Non‐Fossil Origin Explains the Large Seasonal Variation of Highly Processed Organic Aerosol in the Northeastern Tibetan Plateau (3,200 m a.s.l.)

Carbonaceous aerosol plays an important role in climate, but its sources and atmospheric processes are least understood in the Tibetan Plateau (TP), a remote yet climatically sensitive region. This study presents the first seasonal cycle of radiocarbon and stable isotope 13C of organic and elemental carbon (OC and EC) in the atmosphere of the northeastern TP. Large seasonal variations of EC and OC concentrations were explained by non‐fossil sources. Regardless of the season, fossil contribution to OC was strongly correlated with inverse OC concentrations. This allowed the separating a constant background source and a source responsible for OC variability that was mostly of non‐fossil origin. The 13C signature of OC shows that OC was highly atmospherically processed and thus less volatile than OC found near sources or in urban areas. The 13C‐depleted secondary sources contributed strongly to more volatile OC, whereas the 13C‐enriched less volatile OC suggests the influence of atmospheric aging.

. The ability to separate coal combustion from other fossil fuel sources for EC makes 13 C analysis particularly valuable in China, where residential coal combustion is an important pollution source. Furthermore, the 13 C of organic aerosols is modified by atmospheric processes via kinetic isotope effects. Usually, secondary OC (SOC) formation from the oxidation of VOCs precursors is expected to result in a depleted 13 C signature, whereas atmospheric aging releases isotopically lighter carbon (e.g., CO 2 , CO) leading to a gradual 13 C enrichment in the remaining aerosol. Analysis of 13 C in OC thus can give new insight into the formation and transformation processing of organic aerosols (Aggarwal & Kawamura, 2009;Li et al., 2018;Wang et al., 2010). This has inspired 13 C studies on sub-fractions of organic aerosols, for example, more and less volatile OC, that show indications of various atmospheric processes (Masalaite et al., 2017(Masalaite et al., , 2020Ni et al., 2022). While CAs are widely characterized in urban and rural China, sources and formation mechanisms in the pristine and climatically sensitive region of Tibetan Plateau (TP) remain less understood, especially in studies applying dual-carbon isotopic analysis.
The TP (27°-45°N, 70°-105°E) stands 5 km high over a region of ∼3 million km 2 , and is one of the most remote and pristine regions in the world with little industry and sparse population. The high-altitude TP is sensitive to climate change, for example, large areas of glaciers in the TP have retreated in recent decades; the magnitude of climate warming over the TP is greater than the Northern Hemisphere and the globe (Kuang & Jiao, 2016). Despite CAs' important role in climate, their sources and atmospheric processes are least understood in the TP. To date, observation-based isotopic studies on sources and formation of CAs (particularly organic aerosols) in the TP are scarce and rarely extend over more than one season (Li, Bosch, et al., 2022;. In this study, we present a 1-yr study combining radiocarbon and stable carbon isotope analysis of OC and EC in the vastly understudied northeastern TP (Qinghai Lake), aiming to elucidate the seasonal fossil and non-fossil sources and atmospheric processes of organic aerosols.

Sampling
Sampling was conducted at Qinghai Lake (37.04°N, 99.74°E, 3,200 m a.s.l.), a remote site in the northeastern TP. Although it is a remote site, there are two national highways (13-55 km away) as well as four counties (i.e., Gangcha, Haiyan, Gonhe, and Tianjun) around the Qinghai Lake, with a total population of ∼240,000. Local sources include residential burning activities and vehicles. Urban areas of Xining are located ∼180 km southeast of the Qinghai Lake. Total suspended particle (TSP) samples were collected weekly on pre-combusted quartz fiber filters (20.3 cm × 25.4 cm; QM-A, Whatman Inc.) using a high-volume aerosol sampler (TH150-A, Wuhan Tianhong INST Group) from September 2018 to August 2019. Blank filters were collected in each season exactly like the sample filters except that no air was drawn through the filter. The filters were individually wrapped in aluminum foils, packed in airtight polyethylene bags, and stored at −20°C for subsequent analysis. The real-time concentrations of PM 10 and PM 2.5 were measured at Haiyan County and obtained from China National Environmental Monitoring Centre. Equivalent black carbon (eBC) concentrations were measured using an Aethalometer (Model AE33, Magee Scientific) (see Text S1 in Supporting Information S1).

