Joumai of Molecdar Catalysis. 13 (1961) 237 - 247 0 ZLsevier Sequoie S-A., Lausanne --ted Cu The Netherlands

GERALD DOYLE

Exxon Reseuch and Engineetirzg CQ. PSI. Box 45. Linden. h’J 07036 (U.S.A.)

(Received January 30,2981)

S&s of the FeCo,(CO)& anion promoted with methyl iodide have proven to be effective catalysts for the methanol homologation reaction. Tkese cat&y& are similar but superior to other Co-E homoIogation cafz.aLysts and it is suspected that many similar intermediates are involved. Infrared evidence kas shown that the FeCos cluster does not remain intact during the reaction, but tke nature of the active catalyst species is not known. Monftor- ing the product composition at short reaction times has show-n that acetal- dehyde, which is formed simultaneously with a small amount of methyl acetate, is the primary homologation product. By varying many of ‘he reac- tion parameters, relatively high yields of acetaldehyde are obtained. Ethanol can be produced in reasonable yields under more vigorous conditions than those required for acetidehyde, but under suck conditions side reactions become increasingly important.

Introduction

Process which produce clean fuels or chemicals from synthesis gas (CC and Ha) W-S very Likely become increasingly important in the near future. Tke metkanol-homotogation reaction in which ethanol is formed in two steps fiorn CO and Hz is one such process.

When first described by Wender and coworkers [II the cobalt catalysts which they employed were of reIativeIy low a&vim and resulted in low product s&ctivity. Subsequently, a number of improvements and modifica- tions of the original catalysts have been described. A major advance was the addition of iodine or iodides, which greatly improved the reaction ra+~s [2l. Further improvements were achieved by the use of phosphine additives [3,41 and the use of mixtures of cobalt and ruthenium cataIyst.s [4 - 61 - tithougb each of these additives resulted in improved performance, a catalyst system tkat gave both high product seIectivities and high conversion was not yet avaiIabIe. Recently, however, a cataIyst consisting of cobalt acetyEace~nat.e, a ruthenium compound, a pkospkine and iodine has been

238

reported by workers at Gulf and Union Carbide which is fairly efficient and gives good selectivities to ethanol 171. A very similar catalyst system has also been patented by British Petroleum; however, they report considerably lower yields 183 .

A number of workers have postulated that methanol is first converted tc acetaldehyde which is subsequently hydrogenated to ethanol [9 - 13]_ Although this sequence has genera& = been accepted, recent work by Deluzarche et al. [ 141 suggests that acetaldehyde may not be a precursor. Relatively little work has been carried out aimed at producing acetaldehyde by this process compared to ethanol, although acetaldehyde is also a com- mercizlly attractive product. This present study was directed at the investiga- tion of certain bimetallic catalysts for the selective production of either acetaldehyde or ethanol from methanol. Since more detailed knowledge of the effects of reaction parameters on product selectivities and methanol conversions was necessary, much of our effort was concentrated in these areas as well as in a more detailed study of the homologation reaction sequence. Of pvticulv interest in this study were iron-cobalt complexes, especially those based on the FeCo,(CO);s anion first prepared by Chini and co-workers [ 15]_

Experimental

Reagents The metal csrbonyls were purchased from Strem Chemicals and were

used as received. HFeCo,(CO),, and its salts were prepared by the methods of Chini and coworkers [15]. (C,H~),NCO(CO)~ was prepared by the reaction of K-Selectride@* (Aldrich) with Co,(CO)s followed by the addition of (C,H&NCl. Reagent grade methanol and toluene were used without further purificaticn.

Analysis Gas and liquid products were analyzed by gas chromatography using a

Perkin-Elmer Model 900 or a Hewlett-Packed Model 5840A instrument. Columns packed with Chromosorb 102 or Carbowax 2OM on Gas Chrom Q were used with temperature programming. Peaks were identified by compar- ing to known compounds on two different columns if possible. For peaks which could not be identified in this manner, identification was made by gaschromatography - mass spectroscopy performed by the Analytical ar.d Information Division of Exxon Research and Engineering Co.

