The Impact of V Doping on the Carbothermal Synthesis of Mesoporous Mo Carbides

The Impact of V Doping on the Carbothermal Synthesis ofMesoporous Mo CarbidesThomas Cotter,† Benjamin Frank, Wei Zhang,‡ Robert Schlogl, and Annette Trunschke*

Department of Inorganic Chemistry, Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany

*S Supporting Information

ABSTRACT: A series of bimetallic carbides of the form β-(Mo1−xVx)2C (0 < x < 0.12) wassynthesized by carbothermal reduction of corresponding h-Mo1−xVxO3 precursors. The oxideswere synthesized by precipitation, and the subsequent carbide phase development wasmonitored. The reduction mechanism is discussed on the basis of observed structuralevolution and solid-state kinetic data. The reduction is observed to proceed via a complexmechanism involving the initial formation of defective MoIV oxide. Increasing the V contentretards the onset of reduction and strongly influences the kinetics of carburization. Thecarbides exhibit a trend in the growth morphology with V concentration, from a particulate-agglomerate material to a packed, nanofibrous morphology. The high-aspect-ratio crystallitesexhibit pseudomorphism, and in the case of the V-containing materials, some preferentialcrystal orientation of grains is observed. An increasing mesoporosity is associated with thefibrous morphology, as well as an exceptionally high surface area (80−110 m2/g). Thesynthesis was subsequently scaled up. By adapting the heating rate, gas flow, and pretreatmentconditions, it was possible to produce carbide materials with comparable physical properties to those obtained from the smallscale. As a result, it was possible to synthesize Mo2C materials in multigram quantities (5−15 g) with BET surface areas rangingfrom 50 to 100 m2/g, among the highest values reported in the literature.

KEYWORDS: Mo carbide, V carbide, carburization, mechanism

■ INTRODUCTION

Since early transition-metal carbides display noble metal-likeproperties,1,2 there has been a renewed interest in thesematerials in the field of catalysis, especially so in the case ofgroup V and VI carbides. It is hoped that such materials may beable to substitute scarcer and more expensive noble metals in agrowing field of applications, including fuel cells and energy-related catalysis. This resurgence is largely due to the discoveryof facile routes to highly dispersed bulk carbides and supportedcarbides by which transition-metal oxides are reduced using atemperature-programmed reaction in the presence of a carbon-containing gas.3 Depending on the conditions, this approachyields stoichiometric materials of high purity with relativelyhigh surface areas.Molybdenum carbide exhibits interesting activity in a number

of catalytic reactions, including isomerization,4 hydrogenationand hydrogenolysis,5 methanol reforming,6 and hydrotreatingreactions.7 It can also be formed under relatively mildconditions, making it an ideal material for the study of carbidesin catalysis. In addition, Mo participates in an extensivechemistry with oxygen providing various potential approachesto mono- and polymetallic carbides via multimetal oxideprecursors.Various approaches to polymetallic carbides are outlined in

the literature for their potential in hydrotreatment catalysis,including established variations of M6C-type carbides (i.e.,AxByC; x = 2, 3, 4; y = 6 − x; A, B = Ti, V, Cr, Mn, Fe, Co, Ni,Nb, Mo, W, Ta).8,9 Relevant to the present study is the

preparation of carbide solid solutions related to the hexagonalclose-packed (hcp) and face-centered cubic (fcc) carbidephases.10 Oyama et al. prepared bimetallic transition-metaloxide precursors of the form M1M2Ox (M1 = Mo, W; M2 = V,Cr, Fe, Co, Ni, Nb, Mo, W) by the solid-state fusion of simpleoxide components. The oxides were subsequently carburized,resulting in bimetallic oxycarbide materials.11,12 Also, recently,Bastos et al. have synthesized mixed Mo/W carbides by varyingthe synthesis method of the precursor compounds, includingcoprecipitation of mixed metal oxides.13

The carburization of MoO3 in a mixed feed of hydrogen andhydrocarbon is affected by the surface area of the oxide and thechain length of the alkane. Longer-chain hydrocarbons are ableto reduce the oxide at lower temperatures than pure H2. Eachsuccessive increase in chain length decreases the requiredtemperature to decompose the hydrocarbon and incorporatecarbon into the oxide matrix.14−17 Additionally, it is possible tocarburize MoOx using carbon alone as the reducing−carburizing agent.18−21 Depending upon the reaction con-ditions, the reduction−carburization can be directed to formeither the fcc α-Mo2C or the hcp β-Mo2C. Oyama et al.concluded that carbide formation occurs in two consecutivesteps due to similarities in the reaction profiles and effluent gasanalysis in temperature-programmed reduction of MoO3 in

Received: April 25, 2013Revised: June 26, 2013Published: June 28, 2013

Article

pubs.acs.org/cm

© 2013 American Chemical Society 3124 dx.doi.org/10.1021/cm401365y | Chem. Mater. 2013, 25, 3124−3136

pubs.acs.org/cm

hydrogen alone and the reduction−carburization (TPRC) ofMoO3 to β-Mo2C under H2/CH4. The first step comprises thereduction of MoO3 to MoO2, which is carried out by the actionof H2 alone.

3 The second step is the reduction−carburization ofMoO2 to Mo2C, which is performed by H2 and CH4 in concert.The synthesis of binary MoNb carbides seems to occursimilarly in two stages.11 A detailed mechanistic picture of thereduction−carburization process is, however, missing.In the present study, the synthesis of a series of bimetallic

carbides of the form (Mo1−xVx)2C (0 < x < 0.12) is monitoredusing in situ X-ray diffraction (XRD) and reaction gas analysisof temperature-programmed reduction−carburization. Theobserved structural evolution is combined with an analysis ofthe solid-state kinetic data, providing new insight into thereaction mechanism of carburization.With respect to the functional potential of the produced

materials, it is of interest to control the surface termination bysynthetic measures. It can be expected that stability andreactivity of the basal or prismatic surface orientation of thehexagonal Mo carbide are different. Controlling the orientationrequires kinetic control of the high-temperature synthesis. Thefrequently cited “topotactic nature” of the carburizationreaction was used as a positive indication. We further employeda series of precursor compounds to study the possibility ofcontrolling the reactivity of intermediates of the carburizationprocess.

■ EXPERIMENTAL SECTIONSynthesis of Precursor Oxides. Hexagonal molybdate h-MoO3

precursors (of the form h-Mo1−xVxO3) were prepared in an automatedlaboratory reactor (LabMax, Mettler Toledo) by the coprecipitation22

of a 0.032 M aqueous solution (total metals) of ammoniumheptamolybdate (NH4)6Mo7O24·4H2O (AHM, Merck, > 99%) andammonium metavanadate (NH4)VO3 (AMV, Aldrich, > 99%) atconstant pH and temperature. A bimetallic V-substituted modificationof h-MoO3 was chosen as the precursor to reduce potentialsegregation under the conditions of carburization. The precipitationpH of the mixed AHM/AMV solutions was determined using alaboratory titrator (Mettler DL77) at the desired temperature (75 °C)titrated with 1 M HNO3 (Figure 1). From the derivatives, it wasdecided to work at pH 0.5.For coprecipitation, the Mo/V solution (1000 mL) was added to

aqueous HNO3 (pH 1; 300 mL) maintained at a temperature of 75°C. The pH of 0.5 was maintained by automated monitoring and viaaddition of 3 M HNO3. The precipitate was aged for 2 h at 75 °C,collected by filtration, and washed with water (2 × 100 mL) and EtOH(100 mL) before drying overnight at 80 °C in air.23 It was observed

that Mo was preferentially precipitated from the reactant solution, andtherefore, the applied (set) and actual (EDX) Mo/V ratios, and thephysical properties of the oxides, are detailed in Table 1.

An exemplarily LabMax protocol is shown in Figure 2, which can beinterpreted as follows: (1) initial acidification to pH 0.5; (2) additionof metal salt solution (10 mL/min); (3) nucleation; (4) precipitation(endothermic event at t = 3.7 h); (5) aging. The endothermic eventcomprises a sudden decrease of the temperature by 1.5 °C, which isaccompanied by an increase of the pH from 0.5 to 0.53.

