Nocturnal loss and daytime source of nitrous acid through reactive uptake and displacement

Transcription

1 SUPPLEMENTARY INFORMATION DOI: /NGEO2298 Nocturnal loss and daytime source of nitrous acid through reactive uptake and displacement Trevor C. VandenBoer a,ǂ, Cora J. Young b,c,^, Ranajit K. Talukdar b,c, Milos Z. Markovic a,*, Steven S. Brown c, James M. Roberts c, and Jennifer G. Murphy a a Department of Chemistry, University of Toronto, Toronto, Ontario, Canada b Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, USA c Chemical Sciences Division, National Oceanic and Atmospheric Administration, Boulder, Colorado, USA ǂ Now at: Department of Earth Science, Memorial University, St. John s, Newfoundland, Canada ^Now at: Department of Chemistry, Memorial University, St. John s, Newfoundland, Canada * Now at: Air Quality Research Division, Environment Canada, Toronto, Ontario, Canada NATURE GEOSCIENCE 1

2 1. Supplementary Methods 1.1 Instrumentation Two state-of-the-science instruments for the detection of nitrogen oxide species were used to probe the fate of HONO(g) interacting with mineral carbonate salts and real soil suspensions with high time resolution (1 Hz), sensitivity and accuracy. A Negative-Ion Proton-Transfer Chemical- Ionisation Mass Spectrometer (NI-PT-CIMS) was used to detect the gas phase acids HCl(g), HONO(g) and HNO3(g). A detailed description of this design and testing of this instrument can be found in Veres et al. 1 and Roberts et al. 2. Briefly, acetic anhydride gas is passed through a 210 Po source to generate acetate ions that undergo gas phase proton exchange chemistry in (S-R1) with acids of greater gas phase acidity, where HA(g) is a model acid, to produce target acid ions (i.e. ΔG(reaction) < 0 for the ion-molecule reaction). Atmospheric samples are collected in a sample flow of approximately 840 sccm through a critical orifice and inlet assembly. CH3COO - (g) + HA(g) CH3COOH(g) + A - (g) (S-R1) A custom inlet was constructed for the flow tube experiments to allow for rapid online background ion count measurements (Supplementary Figure 1). Using 3-way solenoid valves, the sample flow was either pulled straight in to the flow tube or through an annular denuder (Model URG x150-3CSS, URG, Chapel Hill, NC) coated with Na2CO3(s) according to EPA Compendium Method IO-4.2. The denuder quantitatively removes gas phase acids from the sample flow while not altering the relative humidity of the sample, thereby providing on-the-fly backgrounds. The acid-stripped flow was then directed in to the ion flow tube to collect background spectra. The data were collected by successive counting (0.05 s at each mass) of the ions for: Cl - (m/z 35), NO2 - (m/z 46), and NO3 - (m/z 62) and 0.5 s for CH3COO - (m/z 59), yielding a data collection rate of approximately 1 Hz. This enables highly selective and sensitive mass-selective detection of gasphase acids at atmospherically relevant mixing ratios, on the order of ppt. A diode laser cavity ring-down system (CRDS) was used to determine mixing ratios of NO(g) and NO2(g), as well as record physical parameters of flow tube relative humidity, temperature and pressure. A comprehensive description of this instrument can be found in Wagner et al. 3. Detection limits of NO(g) and NO2(g) under dry conditions were approximately 100 and 50 ppt respectively. 2

3 Samples were collected at a 1 Hz measurement rate in two channel flows of 500 sccm, with slight overflow for a total of 1.2 slpm, during this study. In these experiments, a modified inlet for automated backgrounds every 10 minutes was constructed to allow the introduction of nitrogen gas in to the ring-down cavities during prolonged periods monitoring HONO(g) uptake on reactive substrates. Supplementary Figure 1. Schematic of flow tube and instrumentation setup for reactive uptake of HONO(g) and subsequent acid displacement reactions. The dotted line in the flow tube demarks the initial position for the reactive gas injector. Dry or humidified N2 provided the carrier gas flow in to which reactive gases were mixed. Circles are 3-way solenoid valves used for online background collection for the NI-PT-CIMS and CRDS instruments. Dashed lines on the valves indicate alternate flow paths for backgrounds. The overflow is marked as Patm and the temperature, pressure and relative humidity probes by T-P-%RH. 1.2 Reactive Gas Sources and Calibrations Calibrations of the NI-PT-CIMS were carried out using a method similar to that described in Roberts et al. 2 for HONO(g) and HNO3(g). In this study, two HCl(g) permeation sources were used (VICI Metronics, Poulsbo, WA) and had emission rates of 15 ng min -1 at 40 C certified by the manufacturer and 10.2 ng min -1 at 55 C, confirmed by collection of HCl(g) in to aqueous solution with a bubbler followed by ion chromatographic analysis and intercomparison with a high pressure standard (10.1 ± 5 % ppm, Spectra Gases, Stewartsville, NJ). The 10.2 ng min -1 device was used exclusively in the HONO(g) generation system, and the 15 ng min -1 device for acid displacement experiments and to calibrate the HCl(g) signal in the NI-PT-CIMS. A permeation device was also used to produce known quantities of HNO3(g) for calibration (VICI Metronics, Poulsbo, WA). The emission rate was quantified by UV absorption spectroscopy according to the method of Neuman 3