Analytical Methods
OC and EC concentrations were measured using a thermo-optical carbon analyzer (Model 5, Sunset Laboratory Inc.) applying the EUSAAR_2 protocol (Cavalli et al., 2010), as detailed in Text S1 in Supporting Information S1.
For analysis of carbon isotope, a subset of 12 samples (3 samples per season) with varying loading of CAs were selected to represent different pollution conditions. The 12 samples cover 98 days, including 10 samples with weekly time resolution and 2 composite samples. Each composite sample combines two weekly samples collected during periods with similar TC (= OC + EC) concentrations (relative differences <15%) and similar back trajectories (Table S1 in Supporting Information S1). The analysis of stable carbon isotope composition (δ 13 C) was conducted by linking a carbon analyzer (Model 5, Sunset Laboratory Inc.) with a continuous flow isotope ratio mass spectrometer (IRMS; 652 Optima, Isoprime Ltd.; Dusek et al., 2013), including δ 13 C of EC  as well as OC that desorbs in He at three consecutive temperature steps of 200°C, 350°C, and 550°C .
For 14 C analysis, OC and EC were first converted to CO 2 using an aerosol combustion system , followed by the 14 C measurements using the Mini Carbon Dating System accelerator mass spectrometer at 3 of 11 the University of Groningen (Dee et al., 2019). The 14 C data are reported as fraction modern (F 14 C; Reimer et al., 2004). The detailed measurement procedures and quality controls of δ 13 C and F 14 C measurements were present in Texts S2-S4 of Supporting Information S1.

Source Apportionment Method
F 14 C of OC and EC allows for the clear separation between the relative contributions of fossil (f fossil (OC), f fossil (EC)) and non-fossil sources (f nf (OC), f bb (EC)). f nf (OC) and f bb (EC) can be calculated by dividing the sample F 14 C values through the corresponding values of non-fossil sources (F 14 C nf ). F 14 C nf values are estimated as 1.10 ± 0.05 for EC and 1.09 ± 0.05 for OC, as detailed in Ni et al. (2019), using the long-time observation of atmospheric F 14 C of CO 2 and a tree growth model.
The f fossil (EC) is further divided into the fractions of EC produced from liquid fossil fuel combustion (f liq.fossil ) and coal combustion (f coal ). The two groups have their distinct 13 C source signatures and EC preserves the 13 C signature of emission sources, which allows for isotope-based quantification by the Bayesian Markov chain Monte Carlo (MCMC) simulations (Andersson, 2011;Andersson et al., 2015Andersson et al., , 2020: (1) F 14 C liq.fossil and F 14 C coal are 0, and F 14 C nf is 1.10 ± 0.05, as explained above. The best estimate of 13 C source signatures of EC is established in Text S5 of Supporting Information S1 through a literature search: δ 13 C liq.fossil (−27.0 ± 1.0‰), δ 13 C coal (−23.8 ± 1.2‰), and δ 13 C bb (−27.1 ± 1.7‰ for C3 plants, −14.9 ± 1.7‰ for C4 plant). Uncertainties of both measurements (F 14 C (EC) and δ 13 C EC ) and source signatures of EC are accounted for in the MCMC simulations. The MCMC was run for 10,000 iterations, with a burn-in of 1,000 and a data thinning of 10. With this setup, convergence diagnostics were established to make sure the good MCMC convergence. The MCMC results are the posterior probability density functions (PDFs) of f bb , f liq.fossil , and f coal . The PDFs' median and interquartile range (25%-75%) represent the best estimate and variability, respectively. The 13 C source signatures are not as well constrained as 14 C. The potential influence of 13 C source signatures on the MCMC results was explored in a sensitivity analysis (Text S6 in Supporting Information S1). It shows that our best estimate scenario is relatively robust.
With 14 C-determined f bb (EC) and f nf (OC), EC and OC were resolved into fossil and non-fossil EC and OC (EC fossil , EC bb , OC fossil , and OC nf ). With the MCMC-derived f liq.fossil and f coal , EC fossil can further be separated into EC liq.fossil and EC coal . OC fossil and OC nf were further apportioned into OC fossil from primary and secondary sources (POC fossil and SOC fossil , respectively), OC nf from primary biomass-burning (POC bb ) and other non-fossil sources (OC o,nf ). OC o,nf includes secondary OC nf (SOC nf ), primary biogenic OC, and cooking OC. Due to the sparse population in the TP, cooking OC is expected to be of minor importance. Based on analysis of tracers of primary biogenic aerosol particles (including arabitol, mannitol, and glucose), the concentrations of the fungal-spore-derived OC and plant-debris OC were estimated as 74 ± 49 ng m −3 , accounting for only 1.7% ± 2.1% OC during our sampling campaign. This suggests low abundance of primary biogenic OC. Therefore, OC o,nf is an approximation of SOC nf or a upper limit of SOC nf . Total SOC was thus estimated as the sum of SOC fossil and OC o,nf . The equations for detailed source apportionment are shown in Table S2 of Supporting Information S1.

