Role of methanesulfonic acid in atmospheric particle nucleation and growth Authors: Rima Baalbaki, Jiali Shen, Mario Simon, Hannah Klebach, Samuel Ruhl, Jenna DeVivo, Mingyi Wang, Wiebke Scholz, Lubna Dada, Birte Rörup, Dominik Stolzenburg, Hanna E Manninen, Eva Sommer, Lucía Caudillo-Plath, Guillaume Marie, Martin Friedrich, Wenjuan Yu, Markus Leiminger, Dina Alfaouri, Antonio Amorim, Tatjana Arnoldi-Meadows, Hannah Beckmann, Moritz Berntheusel, Steffen Bräkling, Zoé Brasseur, Randall Chiu, Jonathan Duplissy, Henning Finkenzeller, Martin Heinritzi, Felix Kunkler, Houssni Lamkaddam, Brandon Lopez, Naser Mahfouz, Vladimir Makhmutov, Monica Martinez, Ruby Marten, Dario Massabo, Roy Mauldin, Bernhard Mentler, Markus Müller Published: 2026 ABSTRACT AbstractDimethyl sulfide (DMS; CH3SCH3) from marine phytoplankton is a notable source of atmospheric sulfur1. Its oxidation products include sulfuric acid (SA; H2SO4) and methanesulfonic acid (MSA; CH3SO3H), which has a higher yield than SA below 10 °C (ref. 2). Although SA is known to drive the formation of new particles3, which may subsequently grow and act as cloud condensation nuclei (CCN), the role of MSA remains unclear4. Here, in experiments performed under atmospheric conditions at the CERN CLOUD (Cosmics Leaving OUtdoor Droplets) chamber, we show that MSA nucleates together with ammonia (NH3) below −10 °C, at rates comparable with SA-NH3. Moreover, MSA and SA nucleate synergistically below −10 °C, forming multi-acid molecular clusters with NH3. Even at ultralow NH3 levels, MSA drives particle growth at or near the kinetic limit below 9 °C and above 40% relative humidity (RH). Because MSA and SA generally coexist at similar concentrations in cool marine regions, our findings indicate that nucleation rates may be accelerated up to tenfold and growth rates up to twofold compared with SA-NH3 alone. Our global model simulations indicate that MSA can enhance CCN concentrations, especially in polar regions. We propose that MSA might be an important driver of biogenic particles in cool, pristine marine regions of both the present-day and pre-industrial atmospheres and yet is unaccounted for in global climate models5.Subject terms: Atmospheric science, Climate change Main The importance of DMS from phytoplankton for marine aerosols and clouds has long been recognized. Pioneering work6 in 1978 established the first association between DMS, at pungent concentrations near the Great Barrier Reef, and elevated particle number concentrations. Subsequent observations in marine environments have consistently found an association between DMS and particle formation7–10. This connection underpinned the CLAW hypothesis, which proposed a biogenic feedback loop for climate regulation linking marine DMS emissions, particle formation and cloud brightening11. Although DMS emissions are estimated to contribute to radiative cooling12, the magnitude of their effect remains uncertain13 owing to incomplete understandings of emission sources14, chemical processes15,16 and the role of DMS oxidation products in particle nucleation and growth5,17. New particle formation occurs when low-volatility vapours reach sufficient supersaturation to condense and form stable molecular clusters around 1–2 nm in size. Through a combination of condensation and coagulation, these clusters may grow to sizes above about 50 nm, at which they can act as CCN. Two DMS oxidation products that influence new particle formation are SA and MSA, which are widely found in the marine boundary layer at concentrations of 104–107 cm−3. Although MSA originates mainly from DMS oxidation and, to a lesser extent, from methanethiol (CH3SH)18, SA has further sources, such as volcanic sulfur dioxide (SO2), hydrogen sulfide (H2S)19 and anthropogenic SO2. Nevertheless, in remote marine regions, MSA concentrations are comparable with SA concentrations and may even exceed them in polar regions20–25 and at high altitudes26,27. Whereas the importance of SA for atmospheric particle nucleation and growth is well established3,28, the role of MSA remains inconclusive. Numerous ambient studies have shown that MSA contributes to particle growth and mass21,29–34. However, the contribution of MSA to particle nucleation under atmospheric conditions remains uncertain. Theoretical studies suggest that, although MSA is a weaker acid than SA, it can facilitate nucleation by clustering with SA and base vapours35,36. Experimentally, MSA has been shown to form particles with amines and ammonia (NH3) at high (ppbv) precursor concentrations and warm temperatures, with water playing an important role in the process37–39. More recent flow tube experiments measured SA-MSA and MSA-MSA dimers at 25 °C and 20% RH with 2 × 108 cm−3 MSA, (2–8) × 108 cm−3 SA and base vapours (ammonia and amines)4. However, only relatively small (20–30%) increases (or even decreases) in particle counts were measured when injecting 2.5 × 108 cm−3 MSA into SA-base vapour mixtures4. In summary, ambient observations show that MSA contributes substantially to particle mass and laboratory studies indicate that high RH promotes MSA particle formation at warm temperatures in the presence of base vapours and acid concentrations exceeding typical atmospheric levels. However, MSA remains unrepresented in global climate models because of the lack of laboratory measurements of its contributions to particle nucleation and growth rates under atmospheric conditions. Here we report experiments performed at the CERN CLOUD chamber3 to study the role of MSA in new particle formation under atmospheric conditions between −52 °C and +9 °C. For most experiments, MSA was produced in the chamber from oxidation of DMS with hydroxyl (OH) radicals produced by photolysis of ozone. However, for several experiments at +9 °C, MSA was directly injected into the chamber from a temperature-controlled evaporator. SA was produced from oxidation of both DMS and SO2, separately injected into the chamber, to allow independent control of the SA and MSA concentrations. Experiments were carried out with NH3 between contaminant level (<1 pptv to 4 pptv, depending on temperature) and 760 pptv. Some experiments also involved photolysis of molecular iodine to produce iodine oxoacids (IAx), which we define here as the sum of iodic acid (IA; HIO3) and iodous acid (IOA; HIO2)40. Further details of the CLOUD facility and its analysing instruments are provided in Methods. Particle nucleation rates In Fig. 1, we compare our measured particle nucleation rates at 1.7 nm (J1.7) for MSA-SA-NH3 (data points) at −30 °C, −10 °C and +9 °C with those expected for SA-NH3 (coloured bands), based on previous measurements at CLOUD28. Water is implicit in all of the chemical systems described here. The nucleation rates versus SA, MSA and summed acids (SA + MSA) are shown in Fig. 1a–c, respectively. Experiments in which SA and MSA were both generated from DMS are indicated by filled symbols. To extend the MSA:SA ratio range at +9 °C, some experiments were also carried out with direct injection of MSA vapour and with SA produced by oxidation of SO2 (indicated by hollow red squares). The latter experiments include contaminant NH3 (<4 pptv) and contaminant dimethylamine (DMA; <1 pptv), estimated from an atmospheric pressure interface time-of-flight (APi-TOF) mass spectrometer. Despite such low base vapour concentrations, they were sufficient to reach nucleation rates comparable with those for SA-NH3 with 46–111 pptv NH3 at +9 °C (as seen by the alignment of the filled and hollow red squares in Fig. 1a). All but two measurements (indicated by circle symbols in Fig. 1) were made under galactic cosmic ray (GCR) conditions, so these data are insufficient to distinguish ion-induced effects at low nucleation rates. MSA-SA-NH3 particle nucleation rates, J1.7, versus acid concentrations. a–c, Particle nucleation rates at 1.7 nm (J1.7) under neutral (circles) and GCR (squares) conditions versus SA (a), MSA (b) and summed acid (SA + MSA) (c) concentrations at −30 °C (blue), −10 °C (green) and +9 °C (red). The filled symbols indicate experiments in which SA and MSA were generated by DMS oxidation. The open red squares indicate experiments at +9 °C in which SA was produced by SO2 oxidation and MSA was injected directly from an evaporator. The coloured bands indicate the expected SA-NH3 nucleation rates under GCR conditions according to Dunne et al.28. The bands are identical across all panels (with the x-axis interpreted as SA) and are colour-coded by temperature. The band widths reflect the range of NH3 concentration for the filled symbols at each temperature: 46–111 pptv at +9 °C, 139–460 pptv at −10 °C and 4–121 pptv at −30 °C. Ammonia was not measured during the −30 °C experiments but instead estimated using a box model (Methods). The experimental conditions for the filled symbols were 5–280 ppbv O3, 0–0.5 ppbv SO2, 0.5–2.5 ppbv DMS and 40–80% RH. For the open red squares, the experimental conditions were 38–44 ppbv O3, 5.8–6.2 ppbv SO2, <4 pptv NH3, <1 pptv DMA and 62% RH. Here, and throughout this study, measurements labelled +9 °C correspond to a temperature range of 9–10 °C. Each data point represents the mean values of the steady-state measurement period of a single experiment. Vertical and horizontal error bars show ±1σ measurement uncertainty of J1.7 and acid concentrations, respectively. The systematic scale uncertainties for SA and MSA concentrations (50%, 200%) and J1.7 (±30%) are shown separately in panel a. For all temperatures, nucleation rates in the presence of MSA exceed those expected from SA-NH3 alone. The nucleation rates of the filled squares are around 100 times faster than those expected from SA-NH3 at −30 °C and ten times faster at −10 °C