Catalytic activity of some homogeneous and silica-bound dirhodium complexes having one bridging thiolato and one bridging chloro ligand

Journal of Molecular

Catalysis,

31 (1985) 317 - 326

317

CATALYTIC ACTIVITY OF SOME HOMOGENEOUS AND SILICA-BOUND DIRHODIUM COMPLEXES HAVING ONE BRIDGING THIOLATO AND ONE BRIDGING CHLORO LIGAND MORIS EISEN, JOCHANAN BLUM* Department

of Organic Chemistry,

The Hebrew

University,

Jerusalem 91904

(Israel)

HERBERT SCHUMANN* and STANISLAW JURGIS Institute of Inorganic 12 (F.R.G.)

and Analytical

(Received July 30,1984;

Chemistry,

The Technical

University,

D-l 000 Berlin

accepted October 31,1984)

Summary Several E.c-alkylthiolato- and ~-arylthiolato-~chlorodicarbonylbis(tri-tbutylphosphine)dirhodium complexes, Rh,(C0)2(P[C(CHs)s]s)~(~-Cl)(pSR) 3 - 8, were shown to be effective catalysts for hydrogenation, hydroformylation and isomerization of olefins. Silica-attached versions of these complexes, in which the sulfur was linked to the support by a (CH,),SiOs (n = 2, 3) bridge, could be recycled for use in further runs. Both the homogeneous and the heterogenized catalysts were shown to retain the dinuclear structure during and after the catalytic processes.

Introduction Poilblanc has recently indicated [l] the potential importance of homobimetallic catalysts in organic synthesis. It appears, however, that at present only very few complexes are known that both retain a binuclear structure throughout the process and possess good catalytic activity [2]. Thus e.g., most (p-C1)Z-dirhodium compounds are cleaved into monorhodium species in the presence of nucleophiles [3], while those having two stabilizing thiolato bridges react irreversibly with the common reactants [ 41. Therefore we found it of interest to investigate the catalytic power of dinuclear complexes in which the two metals are linked by one bridging Cl and one bridging SR ligand. We have already shown that dirhodium complexes of type 1 can be synthesized from chlorodicarbonylrhodium dimer (2) provided R is a bulky tri-t-butylphosphine [5, 61 or -arsine group [ 71. By treatment of 2 with the phosphine followed by reaction with the appropriate trimethylsilyl thio *Authors to whom correspondence 0304-5102/85/$3.30

should be addressed. 0 Elsevier Sequoia/Printed in The Netherlands

318

ether, compounds 3 - 10 were obtained. Preliminary experiments that these compounds

IWH3)318,

,Cl,

Rh oc’

indicated

,WWH3)313

\SIRhLCO B

3, R = (CH&CH 4, R = (CH&C

7, R = (CZH50)3Si(CH& 8, R = (CH30)3Si(CH&

5, R = 4-CH3CsH4 6, R = 4-CICsH4

9, R = silica-(O~SiCH&!Hz) 10, R I silica-(O3SiCH&!H2CH2)

promote certain isomerization and transfer reduction reactions at rates comparable to those of conventional mononuclear catalysts [ 61. We have now extended” this study exploring the activities of the pchloro+hiolato dirhodium complexes, both in homogeneous and in silicabound versions, in certain hydrogenation, hydroformylation, and double bond migration processes, (e.g. eqns. (1) - (3)). C6H10 + H2 C6Hlo

+

(1)

C6H12

H2 + CO -

C6HSCH2CH=CH2 -

C6HllCH0 cis- and trans-C6H&H=CHCH3

(2) (3)

Experimental

The homogeneous catalyst cis-Rh2(CO)2(P[C(CH3)3]3)2(~-Cl)(~-SR) where R = (CH3)2CH (3), (CH3)sC (4), 4-CH3CQ-I~(S), 4-Cl&H, (6), (CzH,O)3Si(CH2)2 (7), (CH30)3Si(CH2)3 (8), and the silica-bound complexes where R = silica-(03SiCH2CH2) (9) and R = silica-(03SiCH2CH2CH2) (10) were prepared essentially as described previously [ 5,6]. Hydrogenation of olefins

In a typical experiment, one part of a divided microhydrogenation apparatus was charged under argon with 2.19 X 10M5mol of the catalyst, cooled to -78 “C and evacuated (20 mm). With the aid of an airtight syringe, a solution of 7 X 10m3 mol of freshly distilled cyclohexene in 3.0 ml of degassed toluene was injected into the second part of the reaction vessel. The cooled flask was evacuated, washed three times with hydrogen and immersed

