Inﬂuence of Ambient Atmospheric Environments on the Mixing State and Source of Oxalate-Containing Particles at Coastal and Suburban Sites in North China

: Photodegradation is a key process impacting the lifetime of oxalate in the atmosphere, but few studies investigated this process in the ﬁeld due to the complex mixing and sources of oxalate. Oxalate-containing particles were measured via single-particle aerosol mass spectrometry at coastal and suburban sites in Qingdao, a coastal city in North China in the summer of 2016. The mixing state and inﬂuence of different ambient conditions on the source and photodegradation of oxalate were investigated. Generally, 6.3% and 12.3% of the total particles (by number) contained oxalate at coastal and suburban sites, respectively. Twelve major types of oxalate-containing particles were identiﬁed, and they were classiﬁed into three groups. Biomass burning (BB)-related oxalate–K and oxalate– carbonaceous particles were the dominant groups, respectively, accounting for 68.9% and 13.6% at the coastal site and 72.0% and 16.8% at the suburban site. Oxalate–Heavy metals (HM)-related particles represented 14.6% and 9.3% of the oxalate particles at coastal and suburban sites, respectively, which were mainly from industrial emissions (Cu-rich, Fe-rich, Pb-rich), BB (Zn-rich), and residual fuel oil combustion (V-rich). The peak area of oxalate at the coastal site decreased immediately after sunrise, while it increased during the daytime at the suburban site. However, the oxalate peak area of Fe-rich particles at both sites decreased after sunrise, indicating that iron plays an important role in oxalate degradation in both environments. The decay rates ( k ) of Fe-rich and BB-Fe particles at the coastal site ( − 0.978 and − 0.859 h − 1 , respectively), were greater than those at the suburban site ( − 0.512 and − 0.178 h − 1 , respectively), owing to the high-water content of particles and fewer oxalate precursors. The estimated k values of oxalate peak area for different ambient conditions were in the same order of magnitude, which can help establish or validate the future atmospheric models.


Introduction
Carboxylic acids are important organics in the troposphere. Dicarboxylic acids have high water solubility owing to the existence of two carboxyl groups and mainly exist as liquid and particles in the atmosphere [1,2]. Oxalate is the most abundant dicarboxylic acid in atmospheric aerosols. Oxalate can affect the properties of cloud condensation nuclei and ice nuclei and thus influence the cloud formation, and earth radiation balance [3][4][5][6]. 17 August 2016, while those at OUC were measured from 19 to 23 August and from 31 August to 7 September 2016. During the whole sampling campaign at the LQH site, the average T and RH were 28.5 • C and 87.2%, with ranges of 25.0-33.7 • C and 42.0-99.0%, respectively. OUC site showed a similar average T (27.4 • C), but much less RH (71.1%). The WS is similar at these two sites, with the average WS being 1.7 and 1.3 m/s at LQH and OUC, respectively.

