Insufficient constraining data
The paper explicitly states that current data is insufficient to precisely estimate binding energies using Bayesian inference, which is the core problem addressed. This limits the reliability of current binding energy estimates.
Assumptions about observational uncertainty
The study assumes that all future detected species will have the same level of observational uncertainty, which may not be realistic and could skew the prioritization of species.
Limitations of the chemical network
While the gas-phase network is considered robust, the grain-surface network is less comprehensive, and the chemistry of some species (e.g., sulfur) is viewed sceptically, potentially affecting the accuracy of abundance predictions.
Simplification of diffusion energy
The diffusion energy is assumed to be 0.5 of the binding energy, despite known variations (0.3-0.8), though the paper notes this may not be significant at 10 K. This is an assumption that could impact the model's accuracy.
Simplified characteristic vibration frequency equation
The authors acknowledge that a more accurate equation considering rotational partition function should be used, potentially impacting the precision of rate parameter calculations.
Assumption of well-known activation energies
The study assumes reaction activation energies are well known and independent of ice composition, which might not always hold true.
Focus on most diffusive species
To reduce dimensionality, the analysis concentrates only on the most diffusive species, which might limit the generalizability of the binding energy constraints to all species in the network.
Prioritization bias based on abundance
The recommendations prioritize species with both high 'filter sums' (importance) and 'large abundances,' which might lead to overlooking important but less abundant species that could still offer crucial constraints.