Retention of Nonionic OrganicCompounds on Thermally TreatedSoilsH A I J U N Z H A N G , † S U F A N G Z H A O , ‡ , §

Y I N G Y U , ‡ Y U W E N N I , † X I A N B O L U , †

Y U Z E N G T I A N , † A N D J I P I N G C H E N * , †

Dalian Institute of Chemical Physics, Chinese Academy ofSciences, 457 Zhongshan Road, Dalian 116023, China, Collegeof Environmental Science and Engineering, Dalian MaritimeUniversity, 1 Linghai Road, Dalian 116026, China, andNational Marine Environmental Monitoring Center,42 Linghe Street, Dalian 116023, China

Received June 3, 2009. Revised manuscript received March29, 2010. Accepted April 6, 2010.

To achieve a higher efficiency of thermal remediation of soilcontaminatedwithorganiccompounds, theretentionmechanismsof organic compounds on thermally treated soil need to beunderstood adequately. In this study, a soil-column gaschromatography approach was developed to determine the soil-air partition coefficients (KSA) at 300 °C for a diverse set ofnonionic organic compounds bearing many different functionalgroups; and the retention mechanisms of these organiccompounds on two typical soils, isohumisols and ferralisols,were characterized using a polyparameter linear free energyrelationship (pp-LFER). The KSA values (mL g-1) of typical volatileorganic compounds (VOCs) with lower boiling points were<1.5 and in some cases even below 1.0, suggesting the rapidremoval of VOCs from soils at 300 °C. Moreover, the KSA

values were found to be a strong function of the soil-columntemperature T (K), and be almost independent of the carrier-gas flow rate. Significant differences in molecular interactionswere noted among various soil-solute pairs. The relativecontributions of nonspecific van der Waals forces to theretention of test polar solutes were higher on isohumisols thanon ferralisols. In contrast to the reported pp-LFER models fornatural soils and soil components at normal environmentaltemperatures, our results suggest that elevated temperatureremarkably reduces H-bond interactions between polar organiccompounds and the soil matrix, thus allowing accelerateddesorption of polar organic compounds from soils during thermaltreatment.

IntroductionThermal desorption, also known as thermal volatilizationand soil roasting, is a reliable method for physically reha-bilitating soil contaminated with organic pollutants (1, 2).By heating soil to temperatures of 150-650 °C organiccontaminants will vaporize or turn into gas and be separatedfrom the soil. Some studies aimed at improving decontami-nation efficiencies have focused on the thermodynamic and

mass-transfer processes that occur during the volatilizationand movement of organic contaminants through thermallytreated soils (3–13).

Temperature is a key factor in the thermal decontamina-tion of soil (3, 5). As the soil temperature increases, theequilibrium partition shifts toward the gas phase for allcompounds in the soil, and desorption from the soil isenhanced (6, 7). However, the extent of desorption highlydepends on the characteristics of both the soil and solute.Lighty et al. (4–6) reported the results of their parametricstudies on the desorption of organic contaminants from soilsin a bench-scale particle characterization reactor and a bedcharacterization reactor. They found that the rate of con-taminant evolution from various materials (sand, glass beads,clay, and peat) was a strong function of the sorbent’sproperties, with contaminants removed from nonporousmaterials at a faster rate. Furthermore, thermal desorptionphenomena were shown to vary significantly among differentsolid-organic vapor pairs (8, 9). To date, there have beenonly a few literatures describing the gas-phase transferprocess of organic pollutants in heated soils. Wu et al. (10)used chromatographic response analysis to study masstransfer mechanisms of organic contaminants in heated soilcolumns. Their results suggested that axial dispersion coef-ficients, intraparticle diffusion coefficients, and soil-gasequilibrium constants determined the concentration profilesof organic pollutants in the system. Moreover, several simplekinetic models such as first-order model and 1-D mass-transport model have been developed to simulate the removalof organic contaminants from thermally treated soils (11–13).

These studies established a fundamental appreciation ofthe physicochemical processes involved in the thermaldesorption of organic pollutants from soils or their com-ponents, but the important impact factors, molecularproperties of the pollutants, and chemical properties ofdifferent soils, have not yet been sufficiently considered. Atthe molecular level, the retention of nonionic organiccompounds mainly results from nonspecific van der Waalsforces, including dispersion forces (London forces), dipole-dipole forces (Keesom forces), and dipole-induced dipoleforces (Debye forces), as well as specific H-bond forces (14).During the past decade, some researchers have focused onthe molecular interactions that occur during the vaporadsorption of organic compounds onto natural soils and soilcomponents at normal environmental temperatures with thegoal of understanding the fate of volatile and semivolatileorganic compounds in terrestrial environments (15–24).However, there remains a lack of knowledge regarding therelative strengths of the molecular interactions governingthe gas-phase retention of organic compounds onto naturalsoils at elevated temperatures. To fill these gaps in ourknowledge requires a consistent data set for diverse chemicalsand an appropriate model to describe the relevant molecularinteractions.

