The measurements were conducted between 16 April–26 July 2019 at the University of Helsinki Station for Measuring Ecosystem – Atmosphere Relations (SMEAR) II (Hari and Kulmala, 2005), which is located in a boreal forest in Hyytiälä, southern Finland (61∘51′ N, 24∘17′ E; 181 m a.s.l.). This station is dominated by Scots pine (Pinus sylvestris), and monoterpenes are found to be the dominating emitted biogenic non-methane VOCs (Barreira et al., 2017; Hakola et al., 2012). The measurement station has been considered a rural background site (Manninen et al., 2010; Williams et al., 2011), and the nearest big city is Tampere, with more than 200 000 inhabitants and located ∼ 60 km to the SW of our measurement site. A sawmill which is located 6–7 km away to the SE of our measurement site can contribute significantly to the OA loading in the case of SE winds, and the sawmill OA composition has been found to resemble biogenic OA a lot (Liao et al., 2011; Äijälä et al., 2017; Heikkinen et al., 2020).

All mass spectrometers were set up in a temperature-controlled measurement container kept at ∼ 25 ∘C. Sampling inlets were located about 1.5 m a.g.l. All data are reported in eastern European time (UTC+2).

The meteorological parameters were continuously monitored at this measurement site. Temperature was monitored with a Pt100 sensor (platinum resistance thermometer with a resistance of 100 Ω at 0 ∘C) inside a ventilated custom-made radiation shield, while wind directions and wind speed were measured with a 2D ultrasonic anemometer (Adolf Thies GmbH & Co. KG), and the global radiation was measured with an EQ08 pyranometer (Carter-Scott Manufacturing Pty. Ltd.). The main wind direction above the canopy during the measurement period was southwest (see Fig. S2). The mixing ratios of ozone (O3) and nitrogen oxides (NO and NO2) were measured with an ultraviolet light absorption analyzer (TEI 49C, Thermo Fisher Scientific Inc.) and a chemiluminescence analyzer (TEI 42CTL, Thermo Fisher Scientific Inc.), respectively. The mixing ratios of sulfur dioxide (SO2) were measured with a fluorescence analyzer (TEI 43CTL, Thermo Fisher Scientific Inc.).

An aerosol chemical speciation monitor (ACSM; Aerodyne Research Inc.; Ng et al., 2011) was deployed to continuously measure the non-refractory sub-micrometer aerosol particle chemical composition. The ACSM, which contains a quadrupole mass spectrometer, provided unit-mass-resolution mass spectra every 30 min. This information was chemically speciated to organic, sulfate, nitrate, ammonium, and chloride concentrations by the ACSM analysis software. The mass concentrations of each species were calculated based on frequently conducted ionization efficiency calibrations. The data were corrected for collection efficiency, which was ca. 60 % during the measurement period. The sampling was conducted through a PM2.5 cyclone and a Nafion dryer (RH < 30 %) with a stainless-steel tube of ca. 3 m length and a flow rate of 3 L min-1 (only 1.4 cm3 s-1 into the ACSM). The recorded data were analyzed using the ACSM local v. 1.6.0.3 toolkit (provided by Aerodyne Research Inc.) within the Igor Pro v. 6.37 (WaveMetrics Inc., USA). More details about ACSM operation and data processing can be found in Heikkinen et al. (2020).

