Cl₂⁻ chemical ionization mass spectrometry (Cl₂-CIMS) for the measurement of acyl peroxy radicals

Abstract

Organic peroxy radicals (RO₂) are produced in the atmosphere by oxidation of volatile organic compounds (VOCs) and, in some cases, VOC photolysis. However, photolytic sources of RO₂ are often poorly understood, in part due to challenges in directly detecting RO₂ in both ambient and laboratory settings. We investigated Cl₂⁻ as a chemical ionization mass spectrometry reagent ion (Cl₂-CIMS) for measuring and speciating RO₂ in a laboratory setting. Cl₂-CIMS was more sensitive to the acetyl peroxy radical (CH₃C(O)O₂; 2.30 ± 0.04 ncps/ppt) than iodide CIMS (I-CIMS; 1.54 ± 0.03 ncps/ppt), but high backgrounds in our setup resulted in a slightly higher detection limit of 5 ppt (1 second integration) for Cl₂-CIMS than I-CIMS (2 ppt). We demonstrate the application of Cl₂-CIMS by quantifying the quantum yields of two radical products, CH₃C(O) and C₂H₅C(O), from methyl ethyl ketone photolysis at 254 nm. We identified O₂⁻ and Cl⁻ as possible secondary reagent ions that created unintended product ions in our experiments and thus could complicate the interpretation of Cl₂-CIMS mass spectra for complex atmospheric samples. While several strategies may minimize these effects, Cl₂-CIMS is suitable for measuring RO₂ in controlled laboratory experiments.

---

Environmental significance

Organic peroxy radicals are intermediates in the atmospheric oxidation cycles of organic molecules which impact production of atmospheric free radicals and ozone. Speciated measurements of organic peroxy radicals in the atmosphere are difficult but possible through recent advancements in chemical ionization mass spectrometry. We developed a chemical ionization mass spectrometry method using Cl₂⁻ as the primary reagent ion for detection of select organic peroxy radicals. We demonstrate this technique through laboratory measurements of the acetyl and propionyl peroxy radical quantum yields from methyl ethyl ketone photolysis. Cl₂⁻ reagent ion CIMS may be applicable to ambient measurements of organic peroxy radicals, if methods are developed for better isolation of Cl₂⁻ chemistry from reactions of secondary reagent ions.

Introduction

Organic peroxy radicals (RO₂) are key components of atmospheric oxidation chemistry but are challenging to measure due to their high reactivity and low ambient concentrations (∑RO₂ ≤ 40 ppt).^1,2 RO₂ reactions control hydroxyl (OH) radical production and can contribute to radical termination reactions, thus influencing oxidative capacity of the atmosphere and impacting air quality.³ RO₂ are primarily formed from the oxidation of volatile organic compounds, though certain moieties, such as aldehydes and ketones, can also undergo photolysis to produce RO₂.⁴

Approaches to measuring RO₂ are primarily spectroscopic, including ultraviolet³ and infrared absorbance⁵ measurements and cavity ring-down spectroscopy.⁶ Laser-induced fluorescence measurements are possible following the conversion of RO₂ to OH through reaction with CO and NO.⁷ Chemical amplification of RO₂ is also possible through reaction with CO and NO, or C₂H₆ and NO, to produce and measure NO₂.^8,9 However, these spectroscopic techniques typically cannot quantify speciated radicals and are often limited to measuring HO₂ in the lab¹⁰ (HO₂ can be considered the smallest RO₂ species, given its analogous sources and sinks) or summed concentrations of many RO₂ in the field.¹¹

There is a clear need for new analytical techniques to directly measure and speciate RO₂. Ambient measurements would be improved through the development of a method that could detect individual RO₂ species in the field. Ambient concentrations of RO₂ other than CH₃O₂ and HO₂ are typically a few ppt or less.¹²

Better RO₂ measurements would also improve laboratory data on RO₂ production from organic gas photolysis. Photolytic sources of RO₂ become more important relative to oxidation reactions in the upper troposphere where oxidants (OH, O₃, NO₃, Cl) are scarce and the actinic flux is higher compared to the rest of the troposphere. For example, acetone photolysis is thought to produce HO_x (OH and HO₂) at a similar rate to the O(¹D) + H₂O reaction in the upper troposphere.^13,14 However, the role of acetone in upper tropospheric HO_x production is poorly understood, as laboratory studies of acetone photolysis quantum yields are limited to indirect measurements of the dominant products: acetyl peroxy (CH₃C(O)O₂) and methyl peroxy (CH₃O₂) radicals. This results in substantial uncertainties in acetone photolysis quantum yields.¹⁵ Prior studies typically measure either minor photolysis product channels¹⁶ or stable products (e.g., CO, CO_2, and peroxyacetyl nitrate) of secondary reactions.^17–19 Photolysis quantum yields for larger and more complex ketones are even more limited, as few techniques can speciate CH₃C(O)O₂ and CH₃O₂ radicals from RO₂ of different structures with fast time resolution.¹⁰

One technique that can provide fast, direct, and speciated detection of RO₂ is chemical ionization mass spectrometry (CIMS). CIMS has been used to measure RO₂ in both laboratory and ambient settings. Direct CIMS detection methods include measurements of HO₂ with I⁻ (I-CIMS)²⁰ and Br⁻ reagent ions,^21,22 acyl peroxy radicals (R(O)O₂) and other RO₂ with I-CIMS,^23–26 and highly oxidized molecule RO₂ with NH₄⁺, CH₃C(O)O⁻,²⁷ or C₃H₇NH₃⁺ reagent ions.²⁸ Additional derivatization-aided detection methods exist for H₃O⁺, NH₄⁺,²⁹ and NO₃⁻, but are typically labor-intensive.^12,30 We recently described the use of I-CIMS to measure ketone photolysis product quantum yields via direct detection of RC(O)O₂ produced from photolysis of methyl ethyl ketone (MEK) and 2,3-butadione.²³

However, available CIMS reagent ions, including iodide, are often subject to interferences that impact RO₂ detection such as decreases in sensitivity due to small amounts of water vapor,²² unintended production of radicals inside the instrument,²¹ and secondary ion chemistry that complicates quantification.³¹ The sensitivities of CIMS reagent ions, such I⁻ and Br⁻,^22,25 can be temperature dependent as well. Current detection limits for RO₂ at short integration times (∼1 s) are often too high (>5 ppt) for ambient measurements. Laboratory measurements of ketone photodissociation quantum yields are also challenged by high detection limits. Investigation into alternate CIMS reagent ions to measure speciated RO₂ is warranted.

We describe a reagent ion system for CIMS with Cl₂⁻ as the primary reagent ion (Cl₂-CIMS). We investigate the ion-molecule chemistry and demonstrate the application of Cl₂-CIMS to measurements of RC(O)O₂ from photolysis of acetone and MEK at 254 nm. Through organic acid and woodsmoke sampling, we evaluate the potential and limitations of Cl₂-CIMS in ambient measurements.

Methods

The chemical ionization mass spectrometer

We used a time-of-flight chemical ionization mass spectrometer with atmospheric pressure interface (CIMS; Aerodyne Research, Inc.), which consists of a standard Aerodyne Research Inc. designed ion-molecule reaction region (IMR), ion optics, and a time-of-flight mass analyzer with a multi-channel plate detector.^32,33 In the IMR, sample air interacts with a flow of reagent ions, which ionize analyte molecules through clustering, proton transfer, charge transfer, or other reactions. We operated the CIMS with a typical IMR pressure of 65 mbar. The ion optics consist of one small quadrupole and one large quadrupole, plus multiple skimmers and a segment of ion lenses, operated at increasing voltage (in this case, increasing negative voltages) as sample air moves from the high-to low-pressure regions. These sections maximize ion transport through the instrument as the total pressure decreases to 1 × 10⁻⁶ mbar in the time-of-flight region. The time-of-flight mass analyzer had a mass resolution of 3500 m/Δm (Δm = 0.04 at m/z 144.95, the m/z for Cl₂(CH₃C(O)O₂)^–), adequate for determining the elemental composition of detected ions in the size range described here.

