Addition of a fast GC to SIFT-MS for analyses of individual monoterpenes in mixtures

Soft chemical ionization mass spectrometry (SCI-MS) techniques can be used to accurately quantify volatile organic compounds (VOCs) in air in real time; however, differentiation of isomers still represents a challenge. A suitable pre-separation 10 technique is thus needed, ideally capable of analyses in a few tens of seconds. To this end, a bespoke fast GC with an electrically heated 5 m long metallic capillary column was coupled to a selected ion flow tube mass spectrometry (SIFT-MS) instrument. To assess the performance of this combination a case study of monoterpene isomer (C10H16) analyses was carried out. The monoterpenes were quantified by SIFT-MS using H3O reagent ions (analyte ions C10H17, m/z 137, and C6H9, m/z 81) and NO+ reagent ions (analyte ions C10H16, m/z 136, and C7H9, m/z 93). The combinations of the fragment ion relative 15 intensities obtained using H3O and NO+ were shown to be characteristic of the individual monoterpenes. Two non-polar GC columns (Restek Inc.) were tested: the advantage of MXT-1 was shorter retention times whilst the advantage of MXT-Volatiles was better temporal separation. Thus, it is possible to quantify components of a monoterpene mixture in less than 45 s by the MXT-1 column and to separate them in less 180 s by the MXT-Volatiles column. As an illustrative example, the headspace of three conifer needle samples was analysed by both reagent ions with both columns showing that mainly α-pinene, β-pinene 20 and 3-carene were present.


Introduction
Standard analytical methods used to identify and quantify volatile organic compounds (VOCs) in air, such as thermal desorption gas chromatography mass spectrometry (TD-GC-MS), are often time consuming and cannot be used to investigate temporal changes in chemically evolving systems.In contrast, soft chemical ionization mass spectrometry (SCI-MS) techniques, such as selected ion flow tube mass spectrometry (SIFT-MS) (Smith and Španěl, 2011a;Španěl et al., 2006) and proton transfer reaction mass spectrometry (PTR-MS) (Lindinger et al., 1998;Ellis and Mayhew, 2013;Smith and Španěl, 2011b), represent well-established real time tools to analyse a wide variety of VOCs in ambient air (Amelynck et al., 2013;de Gouw and Warneke, 2007;Rinne et al., 2005;Schoon et al., 2003) and in headspace of biological samples (Shestivska et al., 2015;Shestivska et al., 2011;Shestivska et al., 2012).The advantage of SIFT-MS and PTR-MS lies in the possibility of online, real-time analysis obviating sample collection and pre-concentration of VOCs.In these techniques, defined reagent ions (usually H3O + , NO + or O2 +• ) interact with trace VOCs present in gas samples introduced into a flow tube or a flow/drift tube.
The analytical ion-molecule reactions that produced analyte ions are variously proton transfer, adduct ion formation, charge transfer and hydride ion transfer, principally depending on the type of reagent ions used.This ion chemistry has been thoroughly reviewed in a number of publications, e.g.(Smith and Španěl, 2005).These ion-molecule reactions are not greatly exothermic and so few product (analyte) ions are produced in each reaction, often just one or two, that can readily be identified.
However, chemically similar molecules with the same atomic composition (structural isomers) usually produce identical analyte ions with similar branching ratios and therefore the neutral analyte molecules cannot be easily differentiated using SCI-MS alone (Smith et al., 2012).However, the reactions of the isomeric molecules may have different rate coefficients with the different reagent ions and lead to product ions at recognisably different branching ratios depending on their molecular geometry (Jordan et al., 2009;Pysanenko et al., 2009;Španěl and Smith, 1998;Wang et al., 2003).So the concurrent use of the available reagent ions in SIFT-MS analysis can sometimes be used to analyse and identify particular isomers.
Monoterpenes, mostly emitted from plants, are very important biogenic volatile organic compounds (BVOCs) in the atmosphere.Due to their high reactivity with atmospheric oxidants such as hydroxyl radicals (OH • ), monoterpenes reactions can lead to tropospheric ozone (O3) accumulation as well as to secondary organic aerosol formation, which can affects human health and contribute to global climate change (Chameides et al. (1992); Fehsenfeld et al. (1992); Kulmala et al. (2004)).
Although all monoterpenes comprise two isoprene units and have the same molecular formula, C10H16, their reaction time (or lifetime) with OH • and O3 widely varies from minutes to days (Atkinson and Arey, 2003).The values of the net BVOC/OH • reactivity measured in rainforests have been found to be higher than expected, which could be attributed to undetected monoterpenes or sesquiterpenes (Nolscher et al., 2016).Therefore, it is important to identify and individually quantify these BVOCs at their ambient trace levels.Gas chromatography mass spectrometry (GC-MS) coupled with pre-concentration techniques has been developed to successfully identify and quantify different atmospheric monoterpenes (Janson, 1993;Räisänen et al., 2009;Song et al., 2015).However, the requirements of pre-concentration and long cycle time (more than 1h) are obviously unsuitable for real-time measurements.
A promising approach to the near real time analysis of isomeric molecules is to combine both SCI-MS and fast GC methods.
Pre-separation provided by fast GC involves short columns with thin active layers, fast temperature ramps, fast injection systems and time resolutions below 5 min (Matisová and Dömötörová, 2003).Materic et al. (Materić et al., 2015) established a system using PTR-MS coupled with a fast GC to detect individual monoterpenes in air and achieved the separation of six most common monoterpenes at a limit of detection down to 1.2 ppbv.Pallozzi et al. then compared a fastCG-PTR-ToF-MS system with traditional GC-MS methods, discussing the limitations of the fast GC setup on some BVOCs emitted from plants, including monoterpenes (Pallozzi et al., 2016).SIFT-MS is also widely used in VOCs analyses (Allardyce et al., 2006;Smith andŠpaněl, 2005b, 2011b).It has well-defined analytical reaction conditions and the H3O + , NO + and O2 +• reagent ions can be switched rapidly to analyse time-varying trace gases in air samples.In the present article, we report the results of method