Quantitative measurements were made using toluene as an internal standard. Response factors were either determined experimentally 01 were taken from the compilation-of Dietz [16].

* _KFq CH(CH3)C,H,~ 3H

239

The high pressure reactions (27 MPa) were carried out in a I liter stirred autoclave, which was equipped with a cat&y& blowcase and was directly fed by high pressure syn-gas lines. The autoclave was chvged mith solvent (generally 250 ml methanol with 50 ml totuene as an internal standard) and the appropriate amount of methyl iodides and the solution then preheated to the reaction temperature. The catalyst, dissolved in 100 ml methanol, was then introduced through the blowcase, and the pressure immediately bronught to the desired level. Liquid samples were taken at desired intervals during the reaction and a gas sample taken at the conclusion of the reaction.

Infrared spectra Infrared spectra were recorded on a Perkin-Elmer Model 283 IR Spec-

trophotometer using a heated high pressure infrared cell of custom design. The infrared cell was used in conjunction with a 300 ml stirred autocalve equipped as the P hter reactor described above. The reactions were carried out in the standard manner and infrared spectra and/or liquid samples were taken at desired intervals during the reaction.

Results

Cortzparkmz of Fe(CO)s, Coz(CO), and FeCo,(CO)& A series of reactions were run in order to determine any differences in

activity for FeCos(C0)& and Fe(C0)s or Coz(CO)s. The results of these reactions are shown in Table 1. The conditions chosen, as we shah see later,

TABLE 1

hfethanol homologation with Fe-Co ccmplexes

Catalyst Met&sol Conversion

6)

Percent sdectivity toQ

EtlXWl0i Acetidehydeb

CO,(CO)8 9 WC015 2 Co2GO)g f P(C& 13 29 Coz(CO)e + WC015 11 -=efJ~3(COh 2 26 (C&53 ~4.NC~~o3(~Ohz 1 75

30 10

zx

60

52 50 40 47 43 73 10

Reaction conditions: 27 MPa, 220 “C, Hz/CO ratio 60/40, six hotirs reaction time. meth~oi/mebl ratio 220011, metal1 ratio 1!2.

=of liquid product only becetsldehyde plus dimethylacetal

240

were far from optimal but it can be seen that under these con&tons, of the unsubstituted czrbonyls only the -FeCos salt showed any appr&able activity. The cluster hydride, HF~CO~(CO)~~. did prove to be somewhat more active than either of the individual pure carbonyls, or a mixture of both, but not by fu. The addition of triphenylphosphine to Co,(CO)s improved the methanol conversions considerably, as expected, but the selectivity to ethanol remained quite low. The nature of the products is similar to those reported by earlier workers although the relative amounts of each product vary considerably_ In general at these high temperztures the FeCo,(CO),, catalyst produces more ethanol with onIy small amounts of acetidehyde, whereas Co,(CO:ls gives a much higher percentage of acetidehyde and consequently larger amounts of higher molecular weight condensation pro- ducts derived from it.

Effect of reaction time on product distribufion and meHmzol conuersiorr Of all the vtiables available, reaction time might be expected to have

2 large effect on product distribution, especially if the reaction products are produced sequentially or if the major products are produced independently at widely different rates_ If however, the products are formed independently but at similaz rates, then reaction time should have only a minor effect on product distribution. Koermer and Slinkhard 1131 have obtained some data for ungromoted Co catalysts under sinzilar conditions but only at two relativeiy long residence times where only small changes might be expected.

By repeatedly withdrawing samples during typical homologation runs we have been able to observe the variance of product distribution and methanol conversion uerszzs time under a variety of conditions. Data for typical low temperature and high temperature runs are shown in Table 2. It can be readily seen that acetaldehyde is the dominant product at the early stages of the reaction. After quickly reaching a maximum the concen- tration of acetaltiehyde then gradually diminishes. Ethanol does not appear until substantial quantities of acetaldehyde are present, and products such as 2-buten-l-zl arLd n-butanol start to appear even later. Methyl acetate is present from the veti s’& of the reaction. When samples are taken at very short reaction times (less than 15 tin), the only products detected are acetaldehyde, mainly in the form of dimethylacetal, methyl acetate and water; the selectivity to dimethylacet.4 and acetidehyde can be greater than 99% under cerSn conditions. At these short reaction times the meth- anol conversions are quite low. The prospect of maintalning the high selectiv- ity to acetaldehyde umder conditions of high methanol conversions prompted a more de+ailed investigation.

