1. Cyclodextrins for Desorption and
Solubilization of 2,4,6-Trinitrotoluene
and Its Metabolites from Soil
T A M A R A W . S H E R E M A T A A N D
J A L A L H A W A R I *
Biotechnology Research Institute, National Research Council
of Canada, Montreal, Quebec, Canada H4P 2R2
Heptakis-2,6-di-O-methyl-β-cyclodextrin (DMβCD) and
hydroxypropyl-β-cyclodextrin (HPβCD) (1% w/w solutions)
were investigated for their ability to desorb 2,4,6-
trinitrotoluene (TNT), 4-amino-2,6-dinitrotoluene (4-ADNT),
and 2,4-diamino-4-nitrotoluene (2,4-DANT) from two
artificially contaminated soils (an organic rich topsoil and
an illite shale). The DMβCD (which is highly surface
active) was more effective than HPβCD (negligible surface
activity) for desorption of the three nitroaromatic
compounds (NACs) from the two soils. The efficiency of
both CDs for NAC removal from topsoil decreased
with increasing amino substitution (i.e. TNT > 4-ADNT > 2,4-
DANT), whereas for illite the efficiency generally
decreased with increasing nitro substitution (i.e. TNT <
4-ADNT. 2,4-DANT). In general, the NAC removal efficiency
increased with decreases in the sorption capacity
constant (Kd
s). The two CDs were also assessed for their
ability to remove TNT from a highly contaminated topsoil
(5265 mg/kg) obtained from a former manufacturing facility
that had been aged for 10-40 yr. The estimated association
constants (Ks) were comparable to those obtained for
solubilization of pure TNT into aqueous solution. This study
demonstrates the effectiveness of CDs in decontaminating
a variety of soils containing both high and low levels of
TNT and some of its associated metabolites.
Introduction
Soils contaminated with 2,4,6-trinitrotoluene (TNT) from
activities in the munitions and defense industries is a
worldwide environmental problem. In natural and engi-
neered systems, it has been demonstrated that the nitro
groups of TNT undergo reduction reactions to form amino
derivatives that include 4-amino-2,6-dinitrotoluene (4-
ADNT), 2-amino-4,6-dinitrotoluene (2-ADNT), 2,4-diamino-
6-nitrotoluene (2,4-DANT), and 2,6-diamino-4-nitrotoluene
(2,6-DANT) (1, 2). It has been demonstrated that TNT and
its amino derivatives can be irreversibly sorbed by soil (1,
3-11). Irreversible sorption has been postulated as a possible
explanation for poor mineralization of TNT in soil (1).
Surfactants have been shown to be capable of desorbing
TNT from soil (12) and to subsequently enhance the
mineralization of TNT by Phanerochaete chrysporium (13).
However, there are potential drawbacks to the use of
surfactants for in and ex situ remediation. Potential operating
problems include sorption of the surfactant by soil, pre-
cipitation of the surfactant, phase separation, and foaming
(14). In addition, if surfactants are to enhance the bioavail-
ability of contaminants, then they must be compatible with
the accompanying biological process. For example, nonionic
surfactants have been shown to inhibit mineralization of
phenanthrene in soil water systems (15).
Cyclodextrins (CDs) are cyclic oligosaccharides formed
from the enzymatic degradation of starch (16). CDs are
considered nontoxic and used in pharmaceuticals and as
food additives (17). One class of CDs is able to enhance the
solubility of many organic compounds by forming inclusion
complexes with “guest” molecules. It has been demonstrated
in the laboratory and field that CDs are capable of enhancing
the solubility of a number of common organic contaminants
(18-20). Hydroxypropyl-β-cyclodextrin (HPβCD) was also
shown to significantly enhance the transport of anthracene,
pyrene, and trichlorobiphenyl through soil columns (21). In
a further study, HPβCD increased the solubility of phenan-
threne by 124 times and its removal rate by 6 times by a
phenanthrene degrading isolate (22). In a pilot-scale field
test, HPβCD increased the aqueous solubility of 12 target
compoundsby100to20 000timesfromanonaqueoussource
in an aquifer (20). As discussed by McCray and Brusseau
(20), CDs experience little or no sorption by soil and are not
subject to precipitation, hence they are easily removed from
the subsurface (which may be of regulatory and economic
concern). CDs have also been shown to enhance the
solubilization of the explosive hexahydro-1,3,5-trinitro-1,3,5-
triazine (RDX) from soil (23).
Based on published data, it would appear that CDs are
becoming comparable in cost with surfactants (24, 25).
Particularly, it has been reported that the cost of β-CD is as
low as $4/kg (25). In comparison, the cost of food grade
surfactants, that were evaluated for subsurface remediation,
was estimated to be $2.2/kg (assuming 1% inflation in
converting to 1999 $) (24). A detailed comparison of the costs
of surfactants and CDs for subsurface remediation was not
found in the literature. However, since the cost of CDs has
continuously decreased in recent years (24), investigations
regarding their technical merit for subsurface remediation
are justified.
We recently demonstrated that significant sorption-
desorption hysteresis exists for TNT, 4-ADNT, and 2,4-DANT
sorbed by natural and model soils (3). The purpose of the
presentstudywastoexaminetheabilityoftwoCDs(heptakis-
2,6-di-O-methyl-β-cyclodextrin (DMβCD) and HPβCD) to
desorb TNT, 4-ADNT, and 2,4-DANT from two artificially
contaminated soils examined in our earlier study (an organic
rich topsoil and an illite shale) (3). The DMβCD was selected
because it is highly surface active, and the HPβCD was
selected because it is commonly used and less expensive.
