Mechanistic Aspects of a Highly Regioselective Catalytic Alkene

Bahram Moasser, Wayne L. Gladfelter, and D. Christopher Roe. Organometallics , 1995, 14 (8), pp 3832–3838. DOI: 10.1021/om00008a034. Publication Dat...
0 downloads 0 Views 825KB Size
Organometallics 1995, 14, 3832-3838

3832

Mechanistic Aspects of a Highly Regioselective Catalytic Alkene Hydroformylation using a Rhodium Chelating Bis(phosphite) Complex Bahram Moasser and Wayne L. Gladfelter" Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455

D. Christopher Roe Central Research and Development, E.I. d u Pont de Nemours & Co., Inc., Experimental Station, P.O. Box 80328, Wilmington, Delaware 19880-0328 Received March 24, 1995@ The rhodium-catalyzed hydroformylation of 1-octene using the bis(phosphite) ligand bbmb was studied using in situ, high-pressure 'H and 31PNMR and FT-IR spectroscopy. Four species, Rh(bbmb)(acac),Rh(bbmb)(CO)~H, and two dimeric complexes, appeared sequentially during different stages of the catalysis when Rh(acac)(CO)z was used as the catalyst precursor. These were independently synthesized and their reactivity studied. The major species present during catalysis was determined to be Rh(bbmb)(CO)aH(l),which was fully characterized. This hydride complex was shown to be an effective alkene isomerization catalyst. Using magnetization transfer, the rate of exchange between 1 and the terminal and internal vinyl hydrogens of 3,3-dimethylbutene were 0.62 and 0.51 s-l, respectively. The rapid, reversible nature of the alkene insertion establishes that the regiochemistry of the final aldehyde is not determined at the alkene insertion step or any event prior to it. The dimeric species were shown to convert to 1 via reversible addition of dihydrogen.

Introduction Recently, the class of chelating bis(phosphite1ligands for organometallic complexes has received considerable attention. New reports involving the use of these ligands in regio-1,2and stereo~elective~,~ hydroformylation, as well as asymmetric hydr~cyanation,~,~ have appeared which demonstrate their potential utility. The closely related monosaccharide- and disaccharidederived 1,2- and 1,3-diol phosphinites and phosphine phosphites have been utilized in asymmetric hydrocyanation7-lo and asymmetric hydroformylation," respectively, with success. With the perspective of catalysis in mind, a more careful study of this chemistry appeared warranted. There are numerous advantages of aryl bis(phosphite1 modified catalysis over the currently employed phosphine methodology. These compounds show very good normalliso ( d i ) regioselectivity and they are fairly robust toward hydrolysis. FurtherAbstract published in Advance ACS Abstracts, July 1, 1995. (1)Billig, E.; Abatjoglou, A. G.; Bryant, D. R. (Union Carbide) US. Patent 4,769,498, 1988. (2) Kwok, T. J.; Wink, D. J . Organometallics 1993,12, 1954. ( 3 )Sakai, N.; Nozaki, K.; Mashima, K.; Takaya, H. Tetrahedron: Asymmetry 1992,3,583. (4) Buisman, G. J. H.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Tetrahedron: Asymmetry 1993,4, 1625. (5)Baker, M. J.;Pringle, P. G. J. Chem. Soc., Chem. Commun. 1991, 1292. (6) Baker, M. J.; Harrison, K. N.; Orpen, A. G.; Shaw, G.; Pringle, P. G. J . Chem. SOC.,Chem. Commun. 1991,803. (7) RajanBabu, T. V.; Casalnuovo, A. L. J . Am. Chem. SOC.1992, 114,6265. ( 8 ) RajanBabu, T. V.; Casalnuovo, A. L. Pure Appl. Chem. 1994.66, 1535. (9) RajanBabu, T. V.; Ayers, T. A.; Casalnuovo, A. L. J . Am. Chem. @

SOC.1994,116,4101.

(10)Casalnuovo, A. L.; RajanBabu, T. V.; Ayers, T. A.; Warren, T. H. J . Am. Chem. SOC.1994,116,9869. (11)Sakai, N.; Mano, S.; Nozaki, K.; Takaya. H. J . Am. Chem. SOC. 1993,115,7033.

more, they are readily available from the vast achirall chiral diol synthetic methodology. Here we report our study of the rhodium hydroformylation of 1-octeneusing the bis(phosphite) ligand bbmb (2,2'-bis[(l,l'-biphenyl-2,2'-diyl)phosphitel-3,3'-di-tertbutyl-5,5'-dimethoxy-l,l'-biphenyl), first developed by Billig, Abatjoglou and Bryant1 (see figure 2). These workers observed very high regioselectivity in the bbmbmodified hydroformylation of propene with normal to is0 ratios up to 50:l. Cuny and Buchwald extended the scope of the process, employing this critical bis(phosphite) ligand, t o include the highly regioselective hydroformylation of a variety of functionalized terminal alkenes.12 To probe the mechanism of this important reaction, we employed i n situ, high-pressure spectroscopic studies to identify the critical species present under catalytic conditions. The hydroformylation of 1-octenewas monitored by i n situ, high-pressure 'H and 31P{'H} NMR and FT-IR spectroscopy. The species that were observed in the catalysis were independently synthesized and characterized, and their stoichiometric and catalytic reactivity was studied.

