Interaction between dipotassium hexahalogeno-osmates K2OsX6 (X = Cl, Br) and olefins: metal hydride formation by allylic hydrogen atom abstraction and catalytic isomerization of olefins

Journal of Molecular Catalysis,

22 (1983)

61

61 - 71

INTERACTION BETWEEN DIPOTASSIUM HEXAHALOGENOOSMATES K20sX, (X = Cl, Br) AND OLEFINS: METAL HYDRIDE FORMATION BY ALLYLIC HYDROGEN ATOM ABSTRACTION AND CATALYTIC ISOMERIZATION OF OLEFINS .

ALBERT0

FLAMINI and ANNA MARIA GIULIANI

Zstituto di Teoria e Struttura Elettronica e Comportamento di Coordinazione de1 C.N.R., Area della Ricerca di Roma, 00016 Monterotondo St&one, Rome (Italy)

Spettrochimico dei Composti Via Salaria, Km. 29.5, P.B. 10,

(Received December 10, 1982)

Summary Dipotassium hexahalogeno-osmates, K20sX6 (X = Cl, Br) dissolved in non-aqueous inert organic solvents (benzene, o-dichlorobenzene, bromobenzene) by addition of dicyclohexyl-l&crown-6 ether (C&H,,06) are catalytic precursors for the multiple double bond shift in olefins. Studies on the interaction between K20sX, and several olefins (l-heptene, 1-heptene-3d,, 1-pentene, cis-stilbene, 3,3-dimethyl-l-butene) support an osmium hydride species, probably [OsHX,]‘, as the true catalyst, which affords the olefin isomerization through successive 1,2-addition-elimination steps to the C=C double bond. Such an alleged osmium hydride species is formed by allylic hydrogen atom abstraction from the olefin with formation of the corresponding conjugate diolefin.

Introduction With the aim of studying the catalytic properties of soluble halo-osmium complexes for olefin reactions, we have investigated the interaction between simple and substituted mono-olefins (1-heptene, 1-heptene-3d,, 1-pentene, cl-stilbene, 3,3-dimethyl-1-butene) and the dipotassium hexahalogenoosmates (K,OsX, , X = Cl, Br) dissolved in non-aqueous inert solvents (benzene, o-dichlorobenzene, bromobenzene) upon addition of the dicyclohexyll&crown-6 ether (CJ-I,,O,). There are several reasons for such studies. First, K20sX, compounds have not been investigated from this point of view, although they are widely used as starting materials in studying the coordination and organometallic chemistry of osmium [ 11. Second, many analogous halides of platinum group metals are well-known catalysts for homogeneous organic reactions of olefins [2]. Third, they are potential candidates for homogeneous catalysis, being 1Svalence electron species which undergo 0304-5102/83/$3.00

@ Elsevier Sequoia/Printed

in The Netherlands

62

ligand substitution reactions been reported [ 41.

[3]. A preliminary account of this work has

Results and discussion Isomerization of 1-heptene When l-heptene is added to the organic solution containing OsXi in the ratio [ olefin]/[OsX%] = 103, after heating under nitrogen, the olefin catalytically undergoes a multiple double bond shift and all possible isomers are formed at different rates, as shown in Fig. 1; early in the reaction large

4

2 Time

3== 0s

(h)

Fig. 1. Time course of the catalytic isomerization of 1-heptene in o-dichlorobenzene; 170 “C, [K(CZ&,j06)]20sC16 = 0.86 mM; [heptene] = 1.0 M.

T=

amounts of trans-2-heptene are formed (after 10 min at 170 “C [2-trans]/ ([a-cis] + [3-trans] + [3-cis]) = 4.1); when the system is allowed to reach equilibrium (after 24 h at 170 “C), the thermodynamically most stable mixture of isomers is obtained: 3-trans > 2-trans > 3-cis > 2cis > 1-heptene. The experimental methods used to follow the reaction were GLC (Fig. 2) and proton NMR (Fig. 3a). Identification of the various isomers was made by comparison with retention times or chemical shifts (Table 1) of authentic samples. The concentration of OsXz decreases as the olefin isomerization proceeds; at 30% conversion the amount of OsClz still present is 60% of the initial quantity; the turnover frequency of OsClz at 30% conversion and 170 “C is 23 mol olefin X mol catalyst -r X min-‘. Neither the solvent nor the crown ether is involved in the catalytic cycles; indeed when using C,DsBr as the solvent, after isomerization and distillation the heptene mixture did not show any signal in the 2H NMR spectrum. At first glance it appears that the crown ether behaves as a co-catalyst (increasing its concentration increases the isomerization rate), but this is not the case; its function is rather that of