Figure 1 shows the overview of the time series of meteorological parameters (temperature, global radiation, and wind direction and wind speed), trace gas concentrations (SO2, O3, NO, and NO2), and total gaseous organic compounds measured by MION-Br, MION-NO3, and Vocus, as well as total particulate organics measured by ACSM. Note that relatively long-lived compounds like ethanol, acetone, and acetic acid are excluded from Vocus data presented in this study in order to focus on compounds actively involved in the fast photochemistry (all excluded compounds are listed in Table S1, and the time series of total organic compound concentrations including them are shown in Fig. S3). As we can see from Fig. 1a, most of the measurement days had strong photochemical activities with ambient temperature exhibiting clear diurnal patterns ranging between -3 and 32 ∘C. In general, the time series of the total organics (both gas phase and particle phase; see Fig. 1e–f) measured by MION-Br, MION-NO3, Vocus, and ACSM were similar during the measurement period. Elevated levels of total gaseous and particulate organics (e.g., 17–24 May and 7–10 June; see Fig. 1e–f) were observed on warmer days with strong global radiation and with the main wind direction coming from southeast (the direction of the sawmill, e.g., 17–24 May) or southwest (e.g., 7–10 June; see Fig. 1a–b). Besides, higher concentrations of oxidants of VOCs (such as O3) and/or anthropogenic pollutants (such as SO2 and NOx) also followed some of the elevated concentrations of gaseous and/or particulate organics (e.g., 19 April–3 May, 17–24 May, and 7–10 June; see Fig. 1c–d). The observations of the elevated organics could result from higher VOC emissions (e.g., terpenes, the typically observed VOCs; Li et al., 2021; Fig. S3) influenced by meteorological conditions (i.e., temperature and/or light; Guenther et al., 1995; Kaser et al., 2013), different air mass origins (e.g., terpene pollution from the sawmill in the case of SE winds; Liao et al., 2011; Äijälä et al., 2017; Heikkinen et al., 2020), and chemistry initiated by or related with different trace gases (Yan et al., 2016; Massoli et al., 2018; Huang et al., 2019b; Heikkinen et al., 2020). The results suggest the important roles that meteorological parameters, trace gases, and air masses play in the emission and oxidation reactions of organic compounds. Due to the soft ionization processes of organic molecules in the Vocus, MION-Br, and MION-NO3, the molecular composition of organic compounds was obtained. In the next section we will discuss the molecular composition of the gaseous organic compounds measured by Vocus, MION-Br, and MION-NO3.

During the measurement period, Vocus identified 72 CH compounds (Cx≥1Hy≥1) and 431 CHOX compounds (Cx≥1Hy≥1Oz≥1X0-n), with X being different atoms like N, S, or a combination thereof, while MION-Br and MION-NO3 detected 567 and 687 CHOX compounds, respectively. Substantial overlaps of organic compounds were observed for these three ionization techniques, while distinct organic compounds were also detected with individual methods (Fig. S4). The average mass-weighted chemical compositions for organic compounds measured by Vocus, MION-Br, and MION-NO3 were C5.3H7.5O1.1N0.1, C6.7H10.7O4.3N0.3, and C7.5H11.4O5.4N0.3, respectively. We stress here that the fragmentation of organic compounds inside the Vocus may bias the chemical composition towards a shorter carbon backbone. And the average mass-weighted chemical composition representing the bulk of all measured gaseous organic compounds (with the approach described in Sect. 2.2.1) in this boreal forest was calculated to be C6.0H8.7O1.2N0.1, indicative of the short carbon backbone and relatively low oxidation extent. Similar to previous laboratory results (Riva et al., 2019), MION-NO3 observed the most oxidized compounds with the highest elemental oxygen-to-carbon ratios (O : C; 0.9 ± 0.1, average ± 1 standard deviation), followed by the MION-Br (0.8 ± 0.1); the O : C ratios of the organics detected by Vocus were lowest (0.2 ± 0.1). In addition, the CHO group comprised the largest fraction of the total organic compounds (Vocus: 43.6 ± 9.4 %; MION-Br: 75.4 ± 5.3 %; MION-NO3: 71.8 ± 7.9 %; see Table 1). The second most abundant group for Vocus was the CH group, making up 35.2 ± 15.1 % of its total organic compounds; while it was the CHON group for MION-Br (24.1 ± 5.2 %) and MION-NO3 (28.1 ± 7.9 %; see Table 1), indicating active NOx or NO3 radical-related chemistry (Yan et al., 2016). The CHON group only accounted for 8.1 ± 2.7 % of the total organic compounds measured by Vocus, possibly due to its lower sensitivity towards larger organonitrates (see also Fig. S5) caused by their losses in the sampling lines and on the walls of the inlet (Riva et al., 2019) and/or fragmentation inside the instrument (Heinritzi et al., 2016).