Online Instruments and Data Analysis
The detection methods and principles of SPAMS have been previously reported in detail [23]. Briefly, a single-particle aerosol mass spectrometer mainly consists of an aerodynamic lens, laser system, and bipolar time-of-flight mass spectrometer. During sampling, the inlet was set at a height of 4 m on a roof of a container at the coastal site, and 8 m on an academic building at the suburban site. The atmospheric particles were dried by a drying tube before entering the instrument. Then, the dry particles were introduced into the spectrometer using an aerodynamic lens. The particles were focused onto the centerline of the inlet, and they achieved a specific velocity when exiting the aerodynamic lens. The particles entered a laser system consisting of two continuous-mode diodes (neodymiumdoped yttrium aluminum garnet [Nd:YAG]) laser beams (532 nm), and their aerodynamic diameters were calculated by the laser system. Finally, the particles that arrived at the bipolar time-of-flight mass spectrometer were desorbed/ionized by the Nd:YAG laser (266 nm). Subsequently, the positive and negative ions were detected using the bipolar time-of-flight mass spectrometer. The vacuum aerodynamic diameter (d va ) of the measured individual aerosol particles was calibrated using standard polystyrene latex spheres with diameters of 0.2-2.0 µm.
The single-particle mass spectrum data were processed using the YAADA2.1 toolkit (www.yaada.org, accessed on 30 September 2016), which is based on MATLAB. Moreover, 6,364,529 and 7,319,747 particles were collected from the LQH and OUC sites, respectively, of which 670,887 and 1,718,778 were chemically analyzed to obtain both positive and negative ion spectra. Oxalate-containing particles were identified by a marker ion of m/z −89[HC 2 O 4 ] − according to a relative peak area (RPA) of >0.5% [2,10,25]. According to this definition, 42,226 and 211,180 oxalate-containing particles were obtained, accounting for 6.3% and 12.3% of the total particles at LQH and OUC sites, respectively. These mass spectra of the oxalate-containing particles were classified according to the adaptive resonance theory based on a neural network algorithm (ART-2a), with a learning rate of 0.05, a vigilance factor of 0.7, and 20 iterations. The ART-2a is an effective way to intelligently find clustering in a data set with a large number of variables [26]. In this study, the ART-2a procedure automatically generated 651 clusters of particles based on the presence and intensity of ion peaks. The first 252 clusters, representing over 98% of all of the ionized oxalate-containing particles, were considered important and further classified into 12 major groups according to mass spectrum similarities, particle size distributions, and temporal variations. The clusters that could not be manually classified were labeled "undefined".

Screening of Five HM-Containing Particles
Owing to the higher abundance, and considering their stability constant and solubility of HM complexed with oxalate-containing particles, this study focused on Fe, V, Cu, Zn, and Pb ions [6]. Fe-containing particles were screened according to a peak area (PA) of m/z 56 > 50 and a PA ratio of 56

General Characteristics of SPAMS Particles
The particle statistics of the individual metals are listed in Table 1. Of the total particles (by number) at the LQH sampling site, 6.3% contained oxalate, while 3.4%, 6.4%, 3.4%, 1.7%, and 0.5% contained Fe, V, Cu, Pb, and Zn, respectively. The proportion of oxalate-containing particles at OUC (12.3%) was considerably higher than that at LQH. Likewise, the proportion of HM-containing particles was higher, except for V-containing particles [33,34]. Furthermore, 30.1% and 25.3% of the oxalate-containing particles at LQH and OUC were associated with the abovementioned five HMs. At LQH, Fe was the most abundant HM (by particle number) associated with oxalate-containing particles, accounting for 36.4% of the oxalate-HM particles, followed by Cu (28.4%), V (28.3%), Pb (24.7%), and Zn (8.7%). In contrast, at OUC, Cu was the most abundant HM (by particle number), accounting for 40.4% of the oxalate-HM particles, followed by Fe (31.1%), V (28.9%), Pb (17.7%), and Zn (14.6%). The mixing states of oxalate with HMs at these sites differed from those of cities in South China. In Hong Kong and Guangzhou, Pb was the most abundant HM (by particle number), accounting for 47.7% and 32.0% of the oxalate-HM particles, respectively [6,10]. The different ambient atmospheric environments, e.g., industry and energy consumption structures, may cause the variable mixing states of oxalate and HMs in different regions.

Individual Characteristics of Oxalate-Containing Particles
To identify the characteristics and sources of individual oxalate-containing particles, we separated the oxalate particles into 12 major groups. Table 2 presents the statistics of the classified groups of oxalate particles. The average spectra of these 12 major groups and the undefined group can be found in Figure S2. Figure 1 shows the size-resolved contributions of each group to the total oxalate particles (by particle number). The temporal trends of the meteorological factors and the number fraction of various oxalate-containing particle groups during the sampling period at LQH and OUC are shown in Figure 2a,b, respectively. The time series of the percent contributions by individual groups to total particle number at LQH and OUC are shown in Figure S3a,b, respectively. The number fractions of selected ion markers associated with single-particle types at LQH and OUC are shown in Figure S4a,b, respectively. Atmosphere 2022, 13, x FOR PEER REVIEW particles showed higher contributions during the continental air mass control (Table S1). These findings suggest that the oxalate-containing particles at both sit mainly from BB emission inland air masses. We also noticed at both LQH and OU that almost all oxalate particles mixed with sulfate and nitrate ( Figure S4), sugg secondary formation of oxalate particles [6,15]. However, the OUC site had mor dant nitrate than LQH ( Figures S2 and S4), indicating the suburban site was inf by more human activity compared with that at the coastal site [58].  Here, we briefly introduce the mass spectra of the 12 oxalate groups, as previous studies have already identified similar particle types and discussed their composition in detail [6,10, [35][36][37][38].

