This research area has emerged from the group’s need to make samples on the order of 100 nm thick for studies using the above electron source technology (to make the “molecular movie film”). However, the first work on this length scale was directed towards the development of ultrathin pathlengths of liquid H2O for 2D IR studies as discussed above. The first nanocell developed by the group involved sandwiching two standard SiN windows together. It involved simply placing a drop of water on one window and using the other window as a “hat”. Evaporation closed the gap and the cell had to be quickly sealed. The first cell (RJDM’s last time in the lab) worked the first time... unfortunately. The success rate in assembling cells was not very high and it could be frustrating in the wee hours of the morning trying to change samples. The initial success showed that 100 nm pathlengths of liquid water could be created and led to concerted effort to come up with cell designs based on a thorough understanding of nanofluidics to help scale to ever smaller dimensions.
One might expect that it would not be possible to have liquids enter into a channel only 100 nm or so in dimension. However, this is not the problem. Consider capillary flow on the 10-100 micron scale. Trees can pull water up 100’s of meters via surface adhesion forces (and evaporation to give flow). On the 100 nm scale, there are no viscous drag forces to the flow. It is all surface adhesion. Liquids “roar” into nanocells. The real problem is that for electron probes the windows containing the liquid must be on the order of a few 10’s of nanometers thick at most as to not give excessive background scatter (such as trying to see through snow). The liquid flow gives rise to enormous instabilities in the windows’ positions. In addition, the windows tend to bow under the pressure differential. Electrons require high vacuum conditions to the sample, and from there to the detector, to not lose information in the electron scattering process from environmental background. The problem is further compounded by the fact that femtosecond electron diffraction or optical studies require viewing areas of approximately 100 microns in diameter. Imagine the compliance of a piece of material that is on the order of 10 nm thick by 100 microns in diameter (factor of 10,000). The rapid flow into nanoconfined regions with this extremely small compliance leads to violent motion of the windows that create havoc in the electron transmission. The physics is similar to the nonlinear processes giving rise to violent flag flapping in a strong wind. We discovered that by using negative feedback, like noise eater head phones, with the right bandwidth, we could damp the unstable motion of the windows to get laminar flow conditions with pathlength variations of less than 1 nm (determined optically in the 2D IR setup). This development enabled the first direct observation of the hydrogen bond network of H2O under fully resonant conditions as discussed (2D IR, Cowan el al Nature 2005, Kraemer et al PNAS 2008).
More recently the group has developed nanofluidic cell concepts in which the scale of the liquid pathlength has been reduced to as little as 45 nm with 200 micron areas of view. The SiN windows have been reduced in dimension to 10 nm (with 5 nm possible) so that the overall cell pathlength is on the order of 100 nm. This new cell concept has enabled the direct in situ viewing of nanoscale dynamics using electron probes. The design included well defined hard spacer regions for liquid flow with special contoured entry channels into the cell window region to avoid turbulence. These design concepts in addition to a novel liquid transfer arm to introduce flowing liquids into conventional transmission electron microscopes has now enabled high throughput, in situ, imaging with electrons. We have demonstrated nm spatial resolution with dynamic resolution currently limited only by our detectors. See our early work Mueller et al J. Phys. Chem. Lett 2013 and movie below to appreciate the new imaging modalities.
It is now possible to contemplate imaging biological molecules under physiological conditions. Imagine imaging protein unfolding/folding or DNA unwinding with nm resolution to observe all the correlated motions. This is now possible. In addition, this development has opened up the direct observation of atomic motions for solution phase reaction dynamics using diffractive imaging. Since most chemistry (and all biology) occurs within solution (aqueous) phase this development holds great potential to significantly impact our understanding of chemistry – at the atomic level of detail in which the bath/liquid comes under direct scrutiny.
In addition to nanofluidics, the group has become expert in nanofabrication methods for making specially designed support structures to massive arrays of nanofabricated sample arrays to provide the “film” for our work on atomically resolved structure dynamics. This work has led to a new concept using dynamic liquid wetting in which control of surface properties has enabled the self assembly of up to ~1M-pixel arrays of micron to nanoprotein crystals for high throughput structure determination. This work was primarily driven by our desire to make “molecular movies” of protein functions, to directly resolve the structure-function correlation in biological systems at the molecular level, which, as discussed, is the overarching research objective of the group. One typically needs 100 different crystal projections (using x-rays) and 100 time points to fully capture the atomic motions of interest. In this case, the minimum basis of crystals is on the order of 10,000. This enormous sample load seemed to be impossible but by going to micron to nanoprotein crystals, this problem has now been solved ( see Buecker et al Nat. Commun 11, 996 (2020) ). As it turns out, the original motivation for this chip concept (biodiagnostics) is being supplanted by the use of this M-pixel crystal chip concept for high throughput protein structure determination in which the minimum amount of sample is needed. The amount of sample needed to resolve a structure becomes a major issue with precious protein samples. The key feature of Serial NED is that it eliminates electron induced damage by fractionating the dose in time to see the onset of damage and to distribute the electron dose over thousands of crystals. Effectively, this development represents the equivalent of a Table Top XFEL but based on the much higher scattering cross sections of electrons (105-106 100KeV electrons relative to hard x-rays). The very first protein structure resolved using Serial NED was significantly higher resolution than obtained with XFELs.
The major advantage of Serial NED is that orders of magnitude less material is needed. All chemical synthetic routes need structures to verify the final product with x-ray diffraction being the main method used for current scales of chemical synthesis. This approach requires gram scaleups to give enough material for macrocrystals needed for x-ray diffraction/structure determination. Serial NED promises to reduce the amount of needed sample by orders of magnitude and could very well herald truly Lab-On-A-Chip synthesis to dramatically reduce costs, environmental costs, and increase synthetic approaches/structure determination by orders of magnitude.
Current research in this area is focusing on nanofabrication of sample delivery systems for Serial NED.