The linear free energy relationship (LFER) approach hasbeen widely adopted to interpret or predict the equilibriumpartitioning of neutral organic compounds between any twophases (14, 25). In contrast to one-parameter LFER, poly-parameter LFER (pp-LFER) is based on a concept thatconsiders all significant interactions involved in partitioningby separate parameters. They allow for modeling thecomplete compound variability by a single equation, andthey also provide the possibility to evaluate and predict thevariability in the sorption characteristics of natural phases(25). Nguyen et al. (26) critically reviewed the pp-LFER theoryand presented a pp-LFER that allows interpretation and

* Corresponding author phone/fax: +86 411 84379562; e-mail:chenjp@dicp.ac.cn.

† Dalian Institute of Chemical Physics.‡ Dalian Maritime University.§ National Marine Environmental Monitoring Center.

Environ. Sci. Technol. 2010, 44, 3677–3682

10.1021/es9034705 2010 American Chemical Society VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3677

Published on Web 04/19/2010

accurate estimation of the equilibrium partitioning of organicpollutants between water and the natural organic matter insoils and sediments. Moreover, the pp-LFER approach wasalso applied to characterize the gas-phase sorption propertiesof natural soils (16, 20), organic matters (15, 21–23), minerals(17–19, 27), salts (17–19), and aerosol particles (24). Here, wereport the development of a soil-column gas chromatographymethod to measure the soil-air partition coefficients (KSA)of nonionic organic pollutants with different molecularstructures at elevated temperatures. In addition, the pp-LFERapproach was used to probe the intermolecular interactionsgoverning the thermal desorption of organic pollutants fromsoils. The goals of this work were to evaluate the equilibriumpartitioning between typical nonionic organic compoundsand thermally treated soils, and to better understand thechemical details of how molecular interactions control theretention and removal of soil-bound organic pollutantsduring thermal treatment.

Materials and MethodsSoils. The two natural soils used in the experiments,isohumisols and ferralisols, were collected from the HailunAgro-Ecological Experimental Station, Heilongjiang Provinceand Yingtan Ecological Experimental Station of Red Soil,Jiangxi Province, respectively. Air-dried soil samples weregently ground and then sieved to obtain the fraction of soilparticles with sizes of 80-100 mesh. The mineral composi-tions of the soil were determined by X-ray fluorescencediffraction analysis, and the surface areas, and microporevolumes of the soil were determined by N2-BET analysisusing a Quantachrome NOVA 4000 analyzer (Quantachrome,U.S.) after the soils had been degassed in the vacuum ovenat 200 °C overnight. The total organic carbon (TOC) in soilwas analyzed with an elemental analyzer (Vario EL III). Priorto TOC analysis, the inorganic carbon in soil was first removedused 0.1 M HCl, and then the soil was dried at 105 °C in adry oven for 8 h. The cation exchange capacity (CEC) wasdetermined by extracting a soil sample with 0.1 M ammoniumacetate (pH 7.0). The chemical and physical properties ofthe two test soils are presented in Table 1. Isohumisols hada higher TOC and a higher CEC than ferralisols.

Probe Solutes. The KSA values were measured for 39nonionic organic compounds covering the following com-pound classes: alkanes, chlorinated alkanes, aliphatic alco-hols, esters, ethers, aromatic hydrocarbons, alkyl-aromaticcompounds, halogenated aromatic compounds and otheraryl derivatives. All of the compounds were found to be ofthe highest purity available, as confirmed by the absence ofimpurity peaks during reverse-phase liquid chromatographyon a C18 column. The list of these probe solutes is shownin Table 2, and their solvation parameters, with respect to

polarizability, dipolarity, molecular size and hydrogen bond-ing ability, are reported in Supporting Information (SI) TableS1.

