Fundamental Factors Governing Stabilization of Janus 2D-Bulk Heterostructures with Machine Learning
Tara M. Boland, Rachel Gorelik, Arunima K. Singh · 2025
Read Fundamental Factors Governing Stabilization of Janus 2D-Bulk Heterostructures with Machine Learning by Tara M. Boland, Rachel Gorelik, Arunima K. Singh (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-04603.txt.
Introduction
Tara Boland, Rachel Gorelik, and Arunima Singh ask what makes layered stacks of two-dimensional and bulk materials hold together, in a 2025 paper. Over six thousand 2D materials are predicted to exist, and pairing them with bulk solids could yield devices for electronics, sensing, and energy conversion. The team focused on Janus materials, 2D sheets whose two faces differ, joined to metal surfaces.
Using quantum simulations in a high-throughput pipeline, they built 1,147 candidate stacks from fifty-one Janus sheets and nineteen metals. Of these, 828 proved stable. Machine learning models trained on the results predicted binding strength and layer spacing with small errors. The models showed that the bulk partner's properties dominate the stack's energy and structure. The full dataset is released in a public database, and the authors offer the models as a shortcut for screening thousands of untested combinations.
Catalog record
| CREATOR | Tara M. Boland, Rachel Gorelik, Arunima K. Singh |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-04603
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fundamental Factors Governing Stabilization of Janus 2D-Bulk Heterostructures with Machine Learning Tara M. Boland, Rachel Gorelik, Arunima K. Singh Abstract The more-than-6000 2D materials predicted thus far provide a huge combinatorial space for forming functional heterostructures with bulk materials, with potential applications in nanoelectronics, sensing, and energy conversion. In this work, we investigate nearly 1000 heterostructures, the largest number of heterostructures thus far, of 2D Janus and bulk materials’ surfaces using ab initio simulations and machine learning (ML) to deduce the structure-property relationships of the complex interfaces in such heterostructures. We first perform van der Waals-corrected density functional theory simulations using a high-throughput computational framework on 51 Janus 2D materials and 19 metallic, cubic phase, elemental bulk materials that exhibit low lattice mismatches and low coincident site lattices. The formation energy of the resultant 1147 Janus 2D-bulk heterostructures were analyzed and 828 were found to be thermodynamically stable. ML models were trained on the computed data, and we found that they could predict the binding energy and z z -separation of 2D-bulk heterostructures with root mean squared errors (RMSE) of 0.05 eV/atom and 0.14 Å, respectively. The feature importance of the models reveals that the properties of the bulk materials dominate the heterostructures’ energies and interfacial structures heavily. These findings are in-line with experimentally observed behavior of several well-known 2D materials-bulk systems. The data used within this paper is freely available in the Ab Initio 2D-Bulk Heterostructure Database (aiHD). The fundamental insights on 2D-bulk heterostructures and the predictive ML models developed in this work could accelerate the application of thousands of 2D-bulk heterostructures, thus stimulating research within a wide range of electronic, quantum computing, sensing, and energy applications. keywords American Chemical Society, L a T e X † † affiliation: Computational Atomic-Scale Materials Design (CAMD), Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark † † affiliation: Authors contributed equally † † affiliation: School for Engineering of Matter, Transport and Energy, Arizona State University, Tempe, AZ, USA † † affiliation: Authors contributed equally † † affiliation: Department of Physics, Arizona State University, Tempe, AZ, USA † † email: [email protected] † † suppinfo: T 1 Introduction Two-dimensional (2D) materials are a class of atomically thin materials, ranging from single-to-few atomic layers in thickness, with a wide variety of material properties resulting from quantum confinement effects. Many 2D materials have been experimentally synthesized since the discovery of graphene in 2004 Novoselov et al. 2004 , with various methods subsequently explored for tuning their properties, such as strain engineering, doping, and forming heterostructures (either with other 2D materials or with bulk materials) Singh et al. 2015 ; Singh et al. 2014 ; Houlong L. Zhuang 2017 ; Smirman et al. 2017 ; Paul et al. 2017 ; Singh and Hennig 2014 ; Singh et al. 2015 ; Singh et al. 2015 . Prior works have studied 2D-2D heterostructures but