Seasonal Variability
The particulate air pollution at Qinghai Lake showed large seasonal variability. The yearly minima in concentrations of PM 10 , PM 2.5 , and equivalent BC (eBC, i.e., an optically based analog of mass-based EC) were typically observed during the monsoon season ( Figure S1 in Supporting Information S1). Not only increased wet scavenging ( Figure S2 in Supporting Information S1), but also decreased human activities for heating (e.g., biomass burning, coal combustion) lead to reduced emissions in the monsoon season. These findings are in qualitative agreement with previous investigations at Qinghai Lake, as well as other sites in the TP (Cong et al., 2015;Xiang et al., 2021;Zhao et al., 2013). The considerable seasonal variation in particle concentrations occurs because of varying emission sources and differences in aerosol lifetimes in different seasons. In the pre-monsoon season, the high eBC was associated with strong wind from west and northwest sectors in the bivariate polar plot (Figure 1a), suggesting probable impacts from regional transport. In the rainy monsoon season, the BC lifetime is expected to be shorter, which would mean a weak influence from regional transport and a more pronounced influence from local sources. In post-monsoon and winter, elevated eBC concentrations mainly occur at low wind speed (<3 m s −1 ), indicating the important contribution of local/regional emissions. It has been shown that in addition to regional transport, local sources are also an important contributor to CAs in the TP . This study observed the annual mean eBC of 0.60 ± 0.28 μg m −3 during 2018/2019 at Qinghai Lake, northeastern TP, similar to the concurrent observation in Beiluhe, central TP (0.61 ± 0.32 μg m −3 ). eBC concentrations were higher during 2019/2020 in Ngari, southwestern TP (1.08 ± 0.76 μg m −3 ), where polluted air masses from northern India played a role (Zhu et al., 2021).

14 C-Based Source Apportionment of OC and EC
Radiocarbon allows for quantifying the relative contribution of fossil and non-fossil sources with high precision, regardless of atmospheric processing. For the 1-yr study period, the concentration-weighted 14 C-based relative contribution of fossil sources to EC (f fossil (EC)) was 70% ± 10% (± weighted SD) with a large seasonal variability ( Figure 1b). We find high fossil contributions to EC during the monsoon season (82% ± 3%), with the remaining EC contributed by biomass burning, showing a strong impact from fossil fuel combustion in the monsoon season. This is mainly due to higher contribution of liquid fossil fuel sources to EC in the monsoon season as discussed in Section 3.3. In the pre-monsoon and post-monsoon season, the fossil contributions to EC were lower (68% ± 6% and 60% ± 9%, respectively), and decreased with the increase in EC concentrations (Figure 1c). The f fossil (EC) in winter was relatively stable (70% ± 1%) and did not vary with EC concentrations.
The seasonal variation of f fossil (EC) was driven by a clear seasonal cycle of biomass-burning EC at Qinghai Lake. The seasonal mean biomass-burning EC concentrations (EC bb ) varied by three times from 0.14 μg m −3 in the monsoon season to 0.39 μg m −3 in the post-monsoon season, larger than the variation of fossil-derived EC (EC fossil = 0.57-0.67 μg m −3 ).
The fossil contributions to OC (f fossil (OC)) increased with increasing f fossil (EC), but were consistently smaller than those to EC ( Figure S3 in Supporting Information S1). f fossil (OC) showed a similar seasonal pattern with f fossil (EC), with higher f fossil (OC) in the monsoon season (32% ± 10%) and lower in the post-monsoon season (19% ± 4%).
Despite the seasonality, f fossil (OC) highly correlates with the inverse of OC concentrations (R 2 = 0.83, p < 0.001; Figure 1d), suggesting that the atmospheric OC concentrations reflect the combined contribution of the OC background and the additional non-background OC source that contributed to increases in atmospheric OC at Qinghai Lake (Keeling, 1958(Keeling, , 1961. This inverse relation gives a f fossil (OC) of 12% as OC concentrations approach infinity, showing that the non-background OC source was mainly of non-fossil origin. In winter, although the f fossil (OC) and OC concentrations agreed with the overall inverse trend, they showed the opposite trend that f fossil (OC) increased with the increase in OC concentration, signifying an important contribution of fossil sources to the OC increment in winter. The f fossil (OC) and f fossil (EC) reported here are generally larger than those in other remote areas of southern TP (e.g., Nam Co and Zhongba) that more influenced by biomass burning within the TP and from South Asia, but smaller than those in Lhasa, a typical polluted city in the southern TP ( Figure S4 in Supporting Information S1; Li et al., 2016;Li, Bosch, et al., 2022). Despite being higher than at other remote sites of the TP, our f fossil (EC) was only a little lower than that of East China (e.g., ∼80% in Beijing; Andersson et al., 2015;Zhang et al., 2015), probably associated with contributions of regional transport from east as indicated in the bivariate polar plot (Figure 1a).

