Characteristics, sources, and reactions of nitrous acid during winter at an urban site in the Central Plains Economic Region in China

Abstract. Nitrous acid (HONO) in the core city of the Central Plains Economic Region in China was measured using an ambient ion monitor from 9 to 31 January 2019.
Measurement time intervals were classified into the following periods in
accordance with the daily mean values of PM2.5: clean days (CDs),
polluted days (PDs), and severely polluted days (SPDs). The HONO
concentrations during CD, PD, and SPD periods were 1.2, 2.3, and 3.7 ppbv,
respectively. The contributions of the homogeneous reaction, heterogeneous
conversion, and direct emissions to HONO sources varied under different
pollution levels. The mean values of the net HONO production of the
homogeneous reaction (POH+NOnet) in CD,
PD, and SPD periods were 0.13, 0.26, and 0.56 ppbv h−1, respectively.
The average conversions of NO2 (CHONO) in CD, PD, and SPD periods
were 0.72×10-2, 0.64×10-2, and 1.54×10-2 h−1, respectively, indicating that the heterogeneous
conversion of NO2 was less important than the homogeneous reaction.
Furthermore, the net production of the homogeneous reaction may have been
the main factor in the increase of HONO under high-NOx conditions
(i.e., when the concentration of NO was higher than that of NO2) at
nighttime. Daytime HONO budget analysis showed that the mean values of the
unknown source (Punknown) during CD, PD, and SPD periods were 0.26,
0.40, and 1.83 ppbv h−1, respectively. The values of
POH+NOnet, CHONO, and Punknown
in the SPDs period were comparatively larger than those in other periods,
indicating that HONO participated in many reactions. The proportions of
nighttime HONO sources also changed during the entire sampling period.
Direct emissions and a heterogeneous reaction controlled HONO production in
the first half of the night and provided a contribution that is larger than that of
the homogeneous reaction. The proportion of homogenization gradually
increased in the second half of the night due to the steady increase in NO
concentrations. The hourly level of HONO abatement pathways, except for OH + HONO, was at least 0.22 ppbv h−1 in the SPDs period. The cumulative
frequency distribution of the HONOemission∕HONO ratio (less than
20 %) was approximately 77 %, which suggested that direct emission was
not important. The heterogeneous HONO production increased when the relative
humidity (RH) increased, but it decreased when RH increased further. The
average HONO∕NOx ratio (4.9 %) was more than twice the assumed
globally averaged value (2.0 %).



Introduction
Nitrous acid (HONO) is important in the photochemical cycle and can provide hydroxyl radicals ( q OH; Harrison et al., 1996) as follows: HONO + hv → q OH + NO (300 nm < λ < 405 nm). (R1) According to measurement and simulation studies (Alicke et al., 2002), the contribution of HONO to q OH concentrations can reach 25 %-50 %, especially when the concentration of OH radicals produced by the photolysis of ozone, acetone, and formaldehyde is relatively low (2-3 h after sunrise; Czader et al., 2012). HONO photolysis was the most important primary source of q OH that contributed up to 46 % of the total primary production rate of radicals for daytime conditions (Tan et al., 2018). q OH is an important oxidant in the atmosphere, and it can react with organic substances, control the oxidation capacity of the atmosphere, and accelerate the formation of secondary aerosols in the urban atmosphere (Sörgel et al., 2011). Therefore, the changes in the contribution of the homogeneous reaction, heterogeneous conversion, and direct emissions during pollution can be observed by studying the formation mechanism of HONO.
Several instruments have been used to determine ambient HONO concentrations, and these include the following: a differential optical absorption spectrophotometer (DOAS; Elshorbany et al., 2012;Winer and Biermann, 1994); a long path absorption photometer (LOPAP; Heland et al., 2001); a wet chemical derivatization technique (HPLC) coupled with UV-VIS detection (collectively HPLC/UV-VIS; Michoud et al., 2014); a stripping coil-UV/VIS absorption photometer (SC-AP; Pinto et al., 2014); an incoherent broadband cavityenhanced absorption spectroscope (IBBCEAS; Duan et al., 2018); a chemical ionization mass spectrometer (CIMS; Hirokawa et al., 2009;Roberts et al., 2010); and an ambient ion monitor (AIM;VandenBoer et al., 2014). A result comparison of the different instruments showed that the SC-AP is compatible with two spectral measurement instruments, namely LOPAP and DOAS (Pinto et al., 2014). Compared to HONO measured by the SC-AP deployed on site, HONO measured by AIM has a small error and is within the acceptable analytical uncertainty (VandenBoer et al., 2014). Previous studies have reported that HONO concentrations range from a few parts per trillion by volume (pptv) in clean, remote areas to several parts per billion by volume (ppbv; 0.1-2.1 ppbv) in air-polluted urban areas Michoud et al., 2014).
