Hydroformylation of olefins under mild conditions

Hydroformylation of olefins under mild conditions

Journal of Molecular Catalysis, 24 (1984) 309 - 321 309 HY~ROFORMYLATION OF OLEFINS UNDER MILD CONDITIONS PARTI:THECo,_,Rh,(CO),,+xL(n=0,2,4;x=O-9)S...

996KB Sizes 8 Downloads 61 Views

Journal of Molecular Catalysis, 24 (1984) 309 - 321




di Chimica In~rganica FUMAGALLI

e Metaltorgan~ca, Via Venezian

21, 20133



Centro di Studio sulla Sintesi e la Struttum dei Composti dei Metalli di Transizione nei Bassi Stati di Ossidazione, Via Venezian 21, 20133 Milan (Italy) (Received July 5,1983;accepted

October 6,1983)

Summary The hydroformylation of cyclohexene, 1-pentene and styrene under mild conditions (25 - 50 “C, 1 atm equimolar mixture of CO and H,) has been investigated using as catalyst precursor either the Co4_ .Rh,(CO)i:! f x L (n = 0, 2, 4; x = 0 - 9) system or preformed Rh,(CO),,_,L, (x = 1 -4) substitu~d clusters, where L is a t~substitu~d phosphine or phosphite. The activity of these systems increases as a function of X, and reaches a maximum for a L/Co4_,Rh,(C0)i2 (n = 2, 4) molar ratio of cc. 5 - 6. A further increase in this ratio corresponds to a smooth decrease in the activity. This ratio has apparently a negligible effect on the regioselectivity in the hydroformylation of both l-pentene and styrene. In contrast, both the activity and the regioselectivity are significantly affected by the nature of the liiand employed as cocatalyst . When working with Rhyme as well as Rh,(CO)i6, and t~substitu~d phosphites as ligands, infrared spectroscopy and 31P NMR invariably show the presence of Rh4(CO)gL3 as the most substituted rhodium carbonyl species present in solution, and there is no evidence of fragmentation of the tetranuclear cluster during the catalytic process. In contrast, when using phosphine ligands such as PPh,, evidence of fragmentation to Rh~(CO)~(PPh3)~ or to Rh~(CO)~(PPh~)~ species has been obtained at the higher PPh3/Rh~(CO) I 2 molar ratios. Degradation of the ligand employed as cocatalyst, particularly the arylsubstituted phosphines, is observed, and this is probably at the origin of the loss of catalytic activity of some of these systems with time.

*Author to whom correspondence 0304-5102/84/$3.00

should be addressed. @ Elsevier Sequoia/Printed

in The Netherlands


Introduction Hydroformylation of olefins under mild conditions with rhodium complexes such as HRh(CO)(PPh,), [l] or HRh(CO)(DMBA), (DMBA = N, Ndimethylbenzylamine) [ 21 as catalyst precursor is a well-established method. Although Rh,( CO) I2 and Rh,( CO) 16, sometimes modified with ligands, have been extensively investigated as catalyst precursors for hydroformylation under pressure [ 31, to our knowledge a systematic investigation concerning their application under mild conditions is still lacking. In 1972 the stoichiometric hydroformylation of propene with Rh,(C0)i2 under hydrogen atmosphere (25 “C, 1 atm), according to eqn. (l), has been reported [ 41. 3 Rh4(C0)12 + 4 CHs-CH=CH2

+ 4 H, --

+ 2 CHs-CH2-CH2SH0

2 Rh6(CO),,




The hydroformylation was shown to become catalytic in the presence of aryl- and/or alkylphosphines as cocatalysts and under an equimolar CO and Hz atmosphere (25 ‘C, 1 atm) [4]. This observation apparently was not in keeping with an earlier report, in which preformed substituted tetranuclear rhodium clusters, such as Rh4(C0)s(PPh& and Rh4(C0)s(ETP0)4, (ETPO = 4-ethyl-2,6,7-trioxa-l-phosphabicyclo-[ 2.2.2]-octane), were found to be inactive in the hydroformylation of 1-hexene under mild conditions, whereas addition of excess PPh, to Rh4(C0)i2 was shown to generate Rh2(C0)4(PPh,), as an active precursor [ 51. In the last few years more information about substituted M4( CO)1Z (M = Co, Rh and Ir) clusters became available [6 - 81. Furthermore, the synthesis and, in some cases, the structural characterization of derivatives such as Rh,(C0),4(diene) [9], [HRh,(CO)lS][lo] and [Rh,(CO),,(COR)][ll] has shown that a cluster can at least bear the functional groups required for the hydroformylation of to reinvestigate both the substituted olefins. We decided, therefore, Rh,(CO)l,_,L, and Rh4(C0)12-.L, clusters, as well as the Co,-,,Rh,(C0)12 + x L (n = 0, 2, 4; x = 0 - 9) system with regard to the hydroformylation of olefins under mild conditions. We report here our results concerning the hydroformylation of l-pentene, styrene and cyclohexene.

