Synthesis of tungsten carbides by temperature-programmed reaction with CH4-H2 mixtures. Influence of the CH4 and hydrogen content in the carburizing mixture

Article abstract of DOI:10.1006/jssc.2000.8859

The influence of the composition of a carburizing CH₄-H₂ gas mixture on the process of reduction-carburization over WO₃ has been studied. Bulk tungsten carbide synthesis has been carried out from WO₃ in different CH₄-H₂ mixtures (CH₄-H₂ = 1/1-3/1; CH₄-N₂ 1/1; pure CH₄) at atmospheric pressure by temperature-programmed reduction-carburization (TPRC). The composition of the reaction products has been monitored and quantified by gas chromatography analysis (GCA) and the results have been compared to those obtained for a reference sample WC20 (CH₄-H₂ = 1/4). The solids have been characterized by elemental analysis, XRD, XPS, and BET surface area measurements. The overall process is complex. Considering first the reduction, both H₂ and CH₄ act as oxides reducing agents and are converted respectively into H₂O, CO, and to a less extent CO₂. If the reduction steps follow the same sequence observed under pure H₂, WO₃ → W₂₀O₅₈ → WO₂ → W, with the strong difference that W metal is detected only at the surface to be rapidly carburized, the overall reduction process can be accomplished under CH₄-H₂ mixtures at temperatures all the lower than P(CH4)/P(H2) increases. Prereduction of WO₃ into bulk WO₂ allows an easier reduction in practically pure CH₄ (95% (v/v) CH₄-H₂) as reduction with CH₄ increases the rate of the WO₂ → W transformation. Studies of the carburization suggest that CH₄ decomposes on a metallic surface into C (or CH(x)) species before bulk WO₂ reduction followed by surface carburization. Then carbon diffuses into the bulk of the solid to give first α-W₂C whose formation occurs rapidly. α-W₂C transformation into WC is slower and seems to be very much influenced by the ratio P(CH4)/P(H2) which controls the rate of carbon deposit at the surface of the solid. The best surface area carbide of 27 m².g^-1 consisting of a core of α-W₂C covered with α-WC has been obtained by using WO₂ as starting material. (C) 2000 Academic Press.

Full text of DOI:10.1006/jssc.2000.8859

Journal of Solid State Chemistry 154, 412}426 (2000)
doi:10.1006/jssc.2000.8859, available online at http://www.idealibrary.com on
Synthesis of Tungsten Carbides by Temperature-Programmed
Reaction with CH `H Mixtures. Influence of the CH and
4
2
4
Hydrogen Content in the Carburizing Mixture
J.-M. Giraudon,*ꢀꢁ P. Devassine,* J.-F. Lamonier,- L. Delannoy,* L. Leclercq,* and G. Leclercq*
*
Laboratoire de Catalyse Homog e% ne et H e& t e& rog e% ne, UPRESA no. 8010, Universit e& des Sciences et Technologies de Lille, 59655 Villeneuve d+Ascq C e& dex,
France; and -Laboratoire de Catalyse et Environnement, M.R.E.I.D., 145 Avenue Maurice Schumann, 59140 Dunkerque Cedex, France
Received November 23, 1999; in revised form June 14, 2000; accepted July 13, 2000; published online September 30, 2000
INTRODUCTION
The in6uence of the composition of a carburizing CH
mixture on the process of reduction+carburization over WO
4
+H
2
3
gas
has
Tungsten carbides are interesting catalytic materials
since they are active for several reactions that also occur on
group VIII metals (1). But, as previously pointed out by
M. Boudart et al., their catalytic application requires the
synthesis of materials with high speci"c surface areas free
from pollutants such as polymeric carbon or oxygen (2).
One of the major routes of synthesis for high surface area
tungsten carbides has been developed by Boudart and co-
been studied. Bulk tungsten carbide synthesis has been carried
out from WO in di4erent CH +H mixtures (CH +H
/1+3/1; CH +N ؍ 1/1; pure CH ) at atmospheric pressure by
temperature-programmed reduction+carburization (TPRC).
The composition of the reaction products has been monitored and
quanti5ed by gas chromatography analysis (GCA) and the re-
sults have been compared to those obtained for a reference
3
4
2
4
2
؍
1
4
2
4
sample WC20 (CH
4
+H ؍ 1/4). The solids have been character-
2
workers (2). Their method consists of exposing WO to
ized by elemental analysis, XRD, XPS, and BET surface area
ꢂ
measurements. The overall process is complex. Considering 5rst a reducing}carburizing mixture while heating with a linear
the reduction, both H
are converted respectively into H
CO . If the reduction steps follow the same sequence observed The use of an oxide was believed to by-pass the bulk
under pure H
2
and CH
4
act as oxides reducing agents and temperature program. This method has further been ex-
2
O, CO, and to a less extent tended to the synthesis of other carbides by Oyama (3, 4).
2
2
, WO
3
PW20
O
58PWO PW, with the strong
2
metallic state which is the most prone to sinter (lowest
Tammann temperature, which is 1241 K for W compared to
di4erence that W metal is detected only at the surface to be
rapidly carburized, the overall reduction process can be accomp-
2
744 K for WC) (4). By monitoring changes in the exit gas
4
lished under CH +H
2
mixtures at temperatures all the lower than
into bulk WO allows
(95% (v/v) CH +H
increases the rate of the WO PW
transformation. Studies of the carburization suggest that CH
decomposes on a metallic surface into C (or CH ) species before
bulk WO reduction followed by surface carburization. Then
carbon di4uses into the bulk of the solid to give 5rst a-W
whose formation occurs rapidly. a-W C transformation into WC
composition by gas chromatography it was shown that the
transformation of MoO into an hexagonal Mo C with
P
CH /P
H
increases. Prereduction of WO
3
2
4
2
an easier reduction in practically pure CH
as reduction with CH
4
4
2
)
ꢂ
ꢃ
a CH }H mixture occurred in two steps. In the "rst one
4
2
ꢄ
ꢃ
MoO was reduced into MoO which was clearly identi"ed
4
ꢂ
ꢃ
by XRD measurements. In the second step there was further
water formation and methane consumption. The authors
were led to the conclusion that simultaneous reduction and
carburization occurred (5). Previous works have also been
x
2
2
C
2
is slower and seems to be very much in6uenced by the ratio performed by Leclercq et al. (6}8) in order to elucidate the
P
CH /P
H
2
which controls the rate of carbon deposit at the surface complex transformation of tungsten as metal or oxides into
4
2
؊1
of the solid. The best surface area carbide of 27 m .g consisting the resulting metal carbide by temperature-programmed
of a core of a-W C covered with a-WC has been obtained by experiments (TPEs). From these previous results, it is clear
using WO
2
2
as starting material. ( 2000 Academic Press
that, under the experimental conditions used, carburization
occurs only when there is W metal or a suboxide with a very
low O content at the surface of the particle. Moreover,
carburization of W takes place in two steps, with "rst the
Key Words: tungsten carbide; carbide catalysts; reduction;
carburization.
formation of W C and then its carburization into WC.
ꢃ
Hence, it can be imagined that a way to W carbides with
ꢁ
To whom correspondence should be addressed. Fax:#33 3.20.65.61.
E-mail: Jean-Marc. Giraudon@univ-lille1.fr.
high surface areas would be to form W at the solid surface at
4
12
0
022-4596/00 $35.00
Copyright ( 2000 by Academic Press
All rights of reproduction in any form reserved.

