Introduction

Exposure of humans to arsenic (As)-contaminated soil is a worldwide concern due to a historical legacy of contaminated sites. Its presence in the environment is due to mobilisation of the element during natural processes, such as weathering of the regolith, biological activity, as well as through anthropogenic activities such as mining, agriculture and nonagricultural activities. In Australia, As contamination has resulted from the use of As-based herbicides, pesticides, tanning solutions, timber preservatives as well as through mining and smelting processes (Smith et al. 2003). Surface-soil As contamination in Australia has been reported extensively (McLaren et al. 1998; Smith et al. 1999, 2006) with As concentrations of up to 15,000 mg As kg−1 being described (Ellice et al. 2001). There is considerable community concern due to reports that link the exposure of As to human carcinogenesis and the development of numerous health disorders (Lien et al. 2001; Mandal and Suzuki 2002).

The incidental ingestion of soil and dust by children through dietary and nondietary pathways are important exposure pathways to be considered when assessing the potential health risks associated with exposure to As-contaminated soils. The relative bioavailability of soil contaminants has been considered to be 100% (Roberts 2004) and therefore total As concentrations in contaminated soils have previously been utilised to determine potential As exposure. However, several recent studies have identified that the potential As exposure is considerably less than the total As concentration in the contaminated soil (Ruby et al. 1996; Rodriguez et al. 1999; Juhasz et al. 2007a). In each of these studies, the in vivo swine model was used to assess As bioavailability in contaminated soils and correlated to As bioaccessibility utilising chemical approximations of the human digestive system. The in vitro studies, such as the simplified bioaccessibility extraction test (SBET), the physiologically based extraction test (PBET) (Ruby et al. 1996) or the in vivo gastrointestinal model (IVG) (Rodriguez et al. 1999) utilise chemical studies that are similar to that of the gastrointestinal tract of humans and simulate the gastric (dissolution) and intestinal (absorption) phases. The <250-μm soil particle size fraction is currently the default soil fraction that is utilised for assessing the bioavailability or bioaccessibility of a contaminant for in vivo or in vitro studies (Rodriguez et al. 1999). In a recent review of soil particle size and contaminant partitioning in contaminated sites relevant to human exposure, Bright et al. (2006) identified that there was considerable variability in the distribution of contaminants across different particle size fractions and it cannot be assumed that higher concentrations of contaminants are associated with smaller soil particle fractions. Contaminant exposure through the ingestion or inhalation (particle matter <10 μm) of indoor dust are important exposure pathways for children as their behavioural patterns expose them to greater rates of soil ingestion and elevated inhalation of contaminated soil and dust (Mukerjee 1998; Belluck et al. 2003). Indoor dust generally consists of particles <50 μm in diameter (Edwards et al. 1998) and several Pb studies have identified exposure of Pb-contaminated household dust as a major exposure pathway for children less than 3 years of age (Hunt et al. 1993; Simon et al. 2007). Children’s exposure may therefore be considerably greater if the As concentration in the smaller particle fractions is higher than that in the 250-μm soil particle fraction. Studies by Lombi et al. (2000) and Smith et al. (2006) identified that soil As concentrations increase with decreasing particle size, while Ruby (2004) reported that the bioaccessibility of contaminants in small soil particles is typically greater than that of large soil particles. However, there have been few studies that have investigated the bioaccessibility of As across different soil fractions that may be relevant to human exposure other than the <250-μm soil particle fraction.

The objective of this study was to investigate the total and bioaccessible As across a range of soil particle size fractions in contaminated soils which may be pertinent to various exposure pathways for human health risk assessment (e.g. incidental ingestion, inhalation).

Methods and materials

Study soils

Fifty soils were collected from different regions of Australia where soils had been contaminated with As through historical agricultural or industry practices, or where As is naturally present at elevated concentrations. The pertinent properties of these soils have been previously reported by Juhasz et al. (2007a, b) and will not be reiterated here. The soils collected included: eight soils from areas that contain elevated As of geogenic origins, 18 soils from areas along former railway corridors where As was the active ingredient in herbicides used to control grass growth, 13 soils from areas where As was the active ingredient in pesticide formulations used to control ticks at cattle dips and 11 soils from areas contaminated through the disposal of gold-mine tailings. The collected soils were oven-dried at 40°C, sieved through 2-mm stainless-steel sieves and stored in polyethylene containers until required.

Soil particle fractionation

Particle size separation of different soil fractions was conducted with a subset of 29 of the original 52 soils collected. The 29 soils included six soils railway corridor soils, ten cattle-dip soils, six mine site soils and seven geogenic soils. The soils were separated into different soils particle sizes by sieving the 2-mm soil fraction through 250- and 100-μm stainless-steel sieves. A subsample of the 250-μm soil fraction was further separated into the <10- and <2.5-μm particle fraction using a modified pipette method (Gee and Bauder 1986). In brief, 10 g of the 250-μm soil fraction was added to a 50-ml polyethylene tube with deionised water. The suspensions were mixed overnight on an end-over-end shaker (10 rev min−1), left to stand for 5 min and the suspended particles collected for the <10-μm fraction. The suspension was frozen, then freeze-dried before metal analysis of the particle size fraction was conducted. A similar method was used to separate the <2.5-μm soil particle fraction except that the soils were left to stand for 60 min after mixing overnight before the suspended particles were removed, frozen and freeze-dried. The separation of the <10- and 2.5-μm fractions was conducted in duplicate and the resultant separated fractions combined into one composite sample for each particle size.

