Organometallics 1995, 14, 1510-1513
1510
Selectivity for Hydrogenation or Hydroformylation of Olefins by Hydridopentacarbonylmanganese(1) in Supercritical Carbon Dioxide Philip G. Jessop, Takao Ikariya, and Ryoji Noyori*tt ERATO Molecular Catalysis Project, Research Development Corporation of Japan, 1247 Yachigusa, Yakusa-cho, Toyota 470-03, Japan Received July 19, 1994@ Summary: The stoichiometric and catalytic hydrogenation and hydroformylation of activated olefins by MnH(C0)s are generally believed to proceed by a free radical caged pair mechanism, the selectivity of which is affected by solvent. However, the stoichiometric reaction with a test olefin in supercritical C02 (scCOZ), with its low viscosity, gives a selectivity for hydrogenation almost identical to that found i n alkanes or without solvent. Aside from the possibility of coincidentally equal cage strengths, the most likely explanation is that the aldehydes are primarily formed by nonradical pathways which are independent of solvent viscosity. Introduction
Scheme 1. Free Radical Mechanism Proposed for Hydrogenation and Hydroformylation of Olefins by MnH(CO)a and Related ComplexesaJ1.24
+
M-H
(M.
+
R2C=CR2-----.
-.
cage formation
.**
11
CR2-CHR21
.CR2-CHR2
I
MH -M.
',, pathway 1 I
y%apse
,SCgyM.
', nonradical
- - -- - - - - --- -- + M*
nongeminate combination
j
t
M-CR2-CH R2
1
CO insertion
Recent reports of homogeneous catalytic hydrogenation1s2and hydroformylation3,*reactions in supercritical CHR2-CHR2 M-CO-CR2-CHR2 carbon dioxide (scCO2) are examples of what should hydrogenation become a very important new field: homogeneous catalysis in supercritical fluids (SCF). As a medium for chemical reactions, scCOz has many practical advanHCO-C RTCH R2 tages such as ease of separation of products, and it lacks hydroformylation many of the undesirable properties of liquid solvents a Alternative nongeminate and nonradical routes are indisuch as flammability, solvent residues, and disposal cated by dashed arrows. costs. Among the chemical advantages5s6are the very high concentrations of dissolved gases possible in Comparisons of the cage effects in scCOz versus liquid S C F . ~ I Because ~-~ of the latter advantage, homogeneous solvents are rare in the literature. DeSimone et al.15J6 hydrogenation and hydroformylation in such fluids are reported that the initiation efficiency f of AIBN for likely to be particularly important in the future. Howradical polymerization in scCO2 was 1.5 times higher ever, altered rates or selectivities compared to those in than in benzene, suggesting a weaker cage effect. The liquid solvents may be observed because of the unusual distribution of products from the photolysis of unsymphysical properties of scC02. For example, hydrogenametrical dibenzyl ketones in liquid solvents17 and in tion and hydroformylation reactions of some metal scC0zls shows no evidence of cage effects, while the carbonyls are believed to have radical pair mechanisms results in inclusion complexes show significant cage (Scheme 1)and cage effect-controlled ~ e l e c t i v i t i e s . l ~ - ~ ~ effects.lg Otto et a1.20p21found that the quantum yield scCO2, with its very low viscosity, may have weaker cage of the photolysis of IZ in near-critical liquid C 0 2 at 346 effects than liquid solvents.6 atm is comparable to the value in liquid heptane at lower pressures, suggesting comparable cage effects t Permanent address: Department of Chemistry, Nagoya University, Chikusa, Nagoya 464-01, Japan. despite the lower viscosity of COZ. Reactions which are
