Introduction

Sea spray aerosol (SSA) is the largest source of primary particulate matter, with mass emissions ranging from 2 to 100 Pg yr−1 1. The dominant mechanism underlying the formation of SSA involves the bursting of bubbles at the ocean surface, which is induced by wind-driven air entrainment. During this process, film drops are produced as a result of bubble film rupture, while jet drops are formed due to the collapse of the bubble cavity2. These particles play an important role in the transportation of substances from the ocean to the atmosphere3 and are considered a crucial component of the climate system4,5.

Although wind speed is a driver for SSA formation, other environmental factors are emerging as crucial constraints in accurately assessing the climatic effects of SSA, considering the improved understanding of wind speed. Sea surface temperature (SST), with a global range from −2 °C to 35 °C, has impacts on SSA formation by influencing the surface tension, viscosity and density of seawater. Some investigations confirmed that SST increase enhances SSA mass production6,7 and found that the SSA source functions incorporating SST can improve the prediction accuracy of SSA emission flux8,9,10. However, these results are mostly from field observations, from which it is difficult to exclude external particle transport and the strong correlation between SST and wind speed11,12,13,14.

Laboratory research was conducted to characterize the properties of SSA in controlled artificial environments, and priority was given to the particle size distribution and number concentration (NSSA) of submicron SSA, because these particles determine the concentration of CCN in the remote marine boundary layer15. Although the plunging jet and the plunging waterfall have been proven as good laboratory-generated methods to replicate the dominant formation of SSA in real environments16,17, primary challenges still lie in addressing unexplained discrepancies in the dependency of NSSA on SST, as well as in clarifying the response of SSA particle size distribution to SST17,18,19,20.

There is a growing concern about the enrichment of organic matter in SSA particles. These organics originating from anthropogenic discharges and marine biological production21,22,23,24,25 will not only alter the atmospheric fate of particulate matter26,27,28,29,30,31, but also contribute to the atmospheric organic burden31,32,33. The enrichment factor (EF), defined as the ratio of the concentration of the target substance to that of sodium (Na+) in SSA and seawater, provides a way to quantify the organic matter enrichment in SSA particles and allows for further assessment of the emission fluxes of marine substances via SSA21,34,35. SST may affect the organic enrichment in SSA particles by influencing processes like organic pre-enrichment in the sea surface microlayer (SML), underwater bubble scavenging, and surface bubble bursting, but research in this field is limited. Bates et al.36 reported a potential positive correlation between particulate organic carbon and SST, but without a clear explanation.

In this study, we designed a thermostatic SSA simulation tank (T-SSAT) that incorporates both plunging jet and plunging waterfall, along with a temperature control system. The tank can maintain uniformed experimental settings to comparatively analyze the response of SSA particle size distribution to SST variations in different SSA generation methods. Sodium dodecyl sulfate (SDS), a typical surfactant, and dissolved organic carbon (DOC) produced by Karenia mikimotoi were selected as representatives of terrestrial anthropogenic and marine biogenic sources, respectively, to explore the effect of SST on organic enrichment in SSA.

Finally, we applied our experimental results to a global aerosol simulation for estimating the importance of SST on SSA-derived CCN and the global ocean organic matter emissions.

Results and discussion

A series of laboratory experiments were carried out to investigate the effect of SST on SSA formation, and the overview of experiments is summarized in Table 1. See “Methods” for a detailed description of the experiments.

Table 1 Overview of experiments
Full size table

Effect of SST on SSA size distribution

Plunging jets and plunging waterfall have been proven to reproduce SSA particle size distributions and bubble size distributions in real environments16,17. These methods offer advantages such as moderate volume and controllable environmental conditions, making them widely used in studies related to seawater temperature. Given the discrepancies in the dependencies of NSSA on SST discussed above, the plunging jet and plunging waterfall methods were used to compare and analyze these differences, thereby clarifying SST’s common impact on SSA size distribution. Figure 1 presents the dependencies of NSSA on SST with two generation methods. Owing to the higher entrainment efficiency of the plunging jet37, the SSA production efficiency of the plunging jet is higher than that of the plunging waterfall, even with a lower recirculation flow rate. As for artificial seawater (ASW), NSSA generated by plunging jet shows a nonlinear trend with SST: as SST decreases from 30 °C to 12 °C, NSSA decreases at a rate of 1.79% °C−1; as SST drops further to 0 °C, NSSA instead increases at a rate of 1.63% °C−1. In the plunging waterfall, NSSA decreases almost linearly by 2.16% °C−1 as SST decreases from 30 °C to 0 °C. Similar to previous studies, the SST-dependent results of NSSA exhibit contradictions.

