[GHNNS]

Hi all,
I'm new at the forum. First of all, thanks in advance for your time! My mother language is not English, so don’t shoot me for grammar mistakes.
I am a 26-year old Chemistry student. My current project is an improvement of particle size distribution of PMMA non-porous beads, prepared by a suspension polymerization. Project is still going strong, with some promising results. I started with lab scale tests (< 100 g) before scaling up to “reactor” size (> 1 kg)
However, as I investigated SEM pictures of my polymer beads PSD, I observed something strange when the batch size was scaled up to reactor size, namely the presence of hollow particles. In lab scale polymer batches was the presence of these almost negligible (= we were very lucky to spot one!). Despite the same rotor/stator configuration of our mixer, more hollow particles occur (±0.5 – 1%) in the PSD of the reactor batch. The hollow particles have a half-shell structure and are larger in diameter than the normal distribution (> 10 µm). The influence of these hollow particles on the chromatography of the end product is unsure at this moment, but I want to eliminate them in the PSD. Therefore, I need to know the root cause – and that is unknown at this moment. I hope someone could help me.

FYI:
- aqueous phase is water - PVA – surfactant + Helium flushed before polym
- monomers are methyl methacrylate based, also Helium flushed
- initiator is a peroxide
- Polymerization process is performed without Helium bubbling
- Polym temperature is 77°C
Possible root causes:
- polymerization around an air bubble? And than collapsing during clean-up?
- water droplet in a vesicle-like formation (collapsing during clean-up?)
- a frequently-used polymerization method for hollow particles is a seeded polymerization, can this be the case? As our reactor is not that clean (resin stick on mixer and reactor cannot be cleaned 100% without dissembling), can we have a sort of seed polymerization around a PMMA particle that was already present in the reactor vessel? Or is this impossible?

In case you want more info, just shoot or PM me.
I hope you understand I cannot give all the details as this project is off course company-based (= confidential).
Other root causes are off course welcome!
If someone already know an improvement against this issue, feel free to help me.

Have a nice day! 

[P]

My current project is an improvement of particle size distribution of PMMA non-porous beads, prepared by a suspension polymerization.

Is your reaction flask baffled? I found that essential when undertaking suspension polymerisations. Decent mixing, baffling in the flask and the right surfactant. The baffles will improve the PSD of the particles.

However, as I investigated SEM pictures of my polymer beads PSD, I observed something strange when the batch size was scaled up to reactor size, namely the presence of hollow particles. In lab scale polymer batches was the presence of these almost negligible (= we were very lucky to spot one!). Despite the same rotor/stator configuration of our mixer, more hollow particles occur (±0.5 – 1%) in the PSD of the reactor batch. The hollow particles have a half-shell structure and are larger in diameter than the normal distribution (> 10 µm). The influence of these hollow particles on the chromatography of the end product is unsure at this moment, but I want to eliminate them in the PSD. Therefore, I need to know the root cause – and that is unknown at this moment. I hope someone could help me.

I am not sure I have ever scaled a reaction from lab to larger reactor or industrial mix without tweaks being needed.

Q - Are the target particle sizes the same for the scale up as for the lab reaction? Is it just the larger spheres that are going hollow?

Could the polymerization be reducing the volume of the PMMA in the droplet as it reacts and binds each molecule closer than it would have been in the monmer mix?  This could cause shrinkage of the particle...  which, if the outer shell of the droplet is solid, will happen from the centre of the particle rather from outside.

Q - (Sorry if this is an ignorant question, it has been a while since I done this reaction)  -  Where is the initiator in this reaction? Is it predisolved in the monomer and activated at temperature or is it in the aqueous phase?

[GHNNS]

My current project is an improvement of particle size distribution of PMMA non-porous beads, prepared by a suspension polymerization.

Is your reaction flask baffled? I found that essential when undertaking suspension polymerisations. Decent mixing, baffling in the flask and the right surfactant. The baffles will improve the PSD of the particles.