We also studied the ability of the two CDs to solubilize TNT
from a highly contaminated soil obtained from a former
manufacturing facility, where contamination has existed for
10-40 yr. Desorption and solubilization data for soil were
compared with the association constants that were estimated
from CD-enhanced solubility of the three NACs.
Experimental Section
Chemicals. TNT (>99% purity) was provided by Defense
Research Establishment Valcartier (Valcartier, PQ), 4-ADNT
(>99% purity) was purchased from Omega Inc. (Le´vis, PQ),
and 2,4-DANT (>99% purity) was purchased from Accu-
Standard Inc. (New Haven, CT). The HPβCD and DMβCD
(no reported purity) were purchased from Aldrich (Oakville,
ON). The CD solutions were prepared daily.
Soils. The agricultural topsoil and illite green shale used
are described in detail elsewhere in terms of soil properties
* Correspondingauthorphone: (514)496-6267;fax: (514)496-6265;
e-mail: Jalal.Hawari@nrc.ca.
Environ. Sci. Technol. 2000, 34, 3462-3468
3462 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000 10.1021/es9910659 CCC: $19.00 Published 2000 by the Am. Chem. Soc.
Published on Web 07/08/2000

2. and mineralogy (3). However, the topsoil contained 8.4%
organic matter, and the illite was a pure clay mineral. The
topsoil was sterilized by gamma irradiation at the Canadian
IrradiationCentre(Laval,Que.)withminimumandmaximum
doses of 35.4 to 40.4 kGy, respectively. Sterility was verified
according to the methods of Trevors (26), and all manipula-
tions with sterile topsoil were performed in a laminar flow
biological hood to maintain aseptic conditions.
A contaminated topsoil was obtained from a former
explosives manufacturing site in south eastern Quebec in
November 1992. The soil was contaminated between the
early 1960s and late 1980s, hence it had been aged with TNT
for quite some time. It contained 5265 mg/Kg (range of three
replicates was 4952-5441 mg/Kg) of TNT, as determined by
the EPA SW-846 Method 8330 described subsequently. The
soil was analyzed for TNT at the same time that as the
solubility enhancements using CDs (as described subse-
quently) were conducted.
Sorption-Desorption. Batch sorption experiments were
conducted in triplicate with topsoil and illite (3). The mass
of each nitroaromatic compound (NAC) sorbed was calcu-
lated by difference. The supernatant was decanted, and the
residual supernatant remaining with the soil pellet was
measured gravimetrically.
For topsoil, desorption was carried out by adding 15 mL
of distilled water, 1% HPβCD (w/w in distilled water), or 1%
DMβCD (w/w in distilled water) to the soil pellet (2 g topsoil).
For illite, 15 mL of 0.1 N KCl or the 1% CD solutions (prepared
in 0.1 N KCl) were added to 1 g of clay. The KCl was used
to avoid deflocculation of the clay. The centrifuge tubes were
wrapped in aluminum foil and agitated for 22 h to achieve
equilibrium, as this time was previously determined to be
more than sufficient to achieve equilibrium (3). Samples of
the supernatant were retrieved for analysis, and the masses
of NACs sorbed following desorption were calculated by dif-
ference. For each replicate, the % NAC removed from topsoil
andillitebythetwoCDsandaqueousmediumwascalculated
asthe%differencebetweenthemassofNACsorbedfollowing
sorption and that sorbed following desorption (with cor-
rection for NAC in the residual supernatant).
Control experiments that contained TNT, 4-ADNT, and
2,4-DANT in the absence of topsoil and illite were conducted
for each NAC concentration. Results indicate that there were
no losses from the system over the time periods studied.
Sorption-desorption experiments for 4-ADNT and 2,4-
DANT (both initially at 2.74 mg/L) with sterile and nonsterile
topsoil were also conducted. For desorption, 1% (w/w)
HPβCD, 1% (w/w) DMβCD, and distilled water were em-
ployed. Following desorption, the NAC that remained with
the soil pellet was extracted with acetonitrile, as described
by EPA SW-846 Method 8330 (27). Briefly, the soil pellet was
combined with 10.0 mL acetonitrile, vortexed for 1 min, and
placed in a sonicator bath (20 kHz) that was cooled to 22 °C
(Blackstone Ultrasonics, Jamestown, NY) for 18 h. After 30
min of settling, 5.0 mL of the supernatant was combined
with 5.0 mL of 5 g/L CaCl2. The solutions were manually
shaken and prepared for analysis after 15 min settling.
Solubility Enhancements by Cyclodextrins. Solubility
enhancements of TNT were used to estimate association
constants of the inclusion complexes that form with the two
CDs using the methods of Orienti et al. (28). Data analysis
is described subsequently. An excess of TNT (1500 mg/L)
was combined with increasing concentrations of CD (0, 0.1,
1.0, 2.0, and 4.0 w/w %). Solubility was measured over 12 d,
and there were no increases after the first 2 d. Therefore, the
solutions were equilibrated for 2 d at 23 °C, after which they
were filtered through Millex HV 0.45 µm filter units (Millipore
Corp., Bedford, MA). The filtrates were diluted 1:1 in
acetonitrile for analysis. The same procedure was imple-
mented for the field contaminated topsoil, where 1 g of soil
was combined with 10 mL of the CD solutions.