Construction of a fast GC device for pre-separation
The experimental setup of the bespoke fast GC setup constructed as an addition to SIFT-MS is shown in Fig. 1.In the experiments, two different GC columns were tested.First, a 5 m long nonpolar general-purpose chromatography metallic column MXT-1 (0.28 mm × 0.1 μm active phase, Restek Inc.) using dry air as the carrier gas, which was chosen according to the previous PTR-MS fastGC analyses (Romano et al., 2014).Additionally, a second, application-specific column for volatile organic pollutants, MXT-Volatiles (0.28 mm × 1.25 μm active phase, Restek Inc.) used with helium carrier gas.In order to facilitate direct resistive heating, the coil-shaped stainless steel columns (resistivity ∼4.2 Ω/m) were electrically isolated and connected to a regulated 60 V, 5 A DC power supply.Appearance of cold spots was suppressed by ensuring that the electrical current runs through the entire length of the columns.The temperatures of the columns were monitored by a K-type probe  The routing of the sample and the carrier gases was controlled by solenoid valves (Parker VSONC-2S25-VD-F, < 30ms response), labelled in Fig. 1 as EV1, EV2 and EV3.The needle valve NV1 was used in combination with an overflow relieve tube to fine-adjust the flow rate of the carrier gas (20-50 sccm from a gas cylinder regulator set to about 2 bar) so that the air pressure at the column entrance is held just above ambient.The region of the sampling input line, EV2, EV3 and their connection with the column are permanently heated to ∼60 °C to prevent adsorption of sample gas/vapour and to reduce memory effects.
Three modes of gas flow are possible as illustrated in Fig. 1: • The "normal mode": EV2 is open and both EV1 and EV3 are closed.Carrier gas flows through NV1, partly vented via the overflow relieve but mostly into the column.The pressure at the column entrance is just above that of the ambient atmosphere and a constant flow rate of clean carrier gas (synthetic air or helium) is thus achieved.
• The "sampling mode": EV1 and EV2 are closed and EV3 is open.Sample air is introduced into the column in a short time (1 to 8 s) after which the "normal mode" is resumed.• The "cleaning mode": All valves are open and the carrier gas taken directly from the cylinder regulator is introduced into the column (higher than normal flow) and purges the sample line via EV3.The overflow relieve flow rate is not sufficient to diminish the pressure.
The modes can be switched either manually or controlled from the SIFT-MS software.

