Shaun Kwak - Academia.edu (original) (raw)

Papers by Shaun Kwak

In materials science, accurately computing properties like viscosity, melting point, and glass tr... more In materials science, accurately computing properties like viscosity, melting point, and glass transition temperatures solely through physics-based models is challenging. Data-driven machine learning (ML) also poses challenges in constructing ML models, especially in the material science domain where data is limited. To address this, we integrate physics-informed descriptors from molecular dynamics (MD) simulations to enhance the accuracy and interpretability of ML models. Our current study focuses on accurately predicting viscosity in liquid systems using MD descriptors. In this work, we curated a comprehensive dataset of over 4,000 small organic molecules’ viscosities from scientific literature, publications, and online databases. This dataset enabled us to develop quantitative structure–property relationships (QSPR) consisting of descriptor-based and graph neural network models to predict temperature-dependent viscosities for a wide range of viscosities with considerable accuracy...

Organic and Hybrid Light Emitting Materials and Devices XXVI, Sep 26, 2022

SID Symposium Digest of Technical Papers

Physical Chemistry Chemical Physics

Inter-molecular interactions significantly modulate the electronic properties of radical emitters... more Inter-molecular interactions significantly modulate the electronic properties of radical emitters. The doublet excitons in films demonstrate a significant CT character, impacting both radiative and non-radiative transitions in radical-based OLEDs.

Frontiers in Chemistry

In recent years, generative machine learning approaches have attracted significant attention as a... more In recent years, generative machine learning approaches have attracted significant attention as an enabling approach for designing novel molecular materials with minimal design bias and thereby realizing more directed design for a specific materials property space. Further, data-driven approaches have emerged as a new tool to accelerate the development of novel organic electronic materials for organic light-emitting diode (OLED) applications. We demonstrate and validate a goal-directed generative machine learning framework based on a recurrent neural network (RNN) deep reinforcement learning approach for the design of hole transporting OLED materials. These large-scale molecular simulations also demonstrate a rapid, cost-effective method to identify new materials in OLEDs while also enabling expansion into many other verticals such as catalyst design, aerospace, life science, and petrochemicals.

Frontiers in Chemistry

Data-driven methods are receiving increasing attention to accelerate materials design and discove... more Data-driven methods are receiving increasing attention to accelerate materials design and discovery for organic light-emitting diodes (OLEDs). Machine learning (ML) has enabled high-throughput screening of materials properties to suggest new candidates for organic electronics. However, building reliable predictive ML models requires creating and managing a high volume of data that adequately address the complexity of materials’ chemical space. In this regard, active learning (AL) has emerged as a powerful strategy to efficiently navigate the search space by prioritizing the decision-making process for unexplored data. This approach allows a more systematic mechanism to identify promising candidates by minimizing the number of computations required to explore an extensive materials library with diverse variables and parameters. In this paper, we applied a workflow of AL that accounts for multiple optoelectronic parameters to identify materials candidates for hole-transport layers (HT...

SID Symposium Digest of Technical Papers

The properties of rubber are strongly influenced by the distribution of filler within the polymer... more The properties of rubber are strongly influenced by the distribution of filler within the polymer matrix. We describe a modelling approach to the calculation of dynamic moduli of filled elastomers based on filler morphologies derived from the experimental interface tensions of the material’s components. A Monte Carlo-based morphology generator, developed previously [1,2], is used to build model compounds on the μm-scale. Subsequently the Monte Carlo morphologies are mapped onto on a coarse grained model, allowing to obtain the amplitude and frequency dependence of the dynamic moduli during cyclic deformations. This combination of models ties the experimental surface tensions, characterizing the individual components, to the dynamic behavior of the macroscopic material. We consider selected examples of binary polymer blends containing a single type of filler at variable concentration. In addition to the dynamic moduli we also compute attendant transmission micrographs, wetting envelo...

