Journal of Chromatography A, 1022 (2004) 83–94
Use of reversed-phase high-performance liquid chromatography–diode
array detection for complete separation of 2,4,6-trinitrotoluene
metabolites and EPA Method 8330 explosives:
influence of temperature and an ion-pair reagent
Thomas Borch∗,1, Robin Gerlach
Center for Biofilm Engineering, 366 EPS Building, P.O. Box 173980, Montana State University-Bozeman, Bozeman, MT 59717-3980, USA
Received 30 June 2003; received in revised form 11 September 2003; accepted 23 September 2003
Abstract
Explosives such as 2,4,6-trinitrotoluene (TNT), octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX), and hexahydro-1,3,5-trinitro-1,3,5-
triazine (RDX) are widely distributed environmental contaminants. Complete chromatographic separation is necessary in order to accurately
determine and quantify explosives and their degradation products in environmental samples and in (bio)transformation studies. The present
study describes a RP-HPLC method with diode array detection using a LC-8 guard column, a Supelcosil LC-8 chromatographic column,
and a gradient elution system. This gradient method is capable of baseline separating the most commonly observed explosives and TNT
transformation metabolites including 2,4,6-triaminotoluene (TAT) in a single run. The TNT metabolites separated were 2-hydroxylamino-
4,6-dinitrotoluene, 4-hydroxylamino-2,6-dinitrotoluene, 2,4-dihydroxylamino-6-nitrotoluene, 4,4′,6,6′-tetranitro-2,2′-azoxytoluene, 2,2′,6,6′-
tetranitro-4,4′-azoxytoluene, 4,4′,6,6′-tetranitro-2,2′-azotoluene, 2,2′,6,6′-tetranitro-4,4′-azotoluene, 2-amino-4,6-dinitrotoluene, 4-amino-2,
6-dinitrotoluene, 2,6-diamino-4-nitrotoluene, 2,4-diamino-6-nitrotoluene, and TAT. The same gradient method at a different column tem-
perature can also be used to baseline separate the explosives targeted in the Environmental Protection Agency (EPA) Method 8330 with
approximately 22% reduction in total run time and 48% decrease in solvent consumption compared to previously published methods. Good
separation was also obtained when all TNT metabolites and EPA Method 8330 compounds (a total of 23 compounds) were analyzed to-
gether; only 2,6-DANT and HMX co-eluted in this case. The influence of temperature (35–55 ◦C) and the use of an ion-pair reagent on the
chromatographic resolution and retention were investigated. Temperature was identified as the key parameter for optimal baseline separation.
Increased temperature resulted in shorter retention times and better peak resolution especially for the aminoaromatics investigated. The use
of an ion-pair reagent (octanesulfonic acid) generally resulted in longer retention times for compounds containing amine functional groups,
more baseline noise, and decreased peak resolution.
© 2003 Elsevier B.V. All rights reserved.
Keywords: Explosives; Environmental analysis; EPA Method 8330; Temperature selectivity; 2,4,6-Trinitrotoluene; Hydroxylaminodinitrotoluene;
Tetranitroazoxytoluene; Tetranitroazotoluene
1. Introduction
The nitroaromatic compound 2,4,6-trinitrotoluene (TNT),
has been found to contaminate soils at former and present
∗ Corresponding author. Present address (from January 2004):
Department of Geological and Environmental Sciences, Bld. 320, Rm
118, Stanford University, Stanford, CA 94305-2115, USA.
Tel.: +1-406-994-4770; fax: +1-406-994-6098.
E-mail addresses: thomas b@erc.montana.edu, borch@stanford.edu
(T. Borch).
1 Tel.: +1-650-723-4152; fax: +1-650-725-2199.
munitions manufacturing facilities, storage depots, and
former sites of explosives use [1,2]. Octahydro-1,3,5,7-
tetranitro-1,3,5,7-tetrazocine (HMX) and hexahydro-1,3,5-
trinitro-1,3,5-triazine (RDX) are also contaminants at these
locations as they are routinely components of TNT-based
explosive mixtures [1,3,4].