Gas Chromatographic Measurement. KSA values weredetermined by inverse gas chromatography (IGC) with soilas the stationary phase and N2 as the carrier gas. Air-driedsoil sample with particle sizes of 80-100 mesh was directlypacked into a glass column (3 mm i.d. × 500 mm) to a heightof 200 mm under slightly negative pressure. A small amountof glass wool was packed in the both ends of soil column toprevent the movement of soil particles with the flow of carriergas. The packed soil column had a porosity of 69.1% forisohumisols and 59.9% for ferralisols, and a bulk density of0.86 g mL-1 for isohumisols and 1.12 g mL-1 for ferralisols,respectively. The packed column was installed in a Hewlett-Packard model 5890 series II gas chromatography equippedwith a flame ionization detector (GC/FID). Prior to theexperiments, the soil column was thermally stabilized at atemperature of 320 °C in a stream of nitrogen gas, in orderto remove volatile or semivolatile organic compounds fromthe soil. Nitrogen was used as the carrier gas and the solutewas injected in the splitless mode. The temperatures of boththe injection port and the detector were 280 °C, and thetemperature of soil column was maintained at 240-340 °C.The flow rate of the carrier gas was measured using a soap-film flow meter. Signals generated by flame ionizationdetector were transmitted to a personal computer, and the

TABLE 1. Characteristics of the Two Test Soils

properties isohumisols ferralisols

pH 5.8 4.9total organic carbon

(wt %) 3.10 0.35

clay minerals montmorillonite kaolinite,goethite

SiO2 (wt %) 45.8 38.6Al2O3 (wt %) 15.7 8.8Fe2O3 (wt %) 7.5 13.1CEC (mmol kg-1) 329.1 195.6N2-BET surface

area (m2 g-1) 2.67 3.30

micropore volume(mL g-1) 0.025 0.075

average porediameter (Å) 3.80 2.78

TABLE 2. Soil-Air Partition Coefficients (KSA, mL g-1) ofSelected Solutes on the Two Test Soils at 300 °C

chemicals isohumisols ferralisols

n-pentane 0.2 0.2n-hexane 0.3 0.5cyclohexane 0.3 0.4n-heptane 0.4 0.8n-octane 1.1 1.7n-nonane 1.5 2.7n-dodecane 7.6 18.4di-n-butyl ether 0.8 3.9ethyl propanoate 0.4 0.6ethyl butanoate 0.5 0.81-propanol 0.6 0.81-butanol 0.7 1.01-pentanol 1.0 1.9dichloromethane 0.3 0.1tetrachloromethane 0.5 0.51,2-dichloroethane 0.5 0.3tetrachloroethene 1.4 1.5benzene 1.0 0.6toluene 1.4 1.5ethylbenzene 1.9 2.41,3,5-trimethylbenzene 3.9 5.5n-butylbenzene 5.0 6.8chlorobenzene 1.4 1.41,2- dichlorobenzene 3.8 4.41,3-dichlorobenzene 3.5 2.71,3,5-trichlorobenzene 6.9 7.61,2,4-trichlorobenzene 7.8 5.81,2,3,4-tetrachlorobenzene 15.1 22.3bromobenzene 2.7 2.4iodobenzene 6.7 5.8nitrobenzene 3.1 3.0phenol 4.1 3.14-nitrophenol 12.8 7.84-chlorophenol 7.2 5.94-chloroaniline 8.2 6.84-methylaniline 7.5 11.6acetophenone 5.3 6.5biphenyl 41.9 55.2naphthalene 29.8 31.0

3678 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010

chromatographic data were processed on a DL 800 worksta-tion (Dalian Institute of Chemical Physics, China).

When a solute i is injected onto the soil column underisothermal conditions, its adsorption constant, KSA, can bederived from eq 1 (20):

where VR is the retention volume of solute i, VM is the elutionvolume of a nonretained solute, and Msoil is the dry mass ofsoil in the column. Methane was selected as the nonretainedsolute for VM determination. The volumes VR and VM weredetermined from the elution time of the center of gravity ofthe respective peaks (their first statistical moment) and fromthe volumetric flow rate of the carrier gas corrected for thepressure drop in the IGC system (ref 28, pp 30 and 69). Eachsolute was injected at least four times, and the standarddeviation of the calculated KSA values was lower than 5% ofthe average except n-pentane.

The pp-LFER Model and Its Calculation Method. Thegeneral pp-LFER model can be described as follows:

where capital letters represent solute descriptors. E is thesolute excess molar refraction, which reflects polarizabilitycontributions from n- or π- electrons; S is the solutedipolarity/polarizability descriptor; A is the solute H-bondacidity, which is a measure of the solute’s ability to donatea hydrogen bond; B is the solute H-bond basicity, that is, ameasure of the solute’s ability to accept a hydrogen bond;and L is the logarithm of the gas-hexadecane partitioncoefficient at 298 K, which accounts for the dispersion/cavityformation (14). As these solute descriptors refer to particularphysicochemical interactions, the coefficients in eq 2, e, s,a, b, and l, correspond to the complementary effect of thephase on these interactions. The constant c carries all factorsindependent of the compounds. In this study, all data ofsolute descriptors were taken from the literatures of Abrahamet al (29), and Poole and Poole (16). The values of log KSA

were regressed against all solvation parameters for theselected probe solutes by multiple linear regression usingSPSS 13.0 software. A stepwise method, with entry andremoval criteria at P ) 0.05, was performed. The statistics ofmultiple linear regression included estimates for regressioncoefficients, model fit, collinearity diagnostics, and Durbin-Watson for residuals.

