Three sites (Beijing, Handan, and Zhengzhou) in the NCP and two sites (Nanjing and Hangzhou) in the YRD were selected to collect ambient PM_2.5 and individual aerosol particles in December 2017 and 2020 based on the transport behavior of airborne pollutants under cold fronts in winter (Fig. S1 in the Supplement). Beijing, Handan, and Zhengzhou are located in the northern, central, and southern parts of the NCP, respectively. Beijing is a typical megacity and influenced by local emissions and regional transport, while Handan and Zhengzhou are two typical industrial cities and affected by local industrial, vehicular, and residential emissions. Nanjing and Hangzhou are two megacities located in the northern and southern parts of the YRD. These two megacities can be influenced when haze particles in the NCP invade the YRD. Therefore, the above five cities in the NCP and YRD are representative for exploring transboundary transport of aerosols (e.g., soot particles) in haze plumes. The detailed information of the sampling sites has been described in previous studies (Zhang et al., 2021, 2023).

Ambient PM_2.5 and blank samples (no pumping) were collected on preheated quartz filters with 90 mm diameters (600 °C for 4 h, Whatman) twice a day from 08:30 to 20:00 LT (local time) and from 20:30 to 08:00 LT the next day using medium volume samplers (TH-16A, Wuhan Tianhong, 100 L min⁻¹). Individual aerosol particles were sampled on transmission electron microscopy (TEM) grids and silicon wafers four times a day at around 02:00, 09:00, 14:00, and 20:00 LT utilizing individual particle samplers (DKL-2, Genstar, 1 L min⁻¹) equipped with a 0.5 mm jet nozzle impactor. To avoid particles overlapping on the substrate, the sampling duration of individual particles needs to be adjusted from 30 s to 15 min according to current PM_2.5 concentrations from 17 to 320 µg m⁻³. Hourly PM_2.5 concentrations and meteorological parameters including relative humidity (RH) and winds were derived from local monitoring stations (https://www.aqistudy.cn/, last access: 7 August 2025).

Water-soluble inorganic ions (i.e., NO3-, SO42-, NH4+, F⁻, Cl⁻, Na⁺, K⁺, Ca²⁺, and Mg²⁺), carbonaceous components (organic and elemental carbon), and trace metallic elements in PM_2.5 samples were analyzed using an ion chromatography system (Dionex ICs-90, USA), an OC/EC analyzer (Sunset Laboratory, USA), and inductively coupled plasma mass spectrometry (ICP-MS, Agilent 7500ce). The detailed experimental processes can be found in previous studies (Pan et al., 2013; Zhang et al., 2017a). Considering the contribution of other elements (e.g., O, H, N, and S) to the mass of organic matter (OM, i.e., organic compound), OM concentrations were obtained by multiplying organic carbon (OC) concentrations by 1.91 reported by Xing et al. (2013).

Morphology, mixing states, and compositions of individual aerosol particles collected on TEM grids were examined by TEM equipped with energy-dispersive X-ray spectrometry (EDS) (JEM-2100, JEOL). The acquisition time of TEM images and EDS spectra is usually controlled within 15 s because of the damage of electron beams to non-refractory aerosols. To better observe soot mixing structures and measure soot geometrical parameters, we enhanced the electron beam to sublime non-refractory coatings of indiscernible soot cores after conventional TEM observations. Copper element was excluded from particle EDS spectra because TEM grids are made of copper (Li et al., 2025). Using an image analysis software (Radius, EMSIS GmbH), we further obtained the equivalent circle diameter (ECD), length, and area of particles in TEM images. These parameters for indiscernible soot particles were measured under TEM observations with the enhanced electron beam. In this work, 3642 individual particles were analyzed by TEM in total.

