It is well-known that room temperature ionic liquids (RTILs) often adopt a charge-separated layered structure, i.e., with alternating cation- and anion-rich layers, at electrified interfaces. However, the dynamic response of the layered structure to
temporal variations in applied potential is not well understood. We used in situ, real-time X-ray reflectivity (XR) to study the potential-dependent electric double layer (EDL) structure of an imidazolium-based RTIL on charged epitaxial graphene during potential cycling as a function of temperature. The results suggest that the graphene-RTIL interfacial structure is bistable in which the EDL structure at any intermediate potential can be described by the combination of two extreme-potential structures whose proportions vary depending on the polarity and magnitude of the applied potential. This picture is supported by the EDL structures obtained by fully atomistic molecular dynamics (MD) simulations at various static potentials. The potential-driven transition between the two structures is characterized by an increasing width but with an approximately fixed hysteresis magnitude as a function of temperature. The results are consistent with the coexistence of distinct anion and cation adsorbed structures separated by an energy barrier (~0.15 eV).
The interface between hexagonal ZnO films and cubic MgO (001) substrates, fabricated through molecular beam epitaxy, are thoroughly investigated. X-ray diffraction and (scanning) transmission electron microscopy reveal that, at the substrate temperat
ure above 200 degree C, the growth follows the single [0001] direction; while at the substrate below 150 degree C, the growth is initially along [0001] and then mainly changes to [0-332] variants beyond the thickness of about 10 nm. Interestingly, a double-domain feature with a rotational angle of 30 degree appears for the growth along [0001] regardless of the growth temperature, experimentally demonstrated the theoretical predictions for occurrence of double rotational domains in such a heteroepitaxy [Grundmann et al, Phys. Rev. Lett. 105, 146102 (2010)]. It is also found that, the optical transmissivity of the ZnO film is greatly influenced by the mutation of growth directions, stimulated by the bond-length modulations, as further determined by X-ray absorption Spectra (XAS) at Zn K edge. The XAS results also show the evolution of 4pxy and 4pz states in the conduction band as the growth temperature increases. The results obtained from this work can hopefully promote the applications of ZnO in advanced optoelectronics for which its integration with other materials of different phases is desirable.
The interaction of interfacial water with graphitic carbon at the atomic scale is studied as a function of the hydrophobicity of epitaxial graphene. High resolution X-ray reflectivity shows that the graphene-water contact angle is controlled by the a
verage graphene thickness, due to the fraction of the film surface expressed as the epitaxial buffer layer whose contact angle (contact angle theta_c = 73{deg}) is substantially smaller than that of multilayer graphene (theta_c = 93{deg}). Classical and ab initio molecular dynamics simulations show that the reduced contact angle of the buffer layer is due to both its epitaxy with the SiC substrate and the presence of interfacial defects. This insight clarifies the relationship between interfacial water structure and hydrophobicity, in general, and suggests new routes to control interface properties of epitaxial graphene.
In this paper, we propose the MIML (Multi-Instance Multi-Label learning) framework where an example is described by multiple instances and associated with multiple class labels. Compared to traditional learning frameworks, the MIML framework is more
convenient and natural for representing complicated objects which have multiple semantic meanings. To learn from MIML examples, we propose the MimlBoost and MimlSvm algorithms based on a simple degeneration strategy, and experiments show that solving problems involving complicated objects with multiple semantic meanings in the MIML framework can lead to good performance. Considering that the degeneration process may lose information, we propose the D-MimlSvm algorithm which tackles MIML problems directly in a regularization framework. Moreover, we show that even when we do not have access to the real objects and thus cannot capture more information from real objects by using the MIML representation, MIML is still useful. We propose the InsDif and SubCod algorithms. InsDif works by transforming single-instances into the MIML representation for learning, while SubCod works by transforming single-label examples into the MIML representation for learning. Experiments show that in some tasks they are able to achieve better performance than learning the single-instances or single-label examples directly.
Multi-instance learning attempts to learn from a training set consisting of labeled bags each containing many unlabeled instances. Previous studies typically treat the instances in the bags as independently and identically distributed. However, the i
nstances in a bag are rarely independent, and therefore a better performance can be expected if the instances are treated in an non-i.i.d. way that exploits the relations among instances. In this paper, we propose a simple yet effective multi-instance learning method, which regards each bag as a graph and uses a specific kernel to distinguish the graphs by considering the features of the nodes as well as the features of the edges that convey some relations among instances. The effectiveness of the proposed method is validated by experiments.