Results

In Situ Spectroscopy. Figure 1 shows the 'H and 31P{lH>NMR spectra obtained during a typical catalysis in the high-pressure NMR tube.13 Prior to the addition of CO/H2, all rhodium was present as Rh(bbmb)(acac) (2). Upon addition of a 1:l mixture of CO and H2 (20 atm, 2.1 mol % excess relative to alkene) 2 converted to Rh(bbmb)(CO)sH(1) and alkene isomerization began. ~~

(12) Cuny, G. D.; Buchwald, S. L, J . Am. Chem. SOC.1992,115,2066. (13) Roe, D.C. J . Magn. Reson. 1985,63,388.

0276-733319512314-3832$09.00/00 1995 American Chemical Society

Alkene Hydroformylation using a Rh Complex

Organometallics, Vol. 14, No. 8, 1995 3833 IPI,

3,4

lh

12h

L

--- -

9.70

9.50 ppm

6.0

5.2

ppm -10.64

-10.76 ppm

180

160

ppm

1H

31P (1H) Figure 1. In situ lH NMR (300 MHz) and 31P{1H}NMR (121 MHz) spectra of the hydroformylation of 1-octene using Rh(acac)(CO)Zand bbmb in toluene-&, initially under N2, at 23 "C then under 300 psi of 1:l CO/Hz. Spectra were acquired at 60 "C unless otherwise stated. l p = Rh(bbmb)(CO)zH;2 = Rh(bbmb)(acac);3, 4 = Rhz(bbmb)z(COk isomer. 5i = C6H13CHCH2; 5t = C~HI~CHCH~; 6 = ~ ~ ~ - C ~ H ~ ~ C H C H C H ~ / ~ ~ ~7n ~= ~ n-CeH17CHO; -CEH~~C 7i H=Ci-C8H17CHO; HCH~; l h = Rh(bbmb)(CO)&. Legend: (a) after 74 min, under Nz, at 23 "C. (b) Initially 40 "C.

+

Even in the presence of a very small amounts of 1,most of the 1-octene had isomerized to a mixture of internal alkenes. As more 1was formed, isomerization continued and aldehyde formation began. Heating the tube to 60 "C quantitatively converted 2 to 1 after 95 min and raised the turnover rate. At the same time, the signal for the enolic proton of 2,4-pentanedione appeared at 6 15.91 (br s) in the lH NMR spectrum. Only the vinyl signals of 2-octene were present as steady production of a 15:l ratio of l-nonanal/2-methyloctanal continued. The allylic methyl signals of cis- and trans2-octene were integrated as a 1:l ratio throughout the hydroformylation. As H2 was depleted at high conversions, two new species appeared in the 31P NMR spectrum, 3 and 4, in a 4:l ratio. We attribute these to two dimeric rhodium(0) species formed by the dinuclear reductive elimination of H2 from 1 (vide infra). These observations were corroborated by the study of this reaction in a high-pressure IR autoclave. Initially, the vco region of the spectrum did not show any metal carbonyl or aldehyde absorptions. When the autoclave was charged with 1:lCO/H2 (40 atm, 60 "C), the signals due t o 1 and aldehyde appeared with concomitant disappearance of the alkene C=C stretching signals. At longer times, signals from 3 and 4 appeared as the catalysis slowed down. Under these experimental conditions the turnover rate was 15 mol L-l h-l. The conditions under which our in situ spectroscopic studies were performed differ from those described by Billig and co w0rkers.l Under our experimental condi-

tions we do not observe any ligand degradation via hydrolysis or reaction with aldehyde, as determined by 31PNMR spectroscopy. Also, mass transfer of reactive gases from the head space of the NMR tube might not be efficient enough to replenish the solution, which is rapidly being depleted of CO and H2 by a highly active catalyst. Such mass transfer limited conditions would favor alkene isomerization. In the high pressure FTIR experiments, in which there is efficient stirring (-1500 rpm) t o overcome mass transfer limitations, however, it is difficult to quantify the degree of isomerization due to the overlap of multiple C=C stretching bands. Isolation and Characterization of Rh(bbmb)(C0)2H and Rhdbbmb)dCO)c. The identification of the above species required their independent syntheses and characterization. Figure 2 shows the lH, 31P{H}, and 13C(lH}NMR assignments of 1 that were obtained from the 13CO-enrichedcompound. The IR spectrum in toluene of 1 exhibits two bands at 2074 and 2016 cm-l in the YCO region and a hydride band at 1989 cm-l. These assignments were confirmed by preparation of Rh(bbmb)(CO)zD, whose IR showed no absorption at 1989 cm-' and a slight shift to lower frequency in the other two bands. The decrease in the vco stretching frequencies upon deuteration corroborates the assignment of VRh-H as being lower in energy than YRh-CO and is consistent with previous o b s e ~ a t i o n s . ' ~Definitive assignment of VRh-D, however, was complicated by other (14)Vaska. L. J . Am. Chem. SOC.1966.88, 4100.

