Measurement of Atmospheric Volatile and Intermediate Volatility Organic Compounds: Development of a New Time-of-Flight Mass Spectrometer

Abstract

A new gas inlet port combined with a novel ionization scheme have been developed and coupled to a high-resolution time-of-flight mass spectrometer (TOF MS) for the detection and measurement of atmospheric volatile (VOCs) and intermediate volatility organic compounds (IVOCs). Ions are produced predominantly by charge transfer reactions in a low-temperature plasma ionization source with minimal fragmentation. Enhanced sensitivity is accomplished by incorporating an increased-size inlet capillary in a transverse arrangement to maximize throughput in the ionization source. Additional design aspects of the new mass spectrometer enabling superior transmission include a large acceptance ion funnel and a segmented radio frequency (RF) ion guide with increased space charge storage capacity. An orthogonal TOF analyzer equipped with a two-stage reflectron and tuned to second order is employed for the determination of the mass-to-charge ratio of the ions, with a mass resolving power of >20 k at mass 500 Th. The performance of the instrument was evaluated in tests using VOC standards and in atmospheric chamber experiments to demonstrate the ability to measure a wide range of organic compounds with different functional groups. Linear signal response is demonstrated over a wide range of VOCs used in the calibration processes in the ppb range, while the instrument exhibits linear response in the ppt range as well. Detection limits as low as 1 ppt are accomplished. The potential applications of this new TOF MS instrument were demonstrated in a pilot atmospheric simulation chamber experiment.

1. Introduction

Thousands of organic compounds found in the troposphere are related to anthropogenic or natural emissions and cover a wide range of sizes and functional groups [1]. A general classification is done based on their volatility characteristics by separating them into volatile organic compounds (VOCs) and intermediate-volatility compounds (IVOCs), found almost exclusively in the gas phase and in semi-volatile organic compounds (SVOCs), coexisting in the gas and particulate phases [2]. These compounds are involved in atmospheric chemical processes that alter their chemical and physical characteristics. The larger ones also contribute to the formation of atmospheric aerosols and thus affect human health, visibility, and climate [3]. VOCs are emitted from both natural (vegetation, oceans, etc.) and anthropogenic sources (biomass burning, energy production, transportation, etc.). They are key pollutants for ozone formation [4,5]. The characterization and the management of air quality require not only the quantification of the pollution levels but also the determination of the emission sources, including source type (transportation, industry, agricultural activities, biomass burning, etc.), source location (local versus transported pollution), and atmospheric chemical processing (primary versus secondary pollutants, with the latter produced chemically in the atmosphere). Several attempts have been made to quantify the sources of volatile and/or particulate organic matter [6,7,8,9,10].

While there have been significant steps in our understanding of inorganic air pollutants, the chemical complexity of organics and their continuous chemical reactions still limit dramatically our ability to detect them with traditional analytical techniques and to also address the corresponding air quality problems. Filter, gas canister, and sorbent material sampling techniques have provided useful information about the organic components found in ambient air, though the resolution of the measurements was limited to daily averages, or specific short time periods, overlooking the constantly evolving nature of some of these organic compounds. [11,12,13,14,15,16,17,18,19]. Semi-continuous ‘online’ techniques, such as online gas chromatography coupled to mass spectrometry (GCMS) or FID detectors, partially filled this gap with typical sampling intervals in the range of 20 min to 1 h, but still cannot achieve the necessary time resolution for fast-evolving processes, especially in atmospheric simulation chamber experiments [20,21]. On the positive side, improvements in tandem MS instrumentation provide a more precise identification of the analyzed compounds, while mass spectral libraries listing a plethora of organic compounds are being continuously updated [NIST/EPA/NIH EI-MS Library, 2020 release].

The characterization and quantification of organic pollution requires the development of new instruments and high-quality technologies providing detailed information about the different chemical compounds. Several novel ideas have been implemented in the field of atmospheric pollution, both for the determination of the gas phase and also of the particulate organic components in ambient air. Proton transfer reaction mass spectrometry (PTR MS) [22] addressed many of the challenges mentioned above, especially in the area of gas phase organic composition, and several studies have implemented the PTR MS technique for characterizing the VOC content of the atmosphere [23,24,25,26], helping to update emission models and calculate diurnal fluxes and VOC annual emissions [27]. Moreover, semi-continuous measurements of the particulate phase with appropriate inlets (Charon inlet) have also extended our knowledge of the composition of organic aerosols [28,29,30,31,32]. Other methods based on chemical ionization have also been successfully used (ionization of gas and particulate species using, for example, iodide-adducts) for online measurement of the gas phase composition and the semi-continuous measurement of the particulate composition (FIGAERO inlet) [33,34,35]. Selected ion flow tube mass spectrometry (SIFT MS) is also a powerful technique for detecting and quantifying trace gas species that utilizes a wide range of reagent ions to optimize the ionization reaction process in a controlled manner, enhancing its ability to detect a broad range of compounds [36,37].

The use of the above techniques has allowed us to improve our understanding of the various air pollutants. Analysis of the resulting high-resolution data enabled the assignment of the detected species to specific pollution sources through the use of positive matrix factorization (PMF) or by identifying tracer compounds [38]. For PMF, the measured VOC concentrations are used to derive factors that explain the observed variability and can be connected to sources without requiring the measurement of fluxes via eddy covariance or any other flux measurement techniques. Besides ambient air measurements, the use of these instruments has been extended into other areas, such as gas phase kinetic measurements [39], chamber experiments [40] that simulate specific atmospheric processes, indoor air quality [41], detection of hazardous compounds for security reasons [42], and breath analysis for medical reasons, with the latest development being COVID-19 detection applications [43].