The solubilities of 4-ADNT and 2,4-DANT were deter-
mined in the absence and presence of DMβCD and HPβCD
(4 w/w %) by combining an excess of the NAC (1100 mg/L).
Due to the expense of these two NACs, a broader CD range
was not studied. The solutions were equilibrated for 2 d, and
samples were prepared for analysis as described above.
Analytical Methods. The liquid NAC samples were
analyzed using reversed-phase high-pressure liquid chro-
matography(HPLC)withanultraviolet(UV)photodiodearray
(PDA) detector. An HPLC system (Waters Associates, Milford
MA) consisting of a Model 600 pump, 717 Plus Autosampler,
and a 996 PDA detector (λ ) 254 nm) was used. The system
was outfitted with Millennium data acquisition software.
Separations were performed on a Supelcosil LC-8 column
(25.0 cm, 4.6 mm, 5 µm) (Supelco, Oakville, ON) at 35 °C with
an isocratic mobile phase composed of 18% 2-propanol and
82% water at 1 mL/min. The two CDs had no effect on the
UV spectra of the three NACs in samples diluted 1:1 in
acetonitrile for the CD concentrations that were examined.
Hence, all liquid NAC samples were diluted 1:1 with
acetonitrile prior to analysis.
The DMβCD and the HPβCD were also analyzed by the
HPLCsystemdescribedabovewitharefractiveindexdetector
(Mode 410). For the DMβCD, the Supelcosil LC-8 Supelco
column described above was used. The mobile phase was
50% acetonitrile and 50% water, the column temperature
was 30 °C, and the flow rate was 2 mL/min. For HPβCD
analysis, a Supelcosil LC-NH2 column (25.0 cm, 4.6 mm, 5
µm) (Supelco, Oakville, ON) at 30 °C was used. The mobile
phase consisted of 75% acetonitrile and 25% water and the
flow rate was 2 mL/min. The injection volume was 50 µL for
all cases.
Data Analysis
Sorption-Desorption. The following Freundlich isotherm
is adequate to describe equilibrium sorption and desorption
data for TNT and its metabolites in soil (3, 4)
where x/m is the mass of solute sorbed per unit mass of soil
at equilibrium (µg/g), Kd
s is the sorption capacity constant
(L/kg), C is the aqueous equilibrium phase solute concentra-
tion (mg/L), and n is a constant. The capacity constant for
sorption is denoted as Kd
s, and for desorption it is denoted
as Kd
d. Typically, Kd
s < Kd
d thereby indicating sorption-
desorption hysteresis. In addition to their ability to expedite
the solubilization of contaminants, CDs have the potential
to enhance the desorption of nitroaromatic compounds
(NACs) from soil (23), thereby decreasing sorption-desorp-
tion hysteresis.
InclusionComplexes.Solubilitydiagramsareconstructed
by measuring the increase in solubility of a solute (S) with
increases in concentration of an added ligand or complexing
agent (L) (29). When the increase in S with L is linear, then
the phase solubility diagram is classified as Type A. The
stability of inclusion complexes can be obtained by such
solubilitydiagrams(30).Inparticular,whenthestoichiometry
for inclusion complexes is 1:1, the following complexation
reaction occurs in solution:
The association constant (Ks) of such complexes is expressed
as
x
m
) Kd
s
Cn
(1)
CD + S 98
K
CD - S (2)
Ks )
[CD - S]
[S][CD]
(3)
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3463

3. where S is the concentration of free solute and is regarded
as the aqueous solubility of the solute, So. Hence, the
concentration of solute that complexes with CD is calculated
as the difference between the total aqueous solute concen-
tration and the free aqueous solute concentration (i.e. [CD
- S] ) [St] - [So]). The concentration of free CD is equal to
the concentration initially present minus that which is taken
up as an inclusion complex with the solute (i.e. [CD] ) ([CDt]
- ([St] - [So])). The stability constant can then be reduced
to the following:
The Ks can be determined from the phase solubility diagram
described above (28) using the following expression:
Results and Discussion
Sorption-Desorption Isotherms for Topsoil. The sorption-
desorption isotherms for TNT, 4-ADNT, and 2,4-DANT for
topsoil are shown in Figure 1. For each isotherm, the
Freundlich constants (Kd
s
, Kd
d
, and n) are summarized in
Table 1. The linear constants (i.e. n ) 1) are reported when
they provided a better fit of the data than the nonlinear case.
Sorption-desorption hysteresis for TNT, 4-ADNT, and 2,4-
DANT is evident by the fact that the desorption isotherm lies
above the sorption isotherm, and also since Kd
s
< Kd
d
(95%
confidence) for sorption-desorption of the three NACs with
topsoil (Table 1).
When 1% HPβCD was used to facilitate desorption, the
resulting isotherms remain above the sorption isotherms for
allthreeNACs.However,forTNTand4-ADNT,thedesorption
isotherms lie below the sorption isotherm when 1% DMβCD
was used for desorption. This is indicative of enhanced
desorption and is reflected by the fact that Kd
d < Kd
s when
DMβCD was used to facilitate desorption from topsoil (Table
1).
