Gas phase hydroformylation of propylene with porous resin anchored rhodium complexes part II. the catalytic performance

Gas phase hydroformylation of propylene with porous resin anchored rhodium complexes part II. the catalytic performance

331 JoumaZ of Mohcdar Cutaiysis, Ll(l981) 331- 342 @ Ekevier Sequoia SA., Iaxanne - Printed in the Netherlands GAS PHASE J~YDROFORLMYLATION OF PRCIP...

640KB Sizes 1 Downloads 12 Views


JoumaZ of Mohcdar Cutaiysis, Ll(l981) 331- 342 @ Ekevier Sequoia SA., Iaxanne - Printed in the Netherlands





Depatimerrt of Chemical Technology, 2628 BL Delft (The Netherlands)


J. E. DE LEUR Urriuersify


and J. J. F. SCHOLTEN**

of Technology,

Juliana~aan 136.

Hydridocarbonyhris(tiphenylphosphine)rhodium(I), chemically anchored via phosphine or phosphonite ligands to the surface of macroreticular polystyrene divinylbenzene, is successfully applied in the heterogeneous gas phase hydroformylation of propylene at 90 “C and 0.1 MPa total p:es.sure. Catalysts prepared via chlorophosphonation are stable; at a conversion of 0.9% no deactivation is observed over 2 period of more than 5GO h. Catalysts prepared via chloromethylation deactivate slightly, but have a higher selectivity for n-butyraldehyde than those prepared via chlorophosphonation. The hydroformyIation activity Fr unit weight of rhodium, for catalysts with anchored diphenylphosphines prepared via ch!orophosphonation, decreases with increasing phosphorus ligand coverage. Addition of a small amount of tiphenylphosphine to a catalyst with chemically anchored ligands raises the selectivity for n-butyraldehyde formation. The influence of the type of anchored ligand on the catiytic performance will be discussed_

I. Introduction In Part I of this series [I] the preparation and characterization of porous resin-anchored rhodium complexes are discussed. Moreover, the methods of functionahzation of macroreticular polystyrene diviny~benzene with phosphme and phosphonite tigands through chloromethylation and chlorophosphOmtiOn are presented and correlated with the performance of the catalyst in the heterogeneous gas phase hydroformylation of propylene. Very little work has been published on the heterogeneous gas phase hydroformylation of alkenes w-i’& polymer anchored rhodium complexes. Arti -*Prewnt address: Eksachem HoUmd Inc., P-0. Box 1293. Netherlulds. **Author to whom correspondence shbuld be directed.


LS Rozenburg,



studied the hydroformylation of several alkenes with rhodium phosphine complexes chemically anchored to a poiymer coated silica [2] _ He found a very low hydrof~>rmylation ar.tivity, accompanied by an appreciable akane byproduct formation, and n<, information is given on the stabibty of the catalyst. Most other publications on the hetercgeneous gas phase hydroformylation of alkenes deal with the application of supported liquid phase catalysts [3 - 71 and of physically adsorbed catalysts [S - IO] _ Only in the case of supported as the solvent, were liquid phase catalysts with, e.g., triphenylphosphine highly stable, ac Live, and selective catalysts obtained [5 - 71. A possible drawback of such a system in high temperature applications is the gradual evaporation of the solvent, and chemical anchoring of the ligand to the support might prevent this. Another advantage of chemical anchoring of one of the ligands to a support is the suppression of migration of the rhodium complex over the support surface. The difficulty of migration was encountered by Spek in his stildy of physically adsorbed catzlysts 181. In the present paper a study is presented of the catalytic performace of polymer-anchored rhodi-urn complexes at 90 “C and 0.1 MPd in the gas phase hydroformylation of propylene. The effect of the addition of triphenylphosphine to the catalysts will also be discussed.

2 _ Experimental The preparation and characterization of the catalysts, as well as the method of determining their catalytic performance and the chemicals used, are given in Part I of this series [ 1] . The relevant data characteristic of the catalysts are summarized in Table 1.

