A Study of Two-Dimensional Single Atom-Supported MXenes as Hydrogen Evolution Reaction Catalysts Using DFT and Machine Learning

11 May 2021, Version 1
This content is a preprint and has not undergone peer review at the time of posting.


Enclosed you will find the article entitled “A study of two-dimensional single atom-supported MXenes as hydrogen evolution reaction catalysts using DFT and machine learning”.

Existing studies predominantly focused on the hydrogen evolution reaction (HER) activities and stabilities of oxygen-terminated MXenes with single-atom loading. However, to the best of our knowledge, two-dimensional (2D) MXenes with different terminations (e.g. Br, I, Se, Te, B, Si, P, and NH) have not yet been investigated for the purposes of HER catalysis. Therefore, in this work, we considered the combined effect of the different surface terminations (B, NH, O, F, Si, P, S, Cl, Se, Br, Te, and I) and single atom loading (Ti, V, Fe, Co, Ni, Cu, Zr, Nb, Mo, Tc, Ru, Rh, Pd, Ag, Hf, Ta, W, Re, Os, Ir, Pt, and Au) using DFT calculation. Gibbs free energy of hydrogen adsorption (reflecting activity) and the cohesive energy (a proxy for thermal stability) of these structures (264 in total) were calculated. We demonstrate that 21 uninvestigated 2D single-atom MXene catalysts, among 264 promising candidates, show an electrocatalytic activity surpassing that of platinum and a thermal stability surpassing those of synthesized borophene sheet and MoS2. Moreover, all catalysts examined in this work were further randomly separated into training and test sets with a ratio of 7:3. The HER electrocatalytic performance and thermal stability of the catalysts in the test set were predicted by machine learning algorithms. Most importantly, we present a way to provide a comparable precision (root mean square error values for the activity and thermal stability predictions are 0.158 eV and 0.02 eV, respectively) to the published machine learning works by avoiding their adoption of complex electronic features and the associated high computational cost, and by only using features that are easily available in chemical repositories. The algorithms used in this work are expected to help future researchers quickly screen single atom loaded MXenes HER catalysts at the initial design stage in a cost-effective manner.

We have no financial interest in the subject or instrumentation used and there is no known conflict of interest.


machine Learning Methods Enable Predictive Modeling

Supplementary materials

Supplementary information


Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.