Moasser et al.

3834 Organometallics, Vol. 14, No. 8, 1995

-\\

,

Figure 2. I3C{lH} NMR (125 MHz), 31P{1H}NMR (121 MHz), and lH NMR (500 MHz) of Rh(bbmb)(13CO)zH in CDzClz at 23 'C. CO and Hz. After 30 turnovers 31PNMR spectroscopy features in the region of the spectrum anticipated from force constant calculations. showed that 1 converted irreversibly (under these Heating a brown-red toluene solution of 1 under a conditions) to an unidentified species. Addition of CO purge of CO led to complete conversion t o orange 3 and to the system slowed the rate of isomerization. 4 in a constant 4:l ratio. The 31P{1H}NMR spectrum Alkene Insertion into the Rh-H Bond. To probe showed a pair of AA'AAXX' patterns similar to the the intimate mechanistic process responsible for the patterns for A-frame dirhodium s p e c i e ~ , ~and ~ - lthe ~ lH regioselectivity of the overall reaction, we undertook a NMR spectrum did not contain any hydride signals. The study of the rate of alkene insertion into the Rh-H IR spectrum indicated the presence of both terminal bond. We chose 3,3-dimethylbutene as o u r representa(2078 (m), 2052 (m), 2036 (msh), 2026 (ssh), 2011 (s), tive alkene because (1)the possibility of isomerization and 1968 (wsh) cm-l) and bridging (1865 (w), 1830 (w), did not exist, thus simplifying our analysis, and (2) the 1802 (w), and 1734 (w) cm-l) carbonyls, and the FAEV sterically disproportionate ends of the double bond MS was consistent with their formulation as two provide a higher limit test of this effect. The exchange isomers of Rha(bbmb)z(CO)4. These data are insufficient events shown in Scheme 1were rapid enough at 23 "C, to determine the structures of 3 or 4, but the 31PNMR under vacuum, to measure using magnetization transfer spectrum does favor structures that have the bisexperiments. Scheme 1 shows the exchange processes (phosphite) ligands bridging between the two rhodiums. involving the four-spin system of the rhodium hydride A related chelating-to-bridging transformation was and the internal, cis-terminal, and trans-terminal vinyl observed in the oxidative coupling of Ru(bbmbI(C0)s to hydrogens of 3,3-dimethylbutene. The four sites ingive [R~2(bbmb)2(CO)sl~+,~~ and the X-ray structural volved in the exchange (shown in boxes) are the only analysis of Rh4(bpnap)(CO)l0~~ (bpnap = 2,2'-bis[(l,l'observable species which are related by an underlying biphenyl-2,2'-diyl)phosphitel-l,1'-binaphthyl)verifies the ability of bulky bis(phosphite1ligands to bridge between chemical mechanism (dashed lines), on the basis of welltwo metals. Reports of other dirhodium complexes that established organometallic chemistry. The "outlined" display AA'MAXX' 31PNMR patterns and that are H s represent the (spin) labeled hydrogens leading to structurally characterized have appeared in the case of the magnetization transfer. [Rh(dppe)Cll~~-C1)~(~-CH2)22 and [Rh(dip~e)12@-H)2.~~ Consequently, exchange of Rh-H with the internal The 31PNMR data are not reported in sufficient detail vinyl hydrogen of 3,3-dimethylbutene involves insertion to compare the important fine structural features of the of 3,3-dimethylbutene in a terminal manner, followed spectra of these chelating rhodium diphosphine comby P-hydride elimination of the geminal methylene plexes to the ones in our study. hydrogen of the putative linear rhodium alkyl. ConSolutions of a mixture of 3 and 4 react quantitatively versely, insertion of the alkene in an internal manner, with H2 or H2/CO (20 atm) at 60 "C in 17 min t o leading to a branched rhodium alkyl, followed by regenerate 1. This facile reaction, represented in eq 1, of the terminal methyl hydrogens, brings is consistent with the chemistry of C O ~ ( C and ~ )Rh~ ~ ~ scrambling , ~ ~ about exchange of Rh-H with the terminal vinyl (PPh3)2(CO)zH.26 hydrogens of 3,3-dimethylbutene. 2Rh(bbmb)(CO),H ==H, Rh,(bbmb),(CO), (1) Figure 3 shows the vinylic region of the spectra during the experiment. The time-dependent z-magnetization Alkene Isomerization. 1-Octene isomerization by of the four sites (three of which are shown here) shown 1 was rapid at 23 "C but short-lived in the absence of in this figure were quantitatively expressed by a series (15)Jenkins, J . A.; Cowie, M. Organometallics 1992,11, 2767. of four coupled differential equations, derived from the (16)Kramarz, K. W.; Eisenberg, R. Organometallics 1992,11,1997. Bloch equations modified for chemical exchange. Figure (17)Kubiak, C. P.;Woodcock, C.; Eisenberg, R. Inorg. Chem. 1982, 21, 2119. 4 shows plots of the areas of the internal and terminal (18)Mague, J . T.; Sanger, A. R. Inorg. Chem. 1979,18,2060. vinyl hydrogens vs delay time after selective saturation (19)Shafiq, F.;Eisenberg, R. Inorg. Chem. 1993,32,3287. of the rhodium hydrogen along with the best-fit solu(20) Moasser, B.; Gross, C.; Gladfelter, W. L. J . Organomet. Chem.,

+

1994,471, 201. (21)Moasser, B.; Gladfelter, W. L. Submitted for publication in Inorg. Chim. Acta. (22) Ball, G. E.; Cullen, W. R.; Fryzuk, M. D.; James, B. R.; Rettig, S . J . Organometallics 1991,10, 3767. (23)Fryzuk, M. D.; Jones, T.; Einstein, F. W. B. Organometallics 1984,3,185.