Three-dimensional morphology of individual particles collected on silicon wafers was probed by atomic force microscopy (AFM, Dimension Icon) in tapping mode. Employing a professional image analysis software (NanoScope Analysis), the bearing area (A) and the bearing volume (V) of particles in AFM images were quantified. The ECD and the equivalent sphere diameter (ESD) of these particles can be calculated applying Eqs. (1) and (2). 1ECD=4Aπ2ESD=[3]6Vπ Figure S2 shows that there is a good correlation between the ESD and the ECD with slopes at 0.62 for passing through the inland pathway transport and 0.39 for passing through the sea pathway transport. According to the relationship between the ESD and the ECD, ESDs of particles observed by TEM were computed. The ESDs of soot particles were perceived as equivalent to their ECDs because they are composed of solid carbonaceous spheres that are not affected by substrates in terms of morphology (Barone et al., 2012; Li et al., 2016b).

The size ratio of soot-containing particles to their soot cores (Dp / Dc) was evaluated using Eq. (1): 3Dp/Dc=ESDsoot-containingESDsoot where ESD_{soot-containing} is the ESD of soot cores with their coatings and ESD_soot is the soot core ESD. The volume ratio of soot coatings to soot cores (Vcoating / Vc) was further calculated according to the Dp / Dc.

Based on the scaling law in the following equation, we obtained the fractal dimension (Df) of soot particles, which can be used to reflect the compactness of soot particles: 4N=kg2RgdpDf where N is the monomer number in soot particles, kg is the fractal prefactor, Rg is the gyration radius of soot particles, and dp is the average diameter of soot monomers.

The monomer number in soot particles and the gyration radius of soot particles can be calculated using the following equations: 5N=kaAaApα6Lmax2Rg=1.50±0.05 where Aa is the projected area of soot particles, Ap is the average projected area of soot monomers, ka is a constant, α is an empirical projected area exponent, and Lmax is the maximum length of soot particles.

Because the fractal prefactor is highly affected by the overlap between soot monomers (Lapuerta et al., 2015), an overlap parameter needs to be considered: 7δ=2al where δ is the overlap parameter of soot monomers, a is the average radius of adjacent soot monomers, and l is the spacing of adjacent soot monomers. ka and α depend on δ (Oh and Sorensen, 1997). Details can be found in Wang et al. (2017).

Wind fields coupled with surface PM_2.5 concentrations covering eastern China were obtained from European Centre for Medium-Range Weather Forecasts (ECMWF, https://earth.nullschool.net/, last access: 7 August 2025). Meteorological fields including winds and relative humidity (RH) in eastern China at 1000 hPa were simulated using 1°×1° Final Reanalysis Data (FNL) from the National Centers for Environmental Prediction (NCEP, https://rda.ucar.edu/datasets/ds083.2/, last access: 7 August 2025).

The potential geographic sources of PM_2.5 at observation sites (Nanjing and Hangzhou) in the YRD were identified based on the concentration-weighted trajectory (CWT) analysis. In this study, 72 h of air mass backward trajectories simulated from the wind data sets in the Nation Oceanic Atmospheric Administration (NOAA, ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1, last access: 7 August 2025) were used for the CWT analysis. The resolution of CWT trajectories consists of thousands of grid cells is 0.3°×0.3°. The equation for the CWT analysis is as follows: 8Cij=1∑k=1Nτijk∑k=1NCkτijk where Cij is the average PM_2.5 concentration in a grid cell (i, j); Ck is the measured PM_2.5 concentration for the trajectory k arriving at the observation site; τijk is the number of trajectory endpoints in the grid cell (i, j) for the Ck sample; N is the number of samples with trajectory endpoints in the grid cell (i, j).

In the CWT analysis, a weighing function as shown in Eq. (5) was applied to further improve the CWT accuracy: 9W=1forlog⁡(n+1)≥0.85×maxlog⁡(n+1)0.725for0.6×maxlog⁡(n+1)≤log⁡(n+1)<0.85×maxlog⁡(n+1)0.475for0.35×maxlog⁡(n+1)≤log⁡(n+1)<0.6×maxlog⁡(n+1)0.175forlog⁡(n+1)<0.35×maxlog⁡(n+1) where W is weighting factor; n is the number of all trajectory endpoints in a grid cell; log⁡(n+1) represents the density of trajectories.

