You are on page 1of 22

Applied Catalysis, 33 (1987) 309-330 309

Elsevier Science Publishers B.V., Amsterdam -Printed in The Netherlands

SYNTHESIS OF METHYL MERCAPTAN FROM CARBON OXIDES AND H2S WITH TUNGSTEN-ALUMINA
CATALYSTS

J. BARRAULTa* , M. BOULINGUIEZb, C. FORQUYC and R. MAURELd


aLaboratoire de Catalyse en Chimie Organique, UA CNRS 350, 40, Avenue du
Recteur Pineau, 86022 Poitiers Cedex, France.
b
Centre de Recherches Elf-Solaise (CRES) 60360 St. Symphorien D'Ozon, France.
'Societe Nationale Elf Aquitaine (S.N.E.A.-P) Groupement Recherches de LACQ,
64170 Artix, France.
dTnstitut de Recherches sur la Catalyse, 2, Av. Albert Einstein, 69626
Villeurbanne, France.
*
To whom correspondence should be addressed.

(Received 28 November 1986, accepted 11 March 1987)

ABSTRACT
The direct synthesis of methylmercaptan from carbon oxide(s) (CO or CO2).
hydrogen sulfide and hydrogen was studied in the presence of tungsten-based
catalysts. The optimization of W-K-Al203 catalysts makes it possible to obtain
rather high activity and selectivity while keeping a fair stability,
The main steps of the reaction are the formation of carbonylsulfide and
hydrogenation of carbonylsulfide to methylmercaptan. The second step is
generally the slow one, owing to the low hydrogenating properties of sulfide
catalysts.
Water has a strong inhibiting effect in the (Cop, H2S, H2) reaction and the
activity of the catalyst is lower than in the (CO, H2, H2S) reaction. ESCA
experiments provide some important information on the behaviour of these cata-
lysts during the sulfidation step of the catalytic reaction(s).

INTRODUCTION
Methyl mercaptan is a well known raw material used in the synthesis of
numerous products in the agricultural, plastics, rubber and chemical industries.

The usual process of preparation consists of the catalytic thiolation of


methanol [I] but obviously it would be an advantage if simpler starting materials
could be used, i.e., carbon oxides. Indeed this preparation would be of interest

in the direct transformation of acid gas (C02-H2S = 40/60) derived from the
purification of natural gas.
Thus the direct conversion of carbon oxides (carbon monoxide or carbon
dioxide) to CH3SH with a sulfactive catalyst have been studied

CO + H2S + 2H2 + CH3SH + H20 1

co2 +HS+3H
2 2 + CH3SH + 2 H20 2

01889834/87/$03.50 D 1987 Elsevier Science Publishers B.V.


310

Little work has been carried out on these reactions, the most interesting
being that of the PENNWALT Corp. 12-41 where alkaline promoted nickel catalysts
were used for methylmercaptan synthesis from syngas [S]. But neither the
mechanism of the reaction nor the role of the catalyst have been studied
previously [5-121. In this paper after a thermodynamic analysis of reactions
(1) and (2) and of some intermediate steps or secondary reactions, an investi-
gation into the catalytic transformation of CO or CO2 in the presence of pro-
moted tungsten-alumina catalysts is presented.

TABLE 1
Conversion of carbon monoxide to methylmercaptan: thermodynamic limitations at
30 bars.

500 K 600 K 700 K

CO/H2S/H2 CO Conv/% Se1 ./% CO Conv./% Sel./% CO Conv.% SelJ%

l/1/2 99.3 86.0 98.2 72.8 88.3 64.5


l/4/2 100 89.4 98.8 76.3 86.9 66.3
l/4/4 100 99.8 99.5 94.2 90.0 87.8
l/4/8 100 100 99.8 99.3 94.0 92.7

TABLE 2
Conversion of carbon dioxide to methylmercaptan: thermodynamic limitations at
30 bars.

500 K 600 K 700 K

C0,/H2S/H2 CO2 conv./% Se1 ./% CO2 conv./% Sel./% CO2 conv./% Sel./%

l/1/3 77.0 100 56.5 96.5 45.3 64.2


l/4/3 83.0 99.9 62.3 97.7 47.7 72.5
l/4/4 96.5 100 75.4 98.5 56.1 77.4
l/4/8 99.9 100 95.3 99.5 76.2 85.4

THERMODYNAMICS
Equilibrium values have been calculated for global reactions 1 and 2 and for
some intermediate steps la, 2a, lb, 3 from Gibbs Free energies [I31

CO + H2S + COS + H2 (la)

CDS + 3 H2 + CH3SH + H20 (lb)

+ H2S + COS+H20 (2a)


GO2
CO + H20 -t CO2 + H2 (3)
313

01

500 600 700 T(K)

FIGURE 1 Thermodynamic limitations: influence of total pressure on the forma-


tion of methylmercaptan (CO/H,S/H2 = l/4/4).

TABLE 3

Conversion of carbon monoxide to carbonyl sulfide: thermodynamic limitations for


COS yields from reaction (la).