(24)Alemdaroglu, N. H.;Penninger, J. M. L.; Oltay, E. Monatsh. Chem. 1976,107,1043. (25) UngGary, F.; Marko, L. J . Organomet. Chem. 1980,193, 383. (26)Evans, D.; Yagupsky, G.; Wilkinson, G. J . Chem. SOC.A 1968, 2660.

Alkene Hydroformylation using a Rh Complex

Organometallics, Vol. 14, No. 8, 1995 3835

Scheme 1. Spin Saturation Transfer Mechanism

[Rh] = Rh(bbmb)(CO)

x

of the ultimate rhodium alkyl, it is the latter microscopic step (migratory insertion of alkene), among the complex series of events implied in Scheme 1, which is responsible for the small differences in the overall exchange rates. The stabilities of the intermediate rhodium alkyls are thus reflected in their transition states for insertion. In any event, it is the overall insertion of alkene whose rate is evaluated by this experiment and is relevant to the hydroformylation reaction.

H

Discussion .

5.90

5.80

5.70 ppm

.

5.00 4.90

I

4.80 ppm

Figure 3. 'H NMR (500 MHz) spectrum of the vinylic region showing magnetization transfer between Rh(bbmb)(C0)2H,(1)and 3,3-dimethylbutenein C6D6 at 23 "c,using the selective saturation pulse sequence. tions from the numerical integration. The hexchange's obtained from these calculations are shown in Figure 5. Surprisingly, the rhodium hydride was found to exchange a t nearly the same rate with the cis-terminal (0.62 s-l), trans-terminal (0.62 s-l), and the internal (0.51 s-l) vinyl hydrogens of 3,3-dimethylbutene even though the two sides of the double bond provide a sizeable steric contrast for the insertion step. In three separate experiments changing the concentration of alkene relative to rhodium from 1 to 5 to 10 equiv caused no change in rate of exchange. The rate independence of alkene concentration and the observation of 1 as the only rhodium species during the magnetization transfer experiment are supportive of dissociation of CO from 1 as the rate-determining step in the overall exchange mechanism (Scheme 1). Because initial binding of 3,3-dimethylbutene to rhodium is unlikely to predispose the reaction to any particular regioisomer

The reaction between Rh(acac)(CO)z and bbmb is rapid at room temperature giving Rh(bbmb)(acac),(2), which subsequently converts smoothly to Rh(bbmb)(COkH, (1) under catalytic conditions. It is clear from the in situ lH NMR results that 1 is a very efficient alkene isomerization catalyst, effecting rapid isomerization even at very small concentrations. The isomerization of 1-octene, catalyzed by 1,was independently studied and shown to be first order in 1. The reaction proceeds to equilibrium at a fast rate, the equilibrium composition of the alkenes corresponding favorably with that predicted from calculations based on Benson's thermochemical data.27 Under hydroformylation conditions, as observed by in situ 'H NMR spectra, most of the 1-octene was converted t o cis- and trans-2-octene before hydroformylation began. These experiments, along with the saturation transfer results, demonstrate that the alkene insertion step is rapid, is reversible, and cannot be regioregulating. Because mechanisms involving n-allyl type intermediates could not explain the exchange processes observed with 3,3-dimethylbutene, (27) Benson, S. W.; Cruickshank, F. R.; Golden, D. M.; Haugen, G. R.; O'Neal, H. E.; Rodgers, A. S.; Shaw, R.; Walsh, R. Chem. Reu. 1969, 69, 279.

3836 Organometallics, Vol. 14, No. 8, 1995

Moasser et al. At this stage, we cannot predict which of the remaining steps in the catalytic cycle cause the high regioselectivity observed for bbmb-based catalysts. Considering that rhodium complexes containing bulky monodentate aryl phosphites do not exhibit comparably high regiosele~tivities,~~~~~ the chelating nature of the ligand must be an important factor. One of the difficulties in making such an assignment is the uncertainty in the mechanism itself. While there appears t o be general agreement on the parts of the hydroformylation cycles, some important details remain controversial. Perhaps none has been more discussed than the nature of the aldehyde-forming step. Especially in cobalt-catalyzedreactions there is a substantial body of evidence supporting the binuclear reductive elimination reaction (eq 2) as playing at least a competing role in aldehyde f ~ r m a t i o n . ~ ,One - ~ ~ of the attractive aspects of this mechanism is that the activation of molecular hydrogen can be accomplished by the welldocumented reaction shown in eq 3.24

Trans Terminal Hydrogen 25fJ

5

Cis Terminal Hydrogen

Co(CO),H 5

+ Co(CO),[C(O)Rl - Co,(CO), + RCHO

-r

5

Internal Hydrogen

6

4

1'2 1'6 2'0

2'4

Co,(CO),

2'8

3'2

2 9 s Figure 4. Integrated areas of the internal (bottom),trans-

terminal (middle),and cis-terminal (top)hydrogens vs t m i x , the delay time after selective saturation of the hydride of Rh(bbmb)(CO)zH(1).