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
413
ꢄ
ꢃ
ꢂ
such a temperature that it can be immediately carburized as binding energies (B.E.s) were referenced either to the C1s
soon as it is formed in order to avoid metal sintering. peak of polymeric carbon at 285 eV or to the W4f peak of
ꢅꢆꢃ
Consequently, for optimizing the W carbide surface areas it WO at 38.2 eV.
is of primary importance to determine what parameters can
ꢂ
Elemental analyses (EAs) were performed at the &&Service
in#uence the reduction of WO into W, and the carburiz- Central de Microanalyse'' of the CNRS (Vernaison, France).
ꢂ
ation of W.
Carbon was transformed into CO in a puri"ed oxygen
ꢃ
The purpose of this paper is to improve our understand- #ow, which was quanti"ed by passing through infrared cells.
ing of the mechanism of reduction and carburization of Oxygen was quanti"ed by the determination of carbon
WO by each component of the CH }H mixture in order monoxide resulting from the fusion of the sample. W was
ꢂ
ꢄ
ꢃ
to synthesize tungsten carbides at temperatures as low as quanti"ed by plasma emission spectroscopy after fusion of
possible with the aim of developing high surface areas with- the samples in a NaOH#KNO mixture. Surface areas
ꢂ
out carbon deposit.
were measured with a sorptometer (Quantasorb Jr.) using
Di!erent syntheses of tungsten carbides have been carried single-point BET surface area determination.
out in CH }H mixtures by varying the ratios of the react-
ꢄ
ꢃ
ants. The bulk of the samples has been characterized by
XRD and the related surface by XPS.
Nomenclature
The samples were labeled WCX (where X is the volumet-
ric percentage of methane in the reactive CH }H gas mix-
EXPERIMENTAL SECTION
ꢄ
ꢃ
ture); the su$x N2 is added when dilution of CH is carried
ꢄ
The set-up used for the di!erent experiments was pre- out in nitrogen. WCPX referenced to experiments stopped
viously described (9). To sum up, the syntheses were per- just after the strong peak (P) of methane consumption.
formed in a plug #ow reactor (10 cm long and 2 cm wide) The pre"x IN was added to characterize intermediates of
equipped with a thermowell where a thermocouple was carburization. WCX> (>"A, B, C, D) referred to samples
settled. Flow rates were regulated by membrane #ow regula- submitted to a prereduction of WO into bulk WO or
ꢂ
ꢃ
tors (Brooks 8802) and measured with a soap bubble #ow- WO }W prior carburization in 20% or 95% (v/v) CH in
ꢃ
ꢄ
meter. The composition of the e%uents at the outlet of the H .
reactor was periodically determined by gas chromatogra-
ꢃ
phy using a Shimadzu chromatograph GC9A equipped
with a thermal conductivity detector (TCD). Analyses were
performed every 10 or 20 min with helium as a carrier gas.
RESULTS AND DISCUSSION
1
. Reduction}Carburization of WO3
Figures 1a}1c show three temperature-programmed re-
Products were separated at 1103C in a column (3 m long
and 1/8 in. wide) "lled with carbosphere molecular sieves. duction}carburizations (TPRCs) of WO (Fluka, 3 mꢃ.g\ꢁ)
ꢂ
Peak areas were measured with an integrator (Spectra Phys- performed in CH }H mixtures at atmospheric pressure.
ꢄ
ꢃ
ics 4270).
They are characterized by a set of similar experimental
Hydrogen (Air Liquide, U quality) was puri"ed by #ow- parameters, i.e., an oxide weight of 2 g, a total gas #ow rate
ing through a Chrompack oxygen clean "lter and through of 30 l.h\ꢁ, a heating rate of 13C.min\ꢁ, and an isotherm
1
9
3X molecular sieves. Methane (Air Liquide, N45 quality, duration of 2 h at the "nal temperature of 8003C. On the
9.995%) was puri"ed by passing through 4A molecular other hand they di!er by the methane volumetric percent-
sieves. Nitrogen (Air Liquide, U quality) was passed age which is respectively equal to 50, 75, and 100 in experi-
through copper in short bursts at 3503C in order to remove ments A, B, and C. For all these samples and the others
oxygen impurities and then through 13X molecular sieves. quoted below the synthesis conditions and physical charac-
The passivation mixture (2% O /N , Alphagaz) was used terizations are reported in Tables 1a and 1b.
ꢃ
ꢃ
without further puri"cation.
In order to study the in#uence of the CH partial pressure
ꢄ
The precursor used to synthesize tungsten carbides in this irrespective of that of H , a TPRC under conditions similar
ꢃ
study was monoclinic WO (Fluka, 99.95%, S " to those stated above but in a CH }N 1/1 mixture (experi-
ꢂ
#2
ꢄ
ꢃ
3
mꢃ. g\ꢁ). Details of the di!erent programmed reduction} ment D, compound WC50N2) (Fig. 1d) has been carried
carburizations of WO are given in Table 1a.
The bulk structure of the samples was determined by
out.
For comparison we have added the results of a previous
ꢂ
powder X-ray di!raction (XRD) using a di!ractometer experiment (6) (sample WC1 labeled here WC20, experiment
(
Siemens D 5000) (CuKa"0.154178 nm). XRD patterns E) carried out in a 20% (v/v) CH }H mixture (Fig. 2). It is
ꢄ ꢃ
were ascribed using the JCPDS data base. X-ray photo- noteworthy that some main experimental parameters, i.e.,
electron spectra were obtained with an AEI ES 200B Kratos the oxide precursor weight (3.5 g), the temperature gradient
electron spectrometer equipped with an aluminum anode (0.833C.min\ꢁ), the gas #ow rate (10 l.h\ꢁ), and the "nal
(hl" 1486.6 eV) or with a VG ESCALAB 220XL. The temperature of treatment (9003C), di!er from those given

