Hygroscopic Properties and Respiratory System Deposition Behavior of Particulate Matter Emitted By Mining and Smelting Operations

Abstract

This study examines size-resolved physicochemical data for particles sampled near mining and smelting operations and a background urban site in Arizona with a focus on how hygroscopic growth impacts particle deposition behavior. Particles with aerodynamic diameters between 0.056 – 18 μm were collected at three sites: (i) an active smelter operation in Hayden, AZ, (ii) a legacy mining site with extensive mine tailings in Iron King, AZ, and (iii) an urban site, inner-city Tucson, AZ. Mass size distributions of As and Pb exhibit bimodal profiles with a dominant peak between 0.32-0.56 μm and a smaller mode in the coarse range (> 3 μm). The hygroscopicity profile did not exhibit the same peaks owing to dependence on other chemical constituents. Sub-micrometer particles were generally more hygroscopic than super-micrometer ones at all three sites with finite water-uptake ability at all sites and particle sizes examined. Model calculations at a relative humidity of 99.5% reveal significant respiratory system particle deposition enhancements at sizes with the largest concentrations of toxic contaminants. Between dry diameters of 0.32 and 0.56 μm, for instance, ICRP and MPPD models predict deposition fraction enhancements of 171%-261% and 33%-63%, respectively, at the three sites.

1. Introduction

Aerosol particles have substantial effects on human health, air quality, climate and biogeochemical cycling of nutrients and contaminants. As arid and semi-arid regions cover approximately a third of the global land area, dust is the most abundant particle type on a mass basis that contributes to such effects¹. The highest combination of both dust emission potential and contaminant concentration is at mining sites, where ore handling and processing operations, as well as mine tailings, contribute to the atmospheric load of metal and metalloid contaminants^2-4. Respiratory disease caused by emitted aerosol and especially wind-blown dust from mining operations is a large concern in many regions². As mining and smelting operations continue in areas with population growth, there are elevated concerns regarding the harmful effects of airborne aerosol on vulnerable populations that are susceptible to respiratory and cardiovascular illness, as well as developmental complications. The transport of trace contaminants via aerosol particles is spatially more extensive than transport by other means such as waterways, which is demonstrative of the large-scale implications of contaminant transport by aerosol particles².

A representative natural laboratory to investigate aerosol properties from mining and smelting emissions and subsequent effects on living beings is the southwest region of North America. In particular, Arizona is of major concern due to (i) extensive past and present mining across the state², (ii) high emissions of dust and metal/metalloid contaminants, especially from mining and smelting⁵, and (iii) a rapidly growing population that contains two communities that are especially vulnerable to the effects of aerosol particles: a large number of retirees who suffer from respiratory and cardiovascular diseases; and also infants vulnerable to dust ingestion from contaminated surfaces. A study in the highly populated and important commercial centers, Tucson and Phoenix (populations over one and four million, respectively; U.S. Census Bureau, 2009) reported higher metal and metalloid concentrations in dust⁶. Measurements on the outskirts of these cities near mining operations have revealed high levels of toxic elements (e.g., Pb, As, Cd) owing to condensation and coagulation of smelting vapors and wind-eroded dust from mine tailings⁷. As the dust particles can be transported downwind and serve as cloud condensation nuclei (CCN) and ice nuclei (IN), their associated trace metals and metalloids make their way into cloud and rain water⁸. Consequently, sources of toxic dust can affect regions distant from the source area and enter waterways.

While much attention is placed on chemical characterization of aerosol particles, less effort has focused on determining how the dry size of ambient particles changes upon inhalation in living beings as a result of the high relative humidities (RHs) in the human respiratory system. Respiratory system RH reaches near saturation in the lungs (~99.5%)^9-13 and is more variable in the upper head airways owing to its sensitivity to RH of the inhaled air, health of the individual, and breathing rate¹⁴. Aerosol hygroscopicity is a property that quantifies the growth of particles owing to water vapor uptake at a specific RH, with the magnitude of this growth dependent on chemical composition. Agreement between models predicting particle deposition in the human respiratory tract and experimental measurements improves after accounting for aerosol hygroscopicity in the calculations^15-22.