CO/H2S/H2 500 K 600 K 700 K

l/l/- 17.6 16.9 15.4


1/a/- 34.1 31.4 30.0 COS yield/%
1/4/z 7.9 6.6 6.0
?/4/4 4.3 3.5 3.2

From the results presented in Tables 1 and 2 it can be seen that the forma-
tion of methylmercaptan is thermodynamically possible over a temperature range
of 450-700 K, but yields are higher at low temperatures especially from (CO2,
H2S, H2). Moreover the selectivity for methylmercaptan is increased when
operating with an excess of H2S and H2 and at pressures above 10 bars (Figure 1).
The formation of COS (reactions la and 2a) is severely temperature-limited and
inhibited by hydrogen (Tables 3 and 4). Given these results, the reactions were
studied at low temperature, with a COx/H2S/H2 ratio of about l/4/2 and a pres-
sure between 30 and 50 bars.
312

TABLE 4
Conversion of carbon dioxide in carbonyl sulfide: thermodynamic limitations for
CDS yields from reaction (Za).

CO/H2S/H20 500 K 600 K 700 K

‘/I/- 1.9 3.7 5.9


t/4/- 3.8 7.3 11.6 COS yield/%

l/l/O.1 0.3 1.3 2.8


l/4/0.1 1.3 4.0 7.9

EXPERIMENTAL
Catalyst preparation
Catalysts were prepared by the incipient wetness technique and A1203 from
Rhine-Poulenc (type A, 300 m* g-', up = 2-3 mm) was used as support material.
Impregnated catalysts were dried in flowing air at 400 K and stored. Before
each experiment, the catalyst was sulfided in situ with pure hydrogen sulphide
(flow = 300 cm3 min-' ) at 640 K and atmospheric pressure for 6 hours.

ESCA characterization
ESCA measurements were performed with an AEI-ES 200 B electron spectrometer
using AlKa radiation (hv = 1486.6 eV). The binding energies were calibrated by
assigning 74.8 eV to the A12p transition and 285 eV to the Cls peak position
after adjustment for the surface charging effects. Samples were crushed, then
pressed on an indium foil and manipulated under flowing dry nitrogen. Surface
concentrations were determined by the following equation [14,153

IA nA
-=_
IB nB

where Ii : intensity of the peak related to i species


ni : number of i surface species
oAj : photoexcitation probability of Aj
EAj : energy of Aj level
(0 values were obtained from the literature [16]).

Reactor set-up
Experiments were carried out in a single pass fixed bed reactor designed to
operate under pressures of up to 100 bar and temperatures of up to 773 K.
Feed section. Carbon monoxide (NZO, Air Liquide) carbon dioxide (N20, Air
Liquide), hydrogen (U, Air Liquide) and hydrogen sulfide (S.N.E.A. (P)) were _
313

introduced via separate feed lines. Gas flows were regulated with mass flow
controllers from Brooks. The hydrogen sulfide was cooled and liquified at 263 K
and then pumped off into the preheater (543 K).
Reactor section. In a standard experiment the reactor (I.D. = 3.5 cm) was
filled with 10 to 50 cm3 of catalyst pellets diluted with silicon carbide ($ =
2 mm).
Product section. Hot gases leaving the reactor were passed through heated
lines and a hot condenser (373 K), depressurized, constantly led to a gas
chromatograph sampling valve and vented via a wet gas meter. The products were
separated on two columns: in the first (Porapak S:l = 3 m, 4 = 3.17 mm) CD, CH4,

co2. H2S, COS, H20, CH3SH and CS2 were separated, and the second (Molecular
sieve 5A: 1 = 2.5 m, $ = 3.17 mm) was used to measure Hz.

Experimental conditions and presentation of results


The reactions were studied in a temperature range of 523 to 623 K at a
pressure between 30 and 50 bar. Given the thermodynamic values, an excess of
hydrogen sulfide was used so that the CO/H2S/H2 and C02/H2S/H2 ratios were l/4/2
and l/4/3 respectively.
All the results presented in this paper, namely carbon oxide(s) conversion
(CO(2) conv.), selectivities and yields, were always obtained after stabili-
zation of the activity with time on stream. Generally, after an initial increase
in activity for 5 to 10 hours, there is a long period of relative stability
which allows easy determination of levels of activity and selectivity.

RESULTS
Catalytic properties of tungsten-potassium-alumina catalysts
During the first stage of this work, previous experience in the field of
sulfydrolysis of alcohols [17-191 was called upon and methylmercaptan synthesi-
sed directly from carbon oxides in the presence of tungsten based catalysts.
Preliminary experiments. Preliminary tests showed that under reasonable
temperature and pressure, the activity and selectivity of such solids for
methylmercaptan are quite significant (Table 5). At a reaction temperature of
about 570 K an important activity without significant increase of the byproducts.
methane, carbon disulfide or dimethylsulfide was observed. In agreement with
thermodynamic limitations, the pressure has a decisive effect on carbon monoxide
conversion and methylmercaptan yield. Following these results, the following
experiments have been performed at T = 568 K and P = 30 bars. In the second
step of this study, the selectivity of the same catalyst at low carbon oxide(s)
conversion was examined and from results presented in Table 6, it is observed
that: (i) the carbon dioxide conversion is lower than that of the carbon
monoxide, (ii) the selectivities are very different; with a (CO, HpV H2S)
314

TABLE 5
Synthesis of methylmercaptan from (CO, H2S, H2). Preliminary tests at low and
medium pressures in the presence of a catalyst: W03(3.55%)-K20 (1.44%)/Al203.