We introduced reagent ion flow through the ionizer port on the IMR (Fig. 1), with a sealed Po-210 α-particle source (Model P-2021; NRD, LLC) ion source. Cl₂-CIMS reagent ions were formed by flowing trace concentrations of Cl₂ through the ion source. We used 12 L glass bulbs filled to 1250 mbar with 100 to 400 ppm Cl₂ in ultra-high purity N₂ (99.999%; Airgas) and calibration gas cylinders containing 10 or 40 ppm Cl₂ in N₂ (Gasco) as the reagent Cl₂ sources. A critical orifice limited Cl₂ flows at 10 mL min⁻¹ from the 12 L glass; flow from the Cl₂ cylinders was adjusted by a precision needle valve. We used N₂, mixtures of N₂ and O₂ (99.994%; Airgas), or ultra-zero air (UZA; 99.999%; Airgas) as carrier gases for the Cl₂ reagent mixture. Needle valves controlled total carrier gas flow at >1250 mL min⁻¹, which diluted Cl₂ mixing ratios to ≤ 1 ppm in the ionizer. Unless otherwise stated, we used ∼100 ppb Cl₂ in UZA carrier gas for Cl₂-CIMS reagent ion generation.

Fig. 1 Schematic of flows to introduce Cl₂⁻ reagent ions into the IMR of the CIMS. Describes the major CIMS components and the ionizer; flows of Cl₂, N₂, humidified N₂, zero air, and oxygen; and sample flow. The dashed lines and arrows denote the flow path used to produce I⁻ reagent ions from CH₃I in N₂. This path was not permanently attached to the Cl₂ flow system but replaced the Cl₂ flow through the ionizer during I-CIMS operation.

For I-CIMS, we generated I⁻ by passing a constant flow of N₂ over a CH₃I permeation tube heated to 40 °C through the ionizer. The constant flow and temperature on the permeation tube were maintained when I-CIMS was not in use. We estimate CH₃I concentration at the ionizer to be on the order of 500 ppb. We controlled water vapor partial pressure in the IMR through an additional port by changing the ratio of dry and water-saturated N₂ while keeping total flow constant. The humidified line was typically maintained at 80% RH, providing ∼0.5 mbar H₂O to the IMR.

Calibrations

We calibrated the Cl₂-CIMS for ozone (O₃), formic, acetic, propanoic, and isobutyric acids, and CH₃C(O)O₂. A calibrated O₃ source (Model 306, 2B Technologies) generated O₃ between 25-200 (±2.0%) ppb in a 1600 mL min⁻¹ flow of air, which we diluted with UZA to 2 L min⁻¹ before sampling by the Cl₂-CIMS. Custom permeation tubes containing liquid organic acids were held at 40 °C and served as the acid calibration source. The permeation tube emission rates were determined gravimetrically. We used a custom peroxyacetyl nitrate (PAN) source³⁴ for CIMS calibration to CH₃C(O)O₂. The PAN calibration source photolyzes acetone in UZA to form CH₃C(O)O₂ and CH₃O₂. These radicals then react with NO to form NO₂, CH₃C(O)O and CH₃O. The NO₂ reacts with the remaining CH₃C(O)O₂. In the source we mixed a 60 mL min⁻¹ flow of 18 ppm acetone in UZA (Scott-Marin) with 2 mL min⁻¹ of 2 ppm NO in N₂. NO mixtures were prepared in 12 L glass bulbs. The light source was a phosphorus-coated Hg lamp (Jelight), with primary emission at 285 nm. Photolysis of the acetone-NO mixture occurred in a 200 cm³ quartz cell at 1070 mbar and approximately room temperature, producing a 60 mL min⁻¹ flow containing 40 ppb of PAN. The PAN concentration output of the device was calibrated by GC-ECD prior to our use. Further details on the custom PAN source are available in Flocke et al.³⁴ We diluted this PAN flow into 2 to 5 L min⁻¹ N₂ and thermally decomposed PAN at 150 °C to produce CH₃C(O)O₂ mixing ratios from 0.48 to 0.96 ppb. A heating rope around a 0.25 in O.D. glass sampling tube 1 cm from the CIMS inlet provided the necessary heating for PAN thermal decomposition.²³ We calibrated the Cl₂-CIMS to analytes at a Cl₂ reagent mixture concentration of 100 ppb. Sensitivities are reported for IMR water partial pressure of 0.5 mbar and a total pressure of 65 mbar. All calibration standards were measured by the Cl₂-CIMS as a function of (i) Cl₂ ionizer concentrations (0.1 to 1.5 ppm), (ii) ion-molecule reactor (IMR) pressure between 50 and 100 mbar, (iii) IMR water partial pressure between 0 and 0.8 mbar, and (iv) de-clustering potential³⁵ between the front of the second quadrupole and the adjacent (up-flow) skimmer plate (Fig. S1†).

Sensitivities were calculated as the slope of the line of detected ion signal versus input concentration and reported as normalized ion counts per second per part per trillion (ncps/ppt; per 1 × 10⁶ Hz of summed reagent ions). Unless indicated otherwise, Cl₂-CIMS signals are normalized to the sum of Cl⁻, Cl₂⁻, Cl₃⁻, and O₂⁻, the identified potential reagent ions. I-CIMS signals are normalized to the sum of I⁻ and I(H₂O)^–, the two primary reagent ions for this chemistry. Typical total reagent ion counts for both chemistries were between one and two million Hz. However, the reagent ion counts for Cl₂-CIMS varied, dependent on reagent gas O₂ content, and other instrument conditions as described in calibration tests (i)–(iv) above. All signals were normalized to total reagent ion counts, as described here, for comparisons across changing instrument conditions. Limits of detection (1 s integration) are calculated as three times the background signal-to-noise ratio and assuming a poisson distribution for background count rates.^23,32

Ketone photolysis product quantum yields

To evaluate whether Cl₂⁻ was a better reagent ion for laboratory radical measurements, we compared measurements of CH₃C(O)O₂ and propionyl peroxy radicals (C₂H₅C(O)O₂), generated from photolysis of acetone and MEK, acquired with Cl₂-CIMS versus I-CIMS. In these experiments, ketone photolysis generated acyl peroxy radicals ((R1a) and (R1b)):

RC(O)CH₃ + hv → RC(O) + CH₃ (R1a)

RC(O) + O₂ + M → RC(O)O₂ (R1b)

We produced the MEK and acetone standards in 12 L glass bulbs containing percent-level ketone in N₂ taken from the headspace of purified liquid ketones. Ketone samples in vacuum storage flasks were purified by freezing the liquid and pumping out the headspace. Ketone bulb concentration was determined through absorbance of 254 nm light (Hg Pen-Ray lamp; Analytik Jena) in a 50 cm glass cell with quartz end windows. We measured absorbance by bulb contents relative to background intensity in dry N₂ in triplicate, using a lab-built photodiode detector.

The photolysis reactor was a 1 in O.D. glass tube with three 0.25 in O.D. ports for the addition of photolyte mixture and sampling by the CIMS and exhaust. A Hg Pen-Ray 254 nm lamp was inserted 3 in into the reactor in a 0.5 in O.D. quartz test tube. The CIMS sampled from the photolysis reactor 1 cm from the end of the Pen-Ray lamp through an inserted 0.25 in glass tube.

