A continuous field observation on aerosol optical, hygroscopic, and chemical properties was carried out from 29 July to 19 August 2022. A detailed description of the observation site is available in Supplement S1. During the observation period, urban Chongqing suffered a rare heatwave (Fig. S1; Chen et al., 2024; Wang et al., 2024), which significantly affected local transportation and industrial activities (Hao et al., 2023). The China Meteorological Administration (CMA) defines heatwaves as 3 or more consecutive days with daily maximum temperature (Tmax) above 35 °C (http://www.cmastd.cn/standardView.jspx?id=2103 (last access: 7 October 2025); Guo et al., 2016; Sun et al., 2014; Tan et al., 2007). Since there is no unified definition of heatwaves worldwide, the whole study period was categorized into two stages according to CMA's criteria for daily Tmax records and the excess heat factor (EHF) metric proposed by Nairn and Fawcett (2014) (Fig. S2a): (1) the normally hot period from 29 July to 6 August (marked as P1) and (2) the heatwave-dominated period from 7–19 August (marked as P2), characterized by the consistent occurrence of Tmax exceeding 38 °C (approximately the lowest 25th percentile of temperature records for the whole observation period; Fig. S2b).

Figure 1 displays the time series of the measured aerosol scattering coefficients, f(RH), PNSD, and the corresponding meteorological conditions and air pollutants during the study period. A sharp decrease in aerosol scattering coefficients and PM_2.5, accompanied by continuous excellent visibility over 20 km, was observed after 6 August, indicating a markedly cleaner environment during P2 in comparison to P1 in the summer of 2022 of Chongqing. This can largely be attributed to the reduction in anthropogenic emissions (e.g., NO₂ and CO but not SO₂) from limited outdoor activities influenced by the heatwaves in P2, as well as from the partial suspension of industries and transportation to alleviate the power shortage issue (Chen et al., 2024). Notably, the increased wind speed and enhanced mixing layer height (MLH) also enabled a more favorable atmospheric diffusion condition in P2, facilitating the dilution of surface air pollutants (Zhang et al., 2008). However, a higher mass concentration of SO₂ was observed in the P2 period, likely due to a surge in electricity demand and resulting higher emissions from power plants operating at almost full capacity during the heatwave (Su, 2021; Teng et al., 2022). Moreover, significant discrepancies in aerosol optical and hygroscopic properties were observed under different synoptic conditions (Table S2). Both HBF and SAE were higher during the P2 period, aligning with the smaller Reff (Table S2). f(RH) was found to be relatively higher (p < 0.05) on heatwave days, with mean values of 1.61 ± 0.12 and 1.71 ± 0.15 during the P1 and P2 periods, respectively. In contrast, ALWC was more abundant during the normally hot P1 period than the heatwave-dominated P2 period. This is likely due to the fact that the derivation algorithm of ALWC utilized in this study (Kuang et al., 2018) is partly dependent on (e.g., positively correlated to) the dry aerosol scattering coefficient or, rather, the aerosol volume concentration in the dry condition (refer to Sect. S3 and Fig. S11 of the Supplement). The mean σsca,525 for P2 was about 46.8 % of that for the P1 period, and the corresponding mean level of ALWC was approximately 55.8 % of that for P1. This partly agrees with the stronger aerosol optical hygroscopicity, with a marginally higher fW during the P2 period, highlighting a complex interaction between the optical enhancement and aerosol physicochemical properties.