\

f! kt = 0.62 s-l

ki = 0.51 f1

e3

[Rhl

+

for the internal (ki),trans-terminal (k,) and cis-terminal(k,)hydrogens of 3,3-dimethylbuteneand Rh(bbmb)(CO)zH(1).[Rhl = Rh(bbmb)(CO)z. Figure 5.

kexchange

we conclude that the hydride on Rh(bbmb)(CO)zH is transferred to alkene via a migratory insertionlpelimination mechanism. This contrasts with studies on cobalt systems in which isomerization has been shown to proceed without alkene e l i m i n a t i ~ n . ~ ~ - ~ l

-

2Co(CO),H

(3)

Related reactions, involving rhodium complexed by several phosphine ligands, are not as likely to compete with effective mononuclear steps, and little evidence for the binuclear reductive elimination has been found. Only in situations especially designed to promote such steps is there a real suggestion that binuclear pathways might be competitive. These include coordinating two rhodiums in close proximity using a complex bridging phosphine ligand.37!38The possibility of binuclear reductive elimination in rhodium mono(phosphite) complexes has been c ~ n s i d e r e d . ~ ~ The equilibrium between Rh(bbmb)(CO)zHand Rhz(bbmb)z(CO), (eq 1) occurs rapidly under mild conditions, and this similarity to the cobalt chemistry is striking. Both the small cone angle and the enhanced n-accepting ability of the phosphites (relative t o phosphines) may enhance the rate of the binuclear reaction shown in eq 4. A catalytic cycle involving eq 1 as the Rh(bbmb)(CO),H

k, = 0.62 s-'

+ H,

(2)

+ Rh(bbmb)(CO),[C(O)RI Rh,(bbmb),(CO), + RCHO

(4)

Hz activation step and eq 4 as the aldehyde-producing step can avoid moving into the f 3 oxidation state. The reduced basicity of phosphites vs phosphines could render the complex more disposed to proceed via such (28) Piacenti, F.; Bianchi, M.; Frediani, P.; Matteoli, U.; Lomoro, A. J . Chem. SOC.,Chem. Commun. 1976, 789. (29) Bianchi, M.; Piacenti, F.; Frediani, P.; Matteoli, U. J . Organomet. Chem. 1977, 137, 361. (30) Casey, C. P.; Cyr, C. R. J . Am. Chem. SOC.1971, 93, 1280. (31)Casey, C. P.; Cyr, C. R. J.Am. Chem. SOC.1973, 95, 2248. (32) Van Rooey, A.; Orij, E. N.; Kamer, P. C. J.; van Leeuwen, P. W. N. M. Organometallics 1995,14, 34. (33)Trzeciak, A. M.; Ziotowski, J . J . J . Mol. Catal. 1988, 48, 319. (34) Kovacs, I.; Ungirary, F.; Mark6, L. Organometallics 1986,5,209. (35) Ungirary, F.; Marko, L. Organometallics 1983, 2, 1608. (36) Hoff, C. D.; Ungirary, F.; King, R. B.; Marko, L. J . Am. Chem. SOC.1985, 107, 666. (37) Broussard, M. E.; Juma, B.; Train, S. G., Peng, W.-J., Laneman, S. A.; Stanley, G. G. Science 1993, 260, 1784. (38) Suss-Fink, G. Angew. Chem., Znt. Ed. Engl. 1994,33,67. (39)Jongsma, T.; Challa, G.; van Leeuwen, P. W. N. M. J. Orgunomet. Chem. 1991,421, 121.

Alkene Hydroformylation using a Rh Complex a pathway. Further research is necessary first t o determine whether such a reaction is possible and second whether it is fast enough to compete with mononuclear processes. The normal to is0 aldehyde regioselectivityfor 1-octene hydroformylation, although high, is not as good as what was observed for propene by the Union Carbide gr0up.l The choice of substrate and other differences in experimental conditions (e.g., catalyst concentration, ligand to metal ratios, batch vs continuous reactors, and mass transfer efficiencies) could easily account for such a discrepancy. The likelihood of bimetallic reaction pathways, for example, would be greater at higher rhodium concentrations and under mass transfer limited conditions.