For 2,4-DANT desorption from topsoil using DMβCD, the
desorption isotherm lies above the desorption isotherm
where water alone was used (Figure 1c). Since NAC sorption
was calculated by difference, it was thought that losses of
2,4-DANT by biotransformation processes were mistakenly
attributedtosorptionprocesses.Thishypothesisisconsistent
with the fact that 2,4-DANT was biotransformed to 4-N-
acetylamino-2-amino-6-nitrotoluene (4-N-AcANT) and 2-N-
acetylamino-2-amino-6-nitrotoluene (2-N-AcANT) in non-
sterilized topsoil (3). Hence, further experiments were
conducted to determine the extent of 2,4-DANT desorption
using 1% DMβCD, 1% HPβCD, and distilled water in sterile
and nonsterile topsoil for an initial 2,4-DANT concentration
of 2.74 mg/L. From these data, point estimates of Kd
d were
calculated and are reported in Table 2. For nonsterile topsoil,
apparent Kd
d’s were calculated using the values of n reported
for nonsterile topsoil in Table 1.
According to the values of Kd
d for nonsterile topsoil (Table
2), it appears that 1% HPβCD had little effect on sorption-
desorption hysteresis and that 1% DMβCD increased the
sorption-desorption hysteresis. For sterile topsoil, the Kd
d’s
were calculated using the values of n reported in Table 1 for
nonsterile topsoil and also for the linear Freundlich constant
(i.e. n ) 1). For sterile topsoil, there was reduced sorption-
desorption hysteresis, as indicated by the reduced Kd
d (for
both linear and nonlinear estimates) for the two CDs as
compared to the values of Kd
d for the case of distilled water.
Therefore, the apparent increase in hysteresis illustrated in
Figure1cisanartifactcausedbyindigenousmicrobialactivity
in the soil. This was coupled with the appearance of 4-N-
AcANT in the aqueous phase and acetonitrile extract of the
solid sorbed phases (Table 2). Following desorption of 2,4-
DANT from nonsterile topsoil, there was more 4-N-AcANT
detected in the supernatant when CDs were used as
compared to distilled water, there was also slightly more
4-N-AcANTextractedwithacetonitrilefromnonsteriletopsoil
following desorption with DMβCD compared to the case of
distilled water. Furthermore, 4-N-AcANT was not detected
in either the aqueous or in the solid phases in any of the
replicates involving sterile topsoil. Therefore, it can be
concluded that both CDs enhanced the desorption of 2,4-
DANT from topsoil (i.e. sterile conditions) but that CDs can
also enhance its biotransformation to 4-N-AcANT (i.e.
nonsterile conditions) and possibly other metabolites.
Although there were no reports on the purity of the two
CDsusedinthisstudy,accordingtoWangetal.(22),technical
grade HPβCD can contain 3.5% propylene glycol that can act
as a carbon source for a phenanthrene degrading isolate
obtained from an estuary sediment. Results of the present
study are consistent with the hypothesis that the CD (or an
associated impurity) enhanced the activity of the indigenous
microbial population and that this resulted in partial
transformation of 2,4-DANT to 4-N-AcANT (and possibly
other metabolites). Since the concentration of HPβCD did
not change following 22 h contact with nonsterile topsoil (as
determined for experiments for CD sorption by topsoil/illite
FIGURE 1. Sorption-desorption isotherms for (a) TNT, (b) 4-ADNT,
and (c) 2,4-DANT for nonsterile topsoil using 1% HPβCD, 1% DMβCD,
and distilled water.
Ks )
1
So
( St - So
CDt - (St - So)) (4)
Ks )
slope
So(1 - slope)
(5)
3464 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000

4. thataredescribedsubsequently),itisunlikelythattheamount
of CD degraded was measurable.
According to Szeijtli (17), a necessary prerequisite for the
formation of inclusion complexes is that the molecular
volume of the NACs (i.e. “guest” molecule) be less than that
of the β-CD cavity (i.e. host cavity). However, slight solubility
enhancements with HPβCD have been observed for the case
of DDT, despite the fact that DDT has a larger molecular
volume than the CD cavity (18). This behavior was attributed
to partial entry of DDT within the CD cavity. The molecular
volumeofthethreeNACsstudiedandthe4-N-AcANTproduct
were calculated according to Wang and Brusseau (18) and
are summarized in Table 3. Since these values are all less
than the interior volume of a β-CD cavity (also given in Table
3), the formation of inclusion complexes between the four
NACs and the two β-CDs studied are theoretically possible.
The molecular widths and lengths of the three NACs were
calculated on the basis of bond lengths (Table 3). Although
4-N-AcANT is longer than the β-CD cavity, it may orient
itself within the cavity since its width is less than that of the
cavity.
The effects of the two CDs on the desorption of 4-ADNT
under sterile and nonsterile conditions were also studied
since previous work indicates that this NAC can form
microbially mediated acetylated products (13). However, for
the time span examined, there were no differences between
thesterileandnonsteriletopsoilwhendistilledwater,HPβCD,
and DMβCD were used for desorption. Nonetheless, such
biotransformations must be considered when evaluating the
effectiveness of CDs for soil remediation.