The sources of HONO are direct emissions and homogeneous and heterogeneous reactions (Acker et al., 2005;Grassian, 2001;Kurtenbach et al., 2001). HONO can be directly discharged into the atmosphere during vehicle operations and biomass combustion. Through a tunneling experiment, Kurtenbach et al. (2001) discovered that motor vehicles emit a small amount of HONO, and the HONO/NO x ratio of HONO combustion sources (aside from NO x and other pollutants) is 0.1 %-0.8 %. Another study showed that the homogeneous reaction of NO and OH radicals is a major source of HONO under increased NO concentrations (Spataro et al., 2013). Furthermore, HONO can react with the q OH (Alicke et al., 2003;Vogel et al., 2003). Tong et al. (2015) used NO+OH and HONO+OH homogeneous reactions to calculate the net generation rate of HONO homogeneous reactions at night, which are expressed as follows: Such calculations have been applied in studies on homogeneous reactions and daytime budgets Huang et al., 2017). These are studies of homogeneous reactions, and some researchers have begun to explore the mechanism of NO 2 heterogeneous reactions.  studied the mechanism of chemical adsorption of NO 2 , and H ions were revealed on the adsorbed surface by using isotope-labeled water as follows: In China, most studies of HONO have been focused on the Yangtze River Delta, Pearl River Delta, and Jing-Jin-Ji region. For example, Hao et al. (2006) reported that field measurement results, especially HONO/NO 2 and relative humidity (RH), have a significant correlation and proved that heterogeneous reactions are an important source of nighttime HONO. Although the specific chemical mechanisms of heterogeneous reactions remain unknown, the intensity of HONO formation by NO 2 can be expressed by the HONO conversion frequency (Alicke et al., 2002;Li et al., 2012). Su et al. (2008a) revealed the importance of the q OH from HONO during daytime (09:00-15:00 local time or LT) and found that many unknown sources that are closely related to the solar radiation lead to HONO formation. The unknown sources of HONO may include the NO 2 photolysis of sooty surfaces and adsorbed nitric acid and nitrate at UV wavelengths (Kleffmann et al., 1999). The homogeneous nucleation of NO 2 , H 2 O, and NH 3 is the HONO-formation pathway (Zhang and Tao, 2010). Meanwhile, HONO can deposit and react with amines to form nitrosamines  for sinking. The method of budget analysis needs to include the HONO sources and sinks. Researchers suggested that the method of budget analysis is crucial for obtaining the missing source. Spataro et al. (2013) measured the HONO level in Beijing's urban area and discussed the spatiotemporal changes, meteorological effects, and contributions of HONO from different sources. They used the measured HONO data to compare pollution periods in Beijing's urban and suburban areas. Tong et al. (2015) discovered that the pathway of the HONO-formation mechanism, namely direct emissions, heterogeneous formation, and homogeneous reaction, is the same, but the pathway is different at the two sites. A few studies (Cui et al., 2018;Hou et al., 2016) compared the characteristics and sources of HONO during severe pollution periods and clean periods. Although the definitions of the two periods are different, both can be used to analyze the diurnal variation, source, and daytime budget of HONO during the aggravation of pollution.
There is no study of HONO in the Central Plains Economic Region (CPER) in China, with a total population of 0.18 billion at the end of 2011. CPER is an important region for food production and modern agriculture, as published by the Chinese government (http://www.gov.cn/ zhengce/content/2011-10/07/content_8208.htm, last access: 10 September 2019). The document described the different factors that affect atmospheric pollution, including the level of economic development, energy structure, industrial structure, and geographical location (solar radiation) in the Yangtze River Delta, Pearl River Delta, and Jing-Jin-Ji region. As the core city of CPER, Zhengzhou is characterized by severe PM (particulate matter) pollution (Jiang et al., 2017(Jiang et al., , 2018d and is thus selected for this study. In recent years, comprehensive PM research has been conducted on the chemical characteristics of PM in Zhengzhou (Jiang et al., 2018b; regarding source apportionment (Jiang et al., 2018c, e;Liu et al., 2019), health risks (Jiang et al., 2019a, b), and emission-source profiles Jiang et al., 2018a). However, no study has been performed on the sources and characteristics of HONO in Zhengzhou. Moreover, no synthetic research on different pollution levels in the area is available. In the current study, AIM was used to sample and analyze HONO concentrations. The interactions between HONO and other factors, such as PM 2.5 during pollution, were assessed to understand the formation, influence, and removal of HONO on different pollution periods. The levels of PM 2.5 were divided into three periods to analyze the HONO sources, sinks, and reactions in different periods. Many papers (Huang et al., 2017;Tong et al., 2016) took PM 2.5 as the main control factor of HONO and studied the differences between HONO sources and characteristics during clean and polluted periods. No homogeneous reaction, direct emission, heterogeneous reaction, or daytime budget analysis was conducted during the period of worsening pollution (named the HD period in this paper). Total NO x emissions in cities with different leading factors of emissions have been declining year by year due to the Chinese government's emission control measures, but some Chinese cities are still in high-NO x areas (e.g., Beijing, Shanghai, Guangzhou, and Zhengzhou; Kim et al., 2015;Liu et al., 2017). Under high-NO x conditions, some papers (Cui et al., 2018;Hou et al., 2016) suggested that the heterogeneous reaction was the main source of HONO and did not conduct a quantitative analysis of homogeneous reactionespecially in winter. So, we explore relevant studies of homogeneous reactions. In addition, the source contributions of HONO that varied with the degree of pollution at night were not explained. RH was also analyzed to provide a detailed understanding of HONO-generation intensity under different RH conditions. Analysis of the sources of HONO at night provides strong support for conducting HONO budget analysis during the daytime. To the best of the authors' knowledge, the formation characteristics of HONO at continuous and high time resolutions and different pollution levels have not been studied in Zhengzhou. This work can assist the governments of the CPER in formulating a policy to decrease the level of HONO precursors, i.e., NO and NO 2 , and direct HONO emissions from vehicles.