α-MoO3 used for in situ XRD experiments was obtained by spray-drying a 0.3 M solution of AHM in a spray dryer (Buchi B290; Tinlet =160 °C; Toutlet = 100 °C; pumping rate = 15% of maximum), followedby calcination in a muffle furnace at 500 °C for 10 h. For the TPRCexperiments, α-MoO3 was used as received (Sigma Aldrich).

In Situ Reduction−Carburization of Mo/V Oxides. Carbideswere synthesized via TPRC in a flowing atmosphere of 50 mL/min He(>99.9%), 40 mL/min H2 (>99.9%), and 10 mL/min CH4 (>99.9%).The reduction and subsequent phase formation was investigated by insitu powder X-ray diffraction (XRD). The precursor oxides (170−230mg) were loaded into the in situ cell and initially ramped to 300 °C(no structural changes were observed), where a diffraction pattern wasrecorded under isothermal conditions. Subsequently, the temperaturewas ramped at 1 °C/min and scans were recorded isothermally every50 K to a final temperature of 750 °C, which was held for 4 h beforecooling to room temperature (RT), at which point a final scan wasrecorded (see the Supporting Information, Figure S1). The sample waspassivated under 0.5% O2 in He (total flow = 100 mL/min) for 2 hbefore removing. The conditions of the in situ experiment were chosenin order to balance the temperature resolution and the overall heatingrate (0.56 °C/min) with the apparatus limitations, as well as recordinga functional 2θ range for Rietveld analysis.

The in situ powder XRD studies were performed in a STOE Theta/Theta diffractometer with reflection geometry (secondary graphitemonochromator; Cu Kα1+2 radiation (λ = 1.5418 Å); scintillationcounter) equipped with an Anton Paar XRK 900 in situ XRD cell. Insitu scans were carried out in the range 22° < 2θ < 43° for thetemperature program and a longer scan (5° < 2θ < 60°) at RT afterpassivation. The temperature program was combined with analysis ofthe outgas stream using a Pfeiffer OmniStar quadrupole massspectrometer (QMS).

Carburization of (Mo1−xVx)2C in a laboratory reactor. Thematerials synthesis was scaled up to multigram quantities using acustom-built rotary furnace with three heating zones (XerionAdvanced Heating GmbH), equipped with mass flow controllers forAr, O2, H2, and CH4 (Bronkhorst) and a PID heating controller(Eurotherm 2704). The setup is described in detail elsewhere.24 Ineach case, 10.0 g of h-Mo1−xVxO3 precursor was introduced to themodified SiO2 tube reactor and subjected to a temperature programunder a flowing mixture of 450 mL/min CH4/H2 (1:4) as follows: RTto 350 °C at 1 °C/min; 350 to 500 °C at 0.2 °C/min; 500 to 550 °Cat 1 °C/min; 550 to 675 °C at 0.2 °C/min; hold 4 h, then cool to RT(see the Supporting Information, Figure S1). Following the temper-ature program, the KF butterfly valves on the flange ends were closedand the sealed reactor was introduced to a glovebox.

Temperature-Programmed Reduction−Carburization. TPRCexperiments were carried out using 500 mg of precursor oxide in aSiO2 U-tube reactor heated in the isothermal zone of a tube furnace(Carbolite). The heating was controlled by a PID controller(Eurotherm 2416), and the temperature was monitored by athermocouple in the reactor bed. The gas analysis was carried outwith heated lines using an in-line IR COx detector (ABB) and a QMS(Pfeiffer Omnistar) to monitor reaction products. The oxides werecarburized in a flowing atmosphere of 200 mL/min Ar/H2/CH4(5:4:1) under temperature ramps of 1, 2, and 5 °C/min to a finalhold temperature of 675 °C for 2 h. Quantitative product analysis forH2O and COx was carried out by normalizing the QMS signals to thecalibrated COx detector and by summation of the total oxygen contentof the precursor oxides.

For solid-state kinetic modeling of the reduction process, the TPRCdata was analyzed using “NETZSCH Thermokinetics” software to

Figure 1. Titration curves for 1 M (metals basis) AHM and AHM/AMV solutions at 75 °C.

Chemistry of Materials Article

dx.doi.org/10.1021/cm401365y | Chem. Mater. 2013, 25, 3124−31363125

give the conversion-dependent apparent activation energies using thetechnique of Friedmann.25

In Situ and Ex Situ Characterization. Ex situ powder XRD wascarried out on the precursor oxides and on the as-synthesized carbidesusing a STOE STADI P diffractometer in transmission geometry(primary focusing Ge monochromator, Cu Kα1 radiation (λ = 1.5406Å), linear position sensitive detector). Full pattern analysis of the XRDdata was performed using TOPAS software.26 The pattern fit for eachmaterial was carried out using a Pawley refinement based on ahexagonal unit cell. The information on crystallite sizes wassubsequently obtained as volume-weighted average column heightbased on integral breadth (LVol-IB) according to the double-Voigtapproach.27

Differential scanning calorimetry (DSC) combined with thermog-ravimetric analysis (TGA) was used to study the thermaldecomposition of the precursor oxides. The study was carried outon a Netzsch Jupiter STA 449C calorimeter equipped with a QMS(Pfeiffer Omnistar).The C content of the carbide samples was measured by CHN

analysis carried out using a FlashEA 1112 elemental analyzer. In thecase of V-containing materials, pure oxygen is used to effect totalcombustion of the samples.All oxide and carbide samples were analyzed by full N2 adsorption−

desorption isotherms using a Quantachrome Autosorb AS-6Bmeasured after a pretreatment in vacuum at 200 °C for 2 h. Thespecific surface areas were determined according to the Brunauer−Emmett−Teller (BET) method using 11 data points in the relativepressure p/p0 range of 0.05−0.3. The pore size distributions werecalculated using the Barrett−Joyner−Halenda (BJH) method from thedesorption branch of the N2 sorption isotherm.Analysis of the carbide morphologies was investigated on a

HITACHI S 4000 FEG scanning electron microscope (SEM),operated at 1.5 or 2 kV. To determine elemental concentrations,energy-dispersive X-ray spectroscopy (EDX) was carried out using anEDAX DX-4 analysis system.Transmission electron microscopy (TEM) investigation of the Mo/

V carbide was carried out with a Philips CM 200 FEG (Philips,

Eindhoven, The Netherlands) operated at 200 kV and equipped with aGatan Image Filter (Gatan, Warrendale, PA) and a charge-coupleddevice (CCD) camera. An FEI 80−300 microscope was also employedto conduct elemental maps of the samples.

■ RESULTS AND DISCUSSIONHexagonal Precursor Oxides. h-MoO3 is a metastable,

complex molybdate defined by zigzag edge-sharing MoO6octahedral chains, which extend down the length of the caxis. They are edge-connected to form a superstructure, inwhich hexagonal channels travel down the length of the c axis.The channels are partially occupied by monovalent cationicspecies (e.g., H3O

+, NH4+, K+, Rb+, Cs+) that are intrinsic to the

stability of the structure and presumably serve a templating rolefor the crystallization of the phase. The synthesis was firstdescribed by Olenkova et al. in 1981, and the structure wassubsequently characterized in detail by Caiger et al.28−30 Incontrast, K0.13Mo0.87V0.13O3, a mixed Mo/V analogue, wasdescribed by Darriet et al.31 in 1973 (Figure 3), and the phase

stability of H0.13Mo0.87V0.13O3 is more recently demonstrated byDupont et al.32,33 In this study, we have used a syntheticmethod resulting in NH4

+-containing mixed Mo/V molybdatespreviously described by Mougin et al.22 The characterizationand description of this oxide series are outlined in Table 1.As can be seen from Figure 4a, the result of V substitution is

an increase of structural stability (h-MoO3 → α-MoO3), whichis observed in the profiles of the TGA−DSC analysis of theoxides h-MoO3 and h-Mo0.92V0.08O3. The increase in stability isobserved to increase to a maximum at approximately 10% Vsubstitution and is in accordance with a previous study.22 Thestatistical distribution of V atoms throughout the Mo sublatticeis further indicated in Figure 4b, which shows the divergence of

