Dermally adhered soil: 2. Reconstruction of dry-sieve particle-size distributions from wet-sieve data

Authors


Abstract

In the evaluation of soil particle-size effects on environmental processes, particle-size distributions are measured by either wet or dry sieving. Commonly, size distributions determined by wet and dry sieving differ because some particles disaggregate in water. Whereas the dry-sieve distributions are most relevant to the study of soil adherence to skin, soil can be recovered from skin only by washing with the potential for disaggregation whether or not it is subsequently wet or dry sieved. Thus, the possibility exists that wet-sieving measurements of the particle sizes that adhered to the skin could be skewed toward the smaller fractions. This paper provides a method by which dry-sieve particle-size distributions can be reconstructed from wet-sieve particle-size distributions for the same soil. The approach combines mass balances with a series of experiments in which wet sieving was applied to dry-sieve fractions from the original soil. Unless the soil moisture content is high (i.e., greater than or equal to the water content after equilibration with water-saturated air), only the soil particles of diameters less than about 63 μm adhere to the skin. Because of this, the adhering particle-size distribution calculated using the reconstruction method was not significantly different from the wet-sieving determinations.

INTRODUCTION

Particle-size distributions determined by dry sieving a soil sample are influenced primarily by initial preparation methods (e.g., drying procedures and disaggregation of gravel size aggregates by coarse screening). By contrast, wet sieving provides information on the size distribution of water-stable aggregates or primary particles, depending on whether chemical or mechanical dispersion is employed. Dry- and wet-sieve methods for particle-size analysis of soil are discussed by Bernhardt (1994). Often, wet sieving provides the more useful size information as it is somewhat less dependent on sample preparation methods.

In some situations, particle-size distribution for the dry soil is desired. This is the case for research on dermal exposure to chemically contaminated soil, which requires knowledge of the particle-size distribution of the dry soil that adheres to skin (Choate et al. 2006). In these studies, adhering soil particles are cleaned from the skin by washing with water and then wet sieved to determine size distribution and sometimes properties of adhering soil as a function of particle size. Other authors have dried particles that were washed from the skin and then determined particle size by dry sieving (Holmes et al. 1996; Kissel et al. 1996). However, because particles in many soils disaggregate in the presence of water, the particle-size distributions determined after washing from the skin surface may not represent accurately the particle sizes that actually adhered to the skin. Specifically, particle disaggregation in the wet-sieve measurements could cause the smaller particle sizes to be overrepresented with a compensating underrepresentation of the larger particle sizes. Unlike Holmes et al. (1996) and Kissel et al. (1996), Choate et al. (2006) chose to wet sieve the particles washed from the skin because drying could cause particle aggregation that was different from aggregation in the particles that adhered to the skin.

The goal of this study was to develop a method for reconstructing dry particle-size distributions from the wet-sieve analyses. The approach we used was to conduct a series of experiments in which wet sieving was applied to dry-sieve fractions from the original soil. These data were then used in a series of mass balance equations to reconstruct dry-sieve size distributions from wet-sieve data. The procedure was tested using wet-sieve and dry-sieve results for the same 2 soils at 3 different moisture contents as studied by Choate et al. (2006). Finally, the reconstruction method was applied to the results from the dermal adherence experiments in Choate et al. (2006) to estimate the size distribution of the dry particles that actually adhered to the skin.

MATERIALS AND METHODS

Soil samples

Two soils were investigated: A clay loam from Colorado State University Agricultural Research Station (Fort Collins, CO, USA), referred to as the CSU soil, and a silty-clay loam from the Iowa State University Field Extension Laboratory (Ames, IA, USA), the ISU soil. Particle-size distributions and organic carbon distributions for these soils are described in a companion paper reporting adherence of soil particles on skin (Choate et al. 2006). Other characteristics are described in Choate (2002). Soil moisture of some samples was varied by placing soil samples in closed containers with a relative humidity of 100% (high), 45% (medium), or 15% (low) at 19 to 21 °C as described in Choate et al. (2006).

The <2-mm soil fractions, which were studied in the dermal adherence experiments and also here, were obtained by coarse mechanical shaking, which partially broke up very large (>2 mm) soil aggregates. Subsamples for sieve analysis were obtained using a splitter or by coning and quartering. Particle-size distributions of the <2-mm fractions of the CSU and ISU soil were characterized by dry and wet sieving as described in the companion paper (Choate et al. 2006). All masses in the sieve fractions are reported as a percent of the total dry mass of the sample.