Results and discussion General observations

on the ligand-substituted


cluster systems

Both Rh,(C0)16 and Rh4(C0)i2 are readily degraded by addition of excess phosphine ligands, such as PPhs, to sparingly soluble dimeric clusters [ 5, 131. In contrast, phosphite ligands such as P(OPh),, P(OMe), and ETPO have been reported to cause degradation only of the hexanuclear Rh,(CO)i,_,L, clusters, e.g. eqn. (2), in addition to concomitant progressive carbonyl substitution reactions [ 6 - 81.


2 Rh6(C0)i6

+ 12 P(OPh)3 -

3 Rh,(CO),[P(OPh),],

+ 8 CO


Addition of triphenylphosphite to Rh4(C0)i2 gives rise only to stepwise carbonyl substitution up to Rh,( CO)s[ P( OPh)Js, which is stable also in the presence of excess ligand. Under our working conditions we found that the substituted Rh,(CO),,_.[P(OPh),], (X = 1 - 4) cluster only undergoes ligand substitution equilibria (eqn. (3)), as shown by infrared spectroscopy and 31P NMR,


+ CO 3

Rh,(C0)12-x+ ,D’(OPhM-,

+ P(OPh), (3)

while preformed substituted hexanuclear clusters of Rh,( CO) i6 _,[ P(OPh),], are involved in degradation reactions, fied by eqn. (4): 2 Rh,(CO)i2[P(OPh),],

+ 4 CO -

2 Rh,(CO),[P(OPh),], + Rh,(CO),,]P(OPh),l2

For the above clusters are catalyst mainly focused on well as on the Rh,( good approximation


the type as exempli-

+ (4)

reasons, although the substituted hexanuclear rhodium precursors comparable to the tetranuclear ones, we the behaviour of preformed Rh4(CO)i2_ XL, clusters, as CO)i2 + x L mixture, which may be considered with a coincident with the former for x = 1 -4.

of cyclohexene

As previously reported in the case of propene, reaction (1) becomes catalytic in the presence of carbon monoxide, although the latter exerts a strong inhibiting effect [ 131. Accordingly, we find that Rh4(C0)i2 in toluene catalyzes the hydroformylation of cyclohexene to cyclohexylaldehyde under an equimolar carbon monoxide and hydrogen atmosphere (1 atm) under mild conditions (50 “C). Under these experimental conditions the stoichiometric limit set by eqn. (1) (1.33 mol aldehyde per mol Rh4(CO)r2) is reached in ca. 3 - 4 hours. The hydroformylation rate drops sharply with time, which is due to the precipitation of the sparingly soluble Rh6(C0)i6. After ca. 60 h the system is practically inactive and has produced on the average ca. 6 mol aldehyde per mol starting Rh4(C0)i2. Addition at this stage of ligands such as trisubstituted phosphines or phosphites causes the formation of the more soluble substituted hexanuclear Rh,(C0)i6 _,L, and tetranuclear Rh4( CO) i2 _ XLX clusters, and regenerates an active system. Similar behaviour was previously observed also in the case of propene [ 41. As shown in Fig. 1, the hydroformylation of cyclohexene (50 “C, 1 atm) using Rh,(CO)s[P(OPh),],, either preformed or generated in situ by adding four moles of P(OPh), to Rh,(C0)i2, shows a similar trend in the two cases; this finding is in keeping with the presence essentially of Rh,(CO),[P(OPh)s] 3 throughout the catalytic process in both cases, as detected by IR and 3’P NMR spectroscopy.











Fig. 1. Hydroformylation of cyclohexene (4 ml) at 50 “C and 1 atm of CO + Hz (1 :l) in toluene (20 ml). l, Rhd( CO)12 + 4 P( OPh)a ; 0, RhJ( CO)s[P( OPh)3 14.

