Chemical processing does not always impair heterogeneous ice nucleation of mineral dust particles



[1] Exposing Arizona Test Dust (ATD) particles to nitric acid vapor in an aerosol flow tube impaired subsequent deposition ice nucleation below water-saturation, but promoted condensation/immersion-freezing on approach to water saturation and had no apparent impact on freezing of activated droplets above water saturation. The fraction of particles capable of nucleating ice at −30°C was determined using a continuous flow diffusion chamber. Exposure to HNO3 at 26% relative humidity with respect to water (RHw) reduced the fraction of particles subsequently nucleating ice to below our quantification limit in the deposition nucleation regime below 97% RHw, while leading to a sharper step-wise increase in ice nucleation between 97–100% RHw compared to unreacted dust. These observations contrast with the effect of concentrated sulfuric acid condensation, which in most cases has been reported to reduce ice nucleation of ATD and other dusts both below and above water saturation.

1. Introduction

[2] Mineral dust particles are the most numerous heterogeneous ice nuclei (IN) in the atmosphere. They are frequently detected in ice crystal residues and thus are thought to play an important role in ice cloud microphysics [DeMott et al., 2010; Stith et al., 2009]. The ice nucleating ability of mineral dust depends on its mineralogy. Aluminosilicate clays are present in a large fraction of atmospheric dust particles and display relatively high ice nucleation ability [Zimmermann et al., 2008]. Two principal mechanisms of heterogeneous ice nucleation by insoluble nuclei such as mineral dust are deposition ice nucleation and condensation/immersion-freezing [DeMott et al., 2010]. Below water-saturation deposition-nucleation, the direct formation of ice on a particle's surface from an ice-supersaturated vapor, is expected to be the primary ice nucleation mechanism. Condensation and immersion-freezing occur when an insoluble IN freezes during or following the absorption of water, respectively, typically above water saturation.

[3] Individual atmospheric mineral dust particles frequently accumulate secondary material during atmospheric transport [Shi et al., 2008; Sullivan et al., 2007; Sullivan and Prather, 2007] and this can potentially modify their ice nucleation ability. Previous laboratory studies on this subject have mostly focused on cirrus cloud temperatures below −40°C in the deposition-regime. Recent experiments performed at warmer temperatures and in both the deposition- and condensation/immersion-freezing regimes found that the ice nucleation ability of Arizona Test Dust (ATD) was reduced by the condensation of heated sulfuric acid vapors [Niedermeier et al., 2010; Sullivan et al., 2010]. Nitric acid readily reacts with alkaline components found in mineral dust particles [Liu et al., 2008; Sullivan et al., 2009a; Vlasenko et al., 2006] and alters the warm cloud nucleation properties of dust particles [Sullivan et al., 2009b], but its effect on ice nucleation has not yet been investigated. Here we present experimental results on the effect of nitric acid uptake on the ice nucleation properties of ATD at mixed-phase cloud temperatures of −30°C, both below and above water saturation.

2. Experimental Methods

[4] Arizona test dust (ATD) particles were suspended dry using a fluidized bed generator and size selected at 200 nm with a differential mobility analyzer. The pseudo-monodisperse aerosol then entered a short glass tube held at a fixed relative humidity before entering the longer flow tube. Nitric acid vapor was generated from a permeation tube, diluted, and introduced into the longer flow tube through a Teflon-lined moveable stainless steel tube. A total flow rate of 6–10 vLpm was used corresponding to a maximum reaction time of 90 to 50 s. The reacted aerosol passed through an annular denuder to remove HNO3(g) and was dried by two silica gel diffusion driers before the aerosol was split between a cloud condensation nuclei counter (CCNc) and continuous flow diffusion ice chamber (CFDC). Further details on the flow tube design and operation are provided in the auxiliary material. The system is similar to that described by Sullivan et al. [2009a]. The nitric acid volume fraction of the dust particles after reaction in the flow tube was estimated from the CCN activation curves by deriving the aerosol's hygroscopicity parameter, κapp, following Sullivan et al. [2010]; further details are provided in the auxiliary material.

