Forward and Backward Machine Learning for Modeling Copper Diffusion in Cadmium Telluride Solar Cells

161835-Thumbnail Image.png
Description
To optimize solar cell performance, it is necessary to properly design the doping profile in the absorber layer of the solar cell. For CdTe solar cells, Cu is used for providing p-type doping. Hence, having an estimator that, given the

To optimize solar cell performance, it is necessary to properly design the doping profile in the absorber layer of the solar cell. For CdTe solar cells, Cu is used for providing p-type doping. Hence, having an estimator that, given the diffusion parameter set (time and Temperature) and the doping concentration at the junction, gives the junction depth of the absorber layer, is essential in the design process of CdTe solar cells (and other cell technologies). In this work it is called a forward (direct) estimation process. The backward (inverse) problem then is the one in which, given the junction depth and the desired concentration of Cu doping at the CdTe/CdS heterointerface, the estimator gives the time and/or the Temperature needed to achieve the desired doping profiles. This is called a backward (inverse) estimation process. Such estimators, both forward and backward, do not exist in the literature for solar cell technology. To train the Machine Learning (ML) estimator, it is necessary to first generate a large set of data that are obtained by using the PVRD-FASP Solver, which has been validated via comparison with experimental values. Note that this big dataset needs to be generated only once. Next, one uses Machine Learning (ML), Deep Learning (DL) and Artificial Intelligence (AI) to extract the actual Cu doping profiles that result from the process of diffusion, annealing, and cool-down in the fabrication sequence of CdTe solar cells. Two deep learning neural network models are used: (1) Multilayer Perceptron Artificial Neural Network (MLPANN) model using a Keras Application Programmable Interface (API) with TensorFlow backend, and (2) Radial Basis Function Network (RBFN) model to predict the Cu doping profiles for different Temperatures and durations of the annealing process. Excellent agreement between the simulated results obtained with the PVRD-FASP Solver and the predicted values is obtained. It is important to mention here that it takes a significant amount of time to generate the Cu doping profiles given the initial conditions using the PVRD-FASP Solver, because solving the drift-diffusion-reaction model is mathematically a stiff problem and leads to numerical instabilities if the time steps are not small enough, which, in turn, affects the time needed for completion of one simulation run. The generation of the same with Machine Learning (ML) is almost instantaneous and can serve as an excellent simulation tool to guide future fabrication of optimal doping profiles in CdTe solar cells.
Date Created
2021
Agent

Efficient Inversion of Large-Scale Problems Exploiting Structure and Randomization

158635-Thumbnail Image.png
Description
Dimensionality reduction methods are examined for large-scale discrete problems, specifically for the solution of three-dimensional geophysics problems: the inversion of gravity and magnetic data. The matrices for the associated forward problems have beneficial structure for each depth layer of the

Dimensionality reduction methods are examined for large-scale discrete problems, specifically for the solution of three-dimensional geophysics problems: the inversion of gravity and magnetic data. The matrices for the associated forward problems have beneficial structure for each depth layer of the volume domain, under mild assumptions, which facilitates the use of the two dimensional fast Fourier transform for evaluating forward and transpose matrix operations, providing considerable savings in both computational costs and storage requirements. Application of this approach for the magnetic problem is new in the geophysics literature. Further, the approach is extended for padded volume domains.

Stabilized inversion is obtained efficiently by applying novel randomization techniques within each update of the iteratively reweighted scheme. For a general rectangular linear system, a randomization technique combined with preconditioning is introduced and investigated. This is shown to provide well-conditioned inversion, stabilized through truncation. Applying this approach, while implementing matrix operations using the two dimensional fast Fourier transform, yields computationally effective inversion, in memory and cost. Validation is provided via synthetic data sets, and the approach is contrasted with the well-known LSRN algorithm when applied to these data sets. The results demonstrate a significant reduction in computational cost with the new algorithm. Further, this new algorithm produces results for inversion of real magnetic data consistent with those provided in literature.

Typically, the iteratively reweighted least squares algorithm depends on a standard Tikhonov formulation. Here, this is solved using both a randomized singular value de- composition and the iterative LSQR Krylov algorithm. The results demonstrate that the new algorithm is competitive with these approaches and offers the advantage that no regularization parameter needs to be found at each outer iteration.

