Methanol carbonylation revisited: thirty years on*Peter M. Maitlis," Anthony Haynes," Glenn J. Sunleyb and Mark J. Howardba Department of Chemistry, The University of ShefJield, ShefJield S3 7HF, UKBP Chemicals Ltd., Hull Research and Technology Centre, Salt End, Hull HU12 8DS, UKMonsanto initiated development of its rhodium- and iodide-catalysed process for the carbonylation ofmethanol to acetic acid in 1966. Ownership of the technology was acquired in 1986 by BP Chemicals who havefurther extended it. The work of the Sheffield group, in developing a deeper understanding of the mechanism ofthe process, is reviewed. The rate-determining step in the rhodium-iodide catalysed reaction is the oxidativeaddition of methyl iodide to [Rh(CO),I,] - la: the product from this reaction, the reactive intermediate[MeRh(CO),I,] - 2a has been detected and fully characterised spectroscopically.The rates of the reversiblereactions linking la, 2a and the acetyl complex [(MeCO)Rh(CO)I,]- 3a, as well as activation parameters forseveral of the processes involved, have been measured. The efficiency of methanol carbonylation arisesprimarily from rapid conversion of 2a into 3a, leading to a low standing concentration of 2a, and minimisingside reactions such as methane formation. By contrast, in the iridium-catalysed carbonylation, for whichsimilar cycles can be written, the reaction of [MeIr(C0)213]- 2b with CO to give [(MeCO)Ir(CO),I,]- 4b israte determining. Model studies show that while k,,/k,, is ca. 1 : 150 for the oxidative addition, it is ca.lo5-lo6 : 1 for migratory CO insertion. The migratory insertion for iridium can be substantially accelerated byadding either methanol or a Lewis acid (SnI,); both appear to facilitate substitution of an iodide ligand by CO,resulting in easier methyl migration. The carbonylation of higher alcohols (ROH) has also been successfullymodelled; the corresponding alkyl iodides (RI) react much more slowly than Me1 with [M(CO),I,]-, but againgive the acyls [(RCO)M(CO)I,]- for M = Rh and the alkyls [RM(CO),I,]- for M = Ir. The greater stabilityof [MeIr(CO),I,]- compared with [MeRh(CO),I,] - accounts for the very different characters of the reactionscatalysed by the two metals. It is suggested that the broad features of the Rh/Ir reactivities can be rationalisedsince the M-C bond to a 5d metal is generally stronger than that to the corresponding 4d metal; thus if metal-ligand bond making plays a key role in a step, then the 5d metal is more likely to react faster (e.g.in theoxidative addition), but if a metal-ligand bond-weakening or -breaking step plays a key role in a process(e.g. in the migration), it is likely that the 4d metal will be faster.The rhodium- and iodide-catalysed carbonylation of methanolto acetic.acid is probably the most successful example of anindustrial process homogeneously catalysed by a metal complexin solution that has yet been realised. A recent survey estimatedthat cu. 60% of the current world production of acetic acid (ca.5.5 million tonnes per year) was made in this way, and virtuallyall new plants for acetic acid production now use thistechnology.The older alternative processes include the oxidation offermentation ethanol (still used to make vinegar), acetaldehydeoxidation, followed later by oxidation of butane or naphtha.Significant cost advantages result from the use of carbonmonoxide (derived from natural gas) as a feedstock and fromthe higher selectivity of methanol carbonylation processes.Thefirst of these was commercialised in 1960 by BASF. It usedan iodide-promoted cobalt catalyst but required very highpressures (600 atm) as well as high temperatures (230 "C), andgave ca. 90% selectivity (Table l).,The next major advance occurred in 1966 with the discoveryby Monsanto of the rhodium- and iodide-catalysed reaction,which led to the start-up of their first commercial productionunit in 1970., The advantages over the cobalt-catalysed processare that significantly milder conditions (3&60 atm pressure and150-200 "C) are employed, giving substantial savings inconstruction costs and hence in capital expenditure.Theselectivity is also much better, leading to easier purification, andmaking further savings bn both running and capital costs. Oneproblem is that the process uses rhodium, a rare and veryexpensive metal, but a more significant disadvantage of allthese carbonylation processes is that the reaction medium(acetic acid-HI at > 150 "C) is very corrosive, requiring the useof high cost, exotic materials for plant construction.The reaction, which was then investigated by Forster and hisco-workers at Mon~anto,~ represents one of the triumphs ofmodern organometallic chemistry.It is a classic example of ahomogeneous catalytic process, made up of some six separatestoichiometric reactions, which link to form the cycle shown in* Non-S/ unit employed: atm = 101 325 Pa.Fig. 1methanol to acetic acidCycle for the rhodium- and iodide-catalysed carbonylation ofJ. Chem. SOC., Dalton Trans., 1996, Pages 2187-2196 218Table 1 Comparison of reaction conditions and selectivity for commercial acetic acid processesSelectivityplatm Process (% C per mol) T/*CNaphtha oxidation (BP) 65-70 185 48Methanol carbonylation,cobalt catalyst (BASF) 90 230 600Methanol carbonylation, rhodiumMethanol-methyl acetate carbonylation,catalyst (Monsanto) > 