NAC Removal Efficiencies for Topsoil. The removal
efficiency of the two CDs and water for removing each NAC
from topsoil is depicted as a function of the initial mass of
NAC sorbed in Figure 2. The % removal of 2,4-DANT is also
given for the case of sterile topsoil for HPβCD and DMβCD
(closed symbols in Figure 2c) for one initial sorbed concen-
tration (approximately 11 µg/g). The Kd
s for the three NACs
(Table 1) increases with amino substitution. In particular,
Kd
s for 2,4-DANT is greater than that of 4-ADNT (95%
confidence), and Kd
s for 4-ADNT is greater than that of TNT
(80%confidence).Furthermore,theefficiencyofNACremoval
(Figure 2) decreases with amino substitution for DMβCD. As
discussed previously, the only reactive constituent in the
topsoil is its organic matter (8.4%), and the increase in Kd
s
with amino group substitution may be due to irreversible
interaction with quinoidal structures of soil organic matter
(SOM) (3). The fact that as the amino group substitution
TABLE 1. Freundlich Sorption and Desorption Isotherm Parametersa
topsoilb illitec
operationb NAC Kd, L/Kg n r2 Kd, L/Kg n r2
sorption (Kds) TNT 6.38 ( 1.15 0.82 ( 0.06 0.98 223.63 ( 1.09 0.47 ( 0.05 0.97
4-ADNT 7.91 ( 1.10 0.70 ( 0.09 0.96 58.52 ( 1.30 0.68 ( 0.18 0.84
2,4-DANT 11.96 ( 1.16 0.71 ( 0.13 0.92 19.73 ( 1.11 0.81 ( 0.10 0.96
desorption-aqueous medium TNT 12.01 ( 1.10 0.82 ( 0.08 0.97 265.98 ( 1.05 0.40 ( 0.02 0.99
4-ADNT 14.95 ( 2.06 1.00d 0.78 87.56 ( 14.38 1.00d 0.72
2,4-DANT 36.58 ( 1.59 0.64 ( 0.17 0.84 81.85 ( 7.59 1.00d 0.93
desorption- 1% HPβCD (Kd
d) TNT 12.17 ( 1.07 0.69 ( 0.08 0.97 224.93 ( 1.13 0.43 ( 0.06 0.95
4-ADNT 10.04 ( 1.28 1.04 ( 0.23 0.88 40.04 ( 5.85 1.00d 0.84
2,4-DANT 32.60 ( 1.43 0.53 ( 0.13 0.83 18.56 ( 1.24 0.33 ( 0.09 0.82
desorption- 1% DMβCD (Kd
d) TNT 2.08 ( 1.06 0.84 ( 0.04 0.99 139.72 ( 1.06 0.46 ( 0.04 0.98
4-ADNT 6.07 ( 1.07 0.82 ( 0.06 0.98 26.84 ( 3.20 1.00d 0.87
2,4-DANT 50.25 ( 1.47 0.61 ( 0.13 0.88 16.64 ( 1.29 0.28 ( 0.11 0.70
a 95% confidence intervals are shown for values of Kd and n determined from 18 data points. b Aqueous medium was distilled water. c Aqueous
medium was 0.1 N KCl. d 95% confidence intervals are not shown for cases where the linear isotherm provided a better fit, since in these cases
n was set to 1.00.
TABLE 2. Partition Coefficients for 2,4-DANT Desorption from Nonsterile and Sterile Topsoila
and 4-N-AcANT Measured in
Supernatant and Soil Extractsb
non-sterile topsoil sterile topsoil
medium used
for desorption
Kd
d, L/kg
nonlinearc
aqueous
4-N-AcANT, mg/L
extracted
4-N-AcANT, µg/g
Kd
d, L/Kg
nonlineard
Kd
d, L/Kg
linear (n ) 1)
distilled water 39.23 0.049 0.509 29.30 19.17
(0.005-0.101) (0.127-1.155)
1% HPβCD 35.13 0.091 0.701 19.19 13.06
(0.020-0.196) (0.184-1.509)
1% DMβCD 43.83 0.087 0.769 18.16 13.37
(0.007-0.180) (0.033-1.571)
a Point estimates for Cinitial ) 2.74 mg/L. b Range of three replicates given in parentheses. c Values of n used to calculate Kd
d for nonsterile topsoil
are reported in Table 1. d Values of n used to calculate Kd
d for sterile topsoil are those obtained for nonsterile topsoil, Table 1.
TABLE 3. Selected Properties of NACs and CDs Used in This
Studya
compound
volume,b
nm3
molecular
weight
width,d
nm
length,d
nm
aqueous
solubility,
mg/L (at 25 °C)
TNT 0.258 227 0.71 0.69 128
(124-130)
4-ADNT 0.246 197 0.67 0.69 42
(38-46)
2,4-DANT 0.234 167 0.67 0.65 1597
(1584-1600)
4-N-AcANT 0.305 209 0.67 0.84
HPβCD 0.346c 1331 0.75 0.78
DMβCD 0.346 1500 0.75 0.78 570 000 (34)
a Range of three replicates given in parentheses. b Calculated ac-
cording to Wang and Brusseau (18). c For β-CD cavity (18). d Estimated
on the basis of reported bond lengths (31). The width was calculated
across the two ortho substituents on either side of the methyl group,
and the length was calculated from the methyl group to the para
substituent.
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3465

5. increased (i.e. 2,4-DANT > 4-ADNT > TNT), the efficiency
of DMβCD for desorption decreased is consistent with
irreversible interactions that may include quinoidal group
interaction.
