00:00:00 I'd first like to offer my congratulations to Professor Wender.

00:00:29 receiving the Lubrizol Award this year and also to thank him for his kind

00:00:35 request that we present some of our more recent results in the area of methanol

00:00:40 homologation. The reaction homologation of methanol was originally reported

00:00:46 around 1949 by Professor Wender, Professor Orchin. The reaction to much of

00:00:55 the 50s and 60s received really very little attention and it's only been in

00:00:59 the last five or six years with the Renaissance and interest in single

00:01:04 carbon chemistry that the reaction has again been studied in great detail. Just

00:01:10 as a witness to this from an industrial side we're concerned with patents and

00:01:15 prior to 1977 there were only about six patents addressing the homologation of

00:01:20 methanol to either acetaldehyde or ethanol. At the end of 1981 in that four

00:01:26 year period the number of patents worldwide had grown to be in excess of

00:01:29 70. The reaction itself is attractive for many of the reasons of the interest in

00:01:36 single carbon chemistry. You have methanol, synthesis gas, they can be

00:01:40 produced from a variety of raw materials, petroleum-based, coal-based, residual oil,

00:01:45 etc. You also have the potential of the economic incentive of having a feedstock

00:01:51 particularly when you're producing oxygenated compounds which could have a

00:01:55 significant advantage over a petroleum-based feed such as ethylene or

00:01:59 propylene. We have the first slide.

00:02:09 The reaction we're dealing with is methanol being reacted to carbon

00:02:13 monoxide hydrogen. The initial product formed is acetaldehyde. The oxygen is

00:02:19 rejected as water. This particular reaction is exothermic, has about a

00:02:24 minus 24 kcals per mole. In situ with hydrogen around that acetaldehyde can be

00:02:30 further converted to ethanol. The overall conversion is also very exothermic in

00:02:36 the order of about minus 40 kcals per mole. Now besides producing what we're

00:02:42 after, acetaldehyde or ethanol, the reaction tends to be somewhat messy. It

00:02:47 will produce methyl acetate, a straight carbonylation product, some acetic acid,

00:02:54 generally most of the acetic acid will be present in the ester form of methyl

00:02:59 acetate. This methyl acetate can vary from around 5 mole percent upwards to 30

00:03:05 mole percent of all the methanol which is converted. In addition, one will

00:03:09 produce ethers. The principal ether is dimethyl ether resulting from the

00:03:13 dehydration of the methanol. In addition, one will observe some methyl ethyl ether

00:03:19 and diethyl ether. Principally those ethers are observed when our major

00:03:24 reaction product is ethanol and at times the amount of the diethyl ether can reach

00:03:30 about 10 mole percent. The dimethyl ether is a particularly troublesome

00:03:35 character. You have to perform your experiments carefully, getting a good

00:03:40 accurate mass balance. Being such a volatile ether, missing it can give rise to some

00:03:47 distorted product selectivities. And at times the amount of dimethyl ether formed

00:03:52 can be upwards of 50 mole percent of the methanol which is reacted. Typically it

00:03:57 will be around 5 to 10 percent. In addition, we'll produce higher aldehydes,

00:04:02 alcohols, and esters. The first of these would be the three carbon species such

00:04:07 as propionaldehyde and propanol resulting from the extended homologation

00:04:11 of ethanol. In addition, you can produce C4, C6, and C8 species. These arise

00:04:19 principally from an aldol condensation of your intermediate acid aldehyde and

00:04:23 accordingly they will be represented primarily by N-butyraldehyde, 2-ethyl

00:04:28 butyraldehyde, and the corresponding alcohols. In addition, small amounts of

00:04:33 crotonaldehyde may be observed. Brings us to the hydrocarbons. The principal

00:04:38 hydrocarbon observed in this reaction will be methane. If the order of 3 to 10

00:04:45 mole percent of the methanol converted will go to methane. In addition, one can