The mass defect plots for organic compounds measured by Vocus, MION-Br, and MION-NO3 are shown in Fig. 2. Similar to previous studies (e.g., Yan et al., 2016; Li et al., 2021), multiple series of organic compounds with different numbers of carbon atoms (such as C5, C10, C15, and C20) and oxygen atoms (up to 20; see also Fig. S5) were measured in this boreal forest environment. Organics with the lowest oxidation extent were better observed by Vocus, while organics with the largest molecular weights and highest oxidation extent were better observed by MION-NO3 (Fig. 2a). Figure 2b shows the mass defect plots of organic compounds grouped into different categories. The markers are color-coded with different compound groups, such as CH, CHO, CHON, and others. The size of the markers is proportional to the logarithm of the concentration of each compound. Generally, similar to previous laboratory results (Riva et al., 2019; Rissanen et al., 2019), Vocus and MION-Br detected better the CHO compounds in the mass range of 50–100 Da and CHON compounds in the mass range of 50–150 Da, and MION-Br even detected better the CHON compounds in the mass range of 350–425 Da, which are most likely to be less oxygenated monomers or dimers; MION-NO3 was more sensitive towards the CHO and CHON compounds in the mass range of 425–600 Da, which are most likely to be more oxygenated HOM dimers (see Figs. 2b and S5).

We further investigated the contributions of the measured CHOX compounds with different numbers of oxygen atoms per molecule to total CHOX compounds as a function of the number of carbon atoms (Fig. 3). Organic compounds which were detected with higher sensitivity by Vocus were those with the number of carbon atoms between 3 and 10 and the number of oxygen atoms between 1 and 3 (i.e., less oxygenated monomers). Compounds with a larger number of carbon atoms (i.e., > 10) and oxygen atoms (i.e., > 3) were much better detected by MION-Br and MION-NO3: the former particularly for CHON compounds with the number of carbon atoms between 15 and 20 and oxygen atoms between 4 and 8 (i.e., larger less oxygenated monomers and dimers; see Fig. S5b) and the latter particularly for compounds with the number of oxygen atoms larger than 9 (i.e., HOM monomers and dimers; Rissanen et al., 2019; Riva et al., 2019; Li et al., 2020; see Figs. 3 and S5). In the MION-Br and MION-NO3 data, CHOX compounds with the number of carbon atoms of 5, 10, 15, and even 20 exhibited relatively elevated contributions compared to their neighbors (Fig. 3), indicating contributions of their potential corresponding precursors, i.e., isoprene, monoterpenes, sesquiterpenes, and diterpenes (together accounting for 38.3 ± 12.5 % of total CH compounds; see Table S2, Figs. S3, and S6). We emphasize here that using the number of carbon atoms as a basis to relate the CHOX to their precursor VOCs is a simplified assumption, as negative or positive artifacts can arise from fragmentation or accretion reactions (Lee et al., 2016). A similar pattern was also observed by Huang et al. (2019a) in a rural area in southwest Germany, based on filter inlet for gases and aerosols high-resolution time-of-flight chemical ionization mass spectrometer (FIGAERO-HR-ToF-CIMS) data. The consistency and complement of the results demonstrate the different capabilities of these instruments for measuring gaseous organic compounds with different oxidation extents (from VOCs to HOMs).