Abstract published in Advance ACS Abstracts, February 1, 1995. (1) Jessop, P. G.; Ikariya, T.; Noyori, R. Nature 1994, 368, 231233. (2) Jessop, P. G.; Hsiao, Y.; Ikariya, T.; Noyori, R. J. A m . Chem. SOC.1994.116. 8851-8852. (3) Rathke, J. W.; Klingler, R. J.; Krause, T. R. Organometallics 1991,10, 1350-1355. (4) Rathke, J. W.; Klingler, R. J.; Krause, T. R. Organometallics 1992,11, 585-588. (5) Boock, L.; Wu, B.; LaMarca, C.; Klein, M.;Paspek, S.CHEMTECH 1992,22, 719-723. (6) Subramaniam,B.; McHugh, M. A. Ind. Eng. Chem. Process Des. Deu. 1986,25, 1-12. (7)Howdle, S. M.; Poliakoff, M. J. Chem. SOC.,Chem. Commun. 1989, 1099-1101. (8) Howdle, S. M.; Healy, M. A.; Poliakoff, M. J. A m . Chem. SOC. 1990,112,4804-4813. (9) Tsang, C. Y.; Streett, W. B. Chem. Eng. Sci. 1981,36,993-1000. (10) Nalesnik, T. E.; Orchin, M. Organometallics 1982,1,222-223. (11)Sweany, R. L.; Halpern, J. J.A m . Chem. SOC.1977,99,83358337. @
(12) Wassink, B.; Thomas, M. J.;Wright, S. C.; Gillis, D. J.;Baird, M. C. J. A m . Chem. SOC.1987,109, 1995-2002. (13) Feder, H. M.; Halpern, J. J. A m . Chem. SOC.1975,97, 71867188. (14) Halpern, J. Pure Appl. Chem. 1979,51, 2171-2182. (15) DeSimone, J. M.; Guan, 2.; Elsbernd, C. S.Science 1992,257, 945-947. (16) Guan, 2.; Combes, J. R.; Menceloglu, Y. 2.; DeSimone, J. M. Macromolecules 1995,26, 2663-2669. (17) Robbins, W. K.; Eastman, R. H. J . A m . Chem. SOC.1970, 92, 6077-6079. (18) O'Shea, K. E.; Combes, J. R.; Fox, M. A.; Johnston, K. P. Photochem. Photobiol. 1991, 54, 571-576. (19) Rao, B. N.; Turro, N. J.; Ramamurthy, V. J. Org. Chem. 1986, 51, 460-464. (20) Otto, B.; Schroeder, J.; Troe, J. J. Chem. Phys. 1984,81,202213. (21) Luther, K.; Schroeder,J.;Roe, J.; Unterberg,U.J.Phys. Chem. 1980,84,3072-3075.
0276-733319512314-1510$09.00/00 1995 American Chemical Society
Organometallics, Vol. 14, No. 3, 1995 1511
Notes
Scheme 2. Test Reaction of MnH(C0)b with 3,3-Dimethyl-1,2-diphenylcyclopropene
shields and pressure relief mechanisms, to minimize the risk of personal injury. The cooling of the reactor may cause weakening of the vessel walls. The use of liquid nitrogen for this purpose should be avoided. p h ~ p + h MnH(C0)S Reactions in s c C 0 ~ .The solubility of the olefin (3 x low4 sccoz mol) in scC02 was confirmed by visual inspection using a 5060 "C mL reactor equipped with windows; at room temperature and 20:6 200 atm normal pressure the liquid was clearly visible, while at 60 "C and 200 atm of COz none could be seen. An equivalent test Ph + mol of MnH(C0)s had the same result. The with 2 x qualitative solubility of MnH(C0)s in scCO2 has already been reported.28 Reactions were performed in 50-mL unlined stainless steel reactors manufactured by JASCO. Before each reaction, the hydrogenation hydroformylation reactor was dried at 70 "C and cooled to room temperature product product under strong vacuum. The 3,3-dimethyl-l,2-diphenylcyclopropene (66mg, 3 x mol) was placed in the reactor, which much slower than the rate of cage escape are indepenwas then reevacuated and filled with argon to a slight positive dent of cage effects and solvent viscosity.22 It is mol) pressure. The MnH(C0)s complex (0.11 mL, 1 x therefore not clear whether liquid solvents and SCF was injected by syringe through a threaded opening in the top of the reactor, which was plugged at all other times. C02 was have differing cage effects or whether such differences introduced by pumping from a cooled (-5 "C) reservoir by an could result in differing selectivities. HPLC pump, after which the reactor was heated t o 60 "C for The existence of the radical pair mechanism (Scheme 4 h. The total pressure at the reaction temperature was ca. 1)has been demonstrated for the stoichiometric hydro200 atm. A magnetic stir bar was used to encourage solubigenation of olefins with MnH(C0)5 by observation of lization and mixing; variation of the stir rate had no effect on CIDNP emissions for the hydrogenated p r o d u ~ t s . ~ ~ J ~the p ~ selectivity ~ or yield. An alternative method involved CIDNP absorptions for the hydrometalated intermediheating the olefin in the reactor to 60 "C under 1atm of argon a t e ~suggest, ~ ~ though , ~ ~not~ convincingly, ~ ~ a radical before adding the Mn complex and then C02 pressure. The pair mechanism for hydroformylation by the same results obtained by the two methods were identical. ARer the expiration of the desired reaction time, the reactor reagent. Such evidence does not establish the predomiwas cooled with acetoneldry ice to constant pressure. The nance of these mechanisms over nonradical pathways, remaining pressure was vented and the reactor thawed. but this predominance has on occasion been assumed. CDCl3 solutions of the products were passed through a short The stoichiometric reaction of MnH(C0)5 with 3,3column of silica gel (CDCls as eluent) to remove insoluble Mn dimethyl-l,2-diphenylcyclopropeneat 60 "C (Scheme 2) products, before being analyzed by lH NMR spectroscopy. The was chosen as a test of the selectivity in scCO2, because spectral data of the products have been reported.23 Complete the selectivity of the reaction in liquid solvents is almost conversion of the olefin was observed. From the integrals, evenly split between hydrogenation and hydroformylaproduct ratios were calculated. The experiment in SCCOZ was tion.23,24Previous s t ~ d i e s ~ 3 -of~ 5 this reaction in soluperformed six times in order to measure the scatter, and the tion showed that the selectivity for hydrogenation was highest and lowest values were rejected, the mean percent hydrogenation not being significantlyaltered by such rejection. lower in micelle-containing solutions than in hexane or The standard deviation of the percent hydrogenation was 3.4. pentane, a difference which was attributed to a stronger The experiment under 5 atm of COz without solvent was cage effect in the former solutions. Importantly, these performed once. earlier studies also established that the selectivity is In order to determine the dependence of selectivity on the unchanged whether the reaction is performed under CO MnH(C0)S concentration, a series of experiments were peror argon. formed at a lower concentration of olefin and several concentrations of the complex. However, problems of NMR signal Experimental Section broadening were encountered at high concentrations of Mn, so a slightly different method of analysis was adopted for this Materials. The preparation of 3,3-dimethyl-l,$-diphenyl- series. The Mn products were removed or solubilized by cyclopropene was performed by the published after stirring the CDCl3 products with dilute aqueous HC1, separatwhich the crude product was heated to 45 "C for 1.5 h under ing, drying with &co3,and then analyzing as before by NMR strong vacuum to remove unreacted 1-phenylpropyne. The spectroscopy. Tests showed that this method does not alter desired product was the first fraction obtained by column the observed selectivity. The mass balance was confirmed chromatography (silica gel, hexane eluent) of the remaining (290% recovery of organic products) by adding 1,2-dichloroliquid. MnH(C0)s was obtained by the literature method.27 ethane as an internal standard immediately after the reaction. Safety Warning. Operators of high-pressure equipment such as that required for the following experiments should take Results and Discussion proper precautions, including but not limited to the use of blast
-
y;3)fH0
(22)Brennecke, J. F. In Superitical Fluid Engineering Science: Fundamentals and Applications; Kiran, E., Brennecke, J. F., Eds.;ACS Symposium Series 514;American Chemical Society: Washington, DC, 1993;Chapter 16,pp 201-219. (23)Nalesnik, T.E.;Orchin, M. J. Organomet. Chem. 1981,222,
--
C5-CR
(24)Nalesnik, T.E.;Freudenberger, J. H.; Orchin, M. J.Organomet. Chem. 1982,236,95-100. (25)Matsui, Y.;Orchin, M. J . Organomet. Chem. 1983,244,369373. (26)Friedrich, L. E.;Fiato, R. A. Synthesis 1973,611-612. 1977,99,6243(27)McNeill, E.A,; Scholer, F. R. J.Am. Chem. SOC. 6249.