319

into a thermostat at 50 i: 0.5 “C. After temperature equilibration and adjustment of the pressure to 690 mm, the components were mixed and stirred constantly at the same pressure and temperature. Samples were withdrawn periodically from the reaction mixture and analyzed on a GC column packed either with 10% @,P’-oxydipropionitrile on Chromosorb W or with 10% 1,2,3-tris(2-cyanoethoxy)propane on Chromosorb G. Upon completion of the process in which soluble catalysts were employed, the cyclohexane and toluene were removed from the metal compound by vacuum distillation under strict exclusion of air and the residue was applied immediately in the second run. The silica-bound catalysts were separated from the reaction mixture by decantation under argon. After washing with toluene and drying at 0.5 mm, the beads were stored under anaerobic conditions to await recycling. Hydroformylation of cyclohexene Typically, a mixture of 2.19 X 10m5mol of the catalyst, 7 X low3 mol of cyclohexene and 3 ml of dry toluene was introduced into a mini-autoclave. The apparatus was sealed, purged with argon (20 atm), evacuated and filled with carbon monoxide (40 atm) and hydrogen (40 atm). The mixture was magnetically stirred and heated at 120 * 1 “C. After 15 h the autoclave was cooled and its contents were either distilled under reduced pressure or directly analyzed both by ‘H NMR and by GC on 15% OV-101 on Chromosorb W. When the silica-bound complexes 9 and 10 were used, the apparatus was opened under argon and the used catalysts washed repeatedly with toluene prior to their recycling. Isomerization of allylbenzene A solution of 2.19 X 10m3mol of allylbenzene in 4 ml of dry o-xylene was heated under argon at 120 + 1 “C. After thermal equilibration, 2.19 X 10m5 mol of the catalyst was added and the progress of the reaction monitored by GC on 20% Carbowax 20M on Chromosorb W. After prolonged heating (ca. 50 h) the equilibrium mixture consisted of 88.0% trans- and 5.2% cis-l-propenyl-benzene and 6.8% allylbenzene. Results and discussion Catalytic hydrogenation Under the conditions given in the Experimental section, cyclohexene, cyclooctene and ldecene were hydrogenated in the presence of catalysts 3 - 10. In all experiments we observed an induction period which extended, depending on the catalyst employed, from 8 min to 5.5 h. This period was followed by hydrogen absorption with rather complex kinetics [8]. The maximum rate of hydrogenation of the three olefins was shown to be in the order cyclohexene < cyclooctene < l-decene. For example, in the presence

320

of 10 (conditions as in the Experimental section) the relative rates were 1:1.1:1.7. In contrast to olefins, acetylenic compounds (e.g. phenylacetylene) were not reduced by hydrogen and the bimetallic complexes. Instead, stable substrate-catalyst adducts were formed (IR 1600 cm-i). Among the soluble catalysts 3 - 8, the silicon-containing compound 8 was found to have the highest activity. It gave higher maximum rates than did the lower homologue 7, in spite of the fact that the induction periods with the latter were always shorter. While 3 gave results similar to those of 7 in cyclohexene hydrogenation, complex 4, in which a bulky t-butyl group is attached to the bridging sulfur atom, was found to be inferior; 5 and 6 also proved to have lower activities than 7. We assume that the diminished activities of the complexes with aromatic rings are associated with possible ortho metallation by one of the rhodium atoms. This action would occupy two active coordination sites of the metal atom. It should be noted that under our standard conditions even the less active catalysts 4, 5 and 6 are by far better than the sulfur-free dirhodium complex 2. The latter requires 20 h for the hydrogenation of cyclohexene to an extent of 20% and has an induction period of over 6 h. By attachment of complexes 7 and 8 to a silica support, their activities in a single run decreased with respect to maximum rates and induction periods (see Table 1). However, in the long term 9 and 10 proved superior to 7 and 8. Complex 7 could be used only once, as it deteriorated completely by the end of the first catalytic run, and 8 could be recovered for recycling only if worked up under strict exclusion of oxygen. On the other hand the immobilized complexes were much less sensitive to air; they were practically leach-proof and could be used in further runs without loss of activity. In fact, the catalytic power of 10 increased in the second and third cycles in a manner similar to the polymer-supported Ru, Rh and Ir complexes studied previously [ 91. As the addition of the olefinic substrates to the dirhodium catalysts caused immediate changes in the IR and visible spectra, we concluded that the rather long induction periods are associated with slow oxidative addition of hydrogen to an olefin-catalyst intermediate. This assumption was supported by an experiment in which a mixture of 10 in toluene was treated with hydrogen at 690 mm and 50 “C for 24 h prior to the injection of the cyclohexene. In this experiment the formation of cyclohexane started immediately. Complete absence of induction periods were characteristic also of those processes in which recovered catalysts were used (see ,Table 1). This implies that once the active species is formed it does not dissociate back into the starting catalyst even if the hydrogen source is removed. In the homogeneous reactions in which 3 - 8 were used as catalysts, the color darkened gradually, and in some cases brown or black particles separated towards the end of the processes. These air-sensitive solids dissolved, however, in dimethyl formamide and in dimethyl sulfoxide and were found