99 150-200 30-60rhodium catalyst (BP) High 150-200 30-50The first step (reaction of methanol with HI to give methyliodide) and the last (the reaction of acetyl iodide with water togive acetic acid and regenerate HI) are purely organic.Infraredspectroscopic studies in situ and under pressure showed thatthe major rhodium species present under catalytic conditionswas [Rh(CO),I,] - la; kinetic measurements also showed thatthe rate of the overall reaction was first order in both [Rh]and [Me11 but zero order in [CO] and [MeOH]. The rate-determining step of the cycle was therefore suggested to be theoxidative addition of Me1 to la, to give [MeRh(CO),I,] - 2a.The model studies of Forster and co-workers4 showed thatin practice, the 'first' detectable product is the isomeric acetylcomplex, [(MeCO)Rh(CO)I,] - 3a,* resulting from methylmigration.This monocarbonyl is then carbonylated to the six-co-ordinate dicarbonyl, [(MeCO)Rh(CO),I,] - 4a, that reduc-tively eliminates acetyl iodide and regenerates [Rh(CO),I,] -la, which starts the whole cycle again.In 1986 ownership of the Monsanto technology was acquiredby BP Chemicals who have further developed the process andhave licensed it around the world. BP have also extended thetechnology to the co-production of acetic anhydride and aceticacid, by carbonylation of methyl acetate and methanol, in theirA$ plant at Hull in England, as has been described elsewhere.A related catalytic system is used by Eastman to manufactureacetic anhydride.6Sheffield StudiesThis paper celebrates 30 years since the initial development ofthe process and discusses work carried out at the University ofSheffield in conjunction with BP Chemicals on elucidatingfurther the mechanistic details of the catalytic cycle.Sincecatalytic reactions are, by definition, very fast, intermediates arerarely present in sufficient amounts to be readily detectable.Our approach has therefore been to separate the cycle into itscomponent reactions and to study them individually, bymeasuring their rates and the effects of changes in conditions.Since the species involved are metal carbonyls which exhibitstrong, characteristic infrared absorptions between ca.2 150 and1600 cm-', a key tool for this work has been FTIR (Fourier-transform infrared) spectroscopy. ,Some of the measurements could be carried out underconditions close to ambient, but many required highertemperatures and pressures, for which we used a cylindricalinternal reflectance (CIR) infrared cell (Fig. 2). The CIR cell(Spectra-Tech), originally developed by W. R. Moser,' com-prises a modified Parr reactor fitted with a CIR crystal (asilicon rod with polished conical ends) through which the IRbeam from an FTIR spectrometer is transmitted. Internalreflections from the rod-solution interface lead to a modulationof the IR intensity by solvent and solute absorptions. Theresulting IR spectrum is similar to that obtained from a* The X-ray structure determination shows the acetyl complex tobe dinuclear in the solid state with six-co-ordinate rhodium(rrr) andbridging iodides.Fig.2 Schematic diagram of the CIR cell used for infraredspectroscopic measurements at high temperature and pressureconventional transmission cell but with a very short effectivepathlength (ca. 10 pm). The cell is relatively straightforward tooperate, allows the use of strongly absorbing solvents (such aswater or acetic acid), and can be kept well stirred, ensuringhomogeneous solutions.A further essential tool in our armoury has been isotopiclabelling (mainly with 13C), which enables reaction pathways tobe defined in more detail. The introduction of I3C labels alsoallows the use of I3C NMR spectroscopy to follow reactionsand identify intermediates at low concentration.Oxidative Addition of Me1 to [Rh(C0)212] - l aThe initial studies at Sheffield * focused on the reaction of Me1with [Rh(CO),I,]-, the rate-determining step in the cycle.Thereaction was found to obey second-order kinetics (first order in[la] and in [MeI]) and activation parameters were consistentwith the classic SN2 mechanism commonly proposed foraddition of alkyl halides to square planar d8 complexes. Thereaction rate was found to show substantial dependencies onsolvent, added salts and counter ion.Protic solvents such as methanol accelerated the oxidativeaddition compared to aprotic solvents such as CH,Cl, ormethyl acetate. Addition of iodide salts (LiI or Bu4NI) also ledto enhanced rates, an effect which was suggested to be due tothe formation of a highly nucleophilic dianion, [Rh(CO),I,]' - .However, our attempts to detect this species spectroscopicallyhave so far failed; thus we now think that the effect of iodide ison the transition-state rather than the ground-state complex.By contrast, counter ions containing NH groups (e.g.C,,H,,NH, +) inhibited oxidative addition, and spectroscopygave strong evidence for an interaction between the rhodiumcentre of l a and NH, thus reducing the nucleophilicity of the2188 J.Chem. SOC., Dalton Trans., 1996, Pages 2187-219rhodium complex. Evidence was also obtained for the contaction pair [H' 9 =Rh(CO),I,-].The Key Intermediate, [MeRh(CO)tI3] - 2aAll of the initial reactivity studies described in the previoussection relate to measurements on the net reaction, l a +Me1 - 3a.However, although it was well established thatthis was the