The point estimates for the association constants (as-
suming 1:1 inclusion complex formation) for the two CDs
and the three NACs in aqueous solutions (Table 4) indicate
the following affinities for inclusion complex formation for
both CDs: 4-ADNT > 2,4-DANT > TNT. This preference
may be explained by the geometry of the NAC molecules in
conjunction with a preference of amino groups to coordinate
with ether groups within the CD cavity. For TNT, the width
of the molecule (across the two ortho substituents) is similar
to the width of the β-CD cavity (Table 3). Therefore, TNT
may only be capable of entering the CD cavity on the side
of its para nitro group. Furthermore, the nitro group may
not coordinate as well with the ether groups of the interior
of the CD cavity as compared to the amino groups in the
para positions of 4-ADNT and 2,4-DANT. The difference in
Ks of4-ADNTand2,4-DANTmaybeexplainedbyapreference
of the amino groups for coordination with the ether groups
withintheCDcavityinconjunctionwithmoleculargeometry.
The 2,4-DANT may orient itself with the CD cavity on the
side of its two amino groups and the 4-ADNT on the side of
its one amino group. The 2,4-DANT would encounter steric
hindrance in this orientation, as compared to 4-ADNT.
Furthermore, based on the association constants (Ks) for
TNT and 4-ADNT, complex formation with DMβCD is
favorable to HPβCD; whereas, for 2,4-DANT, complex
formation with HPβCD is favorable to DMβCD. This is in
contrast to what was found for NAC desorption from topsoil
by DMβCD where the removal efficiency was inversely
proportional to sorption affinity (i.e. Kd
s). For HPβCD, the
following trends for removal efficiencies were observed:
4-ADNT > 2,4-DANT-sterile ≈ TNT for similar levels of NACs
initially sorbed. This trend is somewhat comparable to the
association constants reported in Table 4 (i.e. 4-ADNT >
2,4-DANT> TNT). Therefore, for topsoil, removal efficiencies
dependedsomewhatonassociationconstants(Ks)forHPβCD
(negligible surface activity) and on sorption affinity (Kd
s) for
DMβCD (highly surface active).
Sorption-DesorptionIsothermsforIllite.Thesorption-
desorption isotherms for the three NACs and illite are
depicted in Figure 3, and the Freundlich constants are
summarized in Table 1. There was sorption-desorption
hysteresis for all NACs when 0.1 N KCl was used for
desorption. For TNT, desorption hysteresis was slightly
reduced by 1% HPβCD and was significantly reduced by 1%
DMβCD. For the 4-ADNT, both CDs were highly effective in
reducing desorption hysteresis, with DMβCD being slightly
more effective. For 2,4-DANT desorption from illite, both
CDs were nearly equal in reducing desorption hysteresis.
NAC Removal Efficiencies for Illite. The CD enhanced
removal efficiencies for NAC desorption from illite are
depicted in Figure 4. For TNT, the sorption capacity constant
(Kd
s of 223.63 L/Kg, Table 1) was high, and the % removal for
both CDs was lowest for TNT (Figure 4). These results imply
that sorption affinity is inversely proportional to removal
efficiency for TNT and illite. The low removal efficiency of
TNT from illite by both CDs may be explained by the
formation of electron donor acceptor (EDA) complexes
between the electron withdrawing nitro groups of TNT and
the siloxane surface (9, 10). In addition, low affinity for
coordination of thepara nitro group with ether groups within
the CD cavity may have contributed to the low removal
efficiency. For 2,4-DANT, the removal efficiency increased
with the initial sorbed concentration but ultimately reached
the same levels as 4-ADNT that were relatively constant with
the initial sorbed concentration. The fact that the 4-ADNT
removal efficiencies were high and constant with surface
loading may be related to the high values of the association
constant (Ks) for 4-ADNT and the two CDs (Table 4).
Solubility Enhancements for Field Contaminated Soil.
The phase solubility diagrams for TNT in aqueous solution
and for a highly contaminated soil obtained from the field
FIGURE 2. Removal efficiencies of (a) TNT, (b) 4-ADNT, and (c)
2,4-DANT using 1% DMβCD, 1% HPβCD, and water for topsoil (note
that for 2,4-DANT, data for sterile topsoil are also included as closed
symbols).
TABLE 4. Stability Constants of TNT, 4-ADNT, and 2,4-DANT for
Complexation with HPβCD and DMβCD in Aqueous Systems
and Slurries with Field Contaminated Soila
compound system CD Ks µM-1
TNT aqueous 4% HPβCD 14.5b
TNT aqueous 4% DMβCD 55.8b
4-ADNT aqueous 4% HPβCD 254.1b
4-ADNT aqueous 4% DMβCD 443.1b
2,4-DANT aqueous 4% HPβCD 123.3b
2,4-DANT aqueous 4% DMβCD 78.9b
TNT aqueous HPβCD 18.1 ( 9.1c
TNT field contaminated
soil
HPβCD 15.2 ( 3.7c
TNT aqueous DMβCD 66.4 ( 7.9c
TNT field contaminated
soil
DMβCD 62.9 ( 16.3c
a All determined at 23 °C. b Point estimates using eq 4 (n ) 3).
c Calculated using eq 5 and the slopes of the linear trend lines of St
versus CDt (95% confidence intervals shown for 18 data points).
3466 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000

6. with increasing CD concentrations are depicted in Figure 5.
ForpureTNTinaqueoussolution,itisevidentthattherelative
solubility (actual solubility divided by initial solubility or St /
So from eq 4) increased substantially with increasing quanti-
ties of DMβCD (Figure 5b), but the increase in relative
solubility was less pronounced for HPβCD (Figure 5a). Since
the solubility increased linearly with increasing CD con-
centration (i.e. CDt versus St, from eq 4) and the slope of the
two curves is less than unity, the 1:1 inclusion complex
formation is a reasonable assumption (29). The association
constants (Ks), calculated from eq 5 and the slopes of the
solubility diagrams (St versus CDt), are summarized at the
bottom of Table 4. From this, it is evident that TNT complex
formation with DMβCD was more favorable than HPβCD.