Results and DiscussionEstimation of Equilibrium Partitioning. SIFigure S1 showsthe breakthrough curves of 1,2,4-trichlorobenzene in theisohumisol and ferralisol soil columns. As the columntemperature increased from 240 to 340 °C, the peak heightincreased, while the peak width became narrower. All peakshad a modified Gaussian shape. The asymmetry factor (at10% of peak height) of 1,2,4-trichlorobenzene at 300 °C was1.5 for isohumisols and 2.8 for ferralisols, respectively.

KSA is the key parameter for evaluating the efficiency ofthermal remediation of contaminated soil and necessary fordesigning the engineering systems. The KSA values of theprobe solutes at 300 °C are provided in Table 2. Among thetest solutes, biphenyl had the highest KSA, followed bynaphthalene. The KSA values of most of the aliphaticcompounds (straight-chain alkanes, aliphatic alcohols, di-n-butyl ether, and linear esters) were clearly higher onferralisols than on isohumisols. It should be noted that thecalculated KSA values of typical volatile organic compounds(VOCs) with lower boiling points, including n-pentane,n-hexane, cyclohexane, n-heptane, di-n-butyl ether, ethylpropanoate, ethyl butanoate, benzene, toluene, dichlo-

romethane, 1,2-dichloroethane, tetrachloromethane, tetra-chloroethene, 1-propanol, and 1-butanol, were <1.5, and evenbelow 1.0, suggesting that the removal of VOCs from the soilat 300 °C occurs very rapidly.

The retention of methane on the soil matrix would resultin underestimation of the KSA values of the VOCs. To verifywhether the retention phenomenon significantly occurredor not, the variations in the methane peak elution time (tM)with soil-column temperature and carrier-gas flow rate wereinvestigated. The results are shown in SI Figure S2. The tM

at 300 °C varied as a power function of the carrier-gas flowrate, but did not change with variations of the soil-columntemperature. At a certain carrier-gas flow rate, tM value wasalmost constant. These results suggest that methane onlysimply pass through the mobile phase space, from inlet tooutlet; and therefore the KSA values of VOCs with lower boilingpoints were not evidently underestimated.

In this study, KSA was measured by a gas chromatographymethod, which required that the adsorption kinetics wasfast enough to guarantee the local equilibrium conditionsduring the chromatographic process. Thus, it can be ruledout that the measured KSA values may to a large degreeaccount for the fast sorption, that is, sorption of solutes onthe soil surface. The absorption into soil organic matter andthe capture by soil micropores, the so-called slow sorption,may only be partly represented by the measured KSA values.The presence of slow-sorption kinetic processes in the soilcolumn accounted for the gross peak spreading and tailingobserved in our experiments, especially for the solute witha higher boiling point. When the boiling point of a solute isclose to or above the temperature of the soil column, thepeak width becomes so large that it prevents the peak’s gravitycenter from being accurately calculated. This can causedeviations as great as (20% in the determined values of KSA.

Influence of Temperature, Carrier-Gas Flow Rate, andSolute Volatility. Although numerous papers on the sorptionof organic vapors to soils have been published, only a fewhave reported a large data set on the gas-solid equilibriumpartitioning coefficients between organic pollutants andnatural soils (20, 30, 31). In our experiments, the KSA valuesdetermined at 300 °C were generally 2-3 orders of magnitudelower than those measured at normal environmental tem-peratures (20, 30, 31). The dependence of KSA on soil-columntemperature is explained in SI Figure S3. Within the tem-perature range of 240-340 °C, the KSA values of four typicalsolutes, n-octane (apolar compound), 1,3-dichlorobenzene(monopolar compound), and 1,2,4-trichlorobenzene (apolarcompound), and 1-butanol (bipolar compound), were strongfunctions of the soil-column temperature T (K) and obeyedthe van’t Hoff equation,

where R is the universal gas constant (J mol-1 K-1), ∆H0 isthe change in enthalpy of adsorption (kJ mol-1). ∆H0 wascalculated from the slope and intercept of plot of ln KSA versus1/T (SI Figure S3). The results are given in SI Table S2. Thevalues of ∆H0 for the test four solutes are all negative,indicating that the adsorption process is exothermic in nature.

