Structure/metathesis-activity relations of silica supported molybdenum and tungsten oxide

Structure/metathesis-activity relations of silica supported

molybdenum and tungsten oxide

Citation for published version (APA):

Thomas, R., Moulijn, J. A., Beer, de, V. H. J., & Medema, J. (1980). Structure/metathesis-activity relations of silica supported molybdenum and tungsten oxide. Journal of Molecular Catalysis, 8(1-3), 161-174.

https://doi.org/10.1016/0304-5102(80)87015-5

DOI:

10.1016/0304-5102(80)87015-5

Document status and date: Published: 01/01/1980 Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:

openaccess@tue.nl

providing details and we will investigate your claim.

doumal ofMokc~&r G~rdysis. 8 (1980) X61- 174

0 E&e-&r Sequok S A., Lausmn e -printed in the Nethedands

161

R. THOMAS, J. A MOULIJN

Irrstitufe for Chemiial Technoloe. LWwrsify of Arrsterdam, P&zntage Muideqgacht 30, 1018 TVAmsterdam (The Nether&nds)

V. H. J. DE BEER

Labordory ofInorganic Chemise ad Cctapbsis. Eindhown iinfwrsity of Tecknolo~. Postbus 513. 5600 MB EiFdkocen (T’ke ?Jefker&nds)

and J_ l&EDEMA

National Defense Reearck Organiza:ion. T.N.0.. Prbzs Matits Laboratory. Lange

Kleiweg 137, 2280 AA R@wijk (Tne Netierlcnds)

In this study the i&hence

of

content

CaMytic activity has been measured in 8 six&annel microfIow reactor. A combination oE stick and metzlthesis a.div-im data leads tx~ the tag- cb-.sian that in these systems the precursors for the catalytic sites in mek- thesis are surface compounds and not the bulk oxides.

The retits mggest that a prerequisite for cat&&k actity is a com-

bination of a high degree of dispersion and an easy reducibilie.

In the

past

The nature of these surface compounds has been the object of several studies. Castellan et al. [4] conclude tirn electronic spectra and the resuks of titration techniques that on Mo03/Si02 dimolyhdate, polymolyhdate and silicomolybdic acid are present. On the basis of Raman spectn, the presence of polymeric species consisting of distorted moiybdate tungstate octa- hedra has been postulated for MoOB/Si02 [ 51 as well as for W03/Si02 [6] . Raman speAra of the c2tiysts in this project have shown &&e presence of three types of molyjdenum compD-unds: crystalline Mo03, octahedrally coor- dinated surface molybdate and a third, 2s yet unidentifed surface compound, having 2 Rzmvl band at about 860 cm-’ [7] _ On W03/Si02 the spectra again show only the presence of two ccmpounds, namely c~stalline tungsten lzioxide and a surface campound, probably consisting of distorted tungstatz oc&hedr~ Clearly, the nature of the surface compound is still a subject of contioversy. Nevertheless it is worthwhile to determine quantitatively the amounts of the various species present at the catalyst surface.

Quantification of Raman specka is in principle possible only when the relative Raman sensitivities, which are determined mainly by the change in polarizibility during the vibration: are known. The relative Raman sensitiv- ities can be determined by applying an internal standard technique, or by determintig the relative amounts of the several species in an independent wa-.. .

From Raman spectrz of powdered mixtures of catalysts and aluminum nitrate as a? interval standard, it has been established that the sensitiviLy for

bulk tungsten trioxide is 5 - 6 times higher than for surface ~J.F ;&& 171. This ratio is in good agreement with results of reduction experiments in a microbzlance [ 83 .

Temperature nrogmmmed reduction (TPR) showed that crysue mol-jbdemum trioxide is 50 - 100 times more Raman sensitive than the surface molybdatzs 171. From these TPR data a quantitative analysis of

MoOa/Si02 catalysts could be derived. A practical problem is that Raman spectroscopy is not in all cases a bulk technique. An example is that some- times crystailites are present at the boundaries of the catiytic particles. This problem can at iezst putially !X solved by pcwdering. Ano31er problem can be that ‘he sensitivities of the various compounds differ to such a degree that the Ranan bands of the compo=md with low Raman sensitivity are obscured by the intense peaks of another compound. It will be shown that this is the case f9r lMoC& /SiO*.

