Microstructure of the Nitrogen-Induced Localized State in GaAsN Thin

Films Grown by Chemical Beam Epitaxy

Atsuhiko Fukuyama!, Wen Ding!, Goshi Morioka! Akio Suzuki!, Hidetoshi Suzuki2, Masafumi

Yamaguche, and Tetsuo Ikari!

! Faculty of Engineering, and 2Interdisciplinary Research Organization, University of Miyazaki, 1-1 Gakuen­

Kibanadai-Nishi, Miyazaki, 889-2192, Japan 3Toyota Technological Institute, 2-12-1, Hisakata Tempaku, Nagoya, 468-8511, Japan

Abstract - To investigate the microstructure of nitrogen­induced localized state (EN) that perturbs the conduction band of GaAs host material, we adopted the photo reflectance measurements to chemical-beam-epitaxy grown GaAsN thin films. We measured both two split subbands to estimate correct values of EN. It was clearly seen that estimated EN decreased with increasing nitrogen content and temperature. By considering a change of a mean distance between neighboring nitrogen atoms, we concluded that the microscopic structure of EN is not only an isolated nitrogen atom but also nitrogen-related complex, accompanied by low-order pairs of nitrogen atoms and/or nitrogen clusters.

Index Terms - InGaAsN, III-V semiconductors, concentrator photovoltaic solar cells, nitrogen-induced localized state.

I. INTRODUCTION Dilute nitride semiconductor InGaAsN has been expected

as an absorbing layer material for ultra-high-efficiency four­

junction solar cells such as InGaP/GaAs/InGaAsN/Ge

structure [1]. This is because the band gap energy (Eg) of

approximately 1.0 eV and the same lattice constant as GaAs

and Ge are realized by adding a few percent of nitrogen. In

this case, indium can be used to compensate the nitrogen­

induced reduction of the lattice parameter. It is well known

that Eg of GaAs1_xNx drastically decreases with increasing the

nitrogen content (x) and x � 0.03 is required to realize Eg � 1.0

eV. This anomalous behavior of Eg has been understood in

terms of a band anti-crossing (BAC) model proposed by Shan

et al. [2] In this model, an anti-crossing interaction between

electronic states of the localized nitrogen (EN) and the

semiconductor matrix leads to a characteristic splitting of the

conduction band into two non-parabolic subbands of K and

E+. In this case, K corresponds to Eg of GaAsN and E+ is

related to a new and unique conduction-band edge of GaAsN.

Shan et al. also proposed that EN is located at about 1.65 eV

above the top of the valence band from their photoretlectance

(PR) measurements under hydrostatic pressures [3]. They also

fixed at CNM = 2.7 eV, where CNM is the coupling parameter

between EN and unperturbed Eg of GaAsN. However, there is

no insight into the basic parameters EN and CNM• In other

words, it is not clear yet whether these parameters are

determined by only nitrogen content or are modified by other

978-1-4799-4398-2114/$31.00 ©2014 IEEE

factors, such as temperature, distribution of nitrogen atoms,

and its related defects.

We have already reported that EN value decreased as the

nitrogen content increased [4]. In this study, we adopted the

PR measurements to the high-nitrogen-homogeneity GaAsN

thin films and measured perturbed both subbands of E+ and E. in order to estimate correct values of EN. Our results suggested

that consideration of the energy levels of an isolated nitrogen

(Nx), pairs of nitrogen atoms (NN;), where i = 1, 2, ... , in order

of increasing pair separation, and nitrogen clusters may help

explain the experimental results. In the present study, we

investigated the temperature dependences of EN to obtain more

reasonable knowledge of nitrogen induced localized state. Its

microscopic structure will be discussed, especially for the

relationship between EN and nitrogen-related carrier traps.

