El Agente Sólido: A New Age(nt) for Solid State Simulations

TL;DR

El Agente Sólido presents a hierarchical multi-agent framework that autonomously plans, executes, and analyzes solid-state first-principles simulations with Quantum ESPRESSO. By translating natural-language scientific goals into end-to-end pipelines that include structure generation, input construction, workflow execution, and post-processing, it integrates DFT with phonon analyses and MLIP surrogates to enable scalable, reproducible materials discovery. The approach is validated through extensive benchmarking (11 questions across 7 exercises) and diverse case studies (electrocatalysis, thermal properties, electrochemistry, MOFs/COFs), achieving a 3–5 sentence high-level assessment of its capabilities, including an average success rate of 97.9% across trials and the ability to produce physically consistent results via $\Delta G = \Delta E_{DFT} + \Delta ZPE - T\Delta S - eU$ in the computational hydrogen electrode framework and quasi-harmonic free-energy analyses $F(V,T) = E_{DFT}(V) + F_{vib}(V,T)$. The work demonstrates a significant step toward autonomous, reproducible, and scalable materials discovery workflows that can accelerate solid-state chemistry research.

Abstract

Quantum chemistry calculations are a key component of the materials discovery process. The results from first-principles explorations enable the prediction of material properties prior to experimental validation. Despite their impact, the practical use of first-principles methods remains limited by the expertise required to design, execute, and troubleshoot complex computational workflows. Even when workflows are successfully built, they are sometimes rigid and not adaptable to different use cases. Recent advances in large language models (LLMs) and agentic systems offer a pathway to flexibly automate these processes and lower barriers to entry. Here, we introduce El Agente Sólido, a hierarchical multi-agent framework for automating solid-state quantum chemistry workflows using the open-source Quantum ESPRESSO simulation package. The framework translates high-level scientific objectives expressed in natural language into end-to-end computational pipelines that include structure generation, input file construction, workflow execution, and post-processing analysis. El Agente Sólido integrates density functional theory with phonon calculations and machine-learning interatomic potentials to enable efficient and physically consistent simulations. Extensive benchmarking and case studies demonstrate that El Agente Sólido reliably executes a wide range of solid-state calculations, highlighting its potential to improve reproducibility and accelerate computational materials discovery

Figures (31)

• Figure 1: Core computational capabilities of El Agente Sólido, spanning structure generation, simulation, analysis, and advanced materials workflows.
• Figure 2: (a) Graphic representation of El Agente Sólido's hierarchical architecture. At the top of the hierarchy is the Computational Chemist Agent (green node), responsible for planning calculation workflows based on a user's request. It is then assisted by four main subagents (nodes connected by solid and dashed arrows) responsible for accomplishing different parts of a calculation workflow: the Geometry Generator Subagent, which generates initial geometries; the DFT Subagent, which creates input files and runs Quantum ESPRESSO and Phonopy; the File I/O Subagent, which creates folders and moves files; and the Output Analyzer Subagent, which analyzes the output of a calculation, the star symbol indicates a connector in the flowchart. (b) Flowchart that outlines a typical calculation workflow executed by El Agente Sólido, the diamond symbol represents a connector in the flowchart.
• Figure 3: Flowchart summarizing the complete workflow used to calculate the theoretical OER overpotential on the Pt(111) surface.
• Figure 4: Panels (a–c) show the phonon band structures of $\alpha$-Fe, NaCl, and Si at equilibrium volume, respectively. Panel (d) refers to the change in heat capacity of Fe, NaCl and Si with temperature at constant pressure and (e) refers to the thermal expansion coefficient of Fe, Si and NaCl. Plots a-c were generated by El Agente Sólido. Plots d and e were modified from plots generated by El Agente Sólido to include experimental thermal expansion coefficients for Si okada1984precise, $\alpha$-Fe kozlovskii2019linear, and NaCl pashaev2013epitaxial as a reference. Figures S11-S25 in the Supplementary Information showcase all plots generated by El Agente Sólido when using Phonopy for QHA analysis.
• Figure 5: Flowchart summarizing the complete workflow used to compute the delithiation voltage profile of Li_xNi_0.8Co_0.1Mn_0.1O_2 from x = 1.0 to 0.5 with intervals of 0.1. The delithiation voltages for experimental comparison were obtained from marker2019evolution. The delithiation voltage plot was modified to include experimental data from a plot generated by El Agente Sólido. Figures S8-S10 contain details that El Agente Sólido took to generate the delithiated voltage curve of NMC-811.