At these sites, TNT is often being biologically re-
duced stepwise to hydroxylaminodinitrotoluenes (HAD-
NTs), aminodinitrotoluenes (ADNTs), diaminonitrotoluenes
(DANTs), and 2,4,6-triaminotoluene (TAT) (Fig. 1). The
HADNTs, which are intermediates in the transforma-
tion of TNT, include 2-hydroxylamino-4,6-dinitrotoluene
0021-9673/$ – see front matter © 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2003.09.067
84 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
Fig. 1. 2,4,6-Trinitrotoluene (TNT) transformation pathways observed in soil and water.
(2-HADNT), 4-hydroxylamino-2,6-dinitrotoluene (4-HAD-
NT), 2,4-dihydroxylamino-6-nitrotoluene (2,4-DHANT)
and (not shown in Fig. 1) 2,6-dihydroxylamino-4-nitroto-
luene (2,6-DHANT) [5]. The ADNTs are 2-amino-4,6-
dinitrotoluene (2-ADNT) and 4-amino-2,6-dinitrotoluene
(4-ADNT). The DANTs are 2,4-diamino-6-nitrotoluene
(2,4-DANT) and 2,6-diamino-4-nitrotoluene (2,6-DANT)
[1,3,6]. The azoxy compounds, such as 4,4′,6,6′-tetranitro-
2,2′-azoxytoluene (2,2′-azoxy) and 2,2′,6,6′-tetranitro-4,4′-
azoxytoluene (4,4′-azoxy) are believed to be the products of
spontaneous abiotic hydroxylamino–nitroso condensation
reactions (Fig. 1) [1,5,7].
Environmental concerns stem from the mutagenic, car-
cinogenic, and toxic effects of nitroaromatic and aminoaro-
matic compounds. TNT, 2-ADNT, 4-ADNT, 2,4-DANT,
2,6-DANT and TAT have been found to be cytotoxic pre-
sumably due to induced oxidative stress [8,9]. TNT, RDX,
HMX, and the TNT-derived metabolites all demonstrate
mutagenic capability [9] and the EPA classifies TNT and
RDX as possible human carcinogens [10].
In order to develop accurate metabolic pathway maps,
HPLC methods capable of analyzing and separating all ex-
plosives and their transformation products must be available.
The reduced metabolites of TNT exhibit varying degrees of
polarity and chemical stability making the chromatographic
separation extremely challenging. The hydroxylamines are
known to be unstable in aqueous solution in the presence
of molecular oxygen potentially forming azoxy compounds
[5]. Azoxy compounds can undergo further transformation
to form 4,4′,6,6′-tetranitro-2,2′-azotoluene (2,2′-azo) and
2,2′,6,6′-tetranitro-4,4′-azotoluene (4,4′-azo) [11]. The for-
mation of HADNTs, azoxy, and azo compounds may be the
major cause of poor mass balances obtained in bioremedia-
tion systems, where only aminated products are monitored
[5,12].
TAT is one of the most unstable reduced TNT metabo-
lites in aqueous solutions. The three amino groups and the
methyl group are all activating substituents, which increase
the reactivity of the aromatic ring toward electrophiles such
as oxygen. Consequently, TAT is highly oxygen sensitive
and easily degraded under aerobic conditions [1]. However,
even under anaerobic conditions, TAT can participate in oxy-
dimerization and polymerization to form azo and poly-azo
compounds [1,13].
Although GC, solid-phase microextraction (SPME)–GC–
MS, CE-UV, and TLC have been used, HPLC has remained
the major analytical tool for the detection and quantification
of nitroaromatic compounds [1,3,7,11,12,14–18]. The most
T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94 85
commonly used method for the analysis of nitroaromatics is
HPLC with UV detection due to its widespread availability
[3,15], while HPLC combined with MS and electrochemical
detection (ED) are also viable but less frequently available
methods of detection [13,14,19,20].
The analysis of explosives using HPLC continues to
present challenges [3]. Previously published methods have
focused on the use of C-18 RP-HPLC [3,5,20,21] and
acetonitrile as the organic mobile phase. Acetonitrile is of
significant greater health and environmental concern than
methanol and therefore methods avoiding the use of ace-
tonitrile are desirable. Many previously published methods
had difficulties with the separation of the following iso-
mer pairs: 2- and 4-HADNT, 2- and 4-ADNT, 2,4- and
2,6-DANT, 2,2′- and 4,4′-azoxy [11,12,14,15,21,22]. Con-
sequently, the identification of the polar TAT and the less
polar azoxy dimers in TNT degradation studies has often
been performed in separate HPLC runs [23].
