Skip to Content

Analytical Chemistry

All The Way Down to the Hydrogens

I’m always happy to see new techniques for resolving structures of molecules (large and small) and figuring out their behavior. Even if the latest instrument or method doesn’t seem to have any bearing on what I’m doing, or even on drug research in general, it could end up helping. It’s Sydney Brenner‘s 90th birthday today (!), and one of his famous quotes relates to this point: “Progress in science depends on new techniques, new discoveries, and new ideas, probably in that order”. Freeman Dyson has made the same point, most recently in his book The Sun, The Genome, and The Internet, saying that contrary to popular opinion, many scientific revolutions are tool-driven rather than concept-driven.

This paper has me thinking about such issues today. The authors report using electron diffraction to locate hydrogen atoms in small-molecule crystals, and it’s pardonable to wonder what the big deal about that might be. Don’t we pretty much know where the hydrogen atoms are? After all, when you look at a small-molecule X-ray structure, the hydrogens tend to be an afterthought, usually dealt with by a phrase like “hydrogen atoms were fitted using a riding model” or such. But that’s because X-ray diffraction is nearly blind to hydrogen atoms. They hardly diffract. Just because we can’t see them doesn’t mean that they’re not important, though. We use X-ray data to tell us about all the heavier atoms in a crystal, and it’s great stuff, but what are we missing?

Electron diffraction is XRD’s cousin, but it has its tricky aspects. Recent advances in instrumentation and (especially) processing of the data, though, are making it more and more useful (here’s more). I’ve written about a few very impressive structures that have been generated of proteins by electron diffraction, and that first link brings up another very useful aspect of the technique: it can work on ridiculously small crystals. If you’re going to do X-ray diffraction, there’s a size limitation (even on a bright source like a synchrotron, although that’ll certainly help). This new paper demonstrates a great improvement in handling electron diffraction from hydrogen atoms.

It’s demonstrated first by analysis of a very small single crystalline region of an acetominophen (paracetamol) sample, about 500 nm wide. The method nails the hydrogen positions, including the NH and the OH of the phenol. This structure was already worked out, and confirmed by this route, so they then went on to an unknown structure, a zeolitic cobalt aluminophosphate made by a new synthetic technique. X-ray powder diffraction established the general parameters (unit cell, space group), but also revealed that the cobalt atoms were only partially occupying the positions one would have normally assigned. This kind of thing happens a lot, and it can be a real pain to work out – and it is, as fate would have it, often closely related to catalytic activity and other important properties of such solids.

The electron diffraction work on this one was not without trouble – the first data sets they collected showed signs of the structure having been slightly deranged by the electron beam itself, which is a known problem (these are pretty high-energy electrons). Lower-dose data were more reliable, and the hydrogens were tracked down, although it still took plenty of work. This is really nice to see, though, because in these sorts of solids, X-ray data alone are not going to tell you whether that’s an OH or an O-minus coming off of a given metal atom, if something is an acetic acid or an acetate, a water molecule or a hydroxide, and so on, and these can be (as you’d imagine) rather important distinctions. If we’re going to understand the properties of advanced solid-phase samples, which include hugely important industrial catalysts, photovoltaics, superconductors, and structural materials, we’re going to have to track down every last atom. It looks like we may finally have a way to do that.

23 comments on “All The Way Down to the Hydrogens”

  1. Hap says:

    Can it distinguish between different kinds of heavy atoms more reliably? Diazonamide A was misassigned in part because the X-ray structure misassigned the heteroatoms in the fused heterocycles at the bottom. Getting more reliable assignment of structures would be nice.

    Could you use electron diffraction with the “crystal-free” crystallography using molecular sponges, or is there likely to be too much disorder?

    1. James says:

      Hi Derek, Neat post, I agree that the advance of tools is what drives most fields, chemistry, physics, biology, etc. With better tools you can see more and the more you can see the more you can *hopefully* understand.

      I do have an erratum for the post though. You wrote “500 nM”, I am pretty sure that ‘nM’ is nanomolar if I remember things correctly. Since that is not a size I think you meant ‘nm’ which would be nanometer.

