SiO2 hydrodesulphurization catalysts in their oxidic precursor form

Spectroscopy and Related Phenomena, 17 (1979) 121-135 0 Elsevler Sclentlfuz Pubhshmg Company, Amsterdam - Prmted m The Netherlands

Journal of Electron

XI’S STUDY OF THE SUPPORTED PHASE - S102 INTERACTION IN Mo/SlO, AND CoMo/Sx02 HYDRODESULPHURIZATION CATALYSTS IN THEIR OXIDIC PRECURSOR FORM

P GAJARDO,

D PIROTTE,

C DEFOSSE,

P GRANGE

and B DELMON

Unrversrte Catholzque de Louvazn, Groupe de Physrco-Chlmre Place Crow du Sud 1, 1348 Louvam-la-Neuve (Belgzum)

Mu&ale

et de Catalyse,

(Recenred 11 December 1978)

ABSTRACT XPS results on two series of catalysts, namely Mo/S102 and CoMo/SlOz m the oxldlc form, are reported In the first series the molybdenum oxide content deposited on 6102 ranges from 2 8 to 20 6 weight 96 The second series was obtamed by cobalt lmpregnatlon of the first series, in this case the atomic ratio Co/(Co f MO) was fixed at 0 36 2WS bmdmg energy and mtensltles measurements enable estunatlon of the mteractlon and dlsperslon between the oxtde phase and the support to be made In the case of molybdenum deposited alone on the support, the bmdmg energies of the MO and 0 levels both increase with Moo3 content When cobalt 1s also present on the support, the bmdmg energies of the MO and 0 levels remam constant The relative mtensltles &/1si and I,,/lsl remam constant up to 12% active phsse and then drastically increase for higher amounts of oxide These results confirm the extience of a relatively strong Mo/S102 mteractlon Thrs mteractlon ~8substantially weaker when Co 18supported on the MoSIOz solid, leading to the formation of CoMoOa at the expense of the Mo/SlOz mteractlon Furthermore, It 1s shown that at hwh content of oxide the active phase 1sdeported on the outer surface of the particle and not 11181cle the pores

INTRODUCTION

S&a 1s generally considered to be an mert support m hydrodesulphunzatlon (HDS) catalysts There 16 little evidence of reaction with ions like cobalt or molybdenum, and the diffusion of these ions mto the SIOZ 1s not probable at the common calcmatlon temperatures [l] It thus seemed le@tlmate to consider s&a as a model mert support m comparative studies where catalysts with more active supports, hke Al3 OJ and SIOZ-Al, 03, were considered However, m a senes of mvestlgatlons, mcludmg gravlmetnc studies of hydrogen reduction, ESR and diffuse reflectance spectroscopy measurements [2,3], some expenmental evidence has emerged which

122

Indicates that definite interaction takes place 1n catalysts 1nthe oxide precursor form between MO and SIOz, although, generally speakmg, the support IS relatively inert. It was suggested that this dual character of SIOz 1n1tsbehav10~1:towards molybdenum oxide could be attnbuted to a high heterogeneity of the surface Results reported by Castellan et al. [4] on the Mo/SIOz system are consistent with this hypothesis. These authors observed the formation of compounds such as slllcomolybdlc acid, dunolybdate and tetrahedral Mo(VI) adsorbed on SIOz This paper reports XPS results obtamed on the system, m an attempt to e&mate the magnitude of interactions between molybdenum and S102 If strong lnteractlons really take place, they would manifest themselves by shifts 1n the bmdmg energy (BE) values of electrons of the correspondmg elements We also mvestlgated possible changes takmg place on Mo/S10, catalysts when cobalt 1s added The 1nvestlgat1onsreported here were done on the/same senes of catalysts which were previously studied by other physicochemlcal methods [ 2,3]

