Impacts of the “Coal to Gas” Policy on Rural Air VOC Level and Ozone Potentials in North China

A unique study was enacted during the heating season (HS) in 2020 and 2021 at a rural site in the Beijing-Tianjin-Hebei region to evaluate the policy impacts of “Coal to Gas” (CTG) on ambient volatile organic compounds (VOCs). A total of 58 VOCs in air and flue gas from wall-mounted gas stoves (WMGS) were concurrently analyzed. The total VOCs decreased from 38.6 µ g m –3 in 2020 to 32.8 µ g m –3 in 2021, indicating the CTG played a positive role. However, the ozone formation potentials (OFPs) increased from 31.5 to 44.9 µ g m –3 . Toluene, vinylidene chloride, ethylbenzene, o, m, p-xylene, 4-methyl-2-pentanone, n-butylbenzene, trans-1,2-dichloroethylene, and 1,2,4-trimethylbenzene were the main contributors to the OFPs. Halohydrocarbons contributed the most to ∑ 58 VOCs of 54.8% and 54.4% in 2020 and 2021, respectively. It should be noted that the sustained CTG made WMGS the largest VOC source, replacing coal combustion (CC) in 2020. The CC contributions decreased from 33.2% in 2020 to 28.7% in 2021, while the WMGS far increased from 22.5% to 35.6%. Potential source contribution function (PSCF) modelling showed that the WMGS originated mainly from local emissions. High VOCs appeared surprisingly in clean days, because the WMGS and advanced coal-burning stoves with low particle emission prevailed in heating modes. The recognition of WMGS was achieved by coefficients of correlation and divergence between the positive matrix factorization (PMF) identified factor and field measured profiles of WMGS. This study firstly evidenced that the use of WMGS was becoming a major VOC source in rural north China. Meanwhile, the coal combustion for heating was still serious in rural area despite the “Coal Prohibition” law. The study was expected to provide some novel strategies for further VOC control and air quality improvement in rural area.