Sampling site and period
The sampling site is on the rooftop (sixth floor) of a building at Zhengzhou University (34 • 48 N, 113 • 31 E), which is located in the northwestern part of Zhengzhou, China. The observation height is about 20 m from the ground, and the observation platform is relatively open and without any tall buildings around. The site is about 500 m from the West Fourth Ring Road of Zhengzhou and about 2 km from the Lian Huo Expressway to the north. The measurement period was from 9 to 31 January 2019. Daily data were divided into two periods, namely daytime (07:00-18:00 LT) and nighttime (19:00-06:00 LT, i.e., the morning of the next day).

Instruments
AIM (URG-9000D, Thermo Fisher Scientific, USA), an online ion chromatographic monitoring system for particle and gas components in the atmosphere, was used to measure HONO concentration continuously at a temporal resolution of 1 h. The atmospheric airflow entered the PM 2.5 cyclone cutting head through the sample tube, and gas-solid separation was performed with a parallel plate denuder with a new synthetic polyamide membrane. The denuder had no moving parts and could be changed without stopping the sampler. HONO was absorbed by the denuder with an absorption liquid (5.5 mol m −3 H 2 O 2 ). The chemicals that could be oxidized were absorbed by H 2 O 2 on the porous membrane surface, but several gases (e.g., O 2 and N 2 ) were expelled by the air pump. The abundance of other gaseous acids and bases affected the efficiency of HONO collection by AIM due to the relationship between Henry's law constant and pH. This measurement method and its details have been successfully evaluated in many field studies (Markovic et al., 2012;Wang et al., 2019;Yang et al., 2020), as shown in the Supplement. In addition, a QXZ1.0 automatic weather station (Yigu Technologies, China) was used for the synchronous observation of meteorological parameters, including temperature (T ), RH, wind direction (WD), and wind speed (WS). The temporal resolution of the model analysis system that integrates these types of observation instruments (TE -used for measuring O 3 ; 48i -used for measuring CO; 42i -used for measuring NO, NO x , and NO 2 ]; and TEOM 1405 PM 2.5 monitorused for measuring PM 2.5 ; Thermo Fisher Scientific, USA) is 1 h. Detailed information can be found in the work of Wang et al. (2019). Measurement technique, detection limit, and accuracy of measured species are shown in Table S1 in the Supplement.
During the sampling period, all instruments were subjected to strict quality controls to avoid possible contamination. The instrument accessories and sampling process were periodically replaced and calibrated, respectively. The instrument parts and consumables were changed before the observation process, and the sampling flow was calibrated to reduce the negative effect of the accessories. Before this measurement period, the membrane of the denuder was replaced and standard anion and cation solutions were prepared on 3 January. The standard curve should be drawn to ensure the appropriateness of the correlation coefficient (≥ 0.999) and the accuracy of the sample retention time and response value. The minimum detection limit of AIM was 0.004 ppbv. Other detailed information can be found in the work of Wang et al. (2019).

Temporal variations of meteorological parameters and pollutants
The daily changes in the meteorological parameters and PM 2.5 are shown in Fig. 1. In accordance with the daily average concentration level of PM 2.5 , the analysis and measurement process was divided into three periods (clean days -CDs, polluted days -PDs, and severely polluted days -SPDs). The days on which the daily averages of PM 2.5 were lower than the daily average of the second grade of the Chinese National Ambient Air Quality Standards (CNAAQS; 75 µg m −3 ) represented CDs (9,16,17,21,22,23,26, and 31 January), with RH ranging from 5 % to 79 % and WS ranging from 0 to 4.2 m s −1 . The days on which the daily averages of PM 2.5 were between 75 and 115 µg m −3 represented PDs (10,15,18,20,25,27, and 28 January), with RH ranging from 17 % to 86 % and WS ranging from 0 to 4.6 m s −1 . The days on which the daily averages of PM 2.5 were higher than 115 µg m −3 represented SPDs (11,12,13,14,19,24,29, and 30 January), with RH ranging from 30 % to 96 % and WS ranging from 0 to 3.5 m s −1 . Northwesterly or easterly wind was observed in most of the observation periods, except for 21-22 January. WD was northerly, the maximum WS reached 4 m s −1 , the PM 2.5 concentration decreased rapidly, and the effect of pollutant removal was evident. Table 1 lists the data statistics of HONO, PM 2.5 , NO 2 , NO, NO x , HONO/NO 2 , HONO/NO x , O 3 , CO, T , RH, WS, and WD during the measurement period, together with the mean value ± standard deviation. The meteorological parameters in Table 1 show that the average RH in CD, PD, and SPD periods was 33 %, 49 %, and 68 %, respectively. In SPDs, RH was high and WD was low (mean value of 0.4 m s −1 ).