EXPERIMENTAL

Support The support, s111ca(Rhone Pro@), has a specific surface area of 179m2 g -l, with a pore volume of 12 ml g-1 The mean dmmeter of the particles was 170a. Catalysts Two senes of catalysts were studled [ 2,3] The first contamed only molybdenum oxide deposited on S102 The second was obtamed by cobalt unpregnatlon of the first senes and had a fixed atomic compos1tlon, r = Co/(Co + MO). Molybdenum was deposited by the pore volume Impregnation method from a solution of ammonium paramolybdate (APM) After drying (llO°C for 1 h) the catalysts were calclned at 500°C for 24 h Cobalt was deposited from a Co(N03 )2 solution onto the correspondmg calcmed Mo/SrO, catalyst. Impregnation, drymg and calc1nat1onwere identical to the correspondmg steps employed for the molybdenum deposition Complete dettis about the method of preparation of the catalysts, and of unsupported Moo,, , CoMo04 and Cog04, as well as about the BET surface determlnation, are presented elsewhere [ 2,3,5] Table 1 (taken from ref 3) reports the charactenst1cs of the catalysts: colour, molybdenum and cobalt content, atomic composition, rat10 r = Co/(Co + MO) and species detected by X-ray analysis. The catalysts have been designated as SMox and S1MoCoy, where x and y indicate the active phase content, expressed as the weight percentage of Moo3 or Moo3 -ICos 04, respectively, m the catalysts

123 TABLE

1

ACTIVE PHASE CONTENT [(a) EXPRESSED AS WEIGHT% OF Moo3 AND MoOB + Co304 IN THE SlMox AND SlMoCoy CATALYSTS SERIES, RESPECTIVELY], ATOMIC COMPOSITION, BET SPECIFIC SURFACE AREA, COLOUR AND COMPOUNDS DETECTED BY X-RAY ANALYSIS IN S&lox AND SrMoCoy CATALYSTS* Catalyst

Phase actwe content wt % (a)

co Co + MO

Surface FzTg-,

Colour )

X-ray analysis Moo3

Cog O4

b-CoMoO.,

-

-

SiMo28

28

145

Bluish

no

&MO 6 7

67

143

Bluish

ll0

SlMo 116

116

122

Bluish

S&lo 14 8

14 8

118

S&lo 20 6

20 6

0

-

-

yes

-

-

Light blue

yes

-

-

121

Lqjht blue

yes

-

-

3 7

38

0 37

135

Light violet

no

no

no

SlCoMo 8 7

87

0 35

129

Gray violet

no

no

no

SlCoMo 12 3

12 3

0 36

122

Gray

no

no

yes

SlCoMo 19 8

19 8

0 36

105

Gray

yes

no

yes

SlCoMo 27 6

27 6

0 36

97

Gray

yes

no

yes

&CoMo

a This catalyst group IS part of the catalysts studied previously [3] by other methods &MO 11 6 and SlMoCo 12 3 me the same catalysts mvestlgated m previous work [ 21, where they were dessnated as SI 0 00 and S10 36, respectively

XPS Apparatus and experrmental procedure The measurements were performed usmg a Vacuum Generators ESCAZ apparatus. A Tracer Northern NS 560 srgnal averager was used to nnprove the srgnal-to-noise ratio. The samples were dusted onto doublesided adhesive tape and mtroduced into the preparatlon chamber Method and condltrons of spectra recordmg are the same as those employed m a previous work 121. Spectra of the CL,, 01,, SrZp, Mojd, CozP3,2 and Audf levels were recorded. In order to obtam reproducible results, a stnct standardlzatlon of the order and tune of recordmg was used. The AQ~,,~ level, wrth a bmdmg energy (BE) of 82.8 eV, was used as reference lme. In order to evaluate the chargmg effect and, d necessary, to correct the BE values, spectra were recorded several ties. When quantltatwe analysis was necessary, spectra were recorded before gold evaporation.