The goal of the present study was to separate TNT, HAD-
NTs, 2,4-DHANT, ADNTs, DANTs, TAT, azoxy and azo
compounds in a single HPLC run and to improve the chro-
matography of the 14 EPA Method 8330 compounds. The
explosives targeted in EPA Method 8330 were included in
the present studies due to the fact that they often occur as
co-contaminants in environmental samples containing TNT.
They could therefore potentially interfere or co-elute with
TNT or its reduced metabolites during HPLC analysis.
2. Experimental
2.1. Chemicals and sample preparation
The TNT, ADNT, and DANT standards (1000
g/ml in
acetonitrile, purity >99.0%) were obtained from Supelco
(Bellefonte, PA, USA). TAT*3HCl and 4,4′,6,6′-tetranitro-
2,2′-azoxytoluene were obtained from Dr. R.J. Spang-
gord, SRI International, Menlo Park, CA. The chemi-
cals included in EPA Method 8330 (all with a purity
>96.8%), 4-hydroxylamino-2,6-dinitrotoluene (purity
96.0%), 2-hydroxylamino-4,6-dinitrotoluene (purity 97.1%),
2,2′,6,6′-tetranitro-4,4′-azoxytoluene (purity 98.8%),
2,2′,6,6′-tetranitro-4,4′-azotoluene (purity 90.7%) and
4,4′,6,6′-tetranitro-2,2′-azotoluene (purity 94.7%) were
obtained from AccuStandard (New Haven, CT, USA).
2,4-dihydroxylamino-6-nitrotoluene was synthesized bio-
chemically and kindly provided by Dr. J.B. Hughes, Rice
University, Houston, TX [5]. TNT (neat; purity 98.0%)
for the degradation studies was obtained from Chem Ser-
vice (West Chester, PA, USA). HPLC grade methanol
(UV cutoff 205 nm) and Optima acetonitrile (UV cutoff
190 nm) were purchased from Fisher Scientific (Fair Lawn,
NJ, USA). HPLC grade dibasic sodium phosphate hepta
hydrate and enzyme grade monobasic sodium phosphate
were obtained from Fisher Scientific and Fisher Biotech,
respectively (Fair Lawn, NJ, USA). The ion-pair reagent oc-
tanesulfonic acid (sodium salt) was obtained from ACROS
(New Jersey, USA). The ion-pair reagent Low-UV PIC
B8 was obtained from Waters (Milford, MA, USA). The
water used in the preparation of mobile phases and stan-
dards was obtained from a Barnstead NANOpure system
(resistivity ≥ 17.6 M cm).
The TNT metabolite mixture was prepared in a glove
box in the absence of oxygen (90% N2, 5% CO2, 5% H2)
and concentrated by rapid evaporation of acetonitrile. TAT
standards have been found to easily degrade in H2O so-
lutions [24]. However, degradation of the oxygen and pH
sensitive TAT standards was prevented when prepared in
0.025 M phosphate buffer (pH 7) as follows. The phosphate
buffer was boiled after preparation and cooled down on ice
under a constant purge of nitrogen to keep the solution free
of oxygen. The oxygen free buffer solution was transferred
into the anaerobic glove box and aliquots were added to
15 ml vials containing defined amounts of TAT*3HCl pow-
der. The vials were capped with polytetrafluoroethylene
(PTFE) coated butyl rubber septa and crimp-sealed. The
headspace was finally replaced with pure nitrogen. For anal-
ysis, aliquots of these solutions were transferred to HPLC
vials in the glove box and analyzed immediately. The TAT
concentration in the stock solutions was constant for at least
10 days (data not shown).