It is well established that coarse particles deposit most effectively in the upper respiratory tract due largely to inertial impaction, whereas fine particles (D_p < 2.5 μm) better penetrate the extrathoracic, conducting, and pulmonary airways of the human body upon inhalation^23-25. When reaching the lungs, such particles can then be transported to the blood stream by macrophages while contaminants such as metals and metalloids can diffuse into the blood stream^26-27.

The goal of this work is to report on the hygroscopic properties of particles emitted by mining and smelting operations in contrast to an urban background site in Arizona. Size-resolved chemical and hygroscopic data are first reported. Subsequent discussion focuses on using those data in model calculations to quantify the impact of hygroscopicity on deposition behavior in the human respiratory tract.

2. Experimental Methods

2.1 Field Sites

Size-resolved aerosol samples were collected at three field sites in Arizona (Figure 1) including an active smelting site, a legacy mining site, and an urban background site removed from mining. The active site is comprised of open pit mine, copper concentrator, smelter, and tailings impoundment in the twin towns of Hayden-Winkelman (referred to as Hayden hereafter). This site is about ~160 km southeast of Phoenix and 80 km northeast of Tucson. Ongoing activities at this site result in airborne particles containing toxic contaminants such as Pb, As, and Cd^7,28. The inactive mine is the Iron King Mine and Humboldt Smelter Superfund (IK) site, which is still responsible for emissions of contaminated aerosol²⁹. This site is approximately ~120 km north of Phoenix and ~280 km northwest of Tucson. Iron King was active from the late 1800s until 1969 producing Au, Cu, Pb, Zn, and Ag. Its resulting mine tailings pile covers approximately 62 ha near the town of Dewey-Humboldt, Arizona³⁰. Urban background site measurements were conducted at the Tucson Aerosol Characterization Observatory (referred to as TACO), which is on top of a building (30 m AGL, 720 m ASL; 32.2299°N, 110.9538°W) in inner city Tucson on the campus of the University of Arizona. Numerous studies have been conducted at this site with details provided elsewhere^31-34.

Figure 1.

Map of the study region showing the three measurement sites: (i) Tucson Aerosol Characterization Observation (TACO) in inner-city Tucson, (ii) the active smelting operation at Hayden, and (iii) the legacy mining site at Iron King.

2.2 Field Measurements

Sampling was conducted with a micro-orifice uniform deposit impactor (MOUDI, MSP Corporation)³⁵ with aerodynamic cut-point diameters of 0.056, 0.1, 0.18, 0.32, 0.56, 1.0, 1.8, 3.2, 5.6, 10.0, and 18.0 μm. Table 1 summarizes the number of MOUDI samples sets and other measurement details for the three measurement sites. Teflon filters were used for MOUDI sampling (PTFE membrane, 2 μm pore, 46.2 mm, Whatman). The chemical analysis procedure for samples collected at each site was the same. A portion of the filters was extracted in sealed glass vials with aqua regia and sonication at 80° C for 60 min followed by analysis with inductively coupled plasma mass spectrometry (ICP-MS; Agilent 7700 Series) for elemental analysis. A separate portion of the filters was extracted in sealed glass vials with milli-Q water and sonication at 30°C for 20 minutes for analysis with ion chromatography (IC; Thermo Scientific Dionex ICS – 2100 system) for anions and cations. The dual IC system includes AS11-HC (2 × 250 mm) and CS12A (2 × 250 mm) columns for anion and cation analysis, respectively, in addition to a 25-μL injection loop. Anion IC analysis was conducted with a 38-min multi-step gradient with potassium hydroxide eluent (2 mM to 5 mM from 0 to 8 min, 5 mM to 10mM from 8 to 20 min, 10 mM from 20 to 25 min, 10 mM to 18.67 mM from 25 to 38 min). Cation IC analysis was conducted with a 38-min isocratic method with methanesulfonic acid eluent (10 mM). The elements of interest from the ICP-MS analysis include Pb, As, and Cd, while species analyzed with IC included organic acids (acetate, formate, oxalate) and inorganic ions (Cl⁻, NO₂⁻, NO₃⁻, SO₄²⁻, Na⁺, NH₄⁺, K⁺, Mg²⁺, Ca²⁺). Hereafter, organic mass fraction refers to the sum of the three organic acids relative to the total mass of species from IC.