Furthermore, the apparent stability constants for the field
contaminated soil are comparable to those for aqueous
systems for both CDs despite the fact that the field con-
taminated soil had been aged for 10-40 yr. This implies that
for soils containing high levels of TNT, extraction will be
limited by TNT solubility and that this may be enhanced by
the use of CDs. In particular, Hundal et al. (6) found that
whensolidphaseTNTwaspresentinartificiallycontaminated
soil, the concentration present in various extracting solvents
(i.e. 3 mM CaCl2 and CH3CN) was limited by the solubility
of TNT in the respective solvents. For the present system,
the total concentration of TNT in the soil slurries of the
experiments involving the field contaminated soil was
calculated as 526 mg/L, which far exceeds the aqueous
solubility of TNT (128 mg/L, Table 3). Hence, phase solubility
diagrams may be used to predict the efficiency of cyclo-
dextrins for such highly contaminated soils in the field.
SorptionofCyclodextrinsbySoil.TheDMβCDperformed
consistentlybetterthantheHPβCDfordesorptionofallNACs
from illite and topsoil and for solubilization of TNT from the
field contaminated topsoil. This may stem from the high
surface activity of DMβCD as compared to the case of HPβCD
that has negligible surface activity (16). This is also consistent
with the fact that HPβCD was not sorbed by either illite or
topsoil, whereas DMβCD was sorbed by topsoil (2.2 w/w %
of total CD mass) and illite (9.9 w/w % of total CD mass).
These results are consistent with those of Ko et al. (32) where
it was reported that HPβCD was not significantly sorbed by
kaolinite.
Enhanced Desorption of NACs by Cyclodextrins. Al-
though it has been demonstrated that the presence of HPβCD
can retard sorption of contaminants (i.e. anthracene, pyrene,
and trichlorobiphenyl) by soil (21), the present study
demonstrates that both HPβCD and DMβCD were able to
facilitate desorption. Ko et al. (32) suggested that HPβCD
was able to desorb phenanthrene from a model sandy aquifer
by flushing with HPβCD solution. This was based on a one-
dimensional numerical model assuming a Kd
s of 0.411 L/Kg
for the distribution of phenanthrene between aquifer solids
and the aqueous phase. However, it is unlikely that there
FIGURE 3. Sorption-desorption isotherms for (a) TNT, (b) 4-ADNT,
and (c) 2,4-DANT from nonsterile illite using 1% DMβCD, 1% HPβCD,
and 0.1 N KCl.
FIGURE 4. Removal efficiencies for (a) TNT, (b) 4-ADNT, and (c)
2,4-DANT using 1% DMβCD, 1% HPβCD, and 0.1 N KCl.
VOL. 34, NO. 16, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 3467

7. was extensive sorption of phenanthrene by the aquifer solids
considering that such a small value of Kd
s was assumed. In
particular, Huang et al. (33) reported Kd
s values that were
3-4 orders of magnitude larger than the value assumed by
Ko et al. (32) for five EPA reference soils.
Cyclodextrins for Remediation of Soil Contaminated by
NACs. This study demonstrates that CDs can be used to
enhancethesolubilizationofNACsfromhighlycontaminated
soil and that their efficiencies can be accurately predicted
from solubility phase diagrams. An interesting property of
methylated CDs (i.e. DMβCD) is that increased methylation
increases CD solubility in cold water (16). No explanation for
the unusual inverse relationship between solubility and
temperature was given (16). Although this property may be
undesirable in pharmaceutical applications since precipita-
tion will occur at higher temperatures, it is an added benefit
for in situ remediation since subsurface temperatures are
approximately 10 °C. Furthermore, the fact that CDs were
able to desorb TNT from soil is indicative of their potential
application for soil remediation at lower levels of contami-
nation. The DMβCD was generally more effective than
HPβCD. Both CDs enhanced the biotransformation of 2,4-
DANT to 4-N-AcANT in nonsterile topsoil. This last point
must be considered when evaluating the effectiveness of CDs
for soil remediation. Further research will focus on using
these cyclodextrins to enhance the solubility of TNT and its
metabolites in soil, that may in turn enhance their availability
for biodegradation.
Acknowledgments
We thank the Natural Sciences and Engineering Research
Council and the National Research Council (NRC) of Canada
for a Fellowship to T.W.S. and the Department of National
Defence for their continued interest in this work. The authors
would also like to thank Louise Paquet and Annamaria Halasz
for helpful discussions and technical support. This is NRC
publication No. 43312.
Literature Cited
(1) Daun, G.; Lenke, H.; Reuss, M.; Knackmuss, H.-J. Environ. Sci.
Technol. 1998, 32, 1956-1963.
(2) Gorontzy,T.;Drzyzga,O.;Kahl,M.W.;Bruns-Nagel,D.;Breitung,
J.; von Loew, E.; Blotevogel, K,-H. Crit. Rev. Microbiol. 1994,
20(4), 265-284.
(3) Sheremata, T. W.; Thiboutot, S.; Ampleman, G.; Paquet, L.;
Halasz, A.; Hawari, J. Environ. Sci. Technol. 1999, 33, 4002-
4008.
