466
Organometallics 1986, 5 , 466-473
for the band at 1930 cm-'. A 31PNMR spectrum of the reaction mixture was similar to that described above except that two of
the cobalt-phosphine resonances (73.2 and 65.4 ppm) had reversed their intensities. Solubilization of Polyphosphazene-SupportedCobalt (1) under Hydroformylation Conditions. The reaction vessel was charged with the cobalt-supported polyphosphazene (0.67 g, 0.69 mmol of phosphine, 5.0% w t Co), 1, and toluene (30 mL). It was placed in a reactor and subjected to 2000 psig of syn gas at 190 "C for 8 h. An infrared spectrum of the reaction mixture contained weak absorptions at 2040 and 2000 cm-l. A 31PNMR spectrum contained resonances at -5.9 and -19.9 ppm, assigned to the
phosphorus atom of the pendent phosphine units and the phosphazene phosphorus atoms, respectively.
Acknowledgment. We thank the Dow Chemical Co. for its support of the work a t the Pennsylvania State University. Registry NO.2,72796-22-6;4, 1184-10-7;Co,(CO)8, 10210-68-1; 1-hexene, 592-41-6; triphenylphosphine, 603-35-0; hexene, 25264-93-1; hexane, 110-54-3; heptanal, 111-71-7; heptanol, 53535-33-4; heptyl formale, 112-23-2; tri-p-tolylphosphine, 1038-95-5;tribenzylphosphine, 7650-89-7.
CobaWArylphosphine Hydroformylation Catalysts: Substituent Effects on the Stability of the Carbon-Phosphorus Bond Robert A. Dubois and Philip E. Garrou" Dow Chemical Company, Central Research -New
England Laboratory, Wayland, Massachusetts 0 1778
Received June 18, 1985
Triarylphosphines subjected to hydroformylation conditions (190 "C, 2000 psig of CO/H,) in the presence )~ phosphorus-carbon bond scission to give a variety of hydrogenolysis and CO insertion of C O ~ ( C Oundergo products as well as secondary products derived from hydrogenation, hydrogenolysis, and homologation. Electron-withdrawing substituents on the aryl groups enhance the initial rate of formation of both the hydrogenolysis and carbonylation products whereas electron-donating groups inhibit such reactions. Aryl group scrambling is observed when PR3 and PR'3 are subjected to the above hydroformylation conditions. The rate of such scrambling is comparable to the rate of formation of the decomposition products. Product analysis has ruled out contributions by free radical and/or ortho metalation mechanisms. An oxidative addition mechanism is proposed to be operative. Phosphorus-aryl bond cleavage proceeds at a much slower rate when olefin substrate is present.
Introduction It has recently become apparent that tertiary phosphines bound to metal complexes are chemically reactive and liable to undergo carbon-phosphorus bond scission, depending on the specific reaction conditions that they are exposed t0.l It is also becoming more and more apparent that reaction of the phosphorus-carbon bond with the transition metal to which it is bound is a general reaction having profound implications on homogeneous catalysis.2 We have shown in a preliminary communication3 and the preceding paper4 that phosphorus-carbon bond cleavage is a mode of catalyst deactivation during hydroformylation catalyzed by triarylphosphine-substituted cobalt carbonyl species C O ~ ( C O ) ~ / Pand R ~ the heterogenized catalysts C O ~ ( C O supported )~ on (diphenylphosphino)phenoxypolyphosphazene or diphenylphosphine-functionalized polystyrene. In this report we wish to present data concerning substituent effects on the rate of cleavage of the phosphorus-carbon bond and discuss some of the possible mechanistic pathways available for this reaction. The use of homogeneous catalysts in industrial chemical processing necessitates knowledge of the long-term stability of such catalyst systems since reactor downtime to unload Garrou, P. E.; Dubois, R. A,; Jung, C. W. CHEMTECH 1985, 123. Garrou, P.E. Chem. Reu. 1985, 85, 171 and references therein. Dubois, R. A.; Garrou, P. E.; Savin, K.; Allocck, H. R., Organometallics 1984, 3, 649. (4) Dubois, R. A,; Garrou, P. E.; Lavin, K.; Allcock, H. R., preceding paper in this issue.
0276-7333/86/2305-0466$01.50/0
or replace deactivated catalyst can dramatically impact the "operating cost" of the catalyst system. Most researchers are aware of the need to recycle catalysts, but unfortunately very few studies have determined the deactivation vs. time relationship for homogeneous or polymer-supported catalyst systems.