Fig. 1: Dependencies of NSSA on SST.
figure 1

Experimental results were obtained from two widely used SSA laboratory generation methods: a plunging jet and b plunging waterfall. The orange color represents the results from artificial seawater, and the green color represents the results from natural seawater. The y axis error bars come from 4 to 6 measurements of SMPS, and the x axis error bars come from 40 to 60 recordings of a temperature sensor.

Full size image

SSA is produced by the bursting of bubbles on the water surface; hence, surface bubble area (Sbubble) has an important impact on controlling NSSA. We captured images of Sbubble at each temperature and quantified them through manual marking and analysis using Image J, following the methodology in the study by Harb et al.38. Given the subjective marking, obtaining precise data is difficult. Therefore, Fig. S1 displays normalized results that depict the trend of Sbubble with SST. In jet-type experiments, there was a notable decrease in Sbubble as SST increased from 0 °C to 12 °C, and it subsequently exhibited a slight increase as SST further rose to 30 °C. Conversely, waterfall-type experiments show a positive correlation between Sbubble and SST. Our results suggest that Sbubble plays an important role in controlling SSA production. In fact, bubbles on the water surface represent the final stage of air degassing in seawater, resulting from a combination of previous air entrainment and bubble plumes39,40. From a hydrodynamic perspective, different plunging modes can affect air entrainment, and different water entry velocities determine the depth of the bubble plume. For instance, the greater the bubbles depth, the less probable it is to reach the surface of the water41. All these factors may contribute to the different results observed between the plunging jet and the plunging waterfall experiments. In comparison to ASW, the existence of organic matter (367 ± 12 μM C L−1) in natural seawater (NSW) results in a 2–3 times reduction in NSSA for both generation methods, but minimally affecting their dependencies of NSSA on SST (Fig. 1). Certain surface-active organic compounds can modify the properties of seawater, influencing the air entrainment upon water surface impact39,42. Additionally, they can decrease SSA production by stabilizing bubbles at the water surface43.

The responses of SSA particle size distribution to variations in SST appear to be similar in both methods. For ASW, decreasing SST reduces the production of larger SSA and enhances the production of smaller SSA (Fig. 2a, b). One possible explanation is that lower temperatures reduce the drainage speed of the bubble film by increasing the viscosity of seawater41, which allows more liquid to remain and form particles when they burst. Moreover, the decrease in SST can enhance the fragility of the bubble film, further encouraging the formation of smaller particles upon bursting20. For NSW, in addition to SST, the particle size distribution of SSA will also be influenced by surface-active organic matter. As shown in Fig. 2c, d, the presence of organic matter leads to a general shift toward particles with smaller sizes in both generation methods, altering the location of peak number concentration. This shift may occur due to the promotion by surface-active compounds to further thin the bubble film upon rupture, resulting in the production of smaller-diameter particles.

Fig. 2: The responses of SSA particle size distribution to variations in SST under different SSA generation methods and seawater medium conditions.
figure 2

a Plunging jet, ASW; b plunging waterfall, ASW; c plunging jet, NSW; d plunging waterfall, NSW.

Full size image

In order to quantify the effect of SST on the SSA particle size distribution, each particle size bin at 12 °C and 0 °C was normalized to that at 30 °C (Fig. 3). When SST decreased from 30 °C to 0 °C, a significant decrease in SSA particles was observed in both the accumulation mode (100–250 nm) and the coarse mode (>250 nm), while the number of particles in the Aitken mode (10–80 nm) increases substantially at 0 °C. This observed phenomenon is similar to the laboratory results of Salter et al., who reported a decrease in the ratio of 100 nm SSA to 375 nm SSA with increasing SST17, as well as the results from Sea Sweep deployments during the NAAMES2 cruise36. However, it is puzzling that Aitken mode particles seem to have no sensitivity to temperature in the study of Forestieri et al.18. As shown by the results of Salter et al.17, when the temperature decreases from 30.1 °C to −1.3 °C, the number of bubbles with radius (R) < ~1 mm increases significantly. According to the recently updated film drop production mechanism, a large increase in the number of these bubbles with R < ~1 mm does increase the number of Aitken mode particles44. Although our experimental results also show that the decreasing SST (especially from 12 °C to 0 °C) promotes the production of Aitken mode particles regardless of the generation mode, the sensitivity of those in plunging jets is higher. This may be a direct reason for the different dependencies of NSSA on SST between the two generation methods.

Fig. 3: Effect of SST on the production of SSA with different particle sizes.
figure 3

Each particle size bin’s change rate at 12 °C and 0 °C based on 30 °C: a plunging jet, b plunging waterfall. The red bin represents the rate of increase, and the blue bin represents the rate of decrease. The results were obtained from artificial seawater in order to exclude the interference of complex components in natural seawater and to improve the reproducibility of the experiments.