124

Decomposttron of spectra. The decompoetlon of the spectra mto the mdlvldual peaks was camed out by a special computer program, followmg the method employed m a previous work, [ 21. This method rests on the followmg assumptions and constramts (1) the shape of lmes ISgaussmn, (11) the background due to melastlcally scattered electrons 1s assumed to vary lmearly with the BE, (m) the lme mtenslty 1s taken as the normalized area under the gausslan peak, (iv) the Mojd hne 1s decomposed using the followmg constramts - MO3d5,2 and Mo3d3,2 peaks have the same full width at half maxlmum (FWHM) values, their maxima are separated by 3 2eV and then mtenslty ratio (MoJcr5,2 Mo~~~,~) 1s equal to 1 54 (this value was taken from the Mogd spectrum of unsupported Moo3 ), (v) Ols, SlqP and Co2P3,2 (mcludmg prmclpal and satellite peaks of CozP3,2 ) lines are decomposed wlthout any additional constrwnts In some cases, the peaks of the Olo, SlzP, CozP and Mofd levels were clearly asymmetrical, suggestmg the presence of several components. In pnnh clple, such asymmetrical peaks can be decomposed mto several components, however, this was only done m the case of the 01, level, because the presence of several well defined oxygen species was evident from X-ray analysts (Table 1). In this case, if the mtenslty of one component became very low, the decomposltlon was performed without this component, and m consequence the FWHM of the mam peaks was larger than the normal values The Mojd doublet often came out badly resolved, suggestmg the presence of at least two molybdenum species with different BE values We did not try to decompose the species, because such decomposltlon would only be meaningful d a lnnzted number of species 1s present, not if a wide spectrum of different species exist. For the same reason we did not decompose the Slap level

RESULTS

Spectra and BE Table 2 aves the BE and FWHM (m parentheses) values of the Mo~~~,~, 111Table 2 means that BE and Sl2P , 01, and Co2P3,2 levels “Undted.” FWHM were undetermmed, because the spectra were unresolved or the intensities of the signal were too low. S2Mox senes SrZp The iQzP lme 1s asymmetrical,,›revealmg the existence of at least two species of Sl with different BE values (Fig la) This peak was not decomposed, for the reason discussed above However, the BE values at the maxuna of the Slzp p eaks were calculated These values are plotted agamst molybdenum oxide content m Fig 2.

125 TABLE 2 BINDING ENERGIES AND FWHM IN CATALYST FWHM ARE GIVEN IN PARENTHESES Catalyst

Mow/z

s12p

01,

AND MODEL

COMPOUNDS

a

(eV)

cozrrw2

ho,

OMoO

B

undted

-

-

OCoMoO,

SlMo28

undted

1012 (3 3)

530 5 (3 3)

S&lo67

undted

1014 (3 3)

5312 (3 4)

undted

-

-

S&lo

undted

1017 (3 3)

5316 (2 9)

529 6 (2 9)

-

-

&MO 14 8

230 7 (2 8)

102 0 (3 2)

532 0 (2 9)

529 6 (2 9)

-

-

&MO 20 6

230 9 (3 1)

102 3 (3 3)

532 8 (2 8)

629 8 (2 8)

-

-

SlMoCo 3 7

und ted

102 1 13 4)

5317 (3 5)

undted

undted

undted

SlMoCo 8 7

undted

102 2 (3 1)

5316 (3 6)

undted

undted

undted

SlMoCo 12 3

undted

102 3 (3 2)

5317 (3 4)

undted

undted

undted

SLMOCO 19 8

2313 (3 6)

102 0 (3 4)

5316 (2 9)

529 8 (2 91

527 1 (2 9)

780 2 (4 4)

SlMoCo 27 6

230 9 (2 71 231 1 (2 8)

5316 (2 9) -

529 7 (2 9)

Moo3

102 0 (3 4) -

780 8 (4 6) -

a-CoMo04

231 1 (2 5) -

-

-

630 1 (2 9) -

527 8 (2 9) 529 1 (2 9)

779 7 (4 0)

-

-

-

-

1012 (3 21

530 6 (2 9)

-

528 8 (2 51 -

778 3 (2 3) -

-304 s102

116

a Osio, , 0~~0, and 0~~~~0, are the components of the 01, levels, ascrlbed to 0 m S102, MoOa and CoMoO4, respectively (see text)

This peak presents an asymmetry which mcreases with the molyb018 denum content (Fig. lb). Thle lme can be well decomposed mto two peaks 111catalysts w&h x > 11.6. The BE value of the two 01, peaks are plotted, m Fig. 3, agamst molybdenum content, and compared mth those obtamed with pure S102 and

126

Fig 1 SlzP (a) and 01, (b) hnes of &MO 14 8 catalyst

101

IOC

99

0

6

12 18

Mo4(g)

16

24

I catalyst(g)

Fig 2 Dependence of the BE of the 512, level on the MOOS content (S&lox series)

Moo3 (x = 0.0 and 100 0, respectively) The first (mgher) BE value mcreases rapldly wrth the molybdenum content, whereas the second value 1s almost constant. M%d Figure 4 illustrates the evolution of the shape of the MoJd lme m catalysts containmg different amounts of molybdenum. The Mojd doublet 1s clearly resolved only m catalysts with a high molybdenum content (x = 20 6). The poor or nonexlstant resolutron at low molybdenum content mdrcates the presence of several molybdenum species with different BEvalues, The same catalysts m the uncalcmed state (after drying at llO°C) exhibit a very clear MO36 doublet, mdlcatmg the presence of only one molybdenum species.