Table 1. Physico-Chemical Properties of the Precursor Oxides

V/(V + Mo) (at %) content (wt %), and lattice parametersc of the phases H1 and H2

sample IDa set act.b H1 a(H1) (Å) c(H1) (Å) H2 a(H2) (Å) c(H2) (Å) Td (°C) SBET (m2/g)

α-MoO3 5201 0 0 1.6h-MoO3 6600 0 0 75 10.5750(2) 3.7271(1) 25 10.6107(4) 3.7281(3) 403 <1h-Mo0.97V0.03O3 6702 4 2.7 ± 0.3 73 10.5735(2) 3.7231(1) 27 10.6221(4) 3.7232(3) 434 <1h-Mo0.95V0.05O3 6271 10 4.6 ± 0.3 76 10.5861(2) 3.7201(1) 24 10.6324(4) 3.7193(3) 452 1.1h-Mo0.92V0.08O3 6697 15 8.4 ± 1 78 10.5791(1) 3.7119(1) 22 10.6189(4) 3.7117(3) 479 <1h-Mo0.89V0.11O3 6601 20 11 ± 0.5 80 10.5838(1) 3.7091(1) 20 10.6143(5) 3.7095(3) 481 1.8

aNonambiguous internal sample number to distinguish reproductions of sample preparation. bAs determined by SEM-EDX. cAs determined byRietveld refinement. dPhase transformation as determined by TGA.

Figure 2. Typical LabMax protocol for the synthesis of h-Mo1−xVxO3.

Figure 3. Crystal structure of K0.13Mo0.87V0.13O3 as described byDarriet and Galy.31



the overlapping [120] and [101] diffraction peaks at 2θ = 25.8°with increasing V content.The series of Mo/V precursors with a h-MoO3 structure

were comparatively analyzed using a Rietveld refinement basedon the structure for KxMo1−xVxO3 suggested by Galy et al.31

The model parameters were refined after including NH4+ in

place of K+. It was observed that, especially in the case of V-substituted oxides, the structural model best approximated thepattern when a mixture of two identical phases was taken intoaccount. The inclusion of a minority phase (20−30%) helpedto account for various shoulder peaks and asymmetric peakshapes. In Table 1, the two phases are labeled as H1 (majority)and H2 (minority) for distinction. The Rietveld analysis for h-Mo0.92V0.08O3 (for results, see the Supporting Information,Figure S2) shows that the fit is reasonable but, in some cases,does not fully describe the peak shapes. For the c parameter,the oxide series obeys Vegard’s law34 with regards to a linearrelationship between the dopant concentration and the latticeparameter, and the trend is observed to be identical for bothphases. However, we see a relation between the phases H1 andH2 with a distinct maximum for h-Mo0.95V0.05O3. In this case,the a parameter describes the a−b lattice plane, in which theoctahedra are hexagonally arranged. This plane can be expectedto swell/shrink depending upon the occupation of the channelswith H2O and/or NH4

+. Thus, this variation may be assigned toan increasing cation (NH4

+) density in the hexagonal channelswith increasing V content. The a parameter reaches a maximum

and then shrinks as the influence of the smaller ionic diameterof V increasingly dominates.The observed difference in the cell parameters (∼0.05 Å) as

well as the insufficient Rietveld fit may be explained by abicontinuous variation generated by slight inhomogeneities.These can likely originate from the point of nucleation. We see,however, that the c parameter provides a better measure of thestatistical distribution of V throughout the structure as thelayer−layer distances in the c axis exhibit the same trends invariation (Supporting Information, Figure S2).As expected, the crystal morphology of the hexagonal oxides

(Figure 5) is hexagonal prismatic. The increase of the V content

distorts the crystal habit toward flattened hexagonal crystallitesof a more polydisperse nature. The nucleation−precipitation−aging growth of the oxides results in spherical polycrystallites,approximately 100−150 μm in diameter, with flat terminationsand decreased packing density with increasing V content.This is in accordance with the physical attributes of the

oxides, varying from the denser white crystalline h-MoO3 to ayellow-orange more loose powder for the doped materials. TheBET surface areas (Table 1) are uniformly low (∼1 m2/g).Furthermore, the crystal morphology exhibits some delamina-tion and structural distortions from the hexagonal geometry,which may account for the slightly higher surface area.

XRD Analysis of the Carburization. All precursor oxidesform the hcp β-carbides during TPRC under H2/CH4 (Figure6) in the in situ XRD experiments. Their physico-chemicalproperties are outlined in Table 2.The phase evolution of the precursor during TPRC is

illustrated in Figure S3 (Supporting Information). Thecrystalline fractions develop from the precursor oxide structure(α-MoO3, h-MoO3) to MoO2 between 350 and 450 °C, and tothe carbide between 550 and 600 °C. From the plots, itbecomes apparent that the temperature observed for thedisappearence of the h-Mo1−xVxO3 precursors increases withincreasing V content.In Figure 7, the diffraction patterns of the precursor oxides

are presented as a top-down plot of 2θ versus T with thenormalized square root of the intensity to enhance peakvisibility. From this plot, it is evident that the stability of theoxides to reduction increases with increasing V loading. At agiven temperature resolution, no difference in the final

Figure 4. (a) TGA−DSC traces for the degradation of precursoroxides h-MoO3 and h-Mo0.92V0.08O3. (b) XRD patterns of precursoroxides between 25° < 2θ < 27° illustrating the divergence ofoverlapping peaks with increasing V content.

Figure 5. SEM images of the hexagonal oxides: (a) h-MoO3; (b) h-Mo0.97V0.03O3, (c) h-Mo0.92V0.08O3; (d) h-Mo0.89V0.11O3.



carburization temperature is observed. However, significantpeak broadening occurs for the carburization of h-MoO3 ascompared to α-MoO3, and increasingly broad diffraction peaksare seen at higher dopant V concentrations.V-free h-MoO3 shows a by-phase that is presumably formed

in parallel to MoO2. The phase is not observed for the otherprecursors. To determine the nature of this by-phase, the in situcarburization experiments were repeated with an offsettemperature of 25 °C to interpolate the data and provide aneffective increase in time resolution. The intermediate scan at T= 375 °C is shown in the Supporting Information (Figure S4),and the unknown phase was determined to be orthorhombic o-Mo4O11, which has also been observed by Choi et al. in thereduction−carburization of ammonium molybdate.35

Formation of the closely related monoclinic m-Mo4O11 wasobserved in the reduction of α-MoO3 to MoO2 in H2. The roleof m-Mo4O11 in this reaction was long believed to be anintermediate in the reduction of MoVI to MoIV, which isthermodynamically stabilized under certain conditions oftemperature and H2 partial pressure. More recently, this hasbeen disputed, and it is demonstrated in separate studies to be aresult of the comproportionation of α-MoO3 and MoO2 via anO exchange mechanism.36−38 It was shown by Ressler et al. thatthe reaction

+ →3MoO MoO Mo O3 2 4 11 (1)

occurs in reducing conditions at T > 425 °C depending on thepartial pressure of H2.

36 In the same study, an Arrhenius plotgenerated from a series of isothermal reduction experimentsgives two apparent activation energies for the reductiondepending on the temperature. At T > 425 °C, the reduction

of MoO3 to MoO2 is observed to occur with Ea,app = 103 kJ/mol. Below 425 °C, the value lowers to 34 kJ/mol. It wasconcluded from this study that the activation energiescorresponded to the kinetic regime with and without thebyproduction of Mo4O11, respectively. This would imply thatthe reduction of Mo4O11 to MoO2 is kinetically more difficultthan the direct reduction of MoO3.The question as to why Mo4O11 is not usually observed in

the reduction of α-MoO3 (also not in this study) is addressedby Lalik39 in a recent study of the kinetics. It is proposed thatthe layered MoO3 structure is reduced in a topotactic reactionto MoO2 via a shear mechanism of the interlayer planes of α-MoO3. The resulting interfacial boundary between MoVI andMoIV may then collapse to the intermediate Mo4O11 givensufficient O mobility. The thickness of this interfacial layer(and, therefore, the XRD visibility) is dependent upon theexternal conditions.