INTRODUCTION
The Air Pollution Prevention and Control Plans (APPCPs) had been launched in the Beijing-Tianjin-Hebei (BTH) region since 2013, such as weeding out small coal-fired boilers, phasing out small factories, and control VOCs, and so on . Consequently, the main air pollutants drastically decreased in China (Zheng et al., 2018). For instance, the SO2, NOx, and PM2.5 decreased largely by 83. 2%, 42.9%, and 54.7% in Beijing during 2013(Pang et al., 2021Niu et al., 2021). However, new challenges emerged recently in ever increasing levels of O3, VOCs, and mass fraction of secondary organic aerosols (SOA) in PM2.5 Yang et al., 2022). The BTH region has become the most polluted area by O3 in China (Gong and Liao, 2019). O3 has been a pollutant of great concern after PM2.5 due to its negative effects in increasing premature mortality and reducing crop yields (Sun et al., 2021). Consequently, the coordinated control of PM2.5 and O3 is extremely necessary in China after the APPCPs execution Li et al., 2023).
As the precursors of O3 and SOA, VOCs play key roles in producing O3 and PM2.5 (Wu and Xie, 2018;Chen et al., 2022). Thus, VOC control is essential to cut down both of them simultaneously, and identifying the VOC anthropogenic sources and quantifying their contributions are crucial to formulate the control strategies Li et al., 2020;Pinthong et al., 2022). Marked variations in VOC emissions have arisen in recent China in the context of the implementation of APPCPs, indicating that it is paramount to introduce new policies to enable further VOC control (Zheng et al., 2018;Li et al., 2019;Fan et al., 2021;Dao et al., 2022). Meanwhile, significant air pollutant reductions in urban areas with the APPCPs had given prominence to rural emissions (Zheng et al., 2018;Gao et al., 2020). Much evidence accumulated recently indicated that the emissions from residential fuel combustion in rural areas contribute largely to air pollution in China, and these effects were being magnified along with the APPCPs in major cities (Liu et al., 2016b;Liu et al., 2017;Shen et al., 2019). In realizing this crisis, a "Clean Heating" plan was initiated in 28 cities (Beijing, Tianjin, and the other 26 cites in surrounding areas) in 2017, which was aimed at replacing coal by either natural gas (NG) or electricity by 60% for rural households to the end of 2021 (Meng et al., 2020;Zhao et al., 2020). This has led to the drastic variations of VOC levels and sources . Increasingly prominent rural emissions and energy evolution made the research on rural VOC characteristics more and more necessary.
However, the most available studies were mainly about VOC levels and sources in urban area, and the research on changes of VOC emission strength and levels in rural area still remains rare Li et al., 2020). Former studies about urban VOC sources were conducted primarily on vehicle exhaust (Song et al., 2018), industrial production, fuel combustion and evaporation, solvent utility, etc. (Song et al., 2019;Zheng et al., 2020). In addition, the related researches were mainly about the cities in the North China Plain (NCP) (Liu et al., 2016a;Li et al., 2020). For the NCP villages, particularly during heating season, the source of VOCs was susceptible to coal and biomass burning (Shi et al., 2020). However, it is of interest to determine the VOC emission and contribution of natural gas (NG) burning when the coal-fired stoves are replaced by NG-burning wall-mounted gas stoves (WMGS). In a word, a systematic study on rural VOC characteristics with "Coal to Gas" (CTG) project is very imperative for potentiating the further control measures (Meng et al., 2020).
Wangdu County, a large agricultural county, suffers from large amounts of pollutants caused by coal and biomass combustion . Recent researches had focused on VOC characteristics at pollution levels, ozone formation potentials, and health risks in Wangdu Zhang et al., 2021;Xie et al., 2021). However, the rural CTG impacts on VOCs remained unclear. Therefore, a field measurement of ambient VOCs and VOCs discharged from WMGS in Wangdu County during the heating season in 2020 and 2021. The main objectives are to: (1) evaluate variations in the composition profiles and sources of 58 VOCs with CTG; (2) determine the VOC source evolutions and contribution of WMGS; and (3) assess the variations of ozone formation potentials.

Sampling Site Description
A rural site (38.63°N; 115.07°E) in Gudian Town was selected as the sampling site, surrounding by rural villages within 15 km radius (Fig. 1). Wangdu County covers an area of about 374 km 2 and the 90% of population is engaged in agricultural production. The heating facilities have converted from the traditional coal-fired stoves and small boilers to the wall-mounted natural gas stoves (WMGS) for most rural residents (Meng et al., 2020).
Three VOC samples were collected per day with specific sampling periods of 8:00-9:00, 12:00-13:00, and 18:00-19:00. The sample duration was from January 12 th to 31 st , 2021, and from November 15 to December 10, 2021. A total of 138 ambient samples and 11 wall-mounted gas stove (WNGS) samples. The WMGS commonly runs between 42 and 60°C. Teflon bags (Tedlar, Du Pont Co.) were 3 of 14 Volume 23 | Issue 9 | 230136 used for sample collection based on the HJ1006-2018 (Ministry of Ecology and Environment of China). Each bag was ventilated at least three times using sample gas prior to sampling.