The present stildy is part of a pro&&, which aims to clarify the influence of -sition metal ccntent and the method of preparation on the structure and catalytic activitil of silica and y-alumina supported tungsten oxide and molyhdenum oxide. Therefore catalysts wiL& the sane theoretical molyb- denum and tungsten surface coverage, expressed as the number of metal atoms per square nanometer of the support, have been prepared by wet as well as by dry impregnation. The metathesis activity of the silica supported catalysts has been determined and the re4t.s of the activity measurements are compared with the r4e of structural investigations.

163 Experimental

Details of the preparation of the catalysts and their characterization by Raman spectroscopy, X-my d.if&action (XRD), X-ray fluorescence and B.E.T. are rzporkd ekewhere [7].

TPR of a catalyst -sample containing 0.02 mmol of km&ion metal atoms mis performed in a quark tube (irmer diameter 4.5 mm) using a mixture of hydrogen and nitrogen (66 vol.% I&)_ ALI samples were pre- treated at 773 K in dry ZIr (I h), cooled in vacuum to 473 K and than reduced in tSe E&/N, mixture (flowrate IS cm3/min), while the temperakre was linearly increased up to 1333 K at a rate of five Kelvins per minute. Catalytic activity measurem ents were carried out in a six-channel micro flow reactor consisting of Pyrex tubes (inner diarneteer 3 mm).

After purification by molecular sieves (3A) and an oxygen trap, the feed (pmpenc, Matheson CL?.) was divided into six streams. In this way equal reaction conditions in all reac’k~rs was as.zk.z. The catalysts were

activated overnight in oxygen at 773 K and 1 bar. After 16 h the system was evacuated and flushed three t&es with helium dried over molecular sieves (3A). Then the reactor was cooled down to the desired reaction temperature under vacuz~m (1 h).

During the activity test the propene flow and catalyst weight were approximately the same for all reactors. Because of the identical conditio~is, activities can be reliably compared. The reaction mixture was analyzed by gas chromatography.

Results and discussion

In Fig. 1 the reduction patterns and Raman spectra of molybdenum trioxide and two Mo03/Si02 catalysts are shown. CrystaKine MO& is reduced in severzl steps. ??rorn the aess of the various peaks it m be concluded which _wzction actually takes place. Table 1 shows that the East peak represents the reduction of Ma& to Mo& probably via some titer- mediate oxides; the second peak is due ta the reduction of MoOp to MO& while tie thi~I peak is due to the reduction to metallic MO. These results illuskate that TPR is a quanti&&ive technique with high accuracy. The s=tra21 negative peak at 900 K is attributed to the decomposition of a MO hydride. The TPR patterns of the catalysts are complex. They GUI be deeonvoIut& into three pez!!: a sharp peak at 690 K, a bmad peak &om 590 to 970 K and a superimposed peak at 830 K 173. The last band has been attributed to crystalIke Mot&, in good agreement tith XRD results.

The r&k for the complete series of MoO,/SiO. mtalysts are s~- marked in Table 2. The amount of CrystalLine motybdeuum trioxide increases v&h increasing surf&e novae and is independent of the impregnation technique.

&o

sGTi&

Fig. 1. Reduction patterns and Fhnzn spectra of molybdenum triotide and 13~0 hfOOj/sa~ catalysis.