In accordance with the data on trace gases, the average HONO values in CD, PD, and SPD periods were 1.1, 2.3, and 3.7 ppbv, respectively. The mean values of NO 2 were 25, 33, and 42 ppbv (46, 63, and 78 µg m −3 lower than the first grade of the CNAAQS; [80 µg m −3 ]), respectively. The mean values of CO were 1, 1, and 2 ppmv (1, 2, and 2 mg m −3 lower than the first grade of the CNAAQS; [4 mg m −3 ]), respectively. Figure 2 shows the concentration changes in HONO and gas species throughout the measurement period. The variations of the average HONO, PM 2.5 , NO 2 , and CO in the three periods were similar. The mean values of all pollutant concentrations, except O 3 , in the SPDs period were the largest, and those values in the CDs period were the smallest. The highest mean value of O 3 occurred in the CDs period, which is similar to previous observations Huang et al., 2017;Zhang et al., 2019).
The HONO concentrations ranged from 0.2 to 14.8 ppbv and had an average of 2.5 ppbv, which is higher than the average values of 0.6 (Rappenglück et al., 2013), 1.5 , and 1.0 ppbv (Huang et al., 2017) in previous urban studies. The diurnal variations of HONO during the measurement were similar in the three periods, as shown in Figs. 3 and 4. The diurnal variations of HONO, NO, NO 2 , O 3 , HONO/NO 2 , and HONO/NO x are illustrated in Fig. 4. The error bars of Fig. 4 were placed separately in the tables of the Supplement (Table S2). After sunset, the HONO concentrations in CD, PD, and SPD periods began to accumulate due to the attenuation of solar radiation and the stabilization of the boundary layer (Cui et al., 2018). The maximum values of 1.7, 4.1, and 6.9 ppbv were reached in the morning (08:00-10:00 LT) in CD, PD, and SPD periods, respectively. After 10:00 LT, the HONO concentration decreased because of the increased solubility and rapid photolysis, which remained at a low level before sunset (14:00-16:00 LT). The NO concentration decreased rapidly in the forenoon and remained low in the afternoon. After sunset, the concentrations of NO and NO 2 began to increase and remained at a higher level than during the daytime. Furthermore, the diurnal variation of NO in the CDs period was similar to that of NO 2 . The peak was reached at around 09:00 LT due to vehicle emissions in the morning rush hours, and the lowest value was observed at around 16:00 LT. After 18:00 LT, the boundary layer height decreased in the evening rush hours, resulting in an increase in NO and NO 2 concentrations (Hendrick et al., 2014). O 3 showed a diurnal cycle and had maximum values in CD, PD, and SPD periods in the afternoon. The HONO/NO 2 ratio is commonly used to estimate the formation of HONO in NO 2 transformation (Wang et al., 2013). Compared to HONO formation, NO 2 transformation is less affected by the migration of atmospheric air mass during atmospheric migration . The HONO/NO 2 ratio in the CDs period began to increase after sunset and reached its peak at night. Then, it decreased in the morning as a result of the enhancement of NO 2 emissions and photolysis of HONO. However, the mean value of HONO/NO 2 in PD and SPD periods gradually increased from the nighttime level and eventually reached the maximum values of 14.3 % and 18.9 % at 09:00 and 10:00 LT, respectively. The average HONO/NO x ratio (4.9 %) was more than twice the assumed globally averaged value (2.0 %; Elshorbany et al., 2014). This result indicates that the strength of the heterogeneous reaction increased slightly with the exacerbation of pollution. The HONO/NO 2 ratio showed a diurnal cycle with a low level in the afternoon and a high level after sunset due to the heterogeneous reaction of NO 2 on the ground and aerosol surfaces (Su et al., 2008b). For comparison, the daytime and nighttime HONO, HONO/NO 2 , and HONO/NO x mean values in other cities around the world are listed in Ta  21.4 ± 11.5 13.8 ± 10.0 17.6 ± 11.4 17.4 ± 11.9 8.9 ± 8.1 13.1 ± 10.9 15.6 ± 14.2 7.9 ± 7.1 11.8 ± 11.8 14.2 ± 11.7 HONO/NO 2 (%) 4.2 ± 3.6 5.3 ± 2.2 4.7 ± 3.1 6.8 ± 5.8 7.4 ± 3.9 7.1 ± 4.9 9.0 ± 7.7 9.8 ± 5.8 9.4 ± 6.8 7.6 ± 6.4 HONO/NO x (%) 3.3 ± 2.7 6.0 ± 5.6 4.5 ± 4.5 4.4 ± 2.5 4.6 ± 1.7 4.5 ± 2.1 5.3 ± 3.4 5.8 ± 4.7 5.6 ± 4.1 4.9 ± 3.8 RH (%) 30 ± 21 36 ± 20 33 ± 21 44 ± 17 54 ± 18 49 ± 18 64 ± 18 73 ± 13 68 ± 16 50 ± 24 WS (m s −1 ) 0.8 ± 1.0 0.5 ± 0.7 0.7 ± 0.9 1.1 ± 1.4 0.6 ± 0.9 0.9 ± 1.2 0.4 ± 0.7 0.3 ± 0.6 0.4 ± 0.7 0.6 ± 0.9 T ( • ) 4.3 ± 4.6 2.7 ± 3.6 3.5 ± 4.2 3.7 ± 3.3 2.6 ± 3.1 3.1 ± 3.2 4.6 ± 3.2 2.9 ± 2.1 3.8 ± 2.8 3.5 ± 3.5 in Zhengzhou are relatively higher than those in other parts of the world. The reason for this phenomenon is that Zhengzhou is a high-NO x area that provides HONO with abundant precursors (NO 2 and NO) in winter (Kim et al., 2015).