The particle number size distribution data suggested that NPF events occurred on about half of the observation days (Fig. 1i), with an overall occurrence frequency of 52.4 % (Fig. S4a). This indicates that summer NPF events are rather frequent in Chongqing, being notably more common than those observed in other regions of the world, e.g., Beijing (16.7 %, Deng et al., 2020; ∼ 20 %, Wang et al., 2013), Dongguan (4 %, Tao et al., 2023), Hyytiälä (<40 %, Dada et al., 2017), and LiLLE (<20 %, Crumeyrolle et al., 2023). Moreover, the frequent NPF events during heatwaves formed substantially ultrafine particles that contributed less to aerosol optical properties in comparison to large particles (Fig. S13), partially explaining the significantly lower levels of total scattering coefficients observed during the P2 period. It should be noted that the hourly σsca,525 values during the P2 period were exclusively below 100 Mm⁻¹ (approximately the lowest 10th percentile of σsca,525 data, regarded as the threshold value of relatively polluted cases; Fig. S2c), suggesting a much cleaner environment compared to the relatively polluted P1 period. Correspondingly, NPF events occurring during the relatively polluted P1 period (as detailed in Sect. 3.2) are defined as NPF_polluted, while cases during the cleaner and heatwave-dominated P2 period are classified as NPF_clean, HW.

Aside from gaseous precursors (e.g., SO₂, volatile organic compounds), meteorological conditions also play a key role in the occurrence of NPF events. In brief, NPF events are more likely to appear under sunny and clean conditions (Bousiotis et al., 2021; Crumeyrolle et al., 2023; Deng et al., 2021; Wang et al., 2017). The backward trajectory analysis revealed that a southerly breeze was predominant during the study period (Fig. S4b). Although the surface wind vector slightly varied between the P1 and P2 periods, this consistency in air mass origins suggests that some other factors (e.g., changes in environmental conditions and emissions of gaseous precursors under heatwaves) could have played a crucial role in modulating NPF events. To further explore the characteristics of NPF events in different periods, the time-averaged diurnal variations in meteorological parameters and air pollutant concentrations during both NPF events and non-event days are presented in Fig. 2.

As stated in Sect. 3.1, NPF events during the P1 period tended to occur in relatively polluted environments compared to those of P2 NPF_clean, HW events, as evidenced by the frequent occurrence of σsca,525>100 Mm⁻¹, increased air pollutant concentrations, and lower visibility levels during P1 (Table S2, Fig. 1). Additionally, the mean CS of the NPF_polluted events was above 0.015 s⁻¹ (Table S2), which could be considered “polluted” NPF cases (Shang et al., 2023). On P2 NPF_clean, HW days, the overall mean σsca,525 was 33.2±11.7 Mm⁻¹, decreased by 68.0 % (39.3 %) in comparison to that for P1 NPF_polluted days (P2 non-event days). In addition, the mean PM_2.5 concentration was even lower than 10.0 µg m⁻³, and the corresponding visibility level was almost maintained at 30 km (Fig. 1e). All the above implies that the P2 NPF_clean, HW events were generally accompanied by a much cleaner environment. It is notable that the increase in SO₂ concentration after 09:00 LT (Fig. 2d), along with the significant decrease in PM_2.5 mass loadings after 08:00 LT during P1 NPF_polluted events (Fig. 2b), likely favored the occurrence of NPF events. The higher gas-phase sulfuric acid (i.e., H₂SO₄, as estimated with the UVB and SO₂ concentration; Lu et al., 2019; Sect. S4) on the same NPF days (Fig. 2f) further suggests that sulfuric acid concentration was a critical factor for the occurrence of P1 NPF_polluted events.

The diurnal evolutions of meteorological conditions (e.g., T, RH, MLH) for NPF events were distinct between the P1 and P2 periods, although relatively insignificant differences were observed for both NPF days and non-event days within the same period (Fig. 2). This likely suggests that meteorological factors might not be the predominant determining factor of NPF occurrence during the heatwaves of the summer of 2022 in urban Chongqing, while NPF could be accompanied by quite different meteorological conditions depending on gaseous precursors and pre-existing condensation sinks. For instance, the NPF_clean, HW events were typically clean-type NPF, characterized by lower background aerosol loading, higher temperature, and favorable atmospheric dispersion capacity with the higher MLH. However, it is reported that excessive heat can increase the evaporation rate of critical acid-base clusters during the nucleation process and reduce the stability of initial molecular clusters (Bousiotis et al., 2021; Kurtén et al., 2007; Zhang et al., 2012), in line with a recent study reporting that NPF events were weaker during heatwaves in a Siberian boreal forest due to the unstable clusters (Garmash et al., 2024). On the other hand, the emission rate of biogenic VOCs (BVOC_S, e.g., isoprene, monoterpene) from nearby plants and trees would decrease when temperature exceeded around 40 °C (Guenther et al., 1991; Pierce and Waldruff, 1991), despite the fact that BVOCs play a key role in the nucleation mechanism of NPF (Wang et al., 2017; Zhang et al., 2004). Hence, the even higher temperature (e.g., T>40°) likely suppressed the nucleation processes and the subsequent growth of nucleation mode particles on P2 non-event days (Fig. S6b2), in spite of higher concentrations of SO₂ and H₂SO₄.