Sample Preparation and Analysis
The samples were analyzed using a GC (8890)/MSD (5977B) system coupled with a DB-5 column (60 m × 320 µm × 1 µm). Each sample was analyzed for VOCs within 24 h. We consulted HJ644-2013 and analyzed the VOCs using an external standard method. A mixed standard solution (Catalog No. M-502; AccuStandard, USA) was diluted by methanol to obtain the standard calibration curves. The information of specific species were shown in Table S1. The most VOCs correlated well with the calibration curves with linear coefficients above 0.99. Meanwhile, we adjusted the calibration curves every 10 samples. The changes in results should be less than ± 20% compared with the standard values in the curves.
Helium was used as carrier gas at 1.5 mL min -1 , and the splitless injection was adopted. The inlet temperature was 170°C and the chromatographic column temperature increased from 35°C to 140°C (hold for 5 mins) at a rate of 6°C min -1 , and elevated to 220°C (hold for 3 mins) at 15°C min -1 . The ionization selective mode of mass spectrum was adopted with electron energy as 70 eV, and the temperature of ion source and interface were 230°C and 280°C, respectively. Table S1 presented the m/Z values for specific VOCs. An external standard method was used in the sample analysis. A total of 58 VOCs were detected, including 16 aromatics, 37 halocarbons, 4 oxygenated VOCs (OVOCs), and 1 sulfide (carbon disulfide). The strict procedures about quality control and assurance were carried out. All the Tedlar bags were not reused. A system blank experiment was done every ten analyzed samples or after analyzing samples with high concentrations. A parallel sample was also measured every ten samples. The relative deviations (RDs) should be less than or equal to 30%. We checked the reason and reanalyzed the sample when RD was higher than 30%.

Evaluation of the Source Similarity and the Indicative VOCs of WMGS
The coefficient of divergence (CD) was utilized to discriminate the similarities of VOC source profiles between PMF factors and WNGS samples. More details about the CD calculation and related threshold values for similarity discrimination can be found in Li et al. (2018).
The indicative VOCs were used in source apportionment, while the inconsistent indicators for different sources were frequently reported (Yuan et al., 2012). In this study, a formula was used to identify the indicators and shown as follows: More details about information of symbols can be found in Li et al. (2018).

Source Identification by Positive Matrix Factorization (PMF) Model
PMF 5.0 version was employed to quantify the potential VOC sources. More details on PMF can be found in Li et al. (2023). The species with more than 25% missing samples or a high proportion (> 35%) of samples below than the method detection limits (MDLs) were excluded . In total, 29 VOCs were selected as model inputs (Table S2). Each test was conducted 20 times and the lowest Q/Qexpected (Qexp) value appeared when the factor number was six (Mo et al., 2017;Gao et al., 2018;Liu et al., 2020) (Fig. S1). Ultimately, 6 sources with Q(true)/Q(robust) = 1.08 were obtained (Fig. S1).

Ozone Formation Potential (OFP)
The maximum incremental reactivity coefficient (MIR) method is widely employed in OFP evaluation (Hui et al., 2018;Han et al., 2021). OFP is defined as follows: where OFP(i) indicates the ozone formation potential (µg m -3 ), i is the species, and the MIRcoefficient is MIR value (g O3 g -1 VOCs). Carter (2010) provided the MIR values for 34 VOCs in this study (Table S1).

Characteristics of Ambient VOCs and O3
Fig. S1 showed the field measured VOCs and online monitoring O3 (https://aqicn.org/map/ world/cn/) acquired from Wangdu Environmental Monitoring Station. The mean O3 concentration (21.4 µg m -3 ) in the heating season (HS) of 2020 was similar to 20.5 and 15.1 µg m -3 in the HS of 2016 and 2017 in Beijing Han et al., 2021), and far lower than 37.8 µg m -3 of Nanjing in autumn . The relatively low O3 in winter was likely related to the emission reductions on VOCs and NOx, and the suppression of photochemical reaction by low temperature, high humidity, and weak solar radiation (Pu et al., 2017). The O3 peaks frequently lagged 1 or 2 days compared with the VOC peaks, indicating the VOC-limited formation process of O3 . Table S3 listed the top ten VOCs during two periods and the sum of them contributed 84.4% and 32.1% to the total VOC (TVOC) concentration. There were significant variations between 2020 and 2021 due to the emission variations. Fig. 2(a) showed the daily average VOC composition at different times. TVOC trend in 2020 was in agreement with that in 2021, which exhibited an increasing trend in both periods. Also, this trend was consistent with that of the rural North China Plain . Halohydrocarbons contributed most to the TVOCs, with the fractions of 51.3%, 51.9%, and 54.6%, and 53.4%, 55.6%, and 54.0% for 8:00, 12:00, and 20:00, respectively, in two periods. The lowest aromatics appeared at noon in both periods, indicating the decreasing coal consumption used for heating purpose due to relative high temperature. In contrast, OVOCs exhibited the highest levels at noon, which might be related to the promoting effect of formation reactions of OVOCs by high temperature Xie et al., 2021). The high traffic volume during morning peak hours also enhanced the acetone level or concentration. Subsequently, the acetone accumulated at noon (Liu et al., 2017).

