Performing In Situ Closed-Cell Gas Reactions in the Transmission Electron Microscope

Specimens for MEMS-Based Closed-Cell Gas Reactions:
Direct powder deposition by drop casting from a colloidal solution and through a mask
Depending on the material to be studied, there are a number of different ways to prepare E-chips for in situ/operando CCGR-STEM experiments. Preparing the gas cell for catalysis studies typically requires dispersion of the catalyst nanoparticles onto the E-chip either from a colloidal liquid suspension (Figure 2A), or directly from the dry powder itself (Figure 2B). For coarser powders, it may be necessary to crush the particles (e.g., using a mortar and pestle or by placing the powder between glass slides), so the powder aggregates will fit within the 5 µm gap between paired microchips ("sandwich") size without damaging the Si_xN_y membranes. When using a liquid suspension, the deposition of powders results in wider dispersion covering a larger area of the E-chip, which often requires a secondary cleaning ("dusting") step to remove the powder from the gold contacts. Whereas, when depositing dry powder, a mask can be used to directly deposit powder in the desired location (e.g., the electron-transparent Si_xN_y viewing area). In our study, the masks we tested are E-chips with removed Si_xN_y membrane and liquid-cell spacer chip with removed Si_xN_y film. Since the latter one has a narrower opening (50 x 250 µm), a more precise deposition can be achieved directly onto the membrane heater region of the E-chip, and no additional cleaning of the gold contacts is necessary.

Pattern mask and alloy deposition
Deposition of the catalyst on the E-chip is relatively easy when compared with bulk alloys. Since nano-sized particles of random alloy compositions are not readily available and crushing alloy micro-size powders has also been problematic⁶, the evaluation of one more potential new method was addressed for producing alloy specimens of controlled composition and geometry onto gas-cell E-chip membranes³³.

The basic idea for the structural alloy specimens is to deposit "islands" (Figure 2D) of the desired structural material using a suitable vapor deposition technique (e.g., electron beam evaporation, ion sputtering, or magnetron sputtering) where the elemental species are deposited directly onto the E-chip membrane (Figure 3A) through a pattern mask composed of an array of ~2 µm diameter holes (Figure 3). The pattern mask can be produced by FIB-milling techniques, using a Si_xN_y spacer E-chip. Alternatively, it is easier to use a commercially available 50 nm thick SiN Microporous TEM Window, with an array of 2 µm pores in a silicon nitride film in a grid pattern within a single 500 x 500 µm membrane (Table of Materials)³³ as a pattern mask (Figure 3B-b). As shown, it is possible to attach a SiN Microporous TEM Window to an E-chip with removed Si_xN_y membrane (Figure 3B-a) and place it directly on the E-chip (Figure 3C) in a securely tightened E-chip fixture (Figure 3D). This is used to create a perfect alignment of the devices and helps isolate the evaporated species to a small area on the E-chip (Figure 3C-c and 3C-d). Depending on the chemical composition of the depositing alloy/material, each evaporation technique (electron beam evaporation, ion sputtering, or electron magnetron sputtering) has its own advantages and disadvantages³³, which will not be addressed here. Therefore, the idea for gas-reactor sample preparation by the vapor-phase deposition through a pattern mask onto the E-chip surface has potential for further development and experimentation.

FIB milling
E-chip preparation becomes more challenging when investigating solid materials. Comparable studies of bulk structural materials require preparation of the sample as a thin slice or lamella of suitable specimen thickness and geometry (e.g., electron transparent, and a few microns in lateral extent) that can be secured in some fashion to the E-chip membrane. This process can be conducted using FIB-milling procedures and placing the TEM lamella on the Si_xN_y viewing area in the SiC heater membrane (Figure 2C-c)^9,30,36,37 with the caveat that conventional gallium FIB milling typically leaves residual Ga, either as Ga implantation and/or Ga segregation in some material systems (e.g., within grain boundaries and phases in Al and its alloys³⁸) on the milled surface, thereby complicating the reaction process when dynamic events need to be examined at the atomic level. It is essential to evaluate material susceptibility to Ga penetration⁹. To minimize Ga implantation and surface damage, we can use electropolished needles, similar to the ones used for atom probe tomography, which can then be placed on the E-chip (Figure 2C-d) using the FIB by attaching the sample by W or Pt "tack" points³¹. EDS analysis confirms that Ga implantation can be decreased/eliminated (Figure 2C-d); however, the limitation of this method is the geometry of the sample. Only needle-shaped samples can be prepared without exposing the area of interest to Ga ions. As an alternative, new Xe-plasma FIBs can be used to prepare thin lamella without Ga implantation. For example, electron-transparent lamella from 3-mm electropolished discs can be extracted and placed on the E-chip (Figure 2C-e) resulting in a large area of the sample with no issues associated with a residual ion implantation layer (Xe is inert and does not tend to deposit on sample surfaces. It also produces a thinner amorphous layer (~1 nm) than the best FIB procedures with a Ga source)³⁴.

