Desert dust from China’s Tengger Desert freezes clouds more efficiently when mixed with ammonium sulfate

Surface dust collected from the Tengger Desert in China can trigger the freezing of supercooled water droplets at temperatures as high as 264.3 K, and laboratory experiments now show that this ice nucleating ability depends strongly on the dust’s mineral makeup and on how the particles are chemically aged in the atmosphere. In a study published in Atmospheric Chemistry and Physics, researchers led by Jiao Xue and Bingbing Wang of Xiamen University measured immersion freezing, the dominant ice forming pathway in mixed-phase clouds, for dust suspensions prepared from desert surface soil and then tested how mixing with common atmospheric solutes changed that behavior.

The team collected surface dust at a site in the Tengger Desert (37.57° N, 104.93° E), sieved it to below 74 micrometers, and prepared aqueous suspensions at a range of dust mass concentrations. Two versions of the dust were studied: a simply shaken suspension called S-dust, and an ultrasonicated version called Us-dust, in which 30 minutes of treatment at 40 kHz broke up particle aggregates and dispersed finer grains. Freezing experiments were carried out with roughly 20 to 25 microliter-scale droplets on a temperature-controlled cooling stage under an optical microscope, cooled at 10 K per minute, with freezing events recorded frame by frame. Temperature calibration against the melting points of decane and dodecane gave an uncertainty of about ± 0.5 K, and each droplet type was measured in at least three independent samples with more than three repeated runs, yielding roughly 200 to 270 freezing events.

The central result is a clear concentration dependence of freezing. The median freezing temperature, T50, of the ultrasonicated dust rose from 253.6 K to 258.5 K as the dust mass fraction increased from 1.4 × 10⁻⁴ wt % to 1.0 × 10⁻¹ wt %. For the shaken S-dust, T50 climbed from 252.4 K to 254.6 K over its narrower concentration range. At comparable dust surface area concentrations, Us-dust froze about 2 K warmer than S-dust, and T50 increased by roughly 1.1 to 1.2 K per tenfold increase in dust surface area for both dust types. From these measurements the authors computed the ice nucleation active site density, Ns, the number of active sites per unit particle surface area, and found that Ns for Us-dust ran about one order of magnitude higher than for S-dust.

Single particle analysis by computer-controlled scanning electron microscopy with energy dispersive X-ray spectroscopy, covering about 1040 S-dust and 1180 Us-dust particles, helped explain the difference. Particles were classified by a rule-based scheme into illite, kaolinite, chlorite, quartz, and feldspar. Ultrasonication shifted the modal particle diameter from about 0.35 to 0.7 micrometers and reshaped the mineral mixture: compared with S-dust, Us-dust carried higher surface area contributions from chlorite (up 13 %), illite (up 4 %), and feldspar (up 2 %), and lower contributions from quartz (down 3 %) and kaolinite (down 15 %). Feldspar is widely recognized as the most ice-active common dust mineral, particularly K-rich feldspar. The authors argue that the modest 2 % increase in feldspar surface area may account for the roughly 2 K warmer freezing and the order-of-magnitude jump in Ns, a striking illustration of how small changes in mineral composition of complex natural dust can strongly influence its ice nucleation, especially when highly active components are involved.

The most consequential aging effect came from ammonium sulfate, a abundant atmospheric aerosol component. When Us-dust at 1.87 × 10⁻² wt % was mixed with ammonium sulfate for 12 hours before the freezing experiments, the ice nucleating activity rose systematically with solute concentration. The enhancement in T50 relative to untreated dust, ΔT50u, grew from 0.4 K at an ammonium sulfate concentration of 5 × 10⁻⁵ M to 4.7 K at 5 × 10⁻² M. At concentrations above 5 × 10⁻³ M, Ns was approximately two orders of magnitude larger than for untreated dust, and the heterogeneous ice nucleation rate coefficient, Jhet, increased progressively as well.

Notably, the enhancement followed a sigmoidal, or Boltzmann-type, response to ammonium sulfate concentration rather than a simple linear trend. At the lowest concentrations the effect was modest, suggesting that only a small fraction of susceptible ice nucleation sites was activated. As the concentration rose to 5 × 10⁻³ M, the enhancement increased markedly, consistent with substantial activation of additional sites, and at 5 × 10⁻² M it plateaued, implying that most susceptible sites had already been engaged. The authors interpret this threshold-saturation pattern as consistent with adsorption of ammonium ions on negatively charged mineral surfaces, which at the droplet pH of roughly 5.5 to 6.6 applies to feldspar, quartz, and kaolinite alike. Adsorbed ammonium may rearrange interfacial water molecules in ways that facilitate ice nucleation, and ion exchange of ammonium for potassium on feldspar surfaces may also contribute, although the underlying molecular mechanism remains to be confirmed.