2.2. HPLC
2.2.1. Apparatus
HPLC analyses were performed using a Hewlett-Packard
1090 Liquid Chromatograph equipped with an original
autosampler and diode array detector. The Agilent Chem-
Station software (Rev. A.08.01 [783]) was utilized for
instrument control, data acquisition, and analysis. Although
multiple Supelcosil octyl (C-8) 150 mm × 4.6 mm (5
m
particle size) columns were used during the method devel-
opment, all data published herein were obtained using the
same column. Guard columns used for the method devel-
opment included Supelcosil LC-CN, LC-ABZ+, and LC-8
(20 mm × 4.6 mm; 5
m), but only the LC-8 guard column
was used for the final data collection. The injection volume
was 10
l, with an initial syringe draw speed of 83
l/min.
2.2.2. Column temperature, diode array detection, and
mobile phases
The column temperature was controlled using the block
heater built into the HPLC system (circulating heated air
around the column). Stainless steel tubing (35 cm long,
0.17 mm i.d.) was placed upstream of the column in-
side the oven compartment to ensure that the inlet sol-
vent was acclimatized in order to minimize potential
instrument-to-instrument variation [25,26]. Oven tempera-
tures from 35 to 55 ◦C were investigated.
Chromatograms were extracted at absorbencies of 220,
230, 254, 360, and 370 nm. Peaks were scanned from 200
to 600 nm to obtain spectrochromatograms for compound
characterization.
86 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
Two mobile phases were utilized to establish the gra-
dient system. The organic mobile phase was HPLC-grade
methanol. The aqueous mobile phase consisted of either
0.025 M sodium phosphate buffer (pH 7) with or without
0.1% or 0.5% w/w 1-octanesulfonic acid (ion-pair reagent),
or of a phosphate buffer (0.0144 M Na2HPO4) amended with
Low-UV PIC B8 resulting in a 0.1% ion-pair reagent solu-
tion with a pH of 7. No significant differences were observed
for the two different aqueous ion-pair reagent mobile phases.
Since the use of the Low-UV PIC B8 ion-pair reagent re-
sulted in significant time savings, all results presented herein
were acquired using this phosphate buffer solution. Mobile
phases were kept oxygen free by purging them with helium
for at least 30 min prior to the first sample run and continu-
ously throughout analyses.
2.2.3. Gradient
The flow rate of the mobile phase was 1 ml/min. The mo-
bile phase initially consisted of 99% phosphate buffer (with
or without 0.1% ion-pair reagent) and 1% methanol. By
utilizing the Agilent ChemStation “narrow gradient range”
option, the gradient was changed to 30% methanol over
2 min, then to 43% methanol over the next 13 min, finally in-
creased to 100% methanol over 12.5 min, and held constant
for 0.5 min. The solvent ratio was returned to the initial con-
ditions over 1 min and held for an additional 5 min before
injection of the next sample. The total run time including
conditioning time was 34 min.
2.3. Detection and peak performance parameters
Performance assessment was based on calibration stan-
dards of authentic compounds and multipoint standard
calibration curves. All compounds investigated had a lin-
ear detection range of at least 1–100 mg/l and were easily
detected in the range of 5–10 ng per injection. Standard
solutions with a concentration of 10–25 mg/l were used to
generate the chromatograms with the exception of the azo
compounds, which had a concentration of up to 40 mg/l.
Retention times (tr), retention time factors (k′), and the
chromatographic resolution (Rs) were calculated for all com-
pounds. k′ was calculated using a void time which was de-
termined from the column parameters supplied (k′ = (tr −
(Vm/F))/(Vm/F)), where Vm is the column void volume
(L3) and F the flow rate (L3/t). The porosity of the column
packing material was 0.71 based on information from Su-
pelco (Bellefonte, PA, USA). Rs and number of theoretical
plates (N) were calculated using the half-width method.
3. Results and discussion
3.1. Analysis of TNT metabolites
Separation of TNT and its reduced metabolites has been
problematic due to the co-elution of the isomers 2-ADNT
and 4-ADNT, difficulties with the quantification of ionizable
intermediates such as TAT due to speciation, and poor reten-
tion [21]. Octanesulfonic acid has been used as an ion-pair
reagent to increase the retention time of the highly polar
TAT [13].
The development of a gradient elution method for the
separation of the most commonly reported TNT metabolites
was accomplished after extensive investigation of various
types of chromatographic columns (including both C-8 and
C-18 packing materials), guard columns, and mobile phases
(at various pH values). This method was then optimized fur-
ther by the choice of temperature and addition of an ion-pair
reagent.
