Online measurements of cycloalkanes based on NO + chemical ionization in proton transfer reaction time-of-ﬂight mass spectrometry (PTR-ToF-MS)

. Cycloalkanes are important trace hydrocarbons existing in the atmosphere, and they are considered a major class of intermediate volatile organic compounds (IVOCs). Laboratory experiments showed that the yields of secondary organic aerosols (SOAs) from oxidation of cycloalkanes are higher than acyclic alkanes with the same carbon number. However, measurements of cycloalkanes in the atmosphere are still challenging at present. In this study, we show that online measurements of cycloalkanes can be achieved using proton transfer reaction time-of-ﬂight mass spectrometry with NO + chemical ionization (NO + PTR-ToF-MS). Cyclic and bicyclic alkanes are ionized with NO + via hydride ion transfer, leading to major product ions of C n H + 2 n − 1 and C n H + 2 n − 3 , respectively. As isomers of cycloalkanes, alkenes undergo association reactions with major product ions of C n H 2 n (cid:113) (NO) + , and concentrations of 1-alkenes and trans 2-alkenes in the atmosphere are usually signiﬁcantly lower than cycloalkanes (about 25 % and < 5 %, respectively), as a result inducing little interference with cycloalkane detection in the atmosphere. Calibrations of various cycloalkanes show similar sensitivities associated with small humidity dependence. Applying this method, cycloalkanes were successfully measured at an urban site in southern China and during a chassis dynamometer study of vehicular emissions. Concentrations of both cyclic and bicyclic alkanes are signiﬁcant in urban air and vehicular emissions, with comparable cyclic alkanes / acyclic alkanes ratios between urban air and gasoline vehicles. These results demonstrate that NO + PTR-ToF-MS provides a new complementary approach for the fast characterization of cycloalkanes in both ambient air and emission sources, which can be helpful to ﬁll the gap in understanding the importance of cycloalkanes in the atmosphere

Abstract. Cycloalkanes are important trace hydrocarbons existing in the atmosphere, and they are considered a major class of intermediate volatile organic compounds (IVOCs). Laboratory experiments showed that the yields of secondary organic aerosols (SOAs) from oxidation of cycloalkanes are higher than acyclic alkanes with the same carbon number. However, measurements of cycloalkanes in the atmosphere are still challenging at present. In this study, we show that online measurements of cycloalkanes can be achieved using proton transfer reaction time-of-flight mass spectrometry with NO + chemical ionization (NO + PTR-ToF-MS). Cyclic and bicyclic alkanes are ionized with NO + via hydride ion transfer, leading to major product ions of C n H + 2n−1 and C n H + 2n−3 , respectively. As isomers of cycloalkanes, alkenes undergo association reactions with major product ions of C n H 2n q (NO) + , and concentrations of 1-alkenes and trans-2-alkenes in the atmosphere are usually significantly lower than cycloalkanes (about 25 % and < 5 %, respectively), as a result inducing little interference with cycloalkane detection in the atmosphere. Calibrations of various cycloalkanes show similar sensitivities associated with small humidity dependence. Applying this method, cycloalkanes were successfully measured at an urban site in southern China and during a chassis dynamometer study of vehicular emissions. Concentrations of both cyclic and bicyclic alkanes are significant in urban air and vehicular emissions, with comparable cyclic alkanes / acyclic alkanes ratios between urban air and gasoline vehicles. These results demonstrate that NO + PTR-ToF-MS provides a new complementary approach for the fast characterization of cycloalkanes in both ambient air and emission sources, which can be helpful to fill the gap in understanding the importance of cycloalkanes in the atmosphere.