At 60 "C, the selectivity for hydrogenation in SCCOZ (200 atm) or neat under 5 atm of C 0 2 was the same as or slightly higher than that reported for hexane and pentane, and considerably higher than that found in micellar solution (Table 1). As mentioned in the Introduction, metal carbonyl hydrides such as CoH(C0)4and MnH(COI5 are generally believed to catalytically and stoichiometrically hydro(28)Clarke, M. J.; Howdle, S. M.; Jobling, M.; Poliakoff, M. Inorg. Chem. 1993,32,5643-5644.
1512 Organometallics, Vol. 14, No. 3, 1995
Notes
Table 1. Reactions of MnH(C0)s with 3,3-Dimethyl-l,2-diphenylcyclopropene selectivity, % (cis:trans)
a
solvent
gas (pressure, am)
[Mn]/[olefin],mh4
temp, "C
time, h
alkane
aldehyde
hexanez3 pentanez4 micellez5 none sccoz SCCOZ
co Ar or CO co coz (5) coz ( 2 0 )
3200/1100 89/87 812 2016 2016 1816
55 60 50 60 60 35
5 2-4 15 4 3.5 16
66 (7:l) 63 (7:1) 8 (only cis) 66 (51) 66 (6:1) 61 (4:l)
34 (7:l) 37 (7:1) 92 (6:1) 34 (mostly cis)" 34 (mostly cis)" 39 ( E l )
COz (236)
The trans hydroformylation product was observed in very small amounts, so that accurate cis:trans ratios could not be determined.
loo
Scheme 3. Mechanism Consistent with the Observations
F
M-H
+
R2C-CR2 nonradical pathway
M-CR2-CHR2 2 0 1 0
+ 1 [M.
0
10
20
30
40
50
60
70
80
~M~H(CO)SI, mM
Figure 1. Dependence of the selectivity for hydrogenation on the concentration of MnH(C0)5in 200-230 atm of scCO2 at 60 "C, 1.6 mM olefin. The open circle shows the result at 6 mM olefin. genate and hydroformylate activated olefins by radical mechanisms.1°-14 The product-determining steps are escape from the radical cage, leading to hydrogenation, competing with collapse of the radical cage, leading to hydroformylation (Scheme 1). Additional hydroformylation could occur if the escaped alkyl radical encounters a metal complex radical before it reacts with a metal hydride; this process is nongeminate radical combination. If the cage escape and cage collapse processes have comparable rates, then a weaker cage effect in scC02 should increase the selectivity for hydrogenation. However, we found that reactions with scCO2 as a solvent had essentially the same selectivity as that with no solvent under 5 atm of COz and those23J4with hexane or pentane (Table 1). Thus, the selectivity for hydrogenation in scC02 is comparable to that in hydrocarbons, despite the greater diffusivity of the supercritical phase. It is possible that the cage effects are coincidentally identical in all of these media. Perhaps it is more likely that the observed selectivity of 66% hydrogenation represents that to be found in the absence of a significant cage effect. The aldehydes must then be formed mostly by nongeminate combination or nonradica129pathways (Scheme 1). The possibility of nongeminate combination was eliminated by determining the dependence of the selectivity on the concentration of MnH(C0)5 (Figure 1). If nongeminate combination had been significant, one would have observed a saturationtype curve,3owith concentration-dependent selectivity for hydrogenation at low concentrations and constant selectivity at high concentrations of the complex. The observed independence of the selectivity on concentra(29) "reichel, P. M. In Comprehensive Organometallic Chemistry; Wilkinson, G., Stone, F. G. A., Abel, E. W., Eds.; Pergamon Press: Oxford, 1982; Vol. 4, Chapter 29, pp 1-159. (30) (a) Jacobsen, E. N.; Bergman, R. G. J . Am. Chem. Soc. 1985, 107,2023-2032. (b) Hammond, G. S.; Sen, J. N.; Boozer, C. E. J . A m . Chem. SOC.1955,77, 3244-3248.