The Electron-Microscope-to-BC-Simulation (EMBS) tool developed by Wang et al. (2021) was used to model morphology and mixing structures of soot particles. The EMBS tool capable of building various soot-containing particle models can be applied in DDSCAT 7.3 to calculate soot optical properties based on the discrete dipole approximation (DDA). DDA is completely flexible to the geometry of object particles under the condition that the inter dipole separation d follows |m|kd<0.5 and k=2π/λ, where m is the refractive index of particles and λ is the incident light wavelength. To minimize DDA uncertainty, the dipole size is much smaller than the soot monomer size. Based on sizes and mixing structures of soot-containing particles as well as Df and numbers of soot cores obtained from microscopic analyses, we employed EMBS and DDSCAT 7.3 to calculate the light absorption enhancement (Eabs) of soot-containing particles relative to their soot cores at 550 nm λ. In this study, soot-containing particles with one, two, and three or more soot cores were distinguished. The volumes of soot cores and their coatings remained constant in the optical calculation. The refractive indices of soot cores and coatings were considered as 1.85+0.71i (Bond and Bergstrom, 2006) and 1.53+0i (Worringen et al., 2008), respectively. Details about the EMBS and DDSCAT 7.3 can be found in the previous study (Wang et al., 2021).

Figure S3a and b shows variations in hourly winds and PM_2.5 concentrations at the observation sites in the NCP and the YRD from 28 to 31 December  017 and from 5 to 8 December 2020. Based on the association between PM_2.5 concentrations and prevailing winds described in Sect. S1 in the Supplement, we inferred that there was a typical transboundary transport process of pollutants from the NCP to the YRD on 30–31 December 2017 and 7–8 December 2020, respectively. Similar changes in PM_2.5 concentrations accompanied by winds were also found in many transboundary transport events of pollutants (Wu et al., 2022; Xie et al., 2023; Yan et al., 2024; Zhao et al., 2021).

Figure 1 displays meteorological fields coupled with surface PM_2.5 concentrations in eastern China during two transboundary transport events of pollutants. In the first transport event, the wind blew from the NCP through the inland pathway towards the YRD under the high-pressure system located in the west of the NCP (Figs. 1a, c, and S3c). This meteorological field was similar to those of transboundary transport haze events from the NCP to the YRD across inland areas (Hou et al., 2020; Hu et al., 2021; Zhang et al., 2023). It is interesting that there was a significant change in the wind field following the high-pressure system movement to the interior of the NCP during the second transport event compared to the first event, manifested as the wind mainly blowing from the NCP to the East China Sea and then to the YRD (Figs. 1b, d, and S3d). Such wind patterns were observed not only in the second event but also in other periods, as shown in Fig. S4. Although a previous study also discovered comparable wind fields between the NCP and the YRD using a weather model, the changes in chemical compositions and microphysical properties of haze particles have not been defined during the transboundary transport (Wu et al., 2022). To determine whether the transport pathway of pollutants was consistent with the wind field, the PM_2.5 transport pathway was simulated based on the 72 h CWT analysis (Fig. 2). Figure 2 shows that PM_2.5 in Nanjing and Hangzhou was mainly transported from the NCP through the inland pathway during the first transport event but through the sea pathway during the second transport event. Therefore, we concluded two transport models of haze pollutants from the NCP to the YRD, namely passing through the inland and through the sea pathways.

During two transboundary transport events, concentrations of chemical compositions in PM_2.5 in the NCP and the YRD significantly changed (Fig. S5). The concentration of secondary inorganic ions in the NCP decreased from 92–126 µg m⁻³ during the polluted period to 28–30 µg m⁻³ during the clean period (Fig. S5). OM and EC concentrations in the NCP also decreased from 43–76 and 1.5–2.1 µg m⁻³ during the polluted period to 17–31 and 0.7–0.9 µg m⁻³ during the clean period (Fig. S5). Following the transportation of large amounts of pollutants from the NCP to the YRD, the concentrations of secondary inorganic ions, OM, and EC in the YRD increased from 28–37, 13–19, and 1.0–1.4 µg m⁻³ during the clean period to 63–65, 32–36, and 1.6–2.7 µg m⁻³ during the polluted period, respectively (Fig. S5). These results suggest that many primary and secondary aerosols including EC were transported from the NCP to the YRD under cold fronts, both through the inland and the sea pathways. This is consistent with previous studies on the transboundary transport of haze aerosols from the NCP to the YRD (Huang et al., 2020; Li et al., 2019; Xie et al., 2023). In summary, we can confirm that these two events represent typical cases of transboundary transport of haze pollutants from the NCP to the YRD through the inland and the sea pathways.