Pressure GHSV T/K, CO/H2S/H2 CO conv. Yield Selectivity/mole%


/bar /h-l 1% CO+CH3SH/% CH3SH CO2 CH4 DMS+CS2

12 15 513 l/6/5 29.8 12.6 42.3 40.9 - 16.8


II II 608 I' 76.4 24.6 32.2 37.9 20.5 9.4
II I, 563 U 42.0 28.8 44.8 44.8 - 10.4
30 28 " l/6/4 76.6 39.8 52.4 47.6 - -

39 36 " l/5/2 90.4 48.8 53.6 46.4 - -


45 40 " l/7/2 92.6 50.4 54.4 45.6 - -

TABLE 6
Activity and selectivity of a W03 (4.56%)-K20 (0.43%)-A1203 catalyst in methyl-
mercaptan synthesis from a) (CO, H2S, H2): P = 30 bars, T = 56% K, CO/H2S/H2 =
l/4/2. GHSV = 500 h-' and b) (CO 2, H2S. H2): P = 30 bars, T = 568 K, C02/H2S/
H2 = l/4/3, GHSV = 540 h-'.

Reaction Conversion Selectivity/mole%


CO or CO,/% co CDS CH,SH
CO2

CO, H2S, H2 (a) 24.8 39.1 21.9 39.0

C02. H2S, H2 (b) 14.5 77.7 19.4 2.9

reagent mixture carbon dioxide and methylmercaptan are equally produced. On the
other hand, with a (C02. H2, H2S) reagent blend the CO selectivity is very high

(78%) while methylmercaptan formation is low (3%), iii) in both cases a new
product, carbonyl sulfide, is formed with a selectivity of about 20% and this
product could be the main intermediate in the formation of methylmercaptan from

(la) CO + H2S -f COS + H2 or (2a) CO2 + H2 + COS + H2 0


and
(lb) COS + 3 H2 + CH SH + H20
3
Moreover some catalytic experiments of long duration (80 to 150 h) demonstrated
that both the activity and the selectivity remained unchanged. Lastly the
results presented in Figure 2 also show that the catalytic results are re-
producible from one catalyst sample to another.
Influence of catalyst composition. In order to perfect the selectivity of
W-K-A1203 catalysts and to understand the role of potassium or tungsten a
series in which the potassium or tungsten content varied was studied.
315

%A%
30..15

GHSV (h-1)- 500 950


A A A A
3
A .
l A A
t
I

; 20,.10

z 2

00 “,
w
uo 5 co
lo.- 5
; 0
c 5 Cl-$SH
T v v v
V ’ v T
v* \
\
\yy *
v v ” cos

16 2b TIME in hours -

FIGURE 2 Synthesis of methylmercaptan from (CO, H2, H2S); reproducibility of

the catalytic properties of the catalyst. (A,m,v): sample no. 1; (A,lJ,V):


sample no. 2 catalyst: WO3 (4.56%) - K20 (0.43%) - A1203. P q 30 bars, T =
568 K, CO/H2S/H2 = l/4/2.

Influence of potassium. Five catalysts were prepared, maintaining a constant


tungsten content (4.56 wt% of WO3) and varying the potassium content from 0 to
4 wt% (K20). One of these preparations for which the W/K ratio is equal to
2.35 corresponds to the stoichiometry of potassium tungstate K2W04. The cata-
lytic results for identical reaction conditions are presented in Table 7.

(i) Potassium does not have a significant effect on overall activity and a
slight increase is observed. However, when potassium is completely absent, the
catalyst shows intermediate activity.
(ii) Selectivity, on the other hand varies to a far greater degree. The MeSH/
(COS+MeSH) ratio enables one to appreciate the evolution of the hydrogenating
properties of the catalyst. It seems that the best tungsten-potassium ratio
corresponds to the stoichiometry of potassium tungstate. When the potassium
content is increased (W/K ratio < 2.35), a decrease in the hydrogenating pro-
perties is observed and with the highest K20 content (4%) heavy products (CS2 -
DMS) appear and methylmercaptan selectivity decreases.
(iii) In the last example in Table 7, a slight improvement of the CH3SH selecti-
vity is observed when a potassium tungstate precursor is used.
316

TABLE 7
Activity and selectivity of W03-K20-A1203catalysts in the synthesis of methyl-
mercaptan from (CO, H2S, H2). Conditions: P = 30 bars, T = 568 K. CO/H2S/H2 =
l/4/2, GHSV = 950 h-l.

W03 K20 W/K CO conv. Selectivity/% MeSH


/wt% /wt% /% cos CH3SH COS+MeSH
coP

4.56 0 m 16.4 34.6 32.9 32.5 0.50


II
0.43 10 13.3 35.2 28.6 36.3 0.56

1.85 2.35 16.2 37.7 24.6 37.7 0.60


II

,,a
2.23 1.96 16.9 34.2 35.0 30.8 0.47

4.00 1.09 17.7 34.5 36.8 27.lb 0.42


I,

7.10 2.90 2.35 26.7 38.2 23.6 38.2 0.62


8.50 3.50 2.35 22.9 39.1 21.3 39.6 0.65
12.60 5.10 2.35 25.4 39.2 23.4 37.4 0.62

4.56' 1.85 2.35 20.4 38.0 23.8 38.2 0.62

aThis preparation corresponds to the stoichiometry of potassium tungstate.


bselectivity in (Oimethylsulfide and carbon disulfide) = 1.6%.
'Catalyst prepared by impregnating alumina at 350 K with a solution (pH L 12)
containing the K2W04 precursor instead of the mixture of tungsten oxide and

potassium hydroxide.