When the ammonium sulfate effect was normalized by dust loading and expressed as the mass ratio of ammonium sulfate to dust, the desert dust’s enhancement converged with the range reported for K-feldspar in earlier laboratory work, even though K-feldspar-like particles made up only about 13 % of the feldspar fraction, a small slice of the total Us-dust surface area. This suggests feldspar contributes disproportionately to the most active sites, but the enhancement at high ammonium sulfate to dust ratios exceeded what K-feldspar alone could explain, pointing to contributions from other minerals such as kaolinite.

The other aging treatments told a subtler story. Mixing with sodium chloride, treatment with sulfuric acid or nitric acid, and heating the suspension at 95 °C for 30 minutes all left T50 essentially unchanged, indicating no impact on the overall freezing behavior by that metric. However, above 258 K the active site density did shift. Heat treatment, sulfuric acid at all tested concentrations, and the highest sodium chloride and nitric acid concentrations (5 × 10⁻² M) reduced Ns at these warmer temperatures, while lower concentrations of sodium chloride (5 × 10⁻⁴ M) and nitric acid (5 × 10⁻⁴ and 5 × 10⁻³ M) produced modest increases. The authors interpret this as evidence for thermally and chemically labile ice active sites that operate preferentially at warmer temperatures, possibly linked to feldspar, which is both highly active at warm temperatures and known to be sensitive to salt and acid suppression, though the specific origins of these labile sites were not determined.

To connect the laboratory findings to atmospheric modeling, the team developed experimentally constrained parameterizations of Ns and Jhet for Tengger Desert dust, both with and without ammonium sulfate aging. The aged versions express Ns and Jhet as functions of temperature or water activity deviation, with the intercept fitted as a Boltzmann sigmoid of the ammonium sulfate to dust mass ratio, valid for sulfate concentrations below 0.1 M and mass ratios from 0.03 to 30. In an illustrative atmospheric scenario, a 0.5-micrometer ammonium sulfate particle growing into a 10-micrometer droplet (about 0.002 M sulfate) that then collects a 0.5-micrometer dust particle yields a predicted Ns of 2.17 × 10⁶ cm⁻² and Jhet of about 24 cm⁻² s⁻¹ at 260 K, roughly 30 to 40 times higher than for untreated dust under the same conditions.

The study carries clear caveats. The measurements used a single dust source, one aging protocol, geometric rather than BET surface areas for computing Ns and Jhet, and a fixed cooling rate of 10 K per minute, so comparisons with studies at other cooling rates warrant caution. The parameterizations are constrained by inorganic aging only; organic compounds and other atmospheric aging processes remain to be tested. Still, the work provides process-level constraints for representing how sulfate aging modifies ice nucleation by natural Asian dust in cloud and climate models, a notable gap given that Asian dust contributes only 3.4 % of global annual dust mass yet, according to prior modeling, accounts for more than 15 % of global dust ice nucleating particles.

Subject of Research: Immersion freezing of natural desert dust and how mineral composition and atmospheric aging treatments alter its ice nucleating activity

Article Title: Immersion freezing of surface dust particles from Tengger Desert in China: effects of mineral composition and aging treatments on ice nucleating activity

Article References: Xue, J., Wang, S., Qi, H., Yao, Y., Chen, S., Zhou, Y., Lu, J., Zhang, T., Wang, Q., Zhao, D., Hong, Y., Cai, Y., & Wang, B. (2026). Immersion freezing of surface dust particles from Tengger Desert in China: effects of mineral composition and aging treatments on ice nucleating activity. Atmospheric Chemistry and Physics, 26(20), 15081-15099. https://doi.org/10.5194/acp-26-15081-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-15081-2026

Keywords: immersion freezing, ice nucleating particles, mineral dust, Tengger Desert, ammonium sulfate, feldspar, atmospheric aging, mixed-phase clouds, ice nucleation active site density, heterogeneous nucleation, cloud modeling, atmospheric chemistry