To further investigate the effect of heatwaves on NPF events, the diurnal variations in PNSD, Reff, and particle mode diameter (Dmode) are shown in Fig. S6. Aerosol number and volume concentrations, as well as Reff, for different modes are illustrated in Figs. S7–S8, and the relationship between temperature and the duration of NPF events is displayed in Fig. S9. Distinct particle size distributions were observed for different NPF event days. While the number concentrations of Aitken mode particles (NAit.) were comparable during NPF days of both periods, the corresponding number concentration of nucleation mode particles (NNuc.) was significantly higher on P1 NPF_polluted days (1880.8 ± 2261.5 cm⁻³) than on P2 NPF days (1132.0 ± 1333.5 cm⁻³) (Figs. 1i, S7). Different from that of the P1 NPF_polluted cases, the P2 NPF_clean, HW event did not start from the minimum size, and the reduced NNuc. during the P2 period is likely attributable to the influence of transport on the local nucleation and growth processes (Fig. S4; Cai et al., 2018; Lee et al., 2019). Namely, some nucleation mode particles transported from upwind regions or from the mixing layer downwards had undergone atmospheric aging and thereby a certain degree of growth upon arrival (Cai et al., 2018; Lai et al., 2022; Platis et al., 2016), resulting in relatively lower concentrations of smaller-sized particles than in the case of locally formed particles. However, the local formation of sub-25 nm particles and the continuous growth process were still distinctly observed under heatwaves (Figs. 1i, S6, S15). The NPF events under heatwaves usually initiated earlier (Fig. S9), with NNuc. in P2 NPF_clean, HW cases peaking about an hour earlier in comparison to NPF_polluted days (Fig. S8a). Dmode on P2 NPF_clean, HW days also reached its minimum earlier than on P1 NPF_polluted days (Fig. S6). Since the sunrise and sunset times did not significantly vary within the study period (i.e., less than a half-hour discrepancy), heatwaves likely provided more favorable conditions (e.g., enhanced volatile gaseous emissions, low RH; Bousiotis et al., 2021; Hamed et al., 2007; Wang et al., 2024) for the occurrence of NPF events in urban Chongqing. This is supported by the earlier start time of NPF_clean, HW, corresponding to higher temperature ranges (Fig. S9). Furthermore, the end time of subsequent particle growth during the P2 period was even later (i.e., ∼ 21:00 LT) than that of P1 cases (Fig. S9). Given that the growth rates of new particles were generally lower during P2 NPF_clean, HW events (Table S2), these explosively formed new particles could persist longer in the warmer atmosphere and probably underwent aging processes with a relatively higher oxidation degree. This is supported by the commonly higher ratios of secondary organic carbon (SOC) to organic carbon (OC) (i.e., SOC / OC > 0.5) during the NPF_clean, HW days (Fig. S3b). In addition, aerosol Reff was significantly smaller on the NPF_clean, HW days under heatwave conditions. Reff and Dmode were maintained at nearly the same level below/approaching 50 nm during the subsequent growth on the P2 NPF_clean, HW days, while Reff was generally above 50 nm and larger than Dmode for both P1 NPF_polluted cases and non-event days (Fig. S6). The diurnal patterns of aerosol volume concentrations for different size modes were similar to those of aerosol number concentrations during NPF events (Fig. S8b1–b3). However, Reff of both Aitken mode particles (RAit.) and accumulation mode particles (RAcc.) was smaller during P2 NPF_clean, HW events than during P1 NPF_polluted events (Fig. S8c2–c3), which may further influence size-dependent aerosol optical and hygroscopic properties (e.g., σsca,525, HBF, SAE, f(RH)). The decrease in RAit. and RAcc. during heatwaves can be attributed to three factors: (1) evaporation of the outer layer of particles and unstable clusters due to heatwaves (Bousiotis et al., 2021; Cusack et al., 2013; Deng et al., 2020; Garmash et al., 2024; Li et al., 2019), (2) lower FR and GR of particles under cleaner-environment conditions (Table S2), and (3) reduced emissions of larger primary particles during the P2 period.