Many studies have shown higher alkanes (i.e., linear and branched alkanes with 12-20 carbon atoms) to be important chemical components of IVOCs Zhao et al., 2014). Similarly to these acyclic alkanes, cycloalkanes can also account for significant fractions of IVOCs. Cycloalkanes can reach more than 20 % of IVOC concentrations in diesel vehicle exhausts, lubricating oil, and diesel fuels Liang et al., 2018;Lou et al., 2019), which are comparable to or even higher than linear and branched alkanes. In some oil and gas regions, high concentrations of cycloalkanes have also been reported (Aklilu et al., 2018;Gilman et al., 2013;Warneke et al., 2014). More importantly, laboratory studies suggest that SOA yields of cyclic and polycyclic alkanes are significantly higher than linear or branched alkanes with the same carbon number (as high as a factor of 5) (Hunter et al., 2014;Jahn et al., 2021;J. Li et al., 2021;Loza et al., 2014;Yee et al., 2013). As a result, cyclic and bicyclic species are shown to be large contributors to SOA formation potential from vehicles (Xu et al., 2020a, b;Zhao et al., 2015Zhao et al., , 2016. Recently, Hu et al. (2022) proposed that IVOC contributions to SOA formation in an urban region can increase from 8 %-20 % (acyclic alkanes only) to 17.5 %-46 % if cycloalkanes are considered, signifying the importance of cycloalkanes in SOA formation.
Cycloalkanes are mainly measured using gas chromatography-mass spectrometry/flame ionization detector (GC-MS/FID) and two-dimensional gas chromatography techniques (GC × GC) (Alam et al., , 2016de Gouw et al., 2017;Liang et al., 2018;Zhao et al., 2016). Based on measurements of gas chromatographic techniques, the signals of unspeciated cyclic compounds can be determined. This is done by subtracting the total signal for each retention time bin according to the series of n-alkanes (Zhao et al., 2014(Zhao et al., , 2016. The mass of linear alkanes and branched alkanes in each bin is calculated by using the total ion current (TIC) and the fraction of characteristic fragments (C 4 H + 9 , m/z 57) (Zhao et al., 2014(Zhao et al., , 2016. However, this type of quantitative method does not explicitly distinguish individual cycloalkanes, and the determined mass may contain other cyclic compounds, e.g., polycyclic aromatic hydrocarbons and compounds containing oxygen or multifunctional groups (Zhao et al., 2014(Zhao et al., , 2015(Zhao et al., , 2016. Due to the need for collection and pretreatment of air samples, time resolution of GC-MS techniques is usually in the range of 0.5-1.0 h or above. Proton transfer reaction mass spectrometry (PTR-MS) using hydronium ions (H 3 O + ) as the reagent ion is capable of measuring many organic compounds with a high response time and sensitivity (de Gouw and Warneke, 2007;Yuan et al., 2017). However, detection of alkanes and cycloalkanes using PTR-MS with H 3 O + ionization is challenging, as usually only a series of fragment ions (C n H + 2n+1 , C n H + 2n−1 , n ≥ 3) are observed (Erickson et al., 2014;Gueneron et al., Figure 1. Ionization energy (IE) and reaction pathways with NO + ions of organic compounds including alkanes (green pentagon), aromatics (brown triangle), alkenes (red circle), phenolic species (blue circle), aldehydes (brown square), ketones (blue triangle), esters (green diamond), and nitrogen-containing species (black square). The ionization energy of NO (9.26 eV) is represented by the dashed line, with shading representing reported uncertainty. The IE of various organic compounds can be obtained in the NIST Chemistry WebBook (http://webbook.nist.gov, last access: 29 November 2022). The abbreviation "1,2,4-TMB" corresponds to 1,2,4trimethylbenzene.

2015)
. Recently, it was demonstrated that linear alkanes can be measured by PTR-MS with a time-of-flight detector using NO + as reagent ions (NO + PTR-ToF-MS) (Inomata et al., 2014;Koss et al., 2016;Wang et al., 2020). These higher alkanes are ionized by NO + via hydride ion transfer, leading to major product ions of C n H + 2n+1 , with a low degree of fragmentation (Inomata et al., 2014). Meanwhile, it is interesting that cycloalkanes were also quantified using C n H + 2n+1 ions in H 3 O + PTR-MS in oil and gas regions Warneke et al., 2014;Yuan et al., 2014), though sensitivities were substantially low (∼ 10 % of other species) . This evidence suggests that the NO + ionization scheme could provide a possibility for measuring cycloalkanes along with acyclic alkanes, as demonstrated in two recent works (Koss et al., 2016;Wang et al., 2022a).