II. EXPERIMENTAL PROCEDURES The n-type GaAs1_xNx thin films were grown on a semi­

insulating GaAs substrate by chemical beam epitaxy (CBE)

method. The substrate orientation was (100) tilted at 2 degree

toward [110]. This is because the hole mobility in GaAsN

films has been improved by using vicinal GaAs(OOI) wafers as

substrates [5], a low density of nitrogen-related carrier traps

have been expected. Since the growth pressure was

approximately 10-2

Pa, the chemical reaction in the vapor was

neglected and growth was occurred at only the surface of the

sample by decomposing the precursors of Ga (triethylgallium:

TEGa), As (trisdimethylaminoarsenic: TDMAAs), and N

(monomethylhydrazine: MMHy). An attenuated silane gas was

used as a Si dopant. The growth temperature was at 440°C

and the growth rate of the GaAsl_xNx film was about 0.4 !-1m/h.

The thicknesses of GaAsl_xNx thin films were the range from

500 to 630 nm. The nitrogen contents were estimated to be

0.003, 0.008, O.oI 1, and 0.018 by X-ray diffraction

measurements with assuming completely coherent growth and

the validity of Vegard's rule.

The PR measurements were carried out using a standard

setup [6]. A modulated Ar+

laser (488 nm, 3 mW) was used as

the excitation source to generate an electric field. The probing

light for measuring retlectivity was incident on the GaAsN

surface at an angle of 45°. The modulation frequency was set

1148

at 279 Hz. The AC and DC components of the optical

reflection from the sample surface were detected by a lock-in

amplifier and a digital multimeter, respectively. Then, the ratio

of them (�R) was calculated. Transition energies were

estimated by fitting the third derivative Aspnes' function to the

I'lRiR spectrum [7]. The PR measurements were carried out at

temperatures ranging from 4 K to room temperature.

III. REsULTS AND DISCUSSION

As reported in our previous paper [4], Figure 1 shows the

nitrogen content dependences of the K and E+ estimated by

the third derivative Aspnes' fitting analysis at room

temperature. The E_ and E+ denote the transitions between

valence bands and the two split conduction subbands,

respectively. The K showed red-shift whereas E+ moved to

the opposite direction with increasing nitrogen content. Figure

1 also shows the predicted values of K and E+ calculated by

using the following expression,

E ± = � [E N + EM ± {( E N - EM) 2 + 4C NM 2

X r 2 ]

(1)

Fig. 1 (a) Experimental E- and E+ and (b) estimated EN and

CNM from Eqs. (2) and (3) as a function of the nitrogen content at

room temperature [Fig. 3 in Ref. 4].

978-1-4799-4398-2/14/$31.00 ©2014 IEEE

where EM is the energy of the conduction band edge of

unperturbed. Since EM is the conduction band energy of

GaAsN neglecting the interaction with EN, it might be

calculated by linear interpolation between Eg of GaAs and

GaN [8]. In these calculations, we tentatively used EN = 1.64

and CNM = 2.7 eV by reference to previously published data [3,

9]. Although the experimental points seem to coincide to the

expected curves, one can note small but evidence deviations

between them. The experimental data depart higher and lower

sides for low and high nitrogen contents.

Therefore, the correct values of EN and CNM are calculated

by using Eq. (1) as

EN = E+ +E_ -EM (2)

CNM = �{(E+ -Ej -(EN - EM)2 }/ 4x (3)

and were plotted in Fig. l(b). It was found that EN decreased

with increasing nitrogen content. This is contrast to the

discussion of Shan et ai., where they had assumed the constant

energy of EN = 1.65 eV [2, 3]. Our results suggested that the

microstructure of EN may not only be Nx (isolated nitrogen)

but also contain NNj, and/or nitrogen clusters [4].

Figure 2 shows the temperature dependences of �R spectrum for GaAsN film with 0.8% sample. The E- as the

bandgap of GaAsN and E+ were clearly observed. They shifted

to higher photon energies with decreasing temperature. The

transition energy of E - +I'lso that indicates the transition

between the spin-orbit split-off valence band and E- was also

observed and its shift amount was same as that of E- . By

using Eq. (2), we calculate EN for all samples and plotted in

Fig. 3. As shown in the figure, EN for all samples increased

0::: -0::: <l

Photon Energy (eV) Fig. 2 Temperature dependences of the PR spectrum for

GaAsN film with 0.8%.