63

20

30

25

35

6.0

40

5.0

4.0

d fppm

(min)

3.0 from

TMS

2.0

1.0

)

Fig. 2. GL-chromatogram of the isomeric heptenes in o-dichlorohenzene solution derived from catalytic isomerization of 1-heptene (1.0 M) by [K(C~&~06)]aOsC16 (0.8 mM) after heating at 170 “C!for 10 h. Fig. 3. NMR spectra of the isomerization mixture of 1-heptene (a) and 1-heptene-3da (b) in chloroform solution (Ba = 4.69 T). TABLE 1 ‘H NMR data (200 MHz) for heptenesa Compound

C-l

c-2

c-3

c-4

C-5

C-6

c-7

1-heptene trans-2-heptene trans-3-heptene

5.19 1.87 1.35

6.06 5.66 2.22

2.29 5.66 5.67

1.56 2.25 5.67

1.56 1.55 2.22

1.56 1.55 1.62

1.14 1.13 1.13

%hemical

shift in ppm.

64

co-solvent for the catalyst (incidentally, this fact directly catalysis occurs in the homogeneous phase).

proves

that

the

Nature of the true catalyst Although initial isomerization rates appear to show no, or a very short, induction period (see Fig. 1) and much OsXi can be recovered as [K(C&Is,0,)]20sXs, the hexahalogeno-osmates are only catalytic precursors for the C=C double bond migration, as indicated by the small dependence of the isomerization rates on OsXi concentration (Fig. 4). In this figure a comparison is made between isomerization rates of l-heptene and trans-Zheptene: the more crowded olefin reacts more slowly with both OsCli and OsBrz; moreover, addition of potassium halides (KX, 10m3 M) decreases the reaction rate. All these facts are consistent with the formation of a halo-olefin osmium complex by ligand dissociation before catalysis takes place (eqn. (1)):

OsX; + C7Hi4 = [OsX,*C,HrJ-

+ X-

(1)

Reduction in the isomerization rate of trans-Zheptene compared to 1-heptene, or bromide compared to chloride, is probably due to steric hindrance to the coordination of the hexahalogeno-osmates to the olefin. The concentration of the hypothetical [OsX,* C,H& in the equilibrium (1) is so low that no effect on the proton NMR spectrum of 1-heptene could be detected by adding to the solution increasing amounts of the paramagnetic 0~x2, even after heating. In order to clarify the nature of the true catalyst we have carried out some carefully designed experiments with the following results: (i) OsXz does not catalyse the isomerization cis + trans-stilbene . (ii) The presence of 3,3-dimethyl-1-butene does not affect the previous result. (iii) OsXi readily catalyses the above transformation in the presence of lheptene or any other olefin bearing allylic hydrogen atoms. lo

Fig. 4. Rate of isomerization of 1-heptene with varying catalyst concentration.

A, KzOsCl,j ;

W,KzOsBr6. Circles represent the rate of isomerization of trans-2-heptene under the same experimental conditions.

65

(iv) In the presence of Ccl, the isomerization of 1-heptene does not occur, and CHCl, is formed. These facts can be rationalized supposing that the true catalyst is an osmium hydride species and that the OS-H bond is responsible for the catalysis. Such a metal hydride is apparently produced through interaction of OsXz and the allylic hydrogen atoms of the olefin (the failure to secure any isomerization in experiments (i) and (ii) can be ascribed to the absence of allylic hydrogen atoms). The inhibiting effect of CC14can be explained remembering that Ccl, reacts with all transition metal hydrides to give CHCls [5]. In particular, the reaction between OsClg and l-heptene leading to such alleged osmium hydride species can be expressed as an abstraction of an allylic hydrogen from the olefin, as shown in the scheme below:

OsCl: + C7Hi4 1 \

[OsClg*C7Hi4]- + Cl1

[OsCl,H]’

+ H” + C&Hi2(heptadiene)

