Skip to main content

Analytical Chemistry

Droplets, Then Crystals

If you’re a chemist, then you like crystallization. I think that’s pretty much a given; I’ve never met anyone who doesn’t appreciate a good crystal, and watching them form out of a solution never stops feeling a bit like magic. When I was doing a project involving metal-organic frameworks, I had some of the best fun I’ve ever had in the lab making new crystals all the time (and some of the more frustrating times I’ve had while trying to get them to do what I wanted to do afterwards, but that’s another story!)

And of course, many small-molecule pharmaceuticals are produced in some sort of crystalline form, not least because that both helps to purify them and puts them into a defined physical state that we can trust for further formulations. When that solid state changes and a different polymorph forms, chaos can ensue (see the links in that post, and there are plenty more stories besides). One would imagine, since we’ve been crystallizing things for several centuries by now, and since the process is behind the production of hugely important drugs and other substances, that we would have a decent idea of how it happens.

We do not. That link above goes into some of the details, and it’ll show you how tricky the whole process can be when it comes to polymorphs. And even when we’re just talking about one good ol’ crystalline form, the exact process by which those individual molecules come together in solution and start to form orderly lattices is still a matter for hand-waving. This new paper has some details on a molecule that gets crystallized from solution rather often and on a rather large scale: ibuprofen. And the fact that we’re still learning how ibuprofen becomes a solid will tell you all you need to know about where the state of the art is.

As has been seen with some molecules (and often suspected in many more), ibuprofen turns out to go through an interesting liquid-liquid phase separation on the way to becoming a solid. This is “spinodal decomposition“, known to organic chemists as “oiling out” when it happens on a larger scale. You’ll have a solution of some compound, and if you start to adjust the conditions (temperature, salt concentration, addition of another solvent, etc.) you’ll hit a point where suddenly things become cloudy, and that’s usually followed by droplets of a second liquid phase visibly separating as the microscale droplets (which caused the cloudiness) coalesce.

In this work, NMR methods are used to probe the composition of that ibuprofen-rich liquid phase. Molecular dynamics (translational motion and rotation) are slowed compared to the solution phase, which is denser and much more viscous. Intermolecule NOE signals start to show up, indicating close molecular proximity between the ibuprofen molecules that doesn’t exist in the starting solution. In fact, the liquid phase seems to have the molecules about as tightly packed as they are in solid crystalline ibuprofen – they’re just not ordered into a regular lattice, or not yet. This appears to be the first time that anyone’s been able to get such measurements, actually.

It is, in fact, that concentration and reduced motion that allows the crystals to start forming. They don’t nucleate out as tiny solid particles from the initial bulk phase (although this has been demonstrated with other compounds under other conditions). The liquid phase separation is the key, and it’s interesting that there are compounds that can do this without those microdroplets coalescing into a separate oily phase before that crystallization can kick in, and I’m not sure how that works. Perhaps the surfaces of the droplets develop some sort of charge distribution that keeps them from coming together so easily?

The question this paper raises is whether we can exploit this sort of behavior to make new liquid-phase formulations of such drugs, things that are nearly as concentrated as crystals but don’t need to break up a thermodynamically favorable 3D lattice in order to dissolve. Could you really make a gel-cap full of a metastable liquid phase like this (without worrying that they will suddenly solidify into pellets in the bottle?) No one can say it’s impossible, that’s for sure. We just don’t know enough either way!

43 comments on “Droplets, Then Crystals”

  1. Barry says:

    are we any closer to understanding why scratching with a glass rod (sometimes) induces crystalization?

    1. John Wayne says:

      Yeah, magic.

    2. Nick K says:

      I believe that the standard explanation is that scratching produces large numbers of tiny glass fragments which act as nuclei for crystallization. That said, I have found a nickel spatula just as effective for inducing crystallization as a glass rod, and nickel is softer than glass.

      1. Derek Lowe says:

        I always thought that it produced some roughened surface that promoted it somehow, but yeah, magic as far as real explanations go. Perhaps it’s a local concentration effect promoted by the pressure between two surfaces?

        1. Barry says:

          if it’s a local concentration/pressure effect, we should see this with sonication?

          1. Pedwards says:

            Inducing crystallization via sonication is a thing:

        2. wildfyr says:

          I had an idea in my head that it was due to a really large increase in surface area in the newly formed rough spot, along with a bunch of sharp edges, where unique things could happen a la surface enhanced Raman.