Temperature-Programmed Reduction−Carburization.To complement the mechanistic picture of phase trans-formation, TPRC profiles were obtained to determine theoverall extent of reduction α. This extent was calculated byintegrating the total O content, which was observed to evolvefrom the structure according to the overall equations:

+ → + = − °TMoO H MoO H O ( 300 500 C)3 2 2 2 (2)

Figure 6. Normalized powder XRD patterns of the carbide productsmeasured at room temperature.

Table 2. Physico-Chemical Properties of As-Prepared hcp Carbides (TTPRC,max = 750 °C)

lattice parametersb

sample IDa a (Å) c (Å) V/(V + Mo)c (at %) C (theor.) (wt %) C (act.) (wt %) dhkld (nm) SBET (m2/g) Vp (cm

3/g)

β-Mo2C(ref) 5661 3.008(1) 4.761(1) 0 5.89 6.3 71β-Mo2C 6713 3.007(1) 4.778(1) 0 5.89 13.6 4.9 80 0.152β-(Mo0.97V0.03)2C 6708 3.007(1) 4.818(2) 2.7 ± 0.7 5.97 14.1 3.8 99 0.139β-(Mo0.95V0.05)2C 6704 3.006(1) 4.797(2) 4.4 ± 0.5 6.01 13.7 4.1 107 0.175β-(Mo0.92V0.08)2C 6703 3.004(1) 4.786(2) 8.0 ± 0.7 6.11 13.5 3.8 101 0.194β-(Mo0.89V0.11)2C 6701 3.017(1) 4.857(3) 11.3 ± 1 6.18 13.2 3.3 97 0.184

aNonambiguous internal sample number to distinguish reproductions of sample preparation. bAs calculated from XRD carried out using TOPASsoftware. cAs determined by EDX. dLVol-IB as derived from XRD refinement.

Figure 7. Contrast plot of XRD intensity vs T over the phase evolutionof precursor oxides: (a) α-MoO3; (b) h-MoO3; (c) h-Mo0.95V0.05O3;(d) h-Mo0.89V0.11O3. The outlined areas emphasize phase changesduring the TPRC treatment.



+ + → + > °T2MoO CH 2H Mo C 4H O ( 550 C)2 4 2 2 2(3)

This assumption is supported by the characteristic bimodalH2O trace recorded during TPRC (Supporting Information,Figure S6a), but simplified, because MoO2 is a crystallineintermediate, but not necessarily the only one. As well as H2O,CO is observed to evolve from steam reforming of CH4 overthe readily forming carbide via outgassed H2O (eq 4).Additionally, a small amount of CO2 is detected, probablydue to water gas shift reaction 5:

+ → +CH H O CO 3H4 2 2 (4)

+ → +CO H O CO H2 2 2 (5)

The total O content of the precursor systems was calculatedfrom the mass loss observed for the oxides upon heating insynthetic air to 500 °C (decomposed to α-MoO3) and derivedfrom the stoichiometry Mo:O = 1:3 (V:O = 2:5) to give adefined O content for each sample. This was also correlatedagainst the observed mass loss after reaction. The extent ofreduction at each point was determined by integration of thetotal of the QMS signals at m/z = 18 and 28, which werenormalized relative to the known CO concentration from theCOx IR detector. For simplicity, CO2 can be disregarded as itwas found to contribute <1% to the total O evolved.The low-temperature peak of H2O evolution (the only O-

containing product detected in the initial reduction1) for thehexagonal oxide series is shown in Figure 8. In general, all

traces show multiple reactions and clear indications of growthdelays due to nucleation phenomena. Probably, there areseveral suboxides involved. In agreement with the in situ XRDstudy, we observe a shift of the H2O peak to highertemperatures with increasing V content. At lower V loadings,there are additional peaks at 375−400 °C, which appear to berelated to the o-Mo4O11 phase observed by XRD for h-MoO3.This peak disappears entirely for h-Mo0.92V0.08O3, which alsopresents the most uniform reduction peak. At the highest Vloading, a slight shoulder at higher temperatures (475−500 °C)appears, which may be related to the reduction of VOxmoieties. In Figure 9a, the reduction profiles at differentheating rates are contrasted for α-MoO3, h-MoO3, and h-Mo0.92V0.08O3 (see also the Supporting Information, Figure S5).Comparison of the reduction profiles at 1 °C/min shows mostclearly the influence of the precursor structure on the reductionkinetics. Obviously, the onset of reduction is similar for all

samples, indicating a thermodynamic control of the process.54

For the reduction of α-MoO3, a continuous initial increase inreduction is observed at 375−425 °C. This could be associatedwith an increase in reacting surface area, followed by an almostlinear reduction to MoO2 at higher T. This is in contrast to h-MoO3, which exhibits a rapid initial reduction between 350 and400 °C that reflects the parallel formation of o-Mo4O11,followed by a relatively steep reduction curve to MoO2, whichis almost completed by 425 °C. The V-containing precursor h-Mo0.92V0.08O3 has a delayed onset of reduction but is smoothlyreduced to MoO2 at 400−450 °C. Increasing the heating rate inthe case of α-MoO3 and h-MoO3 has the effect of distorting thereduction curve to an overall exponential shape, which is mostpronounced in the case of α-MoO3 but also observed for the 5°C/min curve of h-MoO3. The V-containing precursor h-Mo0.92V0.08O3 shows a delayed onset of reduction but overallretains the same curve shape with only a slight distortion at 5°C/min. The distortion in the reduction curve for the highestheating rate can thus be explained as the delayed onset of Omobility due to V doping, meaning that V acts as an inhibitor ofion diffusion.The second stage of reduction from MoO2 to Mo2C is

illustrated in Figure 9b, which compares the extent of reductionof the precursor oxides versus T at a ramping rate of 1 °C/minover the temperature range of 450−675 °C. Here, we see thatthe influence of V is to retard the reduction−carburizationprocess. We can see from Figure 9c that the initial consumptionof methane (m/z = 15), indicating the onset of Cincorporation, begins at progressively higher temperaturesranging from 580 °C for α- and h-MoO3 to 595 °C for h-Mo0.92V0.08O3. When we correlate these temperatures with theintegrated reduction curve of Figure 9b, it is apparent that thecarburization process does not begin at the completion of thereduction MoO3 → MoO2. The average oxidation state of Mocan be seen to be ∼MoO1.3, which corresponds to Mo3O4 (orMoIVO2.2MoIIO, or 2MoIVO2·Mo0). It can be seen from theXRD patterns (Figure 7 and Figure S3 in the SupportingInformation) that crystalline MoO2 persists until around thistemperature range. Because there are no other crystallinephases observed, the implications of this are that MoIV iscoexisting with a lower oxidation state of Mo, such as metallicMo0 or alternatively “MoIIO”, a cubic oxide phase, which hasbeen observed as microdomains within stabilizing ma-trixes.40−42 Alternatively, an intermediate oxidation state MoIII

could exist, such as that reported by Davies et al.43 To date,there is very little support in the literature for such an electronicstate, and it should be considered as merely speculative withoutstronger evidence.MoIIO is discussed in the context of alkane isomerization

catalysis44,45 and is more widely accepted as an intermediateoxidation state of Mo.46,47 However, from the data at hand, wecannot make a further distinction as to the true nature of thisoxidation state apart from the fact that it lies directly betweenthe disappearance of the MoO2 crystalline phase and theemergence of Mo2C.What is happening in the case of α-MoO3 is less clear due to

the complexity of its reduction profile (see the SupportingInformation, Figure S6). α-MoO3 is known to incorporate Hand form bronzes as well as the so-called Magneli phases, whichcome about through a shear mechanism of reduction. FromFigure 9, it appears that incorporation of C in α-MoO3 beginsat an earlier stage of reduction; however, in situ XRD provesthe persistence of MoO2 until 575−600 °C. However, it still