Effect of pressure mrd temperature on tnefhanol corwemioon tmd product &ZbCfiUity

As can be readily seen eon Table 3, methanol conversion increases with increasing temperature up to a maximum temperature, then declines rather rapidIy. The optimum temperature for highest conversions is about

Xesideace Temp. Methanol time (‘C) conversion

(h) (8)

Product seEectis+#

(S)

Acetaldehyde Ethanol Dimethyt- Methyl CS+ aceta! Acetate

0.25 170 13 34 0.5 170 22 49 1 170 33 60 2 170 52 62 3 170 62 60 0.25 220 3 90 0.5 220 10 58 1 220 22 44 2 220 37 32 3 220 57 18 6 220 73 6

_ 56 _ 38

1 25 1 12 2 il

b .-

33 b

44 b

54 b

63 b

70 b

9 _ 9 - 9 _

9 1 9 2 9 -

9 - 10 -

9 1 8 2 ; 7

Reaction conditions: 27 MPa; CO/H2 ratio, 50/50; metal/iodide ratio, l/4; metail methanol ratio, l/2200. “liq?rid products only bincluded with ace’dldehyde

TABLE 3

Dependerxe of methanol conversion and product selectivity on tempera- tcre and pressure vzith (C,+H,),NF~CO,(CO)~Z catalyst

Product selectivity”

(%)

Temp. Pressure MeOH Acetaidehydeb Ethanol Methyl

(“C) (Wa) Conversion -4cetate

(“-)

140 26.7 23 82 1 15 160 26.7 47 80 1 13 170 26.7 55 79 2 12 180 26.7 64 82 3 LO 190 26.7 68 81 5 8 200 26.7 61 80 9 6 220 26.7 55 69 26 4

150 6.7 12 70 _ 30 150 10.0 16 75 1 20 150 13.3 20 75 1 18 150 16.7 25 75 1 15 150 20.3 30 78 1 15 150 23.3 36 79 1 15 150 26.7 39 81 2 15

Reaction conditions: CO/H2 ratio, 50/50; met&/iodide ratio, l/4; metai/methanaI ratio. l/2200; 2.5 h reaction time.

=Liquid products only; b Total acetaldehyde plus dimethylaceti

242

180 “C at 27 MPa but decreases with lower pressure. The product selectivity also varies considerably with temperature, in general acetaldehyde predom- mating at lower temperatures while ethanol and to some extent acetaldehyde condensation products become an important factor at high temperatures. The relative amounts of methyl acetate also decrease at higher temperatures.

Operating pressures for the homologation reaction play an important role in the rate of product formation and to 2 lesser extent in product distri- bution_ There is a direct correlation between pressure and methanol con- version at 2 given time. In addition, it has been found that high pa&al CO pressures are necessary in order to operate at reasonably high temperatures. Thus the reaction can reasonably be carried out at up to 220 “C at 27 MPa, whereas extensive catalyst decomposition (and consequently low conver- sions) take place at lower pressures.

Catalyst concenlrxztiot2 and composition A range of catalyst concentrations from 1 X lop3 to 20 X 10m3 molar

were studied for the homologation reaction_ This range is considerably lower

than that used in most previous studies [ 1, 10,121. Very little effect was found on methanol conversions and product selectivity at catalyst concen- tration above 5 X 10e3 molar (see Table 4). Below that level, however, a dramatic decrease in both conversions and selectivity to acetidehyde are noted. The ratio of iodide promoter to me’& cat&ys+, shows a similar but

TABLE 4

Effect of catalyst concentration on methanol conversion and product selectivity

Product se&tivitya

(Z)