Homogeneous reaction of NO and OH
The homogeneous reaction of NO and OH (reactions R2 and R3) is the main pathway of HONO formation in the gas phase. Spataro et al. (2013) found that the formation mechanism leads to an increase in HONO in high-pollution areas and an increase in NO at night. P net OH+NO can be understood as the net hourly HONO production amount of the homoge-neous reaction and is calculated as follows: At T = 298 K and P = 101 kPa, the rate constants of k OH+NO and k OH+HONO are 9.8 × 10 −12 and 6.0 × 10 −12 cm 3 molec. −1 s −1 , respectively (Atkinson et al., 2004;Sander et al., 2003).
[OH] is the concentration of q OH that was not measured during the campaign. Tan et al. (2018) found that, during the field measurement, the average concentration of q OH in Beijing at nighttime was about 2.5 × 10 5 molec. cm −3 . Moreover, the same q OH concentration was also used to calculate the homogeneous reaction of HONO in the recent research of Beijing , Shanghai (Cui et al., 2018), and Xi'an (Huang et al., 2017). Nighttime OH concentration increased as the latitude de-   creases ranged from 3 to 6 × 10 5 molec. cm −3 (Lelieveld et al., 2016). Zhengzhou has a lower latitude than Beijing, so the concentration of OH used in this study is 2.5 × 10 5 molec. cm −3 . P net OH+NO primarily depends on the concentrations of NO and HONO because the values of k OH+NO and k OH+HONO are close. Figure 5 shows the nocturnal variations of P net OH+NO , NO, and HONO during CD, PD, and SPD periods. The uncertainties of P net OH+NO , NO, and HONO in Fig. 5 are shown in Table S3. When the NO levels were high, the variations of P net OH+NO followed those of NO during the three periods (Atkinson et al., 2004). The mean value of P net OH+NO was 0.33 ppbv h −1 , and the specific values in CD, PD, and SPD periods were 0.13, 0.26, and 0.56 ppbv h −1 , respectively. We assumed ±50 % q OH values to estimate the uncertainty of P net OH+NO . The q OH values of 1.25 × 10 5 and 3.75×10 5 molec. cm −3 were calculated for the P net OH+NO values of 0.16 and 0.49 ppbv h −1 . P net OH+NO varied from 0.01 to 0.47 ppbv h −1 during the CDs period. The mean value of P net OH+NO increased before midnight, decreased after midnight, and increased slightly at 03:00 LT. In the PDs period, P net OH+NO ranged from 0.07 to 0.44 ppbv h −1 . The situation was similar to that of the CDs period, except that the value remained almost constant. In addition, the contribution of HONO from homogeneous reactions during the SPDs period was larger than those in the CD and PD periods, and the level of P net OH+NO , with an average value of 0.56 ppbv h −1 , was lower than the value in a previous study (2.18 ppbv h −1 in Beijing; Tong et al., 2015). From 19:00 to 03:00 LT, the mean value of P net OH+NO increased from 0.15 to 0.9 ppbv h −1 . HONO increased from 2.84 to 4.59 ppbv and subsequently decreased to 4.43 ppbv. By integrating P net OH+NO during the 8 h period, the homogeneous reaction can provide an accumulated HONO formation of at least 3.36 ppbv (i.e., 0.15 + 0.20 + 0.25 + 0.25 + 0.35 + 0.56 + 0.7 + 0.9 ppbv). However, the mean accumulation value of measured HONO in this nighttime period was merely 1.59 ppbv. With the increase in pollution levels, the HONO accumulation period at nighttime increased. This result indicates that, first, the homogeneous reaction of OH + NO is sufficient to augment HONO in the first half of the night, although NO 2 transformation and other sources may still exist. When the concentration of NO is relatively high, the net production generated by OH + NO may be the leading factor of the increase in HONO at night . Second, the hourly level of HONO abatement pathways, except OH + HONO, should be at least 0.22 ppbv h −1 (i.e., 3.36-1.59 ppbv (8 h) −1 ). This phenomenon may arise because the dry deposition on ground surfaces can be the main HONO removal pathway at night, which is similar to a previous study .