4 et al. 4, and collection in to aqueous solution followed by ion chromatographic analysis, as ng min -1 at 40 C. Gaseous HONO was produced by combining a 7.2 sccm output flow from the dedicated HCl(g) permeation device, with a 15 sccm flow of %RH humidified air from a bubbler. The humidified HCl(g) was passed through a packed ½ (O.D.) PFA tubing containing NaNO2 salt interspersed with 1.5 mm glass beads. The HCl(g) underwent quantitative exchange for HONO as in Roberts et al. 2. By finely tuning the temperature of the HCl(g) permeation device (i.e. the HCl(g) emission rate), loss of HONO due to the self-reaction at ppm HONO(g) levels (S-R2) was avoided. 2 HONO(g,surf) NO(g) + NO2(g) + H2O(g) (S-R2) H2ONO + (aq) + NO2 - (aq) N2O3(g) NO(g) + NO2(g) (S-R3) H3O + (aq) + HONO(aq) H2ONO + (aq) + H2O(l) (S-R4) The presence of water and low HCl(g) levels entering the salt bed also enhanced the release of HONO(g) away from the production of NOx in (S-R3) and (S-R4) during the reactive exchange process. Such problems have been previously reported in the development of these reactive exchange systems 5,6. The HONO(g) source output was determined to be 375 (±23) ppb by catalytic conversion to NO(g), followed by O3(g) chemiluminescent detection, with a conversion efficiency of > 95 % 2. The source output was diluted to concentrations between 0.5 and 5.0 ppb for calibration analysis. The CRDS did not detect any NOx(g) impurities in the HONO(g) source output. Sensitivity of the NI-PT-CIMS towards HONO(g) was 9.6 counts per ppt and calibrations showed high linearity (R 2 = 0.994). The calibrations performed by dilution in N2(g) with the HNO3(g) and HCl(g) permeation devices had sensitivities of 5 and 2 counts per ppt, respectively, across a range of ppb (R 2 HNO3 = 0.991, R 2 HCl = 0.997). 1.3 Reactive Substrate Preparation A variety of solid substrates was utilized to investigate the competing effects of surface ph, water content and acid-base chemistry on the reactive uptake of HONO(g). Supplementary Table 1 4

5 summarizes the experimental conditions and the substrates used in this study and Supplementary Table 2 depicts the available knowledge of the thermodynamics of the systems investigated. Clean Pyrex inserts were used to characterize: i) the effect of the insert on the conduction of HONO(g) through the flow tube, ii) the cleanliness of the flow tube and iii) to determine backgrounds for the system prior to experiments. Each insert was constructed by cutting a Pyrex tube lengthwise and installing small barriers at each end to allow for loading with aqueous solutions and subsequent drying to form reactive surfaces for experiment. The dimensions of each insert was approximately 20 cm in length, with an internal diameter of 1.8 cm. Sodium carbonate and bicarbonate salts were used as reactive substrates representative of the carbonate fractions in real soils. Sodium sulphate was used as a negative control on reactive uptake, since the free energy of the formation of sodium nitrite by reactive uptake of HONO(g) is not favored (Supplementary Table 2). Sodium sulphate also served as a neutral substrate (ph 6) in the relative humidity experiments where the influence of surface water on the partitioning of HONO(g) was investigated. Sodium nitrite was used as a negative control in the HONO(g) reactive uptake experiments and a positive control in the acid displacement experiments. Solid substrates were prepared on 16 cm long Pyrex inserts by pipetting 0.5 to 2.0 ml of 1 mm salt solutions (Na2CO3(s), NaHCO3(s), NaNO2(s), Na2SO4(s); > 98 % purity; ACS Reagent Grade) in to the inserts, followed by dilution and mixing with deionised water to coat the entire surface area. The solutions were dried at 80 C to form solid coatings of the surface. The extent of coverage and homogeneity achieved was evaluated visually. 5