Median diurnal variations of total CH, total CHO, and total CHON compounds measured by Vocus, MION-Br, and MION-NO3 are shown in Fig. 4. In general, the CH and CHO groups measured by Vocus exhibited higher levels during the night (see Fig. 4a–b), mainly driven by the boundary layer height dynamics (Baumbach and Vogt, 2003; Zha et al., 2018). Besides, CHO compounds measured by Vocus were dominated by O1-2 compounds (see Figs. 3 and S5) and have also been reported to follow more the CH trends (Li et al., 2020). Their relatively flat diurnal pattern could result from the smearing effect after summing up the much less oxygenated CHO molecules (mostly peaking at night) and comparatively more oxygenated CHO molecules (mostly peaking during daytime) (Li et al., 2020). In contrast, the CHO and CHON groups measured by MION-Br and MION-NO3 exhibited higher levels during the day (see Fig. 4b), due to strong photochemical oxidation caused by different meteorological parameters (i.e., temperature and global radiation; see Figs. 1a and S7), and/or elevated trace gas levels (e.g., O3 and SO2; see Figs. 1c and S7; Yan et al., 2016; Massoli et al., 2018; Huang et al., 2019b; Bianchi et al., 2017). However, the CHON group measured by Vocus showed relatively stable signals throughout the day (see Fig. 4c). The potential reason could be partly due to its lower sensitivity towards larger organonitrates (see Fig. S5) caused by their losses in the sampling lines and on the walls of the inlet (Riva et al., 2019) and/or their fragmentation inside the instrument (Heinritzi et al., 2016). Another potential reason could result from the smearing effect after summing up the much less oxygenated CHON molecules (mostly peaking at night or early morning) and comparatively more oxygenated CHON molecules (mostly peaking during daytime) (Li et al., 2020).

Different diurnal patterns among different measurement techniques can also be found for individual organic compounds with the same molecular formula, such as several dominant CHO and CHON species, C7H10O4 (molecular formula corresponding to 3,6-oxoheptanoic acid identified in the laboratory as limonene oxidation product by Faxon et al., 2018, and Hammes et al., 2019), C8H12O4 (molecular formula corresponding to terpenylic acid identified in monoterpene oxidation product by Zhang et al., 2015, and Hammes et al., 2019), and C10H15NO6–7 (identified in the laboratory as monoterpene oxidation products by Boyd et al., 2015, and Faxon et al., 2018; see Fig. 5). The inconsistent trends in time series and the varying correlations of these abovementioned dominant CHO and CHON species indicate different isomer contributions detected by different measurement techniques (Fig. S8 and Table S3). Similar behaviors were also evident for OVOCs with varying oxidation extents, like the terpene-related CxHO and CxHON compounds (x=5, 10, 15, and 20; see Fig. S9), which in total accounted for up to 27 % and 39 % of their corresponding CHO and CHON groups (see Table S2). Most of the terpene-related CxHO(N) groups (x=5, 10, 15, and 20) with different oxidation extents behaved similarly among different measurement techniques, but some were also found to vary (see Fig. S9). Compounds with the same number of carbon and oxygen atoms but different numbers of hydrogen atoms (i.e., different saturation level) were also found to behave differently (see Fig. S9c–d), possibly due to different chemistry involved in their formation (Zhao et al., 2018; Molteni et al., 2019). Even compounds with the same molecular formula varied among different measurement techniques (see Fig. S9c–d and also Fig. 5). The differences can likely result from different isomers detected by the different techniques and/or fragmentation products from different parent compounds inside the instruments (e.g., Heinritzi et al., 2016; Zhang et al., 2017).

The results indicate that organic compounds may behave differently among different measurement techniques during different time periods. In the next section, we will investigate the volatility of these gaseous organic compounds, which can influence their lifetime and roles in the atmosphere.

Based on the log⁡10Csat values of all organic compounds parameterized with the modified Li et al. (2016) approach (Daumit et al., 2013; Isaacman-VanWertz and Aumont, 2021) described in Sect. 2.2.2, the gaseous organic compounds were grouped into a 25-bin volatility basis set (VBS; Donahue et al., 2006) (Fig. 6a). Organic compounds with Csat lower than 10-8.5 µg m-3, between 10-8.5 and 10-4.5 µg m-3, between 10-4.5 and 10-0.5 µg m-3, between 10-0.5 and 102.5 µg m-3, between 102.5 and 106.5 µg m-3, and higher than 106.5 µg m-3 are termed ULVOCs, ELVOCs, LVOCs, SVOCs, IVOCs, and VOCs, respectively (Donahue et al., 2009; Schervish and Donahue, 2020). The resulting VBS pie charts for these compound groups and their mean contributions are shown in Fig. 6b–d and Table 2. Organic compounds with Csat of 104 µg m-3 made up the biggest mass contributions for MION-Br and MION-NO3, and the dominating Csat bin measured by Vocus was organic compounds with Csat of 106 µg m-3 (see Fig. 6a). Furthermore, Vocus observed much higher contributions of VOCs with Csat higher than 108 µg m-3, whereas MION-NO3 measured higher contributions of ELVOCs and ULVOCs with Csat lower than 10-8 µg m-3 (See Fig. 6a). And MION-Br and MION-NO3 observed comparable contributions of compounds with Csat between 10-7 and 107 µg m-3. We stress here that the fragmentation of organic compounds inside the Vocus may bias the Csat results towards higher volatilities.