Stable Carbon Isotope Composition of EC
The 13 C signature of EC does not change in the atmosphere, and it is thus the result of source mixing. In our samples, all the δ 13 C of EC overlapped with the endmembers of C3 plant burning, liquid fossil fuel combustion (e.g., traffic) and coal combustion (Figure 2a). C4 plant burning also contributed to EC in Qinghai Lake, evidenced by some enriched δ 13 C EC datapoints (e.g., the maximum δ 13 C EC of −22.7‰ in winter) that were on the higher end of coal combustion endmember (δ 13 CC C coal = −23.8 ± 1.2‰). The δ 13 C EC showed considerable seasonal changes: it was most depleted in 13 C in the monsoon season (−26.9 ± 0.8‰), and higher in winter (−24.8 ± 1.4‰) and the post-monsoon season (−24.5 ± 1.3‰), with the most enriched values in the pre-monsoon season (−23.5 ± 0.5‰). With the depleted 13 C signatures of EC, the higher fossil contribution to EC in the monsoon season was further attributed to the significant influence of liquid fossil fuel combustion (δ 13 C liq.fossil = −27.0 ± 1.0‰). δ 13 C EC in other seasons was comparable to the 13 C source signature of coal combustion, pointing to a larger influence of coal combustion. Compared with the only other seasonal observations on δ 13 C EC in the TP, a much smaller seasonal variation of δ 13 C EC within 1‰ was found in Lhasa, a typical polluted city ( Figure S4 in Supporting Information S1; Li et al., 2016). The Bayesian MCMC results show that liquid fossil fuel combustion accounted for ∼70% of EC in the monsoon season, and the contribution is reduced by half in other seasons (Figure 2b). We also found that coal combustion and biomass burning contributed roughly equally to EC in the monsoon season (13% and 17%, respectively); their contributions increased by around two to three times in other seasons. Despite the fairly constant EC fossil (i.e., the sum of EC coal and EC liq.fossil ) concentrations in different seasons, EC coal and EC liq.fossil showed large seasonal variations, with minimum EC coal and maximum EC liq.fossil in the monsoon season ( Figure S5 in Supporting Information S1). The high EC liq.fossil concentration in the monsoon season was consistent with increased number of tourist vehicles in the tourism rush season at Qinghai Lake (L. Hu et al., 2021).