In this study, we discuss the potential of online measurements of cycloalkanes for ambient air and emission sources utilizing NO + ionization in PTR-ToF-MS. The results of laboratory experiments to characterize product ions, calibration, and response time will be shown. Finally, measurements of cycloalkanes using NO + PTR-ToF-MS will be demonstrated from deployments at an urban site in southern China and a chassis dynamometer study for vehicular emissions.

NO + PTR-ToF-MS measurements
A commercial PTR-ToF-MS instrument (Ionicon Analytik, Austria) equipped with a quadrupole ion (Qi) guide for effective transfer of ions from the drift tube to the time-of-flight Table 1. The formula, purity, and ionization energy (IE) of the species used in product ion characterization experiments are shown. The percentage of each product ion from the reactions with NO + ions is indicated in parentheses, and the major product ions are identified in bold.

Species
Formula Purity (%) IE a (eV) Product ions (%) mass spectrometer is used for this work (Sulzer et al., 2014), and the mass resolution reaches approximately 3000 m/ m ( Fig. S1 in the Supplement). In order to generate NO + ions, 5 sccm ultra-high-purity air (O 2 +N 2 ≥ 99.999 %) is directed into the hollow cathode discharge area of the ion source; NO + ions are produced by ionization as follows (Federer et al., 1985;Karl et al., 2012): For the purpose of ionizing NO + ions to the greatest extent and reducing the generation of impurity ions such as H 3 O + , O + 2 , and NO + 2 , the ion source voltages U S and U SO was set to 40 and 100 V, while drift tube voltages U dx and U drift were set to 23.5 and 470 V with drift tube pressure at 3.8 mbar, resulting in an E/N (electric potential intensity relative to gas number density) of 60 Td. The specific details have been described in Wang et al. (2020). The intensities of primary ions NO + and impurities including O + 2 , NO + 2 , and H 3 O + and the ratio of O + 2 to NO + during the measurements of urban air and vehicular emissions are shown in Fig. S2. The abundances of O + 2 , NO + 2 , and H 3 O + are significantly lower than NO + ions, and the ratio of O + 2 to NO + is basically below 5 % during the measurements of urban air except for the period of 26 October to 2 November 2018 (7 %-10 %), while the ratio of O + 2 /NO + is basically below 2 % during the measurements of vehicular emissions. The measured ToF data are processed for high-resolution peak fitting using Tofware (Tofwerk AG, version 3.0.3), obtaining high-precision signals for cycloalkanes (Fig. S3). The signal of cycloalkanes used for quantification has been subtracted from the contribution of isotopes of other ions and other species such as unsaturated aldehydes that share an identical formula at the unit mass resolution (UMR) with cycloalkanes during the highresolution peak fitting process. A description of the fitting and calculation methods have been fully discussed in previous studies (Stark et al., 2015;Timonen et al., 2016). The raw ion count signals of NO + PTR-ToF-MS are normalized to the primary ion (NO + ) at a level of 10 6 cps to account for fluctuations of the ion source and detector (see Supplement).
Compared to proton transfer reactions occurring mostly between H 3 O + ions and VOC (volatile organic compound) species, NO + ions show a variety of reaction pathways with VOCs, which can be roughly summarized as follows: hydride ion transfer association reaction and bicyclohexyl (e) in NO + PTR-ToF-MS. The major product ions are shown in red, and the fragments are shown in blue.
As shown in Fig. 1, the ionization energy (IE) of VOC species is a determining factor for the reaction pathways with NO + . For example, as the IE of NO is 9.26 eV (Reiser et al., 1988), species with an IE less than 9.26 eV, e.g., benzene and isoprene, will undergo charge transfer reaction (Reaction R5) with NO + Smith, 1999, 1996), while species with IE greater than 9.26 eV, e.g., acetone and n-undecane, will undergo hydride ion transfer (Reaction R6) or association reaction (Reaction R7) with NO + (Amador- Muñoz et al., 2016;Diskin et al., 2002;Koss et al., 2016).

Calibration and correction experiments
In this study, we investigate characteristic ions of cycloalkanes generated by NO + ionization from a series of species identification experiments. The information of cycloalkanes species used in these experiments is listed in Table 1.