The boiling point of a compound is a measure of itsvolatility. The boiling point ranges of target organic pollutantscan be used to estimate the extent to which soil thermaltreatment should be exploited. A significant logarithmiccorrelation was found between the measured KSA values oftest solutes at 300 °C and their boiling points (SI Figure S4).The correlation coefficient (r) was 0.931 on isohumisols and0.930 on ferralisols. However, the standard error (SE) in theestimate of log KSA value from boiling point was 0.220 onisohumisols and 0.227 on ferralisols. The higher SE values

KSA(mL g-1) ) (VR - VM)/Msoil (1)

log KSA ) c + eE + sS + aA + bB + lL (2)

ln KSA ) -∆HRT

+ C (3)

VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3679

suggest that KSA cannot be accurately estimated based onthe boiling points of the solutes of interest.

Furthermore, we investigated the dependence of KSA at300 °C on the flow rate of the carrier gas. The results areshown in SI Figure S5. When the carrier-gas flow rate variedin the range of 5-58 mL min-1, the variation ranges ofdetermined KSA values at 300 °C for n-octane, 1,2-chlo-robenzene and 1-pentanol were 1.0-1.1, 3.7-4.0 and 0.9-1.1on soil column of isohumisols, and 1.4-1.8, 4.2-4.6 and1.7-2.0 on soil column of ferralisols, respectively. Thevariation range of determined KSA values for a certain solutewas very narrow, and there were no significant changes ofKSA values with the variation of the carrier-gas flow rate. Thisresult suggests that the local equilibrium sorption can beachieved quickly.

Robustness of the pp-LFER Model. The regression resultsof the pp-LFER models are listed in Table 3. For the two testsoils, according to the results of F-test, all solute descriptorssignificantly influenced the dependent variable log KSA (P <0.05). Moreover, the colinearity diagnostics indicated thatall values of variance of inflation factor (VIF) were <10,suggesting there is no obvious colinearity among theexplicative variables. However, in view of the larger VIF values

(>5), the individual influences of E and S on log KSA mightnot have been accurately determined.

The two pp-LFER models for isohumisols and ferralisolsshowed a satisfactory goodness of fit between log KSA andthe solvation parameters. The fact that the F values weremuch higher than the F0.01 value (3.65) and the P values wereclose to zero implied that the regression equation was highlysignificant (SI Table S3). The standard error in the estimateof log KSA was 0.110 log units for isohumisols and 0.082 logunits for ferralisols, and the correlation coefficient was 0.984for isohumisols and 0.991 for ferralisols (SI Figure S6).

The residuals (experimental log KSA minus the calculatedvalue) were plotted against all solute descriptors and againstlog KSA (SI Figures S7 and S8) to examine the model’s validityas well as the systematic modeling errors. The correlationcoefficient between the residuals and the values of log KSA

was 0.105 for isohumisols and 0.087 for ferralisols, and theplot of the residuals versus each solute descriptor showed acorrelation coefficient of <0.001. The D-W values of the dataset for these two soils were in the range of 1.4-2.6 (SI TableS3). These results suggest that the regression residuals wereindependent of each explicative variable and occurredrandomly.

According to the above-described robust statistics, it isfeasible to model the KSA values at 300 °C for a diverse setof nonionic organic compounds bearing many differentfunctional groups by the pp-LFER, and most of differencesin log KSA (dependent variable) can be credibly ascribed tothe different solvation parameters (explicative variables).However, the model should be limited to evaluate the KSA

values of solutes that have descriptor values within the studiedranges. Linear extrapolation of the obtained pp-LFER modelsmay produce a larger deviation.

Interpretation for Molecular Interactions GoverningRetention. The pp-LFER models for the two soil types differedmoderately for the descriptor coefficients e, s, a, b, and l(Table 3). To compare the molecular interactions governingthe retention of nonionic organic compounds on these twoheated soils, the relative contributions of the interaction termsto the retention were calculated. The results are shown in SIFigure S9. The molecular interactions differed significantlyamong various soil-solute pairs. It should be noted that thedescriptors E, S and L are often ambiguous in their explana-tions of molecular interactions. The descriptor E largelyaccounts for polarizability effects that are not accounted forby the descriptor S, and the interaction reflected by the eEterm partly incorporates the dispersion effect (lL term) andthe induction effect (part of the sS term); the descriptor Lalso accounts for some polarizability/induction effects

TABLE 3. Regression Results of pp-LFER Models for the TwoTest Soils

explicativevariable coefficient std. error P valuea VIFc

isohumisolsconstant c -1.575 0.079

E 0.936 0.113 0.000b 6.197S -0.632 0.135 0.000 7.456A 0.375 0.120 0.004 1.723B 0.678 0.141 0.000 1.700L 0.396 0.023 0.000 1.874

ferralisolsconstant c -1.784 0.059

E 0.845 0.083 0.000 6.197S -1.109 0.100 0.000 7.456A 0.277 0.089 0.004 1.723B 1.411 0.104 0.000 1.700L 0.523 0.017 0.000 1.874

a Probability value of F test. b P value of <0.05 indicatesthat explicative variable significantly influenced thedependent variable log KSA. c The variance inflation factorthat measuring the impact of collinearity among theexplicative variables in the regression model.

