and depo-sition rate and is especially attractive owing to its low equipment cost and starting materials. In cathodic elec-trodeposition, the metal ions or complexes are hydro-lyzed by electrogenerated base to form oxide, hydro-xide, or peroxide deposits on cathodic substrates. Hyd-roxide and peroxide deposits can be converted to cor-responding oxides by thermal treatment [22].

In this work, Bi2O3-based anodes were synthesized by Bi electrodeposition on stainless steel substrate, at constant current density and during different depo-sition times, followed by calcination. The aim of the work was to investigate the ability of the anodes

obtained during different electrodeposition times to degrade anthraquinone reactive dye Reactive blue 19 and triphenylmethane dye Crystal violet.

EXPERIMENTAL

Materials and equipment

All chemicals were of reagent grade and used without further purification. Bismuth (III) nitrate penta-hydrate was purchased from Carlo Erba (Chezch Rep-ublic), nitric acid, hydrogen peroxide, sodium sulfate; Reactive blue and Crystal violet were purchased from Sigma Aldrich.

All electrochemical experiments were carried out using Amel 510 DC potentiostat (Materials Mates, Italy) furnished with VoltaScope software package. The dye concentrations were determined using UV–Vis spectro-photometer Shimadzu UV-1650 PC (Shimadzu, Japan). Samples were taken during the experiments and their absorbencies were recorded at the wavelengths of 592 nm for Reactive blue 19 and 590 nm for Crystal violet, respectively, at which the dyes show their absorption maxima (Table 1).

Preparation of the anodes

The electrodeposition was performed at room tem-perature in the two-electrode cell with a stainless steel sheet (10 mm×25 mm) as a substrate for the film depo-sition (cathode) and Au sheet (10 mm×20 mm) as auxil-lary electrode (anode). The distance between working and auxiliary electrode was 15 mm. Before the depo-sition, the substrate was polished with different abra-sive papers, ultrasonically cleaned with ethanol and deionized water. One group of stainless steel samples was anodically treated in 0.5 M oxalic acid at current density of 500 mA cm–2 for 30 min. All electrodepo-sition solutions were prepared with distilled water. 0.1

Table 1. Main characteristics of Reactive blue 19 and Crystal violet dyes

Characteristic Reactive blue 19 Crystal Violet Chemical structure

C.I. generic name Reactive blue 19 – Synonym Remazol Brilliant Blue R – IUPAC Name 2-(3-(4-Amino-9,10-dihydro-3-sulpho-9,10-dioxo-

anthracen-4-yl)aminobenzenesulphonyl)vinyl) disodium sulphate

Tris(4-(dimethylamino)phenyl)methylium chloride

Molar mass, g mol–1 626.54 407.98 λmax / nm 592 590


587

M Bi3+ solutions were prepared by dissolving the required amount of bismuth nitrate in 1 M HNO3 water solution. The electrodeposition was carried out at cons-tant current density of 40 mA cm–2 (with regard to geo-metrical surface area of the stainless steel substrate) with the deposition times of: 1, 2, 5, 10, 15, 30 and 45 min. After the deposition, the electrodes covered with the films were washed with distilled water, dried at room temperature for 24 h, calcined at 500 °C for 90 min in air in a furnace and cooled in the open air. Mea-surements of the films thickness and the observation of the anodes surfaces were performed using MZ16 A microscope (Leica), equipped with micrometer scale.

In order to perform a more detailed investigation of the obtained material and processes taking place on its surface, another group of the anodes was prepared by electrodeposition under the same experimental condi-tions as described above, but they were not calcined. Also, following materials were prepared and used as the anodes: a pure stainless steel sheet anodically treated in 0.5 M oxalic acid at current density of 500 mA cm–2 for 30 min (the same conditions as the sub-strates for the Bi deposition); and a stainless steel sheet anodically treated in 0.5 M oxalic acid at current density of 500 mA cm–2 for 30 min and calcined at 500 °C for 90 min in air.

Electrochemical characterization

Characterization of electrochemical processes at the anodes surfaces was performed using linear sweep voltammetry. Voltammograms of the anodes were recorded in the solutions which contained 10 mM H2O2 and 1 mM Na2SO4 by scanning from 0.6 to 3 V at a scan rate of 20 mV s–1, using saturated calomel electrode as a reference electrode and Au sheet as an auxiliary elec-trode. The compositions of the solutions were the same as the ones for the dye degradation, but without the presence of the dyes. All the potentials in this work are given versus standard hydrogen electrode.

Dye degradation experiments

Dye degradation experiments were carried out at room temperature, in two-electrode cell, at constant current density of 10 mA cm–2, with each of the anodes, using Au sheet as a cathode. Reactive blue 19 and Crystal violet solutions of 50 mg L–1 of the dye, pH 7.0±0.1 and 10 mM H2O2 were prepared separately by dissolving the proper amounts of powdered dye and H2O2 in water. Each of the solutions contained a 1 mM Na2SO4, which provided their electrical conductivity. During the decoloration experiments, the dye solutions were stirred on a magnetic stirrer. Dye decoloration experiments were also carried out using non-calcined bismuth anodes and stainless steel anode. Calcined stainless steel anode was not electrochemically stable enough to be used as the anode; at high potentials

which are applied in our experiments, an intensive corrosion of the anode and oxygen evolution takes place. The decoloration time was observed for each of the anodes. All experiments were performed in trip-licate.


All of the films were electrodeposited at the same constant current density, but during the different depo-sition times. The color of the deposited material was middle gray to pale gray.

For the H+ concentration > 0.4 M, which is the case in our work, the Bi3+ prevails, and the predominant cathodic reaction may be [23]:

Bi3+ + 3e– → Bi0 Based on this equation and the color of the depo-

sited films, it can be assumed that bismuth metal was predominantly deposited at the cathode and later, it slowly oxidized in water and air during the drying pro-cess.

After the calcinations at 500 °C, the color of the deposited material changed to pale yellow, indicating that the Bi2O3 was formed.

Adhesion of the films deposited on the surface which was only treated with abrasives and ultrasonic-ally cleaned, was not satisfying; unlike them, films deposited on the surface which was anodically treated with oxalic acid at high current density had very good adhesion and mechanical stability, which was further improved by calcinations.

Thickness of the films

The thickness of the obtained films was calculated from the mass difference before and after the electro-deposition and calcination, assuming the density of the deposited material was the same as that of the bulk material (ρ = 9.17 g cm–3 for Bi2O3). Note that the thick-ness determination was performed only for the cal-cined anodes. Results obtained from this calculation were similar to those obtained by the observation of the cross sections of the stainless steel samples covered with Bi2O3 films under the microscope. The dif-ferences between the results obtained from these two methods for the current density of 40 mA cm–2, were between 5 and 20% and they were the lowest for the films with the thickness of about 1.5 to 6 μm. These films were the ones of the highest compactivity and homogeneity. Figure 1 shows dependence of the film thickness of deposition time. Note that the thickness values given in Figure 1 are the ones obtained by obser-vation and measurement under the microscope. The thinnest film, obtained during 1 min of deposition, was quite inhomogeneous, with significantly lower com-pactivity, and, on some parts of the surface, it could be noted that it wasn’t fully formed; its thickness varied a


588

lot, and it was impossible to measure it with satisfying accuracy, so the thickness value given for it is its cal-culated value. The films thicker than 12 μm were also inhomogeneous, with larger particles attached to the homogenous part of the film and they were not very stable; when their thickness reached more than 15 μm, they became very unstable and peeled off relatively easily.

Figure 1. Dependence of the films thicknesses on deposition time.

It can be observed that during the first 15 min of deposition the film thickness rapidly grows with the deposition time. After that time, the film thickness growth becomes significantly slower. Moreover, as mentioned above, the films obtained during the longer deposition times tend to crack and relatively easily peel off, so it can be assumed that the maximum film thick-ness with the deposition parameters applied in this work was obtained during the first 15–20 min of depo-sition.

Only the anodes with the film thickness of up to 10 μm were mechanically stable enough to be used in the dye decoloration experiments. The measured thickness values of the films deposited during 2, 5, 10 and 15 min are: 1.5±0.3, 2.5±0.3, 5.6±0.5 and 9.6±1 µm, respecti-vely. As mentioned above, the thickness of the film obtained during 1 min of electrodeposition was impos-sible to measure with satisfying accuracy, so the thick-ness value given for it is its calculated value of 0.7 µm. In further text, the anodes will be labeled: 0.7, 1.5, 2.5, 5.6 and 9.6, the numbers which correspond to their thickness. The cross section of the anode with 2.5 µm film thickness, which showed the best decoloration results, is presented in Figure 2.

The surfaces of the stainless steel/Bi2O3 anodes are shown in Figure 3.

As it was expected, some differences between the surfaces obtained during different electrodeposition times are observed. During the first minute of depo-

sition (anode 0.7, Figure 3A), a small aggregates have been formed and randomly attached to the surface, leaving a significant part of the metallic surface uncov-ered. The observation of its cross section confirmed that the Bi2O3 film was not fully formed (result is not shown). During the first two minutes (anode 1.5) rela-tively, the compact layer has been formed (result is not shown), showing that during that time Bi2O3 aggregates have covered practically the whole metallic surface (Figure 3B). Though the aggregates are closely packed and well adhered to the metallic surface, the coat that they have formed still appears to be quite porous and not very homogenous on its exposed surface. Surface of the anode 2.5 (Figure 3C) is covered with larger aggregates than anodes 0.7 and 1.5, indicating that between the third and the fifth minute of deposition the aggregates have significantly grown in size. Its sur-face is also fully covered, though it seems that the layer is little less compact and more porous than that on the anode 1.5. It is, however, also well adhered to the metallic surface, though the exposed surface of the film is inhomogeneous. Surfaces of the anodes 2.5, 5.6 and 9.6 have similar structures (Figure 3C, D and E), mean-ing that no further growth of the aggregates is observed, but only the growth of the thickness of the film they have formed. The surface coatings on the anodes 5.6 and 9.6 appear to be little more compact than that on the anode 2.5 and, as the measurements have shown, they are thicker as well.

Figure 2. Cross section of the anode with the Bi2O3 film thickness of 2.5 µm.

Decoloration of dyes and electrochemical processes at the anodes

Reactive blue 19 is very stable in the solutions used in this work, and it practically does not react with H2O2 without electrochemical treatment during the time period of 24 h. The study of Radović et al. [24] with the same dye gave the similar results. Crystal violet slowly reacts with H2O2 and it takes about 15 h for it to be completely decolorized.

Before the experiments with the calcined anodes, dye decoloration was performed using non calcined bismuth anode and stainless steel anode. All of the experiments were performed under the same decolor-


589

ation conditions. Non calcined bismuth anodes were not chemically stable under the experimental condi-tions; they quickly started to corrode during the experi-ments and the solutions became turbid; after the sedi-mentation, the precipitate was dissolved in HNO3 and Bi metal was detected by atomic absorption spectro-photometry (A Analyst 300 (Perkin Elmer, USA)), indicating that the material obtained by electrodepo-sition was predominantly metal Bi, which was anodi-cally dissolved during the process. With stainless steel anode purple color of Crystal violet and blue color of Reactive blue 19 disappeared within 60 and 130 min, respectively, but the color of the solutions after that time became pale yellowish-green and it has not changed during prolonged electrochemical treatment. Though the solutions looked clear, after 24 h a small amounts of precipitate have been detected; the preci-pitate from both solutions was dissolved in HNO3 and Fe was detected by AAS, leading to the conclusion that the dyes were electrocoagulated with iron originating from stainless steel [2]. Decoloration on the calcined bismuth anodes provided clear, completely colorless solutions; though some Pourbaix diagrams show that Bi might exist in some kind of dissolved form at neutral pHs and high potentials [25], neither Fe nor Bi was detected in the solutions, so it can be assumed that the material obtained after the calcination at the stainless steel surfaces was Bi2O3 and it was chemically stable under those conditions. However, after dye decolor-ation with the anode 0.7, iron was found in the solu-tion. This iron probably originates from the parts of the anode surface which were not covered with Bi2O3.

In order to reveal and compare electrochemical processes at different anode materials, a linear sweep

voltammetry investigation was performed in the abs-ence of the dyes. Figure 4 represents current–potential dependence of the stainless steel anode treated in oxalic acid and Bi2O3 anode 0.7 in the presence and absence of H2O2, as well as the current-potential dependences of Bi2O3 anodes 1.5, 2.5, 5.6 and 9.6 in the absence of H2O2. Figure 5 represents current–pot-ential dependence of Bi2O3 anodes 1.5, 2.5, 5.6 and 9.6 in the presence of H2O2.

Figure 4. Linear voltammograms of: 1) stainless steel anode in 10 mM H2O2 and 1mM Na2SO4; 2) stainless steel anode in 1mM Na2SO4; 3), 4), 5) and 6): anodes 1.5, 2.5, 5.6 and 9.6, respectively, in 1mM Na2SO4.

Significant differences in the corresponding lines between stainless steel anode and Bi2O3 anodes can be observed. Current–potential dependence of stainless steel anode in the presence of H2O2 (Figure 4, line 1)

Figure 3. Surface structure of the anodes with the Bi2O3 film thickness of: A) 0.7, B) 1.5, C) 2.5, D) 5.6 and E) 9.6 µm.


590

shows that the current starts to grow at the potential of about 1.50 V. This growth can be attributed to a corrosion of the anode and oxygen evolution. At the potential of 1.95 V current starts to increase more rapidly, with the higher slope of the corresponding line than that between 1.5 and 1.95 V, indicating the begin-ning of some other electrochemical process. Starting from this potential (1.95 V), the growth of the current is practically linear, and no significant further changes in its shape are observed in the scanned potential region. The current–potential dependence of stainless steel anode in the absence of H2O2 (Figure 4, line 2) has very similar shape, with the little lower current values than those in the presence of H2O2. This indicates that similar processes are taking place in the presence and absence of H2O2 at this anode. A rapid growth of the current starting from about 2V might be attributed to a formation of ferrate ion, FeO4

2–. This anion can be generated by anodic oxidation of iron and its alloys, with the standard electrode potential of 2.20 V in acidic solutions and 0.72 V in alkaline solutions, respectively [26]. Dissolution of iron can be further confirmed by corresponding Pourbaix diagram, which shows that at neutral pHs and electrode potentials higher than 2V (which is the case in our dye degradation experiments), metallic iron exists mainly as FeO4

2– [27]. Bi2O3-based anode 0.7 has a similar electrochemical behavior as the stainless steel anode in the scanned potential region, in the presence and absence of. H2O2 (results are not shown). The rapid current increase is observed between 2.2 and 2.3 V, which is a little higher potential than at the stainless steel anode. This difference can be attri-buted to a different nature of the anodes’ surfaces. The observed current is somewhat lower than that obtained with stainless steel anode. Electrochemical behavior of this anode is also similar in the presence and absence of H2O2. This all indicates that at this anode dominant processes are basically the same as those taking place at the stainless steel anode. Having in mind that the dye decoloration times for stainless steel anode and anode 0.7 are very similar (60 min for Crystal violet for both of the anodes and 130 and 110 min for Reactive blue 19 for stainless steel anode and anode 0.7, res-pectively), it can be assumed that electrogenerated ferrates are responsible for the dyes decoloration at both of these anodes. Current-potential dependences of Bi2O3 anodes 1.5, 2.5, 5.6 and 9.6 (Figure 4, lines 5, 6, 7 and 8, respectively) in the absence of H2O2 is entirely different than that of stainless steel anode and anode 0.7. No significant current is observed within the scan-ned potential region. However, starting from about 2.2 V, a low current can be observed. It increases linearly with the increasing potential and no further change in the shapes of the corresponding curves is observed in the scanned potential region. This current is much

lower than the currents that correspond to electroche-mical processes taking place at the stainless steel and anode 0.7, as well as at Bi2O3 anodes in the presence of H2O2 (see below); it is safe to assume that this current cannot be attributed to any particular electrochemical process which might lead to dyes decoloration; no detectable decoloration of the dyes at anodes 1.5, 2.5, 5.6 and 9.6 in the absence of H2O2 within 2 h at 10 mA cm–2, 1 mM Na2SO4 and 50 mg L–1 initial dyes concen-tration took place. This also indicates that direct dyes oxidation at the anodes did not happen.

Figure 5. Linear voltammograms of the Bi2O3-based anodes: 1.5, 2.5, 5.6 and 9.6 in 10 mM H2O2 and 1mM Na2SO4.

Figure 5 shows current–potential dependence of Bi2O3 anodes 1.5, 2.5, 5.6 and 9.6 in 10 mM H2O2 and 1 mM Na2SO4. The shapes of the lines are very different from the ones obtained in the absence of H2O2. At high potentials, H2O2 decomposes with the formation of •OH, a very strong oxidant which can oxidize many organic compounds. Standard electrode potential for that reaction is 2.8 V [2,28]. This is very close to the potential values of current peaks at 2.65 V for the anode 1.5 and between 2.74 and 2.78 V for the anodes 2.5, 5.6 and 9.6 in the presence of H2O2 (Figure 5). The current peaks values are 8.1, 4.90, 3.76 and 1.85 mA, respectively, which is much higher than the values obtained at the same potentials in the absence of H2O2. Taking into account the potentials of current peaks, and the fact that the dyes decoloration proceeded in the presence of H2O2 (see below), these peaks can be attributed to a formation of oxidative species obtained by the decomposition of H2O2 at the anodes, probably including •OH, which is responsible for the dye oxida-tion in this system. All of described electrochemical experiments with Bi2O3 anodes in the presence and absence of H2O2 were repeated several times and the results were the same, meaning that all the current and potential values were repeatable. No traces of bismuth were detected in any of the solutions after the experi-ments, including all of the dye decoloration experi-ments as well. Also, there were no changes in the weight of the anodes before and after the experiments


591

(the anodes weight was measured before each of the electrochemical experiments, they were cleaned after the experiment and dried at 80 °C for 2 h and then their weight was measured again). This all indicates that the anodes are electrochemically stable in the con-ditions applied in this work.

The current values in their maxima and after reach-ing the maxima are different for different anodes and they decrease as the thickness of the anodes increases. Different current values at the same potentials indicate different abilities of the anodes to produce oxidative species by the decomposition of H2O2 and this might be attributed to different electrical properties of the anodes. It might have occurred due to differences in their thickness, but the reason might be a different crystalline structure of the deposited Bi2O3 films as well: it is known that Bi2O3 may appear in several crys-talline structures: monoclinic α-Bi2O3, stable at room temperature; cubic fluorite type δ-Bi2O3 which exists above 729 °C up to melting point at 825 °C; and two metastable phases, which may occur upon cooling near 650 and 640 °C, respectively: the tetragonal β-phase, and body-centered cubic γ-phase. They usually trans-form to α-phase upon further cooling to the room tem-perature [29,30]. These phases have different electrical conductivities; metastable phases have about one order of magnitude higher conductivity than α-phase [31]. Crystalline structure of Bi2O3 will certainly depend on the synthesis conditions. In our case, the synthesis parameter that varied for different anodes was the time of bismuth electrodeposition and it is possible that the anodes have different crystalline structures. The differences can be already seen in macroscopic structures of the anodes. The macroscopic surface structure of the anode 1.5 is significantly different from the others; the Bi2O3 aggregates at its surface are much smaller (Figure 3). The decomposition of H2O2 at this anode is observed at a little lower potential (2.65 V) than at the others (2.74, 2.75 and 2.78 V for anodes 2.5, 5.6 and 9.6, respectively). Anodes 2.5, 5.6 and 9.6 have also exhibited different potentials for decompo-sition of H2O2 and the formation of corresponding oxi-dative species. Although these differences are very small, they still indicate that the nature and properties of the examined anodes are different. It can also be observed that Bi2O3 films of the anodes 5.6 and 9.6 are little more compact than the others, besides the fact that they are thicker (Figure 2). The reason for different electrical behavior of the anodes obtained within dif-ferent electrodeposition times might be different Bi2O3 film thickness of the anodes, but it may also be exist-ence of differences in their crystalline structures. At this point, it is hard to assume how much impact each of these two factors (thickness or crystalline structure of the films) has. It is certain, however, that the change

of electrodeposition time, as the one of the synthesis parameters in this work, brought the significant differ-ence in abilities of the anodes to decompose H2O2 and form the corresponding species, which could oxidize the dyes.

Decoloration efficiency of Bi2O3 anodes

All of the tested Bi2O3 anodes exhibited the ability to decolorize two dyes in the presence of H2O2. It is assumed that the dyes were decolorized by oxidation with the strong oxidants formed by the decomposition of H2O2 at the anodes at high potentials. Decoloration experiments were also carried out in the absence of H2O2 for both of the dyes, under the same experi-mental conditions as in its presence; after 2 h of elec-trolysis, the dyes decoloration was negligible. In order to ensure that the possible corrosion of the stainless steel trough Bi2O3 film and formation of electro-Fenton reagent (Fe+2) did not occur (which could also lead to the dye degradation), the Bi2O3 films were deposited onto titanium substrate, under the same conditions as with the stainless steel; the decoloration experiments in the presence of anodes with the fully formed films gave very similar results as those with the stainless steel substrate, assuming that, even if the electro-Fe-ton reaction took place, their effect was negligible.

Figure 6 shows the efficiency of the tested Bi2O3 anodes for the decoloration of Reactive blue 19 and Crystal violet. Decoloration efficiency is expressed as decoloration time, i.e., the time needed for 100% deco-loration of the dye solutions.

Figure 6. Decoloration times obtained for the anodes with var-ious film thicknesses (initial dye concentration 50 mg dm–3, 10 mM H2O2, 1 mM Na2SO4, applied current density 10 mA cm–2).

The decoloration times of the anodes significantly differ mutually. Decoloration times of Crystal Violet are shorter then ones of Reactive Blue 19, which can be attributed to the structure, and thus, the stability of the dyes molecules [32]. It can also be observed that two curves have similar shape, which indicates that in


592

both cases the similar processes are taking place at the anodes surfaces.

As can be seen in Figure 6, the shortest decolor-ation times, for both of the dyes, are achieved for the anodes with the film thicknesses between 2 and 4 μm; for the tested anodes, the shortest decoloration time (17±1 min for Crystal violet and 35±1 min for reactive blue 19) was achieved for the anode 2.5 (2.5 μm thick-ness) which was obtained during 5 minutes of deposi-tion. As shown in Figure 5, current values in 10 mM H2O2, attributed to a formation of •OH, are different for different anodes at the same potentials and they dec-rease as the thickness of the anodes increases from 1.5 to 9.6 μm, Since the dyes are not oxidized directly, but via oxidative species obtained by the decomposition of H2O2, decoloration efficiency of the anodes depends on their ability to produce those species. Based on the current values for the anodes in Figure 5, the abilities of the anodes to decompose H2O2, producing the oxi-dative species which could degrade the dyes, can be sorted in descending order: 1.5 > 2.5 > 5.6 > 9.6, and, based on that, it is expected that decoloration effi-ciency decreases in the same way. However, decolor-ation time of anode 1.5 is longer than that of the anodes 2.5 and 5.6, although it is expected to be able to produce the highest concentration of oxidants that can decolorize the dyes. After decoloration experi-ments with anode 1.5, the purple and blue shade could be observed at its surface. Desorption of the dyes was performed by immersing the anode in water-ethanol mixture and stirring it for 3 h at 50 °C, and Reactive blue 19 and Crystal violet were detected in the solu-tion, indicating that the dyes adsorbed onto the anode surface and probably blocked the active places for H2O2 decomposition. The same experiment was performed with the rest of the anodes, although their color did not visually change; neither Reactive blue 19, nor Crystal violet was detected in the solutions, indicating that dyes did not adsorb onto the other anodes. It can be assumed that anode 1.5 has the highest ability to decompose H2O2, forming the oxidants which oxidized the dyes, but also the highest sorption affinity for the dyes, and that the second prevailed in this case, making it less efficient for decoloration than anodes 2.5 and 5.6.

As already mentioned, the shortest decoloration time was achieved for the anode 2.5. As the film thick-ness increases from 2.5 to 9.6 μm, the anode’s ability to decompose H2O2 and form the strong oxidants dec-reases, and decoloration time slightly increases, as it was expected. The current maxima obtained for the anodes 5.6 and 9.6 were 3.76 and 1.85 mA (Figure 5), which is 77 and 38% of the value obtained for anode 2.5, respectively. Decoloration times of Reactive blue 19 with the anodes 5.6 and 9.6 were 47±1 and 100±3

min, which is about 74 and 35% of the efficiency demonstrated with anode 2.5, respectively (having in mind that decoloration time of anode 2.5 for reactive blue 19 was 35±1 min). Decoloration times of Crystal violet with the anodes 5.6 and 9.6 were 24±1 and 46±1 min, which is about 71 and 37% of the efficiency demonstrated with anode 2.5, respectively (having in mind that decoloration time of anode 2.5 for Crystal violet was 17±1 min). This indicates that decoloration efficiency of these three anodes is determined by their ability to decompose H2O2, producing the reactive species which oxidize the dyes and no side effects are observed, as in the case of anode 1.5. The smallest difference in decoloration times is observed between the anode 2.5 and 5.6, for both of the dyes. As it is shown in Figure 6, for the film thicknesses higher than about 6 μm, the decoloration time significantly inc-reases as well, and greater difference is observed between the anodes 5.6 and 9.6, than between 2.5 and 5.6. It is not certain which phenomenon actually caused those differences. They all have different thicknesses, but, as mentioned before, it is also possible that their Bi2O3 films have different crystalline structures.

Decoloration reactions with varying concentrations of peroxide were done with the anode 2.5, which has shown the shortest decoloration times in the presence of 10 mM H2O2. The results are shown in Figures 7 and 8.

Figure 7. Effect of H2O2 concentration on the decoloration time of reactive blue 19 (1 mM Na2SO4, applied current density 10 mA cm–2).

The total decoloration times for reactive blue 19 in the presence of 1, 2, 5, 10 and 20 mM H2O2 are: 90, 60, 45, 35 and 45 min, respectively (note that the dye is considered completely degraded when its concentra-tion in the solution is below 1 mass% of its initial con-centration). The total decoloration times for Crystal


593

violet in the presence of 1, 2, 5, 10 and 20 mM H2O2 are: 45, 35, 20, 17 and 20 min, respectively. For both of the dyes , for the lower peroxide concentrations (up to 10 mM H2O2), the total decoloration time decreases as the peroxide concentration increases, because with the increasing concentration of the peroxide, the concen-tration of the hydroxyl radicals also increases, as the peroxide is their source. Further increase in peroxide concentration to 20 mM causes the slight decrease in the decoloration time related to that in the presence of 10 mM H2O2, probably because of quenching reaction of hydroxyl radicals with H2O2 [24]. Thus, the optimal concentration for the process is assumed to be 10 mmol dm–3.

Figure 8. Effect of H2O2 concentration on the decoloration time of Crystal violet (1 mM Na2SO4, applied current density 10 mA cm–2).

All of the dyes degradation experiments were rep-eated several times, and after each of them, no traces of bismuth were detected in the solutions; the weight of the anodes after cleaning and drying remained cons-tant; no cracks of the films were observed under the microscope, which all indicates that they are electro-chemically and mechanically stable enough under the applied experimental conditions. The anodes with the films thicker than 10 μm were not tested, because they did not possess required mechanical qualities.

As it is well known, the electrodeposition offers a good control and reproducibility of the working para-meters and therefore, the properties of the deposited films; by the proper selection of the electrodeposition conditions, it is possible to obtain material with the desired properties and quality. In this case, it was shown that optimal electrodeposition time in the syn-thesis procedure was 5 min, i.e., that the anode obtained within this time exhibited the highest effi-

ciency for the dye decoloration under the applied expe-rimental conditions.

Further investigation and optimization of dye deg-radation parameters will probably improve the effi-ciency of the process. It would be also interesting to test them as the photo anodes. Since UV irradiation would increase their electrical conductivity and there-fore, the production of higher concentration of oxi-dative species which can degrade the dye molecules their surface, the increase of their efficiency is expected.

CONCLUSION

Bi2O3 based anodes were synthesized by electro-deposition of bismuth, followed by calcination to obtain Bi2O3 films. The films were deposited at constant current density, during various electrodeposition times, and their thickness varied from about 0.7 to 15 μm, depending on the electrodeposition time. Only the films with the thickness of up to 10 μm were mecha-nically stable enough to be tested as the anodes. All of the tested Bi2O3 anodes have shown the ability to degrade Reactive blue 19 and Crystal violet, with the different decoloration times. Bi2O3 film obtained within 1 min of electrodeposition was incomplete, with the significant part of the uncovered stainless steel sub-strate surface. Dyes decoloration with this anode pro-ceeded through electrocoagulation with anodically generated iron. Dyes decoloration with the rest of the tested Bi2O3 anodes proceeded thorough an oxidation with the oxygen species, which was generated from H2O2 decomposition at the anodes surface. The decol-oration time of the anode with Bi2O3 film of 1.5±0.3 μm thickness was longer than expected, which was attri-buted to dyes adsorption onto its surface during decol-oration process. The shortest decoloration time was achieved with the anode obtained during 5 min of elec-trodeposition, with the film thickness of 2.5±0.3 μm and this is assumed to be an optimal electrodeposition time in the synthesis procedure of the anodes for the purpose described in the paper. Dye decoloration times increased as the Bi2O3 film thickness increased above 2.5 μm. Effect of H2O2 concentration on the dyes decol-oration was investigated using the anode with the film thickness of 2.5±0.3 μm. H2O2 concentration affected the decoloration times. For both of the dyes, the opti-mal H2O2 concentration for the process is found to be 10 mmol dm–3.

Acknowledgement

The authors would like to thank the Ministry of Education, Science and Technological Development of the Republic of Serbia for supporting this work (Grant No ТR 34008).


594

REFERENCES

[1] M. Doble, A.K. Kruthiventi, Green Chemistry & Eng-ineering, Elsevier, Burlington, 2007, pp. 286.

[2] C.A. Martínez-Huitle, E. Brillas, Decontamination of wastewaters containing synthetic organic dyes by elec-trochemical methods: A general review, Appl. Catal., B 87 (2009) 105–145.

[3] D. Dogan, H. Türkdemir, Electrochemical oxidation of textile dye indigo, J. Chem. Technol. Biotechnol. 80 (2005) 916–923.

[4] X. Chen, G. Chen, F. Gao, P.L. Yue, High-Performance Ti/BDD Electrodes for Pollutant Oxidation, Environ. Sci. Technol. 37 (2003) 5021–5026.

[5] X. Chen, G. Chen, P.L. Yue, Anodic oxidation of dyes at novel Ti/B-diamond electrodes, Chem. Eng. Sci. 58 (2003) 995–1001.

[6] G. Chen, Electrochemical technologies in wastewater treatment, Sep. Purif. Technol. 38 (2004) 11–41.

[7] W. Chou, C. Wang, C. Chang, Comparison of removal of Acid Orange 7 by electrooxidation using various anode materials, Desalination 266 (2011) 201–207.

[8] J.L. Nava, M.A. Quiroz, C.A. Martínez-Huitle, Electroche-mical Treatment of Synthetic Wastewaters Containing Alphazurine A Dye: Role of Electrode Material in the Colour and COD Removal, J. Mex. Chem. Soc. 52 (2008) 249–255.

[9] J.M. Aquino, R.C. Rocha-Filho, N. Bocchi, S.R. Biaggio, Electrochemical Degradation of the Reactive Red 141 Dye on a β-PbO

2 Anode Assessed by the Response Sur-

face Methodology, J. Braz. Chem. Soc. 21 (2010) 324– –330.

[10] R. F.Yunus, Y. Zheng, K.G.N. Nanayakkara, J.P. Chen, Electrochemical Removal of Rhodamine 6G by Using RuO2 Coated Ti DSA, Ind. Eng. Chem. Res. 48 (2009) 7466–7473.