Oxalate-K-related particles
The positive spectra of K-rich, K-Na, and K-elemental carbon-organic carbon (K-ECOC) oxalate-related particles featured 39[K] + as the highest ion peak ( Figure S2). In addition, the positive spectrum of the K-Na particles featured 23[Na] + as the second-highest ion peak, and that of K-ECOC particles featured typical elemental and organic carbon ion clusters (such as 12   These three oxalate-K-related groups were the dominant particles at both sites, accounting for 68.9% and 72.0% of the total particles at LQH and OUC, respectively ( Table 2). In addition, the temporal variations in K-rich, K-Na, and K-ECOC particles were similar at both LQH and OUC sites ( Figure S3). The particle size of K-Na particles was larger than K-rich and K-ECOC particles at both sites ( Figure 1). Most of these three oxalate-K-related particles were mixed with CN − , CNO − , and levoglucosan (LEV) ( Figure S4), which are considered good tracers for biomass burning (BB), suggesting that these three oxalate-Krelated particles originated from BB [35,[39][40][41]. In addition, this group of particles also showed a signal from −79[HPO4] − ( Figure S2), suggesting vegetative debris may be mixed during the biomass burning process [42].
Oxalate-carbonaceous-related particles Three oxalate-carbonaceous-related particle types were identified. Elemental carbon (EC), EC-Na, and ECOC particles exhibited typical elemental carbon ion clusters ( Furthermore, these three oxalate-carbonaceous-related particle types accounted for 13.6% and 16.8% of the total particles at the LQH and OUC sites, respectively (Table 2). At both LQH and OUC, EC accounted for a more significant proportion of small particles than EC-Na and ECOC (Figure 1). At both LQH and OUC, the three oxalate-carbonaceousrelated particle types exhibited a comparable temporal variation to the three oxalate-K-related particle types (Figures 2 and S3), suggesting similar sources or atmospheric processes of both particle types. Moreover, most of the three oxalate-carbonaceous-related particles were mixed with CN − , CNO − , and LEV ( Figure S4), which confirmed the existence of a BB emission source [35,39,41,43].

Oxalate-HM-related particles
The oxalate-HM particles included Fe-, V-, Pb-, Zn-, Cu-rich particles, and particles containing other metals. The positive spectrum of Fe-rich particles featured strong signals at 54 At LQH and OUC, Fe-rich particles accounted for 8.8% and 5.7% of the total oxalate particles, respectively (Table 2). At both LQH and OUC, the time series of Fe-rich particles is different from these oxalate-K-related particle types (Figures 2 and S3), suggesting that the Fe-rich particles may be from sources other than BB, such as industrial emissions [30,46]. In addition, most of the Fe-rich particles were also mixed with CN − , CNO − , and LEV ( Figure S4), which may be due to the mixing between Fe particles and BB particles.
Compared with the contribution of Fe-rich particles, those of the V-, Pb-, Zn-, and Cu-rich groups were minor. At both LQH and OUC, Pb-and Cu-rich particles exhibited a comparable time series to Fe-rich particles, suggesting that they were from similar sources, such as industrial emissions [47][48][49][50][51]. The time series in Zn-rich particles was different from that in Fe-rich particles. Previous studies have found Zn-rich particles in emissions from waste incineration [52,53]. We also found high CN − and LEV loadings in Zn-rich particles. Bulk measurements revealed relatively abundant Zn emissions from BB [44,45]. Moreover, 98% and 80% (by particle number) of Zn-rich particles exhibited LEV signals at LQH and OUC sites, respectively ( Figure S4), suggesting that they were mostly related to BB. Compared to marine air mass, almost no V-rich particles occurred in the oxalate-containing particles during continental air mass control, suggesting that the V particles were mainly emitted by marine vessels through residual fuel oil combustion [28,29,34,42,[54][55][56][57].
At the two sites, the overall contributions of different sources to oxalate were similar. However, at LQH, there were almost no oxalate-containing particles during marine air mass control (1.4% in Table S1), but 12.4% of the total particles were related to oxalate under continental air mass. At OUC, the number of oxalate particles during the continental air mass control period (13.4%) was approximately twice that during the marine-source air mass-dominated period (7.2%). The oxalate-K and oxalate-carbonaceous particles showed higher contributions during the continental air mass control period (Table S1). These findings suggest that the oxalate-containing particles at both sites were mainly from BB emission inland air masses. We also noticed at both LQH and OUC sites, that almost all oxalate particles mixed with sulfate and nitrate ( Figure S4), suggesting a secondary formation of oxalate particles [6,15]. However, the OUC site had more abundant nitrate than LQH (Figures S2 and S4), indicating the suburban site was influenced by more human activity compared with that at the coastal site [58].