Direct emission
At present, no HONO emission inventory or emission factor database is available for Zhengzhou. As a result, estimating any HONO from direct emissions is difficult. In the current study, directly emitted HONO could have been generated by vehicle exhaust emissions and biomass combustion because the site is close to the West Fourth Ring expressway of Zhengzhou and Lian Huo Expressway in the north. Hence, only night data (17:00-06:00 LT) were considered to avoid the problem of instant photolysis of directly emitted HONO. In a previous study, the HONO/NO x ratio from tunnel measurement was set to 0.65 % to estimate an upper limit of HONO emitted by traffic near the site . The minimum value of HONO/NO x in the SPDs period in the current work was 1.5 %, which is slightly higher than the value measured in the abovementioned study. HONO emitted directly at night was not transformed immediately. The HONO concentrations corrected by direct emissions are given as follows: where [HONO] emission , [NO x ], and 0.0065 are the direct HONO emissions concentration, NO x concentration, and HONO/NO 2 direct emissions ratio, respectively. The direct emissions contribution was estimated by comparing the direct emission HONO with the observed HONO. The ranges of HONO emission /HONO in CD, PD, and SPD periods were 2 %-52 %, 6 %-34 %, and 2 %-41 %, respectively, and the mean values were 17 %, 16 %, and 16 %, respectively. The frequency distribution of the HONO emission /HONO ratio at nighttime is shown in Fig. 6. For this upper-limit estimation, the frequency distribution of HONO emission /HONO (less than 20 %) was approximately 77 %. Hence, direct emissions may not be the main reason for the high growth of HONO levels. Compared to the direct emissions of other sites, the emissions of the measurement site accounted for a lower proportion -possibly because the site is on the campus and relatively far from the highway.

Heterogeneous conversion of NO 2 to HONO
NO 2 is an important precursor for HONO formation. In addition, recent field measurements in many urban locations have shown that a positive correlation exists between HONO and NO 2 (Cui et al., 2018;Hao et al., 2006;Huang et al., 2017;Zhang et al., 2019), which suggests that they have a common source. Moreover, Acker et al. (2005) reported that different meteorological conditions may lead to significant differences in the relationship between the source and receptor, and these differences lead to various types of correlations. During the measurement period, the HONO/NO 2 ratio varied between 1.3 % and 59.0 %, with an average of 7.6 %, which is slightly higher than the averaged value of 6.2 % in a previous study   (Cui et al., 2018). The HONO/NO 2 ratio calculated in this work is much larger than the ratio calculated for direct emissions (< 1 %; Kurtenbach et al., 2001), suggesting that heterogeneous reactions may be a more important pathway for HONO production than direct emissions. With regard to the heterogeneous conversion of NO 2 , several studies Shen and Zhang, 2013) reported that the surface of soot particles is the medium of NO 2 conversion. The contribution of soot surface to HONO production is usually much lower than expected because the uptake efficiency of NO 2 decreases with the prolonged reaction time caused by surface deactivation. The aerosol surface is an important medium for the heterogeneous transformation from NO 2 to HONO (Liu et al., 2014). The mass concentration of aerosols was used as an alternative to identify the influence of aerosols in this study because the surface density of aerosols could not be obtained.
The correlations between PM 2.5 and the HONO/NO 2 ratio in CD, PD, and SPD periods are shown in Fig. 7. With the exacerbation of the PM 2.5 level, the average value of HONO/NO 2 gradually increased, indicating that the aerosol surface occupied an important position in the heterogeneous transformation. A comparison of HONO/NO 2 and HONO with PM 2.5 showed that the correlation between HONO/NO 2 and PM 2.5 (R 2 = 0.23) was weaker than that between HONO and PM 2.5 (R 2 = 0.55) during the entire period. The main source of HONO could not have been the transformation of NO 2 . Notably, the HONO correlation in the PDs period was significantly stronger than that of the other two periods. This result proves that HONO-related reactions occurred more frequently during this period. The fair correlation between HONO and PM 2.5 may pinpoint the mainly anthropogenic origins of these two pollutants with the high direct or indirect contribution of combustion sources. The reason for the increased HONO during the heavy pollution period could be due to the comparatively high loading and large particle surface (Cui et al., 2018). Similar phenomena have been observed in a correlation study on CO and HONO in which CO was used as a tracer for traffic-induced emissions and tested by considering the correlation between HONO and CO over an identical time interval (Qin et al., 2009). The correlation coefficient between HONO and CO was relatively moderate (R 2 = 0.43), indicating that HONO and CO could come from the same source of emissions. Gen- Figure 7. Nighttime correlation studies between PM 2.5 and HONO/NO 2 ; PM 2.5 and HONO; CO and HONO; and RH and HONO/NO 2 during the entire measurement period -including the CD, PD, and SPD periods. The blue lines represent the full measurement period; the light blue points represent the CDs period; the black points represent the PDs period; and the red points represent the SPDs period. erally speaking, CO and NO are mainly related to combustion processes such as vehicle emissions and fossil fuel and biomass combustion . Thus, fossil fuel and biomass combustion may contribute to HONO production, but they cannot be measured directly.
