- Email: [email protected]
Measuring the saturation limit of low-volatility organic compounds in soils: Implications for estimates of dermal absorption
Science of the Total Environment 408 (2010) 6100–6107
Contents lists available at ScienceDirect
Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v
Measuring the saturation limit of low-volatility organic compounds in soils: Implications for estimates of dermal absorption Sandrine E. Déglin a,1, Donald L. Macalady a, Annette L. Bunge b,⁎ a b
Chemistry Department, Colorado School of Mines, Golden, CO 80401, USA Chemical Engineering, Colorado School of Mines, Golden, CO 80401, USA
a r t i c l e
i n f o
Article history: Received 8 February 2010 Received in revised form 21 August 2010 Accepted 26 August 2010 Keywords: Soil Contaminated soil Saturation limit Dermal absorption Skin Risk assessment
a b s t r a c t Estimating dermal absorption from contaminated soils typically requires extrapolations from measurements obtained on soils artificially contaminated at much larger concentrations. Such extrapolations should be constrained by the fact that maximum absorption will occur for the largest possible concentration of chemical on the soil without neat chemical being present; i.e., at the soil saturation limit (Ssoil). Saturation limits of two low-volatility model compounds (4-cyanophenol and methyl paraben) were determined on the 38–63 μm sieve fraction of four soils with different fractions of organic carbon (foc = 0.015–0.45) and specific surface areas (σsoil = 4–34 m2 g− 1) using two methods: equilibrium uptake into silicone rubber membranes and differential scanning calorimetry. Except for Pahokee peat, which had the largest foc, a model assuming contributions from both surface adsorption and organic carbon absorption provided excellent predictions of Ssoil. In all soils, the surface saturation concentration of both chemicals was estimated at 2.2 mg m− 2. The saturation concentration of 4-cyanophenol in the soil organic carbon was 1.7-fold higher than methyl paraben, which is consistent with the estimated solubility limits of these two chemicals in octanol. © 2010 Elsevier B.V. All rights reserved.
1. Introduction In the United States, remediation decisions on contaminated sites are based on assessment of the potential health risks. These assessments often require characterization of the risks associated with exposure to contaminated soils, including skin contact (Kissel et al., 1996). A recent critical review of the literature describing dermal absorption from contaminated soils identified experimental measurements for 33 organic and 4 inorganic compounds (Kissel et al., 2007; Spalt et al., 2009). Although the approaches employed by various research groups have varied widely, the chemical contaminant was commonly added to the soil in a volatile solvent that was removed by evaporation (Spalt et al., 2009). It is also common for dermal absorption to be measured from one chemical concentration on soil that is applied at a single mass per area of skin (i.e., the soil load). Nearly always, the investigated contaminant concentrations and soil loads substantially exceed those that would occur in actual exposures. As a result, risk assessors are faced with decisions about how to extrapolate experimental measurements to a different soil contam-
⁎ Corresponding author. Colorado School of Mines, Chemical Engineering Department, 1500 Illinois Street, Golden, CO 80401-1887. Tel.: +1 303 273 3722; fax: +1 303 273 3730. E-mail addresses: [email protected] (S.E. Déglin), [email protected] (D.L. Macalady), [email protected] (A.L. Bunge). 1 Present address: Exponent Inc., Washington, DC 20036, USA. 0048-9697/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2010.08.042
inated at a different concentration with a different amount in contact with the skin. It is well established that in aqueous suspensions, the equilibrium concentrations of organic compounds in soil are proportional to their aqueous concentrations and the organic matter content of the soil, usually reported as the mass fraction of organic carbon per mass of soil (foc) (Lyman et al., 1990; Boethling and Mackay, 2000). Based on this, it has been proposed that dermal absorption should be inversely proportional to foc (US EPA, 1992; Bunge and Parks, 1998; Spalt et al., 2009); i.e., an increase in the soil sorption capacity would cause the dermal bioavailability to decrease. Of the limited data available for testing this hypothesis, generally an increase in foc produced a less than proportional decrease in absorbed dose (US EPA, 1991, 1992; Duff and Kissel, 1996). In part, this may arise from differences in the soils investigated in dermal absorption studies compared with those equilibrated with organic compounds in water. However, the most obvious difference is that soils in dermal absorption studies usually contain no liquid water. Also, the amount of chemical applied to soils using a volatile solvent can exceed the capacity of the soil to hold chemical, as determined by equilibrating the soil with an aqueous solution. All soils have a limited capacity to hold a chemical in a form that is thermodynamically distinct from the pure form of the chemical. If a chemical is present at a level exceeding the capacity of a given matrix (be it soil, some other solid, a gas, water, or some other liquid), then,
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
at equilibrium, the pure (or neat) chemical is present as a separate phase. For soils equilibrated in water saturated with an organic compound, the soil saturation process mainly involves partitioning into the soil organic matter (as long as foc is not extremely low). For air-dried soils, chemical uptake most probably depends on surface adsorption as well as partitioning into the soil organic matter (Goss et al., 2004). Therefore, chemical saturation of air-dried soils should vary with surface area in addition to the amount of soil organic matter as described by Eq. (1): Ssoil = foc Soc + σsoil Mσ
ð1Þ
which is an extension of the soil-air partition equation proposed by Goss et al. (2004). In this equation, Ssoil is the soil saturation concentration in units of mass of chemical per mass of clean soil, Soc is the saturation concentration of chemical in the soil organic matter (per mass of soil organic matter), σsoil is the specific surface area expressed in area per mass of soil, and Mσ is the mass of chemical adsorbed per area of the saturated soil surface area. Because surface adsorption of water reduces the surface area available for sorption of organic compounds, Mσ will decrease as relative humidity increases (Shih and Wu, 2005). As a result, surface adsorption of organic contaminants will be smaller when the relative humidity is higher (Goss et al., 2004). The rate and amount of dermal absorption of a given chemical is proportional to its thermodynamic driving force, which is maximized when the skin surface is exposed to a saturated medium, whether it is a soil or liquid solution. As a result, for a given soil, the amount of chemical that will absorb into the skin is expected to increase proportionally with the concentration of chemical on the soil (reported per mass of clean soil, Csoil) as long as Csoil is less than Ssoil. Increasing the soil concentration above Ssoil will not increase the thermodynamic driving force and dermal absorption is not expected to exceed that which occurs when Csoil = Ssoil provided the barrier function of the skin is unaltered. Moreover, for a given compound, Ssoil represents an intrinsic characteristic of each soil and therefore, in comparing two soils, the thermodynamic driving force for dermal absorption from each soil should be related to the ratio of Csoil to Ssoil. The purpose of this study was to measure saturation concentrations in soils contaminated using volatile solvents as vehicles, as has been done in many dermal absorption studies. Saturation concentrations of two model chemicals, methyl paraben and 4-cyanophenol, which are both solids at skin temperature (32 °C), were determined for four soils with widely different fractions of organic matter. Two methods were used. In the first method, soil saturation was determined by measuring the amount of chemical uptake into pieces of silicone rubber membrane (SRM) that were equilibrated with the contaminated soil. In this method, equilibrium uptake into the SRM increases with soil concentration to a maximum value that remains constant for all soil concentrations above the soil saturation limit. The soil saturation limit is then the smallest soil concentration of the contaminant that leads to the maximum uptake into the SRM sorbent. In the second method, using differential scanning calorimetry (DSC), the amount of pure chemical present as a separate solid phase within or on the soil sample is quantified by the measurement of its enthalpy of melting. This technique has been used for measuring the solubility of drugs in adhesives (Li et al., 2002) as well as polymers (Theeuwes et al., 1974). In the present study, soil saturation measurements performed using DSC were compared with those from SRM uptake. The results from both methods indicate that the saturation values of the chosen test chemicals for the soils and conditions studied here cannot be adequately estimated using the amount of soil organic matter alone. As expected from prior studies of soil–air uptake, surface area is an important factor.
6101
2. Materials and methods 2.1. Chemicals Reagent grade 4-cyanophenol (CP; molecular weight 119; logarithm of the octanol-water partition coefficient, logKow, is 1.60 (PhysProp Database)) and methyl paraben (MP; molecular weight 152; logKow = 1.96 (PhysProp Database)) were both purchased from Sigma-Aldrich (Milwaukee, WI). Both CP and MP are characterized by low volatility (estimated vapor pressures are 0.89 and 0.032 Pa, respectively (PhysProp Database)). Liquid chromatography grade acetonitrile and acetone were from Mallinckrodt Chemicals (Phillipsburg, NJ). 2.2. Soils The saturation limits of MP and CP were studied in four soils, identified as CSU, ISU, TG and PP, which differed in the amount and nature of their organic matter. The CSU (a clay loam) and ISU (a silty clay loam) soils were collected from the Colorado State University (Fort Collins, CO) and Iowa State University (Ames, IA), respectively. The collection, preparation and characterization of both of these soils are described by Choate et al. (Choate, 2002; Choate et al., 2006). The TG soil is a commercial top soil purchased from Timberline Gardens (Wheat Ridge, CO). Pahokee peat (PP) was originally collected on behalf of the International Humic Substances Society (IHSS) and was provided by Professor Patrick MacCarthy (Colorado School of Mines, Golden, CO). All of the experiments described in this study were performed on the 38–63 μm dry sieve fraction of each soil, which was used in dermal absorption studies performed with the CSU and ISU soils (Deglin, 2007). In a study of the CSU and ISU soils, which contained a wide distribution of particle sizes b250 μm, Choate et al. demonstrated that the majority of particles adhering to the skin have diameters b63 μm (Choate et al., 2006). The lower end of the fraction (38 μm) was chosen to limit and control the extent of the particle size distribution. For the CSU and ISU soils, the 38–63 μm sieve fraction was prepared by sieving the b250 μm fraction from the previous studies by Choate et al. (Choate, 2002; Choate et al., 2006) with stainless steel sieves (W.S. Tyler, Mentor, OH) stacked as follows: 125 μm (No. 120), 63 μm (No. 230), and 38 μm (No. 400). The PP soil was sieved as for the CSU and ISU soils with two sieves added, i.e., 550 μm (No. 35) and 250 μm (No. 60). Two more sieves, the 4 mm (No. 5) and 2 mm (No. 10), were included in the stack used to prepare the TG soil. Using procedures described previously (Choate, 2002), the mass fractions of organic carbon in the 38–63 μm sieve fractions were determined to be 0.015, 0.038, 0.063 and 0.45 for CSU, ISU, TG and PP, respectively. In a study by Choate et al. (Choate, 2002; Choate et al., 2006) performed on different sub-samples of the same size fractions of the CSU and ISU soils, the mass fractions of organic carbon were measured to be 0.011 and 0.031, respectively. The specific surface area (σsoil) of the CSU, TG and PP soil samples were measured by the BET method, using N2 adsorption and calculated from the desorption data (Chiou et al., 1990). The value of σsoil for the 38–63 μm fraction of the ISU soil was obtained from Choate (Choate, 2002; Choate et al., 2006). The MP saturation limit was studied on all four soils, whereas the CP saturation limit was studied on the CSU and ISU soils only. 2.3. Chemical contamination of soils The contamination procedures were similar for all soils except for MP on the CSU soil. The mass of chemical necessary to achieve the desired contaminant concentration was dissolved into the minimum necessary volume of acetone or acetonitrile, usually about 1 to 2 mL. Acetonitrile was only used for MP contamination of the TG soil. A portion of the solution (typically, about 0.5 mL) was added to 1 g of
6102
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