Based on simulated meteorological fields, we noticed that polluted air masses passing through the sea pathway underwent wetter environment during transboundary transport compared to that passing through the inland pathway (Fig. 1c and d). Table S1 in the Supplement also shows much higher average RH at 90 % in the YRD following the transport of haze pollutants from the NCP to the YRD through the sea pathway in contrast to the inland pathway (RH = 83 %). High RH can contribute to the transformation of microphysical properties (e.g., mixing structures, sizes, and morphology) of soot particles in the atmosphere, but the reaction mechanism may vary under different high RH levels (Fu et al., 2022; Zhang et al., 2023). Consequently, we further investigated and compared the microscopic characteristics of soot particles during their transboundary transport through the inland and through the sea pathways.

Based on morphology, components, and mixing states of individual transported particles examined by TEM-EDS, they were classified into soot-containing, S-OM/metal/fly ash/mineral, S-rich, and OM/metal/fly ash/mineral particles (Fig. S6). The specific classification criteria were described in Sect. S2. Figure S6 shows that the number fraction of soot-containing particles in the NCP decreased from 45 % and 51 % during the polluted period to 13 % and 18 % during the clean period following transboundary transport of haze plumes through the inland and the sea pathways. When large amounts of haze pollutants were transported into the YRD from the NCP through the inland and the sea pathways, the number fraction of soot-containing particles in the YRD increased from 38 % and 34 % during the clean period to 53 % and 65 % during the polluted period (Fig. S6). The change in the number fraction of soot-containing particles in the NCP and the YRD during transboundary transport is consistent with the variation of EC concentrations. These results suggest that abundant soot-containing particles in the NCP were transported to the YRD following transboundary transport of haze plumes.

The morphology and mixing structures of soot particles can be changed during transport due to atmospheric aging (Li et al., 2024). Figure 3 shows morphology of soot particles and their mixing structures with other aerosol components observed by TEM. Based on the mixing structure of soot particles, they were divided into three categories: bare-like, partly-coated, and embedded soot particles (Fig. 3). Bare-like soot particles are characterized by being isolated and externally mixed with other aerosols (Fig. 3a). Partly-coated soot particles manifest as a portion of them being coated by other aerosol components (Fig. 3b). Embedded soot particles mean they are completely enveloped by other aerosol materials (Fig. 3c). Among these three types of soot particles, bare-like soot particles were considered to be more freshly emitted, while embedded soot particles were more aged (China et al., 2015). To observe some indiscernible embedded soot particles more clearly, their non-refractory coatings (e.g., S-rich particles) were sublimed under an enhanced electron beam (Fig. 3c). In some individual soot-containing particles, thin halos around aerosol components were observed (Fig. 3c). These thin halos have been confirmed to be water rims left by the dehydrating of aqueous particles because their EDS spectra are similar to the substrate but different from the organic coating (Zhang et al., 2023). Therefore, soot aggregates with water rims were identified as a type of embedded soot particles.