Establishing relationships between basic chemical composition, structure morphology, and macrosco... more Establishing relationships between basic chemical composition, structure morphology, and macroscopic materials properties is the key element in the rational design of more robust, better manufacturable and environmentally friendly polymeric products. Recent advances in atomistic modelling and machine learning methods combined with advances in computing technology make these approaches a method of choice for uncovering key structure-property relationship. Especially, recent advances make it possible to use GPU hardware for MD simulations, enabling simulation time scales that were not previously accessible. Long MD trajectories for molecular systems that contain 10-10 atoms allow to obtain thermodynamic observables with high accuracy. For thermoplastic and rubbery polymers, the ability to model larger systems and longer times, make molecular modeling an increasingly valuable tool for understanding the behavior of industrially relevant polymers. Here, we will use the simulation of the ...

Journal of the Imaging Society of Japan, 2015

Thermoset polymers have gained interest in recent years due to their low cost, ease of processing... more Thermoset polymers have gained interest in recent years due to their low cost, ease of processing and unique physical properties. Molecular simulation represents an avenue to explore the chemical structure-function relationship of these polymers by leveraging advances in the speed and accuracy of molecular dynamics (MD) simulations, due to high performance computing (CPU/GPU), efficient algorithms and modern force fields. We have developed a cross linking algorithm that allows for any chemistry to be defined to break two bonds and form new ones. This feature greatly increases the applicability in forming polymers with different crosslinking chemistries. System properties can be monitored during a cross linking simulation within a single interface, allowing the user to estimate properties like theoretical gel points and reactive group concentrations as curing occurs. After curing, glass transition temperatures (Tg) can be predicted using long MD cooling simulations in excess of 1 mic...

As OLED applications increase, so do the demands on properties of the component materials, active... more As OLED applications increase, so do the demands on properties of the component materials, active layers and devices. The development of flexible OLEDs, a popular future OLED application, require better understanding and control of the mechanical properties of OLED materials and interaction with polymer substrates. Fabrication costs, use of extended classes of materials and the need for large surface area applications drives interest in solution-phase processing techniques; requiring OLEDs with different solubilities and glass transition temperatures than traditional vacuum deposited layers and device stacks. In this era of designing for multiple property requirements, computational techniques can provide important capability to screen new materials and understand the relationship between chemical structure and dependent properties. In this work we show automated molecular dynamics (MD) simulation workflows that efficiently and accurately calculate mechanical and physical properties...

Organic and Hybrid Light Emitting Materials and Devices XXV

Development and characterization of novel OLED materials by traditional computational approaches ... more Development and characterization of novel OLED materials by traditional computational approaches are challenging owing to the complex factors that simultaneously influence the device performance. In this work, we will provide an overview of generative OLED materials discovery using the latest deep neural network formalism, and show an illustrative example to design novel OLED hole-transport materials. The outcome of the work will demonstrate the value of systematic and fundamental understanding of structure-property correlations that can lead to rational design of smart OLEDs with higher efficiency.

Organic and Hybrid Light Emitting Materials and Devices XXV

To date, the development of organic light-emitting diode (OLED) materials has been primarily base... more To date, the development of organic light-emitting diode (OLED) materials has been primarily based on a combination of chemical intuition and trial-and-error experimentation. The approach is often expensive and time-consuming, let alone in most instances fails to offer new materials leading to higher efficiencies. Data-driven approaches have emerged as a powerful tool to accelerate the design and discovery of novel materials with multifunctional properties for next generation OLED technologies. Virtual high-throughput methods assisted by machine learning (ML) enable a broad screening of chemical space to predict material properties and suggest new candidates for OLEDs. In order to build reliable predictive ML models for OLED materials, it is required to create and manage a high volume of data which not only maintain high accuracy but also properly assess the complexity of materials chemistry in the OLED space. Active learning (AL) is among several strategies developed to face the challenge in both materials science and life science applications, where the data management in large-scale becomes a main bottleneck. Here, we present a workflow that efficiently combines AL with atomic-scale simulations to reliably predict optoelectronic properties of OLED materials. This study provides a robust and validated framework to account for multiple parameters that simultaneously influence OLED performance. Results of this work pave the way for a fundamental understanding of optoelectronic performance of emergent layers from a molecular perspective, and further screen candidate materials with superior efficiencies before laborious simulations, synthesis, and device fabrication.

Organic and Hybrid Light Emitting Materials and Devices XXV

In this work, we describe an atomistic-scale modeling and simulation scheme to virtually screen b... more In this work, we describe an atomistic-scale modeling and simulation scheme to virtually screen both host materials and light emitters used in OLEDs while assessing molecular orientations in film. The work also demonstrates the ability to predict wavelength-dependent refractive indices from atomistic-scale up to achieve this goal. These findings would provide valuable guidelines for the development of new material architectures with superior optical loss properties as well as improved outcoupling efficiencies at the device level.