Discussion When triarylphosphines are subjected to hydroformylation conditions (Co,(CO),, 190 "C, 2000 psi of CO/HJ in the presence or absence of olefinic substrate, they undergo C-P bond cleavage to give a number of products, depicted in Figure 1 and quantified in Table I for reactions in the absence of olefin. Four of the seven products (2, 3, 4, and 8) come directly from the tertiary phosphine by C-P bond cleavage, compounds 2 and 8 come by hydrogenolysis, compound 3 comes by coupling of the two aryl groups a t the carbon initially bound to phosphorus, and compound 4 comes by carbonyl insertion into the C-P bond. Reaction profiles for the m-C1- and mCH3-substituted triarylphosphines (Figures 2 and 3) are typical for such reactions. They suggest the combination of parallel and consecutive reactions where the amount of hydrogenolysis product, 2, increases smoothly throughout the reaction while the initial carbonylation product, 4, undergoes relatively rapid reduction to the corresponding alcohol5 5 , which in turn undergoes cobalt-catalyzed ho(5) Aldehydes are only observed early on in the reaction. It is assumed that they are rapidly hydrogenated to the corresponding alcohols.
0 1986 American Chemical Society
CobaltlArylphosphine Hydroformylation Catalysts
Organometallics, Vol. 5, No. 3, 1986 467
Table I. Products of the Cobalt-Mediated Phosphorus-Carbon Bond Cleavage of Triarylphosphines
t\ y& X D-CF., m-cl" p-C1 m-CH3 H p-Me p-OMe
P-F o-CH,
products, mmol time," h 6
Co,mmol 2.58 2.58 2.58 7.8 7.8 7.8 2.58 2.58 2.58
mmol
5.4 6.6 6.2 19.1 19.8 19.8 6.8 7.6 6.6
2
4
3
1.6 3.3 2.8 7.2 6.0 3.4 1.0 0.8 0.9
21
28.5 62 56 47 60 33 27
5 0.3
6
0.6 0.6
tr 1.0 8.6 6.9 2.9 2.0 0.4 2.7
0.5
0.7 0.3 0.4 0.5
7
total, mmol
8
1.9
0.4 0.4
3.9 4.4 16.9 14.0 6.6 3.0 1.6 4.1
tr tr
tr
" A t 190 "C and 2000 psig of CO/H, (1:l).
/
P
1
x
CHO
2
4
3
r-
h
/L
g qlx +
CH20H
+
-
5
I /
! !
QX
I /
I
CH3
CHZCH20H
I
7
6
h
?
I'
/
N
X
,A
h
/
I
8 + (A)
2000
PSIG
C O l H 2 (1/i); 19O.c: CO2(i!l)8
CAT4LYS1
Figure 1. Products of the cobalt-mediated phosphorus-carbon bond cleavage of triarylphosphines. b
,.
Figure 3. Reaction profile for (p-MeC6H4)3Pdecomposition [Coz(CO)8, 3.9 mmol; phosphine, 19.8 mmol; 190 O C ; 2000 psig of CO/H2 (1:l);benzene, 120 mL].
?>De
ihlS,
Figure 2. Reaction profile for (??Z-ClC6H&Pdecomposition [Co,(CO),, 3.16 mmol; phosphine, 16.1 mmol; benzene, 98 mL; 160 "C; 1800 psig of CO/H2 (l:l)]. mologation to 7 or hydrogenolysis to 6.6 Thus compounds 4,5,6, and 7 all derive ultimately from the initial carbonyl insertion into the C-P bond. Compound 3 is rarely seen in more than trace quantities, and compound 8 is the phosphine product of the C-P bond scission. Any mechanisms for C-P cleavage must account for the wide variety of products shown in Figure 1which can be divided into three types: carbonylation, hydrogenolysis,and coupling. We sought to examine potential mechanistic routes to such decomposition products.
Free Radical Pathways. Free radical processes have been proposed to account for the hydrogenolysis products from C O ~ ( C Ocatalyzed )~ hydroformylation reactions of compounds that can form stable free radicals such as polycyclic aromatic hydrocarbons' and styrene.8 However for a system that would give less stable free radicals, for example, the decomposition of methyl triaylphosphine cobalt complexes to methane, the intermediacy of free radicals was precluded based on the lack of deuterium incorporation when run in deuterated t o l ~ e n e . We ~ attempted to determine the involvement of free radicals in Co-catalyzed hydroformylation by subjecting CO,(CO)8/ PR3 to our standard reaction conditions in the presence of ethylbenzene-dIo,a good deuterium source for radicals. A GC/MS examination of the reaction mixtures revealed no deuterium incorporation into the products benzene, diphenylphosphine,or benzyl alcohol. This argues strongly against the presence of phenyl or phosphinyl radicals. An additional argument against the presence of free aryl radicals can be made on the basis of the aryl-coupled products. In all cases where biaryls were produced only (7) Feder, H. M.; Halpern, J. J. Am. Chem. SOC.1975,97, 7186.
