The sample must be maintained at cryogenic temperatures, presumably to make it stay put at the molecular level. Last but not least, they apparently need to rotate the sample stage to keep the sample from being blasted to pieces by the electron beam.
So unfortunately, the hardware isn't going to be easily packaged into something that consumers can afford.
If I had to come up with a drug-identification device, I would probably look into low-field NMR. You aren't trying to visualize the structure, right, just figure out what elements are present in what proportions?
> You aren't trying to visualize the structure, right, just figure out what elements are present in what proportions?
Elemental analysis is both rather easy (just burn it and measure out the CO₂/H₂O/NO₂/SO₂ + other impurities) and rather useless (a cyclohexane and a hexene both have the same chemical formula yet have very different structures and very different reactive potentials).
Inside chemical labs, NMR is the go-to method for identifying "is the compound in this jar what the label says it is?" But that's largely because every chemical lab has an NMR machine already, NMR makes all the functional groups pop out without much work, and if it's not in your database, you can still do a reasonable start on identification with just ¹H NMR. Considering the expense of NMR versus IR spectroscopy, database matching applications probably suffice with IR.
The nootropics community will love this, making it easy to verify the composition of the bulk powders ordered from India or China.