Environ. Sci. Technol. 2003, 37, 2883-2888
Composition and Structure of Agents Responsible for Development of Water Repellency in Soils following Oil Contamination MARINA LITVINA, TIONA R. TODORUK, AND COOPER H. LANGFORD* Department of Chemistry, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4
Soil from the Ellerslie site of experimental oil contamination in Alberta developed water repellency some years after initial remediation. The water-repellent soils were compared to clean soils and contaminated but wettable soils by solidstate nuclear magnetic resonance (NMR). The effects of extraction with CH2Cl2 (for petroleum hydrocarbons), NaOH (for natural organic matter), and 2-propanol/ammonia (IPA/NH3) on wettability were evaluated by the molarity of the ethanol droplet (MED) test. Soil extracts and whole soils, after extraction, were examined using NMR and Fourier transform infrared spectroscopy (FTIR). On the basis of the structure-MED correlations, a model of a thinlayer natural organic matter-petroleum products complex formed under strong drying conditions is proposed to account for the development of water repellency. Studies of two similar soils from accidental oil spills are supportive.
Introduction Water repellency is a phenomenon that develops both naturally (1-3) and in petroleum-contaminated soils (1-3). It is the latter case that is most problematic because of the seemingly random nature of its appearance. Specifically, it can arise following the seemingly successful remediation of the site using a variety of methods. Roy and McGill (1, 2) have proposed that the phenomenon arises from residual contaminants. It is postulated that these contaminants extend from the surface of soil particles following a rearrangement to expose hydrophobic moieties. This approach however fails to identify the specific agents that are causative and does not consider the role of the natural organic matter (NOM) that is well-known to interact with organic contaminants. The use of nuclear magnetic resonance (NMR) to elucidate to the structurally significant components provides an approach to this problem. The most significant samples studied are the series of Ellerslie (ELL) soils, which are collected from a site experimentally contaminated in 1973. The phenomenon of water repellency developed some time following an apparently successful remediation of the site. The patches that developed the phenomenon were those most highly contaminated (4). A clean, uncontaminated soil (ELL-PW), contaminated but wettable soil (ELL-CW), and a waterrepellent soil (ELL-NW) from this site are available for study. Thus, it may be possible to identify structures that can negatively impact soil wettability. * Corresponding author phone: (403)220-3228; fax: (403)289-0344; e-mail:
[email protected]. 10.1021/es026296l CCC: $25.00 Published on Web 05/31/2003
2003 American Chemical Society
NMR (5, 6) is a powerful tool for characterization of soil components. Specifically, humic and fulvic acids and whole soils have been studied (7). It provides a nondestructive method of sample analysis. In 1981, Wilson et al. reported spectra of whole soils obtained using 13C cross polarizationmagic angle spinning (CP-MAS) NMR (8). This solid-state technique quickly became the standard. Since then, the developments in CP-MAS NMR spectroscopy have made it a major tool for investigations of soil organic matter (7, 9-13). NMR methods are used here for structural investigation of water-repellent soils that were previously contaminated with petroleum and subsequently remediated. Soil organic matter (SOM) is a structurally diverse, heterogeneous mixture thought to be composed of biogeopolymers. The “polymers” themselves have a wide range of apparent molecular weights. SOM is often strongly bound to mineral particles. Solid-state NMR on intact, whole soil samples is not always successful for this reason. Good spectra were obtained for all samples considered in this study. 