(8)Marko, L. In "Fundamental Research in Homogeneous Catalysis"; Gragianzi, M., Grongo, M., Eds.; Plenum Press: New York, 1984; Vol. 4.
(6) Wender, I.; Pino, P. "Organic Synthesis Via Metal Carbonyls"; Wiley-Interscience: New York, 1977.
(9) Kemmit, R. D. W.; Russell, D. R. "Comprehensive Organometallic Chemistry"; Wilkinson, G., Ed.; Pergammon Press: New York, 1982 Vol.
5, p 70.
468 Organometallics, Vol. 5, No. 3, 1986
Dubois and Garrou
Scheme I. Phosphorus-Carbon Bond Cleavage by an Ortho Metalation Mechanism Y
9 P-
'/I 10 h
, 2
x
CO/P>
. si Figure 4. Total decomposition products vs. time for PAr,. Y
Table 11. Substituent Effects on Initial Rates of Product Formation from C-P Bond Cleavage initial rate formatn, mmol mL-' h-l X IO3 Ar in hydrogenolysis carbonylatn total ArnP (4 + 5 + 6 + 7) (2 + 4 + 5 + 6 7) (2)
+
R
=
@X
those which could arise by coupling of the aryls through the carbons initially bound to the phosphorus were detected; that is, the aryl ring substitution pattern was preserved. No products of coupling between aryl radicals and solvent, benzene or toluene, were detected.'O Ortho Metalation. Ortho metalation is commonly observed in transition-metal-phosphine catalyst systems; however, P-C bond cleavage products from such intermediates have not been reported.15 The involvement of ortho metalation intermediates during the generation of hydrogenolysis and CO insertion products was considered as follows. The H of intermediate 9 (Scheme I) could transfer to the aryl carbon bound to phosphorus, and the carbon-phosphorus bond could undergo cleavage to give the aryl cobalt species 10. Species 10 could undergo hydrogenolysis to give 2 or CO insertion to give carbonylation products. If such a process were operative, the carbonylation products would have ring substitution patterns different from starting phosphine. For example, metasubstituted triarylphosphine would afford a mixture of ortho- and para-substituted aryl cobalt intermediates and para-substituted triarylphosphines should give meta-substituted products as shown in Scheme I. In all cases the only experimentally observed carbonylation products detected were those with retained ring substitution pattern. It is highly unlikely that ortho metalation could give rise to aryl cobalt intermediates which could only undergo hydrogenolysis to 2, so we therefore discount ortho metalation as a likely mechanistic route to any of the products derived from tertiary phosphine decomposition. Substituent Effects. The oxidative addition of haloaryls to low oxidation state transition metals has been shown to behave in a manner consistent with nucleophilic substitution where electron-withdrawing groups are rate (10) Others have used similar coupling arguments to argue against the presence of free radicals."-" (11) Kikukawa, K.; Yamane, T.;Takagi, M.; Matsuda, T. Bull. Chem. SOC.Jpn. 1979,52, 1187. (12) Lewin, M.; Aizenshtat, Z.; Blum, J. J. Organomet. Chem. 1980,
48
28
76
12
15.5
27.5
8.6
3.6
12.2
5.0
3.5
8.5
4.4
8.2
12.4
1.9
1.8
3.7
NA
0.8
NA
0.8
2.1
1.6
accelerating and electron-donating groups are rate retarding.16J7 Substituent effects on electrophilic substitution processes are essentially the reverse; that is, electron-withdrawing groups are rate retarding and electrondonating groups rate accelerating. The effect of substituents on the overall rate of C-P bond cleavage is shown in Figure 4; a plot of total product formation with time for most of the phosphines are listed in Table I. From such a plot one can construct a relative order of reactivity. Electron-donatingsubstituents (p-CH3, p-OCH3, p-F) effect a roughly fivefold lower rate than
184, 255.
(13) Fahey, D.; Mahan, J. J . Am. Chem. SOC.1976, 98, 4499. (14) Michman, M.; Kaufman, V. R.; Nussbaum, S. J . Orgonomet. Chem. 1979, 182, 547, 555. (15) Parshall, G. W. Acc. Chem. Res. 1970, 3, 139.
(16) Fitton, P; Rick, E. A. J. Organomet. Chem. 1971, 28, 287. (17) Garrou, P. E.; Heck, R. F. J. Am. Chem. SOC.1975, 98, 4115. (18) Wender, I.; Greenfield, H.; Metlin, S.; Orchin, M. J . Am. Chem. SOC.1952, 74, 4079.