Given its efficiency, investigating the new algorithm for the joint inversion of these data sets may be fruitful. Initial research on joint inversion using the two dimensional fast Fourier transform has recently been submitted and provides the basis for future work. Several alternative directions for dimensionality reduction are also discussed, including iteratively applying an approximate pseudo-inverse and obtaining an approximate Kronecker product decomposition via randomization for a general matrix. These are also topics for future consideration.
Date Created
2020
Agent

Computational Methods for Kinetic Reaction Systems

158258-Thumbnail Image.png
Description
This work is concerned with the study and numerical solution of large reaction diffusion systems with applications to the simulation of degradation effects in solar cells. A discussion of the basics of solar cells including the function of solar cells,

This work is concerned with the study and numerical solution of large reaction diffusion systems with applications to the simulation of degradation effects in solar cells. A discussion of the basics of solar cells including the function of solar cells, the degradation of energy efficiency that happens over time, defects that affect solar cell efficiency and specific defects that can be modeled with a reaction diffusion system are included. Also included is a simple model equation of a solar cell. The basics of stoichiometry theory, how it applies to kinetic reaction systems, and some conservation properties are introduced. A model that considers the migration of defects in addition to the reaction processes is considered. A discussion of asymptotics and how it relates to the numerical simulation of the lifetime of solar cells is included. A reduced solution is considered and a presentation of a numerical comparison of the reduced solution with the full solution on a simple test problem is included. Operator splitting techniques are introduced and discussed. Asymptotically preserving schemes combine asymptotics and operator splitting to use reasonable time steps. A presentation of a realistic example of this study applied to solar cells follows.
Date Created
2020
Agent

A Unified 2D Solver for Modeling Carrier and Defect Dynamics in Electronic and Photovoltaic Devices

157838-Thumbnail Image.png
Description
Semiconductor devices often face reliability issues due to their operational con-

ditions causing performance degradation over time. One of the root causes of such

degradation is due to point defect dynamics and time dependent changes in their

chemical nature. Previously developed Unified Solver

Semiconductor devices often face reliability issues due to their operational con-

ditions causing performance degradation over time. One of the root causes of such

degradation is due to point defect dynamics and time dependent changes in their

chemical nature. Previously developed Unified Solver was successful in explaining

the copper (Cu) metastability issues in cadmium telluride (CdTe) solar cells. The

point defect formalism employed there could not be extended to chlorine or arsenic

due to numerical instabilities with the dopant chemical reactions. To overcome these

shortcomings, an advanced version of the Unified Solver called PVRD-FASP tool was

developed. This dissertation presents details about PVRD-FASP tool, the theoretical

framework for point defect chemical formalism, challenges faced with numerical al-

gorithms, improvements for the user interface, application and/or validation of the

tool with carefully chosen simulations, and open source availability of the tool for the

scientific community.

Treating point defects and charge carriers on an equal footing in the new formalism

allows to incorporate chemical reaction rate term as generation-recombination(G-R)

term in continuity equation. Due to the stiff differential equations involved, a reaction

solver based on forward Euler method with Newton step is proposed in this work.

The Jacobian required for Newton step is analytically calculated in an elegant way

improving speed, stability and accuracy of the tool. A novel non-linear correction

scheme is proposed and implemented to resolve charge conservation issue.

The proposed formalism is validated in 0-D with time evolution of free carriers

simulation and with doping limits of Cu in CdTe simulation. Excellent agreement of

light JV curves calculated with PVRD-FASP and Silvaco Atlas tool for a 1-D CdTe

solar cell validates reaction formalism and tool accuracy. A closer match with the Cu

SIMS profiles of Cu activated CdTe samples at four different anneal recipes to the

simulation results show practical applicability. A 1D simulation of full stack CdTe

device with Cu activation at 350C 3min anneal recipe and light JV curve simulation

demonstrates the tool capabilities in performing process and device simulations. CdTe

device simulation for understanding differences between traps and recombination

centers in grain boundaries demonstrate 2D capabilities.
Date Created
2019
Agent

Multiscale modeling of silicon heterojunction solar cells

157640-Thumbnail Image.png
Description
Silicon photonic technology continues to dominate the solar industry driven by steady improvement in device and module efficiencies. Currently, the world record conversion efficiency (~26.6%) for single junction silicon solar cell technologies is held by silicon heterojunction (SHJ) solar cells