Figure 8. TPRC profiles of the normalized H2O (m/z = 18) signal forh-Mo1−xVxO3.



remains unclear whether X-ray phases observed are theintermediates or spectators. As they are least reactive, itappears likely that MoO2 is not the precursor to carburization.Instead, it may decompose into suboxides at temperaturesbelow that of normal MoO2 hydrogen reduction.55 Here, thekinetic control at low temperatures, where little incentive existsfor the system to crystallize in view of shear structures andother local energetic minima that do not form crystals, is acritical aspect when combining XRD and TPR(C) data.In parallel with the removal of O in the form of H2O is the

observation of CO, which is formed via a complex methanesteam reforming process involving lattice carbon substitution,as demonstrated by Green.48 Interestingly, as seen in Figure 9d,formation of CO is observed at lower temperatures and is morepronounced for the α-MoO3 as compared to the hexagonalprecursors. In all cases, the peak is observed to correlate withthe completion of reduction (at which point the H2O reformingsource is depleted). The CO peak area related to the TPRC ofα-MoO3 is considerably larger than those observed for thehexagonal precursor oxides, implying that the reformingreaction is more effectively catalyzed.In Figure 9b, it is seen that h-MoO3 and h-Mo0.97V0.03O3 are

reduced to carbides more easily than h-Mo0.92V0.08O3 and α-MoO3. The trend observed for the h-(Mo,V)O3 samples is inaccordance with the hypothesis that V reduces the O mobilityin the structure and thus retards the removal of lattice O andsubsequently C inclusion.

Solid-State Kinetic Analysis of TPRC Profiles. The fullO reduction curves for α-MoO3, h-MoO3, h-Mo0.97V0.03O3, andh-Mo0.92V0.08O3 can be seen in the Supporting Information(Figure S6), in which α = 1 corresponds to Mo2C. In asimplified case, we can, therefore, define α = 0.33 for thereduction

→MoO MoO3 2 (6)

and the carburization is observed for 0.33 < α < 1

→MoO Mo C2 2 (7)

To picture these mechanistic steps, we have carried out amodel-free kinetic analysis using the Friedmann analysis,25

which is a multipoint analysis determined by plotting log(dα/dt) versus 1/T at different isoconversions for a range of heatingrates (β). The specific advantage of this analysis is that the plotdoes not require a constant rate of temperature change, whichis of relevance because the TPRC profiles were collected usinga temperature ramp, followed by a plateau. In the case of dataobtained at 1 °C/min, the oxide is entirely reduced by the finaltemperature 675 °C; however, at the highest heating rate of 5°C/min, the reduction−carburization is not complete untilsometime after the plateau. For these reasons, we must use theFriedmann analysis if we are to consider the entire reductionMoO3 → Mo2C and identify trends. Figure 10 reveals that thereduction of orthorhombic α-MoO3 differs from that ofhexagonal h-MoO3-type oxides. The reduction of α-MoO3has an initially low Ea,app, which gradually increases to a value

Figure 9. (a) Extent of reduction (expressed as the calculated fraction of MoO2) vs T for α-MoO3, h-MoO3, and h-Mo0.92V0.08O3. Heating rates: 1°C/min (solid lines) and 5 °C/min (dashed lines). (b) TPRC profiles at 1 °C/min showing the reduction of MoO2 to Mo2C. The extent ofreduction is indicated on the y axis (MoO3 = 0%; MoO2 = 33%; Mo2C = 100%). (c) Consumption of CH4 (m/z = 15) vs T at 1 °C/min. (d)Evolution of CO (IR signal) vs T at 1 °C/min.



of ∼100 kJ/mol just before α approaches 0.33. This is followedby an inflection point at the beginning of carburization.The Friedmann analysis of hexagonal oxides shows a

different picture. Here, we see that the initial Ea,app is high(>100 kJ/mol) and decreases rapidly at around α = 0.1.Apparently, for the hexagonal oxides, there are variousoverlaying reactions in the reduction MoO3 → MoO2, whichis further evidenced by the observed o-Mo4O11 phase seen by insitu XRD (Supporting Information, Figure S4). As theconcentration of V increases, we can see that Ea,app of thecarburization step increases (Figure 10). In the case of h-MoO3,Ea,app of carburization is ∼100 kJ/mol close to that observedfor the orthorhombic oxide precursor. For h-Mo0.97V0.03O3 andh-Mo0.92V0.08O3, Ea,app increased to 130 and 150 kJ/mol,respectively. It is also notable that the activation energies forreduction (1) and carburization (2) become increasingly well-defined stepwise with the addition of V (see also Figure S6 inthe Supporting Information). In the case of the carburization ofh-Mo0.92V0.08O3, we can see that there are three mechanisticsteps and that the carburization is clearly separated from thereduction. A further subdivision of regime (2) can be made byconsidering the different onsets of CH4 consumption (Figure9c), which can be interpreted as the beginning of carboninsertion. Indeed, a large extent of reaction overlap (eq 6 vs eq7) exists, which may explain the differences in CO evolutionprofiles (Figure 9d). This supports the assumption that thechemistry, shape, and defect structure of the precursor play animportant role. As the overlap of many reactions is so large, noclear distinction, but rather a complex convolution pattern, isseen in Figure 10. However, the heating rate of 1 °C/min isvery slow, and much lower gradients may even change thereaction mechanism as then stabilization of intermediates withloss of their reactivity may inhibit the final product formation.

Although these trends in activation energies help to paint amechanistic picture, a considerable degree of fluctuation mustbe noted, which is of a systematic nature, meaning thatassumptions in the Friedman model are not fully transferable toour system. If we consider the absolute values of calculatedenergies, we must conclude that they are, in some cases,unrealistic (e.g., Ea,app = 0 kJ/mol in Figure 10c). Oneimportant reason may be the overlap of several reactionprocesses within the same temperature interval rather than astrictly separated step-by-step reaction cascade. A certainimprovement by choosing more complex fit procedures isreported elsewhere;24 the trends observed, however, remainuntouched. Also, it must be noted that additional degrees offreedom lower the significance of information derived. Clearly,the results suggest that the initial reduction step to MoO2involves a more complex mechanism than outlined in eqs 6 and7. From the model-free analyses of Ea,app, we could postulatethat this is in regards to an early onset of carburization ofamorphous material or the formation of less reducible species,which change the diffusional mechanism. This possibility issupported by the TPRC profiles, which show a distortion in thesigmoidal reduction curves for h-MoO3 at higher heating ratesfor β > 1 °C/min (Figure 9a).

Analysis of ex Situ X-ray Diffraction Patterns. Thecarbide derived from h-MoO3 exhibits a broadening of thecarbide peaks as compared to the material derived from α-MoO3. In the case of the V-containing oxides h-Mo1−xVxO3, itis observed that the carbide diffraction patterns are significantlybroadened (Figure 6), implying that the carbide crystallites aresmaller. Analysis of the ex situ XRD data shows a trend in thecrystallite sizes of the as-synthesized carbides that have beenestimated using LVol-IB (volume-weighted mean columnlengths based on integral breadth). The sizes vary from 6.3

Figure 10. Friedmann analysis of TPRC for (a) α-MoO3, (b) h-MoO3, (c) h-Mo0.97V0.03O3, and (d) h-Mo0.92V0.08O3. (1) and (2) indicate the twosteps of reduction−carburization.



nm for the reference carbide β-Mo2C(ref) derived from α-MoO3 to 3.3 nm for the most highly substituted carbide (Table2). From software analysis of the final XRD patterns (asexemplarly shown for an oxidic precursor in the SupportingInformation, Figure S2a), a close fit for β-Mo2C is achievedwith a hexagonal space group (P63/mmc) using a Pawleyrefinement with the addition of Mo metal as a minorityRietveld phase. In the case of the α-MoO3-derived carbide, thePawley indexed phase combined with a minority β-MoC1−xRietveld phase gives the most satisfactory fit. The assignment ofminority phases in these samples can be rationalized byconsidering that, in the case of molybdenum carbidesynthesized from orthorhombic MoO3, pretreatment of theoxide under H2 leads to bronze-like hydrogenation of thestructure, which is subsequently carburized to the less stablecubic α-MoC1−x.