4
14
GIRAUDON ET AL.
TABLE 1a
Experimental Conditions of Preparation of Bulk Tungsten Carbides at Atmospheric Pressure
Preparation?
Passivation
F@
¹
GA
Duration
(h)
Cooling
to RT
F
Duration
(h)
Catalyst
WC20
(l.h\ꢁ)
(3C)
(3C.h\ꢁ)
(l.h\ꢁ)
% O
ꢃ
TCB"10
TAC"8
TC"30
TA"10
TC"30
TA"10
TC"30
TA"10
TC"30
TA"10
TC"30
TC"30
TC"30
TC"30
TC"30
TRD"10
TC"20
TR"10
TC"20
TR"10
TC"20
TR"10
TC"20
900
800
800
800
800
800
800
800
800
800
610
550
700
680
670
635
750
520
700
520
750
520
750
50
No
60
No
60
No
60
No
60
No
60
60
60
60
60
60
60
120
60
120
60
120
60
10
5
2
2
2
2
2
2
2
CH #H
10
15
15
15
15
1
15
5
ꢄ
ꢃ
ꢃ
ꢃ
H
ꢃ
WC50
CH #H
0.4
0.4
0.4
0.4
ꢄ
H
ꢃ
WC75
CH #H
3.5
5
ꢄ
H
CH
ꢃ
WC100
WC50N2
ꢄ
H
ꢃ
CH #H
5
ꢄ
ꢃ
2
H
ꢃ
INWC50
INWC66
WCP50
WCP75
WCP85
WC95A
*
*
*
*
*
*
10
5
10
4.5
27
4
CH #H
3
3
3
3
3
3
2
2
2
2
2
2
2
2.5
2
2
2
ꢄ
ꢃ
ꢃ
ꢃ
ꢃ
ꢃ
CH #H
ꢄ
CH #H
ꢄ
CH #H
ꢄ
CH #H
ꢄ
H #N
2
ꢃ
ꢃ
CH #H
ꢄ
ꢃ
ꢃ
ꢃ
ꢃ
WC95B
WC95C
WC20D
H
3
3
3
2
2
2
2.
2.
2.
ꢃ
CH #H
ꢄ
H
ꢃ
CH #H
ꢄ
H
ꢃ
CH #H
ꢄ
10
Note. All samples were dried under N to 400}5003C for circa 5 h and cooled to RT before any treatment.
ꢃ
?
WO (2 g) is the precursor for all samples except WC20 (ammonium metatungstate, 3 g).
Flow rate (l.h\ꢁ).
ꢂ
@
A Gradient of temperature (3C.h\ꢁ).
B Treatment of carburization.
C Treatment of activation in #owing H .
ꢃ
D Treatment of reduction.
TABLE 1b
Bulk and Surface (XPS) Compositions and Speci5c Surface Areas of Bulk Tungsten Carbides
Composition
Temperature of
carburization/3C (duration/h)
From elemental
analysis
BET surface
areas/mꢃg\ꢁ
Catalysts
From XPS
WC
WC20?
900(8)
800(2)
800(2)
800(2)
800(2)
606
560
702
678
670
WC
WC
WC
WC
WC
O
ꢇꢈꢉꢇ ꢇꢈꢇꢃ
O
O
10
19
3
9
6
ꢂꢈꢄ ꢇꢈꢃꢂ
WC50
WC
WC
WC
WC
O
ꢁꢈꢂ ꢁꢈꢂ
O
ꢇꢈꢊꢃ ꢇꢈꢅꢇ
O
WC75
ꢇꢈꢋꢌ ꢇꢈꢅꢉ
O
ꢁꢈꢁ ꢁꢈꢅ
O
WC100
WC50N2
INWC50
INWC66
WCP50
WCP75
WCP85
WC95A
WC95B
WC95C
WC20D
ꢇꢈꢋꢅ ꢇꢈꢂꢇ
O
ꢁꢈꢃ ꢁꢈꢋ
O
ꢇꢈꢋꢁ ꢇꢈꢃꢇ
ꢁꢈꢁ ꢁꢈꢌ
O
WC
O
ꢇꢈꢁ ꢁꢈꢅꢄ
O
WC
WC
WC
WC
WC
11
4
ꢃꢈꢁꢄ ꢇꢈꢋꢁ
O
WC
ꢇꢈꢇꢁ ꢃꢈꢂꢁ
ꢇꢈꢃꢉ ꢃꢈꢋꢇ
O
WC
WC
WC
O
ꢇꢈꢊꢅ ꢇꢈꢄꢂ
O
14
19
21
*
20
27
23
ꢃꢈꢇꢅ ꢇꢈꢄꢁ
O
ꢁ
ꢈꢁꢋ ꢇꢈꢁꢋ
ꢃꢈꢂꢅ ꢁꢈꢇꢇ
O
O
ꢁ
ꢈꢄꢇ ꢇꢈꢇꢅꢋ
ꢃꢈꢌꢋ ꢇꢈꢄꢅ
*
750(10)
700(10)
750(27)
750(10)
*
WC
O
*
ꢁ
ꢈꢄꢊ ꢇꢈꢁ
*
*
WC
WC
O
O
ꢋꢈꢋꢅ ꢇꢈꢃꢂ
ꢃꢈꢂ ꢇꢈꢃꢁ
?
From Ref. (6).

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
415
ꢄ
ꢃ
ꢂ
FIG. 1. Variations of the areas of CH , H O, and CO chromatographic peaks (normalized, taking into account the relative molar response of the
ꢄ
ꢃ
TCD) during the reduction}carburizations of WO in the di!erent CH }H mixtures (603C.h\ꢁ, 2 g of tungsten trioxide). (a) WC50; (b) WC75; (c)
ꢂ
ꢄ
ꢃ
WC50N2; (d) WC100. (᭡) CH ; (᭿) H O; (᭜) CO.
ꢄ
ꢃ
above. Table 2 reports the temperatures corresponding to
the various events occurring during the experiments, i.e., reduction by hydrogen proceeds by three successive steps
H
partial pressures. They have clearly shown that the
ꢃ
CH consumption, and CO and H O productions. It should which seem to correspond respectively to WO PW O
ꢄ
ꢃ
ꢃ
ꢂ
ꢃꢇ ꢅꢊ
be mentioned that the trace of CO which is similar to that (or similar formula)PWO PW (see for example the curve
ꢃ
of CO but with a signi"cantly lower extent has been omitted in Fig. 4 reprinted from Ref. (6)). The separation between the
for the sake of clarity. di!erent steps is increasingly pronounced as hydrogen pres-
The quanti"cation of the methane consumption and of sure decreases. A decrease of P ꢃ lowers the rates of these
&
the oxygen removal has been obtained from gas chromato- three steps but di!erently since the second step (formation
graphic analyses (6).
of WO ) is more in#uenced than the "rst one (formation of
ꢃ
ꢃꢇ ꢅꢊ
The amount of oxygen (O/W) removed as CO (CO/W) W O ) by a change in P , the third step (reduction of
&
ꢃ
and H O (H O/W) during the reduction of the sample WO into W) being itself still more in#uenced by P . As
ꢃ
ꢃ
ꢃ
&
ꢃ
together with that of carbon held by the solid (C/W) are a consequence, the value of the O/W ratio at the minimum
shown versus time and temperature in Fig. 3.
of water production (see Table 3), which probably corres-
ponds to the disappearance of WO (or W O ), is higher
ꢂ
ꢃꢇ ꢅꢊ
(
a) Tungsten oxides reduction. Let us "rst consider the than 1 for a H pressure of 1 atm which shows that the
ꢃ
reduction of tungsten oxides. Obviously, not only hydrogen reduction of WO into W starts well before all W O has
but also methane take part in the reduction of tungsten been reduced into WO .
ꢃ
ꢃꢇ ꢅꢊ
ꢃ
oxides since not only H O but also CO and traces of CO
Considering now the reduction of WO in CH }H
mixtures (Fig. 1, curves a and b and Fig. 2), it can be seen
ꢃ
ꢃ
ꢂ
ꢄ
ꢃ
are formed.
Let us "rst recall the results of a previous temperature- that water production is noticeably in#uenced by hydrogen
programmed reduction (TPR) of WO (6) at various pressure since it starts at higher temperatures when P ꢃ
ꢂ
&