[11] K. Wang, M. Wei, T. Peng, H. Li, S. Chao, T. Hsu, H. Lee, S. Chang, Treatment and toxicity evaluation of methyl-ene blue using electrochemical oxidation, fly ash adsorption and combined electrochemical oxidation-fly ash adsorption, J. Environ. Menage. 91 (2010) 1778– –1784.

[12] P.A. Carneiro, C.S. Fugivara, R.F.P. Nogueira, N. Boralle, M.V.B. Zanoni, A Comparative Study on Chemical and Electrochemical Degradation of Reactive Blue 4 Dye, Portugaliae Electrochim. Acta 21 (2003) 49–67.

[13] M. Ihos, F. Manea, A. Iovi, Electrochemical Degradation of Aromatic Compounds at Modified SnO2 Anodes, Chem. Bull. Politehnica 54 (2009) 46–49.

[14] E. Brillas, E. Mur, R. Sauleda, L. Sànchez, J. Peral, X. Domènech, J. Casado, Aniline mineralization by AOPs: anodic oxidation, photocatalysis, electro-Fenton and photoelectro-Fenton processes, Appl. Catal. B 16 (1998) 31–42.

[15] A. Socha, E. Sochocka, R. Podsiadly, J. Sokolowska, Electrochemical and photoelectrochemical degradation of direct dyes, Color Technol. 122 (2006) 207–212.

[16] M. Hamza, R. Abdelhedi, E. Brillas. I. Sires, Comparative electrochemical degradation of the triphenylmethane

dye Methyl Violet with boron-doped diamond and Pt anodes, J. Electroanal. Chem. 627 (2009) 41–50.

[17] M. Muthukumar, M. Thalamadai Karuppiah, G. Bhaskar Raju, Electrochemical removal of CI Acid orange 10 from aqueous solutions, Sep. Purif. Technol. 55 (2007) 198–205.

[18] X. Chen, F. Gao, G. Chen, Comparison of Ti/BDD and Ti/SnO2–Sb2O5 electrodes for pollutant oxidation, J. Appl. Electrochem. 35 (2005) 185–191.

[19] G. B. Raju, M. T. Karuppiah, S.S. Latha, D. L. Priya, S. Parvathy, S. Prabhakar, Electrochemical pretreatment of textile effluents and effect of electrode materials on the removal of organics, Desalination 249 (2009) 167–174.

[20] M. Zidan, T.W. Tee, A.l.H. Abdullah, Z. Zainal, G.J. Kheng, Electrochemical Oxidation of Ascorbic Acid Mediated by Bi2O3 Microparticles Modified Glassy Carbon Electrode, Int. J. Electrochem. Sci. 6 (2011) 289–300.

[21] G. Li, H.Y. Yip, C. Hu, P.K. Wong, Preparation of grape-like Bi2O3/Ti photoanode and its visible light activity, Mater. Res. Bull. 46 (2011) 153–157.

[22] I. Zhitomirsky, Cathodic electrodeposition of ceramic and organoceramic Materials. Fundamental aspects. Adv. Colloid Interface. Sci. 97 (2002) 279–317.

[23] I. Valsiûnas, L. Gudavièiûtë, A. Steponavièius, Bi electro-deposition on Pt in acidic medium 1. A cyclic voltam-metry study, Chemija 16 (2005) 21–28.

[24] M.D. Radović, J.Z. Mitrović, D.V. Bojić, M.M. Kostić, R.B. Ljupković, T.D. Anđelković, A.Lj. Bojić, Uticaj parametara procesa UV zračenje/vodonik-peroksid na dekolorizaciju antrahinonske tekstilne boje, Hem. Ind. 66 (2012) 479– –486.

[25] N. Takeno, Atlas of Eh-pH diagrams, National Institute of Advanced Industrial Science and Technology, Tsukuba, 2005, pp. 47.

[26] M.I. Čekerevac, Lj.M. Nikolić-Bujanović, M.V. Simičić, Istraživanje elektrohemijskog postupka sinteze ferata, deo 1. Elektrohemijsko ponašanje gvožđa i nekih njego-vih legura u koncentrovanim alkalnim rastvorima, Hem. Ind. 63 (2009) 387–395.

[27] R.W. Revie, H.H. Uhlig, Corrosion and Corrosion Control, 4th ed., Jonh Wiley & Sons, Hoboken, NJ, 2008, pp. 46.

[28] P.R. Birkin, J.F. Power, M.E. Abdelsalam, T.G. Leighton, Electrochemical, luminescent and photographic charac-terisation of cavitation, Ultrason. Sonochem 10 (2003) 203–208.

[29] M. Zdujić, D. Poleti, Č. Jovalekić, Lj. Karanović, Mechano-chemical synthesis and electrical conductivity of nano-crystalline δ-Bi2O3 stabilized by HfO2 and ZrO2, J. Serb. Chem. Soc. 74 (2009) 1401–1411.

[30] H. Cheng, B. Huang, J. Lu, Z. Wang, B. Xu, X. Qin, X. Zhang, Y. Dai, Synergistic effect of crystal and electronic structures on the visible-light-driven photocatalytic performances of Bi2O3 polymorphs, Phys. Chem. Chem. Phys. 12 (2010) 15468–15475.

[31] H.A. Harwig, A.G. Gerards, Electrical properties of the α, β, γ, and δ phases of bismuth sesquioxide, J. Solid State Chem. 26 (1978) 265–274.

[32] H. Zollinger, Color Chemistry: Syntheses, Properties, and Applications of Organic Dyes and Pigments, 3rd ed., Wiley-VCH, Weinheim, 2003.


595

IZVOD

SINTEZA ANODA BAZIRANIH NA Bi(III)-OKSIDNIM FILMOVIMA ZA ELEKTROHEMIJSKU DEGARDACIJU BOJA REACTIVE BLUE 19 I CRYSTAL VIOLET

Milica M. Petrović, Jelena Z. Mitrović, Miljana D. Radović, Danijela V. Bojić, Miloš M. Kostić, Radomir B. Ljupković, Aleksandar Lj. Bojić

Prirodno–matematički fakultet Niš, Univerzitet u Nišu, Niš,Srbija

(Naučni rad)

Anode bazirane na tankim filmovima Bi2O3 su pripremljene elektrohemijskimtaloženjem bizmuta na podlozi od nerđajućeg čelika, pri konstantnoj gustini strujei tokom različitih vremena taloženja i potonjom kalcinacijom do Bi2O3. Debljine filmova su određene dvema metodama: posmatranjem pod mikroskopom sa mik-rometarskom skalom i na osnovu razlike u masi. Elektrohemijski procesi na ano-dama u prisustvu i odsustvu H2O2 ispitani su tehnikom linearne voltametrije. Ispi-tana je sposobnost anoda za obezbojavanje antrahinonske reaktivne boje Reac-tive blue 19 i trifenilmetanske boje Crystal violet elektrohemijskom oksidacijom.Film Bi2O3 na anodi dobijenoj u toku 1 minuta elektrotaloženja je bio nepotpun iobezbojavanje na njoj se odvijalo elektrokoagulacijom jonima Fe koji su dobijenianodnim rastvaranjem nepokrivenih delova površine nerđajućeg čelika. Debljineanoda dobijenih u toku: 2, 5, 10 i 15 min elektrotaloženja iznosile su: 1,5±0,3, 2,5±0,3, 5,6±0,5 i 9,6±1 μm, redom. Obezbojavanje u prisustvu tih anoda odvijalose oksidacijom •OH, dobijenim razlaganjem H2O2 na anodama na visokim poten-cijalima. U odsustvu H2O2 nije bilo merljivog obezbojavanja. Sve navedene Bi2O3anode su bile mehanički i elektrohemijski stabilne u uslovima obezbojavanja. Sposobnost anoda da generišu •OH opada s poratom debljine Bi2O3 filmova, pa je bilo očekivano da na isti način opada i njihova sposobnost obezbojavanja, tj. da seprodužuje vreme obezbojavanja. Ipak, vreme obezbojavanja na anodi debljine 1,5 μm je bilo duže od očekivanog, i ta pojava je pripisana adsorpciji boja na njenojpovršini, čime su verovatno blokirana aktivna mesta za generisanje •OH. Na osta-lim anodama, debljina 2,5, 5,6 i 9,6 μm nije bilo adsorpcije boja i kod njih su vremena obezbojavanja rasla sa porastom debljine Bi2O3 filma, ali se ne može sa sigurnošću tvrditi da je debljina filma glavni uzrok te pojave; verovatno je da i kris-talna struktura Bi2O3 filmova ima uticaja. Najkraće vreme obezbojavanja postig-nuto je sa anodom debljine 2,5 μm (17±1 min za Crystal violet i 35±1 min za Reactive blue 19). Vremena obezbojavanja dve boje su bila različita na svakojanodi, zbog razlike u molekulskoj strukturi boja. Bi2O3 filmovi deblji od 10 μm nisu bili dovoljno mehanički stabilni da bi te anode bile testirane. Uticaj koncentracijeH2O2 na proces obezbojavanja je ispitan na anodi debljine 2,5 μm. Koncentracija H2O2 je uticala na vremena obezbojavanja i nađeno je da optimalna koncentracijaH2O2 u procesu obezbojavanja iznosi 10 mmol dm–3 za obe boje.

Ključne reči: Bizmut(III)-oksid • Anode •Sinteza • Debljina filma • Obezbojavanje• Vodonik-peroksid

597

Моделовање утицаја температуре и времена хомогенизационог жарења на тврдоћу PdNi5 легуре

Александра Т. Ивановић1, Бисерка Т. Трумић1, Светлана Љ. Иванов2, Саша Р. Марјановић2 1Институт за рударство и металургију Бор, Бор, Србија

2Универзитет у Београду, Технички факултет у Бору,Бор, Србија

Извод

У овом раду датa је анализа тврдоће легуре PdNi5 након жарења, ради одређивањаоптималних услова термо-механичког режима прераде. Варирани су температура ивреме жарења, а као одговор система посматрана је вредност тврдоће. Применом пуног факторског плана експеримента, типа 32, и анализом добијених експеримен-талних података, дефинисан је математички модел функције одзива система којиверодостојно описује експериментално добијене вредности (R2 ˃ 0,95). Израчунати су ефекти главних варијабила а регресиона анализа је коришћена за фитовање одговорасистема.

Кључне речи: PdNi5, експериментални дизајн, Pd катализатори-хватачи.

НАУЧНИ РАД

УДК 519.86/.87:66:662.767

Hem. Ind. 68 (5) 597–603 (2014)

doi: 10.2298/HEMIND130620085I


У многим инжењерским експериментима, неза-висно да ли се обављају у строго контролисаним лабораторијским или погонским условима, истра-жују се производни услови која ће дати оптимум у квалитету производа и његовој економичности. Зависност резултата експеримента од улазних фак-тора може се оптимизирати применом методе одзивне површине (Responce Surface Methodology –RSM). Ова метода представља скуп математичких и статистичких поступака која се користи за форми-рање емпиријских модела и анализу процеса. Циљ је да се на основу пажљиво одабраног плана експе-риманта оптимизују одзиви система (y) који зависе од више независно променљивих фактора (x1, x2,..., xk) [1]. Основна предност коришћења RSM је у сма-њивању броја некада врло скупих експеримената, као приликом истраживања легура из система Pd-Ni, уз обезбеђивање довољног броја информација за одређивање статистички валидних резултата. Општи облик одзивне функције гласи:

y = f(x1, x2,..., xk) + ε (1)

где je ε представља укупну грешку експеримента која садржи случајне грешке мерења и ефекте слу-чајних спољних утицаја који нису обухваећни функ-цијом f. Ова грешка заправо, представља разлику између стварне (измерене) и функционалном завис-ношћу (моделом) израчунате вредности за исте нивое улазних фактора. Апроксимација ће бити уто-лико боља, уколико је грешка мања. Обе компо-

Преписка: Институт за рударство и металургију Бор, Зелени булевар 35, Бор, Србија. E-пошта: [email protected] Рад примљен: 20. јун, 2013 Рад прихваћен: 8. новембар, 2013

ненте грешке експеримента су случајне величине са нормалном расподелом око тачне (истините) вред-ности με = 0 са дисперзијом εσ 2 . Тако је за сваки ниво улазних фактора потребно обавити више поно-вљених мерења како би се добио податак о грешци. Колики ће бити број поновљених мерења за исти ниво фактора зависи од услова експеримента, ње-гове цене и тражене сигурности резултата.

Избор адекватног плана експеримента је кри-тична тачка у примени RSM методе. Најчешће се у истраживањима користе следећи експериментални планови: Box-Behnken дизајн (BBD) [2], централни композитни дизајн (CCD) [3], потпуни факторски дизајн [4] и др. Облик одзивне функције приказује се у облику полинома обзиром да се степен фито-вања експерименталних података може побољшати повећањем степена полинома. Најчешће се приме-њују полиноми првога реда који у обзир узимају само линеарне утицаје појединачних фактора. Ако је крајњи циљ поступка оптимизација процеса, нео-пходно је користити полиноме другог и вишег реда. Тада сваки од улазних фактора мора имати три или пет нивоа вредности у зависности од примењеног експерименталног плана. Приступ експериментал-ном истраживању, уз коришћење планираног експе-римента представља квалитативно нов приступ у теоријско–експерименталној анализи и оптимиза-цији сложених процеса (система и објеката), са уни-верзалном применом и низом предности у односу на класичан пристп експерименталном истражи-вању. Применом планираног експеримента оства-рује се: минимални број потребних (серија) опита, максимална количина информација из датог броја опита, сукцесивно извођење експеримента (корак по корак), идући од једноставних ка сложенијим плановима, једноставна статистичка (регресиона и

А.Т. ИВАНОВИЋ и сар.: УТИЦАЈ ХОМОГЕНИЗАЦИОНОГ ЖАРЕЊА НА ТВРДОЋУ PdNi5 ЛЕГУРЕ Hem. ind. 68 (5) 597–603 (2014)

598

дисперзиона) анализа експерименталних резултата, као и могућност квалитативне и квантитативне оце-не дејства сваког утицајног фактора (и евентуално, њихових интеракција) на одзив система.

Лака оптимизација процеса (система) који је предмет истраживања, на основу добијеног емпи-ријског (регресионог) модела одзивне функције, која обухвата цео експериментални простор, као и минимизација трошкова за реализацију експери-мента, елиминисање субјективног утицаја експери-ментатора, итд. су само неке од предности које пружа палнирани експеримент. Стога, примена пла-нираног експеримента у свим гранама науке зна-чајно расте у последњих двадесет година. Преглед литературе указује да се методологија планираног експеримента уз каснију статистичку анализу, ко-ристи како за повећање фундаменталног знања о металима и легурама [5–12], тако и за оптимизацију процесних параметара [13,14].

Фазни дијаграм бинарног система Pd-Ni пред-ставља дијаграм стања са потпуном растворљи-вошћу компонената у чврстом стању, са минимумом на кривој ликвидус и солидус (1273 °C при 45 at.% Pd) [15]. Ова легура се због тога назива псеудоеу-тектична. Сређеност великог домета у легурама сис-тема Pd–Ni није пронађена, док је сређеност кратког домета присутна на собној температури током изо-хроног жарења[16]. Магнетна трансформација почи-ње на страни Ni па све до Pd стране при конс-тантном смањењу температуре. Проучавано је еле-ктрично и магнетно понашање легура система Pd–Ni [17,18]. Структура и стање површине Pd–Ni легура проучавани су различитим техникама. У легурама овог система изражена је тенденција ка сегрегацији паладијума на нижим темпеаратурама и при нижим садржајима паладијума [19]. Термодинамичка свој-ства легура система Ni–Pd истраживали су Bidwell и Speiser [20]. Испитивана су термодинамичка својства овог система у температурном опсегу 700–1200 °C. На бази ових истраживања, дошло се до закључка да је ентропија мешања позитивна за све саставе и да је ово последица феромагнетних особина легура овог система. Kasprzak и сар. [21] опажају да се током дифузионог жарења микроструктура транс-формише у потпуности уз формирање сферних зрна за разлику од постојеће са издуженим зрнима. Фазне трансформације у систему Pd-Ni током загре-вања и брзог хлађења су проучаване у раду [22]. Установљено да је формирање површински центри-ране кубне решетке могуће при спором хлађењу. Легура PdNi5 налази примену у изради паладијум-ских катализатора за хватање платине у процесу катализе на високим температурама.

Избор нивоа фактора ограничен је законитос-тима дифузионих процеса који се одвијају у легу-

рама приликом загревања доводећи до тога да се ови процеси завршавају за релативно кратко време на високим температурама, док се при ниским тем-пературама ови процеси одигравају врло споро и захтревају дуже време [23].

Циљ овог рада је анализа утицаја температуре и времена жарења на тврдоћу жарене легуре, као и дефинисање математичког модела који даје везу између улазних фактора и одговора система.

ЕКСПЕРИМЕНТАЛНИ ДЕО

За израду узорака коришћен је прах паладијума чистоће 99,99% и никал у облику танких лимова чис-тоће 99,95%. Садржај никла у узорцима износио је 5 мас.%. Полазне сировине су најпре испресоване на хидрауличној преси у циљу постизања боље ком-пактности материјала, а затим је извршено топљење и ливење узорака у средње фреквентној индукцио-ној пећи, у лонцу од MgO, у вакууму. Температура топљења легуре PdNi5 је 1520 °C. Пре ливења, шар-жа се прегрева за 350–400 °C. Термичка обрада узо-рака извршена је коморној електроотпорној пећи типа LP08. Мерење тврдоће извршено је на комби-нованом апарату за мерење тврдоће по Викерсу и Бринелу произвођача WPM (Werkstoffprüfmasch-inen), Немачка, са опсегом мерења тврдоће од 5 до 250 kP.

Корак који по аналогији претходи процесу изво-ђења експеримента јесте избор експерименталног модела по којем ће се експеримент извести и резул-тати математички обрадити. У овом раду примењен је пун факторски план експеримента, типа 32, у којем сваки од два улазна фактора има по три нивоа. Уопштено, примена факторског плана експе-римента дозвољава варирање фактора у широким опсезима од значаја за испитивани процес. У овом раду, избор параметара оптимизације (температура (t у °C) и време (τ у min) условљен је и техничко– –технолошким као и економским разлозима [24]. Температура жарења износила је 800, 900 и 1000 °C, док је време жарења било 30, 60 и 90 min. Овим правим вредностима фактора додељене су коди-ране вредности (нивои фактора) –1, 0 и 1. Као одзив мерена је тврдоћа легуре Y_HV.

За статистичку обраду резултата коришћен је програмски пакет SPSS Statistics [25]. Помоћу наве-деног софтвера израђени су математички модели за описивање утицаја улазних фактора при процесу жа-рења на тврдоћу легуре PdNi5. Прикладност модела одређена је помоћу вредности F-статистике у ANOVA тесту, као и вредности коефицијента детер-минације, R2.


599

РЕЗУЛТАТИ И ДИСКУСИЈА

У табели 1 приказани су утицајни фактори, њи-хови нивои, као и одзив система при различитој комбинацији улазних фактора.

Због лакшег формирања матрице плана експери-мента, физички фактори, Xi, преводе се у бездимен-зионе величине на основу следећих једначина:

X1/30 – 2=x1 (1) X2/100 – 9 = x2 (2)

где су: x1, x2 – кодиране вредности времена и тем-пературе, редом, а X1, X2 – праве вредности вре-мена и температуре, редом.

Како је уочено да независно променљиве узро-кују нелинеаран одговор система, примењена је ме-тода одзивне површине (RSM), што је омогућило оптимизацију процеса [1]. За оптимизацију процеса жарења легуре PdNi5, а у циљу добијања оптимал-них карактеристика за процес даље пластичне пре-раде, експеримeнтални подаци фитовани су поли-номом другог реда облика:

β β β β β β= + + + + +2 20 1 1 2 2 11 1 22 2 12 1 2y x x x x x x (3)

Коефицијенти регресије, βi и βij, се одређују при-меном методе најмањих квадрата. Вредности ових коефицијената добијене коришћењем програмског пакета SPSS Statistics дате су у табели 2.

Резултати из табеле 2 показују да приликом жа-рења легуре PdNi5 оба линеарна члана имају статис-тички значајан утицај на формирање модела одзива Y на нивоу p < 0,05. Температура има већи утицај на промену Y у односу на утицај времена трајања про-цеса. На формирање модела одзива статистички значајан утицај (на нивоу p < 0,05) има и квадратни члан температуре. Квадратни члан времена трајања процеса, као и члан интеракције улазних фактора, немају статистички значајан допринос предикцији зависне променљиве.

Узимајући у обзир само статистички значајне параметре, добија се следећи математички модел којим се описује утицај улазних фактора (темпера-туре и времена жарења) на одзив система (тврдоћа легуре PdNi5):

Y = 76,478 – 0,950X1 – 1,617X2 + 0,983X2X2 (4)

На основу овако дефинисане математичке завис-ности излазне од улазних променљивих, могуће је предвидети вредност одзива уколико су познате вредности улазних величина. Вредности VIF (фак-тора повећања варијансе) при p < 0,05 (Табела 2), указују на задовољавајућу статистичку поузданост резултата [26]. Адекватност модела тестирана је помоћу ANOVA теста. Резултати ANOVA теста разви-јеног модела су приказани у табели 3.

Табела 1. План експеримента и одзив система Table 1. Experimental design and system response

Број експеримента Кодиране вредности Улазне величине Одзив система x1 x2 Време, min Температура, °C Тврдоћа, HV

1 –1 1 30 1000 76,30 2 –1 0 30 900 77,10 3 –1 –1 30 800 79,60 4 0 1 60 1000 75,90 5 0 0 60 900 76,20 6 0 –1 60 800 79,30 7 1 1 90 1000 74,50 8 1 0 90 900 75,30 9 1 –1 90 800 77,50

Tabela 2. Вредности коефицијената регресије Table 2. The values of the regression coefficients

Model Unstandardized coefficients Standardized coefficient t p Semipartial correlation coefficients

Collinearity StatisticsB Standard error Beta Tolerance VIF

Const. 76,478 0,193 395,472 0,000 – 1,000 1,000 X1 –0,950 0,106 –0,479 –8,969 0,003 –0,479 1,000 1,000 X2 –1,617 0,106 –0,815 –15,263 0,001 –0,815 1,000 1,000 X1X2 0,075 0,130 0,031 0,578 0,604 0,031 1,000 1,000 X1X1 –0,417 0,183 –0,121 –2,271 0,108 –0,121 1,000 1,000 X2X2 0,983 0,183 0,286 5,360 0,013 0,286 1,000 1,000


600

Анализа варијансе регресионе једначине потвр-ђује да се на нивоу значајности од 95% (α = 0,05), применом одабране регресионе једначине може предвидети понашање тврдоће легуре PdNi5, током хомогенизационог жарења, при промени времена и температуре жарења (Израчунато F ˃ таблично F(0,05;5;3) = 5,41).

Коефицијент детерминације, R2, који представља процену укупне варијације података објашњених према моделу, износи 0,976 (Табела 4), односно 98% варијансе у резултатима тврдоће легуре PdNi5 је објашњено моделом (4).

Tabela 4. Вредновање модела Table 4. Model summary

Модел R R2 Кориговано R2 Стандардна грешка процене

1 0,988 0,976 0,961 0,33813

На слици 1 приказан је график зависности изра-чунатих и експерименталних вредности.

Према референтној литератури [27], полиномски модели са R2 > 95% могу се узети као задовоља-вајући резултат нелинеарне вишеструке регресије Коефицијент детерминације коначног модела са вредношћу R2 = 0,976 индикује одлично слагање

експерименталних и моделом предвиђених вред-ности.

Одзивна површина (слика 2) приказује зависност одзива система, тврдоћа легуре Y_HV, од промене улазних величина система, времена трајања про-цеса и температуре на којој се процес одвија. Тренд смањивања вредности тврдоће уочен је при порасту времена одвијања процеса и температуре на којој се процес жарења одвија.

Поједине механичке особине различито се ме-њају под утицајем дифузионих процеса који се оди-гравају при жарењу [23,28,29]. Код очвршћавања одливка у неравнотежним условима (као у овом раду), долази до формирања дендритске кристалне структуре одливка, а у самом кристалном зрну до-лази до знатних разлика у хемијском саставу [29]. Меке, тј. никлом сиромашније, гране дендрита јасно се разликују од тврдих, никлом богатијих, међуден-дритских области. Таква неравномерност у струк-тури се може делимично одстранити захваљујући дифузионом процесу приликом загревања матери-јала до температуре испод солидус линије и задр-жавањем на тој температури одређено време, а у циљу побољшања пластичних особина легуре. Тем-пература хомогенизације има такорећи одлучујући утицај на промену особина легуре [29]. Одливак, у овом раду, после ливења има тврдоћу 98 HV. При

Tabela 3. Резултати ANOVA теста коначног модела другог редаTable 3. Results of ANOVA test of the final model of second order

Извор варијабилитета Сума квадрата одступања Број степени слободе Средњи квадрат

одступања F-Тест р-вредност

Модел 23,031 3 7,677 67,144 0,000 Остатак 0,572 5 0,114 Укупно 23,602 8

Slika 2. Зависност између експериментално добијених вредности тврдоће и вредности израчунатих применом једначине (1).Figure 2. Correlation between the experimentally obtained hardness values and the values calculated using Eq. (1).


601

загревању на 800 °C у трајању од 30 min, тврдоћа већ опада на 79,6 HV, док задржавање од 90 min резултује у још нижој вредности тврдоће од 77,5 HV. Са даљим порастом температуре, тврдоћа опада, тако да на 1000 °C вредности тврдоће су: 76,3; 75,9 и 74,5 HV за 30, 60 и 90 min. Време хомогенизације има мањи утицај од температуре, али се ипак зах-тева одређивање оптималног времена хомогени-зације. Све наведене тврдоће добијене су каљењем одливака у води након хомогенизације.

На основу регресионог модела могу се једнос-тавно поставити параметри хомогенизационог жа-рења који ће обезбедити захтеване вредности твр-доће легуре PdNi5 зависно од подручја примене [30]. Температуре и времена која се могу користити у процесима термичке обраде одређене су како тех-нолошким тако и економским разлозима [23]. Са-гледавајући економске и технолошке разлоге као оптимални параметри за даљу палстичну прераду изабрани су: температура жарења 900 °C, време жарења 30 min.

ЗАКЉУЧАК

Планирање експеримента и статистичка анализа примењени су у анализи ефеката параметара жаре-ња који утичу на тврдоћу PdNi5 легуре. Кориш-ћењем пуног факторског плана експеримента испи-тан је и упоређен утицај промена процесних вари-јабли (температура и време жарења). На формира-ње математичког модела промене тврдоће статис-тички значајан утицај (на нивоу p < 0,05) имају ли-неарни чланови улазних променљивих (темпера-туре и времена жарења), као и квадратни члан тем-пературе. Квадратни члан времена трајања процеса, као и члан интеракције температуре и времена,

немају статистички значајан допринос предикцији тврдоће. Дефинисан је емпиријски математички модел погодан за описивање процеса, а на основу кога се може предвидети тврдоћа унутар изабраних нивоа температуре и времена.

Захвалница

Резултати истраживања приказани у овом раду су резултат пројекта технолошког развоја ТР 34029 „Развој технологије производње Pd катализатора-хватача за смањење губитака платине у високотем-пературним процесима катализе“, финансиран од стране Министарства просвете, науке и технолошког развоја Републике Србије.

ЛИТЕРАТУРА

[1] D. C. Montgomery, Design and Analysis of Experiments: Response surface method and designs, John Wiley and Sons, Hoboken, NJ, 2005.

[2] Y. Feng , Z. Cai , H. Li , Z. Du, X. Liu, Рesponse surface optimization of fluidized roasting reduction of low-grade pyrolusite coupling with pretreatment of stone coal ,J. Min. Metall. Sect. B-Metall. 49 (2013) 33–41.

[3] S. Hari Vignesh, Т. Selvaraj, Оptimization of tempera-ture, tool wear and surface finish in turning of 6063 alu-minium alloy using RSM, IJERT 2 (2013) 2535–2541.

[4] I.M. Savić, G.S. Nikolić, I.M. Savić, M.D. Cakić, A. Dosić, Janoš Čanadi, Modelovanje stabilnosti bioaktivnog bakar(II) kompleksa primenom eksperimentalnog di-zajna, Hem. Ind. 66 (2012)693–699.

[5] S. Ivanov, B. Kočovski, B. Stanojević, Ocena uticaja termomehaničkih parametara prerade bakarne žice na izduženje spirale primenom faktornog eksperimenta, Metalurgija 2 (1996) 13–23.

Време, min. Температура, 0C

Тврд

оћа,

HV

Слика 2. Зависност тврдоће легуре PdNi5 од температуре и времена жарења. Figure 2. The dependence of hardness alloy PdNi5 of temperature and annealing time.


602

[6] D. Indumati, G. Purohit, Prediction of hardness of forged Al7075/Al2O3 composites using factorial design of expe-riments, IJERA 2 (2012) 084–090.

[7] S. Hashmi, K. Majumdar, N. Chand, Application of fac-torial design of experiments to the quantitative study of tensile strength of red mud filled PP/LLDPE blends, J. Мater. Sci. Lett. 15 (1996) 1343–1345.

[8] S. Ivanov, B. Stanojević, The influence of density of pres-sed iron powder samples on the quality of boride layers, Sci. Sinter. 2 (2003) 93–98.

[9] Yu. M. Dolzhanskii, Use of the simplex method to select the optimal heat treatment, Met. Sci. Heat Treat. 19 (1977) 652–655.

[10] V.V. Cherkasov, N.S. Postnikov, Yu.M. Dolzhanskii, V.I. Kostyunin, Use of statistical methods for investigation the properties of cast aluminum alloys, Met. Sci. Heat Treat. 17 (1975) 506–508.

[11] E. Požega, S. Ivanov, V. Conić, B. Čađenović, Mogućnost procesa boriranja na presovanim uzorcima od železnog praha, Hem.Ind. 63 (2009) 253–258.

[12] S. Ivanov, E. Požega, Influence of the composition of the boroning mixture on the dimension change of pressed and boroned samples from iron powder, Sci. Sinter. 40 (2008) 197–205.

[13] I.S. Kim, Y.S. Yang, P.K.D.V. Yaragada, Sensitivity analysis for process parameters in GMA welding processes using a factorial design method, Int. J. Mach. Tool. Manu. 43 (2003) 763–769.

[14] I. Mihajlović, N. Štrbac, P. Đorđević, A. Ivanović, Ž. Živ-ković, Technological process modeling aiming to imp-rove its operations management, Serb. J. Manage. 6 (2011) 135–144.

[15] http://resource.npl.co.uk/mtdata/phdiagrams/nipd.htm [16] P.V. Petrenko, A.V. Gavrilyuk, N.P. Kulish, N.A. Mel’ni-

kova, Yu.E. Grabovskii, Structural Changes During Annealing of Deformed Pd-25 at.% Ni Alloy, Phys. Met. Metallography 108 (2009) 449–454.

[17] H. Takahashi, S. Fukatsu, S. Tsunashima, S. Uchiyama, Perpendicular Magnetic Anisotropy of Pd/Co- and

Pd/Ni-multilayers, J. Magn. Magn. Mater. 3 (1992) 1831–1832.

[18] A Tari, B.R. Coles, Electrical Resistivity and the Transition TO Ferromagnetism in the Palladium-nickel Alloys, J. Phys. F: Metal Physics 1 (1971) L69.

[19] S. Helfensteyn et al., Modelling Surface Phenomena in Pd–Ni Alloys, Appl. Surf. Sci. 212–213 (2003) 844–849.