Diurnal Variation in Oxalate-Containing Particles
The average diurnal variations in meteorological parameters, particle number, and PA/RPA of oxalate-containing particles with m/z −89 are illustrated in Figure 3. The oxalate particle number, PA, and RPA showed similar diurnal variations; thus, we selected a PA of m/z −89 to illustrate the diurnal characteristics of oxalate-containing particles. The diurnal variation in oxalate at LQH was different from that at OUC. At the coastal site, the PA of oxalate decreased from 5:00 a.m. onwards after sunrise, which indicated that photodegradation was a key decay mechanism. The PA of oxalate at LQH began to increase from 10:00 a.m. onwards, and approximately 50% of oxalate was decomposed over 5 h. However, the PA of oxalate at OUC increased from 6:00 a.m. onwards, and the PA for daytime was broad. The difference in diurnal variation suggests a distinct source and decay process. A similar sharp decay of oxalate after sunrise has been reported in Hong Kong, which suggests that the oxalate source was limited to the rural coastal site, so that the formation rate was less than the decay rate (k), resulting in a considerable decrease in the number of oxalate particles [6]. In contrast, OUC is a typical suburban site with more human activity and small surrounding restaurants, resulting in the emission of a large number of oxalate-containing particles from BB/cooking [2,[6][7][8][9]. Moreover, both sites showed a decrease in the number of oxalate-Fe-containing particles ( Figure 3f); the particles at LQH decayed faster than those at OUC and were detected 1 h earlier. The sun rises earlier at LQH than at OUC, explaining the difference in the diurnal variations in oxalate-Fe particles at both sites under ambient conditions. Moreover, the abundant source of oxalatecontaining particles at OUC due to the residential areas and small surrounding restaurants may be responsible for the difference in the total diurnal variations in the particles.