The system Rh4(C0)i2 + x P(OPh)3 allowed wider exploration of the effect of the P(OPh)JRh4(C0)i2 molar ratio on the activity. The results of this investigation are collected in Table 1. Unfortunately this system is poorly amenable for kinetic studies, owing to a series of complications. Thus, when working with a low P(OPh),/Rh,( CO) i2 ratio (1 - 2) the system deactivates with time, probably due to the occurrence of equilibria such as (3) followed by transformation of the resulting Rh4(C0)i2 into Rh,(C0)i6. For instance, in run 2 of Table 1, the initial rate of hydroformylation of cu. 0.5 h-’ (expressed in mol aldehyde per mol Rh4(C0)i2 per h) drops to cu. 0.17 h-’ in 18 h, although the olefin concentration is decreased by only cu. 1.3%. In contrast, when working with P(OPh),/Rh4(C0)i2 molar ratios greater than 4, increasing induction periods are observed; afterwards the hydroformylation rate increases sharply, and then decreases again as cyclohexene is consumed. The induction period is virtually eliminated for ratios of 4 and 5 if these systems are initially left to equilibrate under an atmosphere of carbon monoxide and hydrogen at 25 “C for few hours in the absence of olefm. Systems having P(OPh)JRh4(C0)i2 ratios greater than 7 are little affected by this pretreatment, and induction periods even of 4 - 8 h are often observed. Therefore, the activities of these systems are compared in Table 1 as percent of hydroformylation after 48 h. As shown in Table 1, the activity of the system increases on increasing the P(OPh)JRh4(C0)i2 ratio from 1 to 6. A further increase in this ratio apparently corresponds to a smooth decrease in activity, and for ratios greater than 20 the system is substantially inactive. In all cases, the catalytic solutions are redcoloured, and monitoring by IR throughout the catalytic runs shows that the substituted tetranuclear



Effect of the P(OPh)3/Rh4(C0)12 molar ratio (1.65 M) in toluene (20 cm3) with the system and Hz (1 :l)) -__ P( OPh)P


on the hydroformylation of cyclohexene Rh4(C0)12 + x P( OPh)s (50 “C, 1 atm CO

% Hydroformylation

after 48 h

RhdCO)lz 1C 2c 3 4 5 6 7 a 9 .O

20.3 7.9 6.8 6.9 7.4 6.4 8.0 7.6 7.0 7.1

aIn mm01 1-l. bMolar ratio. CPrecipitation of Rhs(CO)ls

1 2 3 4 5 6 7 8 20

5.7 2.4 6.1 23.8 67.2 73.9 90.4 78.2 80.2 0

causes deactivation.

Rh,(CO)n --x F’(OPhM, (x = 1 - 3) clusters are the only carbonyl species detectable in solution when the P(OPh),/Rh4(C0)12 molar ratio is in the range 1 - 4; with ratios greater than 4, the apparently pure trisubstituted Rh,(CO),[P(OPh),] 3 derivative is present. The same compound also is the major component present in solution when using as catalyst precursor a 1:4 molar mixture of Rh,(C0)&12 and P(OPh), in the presence of stoichiometric amounts of triethylamine as base. Working at the same rhodium concentration of run 9 in Table 1, a very close conversion was observed. With the tetrasubstituted Rh,(CO)s[P(OPh),], cluster or the Rh4(C0),2 + 4 P(OPh)3 system, which exhibits negligible complications arising from deactivation processes or induction periods, a linear dependence of the hydroformylation rate on both cluster concentration (lop2 - lop3 M) and cyclohexene concentration (0.8 - 4 M) is found. A similar behaviour also has been observed in the hydroformylation of cyclohexene under pressure with Rh4( CO) 12 as catalyst precursor [ 14 1. Co,(CO) 12_ x [P(OPh),], Although CO,(CO)~~ and the substituted (X = 1 - 3) clusters are essentially inactive under our mild conditions, the bimetallic CozRh2(C0),2 cluster shows an activity greater than that of During the catalytic process, however, the Co2Rh2(CO)12 rearRh,(CO)lz. ranges according to eqn. (5). 6 CozRhz( CO) 12toluene’ 5o “’ 3 COALS ’