Diurnal variations in the aerosol optical and hygroscopic parameters during different NPF days are shown in Fig. 3, and the corresponding results for non-event days are shown in Fig. S10. Generally, σsca,525 possessed a similar bimodal diurnal pattern to that of the accumulation mode aerosol volume concentration (VAcc.) (Fig. S8b3), as supported by the positive correlation between σsca,525 and SMPS-measured aerosol volume concentration (Fig. S12). This is also consistent with Mie theory, with a stronger increase in the scattering efficiency for accumulation mode particles (Titos et al., 2021). The diurnal pattern of σsca,525 also varied distinctly between different NPF days. Specifically, a minor peak in σsca,525 around 12:00 LT (Fig. 3a) was influenced by the newly formed particles during P2 NPF_clean, HW events, which contributed more significantly to the aerosol number and volume concentrations within 100 nm size ranges in markedly clean environments (Fig. S5c1, c2). Instead of a noontime peak, σsca,525 was observed with an early peak around the morning rush hours, and a maximum value similarly occurred at nighttime on P1 NPF_polluted days.

Both HBF and SAE on P2 NPF_clean, HW days were significantly higher than during P1 NPF_polluted cases (Fig. 3c, e), largely due to the smaller Reff observed during the heatwave-dominated period (Table S2). Moreover, the correlation between HBF (or SAE) and particle size in each mode was weaker on NPF days than on non-event days, especially for NPF_clean, HW days (Fig. S14). The strongest negative correlation was found between HBF and Reff of the accumulation mode in comparison to other modes, highlighting that HBF is more sensitive to the size distribution of accumulation mode particles (Collaud Coen et al., 2007). Given that NPF would largely enhance the abundance of both nucleation and Aitken mode aerosols (Fig. S7), no significant variation in HBF was observed during the daytime due to the weakened correlation between HBF and RAcc. of NPF events. SAE is commonly used as an indicator of particle size distribution, almost decreasing monotonously with the increase in aerosol size within 1 µm (Kuang et al., 2017, 2018; Luoma et al., 2019). Accordingly, SAE decreased over the morning and evening rush hours, when coarse particles (e.g., aged particles, road dust, automobile exhaust) generated during anthropogenic activities were present, accompanied by an increase in CO, which is taken as a proxy for primary emissions (Fig. 2l) (Yarragunta et al., 2021). In contrast, the abundant ultrafine particles formed during NPF events led to a continuous increase in SAE during the day.