3. Results and discussion 3.1.


with other perforrnarxe

immobilized catalysts in +the hydroformylation

of propylene, as a function of stream time (tz), for two typical anchored rhodium complexes, uiz., 2 catalyst prepared via chlorophosphonation, and a catalyst prepared via chloromethylation, is shown in Fi g. 1. By way of comparisor the activity and selectivity of XAD-2 supported liquid phase catalyst (SLPcatiyst) with PPhs as the “solvent ligand” are also given (liquid loading = 0.66; 3.78% P, 0.15% Rh; P/Rh = 81.4; particle diameter 0.42 - 0.50 mm). The latter catalyst has been prepared after the method of Gerritsen [S, 71. Figure 1 shows that after 135 h stream time catalyst PS-30 is, per gram of rhodium, five times as active as the SIP-catalyst (SLPW-128) and has twice the activity of PMl. The timedependence of the activity of the c2talysts is remarkable: SLPW-128 is stable during the whole test period and the activation time is short; PS-30 shows an activation of 6090 d&g the first Tine catalytic




The catalysts Catalyst=

Anchored iigand


Anchored (Wt_%)

PS6 PS-13 PS14 PS-15 PS-18 PS-20 PS25” PS-27* PS-30’ PS-31s PS-37* PS4(1 Ps-41 PS-41II

PPh2 P(OMe)z P(OPh)2 PPh;! PPhz PPhp PPha PPh2 PPh2 PPh2 PPh2 CH2PPh2 CH2PPh2 CH2PPh2

0.14 0.05 0.06 0.09 0.09 0.03 0.05 0.17 0.12 0.23 0.17 0.51 0.95 9.95

0.30 0.12 0.13 0.19 0.19 0.07 0.12 0.38 0.26 0.52 0.37 1.14 2.10 2.10


Total PC (wt.%)

[Rh] (epm)

P/Rhd (mol/mol)

0.30 O-19 0.19 0.23 0.89 0.84 0.12 0.38 0.26 0.52 0.37 1.14 2.10 2.19

289 1290 893 496 273 265 172 295 297 575 339 775 1650 1560

25.4 3.1 4.8 12.7 23.1 8.8 21.9 42.3 26.4 30.0 36.3 48.9 42.3 46.5

‘All catalysts have XAD-2 as the support The catalysts marked with Dared from M-2 resin, purified by extraction [ 11. “8, the estimated surface coverage with anchored ligands is calculated 0=


P/g support) .S-

X N,



( 4.9) ( 7.3) ( 15.4) (108.3) (105.3)

( 48.5) are pre-


x S-

of XAD-2

- crosssectional surface area of a PPhz or CHzPPhZ anchored ligand, 0.80 x 10-l” ;Ym = BET surface area of XAD-2. 345 m’/g. N_+ = Avogadro’s number, 6.02 X 10B molecules/mol. =A difference between “total P” and “anchored P” in the Table is due to additional PPh3. added by wet impregnation after coupling of the rhodium complex. dThe chemically anchored phosphorus/rhodium ratio, which is also the total phosphorus/ rhodium ratio without the addition of PPhz. In parentheses, the to&v1 phosphorus/ rhodium ratio is given in the case of additional PPhg_

40 h, and later an additional slow activation of 10%; PS-41 on the other hand deactivates ti 40% of its original activity over the first 40 h and further to 30% &ting the next 200 b. The deactivation of PS-41 is thougbt to be due to a quaternization reaction of excess chloromethyl groups with phospfrine Egands of the complex, as expkned in Park I of this series El]. This poisoning side reactioti proceeds gradually under the reaction conditions during hydroformylation. _ The selectivity of the three catalysts is timedependent, and aldehydes are formed exclusively. Generally speak5ng, with increasing phosphine ligand coverage and with increasing amounts of free PPh, added, the selectivi~ for n-buwraldehyde formation increases. The very high selectiv<~ of the SLP cstAyst is due tu the presence of excess Iiquid PPh, ; further, this excess aLso partly exuIains its lower activi~. Another reason for the Eower activity is the low rhodium effkiency’of the SLP catalyst; the rhodium complexes active in

I pa_ c)F 0


I ‘0




I 120 td

I 160


sLPw-_lze I 200

a I 250


Fig. 1. Catalytic performance. in hydroformylation of propylene, OF N-30 (prepared via chlcrophosphonation, A), of PS-41 (prepared via chloromethylation, 0) and of SLPW-128 (supported liqcid phase system, q)_ T = 90 “C; C3H2/C0 = l/l/l ; P, = 0.1 MPa; W/F = 14.35 (A), 88.56 (a), and 29.95 (a) mg Rh s/cm3 C;_

