DGLD: Domain-Gated Latent Diffusion for the Discovery of Novel Energetic Materials

2026-05-26

AI Summary

This work addresses the challenges in high-energy material design posed by sparse labels, the tendency of existing generative models to memorize known molecules, and their limited extrapolation capability. To overcome these issues, the authors propose a Domain-Gated Latent Diffusion (DGLD) model that integrates label-quality gating during training, multi-task score guidance during sampling, and leverages both SMILES and SELFIES molecular representations within a four-stage chemical validation pipeline—including density functional theory (DFT) calculations and chemical reasonableness filters. For the first time, this approach simultaneously achieves structural novelty and target performance at DFT accuracy, breaking the conventional trade-off between memorization and extrapolation. The method successfully discovers 12 novel high-energy molecules, including L1 with a density of 2.09 g/cm³ and detonation velocity of 8.25 km/s, and E1 with a detonation velocity of 9.00 km/s, substantially outperforming current baselines.

Abstract

Energetic-materials performance gains translate directly into reduced propellant mass, smaller warheads, and more efficient civilian gas-generators, yet no new HMX-class compound has been disclosed in fifteen years. Designing one is a sparse-label problem: of ~66 k labelled CHNO molecules only ~3 k carry experimental or DFT-quality measurements, and naive generative models trained on the full mixture either memorise the high-performance tail or extrapolate without calibration. We introduce Domain-Gated Latent Diffusion (DGLD): a label-quality gate at training time, multi-task score-model guidance at sample time, and a four-stage chemistry-validation funnel ending in first-principles DFT audit. The result is 12 DFT-confirmed novel leads. The headline compound, 3,4,5-trinitro-1,2-isoxazole (L1), reaches \r{ho}_"cal" =2.09 g/cm3 and D_"K-J,cal" =8.25 km/s and is structurally dissimilar from all 65 980 training molecules (nearest-neighbour Tanimoto 0.27). A co-headline lead, E1 (4-nitro-1,2,3,5-oxatriazole), exceeds L1 on calibrated detonation velocity (D_"K-J,cal" =9.00 km/s) from a chemotype family disjoint from L1's. DGLD is the only method to land in the productive quadrant (simultaneously novel and on-target) at DFT level. SMILES-LSTM memorises 18.3% of its outputs exactly; SELFIES-GA's best novel candidate loses 3.5 km/s under DFT audit; REINVENT 4 generates novel high-N heterocycles but peaks at D=9.02 km/s. Code, checkpoints, and 918 mined hard negatives are released on Zenodo (DOI 10.5281/zenodo.19821953); the next compound to enter the HMX-class band can be discovered, validated, and recommended for synthesis at the cost of a few GPU-days.

Problem

• energetic materials
• sparse-label problem
• novel compound discovery
• high-performance prediction
• molecular design

Innovation

• Domain-Gated Latent Diffusion
• energetic materials discovery
• sparse-label generative modeling
• multi-task score guidance
• DFT validation