The lab scale tests are performed with a round-bottom flask (500 ml – 1 ltr).
The reactor is a cylinder-formed vessel with a round-bottom. However the suspension is actually prepared in a separate vessel. The vessel has the same shape as the reactor, equipped with a sort of elbow pipe at the bottom to connect both vessels (off course with a valve to open and/or close). The monomers are brought into the aqueous phase under continuous mixing (250 rpm). The droplets are than mixed at high speed to reduce the droplet size (3000 rpm and higher).

However, as I investigated SEM pictures of my polymer beads PSD, I observed something strange when the batch size was scaled up to reactor size, namely the presence of hollow particles. In lab scale polymer batches was the presence of these almost negligible (= we were very lucky to spot one!). Despite the same rotor/stator configuration of our mixer, more hollow particles occur (±0.5 – 1%) in the PSD of the reactor batch. The hollow particles have a half-shell structure and are larger in diameter than the normal distribution (> 10 µm). The influence of these hollow particles on the chromatography of the end product is unsure at this moment, but I want to eliminate them in the PSD. Therefore, I need to know the root cause – and that is unknown at this moment. I hope someone could help me.

I am not sure I have ever scaled a reaction from lab to larger reactor or industrial mix without tweaks being needed. I don’t have to do many tweaks to scale up as all suspensions and mixing processes were done in the emulsion vessel (qty = 18 ltr), but the polymerization itself was only a small part of this suspension. So the suspension can be assumed in all cases as the same composition as the mixing step was always the same. The particle size distribution is thanks to that also the same. FYI: the monomers were filtered before use on an Alumina layer to remove the inhibitor (hydroquinone). However, the yield of this process is less than 50% (but I’m quite convinced that the reminder of inhibitor doesn’t affect the particle).

Q - Are the target particle sizes the same for the scale up as for the lab reaction? Is it just the larger spheres that are going hollow? Almost all of the hollow particles were larger than 90% of the normal distribution. 

Could the polymerization be reducing the volume of the PMMA in the droplet as it reacts and binds each molecule closer than it would have been in the monmer mix?  This could cause shrinkage of the particle...  which, if the outer shell of the droplet is solid, will happen from the centre of the particle rather from outside.

Q - (Sorry if this is an ignorant question, it has been a while since I done this reaction)  -  Where is the initiator in this reaction? Is it predisolved in the monomer and activated at temperature or is it in the aqueous phase? The initiator is dissolved in the monomer phase before mixing both phases and activated at temperature (> 55°C). The reaction temp is 77°C (for 16h), followed by 6h on 87°C (to harden the particles and receive optimal conversion of monomer).

[P]

Just some notes and speculations - hopefully of help.

1 - You are trying for better PSD. The emulsification vessel that prepares the suspension has baffles, yea? These will improve the PSD by creating more even mixing by removing any laminar flow areas around the vessel (I think that's why it works) - with the baffles it ensures turbulence everywhere so there can be no spots in the vessel where coalescence of the suspension can take place, broadening your PSD.

2 -I would expect that it wont cause a lot of harm...  but once the  emulsion has been transferred to the reaction vessel I would want continuous stirring  - ideally with, you've guessed it, baffles in the vessel to ensure the prep stays emulsified (suspended - sorry - suspension particle size is greater than emulsion, I know). Without them I would assume more coalescence of some of the particles and thus a larger PSD.

3 - If it is the top 10% of the PSD by size that is cracking, are these particles that crack larger or the same size as the top 10 when made in the smaller vessel? Larger vessel without baffles = more coalescence I am afraid.  Can't you just put them through a sieve? Maybe this is impractical due to the volume.


- It sounds to me like you are getting worse coalescence in the larger vessel - especially if the reaction is going for many hours. I would bet that....  you know what I am going to say....   the PSD would be reduced and you would have less of the larger droplets form if the reaction vessel was baffled. It might sound as though it wont make much difference but it definitely will to the PSD. There have been papers written about it.  You could show how important this is and test it out by baffling your lab sized reaction flask (you can buy then baffled) - run the reaction and see the difference...  if it is significant then maybe you could argue the case for the larger reaction vessel to be baffled.