rate-determining step of the catalytic cycle, directobservation of the key first product of oxidative addition, 2a,has only recently been accomplished, by precise choice ofreaction conditions. l oWhen we followed the reaction of l a with neat Me1 verycarefully by IR spectroscopy, we observed a very weak v(C0)peak at 2104 cm-' (Fig. 3) which was not due to la, 3a orany other known rhodiumxarbonyl-iodide species. This weakabsorption appeared early in the reaction and then decayedin direct proportion to [la], the behaviour expected for anintermediate. Comparison with spectra of the stable[MeIr(CO),I,] - 2b strongly suggested assignment of the 2104cm-' band to the high-frequency v(C0) mode of the cis-dicarbonyl complex, [MeRh(CO),I,] - 2a.Using computersubtraction techniques, the low frequency v(C0) band of 2a,exhibiting identical kinetic behaviour, was located near 2065cm-' (Fig. 4), close to absorptions of l a and 3a. The identity ofthe species responsible for these IR absorptions was confirmedby ' 3C NMR spectroscopy on a sample prepared by dissolvingl a in neat l3CH3I at low temperature. After brief warmingto initiate, followed by recooling to freeze the reaction, therhodium methyl-carbon resonance was observed at 6 -0.65['J(Rh-C) 14.6 and 'J(C-H) 143 Hz] (Fig.5) and the methylprotons at 6 2.08. A similar experiment using 13CO-enriched l aallowed observation of a doublet at 6 175.9 ['J(Rh-C) 59.9 Hz]due to equivalent carbonyl ligands in 2a. The conditions (highconcentrations of the reactants and a non-polar medium) forthese experiments were chosen in the expectation that theywould favour formation of a detectable quantity of theintermediate 2a. The IR data indicate that under steady-stateFig. 3(at 2104 cm-', expanded 100 times) due to the intermediate 2aA series of infrared spectra [v(CO) region] recorded during the reaction of Bu,N[Rh(CO),I,] with neat Me1 at 25 "C. Note the weak bandFig. 4 Infrared spectrum [v(CO) region] of 2a generated by computer subtraction of bands due to la and 3a. The relative intensities are consistentwith a cis-dicarbonyl.Inset are kinetic traces showing the growth of the two absorptions of 2a in the initial stages of the reaction ofBu,N[Rh(CO),I,] with neat Me1 at 5 "C. Note that both bands appear at the same rate, indicating that they belong to the same speciesJ, Chem. SOC., Dalton Trans., 1996, Pages 2187-2196 218Fig. 5 Low-temperature I3C NMR spectra ( -reaction of Bu,N[Rh(CO),I,] with Me1 usingdoublets due to lo3Rh-l 3C coupling- 50 "C) illustrating detection of signals due to the rhodium-methyl intermediate, 2a, during the(a) 99% 3C-enriched Me1 and (b) ca. 70% "CO-enriched rhodium complex. All the signals areconditions in neat Me1 the concentration of 2a is ca. 1% of thatof la; this is also in agreement with estimates based on NMRintensities.The IR spectrum of 2a indicated that a cis-dicarbonylrhodium(Ir1) species was present, while the NMR data showedthat the two carbonyl ligands were equivalent; both the IR andthe NMR results were consistent with the fac,cis structureshown for 2a in Fig.I ; the same geometry was also found by acrystal structure determination for the n-hexyliridium complex(Fig. 6 , and see below).The detection of 2a allowed the determination of reactivitydata by monitoring its kinetic behaviour using FTIRspectroscopy. At high Me1 concentrations, pseudo-first-orderconditions apply and application of the steady-state approxim-ation to intermediate 2a [equation (l)] leads to an expression forthe observed rate constant for conversion of l a into 3a[equation (2)].Since the relationship between the concen-trations of la and 2a is given by equation (3), k, can be expressed(3)P a l - k, [Me11[la1 k 1 + k2 Fig. 6 Crystal structure showing the fac,cis geometry of theoctahedral iridium anion in AsPh,[(n-C,H 3)Ir(CO)213]. The tetra-phenylarsonium cation has been removed for clarity. Note the as in equation (4). The rate constant kobs can be measuredhiffeience in Ir-I bond lengths, indicating the high trans influence of(4) the alkyl group(kP2). Forster had already established4 that on heating com-plex 3a regenerated l a and methyl iodide, thus indicating thereversibility of both the oxidative addition and the methylmigration processes. Using specifically labelled [(Me' ,CO)-Rh(CO)I,]-, in the presence of an excess of Me1 to suppressreductive elimination and isolate the migration process, wewere able to measure the first-order rate constant (i.e.k - 2)directly and the ratio [la] : [2a] can be estimated from relativeIR intensities, based on a reasonable assumption concerningrelative molar absorption coefficients. Thus the migratoryinsertion rate constant, k2, is accessible.In addition to measuring the rate of transformation of 2a into3a (k2) we were able to measure the rate of the reverse reaction2190 J. Chem. SOC., Dalton Trans., 1996, Pages 2187-219Fig. 7 Free energy profile constructed from kinetic data for interconversions of complexes la, 2a and 3a at 35 "C in CH2C12-MeI. Equilibriumconstants K , and K , calculated from the forward and reverse rate constants, are also given.Note that the rhodium-methyl intermediate, 2a, isunstable with respect to both reductive elimination and methyl migrationfor exchange of label between acetyl and terminal carbonylpositions. Comparison with the rate of the net reaction,3a - la + MeI, in the absence of Me1 allowed the estima-tion of k , / k (ca. 9 at 35 OC).