Inverse Design of Inorganic Compounds with Generative AI
Authors: Hannes Kneiding, Lucía Morán-González, Nishamol Kuriakose, Ainara Nova, David Balcells
Organizations: Hylleraas Centre for Quantum Molecular Sciences, Department of Chemistry, University of Oslo, P.O. Box 1033, Blindern, 0315 Oslo, Norway
Machine learning is revolutionizing chemistry. Beyond the value of predictive models accelerating virtual screening, generative AI aims at enabling inverse design, reversing the compound-to-property prediction paradigm into property-to-compound generation. Chemists now have access to a rich AI toolbox for organic chemistry, including drug discovery. However, the application of these methods to inorganic compounds remains limited by the challenges posed by their intrinsic nature. This Review analyzes how these challenges have been addressed, considering widely diverse systems ranging from molecules to crystals, including transition metal complexes and microporous materials. The analysis focuses on how generative AI methods have evolved towards data-representation-model pipelines that address the full complexity of inorganic compounds, including their chemical composition, geometry, symmetry, and electronic structure. Future directions, like benchmark standardization and the development of synthesizability metrics, are also discussed.
Figures & tables
Figure 1: Inorganic compounds covered in this Perspective. a) Transition metal complexes are built around a metal center supported by a set of ligands. b) Microporous inorganic materials have porous reticular structures made with metal node-linker (MOFs) or aluminosilicate (zeolites) building blocks. In MOFs, the metal node can also be a single-atom cation. c) Non-porous inorganic crystals form a compact, periodic structure made of metals or ions.
Name a
Size
Contents
Ref.
CSD
1.4M
Experimental composition and structure data
Groom et al. (2016)
COD
0.5M
Experimental composition and structure data
Gražulis et al. (2011)
ICSD
328k
Experimental & computational composition and structure data
Zagorac et al. (2019)
PCD
395k
Experimental & computational composition and structure data
Villars and Cenzual (2024)
MP
785k
Computational structure, quantum properties, and thermodynamic data
Jain et al. (2013)
OQMD
1.4M
Computational structure, quantum properties, and thermodynamic data
Saal et al. (2013)
Table 1: Datasets available for training generative AI models for inorganic compounds.
Representations
Format a
Connect.
Isomerism
3D geom.
Periodic
DL/GA
SMILES b
String
Yes
Partial c
No
No
Both
SELFIES b
String
Yes
Partial c
No
No
Both
SLICES
String
Yes
Yes
Yes
Yes
Both
Graph d
Graph
Yes
Optional e
Optional e
Yes
Both
3D grid f
Point-wise grid
No
Yes
Yes
Yes
DL
Voxel f
Cubic grid
No
Yes
Yes
Yes
DL
Table 2: Selected representations for generative AI methods for inorganic compounds.
Figure 2: Timeline of selected generative AI methods, including genetic algorithms (GAs), generative adversarial networks (GANs), variational autoencoders (VAEs), diffusion models (DMs), and large language models (LLMs), for inorganic compounds, including transition metal complexes (TMCs), non-porous crystals, metal organic frameworks (MOFs), and zeolites.
Figure 3: Selected generative AI methods for TMCs. a) GA using genetic operations, fitness optimization, and selection in Pareto front problems; b) The PL-MOGA GA for directional multi-objective optimization over selected regions of the Pareto front; c) VAE encoding and decoding metal ligands through a compressed latent space representation; d) The CatDRX VAE encoding both TMCs and reaction conditions for the inverse design of novel catalysts.
Figure 4: Diffusion models for non-porous inorganic crystals. a) Noising using of crystal structure representation with Gaussian noise and learned denoising with E(3)-equivariant graph neural networks; b) The MatterGen diffusion model doing unconditional generation with nearly DFT quality and conditional generation on multiple properties, using SUN metrics for evaluation and experiments for verification.
Figure 5: LLM models for MOFs. a) Whereas the classical approach encodes the materials information with tokens, quantum natural language processing (QNLP) uses a qubit representation. b) The generative process leverages a four-category classification task (low, moderate low, moderate high, and high) over two target properties (pore volume and CO 2 Henry’s constant).
Designing novel inorganic materials through generative models remains an important challenge for material science, driven by the complexity and diversity of inorganic structures across expansive chemical compositions and structural landscape. The vast combinatorial space of inorganic compounds demands innovative, AI-driven approaches to overcome limitations in generative accuracy and efficiency. To address this, we introduce a novel method that redefines the encoding and generation of inorganic materials by utilizing domain-specific symmetry-aware representation. Our approach not only refines the representation of intricate inorganic structures but also contributes to the field of material discovery by enhancing the precision and stability of generated candidates. Central to our methodology is a novel padding technique that exploits crystal symmetry information to enhance the encoding process. By integrating Wyckoff position length-aware padding into an encoder architecture, we achieve a more robust informed representation of inorganic materials. This symmetry-driven enhancement improves deep learning models to generate stable, previously unexplored inorganic structures with superior accuracy and computational efficiency. Furthermore, we introduce an end-to-end system that leverages the machine learning potential models to seamlessly generate novel, even those unseen in the training data, and stable inorganic materials from initial data to validated output. This pipeline integrates advanced generative models with stability analysis, marking a significant leap forward in the automated exploration and design of next-generation inorganic materials. Our method improved reconstruction accuracy 5.3% in proton conductor data, and generated 63.5% more novel stable inorganic material to baseline model on the perov-5 dataset.
Modern generative machine learning (ML) models can propose novel inorganic crystalline materials with targeted properties; however, synthesis planning of these materials remains difficult due to the complexity of the associated physical processes and limited availability of computational tools. We introduce a novel hybrid framework to evaluate Large Language Models (LLMs) in inorganic synthesis planning by combining thermodynamic databases with simplified kinetics models to approximate realistic synthesis conditions. As a case study, we focus on the niobium-oxygen system, which features multiple industrially relevant oxide phases with well-characterized data. In computational simulations, we compare LLM-generated synthesis routes with classical path-planning algorithms, showing that the implicit priors in LLMs can yield more viable strategies. In our evaluation setting, classical search methods serve primarily as a foil rather than a direct competitor. This illustrates the relative complexity of the problem and highlights where the LLM's implicit priors add value.
Generative artificial intelligence is rapidly transforming materials design by enabling de novo exploration of immense chemical spaces. Yet a large proportion of AI-generated compositions remain implausible, violating established chemical principles, which limits the reliability and interpretability of generative materials design. Here, we introduce a chemical validity operator that recasts heuristic chemical rules as a configurable algorithmic prior for evaluating and guiding generative materials discovery. Built on the open-source SMACT package, a data-informed oxidation-state model exposes tunable thresholds, allowing users to interpolate continuously between permissive and conservative chemical constraints, while supporting both exploratory and conservative materials-design workflows. Benchmarking six state-of-the-art generative models for inorganic crystals shows that most reproduce stoichiometry but under-represent realistic oxidation-state combinations, and that filtering removes compositions reliant on rarely observed oxidation states while preserving low-energy compounds near the convex hull. Beyond screening, the same operator can also serve as a reinforcement-learning reward, steering a latent diffusion model towards chemically grounded compositions. By encoding chemical heuristics and observations, this work establishes a foundation for oxidation-state-aware generative models.
Kinga O. Mastej, Panyalak Detrattanawichai, Hyunsoo Park +3
Department of Materials, Imperial College London, London SW7 2AZ, United Kingdom