TEM observations showed that there were different numbers of soot cores in individual soot-containing particles during transboundary transport (Fig. 4a). Based on the number of soot cores in individual soot-containing particles, we further divided partly-coated soot-containing particles and embedded soot-containing particles into them with 1 soot core, 2 soot cores, and ≥3 soot cores (Fig. 4a). In this study, less than 10 % of soot-containing particles had both embedded and partly-coated soot cores. To categorize these particles, we classified those with more than 95 % of the total soot volume embedded in host particles as embedded soot-containing particles, and the remainder as partly-coated types. Because these particles were relatively few, they had a limited impact on the statistical results. Figure 4b and c shows the variation in the number fraction of soot-containing particles with different mixing structures and soot core numbers during transboundary transport through the inland and the sea pathways. During the polluted period, partly-coated types were dominant in soot-containing particles in the NCP, accounting for 62 %–67 % (Fig. 4b and c). Following the transboundary transport of haze plumes through the inland pathway, the dominated soot-containing particles changed from partly-coated at 67±4 % by number in the NCP to embedded types at 71±7 % in the YRD (Fig. 4b). Meanwhile, more than 75 % of them had one soot core (Fig. 4b). However, we noticed that the soot core number in the dominated soot-containing particles increased from 1 in the NCP to ≥3 in the YRD in addition to the change in the dominated mixing structures from partly-coated at 62±1 % by number to embedded ones at 72±4 % when plentiful soot-containing particles were transported through the sea pathway (Fig. 4c). These results indicate that soot-containing particles may be subject to different aging processes during their transboundary transport through the inland and the sea pathways. Moreover, large numbers of soot-containing particles with multiple soot aggregates were also observed in an aged atmospheric environment (Wu et al., 2016). However, their aging mechanisms were not effectively elucidated. The potential aging mechanisms for soot-containing particles in two transboundary transport events are discussed in the following section.

Figure 5 shows number percentages of bare-like, partly-coated, and embedded soot-containing particles with different numbers of soot cores in different size bins in the NCP and the YRD during two transboundary transport events. Figure 6 displays size distributions of partly-coated and embedded soot-containing particles during their transboundary transport. Bare-like soot particles were mainly concentrated in the finer size range of 0–200 nm during the transboundary transport (Fig. 5). In the NCP, partly-coated soot-containing particles with 1 soot core dominated soot-containing particles and mainly distributed in the size range of 200–500 nm during the polluted period (Figs. 4b, c and 5a, c). Figure 6a, and c shows consistent results that the size distribution of partly-coated soot-containing particles in the NCP had a peak at 396 nm for the transportation through the inland pathway and at 384 nm for the transportation through the sea pathway. As embedded soot-containing particles became the dominant type during the transboundary transport, their size distribution presented a peak at a larger diameter of 505 nm (inland) and at a much larger diameter of 925 nm (sea) compared to the former diameters at 464 and 446 nm (Fig. 6). The peak diameter at 505 nm for embedded soot-containing particles transported through the inland pathway is close to ∼550 nm of aged soot-containing particles during regional haze reported by Wang et al. (2019). Although number fractions of these embedded soot-containing particles with 1 core were high in both the 500–600 and 600–700 nm bins (Fig. 5b), the absolute number in the 500–600 nm range was approximately twice higher than that in the 600–700 nm range. As a result, the preponderant soot-containing particles in the YRD, i.e., embedded ones with 1 core (inland) and ≥3 cores (sea), dominated in the coarser size range of 500–600 nm and in the much coarser size range of >1600 nm, respectively (Fig. 5b and d). These findings suggest that aging processes of soot-containing particles during the transboundary transport through the sea pathway not only acquired more soot cores but also greatly enlarged their sizes in contrast to the inland pathway. Consistently, high numbers of soot cores were found in coarse particles of ≥800 nm during transboundary transport of biomass burning emissions (Chen et al., 2023).

