Ambient organic peroxy (RO₂) radical is almost the only highly reactive intermediate produced during the oxidation processes of hydrocarbons [1, 2]. As a prominent media, RO₂ radical is closely related to the removal of primary pollutants and the generation of secondary pollutants. Together with hydroperoxy (HO₂) and hydroxyl (OH) radicals, RO₂ radicals determine the atmospheric oxidation capacity, and therefore significantly influence the formation of secondary pollutants, including tropospheric ozone (O₃), organic nitrates, organic acids, and particulate matters [3-7].

The atmospheric abundance and distribution of RO₂ radicals are crucial for a better understanding of tropospheric chemical oxidation [8]. The quantitative detection of RO₂ is not only a critical data foundation but a useful parameter in the chemical model as well [8, 9]. However, since RO₂ is highly reactive with trace gases and surfaces, similar to the other free radicals, a sensitive and precise detection system is required to measure ambient RO₂, which is technically challenging to achieve.

The first field measurement approach of RO₂ radicals was put forward in 1978 [10], known as the matric isolation and electron spin resonance (MIESR). It is a direct measurement technique and can distinguish different RO₂ species. However, complex operation and harsh sample storage conditions limit the application of this technique [10-12]. The peroxy radical chemical amplifier (PERCA), first proposed in 1984 [13], is one of the most widely used methods due to its high operability, but it can hardly separate different RO₂ species [13-17]. The same problem happens to the peroxy radical chemical ionization mass spectrometry (PerCIMS (or RO_xMAS) technique, although it has higher accuracy and shorter chain length requirement [9]. The laser-induced fluorescence (LIF) is a classic technique of measuring OH and HO₂ [18, 19]. Some progress has been made in measuring RO₂ radicals by combining LIF with a chemical flow tube reactor [20], but only a few field observations of RO₂ radicals have been reported before [21, 22].

In this work, a flow tube reactor was developed based on the previous works [20, 23] and was applied together with the Peking University LIF instrument, implementing the measurement of ambient RO₂ radicals. By utilizing the active chemical priorities of some types of RO₂ (long-chain alkane-, alkene- and aromatic hydrocarbon-derived) radicals (known as RO₂^#) [21], speciated detection of these types of RO₂ was achieved.

The measurement of ambient RO₂ by LIF technique involves two steps of chemical conversion: a) RO_x (= RO₂ + HO₂ + RO + OH) radicals are converted into HO₂ radicals inside the flow tube reactor under a low-pressure condition; b) HO₂ radicals obtained at the exit of the flow tube are further converted into OH radicals in the fluorescence cell, whereOH can be detected directly, under an even lower pressure condition. These processes are illustrated in Fig. 1.

Inside the flow tube reactor, nitric oxide (NO) and carbon monoxide (CO) are introduced as reagent gases. With excessive NO, RO₂ can quickly convert to organic oxy radicals (RO), followed by the formation of HO₂ and carbonyl compounds (R'O) with the presence of oxygen (O₂) in sampled air (Reactions 1 and 2). Generated HO₂, as well as the initially existed HO₂ in sampled air, can further react with NO, resulting in the generation of OH radicals (Reaction 3):

(1)

(2)

(3)

As the chemical reactivity of OH is much higher than that of HO₂, the prolonged residence of OH radical in the flow tube may cause significant radical loss. Excessive CO is simultaneously injected into the reactor, thereby the chemical equilibrium between OH and HO₂ can be shifted to the HO₂ side (Reactions 4 and 5), minimizing the radical loss in the flow tube:

(4)

(5)

Nevertheless, the conversion efficiency of RO₂ to HO₂ is still inevitably limited due to several radical loss processes. Firstly, all the RO_x radicals can undergo termolecular recombination reactions with NO (Reactions 6-9):

(6)

(7)

(8)

(9)

Secondly, reactive collisions of RO_x radicals with the inner surface of the reactor are also major loss processes need to be considered (Reactions 10–13):

(10)

(11)

(12)

(13)

Reactions with NO (Reactions 6–9) are pressure-dependent [20], so they can be suppressed effectively by lowering the pressure in the reactor. Based on the rate constants recommended by previous studies (Table S1 in Supporting information), recombination reactions become insignificant under the low-pressure condition. But wall reactions are still dominant with a reduction of total HO₂ yield of ~20% [20], even after coating the inner surface of the reactor with the Teflon layer.

