Technical Note: Effect of varying the λ = 185 and 254 nm photon flux ratio on radical generation in oxidation flow reactors

Abstract. Oxidation flow reactors (OFRs) complement environmental smog chambers
as a portable, low-cost technique for exposing atmospheric compounds
to oxidants such as ozone (O3), nitrate (NO3)
radicals, and hydroxyl (OH) radicals. OH is most commonly generated
in OFRs via photolysis of externally added O3 at
λ=254 nm (OFR254) or combined photolysis of
O2 and H2O at λ=185 nm plus
photolysis of O3 at λ=254 nm (OFR185) using
low-pressure mercury (Hg) lamps. Whereas OFR254 radical generation is
influenced by [O3], [H2O], and photon flux at
λ=254 nm (I254), OFR185 radical generation
is influenced by [O2], [H2O], I185, and
I254. Because the ratio of photon fluxes,
I185:I254, is OFR-specific, OFR185 performance varies
between different systems even when constant [H2O] and
I254 are maintained. Thus, calibrations and models developed
for one OFR185 system may not be applicable to another. To investigate
these issues, we conducted a series of experiments in which
I185:I254 emitted by Hg lamps installed in an OFR was
systematically varied by fusing multiple segments of lamp quartz
together that either transmitted or blocked λ=185 nm
radiation. Integrated OH exposure (OHexp) values
achieved for each lamp type were obtained using the tracer decay
method as a function of UV intensity, humidity, residence time, and
external OH reactivity (OHRext). Following previous
related studies, a photochemical box model was used to develop a
generalized OHexp estimation equation as a function of
[H2O], [O3], and OHRext that is
applicable for I185:I254≈0.001 to 0.1.


Abstract. Oxidation flow reactors (OFRs) complement environmental smog chambers as a portable, low-cost technique for exposing atmospheric compounds to oxidants such as ozone (O 3 ), nitrate (NO 3 ) radicals, and hydroxyl (OH) radicals. OH is most commonly generated in OFRs via photolysis of externally added O 3 at λ = 254 nm (OFR254) or combined photolysis of O 2 and H 2 O at λ = 185 nm plus photolysis of O 3 at λ = 254 nm (OFR185) using low-pressure mercury (Hg) lamps. Whereas OFR254 radical generation is influenced by [O 3 ], [H 2 O], and photon flux at λ = 254 nm (I 254 ), OFR185 radical generation is influenced by [O 2 ], [H 2 O], I 185 , and I 254 . Because the ratio of photon fluxes, I 185 : I 254 , is OFR-specific, OFR185 performance varies between different systems even when constant [H 2 O] and I 254 are maintained. Thus, calibrations and models developed for one OFR185 system may not be applicable to another. To investigate these issues, we conducted a series of experiments in which I 185 : I 254 emitted by Hg lamps installed in an OFR was systematically varied by fusing multiple segments of lamp quartz together that either transmitted or blocked λ = 185 nm radiation. Integrated OH exposure (OH exp ) values achieved for each lamp type were obtained using the tracer decay method as a function of UV intensity, humidity, residence time, and external OH reactivity (OHR ext ). Following previous related studies, a photochemical box model was used to develop a generalized OH exp estimation equation as a function of [H 2 O], [O 3 ], and OHR ext that is applicable for I 185 : I 254 ≈ 0.001 to 0.1.