6 Supplementary Table 1. Summary of experiments performed. The influence of liquid water and the surface solution chemical composition were investigated for enhancement or inhibition of HONO(g) reactive uptake. Mixing ratios of HONO(g), HCl(g), and HNO3(g) are given for their respective experiments, along with the relative humidity values tested for each substrate. Temperature and pressures were generally constant during experiments at 24.7(±0.3) C and 833(±7) mbar. Solid Substrate HONO(g) Uptake (0 % RH) Relative Humidity Effects on HONO(g) Uptake Acid Displacement Pyrex 3 ppb - - NaHCO3 3 ppb 4 ppb, 0/30/50/80/95 Na2CO3 3 ppb 4 ppb, 0/30/50/80/92 HCl(g): 20 ppb, HNO3(g): 60 ppb HCl(g): 20 ppb, HNO3(g): 60 Na2SO4 3 ppb 1 ppb, 0/30/50/80/87 - NaNO2 3 ppb - ppb HCl(g): 20 ppb, HNO3(g): 60 NACHTT Soil 3 ppb 2 ppb, 0/30/50/84/92 - CalNex Soil 3 ppb 3 ppb, 0/30/50/80/95 - ppb Soil samples were collected from the Boulder Atmospheric Observatory tower site in Erie, CO (NACHTT) and the Kern County Cooperative Extension in Bakersfield, CA (CalNex). These soil samples were used to determine if soils may act as sinks for HONO(g) via reactive uptake. Soil substrates for these experiments were prepared by thoroughly mixing 1 kg samples collected from the surface soil layer at both sites (within the top 5 cm of bare soil) and combining a subsample of the soil in deionised water to create a 50 mg L -1 suspension. The suspension was then shaken vigorously for several minutes, sonicated for 1 hour, and left over night to allow low solubility salts to equilibrate in solution. Soil samples were thoroughly mixed to ensure an even suspension of fine solids before removing aliquots when preparing substrates on the Pyrex inserts. No separation of organic content or solids was made before preparing inserts for investigation of the soil reactivities. This experimental setup did not explore the specific area of surfaces with high 6

7 porosities in contact with the gas phase, as would be the case for these real soils, instead they were assumed to have a surface area equal to the geometric surface area they covered. 7

8 Supplementary Table 2. Calculated standard and atmospherically relevant estimations of Gibb s free energy values for the reaction systems investigated in flow tube experiments at 0 % relative humidity from tables, where available 7. Reaction System ΔG (reaction) (kj mol -1 ) ΔG(reaction) (kj mol -1 ) HONO(g) + NaHCO3(s) NaNO2(s) + H2O(g) + CO2(g) HONO(g) + Na2CO3(s) 2 NaNO2(s) + H2O(g) + CO2(g) HONO(g) + Na2SO4(s) 2 NaNO2(s) + H2SO4(l) HONO(g) + NaNO2(s) HONO(g) + NaNO2(s) 0 0 HCl(g) + NaNO2(s) NaCl(s) + HONO(g) HNO3(g) + NaNO2(s) NaNO3(s) + HONO(g) HCl(g) + Ca(NO2)2(s) CaCl2(s) + 2 HONO(g) - - HNO3(g) + Ca(NO2)2(s) Ca(NO3)2(s) + 2 HONO(g) - - Estimates made by using atmospherically relevant mixing ratios of reactant and product gases c d CD and ΔG reaction = ΔG reaction + RT ln Q at 298 K, where Q for aa + bb cc + dd is a b AB. 2. Supplementary Discussion 2.1 HONO(g) Reactive Uptake on Dry Salts and Soil Suspensions: Proof of Concept Inserts loaded with dry substrates (~ 0.05 to 0.20 mg) were placed in the flow tube and the water content altered by modifying the relative humidity of the N2(g) sheath flow. Experiments investigating the reactive uptake of HONO(g) on to salts at 0 % relative humidity were conducted by purging the flow tube with dry N2 for at least 1 hour beforehand to ensure as much surface water was removed as possible. The position of the moveable injector was used to expose the entire surface area of salt to HONO(g). Exposure times ranged from 5 30 min. The surface area of the exposed salt in all experiments was assumed to be equal to the geometric surface area of the exposed salt. The extent of reactive uptake was monitored by observing the change in HONO(g) mixing ratio with the NI-PT-CIMS as a function of substrate surface area available for reaction (Supplementary Figure 2). To perform nitrogen mass balance and derive a mechanism for the fate of lost HONO(g), NO(g), and NO2(g) were monitored by the CRDS and HNO3(g) by the NI-PT-CIMS as the most likely N- containing products to result from decomposition or branching reactions (e.g. Supplementary 8

9 Supplementary Figure 2. Average reactive uptake of HONO(g) observations at 0 % RH on all substrates investigated between the initial injector position and full exposure of the salt/soil coated insert in the flow tube. Shaded region for A - Pyrex (n = 5), B - NaHCO3(s) (n = 9), C - Na2SO4(s) (n = 6), D - Na2CO3(s) (n = 9), E - NaNO2(s) (n = 9), and F - CalNex (n = 9) represent one standard deviation in the measurements. Note the noise in A-C and F is due to 10 Hz observations of m/z 46 for those experiments. Calculated uptake coefficients for each substrate are given in Supplementary Table 3. 9