Soil analyses

Each particle size fraction (<2-mm, <250-μm, <100-μm, <10-μm and <2.5-μm) was digested with aqua regia using USEPA 3052 method. The digested samples were diluted to 20 ml with deionised water and filtered through 0.45-μm filters prior to analysis by inductively coupled plasma mass spectroscopy (ICP-MS). Samples were digested in duplicate, with the inclusion of the appropriate number of blanks and standard reference material (NIST-SRM 2711) for quality assurance and quality control. An average of the blank values was subtracted from the analytical data and the standard reference soil was utilised to assess the quality control of the digestion process. The recovery for As was 97 ± 8% for the NIST-SRM 2711 soil (n = 6).

Bioaccessibility assessment

The simplified bioaccessibility extraction test was used to evaluate the bioaccessibility of As in the <250-, <100- and <10-μm particle size fractions of the contaminated soils. This method only includes gastric phase extraction: extending the method to include an intestinal phase extraction does not improve the estimation of As bioaccessibility (Rodriguez et al. 1999; Basta et al. 2001; Juhasz et al. 2007a). Soil (0.4 g) and 40 ml gastric solution (30.03 g l−1 glycine adjusted to pH 1.5 with concentrated HCl) were added to 50-ml polyethylene screw cap flasks (Kelley et al. 2002). The pH was noted then flasks were incubated at 37°C, 40 rpm on a Ratek suspension mixer. After 1-h incubation, the pH was determined and gastric phase samples (10 ml) were collected, filtered through 0.2-μm filters and analysed by ICP-MS. To ensure internal quality control practices, a standard reference soil (NIST-SRM 2711) was also included in the bioaccessibility assays. Arsenic bioaccessibility was calculated by dividing the SBET extractable As by the total soil As concentration (aqua regia extractable As) as described in Eq. 1:

$$ {\text{In vitro As bioaccessibility}},\;\% = \left[ {\frac{\text{SBET\;As}}{{{\text{Total}} {\text{ As}}}}} \right] \times 100 $$
(1)

Statistical analysis

Statistical treatment of the data was performed using SPSS software and @RISK software. @RISK is a Microsoft Excel add-in software package.

Results and discussion

The total As concentrations in the 2-mm fraction (data not shown) in the soils studied generally exceeded the current Australian National Environmental Protection Measure (NEPM) for the Assessment of Site Contamination guideline value for residential housing of 100 mg As kg−1 (NEPC 1999). The total As concentrations in the <250-μm particle fraction ranged from 13 to 12,781 mg As kg−1, with several soils collected from ore processing sites containing considerable concentrations of As (Table 1). The <250 μm particle size fraction was selected for this study as it has been reported that it is this fraction that adheres to the hands and is thus available for incidental ingestion (Rodriguez et al. 1999).

Table 1 Selected properties of soils used in this study
Full size table

Arsenic bioaccessibility in the <250-μm soil particle fraction has been reported previously (Juhasz et al. 2007a, b); these results are included in Table 1 for comparison with the total As distribution and As bioaccessibility across the other soil particle size fractions studied. Juhasz et al. (2007a) reported that the SBET extractable As correlated with in vivo As bioavailability investigations with immature swine. Assessment of As bioaccessibility in the <250-μm soil particle fraction showed that it varied markedly across the 50 soils investigated (Table 1), with the relative As bioaccessibility ranging from 1% to 89%. Across all soils, the mean relative bioaccessible As fraction was 25 ± 19.5% with upper 95% confidence value of 65% in all the soils studied. However, within the soil database analysed there were several trends in As bioaccessibility observed. When the soils were separated into different sources of As contamination, As bioaccessibility was found to vary depending on the origin of the As (Table 2). A general trend emerged where mean As bioaccessibility in former railway corridor (herbicide source) soils ≥ dip site soils (insecticide source) ≫ mining soils ≫ geogenic soils. This trend was not surprising since previous studies have identified that As bioaccessibility varies between the matrices studied (Hund-Rinke and Kördel 2003; Pouschat and Zagury 2006). Arsenic bioaccessibility in soils containing geogenic sources of As are considerably lower compared with those contaminated through anthropogenic activities due to mineralogical constraints. Arsenic in anthropogenically contaminated soils (railway and dip site soils) have been shown to be associated with the soil amorphous Fe fraction (McLaren et al. 1998; Smith et al. 2002, 2006), while geogenic As is likely to be associated with poorly soluble Fe–S minerals. Studies have shown that the amorphous Fe phases contribute to the majority of the As solubilised during the SBET bioaccessibility assessment (Tang et al. 2007; Smith et al. 2008) through the desorption and/or solubilisation of Fe (oxy) hydroxide surfaces (Rodriguez et al. 2003). However, it is interesting to note that, while the mean As bioaccessibility soil values for all the contaminated sources were all less than 50%, the upper 95% confidence value was markedly higher for the former railway corridor and dip site soils (Table 2). This was attributable to one or two soils in each data set containing As bioaccessible values >80%. These soils contained low Fe concentrations and, in some cases, elevated pH values, which have been reported to be linked to elevated As bioaccessibility in other studies (Yang et al. 2002; Juhasz et al. 2007b; Subacz et al. 2007).