* This showed that mi-gratory insertion in 2a is an order of magnitude faster thanreductive elimination, and that the low steady-state con-centration of 2a during oxidative addition is primarily dueto the rapid migratory insertion, and not to a fast back-reaction.Complete rate data for the interconversions shown inequdtion ( I ) were thus obtained at 35 "C (Fig.7), which allowedthe calculation of the equilibrium constants K , and K,. Theseshow that 2a is unstable with respect to both migratoryinsertion and reductive elimination.Activation parameters measured for the k , and k - , stepshave allowed estimation of the thermodynamic data, A P - 37kJ mol and ASo -54 J mol-' K-' for isomerisation of 2a to3a. The data predict that K , decreases markedly at highertemperatures and is ca. 30 in the range where the carbonylationreaction is operated. Thus 3a becomes less favoured withrespect to 2a at higher- temperatures; however, the trapping of3a by CO to give the dicarbonyl, 4a, must be very efficient undercatalytic conditions. The results also show that the highselectivity of the rhodium-catalysed process (particularly thelow rate of methane formation) can be explained by the verysmall standing concentration of the methyl species, 2a; this isdue to both the rapid migratory insertion in 3a and theunfavourable thermodynamics of Me1 oxidative addition.Thismay be contrasted with the iridium system (see below).Reductive Elimination from [ (MeCO)Rh(C0)J3] -4aThe monocarbonyl acetyl complex, 3a, decomposes by loss ofmethyl iodide rather than of acetyl iodide; however, addition of* Kinetic data indicate that the reaction of methyl iodide with l a togive 3a proceeds ca. 4 times as fast in methanol as in an aprotic solventunder comparable conditions (25 OC).' Closer analysis of a reaction ofla with methyl iodide showed that complex 2a was still detectable onchange of co-solvent from Mel-CH,C12 (80 : 20 v/v) to MeI-MeOH(80: 20 viv), and that k , was increased by ca.50% and k , by ca. 100%.Thus, while a protic solvent does increase the rates both of the oxid-ative addition and of the migration reaction on rhodium, the effectsseem to be comparatively modest.CO (a reversible process ") leads to the dicarbonyl, 4a, whichreductively eliminates acetyl iodide to regenerate la. Thisfinal organometallic step in the catalytic cycle has receivedless attention, but we find that decomposition of 4a occursquite slowly under ambient conditions (tt = 12 h at 25 "C inCH,CI,). A recent study by the du Pont group l 2 has shownthat reductive elimination of acetyl iodide is reversible[equation (5)l.t[(MeCO)Rh(CO),I,]- [Rh(CO),I,]- + MeCOI ( 5 )The reductive elimination of acetyl iodide from 4a isaccompanied by side reactions arising from the extreme facilitywith which acetyl iodide undergoes hydrolysis by traces ofwater.This gives rise to HI, which reacts with [Rh(CO),I,]-[equation (6)] to give a rhodium(m) tetraiodide complex.Thus although traces of water lead. to difficulties inmeasuring accurate kinetics for reductive elimination, approxi-mate activation parameters (AHS 87 kJ mol-' and ASs -43 Jmol-' K-') have been obtained. In addition, dramatic effects areobserved on addition of other nucleophiles. For example,addition of 1 molar equivalent of tetrabutylammonium acetateto a solution of 4a leads to immediate and quantitativeformation of l a and acetic anhydride [v(CO) 1825 cm-'1[equation (7)].[(MeCO)Rh(CO),I,]- + MeC0,- -[Rh(CO),I,]- + (MeCO),O + I - (7)Secondary amines also react with 4a to generate la, this timewith the formation of the corresponding amide [equation (8)l.it Carbon-13 NMR spectroscopy showed that oxidative addition ofacetyl iodide to l a at low temperature gave cis4a as the initial product.On warming, cis4a isomerised to the more thermodynamically stabletruns4a, which is the isomer produced directly by carbonylation of 3a.We have confirmed the du Pont observations by IR spectroscopy.1 The reactions are rapid at room temperature for dialkylamines (R' =R2 = Et or Bun) but slower for the less nucleophilic N-methylaniline(R' = Ph, R2 = Me).Diphenylamine gives no discernible increase inrate over that measured in the absence of nucleophile.J. Chem. SOC., Dalton Trans., 1996, Pages 2187-2196 219Table 2 Activation parameters for catalytic carbonylation processes and individual reactions from the catalytic cyclesReaction Ref. AMlkJ mol-’ AS’/kJ mol-’Catalytic processesMeOH + CO - MeC0,H (Rh-I-)MeC0,Me + CO - (MeCO),O (Rh-I-)MeOH + CO - MeC0,H (Ir-I-)Model reactionsMe1 + [Rh(CO),I,]- - [MeRh(CO),I,]- (MeOH)Me1 + [Rh(CO),I,]- - [MeRh(CO),I,] - (CH,Cl,)4616881010Me1 + [Ir(CO),I,] - - [MeIr(CO),I,]- (CH,CI,) 17[MeIr(CO),I,] + CO - [(MeCO)Ir(CO),I,] - (PhCl) 18[MeIr(CO),I,] ~ + CO - [(MeCO)Ir(CO),I,]- (PhCl-MeOH) 18Me1 + [Rh(CO),I,]- - [MeRh(CO),I,]- (MeI)[MeRh(CO),I,]- - [(MeCO)Rh(CO)I,] - (MeI)63.660.234605350635415533-116-113- 246- 120- 150- 165- 59-11391- 197Fig.8 Two pathways for the reaction of [(MeCO)Rh(CO),I,]- 4awith amines: (i) the unassisted reductive elimination (k, route) and (ii)direct nucleophilic attack by the amine (kN route). Both routes lead tothe same products, since acetyl iodide will react rapidly with the amine[(MeCO)Rh(CO),I,]- + R’R2NH -[Rh(CO),I,]- + R’R2NCOMe + HI (8)Preliminary kinetic studies on the reaction with N-methyl-aniline suggest parallel first- and second-order pathways(kobs = k , + k,[PhNHMe]), which we interpret as evidencefor direct nucleophilic attack on the rhodium acyl competingwith unassisted reductive elimination (Fig. 8).l 3 Theseobservations suggest that direct attack by external nucleophilesat the bound acyl ligand can play a significant role in thereductive elimination process. l 4Relation of the Individual Steps to the OverallCycleOur approach of dissecting the catalytic cycle into itscomponent steps is validated by a comparison of activationparameters found for oxidative addition of Me1 to l a with thosedetermined by other workers for the overall carbonylationprocess (Table 2 ) . 