UV-Spectra of all investigated TNT metabolites were
compared to published spectra [21,24] for positive identifica-
tion, except for the UV-absorption spectra of 4,4′,6,6′-tetra-
nitro-2,2′-azotoluene and 2,2′,6,6′-tetranitro-4,4′-azotoluene,
which are to our best knowledge shown here for the first
time (Fig. 2). The highest UV absorption was obtained at
a wavelength of 230 nm for the majority of TNT metabo-
lites. However, TAT, DANTs, 4,4′-azoxy, and 4,4′-azo had
a higher absorption at 220 nm and 2,2′-azoxy and 2,2′-azo
a slightly increased absorption at 254 nm (Figs. 2 and 3a)
[21,24]. Most chromatograms shown herein are given at
254 nm unless otherwise noted, since it is the wavelength
achievable on most HPLC-UV systems and the detector
wavelength suggested by the EPA [27].
3.1.1. Influence of temperature and ion-pair reagent
The impact of temperature, on the separation of TNT
metabolites in the absence of an ion-pair reagent is shown
in Fig. 3a and b and in the presence of ion-pair reagent
in Fig. 4. Although a temperature increase from 35 to
55 ◦C decreased the overall separation time by less than
one minute, the retention time of compounds such as the
Fig. 2. UV-Spectra of 4,4′,6,6′-tetranitro-2,2′-azotoluene (dashed line) and
2,2′,6,6′-tetranitro-4,4′-azotoluene (solid line).
T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94 87
Fig. 3. (a) Optimal separation of TNT metabolites in the absence of an ion-pair reagent at 37 ◦C (1: 2,6-DANT; 2: 2,4-DANT; 3: 2-HADNT; 4: TNT;
5: 4-HADNT; 6: 2-ADNT; 7: 4-ADNT; 8: 2,2′-Azoxy; 9: 4,4′-Azoxy; 10: 2,2′-Azo; 11: 4,4′-Azo). Note the improved separation of the azoxy and azo
compounds with increased temperature and baseline separation of all compounds at 52.5 ◦C (see inset). 37 ◦C was chosen as the best column temperature
due to the potential for larger instrument-to-instrument variation at higher temperatures. (b) The retention time (tr) as a function of the absolute temperature
for selected compounds.
ADNTs decreased by approximately 4 min which resulted in
a drastically improved overall separation of the compounds.
The isocratic retention as a function of temperature can
often be described by the van’t Hoff relationship (log k′ =
a+b/T ; a and b are constants for a given compound and T is
the absolute temperature) [28]. The following empirical re-
lationship, tr = a′+b′(1/T) where a′ and b′ are constants for
a given solute as T is varied and all other conditions are kept
constant, was derived by Zhu et al. [29] from the van’t Hoff
relationship based on assumptions such as k′0 	 1 (value of
k′ at start of separation) and that S does not vary with tem-
perature; S is a solute parameter (please refer to Zhu et al.
[29] for further details). A linear relationship (R2 > 0.99)
was obtained when plotting the tr as a function of 1/T for the
herein investigated compounds. The following order of tem-
perature sensitivity was observed based on the calculated
88 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
min5 10 15 20 25
mAU
0
10
20
30
40
50
60
70
80
TA
T
1
2
IP
3
4
5
6
7
8
9
10
11
220 nm min10 15 20 25
4,3
5
6 7
8
10
9 11
Tc = 35 ˚C
Tc = 44 ˚C
Fig. 4. Optimal separation of TNT metabolites in the presence of an ion-pair reagent at 44 ◦C; the peak labeled “IP” is an inherent peak caused by
the addition of the ion-pair reagent (1: 2,6-DANT; 2: 2,4-DANT; 3: 2-HADNT; 4: TNT; 5: 4-HADNT; 6: 2-ADNT; 7: 4-ADNT; 8: 2,2′-Azoxy; 9:
4,4′-Azoxy; 10: 2,2′-Azo; 11: 4,4′-Azo). Note the co-elution of TNT and 2-HADNT at 35 ◦C (see inset).
slopes (b′): 4-ADNT (b′ = 21 738) > 2-ADNT (b′ = 21 231)
> HADNTs	 TNT	 2,4-DANT (b′ = 3581) > 2,6-DANT
(b′ = 2464) (Fig. 3b). The linear relationship for the studied
explosives can potentially be used to predict the retention
time as a function of temperature and thereby help improv-
ing the separation of closely eluting compounds, which is
in agreement with detailed studies by Zhu et al. [29,30].