and +9 °C (Fig. 1a). This enhancement reflects the higher MSA:SA ratio at −30 °C (5.8–20) compared with −10 °C (2.6–4.9) and +9 °C (0.7–1.2), owing to increased MSA production at lower temperatures2. Figure 1a,b shows that the nucleation rates at −30 °C and −10 °C depend on MSA as well as SA and they align with those expected for SA-NH3 when MSA is considered equivalent to SA (Fig. 1c). However, the behaviour is different at +9 °C; here the three hollow points at fixed 2 × 107 cm−3 SA show no change of nucleation rate after increasing the MSA concentration tenfold from 2.6 × 107 cm−3 to 2 × 108 cm−3 (Fig. 1b). By contrast, a doubling of the SA concentration under the same base vapour conditions increases the nucleation rate by one order of magnitude (Fig. 1a). The discrepancy between the measured and predicted SA-NH3 nucleation rates at +9 °C (Fig. 1a) is within the combined systematic uncertainties of the measurements and parameterization. We conclude that MSA at these concentrations is contributing to nucleation on an equal basis as SA below −10 °C but is not contributing to nucleation above +9 °C. We also performed experiments to investigate the interaction of MSA with IAx-SA-NH3, as iodine oxoacids are important for new particle formation in marine and polar atmospheres40,41. In Extended Data Fig. 1 (left-hand panels), we show an experiment at −10 °C with fixed IAx (1.9–3.6 × 106 cm−3) and NH3 (0.9 × 109 cm−3, equivalent to 35 pptv), with SA increased in four steps40 from 5 × 105 to 4.1 × 106 cm−3. The nucleation rate, J1.7, increases from about 1 to 20 cm−3 s−1, as shown in our earlier study40. In the right-hand panels, this experiment is repeated with similar IAx (1.7 × 106 cm−3), SA (7.5 × 106 cm−3) and NH3 (1.8 × 109 cm−3, equivalent to 69 pptv) but now including MSA (about 3 × 107 cm−3). Under these conditions, the nucleation rate increases by more than one order of magnitude, reaching about 200 cm−3 s−1. The nucleation rates expected from the parameterization of the IAx-SA-NH3 and SA-NH3 systems are around one-tenth of the measured values (Extended Data Fig. 1e), whereas the nucleation rates expected for the experiment without MSA agree with the measured values (Extended Data Fig. 1b). These experiments demonstrate that MSA participates in multi-acid MSA-IAx-SA-NH3 nucleation at −10 °C and below. It implies that all three acid vapours (SA, IAx and MSA) need to be considered together when assessing marine new particle formation. New particle formation from IAx-SA-NH3(-MSA) at −10 °C. The upper panels show the evolution of vapour concentrations, the middle panels show the particle nucleation rates at 1.7 nm, J1.7, and the lower panels show the size distributions (dN/dlogDp) of negatively charged particles of mobility diameter between 0.8 and 42.0 nm. MSA was absent for the left-hand panels (a–c) and present for the right-hand panels (d–f). The vertical dashed lines in panels a and b represent new stages in which SA was increased. The curves in panels b and e show the measured J1.7 (solid curve) and those expected for SA-NH3 (dotted curve) and IAx-SA-NH3 (dashed curve). The experimental conditions for panels a–c are 41 ppbv O3, 1.6 ppbv SO2 and 72% RH. MSA was generated in the right-hand experiments through the oxidation of DMS. The experimental conditions for panels d–f are 161 pptv DMS, 4.7 ppbv O3, 1.6 ppbv SO2 and 67% RH. The strong enhancement of particle nucleation and growth rates after addition of around 3 × 107 cm−3 MSA demonstrates that MSA participates in multi-acid MSA-IAx-SA-NH3 particle formation. We also conducted experiments at −52 °C without ammonia (< 1 pptv contaminant level) to simulate upper tropospheric conditions. Extended Data Fig. 2 presents an experiment with 5 × 106 cm−3 SA, both before (Extended Data Fig. 2a–c) and after (Extended Data Fig. 2d–f) adding 5 × 106 cm−3 MSA. The measured nucleation rate, J1.7, did not increase, despite doubling the acid concentration (Extended Data Fig. 2b,e). However, the particle growth rate above 3.2 nm was doubled (Extended Data Fig. 2c,f), reaching the kinetically limited condensation rate. The lack of any enhancement of the nucleation rate by MSA in the near absence of NH3 contrasts with the strong enhancements measured at −10 °C and −30 °C in the presence of NH3 (Fig. 1), despite a reduction of MSA saturation vapour pressure by at least an order of magnitude between −30 °C and −52 °C. We also conducted a sensitivity test in which MSA was directly injected into the chamber at concentrations up to 1.4 × 107 cm−3 at −30 °C, in the absence of NH3 and SA. No nucleation was observed until NH3 was introduced to the chamber. These experiments show that MSA at concentrations below around 107 cm−3 does not contribute to particle nucleation in the atmosphere in the absence of a base vapour, even in the presence of SA and at temperatures as low as −52 °C. Effect of MSA on nucleation and growth at −52 °C, in the absence of NH3. The upper panels show the evolution of vapour concentrations, the middle panels show the particle nucleation rates at 1.7 nm, J1.7, and the lower panels show the size distributions (dN/dlogDp) of negatively charged particles of mobility diameter between 0.8 and 42.0 nm. MSA was absent for the left-hand panels (a–c) and present for the right-hand panels (d–f). The experiments were performed with NH3 below the limit of detection, with estimated contaminant concentrations <1 pptv. All nucleation rates, J1.7, are consistent with those expected for SA-NH3 with 1–4 pptv NH3 (grey bands in panels b and e). Addition of MSA has a negligible influence on nucleation rates, J1.7 (panels b versus e), but contributes equally to SA for particle growth rates above 3.2 nm (panels c versus f). These experiments show that MSA is unable to contribute to nucleation in the absence of base vapours such as NH3, even at temperatures as low as −52 °C. However, MSA can contribute to particle growth under these conditions, as the Kelvin (curvature) barrier falls away. The experimental conditions for panels a–c are 320 ppbv O3, 6.1 ppbv SO2 and 71% RH. SA was produced in this experiment from the oxidation of SO2. The experimental conditions for panels d–f are 1,584 ppbv O3, 0.07 ppbv SO2, 6.5 ppbv DMS and 70% RH. MSA and SA were generated in this experiment through the oxidation of DMS. Composition of molecular clusters Figure 2 presents the molecular composition of negatively charged clusters during MSA-SA-NH3 nucleation at +9 °C (Fig. 2a) and −10 °C (Fig. 2b), measured with an APi-TOF mass spectrometer. The vapour concentrations were comparable for both experiments: 1.0 × 107 cm−3 MSA, 1.6 × 107 cm−3 SA and 123 pptv NH3 (Fig. 2a) and 1.1 × 107 cm−3 MSA, 0.8 × 107 cm−3 SA and 71 pptv NH3 (Fig. 2b). Contaminant iodine oxoacid was also present from previous measurements at concentrations around 106 cm−3 but represented a minor nucleation channel. The clusters contain up to ten SA, eight MSA and nine NH3 molecules and reveal 1:1 acid–base pairing as the nucleation mechanism3,42. Molecular composition of negatively charged clusters during MSA-SA-NH3 nucleation. a,b, Mass defect (difference from integer mass) versus m/z (Th) during nucleation events at +9 °C (a) and −10 °C (b) for negatively charged molecular clusters measured with the APi-TOF mass spectrometer. The marker size shows ion signal intensity in counts per second (cps), as indicated in the legend. The vector arrows and dotted grid lines indicate the molecular increments and cluster locations, respectively. The symbol shape indicates the chemical composition: SA-NH3 (triangles), MSA-NH3 (inverted triangles) and MSA-SA-NH3 (diamonds). Clusters that contain contaminant iodine from earlier experiments are indicated by filled grey circles. The symbol colour indicates the MSA molar fraction, MSA/(MSA + SA). The experimental conditions for a are 1.0 × 107 cm−3 MSA, 1.6 × 107 cm−3 SA, 1.2 × 106 cm−3 IA, 771 pptv DMS, 123 pptv NH3, 253 ppbv O3, 0.7 ppbv SO2 and 42% RH. The experimental conditions for b are 1.1 × 107 cm−3 MSA, 0.8 × 107 cm−3 SA, 8.3 × 105 cm−3 IA, 30 pptv DMS, 71 pptv NH3, 7.6 ppbv O3, 0.5 ppbv SO2 and 65% RH. At +9 °C (Fig. 2a), MSA was detected as a monomer and a dimer but did not appear in larger clusters until at least five SA molecules had accreted with NH3, supporting our earlier conclusion that MSA does not contribute to nucleation under these conditions. To further explore the nucleation potential of MSA-NH3 at +9 °C in the absence of SA, we performed further experiments in which elevated MSA (up to 2 × 108 cm−3) was directly injected into the chamber at up to 190 pptv NH3 and 60% RH. No particle formation was observed, confirming that MSA-NH3 cannot drive nucleation at +9 °C. Nevertheless, in the experiment shown in Fig. 2a, MSA was detected in clusters containing five SA, showing that MSA can contribute to particle growth immediately after the nucleation barrier represented by the negatively charged SA tetramer (m/z = 400–450 Th in Fig. 2a) has been crossed. Further insights are provided in Extended Data Fig. 3, which compares the molecular composition of the experiment in Fig. 2a with another experiment at +9 °C with similar MSA concentration but three times lower NH3 concentration (123 pptv versus 43 pptv). The lower NH3 produces a marked reduction of MSA in (SA)5(NH3)n clusters, indicating that MSA is rapidly evaporating from the five SA clusters at +9 °C and is highly sensitive to NH3 concentration, even at base:acid ratios of more than 100. We infer from these molecular measurements that the temperature at which MSA begins to nucleate under atmospheric conditions is not far below +9 °C and it requires the presence of relatively high NH3 concentrations. Molecular composition of