Block copolymers (BCP) are a unique class of polymers, which can self-assemble into ordered microdomains with sizes from 3 nm to about 50 nm making BCPs an appealing meso-scale material. In thin films, arrays of BCP microdomains with long-range lateral order can serve as ideal templates or scaffolds for patterning nano-scale functional materials and synthesizing nanostructured materials with size scales that exceed the reach of photolithography. Among many annealing methods, solvent vapor annealing (SVA) is a low-cost, highly efficient way to annihilate defects in BCP thin films and facilitates the formation of highly ordered microdomains within minutes. Directing the self-assembly of BCPs could, in principle, lead to the formation of domains with near perfect lateral ordering. The mechanism of SVA of BCPs, however, is still ill-understood, albeit it has been widely adopted in research laboratories around the world for the past decade. In the first part of this thesis, the ordering process of BCP thin films during annealing in neutral solvents was investigated mainly by in situ synchrotron X-ray scattering. Briefly, the solvent molecules impart mobility to the BCP and enable a marked improvement in the lateral ordering of the BCP microdomains. Both, BCP concentration in the swollen film and the rate of solvent removal play a key role in obtaining films with well-ordered microdomains. The amount of swelling in a BCP thin film during SVA depends on the chemical nature of the blocks, the quality of the solvent, and the molecular weight of the BCP. A high degree of swelling - still low enough to prevent solvent-induced mixing (disordering) of BCP microdomains,- provides a high chain mobility, and thus results in the formation of arrays of ordered microdomains with large grain sizes after SVA in neutral solvents. The rate of solvent removal is another critical parameter for obtaining long-range lateral order in BCP thin films after SVA in neutral solvents. While in the swollen state ordered structures form with exceptional order, removal of the solvent results in a deterioration of order due to the confinement imposed to a BCP in a thin film by the rigid silicon substrate. It was found, however, that an instantaneous solvent removal can minimize disordering to preserve the order formed in the swollen state. Self-assembled BCP microdomains also serve as ideal template to pattern other materials with exceptional lateral resolution. In this thesis, two examples of BCP lithography was also demonstrated. A reconstruction process was used to enhance the etch contrast between two organic blocks. In one example, a BCP pattern was transferred to a silicon substrate to form high aspect ratio, 5:1, sub-10nm silicon lines or holes with high fidelity. While in a second example, I demonstrated the fabrication of silicon oxide dots with an areal density as high as 2 Tera dots per inch2 by BCP templates, which has the potential to serve as etch mask for bit pattern media applications.

In this article-based thesis, we cover applications of deep learning to different problems in condensed matter physics, where the goal is to either accelerate the computation or design of a nanoscale material. We first motivate and introduce how machine learning methods can be used to accelerate traditional condensed matter physics calculations. In addition, we discuss what designing a material means, and how it has been previously done. We then consider the fundamentals of electronic structure and conventional calculations which include density functional theory (DFT), density functional perturbation theory (DFPT), quantum Monte Carlo (QMC), and electron transport with tight binding. In addition, we cover the basics of deep learning. Afterwards, we discuss 6 articles. The first 5 articles are dedicated to accelerating the computation of nanoscale materials. In Article 1, we use convolutional neural networks to predict energies for diatomic molecules modelled with a Lennard-Jones potential and density functional theory energies of hexagonal lattices with and without defects. In Article 2, we use extensive deep neural networks to represent density functional theory energy functionals for electron gases by using the electron density as input and bypass the Kohn-Sham equations by using the external potential as input. In addition, we use deep convolutional inverse graphics networks to map the external potential directly to the electron density. In Article 3, we use voxel deep neural networks (VDNNs) to map electron densities to kinetic energy densities and functional derivatives of the kinetic energies for graphene lattices. We also use VDNNs to calculate an electron density from a direct minimization calculation and introduce a Monte Carlo based solver that avoids taking a functional derivative altogether. In Article 4, we use a deep learning framework to predict the polarization, dielectric function, Born-effective charges, longitudinal optical transverse optical splitting, Raman tensors, and Raman spectra for 2 crystalline systems. In Article 5, we use VDNNs to map DFT electron densities to QMC energy densities for graphene systems, and compute the energy barrier associated with forming a Stone-Wales defect. In Article 6, we design a graphene-based quantum transducer that has the ability to physically split valley currents by controlling the pn-doping of the lattice sites. The design is guided by an neural network that operates on a pristine lattice and outputs a lattice with pn-doping such that valley currents are optimally split. Lastly, we summarize the thesis and outline future work.