hydroformylation are mainly located in the liquid PPh,-gas phase boundary. The gradual rise in activity with time, for catalyst PS-30, cannot be ascribed to a change in the nature of the rhodium complex, as the selectivity is constant. Probably, it should be attributed to a gradual improvement of the accessibility of the active sites through surface migration. In Table 2 we compare our anchored systems wi’& various other immobilized rhodiu_m complexes as to r;ctivity and selectivity of propylene hydroformylation. Except for cataly c,l A all measurements are performed at the same temperature and total pressure and data, as provided by the authors, are tabulated. Catalyst A is studied under different conditions at 130 “C and at a feed composition of Cg/H,/CO = 3/3/l. Under these conditions Arai observed a sub&an&i alkane formation. In general, the activity per g of rhodium is much higher for our catalysts than for the other immobilized systems. Especially, Arai’s anchored catalyst A [ 21 shows a much lower activity; after correction for the difference ti temperature and in feed composition, the activity is even lower than indicated in Table 2. The activity of the physiczlIy adsorbed catalyst B, studied by Spek [ 8,111, is 2 - 4 times lower, but its selectivity is about equal. It is of interest to note that rhodium cluster an ZnO (Table 2, C and D) [12] show an activity 15 - SO times lower than that of our systems, but a selectivity of the same order, despite the absence of phosphine ligands. Li$tle can be said about the sfrzbility of the variom systems; most authors do not provide any information on this important aspect. The higher activie level of our chemically anchored catalyst+ is first of aLL *de to the better rhodium dispersion. By anchoring phosp-hine ligands to the


TABLE 2 Comparison of out anchored plexes

with regard


rhodium complexes with other immobilized rhodium to the catalytic performance in hydroformyiation of propylene






90 90 90 230 90 90

0.37 0.19 0.075 0.0065 O-0825 0.0124

47.1 77.3 96.1 22.4 60.4 60 72





PS-30 PS-41 SLPW-128 A B c


=Ps30 PS-41 SLPW-128


26.4= 42.3e 81.4 ? 2 0


0.1 0.1 0.1 0.1 0.1 0.1


= anchored RhIiCO(PPh&



+ f ? +d ? ?

prepared via chlorophosphonation

1 of this paper)_ = anchored RhHCO(PPin& prepared vie chloromethylation of this paper)_ = supported liquid Phase catalyst with LAD-2 as a support the iigand sohent (Pig. 1 of this paper).

= RhCl(CO)2


anchored to polystyrenecoated

(see Table

(see Table


and PPh3 as

silica through phosphine

iigam% [ 21. physically adsorbed on y-alumina 000-3P [S, 111. C = Rh,(CO),Z cluster supported on ZnO [lz]. = Rha(CO),e cluster supported on ZnO [i 2]_ Pn-Bur mol% n-butyraldehyde in the converted part of the propylene. CStability: + = stable; f = slightly deactivating; ? = unknown. d At the higher total pressures of 1.2 end 1.6 MPa this catalyst deactivates [ 9,13 ] . eNote that 28-30 and PS-41 contain chemically anchored PPh, whereas SLPW-128 contains only physically adsorbed and capillary condensed PPhz.


= Rh(ir-allyl)CO(PPh3)2

which, as appears from ow keextie study [ I], ze molecularly dispersed over the surface, all rhodium sites are available for catalytic action. However, anchored phosphine ligands are unable to improve the selectivity for n-butyraldehyde formation to the same high level as does free, physically adsorbed PPh, ; the anchored ligands are localized and unable to coordinate to the rhodium centre. support,


Catalysts prepared vtichloromethylation In Pig. 2 the catalytic performance of three catalysts with anchored dipheny:phosphtie ligands prepared via chloromethylation (see also Table 1) is shown. In the preparation, catalyst PS-40 is refLuxed with [email protected], whereas PS-41 is refluxed with phenytiithium. Despite a higher residual chlorine content in PS41(0.67% Cl as against 0.43% CL in PS-40), Thai catalyst shows a dower rate of deactivatipn than PS-40. Obviously, treatment with phenyilkhium is much more effective than treatment with n-butyIlithium in preventing chloromethyl groups reacting with rhodium phosphine







I 63


I :20 ld

I 160



I 240


Fig. 2. Catalytic performance of catalysts prepved via chloromettiylation. x: PS-40 (0.43% Cl); 0: H-41 (0.67% Cl); 0: PS-41 II (0.66% Cl); T = 90 “C- G/Hz/CO l/l/l ;P, = 0.1 MPa; W/F = 25.92 (x), 88.56 (e), and 78.80 (0) mg Rh s/cm’ Cg_


complexes. The more bulky character of the phenyl group might be responsible for this effect.

