Dust-like surfaces sharply reduce UV-driven ice loss in laboratory tests

Why the surface beneath the ice matters

In cold interstellar environments, water accumulates in icy coatings on dust grains. Temperatures are too low for appreciable thermal sublimation, but ultraviolet photons can release water molecules or their fragments into the gas. This process, called photodesorption, helps regulate the balance between frozen and gaseous water.

Laboratory measurements used to describe that balance commonly involve ice on flat substrates. Actual cosmic dust has a more complicated structure. Ph. Parent and C. Laffon, at Aix Marseille University and CNRS’s CINaM laboratory in Marseille, investigated whether replacing flat surfaces with particulate materials changes how readily thin ice loses material under UV irradiation.

Their study, submitted to arXiv on November 6, 2026, is a preprint. The supplied material does not establish that it has completed peer review.

Comparing flat surfaces with dust analogs

The researchers deposited water ice on four substrates: flat gold, a polished olivine crystal, ground olivine particles and soot made of aggregated carbon nanoparticles. Olivine served as a silicate dust analog, while soot represented carbonaceous material. The soot’s primary particles averaged about 25 nanometres across; the ground olivine particles averaged about 0.30 micrometres.

Most measurements involved nominal ice coverages of two, four or six monolayers—equivalent amounts corresponding to a few molecular layers, not necessarily uniform coatings on rough grains. Experiments were conducted in ultrahigh vacuum at 12 and 60 kelvin, using a hydrogen-discharge lamp to supply UV radiation.

Two techniques tracked changes in the ice. X-ray photoelectron spectroscopy, or XPS, measured signals from the ice and substrate. Separate deposition calibrations for each material translated those signals into equivalent ice coverage, avoiding the need to determine the poorly known microscopic surface area of the aggregates independently.

Reflection–absorption infrared spectroscopy, or RAIRS, monitored water’s infrared absorption. It provided a complementary measurement on gold and soot, but substrate absorption prevented quantitative use on olivine.

Strong suppression on soot

On gold, the XPS-derived yield increased from approximately 0.00054 molecule per incident photon at two monolayers to 0.0019 at six monolayers at 12 kelvin. The authors interpret this roughly linear increase as consistent with thicker—but still very thin—ice absorbing more UV photons.

Soot behaved very differently. At two monolayers and 12 kelvin, its yield was approximately 0.000014 molecule per photon, about 39 times lower than gold’s. At four monolayers, the reduction relative to gold was about twentyfold. Both spectroscopic techniques showed strongly reduced ice depletion on soot, although they did not produce identical numerical yields.

The olivine comparison helped separate surface structure from composition. At four monolayers and 12 kelvin, flat olivine and gold gave similar yields, while particulate olivine’s yield was about 2.5 times lower than flat olivine’s. Soot’s yield was another tenfold lower than that of the olivine particles. This pattern supports an important role for morphology in the tested samples.

What the measurements can—and cannot—show

The two methods measure different aspects of UV processing. The authors interpret declining XPS coverage as net material loss from the ice. It does not separately identify escaping intact water molecules and oxygen-bearing fragments.

Infrared absorption can also change when water is converted into products that remain frozen, or when UV exposure rearranges the ice’s hydrogen-bond network. Its decline therefore cannot be equated directly with material escaping into the gas. Moreover, the soot infrared measurements used an initial coverage calibrated through XPS, so the methods were not fully independent in their absolute calibration.

The reported relative uncertainties are approximately 25% for XPS-derived yields and 30% for infrared-derived depletion yields. These uncertainties are much smaller than the largest soot–gold differences.

An unresolved mechanism and a modeling caution

To test whether pores within the ice were chiefly responsible, the researchers deposited a film on soot at 120 kelvin, where compact, essentially nonporous ice is expected, then cooled it to 12 kelvin. Material loss remained strongly suppressed. An additional infrared-only test at ten monolayers also showed low depletion.

The authors propose that soot may encourage patchy ice growth, changing molecular binding and how absorbed energy moves through small ice domains and across ice–grain interfaces. That explanation remains a hypothesis rather than a demonstrated microscopic mechanism.

The results suggest that models should account for both coverage and grain structure when describing photodesorption from thin ice. They do not establish a universal correction for cosmic dust. The authors also stress that rough grains can reduce water sticking: weaker ice loss and weaker ice accumulation have opposing effects, so both processes must be modeled together.