Organometallics, Vol. 5, No. 3, 1986 469
CobaltlArylphosphine Hydroformylation Catalysts
2Jqtq
Table 111. Substituent Effects on the Rate and Selectivity of Cobalt-Catalyzed Homologation of Alcohols
@ CHz-OH
CH3
5
CHzCH20H
6
7
cat.
relative r i t e
p-OMe > p-Me > H > m-Me > m-C1 to*lc0l*lPi~Xi3
p-OMe > p-Me > m-Me > H
C02(CO)8"
> p-C1
617
p-OMe
cat.
p-Me
>loo Co2(CO)f
0.2
o-Me
m-Me
H
53
22
22
2
1.7
p-c1
m-OMe
2.5
10
Reference 12.
Scheme 11. Phosphorus-Carbon Bond Cleavage by an Oxidative Addition Mechanism
11
rJ
CHO
X
2
8
4
,-.-
d
c
?>me
those with somewhat lower electron-donating ability (Ph3P, rn-CH3, p-Cl). The phosphines with strong electronwithdrawing groups, namely, p-CF3 and rn-C1, constitute a third group too reactive to be included in this plot. In order to get some idea of the relative rate of the rn-C1 system, it was necessary to run the reaction thirty degrees lower than the others at 160 OC. With the assumption a doubling of the rate for every ten-degree difference in temperature, the rn-C1-substituted phosphine is about six times more reactive than PPh,. This order of reactivity generally holds for both the initial rates of formation of the hydrogenolysis, 2 , and carbonylation, 5 , products as shown in Table 11. This suggests a common intermediate for both hydrogenolysis and carbonylation as depicted in Scheme 11. The ratio of hydrogenolysis to carbonylation products 215 6 7 roughly parallels the reaction rates in Table 11; that is, electron-withdrawing groups promote hydrogenolysis vs. carbonylation while electron-donating groups have the opposite effect:
+ +
substituent 215
+
6
+
7
p-CF,
rn-C1 p-C1 p-Me m-Me H
p-OMe o-Me
5.5
5.3
0.5
1.9
1.0
0.8
0.8
0.3
A rate plot for the decomposition of (m-MeC,H,),P is shown in Figure 5. Again one clearly observes a pattern of consecutive reactions. Such data can be interpreted in terms of an intermediate such as 11, formed by oxidative addition, initially partitioning between a hydrogenolysis product, 2, and a carbonyl insertion product, 4. The former does not react further, but the latter, present only in very
lbIS
Figure 5. Reaction profile for (rn-MeCsH4)3P[ C O ~ ( C O )3.9 ~, mmol; phosphine, 19.1 mmol; 190 "C; 2000 psig of CO/Hz (1:l); benzene, 120 mL].
low levels when detected, earlier in the reactions, goes on to the benzyl alcohols 5, which often build up to a maximum and then fall off as the hydrogenolysis and homologation products 6 and 7,respectively, build up. This order is remarkably similar to the aforementioned data reported for attack of Pd on substituted aryl halides, an oxidative addition process, and is in contradistinction to that reported for electrophilic processes. Homologation. As noted above the benzaldehydes initially formed by carbonyl insertion into the C-P are subsequently rapidly reduced to the corresponding alcohols 5 under reaction conditions. Benzyl alcohols in turn are known to be susceptible to cobalt carbonyl catalyzed hydrogenolysis and homologation to give the observed products 6 and 7. Wender12proposed a benzylcarbonium ion intermediate formed by reaction between HCo(CO), and the benzyl alcohol to account for the effect of substituents on the rate of Co2(CO),catalyzed homologation of benzyl alcohols. In our system we similarly observe that electron-withdrawing groups inhibit reaction and electron-donating groups enhance it. For example, for the m-Cl-substituted phosphine, only 12% of the benzyl alcohol formed in the reaction mixture had undergone further reaction to 6 or 7 after 10 h of reaction yet, for the triphenylphosphine system, 70% of the benzyl alcohol formed in the reaction had been converted to 6 or 7 in 23 h under identical conditions. Although the orders
470 Organometallics, Vol. 5, No. 3, 1986 Scheme 111. Possible Mechanism for Co,(CO), Catalyzed P-C Cleavage and/or R Group Scrambling
Dubois and Garrou Scheme IV. Possible Cluster-Mediated P-C Bond Cleavage
/'\
12 b
of reactivity for our system vs. Wenders are similar, the substituent effect on the ratio of products 617 derived from the benzyl alcohols are very different (Table 111). For the cobalt-phosphine system not only is the ratio substantially higher but also it increases with the electron-donating ability of the substituents while for Wender's system it decreases. The fact that substituent effects on rates of homologation for the two systems are similar yet the product ratios are very different implies a rate limiting step to a common intermediate which then undergoes hydrogenolysis by two different mechanisms. Certainly a closer examination of these observations is called for. It is interesting to note that the m-CH,-substituted triarylphosphine gives essentially the same homologation product ratio 617 as triphenylphosphine, an expected result based on the electronic similarity of m-CH, to H. Reactions in the Presence of Olefin. For PPh, in the presence of 1-hexene substrate (a working catalyst), 0.93 mol of benzene was produced per mole of cobalt after 8 days, indicating an obviously slower rate of decomposit i ~ n . ~It, *is thus clear that the oxidative addition of the phosphorus-aryl bond when in competition with hydroformylation proceeds at a much slower rate than when the olefin substrate is not present. If the reactions are carried out without the 1-hexene substrate, diarylphosphine, &PH, is detected by 31PNMR. When 1-hexene is present, RzP(hexyl) is observed by 31PNMR. The formation of RzP(hexyl) could occur via catalyzed addition of the 1hexene to the RzPH via a phosphido cobalt hydride such as 12 and subsequent elimination. In fact, it is likely that species such as 11 dimerize to species such as 12 (R = R').