[20] L.R Bidwell, R. Speiser, The Relative Thermodynamic Properties of Solid Nickel-palladium Alloys, Acta Metall. 13 (1965) 61–70.

[21] M. Kasprzak, D. Baither, G. Schmitz, Diffusion-induced Recrystallization in Nickel/palladium Multilayers, Acta Mater. 59 (2011) 1734–1741.

[22] S. Ozdemir Kart, M. Tomak, M. Uludogan, T. Cagin, Molecular Dynamics Studies on Glass Formation of Pd-Ni Alloys by Rapid Quenching, Turk J Phys. 30 (2006) 319–327.

[23] S. Ivanov, Lj. Ivanić, D. Gusković, S. Mladenović, Opti-mizacija režima starenja legura na aluminijumskoj os-novi, Hem. Ind. 66 (2012) 601–607.

[24] G. Slavković, B. Trumić, D. Stanković, Prognoze cena metala platinske grupe u proizvodnji katalizatorskih mreža i hvatača, Rudarski radovi, Bor 2 (2011) 181–192.

[25] SPSS inc. PASW Statistics 18, Predictive Analysis Soft-ware Portfolio (www.spss.com)

[26] I. Đjurić, P. Đjordjević, I. Mihajlović, Dj. Nikolić, Ž. Živ-ković, Predction of Al2O3 leaching recovery in the Bayer process using statistical multillinear regresion analysis, J. Min. Metall., Sect. B-Metall. 46 (2010) 161–169.

[27] И. Михаловић, Ђ. Николић, А. Јовановић, Теорија система, Технички факултет у Бору, Бор, 2009.

[28] S.D. Liu, Y.B. Yuan, C.B. Li, J.H. You, X.M. Zhang, Met. Mater. Int. 18 (2012) 679–683.

[29] H. Schumann, Metallographie, VEB Deutscher Verlag für Grundstoffindustrie, Leipzig, 1975.

[30] C. Hagelüken, Recycling the platinum group metals: A European perspektive, Platinum Metals Rev. 56 (2012) 29–35.


603

SUMMARY MODELING THE EFFECTS OF TEMPERATURE AND TIME OF HOMOGENIZATION ANNEALING ON THE HARDNESS OF PdNi5 ALLOY Aleksandra T. Ivanović1, Biserka T. Trumić1, Svetlana Lj. Ivanov2, Saša R. Marjanović2

1Mining and Metallurgical Institute Bor, Zeleni bulevar 35, Bor, Serbia 2University of Belgrade, Technical Faculty in Bor, VJ 12, Bor, Serbia

(Scientific paper)

Experimental design methodology represents a powerful tool for the analysis and optimization of various processes. PdNi5 alloy is used in the in the production process of nitric acid, as Pd-catalyst-trap. The role of Pd-catalyst-trap consists in reduction of volatile platinum oxide from gas flow to the metal form and reten-tion of platinum metal on the surface of Pd catalyst-trap. Fundamental knowledge about this alloy and its practical use in reduction of volatile platinum oxide can be improved by experimental planning and statistical analysis. In this study, theeffects of annealed temperature and annealed time on the hardness of PdNi5 alloy were evaluated and compared. Full factorial experimental design at three levels was applied. Statistically significant factors were determined considering of hardness of PdNi5 alloy as a system response. By regression analysis, the mathe-matical model for process description was derived. The correlation between pre-dicted and experimental values was high (R2 = 0.976). In the investigated ranges of parameters, the obtained empirical equation can be applied for the prediction of system response.

Keywords: PdNi5 • Experimental design •Pd-catalyst-trap

605

The examination of the seasonal influence on the efficiency in oil and fats removal through primary treatment from the wastewater of edible oil industry

Tatjana Nikolin, Mirjana Sevaljević

Technical college of Applied Sciences in Zrenjanin, Serbia

Abstract This paper investigates the influence of the seasonal change of the air temperature, che-mical oxygen demand, as well as efficiency of suspended matter removal, on the efficiencyof oil and fats removal (η, %) during primary treatment. The parameters are monitored in the period of time from 2006 to 2011. The efficiency of oil and fats removal in the first andin the fourth quarter is proportional to the efficiency of the removal of suspended matterand of total organic matter, measured as chemical oxygen demand (COD). The measured values for oil and fat are: η (IV quarter), 0.96 %–50.8%, and η (I quarter ), 5.06 %–95.97%. The efficiency of oil and fats removal in the second and third quarter is proportional to airtemperature, so the measured efficiency of fat and oil removal is: η (II quarter), 3.93 %–82.86%,and η (III quarter),6.82%–71.51%. The results of investigation have shown the existence of thecorrelation between the air temperature during various seasons and η, as well as the removal of the suspended matter and COD.

Keywords: removal efficiency, wastewater monitoring, primary treated water, edible oiland fat industry, air temperature, seasonal variation, oil and fats.

SCIENTIFIC PAPER

UDC 628.3.034.2:664.3

Hem. Ind. 68 (5) 605–613 (2014)

doi: 10.2298/HEMIND130906089N


The waste water from food industry contains the various contaminates which contaminate the surface and ground water in urban areas. Its concentration and nature depend on technological procedure, raw mate-rials and final products. After the waste water purifi-cation, it can be left in the recipient without any con-sequences or it can be used again. Each user of canal-ization has to remove the damaging and dangerous matters from waste water up to the limited values regulated by law, which can be achieved by convent-ional purification methods. The treatment procedures after previous treatment are [1]: primary, secondary, tertiary treatment and processing of the sludge that is the product of the primary treatment.

Primary purification of wastewater includes the equalization of content and nature of all gathered wastewaters the removal of large material from waste-water and fast sedimented suspended matter, free fat and oil with various mechanisms: sedimentation [2], coagulation, flocculation [3–5]. These procedures fre-quently are defined as the preliminary or pretreatment steps with aim to facilitate the purification. Only the suspended matter removal is considered to be primary treatment. The edible oil industry AD “Dijamant” Zre-njanin releases its own waste water into the canal-

Correspondence: T. Nikolin, Technical college of Applied Sciences in Zrenjanin, Đorđa Stratimirovića 23, Zrenjanin, Serbia. E-mails: [email protected] Paper received: 6 September, 2013 Paper accepted: 10 December, 2013

ization after its purification on plant for the primary treatment.

Secondary purification removes the colloidal par-ticles and a part of dissolved organic matters through the biological or chemical methods [4–9].

The purpose of tertiary treatment is to provide a final treatment stage to further improve the effluent quality before it is discharged to the receiving environ-ment (sea, river, lake, wet lands, ground, etc.).

In the last years the new methods are developed for oil water purification such as alternative cooling and heating [10] and microwaves radiation [11].

Previous investigations examined the correlations between the efficiency of oil and fats removal with its initial concentrations as well as with the concentrations of other pollutant contents, suspended particles, phosphate, iron for the period of time between 1995 and 1999 with coagulant Al-sulphate. The results of monitoring during this period showed that the mecha-nism of removal of oil and fats is in correlation with the mechanism of suspended particles removal. They also show that the efficiency of removal is influenced by the temperature gradient of air bubbles’ input and output of the treatment plant [12]. Also the examined correl-ation between the seasonal water temperature and content of the indicators of self-purification (turbidity, chemical oxygen demand, COD and pH) show the same content minimum and maximum, as indicator of self- -purification in the lakes with 1/3 of primary purified river Begej water in Water Treatment Plant “Begej” [13] (Table 1 and Fig. 1).

T. NIKOLIN, M. SEVALJEVIĆ: THE EFFICIENCY IN OIL AND FATS REMOVAL Hem. ind. 68 (5) 605–613 (2014)

606

The fitted monitoring data in the given figures (Fig. 1) [13] indicate that the ratio between water and air temperatures influence on self-purification indicators content also as indicators of processes enable tempe-rature relaxation in contact surface between two phases. In the I quarter (winter), lower air then water temperature (3.5 °C) favors endothermic processes. Endothermic dehydration of organic macromolecules causes its coagulation, turbidity decreasing and pH increasing. In the II quarter (spring) approximately equal water and air temperatures (9–13 °C) cause parallel endothermic and exothermic temperature rel-axation processes and increase the values of indicators content up to temperature of 13 °C. In the III quarter (summer), higher air then water temperature (22 °C) causes exothermic temperature relaxation processes in water, i.e., organic molecules hydration up to maxi-mum of COD, suspended matter and pH decreasing [13].

The seasonal difference between river water and air temperatures is measured in other papers [14]. The self-purification of atmospheric water with spontane-ous sedimentation in petrochemical refinery [15,17] and ground row water and chlorinated from wells of waterfall in town Zrenjanin [16] were examined also depending on seasonal water temperatures. The sea-

sonal temperature has influence on the turbidity maxi-mum in February, June and October (in minimum clari-fication velocity of and self-purification) and it also indicated increasing phosphate content from January up to June (at stationary iron content) and decreasing from August to December (during increasing iron con-tent) in water of waterfall of Zrenjanin [18].

The results obtained in the previous studies [12–18] are in agreement with the correlations obtained in this paper which indicate:

– Oxygen influence to suspended matter and COD removal efficiency in oxidation processes mostly, namely in oil and fats degradation processes,

– as well as to oxygen influence in the thermal, mechanical and chemical potential relaxation pro-cesses.

The edible oil industry has waste water as its own permanent problem. The sources of organic matter which contaminate the wastewater the most are all plants for oil treatment in any manner. The oils and fat decrease the oxygen solubility and influence on the all steps in technological purification procedures.

The results of the investigation of the processes in primary treatment of oiled waste water are presented in this paper.

Table 1. The represented measured data stand for the stationary seasonal average water temperatures T, pH, turbidity and COD, for lakes I and III which contain 1/3 of added primary purified water [13]

The month Parameter

T / °C pH Turb., NTU COD, mg/l January 2004 4.0 – 6.0 3.2 February2004 1.4 7.6 6.0 4.6 March 2004 3.4 8.0 3.0 3.3 April 2004 9.3 8.3 5.0 3.5 June2004 13.5 8.1 12.0 8.2 June 2005 21.2 8.4 12.0 3.3 p.p. June 2005 21.7 8.4 10.0 22.4 Sep. 2005 27.0 8.3 4.0 15.9 p.p. Sep. 2005 22.6 7.5 18.0 16.5

Figure 1. The seasonal water temperatures have an influence on COD, pH and turbidity for the lakes I and III [13].


607

The sources of contamination of waste water are the part for bottles filling, oil extraction, neutralization, hydrolysis and production of fat acids, refinement, wash waters, water from ions exchange and various washing. The cooling water is considered to be uncon-taminated and can be released directly in canalization, but all other have to be purified previously in the plant for primary treatment.

In the edible oil industry the next technology for primary treatment is applied in the successive steps [19]:

• equalization • gravitational separation of the free fats • coagulation • flocculation • flotation • correction of water pH value • sludge centrifugation. Based on the monitored results, the possible domi-

nant seasonal influence of suspended matter, air tem-perature and chemical oxygen demand on oil and fats removal is examined in this paper due to promotion of the purifying process.

At the stationary beginning concentrations of the solid particles and contaminate content (fats and oil, suspended matter, chemical oxygen demand) in waste-water on input represent the relative difference of contaminants taken simultaneously on input and out-put from primary plant within the two-hour period and it corresponds to the contaminants removal efficiency and can be expressed in percents:

η−

= ×input output

input(%) 100

c cc

(1)

The correlations between the removal efficiencies of the fats and oil, suspended matter, chemical oxygen demand, as well as with the relative change of tem-perature, are examined in this paper.

The knowledge of the dominant physico-chemical mechanisms of complex purification processes (dehyd-ration, adsorption, oxidation, coagulation, flocculation, sedimentation, diffusion, hydration, dissolution and neutralization), as well aeration (oxidation, flotation and degasation) [20–22], make the optimization of purifying processes possible.

Which of the mechanisms will be dominant is based on the statistical correlation coefficients of the oil and fats removal efficiency:

– with the seasonal efficiency of suspended par-ticles removal in the gravitational basin,

– with the seasonal air temperatures which domi-nantly influence the oxidation velocity in aerated basin.

MATERIAL AND METHODS

The coagulant was FeCl3 (Impuls hemija, Novi Sad), the commercial flocculant was Flopam EM 640 CT (Impuls hemija, Novi Sad) and NaOH (Impuls hemija, Novi Sad).

Experimental set-up

The time, i.e., the frequency of sampling influences the obtained results. The samples are taken in the time interval which is short enough for obtaining the repere-sentative sample (for the examination of the concen-tration changes). The samples are analysed immedi-atelly after sampling in order to avoid the change of water composition.

The composite samples used for the results in this paper are taken, examined and measured in intervals of 15 min, simultaneously on waste water input and on purified water output, during two hours of primary plant working, in the period of time from 2006 to 2011.

The measured contents of oil and fats, suspended matter, chemical oxygen demand with the analysis of composite samples, as well as temperatures, represent the average values in two-hour time period, after achieving the stationary state in purifying plant.

In primary plant at constant volume of basin during particularly steps of purification, the change of amount of purified substances performs the concentrations’ change keeping the stationary state between the liquid, solid and gas state and between wastewater input and purified water output:

– in accumulation processes (phase transformation) in flocculation and aggregation or flotation,

– and degradation processes (chemical transfor-mation) oxidation in basins for aeration, hydration and hydrolysis.

The wastewater with flow of 25–50 m3/h, which has to be purified, is placed in equalization basin from gathering basin with pumps, and then into the gravi-tational separator of fats (with retention time 57 min), coagulation basin (retention time 81 min), flocculation basin (retention time, 15 min), flotation basin (reten-tion time 37 min), neutralization basin (retention time 2×11 min) and then purified waters are released in urban canalization collector. The wastewater is anal-yzed according to the methods regulated by law for minimal number of parameters [23]. The water rel-eased into canalization is continual, gravitational and in the given period with flow 55 m3/h. The scheme of primary purifying is given in the Fig. 2.

In basin for coagulation, the coagulant 4 % solution of iron (III) chloride (FeCl3) and flocculant 0.1 % solution of organic polielectrolyte (Flopam EM 640 CT) were added at the flow of 25–50 m3/h during examination.

The samples for analysis are two-hour composites, obtained as the mixture of the samples taken simul-


608

taneously on wastewater input and purified water out-put, each after 15 min. The samples were taken on wastewater input (influent) and on the output from primary plant (efluent) in the four-quarter periods: the first quarter (February and March), the second quarter (May, June), the third quarter (September, October) and the fourth quarter (November, December).

The next parameters are determined with standard methods: oil and fats content, suspended matter con-tent, chemical oxygen demand (COD) and air tempera-ture (suspended matter – JUS.H.Z.1.160, ISO 11923; oil and fats – IR spectroscopy; chemical oxygen demand (COD) – EPA 410.4, US Standard Methods 5220 D, ISO 15705).

The efficiency of fats and oil removal, suspended matter and COD are calculated as the percentage of relative value of removed concentration (the difference between concentrations in effluent and influent) and the beginning concentration in influent according to Eq. (1).


The obtained values of air temperatures, fats and oil removal efficiency, as well as suspended matter and COD removal efficiency, are statistically analyzed through the programme Microsoft Excel and are given in the Tables 2–5.

The obtained results have shown the strong cor-relation:

– between oil and fats removal efficiency, sus-pended matter (Figure 3a and b) and COD (Figure 4a and b) in the first and fourth quarter,

– between oil and fats removal efficiency, sus-pended matter removal efficiencies (Figure 5a and b), in the second and third quarter and

– between oil and fats removal efficiency and air temperature (Figure 6a and b), in the second and third quarter.

Primary purifying is based on oil and fats adsorption and suspended matter on the crystallization nuclei formed after the addition of coagulants and flocculants.

In the 2008, the results (Tables 2–5) have shown

Figure 2.The scheme of primary treatment of edible oil industry waste water.

Table 2. The removal efficiency of the suspended particles, oil and fats, air temperature, water temperature and COD in the first quarter

The month Parameter

ƞ, suspended matter, % ƞ, fats and oil, % T air, °C T water ,°C COD / % February 2006 78.25 77.18 8 15.8 56 March 2007 50.03 19.93 7 14.7 -9.7 March 2008 47.3 51.04 13.8 26.3 -2.25 February 2009 73.73 95.97 9 21 43.7 March 2010 20.69 5.06 3 19.5 14.5 March 2011 92.39 95.03 3.1 15.2 67.2


609

negative values in the first quarter (ηorg. matter = –2.25%), second quarter (ηsusp. matter = –275%), in the third quarter (ηsusp. matter = –20%, ηfats and oil = –42.86%, as well as in the fourth quarter (ηfats and oil = –60.87%).

Those negative values indicate that: – the removal efficiency does not exist, – the negative values, indicate that the generation

of pollution in the system for purification and this water is diverted back to the equalization pool by a bypass pipeline,

– the appearance of the opposite effect, where the amount of fats and oils and suspended solids is greater on the exit than on the entrance and it is possible when the experimental part of the waste water is added to equalization basin in the period shorter than 24 h, prior to the sampling of unpurified water for quality analysis which is needed for waste homogenization (Fig. 2). The composite samples were taken for quality analysis during 2 h daily each in 15 min interval, according to the law regulative [23].

The excess of sulphuric acid in neutralization col-lector of waste water (Fig. 2) could also prevent hydro-lysis of fats acid and favor swelling which prevents its

adsorption and removal during coagulation and aggre-gation processes. Primary purifying is based on oil and fats adsorption and suspended matter of the crystal-lization nuclei formed after the addition of coagulants and flocculants.

Faster velocity of hydration of hydrophilic organic macromolecules could favor their swelling and oil solubility and hydrolysis rate constants decreasing and increasing influence on pH changes rate constants. Besides the seasonal water temperatures which control the hydration velocities of added coagulant and hyd-rophilic components, as well as velocity of dehydration process of hydrophobic components, it also amounts control of the velocities of its successive hydrolysis processes, i.e., turbidity increase and colloidal particles’ destabilization.

For these reasons, the appearance of negative values is not included in examined correlations obtained with aim to explain seasonal influence on the correct primary purifying procedure efficiency.

In the fourth quarter (Fig. 3a) the increased air temperatures up to water temperatures tend to favor exothermic hydration of organic macromolecules up to

Table 3. The removal efficiency of the suspended particles, oil and fats, and air temperature in the second quarter

The month Parameter

ƞ, suspended matter, % ƞ, fats and oil, % T air, °C June 2006 41.2 44.62 20 May 2007 42.88 70.0 18 June 2008 –275 3.93 22 April 2009 48.6 38.94 19 May 2010 20.69 51.0 19 May 2011 71.7 82.86 17.5

Table 4. The removal efficiency of the suspended particles, oil and fats, and air temperature in the third quarter

The month Parameter

ƞ, suspended matter, % ƞ, fats and oil, % T air, °C Sep. 2007 13.44 28.52 15 Sep. 2008 –20.00 -42.86 7 October 2009 6.00 6.82 7 Sep. 2010 55.61 22.34 12 Sep.2011 84.66 71.51 18

Table 5. The removal efficiency of the suspended particles, oil and fats, air temperature, water temperature and COD in the fourth quarter

The month Parameter

ƞ, suspended matter, % ƞ, fats and oil, % T air, °C T water ,°C COD / % Nov.2006 81.01 25.71 19 39.8 22.48 December 2007 78.73 50.80 18 16.9 9.53 Nov.2008 20.55 –60.87 5 21.0 –242.19 Nov.2010 77.60 39.95 7 34.5 6.27 December 2011 88.21 0.96 5 39.2 11.57


610

maximal removal efficiency of COD 22%. Then inc-reased removal efficiency of suspended matter indi-cates inhibition of oil and fats removal.

The results (Fig. 3b) show that in the first quarter, at less air temperatures relating to water, fats and oil removal efficiency increased with increased suspended matter removal due to organic matter endothermic dehydration. The contact of surface hydrolysis and aggregation of added coagulant together with sus-pended matter is enabled.

In the fourth quarter in November 2011, fats and oil removal efficiency in η interval 0.96–50.8% and in the first quarter the fats and oil removal efficiency was 5.06–95.97%. The increased temperature differencies between the air and water temperatures in 2010 were: Tair = 3 °C and Tw = 19.5 °C, in March 2010 and in November 2011 at air temperature of Tair = 5 °C and water temperature Tw = 39.2 °C, COD removal effi-ciency was worse, ηorg. matter, 6.5–22%, decreasing oil and fats removal efficiency ηorg. matter, 50–1%.

Based on the strong correlation coefficient (r = 0.88) obtained for the winter period, fats and oil are mostly adsorbed on suspended matter with the best removal efficiency up to 50%, for suspended matter removal efficiency about 75% and COD removal efficiency (r = = 0.69) up to 25% in the fourth quartal. The strong

correlation coefficient (r = 0.84) indicates linear inc-rease of the fats and oil removal efficiency up to 90% with suspended removal efficiency up to 100% and up to 80 % with 100% COD removal efficiency (r = 0.64) in the first quartal.

In the second quarter, fats and oil removal effi-ciency in interval was ηfats and oil, 3.93–82.86% and in the third quarter fats and oil removal efficiency was 6.82– –71,51%. In the second quarter the increased sus-pended matter removal efficiency, ηsusp. matter, was 20.69–71.7% and in the third quarter suspended mat-ter removal efficiency was 6.00–84.66% (Fig. 5a and b and Tables 3 and 4).

The results (Fig. 6a and Table 3) in the second quar-ter indicate that the fats and oil minimum removal effi-ciency (ηfats and oil = 3.93%) is achieved at maximal air temperature Tair = 22 °C. The optimal temperature that influenced the fats and oil removal efficiency increase is Tair = 17.5 °C; ηfats and oil = 82.86%.

In the third quarter (Fig. 6b and Table 4) air tem-perature favors the hydrations of organic components adsorbed on the surface of suspended matter and prevents the sedimentation of fats and oil along with the suspended matter, as it was obtained for the case, ηfats and oil = 6.82% and ηsusp. matter = 6%, where Tair =7 °C.

(a) (b)

Figure 3. Functional dependences of the removal efficiency of the suspended particles and oil and fats in the fourth and first quarter.

(a) (b)

Figure 4. Functional dependences of the removal efficiency of the oil and fats and COD in the fourth and first quarter.


611

(a) (b)

Figure 5. Functional dependences of the removal efficiency of oil and fats and suspended particles in the second and third quarter.

(a) (b)

Figure 6. Functional dependences of the removal efficiency of the oil and fats and air temperature change in the second and third quarter.

CONCLUSION Based on the monitoring results of the fats and oil

removal efficiency it can be concluded that: − The primary purifying process (adsorption of fats

and oil on the suspended matter and its aggregation) is a dominant mechanism in the winter period in the first and fourth quarter with the average temperature values of less than 10 °C when the best oxygen solu-bility in the purified water is achieved (the first quar-ter). Then, in the first quarter very high efficiencies of fats and oil removal are achieved, up to η = 90% in correlation with the suspended matter and also COD removal efficiency in oxidation processes mostly.

− Strongly decreased removal efficiency ηfats and oil = = 0.96% in the fourth and in the first ηfats and oil = 5,06% were measured during high temperature changes rel-ative to the purified water temperatures on the output of purifying plant (in the IV quarter, Tair = 5 °C, Tw = 39.2 °C in 2011, as well as Tair = 3 °C, Tw = 19.5 °C in 2010 in the I quarter). Then oxygen in oil and fats degradation is consumed in the temperature relaxation processes.

− In the second quarter, the middle air tempera-tures, equal with water temperature (at Tair = 17.5 °C enabled good removal efficiency, ηfats and oil = 82.86%).

However the increased air temperature, (Tair = 22 °C, ηfats and oil = 3.93%), decreased oxygen solubility and the fats and oil removal efficiency.

− Decrease of water temperature up to Tw = 7 °C in the third quarter favors the relaxation of tempe-rature with the exothermic hydration of the fats and oil, on the fats and oil removal efficiency, ηfats and oil = = 6.82%.

− The negative values in the first quarter (ηorg. matter = = –2.25%), in the second quarter (ηsusp. matter = –275%), in the third quarter (ηsusp. matter = –20%, ηfats and oil = = –42,86, –20 and –42.86%), as well as in the fourth quarter (ηfats and oil = –60.87%), were probably caused due to the inadequate sludge removal from the bottom of the flocculation pool.

Abbreviations

η – removal efficiency COD – chemical oxygen demand Tair – air temperature Tw – water temperature


612

REFERENCES

[1] V. Rajaković., LJ. Rajaković, Conventional and contemp-orary methods for water treatment: From wastewater to ultra pure water, Hem. Ind. 57 (2003) 307.

[2] V.B. Lazarević, Ivan M. Krstić, Ljiljana M.Takić, Miodrag L. Lazić, Vlada B. Veljković, Clarification and filtration of the floculated partucle suspension from a chemical treatment of waste oil-in-water emulsions from a non-ferrous metalworking plant, Hem. Ind. 65 (2011) 53–60.

[3] N. Hilal, G. Busca, F. Talens-Alesson, B.P. Atkin, Treat-ment of waste coolants by coagulation and membrane filtration, Chem. Eng. Proc. 43 (2004) 811–821.

[4] D. Deepak, S.G. Roy, K. Raghavan, S. Mukherjee, Effect of ferric chloride on the separation of miscible from waste water, Indian J. Environ. Health 1 (1988) 43–51.

[5] V. Lazarević, I. Krstić, M. Lazić, D. Savić, D. Skala, V. Velj-ković, Scaling up the chemical treatment of spent oil-in-water emulsions from a non-ferrous metal-processing plant, Hem. Ind. 67 (2012) 55–59.

[6] S.A. Muyibi, L.M. Evison, Optimizing physical parameters affecting coagulation of turbid water with Moringa oleifera seeds, Water Res. 29 (1995) 2689–2695.

[7] A. Ndabigengesere, K.S. Narasiah, B.G. Talbot, Active agents and mechanism of coagulation of turbid waters using Moringa oleifera, Water Res. 29 (1995) 703–710.

[8] G.K. Folkard, J.P. Sutherland, Development of a naturally derived coagulant for water and wastewater treatment, Water Sci. Technol.,Water Supply 2 (2002) 89–94.

[9] S.A.A. Jahn, Using Moringa seeds as coagulants in developing countries, Water Works Assoc. 80 (1988) 43–50.

[10] V.N. Rajaković, D. Skala, Demulsification based on the thermal treatment (cooling and heating) of W/O emul-sions, Hem. Ind. 58 (2004) 343–350.

[11] V.N. Rajaković, D. Skala, Separation of oil-in-water emulsions by freeze/thaw method and microwave radi-ation, Sep. Purif. Technol. 49 (2006) 192–196.

[12] L. Jovanović, T. Nikolin, M. Ševaljević, Primary treatment of oily waste water, in Proceedings of 32. Stručno-naučni skup Vodovod i kanalizacija, Kladovo, Serbia, 2011, pp. 85–92.

[13] M. Ševaljević, M. Pavlović, M. Velimirović and D. Tošić , Seasonal temperature variation and the water self–

cleaning in the lake system “Begej's loop”-Zrenjanin, in Proceedings of the International Conference on Environ-mental Management, Engineering, Planning and Eco-nomics, Skiathos, Greece, 2007, pp. 1081–1086

[14] P. Ševaljević, O. Grozdanović, M. Ševaljević, Seasonal electric polatization of the chemisorbed oxygen in the urban river water, in Proceedings of II International Conference „Ecology of urban areas”, Yrenjanin, Serbia, 2012, pp. 164–168.

[15] M.M. Ševaljević, S.N. Simić, P.V. Ševaljević, Thermodyn-amic diagnostic of electron densities in gas bubbles in aerated saturated refinery waste water, Desal. Wat. Treat. 42 (2012)144–154

[16] M. Ševaljević, D. Tošić, T. Manigoda, Seasonal tempe-rature influence on primary purifying of urban river water, in Proceedings of I International Conference ”Ecology of urban areas”, Zrenjanin, Serbia, 2011, pp. 240–246.

[17] M. Ševaljević, M. Minić, Seasonal kinetic aspect of puri-fication of urban atmospheric water in petrochemical refinery, in Proceedings of I International Conference ”Ecology of urban areas”, Zrenjanin, Serbia, 2011, pp. 224–229.

[18] I.M. Protić, M. Ševaljević, N. Aćin, M. Novaković, T. Nikolin, Z. Veljković, The influence of seasonal tempe-rature change rate on the relaxsation processes in the row and chlorinated water of Zrenjanin urban water-work, in Proceedings of II International Conference ”Ecology of urban areas”, Zrenjanin, Serbia, 2012, pp. 220–225.

[19] S. Gaćeša, M. Klašnja, The technology of water and wastewater, 1994.

[20] A. Adin, T. Asano, The role of physical–chemical treat-ment in wastewater reclamation and reuse, Water Sci. Technol. 37 (1998) 79–90.

[21] H. Odegaard, Optimised particle separation in the pri-mary step of wastewater treatment, Water Sci. Technol. 37 (1998) 43–53.

[22] A.F.V. Nieuwenhuijzen, J.H. Graaf, A.R. Mels, Direct influent filtration as a pretreatment step for more sus-tainable wastewater treatment systems, Water Sci. Technol. 43 (2001) 91–98.

[23] Regulations on the minimum number of test quality of wastewaterSl. Glasnik SRS, 47/83 i 13/84


613

IZVOD

ISPITIVANJE SEZONSKIH UTICAJA NA EFIKASNOST UKLANJANJA MASTI I ULJA PRIMARNIM PREČIŠĆAVANJEM OTPADNIH VODA ULJARSKE INDUSTRIJE

Tatjana S. Nikolin, Mirjana M. Ševaljević

Visoka tehnička škola strukovnih studija, VŠSS, Zrenjanin, Srbija

(Naučni rad)

U ovom radu je ispitan uticaj sezonskih promena temperature vazduha,hemijske potrošnje kiseonika, kao i efikasnosti izdvajanja suspendovanih materijana efikasnost izdvajanja masti i ulja u otpadnoj vodi uljarske industrije tokomprimarnog prečišćavanja. Parametri su praćeni u periodu od 2006 do 2011. godi-ne. Efikasnost izdvajanja masti i ulja tokom I i IV kvartala je proporcionalna saefikasnošću uklanjanja suspendovanih čestica i ukupnih organskih materija (HPK).Izmerene vrednosti su za masti i ulja, η(IV kvartala), 0,96–50,8% i η(I kvartala), 5,06––95,97%. Efikasnosit izdvajanja masti i ulja u II i III kvartalu su proporcionalnetemperaturi vazduha te su izmerene efikasnosti izdvajanja masti i ulja, η(II kvartala), 3,93%–82,86% i η(III kvartala), 6,82–71,51%. Rezultati istraživanja su ukazali napostojanje veze između spoljnih temperatura vazduha tokom različitih godišnjihdoba i efikasnosti izdvajanja (η, %) masti, suspendovanih čestica, kao i hemijske potrošnje kiseonika.

