CO Adsorption Sites on Interstellar Water Ices Explored with Machine Learning Potentials. Binding energy distributions and snowline

TL;DR

This study addresses how CO binds to interstellar ASW by deriving a statistically robust binding-energy distribution using a Gaussian-Moment Neural Network potential trained on high-quality DFT data. The approach combines SAPT energy decomposition, large-scale ASW surface models (both porous and non-porous), and extensive binding-site sampling to map BE_d across thousands of adsorption sites, revealing Gaussian-like distributions with mean BE near 900 K and substantial site-to-site variability. The results reconcile experimental TPD trends and demonstrate that incorporating the full BE distribution broadens the CO snowline in protoplanetary disks, impacting gas–ice partitioning and anticipated organic inventories. The work highlights how surface morphology, dangling-OH density, and binding motif heterogeneity shape adsorption, diffusion, and desorption processes in astrochemical environments, with broader implications for modeling molecular evolution in diverse astrophysical regions.

Abstract

Context. Carbon monoxide (CO) is arguably the most important molecule for interstellar organic chemistry. Its binding to amorphous solid water (ASW) ice regulates both diffusion and desorption processes. Accurately characterizing the CO binding energy (BE) is essential for realistic astrochemical modeling. Aims. We aim to derive a statistically robust and physically accurate distribution of CO BEs on ASW surfaces, and to evaluate its implications for laboratory temperature-programmed desorption experiments and interstellar chemistry, with a focus on protoplanetary disks. Methods. We trained a machine-learned potential (MLP) on 8321 density functional theory (DFT) energies and gradients of CO interacting with differently-sized water clusters (22-60 water molecules). The DFT method was selected after extensive benchmark. With this potential we built realistic non-porous and porous ASW surfaces, and computed a BE distribution. We used symmetry-adapted perturbation theory to rationalize the interaction of CO on the different binding sites. Results. We find that both ASW morphologies yield similar Gaussian-like BE distributions with mean values near 900 K. However, the nature of the binding interactions is rather different and critically depends on surface roughness and dangling-OH bonds. Simulated TPD curves reproduce experimental trends across several coverage regimes. From an astrochemical point of view, the application of the full BE distribution has a dramatic influence on the CO distribution in protoplanetary disks, leading to a broader CO snowline region, improving predictions of CO gas-ice partitioning, and suggesting an equally broader distribution of organics in these objects.

Figures (13)

• Figure 1: CO-W_2-3 model systems that we used for the DFT energy and geometry benchmark. Rows report structure names and binding energies (BE). The numbers in parenthesis indicate the ZPVE corrected BE value. The water molecule acting as H-bond donor to CO is labeled $\mathrm{W_D}$. Characteristic distances are in Å; energy values are in K. Color code is black for C, red for O, white for H.
• Figure 2: Upper panel: top view of one of the non-porous (npASW) (left) and porous (pASW) ice models. Each periodic surface contains 500 water molecules. Lower panel: altitude plot relative to the same two structures (see Appendix \ref{['sec:ap_asw']}). Periodic cell dimensions are included. Distances in Å.
• Figure 3: Histograms of the ZPVE corrected BEs obtained for CO adsorbed on five npASW (upper panel) and five pASW surfaces (lower panel). The color‐coded bars below each panel indicate the BE groups (VL: Very Low, Low, Medium, High, and VH: Very High). The Gaussian fits were obtained using the bootstrap procedure detailed in Appendix \ref{['sec:ap_bootstrap']}, that propagates the individual uncertainties from the neural-network models into the Gaussian fitting procedure.
• Figure 4: CO-npASW panel: Left: BE distribution for CO on npASW ice, with color mapping highlighting the dominant interaction contributions at various binding sites, ranging from primarily electrostatic (Elst-Class, blue) to purely dispersion (Disp-Class, orange), while ED-Class (grey) stands for an intermediate group. The coefficient that determines to which class the binding site belongs is the ratio between the dispersion energy and the sum of the attractive interaction energies (electrostatic, induction and dispersion), and is defined in Sec. \ref{['sec:ie']}. Right: correlation between the BE and CO number of neighbors (i.e. number of water molecules in a radius of 4.5 Å from CO center of mass) for each class of binding sites. Averages are represented as points connected by a dashed line and the standard deviation is shown as a shaded region of matching color. CO-pASW panel: analogous for pASW ice.
• Figure 5: Example of Very High (VH)-BE structures belonging to Elst-class (blue, BE = 1370 K) and Disp-class (orange, BE = 1492 K), and their location on the pASW surface, represented as altitude map. The inset figures display a portion of the binding sites comprising 50 water molecules, represented as sticks. The water molecules within 4.5 $\text{\AA}$ of CO center of mass (i.e. nearest neighbors, N$_{neighb}$) have been highlighted.