IVOCs, which include generally less oxygenated VOCs, comprised the significant fraction of total organics (Vocus: 45.8 ± 5.4 %; MION-Br: 65.8 ± 8.5 %; MION-NO3: 56.3 ± 10.6 %), indicating substantial oxidation extent of the precursor VOCs, which made up 53.7 ± 5.5 % of the total organics measured by Vocus but much less by MION-Br (10.4 ± 8.2 %) and MION-NO3 (5.4 ± 2.4 %; see Fig. 6b–d and Table 2). SVOCs, which include slightly more oxygenated VOCs, constituted substantially (Vocus: 0.4 ± 0.2 %; MION-Br: 16.2 ± 4.9 %; MION-NO3: 23.9 ± 5.1 %) to the measured organic compounds. LVOCs and ELVOCs, which include OVOCs with higher oxidation degrees and mainly contribute to the growth of embryonic clusters in the atmosphere (Donahue et al., 2012; Bianchi et al., 2019), accounted for > 8 % of the corresponding total organics measured by MION-Br and MION-NO3, while ULVOCs, which include OVOCs with even higher oxidation extents that are the most effective drivers of pure biogenic nucleation (Schervish and Donahue, 2020; Simon et al., 2020), accounted for 0.5 ± 0.6 % of total organics measured by MION-NO3 (see Fig. 6b–d and Table 2). Differences in the contribution of these compound groups (Fig. 6b–d and Table 2) could be due to different sensitivities of the instruments towards organic compounds with varying oxidation extents (Riva et al., 2019).

With the complementary molecular information of organic compounds from Vocus, MION-Br, and MION-NO3, a combined volatility distribution was plotted to estimate the bulk volatility of all measured organic compounds (with the approach described in Sect. 2.2.1) at our measurement site (Fig. 7). The combined volatility distribution covers very well from VOCs to HOMs, with varying O : C ratios and volatility ranges (Fig. 7a). It therefore provides a more complete picture of the volatility distribution of gaseous organic compounds in this boreal forest. The average mass-weighted log⁡10Csat value representing the bulk of all measured gaseous organic compounds in this boreal forest was ∼ 6.1 µg m-3. In general, MION-NO3 measured > 91 % of the ULVOCs, while MION-Br measured > 70 % of the ELVOCs and Vocus measured > 98 % of the IVOCs and VOCs (Fig. S10). As we can see from Fig. 7b, the VOC class was found to be the most abundant (about 53.2 %), followed by the IVOC class (about 45.9 %), indicating that the bulk gaseous organic compounds observed in this boreal forest were relatively fresh, which is also consistent with the bulk molecular composition's relatively low oxidation extent. Differences in the bulk volatility of organic compounds between daytime (between 10:00 and 17:00) and nighttime (between 22:00 and 05:00) were not significant (Fig. S11). Given the location of the measurement station that is inside a boreal forested area, the gaseous organic compounds were expected to be dominated by VOCs and IVOCs. The abundance of the CH compounds such as terpenes (see Tables 1 and S2, Figs. S3 and S6) as well as less oxygenated VOCs (see Figs. 3 and S5) supports this conclusion. Although the condensable vapors (LVOCs, ELVOCs, and ULVOCs) only comprised about 0.2 % of the total gaseous organic compounds, they contribute significantly to forming new particles via nucleation and further particulate growth and mass via condensation in this boreal forest (Kulmala et al., 2013; Ehn et al., 2014; Mohr et al., 2019). The results from the combined VBS could provide a better basis to test and improve parameterizations for predicting organic compound evolutions in transport and climate models.