[Bu,NFeCo3(C0)12] CH$/knetal Methanol Total Ethanol Methyl Ccc (mmol) ratio conversion ace’taldehyde acetate

(6)

1 4 15 50 - 47 _

2 4 26 65 _ 30 1

3 4 50 73 1 16 3 c

1; 4 4 62 57 73 74 2 2 13 12 4 5

15 4 66 74 3 10 6

23 4 69 73 3 10 6

2.5 1 35 57 2 9 2

2.5 2 48 70 2 10 3

2.5 50 79 2 12 3

2.5

:

53 78 2 10 3

2.5 a 56 77 2 9 4

2.5 12 65 76 2 10 3

2.5 16 20 73 2 9 3

Reaction conditions: 27 _&Pa; CC)/Hz ratio, 50/50; 180 “C: 1 h reaction time. =Liquid products only

243

less dramatic effect on lowering the iodine-metal ratios. The form in which the iodine is added appears to be of some importance with regard to meth- ancl conversions, although it has no effect at ah in the selectivity to acetal- dehyde. As can be seen in Table 5, both CH,I and HI appear to be equivalent but iodide s&s result in considerably lower conversion.

A comparison of the activity of various salts of the FeCo,(C0),2 anion was made, and the results are shown in Table 6. Also included for compar- ison are a mixture of Fe(CO)S and Co2(CO)s and the same combination with (C,I&),NI added. Under the conditions chosen, all the salts are roughly equivalent and superior to the Fe(CO)&o,(CO)s mixture, although by a smaher factor than observed at higher temperatures. fnteresth@y, the addition of (C,H,),NI to the mixed carbonyls results in a catalyst that very nearly abproaches the activity of the FeCos(CO)ya salts.

Discus.sioo

One of the factors affecting the potential utility of the methanol ho- mologation reaction is its lack of selectivity to a single product. The major by-products observed can be acccunted for by a series of simple reaction steps (Scheme 1). In the production of acetaldehyde, the by-products of greatest concern are methyl acetate, ethanol and the higher boiling Ci compounds formed by condensation reactions from acetaldehyde”.

TABLE 5

Alternate iodine sources used with (C&)eNFeC03(C0)12 catalyst

Product selectivity (W)

Iodine Methanol Source conversion

(S)

Total Ethanol Metinyl acetate acetaldehyde

CH31_ 48 87 3 8 HL 44 83 5 8 LiI 26 85 3 9 NaI 31 86 3 8 Bu4Nf 23 87 3 10

2eaction conditions: 27 ME%; CO/He ratio 50/50; metal/iodide ratio l/S; mekd/methanol ratio l/2200; 160 “C; 1 h reaction time.

*Very little attention has been paid to the kss volatile by-products of the homotogation reaction in the Literature [9]_ By careful GC-MS analysis we have found a number of com- ponents which can aR be derived &om an initial condensatiou reactioa of aceteldehyde. The major constituents of this fraction are typically n-butanot, butanal, 2-but-en-I4 and L-mefhoxy-2-bzutene, but a number of other compounds are also present.

244

TABLE6

Methanol conversion for various cztalysts

Catalyst Temp. Residence Methanol

W) time conversion

(mh) (5)

(CsH5)3?NP(C6H5)3FeCo3(CO)12 180 60 38 :CzHS)rNFeCos;(CO)12 180 60 39

(‘%H5)&~Fd=o3W0)12 180 60 42 CSFeCO3(CO)lp 180 60 37 (C4HQ)4NFeCo3(C3)12 180 60 40 Co2(CO)s + Fe(CO), 180 60 I5

Co2(COh +FtiCO)5+(C4Hg).tNI 180 60 37

(C,Hg),NCo(C0)4 1so 60 30

(CaHgkNF~o3(Co),z 220 150 60

Coa(CO)e 220 150 18

Cozy +Fe(C0)5 220 150 i6 CO~(CO)~ + (CsHg)4NI 220 150 48

Reaction conditions: 27 MPa; CO/H2 ratio, 50/50; me’bl/iodide ratio, l/4; total metal (Fe + Co)conc. = constant= 1/2200methanolconc.