Frontiers in Chemistry

Quantum chemical calculations are necessary to develop advanced emitter materials showing thermal... more Quantum chemical calculations are necessary to develop advanced emitter materials showing thermally-activated delayed fluorescence (TADF) for organic light-emitting diodes (OLEDs). However, calculation costs become problematic when more accurate functionals were used, therefore it is judicious to use a multimethod approach for efficiency. Here we employed combinatorial chemistry in silico to develop the deep blue TADF materials with a new concept of homo-junction design. The homo-junction materials containing TADF candidates designed by calculation were synthesized and analyzed. We found that these materials showed the emission from charge transfer (CT) state, and the clear delayed emission was provided in solid state. Because the homo-junction TADF materials showed three exponential decayed emission in solid state, we employed novel four-state kinetic analysis.

The Journal of Physical Chemistry A

Materials exhibiting higher mobilities than conventional organic semiconducting materials such as... more Materials exhibiting higher mobilities than conventional organic semiconducting materials such as fullerenes and fused thiophenes are in high demand for applications in printed electronics. In order to discover new molecules in the heteroacene family that might show improved charge mobility, a massive theoretical screen of hole conducting properties of molecules was performed by using cloud computing environment. Over 7,000,000 structures of fused furans, thiophenes and selenophenes were generated and 250,000 structures were randomly selected to perform DFT (Density Functional Theory) calculations of hole reorganization energies. The lowest hole reorganization energy calculated was 0.0548 eV for a fused thioacene having 8 aromatics rings. Hole mobilities of compounds with the lowest 130 reorganization energy were further processed by applying combined DFT and MD (Molecular Dynamics) methods. The highest mobility calculated was 1.02 cm2/Vs and 9.65 cm2/Vs based on percolation and disorder theory, respectively for compounds containing selenium atoms with 8 aromatic rings. These values are about 20 times higher than that for DNTT (dinaphthothienothiophene).

In materials science, accurately computing properties like viscosity, melting point, and glass tr... more In materials science, accurately computing properties like viscosity, melting point, and glass transition temperatures solely through physics-based models is challenging. Data-driven machine learning (ML) also poses challenges in constructing ML models, especially in the material science domain where data is limited. To address this, we integrate physics-informed descriptors from molecular dynamics (MD) simulations to enhance the accuracy and interpretability of ML models. Our current study focuses on accurately predicting viscosity in liquid systems using MD descriptors. In this work, we curated a comprehensive dataset of over 4,000 small organic molecules’ viscosities from scientific literature, publications, and online databases. This dataset enabled us to develop quantitative structure–property relationships (QSPR) consisting of descriptor-based and graph neural network models to predict temperature-dependent viscosities for a wide range of viscosities with considerable accuracy...

Organic and Hybrid Light Emitting Materials and Devices XXVI, Sep 26, 2022

SID Symposium Digest of Technical Papers

Physical Chemistry Chemical Physics

Inter-molecular interactions significantly modulate the electronic properties of radical emitters... more Inter-molecular interactions significantly modulate the electronic properties of radical emitters. The doublet excitons in films demonstrate a significant CT character, impacting both radiative and non-radiative transitions in radical-based OLEDs.

Frontiers in Chemistry

In recent years, generative machine learning approaches have attracted significant attention as a... more In recent years, generative machine learning approaches have attracted significant attention as an enabling approach for designing novel molecular materials with minimal design bias and thereby realizing more directed design for a specific materials property space. Further, data-driven approaches have emerged as a new tool to accelerate the development of novel organic electronic materials for organic light-emitting diode (OLED) applications. We demonstrate and validate a goal-directed generative machine learning framework based on a recurrent neural network (RNN) deep reinforcement learning approach for the design of hole transporting OLED materials. These large-scale molecular simulations also demonstrate a rapid, cost-effective method to identify new materials in OLEDs while also enabling expansion into many other verticals such as catalyst design, aerospace, life science, and petrochemicals.