The absorbed water influences the heterogeneous formation (Stutz et al., 2004). The influence of RH on the heterogeneous conversion is shown in Fig. 7d. When RH was less, the HONO/NO 2 ratio slowly increased. When RH was increased, the HONO/NO 2 ratio began to increase rapidly with RH. The HONO/NO 2 ratio decreased when RH reached a certain high level. Similar variation patterns have been observed in previous studies (Huang et al., 2017;Qin et al., 2009;Tong et al., 2015). Surface adsorbed water functions not only as the source but also as the sink of HONO by affecting the hydrolysis of NO 2 and the sedimentation of HONO to generate HONO (Ammann et al., 1998). When RH ranged at the middle level, the heterogeneous conversion of NO 2 to HONO was more significant than that of the deposition. This phenomenon confirms that RH improved the conversion efficiency (Stutz et al., 2004). However, the surface reached saturation when RH reached a certain high level. The excess wa-ter restricted NO 2 transformation (Wojtal et al., 2011). The absorption and dissolution of HONO by the saturated surface water layer caused the HONO/NO 2 ratio to decrease drastically.
The correlation between HONO correct and NO 2 at nighttime is shown in Fig. S1. HONO correct was used in the calculation to exclude the influence of direct emissions on NO 2 conversion. The nocturnal variations of HONO correct , NO 2 , and HONO correct /NO 2 ratios in the CD, PD, and SPD periods are presented in Fig. 8. The uncertainties of HONO correct , NO 2 , and HONO correct /NO 2 ratios in Fig. 8 are shown in Table S4. In general, the HONO correct /NO 2 ratio reached its maximum at or before midnight but decreased after midnight. In the PD and SPD periods, HONO was generated by heterogeneous reaction (R4), and NO 2 decreased after midnight. The production of HONO was equal to its loss (mainly night deposition), and the HONO concentration reached a relative balance. The weak correlation between nighttime HONO/NO 2 and PM 2.5 can be reasonably explained by the stable HONO correct /NO 2 ratio after midnight (Qin et al., 2009). A previous study (Xu et al., 2015) found that a low HONO correct in the first half of the night (19:00-00:00 LT) indicates the important contribution of automobile exhaust emissions, and a low HONO correct in the second half of the night means heterogeneous reactions dominate. Therefore, the heterogeneous reaction conversion rate of HONO was calculated in the current study by using the data of HONO correct .
The conversion rate of HONO (C HONO ) is usually used as an indicator to test the efficiency of NO 2 heterogeneous reactions. Total HONO correct was assumed to be generated by the heterogeneous transformation of NO 2 . The formula for the conversion rate of NO 2 (C HONO ) is as follows (Su et al., 2008a;Xu et al., 2015): where [NO 2 ] is the average concentration of NO 2 within the t2 − t1 time interval (1 h). In this study, the averaged conversion rate of NO 2 was 1.02 × 10 −2 h −1 . The mean values of C HONO in the CD, PD, and SPD periods were 0.72 × 10 −2 , 0.64 × 10 −2 , and 1.54 × 10 −2 h −1 , respectively. The averaged conversion rates in this study were 0.58 × 10 −2 and 1.46 × 10 −2 h −1 higher than those of Beijing I (polluted) and II (heavily polluted) periods, respectively. The increase in the conversion rate demonstrates that NO 2 had high reaction efficiency through the process from NO 2 to HONO in the aggravation of pollution, which could have led to the high utilization efficiency of the aerosol surface. The exact uptake coefficients of NO 2 on the ground and aerosol surfaces are variable and should be different (Harrison and Collins, 1998). The present analysis simplified this process by treating the ground and aerosol surfaces in the same way. The uptake co-efficient is mainly dependent on the surface characteristics, e.g., surface type and moisture .

Daytime HONO budget
The expression of dHONO/dt represents the observed variations of hourly HONO concentrations, for which we can use HONO/ t in the following way instead: dHONO/dt = sources − sinks = (P unknown + P OH+NO + P emi + P het ) The dHONO/dt calculated from the measurements was small and evenly distributed at around zero . P unknown is the production rate by an unknown daytime HONO source. P OH+NO is the rate of reaction of NO and OH. P emi represents the direct emission rate of HONO from combustion processes. The daytime HONO budget was analyzed with Eq. (4) by studying the source and reduction (Su et al., 2008b). The heterogeneous transformation mechanism was assumed to be the same for day and night. Therefore, the daytime heterogeneous productivity (P het = C HONO × [NO 2 ]) was calculated with the nighttime mean values of C HONO in different periods. L OH+HONO is the rate of the reaction between OH and HONO (reaction R3). The calculation formulas of P OH+NO and L OH+HONO have been provided in Sect. 3.2.1. Upon sunlight irradiation, q OH and NO were formed as reaction R1. L photo represents the photolysis loss rate of HONO (L photo = J HONO × [HONO]). The   (Rohrer and Berresheim, 2006). Aerosol effects were considered by using aerosol optical thickness (AOD), single scattering albedo (SSA), and the Ångström exponent as inputs in the TUV model. Typical AOD, SSA, and Ångström exponent values of 1.32, 0.9, and 1.3, respectively, were adopted for the PD and SPD periods. In the CDs period, the respective values were 0.66, 0.89, and 1.07 (Che et al., 2015;Cui et al., 2018;Hou et