In addition, we also evaluated potential interferences from mono-alkenes, the isomers of cycloalkanes. Calibration experiments were carried out to obtain the sensitivities of cycloalkanes in both the laboratory and the field, using a customized cylinder gas standard (Apel-Riemer Environmental, Inc., USA), containing five different alkyl-cyclohexanes (C 10 -C 14 ) and eight n-alkanes (C 8 -C 15 ) (Table S1). Furthermore, some additional experiments were performed to explore the influence of humidity and tubing delay effects on measurements of cycloalkanes. Previously, it was shown that response factors of higher alkanes in NO + PTR-ToF-MS are slightly affected by air humidity, and the degree of influence is related to carbon number . Therefore, we evaluate the influence of humidity on sensitivities of cycloalkanes in NO + PTR-ToF-MS using a custombuilt humidity delivery system (Fig. S4), and the results are applied to explore the relationship between sensitivities of cycloalkanes and humidity. The perfluoroalkoxy (PFA) Teflon tubing is used for inlets in this study, but gas-wall partitioning can be important for low-volatility compounds (Pagonis et al., 2017). As a result, measurements from controlled laboratory experiments and field deployments were analyzed to systematically quantify and characterize the tubing delay time of cycloalkanes.
3 Results and discussion 3.1 Characterization of product ion distribution NO + PTR-ToF-MS was used to directly measure high-purity cycloalkane species and identify the characteristic product ions produced by cycloalkanes under NO + ionization. Here, the major product ions, fragments, and their contributions for different cycloalkanes are shown in Table 1. Chemical formulas of the product ions are determined based on the positions of measured mass peaks. Figure 2 shows mass spectra within the relevant range (m/z 0 + to 200 + Th) for cycloalkanes. The signals are normalized to the largest ion peak for better comparison. As shown in Fig. 2, no significant fragmentation appears for cycloheptane and methylcyclohexane (C 7 H 14 ), and the dominating product ions are observed at m/z 97 Th, corresponding to C 7 H + 13 . Similarly, the product ions generated by hexyl- cyclohexane and cyclododecane (C 12 H 24 ) under NO + ionization mainly appear at m/z 167 Th (Fig. 2c-d), corresponding to C 12 H + 23 , and fragments occur at m/z 97 Th and m/z 111 Th, corresponding to C 7 H + 13 and C 8 H + 15 , respectively. The product ions generated by cyclopentadecane and nonylcyclohexane (C 15 H 30 ) mainly appear at m/z 209 Th, corresponding to C 15 H + 29 , with slightly more fragmentation than C 12 cyclic alkanes (Fig. S5). Bicyclic cycloalkanes undergo the same ionization channel from NO + ionization, as demonstrated by major product ions at m/z 165 Th (C 12 H + 21 ) and other fragmentation ions from bicyclohexyl (C 12 H 22 ) (Fig. 2e). For instance, the mass spectra for C 10 -C 14 alkylcyclohexanes during the calibration experiments are shown in Fig. S6, with the same C n H + 2n−1 as the major product ions. The fractions of M-H ions generated by high-purity cycloalkane species including C 7 , C 12 , and C 15 cyclic alkanes and C 12 bicyclic alkanes are summarized in Fig. S7. We observe that M-H ions account for ∼ 100 % of total ion signals for C 7 cyclic alkanes and lower but comparable fractions (74 %-82 %) for C 12 and C 15 cyclic alkanes. These results verify that reactions of cyclic and bicyclic alkanes with NO + ions follow the hydride ion transfer pathway to yield C n H + 2n−1 and C n H + 2n−3 product ions, respectively. As mentioned above, the characteristic peaks of cycloalkanes under NO + ionization are consistent with the ions that are received from the attempts to utilize H 3 O + PTR-MS. For the latter method though, sensitivities are reported to be lower (Erickson et al., 2014;Gueneron et al., 2015;Warneke et al., 2014;Yuan et al., 2014). As the result, we speculate that reactions of NO + with cycloalkanes may present large contributions to cycloalkane M-H product ions in the H 3 O + chemistry mode of PTR-MS (Španěl et al., 1995).  1,1,3,5tetramethylcyclohexane (red), pentylcyclohexane (blue), hexylcyclohexane (orange), heptylcyclohexane (green), and octylcyclohexane (black). (b) Humidity dependence of sensitivity for various cycloalkanes, including measurement results for hexylcyclohexane (red markers) and the fitted lines for C 10 cyclic alkane (green), C 12 cyclic alkane (black), and C 14 cyclic alkane (blue), with the corresponding fitted functions of y = 0.82 + 0.19× exp(−0.06x), y = 0.87 + 0.14× exp(−0.06x), and y = 0.90 + 0.11× exp(−0.07x), respectively.