4 3 6 * 1 5 The parameters for the model reaction inmethanol are particularly close to those for the catalyticreactions, and even though the stoichiometric oxidativeaddition is significantly slower in aprotic solvents the agreementis still reasonable.It is interesting to compare reactivity data for the oxidativeaddition of Me1 to l a with that for the reductive elimination4a - l a + MeCOI.We estimate that in dichloromethanecontaining 1 mol dm-, Me1 the two reactions will have equalrates at ca. 30°C. Above this temperature the oxidativeaddition is rate limiting, whilst below it the reductiveelimination would be the rate-determining step of the cycle.This demonstrates how changes in reaction conditions can leadto dramatic consequences for the rate-determining step of acatalytic process.Since oxidative addition to a rhodium(1) centre ‘controls’catalytic activity under normal operating conditions, it shouldbe possible to accelerate the process by making the rhodiumcentre more nucleophilic, for example by the addition ofphosphines or other electron releasing ligands.l 9 It has recentlybeen found that chelating ligands containing both phosphorusand sulfur donors lead to enhanced catalytic activity in batchreactions.,’ The main problem with these modified systems isthe long term stability of the complex and the ligand underprocess conditions.Comparisons with IridiumOther Group VIII metals have been reported to be active formethanol carbonylation, and in their original work theMonsanto group noted the effectiveness of iodide-promotediridium catalyst^.^.^ This has been confirmed by otherworkers,16 and thus it is instructive to compare the kinetics andmechanism of catalysis for the rhodium and iridium systems.Forster’s mechanistic studies of iridium-catalysed methanolcarbonylation showed many similarities to the rhodium system,but with a greater degree of complexity, due to the participationof both neutral and anionic species.2 1 High-pressure IRspectroscopic studies showed that the main species present werethe iridium(Ir1) complexes, [MeIr(CO),I,] - 2b, which is‘active’, and [1r(CO),I4] - . The latter is ‘inactive’ and needs tobe reduced to l b before it can further participate in themethanol carbonylation cycle. A similar inactive species,[Rh(CO),I,] -, forms in the rhodium-catalysed reactions, butthis is more easily reduced and is therefore less troublesome.A simplified version of the ‘anionic’ iridium cycle is shown inFig. 9. The principal difference between rhodium and iridium isa change in the rate-determining step.Oxidative addition ofMe1 to the iridium(1) anion, lb, is fast, the slow step now beingthe carbonylation of 2b to give 4b. In contrast to the rhodiumsystem, where specially chosen conditions were required todetect the rhodium-methyl complex 2a, salts of the iridiumanalogue, 2b are stable and isolable, and there is no tendencyfor spontaneous isomerisation to an acetyl complex.**22 A* The complex 2b shows a ‘J(C-H) of 139 Hz for the methylattached to iridium,’, this agrees well with the value reported for‘J(C-H) of 143 Hz for the methyl attached to rhodium in theintermediate 2a.2192 J. Chem. SOC., Dalton Trans., 1996, Pages 2187-219crystal-structure determination ' of the n-hexyliridium ana-logue [(n-C,H, 3)Ir(CO)213] - (from oxidative addition of n-iodohexane to lb) shows it to have the fac,cis octahedralgeometry (Fig.6). Kinetic measurements for the reaction ofMe1 with l b gave the activation parameters (Table 2), anddirect comparison of rate data shows that oxidative addition ofmethyl iodide is 120 times faster for iridium than for rhodium inCH,Cl, at 25 0C.24By contrast to the rapid methyl migration in the rhodiumsystem, Forster found that 2b reacted only slowly with CO inchlorobenzene at 80 "C [equation (9)l;" the reaction was alsoinhibited by added iodide.