During photolysis experiments, 10 mL min⁻¹ of ketone in N₂ was mixed with 7 L min⁻¹ of 21% O₂ in N₂ before entering the reactor adjacent to the Pen-Ray lamp housing. We measured the background flow of O₂ and N₂ before turning on the Pen-Ray Lamp for 100 s and then adding the ketone. After sampling during ketone photolysis for 5 min, we turned off the lamp, turned off the ketone flow, and sampled for at least one minute before turning the lamp on again. Experiments were run with both acetone and MEK. The Cl₂-CIMS detected CH₃C(O)O₂ primarily as the cluster Cl₂(CH₃C(O)O₂)^– (R2):

CH₃C(O)O₂ + Cl₂⁻ → Cl₂(CH₃C(O)O₂)⁻ (R2)

By maintaining constant photolysis reactor conditions over the course of the experiment, we determined MEK photolysis product quantum yields from the ratio of Cl₂(RC(O)O₂)^– signals between the two photolysis steps without needing to consider the reaction time of photon flux components of the photolysis rate coefficient eqn (E1).²³where S represents the Cl₂-acylperoxy radical cluster ion signals produced during photolysis steps (Hz), σ and φ are the terms for ketone 254 nm absorption cross section and RC(O) photolysis product quantum yield, respectively, and [acetone] and [MEK] are ketone concentrations in the photolysis reactor. Details on experiments performed to ensure negligible impact of radical–radical and wall loss reactions on RC(O)O₂ yields with this reactor and experimental design are described by Link et al.²³

Woodsmoke experiments

The Cl₂-CIMS sampled woodsmoke from a 1 m³ Teflon chamber that had ports for air sampling, injection of smoke, and a constant flow of 3 L min⁻¹ zero air (produced by zero air generator; Model 7000, Environics). Following the method described in Li et al.,³⁶ we burned ∼1 g of Douglas fir woodchips in a Breville (BSM600SILUSC) cocktail smoker and injected the smoke directly into the chamber. This injection produced approximately 300 μg m⁻³ of smoke aerosol along with concentrated gases that could be analyzed by the CIMS. We connected the CIMS to a sampling port through ∼1 m of PFA tubing (0.25 in O.D.) with a PTFE filter to remove particles. No lights were used to induce chemical aging of the smoke. A manual three-way valve allowed zero air to bypass the chamber directly to the CIMS. During measurements, we first flowed only zero air into the chamber for a background measurement before injecting smoke. The Cl₂-CIMS sampled smoke from the chamber for 2 hours. We then switched to I-CIMS mode and performed the same woodsmoke experiment with a new smoke addition. For analysis of woodsmoke samples, we identified ions that increased during smoke sampling with m/z < 300 for Cl₂-CIMS and < 350 for I-CIMS. We used a larger mass range for I-CIMS due to the difference in m/z of cluster ions formed with Cl₂⁻ and I⁻. Thus, we account for approximately the same range of analyte masses with both reagent ion chemistries.

Results and discussion

Analyte detection

The Cl₂-CIMS produced ions through a variety of mechanisms including proton abstraction; adduct formation with Cl⁻, Cl₂⁻, and O₂⁻; charge transfer; and secondary chemistry (Table 1 and Fig. S2†). Little fragmentation was observed. The dominant product ion varied by analyte. Cl₂⁻ cluster ions provided the largest sensitivity for formic acid and CH₃C(O)O₂, while product ions from secondary chemistry and O₂⁻ clustering provided the largest sensitivities for O₃ and C₂–C₄ organic acids, respectively. CO₃⁻ is a known product ion of O₃ for O₂-CIMS and is formed from the reaction of O₃⁻ with CO₂.³⁷ Though C₂–C₄ organic acids were detected as O₂⁻ clusters with sensitivities on the order of 1 to 2 ncps/ppt, Cl₂⁻ clusters produced much lower sensitivities at <0.02 ncps/ppt. C₂–C₄ organic acid clusters with Cl⁻ showed larger sensitivity than Cl₂⁻ at 1.2 ± 0.2 ncps/ppt for acetic acid, 0.215 ± 0.002 ncps/ppt for propanoic acid, and 0.142 ± 0.005 ncps/ppt for isobutyric acid.

Table 1 Analyte product ion detection and sensitivities

Analyte	Molecular formula	Product ions	Primary product ion	Primary ion sens. (ncps/ppt)^a	LOD (ppt)	Calibration range (ppb)
a Sensitivities normalized (per million ions) to sum of reagent ion signal (O2^− + Cl^− + Cl2^− + Cl3^−).b Sensitivity normalized (per million ions) to Cl2^− signal.c General formulas of observed product ions for all organic acids.
Acetyl peroxy radical	CH3C(O)O2	Cl2CH3C(O)O2^−	Cl2CH3C(O)O2^−	2.30 ± 0.04	5	0.48–0.96
				3.19 ± 0.05^b
Ozone	O3	O3^−, CO3^−	CO3^−	0.78 ± 0.02	291	18.4–147
Formic acid	HC(O)OH	ClRC(O)OH^−, Cl2RC(O)OH^−, O2RC(O)OH^−, RC(O)O^−^c	Cl2HC(O)OH^−	1.01 ± 0.02	73	33–65
Acetic acid	CH3C(O)OH		O2CH3C(O)OH^−	1.6 ± 0.2	50	23–46
Propanoic acid	C2H5C(O)OH		O2C2H5C(O)OH^−	2.00 ± 0.01	9	8.7–17.2
Isobutyric acid	C3H7C(O)OH		O2C3H7C(O)OH^−	1.11 ± 0.01	22	0.97–1.9

---

Sensitivities to all analytes were independent of IMR pressure within the range of 50-100 mbar (Fig. S3†) but did vary with IMR water partial pressure (0-0.6 mbar; Fig. S4 and S5†), and de-clustering voltage (Fig. S6†). The largest water partial pressure impacts were observed for Cl⁻ and Cl₂⁻ clusters. Thus, we operated the Cl₂-CIMS optimized for Cl₂(RC(O)O₂)^– (∼0.5 mbar H₂O) and overall cluster ions production (dV = 1.4).

Sensitivity to CH₃C(O)O₂, detected as Cl₂(CH₃C(O)O₂)^–, was 2.30 ± 0.04 ncps/ppt when normalized to the total reagent ion signal, and 3.19 ± 0.05 ncps/ppt when normalized to the Cl₂⁻ reagent ion signal. In contrast to other analytes, alternate ion formation pathways were negligible for the acetyl peroxy radical. The sensitivity of Cl₂-CIMS was higher than that for I-CIMS (1.54 ± 0.03 ncps/ppt for ICH₃C(O)O₂⁻) under the same measurement conditions. However, the LOD for I-CIMS was lower (2 ppt) than we obtained for Cl₂-CIMS (5 ppt) due to larger backgrounds for Cl₂-CIMS.

We note that CIMS users employ various normalization techniques; either normalizing data to total ion counts to account for slight variations in instrument performance, or normalizing data to the relevant reagent ion signals acknowledging that variations in just those signals can modulate the sensitivity to an analyte.^25,38,39 Different approaches to normalization impact the calculated sensitivity and detection limit. For example, we add a second Cl₂-CIMS sensitivity to CH₃C(O)O₂ (normalized to the Cl₂⁻ counts only) in Table 1 and Fig. S7,† due to a consistent relationship between Cl₂⁻ and Cl₂(CH₃C(O)O₂)^– observed during CH₃C(O)O₂ calibration and MEK photolysis (Fig. S8†). This normalization method increased the normalized sensitivity, but the total sensitivity observed during the test remains the same. The Cl₂-normalized sensitivity serves as both the maximum possible sensitivity in a reagent ion spectrum of entirely Cl₂⁻ and the most accurate method of calculating sensitivity to CH₃C(O)O₂ between tests. The CH₃C(O)O₂ sensitivity normalized to the sum of Cl⁻, Cl₂⁻, Cl₃⁻, and O₂⁻ did not reflect the difference of signal observed between Cl₂-and I-CIMS during MEK photolysis.