negatively charged clusters during MSA-(IAx)-SA-NH3 nucleation at +9 °C at varying NH3 concentration. Mass defect versus m/z during nucleation events at 123 pptv NH3 (a) and 43 pptv NH3 (b) for negatively charged molecular clusters measured with the APi-TOF. The marker size shows ion signal intensity in counts per second (cps), as indicated in the legend. The symbol shape indicates the chemical composition. The symbol colour indicates the MSA content, expressed as MSA/(MSA + SA). MSA was generated in both experiments through the oxidation of DMS. The experimental conditions for panel a are 1.04 × 107 cm−3 MSA, 1.6 × 107 cm−3 SA, 1.2 × 106 cm−3 IA, 771 pptv DMS, 123 pptv NH3, 253 ppbv O3, 0.7 ppbv SO2 and 42% RH. The experimental conditions for panel b are 1.4 × 107 cm−3 MSA, 2 × 107 cm−3 SA, 6.8 × 105 cm−3 IA, 827 pptv DMS, 43 pptv NH3, 261 ppbv O3, 0.72 ppbv SO2 and 42% RH. The cluster composition is very different at −10 °C (Fig. 2b). MSA is now seen in the initial nucleating clusters and a clear sequence of pure MSA-NH3 clusters is observed (inverted green triangles), extending up to (CH3SO3H)5·(NH3)4·CH3SO3−. This demonstrates that ion-induced nucleation of MSA-NH3 takes place at −10 °C, without requiring SA. The first MSA cluster with NH3 is the dimer, that is, CH3SO3H·NH3·CH3SO3−, whereas the first SA cluster with appreciable NH3 is the tetramer, that is, (H2SO4)3·NH3·HSO4−, which is probably because of CH3SO3− being a weaker Lewis base than HSO4−. Pure SA-NH3 clusters are also detected (red triangles). However, the dominant nucleation pathway involves multi-acid MSA-SA-NH3 clusters (diamonds), which follow a 1:1 acid:base ratio, in which acid is the sum of MSA and SA and they become increasingly abundant as the clusters grow. This illustrates the symbiotic and freely mixing nature of these two acids in forming molecular clusters at −10 °C and below. To further explore the effect of acid composition, we increased the MSA:SA ratio from 1.4:1 (Fig. 2b) to 16:1 at −10 °C and to 13:1 at −30 °C (Extended Data Fig. 4a,b, respectively). At these high MSA:SA ratios, the dominant nucleation mechanism becomes MSA(-NH3), with only a minor presence of SA in the clusters, once again demonstrating the interchangeability and synergy between MSA and SA in multi-acid nucleation. Molecular composition of negatively charged clusters during MSA-(IAx)-SA-NH3 nucleation at high MSA:SA ratio. Mass defect versus m/z during nucleation events at −10 °C (a) and −30 °C (b) for negatively charged molecular clusters measured with the APi-TOF and MION2-APi-TOF. The marker size shows ion signal intensity in counts per second (cps), as indicated in the legend. The symbol shape indicates the chemical composition. The symbol colour indicates the MSA content, expressed as MSA/(MSA + SA). MSA was generated in both experiments through the oxidation of DMS. The experimental conditions for panel a are 8.1 × 106 cm−3 MSA, 5 × 105 cm−3 SA, 1.8 × 106 cm−3 IA, 2.9 ppbv DMS, 88 pptv NH3, 122 ppbv O3, 0.5 ppbv SO2 and 68% RH. The experimental conditions for panel b are 5.8 × 106 cm−3 MSA, 4.5 × 105 cm−3 SA, 2.7 ppbv DMS, 47 ppbv O3, 0.06 ppbv SO2 and 63% RH. NH3 was estimated from MFC settings to be between 4 and 119 ppbv (see Methods for further details). The experiments shown in Extended Data Fig. 4a included contaminant iodine oxoacids (1.8 × 106 cm−3 IA and 9.6 × 104 cm−3 IOA), leading to the formation of IA-MSA-NH3 and IA-MSA-SA-NH3 clusters (squares). Here IA may either provide the core ion (IO3−) or else assist cluster formation as a neutral molecule. Notably, our nitrate chemical ionization mass spectrometer (CIMS) measurements revealed a prominent MSA-IOA dimer, indicating that IOA may substitute for ammonia, as seen previously for SA-IA nucleation40,41. This could be important for low-NH3 environments, such as the upper free troposphere and remote polar regions, away from localized sources such as seabird colonies. Quantifying the kinetics and stability of MSA-IOA clusters will be critical to understanding their atmospheric relevance. Particle growth rates In Fig. 3, we show four experiments to study particle growth at +9 °C and constant SA (2 × 107 cm−3) but with different MSA concentrations: <3 × 105 cm−3 (Fig. 3a), 2 × 107 cm−3 (Fig. 3d), 7 × 107 cm−3 (Fig. 3g) and 2 × 108 cm−3 (Fig. 3j). Here we injected MSA vapour directly from an evaporator, so its concentration could be adjusted without changing SA. For these experiments, the base vapours consisted of around <1 pptv contaminant DMA plus <4 pptv NH3. The particle growth rates between 3.2 and 8.0 nm (GR3.2–8.0) increased progressively from 3.5 nm h−1 with low MSA (Fig. 3c) to 5.3, 9.8 and 22.3 nm h−1 (Fig. 3f,i,l, respectively). These growth rates correspond to 0.9–0.96 nm h−1 per 107 cm−3 of MSA, which indicates kinetically limited growth by MSA condensation, as we previously measured for SA43 and IA41. Influence of MSA on particle growth rates at +9 °C. a,d,g,j, Evolution of the SA and MSA concentrations for four experiments at constant SA and increasing MSA. b,e,h,k, Particle nucleation rates at 1.7 nm, J1.7. c,f,i,l, Corresponding particle size distributions (dN/dlogDp) between 2 and 42 nm. For these experiments, SA was generated by oxidation of SO2, while MSA vapour was directly injected into the chamber from an evaporator so that it could be adjusted without changing SA. The base vapours for these experiments consisted of <1 pptv contaminant DMA and <4 pptv contaminant NH3. The experimental conditions for a–c are 38 ppbv O3, 4.9 ppbv SO2 and 63% RH. The experimental conditions for d–f are 38 ppbv O3, 6.4 ppbv SO2 and 62% RH. The experimental conditions for g–i are 39 ppbv O3, 6.7 ppbv SO2 and 62% RH. The experimental conditions for j–l are 44 ppbv O3, 6.3 ppbv SO2 and 62% RH. The UV intensity was held constant during all experiments. In Fig. 4, we show all of our measurements of particle growth rates with various mixtures of SA, IAx and MSA between −52 °C and +9 °C. The upper panels show growth rates from 1.8 to 3.2 nm and the lower panels from 3.2 to 8.0 nm. In Fig. 4a,b, the measured growth rates greatly exceed those expected solely from kinetic condensation of SA and IAx (indicated by the dashed lines). However, once MSA is included in the total acid concentration, the measurements either align with or approach the kinetic limit (Fig. 4c,d), indicating that MSA drives particle growth above 1.8 nm size at or near the kinetic limit, below +9 °C. Growth rates measured above the kinetic limit result from the accretion of acid–base molecular clusters at base:acid ratio >1, which are unaccounted by the acid monomer concentration (x axis)41. Two measurements (green squares) at −10 °C that fall slightly below the kinetic line in Fig. 4d correspond to experiments with base:acid ratios near or below unity (NH3 < 4 pptv) and with nucleation rates below 10 cm−3 s−1. Under these conditions, molecular clusters are not expected to contribute substantially to particle growth rate. The situation is similar at +9 °C because, at this temperature, MSA does not contribute appreciably to small molecular clusters. In summary, our measurements show that MSA drives particle growth at or near the kinetic limit across the full temperature range from −52 °C to +9 °C, even at extremely low NH3 concentrations with a base:acid ratio below unity. Growth rates versus acid concentrations. a–d, Particle growth rates versus SA (+IA) (a,b) and SA (+IA) + MSA (c,d) concentrations. The ‘(+IA)’ is enclosed in parentheses to denote that it was only present in certain experiments (indicated by symbols with a thick outline). The upper panels show the growth rates between 1.8 and 3.2 nm, whereas the bottom panels show the growth rates between 3.2 and 8.0 nm. The CLOUD measurements are represented by inverted triangles and squares (corresponding to negatively charged and total particles, respectively). For comparison, ambient observations from Ny-Ålesund are represented by circles (total particles). The symbol colour indicates temperature. Open circles indicate that temperature measurements were not available from Ny-Ålesund. The dashed lines show the theoretical growth rates for SA condensation at the kinetic limit, after accounting for van der Waals interactions, which we have verified experimentally43. The alignment of all measurements near the dashed lines when all acids are summed (c,d) indicates that MSA drives particle growth above 1.8 nm at or near the kinetic limit for all temperatures below +9 °C. Each data point represents the growth rate measured in a single experiment. The acid concentrations plotted on the x axis are the mean values measured during the corresponding growth period. Vertical error bars represent ±1σ statistical uncertainties in the growth rate measurement, whereas the horizontal error bars represent ±1σ uncertainty in vapour concentrations. The systematic scale uncertainties for acid concentrations (50%, 200%) and growth rates (±50%) are shown separately in panel d. The experimental condition ranges are 40–79% RH, 0–1.8 × 107 cm−3 IA, <1 to 492 pptv NH3, 4–255 ppbv O3 and 0–7 ppbv SO2. Ammonia was not measured during the −30 °C experiments but instead estimated using a box model (Methods). We have also compared our growth rate measurements with ambient observations from Ny-Ålesund, Svalbard (circle symbols in Fig. 4b,d). To enable a direct comparison with our results, we reanalysed the ambient data21, to extract the growth rates for 3.2–8.0 nm. The measured growth rates during spring in Ny-Ålesund exceed those expected from SA alone and closely match our experimental results, within measurement uncertainties, supporting the important role of MSA for particle growth in polar environments. Finally, we