As the isomers of cyclic alkanes, alkenes may interfere with measurements of cycloalkanes. Here, we use 1-heptene (C 7 H 14 ) and 1-decene (C 10 H 20 ), the isomers of C 7 and C 10 cyclic alkanes, to explore the ionization regime of alkenes in NO + chemical ionization ( Fig. 3 and Table 1). These two alkenes produce more fragments than cycloalkanes under NO + chemistry but mainly react with NO + via association reactions to yield C n H 2n q (NO) + product ions. The major product ions of 1-heptene and 1-decene appear at m/z 128 Th and m/z 170 Th, corresponding to C 7 H 13 HNO + and C 10 H 19 HNO + , respectively. Based on the mass spectra, alkenes produce the C n H + 2n−1 product ions at fractions of < 5 %, which are similar to fragmentation ions from NO + ionization of the two species and other 1-alkenes determined from a selected ion flow tube mass spectrometer (SIFT-MS) (Diskin et al., 2002). The same study (Diskin et al., 2002) also demonstrated that trans-2-alkenes might produce more C n H + 2n−1 ions under NO + ionization (e.g., trans-2heptene contributes 40 % of C 7 H + 13 ions). However, concentrations of 1-alkenes and trans-2-alkenes in the atmosphere are usually significantly lower than cycloalkanes (about 25 % and < 15 %, respectively) Yuan et al., 2013). We further compare the signals of C n H + 2n and C n H + 2n−1 from calibration experiments, urban air measurements, and a chassis dynamometer study (Fig. S8). The typical ratios of C n H + 2n / C n H + 2n−1 for cyclic alkanes are in the range of 2 %-6 %, with similar ratios determined from urban air measurements (3 %-7 %). The ratios of C n H + 2n to C n H + 2n−1 from vehicular emissions are maintained at 6 %-16 % for C 10 -C 14 ions, which is a little bit higher than those determined from cyclic alkanes, while the ratios of C 15 -C 20 ions are comparable with pure cyclic alkanes (4 %-8 %). Even though all of these differences are attributed to potential interference from 2-alkenes and assume the same quantity of C n H + 2n−1 and C n H + 2n ions from NO + ionization from 2-alkenes, the upper limits of the interference from alkenes should be in the range of 1 %-2 % for urban air measurements and 2 %-12 % for measurements of vehicular emissions. Therefore, we conclude that the interferences from alkenes to cyclic alkane measurements in most environments are minor.

Sensitivity, humidity dependence, and detection limits
The calibration experiments of cycloalkanes (see details of gas standard in Table S1) are carried out in both dry conditions (< 1 % RH) and humidified conditions (Fig. S9). Figure 4 illustrates the results from a typical calibration experiment for five different alkyl-substituted cyclohexanes with carbon atoms of 10-14 in dry air (relative humidity < 1 %) for NO + PTR-ToF-MS. There is a good linear relationship between ion signals and concentrations of various cycloalkanes (R = 0.9996 to 0.9999). Sensitivities (ncps ppb −1 ) of cycloalkanes are in the range of 210 to 260 ncps ppb −1 (Table 2). The sensitivity of each cycloalkane remained stable in the long-term calibration conducted in the laboratory and in the field (Fig. S10). Table 2 further shows the detection limits of cycloalkanes by NO + PTR-ToF-MS, which are calculated as the concentrations with a signal-to-noise ratio of 3 (Bertram et al., 2011;Yuan et al., 2017). The detection limits of cycloalkanes at the integral times of 10 s and 1 min are in the range of 6.2-8.2 ppt and 2.4-3.6 ppt, respectively. For comparison, the detection limits of NO + PTR-ToF-MS for 24 h of integral time are also determined, obtaining comparable results (< 0.1 ppt) with detection limits of GC × GC-ToF-MS (0.1-0.3 ppt for daily sample) Xu et al., 2020a),.  Figure 4b shows that normalized signals of hexylcyclohexane are relative to dry conditions as a function of different humidity. The relative signals of the explored cycloalkanes show a minor decrease (< 10 %) at the highest humidity (∼ 82 % RH at 25 • ) compared to the dry condition, and the observed changes for cycloalkanes with different carbon numbers are similar, suggesting little influence of humidity on measurements of cycloalkanes. The humidity-dependence curves determined in Fig. 4b are used to correct variations in ambient humidity in the atmosphere.
