Disintermediation

Historically, scientific journals got started as a way to share information. They were the most effective ways to tell other researchers what you were doing and find out what they were doing. Nowadays, they aren’t really. They are complicated intermediaries – which add some value, true – between you and the people you want to share information with. They have rules which are largely arbitrary and which impact negatively on how useful  they are as media for sharing information with people – rules about how long a paper should be, about how it should be structured, about what should be put in and what should be left out. More importantly, their primary function nowadays is not to share information, but to score points in The Academia Game^TM, one of the first and most dramatic successes of the ‘gamification’ craze.

Anyhow, I think the search engines we have nowadays are good enough to cope with a bit of disintermediation; so I thought I would have a go at sharing my information here instead. Some of it, anyways. I’ve got a piece of work, you see, that I can’t see scoring any points, and I want to tell you about it.

Back in my PhD I came across a paper on work done by two researchers at the University of Texas El Paso in the 1970s, Wang and Cabaness. In this paper they had investigated the copolymerisation of acrylic acid (AA) and acrylamide (AAm)  in the presence of a number of Lewis Acids of general formula XCl₄, and reported that tin tetrachloride could induce the formation of a copolymer with the regular repeating formula ((AA)₄(AAm)) – four acrylic acid residues in a row, followed by an acrylamide unit, rinse repeat – which they attributed to a 1:1 alternating copolymerisation of a SnCl₄(AA)₄ complex and acrylamide.

Structure postulated by Cabaness and Wang 1975

I was intrigued by this article, because I was also studying polymerisation with Lewis Acids (only alkyl aluminium chlorides rather than the XCl₄ species), in systems where we got 1:1 alternating copolymers of donor monomers (like styrene, butadiene, vinyl acetate, or acenaphthalene) with acceptor monomers (like acrylates, methacrylates, and oh yes, acrylamides and methacrylamides). The whole basis of our understanding of these systems was that complexes of acceptor monomers with Lewis Acids made them fantastically better at being acceptor monomers, and I couldn’t see how four monomers that were all complexed to a single Lewis Acid ought to polymerise together: if they were better acceptors, they would be less likely to polymerise with each other, and once one added onto a growing polymer chain it seemed to me that it would be more likely to add a (relatively donor-ish) free acrylamide rather than one of its fellow complexed acrylic acids that was probably held in a sterically unfavourable position, as well as an electronically unfavourable condition. And 4:1 regular polymers had never been reported with acrylic acid and any more conventional donor monomers that would be more likely to behave themselves. It was all very mysterious. Nobody had ever confirmed or followed up on this work of Wang and Cabaness; which was disappointing, but not very surprising, because the whole field of playing with various Lewis Acids and donor and acceptor monomers had been a big thing over approximately the years 1968-1975 and had then petered out for no good reason.



So many years later I found a bottle of SnCl₄ lying around (it’s a liquid; it comes in bottles – the Sn(IV)-Cl bond has a lot of covalent character) and remembered this paper and thought I would have a go. Wang and Cabaness had only looked at their polymers using elemental analysis, which realistically tells you about 2/10 of not very much about polymer structure, whereas I had spent quite a lot of time looking for regularity in polymer sequences using Nuclear Magnetic Resonance Spectroscopy, which tells you an awful lot, and I thought I would make the polymers they made and have a look at them with NMR. The proton NMR spectra of polymers are always broad, and the backbone protons of acrylic acid and acrylamide residues (as you can guess from their structures) end up on top of each other in a messy way.  So the way to tell what is what is to do carbon-13 NMR spectra, which gives you nicely resolved peaks in the carbonyl region, and reasonably well resolved peaks for the methine carbons.  If you look at the carbonyl region of a copolymer of AA and AAm, you can pick out the six different nearest-neighbour environments quite nicely. In the figures below, for example, you can see clearly how base hydrolysis of PAAm generates isolated AA units on the chain, which have a protective effect on neighbouring AAm and make it much less likely that they will be hydrolysed.

Carbonyl (left) and methine (right) regions of  13C-NMR spectra of polyacrylamide at different levels of base hydrolysis. From: Yasuda K, Okajima K, Kamide K. Study on alkaline hydrolysis of polyacrylamide by carbon-13 NMR. Polym J (Tokyo) 1988:20(12):1101-1107.

Carbonyl region of 13C-NMR spectra of AA-AAm copolymers. From: Candau F, Zekhnini Z, Heatley F. ^I3C NMR Study of the Sequence Distribution of Poly(acrylamide-co-sodium acrylates) Prepared in Inverse Microemulsions. Macromolecules 1986:19:1895-1902.

 Now, when I got back and had a look at the paper again, I was troubled by the times given by Wang and Cabaness for these reactions. These sort of radical reactions usually have an inhibition period when nothing much happens at the beginning, even if you take pains to get rid of dissolved oxygen from the system first. The reported reactions were done under nitrogen. So, if you bubble nitrogen through a reaction mixture enough to get rid of oxygen, to do a halfway decent job you need to do it a lot longer than the 100 s or less quoted for these reactions. So you would have to bubble nitrogen through at room temperature, then shift to a higher temperature, at which it would take a lot longer than 100 s to warm up to the quoted temperature values. Maybe the times quoted were the time it took after the inhibition period finished, but before the reactions were quenched? Which would mean Wang and Cabaness would have had to have been watching their reactions like hawks, and even so quoting reaction times to a precision of one second was a bit ridiculous. So, anyhow, I resolved to cut the temperature down to 60 °C to give me a bit more time to work with and quote reaction times in minutes rather than seconds.