Table 1.

Sampling details for each MOUDI set collected in the study region shown in Figure 1.

ID	Site	Start Date	End Date	Total Hours	Flow Rate (LPM)
IK 1	Iron King	10/24/2011	10/28/2011	96	30
IK 2	Iron King	10/14/2014	10/21/2014	84	30
IK 3	Iron King	10/21/2014	10/30/2014	108	30
H 1	Hayden	11/15/2011	11/19/2011	96	30
H 2	Hayden	8/30/2012	9/3/2012	81.7	30
TACO 1	TACO	8/19/2013	8/23/2013	96	26.5
TACO 2	TACO	8/26/2013	8/30/2013	96	26.3
TACO 3	TACO	9/3/2013	9/7/2013	96	27.0
TACO 4	TACO	9/9/2013	9/13/2013	96	26.8
TACO 5	TACO	9/16/2013	9/20/2013	97	26.5

Parts of the aqueous extracts were atomized to a Differential Aerosol Sizing and Hygroscopicity Spectrometer Probe (DASH-SP, Brechtel Manufacturing Inc.)³⁶ for quantification of hygroscopic growth factors (GFs), which are the ratio of particle diameter at humidified conditions relative to dry conditions (<20% RH in this study). Atomization was conducted with a constant-rate atomizer and a desiccant dryer (Brechtel Manufacturing Inc. Model 9200). Dried aerosol particles were fed to the DASH-SP, which was used to select the diameter mode of the resultant dry aerosol size distribution (~ 225 nm in this study). Blank samples with the same water used in extraction of field samples were also aerosolized to the DASH-SP to confirm the absence of any interferences associated with the atomization process. The DASH-SP is comprised of the following in order of the direction of flow: (i) aerosol dryer, (ii) aerosol charge neutralizer, (iii) differential mobility analyzer (DMA) for selection of particles at a fixed diameter based on their electrical mobility, (iv) diffusion-based humidifier to bring particles to equilibrium at a pre-selected RH, (v) optical particle counters (OPCs) that obtain the light-scattering distribution of both dry and humidified particles, in separate channels, to ultimately retrieve the final humidified diameter using a data processing algorithm³⁷. The OPCs use diode lasers at a wavelength of 532 nm (World Star Technologies, Model TECGL-30). For these experiments, the pre-selected RHs used for humidification were 70%, 80%, and 90%. Details of the DASH-SP operation, data processing algorithm, and quality control procedures are summarized in past work^36-38.

Since hygroscopic growth factors are a sensitive function of the RH that the humidified diameter is measured at, it is convenient to quantify a single parameter value for hygroscopicity that represents water uptake across all RHs. This allows all the measurements in these experiments at the three RHs to be directly intercompared and to be extrapolated to other RHs of interest such as at the high value in lungs that is hard to reach with hygroscopicity instruments³⁹. Growth factor values, which are provided for each site and RH tested in the Supporting Information (Table S1), were converted into kappa (κ) values using the approximation in Equation 1⁴⁰:

$${\left[\mathit{GF}\right]}^3=1+\kappa \left(\frac{\frac{\mathit{RH}}{100\%}}{1-\frac{\mathit{RH}}{100\%}}\right)$$ (1)

Kappa is frequently reported by measurement studies and can range from 0 (i.e., GF = 1) for insoluble particles (e.g., CaCO₃) that do not grow to ~1.2 for NaCl particles, which are highly hygroscopic and exhibit significant growth⁴¹.