1149

Fig. 3 Temperature dependences of EN for all GaAsN thin film

samples.

0.40 ....--------------...,

Fig. 4 The nitrogen-content-change of -dEN/dT. The dashed

line is just to guide the eye.

with decreasing the temperatures due to increasing the

bandgap energies of host GaAs. For detailed discussion, we

evaluated the temperature coefficient of EN (dEN/dT) by the

least square method for the temperature range from 100 K to

room temperature. Results were shown in Fig. 3 as a function

of nitrogen content. We found -dEN/dT = 0.33 meV/K for a

present GaAsO.992No.008 thin film sample, in quite good

agreement with previously published data of 0.33 meV/K for a

GaAsO.9935No.0065 thin film [10]. The most important fmding

was that -dEN/dT showed a negative correlation with nitrogen

content.

The Eg of semiconductors tend to decrease with increasing

temperature, it can be described by Varshni's empirical

978-1-4799-4398-2/14/$31.00 ©2014 IEEE

expression. When the temperature of the sample increases,

amplitude of atomic vibration increases, this leads to a larger

interatomic spacing. The temperature coefficient of Eg of

GaAs was reported to be 0.452 meVIK [11]. If EN is a

localized level caused by an isolated nitrogen atom, its energy

is kept constant at 1.65 eV above the conduction band and the

-dEN/dT = 0.452 meVIK is also expected following the

temperature dependence of host semiconductor matrix.

However, -dEN/dT decreased as increasing nitrogen content.

To explain present nitrogen-content dependencies of -dEN/dT, we have to consider an interaction between adjacent nitrogen

atoms, especially a change in distance of neighboring nitrogen

atoms in GaAsN. Decrease of a lattice constant accompanied

by a temperature decrease results in the decrease of the

distance between neighboring nitrogen atoms. It is expected

that -dEN/dT may change with a mean distance between

neighboring nitrogen atoms. These experimental results also

imply that the microscopic structure of EN is not only an

isolated nitrogen atom but also nitrogen cluster as proposed in

our previous paper [4].

Next, we discuss a coupling parameter of CNM• The CNM were calculated by using Eq. (3) and results at room

temperature are shown in Fig. l(b) by open diamonds. We

obtained CNM = 2.72 eV for GaAsO.989No.Olb this qualitatively

confirms the CNM = 2.7 eV in the previously published data [9].

Although CNM showed an increasing tendency with nitrogen

content, we concluded that there is not a nitrogen-content

dependency of CNM• The temperature dependencies of CNM were also calculated

by using the low-temperature PR measurements. Results were

plotted in Fig. 5. In contrast to the EN, CNM of all samples did

not change by temperature. On this point, Grau et al. [10]

Fig. 5 Temperature dependences of CNM for all GaAsN thin

film samples

1150

reported the temperature dependencies of CNM for

GaAsO.9935No.0065 and GaAsO.988No.OI2 thin film samples. As

mentioned above, however, we considered that they mistook

about EM, which is the conduction band energy of GaAsN

neglecting the interaction with EN. We then concluded that

CNM does not depend on the nitrogen content or temperature.

IV. SUMMARY

We have investigated the microstructure of nitrogen­

induced localized level EN that perturbs the conduction band

of GaAs host material. To estimate correct values of EN, we

adopted the PR measurement to high-nitrogen-homogeneity

CBE-grown GaAsN thin films and measured perturbed both

subbands of E+ and K . It was clearly seen that calculated EN decreased with increasing nitrogen content. From the low­

temperature PR measurements, it was found that EN increased

with decreasing temperature. Most important fmding was that

the temperature coefficient of -dEN/dT showed a negative

correlation with nitrogen content. By considering a change of a

mean distance between neighboring nitrogen atoms, we then

concluded that the microscopic structure of EN is not only an

isolated nitrogen atom but also nitrogen-related complex.

