Eight indium-containing materials were collected at the facility as reported in a previous study (Badding et al., 2014). Three bulk materials were collected from the refinery (feedstock In(OH)₃, In₂O₃ and SnO₂ powders) at this facility. Two bulk material samples were collected from production processes: unsintered ITO (UITO) dust from a sanding operation and sintered ITO (SITO) dust from a machining operation. Three additional bulk materials were collected from the reclamation process: a mixture of waste sintered and unsintered (SUITO) dust, aerosol particles captured in a ventilation dust (VD) collector and reclaim by-product (RB) dust (Badding et al., 2014).

Particle size, density, surface area and surface chemistry of these materials were summarized previously (Badding et al., 2014). For the current study, materials were additionally characterized using scanning electron microscopy (SEM) with semiquantitative energy-dispersive X-ray analysis (EDX) to identify elemental constituents and with powder X-ray diffraction to identify crystalline constituents and specific phases (EB Scientific Enterprises, Golden, CO). Phase identifications were made by comparing the measured diffraction spectra to patterns in the International Centre for Diffraction Data Powder Diffraction File (ICDD, Newtown Square, MA) and to pertinent literature. The mass fraction of powder that was elemental indium and/or tin was determined using proton-induced X-ray emission (PIXE) spectroscopy (Elemental Analysis Corp., Lexington, KY).

To evaluate the dissolution of indium and tin from particles that are inhaled and deposited in the conducting airways, we used a model of airway epithelial-lining fluid referred to as serum ultrafiltrate (SUF) (Finch et al., 1988). This formulation includes sodium citrate and the chelator DTPA, originally added to ensure that dissolved trivalent cations remained in solution. Results of animal studies (Gwinn et al., 2013; Nagano et al., 2011; Tanaka et al., 2002), cellular in vitro studies (Gwinn et al., 2013) and human lung biopsy specimens (Homma et al., 2003) indicate that indium-containing particles with size capable of penetrating to the alveolar region of the lung are engulfed and sequestered by macrophages. As such, dissolution of indium from particles was also evaluated using phagolysosomal simulant fluid (PSF) to mimic the acidic environment inside of alveolar macrophage phagolysosomes (Stefaniak et al., 2005). In a preliminary study with PSF, it was observed that up to 70% of a dissolved indium standard adhered to the containers used to perform the dissolution experiment and for sample storage prior to quantification; however, when sodium citrate and DTPA were added to PSF (at the same concentrations as prescribed for SUF), recovery of the indium standard was 90%. Hence, the PSF formulation originally described by Stefaniak et al. (2005) was modified to include sodium citrate (0.0588 g/L) and DTPA (0.0787 g/L) for all dissolution studies.

Each study material was evaluated using the well-established static dissolution technique as described previously (Kanapilly et al., 1973). A known mass of powder was weighed in a static dissolution chamber and each chamber immersed in 80 mL of artificial lung fluid maintained at 37 °C to mimic lung temperature. The SUF solvent was maintained at pH 7.4 ± 0.2 using 5% CO₂/95% air and samples of the liquid were collected at 3, 6, 12, 24, 48, 96 and 168 h. The PSF solvent remained at pH 4.5 ± 0.1 and samples were collected at 3, 6, 12, 18, 24, 48 and 96 h then weekly for 28 days to simulate long-term phagolysosomal sequestration during the lifespan of alveolar macrophages. Triplicate samples of each study material and field blanks (dissolution chambers without powder) were prepared for each dissolution experiment.

The initial mass of elemental indium and/or tin in each sample was calculated as the product of powder mass weighed into each dissolution chamber multiplied by the fraction of metal determined by PIXE analysis. Artificial lung fluid samples were analyzed without digestion to quantify the mass of dissolved metal using inductively coupled plasma-optical emission spectroscopy (ICP-OES). The matrix-specific limit of detection (LOD) and limit of quantification (LOQ) for indium were 10 and 48 μg/L (SUF) and 6 and 20 μg/L (PSF). For tin in SUF, the LOD and LOQ values were 10 and 47 μg/L (SUF). Note that based on the results from the study with SUF, it was observed that the dissolution of indium from particles was independent of the dissolution of tin (data not shown). Recently, Andersen et al. reported that the dissolution of indium and tin were independent from ITO particles (Andersen et al., 2017). Given that there was no interference with indium dissolution, tin was not quantified in the study with PSF. Neither indium nor tin was detected in any of the field or reagent blank samples.