127

to2 MoOa (g) /catalyst

(9)

Fig 3 Dependence of the BE of the 01, levels on the MOOS content (SMox series) (a) BE of 01, ascribed to 0 m SlO2, (b) BE of 01~ ascrlbed to 0 m supported Moo3 (o ) BE of OIs of unsupported MoOa

I

2250

I

2500

I

2350

1

2400

Fig 4 Evolution of the Mojd hne shape m S&lox catalysts (a) SrMo 20 6, (b) SiMo 116, (c) SJMO 2 8

a

co 2p

b

51 2p

Fig 5 Computer decomposrtlon of MO 3d 27 6 catalyst

SlMoCo

8

103

Cozp

@A

1I

01,

(~1,

and

S1zp

(d)

spectra

of

51 2p

8

5 ;

(a),

-

102

l

P f 101

0

5

10

15

ld

Fa 6 Dependence catalysts)

.

actlva

.

25

2-o

phase

(0 1 I

catalyst

30

(0)

of the BE of the SIDE level on the actwe phase content (SIMoCoy

129

SaMoCo y series SizP The S12P signal H sh&tiy asymmetrical,,›as it was in the SlMox senes (Fig. 5d) The correspondmg BE values measured at the maxmum of the peak are plotted m Fig. 6 vs the amount of active phase, the BE value of pure silica is also reported. The Slzp BE is mdependent of the active phase content y ; the value is higher than that found m Sr02 . The Oi, lme of catalysts havmg a high active phase loadmg shows a 01, strong asymmetry. A decomposition mto two peaks did not gwe any good fittmg between experimental and calculated curves Furthermore, the FWHM was larger than the normal value, suggesting the presence of another component. Good results were obtamed with a three-peak decomposition for a catalyst of active phase content higher than 12.3% (Fig 5~). In catalysts with lower molybdenum content, the spectrum is satisfactorily accounted for by only one peak. The BE dependence of the Oi, level of the v8z1ous species on the catalyst molybdenum content ls reported m Fig 7. The BE value of the mam component (curve a) ls independent of the amount of active phase y The Oi, BE value of the model compounds SiOz , Co3 04, a-CoMo04 and MoOa are also reported m Fig 7. Mojd. In parallel with the situation m the S&lox senes, the Mojd doublet is well resolved only 11~catalysts with a high active phase content (Fig. 5a). In catalysts with a smaller loadmg of active phase, the spectrum suggests the superposition of various molybdenum species havmg different BE values. signals are very weak. The spectrum Co,, The co2~3,2 and cO2,1/2 decomposition was only possible at high active phase content (y = 19 8 and 532’ a 531 2

n

B 530i-9 z

r"529 _ii c m

t

A

Co30,

D

CoMoo4

OMooj

520 -

527

I

5

I

I

10

15 lo*

actlve

20 phase(g)

25 lcatatyst(g)

A

100

Fg 7 Dependence of the BE of 01, levels on the active phase content (SMoCoy catalyst) BE value of 01, levels for C&O.+, aCoMo04 and MOOS are given at y = 100 (A) Co304,(o)MoO~,(o)aCoMo04

130

27 6), where the presence of only one cobalt species was observed. Figure 5b shows spectra of CO,,~,, and Co2P1,2 bands, each of them followed by a shake-up satelhte Relative mtensrtws of Mo3d, c02p3,2 and SzzD sgnals The relative mtensltles, expressed as I Mo/Isi and Ic, /Isi, were determined m both catalyst senes These values are plotted agamst active phase content m Figs 8 and 9, correspondmg to the SMox and SlMoCoy selples,respectlvely . The I Mo/Isi values remam more or less constant up to a loadmg of “12% III both semes of catalysts At an active phase content higher than 12, a 2 I

MO

I SI 15

10

20 ld

F@

8 &/&

(g)

