Wednesday, March 30, 2011

Process Wednesday: Kilomentor's questions on crystallization

Continuing on in Kilomentor's questions for scaling up, let us confront crystallization:
  1. Is a method in place to determine the solvent composition?
  2. Is there a method in place to determine the ratio of volatile solvent to non-volatile residue?
  3. Is there provision to initiate crystal formation? (seeding)
  4. Is there a provision for managed crystal growth?
  5. Is there provision for crystal size assessment?
  6. Is there a study of crystal aging(ripening)? Does the purity change depending on aging?
  7. Is there an IPC (in-process control -- CJ) to verify that the crystallization is complete and the correct quantity of residual product is left?
Considering I've never really done crystallization as part of a formal process (as opposed to the informality of looking at the flask and saying "Yep, them's crystals!"). I think it is interesting to learn all the different things that you have to control for. I'll be honest and say that I can't really begin to answer questions 1, 2, 5 and 7 HPLC?).

Crystal aging/ripening is something that can be an unpleasant surprise. You have to prepare yourself for the potentiality of leaving a tray of nice white/yellow crystals and coming back to a (figuratively) hot brown mess. Of course, Kilomentor's talking about something a lot less dramatic than that, I'm sure.

As a relatively young chemist, the one thing that I was surprised to learn was the density of seeding. I think we're all used to the concept of just a pinch of seed crystal from school. I was surprised to learn that seed crystals may need to be added in a much higher amount on larger scale reactions; 10% of total mass (or higher) may be necessary to induce appropriate levels of crystallization.


  1. For solvent composition and residual solvents (1 & 2) you can use NMR for a rough number and GC if you want to be more accurate. There are probably a lot of ways to check 7; one easy one is just to use HPLC to quantify the mother liquor (i.e. mg/mL remaining).

    Crystallization is one of the more interesting topics that isn't covered well in most grad programs, where it still has an aspect of voodoo about it (like heterogeneous catalysis for hydrogenation, another area much more well-understood in industry). There are a couple surprises when you start looking at it in process chemistry -- e.g. normally you stir during a crystallization, and the type of stirring matters, since magnetic stirbars can grind crystals into small particles that further seed the mixture. Seeding, controlled antisolvent charges, and temperature ramping can all help a lot in getting the right morphology and particle size, which are needed for a good filtration and wash, something that you covered in a previous post. With the right tools you can also develop a more fundamental understanding of the metastable zone width.

    I think part of the reason that most people don't get a lot of exposure to this in organic synthesis programs is that the class of compounds are different. In pharmaceutical chemistry where you have a lot of polyaromatics and heterocycles it's almost a given that your compound will be crystalline, and in fact you might face the problem of poor solubility.

  2. A long suffering crystallization development chemistMarch 30, 2011 at 9:31 AM

    "it's almost a given that your compound will be crystalline"

    Hehehehe thanks for the Wednesday chuckle

  3. Ha! Now that I think about it, just last year I had a compound that we could never crystallize; it always came out amorphous. I guess my recent projects made me forget that one :-)

  4. Something else that's nice to know as early as possible is polymorph control. It's really no fun to find a polymorph issue once you've scaled up to kilos, much less hit the issue all at once in the pilot plant.

  5. regarding point 3, especially from a GMP aspect seeding is extremely crucial. It is something that makes an auditor cringe if you tell him "well we just cool the batch down and the stuff will crystallize eventually" - the seeding must be a very well defined process, temperature at time of seeding, resting time after addition of seed material before ramping down the temperature, amount & polymorphism of seed crystals added...

    Re point 6, you should definitely know what happens if you are forced to keep your material longer in the reactor than you initially anticipate. From my experience, isolation equipment such as filter driers and centrifuges tends to be rather prone to equipment failures and getting things to work again can sure take some time. It really puts your mind at ease when you know your material still has the same purity after two or three days and helps you supporting your case when filling out deviation reports. Behaviour of the filterability of your nice crystals on scale is a whole different story.

    I personally really started to like the process of crystallization when getting in touch with process chemistry, it looks really impressive through the manhole of a 6 cbm vessel, watching your crystals grow after seeding.

    Sorry for the rambling. Very nice blog CJ, I really enjoy reading it.

  6. NS29 -- Thanks for the kind words. And thank you for coming and making it better with your informed comments!

  7. As noted above crystallization is rarely taught or experienced in school but in large scale it can be very exciting to learn and implement. Maybe I have been fortunate with majority of crystallization experiences have been positive where bigger challenges in other scale-up operations (or perhaps because at stage get to see/touch a product I have forgotten any problems).

    I won't dispute that this a good set of questions to consider for crystallization however temper for each one has to be put the the context of does it matter for the particular case. If there is polymorphism or wide variations in yield and/or purity then such answers will likely provide the keys to control.

    #3 10% seeding seems extreme (1% or less customary) and would only apply that excess if slow or polymorph considerations.

    #4 Beside solvents and concentration the physical parameters like stirring, hold temp or cooling rate have big influneces. Unless otherwise dictated faster is better as in Plant time is serious cost contribution.

    #5 Size determination can be either light scattering in critical situation with a microscope or sieve evaluation servicing for quick tests.

    #6 Assume this means aging in solution during crystallization as if impurities begin to drop out or co-crystallize after a certain point or longer time. Rates and hold time factors get balanced. Darkening on filters/tray would be more of a post-isolation decomposition event and addressed differently (inert gas blanket?)

    #7 Use of product assay method good if fast turn around even if must strip mother liquor sample first however sometime good old TLCs are great approximations because just like with reaction can monitor disappearance.



looks like Blogger doesn't work with anonymous comments from Chrome browsers at the moment - works in Microsoft Edge, or from Chrome with a Blogger account - sorry! CJ 3/21/20