Full size image

Effect of SST on organic enrichment

The organic enrichment of individual SSA is primarily influenced by individual bubble size and rupture process. Thus, in both plunging jet and plunging waterfall, the organic enrichment of SSA produced by same-size bubbles undergoing the same rupture process could be similar under identical environmental conditions. In organic enrichment experiments, we chose the plunging jet for its higher SSA production efficiency, which can reduce sampling time and minimize potential transformations of organic matter during sampling. The EFs of SDS in size-resolved SSA particles at different SSTs are presented in Fig. 4. Elevated SST strongly increases SDS enrichment in SSA particles. The EF of SDS in SSA with the same particle size at 25 °C increased by approximately an order of magnitude compared to that at 0 °C. Faik et al.45 indicated that bubble terminal velocity was directly proportional to the variation of SST, which could be attributed to the reduction in water surface tension as SST rises. Besides, higher SST accelerates molecular diffusion in water, and reduces the adsorption equilibrium time of surfactants at the interface46. The combined influence of these two factors enhances the scavenging rate of surfactant by bubbles, thereby promoting surfactant enrichment in SSA particles. The largest increase in submicron particles was observed at 25 °C compared to 0 °C, especially in the 0.016-0.60 μm and 0.60-0.94 μm size ranges, with EFs of 12.2 ± 4.6 and 10.7 ± 5.9, respectively. Studies have demonstrated that high SST tends to extend the lifetime of water surface bubbles47 and increase film drainage speed48. This means that bubble film expels liquid longer and faster, thereby increasing the enrichment of SDS in bubble film and film drops. As film drops are the main contributor of submicron particles, the enrichment in submicron particles is thus more sensitive to SST. Additionally, SST does not change the enrichment trend of SDS on size-resolved SSA particles: EF decreases with increasing particle size. Among the six different SSTs, the average EF of SDS in the 0.016-0.60 μm particle size range is 21.4 ± 6.7 times higher than that in 5.34–10 μm range. Due to the amphiphilic nature, SDS can easily accumulate in the SML. Film drops produced by the bubble film can efficiently enrich a significant amount of SDS from the SML, while jet drops (the main contributor of supermicron particles) primarily originate from the air-water interface within the bubbles, which have lower EF values49. Our conclusion is that increasing SST promotes SDS enrichment in SSA particles of all sizes, with a greater sensitivity in submicron particles. This can increase the disparity in EFs of SDS among submicron and supermicron particles.

Fig. 4: EFs of SDS in size-resolved SSA particles at different SSTs.
figure 4

The y axis error bars from three repetitions of the experiment.

Full size image

As DOC occupies ~66.2% of the total mass of the marine organic carbon pool and is widely distributed in all ocean areas50, we proceeded to investigate the effect of SST on the enrichment of DOC in SSA particles. In Fig. 5a, it is shown that the EF of DOC in each sample type exhibits good positive correlations with SST increasing (r > 0.5). By analyzing the slope, k, of the fitted curve to represent the sensitivity of EF to SST, we observed the highest sensitivity for submicron SSA particles (k = 16.38 ± 3.53), followed by supermicron SSA particles (k = 5.36 ± 2.45). Compared to submicron SSA, the relatively weaker correlation and sensitivity between the EF of DOC and SST in supermicron SSA may be attributed to the formation mechanism of the SSA, which makes the composition of the supermicron SSA closer to that of bulk seawater, whereas the composition of submicron SSA is closer to that of SML25. The pre-enrichment of DOC in the SML also maintains a better positive correlation with SST than that of supermicron SSA. In contrast to SDS, the EF of algal DOC is at least two orders of magnitude lower, and the difference in EFs between submicron and supermicron particles is less pronounced. Indeed, there are wider varieties and higher concentrations of organic matter in algal DOC. As a result, the EF for DOC represents an average of all compounds, potentially obscuring those with very high EF, such as amino acids51. Furthermore, algae produce not only aliphatic-rich compounds that are easily enriched in the SML, but also water-soluble oxygen-rich compounds during their growth52. This contributes to a more even distribution of DOC concentration in both SML and bulk seawater. The organic volume fraction (OVF) is defined as the ratio of organic matter to the total particle volume, with the specific OVF formula provided in the SI. It is assumed that the core of the SSA particle is primarily inorganic, while the shell is predominantly organic53. OVF is more suitable for single-particle SSA morphology and phase analysis compared to EF. In Fig. 5b, as SST increases, OVF of submicron SSA particles increases from 0.12 to 0.24, while for supermicron SSA particles, it gradually increases from 0.06 to 0.11. Notably, good positive correlations between OVF and SST are observed for both particle size ranges (r > 0.5).