VOCs at Different Pollution Levels
The VOC variations under different PM2.5 levels were shown in Fig. 2(b). The clean day (C), moderately polluted day (MP), and heavily polluted day (HP) refer to the days with PM2.5 concentrations of < 75 µg m -3 , 75-150 µg m -3 , and ≥ 150 µg m -3 , respectively. A large divergence in VOC homolog fractions in the different PM2.5 levels emerged between 2020 and 2021.  reported that the TVOCs increased with the rising PM2.5 in 2018 at a same sampling site adopted in this study. The results in this study indicated that no similar trends appeared in both 2020 and 2021. The TVOC peaks occurred in the C and the MP in 2020 and 2021, respectively. No TVOC peaks in the HP period in 2020 and 2021 indicated that PM2.5 decreased more than VOCs due to the implementation of "Coal to Gas" (CTG) policy (Li et al., 2023). Sulfur compounds increased in the HP of 2021 compared with 2020 evidenced the CTG impacts. This would be described in detail in the section of source apportionment. Factor 1 was distinguished by 4-methyl-2-pentanone, 2-hexanone, 1,2-dichloroethane, 1,2dichlorobenzene, benzene, toluene, ethylbenzene, m, p-xylenes, and o-xylene. Benzene and toluene are abundant in coal burning as relatively simple aromatics (Lyu et al., 2016;Hui et al., 2018). 1,2-dichloroethane is also considered as another typical marker from coal burning. Thus, Factor 1 was identified as CC. Factor 2 was characterized by mesitylene, 1,2,4-trimethylbenzene, and isobutylbenzene. 1,2,4-trimethylbenzene are markers of gasoline vehicle exhaust (Liu et al., 2008). Factor 2 could be explained as vehicle exhaust. Factor 3 was characterized by high loads of carbon tetrachloride, o-and 4-chlorotoluene, n-propylbenzene, and p-isopropyltoluene. The carbon tetrachloride and other chlorinated VOCs are more emitted from the chlor-alkali process of petrochemical complexes (IPPC, 2001). Consequently, Factor 3 was attributed to the petrochemical industry. Factor 4 was highly enriched with dichloromethane, 1,2-dichloropropane, and perchloroethylene, which was identified as solvent use (Guo et al., 2009;Han et al., 2019). and few kinds of aromatics (toluene, cumene, p-isopropyltoluene, and n-butylbenzene) and recognized as industrial processes (Zou et al., 2016;Hui et al., 2018). Factor 6 was dominated by carbon disulfide (CS2). Though CS2 is an important product of the rubber industry (Han et al., 2011), no rubber-associating factories existed near the sampling site. Hence, we attributed Factor 6 to WMGS, which was further discussed in Section 3.3.2.

