Machine Learning Refinement of In Situ Images Acquired by Low Electron Dose LC-TEM.

We have studied a machine learning (ML) technique for refining images acquired during in situ observation using liquid-cell transmission electron microscopy. Our model is constructed using a U-Net architecture and a ResNet encoder. For training our ML model, we prepared an original image dataset that contained pairs of images of samples acquired with and without a solution present. The former images were used as noisy images, and the latter images were used as corresponding ground truth images. The number of pairs of image sets was 1,204, and the image sets included images acquired at several different magnifications and electron doses. The trained model converted a noisy image into a clear image. The time necessary for the conversion was on the order of 10 ms, and we applied the model to in situ observations using the software Gatan DigitalMicrograph (DM). Even if a nanoparticle was not visible in a view window in the DM software because of the low electron dose, it was visible in a successive refined image generated by our ML model.

In Situ Transmission Electron Microscopy Study of Bubble Behavior Near the Surface of Ice Crystals by Using a Liquid Cell With a Peltier Cooling Holder.

Liquid cell transmission electron microscopy (LC-TEM) is a unique technique that permits in situ observations of various phenomena in liquids with high spatial and temporal resolutions. One difficulty with this technique is the control of the environmental conditions in the observation area. Control of the temperature ranging from room temperature to minus several tens of degrees Celsius, is desirable for controlling the supersaturation in various materials and for observing crystallization more easily. We have developed a cooling transmission electron microscopy specimen holder that uses Peltier devices, and we have combined it with a liquid cell to realize accurate temperature control in LC-TEM. We evaluated this system by using water as a specimen. Motionless bubbles, shown to be voids containing pressurized gas, formed in the specimen sometime after the temperature had reached -12°C. An electron diffraction pattern showed that the specimen turned into ice Ih after the formation of these bubbles, confirming that our system works properly and can induce crystallization. In addition, we analyzed the behavior of bubbles formed in the ice Ih, and we discussed the formation of these bubbles and their internal pressure.

Possible embryos and precursors of crystalline nuclei of calcium carbonate observed by liquid-cell transmission electron microscopy.

Several different building blocks or precursors play an important role in the early stages of the crystallization of calcium carbonate (CaCO3). Many studies have been conducted over a long period to elucidate the fundamental processes involved in this crystallization. Here, we report the role of an amorphous phase and embryo at the beginning of the nucleation of CaCO3 from solutions of relatively low supersaturation. Prenucleation crystals formed in amorphous calcium carbonate (ACC) at a significantly large formation rate of 2 × 1022 m-3 s-1, suggesting that a low interfacial energy exists between the ACC and crystals. Only one calcite crystal exceeded the size for a critical nucleus (∼104 molecules) in 150 pre-nucleation crystals. Each pre-nucleation crystal might consist of a different polymorph, and ACCs have a similar composition and structure. A particle-detection algorithm, used in conjunction with machine learning, suggested that an embryo with a characteristic structure exists in solution and might play a crucial role in nucleation. No similar embryonic structure could be observed immediately after the dissolution of pre-nucleation particles, implying that their dissolution process is not simply the reverse process of their growth. This method should provide a new approach to understanding nucleation processes.

Fast Improvement of TEM Images with Low-Dose Electrons by Deep Learning.

Low electron dose observation is indispensable for observing various samples using a transmission electron microscope; consequently, image processing has been used to improve transmission electron microscopy (TEM) images. To apply such image processing to in situ observations, we here apply a convolutional neural network to TEM imaging. Using a dataset that includes short-exposure images and long-exposure images, we develop a pipeline for processed short-exposure images, based on end-to-end training. The quality of images acquired with a total dose of approximately $5$$e^{-}$ per pixel becomes comparable to that of images acquired with a total dose of approximately $1{,}000$$e^{-}$ per pixel. Because the conversion time is approximately 8 ms, in situ observation at 125 fps is possible. This imaging technique enables in situ observation of electron-beam-sensitive specimens.

Conversion of stable crystals to metastable crystals in a solution by periodic change of temperature.

Using a Becker-Döring-type model including cluster incorporation, we study the possibility of conversion of stable crystals to metastable crystals in a solution by a periodic change of temperature. At low temperature, both stable and metastable crystals are assumed to grow by coalescence with monomers and corresponding small clusters. At high temperature, a large amount of small clusters produced by the dissolution of crystals inhibits the dissolution of crystals, and the imbalance in the amount of crystals increases. By repeating this process, the periodic temperature change can convert stable crystals into metastable crystals.

Dynamics of Poly(3-hexylthiophene) Monolayers at Solution/Graphite Interfaces.

We have studied the dynamics of poly(3-hexylthiophene) (P3HT) monolayers at the interfaces between highly oriented pyrolytic graphite (HOPG) and P3HT solutions in 1,2,4-trichlorobenzene, using scanning tunneling microscopy (STM). Real-time STM observation at room temperature reveals that P3HT molecules adsorbed on graphite are substantially mobile even in densely packed conditions, causing significant fluctuations of the self-organized monolayer of P3HT. We find that in the monolayers, the orientational order is limited to a short range comparable to the polymer chain length. We show that our observations can be understood based on a 2D semiflexible lattice polymer model by performing kinetic Monte Carlo simulations.

Early Detection of Nucleation Events From Solution in LC-TEM by Machine Learning.