Fig. 5: Effect of SST on DOC and Ca2+ enrichment.
figure 5

a EFs of DOC, b OVF of SSA particles, c EF of Ca2+, d EFs of Ca2+ versus EFs of DOC.

Full size image

SST also affects the EFs of divalent cations. As shown in Fig. 5c, an increase in SST from 0 °C to 25 °C results in approximately 100% and 50% increases in the EFs of Ca2+ in submicron and supermicron particles, respectively. These EFs exhibit good positive correlations with SST (r > 0.5), while the increase in the EF of Mg2+ and its correlation with SST is less pronounced compared to Ca2+, as shown in Fig. S2a. Previous studies have reported the enrichment of Ca2+ and Mg2+ in SSA particles54,55,56,57,58. Building on the hypothesis that divalent cations can serve as bridges between organic molecules, and co-adsorbed into SML and SSA59,60, we conducted further investigation to examine the correlation between the EFs of divalent cations and DOC in different samples. In Fig. 5d, a strong correlation is observed between the EFs of Ca2+ and DOC in submicron particles (r = 0.95), while the correlation is weak in supermicron particles (r = 0.44). In Fig. S2b, the EF of Mg2+ exhibits a weaker correlation with the EF of DOC in all samples compared to Ca2+. This can be attributed to Mg2+ having a strong tendency to hydrate, which leads to weaker interactions with organic molecules60.

Transmission electron microscopy (TEM) can provide us with a better visual representation of the effect of SST on organic enrichment in SSA. In TEM images, the sea salt cores appear as dark and regular structures, while the organic matter is irregularly wrapped around the core35,61. This core-shell structure arises from the liquid-to-solid phase transition in which sea salts are the first to crystallize in the central region62. In Figs. S3 and S4, the thickness of the organic shell surrounding the sea salt core for SDS and DOC shows a minimum at 0 °C and a maximum at 25 °C. Increasing the organic shell thickness affects the phase state63, surface tension64, hygroscopicity29, and CCN activity3,30 of SSA particles, potentially increasing the uncertainty of climate prediction for SSA on global scales. Additionally, increased organic shell thickness with higher SST raises the question of whether higher SST enhances the emission of organic matter to the atmosphere via SSA.

Atmospheric implications

Aitken mode SSA-derived CCN

The involvement of SSA particles in cloud formation is directly influenced by their sizes and compositions. According to Köhler’s theory, for each dry particle, there is a defined critical diameter at a defined supersaturation. It is generally accepted that SSA particles with dry diameters larger than critical diameter are activated into CCN and subsequently grow to raindrop size before precipitation. Quantifying the contribution of SSA to CCN is essential to comprehending the radiation budget. A recent study revisited the global SSA-derived CCN, emphasizing the significant contribution of Aitken mode SSA to the formation of CCN5. SST variations cause changes in the size distributions of SSA particles. Consequently, we conducted additional investigations to assess how SST influences the contribution of the Aitken mode to submicron SSA-derived CCN.

The contributions of Aitken mode particles to CCN were calculated at different SST conditions, with supersaturation ranging from 0.1% to 1.0% (covering most supersaturation ranges in the marine boundary layer65). Detailed calculations are provided in the SI. Given that the strength of the effect of organic matter and biological activity in seawater on CCN activity remains inconclusive29,66,67,68,69, we targeted SSA produced by ASW to simplify these components’ effects. Figure 6 shows that the CCN contribution of the Aitken mode particles increases with decreasing SST at each supersaturation level. Compared to SSA produced at 30 °C, the Aitken mode SSA generated at 0 °C in ASW exhibits a higher CCN contribution, ranging from 145% to 236% across various supersaturation levels from the results of the two generation methods. However, this result has limitations when applied to equatorial and polar waters. In real environments, the particle size distribution, hygroscopicity, and surface tension of SSA can introduce uncertainties in its CCN activation. First, organic matter in NSW can influence the particle size distribution and number concentration of SSA, and this effect may depend on the species and concentration of organic matter. Second, in-situ and laboratory studies fail to systematically constrain the relationship between organic aerosols and CCN, with replicated experiments potentially yielding varying results50. Finally, although particle surface tension is crucial for CCN activity, quantifying the influence of surfactants on this surface tension remains challenging29. Nevertheless, at comparable supersaturation levels and DOC concentrations, the contribution of Aitken mode SSA to CCN tends to be lower in warm waters compared to cold waters. This occurs because higher SST reduces Aitken mode SSA production and increases organic enrichment in SSA (higher organic enrichment generally inhibits SSA’s CCN activity29,69).

Fig. 6: Contribution of Aitken modal SSA to submicron SSA-derived CCN at different SSTs.
figure 6

The results are obtained from a plunging jet and b plunging waterfall methods, respectively.