+ 2 Rh6(C0)16 + 4 CO


Probably owing to this rearrangement, the hydroformylation rate decreases in a few hours from an initial value of cu. 2 to cc. 0.5 h-‘; afterwards, the


system continues to produce cyclohexylaldehyde at this rate for several hours (>200). It is worth mentioning that when hydroformylating ethylene and propene (25 “C, 1 atm), the deactivation apparently corresponds to the appearance in solution of Co(CO),(COR) species, as monitored by IR [15]. It appears likely that the slight residual activity in the case of cyclohexene could be ascribed to the presence of traces of soluble Co-Rh mixed-metal derivatives such as the Coq-nRh,(CO)lz clusters [ 161, or the recently reported binuclear CoRh( CO), derivative [ 171. In keeping with this interpretation, addition of either Coz(CO)s or Coq(CO)i2 to an inactive suspension of Rh,( CO)i6 induces a slow hydroformylation reaction to take place. Hydroformylation rates approaching those observed with preformed CozRhz( CO) i2 are obtained by adding either Co2(CO)s or Coq(CO)i2 to Rh4(C0)i2. A related ‘synergetic effect’ has been previously reported by Pino in the hydroformylation of diketene to methylsuccinic anhydride with cobalt and rhodium carbonyl mixtures (Co :Rh = 12:l) [ 181. An increasing Co,(CO)s/ Rh4(CO)i2 molar ratio up to 8 apparently slows down the deactivation processes. As shown in Table 2, addition of P(OPh), to Co2Rh2(CO)i2 enhances the activity of the system. The IR spectra of these solutions are in keeping with the presence of CO~_~R~,(CO),~_,[P(OP~),], species [19]; in these cases, however, it is difficult to state the most probable values for both n and X, since metal scrambling has been observed to occur on carbonyl substitution reactions [19]. On considering the rhodium as the only active metal center in these systems, Table 2 suggests the presence of a synergetic effect, which apparently decreases on increasing the P(OPh)$Co2Rh2( CO) i2 molar ratio when comparing the percent of hydroformylation after 48 h for identical L/M4( CO) i2 molar ratios given in Tables 1 and 2. TABLE 2 Effect of the P(OPh)3/Co2RhZ(CO)IZ molar ratio on the hydroformylation of cyclohexene (1.65 M) with the system CozRhz(CO)lz + x P(OPh)3 in toluepe (50°C, 1 atm of CO + Hz) Run



% Hydroformylation

after 48 h

CozRhz(CO)lz 1 2 3 4

7.4 6.0 5.1 7.0

2 4 6

9.4 14.2 32.5 51.2

a In mm01 1-l. bMolar ratio.

The influence of the nature of the ligand in the rhodium-catalyzed hydroformylation reactions has been thoroughly investigated in the past [ 31. The influence of the nature of the ligand in a cluster system could be further


complicated by its effect on the ligand substitution reactions, e.g. eqn. (3), and/or on fragmentation equilibria, e.g. eqn. (4). Nevertheless, a few spot tests have been undertaken in order to get a relative scale in cocatalytic ability of the ligands for cyclohexene hydroformylation with Rh,( CO) 12, obtaining roughly the sequence : P( OPh), - P(O-p-tolyl),

- PPh, > P(OMe), - PBus - AsPhs > ETPO = 0

The inactivity of the system when using ETPO is well in keeping with the previously reported inactivity of Rh4(CO)s(ETP0)4 [ 51. The systems Rh4(C0)12 + x L-L, where L-L is a diphosphine such as 1,2-bis(diphenylphosphine)benzene (DPPB) or ethane (DPPE) and bis(diphenylphosphine)methane (DPPM), are similarly inactive. A further complication in attempting kinetic studies or rationalization of the effect of the ligands arises from facile degradation reactions of most ligands under these experimental conditions. Thus, for instance, when employing phenyl-substituted phosphines or arsines as cocatalysts, gas chromatographic analyses during catalysis show the formation of both benzene and benzaldehyde [20]. In the case of PPh, and AsPh,, and depending on the L/Rh4(C0)i2 molar ratio, these products may also account for cu. 30 - 40% of the phenyl groups initially introduced as ligands. As an apparent consequence, the system Rh4(C0)i2 + x PPh3, which for x = 3 - 5 shows an initial hydroformylation rate close to that of the corresponding Rh,(CO) i2 + x P(OPh), system, becomes inactive in cu. 30 - 40 h when only ca. 35 - 40% of cyclohexene has been hydroformylated. Addition of either excess PPh, or cyclohexene at this stage has no effect. On the contrary, although trace amounts of phenol have been detected when using P(OPh)s, both and Rh4(C0)i2 + 4 P(OPh), systems, for instance, are Rh,(COhEP(OW,la able to hydroformylate 90 - 95% of cyclohexene in ca. 60 h, and addition at this stage of another 4 cm3 of cyclohexene restores the hydroformylation rate back to a value very close to the initial one. Up to 4 cycles of this kind have been performed, and cu. 600 mol aldehyde per mol cluster have been obtained. Hydroformylation of l-pentene and styrene Hydroformylation of 1 -pentene As shown in Table 3 the CozRh2(CO)i2 cluster alone is poorly active in the hydroformylation of l-pentene at 25 “C, under an equimolar mixture of CO and HZ (1 atm). A similar behaviour is observed for RhdCO)i2 and mixtures of Rh4(C0)i2 and Co2(CO)s. Higher activity is shown by the Rh4(C0)i2 + x L system particularly when L = PPh3. Tributylphosphine and triphenylphosphite are less active cocatalysts and DPPB is very poor. The Rh4(C0)i2 + x PPh, (x = 1 - 9) system has been investigated in more detail. Table 4, where the initial rates, the percent hydroformylation after 6 h and the molar ratios between linear and branched aldehyde are tabulated as a function of the PPh3/Rh,(C0)i2 molar ratio, shows a trend sufficiently comparable to that reported in Table 1. The

