Frequent new particle formation over the high Arctic pack ice by enhanced iodine emissions

Arctic amplification has made the region warm at least twice as fast as the global average, with clouds exerting large radiative effects over the sea-ice–covered central Arctic Ocean. Cloud formation there can be limited by the scarcity of cloud condensation nuclei (CCN), so small changes in CCN can strongly affect surface energy budgets and sea-ice extent. While nucleation-mode and Aitken-mode particles have been observed over the central pack ice for decades, the vapors responsible for new particle formation (NPF) remained unidentified and climate models lack key local processes, including iodine nucleation. This study aims to identify the molecular drivers of NPF over the high Arctic pack ice during late summer to early autumn, quantify their variability and sources, and assess implications for particle growth and potential CCN formation.

Previous studies identified several CCN sources in the central Arctic, including secondary marine organics, primary marine emissions (sea salt and organic aerosol), and long-range transported continental aerosol and its downward mixing. Nucleation-mode particles over the pack ice have been reported since 1991 and are associated with air masses that have resided over ice, suggesting local sources; however, the causative vapors remained unknown. Around the marginal ice zone (MIZ), NPF has been attributed to sulfuric acid, ammonia, marine organics, and sporadically iodine. Models have suggested substantial NPF contributions to Arctic CCN but generally omit iodine nucleation and lack constraints for central Arctic precursors, which are lower than in the MIZ. Hypotheses of primary production of ultrafine particles have also been proposed. Together, these gaps motivated direct molecular-level measurements in the central pack ice to resolve the NPF mechanism and its seasonal behavior.

Field campaign: Measurements were conducted during the MOCCHA campaign within the Arctic Ocean 2018 expedition aboard the Swedish icebreaker Oden in August–September 2018, including over four weeks of ice-drift operations north of 88°N. Instrumentation and inlets: Three aerosol inlets were used: (i) an NPF inlet minimizing diffusional losses for nascent particles, (ii) a whole-air inlet sampling interstitial and activated aerosol, and (iii) a counterflow virtual impactor (CVI) inlet sampling only cloud droplets/ice crystals larger than ~7.8 µm aerodynamic diameter (enrichment factor ~6.5). Gas-phase measurements: Iodic acid (HIO3), sulfuric acid (H2SO4), and methanesulfonic acid (MSA) were measured with a nitrate chemical ionization mass spectrometer (CIMS). Calibration was performed post-campaign for H2SO4 and applied to HIO3 and MSA under kinetic ionization assumptions. Diffusion-loss corrections used the H2SO4 diffusion coefficient. Data were integrated over 10 min; LOD ~5×10^3 molecules cm^-3; for statistics, sub-LOD values were set to LOD/2. Uncertainty for field concentrations: −50%/+100%. Cluster ions: Naturally charged negative ions and clusters were measured with an atmospheric pressure interface time-of-flight mass spectrometer (API-TOF) without chemical ionization; data reported as counts per second (cps) due to unknown absolute transmission. Aerosol size distributions and number: Neutral/charged ions and particles <40 nm were measured with a Neutral cluster and Air Ion Spectrometer (NAIS). Submicron aerosols were characterized by a scanning mobility particle sizer (SMPS; 18–660 nm) and a differential mobility particle sizer (DMPS; 10–959 nm). Total particle number above ~2.5 nm was measured by a CPC (TSI 3776) and a Particle Size Magnifier (PSM). The UFP (2.5–15 nm) time series combined CPC3776 and PSM. Size distributions for broad ranges combined NAIS and DMPS. Loss corrections for diffusion/impaction were applied. Cloud residuals and CCN: Cloud residuals were sampled behind the CVI and measured with a CPC (TSI 3772) and a DMPS (17–959 nm). CCN spectra were measured with a DMT CCN counter scanning supersaturations of 0.1–1.0%. Residual size distributions were not corrected for CVI droplet efficiency due to lack of FSSP data for the highlighted event. Trace gases and meteorology: Ozone was measured with a 2B Tech Model 205 (±5%). Meteorological data were obtained from a ship-based weather station (~25 m ASL). Data were screened for ship pollution using an algorithm employing UFP derivatives, PSD features, CO2, black carbon, and wind direction. Boundary-layer diagnostics: Surface mixed layer height was inferred from radiosondes launched every 6 h using an established inversion-detection algorithm, with linear interpolation between launches and manual adjustment when needed. Dry deposition velocity: Computed as the inverse of the sum of aerodynamic and quasi-laminar resistances (canopy resistance assumed zero over snow). A neutral boundary layer and H2SO4 diffusion coefficient were assumed; friction velocity derived from ship wind speed. Median (IQR) va = 0.67 (0.39–0.96) cm s^-1. Back trajectories: 10-day LAGRANTO trajectories using ECMWF analyses (0.5° grid, 137 levels) were used to assess air mass origins, focusing on boundary-layer residence. Iodic acid steady-state model: Under clear conditions, HIO3 concentration was interpreted via a simple steady-state balance linking a constant iodine emission rate E (atoms cm^-2 s^-1), dry deposition (va), boundary-layer height (h), and particle condensation sink (CS): log10[HIO3] = log10 E − log10(va·h·CS). Eleven periods met steady-state criteria; data were analyzed against model predictions to infer E. Growth-rate analysis: Growth rates (GR) were derived by fitting multi-modal lognormal distributions to 10-min NAIS PSDs and tracking mode diameter evolution for periods of continuous growth of at least 2 h, excluding fog/pollution/air-mass change intervals. Contributions of HIO3 and H2SO4 to GR were estimated using a hard-sphere kinetic condensation model accounting for hydration (assumed three water molecules per acid), converting mass to mobility diameter (+0.3 nm), and applying a reported enhancement factor (EF) for sub-10 nm condensation (from sulfuric acid) as an upper-bound proxy for HIO3. Critical diameter and CCN activation: Critical diameters were estimated from CCN measurements assuming internal mixing by integrating the size distribution downward to match CCN concentrations at 0.2–1.0% supersaturation. CVI-derived cloud residual number and size distributions were compared with dry PSDs to infer activation of small Aitken-mode particles during fog. Event characterization and spatial extent: Eleven NPF events over pack ice were identified; two pollution-influenced events and one MIZ event were excluded from analysis. The spatial scale of events was estimated from duration and wind speed; e.g., a 5 September event implied a minimum air-mass diameter of ~160 km. Uncertainties: CIMS absolute uncertainties (−50%/+100%), API-TOF cps non-quantitative, PSM lower cutoff (~2.5 nm), CVI droplet cut (>7.8 µm) and lack of concurrent FSSP for some periods, and unknown HIO3 EF for growth were acknowledged.