FIGURE 1. Relative contributions of specific H-bond interactions and van der Waals forces to the total interaction strengths betweentest polar organic compounds and heated soils. The molecular interactions were calculated from the interaction terms of thepp-LFER models. van der Waals forces were the sum of eE, sS, and lL terms, and specific H-bond interactions were the sum of aAand bB terms. Soil-column temperature: 300 °C.

3680 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010

(14, 25). Therefore, it is difficult to discern the exactdistribution of polarity, dispersion effects, and inductioneffects in the coefficients of these descriptors. Moreover, theevaluation of the descriptor coefficients e and s may not yieldaccurate information because of the moderate colinearitiesof E and S with log KSA. Accordingly, to avoid overinterpre-tation of the pp-LFER model, the sum of the eE, sS, and lLterms was interpreted as representing the nonspecific vander Waals forces.

For apolar compounds, straight-chain alkanes, 1,3,5-trichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlo-robenzene, tetrachloromethane, and tetrachloroethane, aAand bB terms were both equal to zero, and thus theinteractions between apolar solute and heated soil dependedonly on the nonspecific van der Waals forces. For the test 27polar solutes, nonspecific van der Waals forces also pre-dominated in the retention strengths (Figure 1). The relativecontribution of nonspecific van der Waals forces to theretention of the test polar solutes was higher on isohumisols(in the range of 62.1-99.4%) than on ferralisols (50.5-98.7%)when soils were heated to 300 °C.

The specific H-bond interactions can be clearly explainedby the aA term and bB term of the pp-LFER model. Thedescriptor coefficient a for isohumisols (0.375 ( 0.120) wasnot obviously different from that for ferralisols (0.277(0.089)(Table 3); however, the descriptor coefficient b for ferralisols(1.411 ( 0.104) was much higher than that for isohumisols(0.678 ( 0.141). This result suggests that ferralisols have therelatively high capacity to adsorb the solute with a greaterability to accept a hydrogen bond.

Comparison with Reported or Derived pp-LFER Modelsin the Literature. During the past decade, pp-LFER modelshave been used to characterize the molecular interactionsthat take place during the vapor adsorption of organiccompounds onto soils and soil components (15–24). Thereported pp-LFER models and the pp-LFER models derivedfrom the reported data are listed in SI Table S4. Comparedwith these reported and derived pp-LFER models for naturalsoils or soil components at normal environmental conditionspreviously, our pp-LFER models for thermally treated soilsgave the values of relatively smaller magnitude for the a andb coefficients. This suggests that the vapor adsorptionmechanisms of organic compounds on the heated soilssignificantly differ from those on natural soils at normalenvironmental conditions.

As well-known, the sorption characteristics of soils dependnot only on their own chemicophysical properties (soilcomposition, pH, CEC and etc.), but also on their environ-mental conditions, such as humidity and temperature. Asshown in SI Table S4, at the conditions of normal environ-mental temperatures (15-30 °C) and relative humidity(50-100%), all of the pp-LFER models obtained from 10 kindsof humic and fulvic acids (21, 22), three kinds of inorganicmatter (quartz, R-Al2O3, and CaCO3) (18), two kinds of air-dried soils (20), and one kind of water saturated soil (16), hadthe relatively large magnitudes of coefficients a and b.Similarly, the relatively large magnitude of a and b coefficientswere also found in pp-LFER models that describe the vaporadsorption of organic compounds on active carbon andorganobentonites (15, 27).

Several studies have indicated that the relative humiditycan not cause a considerable variation in the relativecontributions of the van der Waals forces and H-bondstrengths to the vapor adsorption of organic compounds onsoil components. Goss et al. (17, 18) found that the van derWaals parameter and hydrogen-donor parameter of themineral surface both decreased from a value close to that ofa pure mineral surface to the value of a bulk water surfacewhen the relative humidity increases from 0 to 100%. Thestudies of Niederer et al. (21, 22) indicated that the adsorption

of polar compounds on humic and fulvic acids did not showany uniform variation with the change of relative humidity.Therefore, temperature should be a crucial factor for thevariation in the relative contribution of H-bond strengths tothe vapor adsorption of organic compounds on the naturalsoils or soil components. The remarkable decrease ofH-acceptor parameter (coefficient a) with increasing tem-perature from 60 to 240 °C have been observed for thestationary phase of wall-coated open-tubular columns forgas chromatography (32). In this study, the elevated tem-perature also causes a very small coefficient a for the naturalsoils. As a result, the removal of bipolar organic compoundsfrom soils can be greatly accelerated by thermal treatment.