To investigate the existence  of  molecular heterogeneties in mixtures of DMSO-water, the dynamics were measured with the method of dynamic light scattering (DLS). Three different compositions (20 mol %, 33 mol% and 60 mol%) were included in the study and measured at room temperature (295 K) and for one composition (33 mol %) also a temperature dependence (from 295 K down to 263 K) was measured. Measurements were done on samples both with and without nanoparticles acting as tracers for the DLS. The diffusion coefficients of DMSO in water was extracted from the analysis and the results from samples without nanoparticles are consistent with diffusion of DMSO molecules reported previously, except for the highest concentration,  and showed a clear Arrhenius behaviour with an activation energy of 26±1kJ/mol. The viscosity was extracted from the diffusioncoefficient of the nanoparticles in the solutions and followed an expected trend regarding the concentration as well as for the higher temperatures, but deviated for lower temperatures due to an unexpected drastic change in the diffusion coefficient around the temperature T= 273 K. The reason for the drastic change could be connected to a possible liquid-liquid phase separation in the DMSO-water  mixture. The hydrodynamical radii was estimated using Stokes-Einsteins equation and had a small but unsure concentration dependence so the size could only be confirmed to be around 0.43 nm at T= 295 K and increased with temperature up to 1.6 nm atT= 263 K, indicating clustering effects and supporting the theory of molecular heterogeneity in DMSO-water mixtures. This was a pilot study to aproposed x-ray experiment at NanoMax at MaxIV to capture the nanoscale fluctuations present in binary solutions.

Recent progress and interest have bought considerable effort to bear on the synthesis and opportunities of magnetic thin films in different fields. There are applications in many fields, including remote sensing, waveguide applications, hard drive applications, etc. at the College of William and Mary, we have focused on utilizing magnetic thin films in some of these applications and are deeply involved in the optimization process of the thin films.

This thesis presents optical characterizations of interfaces in ferromagnetic heterostructures and thin films used for spin polarized electronic devices. In these experiments, femtosecond laser spectroscopies are exploited to investigate the interface magnetization reversal, spin precession, and band offset, which are crucial in determining the performances of spintronic devices.;First, magnetization-induced second-harmonic-generation (MSHG) is applied to study interface magnetism in a hybrid structure containing a noncentrosymmetric semiconductor---Fe/AlGaAs. The reversal process of Fe interface layer magnetization is compared with the bulk magnetization reversal. In Fe/AlGaAs (001), the interface magnetization is found to be decoupled from the bulk magnetization based on the different switching characteristics---single step switching occurs at the interface layer, whereas two-jump switching occurs in the bulk. In contrast, the interface layer in Fe/AlGaAs (110) is rigidly coupled with the bulk Fe, indicating a strong impact of electronic structure on the magnetic interaction despite the same chemical composition. Furthermore, a time-resolved MSHG study demonstrates a coherent interface magnetization precession in Fe/AlGaAs (001), implying the feasibility of fast precessional control of interfacial spin. The interface magnetization precession exhibits a higher frequency and opposite phase for a given applied field compared to the bulk magnetization precession.;Second, uniform magnetization precession in the Lac0.67Ca 0.33MnO3 (LCMO) and La0.67Sr0.33MnO 3 (LSMO) films grown on different substrates are investigated by time-resolved magneto-optic Kerr effect. The parameters of magnetic anisotropy are determined from the field dependence of the precession frequency. The strain-free LCMO films grown on NdGaO3 exhibit a uniaxial in-plane anisotropy induced by the tilting of the oxygen octahedra in NdGaO3 An easy-plane magnetic anisotropy is found in the tensile-strained films grown on SrTiO 3, whereas the compressive-strained film grown on LaAlO3 exhibits an easy normal-to-plane axis.;Third, a table-top internal photoemission system is developed to measure the band offsets across semiconductor heterointerfaces by utilizing an optical parametric amplifier as the bright light source. The conduction band offsets DeltaE c = 660 meV and 530 meV at the CdCr2Se4-GaAs and CdCrZSe4-ZnSe interfaces are determined from the threshold energies of the photocurrent spectrum. The band offset is shown to be reduced by engineering the interface bonding and stoichiometry.