In a recent review by Dolan [25], it was reported that a
change of temperature especially influenced the selectivity
of ionizable and polar compounds due to a possible con-
current change of the pKa, which is consistent with the
results observed in this study. The impact of temperature
on the DANTs was attenuated compared to the ADNTs
possibly due to a significantly higher pKa value or other
physical–chemical interactions, which were not further in-
vestigated as part of this study [25,29–32]. However, when
comparing the relative shift in retention time of the polar
TAT with TNT and 4-ADNT as a result of an increase in col-
umn temperature from 37 to 52.5 ◦C similar relative shifts
were observed (i.e. 13, 16, and 19%, respectively).
The increased temperature sensitivity of 2- and 4-HADNT,
as compared to TNT, was used to optimize the peak sep-
aration (Fig. 3b). Most compounds were separated with
baseline resolution (Rs ≥ 1.5) at a temperature of 37 ◦C
except for 2,2′-azo and 4,4′-azo which had Rs values of 1.3
and 1.2, respectively (Fig. 3a). The resolution of the azo
isomers was improved to Rs ≥1.4 by increasing the column
temperature to 52.5 ◦C (inset in Fig. 3a and Table 1). How-
ever, 37 ◦C was chosen as the best-suited chromatography
temperature due to potentially increased relative standard
deviation, and potentially larger instrument-to-instrument
variation at elevated column temperatures (Table 1) [26].
Increased temperature decreased the peak width (W) of
all peaks and increased the number of theoretical plates (N)
for the majority of peaks. In a few cases, a decrease of
N was observed. In the absence of an ion-pair reagent for
instance, a temperature increase from 35 to 52.5 ◦C resulted
in an approximately 33% increase in N for 2,6-DANT, but a
13% decrease in N for TNT. N should in theory increase as
temperature increases, however, experimental studies have
shown that this is not always the case in gradient elution
[29].
An ion-pair reagent has often been used for the chro-
matography of TNT metabolites in order to prolong the
retention time of the polar TAT [13]. The addition of an
ion-pair reagent to the aqueous mobile phase in the present
study resulted in the co-elution of 2-HADNT and TNT at
the commonly used column temperature of 35 ◦C (inset in
Fig. 4). However, the co-elution was completely avoided by
increasing the temperature to 44 ◦C (Fig. 4).
A relatively low shift in retention time was observed in
the presence of an ion-pair reagent compared to previously
published results with Supelcosil C-18 and ABZ+ columns
(25 cm by 4.6 mm; 5
m particles) [13,24]. The relatively
low shift in tr compared to these results could be due to a
lower ion-pair reagent affinity to the (15 cm long) C-8 col-
umn packing material, differences in the mobile phase, or
differences in the gradient elution. However, the exact rea-
sons remain unknown and were beyond the scope of this
investigation. The retention time of TNT decreased by 2.6%
in the presence of the ion-pair reagent possibly due to a
change of polarity in either the mobile phase or the station-
ary phase. The temperature sensitivity in the presence of
an ion-pair reagent decreased by 8–17% for all compounds
T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94 89
Table 1
Retention time factor (k′) and peak resolution (Rs) calculated for the TNT
metabolites (37 ◦C; Fig. 3a), EPA Method 8330 compounds (50.5 ◦C;
Fig. 7a), and TNT metabolites combined with the EPA Method 8330
compounds (50.5 ◦C; Fig. 8a)
Compound
(n = 6)
TNT
metabolitesa
EPA
compoundsa
Metabolites and
EPA compoundsa
k′b Rsb k′b Rsb k′b Rsb
TATc 1.05 >1.5 0.83 >1.5
2,4-DHANTd 1.95 n.a.