Full size image

Conversely, the reduction in SSA particle size due to lower SST may inhibit SSA CCN activation. According to Köhler’s theory, smaller particles require higher supersaturation to grow to critical diameter. This implies that at the same supersaturation, larger SSA particles from warm seawater are more likely to activate into CCN. Increased CCN concentration can enhance cloud albedo, thereby reducing solar radiation reaching the Earth’s surface4. This conclusion assumes that the effects of increased organic enrichment in SSA due to elevated SST and increased organic enrichment due to reduced SSA particle size as a result of reduced SST might offset each other in terms of CCN activation. The impact on precipitation is less clear because particles continue to absorb water vapor after reaching critical diameter until they become raindrops. This process involves competition for water vapor both within and between populations of CCN70, with factors such as number concentration, size, and chemical composition playing critical roles71. In summary, our laboratory results indicate the atmospheric implication of SST in influencing SSA-derived CCN, but more field observations are needed for validation.

Global substance emissions via SSA

The significant role of SST in organic enrichment has the potential to restructure the global distribution of oceanic substance emissions. Based on previous research results, SDS and algal DOC in this study can be used to represent the high- and low-enriched substances, respectively (Table S1). EFmixed is the EF for the full-size SSA, which is obtained from a Na+ mass-weighted average of size-resolved EF for different SSA particle sizes. The continuous variations in EFsmixed for high- and low-enriched substances over the global SST range result from polynomial fits to the data in Table S2 and S3. Figure 7a–d shows the global distributions of EFs for the two substances in January and July. High-enriched substances show greater differences in latitudinal distribution than low-enriched substances. Enclosed seas like the Gulf of Mexico, the Mediterranean Sea, and the Red Sea tend to experience stronger seasonal fluctuations in the EF. Transfer potential can be used to predict the emission capacity when the oceanic distribution of substance concentrations is unknown. Here, transfer potential is defined as the multiplication of the EF and SSA emission flux (see SI), and its results are presented in Fig. 7e–h. The SSA emission flux used here is derived from the results of the GEOS-Chem model. Although high SSA emission fluxes occurred over the North Atlantic in January and over the Indian and Southern Oceans in July (Fig. S5), their low SST significantly diminished the transfer potential of substances. Despite the seasonally high EFs values, the transfer potential of these enclosed seas is constrained by their relatively small SSA emissions. Previous studies overlooked the effect of SST and estimated the mass emissions of substances via SSA using EF obtained at specific temperatures. In contrast, for the first time, our assessment method takes SST into consideration. We use temperature-specific EF and temperature-considered EF to estimate mass emissions as calculated in the SI. The results presented in Table S4 indicate that the mass emissions calculated for the two substances using temperature-considered EF fall between those obtained at 20 °C and 25 °C.

Fig. 7: The influence of SST on the transfer of the high- and low-enriched substances at sea-air interface on a global scale.
figure 7

The results show (a–d) the global distributions of EF, and (e–h) the global distributions of transport potential. The transfer potential is the multiplication of the EF and SSA emission flux and represents the emission potential of a substance in a defined sea area.

Full size image

Based on transfer potential results, obtaining the concentration distribution of organics in global surface seawater would enable us to calculate the global emission flux of these organics via SSA. However, determining the concentration distribution of organics in global surface seawater is currently difficult due to their uneven spatial distribution. Even for the most abundant and well-mixed DOC in seawater, its concentration is influenced by the proximity to shorelines, river estuaries, biological activity, and ocean currents. In the open ocean, the DOC concentration in surface seawater is relatively constant at 63.4 ± 0.2 μmol C L−1 72, whereas in the open coastal waters, the average DOC concentration rises to 139 ± 8 μmol C L−1 73. Thus, we estimated the atmospheric emission flux of DOC using these concentrations as the lower and upper limits. Our estimation results indicate that the global DOC mass emitted to the atmosphere via SSA ranges from 1.95 to 4.65 Tg C month−1 or 23.45 to 55.78 Tg C yr−1 (see SI for more details on the calculation).