[5] The ice nucleation properties of 200 nm ATD particles with and without reaction with nitric acid vapor were determined using the CSU CFDC [Petters et al., 2009], which permits the observation of ice nucleation from a continuously flowing aerosol sample. Briefly, the aerosol is passed between two ice-coated temperature controlled walls that determine the temperature and RHw (relative humidity with respect to water) the aerosol experiences. Particles that nucleate ice are counted with an optical particle counter at the outlet. Above 108–112% RHw droplets can survive a droplet evaporation section and be erroneously counted as IN. A 3% absolute uncertainty in RHw is estimated [Petters et al., 2009]. Further details are provided in the auxiliary material. We report immersion-freezing data at 105% RHw, which is high enough to produce cloud droplet at sizes ten times the selected aerosol size, captures close to the maximum number of potential IN at that temperature, and still separates ice and water droplets [DeMott et al., 2010]. RHw was scanned from 75% to 112% at a rate of ∼1% per minute at −30°C in all experiments. The total particle concentration (CN) was determined by a CPC sampling the aerosol flow entering the CFDC, to determine the fraction of total particles serving as IN: IN fraction, fIN = IN/CN. Typical background IN counts were ≤5 L−1. For typical minimum CN concentrations of 100 cm−3, this corresponds to fIN = 5 × 10−5. We therefore assign fIN = 10−4 as our limit of quantification for these experiments and do not ascribe any meaning to differences in measured fIN below this threshold. fIN was calculated at a few selected RHw values by averaging all the 1 Hz data for ±0.3% RHw around that value, and the uncertainty was estimated from one standard deviation of this average.

3. Effect of Nitric Acid on Ice Nucleation Properties of Mineral Dust

[6] Arizona Test Dust mineral particles were exposed to nitric acid vapor in the aerosol flow tube at 0% and 26% RHw using different HNO3(g) concentrations and residence times, producing a range of HNO3 exposures (= [HNO3] × residence time). The amount of nitrate added and the resulting increase in particle hygroscopicity of CaCO3 particles was previously found to depend linearly on the HNO3 exposure in a similar flow tube system [Sullivan et al., 2009a]. ATD contains about 5% by mass each of the alkaline carbonates CaCO3 and MgCa(CO3)2 [Vlasenko et al., 2005] which readily react with HNO3, especially under moist conditions, producing soluble Ca(NO3)2 and Mg(NO3)2 [Liu et al., 2008; Vlasenko et al., 2006]. Vlasenko et al. [2005] reported that calcite was not detected in individual ATD particles they analyzed, though it was present in bulk particle samples. Therefore, we assume a heterogeneous distribution of carbonate minerals between individual ATD particles; some likely contain no carbonates. We expect the primary HNO3 uptake mechanisms were heterogeneous reaction with carbonate-containing particles, and physisorption of HNO3.

[7] The amount of nitric acid added to the 200 nm ATD particles was estimated from the particle hygroscopicity measurements. For this purpose, we ignore the formation of products and treat the exposed particles as composites of the original ATD and nitric acid, using κHNO3 = 0.8 to describe the nitrate material produced. For the highest HNO3 exposure of 2087 ppb s at 26% RHw κapp was determined to be ∼0.07, corresponding to an average HNO3 particle volume of 8.3%. Dust particles reacted at 26% RHw contained significantly more HNO3 than similar HNO3 exposures at 0% RHw; the κapp values and HNO3 volume fraction estimates are listed in Table S1. The reaction probability of HNO3 with ATD aerosol, and also CaCO3 aerosol, increases as RHw increases [Liu et al., 2008; Vlasenko et al., 2006]. Further details on the HNO3 volume fraction estimates are provided in the auxiliary material.