FeCo3 (CO),, + 5C0 -+ Fe(C0)5 + Co2(CO)s + COG-

3C02(CG), I 12CH,OH -, 2Co(CH,OH),= + 4Co(CO),- + SC0

COAX + HZ + 2HCo(CO),

CHBI + Co(CO), + CH,Co(CO), -+

. CH,Co(CO), 7 CO 2 CH COCo(CO:- 3 4

CHCHO-

3 1

CH,CHO + COAX

CH,COOCH3 + HCO(CO)~

CH,CH( OCH,), +

CHsCH20H

CH,CH= CHCHO

H2O

H2 - CH3CH2CH2CH20H

Reaction scheme 1: Forknation of major products and catalyst species in FeCoJ-cat.alyzed meASxnol homologation.

Since both acetaldehyde and methyl acetate are formed from a common intermediate some of the ester will always be present. In the absence of hydrogen, methyl acetate is formed in very high sek&ivities under typical

homologation reaction conditions. Secondary products such as ethanol and condensation products can virtually be elimhtated, but this usuahy results in lower methanol conversions_

Choosing the optimum conditions for product yield inevitably involves a compromise, since some process variables are interre&ed and can often affect conversion and selectivity in opposite ways. Nevertheless if the reaction is carried out at relatively high pressures, moderate temperatures and with sufficient catalyst for short periods, it is possible to obtain aceti- dehyde in 80% selectivity at methanol conversions of 75%. Under such con- ditions only small amounts of ethanol end its derivatives are fo?ea and only traces of methane are detected.

High ethanol yields are more difficult to achieve with these caralyst.~. For optimum yields of ethanol, long reaction times at high temperatures are required, but this results in some catalyst decomposition along with relatively large amounts of methane and high boil&g condensation produets derived from acetaldehyde. Catalysts employing ruthenium in add&i& to cob& are preferred for ethanol production, presumably due to their ability to readily hydrogenate acetaldehyde thereby eliminating any of its undesired by-product derivatives.

In order to identify the major metal-contaming species present in solution, and therefore perhaps identify important reaction intermediates, an infrared study of this reaction was made under operating conditions. We analyzed solutions in which (C&&NFeCo3(C0)22 and also a mixture of Fe(CO)s and CO~(CO)~ were used as catalysts, to gain data which would allow us to explain the observed differences in catalytic activity.

Unhke Ccs(CO)s which disproportionates in methanol to form Co(CH,OH),2 + [Co(CO), ] 2, s&s of the FeCos(CO);Z anion show no signs of reaction with this solvent. When a methanol solution of (C&),NFeCo,-

(CO),,, which has characteristic infrared bands at 2008, 1971, 1923 and 1828 cm-‘, is introduced into an autoclave under reaction conditions, new species are very rapidly formed which have infrared bands at 2050, 2010, 199O,f985 and 1880 cm-‘. This is simiIar to, but not identical with, the spectrum observed when a mixture of Fe(CO)s and Co,(CO)s are introduced under the seme conditions (see Fig. I}. Based on these spectra it appears that in both cases the major species present arc\ Fe(CO),, HCo(CO), and some salt of the Co(CO)z anion. AMrough metal acyI species have been observed in some Rhcatalyzed methanol carbonylation reactions, we cannot detect any adyIcob& complexes [17]. This is not entirely unexpected, since it is Likely that an acyl complex would react readily with solvent or HCo(CO), to form methyl acetate or acetaldehyde respectively.

Tht: intensity of the infrared bands decrease with time, eventually resulting in the complete disappearance of alI the bands which can be atti- buted to the initial cobaltcontaining species. The rate at which the bands disappear is noticeabIy slower for solutions derived from (C.&&NFeCos- (CO)I2 than those fo_rmed from the COAX - Fe(CO)s mixture (see Fig.-2). Since the catalytic activity of these sohrtions decreases in a similar manner

f

I F&0~lc0),;

!

~200 7000 ,PCO l,CO

Y. cm-’

Fig. 1. Comparison of the infrared spectra of the FeCo3(C0)12anion and a mixture OF Fe(CO)s and C%(CO)g under reaction coaditions in methanol.