2.3 Aerosol Deposition Modeling

Modeling of particle deposition in human airways currently is conducted with a variety of different models ranging in complexity⁴². Computational fluid dynamics (CFD)-based models are among the most complex models that previously were limited to the upper respiratory tract and generally less accurate than other models even though they can cover a wider range of parameters⁴². However, in recent years more data has been collected to validate CFD-based models and they can be used for whole lung studies^43-44. On the simpler side are a variety of semi-empirical models that can provide accurate results within a narrow range of parameters. Here we use two models that fit into the latter and simpler category of whole-lung models to estimate the impact of hygroscopic growth, as measured with the DASH-SP, on particle deposition in the human respiratory system. The first model is the Multiple-Path Particle Dosimetry (MPPD, v2.1)^45-46, which quantifies deposition of aerosol particles ranging in size from 10 nm up to 20 μm for the entire lung and also regional sections (lobar, tracheobronchial, pulmonary, alveolar). Results from MPPD are compared with those from perhaps the most widely referenced whole-lung model, the International Commission on Radiological Protection (ICRP) model⁴⁷. A key difference between the models that impacts results is that ICRP accounts for the inhalable fraction of particles, which is the fraction of particles in the original volume of air that enters the nose or mouth. Full details of ICRP and MPPD model calculations are provided in the Supporting Information. These two models have been used for various applications such as examining health effects of simple salts, purely organic aerosol, pathogenic bioaerosol, and emissions from biomass burning, indoor sources, cigarettes, welding, and vehicular traffic^{14,20,39,48-51}.

3. Results and Discussion

3.1 Contaminant Mass Size Distributions

Mass size distributions of aerosol species are of importance owing to the dependence of particle deposition in human airways on particle size. Such distributions for contaminants such as Pb, As, and Cd are of special importance owing to their adverse effect on human health². Data indicate that communities close to smelters and mine tailings, especially children, have elevated levels of As and Pb in their blood and urine, respectively, due to dust ingestion and inhalation^52-59. Here we discuss the characteristic mass size distribution of these toxic species to isolate the dry diameter range of particles with the highest concentrations and thus potential for leading to adverse health effects.

Figure 2 summarizes MOUDI data for Pb, As, and Cd at two of the three measurement sites. Although not shown in Table 1, eight sample sets were collected at Hayden between August and September of 2009 for only chemical characterization purposes. For Tucson and Hayden, mass size distributions reveal a dominant concentration peak in the sub-micrometer range, specifically between aerodynamic diameters of 0.32-0.56 μm. A second, albeit smaller, concentration peak exists between 3.2-10 μm. The exception to these profile patterns is that Cd in Tucson was more abundant in the coarse mode (peak between 3.2-5.6 μm). The mass concentrations at Hayden are expectedly higher than Tucson owing to the elevated levels of contaminants emitted from the smelter site. It is further noted that the mass size distributions of these contaminants are similar for other seasons of the year at Hayden, but with higher concentrations generally in the winter months due to a shallower mixing layer⁷.

Figure 2.

Average mass size distributions of Pb, As, and Cd for (a) eight MOUDI sets collected at the Hayden mining operation in the months of August and September in 2009 and (b) five MOUDI sets collected at TACO. Error bars signify one standard deviation and markers without bars indicate only one data point used in the calculation.

Csavina et al.⁷ previously characterized the composition of particles emitted by Hayden and concluded that the sub-micrometer mode for these species is due to condensation and coagulation of smelting vapors, whereas in the coarse mode these species stem from aeolian transported fugitive dust, including especially from tailings. Similar observations of toxic metals and metalloids being associated with dust have been observed in other regions such as Australia⁵⁸. Although not shown in Figure 2, the same type of bimodal distribution with a peak in the sub- and super-micrometer modes has been observed at Hayden for crustal dust tracer species (e.g., Sc, Al, Fe)^7,61. Measurements near Australian copper and lead smelters in Mount Isa (Queensland) and a lead smelter in Port Pirie (South Australia) revealed similar mass size distributions for Pb and As with a dominant mode in the sub-micrometer fraction and a second smaller mode in the coarse fraction⁶². This suggests that a general profile of these species exists across different regions with active smelters. It is likely that Iron King exhibits a similar profile in the coarse mode owing to coated dust; however, it is possible that the sub-micrometer mode is absent as there is no active smelter. The characteristic peaks observed in the sub- and super-micrometer ranges motivate a closer look at the hygroscopic profiles of the sampled aerosol.