It is tempting to assign the formula [OsCl,H]= to the catalytic hydride species, even if it has been neither isolated nor directly detected in solution; and indeed such formulation is justified by the reactions which it undergoes: it reacts with Ccl4 to give the starting OsCli ; because under these conditions the isomerization of 1-heptene does not occur, CHCl, and heptadiene are formed in small amounts; and, most relevant, the concentration of OsClz does not change. The heptadiene has been directly observed in the reaction mixture using the GC-MS technique, as shown in Fig. 5. Here we notice that both solutions contain (apart from heptenes) heptane and C,H,Cl as impurities of the olefin and the solvent respectively; the catalyst solution contains in addition several species with a molecular ion of mass number 112, which appear to be octene isomers probably deriving from 4-octene contained in the blank; the catalyst solution also exhibits the presence of three species of molecular weight 96, which may be identified as 2,4-heptadiene isomers: their mass spectra are very similar to each other and closely resemble those reported in the literature for the 2,6heptadiene isomers (Table 2). Isomerization of 1-heptene-3d2 To better understand the mechanism of the catalysis, we prepared the olefin labeled with deuterium in the allylic position (l-heptene-3d,) and, after isomerization, we recorded the ‘H NMR spectrum (Fig. 3b). The deuterium label in position 3, whose chemical shift is 2.29 ppm, appears to have spread to all positions, since seven groups of resonances are observed. The assignments are based on the chemical shifts of the corresponding pure unlabeled heptenes (Table l), since ‘H and ‘H spectra for the same

I

I

20

25

30

35

min

Fig. 5, GL-MS chromatographs of solutions of l-heptene in o-dichlorobenzene after 20 ), K20sC1e added as catalyst. The number min; (---), blank with no catalyst; (above each peak is the mass number of the molecular ion appearing in the mass spectrum of the species originating the peak itself. TABLE 2 Mass numbers of the eight most intense peaks of the species with molecular ion m/e = 96 in Fig. 5 (a), and of the hepta-2,4-dienes (mixed isomers) reported in the literature (b); normalized intensities in parentheses (a)

(b)

96 (38) 81(100) 79 (17) 67 (13) 54 (20) 41(31) 39 (13)

96 81 79 67 54 41 39

[f-31 (46) (100) (25) (21) (22) (32) (22)

compound exhibit the same chemical shifts, except for very small and, for our purpose, unimportant isotope shifts [ 71. The analogous proton NMR spectrum of the unlabeled mixture of isomers (Fig. 3a) exhibits seven groups of signals which, starting from low field, are assigned as follows: (I) non-terminal vinylic of 1-heptene; (II) vinylic of 2- and 3-heptenes; (III) terminal vinylic of 1-heptene; (IV) allylic methylenic; (V) allylic methylic of 2-heptene; (VI) aliphatic methylenic; (VII) aliphatic methylic. The broad-band proton-decoupled *H spectrum in Fig. 3b deserves some comment. Since each resonance corresponds to a well-defined chain positon in a specific isomer, only one signal is found at 6.06 ppm, which

67 TABLE 3 Experimental (a) and calculated (statistical) (b) deuterium distributions (%) for the seven signals of Fig. 3b Signal

(a)

(b)

I II III IV V VI VII

0.7 16.6 1.5 19.9 21.4 16.6 16.6

0.6 13.1 1.2 13.1 18.6 27.9 25.5

corresponds to the C-2 position on the double bond of l-heptene (for which no isomers exist). Instead, two resonances appear for deuterium in position 1 of 1-heptene (6 cu. 5.2 ppm) corresponding to the two geometric isomers with the 2H atom cis or truns to the alkyl chain. Two resonances are also found for the methyl groups, one at 1.35 ppm assigned to the C-l methyl group of 3-heptene and the other at 1.13 ppm assigned to the terminal methyl groups of l-, 2- and 3-heptenes. While the chemical shifts are the same for the mixtures of isomers obtained from the labeled and the unlabeled starting reagent, the relative intensities are not (thermodynamic equilibrium was reached in both cases). This results from a deuterium distribution over the various chain positions different from that calculated for a random distribution (Table 3), the vinylic and the allylic positions, which are closer to the initially-labeled position, being more deuterated than expected for random distribution. It can be concluded that the deuterium movements occur during the isomerization of the olefin, leading to an isomeric mixture labeled on all carbons of all isomers. Two others remarkable features of the isomerization should be mentioned: (i) the isomerization rate is the same for l-heptene-3d2 and l-heptene within experimental error, and (ii) deuterium has very little effect on isomer distribution as determined by GLC. The first result indicates that the ratedetermining step in the isomerization cannot involve cleavage of a C-H bond to give the metal hydride as the active catalyst, since for reactions involving such a process the Kn/Kn value may be as high as 7 [ 81. The second result suggests that both the cleavage and the formation of the same bonds C-H(D) and/or OS-H(D) are important in the transition states of the catalytic cycles. Isomerization of l-pentene in the presence of I-heptene-3d2 In order to clarify the type of deuterium transfer (inter- or intramolecular) we have isomerized l-pentene in the presence of 1-heptene-3d2. After separation from the reaction mixture, the pentene isomers show the 2H spectrum reported in Fig. 6. The formation of deuterated pentenes demonstrates the intermolecularity of deuterium shifts involved in the isomerization.