Diurnal Variation in Oxalate-Containing Particles
The average diurnal variations in meteorological parameters, particle number, and PA/RPA of oxalate-containing particles with m/z −89 are illustrated in Figure 3. The oxalate particle number, PA, and RPA showed similar diurnal variations; thus, we selected a PA of m/z −89 to illustrate the diurnal characteristics of oxalate-containing particles. The diurnal variation in oxalate at LQH was different from that at OUC. At the coastal site, the PA of oxalate decreased from 5:00 a.m. onwards after sunrise, which indicated that photodegradation was a key decay mechanism. The PA of oxalate at LQH began to increase from 10:00 a.m. onwards, and approximately 50% of oxalate was decomposed over 5 h. However, the PA of oxalate at OUC increased from 6:00 a.m. onwards, and the PA for daytime was broad. The difference in diurnal variation suggests a distinct source and decay process. A similar sharp decay of oxalate after sunrise has been reported in Hong Kong, which suggests that the oxalate source was limited to the rural coastal site, so that the formation rate was less than the decay rate (k), resulting in a considerable decrease in the number of oxalate particles [6]. In contrast, OUC is a typical suburban site with more human activity and small surrounding restaurants, resulting in the emission of a large number of oxalate-containing particles from BB/cooking [2,[6][7][8][9]. Moreover, both sites showed a decrease in the number of oxalate-Fe-containing particles ( Figure 3f); the particles at LQH decayed faster than those at OUC and were detected 1 h earlier. The sun rises earlier at LQH than at OUC, explaining the difference in the diurnal variations in oxalate-Fe particles at both sites under ambient conditions. Moreover, the abundant source of oxalate-containing particles at OUC due to the residential areas and small surrounding restaurants may be responsible for the difference in the total diurnal variations in the particles. According to previous studies, PA provides more information than particle number and is recommended over RPA to track the variations in low-intensity species [5,6]. The average diurnal variations in the relative contributions to different classified oxalate groups of PA (m/z −89) are shown in Figure 4. The diurnal variations in oxalate particle number and RPA exhibited similar trends (Figures S7 and S8). At LQH, Fe-rich particles accounted for the main decrease in oxalate PA after sunrise (Figure 4). At OUC, although the PA of oxalate increased after sunrise owing to the contribution of oxalate-K-related particles, the oxalate PA in Fe-rich particles decreased, consistent with the trend for LQH. On average, the contribution of Fe-rich particles decreased from 27% to 1% and from 28% According to previous studies, PA provides more information than particle number and is recommended over RPA to track the variations in low-intensity species [5,6]. The average diurnal variations in the relative contributions to different classified oxalate groups of PA (m/z −89) are shown in Figure 4. The diurnal variations in oxalate particle number and RPA exhibited similar trends (Figures S7 and S8). At LQH, Fe-rich particles accounted for the main decrease in oxalate PA after sunrise (Figure 4). At OUC, although the PA of oxalate increased after sunrise owing to the contribution of oxalate-K-related particles, the oxalate PA in Fe-rich particles decreased, consistent with the trend for LQH. On average, the contribution of Fe-rich particles decreased from 27% to 1% and from 28% to 7% after sunrise at LQH and OUC, respectively. Similar diurnal variations were found in oxalate particle number and RPA (m/z = −89, SI Figures S7 and S8). to 7% after sunrise at LQH and OUC, respectively. Similar diurnal variations were found in oxalate particle number and RPA (m/z = −89, SI Figures S7 and S8).

Decay of Oxalate in Different Oxalate-Fe Particle Types
The PAs of the oxalate in all Fe-containing oxalate particles decreased after sunrise at both LQH and OUC (Figures 3 and 4). To identify the relative contributions of different Fe-containing oxalate particles to oxalate decay in the morning, we examined the diurnal variations in all 12 groups of oxalate particles. Only Fe-rich particles and oxalate-K-related particles (from BB) showed considerable diurnal variations. Other groups had few particle numbers, and the uncertainty was too large; hence, we could not consider them. To determine whether the presence of Fe influenced the diurnal variation in BB particles, we separated BB particles into two subgroups: BB-Fe (containing Fe [m/z 56] particles) and BB-noFe (without Fe [m/z 56] particles). Both sites exhibited a rapid decay of BB-Fe particles ( Figure S5c,d), suggesting the importance of Fe in oxalate photodegradation. Only a slight decrease in the number of oxalate particles associated with BB-noFe was observed at LQH, while the number of oxalate particles increased at OUC. Similar diurnal variations of oxalate were observed at LQH and the Hong Kong site, as both Hong Kong and the coastal LQH sites are similar receptor sites. To compare the decay characteristics of oxalate in different ambient environments, we calculated the decay rate k of oxalate PA (m/z −89) ( Figure 5) and RPA (m/z −89) ( Figure S6).
The oxalate PA intensity decay points for Fe-rich and BB-Fe particles were fitted to exponential curves ( Figure 5). The k values of Fe-rich and BB-Fe particles at LQH were −0.978 and −0.859 h −1 , respectively, which were greater than those of the corresponding particles at OUC (−0.512 and −0.178 h −1 , respectively). The k values of Fe-rich particles were greater than those of BB-Fe particles, especially at OUC. The k of BB-Fe particles at OUC was very low, which may be because the residents began to cook after sunrise and primary emissions of some BB-Fe particles or BB particles were mixed with Fe-containing particles. According to our previous studies, the k of oxalate in Fe-rich particles and BB-Fe particles in Hong Kong were −0.6 and −0.23 h −1 , respectively, from February to April [6].
The k of Fe-containing oxalate particles depends on several factors, including aerosol acidity, particle water content, Fe content, oxalate speciation, and the presence of other ligands that may form complexes with oxalate [6, [59][60][61][62]. Several factors can cause a higher k of oxalate at LQH. First, the marine-source air mass usually has a high particle water content, which enhances the Fe-mediated photochemical degradation of oxalate. Moreover, BB is low, and thus, only a few oxalate precursors are emitted, such as methylglyoxal and glyoxal, which cannot produce many oxalate particles compared with the case in OUC [10][11][12]. The previously studied Hong Kong site was also beside a marine