A low-temperature plasma is a non-equilibrium system of partially ionized gas created by free electrons at low gas temperatures. Also present in the plasma are reactive species, such as metastable atoms and molecules, radicals, and high-energy photons. Typically, the produced radicals and other excited species fall to lower (more stable) energy states with the release of UV radiation and heat [44]. The use of several gas combinations has been studied for the production of low-temperature plasma ranging from typical synthetic air (mixtures of N₂, O₂, CO₂) to pure gases (e.g., N₂, H₂O, or H₂,) or noble gases (Ne, Ar, He) [45,46,47]. Besides the use of water vapor for VOC ionization through proton transfer reactions, other ions, such as NO⁺, O²⁺, and NH₄⁺, are able to generate organic ions, mainly via charge transfer reactions. The following schemes have been proposed related to VOC ionization through the use of oxonium ions (H₂O)H⁺ and/or other exited species: proton transfer reactions leading to the formation of the protonated VOC molecule as [M+H]⁺ (1), dissociative proton transfer reactions where the parent VOC (e.g., MX) fragments into one or more parts that can be ionized through proton transfer as [M+H]⁺ (2), and association reactions, where the ionized VOC molecule appears in the form of [M+(H₂O)H]⁺ (3); use of O₂⁺ ions, where the VOC is ionized through charge transfer reactions leading to the formation of the cation M⁺ (4), or dissociative charge transfer reactions leading to the fragmentation of the parent VOC (5); use of NO⁺ ions, where the VOC molecule can undergo charge transfer reactions (6); or hydride ion transfer, which can produce a [M-H]⁺ ion, probably due to charge transfer forming M⁺ initially and, subsequently, loss of a hydrogen (7); or hydroxide ion transfer (8); or three-body association reactions (9) [48].

M + (H₂O)H⁺ → [M+H]⁺ + H₂O

(1)

MX + (H₂O)H⁺→ [M+H]⁺ + X + H₂O

(2)

M + (H₂O)H⁺→ [M+(H₂O)H]⁺

(3)

M + O₂⁺ → M⁺ + O₂

(4)

MX + O₂⁺ → M⁺ + X + O₂

(5)

MH + NO⁺ → [M+H]⁺ + NO

(6)

MH + NO⁺ → [M-H]⁺ + HNO

(7)

MOH + NO⁺ → [M-OH]⁺ + HNO₂

(8)

M + N₂ + NO⁺ → [M+(NO)]⁺ + N₂

(9)

Ongoing developments in the field of MS instrumentation continue to provide improvements in speed, detection limits, mass resolving power, and tandem MS capabilities. Here the focus is on the development of a state-of-the-art TOF MS platform equipped with a novel ionization source for the measurement of organic compounds with a wide range of volatilities, ranging from VOCs to IVOCs, and extending to gas phase SVOC species. The extremely low detection limits accomplished with this new design are highlighted. A detailed description of the instrument enabling high-resolution TOF measurements is provided.

2. Instrument Description

Key features of the ionization source and all critical ion optical elements of the new instrument are presented in Figure 1. The ionization source is operated in the low mbar pressure range and comprises (a) of a first set of electrodes to confine the nitrogen/water vapor at low pressure, low-temperature plasma, forming a primary flow of reagent ions in the axial direction; (b) a high-temperature, high-throughput inlet capillary for sampling ambient air, forming a secondary flow of neutral molecules in the transverse direction; and (c) an interaction zone where mixing of the two flows leads to charge transfer and proton transfer reactions between reagent ions produced in the plasma and analyte species entrained in the secondary flow with high efficiency. An ion funnel, with the entrance end adjacent to the interaction zone, receives and focuses product ions through a differential aperture into downstream RF optics. Ions are subsequently transferred and accumulated at the end of a hybrid RF ion guide designed with different field-orders and multiple segments for enhanced transmission across a wide range of pressures (10⁻³ to >10⁻² mbar). Radial confinement at lower pressure is enabled further by an additional RF hexapole ion guide followed by a DC lens to form a planar ion beam and match the requirements of the orthogonal TOF mass analyzer for high resolution mass-to-charge ratio measurements. A more detailed description of the different sections of the instrument and related processes is provided below.

2.1. Ionization Source

Ionization of organic vapors is accomplished via reactions with reagent ions produced in a low-temperature plasma ionization source within a pressure range of 1 to 5 mbar. Figure 1 provides a cross-sectional view of the ionization source. The source comprises a set of closely spaced electrodes to generate the low-temperature plasma using relatively low voltage settings (<500 V). Radial confinement of the plasma is accomplished by ceramic rings installed between the electrodes. A ceramic tube with a 2 mm i.d. supported on the first plate electrode delivers a carrier gas (high-purity nitrogen) inside the electrode structure, forming a primary low-pressure plasma jet sharing a common axis with downstream optics. The flow of the carrier gas is regulated through a stainless-steel micro needle valve and flows between 0.1 and 2 L min⁻¹ can be achieved. The penetration depth of the plasma jet is pressure dependent and can be extended to overlap with the transverse gas jet introduced through the sampling inlet. The carrier gas is humidified upstream of the ionization source by passing nitrogen through a high-purity water reservoir enclosed in a stainless-steel container. The presence of water vapor in the plasma results in the formation of oxonium ions (H₂O)H⁺, which can subsequently transfer protons to VOCs and IVOCs. In addition to the proton transfer reactions driven by the concentration of oxonium ions, the background mass spectrum includes ions such as NO⁺, O₂⁺, and NO₂⁺, which can participate in charge transfer reactions via electron exchange with analyte molecules. The relative concentration of NO_x⁺ ions can be partly controlled by adjusting the flow rate of oxygen into vacuum through the sampling inlet. Similarly, the relative concentration of oxonium ions can be affected by the flow rate of the carrier gas. Proton and/or charge transfer ionization is observed based on the chemical properties of each analyte. Proton transfer reactions dominate for species such as acetonitrile or monoterpenes. In contrast, chlorobenzene and most aromatic species undergo predominantly charge transfer ionization.