316 TABLE 3 Hydroformylation

of l-pentene (1.5 M) in benzene at 25 “C, 1 atm

Compound a




Rh4(CO)lz CozRhz(CO)lz Rh4(C0)12 Rh4(C0)12 Rh4(C0)12 Rh4(C0)12 Rh4(CO)u Rh4(C0)12






0.6 (0.12) 0.3 (0.12) cu. 0


4 4 4 2

11.2 3.1 3.1 0.2





% Aldehyde after 24 h 4.1 2.3


98.1 30.0 28.1 2.1



0.8 (0.5) 0.9 (0.6)

3.7 4.3 16.3 0.5


lop3 M.

bMolar ratio. ‘Initial hydroformylation rate expressed as mol aldehyde per mol cluster per h; the corresponding values after deactivation are reported in parentheses. dn-Hexanal/2-methylpentanal molar ratio. el,2-Bis(diphenylphosphine)benzene.

activity of this system increases on increasing the PPhs/Rh4(C0)12 ratio from 1 to 5; a further increase apparently causes a smooth decrease in the activity. No relevant induction periods or deactivation process have been observed in any case. Up to cu. 2000 mol aldehyde per mol Rh,(CO)lZ can be obtained in one batch with the most active system by repeated addition of fresh l-pentene at the end of each cycle. The highest regioselectivity has been observed using P(OPh),, which produces the linear aldehyde with more than 94% selectivity. The selectivity drops on using PBu,, PPhs or DPPB as cocatalyst. In all these hydroformylations the L/Rh4(C0)i2 ratio apparently has only a minor effect on the regioselectivity. Although complications hampering a comparison in terms of rates, such as those experienced with cyclohexene (induction for L = P(OPh),, deactivation for L = PPh,), are not so evident with 1-pentene, the significance of the lower activity on increasing the PPh3/Rh4(C0)i2 molar ratio should not be overstressed owing to the following observations. First of all, degradation of the PPh3 ligand also occurs under these milder conditions. Secondly, the highest PPh3/Rh4(C0)12 ratios (7 - 9) favour precipitation of a yellow compound, apparently as the olefin concentration decreases. This sparingly soluble yellow derivative shows an IR spectrum in nujol mull (v(C0) at 203O(s, sharp), 1995(vs), 1975(sh), 1950(w), 1803(s, sharp), 178O(vs), 1750(w) and 1740(w) cm-‘), which corresponds fairly well with that of the previously reported binuclear Rh2(C0)4(PPh,)4 derivative, obtained both from the Wilkinson catalyst HRh(CO)(PPh,), [21], and by reaction of Rh4(C0)i2 with excess PPh, [ 51. This compound is a well-established hydroformylation catalyst precursor active under mild conditions [ 51; also under our experimental conditions, once resuspended in fresh toluene in the presence of 1-pentene, it redissolves and catalyzes hydroformylation.

317 TABLE 4 Effect of PPh3/Rh4(C0)12 molar ratio on the hydroformylation of 1-pentene (1.5 M) with the system Rh4(C0)12 + PPh3 in benzene (25 “C, 1 atm of CO + Hz (1 :l)) Run


PPh3 b


% Aldehyde after 6 h


2.4 5.5 11.2 40.0 35.2 27.5 18.1 17.7

5.5 12.2 24.6 98.7 81.2 62.0 41.7 41.3

3.8 3.7 3.7 3.7 3.8 3.9 4.5 4.6

RhdCOh 1 2 3 4 5 6 7e 8e

5.9 5.9

5.8 6.6 6.2 6.0 6.1 6.2

a In mm01 I-‘. bMolar ratio. ‘Initial hydroformylation rate expressed in mol aldehyde per mol RhQ(C0)12 per h. dn-Hexanal/2-methylpentanal molar ratio. ePrecipitation of Rh2(C0)4(PPh3)4 occurs during the run.

