Publications

We have developed a mold-to-mold cross imprint (MTMCI) process, which redefines an imprint mold with another imprint mold. By performing MTMCI on two identical imprint molds with silicon spacer nanowires in a perpendicular arrangement, we fabricated a large array of sub-30-nm silicon nanopillars. Large-area arrays of Pt dots are then produced using nanoimprint lithography with the silicon nanopillar mold.

A method for growing vertical ZnO nanowire arrays on arbitrary substrates using either gas-phase or solution-phase approaches is presented. A approximately 10 nm-thick layer of textured ZnO nanocrystals with their c axes normal to the substrate is formed by the decomposition of zinc acetate at 200-350 degrees C to provide nucleation sites for vertical nanowire growth. The nanorod arrays made in solution have a rod diameter, length, density, and orientation desirable for use in ordered nanorod-polymer solar cells.

Graph neural networks are powerful architectures for structured datasets. However, current methods struggle to represent long-range dependencies. Scaling the depth or width of GNNs is insufficient to broaden receptive fields as larger GNNs encounter optimization instabilities such as vanishing gradients and representation oversmoothing, while pooling-based approaches have yet to become as universally useful as in computer vision. In this work, we propose the use of Transformer-based self-attention to learn long-range pairwise relationships, with a novel "readout" mechanism to obtain a global graph embedding. Inspired by recent computer vision results that find position-invariant attention performant in learning long-range relationships, our method, which we call GraphTrans, applies a permutation-invariant Transformer module after a standard GNN module. This simple architecture leads to state-of-the-art results on several graph classification tasks, outperforming methods that explicitly encode graph structure. Our results suggest that purely-learning-based approaches without graph structure may be suitable for learning high-level, long-range relationships on graphs. Code for GraphTrans is available at https://github.com/ucbrise/graphtrans.

We present the spectroscopic and photometric evolution of the nearby (z = 0.059) spectroscopically confirmed type Ic supernova, SN 2010bh, associated with the soft, long-duration gamma-ray burst (X-ray flash) GRB 100316D. Intensive follow-up observations of SN 2010bh were performed at the ESO Very Large Telescope (VLT) using the X-shooter and FORS2 instruments. Owing to the detailed temporal coverage and the extended wavelength range (3000--24800 A), we obtained an unprecedentedly rich spectral sequence among the hypernovae, making SN 2010bh one of the best studied representatives of this SN class. We find that SN 2010bh has a more rapid rise to maximum brightness (8.0 +/- 1.0 rest-frame days) and a fainter absolute peak luminosity (L_bol~3e42 erg/s) than previously observed SN events associated with GRBs. Our estimate of the ejected (56)Ni mass is 0.12 +/- 0.02 Msun. From the broad spectral features we measure expansion velocities up to 47,000 km/s, higher than those of SNe 1998bw (GRB 980425) and 2006aj (GRB 060218). Helium absorption lines He I lambda5876 and He I 1.083 microm, blueshifted by ~20,000--30,000 km/s and ~28,000--38,000 km/s, respectively, may be present in the optical spectra. However, the lack of coverage of the He I 2.058 microm line prevents us from confirming such identifications. The nebular spectrum, taken at ~186 days after the explosion, shows a broad but faint [O I] emission at 6340 A. The light-curve shape and photospheric expansion velocities of SN 2010bh suggest that we witnessed a highly energetic explosion with a small ejected mass (E_k ~ 1e52 erg and M_ej ~ 3 Msun). The observed properties of SN 2010bh further extend the heterogeneity of the class of GRB supernovae.

The vibrational spectroscopy sum frequency generation (SFG) is used to investigate the adsorption of carbon monoxide on the single crystal (111) and polycrystalline platinum surfaces. By varying the frequency and polarization of the light beams, different surface species of CO species are probed. SFG signal intensities for different polarization indicate that adsorbed CO polarizability is significantly perturbed from the gas-phase molecule. The SFG signal of CO disappears well below the main oxidation potential of CO to CO2. The disappearance of the CO signal is interpreted as a transformation in the CO layer to a state which is invisible to SFG. The invisible state is suggested to be CO with the bond axis nearly parallel to the platinum surface.