[MeIr(CO)213]- + CO - [(MeCO)Ir(CO),I,]- (9)We have confirmed Forster's observations, and haveobtained kinetic data for the carbonylation of 2b using the high-pressure CIR cell (typical spectra are shown in Fig.10). Theactivation parameters (Table 2), derived from rates measuredin chlorobenzene over the temperature range 80-122 OC,* wereindependent of pco above a threshold of ca. 3.5 atm. Acomparison between the metals can again be made: using theactivation parameters for rhodium (k,, see above) to predict arate, under conditions comparable to those used for the iridiummeasurements, indicates that migration of methyl ontocarbonyl is a factor of ca. lo5 faster for rhodium(Ir1) than foriridium(ii1) at 100 "C in an aprotic solvent. The slow rate ofcarbonylation of [Me1r(CO),I3]- may also account for onepotential problem that has been identified with iridiumcatalysts, namely that methane by-product formation is moresignificant.Improvements in both the activity and selectivity of aniridium-catalysed process are therefore dependent on enhancingthe rate of carbonylation of [MeIr(CO),I,] - .We have recentlyfound that this reaction is dramatically accelerated on additionof small amounts of methanol, and to a somewhat smallerextent by higher alcohols (Fig. 11). l 8 Further, kineticmeasurements in a chlorobenzene-methanol (3 : 1) solventsystem gave quite different activation parameters from thosein neat chlorobenzene, with a much lower AHt and a largenegative AS: (Table 2). These data suggest a change in the* For example. at 1 1 5 OC, 5 atm CO; kobs = ca. 2 x s-'nature of the rate-determining step of the migration reactionfrom dissociative in neat chlorobenzene to associative inchlorobenzene-methanol. Our data for the model reaction aregratifyingly similar to those for the overall catalytic reactionin acetophenone-methanol, calculated from the rate datareported by Matsumoto.l6Addition of ionic iodide resulted in a decrease in rate, aspreviously reported by Forster." These data are all consistentwith the mechanism shown in Fig. 12, involving primarysubstitution of I- by CO. We suggest that the iodide ligandtrans to methyl is the most labile, and we note that in the crystalstructure of the n-hexyl analogue (Fig. 6) the equivalent iodide(trans to hexyl) has the longest bond to iridium.Thus, as for rhodium, but in a different context, methanoloffers very large rate enhancements.In each case these appearFig. 9 A simplified catalytic cycle for carbonylation processesinvolving anionic iridiumxarbonyl-iodide complexes. The organicreactions of Me1 and MeCOI (shown in Fig. 1) are omitted for clarityFig. 10the CIR cellA series of IR spectra illustrating the reaction of [MeIr(CO),I,]- 2b with carbon monoxide in chlorobenzene at 93 "C and 6 atm CO usingJ. Chem. SOC., Dalton Trans., 1996, Pages 2187-2196 219Added solventFig. 11 The relative degrees of acceleration on the rate ofcarbonylation (5.7 atm CO) of [MeIr(CO),I,]- 2b effected bymethanol and higher alcohols (25% v/v) in chlorobenzene at 33 "Cto be rather specific solvation effects; other authors have madesimilar observation^.^^ It is also of interest that an excess ofionic iodide has opposite effects on the rate-determining stepsfor the rhodium and iridium processes.For rhodium, iodidehas a beneficial effect on the oxidative addition reaction,whereas for iridium, added iodide inhibits the crucial migratoryinsertion step.In view of the dramatic accelerations of migratory inser-tions induced by Lewis acids that Shriver and his collaboratorshave reported,26 we also tested a range of metal iodides."Significant acceleration of the migratory insertion was obtainedfor SnI,, where we found that addition of 2 molar equivalentsgave a rate enhancement of ca. 200, at 93 "C and 6 atm CO.Spectroscopic data suggested formation of an intermediatespecies, [MeIr(CO),I,(SnI,)] -, which is activated for COinsertion.Extrapolation of data for the methanol-promotedreaction in chlorobenzene indicates a very similar rate enhance-ment at 93 OC, making the two promoters comparable. Thus,activation of an Ir-I bond in [MeIr(CO),I,]- by methanol orby SnI, in the presence of CO leads to dramatic accelerationof migratory insertion.Key Steps in Higher Alcohol CarbonylationsRhodium- and iridium-iodide catalysts also promote thecarbonylation of higher alcohols to linear or branched chaincarboxylic acids,27- 29 for example, ethanol is converted intopropanoic acid, and n-propanol yields a mixture of n- and iso-butanoic acids. Key steps in these catalytic cycles have also beeninvestigated by our group.The relative rates for the rhodium-catalysed carbonylationswere found to decrease sharply in the order, methanol % eth-anol > propanol.Although [Rh(CO),I,]- was very much lessreactive towards the higher alkyl iodides than towards methyliodide,' other key features of the stoichiometric reactions weresimilar: they gave the acyls [(RCO)Rh(CO)I,]- as stableproducts [equation( lo)],* and the rates were first order in [la]and [RI].RI + [Rh(CO),I,]- - [(RCO)Rh(CO)I,]- (10)The mechanism of the alkyl isomerisation step in therhodium reactions was investigated. The oxidative additionsproceeded smoothly and cleanly at temperatures up to 80 OCwith no evidence for alkyl group isomerisation. Further* We suggest that the acyl complexes have the same dimeric structuresin the solid state as was found for [{(MeCO)Rh(CO)I,},]2-.5Fig.12the carbonylation of complex 2bProposed mechanism for acceleratory effect of methanol onFig. 