00:04:51 observe small amounts of ethane, 1 or 2 mole percent, when ethanol is your major

00:04:56 product. Now the goal of most of the research which has been conducted is to

00:05:01 clean up this reaction, make it selective either to the acid aldehyde or to the

00:05:06 ethanol, and one now tackles that problem by addressing the catalyst. Now

00:05:12 traditionally, cobalt complex has been used and indeed the first catalyst was

00:05:17 reported by Professor Wender in 49 was cobalt carbonyl. There's been some recent

00:05:22 reports in the last several years by workers at Argonne, Fader, and Chen using

00:05:27 an iron-based catalyst in basic media and producing ethanol as their dominant

00:05:33 homologation product. That reaction rejects the oxygen as carbon dioxide

00:05:38 rather than water as in the cobalt systems. Besides cobalt carbonyl, about

00:05:44 any other form of cobalt can be used. Cobalt carboxylates, cobalt acetyl

00:05:49 acetates, cobalt oxide, inorganic sources of cobalt. A complicating factor arises

00:05:57 is that not all these cobalt sources are equal in their ability to promote the

00:06:01 reaction and you'll get some discrepancies or variances in the rate

00:06:06 of conversion as well as the product selectivity. A recent innovation

00:06:10 introduced in about the 1960s was the use of a halogen promoter, principally

00:06:15 iodide, and again the source of the halogen promoter can be varied. It can be

00:06:21 elemental iodine itself, can be sodium iodide, can be alkyl iodide since it's

00:06:27 methyl iodide, ethyl iodide, hydrogen iodide, and again the effect of these

00:06:33 halogen promoters is not equal. Some can promote the reaction better than others.

00:06:37 Some have different effects on the product selectivity. The ligand

00:06:41 modifiers have been used. They generally, as will a major portion of this talk, will

00:06:46 be addressing their effects on product selectivity. They also introduce

00:06:50 considerable catalyst stability. Co-catalysts have also been employed. The

00:06:55 most successful one has perhaps been ruthenium. It functions primarily to

00:07:00 reduce your intermediate product, acetaldehyde, and convert it very rapidly

00:07:04 to ethanol. Reaction conditions are two to three hundred atmospheres at the

00:07:09 reaction temperature of 175 to 225 degrees C. Now we have here a very

00:07:15 complex reaction. We have a variety of products which can be formed. We have

00:07:20 catalysts which can consist of at least four individual components. We can vary

00:07:26 the CO hydrogen ratio. We can vary the reaction time, the reaction temperature,

00:07:30 and quite often the interaction of these variables is a synergistic effect. They

00:07:36 do not always operate in an independent manner. Now for the talk today, we'll try

00:07:41 to simplify things to some extent by only using cobalt acetyl acetonate as

00:07:46 the cobalt source. We will only use iodine as the source of the halogen

00:07:51 promoter. The ligands will vary. We will not use any cold catalyst. We'll hold our

00:07:57 pressure at always 4,000 pounds using a one-to-one blend of CO and hydrogen and

00:08:02 we will vary some of the reaction temperatures. The results shown here are

00:08:08 by no means optimized results. What we're interested in is what happens when we

00:08:13 change one or more of these particular parameters. The next slide please. We

00:08:19 show here the product selectivity, mole percent, using a system in which our

00:08:24 loading of iodine to cobalt is low, less than one. Under these conditions, the

00:08:30 catalyst will generally produce higher amounts of ethanol in the absence of

00:08:35 ligand. We introduce triphenylphosphine. We get some enhancement in the amount of

00:08:40 ethanol, a little bit drop in the amount of acetaldehyde. Now a very startling

00:08:45 change occurs when one introduces triphenylarsine. The cobalt catalyst is

00:08:51 greatly inhibited in its ability to reduce the intermediate acetaldehyde and

00:08:55 we see in fact acetaldehyde is our dominant product. This holds true as well

00:08:59 for the triphenylstibine. We go to triphenylbismuth and it behaves very much