Many linear accelerator (linac) applications rely on the use of superconducting radio frequency (SRF) cavities. In order to overcome the current field gradient limits imposed by the use of bulk niobium, a model involving the deposition of alternating superconducting-insulating-superconducting (SIS) thin films onto the interior surface of SRF cavities has been proposed. Since SRF performance is a surface phenomenon, the critical surface of these cavities is less than 1 micron thick, thus enabling the use of thin films. Before such approach can successfully be implemented fundamental studies correlating the microstructure and superconducting properties of thin films are needed. to this end the effect of grain boundary density and interfacial strain in thin films has been explored. Thin films with a smaller grain boundary density were found to have better superconducting properties than films with a larger grain boundary density. Interfacial strain due to a lattice mismatch between the film and substrate lead to two regions in films, one strained region near the interface and one relaxed region away from the interface. The presence of two regions in the film resulted in two types of superconducting behavior. Niobium films were deposited onto copper surfaces to help understand why previous attempts of implementing niobium coated copper cavities in order to exploit the better thermal properties of copper had varying degrees of success. It was found that an increased growth temperature produced niobium films with larger grains and correspondingly better superconducting properties. Proof of principle multilayer samples were prepared to test the SIS model. For the first time, multilayers were produced that were capable of shielding an underlying niobium film from vortex penetration beyond the lower critical field of bulk niobium. This result provides evidence supporting the feasibility of the SIS model.

This thesis presents results from large scale Monte Carlo simulations of systems subject to a superfluid phase transition in the presence of disorder. The simulations are performed by state-of-the-art, collective Monte Carlo algorithms treating phase degrees of freedom in effective models with amplitude fluctuations integrated out. In Paper I a model system for the possible solid to supersolid transition in 4He is presented.The Wolff cluster algorithm is used to study how the presence of linearly correlated random defects is able to alter the universality class of the 3-dimensional XY-model. In the pure case the superfluid density and heat capacity have singular onsets, which are not seen in the supersolid experiments where instead a smooth onset is obtained. Using finite size scaling of Monte Carlo data, we find a similar smooth onset in our simulations, governed by exponents  ν=1 for the superfluid density and α=-1 for the heat capacity. These results are in qualitative agreement with experiments for the observed transition in solid 4He. In Paper II a systematic investigation of the scaling result z=d for the dynamic critical exponentat the Bose glass to superfluid quantum phase transition is performed. The result z=d has been believed to be exact for about 20 years, but although it has been questioned lately no accurate estimate of z has been available. An effective link current model of quantum bosons at T=0 with disorder in 2D is simulated using highly effective worm Monte Carlo simulations.The data analysis is based on a finite size scaling approach todetermine the quantum correlation time from simulationdata for boson world lines without any a priori assumption on the critical parameters. The resulting critical exponents are z=1.8 \pm 0.05, ν=1.15 \pm 0.03, and η=-0.3 \pm 0.1. This suggests that z=d is not satisfied. / <p>QC 20111206</p>