68

d f ppm

from

TMS

)

Fig. 6. ‘H NMR spectrum of 1-pentene isomerized in the presence of 1.-heptene-3dz (Bo= 4.69 T).

In comparison to the spectrum of Fig. 3b, we notice the difference in the relative intensities of the signals, due to a different deuterium distribution and to a different number of positions of the same type in the carbon chain as well. Mechanism of the catalysis The proposed catalytic mechanism (Scheme 1) consists of successive 1,2-addition-elimination steps of the osmium hydride to the olefin with formation of alkyl intermediates, and it is consistent with all known facts, as summarized here: (i) A multiple double bond migration occurs, that is, all possible isomers are formed from the very beginning of the catalysis, because the decomplexation of the isomerized alkene is the slowest step in cycle 1. (ii) The catalyst is responsible not only for the double bond shift in olefins, but also for the geometrical isomerization, as for stilbenes, which can occur in cycles 2 or 3 of the scheme. (iii) The bans isomers predominate initially during the catalytic process, because the alkyl intermediates do not contain bulky ligands [9]. (iv) The appearance of deuterium on the second carbon atom of 1-heptene during the isomerization of 1-heptene-3dz can only occur in cycle 2 through anti-Markownikov addition of the osmium hydride to the olefin [lo]. In particular this finding, that is, the shift of deuterium between adjacent carbon atoms, definitely requires the participation of alkyl intermediates; the presence of a *H signal for the 2 position is specifically important to substantiate the proposed mechanism for the catalysis: if

69

Ii - OS + CH, = CHCH,R

CH2=CHCH2R

Scheme 1. Propoed scheme for the reactions occuring between the catalyst, OS-H (the other ligands, halogens, are omitted for clarity) and the olefin. They can be analyzed in cycles 1, 2 and 3.

all but this signal were present in Fig. 3, the 1,3-intermolecular-hydrogen shift, a recently discovered olefin isomerization mechanism [ 111, could equally well account for the double bond migration in our case.

Conclusion This work was mainly undertaken to seek organic reactions of olefins for which the hexahalogeno-osmates could act as homogeneous catalysts. Our results indicate that the hexahalogeno-osmates can be used as homogeneous catalysts for those olefin reactions for which metal hydrides play an important role; to this purpose they should be used together with a hydride source (the co-catalyst), and such species could even be an olefin bearing allylic hydrogen atoms.

Experimental Measurements The gas-liquid chromatographic measurements (GLC) were performed with a Carlo Erba Fractovap 2150 instrument equipped with a Hewlett Packard automatic integrator, model 3380 A. Conditions: Carbopack C + 0.19% picric acid; 1.6 m X 2 mm ID X 6 mm OD, glass; column temperature:

70

80 “C; inlet and detector: 150 “C; flow rate: 30 ml min-’ N,; AP (N,): 3.1 kg cmm2;detector: FID; sample size 0.04 ~1. The nuclear magnetic resonance (NMR) spectra were recorded with a WP 200 Bruker instrument operating at 4.69 T in the FT mode. Deuterium field-frequency lock was used for ‘H measurements. All chemical shifts are in ppm from TMS, positive shifts being to low field of the reference. The measurements were performed at room temperature (ca. 24 “C!). The mass spectra after gas-liquid chroma~~aphic separation (GC-MS) were measured with a GC-MS data system (2350-PDP 11/23), located at Servizio GC-MS of Area della Ricerca di Roma. Mass spectrometer: V.G. micromass 7070 F; E. I., 70 eV; emission current, 200 PA. Gas chromatograph: Dani 3900; column: Carbopack C + 0.2% Carbowax, 1.4 m X 1.5 mm ID X 6 mm OD, glass; column temperature: 3 min at 100 “C, then to 170 “C at 2 “C!min-‘; flow rate: 20 ml min-‘, N,; AP (N,): 3.0 kg cm-*; sample size: 1 jll. UV and visible spectrophotomet~c data were recorded on a Perkin Elmer 330 instrument. Materials The olefins (Fluka) l-heptene, tmns-2-heptene, 1-pentene, cis-stilbene and 3,3-dimethyl-l-butene were treated with an aqueous solution of FeSO,+, distilled, dried over KOH and stored under nitrogen in the dark at low temperature. I-Heptene-3d2 [12] and K20sX, 1131 were prepared by known methods. The solvents were dried over 4A molecular sieves. Isomerization runs The diagrams in Fig. 1 were obtained as follows: in a 50 ml two-necked flask equipped with a condenser, 1 ml of l-heptene was added to 10 ml of odichlorobenzene containing K20sC16 (5 mg, - 10F3 M) dissolved upon addition of the dicyclohexyl-l~cro~-6 ether C,&,O, (Fluka); the solution was refluxed in a silicone oil bath (170 “C) under nitrogen; samples for GLC analysis were taken periodically by syringe under nitrogen, The relative catalytic activities of K,OsX, (X = Cl, Br), reported in Fig. 4, were measured on 1 ml C,H,Br solutions containing 1-heptene or trans-2heptene (1.0 M) and variable amounts of the catalyst and the crown ether, sealed in vials under vacuum and heated at 80 “C for 17 days. The exact concentration of OsXz was measured by visible spectrophotometry. The cross-over experiment (isome~zation of l-heptene-3dz in the presence of I-pentene) was carried out in a Berghof 75 ml autoclave in 1.0 M o-dichlorobenzene solution for both the olefins in the presence of K,OsCl, (- 10v3 M) after heating at 150 “C for 12 h. The pen&me isomers were separated by fractional distillation. The isomerization of 1-heptene-3d, was carried out in a sealed vial containing 3 ml of o-dichlorobenzene solution, K,OsCl, ( 10m3M) and the olefin (1.0 M), which was heated at 170 “C for 24 h. The olefin collected after distillation (0.1 ml) was analyzed by GLC and 2H NMR.

