Comparison of Two Photolytic Calibration Methods for Nitrous Acid

. Nitrous acid (HONO) plays an important role in tropospheric oxidation chemistry as it is a precursor to the hydroxyl radical. Measurements of HONO have been historically difficult due to instrument interferences and difficulties in sampling and calibration. The traditional calibration method involves generation of HONO by reacting hydrogen chloride vapor with sodium nitrite followed by quantification by various methods (e.g., conversion of HONO to nitric oxide (NO) followed by chemiluminescence detection). Alternatively, HONO can be generated photolytically in the gas-phase by reacting NO with OH 10 radicals generated by H 2 O photolysis. In this work, we describe and compare two photolytic HONO calibration methods that were used to calibrate an iodide adduct chemical ionization mass spectrometer (CIMS). Both methods are based on the water vapor photolysis method commonly used for OH and HO 2 calibrations. The first method is an adaptation of the common chemical actinometry HOx calibration method, in which HONO is calculated based on quantified values for [O 3 ], [H 2 O], [O 2 ], and the absorption cross sections for H 2 O and O 2 at 184.9 nm. In the second, novel method the HONO concentration is simply determined 15 based on the simultaneous measurements of NO 2 formed by the reaction of NO with HO 2 from the H 2 O photolysis. This second, novel approach generally has an improved (lower) calibration uncertainty and is simpler to apply. Calibration uncertainties are typically 30 to 36% (2σ) for the actinometric method and as low as 9% (2σ) for the NO 2 proxy method, limited by the uncertainty of the NO 2 measurements.

1 Supplementary Information

S1 Detection of HONO by Chemical Ionization Mass Spectrometry
The following sections detail the analytical parameters and humidity effects of the chemical ionization mass spectrometry (CIMS) HONO measurements.

5
The signal-to-noise ratio (SNR) and limit of detection (LOD) for our CIMS measurements of HONO are determined by the precision of the total and background signals, both of which are dominated by shot noise.The SNR is given by SNR = (ST -SB)/( ½ where ST is the total signal, SB is the background signal, σT is the precision of the total signal, and σB of the background signal.For our lab experiments the background signal was elevated due to HONO impurities associated with the high NO mixing ratios used (1.58 ppmv).For field measurements from 2019 in Boise, Idaho, our unnormalized I(HONO) - 10 background signal was 228 counts/s, corresponding 76 ncps.With a sensitivity of 2 ncps ppt -1 (that of a typical ambient humidity), the resulting 1 s LOD was 8.0 pptv (SNR = 2).An upcoming manuscript focused on the 2019 field measurements will have further discussion of limit of detection.
The CIMS I(HONO) -signal responds linearly to [HONO] for a given humidity as demonstrated by the multipoint calibration curve in the main text (Fig. 3).We have yet to experience HONO concentrations that deviate from a linear trend.Linear response 15 in I(HONO) -signal is shown in the multipoint calibration figure (Fig. 3) for a 450 to 3,400 pptv range, and past multipoint calibrations have shown linear response for wider ranges that extend above 10,000 pptv.

S1.2 Humidity Effects
The ionization chemistry utilized is sensitive to humidity.HONO can be ionized by reaction with either I -or the iodide-water adduct I(H2O) -to form the detected adduct I(HONO) -: 20 I -+ HONO → I(HONO) - RS1 I(H2O) -+ HONO → I(HONO) -+ H2O RS2 The reverse reaction of S2 also occurs: I(HONO) -+ H2O → I(H2O) -+ HONO RS3 Reactions RS2 and RS3 lead to variation in the sensitivity depending on ambient water vapor levels.The CIMS sensitivity to 25 HONO decreases with ambient water vapor concentration as shown in Fig. 4 of the main text.Many compounds sampled by I - CIMS exhibit similar sensitivity-water vapor trends, though to varying degrees (Lee et al., 2014).As a result, sampling in dry conditions with low ambient water vapor mixing ratios allows for more sensitive detection of many compounds by I -CIMS including HONO.Drastic changes in sensitivity from atmospheric variability can be suppressed by constant dilution of the IMR with humidified nitrogen (as performed here) or by maintaining constant water vapor concentration (Veres et al., 2020).