Table Table 1.. Example calculation for estimating the dry-sieve fractions Dm from wet-sieve fractions Wm
Sieve fraction
Nr (m)Size (μm)F3,k (%)F2,k (%)Wm (%)z3,k (%)Z2,k (%)z1,k (%)Dm (%)
32,000–25030015.815.8 × 30/30 = 1 5.8(72.7–32.1) × 0/93 = 0(11.5–4.7–3.1) × 0/100 = 052.6
2250–38619372.715.8 × 61/30 = 32.1(72.7–32.1) × 93/93 = 40.6(11.5–4.7–3.1) × 0/100 = 043.7
1<389711.515.8 × 9/30 = 4.7(72.7–32.1) × 7/93 = 3.1(11.5–4.7–3.1) × 100/100 = 3.73.7
Sum 10010010052.643.73.7100

Preparation of individual dry size fractions

The strategy was to measure the extent to which each dry-sieve fraction of a given soil disaggregates into smaller sieve fractions during wet sieving. Therefore, a sizable quantity of each dry-sieve fraction was needed for subsequent wet-sieve analysis. This was prepared by dry sieving subsamples of approximately 17 kg of the CSU and 20 kg of the ISU bulk (<2 mm) soils. The moisture content of these soil samples was not controlled but was usually similar to soil samples equilibrated at 45% relative humidity. Standard 20-cm-diameter stainless-steel testing sieves (ASTM E-11 specification) were stacked as follows: 500 μm (No. 35), 250 μm (No. 60), 125 μm (No. 120), 63 μm (No. 230), 38 μm (No. 400), and 25 μm (No. 500). An approximately 100-g soil sample was placed in the sieve stack and shaken on an RX-29 ROTAP vibratory shaker for 10–15 min. This procedure was repeated for another 100 g of soil. Following the 2nd shaking, the sieves were removed and the fractions collected and weighed. This was continued until approximately 1 kg of the CSU soil and 3 kg of the ISU soil had been dry sieved. The sieves were then cleaned with a detergent solution, Liqui-Nox (Alconox), and the 125-, 63-, 38-, and 25-μm sieves were sonicated to remove lodged particles. When the sieves were dry, the procedure described previously was repeated until the entire mass of each soil was sieved.

Once all the subsampled bulk soil had been sieved, the 3 sieve fractions with particle sizes between 25 and 125 μm were resieved to ensure complete passage of fine particles. Approximately 50 g (125–63 μm) or 25 g (63–38 μm, 38–25 μm) of soil were placed in the appropriate sieve stack and shaken on the ROTAP for 10 min. Another 50 or 25 g of soil were added to the stack and shaken for another 10 min. Following the 2nd shaking, the sieves were removed and the fractions weighed and collected. This procedure was continued until the fraction had been totally resieved.

Wet sieving the individual dry particle-size fractions

Particle-size distributions of the individual dry-sieve fractions were obtained by wet sieving 4- to 12-g subsamples. Approximately 100 mL of distilled water were added to each sample and allowed to stand for approximately 2 h. After this, the slurry was poured through a 7.6-cm-diameter sieve stack using the appropriate sieves (i.e., sieve sizes less than or equal to the size of the dry-sieve fraction being wet sieved). The particles on each sieve were rinsed with distilled water to move smaller particles through to the next sieve. Once the smaller particles were transferred, the retained mass on each sieve was washed into a preweighed Petri dish and dried at 50 °C. The drying temperature of 50 °C was chosen for the wet-sieve fractions to obtain consistent moisture content without altering the organic matter.

Particles that passed through the 25-μm sieve were treated slightly differently. This fraction was transferred to a plastic container and allowed to settle until little or no turbidity was observed in the supernatant. The water was siphoned off, and the settled soil and remaining entrapped water were poured into a preweighed Petri dish and allowed to air-dry. The plastic container was rinsed 3 times with decantate, and the rinse was placed in a beaker and allowed to settle. After settling, the water was siphoned from the beaker, and the settled soil was poured into the appropriate Petri dish and dried at 50 °C.