      1. Derek Lowe says:

        Ugh, that’s completely correct. Fixing it now!

    2. James says:

      Sorry, didn’t mean to click reply on your post. *hangs head in shame*

    3. Derek Lowe says:

      I’ve wondered about that last question, too. Right now, the technique is so experimentally demanding that I’m not sure if anyone’s even tried that combination, but it would be very interesting to see.

    4. anon says:

      I am sorry but enough attention, good experiments (1H/15N-HSQC, COSY etc.) and good communication between x-ray experts and chemists could have caught it.

    5. anon electrochemist says:

      No, the structure factors for the x-ray and electron scattering are about the same. If you want better discrimination between similar atomic number, you need to use resonant x-ray diffraction (or neutrons). It’s like the Se labelling and MAD phasing for proteins, originally developed back in 1925 by Szillard (Z. Phys, 33, 68). Nanofocus synchrotron beamlines are now well-established, and can run diffraction using soft x-rays.

      The biggest hold-up by far is researchers willing to spend the time to learn and use the tools that we’ve developed.

      1. Chemperor says:

        I’d also add that a big holdup is the cost associated with acquiring these fancy new toys. Instruments and techniques that are accessible only to R1 universities, national labs, and big companies are going to be of limited broad utility until they become accessible to the masses. For example, most PUI’s now have access to NMR of some kind, even if it’s just a permanent magnet 90 MHz instrument. I’ve seen many small teaching colleges with SEM capabilities as well, and I’m seeing more and more benchtop single xtal X-ray diffraction units being deployed. Students who grow up using these instruments and who gain an appreciation and familiarity early in their training are more likely to seek out the next level of sophistication later in their careers.

  2. mallam says:

    “If you build it, they will come.”

  3. CMCguy says:

    Is there a Chicken and Egg argument here? Since would suggest most the new tools and technique are spawned by people seeking ideas on how to do something better, differently or because nothing currently exists to address a particular line of inquiry. Therefore yes once those become established and available tools greatly aid and accelerate discoveries and encourage expansion into other avenue however the development of new tools and techniques requires certain ingenuity. With everything being intertwined, largely increment advancement build off previous knowledge not sure one can make distinctions on relative starting points and importance in any stage in the process.

    1. Fred says:

      Not much of chicken and the egg problem, more one of a chicken and the grain field problem, with some chicken and the egg thrown in.

      James Burke’s “Connections”, the 1978 series rather than the later ones, might cover much of what you’re looking for in an answer. The last episode would be a focus if you are time constrained. But, the whole series is worth the binge watch.

  4. myma says:

    This is really neat. It will be even cooler if/when they get to do proteins. I remember back in the day our protein modeler guy spending days “cleaning” the pdb structure, and even more days populating/checking likely hydrogen positions and protonation states.

  5. Aole says:

    Did they find Waldo, or just the hydrogens so far?

  6. oldnuke says:

    Here is some information on electron diffraction work underway at LBL:


  7. skeptic says:

    Does anyone know offhand what the current requirements (sample size, time) for neutron diffraction are?

  8. Dr. Lloyd T J Evans says:

    I thought that neutron diffraction analysis could already be used to nail down exactly where hydrogen atoms are in a crystal? That is not without its disadvantages though. You need to grow much bigger crystals than those used for X-ray crystallography. But you also need a neutron spallation source or a nuclear reactor. Not many crystallography departments have access to either of those.

    1. zero says:

      Perhaps not, but a fusor is a benchtop device capable of generating neutrons a’plenty. It’s a simple device (few/no moving parts) , but with complex supporting equipment (vacuum pumps, high-voltage supply, low-pressure deuterium supply) and radiation shielding requirements. I’m guessing there isn’t a commercial device on the market that could ‘plug and play’ into neutron diffraction equipment, but with a bit of applied engineering one could put together all the necessary equipment.

      1. zero says:

        (then I read the rest of the comments and found that you had already addressed fusors… sorry about that.)

  9. gippgig says:

    Shouldn’t that be acetaminophen?