Influence of tungsten. Since the best result was obtained with a catalyst
with the stoichiometry of potassium tungstate, this ratio was used in the study

of the influence of tungsten content on catalytic properties. From the results


presented in Figure 3 it is evident that activity levels out as soon as one

reaches 10% K2W04 i.e., 5% W. It can also be seen that the metal content in no
way modifies selectivity (Table 7). Consequently, a 10% K2W04-A1203 catalyst is

the optimal composition for the synthesis of methylmercaptan from (CO, H2S, Hz).
Applications to methylmercaptan synthesis. Some experiments have been
carried out with a K2W04 (IO%)-Al,03 catalyst at high carbon oxide conversion
by decreasing the GHSV to 50 h-' (Table 8). It is observed that i) the selecti-

vity to methylmercaptan is the same whatever the carbon oxide used and ii) this
compound is the only sulfided product obtained. The objective of direct
synthesis of methylmercaptan from acid gas is thus obtained with this catalyst.

But it must be pointed out that even here the rate of CO2 transformation is
lower than that of CO transformation. Moreover, in these experimental conditions,

thermodynamic limitations would be ruled out. Consequently, before pursuing the


317

A mol. h:’ I:’ of cat. 10~ mol. h?g:l of W A’

10 20

5 10

FIGURE 3 Synthesis of methylmercaptan from (CO, H2S, H2). Influence of tungsten


content on catalyst activity. Catalyst: "K2W02"-A1203. P = 30 bars, T = 568 K,

CO/H2S/H2 = l/4/2.

TABLE 8
Activity and selectivity of K2WQ410-A1203 catalyst in the direct synthesis of
methylmercaptan from (CO, H2, H2S) or (C02, H2, H2S). Conditions: P = 30 bars,
T = 568 K, CO(C02)/H$/H2 = l/4/4, GHSV = 50 h-'.

Reaction CO-H2-H2S C02-H2-H2S

Carbon oxide conversion/% 75.8 21.0

Selectivity/%

0.7
CH4
44.1
co2
co 48.9

cos
CH3SH 55.2 51.1
CS2-DMS

formulation of more effective catalysts, an attempt was made to determine the


main steps in the formation of methylmercaptan and secondary products from
carbon oxide(s) and to explain the differences in activity.
318

co

*cklp
‘CO*

10
cos

6 (s)
10 20 30 40 6 (g/h/mole CO)

FIGURE 4 Synthesis of methylmercaptan from (CO, H2S, H2): influence of


residence time on CO conversion and product formation. Catalyst and conditions:
see Figure 2,

Formation of methylmercaptan
The results presented below were obtained using a WO3 (4.56%)-K20 (0.43%)-
Al203 catalyst prepared by the classical impregnation method and sulfided in
sz'tu with hydrogen sulphide at 643 K for 5 l/2 h.
With a CO-H,fi,S reagent. The evolution of the CO conversion as well as the
yields of main products in relation to the residence time of the reagents(a) is
shown in Figure 4. Such evolution shows that the CO conversion linearly varies
with residence time. Under such conditions, it would thus seem that the reaction
319

SELECTIVITY O/o
60

50
-fl-

-8-
I I l
0 10 20 30 " 70 co CONV,O/Q

FIGURE 5 Synthesis of methylmercaptan from (CO, H25, H2): evolution of


selectivity with carbon monoxide conversion. Catalyst and conditions: see Figure
2.

is not 1 imited by diffusion or by thermodynamics of the formation of carbonyl


sulfide (COS yield = 7 to 10% for molar ratios for CO/H2S/H2 varying from
l/4/1.5 to i/4/2 at 598 K).
Carbonyl sulfide is a primary product of the reaction and is formed according
to the following equation:

co tH2S + COS + H
2 (la)

As the methylmercaptan is a byproduct, it is obviously a result of the hydro-


genation of carbonyl sulfide via the following reaction:

COS + 3H2 + CH3SH + H20 (lb)

The carbon dioxide shows up as a byproduct. Within this residence time zone,
the amount of CO2 evolved is similar to that of CH3SH and, moreover, there is no
trace of water in the effluents. This points to the carbon dioxide being pro-
duced by reaction with the water which forms during the production of methyl-
mercaptan (lb) either with the CO by water gas shift (3) or with the CDS when it
is hydrolyzed (-2a)
320

FIGURE 6 Synthesis of methylmercaptan from(C02, H2S, H2): influence of residence


time on CO2 conversion and product formation. Catalyst: WO3(4.56%)-K20(0.43%)/
A1203. P = 30 bar, T = 568K, C02/H2S/H2 = l/4/3.