CR2-CHR21
cagel
M-CO-CRTCHR~
escape
-Mm
HCO-CR2-CHR2 hydroformylation CHR2-CHR2 hydrogenation
tion allows us to reject the nongeminate pathway.31We are thus left with two pathways to aldehyde formation: the cage collapse and the nonradical routes. It is possible that both pathways operate, the cage collapse pathway being favored in viscous media (Scheme 1).The selectivity could also be determined by a competition between homolysis and CO insertion reactions of the metal alkyl complex, which would be formed by a nonradical pathway (Scheme 3).31 If the geminate recombination of M and 'CR2CHR2 occurs to some extent, then this would account for both the CIDNP results and the dependence of selectivity on the medium. It is only possible to roughly estimate whether a cage effect should have been expected. Although calculations have not been made for scC02, the lifetime of the solvent cage has been estimated to be on the order of picoseconds for other SCF.32 This is too short to be an effective cage for all but the very fastest reactions. The rate of the geminate recombination reaction in the present system is not known, but it should be at least as fast33 as the rates of the self-dimerizationsof Mn(C0)5or alkyl radicals (second-order rate constants -1 x lo9 M-l s-1).33934 Thus, the rate of reaction of the alkyl radical with the Mn(C0)5 radical must be close to the diffusion limit in liquid solvents but would have to be a few orders of magnitude faster in order to be affected by the cage of a SCF. (31) Examination of the dependence of selectivity on MnH(C0)E concentration was suggested by a reviewer. We also appreciate the comments suggesting Scheme 3. (32)Petsche, I. B.; Debenedetti, P. G. J . Chem. Phys. 1989,91, 7075-7084. (33) Ingold, K. U. In Free Radicals; Kochi, J. K., Ed.; John Wiley and Sons: New York, 1973; Vol. 1, Chapter 2, pp 37-112. (34) Baird, M. C. Chem. Rev. 1988,88,1217-1227.
Organometallics, Vol. 14, No. 3, 1995 1513
Notes
The effect of clustering near the critical point of a SCF has been greatly d i s c ~ s s e d , but ~ ~ -the ~ ~effect is reportedly strongest near the critical point. The conditions of the reactions in Table 1 are too far from the critical point to exhibit significant clustering effects. The selectivity for cis rather than trans hydrogenation products is similar in SCCOZ,neat, and hydrocarbon solutions. However, the trans hydroformylation product was barely detectable in our experiments. In all media, the cis products are preferred for both hydrogenation and hydroformylation. The reason for the preference for cis addition is not known, but steric approach ~ o n t r o Pseems ~ , ~ ~likely. This idea assumes that steric effects acting in an early transition state would direct (35) Ikushima, Y.;Saito, N.;Arai, M. J.Phys. Chem. 1992,96,22932297. (36)Ellington, J. B.;Brennecke, J. F. J. Chem. Soc., Chem. Commun. 1993, 1094-1095. (37) Combes, J. R.;Johnston, K. P.; O'Shea, K. E.; Fox, M. A. In Supercritical Fluid Technology; Bright, F. V., McNally, M. E. P., Eds.; ACS Symposium Series 488; American Chemical Society: Washington, DC, 1992; Chapter 3, pp 31-47. (38) Dauben, W.G.;Fonken, G. J.; Noyce, D. S. J . Am. Chem. SOC. 1956, 78,2579-2582.
the entering reagent to the less hindered trans position. An experiment at 35 "C in scC02 was considerably slower, giving 15%conversion afker 16 h. The selectivity for hydrogenation was only slightly lower. In conclusion, identical selectivities for the stoichiometric reaction of MnH(C0)5with the test olefin have been found in three media, hexane, neat olefin, and scCO2, suggesting that cage effects are either coincidentally identical or that the aldehydes are primarily formed by nonradical pathways (Scheme 3). The cage collapse and nonradical pathways could be in competition, with viscous media favoring the former. Nongeminate radical combination pathways are not significant under the conditions tested. By analogy to the present results, cage effects are not expected to be a factor in catalytic hydrogenations or hydroformylations by carbonyl catalysts in scCO2. OM9405731 (39) Skell, P.S.;Allen, R. G. J . Am. Chem. SOC. 1968,80, 59976000.