Ključne reči: Efikasnost izdvajanja • Mo-nitoring otpadne vode • Industrija jesti-vog ulja • Temperatura vazduha • Sezon-ske varijacije • Masti i ulja

615

Predlog za određivanje promene entropije poluidealnog gasa primenom srednjih vrednosti temperaturnih funkcija

Branko B. Pejović, Vladan M. Mićić, Mitar D. Perušić, Goran S. Tadić, Ljubica C. Vasiljević, Slavko N. Smiljanić

Tehnološki fakultet, Univerzitet u Istočnom Sarajevu, Zvornik, Republika Srpska, BIH

Izvod Kod poluidealnog gasa, koji u tehničkoj praksi ima svoje mesto i značaj, promena entropijene može se odrediti preko srednjeg specifičnog toplotnog kapaciteta na način kao što seodređuje promena unutrašnje energije i entalpije, odnosno razmenjena količina toplote.Uzimajući ovo u obzir, u radu su izvedena dva modela preko kojih je moguće odrediti pro-menu specifične entropije poluidealnog gasa za proizvoljan temperaturni interval prime-nom tablične metode, koristeći srednje vrednosti pogodno izabranih funkcija.Ideja je da seintegriranje koje se ovde neminovno javlja, zameni srednjim vrednostima prethodnihfunkcija.Modeli su izvedeni na bazi funkcionalne zavisnosti stvarnog specifičnog toplotnogkapaciteta od temperature. Takođe, izvršena je analiza usvajanja pogodne početne tem-perature.Pri ovome korišćena je teorema o srednjoj vrednosti funkcije kao i matematičkeosobine određenog integrala. Srednja vrednost razlomljene funkcije određena je direktno preko njene podintegralne funkcije dok je kod logaritamske funkcije izvršena pogodnatransformacija primenom diferencijalnog računa. Izvedene relacije, primenom računarskogprograma, omogućile su sastavljanje odgovarajućih termodinamičkih tablica preko kojih je moguće odrediti promenu entropije proizvoljne promene stanja na efikasan odnosno ra-cionalan način bez primene integralnog računa, odnosno gotovih obrazaca. Na ovaj način,promena entropije poluidealnog gasa, određena je za proizvoljan temperaturni interval analognom metodom koja se primenjuje i kod određivanja promene unutrašnje energije ientalpije odnosno razmenjene količine toplote, što je bio i cilj rada. Verifikacija predloženemetode za obe gore navedene funkcije, izvedena je za nekoliko karakterističnih poluideal-nih gasova kod kojih je izraženija nelinearnost funkcije cp(T), za tri usvojena temperaturska intervala, za karakterističnu promenu stanja. Pri ovome izvršeno je poređenje rezultataprema klasičnoj integralnoj i predloženoj metodi preko sastavljenih tablica za razlomljenufunkciju. Prema drugom modelu s obzirom na logaritamsku funkciju izvršeno je poređenjesa prvim modelom pri čemu je dobijena zadovoljavajuća tačnost. Prikazanu metodu, uodređenim odnosno posebnim slučajevima, moguće je primeniti i kod određivanja pro-mene entropije realnog gasa. Isto tako, u radu je pokazano da je promenu entropije zaposmatrani karakterističan slučaj, moguće predstaviti odnosno grafički odrediti planimetrij-skom metodom u dijagramima sa pogodno odabranim koordinatama.

Ključne reči: poluidealan gas, promena entropije, srednji i pravi specifični toplotni kapacitet,srednja vrednost funkcije, diferencijalni i integralni račun, tablične vrednosti funkcije, apro-ksimativne funkcije, grafičke metode.

STRUČNI RAD

UDK 544.3/.322:517.5:66

Hem. Ind. 68 (5) 615–628 (2014)

doi: 10.2298/HEMIND130825090P


Poluidealni gasovi pripadaju grupi gasovitih sup-stanci kod kojih se uzima u obzir zavisnost specifičnih toplotnih kapaciteta od temeperature. U velikom broju slučajeva zavisnost specifičnog toplotnog kapaciteta od temperature je dosta izražena i mora se kod proračuna uzeti u obzir [1–5].

Vrednost srednjeg specifičnog toplotnog kapaciteta u datom temperaturnom intervalu ne zavisi samo od

Prepiska: V.M. Mićić, Tehnološki fakultet Univerziteta u Istočnom Sarajevu, Karakaj bb, 75400 Zvornik, Republika Srpska, Bosna i Hercegovina. E-pošta: [email protected] Rad primljen: 25. avgust, 2013 Rad prihvaćen: 18. december, 2013

veličine tog intervala tj. temperaturne razlike T2–T1, već i od položaja tog intervala na temperaturnoj skali.

Ovde treba naglasiti da se u nizu stvarnih termo-dinamičkih procesa, na primer: kod kompresora, mo-tora sa unutrašnjim sagorevanjem, gasnih turbina, itd., kod kojih su radne materije gasovi bliski idealnim gaso-vima (vazduh i produkti sagorevanja), može se usvojiti da se realna materija ponaša kao poluidealan gas, pa razmatranje specifičnih toplotnih kapaciteta kao isklju-čivih funkcija temperature ima veliki praktični značaj, [6–9].

Specifična količina toplote (q12), koja se dovodi poluidealnom gasu (i uopšte nekom telu čiji specifični toplotni kapacitet zavisi od temperature), od početne

B.B. PEJOVIĆ i sar.: ODREĐIVANJE PROMENE ENTROPIJE POLUIDEALNOG GASA Hem. ind. 68 (5) 615–628 (2014)

616

temperature T1 do krajnje temperature T2, određena je izrazom [10–13,31]:

= = − 2

1

2

12 2 11

( )d ( )TTq c T T c T T (1)

Ovde c(T) predstavlja stvarni odnosno pravi, dok je 2

1

TTc srednji toplotni kapacitet u intervalu tempe-

rature T1 i T2, određen relacijom:

− − −=

−

2 1

2 0 0

1

2 0 1 0

2 1

( ) ( )T TT T TT

c T T c T Tc

T T (2)

Prema relacijama (1) i (2) na slici 1 dat je grafički prikaz predstavljanja razmenjene specifične količine toplote u dijagramu (c,T).

Promena unutrašnje energije (du) i entalpije (dh) poluidealnog gasa može se izraziti kao [2,5,9,16–19,31– –33]:

=d ( )dvu c T T , =d ( )dph c T T

odnosno:

− = = − 2

1

2

2 1 2 11

( )d ( )Tv v Tu u c T T c T T (3)

− = = − 2

1

2

2 1 2 11

( )d ( )Tp p Th h c T T c T T (4)

Uzimajući u obzir relaciju (1), odnosno diferencijalni oblik dq = cdT, razmenjena specifična količina toplote za izobarsku i izohorsku promenu, kao karakteristične promene stanja, biće [1,3,20]:

= 2

121

( )dpq c T T , = 2

121

( )dvq c T T (5)

= −2

112 2 1( )Tp Tq c T T , = −2

112 2 1( )Tv Tq c T T (6)

U prethodnim relacijama cp i cv predstavljaju speci-fične toplotne kapacitete pri konstantnom pritisku i konstantnoj zapremini.

Iz relacija (3) i (4), odnosno (5) i (6), sledi da je za slučaj poluidealnih gasova, promenu unutrašnje ener-gije i entalpije, odnosno razmenjenu količinu toplote, moguće odrediti preko srednjeg specifičnog toplotnog kapaciteta. Kao što će se kasnije videti, kod određivanja promene entropije, nije moguć ovakav pristup.

Pre nego što se pređe na izvođenje glavnih relacija koje se odnose na temu rada, biće ukratko data mate-matička interpretacija srednje vrednosti funkcije na bazi koje će biti rešen postavljeni problem.

Srednja vrednost funkcije

Prema teoremi o srednjoj vrednosti neprekidne funkcije, y = f(x), u intervalu x1 i x2 važi da je: [14,15,21,23]:

= −2

1

2 1( )d ( )x

x

f x x x x y (7)

Geometrijski smisao relacije (7) je u tome da je površina ispod krive 12 tj. površina a12ba, jednaka površini pravougaonika acdba, slika 2.

Isto tako, prema osobini određenog integrala biće:

= − 2 2 1

1 0 0

( )d ( )d ( )dx x x

x

f x x f x x f x x (8)

Pri ovome, početna tačka x0 ne mora biti koordi-natni početak. Iz relacije (7) sledi da je srednja vrednost funkcije y = f(x) u zadatom intervalu:

Slika 1. Predstavljanje razmenjene količine toplote u dijagramu (c,T). Figure 1. Presentation of the amount of exchanged heat in the diagram (c,T).


617

=−

2

12 1

1( )

x

x

y f xx x

(9)

što predstavlja srednju ordinatu krive f(x) u istom intervalu, za apscisu x .

Relacije (8) i (9) biće korišćene u razmatranjima koja slede.

Promena specifične entropije poluidealnog gasa

Iz relacija (5) i (6) sledi da je:

=−

2

2

1

12 1

1 ( )dT

Tp pT

T

c c T TT T

(10)

što, s obzirom na relaciju (9), u stvari predstavlja srednju vrednost funkcije stvarnog specifičnog toplot-nog kapaciteta u dijagramu cp = cp(T) u intervalu tem-peratura T1–T2.

Promena specifične entropije (s2–s1), za opšti slučaj promene stanja s obzirom da je =d dq T s , može se napisati kao [2,5,7,10]:

− = +2

22 1

11

dlnv

vTs s c RT v

(11)

odnosno za slučaj poluidealnog gasa:

− = +2

22 1

11

( )d lnvc T v

s s T RT v

(12)

Isto tako može se pokazati da važi i relacija:

− = −2

22 1

11

dlnp

pTs s c RT p

(13)

odnosno za slučaj poluidealnog gasa:

− = −2

22 1

11

( )d lnpc T p

s s T RT p

(14)

U relacijama (11)–(14) v i p predstavljaju specifičnu zapreminu, odnosno pritisak.

Iz relacija (11) i (13) sledi da je određivanje pro-mene entropije kada je u pitanju idealan gas pojed-nostavljeno, jer su specifični toplotni kapaciteti cv i cp konstantni, što dovodi do jednostavnog integriranja izraza.

Iz relacija (12) i (14), za slučaj izohorske i izobarske promene kao karakterističnih promena stanja kada je u pitanju poluidealan gas, promena specifične entropije biće:

− = 2

2 11

( )dvc T

s s TT

(15)

− = 2

2 11

( )dpc T

s s TT

(16)

U nastavku, analiza problema biće izvedena s obzi-rom na relacije (15), odnosno (16), ali je očigledno da će analiza važiti i za opšti slučaj promene stanja, relacije (12), odnosno (14).

Kada bi se promena entropije izračunavala prema relacijama (13), (14), odnosno (15) i (16), primenom srednjeg specifičnog toplotnog kapaciteta, činila bi se principijelna greška. Naime, mora se uzeti u obzir srednja vrednost cele podintegralne funkcije, a ne samo jednog njenog dela.

Slika 2. Srednja vrednost funkcije y = f(x). Figure 2 The average value of y = f(x).


618

Prema tome, s obzirom na relacije (15) i (16) za izohorsku i izobarsku promenu stanja, za promenu entropije poluidealnog gasa, bilo bi pogrešno napisati da je:

− = =2 2

1 1

22

2 111

dlnT T

v vT TTTs s c c

T T (17)

− = =2 2

1 1

22

2 111

dln

T Tp pT T

TTs s c cT T

(18)

U nastavku, biće dat predlog za rešavanje prethod-nog problema i to na dva načina, primenom dveju pogodno izabranih karakterističnih temperaturnih funk-cija.

Napomenimo ovde da i za slučaj određivanja pro-mene specifične entropije van der Valsovog gasa (realan gas za koji važi van der Valsova jednačina stanja), nailazimo na sličan problem s obzirom da je u jednačini takođe prisutna slična podintegralna funkcija, [11,13,18]:

−− = +

− 2

22 1

1 1

dln ( )g V

v b Ts s R c Tv b T

(19)

gde b predstavlja konstantu.

Rešavanje postavljenog problema primenom razlomljene funkcije

U termodinamičkim tablicama se nalaze srednje vrednosti specifičnog toplotnog kapaciteta, 0|

Tpc , za

različite gasove i za različite temperature, dok se vrednosti 2

1|T

p Tc određuju preko relacije (2). Vrednost 2

1|T

V Tc , može se dobiti koristeći Majerovu jednačinu. Definisane srednje vrednosti specifičnih toplotnih

kapaciteta, kao što je pokazano, proizilaze iz naznače-nih integrala i zamenjuju u računu samo naznačeni pod-integralni izraz.One se, kao što je rečeno, ne mogu koristiti za izračunavanje promene entropije poluideal-nog gasa preko relacija (12) i (14), odnosno (15) i (16).

Ako bi se želelo zameniti integriranje u ovim rela-cijama računanjem srednje vrednosti, tada bi se morala poznavati srednja vrednost 2

1( ( ) / )|T

Tc T T , kao podinte-gralne funkcije. Pošto takvih podataka nema u tabli-cama, mora se izračunati integral

2

1( ( ) / )dc T T T .

U praktičnim problemima, gde je radna materija poluidealan gas često se ne koriste analitički proračuni nego dijagrami ili tabele iz kojih se očitavaju mnoge termodinamičke veličine. Međutim, u mnogim sluča-jevima kada je potrebno odrediti promenu entropije, neophodna je primena analitičke metode [10,12,17,19].

Koristeći osobinu određenog integrala (8), prema prethodnoj analizi, promena entropije poluidealnog gasa, naprimer za izobarsku promenu, može se napisati kao:

Δ = − = = − 2 2 1

1 0 0

2 1( ) ( ) ( )

d d dT T T

p p p

T T T

c T c T c Ts s s T T T

T T T (20)

Ovde je T0 neka unapred pogodno odabrana po-četna temperatura.

Uzimajući u obzir prethodnu analizu, promena en-tropije izobarske promene stanja, može se izraziti preko srednje vrednosti razlomljene funkcije ( ) /pc T T za interval temperatura T1–T2:

Δ = = −22

1 1

2 1( )

d ( )TT

p p

T T

c T cs T T T

T T (21)

odnosno, s obzirom na relaciju (20):

Δ = − − −2 1

0 0

2 0 1 0( ) ( )T T

p p

T T

c cs T T T T

T T (22)

što predstavlja glavnu relaciju za određivanje promene entropije postavljenog problema.

Iz relacije (21) sledi da je promenu entropije mo-guće principijelno predstaviti u dijagramu (Y,T) gde je razlomljena funkcija = ( ) /pY c T T . Može se pokazati da ova funkcija za poluidealne gasove monotono opada sa porastom temperature. Promenu entropije za prethod-ni slučaj, prema relaciji (21), moguće je odrediti plani-metrijskom metodom kao površinu ispod krive Y = Y(T) ograničene temperaturama T1 i T2 (površina T112T2), odnosno kao površinu pravougaonika T1deT2, slika 3.

Pri ovome, očigledno je:

= −

= −

22

0 0

11

0 0

2 0

1 0

( )d ( ),

( )d ( )

TTp p

T T

TTp p

T T

c T cT T T

T T

c T cT T T

T T

(23)

dok se srednja vrednost funkcije ( ) /pc T T za interval temperatura T1–T2 može izračunati prema prethodnim relacijama uzimajući u obzir relaciju (2), kao:

− − −

=−

2 1

2

0 0

1

2 0 1 0

2 1

( ) ( )T T

p pT

T Tp

T

c cT T T T

T TcT T T

(24)

Isto tako prema prethodnom, za srednje vrednosti iste funkcije prema relaciji (23) važi da je:

=−

1

10

01 0

( )d

Tp

TTp

T

c TT

TcT T T

, =−

2

20

02 0

( )d

Tp

TTp

T

c TT

TcT T T

(25)


619

odnosno za interval T0 do Ti, gde je Ti neka proizvoljna krajnja temperatura, biće:

=−

0

00

( )d

i

i

Tp

TTp

iT

c TT

TcT T T

(26)

Isto tako, iz relacije (26) sledi da je:

= −0 0

0( )

d ( )ii TT

p pi

T T

c T cT T T

T T (27)

Prikaz karakterističnih srednjih vrednosti razlom-ljenih funkcija 1

0( / )|T

p Tc T , 2

0( / )|T

p Tc T i 2

1( / )|T

p Tc T , odno-sno njihov položaj u dijagramu Y = Y(T), dat je na slici 3.

Promena entropije izobarske promene stanja, za interval temperatura T1–T2, može se na jednostavan način odrediti planimetrijski, preko srednjih vrednosti

1

0( / )|T

p Tc T i 2

0( / )|T

p Tc T kao razlika površina pravo-ugaonika abgf i T0T1nm u dijagramu (Y,T), što je prikazano na slici . (šrafirana površina abde):

Δ = − − −2 1

0 0

2 0 1 0( ) ( )T T

p p

T T

c cs T T T T

T T (28)

Pri ovome temperatura T0 je usvojena u koordi-natnom početku. Primena razlomljene funkcije na grupu poluidealnih gasova

U termodinamici i termotehnici često se koristi zavisnost stvarnog odnosno pravog specifičnog toplot-nog kapaciteta od temperature u obliku funkcije čet-vrtog stepena [13,17,20,25,32]:

= + + + +2 3 4( ) 2 3 4 5pc T B CT DT ET FT [ ]kJ/(kg K) (29)

Pri ovome, moguće je usvojiti funkciju ( )pc T i u obliku drugačijeg polinoma.

Konstante B, C, D, E i F u funkciji (29), dobijaju se eksperimentalno i nalaze se za različite gasove u odgo-varajućoj literaturi. Zamenom funkcije (29) u relaciju (21) dobija se da je:

[ ]

Δ = − = =

= + + + +

2

1

2

1

2 1

2 3

( )d

( 2 3 4 5 )d kJ/(kg K)

Tp

T

T

T

c Ts s s T

T

B C DT ET FT TT

(30)

Integriranjem izraza (30), s obzirom na temperaturni interval od T1 do T2 biće:

Δ = + + + +

2

1

2 3 4ln 2 3 4 5 |

2 3 4TT

T T Ts B T CT D E F (31)

odnosno nakon zamene granica integrala, dobija se konačno:

Δ = + − + − +

+ − + −

2 222 1 2 1

1

3 3 4 42 1 2 1

3ln 2 ( ) ( )

24 5

( ) ( )3 4

T Ds B C T T T TT

E FT T T T (32)

Relacija (32) se koristi često u termodinamičkoj praksi za određivanje promene entropije poluidealnog gasa [5,13,17,18,33]. Očigledno, ovakav način određi-vanja promene entropije, pored toga što dugo traje, može biti i izvor računskih grešaka.

Slika 3. Karakteristične srednje vrednosti funkcije ( ) /pc T T i predstavljanje entropije u dijagramu ( )( ) / ,pc T T T . Figure 3. Typical mean values of the function ( ) /pc T T and entropy in diagram ( )( ) / ,pc T T T .


620

Promena entropije izobarske promene od neke početne temperature T0 do proizvoljne temperature Ti, preme prethodnoj analizi, može se napisati kao:

Δ = − = 0

0( )

diT

pi

T

c Ts s s T

T (33)

Uzimajući u obzir rešenje (32), biće, prema relaciji (33):

Δ = − = + − + − +

+ − + −

2 20 0 0

0

3 3 4 40 0

3ln 2 ( ) ( )

24 5

( ) ( )3 4

ii i i

i i

T Ds s s B C T T T TT

E FT T T T (34)

Sada prema izrazu (26), odnosno (30), uzimajući u obzir relaciju (34), dobijamo konačnu formulu za izra-čunavanje srednje vrednosti funkcije cp(T)/T za interval temperatura T0 do Ti:

= ×−

× + − + − +

+ − + −

00

2 20 0

0

3 3 4 40 0

1

3( ln 2 ( ) ( )

24 5

( ) ( ))3 4

iTp

iT

ii i

i i

cT T T

T DB C T T T TT

E FT T T T

(35)

Očigledno, relacija (35) se može primeniti za proiz-voljnu promenu stanja.

Određivanjem promene entropije za proizvoljna stanja 1 i 2 kao s2–s1 = (s2–s0)-(s1–s0) koristeći opštu relaciju (34), lako se može pokazati da ona ne zavisi od početne temperature T0. Ovo sledi iz činjenice da je entropija veličina stanja koja ima totalni diferencijal [26–28].

U hemijskoj termodinamici često se koristi zavisnost stvarnog specifičnog toplotnog kapaciteta od tempe-rature u obliku funkcije [19,20,29]:

−= + + = + +22( )p

dc T a bT dT a bTT

(36)

pri čemu konstante a, b i d zavise od vrste poluidealnog gasa.

Po istoj proceduri kao kod funkcije (29), integrira-njem se dobija da je:

Δ = = − +

+ − − −

2

2 11

2 1 2 22 1

( )d (ln ln )

1 1( ) ( )

2

pc Ts T a T T

T

db T TT T

(37)

Promena entropije za interval temperatura od T1 do T2 i ovde se može napisati u obliku:

Δ = − = − − −2 1

0 0

2 1 2 0 1 0( ) ( )T T

p p

T T

c cs s s T T T T

T T

Opšta formula za izračunavanje funkcije cp(T)/T za proizvoljni interval temperatura od T0 do Ti za ovaj slučaj biće, s obzirom na relacije (28) i (37):

− + − − −=

−0

0 0 2 20

0

1 1(ln ln ) ( ) ( )2iT i i

p i

iT

da T T b T Tc T TT T T

(38)

Relacija (38) važi za interval temperatura T0–Tmax, gde je T0 = 300 K, a Tmax od 1800 do 2500 K, zavisno od vrste poluidealnog gasa.

U tabeli 1 date su konstante za funkciju (29) za neke poluidealne gasove kao i interval temperatura T0–Tmax, za koji iste važe [13,17,30].

Koristeći izvedeni opšti izraz (35), preko sastav-ljenog numeričkog programa za četiri karakteristična poluidealna gasa CO2, O2, N2 i H2, kod kojih je promena specifičnog toplotnog kapaciteta od temperature izra-žena, određene su srednje vrednosti

0( / )| iT

p Tc T za inter-val temeperatura za koji važi funkcija (29). Izračunate vrednosti su date tabelarno (tabela 2), za razmak tem-peratura ΔT = 100 K pri čemu je početna temperatura T0 = 100 K. Tabela 2 je sastavljena sa tačnošću od 6 decimala. Ovde treba zapaziti da je početnu tempe-raturu, T0, pogodno usvojiti na početku intervala za koji

Tabela 1. Konstante u funkciji cp = B+2CT+3DT2+4ET3+5FT4 za neke poluidealne gasove Table 1. The constants in the function cp = B+2CT+3DT2+4ET3+5FT4 for some semi-ideal gases

Gas B C×103 D×106 E×109 F×1013 Opseg temperature, °C Azot 1,06849 –0,134096 0,215569 –0,078632 0,06985 –175–1200 Kiseonik 0,95244 –0,28114 0,655223 –0,452316 1,087744 –175–1200 Sumpor-dioksid 0,46165 0,248915 0,12090 –0,18878 0,568232 –175–1200 Ugljen-dioksid 0,479107 0,762159 –0,359392 0,084744 –0,057752 –175–1200 Ugljen-monoksid 1,074015 –0,172664 0,302237 –0,137533 0,200365 –175–1200 Voda 1,915007 –0,395741 0,876232 –0,495086 1,038613 –175–1200 Vodonik 13,396156 2,960131 –3,980744 2,661667 –6,099863 –175–1200 Acetilen 0,094773 4,114197 –4,037767 2,133692 –4,415085 –20–1200 Etilen 0,60693 1,288788 1,033636 –1,099537 2,929326 –20–1200


621

važi funkcija (29). Razlog ovome je taj što ako bi se, na primer, usvojila temperaturaT0 = 0 K, izraz (35) postaje nedefinisan. Isto tako za ovaj slučaj, integral (33) pos-taje nesvojstveni [14,15,24,29]. Na ovo treba obratiti posebnu pažnju. Isto tako, zapaža se da razlomljena funkcija prema tabeli 2 uvek opada sa porastom apso-lutne temperature.

Za verifikaciju modela prema razlomljenoj funkciji cp(T)/T, usvojena su tri karakteristična temperaturna intervala za gore navedena četiri poluidealna gasa. Za svaki interval, izračunata je promena entropije za izo-barsku promenu stanja prema klasičnom postupku pri-menom integralnog računa, relacija (34), odnosno prema predloženom postupku koristeći podatke iz ta-bele 2 prema relaciji (22). Rezultati su sistematizovani u tabeli 3, odakle sledi da su dobijene gotovo identične vrednosti. Dobijena razlika u rezultatima je posledica zaokruživanja vrednosti u tabeli 2. S obzirom na pos-tupak izvođenja modela i grafičkoj interpretaciji pro-mene entropije prema slici 3, zbog jednakosti površina u dijagramu, ovo se moglo i očekivati.

Rešavanje problema primenom logaritamske funkcije

Kada bi, na primer, u relaciji (14) kod prvog inte-grala, odnosno u relaciji (16), hteli koristiti srednje vrednosti za cp, onda bi se prema drugom predlogu mogla razmatrati kao pogodna funkcija cp = cp(lnT). Pošto takvih funkcija nema, ideja je da se ista odredi na bazi promene poznate funkcije cp = cp(lnT) za određeni poluidealan gas.

Promena entropije poluidealnog gasa od stanja 1 do 2 prema relaciji (16), za izobarsku promenu stanja, može se napisati kao [14,15,25,26]:

Δ = − = = 2 2

2 11 1

d( ) ( )d(ln )p p

Ts s s c T c T TT

(39)

gde je diferencijal logaritma:

= dd(ln )

TTT

(40)

Funkciju cp(T) možemo predpostaviti s obzirom na relaciju (29) u logaritamskom obliku:

= = + + + +

+

2 31 1 1 1

41

( ) (ln ) ln ln ln

lnp pc T c T B C T D T E T

F T (41)

gde su B1, C1, D1, E1 i F1 konstante koje treba odrediti. Pri ovome moguće je usvojiti i drugačiji oblik logari-tamske funkcije.

S obzirom na funkciju (41), relacija (39) prelazi u:

Δ = 2

1

(ln )d(ln )ps c T T (42)

odnosno zamenom relacije (41) u relaciju (42) biće:

Δ = + + + +2

2 3 41 1 1 1 1

1

( ln ln ln ln )d(ln )s B C T D T E T F T T (43)

Promena entropije prema relaciji (42) može se napi-sati u jednostavnijem obliku:

Δ = 2

1

( )dps c z z

(44)

uvođenjem smene [14,15]:

Tabela 2. Srednje vrednosti 0

( / )| iTp Tc T (kJ/(kg K2)) za CO2, O2, N2 i H2 kao poluidealne gasove pri početnoj temperaturi T0 = 100 K

Table 2. The mean values 0

( / )| iTp Tc T (kJ/(kg K2)) for CO2, O2, N2 and H2 as semi-ideal gases at the initial temperature T0 = 100 K

Ti / K CO2 O2 N2 H2

100 0 0 0 0 200 0,004691 0,006294 0,007228 0,097221 300 0,003955 0,004990 0,005717 0,077549 400 0,003492 0,004215 0,004810 0,065519 500 0,003162 0,003694 0,004194 0,057220 600 0,002911 0,003315 0,003744 0,051074 700 0,002708 0,003024 0,003400 0,046307 800 0,002541 0,002791 0,003125 0,042486 900 0,002398 0,002599 0,002902 0,039346 1000 0,002274 0,002437 0,002714 0,036717 1100 0,002166 0,002298 0,002555 0,034481 1200 0,002069 0,002176 0,002418 0,032554 1300 0,001983 0,002069 0,002297 0,030874 1400 0,001905 0,001975 0,002191 0,029393 1470 0,001855 0,001915 0,002095 0,028075


622

=lnT z , =d(ln ) dT z (45)

S obzirom na početnu tačku i tačke promene stanja 1 i 2 biće:

=0 0lnz T , =1 1lnz T , =2 2lnz T (46)

Sa ovom smenom, problem je u suštini sveden na slučaj koji se javlja kod određivanja promene unu-trašnje energije i entalpije odnosno razmenjene koli-čine toplote poluidealnog gasa, relacije (3)–(6).

Uzimajući u obzir novu smenu, relacija (43) prelazi u:

Δ = + + + +2

1

2 3 41 1 1 1 1( )d

z

z

s B C z D z E z F z z

(47)

Integriranjem izraza (47) za promenu entropije od stanja 1 do stanja 2 biće:

Δ = + + + +

2

1

2 3 4 5

1 1 1 1 12 3 4 5

z

z

z z z zs B z C D E F

(48)

odnosno nakon zamene granica integrala:

Δ = − + − + − +

+ − + −

2 2 3 31 11 2 1 2 1 2 1

4 4 5 51 12 1 2 1

( ) ( ) ( )2 3

( ) ( )4 5

C Ds B z z z z z z

E Fz z z z

(49)

Uzimajući u obzir smenu (46), relacija (49) prelazi u:

Δ = − + − +

+ − + − +

+ −

2 211 2 1 2 1

3 3 4 41 12 1 2 1

5 512 1

(ln ln ) (ln ln )2

(ln ln ) (ln ln )3 4

(ln ln )5

Cs B T T T T

D ET T T T

FT T

(50)

Relacija (44), s obzirom na srednju vrednost funkcije cp(z), može se sada napisati kao:

( )Δ = = −2

2

11

2 1( )d ( )z

zp p z

z

s c z z c z z z (51)

Prema osobini određenog integrala (8), odavde sledi da je:

( )

( )

Δ = − = − −

− −

2 1

2

00 0

1

0

2 0

1 0

( )d ( )d ( )

( )

z zz

p p p zz z

zp z

s c z z c z z c z z z

c z z z

(52)

Ovde je direktno iskorišćena definicija srednje vred-nosti funkcije (9):

Tabela 3. Promena entropije izobarske promene za CO2, O2, N2, H2 kao poluidelne gasove, izračunata primenom integralne i tabelarne metode (kJ/kg K)

Temperaturni interval Opseg ΔT = T1–T2, K Δ = 2

1

( )d

Tp

T

c Ts T

T Δ = − − −

2 1

0 0

2 0 1 0( ) ( )T T

p p

T T

c cs T T T T

T T

CO2 Širi 200–1300 1,910502 1,9105 Srednji 400–1100 1,118401 1,1184 Uži 600–800 0,323200 0,3232

cp(T) = 0,479107 + 1,524318×10–3T – 1,078176×10–6T2 + 0,338976×10–9T3 – 0,28876T4, T0 = 100 K O2

Širi 200–1300 1,853404 1,8534 Srednji 400–1100 1,033501 1,0335 Uži 600–800 0,296206 0,2962

cp(T) = 0,95244 – 0,28114×10–3T + 0,655223×10–6T2 – 0,452316×10–9T3 + 1,087744×10–13T4, T0 = 100 K Azot 2

Širi 200–1300 2,03362 2,0336 Srednji 400–1100 1,11204 1,112 Uži 600–800 0,31551 0,3155

cp(T) = 1,0684 – 0,134096×10–3T + 0,215569×10–6T2 – 0,078632×10–9T3 + 0,06985×10–13T4, T0 = 100 K H2

Širi 200–1300 27,326702 27,3267 Srednji 400–1100 14,825304 14,8253 Uži 600–800 4,203205 4,2032

cp(T) = 13,396156 + 2,960131×10–3T – 3,980744×10–6T2 + 2,661667×10–9T3 – 6,099863×10–13T4, T0 = 100 K


623

( )= −2

2

00

2 0( )d ( )z

zp p z

z

c z z c z z z

( )= −1

1

00

1 0( )d ( )z

zp p z

z

c z z c z z z (53)

Karakteristične srednje vrednosti nove uvedene funkcije, prema relacijama (53) biće:

=−

2

2 0

0 2 0

( )d

( )

z

pz z

p z

c z z

c zz z

, =−

1

1 0

0 1 0

( )d

( )

z

pz z

p z

c z z

c zz z

(54)

=−

2

2 1

1 2 1

( )d

( )

z

pz z

p z

c z z

c zz z

(55)

S obzirom na uvedenu smenu (46), promena entro-pije prema (52) biće:

Δ =

= − − −2 1

0 0

ln ln2 0 1 0ln ln

(ln ) (ln ln ) (ln ) (ln ln )T T

p pT T

s

c T T T c T T T (56)

odnosno konačno:

Δ = − = −2 1

0 0

ln ln2 12 1 ln ln

0 0(ln ) ln (ln ) ln

T Tp pT T

T Ts s s c T c T

T T (57)

Relacija (57) predstavlja glavnu relaciju za određi-vanje promene entropije prema drugompredlogu, gde će srednje vrednosti funkcija biti određene u nastavku koji sledi.