00:09:04 as the system shown on the extreme left in which we have no ligand present at

00:09:09 all. And to a certain degree, this results can be rationalized based on the

00:09:13 relative strengths of these various ligands. The phosphine is a much stronger

00:09:17 donor than the arsine. Your resulting cobalt complex will have a greater

00:09:22 electron density and might be expected to be a better reducing agent. In fact, the

00:09:26 aldehyde is converted rather well to ethanol. In the case of the arsine, we're

00:09:31 very close to the borderline between where we can reduce the aldehyde and

00:09:37 where we can't. And this is illustrated more in the next slide where we take a

00:09:42 family of arsine, starting with triphenylarsine, systematically

00:09:47 replacing the phenyl with an alkyl group until we finally reach at this end

00:09:52 triethylarsine. In doing so, we will increase the ligand strength and we see

00:09:59 a very nice progression as the catalyst becomes more and more selective in

00:10:03 producing ethanol. In fact, at the end here, we see that triethylarsine produces

00:10:08 dominantly ethanol as opposed to acetaldehyde. A similar trend can be

00:10:13 observed for the stibines. Triphenylstibine, as you recall, produces predominantly

00:10:17 aldehyde. In the case of trienbutylstibine, it produces more amounts of

00:10:23 ethanol. The next slide, please. In the case of phosphines, the effect is less

00:10:33 pronounced. Trienbutylphosphine, being a stronger donor than triphenyl, we only

00:10:39 observed a slight enhancement in the selectivity. The amount of aldehyde was

00:10:43 diminished, however. Going to triphenylphosphite, we have here a ligand

00:10:50 which is a reasonably good pi acid. As such, you might expect it would be a poor

00:10:55 reducing agent. Indeed, we do observe less ethanol. We did not see much increase in

00:11:00 the acetaldehyde. And this is because in the presence of this phosphite, under

00:11:05 these particular reaction conditions, that acetaldehyde is converted to the

00:11:09 aldol products, principally n-butyraldehyde and 2-ethylbutanol. Now we

00:11:15 haven't discussed nitrogen ligands. Several we've looked at, such as

00:11:19 trienbutylamine and various substituted pyridines, by and large, totally inhibit

00:11:25 that homologation catalyst and the conversions are very, very low. Now the

00:11:29 next slide should show us the effects of some sulfur compounds and we have a

00:11:36 representative group of among those we've examined. The pentafluorothiophenol,

00:11:41 pentafluorothiophenol, diphenyl sulfide, and the tetraethyl diphosphine disulfide.

00:11:48 The key feature to note here is to compare them with the systems in which

00:11:52 there is no ligand present. In those systems, these condensate products,

00:11:57 butyraldehyde, 2-ethylbutanol, etc., are very low. When we introduce the sulfur

00:12:03 ligands, they have a very strong promotional effect of that aldol

00:12:07 reaction. And accordingly, our amount of C2 products drop. Interestingly, the

00:12:14 amount of ester, principally here methyl acetate, remains pretty much constant

00:12:19 whether that sulfur ligand is there or there isn't any ligand there. Now this

00:12:23 effect and ability to produce these C4 plus materials is shown on the next

00:12:29 slide using the triphenyl arsine catalyst, which, as we've called previously, gave

00:12:34 us a high amount of acetaldehyde. That was at low loadings of iodine. What we're

00:12:40 plotting down here is the iodine-to-cobalt ratio. As we increase the

00:12:46 amount of iodine in that system, there is a rather dramatic fall in the amount of

00:12:51 the C2 products which are produced. The amount of the C4 plus materials

00:12:56 increased rather dramatically. We've made the system more acidic and quite

00:13:01 possibly what we have operating here is an acid-catalyzed aldol process. If we go

00:13:06 higher in the iodine ratio, we see now that the amount of ester, it's the

00:13:11 straight carbonylation product, begins to grow. And indeed, this is at a two-to-one

00:13:15 ratio. If we were to go up to five-to-one, the ester becomes the dominant product.