10 Figure 3). Preliminary investigations of this reaction system were targeted to determine: i) whether the hypothesized reactive mechanism (R1 or R2) was active; and ii) if results between substrates were reproducible. Reproducibility of results for individual inserts was determined by performing a minimum of three reactive uptake experiments on up to three separately coated inserts (e.g. Supplementary Figure 4). Supplementary Figure 3. Mixing ratios of NO(g), and NO2(g) measured during n = 3 HONO(g) reactive uptake trials on NaHCO3(s) at 0 % RH. Data gaps are due to collection of backgrounds between trials. Solid lines indicate data averaged to 10 s, with error bars of one standard deviation. Shading in the upper bar indicates when the injector was retracted to expose the surface (grey) to HONO(g) or returned to its starting position (white). 10

11 Supplementary Figure 4. Triplicate analysis of HONO(g) reactive uptake on three separate pyrex inserts coated with NaHCO3(s) at 0 % relative humidity and 830 mbar. Absolute HONO(g) number densities have been normalized to the average of the initial conditions. Sharp decrease indicates retraction of the injector to expose the full insert surface to HONO(g) and the sharp increase in each trace indicates returning the injector to its initial location via adsorption and desorption on the uncoated components of the flow tube. 2.2 Calculating Reactive Uptake Coefficients Reactive uptake efficiency of HONO(g) to the prepared substrate surfaces is a function of the number of collisions between the reactant and the surface, which can be determined by the residence time of the reactant HONO(g) over the substrate surfaces. The residence time (t, s) spent in contact with the substrate surface was calculated by 11

12 t = A rd V f (E1) Where Ar is the total surface area (cm 2 ) available for HONO(g) exchange in the flow tube (changed by < 1 % with the presence of an insert), d is the injector distance (cm) from its initial location in the flow tube, and Vf is the volumetric flow rate (cm 3 s -1 ) of gases through the flow tube. The uptake rate was determined using: C t = C 0 e k mt (E2) Such that, ln ( C 0 C t ) = k m t (E3) Where Co is the observed number density of HONO(g) (molec cm -3 ) with the injector fully inserted, Ct is the number density observed with the injector at a distance, d (cm), and km is the measured first order loss rate coefficient (s -1 ). Since our experiments were carried out at atmospheric pressure, the calculated uptake coefficient required correction for potential diffusion limitations: 1 k c = 1 1 k m k D (E4) Where kd is the first order diffusion limited rate coefficient of HONO(g) to the surface in the flow tube and kc is the corrected first order loss rate coefficient (s -1 ). The value of kd was determined to be 1.1 s -1 for this flow tube setup using: k D = 3.6 D HONO r 2 (E5) 12

13 Where the diffusion coefficient for HONO(g), DHONO, is corrected from 0.57 cm 2 s -1 as determined by Hirokawa, et al. 8 in He to 0.24 cm 2 s -1 in N2 at 622 Torr N2 at 298 K, and r is the flow tube radius (cm). The correction was made for the dissimilar molecules by calculating the binary diffusion coefficient for HONO(g) (1) in N2 (2) 9-11 : 3/2 D T M1 M (1,1)* P ( 12) 12 2 M M 1 2 1/2 Where M is the molecular mass, P is the pressure (atm), Ω12 (1,1)* is the collision integral (0.90 for neutral-polar molecule interactions), σhono = 3.60 Å, σn2 = 3.61 Å are the Lennard-Jones collision diameter characteristics, yielding σ12 = 3.61 Å. The corrected rate coefficient was then used to determine the effective uptake coefficient (γ) of HONO(g) on to a given substrate: γ = 2rk c ωa f (E6) Where r is, again, the radius of the flow tube reactor (cm) and Af is the fractional geometric area of the flow tube coated with substrate on the Pyrex inserts. Diffusion limitations of this system indicate that the upper limit of measureable uptake coefficients is 2.3 x 10-5, well above our observations. The mean molecular speed (ω, cm s -1 ) of HONO(g) (M = kg mol -1 ), using the gas constant (R = J mol -1 K -1 ), and ambient temperature (T, K) was determined by: ω = 8RT πm (E7) The uncertainty in the HONO(g) source output on the observed uptake rate of HONO(g) is depicted in Supplementary Figure 5 for the NACHTT soil sample as a function of HONO(g) residence time in the flow tube. The overall increasing uptake of HONO(g) on to the substrate as the reactive surface area increases (i.e. residence time increases) is confirmation of the reactive uptake mechanism. The scatter of markers and the best fit line for the NACHTT soil suspension in 13