!2 Ty
This is consistent with the observations of Geoffroy and coworkers,lgwho found that Co2(h-PPh2)z(CO)6 (13) reacts (19) Harley, A. D.; Guskey, G. J.; Geoffroy, G. L. Organometallics 1983, 2, 53.
with 1-hexene under synthesis gas to give C O ~ ( C O ) ~ (PPhz(hexyl))z(14). This reaction could occur via an intermediate such as 12 which would give HCO(CO)~(PPh2(hexyl)). Removal of the CO/H2 pressure from the solution of HC0(C0)~(PPh,(hexyl))would result in dimerization to give 14, the observed product. Proposed Oxidative Addition Mechanism. Scheme I11 is offered as a plausible mechanism for catalytic formation of arene (R = Ph) and PPh,(hexyl) (R' = C,H9). PPh3could oxidatively add to an unsaturated Co-H to give an intermediate such as 12a (R = Ph). This could eliminate benzene to give 13. In the absence of olefin an intermediate such as 13 (R = Ph) could add H2 to give a species such as 12a (where R bonded to cobalt is H) which would eliminate HCo(C0)3(PPh2H).Under syn gas, in the presence of excess PR,, HCo(C0)&PPh2H)would exchange phosphine, leading to the observation of free PPhzH in solution. Insertion of CO into the Co-CHzCH2R bond to 12b would lead to the elimination of CH(0)CHzCH2R (hydroformylation). Insertion of CO into the Co-Ph bond of 12a would lead to the elimination of benzaldehyde and under reaction conditions subsequent reduction to benzylalcohol. A cluster-mediated reaction can also be envisaged generating hydrogenolysis and/or CO insertion products. Fachinetti and co-workers20,21 have observed the presence of the trinuclear cluster HCo,(CO), under hydroformylation conditions and have shown that H C O ~ ( C O ) ~ and HCo(CO), must both be present for the stoichiometric hydroformylation of 3,3-dimethylbutene. Scheme IV depicts a trimeric cluster, 15 (M = Co). The interactive phenyl group could easily transfer to the adjacent cobalt atom resulting in a cluster containing a bridging phosphido group, 16. Species 16 could then undergo reaction with either Hz or CO/H2 to give the experimentally observed products. Such a mechanism would also result in the maintenance of the stereochemistry around the phenyl ring, i.e., para products from para-substituted substrates." Aryl Group Scrambling. Scheme V is proposed to account for all of the above observations. It is not clear, at this point, which step is rate controlling; the reaction kinetics appear very complex. Since there is a slow plate c u t of cobalt metal as the reaction progresses, we cannot even say with certainty that the reaction is first order in cobalt concentration. We did want to have a feel for the relative rate of the back reaction kl since it has relevance to the rate of aryl group exchange if mixed phosphines were used, Le., PR2R'. We subjected equimolar mixtures of three different pairs of triaylphosphines to our standard hydroformylation conditions and monitored the reactions (20) Fachinetti, G.; Stefani, A. Angew. Chem., Int. Ed. Engl. 1982,21, 925. (21) Bradamante, P.; Stefani, A,, Fachinetti, G. J . Organomet. Chem. 1984, 266, 303. (22) Carty has recently described a trimeric hydrido ruthenium cluster similar to 15 which contains a weak phosphorus-phenyl interaction with a ruthenium atom. Reaction of this ruthenium species with H, at 80 "C resulted in the elimination of benzene.23 (23) Maclaughlin, S. A,; Carty, A. J.; Taylor, N. J . Can. J . Chem. 1982, 60, 88.