For each material, values of the mass fraction of material remaining were plotted versus time and multiple-component negative exponential functions fitted to the data using the Proc NLIN procedure in SAS version 9.3 (SAS Institute, Cary, NC) and Sigma Plot version 12.5 (Systat Software Inc., San Jose, CA) (Chazel et al., 2000): (1)fractionremaining=fr·exp(srt)+fs·exp(sst) where f = the proportion of the total element dissolved in the initial rapid (f_r) or latter slower (f_s =1 − f_r) dissolution phase, s = the dissolution rate for the initial rapid (s_r) or latter slower (s_s) phase in d⁻¹ and t = elapsed time (d). Values of f_r and f_s were determined from the intercepts of the fitted curves. The material fractional dissolution rates (s_r and s_i) were equal to the slopes of the fitted curves. Particle dissolution half-times (t_1/2) were calculated from the relationship ln2slope.

A major objective of this study was to highlight the effect of material form on dose. Though particle size distribution, density and other physical factors are important for estimating pulmonary deposition, the focus herein was not to estimate dose in a specific region of the lung to evaluate some toxicological end point, rather it was to illustrate the effect of chemical form and its importance for exposure assessment and generation of metrics for epidemiologic studies. To achieve this objective, particle deposition and clearance in the respiratory tract were modeled using the Activity and Internal Dose Estimates (AIDE) software (Bertelli et al., 2008). The software uses the International Commission on Radiological Protection (ICRP) human respiratory tract model (HRTM) to calculate the mass of indium remaining in the different respiratory tract compartments (lung burden) and accounts for mechanical clearance and dissolution clearance in biological fluids (ICRP, 1994). The ICRP model uses a simplistic lung morphometry assuming the lung acts as a series of filters that dictate regional lung deposition for particles across a range of sizes (Supplemental Figure S1a). Since clearance from the tracheobronchial and head airways compartments are relatively rapid and indium lung disease is manifest primarily in the alveoli, we focused on long-term dose from exposure to respirable size particles, that is, mass median aerodynamic diameter (MMAD) = 4 and geometric standard deviation (GSD) = 2.5. Rather than use the ICRP default MMAD of 5 μm, we defined respirable size particles as having a MMAD of 4 μm based on the convention for “respirable” air samplers, which have a 50% collection efficiency at that diameter (Lippman, 1999). In this situation, the GSD of 2.5 has more influence than the MMAD in modeled particle lung deposition. For example, at a GSD of 2.5 the alveolar regional deposition (compartment of interest for indium lung disease) values are 9.9% and 9.6% for 4 and 5 μm particles, respectively. Note that in addition to the size distribution, density can exert significant influence on particle regional lung deposition, especially for scenarios in which particle sizes are small and in the Stokes regime (Miller et al., 2013). A range of densities were previously measured for the indium study materials (Badding et al., 2014). For the case when the exposure material is In₂O₃ particles with a density of 7.18 g/cm³, then a particle with aerodynamic diameter of 4 μm will have a physical particle size of 1.44 μm. At the lower end of measured densities, if the exposure material is particles of In(OH)₃ with a density of 4.38 g/cm³, then the physical diameter for these lower density particles is 1.87 μm. Hence, in this situation, for the indium facility under study, even though the different densities of these materials results in different physical diameters for particles of the same aerodynamic diameter, those particles will behave the same in terms of their regional deposition in the human respiratory tract. For mechanical clearance, the default rate constants were obtained from the ICRP HRTM (see Supplemental Figure S1b). For the biokinetic components of the model, additional code was obtained from the author of the software to implement the systemic distribution and clearance of indium by ICRP 30. The fractional coefficients were: marrow 0.3, liver 0.2, kidney 0.07, spleen 0.02 and other organs 0.41 with indefinite retention in the organs (ICRP, 1981). Hoet et al. and Castronovo and Wegner reported that indium kinetics in the blood followed a biphasic clearance with half-lives of 1.9 to 2.8 days for the fast phase and 60 to 69 days for the slow phase (Castronovo & Wagner, 1973; Hoet et al., 2012). Thus, rather than using a fractional clearance rate for indium in blood of 2.77 d⁻¹ (half-life of 0.25 days) we used 0.011 d⁻¹ (half-life of 65 days) as the slower rate is more consistent with the protracted release of poorly soluble forms of indium from the lungs as well as evidence from human data showing prolonged presence of indium in blood years after the cessation of exposure (further, the software cannot accommodate biphasic clearance) (Hoet et al., 2012). The use of a single clearance rate constant of 0.011 d⁻¹ provides a reasonable estimate for modeling clearance from the respiratory tract of both acutely and chronically exposed workers. For chemical clearance (see Supplemental Figure S1c), the material-specific indium dissolution rates (d⁻¹) and fractions were obtained from the dissolution studies. The AIDE software only permits input of a single set of dissolution parameters for all respiratory compartments. For modeling purposes, we used the rates and fractions determined from the PSF studies because, as noted above, indium lung disease is primarily manifest in the alveoli.