Isto tako, prema uvedenoj smeni, relacije za srednje vrednosti funkcije cp(lnT) biće prema relacijama (54) i (55):

=−

2

2 0

0

ln

ln ln

ln2 0

(ln )d(ln )

(ln )ln ln

T

pT T

p T

c T T

c TT T

=−

1

1 0

0

ln

ln ln

ln1 0

(ln )d(ln )

(ln )ln ln

T

pT T

p T

c T T

c TT T

=−

2

2 1

1

ln

ln ln

ln2 1

(ln )d(ln )

(ln )ln ln

T

pT T

p T

c T T

c TT T

(58)

odnosno za opšti slučaj, od početne temperature T0 do neke proizvoljne temperature Ti:

=−

0

0

ln

ln ln

ln0

(ln )d(ln )

(ln )ln ln

i

i

T

pT T

p Ti

c T T

c TT T

(59)

Brojilac u relaciji (59), uzimajući u obzir relaciju (39), s obzirom na izračunatu promenu entropije (50) biće:

= − +

+ − + − +

+ − + −

0

ln

1 0ln

2 2 3 31 10 0

4 4 5 51 10 0

(ln )d(ln ) (ln ln )



iT

p iT

i i

i i

c T T B T T

C DT T T T

E FT T T T

(60)

Zamenom vrednosti dobijene iz relacije (60) u izraz (59) dobija se:

= − +−

+ − + − +

+ − + −

0

ln1 0ln

0

2 2 3 31 10 0

4 4 5 51 10 0

1(ln ) ( (ln ln )

ln ln


(ln ln ) (ln ln ))4 5

iTp iT

i

i i

i i

c T B T TT T

C DT T T T

E FT T T T

(61)

što predstavlja opštu formula za izračunavanje srednje vrednosti funkcije cp(ln T) prema temperaturskoj funk-ciji (41), za proizvoljni interval temperatura T0 do Ti. Pri ovome, konstante B1, C1, D1, E1 i F1 u istoj funkciji, određuju se nekom od numeričkih metoda, na bazi po-znate funkcije cp = cp(T).

Do približnog ali zadovoljavajućeg rešenja može se doći ako se za poznatu funkciju cp = cp(T) usvoji pet tačaka sa koordinatama (Ti,cpi), na približno istom raz-maku. Tako se s obzirom na relaciju (41) dobija sistem od pet linearnih jednačina sa pet nepoznatih: B1, C1, D1, E1 i F1. Dobijeni sistem se rešava na jednostavan način.

Prema tome, promena entropije od stanja 1 do stanja 2, izračunava se prema izvedenoj relaciji (57). Primena relacije (61) očigledno je moguća za proiz-voljnu promenu stanja.

Uvedena funkcija φ = cp(z), s obzirom na smenu (45) može se prikazati dijagramski kao na slici 4. Očigledno, u pitanju je monotono rastuća funkcija. S obzirom na relaciju (44) sledi da je u istom dijagramu moguće predstaviti odnosno planimetrijski odrediti promenu entropije, za izobarsku promenu stanja kao površinu ispod krive φ = cp(z), u granicama Z1 i Z2. Promena entropije prikazana je principijelno šrafiranom povr-šinom. Isto tako, prema istom dijagramu sledi i odre-đivanje promene entropije korišćenjem srednje vred-nosti prikazane funkcije 2

1

zp z

c , preko površine pravo-ugaonika Z1Z22’1’.


624

Primena logaritamske funkcije na grupu poluidealnih gasova

Prikazani model na bazi logaritamske funkcije (41), primenićemo za izračunavanje promene entropije za poluidealne gasove CO2, O2, N2 i H2, usvojene prema prvom modelu.

Pri ovome može se u opštem slučaju poći od po-znate eksperimentalne funkcije cp(T), slika 5.

Za konkretan slučaj može se iskoristiti relacija (29) koja je korišćena za istu grupu poluidealnih gasova kod primene razlomljene funkcije.

Prema dijagramu na slici 5, interval T0–Tmax može se, radi veće tačnosti, podeliti, na primer na 7 delova, pri razmaku ΔT = 200 K, odnosno ΔT = 170 K na kraju intervala. Svakoj tački 0, 1, 2, 3,…, 7, na ordinati odgo-vara neka vrednost cpi. Koristeći poznate koordinate prethodnih tačaka, s obzirom na logaritamsku funkciju (41), sastavljen je odgovarajući numerički program [21– –24], koji je omogućio izračunavanje konstanti B1, C1, D1, E1 i F1, odnosno definisanje iste funkcije u aproksi-mativnom obliku.

Dobijeni rezultati su prikazani u tabeli 4. Lako se može pokazati da je pri ovome postignuta

zadovoljavajuća korelaciona tačnost.

Slika 4. Planimetrijsko određivanje promene entropije u dijagramu φ = cp(z). Figure 4. Planimetric determination of the change in entropy in the diagram φ = cp(z).

Slika 5. Eksperimentalna funkcija cp(T) za poluidealan gas. Figure 5. The experimental function cp(T) for a semi-ideal gas.


625

Prema funkciji (61), koristeći odgovarajući nume-rički program, sastavljena je tabela 5 za CO2, O2, N2 i H2 kao poluidealne gasove.

Zapaža se da logaritamska funkcija prema istoj tabeli raste sa temperaturom za sve gasove.

Korišćenjem vrednosti iz tabele 5, izračunata je pro-mena entropije izobarske promene stanja za tri tempe-raturska intervala, za gasove CO2, O2, N2 i H2, radi pore-đenja rezultata. Za izračunavanje, korišćena je relacija (57) pri početnoj temperaturi T0 = 100 K. Rezultati su sistematizovani u tabeli 6.

Radi procene greške prema približnom modelu (ta-bela 6) izračunate su apsolutne i relativne greške za četiri karakteristična gasa i sve temperaturne intervale. Rezultati su sistematizovani u tabeli 7.

Apsolutna greška, Δxi, izračunata je kao razlika rezultata dobijenih prema približnom modelu (tabela 6) i rezultata prema tačnom modelu (tabela 3).

Relativna greška, δxi, je dobijena kao količnik apso-lutne greške i rezultata dobijenih prema približnom modelu (tabela 3).

Tabela 5. Srednje vrednosti0

lnln(ln ) iT

p Tc T za CO2, O2, N2 i H2 kao poluidealne gasove pri T0 = 100 K, u kJ/(kg K2) Table 5. The mean values

0

lnln(ln ) iT

p Tc T of CO2, O2, N2 and H2 as a semi ideal gases at T0 = 100 K, in kJ/(kg K2)

Ti / K CO2 O2 N2 H2 100 0 0 0 0 200 0,7967 0,8874 0,3305 13,0037 300 0,8082 0,9414 0,5402 13,5194 400 0,8355 0,9602 0,6710 13,7294 500 0,8637 0,9703 0,7405 13,8374 600 0,8898 0,9781 0,8101 13,9050 700 0,9133 0,9850 0,8263 13,9552 800 0,9344 0,9917 0,8422 13,9979 900 0,9533 0,9982 0,8482 14,0376 1000 0,9704 1,0044 0,8502 14,0763 1100 0,9859 1,0103 0,8591 14,1151 1200 1,0000 1,0159 0,8742 14,1545 1300 1,0130 1,0210 0,8890 14,1945 1400 1,0250 1,0258 0,9130 14,2353 1470 1,0329 1,0288 0,9234 14,2642

Tabela 4. Koeficijenti u funkciji cp = cp(lnT) za CO2, O2, N2 i H2 kao poluidealne gasoveTable 4. Coefficients in function of cp = cp(lnT) for CO2, O2, N2 and H2 as semi ideal gases

Gas B1 C1 D1 E1 F1 CO2 59,3340 –36,1959 8,2216 –0,8155 0,0301 O2 187,1310 –116,3438 27,0968 –2,7912 0,1075 N2 –682,2346 452,9835 –112,5957 12,4479 –0,5171 H2 –429,6836 258,9139 –55,4505 5,1370 –0,1718

Tabela 6. Promena entropije izobarske promene (kJ/(kg K)) za CO2, O2, N2 i H2 kao poluidealne gasove, izračunata na bazi logaritamske funkcije Table 6. The change in entropy of the isobaric change (kJ/(kg K)) of CO2, O2, N2 and H2 as semi-ideal gases calculated on logarithmic function

Temperaturni interval Opseg T1–T2, K Δ = −2 1

0 0

ln ln2 1ln ln0 0

(ln ) ln (ln ) lnT T

p pT T

T Ts c T c TT T

CO2 Širi 200–1300 2,0161 Srednji 400–1100 1,1684 Uži 600–800 0,3387

O2 Širi 200–1300 1,9837 Srednji 400–1100 1,0915 Uži 600–800 0,3097

N2 Širi 200–1300 2,0510 Srednji 400–1100 1,0991 Uži 600–800 0,3011

H2 Širi 200–1300 27,3947 Srednji 400–1100 14,8105 Uži 600–800 4,1934


626

Maksimalna apsolutna greška je 0,1303 kJ/(kg K), dok je maksimalna relativna greška 7,03%.

Isto tako tačnost dobijenih rezultata nije ista kod svih posmatranih gasova. Tačnost je moguće poboljšati, korišćenjem više od 8 eksperimentalnih tačaka, prema dijagramu na slici 5.

ZAKLJUČAK

Predložena metoda rešava postavljeni problem direktno, primenom tabličnih vrednosti,izračunatih na bazi temperaturnih funkcija koje zavise od stvarnog specifičnog toplotnog kapaciteta, za proizvoljan tempe-raturni interval odnosno razmak. Kod postojeće inte-gralne metode, za svaki konkretan slučaj sprovodi se obimna i glomazna procedura koja je često izvor gre-šaka, relacija (32).

Predložena razlomljena funkcija daje gotovo tačne rezultate pa se ona preporučuje za korišćenje posebno zbog jednostavnije matematičke procedure, odnosno jednostavnog računarskog programa. Logaritamska funkcija je takođe primenljiva ali je nešto složenija u matematičkom smislu. Pri njenoj primeni javlja se odre-đena greška, usled aproksimacije eksperimentalne funkcije, koja je u dozvoljenim granicama ukoliko se usvoji dovoljno veliki stepen funkcije, odnosno dovoljan broj eksperimentalnih tačaka.

Isto tako oblik polazne funkcije zavisnosti stvarnog specifičnog toplotnog kapaciteta od temperature u principu nema uticaja sa aspekta primene metode.

Prikazani model za razlomljenu funkciju može se direktno koristiti za sastavljanje tabela srednjih vred-

nosti funkcija za svaku vrstu poluidealnog gasa pri čemu se razmak temperatura može usvojiti po želji, zavisno od tražene tačnosti (na primer ΔT = 10 K). Model sa logaritamskom funkcijom, takođe je moguće primeniti za rešavanje problema ali on zahteva odre-đivanje aproksimativne funkcije za svaki gas posebno.

Početna temperatura, T0, mora biti usvojena tako da bude različita od nule zbog pojave nesvojstvenog integrala u računu. S obzirom da ona ne utiče na pro-menu entropije, najpogodnije je početnu temperaturu usvojiti kao najnižu temperaturu, na početku razmatra-nog temperaturnog intervala. Za slučajeve kada se izra-čunava entropija u nekoj određenoj tački, neophodno je naglasiti usvojenu početnu tačku. Isto tako, predlo-ženi model u opštem slučaju obuhvata široko tempera-turno područje, uključujući i negativne temperature, što može imati praktični značaj.

U termodinamici kao što je poznato, veličine pro-cesa, rad i razmenjena toplota predstavljaju se i odre-đuju planimetrijski preko odgovarajućih površina u rad-nom i toplotnom dijagramu. U radu je pokazano da je na isti način, koristeći odgovarajuće koordinate, mo-guće predstaviti odnosno odrediti i promenu entropije poluidealnog gasa pri čemu treba posebnu pažnju obra-titi na koeficijente razmere pri konstruisanju dijagrama. Ovo omogućuje potpunije sagledavanje i praćenje pos-matranog procesa.

Verifikacija modela izvedena je s obzirom na karak-terističnu izobarsku promenu stanja, ali je promenu entropije poluidealnog gasa prema oba prikazana modela, moguće odrediti koristeći isti postupak, za proizvoljnu promenu stanja.

Tabela 7. Apsolutna i relativna greška približnog modelaTable 7. Absolute and relative error of approximate model

Temperaturni interval

Opseg T1–T2, K

Apsolutna greška -1 -1/ kJ kg Ki ix x xΔ = −

Relativna greška

100 / %ii

xxx

δ Δ=

CO2 Širi 200–1300 0,1056 5,50 Srednji 400–1100 0,0500 4,47 Uži 600–800 0,0155 4,79

O2 Širi 200–1300 0,1303 7,03 Srednji 400–1100 0,0580 5,61 Uži 600–800 0,0135 4,55

N2 Širi 200–1300 0,0174 0,85 Srednji 400–1100 0,0129 1,16 Uži 600–800 0,0144 4,56

H2 Širi 200–1300 0,0680 0,25 Srednji 400–1100 0,0148 0,10 Uži 600–800 0,0098 0,23


627

Za slučaj realnog gasa, za koji važi van der Valsova jednačina stanja, ukoliko su za određeni temperaturni interval poznati koeficijenti u korelacionoj jednačini specifičnog toplotnog kapaciteta, moguće je takođe primeniti prikazanu metodu.

Na kraju, prema prikazanoj metodi, koristeći izve-dene opšte izraze i odgovarajuće računarske programe, predlaže se sastavljanje termodinamičkih tablica u pri-ručnicima za različite poluidealne gasove, što bi znatno ubrzalo rešavanje postavljenog problema u tehničkoj praksi, na efikasniji i brži način.

LITERATURA

[1] M.M. Abbott, H.C.Van Ness, Thermodynamics, Sehanm outline Series, McGrow-Hill, Book Co, New York, 1986.

[2] H.D. Baehr, Termodynamik, Springer-Verlag, Berlin, 1983.

[3] I.P. Bazarov,Termodinamika, V.Š., Moskva, 1993. [4] W.Z. Black, J.G. Hartlay, Thermodynamics, Harper and

Row, New York, 1995. [5] J.M. Michael, Fundamentals od Engineering Thermo-

dynamics, Wilay, New York, 2004. [6] J. Szargut, Termodynamika, PWN, Warszawa, 1995. [7] Z. Rant, Termodinamika-knjiga za uk i prakso, Univerza v

Ljunljane, Ljubljana, 1993. [8] Đ. Kozić, B.Vasiljević, B. Bekavoe, Priručnik za termo-

dinamiku, Mašinski fakultet, Beograd, 2007. [9] Đ. Kozić, Termodinamika, inženjerski aspekti, Mašinski

faklutet, Beograd, 2007. [10] F. Bošnjaković, Nauka o toplini, Tehnička knjiga, Zagreb,

1982. [11] G. Büki, Energetika, MK, Budapest, 1997. [12] J.B. Fenn, Engines, Enegry and Entropy, W.H. Freeman

and Comp., New York, 1982. [13] B.V. Karleker, Thermodynamics for Engineers, Prentice-

Hall, Inc., Englewood Cliffs, 1993. [14] D.S. Mitrinović, Matematika I i II, Građevinska knjiga,

Beograd, 1987. [15] C.B. Allendoerter, Principlls of mathematics, BCH, New

York, 1985. [16] J.M. Smith, H.C. Van Ness, Internetion to Chemical Eng-

ineering Thermodynamics, third Edition, McGraw-Hill, New York, 1985.

[17] B. Đorđević,V. Valent, S. Šerbanović, N. Radojković, Ter-modinamika i termotehnika, priručnik, Građevinska knjiga, Beograd, 1989.

[18] B. Đorđević, V. Valent, S. Šerbanović, Termodinamika i termotehnika, Građevinska knjiga, Beograd, 1997.

[19] S. Perry, Chemical Engineers Handbook, McGraw-Hill, New York, 1984.

[20] VDI Wärmeatlas Berechenugsblätter für den Wärme-übergang, VDI Verlag GmbH, Düsseldorf, 1984.

[21] D. Herceg, Numeričke metode linarne algerbe, Građe-vinska knjiga, Beograd 2003.

[22] H. Schenck, Theories of Engineering Experimentation, Mc Graw-Hill Book Company, New York, 2002.

[23] L. Rumnšiskij, Matematičeskaja obrabotka rezultatov eksperimenta, Mašinostvorenie, Moskva, 2001.

[24] L. Collatz, Numerische Behandlung von Differentiol-gleichungen, KTS, Berlin, 1998.

[25] A. Bejan, Advanced Engineering Thermodynamics, John Wiley and Sons, New York, 1997.

[26] R.K. Rajput, Engeneering thermodynamics, Jones and Bartlett Publishers, London, 2010.

[27] R.K. Rajput, Thermal Engeneering, Jones and Bartlett Publlishers, London, 2009.

[28] M. J.Moran, H.N. Shapiro, B.B. Boettner, M.B. Bailey, Engeneering thermodynamics, John Wiley and Sons, New York, 2011.

[29] Y.A. Cengel, M.A. Boles, Thermodynamics, McGraw Hill Higher Education, New York, 2010.

[30] C. Borgnakke, R. E. Sonntag, G. J. Wylen, Fundamentals thermodynamics, John Wiley and Sons, New York, 2009.

[31] B. Pejović, M. Perušić, V. Mićić, G. Tadić, S. Pavlović, Jedna mogućnost grafičkog predstavljanja energetskih veličina realnog gasa za karakterističnu promenu stanja, Termotehnika XXXIX (2013) 11–25.

[32] B. Pejović, Lj. Vasiljević, V. Mićić, M. Perušić, Jedan pogodan model za određivanje korelacije između stvar-nog i srednjeg specifičnog toplotnog kapaciteta sa mo-gućnostima njegove primene, Hem. Ind. 67 (2013) 495– –511.

[33] B. Pejović, M.Tomić, V. Mićić, Određivanje korelacije između toplotnog kapaciteta i osnovnih termodinamič-kih veličina stanja primenom diferencija drugog reda, KGH 4 (2012) 43–48.


628

SUMMARY PROPOSAL FOR DETERMINING CHANGES IN ENTROPY OF SEMI IDEAL GAS USING MEAN VALUES OF TEMPERATURE FUNCTIONS Branko B. Pejović, Vladan M. Mićić, Mitar D. Perušić, Goran S. Tadić, Ljubica C. Vasiljević, Slavko N. Smiljanić

Faculty of Technology Zvornik, University of East Sarajevo, Republic of Srpska, BH

(Professional paper)

In a semi-ideal gas, the entropy changes cannot be determined through themedium specific heat capacity in a manner as determined by the change of inter-nal energy and enthalpy, i.e., the amount of heat exchanged. Taking this intoaccount, the authors conducted two models through which it is possible to deter-mine the change in the specific entropy of a semi-ideal gas for arbitrary tempe-rature interval using the spread sheet method, using the mean values of theappropriate functions. The idea is to replace integration, which occurs here inevitable, with mean values of the previous functions. The models are derivedbased on the functional dependence of the actual specific heat capacity on thetemperature. The theorem used is that of the mean value of a function, as well as the mathematical properties of the definite integral. The mean value of a frac-tional function is determined via its integrand, while the logarithmic functionswere performed by applying a suitable transformation of the differential calculus. The derived relations, using the computer program, have enabled the design ofappropriate thermodynamic tables through which it is possible to determine thechange in entropy of arbitrary state changes in an efficient and rational manner,without the use of calculus or finished forms. In this way, the change in theentropy of a semi-ideal gas is determined for an arbitrary temperature intervalusing the method which is analogous to that applied in determining the change ofinternal energy and enthalpy or the amount of heat exchanged, which was thegoal of the work. Verification of the proposed method for both the above func-tions was performed for a few characteristic semi-ideal gases where change c(T) is significant, for the three adopted temperature intervals, for the characteristicchange of state. This was compared to the results of the classical integral and theproposed method through the prepared tables. In certain or special cases, it ispossible to apply the presented method also in determining the change in entropyof the real gas. Apart from that, the paper shows that the change in entropy forthe observed characteristic case can be represented or graphically determinedusing the planimetric method of diagrams with suitably selected coordinates.

Keywords: Semi-ideal gas • The change in entropy • Mean and true specific heat capacity • The mean value of the func-tion • Differential and integral calculus •Table value functions • Approximate functions • Graphical methods

629

Mikro-elektro-mehanički sistemi (MEMS) – Tehnologija za 21. vek

Tatjana A. Djakov, Ivanka G. Popović, Ljubinka V. Rajaković

Tehnološko–metalurški fakultet, Univerzitet u Beogradu, Beograd, Srbija

Izvod Mikro-elektro-mehanički sistemi (MEMS) pripadaju minijaturnim elektromehaničkim siste-mima (uređaji i strukture) koji mogu da registruju promene u okolini, da ih analiziraju iprocesiraju pomoću mikroelektronike. MEMS čine mehanički elementi, senzori, pojačivači,električni i elektronski uređaji koji su smešteni na silicijumov supstrat (čip). MEMS uređajisu minijaturni (dimenzije), optimalni (energetski i ekonomski), adaptabilni (lako se inte-grišu u druge sisteme i lako se modifikuju), niskog koeficijenata toplotnog širenja, velikeotpornosti na vibracije, udar i zračenje. Tehnologije koje se primenjuju za proizvodnju MEMS i MEMS prodiru u svakodnevni život, na sličan način kao mikroelektronika. Moguć-nost šaržne proizvodnje u velikim serijama otvorila je široku, komercijalnu primenu MEMSuređaja u biomedicini, telekomunikacijama, bezbednosti i zabavi. MEMS povezuje oblasti koje nisu imale dodirne tačke kao što su biologija, mikroelektronika i nanotehnologija.MEMS se razvija u pravcima koji nadrastaju trenutna saznanja i otkrića tako da se smatrada je MEMS osnova za uređaje i tehnologije 21. veka.

Ključne reči: mikro-elektro-mehanički sistemi (MEMS), senzori, pokretači, čip.

PREGLEDNI RAD

UDK 621.38:681.586

Hem. Ind. 68 (5) 629–641 (2014)

doi: 10.2298/HEMIND131008091D


O MEMS

Mikro-elektro-mehanički sistemi (skraćeno MEMS) su mehanički ili elektromehanički sistemi (uređaji i strukture) koje pokreće električna energija [1]. Ovi sis-temi mogu da registruju, kontrolišu i aktiviraju meha-ničke procese (koji se odigravaju na nevidljivoj mikro-skali) koje dalje pojedinačno ili u nizu generišu u efekte vidljive na makro skali. Svaki deo naziva za MEMS ima poseban smisao: mikro (strukture malih dimenzija koje se dobijaju mikrofabrikacijom), elektro (električni sig-nal/kontrola), mehanički (mehanička funkcionalnost), sistemi (strukture, uređaji ili sistemi) [2]. MEMS su se razvili kao logičan nastavak mikroelektronike i integri-sanih kola [3]. Za električna kola je karakteristično da su čvrste i kompaktne strukture, MEMS strukture imaju šupljine, kanale, konzole, membrane. MEMS se razli-kuje od mikroelektronike, od molekularne nanotehno-logije, kao i od molekularne elektronike. Na slici 1 dat je ilustrativan prikaz međusobnog prožimanja optike, mehanike i elektronike koje su osnova za razvoj MEMS, mikro-opto-elektro-mehaničkih sistema (MOEMS), optoelektronike i optomehanike [3].

MEMS čine mehanički elementi, senzori, pokretači (aktuatori), električni i elektronski uređaji koji su smeš-teni na supstrat silicijuma (čip) [4]. Senzori u okviru MEMS sakupljaju informacije iz okoline merenjem mehaničkih, toplotnih, bioloških, hemijskih, optičkih i magnetnih efekata. Elektronika procesira ove infor- Prepiska: T.A. Djakov, Tehnološko–metalurški fakultet, Karnegijeva 4, 11000 Beograd, Srbija. E-pošta: [email protected] Rad primljen: 8. oktobar, 2013 Rad prihvaćen: 10. decembar, 2013

macije i uz mogućnost donošenja odluka usmerava pokretače (aktuatore) da svrsishodno željenom odzivu reaguju prema okolini (na primer pomeranjem, pozicio-niranjem, regulisanjem, pumpanjem i/ili filtriranjem) [5].

Slika 1. Osnovne tehnike (optika, mehanika i elektronika) za razvoj MEMS, MOEMS, optoelektronike i optomehanike. Figure 1. Basic technics (optics, mechanics and electronics) for MEMS, MOEMS, optoelectronic and optomechanic development.

Veličina MEMS komponenti (senzora, pokretača i elektronike) iznosi od 1 do 100 μm, dok je sam MEMS uređaj veličine od 20 μm do 1 mm. Tipovi MEMS ure-đaja mogu da variraju od relativno jednostavnih – bez pokretnih delova, do izuzetno kompleksnih elektrome-haničkih sistema sa mnogo pokretnih elemenata koje kontrololiše integrisana mikroelektronika. U MEMS ure-đaje spadaju senzori za pritisak, merači ubrzanja-akce-lerometri kao inertni senzori, mikroogledala, minija-turni roboti, zupčanici, pumpe za fluide, senzori za pro-tok, generatori mikrokapljica, optički skeneri, uređaji za analizu i dobijanje slika, hemijski senzori, sonde i šiljci za ispitivanje površine [5].

T.A. DJAKOV i sar.: MIKRO-ELEKTRO-MEHANIČKI SISTEMI Hem. ind. 68 (5) 629–641 (2014)

630

Istorijat MEMS

Istorijat MEMS je vezan za razvoj svake od tehnika na kojima se zasniva MEMS tehnologija, ali se smatra da su prvi fenomeni koji su posledica ovih integrisanih sistema zabeleženi pedesetih godina dvadesetog veka [6]. U tabeli 1 prikazan je hronološki istorijat otkrića i proizvodnje uređaja koji su preteča MEMS.

Šest najvećih kompanija za proizvodnju MEMS ure-đaja u ovom trenutku u svetu su: Robert Bosch (Ne-mačka), ST MIKRO (Italija/Francuska), Lexmark (SAD) – ranije IBM, SEIKO-EPSON (Japan), Bei Technologies (SAD, Kalifornija) i Analog Devices (SAD, Boston). One proizvode merače ubrzanja, žiroskope, raznovrsne sen-zore za pritisak, štampače i glave za mlazne (ink džet) štampače.

Osnovne ideje za razvoj i primenu MEMS

Tri su osnovne ideje za razvoj i primenu MEMS: 1. minimizacija postojećih uređaja (primer proiz-

vodnje žiroskopa na bazi silikona: masa postojećih ure-

đaja od kilograma i zapremine od 1 dm3 smanjena je na čip mase nekoliko grama i zapremine 0,5 cm3),

2. primena principa i otkrića moderne fizike (bio-čipovi koji koriste električno polje za doziranje reak-tanta oko čipa, na osnovu elektro-osmotskog efekta u kanalima prečnika do 1 mm) i

3. primena mikrotehnika (izgradnja i razvoj uređaja kao što su delovi skenirajućeg mikroskopa sa tunelskim efektom, STM i mikroskopa atomskih sila, AFM). MEMS uređaji se primenjuju kao mikrogriperi (hvatači) za hva-tanje ćelija koje treba analizirati.

MEMS uređaji se primenjuju u [3]: • automobilskoj industriji (u sistemima vazdušnih

jastuka, bezbednosnim sistemima u vozilu, svetlima za kočnice, uređajima za pozicioniranje prednjih farova, detektovanje proklizavanja i automatsko zaključava-nje). Na slici 2 prikazani su delovi automobila u kojima se nalaze MEMS uređaji,

• proizvodima široke potrošnje (u aparatima i spravama za sportske treninge, perifernim uređajima za

Tabela 1. Hronološki istorijat otkrića i izgradnje uređaja koji su preteča MEMS Table 1. Historical overview of most important discoveries for development of MEMS (chronological order)

Godina Otkriće, fenomen ili uređaj bitni za razvoj i ugradnju MEMS Ref. 1947. U laboratoriji AT&T Bell je napravljen prvi tranzistor [1] 1954. Otkriven je piezootporni efekat silicijuma [2] 1958. U kompaniji Texas Instruments napravljeno je prvo integrisano kolo [2] 1960. Počinje proizvodnja senzora [6] 1979. Dobijene su prve mlaznice za ink-džet patrone (kertridže) procesom mikrofabrikacije [3] 1988. Napravljen je prvi mikromotor sa elektrostatičkom površinom [1] 1994. Počela je proizvodnja uređaja koji se zasnivaju na digitalnom procesiranju svetla (Digital Light Processing,

DLP tehnologija) [4]

2001. U kompaniji Apple je proizveden i predstavljen prvi iPod [2,5] 2006. Proizveden je uređaj za igrice, Nintendo Wii [2,5] 2007. U kompaniji Apple je proizveden i predstavljen prvi iPhone [2,4] 2010. Proizveden je prvi iPad [2,5]

Slika 2. Primena MEMS u automobilu. Figure 2. Automotive applications of MEMS.


631

kompjutere, navigacionim uređajima za automobile i ličnu upotrebu, u mobilnim aplikacijama),

• industriji (u uređajima za detektovanje zemljo-tresa, za regulaciju/isključivanje gasa, za testiranje ispravnosti aparata i mašina, registrovanje udara i po-tresa),

• vojnoj industriji (u uređajima u tenkovima, avio-nima i opremi za vojnike),

• medicini, biomedicini i biotehnologijama (u koje spadaju i biočipovi za detektovanje opasnih hemijskih i bioloških supstanci, mikrosistemi za DNK identifikaciju, mikrofabrikovani STM uređaji, mikrosistemi za sorti-ranje i selekciju lekova i bio-MEMS uređaji u medicini i tehnologijama bliskim medicini (od tzv. „laboratorije na čipu” do biosenzora i hemosenzora)) i

• vazduhoplovstvu i istraživanju svemira (to su merači ubrzanja i žiroskopi za unutrašnju navigaciju, senzori za pritisak, radiofrekventni prekidači i filteri za komunikaciju, harmonična ogledalca za optiku, mikro-energetski izvori i turbine, uređaji za kontrolu pokreta i položaja, bio-reaktori i bio-senzori, uređaji sa mikro-fluidima, uređaji sa toplotnom kontrolom i atomski satovi).

Pravci razvoja MEMS do 2017. godine

Tržište MEMS uređaja predstavlja deo velikog tržišta poluprovodničkih čipova, a to je tržište koje pokazuje najbrži rast. U početku je automobilska industrija bila glavni pokretač tržišta za MEMS uređaje, ali se rast ove grane poslednjih godina industrije usporio. S druge strane, novije oblasti primene (primer su oblast medi-cine, telekomunikacija i proizvodi široke potrošnje) imaju značajniji uticaj na tržište MEMS [7].

Veliki zahtevi tržišta za elektronskim uređajima najnovije generacije, kao što su inteligentni telefoni (smart phones) i tablet uređaji utiču na porast proiz-vodnje i razvoj MEMS. Procena je da će tržište MEMS uskoro dostići nivo od 15-20 milijardi dolara. Ukoliko se u analizu tržišta uključi i oblast mikrofluidike na staklu,

polimerima i SiO2, kao i digitalni kompasi koji nisu u pravom smislu reči MEMS uređaji, ali se sve više inte-grišu sa njima, predviđanja su da će godišnji finansijski rast biti oko 15, a čak 20% kada je reč o količini proiz-vedenih MEMS uređaja. Ako se razmatraju glavni MEMS uređaji, tržište za radiofrekventne (RF) MEMS uređaje će imati najveći rast (50%), zatim slede čipovi sa mikrofluidikom koji služe za distribuciju lekova (42%), mikrofoni na bazi silicijuma (32%), čipovi sa mik-rofluidikom za dijagnostiku (23%), mikrosonde i mikro-špricevi (22%) i mikrobolometri (20%) [8].