49 This preconditioning seems to occur tosome extent during regular carburization of the orthorhombicoxide, leading to a minority phase. In the case of the hexagonaloxides, this minority phase is not observed.As the diffraction peaks of the carbides are extremely broad,

the cell parameters obtained from the diffraction patterns donot show a systematic trend as did the oxides, but it can be seen

that there is no evidence based on XRD of crystalline VxOy orVC. Further evidence for a homogeneous solid solution ispresented in a later section, as demonstrated by elementalmapping using energy filtered TEM.

Carbon Determination. CHN analysis of the carbidesshows an excess of C with respect to the theoretical valuescalculated for stoichiometric M2C (Table 2). The C excess isdue to the synthetic conditions applied in the in situ XRD cell,in which the materials are formed. These applied conditionswere derived by consideration of the ideal synthesis conditionsby Lee et al.3 combined with consideration of thediffractometer cell and combined analytics. The resultingdilution of the feed gas decreased the partial pressure of H2,resulting in significant formation of carbonaceous deposits overthe course of the synthesis.

Surface Area, Porosity, and Microstructure of theResulting Carbides. The samples exhibit relatively highsurface areas (Table 2), ranging from 71 m2/g for the referencecarbide and 80 m2/g for the undoped β-Mo2C up to amaximum of 107 m2/g for β-(Mo0.95V0.05)2C. This is in contrastto the observed LVol-IB values that show no systematic trend,but may be explained by analysis of the N2 sorption isotherms

Figure 11. (a) Comparison of N2 adsorption−desorption isotherms and (b) BJH pore size distributions.

Figure 12. SEM images of the carburized materials: (a) β-Mo2C(ref); (b) β-Mo2C; (c) β-(Mo0.97V0.03)2C; (d) β-(Mo0.95V0.05)2C; (e) β-(Mo0.92V0.08)2C; (f) β-(Mo0.89V0.11)2C.



in conjunction with the morphologies observed by SEM. Thefull adsorption−desorption isotherms indicate an increasinglymesoporous character of the carbide materials with increasingpore volume to a maximum of ∼0.19 cm3/g for β-(Mo0.92-V0.08)2C (Figure 11a). The shape of the N2 sorption isothermsis consistent with an irregular pore structure and the occurrenceof slit pores. This observation agrees with the morphologicalfeatures seen by SEM (Figure 12). Isotherms evolve from apurely type III adsorption isotherm with a type H3 hysteresisloop (polydisperse pore sizes, consistent with slit micropores)in the case of β-Mo2C to a type IV isotherm with a type H2hysteresis loop (tighter pore size distribution, more regularmesoporous character) in the case of β-(Mo0.92V0.08)2C. Thepeak in calculated surface area could be attributable to thecontributions of microporous and mesoporous character, whichis apparently maximized in sample β-(Mo0.95V0.05)2C.The mesoporosity evolves with increasing V content and

results in a material with an average pore diameter of 40−50 Å(Figure 11b). Such mesoporosity is highly relevant to the use ofmaterials in heterogeneous catalysis because it allows forreactants to be transported to, and products to be transportedfrom, the active sites of the catalyst.The SEM images of the as-synthesized carbides show an

interesting morphological trend. With increasing V content, it isobserved that the morphology evolves from that of anagglomerated-particulate material to a densely packed nano-fibrous material with a visible mesoporosity (Figure 12). Theapparent fibers propagate lengthwise along the c axis relative tothe original parent oxide crystals. To this extent, pseudo-morphism is evident in the preservation of the overallhexagonal prismatic structure observed in the parent oxidespecies.With increasing V content, the carbides present an evolved

mesoporosity, which originates from a smaller, more organizedmorphology composed of nanocrystalline carbide (Figure 12).The fiber-like nature becomes increasingly dense with a trendtoward a finer and more regular morphology. With reference tothe TPRC profiles (Figures 8 and 9), β-(Mo0.92V0.08)2C, whichexhibits the most uniform mesoporosity as seen in the N2physisorption isotherms, is characterized by the cleanestreduction profile. The trend is observed toward increasingmesoporosity with decreasing formation of intermediate o-Mo4O11. It is hypothesized that the kinetic interference of Mosuboxide formation is linked to sintering and aggregation ofparticles in what is otherwise described as the topotacticreduction of MoO3 to MoO2.

39 There are conflicting reports ofthe topotactic/pseudomorphic nature of this reduction;3,50,51

however, these are not incommensurable with mechanismsdiscussed in the literature, which are highly dependent on theapplied conditions and mass transport limitations.36,39,52,53

Characterization of the intermediately doped sample β-(Mo0.92V0.08)2C by TEM is consistent with the morphologyobserved by SEM. Figure 13 shows a high-resolution TEMimage of the high-aspect-ratio crystallites, which make up thefibrous bundles observed in the SEM (Figure 12). Many lath-like crystals, illustrated by the arrows, were found. Thecharacteristic crystal planes (101) and (011) of the hcpstructure were identified with their acute angle of 81°.Consistent with the elongated diffraction spots, the mismatchangles between the crystallites are small, indicating their originfrom one large grain. Such preferential orientation is inagreement with powder XRD analysis. The low-magnificationimage in Figure 14 is a micrograph of β-(Mo0.92V0.08)2C taken

in high-angle annular dark-field scanning TEM (HAADF-STEM), which shows a homogeneous distribution of Mo, V,and C throughout the observed region of the sample. Thepresence of O originates from the passivation procedure.Aggregation of V can be excluded from this analysis.Of particular interest in the TEM study was the observation

that the spot diffraction pattern of the parent polycrystalliteexhibits a circular pattern of elongated spots, indicating that thecoformed carbide particles have a preserved crystallographicorder and exhibit preferential orientation. This retention ofcrystallographic order is not observed in carbides derived fromundoped h-MoO3 or from the carburization of α-MoO3. Inprevious studies, a range of Mo-containing oxidic materials havebeen used as precursors to mono- and polymetallic carbides viaa reductive carburization. These include α-MoO3,

3 MoOxHy,49

and (NH4)6Mo7O24·4H2O,51 and mixed metal oxides of the

form M5O14 (M = Mo/V/Nb/Ta/W).12 Topotaxy is reportedin the case of the mixed metal oxide precursors, and also, in thecase of the reduced oxyhydride (MoOxHy); the reaction underCH4 proceeds topotactically to the cubic carbide phase MoC1−xwith no observed intermediate phases. Depending on thereaction conditions used, it is observed in many cases that thereaction products show pronounced pseudomorphism.

Carbide Synthesis in a Laboratory Reactor. The up-scaled synthesis of β-(Mo1−xVx)2C materials was carried outusing a series of oxide precursors analogous to that used for thein situ XRD study. A summary of the properties is given inTable 3. It can be seen from the specific surface areas of the as-synthesized carbides that a similar trend is observed for thecarbides synthesized in the rotary furnace with the lowest beingthe Mo-only sample. Notably, the maximum temperature used

Figure 13. (a) HRTEM image of β-(Mo0.92V0.08)2C. (b) Low-magnification TEM image. (c) Electron diffraction pattern appliedover a large area.

Figure 14. HAADF-STEM image of β-(Mo0.92V0.08)2C combined withEDX elemental maps for V, Mo, C, and O.



for the synthesis is 75 °C lower than that used in the in situXRD study. This was done to reduce carbon deposition duringthe later stages of synthesis, and by direct comparison withTable 2, it is demonstrated that the carbides exhibit astoichiometry much closer to Mo2C. The discrepancy betweenstoichiometry and measured composition likely originates fromthe passivation procedure, during which surface C is oxidized toCOx.SEM images (Figure 15) show very similar features to those

observed for the in situ study. In Figure 15c, the spherical

aggregates reflect the microstructure of the precipitatedcrystallite material of the precursor oxides (Figure 5). In thecase of the highest V loading, surface ridging can be seen inFigure 15d. In general, the mesostructure as determined by N2sorption and SEM reflects in every way the syntheses carriedout in small scale in the in situ XRD.