The fast GC operation
The operation sequence for air analysis is as follows: A column is first heated up to 200 °C in the "cleaning mode" for three minutes prior to commencing the "normal mode" with an appropriate heating voltage setting (e.g. 10 V as shown in Fig. 2).
Whilst the column cools down, a pre-sampling interval (8-10 s "sampling mode", see Figure 2) is activated in order to refill the "dead volume" comprising the EV3 valve and the sampling inlet by air at its entrance.After the column reaches working temperature and a steady flow of clean carrier gas is established, the sample for actual analysis is introduced by enabling the "sampling mode" for 1 to 8 s.The GC separation then takes place over typically 60 -300 s whilst the eluent is continuously analysed by SIFT-MS.It is possible to apply a heating ramp during this period.
In the initial tests with the first generic MXT-1 column, the "sampling mode" duration was fixed at 1.8 s due to SIFT-MS software limitations.For the later tests with the second MXT-Volatiles column, the SIFT-MS operational software was upgraded to provide an arbitrary timing of the "sampling mode" duration.Sampling was repeated several times to improve sensitivity.
Several heating ramp profiles were tested (see data for MXT-1 column in Fig. S1 in the Supplement); however, due to the short GC column and relatively long injection time, the monoterpene chromatogram peaks coalesced when the column temperature exceeded 60 °C and it was found that optimal chromatograms were obtained isothermally at 40 °C (15 V heating voltage).Effects of the heating voltage on the retention time and the chromatogram profile are illustrated in Fig. S3 in the Supplement (data for MXT-Volatiles column).

SIFT-MS analyses of the eluent
In the present study, the Profile 3 SIFT-MS instrument (Instrument Science, Crewe, UK) was used (Smith et al., 1999).Reagent ions are formed in a microwave discharge through a mixture of water vapour and atmospheric air at a pressure of about 0.3 mbar (see Fig. 1).A mixture of ions is extracted from the discharge and focused into a quadrupole mass filter where they can be analysed according to their mass-to-charge ratio, m/z.Thus, the reagent ions H3O + , NO + or O2 +• can be selected (O2 +• was not used in the present experiment) and separately injected into flowing helium carrier gas (pressure p = 1.4 mbar, temperature T = 24 °C).Any internal energy possessed by the reagent ions is rapidly quenched in collisions with helium atoms leaving a thermalized ion swarm that is convected down the flow tube.Sample gas is introduced into the helium/thermalized swarm at a known flow rate that changes with the GC column temperature.The reagent ions react with the VOC molecules in the sample gas during a time period defined by the known flow speed of the ion swarm and the length of the flow tube.At the end of the flow tube, the reagent ions and the ionic products (analyte ions) generated by ion-molecule reactions are sampled by a pinhole orifice into the analytical quadrupole mass spectrometer.The count rates of the reagent and analyte ions are obtained using a channeltron multiplier.Thus, full scan (FS) spectra can be obtained over a chosen m/z range to identify the analyte ions or rapidly switched between selected m/z values using the multiple-ion monitoring mode (MIM) (Španěl and Smith, 2013;Smith and Španěl, 2011a).For the monoterpene study, the FS mode was used for SIFT-MS analyses, whilst the MIM mode was used for fast GC-SIFT-MS setup.