14 Supplementary Figure 5 illustrates that a representative uptake coefficient can be derived by incremental exposure of the coated surface for each substrate, rather than using only the measure of uptake on a fully exposed surface, since the reactive uptake coefficients found were small. Uptake on dry substrates in this work was only determined by fully exposing the surface approach and comparing to initial conditions. Statistical reproducibility, however, was pursued in these analyses with three separately coated surfaces of each soil proxy and soil suspension exposed to HONO(g) in triplicate. Thus, the dry uptake coefficients may have greater uncertainty than the remaining values which all utilize the incremental increase in exposed surface area to determine the uptake coefficient. The corrected rate coefficient value determined using E4 was <1-5 % larger than measured values, resulting in additional error up to 5 % in the calculated uptake coefficient from diffusion limitation in these experiments. In comparison to the relative error found in the linear fits of the data (12 20 %), uncertainty from diffusion limitations is appropriately captured in the reported uptake coefficients. Clearly, if there was a diffusion limitation in this system, it would not be possible to identify changes in the HONO(g) uptake value between different substrates or under changing conditions of relative humidity. This is not the case here (Supplementary Table 3). 14

15 Supplementary Figure 5. Uptake of HONO(g) on the NACHTT soil suspension determined by change in signal as a function of residence time of HONO(g) over the substrate (i.e. surface area). Note the potential variability in the calculated uptake from incremental loss of HONO(g) as a function of residence time. The dominant error in these experiments is driven by variability in the HONO(g) source output and the scan rate of the mass spectrometer. To account for this error, we collected n = 342 observations for each residence time to calculate the mean and reduce its standard error (black squares, ±s.e. ). The uptake coefficient was calculated using a linear least-squares regression fit (R 2 = 0.7) of the rate determined over the range of residence times (red line). The soils investigated were found to exhibit an alkaline nature in the presence of water. The NACHTT soil had a ph of 7.67 and the CalNex soil a ph of 7.51 in 1:1 mass ratios of soil and deionised water. The soils exhibited similar relative humidity dependences on the calculated uptake coefficients. In each, there is a general trend of increasing uptake with increasing relative humidity. The NACHTT soil shows the highest uptake coefficients as a function of relative humidity, ranging from x This value is consistent with the lower limit of the ground uptake coefficient range derived from field observations at this site 12. CalNex soil 15

16 showed reactivity with uptake coefficients ranging from x These results suggest that there may be other, more reactive, soil components or higher soil particle porosities 13 that increase the uptake or adsorption of HONO(g) to the surface at the NACHTT and CalNex field sites. Supplementary Table 3. Summary of sustained reactive uptake coefficients for HONO(g) on surfaces, salts and soil suspension under dry and variable relative humidities. The 95 % confidence interval limits are listed below each measured uptake coefficient in brackets where incremental reactive surface exposure was explored. γhono Measured (% RH) Substrate Pyrex 2.0 x Na2SO4 2.1 x 10-6 (6.2 x 10-6 ) < 1 x 10-6 (1.9 x 10-6 ) <1 x 10-6 (6.7 x 10-6 ) 1.1 x 10-6 (7.1 x 10-6 ) 2 x 10-6 (8.1 x 10-6 ) NaNO2 7.6 x NaHCO x 10-6 (2.9 x 10-6 ) Na2CO3 3.3 x 10-6 NACHTT Soil CalNex Soil (n = 2) 1.2 x 10-5 (0.7 x 10-5 ) x 10-6 (3.2 x 10-6 ) 1.0 x 10-5 (0.4 x 10-5 ) 4.9 x 10-6 (3.9 x 10-6 ) 1.4 x 10-5 (0.5 x 10-5 ) x 10-6 ( x 10-6 ) 1.1 x 10-5 (0.4 x 10-5 ) 6.3 x 10-6 (4.5 x 10-6 ) 1.4 x 10-5 (0.4 x 10-5 ) x 10-6 (3.6-5 x 10-6 ) 1.0 x 10-5 (0.4 x 10-5 ) 1.4 x 10-5 (0.5 x 10-5 ) 1.5 x 10-5 (0.3 x 10-6 ) x 10-6 ( x 10-6 ) 1.1 x 10-5 (0.5 x 10-5 ) 1.3 x 10-5 (0.5 x 10-5 ) 1.4 x 10-5 (0.6 x 10-5 ) x 10-6 ( x 10-6 ) Since good reproducibility in reactive uptake of HONO(g) was seen between inserts under dry conditions, multiple inserts were not investigated in the variable relative humidity. Salts were equilibrated at the relative humidity of interest for 10 minutes prior to measuring HONO(g) uptake on the surface. Reactive uptake kinetics were determined in these experiments by increasing the exposed reactive surface area in 2 cm intervals, collecting data for 5 min at each interval (e.g. 16