Conclusion Rh(bbmb)(CO)zH,(1)was the predominant rhodium species present in the bbmb-modified hydroformylation of 1-octene starting with Rh(acac)(CO)n. Isomeric nonanals were formed in a constant d i of 15:l during this time. In the depleted H2 regime of the catalysis, 1 was converted to a pair of closely related dimeric species, 3 and 4. This equilibrium was independently studied, and it was shown that under CO or N2 1 was converted to a constant 4:l ratio of 3 t o 4. This reaction was reversed by addition of H2. These observations raised the possibility of binuclear reductive elimination of aldehyde as the regiochemically significant element in the catalysis. Isomerization of 1-octeneto internal octenes was also rapid under hydroformylation conditions, even in the presence of small amounts of 1. During the bulk of the catalysis, the substrate consisted entirely as a mixture of internal alkenes on the time scale of the NMR experiment. The 1-catalyzed isomerization of 1-octene was also rapid, although short-lived, under N2. The migratory insertioddeinsertion of 3,3-dimethylbutene into the Rh-H bond of 1 was studied using magnetization transfer experiments and found to be rapid and nonselective with regards t o regiochemistry of the insertion step. The results require that aldehyde regioselectivity in the hydroformylation reaction be determined subsequent to the alkene insertion step.

Experimental Section General Remarks. The preparation ,and purification of materials were performed under prepurified nitrogen using standard Schlenk-type techniques. CP grade CO, W O (99.2%), Hz, and CO/H2 (49.1% CO) were purchased from Matheson, Isotec, Genex, and Air Products, respectively. Rh(acac)( C O ) Z , ~Rhz(C0)4(p-C1)~,~~ ~,~~ Rhz(C~H4)4(p-C1)2,~~ Rh(acac1( C Z H ~ 2,4-pentanedione-d~~~ )~,~~ and bbmbZ0were prepared according to reported procedures. 1-Nonyl aldehyde, 1-octene, and 3,3-dimethyl-l-butene were purchased from Aldrich and distilled prior to use. RhC13.3HzO was obtained from Strem. Toluene, o-xylene, hexane, tetrahydrofuran, and Et20 were distilled from sodium benzophenone ketyl. o-Xylene used for IR studies was additionally distilled from sodium. Methylene (40)Varshavskii, Yu.S.;Cherkasova, T. G. Russ. J . Inorg. Chem. (Engl. Transl.) 1967,12,899. (41)Bonati, F.; Wilkinson, G. J . Chem. SOC.1964,3156. (42)McCleverty, J. A,; Wilkinson, G . Inorg. Synth. 1991,28,84. (43) Cramer, R.Inorg. Synth. 1990,28,26. (44)Cramer, R. Inorg. Synth. 1974,15,14. (45)Doyle, G.;Tobias, R. S. Inorg. Chem. 1968, 7,2479.