The response time of various cycloalkanes in the instrument and also the sampling tubing is determined from laboratory experiments (Fig. S11). For the species not in the gas standard, we also take advantage of vehicular emissions measurements associated with high concentrations of cycloalkanes, and the sampling methods are the same as mentioned in T. Li et al. (2021) and Wang et al. (2022b). Here, we use the delay time to determine response of cycloalkanes, which is calculated based on the time it takes for signals to drop to 10 % of their initial value (Fig. S12) (Pagonis et al., 2017). The delay time of cycloalkanes is summarized in Fig. 5. The delay time of various cycloalkanes generally increases with the carbon numbers, ranging from a few seconds to a few minutes, but are lower than determined for those acyclic alkanes within the C 10 -C 15 range . These results suggest that measured variability in cycloalkanes with higher carbon numbers, especially for C 19 -C 20 or above, may only be reliable for timescales longer than 10 min. Figure 6. (a) Time series of CO, cyclic, bicyclic, and acyclic alkanes measured at the urban site in Guangzhou. (b) Normalized diurnal variations in CO, benzene, C 8 aromatics, C 10 cyclic alkanes, C 10 acyclic alkanes, C 12 cyclic alkanes, C 12 bicyclic alkanes, and C 12 acyclic alkanes. The measurement data for each species are normalized to midnight concentrations. (c) Scatterplots of cyclic and bicyclic alkanes to acyclic alkanes with carbon atoms of 12 and 15.

Applications in ambient air and vehicular exhausts
Based on the results shown above, we deployed NO + PTR-ToF-MS to measure concentration levels and variations in cycloalkanes at an urban site in Guangzhou, southern China. Details of the field campaign have been described in previous studies Wu et al., 2020). The average sensitivities of long-term calibration (dry condition) during the field observations were used to quantify cycloalkanes after corrected variations in ambient humidity in the atmosphere. For the cycloalkanes that are not contained in the gas standard, we employ average sensitivity for calibrated cycloalkanes in the gas standard.
The concentration levels and diurnal profiles of cyclic and bicyclic alkanes are illustrated in Fig. 6a, along with CO. In general, cyclic and bicyclic alkanes demonstrated similar temporal variability to CO, suggesting cyclic and bicyclic alkanes may be mainly emitted from primary combustion sources. Concentrations of C 12 bicyclic alkanes are observed to be comparable with C 12 cyclic alkanes. As shown in Fig. 6b, selected C 10 and C 12 cycloalkanes show diurnal variations with lower concentrations during the daytime.
Compared to diurnal variations in other species with different reactivity , the decline fractions of cycloalkanes are more comparable to the reactive species (e.g., C 8 aromatics) than the inert ones (e.g., CO, benzene), indicating significant daytime photochemical removal of these cycloalkanes. Diurnal patterns of other cyclic and bicyclic alkanes exhibit similar results (Fig. S13). As discussed in Wang et al. (2020), similar diurnal profiles of cycloalkanes with different carbon numbers also imply that the tubing delay effect may not significantly affect temporal variations in cycloalkanes reported here. Based on both time series and correlation analysis (Fig. 6c), cyclic and bicyclic alkanes showed strong correlation with acyclic alkanes, suggesting that they predominantly came from same emission sources.