Reactions of H3O + and NO + reagent ions with monoterpenes
In the present study, SIFT-MS analyses of monoterpenes were carried out using the previously investigated reactions of monoterpenes with H3O + and NO +. ions (Schoon et al., 2003;Wang et al., 2003).The H3O + reactions are known to proceed via proton transfer forming C10H17 + (m/z 137) that partially fragments to C6H9 + (m/z 81) due to elimination of a C4H8 moiety from the nascent (C10H17)* excited ion: NO + reacts with monoterpenes by charge transfer forming the parent cation C10H16 +• (m/z 136) and a number of fragment ions, including C7H9 + : The exothermicity of charge transfer (2a) is represented by the difference between the ionization energies of the neutral NO (9.26 eV) and of the particular monoterpene (ranging from 8.07 eV for α-pinene to 8.4 eV for (R)-(+)-limonene) (Garcia et al., 2003;NIST).Other fragments, including C7H8 + , C7H10 + , C9H13 + and C10H15 + , are also formed and the branching ratios between the channels (2a) to (2b) and other fragments depend on the isomeric structure of the monoterpene (Schoon et al., 2003;Wang et al., 2003) and are given in Table S1 in the Supplement.Based on this known ion chemistry, for the present study it was decided to analyse monoterpenes using both the H3O + reagent ions by recording the C10H17 + (m/z 137) and C6H9 + (m/z 81) analyte ions and the NO + reagent ion by using the C10H16 + (m/z 136) and C7H9 + (m/z 93) analyte ions.To facilitate the identification of monoterpenes on the basis of the branching ratios of reactions ( 1) and ( 2), the product ion signal ratios [m/z 81]/[m/z 137] and [m/z 93]/[m/z 136] were determined under the conditions of the Profile 3 SIFT-MS instrument using standard monoterpene mixtures, and these ratios (r) are given in Table 1.
The interaction of the primary ions with monoterpenes may be affected by the presence of neutral water molecules, and thus by different humidity of the sample, as reported previously by Wang et al. (Wang et al., 2003) when decreased fragmentation of monoterpene product ions was observed in humid air samples.For H3O + regent ions, this change was significant for βpinene (r reducing from 0.75 to 0.51), (R)-(+)-limonene (r from 0.45 to 0.34) or 3-carene (r from 0.33 to 0.23).For the NO + regent ion, a significant effect was observed only for α-pinene (r from 0.32 to 0.08) and β-pinene (r from 0.25 to 0.05).
The weights (wi) applied to each of several discreet measurements were based on the total signal of both ions fi and gi in order to emphasise the area within the peak.Time intervals t1 to t2 were chosen for each isomer as the area of the chromatographic peak where the total ion signal was >10% of the peak value.
The quality of the ratio estimation was assessed from the variation of the fi/gi ratio estimated as: where µf and µg represent intensities of the selected fragments and   2 and   2 are the variances of the µf and µg intensities estimated according to the Poisson distribution as the sum of distribution variance equal to the expected value λ = µ and background variance   2 (Van Kempen and Van Vliet, 2000).
From this variation, the standard error of the weighted mean was calculated as: The weighted standard deviation of the fi/gi ratios was also routinely calculated as:
Thus, for individual standards, about 5 µl of each monoterpene was placed in a 2 ml vial closed by PTFE septum caps.Each vial was then penetrated with a diffusion tube (1/16" OD x 0.25 mm ID x 5 cm length PEEK capillary) and placed into a 15 ml glass vial closed by a PTFE septum.The headspace of the 15 ml vial was sampled after stabilization (> 30 minutes) of the concentration.Humidity of the headspace was typically 1.5% water vapour by volume as determined by SIFT-MS.For the αpinene, the intensities were too high and thus they had to be reduced by placing only trace amount of sample into the 2 ml vial.
For the mixture preparations, a similar approach was used; several vials containing different monoterpene liquid samples, To demonstrate the applicability of the fast GC/SIFT-MS analyses to real samples, three different species of coniferous tree needles were prepared: Spruce (Pincea punges), Fir (Abies concolor) and Pine (Pinus nigra) (see Fig. S4 -S6 in the Supplement).For the first study using the MXT-1 column, the needle samples (0.26 g Spruce, 0.42 g Fir and 0.32 g Pine) were collected in the urban area of Prague in June 2017 and stored in 10 ml vials from which the headspace was sampled.For the later study using the MXT-Volatiles column, pine tree twigs were collected in June 2018 from the same trees (21.8 g Spruce, 21.4 g Fir and 20.6 g Pine).The exposed cuts of the twigs were sealed by parafilm.The samples were placed into a Nalophan bag of volume approximately one litre.During the analyses, the laboratory was thermalized to the outdoor temperature (about 30 °C) to reduce thermal shock to the samples.