00:13:21 Now the next slide shows a system of phosphite, which at very low loadings of

00:13:28 your iodine, principally yields ethanol as your product. You begin to put more

00:13:33 iodine into this system. The amount of ethanol falls. The amount of acid

00:13:38 aldehyde begins to increase. This iodine is evidently forming a cobalt complex

00:13:44 which has a very difficult time in reducing the acid aldehyde. The

00:13:50 amount of ethanol accordingly falls. So we add even more iodine to the system.

00:13:57 The acidity goes up. We begin to produce more of the C4 aldol products. Now not

00:14:03 shown on this one is what the ester is doing. And it is pretty much tracking

00:14:07 parallel to the C4 plus materials. And if we go to ten-to-one ratios of iodine to

00:14:12 cobalt, the ester again becomes the dominant product. So what we've observed

00:14:17 with this particular system as we increase the amount of iodine relative

00:14:20 to cobalt is initially we inhibit the soluble cobalt complexes' ability to

00:14:26 reduce intermediate acid aldehyde to ethanol. And then at later stages, out at

00:14:31 five-to-one and higher ratios, we've even inhibited the ability of the acyl

00:14:37 intermediate to be hydrogenated to the acid aldehyde product. And instead the

00:14:42 competing reaction, the reaction of the acyl cobalt complex with methanol,

00:14:46 becomes rather dominant. And the methyl acetate is our predominant product.

00:14:51 Next slide please. Now temperature also plays a rather dominant role in the

00:14:59 formation of these C4 plus materials. And quite often if you look in the

00:15:04 patent literature, you see a reaction performed at 185 and you say well if I

00:15:09 do it at 200 it probably won't make much difference. Well this shows that a 25

00:15:14 degree C change can make quite a significant difference. At 175 degrees

00:15:20 there are very little of these aldol products observed, less than 1%. If we go

00:15:25 to 200 degrees, they jump quite dramatically and comprise about 34%

00:15:30 based on our homologated products of the reaction mixture. The amount of C2

00:15:37 products accordingly suffers. We go even higher in temperature, the C2 products

00:15:42 become favored among the homologated products, the amount of C4s fall. Now at

00:15:46 225 degrees your principal product is really dimethyl ether. That's not a

00:15:51 homologation product. We haven't inserted CO or hydrogen into that and so

00:15:56 it's not reflected in this particular selectivity. But at 225 degrees you'll

00:16:00 produce about 50 to 55 percent of the methanol converted is converted to

00:16:04 dimethyl ether. The next slide please.

00:16:11 Well another parameter which can be followed is reaction time or I could

00:16:16 have put in here methanol conversion and we'll track the how the extent of the

00:16:21 which the feed is converted. Basically we have a time here in hours that could

00:16:27 be also methanol conversion and that would roughly correspond to about 30%

00:16:31 conversion, 50% conversion, and 65% conversion. So what's observed at the low

00:16:37 conversions of the methanol for this system which will produce the C4 plus

00:16:42 materials is an awful lot of acetal. Principally the dimethyl acetal 1,1

00:16:48 dimethoxyethane. As the reaction proceeds that acetal falls and we do see an

00:16:53 increase in the amount of the C4 plus materials and also the C2 products.

00:17:01 The next slide please.

00:17:05 One last parameter I'll discuss before going on to the final topic I wish to

00:17:11 cover this afternoon is the effect of the ligand to cobalt ratio and this can

00:17:16 also be quite pronounced and it's very sensitive to the degree to which these

00:17:21 aldol products can be formed. We see for the pentafluorothiophenol and the

00:17:27 tetraethyl diphosphine disulfide at very low loadings of those materials

00:17:33 relative to the cobalt, six times as much cobalt, they will promote that C4

00:17:39 production. Go to a one-to-one ratio and the system is more selective back to

00:17:45 your C2 products. Now going to the one-to-one ratio for both those systems

00:17:49 does drop your total methanol conversion from around 65% down to 40 but the

00:17:55 product distribution as well has changed. In the case of diphenyl sulfide

00:18:01 more or less the opposite trend has been observed. At the low loadings there's

00:18:06 very little effect of that sulfide being present. So we go to the one-to-one

00:18:10 loading however we now see that again the aldol reaction is being promoted and

00:18:14 we're getting around 30%, 31% of the product existing as the aldol material.

