Liquid cell Transmission Electron Microscopy (TEM) provides the opportunity to view nanocrystal dynamics in their native liquid environment with the necessary spatial resolution. In this dataset, gold nanocubes and nano-rhombic dodecahedra are etched inside graphene liquid cells while viewing using TEM. The initial gold nanocrystals are synthesized and then loaded in water pockets sandwiched between sheets of graphene. These graphene pockets protect the liquid from the vacuum of the TEM column while not preventing the electron beam from imaging the sample. The etching species in the reaction is a combination of FeCl₃ preloaded in the liquid cell and oxidative radiolysis species generated by the electron beam. For this experiment, the initial concentration of FeCl₃ was modulated to determine its effect on the etching process. A previously published paper using this dataset showed that the concentration of FeCl₃ acts as the potential in the system and this controls the intermediate facets on the nanocrystal. By watching nanocrystals as they undergo dynamic processes, the pathways and mechanisms of nanoscale processes can be better understood.

Taken from the Supporting Information of the previously published paper investingating the effect of the concentration of FeCl₃ on the intemediate facets:

Hauwiller, M. R.; Frechette, L.; Jones, M.; Ondry, J.; Rotskoff, G.; Geissler, P.; Alivisatos, A. P. Unraveling Kinetically-Driven Mechanisms of Gold Nanocrystal Shape Transformations using Graphene Liquid Cell Electron Microscopy. Nano Letters. 2018.

Synthesis of Initial Gold Nanocrystals

Gold cubes and rhombic dodecahedra were synthesized following a previously reported method for growing polyhedral nanocrystals from uniform seeds. To prepare the seeds, nanorods were synthesized and then etched down to spherical seeds. Briefly, 125 µL of 10 mM HAuCl₄ was mixed with 5 mL of 100 mM CTAB before adding 300 µL of 10 mM ice cold NaBH₄. This nanorod seed reaction mixture was stirred for 1 minute, and then let sit for 30 minutes. A separate solution of 20 mL of 100 mM CTAB, 1 mL of 10 mM HAuCl₄, 0.18 mL of 10 mM AgNO₃, and 0.114 mL 100 mM Ascorbic Acid was prepared before injecting 24 µL of nanorod seed reaction mixture. The resulting growth solution sat in a water bath at 28° C for 2 hours. Nanorods were centrifuged two times for 15 minutes at 8,000 rpm, resuspending each time in 50 mM CTAB. The final nanorod solution was brought to 2 OD with 50 mM CTAB. Spherical gold seeds were synthesized from the nanorods by adding between 0.087 and 0.105 µL of 1 mM HAuCl₄ to each mL of nanorod solution and then let stir gently for 4 hours at 40° C. The exact amount of HAuCl₄ added to each rod sample was determined from small test batches to account for slight differences in the nanorod samples. The spherical seeds were centrifuged two times for 30 minutes at 11,000 rpm, each time resuspending in 100 mM cetylpyridinium chloride (CPC) before bringing the final solution of spherical seeds to a 1 OD concentration.

Gold nanocubes were synthesized by mixing 5 mL of 100 mM CPC, 500 µL of 100 mM KBr, 100 µL of 10 mM HAuCl4, and 150 µL of 100 mM Ascorbic Acid before injecting between 350 µL and 400 µL of spherical seeds. The reaction was left to gently stir for 1 hour and then centrifuged twice at 10,000 rpm for 10 minutes before redispersing in Millipore-filtered water.

Gold rhombic dodecahedra were synthesized by mixing 5 mL of 100 mM CPC, 250 µL of 1 M HCl, 250 µL of 10 mM HAuCl4, 13 µL of 10 mM AgNO3, and 30 µL of 100 mM of Ascorbic Acid before injecting between 50 and 100 µL of spherical seeds. The reaction was left to stir for 5.5 hours and then centrifuged twice at 10,000 rpm for 10 minutes before redispersing in Millipore-filtered water.