of some soluble

0.334 0.159(4.7) 0.223(l)

0.175

1st run

0.349 O.lll(5.0) 0.381(l)

-

2nd run

dirhodium

220 550 450

-

2nd run

50 “C, 0.9 atm hydrogen.

230 1060 560

375

1st run

200 550 450

-

3rd run

(min)

conditionf

for full conversion

under comparable

Time required

complexes

3 ml of dry toluene,

0.365 O.lll(3.5) 0.413(l)

-

3rd run

and heterogenized

Maximum rate (mmol s-l 1-l); metal leaching in parentheses (ppm)

in the presence

conditions: 7 mmol of cyclohexene, 2.19 x 10m2 mmol of catalyst, content 0.438 meq g-l silica gel 60 (230 - 400 mesh). content 0.582 meq g-l silica gel 60 (230 - 400 mesh).

30 40 70

8 9b 10:

aReaction bRhodium ‘Rhodium

8

Induction period in 1st run (min)

Catalyst

7

of cyclohexene

Hydrogenation

TABLE 1

322

to be highly active hydrogenation catalysts. Their IR spectra, in which new bands at -2040 and -2110 cm-’ appear, suggest the formation of Rh(II1) car-bony1 hydrides. Hydroformylution

Under pressure of 40 atm H, and 40 atm of CO, both the homogeneous and the heterogenized dirhodium complexes were shown to act as good hydroformylation catalysts. For example, cyclohexene (0.01 mol) was converted at 120 “C in quantitative yield into cyclohexanecarboxaldehyde when reacted for 23 h in the presence of either 5 X 10m5 mol,of Rh,(CO)z(P[C(CH,),I~)~(~~~-C~)[C(-SC(CH~)~I (4) or of the silica-supported catalyst 9. The other soluble complexes were somewhat less effective: the yields with 3 and 5 were 43 and 93%, respectively. It should be noted that under these conditions the sulfur-free dirhodium complex 2 gives only 12% C6H&H0 along with 68% high boiling material. The immobilized catalyst could be recycled several times with hardly any loss in activity (maximum metal leaching in the first two runs was 3.5 ppm) as long as the silica support did not lose its mechanical properties. Under 80 atm pressure the coarse particles of Silica gel 40 (35 - 70 mesh) support had become a fine powder after 60 - 70 h from constant impact with the Tefloncoated magnetic stirring bar. Thus, during the third or fourth run, elimination of the dirhodium complex from the surface of the silica became significant, and usually in the fifth run only metal-free silica powder was left. The catalyst destruction could be reduced to some extent by using Silica gel 60 of smaller particle size (230 - 400 mesh) instead of Silica gel 40, or even eliminated completely by removal of the magnetic bar from the reaction mixture. However, without stirring low and h-reproducible results were obtained due to lack of homogeneity. It is thus not unexpected that the activity of the supported catalysts depends considerably on the mode of attachment of the soluble bimetallic compound onto the support. It was shown [6] that the catalysts 9 and 10 can be prepared either by direct interaction of 7 and 8 with silica according to eqn. (4) (method A) or by treatment of the presilylated silica derivatives 11 and 12 with 2 and tri-t-butylphosphine (eqns. (5) and (6)) (method B).