NO + PTR-ToF-MS was also applied to measure cycloalkane emissions along with other organic compounds from vehicles using gasoline, diesel, and liquified petroleum gas (LPG) as fuel, by conducting a chassis dynamometer measurement (Wang et al., 2022b). Figure 7 shows the time series of C 12 cyclic and bicyclic alkanes, C 12 alkanes, toluene, and acetaldehyde measured by NO + PTR-ToF-MS for a gasoline vehicle with emission standard of China III Figure 7. Concentrations of C 12 cyclic, bicyclic, and acyclic alkanes; acetaldehyde; toluene; and CO 2 for (a) a gasoline vehicle with emission standard of China III and (b) a diesel vehicle with emission standard of China IV. The gray shadows represent the speeds of the vehicles on the chassis dynamometer. The data of toluene and acetaldehyde were detected by NO + PTR-ToF-MS. and a diesel vehicle with emission standard of China IV. Both test vehicles were started with hot engines. As shown in Fig. 7a, high concentrations of selected cycloalkanes emitted by the gasoline vehicle were observed as the engine started. Compared with typical VOCs exhausted by vehicles (e.g., toluene and acetaldehyde), concentrations of cycloalkanes were lower but showed similar temporal variations. In comparison, cycloalkane emissions from diesel vehicles are obviously different. As shown in Fig. 7b, concentrations of cycloalkanes showed relatively moderate enhancements as the engine started but were significantly enhanced with high vehicle speed, obtaining periodic pattern variations within each test cycle. Though the highest concentrations of cycloalkanes observed for gasoline and diesel vehicles are similar, determined emission factors of diesel vehicles are significantly higher than gasoline vehicles, since emissions of cycloalkanes are mainly concentrated during a short period at the engine start for gasoline vehicles, whereas emissions of cycloalkanes remain high during hot running for diesel ve-hicles. For instance, the determined emission factors of C 12 cyclic alkanes are 0.06 mg km −1 for gasoline vehicles and 1.17 mg km −1 for diesel vehicles. In addition to cycloalkanes and alkanes, other compounds including aromatics and oxygenated VOCs also demonstrate large differences between gasoline and diesel vehicles, which were mainly attributed to different chemical compositions of gasoline and diesel fuels (Wang et al., 2022b). Recent studies reported that cyclic compounds occupied a large proportion of IVOCs emitted from vehicles, with prominent SOA formation potentials (Fang et al., 2021;Huang et al., 2018;Zhao et al., 2016), but emissions of individual cycloalkanes have not been not reported yet. As a result, high-time-resolution measurements of cycloalkanes from vehicular emissions by NO + PTR-ToF-MS can improve the characterization of the emission mechanisms of these species.

Insights from simultaneous measurements of cycloalkanes and alkanes
Since NO + PTR-ToF-MS can provide simultaneous measurements of cycloalkanes and acyclic alkanes, we can use this information to explore the relative importance of cycloalkanes. Figure 8 shows mean concentrations of cycloalkanes (cyclic and bicyclic) and alkanes with different carbon numbers (C 10 -C 20 ) measured by NO + PTR-ToF-MS in urban air and emissions from diesel vehicles. In the urban region, concentrations of cyclic and bicyclic cycloalkanes are comparable but lower than acyclic alkanes, with concentration ratios relative to acyclic alkanes in the range of 0.30-0.46 for cyclic alkanes and 0.23-0.51 for bicyclic alkanes. Similar results are obtained for gasoline vehicles, with cyclic alkanes / acyclic alkanes and bicyclic alkanes / acyclic alkanes ratios of 0.27-0.53 and 0.21-0.52, respectively (Fig. 8).
The contributions of cycloalkanes in diesel vehicular emissions are relatively high, with the concentration ratios relative to acyclic alkanes in the range of 0.42-0.66 for cyclic alkanes and 0.37-0.95 for bicyclic alkanes.