The sampling inlet for introducing ambient air in the vacuum is a 70 mm long stainless-steel transfer tube designed with a 0.7 mm internal diameter and typically operated at 250 °C. Temperature is controlled by a pair of heater cartridges attached to an aluminum bronze block supporting the tube. The gas flow rate through the inlet at operating temperature can be in the 0.5 to 1.5 L min⁻¹ range. The supporting block is designed with a lock-in mechanism whereby the transfer tube can be removed without venting the mass spectrometer in order to facilitate regular cleanups of the inlet system. Instrument downtime is therefore minimized, and performance restored by eliminating memory effects. Upstream of the transfer tube, a PTFE filter (PTFE 0.22 µm) is used to prevent particulate matter from entering the instrument and a micro needle valve controls the sample flow rate.

The gas jet from the sampling inlet established at the exit of the transfer tube is radially confined by a carefully dimensioned duct [49], controlling dispersion of the low-pressure flow and increasing the concentration of VOCs and other analyte species in the interaction zone of the ionization source. The flow of gas from the sampling inlet is directed across the interaction zone and into a 40 mm diameter pumping port connected to a 35 m³ h⁻¹ dry scroll pump (Edwards, XDS35i). The transverse arrangement of the sampling inlet allows transfer tubes with a larger i.d. to be utilized without significantly affecting the operating pressure of downstream ion optics. In addition, the ion funnel and the first differential aperture surfaces are kept clean and free of residues, which is a typical problem for inlet systems sharing a common axis with downstream ion optics.

2.2. Radio Frequency Transfer Line

The RF transfer line is designed to receive ions from the ionization source and produces a collimated ion beam at the entrance of the TOF mass analyzer. The ion optical components of the transfer line are tailored to the different pressure regimes to maximize transmission through differential apertures. The ion funnel is operated at >1 mbar pressure and consists of a stack of 0.4 mm thick stainless-steel rings with 0.5 mm spacing. The acceptance area of the electrodes is progressively reduced from 25 mm diameter to 2.5 mm, followed by a 2 mm, DC-only exit aperture to maintain a pressure difference while extending the low-mass cut-off to the lower m/z region for the detection of background ions formed in the low-temperature plasma. A pair of PCBs is integrated onto the structure of the ion funnel to distribute RF and DC signals through a resistor–capacitor network. A pair of antiphase waveforms with a frequency of 3 MHz and a maximum amplitude of 250 V are applied to the two phases of the stack. A DC gradient is also established across the ion funnel to guide ions towards the differential aperture. To control undesired effects associated with heavy space charge conditions in RF devices, the DC signal applied to the differential aperture at the end of the ion funnel is switched between a transmission voltage level and deflection voltage level. The continuous ion beam produced in the ionization source can therefore be converted into packets while the number of ions made available in every scan can be controlled by adjusting the timing sequence of the experiment through the instrument control software.

A segmented RF ion guide with an octapole structure is installed in the subsequent compartment evacuated by a water-cooled 400 L s⁻¹ turbomolecular pump (Edwards, nEXT series) and operated in the pressure range of 10⁻³ mbar to 10⁻² mbar. The RF octapole is designed to provide an octapole field-order at the entrance for enhanced phase space acceptance to accommodate ions jetting out from the higher-pressure region of the ion funnel and having strong radial velocity components. To enhance radial compression and maximize transmission through the next differential aperture, the octapole structure is wired to create a quadrupole field-order across the end segments of the device. Axial segmentation allows operating the octapole ion guide over an extended range of pressures to enhance sampling and ion transfer efficiency from the ionization source to the TOF mass analyzer. In parallel, segmentation allows ions to be trapped and compressed axially to improve gating into downstream optics. The subsequent exit lens installed at the end of the octapole ion guide, designed with a 2 mm aperture to maintain a pressure difference, is similarly switched between two DC levels. The first level enables storage and accumulation of ions while the second enables the release and synchronization with the pulsed extraction scheme of the TOF mass analyzer. The final RF hexapole ion guide is installed in the subsequent compartment evacuated by an 85 L s⁻¹ turbo molecular pump (Edwards, nEXT series) and operated at 5 × 10⁻⁵ mbar to reduce pressure further in the TOF section.

2.3. Time-of-Flight Mass Analyzer

The TOF analyzer consists of a two-stage orthogonal gate and the time-focus formed in the field-free region of the analyzer is reproduced by a two-stage reflectron onto the detector plane. The two-stage orthogonal gate and the two-stage reflectron are both tuned to second order according to the principles of space focusing [50]. A set of DC lenses is designed to shape the ions into a planar beam. Ions with stray trajectories are prevented from entering the analyzer by a dual 2 mm slit. The electrodes of the gate are designed with parallel wire grids running in the direction of the incoming ion beam. The mass resolving power of the TOF mass analyzer is enhanced by minimizing the velocity component of the ions perpendicular to the wire-grids. This is accomplished by controlling the voltage settings applied to the DC lens geometry. The axial energy of the ions is 20 eV with an axial energy spread of <5 eV and for a 10 kV acceleration in the TOF direction, the flight angle is approximately 87 degrees. Parallel wire grids are also used to define electric fields in the two-stage reflectron geometry. Ions are detected using a magneTOF electron multiplier (ETP) with a pulse width of 400 ps and the signal is sampled by a 14-bit, 2 GS s⁻¹ analog-to-digital converter (Teledyne–SP Devices). The overall length of the TOF is approximately 2 m and the full-width, half-maximum mass resolving power achieved is 20,000 at mass-to-charge 500 Th. Pressure in the TOF is maintained at 10⁻⁷ mbar by a 400 L s⁻¹ turbomolecular pump (Edwards, nETX).