Frontiers in Chemistry

Data-driven methods are receiving increasing attention to accelerate materials design and discove... more Data-driven methods are receiving increasing attention to accelerate materials design and discovery for organic light-emitting diodes (OLEDs). Machine learning (ML) has enabled high-throughput screening of materials properties to suggest new candidates for organic electronics. However, building reliable predictive ML models requires creating and managing a high volume of data that adequately address the complexity of materials’ chemical space. In this regard, active learning (AL) has emerged as a powerful strategy to efficiently navigate the search space by prioritizing the decision-making process for unexplored data. This approach allows a more systematic mechanism to identify promising candidates by minimizing the number of computations required to explore an extensive materials library with diverse variables and parameters. In this paper, we applied a workflow of AL that accounts for multiple optoelectronic parameters to identify materials candidates for hole-transport layers (HT...

SID Symposium Digest of Technical Papers

The properties of rubber are strongly influenced by the distribution of filler within the polymer... more The properties of rubber are strongly influenced by the distribution of filler within the polymer matrix. We describe a modelling approach to the calculation of dynamic moduli of filled elastomers based on filler morphologies derived from the experimental interface tensions of the material’s components. A Monte Carlo-based morphology generator, developed previously [1,2], is used to build model compounds on the μm-scale. Subsequently the Monte Carlo morphologies are mapped onto on a coarse grained model, allowing to obtain the amplitude and frequency dependence of the dynamic moduli during cyclic deformations. This combination of models ties the experimental surface tensions, characterizing the individual components, to the dynamic behavior of the macroscopic material. We consider selected examples of binary polymer blends containing a single type of filler at variable concentration. In addition to the dynamic moduli we also compute attendant transmission micrographs, wetting envelo...

Establishing relationships between basic chemical composition, structure morphology, and macrosco... more Establishing relationships between basic chemical composition, structure morphology, and macroscopic materials properties is the key element in the rational design of more robust, better manufacturable and environmentally friendly polymeric products. Recent advances in atomistic modelling and machine learning methods combined with advances in computing technology make these approaches a method of choice for uncovering key structure-property relationship. Especially, recent advances make it possible to use GPU hardware for MD simulations, enabling simulation time scales that were not previously accessible. Long MD trajectories for molecular systems that contain 10-10 atoms allow to obtain thermodynamic observables with high accuracy. For thermoplastic and rubbery polymers, the ability to model larger systems and longer times, make molecular modeling an increasingly valuable tool for understanding the behavior of industrially relevant polymers. Here, we will use the simulation of the ...

Journal of the Imaging Society of Japan, 2015

Thermoset polymers have gained interest in recent years due to their low cost, ease of processing... more Thermoset polymers have gained interest in recent years due to their low cost, ease of processing and unique physical properties. Molecular simulation represents an avenue to explore the chemical structure-function relationship of these polymers by leveraging advances in the speed and accuracy of molecular dynamics (MD) simulations, due to high performance computing (CPU/GPU), efficient algorithms and modern force fields. We have developed a cross linking algorithm that allows for any chemistry to be defined to break two bonds and form new ones. This feature greatly increases the applicability in forming polymers with different crosslinking chemistries. System properties can be monitored during a cross linking simulation within a single interface, allowing the user to estimate properties like theoretical gel points and reactive group concentrations as curing occurs. After curing, glass transition temperatures (Tg) can be predicted using long MD cooling simulations in excess of 1 mic...

As OLED applications increase, so do the demands on properties of the component materials, active... more As OLED applications increase, so do the demands on properties of the component materials, active layers and devices. The development of flexible OLEDs, a popular future OLED application, require better understanding and control of the mechanical properties of OLED materials and interaction with polymer substrates. Fabrication costs, use of extended classes of materials and the need for large surface area applications drives interest in solution-phase processing techniques; requiring OLEDs with different solubilities and glass transition temperatures than traditional vacuum deposited layers and device stacks. In this era of designing for multiple property requirements, computational techniques can provide important capability to screen new materials and understand the relationship between chemical structure and dependent properties. In this work we show automated molecular dynamics (MD) simulation workflows that efficiently and accurately calculate mechanical and physical properties...