soil in a glass vial, which was briefly shaken to optimize the contact between the soil and the solution. The solvent was allowed to evaporate at room temperature for 12 to 24 h, and then the vial containing the soil was shaken to break up the clumps. This process was repeated until all of the solvent containing the chemical of interest had been added to the soil. This procedure of adding the solvent in increments was followed to reduce the amount of chemical that was lost to the walls of the vial that contained the soil. After the final increment of the solvent was allowed to evaporate for at least 24 h, the soil was again shaken. Large particle aggregates that were sometimes present in the larger concentration samples (> about 100 mg/g of clean soil) were broken up by grinding with a mortar and pestle. The soils were allowed to incubate at room temperature without agitation for at least 48 h before being sampled to determine the actual concentration, and usually for 7 to 10 days before being used to assess soil saturation. For MP contamination of the CSU soil, MP was dissolved into the minimal volume of acetonitrile and then delivered all at once to 5 g of soil. After the solvent was allowed to evaporate at room temperature for at least 24 h, the soil was broken up by shaking the vial and, when necessary, large aggregates present at the larger concentrations were ground with a mortar and pestle. The contaminated soil samples were then mixed for 24 h on a wrist-action shaker (Rotoshake Genie from Scientific Industries Bohemia, NY), after which they were incubated at room temperature for 2 to 3 days without any further agitation before starting the saturation measurements. The specific concentrations that were prepared for MP and CP in each soil varied. Determinations of soil saturation by SRM uptake required concentrations that are both smaller and larger than saturation. However, soil saturation measured by the DSC method only required concentrations that exceed saturation. Saturation limits of MP were determined by DSC in all four soils and by SRM uptake in the CSU and ISU soils. Soil saturation limits of CP were measured by DSC alone in only the CSU and ISU soils. The DSC technique was also used to measure the saturation limit of CP in the less than 250 μm fraction of the CSU and ISU soils that were contaminated 5.5 years earlier using acetone, which was removed by evaporation (unpublished report, C-P Chen, CSM, Golden, CO, 2002). These samples are designated as aged CSU and aged ISU soils. Nominal concentrations of MP were 2, 8, 12, 30, 60, 100, 185, and 320 mg MP (g clean soil)− 1 for the ISU soil and 20, 30, 40, 100, 200 and 250 mg MP (g clean soil)− 1 for the CSU soil. Concentrations of MP were 65, 450 and 550 mg (g clean soil)−1 for the TG soil and 70 and 100 mg (g soil)− 1 for the PP soil. The nominal CP concentrations were 80, 190 and 260 mg (g soil)− 1 for the ISU soil and 100, 140, 200 and 250 mg (g soil)− 1 for the CSU soil. The actual concentrations of chemical in each soil were measured by extracting 15 to 20 mg of soil with 10 mL acetonitrile and analyzing the extract by HPLC using the method described below. Although relative humidity was not controlled in this study, it is relatively constant at approximately 50% in the laboratory where these experiments were conducted. This estimate is also consistent with determinations of moisture content for air-dried and controlled humidity samples of the CSU and ISU soils from an earlier study (Choate et al., 2006). In the following discussion, soil samples contaminated with different amounts of either CP or MP are designated by the soil acronym (CSU, ISU, TG or PP) followed by the chemical (MP or CP) and the nominal contaminant concentration given in units of mg of chemical per gram of clean soil. For example, the soil sample identified as CSU-CP100 is the CSU soil contaminated with approximately 100 mg of CP per gram of clean soil. 2.4. Equilibrium uptake experiments Saturation limits of MP in the CSU and ISU soils were evaluated by measuring the equilibrium uptake into SRM. Pieces (about 1 or
2.5 cm2) of SRM sheets (88 ± 6 μm thick) from Membrane Products Corporation (Albany, NY) were weighed and then equilibrated at room temperature (approximately 23 °C) with neat MP powder as well as with CSU and ISU soils with different values of soil concentration. After a waiting period of 18 days for the CSU soil and 21 or 35 days for the ISU soil, each SRM piece was extracted in 5 mL of acetonitrile, and the extract was analyzed by HPLC. Preliminary tests confirmed that all chemicals were recovered from the SRM using a single extraction. If the SRM has direct contact with the soil or the neat chemical powder, it is difficult to insure that no particles are left on the SRM surface after the soil or chemical have been removed. To avoid this problem, the SRM samples were equilibrated with the soils or powders using an indirect method of contact. Specifically, one SRM piece was sandwiched between a second piece of SRM and aluminum foil. This membrane-foil sandwich was mounted into a Franz-type vertical, static diffusion cell (9 mm, jacketed cells) from Permegear (Bethlehem, PA) with the aluminum foil facing down towards the receptor chamber, which was left empty. A large excess of contaminated soil or neat chemical powder was then placed on top of the upper SRM piece. The aluminum foil provided a barrier for chemical loss from the bottom side of the membrane and the top piece of SRM allowed chemical transfer to the second SRM piece, while preventing soil from adhering. After carefully removing the top protective SRM piece, chemical uptake into the protected second piece of SRM was measured. As many as seven of these systems could be equilibrating at one time and therefore, for a particular soil the whole range of concentrations could be tested in a single experiment. The uptake results from these experiments were compared to SRM uptake when the same membranes were equilibrated for 5 days at room temperature with a saturated aqueous solution containing excess solid MP. In this case, the SRM pieces were protected from the chemical particles by placing the membranes inside dialysis bags (Spectrapore Membrane Tubing, 25 mm, 12,000–14,000 molecular weight cut-off). 2.5. High performance liquid chromatography (HPLC) analysis Solutions containing MP or CP were analyzed using a Hewlett Packard 1100 series modular HPLC system equipped with a Zorbax Eclipse XDB-C18 analytical column (Hewlett Packard, 4.6 mm × 25 cm) and a guard column (4.6 mm × 12.5 mm) with the same packing. The injection volume was 10 μL and the mobile phase was 70% HPLC grade acetonitrile and 30% de-ionized water pumped at 1 mL min− 1. Retention times for CP and MP were about 3.1 and 2.9 min, respectively. 2.6. Soil saturation measurements by DSC The method involves the measurement of the enthalpy of melting of pure chemical present within or on the sample, in this case the soil. The approach relies on two assumptions: (1) only the neat chemical contributes to the enthalpy change, and (2) the amount of chemical dissolved into the soil matrix does not change for soil concentrations exceeding the saturation limit. Based on these assumptions, the applicable mass and energy balances are expressed by Eqs. (2) and (3): Mtot = Msorb + Mneat
ð2Þ
ΔHobs ðMsoil + Mtot Þ = ΔHfus Mneat
ð3Þ
For a given mass of clean soil (Msoil), Mtot is the total mass of chemical distributed on the soil, Msorb is the mass of chemical dissolved into or adsorbed within the soil, Mneat is the mass of chemical present as a separate phase. In the energy balance, ΔHobs is the observed enthalpy of melting of the chemical in Joules per mass of
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
ΔHobs = ΔHfus
M Mtot 1− sorb Mtot Msoil + Mtot
ð4Þ
which can be written in terms of the mass fraction of all chemical in the contaminated soil (i.e., Ĉsoil = Mtot / (Msoil + Mtot)) and the mass fraction of chemical at the saturation limit in a mass of contaminated soil (i.e., Ŝsoil = Msorb /(Msoil + Msorb)) as follows (Li et al., 2002): ΔH obs + Sˆsoil Cˆ soil = 1− Sˆsoil ΔHfus
ð5Þ
Eq. (5) was derived by relating the ratio of Msorb to Mtot in Eq. (4) to the contaminant mass fractions, expressed as a mass of chemical per mass of clean soil, at concentrations above (Csoil) and at soil saturation (Ssoil); i.e., Msorb / Mtot = Ssoil / Csoil. Finally, from the mass balance of contaminant on the soil, Csoil and Ssoil were related to Ĉsoil and Ŝsoil as given by Eqs. (6) and (7):
Csoil =
Cˆ soil Mtot = Msoil 1−Cˆ soil
ð6Þ
Ssoil =
Msorb Sˆsoil = Msoil 1− Sˆsoil
ð7Þ
According to Eq. (5), the observed enthalpy and the chemical concentration on the soil are linearly correlated. Therefore, on a plot presenting Ĉsoil as a function of ΔHobs, the intercept of the line at ΔHobs equal to zero corresponds to Ŝsoil and the slope is the product of (1 − Ŝsoil) and the reciprocal of the enthalpy of fusion of the pure chemical (ΔHfus). The DSC analyses were performed on pure MP and CP powder as well as clean and contaminated soils using a Pyris 6 differential calorimeter from Perkin-Elmer (Waltham, MA). Samples of pure chemical or soil (approximately 5 or 10–20 mg, respectively) were weighed into an aluminum DSC pan with crimped cover and analyzed by increasing the temperature from 90 to 135 °C at a rate of 10 °C per minute. Prior to the DSC analysis, contaminated soil samples were ground with a mortar and pestle to ensure sample homogeneity. Measurements were performed on different sub-samples of each contaminated soil (replicate measurements) and sometimes two to three times on the same sub-sample (repeated measurements). For most concentrations of the contaminated soils, the DSC measurements were replicated at least twice. Measurements of two samples (CSU-MP200 and ISU-MP60) were repeated (three times and two times, respectively) but not replicated. Three samples (TG-MP65, CSU-CP190 and CSU-CP200) were replicated three times and four samples (CSU-MP100, CSU-MP250, TG-MP450 and TG-MP550) were replicated four times. Two replicate measurements were also performed on the aged CSU and ISU soils contaminated with CP. Data points plotted with error bars represent the mean value of replicated measurements for a given soil concentration plus or minus one standard deviation. The repeated measurements were averaged before they were combined with any other replicate measurements to calculate the standard deviations. Throughout this work, numerical values are reported as the mean ± one standard deviation along with the number of replicated measurements designated by n. Reported values of standard deviations for slopes and intercepts derived from linear regression were estimated as described elsewhere (Montgomery and Runger, 2003).
3. Results 3.1. Equilibrium uptake experiments The results of the SRM measurements on MP contaminated CSU soil are presented in Fig. 1. Uptake into the SRM increases with increasing soil concentration to a maximum value of approximately 0.77 mg MP (g SRM)−1, which is also the amount of uptake measured in SRM pieces that were equilibrated with either a saturated aqueous solution of MP or pure MP powder. These results are consistent with the fact that the fugacity should be the same for the neat powder, the saturated aqueous solution and saturated or supersaturated soil. When the chemical concentration in the soil, given as mass of chemical per mass of clean soil (Csoil), is less than Ssoil, the amount of uptake appears to vary linearly with Csoil. Assuming this is the case, Ssoil was estimated from the intersection of the maximum uptake and a straight-line that was forced through the origin and fitted to the uptake results for soil samples with Csoil less than Ssoil. The resulting value for Ssoil for MP in CSU soil is 88.5 mg MP (g soil)− 1. It should be noted that uncertainty in this estimate for Ssoil is potentially large due to uncertainty in the slope of the regression line arising from the small number measurements at concentrations smaller than Ssoil. The SRM uptake of MP from the ISU soil was the same after 21 and 35 days, indicating that 21 days was long enough to reach equilibrium (Fig. 2). However, the maximum SRM uptake was about 20% smaller than the uptake from both saturated water and neat powder, which were measured during the CSU soil experiments. The cause of this difference is unknown. Possible explanations include a shortage of neat chemical and biotic or abiotic degradation, but these are inconsistent with the observations that: (a) uptake was the same after 21 and 35 days; (b) uptake was equivalent for the three highest soil concentrations; and (c) the presence of neat chemical in the samples was confirmed by the DSC measurements conducted coincidentally with SRM uptake from the ISU soil. There are two other possible explanations for the discrepancy. The more likely cause is that, although the membrane area in direct contact with soil or powder was always the same, the total surface area of the membranes used in the ISU soil experiments were 2.5-fold larger than those used in the CSU soil and powder experiments. Since equilibration time was adequate (i.e., uptake was the same after 21 and 35 days), MP concentration in the membrane would be the same everywhere, including the area with
Uptake from MP-saturated water
0.8
Uptake from neat MP powder
0.6
mg MP / g SRM
the contaminated soil (i.e., Msoil + Mtot) and ΔHfus is the specific enthalpy of fusion of the neat chemical (Joules per mass of neat chemical). Combining Eqs. (2) and (3) gives:
6103
0.4
0.2
Ssoil 0 0
100
200
300
mg MP / g soil Fig. 1. Equilibrium uptake of MP into SRM measured as a function of MP concentration on the CSU soil after 18 days of equilibration compared with uptake from neat powder and a saturated aqueous solution (designated by the horizontal solid and dash-dot lines, respectively). The dashed lines represent the linear fit to the uptake results for Csoil less than Ssoil and the intersection of this line with the maximum uptake.