Stabile, zeolithartige Mikroporen weist ein aus Lanthanoid-Ionen und Benzoldicarboxylat(BDC)-Gruppen aufgebautes Gerüst auch in Abwesenheit von Gastmolekülen auf. Das Gerüst wurde durch Copolymerisation von TbIII-Ionen und 1,4-Benzoldicarbonsäure erhalten. Das offene Tb-BDC-Gerüst weist die Struktur eines erweiterten PtS-Netzwerks auf (siehe Bild; Tb: offene, C: gefüllte Kreise. Tb- und C-Atome verbindende Linien stellen O-Atome dar, C-Atome verbindende Linien Benzolringe) und kann reversibel aus der Gasphase und aus Lösungen Gastmoleküle aufnehmen.

Biopharmaceuticals, such as proteins and DNA, have demonstrated their potential to prevent and cure diseases. The success of such therapeutic agents hinges upon their ability to cross complex barriers in the body and reach their target intact. In order to reap the full benefits of these therapeutic agents, a delivery vehicle capable of delivering cargo to all cell types, both phagocytic and non-phagocytic, is needed. This article presents the synthesis and evaluation of a microparticle delivery vehicle capable of cell penetration and sub-cellular triggered release of an encapsulated payload. pH-sensitive polyacrylamide particles functionalized with a polyarginine cell-penetrating peptide (CPP) were synthesized. The incorporation of a CPP into the microparticles led to efficient uptake by non-phagocytic cells in culture. In addition, the CPP-modified particles showed no cytotoxic effects at concentrations used in this study. The results suggest that these particles may provide a vehicle for the successful delivery of therapeutic agents to various cell types.

Artificial intelligence for scientific discovery has recently generated significant interest within the machine learning and scientific communities, particularly in the domains of chemistry, biology, and material discovery. For these scientific problems, molecules serve as the fundamental building blocks, and machine learning has emerged as a highly effective and powerful tool for modeling their geometric structures. Nevertheless, due to the rapidly evolving process of the field and the knowledge gap between science (e.g., physics, chemistry, & biology) and machine learning communities, a benchmarking study on geometrical representation for such data has not been conducted. To address such an issue, in this paper, we first provide a unified view of the current symmetry-informed geometric methods, classifying them into three main categories: invariance, equivariance with spherical frame basis, and equivariance with vector frame basis. Then we propose a platform, coined Geom3D, which enables benchmarking the effectiveness of geometric strategies. Geom3D contains 16 advanced symmetry-informed geometric representation models and 14 geometric pretraining methods over 46 diverse datasets, including small molecules, proteins, and crystalline materials. We hope that Geom3D can, on the one hand, eliminate barriers for machine learning researchers interested in exploring scientific problems; and, on the other hand, provide valuable guidance for researchers in computational chemistry, structural biology, and materials science, aiding in the informed selection of representation techniques for specific applications.

Carbon aerogels are distinguished by a series of remarkable qualities, including ultralow density, low thermal conductivity, and high specific surface area; however, their low strength and inherent brittleness present a problem that severely constrains their application. Here a design strategy is proposed to overcome these shortcomings by synergistically carbonizing flexible organic fibers and aerogel matrix while elaborately engineering their interfacial bonding; the result is the creation of an ultra-strong, damage-tolerant carbon aerogel composite. The material exhibits a nanoporous carbon aerogel matrix reinforced by in situ-formed carbon fibers, featuring low residual stress and gradient atomic-scale bonding between the fibers and matrix. This structure imparts ultrahigh strength, with high specific strength normalized by density, all surpassing the benchmark properties of existing aerogels; it also displays exceptional fracture toughness with unprecedented stability in crack propagation. These attributes are further combined with excellent thermal stability and insulation, high-temperature mechanical properties, good machinability, and ease of large-scale production.