3. Instrument Performance

In this section, we present examples of detection of VOCs typically found in ambient air with a focus on the sensitivity and quantification specifications achieved with the new instrument. The underlying ionization mechanism and the extent of fragmentation observed are discussed. A general evaluation of the quantification process and other stress tests are also presented.

3.1. Ionization of Species

A series of standard compounds were sampled to gain insights into the ionization processes in the new instrument. Tests were performed using a gas standard mixture of these compounds in N₂ at concentrations in the range of 15 ppb (14 to 16 ppb depending on the compound). The following compounds were examined: acetonitrile, acrolein, acetone, isoprene, methacrolein, 2-butanone, benzene, toluene, o-xylene, chlorobenzene, α-pinene, and dichlorobenzene. The measured mass spectra for each of these compounds are shown in Figure 2. Additional information on the measured compounds presented in

Table 1. Organic compounds investigated in this study and related information.

Compound	Chemical Formula	Theoretical Masses for Monoisotopic Peak and the Second Isotope (Percentage Abundance)	Formulas for Ionic Compounds Observed Experimentally (Theoretical Masses for the Monoisotopic Peak and the Second Isotope)	Experimental m/z of Detected Peaks (Percentage Abundance)	Δppm
Acetonitrile	C2H3N	41.0265 (100) ǂ	[C2H3N+H]+ (42.0338)	42.0342 (100)	9.5
		42.0299 (2.2)	[C2H3N+H]+ (43.0371) *	N/D	N/A
Acrolein	C3H4O	56.0262 (100) ǂ	[C3H4O+H]+ (57.0335)	57.0333 (100) Ɫ	−3.5
		57.0295 (3.2)	[C3H4O+H]+ (58.0368) *	58.0366 (0.6)	−3.4
Acetone	C3H6O	58.0418 (100) ǂ	[C3H6O+H]+ (59.0491)	59.0493 (100) Ɫ	3.4
		59.0452 (3.2)	[C3H6O+H]+ (60.0525) *	60.0529 (2)	6.7
Isoprene	C5H8	68.0626 (100) ǂ	[C5H8-H]+ (67.0542) C5H8+ (68.0621) [C5H8+H]+ (69.0699)	67.0546 (30) 68.0617 (99) 69.0697 (100) Ɫ	6 −5.9 −2.9
		69.0659 (5.4)	[C5H8-H]+ (68.05760) * C5H8+ (69.0654) * [C5H8+H]+ (70.0732) *	N/D N/D N/D	N/A N/A N/A
Methacrolein	C4H6O	70.0418 (100) ǂ	C4H6O+ (70.0413) [C4H6O+H]+ (71.0491)	70.0409 (1) 71.0488 (100) Ɫ	−5.7 −4.2
		71.0452 (4.3)	[C4H6O+H]+ (72.0525) *	N/D	N/A
2-butanone	C4H8O	72.0575 (100) ǂ	C4H8O+ (72.0570) [C4H8O+H]+ (73.0648)	72.056 (5) 73.0644 (100) Ɫ	−13.9 −5.5
		73.0608 (4.3)	[C4H8O+H]+ (74.0681) *	74.0677 (1.6)	−5.4
Benzene	C6H6	78.0469 (100) ǂ	C6H6+ ( 78.0464) [C6H6+H]+ (79.0542)	78.0467 (100) Ɫ 79.0543 (16.5)	3.8 1.3
		79.0503 (6.5)	C6H6+ (79.0498) *	N/D	N/A
Toluene	C7H8	92.062 (100) ǂ	C7H8+ (92.0621) [C7H8+H]+ (93.0699)	92.0621 (100) Ɫ 93.0698 (43)	0.0 −1.1
		93.0659 (7.6)	C7H8+ (93.0654) * [C7H8+H]+ (94.0732) *	N/D N/D	N/A N/A
O-xylene	C8H10	106.0782 (100) ǂ	C8H10+ (106.0777) [C8H10+H]+ (107.0855)	106.0784 (100) Ɫ 107.0849 (45)	6.6 5.6
		107.0816 (8.7)	C8H10+ (107.0811) * [C8H10+H]+ (108.0889) *	N/D 108.087 (26)	N/A −17.6
Chlorobenzene	C6H5Cl	112.0079 (100)	C6H5Cl+ (112.0074)	112.0079 (100) Ɫ	4.5
		113.0113 (6.5)	2nd*C6H5Cl+ (113.0108) *	113.0118 (6.7)	8.8
		114.0050 (32.4)	3rd*C6H5Cl+ (114.0045) **	114.0051 (13.5)	5.3
α-Pinene	C10H16	136.1252 (100) ǂ	C10H16+ (136.1247) [C10H16+H]+ (137.1325)	136.1258 (14) 137.1326 (100) Ɫ	8.1 0.7
		137.1285 (10.8)	[C10H16+H]+ (138.1358) *	BQL	N/A
Dichlorobenzene	C6H4Cl2	145.9690 (100) ǂ	C6H4Cl2+ (145.9685)	145.9695 (100) Ɫ	6.9
		146.9724 (6.5)	C6H4Cl2+ (146.9718) *	146.9732 (3.9)	9.5
		147.9661 (64.8)	C6H4Cl2+ (147.9655) **	147.964 (55)	−10.7
		148.9694 (4.2)	C6H4Cl2+ (148.9689) ***	148.9697 (1.5)	5.4
		149.9631 (10.5)	C6H4Cl2+ (149.9626) ****	N/D	N/A