MEMS u proizvodima široke potrošnje, mobilne aplikacije, na primer iPhone, iPad i iPod, koji su pri-kazani na slici 3, predstavljaju 50% ukupne količine MEMS uređaja. U 2012. godini je proizvedeno 7 mili-jardi MEMS uređaja, što predstavlja rast od 17% u odnosu na 2011. godinu, a čak 54% u odnosu na 2010. godinu.

Detekcija pomeraja (rotiranje ekrana od vertikalnog ka horizontalnom položaju) je našla široku primenu u mobilnim telefonima, pejsmejkerima, inteligentnoj municiji što je uslovilo rast potražnje i proizvodnje merača ubrzanja, žiroskopa i elektronskih kompasa. MEMS mikrofoni se ugrađuju u mobilne telefone i zamenjuju do skora korišćene mikrofone na bazi elek-trokondenzatora. Mikrofoni, merači ubrzanja, žiroskopi i magnetometri čine više od polovine svih MEMS ure-đaja napravljenih u 2011. godini [8].

Razvoj novih MEMS

Razvoj novih MEMS uređaja uključuje termoniti, mikrodispleje, mikroogledala za mobilne telefone i pikoprojektore u tablet uređajima, autofokus, RF MEMS prekidače i promenljive kondenzatore za mobil-ne uređaje, oscilatore, mikrozvučnike. Primena MEMS zapravo nema ograničenja. Tržišta novih MEMS uređaja će intenzivno rasti, a procenjeni rast u sledećem peto-godišnjem periodu je najveći za mikroogledala koji se koriste za mobilne telefone i pikoprojektore u tablet

Slika 3. Proizvodi široke potrošnje u kojima se koriste MEMS uređaji: a) iPhone i iPad i b) iPod. Figure 3. Consumer/mobile MEMS applications: a) iPhone and iPad and b) iPod.


632

uređajima, za RF MEMS prekidače i promenljive kon-denzatore za mobilne uređaje, za oscilatore, za ter-moniti i za MEMS uređaje kao što su mikrozvučnici [8].

TEHNOLOGIJE I POSTUPCI ZA PROIZVODNJU MEMS

Proizvodnja MEMS proističe iz procesnih tehnolo-gija razvijenih u proizvodnji poluprovodnika kao što su: depozicija slojeva materijala, fotolitografija za dobijanje šablona i nagrizanje za dobijanje željenih završnih formi. MEMS odlikuje velika vrednost odnosa povr-šina/zapremina (takozvanog „aspekt odnosa”), pa su površinski efekti (kao što su kvašenje i elektrostatički efekti) dominantni u odnosu na zapreminske efekte (kao što su inercija i toplotna energija) [9].

Procesne tehnologije, razvijene za dobijanje polu-provodnika koriste se i za proizvodnju MEMS uređaja od kojih su tri najčešće primenjivane:

I. mikrofabrikacija u masi, II. mikrofabrikacija površine i III. LIGA proces. Ove tehnologije su bile predmet intenzivnih istraži-

vanja i razvijale su se u poslednjih dvadeset pet godina, a podrobno su opisane u literaturi [1,6,9–13].

Proces mikrofabrikacije u masi

Mikrofabrikacija u masi se zasniva na tehnikama nagrizanja pripremljenih silikonskih slojeva (pravac je od vrha ka dnu) sa ciljem da se stvore trodimenzionalne (3D) MEMS strukture [11]. To je proces u kojem mogu da se koriste mokri anizotropni postupak ili suvi postupak nagrizanja (pomoću reaktivnih jona, RIE – reactive ion etching), kako bi se dobile velike šupljine, brazde i kanali [4,9]. Materijali koji se koriste u mokrom postupku nagrizanja su silicijum i kvarc, dok se u pos-tupku suvog nagrizanja koriste silicijum, metali, poli-meri i keramika [11].

Mokri postupak nagrizanja Kod mokrog postupka nagrizanja material se potapa

u hemijsko sredstvo za nagrizanje [13] koje može biti izotropno (smeša HF, HNO3 i CH3COOH) ili anizotropno (KOH) [14]. Anizotropna sredstva za nagrizanje imaju svojstvo da brže nagrizaju u željenom pravcu. Brzina nagrizanja i kvalitet dobijene površine zavise od hemij-skog sastava i kristalne orijentacije supstrata [15].

Suvi postupak nagrizanja Kod suvog postupka nagrizanja koristi se plazma –

joni se ubrzavaju i usmeravaju ka materijalu koji treba da se nagrize obezbeđujući energiju potrebnu za reak-ciju [4]. Najčešći korišćeni postupak za dobijanje MEMS struktura je nagrizanje pomoću reaktivnih jona (RIE)

primenom dopunske energije radio frekvencija [16–18].

Proces površinske mikrofabrikacije

Kod površinske mikrofabrikacije, trodimenzionalne strukture se dobijaju simultanim dodavanjem ili ukla-njanjem slojeva tankih filmova na površinu strukture [2,4]. Slojevi čiji se delovi uklanjaju su nazvani „žrtveni” slojevi, a nanose se i zatim uklanjaju sa strukturnih slojeva da bi se formirale mehaničke rupe ili otvori između strukturnih slojeva [11]. Materijali koji se naj-češće koriste za filmove kod površinskog mikroprocesi-ranja MEMS i tradicionalnih mikroelektričnih uređaja su silicijum-dioksid, silicijum-nitrid, polikristalni silicijum-dioksid (polisiloksan) i metali.

Dubinsko nagrizanje pomoću reaktivnih jona (DRIE) Metoda dubinskog nagrizanja pomoću reaktivnih

jona (DRIE – deep reactive ion etching) uključuje alter-nativni proces nagrizanja struktura jonizujućim gasom tzv. plazmom velike gustine (kao kod RIE), kao i pos-tupak deponovanja polimera kao zaštitnog sloja, pa je odnos površina/zapremina mnogo više izražen [11].

LIGA i UV-LIGA procesi

LIGA je trostepeni proces – sastoji se od litografije X-zracima, galvanizacije i polimernog kopiranja-replici-ranja [19]. Naziv procesa je nastao kao nemački akro-nim od slova LI (Roentgen Litographie, odnosi se na litografiju X-zracima), G (Galvanik, odnosi se na proces galvanizacije) i A (Abformung, odnosi se na modelova-nje (izlivanje) struktura, koje imaju velike vrednosti aspekt odnosa, od potpuno različitih materijala). U poslednje vreme LIGA proces je prerastao u dvostepeni proces – sastoji se od litografije X-zracima i galvani-zacije. U ovoj tehnici se debeli slojevi fotootpornog materijala izlažu dejstvu X ili UV zraka da bi se dobile matrice, koje se zatim koriste za formiranje trodimen-zionalnih struktura sa velikim odnosom površina/za-premina, elektrolitičkom depozicijom [20]. Materijali koji se koriste u LIGA procesu za litografiju X-zracima su poli(metil metakrilat) (PMMA) i epoksidne smole, a u UV-LIGA procesu se pored epoksidnih smola koriste i materijali osetljivi na UV zračenje [21]. Pomoću LIGA procesa mogu da se dobiju mikrostrukture, koje su manje od onih dobijenih konvencionalnim procesima fabrikacije, a veće od površinski mikrofabrikovanih delova.

MATERIJALI

MEMS i delovi za MEMS uređaje proizvode se od silicijuma, polimera, metala, keramike i kompozitnih materijala.

Silicijum je, kao materijal koji se koristi za proizvod-nju integrisanih kola, najzastupljeniji u modernom svetu. Dostupnost finansijski povoljnih visokokvalitet-nih materijala i mogućnost ugrađivanja elektronske


633

funkcionalnosti, omogućile su da silicijum bude atrak-tivan za široku primenu kod MEMS aplikacija [4,10,22].

Iako se veliki deo elektronske industrije oslanja na industriju silicijuma, proizvodnja kristalnog silicijuma je još uvek kompleksna i relativno skupa. Pored toga što je krt i nefleksibilan, silicijum mora da se prečisti pre primene. Zamena za silicijum su polimeri, jer mogu da se proizvode u velikim količinama uz raznovrsne karak-teristike.

Polimeri se intenzivno koriste kao strukturni i funk-cionalni materijali za mikrouređaje. Osnovne prednosti polimera su elastičnost, optička svojstva i biokompa-tibilnost. Različiti polimerni uređaji se prave od tankih i/ili debelih polimernih filmova i trodimenzionalnih (3D) polimernih mikrostruktura. Neki polimeri koji se koriste su poliimidi [23–25], polisiloksani (na primer poli(dime-til siloksan) – PDMS) [26,27], epoksidne smole (na pri-mer SU-8) [28–35], poli-p-ksilen tzv. parilen C [36–38], poliuretani, polisulfoni, polimetakrilati [39], silikonske gume, poliakrilati, poliestri i polikarbonati. Ovi polimeri mogu takođe da se koriste za konstrukciju osetljivih i pokretnih komponenti za MEMS, kao što su mikrosen-zori i mikropokretači. Procesi koji se koriste za dobi-janje MEMS uređaja su injekciono brizganje, utiskivanje ili stereolitografija. Polimeri su kao materijali posebno pogodni za primenu u mikrofluidici [40–48] (primer su patrone za jednokratnu upotrebu pri testiranju krvi).

Metali za proizvodnju MEMS elemenata su zlato, nikl [49], bakar, volfram, aluminijum, hrom, titan, pla-tina i srebro.

Keramički materijali koji se sve više koriste za pro-izvodnju MEMS su nitridi silicijuma [50], aluminijuma i titana, kao i silicijum-karbid [51,52].

MIKROSENZORI I MIKROPOKRETAČI

Najznačajniji i najinteresantniji elementi MEMS sis-tema su mikrosenzori i mikropokretači. To su uređaji koji konvertuju energiju iz jednog oblika u drugi. Kod mikrosenzora, tipično je da uređaj pretvara mereni mehanički signal u električni signal. Mikrosenzori takođe mogu da detektuju promene u okolini mere-njem toplotnih, magnetnih, hemijskih ili elektromag-netnih efekata. Ove informacije se obrađuju pomoću mikroelektronike, koja daje signal mikropokretačima da reaguju u vidu promene prema okolini.

Mikrosenzori

Poslednjih decenija razvijeni su brojni mikrosenzori koji omogućavaju očitavanje temperature, pritiska, inercijalnih sila, hemijskih vrsta, magnetnih polja, zra-čenja [3,25,53–57]. Sposobnost MEMS senzora da mere različite parametre se zasniva na ograničenom broju mehanizama pretvaranja koji su kompatibilni sa pro-cesom minimizacije, a u koje spadaju piezootporni [53],

kapacitivni [54], piezoelektrični [55,56] i u nekoliko slučajeva induktivni efekat.

Tri osnovna tipa senzora su: senzori za pritisak (piezootpornost), hemijski senzori (kapacitivnost) i iner-cijalni senzori (akcelerometri zasnovani na kapacitiv-nom efektu i žiroskopi zasnovani na piezoelektričnom efektu).

MEMS senzori za pritisak koriste fleksibilnu dija-fragmu za detektovanje promene. Ovaj tip senzora je našao veliku primenu u automobilskoj industriji (pri-tisak u gumama, ulju i gorivu, merenje protoka vazduha i u vazdušnim jastucima), u biomedicini (za merenje krvnog pritiska, kod endoskopa, infuzionih pumpi i mik-rosenzori koji se ugrađuju unutar lobanje). Barome-tarski mikrosenzori za pritisak se koriste za praćenje vremenskih prilika, kao i u tunelima gde ima vetra. Milioni MEMS senzora za pritisak se ugrađuju u putnu infrastrukturu za dobijanje i skupljanje podataka o uslo-vima na putevima. Primenjuju se u raznim oblastima, od sporta preko igrica široke potrošnje (Playstation) do vazdušnog prostora i svemira.

U inercijalne MEMS senzore spadaju merači ubrza-nja i žiroskopi pomoću kojih se mere promene u ubr-zanju, vibraciji, orijentaciji i nagibu. Najjednostavniji MEMS senzor merač ubrzanja je inercijalna masa koja je zakačena za opruge. Pri ubrzanju, dolazi do otklona mase od polazne pozicije što se prevodi u električni signal. MEMS žiroskopi imaju oblik diska ili točka koji rotira. Kod pojedinih MEMS žiroskopa koriste se vibri-rajuće strukture umesto tradicionalnih rotirajućih dis-kova. U automobilu se inercijalni senzori koriste kod vazdušnih jastuka, za navigaciju, podešavanje i pozicio-niranje visine prednjih farova, sistem protiv krađe, za izbegavanje sudara i detektovanje proklizavanja. Tako-đe se primenuju kod merenja i detektovanja vibracija, nagiba, pokreta i udara, a koriste se kod uređaja za zumiranje u kompjuterima, kao stabilizatori slike u tele-fonima i kamerama i za igrice.

Hemijski senzori

Minijaturni senzori dobijeni mikrofabrikacionim procesima mogu da se koriste za detektovanje gasova, u medicinskim, biomedicinskim i biohemijskim anali-zama, kod kontrole kvaliteta i procesa. Razvijeni su izuzetno osetljivi postupci mehaničke konverzije bio-(hemijskih) ili fizičkih procesa u signale koji mogu da se zabeleže, korišćenjem mikrofabrikovanih senzora.

Senzori su konzole (štapići pravougaonog oblika debljine ne veće od 1 μm) koje su najčešće napravljene od silicijuma [57]. Adsorpcija molekula na površini ovih mikromehaničkih konzola (funkcionalizovanih recep-torskim molekulima) izaziva savijanje konzole usled površinskog naprezanja. Konzolni senzori mogu da se koriste u sredinama kao što su vakuum, vazduh ili razne tečnosti. Glavne prednosti ovih minimizovanih senzora


634

su njihova mala veličina, kratak vremenski odziv, velika osetljivost i direktno prevođenje pobude u odziv.

Mikrofabrikovane konzole se uglavnom koriste kao senzori sile za snimanje topografije površine korišće-njem tehnika kao što su SFM (skenirajuća mikroskopija sila) [58] i AFM [59]. Kod ovih metoda se koristi konzola sa oštrim šiljkom koji skenira površinu uzorka (piezo-električni skener). Šiljak na konzoli može da bude u direktnom kontaktu sa površinom (kontaktni) ili da osciluje i bude u interakciji sa površinom u jako kratkim intervalima tokom oscilatornog ciklusa (dinamički, nekontaktni). Savijanje konzole se meri optičkim detek-tovanjem pozicije skrenutog laserskog zraka na vrhu konzole ili merenjima piezootpornih deformacija. Inter-akcija vrha konzole sa površinom je uobičajena kod svih SFM tehnika.

AFM se razvila u najjači i najsvestraniji instrument za karakterisanje površine na molekulskom i atomskom nivou. Metode su dobro razvijene i intenzivno se pri-menjuju.

Korišćenjem konzola u nizu koje su prevučene slo-jem osetljivim na specifične molekule dobijaju se ultra-osetljivi nanomehanički senzori za detektovanje hemij-skih ili biohemijskih reakcija u gasnoj fazi i tečnom okruženju [57]. Prevučeni sloj može da bude potpuno ili delimično osetljiv za prepoznavanje pojedinih mole-kula. Ukoliko je svaka od konzola u nizu prevučena različitim (delimično specifičnim) senzorskim slojem, može da se dobije odziv za različite analite [60].

Kod senzora prikazanog na slici 4 gornja površina konzola je prevučena slojem titana (debljine 2 nm) i slojem zlata (debljine 20 nm) da bi se obezbedila refleksija površine i omogućila adsorpcija funkcionalnih grupa molekula iz probnih uzoraka. Uticaj tankih metal-nih slojeva na naprezanje konzole je zanemarljiv pri konstantnoj temperaturi. U gasnoj sredini, ova konfigu-racija može da se koristi kao „veštački nos” za prepo-znavanje i karakterisanje isparljivih mirisa i para [60]. U tečnoj sredini, konzolni senzori omogućavaju brzo, kvantitativno i kvalitativno detektovanje neobeleženih biomolekula (na primer specifičnih delova DNK, kod molekulskog prepoznavanja antitelo–antigen i protein– –protein) [61].

Slika 4. SEM slika senzora na bazi silicijuma sa osam konzola u nizu (dimenzija konzole 500/100/0,5 μm dužina/širina/debljina). Figure 4. Scanning electron micrograph of a cantilever sensor array (console dimensions 500/100/0.5 μm l/w/t).

Konzolni senzori mogu da se koriste u statičkom, dinamičkom i toplotnom režimu [61,62]. Na slici 5 prikazan je konzolni senzor u statičkom, dinamičkom i toplotnom režimu.

U statičkom režimu se na gornjoj površini konzole nalazi sloj osetljiv na adsorpciju. Površinsko naprezanje tokom procesa adsorpcije utiče na statičko savijanje konzole. U dinamičkom režimu konzola osciluje spolja na svojoj rezonantnoj frekvenci (koriščenjem piezoelek-tričnog pobuđivača). Konzola može da bude prevučena sa gornje ili donje strane molekulskim slojem osetljivim na adsorpciju. Kada se na konzoli adsorbuje masa, rezonantna frekvencija se pomera ka nižim vrednos-tima. Iz ovog pomeraja u frekvenciji moguće je izraču-nati masu koja se adsorbovala na konzoli [61]. Meha-nička svojstva konzole se ne menjaju značajno sa adsor-bovanjem mase. Promena rezonantne frekvence od 1 Hz, grubo odgovara promeni mase od 1 pg [60].

U toplotnom (bimetalnom) režimu, razlika u koefici-jentima toplotnog širenja materijala od kojih je na-pravljena konzola (uobičajeni materijal je kristalni Si koji je prevučen metalnim slojem debljine 100 nm na jednoj od površina) izaziva savijanje konzolnog senzora, ako se temperatura promeni. Promena temperature od 10–5 K izaziva skretanje konzole od nekoliko nm. Piezo-otporne mikrokonzole u nizu mogu da se koriste za po-buđivanje i za čuvanje podataka.

Mogućnosti primene konzolnih senzora su velike, a osnovni pravci razvoja su usmereni ka:

1. detektovanju para i isparljivih komponenata po-moću pojedinačnih konzola koje se koriste u statičkom režimu [61,63],

2. primeni za detektovanje gasne faze kod rastva-rača korišćenjem piezootpornih konzola [60],

3. biohemijskoj primeni u statičkom režimu [64] (senzor za glukozu) i

4. detektovanju proteina pomoću piezootpornih konzola koje rade u dinamičkom režimu [61,65].

Razvijeni su mikrokonzolni senzori kao detektori za eksplozive [66] i za bakteriju Escherichia coli [67]. Kada se mikrokonzola prevuče slojem hidrogela dobija se pH mikrosenzor [64]. Teorija adsorpcije molekula na mik-rokonzole [68], kao i elektrohemijske redoks reakcije [69] su opisane u literaturi.

MEMS senzori mogu da se koriste u kombinaciji sa drugim senzorima. Na primer MEMS može da se dizaj-nira sa senzorima tako, da meri protok tečnog uzorka, a u isto vreme da detektuje i najmanju količinu zagađu-jućih materija, ako ih u uzorku ima.

Razvoj mikrokonzolnih senzora će ići u pravcu teh-noloških primena, odnosno novih načina karakterisanja realnih materijala (primer su klinički uzorci krvi). Razvoj sredstava za medicinsku dijagnostiku će zahtevati pove-ćanje osetljivosti određenog broja genetskih testova koji se obavljaju sa malim količinama tečnih uzoraka


635

(krvi ili telesnih tečnosti). Sa naučne strane, izazovi leže u optimizaciji [57,62,65] konzolnih senzora (sa ciljem da se poboljša osetljivost do krajnjih granica – a to je nanomehanička detekcija pojedinačnih molekula [70]). Značajno je to da su mnogi od ovih mikrosenzora poka-zali performanse koja prevazilaze performanse svojih makroduplikata.

Mikropokretači

Mikropokretači predstavljaju jednu od osnovnih komponenti MEMS uređaja. Oni mogu da se koriste za mehaničko pokretanje delova uređaja, za merenje određenih fizičko–hemijskih svojstva iz čovekove okoli-ne ili čak da obezbede kretanje robota [71]. U osnovi, mikropojačivači proizvode mehanički otklon kao odziv na pobudu nastalu iz energetskog izvora. Uobičajeni oblici pobude su elektrostatička, toplotna i magnetna. Tehnike koje se koriste za detektovanje mehaničkog otklona zasnivaju se na kapacitivnim, piezootpornim i optičkim efektima [72].

Na slici 6 prikazane su uobičajene tehnike za detek-tovanje skretanja (otklona) konzole: a) kapacitivna, b) piezootporna i c) optička.

Slika 6. Uobičajene tehnike za detektovanje skretanja (otklo-na) konzole: a) kapacitivna, b) piezootporna i c) optička. Figure 6. Common techniques of detecting beam deflection: a) capacitive, b) piezoresistive and c) optical.

U slučaju elektrostatičke pobude, tehnike kojima se detektuju otkloni zasnovani su na kapacitivnosti. Detek-tovanje na bazi piezootpornosti zasniva se na koriš-ćenju integrisanih materijala čiji se otpor menja pod dejstvom naprezanja. Postoji nekoliko načina upotrebe izvora svetlosti sa odgovarajućim detektorom koji ko-riste za registrovanje savijanja mikropojačivača kod optičkog detektovanja otklona.

Jedna od tehnika za primenu senzora zasniva se na upotrebi nekoliko elektrostatičkih pokretača koji se preklapaju. Pokretači se još nazivaju elektrostatički češ-ljevi [73] u skladu s karakterističnim oblikom. Na slici 7 prikazan je elektrostatički češalj-pokretač. Veće prekla-panje između stacionarnih i pokretnih naelektrisanih konzola daje veće otklone i jači povratni signal. Elek-trostatički češalj pokretači su postali popularan izbor kod savremene primene senzora i pokretača zbog nji-hove sposobnosti kontrole otklona preko povratne kapacitivnosti i linearno pojačanih signala jednostavnim dodavanjem više prstasto isprepletenih konzola.

Glavne prednosti elektrostatičke pobude su mali utrošak energije, velike sile i gustina energije, koji mogu da se dobiju smanjenjem rastojanja između naelektri-sanih ploča. Osnovni princip pobude pod dejstvom toplote je da se oblik i zapremina materijala menjaju pri grejanju ili hlađenju. Veličina promene dimenzija zavisi od koeficijenta toplotnog širenja materijala. Tehnika se zasniva na korišćenju učvršćenih konzola koje mogu da se savijaju pri grejanju. U ovoj geometriji, savijanje može da se ostvari kombinovanjem slojeva dva ili više materijala sa različitim koeficijentima toplotnog širenja. Pri grejanju se svaki sloj širi drugačije, generišući line-arnu razliku u naprezanju, a kao rezultat se dobija pobuda u strukturi. U početnim istraživanjima predlo-ženi su toplotni pokretači sa dve pokretne konzole (koje su napravljene od istog materijala, ali su različite deb-ljine). Pokretač se savija pri grejanju, jer se tanji deo konzole širi više nego deblji deo konzole [74]. Koncept pomeranja u jednoj ravni (zasnovan na savitljivim top-lotnim pokretačima sa dve uniformne konzole) kasnije je razvijen u smeru dobijanja toplotnih pokretača koji se kreću u dva pravca i to van jedne ravni.

Toplotni bimorfni pokretači

Alternativni pristup korišćenju tehnike ,,dve kon-zole” je slaganje slojeva u bimorfnim strukturama [75].

Slika 5. Radni režimi za konzolni senzor: A) režim statičkog savijanja, B) režim dinamičkog rezonovanja i C) toplotno bimetalni režim.Figure 5. Cantilever sensor operating modes: A) static deflection mode, B) dynamic resonance mode and C) bimetallic heat mode.


636

Toplotna pobuda može da se ostvari provođenjem električne struje kroz integrisani (umetnuti) otpornik u obliku trake ili grejanjem preko spoljnog izvora (optičko grejanje usmerenim laserom ili pomoću ploče koja se kontrolisano greje).

Na slici 8 prikazan je toplotni bimorfni pokretač (TBP) sa integrisanom toplotnom trakom. Ovakve struk-ture generišu velike otklone pri grejanju usled razlike vrednosti koeficijenata toplotnog širenja materijala od kojih su napravljene. Tehnike za registrovanje otklona kod konzole uključuju integrisane piezootporne ele-mente, kao i optičke metode detektovanja. Strukture se inicijalno uvijaju na gore u odnosu na supstrat, što je posledica naprezanja koje preostaje između bimorfnih slojeva (slika 8b).

TBP su pogodni za primenu u MEMS/MOEMS uređajima kod kojih su potrebni veliki otkloni van ravni (nekoliko stotina mikrometara) i mala potrošnja energije (nekoliko milivati). Jednu od prednosti predstavlja činjenica da toplotni bimorfi daju skoro linearan odziv na primenjenu silu. TBP koji su zasnovani na ,,hodajućim” mikronožicama počeli su da se razvijaju sredinom poslednje decenije prošlog veka sa ciljem da se primene kod nove generacije MEMS uređaja. Istraživanja su se vodila u pravcu dobijanja velikih otklona mikrooptičkih komponenti [76], u specifičnim AFM merenjima (na primer ćelija raka [77]). Oblast istraživanja toplotnih bimorfnih pojačivača je široka, a prednosti njihove primene se stalno potvrđuju (koriste se kod barkod skenera, za čuvanje podataka, kao mikroventili, u displej tehnologiji, u optici).

Konvencionalni bimorfni pojačivači su sastavljeni od sloja silicijuma i sloja odgovarajućeg metala [78]. Deponovanje slojeva se odigrava na uslovima visokih temperatura i visokog vakuuma. Ovakvi procesni para-metri zahtevaju duže vreme u procesu proizvodnje (pri-tisak treba da se spusti na zadovoljavajuće nisku vred-nost), pa troškovi proizvodnje rastu. Polimeri, s druge strane predstavljaju alternativu konvencionalnim struk-turnim materijalima, nudeći niz prednosti kao što su jednostavne metode proizvodnje, koje su brze i finan-sijski povoljne. Polimeri obezbeđuju izolaciju bilo kog metalnog elementa (koji služi za grejanje), pa su pri-menljiviji za primenu u vodenim medijumima, kao he-mijski ili biološki senzori.

Polimeri mogu da se koriste kao jedan ili dva sloja u bimorfnim strukturama [78]. Uobičajeno je da se kom-binuje jedan sloj polimera (na primer poliimida ili epo-ksidne smole) sa strukturnim slojem zlata [79,80] ili aluminijuma [78], iako je moguće koristiti i dva različita polimerna (na primer poliimidna) sloja [81]. Ključna stvar za dobijanje ovakvih struktura je specifičan izbor svojstava materijala (sa ciljem da se obezbedi dobar početni otklon van ravni s jedne i osetljivost s druge strane).

Piezoelektrični pokretači

Piezoelektrična pobuda se izaziva električno induko-vanim naprezanjem duž piezoelektričnog materijala (na primer kvarca). Složene piezoelektrične konzole obez-beđuju izuzetno dobru kontrolu samog otklona, pa se koriste kod preciznih tehnologija (kod AFM). Piezoelek-trični pokretači su razvijani da obezbede velike otklone

Slika 7. Elektrostatički češalj pokretač. Levo: princip funkcionisanja, desno: u aktivnom – radnom stanju. Figure 7. Left: operating principal of electrostatic comb drive actuators. Right: video image of electrostatic comb drive actuators in operation.

Slika 8. Toplotni bimorfni pokretač sa integrisanom toplotnom trakom. a) Šematski prikaz; b) mikrografija TBP u radu (SEM prikaz).Figure 8. Thermal bimorph actuator with integrated heater track. a) Schematic view; b) photograph of a TBA system (working mode).


637

koji mogu da se porede sa onima dobijenim pomoću toplotnih pojačivača ili memorijski oblikovanih legura [4,56].

IZAZOVI, PRAVCI RAZVOJA I ZAKLJUČCI

Sadašnji izazovi za MEMS

U razvoju MEMS minimizacija je najvažniji zadatak, jer se smanjuju troškovi (potrošnja materijala je manja i omogućena je šaržna proizvodnja). Smanjenjem mase i veličine omogućava se smeštanje MEMS na mesta gde tradicionalni sistemi zbog velikih dimenzija ne mogu da se smeste. Najbolji primer za modernu primenu i izazov za dalji razvoj MEMS su merači ubrzanja, koji su se devedesetih godina prošlog veka počeli uspešno koris-titi, a danas se razvijaju i unapređuju kao senzori u vaz-dušnim jastucima, u kvalitetnim kamerama, fotoapara-tima i sličnim uređajima za stabilizaciju slike, u posled-njim generacimama mobilnih telefona (smartphone). Nedavno su razvijeni MEMS mikropretvarači koji sadr-že: mikroventile za kontrolu protoka gasa ili tečnosti, optički prekidači ili ogledala koji preusmeravaju ili me-njaju snop svetla, nezavisno kontrolisano polje mikro-ogledala za displeje, mikrorezonatori za različite pri-mene, mikropumpe za ostvarivanje pozitivnog pritiska u fluidu, mikrozakrilca za korigovanje vazdušnih struja na krilu aviona, kao i mnogi drugi. MEMS tehnologije omogućavaju ugradnju smanjenih, „inteligentnih” kom-ponenti. Široka primena moguća je zbog niskih troškova proizvodnje. Ali procesi mikrofabrikacije u velikom broju slučajeva još uvek nisu konkurentni kada su u pitanju troškovi sa konvencionalnim metodama ma-sovne proizvodnje.

Naučni aspekti

U ovom preglednom radu prikazan je razvoj MEMS tehnologija i uređaja, od polazne ideje koja se javila devedesetih godina prošlog veka do primena u raznim oblastima koje se svakodnevno proširuju. Dat je pre-gled istorijski bitnih naučnih i tehnoloških dostignuća za razvoj MEMS, kao i pregled tehnika dobijanja, mate-rijala od kojih se prave i sastavnih delova MEMS ure-đaja sa osvrtom na mikrosenzore i mikropokretače.

S naučne tačke gledišta, postoji više izazova u tehnološkom smislu na koje bi se trebalo fokusirati u cilju poboljšanja performansi MEMS-a, a najvažniji su:

• optimizacija fizičko–hemijskih svojstava materi-jala koji se koriste kao gradivni elementi MEMS uređaja (na primer: uvođenjem novih materijala, hemijskim ili fizičkim modifikovanjem materijala ili njegove povr-šine). Problemi koje bi trebalo rešiti su slaba adhezija materijala za supstrat, pojava apsorpcije i/ili adsorpcije vode, kvašenje površine organskog materijala, odnosno svih efekata koji imaju veliki uticaj na mikrostruk-

turnom nivou (mogući izvor strujnih udara, kratkih spojeva),

• ispitivanje mogućnosti zamene klasičnog silikon-skog supstrata MEMS uređaja, novim materijalima (na primer polimerima),

• proširenje primene MEMS uređaja (mikrosen-zora i mikropokretača) na nivo ultraosetljivog detekto-vanja komponenti (na molekulskom nivou) i

• modifikacija i optimizacija postojećih tehnološ-kih postupaka dobijanja MEMS uređaja nove generacije

U sledećem periodu akcenat istraživanja grupe sa TMF [82–92], biće optimizacija svojstava materijala (različitih tipova polimera) od kojih su napravljeni MEMS uređaji, mikrosenzori i mikropokretači, a najveći doprinos u radu očekuje se u izgradnji fleksibilnih mik-rokonzola. Odgovarajućim izborom materijala za poli-merne bimorfne konzole optimizovaće se njegove per-formanse, sa ciljem da se proširi oblast primene nove generacije MEMS uređaja. Dobijanje super hidrofobnih slojeva kao novih materijala u cilju zaštite organskog supstrata od vlage (kako bi se sprečila adsorpcija/ap-sorpcija vode i neželjeni efekti kao što su kratki spojevi) i razvoja novih polimernih materijala modifikacijom unutar strukture polimera jedan je od važnijih pravaca istraživanja.