f(RH) exhibited a similar diurnal pattern on P1 and P2 NPF days (Fig. 3b). During the daytime, f(RH) remained relatively stable and gradually increased until peaking around 16:00–18:00 LT, with a generally higher f(RH) particularly after 15:00 LT during P2 NPF_clean, HW days than during P1 cases. The insignificant fluctuations in relatively lower f(RH) levels before noon could be attributed to the continuous development of the mixing layer (Fig. 2k), leading to efficient mixing of particles in the nocturnal residual layer with anthropogenic emissions near the ground. Additionally, photochemical reactions in the afternoon facilitated the formation of more hygroscopic secondary aerosols with a higher oxidation level (Liu et al., 2014; R. Zhang et al., 2015). The diurnal patterns of O₃ and the O3/ O_X ratio (i.e., an indicator of atmospheric oxidation capacity, where OX= O3+ NO₂, Tian et al., 2021) also showed similar trends (Fig. 2g, h). The presence of black carbon (BC) mixed with organic compounds (e.g., from traffic emissions and residential cooking activities) explained the rapid decrease in f(RH) during the evening rush hours (Liu et al., 2011). Furthermore, the daily mean f(RH) for NPF days was higher than that of non-event days (Table S2), particularly after the end of NPF events around 12:00 LT. Given that newly formed particles were too small to significantly impact the total light scattering (Fig. S11a), this indicates that the atmospheric conditions conducive to the occurrence of NPF may promote further growth (e.g., via intensified/prolonged photooxidation or atmospheric aging processes) of pre-existing particles and newly formed ones, partly contributing to enhanced aerosol optical hygroscopicity, as indicated by the concurrent variations in ALWC and fW in urban Chongqing during the hot summer (Asmi et al., 2010; Wang et al., 2019; Wu et al., 2016). The diurnal pattern of ALWC closely mirrored the variation in σsca,525, while fW followed a similar evolution to f(RH). This suggests that ALWC was more sensitive to changes in the aerosol volume concentration, as determined by the corresponding retrieval algorithm (Kuang et al., 2018). fW levels were slightly higher during NPF days in comparison to those on non-event days (Table S2). This difference was more pronounced in the afternoon of NPF days (e.g., even exceeding 50 %; Fig. 3f), verifying the enhancement of aerosol hygroscopicity during the subsequent growth and atmospheric aging of both pre-existing and newly formed particles.

To further explore the impacts of heatwaves on f(RH) during diverse NPF events, data mainly within the time window of 08:00–22:00 LT (i.e., typically covering the complete process of NPF and subsequent growth, while excluding higher-RH conditions at night) were utilized for the following discussion.

Although ultrafine particles exhibited higher number concentrations during the study period, accumulation mode particles dominated the aerosol volume concentration and contributed predominantly to the total light scattering (Figs. S7, S13). A positive correlation between f(RH), Reff, and the volume fraction of accumulation mode particles (VF_Acc.) was found on non-event days (Fig. 4c–d), when the aerosol size distribution was undisturbed by newly formed ultrafine particles and the corresponding VF_Acc. was maintained at a high level of approximately 0.95 (Fig. 4a–b). The notably positive correlation between f(RH) and Reff could be linked to the secondary formation of hygroscopic particles within the accumulation mode, primarily via photochemical reactions, and further intensified by heatwaves during the non-event day, particularly in the P2 period (Gu et al., 2023; Liu et al., 2014; R. Zhang et al., 2015; Zhang et al., 2024). Consequently, f(RH) at a specific Reff was generally higher during the P2 period in comparison to that of P1 (Fig. 4c–d), also with high f(RH) levels observed for smaller-size cases of Reff < 110 nm under some extremely high temperature conditions (T>40°, as highlighted by the dashed red circle in Fig. 4d). The higher SOC / OC on P2 non-event days further demonstrated the stronger secondary aerosol formation in comparison to P1 non-event days (Fig. S3b).