13C CP-MAS NMR is useful for the comparison of functional groups in related samples without requiring the use of more complicated and time-consuming methods. Errors arising from differences in the Hartmann-Hahn match and variable relaxation times can cancel each other in a comparative analysis of samples. To reduce the effect of dipolar-dipolar interactions between protons and between 13C and 1H nuclei high power proton decoupling was used. The quantitative reliability of 13C CP-MAS analysis of SOM and whole soils has been discussed by several authors (1215). There are at least three factors. First, carbon atoms in different functional groups have varying efficiencies of crosspolarization. These variations arise because of nonuniform cross-polarization and variable proton spin-lattice relaxation time (T1) in heterogeneous organic matter. Usually this results in carbon atoms that are in close proximity to protons gaining energy faster. Second, carbon atoms in close proximity to paramagnetic centers (such as Fe3+) may not be observed because proton relaxation is rapid, minimizing CP efficiency. If distribution of paramagnetic species is homogeneous throughout the sample, however, quantification is not affected. Third, some types of carbon atom signals can produce side bands that can interfere with other signals. These problems can be adequately minimized to allow comparative analysis of related samples using proper experimental parameters (12). For reasonable quantitative representation of solid humic substances, the sample should be spun at a rate exceeding 7 MHz in a 300 MHz spectrometer (15). The present study presents CP-MAS spectra that allow quantitative comparison between functional groups in similar samples but do not necessarily provide absolute values of functional group distribution. The use of solid-state 13C CP-MAS technique allows for the collection of well-resolved 13C spectra of whole soils and soil organic matter with sufficient signal-to-noise ratio in reasonable experimental time frames. The time required to record good quality spectra is significant for extended studies. Spectra reported here required 29 h for acquisition. The purposes of this study are (i) to compare the differences in functional group distribution between a series of clean, contaminated, but wettable and water repellent soils. The most important series of soils to be studied are the ELL soils. (ii) to attempt identification of structural components that are associated with the development of water repellency using 13C CP-MAS NMR. (iii) to attempt to elaborate upon the proposed mechanisms for the development of water repellency using extraction work. VOL. 37, NO. 13, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2883
TABLE 1. Total Carbon, Hydrogen, Nitrogen, Iron Content (% w/w), pH, MED Values, and Exchangeable Cations of the Soils Examined soil
C
H
ELL-NW ELL-CW ELL-PW DEV-NW DEV-CW STE-NW STE-CW
5.6 4.8 4.7 4.5 5.7 2.3 1.6
0.5 0.8 0.8 0.5 0.6
N
Fe
pH
MED value
0.4 1.3 5.0 3.8 ( 0.1 0.4 5.6 0.0 ( 0.1 0.4 5.4 0.2 1.4 6.4 4.2 ( 0.1 0.4 7.2 0.0 ( 0.1 0.5 5.1 4.4 ( 0.1 5.7 0.0 ( 0.1
Ca2+ Mg2+ Na2+ K+ 27.7 9.0 32.7 12.2
0.7 0.7
0.3 0.4
31.0 11.8 55.3 21.2 5.3 2.4 6.5 1.6
1.0 0.7 0.7 0.6
1.6 2.0 0.3 0.2
Experimental Section Soil Samples. The main soil samples used in this study were collected from the Ellerslie site in Alberta, 18 km south of Edmonton (24°51′25′′ NE, 4° W). The site was the subject of an experimental oil contamination in 1973 (16). Three samples from this site are reported on here. Ellerslie pristine wettable (ELL-PW) is a clean soil from outside the boundaries of the experimental contamination. Ellerslie control wettable (ELL-CW) is a wettable soil contaminated in a manner similar to the water-repellent soil. Ellerslie water-repellent soil (ELLNW) is a contaminated, water-repellent soil that developed the phenomenon some years following remediation of the site. The detailed history of these samples is given in the accompanying paper (17). The soil is an Eluviated Black Chernozem of the Malmo silty clay loam series (1). These samples have been studied extensively by Roy and McGill (1, 2). Extensive podological characterization of the soils is found in refs 1-3 and the several earlier studies sited therein. The notation and nomenclature used for the samples in this study is consistent with that used by Roy and McGill in order to allow easy comparison of studies. The results obtained with these samples are supplemented by samples from oil spill sites in Devon, AB (DEV), and Stettler, AB (STE), where NW and CW samples could be obtained. Less information about the history of these sites is available than for the ELL spill site. The DEV soil is a Gleyed Eluviated Black Chernozemic soil developed on