The central part of the airflow is collected into the fluorescence cell at the exit of the reactor. Additional NO is injected into this unit to shift the chemical equilibrium between HO₂ and OH back to the OH side, which can be detected by the LIF directly. A detailed description of the detection principle can be found in previous papers [18, 19]. In brief, OH radicals can be excited by narrow-bandwidth UV laser (λ =308 nm) on a single rovibronic transition, and therefore fluoresce. Gated photon counting is used to produce a time delay to discriminate the emitted fluorescence by OH and laser stray light. The background signal is determined by turning the laser resonance on and off periodically. The conversion between fluorescence signals and radical concentrations requires a calibration.

When NO and CO are added as reagent gases into the reactor, all the RO_x radicals are converted and detected in the fluorescence cell, so this measurement mode is called "RO_x mode". When only CO is added, RO₂ and RO cannot be converted into HO₂ theoretically, so this mode was called "HO_x mode" in previous studies [20]. However, it has been found afterward that for longer chain (C > 3) alkane-, alkene- and aromatic hydrocarbon-derived RO₂ radicals (RO₂^#), they can rapidly convert to HO₂ and then OH in the fluorescence cell when the concentration of NO reaches a certain level in the fluorescence cell [23, 24]. This characteristic may elevate the signal of HO₂ radicals as an "RO₂ interference", yet it is also a potentiality to measure different RO₂ species by rising NO concentration in the fluorescence cell. In consequence, when only CO is introduced, RO₂^#, RO, OH, and HO₂ are measured (named as "RO₂^# mode"). Because the concentration of RO is negligible compared to that of RO₂ and HO₂, so RO₂ and RO₂^# radical concentrations can then be calculated by simultaneously measured OH and HO₂ data.

Based on the measurement principle mentioned above, the flow tube reactor was designed, as can be seen in Fig. 2. The ambient air was sampled into the reactor through a 1 mm orifice (nickel, 70°, Beam Dynamics, Inc.) with a flow rate of 7 slm (standard liters per minute at 1 atm, 20 ℃). Reagent gases, NO and CO, were immediately injected into the sampled air through a glass tube (λ = 100 mm, i.d. = 4 mm) after the inlet nozzle, attaining ultimate mixing ratios of 0.7 ppmv and 0.17% respectively. These concentrations were defined based on the simulation results of MCM 3.2 ([20]). The flow tube (λ = 830 mm, i.d. = 66 mm) is made of quartz, and its inner surface is coated with the Teflon layer. The reactor was operated at ~25 hPa. The pressure was controlled by a butterfly valve and monitored by a vacuum gauge (Edwards, APG100). Excess air in the reactor was removed by an oil-free vacuum pump (Edwards, XDS35i).

At the end of the flow tube, another nozzle (4 mm orifice, stainless steel, 70°) was used to sample the central part of the reacted gas to the fluorescence cell, in which the pressure was kept at ~4 hPa. Pure NO (Linde AG, 99.5%) was used to convert HO₂ to OH, so that both RO₂ and RO₂^# can be detected in this work.

All the reagent gases were controlled by mass flow controllers (HORIBA METRON, S48 300/HMT). Electromagnetic valves and time relays were used to achieve periodically (5 min) switch between the two measurement modes. Thus, continuous online measurement with alternating modes has a time resolution of 10 min for both RO₂ and RO₂^#.

Based on previous studies, equal concentrations of OH and HO₂ can be generated by the photolysis (λ =185 nm) of H₂O at 1 atm [19, 25]. It has also been proven that equal amounts of HO₂ and RO₂ can be produced when the suitable hydrocarbon is mixed with humid air [8, 20, 26], as long as the introduced hydrocarbon is stabilized under 185 nm ultraviolet (UV). Hence, it is feasible to calibrate the system by using a radical source designed for OH and HO₂ calibration (for a detailed description of the calibration source, see reference [20]).

The calibration method depends on the approaches to obtaining OH and HO₂ data. If all the RO₂, HO₂, and OH are measured simultaneously by one LIF instrument, the detection sensitivities can be determined in three steps by altering reagent gases in the calibration source: a) Humid air with CO; b) humid air with hydrocarbon; c) only humid air. The fluorescence photon count rate (counts/s) of each radical species can then be calculated directly. The calculation method can be found in the previous paper [20].

However, if the OH and HO₂ data are obtained from another instrument, which is the case of this work, the photon count rate cannot be calculated directly. Concentrations of RO_x, HO₂, and OH need to be calculated before doing the subtraction. In this case, three parameters should be considered: a) Conversion efficiency of RO₂ to HO₂ in the reactor (r₁), b) conversion efficiency of HO₂ to OH in the fluorescence cell (r₂), c) the sensitivity of HO₂ radicals (S_HO2/ [HO₂]₀).