TABLE 1

hysis of the TPR pattern of molybdenum tioxide Reduction range WI Fraction of total 2~~2 Fxperimcntal Theoreticel- Reaction 500 - 855 0.32 0.33 Moo3 -, Moo2 855 - 1100 0.59 0.57 XI002 + MOJO 1100 - 1200 0.10 O-l@ MojO+Mo

Rom tie Raman spectra alone the conclusion might be drawn that at high surf&e coverages crystalline molybdenum trioxide is &most exclusively present. Clearly these different results are due to tte very high Raman sex&tiviQ of molybdenum frioxide and probably &so t.u a dispropotianate amount of these crystals at the oukide of the primary silica particles.

Similar effects are observed in the TPR- aad E&man patterns of WOJ SiOp catalysts (Fig. 2). The Raman sensitivities for crystAlhe WO3 and surface tigstate differ much less dramatically than for the corresponding MO series; therefore, for WOJ/Si02 Raman spectroscopy gives reasonably

accurate qusnthtive dwk Figure 3 shows the TPE patterns of crystalline

W03 and the W03/Si02 catalysts_

CrystaJine WO, is reduced under these conditions essentially in one

step ai; 770 K. In most catalysts a peak can be observed at III& temperature indic&ing the presence of crgs&_lline W03. In this respect there is a difference betvzen wet aml dry impreg~~&on_ In the case of wet impregnation this peak is observed in all catalysts, while k~ the dry impregmted catalysts it is only present at higher coverages Another difference between the two series

0.8 0.10 1.8 022 3.7 0.46 7.5 0.97 13.1 I.&l 28.9 4.89 Surface moIybd.ate Crystdline MC03 1.00 0

l.ao

'DetetinedbyTPR_ bCakul&edas MoO3. eMonolayer assumed.

Fig. 2 Fig.3 !

!i :i

seems to be present in catalysts with a low tin&en content and prepared

by wet impregnation.

Furthermore, a compound is observed to reduce at a hi&r temperature

then crysta19ine W03 itself. If it is as.smned that this compound represents a surface tungsbte, and if the deconvolution is carried out analogously to Mo03/Si02, the mount of surface and crystalline mateti& can be calculated

(Table 3). The composition of the catiysts determined by Raman spectra- scopy is alw given in this table.

It is evident that at high tungsten coverages a discrepancy exists between the TPR and Raman results. The amount of crystalline tungsten trioxide, as

determined by Raman spectroscopy, is much higher. A priori this Cgkt be

due to a relatively high amount of crysMline tungsten trioxide on the out-

side of the primary silica particles. In principle, XRD gives direct informatio;?

on the amount of crystalline material. In Fig. 4 the intensity of the XRD pea!! (between 26 = 22” aud 25”) is compared with the Raman intensity of

WOJ (.baud at- 8G5 cm -‘I Deviation from a linear relationship would be an

indication of a nonunikm distribution of the cryskllites over the particles.

As this is not the case this explanation does not apply. Obviously, the Raman data are reliable and the interpretation of .&he TPR patterns is not

correct. The amount of crystAke material is underestimated by TPR. This

misinteLpretation might result from ‘ihe fact that lin some catalysts the

crystals are not reduced in one single step. In the literature it has been

reported that this is the case when the w2ter concentration in the HJN-, mixture is high 191. This effect can he expected in catalysts with a high

kngsten coctent as has been discussed earlier for W03/Si02 catiysts [lo].

TABLE 3

Compasiti@n of WO;/SiO* catalysts

Total Wa content Surfaceb Impregn0tion Fractional composition (wt.%) COF3-Zlg~

(W &/Id)

technique

surface tungstate Cry&dine WO3 RaInan TPR Raman TPR ‘1.2 0.09 1.00 1.00 0 0 3.0 O-23 i_OC 1.00 0 0 6.0 0.47 z 1.00 1.00 0 0 6.3 0.53 1.00 l.@O 0 0 13.7 127 z 0.87 0.74 0.13 0.26 39.7 a-90 0.25 0.80 0.75 0.20 1.0 0.07 Wet 1.00 0.97 0 0.03 2.7 0.21 Wet 0.90 0.92 0.10 0.08 5.9 0.46 Wet 0.91 0.90 0.09 0.10 11.6 0.98 Wet 0.56 0.76 0.44 0.24 20.1 1.86 Xet 0.46 0.63 0.5: 0.37 36.9 4.56 Wet 0.32 0.52 0.68 0.48 aCakulaAd as WC3. %omAayerassumed.

