[curiouscat]

I was trying to debug an incident where during vacuum batch distillation of Molecule A (see image below) there was a brief exotherm towards the very end and a polymeric by product was formed along with the liberation of free water. What could have been any possible reactions, any ideas?

A is present in approximately 97% w/w in the feed. Molecules B,C and D make up the rest but the proportions can vary.  Towards the end these B,C,D are mainly left in the still. Water content of feed was negligible.

I can think of polymerization products (e.g. see fig. ) but none seem to eliminate H2O. Any other condensation possibilities? The liberated water seems acidic (pH~3) if that helps. The same distillation has been carried out without incident before but some protocol changes (time, temp., packing etc.) may have happened.

Some of this data may be red herrings. 

[discodermolide]

Any metal contaminants such as iron (rust)?

[discodermolide]

Do you have DSC data for the components and the mixture?

[curiouscat]

Any metal contaminants such as iron (rust)?

Very likely. It's a SS column & long in use. Rusting can't be entirely eliminated. Had gone through a recent retrofit so debris may have been dislodged.

To its defense same distillation has been carried out before with no incident in same column. 

[curiouscat]

Do you have DSC data for the components and the mixture?

Sadly not.

Tried replicating the distillation in a glass lab column  but couldn't see this phenomenon.

One factor is duration: in plant this batch can stretch up to 24 hours. Our stupid lab gave up within a shift (8 hrs run) I think.

[discodermolide]

I think you were lucky this time.
Try and get some thermal data on everything with and without rust.
Benzyl halides are known for a low onset temperature, even lower when rust is present.

[curiouscat]

I think you were lucky this time.
Try and get some thermal data on everything with and without rust.
Benzyl halides are known for a low onset temperature, even lower when rust is present.

Indeed. I mean with (A) itself we do know we are teetering on the edge of a polymerization but with careful batch size control, inerting & temp. control (and emergency dumping valves) so far it has never been a problem. (Personally I never walk near that column when in use.  )

One part still puzzles me: Could any of this form free water?

[curiouscat]

Benzyl halides are known for a low onset temperature, even lower when rust is present.

Thanks! I've had them heat this reaction with & without catalytic acid in the lab. Now we'll test with some rust.

[discodermolide]

Sure, you can eliminate H₂O and get the chlorostyrene, which is even more thermally labile.
Try and get the DSC data, outsource it, otherwise you are, safety wise, on thin ice.
What is the distillation temp. and batch size?

[discodermolide]

You could also eliminate HCl and get the aldehyde!

[curiouscat]

Sure, you can eliminate H₂O and get the chlorostyrene, which is even more thermally labile.
Try and get the DSC data, outsource it, otherwise you are, safety wise, on thin ice.

Will do. I agree with you though. I've had my reservations about this process and have let them be known.

What is the distillation temp. and batch size?

I'll get you the exact data from the logs in a day or two. The vacuum is ~10 mmHg abs.

[curiouscat]

You could also eliminate HCl and get the aldehyde!

Sure, you can eliminate H2O and get the chlorostyrene, which is even more thermally labile.

Thanks! That's very useful. I'll try and get standards for both chlorostyrene & the aldehyde and see if those match any of our GC peaks in the problem batch.

PS. By chlorostyrene do you mean these?

I'm confused because Sigma calls this a Chlorostyrene too, but I can't see how elimination from my molecules can get me there.

[discodermolide]

Both of the ones you drew, not the sigma one.
I had a 2,6-difluorobenzyl alcohol, it had a decomposition energy of 1050KJ/Kg, which gave us a calculated adiabatic temp increase of >500°C,
Unfortunately I don't have the DSC data here, but I can remember that with rust the decomposition started about 60°C lower than without.

[curiouscat]

Both of the ones you drew, not the sigma one.
I had a 2,6-difluorobenzyl alcohol, it had a decomposition energy of 1050KJ/Kg, which gave us a calculated adiabatic temp increase of >500°C,
Unfortunately I don't have the DSC data here, but I can remember that with rust the decomposition started about 60°C lower than without.

Indeed that's very likely here too. I do know that Fe catalyzes the self polymerization of my (A). Though literature shows that this seems to need quite a bit of heating and isn't a very fast reaction.

One thing that may ease our situation is that none of B, C, D are the major products. Early on they are diluted to such excess that the adiabatic temperature will be self limited by the specific heat capacity of the mass.

Our critical phase will be towards the end when they start getting concentrated. My guess is we may have a total (B+C+D) inventory of ~20 kgs which while not negligible isn't huge either. So perhaps one mode of inherent safety could be if we find a high boiling diluent that can be added during the distillation so that the maximum adiabitic rise becomes self limiting, worst case. Need to find  a suitable inert material though.

Thoughts?

[discodermolide]

Sure adding a diluent would be good, I come back to the carbowax 400 which we used for our mol dest.