Organometallics, Vol. 5, No. 3, 1986 471
CobaltlArylphosphine Hydroformylation Catalysts
ArArPli
l ' ' ' ' / " ' ' i " " ~
-5.0
-6.0
-7.0
ppm
I
-41.0
~ ppm
-43.0
-42.0
~
Figure 6. 31PNMR of (m-C1C6H4)3P+ (P-CF,C~H,)~P reaction mixture [1.3 mmol of (m-CIC6H4)3P+ 1.3 mmol of (p-CF,C6H,),P + 0.47 mmol of C O ~ ( C O3) ~h;; 190 " C ; 2000 psig of C O / H z ;35 mL of benzene]. Scheme V. Proposed Mechanism for Phosphorus-Aryl Cleavage by Cobalt Carbonyl
7 ru
by 31PNMR. The pairs were selected on the basis of relative rate of product formation from carbon-phosphorus bond cleavage, that is, one pair of very reactive phosphines, one of moderate reactivity, and one of low reactivity. This order also corresponds roughly with the order of pK,'s and thus allows us to minimize complications that might arise from substantial differences in complexing ability within the phosphine pairs. In each case substantial scrambling was observed. Figure 6 shows the 31P NMR spectrum obtained from subjecting 1.3 mmol of (m-C1C6H4)3P,1.3 mmol of (p-CF3C&)3P, and 0.47 mmol of C O ~ ( C Oto) ~3 h at 190 " C and 2000 psig of CO/H,. Figure 7 reveals 31P spectra taken over a period of time for the reaction of PPh, and @-ClC&)3P (1.3 mmol each) under the same conditions. Table IV compares the time to get to 50% scrambling equilibration vs. the time needed to produce 50% of the experimentally observed cleavage products. It is clear that the rates of R group scrambling and the rates of formation of the hydrogenolysis and carbonylation products are comparable. Abatjoglou and Bryant% have recently described similar aryl group exchange catalyzed by group 8 transition metals. It is clearly from their work and ours that the use of "exotic" phosphine modifiers in cobalt- and/or rhodium(24) Abatjoglou, A. G.; Bryant, D. R. Organometallics 1984, 3, 932.
Table IV. Comparison of R Group Scrambling Rates to Phosphorus-Carbon Cleavage Rates for Ar3P - Ar'3P
Reactions Studied Ar3P t A r i P A
0
A r n A r ' P t ArArp'P
D
C
time to, h
Ar
Ar'
50% equilibratn
50% Droduct formation 12
0-
2.75
3.25 0.9
u a
Run at 3-fold dilution.
catalyzed hydroformylation would be commercially useless since the R group on the tertiary phosphine will eventually be the same as the olefin substrate.
Conclusions From this study and others',' it is now clear that phosphorus-carbon bond cleavage is a general reaction. Such cleavage does not appear to be free radical in nature
472 Organometallics, Vol. 5, No. 3, 1986
Dubois a n d Garrou
Ar =
+J
A:
=
@c 1
n
I
l""""""'"-''~'"';'
- , -6.0
-9.0
-6.0
T = 0 hr.
-9.0
~ ~ " " I " " ~ ' " ' ;
T
'
I
I
T
,
.
-6.0
-9.0
-6.0
T = 1 hr.
n
= 3 hr.
+
"""'T -9.0
T = 5 hr.
Figure 7. 31PNMR spectra of the reaction of PPh3 + (p-CIC6H4)3Pvs. time [3.8 mmol of (p-C1C6H4)3P 3.8 mmol of (PPh3) mmol of Co2(CO),; 190 "C; 2000 psig of CO/H2; 35 mL of p-xylene]. Table V 31PNMR chemical shifts, 6
phosphines Ar
Ph m-PhC1 p-PhMe
Ar' p-PhC1 P-PhCF, p-PhOMe
+ 1.3
PAr, -5.9 -5.3
PAr2Ar' -6.9 -7.5
PArAr', -8.0 -6.3
-8.5
-9.2
-10.0
but does appear to be influenced electronically by the substituents present on the aryl ring. The reaction appears to fit the traditional requirements of a "oxidative addition" both in its electronic response and its subsequent product distribution. The cleavage of phosphorus-carbon bonds in tertiary phosphines has obvious implications in the area of homogeneous catalysis in general and more specifically in the areas of polymer-supported homogeneous catalysis4 and assymetric catalysis when the site of assymetry resides on the phosphorus. Future studies in this area should address reactions where extended lifetime studies can be carried out and the specific question of deactivation addressed. Understanding the underlying causes for phosphoruscarbon bond breakage will hopefully point out ways to prevent or retard it.