Stable Carbon Isotope Composition of OC as a Function of Thermal Refractiveness
Unlike EC, the 13 C signature of OC reflects the integrated effects of source mixing as well as atmospheric processing (e.g., secondary formation and photochemical oxidation) via the kinetic isotope effect. OC desorbed at three different temperature steps of 200°C, 350°C, and 550°C (i.e., OC 200°C , OC 350°C , and OC 550°C ), approximately separating OC according to volatility (Ma et al., 2016), had different δ 13 C values ranging from −29.2 ± 0.9‰ for δ 13 C OC,200 to −27.9 ± 0.7‰ for δ 13 C OC,350 to −26.2 ± 1.5‰ for δ 13 C OC,550 . In contrast, source samples from a single source usually do not show such large differences between δ 13 C OC values at different temperature steps (<2‰; Zenker et al., 2020). At Qinghai Lake, this large difference in δ 13 C OC values at different desorption temperatures indicates the varying contribution of sources and atmospheric processes to the different OC volatility fractions.
Overall, OC 550°C and OC 350°C were comparable large mass fraction of OC, constituting 25% and 26% of total OC, respectively. OC 200°C was the smallest fraction for all seasons (∼6% of total OC). OC 200°C , OC 350°C , and OC 550°C do not add up to total OC, because a portion of OC forms charred OC that cannot be desorbed in helium. The larger mass fraction of OC 200°C , the more volatile OC is, because more volatile compounds tend to desorb at lower temperatures. Usually, freshly emitted POC and newly formed SOC are more volatile than aged OC that has been modified in the atmosphere by extensive photochemical processing (Keller & Burtscher, 2017;Masalaite et al., 2020;Meusinger et al., 2017). The mass fraction of OC 200°C in total OC at Qinghai Lake was smaller than that in urban Beijing and much smaller than for primary emissions (Figure 3a), probably due to active photochemical processing of OC emissions under strong solar radiation over the TP (Yang et al., 2014). On the other hand, the mass fraction of OC 200°C was higher (i.e., OC was more volatile) in the post-monsoon season and winter than in the pre-monsoon and monsoon season. This could be caused by lower temperature in winter ( Figure S2 in Supporting Information S1) that promotes partitioning of semi-volatile OC to the condensed phase. In addition, increased solar radiation in the pre-monsoon and monsoon season (Table S1 in Supporting Information S1) led to enhanced photochemical activity and thus less volatile OC (Masalaite et al., 2020). This is also consistent with in Qinghai Lake (this study) and Beijing , as well as in the primary emission of C3 plant burning, coal combustion and traffic in China . The symbol size in panel (a) is proportional to the mass fraction of OC 200°C in total OC (i.e., OC determined by the EUSAAR_2 protocol). Panels (b and c) show the scatter plots of SOC/OC ratios versus δ 13 C of OC 200°C , OC 350°C , and OC 550°C in different seasons. For secondary OC (SOC) estimation, see details in Section 2.3. (SOC/OC) nf , non-fossil SOC/OC ratio; (SOC/OC) fossil , fossil SOC/OC ratio. 7 of 11 our laboratory studies, where source samples were aged in a small reactor under UV light, that photolysis causes mainly mass loss in OC 200°C relative to OC that desorbs at higher temperatures (Ettinger, 2022).
Recently formed SOC is considered to be 13 C-depleted, whereas aged OC is enriched in 13 C (Dasari et al., 2019;Fisseha et al., 2009;Irei et al., 2006;Kirillova et al., 2013;Pavuluri & Kawamura, 2016). In our samples, the most strongly depleted δ 13 C OC,200 values fall out of the ranges of 13 C signatures of anthropogenic primary emission (Figure 3a). The most likely explanation for the strongly depleted δ 13 C OC,200 is the formation of (biogenic) SOC that has depleted (more negative) 13 C signature and contributes strongly to the more volatile fraction. More depleted δ 13 C values of OC at lower desorption temperature was also found for urban aerosols in Beijing, China  and Naples, Italy , as well as urban, coastal, and forest aerosols in Lithuania (Masalaite et al., 2017(Masalaite et al., , 2020. Our δ 13 C OC,200 values varied with seasons, with the less depleted values in winter compared with other seasons (Figure 3). In winter, OC contained less SOC than that in other seasons (Figure 3b), and the δ 13 C of OC 200°C showed a tendency toward depleted values with increased SOC/OC ratios. The largest SOC/OC ratios was 0.38 in winter, characterized by the most depleted δ 13 C OC,200 of −29.2‰. The SOC/OC ratios were further increased in other seasons (i.e., non-winter seasons) up to 0.57; however, δ 13 C OC,200 values did not further decrease, and converged to a narrow range of −30.4‰ to −29.1‰. A similar trend was also observed for δ 13 C OC,200 versus non-fossil SOC/OC ratio, but not present for δ 13 C OC,200 versus fossil SOC/OC ratio. This again indicates the importance of non-fossil secondary sources on the variation of δ 13 C OC,200 .
Much more depleted than δ 13 C OC,200 for anthropogenic primary emissions, our δ 13 C OC,200 in non-winter seasons (weighted average: −29.6‰; range: −30.4‰ to −29.1‰) is a potential 13 C signature for SOC under strong solar radiation in a remote area. The relatively depleted δ 13 C OC,200 values in non-winter seasons are in line with the δ 13 C of OC 200°C during the clean period in summer at a forest site in Lithuania (−29.6 ± 0.5‰; Masalaite et al., 2020) and total SOC from β-pinene oxidation (−29.6‰;Fisseha et al., 2009), and are a slightly heavier than the δ 13 C values of SOC from toluene (−31.6‰ to −32.3‰; Irei et al., 2006).
In winter, the OC 200°C fraction at Qinghai Lake was lower and ∼1.7‰ depleted in 13 C compared to OC 200°C in urban Beijing (Figure 3a). This is probably due to the proximity of primary emission sources in the urban environment, which have higher δ 13 C values than SOC and lead to higher OC 200°C amounts due to less aging times. δ 13 C OC,350 values varied in a small range within 2‰ for all samples in all seasons, and did not change significantly with SOC/OC ratios (Figure 3b and Figure S6 in Supporting Information S1). In contrast, δ 13 C of OC 550°C , the less volatile OC fraction, varied in a wide range from −28.5‰ to −22‰, but did not show any correlation with SOC/OC ratios. δ 13 C OC,550 was more enriched in 13 C (up to −22‰) compared with δ 13 C OC,350 and δ 13 C OC,200 , suggesting the contribution of specific sources with high δ 13 C values to OC 550°C . However, our 13 C signature of EC shows that coal combustion (−21‰ to −25‰; Widory, 2006), a relatively 13 C-enriched source, was not a dominant source. The enriched δ 13 C OC,550 was therefore related to the atmospheric aging processes that leading to 13 C enrichment (Dasari et al., 2019;Irei et al., 2011;Pavuluri & Kawamura, 2016). This finding is consistent with previous observations in ambient atmosphere and our laboratory studies (Ettinger, 2022;Masalaite et al., 2020;Ni et al., 2022).