17 Na2CO3(s) in Supplementary Figure 6) and calculating the uptake coefficient as for the dry experiments (Supplementary Table 3). Supplementary Figure 6. Increasing reactive uptake of HONO(g) on to Na2CO3(s) with increasing relative humidity. Solid lines represent linear least-squares regressions and error bars are the standard error for each HONO reaction time tested (n = 342). 2.3 Acid Displacement of HONO from Reacted and Amended Nitrite Salts The final component of the HONO(g)-surface chemical system explored in these experiments was determining if reactive substrates, on to which HONO(g) was found to be taken up irreversibly, could be sources of HONO(g). In particular, a mechanism applicable towards the unknown daytime HONO(g) source - via acid displacement of nitrite salts with nitric and hydrochloric acids - was conducted on the reactive salt substrates. Efficient acid reactions of HCl(g) and HNO3(g) on mineral dust and calcite aerosols have been recently shown in a number of field and laboratory studies Furthermore, the photochemical production of HNO3(g) and displacement of HCl(g) from sea salt aerosols in coastal regions is a well-known phenomenon 22,23. Supplementary Figure 7 shows that NaHCO3(s) has undergone reactive uptake of HONO(g) at 80% RH over 16 hours. Under dry conditions the reacted salt was exposed to 35 ppb of HCl(g) and displacement of HONO(g) was found to take place immediately, reaching a maximum emission rate of 0.8 ppb s -1. The 17

18 displacement mechanism was confirmed by modulating soil exposure to HCl(g). In the absence of HCl(g) exposure, an immediate decrease in measured HONO(g) mixing ratios from 0.8 ppb s -1 to 0.15 ppb s -1 was observed, yielding a displacement efficiency of 20 % assuming the surface was entirely reacted to NaNO2(s). The continued emission of HONO(g) from the surface when not exposed to HCl(g) may be explained by desorption of HCl(g) from the flow tube surface (i.e. surface not coated with reacted salt) and subsequent deposition to the reactive surface of the coated Pyrex insert. Alternatively, the displacement and emission of HONO(g) from the surface could experience some kinetic limitations, resulting in the persistent emission of HONO(g) as the system approaches equilibrium. This experiment was repeated with Na2CO3(s) as the substrate exposed to HONO(g) over a similar time period and using HCl(g) as the displacement acid (data not shown). The displacement of HONO(g) from the reacted substrate in this case was found to be 30 ppt, much lower than the observed 0.8 ppb on the reacted NaHCO3(s), yielding a displacement efficiency of 1 % assuming that the surface was entirely reacted to NaNO2(s). The reacted ratios of HONO(g) to total Na + -salt were similar in each case, at 2 % for Na2CO3(s) and 4.5 % for NaHCO3(s). Therefore, the displacement efficiency of nitrite from various salts may be linked to the chemical structure of the salt on which the displacement is taking place. In the case of NaHCO3(s) reacting with HONO(g), the primary product is expected to be NaNO2(s), while the reaction product of Na2CO3(s) is likely a more complex mixture of products, such as Na2(NO2)(HCO3)(s) and NaNO2(s) (see R6-R8). In either case, it was found that HONO(g) can be displaced from the nitrite salt product of these reacted substrates. To determine the efficiency of HONO(g) displacement mechanism more certainly for HCl(g) and HNO3(g), and the effects different salts may exert on the process, explicit control over the form and mass fraction of the reactive precursor was undertaken by adding NaNO2(s) to the reactive salt coating such that the mole fraction of NaNO2(s) was at 40 % in NaHCO3(s) and 60 % in Na2CO3(s), instead of reacting the carbonate salt prior to the displacement. Therefore, the acid displacement reactions found here are consistent with those commonly used in HONO(g) generation sources 2,5, The displacement mechanism was confirmed by modulating the exposure of a NaNO2(s)- amended substrate to atmospherically relevant acids HCl(g) and HNO3(g) (Supplementary Figures 8 and 9, respectively). 18

19 Supplementary Figure 7. Reactive uptake of HONO(g) on to NaHCO3(s) over 16 hours at 80 % RH (top), followed by acid displacement of HONO(g) with HCl(g) at 0 % RH (bottom). A - Three HONO(g) (red, left axis) uptake trials preceded (14:00 18:00) the long duration uptake. The HONO(g) source was turned off at 10:05 and the relative humidity lowered to 0 % prior to addition of HCl(g) (green, right axis; HNO3(g) impurity, blue, left axis) at 10:30 and displacement of HONO(g) from the entire surface. The surface exposure was turned off at 11:35 and resumed at 12:10. B Magnified region of upper panel for the period when the HCl(g) was deposited to the NaHCO3(s) reacted with HONO(g), displacing HONO(g) from the surface. Note that HNO3(g) observed during this experiment is derived entirely from a trace impurity in the HCl(g) permeation source. 19