MEK photolysis product quantum yield measurements at 254 nm

The Cl₂-CIMS detected both CH₃C(O)O₂ and C₂H₅C(O)O₂ from MEK photolysis as clusters with Cl₂⁻. We assume that the Cl₂-CIMS sensitivity to CH₃C(O)O₂ and C₂H₅C(O)O₂ are equivalent based on voltage scanning data. Briefly, by changing the voltage difference between two components in the ion-focusing regions of the CIMS, we provide adequate energy to de-cluster reagent ion-analyte adducts.^35,41 The voltage difference that corresponds to a 50% loss in adduct ion signal (dV50) correlates to the sensitivity of the instrument to the clustering mechanism.⁴² The voltage scan provides a dV50 of 4.58 ± 0.05 V for Cl₂(CH₃C(O)O₂)^– and 4.21 ± 0.05 V for Cl₂(C₂H₅C(O)O₂)^– (Fig. S9†). This is quite close to the 0.3 V difference in dV50 found with I-CIMS,²³ indicating that the sensitivities of Cl₂-CIMS to CH₃C(O)O₂ and C₂H₅C(O)O₂ are similar. This assumption is further supported by the agreement of quantum yields between Cl₂-CIMS and I-CIMS and with prior literature.^19,40

Cl₂-CIMS derived MEK photolysis quantum yields at 254 nm (Φ^{MEK, RC(O)}₂₅₄) were consistent with previous measurements^19,23,40 and with I-CIMS Φ^{MEK, RC(O)}₂₅₄ obtained as part of this work. and ΣΦ^{MEK, RC(O)}₂₅₄ from Cl₂-CIMS and I-CIMS (Fig. 2) agreed within the measurement precision of 3%, while agreed within the experimental uncertainties of 26% and 24% calculated for Cl₂-CIMS and I-CIMS, respectively. The experimental uncertainties were 20% for and 21% for ΣΦ^{MEK, RC(O)}₂₅₄ for both Cl₂-CIMS and I-CIMS. Signals for Cl₂(CH₃C(O)O₂)^– and Cl₂(C₂H₅C(O)O₂)^– obtained by Cl₂-CIMS during MEK photolysis were larger than the corresponding I⁻ cluster signals from I-CIMS (Fig. S8†). Cl₂-CIMS measured radical signals normalized to only Cl₂⁻ were a factor of 2.1 ± 0.1 larger than I-CIMS (normalized to the sum of I⁻ and I(H₂O)^–), in agreement with the sensitivity ratio (2.07 ± 0.05) obtained during calibrations. This indicates that the Cl₂-CIMS sensitivity to RC(O)O₂ was dependent on Cl₂⁻ and not related to the total reagent ion count.

Fig. 2 Comparison of MEK photolysis product quantum yield values from the Cl₂-CIMS (dark red) and I-CIMS (light red) from this study to recent literature values. The x-axis consists of segments for each MEK photolysis product: CH₃C(O) on the bottom-left, C₂H₅C(O) on top-center, ΣRC(O) on the bottom-center, and branching ratios on the right (C₂H₅C(O)/∑RC(O) top; (CH₃C(O)/∑RC(O) bottom) orange diamonds represent values from Link et al.;²³ gold diamond represents the Rajakumar et al.⁴⁰ value for Φ_CH3C(O); light-brown diamonds represent branching ratios derived from C₂H₅C(O) : CH₃C(O) signal ratios in Zborowska et al.¹⁹ Error bars represent experimental uncertainties.

Cl₂-CIMS reagent ion chemistry

Cl₂-CIMS reagent ion chemistry is a complex system of ionizing reactions involving Cl⁻, Cl₂⁻, Cl₃⁻, and O₂⁻, which is best described by first examining the ion chemistry of Cl₂ in N₂. In a reagent gas mixture with dry N₂, collisions of α-particles from the Po-210 source with N₂ produce free electrons⁴³ (R3a). Cl₂ is the primary molecule that attaches free electrons within the ionizer flow (R3b).

α + N₂ → N₂⁺ + e⁻ + α (R3a)

e⁻ + Cl₂ → Cl₂⁻∗ (R3b)

Excited Cl₂⁻* formed by electron attachment can decompose to Cl⁻ and Cl (R4a),⁴⁴ though the example reagent ion spectrum in N₂ (Fig. 3(a)) shows a significant portion of Cl₂⁻* in our system is stabilized by collisions (R4b). Cl₂⁻* is the source of both Cl⁻ and Cl₂⁻ in the reagent ion spectra through either decomposition or collisional stabilization.

Cl₂⁻∗ → Cl + Cl⁻ (R4a)

Cl₂⁻∗ + M → Cl₂⁻ + M∗ (R4b)

Cl + Cl⁻ + M → Cl₂⁻ (R5)

Fig. 3 (a) Mass spectrum for dry N₂ with a reagent ion flow of Cl₂ in N₂; fractional contributions of reagent ions (dark red: Cl⁻, light red: Cl₂⁻, orange: Cl₃⁻, yellow: O₂⁻) to total reagent ion counts in the right-hand inset. (b) Mass spectrum for dry N₂ with reagent ion flow of Cl₂ in UZA and the respective fractional reagent ion contribution inset. (c) Ion counts for Cl⁻, Cl₂⁻ and Cl₃⁻ as a function of O₂ mixing ratio (%) in the reagent gas mixture (dry N₂ sampling). The black line represents the sum of these three ions.

The recombination reaction (R5) is a significant source of Cl₂⁻ for liquid phase studies but is not expected to be important here.⁴⁵ Instead, Cl⁻ can go on to react with Cl₂, forming the observed Cl₃⁻ signal (R6a), with Cl₂⁻ + Cl₂ (R6b) and Cl₂⁻ + Cl (R6c) reactions also being possible sources of Cl₃⁻.⁴⁶

Cl⁻ + Cl₂ → Cl₃⁻ (R6a)

Cl₂⁻ + Cl → Cl₃⁻ (R6b)

Cl₂⁻ + Cl₂ → Cl₃⁻ + Cl (R6c)

We found an upper limit of 400 ppb Cl₂ in the reagent ion flow (Fig. 4(c)), above which point the Cl₂⁻ signal stopped increasing and Cl₃⁻ became the largest reagent ion signal. We were unable to isolate production of Cl₂⁻ through modulation of Cl₂ concentration in N₂ alone.

Fig. 4 (a) Fraction of total reagent ion signal for Cl⁻ (dark red), Cl₂⁻ (light red), and Cl₃⁻ (orange), as a function of reagent gas Cl₂ mixing ratio (ppb) in N₂. (b) Fraction of total reagent ion counts for Cl⁻, Cl₂⁻, and Cl₃⁻, and O₂⁻ (yellow) as a function of reagent gas Cl₂ mixing ratio in UZA. (c) Reagent ion signal, relative to 300 ppb Cl₂, as a function of reagent gas Cl₂ mixing ratio in N₂. (d) Reagent ion signal, relative to 100 ppb Cl₂, as a function of reagent gas Cl₂ mixing ratio in UZA. All reagent ion data was collected while sampling dry N₂ to the CIMS inlet.

Impacts of O₂ and Cl₂ concentrations

The reagent ion distribution changed with the addition of O₂ (Fig. 3b and c). Adding ∼21% O₂ to the ionizer flow increased total ion counts by 58%, increased the fraction of Cl₂⁻ ions in the total reagent ion signal from 60 to 90%, reduced Cl⁻ and Cl₃⁻ signals, and increased O₂⁻. Due to the direct dependence of Cl₂(RC(O)O₂)^– on Cl₂⁻, the net result of these changes was a factor of 2.4 increase in total sensitivity to RC(O)O₂. Addition of O₂ also caused a 32% decrease of non-reagent ions in background mass spectra (Fig. 3a and b), leading to overall improvements in LODs.

