A versatile nanoreactor for complementary in situ X-ray and electron microscopy studies in catalysis and materials science

A series of nanoreactors have been developed for complementary in situ X-ray and electron microscopy. The setup was utilized to monitor thermal annealing processes of porous materials under controlled atmospheres with hard X-ray ptychography.

In ANSYS AIM (Student Version, ANSYS, Inc.), Fluid Flow template module was used to simulate the gas flow in the 2G and 3G cell. Specific parameters were kept as default unless specifically mentioned. The geometry was generated by the integrated Space Claim Direct Modeler (SDCM) and "Define mesh manually" was chosen. In Fluid-specific options, "Swirling flow" was ticked to enable any significant rotating or twisting movement of the gas. The simulation focus was set as "Steadystate fluid flow". Number of fluid regions was selected to be "1", i.e. Air as the Material Sample, and "Compressible flow (ideal gas)" was ticked as an additional flow physics. For the meshing process settings, the resolution was set to 4, "Fluid flow or fluid-solid heat transfer" was selected for Engineering intent, and the refinement mechanism of "Curvature and proximity" was used. In Mesh controls, the element shape was set to "Automatic", where it meshed sweepable bodies with hexahedrons and provided a tetrahedral mesh on bodies that were not sweepable. The number of generated nodes was 46556 for 2G cell and 95272 for 3G cell, respectively.
For the Physics parameters, the whole SDCM-generated volume was specified as the location of Physics Regions. "Laminar" flow model was selected as the Flow Model since a Reynolds number (Re) of 9 was estimated according to: = * * , where  is the mass density of air (1.1839 kg m -3 at T = 298 K),  is the mean velocity of air (0.236 m s -1 in this case from a volumetric flow of 1 ml min -1 ), DH is the hydraulic diameter of the inlet/outlet channel (0.3 mm), and  is the dynamic viscosity of air (18.6 µPa s). The Inlet/Outlet Boundary was located on either one cross-section face of the long cylindrical channels with "Mixed" subsonic-supersonic regime and Gauge static pressure of 0 Pa. Additional parameter settings were for the inlet with gas Velocity magnitude of 0.236 m s -1 and Temperature of 298 K. The rest of the faces on the SDCM-generated volume was determined as the Wall Boundary. The Solver Options was left as default, where the solution would converge with Residual Tolerance of 1E-05 and Maximum iteration number of 1000; Solution advancement exponent was kept 0 under Calculation Control; and "Least squares cell-based" Gradient method was used for Discretization Control with "Second-order upwind" Advection and "Automatic" Pressure scheme. As the results, the convergence criteria were met after 45 iterations with residual tolerances up to 9.69E-06 for 2G cell, and 773 iterations with residual tolerances up to 2.15E-07 for 3G cell. The fluid flow profiles of the 2G and 3G cell are displayed in Figure 3 of the manuscript.

S2. Sample preparation of np-Au series via FIB
The description below ( Figure S1-S4) only applies to the preparation of np-Au samples. Each original sample was a solid pellet, a piece of which was deposited on a SEM stub and then on a lift-out grid, which acted as an intermediate stage to the final sample holder, i.e. MEMS chip ( Figure S1). All sample preparation steps were performed in a dual-beam focused ion beam. For details in each stage after mounting the stub, the piece was cut again and shaped into a smaller block / trapezoidal prism ( Figure S2). The block was then transferred onto the lift-out grid, where a smaller wedge shape was eventually obtained ( Figure S3). Last, the final transfer onto the MEMS-chip was made on the spiral center area of the chip and glued with Pt on the corner to stabilize the position ( Figure S4).

Figure S1
Overview of np-Au sample preparation on the MEMS chip via FIB.

Figure S2
Initial np-Au extraction on a SEM stub to make a block for preparing several samples.

Figure S3
Final cut after transfer to intermediate lift-out grid and thinning of block into a wedge.

Figure S4
Transfer and deposition of the wedge on the MEMS chip.