1 2 3 4 XRO -inlEnsity

au.

Fig. 4. Relation between XRD intetity and Raman intetity of crystaLline WOj in WO3/.%02 czklysts prepared by wet and dry impregnation.

A

correct

of the

band at a temperature above 770 K also to the reduction of crysklline ‘iVQ3_

Therefore, it was decikkd to use Raman spwtra for the quanti&ation of

FV03/Si02 cataly_&s and TPR patterns for M303iSi02 catalysts.

In condusion it is evident that, on .silicq dispersion of molybdenwn is

much higher than in the tungsten based catalysts.

Afebthesis

ffctivity

The conversion of propene over LMo&/SiO. catalysks has been measured

at 680 K; in general it vzzs less than about X5%. The determination of the

activiQ is compk&ed becawz, as C~JZ be seen from Fig. 5, no conskmt

comersion is reached. -4 deactivation takes place, in p&cular at higher MO

contents.

Mixing the catalyst particIes w3.h silica gave a remarkable improvement of the performance of the catalysts with high MO coverage. This is shown in

Fig. 6 for a hio&/Si02 c&alyst with high MO content. Evidently the con-

version is higher the more silica is added. The effect produced by the addi- tion of silica could be due to various mechanisms. For example, &ca might

remove poisons from the feed; tierefore experiments were carried out witi

silica placed before and after the cz&lyst bed.

F&me 7 &es the results_ Clearly orrEy diLution with s&x gives a

drastic iucrease in activi~. That it is essential for the &Ming material to be

porous is proven by the ewerim ent in which the catalyst is diMed with

non-porous quartz beads w& the Same diameter as the Slica particks.

Dflution with these nonporous beads has essentially no effect. Th- results

sugget that the mechanism is by adsorption of poisotig material formed

Fig. 5_ Conversion of propme over some MoO3/SiO:! catalysb.

Fig. 6. Effect of mixing with silica on conversion of propne over 28 vet.% hZoO3lSiO2; numbera indicate the s?lica/cablyst ratio.

IMking of W03/SiOz catalysts with silica had no influence cn the activity. A tentative erpknation may be derivH1 from the fact that meti oxides such as WOs and MOO3 are reduced by propene, leading to the forma- tion of oxidation products of propene [ll] . Over MuO3, maidy acmlein is formed, ;iifriIe cn W03, pmpene is converted mainly into CO, CO2 and H,O. Morecww, MoOa is reduced to a much larger extent than WO3. As it is known that metathesis catiysts are poisoned by polar sutices, such as water, akohols and aldehydes [12], the influence of silica might be ta trap WirondXG pmpene oxidation products, which are only formed in the case of MoQ,,j siq G3taiyEt.s.

.:i more M e~pl2nation for the effect of silica might k fiat, during &ti~ation, redkpe&on in the bed occurs of the molybdenum oxide, invoking conversion of ELI& into active material. This might be caused by

169

Fig. 7. Analysis of the influence OF dilution ~5th silica of a 28 w-k_% bfoO3/SiO2 catiyst

(silica/calyst r-&o = 10): (a) diluted with sXc2 par&k; (b) silica zfter catalyst bed;

(c) silica beFore c&a&t bed ; (d j csaAnlr_st ocly; (e) diluted titb quark be&s_

Figure 6 shows that, for Mo03/Si02, a decline in activity still fakes

place after dihkion with SiLica Therefore, as the activity of the catalyst is arbitrary the conversion is taken after 200 min.

In Fig. 8 the turnover frequency is plotted as a function of toti MO surface coverage; alI MO at43m.s are assumed to be active. Obvio~&y, the way of impregnation hardly affects the cataEytic activii$. IQ both series of cata- lysts t&e turnover kzquency increases with SurEace coverage up to about 2 MO &oms/nm2 and is constant at higher MO contents.