Table Table 2.. Mass fraction from dry-sieve fraction number m in each wet-sieve fraction k (Fm,k) measured for the Colorado State University (Fort Collins, CO, USA) soil
Sieve fraction
Nr (m)Size (μm)F7,m (%)F6,m (%)F5,m (%)F4,m (%)F3,m (%)F2,m (%)
72,000–5006.15 ± 1.3400000
6500–25019.38 ± 1.7934.90 ± 3.590000
5250–12528.78 ± 0.6524.95 ± 3.4867.36 ± 1.64000
4125–6329.13 ± 1.6424.91 ± 0.2918.60 ± 1.3091.46 ± 0.7400
363–387.46 ± 1.297.30 ± 0.346.57 ± 0.423.41 ± 0.4392.44 ±1.180
238–252.48 ± 0.351.97 ± 0.111.99 ± 0.161.15 ± 0.111.22 ± 0.3888.58 ± 0.73
1<256.62 ± 0.515.95 ± 0.165.49 ± 0.243.98 ± 0.216.34 ± 0.8111.42 ± 0.73
Figure Figure 1..

The percent mass of particles that would be in each wet-sieve fraction from a constructed soil containing equal amounts of the dry-sieve fractions from either the Colorado State University (CSU; Fort Collins, CO, USA) or the Iowa State University (ISU; Ames, IA, USA) soil. The dashed line indicates the percent mass that would be in each wet-sieve fraction if particle disaggregation does not occur.

All the dried wet-sieve fractions were weighed and the results used to calculate Fm,k (see next section), defined as the fraction of the mass from dry-sieve fraction number m that is in each wet-sieve fraction number k. For a soil divided into N sieve fractions, the smallest sieve fraction is number 1, and the largest sieve fraction is N. We assume that particle aggregation (i.e., particle growth) during wet sieving is minimal. Thus, each dry-sieve fraction number m is wet sieved into m sieve fractions of equal or lesser size than fraction number m. Since the smallest dry-sieve fraction cannot be divided into more sieve fractions, F1,1 is by definition always equal to 1. Thus, only dry-sieve fractions 2 through N are wet sieved. Triplicate samples of the dry-sieve fractions were analyzed, and the mean values and standard deviations for Fm, k were used in the reconstruction calculations.

RECONSTRUCTION ALGORITHM

An algorithm was developed to reconstruct the particle-size distribution of the original dry soil from the wet-sieve particle-size fractions, Wm, where m is the number of the sieve fraction with 1 being the smallest fraction and N being the largest. The particles collected in a wet-sieve fraction number m come from either 1) dry particles of the same size that did not disaggregate or 2) dry-sieve fraction numbers larger than m that did disaggregate. If no particles in dry-sieve fraction number m disaggregated, then Fm,m = 1 and Fm,k. = 0 for k = 1 to m −1. We assume that particles do not aggregate into larger particles, and therefore Fm,k = 0 for k > m. Consequently,

equation image((1))

and F1,1 = 1. Thus, the mass fraction of particles, Dm, that would be part of the mth dry-sieve fraction is the sum of all particles in the wet-sieve fractions that originated in dry-sieve fraction m. That is,

Table Table 3.. Mass fraction from dry-sieve fraction number m in each wet-sieve fraction k (Fm,k) measured for the Iowa State University (Ames, IA, USA) soil
Sieve fraction
Nr (m)Size (μm)F7,m (%)F6,m (%)F5,m (%)F4,m (%)F3,m (%)F2,m (%)
72,000–50060.14 ± 3.8000000
6500–25019.10 ± 3.1184.39 ± 2.840000
5250–12510.47 ± 0.518.70 ± 1.8690.54 ± 0.31000
4125–634.97 ± 0.173.60 ± 0.544.66 ± 0.1497.01 ± 0.0600
363–382.20 ± 0.041.36 ± 0.331.90 ± 0.070.98 ± 0.0298.08 ± 0.100
238–250.84 ± 0.020.63 ± 0.110.95 ± 0.090.59 ± 0.020.43 ± 0.0598.60 ± 0.27
1<252.27 ± 0.061.32 ± 0.161.95 ± 0.021.49 ± 0.031.49 ± 0.061.40 ± 0.27
Figure Figure 2..