6104
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
for the ISU soil was thus estimated to be 36.1 mg (g soil)− 1 and 33.9 mg (g soil)− 1 from the data collected after 21 days and 35 days, respectively. On average, the estimated saturation concentration of MP in ISU soil was 35.0 mg (g soil)− 1. For the ISU soil, there is some evidence that SRM uptake might not vary linearly with Csoil when Csoil is less than Ssoil (see Fig. 2). The intersection of a straight-line fit to the ascending part of the uptake curve (i.e., without the uptake determination from the smallest soil concentration from the 21-day experiment and without forcing the regressed line through the origin) with the maximum uptake provides an estimate of the saturation concentration at 32 mg MP (g soil)− 1 and 31 mg MP (g soil)− 1 for the 21-day and the 35-day experiments, respectively. These estimates are only slightly smaller than the estimate derived assuming uptake was linearly proportional to Csoil for Csoil b Ssoil. These saturation limits for the ISU soil along with those for the CSU soil are listed in Table 1.
Uptake from MP-saturated water
0.8
Uptake from neat MP powder
mg MP / g SRM
0.6
0.4
0.2
Ssoil 0 0
40
80
120
mg MP / g soil Fig. 2. Equilibrium uptake of MP into SRM measured as a function of MP concentration on the ISU soil after 21 days ( ) and 35 days ( ) of equilibration compared with uptake from neat powder and a saturated aqueous solution (designated by the horizontal solid and dash-dot lines, respectively). The dashed lines represent the linear fit to the uptake results for Csoil less than Ssoil and the intersection of this line with the maximum uptake.
no soil contact, if we assume that evaporation of MP from the edges of the membrane is zero. However, although its vapor pressure is low, evaporation could be sufficient to reduce the SRM concentration in membranes that have a large area without chemical contact. Based on diffusion calculations accounting for the membrane area without direct chemical contact and assuming local equilibrium in the membrane area in contact with soil, we estimate that the observed smaller uptake from soil would occur if evaporation caused the concentration of chemical at the outer edge of the membrane to be 50% smaller than the concentration of MP in the portion of the membrane in direct contact with soil (calculation not shown). Note that for volatile chemicals, the concentration at the outer edge would be nearly zero (i.e., a 100% reduction). If this is the cause of the reduced SRM concentration, the soil saturation estimate will not be affected since the measurement was made at steady state. A second possible explanation is that room temperature, which was not recorded, was different during the SRM uptake studies of the CSU and ISU soils, which were separated by several months. If the temperature was lower during the SRM uptake studies of the ISU soil, then the SRM saturation limit measured from the ISU soil would be low relative to what it would be at the temperature of the CSU experiments. This would cause the intersection of the maximum uptake and the linear fit through the uptake results obtained for Csoil less than Ssoil to shift towards a lower value of Csoil. The difference between the maximum uptake from soil and the uptake from saturated solution and powder indicates that the underestimate in Ssoil is at most 10–15 mg g− 1. Even with such an increase, the estimated saturation limit of MP in the ISU soil (see below) would still be significantly lower than in the CSU soil. In the DSC studies, which are described below, the CSU and ISU soils were studied at the same time, and the soil saturation limit determined for the CSU soil was slightly smaller than was measured previously by SRM uptake, although the difference is probably not statistically significant (i.e., 85 ± 2 compared to 88.5 mg (g soil)− 1). Additional experiments show that the rate of uptake into the SRM is proportional to the contaminant concentration when it is below the soil saturation limit and is essentially constant for concentrations above the soil saturation limit (Deglin, 2007). As a result, even if SRM uptake in Fig. 2 was not at equilibrium, the saturation concentration of MP in the ISU soil can still be estimated from the intersecting lines, following the same procedure used for the CSU soil. The value of Ssoil
3.2. DSC experiments Fig. 3 shows a typical set of DSC endotherms obtained from the clean and MP-contaminated ISU soil compared with pure MP. Fig. 4 shows endotherms for the clean and CP-contaminated CSU soil as well as pure CP. DSC endotherms for the other soils and chemicals can be found elsewhere (Deglin, 2007). The enthalpies of melting for pure MP and CP were determined to be 168.5 ± 5.0 J g− 1 (n = 8) and 151.4 ± 4.1 J g− 1 (n = 3), respectively, with melting points at 128.1 ± 1.7 °C for MP and 114.2 ± 1.7 °C for CP. These results compare favorably to reported values of 159.9 J g− 1 (Manzo and Ahumda, 1990) for MP and melting point values of 125.4 °C (Manzo and Ahumda, 1990) and 131 °C (PhysProp Database) for MP and 110– 113 °C (PhysProp Database) for CP. Literature values of the heat of fusion for CP were not found. None of the DSC analyses of the clean soils showed any thermal transitions over the temperature range investigated (90–135 °C). When contaminated soils contained neat chemical, the enthalpy of melting was determined from the peak area. Repeated measurements performed on six different soil samples (CSU-MP200, PP-MP100 and all four concentrations of MP on the ISU soil) were on average within 3% of each other. Replicate measurements performed on different subsamples of the same soil were on average within 10% of each other. The larger variation of the replicated measurements compared with the repeated determinations suggests that a significant fraction of the variability is introduced by the soil sampling. The small sample size required for DSC analyses (typically between 10 and 15 mg) and also the possible variation in the distribution of neat chemical within the soil could be causes of the observed variability in the DSC measurements. Table 1 Values of the organic carbon fraction (foc), BET surface area (σsoil), and Ssoil. Soil
CSU ISU TG PP
foc
0.015 0.038 0.063 0.45
σsoil (m2 g− 1)
34.3 6.8 5.1 4.0
Ssoil (mg per g of clean soil) MP
CP
DSC
SRM
DSC
85 ± 2 34 ± 16 56 ± 16 41
88.5 35.0a NDb ND
91 ± 34 58 ± 22 ND ND
Σ = σsoil MW / (NA Ssoil) is the average surface area occupied by a molecule of contaminant with a soil saturation of Ssoil where MW is the molecular weight of the contaminant, and NA is Avogadro's number. Area based on the average value of Ssoil measured with both SRM uptake and DSC methods. a Average of the Ssoil measured on ISU after 21 days (36.1 mg g− 1) and after 35 days (33.9 mg g−1) calculated using a line forced through the origin and Csoil b Ssoil. If Ssoil is calculated using a line forced through 10 mg g− 1 b Csoil b Ssoil, then the average Ssoil is 31.5 mg g− 1, measured on ISU after 21 days (32 mg g− 1) and after 35 days (31 mg g− 1). b Not determined.
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
6105
Heat Flow
Pure MP ISU-MP 300 ISU-MP 180 ISU-MP 100 ISU-MP 60 Clean ISU
^ Csoil (mg MP / g contaminated soil)
400
300
200
100
0 124
128 (oC)
0
132
Fig. 3. DSC endotherms of clean and MP-contaminated ISU soil compared with pure MP.
Fig. 5 shows a plot of Ĉsoil as a function of the observed enthalpy of melting for MP on the CSU, ISU, TG and PP soils. In most cases, the error bars representing the standard deviation are smaller than the data symbols. Error bars relative to concentration are only shown for two values of Ĉsoil for the CSU and TG soils because these were the only soil samples for which a sufficient number of replicates was obtained for meaningful statistics. The saturation concentration of MP in CSU, ISU, TG and PP are reported in Table 1. The average enthalpy of fusion of MP obtained by averaging the enthalpy of fusion calculated as specified in Eq. (5) from the slope of each line shown in Fig. 5 was estimated to be 172.5 ± 6.9 J g−1. This value is about 2% larger than the average enthalpy of fusion measured for the pure compound (168.5 ± 5.0). The saturation concentration of CP was only determined for the CSU and ISU soils. Fig. 6 shows the results. From the intercept of each line, Ssoil in CSU and ISU soils was estimated to be 91 and 58 mg CP per g of clean soil, respectively. These values are compared with the measurements performed on MP-contaminated soils reported in Table 1. The CP saturation concentration differences observed in the freshly contaminated CSU and ISU soils (i.e., soils contaminated less than 1 month prior to the measurements) seems to disappear when the CSU and ISU soils have aged approximately 5.5 years after CP contamination (Fig. 6). The saturation concentration of CP in both aged soils determined as the intercept of a straight line going through
Heat Flow
Pure CP CSU-CP 250 CSU-CP 200
Clean CSU 105
110
115
40
60
Fig. 5. Determinations of the observed enthalpy of melting for MP plotted as a function of Ĉsoil for four contaminated soils: CSU ( , Ĉsoil = 5.1 ΔHobs + 78.4, R2 = 1.00), ISU ( , Ĉsoil = 5.9 ΔHobs + 33.0, R2 = 0.99), TG ( , Ĉsoil = 5.5 ΔHobs + 53.3, R2 = 1.00), and PP ( , Ĉsoil = 5.6 ΔHobs + 29.6).
the single data point obtained for each soil with the slopes obtained from the freshly contaminated soils, was estimated to be 85 and 96 mg g− 1 for the CSU and ISU soils, respectively. The enthalpy of fusion of CP estimated from the slope of the lines for the CSU and ISU soils was 169.4 ± 5.0 J g− 1, which is larger than that measured for pure CP (151.4 ± 4.1 J g− 1) by almost 12%. 4. Discussion As shown in Table 1, the soil saturation results determined from SRM uptake and DSC were in close agreement. For both the CSU and ISU soils, the soil saturation was larger for CP than MP. However, the difference was small for the CSU soil and approximately a factor of 1.7 for the ISU soil. For comparison, the ratio of the saturation concentrations of CP and MP in octanol is 1.5 (see Table 2), where the octanol saturation (Soct) is estimated as the product of the octanol-
300
200
100
0 0
CSU-CP 100
100
20
ΔHobs (J / g contaminated soil)
^ Csoil (mg CP / g contaminated soil)
120
120
125
(oC) Fig. 4. DSC endotherms of clean and CP-contaminated CSU soil compared with pure CP.
10
20
30
40
ΔHobs (J / g contaminated soil) Fig. 6. DSC determinations of the observed enthalpy of melting for CP plotted as a function of Ĉsoil for freshly contaminated soils: CSU ( , Ĉsoil = 5.3 ΔHobs + 83.0, R2 = 0.96), ISU ( , Ĉsoil = 5.7 ΔHobs + 54.6, R2 = 0.99). Also shown are measurements from samples that were aged for approximately 5.5 years: CSU ( ) and ISU ( ) soils. The dashed lines extrapolate the results obtained from the aged CSU and ISU soils to ΔHobs = 0 based on a slope that is the average of the slopes for the freshly contaminated ISU and CSU soils. From this, Ŝsoil is estimated to be 85.0 and 96.1 mg g− 1 for the CSU and ISU, respectively.