Co2(CO)~ + Fe(CO)5

IO06 IEOO 2200 2000 1qoo 1uca

Y, cm-

Fig. 2. Infrared .spccb of Bu~NF+CO~(CO) 12 and Coa(CO)g + Fe(CO)s in methanol under reaction condition

247

it is tempting to correlate this decrease witk the decrease in the concentra- tion of the cobalt ca=bonyl species, although other factors may also be invoked. The or& obvious difference in the composition of these catalyst solutions whick could account for the greater stability of those derived from

(C4Hq)dNFeCo3(C0)12 is t.ke presence of the ktrabutylammonium c&ion. The deliberate addition of a tefxabuty~ammonium salt to a Co&CO),,- FejCO)s cat&y& solution kas a signiEcant effect on the observed methanol conversions (see Table 6). Suck mixtures have nearIy the same activity as (C.&).&JF~CO&O)~~ and are far more active than mixtures of Co2(CQ)~ and Fe(CO)s done. Improvements in methanol conversion are also observed when a (C,H,),W salt is added to Co2(COjB or if (C&),NCO(CO)~ is employed as the initial catalyst. Similar improvements in product yields or enkancement in rates have been observed in kydroformylation reactions when tetxaalkyIa.mmonium salts are added to COAX catalysts [IS].

Et seems Iikdy that the function of the tetraalkylammonium cations is not to alter the composition of the catalytic species but rather to stabilize, perhaps through ion pairing, aniotic complexes wkick are important in the catiytic cycle. The rote of the Fe(CO)5 or other iron conking species derived from it is not clear. Iron pentacarbonyl is observed at the conclusion of the reaction and no other iron carbonyi ccmplexes could be detected during the reaction. However, tkis does not preclude the possibility that smd amounts of catiytically important species are present under reaction conditions.

References

6

7

8

3

I. Wender, R. 4. Friedel and M. Orchin, Scierrce. ZZ3 (1951) 206. J. Eerty, L. Marko and D. Kallo, Chem. Tech. (Berlin). 8 (1956) 260. 1. Wender. Catal. Rev. Sci. Eng., 14 (1976) 97. G. N. Butter (Commercial Solvents Corp.) U.S. Pat. 3 265 948 (1966). G. Brace, G. Sbrarz, G. Valentini, G. Andrich and G. Gregorio, J_ Am. Ckem. Sot.. 100 (1978) 6238. S. J. Metlin, R. R. Anderson and F. W. Steffgen, a’oshzcts of papers presented at the Z 7th Spring Symposium of the Pittsburgh Catalysis Society, Pittsburgh, PA, 26-28 Aprii, 1978. W. R. Pretzer, T. Kobylbki and J. E. Eozik. (Gulf Research and Development Co.) U.S. Pat. 4 133 966 (1979)_ R. Fiato (Union Carbide Carp_) U.S. Pat. 4 253 987 (X980). British Petrolelrm Lid. British Pat. 2 036 739 (1980). D. W. Siocum in W. H. Jones (ed.), Cataiysb in Organic Chemistry, Academic Press, New York, 1980, p. 245. K. H. Ziesecke, Brenn.sC. Chem., 33 (1952) 385. T. Mizoroki and M. Nakayanra, Bull. Chem Sot. Jprr. 37 (2964) 236. G. Alban&, Chim. Ztzb (&filarz). 55 (1973) 319. G. S. Koermer and W. E_ Sli&ard, fnd. Errg. Cfrem. Prod Res_ Dew.. 17 (1978) 23X_ A. Dekwarche, G_ Jenns, Z. Kiennennaun and F. Abou Samoa, Erdiil K. Kohle, 32 (1979) 436.

15 P. ChM L. ColIi and M. Perddo, Gazz. Chin ItaL, 90 (196C) 1005. 16 W. A_ Deitz, J. G%s. Chrom_ 6. 66 (1967). 17 D. Forster, 6. Am. Chem. Sot., 98 (1976) 846. 18 E&hyT. co-. US. Pat 4 209 643 (1980).