When working with PPh,/Rh4(C0)i2 ratios in the range 1 - 4, the catalytic solutions are homogeneous and red-colored. Throughout the show essentially the presence of substituted catalytic run, spectra when using Rh,(CO)iz -,(PPh,), (x = 1 - 3) clusters. However, particularly Rh,( CO)s(PPh& or Rh,(CO)i* + 4 - 5 PPh3 as precursors, small amounts of a yellow precipitate may be present at the end of the reaction. This yellow precipitate, in contrast to the previous one, shows an IR spectrum in nujol mull (v(C0) at 2lOO(vw), 2040(w), 2010(w), 1995(w), 1985(m, sh), 1962(vs), 1915(m, sh), 1805(w), 1775(vw) cm-‘) which matches sufficiently well that of a yellow compound previously isolated by carbonylation at 80 atm of Rh4(CO)i,,(PPh,), in the presence of stoichiometric amounts of triphenylphosphine. Its formulation as Rh2(C0)6(PPh3)2 was inferred from elemental analysis and from the similarity of its IR spectrum with that of the corresponding CO~(CO)~(PBU,)~ [ 121. The original solutions, from which Rha(C0)6(PPh3)2 separated out, and this latter, once suspended in benzene, hydroformylate freshly added l-pentene with comparable rates. This is in keeping with the observation that all these solutions show IR spectra in the carbonyl stretching region very similar to those expected for a mixture of Rh,(CO)iO(PPh& and reported by previously Rh4(C0)9(PPh,)3. As Whyman [W, Rh,(CO),(PPh& condenses on dissolution also under an atmosphere of carbon monoxide or nitrogen. Hydroformylation of styrene Results referring to the hydroformylation of styrene are reported in Table 5. In this case the unsubstituted CozRhz(CO)iz and Rh4(C0)i2 clusters

318 TABLE 5 Hydroformylation and Ha mixture Run

of styrene (2 M) in toluene at 25 “C and 1 atm with an equimolar CO



Lb M4(COhz



ca. 0







25 - 50

2 3 4

CoaRhz(CC)iz CozRhz(CC)lz Rh4(CC)12

PPha PPhs -

3 3

25 50 25 - 50

ca. 0


Rh4(CC)lz Rh4(CC)r2 Rh4(CC)lz Rh4(CC)12 Rh4(CC)12 Rh4(CC)la Rh4(CC)ra Rh4(CC)lz Rh4(CO)l2

PPha PPha PPha PPha PBua DPPBf DPPBf P(CPh)a P(CPh)s

3 5 5 4 5 4 4 8 8

25 25 50 25 50 25 50 25 50

10.2 43.0 110.6 6.0 10.0 4.2 53.0 18.8 23.8

14.6 14.5 1.6 1.2 15.1 30.0 15.0 4.0 3.8

5 6 7 ge 9 10 11 12 13

4.3 25.6

8.6 1.6

a Concentration 7 x lop3 M. bMolar ratio. c Initial hydroformylation rate expressed as mol aldehyde per mol cluster per h. d2-Phenylpropanal/3-phenylpropanal molar ratio. e In THF as solvent. fl,2-Bis(diphenylphosphine)benzene.

are essentially inactive. In contrast, good hydroformylation rates are achieved with the substituted clusters. The ligand inducing the highest activity is again the triphenylphosphine and, as previously found [ 31, the branched 2-phenylpropanal is the isomer prevailing in the hydroformylated mixture (iso/n - 14). An even better regioselectivity is induced by the DPPB chelating ligand, which also induces a relevant activity in comparison with the poor results of the diphosphines with both cyclohexene and 1-pentene. Parallel experiments run at 25 and 50 “C show that an increase in the temperature in most cases is detrimental to the regioselectivity. Thus, at 25 “C the branched isomer is obtained in cu. 90 - 95% with either CozRhz(CO)iz or Rh4(C0)i2 and PPh, as cotiatalyst; this percent decreases to 55 - 60% on increasing the temperature to 50 “C. A similar loss of regioselectivity, in addition to a loss in activity, is observed when performing the hydroformylation in a more polar solvent such as tetrahydrofuran (THF) (run 8 in Table 5). Up to 2000 mol aldehyde per mol Rh4( CO)iz can be obtained in one batch with the most active mixture by repeated addition of fresh styrene at the end of each cycle.