MO%

(gl

/catalyst

vs amount of MoOB (x) deposited on the catalyst (Srlwox catalysts) T

3

3

ha IS,

kQ

IS,

2

.2

1, %I

0

0 10

Id aCtWe

:

30

PtlSG (Q) /Gif?iYJt (Q)

Fig 9 IM,/Is~ ad IcO/lsi (SlMoCoy catalysts)

vs amount MOOS + Co3O.q (y) depos&d

on the catalysts

131

steep build-up of the curve IS observed (Figs. 8 and 9). The Ic,/&i ratio presents a snnllar dependence on the active phase content m SlMoCoy catalysts (Fig 9).

DISCUsSION

Molybdenum-EMI2 mteractwn In previous work [!2, 31, we mdlcated that the S102 surface UJrelatively inert mth respect to MO(W) when it 1s compared to the r-Al2 O3 surface, but It seems highly heterogeneous. It was suggested that, owmg to this heterogeneity, the srmultaneous presence of strongly and weakly adsorbed Mo(V1) species on SIOz 1s quite probable. SzMox. The above hypothesis 1~iconfurned by the XPS results reported in this work. A strong mdlcatlon of this heterogeneity comes from the changes that the Mosd hne undergoes when the molybdenum content mcreases (Fig 4). For example, the absence of resolution of the Mosd doublet m S&IO 2.8 (Fig. 4c) can only be understood by the sunultaneous presence of at least two, and presumably more, molybdenum species wth different BE values. At 3c= 116, the Mojd doublet becomes more resolved (curve b), and at x = 20 6 it becomes qmte clear (curve a). The mterpretatlon 1s that the contnbutlon of the Moo3 phase to the MO3d spectrum becomes progressively more nnportant, m agreement vvlth the fact that X-ray dlffractlon data detect Moo3 m the correspondmg catalysts (Table 1). This mterpretatron ls also supported by the fact that the BE values of the Mo3d5,2 levels m catalysts vvlth 3c> 14.8 and MoOa are sun&~ (Table 2). A quite surpnslng fact 18 the steep lmear mcrease of BE of the SlzP level mth molybdenum content (Fig 2). This mcrease corresponds to an unexpectedly high varlatlon of 1.1 eV between SIOz and S&lo 20.6 It 1s difficult to interpret this linear dependence m terms of a progressive displacement of the electron density from Sl to MO. A possible explanation 16 that various Sl species (at least two of them, separated by 1 1 eV) with different BE values are present. This ls cons-tent with the asymmetical shape of the SlzP lme If the relative amount of these species depends on the molybdenum content of the catalyst, the BE value of the envelope of the Slzr, peak will clearly depend on the molybdenum loadmg. If this picture 1s correct, we must expect the presence of vmous molybdenum species correspondmg to the vmous kmds of mteractlon unth Sl to be detected by displacement of the Slzp level, there should be at least two species of MO, and these must have a difference m BE snnilar to that observed for the Slz2, level (i.e. 1.1 eV) The broadenmg of the MoJd lme (Fig. 4) clearly mdlcates the posslblllty that the observed Mojd hne contams (at least) two normal MO3d doublets separated by 1 .l eV