3.2 Size-Resolved Hygroscopicity Profile

The aerosol at the three measurement sites expectedly displayed the ability to take up water vapor (i.e., GF > 1 and κ > 0) at all RHs examined with variability as a function of size and location (Figure 3). The average (± 1 standard deviation) κ values were 0.12 (± 0.05), 0.16 (± 0.06), and 0.07 (± 0.03) for Hayden, Iron King, and TACO, respectively. Hayden aerosol were most hygroscopic in the sub-micrometer range (κ: 0.12 – 0.18) with the peak value between 0.056 - 0.1 μm. In the super-micrometer range, κ varied between 0.07 - 0.12 with its lowest value between 10 - 18 μm. Iron King aerosol differed from Hayden in that the peak κ (0.19) was highest between 0.56 - 1 μm, but similarly the lowest κ (0.12) was between 10 - 18 μm. For TACO samples, κ exhibited a maximum value for diameters between 0.32 - 0.56 μm coinciding with the peak concentrations of Pb and As. Otherwise, κ exhibited very little variability (0.054 - 0.064) at sizes below 0.1 μm and above 1 μm. On average, the sub-micrometer stages at all sites exhibited a higher κ as compared to super-micrometer sizes.

Figure 3.

Single hygroscopicity parameter κ and inorganic mass fraction as a function of aerodynamic cut-point diameter for (a) Hayden, (b) Iron King, and (c) TACO. Black marker sizes are proportional to sulfate mass fraction (range: 0.04-0.44)

3.3 Hygroscopicity and Composition Comparison

While the focus of this study is mainly on the hygroscopicity measurements and their ultimate impacts on the particle deposition model predictions, κ values are still compared here to composition data. The goal is to see how well the measured constituent concentrations govern hygroscopic growth at a fixed RH. Previous work has shown that the relative amount of organic matter to ambient aerosol is a good predictor of hygroscopic growth³⁸ with the caveat that their results were for the marine boundary layer, which is characterized by different aerosol composition. A limitation of our study is that only a subset of all species in the ambient aerosol was measured, especially among the organic fraction of the aerosol.

The data show that there was low variability in inorganic mass fraction as a function of either dry diameter or location. For example, the cumulative stage-averaged inorganic mass fraction was 0.87 (± 0.04), 0.91 (± 0.04), and 0.87 (± 0.05) for Hayden, Iron King, and Tucson, respectively. The most abundant constituent in the sub-micrometer range was sulfate at the three sites with stage-averaged mass fraction of 0.37, 0.36, and 0.44 for, Hayden, Iron King, and Tucson, respectively. The second most abundant sub-micrometer species was either sodium (Tucson), potassium (Hayden), or chloride (Iron King). The highest mass fraction in the super-micrometer range was for either sodium (Hayden: 0.26), chloride (Iron King: 0.31), or sulfate (Tucson: 0.21). Perhaps, the clearest signature of a relationship between κ and the IC measurements in Figure 3 is for the sub-micrometer range at Tucson, which is the site with the highest relative contribution of sulfate to the overall mass. Sulfate’s highest mass fraction (0.56) between 0.32-0.56 μm exceeded the maximum mass fraction at the other sites (0.44-0.49), albeit not by a large margin. When sulfate’s relative contribution decreases, a number of other species increase in their contribution, including nitrate, potassium, and calcium, which exhibited the greatest enhancement in mass fraction relative to other species above 1 μm relative to below 1 μm in Tucson.

