About (Updated: 2024-07-08)
My name is Longkun Xu, people also call me Luke.
I earned my PhD in computational chemistry at the Australian National University in 2022, under the guidance of Prof. Michelle Coote. After that I joined Samsung Research China-Beijing (SRC-B) as an engineer. My research interests focus on method development of computational chemistry and computational materials, AI for Science, HPC and quantum computing, as well as the application of their related algorithms in industry scenarios such as batteries, OLEDs, semiconductors, and drugs. I published 10+ papers in journals including JACS, Nature Communication, JCTC, JPCL, JPCC, JPCA. My research was highlighted in media including AI for Science Global Outlook 2023 Edition, EurekAlert!, PHYS.ORG, Chemistry in Australia. I server as a reviewer for ICML(2024, 2023), NeurIPS(2023, 2022), JCTC, and JPCA.
You can find my blog posts here, which are some tutorials used to answer questions I have been frequently asked, feel free to let me know your comments and questions.You can contact me via longkunx@gmail.com
Education and Employment
[12/2021]-[07/2024], Engineer (05/2022 - 07/2024) <- Intern (12/2021 - 05/2022), Samsung Research China-Beijing (SRC-B), Beijing, China
[10/2018]-[04/2022], PhD student (Computational Chemistry), Australian National University, Canberra, Australia
[07/2018]-[09/2018], Assistant Editor, MDPI publisher, Beijing, China
[09/2015]-[07/2018], Master’s degree (Applied Chemistry), Sichuan University, Chengdu, China
[09/2011]-[07/2015], Bachelor’s degree (Material Chemistry), Qingdao Agricultural University, Qingdao, China
Research Interests
More introductions to my current and previous research projects can be found in my PhD thesis and our book chapter.
Following are some related keywords often appeared in my papers:
- Computational Chemistry: quantum chemistry, MD simulation, multi-scale calculation, …
- AI for Science: machine learning potentials, generative AI for molecular generation, …
- Quantum Computating and HPC: quantum computing for quantum chemistry, HPC, …
- Molecular and Materials Science:
- Solvation: electrolyte solution, ionic liquids, electrostatic and polarization interactions, external electric field, …
- Materials: battery, OLED, semiconductors, …
- Biology: drug, peptide, protein, synthetic accessibility, retrosynthesis, …
Research Value: I think a good computational method development should be theoretically fundamental (use math and physics to solve issues, not by tuning parameters), easy to use (black-box, automatic and open-source) and generally effective for most chemical systems (versatile). A good calculational applications should aim to offer new physical insights and guide the design of products.
I am particularly interested in the following topics:
1. AI for Science
Specifically my current interests in AI for Science include machine learning potential (with long-range interactions and field effects), automatic differentiation algorithms, deep generative models, graph theory, and their applications into interface modelling, enhanced sampling method, force field optimization, molecular generation, material property prediction and chemical reaction networks, etc.
Particularly I am interested in developing data efficient deep learning models. With my co-workers, we propose two strategies: stability indicated sampling (SIS) and using multi-fidelity training data.
For the first one, we developed SIS-MLPMD, a new method to sample training data of machine learning potentials based on temperature stability in MD simulation. Applied the method to study the long-time reactive dynamics of SEI formation process at the interface of Lithium metal batteries, see 2023 JPCC.
For the second one, we developed a method to use multi-fidelity training data of JAX-ReaxFF and DFT to reduce the computational costs in training SchNet and MACE based machine learning potentials. The method was employed to study a diverse range of properties of 2D semiconductor materials, see 2024 JPCL.
2. Computational Solvation Science
2.1. Structures and properties of ionic liquids under external electric field;
Structures and properties of ionic liquids under external electric field is closely related with the design of ionic liquids based electrochemical devices. In 2021JACS, we analysed details of this topic using both computational and experimental techniques. A plateau in the open circuit potential (OCP) was observed upon application of an external electric field and its removal, which persisted in some ionic liquids over several hours. We propose this method as an easy and straightforward technique to characterize and compare the degree of order of different ionic liquids. Several parameters were also calculated using the trajectory of the polarizable molecular dynamic simulations to characterise the ordering of different ionic liquids. By considering ion dipole projection, the diffusion coefficient of the cation and its volume, a good correlation was found between the measured OCP and the calculated quantities for different ionic liquids, which offers a way to choose suitable ionic liquids for electrochemical applications.