Conclusions Several observations arising from the above described experiments deserve some further comments. First of all, this work has shown that some substituted Co4_nRhn(CO)1Z_xLx (n = 2, 4) clusters, as well as Co4_,Rh,(CO)iz + x: L mixtures, are precursors to hydroformylation catalysts very active also under mild conditions. The activity of these systems in a few cases is comparable to that reported for HRhtCO)(PPh~)~ Cl], although inferior to that achieved with HRh(CO)(DMBA), [Z]. When L = PPh,, extensive overlapping of the above system with the Wilkinson catalyst is suggested by isolation of Rh~~CO)~~PPh~)~and Rh*tCO)~(PPh~)~ complexes deriving from fr~entation of the tetranuclear substituted Rhea*-~~(PPh~)~ (X = 1 - 4) clusters; in this case their observation is favoured by separation from the reaction mixture because of their low solubility in toluene. Although the observation of corresponding fragmentation products when using phosphites as cocatalysts may be disfavoured by a greater solubility, recent NMR experiments have shown that tetranuclear Rh~~CO)~~-~[P(OPh)~]~ clusters are stable in solution also under 800 atm of either carbon monoxide or a 1:1 mixture of CO and hydrogen at room temperature. Dimeric products were observed only when decreasing the temperature to -25 or -50 “C, and there was no evidence of monomeric hydridic species under these con~tions 1221. Significant degradation of most ligands employed as cocatalysts occurs during catalysis. This corresponds to a loss of activity in the case of cyclohexene, while such deactivation processes are not so evident with other investigated olefins. Since an increasing number of papers report examples of syntheses of clusters containing the I,~~-PRor the p3- or pcPR moieties by degradation of coordinated ligands [ 23,241, it appears reasonable to suggest that clusters of this kind could also be present in our experiments, and that their activity greatly depends on the substrate. All these obse~ations are in keeping with more than one rhodium species being involved in the catalysis, depending on the experimental conditions and the substrate.

Experimental All the reactions were carried out under strict anaerobic conditions. Cyclohexene, I-pentene and styrene were Merck-Schuchardt products and were used either as purchased or distilled prior to use, with no significant differences in the two cases. Rh4(CO)r2 [25], Co*Rh~(CO)~~ [16], Rh,(CO),,-,[P(OPh),], (X = 1 -4) and Rh4(CO)1z_X(PPh,), (X = 1 -4) [?I were prepared by literature methods. The 1:l mixture of CO (99.5%) and H2 (99.99%) was purchased from S.I.A.D. IR spectra were recorded on both Perkin Elmer 457 and 781 grating spectrophotometers using CaF, cells. Mass spectra were measured on a


Varian Mat 114 spectrometer. Gas chromatographic analyses were performed on a Perkin Elmer SIGMA 115 or F 20 instrument with the internal standard methods under the following conditions: Separations


T (“C)

cyclohexene, cyclohexylaldehyde, 1,3,5triethylbenzene, toluene

Carbowax 1540 (15%) on Chromosorb G (2 m)


1-pentene, cis-trans-2-pentene, pentane

propylene carbonate (20%) on AlzOs (10 m)


hexanal, 2-methylpentanal, mesitylene, benzene

Carbowax 400 (8%) on Chromosorb G (3 m)


styrene, 3-phenylpropanal, 2-phenylpropanal, ethylbenzene, 1,3,5-trimethylbenzene, toluene

Carbowax 20 M (15%) on Chromosorb G (2 m)


Hydrogenation products of the aldehydes derived from cyclohexene, l-pentene and styrene were not detectable in any of the reported experiments. On the contrary, hydrogenation of the starting olefin was noticed, which was generally contained under a 0.5% upper limit. In the case of lpentene, we observed in addition that the residual olefin at the end of the hydroformylation was often isomerized to cis- and trans-2-pentene. The catalytic reaction was performed in jacketted glass reactors of 500 cm3 equipped at the top and bottom with two stopcocks fitted with self-sealing rubber serum caps and connected by polyethylene tubing to the 1:l CO/H2 gas cylinder; a take-off bubbler filled with ethyleneglycol was inserted near the reactor to ensure a slightly positive pressure during catalysis. In a typical experiment 20 cm3 of anhydrous toluene or benzene (Merck-Uvasol) and 0.132 g of Rh4(C0)i2 were charged under nitrogen atmosphere in the reactor. The desired amount of ligand was added in small portions with nitrogen flow under vigorous stirring. The reactor was evacuated, filled with the carbon monoxide and hydrogen (1 :l, 1 atm) and thermostatted to the required temperature. The reactions were started by injection into the reactor of deoxygenated olefin and internal standard using a hypodermic syringe through the top serum-capped stopcock. The reaction was periodically tested by removing samples from the bottom serum-capped stopcock with a microsyringe for gas chromatographic analyses and/or IR measurements.