00:18:21 Now to summarize what we've done here we've seen that the reactions can be

00:18:27 highly dependent upon the nature of the ligand, the product selectivity can be

00:18:32 highly dependent on the amount to which that ligand can be present, the reaction

00:18:36 can also be very highly dependent on the amount of the iodine relative to cobalt.

00:18:44 Now let's go to the next slide and the last topic I'll be discussing is the

00:18:49 effect of various chelating ligands on this homologation reaction.

00:18:55 In principle what we've done here is change our reaction conditions, the amount of

00:19:01 iodine loading that we have principally to somewhat higher ratio, an excess of

00:19:06 one-to-one. Under these conditions a phosphine can be made to more

00:19:10 selectively produce acetaldehyde than ethanol. We're holding on all these

00:19:15 experiments the phosphorus to cobalt ratio constant. Accordingly there will be

00:19:20 on a molar basis twice as many moles of triimbutylphosphine and triphenylphosphine

00:19:25 as the diphos which is our prototypical chelating ligand. Now several

00:19:30 dramatic effects are observed in the use of the chelate. First the conversion of

00:19:35 ethanol is much higher, ten to twenty percent higher. The selectivity in a

00:19:40 weight percent basis to acetaldehyde is about twenty percent higher. The amount

00:19:44 of ethanol is lower, it's around one percent and under certain conditions we

00:19:49 cannot even detect ethanol. In the case of the monophosphines they're three to

00:19:54 four percent. Well that doesn't sound like a big difference but it's a

00:19:57 relative basis that's three or four times as much ethanol. The C4 material is

00:20:02 relatively constant. Now the dimethyl ether again is a big difference. The

00:20:07 diphos system is only around four and a half percent dimethyl ether is formed.

00:20:11 With these monophosphines it's upwards of twenty percent. Now the next slide

00:20:16 shows the effect of varying that hydrocarbon backbone on the diphos or

00:20:21 another measure might be the extent of the bite of the chelate. Diphos of

00:20:26 course will have n equal to two and that's our best ligand. Relative rate of

00:20:30 acetaldehyde formation is what we're plotting. If we go to either side where

00:20:35 the hydrocarbon backbone is one or three we're getting a rather steep fall-off

00:20:39 and the relative rate at which we can produce acetaldehyde in this reaction. If

00:20:44 we go further out it falls even more steeply. Now there have been some recent

00:20:49 reports by Japanese workers and in some of the British petroleum patents for

00:20:53 this reaction that in particular for n equal four and n equal six those systems

00:20:58 are highly selective through the production of ethanol and our results

00:21:03 would perhaps somewhat agree with that under conditions which would normally

00:21:07 force a phosphine to be an aldehyde producer those substances four and

00:21:12 particularly six do not do this very efficiently. The next slide shows the

00:21:20 effects of taking our basic diphos molecule and starting to vary the groups

00:21:25 that can be put around it and for a comparison we have diphos at the top. We

00:21:30 got a rather unexpected result and I still don't have an explanation for it

00:21:34 when we replaced the phenyl with paratolil. The conversion fell and more

00:21:39 importantly our acetaldehyde selectivity fell by around 20 points. If we replace

00:21:45 the phenyls with ethyl groups we get a nice rise in conversion it's an excess

00:21:50 of 92% the selectivity stays in the mid 40s. Now leaving the phenyls attached