The differences we observed in the reagent ion spectrum upon addition of O₂ result from a change in the primary Cl₂ ionization mechanism. At percent-level concentrations in the reagent gas mixture, O₂ becomes the primary electron capturing molecule (R7).

e⁻ + O₂ → O₂⁻ (R7)

Cl₂⁻ is a known product of Cl₂ ionization by O₂-CIMS,⁴⁷ and we hypothesize Cl₂⁻ is primarily formed here in a multiple step process. Electrons formed by ionization in the Po-210 source may predominately attach to O₂ molecules present in the reagent gas mixture at atmospheric mixing ratios. The electron affinities of Cl₂ and O₂ are 2.38 eV⁴⁸ and 0.451 eV,⁴⁹ respectively, indicating the following favorable electron transfer reaction as the primary source of Cl₂⁻ with O₂ in the reagent gas mixture (R8).

O₂⁻ + Cl₂ → O₂ + Cl₂⁻ (R8)

The reagent ion distribution impacts the chemical ionization mechanisms and thus sensitivity. We found that modulation of Cl₂ reagent gas concentration in UZA produced a negative correlation between Cl₂⁻ and O₂⁻ (Fig. 4(d)). Analyte ions for the organic acids, O₃, and CH₃C(O)O₂ responded to either Cl₂⁻ or O₂⁻, depending on which ion chemistry produced each ion. The sensitivity of these analyte ions to changes in the Cl₂ mixing ratio in the reagent gas mixture (in UZA) are shown in Fig. 5, with the slopes and r² of the linear least squares fits in Table S10.† Cl⁻ and Cl₂⁻ signals correlated well to both the Cl⁻ and Cl₂⁻ cluster analyte ions (r² > 0.73 for Cl⁻ and > 0.59 for Cl₂⁻). The exceptions were Cl₂(C₂H₅C(O)OH)^–, which had low sensitivity, and Cl₂(CH₃C(O)O₂)^– at Cl₂ concentrations above 400 ppb (Fig. S11†). For Cl₂ reagent gas concentrations below 400 ppb, Cl₂⁻ correlated to Cl₂(CH₃C(O)O₂)^– with slope = 0.53 and r² = 0.99. Fitting of O₂⁻ cluster and H⁺ abstraction ion signals to Cl⁻ and Cl₂⁻ resulted in negative slopes for all sampled acids. We also observed negative Cl⁻ and Cl₂⁻ correlations for O₃⁻ and CO₃⁻ from O₃ and for CH₃C(O)O⁻ and CH₃C(O)O₂⁻ detected during CH₃C(O)O₂ sampling. The O₂⁻ clusters and H⁺ abstraction products, instead, correlated strongly to O₂⁻ signal, as did O₃⁻, CO₃⁻, and CH₃C(O)O₂⁻ (r² > 0.96; slopes = 1.0). Changes in Cl₃⁻ signal did not correlate to any analyte ion, returning slopes between 0.09 and −0.15.

Fig. 5 (a) Correlation plot of Cl⁻ (diamonds) and Cl₂⁻ (hexagons) cluster ion signals for formic (dark red), acetic (light red), and propionic (orange) acids against background signals of Cl₂⁻ over the Cl₂ concentration changes. Linear correlation fits shown with dashed (Cl⁻ clusters) and solid (Cl₂⁻ clusters) lines. (b) Correlation plot of proton abstraction (circle) and O₂⁻ cluster (square) ion signals against background signals of O₂⁻ over the Cl₂ concentration changes. Acid signals are indicated by the same colors as (a), with gray markers for O₃ product ion correlations (squares for CO₃⁻ and circles for O₃⁻). Linear fits are shown as dashed lines for proton abstraction and solid lines for O₂⁻ cluster. The y-axis represents ratios of the observed signal to the signal at the lowest Cl₂ concentration; all error bars represent propagated uncertainty of one standard deviation in signal at each Cl₂ concentration step. Isobutyric acid is left off to reduce graph clutter, as these signals had large uncertainty for Cl₂⁻ reactions. Fit information for isobutyric acid is provided in Table S10 and Fig. S12.†

The formation of Cl⁻ clusters from Cl⁻ ionization is straightforward. However, Cl⁻ cluster formation from Cl₂⁻ is also possible, due to the instability Cl₂⁻. Here, the reaction between Cl₂⁻ and an analyte would act more like a ligand exchange, producing Cl-analyte ion and a Cl atom (R9).

Cl₂⁻ + RC(O)OH → Cl + (RC(O)OH)Cl⁻ (R9)

A similar ionization mechanism occurs with SF₆⁻ for some analytes.⁵⁰ Cl₃⁻ formation acts as a sink of sensitivity in the Cl₂-CIMS, as Cl₃⁻ signal did not show a positive correlation with any analyte ion. Cl₃⁻ is a cluster ion of Cl₂ and Cl⁻, which has a high binding energy.⁵¹ This binding energy was larger than that of the acid-Cl^- clusters (Fig. S6†). In addition, Cl₃⁻ has an electron affinity ≥ 3.8 eV,⁴⁶ making electron transfer ionization unfavorable for most analytes.

The results in Fig. 5 demonstrate that analyte ion production is directly dependent on the availability of reagent ions that produce each analyte ion, and not necessarily on total reagent ion count. This observation suggests that the proper normalization for analyte sensitivities is to the corresponding reagent ion, as we provide for CH₃C(O)O₂ in Table 1. Sensitivities for all organic acid and O₃ product ions are in Table S13.† Under the optimized conditions for Cl₂⁻ production in our system, the sensitivities normalized to Cl⁻ and Cl₂⁻ shift only slightly in Table S13† from those in Table 1, as Cl₂⁻ accounted for ≥ 80% of total reagent ion counts across these experiments and the sum of Cl⁻ and Cl₂⁻ accounted for ≥ 90%. Normalization to O₂⁻ increases the Cl₂-CIMS sensitivity to proton abstraction and O₂⁻ cluster products by at least an order of magnitude compared to Table 1. The O₂⁻ normalized sensitivities are near the collision limit, but sensitivities to O₂(CH₃C(O)OH)^– and CO₃⁻ are much larger than that limit and thus imply that additional ionization mechanisms or interferences are occurring.

O₂⁻ normalized sensitivities of 98 ± 12 ncps/ppt for O₂(CH₃C(O)OH)^– and 56 ± 4 ncps/ppt for CO₃⁻ suggest that the measured O₂⁻ signals may not be representative of the O₂⁻ concentrations in the IMR. These inflated sensitivities may be due to changes in transmission efficiency with mass, as CIMS of similar design have lower transmission efficiencies at low m/z.⁵² If the reaction of O₂⁻ with Cl₂ does not run to completion before sample introduction, analyte ions will appear at signals consistent with a larger O₂⁻ concentration than was measured, also explaining the surprisingly high O₂⁻ normalized sensitivities. Despite this problem, sensitivities in Table S13† may provide more consistent comparison of Cl₂-CIMS response across changing instrument and sample conditions, as was the case for our CH₃C(O)O₂ calibration and MEK photolysis measurements.