■ GENERAL DISCUSSIONIn comparing and contrasting the carburization of α-MoO3 andvarious h-(Mo,V)O3-type oxides, we observe a number ofsignificant differences. With respect to the reduction−carburization of α-MoO3, the measured TPRC curves alongwith the model-free kinetic analysis support a complexmechanism whereby increasing the reduction temperatureincreases the apparent activation energy. This phenomenonmay be explained as O becoming mobilized in the oxide latticeabove a critical temperature. Above this temperature, the latticemay assume locally more stable configurations through theformation of shear structures. The layered α-MoO3 (Figure 16)is initially reduced via facile transport of O from the interlayerspacings of the oxide, forming O vacancies, which lead to a

slipping of the planes and reoxidation of the vacancies byparallel terminal oxygen groups.Increasing the temperature drives the fast reconstruction of

the oxide to form a suboxide, which, depending upon reactionconditions (mass-transport limitations), may form a bulkcrystalline phase or, alternatively, is formed as a microdomainof corner-linked Mo octahedra at the reaction front. Thecalculated apparent activation energies in Figure 10 reflect this;as the conversion increases, the mechanism is changing, leadingto strongly interpenetrating kinetic regimes, which may accountfor a large degree of observed error.The TPRC of h-(Mo,V)O3-type oxides appears mechanisti-

cally different to that of the orthorhombic oxide. This originatesfrom the different structure that is a 3D oxidic lattice, in whichall oxygen atoms are connected to at least two Mo atoms(Figure 3). Consequently, the initial reduction step is observedto have a high apparent activation energy (90−100 kJ/mol),which is because the oxide lattice has no low-energy mechanismby which it may be reduced. Initial removal of O results indestabilization of the crystal structure. The temperature ofreduction increases with increasing V content, which reflectsthe increased intrinsic stability of the V-doped materials.Because of the presence of cations in the hexagonal channels, h-MoO3 is nonstoichiometric with respect to Mo in order tocompensate the charge in the crystal. By substituting Mo for V,the number of crystallographic metal vacancies in the structuredecreases, resulting in a more stable material. On the basis ofthis analysis, we suggest the following reaction cascade for thecarburization of orthorhombic and hexagonal MoO3 (Figure17). We assign MoO2 the role of a spectator, which crystallizesin competition with further reduction of defective MoIV species.Furthermore, the presence of Mo metal, at least on the surface,cannot be excluded. Different colors in Figure 17 indicate thesame reaction network for both precursors, but different qualityof the real structure in terms of particle size or orientationdefects. The deconvoluted steps in the reaction networkpreclude a clear quantitative description. The differences in the

Table 3. Physico-Chemical Properties of β-(Mo1−xVx)2C Materials Synthesized in Large Scale (TTPRC,max = 675 °C)

V/(V + Mo) (at %)a IDb Ctheor (wt %) Cacta (% w/w) dhkl

c (nm) SBET (m2/g)

0.00 8425 5.89 5.2 7.1 690.03 8478 5.97 5.2 6.4 1100.08 8447 6.01 5.1 5.5 1060.11 8486 6.11 5.3 97

aAs determined by EDX. bNonambiguous internal sample number to distinguish reproductions of sample preparation. cLVol-IB from XRDrefinement.

Figure 15. SEM micrographs of highly V-doped carbides: (a, b) β-(Mo0.92V0.08)2C and (c, d) β-(Mo0.89V0.11)2C.

Figure 16. View of α-MoO3 looking down the (100) plane at theinterlayer spacings.



kinetic analysis profiles (Figure 10) indicate, however, clearlythat the products are not formed with identical temporal detailsand hence will show different structural properties. Thecommon qualitative nature of the reaction network can bederived from the similarities of the overall shape of the plots inFigure 10.The carburization is seen to increase in activation energy

with the inclusion of V (Figure 10). This phenomenon can beexplained in two ways. The first is to reduce the mobility of Oin the sublattice, thereby increasing the energy required toreduce the oxide. The more oxophilic nature of V serves tocreate strongly bound VO moieties, which could kineticallyhinder the reduction process and the carbide formation.Another explanation is that the reduced Mo/V oxide lackssufficient metallic character to decompose CH4 and therebyincorporate carbon. It is well-known that MoO2 is a metallicoxide, which has electron density at the Fermi edge. For acarbide to be formed, the reacting species must be metallic. Ineither case, we observed that, in the case of h-Mo0.92V0.08O3, thereduction step MoO3 →MoO2 is completed entirely before thefirst carbide is formed. This is supported by the TPRC profiles(Figure 9), which show that, in the V-free oxides, CH4 isconsumed at an earlier stage of the reaction.

■ SUMMARYReductive carburization of a mixed metal oxide based onhexagonal MoO3 results in bimetallic carbides with a novelhigh-aspect-ratio mesostructure and ordered crystallographicnature. The Mo/V precursor oxides were synthesized by asimple precipitation reaction, and the carburization reactionwas monitored using the in situ powder XRD technique. Thekinetics of reduction for the h-(Mo,V)O3 precursor oxides wereinvestigated and contrasted to α-MoO3, yielding insight intothe mechanism of reduction. The observed TPRC profilessuggest that the mobility of atomic O in the lattice is affected bythe inclusion of V as a dopant. It was observed that V inclusionresulted in a sharper, single reduction peak that was correlatedwith a more uniform evolved mesoporosity in the productcarbide.The use of multimetal oxides as precursors for carbides is a

largely unexplored area of transition-metal carbide chemistry. Inparticular, it offers interesting possibilities to produce tailoredcarbides with novel material properties, higher surface areas,and/or improved catalytic selectivity and performance. Thesegoals can be achieved either by doping of bulk carbides withmore active metals or by choosing precursor oxides with

structural properties that influence the carbide product. Therelatively high surface area of this material and the inclusion ofV atoms in the structure provide a basis for the investigation ofa number of industrially useful reactions.In this article, we also present a methodology for the up-

scaled synthesis and handling of these high-surface-area air-sensitive materials that is all-important in view of anyapplication. This technique allows using the materials directlyin catalytic reactions without pretreatment, providing theopportunity to characterize the materials extensively prior toreaction. This contributes to our understanding of the surface−structure relations with catalytic activity.

■ ASSOCIATED CONTENT*S Supporting InformationDetails of XRD analysis (PDF). This material is available free ofcharge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: [email protected].

Present Addresses†T.C.: Clariant Int. AG, Waldheimer Str. 13, 83052 Heufeld,Germany.‡W.Z.: Department of Energy Conversion and Storage,Technical University of Denmark, Frederiksborgvej 399, 4000Roskilde, Denmark.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTST.C. gratefully acknowledges support by the International MaxPlanck Research School (IMPRS) Complex Surfaces in MaterialsSciences. Financial support by the Federal Ministry of Educationand Research (BMBF) within the CarboKat project (FKZ03X0204C) of the Inno.CNT alliance is gratefully acknowl-edged. The authors thank G. Lorentz, G. Weinberg, F.Girgsdies, M.E. Schuster, and A. Tarasov for assistance onexperiments and scientific discussion.