          1. Derek Lowe says:

            That makes sense – pretty tough to prove, though, like a lot of nanoscale surface hypotheses!

          2. wildfyr says:

            Good thing I’m a nanoscale surface chemist! Though no expertise in crystallization dynamics. Give me $2 million and a couple years and I’ll get back to you!

        3. Chris Phoenix says:

          Wild guess: Liquid water forms locally ordered arrangements of hydrogen-bonding. Probably more so near a glass surface.

          Glass can be extremely smooth. Maybe also that magic nickel spatula. When you scrape a glass curve against a glass surface, you’re probably shearing the water at the nanoscale.

          If you disrupt (shear / “crack”/separate) the local-quasi-crystals, you may be briefly exposing fields of unbalanced charge. I’m imagining an analogy with triboelectricity.

          Some combination of exposed unbalanced charge, and the rapid rearrangement of water to rebuild the structure, might change the local conditions enough to nucleate a crystal.

          Probably not the real explanation – but might be useful grist for the insight mill.

      2. A Nonny Mouse says:

        I was demonstrating in undergraduate lab (must have been the year below you) while doing my PhD and suggested to someone that they scratch the test tube with the solution of the compound, which she did. I then suggested that it might be better doing INSIDE the test tube where the solution was.

        1. Derek Lowe says:

          Yes! I have seen this one in person as well, as a TA. That’s one of the things, along with spotting the aluminum back of the TLC plates rather than the silica side, that made me realize that I could not possibly anticipate all end-user mistakes, and that such things had to be tested in the field.

          1. Barry says:

            a friend (Hi Mike!) relates a fellow TA was interrupted by an undergrad needing help “I can’t get my steam-bath lit”. Sure enough, he had plumbed a water-line into the upper nipple and a gas line into the lower nipple and was frustrated that he wasn’t making steam
            There is no such thing as fool-proof explanation. Especially when your audience lacks a lot of your working vocabulary

          2. Hap says:

            When I was watching fellow grad students TAing the advanced organic lab (though held a year after the initial organic lab sequence), I hit the jackpot. One student was doing a column and pouring something onto the top of the column (not that big a column) from a beaker. It turned out he was loading his compound onto the column in about 200 mL of solvent. The TA advised him to blow the material through the column, since he was unlikely to get preparatively useful separation. Another student the same day was spotting TLC plates and getting really large blobs. The TA asked him if he could make smaller spots and he said that he couldn’t. The TA then asked what he dissolved the material in. (blank stare)

            Unfortunately I don’t have room to talk because I tried (for some reason which escapes me now) to blow-dry my ferrocene in first-semester organic lab. Eventually, I scraped the powder off the benchtop, filtered out the paint chips and assorted dirt, and recrystallized it again. My TA was not happy with me.

          3. BuLi fail says:

            … I have heard of a student supposed to use BuLi for the first time. Got a fresh batch, delivered in the usual aluminum cans…. filled with styrofoam to protect the bottle. Being the first thing visible once the can was opened, the student started weighing out the styrofoam!!!

          4. Tocrat says:

            Derek, you need to do a post on amusing things seen whilst demonstrating undergrad labs

          5. Nick K says:

            This kind of practical incompetence is not limited to undergrads, alas. Just last week I found a co-worker (PhD, age 56) attempting to withdraw a solution from a sure-seal bottle into a syringe. He couldn’t understand why the solution kept flowing back until I pointed out that he was creating a partial vacuum in the bottle.

          6. Escapee says:

            After patiently explaining to a student in undergrad labs that their solution would eventually pellet if they left it in the centrifuge long enough, I was confused that they said they’d done it as long as they could bear it but now their arm hurt too much.

            When I asked them to show me what was going wrong, I realised they’d been holding the sample on a vortex for half an hour straight.

        2. Larry says:

          Here’s one- I was TA’in in my senior year as an undergrad. This was back in the day when we had the students bend glass tubes with the bunsen burner to make wash bottles. A student comes up to me and say’s “Can you look at mine”, then proceeds to drop the thing in my hand. Of course, it just came out of the flame, cursing ensued, yada, yada, yada…..

        3. Jon says:

          How do I make dry ice? Academic in charge of lab: Use a paper towel
          I spent 2 minutes watching the student blotting ice with a paper towel before realising all his fellow students were using solid CO2….