Comparison of the dry-sieve and the reconstructed dry-sieve size distribution for the Colorado State University (CSU; Fort Collins, CO, USA) soil at different moisture contents.

equation image((2))

in which zm,k is the mass of soil in a given wet-sieve fraction number k that came from dry-sieve fraction m. Values of zm,k are calculated by applying a mass balance to each of the wet-sieve fractions and by knowing FM,K from separate experiments in which the dry-sieve fractions of the bulk soil have been wet sieved. In this analysis, we assume that particle disaggregation of the dry-sieve fractions from the bulk soil is the same as occurred to produce the wet-sieve mass fractions WM.

The procedure for calculating zm,k from WM begins with the largest sieve fraction (i.e., number N) as described by the following equation:

equation image((3))
Figure Figure 3..

Comparison of the dry-sieve and the reconstructed dry-sieve size distribution for the Iowa State University (ISU; Ames, IA, USA) soil at different moisture contents.

All the particles in the largest wet-sieve fraction (number N) came from the dry-sieve fraction N, and thus zN,N = WN Nonzero values of zN,k indicate that some portion of the particles in dry-sieve fraction number N should have disaggregated into wet-sieve fraction number k. This means that only a portion of the mass in the next largest wet-sieve fraction (i.e., WN-1) originated from particles in dry-sieve fractions smaller than N. Thus, the mass in sieve fraction N −1 is reduced by an amount equal to WN-1 -zN,N-1 The algorithm for calculating zm,k, therefore, begins with the largest sieve size and works down to the smallest sieve size, accounting for the mass of larger particles that have disaggregated into the smaller size sieve fractions. The procedure is summarized by the following equation:

equation image((4))

which uses values of zN,k calculated according to Equation 3.

Figure Figure 4..

Particle-size distributions of the wet-sieve adhered soil compared with the reconstruction estimate of the dry-sieve adhered soil for the Colorado State University (CSU; Fort Collins, CO, USA) soil at different moisture contents.

Use of the reconstruction algorithm is illustrated in Table 1 with an example of a hypothetical soil divided into 3 sieve fractions. In this example, a significant amount of the largest dry-sieve fraction disaggregates and contributes to the smaller wet-sieve fractions. The estimated amount of the dry-sieve fraction number 3 was 52.6%, whereas the wet-sieve fraction was only 15.8%.

RESULTS AND DISCUSSION

The particle-size distributions that resulted from wet sieving the dry fractions of the CSU and ISU soils are listed in Tables 2 and 3, respectively. The relative amount of mass in the wet-sieve size fractions arising from the same dry-sieve size fractions is greater for the ISU soil compared with the CSU soil. These results show that when wet sieved, the original fractions hold together more for the ISU soil than for the CSU soil. Of the 2 soils studied here, the ISU soil is more water stable. For the CSU soil, the contribution of disaggregated particles is relatively small for size fractions less than 63 μm and greater than 25 μm, indicating that the sizes of most disaggregated particles are either greater than 63 μm or less than 25 μm.

Figure Figure 5..

Particle-size distributions of the wet-sieve adhered soil compared with the reconstruction estimate of the dry-sieve adhered soil for the Iowa State University (ISU; Ames, IA, USA) soil at different moisture contents.

Figure 1 shows the expected particle-size distribution of the wet-sieve fractions of the CSU and ISU soils if the 7 dry fractions were mixed together in equal amounts. That is, each dry-sieve fraction of these new soils contains 14.3% (one-seventh) of the total mass. Figure 1 also shows the original dry-sieve source of disaggregated particles. For example, the 250- to 125-μm wet-sieve fraction of the CSU soil includes significant amounts of disaggregated particles from the 2 larger dry-sieve fractions. If particle disaggregation does not occur, then each wet-sieve fraction would contain 14.3% of the total mass and only particles of the same size.

For the ISU soil, only the largest size fraction disaggregates significantly. Furthermore, at least 80% of the mass of each wet-sieve fraction of the ISU soil is from dry particles of the same size fraction. In contrast, for only the 63- to 38-μm wet-sieve fraction of the CSU soil does at least 80% of the mass come from dry particles of the same dry-sieve fraction. For the CSU soils, particle disaggregation is significant for the sieve sizes larger than 125 μm, whereas particles smaller than this disaggregate only slightly. Disaggregated larger particles contribute almost 30% of the <25 μm wet-sieve fraction of the CSU soil compared with only 10% for the ISU soil. After wet sieving, only 6% of the largest dry-sieve fraction of the CSU soil remains compared to 60% for the ISU soil. Disaggregated larger particles account for 45% of the mass in the 250- to 125-μm and 125- to 63-μm wet-sieve fractions of the CSU soil.