Although DOC is ubiquitous in the ocean, the available primary organic spray function relies on an empirical equation relating water-insoluble organic carbon and chlorophyll-a74. This function was summarized in the results of long-term observational data from Mace Head and had been applied to many model simulations33,75,76,77,78. As a land-based station in the northeastern Atlantic, observations at Mace Head are susceptible to high biological productivity levels in coastal waters. During high phytoplankton growth, large amounts of insoluble organic matter from algal photosynthesis are readily enriched in SSA, so chlorophyll-a concentration can be effectively used as a measure in the organic spray source function. In contrast, open ocean seawater is typically oligotrophic, and the concentration of refractory DOC, which has undergone multiple carbon cycles, is weakly correlated with chlorophyll-a concentration. For example, several large-scale oceanographic surveys in the open ocean revealed that chlorophyll-a concentration almost had no correlation with the level of organic enrichment in SSA, and DOC in seawater was found to be the primary source of organic matter in the SSA32,36. Studies using the chlorophyll-a organic spray function estimated that the global total organic spray could range from 20 to 50 Tg C yr−1 75,79. While these estimates are similar to ours, some discrepancies persist. For instance, their results show that high organic matter emission fluxes occur in waters with high chlorophyll-a concentrations, whereas our findings indicate high organic matter emission fluxes occur in waters with high SST and high SSA production. Objectively, both differences are reasonable as organic matter emission flux is affected by biological activity, SST, and SSA emission flux. Attempting to apply both estimations to coastal and open ocean waters in future models may improve the accuracy of the global distribution of organic spray emission flux. The IPCC report predicts a global average ocean temperature increase of ~2.5 °C by 210080. According to our estimates, considering only the impact of SST on EF and neglecting the effect of DOC concentration in seawater, global oceanic DOC emissions could increase by 1.70 to 3.92 Tg C yr−1 by the end of the century. However, rising SST also promotes algal bloom expansion, particularly in coastal waters81,82,83. The strong correlation among SST, chlorophyll-a, and SSA organic enrichment in coastal regions may render our estimation method inaccurate. Given global warming, the increase in atmospheric organic burden from oceanic sources may exceed our current expectations. Increasing organic matter will directly alter the properties of SSA particles on a global scale, thereby profoundly affecting SSA-cloud interactions and atmospheric chemistry.

We recognize several sources of uncertainty in our estimates. First, our approach simplified the DOC distribution on the ocean surface by using average DOC concentrations from the open ocean and coastal seawater as the bounds for emissions. Second, our estimation method is applicable mainly to oligotrophic open ocean waters, while the chlorophyll-a-based estimation is more appropriate for coastal waters with high biological productivity. Finally, our prediction of future increased organic emission via SSA does not adequately account for the effects of elevated SST on DOC concentrations in seawater and on coastal algal blooms, which may lead to an underestimate.

In this study, we investigated the impact of SST on the formation and climate effects of SSA by combining both laboratory and model simulations. Our results show that lower SST results in smaller SSA particle size, which may cause a higher concentration of Aitken modal SSA-derived CCN in cold waters compared to warm waters. Besides, SST significantly affects the organic enrichment in SSA particles, especially for small SSA particles and high-enriched substances. Building on this foundation, we provide a method to calculate the emission flux of oceanic substances by incorporating the global SST distribution. This study confirms and quantifies the important role of SST in SSA formation, substance transport, and climate effects, emphasizing the need to consider the critical contribution of SST in future large-scale studies of SSA.

Methods

Thermostatic SSA simulation tank

The main body of T-SSAT container that holds seawater is constructed from stainless steel and measures 30 cm × 30 cm × 55 cm (length × width × height). A jet nozzle is located at the center of the lid, and an overflow pipe with a lateral slit extends across the upper space of the container. Fig. S6 illustrates the perspective 3D structure of T-SSAT and annotates the function of each part. SSA can be generated independently using two methods: plunging jet and plunging waterfall, as shown in Fig. S7. Considering that the maximum height for jet drops ejection can reach 20 cm84 and the typical natural plume depth of oceanic bubbles is ~30 cm85,86, we positioned the distance from the water surface to the lid at 25 cm and maintained a seawater depth of 30 cm. A more comprehensive overview of the physical characteristics can be found in Table S5. An enclosed stainless-steel jacket was welded to the exterior of the tank. During operation, a coolant mixture (ethylene glycol: water = 1:1) acts as the heat transfer medium for the seawater, circulating between a thermostatic circulator (Alpha RA 24, LAUDA, Germany) and the jacket. To minimize heat loss in the system, we applied a layer of polyurethane foam to the exterior of the jacket and all pipelines. A temperature probe (HOBO MX2303, with an accuracy of ±0.02 K) monitored seawater temperature at 30-second intervals. Fig. S8 shows the effectiveness of temperature control.