The activity of W0s/Si02 catalyst.s has been determined at 760 K. Typic&y, a break-in period is observed, folIowed by a coastant conversion level of Less than C5-k In Fig. 9 the tumover frequency is plotted as a function of the surfs coverage, assuming that & W atoms form an active site.

The catalysts prepared by wet impregnation aze slightly more active than those prepared by dry impregnation. These resuks are essentially in agreement witi previous meziurements [IO].

Metathesis catiyst.5 alI show a transient bebax-iaur, a ‘break-in’-period during which the z&t--~ is incrzasin g. Therefore, any relation between met&xsis activi~ and the amount of a cerki~ hantition meti species, determined on catalysts tiich are not yet active in metathesis, o&y indicates that such a .species is a possible pmxrsor for a c&&&c site_

On tie basis of tie described .skuctu& information obtained from TPR and E&man spectmscopy, relations can be obtained between the amount of the various compounds and the tzr-sver Frequencies, caIcu.Iat~~I on the basis

10-%-'

.

:

.

\

I

Fig. 8_ Turnove= frequency in the metathh of pro_wne oser MOO, jSiO2 as a function of the theoretica! surface coverage. For the &cd&ion of the turnover frequency all MO atoms ere asumed t6 be active. m/@ = 1.8 g s molbL.

Fig. 9. Turnover frequency in the mrbathsis of propane over WO3/SiO2 as a function of the theoretical surface ccserage. For the calcu!ation of the turnover frequency dl Vi atomsarea5um edto~encti~.m/~=l_lgsm~l-l.

crystalline molybdenum trioxide clearly shows that the oxide does not

contribute significantly +_43 c&~Iytic activity in met&he& because the

turxover kquency decreases with incretig crysklline oxide content

(Fig_ IO). This conclusion is conErmed by activity me asurements on prve

cry&dine molybdenum krioxide, which did not show metathesis aetivityy.

The relation between tie mover frequency, based on the surface

molyWenum atoms only, and the amount of non4zrystAline moly5date is

given in Fig. 11. TiCs Egure resembles Fig. 8 because of tie relatiely small

amount of crystalline material.

With respect to W03/SiO-,, it is improbable that the compounds detected by TPR at a reduction temperature below the reduction temperature of cryst&ine WO, are precursors for the cata$tic sites. The relative amount of these species decreases at in creasing tungsten content wbSe metithesis activity increases up to at least 20 wt.% W03. CrystaUine W03 itself is ako not a precursor for the active site, since the measured activif;y decrease tim

20 to 40 wt.% W03 is not in agreement with the increase of the amount of

crystalline WOB in this rage. Therefore it seems plausible that the tungsten

species, which are reduced above 800 EC, have a link with tie catiyGc site. The relation between the Camoun~ of these species and the turnover kqueney

Fig. LC. Tcrrnoeer frequency of propene o- MoO3/SiOp catdysk as a function of the

amount of cry&me molybdenum trioxide (kg Mo03/kg oabdyst). The turnover 67e quency calculated OQ basis of cry&ilhe MoOg.

10

1

Fig_ 12. Tcmo-~r frequency of pro_pene over WO~/.SiO2 as a function of non-xyskdine tungstate surface coveraq2. For the c2lculation of the turnover frequency, only ?2 atoms of surEz02 tungstate are assumed to be active.

evident that again the catalyst.s prepued by wet knpregnstion are digidly

more ackive thvl the caklysts prepared by dry impregnation.

As the significvlce of the points in this figure is least at high coverages,

because of the idense bands of crystalline W03 in the Raman spectra, a

horizontal plot is suggested_ Principally a horizontal plot suggests a single site mechanism and equal activity for all sites. The deviation from this picture

for MoOJ/Si02 at 10~ surface coverage (Fig. 11) may be atibuted to

heterogeneity. The fact &at at low MO loadings the turnover kquency is

relatively low, suggests that the favorable influence of diluting with silica is

not caused by redispetion. Redispersion would lead to catalytic particles

with a low surface coverage, which sould contribute little to catiytic

activity.