In situ reaction experiments
In order to capture dynamic events, first, it is necessary to bake and pump down the system overnight. During the actual experiment, the holder is connected to the gas manifold system and is pumped and purged several times. The system is initially pumped down twice from 100 Torr to 0.5 Torr and purged with an inert gas (e.g., N₂, Ar); the third cycle involves pumping down to 0.001 Torr. The internal conditions are monitored by an RGA system (Table of Materials), which is equipped with an electron multiplier¹⁰. The RGA is integrated into the gas-control system through connection to the return side of the CCGR-TEM holder (Figure 4B). To remove residual water vapor and other gases from the RGA chamber, heating tape is used which permits for bake-outs between experiments. An ultra-high vacuum in the RGA of < 2×10^-8 Torr can be achieved. An electronically controlled leak valve (LV) is used to control the amount of gas from the holder and into the RGA chamber, and a return capillary line to the manifold is isolated from the leak valve with a hand valve (HV).

An example of the recorded gas partial pressures measured in the RGA chamber, CCGR-TEM holder (LV open), manifold (H1 open), and Tank 1 (T1 open) before in situ experiments is shown in Figure 4. This demonstrates that even though overnight baking, pumping down and purging were performed, there is still some degree of residual water vapor. Thus, for experiments particularly with water vapor, it is important to establish the baseline for initial conditions of the system and record the initial partial pressures. For our system shown in Figure 4, the partial pressure of water vapor measured all the way to Tank 1 reads 1.1 x 10^-7 Torr. The atomic mass spectrum versus partial pressure shows the water vapor peak at 18 amu, reaching 1.1 x 10^-7 Torr (Figure 3C). Comparing the spectrum with the one from the experiments that contained O₂ and water vapor, there is a significant increase in the partial pressure (2.5 x 10^-7 Torr) of the peak at 18 amu. Note that by further opening the leak valve, more gas flow is introduced into RGA chamber where measurements are performed. It is important to adjust the leak valve in such a way that the total pressure of the experiment is kept constant in order to compare results between conditions. Gas composition measurements are possible when the RGA chamber pressure is in the ≤10^-5 Torr range, which is less than a billionth of an atmosphere due to the high reactivity of ions and their short life; therefore, the pressures in the RGA are much lower than within the gas cell.

The attachment of the water vapor delivery system to the manifold requires purging the VDS with inert gas until no liquid is present (it is also important to clean the VDS right after the experiment to make this step simpler) and keep it purged during the connection to the manifold. Before the VDS is filled with the desired liquid (e.g., water, methanol, or ethanol), first, the VDS is pumped down to vacuum. Then, the liquid is added using a syringe and tubing. To improve the quality of the vapor (with reduced oxygen content) the experimental supply tank can be filled with the vapor and pumped down two or three times; otherwise, it is ready to be used.

The gas-control software guides the user through the settings during all phases of the experiment. In the beginning, the correct gases and pressures need to be selected. The resistance of the E-chip must be checked to ensure that the E-chip was not damaged during loading into the CCGR-TEM holder. In the manifold, there are two supply tanks (Tank 1 and Tank 2) that hold and supply gas with a final composition for the reaction. The desired gas composition can be obtained by mixing the media directly in one of the supply tanks (Tank 1 or Tank 2 in Figure 1D and 4B). The manifold system has three ports that introduce gases to the manifold. However, if more than three gases are desired to be mixed, one or more of the input lines need to be split. Alternatively, if the gas composition is very complicated, pre-mixed gases should be used, which allows their mixing during the experiment with the desired vapor composition.

After setting the desired gas composition for the in situ experiment, the gas-control software will first introduce the lower percentage gas; then after reaching the desired pressure, it will feed the second gas into the supply tank. Afterward, depending on the experiment, the gas can be introduced into the gas cell at either room temperature or after heating the sample to the desired temperature at a certain/desired heating rate. This depends on each user's experiment. The heating can occur in a vacuum, under inert gas, or under the premixed gas that will be used in the experiments. When the gas needs to be changed while running experiments, the system is pumped down and purged with inert gas to avoid any hazard of mixing two incompatible gases.