Graphene Liquid Cell Fabrication

Graphene coated TEM grids were prepared by direct transfer of graphene onto holey quatifoil TEM grids. Breifly, CVD grown 3-5 layer Graphene on copper (ACS Materials) was washed three times in Acetone at 50° C for five minutes to dissolve any residual polymer on the graphene surface. After air drying, the graphene on copper was carefully flattened between two glass slides to smooth out bumps and rough areas. Holey carbon (1.2 µm diameter with 1.3 µM separation between holes), gold 300 Mesh quantifoil TEM grids (Structure Probe, Inc.) were placed carbon-side down on the graphene. A couple droplets of isopropanol were placed on top of the grids on graphene and let dry to bring the holey carbon in close contact with the graphene. After drying, the TEM grids bonded to graphene/copper substrate were floated on sodium persulfate with the copper side down to etch the copper. Following overnight etching, the graphene coated TEM grids were floated on Millipore-filtered water three times to rinse off the sodium persulfate and placed graphene-side up on filter paper for future use.

Graphene liquid cell pockets were made by encapsulating the solution of interest between two graphene coated TEM grids. The solution contained 0.25 mL of a FeCl₃ and Tris Buffer-HCl mixture with 10 uL of highly concentrated nanoparticles. The ratio of FeCl₃ and Tris Buffer was tuned to yield the desired FeCl₃ concentration for each trial. Tris Buffer-HCl was used to stabilize pockets although the exact mechanism by which Tris Buffer-HCl increases pocket success is not well understood. An approximately 0.5 uL droplet of the FeCl₃/Tris Buffer/gold nanoparticle mixture was deposited on the graphene side of one grid. A second graphene-coated grid with a corner cut off was quickly placed graphene side down on top of the droplet on the first grid using tweezers. After waiting a couple minutes, the grid was taken to the TEM for imaging.

TEM Imaging

TEM imaging was performed on a FEI Tecnai T20 S-Twin TEM operating at 200kV with a LaB₆ filament. All videos were taken between 30 minutes and 3 hours after encapsulation. Timeseries of TEM images were collected with a Gatan Orius SC200 camera using a custom digital micrograph script with full 2048 x 2048 pixel readout with a binning of 2 pixels in each direction, at a nominal magnification of 71kx resulting in a pixel resolution of 1.6 Å/pixel, an exposure time of 0.5 s, and a readout time of 0.8 s yielding a framerate of 0.77 fps. Since etching is beam initiated, care was taken to minimize electron dose prior to imaging. Searching for suitable nanoparticles in liquid pockets was performed with a spread beam. Once a etchable nanocrystal was found, a custom Digital Micrograph script was used to reproducibly condense the electron beam to the same dose rate and start the TEM movie acquisition.

 Briefly, at the start of a TEM session, a calibration script was run which fits the excitation of the C2 condenser lens to the dose rate of electrons recorded on the camera. This was done without a sample in the beam path. The extrema (most condensed and most spread) of the beam at the desired magnification were input into the script, and the C2 excitation was recorded by the script for both those states.  The script divided this interval into 10 equally spaced C2 lens excitation values, then it cycled through the C2 lens values and recorded an image at each value.  The image was used to calculate the dose rate at the given C2 lens excitation.  This data was linearized by plotting the C2 lens excitation vs. 1/sqrt(dose rate) and fit using a linear regression algorithm script.  The linear regression results were then saved as a persistent number in Digital Micrograph. The dose rate calibration results were then used by a separate data acquisition script which the user could input an arbitrary dose rate within the calibrated range for data acquisition. Then, the script would convert the desired dose rate to a C2 lens excitation that would deliver the desired dose rate.  Once the script set the required C2 lens excitation, the script would begin collecting a timeseries of images to record the beam induced in situ reaction.  The C2 lens calibration script was run before each imaging session since the brightness of our LaB₆ source decreases over time. We estimate that the dose rate is reproducibly returned to the same value within ±10%.  We note that our dose rate control method reproducibly provides the same dose rate over months of experiments (assuming camera sensitivity does not change over time) although it’s absolute magnitude is limited by the accuracy of the detector sensitivity measurement. Dose rate during etching was 800 e^-/Ųs using an incident electron to counts conversion of 6.26 counts per incident electron. 

Image Analysis

The .dm3 are the original video files from the TEM imaging. The .avi files were made by opening the .dm3 files in ImageJ and saving as an .avi with TIFF compression.