6106
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
Table 2 Estimated saturation concentrations in octanol (Soct), in soil organic carbon (Soc), and on the soil surface (Mσ) and the corresponding estimates of the area occupied per molecule (Σ) at surface saturation compared with estimated planar area of a molecule. logKow
Soct (mg/mL)a
Soc (mg g− 1)b
Mσ (mg m− 2)b
Σ (Å2 molecule− 1)c
CP MP Ratio CP/MP
1.60 1.96
340 230 1.5
1140 660 1.7
2.15 2.17 1.0
9.19 11.6 0.79
Predicted Saturation Concentration (mg / g clean soil)
Chemical
100
a Saturation concentration of MP or CP in octanol (Soct) was estimated as the product of the water saturation and Kow. b Soc and Mσ were determined for each chemical by regressing Ssoil values for all but the PP soil to Eq. (1). c Σ = MW / (NA Mσ) is the average surface area occupied by a molecule at the surface saturation Mσ where MW is the molecular weight of the contaminant, and NA is Avogadro's number.
water partition coefficient (Kow) and the water solubility (i.e., 8.5 mg mL− 1 for CP and 2.5 mg mL− 1 for MP (PhysProp Database)). Correlation of soil saturation to octanol saturation would be expected if soil saturation arose primarily from partitioning into the soil organic matter. The absence of correlation between soil and octanol saturation limits for CP and MP in the CSU soil while a correlation was observed in ISU soil suggests that saturation is controlled by different factors for the two soils. Significantly, the saturation concentrations of both CP and MP were largest in the soil with the largest surface area (CSU), even though it has the smallest organic carbon content (foc = 0.015) of all the soils studied. Overall, the saturation concentrations of MP varied by less than 2.5 despite a 30-fold variation in the organic carbon content of the different soils. It is evident that soil organic carbon alone does not explain the measured soil saturation values. Surface area also appears to contribute. For example, the saturation value of MP in the ISU, TG and PP soils varied by a factor of 1.6, which correlates with the 1.7-fold variation in surface area, but not the 12-fold variation in foc for these three soils. Also, soil surface areas for the ISU, TG and PP soils are smaller than the surface area of the CSU soil by at least a factor of 5, which, as shown below, can explain the larger soil saturation values in the CSU soil. Overall, the soil saturation values appear to be related to both organic carbon and surface area, as suggested by Eq. (1), except for the saturation concentration measured in PP, which is inconsistently low for a soil with such a large foc. Assuming that the absorption into the organic matter and adsorption on the mineral surfaces were similar for all soils except the PP, Soc and Mσ were determined by regressing the soil saturation values (Ssoil) to Eq. (1) for MP and CP, respectively. As shown in Fig. 7, the resulting equation provides an excellent prediction of the soil saturation concentrations determined in this study. The quantitative results, listed in Table 2, indicate that Soc for CP is 1.7-fold larger than for MP, which is similar to the ratio of 1.5 between the estimated octanol saturation concentrations of CP and MP. Also, the saturated surface concentration Mσ is essentially the same for CP and MP as expected for molecules of similar size. The average surface areas occupied by a molecule at Mσ are 9.2 and 11.6 Å2 molecule−1 for CP and MP, respectively. Notably, this is smaller than the area that a nitrogen molecule (16.2 Å2) or benzene molecule in its perpendicular orientation (25.0 Å2) would occupy assuming uniform monolayer distribution on the soil surface (Chiou et al., 1990; Bambrough et al., 1998). This suggests that (i) the surface saturation concentration estimated for these molecules results from a larger than monolayer coating of the soil surface, (ii) these molecules can reach surface areas that are not accessed in the nitrogen-BET measurements, or (iii) the orientation of the adsorbed molecules allows each of them to occupy a smaller area than nitrogen or benzene molecules. The estimated fraction of the soil saturation concentration that is due to the surface area is listed in Table 3 for the four soils studied.
CSU 80
60
TG
ISU
ISU
40
20
0 0
20
40
60
80
100
Measured Saturation Concentration (mg / g clean soil) Fig. 7. Predicted versus measured saturation values for MP (circles) in ISU, CSU and TG soils, and for CP (triangles) in ISU and CSU soils; the line represents the identity.
According to the calculations summarized in Tables 2 and 3, surface adsorption is responsible for more than 80% of the saturation concentration on the CSU soil but less than 37% for the other soils, including the PP. The estimate for PP was calculated assuming that Soc and Mσ were the same as for the other soils, but that the organic carbon in the PP is not all available. By this calculation, 11% of the actual foc of 0.45 was available for chemical absorption. This reduced sorption capacity is probably related to the fact that PP is dramatically different from the other soils in that it is virtually pure organic matter. Therefore, the organic carbon is not merely a surface coating on inorganic particles as is often the case in more mineral soils. Thus, the association between the organic carbon and contaminant molecules necessarily involves diffusion into the particle interiors, which is expected to be a very slow process. There is also the likelihood of dramatically different sorption capacities for organic carbon in the interior of particles. The latter scenario could be due, for example, to intra-particle associations of organic functional groups that alter the sorption capacity. Another possible factor contributing to this reduced sorption is that the organic matter in peat is in an earlier stage of decomposition than the organic carbon in the other soils. Less decomposition means a less advanced stage of humification and the likely presence of residual cellulosic particles that, because of their polar nature, would almost certainly have a significantly lower capacity for hydrophobic sorption of molecules such as MP and CP. Separate studies of dermal absorption from CSU and ISU soils contaminated with MP using acetone (Deglin, 2007) are consistent with the measured saturation concentrations determined in this
Table 3 Fraction of the soil saturation concentration associated with the surface area for MP and CP on the soils studieda. Soil
MP
CP
CSU ISU TG PPb
0.88 0.37 0.21 0.21
0.81 0.25 NDc ND
a
Calculated as (σsoil Mσ)/Ssoil for each soil and chemical. Calculated value for MP in PP was derived assuming that Mσ = 2.17 mg m− 2 and Soc = 660 mg mL− 1 (which are the same as for the other soils), but that only part of the organic carbon was available for absorption. The available oc (foc,avail) was estimated to be 0.049, which is about 11% of the actual foc = 0.45. c Not determined. b
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
study: dermal absorption of MP increased proportionally to soil concentration when the contamination level was less than saturation and remained constant for contamination levels above the saturation concentration. Based on these results as well as thermodynamic arguments, the soil saturation limit of chemicals that absorb through the skin is an important parameter for the exposure assessments involving contaminated soils. In this study, saturation concentrations varied with soil surface area in addition to the amount of soil organic matter. Confirmation that these observations are also observed on different soils contaminated with different chemicals is still needed. However, it is not clear whether this will be true for contaminated soils at hazardous waste sites, which would have undergone exposure to water from rainfall or other sources over an extended period of time. It is possible that surface area is more important for soils that have been contaminated with a volatile solvent like acetone, which also has water solubility and may, therefore, increase the surface area available for chemical sorption by removing adsorbed water from the surface. It is important to determine if this is the case, since most soils used in dermal absorption studies have been contaminated in this way (Kissel et al., 2007; Spalt et al., 2009). It is evident that the understanding and meaningful use of dermal absorption measurements from contaminated soil is more complicated than has been assumed in risk assessment practice. Extrapolations from one soil to another and from one concentration to another will most probably depend on the soil saturation. At present, the soil characteristics that affect soil saturation for actual contaminated soils are not clear, nor are the effects of variations in methods used for creating contaminated soils used in dermal absorption studies. Once these two issues are clarified, then protocols can be specified for preparing contaminated soils used in dermal absorption studies and for estimating soil saturation concentrations for use in risk assessment. Therefore, there is a need to understand how soil saturation varies with soil characteristics such as organic carbon content, surface area as well as the chemistry and physical condition of the organic and mineral phases. It is also important to determine how solvents commonly used to deliver contaminants to soils affect the capacity of the soils to absorb these compounds. Acknowledgments This work was supported in part by the U.S. Environmental Protection Agency (EPA) under agreement R830131-01. No endorsement from the EPA should be inferred.
6107
References Bambrough CM, Slade RCT, Williams RT, Burkett SL, Sims SD, Mann S. Sorption of nitrogen, water vapor, and benzene by a phenyl-modified MCM-41 sorbent. J Colloid Interface Sci 1998;201:220–2. Boethling RS, Mackay D. Handbook of Property Estimation Methods for Chemicals. Boca Raton: Lewis Publishers; 2000. 481 pp. Bunge AL, Parks JM. Soil contamination: theoretical descriptions. In: Roberts MS, Walters KA, editors. Dermal Absorption and Toxicity Assessment. New York, NY: Marcel Dekker; 1998. p. 669–96. Chiou TC, Leet J-F, Boyd SA. The surface area of soil organic matter. Environ Sci Technol 1990;24:1164–6. Choate LM. The effect of variations in soil organic matter with particle size on organic contaminant sorption and its relationship to dermal exposure. PhD thesis. Colorado School of Mines, Golden, CO, 2002. Choate LM, Ranville JF, Bunge AL, Macalady DL. Dermally adhered soil: 1. Amount and particle-size distribution. IEAM 2006;2:375–84. Deglin SE. Dermal Absorption of Low Volatility Chemicals from Soils. PhD Thesis. Colorado School of Mines, Golden, Colorado, 2007. Duff RM, Kissel JC. Effect of soil loading on dermal absorption efficiency from contaminated soil. J Toxicol Environ Health 1996;48:93-106. Goss KU, Buschmann J, Schwarzenbach RP. Adsorption of organic vapors to air-dry soils: model predictions and experimental validation. Environ Sci Technol 2004;38: 3667–73. Kissel JC, Richter KY, Fenske RA. Field measurement of dermal soil loading attributable to various activities: implications for exposure assessment. Risk Anal 1996;16: 115–25. Kissel JC, Spalt EW, Shirai JH, Bunge AL. Dermal absorption of chemical contaminants from soil. In: Roberts MS, Walters KA, editors. Dermal Absorption and Toxicity Assessment. Second. New York: Informa Healthcare; 2007. p. 563–73. Li J, Masso JJ, Guertin JA. Prediction of drug solubility in an acrylate adhesive based on the drug-polymer interaction parameter and drug solubility in acetonitrile. J Controlled Release 2002;83:211–21. Lyman WJ, Reehl WF, Rosenblatt DH. Handbook of Chemical Property Estimation Methods: Environmental Behaviour of Organic Compounds. Washington, DC: American Chemical Society; 1990. Manzo RH, Ahumda A. Effects of solvent medium on solubility. V: Enthalpic and entropic contributions to the free energy changes of di-substituted benzene derivatives in ethanol-water and ethanol-cyclohexane mixtures. J Pharm Sci 1990;79:1109–15. Montgomery DC, Runger GC. Applied Statistics and Probability for Engineers. New York: John Wiley & Sons, Inc; 2003. 706 pp. PhysProp Database. Syracuse Research Corporation. http://www.syrres.com/esc/ physdemo.htm. Accessed Aug. 18, 2010. Shih Y-H, Wu S-C. Distinctive sorption mechanisms of soil organic matter and mineral components as elucidated by organic vapor uptake kinetics. Environ Toxicol Chem 2005:2827–32. Spalt EW, Kissel JC, Shirai JH, Bunge AL. Dermal absorption of environmental contaminants from soil and sediment: a critical review. J Expo Sci Environ Epidemiol 2009;19:119–48. Theeuwes F, Hussain A, Higuchi T. Quantitative analytical method for determination of drugs dispersed in polymers using differential scanning calorimetry. J Pharm Sci 1974;63:427–9. US EPA. Percutaneous absorption of 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin (TCDD) and 3,3',4,4'-tetrachlorobiphenyl (TCB) applied in soil. OHEA-E-453. Washington, DC, December 1991. US EPA. Dermal Exposure Assessment: Principles and Applications. EPA/600/891/011B. Washington, DC, January 1992, 389 pp.