As there are only limited measurements of bicyclic alkanes in the literature, we compare concentration ratios of cyclic alkanes to acyclic alkanes with results in previous studies, mainly using gas chromatography techniques (GC-MS/FID and GC × GC). The details of the technique used in these earlier studies are summarized in Table S2. As shown in Fig. 9, the ratios obtained in the urban region of Guangzhou in this work (0.2-0.4) are similar to other measurements in urban areas, including Algiers, Algeria (Yassaa et al., 2001). The ratios obtained in London, UK (Xu et al., 2020b), are higher than the ratios obtained in Guangzhou but similar to the diesel exhaust in our work for the C 13 -C 18 range. These results are likely due to the fact that the measurement location in London is in close proximity to a main road; cyclic and acyclic alkanes may be dominated by traffic emissions with high fractions of diesel vehicles in fleets. Although some variations are observed in different urban environments, these ratios are broadly within the range between gasoline and diesel emissions. As for emissions from diesel vehicles, the ratios measured in this study are similar to GC × GC measurements in the UK (Alam et al., 2016) for the C 12 -C 14 range and GC-MS measurements in the USA (Gentner et al., 2012), whereas the ratios from this study are higher than in Alam et al. (2016) for larger carbon numbers. The relative fractions of cycloalkanes measured from oil and gas regions (Gilman et al., 2013) and emissions from lubricating oil ) (> 0.7) are higher than ambient air and vehicular emissions. The variability pattern of ratios of cyclic alkanes to acyclic alkanes for different carbon numbers may be used for a source analysis of these IVOCs in the future.

Conclusions
In this study, we show that online measurements of cycloalkanes can be achieved using proton transfer reaction time-of-flight mass spectrometry with NO + chemical ionization (NO + PTR-ToF-MS). Our results demonstrate that cyclic and bicyclic alkanes are ionized via hydride ion transfer, leading to characteristic product ions of C n H + 2n−1 and C n H + 2n−3 , respectively. As isomers of cycloalkanes, alkenes undergo association reactions mainly yielding product ions C n H 2n q (NO) + , which induce little interference with cycloalkane detection. Calibrations of various cycloalkanes show similar sensitivities with low humidity dependence. The detection limits of cycloalkanes are in the range of 2-4 ppt at the integral time of 1 min.
Applying this method, cycloalkanes were successfully measured at an urban site in southern China and in a chassis dynamometer study for vehicular emissions. Concentrations of both cyclic and bicyclic alkanes are substantial in urban air and vehicular emissions. Diurnal variations in cy- Figure 9. The concentration ratios of cyclic alkanes to acyclic alkanes for different carbon numbers. Measurements in various urban areas, including Guangzhou in China, London in the UK (Xu et al., 2020b), Los Angeles in the USA , Algiers in Algeria (Yassaa et al., 2001), and an oil and gas region in Colorado in the USA (Gilman et al., 2013), are also shown for comparison. Emission sources, including vehicle exhausts (Alam et al., 2016;Gentner et al., 2012) and lubricating oils , are also included. cloalkanes in the urban air indicate significant losses due to photochemical processes during the daytime. The concentration ratios of cyclic alkanes to acyclic alkanes are similar between urban air and gasoline vehicle emissions but higher for diesel vehicles, which could potentially be used for source analysis in future studies. Our work demonstrates that NO + PTR-ToF-MS provides a new complementary approach for the fast characterization of cycloalkanes in both ambient air and emission sources, which can be helpful to investigate sources of cycloalkanes and their contribution to SOA formation in the atmosphere. Measurements of cycloalkanes in the particle phase may also be possible by combining NO + PTR-ToF-MS with "chemical analysis of aerosols online" (CHARON) or other similar aerosol inlets (Muller et al., 2017), which could further provide particle-phase information of cycloalkanes and gas-partitioning analysis of cycloalkanes.
Data availability. Data are available from the authors upon request.
Author contributions. BY designed the research. YBC, CMW, SHW, XJH, CHW, XS, YBH, XBL, YJL, and MS contributed to data collection. YBC performed data analysis, with contributions from CMW, SHW, XJH, and CHW. YBC and BY prepared the article with contributions from other authors. All the authors reviewed the article.
Competing interests. The contact author has declared that none of the authors has any competing interests.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.