20 Supplementary Figure 8. Displacement of HONO(g) by HCl(g) at 0 % RH from A) NaNO2(s), B) 30 % mole fraction NaNO2(s) in NaHCO3(s) and C) 60 % mole fraction NaNO2(s) in Na2CO3(s) For both HCl(g) and HNO3(g), the major losses of the acids were dominated by surface adsorption processes, and so deposition of the acids to the substrates may not have been directly related to the fractional coating of the flow tube. The resulting HONO(g) displacements observed for each acid were not similar for each substrate, as found in the previous reactive uptake and displacement experiments (Supplementary Figures 8 and 9). The displacement rate of HONO(g) from NaNO2(s) was determined by observing that i) HCl(g) and HNO3(g) surface loss could be assumed constant to all available surface area, reactive or not; ii) the fraction of nitrite salts in the reactive substrates was known and assumed to cover an area of the reactive surface proportional to its mole fraction; 20

21 iii) HONO(g) emission rates were measured; and iv) displacement efficiency is the ratio of HONO(g) emitted per unit of HCl(g) or HNO3(g) lost to the fractional area of nitrite in the flow tube. Overall, HCl(g) showed a higher displacement efficiency than HNO3(g) across all substrates, reaching values as high as 20 % for amended Na2CO3(s) and the NaHCO3(s) reacted with HONO(g) overnight (Supplementary Figure 7) and always higher than 5 %, except in the reacted Na2CO3(s) experiment (Supplementary Table 4). Supplementary Table 4. Summary of displacement efficiencies for HONO(g) from reacted and NaNO2(s)-amended salts under dry conditions by HCl(g) and HNO3(g). Amended salts were all investigated in triplicate and pure NaNO2(s) in duplicate. Propagated uncertainty in the displacement efficiencies is listed in brackets. Displacement Efficiency (%) Acid NaHCO3(s) Na2CO3(s) NaHCO3(s)-amended Na2CO3(s)-amended NaNO2(s) HCl(g) 19(5) 1(1) 9(5) 20(5) 10(3) HNO3(g) - - 5(2) 6(3) 3 (1) The cause of the difference between substrate HONO(g) emissions is unclear, and this uncertainty precludes the ability to determine consistent substrate-specific displacement efficiency for these reactions. Future work aimed at resolving this should utilize a uniformly coated tube insert in place of Pyrex inserts used here so that all observed acid losses may be attributed directly to uptake on to the surface area of the exposed substrate matrix. Overall, it can be concluded that acid displacement reactions producing HONO(g) from nitrite salts are likely to take place during the day under real atmospheric conditions from the dry deposition of strong acids to the ground surface where nitrite salts are likely to be found; in particular, derived from the reactive uptake of HONO(g) on to mineral salts at night via the reactions described for the reaction product NaNO2(s) in these experiments. 21

22 Supplementary Figure 9. Displacement of HONO(g) by HNO3(g) at 0 % RH from A) NaNO2(s), B) 30 % mole fraction NaNO2(s) in NaHCO3(s) and C) 60 % mole fraction NaNO2(s) in Na2CO3(s) 22

23 2.4 Calculation of Compensation Point The compensation point ([HONO(g)]*) for emission of HONO(g) from soil pore water was calculated according to the following equation taken from Su et al. 27 : [HONO (g) ] = [HONO (aq)]+[no 2(aq) ] ( 1+ K a,hono [H + )H ] HONO (E8) where the acid dissociation constant of HONO(g) is given by Ka,HONO and the Henry s Law constant for HONO(g) is given by HHONO. 2.5 X-Ray Powder Diffraction Analysis Supplementary Figure 10. Diffraction pattern of minerals detected in a ground surface soil sample collected from the CalNex-SJV Bakersfield observation site by x-ray powder diffraction. The diffraction pattern angles of calcite (CaCO3) are overlaid in orange. The inlay shows the major CaCO3 diffraction features inside the dashed box. 23