Woodsmoke

While Cl₂-CIMS is an effective detection technique for RO₂ in controlled laboratory systems, woodsmoke demonstrates the problems of using Cl₂-CIMS for complex mixtures. We use ion mass defects (defined as the difference between the exact m/z of an ion and the nearest unit mass) to visualize the differences in ions that appear in the Cl₂-CIMS and I-CIMS mass spectra during woodsmoke experiments. Multiple reagent ions and potential reaction pathways produced more analyte ions with Cl₂-CIMS (>300 ions; Fig. 6(a)). The I-CIMS spectra contained about half as many ions, which were mostly produced from iodide adduct formation (177 ions; Fig. 6(b)). Though more peaks appeared for Cl₂-CIMS, these were not indicative of compounds unique to Cl₂-CIMS detection. Instead, the larger peak count in Cl₂-CIMS is a byproduct of multiple reagent ion chemistries, where a single analyte formula would appear in up to four product ions. We fit 103 Cl₂⁻ cluster analyte ions, which represented 39.8% of analyte ion signals measured with Cl₂-CIMS. 82 Cl⁻ clusters accounted for 21.6% of analyte ion signals, while 38.6% of analyte signal was from 128 ions which did not contain Cl. In contrast, the 111 identified I⁻ containing ions accounted for 99.8% of analyte signals measured with I-CIMS.

Fig. 6 (a) Mass defect plot for identified ions from woodsmoke sampling with Cl₂-CIMS. Blue circles indicate O₂⁻ cluster and non-clustered product ions, orange indicates Cl-containing, and red indicates Cl₂-containing formulas. (b) Mass defect plot for detected ions during I-CIMS sampling of woodsmoke. Purple circles indicate peaks that did not contain iodide while green indicates I-containing formulas. Dot size represents the relative signal increase between background and smoke sampling steps.

The largest signals for both reagent ions were small organic acids and peracids. Specifically, C₂H₄O₂, CH₂O₂ and C₃H₆O₃ were all measured at signals that were about two times greater than any other product by I-CIMS. These species were also among the most prominent Cl₂-CIMS signals. However, the highest analyte signal for Cl₂-CIMS was O₂(CH₄O₂)^– (∼200 000 cps), corresponding to the oxygen cluster for acetic acid. This observation again indicates that the conditions in the IMR do not reflect the final spectra, and that the Cl₂-CIMS acts largely as an O₂-CIMS under polluted sampling conditions.

The appearance of Cl₂⁻, Cl⁻, and O₂⁻ clusters, and molecular ions (or H⁺ abstraction products) led to an array of possible mass defects. Overlap of ions at the same mass-to-charge occurred more frequently for Cl₂-CIMS. I-CIMS provided I⁻ clusters with large mass defects and a lack of competing ion-molecule reactions. Thus, ions detected by I-CIMS were more often separated from interfering peaks than those detected by Cl₂-CIMS. For example, nitrogen-containing products of biomass burning, including C₂H₅NO, CH₃NCO, C₃H₇NO,⁵³ could be identified with I-CIMS, but not clearly with Cl₂-CIMS. This limitation was largely due to the tendency for overlap of Cl₂⁻ (even mass) clusters, which might contain N, with Cl⁻ (odd mass) clusters, which did not (and vice versa). We saw little evidence for the identification of RO₂ in our data due to spectral complexity. It is also possible in a complex sample that side reactions of Cl₂-CIMS would produce ion formulas that are not representative of sampled molecular formulae, particularly if Cl radicals are formed from the decomposition of Cl₂⁻. We could derive no clear evidence of this from the woodsmoke experiments but provide evidence of possible Cl radical impacts in Fig. S11.† Due to the detrimental complexities of the current Cl₂-CIMS ion chemistry, we could draw little else from the analysis of these smoke samples without significant alterations to the CIMS IMR design.

Conclusion

Direct measurements of RO₂ in both ambient and laboratory settings require low limits of detection (1 ppt). Cl₂-CIMS is an effective reagent ion chemistry for laboratory measurement of CH₃C(O)O₂ and C₂H₅C(O)O₂ with better sensitivity than existing I-CIMS measurements – capable of producing quantum yields for MEK photolysis consistent with results obtained from other measurement techniques in the literature. However, the reported LOD of 5 ppt, higher than that found for I-CIMS, is too high for ambient measurement of RO₂. The sensitivity of Cl₂-CIMS to CH₃C(O)O₂ and C₂H₅C(O)O₂ would be increased through isolation of Cl₂⁻ ionization chemistry, without the non-selective ionization of O₂⁻ or the potential for secondary chemistry of Cl atoms in the IMR. Such alterations to Cl₂-CIMS would lower LODs and expand its applications.

Reactions of O₂⁻ may be avoided by using a different electron donor. For the electron transfer with Cl₂ to occur, the electron donor must be more reactive than Cl₂⁻. High purity air requires few considerations for use in CIMS reagent gas mixtures, compared to, for example, SF₆ (electron affinity = 1.05 eV).⁵⁰ In addition, O₂⁻ reacts with Cl₂ in our system to form Cl₂⁻ with few notable secondary products other than Cl⁻. This is not a guarantee for other electron donors. A more effective means of isolating Cl₂⁻ chemistry may be to alter the IMR design to add an initial reaction chamber where the O₂⁻ + Cl₂ reaction would run to completion prior to sample introduction, thus minimizing O₂⁻ chemistry. However, such modifications were outside the scope of this study. With the current instrumental design, the multiple reagent ion chemistries present in the Cl₂-CIMS system prevent the reagent ion chemistry from being useful for ambient studies, as evidenced by the woodsmoke measurements. Alternatively, this work highlights the utility of dual reagent ions (Cl₂⁻ and O₂⁻) to provide a selective measurement of acyl peroxy radicals.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

The authors declare no conflicts of interest regarding the content of this manuscript.

Acknowledgements

We thank Dr Kathryn Mayer for assisting with the woodsmoke sampling tests. We thank Dr Frank Flocke for providing the PAN source used for calibration tests. We thank the National Science Foundation (Grant 1922619) for funding.