Modeling was performed assuming a reference worker at light work, breathing through their nose only and following a pattern of 8 h of exposure (deposition and clearance) followed by 16 h of clearance each day for 5 work days followed by 2 days off representing clearance only. Two exposure scenarios were modeled. In the first, exposure duration was taken to be 2 years, the median tenure for the workforce at this facility (Cummings et al., 2014), followed by a clearance period of 53 years. In the second scenario, exposure duration was 40 years (working lifetime) followed by a clearance period of 15 years for the faster clearing VD dust to 60 years for the slower clearing indium oxide. For the purposes of this study, clearance was considered to be complete when the amount of indium remaining in a respiratory tract compartment or blood reached 0.01% of the initial mass inhaled. Though we focused on a light work scenario as this is the predominate type of work at this facility, breathing rate influences deposition and the same scenarios described herein were modeled for a heavy breathing (100% nose breather) worker and results are provided in Supplemental File 2. The mass fraction of indium distributed among various respiratory tract compartments and to blood were plotted against time using Sigma Plot to explore dose and retention patterns for the different exposure scenarios.

All bulk materials were composed of particles with sizes capable of penetrating to the alveolar region of the lung (Badding et al., 2014). On the surfaces of the three ITO materials, indium was present as a mixture of indium metal (In⁰) and In₂O₃ at similar proportions; tin was present in the form of SnO (not SnO₂). The VD collector and RB particle surfaces were consistent with indium salt and In₂O₃ (also In⁰ on the surface of the RB dust). Elemental analysis by EDX confirmed that the feedstock powders were high purity (i.e. indium to oxygen ratios approached values expected from stoichiometry). Assuming that the ITO powders were a 90:10 mixture of In₂O₃ and SnO₂, the ratios of indium to oxygen and tin to oxygen also indicated high-purity materials. The VD particles gave strong EDX signals indicating that this material contained an indium salt and the RB dust produced weaker signals indicative of lower amounts of the indium salt. From X-ray diffraction analysis, the feedstock indium hydroxide, indium oxide and tin oxide were pure crystalline In(OH)₃, In₂O₃ and SnO₂, respectively. All ITO materials were crystalline cubic In₂O₃ and SnO₂. The particles from the VD collector contained crystalline In₂O₃. The diffraction pattern of RB dust was consistent with a highly crystalline material though none of the peaks matched reference patterns for In₂O₃, SnO₂ or ITO. PIXE analysis revealed that the indium or tin content of the feedstock powders matched values based on stoichiometry. Levels of indium in the ITOs were similar. Among the eight study materials, levels of tin ranged from 0.3% (VD) to 8% (SUITO, SITO).