00:21:58 directly to the phosphorus and substituting onto that backbone linking

00:22:02 the two phosphorus atoms putting in methyl groups we're getting conversions

00:22:06 now approaching into the mid 90s the selectivity acetaldehyde is still

00:22:11 staying in the low to mid 40s. We go to the next slide we again stick with the

00:22:18 basic configuration of having two carbon atoms linking the phosphorus but

00:22:23 we'll vary that from being a sigma bond as in diphos to a non-saturated bond

00:22:30 double bond in this particular case a cis configuration should be a good

00:22:34 chelate and in our reaction we see that the selectivity conversion has jumped

00:22:39 to 92% selectivity is now around 54%. You go to an acetylenic linkage where now

00:22:46 the phosphorus carbon-carbon phosphorus is linear that system would be a very

00:22:51 poor chelate and indeed our results would suggest that it's not even

00:22:56 functioning as a chelate at all the conversion has fallen to that observed

00:22:59 for monophosphenes as is the selectivity to acetaldehyde. Incorporating the two

00:23:06 carbons linking the phosphorus atoms it's part of a benzene ring again that

00:23:10 should be a good chelate the conversion is mid 90s and the selectivity is over

00:23:15 51%. We introduce more carbons so there are now four carbons linking those

00:23:21 phosphorus the methylene groups also introduce a some more degrees of freedom

00:23:25 to the diphenyl phosphine groups the system evidently is not behaving very

00:23:30 well as a chelate the acetaldehyde selectivity is apparently dropped to

00:23:34 around 28%. Now the next slide shows that we can use other chelates having

00:23:44 three and even four phosphorus atoms variously termed triphos and tetraphos

00:23:54 with Roman numerals after those and I've forgotten which one is which at this

00:23:58 point the triphos is at the top the conversions are now end up into the mid

00:24:02 90s. The selectivity for the triphos is a little lower than what was observed for

00:24:08 diphos but for the other systems for very good chelates we're now getting our

00:24:13 selectivity up into the around 56% under these particular reaction

00:24:19 conditions. The next slide shows some other potential chelates and also what

00:24:24 we thought was a disappointing result diphos is here for comparison using this

00:24:31 bis diphenyl phosphine ferrocene potentially it could behave as a chelate

00:24:35 we would argue that in our reaction system that is not the acetaldehyde

00:24:39 selectivity is very similar to that observed for monophosphine. Going to the

00:24:44 tetraethyl ethylene diphosphonate showing here more or less marginal

00:24:49 results the conversion is a little lower than for diphos and so is the

00:24:53 selectivity. Our disappointing results came when we replaced the phosphorus

00:24:58 atoms with arsenic based on our other work the arsines have been shown to

00:25:03 really produce acetaldehyde and good selectivity we were confident that we

00:25:09 would get a tremendous enhancement in that acetaldehyde selectivity and

00:25:12 strangely in this particular system replacing one of the phosphorus is one

00:25:17 arsenic the conversion went way down as did the selectivity. In fact we

00:25:23 repeated that experiment several times and each time the conversion was truly

00:25:28 this low and we have no explanation as to why that occurred. Replacing both

00:25:34 phosphorus atoms with arsenic again gave a result very comparable to that of

00:25:40 diphos around 86% conversion and the selectivity in the mid 40s. I think in

00:25:48 summarizing here over the 33 years this reaction has been known we've learned

00:25:55 considerable things about the reaction effects of many of the reaction

00:26:00 parameters but yet based on our experience I would say that the

00:26:05 successful practice of this reaction because of its complexity and all the

00:26:10 interactions between the variables still remains very much an art and not so much

00:26:14 a science. The last slide I would like to acknowledge the very fine technical help

00:26:20 of Mr. William Faust and Mr. Dennis Gorley as well as we'd like to express

00:26:26 my thanks to the Gulf Oil Corporation and Gulf Research and Development

00:26:29 Company for support and permission to publish this work. Thank you.