Organometallics, Vol. 14,No. 8, 1995 3837 chloride was distilled from CaHz. Infrared spectra were recorded on a Mattson Polaris FTIR spectrometer equipped with an HgCdTe detector. 'H, 31P,and 13CNMR were recorded at 300,121, and 75 MHz, respectively, on a Varian VXR-300s spectrometer. Chemical shifts are reported in ppm and referenced to residual deuterated solvent signals for lH and 13C NMR and external 85% H3P04 (6 0.00 ppm) for 31PNMR. Ultraviolet spectra were obtained on a HP 8452A diode array spectrometer at various dilutions and 6's were obtained from Beer's law plots. Low-resolution FAB mass spectra were obtained on a VG 7070E-HF instrument. Microanalyses were performed by M-W-H Laboratories. Melting points are uncorrected. Synthesis of Rh(bbmb)(acac). Dissolution of Rh(acac)(C0)z and bbmb in toluene under Nz led to effervescence and formation of yellow-green 2. This was recrystallized from 5:l PhCHhexane to yield a yellow-brown microcrystalline material in 85% yield. 31P{1H}NMR (PhCH3, 121 MHz): 6 146.5 (d, 302 Hz). FAB/MS W e ) : [Rh(bbmb)(acac)l+, 988; [Rh(bbmb)l+, 889. Mp (sealemz): 110 "C dec. Synthesis of Rh(bbmb)(CO)sH. Rh(bbmb)(acac) was generated in the manner described above. Exchanging the atmosphere with 1:l CO/H2 in the sealed flask for 45 min led to darkening of the solution to yellow, orange, red, and redbrown. Vacuum distillation of solvent gave a brown oil, which was recrystallized from 1 O : l CHzClz/hexane. Methylene chloride was added, and the resulting solution was evaporated to dryness. A further three coevaporations with methylene chloride afforded tan-brown microcrystals of the monomeric rhodium(1) hydride Rh(bbmb)(CO)zHin 82% yield. lH NMR (C6D6, 300 MHz): 6 3.19 (s, Ar-OCH31, 1.60 (s, Ar-C(CH3)3), 6.8-7.4 (m, Ar-H,), -10.60 (dt, 3.5, 3.5 Hz, Rh-H). IHC31P} NMR (C6D6, 300 MHz): -10.60 (d, 'JR~-H = 3.5 Hz). 31P{1H} NMR (C6D6, 121 MHz): 6 174.7 (d, 236 Hz). IR (PhCH3, cm-l): VCO, 2074 (VS), 2016 (VS); YRh-H, 1989 (W). IR ( m r , Cm-'): Yco, 2090 (vs),2016 (vs). Anal. Calcd for C48H45010P~Rh:C, 60.89; H, 4.79; P, 6.54. Found: C, 61.03; H, 4.96; P, 6.33. FAB/MS ( d e ) : [Rh(bbmb)(CO)z]+,945; [Rh(bbmb)(CO)]+,917; [Rh(bbmb)Hl+,890; [Rh(bbmb)(CO)l-,917; [Rh(bbmb)]-, 889. Mp (sealemz): 110 "C dec. Synthesis of Rh(bbmb)(CO)zD. Rh(acac-&)(CO)Z was synthesized from 2,4-pentanedione-d* and Rh~(C0)4(p-C1)~.~~ Rh(bbmb)(CO)zD was synthesized in the same fashion as 1 using bbmb, Rh(acac-d,)(CO)p and 1:l CO/Dz, in C6D6. IR (CHzClZ, cm-l): YCO, 2056 (SI, 2005 (s). Synthesisof Rh(bbmb)(lSCO)2H.Rh(bbmb)(13CO)zH was synthesized from Rh(acac)(WO)z and bbmb, as described above, in 85% yield. IR (hexane, cm-l): W C O , 2035 (vs), w c o 2019 (vs). R h ( a ~ a c ) ( ~ ~ Cwas O ) zsynthesized by condensing -2 atm (1.6 mmol) of 13C0 unto an evacuated toluene solution of Rh(acac)(CzH& (120 mg, 0.46 mmol, 0.15 M) which had been subjected to three freeze-pump-thaw cycles. When the temperature was raised, the characteristic dichroism of the product appeared. Removal of solvent under vacuum and recrystallization from hexane gave 108 mg (0.42 mmol, 91%) of Rh(bbmb)(l3C0)2. Rh(acac)(CzH4)~~~ was obtained from Rh(CzH4)4(~-Cl)z (71%),which was synthesized from RhC13.3Hz0, (39%ha Rh(bbmb)(13CO)zHwas thus available from RhC4.3HzO in a total 21% yield. Synthesisof Rhz(bbmb)s(CO)dand Isomer. A red-brown toluene solution of 1 was stirred vigorously at 60 "C, under a purge of CO. 31P{1H}NMR indicated complete conversion t o orange 3 and 4 in a constant 4:l ratio after 30 min. Recrystallization from 5: 1PhCHhexane afforded orange microcrystalline 3 and 4 in 93% yield. lH NMR (C6D6,300MHz): major isomer, 6 3.05 (s, Ar-OCH31, 1.45 (s, Ar-C(CH3)3), 6.6-7.7 (m, Ar-H,); minor isomer, 6 3.15 (s, Ar-oc&), 1.46 (s, ArC(CH3)3), 6.6-7.7 (m, Ar-H,). 31P{1H} NMR (C6D6, 121 MHz): major isomer, 6 170.4, 1 l J ~ h - p+ 2 J ~ = ~ 250.6 - ~ lHz; p /257.9 Hz. l3C{lH) minor isomer, b 167.2, 1 l J ~ h - p+ 2 J ~ ~ - = NMR (CDzClZ, 75 MHz): major isomer ( W O enriched), 203.8, 187.1 (these are complex second-order multiplets similar to

3838 Organometallics, Vol. 14, No. 8, 1995 those discussed in ref 18); minor isomer P3C0 enriched), 197 (br m), 182 (br m). IR (PhCH3, cm-'): VCO, 2078 (m), 2052 (m), 2036 (msh), 2026 (ssh), 2011 (SI, 1968 (wsh), 1865 (w), 1830 (w),1802 (w), and 1734 (w). FAB/MS (mle):[Rhz(bbmb)z(C0)4Kl+,1929; [Rh(bbmb)21+,1676; [Rhz(bbmb)(CO)#, 1105; nm (E, M-l cm-I)): [Rhz(bbmb)]+,992; UV-vis (CH3CN; A,, 326 (58901, 278 (13 434). Mp (sealedN2): 125 "C dec. Reversible Interconversion of Rhz(bbmb)z(CO)rand Rh(bbmb)(CO)nH.Stirring an orange toluene solution of Rhz(bbmb)z(CO)4,under a purge of H2, at 60 "C caused a color change to red-brown within 5 min. 31P{1H} NMR shows complete conversion to Rh(bbmb)(CO)nH,without decomposition. The same 4:1mixture of dimers is re-formed as described above. Spin Saturation Transfer Experiment with Rh(bbmb1(C0)zH and 3,3-Dimethyl-l-butene.A rigorously anaerobic NMR sample was prepared in the following fashion. A 5 mm NMR tube was charged with 25 mg (0.026 mmol) of Rh(bbmb1(C0)2H and evacuated. This was cooled to -78 "C, and approximately 3.5 p L (2.3 mg, 1equiv relative to Rh) of freshly distilled 3,3-dimethyl-l-butene and 0.45 mL of a CsDs (dried over Nao/Ph2CO)were vacuum-distilled into the tube to form a final 58 mM solution. The NMR tube was flame-sealed under vacuum. Spectra were acquired on a 500 MHz Varian Unity spectrometer over 2 h. A selective 90" saturating pulse was applied to the center of the Rh-H quartet, while the time dependencies of all the signals were monitored using a 90" nonselective pulse, applied at variable intervals (t,,,) after application of the selective pulse. A delay of 10.0 s was used before application of each subsequent selective pulse to ensure thermal equilibration of all spin states. The nonselective observation pulse was 12.8 ,us long. The described pulse was used to acsequence (RD-d2,,~-t,,,-,zI2-acquisition) quire 18 spectra. A total of 16 scans were accumulated for each value of t,,,, from 0.00 to 32 s. The temperature was regulated at 23 "C, and spectra were acquired without sample spinning. The area under the internal and terminal alkene multiplets was determined in an absolute integration mode for each individual spectrum. Plots of these values vs t,,, provided the necessary data for determination of kinetic pseudo-first-order rates. Difference spectra from a "dummy" decoupling frequency showed no detectable effect on the intensity or integrals of the hydride or olefinic signals. Using longer preacquisition delays showed no change as well, ensuring complete relaxation of the nuclei within the given acquisition conditions. Higher selective pulse power levels also had no effect, guaranteeing complete saturation of the hydride signal. The extent of NOE contributions to the magnetizations of the vinylic hydrogens of 3,3-dimethyl-l-butene was evalu.~~ prolonged, lowated using the method of N e u h a u ~Selective, power saturation of individual lines of each multiplets was used t o suppress selective population transfer and measure NOE exclusively. NOE effects measured using this highly sensitive method were found to be less than 2%. The rate constants for the insertion of 3,3-dimethyl-l-butene into the rhodium-hydride bond of Rh(bbmb)(CO)zH were determined by evaluating the set of coupled Bloch equations (46) Neuhaus, D.; Sheppard, R. N.; Bick, I. R. C. J . Am. Chem. SOC. 1983,105, 5996.