As shown in Figure 1(a–c), dissolution of indium from all indium-containing study powders in SUF was biphasic, that is, consisted of an initial rapid phase followed by a later slower phase. Tin dissolution followed the same bi-phasic pattern (data not shown). In SUF, the total mass of indium dissolved was greatest from the VD and RB materials (Figure 1(a)), intermediate for In(OH₃), UITO and SITO (Figure 1(b)) and least for In₂O₃ and SUITO (Figure 1(c)). Interestingly, the mass fractions of dissolved indium differed among the ITOs despite having the same crystalline phase of indium and nearly identical surface and bulk chemistries. As shown in Table 1, the rank order of specific surface areas of the ITOs were (from highest to lowest) UITO >SITO >SUITO. Hence, the greater the available surface area for solvation, the more indium that dissolved, which explains the observed differences in fractional dissolution rates among the ITO materials. Table 1 summarizes the cumulative amount of indium and/or tin dissolved, and the fitted dissolution parameters (f_r, f_s, s_r and s_s) for the industrially sampled powders in SUF. The calculated values of s_r for indium varied from 0.25 to 6.15 d⁻¹ among study materials. Values of s_s for indium spanned two orders of magnitude.

Based on the SUF data, three representative indium-containing materials that spanned the range of observed solubilities (VD, SITO and In₂O₃) were further evaluated in PSF to understand the alveolar fate of indium. As shown in Figure 1(d), in PSF dissolution of indium from In₂O₃, SITO and VD particles was biphasic. As summarized in Table 2, the mass fractions of indium that dissolved differed by material and followed the rank order (from greatest to least): VD >SITO >In₂O₃. For each material tested in the acidic PSF model, the mass fraction of indium dissolved was greater than in SUF at near-neutral pH. Table 2 also summarizes the fitted dissolution parameters for these materials that were used to model lung dosimetry.

Figure 2(a) is a plot of alveolar indium retention predicted by the ICRP HRTM for a two-year (median tenure) exposure to respirable sized particles of In₂O₃, SITO and VD particles. The clearance aspect of the model includes the contributions from both indium chemical dissolution (as measured in PSF) and mechanical particle removal mechanisms (i.e. particle biopersistence). From this plot, indium lung clearance rates in the alveolar compartment were slow for all study powders though long-term indium retention differed among chemical forms. Specifically, three distinct alveolar retention profiles for indium were evident. Following cessation of exposure, indium in the form of salts in VD particles would need 4 years to clear, indium in the form of SITO would clear in 9 years and indium as In₂O₃ would require 48 years.

Of interest is the slow clearance of indium from blood following a two-year exposure as shown in Figure 2(b) because this matrix is often used as a biological marker of exposure (Harvey et al., 2016). In the ICRP HRTM, indium clearance to blood is a reflection of only the chemical dissolution rate. Note that our modeling results are based only on measurements in PSF because our emphasis was on alveolar aspects of indium lung disease and in vivo dissolution in the airway lining fluid (mimicked by SUF) is only relevant for the first few days following particle deposition, as such, dissolution by alveolar macrophages (mimicked by PSF) would mainly contribute to indium clearance to blood. The ICRP model predicted that following two years of exposure to indium in the form of VD, SITO and In₂O₃ indium would remain in blood for 3.5, 6.5 and 18 years, respectively and is systemically distributed to other organs in the body over time.

For a 40-year exposure (Figure 2(c)), the ICRP HRTM predicted that indium would quickly accumulate in the alveoli before reaching a plateau. The same three general alveolar retention profiles by chemical form of indium compound were also observed for a 40-year exposure. After cessation of 40 years of exposure, the model predicted it would take 5 years to clear the indium from the alveoli for VD particles, 10 years for SITO particles and 60 years for In₂O₃.

Figure 2(d) presents plots of blood indium retention predicted by the ICRP model for a 40-year (working lifetime) exposure to respirable size particles. For this scenario, the model predicted that following exposure to indium in the form of VD, SITO and In₂O₃ indium would remain in blood for 5, 10 and 53 years, respectively.

The model results presented in Figure 2 are for a reference worker at light work; however, some jobs at this facility require more exertion, which will affect breathing patterns and therefore particle deposition and indium retention. To understand the influence of work pattern on indium retention, we modeled alveolar and plasma indium on fractional retention for VD, SITO and In₂O₃ indium dusts under heavy exercise and 100% nose breather (see Figure S1 in the Supplemental File) for the same 2- and 40-year exposure scenarios as described earlier. Among the forms of indium dusts, for the 2-year exposure scenario, heavy exercise increased indium retention in the alveoli by 31–36% compared to light work and increased plasma retention by 22–35%. For the 40-year exposure scenario, heavy exercise increased indium retention in the alveoli by 34–37% and indium retention in the plasma by 24–35%.