Experiment summary

Before each experiment, the entire system was rinsed several times with different concentrations of isopropanol and ultrapure water (MilliQ, >18.2 MΩcm) to minimize organic residues. The system was considered clean when the surface tension of water in the tank equaled that of ultrapure water16. The tank was then filled with seawater, and the headspace was purged with purified zero air (Zero Air Supply, Model 111, Thermo Scientific) until the background particle concentration was reduced to ~0 cm−3. During SSA generation, zero air was introduced into the headspace as a sweeping airflow for the nascent SSA. The outlet port was fitted with a T-joint to regulate fluctuations between the airflow rates of the gas supply system and the sampling system. The first 15 min was the stabilization time of T-SSAT operation, with no measurements conducted. Afterwards, the sample airflow was dried by a diffusion drying tube (MD700-12F-3, Perma Pure, USA) to reduce the relative humidity to ~30%. To elucidate the effect of SST on SSA formation, three sets of experiments were carried out. Comprehensive details are presented in Table 1. The operation of T-SSAT is schematized in Fig. S9.

Experiments A were conducted using two-generation methods for two aqueous media compositions: ASW and NSW. ASW was prepared by dissolving 10 inorganic components in specific proportions into ultrapure water (as detailed in Table S6), which eliminates the interference of salinity and organic fluctuations in the experimental results. NSW collected from Shazikou, Qingdao, China, was filtered through 0.2 μm filter (Supor®-200, Pall Life Sciences, USA) and stored in pre-cleaned polyethylene containers to eliminate initial biological activity. When running a plunging jet or plunging waterfall, we keep the impact of the plunging water column on the water surface at a relatively constant state by using a constant circulation flow of seawater, which can eliminate the wind speed variable. During the experiments, SST was initially increased to 30 °C and maintained for 1 h. Then, the cooling mode was activated to gradually reduce the system temperature from 30 °C to 0 °C, and the sensitivity of SSA number concentration and particle size to SST was explored.

In Experiments B and C, SDS (≥99.0%, Aladdin) and algal DOC were used to represent terrestrial anthropogenic and marine biogenic sources, respectively. Alkyl sulfates (AS), including SDS, are among the most widely used surfactants. They are commonly found in industrial auxiliaries and personal toiletries, and are present in rivers and coastal ecosystems at concentrations reaching the μg L−1 level87,88,89,90. SDS has a high enrichment capacity in SSA and can represent the transfer patterns of organic pollutants such as perfluoroalkyl acids21,34,91,92, polycyclic aromatic hydrocarbons93, and human lung surfactants22, which also have high enrichment capacities in SSA. Shaharom et al. detected an average concentration of total surfactant around 0.5 µM in seawater at several nearshore stations, and this concentration was highly influenced by anthropogenic sources94. Therefore, SDS with a concentration of 0.5 µM was selected for the experiments. DOC is the largest reservoir of organic carbon in the oceans, primarily sourced from phytoplankton photosynthesis50. Karenia mikimotoi, as a widely distributed alga in the global ocean, has been highlighted for its ability to cause red tides in coastal areas around the world95. This makes it an ideal representative of the algal species in our experiments. See SI for details of algae acquisition, cultivation and treatment procedures. To isolate SST as the sole variable in the organic enrichment experiment, we used ASW, which can effectively eliminate the effects of organic matter and salinity fluctuation in NSW, providing a clearer reflection of SST’s impact on organic enrichment in SSA. The experiments were conducted at six temperature gradients: 0, 5, 10, 15, 20, and 25 °C. Each enrichment experiment was repeated three times at each temperature gradient.

Sampling and analysis

Number concentrations and size distributions of SSA particles were real-time monitored by a scanning mobility particle sizer (SMPS), consisting of a differential mobility analyzer (Model 55-L, GRIMM, Germany) and a condensation particle counter (CPC, Model 5416, GRIMM, Germany). The calibration and measurement methods are referenced in Wiedensohler et al.96. The SMPS scanned every 5 min at a sampling flow rate of 0.3 L min−1, covering a particle electrical mobility diameter range from 10.6 to 1055.5 nm.

SSA single particles were collected on carbon-coated Cu-TEM grids (T11023, Tianld Co., Ltd., China) using a single-particle sampler (DKL-2, Genstar electronic technology Co., Ltd., China) at a flow rate of 1 L min−1 for 30 min or 60 min. We followed the sampling method and TEM measurements of Mishra et al.97. The copper grids were stored in a dry environment at −20 °C until the particle morphology was analyzed by TEM (FEI Tecnai G2 F20, Thermo Fisher Scientific, USA). Subsequently, SSA particles were collected using a low-pressure cascade impactor (DLPI+, Dekati Ltd., Finland) with a flow rate of 10 L min−1 for 10 h. These particles were categorized by aerodynamic cutoff diameter within the range of 0.016 to 10 μm, distributed across 14 stages (Table S7). Each stage was loaded with 25 mm diameter quartz fiber filters (QMA, Whatman) which were pre-baked at 550 °C for 6 h. After collection, the filters were separated into several fractions. In experiment B, the initial 8 stages collected low-mass SSA particles, so these filters were combined and extracted with 12 mL ultrapure water, while filters from stages 9–14 were each extracted with 4 mL ultrapure water. In experiment C, all filters were divided into two parts: a submicron fraction of 1–9 stages (0.016–0.94 μm) and a supermicron fraction of 10–14 stages (1.62–10 μm), and each fraction was extracted with 12 mL ultrapure water. All extractions were assisted by ultrasound for 30 min and filtered through 0.2 μm filter before determinations. Sea SML, which is a thin layer of water at the sea-air interface, was sampled using the glass plate method98. An ion chromatography (IC, Dionex ICS-6000, Thermo Scientific, USA) was used to measure the concentrations of inorganic cations, including Na+, Mg2+, and Ca2+. A high-performance liquid chromatography-mass spectrometer (Thermo Scientific, USA) and a total organic carbon analyzer (TOC-5000, Metash, China) were conducted to quantify SDS and DOC concentrations, respectively, in the samples. Detailed instrumental analysis methods can be found in the SI. EF is the ratio of the target substance to Na+ in SSA particles or SML versus the ratio of the target substance to Na+ in seawater, given by Eq. (1):