Decay of Oxalate in Different Oxalate-Fe Particle Types
The PAs of the oxalate in all Fe-containing oxalate particles decreased after sunrise at both LQH and OUC (Figures 3 and 4). To identify the relative contributions of different Fe-containing oxalate particles to oxalate decay in the morning, we examined the diurnal variations in all 12 groups of oxalate particles. Only Fe-rich particles and oxalate-K-related particles (from BB) showed considerable diurnal variations. Other groups had few particle numbers, and the uncertainty was too large; hence, we could not consider them. To determine whether the presence of Fe influenced the diurnal variation in BB particles, we separated BB particles into two subgroups: BB-Fe (containing Fe [m/z 56] particles) and BB-noFe (without Fe [m/z 56] particles). Both sites exhibited a rapid decay of BB-Fe particles ( Figure S5c,d), suggesting the importance of Fe in oxalate photodegradation. Only a slight decrease in the number of oxalate particles associated with BB-noFe was observed at LQH, while the number of oxalate particles increased at OUC. Similar diurnal variations of oxalate were observed at LQH and the Hong Kong site, as both Hong Kong and the coastal LQH sites are similar receptor sites. To compare the decay characteristics of oxalate in different ambient environments, we calculated the decay rate k of oxalate PA (m/z −89) ( Figure 5) and RPA (m/z −89) ( Figure S6).
The oxalate PA intensity decay points for Fe-rich and BB-Fe particles were fitted to exponential curves ( Figure 5). The k values of Fe-rich and BB-Fe particles at LQH were −0.978 and −0.859 h −1 , respectively, which were greater than those of the corresponding particles at OUC (−0.512 and −0.178 h −1 , respectively). The k values of Fe-rich particles were greater than those of BB-Fe particles, especially at OUC. The k of BB-Fe particles at OUC was very low, which may be because the residents began to cook after sunrise and primary emissions of some BB-Fe particles or BB particles were mixed with Fe-containing particles. According to our previous studies, the k of oxalate in Fe-rich particles and BB-Fe particles in Hong Kong were −0.6 and −0.23 h −1 , respectively, from February to April [6].
The k of Fe-containing oxalate particles depends on several factors, including aerosol acidity, particle water content, Fe content, oxalate speciation, and the presence of other ligands that may form complexes with oxalate [6, [59][60][61][62]. Several factors can cause a higher k of oxalate at LQH. First, the marine-source air mass usually has a high particle water content, which enhances the Fe-mediated photochemical degradation of oxalate. Moreover, BB is low, and thus, only a few oxalate precursors are emitted, such as methylglyoxal and glyoxal, which cannot produce many oxalate particles compared with the case in OUC [10][11][12]. The previously studied Hong Kong site was also beside a marine environment, but the studies were from February to April, during which solar radiation is weaker than that in Qingdao during summer; thus, the k of oxalate at the Hong Kong site was less than that at LQH. However, a previous study observed similar k values between the Hong Kong and OUC sites [6]. environment, but the studies were from February to April, during which solar radiation is weaker than that in Qingdao during summer; thus, the k of oxalate at the Hong Kong site was less than that at LQH. However, a previous study observed similar k values between the Hong Kong and OUC sites [6].