For PS-41 the selectivity

is higher than for PS40,


may be ascribed to the higher phosphorus content of PS-41 (2.10% against l-14%)_ After a test period of 400 h (final activity 0.10 cm3 Ci/g Rh s) catalyst was taken from the reactor and treated with a solution of PPhB in diethyl ether (1.2 g in 100 ml) for 0.5 h, in order to enrich the catiyst with a known amount of PPh,. After collection by filtration, the catalyst (now indicated as PS-d,l II) was dried for 1 h at 90 “C in a hydrogen atmosphere. An amount of 0.09% P in the form of PPh, was added. Then PS-41 II was tested in hydroformylation. An initial rise in activity due to PPh3 addition, from 0.1 (final activity of PS-41) to 0.25 cm3 Cz/g Rh s, is found, but the deactivation again proceeds in the same manner as for PS41 (Fig. 2). After 40 h the rate of deactivation is the same for both catalysts. The advantage of the addition of PPhs lies in an increase in selectivity, S (from 3.4 to 5.7). on addition of 2 mol PPh3/mol Rh (see Table 1, last column). PS-41


Catalysts prepared via chlorophosphonatior Six catalysts with anchored diphenylphosphme

ligands were prepared via chlorophosphonation [ 11, with BFs. Eta0 as Friedel-Crafts catalyst, and these systems were &o tested in propylene hydroformylation. Results are shown in Fig. 3 and listed in Table 3 (see also Table 1 for the andyticd data of these ca+Hysts). PS-6 (not shown in Fig. 3), tested for 20 h, shows a stable activie of 0.24 cm3 Ca/g Rh s and a selectivity of 1.0. Tire stability of all catalysts, prepared via chlorophosphonation, is high; ca’dyst PS-27, for insLtance, shows no sign of deactivation at a conversion of 0.9% over a period of more than 200 h (see Fig. 3) but, even after 500 h testing, the activity was still at the same level, and this holds for all other catalysts in Fig. 3. PS-27 was tested in the tempexture range 70 - IGO “C and from the results an apparent activation energy of 33.0 kJ/mol is c&xMed.



Fig. 3. Performance of catalysts = 1/1/1;P, = 0.1 ME%. TABLE Results nation





PS-27 PS-37

S=O85 s=o75




S=O62 I 240

vi2 chlorophosphonztion.

T = 90 ‘C; Cz/H&O

3 of the test of the catalytic





of catalysts





(wt.%) PS-25 PS-30 PS-6 PS-3 7 PS-27 PS-31 asee ‘0

aLso Table = ~timeted

0.05 0.12 0.14 G.17 0.17 0.23

o.i2 C-26 0.30 0.37 0.38 0.52

0.18 0.37 0.24 0.13 0.14 0.033

0.60 0.89 1.0 0.75 0.85 G-62

21.9 26.4 34.5 36.3 42.8 30.0

7.58 14.35 12.8 20.4 15-03 26.04


PPh2 coverage_

Figure 4 shows that above a = 0.1, the final activity per g of rhodium decreases with increasing coverage, 8, whereas the selectivity (Table 3) is low and decreases slightly with increasing 8 (only the activity of PS-25 does not show a clear correIation with 6 _ The phosphorus content of PS-25 is too low for an immediate stabikzation of the catalyst; it deactivates in 30 h to a stable level). For the decrease in activity with increasing 8 no direct explanation is at hand. However, in Table 3 we see that the selectivity decreases slightly with increasing 0, whereas the phosphorus to rhodium ratio does not influence the activity or the seIectivity. The rhodium complex is bound to one anchored Llgard El] ;the decrease in [email protected] with irmeasing 8 czumot be due





OS 1





w ‘/. P 3PPk2

Fig. 1. The final hydroformylation activity per unit weight of rhcdium of catalysts prepared via chlorophosphonation, as a function of 6_ Dashed part: not measured; expected behaviour. The rhodium concentrations in ppm are

indicated in the figure between parentheses for each catalyst. to a change in the number of anchored phosphke ligands coordinated to the rhodium complex, but must be due to a change in the nature of the anchored ligand as a function of 8,. At low the surface of the resin have reacted