71

All other experiments (interaction of K20sC1, with cis-stilbene, cis-stilbene and 3,3-dimethyl-1-butene or l-heptene, 1-heptene and Ccl,) were performed in sealed vials in o-dichlorobenzene solution (1 ml, [ OsClz] = 10e3 M), containing the olefins (1.0 M) or Ccl, (1.0 M), and heated at 160 “C for 12 h. The cis + trans-stilbene transformation was followed spectrophotometrically (UV). The amount of CHCI, formed in the solution of 1-heptene and Ccl4 was determined by GLC, and it was found that [CHCl,] 103/[CC14] = [ CHCl,] /[ OsCl;] = 3.

Acknowledgement We thank Mr. Enzo Brancaleoni

for the GC-MS measurements.

References 1 K. R. Seddon, Coord. Chem. Rev., 41 (1982) 159. (RuCld): J. Halpern, J. F. Harrod and B. R. James, J. Am. Chem. 2 Hydrogenation Sot., 88 (1966) 5150; acetylene hydration (RuClz): J. Halpern, B. R. James and A. L. W. Kemp. J. Am. Chem. Sot., 83 (1961) 4097; Wacker process (PdCl;); addition of alcohols to butadiene (RhCls): H. Kawazura and T. Ohmori, Bull. Chem. Sot. Jpn., 45 (1972) 2213; Hydrogen-deuterium exchange between D20/CDsCOaD and aromatic compounds (PtCl;, IrClz): J. L. Garnett, R. J. Hodges, R. S. Kenyon and M. A. Long, J. Chem. Sot., Perkin ZZ(1979) 885; isomerization of allylic compounds (IrCls, RhCls, IrC1z3, RuC13, PdC12, PtClT, PtClz): Y. Sasson, A. Zoran and J. Blum, J. Mol. Catal., II (1981) 293. 3 Von W. Preetz and H. D. Zerbe, 2. Anorg. Allg. Chem., 479 (1981) 7. 4 A. Flamini,‘A. M. Giuliani and G. Bertoni, Proc. XVth Nat. Symp. Inorganic Chemistry, Bari 1982, B5. 5 E. L. Muetterties, Transition Metal Hydrides, Marcel Dekker, New York, 1971, p. 246. 6 Eight Peak Index of Mass Spectra, Vol. 3, 2nd edn., Mass Spectrometry Data Centre, Reading, 1974, p. 43. 7 H. H. Mantsch, H. Saito and I. C. P. Smith, Prog. Nucl. Magn. Reson. Spectrosc., 11 (1971) 211. 8 S. M. Taylor and J. Halpern, J. Am. Chem. Sot., 81 (1959) 2933. 9 D. Bingham, D. E. Webster and P. B. Wells, J. Chem. Sot., Dalton Trans. (1974) 1514. 10 G. Henrici-Olive and S. Olive, Coordination and Catalysis, in Monographs in Modern Chemistry, Vol. 9, Verlag Chemie, Weinheim, 1977, p. 157. 11 A. Sen and T.-W. Lai, Znorg. Chem., 20 (1981) 4036. 12 J. F. Harrod and A, J. Chalk, J. Am. Chem. Sot., 88 (1966) 3491. 13 A. G. Turner Jr., A. F. Clifford and C. N. Ramachandra Rao, Anal. Chem., 30 (1958) 1709.