Contents lists available at ScienceDirect
Science of the Total Environment j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s c i t o t e n v
Measuring the saturation limit of low-volatility organic compounds in soils: Implications for estimates of dermal absorption Sandrine E. Déglin a,1, Donald L. Macalady a, Annette L. Bunge b,⁎ a b
Chemistry Department, Colorado School of Mines, Golden, CO 80401, USA Chemical Engineering, Colorado School of Mines, Golden, CO 80401, USA
a r t i c l e
i n f o
Article history: Received 8 February 2010 Received in revised form 21 August 2010 Accepted 26 August 2010 Keywords: Soil Contaminated soil Saturation limit Dermal absorption Skin Risk assessment
a b s t r a c t Estimating dermal absorption from contaminated soils typically requires extrapolations from measurements obtained on soils artificially contaminated at much larger concentrations. Such extrapolations should be constrained by the fact that maximum absorption will occur for the largest possible concentration of chemical on the soil without neat chemical being present; i.e., at the soil saturation limit (Ssoil). Saturation limits of two low-volatility model compounds (4-cyanophenol and methyl paraben) were determined on the 38–63 μm sieve fraction of four soils with different fractions of organic carbon (foc = 0.015–0.45) and specific surface areas (σsoil = 4–34 m2 g− 1) using two methods: equilibrium uptake into silicone rubber membranes and differential scanning calorimetry. Except for Pahokee peat, which had the largest foc, a model assuming contributions from both surface adsorption and organic carbon absorption provided excellent predictions of Ssoil. In all soils, the surface saturation concentration of both chemicals was estimated at 2.2 mg m− 2. The saturation concentration of 4-cyanophenol in the soil organic carbon was 1.7-fold higher than methyl paraben, which is consistent with the estimated solubility limits of these two chemicals in octanol. © 2010 Elsevier B.V. All rights reserved.
1. Introduction In the United States, remediation decisions on contaminated sites are based on assessment of the potential health risks. These assessments often require characterization of the risks associated with exposure to contaminated soils, including skin contact (Kissel et al., 1996). A recent critical review of the literature describing dermal absorption from contaminated soils identified experimental measurements for 33 organic and 4 inorganic compounds (Kissel et al., 2007; Spalt et al., 2009). Although the approaches employed by various research groups have varied widely, the chemical contaminant was commonly added to the soil in a volatile solvent that was removed by evaporation (Spalt et al., 2009). It is also common for dermal absorption to be measured from one chemical concentration on soil that is applied at a single mass per area of skin (i.e., the soil load). Nearly always, the investigated contaminant concentrations and soil loads substantially exceed those that would occur in actual exposures. As a result, risk assessors are faced with decisions about how to extrapolate experimental measurements to a different soil contam-
⁎ Corresponding author. Colorado School of Mines, Chemical Engineering Department, 1500 Illinois Street, Golden, CO 80401-1887. Tel.: +1 303 273 3722; fax: +1 303 273 3730. E-mail addresses: [email protected] (S.E. Déglin), [email protected] (D.L. Macalady), [email protected] (A.L. Bunge). 1 Present address: Exponent Inc., Washington, DC 20036, USA. 0048-9697/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.scitotenv.2010.08.042
inated at a different concentration with a different amount in contact with the skin. It is well established that in aqueous suspensions, the equilibrium concentrations of organic compounds in soil are proportional to their aqueous concentrations and the organic matter content of the soil, usually reported as the mass fraction of organic carbon per mass of soil (foc) (Lyman et al., 1990; Boethling and Mackay, 2000). Based on this, it has been proposed that dermal absorption should be inversely proportional to foc (US EPA, 1992; Bunge and Parks, 1998; Spalt et al., 2009); i.e., an increase in the soil sorption capacity would cause the dermal bioavailability to decrease. Of the limited data available for testing this hypothesis, generally an increase in foc produced a less than proportional decrease in absorbed dose (US EPA, 1991, 1992; Duff and Kissel, 1996). In part, this may arise from differences in the soils investigated in dermal absorption studies compared with those equilibrated with organic compounds in water. However, the most obvious difference is that soils in dermal absorption studies usually contain no liquid water. Also, the amount of chemical applied to soils using a volatile solvent can exceed the capacity of the soil to hold chemical, as determined by equilibrating the soil with an aqueous solution. All soils have a limited capacity to hold a chemical in a form that is thermodynamically distinct from the pure form of the chemical. If a chemical is present at a level exceeding the capacity of a given matrix (be it soil, some other solid, a gas, water, or some other liquid), then,
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
at equilibrium, the pure (or neat) chemical is present as a separate phase. For soils equilibrated in water saturated with an organic compound, the soil saturation process mainly involves partitioning into the soil organic matter (as long as foc is not extremely low). For air-dried soils, chemical uptake most probably depends on surface adsorption as well as partitioning into the soil organic matter (Goss et al., 2004). Therefore, chemical saturation of air-dried soils should vary with surface area in addition to the amount of soil organic matter as described by Eq. (1): Ssoil = foc Soc + σsoil Mσ
ð1Þ
which is an extension of the soil-air partition equation proposed by Goss et al. (2004). In this equation, Ssoil is the soil saturation concentration in units of mass of chemical per mass of clean soil, Soc is the saturation concentration of chemical in the soil organic matter (per mass of soil organic matter), σsoil is the specific surface area expressed in area per mass of soil, and Mσ is the mass of chemical adsorbed per area of the saturated soil surface area. Because surface adsorption of water reduces the surface area available for sorption of organic compounds, Mσ will decrease as relative humidity increases (Shih and Wu, 2005). As a result, surface adsorption of organic contaminants will be smaller when the relative humidity is higher (Goss et al., 2004). The rate and amount of dermal absorption of a given chemical is proportional to its thermodynamic driving force, which is maximized when the skin surface is exposed to a saturated medium, whether it is a soil or liquid solution. As a result, for a given soil, the amount of chemical that will absorb into the skin is expected to increase proportionally with the concentration of chemical on the soil (reported per mass of clean soil, Csoil) as long as Csoil is less than Ssoil. Increasing the soil concentration above Ssoil will not increase the thermodynamic driving force and dermal absorption is not expected to exceed that which occurs when Csoil = Ssoil provided the barrier function of the skin is unaltered. Moreover, for a given compound, Ssoil represents an intrinsic characteristic of each soil and therefore, in comparing two soils, the thermodynamic driving force for dermal absorption from each soil should be related to the ratio of Csoil to Ssoil. The purpose of this study was to measure saturation concentrations in soils contaminated using volatile solvents as vehicles, as has been done in many dermal absorption studies. Saturation concentrations of two model chemicals, methyl paraben and 4-cyanophenol, which are both solids at skin temperature (32 °C), were determined for four soils with widely different fractions of organic matter. Two methods were used. In the first method, soil saturation was determined by measuring the amount of chemical uptake into pieces of silicone rubber membrane (SRM) that were equilibrated with the contaminated soil. In this method, equilibrium uptake into the SRM increases with soil concentration to a maximum value that remains constant for all soil concentrations above the soil saturation limit. The soil saturation limit is then the smallest soil concentration of the contaminant that leads to the maximum uptake into the SRM sorbent. In the second method, using differential scanning calorimetry (DSC), the amount of pure chemical present as a separate solid phase within or on the soil sample is quantified by the measurement of its enthalpy of melting. This technique has been used for measuring the solubility of drugs in adhesives (Li et al., 2002) as well as polymers (Theeuwes et al., 1974). In the present study, soil saturation measurements performed using DSC were compared with those from SRM uptake. The results from both methods indicate that the saturation values of the chosen test chemicals for the soils and conditions studied here cannot be adequately estimated using the amount of soil organic matter alone. As expected from prior studies of soil–air uptake, surface area is an important factor.
6101
2. Materials and methods 2.1. Chemicals Reagent grade 4-cyanophenol (CP; molecular weight 119; logarithm of the octanol-water partition coefficient, logKow, is 1.60 (PhysProp Database)) and methyl paraben (MP; molecular weight 152; logKow = 1.96 (PhysProp Database)) were both purchased from Sigma-Aldrich (Milwaukee, WI). Both CP and MP are characterized by low volatility (estimated vapor pressures are 0.89 and 0.032 Pa, respectively (PhysProp Database)). Liquid chromatography grade acetonitrile and acetone were from Mallinckrodt Chemicals (Phillipsburg, NJ). 2.2. Soils The saturation limits of MP and CP were studied in four soils, identified as CSU, ISU, TG and PP, which differed in the amount and nature of their organic matter. The CSU (a clay loam) and ISU (a silty clay loam) soils were collected from the Colorado State University (Fort Collins, CO) and Iowa State University (Ames, IA), respectively. The collection, preparation and characterization of both of these soils are described by Choate et al. (Choate, 2002; Choate et al., 2006). The TG soil is a commercial top soil purchased from Timberline Gardens (Wheat Ridge, CO). Pahokee peat (PP) was originally collected on behalf of the International Humic Substances Society (IHSS) and was provided by Professor Patrick MacCarthy (Colorado School of Mines, Golden, CO). All of the experiments described in this study were performed on the 38–63 μm dry sieve fraction of each soil, which was used in dermal absorption studies performed with the CSU and ISU soils (Deglin, 2007). In a study of the CSU and ISU soils, which contained a wide distribution of particle sizes b250 μm, Choate et al. demonstrated that the majority of particles adhering to the skin have diameters b63 μm (Choate et al., 2006). The lower end of the fraction (38 μm) was chosen to limit and control the extent of the particle size distribution. For the CSU and ISU soils, the 38–63 μm sieve fraction was prepared by sieving the b250 μm fraction from the previous studies by Choate et al. (Choate, 2002; Choate et al., 2006) with stainless steel sieves (W.S. Tyler, Mentor, OH) stacked as follows: 125 μm (No. 120), 63 μm (No. 230), and 38 μm (No. 400). The PP soil was sieved as for the CSU and ISU soils with two sieves added, i.e., 550 μm (No. 35) and 250 μm (No. 60). Two more sieves, the 4 mm (No. 5) and 2 mm (No. 10), were included in the stack used to prepare the TG soil. Using procedures described previously (Choate, 2002), the mass fractions of organic carbon in the 38–63 μm sieve fractions were determined to be 0.015, 0.038, 0.063 and 0.45 for CSU, ISU, TG and PP, respectively. In a study by Choate et al. (Choate, 2002; Choate et al., 2006) performed on different sub-samples of the same size fractions of the CSU and ISU soils, the mass fractions of organic carbon were measured to be 0.011 and 0.031, respectively. The specific surface area (σsoil) of the CSU, TG and PP soil samples were measured by the BET method, using N2 adsorption and calculated from the desorption data (Chiou et al., 1990). The value of σsoil for the 38–63 μm fraction of the ISU soil was obtained from Choate (Choate, 2002; Choate et al., 2006). The MP saturation limit was studied on all four soils, whereas the CP saturation limit was studied on the CSU and ISU soils only. 2.3. Chemical contamination of soils The contamination procedures were similar for all soils except for MP on the CSU soil. The mass of chemical necessary to achieve the desired contaminant concentration was dissolved into the minimum necessary volume of acetone or acetonitrile, usually about 1 to 2 mL. Acetonitrile was only used for MP contamination of the TG soil. A portion of the solution (typically, about 0.5 mL) was added to 1 g of
6102
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
