A fast machine learning tool to predict the composition of interstellar ices from infrared absorption spectra

¹ Centro de Astrobiología (CAB), CSIC-INTA, Carretera de Ajalvir, km 4, 28805 Torrejón de Ardoz, Spain
² Facultad de Ciencias Físicas, Universidad Complutense de Madrid, 28040 Madrid, Spain
³ LIRA, CY Cergy Paris Université, Mail Gay-Lussac, 5, 95000 Neuville-sur-Oise, France
⁴ Instituto de Estructura de la Materia (IEM), CSIC, Calle Serrano, 121, 28006 Madrid, Spain
⁵ MasOrange, Paseo del Club Deportivo, 1, 28223 Pozuelo de Alarcón, Spain
⁶ Laboratory for Astrophysics, Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, The Netherlands

^★ Corresponding author: amegias@cab.inta-csic.es

Received: 31 December 2024
Accepted: 14 August 2025

Abstract

Context. Current observations taken by James Webb Space Telescope (JWST) allow us to observe the absorption features of icy mantles that cover interstellar dust grains, which are mainly composed of H₂O, CO, and CO₂, along with other minor species. Thanks to its sensitivity and spectral resolution, JWST has the potential to observe ice features towards hundreds of sources at different stages along the process of star formation. However, identifying the spectral features of the different species and quantifying the ice composition is not trivial and requires complex spectroscopic analysis.

Aims. We reduce the difficulty and the time employed in doing this task by developing a new software analysis tool based on machine learning.

Methods. We present Automatic Ice Composition Estimator (AICE), a new tool based on artificial neural networks. Based on the infrared (IR) ice absorption spectrum between 2.5 and 10 μm, AICE predicts the ice fractional composition in terms of H₂O, CO, CO₂, CH₃OH, NH₃, and CH₄. To train the model, we used hundreds of laboratory experiments of ice mixtures from different databases, which were reprocessed with baseline subtraction and normalisation.

Results. Once trained, AICE takes less than one second on a conventional computer to predict the ice composition associated with the observed IR absorption spectrum, with typical errors of ~3% in the species fraction. We tested its performance on two spectra reported towards the NIR38 and J110621 background stars observed within the JWST Ice Age program, demonstrating a good agreement with previous estimations of the ice composition.

Conclusions. The fast and accurate performance of AICE enables the systematic analysis of hundreds of different ice spectra with a modest time investment. In addition, this model can be enhanced and re-trained with more laboratory data, improving the precision of the predictions and expanding the list of predicted species.