The Dp / Dc and Vcoating / Vc ratios of transboundary soot-containing particles were calculated to reflect the coating thickness of soot particles and to quantify the aging degree of soot particles (Fig. 7 and Table S2). During two transboundary transport events, the mean Dp / Dc ratios of partly-coated and embedded soot-containing particles presented similar levels at 2.37–2.41 and 2.85–2.92 in the NCP (Fig. 7). Correspondingly, the mean Vcoating / Vc ratios remained at 12–13 and 22–24 (Table S2). Following the transboundary transport of soot-containing particles through the inland pathway, the mean Dp / Dc ratios of partly-coated and embedded soot-containing particles increased from 2.37±1.27 and 2.85±1.89 in the NCP to 2.79±1.37 and 3.41±1.87 in the YRD (P<0.05, Fig. 7a). Their mean Vcoating / Vc ratios also increased from 12 and 22 in the NCP to 21 and 39 in the YRD (Table S2). This amount of increase for the Dp / Dc ratio of soot-containing particles is comparable to that from ∼1.8 to ∼2.2 during haze evolution (Zhang et al., 2019b) and from 1.42 to 1.78 during dust storm transport (Xu et al., 2020). Moreover, consistent with these studies, we observed a transition in the dominant mixing structure of soot particles with secondary coatings from partly-coated with single soot core to embedded with single soot core configurations during the transboundary transport through the inland pathway (Fig. 4b), indicating that coagulation played a negligible role in the aging process (China et al., 2015). Soot particles have been demonstrated to provide a substrate for the formation of secondary aerosols via heterogeneous or aqueous-phase reactions (Farley et al., 2023; Han et al., 2013; Zhu et al., 2025). Figure 8 displays mixing structures of soot-containing particles when they invaded into the YRD through the inland and the sea pathways. It is noted that water rims around soot-containing particles transported through the inland pathway were not observed (Fig. 8a, c and d). This implies that aqueous-phase chemistry contributed minimally to secondary aerosol formation on soot particles during the transboundary transport through the inland pathway. As a result, heterogeneous aging processes might mainly drive the enhancement of secondary aerosols on soot-containing particles transported through the inland pathway and enlarged their Dp / Dc ratios.

When soot-containing particles were transported from the NCP to the YRD through the sea pathway, the partly-coated Dp / Dc ratio slightly increased from 2.41±1.37 to 2.66±1.58, but the embedded Dp / Dc ratio significantly increased from 2.92±2.01 to 4.38±2.92 (P<0.001, Fig. 7b). Consistently, the mean Vcoating / Vc ratios increased from 13 for the partly-coated structure and 24 for the embedded structure to 18 and 83 (Table S2). Similar results were also found in cloud processes with the Dp / Dc increase from 2.3 to 4.4 for embedded soot-containing particles reported by Fu et al. (2022). Moreover, Xu et al. (2020) showed a relatively high Dp / Dc increase proportion of soot-containing particles during the transportation of dust storms from China across the East China Sea to Japan. Based on observed and simulated RH in eastern China (Table S1 and Fig. 1c and d), soot-containing particles could experience wetter environments with >90 % RH during transboundary transport through the sea pathway compared with the inland pathway. We indeed observed obvious water rims around soot-containing particles transported through the sea pathway compared to the inland pathway (Fig. 8). The presence of water rims indicates that those soot-containing particles were in the aqueous phase prior to being analyzed by TEM under vacuum (Zhang et al., 2023). Laboratory studies have also observed water rims after aqueous-phase particles are dehydrated (Sun et al., 2018). In other words, if particles undergo aqueous-phase processing during transport, and then effloresce under low RH conditions, the water rim will be present as a marker. Additionally, Liu et al. (2018) revealed pronounced aqueous-phase signatures surrounding cloud droplet residuals, as indicated by water rims. AFM measurements further confirmed that the particles transported through the sea pathway exhibited a droplet morphology (Fig. S2b). It is noted that the peak diameter and core number of embedded soot-containing particles largely shifted from 446 nm and 1 to 925 nm and ≥3 during transboundary transport through the sea pathway (Figs. 4c and 6c, d). The evolution implies that simple coagulation or condensation was not the primary aging processes of soot-containing particles in high RH environments, because these mechanisms are insufficient to explain the observed micron-sized particles with multiple cores (Liu et al., 2018). Instead, cloud processing likely played a more important role. Figure S7 shows the satellite image combined with the backward trajectory of haze masses during 7–8 December 2020. We found the presence of clouds over the East China Sea during the transport of haze masses through the sea pathway (Fig. S7). Moreover, the observed phenomenon of two or more soot cores within individual particles transported through the sea pathway aligns with the findings that a single cloud droplet can entrain numerous refractory aerosol particles (e.g., soot) (Ding et al., 2025; Liu et al., 2018). We further noticed that soot-containing particles did not pass areas with high emissions during transboundary transport through the sea pathway compared to those transported through the inland pathway (Fig. 1a and b). These findings suggest that the aging process of soot-containing particles was primarily driven by the meteorological change (i.e., cloud), with minimal contribution from additional industrial and urban emissions along the sea pathway. Therefore, soot-containing particles predominantly underwent cloud process aging during the transboundary transport through the sea pathway, resulting in a significant thickening of coatings on soot cores.