Results and discussion
To assess whether the various monoterpenes in a mixture could be effectively distinguished using SIFT-MS enhanced by the fast GC pre-separation, eight common biogenic monoterpenes were investigated, as identified in 3.3 above.The mixture of monoterpene standards was analysed using isothermal GC with two different columns at temperature of 40 °C.The elution times of all eight monoterpenes were within 45 s of total retention time for the MXT-1 column and within 180 s for the MXT-Volatiles column.Using the information on the ratios of the ion products for the H3O + and NO + reactions together with the GC retention times, it was possible to identify the components of the reference mixture.Finally, the same procedure was used to analyse the three fresh pine tree needle samples.

Comparison of columns: MXT-1 vs. MXT-Volatiles
The retention times determined from the chromatograms obtained for individual monoterpenes are given in Table 1 together with their   � values (see equation 3).For the MXT-1 column, the apparent difference in retention times observed between the two reagent ions was probably caused by the temperature fluctuations of the column.Whilst the retention times for individual monoterpenes are different, they are not sufficiently stable (fluctuate by > 1 s, see Table 1) in the present fast GC device for analyses based on retention time only to be reliable.Use of the MXT-Volatiles column resulted in about five times longer retention times and better GC peaks separation at the same operational conditions (flow rate, temperature and pressure) due to the higher efficiency of the 1.25 μm active phase (compared to 0.1 μm for MXT-1 column).Due to the different sampling times used with each column (1.8 s for MXT-1 and 5 to 12 s for MXT-Volatiles), the peak shapes cannot be compared directly but the peak width (FWHM) increased only two times for the MXT-Volatiles column.
The performance of both MXT-1 and MXT-Volatiles columns were compared by analyses of a gas mixture of the eight monoterpenes.For the MXT-1 column, four characteristic GC peaks were identified for both reagent ions, marked as A, B, C and D with retention time of 17.6 s, 20.8 s, 26.3 s and ∼30 s for H3O + , and 17.5 s, 20.7 s, 26.3 s and ∼30 s for NO + (see Fig. 3).pressures (see Table 1).Using the MXT-1 column under these conditions it was not possible to achieve separate GC peaks for individual monoterpenes, however qualitative analysis was possible.5  (Schoon et al., 2003); b (Wang et al., 2003); c Present result based on SIFT-MS measurement; d Present result based on fast GC-SIFT-MS measurement; saturated vapour pressures in Torr at 25 °C according to e (Daubert, 1989), f (Haynes, 2014), 10 g (Yaws, 1994), h (TGSC), i (Takasago, 2011), and at 20 °C according to j (ChemicalBook, 2016).The separation of the chromatographic peaks could be improved using hydrogen or helium as a carrier gas and by faster sample injection, as demonstrated by Materic et al. (Materić et al., 2015) with fastGC PTR-MS, where complete separation of monoterpenes was achieved.As observed for both columns, separation can be improved by decreasing the column temperature (see Fig. S3 in the Supplement), however this may increase the chromatogram width and thus decrease the sensitivity of the technique.Additional sensitivity can be achieved by increasing the injection time, which will, however, increase the peak width.In the present experiment we used heated columns isothermally to the temperature about 40 °C due to the behaviour of the MXT-1 column.For higher temperatures, the monoterpene chromatogram peaks coalesced.For lower temperatures a significant influence of lab air temperature fluctuations was apparent.Under these conditions for the MXT-1 column, The MXT-Volatiles column facilitates identification of all monoterpenes present in the mixture for temperatures close to room temperature (see Fig. S3 in the Supplement).For the MXT-Volatiles tests, the sampling mode was extended to 12 s, representing the collection of approximately 0.6 ml of the monoterpene mixture headspace.A noticeable effect of ambient temperature on the rate of passive column cooling was observed resulting in changes of the column temperature profile and thus in variations of the monoterpene retention times.It is interesting to note that the chromatogram (see Fig. S3 in the Supplement) changes with the temperature of the column and additional peaks appear at higher temperatures probably resulting from the presence of different conformers.It thus seems that at the column temperature ~45 °C using 20 V heating voltage (see Fig. 4) the small β-pinene is hidden behind the second camphene