there has been little exploration of the potentially differing phenomena of 2D-bulk interfaces Pakdel et al. 2024 ; Zhang et al. 2023 ; Rhodes et al. 2019 ; Akiyama et al. 2021 ; Özçelik et al. 2016 ; Geim and Grigorieva 2013 ; Willhelm et al. 2022 ; Zeng and Liu 2018 . Heterostructures of 2D and bulk materials could provide a huge combinatorial space for the formation of functional heterostructures that will facilitate the low-barrier integration of novel 2D materials with existing bulk technologies such as nanoelectronics, quantum computing, chemical sensing, and energy conversion. More recently, the high-throughput ab initio /theory-based approach to materials discovery has led to significant advancements in 2D materials–i.e. the prediction of over 7000 2D materials and the experimental realization of over 5 % of these materials. Haastrup et al. 2018 ; Ashton et al. 2017 ; Özçelik et al. 2016 ; Choudhary et al. 2017 ; Zhang et al. 2019 ; Campi et al. 2022 Crystal structure and properties of 2D materials have been systematically curated in more than five open-source databases Haastrup et al. 2018 ; Ashton et al. 2017 ; Özçelik et al. 2016 ; Choudhary et al. 2017 ; Zhang et al. 2019 ; Campi et al. 2022 , guiding future studies towards further experimental realization and application of 2D materials. Similarly, there are currently more than 10 databases that curate the structural, electronic, mechanical, and optical properties of bulk materials obtained from ab initio methods such as density functional theory (DFT) that also facilitate machine learning (ML) and data-driven sciences. Curtarolo et al. 2012 ; Jain et al. 2013 ; Kirklin et al. 2015 ; of Standards and Technology However, there are no systematic high-throughput simulations or open-source databases of 2D-bulk heterostructures. Figure 1: 1H -Janus 2D-bulk heterostructure model showing the orientation of the 2D material on top of the bulk surface. The Y-M-X represents the atomic layers of the Janus 2D material. In this report, we introduce the Ab Initio 2D-Bulk Heterostructure Database (aiHD), the first freely available database of ab initio -calculated data on 2D-bulk heterostructures and their post-adsorbed properties. To construct, optimize, and analyze the 2D-bulk heterostructures, we use Hetero2d Boland and Singh 2022 , a high-throughput computational synthesis package designed for computing heterostructure properties within the DFT framework. We explore the interaction of metastable 2D materials with various bulk materials’ surfaces through DFT calculations for a set of 51 Janus 2D materials and 19 elemental bulk substrates in lattice- and symmetry-matched 2D-bulk heterostructures. Janus 2D materials with formula Y-M-X – which are 3-atomic-layers-thick, with each atomic layer consisting of a distinct element – exhibit unique properties, including finite out-of-plane dipole moments, in-plane piezoelectricity Riis-Jensen et al. 2019 ; Zhang et al. 2020 , and magnetic skyrmions Hou et al. 2022 . From our computed dataset of 1147 Janus 2D-bulk pairs, we identify 828 2D-bulk configurations that stabilize the metastable Janus 2D materials. ML random forest regression models developed with this ab-initio -computed data predict the binding energy and z -separation distance for 2D-bulk heterostructures with low root mean squared errors (RMSE) of 0.047 eV/atom and 0.139 Å , respectively. Through these models, the bulk material’s electronegativity and surface energy are each found to be the leading fundamental factor that provides bulk-induced stabilization of a 2D-bulk pair’s binding energy and z -separation distance, respectively. This work makes a major advancement in the understanding of heterostructures of 2D materials and bulk materials. It is poised to strongly guide the experimental realization and utilization of the thousands of previously unsynthesized 2D materials for a wide variety of applications such as photocatalysis, electronics, and batteries. The ML models developed in this work will significantly reduce the computational costs of extending such studies to other 2D-bulk heterostructures. Furthermore, the aiHD and the ML models can facilitate targeted searches of large combinatorial spaces of heterostructures for the exploration of stable 2D-bulk pairs and their corresponding properties. 2 Results and Discussion 2.1 Screening Parameters and Materials Selection 2.1.1 Materials Selection The 51 Janus 2D materials, Y-M-X, (Y=[S, Se, Te], M=[As, Bi, Cr, Hf, Mo, Nb, Sb], and X=[Br, Cl, I, Se, Te]) were obtained from the Computational 2D Materials Database (C2DB) Haastrup et al. 2018 , which contains the DFT-computed structures of these Janus 2D materials which exhibit a diverse range of predicted properties. Two different Janus 2D