CO t H20 + CO2 + H2

COS + H20 + CO2 + H2S (-2a)

It is not easy to determine exactly how the CO2 is formed, because neither
reaction (3) nor (-2a) is thermodynamically limited, and both are favoured by
acid-base catalysts [20,21]. An alternative representation is given in Figure 6,
where the evolution of selectivity in relation to the overall transformation
rate of CO is shown. As well as giving the results for a short residence time
(6<8s) the values observed for a much longer time: 6 = 31 s are also given.
Figure 5 shows the carbonyl sulfide is indeed a primary product and, moreover,
that its hydrogenation to CH3SH is the kinetically limiting step, since
appreciable amounts of CDS are still evident (around 6%) at a global trans-
formation rate of about 70% for a fourfold residence time (6 = 31 s). Carbon
dioxide is produced by the water gas shift or by COS hydrolysis much faster
than CH3SH, since water never shows up in the reaction effluents, even when
contact time is short.
Estimation of the respective rates of these reactions by studying the in-
fluence of the partial pressures of the reagents was not possible. However, the
rate of formation of carbonyl sulfide (COS) is initially equal to the overall
CO reaction rate.
r"C0 = rCOS = 0.63 x 10-2 Mole h-l g-1 catalyst

The rate of formation of COS from (CO, H2S, Hz) is compared with that from
(C02, H2S, Hz) at a later stage.
When determining the effect temperature has on activity and selectivity, all
other parameters were regarded as constant. The evolution of the overall
transformation rate (CO COIN.) corresponds to an apparent activation energy:
Ea q 15.5 kcal/mole, which is within the range of activation energies for
hydrogenation and which confirms the nature of the slow step, i.e., the hydro-
genation of COS to CH3SH.
With a (C02, H$i, Hz) reagent. The evolution of the CO2 conversion as well
as the yields of the main products in relation to residence time are shown in
Figure 6. It'is observed that: COP conv. does not vary linearly within this

residence-time contrary to the CO conv. examined above. It would seem that,


in this case, the rate is initially very high for some steps (for residence
time < 2 s). For large residence-time, the thermodynamic limitations related
to the formation of products such as CO or COS are not reached (CO yield z 24%
for C02/H2S/H2 = l/4/3 and CDS yield z 7%) nor can they be used to explain the
evolution of such products in terms of residence time. The amount of water
formed at the same time as CO, COS and CH3SH has little effect on the reaction
equilibria. However, water is no longer eliminated from the reaction and can
thus have an inhibiting effect on the formation of CO and COS and, consequently,
on the formation of CH3SH. This would seem to be borne out by certain aspects
of industrial production of COS from CO2 during which the water is constantly
eliminated [lO,ll].
Carbonyl sulfide definitely shows up as a primary product whose hydrogenation
leads to methylmercaptan. The comparison of the CO and COS yields curves for
short residence times (<2 s) show that it is possible to obtain carbonyl
sulfide from CO2 by the following reaction:

coP+HS 2 -P COS + Hz0

Moreover, this step in COS formation has been independently confirmed in the
absence of hydrogen and also led to a rate of transformation to COS of about
7%. It thus seems that this reaction is initially rapid, since the thermodynamic
limit is reached for very short residence times (6<2 s).
As for carbon monoxide, which is initially the minor product (6 = 2 s) it
rapidly becomes the major product, then seems to level off with CO conversion.
2
Two reactions can lead to its formation, either from CO2 via inverse water-
gas shift,

CO2 + H2 -f CO+HO (-3)


2
322

or from COS via the inverse reaction of COS formation.

CDS + H + CO + H2S (-la)


2

However, from the experimental results, discrimination between these two


reactions is not possible. The reduction of COS to CO by hydrogen or the in-
hibiting effect of water could account for the sudden drop in the transformation
rate of the carbonyl sulfide. The latter is in no way solely a result of the
formation of methylmercaptan. Nevertheless, the significant level of CO pro-

duced shows that, apart from the hydrogenation of COS to CO, reaction (-3) also
occurs. Although the residence-time zone in which the reaction was performed did
not enable the initial evolution of CO2 conv. to be followed as well as it was
followed for the (CO, H2S, H2) mixture, nevertheless an initial overall rate
r°C02 = 1.5 x IO -2 moles of C02/h/g of catalyst was obtained, i.e., about twice

the overall rate (P CO) obtained with the (CO, H2,H2) mixture. The initial rate
of formation of COS is of the same order.

DISCUSSION
This study shows that carbonyl sulfide is the intermediate compound whose
hydrogenation leads to methylmercaptan, be it from a (CO, H2S, H2) mixture or a

(C02, H2S, H2) mixture. The hydrogenation of COS to methylmercaptan clearly


appears as the kinetically limiting step. Moreover, various equilibria between

the initial carbon oxide, the COS intermediate and the main byproduct, i.e.,
the carbon dioxide from (CO, H2S, H2) and the carbon monoxide from (C02, H2S,

H2) are involved.