ǂ monoisotopic peak. Ɫ peak used for calibration. *: second isotope of compound. **: third isotope of compound. ***: fourth isotope of compound. ****: fifth isotope of compound. N/D: not detected. N/A: not available. BQL: Below quantification limit.

includes the chemical formula, theoretical and experimental masses for the monoisotopic peak, the second isotope, and the corresponding mass accuracy error. While Table 1 shows mass numbers with four-digit precision, for simplicity reasons we use single mass precision in the text. Additionally, protonated ions [M+H]⁺ are expressed as [C_xH_y+H]⁺ while for those produced via charge transfer to form the cation M⁺, the preferred expression is C_xH_y⁺, keeping track of the ionic compound formulas in each case. Finally, the [M-H]+ ion is depicted as the parent compound minus one hydrogen, e.g., [C_xH_y-H]⁺. The theoretical monoisotopic mass (the first peak of every isotopic distribution) of the ions observed experimentally is depicted without an asterisk. The second highest peak of every isotopic distribution is depicted with an asterisk as C_xH_y⁺ * (or is stated in the text). The percentage abundance of the isotopes and the theoretical masses of the compounds observed experimentally are provided in brackets. The theoretical masses of the isotopes listed in Table 1 were generated using an online calculator (https://www.envipat.eawag.ch/, accessed on 4 February 2023) [51].

Acetonitrile (C₂H₃N) was detected in the protonated form [C₂H₃N+H]⁺ (m/z 42) (Figure 2a), indicating that the molecule is ionized exclusively by proton transfer reactions. No peak was observed at m/z 41 that would indicate a charge transfer ionization mechanism C₂H₃N⁺. Additionally, the second isotope (m/z 43) expected with a relative abundance of less than 2% with respect to the main peak was below the noise threshold (Table 1). Acrolein (C₃H₄O) was also detected almost exclusively in the protonated form [C₃H₄O+H]⁺ (m/z 57) (Figure 2b). For the second (3%) isotope of protonated acrolein [C₃H₄O+H]⁺ the corresponding signal at m/z 58 was 0.6% of the main peak. Acetone (C₃H₆O) was detected in the clean air used in our experiments (Linde, synthesized clean air, H/C free) at m/z 59 (protonated form, [C₃H₆O+H]⁺). The corresponding signal increased, as expected, for the VOC sample (Figure 2c). The protonated second isotope of acetone ([C₃H₆O+H]⁺) at m/z 60 (Table 1) was 2% of the monoisotopic peak (with literature values indicating a 3.2%).

Three isoprene-related ions were detected (Figure 2d). The non-protonated C₅H₈⁺ and protonated [C₅H₈+H]⁺ ions yielded similar signals at m/z 68 and 69, respectively (Table 1). The peak at m/z 69 attributed here to the protonated ion [C₅H₈+H]⁺ does not contain the second isotope C₅H₈⁺. One additional peak was observed at m/z 67.0546 that has been attributed to the [C₅H₈-H]⁺ ion, an indication of isoprene losing one hydrogen possibly undergoing a hydride ion transfer, as indicated by reaction (7) in Section 1. A peak at 69.0067, highlighted with a different color in Figure 2d, was identified as a fragment ion and will be discussed in the next section.

Methacrolein (C₄H₆O) was detected in the protonated form [C₄H₆O+H]⁺ (m/z 71), though a small percentage (1% of the protonated signal) was also observed at m/z 70 in the radical cation form, C₄H₆O⁺ (Figure 2e, Table 1). A similar behavior was observed for 2-butanone (C₄H₈O) with the protonated form [C₄H₈O+H]⁺ dominating at m/z 73 and the signal for C₄H₈O⁺ at m/z 72 accounting for less than 5% of the protonated ion signal (Figure 2f). Additionally, a small peak at 74 (1.6%) was found and is related to the protonated second isotope of 2-butanone, [C₄H₈O+H]⁺ (Table 1).

The ionization of aromatic compounds such as benzene (C₆H₆), toluene (C₇H₈), and o-xylene (C₈H₁₀) was manifested through the loss of one electron (Figure 2g–i), producing the following radical cations: C₆H₆⁺, C₇H₈⁺, and C₈H₁₀⁺, respectively. A limited degree of protonation was also observed in all cases (Table 1). More specifically, for benzene, the signal at m/z 79 is 16.5% of the principal signal at m/z 78 C₆H₆⁺, which is significantly higher than the 6.5% expected due to the natural relative abundance of the isotopes based on the theoretical isotopic distribution of the cation (Table 1). This suggests that the benzene molecules are protonated [C₆H₆+H]⁺, with a 10% efficiency relative to charge transfer ionization via electron exchange. Toluene ions are detected at m/z 92 (C₇H₈⁺) as a principal peak and m/z 93 corresponding mainly to the protonated form [C₇H₈+H]⁺ and less at the toluene second isotope C₇H₈⁺ (at 43% related to the principal peak), as shown in Table 1. The expected percentages would be 100% and 7.6% respectively if no proton transfer reactions took place. This indicates that roughly a third of the detected toluene ions are in the protonated form. o-Xylene follows a similar pattern with benzene with the ions detected at m/z 106 (monoisotopic peak or 100%), m/z 107 (45%), and 108 (26%). The expected percentages in the absence of proton transfer reactions would be 100%, 8.8%, and 0.3%, respectively, which again shows that more than a third of o-xylene ions are produced via protonation. Some background peaks are also seen (Figure 2i) in the spectrum of o-xylene, which are not related with this molecule. These peaks were detected in zero air as well and are discussed below.