Moasser et al. modified for chemical exchange4? describing the time dependence of the longitudinal magnetizations (M(t))of a four-site system (Rh-H, tBuCH=CH2, C~S-~BUCH=CHH, and trans'BuCH=CHH):

+

dM(t)/dt = (K R)(M(t) - M") Here, K is the exchange matrix whose diagonal elements from a given describe the rate of loss of magnetization (M,) site and whose off-diagonal elements describe the rate of transfer of magnetization between different sites. Statistical effects were incorporated into K,by means of a Kubo-Sack probability matrix.48 The relaxation matrix, R, contains the spin-lattice relaxation rates (1/T1)for the individual sites and M" contains the magnetizations of the exchanging species at thermal equilibrium. Individual 2'1's were measured under conditions of temperature, concentration, and medium identical with those in the spin saturation transfer experiment. A Runge-Kutta routine within M a t h e m a t i ~ awas ~ ~ used to integrate the system of four differential equations. The parameters of initial intensities, M(O),equilibrium intensities, M", TI'Sand exchange rates were varied to provide a best fit to the experimental time-dependent magnetizations, which were plotted as absolute integrations of the lH NMR signals. High-pressureNMR and IR Experiments. In a typical NMR experiment, Rh(acac)(CO)z(17 mg, 0.066 mmol, 0.17 M, 18.7 ppt), bbmb (57 mg, 0.072 mmol, 0.18 M), and freshly distilled l-octene (200 pL, 143 mg, 1.27 mmol, 3.18 M) were dissolved in benzene-&, in a Nz-filled glovebox, transferred to a high-pressure sapphire NMR tube, and sealed with a titanium alloy pressure ~ a 1 v e . lThe ~ atmosphere in the tube was exchanged by several purge cycles from a 1:l COH2 tank and finally set to 20 atm. The tube was then placed in the probe of a 300 MHz Varian Unity spectrometer as expeditiously as possible. Spectra were acquired at various temperatures (see text) while the tube at was spun at -26 Hz. Similarly, a cylindrical internal reflection cel150 (a Parr autoclave embedded with a silicon crystal) was charged with Rh(acac)(CO)z (139 mg, 0.54 mmol, 60 mM) and bbmb (425 mg, 0.54 mmol, 60 mM), to which a solution of l-octene (4.2 mL, 3.0 g, 27 mmol, 3.0 M) in 4.8 mL of o-xylene was added under N2, at ambient temperature. The autoclave was first purged with and finally set t o 40 atm of 1:l CO/H2. This corresponds to 28 mmol each of CO and Ha (based on head space calculations). The IR autoclave was placed in the external bench of a Mattson Polaris FTIR spectrometer, and the spectra were collected as the well-stirred solution (-1500 rpm) was heated a t 60 "C. Background spectra at 60 "C and various CO pressures were recorded and used for spectral subtraction.

Acknowledgment. This work was supported by a grant from the National Science Foundation (Grant No.

CHE-9223433). OM9502190 (471McConnel1, H. M. J . Chem. Phys. 1958,28,430. (48) Johnson, C. S., Jr.; Moreland, C. G . J. Chem. Educ. 1983,50,

477.

(49)Wolfram Research, Inc.,Champaign, IL. (50)Moser, W. R.; Cnossen, J. E.; Wang, A. W.; Krouse, S. A. J . Catal. 1985,95,21.