Introduction
Hydroxyl (OH) radicals govern the concentrations of most atmospheric organic compounds, including those that lead to secondary organic aerosol (SOA) formation. For decades, environmental chambers and oxidation flow reactors (OFRs) have been used to simulate atmospheric aging processes through the controlled exposure of trace gases and aerosols to OH radicals. Environmental chamber studies are typically conducted over experimental timescales and equivalent atmospheric exposure times of hours up to 1 or 2 d. OFRs with residence times on the order of minutes achieve multiple days of equivalent atmospheric OH exposure (OH exp ), typically through the following reactions: This method is referred to as OFR254 and relies on addition of externally generated O 3 at the OFR inlet. In some cases, OFRs have additionally employed the secondary λ = 185 nm emission line present in low-pressure mercury (Hg) lamps to generate radicals from the following reactions in addition to those listed above that are employed in OFR254: This method is referred to as OFR185. Recent modeling studies suggest that OFR185 is less affected by experimental artifacts than OFR254 such as SOA photolysis and unwanted reactions with non-OH oxidants (Peng et al., 2016(Peng et al., , 2018

Experimental
Experiments were conducted using an Aerodyne Potential Aerosol Mass (PAM) OFR, which is a horizontal aluminum cylindrical chamber (46 cm long × 22 cm ID) operated in continuous flow mode (Lambe et al., 2011). A simplified schematic is shown in Fig. S1  ) that were isolated from the sample flow using type 214 quartz sleeves. Nitrogen purge gas was flowed over the lamps to prevent O 3 buildup between the lamps and sleeves. A fluorescent dimming ballast was used to regulate current applied to the lamps. The dimming voltage applied to the ballast ranged from 0.8 to 10 V direct current (DC). Below ∼ 0.8 V DC, the lamp output was unstable due to flickering, and 10 V DC was the maximum control voltage permitted by the ballast. Figure 1 shows the Hg fluorescent lamp configurations that were used in this study. Lamp type A is an ozoneproducing low-pressure Hg germicidal fluorescent lamp (GPH436T5VH/4P, Light Sources Inc.) in which type 214 quartz that transmits λ = 185 and 254 nm radiation is present along the entire 356 mm arc length. This lamp type is a standard component of the Aerodyne PAM OFR. The relative transmissivity of λ = 185 nm radiation (T 185 ) in lamp type A is thus equal to 1. Lamp type B is equivalent to lamp type A with added segments of opaque heat shrink tubing applied to approximately 86 % of the arc length (T 185 ≈ 0.14; see also Fig. S2 in the Supplement) to reduce I 185 and I 254 to levels below what is achievable using the ballast dimming voltage. A different type of quartz is available (type 219) that blocks λ = 185 nm and transmits λ = 254 nm radiation (T 185 = 0). To cover the largest possible range of I 185 : I 254 , lamp types C, D, and E (GPH436T5L/VH/4P 90/10, GPH436T5L/VH/4P 96/4, and GPH436T5L/VH/4P 98.5/1.5; Light Sources, Inc.) fused one segment each of quartz with T 185 = 0 and T 185 = 1 to provide reduced I 185 relative to lamp type A while maintaining constant I 254 . Finally, to evaluate the effect of lamp design at fixed T 185 and I 254 , lamp types F and G contain the same ratios of T 185 = 0 and T 185 = 1 quartz as types C and D, but with 5 and 13 total segments instead of 2 segments. These different designs isolate the effect of discretized λ = 185 nm irradiation across the entire arc length of the lamp vs. having all λ = 185 nm radiation near the entrance of the OFR.