Predviđanja i razvoj MEMS u budućnosti

Pravci razvoja MEMS i bio-MEMS tehnologija u budućnosti će biti usmereni i oslonjeni na multidisci-plinarne naučne timove sastavljene od stručnjaka iz oblasti elektro, mašinskog i hemijskog inženjerstva, kao i naučnika iz oblasti fizike i materijala, kliničke medicine i biohemije, i na njihovu kreativnu razmenu znanja. Ovakvim pristupom bi se omogućio razvoj i primena MEMS i bio-MEMS sistema u pravcu iznalaženja novih metoda, primarno, iz oblasti koja sjedinjuje naučni inte-res i interes čoveka za zdrav i kvalitetan život, to je oblast medicinskog dijagnostifikovanja jer su MEMS sis-temi osetljivi, selektivni, brzi, finansijski povoljni i lako primenljivi. MEMS sistemi se lako mogu prilagoditi i za novi pristup u korišćenju lekova. U tradicionalnim oblastima poput elektronske industrije, integracija MEMS sa mikroprocesorima i postojećim tehnologi-jama omogućiće stvaranje nove generacije interaktiv-nih, kompaktnih, mobilnih uredjaja (kao što su na pri-mer inteligentni mobilni telefoni i interaktivne konzolne igre). Nove primene MEMS i nanotehnologija se oče-kuju već u narednim godinama, a razvoj će ići u prav-cima koji se mogu nazreti na osnovu današnjih, poz-natih i definisanih saznanja, ali će svakako razvoj biti uslovljen novim znanjima i budućim tehnologijama.

LITERATURA

[1] C.W. Pouton, Formulation of poorly water-soluble drugs for oral administration: physicochemical and physio-


638

logical issues and the lipid formulation classification system, Eur. J. Pharm. Sci. 29 (2006) 278–287.

[2] M. Kahlweit, Microemulsions, Annu. Rep. Prog. Chem., C 95 (1999) 89–115.

[3] J. Sjöblom, R. Lindbergh, S.E. Friberg, Microemulsions - phase equilibria characterization, structures, appli-cations and chemical reactions, Adv. Colloid Interfac. 65 (1996) 125–287.

[4] R. Strey, Microemulsion microstructure and interfacial curvature, Colloid Polym. Sci. 272 (1994) 1005–1019.

[5] C.J.H Porter, C.W. Pouton, J.F. Cuine, W.N. Charman, Enhancing intestinal drug solubilisation using lipid-based delivery systems, Adv. Drug Delivery Rev. 60 (2008) 673–691.

[6] J.A. Yáñez, S.W.J. Wang, I.W. Knemeyer, M.A. Wirth, K.B. Alton, Intestinal lymphatic transport for drug deli-very, Adv. Drug Delivery Rev. 63 (2011) 923–942.

[7] R. Holm, A. Müllertz, H. Mu, Bile salts and their impor-tance for drug absorption, Int. J. Pharm. (2013), doi: 10.1016/j.ijpharm.2013.04.003.

[8] K. Kawakami, Modification of physicochemical charac-teristics of active pharmaceutical ingredients and appli-cation of supersaturatable dosage forms for improving bioavailability of poorly absorbed drugs, Adv. Drug Deli-very Rev. 64 (2012) 480–495.

[9] Nacionalni registar lekova, NRL 2012. Agencija za lekove i medicinska sredstva Srbije, Beograd, 2012.

[10] European Pharmacopoeia, seventh ed., Council of Europe, Strasbourg, 2010.

[11] M. Meloun, S. Bordovská, L. Galla, The thermodynamic dissociation constants of four non-steroidal anti-inflam-matory drugs by the least-squares nonlinear regression of multiwavelength spectrophotometric pH-titration data, J. Pharmaceut. Biomed. 45 (2007) 552–564.

[12] H. Potthast, J.B. Dressman, H.E. Junginger, K.K. Midha, H. Oeser, V.P. Shah, H. Vogelpoel, D.M. Barends, Bio-waiver Monographs for Immediate Release Solid Oral Dosage Forms: Ibuprofen, J. Pharm. Sci. 94 (2005) 2121– –2131.

[13] B.A. Khan, S. Bakhsh, H. Khan, T. Mahmood, A. Rasul, Basics of Self Micro Emulsifying Drug Delivery System, Journal of Pharmacy and Alternative Medicine, 2012 (dostupno na: www.iiste.org).

[14] H. Araya, S. Nagao, M. Tomita, M. Hayashi. The novel formulation design of self-emulsifying drug delivery systems (SEDDS) type O/W microemulsion I: enhancing effects on oral bioavailability of poorly water soluble compounds in rats and beagle dogs, Drug. Metab. Phar-macokinet. 20 (2005) 244–256.

[15] B.B. Subudhi, S. Mandal, Self-Microemulsifying Drug Delivery System: Formulation and Study Intestinal Per-meability of Ibuprofen in Rats, J. Pharm. (2013), Article ID 328769 (http://dx.doi.org/10.1155/2013/328769).

[16] The United States Pharmacopoeia, 30/National Formul-ary 25, Pharmacopoeia Convention Inc., Rockville, MD, 2007.

[17] M.J. Madou, Fundamentals of Microfabrication: The Science of Miniaturization, 2nd ed., CRC Press, Boca Raton, FL, 2002.

[18] R.Ghodssi, L. Pinyen: MEMS Materials and Processes Handbook, Springer, New York, 2011.

[19] F. Chollet, H. Liu, A Short Introduction to MEMS, CC Publ. Franche-Comte, France, 2011.

[20] G. Jia, M.J. Madou, MEMS Fabrication, in M. Gad-el-Hak (Eds.), The MEMS Handbook, 2nd ed., Vol. 2: MEMS Design and Fabrication, CRC Press/Taylor and Francis, Boca Raton, FL, 2006.

[21] M.R. Douglass, DMD Reliability: A MEMS Success Story, Proceedings of the Reliability, Testing and Character-ization of MEMS/MOEMS II, Bellingham, WA, 2003.

[22] S.A. Campbell, Fabrication Engineering at the Micro- and Nanoscale, Ch. 12, Oxford University Press, New York, 2008.

[23] J.-C. Eloy, MEMS market outlook, Yole Development, 2011.

[24] J.-C. Eloy, MEMS market outlook, Yole Development, 2012.

[25] N. Maluf , An introduction to micro-electro-mechanical systems engineering, Artech House, Boston, 2000.

[26] K.A. Reinhardt, W. Kern (Eds.), Handbook of Silicon Wafer Cleaning Technology, 2nd ed., William Andrew, Norwich, NY, 2008.

[27] M. Gad-el-Hak (Ed.), MEMS: Design and Fabrication, Chapter 3, CRC Press, Boca Raton, FL, 2006.

[28] H. Geng (Ed.), Semiconductor Manufacturing Handbook, Chapter 11, McGraw-Hill, New York, NY, 2005.

[29] T. Ohmi (Ed.): Scientific Wet Process Technology for Innovative LSI/FPD Manufacturing, CRC Press, Boca Raton, FL, 2006.

[30] A. Goyal, V. Hood, S. Tadigadapa, High speed anisotropic etching of Pyrex for microsystems applications, J. Non-Cryst. Solids 352 (2006) 657–663.

[31] K.R. Williams, K. Gupta, M. Wasilik: Etch rates for mic-romachining processing – Part II, J. Micromech. Syst. 12 (2003) 761–778.

[32] H. Liu, F. Chollet, Layout Controlled One-Step Dry Etch and Release of MEMS Using Deep RIE on SOI Wafer, IEEE/ASME J. MEMS 15 (2006) 541–547.

[33] K. Breitschwerdt, V. Becker, F. Laermer, A. Schilp, Device and Method for Etching a Substrate by using an Induc-tively Coupled Plasma Patent 6,709,546 and 7,094,706, Office, U.S.P., US (2004).

[34] K.O. Abrokwah, P.R. Chidambaram, D.S. Boning, Pattern based prediction for plasma etch, IEEE Trans. Semicond. Manuf. 20 (2007) 7–86.

[35] J. Goettert, The LIGA Process, A Fabrication Process for High-Aspect-Ratio Microstructures in Polymers, Metals, and Ceramics, In S. Soper, W. Wang (Eds.), BioMEMS: Technologies and Applications,CRC Press, Boca Raton, FL, 2006, pp. 46–94.

[36] Y. Desta, J. Goettert, X-ray Mask for LIGA Microfab-rication, In V. Saile, U. Wallrabe, O. Tabata, J.G. Korvink (Eds.), LIGA and Its Applications, Advanced Micro & Nanosystems, Vol. 7, Wiley-VCH, New York, 2009.

[37] U. Gengenbach, I. Sieber, U. Wallrabe, Design for LIGA and Safe Manufacturing, in: O. Brand, G. Fedder, C. Hierold, J. Korvink, O. Tabata (Eds.), Advanced Micro &


639

Nanosystems, Vol. 7, LIGA and Its Applications, Wiley- -VCH, Weinheim, 2009.

[38] S.B. Patil, T. Adrega, V. Chu, J.P. Conde, Thin film silicon MEMS microresonators fabricated by hot-wire chemical vapor deposition, J. Micromech. Microeng. 16 (2006) 2730–2735.

[39] A. Georgiev, E. Spassova, J. Assa, G. Danev, Preparation of Polyimide Thin Films by Vapour Deposition and Solid State Reactions, Polymer Thin Films, Abbass A Hashim ed., 2010.

[40] A. Georgiev, D. Dimov, E. Spassova, J. Assa, P. Dineff, G. Danev, Chemical and Physical Properties of Polyimides: Biomedical and Engineering Applications, in M.Abadie (Ed.): High Performance Polymers - Polyimides Based – From Chemistry to Applications, In Tech ed., 2012.

[41] A. Pashan, Sensor Applications of Polyimides, in: M. Abadie (Ed.), High Performance Polymers - Polyimides Based – From Chemistry to Applications, In Tech ed., 2012.

[42] A.A.S. Bhagat, P. Jothimuthu, I. Papautsky, Photodefin-able polydimethylsiloxane (PDMS) for rapid lab-on-a- -chip prototyping, Lab Chip 7 (2007) 1192–1197.

[43] H.L. Cong, T.R. Pan, Photopatternable conductive PDMS materials for microfabrication, Adv. Funct. Mater. 18 (2008) 1912–1921.

[44] D. Bachmann, B. Schoberle, S. Kuhne, Fabrication and characterization of folded SU-8 suspensions for MEMS applications, Sens. Act. A Phys. 130 (2006) 379–386.

[45] S. Jiguet, M. Judelewicz, S. Mischler, SU-8 nanocom-posite coatings with improved tribological performance for MEMS, Surf. Coatings Technol. 201 (2006) 2289– –2295.

[46] Y.L. Wang, J.H. Pai, H.H. Lai, Surface graft polymerization of SU-8 for bio-MEMS applications, J. Micromech. Mic-roeng. 17 (2007) 1371–1380.

[47] P. Abgrall, V. Conedera, H. Camon, SU-8 as a structural material for labs-on-chips and microelectromechanical systems, Electrophoresis 28 (2007) 4539–4551.

[48] A. Mata, A.J. Fleischman, S. Roy, Fabrication of multi-layer SU-8 microstructures, J. Micromech. Microeng. 16 (2006) 276–284.

[49] S.L. Tao, K.C. Popat, J.J. Norman, Surface modification of SU-8 for enhanced biofunctionality and nonfouling pro-perties, Langmuir 24 (2008) 2631–2636.

[50] M. Joshi, N. Kale, R. Lal, A novel dry method for surface modification of SU-8 for immobilization of biomolecules in bio-MEMS, Biosens. Bioelectron. 22 (2007) 2429– –2435.

[51] D. Sameoto, S.H. Tsang, I.G. Foulds, Control of the out-of-plane curvature in SU-8 compliant microstructures by exposure dose and baking times, J. Micromech. Mic-roeng. 17 (2007) 1093–1098.

[52] S.W. Youn, H. Goto, M. Takahashi, Thermal imprint process of parylene for MEMS applications, Key Eng. Mater. 340–341 (2007) 931–936.

[53] E. Meng, P.Y. Li, Y.C. Tai, A biocompatible parylene MEMS thermal flow sensing array, Sens. Act., A Phys. 144 (2008) 18–28.

[54] P.J. Chen, D.C. Rodger, R. Agrawal, Implantable mic-romechanical parylene-based pressure sensors for unpowered intraocular pressure sensing, J. Micromech. Microeng. 17 (2007) 1931–1938.

[55] B. Bilenberg, T. Nielsen, B. Clausen, PMMA to SU-8 bonding for polymer based lab-on-a-chip systems with integrated optics, J. Micromech. Microeng. 14 (2004) 814–818.

[56] K. Tsougeni, A. Tserepi, E. Gogolides, Photosensitive poly(dimethylsiloxane) materials for microfluidic appli-cations, Microelectron. Eng. 84 (2007) 1104–1108.

[57] S. Rajaraman, S.O. Choi, R.H. Shafer, Microfabrication technologies for a coupled three-dimensional micro-electrode, microfluidic array, J. Micromech.Microeng. 17 (2007) 163–171.

[58] J.H. Park, Y.K. Yoon, S.O. Choi, Tapered conical polymer microneedles fabricated using an integrated lens tech-nique for transdermal drug delivery, IEEE Trans. Biomed. Eng. 54 (2007) 903–913.

[59] R. Lo, E. Meng, Integrated and reusable in-plane micro-fluidic interconnects, Sens. Act., B Chem. 132 (2008) 531–539.

[60] P. Abgrall, A.M. Gue, Lab-on-chip technologies: making a microfluidic network and coupling it into a complete microsystem – a review, J. Micromech. Microeng. 17 (2007) 15–49.

[61] P.J. Chen, D.C. Rodger, E.M. Meng, Surface-micro-machined parylene dual valves for on-chip unpowered microflow regulation, J. Microelectromech. Syst. 16 (2007) 223–231.

[62] P.J. Chen, D.C. Rodger, M.S. Humayun, Floating-disk parylene microvalves for selfpressure- regulating flow controls, J. Microelectromech. Syst. 17 (2008) 1352– –1361.

[63] P.Y. Li, J. Shih, R. Lo, An electrochemical intraocular drug delivery device, Sens. Act., A Phys. 143 (2008) 41–48.

[64] P. Abgrall, A.M. Gue, Lab-on-chip technologies: making a microfluidic network and coupling it into a complete microsystem – a review, J. Micromech. Microeng. 17 (2007) 15–49.

[65] Y. Sverdlow, Y. Rosenberg, Y.I. Rozenberg, R. Zmood, R. Erlich, S. Natan, Y. Shacham- Diamand, The electrodepo-sition of cobalt-nickel-iron high aspect ratio thick film structures for magnetic MEMS applications, Micro-electron. Eng. 76 (2004) 258–265.

[66] A. Kaushik, H. Kahn, A.H. Heuer, Wafer-level mechanical characterization of silicon nitride MEMS, J. Microelec-tromech. Syst. 14 (2005) 359–367.

[67] J. Zhang, C. Carraro, R.T. Howe, R. Maboudian, Electrical, mechanical and metal contact properties of polycrys-talline 3C-SiC films for MEMS in harsh environments, Surf. Coatings Technol. 201 (2007) 8893–8898.

[68] G. Rehder, M.N.P. Carreño, Thermally actuated a-SiC:H MEMS fabricated by a PECVD process, J. Non-Crystalline Solids 352 (2006) 1822–1828.

[69] J.C. Doll, S.-J. Park, B.L. Pruitt, Design optimization of piezoresistive cantilevers for force sensing in air and water, J.Appl. Phys. 106 (2009) 064310.


640

[70] J.N. Palasagaram, R. Ramadoss, MEMS-capacitive pres-sure sensor fabricated using printed circuit-processing techniques, IEEE Sens. J. 6 (2006) 1374–1375.

[71] S. Tadigadapa, K. Mateti, Piezoelectric MEMS sensors: state of art and perspectives, Meas. Sci. Technol. 20 (2009) 092001.

[72] P. Muralt, R.G. Polcawich, S. Trolier-McKinstry, Piezo-electric thin films for sensors, actuators, and energy harvesting, MRS Bull. 34 (2009) 658–664.

[73] A. Boisen, S. Dohn, S.S. Keller, S. Schmid, M. Tenje, Can-tilever like micromechanical sensors, Rep. Prog. Phys. 74 (2011) 036101.

[74] G. Julius Vancso, H. Schoenherr, Scanning Force Micro-scopy of polymers, Springer, Berlin, 2010.

[75] P.Eaton, P.West, Atomic Force Microscopy, Oxford University Press Inc., New York, 2010.

[76] P.Xu, X. Li, H. Yu, M. Liv, J.Li, Self-assembley and sensing group of pre-modified CNTs of resonant micro-canti-levers for specific detection of volatile organic com-pound vapors, J. Micromech. Microeng. 20 (2011) 115003.

[77] H. Pan, Y. Xu, S. Wu, B. Zhang, J. Tang, Molecular interactions in self-assembly monolayers on gold-coated microcantilever electrodes, Nanotechnology 22 (2011) 225503.

[78] X. Li, D.W. Lee, Integrated microcantilevers for high resolution sensing and probing, Meas. Sci. Technol 23 (2012) 022001.

[79] N. Nelson-Fitzpatrick, S. Evoy, Highly compliant static micro cantilevers fabricated in gold nanocomposite materials, J. Micromech. Microeng. 21 (2011) 115021.

[80] O. Korostynska, K. Arshak, A. Arshak, E. Gill, P. Creedon, S. Fitzpatrick, Polymer-based microsensors arrays for pH and glucose monitoring, Key Eng. Mater. 437 (2010) 354–358.

[81] M.Godin, V. Tabard-Cossa, Y. Miyahara, T. Monga, P.J. Williams, L.Y. Beaulieu, R. Bruce Lennox, P.Gruffer, Can-tilever-based sensing: the origin of surface stress and optimization strategies, Nanotechnology 21 (2010) 075501.

[82] Y. Yang, X.Xia, X. Gang, P,Xu, H. Yu, X. Li, Self-assembling siloxane bilayer on SiO2 surface of microcantilevers for long term highly repeatible sensing to trace explosives, J. Micromech. Microeng. 20 (2010) 050101.

[83] K.Y. Gfeller, N. Nugaeva, M. Hegner, Micromechanical oscillators as rapid biosensor for the detection of active growth of Escherichia coli, Biosens. Bioelectron. 21 (2005) 528–533.

[84] M.Z. Ansari, C.Cho, Thermal characteristics of micro-cantilever biosensors, in: A. Fred, J. Felipe, H. Gamboa (Eds.), Biomedical Engineering Systems and Techno-logies, Springer, Berlin, 2011.

[85] F. Tian, J.H. Pei, D.L. Hedden, G.M. Brown, T. Thundat, Observation of the surface stress induced in microcan-tilevers by electrochemical redox processes, Ultramic-roscopy 100 (2004) 217.

[86] Y. Yang, X. Xia, X.Gang, P.Xu, H.Yu, X. Li, Nano-thick resonant cantilever with a novel specific reation-induced frequency-increase effect for ultra-sensitive

chemical detection, J. Micromech. Microeng. 20 (2010) 055022.

[87] K. Oldham, J. Pulskamp, R. Polcawich, Thin-film piezo-electric actuators for bio-inspired micro-robotic appli-cations, Integr. Ferroelectr. 95 (2007) 54–65.

[88] D. Yan, A. Khajepour, R. Mansour, Design and modelling of a MEMS bidirectional vertical thermal actuator, J. Micromech. Microeng. 14 (2004) 841–850.

[89] V. Mukundan, P. Ponce, H.E. Butterfield, B.L. Pruitt, Modeling and characterization of electrostatic comb-drive actuators in conducting liquid media, J. Micromech. Microeng. 19 (2009) 1–9.

[90] R. Hickey, D. Sameoto, T. Hubbard, M. Kujath, Time and frequency response of two-arm micromachined thermal actuators, J. Micromech. Microeng. 13 (2003) 40-46.

[91] F. Carta, Y.J. Hsu, J. Sarik, I. Kymissis, Bimorph actuator with monolithically integrated CMOS OFET control, Org. Electr. 14 (2013) 286–290.

[92] G. Lammel, S. Schweiser,P. Renaud, Optical Microscan-ners and Microspectrometers using Thermal Bimorph Actuators, Kluwer Acad. Publ., Boston, MA, 2002.

[93] J. Henriksson, M.R. Gullo, J. Brugger, Integrated long-range thermal bimorph actuators for parallelizable Bio-AFM applications, Sens. J. 13 (2013) 2849–2856.

[94] S. Mouaziz, G. Boero, R.S. Popovic, Polymer-based canti-levers with integrated electrodes, J. Microelectromech. Syst. 15 (2006) 890–895.

[95] R.H. Ibbotson, R.J. Dunn, V.A. Djakov, P.Ko. Ferrigno, S.E. Huq, Polyimide microcantilever surface stress sensor using low-cost, rapidly-interchangable, spring-loaded microprobe connections, Microelectron. Eng. 85 (2008) 1314–1317.

[96] M.C. Mieux, M.E. McConney, Y.H. Lin, S. Singamaneni, H. Jiang, T.J. Bunning, W. Tsukruk, Polymeric nanolayers as actuators for ultrasensitive thermal bimorphs, Nano. Lett. 6 (2006) 730–734.

[97] J.W.L. Zhou, H.Y. Chan, Tony K.H To, Wen J. Li, Polymer MEMS Actuators for Underwater Micromanipulation, IEEE-ASME Transactions Mechatronics 9 (2004) 334–342.

[98] Lj.V. Rajaković, A New Metodology for Testing and Cha-racterization of Sorption Materials in a gas Flow System Based on Piezoelectric Sensors, Ch. 12, pp. 139–149; N. Akmal, A. Usmani (Eds.), Polymers in Sensors – Theory and Practice, ACS Symposium Series 690, ACS Books, Washington DC, 1998.

[99] I.G. Popović, L. Katsikas, S. Čurović, B. Ćosić, Lj.S. Čero-vić, The thermal degradation of poly(methyl methac-rylate)/silicon carbide nanocomposites, Hem. Ind. 56 (2002) 478–482.

[100] A.Onjia, Lj.V.Rajaković, The Potential of Piezoelektric Sensor for Characterization of Activated Carbon Cloth Applied in Adsorption of Phenols from Air, Ch. 15, pp. 168–173; N. Akmal, A. Usmani (Eds.), Polymers in Sen-sors – Theory and Practice, ACS Symposium Series 690, ACS Books, Washington DC, 1998.

[101] J.M. Filipović, L. Katsikas, I.G. Popović, S. J. Veličković, T.A. Djakov, D.M. Petrović-Djakov, The Thermal Degrad-ation of Some Alkali Metal Salts of Poly(Itaconic acid), J. Therm. Anal. 49 (1997) 335–341.


641

[102] Lj.V. Rajaković, S.B. Štrbac, Surface Morfology and the Response of Piezoelectric Gas Sensor, Anal. Chim. Acta 315 (1995) 83–91.

[103] T.A. Djakov, I.G. Popović, The Kinetics of the Radical Polymerisation of Di-2-Chloroethyl Itaconate, Hem. Ind. 53 (1999) 361.

[104] Lj.V. Rajaković, M. Bastić, S.A.Tunikova, N.V. Bel'skih, Ya.I. Korenman, Potential of the Application of Piezo-electric Sensor for the Phenol Determination in the Air, Anal. Chim. Acta 318 (1995) 77–87.

[105] S.J. Veličković, M.T. Kalagasidis-Krušić, R.V. Pjanović, N.M. Bošković-Vragolović, I.G. Popović, The diffusion of water in poly(ditetrahydrofurfuryl itaconate), Polymer 46 (2005) 7982–7988.

[106] S. Simić, B.M. Dunjić, S.V. Tasić, B.R. Božić, D.M. Jovanović, I.G. Popović, Synthesis and characterization of interpenetrating polymer networks with hyper-branched polymers through thermal-UV dual curing, Progr. Org. Coat. 63 (2008) 43–48.

[107] L. Katsikas, N. Nišević, M. Ignjatović, V. Adamović, T. Djakov, I. Popović, Radikalna polimerizacija monoetil itakonata, Hem. Ind. 57 (2003) 553–558.

[108] S. Štrbac, M. Nenadović, Lj.V. Rajaković, Z. Rakočević, Chemical surface composition of the polyethylene implanted by Ag+ ions studied by phase imaging atomic force microscopy, Appl. Surf. Sci. 256 (2010) 3895–3899.

SUMMARY MICRO-ELECTRO-MECHANICAL SYSTEMS (MEMS) – TECHNOLOGY FOR THE 21ST CENTURY Tatjana A. Djakov, Ivanka G. Popović, Ljubinka V. Rajaković

Faculty of Technology and Metallurgy, University of Belgrade, Belgrade, Serbia

(Review paper)

Micro-electro-mechanical systems (MEMS) are miniaturized devices that can sense the environment, process and analyze information, and respond with avariety of mechanical and electrical actuators. MEMS consist of mechanical elements, sensors, actuators, electrical and electronics devices on a common sili-con substrate. Micro-electro-mechanical systems are becoming a vital technologyfor modern society. Some of the advantages of MEMS devices are: very small size,very low power consumption, low cost, easy to integrate into systems or modify,small thermal constant, high resistance to vibration, shock and radiation, batchfabricated in large arrays, improved thermal expansion tolerance. MEMS technol-ogy is increasingly penetrating into our lives and improving quality of life, similarto what we experienced in the microelectronics revolution. Commercial oppor-tunities for MEMS are rapidly growing in broad application areas, including bio-medical, telecommunication, security, entertainment, aerospace, and more inboth the consumer and industrial sectors on a global scale. As a breakthroughtechnology, MEMS is building synergy between previously unrelated fields such asbiology and microelectronics. Many new MEMS and nanotechnology applications will emerge, expanding beyond that which is currently identified or known. MEMSare definitely technology for 21st century.

Keywords: Micro-electro-mechanical systems (MEMS) • Sensors • Actuators •Chip

643

Heavy metal content of soil in urban parks of Belgrade

Maja M. Kuzmanoski, Marija N. Todorović, Mira P. Aničić Urošević, Slavica F. Rajšić

Institute of Physics, University of Belgrade, Belgrade, Serbia

Abstract This study focuses on soil pollution in four urban parks of Belgrade. The sampling locationswithin each park were chosen based on proximity to streets characterized by heavy traffic,and soil samples were taken at different depths down to 50 cm. Concentrations of sixheavy metals (Cr, Cu, Fe, Mn, Ni and Zn) were measured using Energy Dispersive X-Ray Fluorescence (EDXRF) spectrometer. The following average abundance order of heavymetals was found: Fe >> Mn > Zn > Cr > Ni > Cu in topsoil samples. The highest enrichmentin topsoil was observed for Zn. Copper and Zn, metals mainly related to traffic emissions,exhibited the highest concentrations at the sampling location close to a bus and trolleybustermini. The highest Ni and Cr concentrations were observed in a park located in a citysuburb, where a large number of individual heating units is present. The largest decreasein concentrations with soil dept was observed for Zn and Cu, followed by Ni and Cr, in theparks with the highest concentrations of these elements in topsoil. Generally high topsoilCr and Ni concentrations were observed in comparison with average values reported inliterature for other world cities.

Keywords: heavy metals, soil, urban parks, EDXRF.

SCIENTIFIC PAPER

UDC 712.25(497.111):504.5

Hem. Ind. 68 (5) 643–651 (2014)

doi: 10.2298/HEMIND131105001K


Heavy metals are naturally present in soil and some of them (e.g., Cu, Cr, Mn and Zn) are essential elements for plants, animals and humans, as they are important for various physiological functions [1]. However, in elevated concentrations they can have harmful effects on the environment [2]. In general, soil acts as a sink for heavy metals released from various anthropogenic sources, including industrial and traffic emissions, fossil fuel combustion, sewage sludge, waste disposal, mining and agricultural activities [3]. Besides the inputs from local pollution sources, long-range transported pollutants can also contribute to soil contamination through atmospheric deposition [4]. Since heavy metals are not biodegradable, they tend to accumulate in the environment [5] and therefore soil presents an evi-dence of pollution over a longer period of time.

Contamination of soil with heavy metals is of con-cern due to their toxicity. They can move towards ground water and pollute water supplies, or accumul-ate in plants thus entering the food chain [6,7] and affecting human health. Humans can also be exposed to heavy metals in soil more directly through inhal-ation, ingestion and dermal contact absorption [8]. This is of particular concern in urban environment, as urban soil can easily be disturbed and re-suspended due to dense population and heavy traffic activities. Children can be more affected by soil contamination, particul-arly in urban parks, due to frequent hand-mouth acti- Correspondence: M.M. Kuzmanoski, Institute of Physics, University of Belgrade, Pregrevica 118, 11080 Belgrade, Serbia. E-mails: [email protected] Paper received: 5 November, 2013 Paper accepted: 20 December, 2013

vity and higher absorption rate in comparison to adults. The urban soil is subjected to significant level of pol-lution, mainly originating from traffic emissions (vehicle exhaust, tire and brake wear, weathered street surface particles), industrial (coal combustion, metallurgical and chemical industry, etc.), domestic emissions, weathering of building and pavement surface and waste incinerators [9].

The objective of this study is to estimate the heavy metal content of soil samples taken at different depths down to 50 cm from four parks in urban part of Bel-grade. Elemental content of topsoil in Belgrade urban area (including city parks), using another analytical method – atomic absorption spectroscopy (AAS), has been reported in earlier studies [10,11]. In a study by Crnković et al. [12] analysis of samples of different soils in Belgrade wider area, taken at 0–10 cm and 40–50 cm depths, was discussed. In the present study total ele-mental content of heavy metals in soil in Belgrade urban parks was determined using an Energy Dispersive X-Ray Fluorescence (EDXRF) spectrometer. One of the main advantages of EDXRF spectrometry is that it does not require sophisticated sample preparation and allows analysis of samples for total element concen-trations.

EXPERIMENTAL

Study area

Belgrade is located at 20°27'44'' Eastern longitude and 44°49'14'' Northern latitude, at the average alti-tude of about 120 m above sea level. It covers the area of over 3000 km2, and has a population of about two

M.M. KUZMANOSKI et al.: HEAVY METAL CONTENT OF SOIL IN URBAN PARKS OF BELGRADE Hem. ind. 68 (5) 643–651 (2014)

644

million. Major pollution sources in Belgrade are heating plants run with crude oil or natural gas, domestic heat-ing (using coal and crude oil as fuel), gasoline and diesel vehicle exhaust, as well as other vehicle emissions. Traffic has been recognized as the main source of air pollution in the central area of Belgrade [13]. It should be noted that there are many old buses and trucks on the streets of Belgrade, and a large number of pas-senger cars is over 10 years old.

