Dynamic behavior of nanoscale liquids in graphene liquid cells revealed by in situ transmission electron microscopy.

Recent advances in graphene liquid cells for in situ transmission electron microscopy (TEM) have opened many opportunities for the study of materials transformations and chemical reactions in liquids with high spatial resolution. However, the behavior of thin liquids encapsulated in a graphene liquid cell has not been fully understood. Here, we report real time TEM imaging of the nanoscale dynamic behavior of liquids in graphene nanocapillaries. Our observations reveal that the interfaces between liquid and gas bubble can fluctuate, leading to the generation of liquid nanodroplets near the interfaces. Liquid nanodroplets often show irregular shape with dynamic changes of their configuration under the electron beam. We consider that the dynamic motion of liquid-gas interfaces might be introduced by the electrostatic energy from transiently charged interfaces. We find that improving the wettability of graphene liquid cells by ultraviolet-ozone treatment can significantly modify the dynamic motion of the encapsulated liquids. Our study provides valuable information of the interactions between liquid and graphene under the electron beam, and it also offers key insights on the nanoscale fluid dynamics in confined spaces.

Herein we present an in situ TEM study of the dynamic behavior of aqueous solution encapsulated between two graphene sheets in a graphene liquid cell. The bubble formation and nanoscale fluctuations of liquid-gas interfaces are observed. We find that the dynamic motion of interfaces can lead to the formation of liquid nanodroplets near the interface and inside the gas bubbles. These liquid nanodroplets often show irregular shape with dynamic changes. We find that improving the wettability of graphene liquid cells through ultraviolet-ozone treatment has drastic effects on the dynamic behavior of liquid. This study facilitates fundamental understanding of nanoscale liquid dynamics during liquid cell TEM measurements, and it demonstrates in situ liquid cell TEM as a promising tool for studying nanofluidics.

Fabrication of graphene liquid cells
Graphene liquid cells containing aqueous solutions were prepared by encapsulating each solution between a pair of graphene-coated TEM grids. First, multilayer graphene sheets on copper foil were transferred to Quantifoil film TEM grids (Ted Pella Inc., US) by the modified direct wet transfer method (Regan et al., 2010). Specifically, multilayer graphene sheets (~3−5 layers, ACS Materials, US) synthesized by chemical vapor deposition on copper foils were used in this study. Quantifoil carbon film TEM grids were placed on top of multilayer graphene on copper foils. Several drops of isopropanol (>99.5%, Sigma-Aldrich, US) were introduced between the Quantifoil carbon films and graphene sheets and dried under mild heating (80 o C) to enhance the adhesion of graphene to the TEM grids. The copper substrate was etched by sodium persulfate (<98%, Sigma-Aldrich) solution (50 mg/mL in Milli-Q water) for ~8 hours to completely remove copper from the surface of graphene. The resulting graphene-coated TEM grids were mildly washed with Milli-Q water at least three times. Finally, graphene liquid cells were obtained by encapsulating the aqueous solution between a pair of graphene-coated TEM grids. The aqueous solution was prepared by dissolving NaCl in Milli-Q water (~3 wt%).
The addition of salt is expected to result in the increase of surface tension, which decreases the wettability of the water solution on hydrophobic surfaces (Zisman, 1964;Nayar et al., 2014).
The relation between the wettability and the liquid dynamics is discussed in the Section 3.3. A small amount of the solution (< 0.1 µL) was placed on the graphene side of the as-prepared graphene-coated TEM grid. Another graphene-coated TEM grid was placed onto the liquid solution dispensed on the first TEM grid and the excess solution squeezed out was removed by a piece of filter paper. The top and bottom TEM grids were held together by tweezers for more than one hour, which resulted in the formation of many isolated liquid pockets in the graphene liquid cell.
Comparison experiments were performed using surface modified graphene liquid cells. The ultraviolet-ozone (UVO) treatment was conducted using a Jelight UVO-Cleaner® Model 42.
Graphene-coated TEM grids were treated by UVO for various exposure times (30, 90, 150, 300, and 600 s) before the fabrication of liquid cells.