have directly confirmed the presence of MSA in the particle phase by analysing thermally evaporated particles with a Filter Inlet for Gases and AEROsols (FIGAERO) coupled to a CIMS. Extended Data Fig. 5 shows our measurements at +9 °C (Extended Data Fig. 5a–c) and −10 °C (Extended Data Fig. 5d–f). MSA is detected in particles smaller than 10 nm and at particle mass concentrations of only 1 ng m−3 (Extended Data Fig. 5c,f). The production of SA is turned off during the experiment at +9 °C but the particles continue to grow and particulate MSA continues to increase (Extended Data Fig. 5a–c), demonstrating that MSA can drive particle growth independently of SA. The only particle-phase compounds detected by the FIGAERO in these experiments were SA and MSA. This provides direct evidence that particle growth is driven by SA and MSA alone rather than by some other, unidentified compound. Measurement of particle composition below 10 nm. Evolution of vapour concentrations (upper panels), particle size distributions (dN/dlogDp) for mobility diameters between 4 and 30 nm (middle panels, measured by the nanoSMPS) and particulate MSA signals and mass concentrations (lower panels, measured by the FIGAERO-CIMS). The experiments were performed at +9 °C (a–c) and −10 °C (d–f). MSA was either directly injected into the chamber (a–c) or else generated by oxidation of DMS (d–f). The bin widths in panels c and f represent the periods when particles were collected on the filters of the FIGAERO-CIMS. The dashed vertical line in the left-hand panels indicates when the UV illumination was switched off and SA production ended; nevertheless, MSA continued to drive particle growth. At −10 °C, MSA was detected in the particle phase using a Br-FIGAERO-CIMS when the particles were all below 10 nm (panel f). Although SA was measured in the particle phase, it is not included in panels c and f because direct comparison with MSA requires compound-specific calibrations, which were not available. The experimental conditions for panels a–c are 38 ppbv O3, 6.4 ppbv SO2, <4 pptv NH3, <1 pptv DMA and 62% RH. The experimental conditions for panels d–f are 247 ppbv O3, 0.6 ppbv SO2, 491 pptv DMS, 472 pptv NH3 and 73% RH. Atmospheric implications In summary, we find that MSA nucleates together with SA-NH3 at −10 °C and below, at atmospheric concentrations. Moreover, MSA and SA show comparable strengths, demonstrating their interchangeability and synergy when quantitatively assessing acid–base nucleation rates. However, at slightly warmer temperatures (+9 °C), MSA does not nucleate with NH3, although our molecular measurements suggest that the onset of nucleation is not far below +9 °C. Furthermore, MSA does not nucleate in the absence of a base vapour. We also find that MSA drives rapid growth of particles above 1.8 nm at or near the kinetic limit across the full temperature range from −52 °C to +9 °C. Rapid growth rates persist even at extremely low NH3 concentrations, with base:acid ratios well below unity, indicating that NH3 is not required for particle growth. Together, our findings therefore suggest that MSA should be considered as being comparable with SA when evaluating acid–base new particle formation in cool regions of the marine atmosphere. To assess the potential impact of MSA on new particle formation, we have assembled global measurements of SA, MSA and IA in cool marine regions, made from ground stations and ships (Fig. 5) and from aircraft (Extended Data Fig. 6). These vapours generally coexist at all of these locations with concentrations between 104 and 107 cm−3. Moreover, the MSA and SA concentrations are closely matched, with simultaneous MSA:SA ratios predominantly in a narrow range between 0.7 and 1.3 (bottom panels in Fig. 5 and Extended Data Fig. 6). When combined with our results, it implies that present model studies may be underestimating the effective acid concentrations that drive particle nucleation and growth in cool marine regions by around a factor of 2, which translates to underestimating the nucleation rates by an order of magnitude and growth rates by twofold. Global SA, MSA and IA observations from ground stations and ship expeditions. Violin plots showing ground-based and ship-based measurements of SA (red), MSA (green) and IA (purple) at the observatories or along the ship paths indicated on the map. Dates and locations (or titles) are indicated on each plot. The lower plot shows the simultaneous MSA:SA ratio for each site, with a linear y scale. The width of a violin distribution is proportional to the frequency of observation at the corresponding vapour concentration indicated on the y-axis. The white circles represent the median of the distribution and the attached vertical lines span the 25th to 75th percentiles. All data presented here have hourly time resolution. Details of the measurement sites and data sources are provided in Methods. Global SA and MSA observations from aircraft measurements. Violin plots showing airborne measurements of SA (red) and MSA (green) along the flight paths indicated on the map. Dates and campaign names are indicated on each plot. The lower plot shows the simultaneous MSA:SA ratio for each campaign, with a linear y scale. The width of each violin is proportional to the frequency of observation at the corresponding vapour concentration indicated on the y-axis. The white circles represent the median and the attached vertical bars denote the interquartile range (25th to 75th percentiles). Details on the flight campaigns and data sources are provided in Methods. This conclusion for nucleation rates depends on the presence of base vapours. However, we note that SA-NH3 ternary nucleation is also considered to be the dominant mechanism for SA-driven nucleation in the remote atmosphere44. An important source of NH3 in polar regions is guano from ubiquitous seabird colonies, which can produce between 10 and 1,000 pptv NH3 in both the Arctic45 and Antarctic44 boundary layers. The ammonia emissions persist even after migration of the birds, as they leave behind so-called ornithogenic soils, enriched with guano and vast amounts of organic matter and nutrients brought from the ocean to the land. In the Arctic, long-range transport of anthropogenic emissions can further increase NH3 levels46. In the cold temperatures of the free troposphere, MSA is produced with a high yield from DMS oxidation2 and can then drive new particle formation after mixing with NH3 convectively transported from the boundary layer. On the other hand, even in the absence of base vapours, we find that MSA accelerates the growth rates of particles above 1.8 nm in size, below +9 °C. This may be important for CCN concentrations over extensive regions of the cool boundary layer and free troposphere, because the main loss mechanism of small, highly mobile new particles is scavenging by pre-existing larger particles. Accelerated growth rates will result in reduced losses of the newly formed particles and a higher likelihood that they survive to reach CCN sizes. To evaluate the impact of MSA on global new particle formation, we have carried out an initial study with the EMAC (ECHAM/MESSy Atmospheric Chemistry) Earth system model (Extended Data Fig. 7). Our baseline simulation considers only SA(-NH3) nucleation28, (labelled ‘without’ MSA). We compare this with a simulation that adds MSA-SA-NH3 nucleation (labelled ‘with’ MSA). Here the sum of MSA and SA drives nucleation when the temperature is below −10 °C and [NH3] > 3 × [MSA]; outside these conditions, MSA does not contribute to nucleation. Under conditions in which MSA volatility is low5, MSA is considered to contribute equally with SA to particle growth, whereas it is treated as semi-volatile at higher temperatures and dry conditions (see Methods for further details). Our simulations indicate that MSA enhances nucleation rates and CCN concentrations especially strongly in the Arctic and Antarctic coastal regions, from the boundary layer up to at least 4 km altitude (Extended Data Fig. 7). These results are supported by recent observations of elevated gaseous MSA from aerosol evaporation during katabatic outflow from the Antarctic continent, which draws in upper tropospheric air from over the Southern Ocean25. Earth system models systematically underestimate CCN by more than 50% over the Southern Ocean, producing a large positive radiation bias in this climatically important region47,48. Our findings suggest that inclusion of MSA-driven new particle formation, which is not considered in present climate models, may substantially reduce their radiation bias in the Southern Ocean. Yearly averaged contribution of MSA to particle nucleation rates, N50 and CCN concentration simulated with the EMAC model. Panels a–c show the fraction of the total particle nucleation rates (J) attributable to MSA (Jwith MSA/(Jwith MSA + Jwithout MSA)) at the surface (a), 2 km altitude (b) and 4 km altitude (c) for the period 2016–2019. Panels d–f show the corresponding yearly averaged enhancement in N50 concentration owing to MSA at the surface (d), 2 km altitude (e) and 4 km altitude (f) for the period 2016–2019. Panels g–i show the corresponding yearly averaged enhancement in CCN concentration at 0.4% supersaturation owing to MSA at the surface (g), 2 km altitude (h) and 4 km altitude (i) for the period 2016–2019. The CCN enhancement ratio represents CCN concentrations relative to the baseline simulation, after accounting for SA-MSA-NH3 nucleation and MSA-driven particle growth measured in this study (see Methods for details). The baseline simulation considers only SA(-NH3) particle formation, calculated according to the Dunne et al. parameterization28, together with all primary aerosol