alluvial parent material. Patches of waterrepellent soil were first reported over 30 yr ago (3). The high C content (Table 1) of the CW soil reflects consequences of cultivation and recent fertilization on the site. The STE site has a Gleyed Black Chernozem soil developed on an alluvial parent. The site has rolling topography. Contamination occurred in 1970 (3). Substantial information on these two sites is presented in the Supporting Information. Soil Samples Characterization. Total carbon, hydrogen, and nitrogen content were determined using a combustion elemental analyzer (440 CHN-O/S elemental analyzer, Exeter Analytics, Chelmsford, MA) on air-dried samples. Results are shown in Table 1. Iron content and exchangeable cations were obtained using an inductively coupled plasma (ICP) atomic emission spectrometer Atom/Scan 16/25 (Thermo Jarrell Ash Corp., Franklin, MA). For ICP analysis of iron, 1.000 g of air-dried soil was boiled in 10 mL of concentrated HCl for 30 min. The slurry was filtered, diluted to 100.0 mL, and then analyzed. Exchangeable cations were determined by saturating soil cation-exchange sites with NH4+ by 1 M NH4+-OOCCH3 (pH 7). The solution obtained was analyzed for Ca2+, Mg2+, K+, and Na+. In all cases, 18 MΩ of water (Barnstead Nanopure system) was used. Soil pH was measured in water at a soil:water ratio of 1:2 after 30 min of mixing. A Fisher Accumet meter was used. Results are presented in Table 1. Empirical Measurement of Soil Water Repellency (MED Test. Soil water repellency assessment was performed on the set of water-repellent and corresponding CW soils using 2884
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 13, 2003
the widely exploited Molarity of Ethanol Droplet (MED) test (18). The practical field reference for wettability is the MED value. The MED test is conceptually based on the reduction of the surface tension of water in ethanol solutions but is essentially an “empirical” measure validated by correlation with field behavior. Ethanol solutions with concentrations ranging from 0 to 5 M increasing in 0.2 M increments were prepared. HPLC-grade ethanol, purchased from Aldrich, and distilled water with resistance of 18 MΩ (Barnstead Nanopure system) were used. To assess soil water, repellency droplets of ethanol solutions of different concentrations were placed on a smoothed soil surface. The molarity of the droplets of the lowest ethanol concentration that are completely absorbed within 10 s is considered to be MED index of the soil. Soils with 0 < MED index 2.2 M are severely water repellent (18). The test was performed on soil samples airdried to constant mass. At higher soil moisture content (fieldmoist soil), cohesive forces between the water initially present in the soil and the water placed on the surface make MED values unreliable. It has been shown that the MED indices are essentially the same for air-dried and oven-dried soils (3). MED results for NW and CW soils are presented in Table 1. None of the CW soils has an MED significantly different from zero. All identified NW soils have values near 4. 13C CP-MAS NMR. Samples were tightly packed into a cylindrical 4/18 mm zirconia rotor with Kel-F caps. Solidstate NMR measurements were performed on a Bruker AMX300 spectrometer with a BL4 probe, operating at 300 MHz for 1H and 75.5 MHz for 13C. Solid-state 13C CP-MAS spectra were obtained (15) with 80 000-100 000 scans with a contact time of 1 ms and a spinning rate of 8 kHz. Calibration used the [13C]glycine ketonic signal at 176.32 ppm. Processing of spectra with Bruker 1D WIN NMR software included phasing and background correction (15). Solid-state spectra were processed with 100 Hz line broadening. The wide range of chemical structures in soil organic matter result in broad overlapping bands in the solid-state NMR spectra. Because of this, the intensity is integrated within a region corresponding to similar types of carbon atoms. Chemical shift regions were assigned to carbon types as follows (19): 0-50 ppm were assigned to aliphatic carbon atoms; 50-96 ppm were assigned to carbohydrate-like carbon atoms; 96-108 ppm were assigned to O-C-O linkages; 108162 ppm were assigned to aromatic carbon atoms; 162-190 ppm were assigned to carboxyl carbon atoms; 190-220 ppm were assigned to ketonic carbon atoms. Liquid-State 1H NMR. Liquid-state NMR spectra were acquired on