0 /OH SiOz

~g;

+

OC,

(ROMWWnS\Rh,

7, R = 2 8,n= 3

,W(CH3)313

Rh ’

R = C2H5 R=CH3 oc’

-3ROH+

‘Cl ‘P[C(CH3)313

0

3

Si02 fO:Si(CH&S< 0 0Ckh/‘W(CH3M >Cl 0 Rh oc’ ‘W(CH,),I 9, n = 2 10, n = 3

R = C2H5 R = CH3

(4) 3

323

/OH

SiOs qOH + (CHs)sSiS(CH,),Si(O&)s OH n=2.3 R = CHs, CsHs

-3ROH

A

0 SiOs /--O%i(CH&$Si(CHs)s ‘oy

(5)

11, n = 2 12, n = 3

2

1) PIC(CW& 2) 11 or 12

’

9orlO

(6)

Indeed, when we applied 9 and 10 which had been prepared by the different methods as hydroformylation catalysts, we observed substantially different activities. Some examples are summarized in Table 2. TABLE 2 Hydroformylation of cyclohexene catalysts 9 and 1Oa Catalyst

9

9 10 10

Method of preparation

A B A -B

by two

different

versions of the silica-supported

Rhodium content (mmol g-l Silica 60)

Yield of cyclohexanecarboxaldehyde after 20 h (%)

0.438 0.276 0.582 0.416

75 30 85 42

*Reaction conditions: 7 X 10d2 mol of cyclohexene, Silica gel 60). 40 atm H2,40 atm CO, 120 f 1 “C.

2.19

X

lob5 mol of catalyst

(on

Several features of the catalytic hydroformylation were shown to resemble those of the above hydrogenation process. As for the hydrogenation, the substrate effectiveness of most of the bimetallic catalysts for Hz and CO addition increases in the order cyclohexene > cyclooctene > 1-decene. A slight discrepancy was found when 10 was employed. Under comparable conditions cyclooctene was hydroformylated faster than 1-decene (see Table 3). The acyclic olefin gave a mixture of both the n- and the iso-aldehydes in practically equal amounts. As in the above process, acetylenic compounds -gave only substratecatalyst adducts but did not undergo catalytic hydroformylation. The catalytic activity of the homogeneous silicon-containing complex 7 with the short ethylene moiety was found to be exceptionally high. However, it lost most of its activity on recycling. On the other hand, the higher homologue 8 was less efficient than 7 in the first run, but retained its activity through many further cycles. Under exclusion of air, 8 proved to be even

324 TABLE

3

Hydroformylation Substrate

of some olefins

in the presence

of 9 and 10’ Yield in first run (%)

Products

Catalyst cyclohexene cyclooctene l-

decene

Catalyst

53 80 46 45

CeHlrCHC CsHrsCHO CH&CH2)&H0 CHa(CH,)$H(CHO)CHa

aReaction conditions: 7 x 10e3 mol substrate, 2.19 total Hz and CO pressure 80 atm, 120 f 1 “C!, 15 h. bhepared according to eqn. (4).

gb

X

10b

50 100 46 46

1O-5 mol catalyst

(on Silica gel 60),

TABLE 4 Hydroformylation soluble counterparts Catalyst

7 8 gc loc

of cyclohexene by silica-supported under comparable conditionsa

dirhodium

Yield of cyclohexanecarboxaldehyde parentheses (ppm)

complexes

and by their

(X); metal leaching

in

1st run

2nd run

3rd run

95 46 53(19) 50(11)

10 40 68(16) 62(16)

5 41b 33(4) 36(35)

aReaction conditions as in Table 3. Each run was conducted for 15 h. bYields of 40 - 42% were recorded also in the following 4 runs. ‘Prepared according to eqn. (4).

longer-lived than the silica-supported catalysts 9 and 10. The activity of the latter two increased in the second run but dropped thereafter, owing to gradual destruction of the silica particles and consequent enhanced metal leaching (Table 4). Catalytic double bond migration Although no double bond migration was observed when 1-decene was subjected to Hz or to a mixture of H2 and CO under the above conditions, in the absence of these gases compounds 9 and 10 proved to be excellent isomerization catalysts. For example, allylbenzene was converted into cisand trans-l-propenylbenzene (eqn. (3)) at rates which are 2 - 3 orders of magnitude higher than the maximum rates of isomerization by homogeneous and by polystyrene-bound RhC!I[P(C,Hs)s] 3 [lo]. In typical experiments in. which 2.19 X 10m3 mol of the starting olefin in 4 ml of o-xylene was reacted at 120 “C.with 2.19 X 10m5 mol of 4, 9 or 10,the corresponding initial rates were 0.341, 0.0165 and 0.0214 mmol s-r 1-l. The sulfur-free dirhodium