4
16
GIRAUDON ET AL.
FIG. 2. Variations of the areas of CH , H O, and CO chromatographic peaks (normalized, taking into account the relative molar response of the
ꢄ
ꢃ
TCD) during the temperature-programmed reduction}carburization of ammonium metatungstate (503C.h\ꢁ, 3.51 g of ammonium metatungstate) in 20%
CH }H (6). (᭡) CH ; (᭿) H O; (᭜) CO.
ꢄ
ꢃ
ꢄ
ꢃ
decreases (330, 360, and 5303C for a H pressure of respec- production occurs in, at least, two steps (the initial "rst step
ꢃ
tively 0.8, 0.5, and 0.25 atm). It also starts before CO forma- seen in Fig. 4 being mixed with the second step, we will now
tion (whatever the H and the CH pressures, at least in the refer to these two undistinguishable steps as the "rst one).
ꢃ
ꢄ
pressure range studied here) which indicates that, at low The shapes of the di!erent pro"les of H O formation reveal
ꢃ
temperature, the reduction of WO by H is favored. Water signi"cant changes. Thus the two peaks of water production
ꢂ
ꢃ
are well resolved for H pressures of 0.8 or 0.5 atm whereas
this is not the case for a hydrogen pressure of 0.25 atm.
ꢃ
TABLE 2
Temperatures Relative to the CH
O, CO Productions
Furthermore, the relative extent of water production during
4
Consumption and to the
H
2
the second step decreases noticeably as the ratio P /P is
&
ꢃ
!&
ꢄ
lowered, this second step of water production appearing
WC20 WC50 WC75 WC100 WC50N2
only as a weak shoulder when P /P "1/3 (WC75).
&
ꢃ
!&
ꢄ
The onset of CO formation occurs during the "rst peak of
water production and seems to be little in#uenced by
P /P in contrast to the amount of CO produced, which
P
P
P
(R"CH #H ) 1 atm
1 atm 1 atm 1 atm
0.5 atm
0
ꢄ
ꢃ
/P
0.2
0.8
0.5
0.5
0.75
0.25
1
0
1
0
!
&
ꢄ
0
ꢃ/P
&
ꢃ
!&
ꢄ
&
0
increases as P /P decreases (Table 4). This is a result of
!&
the competition between H and CH for the reduction of
CH consumption
&
ꢃ
ꢄ
ꢄ
h
*
550
680
810
550
640
730
580
655
690
600
755
780
770
780
790
ꢃ
ꢄ
h
tungsten oxides. The CO production starts for di!erent
ꢀ
h&
!ꢁ
global degrees of WO reduction (ratios H O/W), as can be
ꢂ
ꢃ
seen in Table 4. It is noticeable that the trace of CO produc-
tion is very di!erent from that of water, since a single peak is
observed even if its shape varies with the gas mixture com-
position. Moreover, when the step WO PW O can be
H O production
ꢃ
h
*
330
360
530
580
750
690
800
h
610}720 565}640 610
830
ꢀ
h&
!ꢁ
730
700
780 800 (25 min)
ꢂ
ꢃꢇ ꢅꢊ
CO production
distinguished as a shoulder on the "rst water production
peak (Fig. 2 and solids CarbA, B, and C in Ref. (10)), the
onset of CO production seems to coincide with the end of
that step.
h
h
h&
550
700
810
540
640
730
560
630
690
650
750
780
770
780
790
*
ꢀ
!ꢁ

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
417
ꢄ
ꢃ
ꢂ
FIG. 3. Evolution of the total oxygen to tungsten: O/W (᭹) (as water H O/W (᭿) and carbon monoxide CO/W (᭜) and carbon (polymeric and
ꢃ
carbidic: C/W (᭝)). Mass balance integrations versus temperature during the temperature-programmed reduction}carburizations of WO in the di!erent
ꢂ
CH }H mixtures (603C.h\ꢁ, 2 g of tungsten trioxide). (a) WC50; (b) WC75; (c) WC100; (d) WC50N2.
ꢄ
ꢃ
To summarize, two points are remarkable in the W oxides spectively with H }CH mixtures containing 66% and 50%
ꢃ
ꢄ
reduction process: the beginning of CO production and the (v/v) of CH and where the reduction process has been
ꢄ
minimum between the two peaks of water formation. In stopped respectively when CO started to be detected
order to get more information we have performed two (5603C) and when the H O partial pressure was minimum
ꢃ
experiments called INWC66 and INWC50 carried out re- (6103C). After rapid cooling to room temperature in the
reactive mixture #ow the solids were passivated as indicated
in Table 1a.
First for sample INWC66, at the "nal temperature of
5
603C, the H O/W and CO/W ratios respectively of 0.12
ꢃ
and 0.003 lead to a global formula of WO
tion its XRD pattern shown in Fig. 5 exhibits mainly the
. After passiva-
ꢃꢈꢊꢊ
characteristics of W O . The decomposition of the com-
ꢃꢇ ꢅꢊ
plex XPS W4f signal (Fig. 6a) reveals the presence of di!er-
ent oxidic phases (Wꢋ>+71%, WV>+18%, Wꢄ>+10%)
and of a very small amount of Wꢇ (about 1%) at a B.E. of
3
1.6 eV for the W4f photoelectron peak. Otherwise the
ꢌꢆꢃ
C1s signal is very weak and broad (FWHM"2.7 eV), its
position centred at 284.9 eV being characteristic of poly-
meric carbon.
From the XRD pattern it is clear that CO appear-
FIG. 4. Temperature-programmed reduction of WO in 30% H }Ar
ꢂ
ꢃ
(
3003C.h\ꢁ, 64 mg). TCD signal (in arbitrary units) versus temperature (6). ance which indicates the beginning of reduction by CH
ꢄ