emissions103. Our results indicate that purely biogenic marine sources produced more CCN in the pristine pre-industrial marine atmosphere than thought at present. Evaluating the influence of MSA on the baseline pre-industrial aerosol state will require careful consideration of the effects of cooler temperatures, which would enhance the yield of MSA from DMS compared with the present day, and of a greater extent of sea ice, which would displace phytoplankton and their DMS emissions. Increased biogenic CCN will tend to reduce estimates of Earth’s effective radiative forcing from present-day anthropogenic aerosol and, in turn, affect projections of warming later this century. SO2 emissions are now decreasing, which may return the future aerosol state of the atmosphere towards pre-industrial levels, buffered by biogenic particle formation in a warmer climate. Although future globally averaged seawater DMS concentrations are predicted to decrease, emissions of DMS into the atmosphere are expected to increase as a result of rising surface wind speeds and sea surface temperatures49. Together with falling SO2 levels, this suggests an increasing importance of MSA and of ammonia emissions for aerosol effective radiative forcing in the post fossil fuel climate. Methods The CLOUD chamber The experiments presented in this study were performed in the CERN CLOUD chamber. CLOUD is a cylindrical stainless-steel chamber measuring approximately 3.5 m in height and 2.5 m in diameter, with a total volume of 26.1 m3. It provides a controlled environment for studying the nucleation and growth processes of aerosol particles featuring an advanced environmental control system that regulates temperature, humidity and trace gas concentrations across the full range of boundary layer, tropospheric and lower-stratospheric conditions. The thermal housing surrounding the chamber ensures uniformity and stability of the chamber temperature over a wide range. In this study, experiments were conducted at temperatures of +9 to 10 °C, −10 °C, −30 °C and −52 °C. The chamber is operated in a continuous mode with ultrapure synthetic air, produced by mixing cryogenic liquids of oxygen (21% O2) and nitrogen (79% N2), continuously injected into the chamber. The gas-handling system is equipped with a variety of mass flow controllers (MFCs) covering different ranges and several dilution steps that enables the precise control and injection of trace gases at the desired level. Sulfur dioxide (100 ppmv SO2 in N2), ammonia (1% NH3 in N2), carbon monoxide (3% CO in N2) and dimethyl sulfide (1,000 ppmv DMS in N2) were injected into the chamber from pressurized gas cylinders, whereas iodine (I2) and MSA were injected into the chamber from a temperature-controlled evaporator containing crystalline iodine (Sigma-Aldrich, 99.999% purity) and liquid MSA (Sigma-Aldrich, >99.95% purity), respectively. Ozone (O3) was introduced into the chamber by passing synthetic air through an ozone generator. O3 was photolysed in the chamber to produce OH radical using four 200-W Hamamatsu mercury–xenon ultraviolet (UV) lamps of wavelength range between 250 and 450 nm or using a xenon fluoride excimer laser (UVX) at 248 nm. To prevent cross-contamination, all gases are delivered through independent gas lines. Further, the chamber undergoes a thorough cleaning process involving rinsing the chamber walls with ultrapure water and subsequent heating to 373 K for more than 24 h in between different experimental programmes to maintain minimal cross-contamination between different runs. The chamber is also operated at different ionization levels. In neutral (N) experiments, a powerful electric field (±30 kV) is applied, rapidly eliminating all small ions from the chamber within a second to establish an ion-free environment. In ground-level GCR experiments, the natural ionization from cosmic rays is maintained. Furthermore, in BEAM experiments, the chamber is exposed to a 3.5 GeV c−1 secondary pion beam from the CERN Proton Synchrotron, which substantially increases the ionization to levels akin to those observed in the upper troposphere–lower stratosphere. The experiments presented here were performed largely at GCR conditions, with few experiments performed at neutral conditions. Uniform mixing of particles and vapours within the chamber is achieved using two magnetically coupled stainless-steel fans positioned at the top and bottom of the chamber. This set-up ensures rapid and efficient mixing within a few minutes. The data presented in this study were collected during CLOUD 13, CLOUD 14, CLOUD 15 and two experiments from CLOUD 16. The CLOUD 13 and CLOUD 14 campaigns were conducted consecutively in 2018 and 2019, whereas CLOUD 15 and CLOUD 16 took place in 2022 and 2023, following the renovation of the building housing the CLOUD chamber. The renovation focused on upgrading the facilities in the building, whereas the CLOUD chamber itself remained unchanged. To ensure consistency, pure SA and iodine oxoacid new particle formation experiments were conducted during each campaign. The results were compared with well-established parameterizations, confirming the consistency of the CLOUD chamber before and after the renovation. The new particle formation experiments To investigate the role of MSA in new particle formation, we conducted experiments at different temperatures. Before each experiment, precursor gases (O3, SO2, DMS) were introduced into the chamber and allowed to stabilize, with fans operating at maximum speed (100%) to ensure efficient turbulent mixing and light illumination switched off. Once particle-free conditions and stable precursor concentrations were established, the experiments were initiated by reducing the fan speed to 12% and turning on the lights. This triggered photochemical OH radical production, leading to the oxidation of DMS and the subsequent formation of SA and MSA. Varying the MSA:SA ratio from DMS oxidation over a wide range is challenging, particularly for achieving higher MSA:SA ratios at warmer temperatures. To address this, further experiments were conducted at +9 °C, at which MSA was directly injected into the chamber from an evaporator, while SA was generated through the photo-oxidation of SO2. Following particle generation in the chamber, the experiment was continued until a stable particle nucleation rate (for nucleation rate experiments) or the desired particle size growth (for growth rate experiments) was achieved. During the MSA injection experiments at +9 °C, high concentrations of MSA may accumulate near the bottom of the chamber before mixing throughout its total volume. To evaluate the effect of this phenomenon, blank experiments with pure MSA injection were conducted, confirming that no particles formed in the absence of SA. Furthermore, in some experiments, a small amount of iodine was added alongside MSA and SA to simulate the marine and polar atmospheric conditions. During these experiments, a green light (528 nm) was used to photolyse I2 into iodine atoms, which were then rapidly oxidized in the presence of water vapour and ozone, leading to the formation of both IA and IOA. Throughout the experiments, environmental conditions, trace gas concentrations, particle concentrations and composition were continuously monitored in real time, as explained in subsequent sections. To quickly clean the chamber between experiments at the same temperature, we turned off the lights or gas injection, increased the fan speed to 100%, activated the high-voltage clearing field in alternating field mode and sometimes increased the ionization rate for small particles to facilitate their removal by the alternating field. The experiments were always conducted from the highest to the lowest temperatures to prevent contamination release from the chamber walls. Instrumentation SA, MSA and iodine species The concentrations of SA, MSA, IOA and IA are critical in this study and were monitored using the nitrate CIMS (NO3-CIMS; TOFWERK AG) coupled with a corona nitrate ion source50 and the bromide CIMS coupled with a multi-scheme chemical-ionization inlet (Br-MION-CIMS)51,52. The NO3-CIMS, identical across the CLOUD 13, CLOUD 14, CLOUD 15 and CLOUD 16 campaigns, is described in detail in a previous publication53. The Br-MION-CIMS was not used during CLOUD 14 and we used different instruments and inlets for CLOUD 13 (Br-MION1-CIMS) and CLOUD 15 and CLOUD 16 (Br-MION2-CIMS), although the underlying principles are identical. Therefore, we relied on the NO3-CIMS measurements for the reported concentrations of SA, MSA, IA and IOA, with cross-checks against Br-MION-CIMS or against parameterizations of new particle formation from iodine oxoacid (IAx)40,41. For CLOUD 13, CLOUD 15 and CLOUD 16, we performed several SA calibration experiments using two independent calibration systems before, during and after the campaigns54. Although the instrument is mainly calibrated for SA, the same calibration factor is applied to quantify IA55, IOA40 and MSA2, as extensive research has shown that these compounds can be detected at the collision limit. After applying the respective calibration factors, we compared the measured concentrations of SA, MSA and IA from both instruments across different temperatures. These comparisons demonstrated strong consistency, validating the accuracy of the SA calibrations. The calibration coefficients applied for the CLOUD 13 and CLOUD 15 campaigns were 1.1 × 1010 and 6.2 × 109 cm−3 per normalized signal (cps cps−1), with uncertainties of 50–150% (ref. 2) and 88–164% (ref. 56), respectively. During CLOUD 16, two SA calibration experiments were performed before and after the experiments, yielding calibration factors of 1.25 × 1010 and 1.85 × 1010 cm−3 cps cps−1. An