■ REFERENCES(1) Boehm, H. Electrochim. Acta 1970, 15, 1273.(2) Levy, R. B.; Boudart, M. Science 1973, 181, 547.(3) Lee, J. S.; Oyama, S. T.; Boudart, M. J. Catal. 1987, 106, 125.(4) Leclercq, L.; Provost, M.; Pastor, H.; Leclercq, G. J. Catal. 1989,117, 384.(5) Ranhotra, G. S.; Bell, A. T.; Reimer, J. A. J. Catal. 1987, 108, 40.(6) York, A. P. E.; Claridge, J. B.; Brungs, A. J.; Tsang, S. C.; Green,M. L. H. Chem. Commun. 1997, 39.(7) Dolce, G. M.; Savage, P. E.; Thompson, L. T. Energy Fuels 1997,11, 668.(8) Alconchel, S.; Sapina, F.; Martinez, E. Dalton Trans. 2004, 2463.(9) Wang, X.-H.; Zhang, M.-H.; Li, W.; Tao, K.-Y. Catal. Today 2008,131, 111.(10) Oyama, S. T. J. Solid State Chem. 1992, 96, 442.(11) Yu, C. C.; Ramanathan, S.; Dhandapani, B.; Chen, J. G.; Oyama,S. T. J. Phys. Chem. B 1997, 101, 512.(12) Oyama, S. T.; Yu, C. C.; Ramanathan, S. J. Catal. 1999, 184,535.(13) Bastos, L.; Monteiro, W.; Zacharias, M.; da Cruz, G.; Rodrigues,J. Catal. Lett. 2008, 120, 48.(14) Xiao, T. C.; York, A. P. E.; Williams, V. C.; Al-Megren, H.;Hanif, A.; Zhou, X. Y.; Green, M. L. H. Chem. Mater. 2000, 12, 3896.

Figure 17. Suggested cascade of reaction steps during carburization ofMoO3.



http://pubs.acs.org

mailto:[email protected]

(15) Xiao, T.; York, A. P. E.; Coleman, K. S.; Claridge, J. B.; Sloan, J.;Charnock, J.; Green, M. L. H. J. Mater. Chem. 2001, 11, 3094.(16) Xiao, T. C.; Wang, H. T.; Da, J. W.; Coleman, K. S.; Green, M.L. H. J. Catal. 2002, 211, 183.(17) Hanif, A.; Xiao, T. C.; York, A. P. E.; Sloan, J.; Green, M. L. H.Chem. Mater. 2002, 14, 1009.(18) Wang, H. M.; Wang, X. H.; Zhang, M. H.; Du, X. Y.; Li, W.;Tao, K. Y. Chem. Mater. 2007, 19, 1801.(19) Li, X.; Ma, D.; Chen, L.; Bao, X. Catal. Lett. 2007, 116, 63.(20) Yao, Z. W. J. Alloys Compd. 2009, 475, L38.(21) Frank, B.; Friedel, K.; Girgsdies, F.; Huang, X.; Schlogl, R.;Trunschke, A. ChemCatChem 2013, DOI: 10.1002/cctc.201300010.(22) Mougin, O.; Dubois, J.-L.; Mathieu, F.; Rousset, A. J. Solid StateChem. 2000, 152, 353.(23) FHI internal sample numbers in parenthesis: α-MoO3 (5201),h-MoO3 (6600), h-Mo0.97V0.03O3 (6702), h-Mo0.95V0.05O3 (6271), h-Mo0.92V0.08O3 (6697), h-Mo0.89V0.11O3 (6601), β-Mo2C(ref) (5661),β-Mo2C (6713), β-(Mo0.97V0.03)2C (6708), β-(Mo0.95V0.05)2C (6704),β-(Mo0.92V0.08)2C (6703), β-(Mo0.89V0.11)2C (6701); large-scale syn-thesis: β-Mo2C (8425), β-(Mo0.97V0.03)2C (8478), β-(Mo0.92V0.08)2C(8447), β-(Mo0.89V0.11)2C (8486).(24) Cotter, T. The reducibility of mixed Mo/V oxides to carbides andtheir reactivity in the activation of propane. Ph.D. Thesis, TechnischeUniversitat Berlin, Berlin, Germany, 2011.(25) Friedman, H. L. J. Polym. Sci., Part C 1964, 183.(26) DIFFRACplus TOPAS: Profile and Structure Analysis Software,V3.0; Bruker AXS: Karlsruhe, Germany, 1999.(27) Balzar, D. In Defect and Microstructure Analysis by Diffraction;Snyder, R. L., Fiala, J., Bunge, H. J., Eds.; IUCR Monographs onCrystallography; Oxford University Press: Oxford, U.K., 1999; Vol. 10.(28) Olenkova, I. P.; Plyasova, L. M.; Kirik, S. D. React. Kinet. Catal.Lett. 1981, 16, 81.(29) Plyasova, L. M.; Kirik, S. D.; Olen’kova, I. P. J. Struct. Chem.1983, 23, 590.(30) Caiger, N. A.; Crouch-Baker, S.; Dickens, P. G.; James, G. S. J.Solid State Chem. 1987, 67, 369.(31) Darriet, B.; Galy, J. J. Solid State Chem. 1973, 8, 189.(32) Dupont, L.; Larcher, D.; Portemer, F.; Figlarz, M. J. Solid StateChem. 1996, 121, 339.(33) Dupont, L.; Larcher, D.; Touboul, M. J. Solid State Chem. 1999,143, 41.(34) Denton, A. R.; Ashcroft, N. W. Phys. Rev. A 1991, 43, 3161.(35) Choi, J.-S.; Bugli, G.; Djega-Mariadassou, G. J. Catal. 2000, 193,238.(36) Ressler, T.; Jentoft, R. E.; Wienold, J.; Gunter, M. M.; Timpe, O.J. Phys. Chem. B 2000, 104, 6360.(37) Lalik, E. Catal. Today 2011, 169, 85.(38) Lalik, E.; David, W. I. F.; Barnes, P.; Turner, J. F. C. J. Phys.Chem. B 2001, 105, 9153.(39) Lalik, E. Catal. Today 2011, 169, 85.(40) Wang, D.; Su, D. S.; Schlogl, R. Z. Anorg. Allg. Chem. 2004, 630,1007.(41) Torres-García, E.; Rodríguez-Gattorno, G.; Ascencio, J. A.;Aleman-Vazquez, L. O.; Cano-Domínguez, J. L.; Martanez-Hernandez,A.; Santiago-Jacinto, P. J. Phys. Chem. B 2005, 109, 17518.(42) Rodriguez-Gattorno, G.; Martinez-Hernandez, A.; Aleman-Vazquez, L. O.; Torres-Garcia, E. Appl. Catal., A 2007, 321, 117.(43) Watt, G. W.; Davies, D. D. J. Am. Chem. Soc. 1948, 70, 3751.(44) Sakagami, H.; Asano, Y.; Ohno, T.; Takahashi, N.; Itoh, H.;Matsuda, T. Appl. Catal., A 2006, 297, 189.(45) Ohno, T.; Li, Z.; Sakai, N.; Sakagami, H.; Takahashi, N.;Matsuda, T. Appl. Catal., A 2010, 389, 52.(46) McIntyre, N. S.; Johnston, D. D.; Coatsworth, L. L.; Davidson,R. D.; Brown, J. R. Surf. Interface Anal. 1990, 15, 265.(47) Yamada, M.; Yasumaru, J.; Houalla, M.; Hercules, D. M. J. Phys.Chem. 1991, 95, 7037.(48) Claridge, J. B.; York, A. P. E.; Brungs, A. J.; Marquez-Alvarez, C.;Sloan, J.; Tsang, S. C.; Green, M. L. H. J. Catal. 1998, 180, 85.

(49) Bouchy, C.; Pham-Huu, C.; Ledoux, M. J. J. Mol. Catal. A 2000,162, 317.(50) Bouchy, C.; Pham-Huu, C.; Heinrich, B.; Chaumont, C.;Ledoux, M. J. J. Catal. 2000, 190, 92.(51) Patt, J.; Moon, D. J.; Phillips, C.; Thompson, L. Catal. Lett.2000, 65, 193.(52) Ressler, T.; Timpe, O.; Neisius, T.; Find, J.; Mestl, G.; Dieterle,M.; Schlogl, R. J. Catal. 2000, 191, 75.(53) Ressler, T.; Wienold, J.; Jentoft, R. E. Solid State Ionics 2001,141−142, 243.(54) Brungs, A. J.; York, A. P. E.; Green, M. L. H. Catal. Lett. 1999,57, 65.(55) Kim, B.-S.; Kim, E.-Y.; Jeon, H.-S.; Lee, H.-I.; Lee, J.-C. Mater.Trans. 2008, 49, 2147.



Documents

The Impact of V Doping on the Carbothermal Synthesis of Mesoporous Mo Carbides