The Df of soot particles serves as a critical metric for assessing their compactness and degree of atmospheric aging processes, providing a quantitative basis for black carbon characterization (Li et al., 2024; Pang et al., 2022). Figure 9 presents the evolution of Df for partly-coated and embedded soot particles during atmospheric transport through the inland and sea pathways. The Df of partly-coated and embedded soot particles increased from 1.81±0.03 and 1.90±0.03 in the NCP to 1.84±0.06 and 1.93±0.04 in the YRD following the transboundary transport through the inland pathway (Fig. 9a and b), suggesting that secondary coatings formed via heterogeneous aging processes enhanced soot compactness during the transport. This varied result of soot Df during the transboundary transport through the inland pathway is similar to that during the dynamic progression of regional heavy haze pollution in winter (Zhang et al., 2023). However, when soot particles were transported to the YRD from the NCP through the sea pathway, their Df increased from 1.81±0.05 for partly-coated soot and 1.89±0.06 for embedded soot to 1.85±0.06 and 2.07±0.03 (Fig. 9c and d). This suggests that the structural collapse of embedded soot particles was more pronounced compared to partly-coated soot particles during the transport through the sea pathway. Moreover, in contrast to the inland pathway, the Df of embedded soot particles transported through the sea pathway showed a 9.5 % greater amplitude (Fig. 9). This comparative result is consistent with the observed differences in the Dp / Dc of soot-containing particles during two distinct atmospheric transport events (Fig. 7), indicating that cloud process aging can greatly promote the structural collapse of soot aggregates. This mechanism can be ascribed to surface tension induced by the hygroscopic growth of secondary coatings on soot under elevated RH, which collapse the soot fractal morphology through water-mediated structural restructuring (Schnitzler et al., 2017). Therefore, cloud process aging of soot-containing particles during the transboundary transport through the sea pathway acted synergistically to (1) facilitate the entrainment of multiple soot cores, (2) substantially enhance their Dp / Dc ratios by ∼50 %, and (3) induce more pronounced collapse of soot fractal structures with Df from 1.89±0.06 to 2.07±0.03.

Based on mixing structures of soot particles during the transboundary transport, the light absorption enhancement (Eabs) of soot-containing particles with 1–3 cores and different mixing structures (partly-coated and embedded configurations) was calculated by the DDA combined with the EMBS. Considering that embedded soot cores were often distributed at the periphery rather than the center within individual particles (Fig. 8), we conducted optical simulations of embedded soot-containing particles based on this realistic mixing structure. In the optical calculation, the diameters of single soot cores and coatings were presumed to 194 and 925 nm according to their size distribution (Figs. S8 and 6d), and the total volume of soot cores in individual constructed particles was constant when their numbers were changed. Moreover, the Df of soot cores was presumed to 2.07 and their coatings were assumed to be non-absorbing materials in the optical calculation.

Figure 10a shows the change in the Eabs of soot-containing particles following their aging from bare-like to partly-coated, and then to embedded states. Compared to soot cores, partly-coated and embedded soot-containing particles present higher Eabs (Fig. 10a), due to the lensing effect under conditions of thick coating and compact soot morphology (Fierce et al., 2020; Wang et al., 2021, 2025). The Eabs of soot-containing particles with one soot core increased from 1.80 for the partly-coated structure to 2.83 for the embedded structure (Fig. 10a). When soot-containing particles had two soot cores, the Eabs increased from 1.74 to 2.44, representing a 0.4-fold increase, with soot aging from partly-coated to embedded configurations (Fig. 10a). Following the soot core number increase to three, the Eabs of soot-containing particles increased by 117 % (from 1.04 to 2.26) when their mixing structures changed from partly-coated to embedded status (Fig. 10a). These results suggest that individual particles containing higher numbers of soot cores demonstrate larger optical absorption amplification during atmospheric aging processes although their Eabs values were lower. Previous studies also found lower absorption efficiency in cloud drops with higher numbers of soot cores (Jacobson, 2006) and smaller Eabs in simulated ambient particles with larger number density of soot cores (Fierce et al., 2016). In addition, comparable radiative absorption changes for soot-containing particles with different numbers of soot cores were observed during the transformation of soot core positions (Zhang et al., 2022).