132

Accordmg to our explanation, the Si species, with low BE, should be ascribed to s102; the other species, with the hrgher BE values, mrght be assigned to Sl m strong mteraction with MO The results reported here are quite surprismg. Alternative explanations should therefore be considered The vmation observed m Fig. 2 might have its ongm m a physical side phenomenon dependmg on the molybdenum loadmg. The conductlvrty of the sample, for mstance, mcreases with the molybdenum content, probably mducmg a change m the charge effect durmg XPS measurement, from sample to sample Charge effect was observed and it was always estimated by measurmg the BE of the level studied III relation to the standard which was recorded immediately before and after the corresponding spectrum. The standard was Au~~,,~ (or Cls, the BE of which was measured, m turn, m relation to Au~~,,~ ) The 01, peaks are strongly asymmetrical for loadmgs with x > 12. They were resolved m two components (Fig. lb and Fig. 3a,b). The peak placed at the low BE side was attributed to 01, 111 MoOJ , because the BE value 1s slmrlar to that observed in unsupported Moos and because this component 1s almost msensrtlve to molybdenum content. This is quite loglcal m the frame of our asslgnatlon. In contrast, the OJ, BE of the mam component, placed on the higher energy side, depends on the molybdenum content (Fig 3a). This dependence can be explamed, assummg, m conformity mth the previous discussion, that curve (a) corresponds to a multicomponent signal, caused by a spectra of several specres, the proportion of which varres with catalysts loadmg. It should be noted that the smgle signal at 3c< 12 UJ asymmetrical, thus suggesting the presence of another species (in minor proportion) m the corresponding samples. Retummg to the discussion of the high BE value of the 01, species, it IS difficult to assrgn this signal to a definite oxygen species because of the presumed h@ heterogeneity of the surface One may sunply speculate that the mam component of the peak should correspond to 0 m the normal SrOz surface and whrle the other corresponds to 0 m an Sr-MO(W) complex or SiMo(V1) complexes [ 2] The broadening of the MOJd hne (Fig. 4) and the apparent shift of BE values of the Slzp and 01, levels w&h MO content (Figs. 2 and 3, curve a), are, generally speakmg, a strong mdicatlon of the high diversity in strength of the interaction between Mo(V1) and the SIOz surface. More specifically, it could also mdxcate the simultaneous presence of various MO and MO-& compounds. The presence of compounds such as &co-molybdlc acid (SMA), dnnolybdate, polymolybdate and MOO, in Mo/SlOs catalysts has been reported by Cast&an et al [4] and by Prahaud [6] Tl~s kind of compound can be formed at the molybdenum unpregnation stage Actually, owmg to the fact that amorphous Si02 is soluble m water to the extent of about 0 01% (expressed as S102 ) 171, the addition of the Mo(VI) ion mto the solutron

133

could bnng about the formation of the complex Si(Mo, 2040)4- [8]. Dunng the drymg stage, this complex would be deposited on the surface; Its unportante, m terms of monolayer coverage, might be considerable. This kmd of complex would be decomposed m the calcmatlon step For example, according to Prahaud [6], the heatmg of SMA (formed on the S10, surface) at 500°C under an oxygen atmosphere leads to the mcorporatron of Mo(V1) mto Si02 vacancies. Our results and the previous discussion should be put m relation to the broadenmg of the W4f lme observed by Blloen and Pott 191 on oxide W/SiOz catalyst, whrch exhlblts much sumhtude vnth our Mo/Si02 system These authors have shown that 30% of the intensity of the W4f lme shows up at a low BE which 1s ” 2 eV below the mam lme SlMoCoy The addition of cobalt to SlMox catalysts bnngs about the formation of b-CoMo04 This compound was detected by X-ray analysis at y > 12 (Table I), but a compound which can be described as a cobalt molybdate is formed m all catalysts [ 2,3]. Cog O4 1s absent as confumed by the BE of the Co2p3,2 level of catalysts (Table 2), they are slgnlflcantly different from the BE of unsupported Co304 The myor part of the cobalt reacts with the molybdenum formmg cobalt molybdates A conspicuous difference between the SlMox and the SlMoCoy senes 1s the stability of the Slpp BE value m the latter senes (Fig 7) compared to the large vanation mth MO content m the former (Fig. 2) If the vanation m the case of the SlMox senes has to be attnbuted to mteractlon between Si02 and deposited MO, this would suggest that the introduction of cobalt provokes the destruction of the SiOz -MO assoclatlon. This corroborates the formation of an association of Si and MO the fonnatlon of CoMo04 occurs at the expense of the SIOz -MO association The peak conespondmg to the Oi, level was decomposed mto three components (Fig 5~). The mam one (Fig. 7, curve a), situated on the high BE side, must be attributed to the support because it 1s present m catalysts with a low active phase content (m the case of the SIMOX senes, It 18 also the high BE component which was attnbuted to the support). The second component (curve b) can be assigned to oxygen m Moos, it has BE values sunllar to those of MOO, (Fig 7) and of the component correspondmg to curve (b) m Fig 3, for SlMox catalysts, and also assigned to MoOa. The third component, which IS the smallest one, 1s situated on the low BE side, it must be attnbuted to 0 m cobalt molybdates The second and thud components only show up clearly at y > 19 8, 1 e when the presence of both MOO, and CoMoO, is evident from X-ray analysis. The 0 Is BE of the third component 1s lower than the BE of a-CoMoO,, this difference may be due either to errors made m decomposition procedure of the spectra or to a contnbution from other cobalt molybdate species. One has to note that both a- and b-CoMo04 were detected by electron dlffractlon study [12] and sulphuratlon reduction