Silicon photonic technology continues to dominate the solar industry driven by steady improvement in device and module efficiencies. Currently, the world record conversion efficiency (~26.6%) for single junction silicon solar cell technologies is held by silicon heterojunction (SHJ) solar cells based on hydrogenated amorphous silicon (a-Si:H) and crystalline silicon (c-Si). These solar cells utilize the concept of carrier selective contacts to improve device efficiencies. A carrier selective contact is designed to optimize the collection of majority carriers while blocking the collection of minority carriers. In the case of SHJ cells, a thin intrinsic a-Si:H layer provides crucial passivation between doped a-Si:H and the c-Si absorber that is required to create a high efficiency cell. There has been much debate regarding the role of the intrinsic a-Si:H passivation layer on the transport of photogenerated carriers, and its role in optimizing device performance. In this work, a multiscale model is presented which utilizes different simulation methodologies to study interfacial transport across the intrinsic a-Si:H/c-Si heterointerface and through the a-Si:H passivation layer. In particular, an ensemble Monte Carlo simulator was developed to study high field behavior of photogenerated carriers at the intrinsic a-Si:H/c-Si heterointerface, a kinetic Monte Carlo program was used to study transport of photogenerated carriers across the intrinsic a-Si:H passivation layer, and a drift-diffusion model was developed to model the behavior in the quasi-neutral regions of the solar cell. This work reports de-coupled and self-consistent simulations to fully understand the role and effect of transport across the a-Si:H passivation layer in silicon heterojunction solar cells, and relates this to overall solar cell device performance.
Date Created
2019
Agent

The economics of need-based transfers

156315-Thumbnail Image.png
Description
Need-based transfers (NBTs) are a form of risk-pooling in which binary welfare exchanges

occur to preserve the viable participation of individuals in an economy, e.g. reciprocal gifting

of cattle among East African herders or food sharing among vampire bats. With the

broad goal

Need-based transfers (NBTs) are a form of risk-pooling in which binary welfare exchanges

occur to preserve the viable participation of individuals in an economy, e.g. reciprocal gifting

of cattle among East African herders or food sharing among vampire bats. With the

broad goal of better understanding the mathematics of such binary welfare and risk pooling,

agent-based simulations are conducted to explore socially optimal transfer policies

and sharing network structures, kinetic exchange models that utilize tools from the kinetic

theory of gas dynamics are utilized to characterize the wealth distribution of an NBT economy,

and a variant of repeated prisoner’s dilemma is analyzed to determine whether and

why individuals would participate in such a system of reciprocal altruism.

From agent-based simulation and kinetic exchange models, it is found that regressive

NBT wealth redistribution acts as a cutting stock optimization heuristic that most efficiently

matches deficits to surpluses to improve short-term survival; however, progressive

redistribution leads to a wealth distribution that is more stable in volatile environments and

therefore is optimal for long-term survival. Homogeneous sharing networks with low variance

in degree are found to be ideal for maintaining community viability as the burden and

benefit of NBTs is equally shared. Also, phrasing NBTs as a survivor’s dilemma reveals

parameter regions where the repeated game becomes equivalent to a stag hunt or harmony

game, and thus where cooperation is evolutionarily stable.
Date Created
2018
Agent

Numerical issues arising in the simulations of transient water flow in layered unsaturated soils

155496-Thumbnail Image.png
Description
The geotechnical community typically relies on recommendations made from numerical simulations. Commercial software exhibits (local) numerical instabilities in layered soils across soil interfaces. This research work investigates unsaturated moisture flow in layered soils and identifies a possible source of numerical

The geotechnical community typically relies on recommendations made from numerical simulations. Commercial software exhibits (local) numerical instabilities in layered soils across soil interfaces. This research work investigates unsaturated moisture flow in layered soils and identifies a possible source of numerical instabilities across soil interfaces and potential improvement in numerical schemes for solving the Richards' equation. The numerical issue at soil interfaces is addressed by a (nonlinear) interface problem. A full analysis of the simplest soil hydraulic model, the Gardner model, identifies the conditions of ill-posedness of the interface problem. Numerical experiments on various (more advanced and practical) soil hydraulic models show that the interface problem can also be ill-posed under certain circumstances. Spurious numerical ponding and/or oscillations around soil interfaces are observed consequently. This work also investigates the impact of different averaging schemes for cell-centered conductivities on the propensity of ill-posedness of the interface problem and concludes that smaller averaging conductivities are more likely to trigger numerical instabilities. In addition, an agent-based stochastic soil model, with hydraulic properties defined at the finite difference cell level, results in a large number of interface problems. This research compares sequences of stochastic realizations in heterogeneous unsaturated soils with the numerical solution using homogenized soil parameters. The mean of stochastic realizations is not identical to the solution obtained from homogenized soil parameters.
Date Created
2017
Agent

Swarming in bounded domains

154089-Thumbnail Image.png
Description
Swarms of animals, fish, birds, locusts etc. are a common occurrence but their coherence and method of organization poses a major question for mathematics and biology.The Vicsek and the Attraction-Repulsion are two models that have been proposed to

Swarms of animals, fish, birds, locusts etc. are a common occurrence but their coherence and method of organization poses a major question for mathematics and biology.The Vicsek and the Attraction-Repulsion are two models that have been proposed to explain the emergence of collective motion. A major issue for the Vicsek Model is that its particles are not attracted to each other, leaving the swarm with alignment in velocity but without spatial coherence. Restricting the particles to a bounded domain generates global spatial coherence of swarms while maintaining velocity alignment. While individual particles are specularly reflected at the boundary, the swarm as a whole is not. As a result, new dynamical swarming solutions are found.

