Here’s more evidence of the power of the MicroED electron diffraction technique: this new paper reports the structure of two reactive organometallic species whose structures could not be determined by either NMR methods or conventional X-ray crystallography. One of them is the zirconium hydride species known as Schwartz’s reagent (zirconcene chloride hydride) – its problem is that it’s not soluble in nonpolar solvents (no NMR, no crystal growth solution) and it reacts with most anything polar (including things like dichloromethane).
But the authors found by examining a commercial sample of the reagent that it had microcrystalline domains in it that were perfectly suited for electron diffraction, and they obtained a data set at 1.15-Angstrom resolution. Importantly, you’ll note that the hydrides are visible in the data – one advantage of electron diffraction is that it can pick up atoms like these which are often lost in the X-ray data due to poor signal/noise. Another feature to note is that this structure (and the others in this paper) were obtained at room temperature and with very low-flux electron beams (to avoid damage to the samples). The high signal/noise of electron diffraction lets you get away with this protocol as well, which opens things up to both ease of experimentation and to samples that otherwise would yield no useful data.
That’s illustrated by the second structure in the paper, which is a reactive intermediate formed when a dimeric palladium reagent is exposed to ethylene gas. The authors had noted this reaction producing a bright yellow precipitate, but could not determine its structure due to the same combination of insolubility in most solvents and reactivity in others. But a sample of it, as prepared by precipitation, again had actual crystals scattered through it that were large enough for structure determination (but so small as to be totally hopeless for X-ray work). It turned out to be an unusual insertion product whose structure would have been basically impossible to prove in any other fashion. Both of these structures were obtained by direct methods, and without the need for molecular replacement or any other adjustments to the collected data. The paper goes on to apply the right-out-of-the-container ambient-temperature method to several other organometallic compounds.
Now, the crystallographers in the crowd will take a look at the R-values and goodness-of-fit on these data sets and not be that impressed. They’re not terrific (R1 of 11 to 16% in the various structures reported here) but when you overlay the microED structures of the other complexes where there are known X-ray structures, the agreement is very good. And the data are certainly good enough for unambiguous structure assignment, which is the main consideration here.
This illustrates the interesting gray area where such work lands. I find that chemists are quite enthusiastic about the possibilities of this technique, but both crystallographers and biologists can be more puzzled by it. Those two groups are looking at structure determination from very different perspectives, but they both tend to underestimate the number of times that chemists honestly do not know the structure or even the identity of the compound they have produced. Biologists are used to thinking of diffraction-derived structures in terms of ligand-bound protein binding sites, so when they see small-molecule single-crystal work they’re always asking “But will this be the same conformation that it has when it’s bound to the target? If it isn’t, why do you care?” And that’s because they don’t realize that we’re not talking about anything even so refined as conformation, we just want to know what atoms are attached to what, because we really don’t know. Meanwhile, the crystallographers (the most precision-oriented of all analytical chemists, absolute falcons when it comes to data quality) are used to being able to draw firm conclusions from precise atomic distances and bond angles, and can wonder at a synthetic chemist’s indifference to these refinements in favor of just knowing what atoms are attached to what. So crude! But still extremely useful, and more often than people outside of synthetic chemistry realize.
So as one of those synthetic chemists, I say bring on the electron beams. Scrape stuff out of flasks and smear it on the grid. You can find tiny flecks of real crystalline material in there, less than a micron wide, and that’s all that you need to get the data that can tell you without a doubt what atoms are attached to what. It’s a great thing.