Experimental Section 1. Equipment. All reactions were run in a 300 cm3 Autoclave
Engineers, Inc., magnedrive packless autclave, one of which was equipped with a sampling system. Glass liners when used (147 mm long, 41-mm i.d., 45-mm 0.d.) were wrapped with a few turns of Teflon tape near the top to provide a tight fit that minimizes spillover of the contents. An external heater was controlled with a LFE series 230 controller, and reaction temperature was maintained by a Parr temperature controller which circulated water through internal cooling coils. The stirring shaft and cooling coils were immersed between runs in concentrated H N 0 3 to remove cobalt. Infrared spectra were obtained on a Beckman 4240 spectrometer. Reaction mixtures were analyzed on a HewlettPackard ( H P ) 5710A (FID) gas chromatograph connected to a H P 3390A reporting integrator and equipped with one of the following 25-m H P fused silica capillary columns: A, high-temperature SE-30 (0.11-gm film); B, Carbowax 20M (0.33-pm film, 0.2-mm i.d.); C, high-temperature 5% methylphenylsilicone (0.33-gm film, 0.2-mm i.d.). Carrier gas (helium) flow rate was
PAr', -9.2 -7.0 -10.8
HPAr,
HPArAr'
HPAr',
-41.5 -43.3
-42.0 -44.4
-42.7 -45.4
1 mL/min a t 15 psi and 1OO:l split ratio. Injection port and detector temperatures were 250 and 300 "C, respectively. 31P NMR analyses were obtained on a JEOL FX-9OQ spectrometer equipped with a broad-band, tunable probe operating at 36.2 MHz. Chemical shifts were measured relative to 85% H3P0,. Mass spectra were obtained on a H P Model 5985 capillary gas chromatograph-mass spectrometer. 2. Reagents. Unless otherwise specified all chemicals were reagent grade and used as received. Solvents were degassed prior to use. The phosphines were obtained from Strem Chemicals. Triphenylphosphine was recrystallized three times from ethanol. Hydrogen (UHP) and carbon monoxide (UHP) were obtained from Matheson and used as received. 3. Typical Reaction. Decomposition of Tri-m -tolylphosphine under Hydroformylation Conditions. A reactor was charged under an argon atmosphere with 5.8 g (19.1 mmol) of the phosphine, 1.32 g (3.9 mmol) of CO,(CO)~,120 mL of benzene, and 320 pL (224 mg) of n-octane as internal GLC standard. After being sealed and pressurized to 1500 psig with a 1:l mixture of CO and HP,the reactor was heated with stirring to 190 "C at which point the pressure was about 2000 psig. Liquid samples were withdrawn periodically through a sampling port for GLC analysis. 4. Product Characterization. The hydrocarbon products 2-7 were characterized by comparison with authentic compounds using a combination of GLC coinjection and GLC-mass spectrometry. The only instances this approach did not work satisfactorily was for the isomers of xylene. Although we were not able to get separation of meta from para, mass spectra confirmed the xylene structure. Assignment of meta or para was based on the fact that the xylenes were derived from the corresponding m- or p-methylbenzyl alcohols. Similarly for (trifluoromethy1)toluene, although we did not have any authentic samples, the para assignment was made on the basis of its derivation from the corresponding p-(trifluoromethyl) benzyl alcohol. The secondary phosphine products 8 , were characterized by CLC-mass spectrometry and 31PNMR. Although the only authentic compound available to us was diphenylphosphine. In every
Organometallics 1986, 5 , 473-481 other GLC analysis and 31PNMR spectra revealed only one peak, indicating only one isomer which we assume to be the one with ring substitution pattern intact. 31PNMR and mass spectral data for secondary phosphines 8 are as follows [aryl functionality (6; JP-H, Hz, M')]: p-C6H4F (-44.96; 216.3; 222); C6H5(-41.35; 210.0; 186),p-C6H4Me(43.14; ...; 214); rn-C&4C1 (-41.5; ...; ...), p-C,H,CF, (42.0; ...; ...); pC6H40Me(-45.4; ...; ...). 5. Free Radical Trapping. A glass liner was charged with 2.5 g (9.54 mmol) of triphenylphosphine, 0.44 g (1.29 mmol) of CO~(CO)~, 112 mg of n-octane as internal standard, 10 g of ethylbenzene-d,,, and 20 mL of heptane and pressurized in an autoclave to 1500 psig of a 1:l mixture of CO and Hz. The temperature was raised to 180 "C and held there while stirring for 16 h and then raised to 190 "C for another 15 h. Capillary GLC analysis showed the normal production of benzene, benzyl alcohol, diphenylphosphine, and toluene. A GLC-mass spectral analysis showed no deuterium incoroporation into the first three products. We were not able to obtain a mass spectrum of the toluene due to masking by heptane solvent. 6. Aryl Scrambling. Three pairs of triarylphosphines were subjected to hydroformylation conditions in the presence of