4
18
GIRAUDON ET AL.
TABLE 3
at considerably higher temperatures in the absence of H in
ꢃ
O/W (O Consumed per W during Reaction) Ratio as a Function the initial gas feed (Figs. 1c and 1d and Table 2). In that case
of P
H
from Ref. (6)
2
it can be noticed that gas evolution occurs in a short time
which probably shows that the slow step of the process is
P ꢃ (atm)
O/W (minimum of H O production)
&
ꢃ
the reduction of WO , the further suboxides reduction
ꢂ
being fast. Considering the reduction of W oxides by CH ,
0
0
1
.1
.3
1
1
1.7
ꢄ
evidenced by CO formation, its participation increases with
the ratio P ꢄ/P ꢃ since the ratio H O/CO of the number of
!
ꢃ
&
&
ꢃ
moles of H O and CO produced during the reaction is 0.3,
0
.5, 1.4, and 3.8 for ratios P ꢄ/P ꢃ of 4, 1, 0.33, and 0.
!
&
&
Then a question arises: why does CO appear later than
corresponds to the WO exhaustion. The bulk of the solid is H O? There could be three possibilities:
ꢂ
ꢃ
a suboxide WO , but XPS results show that the surface is
ꢃꢈꢉ
(i) CH and H can reduce tungsten oxides but the ener-
more extensively reduced and that CO appearance co-
ꢄ
ꢃ
gies of activation of the rate constants of reduction by CH
are higher than those related to reduction by hydrogen.
incides also with the formation of Wꢇat the sample surface.
ꢄ
For sample INWC50, the ratios H O/W and CO/W are
ꢃ
(ii) CH is only slightly active for WO reduction (see
respectively 1.2 and 0.2, giving a global formula of WO
.
ꢄ
ꢂ
ꢁ
ꢈꢋ Figs. 1c and 1d), but it is better for the reduction of a W sub-
Its bulk characterization by XRD (Fig. 5) does not reveal
any WO or W O phase and shows WO as the only
oxide phase together with small amounts of a reduced phase
where W is at zero oxidation degree and which could be
oxide. This suboxide would probably not be W O since
ꢃꢇ ꢅꢊ
CO is not seen during the "rst step of reduction (WO P
ꢂ
ꢂ
ꢃꢇ ꢅꢊ
ꢃ
W O ) but more likely WO , since CO production
ꢃꢇ ꢅꢊ
ꢃ
increases rapidly during the second H O peak.
W metal or, more probably, a-W C. This shows that, as
ꢃ
ꢃ
(iii) CO appearance could be related to the formation of
expected from previous results, the subsequent transforma-
a metal phase at the solid surface where CH could be
tion of WO into W has already begun. The XPS W4f signal
ꢄ
activated by chemisorption (either as CH or as C) to
V
ꢃ
(
Fig. 6b) is very complex and evidences the presence of, at
further react for the reduction of the remaining W subox-
least, four doublets corresponding to Wꢋ>, WV>, Wꢄ>, and
ides. Here, let us recall a previous study (10), which shows
that CO production during a TPRC of WO with a CH /D
Wꢇ, the Wꢇ phase percentage being approximately 11%.
The B.E. of 31.9 eV for the photoelectron W4f peak of the
ꢂ
ꢄ
ꢄ
ꢃ
ꢃ
ꢌꢆꢃ
mixture begins simultaneously at the onset of CH }D
reduced component could correspond to a mixture of
exchange. Since this exchange reaction is likely to occur
only on metal, this is probably another indication, in addi-
tion to the XPS study on the INWC66 sample, of the high
surface reduction when CO formation starts. However, if it
is clear that CO appearance is an indication of the presence
of a metal phase at the solid surface, at that stage it is not
possible to know whether CO and metal are simultaneously
formed during WO reduction by gaseous CH or if the
W metal and of W carbide. These results con"rm that the
minimum of water production corresponds to the end of
W O reduction into WO and, although the reduction of
ꢃꢇ ꢅꢊ
ꢃ
ꢃ
WO into W is slower, it has readily started at that point.
Let us note that the CO production is faster during the
second peak of H O production which corresponds to WO
reduction.
ꢃ
ꢃ
All of the above information can be interpreted in terms
of competitive reduction of W oxides by H and CH . It is
ꢃ
ꢄ
presence of W metal at the surface due to the WO reduc-
ꢃ
ꢃ
ꢄ
tion is a condition for CH adsorption before it can reduce
clear that the reduction of WO by H is favored at low
ꢄ
ꢂ
ꢃ
WO .
temperatures compared to that by CH since H O forma-
tion occurs well before that of CO. This observation is
ꢃ
ꢄ
ꢃ
In order to check the "rst hypothesis (i) we have per-
further supported by the fact that reduction of WO occurs formed a TPRC of WO (Fig. 7). It was prepared by TPR of
ꢂ
ꢃ
WO in a H }N (1/1) mixture in order to have a better
ꢂ
ꢃ
ꢃ
separation between the two H O peaks, from room temper-
ature until the minimum of water production was reached
ꢃ
TABLE 4
Characteristics of the Reduction of WO
3
(6353C). The corresponding O/W ratio was 1.1 and the
e!ectiveness of the complete transformation of WO into
Beginning of
CO formation
ꢂ
WO was ascertained by XRD. After cooling to room tem-
ꢃ
P
P
H O/W
!
&
ꢄ
&
ꢃ
ꢃ
perature the sample was heated to 7503C (603C.h\ꢁ) in
0
0
0
1
0
.2
.5
.75
.5
0.8
0.5
0.25
0
550
540
560
650
770
0.85
0.5
0.2
*
a CH }H (95/5) gas mixture (20 l.h\ꢁ) and kept in an
ꢄ
ꢃ
isotherm for 10 h (sample WC95A). Considering the sub-
sequent oxygen removal, CO was the only oxygen-contain-
ing product in the gas phase (CO/W"1.8). It began to be
detected at 4703C, which is much lower than for samples
0
*

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
419
ꢄ
ꢃ
ꢂ
FIG. 5. (a) X-ray di!raction pattern of the INWC66 sample; beneath are the lines of W
beneath are the lines of WO (*) and W C (} }).
O . (b) X-ray di!raction pattern of INWC50 samples;
ꢃꢇ ꢅꢊ
ꢃ
ꢃ
WC20, !50, and !75 (5403C to 5603C). CO production allow us to choose between the second and the third
was maximum around 6303C before disappearing at 6703C. hypotheses.
Consequently, the overall reduction process occurred at
Going further, if it is assumed, as suggested above, that
a lower temperature than in experiments WC20, -50 and H is the only species which reduces WO into WO via
ꢃ
ꢂ
ꢃ
-
75, which implies that explanation (i) may be ruled out. On W O formation, at least when its partial pressure is not
ꢃꢇ ꢅꢊ
the other hand, it is clear that CH is more e!ective in the too low, it is possible to calculate the ratio [(H O/W)!
ꢄ
ꢃ
reduction of WO than in that of WO (compare with 1/(CO/W)] of the number of moles of H O and CO
experiment C or D). However, this experiment does not produced during the reduction of WO into W (Table 5).
ꢃ
ꢂ
ꢃ
ꢃ

4
20
GIRAUDON ET AL.
carbidic or polymeric) is observed in the same temperature
range as that of CO with a lower ending temperature as
CH pressure increases (810, 730, and 6903C for CH pres-
ꢄ
ꢄ
sures of respectively 0.2, 0.5, and 0.75 atm). Finally the
retention of carbon by the solid (C/W) increases as P /P
drops (Table 6).
&
ꢃ
!&
ꢄ
Consequently, it seems that carburization begins to pro-
ceed as soon as some W metal is formed if CO is really an
indication of W formation. This observation is very much in
agreement with previous conclusions reported in Ref. (6)
where we mentioned that carburization takes place only
when W metal is present at the surface of the solid. Let us
note that the simultaneous reduction of WO into W by
ꢃ
FIG. 6. (a) XPS W4f signals of the INWC50 sample. (b) XPS W4f CH and CH conversion into carbidic or amorphous car-
ꢄ
ꢄ
signals of the INWC66 sample.
bon can be another argument for CH decomposition be-
ꢄ
fore bulk WO reduction, C (or CH ) species deposited at
the surface being either oxidized by O species of the tung-
Considering, as an example, the value obtained for equal sten oxides, incorporated into W metal for carburization, or
pressures of H and CH of 0.6, it is clear that CH is left at the solid surface as an amorphous deposit.
In order to understand what happens at the end of the
Concerning the reduction by CH , several possibilities strong methane consumption and to characterize the solids
ꢃ
V
ꢃ
ꢄ
ꢄ
a better reducing agent for WO reduction than H .
ꢃ
ꢃ
ꢄ
can be considered:
i) 3&&O''#CH "CO#2H O
at that stage, three TPRCs (samples WCPX, X"50, 75, 85)
have been carried out and stopped at the end of the strong
(
[1]
[2]
[3]
[4]
[4ꢀ]
[5]
[5ꢀ]
[3]
ꢄ
ꢃ
(ii) &&O''#CH "CO#2H
ꢄ ꢃ
&
&O''#H "H O
ꢃ ꢃ
(iii) CH "C#2H or
ꢄ ꢃ
CH "CH #(2!x/2)H
ꢄ
V
ꢃ
&
&O''#C"CO
(1#x/2)&&O''#CH "CO#x/2H O
V
ꢃ
&
&O''#H "H O.
ꢃ
ꢃ
where, &&O'' is an oxygen moiety in tungsten oxides. Obvi-
ously, the "rst hypothesis does not agree since the ratio
H O/CO for WO reduction by CH }H mixtures should
ꢃ
ꢂ
ꢄ
ꢃ
be higher than or equal to 2, which is obviously not the case
for WC50, !75, !100, and !50N2 (Table 5). Hypotheses
(ii) and (iii) are both reasonable.
(
b) Carburization. We will now examine the results of
the carbon formation from CH (either for carburization
ꢄ
r as polymeric carbon). All of the CH consumption pro"les
ꢄ
(
Fig. 1) exhibit a strong peak (except WC20 which shows
FIG. 7. (a) Temperature-programmed reduction of WO (m"2.046 g,
a supplementary shoulder) which evidences a residual con-
ꢂ
6
(
03C.h\ꢁ) to WO in a 1/1 H }N mixture (total #ow rate, 10 l.h\ꢁ).
ꢃ
ꢃ
ꢃ
sumption, the peak being sharper as P ꢄ/P is higher. It is
b) Subsequent carburization pro"le of WC95A in a 95/5 CH }H mixture
!
&
&
ꢃ
ꢄ ꢃ
remarkable that the strong production of carbon (either (20 l.h\ꢁ, 603C.h\ꢁ, hꢀ!ꢁ"7503C for 10 h).