With a (CO, H2S, H2) feed mixture, the various reactions likely to be

involved are:

kl
CO + H2S = COS + H (la)
2
k-l

COS + 3 H2 + CH3SH + H20 (lb)

CO+HO = CO2 + H2 (3)


2

COS + H20 = CO2 + H2S (-2a)

The water produced in reaction (lb) is consumed to form C02. From the experi-
mental results, the rate constants of the first step are calculated (formation
of COS) as kl E 8 x 10m5 (mole h-' g-l atmm2) and k-l E 1.9 10m3 (mole h-' g-'
atm-') if the rate of formation of COS is initially equal to that of CO.
323

With a (CO2, H2S, H2) feed mixture, the main reactions are the following:

k-4
CO2 + H2S = COS + H20 (2%)
k4
COS + 3H2 + CH SH + H20 (lb)
3
k-3
CO2 + H2 = CO+HO (-3)
2
k3
COS + H : CO+HS (-1)
2 2

The carbonylsulfide is definitely formed from carbon dioxide without involving


carbon monoxide. The water formed is not consumed during the overall reaction
and seems to have an inhibiting effect. Using our experimental results the rate
constant of the formation of CDS from CO2 (2a) can be calculated as approxi-
mately k-4 g 2.7 x low4 (mole h-' g-' atmm2) and k4 E 0.255 (mole h-' g-' atmm2),
Moreover if it is assumed that the CO is also a primary product formed by the.
reverse water gas shift reaction, the rate constants can be calculated: k-3 E
9.5 x 1o-5 (mole h-' g-' atmw2) and k3 = 3.85 x 10m3 (mole h-' g-' atmD2).
(These values represent the upper limits of the rate constants on account of the
possible transformation of COS into CO when a significant amount of COS is
formed). Finally, a general diagram for the formation of methylmercaptan from
(CO, H2S. H2) or (CO, H2S, H2) would be:

cf\
0.4foY4>; cos < ' 3 CH3SH

As it was very difficult to carry out the reaction involving COS under the same
conditions there is no precise value for the rate constant of hydrogenation of
COS to methylmercaptan. Nonetheless, an approximation gives it as of the order
of the rate constants of COS formation.
Thus, methylmercaptan can be produced either from the syngas or from the acid
gas: moreover, the intermediate COS is obtained more rapidly from CO2 than CO.
However, the most rapid steps are found in the elimination of COS to CO2 or CO.
There is an essential difference between the two procedures: when starting from
carbon monoxide, the water formed along with the methylmercaptan is immediately
consumed by the hydrolysis of CDS and by the water gas shift reaction to form
C02. When starting from carbon dioxide the water that is formed is not
eliminated from the medium and, what is more, it seems to inhibit the formation
324

of COS, as a consequence of which the amount of mercaptan produced is small.


According to these results, a technical improvement in the (CO*, Hz, H2S)
reaction resulted in the use of a recycling reactor, with a low conversion per
cycle, which allowed continuous elimination of water [22]. In the following
stage of this work, some information was obtained on the superficial composition
of some of the catalysts and the evolution during the catalytic reaction.

-!

I .
AFTER #

SULFIDE
,.*-‘I”..I.” ’
LS LO EL 35 36 170 160
EL

FIGURE 7 ESCA spectra (W4f binding energy) of a K2W04-Al203 catalyst after

(CO, H2S, H2) reaction.

FIGURE 8 ESCA spectra (S2p binding energy) of a K2W04-Al203 catalyst after

(CO, HpS, Hz) reaction.

PHYSICOCHEMICAL CHARACTERIZATION OF THE CATALYSTS BY ESCA


The present study concerns two particular catalysts: tungsten oxide alone
(WO3) impregnated on alumina and a catalyst whose composition corresponds to
the stoichiometry of K2W04 on alumina, before and after sulfidation (in pure
H2S at 643 K for 5 h 30) and in the case of the K2W04/A1203 catalyst, after 85 h
time-on-stream in the (CO, H2S. H2) reaction. The results obtained are presented
in Figures 7 and 8 as well as Tables 9, 10 and 11. The catalysts were examined
before sulfidation, after sulfidation, and then after reaction for the K2W04-
Al203 catalyst.
325

Before sulfidation
The binding energies presented in Table 9 show that:

(i) The 4f5,2_7,2 doublet is not resolved, thus the position of the apparent
maximum is attributed to that of the 4f transition. For both catalysts the
7/2
6+
binding energies (36.0 + 0.2 eV) are characteristic of W species in an oxygen
environment.

(ii) The Kzp3,* binding energy (293.6-293.8 eV) for the K2W04-A1203 catalyst
corresponds to that of K+. It is not easy to determine precisely the chemical
environment of K+, since, the K binding energy is hardly sensitive to it [23].
2P

TABLE 9
Elemental binding energies for W-(K)-A1203 samples/eV.

Catalyst

wo3/~1203 non sulfided 35.9-36.2 - 531.4 - 74.5-74.8


531.7
sulfided 35.8-36.0 - 531.6 168.6 74.8
32.0-32.2 163.4
162

K2W04/A1203 non sulfided 36.2 293.6 531.7 - 74.8


sulfided 36.2 293.8 531.8 169.1 74.8
163.6

The various bulk stoichiometries reported in Table IO are compared to the


surface compositions determined by ESCA. It is observed that: (i) the surface
composition in tungsten islequivalent to the bulk one for the W03-A1203 cata-
lyst and distinctly lower for the K2W04-A1203 catalyst, (ii) with the K2W04-
A1203 catalyst there is also a surface over-stoichiometry of potassium (cf nk/
nAl and nk/nw in Table 10).

