What’s another (kilo)gram?

Prior to my current post, I’d not given too much thought to scale up.  I suspect to the majority of early career synthetic chemists, “large-scale” is synonymous with breaking out the one-liter round bottom flask.  That’s pretty much the comfortable upper bound of what you can work with 1) on a benchtop; 2) with magnetic stirring; and 3) with a oil bath heat source.

Your efforts on this scale will yield somewhere in the ballpark of 100 grams of product, depending on formula weight and a slew of other variables.  And what’s more, purification and workup has now ventured into the realm of things that are no longer routine.  A one-liter reaction volume is going to require a rather large separatory funnel (as a side note, Chemglass sells them up to 22-L — good luck with that).  And unless your starting materials and product have wildly different silica affinities, you’re going to have quite a bit of fun trying to run a 100-gram flash column, so you’ll likely have to break it into a couple runs.

And that’s all great until you need to crank out a kilogram of material.  You can now forget about running things in round bottomed flasks (Chemglass also sells a 22-L round bottomed flask, a testament to the age-old adage “just because you can does not mean you should“).  You’re also not going to have much luck trying to fit a vessel that size onto a hot plate, so that rules out both magnetic stirring (which would be ineffective anyway) and conventional heating baths or mantles.

Things like efficient mixing and heat transfer — which we hand wave away at the gram-scale — start to matter quite a bit once you cross the kilogram threshold.  So you’re going to need a specialized, jacketed reactor, through which you can recirculate a heated (or cooled) thermal transfer media.  And because surface area to volume ratios are the way they are, the temperature gradient between the outside of the reactor and the inside can be pretty dramatic.  So you’ve really got to get things mixed well, which means you need motorized stirring and a decent sized impeller.

Next on your synthetic checklist is workup, which now takes an entire day in and of itself.  Pray you don’t need to purify anything chromatographically.  Your precipitation that required 10 ml of solvent X per ml solvent Y suddenly won’t fit in any container in the lab, save the 55-gallon waste drum.  I’m not ashamed to admit I’ve MacGyvered a workup involving a 5-gallon orange Home Depot paint bucket at a previous position.

All this, and I haven’t even touched on time yet.  Everything at the kilo-scale takes longer.  A reaction which you could comfortably set up in 20 minutes at the gram scale will take you all morning to get going.  And you’d best triple check your work here, as mistakes on this scale are costly.

Of course, the proper process chemists will scoff at the struggles of the kilo-scale.  Steel reactors replace glass, drum evaporators replace rotavaps, and somehow I doubt the tried and true paint bucket workup would pass cGMP muster.

Inventories of the chemical variety

I’ve been thinking a lot about chemical inventory systems recently.  Derek’s post yesterday made for an interesting read, especially the comments section.

Over the years I’ve seen most combinations of storage systems described there.  But what I’m much more interested in now is the software (or lack thereof) which runs these systems.  There’s no shortage of LIMS vendors out there.  There are even some free app-based inventory management solutions that I’ve played with (see Quartzy), which may work well for a small academic lab but lack functionality required by a modest sized company.

There’s a constant struggle, as others have pointed out, with compliance in any inventory management system.  The more work you require of a scientist to use the system, the more likely they are to ignore it entirely.  Barcoding, or more recently RFID tagging, attempt to alleviate some of the burden with logging and tracking materials.

But these attempts to automate inventory management all seem to suffer from what I’ll dub the problem of granularity.  The granularity problem is simply that for an inventory management system to be useful*, it must be sufficiently granular to describe the location of a material with both precision and accuracy.  In other words, a system which correctly identifies the location of a bottle of pyridine as “Chemistry lab 1,” is not precise enough to be useful.  Similarly, a full site map placing that bottle of pyridine in Bin 1, on Shelf 2, in Flammable Cabinet A, in Chemistry lab 1 is only useful if John Smith hasn’t used the last of it and forgot to remove the container from the inventory.

One might envision a system in which each storage location is fitted with an RFID reader, and each reagent bottle tagged with chip (which cost less than a quarter each now).  This system would be able to identify where exactly each reagent is simultaneously, provided it’s within range of an RFID reader.

And indeed, it seems like something like this has been done at least once.  The issue I envision with such a set up is one of granularity — you can’t practically put a reader in each bin on each shelf of each cabinet in your entire facility.  That being said, such a system would probably be able to distinguish whether or not a particular reagent is in the proper flammable cabinet, or if John Smith moved it to his fume hood.

My searching hasn’t uncovered any turnkey solutions involving RFID chemical tracking — but it must be possible if not feasible.  After all, manufacturing operations and logistics companies have been employing this sort of technology for years.

Readers, what’s the best inventory system you’ve seen employed?  Have you ever seen a system that manages to solve both the granularity problem and the compliance problem simultaneously?


*There are inventory management systems that are not useful, and serve simply to allow administration to be in regulatory compliance.