[8] The fIN measured at −30°C as a function of RHw for the lowest and highest HNO3 exposures studied at 26% RHw, along with an unreacted ATD reference experiment, are shown in Figure 1a. In the unreacted ATD scan fIN began to increase gradually above ∼80% RHw, increasing more strongly above 95% RHw, and approaching its maximum just above water saturation. When ATD was exposed to HNO3 at 26% RHw, ice nucleation at <97% RHw, in the deposition regime, was deactivated below our quantification limit. For the two extremes in HNO3 exposure at 26% RHw, fIN increased sharply just below water saturation at ∼97% RHw, quickly reaching a relative plateau in fIN near 100–101% RHw. The ice nucleation ability of mineral dust in the condensation/immersion-freezing regime was therefore not impaired by the uptake of nitric acid. Similar exposure levels at 0% RHw (Figure 1b) produced a smaller decrease in fIN below water saturation, attributed to the smaller amount of nitric acid deposited, and possibly due to variations in the amount of nitric acid added to individual particles. We discuss these less atmospherically relevant dry exposures further in the auxiliary material.

Figure 1.

Fraction of 200 nm ATD mineral particles that nucleated ice at −30°C as RHw was scanned up after different levels of nitric acid vapor exposure. (a) Highest and lowest HNO3 exposures tested at 26% RHw, and an unexposed reference scan at 26% RHw. (b) Lowest HNO3 exposures tested at 0% and 26% RHw, and a reference scan at 0% RHw. 1 Hz data have been smoothed by a 30 s adjacent average. Grey box indicates the region where droplets can interfere with our IN measurement. Ice saturation (100% RHi) occurs at 74.6% RHw.

[9] The fractions of particles that nucleated ice at different representative RHw values following exposure to nitric acid at 0% and 26% RHw are shown in Figure 2. HNO3 exposure reduced fIN from ∼10−3 to as low as <10−4 at 95% RHw where deposition ice nucleation is expected to dominate. fIN was reduced to a lesser degree at 98% RHw, where condensation-freezing had ensued for particles containing nitric acid. Exposures at 26% RHw reduced fIN to a larger degree compared to the 0% RHw experiments, both at 95 and 98% RHw. The spread in fIN values can be partly attributed to the large range of HNO3 exposures used here. The scans in Figure 1 indicate that the sudden increase in ice nucleation following nitric acid uptake began at ∼97% RHw. Above 100% RHw, where condensation and immersion-freezing dominate, the ice nucleation abilities of reacted and unreacted mineral dusts were not significantly different. Also included in Figure 2 are results from the recent FROST-2 experiments in which 300 nm ATD particles were coated by condensation of sulfuric acid generated from a heated vapor source, resulting in ∼4% sulfate particle volume [Sullivan et al., 2010]. Exposure to sulfuric acid in this manner produced a two orders of magnitude decrease in fIN at 98% RHw. fIN was also reduced by more than one order of magnitude at 105% RHw following sulfuric acid treatment, while these nitric acid exposures had no discernable effect on immersion-freezing.

Figure 2.

Fraction of total ATD mineral particles that nucleated ice with (colored symbols) or without (grey symbols) exposure to acid, at selected RHw values at −30°C. Blue symbols: 200 nm ATD with nitric acid from these flow tube experiments at 0% (circles) or 26% (squares) RHw; Triangles (triangles): 300 nm ATD with (red) or without (grey) sulfuric acid at 0% RHw from the FROST-2 experiments described by Sullivan et al. [2010]. Numbers indicate the minimum and maximum nitric acid exposures (ppb s) used at the two RHw levels studied. The horizontal numbered bar indicates the relative humidity of the acid exposure. The grey box displays the range of fIN <10−4 which is below our quantification limit for these experiments.

4. Interpretation of Ice Nucleation Response

[10] We interpret our experimental observations of nitric acid uptake on the ice nucleation properties of mineral dust particles under the framework that heterogeneous ice nucleation begins with the formation of a critical ice embryo at a specific ice-active surface. Chemical or physical modification of these ice active surface sites can alter their critical temperature and RHw thresholds, thereby changing fIN.