Further thought. 4 - when adding the prep to the water in the reaction vessel you add drop wise with stirring yea?  Don't just dump the prep into the water in one go. It may cause a greater percentage of the droplets to coalesce into larger particles that if you add the prep drop wise to a vigorously stirred vessel.


 

[P]

I'd shop around, but this is the sort of thing I am on about.  https://www.wilmad-labglass.com/ProductList.aspx?t=1265

This is some random info about baffles: https://www.ddpsinc.com/blog-0/baffle-options-for-glass-lined-reactors

[GHNNS]

1. We use an inline mixer as emulsifier. The pipe of the vessel is connected with the inline mixer and the outlet of the inline mixer is put back in the vessel, creating a closed circuit (= recirculation). Off course, the composition is during the process a mix of reduced-in-size droplets, probably broadening the droplet size distribution, which will be indicative for our final polymer bead distribution. So the emulsifying step is actually outside the vessel, while the stirrer just keeps everything in suspension (250 rpm). The vessel has no baffles, but do you think it would be a surplus in our mixing set-up with our inline mixer?

2. The polymerization process is at continuous stirring, 250 rpm. The stirrer is equipped with 2 stirrer blades. At different heights.
3. No difference in hollow particle size between lab scale and reactor. We cut off the large particles, but this is based on weight. The hollow particles are in the same range of our desired particle size range, so we cannot get rid of them in that way. It’s not really a point to buy a new reactor vessel, because we actually have no idea if the hollow particles affecting our final product (= chromatography). Do you have an idea what the influence could be of hollow particles in a non-porous stationary phase on HPLC? And do you have a proposal to generate baffles in our reaction vessel (= something homemade?)

4. Actually I do. The volume is about 6 ltr, so dropwise will be taking too much time. Nonetheless this is something I can try, by putting it much slower in the aqueous phase. Thanks!

Thanks for your advice!

Extra Q: Do you know why should the larger particles be more favorable for collapsing?

[P]

Q - Is the 10% part of the PDS present before being added to the reaction vessel or does it form in the vessel due to coalescing? Maybe light scattering would give you a particle size distribution for the prep mix? If you knew this then you would know if the problem was occurring due to coalescence in the reaction or if it was just part of the prep you are mixing.

-If it is part of the prep you are mixing then can you use a smaller droplet size to just get away from the larger ones splitting open? I guess you might not be able to if this is the specific particle size you are aiming for.
-If the problem is being caused by coalescing in the reaction vessel then the baffles should help for sure.

Q - How pure does the particle have to be?   Can you use a filler in the polymer mix to stop shrinkage during polymerization? I am assuming the case here where you actually want to keep the particle size large and can't reduce the size to remove the problem. I guess the shrinkage in volume of the droplet is due to all the monomer molecules now being more closely packed after the reaction compared to before is what might be causing the splitting of the particle and the hollowness.  A totally inert filler may reduce the shrinkage of the droplet during the reaction, thus reducing the stress on the newly formed solid particle...  which should allow a larger particle size before splitting occurs. - Just speculation here though - I do not know what you would use a filler powder...  Maybe some PMMA in a powdered form or something thrown in with the monomer dispersion, idk.

[wildfyr]

I agree that DLS could be a valuable tool for some more granular information on these particles. Its more statistically stringent than counting things on SEM.

[GHNNS]

Thanks for your advice.

We actually analyze our resin particles with DLS (PSD = diameter vs volume%) and counting (PSD = diameter vs number%). SEM analysis is only for the morphology – and now off course identifying the hollow particles.

As the root causes is unsure, I suggest following tests:
- Not only the root cause is unsure, the effect of these particles on our final chromatography is unsure. Does it affect the performance, does it decrease our lifetime? Idk. I can try to prepare a resin batch full of hollow particles. I found some literature to do it with seeded suspension polymerization; I can use PMMA beads of our desired range as seeds, with hopefully some larger hollow particles as result. I can do a full PSD and SEM analysis. Followed by a preparation of a blend of non-porous PMMA containing no hollow particles and a small amount of hollow particles and compare the c’graphy with the non-containing hollow particles batch, to actually frame the effect of the hollow particles. If there is no effect, it is not necessary to eliminate them as they still occur in a very small percentage.