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
421
ꢄ
ꢃ
ꢂ
TABLE 5
Mass Balance Calculations Concerning the Reduction of WO
3
WC20 WC50 WC75 WC100 WC50N2
P
(atm)
ꢄ
(atm)
0.2
0.8
2.1
3.8
0.5
0.5
0.6
1.4
0.75
0.25
+0
0.47
1
0
*
0.27
0.5
0
*
!
&
P
&
ꢃ
[
(H O/W!1)/(CO/W)]
ꢃ
H O/CO
0.12
ꢃ
CH consumption which corresponds to an ending temper-
ꢄ
ature of 702, 678, and 6703C for respectively WCP50, -75,
and !85. The samples were then rapidly cooled to room
temperature before being passivated. Mass balance calcu-
lations from elemental and chromatographic analyses give
C/W ratios of 0.85 (0.92), 1.16 (1.19), 1.40 (!), respectively,
for WCP50, !75 and !85. The resulting XRD are in Fig.
8
. That of WCP50 clearly exhibits only the hexagonal
closed- packed a-W C phase whereas those of WCP75 and
ꢃ
WCP85 are poorly resolved, showing particularly broad
lines in the 30}453 2h windows. The very broad lines cannot
be explained only by the presence of small W C particles
ꢃ
and probably attest that the solid is not well organized. One
notices besides a very small contribution of the hexagonal
a-WC phase, this one being more extensive as P ꢄ/P ꢃ is
!
&
&
higher.
From these results it is clear that carburization into WC
is far from being complete at the end of the CH consump-
ꢄ
tion peak. When all oxygen has been practically removed
FIG. 8. XRD of the passivated WCP50 (a), WCP75 (b), and WCP85 (c)
from the initial solid, only a-W C is formed for a ratio samples prepared from temperature-programmed reduction}carburiz-
ꢃ
P
ꢄ/P of 1 whereas the successive transformation of W C ations of WO respectively in CH 50}75}85 (v/v) in H and stopped at
ꢂ ꢄ ꢃ
ꢃ
!
&
&
ꢃ
7
003C (WCP50), 6803C (WCP75), and 6703C (WCP85). Above are repre-
into WC has already begun for the higher ratios.
sented the characteristic patterns of WC (*), a-W C (#), and W (£)
In order to remove the excess polymeric carbon which is
always deposited at the sample surface during the carburiz-
ation process (11) samples WC50, !75, !100, and !50N2
ꢃ
extracted from the JCPDS data "les.
have been submitted in situ to a cleaning treatment in pure Fig. 1 of Ref. 11, which showed a slow CH decrease until
ꢄ
#
2
owing H (10 l.h\ꢁ) just after carburization at 8003C for a weak and stable value of CH was reached. It has been
hours. The CH trace is rather similar to that given in shown that the strong CH production at the beginning of
ꢃ
ꢄ
ꢄ
ꢄ
the treatment probably arises from the excess carbon elim-
ination, while the residual CH pressure at the stationary
ꢄ
state has been ascribed to a slow process of decarburization
TABLE 6
Characteristics of the Di4erent Carbides: Mass Balance (12, 13). The elimination of carbidic carbon can result from
Ratios of Carbon Held by the Solid at the End of the Strong Peak the two successive steps:
of Methane Production* and at the End of the Carburization**
(
W)Cꢂꢃꢄ&#2H "Wꢂꢃꢄ&#CHꢄ
[1]
[2]
ꢃ
Samples
(C/W)*
(C/W)**
.
$
(
W)C"ꢃꢅꢆ#Wꢂꢃꢄ&"(W)Cꢂꢃꢄ&#W"ꢃꢅꢆ
.
WC50
WCP50
WC75
WCP75
WCP85
WC100
WC50N2
0.94
0.92
1.40
1.19
1.4
1.3
*
2.1
*
*
2.1
2.2
At 8003C, the experimental P (6}7;10\ꢄatm) is very
!&
ꢄ
close to that of the equilibrium of reaction [1] (about 8.7;
0\ꢄatm). Consequently, steps [1] and [2] respectively of
1
1.8
1.6
partial surface decarburization and of carbidic carbon di!u-
sion from the bulk to the surface are probably fast.