independent SA calibration of the Br-MION2-CIMS yielded a calibration factor that was 74% of the lower value (1.25 × 1010 cm−3 cps cps−1). Accordingly, we used a final SA calibration factor of 1.25 × 1010 cm−3 cps cps−1 with an estimated uncertainty of 74–148%. During CLOUD 14, we were unable to cross-check the measured acids concentration with a second instrument owing to the unavailability of Br-MION-CIMS. However, we conducted the same set of iodine oxoacid nucleation experiments at −10 °C during CLOUD 13, CLOUD 14 and CLOUD 15. The measured nucleation rates as a function of IAx concentration across these campaigns were consistent, indicating that the SA calibration experiments performed for NO3-CIMS were reliable and accurate. The calibration factor used during the CLOUD 14 campaign was 4.8 × 1010 cm−3 cps cps−1, with uncertainty of 50–150% (ref. 2). DMS DMS concentration was measured during our experiments using two separate instruments: a proton-transfer reaction mass spectrometer 3 operating in hydronium chemical-ionization mode (H3O-PTR3-CIMS)57 and a proton-transfer reaction mass spectrometer using hydronium as the reagent ion (H3O-PTR-MS)58. Throughout all campaigns, the instruments were regularly calibrated with a standard gas mixture containing various volatile organic compounds to correct for potential drifts in transmission efficiency or variations owing to absolute humidity. The estimated uncertainty in DMS concentrations is approximately 30% (ref. 59). During CLOUD 15 at −30 °C, DMS concentration was measured with a FUSION PTR-TOF60 (IONICON Analytik GmbH). Quantification was based on the calibrated sensitivity to acetone, which was 3,600 cps ppbv−1. The DMS signal, recorded counts per second, was normalized by acetone sensitivity and corrected for the reaction rate constants, mass-dependent transmission efficiency and isotope abundance. However, because DMS serves as the precursor to SA and MSA, both of which are directly measured, the concentration of DMS is not crucial to the results reported in this study. NH3 Real-time measurements of gas-phase ammonia concentrations is challenging, especially at low concentrations owing to its high water solubility and strong surface affinity. In this study, NH3 concentrations were either directly measured or estimated from the injection rate. During the CLOUD 13 campaign, NH3 was measured using a protonated water cluster CIMS (CI-APi-TOF), which achieved a background level of approximately 4 pptv at 278 K and 80% RH. Detailed descriptions of the measurement set-up at CERN, including calibration and quantification procedures, are described elsewhere61. During the CLOUD 14 and CLOUD 15 campaigns, NH3 was measured using the H3O-PTR3-CIMS, which is not optimized for NH3 measurements. The NH3 concentrations in the chamber at low pptv level were within the same range as the background signal of the instrument. The instrument background was also not stable but influenced by NH3 interactions with the internal surfaces, which depend not only on the surface materials used but also by changes in absolute humidity and temperature in the sample air. Specifically, ammonia can accumulate on the instrument surfaces and be released when humidity and temperature increase. In this study, we must also consider the stainless-steel chamber walls, which can store ammonia and subsequently release it at high temperatures. By contrast, under very cold and dry conditions, the chamber walls act almost as a perfect sink for NH3. Detailed information on instrument set-up, calibration and the effects of humidity and temperature is available elsewhere62. Given these measurement challenges and instrument limitations, the NH3 concentrations reported from H3O-PTR3-CIMS represent an upper limit of expected concentrations. Furthermore, the H3O-PTR3-CIMS was unavailable during the CLOUD 15 −30 °C experiments, presented in this study. Therefore, for these experiments, the NH3 concentration in the chamber was estimated by numerically integrating a simple box model (gas and wall) with 1-s time steps. The model accounts for (1) continuous dilution owing to chamber flow, (2) reversible uptake to and re-evaporation from the chamber walls and (3) the NH3 injection rate (ppt s−1). The two equations solved are: in which C(t) denotes the gas-phase NH3 mixing ratio (pptv), S(t) is the NH3 injection rate calculated from the MFCs (ppt s−1), λ(t)=QV= Chamber flowChamber volume is the dilution rate (s−1) and kon and koff are the effective gas to wall uptake and wall to gas release rates, respectively (s−1). These values were estimated by fitting NH3 decay at −30 °C when NH3 measurements were present. The fit was performed using a weighted least-squares approach that accounted for the 30% measurement uncertainty of the H3O-PTR3-CIMS signal, yielding best-fit values and confidence intervals for the kon and koff rate constants. These uncertainties combined with a 10% uncertainty in the MFC-derived injection rate were propagated through the model using Monte Carlo analysis to obtain confidence intervals for predicted ammonia concentrations. The H3O-PTR3-CIMS was also not available during CLOUD 16. However, the CLOUD 16 experiments were conducted under low-NH3 conditions, with no NH3 injected into the chamber in the preceding months. The resulting background NH3 concentration remained below 4 pptv, consistent with previous characterizations of the chamber contaminant level61. Charged clusters Naturally charged clusters were measured using two APi-TOF mass spectrometers from Aerodyne Research Inc., both operating in negative ion mode. The first instrument, a standard APi-TOF63, was used in the CLOUD 13, CLOUD 14 and CLOUD 15 campaigns. It was connected to the chamber by means of a standard stainless-steel inlet. During the CLOUD 15 campaign, the APi-TOF mode51 of the Br-MION2-CIMS was also used to measure negatively charged clusters by deactivating the inlet voltages responsible for directing the charged reagent ions into sample flow. The MION2-APi-TOF provided higher sensitivity compared with the standard APi-TOF. Consequently, the charged cluster data for CLOUD 15 were mainly obtained from the MION2-APi-TOF, with the standard APi-TOF serving as a validation instrument. Particle-phase measurements To analyse the chemical composition of small particles, we used a FIGAERO coupled with a CIMS. This technique involves collecting particles on a polytetrafluoroethylene filter, followed by a controlled heating process to thermally desorb lower-volatility molecules back into the gas phase64. We confirmed complete desorption in our experiments by observing the return of all signals to baseline levels at the end of each heating cycle. Although FIGAERO-CIMS typically uses iodide (I−) chemical ionization for detecting low-volatility compounds, we used bromide (Br−) ionization during CLOUD 14, using the same instrument set-up, to differentiate between the chemical ionization reagent and iodine species within the CLOUD chamber. The collection times for particle samples varied across experiments: 15 min for CLOUD 15 and 30 min for both CLOUD 13 and CLOUD 14. Trace gases Trace gas concentrations were monitored using a Thermo Environmental Instruments TEI 49C chemiluminescence analyser for measuring O3 and a Thermo Fisher Scientific 42i-TLE pulsed fluorescence analyser for measuring SO2. Relative humidity A chilled dew-point mirror hygrometer (DewMaster, Edgetech Instruments) measured the dew point (above 0 °C) or frost point (below 0 °C) inside the CLOUD chamber. The measured dew or frost point was converted to the actual water vapour pressure using the saturation vapour pressure over liquid water or ice, respectively, calculated from the Goff–Gratch equations65. RH was then calculated at the chamber temperature either with respect to liquid water or with respect to ice, as specified. Temperature The chamber temperature was determined by averaging readings from six thermocouple sensors and one platinum resistance thermometer (PT100). These sensors are mounted at equal intervals along a horizontal line, positioned radially near the midplane of the chamber. The chamber temperature uniformity is within less than ±0.06 °C in the radial direction and ±0.1 °C in the vertical direction for temperatures ranging from −70 °C to 20 °C (ref. 66). Particle number size distribution Several instruments were used to measure the number size distribution of particles and ions across a range of about 1 to 1,000 nm. The Airmodus A11 nano Condensation Nucleus Counter (nCNC) measured particle number size distributions between 1 and 3 nm (ref. 67). This system consists of a particle size magnifier (PSM; Airmodus A10) and a condensation particle counter (CPC; Airmodus A20). The PSM operated in a scanning saturator flow mode between 0.1 and 1.3 l min−1, with data processed using the step inversion method. To minimize particle losses in the sampling line, a core sampling inlet was used, increasing total flow from 2.5 to 7.5 l min−1, with 5 l min−1 directed to a bypass and 2.5 l min−1 sampled by the PSM. The data were further corrected for losses within the 12-cm-long, 6-mm outer diameter core sampling piece. The Neutral cluster and Air Ion Spectrometer (NAIS) from Airel OÜ was used to determine the number size distribution of ions within a mobility diameter range of 0.8–42 nm, as well as the total aerosol particles (both naturally charged and neutral) in the range of about 2–42 nm (refs. 68,69). Like any other electrical mobility spectrometers, the NAIS classifies and detects ions based on their electrical mobility. The total flow rate of the instrument is 54 l min−1, with 20 l min−1 sampled from the CLOUD chamber. The remaining flow was supplied