325

complex 2 seems to be cleaved into monorhodium compounds during an induction period of 90 min which precedes a very slow isomerization reaction (23% conversion after 7 h). In contrast to some Rh(1) and Rh(II1) complexes that are deactivated before completion of the process [ 111, the dirhodium catalysts retain full activity until an equilibrium mixture of 5.2% cis-, 88.0% truns-l-propenylbenzene and 6.8% allylbenzene has been obtained. Furthermore, the supported complexes 9 and 10 could be recycled numerous times without significant loss of activity or of metal content. Infrared analysis indicated that during all three catalytic processes, the original ,dirhodium complexes undergo structural changes. Attempts to fully characterize the modified catalysts were unsuccessful. However, we believe that they retain the bridged bimetallic structures. Support for this assumption was found in their mass spectra, in which fragment ions with two rhodium atoms could be detected (e.g., in the EI mass spectrum of used 3 ions appear-d of m/z 475, 477 [Rh&lS(PC$H&], 329, 331 [Rh,CIS(CO),], 281 [Rhz(SC3H,)], 238 [Rh,S]). Moreover, destruction of the chloroand thiolato-bridges under hydrogenation and hydroformylation conditions would be expected to be associated with the formation of HCl, H2S and/or mercaptans. We found, however, that none of these molecules was formed in our catalyses. Additional support for the survival of the bridged dirhodium structure at elevated temperature was found from decarbonylation experiments of l-naphthaldehyde in xylene at 120 “C. Both 5 and the silica-bound complexes 9 and 10 led to the formation of exactly two mol of naphthalene per mol of organometallic compound. During the reaction conducted with 9 and 10 no CO evolution was observed. The carbon monoxide was absorbed by the rhodium complexes, whose IR metal carbonyl bands were shifted to 1965 - 1970 and 2030 cm- ‘. These absorptions of the transformed complexes closely resemble those of tetracarbonyl dirhodium phosphine compounds having two bridging thiolato ligands [ 121, and their fragmentation patterns under electron impact indicated the presence of two linked metal atoms. Acknowledgements We are grateful to the Deutsche Forschungsgemeinschaft (DFG) and the exchange program between the Technical University of Berlin and the Hebrew University of Jerusalem for financial support of this study. References 1 R. Poilblanc, Znorg. Chim. Acta, 62 (1982) 75. 2 (a) P. Kalck, R. Poilblanc, A. Gaset, A. Rovera and R. P. Martin, Tetrahedron Lett., 21 (1980) 459; (b) P. Kalck, R. Poilblanc, P. R. Martin, A. Rovera and A. Gaset, J. Organometall. Chem., 195 (1980) C9.

326 3 See e.g., R. P. Hughes, in G. Wilkinson, F. G. A. Stone and E. W. Abel (eds.), Comprehensive Organometallic Chemistry, Vol. 5, Pergamon Press, Oxford, 1982, pp. 277 - 540 and references therein. 4 E.g., A. Thorez, A. Maisonnat and R. Poilblanc, J. Chem. Sot., Chem. Commun., (1977) 518. 5 H. Schumann, G. Cielusek and J. Pickardt, Angew. Chem., 92 (1980) 60. 6 H. Schumann, G. Cielusek, S. Jurgis, E. Hahn, J. Pickardt, J. Blum, Y. Sasson and A. Zoran, Chem. Ber., 117 (1984) 2825. 7 H. Schumann, S. Jurgis, E. Hahn, J. Pickardt, J. Blum and M. Eisen, Chem. Ber., 118 (1985), in press. 8 Cf. e.g., J. Halpern, Relations between Homogeneous and Heterogeneous Catalysis, CNRS, Paris, 1978, pp. 27 - 47. 9 J. Azran, 0. Buchman, M. Orchin and J. Blum, J. Org. Chem., 49 (1984) 1327 and references therein. 10 A. Zoran, Y. Sasson and J. Blum, J. Org. Chem., 46 (1981) 255. 11 J. Blum and Y. Pickholtz, Zsr. J. Chem., 7 (1969) 723. 12 E.g., P. Kalck and R. Poilblanc, Znorg. Chem., 14 (1975) 2779.