soil in a glass vial, which was briefly shaken to optimize the contact between the soil and the solution. The solvent was allowed to evaporate at room temperature for 12 to 24 h, and then the vial containing the soil was shaken to break up the clumps. This process was repeated until all of the solvent containing the chemical of interest had been added to the soil. This procedure of adding the solvent in increments was followed to reduce the amount of chemical that was lost to the walls of the vial that contained the soil. After the final increment of the solvent was allowed to evaporate for at least 24 h, the soil was again shaken. Large particle aggregates that were sometimes present in the larger concentration samples (> about 100 mg/g of clean soil) were broken up by grinding with a mortar and pestle. The soils were allowed to incubate at room temperature without agitation for at least 48 h before being sampled to determine the actual concentration, and usually for 7 to 10 days before being used to assess soil saturation. For MP contamination of the CSU soil, MP was dissolved into the minimal volume of acetonitrile and then delivered all at once to 5 g of soil. After the solvent was allowed to evaporate at room temperature for at least 24 h, the soil was broken up by shaking the vial and, when necessary, large aggregates present at the larger concentrations were ground with a mortar and pestle. The contaminated soil samples were then mixed for 24 h on a wrist-action shaker (Rotoshake Genie from Scientific Industries Bohemia, NY), after which they were incubated at room temperature for 2 to 3 days without any further agitation before starting the saturation measurements. The specific concentrations that were prepared for MP and CP in each soil varied. Determinations of soil saturation by SRM uptake required concentrations that are both smaller and larger than saturation. However, soil saturation measured by the DSC method only required concentrations that exceed saturation. Saturation limits of MP were determined by DSC in all four soils and by SRM uptake in the CSU and ISU soils. Soil saturation limits of CP were measured by DSC alone in only the CSU and ISU soils. The DSC technique was also used to measure the saturation limit of CP in the less than 250 μm fraction of the CSU and ISU soils that were contaminated 5.5 years earlier using acetone, which was removed by evaporation (unpublished report, C-P Chen, CSM, Golden, CO, 2002). These samples are designated as aged CSU and aged ISU soils. Nominal concentrations of MP were 2, 8, 12, 30, 60, 100, 185, and 320 mg MP (g clean soil)− 1 for the ISU soil and 20, 30, 40, 100, 200 and 250 mg MP (g clean soil)− 1 for the CSU soil. Concentrations of MP were 65, 450 and 550 mg (g clean soil)−1 for the TG soil and 70 and 100 mg (g soil)− 1 for the PP soil. The nominal CP concentrations were 80, 190 and 260 mg (g soil)− 1 for the ISU soil and 100, 140, 200 and 250 mg (g soil)− 1 for the CSU soil. The actual concentrations of chemical in each soil were measured by extracting 15 to 20 mg of soil with 10 mL acetonitrile and analyzing the extract by HPLC using the method described below. Although relative humidity was not controlled in this study, it is relatively constant at approximately 50% in the laboratory where these experiments were conducted. This estimate is also consistent with determinations of moisture content for air-dried and controlled humidity samples of the CSU and ISU soils from an earlier study (Choate et al., 2006). In the following discussion, soil samples contaminated with different amounts of either CP or MP are designated by the soil acronym (CSU, ISU, TG or PP) followed by the chemical (MP or CP) and the nominal contaminant concentration given in units of mg of chemical per gram of clean soil. For example, the soil sample identified as CSU-CP100 is the CSU soil contaminated with approximately 100 mg of CP per gram of clean soil. 2.4. Equilibrium uptake experiments Saturation limits of MP in the CSU and ISU soils were evaluated by measuring the equilibrium uptake into SRM. Pieces (about 1 or
2.5 cm2) of SRM sheets (88 ± 6 μm thick) from Membrane Products Corporation (Albany, NY) were weighed and then equilibrated at room temperature (approximately 23 °C) with neat MP powder as well as with CSU and ISU soils with different values of soil concentration. After a waiting period of 18 days for the CSU soil and 21 or 35 days for the ISU soil, each SRM piece was extracted in 5 mL of acetonitrile, and the extract was analyzed by HPLC. Preliminary tests confirmed that all chemicals were recovered from the SRM using a single extraction. If the SRM has direct contact with the soil or the neat chemical powder, it is difficult to insure that no particles are left on the SRM surface after the soil or chemical have been removed. To avoid this problem, the SRM samples were equilibrated with the soils or powders using an indirect method of contact. Specifically, one SRM piece was sandwiched between a second piece of SRM and aluminum foil. This membrane-foil sandwich was mounted into a Franz-type vertical, static diffusion cell (9 mm, jacketed cells) from Permegear (Bethlehem, PA) with the aluminum foil facing down towards the receptor chamber, which was left empty. A large excess of contaminated soil or neat chemical powder was then placed on top of the upper SRM piece. The aluminum foil provided a barrier for chemical loss from the bottom side of the membrane and the top piece of SRM allowed chemical transfer to the second SRM piece, while preventing soil from adhering. After carefully removing the top protective SRM piece, chemical uptake into the protected second piece of SRM was measured. As many as seven of these systems could be equilibrating at one time and therefore, for a particular soil the whole range of concentrations could be tested in a single experiment. The uptake results from these experiments were compared to SRM uptake when the same membranes were equilibrated for 5 days at room temperature with a saturated aqueous solution containing excess solid MP. In this case, the SRM pieces were protected from the chemical particles by placing the membranes inside dialysis bags (Spectrapore Membrane Tubing, 25 mm, 12,000–14,000 molecular weight cut-off). 2.5. High performance liquid chromatography (HPLC) analysis Solutions containing MP or CP were analyzed using a Hewlett Packard 1100 series modular HPLC system equipped with a Zorbax Eclipse XDB-C18 analytical column (Hewlett Packard, 4.6 mm × 25 cm) and a guard column (4.6 mm × 12.5 mm) with the same packing. The injection volume was 10 μL and the mobile phase was 70% HPLC grade acetonitrile and 30% de-ionized water pumped at 1 mL min− 1. Retention times for CP and MP were about 3.1 and 2.9 min, respectively. 2.6. Soil saturation measurements by DSC The method involves the measurement of the enthalpy of melting of pure chemical present within or on the sample, in this case the soil. The approach relies on two assumptions: (1) only the neat chemical contributes to the enthalpy change, and (2) the amount of chemical dissolved into the soil matrix does not change for soil concentrations exceeding the saturation limit. Based on these assumptions, the applicable mass and energy balances are expressed by Eqs. (2) and (3): Mtot = Msorb + Mneat
ð2Þ
ΔHobs ðMsoil + Mtot Þ = ΔHfus Mneat
ð3Þ
For a given mass of clean soil (Msoil), Mtot is the total mass of chemical distributed on the soil, Msorb is the mass of chemical dissolved into or adsorbed within the soil, Mneat is the mass of chemical present as a separate phase. In the energy balance, ΔHobs is the observed enthalpy of melting of the chemical in Joules per mass of
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
ΔHobs = ΔHfus
M Mtot 1− sorb Mtot Msoil + Mtot
ð4Þ
which can be written in terms of the mass fraction of all chemical in the contaminated soil (i.e., Ĉsoil = Mtot / (Msoil + Mtot)) and the mass fraction of chemical at the saturation limit in a mass of contaminated soil (i.e., Ŝsoil = Msorb /(Msoil + Msorb)) as follows (Li et al., 2002): ΔH obs + Sˆsoil Cˆ soil = 1− Sˆsoil ΔHfus
ð5Þ
Eq. (5) was derived by relating the ratio of Msorb to Mtot in Eq. (4) to the contaminant mass fractions, expressed as a mass of chemical per mass of clean soil, at concentrations above (Csoil) and at soil saturation (Ssoil); i.e., Msorb / Mtot = Ssoil / Csoil. Finally, from the mass balance of contaminant on the soil, Csoil and Ssoil were related to Ĉsoil and Ŝsoil as given by Eqs. (6) and (7):
Csoil =
Cˆ soil Mtot = Msoil 1−Cˆ soil
ð6Þ
Ssoil =
Msorb Sˆsoil = Msoil 1− Sˆsoil
ð7Þ
According to Eq. (5), the observed enthalpy and the chemical concentration on the soil are linearly correlated. Therefore, on a plot presenting Ĉsoil as a function of ΔHobs, the intercept of the line at ΔHobs equal to zero corresponds to Ŝsoil and the slope is the product of (1 − Ŝsoil) and the reciprocal of the enthalpy of fusion of the pure chemical (ΔHfus). The DSC analyses were performed on pure MP and CP powder as well as clean and contaminated soils using a Pyris 6 differential calorimeter from Perkin-Elmer (Waltham, MA). Samples of pure chemical or soil (approximately 5 or 10–20 mg, respectively) were weighed into an aluminum DSC pan with crimped cover and analyzed by increasing the temperature from 90 to 135 °C at a rate of 10 °C per minute. Prior to the DSC analysis, contaminated soil samples were ground with a mortar and pestle to ensure sample homogeneity. Measurements were performed on different sub-samples of each contaminated soil (replicate measurements) and sometimes two to three times on the same sub-sample (repeated measurements). For most concentrations of the contaminated soils, the DSC measurements were replicated at least twice. Measurements of two samples (CSU-MP200 and ISU-MP60) were repeated (three times and two times, respectively) but not replicated. Three samples (TG-MP65, CSU-CP190 and CSU-CP200) were replicated three times and four samples (CSU-MP100, CSU-MP250, TG-MP450 and TG-MP550) were replicated four times. Two replicate measurements were also performed on the aged CSU and ISU soils contaminated with CP. Data points plotted with error bars represent the mean value of replicated measurements for a given soil concentration plus or minus one standard deviation. The repeated measurements were averaged before they were combined with any other replicate measurements to calculate the standard deviations. Throughout this work, numerical values are reported as the mean ± one standard deviation along with the number of replicated measurements designated by n. Reported values of standard deviations for slopes and intercepts derived from linear regression were estimated as described elsewhere (Montgomery and Runger, 2003).
3. Results 3.1. Equilibrium uptake experiments The results of the SRM measurements on MP contaminated CSU soil are presented in Fig. 1. Uptake into the SRM increases with increasing soil concentration to a maximum value of approximately 0.77 mg MP (g SRM)−1, which is also the amount of uptake measured in SRM pieces that were equilibrated with either a saturated aqueous solution of MP or pure MP powder. These results are consistent with the fact that the fugacity should be the same for the neat powder, the saturated aqueous solution and saturated or supersaturated soil. When the chemical concentration in the soil, given as mass of chemical per mass of clean soil (Csoil), is less than Ssoil, the amount of uptake appears to vary linearly with Csoil. Assuming this is the case, Ssoil was estimated from the intersection of the maximum uptake and a straight-line that was forced through the origin and fitted to the uptake results for soil samples with Csoil less than Ssoil. The resulting value for Ssoil for MP in CSU soil is 88.5 mg MP (g soil)− 1. It should be noted that uncertainty in this estimate for Ssoil is potentially large due to uncertainty in the slope of the regression line arising from the small number measurements at concentrations smaller than Ssoil. The SRM uptake of MP from the ISU soil was the same after 21 and 35 days, indicating that 21 days was long enough to reach equilibrium (Fig. 2). However, the maximum SRM uptake was about 20% smaller than the uptake from both saturated water and neat powder, which were measured during the CSU soil experiments. The cause of this difference is unknown. Possible explanations include a shortage of neat chemical and biotic or abiotic degradation, but these are inconsistent with the observations that: (a) uptake was the same after 21 and 35 days; (b) uptake was equivalent for the three highest soil concentrations; and (c) the presence of neat chemical in the samples was confirmed by the DSC measurements conducted coincidentally with SRM uptake from the ISU soil. There are two other possible explanations for the discrepancy. The more likely cause is that, although the membrane area in direct contact with soil or powder was always the same, the total surface area of the membranes used in the ISU soil experiments were 2.5-fold larger than those used in the CSU soil and powder experiments. Since equilibration time was adequate (i.e., uptake was the same after 21 and 35 days), MP concentration in the membrane would be the same everywhere, including the area with
Uptake from MP-saturated water
0.8
Uptake from neat MP powder
0.6
mg MP / g SRM
the contaminated soil (i.e., Msoil + Mtot) and ΔHfus is the specific enthalpy of fusion of the neat chemical (Joules per mass of neat chemical). Combining Eqs. (2) and (3) gives:
6103
0.4
0.2
Ssoil 0 0
100
200
300
mg MP / g soil Fig. 1. Equilibrium uptake of MP into SRM measured as a function of MP concentration on the CSU soil after 18 days of equilibration compared with uptake from neat powder and a saturated aqueous solution (designated by the horizontal solid and dash-dot lines, respectively). The dashed lines represent the linear fit to the uptake results for Csoil less than Ssoil and the intersection of this line with the maximum uptake.