13 Proposed mechanism for alkyl group isomerisation and alkeneexchange reactions between rhodium(1rr) acyl complexes. The isomeris-ations lead to isomeric butanoic acids in the catalytic carbonylationof propanolexperiments showed that the isomerisation observed in thecarbonylation of the higher alcohols occurs in the acylcomplexes and a mechanism has been suggested (Fig. 13).Recent work by the du Pont group has shown that[(EtCO)Rh(CO)I,] - can also be generated by the reactionof [Rh(CO),I,]- with HI and ethylene, the hydride[HRh(CO),I,]- being detected as an intermediate at lowtemperature.,' They suggested that this route may be favoured2194 J.Chem. SOC., Dalton Trans., 1996, Pages 2187-219in the catalytic carbonylation of ethanol and of the higher such analyses give useful information about the step in the cyclealcohols. However, our activation parameters for oxidativeaddition of higher alkyl iodides are very similar to those for thecorresponding alcohol carbonylations, suggesting that a routeuiu ethylene may be of minor significance.Addition of alkyl iodides to [Ir(CO),I,]- was much fasterthan for the rhodium analogue, and proceeded under mildconditions (30 "C, 16 h, N, atmosphere) to give the alkylcomplexes, [RIr(CO),IJ ~. As for rhodium, reactions withhigher alkyl iodides were much slower than with methyl iodide;again the reactions were first order in both [lb] and [RI].*Comparison of oxidative addition kinetics for Rh and Ir gavethe relative rates: k,,/k,, = 150 for MeI, 220 for EtI and 140for Pr"1 (80 "C).The similarity in refutiue reactivity isunderstandable in view of the nearly identical radii of rhodiumand iridium which lead to similar steric effects. For both metalsthe largest difference in reaction rate is observed between Me1and EtI. with much smaller changes on further increases in thealkyl chain length. Based on a rate arbitrarily chosen as 1000 forMel, those for EtI are 3 for Rh and 2.3 for Ir. These valuesfall into the range reported for other transition-metal basednucleophilcs,3' where Me1 is a factor of 102-103 more reactivethan EtI.Non-transition-metal nucleophiles, such as iodideanion 3 2 and organic amines 3 3 also exhibit a large step changein reactivity between Me1 (1000) and EtI (33), although theeffect is somewhat less pronounced. The similarity in reactivitytrends for all these reactions supports the proposal thatoxidative addition to both Rh and Ir proceeds by a classic S,2process at carbon.34 (Recent isotope effect measurementscombined with ub initio calculations confirm this conclusion.)which determines the overall rate of the catalysis. There are,of course, some quantitative differences, which arise from theeffects of different temperatures and different media in the twosets of experiments. However, the approach is clearly validatedand one can now analyse other systems with reasonableconfidence.The key finding is that while [Ir(CO),I,]- oxidatively addsorganic iodides cu.150 times faster than [Rh(CO),I,]-,migratory insertion on rhodium(rI1) is 10s-106 faster than oniridium(iI1). One consequence of this is that whereas the alkyl-iridium(IIr)-carbonyI-iodide complexes are reasonably stable,the corresponding rhodium alkyls are very unstable and rapidlyundergo migratory insertion to give the acyls. The increasedstability of the alkyl-iridium complexes is probably amanifestation of the well known greater stability of the higheroxidation states for the 5d metals. Since the M-C bonds are inthe same environments in the rhodium and the iridiumcomplexes, the easier migration on rhodium may be aconsequence of the relative strengths of the M-C bonds.Although no data are available for rhodium and iridium, bondsto 5d elements are generally stronger than those to thecorresponding 4d Theoretical studies suggest that suchdifferences arise principally from relativistic effects for theheavier metals.36One can try to generalise the information available fromthese data.Thus, if a metal-ligand bond-making step playsa key role, then it is more likely that k(,,, > k(4d); this ispresumably what happens in the oxidative addition. Bycontrast, if a metal--ligand bond-weakening or -breaking stepThe greater difference in reactivity between methyl iodide andthe longer chain alkyl iodides exhibited by metal complexes bycomparison with organic nucleophiles, probably arises fromsteric congestion around the metal centre.plays a key role in a process, such as the migration, it is likelythat k(4d) > k(5dl.AcknowledgementsRelevance to Catalytic Alcohol CarbonylationThe relative rates for reaction of alkyl halides with[Rh(CO),IJ (Me1 1000, EtI 3, Pr"I 1.7 and Pr'I 4, allmeasured at, or extrapolated to, 80 "C) correlate reasonablywell with the relative rates for the rhodium-catalysedcarbonylation of the corresponding alcohols [ 1000 (MeOH),48 (EtOH).22 (Pr"0H). 57-180 (Pr'OH) at 170 "C and 35 atmCO]. The principal difference between the two series is thesmaller step change in carbonylation rate between methanoland the higher alcohols, compared with the oxidative additionreactions. This may arise from the different temperatures andthe different reaction media used in the two sets of experiments.The activation parameters obtained for the oxidative additionreactions also resemble those for catalytic carbonylation; theincrease in AHx for higher alkyls compared with methyl issimilar for both catalytic carbonylation and oxidative addition(Table 2).Our data therefore fully support the conclusion thatthe oxidative addition step is also key in the rhodium-iodidecatalysed carbonylation of the higher alcohols.ConclusionWe have shown that catalytic carbonylation reactions can beanalysed into a cyclic series of stoichiometric reactions, and that* Although clean second-order kinetics were observed for the reactionof [lr(C0)21,] with MeI.those with an excess of EtI and Pr"1exhibited more complex kinetics, with substantial deviations frompseudo-first-order behaviour. particularly