[11] The nitric acid concentration in the flow tube was well below its saturation vapor pressure. Therefore, the majority of the nitric acid added to the ATD in the flow tube must have been due to heterogeneous reactions and physisorption, not condensation. HNO3 is known to react efficiently with carbonates, and the reaction rate is accelerated above the 10% deliquescence RHw of Ca(NO3)2 [Liu et al., 2008]. The resulting increase in hygroscopicity could have permitted a thin aqueous layer to form around the particle during the 26% RHw exposures. More nitric acid vapor would have dissolved into this aqueous layer, and a nitric acid coating would then have formed when the aerosol was dried before analysis. Aqueous partitioning and the accelerated reaction kinetics explain the larger change in both hygroscopicity and fIN for exposures conducted at 26% RHw versus 0% RHw. By concealing active sites the nitric acid coating impaired deposition nucleation below water saturation. Some IN ability presumably remained for the smaller 0% RHw exposures due to an incomplete surface coating or heterogeneous treatment at low HNO3 exposure. The concealed active sites were restored when water was absorbed and the coating diluted during hygroscopic growth and droplet activation, allowing the nitric acid to desorb from active sites, causing freezing of diluted drops to be unaffected. This hypothesis requires that nitric acid did not irreversibly alter active sites through chemical reactions. Alternatively, new unaltered active sites could have been revealed when the soluble reaction products were desorbed from the surface by dissolution. The observed complete deactivation of deposition-nucleation following HNO3 exposure at 26% RHw suggests that all IN-active ATD particles contained enough nitric acid to inhibit ice nucleation, either through physisorption, or heterogeneous uptake if the particle contained carbonate minerals.

[12] We interpret the sharp increase in ice nucleation ability just below water saturation at ∼97% RHw caused by nitric acid exposure as a shift from deposition nucleation to a condensation/immersion-freezing mechanism. The observed increase in hygroscopicity facilitates the absorption of water below water saturation, which can then nucleate ice when it encounters an active site, once the droplet solution is dilute enough to overcome freezing point depression impacts. This is readily accomplished in the water subsaturated regime at −30°C. The increase in hygroscopicity homogenized the population's ice nucleation behavior below water saturation, whereas unprocessed ATD required water supersaturation for freezing of absorbed water to contribute to ice formation. The sudden onset of condensation/immersion-freezing indicates that the IN-active particles did accumulate HNO3, which inhibited deposition-nucleation, but facilitated condensation- and immersion-freezing. Aluminosilicate mineral phases, which are not alkaline, may compose the most efficient active sites in ATD, and explain why chemical reaction of carbonate minerals with HNO3 did not alter immersion-freezing.

[13] The gradual increase in fIN versus RHw for unreacted ATD is likely the result of a range of active site nucleation energies in the deposition-regime, and an unresolved contribution from condensation-freezing on approach to water saturation. Exposure to nitric acid inhibits deposition-nucleation while homogenizing the aerosol population's hygroscopicity, causing the increase in fIN to occur over a much narrower range of RHw, as most particles can now freeze at the water activity limit necessary to overcome solute effects [Zobrist et al., 2008]. Exposure to nitric acid therefore produces a step-function shift between heterogeneous ice nucleation mechanisms: inhibiting deposition nucleation while facilitating condensation and immersion-freezing by increasing particle hygroscopicity without irreversibly altering ice active surface sites. To our knowledge, this is the first report of such a distinct shift between ice nucleation mechanisms caused by the uptake of an acidic vapor.