a Bruker AMX-300 spectrometer with a BBI5 probe, operating at 300 MHz. Glass NMR tubes with 5 mm diameter were used. For NMR analysis, samples of Redwater crude oil (RW) and dried material extracted from soil were dissolved in deuterated chloroform, CDCl3 (purchased from Aldrich). Chemical shift assignments for various types of protons were made as follows (19, 20): 0.4-1.0 ppm were assigned as terminal methyl protons of methylene chains; 1.0-2.0 ppm were assigned as protons on aliphatic carbons that are two or more carbons removed from an aromatic ring; 2.0-4.2 ppm were assigned as protons bound to aliphatic carbon (methyl and methylene groups) that are assigned to an aromatic ring or electronegative functional group; and 6.0-9.0 ppm were assigned as aromatic protons. Extractions and Treatment with Solvents. For extractions under Soxhlet reflux conditions, an air-dried sample of soil (20-30 g) in a cellulose thimble was placed into a Soxhlet apparatus with a capacity of 50 mL and extracted with 150 mL of reagent-grade solvent for 6 h. The approximate extraction rate was 20 cycles per hour. It was possible to achieve this rate due to the small apparatus capacity. The
apparatus capacity is defined as the maximum solvent volume accumulated in the sample compartment of an upper vessel before the solvent returns to the bottom heated vessel with bulk solvent. Poly(tetrafluoroethylene) boiling chips (Teflon) were placed into the heated vessel of the Soxhlet apparatus. The quantity of extracted material was determined gravimetrically. Solvent with extracted material was concentrated on a rotary evaporator and placed into preweighted aluminum dishes. After being air-dryed to constant mass in the fume hood, the mass of extracted material was obtained and its concentration calculated. The extracted soil samples were air-dried for 7 d, and then soil water repellency was assessed using the MED test. Chloroform and 2-propanol/ammonia (IPA/NH3) treatment of soil consisted of saturating a sample with a corresponding solvent, removing the solvent from the particulate, and vortex spinning until the solvent evaporated. The treated samples were air-dried for several days before soil water repellency was assessed using the MED test. Fourier Transform Infrared (FTIR) Spectroscopy of Soil Extracts. FTIR spectra were obtained on a Nicolet Nexus470 FTIR spectrometer. The extract from soil material was air-dried, quantitated, and then redissolved in a small volume of CH2Cl2. This solution was placed dropwise on a 4 mm thick KBr window and solvent evaporated under a flow of N2. After complete solvent evaporation, the KBr window was covered with the second window, and FTIR analysis of the film was performed. All organic solvent solutions were dried with a drying agent (anhydrous Na2SO4) before FTIR analysis. Nonpolar fractions of soil extracts were also analyzed by FTIR spectroscopy. To obtain these fractions, CH2Cl2 soil extracts were sorbed on a cartridge (Varian Bond Elute PCB, Varian) that contained a polar sorbent. The nonpolar fraction of the extract was eluted from the cartridge using several milliliters of CH2Cl2, while a polar fraction was left in the cartridge. Results are reported in the Supporting Information. The IPA/NH3 soil extracts were air-dried, redissolved in CH2Cl2, filtered, and separated on the cartridge as described above. Band assignments are presented below where the spectra are discussed. Gas Chromatography (GC) Analysis. CH2Cl2 extracts and a sample of Redwater oil, which was the type of oil used in the experimental contamination, were analyzed by conventional GC analysis of oils at Enviro-Test Laboratory, Calgary, AB.
Results and Discussion Soil Composition. To obtain background information, some chemical properties of the selected soils were characterized. These complement data from refs 1-3. Total carbon content tends to be higher for the NW samples than the CW samples. This is probably a result of the higher content of the oil and diagenetic products from the contamination event. This is confirmed by the darker color of the extracts obtained in “petroleum hydrocarbons” extractant CH2Cl2 solution. The concentration of iron in soil was also quantified since it is commonly a factor in line-broadening in solid-state NMR spectroscopy. Quantification problems are generally observed in soils with a C/Fe ratio