Table 3 - the tin tetrachloride data - from Wang and Cabaness 

I first tried doing what I would usually do, which is freeze/thaw degassing.  I still had a significant inhibition period after I put the samples in a 60 °C oil bath. When anything happened, it happened too fast for me to stop it, and what it was, was my polymer ‘popcorning’ into a solid mass. This is something that happens with monomers that polymerise very quickly. Polymer solutions are viscous, and transfer heat worse the more viscous they get.  Polymerisation reactions are exothermic. So, a polymerisation that goes quickly generates a lot of heat and a viscous solutions which makes it difficult for the heat to dissipate, and as the temperature increases the reaction goes faster, making it even hotter and more viscous, and you get very quickly to a temperature where your solvent vaporises, and your polymer chars, and if you are doing a reaction in a 20 tonne reactor instead of a tiny little tube you ring the insurance company. That is what happened to my first attempts: they polymerised into intractable masses that I couldn’t get out of my reaction vessels without smashing them and then wouldn’t dissolve once I’d smashed them out.

I decided to drop the freeze/thaw degassing and use the dodgier ‘sparge with nitrogen’ method in round bottomed flasks that would be (should be) easier to get the polymer out of.

Yes, I could get the polymer out without smashing anything. But the reactions that formed it were the same: the vessel sat there for a while without doing anything, then suddenly there was the popcorny noise of solvent vaporising and insoluble masses with yields of approximately 100%.

I cut the temperature a bit further, and cut the concentration of everything a bit, and still couldn’t get any useful polymer. 

Then my colleague Daniel Keddie suggested something that made a lot of sense which I should have thought of a long time before: why not put in a chain transfer agent? This is something that cuts the length of the polymer chains but shouldn’t (knock wood) have any more dramatic effects on the chemistry of the reaction: it just introduces a ‘jump to a new chain’ step that replaces some of the propagation steps. So I put in enough butanethiol to reduce the degree of polymerisation to about 25 and had another go. And I got polymers I could remove from round bottom flask, which actually dissolved up okay! These proceeded to a conversion of about 40-60% when heated at 60 °C for 30 minutes – again, almost all of this time was inhibition period, so I couldn’t actually stop any of the reactions at a low enough conversion to get an unambiguous relation between the composition of the polymer and the feed composition of the monomers.

Following the next step of the procedure – dissolving in water and reprecipitating in methanol – was a little tricky, because the polymers were reluctant to dissolve in plain water. So I added a bit of base to convert the acrylic acid residues to sodium acrylate – and crossed my finger and kept the temperature low to avoid hydrolysing any of the acrylamide residues to sodium acrylate. And I managed to get some halfway decent carbon-13 NMR by running these overnight, like so:

 

Carbonyl region of 13C NMR spectra of AA:AAm:SnCl₄ polymers made by lil' ole me
 

Now the area of these peaks should be pretty much proportional to the amount of carbon contributing, so what you can tell straight away from these results is that the products  don’t have a constant  4:1 AA:AAm composition, so the mysterious result of Wang and Cabaness is artefactual.

The spectra aren’t similar enough to be of 1:1 alternating copolymers, either – there is always excess AA, and the main AA peak always starts out as  the one with one AA and one AAm neighbour. If I was getting a 1:1 alternating copolymer and then hydrolysis, I would expect to see significant amounts of AA with two AA neighbours coming in as soon as the AA with two AAm neighbours started to disappear.

Besides these negative results, though, there isn’t a lot that I can say. The tin tetrachloride is doing something: it is making the reaction go a lot faster than it would.  It *might* be encouraging a tendency towards alternation. Because of composition drift, though, and because of the uncertainty in the literature reactivity ratios, I can’t tell for sure whether there is any shift in the polymer composition compared to what I would see without the tin tetrachloride.  It looks to me like I am getting a significant amounts of base hydrolysis of any acrylamide residues which aren’t alternating – which is what you would expect from the literature.

In order to publish this, I would need to make sure my 13C NMR was quantitative, which wouldn’t be too hard. I would also need to work out a way to kill my reactions quickly, and I would need to figure out a way to reprecipitate the polymers without hydrolysing the acrylamide residues. Obviously these are soluble problems, but this system isn’t doing the really exciting thing it was reported as possibly doing, and doesn’t appear to be doing anything moderately exciting, so I don’t know if it is worth carrying on with.

 

Conclusions:

1)      I have no idea how Wang and Cabaness got something out their system that they could dissolve and reprecipitate and find viscosities of.

2)      There is no evidence that the 4:1 acrylic acid: acrylamide composition they report can be reproduced.

3)      People probably tried to repeat their work before, and came to the same conclusion, but didn’t put it on the internet.