A sensitivity analysis of interstellar ice chemistry in astrochemical models

TL;DR

The paper tackles the challenge of numerous poorly constrained parameters in interstellar ice chemistry by performing a global Monte Carlo sensitivity analysis on 630 chemical parameters within UCLCHEM across 225 physical environments defined by temperature, density, cosmic-ray ionization, and UV flux. It finds that diffusion barriers for small, mobile species such as H, N, O, HCO, and CH primarily drive the abundances of main ice species, while many reaction energy barriers have limited impact due to competition with diffusion. These findings guide where experimental and computational efforts should focus, emphasizing accurate diffusion barriers and careful treatment of binding-energy distributions over precise reaction barriers in many cases. The study also enhances UCLCHEM with updated desorption formalisms and tunneling considerations, providing a robust framework for interpreting ice observations and guiding future laboratory and theoretical work. Overall, the work clarifies which ice-chemistry parameters matter most and why, enabling more reliable astrochemical modeling across a range of prestellar environments.

Abstract

Astrochemical models are essential to bridge the gap between the timescales of reactions, experiments, and observations. Ice chemistry in these models experiences a large computational complexity as a result of the many parameters required for the modeling of chemistry occurring on these ices, such as binding energies and reaction energy barriers. Many of these parameters are poorly constrained, and accurately determining all would be too costly. We aim to find out which parameters describing ice chemistry have a large effect on the calculated abundances of ices for different prestellar objects. Using Monte Carlo sampled binding energies, diffusion barriers, desorption and diffusion prefactors, and reaction energy barriers, we determined the sensitivity of the abundances of the main ice species calculated with UCLCHEM, an astrochemical modeling code, on each of these parameters. We do this for a large grid of physical conditions across temperature, density, cosmic ray ionization rate and UV field strength. We find that, regardless of the physical conditions, the main sensitivities of abundances of the main ice species are the diffusion barriers of small and relatively mobile reactive species such as H, N, O, HCO, and CH$_3$. Thus, these parameters should be determined more accurately to increase the accuracy of models, paving the way to a better understanding of observations of ices. In many cases, accurate reaction energy barriers are not essential due to the treatment of competition between reactions and diffusion.

Figures (9)

• Figure 1: Abundances of all samples at $T=10$ K, $n_{\text{H}}=10^5$ cm$^{-3}$, $\zeta=1.3\times10^{-17}$ s$^{-1}$ and $F_{\mathrm{UV}} = 1$ Habing at $10^4$ years as function of the hydrogen diffusion barrier. Numbers at the bottom indicate $r_{\mathrm{RIN}}$ and its 95% confidence intervals. Pink crosses indicate nominal abundances and hydrogen diffusion barrier.
• Figure 2: Abundances (top) and correlation coefficients (bottom) of H2O, CO, CO2, CH3OH, and NH3 ices at $T=10$ K, $n_{\text{H}}=10^5$ cm$^{-3}$, $\zeta=1.3\times10^{-17}$ s$^{-1}$ and $F_{\mathrm{UV}} = 1$ Habing. In the top panels, the dashed pink line indicates the abundance using the nominal network, black lines indicate the individual samples, and light green is the log-average of all samples. In bottom subplots, the dashed gray line indicates perfectly uncorrelated parameters ($r_{\mathrm{RIN}}=0$), and the filled gray area indicates "weakly" correlated parameters ($\left|r_{\mathrm{RIN}}\right|<0.4$). The colored filled areas correspond to the 95% confidence intervals of the correlation coefficients.
• Figure 3: Abundances (top) and correlation coefficients (bottom) of H2O, CO, CO2, CH3OH, and NH3 ices at various physical conditions indicated by the text above each subfigure. In the top panels, the dashed pink line indicates the abundance using the nominal network, black lines indicate the individual samples, and light green is the log-average of all samples. In bottom subplots, the dashed gray line indicates perfectly uncorrelated parameters ($r_{\mathrm{RIN}}=0$), and the filled gray area indicates "weakly" correlated parameters ($\left|r_{\mathrm{RIN}}\right|<0.4$). The color indicates which species or reaction the parameter belongs to, and the linestyle indicates its type. The colored filled areas correspond to the 95% confidence intervals of the correlation coefficients. Legends are the same for all panels.
• Figure 4: Width of the 68.27% confidence intervals of abundances of selected ice species at various temperatures (averaged over all densities) over time at $\zeta=1.3\times10^{-17}$ s$^{-1}$ and $F_{\mathrm{UV}}=1$ Habing. The solid, dashed, dotted, dashed-dotted, and densely dotted lines correspond to 10, 20, 30, 40 and 50 K, respectively. The dashed gray line indicates a width of 2 orders of magnitude.
• Figure 5: Ratio between the reaction probability (Eq. \ref{['eq:reactionProbability']}) and the Boltzmann factor for diffusion at 10 K, for reactions with different energy barriers and diffusion barrier of the most mobile reagent. Solid lines indicate a reaction where tunneling is efficient ($\mu=1$ amu), and dashed lines indicate a reaction where tunneling is inefficient ($\mu=12$ amu).