Nevertheless, f(RH) was almost independent of the two parameters (i.e., Reff and VF_Acc.) during NPF events (Fig. 4e–f). This is mainly due to the explosive formation of ultrafine particles and their subsequent growth on NPF days, significantly altering aerosol size distributions and inducing large fluctuations in NF_Acc. and VF_Acc. compared with non-event days, especially during the P2 period (as shaded in Fig. 4a–b). Therefore, characterizing f(RH) by the corresponding Reff of aerosol populations was no longer applicable. Alternatively, SAE was commonly used to estimate or parameterize f(RH) (Titos et al., 2014; Xia et al., 2023; Xue et al., 2022), in line with the similar diurnal patterns of f(RH) and SAE observed in this study. Figure 5 demonstrates a significantly positive correlation between f(RH) and SAE during NPF days in comparison to non-event days, with a similar slope of approximately 0.65, suggesting the consistent variation in f(RH) with SAE across both periods. As larger particles contributed more to aerosol volume concentrations (Fig. S5), the decrease in SAE also corresponded to an increase in σsca,525 (Fig. 5a3, b3). Given that larger σsca,525 values typically indicate the condition of higher aerosol loading, f(RH) increased with SAE, whereas it decreased with σsca,525 or rather, the pollution level, during NPF days. The cleaner environment of the P2 period may further favor the occurrence of NPF events. Both f(RH) and SAE exhibited a higher level on P2 NPF_clean, HW days (as shown by the dashed lines in Fig. 5), probably due to the following two aspects. One is related to the smaller aerosol Reff (with a larger SAE) due to the lower FR and GR, likely influenced by the evaporation of newly formed unstable clusters and particle coatings under heatwaves (Bousiotis et al., 2021; Cusack et al., 2013; Deng et al., 2020) during the subsequent growth of aerosols. Secondly, the higher temperature was normally associated with stronger photochemical oxidation, which could intensify the formation of secondary aerosol components with a higher hygroscopicity (Asmi et al., 2010; Gu et al., 2023; Liu et al., 2014; Wu et al., 2016; R. Zhang et al., 2015; Zhang et al., 2024). This is further supported by the slightly higher levels of UVB (P1: 2.6 ± 1.9 W m⁻² versus P2: 2.7 ± 2.0 W m⁻²) and O3/ O_X (P1: 0.81 ± 0.17 versus P2: 0.82 ± 0.17) during P2 heatwave days, also in line with a recent study which demonstrated that heatwaves affected secondary organic aerosol (SOA) formation and aging by accelerating photooxidation in Beijing (Zhang et al., 2024).

It is worth noting that f(RH) did not show a consistently higher level after the NPF_clean, HW occurrence during the P2 period, and it was slightly higher within the first few hours of NPF occurrence (i.e., ∼ 12:00–15:00 LT) on P1 NPF_polluted days (Fig. 3b). In fact, aerosol optical hygroscopicity does not fully correspond to the bulk hygroscopicity, primarily determined by aerosol chemical components, and the variability in aerosol optical features also plays a key role in f(RH). Hence, the size dependency of aerosol optical properties should be considered. The size-resolved σsca,525 distribution and size-resolved cumulative frequency distribution (CFD) of σsca,525 over different NPF days were calculated using Mie theory, with good agreement between the theoretically calculated and measured σsca,525 values (R2=0.99). As shown in Figs. S11a and S13, new particles must grow into the accumulation mode size at least before they can exert a significant influence on the total scattering coefficient. The critical sizes corresponding to the cumulative frequency of 50 % in σsca,525 were 358.7 and 333.8 nm on P1 and P2 NPF days, respectively. This indicates that relatively smaller particles – including the newly formed and grown particles mixed with pre-existing and aged particles – contributed a slightly higher portion to σsca,525 on P2 NPF_clean, HW days, while σsca,525 was mainly contributed by larger particles on P1 NPF_polluted days. Nevertheless, Mie theory suggests that these smaller particles generally have a weaker enhancement in total scattering after hygroscopic growth in comparison to larger size particles (Collaud Coen et al., 2007, Fig. S11a). Consequently, the changes in aerosol optical and hygroscopic properties necessitate consideration of both aerosol optical and chemical characteristics during different NPF events. Newly formed ultrafine particles contributed less to aerosol optical properties, resulting in lower f(RH) during the initial hours of P2 NPF_clean, HW events compared to P1 NPF_polluted events (Fig. 3b), as evidenced by smaller Reff for P2 NPF_clean, HW events (Fig. S6). In contrast, the growth of pre-existing and newly formed particles into larger sizes would subsequently affect bulk aerosol optical properties, which was evidenced by the enhancement in the aerosol extinction coefficient observed after NPF occurrence in a recent study (Sun et al., 2024). Specifically, particles could undergo a longer and more intensified photochemical aging process during NPF_clean, HW days, as influenced by persistent heatwaves, which facilitated the secondary formation of hygroscopic aerosols and jointly contributed to higher f(RH) after 15:00 LT (Fig. 3b).