AcknowledgmentsThis study was funded by the Key Projects in the NationalScience & Technology Pillar Program during the EleventhFive-Year Plan Period (2008BAC32B03) and the National BasicResearch Program of China (973 Program) (2009CB421602).We are grateful to the insightful comments from threeanonymous reviewers for improving our manuscript.

Supporting Information AvailableSolvation parameters of the test solutes; ANOVA lists for themultiple linear regression of the pp-LFER models and residualanalysis of the pp-LFER model; enthalpy changes for theadsorption of four typical solutes on heated soils; the reportedpp-LFER models and the pp-LFER models derived from thereported data; breakthrough curves of 1,2,4-trichlorobenzenethrough heated soil-columns; dependences of methane peakelution time and KSA values on the flow rate of carrier gas andthe temperature of soil column; the relationship betweenKSA values of the test solutes and their boiling points; plotsof predicted vs experimental log KSA values for test solutes;relative contributions of different interaction terms to theretention. This material is available free of charge via theInternet at http://pubs.acs.org.

Literature Cited(1) de Percin, P. R. Application of thermal desorption technologies

to hazardous waste sites. J. Hazard. Mater. 1995, 40, 203–209.(2) U.S. Naval Facilities Engineering Service Center. Application

Guide for Thermal Desorption Systems, Technical Report TR-2090-ENV; NFESC: Port Hueneme, CA, 1998.

(3) Pina, J.; Merino, J.; Errazu, A. F.; Bucala, V. Thermal treatmentof soils contaminated with gas oil: Influence of soil compositionand treatment temperature. J. Hazard. Mater. 2002, 94, 273–290.

(4) Lighty, J. S.; Pershing, D. W.; Cundy, V. A.; Linz, D. G.Characterization of thermal desorption phenomena for thecleanup of contaminated soil. Nucl. Chem. Waste Manage. 1988,8, 225–237.

(5) Lighty, J. S.; Silcox, G. D.; Pershing, D. W.; Cundy, V. A.; Linz,D. G. Fundamental experiments on thermal desorption ofcontaminants from soils. Environ. Progress 1989, 8, 57–61.

(6) Lighty, J. S.; Silcox, G. D.; Pershing, D. W. Fundamentals for thethermal remediation of contaminated soils. Particle and beddesorption models. Environ. Sci. Technol. 1990, 24, 750–757.

(7) Wolters, A.; Linnemann, V.; Smith, K. E. C.; Klingelmann, E.;Park, B.-J.; Vereecken, H. Novel chamber to measure equilibriumsoil-air partitioning coefficients of low-volatility organic chemi-cals under conditions of varying temperature and soil moisture.Environ. Sci. Technol. 2008, 42, 4870–4876.

(8) Keyes, B. R.; Silcox, G. D. Fundamental study of the thermaldesorption of toluene from montmorillonite clay particles.Environ. Sci. Technol. 1994, 28, 840–849.

(9) Tognotti, L.; Flytzani-Stephanopoulos, M.; Sarofim, A. F.;Kopslnis, H.; Stoukldes, M. Study of adsorption-desorption ofcontaminants on single soil particles using the electrodynamicthermogravimetric analyzer. Environ. Sci. Technol. 1991, 25,104–109.

(10) Wu, Y. P. G.; Dong, J. I.; Bozzelli, J. W. Mass transfer of hazardousorganic compounds in soil matrices: experiment and model.Combust. Sci. Technol. 1992, 85, 151–163.

VOL. 44, NO. 10, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3681

(11) Silcox, G. D.; Larsen, F. S.; Owens, W. D.; Choroszy-Marshall,M. Kinetics of hydrocarbon and pesticide removal from claysoils during thermal treatment in a pilot-scale rotary kiln. WasteManage. 1995, 15, 339–349.

(12) Di, P.; Chang, D. P. Y.; Dwyer, H. Modeling of polychlorinatedbiphenyl removal from contaminated soil using steam. Environ.Sci. Technol. 2002, 36, 1845–1850.

(13) Heron, G.; Vanzutphen, M.; Christensent, H.; Enfield, C. G. Soilheating for enhanced remediation of chlorinated solvents: Alaboratory study on resistive heating and vapor extraction in asilty, low-permeable soil contaminated with trichloroethylene.Environ. Sci. Technol. 1998, 32, 1474–1481.