473
C O ~ ( C Oas) ~described here for the pair tris(p-chloropheny1)phosphine and triphenylphosphine. A glass liner was charged under argon with 1.39 g (3.8 mmol) of tri-p-chlorophenylphosphine, 0.99 g (3.8 mmol) of triphenylphosphine, 0.44 g (2.58 mmol) of CO~(CO)~, 112 mg of n-octane as internal GLC standard,. and 35 mL of p-xylene as a solvent, then sealed in an autoclave, and brought up to 190 "C and 2000 psig of a 1:l mixture of CO and Hz. Liquid samples were periodically withdrawn for GLC analysis on capillary column C to monitor the decomposition reaction and for 31PNMR analysis to monitor aryl scrambling. It was not possible to monitor both by GLC because aryl scrambling takes place in the gas chromatograph. The chemical shifts for each of the phosphines are listed in Table V. Note that the secondary phosphines 8 also suffered aryl scrambling. Registry No. 1 (X = p-CF,), 13406-29-6; 1 (X = rn-Cl), 29949-85-7;1 (X = p-Cl), 1159-54-2;1 (X = p-Me), 1038-95-5;1 (X = rn-Me), 6224-63-1;1 (X = H), 603-35-0; 1 (X = p-OMe), 855-38-9;1 (X = o-OMe),6163-58-2;8 (X = p-F), 25186-17-8;8 (X = H), 829-85-6; 8 (X = p-Me), 1017-60-3;8 (X = rn-Cl), 99665-67-5;8 (X = p-CF,), 99665-68-6 8 (X = p-OMe), 84127-04-8; CO~(CO)~, 10210-68-1;1-hexene, 592-41-6.
Intermediates in the Palladium-Catalyzed Reactions of 1,3-Dienes. 3.' The Reaction of ($,q3-Octadienediyl)palladium Complexes with Acidic Substrates P. W. Jolly," R. Mynott, B. Raspel, and K.-P. Schick Max-Planck-Institut fur Kohlenforschung, 0-4330 Mulheim a.d. Ruhr, West Germany Received April 11, 1985
The stoichiometric reactions of (q1,q3-octadienediyl)palladium ligand complexes with a variety of acidic substrates, e.g., active methylene compounds, acetylacetone, and alcohols, have been followed by variable-temperature 31Pand 13C NMR spectroscopy and a number of intermediates, viz., (q'-octadienyl)-, ligand complexes, have been identified (q3-octadienyl)-, (q2,q3-octadienediy1)-,and (q2,v2-octadiene)palladium and characterized. The role of these species in the palladium-catalyzed telomerization of butadiene with the acids to give octadiene derivatives is discussed.
Introduction Although a number of transition metals, e.g., Co, Rh, and Ni, are known t o form active catalysts for the linear dimerization and telomerization of butadiene, ligand-modified palladium systems are the catalysts of choice. The product of the telomerization reaction is a mixture of 1and 3-substituted octadiene derivatives in which the terminally substituted isomer predominates (Scheme I). These reactions, first reported in 1967 by Japanese2 and American3 groups, have since been extended to substituted 1,3-dienes and to a wide range of nucleophiles. Although, as far as we are aware, they have not found industrial application, the products have been used as starting materials for a number of natural product syntheses (Scheme 11). The field is the subject of a recent re vie^.^ (1) Part 2, see: Benn, R.; Jolly, P. W.; Mynott, R.; Raspel, B.; Schenker, G.; Schick, K. P.; Schroth, G. Organometallics 1985,4, 1945. (2) Takahashi, S.; Shibano, T.; Hagihara, N. Tetrahedron Lett. 1967, 2451. (3) Smutny, E. J. J . Am. Chem. SOC.1967,89,6793. (4) Behr, A. Aspects Homogeneous Catal. 1984,5, 3. (5) Tsuii, J.; Kasuaa, K.: Takahashi, T. Bull. Chem. SOC.J m . 1979. 52, 216. (6) Tsuji, J.; Mizutani, K.; Shimizu, I.; Yamamoto, K. Chem. Lett. 1976, 773.
0276-7333/86/2305-0473$01.50/0
Scheme I
Scheme I1 0 II
The palladium component in these reactions can either be a zerovalent palladium complex, e.g., [Pd(PPh3)2(q2maleic anhydride)], or a palladium(I1) salt, e.g., PdCl,/py, whereby systems containing a P-donor ligand have the (7) Tsuji, J.; Yamakawa, T.; Mandai, T. Tetrahedron Lett. 1978,565.
0 1986 American Chemical Society