[14] The observed effect of nitric acid is in stark contrast to recent experiments in which ATD particles were exposed to hot sulfuric acid vapor; this treatment reduced ice nucleation at −30°C both above and below water saturation, although the deposition regime was impaired more than the condensation/immersion-freezing-regime (Figure 2) [Sullivan et al., 2010]. A heated 45–85 °C tube was used to generate sulfuric acid vapor and coated particles were also exposed to these elevated temperatures. Those experiments revealed the accelerated loss of ice nucleation ability when coated particles were subsequently heated in a thermodenuder at 250 °C. Thus, we postulated that the heated vapor tube may have promoted the irreversible loss of active sites but that this loss may not occur from natural sulfuric acid condensation alone [Sullivan et al., 2010]. Alternatively, the differing response may be due to different mechanisms of acid uptake by ATD (e.g. condensation versus heterogeneous reactions), or differences in reactions and product properties of nitric acid and sulfuric acid reacting with ATD components. We note that while almost all relevant nitrate salts are soluble under the particle water contents experienced in the CFDC, many sulfate compounds have low solubility [Sullivan et al., 2009b]. The neutralization of nitric acid by alkaline dust components such as CaCO3 and CaMg(CO3)2 which form soluble products including Ca(NO3)2 and Mg(NO3)2 that are subsequently desorbed by dilution from the particle surface to uncover new unaltered surface sites remains a likely possibility. In no case did nitric acid or sulfuric acid exposure increase fIN compared to untreated dust.

5. Atmospheric Implications and Conclusions

[15] Previous experiments have supported the notion that mixing of mineral dust with secondary pollutant acids, such as sulfuric acid, impairs their ability to nucleate ice [Cziczo et al., 2009; Eastwood et al., 2009; Niedermeier et al., 2010; Sullivan et al., 2010]. This study indicates that not only does exposure to nitric acid not impair particle freezing above water saturation, the addition of hygroscopic material can at some temperatures shift the onset of condensation-freezing to just below water saturation and homogenize the freezing behavior. This effect was achieved from just 8 ppb s HNO3 exposure at 26% RHw. We previously calculated a timescale of only 4 hours for the near complete reaction of pure CaCO3 particles with 10 ppt HNO3 [Sullivan et al., 2009a]. As tropospheric humidity levels frequently exceed 26% RHw and nitric acid is present generally at >100 ppt, we can expect that most atmospheric dust particles that contain alkaline minerals will have already achieved this degree of processing before experiencing cloud nucleation events. Differences in dust mineralogy have been observed to influence the type of atmospheric processing that dust experiences [Sullivan et al., 2007], and the resulting changes in the dust's apparent hygroscopicity [Sullivan et al., 2009b]. Alkaline calcite and other carbonates can undergo heterogeneous reactions with nitric acid, and with other pollutant acids such as hydrochloric and oxalic acids. Following the mechanism we have proposed, the resulting increase in hygroscopicity that facilitates further uptake of acids can produce a secondary coating that conceals ice active sites and prevents deposition ice nucleation. Immersion-freezing is unaffected if the coating is reversibly desorbed off the particle surface during droplet nucleation. The mineralogy of the source dust will therefore play an important role in the ice nucleation properties of unprocessed dust, the types of processing the dust experiences, and the resulting ice nucleation properties of aged dust.

[16] All acid exposures (sulfuric acid, nitric acid, secondary organic aerosol) experimentally tested thus far in this and other studies were observed to impair ice nucleation occurring well below water saturation in the deposition-regime [Cziczo et al., 2009; Eastwood et al., 2009; Möhler et al., 2008; Sullivan et al., 2010]. The alteration of the ice nucleation mechanism by nitric acid suggests that the effect of secondary pollutants mixing with mineral dust, and other ice nuclei, warrants further exploration and reconsideration. The addition of secondary material cannot be assumed to always impair ice nucleation, particularly when water saturation is achieved, as it is in mixed-phase clouds. As it appears that the degree of chemical processing required to deactivate deposition nucleation occurs readily under typical atmospheric conditions and residence times, it seems unlikely that a significant fraction of atmospheric dust particles will be able to nucleate ice below water saturation. However, atmospherically processed dust particles may remain efficient IN above water saturation.


[17] This work was funded by the NSF (ATM-0611936 & ATM-0841602). LM thanks UPV/EHU for a research grant that enabled her participation in this work. We thank Jeff Collett for the use of the permeation tube oven, and Amy Sullivan for analyzing permeation tube calibration samples.