Finally, α-pinene (C₁₀H₁₆) was mostly detected in the protonated form [C₁₀H₁₆+H]⁺ at m/z 137 and only a small fraction (14%) was ionized via charge transfer as C₁₀H₁₆⁺ at m/z 136 (Figure 2k and Table 1). A low-intensity peak was also observed at 81.0704 corresponding to the C₆H₉⁺ fragment of α-pinene.

The loss of an electron to form a cation rather that the addition of a proton is strongly favored for compounds such as chlorobenzene and dichlorobenzene. For chlorobenzene (C₆H₅Cl), the main peaks detected (Figure 2j) were observed at m/z values 112 (monoisotopic peak 100%) corresponding to C₆H₅Cl⁺, m/z 113 (6.7%) corresponding to the second isotope C₆H₅Cl⁺, and m/z 114 (13.5%) corresponding to the third isotope of C₆H₅Cl⁺, which are close to the expected theoretical values of 100%, 6.5%, and 32%, respectively (Table 1). For dichlorobenzene (C₆H₄Cl₂) a similar ionization pattern to chlorobenzene was seen with the formation of the M⁺ radical cation. The reported results and the literature values for the isotopes of the compound (Table 1) show good agreement between the monoisotopic C₆H₄Cl₂⁺ and the second, third, and fourth isotopes. The detected ions (Figure 2l) were at m/z 146 (100%), m/z 147 (3.8%), m/z 148 (55%), and m/z 149 (1.5%). This is close to the expected values of 100%, 6.5%, 64.2%, and 4% (Table 1), though no peak was observed at m/z 150. Overall, chlorinated aromatic compounds do not appear to undergo proton transfer reactions.

3.2. Fragmentation of Detected Molecules

Higher molecular weight compounds and compounds that are more difficult to ionize appear to undergo dissociation in the ionization source, producing fragment ions with a wide range of efficiencies. A series of chamber experiments were performed for three sesquiterpenes (C₁₅H₂₄), namely β-caryophyllene, α-humulene, and δ-cadinene sharing common properties in terms of the ionization scheme but exhibiting differences in the fragmentation patterns observed experimentally. For β-caryophyllene, major peaks were observed at m/z 204 corresponding to the radical cation M⁺, at m/z 205 corresponding to the protonated form [M+H]⁺, and at m/z 206, which corresponds primarily to the second isotope of the protonated form [C₁₅H₂₄+H]⁺. Besides these parent ions, β- caryophyllene ion signals were observed at m/z 147, 148 and 149. These ions are also present in the electron ionization mass spectrum obtained from the NIST library for β-caryophyllene (NIST Chemistry WebBook, https://webbook.nist.gov/chemistry, accessed on 30 October 2022), indicating a similar fragmentation process, presumably from the radical cation parent ion. Additionally, signals were detected at m/z 175, 176, 161, and 189 and are also present in the electron ionization mass spectrum of β-caryophyllene available in the NIST library. Finally, a peak was seen at m/z 222.25. The production mechanism for this ion, which is at a higher value than the parent molecule, is unclear.

Similarly to β-caryophyllene, the α-humulene parent ions were detected at m/z values 204 corresponding to the M⁺ species (C₁₅H₂₄⁺), 205 to the protonated ions [M+H]⁺ or [C₁₅H₂₄+H]⁺, and 206 corresponding to the second isotope of the protonated form or [C₁₅H₂₄+H]⁺. However, no fragment ions were observed in the mass spectrum of α-humulene. Nevertheless, ions were detected at higher m/z values, namely 221, 222, and 223. These could be due to a water cluster addition to the parent molecule or are some byproducts of the ionization process due to molecular rearrangement reactions.

For δ-cadinene, the parent ion peaks were similarly observed in the radical cation form M⁺ at m/z 204, in the protonated form [M+H]⁺ at m/z 205, and the heavier isotope at m/z 206, similarly to the β-caryophyllene and α-humulene. The formation of the m/z 203 ion is attributed to a loss of a hydrogen atom during the ionization process. Abundant δ-cadinene fragments were also detected at m/z 161, which is also present in the electron ionization mass spectrum of δ-cadinene (NIST).

The results produced for structural isomers of sesquiterpene ions such as α-humulene and δ-cadinene show that it is possible to differentiate these compounds in a mass spectrum based on the observed fragmentation pattern, while it may also be possible to identify and quantify these compounds in complex mixtures based on the relative abundance of the different fragment ions. This approach would require producing a mass spectral library of all the member compounds of the sesquiterpene family. It is noted that for the detection of the sesquiterpenes, the settings applied to the ion funnel and to the RF-hexapole were tuned to maximize transmission for ions in the 200 m/z range, which led to information loss in the m/z region below 100 due to a change in the low-mass cut-off. Any additional fragments appearing in the lower m/z region of the mass spectrum would therefore not be observed and are not reported here.