by the NAIS Integrated Dilution System from Airel. The number size distributions obtained from the NAIS were corrected for line losses, dilution and the ion transmission correction developed by Wagner et al.70, which is applicable to the NAIS model used in the CLOUD experiments. The NAIS was operated only in the ion mode measuring naturally charged particles during CLOUD 13 and CLOUD 14 whereas in CLOUD 15 and CLOUD 16, it was operated in both ion mode (measuring naturally charged particles) and total mode (measuring both naturally charged and neutral particles). Two commercially available Scanning Mobility Particle Sizer (SMPS) spectrometers71 from TSI were used to measure particle size distributions in the range 6–1,000 nm. The first instrument, termed the nanoSMPS, is a model 3936 equipped with a 3085 Nano DMA and a 3750 CPC. The second instrument, referred to as the longSMPS, is a model 3983 that featured a 3082 electrostatic classifier, a 3081a DMA and a 3775 CPC for particle detection. Both systems were connected to the chamber by means of core-sampling inlets to minimize particle losses. The measured particle size distributions were then corrected for expected line losses. During the CLOUD 13 campaign, a differential mobility analyser-train (DMA-train) was also available. The DMA-train measures the particle number size distribution within a mobility diameter range 1.8–8 nm (ref. 72). It consists of six DMAs operating in parallel, each connected to a CPC for particle detection. The DMA-train has an inlet flow rate of 11 l min−1. To increase the flow in the main sampling line and reduce diffusional sampling losses, an extra 9 l min−1 of air was drawn in through a core-sampling inlet. In this study, the DMA-train was used to cross-validate the growth rates determined from the NAIS (see below). Determination of the particle nucleation rate (J1.7) The particle nucleation rate reported here corresponds to the flux of nanoparticles that grow to a mobility diameter of 1.7 nm. This follows previous CLOUD studies, as most chemical systems are expected to have formed stable clusters by this size, at which evaporation is negligible under the studied conditions. The rate is derived from the time evolution of the particle concentration (Ndp), under the assumption that new particle formation is the dominant source term73. Other potential sources, such as coagulation of smaller clusters or evaporation from larger particles, are typically negligible in new particle formation chamber experiments and are therefore neglected. In chamber experiments, particle losses occur through three main mechanisms: (1) coagulation with equal-sized or larger particles; (2) deposition onto chamber walls; and (3) dilution resulting from chamber volume exchanges. J1.7 is therefore expressed as: in which dN1.7dt is the time derivative of the total particle concentration above 1.7 nm, Scoagulation is the particle loss owing to coagulation with equal or bigger sized particles, Swall loss is the particle loss to the chamber wall and Sdilution is the particle loss owing to dilution, which depends on the chamber volume and the continuous replenishment flow into the chamber. Further details on calculating each term for the CLOUD chamber are provided in ref. 74. To calculate the coagulation sink, we used the combined particle size distribution from three instruments (NAIS, nanoSMPS and longSMPS), whereas the nCNC data were used to provide the number concentration of particles larger than 1.7 nm. The measured particle nucleation rates were compared with those expected for the SA-NH3 system using the parameterization in ref. 28 and with those expected for the IAx-SA-NH3 system based on ref. 40. In the latter case, J1.7 was estimated using the empirical relationship between the nucleation rate and the concentrations of SA, IA and IOA. Specifically, He et al.40 showed that log10(J) scales almost linearly with log10((HIO3 + H2SO4) × HIO2), such that the nucleation rate can be approximated as: Determination of the growth rate The growth rates were calculated using the 50% appearance time method43,74,75 over two size intervals: 1.8–3.2 nm and 3.2–8.0 nm in mobility diameter, consistent with the approach in ref. 43. The main instrument used for growth rate determination was the NAIS. However, during the CLOUD 13 campaign, when the DMA-train was available, further growth rates were calculated using DMA-train data. The DMA-train growth rates represent the growth of the total particle population, whereas the NAIS provided growth rates for both negatively and positively charged particles (ions) during CLOUD 13, CLOUD 14 and CLOUD 15. Furthermore, during CLOUD 15 and CLOUD 16, the NAIS was also operated in total mode, enabling calculating the growth rate for the total particle population in the 3.2–8.0 nm range. As shown in Extended Data Table 1, the growth rates calculated from the two instruments demonstrate strong agreement, even when comparing total particle growth rates to those of naturally charged particles. We also compared growth rates between the DMA-train and the NAIS in total mode for experiments with α-pinene and SA, confirming that growth rates measured by the NAIS in total mode closely align with those measured by the DMA-train. Extended Data Table 1Comparison of growth rates measured using the DMA-train with those obtained from the NAIS in both negative and positive ion modes Comparison of growth rates measured using the DMA-train with those obtained from the NAIS in both negative and positive ion modes The appearance time method can overestimate growth rates when coagulation growth is pronounced relative to condensational growth76. To evaluate this, we calculated the coagulation growth rate using the equations in ref. 76 for selected runs with GR3.2–8.0 values ranging from 1.79 to 26.0 nm h−1. The results showed that coagulation growth contributed at most 0.01% to the growth rate derived from the appearance time method. Consequently, coagulation growth is negligible in our experiments and can be disregarded. Atmospheric observations Global observations of SA, MSA and IA from ground-based stations and shipborne expeditions, as shown in Fig. 5, were compiled from the literature. To facilitate comparison across sites, all data were averaged to hourly resolution. The following locations and sources were included: Villum (81.6023° N, −16.6620° E, 24 m above sea level (a.s.l.)) and Ny-Ålesund (78.9255° N, 11.9323° E, 45 m a.s.l.) from ref. 21; the MOSAiC expedition in the central Arctic Ocean from ref. 77; Värriö (67.7481° N, 29.6102° E, 370 m a.s.l.) from ref. 78; Helsinki (60.2049° N, 24.9636° E, 25 m a.s.l.) from ref. 79; Réunion Island (−21.0800° N, 55.3830° E, 2,150 m a.s.l.) from ref. 27; the Southern Ocean expedition from ref. 20; Aboa (−73.0500° N, −13.4167° E, 480 m a.s.l.) from ref. 23; Marambio (−64.2411° N, −56.6267° E, 198 m a.s.l.) from ref. 24; Chacaltaya (−16.3498° N, −68.1307° E, 5,240 m a.s.l.) from ref. 59; Utqiaġvik (formerly Barrow) (71.3248° N, −156.6739° E, 3 m a.s.l.) from ref. 80; and CAPRICORN2 expedition from ref. 25. The aircraft measurements presented in Extended Data Fig. 6 were retrieved from several open-access repositories, in which ACE-1 stands for the First Aerosol Characterization Experiment81, PEM-Tropics B stands for the Pacific Exploratory Missions82, TRACE-P stands for the NASA Transport and Chemical Evolution over the Pacific83, INTEX-B stands for the Intercontinental Chemical Transport Experiment – Phase B84, PASE stands for the Pacific Atmospheric Sulfur Experiment85, ARCTAS stands for the Arctic Research of the Composition of the Troposphere from Aircraft and Satellites86 and SAS stands for Southeast Atmosphere Study87. The sampling periods as well as corresponding temperature and RH conditions of all atmospheric observations presented in Fig. 5 and Extended Data Fig. 6 are summarized in Extended Data Table 2. Extended Data Table 2Sampling periods and corresponding temperature and RH conditions for the atmospheric observations in Fig. 5 and Extended Data Fig. 6 Sampling periods and corresponding temperature and RH conditions for the atmospheric observations in Fig. 5 and Extended Data Fig. 6 EMAC model simulation Global simulations were performed using the EMAC model, a numerical chemistry and climate simulation system that includes submodels describing tropospheric and middle atmosphere processes and their interaction with oceans, land and human influences88. The core atmospheric model is the fifth-generation European Centre Hamburg general circulation model (ECHAM5 (ref. 89)), nudged using Newtonian relaxation towards ECMWF ERA5 reanalysis over the years 2016–2019. For the present study, we applied EMAC with a spherical truncation of T63 (corresponding to a quadratic Gaussian grid of approximately 1.88° at the equator) and 31 vertical hybrid pressure levels up to 10 hPa (about 32 km). The applied model set-up consisted of the submodels GMXe for aerosol microphysics using seven log-normal modes90, including: (1) treatment of multi-phase processes on deliquescent aerosols within the GMXe sub-submodel AERCHEM91; (2) NAN for the nucleation mechanisms92; (3) IONS for ion pair production rates from GCRs and radon decay92; (4) AEROPT for aerosol optical properties93; (5) MECCA for gas-phase chemistry94, using detailed DMS oxidation mechanism with extended halogen DMS reactions from the literature2; (6) JVAL for photochemistry95; (7) SCAV for cloud droplet multi-phase chemistry and the wet deposition of gases and aerosols96; (8) DRY-DEP for dry deposition97; (9) SEDI for aerosol sedimentation97; and (10) AIRSEA for emissions over the ocean98, following surface DMS climatology14. SA binary and ternary new particle