4
22
GIRAUDON ET AL.
After this cleaning procedure, the solids were cooled in the WC20 sample (6). By way of example a decomposition of
#
owing H to room temperature and submitted to a simple the C1s, W4f, and O1s photoelectron peaks for the WC50
ꢃ
nitrogen sweep before being passivated in an O }N
sample is exhibited in Fig. 10. The C1s peak shows a well-
ꢃ
ꢃ
(0.4/99.6) mixture. Finally the samples WC50, !75, !100, de"ned carbidic component at low B.E. (283.2$0.2 eV); the
and 50N2 were characterized by EA, XRD, and XPS. The other components at higher B.E. are associated with carbon
elemental analysis results given in Table 1b show that the linked with oxygen as previously noted (6). The W4f peak
synthesis process for these samples leads to incomplete consists of, in any case, two main doublets characteristic
carburization (C/W atomic ratio less than 1). The XRD respectively of the reduced and oxidic forms (Wꢋ>). The
patterns of the four samples are displayed in Fig. 9. Sample B.E. of the W4f photoelectron peak from the more re-
ꢌꢆꢃ
WC50 corresponds to a major a-WC phase whith a minor duced doublet arises at 32.2 $0.2 eV for all samples, con-
phase which could be a-W C phases or a mixture of a-W C "rming that tungsten is in its carbidic form (6). Besides, the
ꢃ
ꢃ
with W metal. The others patterns are qualitatively similar, O1s signal is typically asymmetric and has a B.E. of 531 eV
but with a higher proportion of a-W C as the CH content characteristic of oxygen bound to tungsten in transition
ꢃ
ꢄ
in the feed is higher. This is probably due to a too short (#6) oxides.
duration of the carburization isotherm at 8003C. These
The surface areas of the solids are listed in Table 1b. It can
results may seem to be in contradiction with the previous be seen that the surface area of WC50 is the highest of the
ones on the samples WCPX. We will come back later on four samples (19 mꢃ.g\ꢁ) and higher than those of most
these points.
WCs obtained in a H }CH 1/4 gas mixture (6). In contrast,
ꢃ
ꢄ
Surface characterizations performed by XPS experiments WC75, !100, and !50N2 have low surface areas, obvious-
led to corresponding photoelectron W4f, C1s, and O1s ly smaller than those obtained without hydrogen cleaning
peaks having a general aspect similar to those observed for treatment.
2
. Reduction}Carburization of WO2
The "nal objective of these studies is to try to select
optimum experimental conditions for the synthesis of
W carbides (mainly a-WC) with surface areas as large as
possible. In order to improve the surface areas of the car-
bides it seems desirable to carburize at the lowest temper-
ature possible. Thus W metal should be carburized as soon
as it is formed in order to decrease the temperature of the
overall process and to avoid sintering at high temperature.
Since, in the "rst part of this study, it has been seen that CH
ꢄ
is likely to be more e$cient for the WO PW transforma-
ꢃ
tion than H and that the carburization proceeds faster as
ꢃ
P
ꢄ/P ꢃ increases, one might envision separating the over-
!
&
&
all reduction}carburization process into two steps. The "rst
one will be the formation of a well-de"ned WO species that
ꢃ
will further be reduced and carburized in a CH -enriched
ꢄ
methane/hydrogen mixture.
Our purpose being to synthesize a-WC, taking into ac-
count the previous results from experiments WCPX, seem-
ing to indicate that a higher P
in the early stages of
!
&
ꢄ
carburization led to the formation of higher WC percent-
ages, we "rst performed the reduction}carburization of
WO in a CH }H mixture (95/5). The procedure was the
ꢃ
ꢄ
ꢃ
following:
(
i) WO was "rst reduced in pure #owing hydrogen from
ꢂ
room temperature to 5203C until water production signi"-
cantly decreased and was left for about 2 h at that temper-
ature. The ratio O/W corresponding to WO reduction by
ꢂ
H was about 1.4}1.5. XRD results show the presence of
FIG. 9. X-ray di!raction patterns of the di!erent carbides. (a) WC50;
ꢃ
(
b) WC75; (c) WC100; (d) WC50N2. WC (*), a-W C (#), W (£).
WO and of some W metal (Fig. 11).
ꢃ
ꢃ

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
423
ꢄ
ꢃ
ꢂ
FIG. 10. Decomposition of the W4f (a), C1s (b), and O1s (c) XPS peaks of WC50.
(ii) After the reactor was cooled to room temperature, CO formation is the same for both experiments (470}
H was replaced by the CH }H 95/5 mixture (20 l.h\ꢁ). 6703C), CO partial pressure is, in the early stages of the
ꢃ
ꢄ
ꢃ
The solid was then heated (603C.h\ꢁ) to a chosen temper- reduction, signi"cantly lower for experiment WC95A than
ature (7003C, WC95B or 7503C, WC95C, and left for several for WC95B. Considering that the extent of the initial oxide
hours in an isotherm before being cooled to room temper- reduction was lower for the "rst experiment (O/W"1.1) as
ature and passivated.
calculated from the amount of water evolved during the
production of WO than for the second one (1.4) as also
ꢂ
(
a) Reduction of WO2. The trace of CO for experiment evidenced by XRD (hence the amount of W metal at the
WC95B is shown in Fig. 12 and compared to that of sample sample surface was probably lower), this can be explained
WC95A. It is shown that, while the temperature range for by the necessity for CH to be activated by adsorption on
ꢄ

4
24
GIRAUDON ET AL.
FIG. 11. XRD of WO (m"2 g) reduced in #owing H (10 l.h\ꢁ) from
ꢂ
ꢃ
room temperature to 5203C (1203C.h\ꢁ, 4 h in isotherm). WO (᭛), W (£).
ꢃ
W metal before being active in W oxides reduction. Conse-
quently, this would be in favor of the third explanation (iii)
proposed above for the in#uence of the ratio P ꢄ/P ꢃ on
!
&
&
H O and CO formation during oxide reduction. However,
ꢃ
we think that this argument should not be considered as
a de"nite proof.
(
b) Carburization. A "rst attempt of carburization
following the above procedure was performed with a max-
imum temperature of 7003C (WC95B), which was main-
tained for 10 h. XRD characterization (Fig. 13) shows that
carburization was far from being complete since a-W C was
the major phase formed during this experiment with only
along a small percentage of a-WC. The global formula
ꢃ
FIG. 13. XRD patterns of the passivated samples WC95A (a), WC95B
b), WC95C (c), and WC20D (d) prepared by temperature-programmed
(
reduction}carburizations of prereduced WO (2 g) (see Table 1a) in
obtained from EA which is WC
carbon deposit of polymeric carbon. Another preparation
O
attests to a strong
ꢂ
ꢁ
ꢈꢄꢊ ꢇꢈꢁ
CH }H (95/5) gas mixtures, except for WC20D in (20/80) (603C.h\ꢁ,
ꢄ
ꢃ
3
0 l.h\ꢁ) respectively to 750}700}750}7503C for 10}10}27}10 h (WC95A,
starting from a very similar WO #W mixture after reac-
WC95B, WC95C, WC20D).
ꢃ
tion at 7503C for 27 h (WC95C) led to a more extensive
carburization (Fig. 13) but it is still incomplete since a-W C
ꢃ
is again detected which can seem rather surprising consider- ing that carburization in a mixture CH }H 1/4 led to the
ꢄ
ꢃ
complete carburization into WC at 7503C, to a high per-
centage of WC at 7003C (6), and to the apparently faster
carburization in CH }H mixtures when the ratio P ꢄ/P ꢃ
ꢄ
ꢃ
!&
&
increases as seen earlier.
However, these observations correlated with those ob-
tained from WCX and WCPX syntheses can be rational-
ized in the following way:
Methane starts to be adsorbed on the reduced surface
(either W or surface W carbide) where it dissociates into
H
and CH fragments or C atoms at the solid surface.
!
$ꢂ
V
Such an adsorption and decomposition is favored by higher
P
(adsorption) and lower P (decomposition). As soon as
!
&
ꢄ
&
ꢃ
CH or C is formed it can be incorporated into the solid as
V
surface carbide and di!use into the bulk, leading to bulk
carbides, or polymerize to form the carbon deposit. The
competition of these two processes is a function of temper-
ature. Of course, when the solid surface gets covered by the
FIG. 12. CO traces of WC95A (᭝) and WC95B (᭛) versus temperature
obtained during the reduction}carburizations of these samples in CH }H
ꢄ
ꢃ
(95/5) gas mixtures (603C.h\ꢁ).