  10. Anonymous says:

    Reaction of dimethyl sodio-3-ketoglutarate with glyoxal and substituted glyoxals: First expeditious preparation of bicyclo [3.3. 0] octane-3, 7-dione…, Bertz, Rihs, Woodward, Tetrahedron, 1982, 63-70, reported “The X-ray crystal structure of the indanone reveals a novel hydrogen bonding phenomenon.” I think it was claimed to be one of the first X-ray structures where a crucial H was resolved. (It’s a Woodward paper, so I remembered it better than others.)

    I agree somewhat with the chicken-egg comment from CMCGuy. Technique-Discovery-Idea. A new discovery leads to an idea to make a new tool. The new tool enables experiments that lead to new discoveries that lead to new ideas. And continue the permutations. I think the big difference is in numbers: everybody uses a new technique, even some who do stupid, lousy experiments; a handful of those make actual new discoveries (often by sheer luck); a smaller handful come up with truly new ideas.

  11. MattC says:

    This is really impressive work and I’m glad it made it onto the Pipeline.

    I think the key thing to point out here is that this is really a supplement to powder diffraction rather than single-crystal diffraction. Single crystal diffraction will still be the gold standard technique for a long time as although electrons have a sensitivity gain compared to X-rays for hydrogen (scattering factors scale as ~Z^(1/3) for electrons, and Z for X-rays), the dynamical effects (electrons, unlike X-rays, tend to interact multiple times, so your diffraction patterns are much harder to interpret) along with difficulties in getting accurate intensities mean electron structures are probably not going to be the gold standard.

    SXD relies on you having a single crystal though – as Derek says, at a synchrotron you’re probably going need a crystal of at least 10 microns (ideally much larger). Some materials don’t even form micron sized crystals, and including many extremely useful materials.

    Minor point – ‘riding’ is a comment about how an atom was refined not located (briefly riding Hs means they move when the atom they’re bonded to moves). In the zeolitic structure in this paper they apply riding restraints (entirely sensibly). This is different to whether the hydrogens were found in the difference map or added geometrically. It’s actually pretty common for all the hydrogens to be located in a difference map in SXD, nowadays.

    some replies:
    Hap: electron diffraction will be less sensitive to C-N differences in general (hence why it can see H a little more easily). I don’t know about that particular problem though.

    re: neutron diffraction – typical samples would be about a mm in linear dimension, and your collection would probably be about 24hrs. This is very dependent though on the crystal quality, beamline, etc. As Lloyd says, no university has a neutron single crystal instrument – but there are maybe half a dozen which you can apply for time at. If you’re just trying to find Hs in a structure that you have a good X-ray structure for, you can probably work out what’s going on using powder diffraction though (where 100mg of sample would do.

    1. Derek Lowe says:

      Thanks! Neutron diffraction is something I know even less about – I can imagine that good sources of high-energy neutrons aren’t so common, or fun to be around. And how do you “focus” them?

      1. Dr. Lloyd T J Evans says:

        As for focusing neutrons, generally you can’t. A nuclear fission reactor neutron source is simply collimated. When operating, the reactor is producing a high neutron flux in every direction. The reactor basically has an intentional weak point in the shielding which allows neutrons to come out in one place. These are channeled down a beam line long enough that only neutrons travelling within an extremely narrow directional cone are allowed to hit the target.

        A spallation source is somewhat better controlled. These work by firing a high velocity proton beam at a mercury target, which then spits out neutrons. The proton beam can be focused in one direction, so a decent percentage of the spalled neutrons are emitted in more or less the same direction. They still need to be collimated down a beam line though, so a lot of the neutrons are wasted to produce a fine beam.

        A rather easier device to make is the fusor neutron source – some people have even built these in their basements. This is basically a miniature fusion reactor, firing deuterium ions at each other. These then fuse and spit out helium-3 and neutrons. Despite achieving fusion, these cannot be a net power source, because more energy is used for accelerating the deuterium than is produced by the fusion reactions. Again, because the ions can be focused, there is some degree of control over the direction of the emitted neutrons. However, the rate of neutron production by a fusor source is nowhere near high enough to do neutron diffraction crystallography with.

Comments are closed.