formation is included28. For MSA nucleation, we use the parameterization based on the E-AIM thermodynamic model to determine the MSA volatility transition5, further requiring that the ambient NH3 concentration is greater than three times MSA to nucleate. Also, we restrict MSA nucleation to only happen at ambient temperatures below −10 °C. Under these conditions, the sum of MSA and SA concentrations available for nucleation are given to the ‘ternary’ (neutral and ion-induced) J1.7 nucleation channel described in ref. 28. If the conditions are not met, only SA concentrations are used to calculate the ‘ternary’ J1.7 rate. Particle growth is processed by GMXe’s modal 2 moments aerosol scheme, which includes condensation and coagulation processes. Gasaerosol partitioning is treated in ISORROPIA II (ref. 99) for semi-volatile inorganic species, whereas the submodel ORACLE100 treats the partitioning of organic species using a volatility-based approach. Condensation of inorganic and organic vapours increases particle mass and size within modal size bins, whereas coagulation transfers mass to larger modes and reduces number concentration. Under cold conditions, MSA condensation is treated analogously to SA, following ref. 101, otherwise MSA is treated as semi-volatile inorganic species. Low-volatility vapours that do not condense onto existing particles are supplied to the NAN submodel, which computes nucleation rates to form new particles added to GMXe’s nucleation mode. Finally, AERCHEM affects aerosol composition by means of non-equilibrium aqueous phase chemistry, indirectly influencing particle growth by altering hygroscopicity and water uptake. This integrated approach enables realistic simulation of aerosol size distribution dynamics and growth102,103. Online content Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41586-026-10810-2. Supplementary information Peer Review File Acknowledgements We thank CERN for supporting CLOUD with important technical and financial resources and for providing a particle beam from the CERN Proton Synchrotron. Extended data figures and tables Author contributions X.-C.H., R.B., J.S., K.L. and J.K. planned the experiments. R.B., J.S., M. Simon, H.K., J. DeVivo, M.W., W.S., L.D., B.R., D.S., H.E.M., E.S., L.C.-P., G.M., M.F., W.Y., M.L., D.A., A.A., T.A.-M., H.B., M.B., S.B., Z.B., R.C., J. Duplissy, H.F., M.H., F.K., H.L., B.L., N.M., V.M., M. Martinez, R. Marten, D.M., R. Mauldin, B.M., M. Müller, M.P., A.A.P., P.R., T.R., S. Richter, D.M.R., B.S., M. Surdu, R.C.T., Y.J.T., P.T., A.T., Y.T., J.T., A.C.W., D.S.W., Y. Wang, R.X.W., S.K.W., A.W., Y. Wu, M.Z.-W., J.Z., T.C., J.C., N.M.D., I.E.H., R.C.F., A.H., H.H., A.K., T.P., A.P., S.S., M. Sipilä, R.V., P.M.W., D.R.W., M.K., J.K., K.L. and X.-C.H. prepared the CLOUD facility or measuring instruments. R.B., J.S., M. Simon, H.K., J. DeVivo, M.W., W.S., L.D., B.R., D.S., H.E.M., E.S., L.C.-P., G.M., M.F., W.Y., M.L., A.A., T.A.-M., H.B., M.B., S.B., R.C., J. Duplissy, H.F., M.H., H.L., B.L., R. Mauldin, B.M., M. Müller, T.R., S. Richter, M. Surdu, A.T., Y.T., J.T., A.C.W., D.S.W., S.K.W., Y. Wu, M.Z.-W., J.Z., P.M.W. and X.-C.H. collected the data. R.B., J.S., M. Simon, H.K., J. DeVivo, M.W., W.S., L.D., D.S., G.M., M.F., W.Y., M.L., B.M., M. Müller, T.R., S.K.W., P.M.W. and X.-C.H. analysed the CLOUD data. R.B. collected and analysed the atmospheric observations data. S. Ruhl, A.P. and T.C. carried out the global model simulations. R.B., J.S., M. Simon, H.K., S. Ruhl, J. DeVivo, M.W., W.S., L.D., R. Mauldin, J.C., N.M.D., I.E.H., A.H., A.K., S.S., R.V., T.C., A.P., J.K., K.L. and X.-C.H. contributed to the scientific discussion. R.B., J.S. and S. Ruhl prepared the manuscript figures. R.B., J.S., J.K., K.L. and X.-C.H. wrote the manuscript. Finally, R.B., J.S., H.K., J. DeVivo, W.S., L.D., D.S., Z.B., H.F., A.A.P., J.C., N.M.D., I.E.H., A.H., A.K., A.P., S.S., J.K., K.L. and X.-C.H. commented on and edited the manuscript. Peer review Peer review information Nature thanks Hugh Coe and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available. Funding This work was supported by the EMME-CARE project, which received funding from the European Union’s Horizon 2020 Research and Innovation Programme under grant agreement no. 856612 and from the Cyprus Government. The Research Council of Finland supported this work through grant numbers 359331, 349659, 371185, 307331, 310682, 337550, 346372 and 296628. Further support was provided by the European Union’s MSCA Doctoral Networks ‘CLOUD-DOC’ project (grant no. 101073026), the MSCA Innovative Training Networks ‘CLOUD-Motion’ project (grant no. 764991) and the ERC FP7 ‘NANODYNAMITE’ project (grant no. 616075). We also acknowledge support from the German Federal Ministry of Education and Research (Bundesministerium für Bildung und Forschung; BMBF) for projects CLOUD-16 (grant no. 01LK1601A) and CLOUD-22 (grant no. 01LK2201A). Funding was also received from the European Research Council (grant no. 714621), the Swiss National Science Foundation (project numbers 200021_213071), the US National Science Foundation (AGS-1801574, AGS-1801897, AGS-2132089, AGS-2215489, AGS-2431817, AGS-1801280, AGS-2027252, AGS-2215522 and CHE2336463), the NASA fellowship awards (NASA-80NSSC17K0369 and NASA-NNX16AP36H), the Wallace Research Foundation and the CMU Scott Institute for Energy Innovation. W.S. acknowledges support from the Doctoral Scholarship of the University of Innsbruck (2021/2, no. 346360). J.Z. acknowledges support from the Research Council of Finland (grant no. 345982) and the Jane and Aatos Erkko Foundation. D.S.W. acknowledges funding from the MSCA-COFUND fellowship programme (grant agreement no. 701647). Open Access funding provided by University of Helsinki (including Helsinki University Central Hospital). Data availability The datasets underlying the experimental figures presented in this study are publicly available from Zenodo at 10.5281/zenodo.18997474 (ref. 104). Global atmospheric observations of sulfuric acid, methanesulfonic acid and iodic acid used in Fig. 5 are publicly available from the following repositories: Villum and Ny-Ålesund (10.5281/zenodo.4292239 (ref. 105)), MOSAiC expedition (https://doi.pangaea.de/10.1594/PANGAEA.963321 (ref. 106)), Värriö (10.5281/zenodo.5879549 (ref. 107)), Helsinki (10.5281/zenodo.6426198 (ref. 108)), Réunion Island (10.5281/zenodo.10302533 (ref. 109)), Southern Ocean expedition (10.5281/zenodo.2636771 (ref. 110); 10.5281/zenodo.3265832 (ref. 111); 10.5281/zenodo.5176217 (ref. 112)), Marambio (10.5281/zenodo.6560413 (ref. 113)), Chacaltaya (10.5281/zenodo.7429639 (ref. 114)), Utqiaġvik (10.5065/D6TM787P (ref. 80)) and CAPRICORN2 (10.25919/enc5-wh07 (ref. 115)). Aircraft measurements used in Extended Data Fig. 6 are publicly available from open-access repositories associated with the following campaigns: ACE-1 (10.26023/FRZK-NVH9-CT0S (ref. 81)), PEM-Tropics B (https://asdc.larc.nasa.gov/data/PEM/Tropics-B/TraceGas_AircraftInSitu_P3B_Data_1 (ref. 82)), TRACE-P (https://www-air.larc.nasa.gov/pub/TRACEP/P3B-AIRCRAFT/AIRCRAFT/EISELE.NCR/ (ref. 83)), INTEX-B – Phase B (https://www-air.larc.nasa.gov/cgi-bin/ArcView/intexb?C130=1#MAULDIN.LEE/ (ref. 84)), PASE (10.26023/P72H-HDDT-2Z0M (ref. 85)), ARCTAS (https://asdc.larc.nasa.gov/data/ARCTAS/TraceGas_AircraftInSitu_DC8_Data_1/ (ref. 86)) and SAS (10.26023/A58C-EF19-A80V). Code availability The custom MATLAB and Python codes used for data analysis and figure generation in this study are available from the corresponding author on reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Rima Baalbaki, Jiali Shen Contributor Information Katrianne Lehtipalo, Email: katrianne.lehtipalo@helsinki.fi. Xu-Cheng He, Email: xucheng.he@helsinki.fi. Extended data is available for this paper at 10.1038/s41586-026-10810-2. Supplementary information The online version contains supplementary material available at 10.1038/s41586-026-10810-2. References Associated Data Supplementary Materials Peer Review File Data Availability Statement The datasets underlying the experimental figures presented in this study are publicly available from Zenodo at 10.5281/zenodo.18997474 (ref. 104). Global atmospheric observations of sulfuric acid, methanesulfonic acid and iodic acid used in Fig. 5 are publicly available from the following repositories: Villum and Ny-Ålesund (10.5281/zenodo.4292239 (ref. 105)), MOSAiC expedition (https://doi.pangaea.de/10.1594/PANGAEA.963321 (ref. 106)), Värriö (10.5281/zenodo.5879549 (ref. 107)), Helsinki (10.5281/zenodo.6426198 (ref. 108)), Réunion Island (10.5281/zenodo.10302533 (ref. 109)), Southern Ocean expedition (10.5281/zenodo.2636771 (ref. 110); 10.5281/zenodo.3265832 (ref. 111); 10.5281/zenodo.5176217 (ref. 112)), Marambio (10.5281/zenodo.6560413 (ref. 113)), Chacaltaya (10.5281/zenodo.7429639 (ref. 114)), Utqiaġvik (10.5065/D6TM787P (ref. 80)) and CAPRICORN2 (10.25919/enc5-wh07 (ref. 115)). Aircraft measurements used in Extended Data Fig. 6 are publicly available from open-access repositories associated with the following campaigns: ACE-1 (10.26023/FRZK-NVH9-CT0S (ref. 81)), PEM-Tropics B (https://asdc.larc.nasa.gov/data/PEM/Tropics-B/TraceGas_AircraftInSitu_P3B_Data_1 (ref. 82)), TRACE-P (https://www-air.larc.nasa.gov/pub/TRACEP/P3B-AIRCRAFT/AIRCRAFT/EISELE.NCR/ (ref. 83)), INTEX-B – Phase B (https://www-air.larc.nasa.gov/cgi-bin/ArcView/intexb?C130=1#MAULDIN.LEE/ (ref. 84)), PASE (10.26023/P72H-HDDT-2Z0M (ref. 85)), ARCTAS (https://asdc.larc.nasa.gov/data/ARCTAS/TraceGas_AircraftInSitu_DC8_Data_1/ (ref. 86)) and SAS (10.26023/A58C-EF19-A80V). The custom MATLAB and Python codes used for data analysis and figure generation in this study are available from the corresponding author on reasonable request.