24 3. References 1 Veres, P. et al. Development of negative-ion proton-transfer chemical-ionisation mass spectrometry (NI-PT-CIMS) for the measurement of gas-phase organic acids in the atmosphere. Int J Mass Spectrom 274, (2008). 2 Roberts, J. M. et al. Measurement of HONO, HNCO, and other inorganic acids by negative-ion proton-transfer chemical-ionization mass spectrometry (NI-PT-CIMS): application to biomass burning emissions. Atmos Meas Tech 3, (2010). 3 Wagner, N. L. et al. Diode laser-based cavity ring-down instrument for NO3, N2O5, NO, NO2 and O3 from aircraft. Atmos Meas Tech 4, , doi: /amt (2011). 4 Neuman, J. A. et al. Calibration and evaluation of nitric acid and ammonia permeation tubes by UV optical absorption. Environ Sci Technol 37, (2003). 5 Febo, A., Perrino, C., Gherardi, M. & Sparapani, R. Evaluation of a high-purity and highstability continuous generation system for nitrous acid. Environ Sci Technol 29, (1995). 6 Schiller, C. L., Locquiao, S., Johnson, T. J. & Harris, G. W. Atmospheric measurements of HONO by tunable diode laser absorption spectroscopy. J Atmos Chem 40, (2001). 7 Lide, D. R. CRC Handbook of Chemistry and Physics, Internet Version (CRC Press, 2005). 8 Hirokawa, J., Kato, T. & Mafune, F. Uptake of gas-phase nitrous acid by ph-controlled aqueous solution studied by a wetted wall flow tube. J Phys Chem A 112, (2008). 9 Mason, E. A. & Monchick, L. Transport properties of polar gas mixtures. J Chem Phys 36, , doi: / (1962). 10 Monchick, L. & Mason, E. A. Transport properties of polar gases. J Chem Phys 35, , doi: / (1961). 11 Rudich, Y., Talukdar, R. K., Imamura, T., Fox, R. W. & Ravishankara, A. R. Uptake of NO3 on KI solutions: rate coefficient for the NO3 + I - reaction and gas-phase diffusion coeffcients for NO3. Chem Phys Lett 261, (1996). 24

25 12 VandenBoer, T. C. et al. Understanding the role of the ground surface in HONO vertical structure: High resolution vertical profiles during NACHTT-11. J Geophys Res 118, , doi: /jgrd (2013). 13 Donaldson, M. A., Berke, A. E. & Raff, J. D. Uptake of gas phase nitrous acid onto boundary layer soil surfaces. Environ Sci Technol 48, , doi: /es404156a (2014). 14 Sullivan, R. C. et al. Timescale for hygroscopic conversion of calcite mineral particles through heterogeneous reaction with nitric acid. Phys Chem Chem Phys 11, , doi: /b904217b (2009). 15 Sullivan, R. C. et al. Mineral dust is a sink for chlorine in the marine boundary layer. Atmos Environ 41, (2007). 16 Sullivan, R. C., Guazzotti, S. A., Sodeman, D. A. & Prather, K. A. Direct observations of the atmospheric processing of Asian mineral dust. Atmos Chem Phys 7, (2007). 17 Vlasenko, A. et al. Effect of humidity on nitric acid uptake to mineral dust aerosol particles. Atmos Chem Phys 6, (2006). 18 Vlasenko, A., Huthwelker, T., Gäggeler, H. W. & Ammann, M. Kinetics of the heterogeneous reaction of nitric acid with mineral dust particles: an aerosol flowtube study. Phys Chem Chem Phys 11, , doi: /b904290n (2009). 19 Ooki, A. & Uematsu, M. Chemical interactions between mineral dust particles and acid gases during Asian dust events. J Geophys Res 110, doi: /2004jd (2005). 20 Hatch, C. D. & Grassian, V. H. 10th Anniversary Review: Applications of analytical techniques in laboratory studies of the chemical and climatic impacts of mineral dust aerosol in the Earth's atmosphere. J Environ Monit 10, (2008). 21 Cwiertny, D. M., Young, M. A. & Grassian, V. H. Chemistry and photochemistry of mineral dust aerosol. Annu Rev Phys Chem, (2008). 22 Dasgupta, P. K. et al. Conversion of sea salt aerosol to NaNO3 and the production of HCl: Analysis of temporal behaviour of aerosol chloride/nitrate and gaseous HCl/HNO3 concentrations with AIM. Atmos Environ 41, (2007). 25

26 23 Keene, W. C. et al. Composite global emissions of reactive chlorine from anthropogenic and natural sources: Reactive chlorine emissions inventory. J Geophys Res 104, (1999). 24 Vecera, Z. & Dasgupta, P. K. Measurement of ambient nitrous acid and a reliable calibration source for gaseous nitrous acid. Environ Sci Technol 25, (1991). 25 Taira, M. & Kanda, Y. Continuous generation system for low-concentration gaseous nitrous acid. Anal Chem 62 (1990). 26 Braman, R. S. & de la Cantera, M. A. Sublimation sources for nitrous acid and other nitrogen compounds in air. Anal Chem 58, (1986). 27 Su, H. et al. Soil nitrite as a source of atmospheric HONO and OH radicals. Science 333, (2011). 26