Control of film thickness in liquid cell transmission electron microscopy depending on mechanical clamping conditions

Abstract text (incl. figure legends and references)

Introduction

Liquid-cell transmission electron microscopy (LC-TEM) is a rapidly developing electron microscopy technique employed to study the structure and dynamics of samples in a liquid environment [1]. The achievable spatial resolution is typically limited by elastic and inelastic electron scattering in the liquid layer and the enclosing membranes and ultrathin layers are therefore desirable.

In early silicon-nitride(SiN)-based LC-TEM sample designs utilizing large membrane areas, the thinnest achievable liquid layer was typically determined by the bulging of the viewing membranes and partially by the height of membrane spacers defining the liquid flow channel. In contrast, for novel LC-TEM systems with largely reduced membrane areas other secondary effects start to determine the liquid thickness.

Objectives

To characterize the achievable film thickness in a commercial liquid cell TEM sample holder, we analyzed in detail the impact of mechanical clamping conditions of the cell stack and the effect of relative positioning and orientation of the membranes in the stack.

Materials & Methods

In the utilized LC TEM holder (Stream, DENSSolution; for JEOL TEM), the liquid is encapsulated by two 50-nm thick SiN-membranes (200 µm x 12 µm area) with patterned Pt electrodes. For determining the thickness of the enclosed water layer, we used a mass-thickness-contrast method [2] and electron-energy-loss spectroscopy. Different experimental conditions were adopted, including variations in the liquid inlet/outlet pressure, flow speed, relative chip orientation and various designs for the mechanical components clamping the two membranes. The resulting film thickness is compared to numerical models.

Results

For the geometry of the DENS Stream holder, numerical calculations including bulging effects and spacer height, result in an estimated liquid thickness of ∿600 nm for an internal absolute liquid pressure of 1600 mbar. However, experimentally, we find liquid thicknesses of about 1500 nm, largely independent from inlet/outlet pressure and flow speed (Fig. 1). Changing the mechanical clamping conditions by different top lid geometries and additional foil spacers drastically reduces the film thickness to ∿900 nm (Fig. 1). Optical microscopy images of the assembled stack demonstrate that for the optimized clamping conditions chip bending is minimized. Curiously, we also observed that the bending is depending on the relative orientation of the rectangular membranes (cross/parallel configuration), putatively due to the difference in the mechanical properties of the assembly, and a generally lower thickness is observed in the parallel case. With these modifications and utilizing the inhomogeneous bending across the membranes, a minimal liquid thickness of ∿420 nm was found in the parallel configuration (Fig. 2).

Conclusion

We demonstrated that mechanical clamping conditions of the liquid cell stack can be the limiting factor determining the film thickness for small area membrane cells. With optimized clamping close to ideal liquid thicknesses are experimentally achieved.

References

[1] S. Pu et al., Roy. Soc. Open Sci. 7, 191204 (2020).

[2] H. Wu et al., Small Methods 5, 2001287 (2021).