References

1. C. A. Cantrell, R. E. Shetter and J. G. Calvert, Comparison of Perxoy Radical Concentrations at Several Contrasting Sites, J. Atmos. Sci., 1995, 52, 3408–3412.
2. Z. Tan, A. Hofzumahaus, K. Lu, S. S. Brown, F. Holland, L. G. Huey, A. Kiendler-Scharr, X. Li, X. Liu, N. Ma, K. E. Min, F. Rohrer, M. Shao, A. Wahner, Y. Wang, A. Wiedensohler, Y. Wu, Z. Wu, L. Zeng, Y. Zhang and H. Fuchs, No Evidence for a Significant Impact of Heterogeneous Chemistry on Radical Concentrations in the North China Plain in Summer 2014, Environ. Sci. Technol., 2020, 54, 5973–5979.
3. G. S. Tyndall, R. A. Cox, C. Granier, R. Lesclaux, G. K. Moortgat, M. J. Pilling, A. R. Ravishankara and T. J. Wallington, Atmospheric chemistry of small organic peroxy radicals, J. Geophys. Res.:Atmos., 2001, 106, 12157–12182.
4. A. Mellouki, T. J. Wallington and J. Chen, Atmospheric Chemistry of Oxygenated Volatile Organic Compounds: Impacts on Air Quality and Climate, Chem. Rev., 2015, 115, 3984–4014.
5. S. Y. Chen and Y. P. Lee, Transient infrared absorption of t-CH₃C(O)OO, c-CH₃C(O)OO, and α-lactone recorded in gaseous reactions of CH₃CO and O₂, J. Chem. Phys., 2010, 132, 114303.
6. M. George, M. Dolores Andrés Hernández, V. Nenakhov, Y. Liu and J. Philip Burrows, Airborne measurement of peroxy radicals using chemical amplification coupled with cavity ring-down spectroscopy: The PeRCEAS instrument, Atmos. Meas. Tech., 2020, 13, 2577–2600.
7. H. Fuchs, F. Holland and A. Hofzumahaus, Measurement of tropospheric RO₂ and HO₂ radicals by a laser-induced fluorescence instrument, Rev. Sci. Instrum., 2008, 79, 084104.
8. E. C. Wood, B. L. Deming and S. Kundu, Ethane-based chemical amplification measurement technique for atmospheric peroxy radicals, Environ. Sci. Technol. Lett., 2017, 4, 15–19.
9. S. Kundu, B. L. Deming, M. M. Lew, B. P. Bottorff, P. Rickly, P. S. Stevens, S. Dusanter, S. Sklaveniti, T. Leonardis, N. Locoge and E. C. Wood, Peroxy radical measurements by ethane-nitric oxide chemical amplification and laser-induced fluorescence during the IRRONIC field campaign in a forest in Indiana, Atmos. Chem. Phys., 2019, 19, 9563–9579.
10. J. J. Orlando and G. S. Tyndall, Laboratory studies of organic peroxy radical chemistry: An overview with emphasis on recent issues of atmospheric significance, Chem. Soc. Rev., 2012, 41, 6294–6317.
11. Z. Tan, H. Fuchs, K. Lu, A. Hofzumahaus, B. Bohn, S. Broch, H. Dong, S. Gomm, R. Häseler, L. He, F. Holland, X. Li, Y. Liu, S. Lu, F. Rohrer, M. Shao, B. Wang, M. Wang, Y. Wu, L. Zeng, Y. Zhang, A. Wahner and Y. Zhang, Radical chemistry at a rural site (Wangdu) in the North China Plain: Observation and model calculations of OH, HO₂ and RO₂ radicals, Atmos. Chem. Phys., 2017, 17, 663–690.
12. R. S. Hornbrook, J. H. Crawford, G. D. Edwards, O. Goyea, R. L. Mauldin, J. S. Olson and C. A. Cantrell, Measurements of tropospheric HO₂ and RO₂ by oxygen dilution modulation and chemical ionization mass spectrometry, Atmos. Meas. Tech., 2011, 4, 735–756.
13. M. Neumaier, R. Ruhnke, O. Kirner, H. Ziereis, G. Stratmann, C. A. M. Brenninkmeijer and A. Zahn, Impact of acetone (photo)oxidation on HOx production in the UT/LMS based on CARIBIC passenger aircraft observations and EMAC simulations, Geophys. Res. Lett., 2014, 41, 3289–3297.
14. S. A. McKeen, T. Gierczak, J. B. Burkholder, P. O. Wennberg, T. F. Hanisco, E. R. Keim, R.-S. Gao, S. C. Liu, A. R. Ravishankara and D. W. Fahey, The photochemistry of acetone in the upper troposphere: a source of odd-hydrogen radicals, Geophys. Res. Lett., 1997, 24, 3177–3180.
15. P. O. Wennberg, T. F. Hanisco, L. Jaeglé, D. J. Jacob, E. J. Hintsa, E. J. Lanzendorf, J. G. Anderson, R. S. Gao, E. R. Keim, S. G. Donnelly, L. A. Del Negro, D. W. Fahey, S. A. McKeen, R. J. Salawitch, C. R. Webster, R. D. May, R. L. Herman, M. H. Proffitt, J. J. Margitan, E. L. Atlas, S. M. Schauffler, F. Flocke, C. T. McElroy and T. P. Bui, Hydrogen radicals, nitrogen radicals, and the production of O₃ in the upper troposphere, Science, 1998, 279, 49–53.
16. M. A. Blitz, D. E. Heard, M. J. Pilling, S. R. Arnold and M. P. Chipperfield, Pressure and temperature-dependent quantum yields for the photodissociation of acetone between 279 and 327.5 nm, Geophys. Res. Lett., 2004, 31, 1–5.
17. T. Gierczak, J. B. Burkholder, S. Bauerle and A. R. Ravishankara, Photochemistry of acetone under tropospheric conditions, Chem. Phys., 1998, 231, 229–244.
18. M. Emrich and P. Warneck, Erratum: Photodissociation of acetone in air: dependence on pressure and wavelength, behavior of the excited singlet state, J. Phys. Chem. A, 2000, 104, 9436–9442.
19. A. G. Zborowska, C. Y. MacInnis, C. Z. Ye and H. D. Osthoff, On the photolysis branching ratio of methyl ethyl ketone, Atmos. Environ., 2021, 254, 118383.
20. P. R. Veres, J. M. Roberts, R. J. Wild, P. M. Edwards, S. S. Brown, T. S. Bates, P. K. Quinn, J. E. Johnson, R. J. Zamora and J. De Gouw, Peroxynitric acid (HO₂NO₂) measurements during the UBWOS 2013 and 2014 studies using iodide ion chemical ionization mass spectrometry, Atmos. Chem. Phys., 2015, 15, 8101–8114.
21. S. R. Albrecht, A. Novelli, A. Hofzumahaus, S. Kang, Y. Baker, T. Mentel, A. Wahner and H. Fuchs, Measurements of hydroperoxy radicals (HO₂) at atmospheric concentrations using bromide chemical ionisation mass spectrometry, Atmos. Meas. Tech., 2019, 12, 891–902.
22. J. Sanchez, D. J. Tanner, D. Chen, L. G. Huey and N. L. Ng, A new technique for the direct detection of HO₂ radicals using bromide chemical ionization mass spectrometry (Br-CIMS): initial characterization, Atmos. Meas. Tech., 2016, 9, 3851–3861.
23. M. F. Link, D. K. Farmer, T. Berg, F. Flocke and A. R. Ravishankara, Measuring Photodissociation Product Quantum Yields Using Chemical Ionization Mass Spectrometry: A Case Study with Ketones, J. Phys. Chem. A, 2021, 125, 6836–6844.
24. S. Iyer, X. He, N. Hyttinen, T. Kurtén and M. P. Rissanen, Computational and Experimental Investigation of the Detection of HO₂ Radical and the Products of Its Reaction with Cyclohexene Ozonolysis Derived RO2 Radicals by an Iodide-Based Chemical Ionization Mass Spectrometer, J. Phys. Chem. A, 2017, 121, 6778–6789.
25. M. A. Robinson, J. Andrew Neuman, L. G. Huey, J. M. Roberts, S. S. Brown and P. R. Veres, Temperature-dependent sensitivity of iodide chemical ionization mass spectrometers, Atmos. Meas. Tech., 2022, 15, 4295–4305.
26. D. L. Slusher, L. G. Huey, D. J. Tanner, F. M. Flocke and J. M. Roberts, A thermal dissociation - chemical ionization mass spectrometry (TD-CIMS) technique for the simultaneous measurement of peroxyacyl nitrates and dinitrogen pentoxide, J. Geophys. Res.: Atmos., 2004, 109, 1–13.