Since pure NO is used in the fluorescence cell, 100% HO₂ is assumed to be converted into OH (r₂ = 1). The conversion efficiency of RO₂ to HO₂ is referenced by CH₃O₂, which was simulated by the Master Chemical Mechanism (MCM v3.3). The sensitivity of RO_x radicals can be calculated via Eq. 1:

(1)

An ambient air measurement was performed in summer 2019, at a wetland park in Chengdu, China (30.4259°N, 103.847126°E). The site is in the suburbs, surrounded by plenty of vegetation. However, because of the nearby China Civil Aviation Flight Academy, many helicopters were flying above the site during the campaign, which may lead to primary emissions of pollutants. The instrument was placed ~10 m above ground, adjacent to the other LIF (used to measure HO₂ and OH).

Fig. 3 shows the averaged diurnal profiles of RO₂ and RO₂^# radicals from 26^th August to 12^th September. The variations of RO₂ and RO₂^# were highly correlated with HO₂ radicals. These photochemically formed radicals peaked at around 14:00 with concentrations of 1.55 × 10⁸ (RO₂), 1.04 × 10⁸ (RO₂^#), and 1.50 × 10⁸ (HO₂) molecule/cm3, when the solar actinic flux reached the highest value. In the meanwhile, NO concentration reached the lowest values at noon and peaked in the early morning with a concentration of ~8 ppbv. An apparent negative correlation between peroxy radicals and NO can be observed. A small peak of RO₂ and RO₂^# radicals was found at around 20:00, indicating the existence of radical source in the evening. A strong positive correlation between NO₃ and RO₂ radicals has been found in a previous study [27], implying the rising NO₃ radicals may be the possible reason for this small peak at night. The concentration level and diurnal pattern of RO₂ are similar to what has been reported in previous studies [20-23], besides the RO₂^#/RO₂ (~60% in this case) was a little higher than before. The divergence between RO₂^# and RO₂ can hardly be observed in the early morning when NO concentration was increasing rapidly. A possible reason is that the ambient NO promoted the conversion of RO₂ in the converter or even in the atmospheric, invalidating the separation of RO₂^# radicals. Volatile organic compounds (VOCs) were also measured in this campaign (Table S2 and Fig. S1 in Supporting information). Ethene was the most abundant VOC with a proportion of 4.49% that can generate RO₂^#, while ethyne was the most abundant among all the VOCs. The ratio of RO₂^#-produced VOCs to total VOCs was about 35.5%.

The photochemical O₃ production rate (P(O₃)) can be calculated by the oxidation rate of NO to NO₂ through its reaction with peroxy radicals (HO₂ and RO₂, Reactions 1–3) [28-30]. Without considering the loss processes of NO₂ (e.g., the reaction of NO₂ and OH or the formation of organic nitrates by RO₂), the upper limit for P(O₃) can be estimated by Eq. 2 [31],

(2)

where k_RO2+NO and k_HO2+NO denote the rate constants of the reaction of RO₂ + NO and HO₂ + NO. Fig. 4 shows the dependence of P(O₃) on NO. The P(O₃) increased with the rising of NO concentration, except for a small decline of around 1 ppbv of NO. The correlation between NO and P(O₃) in pollution days has also been reported in Fig. 4 with a relatively strong positive correlation (R² = 0.78), indicating that the suburb Chengdu is NOx sensitive for the ozone production.

In conclusion, a low-pressure flow tube reactor coupling with the LIF instrument was developed and successfully applied in the field measurement of RO₂ and RO₂^#. RO₂ radicals were firstly converted to HO₂ radicals in the reactor and then to OH in the fluorescence cell. Longer chain (C > 3) alkane, alkene and aromatic hydrocarbon-derived RO₂ radicals (RO₂^#) can be determined separately due to their selective chemical properties. The results of field measurements have proven that the designed reactor is an effective and convenient extension for the LIF technique. High NO concentration may affect the speciated measurement results, which limits the application of this technique in urban areas. Further studies on improving the performance of speciated measurements are still needed.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The work was supported by the National Key R&D Program of China (No. 2017YFC0209402) and the Beijing Natural Science Foundation, China (No. JQ19031). The authors gratefully acknowledge the support and assistance from Prof. Limin Zeng and Prof. Xin Li in the field campaign of Chengdu.

Appendix A. Supplementary data

Supplementary material related to this article can befound, in the online version, at doi:https://doi.org/10.1016/j.cclet.2020.07.051.