6104
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
for the ISU soil was thus estimated to be 36.1 mg (g soil)− 1 and 33.9 mg (g soil)− 1 from the data collected after 21 days and 35 days, respectively. On average, the estimated saturation concentration of MP in ISU soil was 35.0 mg (g soil)− 1. For the ISU soil, there is some evidence that SRM uptake might not vary linearly with Csoil when Csoil is less than Ssoil (see Fig. 2). The intersection of a straight-line fit to the ascending part of the uptake curve (i.e., without the uptake determination from the smallest soil concentration from the 21-day experiment and without forcing the regressed line through the origin) with the maximum uptake provides an estimate of the saturation concentration at 32 mg MP (g soil)− 1 and 31 mg MP (g soil)− 1 for the 21-day and the 35-day experiments, respectively. These estimates are only slightly smaller than the estimate derived assuming uptake was linearly proportional to Csoil for Csoil b Ssoil. These saturation limits for the ISU soil along with those for the CSU soil are listed in Table 1.
Uptake from MP-saturated water
0.8
Uptake from neat MP powder
mg MP / g SRM
0.6
0.4
0.2
Ssoil 0 0
40
80
120
mg MP / g soil Fig. 2. Equilibrium uptake of MP into SRM measured as a function of MP concentration on the ISU soil after 21 days ( ) and 35 days ( ) of equilibration compared with uptake from neat powder and a saturated aqueous solution (designated by the horizontal solid and dash-dot lines, respectively). The dashed lines represent the linear fit to the uptake results for Csoil less than Ssoil and the intersection of this line with the maximum uptake.
no soil contact, if we assume that evaporation of MP from the edges of the membrane is zero. However, although its vapor pressure is low, evaporation could be sufficient to reduce the SRM concentration in membranes that have a large area without chemical contact. Based on diffusion calculations accounting for the membrane area without direct chemical contact and assuming local equilibrium in the membrane area in contact with soil, we estimate that the observed smaller uptake from soil would occur if evaporation caused the concentration of chemical at the outer edge of the membrane to be 50% smaller than the concentration of MP in the portion of the membrane in direct contact with soil (calculation not shown). Note that for volatile chemicals, the concentration at the outer edge would be nearly zero (i.e., a 100% reduction). If this is the cause of the reduced SRM concentration, the soil saturation estimate will not be affected since the measurement was made at steady state. A second possible explanation is that room temperature, which was not recorded, was different during the SRM uptake studies of the CSU and ISU soils, which were separated by several months. If the temperature was lower during the SRM uptake studies of the ISU soil, then the SRM saturation limit measured from the ISU soil would be low relative to what it would be at the temperature of the CSU experiments. This would cause the intersection of the maximum uptake and the linear fit through the uptake results obtained for Csoil less than Ssoil to shift towards a lower value of Csoil. The difference between the maximum uptake from soil and the uptake from saturated solution and powder indicates that the underestimate in Ssoil is at most 10–15 mg g− 1. Even with such an increase, the estimated saturation limit of MP in the ISU soil (see below) would still be significantly lower than in the CSU soil. In the DSC studies, which are described below, the CSU and ISU soils were studied at the same time, and the soil saturation limit determined for the CSU soil was slightly smaller than was measured previously by SRM uptake, although the difference is probably not statistically significant (i.e., 85 ± 2 compared to 88.5 mg (g soil)− 1). Additional experiments show that the rate of uptake into the SRM is proportional to the contaminant concentration when it is below the soil saturation limit and is essentially constant for concentrations above the soil saturation limit (Deglin, 2007). As a result, even if SRM uptake in Fig. 2 was not at equilibrium, the saturation concentration of MP in the ISU soil can still be estimated from the intersecting lines, following the same procedure used for the CSU soil. The value of Ssoil
3.2. DSC experiments Fig. 3 shows a typical set of DSC endotherms obtained from the clean and MP-contaminated ISU soil compared with pure MP. Fig. 4 shows endotherms for the clean and CP-contaminated CSU soil as well as pure CP. DSC endotherms for the other soils and chemicals can be found elsewhere (Deglin, 2007). The enthalpies of melting for pure MP and CP were determined to be 168.5 ± 5.0 J g− 1 (n = 8) and 151.4 ± 4.1 J g− 1 (n = 3), respectively, with melting points at 128.1 ± 1.7 °C for MP and 114.2 ± 1.7 °C for CP. These results compare favorably to reported values of 159.9 J g− 1 (Manzo and Ahumda, 1990) for MP and melting point values of 125.4 °C (Manzo and Ahumda, 1990) and 131 °C (PhysProp Database) for MP and 110– 113 °C (PhysProp Database) for CP. Literature values of the heat of fusion for CP were not found. None of the DSC analyses of the clean soils showed any thermal transitions over the temperature range investigated (90–135 °C). When contaminated soils contained neat chemical, the enthalpy of melting was determined from the peak area. Repeated measurements performed on six different soil samples (CSU-MP200, PP-MP100 and all four concentrations of MP on the ISU soil) were on average within 3% of each other. Replicate measurements performed on different subsamples of the same soil were on average within 10% of each other. The larger variation of the replicated measurements compared with the repeated determinations suggests that a significant fraction of the variability is introduced by the soil sampling. The small sample size required for DSC analyses (typically between 10 and 15 mg) and also the possible variation in the distribution of neat chemical within the soil could be causes of the observed variability in the DSC measurements. Table 1 Values of the organic carbon fraction (foc), BET surface area (σsoil), and Ssoil. Soil
CSU ISU TG PP
foc
0.015 0.038 0.063 0.45
σsoil (m2 g− 1)
34.3 6.8 5.1 4.0
Ssoil (mg per g of clean soil) MP
CP
DSC
SRM
DSC
85 ± 2 34 ± 16 56 ± 16 41
88.5 35.0a NDb ND
91 ± 34 58 ± 22 ND ND
Σ = σsoil MW / (NA Ssoil) is the average surface area occupied by a molecule of contaminant with a soil saturation of Ssoil where MW is the molecular weight of the contaminant, and NA is Avogadro's number. Area based on the average value of Ssoil measured with both SRM uptake and DSC methods. a Average of the Ssoil measured on ISU after 21 days (36.1 mg g− 1) and after 35 days (33.9 mg g−1) calculated using a line forced through the origin and Csoil b Ssoil. If Ssoil is calculated using a line forced through 10 mg g− 1 b Csoil b Ssoil, then the average Ssoil is 31.5 mg g− 1, measured on ISU after 21 days (32 mg g− 1) and after 35 days (31 mg g− 1). b Not determined.
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
6105
Heat Flow
Pure MP ISU-MP 300 ISU-MP 180 ISU-MP 100 ISU-MP 60 Clean ISU
^ Csoil (mg MP / g contaminated soil)
400
300
200
100
0 124
128 (oC)
0
132
Fig. 3. DSC endotherms of clean and MP-contaminated ISU soil compared with pure MP.
Fig. 5 shows a plot of Ĉsoil as a function of the observed enthalpy of melting for MP on the CSU, ISU, TG and PP soils. In most cases, the error bars representing the standard deviation are smaller than the data symbols. Error bars relative to concentration are only shown for two values of Ĉsoil for the CSU and TG soils because these were the only soil samples for which a sufficient number of replicates was obtained for meaningful statistics. The saturation concentration of MP in CSU, ISU, TG and PP are reported in Table 1. The average enthalpy of fusion of MP obtained by averaging the enthalpy of fusion calculated as specified in Eq. (5) from the slope of each line shown in Fig. 5 was estimated to be 172.5 ± 6.9 J g−1. This value is about 2% larger than the average enthalpy of fusion measured for the pure compound (168.5 ± 5.0). The saturation concentration of CP was only determined for the CSU and ISU soils. Fig. 6 shows the results. From the intercept of each line, Ssoil in CSU and ISU soils was estimated to be 91 and 58 mg CP per g of clean soil, respectively. These values are compared with the measurements performed on MP-contaminated soils reported in Table 1. The CP saturation concentration differences observed in the freshly contaminated CSU and ISU soils (i.e., soils contaminated less than 1 month prior to the measurements) seems to disappear when the CSU and ISU soils have aged approximately 5.5 years after CP contamination (Fig. 6). The saturation concentration of CP in both aged soils determined as the intercept of a straight line going through
Heat Flow
Pure CP CSU-CP 250 CSU-CP 200
Clean CSU 105
110
115
40
60
Fig. 5. Determinations of the observed enthalpy of melting for MP plotted as a function of Ĉsoil for four contaminated soils: CSU ( , Ĉsoil = 5.1 ΔHobs + 78.4, R2 = 1.00), ISU ( , Ĉsoil = 5.9 ΔHobs + 33.0, R2 = 0.99), TG ( , Ĉsoil = 5.5 ΔHobs + 53.3, R2 = 1.00), and PP ( , Ĉsoil = 5.6 ΔHobs + 29.6).
the single data point obtained for each soil with the slopes obtained from the freshly contaminated soils, was estimated to be 85 and 96 mg g− 1 for the CSU and ISU soils, respectively. The enthalpy of fusion of CP estimated from the slope of the lines for the CSU and ISU soils was 169.4 ± 5.0 J g− 1, which is larger than that measured for pure CP (151.4 ± 4.1 J g− 1) by almost 12%. 4. Discussion As shown in Table 1, the soil saturation results determined from SRM uptake and DSC were in close agreement. For both the CSU and ISU soils, the soil saturation was larger for CP than MP. However, the difference was small for the CSU soil and approximately a factor of 1.7 for the ISU soil. For comparison, the ratio of the saturation concentrations of CP and MP in octanol is 1.5 (see Table 2), where the octanol saturation (Soct) is estimated as the product of the octanol-
300
200
100
0 0
CSU-CP 100
100
20
ΔHobs (J / g contaminated soil)
^ Csoil (mg CP / g contaminated soil)
120
120
125
(oC) Fig. 4. DSC endotherms of clean and CP-contaminated CSU soil compared with pure CP.
10
20
30
40
ΔHobs (J / g contaminated soil) Fig. 6. DSC determinations of the observed enthalpy of melting for CP plotted as a function of Ĉsoil for freshly contaminated soils: CSU ( , Ĉsoil = 5.3 ΔHobs + 83.0, R2 = 0.96), ISU ( , Ĉsoil = 5.7 ΔHobs + 54.6, R2 = 0.99). Also shown are measurements from samples that were aged for approximately 5.5 years: CSU ( ) and ISU ( ) soils. The dashed lines extrapolate the results obtained from the aged CSU and ISU soils to ΔHobs = 0 based on a slope that is the average of the slopes for the freshly contaminated ISU and CSU soils. From this, Ŝsoil is estimated to be 85.0 and 96.1 mg g− 1 for the CSU and ISU, respectively.