After sulfidation
(i) For the W03-A1203 catalyst, a peak whose binding energy (32.2 eV) is
characteristic of-tungsten disulfide (WS2) is observed. Based on the relative
intensity of the peaks of each type of tungsten, the sulfidation rate of
tungsten can be considered to be about 30% (see Table 9).
(ii) for the K2W04-A120j catalyst no formation of tungsten sulfide is observed.
(iii) in both cases SO4 - species (EL F 170 eV) whose relative intensity is
of the order of 20% of the total sulfur (Table 9) are observed.
(iv) for the W03-A1203 catalyst the most important S2p peak seems to be composed
TABLE 10
Mass and surface compositions for W-(K)-A1203 samples.

(1001 nk/nA, 0;;) ns/nA, (100) n /n


%'"A1 nk'"w s w
mass surf. mass mass surf. mass surf mass surf.

W03/A1203 non sulfided 0.92 0.81 - - -

sulfided 0.92 1.04 - - 3.4 5.35 - - 3.7 5.14

K2W04/A1203 non sulfided 1.07 0.77 2.15 3.3 - 2 4.28 - -

sulfided 1.07 0.79 2.15 3.3 3.3 4.67 2 4.18 3.08 5.91
321

of two components: one around 162 eV corresponding to S2- species, the other
around 164 eV corresponding very likely to sulfur atoms in short chains Sn
[24-261.
(v) for the K2W04-Al203 catalyst, the S2p peak is more or less symmetrical
corresponding to a binding of 163.6 eV. The sulfur must be present in various
forms-s2-, Sn or an intimate association of tungsten sulfide and amorphous
sulfur.
For both catalysts, a surface enrichment in sulfur (cf ns/nAl and ns/nw
ratios in Table 10) is observed.

TABLE 11
Surface composition of K2W04-A1203 catalysts after sulfidation or after (CO,
H2S, H2) reaction.

K2W04/A1203 I(W)/I(Al) I(O)/I(Al) I(S)tot/I(Al) I(K)/I(Al) I(S)/I(W)

sulfided 0.163 4.89 0.13 0.185 0.786


after reaction 0.200 5.00 0.15 0,176 0.750

After,H$,H?) reaction (for K2WO4-A1203)


Contrary to the preceding results obtained after simple pretreatment with
H2S, a distinct sulfidation of the tungsten is observed after the catalytic
reaction:
(i) in Figure 7 a W4f7,2 peak corresponding to tungsten disulfide (WS2) (ELe32.1
eV) is apparent. However, the sulfidation is not complete as shown by the
6+
shoulder at 36 eV which characterizes W species in an oxygen environment.
(ii) at the same time there is a S2p peak at 162 eV characteristic of the S*-
species (Figure 8).
(iii) finally the position of the K2p. Ols and A12p peaks are not modified.
The various intensity ratios for these peaks are given in Table 11. It is
evident that, except for tungsten, where a definite increase of the I(W)/I(A)
ratio is observed, these ratios do not vary significantly, indicating that the
surface dispersion of the elements is not changed by reaction.
After the catalytic reaction the surface ratio nw/nAl (x 100) is equal to
1 and corresponds to the value of the bulk composition. Thus the reaction would
seem to induce a better surface dispersion of the tungsten.

Conclusion
A study of the surface stoichiometry shows an enrichment in potassium to the
detriment of tungsten in the case of the K2W04/alumina catalyst. Thus, the W-K
328

stiochiometry is not respected and the denomination "K2W04' is purely arbitrary.


After simple pretreatment in H$, the sulfidation of the tungsten is only
partial (30%) with W03/A1203 and non-existent in the presence of potassium,
since it seems that the sulphur deposited is not bound to the tungsten (no WS2,
no shift of the WJf7,2 peak) in the case of the K2WU4/A1203 sample. On the
other hand, the results obtained with K2W04/A1203 after 85 h on stream indicate
that the catalyst undergoes significant modification on contact with the
reaction mixture (CO, H$, HZ). In particular, the sulfidation of the tungsten
takes place during the reaction and seems to be associated with a better
surface dispersion of the tungsten.

CONCLUSION
In this study, it has been demonstrated that carbon monoxide or carbon
dioxide (with hydrogen and hydrogen sulfide) can be directly transformed into
methylmercaptan. This reaction is performed at a temperature and pressure which
make it possible to be free of thermodynamic limitations and in the presence
of a presulfided W-K-A1203 catalyst. This represents a new way for the valori-
zation of acid gas (CO*, H2S) available in natural gas purification plants.
So as to be able to perfect catalysts and to explain the role of each com-
ponent the kinetically important steps were determined. Our study shows up the
intermediate carbonyl sulfide (CDS), whose formation (directly from CO or CO2)
is subject to thermodynamic limitations. Moreover, the hydrogenation of COS
to methylmercaptan turns out to be the limiting step. What is more, various
equilibria between the initial carbon oxide, the COS reactive intermediate and
the main byproduct are involved, the latter being the carbon dioxide obtained

from (CO, H3S. HZ). The rates of some of these reactions can be calculated and
a general scheme for the formation of methylmercaptan from the (CO or COP, H2S,
HZ) mixture is suggested. There is however, an important difference between
the two procedures: with carbon monoxide, the water formed with the methyl-
mercaptan is immediately consumed to form C02; with carbon dioxide, however,
the water formed is not eliminated and, inhibits the formation of COS resulting
in low methylmercaptan production.
This work also showed that the tungsten/potassium ratio influenced catalytic
properties and that a catalyst with a bulk composition corresponding to that of
potassium tungstate seemed more efficient. Nevertheless, characterization of
some catalytic systems before and after the catalytic reaction demonstrated that:
(i) the composition of the optimized catalyst, i.e., K2W04, is not preserved on
the surface since is observed an enrichment in potassium.
(ii) in the presence of potassium, the tungsten does not appear to be affected
by sulfidation, whereas it is partially sulfided in WS2 when potassium is absent
and under the same conditions of pretreatment.
329

