Combining Numerical Simulations, Machine Learning and Genetic Algorithms for Optimizing a POCl<sub>3</sub> Diffusion Process

Wagner-Mohnsen, Hannes; Esefelder, Sascha; Klöter, B.; Mitchell, Bernhard; Schinke, Carsten; Bredemeier, Dennis; Jäger, Philip; Brendel, Rolf

doi:10.1109/pvsc43889.2021.9518450

Cited by 5 publications

(4 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The method was validated with experimental results, where the parameters obtained by the ML-based and DPSS approaches were shown to agree within an acceptable uncertainty range . ML-based regression methods for solar cell parameters to evaluate the impact of the thickness of different layers on the efficiency, , understand material properties, , and understand current–voltage curve analysis ,− were recently published. Regression tasks have also been performed on luminescence images. , Classification methods have mainly revolved around automated image analysis using deep learning algorithms such as convolutional neural networks (CNNs), where the objective is to classify defects − or identify their position in luminescence images. − …”

Section: Introductionmentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

Buratti¹,

Dick

Gia

et al. 2022

ACS Appl. Mater. Interfaces

View full text Add to dashboard Cite

Bulk defects in silicon solar cells are a key contributor to loss of efficiency. To detect and identify those defects, temperature-and injection-dependent lifetime spectroscopy is usually used, and the defect parameters are traditionally extracted using fitting methods to the Shockley−Read−Hall recombination statistics. In this study, we propose a deep learning-based extraction technique that is based on an alternative representation of the lifetime curves: lifetime mapping in the temperature and minority carrier concentration space. The deep learning approach successfully predicts all the defect parameters while addressing one of the main limitations of the traditional approach of locating the defect in the energy spectrum, which usually outputs two possible solutions. Furthermore, the approach is applied to temperature-dependent defect parameters where the traditional approach is not applicable, achieving satisfying levels of prediction of the defect parameters. Image representation and deep learning have the potential to bolster solar cell characterization techniques by extracting more insights from the characterization data.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

Buratti¹,

Dick

Gia

et al. 2022

ACS Appl. Mater. Interfaces

View full text Add to dashboard Cite

show abstract

“…Kaya and Hajimirza [10] also show that a black box approach of using ML models to optimize device performance is efficient compared to fitting a numerical simulation model, the emphasis was not on creating a perfectly calibrated simulation model but to create a pool of data set by varying individual parameters and predicting device performance to identify the optimized device characteristics. The concept of using machine learning techniques to also optimize a processing recipe was demonstrated in Mohnsen et al [11].…”

Section: Introductionmentioning

confidence: 99%

A semi-empirical approach to calibrate simulation models for semiconductor devices

Jaiswal¹,

Ramon-Martinez²,

Busani³

2023

Preprint

View full text Add to dashboard Cite

Semiconductor device optimization using computer-based prototyping techniques like simulation or machine learning digital twins can be time and resource efficient compared to the conventional strategy of iterating over device design variations by fabricating the actual device. However, simulation models require perfect calibration of material parameters for the model to represent a particular semiconductor device. This cali- bration process itself can require characterization information of the device and its precursors and extensive expert knowledge of non char- acterizable parameters and their tuning. We propose a hybrid method to calibrate multiple simulation models for a device using minimal characterization data and machine learning-based prediction models. A photovoltaic device is chosen as the example for this technique where optical and electrical simulation models of an industrially manufactured silicon solar cell are calibrated and the simulated device performance is compared with the measurement data from the physical device.

show abstract

“…Kaya and Hajimirza 10 also show that a black box approach of using ML models to optimize device performance is efficient compared to fitting a numerical simulation model, the emphasis was not on creating a perfectly calibrated simulation model but to create a pool of data set by varying individual parameters and predicting device performance to identify the optimized device characteristics. The concept of using machine learning techniques to also optimize a processing recipe was demonstrated in Mohnsen et al 11 . Our proposed methodology is based on the hypothesis that ML models can learn from a minimal data set and can try to predict a target value by interpolating or extrapolating the training data points.…”

mentioning

confidence: 99%

A semi-empirical approach to calibrate simulation models for semiconductor devices

Jaiswal

Martínez‐Ramón

Busani

2023

Sci Rep

View full text Add to dashboard Cite

Semiconductor device optimization using computer-based prototyping techniques like simulation or machine learning digital twins can be time and resource efficient compared to the conventional strategy of iterating over device design variations by fabricating the actual device. Ideally, simulation models require perfect calibration of material parameters for the model to represent a particular semiconductor device. This calibration process itself can require characterization information of the device and its precursors and extensive expert knowledge of non characterizable parameters and their tuning. We propose a hybrid method to calibrate multiple simulation models for a device using minimal characterization data and machine learning-based prediction models. A photovoltaic device is chosen as the example for this technique where optical and electrical simulation models of an industrially manufactured silicon solar cell are calibrated and the simulated device performance is compared with the measurement data from the physical device.

show abstract

Combining Numerical Simulations, Machine Learning and Genetic Algorithms for Optimizing a POCl₃ Diffusion Process

Cited by 5 publications

References 4 publications

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

A semi-empirical approach to calibrate simulation models for semiconductor devices

A semi-empirical approach to calibrate simulation models for semiconductor devices

Contact Info

Product

Resources

About

Combining Numerical Simulations, Machine Learning and Genetic Algorithms for Optimizing a POCl3 Diffusion Process

Cited by 5 publications

References 4 publications

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

Deep Learning Extraction of the Temperature-Dependent Parameters of Bulk Defects

A semi-empirical approach to calibrate simulation models for semiconductor devices

A semi-empirical approach to calibrate simulation models for semiconductor devices

Contact Info

Product

Resources

About

Combining Numerical Simulations, Machine Learning and Genetic Algorithms for Optimizing a POCl₃ Diffusion Process