Hexane and heptane were also detected using the new instrument, but with significant fragmentation. These two molecules showed different fragmentation patterns. For n-hexane (C₆H₁₄, MW 86.17), the main ions detected were 57, 59, 67, 69, 85, and 101. For n-heptane (C₇H₁₆, MW 100.21), they were: 55, 57, 59, 69, 70, 71, 72, 74, 83, 85, 88, and 97. Figure 3 shows the related spectra obtained for n-hexane and n-heptane.

Due to the lack of reliable standards in the IVOC range, we focused mainly in the VOC range since there is extensive relevant literature to evaluate the performance of the instrument and the efficiency of the new ionization source in terms of sensitivity and degree of fragmentation produced.

While a multimode ionization scheme may complicate the interpretation of the results when certain features of the mass spectrum must be attributed to a single compound, such complex features may also help to clarify/validate the result since, for example, a principal peak at m/z 78.0464 corresponding to benzene must be accompanied by a peak at 79.0542 with a specific relative abundance. Variations in the relative abundance of isotopes different from those expected using the new setup would point out the presence of an isobaric interference. It is emphasized, however, that this approach is prone to variations in the pressure of the ionization source and to the RF settings applied to the ion funnel, which affect the relative abundance of the cation M⁺ and the [M+H]⁺ species.

The low-abundance isotopes of a specific VOC are generally overlooked and largely excluded in the literature during data interpretation, which may lead to loss of useful information. For example, in the case of benzene, a high-resolution PTR TOF instrument should detect not only the protonated benzene peak at m/z 79.0542, but also a peak at 80.0576 with a 6.5% relative intensity, something that is rarely used for validation.

3.3. Quantification and Limit of Detection

For most of the air pollutants discussed in Section 3.1, five series of calibration experiments were performed on different days in order to check the linearity and reproducibility of the results at 20 ms total injection time. The most abundant peak of a given compound was used for mass calibration. Different detection limits were observed depending on the compound examined. For example, a linear response (R² of 0.99) was produced for compounds such as benzene within the ppt range, while methacrolein was below quantification limits in the low ppt range (e.g., below 200 ppt). In general, the limit of detection is in the 100 ppt range, while for certain compounds it can be extended to below 10 ppt. Figure 4 shows calibration curves produced for different compounds in the (a) ppb range (1–20 ppb) as well as (b) in the ppt range (25–1000 ppt). The use of log scale in the concentration was selected for better illustration in the ppt range. A linear response was observed in the ppb range with R² > 0.96, reaching a maximum of 0.99 for most of the compounds tested. The linearity in the ppt range was reduced with the value of R² < 0.95 for most of the compounds (Figure 4b). Although acetone was not included in these measurements due to the high background concentration, a linear response was also observed for this compound in the ppb range. Acetonitrile (m/z 45) was excluded from this set of measurements due to mass discrimination effects induced by RF fields in the very low m/z range. Figure 4c shows the instrument response for benzene at different concentrations in the ppt range. Based on the intensity of the ion signal and using a signal-to-noise ratio of 3, concentrations of benzene as low as 1.4 ppt could be detected. The radical cation of benzene, C₆H₆⁺, is one of the best compounds to demonstrate the extremely low levels of detection accomplished in the new instrument.

3.4. Background Signals and Stress Tests

Besides the instrument evaluation experiments using the targeted VOCs and IVOCs, a series of background ions formed in the low temperature plasma were observed in the mass spectrum. The strongest peak present during the analysis of hydrocarbon-free air is attributed to NO₂⁺ (45.993). Additionally, the [C₃H₆O+H]⁺ ion (m/z 59) related to background concentrations of acetone and the [C₂H₄O₂+H]⁺ ion (m/z 61) related to background concentrations of acetic acid, both in the protonated form, were also detected in zero air (hydrocarbon-free, zero air cylinder). In most of the measurements performed with the new system so far, the low-mass cut-off was adjusted to m/z ~30 and background ions below this threshold were not observed. As a result, the concentration of oxonium (H₂O)H⁺ ions (m/z 19) for the different settings applied to the ionization source were not explored in detail. Nevertheless, by dropping the RF amplitude applied to the ion funnel and further adjusting parameters in the RF transfer line, oxonium ions are visible in the mass spectrum. The water cluster of the oxonium ion (H₂O)₂H⁺ at m/z 37 was also readily detected. Finally, NO⁺ (m/z 30) and O₂⁺ (m/z 32) ions with masses close to the low-mass cut-off were also observed; however, these species were not quantified.

Additional ions were observed in the background mass spectrum and were present both when sampling clean air and/or also when sampling high-purity nitrogen. These ions may originate from the outgassing of materials installed in the ionization source or from organic solvents used for cleaning parts during the instrument assembly process. A dry pump was employed to control pressure in the ionization source and eliminate back-streaming of oil contaminants. The m/z values of the main ions observed in the background mass spectrum are: 43.0039 (possibly related to small alkanes), 47.0090 (possibly related to ethanol and or formic acid), 63.0002, 88.0071, 88.9908, 89.9316 and 90.9112 (of unknown origin), 106.9531, 107.9340, and 108.9190 (possibly related to aromatic ring compounds other than the o- xylene). Finally, a group of ions with the m/z values of 117.9391 118.9198, 118.9495, 119.9304, 120.9356, 121.9319 121.9994, 122.9148, 122.9390, 123.8995, 123.9415, 124.9294, and 125.9139 can be attributed to trimethyl benzene and other similar poly-substituted aromatics that were used in high concentrations during the development of the prototype instrument. This contamination tends to be persistent even though the gas sampling line is operated at 100 °C and is regularly flushed with high-purity N₂. An additional peak observed at m/z 69.0067, which appeared during the analysis of the VOC gas standard, remains unassigned.