      3. milkshake says:

        even better than scratching the flask with glass rod or a spatula is to put a drop of the mixture on the inner surface of a ground joint, stopper it with a glass stopper, and gently turn the stopper around

        1. Nick K says:

          I have inadvertently used this technique when some of the oil has crept between the stopper and the neck of the flask. Unfortunately, this usually freezes the stopper in the flask.
          A number of times I have had a compound start to crystallize while I was trying to load it on a column. This is extremely irritating.

          1. milkshake says:

            yes, it is best not to do this with neat oiled out compound, but start with solution that is becoming cloudy, before oil comes out (and re-heat if necessary).

            With stuff crystallizing on the column – I have a better story. We bought an early model of Combiflash ISCO LC, just when ISCO started, in 2000. It had ports for 16 columns (it was built for parallel purification on multiple pre-loaded columns – pretty over-engineered, the fraction collector couldn’t handle that many fractions anyway). And two days later, my colleague injected a crude phthalimide compound as neat oil, into that Combiflash, where the oil finally solidified. The poor sales rep/field engineer who just installed it in our lab took it apart and try to undo the blockage. There with the cover off, I could see where all the money went, building this LC – two ginormous solenoid valves for distribution, each with 32 ports, and additional 3-way main control valve. And bundles and oodles of Teflon tubing – now with our phthalimide solidified all the way through! Totally hopeless. We had to pack the thing in a box and ship it back to the factory. A week later, we got an angry phone call from some engineer in ISCO plant in Illinois: he wanted to know if we were injecting some kind of superglue on purpose…

        2. Jamil says:

          Sometimes a chip of dry ice provides enough edges to kick the crystals out. But I think usually an impurity is causing the problem if the crystallization worked last time. Sometimes a temperature or even solvent gradient across the crystallization vessel really helps as the nucleation and growth events require slightly different conditions.

        3. Anonymous says:

          As an undergrad doing summer research in a pretty good synthesis lab, a senior researcher saw my oily crystals. He got an unglazed porcelain dinner plate from a drawer. He put the plate in a sturdy bag and broke it into pieces. He placed some oil-crystals on a piece of the plate, moved them around with a spatula, and much of the oil migrated into the porcelain. (Unglazed porcelain is really good at absorbing oils, hence need to glaze it for normal, non-lab use.)

          The de-oiled crystals were much easier to wash with some cold solvent. … Years later, scavenging stuff from a retired prof’s lab that was being cleaned out for a refurb, I salvaged a box of small (~3″ diameter) unglazed porcelain saucers (as for tiny tea cups; with a science supply company label on the box). Nobody knew why I was salvaging them until I explained.

          You can still buy unglazed porcelain from Fisher, etc., but they are rectangular slab “plates” made for the lab, not unglazed kitchen or craft project plates.

          1. Nick K says:

            The use of porous pot to dry oily crystals works admirably well. It’s far easier and less wasteful than trying to wash the crystals with solvent. Unfortunately, the technique seems to have been forgotten.

            A good substitute for porous pot is ordinary filter paper.

  2. Eugene says:

    It would be interesting to apply an external electrical field to this phase and see what happens. Inhibition or promotion? Or nothing?

    1. Barry says:

      I expect you’d inhibit the formation of any crystal in which some of the units pair head-to-tail but might promote crystalization in which all the units have the same orientation

  3. Dave says:

    Twinning is a real pain. To the best of my knowledge, it’s unfortunately not controllable, at least in the majority of cases. I’ve grown some decently large crystals, only to have them marred by being twinned.

    Expanding on the earlier comment about crystallization using electric fields, there has been some work done with crystallization using magnetic fields. One would presume that there’s also been work with various frequencies of electromagnetic fields, perhaps including light of various wavelengths (linear and/or circular polarization?).

    For an interesting look at crystallization, consider the various liquid crystal materials. I dealt rather extensively with Twisted Nematic liquid crystals in the early part of my career, and the alignment of these is highly dependent upon electric fields (thus the Liquid Crystal Display).

  4. MattF says:

    Note, though, that there can be close-packed non-crystalline structures such as Weare-Phelan foams— crystallization is not inevitable. I imagine that oddly-shaped molecules could promote solidification into amorphous solid phases.

  5. bodrell says:

    I think spinodal decomposition is more subtle than “oiling out” as you describe. In the phase diagram, there are regions where a change in temperature or composition causes phase separation into droplets – i.e., both phases coexist but at least one phase is discontiguous (droplets). For spinodal decomposition, you go through the part of the phase diagram where there is no intermediate binary phase. This causes a bicontinuous separation – every “oil” region is connected to every other oil region, and vice versa with the “water” regions.