27. A. Hansel, W. Scholz, B. Mentler, L. Fischer and T. Berndt, Detection of RO₂ radicals and other products from cyclohexene ozonolysis with NH₄⁺ and acetate chemical ionization mass spectrometry, Atmos. Environ., 2018, 186, 248–255.
28. T. Berndt, B. Mentler, W. Scholz, L. Fischer, H. Herrmann, M. Kulmala and A. Hansel, Accretion Product Formation from Ozonolysis and OH Radical Reaction of α-Pinene: Mechanistic Insight and the Influence of Isoprene and Ethylene, Environ. Sci. Technol., 2018, 52, 11069–11077.
29. A. Zaytsev, M. Breitenlechner, A. Novelli, H. Fuchs, D. A. Knopf, J. H. Kroll and F. N. Keutsch, Application of chemical derivatization techniques combined with chemical ionization mass spectrometry to detect stabilized Criegee intermediates and peroxy radicals in the gas phase, Atmos. Meas. Tech., 2021, 14, 2501–2513.
30. G. D. Edwards, C. A. Cantrell, S. Stephens, B. Hill, O. Goyea, R. E. Shetter, R. L. Mauldin, E. Kosciuch, D. J. Tanner and F. L. Eisele, Chemical Ionization Mass Spectrometer Instrument for the Measurement of Tropospheric HO₂ and RO₂, Anal. Chem., 2003, 75, 5317–5327.
31. W. Zhang and H. Zhang, Secondary Ion Chemistry Mediated by Ozone and Acidic Organic Molecules in Iodide-Adduct Chemical Ionization Mass Spectrometry, Anal. Chem., 2021, 8595–8602.
32. T. H. Bertram, J. R. Kimmel, T. A. Crisp, O. S. Ryder, R. L. N. Yatavelli, J. A. Thornton, M. J. Cubison, M. Gonin, D. R. Worsnop and A. field-deployable, chemical ionization time-of-flight mass spectrometer, Atmos. Meas. Tech., 2011, 4, 1471–1479.
33. P. Brophy and D. K. Farmer, A switchable reagent ion high resolution time-of-flight chemical ionization mass spectrometer for real-time measurement of gas phase oxidized species: Characterization from the 2013 southern oxidant and aerosol study, Atmos. Meas. Tech., 2015, 8, 2945–2959.
34. F. M. Flocke, A. J. Weinheimer, A. L. Swanson, J. M. Roberts, R. Schmitt and S. Shertz, On the measurement of PANs by gas chromatography and electron capture detection, J. Atmos. Chem., 2005, 52, 19–43.
35. P. Brophy and D. K. Farmer, Clustering, methodology, and mechanistic insights into acetate chemical ionization using high-resolution time-of-flight mass spectrometry, Atmos. Meas. Tech., 2016, 9, 3969–3986.
36. J. Li, M. F. Link, S. Pandit, M. H. Webb, K. J. Mayer, L. A. Garofalo, K. L. Rediger, D. G. Poppendieck, S. M. Zimmerman, M. E. Vance, V. H. Grassian, G. C. Morrison, B. J. Turpin and D. K. Farmer, The persistence of smoke VOCs indoors: Partitioning, surface cleaning, and air cleaning in a smoke-contaminated house, Sci. Adv., 2023, 9, eadh8263.
37. I. Dotan, J. A. Davidson, G. E. Streit, D. L. Albritton and F. C. Fehsenfeld, A study of the reaction O₃⁻+CO₂ ⇌ CO₃⁻ + O₂ and its implication on the thermochemistry of CO₃ and O₃ and their negative ions, J. Chem. Phys., 1977, 67, 2874–2879.
38. M. Wang, X.-C. He, H. Finkenzeller, S. Iyer, D. Chen, J. Shen, M. Simon, V. Hofbauer, J. Kirkby, J. Curtius, N. Maier, T. Kurtén, D. Worsnop, M. Kulmala, M. Rissanen, R. Volkamer, Y. J. Tham, N. Donahue and M. Sipilä, Measurement of iodine species and sulfuric acid using bromide chemical ionization mass spectrometers, Atmos. Meas. Tech. Discuss, 2020, 1–23.
39. L. Xu, M. M. Coggon, C. E. Stockwell, J. B. Gilman, M. A. Robinson, M. Breitenlechner, A. Lamplugh, J. D. Crounse, P. O. Wennberg, J. A. Neuman, G. A. Novak, P. R. Veres, S. S. Brown and C. Warneke, Chemical ionization mass spectrometry utilizing ammonium ions (NH₄⁺ CIMS) for measurements of organic compounds in the atmosphere, Atmos. Meas. Tech., 2022, 15, 7353–7373.
40. B. Rajakumar, T. Gierczak, J. E. Flad, A. R. Ravishankara and J. B. Burkholder, The CH₃CO quantum yield in the 248 nm photolysis of acetone, methyl ethyl ketone, and biacetyl, J. Photochem. Photobiol., A, 2008, 199, 336–344.
41. J. M. Mattila, P. S. J. Lakey, M. Shiraiwa, C. Wang, J. P. D. Abbatt, C. Arata, A. H. Goldstein, L. Ampollini, E. F. Katz, P. F. Decarlo, S. Zhou, T. F. Kahan, F. J. Cardoso-Saldaña, L. H. Ruiz, A. Abeleira, E. K. Boedicker, M. E. Vance and D. K. Farmer, Multiphase Chemistry Controls Inorganic Chlorinated and Nitrogenated Compounds in Indoor Air during Bleach Cleaning, Environ. Sci. Technol., 2020, 54, 1730–1739.
42. F. D. Lopez-Hilfiker, S. Iyer, C. Mohr, B. H. Lee, E. L. D’ambro, T. Kurtén and J. A. Thornton, Constraining the sensitivity of iodide adduct chemical ionization mass spectrometry to multifunctional organic molecules using the collision limit and thermodynamic stability of iodide ion adducts, Atmos. Meas. Tech., 2016, 9, 1505–1512.
43. J. D. Crounse, K. A. McKinney, A. J. Kwan and P. O. Wennberg, Measurement of gas-phase hydroperoxides by chemical ionization mass spectrometry, Anal. Chem., 2006, 78, 6726–6732.
44. L. G. Christophorou and J. K. Olthoff, Electron Interactions With Cl₂, J. Phys. Chem. Ref. Data, 1999, 28, 131–169.
45. J. C. Han, M. Suto, J. C. Lee and Z. L. Petrović, Transient signals induced by laser irradiation of negative ions in hollow electrode discharges of Cl₂ and HCl in N₂, J. Appl. Phys., 1990, 68, 2649–2656.
46. L. M. Babcock and G. E. Streit, Ion–molecule reactions of Cl₂ with Cl⁻ and F⁻, J. Chem. Phys., 1982, 76, 2407–2411.
47. B. D. Finley and E. S. Saltzman, Observations of Cl₂, Br₂, and I₂ in coastal marine air, J. Geophys. Res.:Atmos., 2008, 113, D21301.
48. W. A. Chupka, J. Berkowitz and D. Gutman, Electron affinities of halogen diatomic molecules as determined by endoergic charge transfer, J. Chem. Phys., 1971, 55, 2724–2733.
49. M. J. Travers, D. C. Cowles and G. B. Ellison, Reinvestigation of the Electron Affinities of O₂ and NO, Chem. Phys. Lett., 1989, 164, 450–455.
50. L. G. Huey, D. R. Hanson and C. J. Howard, Reactions of SF₆⁻ and I⁻ with atmospheric trace gases, J. Phys. Chem., 1995, 99, 5001–5008.
51. C. Amelynck, E. Arijs, N. Schoon and A.-M. Van Bavel, Gas phase reactions of HNO₃ with Cl⁻, Cl⁻H₂O, and Cl⁻HCl, of Cl₂ with Cl⁻H₂O and Cl⁻HCl, and of HCl with Cl⁻H₂O, Int. J. Mass Spectrom., 1998, 181, 113–121.
52. M. Heinritzi, M. Simon, G. Steiner, A. C. Wagner, A. Kürten, A. Hansel and J. Curtius, Characterization of the mass-dependent transmission efficiency of a CIMS, Atmos. Meas. Tech., 2016, 9, 1449–1460.
53. M. Priestley, M. Le Breton, T. J. Bannan, K. E. Leather, A. Bacak, E. Reyes-Villegas, F. De Vocht, B. M. A. Shallcross, T. Brazier, M. Anwar Khan, J. Allan, D. E. Shallcross, H. Coe and C. J. Percival, Observations of Isocyanate, Amide, Nitrate, and Nitro Compounds From an Anthropogenic Biomass Burning Event Using a ToF-CIMS, J. Geophys. Res.:Atmos., 2018, 123, 7687–7704.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ea00043b

‡ Now at Engineering Laboratory, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, USA.

---