https://doi.org/10.1016/j.xcrp.2022.100772
“Direct imaging of the dynamics of polymeric assemblies in organic solvents is an outstanding challenge. Herein, we apply liquid cell transmission electron microscopy (LCTEM) to study polymeric nanomaterials in organic solvents. LCTEM is distinct from other TEM methods as it can be applied to characterize the morphology of nanomaterials in organic solvents. To enable this demonstration, we examined electron-solvent interactions for two common organic solvents, N,N-dimethylformamide and methanol, and compared these solvents to water. For each solvent, we developed Monte Carlo simulations and kinetic radiolysis models, providing scattering and chemical insight, respectively. Guided by theoretical results, we applied LCTEM and postmortem mass spectral imaging of poly(styrene)-b-poly(4-vinylpyridine) assemblies in each solvent. Then, a worm-to-micelle transformation in poly(styrene)-b-poly(4-vinylpyridine) was triggered via organic solvent mixing during LCTEM, enabling an experiment not possible through a cryogenic TEM time series. Our work provides a pathway for an expanded examination of nanomaterials in organic solvents via LCTEM, a neglected research area despite the obvious prevalence of such materials across chemistry and materials science.”
”
To develop a GEANT4 simulation representative of LCTEM conditions, a literature precedent employing GEANT4 to model a scanning transmission electron microscope (STEM) was amended.34 In brief, the amended Monte Carlo simulation incorporates an electron source operated at 300 keV, as is typical for an LCTEM microscope,18,24,33,35,36 and the underlying physics accounts for multiple scattering under the penetration and energy loss of positions and electrons (Penelope) system.37 The three-dimensional sample geometry mirrors the typical liquid cell window size of 50 × 50 μm in the x and y dimensions (Figure 1). Given the variability of the liquid thickness across liquid cell experiments, we simulated liquid cells with liquid layers between 100 nm and 10 μm containing DMF, methanol, or water (Figure 1A). Each simulation was repeated three times with one million primary electrons. To assess electron-solvent interactions for each solvent, the average number of secondary electrons (SEs), the average absorbed energy, and the average absorbed dose were measured by a detector placed in the center of the sample geometry (Figures 1B–1D and S1–S3).
Figure 1. GEANT4 Monte Carlo simulation setup and results
(A) Liquid-cell assembly with the inset showing the geometric setup of a simulated liquid cell.
(B–D) (B) Average secondary electron generation, (C) absorbed energy, and (D) absorbed dose for methanol, N,N-dimethylacrylamide, and water over liquid layers in the range of 100 nm to 10 μm.
Across the examined liquid thicknesses, the average number of SEs is close for water and methanol, while DMF slightly surpasses these solvents at higher thicknesses (Figure 1B). The increased SE generation in DMF may be due to the comparatively high molar concentration of DMF (77.4 mol L−1) relative to methanol (40.5 mol L−1) and water (55.5 mol L−1), the pure solutions of which these systems approximate. Given the increased SE generation in DMF, the simulated results indicate that the contribution of SEs to solvent damage should be highest for DMF, whereas water and methanol should be similarly affected. For all three solvents, the average number of SEs generated increases with increasing thickness and, at lower thicknesses, the difference in SE generation for the three solvents is minimized. However, given the similarity in SE generation for all three solvents at the low liquid thicknesses most relevant for LCTEM (<1 μm), the simulated results suggest that SE generation contributes similarly to electron beam-induced damage in all three solvents.38
On the other hand, the absorbed energy is highest for water, followed by DMF, and finally methanol (Figure 1C). Since water has the highest density, electrons penetrating through the sample lose more energy compared to less dense solvents. Likewise, for thicker samples, more energy is lost when electrons travel through the sample, manifesting in an increase in the absorbed energy (Figure 1C). Similarly, the absorbed dose is highest for water, followed by methanol, and then DMF (Figure 1D). The absorbed dose results can be rationalized by considering the elemental composition of each solvent, particularly the percentage of oxygen, the heaviest constituent element. Water is roughly composed of 88% oxygen, compared to 50% for methanol, and 22% for DMF. Since the absorbed dose measures the energy absorbed per unit mass, the results correlate with the percentage of oxygen, thus resulting in water having the highest absorbed dose and DMF the lowest. As with the absorbed energy and number of SEs, the absorbed dose increases with liquid thickness when each solvent is examined individually (Figures S1–S3).
Taken together, the Monte Carlo simulation results suggest that water should undergo the most damaging electron-solvent interactions. For all three solvents, the number of generated SEs does not vary greatly, specifically at the low liquid thicknesses typical for LCTEM (100 nm to 1 μm).39 However, for all liquid thicknesses, both the absorbed dose and energy are far higher for water compared to the two organic solvents. Ultimately, these results indicate that, from a scattering perspective, aqueous solutions are likely more prone to electron beam-induced damage. Correlated with this increased electron beam-induced solvent damage, the more strongly scattered electrons would be expected to lead to lower image contrast compared to organic solvents when attempting to image low contrast solvated organic materials, pointing to the value in expanding LCTEM to the study of organic nanomaterials in organic solvents.
“