Based on the percentage, Dp / Dc, and Eabs of soot-containing particles with different mixing structures and core numbers, we can compare the change in radiative absorption capacity per unit the change in coating thicknesses of soot-containing particles during the transboundary transport through the inland and the sea pathways. Figure 10b shows ΔEabs/Δ(Dp / Dc) of transboundary soot-containing particles transported through the inland and sea pathways. When soot-containing particles were transported from the NCP to the YRD through the inland pathway, their ΔEabs/Δ(Dp / Dc) reached 0.6 (Fig. 10b). However, the ΔEabs/Δ(Dp / Dc) of soot-containing particles was only 0.17 following their transboundary transport through the sea pathway (Fig. 10b). Previous studies have revealed that the Eabs of soot-containing particles first increases and then tends to stabilize with their coating thickness (e.g., Dp / Dc) increases (Beeler et al., 2024; Fu et al., 2022). We found that the mean Dp / Dc of embedded soot-containing particles exhibited a large value at 4.38 when haze masses were transported through the sea pathway (Fig. 7b). In addition, cloud processes induced multiple soot cores within single particles during the transboundary transport through the sea pathway in contrast to the inland pathway, reducing their optical absorption (Fig. 10a). Beeler et al. (2024) also found consistent results that much lower Eabs variation for soot-containing particles with the thickening of coatings in pyrocumulonimbus clouds compared with urban air. Therefore, the larger Dp / Dc change and the smaller Eabs change of soot-containing particles transported through the sea pathway should result in the lower ΔEabs/Δ(Dp / Dc) compared to those transported through the inland pathway. These findings further suggest that the radiative absorption amplification per unit Dp / Dc change of transboundary soot-containing particles reduced by 72 % with the change in their transport pathways from inland to sea. A sensitivity test was conducted for the ΔEabs/Δ(Dp / Dc) reduction caused by the transport pathway change through varying the embedded soot core position in host particles, as shown in Fig. S9. It was found that the ΔEabs/Δ(Dp / Dc) reduction is 68 % when soot cores are randomly distributed in host particles (Fig. S9). This is close to the 72 % reduction calculated when soot cores are distributed at the periphery of host particles (Fig. S9). Because over 80 % of the embedded soot cores were observed to be distributed at the periphery of transboundary particles, and the remainder was primarily randomly distributed, the 72 % ΔEabs/Δ(Dp / Dc) reduction can be considered reliable. If embedded types in soot-containing particles were presumed as the traditional core-shell model, the ΔEabs/Δ(Dp / Dc) of transboundary soot-containing particles was extremely low at 0.01–0.03 (Table S3). This result shows a large difference from the optical absorption simulated with real mixing structures of soot-containing particles. Therefore, the traditional core-shell assumption significantly underestimates the optical absorption of transboundary soot-containing particles, which is consistent with findings in a wildfire smoke episode (Chen et al., 2025). It should be noted that the ΔEabs/Δ(Dp / Dc) derived from optical simulation is a conceptual metric, which depends on the chosen definition of Dp and Dc (e.g., ESD). While model–model comparisons are likely robust, model–measurement comparisons would likely require careful matching of definitions. In summary, the atmospheric humidity condition during the transport of soot particles not only affects their aging processes but also influences their radiative absorption (Fig. 11). In view of that soot particles can be exposed to high-humidity or supersaturated atmospheric environments during transboundary transport, climate models should incorporate multicore soot-containing particles to refine current simulations of climate effects.