134

measurements [13]. However the BE of the a- and b-CoMo04 phases are nearly the same. The BE values of the O,, level of the mam component do not vary mth actwe phase load (m contrast to the steep positwe slope of the same lme m the SlMox senes (Fig 3a) which we interpreted to be due to the existence of a relatively strong &OZ -MO mteractlon). This fact suggests the absence of perturbation of the oxygen of SIOZ by the deposrted species, and, consequently, the absence of SIOZ-MO mteractron m the SlMoCox catalysts. This result corroborates the above concluaon of the dlscussslon concermng the SIZp hne the addition of cobalt mduces a suppression of the SIC&-MO rnteractions Relu twe m tens1ttes

In other studres of our group, we have used measurements of the relative mtensltles of peaks to gam valuable mformatlon about the dlstnbutlon of the actwe phase on the support [2,10,11] The constant value of the &, /Isi and IcO/Isi ratios up to x and y equal to 12 (Figs 8 and 9) suggests that the species contammg MO and/or Co increase m particle size when the total active phase load mcreases The sudden rarsmg of the curves at 3e= y = 12 clearly mdlcates either a change of the site or a change of the overall repartltlons III the mdlvldual catalyst particles A probable mterpretatlon 1s that, for x and y hrgher than 12, mcreasmg portlons of the active phase are deposited at the outer surface of the particle and not mslde the pores. Electron mlcroscopy measurements [12] on these catalysts support this hypothesis

CONCLUSIONS

The results reported m this work confirm the existence of the relatively strong SiOZ -MO mteractlon assumed to take place m these sohds. This mteraction 1s substantially weaker than CO-MO mteractlon The formation of cobalt molybdates IS thus possible, and takes place at the expense of the SlOz -MO association The Inverse picture may explam the absence of cobalt molybdate on an “active” support such as Al2 O3 [ 5,141, In this case the molybdenum-upport mteractron 1s stronger than the one between Co and MO, hmdenng the formatIon of CoMo04, m such condrtlons molybdenum remams attached to the support as a monolayer

ACKNOWLEDGEMENTS

The authors gratefully acknowledge fmanclal support from the “Services de la Programmatlon de la Pohtlque Sclentlflque” m the framework of the

135

“Actions Concert&es Interunlversrtalres de Catalyse” One of us (P Ga ) acknowledges support from C.T.M. durmg the course of this mvestlgatlon and C D thanks the Natlonal Fund for Sclentlflc Research (F.N R S., Belgmm) for an “Aspirant Fellowship”.

REFERENCES 1

8 9 10 11 12 13 14

G T Pott and W H J Stork, m B Delmon, P A Jacobs and G Poncelet, (Eds ), Preparation of Catalysts, Elsevler, Amsterdam, 1976, p 537 P GaJardo, P Grange and B Delmon, J Phys Chem , m press P GaJardo, D Plrotte, P Grange and B Delmon, Trans Faraday Sot , m press A Castellan, J C J Bart, A Vaghl and N Glordano, J Catal ,42 (1976) 162 P Galardo, P Grange and B Delmon, submltted for pubhcatron H Prahaud, J LessCommon Metals, 54 (1977) 387 R K Iler, m E Matljevlc (Ed ), Surface and CoIlold Science, Vol 6, Wiley, New York, 1973, p 3 G Charlot and D Bdzler, Les M6thodes de la Chlmle Analytlque Analyse Quantltatnre Mmerale, Masson et Cle, Paris, 1955, p 680 P Blloen and G T Pott, J Catal , 30 (1973) 169 P Galardo, P Grange and B Delmon, submitted to J Catal C Defos&, P Canesson, P G Rouxhet and B Delmon, J Catal ,51(1978) 269 F Delannay, P GaJardo and P Grange, J Mlcrosc Spectrosc Electron, 3 (1978) 411 D Plrotte, M&moire, Untverslte Cathohque de Louvam (1978) P Gajsrdo, Ph D Thesis, Umverslte Cathohque de Louvam (1978)