8 the more accessible with phosphorus

phenyl groups on trichloride and, hence,

more accessible diphenylphosphine groups are available for coordination to rhodium. At higher values of 0 less accessible anchored ligands are formed which cannot have an ideal coordination to the rhodium, resulting in a lower activity for the hydroformylation reaction. This also explains the slightly lower selectivities observed at higher values of 0. In order to test our catalyst system for a possible application in liquid phase hydrofonnylation, the spent catalyst PS-30 was extracted in a Soxhlet apparatus with zJly1 alcohol under a 1:l hydrogen/carbon monoxide atmosphere. Over a period of 36 h the rhodium content cf the catalyst decreased from 297 ppm to 57 ppm, whereas after 96 h of extraction the rhodium content was only 45 ppm. Hence, despite the stabilizing action of the anchored ligands in gas phase hydroformylation, they are unable to anchor the rhodium complex in liquid-phase applications_ PhthaIocyanine systems are likely to be much more suitable for this purpose [14], but it is expected that due to stronger ar,choring the catalytic activity will be lower.

3.4_ The influence of additional friphenylphosphine In Part I of this series evidence was presented that the rhodium comples is bound to one anchored ligand only [ 1,15]_ Under our hydroformylation conditions the same situation will exist, and this explains the low selectivities observed for the catalysts mentioned in the previous section. It is to be expected that on addition of PPb, a bigher number of phosphine ligands become coordinated to the rhodium complex, and that 2 higher selectivity can be expected:



BF \

-= -(=.



s =I8

Fig. 5. Catalytic performance cf catalysts prepared via chlorophosphonation with additional triphenylphosphine. T = 90 “C; Ci/H#ZO = l/l/l; P, = 0.1 ME%.A: PS-18; 0: PS20; W/F = 13.16 (A) and 13.39 (0) mg Rh s/cm” Cl-

In Section 3.2. we demonskated that the addition of an amount of PPh, molecules equal to twice the number of Rh atoms, to a catalyst prepared via chloromethylation, improves the selectivity and, to a smaller extent Aso, the activity. We then performed the same treatment with two catalysts prepared via chlorophosphonation, though in this case an excess of PPhs was added. The performance of these catalysts is shown in Fig. 5. PS-18 and PS-20 are freshly prepared catalysts, impregnated with a PPhs solution (2.5 g in 100 ml [email protected]) after coupling the complex to the support. In Table 1 the characteristic data of these catalysts are given. Figure 5 shows that, on addition of excess PPh,, very active and se&ctive catalysts are obtained. Although catalysts PS-18 and PS-20 have almost the same to-&l phosphorus content, they differ by a factor of three in activity, which is apparently caused by the difference in anchored phosphme coverage (0.19% P for PSI8 and 0.07% P for PS-20). The dashed line in Fig. 4 suggests that without addition of PPh,, catalyst PS-20 wiJ.lhave a very low (m-&able) activity of 0.06, similar to PS-25, whereas catalyst PS-18 is expected to have an activity of 0.43 cm3 C;/g Rh s. In the case of PS-20 the stzbility of the catalyst is improved upon addition of PPh, ; for both catalysts a remarkable improvement in activity as well as in selectivity is obtained. It is of interest to note &at the addition of two to three PPh, molecules per Rh atom (chem.ica.lly anchored via diphenylphosphine Iigands) to

340 catalyst PS-15 also influences activity and selectivity in a positive sense (uidc infra, Section 3.5). In Part III of this series the improvement in activity and selectivity will be discussed by comparing the performance of the above mentioned catalysts with that of catalysts without anchored phosphine Iigands [Is]. ligcnd In order to study the influence of the type of anchored ligand we synthesized catalys.ts with anchored P(OMe),, P(OPht2 and PPhz (Table 1, catalysts PS-13. PS-14, and PS-15). To all catalysti small amounts of extra PPhs were added (0.04 - 0.07% P), but these additional amounts did not surpass the amount; which can be ligated by the anchored rhodium complex. Figure 6 shows the activim as a function of time. From this Figure we see that both PPh,- and P(OMe).+nchored ligand catalysts reach a stable activity level; P(OPh), deactivates continuously, which is, in our view, due to an The influerrce of the type of anchored



















Fig. 6. The influence of the type of anchored ligand on stability and activity in hydroformylation of propylene. T = 75 “C; Ci/H#ZO = 1:1/l (x, 0, a) and l/1.5/0.5 (0. =); P, = 0.1 MPa. x: PS-13 P(OMe)z, W/F 31.61;0, 0: PS-14 P(OPh)% W/F 13.80; 0, I: PSI 5 PPhl, IV/F 9.26 mg Rh s/cm3 Ci.