$$\text{EF}=\frac{{(\text{X})}_{\text{SSA}\; {\text{or}}\; {\text{SML}}}/{({\text{Na}}^{+})}_{\text{SSA}\; {\text{or}}\; {\text{SML}}}}{{(\text{X})}_{\text{SW}}/{({\text{Na}}^{+})}_{\text{SW}}}$$
(1)

where, \({(\text{X})}_{\text{SSA}\; {\text{or}\; {\text{SML}}}}\) is the concentration of the target component in SSA particles or SML, (Na+)SW or SML is the concentration of Na+ in SSA particles or SML, \({(\text{X})}_{\text{SW}}\) is the concentration of the target component in bulk seawater, (Na+)SW is the concentration of Na+ in bulk seawater.

QA/QC

The calculation of the EF involves detections from different instruments, which can lead to uncertainties in the results. Therefore, we repeated the experiment three times for each condition to improve the reliability of the results. Furthermore, the following measures were taken to test the accuracy and precision of the analytical methods. The SDS standard solutions were configured at concentrations of 0.3, 0.5, 1.0 μM and determined by the analytical conditions provided in the SI. The recoveries were calculated in the range 64–105%, and the standard deviations (RSD) of the peak areas for the three replicates were in the range 4.5–10.1% (Table S8). The accuracy of the DOC detection was guaranteed using a midpoint concentration correction of the calibration curve, and if the relative error between the measured value and the true value was >10%, the standard curve was recreated. The precision of the DOC assay is accomplished by the instrument itself, which takes at least three measurements and gives a mean value with RSD < 3% (n ≥ 3). Additionally, two sets of blank samples were treated with the same steps of combination and extraction as in experiments B and C, respectively, to avoid any potential interference from DLPI+ substrate and quartz fiber filter residues. Na+ and organic carbon were observed above their corresponding instruments’ quantification limit but at low concentrations, which have been deducted from our results.

Model implementation

To extend the effect of SST on organic enrichment in SSA particles to a global scale, we combine global SST and SSA emission fluxes. SST data were sourced from the International Comprehensive Ocean-Atmosphere Data Set (ICOADS) via the National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL), and this dataset comprises long-term monthly averages of SST spanning from 1991 to 2020. The spatial distribution of global SSA emissions in 2018 was obtained from the GEOS-Chem Support Team at Harvard University. The range of SSA particle size included in this database is comparable to that in our enrichment experiments; see SI for a detailed comparison. To estimate the global transport potential of artificial surfactant and algal DOC via SSA, their EFmixed in SSA particles from 0.016 to 10 µm need to be calculated. We conducted a Na+ mass-weighted average of EFs for different particle size ranges:

$${\text{m}}_{{\text{Na}}^{+},\text{i}}={\text{c}}_{{\text{Na}}^{+},\text{i}}\times {\text{V}}_{\text{i}}$$
(2)
$${\text{EF}}_{\text{mixed}}=\sum {\text{EF}}_{\text{i}}\times \frac{{\text{m}}_{{\text{Na}}^{+},\,\text{i}}}{\sum {\text{m}}_{{\text{Na}}^{+},\,\text{i}}}$$
(3)

where, \({\text{c}}_{{\text{Na}}^{+},\,\text{i}}\) is Na+ concentration in the extract of fraction i; \({V}_{i}\), the volume of the extract; \({\text{m}}_{{\text{Na}}^{+},\text{i}}\), the mass of Na+ in fraction i; \({\text{EF}}_{\text{i}}\), the EF value for fraction i; \(\sum {\text{m}}_{{\text{Na}}^{+},\text{i}}\), the sum of Na+ masses of all fractions.