Conclusions
In this study, we conducted SPAMS measurements to analyze individual oxalate-containing and HM-containing particles in the summer of 2016 at a coastal site LQH and a suburban site OUC in Qingdao. Generally, 6.3% and 12.3% of the total particles (by number) at LQH and OUC were associated with oxalate, with 30.1% and 25.3% of oxalate-containing particles being associated with HMs, respectively. Oxalate-K-related particles (68.9% at LQH and 72.0% at OUC) and oxalate-carbonaceous-related particles (13.6% at LQH and 16.8% at OUC) were associated with BB during the study campaign. Fe-, Cu-, and Pb-rich particles in oxalate-HM-related particles were mainly from industrial emissions. The number of BB-related particles increased during continental air mass control at both sites.
The diurnal variation in oxalate-containing particles showed a decrease at LQH and an increase at OUC in oxalate PA after sunrise. However, the oxalate PA in Fe-rich and BB-Fe particles at both LQH and OUC decreased after sunrise, suggesting that Fe plays an important role in oxalate degradation. In summer, the intense solar radiation in Qingdao may cause higher k of Fe-rich (−0.978 h −1 ) and BB-Fe particles (−0.859 h −1 ) at LQH than in winter/spring in Hong Kong. Compared with LQH, the OUC site with heavy traffic and human activities had higher primary emissions and oxalate precursors, resulting in higher oxalate production during the early morning and a higher oxalate gen- Figure 5. Average decay estimation of oxalate intensity in different oxalate-Fe complexes. (Plotted by particle PA of m/z −89; box and error bars represent the 10th, 25th, 50th, 75th, and 90th percentiles; and the red diamond represents the mean).

Conclusions
In this study, we conducted SPAMS measurements to analyze individual oxalatecontaining and HM-containing particles in the summer of 2016 at a coastal site LQH and a suburban site OUC in Qingdao. Generally, 6.3% and 12.3% of the total particles (by number) at LQH and OUC were associated with oxalate, with 30.1% and 25.3% of oxalate-containing particles being associated with HMs, respectively. Oxalate-K-related particles (68.9% at LQH and 72.0% at OUC) and oxalate-carbonaceous-related particles (13.6% at LQH and 16.8% at OUC) were associated with BB during the study campaign. Fe-, Cu-, and Pb-rich particles in oxalate-HM-related particles were mainly from industrial emissions. The number of BB-related particles increased during continental air mass control at both sites.
The diurnal variation in oxalate-containing particles showed a decrease at LQH and an increase at OUC in oxalate PA after sunrise. However, the oxalate PA in Fe-rich and BB-Fe particles at both LQH and OUC decreased after sunrise, suggesting that Fe plays an important role in oxalate degradation. In summer, the intense solar radiation in Qingdao may cause higher k of Fe-rich (−0.978 h −1 ) and BB-Fe particles (−0.859 h −1 ) at LQH than in winter/spring in Hong Kong. Compared with LQH, the OUC site with heavy traffic and human activities had higher primary emissions and oxalate precursors, resulting in higher oxalate production during the early morning and a higher oxalate generation rate than the decay rate. Low liquid water content and solar radiation may also induce lower k of Fe-rich and BB-Fe particles: −0.512 and −0.178 h −1 , respectively.
Our results highlight the importance of involving transition metals and their chemical processes with oxalate or other low-molecular-weight organic acids in air quality and global climate models. Oxalate is an effective CCN candidate to form cloud condensation nuclei and ice nuclei due to its high-water solubility, affecting the climate [3][4][5]. Iron-mediated photodegradation of oxalate is a crucial factor affecting its lifetime [10,22]. Recent models that do not include the iron-mediated photodegradation of oxalate usually overestimate the oxalate concentration, while models including this process usually underestimate the