The Attraction-Repulsion Model set with a long-range attraction and short-range repulsion interaction potential typically stabilizes to a well-studied flock steady state solution. The particles for a flock remain spatially coherent but have no spatial bound and explore all space. A bounded domain with specularly reflecting walls traps the particles within a specific region. A fundamental refraction law for a swarm impacting on a planar boundary is derived. The swarm reflection varies from specular for a swarm dominated by

kinetic energy to inelastic for a swarm dominated by potential energy. Inelastic collisions lead to alignment with the wall and to damped pulsating oscillations of the swarm. The fundamental refraction law provides a one-dimensional iterative map that allows for a prediction and analysis of the trajectory of the center of mass of a flock in a channel and a square domain.

The extension of the wall collisions to a scattering experiment is conducted by setting two identical flocks to collide. The two particle dynamics is studied analytically and shows a transition from scattering: diverging flocks to bound states in the form of oscillations or parallel motions. Numerical studies of collisions of flocks show the same transition where the bound states become either a single translating flock or a rotating (mill).
Date Created
2015
Agent

An aggregate second order continuum model for transient production planning

154081-Thumbnail Image.png
Description
Factory production is stochastic in nature with time varying input and output processes that are non-stationary stochastic processes. Hence, the principle quantities of interest are random variables. Typical modeling of such behavior involves numerical simulation and statistical

Factory production is stochastic in nature with time varying input and output processes that are non-stationary stochastic processes. Hence, the principle quantities of interest are random variables. Typical modeling of such behavior involves numerical simulation and statistical analysis. A deterministic closure model leading to a second order model for the product density and product speed has previously been proposed. The resulting partial differential equations (PDE) are compared to discrete event simulations (DES) that simulate factory production as a time dependent M/M/1 queuing system. Three fundamental scenarios for the time dependent influx are studied: An instant step up/down of the mean arrival rate; an exponential step up/down of the mean arrival rate; and periodic variation of the mean arrival rate. It is shown that the second order model, in general, yields significant improvement over current first order models. Specifically, the agreement between the DES and the PDE for the step up and for periodic forcing that is not too rapid is very good. Adding diffusion to the PDE further improves the agreement. The analysis also points to fundamental open issues regarding the deterministic modeling of low signal-to-noise ratio for some stochastic processes and the possibility of resonance in deterministic models that is not present in the original stochastic process.
Date Created
2015
Agent

A: kinetic approach to anomalous diffusion in biological trapping regions

153170-Thumbnail Image.png
Description
Advances in experimental techniques have allowed for investigation of molecular dynamics at ever smaller temporal and spatial scales. There is currently a varied and growing body of literature which demonstrates the phenomenon of \emph{anomalous diffusion} in physics, engineering, and biology.

Advances in experimental techniques have allowed for investigation of molecular dynamics at ever smaller temporal and spatial scales. There is currently a varied and growing body of literature which demonstrates the phenomenon of \emph{anomalous diffusion} in physics, engineering, and biology. In particular many diffusive type processes in the cell have been observed to follow a power law $\left \propto t^\alpha$ scaling of the mean square displacement of a particle. This contrasts with the expected linear behavior of particles undergoing normal diffusion. \emph{Anomalous sub-diffusion} ($\alpha<1$) has been attributed to factors such as cytoplasmic crowding of macromolecules, and trap-like structures in the subcellular environment non-linearly slowing the diffusion of molecules. Compared to normal diffusion, signaling molecules in these constrained spaces can be more concentrated at the source, and more diffuse at longer distances, potentially effecting the signalling dynamics. As diffusion at the cellular scale is a fundamental mechanism of cellular signaling and additionally is an implicit underlying mathematical assumption of many canonical models, a closer look at models of anomalous diffusion is warranted. Approaches in the literature include derivations of fractional differential diffusion equations (FDE) and continuous time random walks (CTRW). However these approaches are typically based on \emph{ad-hoc} assumptions on time- and space- jump distributions. We apply recent developments in asymptotic techniques on collisional kinetic equations to develop a FDE model of sub-diffusion due to trapping regions and investigate the nature of the space/time probability distributions assosiated with trapping regions. This approach both contrasts and compliments the stochastic CTRW approach by positing more physically realistic underlying assumptions on the motion of particles and their interactions with trapping regions, and additionally allowing varying assumptions to be applied individually to the traps and particle kinetics.
Date Created
2014
Agent