In general, there is little or no drift in the x and y directions during the experiments, but during heating and/or pressure changes, significant variation in the specimen height is observed (which poses a challenge to capture the initiation of a reaction). If possible, heat to the desired temperature under vacuum or inert gas, adjust all the alignments and then introduce the gas media. Experiments below 200 °C are also challenging with the closed-cell due to contamination build up on the surface of the E-chip viewing area.

As an example, the evolution of the surface of Pt nanoparticles on a TiO₂ support was captured when exposed to 100% water vapor at 17 Torr at 300 °C (Figure 5). The structural changes in the Pt particle and rearrangement of the structure to expose {111} surfaces (Figure 6) were observed (Figure 6A vs. Figure 6B vs. Figure 6C).

Figure 1: E-chip paired device with CCGR-TEM holder. (A) Pair of MEMS-based silicon microchip devices (spacer chip and E-chip (heater)) for in situ CCGR-STEM experiments. (B) Schematic of the CCGR-TEM holder tip with paired microchip devices being secured with a clamp. (C) Cross-section of the CCGR-TEM holder tip showing E-chip placed on top of the spacer chip creating the closed-cell (sandwich) that isolates environment around the specimen from TEM column. (D) Closer view of manifold that encloses three gas input lines on the side, two experimental gas delivery tanks, and a receiving tank for gas flow control during the experiment. (Images provided by the CCGR system manufacturer). Please click here to view a larger version of this figure.

Figure 2: Example of different deposition techniques for preparing samples on the E-chip. (A) E-chip with catalyst deposited by drop-casting from a colloidal solution. (B) E-chip after dry powder deposition using two different masks (a) E-chip with removed Si_xN_y membrane and (b) liquid-cell E-chip with removed Si_xN_y membrane. (C) E-chip prepared by (a) standard FIB-milling procedures and placement of FIB lamella on Si_xN_y electron transparent viewing areas, (b) electropolished needle, (c) electron-transparent section of a grain in a 3 mm electropolished disc extracted by Xe-plasma FIB and placed on E-chip. (D) Higher magnification image of E-chip with alloy deposited through pattern mask. Please click here to view a larger version of this figure.

Figure 3: E-chip preparation using sputtering techniques. (A) Schematic of the E-chip. (B) Pattern mask fabricated from liquid-cell E-chip with Si_xN_y membrane removed and 50 nm thick SiN Microporous TEM Window (B-a), with arrays of 2 µm pores in silicon nitride film in a grid pattern within single 500 x 500 µm membrane window ³⁴ that overlaps the 50 x 250 micron opening of the liquid-cell microchip (B-b). (C) Pattern mask directly placed on the E-chip (C-c) with higher magnification image showing the alignment of Si_xN_y viewing area with 50 x 250 µm opening in liquid-cell E-chip that is covered with SiN Microporous TEM Window (3C-d and also 3B-b). (D) E-chip cross-section within the fixture (D-e), top view (D-g), and (D-f) closeup view of the pattern mask in the E-chip fixture. E-chip fixture holds pattern mask placed on E-chip in secure manner during vapor-phase deposition. Please click here to view a larger version of this figure.

Figure 4: Measuring gas compositions using a residual gas analyzer. (A) Example of gas partial pressures measured in RGA chamber, CCGR-TEM holder (LV open), manifold (H1 open), and Tank 1 (T1 open) before in situ experiments. (B) Schematic of the gas-control software showing locations for RGA measurements before experiments. (C) Mass spectra generated in a vacuum before the experiment (red) with water vapor peak at 18 amu reaching 1.1 x 10^-7 Torr and during the experiment (blue) with a mixture of O₂ with water vapor showing an increase in the partial pressure for OH, H₂O, and O₂. RGA confirms the presence of water vapor in in situ closed-cell. Please click here to view a larger version of this figure.

Figure 5: Measuring water vapor content. (A) Schematic of the gas-control software showing an example of 100% water vapor introduced to Tank 1 with test parameters recorded by the gas-control system at room temperature before the experiment. (B) Gas partial pressure spectra acquired using the RGA before (red) and during (blue) reaction with 100 % water vapor at 17 Torr and 300 °C. Please click here to view a larger version of this figure.

Figure 6: Experimental results of water vapor exposure effects on Pt nanoparticle structure. (A-C) BF-STEM images showing the reconstructed surface of a Pt nanoparticle on TiO₂ support when exposed to 100 % water vapor at 17 Torr and 300 °C. Please click here to view a larger version of this figure.