30
S2 Impact of Product Branching Ratio for HO2 + NO 2 We account for the impact of the small yield of HNO3 from the HO2 + NO reaction (R6b in main text) on the two HONO calibration methods.The branching ratio for these reactions is determined using literature temperature, pressure, and humidity dependences.Butkovskaya et al. (2007) present the quantity β, which is the ratio of reaction rates (kR7b/kR7a), rather than a traditional branching ratio (kR6b/(kR6a + kR6b).First, we determine β under dry conditions using only the temperature and pressure 35 dependences.This first form will be referred to as β * .
* (, ) = 0.01 where T is temperature in Kelvin and P is pressure in Torr.The 2σ uncertainties for the constants of the three numerical terms are ±10, ±1.3 and ±0.07, respectively.A humidity factor ƒH2O is then applied to β * to determine β (Butkovskaya et al., 2009).
Here, [H2O] is expressed in number concentration (molecules cm -3 ).β is useful for accounting for the [HNO3] if the final [NO2] is observed after all HO2 is processed by R6a and R6b: (S4)

45
In the photolytic calibration involving standard O3 actinometry, the [HOx] generated at the point of H2O photolysis is quantified.
Therefore, a branching ratio is required to account for the small portion of the HO2 initially formed that does not form HONO because of R6b.Here, we define branching ratio as βBR.

S3 Uncertainty Propagation for Actinometric Calibration
The uncertainty in [HONO] quantified by the actinometric method involves adding in quadrature the relative uncertainties of each variable in Eq. (3) (see Sect. 2.2.1 of main text) as shwon in Table S1.The resulting [HONO] relative uncertainty is 26.9 % 55 (2σ).A small uncertainty is also associated with the term β used in the correction equation that converts [HOx] to [HONO] (Eq. (4) in the main text).Because of this small correction, we round up and assign an overall uncertainty of 27 % (2σ) to the calibration.
The σO2 uncertainty is the largest contributor to this overall [HONO] uncertainty and was determined experimentally.The uncertainty in σH2O was the same as used by Dusanter et al. (2008).We choose a [H2O] uncertainty of 5 % (2σ) because the Vaisala HMP60 RH/T probe used for [H2O] quantification in this manuscript had not recently been factory calibrated but agreed to within 60 1.1 % with a brand new Vaisala HMP60 sensor that has a manufacturer-stated accuracy of 3 %.In a separate experiment, the RH/T probe was compared to the CIMS exhaust RH/T probe (also a Vaisala HMP60) and agreed to within 3.5 %.The [O2] uncertainty is based on the range stated by the gas manufacturer (Airgas), and the uncertainty of [O 3 ] is that of the CAPS NO2 instrument (see main text Sect.2.2), which effectively measures Ox (O3 + NO2).

S4 Uncertainty Propagation For NO2 Proxy Calibrations
Uncertainty calculations for the NO2 proxy calibrations are mentioned in Sect. 3 of the main text.In the following sections, uncertainty equations are provided for the multipoint calibration curve (Fig. 3 of main text) and the sensitivities shown in Fig. 4 70 (spanning several humidity values) determined by single point calibrations.
The terms added in quadrature are 1.) the background subtracted [NO2] relative uncertainty, 2.) the CAPS NO2 measurement accuracy (i.e.3%), and 3.) the relative uncertainty of β as applied in Eq. ( 5) of the main text (i.e.including the addition of 2 as a constant).The uncertainty propagation for the NO2 background subtraction is accomplished by combining absolute NO2 uncertainties in quadrature (see numerator of first term) using a set value of 27 pptv for σNO2 based on the 5 s average precision.
The resulting absolute uncertainty in NO2 is converted to relative uncertainty by its quotient with the background subtracted [NO2] 85 value (i.e. the denominator where S and bkg represent the signal and background values of [NO2], respectively).The relative uncertainty of the third term in Eq. ( S9) is very small (typically 0.14%, 2σ) due to the constant addition of 2. The value of σβ is determined by combining the uncertainties for the variables of Eq. (S1).

S4.2 Single Point Calibration Uncertainty Calculations
represented by σ, and the CIMS signals in counts per second are represented with S. Normalized CIMS HONO signals are scaled by a 10 6 factor for this manuscript, in which the relative error calculated by Eq. (S8) is maintained.