The changes in f(RH) have significant implications for aerosol direct radiative forcing. Despite considerably lower σsca,525 results during heatwaves, the corresponding mean fRF(RH) levels, particularly for P2 NPF_clean, HW days, were higher than those of the P1 cases (Fig. 6a). A robust positive correlation (R2=0.68) was observed between f(RH) and the aerosol radiative forcing enhancement factor, fRF(RH) (Fig. 6b). This is likely due to the enhanced fRF(RH) with the larger forward scattering ratio β, or rather higher HBF for smaller particle sizes, as supported by a generally negative correlation between fRF(RH) and Reff. Specifically, the highest fRF(RH) value of 2.21 ± 0.23 was observed on P2 NPF_clean, HW days, characterized by the highest f(RH) and smallest Reff (i.e., highest HBF) of the entire study period.

The definition of fRF(RH) in Eq. (5) implies its dependence on both f(RH) and HBF-derived β(RH) and β(dry), or, more precisely, the ratio of HBF_525, RH / HBF₅₂₅. The mean HBF_525, RH was generally larger than HBF₅₂₅ in this study, specifically with the HBF_525, RH / HBF₅₂₅ ratios centered around 1.8, and even approached 2.5 on P2 NPF_clean, HW days (Fig. 6c, Table S2). This could differ from classical Mie theory with its spherical-particle premise; i.e., the observed light backscattering was enhanced after hydration, likely as a result of the evolution in particle morphology that significantly influences its optical properties (Mishchenko, 2009). Additionally, the predominant organic components, when heterogeneously mixed with diverse chemical compositions (e.g., inorganics and black carbon), likely introduced the heterogeneity in aerosol hygroscopicity (Yuan and Zhao, 2023), which may alter particle morphology and thereby optical properties upon water uptake (Giordano et al., 2015; Tan et al., 2020; Tritscher et al., 2011). The efficient evaporation of organic coatings under extremely hot conditions could also contribute to the change in particle morphology (e.g., non-spherical irregular shapes) upon humidification, as evidenced by a recent study that reported that high-temperature conditions could accelerate the evaporation rate of SOA (Li et al., 2019). Given that the backward scattering intensity of non-spherical particles is suggested to be much larger than that of their spherical counterparts at scattering angles between 90 and 150° (Mishchenko, 2009; Yang et al., 2007) and that the HBF-derived asymmetry parameter (g) normally correlates positively with the aerosol forward scattering (Andrews et al., 2006; Marshall et al., 1995), the generally smaller gRH results (in comparison to g) confirmed the decrease (increase) in forward (backward) light scattering after water uptake (Fig. S11c), likely implying a change in the morphological structure of particles. This is particularly evident for P2 NPF_clean, HW days, during which a much lower level of gRH was observed (Fig. S11c). Another possible reason is the distinct size dependences of both light scattering and backscattering efficiencies (Fig. S11a), with much more significant enhancements in the backscattering efficiency and thereby HBF, specifically of accumulation mode particles after hygroscopic growth (Fig. S11b). As reflected by the Mie model, although the abundant newly formed particles were generally optically insensitive (e.g., below 100 nm), their contributions to σsca,525 and especially to σbsca,525 could be amplified upon humidification (Fig. S11b). Moreover, the shift of the size distribution towards larger accumulation mode particles could also result in a significant elevation in HBF_525, RH / HBF₅₂₅ ratios, especially under the condition of a smaller-mode diameter and narrower distribution of ultrafine-mode particles (e.g., during NPF events) (Fig. S16a1–b2 for the theoretical sensitivity tests of Sect. S9 in the Supplement). Furthermore, the HBF_525, RH / HBF₅₂₅ ratio exhibited a significant positive correlation with the real part of the complex refractive index (n) of bulk aerosols (Fig. S17), and n tends to increase with the aging process of organic species (Moise et al., 2015; G. Zhao et al., 2021). In this sense, the evolution of both aerosol size distribution patterns and chemical compositions, combined with the heterogeneity in aerosol hygroscopicity, could potentially change particle morphology and optical properties (e.g., complex refractive index and elevated HBF_525, RH), particularly during heatwave-influenced NPF_clean, HW days, characterized by the smallest aerosol Reff (102.8 ± 12.4 nm) (Fig. S6), lowest number concentration (1897.0 ± 680.8 cm⁻³) and fraction (0.20 ± 0.10) of accumulation mode particles, intensified photooxidation, and a higher SOC / OC ratio. The higher HBF_525, RH / HBF₅₂₅ ratios increased the HBF-derived β(RH)/β(dry) levels, in combination with the elevated f(RH), further resulting in the highest fRF(RH) observed during P2 NPF_clean, HW events. Given that previously observed HBF_525, RH was typically lower than HBF₅₂₅ (Titos et al., 2021; Xia et al., 2023; L. Zhang et al., 2015), the mean fRF(RH) results of this study (fRF(85 %) = 2.05 ± 0.24) were significantly higher than those observed in the Yangtze River Delta (fRF(85 %) = 1.5, L. Zhang et al., 2015), the North China Plain (fRF(80 %) = 1.6 ± 0.2, Xia et al., 2023), and some other regions in the world (Titos et al., 2021, Fig. 6d). It should be noted that the reported fRF(RH) for the UGR site (Spain) was even higher, likely due to the higher Rs and αs used in the derivation of fRF(RH) in that area (Titos et al., 2021).