Fig. 7. Ortho-melallation phozphonite (bj.

of anchored


(a) and anchored


_ 341

orthometahation reaction of ^&e P(OPh)2 ligand with the rhodium met.aI centre, resulblg in a complex inactive for hydroformylation [17]. For P(OPh)2, orthometaRation goes much more easily than for PPh*, because a less strained complex is obtained (Fig. 7). Et follows from Fig. 6 that, within the framework of this research, diphenylphosphine is the preferred anchored &and, giving the highest activity and selectivity. We have also shown that upon addition of a stoichiometric amount of PPh, a very active catiyst with a sufficiently high selectivity is obtained.

4. Conclusions Rhodium phosphine complexes chemically anchored via phosphine or phosphonite hgands to the surface of XAD-2 macroreticular resins are very attractive as catiysts in the heterogeneous gas phase hydroformylation of propylene at low pressures. They possess a very high activity per gram of rhodium together witb a good stability, especially the types prepared by functionahzation of XAD-2 via chlorophosphonation. Catiysts prepared by functionahzation via chloromethylation will also show a stable behaviour, provided that the remaining chlorine is totally removed. Additional triphenylphosphine improves the activity as well as the selectivity. The addition of only a small amount of PPh, (2 - 4 mo’r PPh,/mol Rh complex) results in a strong increase in activity together with a moderate mcrease in seIectivity . Our chemically anchored catalysts do not display the extremely high selectivities characteristic for SLP catalysts 16, 7). However, their substantialEy higher activity per gram of rhodium, due to a molecular dispersion of the rhodium complexes over the support surface, is an important feature in relation to the rhodium economy, and may well counterbalance the disadvantage of a lower n-aIdehyde selectivity.

Acknowledgments We thank Prof. Dr. W. Drenth (State University at Utrecht, The Netherlands) for stimulating discussions d&12 the preparation of the manuscript, ~Mrs. M. W. Zeelenberg-Tent for carrying out the phosphorus an-Gysis, Mr. P. Bode for the rhodium neutron actiiration analysis, and Mr. P. J. Gommers for. experimental assistance.


F P/Rh

of .symbok Flow of propylene at 0.1 MPa and 25 “C Molar phosphorus to rho&& ratio

cm”/s moi/mol


Total pressure Reaction rate Selectivity = njiso ratio Reaction temperature Weight of rhodium metal in the reactor Surface coverage Avogadro's number Propylene

MPa cm 3 /gRhs mol/mol

°c g molecules/mol

References 1 N. A. de Munck, M. W. Verbruggen and J. J. F. Scholten, J. Mol. Calal., 10 (1981) 313. 2 H. Arai, J. Catal., 51 (1978) 135. 3 P. R. Rony, J. Catal., 14 (1969) 142. 4 P. R. Rony and J. F. Roth, J. Mol. Catal., 1 (1975/76) 13. 5 J. Hjortkjaer, M. S. Scurrell and P. Simonsen, J. Mol. Catal., 6 (1979) 405. 6 L. A. Gerritsen and J. J. F. Scholten, German Patent Appl. 2802276 (1978). 7 L. A. Gerritsen, A. van Meerkerk, M. H. Vreugdenhil and J. J. F. Scholten, J. Mol. Calal., 9 (1980) 139. 8 Th. G. Spek and J. J. F. Scholten, J. Mol. Calal., 3 (1977/78) 81. 9 P. W. H. L. Tjan and J. J. F. Scholten, Proc. Sixth Int. Congr. Catal., The Chemical Soc~ty,London, 1977,p.488. 10 K. K. Robinson, F. E. Paulik, A. Hersman and J. F. Roth, J. Catal., 15 (1969) 245. 11 Th. G. Spek, Ph.D. Thesis, Delft, 1976. 12 M. Ichikawa, J. Catal., 59 (1979) 67. 13 P. W. H. L. Tjan,Ph.D. Thesis, Delft, 1976. 14 E. H. Homeier (DOP), German Patent 2639755 (1977). 15 J. Reed, P. Eisenberger, B. Teo and B. M. Kincaid, J. Am. Chern. Soc., 100 (1978) 2375. 16 N. A. de Munck and J. J. F. Scholten, Part III of this series. 17 E. F. Barefield and G. W. Parshall, Inorg. Chern., 11 (1972) 964.