Sampling

The soil samples were collected in May 2011, in four parks situated in different areas of urban part of Belgrade: Karađorđev Park (KP), Studentski Park (SP), Zemun Park (ZP) and Botanical Garden (BG). The loca-tions of the parks are shown on the map in Figure 1: three of them (KP, SP and BG) are in the city central area, while one (ZP) is in the center of one of the city suburbs − Zemun. Detailed descrip on of the area sur-rounding each park is given in Table 1. Within each park a location close to streets with high traffic density was selected for taking samples.

The analysis discussed here includes topsoil samples, as well as those taken from different depths down to 50 cm. Steel corer was used for collecting the samples

from different depths. In each sampling area several sub-samples were taken from a rectangular of approxi-mately 2 m×2 m, and mixed to obtain a composite sample. The samples were taken at depths of 0–10 cm, 10–20 cm, 20–30 cm, 30–40 cm and 40–50 cm. A total of 22 composite samples were made by mixing and homogenizing the collected sub-samples, one for each of these layers, for each park, and placed into poly-ethylene bags. In laboratory, the samples were dried at air temperature, and large rock and organic debris were removed before sieving. They were crushed and sieved using a nylon sieve to obtain fraction smaller than 2 mm, which was then ground to a fine powder. The soil samples, as well as the standard reference materials, have been prepared in pellet form: 5 g of material was pressed for 1 min at 30 t into a pellet of 3.5 cm in diameter, using wax backing.

Analysis

The measurements of heavy metal content in the soil samples were performed using PANalytical´s MiniPal 4 spectrometer and the analysis of the spectral data was performed using MiniPal/MiniMate software. The XRF spectrometer is equipped with 9W Rh tube and silicon drift detector, with resolution FWHM = 145

Figure 1. Sampling locations: Karađorđev Park (KP), Studentski Park (SP), Zemun Park (ZP), and Botanical Garden (BG).

Table 1. Description of the sampling areas

Studied park Area description Karađorđev Park (KP)

Located in a populated part of the city, near Clinical Center of Serbia, this park is surrounded by two-way traffic roads and is in close proximity (~200 m) of the section of the highway E-75 which passes through Belgrade. The tree coverage is approximately two thirds of the park area.

Studentski Park (SP)

It is located in the very city center, near university buildings and the main promenade. The park is smalland surrounded by one way streets, with the exception of the southwest side, where bus and trolleybusterminus is located. The tree coverage is much lower compared to the other three examined sites.

Zemun Park (ZP)

The park is located in a residential part of the city with low private houses, surrounded by two schools,hospital and sports center. In comparison with the other three locations, the intensity of traffic is lower, though one must have in mind local traffic and the closeness of public transport stations. Studied area islargely covered with trees.

Botanical Garden (BG)

It is located in a residential part of the city with multistory buildings, approximately 1 km from the port of Belgrade and small chemical industries. The northern side of the park is exposed to the impact oftraffic (cars and buses). The examined area is mostly covered with treetops.


645

eV for 5.9 keV 55Fe. We focused primarily on Cr, Cu, Fe, Mn, Ni and Zn, as common heavy metal pollutants in urban soil. The calibration of the instrument was per-formed using six certified reference materials (ERM-CC135a, IAEA-PTXRF, MINTEK-SARM 42, MINTEK-SARM 69, IAEA-SL 1 and NIM-GBW07406) pressed into pellets as described above. In addition, NIST (US Department of Commerce, National Institute of Standards and Tech-nology) CRM 2711 soil certified reference material was included in calibration for Fe, in order to extend the calibration range towards lower concentrations of this element. Optimal measurement parameters (tube vol-tage, current and filter between the tube and the sample) were determined [14] for different sets of ele-ments (Table 2). For each of the three sequences mea-surement time was 1800 s, and all measurements were carried out in air. Three repeated measurements were performed for each sample. The detection limits for the measured elements were calculated following the International Union of Pure and Applied Chemistry (IUPAC) method, as 3 times the standard deviation of blank signal divided by the sensitivity (change in inten-sity of the spectral line per unit concentration). For that purpose, a pellet made of Merck Hoechst wax C micro-powder (C38H76N2O2) was measured 10 times conse-cutively as a blank sample. It has low heavy metal content – the concentration of each element examined here is less than 5 mg kg–1. Generally, the wax micro-powder is used as binder in preparation of sample pel-lets for XRF analysis. The obtained values are the fol-lowing: 3.7, 2.5, 2.3, 1.5, 5.6 and 1.8 mg kg–1 for Cr, Cu, Fe, Mn, Ni and Zn, respectively.

Enrichment factors

The enrichment factor (EF) was calculated from the measured element concentrations. It represents the ratio of concentration of an element associated with anthropogenic pollution and a reference element in analyzed sample, relative to the corresponding ratio for the background concentrations. Usual reference elements are Al, Fe, Li and Mn [15-17]. In this work EF was calculated as follows, using Fe as a reference element:

= M Fe soil

M Fe background

( / )( / )

c cEFc c

(1)

Here, cM and cFe are concentrations of the test ele-ment and Fe, respectively. Background concentrations

used in these calculations are discussed in the next section.

We consider EF > 2 indicative of enrichment asso-ciated with anthropogenic pollution, while EF > 5 is assumed to imply significant enrichment [18].


In order to verify the accuracy and precision of the results of the sample analysis, we performed the measurements of elemental content of ERM (European Reference Materials) CC135a soil reference material, included in the calibration for all measured elements, as an unknown sample. In addition, we measured NIST CRM 2711 soil certified reference material, which was not included in the calibration (except for Fe, as explained earlier) and thus could represent a real unknown sample. In this reference material the cer-tified concentrations of Cu, Fe, and Mn are within the calibration concentration range (Table 3), and thus it can be used here to test the accuracy of the measure-ments for these elements. The average values of ele-ment concentrations in the two reference materials obtained from measurements conducted on 10 conse-cutive days were compared with the corresponding certified values, as shown in Table 3.

The results of measurements of ERM CC135a showed that the measurement precision was better than 1% for all elements of interest.

Elemental concentrations in park topsoil

The elemental concentrations measured in the topsoil samples taken from the four studied parks (KP, SP, ZP and BG) are given in Table 4. The shown values are the average concentrations and relative standard deviations calculated from the measurements of three replicate samples. The average abundance order of heavy metals in the samples was: Fe >> Mn > Zn > Cr > > Ni > Cu. Copper exhibited the lowest concentration at all depths, in all parks except for SP, as discussed later in this section. The lowest concentrations of all anal-yzed elements, except for Fe, were found in BG. This could be explained with high tree coverage of this park (Table 1) in comparison with the other three parks. Note that the concentration of Fe in SP is below the lower limit of the calibration range.

For comparison with our measurements, average heavy metal concentrations in urban soil, reported in earlier studies for Belgrade and for other world cities,

Table 2. Optimal measurement parameters of XRF spectrometer for different groups of measured elements in soil

Element Voltage, kV Current, μA Filter material and thickness Cr 15 540 Al (50 μm) Mn, Fe 20 405 Al (200 μm) Cu, Ni, Zn 30 270 Mo (100 μm)


646

are also given in Table 4. It is apparent that the con-centrations of Cu, Ni and Zn obtained in this study are lower than the average values previously reported for the Belgrade urban area [10,11]. They are however, within the reported ranges of concentrations of these elements [10,11].

Although Ni concentrations obtained in this study are close to the lower limit of the range of values reported for Belgrade urban soil [11], they are high in comparison with those reported for many other world cities, except for cities in China (Table 4). This is in agreement with the earlier studies of soil and air pol-

lution in Belgrade urban area [12,13], where it was noted that soil in Serbia is characterized by high Ni content due to its geological origin [26]. In addition, fossil fuel combustion is reported to be the major anthropogenic source of atmospheric Ni in Europe and the world [27,28]. The highest Ni concentration in our study was found in ZP and a possible explanation of this result is a larger number of individual heating units in this part of the city. To support this assumption, we refer to a recent study of air pollution in Belgrade urban area [29], where strong negative correlation between Ni concentration in airborne particulate mat-

Table 3. Calibration concentration ranges, certified concentrations (average value and relative uncertainty) of elements of interest in two soil reference materials − CC135a (ERM) and CRM 2711 (NIST), and corresponding measured concentrations (average value, relative standard deviation (RSD) and recovery range, N = 10)

Metal Cr Cu Fe Mn Ni Zn

Calibration concentration range,mg kg–1 100–455 17–390 28900– –67400

390–1450 28–291 44–345

ERM CC135a

Certified Average, mg kg–1 455 107 47500 390 291 345 Rel. uncertainty, % 13.0 4.7 9.7 10.3 7.6 14.2

Measured Average, mg kg–1 482 105 46994 411 293 354 RSD / % 0.2 0.7 0.1 0.5 0.9 0.6

Recovery range, % 105.6–106.3 96.8–98.9 98.8–99.1 104.8–106.1 99.2–102.3 101.8–103.3 NIST CRM 2711

Certified Average, mg kg–1 47a 114 28900 638 21 350 Rel. uncertainty, % – 1.8 2.1 4.4 5.3 1.4

Measured Average, mg kg–1 45 108 28122 554 28 364 RSD / % 0.4 1.0 0.1 0.2 5.6 0.4

Recovery range, % 94.3–96.3 94.6–97.8 97.2–97.4 86.1–86.6 196.1–232.4 102.9–104.1 aNoncertified value

Table 4. Heavy metal topsoil concentrations (average value in mg kg–1 and percentage of relative standard deviation of measurements of three replicate samples in parentheses (%)) obtained in this study for the four Belgrade urban parks (KP, SP, ZP and BG) and average topsoil concentrations reported in earlier studies for Belgrade and other world cities

Cite Cr Cu Fe Mn Ni Zn Reference Belgrade – this study

KP 100 (1.6) 27 (1.9) 35507 (0.3) 771 (0.1) 56 (1.2) 104 (0.7) – SP 99 (2.8) 86 (1.0) 24835 (0.1) 671 (0.4) 55 (1.1) 165 (0.6) – ZP 121 (2.7) 40 (0.4) 30324 (0.1) 691 (0.6) 81 (1.6) 121 (1.3) – BG 92 (5.4) 25 (1.3) 29224 (0.1) 618 (0.9) 54 (1.1) 95 (0.4) –

Belgrade – previous studies – – 46 – 418 – 174 [10] – 70 122 – 642 124 268 [11] – 33 29 – – 67 129 [12]

Other world cities Palermo, Italy 34 63 – 519 18 – [19] Naples, Italy 11 74 – – – 251 [20] Seville, Spain 39 68 20100 471 22 145 [21] Tallin, Estonia 70 – – 821 23 121 [22] Cities in China 78 115 – – 100 266 [9] Mexico City, Mexico 116 54 – – 39 219 [23] Jersey City, NJ, USA 52 250 – – – 236 [24] Annaba, Algeria 31 39 24270 355 – 68 [25]


647

ter and air temperature was observed (based on two-year observations) and explained with Ni originating from fossil fuel combustion in local heating units.

Chromium concentrations obtained in this work for the four parks exceed the previously reported average values for Belgrade urban area, and those in many other world cities. This element also shows higher con-centration in ZP, in comparison with the other three parks. Major sources of Cr are coal combustion and vehicle exhaust [30,31]. However, it should be noted that the calibration concentration range for Cr does not include values lower than 100 mg kg–1 (Table 3).

The concentration of Cu in the topsoil sample from SP shows the most notable departure from concen-trations in other samples. This result is in agreement with results of earlier studies of air pollution moni-toring in the Belgrade urban area using both the ins-trumental and biomonitoring techniques [13,32]. Both studies reported higher Cu concentration in the area of SP than in KP and BG. It is worth mentioning that the mutuality between air and soil pollution has been noted in other studies: through wet and dry deposition heavy metals are transferred from atmosphere to soil [33], while heavy metals in topsoil contribute to the corresponding concentrations in the atmosphere [34]. The sources of Cu emission are mainly related to traffic, such as brake abrasion and corrosion of metallic parts of cars [29,31]. In an air pollution study by Rajšić et al. [29], based on data collected for two years, significant negative correlation between Cu in airborne particulate matter in Belgrade central area and wind speed was reported, indicating that Cu was mainly emitted from some local sources. Bearing this in mind, and noting that bus and trolleybus terminus is located in vicinity of this sampling location (Table 1), a possible explanation for high Cu concentration is that vehicle-related emis-sion sources, such as brake and trolley pole abrasion, could be responsible for high Cu concentration in SP.

In topsoil samples from SP the largest concentration of Zn was also observed, in comparison with the other three parks. Zinc is mainly emitted by wear of tires and brake pads [35,36]. Note that other studies showed higher Cu and Zn concentrations at sites with more frequent start-stop driving [37,38], which is also char-acteristic for streets close to SP.

Apart from Fe, concentrations of Mn and Cr exhi-bited the least scatter of values between different parks. Iron and Mn are dominantly of crustal origin, but they can also be emitted from anthropogenic sources, such as motor vehicles and metallurgical industry [30,39].

Distribution of heavy metals with soil depth

In addition to concentrations of heavy metals in topsoil, variations of their content with soil depth were analyzed and the results are presented in Figures 2 and

3. The metal concentrations show variations with depth, but generally without uniformity of trends.

According to our results (Figure 2), Cu concentra-tions at all depths in SP are higher than those in other parks. Zinc concentrations in the top 0–30 cm layer are also higher in SP than in other parks. The average Zn concentration from 30–50 cm is by about 20% lower than in the upper layer and it is within the range of the concentrations in the other parks. The range of Ni con-centrations at all depths in ZP exceeds those in the other parks.

Figure 2. Distribution of Cu, Ni and Zn concentrations with soil depth in the four parks (KP, SP, ZP and BG).

In SP and ZP, the concentrations of Cu and Zn showed markedly lower values at 40–50 cm depth in comparison with those in the topsoil samples. Both elements had the highest topsoil concentrations in these two parks. This is particularly evident for Cu in SP, the park in which its concentration in topsoil is large,


648

indicating its considerable enrichment at the surface. Note that Cu is among metals with very low mobility, and has been reported in literature to accumulate in the topsoil [1,40,41]. Conversely, Zn has been reported to have high mobility and is likely to migrate down through the soil profile [42]. However, urban topsoil is constantly being enriched with Zn, particularly in area with high traffic density [43]. These two metals showed the opposite, increasing trend with depth in KP. Since the degree of Cu and Zn retention (and thus variation with depth) in soil depends on various parameters such as pH, amount of metal, organic matter content and soil mineralogy, it is clear that further analysis is neces-sary to explain these opposed results. Such analysis will be subject of our further studies.

Figure 3. Distribution of Cr and Mn concentrations with soil depth in the four parks (KP, SP, ZP and BG).

Nickel content showed moderate decrease with depth, most significant in ZP, the park in which its con-centration is the highest. This is in agreement with its high mobility in soil [1]. As Figure 3 shows, low enrich-ment at the surface was observed for Cr and Mn. How-ever, note that for SP and BG, as well as KP except for the topsoil, the measured Cr concentrations are below the lower limit of the calibration range (100 mg kg–1).

Enrichment factors The enrichment factor (EF) values have been cal-

culated in order to assess the contamination levels of heavy metals in the studied parks. Background concen-trations, necessary for this calculation, are usually

obtained from measurements in an area with little impact of anthropogenic pollution [44], measurements of deep soil layers [45], or average Earth’s crust com-position from literature [46–50]. Since in this study there were no park areas that could be considered as reasonably unpolluted, and due to lack of data from deep soil layers, we used average soil crust compo-sition given by Mason [49], which was also used for this purpose in other studies on soil contamination [51–53].

The EF values were calculated for the topsoil samp-les and those taken from 40–50 cm depth (except for KP, where samples were taken down to 30 cm), and presented in Figure 4. Since the concentration of Fe (the reference element here) in SP is somewhat below the calibration range, the EF values for this park are not shown in Figure 4. The results show that Zn is the most enriched heavy metal in topsoil in all four parks. Its enrichment factor is higher than 2 in all parks and show

Figure 4. Heavy metal enrichment factor (EF) values for top-soil (0–10 cm) and at 40–50 cm depth, in analyzed parks (KP, ZP and BG).


649

values above or close to 2 even below the topsoil. Copper has generally the smallest EF value among the analyzed metals.

The presented results show that local emission sources have strong influence on park topsoil concen-trations of Zn, Cu, Cr and Ni. More information on the effect of traffic and other urban emissions on urban soil elemental content could be obtained from the analyses of distribution of heavy metals in park topsoil, as well as at higher depths.

CONCLUSION

We performed the analysis of heavy metal content of urban soil in four Belgrade parks: Karađorđev Park (KP), Studentski Park (SP), Zemun Park (ZP) and Bota-nical Garden (BG). The samples were taken from a location in each park close to busy streets, at different depths down to 50 cm. The analysis of the samples was carried out using the EDXRF spectrometry. The results show high Cr concentration in comparison with average values for urban soil in Belgrade and other world cities, reported in earlier studies. Zinc was the most enriched element in topsoil in all parks, and together with Cu showed the highest concentration in SP, a park close to a bus and trolleybus termini, suggesting high influence of traffic emissions at this location. The highest Ni and Cr concentrations were observed in ZP, and are pro-bably related to fossil fuel combustion processes, as this park is located in the part of the city characterized by a larger number of individual heating units, in com-parison with the other three parks. These metals exhi-bited the most significant decrease with soil depth: Zn and Cu in SP, and Ni and Cr in ZP. The results presented here show strong influence of local emission sources on park topsoil concentration of these four metals (Zn, Cu, Cr and Ni). However, for more conclusive results the analysis should include the total area of each park. The analyses of distribution of heavy metals in park topsoil, as well as at higher depths, would provide more insight into the effect of traffic and other urban emissions on urban soil elemental content.

Acknowledgement

This paper was realized as a part of the project No III43007 financed by the Ministry of Education, Science and Technological Development of the Republic of Serbia within the framework of integrated and inter-disciplinary research for the period 2011–2014.

REFERENCES

[1] D.C. Adriano, Trace Elements in Terrestrial Environ-ments: Biogeochemistry, Bioavailability and Risks of Metals, Springer, New York, 2001.

[2] B. Wei, F. Jiang, X. Li, S. Mu, Heavy metal induced ecological risk in the city of Urumqi, NW China, Environ. Monit. Assess. 160 (2010) 33-45.

[3] B.J. Alloway, Trace Metals in Soil, Chapman & Hall, London, 1995.

[4] E. Steinnes, in: T. C. Hutchinson and K. M. Meema (Eds.), Lead, Mercury, Cadmium and Arsenic in the Environ-ment, John Wiley & Sons Ltd., 1987, pp. 107–117.

[5] G. Schüürmann, B. Markert, Ecotoxicology: Ecological Fundamentals, Chemical Exposure, and Biological Effects, John Wiley & Sons, New York, 1998.

[6] O.M. Saether, R. Krog, D. Segar, G. Storroe, Contam-ination of soil and ground water at former industrial site in Trondheim, Norway, Appl. Geochem. 12 (1997) 327– –332.

[7] O. Selinus, B. Alloway, J. A. Centeno, R. B. Finkelman, R. Fuge, U. Lindh, and P. Smedley, Essentials of Medical Geology: Impacts of the Natural Environment on Public Health, Academic Press, 2005.

[8] P.W. Abrahams, Soils: Their implications to human health, Sci. Tot. Environ. 291 (2002) 1–32.

[9] B. Wei, L. Yang, A review of heavy metal contaminations in urban soils, urban road dusts and agricultural soils from China, Microchem. Jour. 94 (2010) 99–107.

[10] M.D. Marjanović, M.M. Vukčević, D.G. Antonović, S.I. Dimitrijević, Đ.M. Jovanović, M.N. Matavulj, M.Đ. Ristić, Heavy metals concentration in soils from parks and green areas in Belgrade, J. Serb. Chem. Soc. 74 (2009) 697–706.

[11] I. Gržetić, R.H.A. Ghariani, Potential health risk assess-ment for soil heavy metal contamination in the central zone of Belgrade (Serbia), J. Serb. Chem. Soc. 73 (2008) 923–934.

[12] D. Crnković, M. Ristić, D. Antonović, Distribution of heavy metals and arsenic in soils of Belgrade (Serbia and Montenegro), Soil Sediment Contam. 15 (2006) 581– –589.

[13] Z. Mijić, A. Stojić, M. Perišić, S. Rajšić, M. Tasić, M. Radenković, J. Joksić, Seasonal variability and source apportionment of metals in the atmospheric deposition in Belgrade, Atm. Environ. 44 (2010) 3630–3637.

[14] MiniPal 4 User’s Guide, PANalytical BV, Amelo, Nether-lands, 2006.

[15] K.D. Daskalakis, T.P. O’Connor, Normalization and ele-mental sediment contamination in the coastal United States, Environ. Sci. Technol. 29 (1995) 470–477.

[16] L.F.H. Niencheski, B. Baraj, R.G. Franca, N. Mirlean, Lithium as a normalizer for assessment of anthropo-genic metal contamination of sediments of the southern area of Patos Lagoon, Aquat. Ecosyst. Health 5 (2002) 473–483.

[17] K. Loska, J. Cebula, J. Pelczar, D. Wiechula, J. Kwapil-inski, Use of enrichment and contamination factors together with geoaccumulation indexes to evaluate the content of Cd, Cu, and Ni in the Rybnik water reservoir in Poland, Water Air Soil Poll. 93 (1997) 346–365.

[18] R.A. Sutherland, Bed sediment-associated trace metals in an urban stream, Oahu, Hawaii, Environ. Geol. 39 (2000) 611–627.


650

[19] D.S. Manta, M. Angelone, A. Bellanca, R. Neri, M. Spro-vieri, Heavy metals in urban soils: a case study from the city of Palermo (Sicily), Italy, Sci. Tot. Environ. 300 (2002) 229–243.

[20] M. Imperato, P. Adamo, D. Naimo, M. Arienzo, D. Stanzione, P. Violante, Spatial distribution of heavy metals in urban soils of Naples city (Italy), Environ. Pollut. 124 (2003) 247–256.

[21] L. Madrid, E. Díaz-Barrientos, F. Madrid, Distribution of heavy metal contents of urban soils in parks of Seville, Chemosphere 49 (2002) 1301–1308.

[22] L. Bityukova, A. Shogenova, M. Birke, Urban geochemis-try: A study of element distributions in the soils of Tallinn (Estonia), Environ. Geochem. Hlth 22 (2000) 173–193.

[23] O. Morton-Bermea, E. Hernández-Álvarez, G. González-Hernández, F. Romero, R. Lozano, L.E. Beramendi-Orosco, Assessment of heavy metal pollution in urban topsoils from the metropolitan area of Mexico City, J. Geochem. Explor. 101 (2009) 218–224.

[24] F.J. Gallaghera, I. Pechmann, J. D. Bogden, J. Grabosky, P. Weis, Soil metal concentrations and vegetative assemblage structure in an urban brownfield, Environ. Pollut. 153 (2008) 351–361.

[25] S. Maas, R. Scheifler, M. Benslama, N. Crini, E. Lucot, Z. Brahmia, S. Benyacoub, P. Giraudoux, Spatial distribu-tion of heavy metal concentrations in urban, suburban and agricultural soils in a Mediterranean city of Algeria, Environ. Pollut. 158 (2010) 2294–2301.

[26] Environmental quality in the city of Belgrade, 2011, http://www.zdravlje.org.rs/publikacije/Zivotna-sredina-bgd-2011-II%20korektura.pdf (in Serbian).

[27] C. Barbante, C. Boutron, A.-L. Moreau, C. Ferrari, K. Van de Velde, G. Cozzi, C. Turetta, P. Ceson, Seasonal vari-ations in nickel and vanadium in Mont Blanc snow and ice dated from the 1960s and 1990s, J. Environ. Monit. 4 (2002) 960–966.

[28] ATSDR, Toxicological Profile for Nickel, Atlanta, GA: US Public Health Service, Agency for Toxic Substances and Disease Registry, 2005.

[29] S. Rajšić, Z. Mijić, M. Tasić, M. Radenković, J. Joksić, Evaluation of the levels and sources of trace elements in urban particulate matter, Environ. Chem. Lett. 6 (2008) 95–100.

[30] C. Samara, Th. Kouimtzis, R. Tsitouridou, G. Kanias, V. Simeonov, Chemical mass balance source apportion-ment of PM10 in an industrialized urban area of North-ern Greece, Atm. Environ. 37 (2003) 41–54.

[31] J.M. Pacyna, E.G. Pacyna, An assessment of global and regional emissions of trace metals to the atmosphere from anthropogenic sources worldwide, Environ. Rev. 9 (2001) 269–298.

[32] M. Aničić, T. Spasić, M. Tomašević, S. Rajšić, M. Tasić, Trace elements accumulation and temporal trends in leaves of urban deciduous trees (Aesculus hippocast-anum and Tilia spp.), Ecol. Indicat. 11 (2010) 824–830.

[33] K.S. Patel, A. Shukla, A.N. Tripathi, P. Hoffman, Heavy metal concentrations of precipitation in east Madhya

Pradesh of India, Water Air Soil Poll. 130 (2001) 463– –468.

[34] A. Pietrodangelo, R. Salzano, E. Rantica, C. Perrino, Characterization of the local topsoil contribution to air-borne particulate matter in the area of Rome (Italy). Source profiles, Atm. Environ. 69 (2013) 1–14.

[35] T.B. Concell, K.U. Duckenfield, E.R. Landa, E. Callender, Tire-wear particles as a source of zinc to the environ-ment, Environ. Sci. Technol. 38 (2004) 4206–4214.

[36] K. Adachi, Y. Tainosho, Characterization of heavy metal particles embedded in tire dust, Environ. Int. 30 (2004) 1009–1017.

[37] E. Furusjö, J. Sternbeck, A. Palm Cousins, PM10 source characterization at urban and highway roadside loc-ations, Sci. Tot. Environ. 387 (2007) 206–219.

[38] C. Laschober, A. Limbeck, J. Rendl, H. Puxbaum, Parti-culate emissions from on-road vehicles in the Kaiser-muhlen-tunnel (Vienna, Austria), Atm. Envron. 38 (2004) 2187–2195.

[39] Y.-S. Wu, G.-C. Fang, W.-J. Lee, J.-F. Lee, C.-C. Chang, C.-Z. Lee, A review of atmospheric fine particulate mat-ter and its associated trace metal pollutants in Asian countries during the period 1995–2005, J. Hazard. Mater. 143 (2007) 511–515.

[40] P. Hooda, Trace Elements in Soils, Blackwell Publishing Ltd., 2010.

[41] A. Kabata-Pendias, H. Pendias, Trace Elements in Soils and Plants, CRC Press, Boca Raton, FL, 2001.

[42] L. Citeau, I. Lamy, F. van Oort, F. Elsass, Colloidal faci-litated transfer of metals in soils under different land use, Colloids Surfaces, A 217 (2003) 11–19.

[43] F. Napier, B. D’Arcy, C. Jefferies, A review of vehicle related metals and polycyclic aromatic hydrocarbons in the UK environment, Desalination 226 (2008) 143–150.

[44] B. Škrbić, S. Milovac, M. Matavulj, Multielement pro-files of soil, road dust, tree bark and wood-rotten fungi collected at various distances from high-frequency road in urban area, Ecol. Indicat. 13 (2012) 168–177.

[45] T.T.T. Dung, V. Cappuyns, R. Swennen, E. Vassilieva, N. K. Phung, Leachability of arsenic and heavy metals from blasted copper slag and contamination of marine sedi-ment and soil in Ninh Hoa district, south central of Viet-nam, Appl. Geochem. (2013), in press (http:// //dx.doi.org/10.1016/j.apgeochem.2013.07.021)

[46] A.P. Vinogradov, The Geochemistry of Rare and Dis-persed Chemical Elements in Soils, Chapman & Hall, New York, 1959.

[47] K.H. Wedepohl, in: L.H. Ahrens (Ed.), Origin and dis-tribution of the elements, Pergamon Press, London, 1968, pp. 999–1016.

[48] K.H. Wedepohl, The composition of the continental crust, Geochim. Cosmochim. Acta 59 (1995) 1217–1232.

[49] B. Mason, Principles of Geochemistry, John Wiley & Sons Inc., New York, 1966.

[50] S.R. Taylor, Abundance of chemical elements in the continental crust: a new table, Geochim. Cosmochim. Acta 28 (1972) 1273–1285.


651

[51] H. Pekey, Heavy metal pollution assessment in sedi-ments of the Izmit Bay, Turkey, Environ. Monit. and Assess. 123 (2006) 219–231.

[52] N. Guvenç, O. Alagha, G. Tuncel, Investigation of soil multi-element composition in Antalya, Turkey, Environ. Int. 29 (2003) 631–640.

[53] U.M. Joshi, K. Vijayaraghavan, R. Balasubramanian, Ele-mental composition of urban street dusts and their dis-solution characteristics in various aqueous media, Che-mosphere 77 (2009) 526–533.

IZVOD

SADRŽAJ TEŠKIH METALA U ZEMLJIŠTU URBANIH PARKOVA BEOGRADA Maja M. Kuzmanoski, Marija N. Todorović, Mira P. Aničić Urošević, Slavica F. Rajšić

Institut za fiziku, Univerzitet u Beogradu, Beograd, Srbija

(Naučni rad)

Teški metali su prirodna komponenta zemljišta, ali njihovo prisustvo u povi-šenim koncentracijama može imati štetan efekat na okolinu, kao i na zdravljeljudi. Ovo je naročito problem u urbanim sredinama, gde emisije iz saobraćaja iindustrije značajno doprinose povećanom sadržaju teških metala u životnoj sre-dini, uključujući i zemljište. Osnovni cilj ovog rada je određivanje koncentracijateških metala u zemljištu parkova u urbanom području Beograda. Uzorci zemljištasu prikupljeni u maju 2011. godine, u Karađorđevom, Studentskom, Zemunskomparku i Botaničkoj bašti, do dubine od 50 cm. Koncentracije Cr, Cu, Fe, Mn, Ni i Znu uzorcima merene su energetski disperzivnom fluorescentnom spektroskopijomX-zracima (EDXRF). U površinskom sloju zemljišta (0–10 cm dubine) dobijen je sledeći redosled koncentracija: Fe >> Mn > Zn > Cr > Ni > Cu. Najviše koncentracijeCu i Zn, metala koji se uglavnom emituju iz saobraćaja, dobijene su u Studentskomparku koji se nalazi u neposrednoj blizini okretnice autobusa i trolejbusa. S drugestrane, najviše koncentracije Ni i Cr dobijene su u Zemunskom parku, u delu gradasa velikim brojem individualnih grejnih jedinica i verovatno potiču od sagorevanjafosilnih goriva. Najznačajnije smanjenje koncentracije sa dubinom pokazali su Zn iCu, a zatim Ni i Cr, i to u parkovima u kojima su njihove koncentracije u povr-šinskom sloju zemljišta najviše. Koncentracije Cr i Ni su visoke u poređenju saodgovarajućim vrednostima u drugim svetskim gradovima. Rezultati predstavljeniu ovom radu ukazuju na značajan doprinos lokalnih izvora emisije koncentraci-jama Cu, Cr, Zn i Ni u zemljištu u urbanim parkovima Beograda. Dalja ispitivanja bitrebalo usmeriti na analize zemljišta sa celokupnih površina parkova. Analizadistribucije teških metala u površinskom sloju zemljišta, kao i na većim dubinama, pružila bi više informacija o uticaju saobraćaja i drugih urbanih izvora zagađenjana sadržaj teških metala u zemljištu parkova.

Ključne reči: teški metali • zemljište •urbani parkovi • EDXRF

Documents

Vol. 68 Èasopis Saveza hemijskih inženjera · 2017. 2. 3. · 5 UDK 66:54(05) CODEN HMIDA 8 ISSN 0367–598 X Vol. 68 Èasopis Saveza hemijskih inženjera BEOGRAD, SEPTEMBAR-OKTOBAR