Organic and Hybrid Light Emitting Materials and Devices XXV

Development and characterization of novel OLED materials by traditional computational approaches ... more Development and characterization of novel OLED materials by traditional computational approaches are challenging owing to the complex factors that simultaneously influence the device performance. In this work, we will provide an overview of generative OLED materials discovery using the latest deep neural network formalism, and show an illustrative example to design novel OLED hole-transport materials. The outcome of the work will demonstrate the value of systematic and fundamental understanding of structure-property correlations that can lead to rational design of smart OLEDs with higher efficiency.

Organic and Hybrid Light Emitting Materials and Devices XXV

To date, the development of organic light-emitting diode (OLED) materials has been primarily base... more To date, the development of organic light-emitting diode (OLED) materials has been primarily based on a combination of chemical intuition and trial-and-error experimentation. The approach is often expensive and time-consuming, let alone in most instances fails to offer new materials leading to higher efficiencies. Data-driven approaches have emerged as a powerful tool to accelerate the design and discovery of novel materials with multifunctional properties for next generation OLED technologies. Virtual high-throughput methods assisted by machine learning (ML) enable a broad screening of chemical space to predict material properties and suggest new candidates for OLEDs. In order to build reliable predictive ML models for OLED materials, it is required to create and manage a high volume of data which not only maintain high accuracy but also properly assess the complexity of materials chemistry in the OLED space. Active learning (AL) is among several strategies developed to face the challenge in both materials science and life science applications, where the data management in large-scale becomes a main bottleneck. Here, we present a workflow that efficiently combines AL with atomic-scale simulations to reliably predict optoelectronic properties of OLED materials. This study provides a robust and validated framework to account for multiple parameters that simultaneously influence OLED performance. Results of this work pave the way for a fundamental understanding of optoelectronic performance of emergent layers from a molecular perspective, and further screen candidate materials with superior efficiencies before laborious simulations, synthesis, and device fabrication.

Organic and Hybrid Light Emitting Materials and Devices XXV

In this work, we describe an atomistic-scale modeling and simulation scheme to virtually screen b... more In this work, we describe an atomistic-scale modeling and simulation scheme to virtually screen both host materials and light emitters used in OLEDs while assessing molecular orientations in film. The work also demonstrates the ability to predict wavelength-dependent refractive indices from atomistic-scale up to achieve this goal. These findings would provide valuable guidelines for the development of new material architectures with superior optical loss properties as well as improved outcoupling efficiencies at the device level.

Frontiers in Chemistry

Quantum chemical calculations are necessary to develop advanced emitter materials showing thermal... more Quantum chemical calculations are necessary to develop advanced emitter materials showing thermally-activated delayed fluorescence (TADF) for organic light-emitting diodes (OLEDs). However, calculation costs become problematic when more accurate functionals were used, therefore it is judicious to use a multimethod approach for efficiency. Here we employed combinatorial chemistry in silico to develop the deep blue TADF materials with a new concept of homo-junction design. The homo-junction materials containing TADF candidates designed by calculation were synthesized and analyzed. We found that these materials showed the emission from charge transfer (CT) state, and the clear delayed emission was provided in solid state. Because the homo-junction TADF materials showed three exponential decayed emission in solid state, we employed novel four-state kinetic analysis.

The Journal of Physical Chemistry A

Materials exhibiting higher mobilities than conventional organic semiconducting materials such as... more Materials exhibiting higher mobilities than conventional organic semiconducting materials such as fullerenes and fused thiophenes are in high demand for applications in printed electronics. In order to discover new molecules in the heteroacene family that might show improved charge mobility, a massive theoretical screen of hole conducting properties of molecules was performed by using cloud computing environment. Over 7,000,000 structures of fused furans, thiophenes and selenophenes were generated and 250,000 structures were randomly selected to perform DFT (Density Functional Theory) calculations of hole reorganization energies. The lowest hole reorganization energy calculated was 0.0548 eV for a fused thioacene having 8 aromatics rings. Hole mobilities of compounds with the lowest 130 reorganization energy were further processed by applying combined DFT and MD (Molecular Dynamics) methods. The highest mobility calculated was 1.02 cm2/Vs and 9.65 cm2/Vs based on percolation and disorder theory, respectively for compounds containing selenium atoms with 8 aromatic rings. These values are about 20 times higher than that for DNTT (dinaphthothienothiophene).