Considering a microstructure that EN consists of Nx, NN" and

cluster may explain the present result why the estimated EN decreased and -dEN/dT showed a negative correlation with

increasing nitrogen content.

The coupling parameter of CNM were also calculated. We

concluded that there is not a nitrogen-content dependency of

CNM• The temperature dependencies of CNM were also

investigated. In contrast to the EN, CNM of all samples did not

change by temperature. We then concluded that CNM does not

depend on the nitrogen content or sample temperature.

This work was supported in part by the Japan-EU joint

R&D project on the NGCPV under NEDO in the Ministry of

Economy, Trade Industry Japan, and in part by Grant-in-Aid

for Scientific Research (B) and (C) under the Ministry of

Education, Culture, Sports, Science and Technology.

978-1-4799-4398-2/14/$31.00 ©2014 IEEE

REFERENCES [1] D. J. Friedman, J. F. Geisz, S. R. Kurtz, and J. M. Olson, "1-ev

solar cells with GaInNAs lattice-matched to GaAs," J. Cryst. Growth 195, 409 (1998).

[2] W. Shan, W. Walukiewicz, J. W. Ager III, E. E. Haller, J. F. Geisz, D. J. Friedman, J. M. Olson, and S. R. Kurtz, "Band anticrossing in GaInNAs alloys," Phys. Rev. Lett. 82, 1221 (1999).

[3] W. Shan, W. Walukiewicz, K. M. Yu, J. W. Ager III, E. E. Haller, J. F. Geisz, D. J. Friedman, J. M. Olson, S. R. Kurtz, C. Nauka, "Effect of nitrogen on the electronic band structure of group III-N-V alloys," Phys. Rev. B 62, 4211 (2000).

[4] H. Suzuki, A. Suzuki, A. Fukuyama, and T. Ikari, "Nitrogen­induced localized level observed by photoreflectance in GaAsN thin films grown by chemical beam epitaxy," J. Cryst. Growth 384, pp. 5-8 (2013).

[5] H. Suzuki, M. Inagaki, T. Honda, Y. Ohshita, N. Kojima, and M. Yamaguchi, "Improvements in optoelectrical properties of GaAsN by controlling step density during chemical beam epitaxy growth," Jpn. J. Appl. Phys. 49, 04DP08 (2010).

[6] A. Suzuki, T. Mori, A. Fukuyama, T. Ikari, J. H. Paek, and M.

Yamaguchi, "Effect of Silicon doping on the photoluminescence

and photo reflectance spectra of catalyst-free molecular beam

epitaxy-vapor liquid solid grown GaAs nanowires on (111 )Si

substrate," Jpn. J. Appl. Phys. 50, 06GH08 (2011).

[7] D. E. Aspnes, "Third-derivative modulation spectroscopy with

low-field electrorefrectance," Surf. Sci. 37, 418 (1973).

[8] J. S. Blakemore, "Semiconducting and other major properties of

gallium arsenide", J. Appl. Phys. 53, R123 (1982).

[9] R. Chtourou, F. Bousbih, S. B. Bouzid, F. F. Charfi, J. C.

Harmand, G. Ungaro, and L. Largeau, "Effect of nitrogen and

temperature on the electronic band structure of GaAsN alloy,"

Appl. Phys. Lett. 80, 2075 (2002).

[10] A. Grau, T. Passow, and M. Hetterich, "Temperature

dependence of the GaAsN conduction band structure," Appl.

Phys. Lett. 89, 202105 (2006).

[11] J. Plaza, J. L. Castano, B. J. Garcia, H. Carrere, and E. B.­

Pereira, "Temperature dependence of photoluminescence and

photoreflectance spectra of dilute GaAsN alloys," Appl. Phys.

Lett. 86, 121918 (2005).

1151