서브메뉴
검색
Design of Organic-Inorganic Hybrid Membranes Using Density Functional Theory and Machine Learning
Design of Organic-Inorganic Hybrid Membranes Using Density Functional Theory and Machine Learning
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260209102902
- ISBN
- 9798265405272
- DDC
- 000
- 저자명
- Liu, Yifan.
- 서명/저자
- Design of Organic-Inorganic Hybrid Membranes Using Density Functional Theory and Machine Learning
- 발행사항
- [Sl] : Georgia Institute of Technology, 2023
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2023
- 형태사항
- 165 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Ramprasad, Rampi.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2023.
- 초록/해제
- 요약Novel organic-inorganic hybrid membranes processed through vapor phase infiltration (VPI) incorporate the advantages of both organic and inorganic materials. Compared to conventional organic membranes, these hybrid materials offer significant improvements in stability when exposed to organic solvents while retaining desirable membrane properties such as high permeability and selectivity. However, the extensive design space involved in developing such membranes, which encompasses polymer chemistry, inorganic chemistry, and hybrid microstructures, poses challenges to traditional trial and error methods. To surmount these obstacles, this work develops a more efficient and systematic approach. It involves three steps that leverage density functional theory (DFT) and machine learning (ML) to develop the knowledge and tools necessary to predict and explore novel VPI organic-inorganic membranes:1. This research entails an in-depth investigation into the interactions between three metal precursors and the prototype polymer of intrinsic microporosity 1 (PIM-1) during the VPI process. Our primary objective was to identify crucial characteristics of polymer-inorganic interactions, decipher structure-property relationships, and unveil significant properties that could contribute to ML model predictions for future materials selection. Our work uncovered two atomic-level mechanisms for solvent stability.2. An ML-based tool predicting sublimation enthalpy was developed to aid chemistry selection and experimental design for precursors. Initial training used a comprehensive DFT dataset of organic molecules constructed in this work due to a lack of metal precursor parameters in the literature. As new data emerged, an active learning algorithm incorporated new chemical species into the model, dynamically improving its accuracy and expanding its applicability.3. An ML model, incorporating multi-task learning and meta-learning, was trained on a new DFT dataset to predict binding energy between metal precursors and polymers. This enhanced the understanding of polymer-inorganic interactions' strength and stability, aiding in the selection of potential precursors. The model provides a promising route for informed precursor selection, VPI process optimization, and the design of hybrid materials with custom properties.This foundational work provides automated and effective tools for the design and development of VPI organic-inorganic hybrid membranes, leveraging the combined capabilities of DFT and ML. The predictive models developed here can be employed alongside the insights derived from our atomic-level mechanistic studies in the selection of suitable polymers and metal precursors for designing energy-efficient organic-inorganic hybrid membranes for chemical separation. In addition, the DFT database and ML models developed in this project serve as valuable instruments to be utilized by researchers for future studies on the sublimation enthalpy and binding energy of organic-inorganic systems, facilitating further advancements in the field of material science. This thesis presents and executes a methodical framework through which future models can be developed for the exploration of novel material spaces.
- 일반주제명
- Heat treating
- 일반주제명
- Humidity
- 일반주제명
- Carbon
- 일반주제명
- Adsorption
- 일반주제명
- Water
- 일반주제명
- Hollow fiber membranes
- 일반주제명
- Aluminum
- 일반주제명
- Energy
- 일반주제명
- Thermodynamics
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260203s2023 us c eng d■001000017365955
■00520260209102902
■006m o d
■007cr#unu||||||||
■020 ▼a9798265405272
■035 ▼a(MiAaPQ)AAI32315541
■035 ▼a(MiAaPQ)GeorgiaTech73099
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a000
■1001 ▼aLiu, Yifan.
■24510▼aDesign of Organic-Inorganic Hybrid Membranes Using Density Functional Theory and Machine Learning
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2023
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2023
■300 ▼a165 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Ramprasad, Rampi.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2023.
■520 ▼aNovel organic-inorganic hybrid membranes processed through vapor phase infiltration (VPI) incorporate the advantages of both organic and inorganic materials. Compared to conventional organic membranes, these hybrid materials offer significant improvements in stability when exposed to organic solvents while retaining desirable membrane properties such as high permeability and selectivity. However, the extensive design space involved in developing such membranes, which encompasses polymer chemistry, inorganic chemistry, and hybrid microstructures, poses challenges to traditional trial and error methods. To surmount these obstacles, this work develops a more efficient and systematic approach. It involves three steps that leverage density functional theory (DFT) and machine learning (ML) to develop the knowledge and tools necessary to predict and explore novel VPI organic-inorganic membranes:1. This research entails an in-depth investigation into the interactions between three metal precursors and the prototype polymer of intrinsic microporosity 1 (PIM-1) during the VPI process. Our primary objective was to identify crucial characteristics of polymer-inorganic interactions, decipher structure-property relationships, and unveil significant properties that could contribute to ML model predictions for future materials selection. Our work uncovered two atomic-level mechanisms for solvent stability.2. An ML-based tool predicting sublimation enthalpy was developed to aid chemistry selection and experimental design for precursors. Initial training used a comprehensive DFT dataset of organic molecules constructed in this work due to a lack of metal precursor parameters in the literature. As new data emerged, an active learning algorithm incorporated new chemical species into the model, dynamically improving its accuracy and expanding its applicability.3. An ML model, incorporating multi-task learning and meta-learning, was trained on a new DFT dataset to predict binding energy between metal precursors and polymers. This enhanced the understanding of polymer-inorganic interactions' strength and stability, aiding in the selection of potential precursors. The model provides a promising route for informed precursor selection, VPI process optimization, and the design of hybrid materials with custom properties.This foundational work provides automated and effective tools for the design and development of VPI organic-inorganic hybrid membranes, leveraging the combined capabilities of DFT and ML. The predictive models developed here can be employed alongside the insights derived from our atomic-level mechanistic studies in the selection of suitable polymers and metal precursors for designing energy-efficient organic-inorganic hybrid membranes for chemical separation. In addition, the DFT database and ML models developed in this project serve as valuable instruments to be utilized by researchers for future studies on the sublimation enthalpy and binding energy of organic-inorganic systems, facilitating further advancements in the field of material science. This thesis presents and executes a methodical framework through which future models can be developed for the exploration of novel material spaces.
■590 ▼aSchool code: 0078.
■650 4▼aHeat treating
■650 4▼aHumidity
■650 4▼aCarbon
■650 4▼aAdsorption
■650 4▼aWater
■650 4▼aHollow fiber membranes
■650 4▼aAluminum
■650 4▼aEnergy
■650 4▼aThermodynamics
■690 ▼a0791
■690 ▼a0800
■690 ▼a0348
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2023
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17365955▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