2.2. Electrostatic catalysis in unusual solvent environment;
Achieving electrostatic catalysis in polar solvent is highly desirable, especially in some unusual and complex solvent environment where the mean field approximation is invalid. For example, under some conditions, the normal solvent networks can be broken and the ordered solvent is formed, this is a very interesting phenomenon. In 2020JACS working with the group of Prof.Ekaterina I Pas employing classical molecular dynamic simulations with the Drude oscillator-based polarizable force field, quantum chemical calculations, and ONIOM (DFT:semi-empirical) multiscale calculations, we investiagted the potential of ordered solvent and ionic liquids in catalysing chemical reactions. Bubble surface is another unique electrostatic environment where high concentration of excess OH- exist, in 2020 Nat. Commun., we use GFN-XTB method based MD simulation and ONIOM(CCSD(T)/CBS:DHDFT) methods to help experimental collaborators (Dr Simone Ciampi group) prove that the high concentration of OH- can promote the oxidation potential of itself, which might be useful for further works in the fields of electrocatalysis and electrosynthesis.
2.3. Improving the accuracy of implicit solvent models;
The implicit solvent models together with quantum chemical methods is the main strategy in modelling solution-phase properties and reactions. Thus, its accuracy in producing the solvation free energies is very important. In 2019JPCA, we disccussed different methods to improve the SMD solvation free energies and associated pKa values. In 2019JCTC, we presented the optimal electrostatic scaling factor (ESF) values in a wide range of solvents using PCM-UAHF and PCM-UAKS methods, we proposed the mixed-ESF method to improve the accuracy in calculating solution-phase properties.
2.4. Non-equilibrium solvation and solvent reorganization;
Non-equilibrium solvation is an important conecpt originaly proposed by Prof. Rudolph A. Marcus in his well-known Marcus theory. It is a key factor in modelling the superfast processes in solution phase, for example, electron transfer and electronic excitation. Prof. Xiang-Yuan Li and co-workers have been working on implementing the constrained-equilibrium method in the non-equilibrium solvation theory. In 2017CPL, we derived the expressions of non-equilibrium solvation free energy within the framework of analytical Onsager model and applied it to model charge-transfer excited state. The theory was developed and implemented in the local Q-Chem program, see 2017PCCP, 2018 PCCP HOT article and 2017CP.
The full list of my publications can be found from the google scholar in the left sidebar.
Peer Review Service
Reviewers for the following journals and conferences: ICML 2024 workshop (AI4Science, SPIGM), NeurIPS2023 workshop (AI4Science, GenBio), ICML 2023 workshop (SPIGM), NeurIPS2022 workshop (AI4Science), JCTC, JPCA
Talks and Posters
[12/2020], ASIL9, poster and video, Zoom
[10/2019], APATCC2019, poster, Sydney, Australia
[06/2017], ncqc2017, poster, Dalian, China
Awards and Fellowships
Samsung Excellent Proposal Award (2023)
Postgraduate Research Support (2020)
HDR Fee Remission Merit Scholarship (2018-2021)
ANU PhD Scholarship (International) (2018-2021)
Second Class Scholarship for Gruduate Student (2015-2018)
Hailier Scholarship for Outstanding Students (2013)
Teaching Service
Mentored Interns:
- Kehan Wang (2022-2023 @Samsung, Now: PhD student @Tsinghua)
- Mingwei Ge (2022 @Samsung, Now: PhD student @Yale)
Teaching assistant for undergraduate course Physical Chemistry (2016)
News
[8/2023] Our 2023 JPCC was reported by AI for Science Global Outlook 2023 Edition, see page 179 in here
[12/2021] Our 2021 JACS was reported by Chemistry in Australia, see page 15 in here
[12/2020] Our 2020 Nat. Commun. was reported by several news platforms including EurekAlert!, PHYS.ORG and Chemistry in Australia.
[11/2020] Our 2020 JACS paper was reported by Chemistry in Australia, see page 14 in here