at higher temperatures in thepresence of light or air. Addition of a radical scavenger restored simplekinetics, suggesting that the complex rate behaviour is due to a radicalprocess which competes with the second-order reaction. We estimatethat the radical component accounts for cu. 20% of the overall reactionof I b with ethyl iodide at 50 "C.It is a great pleasure to thank our colleagues and co-workers,Robb Watt and George Morris (BP Chemicals). Brian Mann,Cathryn Hickey, Tony Fulford, Jean Pearson and Paul Ellis inSheffield, and Mauro Bassetti and Donato Monti in Rome, fortheir help in the work described in this article.We also thankBP Chemicals and the EPSRC (SERC) for support and theEU Human Capital and Mobility programme (Contract no.ERBCHRXCT930147) for additional assistance.References1 M. J. Howard, M. D. Jones, M. S. Roberts and S. A. Taylor, Cutul.Today, 1993, 18, 325.2 R. T. Eby and T. C . Singleton, Applied Industrial Catalysts, ed.B. E. Leach, Academic Press, 1983, vol. I , p. 275.3 F. E. Paulik and J. E. Roth, Chern. Comnrun.. 1968, 1578; J. F. Roth,J. H. Craddock, A. Hershman and F. E. Paulik, C'hcm. Techno/.,1971, 600; K. K. Robinson, A. Hershman, J. H. Craddock andJ. F. Roth, J. Curul., 1972, 27, 389; F. E. Paulik. A. Hershman,W. R. Knox and J. E. Roth, Monsanto Company, US Pot.,3 769 329, 1973.4 T.W. Dekleva and D. Forster, A h . Curd., 1986, 34. 8 I .5 G. W. Adamson, J. J. Daly and D. Forster, J . Orgunornet. Chem.,1974.71. C17.6 S. W. Polichnowski, J . Clretn. Eriltc., 1986, 63, 204; V. 13. Agreda,Cliem. Technol., 1988, 250; J. R. Zoeller. V. H. Agreda, S. L. Cook,N. L. Lafferty, S. W. Polichnowski and D. M. Pond. CutLrl. Tocltrj,1992, 13, 73.7 See, for example, Homogeneous Trunsition Metal CutulysedReactions, eds. W. R. Moser and D. W. Slocum, Advances inChemistry Series 230, p. 3, and refs. therein.8 A. Fulford, C. E. Hickey and P. M. Maitlis, J . Orgnrzomt>t. Chem.,1990,398, 31 1.9 A. Fulford and P. M. Maitlis, J. Organomet. Ciiem.. 1989, 366, C20.10 A. Haynes, B. E. Mann, D. J. Gulliver, G. E. Morris and P.M.Maitlis, J. Am. Chem. SOL.., 1991, 113,8567; A. Haynes, B. E. Mann,G. E. Morris and P. M. Maitlis, J. Am. Chem. Suc., 1993, 115, 4093.1 1 H. Adams, N. A. Bailey, B. E. Mann, C. P. Manuel. C. M. Spencerand A. G. Kent, J . Chem. SOL.., Dalton Truns., 1988. 489.J. Chem. SOC., Dalton Truns., 1996, Pages 2187-2196 21912 L. A. Howe and E. E. Bunel, Polyhedron, 1995,14, 167.13 A. Haynes, D. Monti and P. M. Maitlis, unpublished work.14 D. Forster, J. Am. Chem. Soc., 1976,98, 846.15 J. Hjortkjaer and 0. R. Jensen, Ind. Eng. Chem. Prod. Res. Dev.,1976,15,46; J. Hjortkjaer and V. W. Jensen, Ind. Eng. Chem. Prod.Res. Dev., 1977, 16, 281; J. Mol. Catal., 1978,4, 199.16 D. Brodzki, B. Denise and G. Pannetier, J. Mol. Catal., 1977,2, 149;T.Matsumoto, K. Mori and A. Ozaki, J. Catal., 1978, 51, 96;T. Mizoroki, T. Matsumoto and A. Ozaki, Bull. Chem. SOC. Jpn.,1979,52,479.17 P. R. Ellis, J. M. Pearson, A. Haynes, H. Adams, N. A. Bailey andP. M. Maitlis, Organometallics, 1994, 13, 3215.18 J. M. Pearson, A. Haynes, G. E. Morris, G. J. Sunley andP. M. Maitlis, J. Chem. SOC., Chem. Commun., 1995, 1045.19 M. Schrod and G. Luft, Ind. Eng. Chem. Prod. Res. Dev., 1981,20,649; G. Luft and M. Schrod, J. Mol, Catal., 1983,20, 175; 1983,22,169; E. Lindner and B. Andre, Chem. Ber., 1988, 121,829.20 M. J. Baker, M. F. Giles, A. G. Orpen, M. J. Taylor and R. J. Watt,J. Chem. SOC., Chem. Commun., 1995, 197; J. R. Dilworth, J. R.Miller, N. Wheatley, M. J. Baker and J. G. Sunley, J. Chem. Soc.,Chem. Commun., 1995, 1579.21 D. Forster, J. Chem. SOC., Dalton Trans., 1979, 1639.22 D. Forster, Inorg. Chem., 1972, 11,473.23 J. M. Pearson, Ph.D. Thesis, University of Sheffield, 1994.24 M. Bassetti, D. Monti, A. Haynes, J. M. Pearson, I. A. Stanbridge25 0. Blum and D. Milstein, J. Am. Chem. SOC., 1995,117,4582.26 S. B. Butts, S. H. Straus, E. M. Holt, R. E. Stimson, N. W. Alcockand D. F. Shriver, J. Am. Chem. SOC., 1980, 102, 5093; T. G.and P. M. Maitlis, Gazz. Chim. Ital., 1992, 122, 391.Richmond, F. Basolo and D. F. Shriver, Inorg. Chem., 1982, 21,1272.27 T. W. Dekleva and D. Forster, J. Am. Chem. SOC., 1985, 107, 3565,3568; Abstract from the 187th National Meeting of the AmericanChemical Society, INDE 0108, 1984; J. Mol. Catal., 1985, 33, 269.28 J. Hjortkjaer and J. C. Jorgensen, J. Chem. SOC., Perkin Trans. 2,1978, 763; J. Mol. Catal., 1978,4, 199.29 S. B. Dake, D. S. Kohle and R. V. Chaudhari, J. Mol. Catal., 1984,24, 99; S. B. Dake and R. V. Chaudhari, J. Mol. Catal., 1986, 35,91 19.30 D. C. Roe, R. E. Sheridan and E. E. Bunel, J. Am. Chem. SOC., 1994,116, 1163.3 1 A. J. Hart-Davis and W. A. G. Graham, Inorg. Chem., 1970,9,2658;P. K. Monaghan and R. J. Puddephatt, J. Chem. SOC., Dalton Trans.,1988,595; G. N. Schrauzer and E. Deutsch, J. Am. Chem. SOC., 1968,90, 2441.32 J. Hine, Physical Organic Chemistry, McGraw-Hill, New York,1962, p. 176.33 J. March, Advanced Organic Chemistry, J. Wiley, New York, 4thedn., 1992, p. 339.34 T. R. Griffin, D. B. Cook, A. Haynes, J. M. Pearson, D. Monti andG. E. Morris, J. Am. Chem. SOC., 1996, 118, 3029.35 C. Mancuso and J. Halpern, J. Orgnnomet. Chem., 1992,428, C8.36 T. Ziegler and V. Tschinke, ACS Symp. Ser., 1990,428.Received 18th December 1995; Paper 5/08206D2196 J. Chem. SOC., Dalton Trans., 1996, Pages 2187-219