Table 2 Variation in As bioaccessibility in the <250-μm soil particle size fraction with different sources of As contamination
Full size table

To date, few studies have investigated As distribution across various soil particle sizes in contaminated soils (Bright et al. 2006; Laird et al. 2007). To investigate As distribution in various soil particle fractions, 29 of the 50 soils were randomly selected and separated into different particle sizes. The selected soils consisted of seven geogenic soils, ten former cattle dip site soils, six former railway corridor soils and six mining soils, with total As concentration in the <250-μm particle size ranging from 34 to 11,280 mg As kg−1. The total As concentration showed limited variability across the <250-, <100- and <10-μm particle size fractions in the 29 soils studied (Fig. 1). However, considerable variability was observed between total As concentrations found in the <2.5-μm particle size fraction and the other soil particle fractions examined. Particle size distribution studies for As are limited, but Lombi et al. (2000) and Smith et al. (2003) reported that As concentrations in the <2-μm particle size fraction increase significantly compared with the 2-mm particle fraction. In another study, Cai et al. (2002) reported similar total As concentrations in the <250-μm and the 250- to 750-μm soil particle size fractions. The increase in total As concentrations in the <2.5-μm particle size fraction was accompanied by a corresponding increase in the total Fe concentration in the <2.5-μm particle size fraction (data not shown), which concurs with previous findings of Lombi et al. (2000) and Smith et al. (2003).

Fig. 1
figure 1

As distribution across various particles sizes in 29 contaminated soils

Full size image

It is commonly assumed that As bioaccessibility will increase with decreasing soil particle size (Bright et al. 2006) because of enhanced dissolution as a result of increased surface area of the soil particle fraction. However, there have been limited studies investigating As bioaccessibility across various soil particle sizes (Belluck et al. 2003). In the 29 soils studied, As bioaccessibility increased with decreasing particle size (Table 3) with mean As bioaccessibility increasing from 25 ± 16% in the <250-μm particle fraction, to 33 ± 22% and 42 ± 23% in the 100- and 10-μm particle fractions. The increase in As bioaccessibility was consistent across all the soils studied even though the total As concentrations in these fractions remained similar. The mean and 95% upper confidence level values were similar for the <250-μm particle fractions in the 50 soils and the subset of 29 soils that were evaluated. However, it is notable that the 95% upper confidence level bioaccessibility for As increases markedly to 89% in the <10-μm particle fraction. Although the <10-μm particle fraction only constitute a small proportion of the <250-μm soil mass, this soil particle fraction is important for inhalation exposure, where the 3- to 10-μm particle fraction is intercepted or deposited in the upper respiratory tract and subsequently swallowed and ingested (Bright et al. 2006). In addition, Yamamoto et al. (2006) reported that only soil particles <39 ± 26 μm (mode and standard deviation) adhered to children’s hands after a variety of activities. For young children, the ingestion of household dust may be a major exposure route as a large proportion of their early years is spent indoors in the home environment (Duggan et al. 1985; Lioy et al. 2002; Simon et al. 2007). Studies of Pb exposure in children have identified that there is a strong relationship between Pb-contaminated surfaces in the home and blood Pb levels in children (Hunt et al. 1993; Lanphear and Roghmann 1997; Simon et al. 2007).

Table 3 Variation in As bioaccessibility in various particle size fractions in 29 contaminated soils
Full size table

Therefore, in exposure scenarios where small particles contribute a greater proportion of the bulk soil ingested (e.g. dust deposition in houses), As bioaccessibility may be considerably higher compared with bioaccessibility values derived using the “bulk” 250-μm soil fraction.

Conclusions

The results of this research indicate that As bioaccessibility across a wide range of As-contaminated soil (<250 μm) is generally less than 50%; however, As bioaccessibility may be attributable to the source of contamination. In general, As bioaccessibility in soils contaminated through anthropogenic sources were higher than As derived from geogenic origin. Furthermore, this research showed that As concentrations were uniformly distributed across the 10- to 250-μm soil particle fractions but elevated concentrations were found in the <2.5-μm fraction. In addition, As bioaccessibility within the 10- to 250-μm particle size fractions increased with decreasing soil particle size. Since the bioaccessibility of As varied across soil particle size fractions, these suggest that estimation of human exposure through incidental soil ingestion via bioaccessibility may vary depending on the soil sample’s particle size composition.