If effect:
- Several samples during the polymerization process (for example each 15 min.). Analyze PSD and check SEM images. Increase of hollow particles during the polymerization??
- I am also thinking about a suspension analysis: measuring the droplet size distribution during the emulsification process. Can we correlate the droplet size distribution to our final PSD? Can we have an indication in our suspension matching with the amount of hollow particles? Idk.

I’m still brainstorming.

What do you think guys?

[DrCMS]

Some thoughts:

Clean your reactor.  If that means taking it apart then do it; using equipment covered in previous batches is lazy and a source of random errors.

Why do you remove the inhibitor with alumina? Is this work ever going to have an application outside of the lab? I despair seeing lab chemists taking "short cuts" in the lab that we'd never do at a larger scale.

Are the hollow large beads due to too high shear at the point of polymerisation.  As the droplets of monomer start to polymerise from the outer edge are they being hit by your stirrer and getting ripped open before they have enough mechanical strength.  How big is the stirrer in your larger reactor compared to the small scale work; consider not the stirrer shaft speed but the tip speed of the two stirrers.

[GHNNS]

Hi DrCMS, thanks for your reply.

It is not really an option to clean the reactor by dismantling. The entire set-up is a build-in design. We have to dissamble everything, stirrer frame etc. The reactor is actually used to produce resin for industrial sciences as the project is part of a company. So it is not about laziness, believe me. It is just out of my power, but yeah, it is something to keep in my mind. Maybe I can do a polymerization with already some PMMA resin present in powder form, and on lab scale. To simulate the possible root cause on larger scale? What do you think?

The removal with alumina is actually taken from the industrial process. But the yield of this reaction is only 40-50%, so there is still some inhibitor left in the monomer mixture. I could increase the amount of initiator to limit the effect of the inhibitor?

The stirrer blade on lab scale is a round blade, sort of moon shape. On larger scale is the surface stirrer vs mixture a lot smaller (less contact). The stirrer is equipped with 2 blades. One at the bottom and one a little bit higher. I’m not sure how I can simulate the conditions on lab scale. Maybe with a homemade stirrer, equipped with two straps or so? Hopefully inert straps 😅

[wildfyr]

I like your idea to see what a contaminated reactor does on the small scale.

I would definitely try to do some work without the alumina plug purification. In my industry we never remove inhibitor from monomers. You can very easily overwhelm inhibitor with initiator, in a lot of cases you can't even tell the difference. Your initiator concentration is often 2-3 orders of magnitude higher than the inhibitor concentration.

Its a great way to save money if you are currently losing monomer and paying for high purity alumina.

[P]

I would definitely try to do some work without the alumina plug purification. In my industry we never remove inhibitor from monomers. You can very easily overwhelm inhibitor with initiator, in a lot of cases you can't even tell the difference. Your initiator concentration is often 2-3 orders of magnitude higher than the inhibitor concentration.

I've not done a polymerization for some time  -  I always used to remove the inhibitor via reduced pressure distillation. Not sure how things would have gone with using excess initiator - we were trying to precisely control polydisperities, molecular weights and co-polymer ratio. Everything was high purity and PDs were temperamental to impurities, reaction times, % yield and just about everything.

If you were going for a tight PD would you consider removing the inhibitor wildfyr? I can imagine the inhibitor terminating chains randomly during the reaction which might broaden the PD   -  idk - just speculating - I never ran one with the inhibitor still in it.

[wildfyr]

I've only done free radical polymerization with inhibitor present. Whenever I've done controlled polymerizations I purified it. Like ATRP or NMP I would certainly distill the monomer to get rid of inhibitor or oligomer.

In in industry now and controlled polymerization is quite rare. Requires a lot of extra effort.