al., 2016). We wanted to study the produc-tion and loss rate of HONO under the same output conditions from the TUV model in the PD and SPD periods; however, the impact of different pollution levels changed the daytime budget. Hence, the average profiles of J HONO and J O 1 D concentrations in the CD, PD, and SPD periods are shown in Fig. 9. The mean values of J HONO and q OH concentration at noon in the CD, PD, and SPD periods were 5.93 × 10 −4 , 3.79×10 −4 , and 3.79×10 −4 and 4.10×10 6 , 2.93×10 6 , and 3.76 × 10 6 molec. cm −3 , respectively. The results of the calculated OH radicals ranged (0.58-11.49) ×10 6 molec. cm −3 , and the mean value was 3.57 × 10 6 molec. cm −3 at noon in Zhengzhou. Each production and loss rate of daytime HONO during CD, PD, and SPD periods is illustrated in Fig. 9m, together with dHONO/dt. P unknown was at a high level before midday. P unknown approached 0 ppbv h −1 after midday. In the CD, PD, and SPD periods, the mean values of P unknown were 0.26, 0.40, and 1.83 ppbv h −1 , respectively; the mean values of P OH+NO were 1.14, 2.07, and 4.03 ppbv h −1 , respectively; the mean values of P emi were 0.17, 0.30, and 0.43 ppbv h −1 , respectively; and the mean values of P het were 0.14, 0.18, and 0.55 ppbv h −1 , respectively. The midday time P unknown (1.83 ppbv h −1 ) calculated in Zhengzhou during the winter haze pollution period was close to the result obtained from Beijing's urban area  1.85 ppbv h −1 ). The P unknown contribution to daytime HONO sources in CD, PD, and SPD periods accounted for 15 %, 14 %, and 28 % of the HONO production rate (P unknown + P OH+NO + P emi + P het ), respectively. Previous studies (Spataro et al., 2013;Yang et al., 2014) showed that meteorological conditions, such as solar radiation and WS, can affect unknown sources. The low P unknown contribution of daytime HONO concentration may be related to the low solar radiation and low wind speed during severe pollution. The concentration of NO has a great influence on P OH+NO , so the homogeneous reaction is still an important pathway of HONO production during the daytime. In addition to the photolysis of HONO and the homogeneous reaction of HONO and OH, one, or more, important sink might exist to control the variation between the sources and sinks of the daytime HONO during complex contamination. However, further research is needed to analyze the unknown sources of daytime HONO.

Conclusions
Ambient HONO measurement, using AIM with other atmospheric pollutants and meteorological parameters, was conducted in the CPER. The HONO concentrations during the entire measurement varied from 0.2 to 14.8 ppbv, with an average of 2.5 ppbv. The HONO concentrations in the CD, PD, and SPD periods were 1.1, 2.3, and 3.7 ppbv, respectively, and the HONO/NO 2 ratios were 4.7 %, 7.1 %, and 9.4 %, respectively. HONO concentration was a combined action of direct emissions and heterogeneous reaction, and the contributions of these two were higher than those of the homogeneous reactions in the first half of the night. However, the proportion of homogenization gradually increased in the second half of the night due to the steady increase in NO concentration. The hourly level of other HONO abatement pathways, aside from OH + HONO, should be at least 0.22 ppbv h −1 in the SPDs period. The sum of the frequency distributions of the HONO emission /HONO ratio (less than 20 %) was approximately 77 %, indicating that the direct emission of HONO was not the main source of the observed HONO level at night. The mean values of HONO emission /HONO in the CD, PD, and SPD periods were 17 %, 16 %, and 16 %, respectively. This phenomenon means that the policy of restricting motor vehicles, published by the local government in January 2019, had a good effect on decreasing HONO emissions. In addition, when RH increased at the middle level, the heterogeneous HONO production increased, but it decreased when RH increased further due to the effect of surface water. The contribution of the three sources varied with different pollution levels.
The mean values of C HONO in the CD, PD, and SPD periods were 0.72 × 10 −2 , 0.64 × 10 −2 , and 1.54 × 10 −2 h −1 , respectively. In the SPDs period at nighttime, the heterogeneous conversion of NO 2 appeared to be unimportant. Furthermore, the net production generated by homogeneous re-action may be the leading factor for the increase in HONO under high-NO x conditions (i.e., the concentration of NO was relatively higher than that of NO 2 ) at nighttime. The mean values of P net OH+NO in the CD, PD, and SPD periods were 0.13, 0.26, and 0.56 ppbv h −1 , respectively. Daytime HONO budget analysis showed that the mean values of P unknown in the CD, PD, and SPD periods were 0.26, 0.40, and 1.83 ppbv h −1 , respectively. Although the values of P OH+NO had high uncertainty because of the variation of NO concentrations, P OH+NO contributed the most to HONO production during the daytime. After analysis, C HONO , P net OH+NO , and P unknown in the SPDs period were larger than those in the other periods, indicating that HONO participated in many reactions.
Data availability. All the data used in this paper are available from the corresponding author upon request (jiangn@zzu.edu.cn).
Author contributions. NJ, RZ, and, SL conceived and designed the study. QH analyzed the data and wrote the paper. LY performed aerosol sampling and data analyses.  Review statement. This paper was edited by Eliza Harris and reviewed by three anonymous referees.