TEM analysis
The graphene liquid cells were characterized by a JEOL JEM-2100 transmission electron microscope with a high-resolution pole piece, a LaB6 filament and a Gatan Orius SC200 CCD camera. The operating voltage of TEM was 200 kV. The electron dose rate was controlled (100−500 e -/Å 2 s) to limit bubble generation which can interrupt experiments. In situ TEM movies were acquired at 2-5 frames per second; the frame rate for each movie is indicated in the supplementary information. For all the movies in the supplementary information, the playback speed is 4 times faster than their original speed. Movies were compressed into MP4 format for publication, but the original data were used for the data analysis. Area, center of the mass, and circularity of the objects of interest were analyzed by FIJI software (Schindelin et al., 2012). The circularity is defined as, circularity=4area/perimeter 2 . The value is 1.0 for a perfect circle and is decreasing as a boundary become irregular or shape elongates.
Electron energy loss spectroscopy (EELS) analysis was performed using a monochromated FEI Tecnai F20 UT operated at 200 kV at the National Center for Electron Microscopy within the Molecular Foundry in Lawrence Berkeley National Laboratory. The microscope is equipped with a Tridiem Gatan imaging filter and a double-focusing Wien filter acting as a monochromator below the field-emission gun. The spectra were acquired with an energy dispersion of 0.1 eV/channel and with an exposure time of ~4 s.

Other characterization methods
Raman spectra of graphene-coated grids were acquired by a Horiba Jobin Yvon LabRAM

Dynamics of liquid-gas interfaces and nanodroplet formation
We characterize the aqueous solution encapsulated in graphene liquid cells using TEM operated at 200 kV. In situ observation reveals the dynamic behavior of the nanoscale liquid between two graphene sheets (Movie S1 and S2). The schematic illustration of the overall process is described in Figure 1(a), and the representative images from each movie are captured and displayed in Figure 1  (400 e -/Å 2 s, which is more than twice compared to Movie S1 and S2). A series of representative TEM images, from the early stages of Movie S4, are displayed in Figure 2

Motion and fluctuations of nanodroplet interfaces
Another interesting feature is the dynamic behavior of the liquid droplets after their formation in the interior of a gas bubble. For example, in the mid-and-late part (68−86 s) of Movie S4 (displayed in Figure 3(a)), as the experiment proceeds, the highly deformed liquid part is finally detached from the outer liquid and forms the liquid droplet in the interior of the gas bubble.
TEM images of this liquid nanodroplet clearly show two boundaries, which can be attributed to the boundaries either formed by the liquid-gas interface contacting with graphene windows or maximum lateral size of the droplet. Schematics showing the cross section of the droplet is illustrated in Figure S6 for two possible models: the cylinder and the (hemi-)sphere model. The two boundaries are highlighted as red and blue dashed lines in Figure 3(a) and their contour plots are shown in Figure 3  In addition, the electron dose rate in our experiment is quite low (100−400 e -/Å 2 s).
Consequently, heating effects are negligible in graphene liquid cells. Secondly, we have checked whether the radiation pressure can exceed Laplace pressure of the droplets, or not. The Laplace pressure is determined from the Young-Laplace equation as follows: where l is the thickness of the liquid and is the mean free path of the 200 keV electron where the surface area (A) of the droplet is approximately estimated from the cylindrical model (i.e., l is the thickness of the liquid and Rdroplet is the radius of the droplet). The electrostatic energy of cylindrical droplets can be expressed by eq. (4): where q is charge and ε 0 is permittivity (8.8510 -12 F/m). So, the electrostatic energy exceeds the surface energy of the droplet when the portion of the charged water molecules is higher than the critical point. Similarly, the critical fraction is also estimated for the sphere model by simply modifying eq. (3) and (4). For both models, the calculated results are plotted in Figure   3

Impact of graphene wettability on the dynamic behavior of liquid
We investigate the impact of the surface wettability on the liquid dynamics. The change in total surface free energy (∆G) of the cylindrical droplet due to the movement (∆R) can be To determine the optimum UVO treatment condition, we measured the contact angle of the pure water on the graphene-coated grids with various exposure times ( Figure S8) and the result is summarized in Figure 4

Conclusions
In summary, we have studied the dynamic behavior of aqueous liquid encapsulated between two graphene nanosheets using in situ TEM. The morphology of liquid can change under electron beam exposure, which can be attributed to the bubble growth and the interface

Supplementary Information
Supplementary data associated with this article can be found, in the online version. of Movie S1, which is shown in Figure S1.