OH exp characterization studies
OH exp , defined here as the product of the average OH concentration and the mean OFR residence time (τ OFR ), was characterized by measuring the decay of carbon monoxide (CO) and/or sulfur dioxide (SO 2 ) tracers using Thermo 48i and 43i CO and SO 2 analyzers (e.g., Lambe et al., 2011). Tracer mixing ratios entering the reactor were 6-9 ppmv for CO and 288-629 ppbv for SO 2 , each diluted from separate gas mixtures of 0.5 % CO or SO 2 in N 2 (Praxair). The corresponding total external OH reactivity (OHR ext ), which is the summed product of each tracer mixing ratio and its bimolecular OH rate coefficient, ranged from approximately 9 to 64 s −1 . Tracer concentrations were allowed to stabilize before initiating OH exp measurements, during which steadystate levels of CO and/or SO 2 were obtained with the lamps turned off. Then, the lamps were turned on, and tracer concentrations were allowed to equilibrate before being measured at illuminated steady-state conditions. In most experiments, the calculated mean residence time was τ OFR = 124 s, which was obtained from the ratio of the internal OFR volume (≈ 13 L) and the total sample and makeup flow rate through the OFR (6.4 L min −1 ). This calculation implicitly assumes plug flow conditions, with associated uncertainty of approximately 10 % compared to an explicit residence time distribution measurement at a specific OFR condition . Variability in OFR parameters (e.g., temperature, flow rate) may increase the uncertainty in this assumption across a continuum of conditions (Huang et al., 2017;Lambe et al., 2019). To characterize the uncertainty in our plug flow approximation across multiple sample flow conditions, we measured integrated OH exposure (OH exp ) values of 3.3 × 10 11 , 7.8 × 10 11 , and 2.0 × 10 12 molec. cm −3 s at sample flow rates of 12.5, 6.4, and 3.1 L min −1 , respectively, using the tracer decay method (Sect. 2.2) with the OFR operated at the same humidity and lamp intensity. Thus, perturbing the "plug flow" τ OFR = 124 s by a factor of 2 in either direction changed OH exp by factors of 2.36 and 2.56. Based on these results, an upperlimit estimated uncertainty in τ OFR and corresponding OH exp is approximately 30 %.

Photochemical model
We used a photochemical model implemented in MATLAB and Igor Pro to calculate concentrations of radical/oxidant species produced in the reactor . The Kin-Sim chemical kinetic solver was used to compile the version of the model that was implemented in Igor Pro . Model input parameters are shown in Table 1, and reactions and associated kinetic rate coefficients 0.4-156 CO (ppmv) 0 or 6-9 SO 2 (ppbv) 0 or 288-629 I 185 (photons cm −2 s −1 ) 1.1 × 10 12 -3.2 × 10 14 I 254 (photons cm −2 s −1 ) 6.0 × 10 13 -4.2 × 10 15 that were included in the model are summarized in Table S1 in the Supplement (Peng and Jimenez, 2020a). For cases where [H 2 O] ≤ 0.1 % and the RH sensor accuracy became a limiting factor, we systematically adjusted the [H 2 O] value that was input to the model to a value between 0.01 % and 0.1 % to achieve better agreement between measured and modeled OH exp . I 254 and I 185 values input to the model were adjusted to match the measured OH exp values as best as possible within the following constraints: 1. I 254,max = (3.5 ± 0.7) × 10 15 photons cm −2 s −1 for two lamps operated at maximum output (Lambe et al., 2019).
2. At reduced lamp output, I 254 was calculated by multiplying I 254,max by the ratio of photodetector-measured irradiance values measured at maximum and reduced lamp output at λ = 254 nm. Within these constraints, the mean (±1σ ) ratios of modeled to measured CO and SO 2 concentrations remaining at the exit of the OFR were 1.02 ± 0.06 and 0.97 ± 0.17, respectively.  (Fig. 1).  Fig. 2 shows that [O 3 ] generated using lamp types D and G was approximately 1.7 and 1.8 ppmv; here, lamp type D had one 15 mm quartz segment with T 185 = 1, whereas lamp type G had three 5 mm quartz segments with T 185 = 1. At the same OFR conditions, [O 3 ] generated using lamp types C and F was 4.5 and 2.7 ppmv; these lamps had one 35 mm and seven 5 mm quartz segments with T 185 = 1. Despite the discrepancy in measured [O 3 ], corresponding OH exp obtained with lamp types C and F were 2.5×10 12 and 2.8×10 12 molec. cm −3 s, respectively. Thus, the worse agreement in [O 3 ] measured between lamp types C and F may be associated specifically with O 3 measurements from these experiments. We hypothesize that the OFR-volume-averaged I 185 is sufficient to describe associated HO x production for these cases. Figure 3 plots OH exp as a function of T 185 at [H 2 O] = 1.90 % ± 0.26 %. The corresponding equivalent photochemical age shown on the right y axis assumes a 24 h average OH concentration of 1.5 × 10 6 molec cm −3 (Mao et al., 2009). Results obtained with lamp types D and G, and C and F were averaged together at T 185 = 0.04 and 0.1, respectively, due to their similar OH exp values. Over the range of T 185 shown in Fig. 3, excluding lamp type B, OH exp increased by approximately a factor of 5 at I 254 = (3.7 ± 0.6) × 10 15 photons cm −2 s −1 ) and a factor of 17 at I 254 = (2.1 ± 0.3) × 10 14 photons cm −2 s −1 . Maximum OH exp also decreased by about a factor of 5 between lamp types A and B due to reduction in both I 254 and I 185 (not shown in Fig. 3). Similar trends were observed for OH exp measurements at [H 2 O] = 0.93 % ± 0.06 % and 3.42 % ± 0.30 %, but at [H 2 O] = 0.09 % ± 0.07 %, the sensitivity of OH exp to T 185 was weaker due to suppressed OH production at lower humidity.