Identification of WMGS by the related markers
Similarity between the emission profile for WMGS and factor 6 were evaluated using the coefficient of divergence (CD) . The CD value of 0.3 was widely used to identify the similarity. Fig. 4 showed the CD value for 29 VOCs between the WMGS and factor 6 was 0.268, indicating their similarity. This was in accordance with the sustained increase in natural gas utility in rural area (Li et al., 2023). In addition, the VOC markers of WMGS were recognized by a parameter of ratioj,i . Carbon disulfide and m-xylene were the indicative VOCs. Thus, this study provided direct emission evidence of wall-mounted gas stoves for carbon disulfide.  Coal combustion (CC) was a dominant source in the heating season, which was consistent with the previously reported results in Wangdu County Xie et al., 2021). CC (coal combustion) was the largest contributor in 2020, indicating that CC was still an important VOC source in rural area without lots of industries . With the sustained "Coal to Gas" policy and increasing coal price in 2021, CC fractions decreased from 33.2% in 2020 to 28.7% in 2021. As a result, the WMGS contributed most to the TVOCs of 35.6% in 2021. CC has become the second largest source in 2021. The contributions of solvent use (SU), industrial processes (IP), and vehicle exhaust (VE) in this study were far lower than those in urban areas (Mo et al., 2017;Hui et al., 2020;Fan et al., 2021).

Potential source contribution function (PSCF) analysis
We used a PSCF model to evaluate the potential regions of six sources. More details about the PSCF model were illustrated in Li et al. (2023). Fig. S2 showed the PSCF result for six sources. Fig. 6 illustrated the potential source regions for WMGS and CC. The source region of WMGS was narrower than that of CC, implying that the WMGS was more inclined to local emissions. However, CC's source region originated mainly from rural area of surrounding counties and local Wangdu County. In a word, the natural gas combustion in WMGS was becoming an important VOC source in rural area with the "Coal to Gas" policy.

Ozone Formation Potentials (OFPs) of VOCs
We calculated the OFPs for 34 VOCs based on the available maximum incremental reactivity (MIR) coefficients (Fig. 7). Though the total VOC concentration in 2020 were higher than 2021, whereas higher OFPs were found in 2021. The total OPFs were 31.5 and 44.9 µg m -3 in 2020 and 2021, respectively. Halohydrocarbons were predominant VOCs, however, they possessed low OFPs due to their low MIRs. The main chemical bonds of halohydrocarbons were stronger and more stable (Kumar et al., 2018). The aromatics contributed most to the total OFPs of 68.1% and 79.8% in 2020 and 2021 in term of high MIRs. Ou et al. (2015) reported a similar result for OFP fractions in the Pearl River Delta region. Furthermore, the higher OFPs and lower TVOC concentrations in 2021 compared with 2020 indicated that VOCs with high MIRs should be more concerned in term of O3 control. Fig. 5(b) showed the source contributions to the total OFPs in both periods. Compared with the apportionment result for TVOCs, the OFP contribution of WMGS exhibited a smaller increase in 2021 relative to 2020. The CC contribution in both periods were comparable. Industrial processes and petrochemical industry contributed more to the total OFPs than to the TVOCs. In conclusion, controlling the VOC emissions from wall-mounted gas stoves would benefit the reductions for both TVOCs and OPFs.

CONCLUSIONS
We accomplished a field observation on ambient VOCs and VOCs released from the wall-mounted gas stoves were observed at a rural site in North China in the heating seasons of 2020 and 2021. Our main starting points were to examine the policy impacts of "Coal to Gas" (CTG) on VOC evolution to enlighten the improvement of VOC control measures.
The total VOC (TVOC) concentration decreased by 15.0% in the heating season (HS) of 2021 compared with that of 2020, indicating the "Coal to Gas" policy impacts. At the same time, however, the ozone formation potentials (OPFs) increased by 42.5%. Wall-mounted gas stoves (WMGS) Volume 23 | Issue 9 | 230136 has been an important VOC source in rural area without industries. WMGS contributed most to the TVOCs of 36.5% and to the OFPs of 25.3% in 2021, respectively. WMGS was also an important origin of CS2. The potential source region of WMGS originated from local emissions. The contributions of coal combustion (CC) to TVOCs and OFPs decreased by 13.7% and 7.43% in the HS of 2021 compared to 2020. However, coal combustion was still a dominated contributors despite that the implementation of CTG policy. In a word, the comprehensive utility of different energy would be an effective way for both the VOC control and cost reduction of rural residents.