(4) Selim, H. M.; Iskandar, I. K. Sorption-Desorption and Transport
of TNT and RDX in Soils; CRREL Report 94-7; U.S. Army Corps
of Engineers, Office of the Chief of Engineers: 1994.
(5) Li,A.Z.;Marx,K.A.;Walker,J.;Kaplan,D.L.Environ.Sci.Technol.
1997, 31, 584-589.
(6) Hundal, L. S.; Shea, P. J.; Comfort, S. D.; Powers, W. L.; Singh,
J. J. Environ. Qual. 1997, 26, 896-904.
(7) Pennington, J. C.; Patrick, W. H., Jr. J. Environ. Qual. 1990, 19,
559-567.
(8) Comfort, S. D.; Shea, P. J.; Hundal, L. S.; Li, Z.; Woodbury, B.
L.; Martin, J. L.; Powers, W. L. J. Environ. Qual., 1995, 24, 1174-
1182.
(9) Haderlein,S.B.;Weissmahr,K.W.;Schwarzenbach,R.P. Environ.
Sci. Technol. 1996, 30, 612-622.
(10) Haderlein, S. B.; Schwarzenbach, R. P. Environ. Sci. Technol.
1993, 27, 316-326.
(11) Xue, S. K.; Iskandar, I. K.; Selim, H. M. Soil Sci. 1995, 160(5),
317-327.
(12) Taha, M. R.; Soewarto, H.; Acar, Y. B.; Gale, R. J.; Zappi, M. E.
Water, Air, Soil Pollut. 1997, 100, 33-48.
(13) Hawari, J.; Halasz, A.; Beaudet, S.; Paquet, L.; Ampleman, G.;
Thiboutot, S. Appl. Environ. Microbiol. 1999, 65(7), 2977-2986.
(14) Deshpande, S.; Shiau, B. J.; Wade, D.; Sabatini, D. A.; Harwell,
J. H. Water Res. 1999, 33, 351-360.
(15) Laha, S.; Luthy, R. G. Environ. Sci. Technol. 1991, 25, 1920-
1930.
(16) Fro¨mming, K.-H.; Szejtli, J. Cyclodextrins in Pharmacy; Kluwer
Academic Publishers: Dordrecht, The Netherlands, 1994.
(17) Szejtli, J. In Cyclodextrins and their Industrial Uses; Ducheˆne,
D., Ed.; Editions de Sante´: Paris, France, 1987; Chapter 5.
(18) Wang, X.; Brusseau, M. L. Environ. Sci. Technol. 1993, 27, 2821-
2825.
(19) Wang, X.; Brusseau, M. L. Environ. Sci. Technol. 1995, 29, 2346-
2351.
(20) McCray, J. E.; Brusseau, M. L. Environ. Sci. Technol. 1998, 32,
1285-1293.
(21) Brusseau, M. L.; Wang, X.; Hu, Q. Environ. Sci. Technol. 1994,
28, 952-956.
(22) Wang, J.-M, Marlowe, E. M.; Miller-Maier, R. M.; Brusseau, M.
L. Environ. Sci. Technol. 1998, 32, 1907-1912.
(23) Hawari,J.;Paquet,L.;Zhou,E.;Halasz,A.;Zilber,B.Chemosphere
1996, 32, 1929-1936.
(24) Sabatini, D. A.; Knox, R. C.; Harwell, J. H. Surfactant-Enhanced
DNAPL Remediation: Surfactant Selection, Hydraulic Efficiency,
and Economic Factors; Environmental Research Brief; United
States Environmental Protection Agency, EPA/600/5-961002;
NationalRiskManagementResearchLaboratory: Ada,OK,1996.
(25) McCoy, M. Chem. Eng. News 1999, March 1, 25-28.
(26) Trevors, J. T. J. Microbiol. Methods 1996, 26, 53-59.
(27) US EPA. Method 8330: Nitroaromatics and Nitramines by High
Performance Liquid Chromatography (HPLC), Test Methods for
Evaluating Solid Waste, SW-846 update III, Part 4: 1(B); Office
of Solid Waste: Washington, DC, 1997.
(28) Orienti, I.; Fini, A.; Bertis, V.; Zecchi; V. Eur. J. Pharm. Biopharm.
1991, 37, 110-112.
(29) Higuchi, T.; Connors, K. Adv. Anal. Chem. Instrum. 1965, 4,
117-212.
(30) Szejtli, J. Cyclodextrin Technology; Kluwer Academic Publish-
ers: Dordrecht, The Netherlands, 1988.
(31) Lide, D. R. CRC Handbook of Chemistry and Physics, 72nd ed.;
CRC Press: Boca Raton, FL, 1991; Section 9.
(32) Ko, S.; Schlautman, M. A.; Carraway, E. R. Environ. Sci. Technol.
1999, 33, 2765-2770.
(33) Huang, W.; Yu, H.; Weber, W. J., Jr. J. Cont. Hydrol. 1998, 31,
129-148.
(34) Uekema, K.; Irie, T. In Cyclodextrins and their Industrial Uses;
Ducheˆne,D.,Ed.;EditionsdeSante´: Paris,France,1987;Chapter
10.
Received for review September 15, 1999. Revised manuscript
received April 10, 2000. Accepted May 10, 2000.
ES9910659
FIGURE 5. Phase-diagrams for solubilization of pure TNT in solution
and for a highly contaminated soil from the field (i.e. 5265 mg/kg)
using (a) HPβCD and (b) DMβCD (vertical bars denote range of
three replicates).
3468 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 16, 2000