While there is a lack of relationship between either organic or inorganic mass fraction with κ at a particular site, there is an improved relationship when combining data from the three sites. When examining sub- and super-micrometer diameters independently, κ and organic mass fraction are inversely related with statistical significance at 95% confidence based on a student’s t-test (Figure 4). Future measurements with more speciation can likely lead to an improved assessment of what chemical parameters best predict hygroscopicity at these sites.

Figure 4.

A comparison of κ and organic mass fraction for (a) sub- and (b) super-micrometer aerosol at the three measurement sites. The linear best-fit lines have their corresponding r² values shown, representing 50 data points for each panel.

3.4 Particle Deposition Results

The nature of how particles grow and deposit in the human respiratory system is complex and the subsequent discussion focuses on qualitative differences in predicted deposition behavior between dry MOUDI cut-point diameters and their associated humidified diameter after multiplication with measured hygroscopic GFs. While each MOUDI stage represents a range of diameters, we focus on the cut-point size when referring to each stage for simplicity. Since the RH experienced by inhaled particles can reach as high as 99-99.8% in the lungs^9,63, we utilize measurement-based κ values at 90% RH to derive growth factors at 99.5% using Equation 1.

Figure 5 summarizes results for the three measurement sites using the ICRP and MPPD models, with numerical values provided in Table S2 (Supporting Information). Before GF adjustment, the total deposition fraction follows the well-documented curve with minimum values in the sub-micrometer range (~0.1-1.0 μm) owing to the enhanced diffusional losses for ultrafine particles (< 0.1 μm) in the alveolar regions and inertial impaction and settling above 1 μm. After hygroscopic growth size adjustment, the cut-point diameter exhibiting the greatest change in deposition fraction is 0.56 μm with the ICRP model and 1 μm with the MPPD model; other computational work has shown the greatest change in deposition fraction is for 1 μm salt particles⁶⁴. With regard to the highest concentration of Pb and As, the cut-point diameter of 0.32 μm is of significance in terms of how humidification impacts deposition. Of the two models, ICRP predicts the greatest percent enhancement of deposition: 261% at Iron King, 207% at Hayden, 171% at Tucson. MPPD still predicts a considerable enhancement though: 63% at Iron King, 45% at Hayden, 33% at Tucson.

Figure 5.

Comparison of total deposition fraction between dry and humidified particles based on hygroscopicity data collected at (a-b) Hayden, (c-d) Iron King, (e-f) TACO. Panels a/c/e show ICRP results and panels b/d/f show MPPD results. The percent values shown signify the percent change in deposition fraction after accounting for hygroscopic growth.

Figure 6 extends the analysis to specific regions of the respiratory tract including the Head Airway (HA), Tracheobronchial (TB), and Alveolar (AL) regions. Since impaction and settling are most efficient in the HA region, humidification has the effect of increasing deposition fraction for almost all cut-point diameters examined except the smallest one (0.056 μm) when using both models, in addition to the two largest ones (5.6 μm and 10 μm) when using ICRP. In the TB region, humidification is shown the reduce deposition fraction for cut-point diameters ≤ 0.18 and ≥ 1.8 μm with ICRP, and ≤ 0.32 μm and ≥ 5.6 μm with MPPD. The AL results are qualitatively similar to TB with the exception that the aforementioned sizes are all shifted to the next smallest cut-point size.

Figure 6.

Regional deposition fraction in different airway regions: (a/d) Head Airways (HA), (b/e) Tracheobronchial (TB), (c/f) Alveolar (AL). Panels a-c and d-f show ICRP and MPPD results, respectively.

The regionally-specific results have major implications for the deposition of particles enriched with toxic contaminants in the measurements regions. For example, if particles were inhaled instead of being collected in the 0.32-0.56 μm MOUDI stage, they would change in where they deposit after accounting for hygroscopic growth. Both 0.32 μm and 0.56 μm are the sizes exhibiting the greatest increase in deposition fraction in the AL region based on the ICRP model. In terms of extensions of this work, future analysis can extend such measurements to quantify lung doses at a given site for an average subject.