(14) Vitha, M.; Carr, P. W. The chemical interpretation and practiceof linear solvation energy relationships in chromatography.J. Chromatogr. A 2006, 1126, 143–194.

(15) Burg, P.; Fydrych, P.; Abraham, M. H.; Matt, M.; Gruber, R. Thecharacterization of an active carbon in terms of selectivitytowards volatile organic compounds using an LSER approach.Fuel 2000, 79, 1041–1045.

(16) Poole, S. K.; Poole, C. F. Chromatographic models for the sorptionof neutral organic compounds by soil from water and air.J. Chromatogr. A 1999, 845, 381–400.

(17) Goss, K.-U.; Schwarzenbach, R. P. Quantification of the effectof humidity on the gas/mineral oxide and gas/salt adsorptionof organic compounds. Environ. Sci. Technol. 1999, 33, 4073–4078.

(18) Goss, K.-U.; Schwarzenbach, R. P. Adorption of a diverse set oforganic vapors on quartz, CaCO3, andR-Al2O3 at different relativehumidities. J. Colloid Interface Sci. 2002, 252, 31–41.

(19) Goss, K.-U.; Buschmann, J.; Schwarzenbach, R. P. Determinationof the surface sorption properties of talc, different salts, andclay minerals at various relative humidities using adsorptiondata of a diverse set of organic vapors. Environ. Toxicol. Chem.2003, 22, 2667–2672.

(20) Goss, K.-U.; Buschmann, J.; Schwarzenbach, R. P. Adsorptionof organic vapors to air-dry soils: Model predictions andexperimental validation. Environ. Sci. Technol. 2004, 38, 3667–3673.

(21) Niederer, C.; Goss, K.-U.; Schwarzenbach, R. P. Sorptionequilibrium of a wide spectrum of organic vapors in Leonarditehumic acid: Modeling of experimental data. Environ. Sci.Technol. 2006, 40, 5374–5379.

(22) Niederer, C.; Schwarzenbach, R. P.; Goss, K.-U. Elucidatingdifferences in the sorption properties of 10 humic and fulvicacids for polar and nonpolar organic chemicals. Environ. Sci.Technol. 2007, 41, 6711–6717.

(23) Endo, S.; Grathwohl, P.; Haderlein, S. B.; Schmidt, T. C.Compound-specific factors influencing sorption nonlinearityin natural organic matter. Environ. Sci. Technol. 2008, 42, 5897–5903.

(24) Arp, H. P. H.; Schwarzenbach, R. P.; Goss, K.-U. determinationof ambient gas-particle partitioning constants of non-polar andpolar organic compounds using inverse gas chromatography.Atmos. Environ. 2008, 42, 303–312.

(25) Goss, K.-U.; Schwarzenbach, R. P. Linear free energy relation-ships used to evaluate equilibrium partitioning of organiccompounds. Environ. Sci. Technol. 2001, 35, 1–9.

(26) Nguyen, T. H.; Goss, K.-U.; Ball, W. P. Polyparameter linear freeenergy relationships for estimating the equilibrium partition oforganic compounds between water and the natural organicmatter in soils and sediments. Environ. Sci. Technol. 2005, 39,913–924.

(27) Tian, S.; Zhu, L.; Shi, Y. Characterization of sorption mechanismsof VOCs with organobentonites using a LSER approach. Environ.Sci. Technol. 2004, 38, 489–495.

(28) Conder, J. R.; Young, C. L. Physicochemical Measurement byGas Chromatography; Wiley: New York, 1979.

(29) Abraham, M. H.; Andonia-Haftvan, J.; Whiting, G. S.; Leo, A.;Taft, R. S. Hydrogen bonding. Part 34. The factors that influencethe solubility of gasses and vapours in Water at 298 K, and anew method for its determination. J. Chem. Soc. Perkin Trans.2 1994, 1777–1791.

(30) Chiou, C. T.; Shoup, T. D. Soil sorption of organic vapors andeffects of humidity on sorptive mechanism and capacity.Environ. Sci. Technol. 1985, 19, 1196–1200.

(31) Thibaud, C.; Erkey, C.; Akgerman, A. Investigation of the effectof moisture on the sorption and desorption of chlorobenzeneand toluene from soil. Environ. Sci. Technol. 1993, 27, 2373–2380.

(32) Poole, C. F.; Poole, S. K. Separation characteristics of wall-coatedopen-tubular columns for gas chromatography. J. Chromatogr.A 2008, 1184, 254–280.

ES9034705

3682 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 10, 2010