(iii) finally, the superficial composition of the catalyst evolves on contact


with the (CO, H2S, H2) mixture and partial sulfidation associated with a better
tungsten dispersion after time on stream is observed. Thus determining the
active sites turns out to be an extremely delicate operation in the case of this
reaction and would seem to require a physico-chemical characterization while
the reaction is taking place,

ACKNOWLEDGEMENTS
The authors gratefully acknowledge the support of Societe Nationale Elf
Aquitaine. Centre de Recherches de LACQ 64170 Artix, France and would also like
to thank Prof. J.P. Bonnelle and L. Gengembre (Laboratoire de Catalyse Homogene
et H&Crog&ne de 1'Universite des Sciences et Techniques de Lille) for performing
ESCA analysis.

REFERENCES

1 a) P. Sabatier and A. Mailhe, C.R. Acad. Sci., 150 (1910), 1217 and 1569.
b) R.L. Kramer and E.E. Reid, J. Amer. Chem. Sot., 43 (1921) 880.
c) M. Constantinescu and T. Constantinescu, Petrol., si gaze, 7 (1959) 298.
d) H.O. Folkins and E.L. Miller, Ind. Eng. Chem., 49 (1957) 241.
e) Pure Oil Company (H.O. Kolkins, E.L. Miller) US patent 2820062 (1958).
f) H.O. Folkins and E.L. Miller, I & E.C. Process Design and Develop., I-4
(1962) 271.
g) H.O. Folkins and E.L. Miller, Proc. Am. Petr. Inst. Sect. III, 42 (1962)
188.
h) M. Miki, T. Ito, M. Hatta and H. Tsuchiya, Yuki Gasei Kagaku Kyokai Shi,
24-6 (1966) 476. (CA: 65-4, 5354g-55d) (1966).
2 Pennwalt Corp. (J.F. Olin, B. Buchholz, B. Loev, R.H and R.H. Goshorn), U.S.
Patent 3 070'632 (1962).
Pennwalt Corp. (6. Buchholz), Belg. Pat. 874 616 (1979).:'
Pennwalt Corp. Eur. Pat, 0104 507 Al.
E.E. Reid, "brganic Chemistry of bivalent sulfur", 4 (1958) 386.
R.J. Ferm, Chem. Rev., 57 (1957) 621.
Stanolind Oil and Gas Companv-. (W.A. Adcock and W.C.Lake), US Pat. 2.767.059
(1956).
8 Agency of Industrial Science and Technology, Tokyo (Y. Kodera, N. Todo and
K. Fukuda). US Pat. 385 6925 (1974).
9 a) K. Fukuda, M. Dokiya and T: Kamkyama, J. Catal., 49 (1977) 379.
b) M. Dokiya, K. Fukuda and H. Yokokawa, Bull. Chem. Sot., Jpn. 51(l) (1978)
150.
C. Otto and G. Ruhl, Ger. Pat. 1 196 624 (1956).
Hamburger Gaswerke G.m.b.H. and Salzgitter Industriebau G.m.b.H. Get-. pat.
1 196 624 (1965).
12 Phillips Petroleum (D.H. Kubicek), US Pat. 4 120 944 (1978).
13 D.R. Stull, E.F.Vestrum and G-C. Sinke, "The Chemical Thermodynamics of
Organic Compounds", Wiley, New York (1969).
14 D.R. Penn, J. Electron. Spectrosc. Relat. Phenom., 9 (1976) 29.
C.D. Wagner, L.E. Davis and W.N. Riggs, Surf. Interf. Anal., 2 (1980) 53.
ii J.H. Scofield, J. Electron Spectrosc. Relat. Phenom., 8 (1976) 129.
17 J. Barrault, M. Guisnet and R. Maurel, Bull. Sot. Chim., (1975) 1592.
18 J. Barrault, M. Guisnet, J. Lucien and R. Maurel, Bull. Sot. Chim., (1977)
362.
19 J. Lucien, J. Barrault, M. Guisnet and R. Maurel, Nouveau Journal de Chimie,
3-l (1979) 15.
20 Y. Amemomiya and G. Pleizier, J. Catal., 76 (1982) 345.
330

21 R. Fiederow, R. Leaute and T.G. Dalla Lana, J. Catal., 85 (1984) 339.


22 S.N.E.A. (P) (M. Boulinguiez, C. Forquy and J. Barrault, Fr 84 10928 (1984).
23 H.P. Bonzel, G. Broden and H.J. Krebs, Appl. Surf. Sci., 16 (1983) 373.
24 K.S. Seshadri, F.E. Massoth and L. Petrakis, J. Catal., 19 (1970) 95.
25 M.L. Jacono, J.L. Verbeek and G.C.A. Schuit, Proc. Vth Int. Cong. Catal.,
2 (1972) 1409.
26 G.C. Stevens and T. Edmonds, J. Catal., 37 (1975) 544.

You might also like