- Molecular driver: All 11 major NPF events observed over the high Arctic pack ice during Aug–Sep 2018 were driven by iodic acid (HIO3), with sulfuric acid playing a minimal role. Negative-ion mass spectra showed iodine-oxide clusters containing up to eight iodine atoms; mixed S–I clusters were present but minor. - Nucleation polarity: Growth occurred only on negative ions; no positive-ion growth was observed, consistent with HIO3’s low proton affinity and a purely negative ion-induced pathway. - Concentrations: During a representative event (17 Sep), HIO3 exceeded 8×10^6 molecules cm^-3 while H2SO4 was 6–10× lower. The sulfuric acid level was >2 orders of magnitude too low for binary H2SO4–H2O nucleation; lack of amine/ammonia clusters ruled out ternary base-stabilized H2SO4 nucleation. - Seasonal shift and linkage to freeze-up/ozone: HIO3 increased steadily from late August into September, coinciding with near-surface temperature dropping below 0 to −2°C and the onset of freeze-up (identified ~28 Aug) and with a concurrent rise in boundary-layer ozone. UFP (2.5–15 nm) concentrations increased by more than an order of magnitude from summer (August) to autumn (September). Median HIO3 was >5× higher in autumn, whereas H2SO4 and MSA increased <2×. - Meteorological control and fog sink: Over the central pack ice, extremely low background aerosol led to very low condensation sinks (10^-5–10^-4 s^-1). Fogs/clouds dominated HIO3 removal; even tenuous fogs (visibility <4–5 km) reduced HIO3 enough to suppress NPF. Under clear conditions, HIO3 variability was explained by dry deposition velocity (va), boundary-layer height (h), and CS. - Emission rates: A simple steady-state model reproduced HIO3 variability across 11 clear periods, yielding iodine emission rates E with median (IQR) 5.0 [3.2–7.6]×10^6 atoms cm^-2 s^-1 and an overall range of ~1.5–15.4×10^6 atoms cm^-2 s^-1 (±2σ). These rates are likely lower limits since not all emitted iodine converts to HIO3. - Ultrafine particle growth: UFP growth rates were low, 0.2–1.2 nm h^-1 (median 0.5 [0.4–0.6] nm h^-1), yet persistent under intermittent fog, allowing growth to ~15–20 nm over hours. Kinetic condensation estimates indicate HIO3 alone accounted for >50% of observed growth in most events; applying a sulfuric-acid-based enhancement factor suggests HIO3 could explain most growth within uncertainties. Two events required additional growth from likely organics. - Survival and spatial extent: Despite fog-induced interruptions, UFP lifetimes (e.g., ~2.5 h for 5 nm; ~10 h for 10 nm under Arctic fog droplet concentrations <30–50 cm^-3) allowed recurring growth as fog patches advected. NPF was spatially extensive, with a minimum event scale of ~160 km diameter. - CCN relevance: Background aerosol was extremely low (median [IQR] N>30 nm: 26 [10–48] cm^-3; CCN at 0.3% SS: 17 [6–33] cm^-3). Direct CVI measurements during fog showed activation of Aitken-mode particles, with residual counts matching integrated dry aerosol above ~37–44 nm, implying activation of particles down to <30–40 nm when accumulation-mode particles were scarce. CCN/size data suggest ~1% SS was sufficient for activation during the event, indicating iodine-driven NPF can supply CCN under Arctic conditions.