peak and the α-terpinene peak also disappears (see also the fragmentation analyses later in section 4.2).According to the elution time, the first chromatographic peak A consist of three monoterpenes: α-pinene, camphene and myrcene.For the H3O + reagent ions, the   � value corresponds to both α-pinene and myrcene considering the   � value for peak A (0.49) or   close to the peak maxima (0.55-0.6).However, a more obvious difference between α-pinene and myrcene is observed with the NO + reagent ions.The value of the weighted mean ratio for the peak A (0.21) is close to the ratio for αpinene.In the maxima of peak A, however,   approaches the value of 0.3, which is close to the value expected for a combination of both these monoterpenes (0.32, considering the data from fast GC measurement and the vapour pressure in Table 1).For camphene,   in the chromatograph did not reach the low values expected for both reagent ions.However, its presence is clearly visible as a dip in   situated between the peaks A and B. In the absence of camphene, the ratio should linearly move to values characteristic for the peak B without any dip.The depth of the dip does not reach the ratio expected for camphene due to a persistent tails of the peaks for both α-pinene and myrcene.
Peak B in the chromatograms is identified as β-pinene by its retention time.The   � values for the H3O + and NO + reagent ions are 0.63 and 0.21, respectively.The values   are similar to   � and slightly higher than to the fast GC standard values for βpinene (see Table 1).
Peaks C and D are not clearly separated in the chromatogram.For the H3O + reagent ions, the   � value is similar for both peaks; thus, the presence of (R)-(+)-limonene, 3-carene or α-terpenine is likely since the   � values for the peaks C (0.45) and D (0.4) are comparable with the analyte signal ratios (see Table 1) for (R)-(+)-limonene and 3-carene.A lower   for α-terpinene might be observed as a dip similar to that for camphene.However, the observed dip in   at the D peak is not as statistically significant as the dip for camphene, and the vapour pressure for both α-and γ-terpinene are lower than other monoterpenes.Analysis of the C and D peaks using the NO + reagent ion shows a clearer difference between them.The calculated   � for the peak C (0.27) as well as the maximum   (0.35) are, unexpectedly, much higher than for the remaining monoterpenes.This can be explained only by the influence of myrcene or by the presence of impurities in the form of an additional monoterpene in the mixture (for example ocimeme has high   of 0.62 (Wang et al., 2003)).Amongst the eight monoterpenes, 3-carene has the highest   within the retention time of peak C. The second peak D (0.14) can be then associated with (R)-(+)-limonene, which has a low   (0.06) for NO + reagent ions, with some contribution by α-terpinene.The presence of γ-terpenine is not visible due to its low vapour pressure, but there may be some contribution in the D peak, but much smaller than the contribution by (R)-(+)limonene.
To summarize, combining analyses using both H3O + and NO + reagent ions with dynamic variations of   allows the identification of α-pinene, camphene and myrcene in peak A followed exclusively by β-pinene in peak B. Peak C is characterized as 3-carene and peak D as (R)-(+)-limonene and/or α-terpinene.γ-terpenine contributes only weakly due to its low vapour pressure and has no recognisable response in the chromatogram compared to the remaining monoterpenes.NO + ) were again imprecise due to the low intensity and do not fully agree with the unique   � for myrcene (see Table 1).The observed weak peak could therefore be due to other monoterpenes other than those eight included in Table 1.The last peak corresponds to 3-carene with   � as 0.48 for H3O + and 0.16 for NO + reagent ions  The pine sample chromatogram shows three clear peaks of α-pinene (0.73 for H3O + , 0.30 for NO + ), β-pinene (0.92 for H3O + , 0.26 for NO + ) and 3-carene (0.49 for H3O + , 0.13 for NO + ) with just a very small and statistically insignificant indication of camphene.The retention times for α-pinene, β-pinene and 3-carene were 69.6 s, 97 s and 141 s, respectively.
Some differences can be seen between the results from the MXT-1 and MXT-Volatiles columns.The most significant difference is the presence of a camphene peak in the fir sample headspace, and the presence of β-pinene and 3-carene in the pine sample headspace when the MXT-Volatiles column was used.However, samples were collected at different times of the year and the character of the samples was also different (only needles for MXT-1 and whole twigs for the MXT-Volatiles analyses).

