Accurate large-scale simulations of siliceous zeolites by neural network potentials
Open accessAbstract
Abstract The computational discovery and design of zeolites is a crucial part of the chemical industry. Finding highly accurate while computational feasible protocol for identification of hypothetical siliceous frameworks that could be targeted experimentally is a great challenge. To tackle this challenge, we trained neural network potentials (NNP) with the SchNet architecture on a structurally diverse database of density functional theory (DFT) data. This database was iteratively extended by active learning to cover not only low-energy equilibrium configurations but also high-energy transition states. We demonstrate that the resulting reactive NNPs retain DFT accuracy for thermodynamic stabilities, vibrational properties, as well as reactive and non-reactive phase transformations. As a showcase, we screened an existing zeolite database and revealed >20k additional hypothetical frameworks in the thermodynamically accessible range of zeolite synthesis. Hence, our NNPs are expected to be essential for future high-throughput studies on the structure and reactivity of siliceous zeolites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The total NNP errors are similar to other state-of-the-art (rotationally invariant) MLPs for the modeling of reactive events.
“The total NNP errors are similar to other state-of-the-art (rotationally invariant) MLPs38,42,44 for the modeling of reactive events.”
From A reactive neural network framework for water-loaded acidic zeolites (2024) · cites this paper as [38]
The NNP errors are not larger than the errors between two different flavors of (meta)GGA exchange-correlation DFT functionals.
“Note also that such errors are not larger than the errors between two different “flavors” of the (meta)GGA exchange-correlation DFT functionals (see, e.g., Supplementary Table 4 in ref. 38 comparing the energy and force errors of PBE and SCAN functionals).”
From A reactive neural network framework for water-loaded acidic zeolites (2024) · cites this paper as [38]
These errors are similar to other reactive (rotationally invariant) MLPs and about the same as for previously developed silica NNPs.
“These errors are similar to other reactive (rotationally invariant) MLPs42,44,52 and about the same as for our previously developed silica NNPs38.”
From A reactive neural network framework for water-loaded acidic zeolites (2024) · cites this paper as [38]
A set of seven silica structures was optimized at the NNP level to check whether the current NNPs have the same accuracy as previously developed NNPs for silica and to evaluate how much NNP energy/force errors change the structure and energetics of DFT-optimized zeolites.
“This set of structures was chosen to check whether the current NNPs have the same accuracy as our previously developed NNPs for silica38 and to evaluate how much the NNP energy/force errors change the structure and energetics of the DFT-optimized zeolites.”
From A reactive neural network framework for water-loaded acidic zeolites (2024) · cites this paper as [38]
The citation lists haven’t been fetched yet
References 109 · Cited by 58
- Authors
- Erlebach, A., Nachtigall, P., & Grajciar, L.
- Affiliations
- Charles University (CZ)
- Venue
- npj Computational Materials
- Published
- 2022-08-19
- Volume
- 8
- Issue
- 1
- ISSN
- 2057-3960 (ISSN-L)
- OpenAlex
- W3131437635
- Cited by
- 58
- References
- 109
- Citations by year
- 2026: 11 · 2025: 13 · 2024: 17 · 2023: 12 · 2022: 3 · 2021: 2
- FWCI
- 3.16
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- npj Computational Materials (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University) (cc-by)
- Funders
- Ministerstvo Školství, Mládeže a Tělovýchovy (CZ.02.1.01/0.0/0.0/15_003/0000417, ID:90140)
- Univerzita Karlova v Praze (PRIMUS/20/SCI/004)
- Grantová Agentura České Republiky (19-21534S)
- Topics
- Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
- Physical Sciences › Chemistry › Inorganic Chemistry › Zeolite Catalysis and Synthesis
- Physical Sciences › Physics and Astronomy › Atomic and Molecular Physics, and Optics › Advanced Chemical Physics Studies
- Keywords
- siliceous zeolites, neural network potentials, SchNet, active learning (machine learning), phase transformation, thermodynamic stability, vibration characteristics, reactive potentials, high-throughput screening
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