EFFECT OF CH }H CONTENT ON CARBURIZATION OVER WO
425
ꢄ
ꢃ
ꢂ
CH or C atom formationPsurface W carbide formation PC di!usion into the bulk
V
(
C deposit).
SCHEME 1
polymerization carbon deposit, CH adsorption and conse- the preparation conditions will have to be done by playing
ꢄ
quently CH or C atom formation is decreased as is the on the CH /H mixture composition, the maximum tem-
V
ꢄ
ꢃ
feeding of the solid bulk with C atoms; hence, WC forma- perature, the duration of carburization, and the space velo-
tion would stop even under conditions where usually it city of the carburizing mixture. Until now, the best surface
should be complete in the absence of such surface blocking area obtained was 27 mꢃ.g\ꢁ for sample WC95C, which is
by C deposit. The layer of polymerized carbon is assumed to a core of W C surrounded by WC at its surface.
ꢃ
be ine$cient in the carburization processes. The reaction
scheme is shown in Scheme 1.
CONCLUSION
These four steps are in#uenced di!erently by temperature
and P /P . At a given temperature the "rst step (CH or
The in#uence of the composition of a carburizing CH }
!
&
ꢄ
&
ꢃ
V
ꢄ
C formation) is favored by high P ꢄ/P ratios as seen H gas mixture on the process of reduction}carburization
!
&
&
ꢃ
ꢃ
above which will increase the surface coverage with over tungsten oxides has been studied. This process has
individual C adsorbed species (with only one carbon). been shown to be very sensitive to the ratio P ꢄ/P ꢃ of the
!
&
&
Consequently, the rates of the two others steps of surface reactive gas. The solids have been characterized by XRD
W carbide formation and of C polymerization will also and XPS at various extents of reaction. The overall process
increase as P ꢄ/P increases. But while the W carbide is complex and starts at low temperature with the reduction
!&
&
ꢃ
formation is probably "rst order in the CH (or C atom) of WO and W O by hydrogen. Then, as soon as WO is
V
ꢂ
ꢃꢇ ꢅꢊ
ꢃ
surface concentration, the carbon polymerization will in- formed at the surface of the solid, reduction by CH occurs
ꢄ
crease faster as the CH surface concentration increases, and as evidenced by CO formation, this reduction being favored
V
also the blocking of the solid surface, resulting in a decrease compared to that by H . This step is likely to occur via
of the rates of CH adsorption and decomposition and of adsorption and decomposition of CH on W metal formed
ꢃ
ꢄ
ꢄ
surface and bulk carbides formation. When the solid surface at the surface by WO reduction and reaction of C atoms or
ꢃ
is completely covered with polymerized carbon the carbu- of CH fragments with oxygen atoms of the solid, although
V
rization process stops because the solid surface can no longer this has not yet been de"nitively proved. The carburization
be fed by monoatomic C species. Hence, this mechanism process starts with monoatomic carbon deposition (as
explains well why carburization starts at lower temperatures C atoms or as CH adsorbed species) at the W surface and
V
when P ꢄ/P is increased since the rates of the processes surface carburization. Carbon species then di!use into the
!&
&
ꢃ
increase but it also explains why carburization can stop before bulk of the solid to form, "rst, a-W C whose formation
ꢃ
being complete.
occurs rapidly. WC formation by carbon di!usion into
A consequence of the above observations is that a mixture W C is slower and can be incomplete when the ratio
ꢃ
!&
H }CH with lower CH concentrations could be prefer-
P ꢄ/P ꢃ is higher. This phenomenon is probably related to
ꢃ
ꢄ
ꢄ
&
able to obtain more easily pure a-WC; since it is well known the extent of polymeric carbon formation at the solid sur-
that H is an inhibitor for alkanes adsorption (14), a higher face. The reduction}carburization of WO has been com-
ꢃ
ꢂ
H
pressure will lower CH adsorption and its further pared to that of WO prepared by previous reduction of
decomposition. Consequently, the rate of carbon deposition WO by hydrogen. The overall process occurs at lower
could be adapted to that of C di!usion into the bulk of the temperatures when using WO and higher surface areas are
solid by adjusting H and CH pressures to an optimum obtained for the resulting W carbides; the best surface area
ꢃ
ꢄ
ꢃ
ꢂ
ꢃ
ꢃ
ꢄ
value. A "nal experiment has been performed on obtained was 27 mꢃ.g\ꢁ for a solid composed of a core of
a prereduced sample in a CH }H 1/4 mixture at 7503C for W C covered with WC. This surface is about twice that of
ꢄ
ꢃ
ꢃ
1
0 h (WC20D, Table 1a). The solid obtained was mainly the WC usually obtained by WO carburization. All of these
ꢂ
a-WC (XRD, Fig. 13) with only a very small amount of observations will be a guideline to control and optimize the
a-W C. It surface area was 23 mꢃ.g\ꢁ. This result seems to synthesis of the carbides.
ꢃ
con"rm the above reasoning.
These attempts at improving the W carbides surface areas
ACKNOWLEDGMENTS
by using WO prereduced into a mixture WO }W were
ꢂ
ꢃ
successful since the solids obtained had higher surface areas
than those obtained in our laboratory by reduction}carbu-
rization of WO by CH }H mixtures. The optimization of
We thank Mr. L. Gengembre for performing the XPS analysis, Ms.
Burylo for performing the X-ray di!raction studies, and Dr. A. Lofberg for
helpful discussions.
K
ꢂ
ꢄ
ꢃ

4
26
GIRAUDON ET AL.
Dufour, M. Kamal, L. Feigenbaum, J.-M. Giraudon, V. Keller,
REFERENCES
P. Wehrer, M. Cheval, F. Garin, P. Kons, P. Delcombe, and L. Binst,
in &&Proceedings of the 10th Congress on Catalysis'' (L. Guczi, F.
Soymost, and P. Tetenyi, Eds.), pp. 927}937. Akademia Kiado,
Budapest, 1993.
1
. S. T. Oyama, Catal. ¹oday 15, 179 (1992).
2
. E. Iglesia, F. H. Ribeiro, M. Boudart, and J. E. Baumgartner, Catal.
¹oday 15, 445 (1992).
3
4
. S. T. Oyama, Ph.D. Dissertation, Stanford University, 1981.
9. G. Leclercq, L. Leclercq, and R. Maurel, J. Catal. 5, 87 (1977).
. S. T. Oyama, J. C. Sclatter, J. E. Metcalfe, and J. M. Lambert, Jr., Ind. 10. A. Lo
K
fberg, A. Frennet, G. Leclercq, L. Leclercq, and J.-M. Giraudon,
Eng. Chem. Res. 27, 1639 (1988).
J. Catal. 189, 170 (2000).
5
6
. J. S. Lee, S. T. Oyama, and M. Boudart, J. Catal. 106, 125 (1987).
. G. Leclercq, M. Kamal, J.-M. Giraudon, P. Devassine, L. Feigenbaum,
1
1
1
1. G. Leclercq, M. Kamal, J. F. Lamonier, L. Feigenbaum, P. Malfoy, and
L. Leclercq, Appl. Catal. 169, 121 (1995).
L. Leclercq, A. Frennet, J. M. Bastin, A. LoK fberg, S. Decker, and
2. F. H. Ribeiro, R. A. Dalla Betta, M. Boudart, J. E. Baumgartner, and
E. Iglesia, J. Catal. 130, 86 (1991).
M. Dufour, J. Catal. 158, 142 (1996).
. G. Leclercq, M. Dufour, A. El Gharbi, J.-M. Giraudon,
M. Kamal, J. F. Lamonier, and L. Leclercq, Bull. Soc. Chim. Belg. 105,
7
8
3. F. H. Ribeiro, R. A. Dalla Betta, G. J. Guskey, and M. Boudart, Chem.
Mater. 3, 805 (1991).
2
(1996).
. A. Frennet, G. Leclercq, L. Leclercq, G. Maire, R. Ducros, M. Jar- 14. A. Frennet, in &&Hydrogen E!ects in Catalysis'' (Z. Paal and P. G.
dinier-O!ergeld, F. Bouillon, J. M. Bastin, A. Lofberg, P. Blehen, M.
Menon, Eds.), pp. 399}423. Dekker, New York, 1988.
K