To support the detection, recording, and analysis of nucleation events during in situ observations, we developed an early detection system for nucleation events observed using a liquid-cell transmission electron microscope. Detectability was achieved using the machine learning equivalent of detection by humans watching a video numerous times. The detection system was applied to the nucleation of sodium chloride crystals from a saturated acetone solution of sodium chlorate. Nanoparticles with a radius of more greater than 150 nm were detected in a viewing area of 12 µm × 12 µm by the detection system. The analysis of the change in the size of the growing particles as a function of time suggested that the crystal phase of the particles with a radius smaller than 400 nm differed from that of the crystals larger than 400 nm. Moreover, the use of machine learning enabled the detection of numerous nanometer sized nuclei. The nucleation rate estimated from the machine-learning-based detection was of the same order as that estimated from the detection using manual procedures.

UV-Induced Formation of Ice XI Observed Using an Ultra-High Vacuum Cryogenic Transmission Electron Microscope and its Implications for Planetary Science.

The occurrence of hydrogen atom-ordered form of ice Ih, ice XI, in the outer Solar System has been discussed based on laboratory experiments because its ferroelectricity influences the physical processes in the outer Solar System. However, the formation of ice XI in that region is still unknown due to a lack of formation conditions at temperatures higher than 72 K and the effect of UV-rays on the phase transition from ice I to ice XI. As a result, we observed the UV-irradiation process on ice Ih and ice Ic using a newly developed ultra-high vacuum cryogenic transmission electron microscope. We found that ice Ih transformed to ice XI at temperatures between 75 and 140 K with a relatively small UV dose. Although ice Ic partially transformed to ice XI at 83 K, the rate of transformation was slower than for ice Ih. These findings point to the formation of ice XI at temperatures greater than 72 K via UV irradiation of ice I crystals in the Solar System; icy grains and the surfaces of icy satellites in the Jovian and Saturnian regions.

Effect of impurities on chirality conversion by grinding.

We study the effect of chiral impurities on Viedma ripening using a simple reaction model. The exponential amplification of the enantiomeric excess (EE) is greatly accelerated by a smaller growth rate of solids with monomers caused by the chiral impurities. From the analysis of the model, it is found that the time evolution of the EE is essentially described by a second-order differential equation. The effect of chiral impurities is that the unstable fixed point is shifted from the racemic point, which leads to the linear amplification behavior observed experimentally. The analysis also shows a possibility of an oscillatory decay of the EE.

Mechanism of chirality conversion by periodic change of temperature: Role of chiral clusters.

By grinding crystals in a solution, the chirality of crystal structure (and the molecular chirality for the case of chiral molecules as well) can be converted, and the cause of the phenomenon is attributed to crystal growth with chiral clusters. We show that the recently found chirality conversion with a periodic change of temperature can also be explained by crystal growth with chiral clusters. With the use of a generalized Becker-Döring model, which includes enantio-selective incorporation of small chiral clusters to large solid clusters, the change of cluster distribution and the mass flow between clusters are studied. The chiral clusters act as a reservoir to pump out the minority species to the majority, and the exponential amplification of the enantiomeric excess found in the experiment is reproduced in the numerical calculation.

Effect of container shape and walls on solidification of Brownian particles in a narrow system.

We carry out Brownian dynamics simulations and study the ordering of particles under a uniform external force in a narrow system. In our previous studies [M. Sato et al., Phys. Rev. E 87, 032403 (2013); J. Phys. Soc. Jpn. 82, 084804 (2013)], we showed that the ordering of particles depends on the direction of the external force. In the studies, however, the system size and the number of particles are small, so the behaviors we observed corresponds to the motions in the initial stage of crystallization. In this paper, using a longer container and more particles, we investigate how solidification in a narrow system proceeds. We also study the effect of the shape of simulation box on the ordering of particles. When we use a rhomboid as a simulation box, the ratio of the particles with the face-centered cubic structure to those with the hexagonal close-packed structure is larger than that in a cuboid system.

Crystallization of Brownian particles in thin systems constrained by walls.

Keeping formation of a colloidal crystal by a centrifugal force in mind, we carry out Brownian dynamics simulations in thin systems and study ordering of particles induced by an external force. During solidification, the two-dimensional ordering along walls initially occurs. Then, the ordered particles on the walls act as substrates, and crystallization proceeds into bulk. When the external force is weak, the close-packed face of the crystal structure is parallel to the bottom wall. The direction of the close-packed face depends on the strength of the external force: The close-packed face becomes parallel to the side walls with a strong external force.

Appearance of a homochiral state of crystals induced by random fluctuation in grinding.

We study crystallization of chiral crystals from achiral molecules using a master equation based on a simple reaction model. Although there is no chiral symmetry breaking in the reaction model, random fluctuations drive the system to a homochiral state. The time necessary for the appearance of the homochiral state is proportional to the total number of molecules in the system. This behavior is described by a diffusion equation in a size space with a position-dependent diffusion coefficient. We also study the effect of chiral impurities, which affect the crystal growth. Depending on the type of impurities, the chiral symmetry breaking occurs either deterministically or with the help of random fluctuations.

Effect of immobile impurities on two-dimensional nucleation.

We study the dependence of the critical size of the nucleus on the density of impurity by carrying out a Monte Carlo simulation. We assume that the impurities are fixed on a crystal surface, and neglect restoration of bonds between an impurity atom and an adsorbed atom. We initially prepare one cluster in the system and investigate the change of the cluster size. When the cluster size is the critical value, the frequency of increasing the cluster size is equal to that of decreasing the size. With increasing the impurity density, the critical nucleus becomes large. When the impurity density is sufficiently high, regardless of the initial cluster size, the cluster vanishes after a long time interval, namely, the critical cluster size is diverged.