On the basis of the present information it is not yet possible to give a detailed stmcture of the precursor fc-r catalytic activity. It is however clear that a good dispersion of the transition metal is necessary for its ability to

catiyze the metathesis reaction. In addition, the reducibility of the surface

species is probably an important parameter. A good exampIe of a favorable

combination of high diqersion auci high reducibility is the catalyst Re,O,/

y-Al,O,. The dispersion is complek acd TPR showed that the rhenium

The nakuzt of the catiysts render reaction conditions is still fhe subject offKrfAerre!%earch_

(1) Silica supported molybdenum and tuugsten oxide are composed uf crysee trioxide and surface com~oumis_

(2) The composition can be a_uantit&iveIy determined by temperature programmed reduction and Raman speckoscopy.

(3) The relative amounts of crystalline trioxide increase tit21 the transition metal concenk2tion.

(4) In Mo03/Si02, less cqwkU.ine kioside is formed &an in W03/Si02. (5) Dilution ti$h silica results in 2 considerable improvement in the

performance of Mo03/Si02 catalysts witi high molyhdenwn concentrations_ (6) The activity for both WOs/SiOz and Mo08/Si02 correlates with the amounts of surface compounds and not with the amounts of crystalline

triotide.

(7) The turnover &equency for WQ3/Si02 ca+Aysts is independent of the loading, while for MoOs/SiQ, it increases up to a surface coverage of about 2 MO atoms/m2 and then remains constant.

(8) A prerequisite for catalytic actkty in metAhesis is a combination of high dispersion and low reduction temperature of the transition metal compound.

Thanks are due to M. C_ Mittehneijer-Hazeleger for the TPR measure- ments, M. Vos for the recording of tie Raman speck2 and P. ScheUingerliout for the activity measurements.

Referemes

1 J. C. Mol and J. A. Moulijn, Adu. CatcL, 24 (1975) 131.

2 Froceedina ZXXf-2. Rec. Tmv. Chin. Pavs-Sas. 96 (1977) ml - mi4X. F. P. J. M~Ke&kof,~2Tw.sis, Llr~iuerssity of-&z&f~rk~-l9’T3.

A. CasblGn. 5. C. d. Eat, A. Vagbi and N. Giorthne, 6. CataL. 42 (1976) 162. 5. Medema, C. Van Skm, V. H. J. de Beer, A. J. A. Korziugs and D. C. Koningsbeger, d. CataL. 53 (1978) 386.

R. T~OLUZE., 6. A. Modijn and F. P. d. M. Kerkhof, Rec. w. Chirn. Pays-l&q 96 (1Si7) m134.

R. Thomas, M. C. Mitteheijer-Efaze?egeger, F. P. 6. M. KerkImE, J. A. Eodijn, J. bktema and V_ H_ S. de Eker, Rot_ Third Znf. Conf_ orz The Chemistry azd Uses of Molybdenum. Ann Arbor, hfichig~~~, USA, 1979.

F. P. J. M. Kerkhof, R. Thomas and J. A. Motijn, in E. Dehoo, B. Grange, P. Jacobs sad G. PonceIet (e&J, Rot. 2nd Znf. Synrp. Scientific Bcses for Cite B-epurdio~ of

9 P. Rarret and C. C. Dufour, C.R. Acadd. Sci. Pcnk. 258 (1964) 2 337.

10 F. P. J. M. KerkhoE, R. Thomas end J. A. Modijn, Rec. Tmu. Cfzim. Pays-Bus 96 (1977) m121.

11 i. AM, N. hhashi, N. Ym end T. Seyama, J_ CutaL, 57 (1979) 28’7. 12 R. C. Ranks. Fmh-chr. Chem. Forsch.. 25 (1972) 39.