Comparison with previous studies
The present experiments indicate that using the fast GC-SIFT-MS combination, it is possible to achieve only qualitative analysis of the monoterpene mixture with a limit of the detection of about 100 ppb.This is inferior to the previously described fastGC-PTR-MS systems (Materić et al., 2015;Pallozzi et al., 2016), which achieved full separation with limit of the detection up to 1-2 ppt.However, one advantage of SIFT-MS is the facility to use two reagent ions, and the analysis of product ion ratios provides additional information.Thus, the combination of the data from the two reagent ions together with the analyses of the product ion signal ratios ri can be shown to improve the identification of monoterpenes.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.development aimed at selective analyses of individual monoterpenes in mixtures in air using a bespoke fast GC/SIFT-MS combination with H3O + and NO + reagent ions.This involved the analysis of both prepared laboratory monoterpene/air mixtures and headspace of the foliage of different pine trees.

Figure 1 :
Figure 1: Schematic visualization of the fast GC-SIFT-MS experiment.Coloured dashed lines in the inlet part of the fastCG represent gas flow through the system of the valves EV1-3.The blue line traces the "normal mode" regime, the green line represents the "sampling mode" and the red line represents the "cleaning mode".

Figure 2 :
Figure 2: Left: the applied heating voltage (dashed) and the temperature profile of the column (red) during the fast GC cycle.The pulses indicate the opening of the valve EV3 during the pre-sampling and the sampling periods.Right: The increase of the column temperature and the related decrease of the carrier gas flow rate with the heating voltage.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.3.2Analysis of the product ion intensity ratiosTo facilitate assignment of the fast GC elution peaks to specific monoterpenes, mean fragment ion fractions   = fi/gi = [m/z 81]/[m/z 137] (or for NO + ,   = fi/gi = [m/z 93]/[m/z 136]) were calculated for each interval of retention times t1 to t2, as the weighted mean of the product ion signal ratios   � : Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.penetrated by PEEK capillaries, were placed together into a 500 ml bottle.Note that the concentrations of the individual isomers in the mixture are different due to the variations in their saturated vapour pressures.
Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.Based on the retention times obtained for individual monoterpenes (see Fig. S2 in the Supplement), peak A is due to co-elution of α-pinene, camphene and myrcene.Peak B is due to the presence of β-pinene exclusively and peaks C and D are due to the remaining four monoterpenes.Note that the individual peak heights are influenced by the monoterpene saturated vapour Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.

Figure 3 :
Figure 3: Chromatograms of the mixture of monoterpenes (upper figures) measured by H3O + (left) and NO + (right) reagent ions, obtained using the MXT-1 column.A, B, C, D represent characteristic peaks in the chromatogram.For each chromatogram, the product ion signal ratio ri is presented in the lower figures.The grey data background represents the calculated standard deviation of the data by Savizky-Golay smoothing between 15 s and 40 s.The position and value of the ratio for individual monoterpenes is based on the fast GC MXT-1 measurements presented in Table 1.Note that the retention times are determined by the fast GC conditions and do not depend on which SIFT-MS reagent ion is used.

Figure 4 :
Figure 4: A chromatogram for a prepared mixture of monoterpenes obtained with the MXT-Volatiles column as derived by SIFT-MS using H3O + reagent ions.For the chromatogram, the product ion signal ratio ri is presented in the bottom figure.This chromatogram was obtained at the column temperature ~45 °C using 20 V heating voltage.

Figure 5 :
Figure 5: Chromatograms derived using the product ions for the reactions of H3O + (upper row) and NO + (lower row) reagent ions 5

Figure 6 :
Figure 6: SIFT-MS selected ion mode/fast GC/SIFT-MS chromatograms for monoterpene emissions from pine tree samples (s1, s2 and s3) obtained using the MXT-Volatiles column.The upper and lower rows were obtained using H3O + and NO + reagent ions respectively.The signal intensities are the analyte ion count rates normalized to a reagent ion count rate of 10 6 s -1 .The black and red curves stand for monitored ions     + (m/z 81) and     + (m/z 137) for H3O + reagent ions of     + (m/z 93) and     + (m/z Atmos.Meas.Tech.Discuss., https://doi.org/10.5194/amt-2019-12Manuscript under review for journal Atmos.Meas.Tech.Discussion started: 21 March 2019 c Author(s) 2019.CC BY 4.0 License.