3.2 I 185 : I 254 determination and derivation of OH exp estimation equations Figure 4 plots I 185 as a function of I 254 for the Hg lamps used in this study and a different model of Hg lamps used in an earlier-generation PAM OFR . As with OH exp values shown in Fig. 3, I 185 and I 254 values obtained with lamp types D and G, and C and F were combined together into T 185 = 0.04 and 0.1 symbols following our hypothesis that the OFR-volume-averaged I 185 was sufficient to describe HO x production. Linear fits applied to the data shown in Fig. 4 Li et al. (2015), with a lower apparent sensitivity of I 185 : I 254 to lamp power. This is presumably due to differences in the specific Hg lamp types, potential variability in lamp output within the same lamp type, and/or the method of dimming used in the two studies. Previous studies reported empirical OH exposure algebraic estimation equations for use with OFRs Peng et al., 2015Peng et al., , 2018Lambe et al., 2019). These equations parameterize OH exp as a function of readily measured experimental parameters, therefore providing a simpler alternative than detailed photochemical models for experimental planning and analysis. Here, we expand on those studies by deriving OH exp estimation equations for the lamp types that were used in this study. We adapted the estimation equation format introduced by Li et al. (2015):   Table 2.
To generalize the results shown in Figs. 5 and S3 to OFR185 systems with other I 185 : I 254 values, Fig. 6 plots fit coefficients a-f as a function of I 185 : I 254 . Each of these coefficients changes monotonically as a function of I 185 : I 254 , enabling the usage of simple exponential regression functions to parameterize the a-f values as a continuous function of I 185 : I 254 . Exponential function coefficients for the regression curves shown in Fig. 6 are presented in Table 3. Figure 7 compares the equation-estimated OH exp (obtained using Eq. 1 with Table 3 fit coefficients) and the measured OH exp obtained using the tracer decay method. The mean (±1σ ) ratios of equation-estimated and measured OH exp values were 0.94±0.55, 1.13±0.48, 1.03±0.37, and 1.32±0.71 for I 185 : I 254 = 0.00167, 0.00242, 0.00595, and 0.0664.  (Fig. S4 in the Supplement).
Similarly, F photolysis, toluene ≤ 0.07 and 0.04 at low and high I 185 : I 254 (Atkinson, 1986;Serralheiro et al., 2015). Code and data availability. Data presented in this paper are available upon request. The KinSim mechanism can be downloaded from the Supplement. The kinetic solver is freely available at https://tinyurl.com/kinsim-cases#bookmark=kix.6zu8zdwq2lce. (Peng and Jimenez, 2020b).
Author contributions. AL conceived and planned the experiments. JR performed the experiments. JR and AL performed the data analysis. JR, AL, and WB conceived and planned the model simulations, and JR and AL carried out the model simulations. JR, AL, and WB contributed to the interpretation of the results. AL took the lead in writing the paper. All authors provided feedback on the paper.