Data-driven automated control algorithm for floating-zone crystal growth derived by reinforcement learning.

The complete automation of materials manufacturing with high productivity is a key problem in some materials processing. In floating zone (FZ) crystal growth, which is a manufacturing process for semiconductor wafers such as silicon, an operator adaptively controls the input parameters in accordance with the state of the crystal growth process. Since the operation dynamics of FZ crystal growth are complicated, automation is often difficult, and usually the process is manually controlled. Here we demonstrate automated control of FZ crystal growth by reinforcement learning using the dynamics predicted by Gaussian mixture modeling (GMM) from small numbers of trajectories. Our proposed method of constructing the control model is completely data-driven. Using an emulator program for FZ crystal growth, we show that the control model constructed by our proposed model can more accurately follow the ideal growth trajectory than demonstration trajectories created by human operation. Furthermore, we reveal that policy optimization near the demonstration trajectories realizes accurate control following the ideal trajectory.

Pharmacokinetics of core-polymerized, boron-conjugated micelles designed for boron neutron capture therapy for cancer.

Core-polymerized and boron-conjugated micelles (PM micelles) were prepared by free radical copolymerization of a PEG-b-PLA block copolymer bearing an acetal group and a methacryloyl group (acetal-PEG-b-PLA-MA), with 1-(4-vinylbenzyl)-closo-carborane (VB-carborane), and the utility of these micelles as a tumor-targeted boron delivery system was investigated for boron neutron capture therapy (BNCT). Non-polymerized micelles (NPM micelles) that incorporated VB-carborane physically showed significant leakage of VB-carborane (ca. 50%) after 12 h incubation with 10% fetal bovine serum (FBS) at 37 °C. On the other hand, no leakage from the PM micelles was observed even after 48 h of incubation. To clarify the pharmacokinetics of the micelles, (125)I (radioisotope)-labeled PM and NPM micelles were administered to colon-26 tumor-bearing BALB/c mice. The (125)I-labeled PM micelles showed prolonged blood circulation (area under the concentration curve (AUC): 943.4) than the (125)I-labeled NPM micelles (AUC: 495.1), whereas tumor accumulation was similar for both types of micelles (AUC(PM micelle): 249.6, AUC(NPM micelle): 201.1). In contrast, the tumor accumulation of boron species in the PM micelles (AUC: 268.6) was 7-fold higher than the NPM micelles (AUC: 37.1), determined by ICP-AES. Thermal neutron irradiation yielded tumor growth suppression in the tumor-bearing mice treated with the PM micelles without reduction in body weight. On the basis of these data, the PM micelles represent a promising approach to the creation of boron carrier for BNCT.

On-off regulation of 19F magnetic resonance signals based on pH-sensitive PEGylated nanogels for potential tumor-specific smart 19F MRI probes.

The pH-sensitive PEGylated nanogels constructed from tethered PEG chains and a polyamine gel core containing 19F compounds showed remarkable on-off regulation of 19F MR (magnetic resonance) signals in response to the extracellular pH (6.5) of the tumor environment, even in the presence of 90% fetal bovine serum, due to the increase in the molecular motion of the 19F compounds through the hydrophilic-hydrophobic (volume-phase) transition of the polyamine gel core. Eventually, an appreciably enhanced 19F MR signal at an extremely low 19F compound concentration (approximately 55 microM) was achieved, demonstrating the utility of these nanogels as solid tumor-specific smart 19F MRI probes.