The performance of the instrument was also investigated under both typical and elevated concentrations of gas pollutants found in urban environments, such as NO and NO_2, as well as the usual oxidants found in the atmosphere, such as O₃ and OH radicals. Stress tests performed using the FORTH atmospheric simulation chamber at varying levels of NO, NO_2, and O₃ showed no effect on the accuracy of concentration measurements using stable levels of toluene and performed over prolonged periods of time.

4. Application Examples

A smog chamber experiment of the ozonolysis of β-caryophyllene under dark conditions was performed to demonstrate the process-monitoring capabilities of the new instrument. β-caryophyllene reacts rapidly with ozone to form products that can be found both in the gas and particulate phases. Here, only the gas phase products were examined as the sampled air was filtered upstream of the MS inlet and the ionization source. The experimental procedure has been described in detail [52], and only a brief summary is provided here. Initially, the chamber was filled with clean air, free of NO_x, O₃, CO, VOCs, and particulate matter. After a period of measuring the background concentrations, ammonium sulfate particles in the 100 nm diameter range were introduced into the smog chamber through the atomization of a 1 g L⁻¹ water solution. The ammonium sulfate particles facilitate the condensation of secondary organic aerosol components produced in the chamber. After an equilibration period, β-caryophyllene was added by injecting 2 µL of the liquid compound into a heated jet or air releasing into the smog chamber, establishing a 22 ppb concentration (based on the amount injected). Due to the difficulty in obtaining reliable β-caryophyllene gas standards, signal intensity was used instead of the actual concentration values. After another equilibration/mixing period, O₃ was injected at a concentration of 175 ppb with the use of an ozone generator. As soon as O₃ was introduced, rapid depletion of β-caryophyllene was observed due to its fast reaction with O₃. At the same time, the oxidation products of these reactions are distributed between the gas and particulate phases. An increase of the particles’ volume concentration was monitored by a Scanning Mobility Particle Sizer (SMPS, TSI) and a simultaneous increase in the ion signal of the organics detected using a High-Resolution Time-of-Flight Aerosol Mass Spectrometer (TOF-AMS, Aerodyne). On the other hand, volatile and intermediate volatility products were present in the gas phase. Besides the first-generation products, next-generation oxidation products were also formed, as seen by the developed prototype. Figure 5 shows the detected ions related to β-caryophyllene over the course of this experiment. The ions shown here, both for the parent molecule and the fragments of β-caryophyllene, have been discussed in Section 3.2. The rapid depletion of β-caryophyllene is apparent in the signal attributed to the parent molecule, validating peak assignment Figure 5a–d. Some of the produced species detected in the gas phase are shown in Figure 5e–h.

Several studies related to the ozonolysis of β-caryophyllene have been carried out over the last decade [40,53,54,55,56] focusing on the gas and particulate products under different reaction conditions. In our study, rapid formation of products is apparent across several m/z values, with the most distinctive peaks at m/z 235, 253 and also 177, 197, 207, and 272. The ion at m/z 235 has been reported by Gao et al. (2022) as well as in other studies [40,51,53] and is attributed to C₁₅H₂₂O₂.H⁺ formed in the early stages of oxidation of β-caryophyllene, which is subsequently consumed by further reactions (Figure 5e). The signal at m/z 253 is attributed to C₁₅H₂₄O₃.H⁺ and has also been reported as a possible β-caryophyllene oxidation product [40,53,54,55,56]. The peak at m/z 272 is formed with a considerable delay after the O₃ addition as a second-generation product and is detected partially in the gas phase. The peaks observed at m/z 272, 177, and 207 have not been positively identified. Further studies are necessary under various conditions (different concentrations of β-caryophyllene and O₃, temperature, the presence of NO_x, etc.) to better characterize the oxidation process; however, these fall outside the scope of this manuscript.

5. Discussion and Conclusions

A new prototype instrument for the measurement of VOCs and IVOCs has been developed and evaluated experimentally. Ionization of the organic pollutants in low-temperature plasma is achieved either by proton transfer reactions due to the presence of oxonium ions or by charge transfer reactions due to the presence of NO⁺ and O₂⁺ ions. The prototype mass spectrometer described in this work has several new and distinguishing features compared to existing PTR TOF MS technology. Most notable are (a) the transverse arrangement in the ionization source permitting the use of an enlarged inlet with increased sampling flow rate (adjustable within a range of 0.2–5 L min⁻¹), leading to enhanced sensitivity; (b) a high-efficiency charge transfer ionization scheme established in the extended volume occupied by the low-temperature plasma, occurring in parallel to proton transfer reactions driven by the presence of oxonium ions; (c) a segmented RF ion guide operable over an extended range of pressure settings to accommodate the high pressure in the ionization source and to bunch the continuous ion beam received from upstream ion optics; and (d) a high-performance TOF analyzer with a mass resolving power of 15,000 (fwhm) at mass-to-charge 250 Th and with a mass accuracy of typically lower than 10 ppm in the lower m/z region.

For typical organic pollutants found in ambient air, minimum fragmentation has been observed. Initial measurements show enhanced capabilities for the detection and identification of VOCs and IVOCs down to the ppt level. The detected ions cover a broad range of organic molecules, including aromatic species, terpenes, oxygenated VOCs, and aliphatic compounds. For the latter, alkanes can also be detected, though fragmentation is observed. The ozonolysis of ẞ-caryophyllene is also presented as an application example for of the new instrument.