These results resolve a three-decade uncertainty regarding the source of frequent nucleation over the central Arctic pack ice by identifying iodic acid as the dominant nucleating vapor in late summer to early autumn. The clear autumnal rise in HIO3, coinciding with the onset of freeze-up and increased ozone, indicates that sea-ice formation processes and oxidant availability likely enhance iodine emissions across the pack ice, producing widespread NPF. The negative-ion signature and iodine-rich cluster chemistry substantiate an iodine-oxide nucleation pathway distinct from classical sulfuric acid mechanisms, which are insufficient in this environment. The interplay of very low condensation sinks, fog/cloud sinks, boundary-layer dynamics, and surface dry deposition controls HIO3 and NPF occurrence. A simple physically based steady-state model captures much of the observed HIO3 variability and provides emission-rate constraints that Earth system models can adopt to represent iodine-driven nucleation in the high Arctic. Although growth rates are modest, the exceedingly clean background and long UFP lifetimes in CCN-limited foggy conditions allow survival and growth of nucleation-mode particles into the Aitken mode. Direct observations of activation of <30–40 nm particles in fog show that, when larger particles are scarce, iodine-driven NPF can contribute to the CCN population influencing cloud properties and thus the surface energy budget. The observed seasonality and similarity to multidecadal iodine records at Alert suggest these processes are not unique to 2018 and may be a recurring high Arctic feature. However, Aitken-mode populations can also originate from transported particles formed nearer the MIZ, underscoring that iodine NPF is an important, but not sole, contributor to the CCN budget over pack ice.

The study provides direct molecular evidence that iodic acid drives frequent new particle formation over the high Arctic pack ice during late summer to early autumn, with sulfuric acid playing a minor role. It documents a pronounced increase in HIO3 and UFP from summer to autumn, likely linked to freeze-up and elevated ozone, and demonstrates that particles in the 20–40 nm range can activate as CCN under CCN-limited Arctic fog conditions. A parsimonious steady-state framework quantifies iodine emission rates and explains much of the observed HIO3 variability, offering a pathway for incorporating iodine nucleation into climate models. Future research should (i) disentangle the relative roles of sea-ice formation processes versus ozone in driving iodine emissions, (ii) directly quantify neutral versus ion-induced nucleation pathways and the enhancement factors for HIO3 condensation, (iii) measure other iodine oxides and low-volatility organics contributing to growth, (iv) better constrain supersaturation dynamics and activation of very small particles in Arctic clouds, and (v) assess the seasonal and interannual variability and relative contributions of local iodine NPF versus transported Aitken-mode particles to the Arctic CCN budget.

- Seasonal and spatial scope: Observations are limited to August–September 2018 over the central pack ice; results may not represent other seasons or regions (e.g., springtime bromine/iodine chemistry or winter conditions). - Nucleation pathway quantification: While ion-induced nucleation was clearly negative, the relative importance of ion-induced versus neutral pathways could not be quantified. - Chemical speciation constraints: API-TOF measurements are in cps (uncalibrated for absolute concentrations). Nitrate-CIMS absolute uncertainties are large (−50%/+100%); calibration for HIO3 and MSA was inferred from H2SO4. Other iodine oxides likely contributing to growth were not quantified; the condensation enhancement factor specific to HIO3 is unknown (sulfuric-acid EF used as proxy). - Microphysical sampling: The CVI inlet sampled droplets >7.8 µm only; lack of concurrent droplet sizing for the highlighted event biases activation-diameter inference high. FSSP data were unavailable at times. PSM did not measure below ~2.5 nm. - Meteorological controls: The simple steady-state model assumes constant emission rates during events and steady boundary-layer conditions; it cannot separate the impacts of freeze-up versus ozone increases on emissions and neglects chemical intermediates. - Intermittent fog and advection: Frequent fog and air-mass changes limit continuous tracking of particle growth to CCN sizes and complicate attribution of survival probabilities.