QC 20110309

Special quasi-random structure (SQS) and coherent potential approximation (CPA) are techniques widely employed in the first-principles calculations of random alloys. The aim of the thesis is to study these approaches by focusing on the local lattice distortion (LLD) and the crystal symmetry effects. We compare the elastic parameters obtained from SQS and CPA calculations. For the CPA and SQS calculations, we employ the Exact Muffin-Tin Orbitals (EMTO) method and the pseudopotential method as implemented in the Vienna Ab initio Simulation Package (VASP), respectively. We compare the predicted trends of the VASP-SQS and EMTO-CPA parameters against composition. As a first case study, we investigate the elastic parameters of face centered cubic (fcc) Ti1−xAlx(0≤x≤100at.%) random solid solutions as a function of Al content (x). The EMTO-CPA and VASP-SQS results are in good agreement with each other. Comparing the lattice constants from SQS calculations with and without local lattice relaxations, we find that in Ti-rich (Al-rich) side the lattice constants remain almost unchanged (slightly increase) upon atomic relaxations. Taking local lattice distortions into consideration decreases the C11 and C44 elastic parameters, but their trends are not significantly affected. The C12 elastic constant, on the other hand, is almost unchanged when atomic relaxations are included. In general, the uncertainties in the elastic parameters associated with the symmetry lowering in supercell studies turn out to be superior to the differences between the two alloy techniques including the effect of LLD. We also investigate the elastic properties of random fcc Cu1−xAux(0≤x≤100 at.%) alloys as a function of Au content employing the CPA and SQS approaches. It is found that the CPA and SQS values forC11andC12 are consistent with each other no matter whether the atomic relaxations are taken into account or not. On the other hand, the EMTO-CPA values for C44 are slightly larger than those from SQS calculations especially for Cu-rich alloys which we ascribe to the differences in the DFT solvers rather than the differences between CPA and SQS. The Perdew-Burke-Ernzerhof (PBE) approximation to the exchange-correlation term in density functional theory (DFT) is a mature approach and have been adopted routinely to investigate the properties of metallic alloys. In most of the cases, PBE provides theoretical results in good agreement with experiments. However, the ordered Cu-Au system turned out to be a special case where large deviations between the PBE predictions and observations occur. In this work, we make use of a recently developed exchange-correlation functional, the so-called quasi-non-uniform exchange-correlation approximation (QNA), to calculate the lattice constants and formation energies for ordered Cu-Au alloys as a function of composition. The calculations are performed using the EMTO method. We find that the QNA functional leads to excellent agreement betweent heory and experiment. The PBE strongly overestimates the lattice constants for ordered Cu3Au, CuAu, CuAu3 compounds and also for the pure metals which is nicely corrected by the QNA approach. The errors in the formation energies of Cu3Au, CuAu, CuAu3relative to the experimental data decrease from 38-45% obtained with PBE to 5-9% calculated for QNA. / <p>QC 20151216</p>

Ab initiocalculation based on density function theory (DFT) is an accu-rate and efficient method for modelling material properties. It is performedby solving the Shrödinger equations with a few assumptions to obtain thephysical properties of the system. It is very computational demanding whendealing with large systems or long-time simulations. Developing empiricalpotentials on the basis ofab initiocalculations on smaller systems is a possi-ble way to solve this problem. The empirical potentials will benefit from theaccuracy ofab initiosimulations and can facilitate applications to large sys-tems and long-time simulations. In this thesis, we have performed two studiesregarding fitting empirical potentials: one is fitting an empirical Sutton-Chenpotential based onab initiosimulations for iron under extreme conditionsand the other one is fitting an improved Finnis-Sinclair potential for ternaryV-Ti-Cr alloy.In the first part, we focus on fitting a Sutton-Chen potential for solid Feunder the Earth’s inner core condition. Based onab initiomolecular dynam-ics (MD) simulation results, the Sutton-Chen potential is fitted to energies ofthe configurations obtained fromab initioMD simulations at the pressure of360 GPa and temperature of 6000 K. The method applied for the fitting isthe Particle Swarm Optimization (PSO) algorithm. The Sutton-Chen poten-tial can reproduce theab initioenergies with an error of 6.2 meV/atom. Setas the same withab initioMD simulations, classical MD using Sutton-Chenpotential can obtain the consistent results with those fromab initioMD sim-ulations at the pressure of 360 GPa and temperature of 6000 K. In order toexplore the size effect on the results, we extend the classical MD to large-sizesystems (from 1024 atoms to 65536 atoms). We also extend the temperaturerange to see the temperature effect on the results.In the second part, we develop an improved Finnis-Sinclair (IFS) potentialfor ternary V-Ti-Cr alloys. The interaction parameters of V-V, Ti-Ti andCr-Cr are fitted to the experimental lattice constants, cohesive energies andelastic constants. The binary alloy potential parameters are obtained byconstructing 3 binary alloy models (V15Ti, V15Cr, V8Ti8) and fitting to theirtheoretical lattice constants, cohesive energies and elastic constants. Finally,the IFS potential is successfully used to calculate mechanical properties andthe monovacancy formation energy in V-Ti-Cr alloy. It is also applied toinvestigate the composition effect on the mechanical properties of ternaryV-Ti-Cr alloys. / <p>QC 20160815</p>