The ability of the reconstruction algorithm, Equations 2 to 4, to estimate dry-sieve size distributions from wet-sieve data was tested using wet-sieve and dry-sieve results of the bulk CSU and ISU soils at 3 different moisture contents from Choate et al. (2006). The reconstructed dry-sieve distributions calculated from the wet-sieve data are compared with the actual dry-sieve data in Figures 2 and 3 for the CSU and ISU soils, respectively. The error bars in Figures 2 and 3 and also Figures 4 and 5 represent ±1 standard deviation. They were computed using standard statistical methods of propagation of error (Skoog et al. 1992) and included assessment of the standard deviations of sieve fraction measurements incorporated into the calculations. Some error bars are too small to be observed at these scales.

As shown in Figures 2 and 3, there is good agreement for the CSU soil at the medium and high moisture contents, but for the low moisture content the mass of dry particles in the 2,000- to 500-μm and the 500- to 250-μm fractions are underestimated by the reconstruction algorithm. Although the very small amounts of mass in the smaller size fractions of the ISU soil make comparison difficult, the ISU soil has good agreement at the medium moisture content (Figure 3). However, the technique overestimates the mass of dry particles in the 2,000- to 500-μm fraction for the high-moisture-content samples and underestimates the mass of dry particles in the 2,000- to 500-μm fraction for the low-moisture-content samples. In the companion paper (Choate et al. 2006), there was evidence that the largest particles of the high-moisture-content ISU soil were less prone to disaggregation than when they were in low- and medium-moisture-content soils. This observation is consistent with the reconstruction estimate of the largest dry-sieve fraction for the medium- and high-moisture ISU soils.

The reconstruction method was then applied to the adhered soil fractions, from the dermal adherence experiments in Choate et al. (2006), to estimate the size distribution of the dry particles that adhered to the skin. Figures 4 and 5 show a comparison of the size distributions of the wet-sieve adhered particles from Choate et al. (2006) and the reconstructed estimate of the adhered dry particle-size distribution for the CSU and ISU soils, respectively. Good agreement exists for both soils and all 3 moisture contents. This shows that the particles that adhered are in the water-stable size ranges. Therefore, the particle sizes that adhered dry are similar to the particles collected from wet sieving the adhered soil. This is consistent with the fact that smaller sizes preferentially adhere, and, as shown in Figure 1, these smaller dry size fractions are more stable in water. Thus, wet sieving of the adhered soil did not noticeably change the particle-size distribution of the particles that initially adhered.

The wet-sieve size distributions of the adhered soil and the reconstructed estimates of the size distributions of the dry particles that adhered were similar. Therefore, the adhered soil particle-size distributions, determined for these 2 soils by wet sieving, are in reality essentially the same as those of the dry particles that originally adhered to the skin. While this is likely to be true for many soils, investigators studying particle sizes adhering to skin should confirm that the size distributions of the dry-sieve fractions corresponding with the adhering particle sizes change insignificantly when wet sieved. In studies of soils that are saturated or more with water, a complete reconstruction analysis as described in this paper may be required because larger particles, which are prone to disaggregation, can adhere to skin from these soils.

CONCLUSIONS

The method described in this paper provides a procedure for estimating the particle-size distributions in dry-sieve soils of medium moisture content from data obtained from wet sieving of the same soils. This is especially true for the smaller size particles that adhere to human skin, so the method is especially relevant for studies of dermal adsorption from dermal exposure to contaminated soils. For larger particles, particularly those in very dry or very wet soils, the method should be applied with caution, as statistically significant errors can occur.

Acknowledgements

This work was supported in part by the US Environmental Protection Agency (Assistance Agreements CR817451, CR822757, R825398, and R826651), by the US Air Force Office of Scientific Research under agreement F49620–95–1–0021, and by the National Institute of Environmental Health Sciences under grant R01-ES06825. LM Choate acknowledges support from the Department of Chemistry and Geochemistry and the Edna Baily Sussman Fellowship.

Disclaimer—Any use of trade, product, or firm names is for descriptive purposes only and does not constitute endorsement by the US government.

Ancillary