Conclusions and Implications
Observational constraints on sources and formation of CAs are scarce and rarely extend over more than one season in the TP. This study presents the first year-round dual-carbon isotopic analysis of CAs in the northeastern TP (Qinghai Lake). EC was dominated by fossil sources (70% ± 10%), while non-fossil sources contributed more strongly to OC (76% ± 7%). 14 C results also show that there was a relatively stable background aerosol with a relatively strong contribution of fossil sources and a non-background source of non-fossil origins. This drives a large seasonality in both concentrations and sources of EC and OC, with lower concentrations and higher fossil contribution in the monsoon season. An opposite seasonality of EC sources was previously found in the Arctic, though with comparable fossil contribution (69% ± 19%; Winiger et al., 2017). Fossil contributions to EC and OC were lower in the southern TP that are more influenced by biomass burning within the TP and from South Asia Li, Bosch, et al., 2022), highlighting the spatially heterogeneous of local sources and transport over the TP. Therefore, high-resolution emission inventories specifically for TP are needed to better characterize locally sourced CAs.
The observed OC at Qinghai Lake was less volatile than OC found in primary emissions and in urban areas, reflecting highly processed OC under strong solar radiation. Large differences were found in 13 C of different OC volatility fractions. More volatile OC was depleted in 13 C by as much as 3‰ relative to less volatile OC. This difference is caused by the kinetic isotopic effect in atmospheric processes. First, SOA compounds, which are depleted in 13 C, contributed strongly to more volatile OC (−29.2 ± 0.9‰). This influence of secondary formation was further justified by the negative correlation between δ 13 C and SOC/OC ratios, providing observational evidence for 13 C-depleted SOC besides few laboratory studies, for example, δ 13 C of SOC from β-pinene oxidation (−29.6‰; Fisseha et al., 2009) and toluene (−31.6‰ to −32.3‰; Irei et al., 2006). On the other hand, photochemical aging played an important role in the less volatile OC, evidenced by the enriched δ 13 C values (up to −22‰) that cannot be explained by primary source mixing. This is consistent with what we expect from photolysis with strong UV light (Ettinger, 2022). The 13 C signature of OC thus provides a strong basis to investigate aerosol processing. However, isotopic fractionation induced by atmospheric processes is not quantitatively constrained and needs further laboratory investigations.
TSP samples are used in this study, however, sources and transformations of CAs in PM 2.5 are probably different due to different particle sizes . Therefore, in the future, PM 2.5 needs to be investigated to more comprehensively understand CAs over TP. Together, this study provides observational constraints on sources and atmospheric processes of CAs in a climatically sensitive region, thereby facilitating improved modeling of aerosol climate effects.

Data Availability Statement
Data used in this study can be accessed online (at https://doi.org/10.5281/zenodo.7380212).