Acknowledgements This work has been inspired by the late P. Chini and made possible by a grant of CNR (Italian Council of Research) under the Progetto Finalizzato per la Chimica Fine e Secondaria.



4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

D, Evans, J. A. Osborn and G. Wilkinson, J. Chem. Sot. (A), (1968) 3133; C. K. Brown and G. Wilkinson, ibid., (1970) 2753. B. Fell, A. Geurts and E. Muller, Angew. Chem. Znt. Ed. Engl., 10 (1971) 828. B. Cornils, in J. Falbe ted.), New Syntheses with Carbon Monoxide, Springer, Berlin, Heidelberg, New York, 1980, pp. 1 - 225; P. Pino, F. Piacenti and M. Bianchi, in I. Wender and P. Pino (eds.), Organic Syntheses via Metal Carbonyls, Vol. II, Wiley, New York, 1977, pp. 43 - 232. P. Chini, S. Martinengo and L. Garlascheili, J. Chem. Sot., Chem. Commun., (1972) 709. B. L. Booth, M. J. Else, R. Fields and R. N. Haszeldine, J. O~anometal. Chem., 27 (1971) 119. P. Chini and B. T. Heaton, Tetranuclear Carbonyl Clusters, in Topics in Current Chemistry, Vol. 71, Springer, Berlin, Heidelberg, New York, 1977, p. 1. B. T. Heaton, L. Longhetti, D. M. P. Mingos, C. L. E. Briant, P. C. Minshall, B. R. C. Theobald, L. Garlaschelli and U. Sartorelli, J. Organometal. Chem., 192 (1980) 330. B. T. Heaton, L. Longhetti, L. Garlaschelli and U. Sartorelli, J. Organometal. Chem., 192 (1980) 431. T. Kitamura and T. Joh, J. Organometal. Chem., 65 (1974) 235. B. T. Heaton, L. Strona, S. Martinengo, D. Strumolo, R. J. Goodfellow and I. H. Sadler, J. Chem. Sot., Dalton Trans., (1982) 1499. G. Ciani, A. Sironi, P. Chini and S. Martinengo, J. Organometal. Chem., 213 (1981) c37. R. Whyman, J. Chem. Sot., Dalton Trans., (1972) 1375. G. Csontos, B. Heil, L. Mark6 and P. Chini, Hung. J. Ind. Chem. (Veszprem), 1 (1973) 53. B. Heil and L. Marko, Chem. Ber., 101 (1968) 2209; G. Csontos, B. Heil and L. Mark6,Ann. N. Y. Acad. Sci., 239 (1974) 47. P. Chini and L. Garlaschelli, unpublished results. S. Martinengo, P. Chini, V. G. Albano, F. Cariati and T. Salvatori, J. Organometal. Chem., 59 (1973) 379. F. Spindler, G. Bor, U. K. Dietler and P. Pino, J. Organometal. Chem., 213 (1981) 303. P. Pino, Froc. Symp. Rhodium Homog. Catal., Sept. 1978, Veszprem, 98 F. D. Labroue, R. Queau and R. Poilblanc, J. Organometal. Chem., 186 (1980) 101. G. Gregorio, G. Montrasi, M. Tampieri, P. Cavalieri d’Oro, G. Pagani and A. Andreetta, Chim. Ind. ~Milan~, 62 (1980) 389, D. Evans, G. Yagupsky and G. Wilkinson, J. Chem. Sot. (Al, (1968) 2660. B. T. Heaton, IIZ Internat. Symp. Homog. Catal., August 30 - September 3, 1982, Milan, Italy. P. L. Bellon, A. Ceriotti, F. Demartin, G. Longoni and B. T. Heaton, J. Chem. Sot., Dalton Trans., (1982) 1671 and references therein. F. Demartin, M. Manassero, M. Sansoni, L. Garlaschelli and U. Sartorelli, J. Organometal. Chem., 204 (1981) ClO. S. Martinengo, P. Chini and G. Giordano, J. Organometal. Chem., 27 (1971) 389.