S3. Temperature stability of the heating interface system
Since the MEMS chips and parts of heating interface system are commercially available and tested, the temperature readout and control process proved to be robust as shown in Figure S5. Excellent stability can be observed in the step functions between heating points, where fluctuations of <1 K were observed during temperature adjustment.

Figure S5
Plot of temperature readout vs time via the integrated temperature control system, recorded during thermal annealing study of nanoporous gold. The valley between the fourth and fifth hour of the measurement resulted from pausing the experiment due to loss of beam.

S4. Determination of the spatial resolution
To estimate the spatial resolution of the ptychograms obtained, a Fourier Ring Correlation (FRC) analysis was performed (Van Heel & Schatz, 2005), depicted in Figure S6 for colloidal Au, Figure S7 for nanoporous gold, and Figure S8 for the hierarchical zeolite particle. The intersection of the FRC datapoint with the 1 -bit threshold curve provides an estimation of the spatial resolution in nm averaged over the whole image and all directions.   In addition to FRC, the resolution of the images was also estimated with a sharp edge fitting as shown in Figure S9. This was performed by drawing a line profile across a sharp contrast feature and counting the number of pixels through which the line passes. It should be noted that this method relies on the presence of sharp features, and is therefore dependent on the specific sample and image contrast. In general, lower estimations of resolution can be expected than for FRC. In particular, since the colloidal gold and zeolite could be clearly measured against the background, a sharper contrast gradient was observed. On the other hand, the np-Au sample is significantly thicker, while the void measured probably does not penetrate through the entire sample. Colloidal Au particles with a size of 50 nm were utilized to confirm the applicability and feasibility of the 2G cell for in situ XRP and XRF. For this study, an early prototype of the 2G cell was used. This prototype consisted of the same physical cell profile shown in Figure 2 including electrical contacts, but without the gas tight O-ring and window assembly. Experiments were therefore performed only at elevated temperature up to 773 K but with ambient gas conditions. The ptychography images ( Figure   S10 top row) imply that the spatial resolution should be at least 50 nm as colloidal Au can be clearly resolved, where high local concentration of colloidal Au is indicated by the darker intensity. Using FRC, the spatial resolution was in fact estimated to be 45 nm as shown in Figure S6. However, the Au fluorescence maps ( Figure S10 bottom row) are more suitable for locating colloidal Au, with increased local concentration leading to higher fluorescence counts. The Au colloids showed a distribution of larger connected ligaments, likely from clustered bunches of colloids, together with a homogeneous distribution of smaller particles, particularly visible in the upper middle of Figure S10a the apparent disappearance of smaller particles in close vicinity to the more densely concentrated ones. These observations are typically expected for unsupported particles at elevated temperatures, at which the organic surfactant likely decomposes leading to aggregation. The complete disappearance of many smaller Au particles was also observed particularly at 773 K, which was somewhat unexpected since particle migration may not be expected to proceed across several microns. However, this could be due to the unstable deposition of unsupported particles directly on the Si3N4 window of the MEMS-chip, so the thermal effect may have caused them to significantly drift or fall away from the chip. As noted above, instead of the Digiheater temperature control unit, a variable voltage power supply and an IR thermographic camera were used as in previous work (Baier, Damsgaard et al., 2016). A typical IR thermography image captured from the cell during operation is shown in Figure   S11. In this case, wires connected to the power supply were soldered directly to the power in/out contact of the printed circuit. As such, accurate control of the sample temperature was not guaranteed during the experiment. As noted previously, a sudden variation in local gas composition or local chemical reactions on the chip may cause significant temperature spikes, as observed in previous work (Baier, Damsgaard, et al., 2016). This might also contribute to the sudden disappearance of colloidal Au at 773 K. This early case study highlights the necessity of a stable and dependable temperature control system, which was later integrated into both the 2G and 3G setups detailed in Figure 1 and Figure 2. Figure S12 TEM-EDX data recorded on the nanoporous gold sample used for in situ ptychography.
Analysis shows the presence of minor Si and Pt contaminants close to the Au ligament edges.

Table S1
Quantification of EDX spectra in Figure S12.