A recent study has indicated that continuous reduction in PM_2.5 mass loadings can increase the net solar radiation, thereby promoting NPF events (S. Zhao et al., 2021). Given the complexity and dynamic evolution of the atmospheric environment, these can further alter the intrinsic properties of aerosol particles (e.g., f(RH), HBF, morphology), potentially feeding back into aerosol–radiation interactions. Our findings suggest that NPF days may possess a relatively higher aerosol optical hygroscopicity in rather hot environments, e.g., the basin area and tropical regions. Meanwhile, NPF serves as a crucial secondary transformation process in the atmosphere (Zhu et al., 2021). The favorable atmospheric diffusion capability ensured the mixing of newly formed particles into the upper boundary layer, where it is colder and more humid compared to near the surface during heatwaves (Jin et al., 2022). Hence, the enhancement of aerosol optical hygroscopicity during the subsequent growth and aging of both pre-existing and newly formed particles possibly exacerbates secondary pollution and even triggers haze events (Hao et al., 2024; Kulmala et al., 2021). On the other hand, a large number of studies have demonstrated that the new particles with higher hygroscopicity could contribute more to the activation of CCN (Ma et al., 2016; Ren et al., 2021; Rosati et al., 2022; Sun et al., 2024; Wu et al., 2015), thereby modulating aerosol–cloud interactions and further the global climate (Fan et al., 2016; Merikanto et al., 2009; Westervelt et al., 2013). Additionally, the simultaneous decrease in aerosol effective radius and possibly evaporation-induced non-spherical particle morphology further enhance the aerosol direct radiative forcing enhancement factor, potentially amplifying the cooling effect mainly caused by light scattering aerosols. This highlights the need for further in-depth exploration of aerosol radiative impacts under heatwaves with a changing climate, given the continuous reductions in anthropogenic emissions and more intense emissions of biogenic origins with global warming. Moreover, more detailed information on the evolution of particle morphology with the changing environment (e.g., varied temperature and RH) would enrich insights into aerosol radiative forcing.