    I learned about spinodal decomposition via some clever scientists in Scotland who figured out how to pause the phase separation process by “jamming” the interfaces with small particles. These are very interesting materials that are still looking for a good application. Do a search for “Bicontinuous Interfacially Jammed Emulsion Gels (bijels)” for more info.

    1. Chris Phoenix says:

      With that much surface area, and that highly connected a topology? There’s got to be a way to make an ultracapacitor out of it.

      1. Barry says:

        but to make a capacitor out of it, you would have to wire together all the nano-droplets

  6. Peter S. Shenkin says:

    This post is reminiscent of so many things (not even to mention blunders made by newbies, which frankly all of us were at one time, and probably still are when we try to do something new).

    Aside from those then, this is reminiscent of protein folding (remember the “molten globule”?) and the idea of distributing the drug in the form of small concentrated globules of course makes me thing about glass transitions. Would it make sense to distribute any drugs in the form of a glass? You know, like hard candy…. If you could form a “stable” glass that is, or for that matter introduce it as an impurity in a glassy substance – yes, such as hard candy.

    Imagine reading the label: “Added sugar: 99.99%”….

    1. loupgarous says:

      In an oncology application – say, a glycolysis inhibitor, if you chose mannose as the sugar you embedded the drug in, so the cancer cells were full of a sugar they couldn’t use….

  7. MTK says:

    I conduct MALDI-MS often and never fail to get a kick out of watching the matrix crystallize a minute or two after the solution is deposited on the spot. The particular matrix I use the most forms needles and seeing it start and spread is rather mesmerizing.

  8. myma says:

    I have been away from the bench for many many years, but I shall never forget watching n-bromosuccinimide recrystallizing in a giant beaker. The little baby crystals riding a bubble of gas up, popping at the top, then floating back down, all while the solution color was slowly changing from colorless to yellowy to orangey.

  9. Alex says:

    I think Peggy Etter said it best:

    “The scientists’ role in the crystal growth process is that of an assistant who helps molecules to crystallize. The scientific challenge is to learn how to intervene in the process in order to improve the final product.” – M. C. Etter, in Preprints Inf. Workshop Cryst. Growth Org. Muter., Tokyo University,
    Tokyo, 1989, p. 1.

  10. shard says:

    Too bad most chemists can’t do it. I remember once…my boss, an “esteemed” PhD out of Berkeley(top lab LOL), spent the whole day trying to crystallize a simply heterocycle.
    He spent the whole day wondering why he couldn’t get a small amount of intractable material to keep “crystallizing” to yield his product.
    He eventually gave up and went home.
    I filtered his solution…stripped it down on the rotovap and harvested the crystal, clean as can be in good yield, right from the flask as the solvent volume reduced.
    I put the bottle of dried white powder with the NMR on his bench for him to find in the morning.
    He comes in and says….”what happened to my reaction” I just pointed to the bottle and NMR and said nothing. He never said a word about it to me.
    I quit working for that place shortly thereafter, just another group of mediocre hacks trying to bad mouth me cause I made them look stupid in the lab.

    Investors should pay attention though, all to many startups are staffed by such clowns, they are just trying to dupe you and steal your money.

    It turns out that that particular bay area startup fleeced investors…..good. Absolutely nothing every came from the place.

  11. An Old Chemist says:

    About obtaining crystals, Professor Phillip Eaton, who was the first one ever to synthesize the first cage compound ever, cubane, told this story in his class, at the university of Chicago. There once was this chemist who managed to obtain good crystals of a certain compound which no one else had been able to do despite many arduous attempts. No one could figure out the reason, but the reason could simply have been that he happened to have a long beard (Derek take note!), and often stood on top of his flask wondering, while scratching his beard, why crystals were not forming. Likely, something fell from his beard into the flask and acted as the nuclei for the crystals to form. Prof Eaton also suggested that if you leave the flask open to air, and hence to dust particles, then crystallization can get aided by the dust particles falling in and acting as nuclei.

  12. Retro Sinner says:

    Of course, we do not want dust in our product. For an API, a polishing filtration occurs prior to crystallisation to avoid such contamination and seeding is preferred to induce nucleation. The batch is also stirred to aid uniformity in the crystals unlike most academic preps. One can always question how you get your seed crystals in the first place but any process chemist worth their salt is part magician.

  13. Clare says:

    Well THATS crystsl clear.

Comments are closed.