6106
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
Table 2 Estimated saturation concentrations in octanol (Soct), in soil organic carbon (Soc), and on the soil surface (Mσ) and the corresponding estimates of the area occupied per molecule (Σ) at surface saturation compared with estimated planar area of a molecule. logKow
Soct (mg/mL)a
Soc (mg g− 1)b
Mσ (mg m− 2)b
Σ (Å2 molecule− 1)c
CP MP Ratio CP/MP
1.60 1.96
340 230 1.5
1140 660 1.7
2.15 2.17 1.0
9.19 11.6 0.79
Predicted Saturation Concentration (mg / g clean soil)
Chemical
100
a Saturation concentration of MP or CP in octanol (Soct) was estimated as the product of the water saturation and Kow. b Soc and Mσ were determined for each chemical by regressing Ssoil values for all but the PP soil to Eq. (1). c Σ = MW / (NA Mσ) is the average surface area occupied by a molecule at the surface saturation Mσ where MW is the molecular weight of the contaminant, and NA is Avogadro's number.
water partition coefficient (Kow) and the water solubility (i.e., 8.5 mg mL− 1 for CP and 2.5 mg mL− 1 for MP (PhysProp Database)). Correlation of soil saturation to octanol saturation would be expected if soil saturation arose primarily from partitioning into the soil organic matter. The absence of correlation between soil and octanol saturation limits for CP and MP in the CSU soil while a correlation was observed in ISU soil suggests that saturation is controlled by different factors for the two soils. Significantly, the saturation concentrations of both CP and MP were largest in the soil with the largest surface area (CSU), even though it has the smallest organic carbon content (foc = 0.015) of all the soils studied. Overall, the saturation concentrations of MP varied by less than 2.5 despite a 30-fold variation in the organic carbon content of the different soils. It is evident that soil organic carbon alone does not explain the measured soil saturation values. Surface area also appears to contribute. For example, the saturation value of MP in the ISU, TG and PP soils varied by a factor of 1.6, which correlates with the 1.7-fold variation in surface area, but not the 12-fold variation in foc for these three soils. Also, soil surface areas for the ISU, TG and PP soils are smaller than the surface area of the CSU soil by at least a factor of 5, which, as shown below, can explain the larger soil saturation values in the CSU soil. Overall, the soil saturation values appear to be related to both organic carbon and surface area, as suggested by Eq. (1), except for the saturation concentration measured in PP, which is inconsistently low for a soil with such a large foc. Assuming that the absorption into the organic matter and adsorption on the mineral surfaces were similar for all soils except the PP, Soc and Mσ were determined by regressing the soil saturation values (Ssoil) to Eq. (1) for MP and CP, respectively. As shown in Fig. 7, the resulting equation provides an excellent prediction of the soil saturation concentrations determined in this study. The quantitative results, listed in Table 2, indicate that Soc for CP is 1.7-fold larger than for MP, which is similar to the ratio of 1.5 between the estimated octanol saturation concentrations of CP and MP. Also, the saturated surface concentration Mσ is essentially the same for CP and MP as expected for molecules of similar size. The average surface areas occupied by a molecule at Mσ are 9.2 and 11.6 Å2 molecule−1 for CP and MP, respectively. Notably, this is smaller than the area that a nitrogen molecule (16.2 Å2) or benzene molecule in its perpendicular orientation (25.0 Å2) would occupy assuming uniform monolayer distribution on the soil surface (Chiou et al., 1990; Bambrough et al., 1998). This suggests that (i) the surface saturation concentration estimated for these molecules results from a larger than monolayer coating of the soil surface, (ii) these molecules can reach surface areas that are not accessed in the nitrogen-BET measurements, or (iii) the orientation of the adsorbed molecules allows each of them to occupy a smaller area than nitrogen or benzene molecules. The estimated fraction of the soil saturation concentration that is due to the surface area is listed in Table 3 for the four soils studied.
CSU 80
60
TG
ISU
ISU
40
20
0 0
20
40
60
80
100
Measured Saturation Concentration (mg / g clean soil) Fig. 7. Predicted versus measured saturation values for MP (circles) in ISU, CSU and TG soils, and for CP (triangles) in ISU and CSU soils; the line represents the identity.
According to the calculations summarized in Tables 2 and 3, surface adsorption is responsible for more than 80% of the saturation concentration on the CSU soil but less than 37% for the other soils, including the PP. The estimate for PP was calculated assuming that Soc and Mσ were the same as for the other soils, but that the organic carbon in the PP is not all available. By this calculation, 11% of the actual foc of 0.45 was available for chemical absorption. This reduced sorption capacity is probably related to the fact that PP is dramatically different from the other soils in that it is virtually pure organic matter. Therefore, the organic carbon is not merely a surface coating on inorganic particles as is often the case in more mineral soils. Thus, the association between the organic carbon and contaminant molecules necessarily involves diffusion into the particle interiors, which is expected to be a very slow process. There is also the likelihood of dramatically different sorption capacities for organic carbon in the interior of particles. The latter scenario could be due, for example, to intra-particle associations of organic functional groups that alter the sorption capacity. Another possible factor contributing to this reduced sorption is that the organic matter in peat is in an earlier stage of decomposition than the organic carbon in the other soils. Less decomposition means a less advanced stage of humification and the likely presence of residual cellulosic particles that, because of their polar nature, would almost certainly have a significantly lower capacity for hydrophobic sorption of molecules such as MP and CP. Separate studies of dermal absorption from CSU and ISU soils contaminated with MP using acetone (Deglin, 2007) are consistent with the measured saturation concentrations determined in this
Table 3 Fraction of the soil saturation concentration associated with the surface area for MP and CP on the soils studieda. Soil
MP
CP
CSU ISU TG PPb
0.88 0.37 0.21 0.21
0.81 0.25 NDc ND
a
Calculated as (σsoil Mσ)/Ssoil for each soil and chemical. Calculated value for MP in PP was derived assuming that Mσ = 2.17 mg m− 2 and Soc = 660 mg mL− 1 (which are the same as for the other soils), but that only part of the organic carbon was available for absorption. The available oc (foc,avail) was estimated to be 0.049, which is about 11% of the actual foc = 0.45. c Not determined. b
S.E. Déglin et al. / Science of the Total Environment 408 (2010) 6100–6107
study: dermal absorption of MP increased proportionally to soil concentration when the contamination level was less than saturation and remained constant for contamination levels above the saturation concentration. Based on these results as well as thermodynamic arguments, the soil saturation limit of chemicals that absorb through the skin is an important parameter for the exposure assessments involving contaminated soils. In this study, saturation concentrations varied with soil surface area in addition to the amount of soil organic matter. Confirmation that these observations are also observed on different soils contaminated with different chemicals is still needed. However, it is not clear whether this will be true for contaminated soils at hazardous waste sites, which would have undergone exposure to water from rainfall or other sources over an extended period of time. It is possible that surface area is more important for soils that have been contaminated with a volatile solvent like acetone, which also has water solubility and may, therefore, increase the surface area available for chemical sorption by removing adsorbed water from the surface. It is important to determine if this is the case, since most soils used in dermal absorption studies have been contaminated in this way (Kissel et al., 2007; Spalt et al., 2009). It is evident that the understanding and meaningful use of dermal absorption measurements from contaminated soil is more complicated than has been assumed in risk assessment practice. Extrapolations from one soil to another and from one concentration to another will most probably depend on the soil saturation. At present, the soil characteristics that affect soil saturation for actual contaminated soils are not clear, nor are the effects of variations in methods used for creating contaminated soils used in dermal absorption studies. Once these two issues are clarified, then protocols can be specified for preparing contaminated soils used in dermal absorption studies and for estimating soil saturation concentrations for use in risk assessment. Therefore, there is a need to understand how soil saturation varies with soil characteristics such as organic carbon content, surface area as well as the chemistry and physical condition of the organic and mineral phases. It is also important to determine how solvents commonly used to deliver contaminants to soils affect the capacity of the soils to absorb these compounds. Acknowledgments This work was supported in part by the U.S. Environmental Protection Agency (EPA) under agreement R830131-01. No endorsement from the EPA should be inferred.
6107
References Bambrough CM, Slade RCT, Williams RT, Burkett SL, Sims SD, Mann S. Sorption of nitrogen, water vapor, and benzene by a phenyl-modified MCM-41 sorbent. J Colloid Interface Sci 1998;201:220–2. Boethling RS, Mackay D. Handbook of Property Estimation Methods for Chemicals. Boca Raton: Lewis Publishers; 2000. 481 pp. Bunge AL, Parks JM. Soil contamination: theoretical descriptions. In: Roberts MS, Walters KA, editors. Dermal Absorption and Toxicity Assessment. New York, NY: Marcel Dekker; 1998. p. 669–96. Chiou TC, Leet J-F, Boyd SA. The surface area of soil organic matter. Environ Sci Technol 1990;24:1164–6. Choate LM. The effect of variations in soil organic matter with particle size on organic contaminant sorption and its relationship to dermal exposure. PhD thesis. Colorado School of Mines, Golden, CO, 2002. Choate LM, Ranville JF, Bunge AL, Macalady DL. Dermally adhered soil: 1. Amount and particle-size distribution. IEAM 2006;2:375–84. Deglin SE. Dermal Absorption of Low Volatility Chemicals from Soils. PhD Thesis. Colorado School of Mines, Golden, Colorado, 2007. Duff RM, Kissel JC. Effect of soil loading on dermal absorption efficiency from contaminated soil. J Toxicol Environ Health 1996;48:93-106. Goss KU, Buschmann J, Schwarzenbach RP. Adsorption of organic vapors to air-dry soils: model predictions and experimental validation. Environ Sci Technol 2004;38: 3667–73. Kissel JC, Richter KY, Fenske RA. Field measurement of dermal soil loading attributable to various activities: implications for exposure assessment. Risk Anal 1996;16: 115–25. Kissel JC, Spalt EW, Shirai JH, Bunge AL. Dermal absorption of chemical contaminants from soil. In: Roberts MS, Walters KA, editors. Dermal Absorption and Toxicity Assessment. Second. New York: Informa Healthcare; 2007. p. 563–73. Li J, Masso JJ, Guertin JA. Prediction of drug solubility in an acrylate adhesive based on the drug-polymer interaction parameter and drug solubility in acetonitrile. J Controlled Release 2002;83:211–21. Lyman WJ, Reehl WF, Rosenblatt DH. Handbook of Chemical Property Estimation Methods: Environmental Behaviour of Organic Compounds. Washington, DC: American Chemical Society; 1990. Manzo RH, Ahumda A. Effects of solvent medium on solubility. V: Enthalpic and entropic contributions to the free energy changes of di-substituted benzene derivatives in ethanol-water and ethanol-cyclohexane mixtures. J Pharm Sci 1990;79:1109–15. Montgomery DC, Runger GC. Applied Statistics and Probability for Engineers. New York: John Wiley & Sons, Inc; 2003. 706 pp. PhysProp Database. Syracuse Research Corporation. http://www.syrres.com/esc/ physdemo.htm. Accessed Aug. 18, 2010. Shih Y-H, Wu S-C. Distinctive sorption mechanisms of soil organic matter and mineral components as elucidated by organic vapor uptake kinetics. Environ Toxicol Chem 2005:2827–32. Spalt EW, Kissel JC, Shirai JH, Bunge AL. Dermal absorption of environmental contaminants from soil and sediment: a critical review. J Expo Sci Environ Epidemiol 2009;19:119–48. Theeuwes F, Hussain A, Higuchi T. Quantitative analytical method for determination of drugs dispersed in polymers using differential scanning calorimetry. J Pharm Sci 1974;63:427–9. US EPA. Percutaneous absorption of 2, 3, 7, 8-tetrachlorodibenzo-p-dioxin (TCDD) and 3,3',4,4'-tetrachlorobiphenyl (TCB) applied in soil. OHEA-E-453. Washington, DC, December 1991. US EPA. Dermal Exposure Assessment: Principles and Applications. EPA/600/891/011B. Washington, DC, January 1992, 389 pp.