2,6-DANT 2.05 >1.5 1.91 >1.5
HMX 1.89 >1.5 1.91 0.0
2,4-DANT 2.32 >1.5 2.11 >1.5
RDX 2.74 >1.5 2.74 >1.5
1,3,5-TNB 3.46 >1.5 3.47 >1.5
1,3-DNB 4.17 >1.5 4.19 >1.5
NB 4.94 >1.5 4.97 >1.5
2-HADNT 6.27 >1.5 5.08 0.82
Tetryl 5.27 >1.5 5.28 >1.5
4-HADNT 6.77 >1.5 5.49 1.4
TNT 6.51 >1.5 5.61 >1.5 5.63 0.94
2-ADNT 7.27 >1.5 5.89 >1.5 5.91 >1.5
4-ADNT 7.53 >1.5 6.13 >1.5 6.14 >1.5
2,4-DNT 6.36 >1.5 6.38 >1.5
2,6-DNT 6.60 >1.5 6.62 >1.5
2-NT 7.52 >1.5 7.55 >1.5
4-NT 7.73 1.3 7.75 1.3
3-NT 8.06 >1.5 8.08 >1.5
2,2′-Azoxy 11.7 >1.5 11.4 >1.5
4,4′-Azoxy 12.2 >1.5 11.8 >1.5
2,2′-Azo 12.3 1.3 11.9 1.4
4,4′-Azo 12.4 1.2 12.0 >1.5
a Ranges of relative standard deviation (R.S.D.) of the retention time
(tr): T = 37 ◦C (0.01–0.06%); T = 50.5 ◦C (0.01–0.17%).
b Relative standard deviation (R.S.D.) of k′ and Rs 
 1% for all
compounds.
c TAT was measured at 220 nm and run separately (n = 7).
d 2,4-DHANT was run separately because the chemical was obtained
just before submission of this manuscript (n = 3).
based on comparisons of the calculated slopes (b′) (4-ADNT
(b′ = 19 985) > 2-ADNT (b′ = 19 483) > 4-HADNT (b′ =
18 103) > 2-HADNT (b′ = 16 269) > TNT (b′ = 12 014) >
2,4-DANT (b′ = 3031) > 2,6-DANT (b′ = 2054)) with the
calculated slopes in the absence of an ion-pair reagent (see
above and Fig. 3b).
The use of an ion-pair reagent for improved compound se-
lectivity often has an unfavorable impact on various factors
such as slow column equilibration and method ruggedness,
however it is still one of the favored ways to increase the
selectivity factor (α) of ionizable compounds in RP-HPLC
[29]. The use of an ion-pair reagent in this study resulted
in the appearance of an inherent peak (IP). In this case, the
inherent peak did not interfere with other peaks of interest
(e.g. IP observed at 8.45 min in Fig. 4), however the pres-
ence of such an IP could complicate the detection of TNT
related compounds in certain situations. Since the overall
chromatography was not improved by use of an ion-pair
reagent in the aqueous mobile phase, it is suggested to per-
form the chromatography in its absence, which will result
in a significant cost reduction.
3.1.2. Method performance
An acceptable level of reproducibility must be established
before any separation method can be applied to the analysis
of intricate environmental samples. Therefore, the chromato-
graphic performance and reproducibility of the developed
gradient method for the analysis of TNT and its metabo-
lites was investigated at 37 and 52.5 ◦C (data not shown),
which had demonstrated optimal separation in the absence
of an ion-pair reagent, and at 44 ◦C (data not shown) in the
presence of an ion-pair reagent. A mixture of TNT and its
metabolites (solubilized in acetonitrile) was repeatedly in-
jected and monitored for variability in retention time (tr),
retention time factor (k′), and peak resolution (Rs) (Table 1).
TAT was run separately, due to its poor solubility and sta-
bility in acetonitrile.
The retention time factors (k′) for the less polar azoxy and
azo compounds were significantly higher than for the other
metabolites (Table 1). However, in order to obtain sufficient
separation of the azoxy and azo compounds, a slow increase
of the methanol concentration was necessary. If the chro-
matography of the azo or both the azo and azoxy compounds
is not desired, the total run time can be significantly de-
creased by use of a steeper final gradient. If the chromatog-
raphy of the azo compounds is not desired, the methanol
concentration can be increased to 100% over 8.5 min instead
of 12.5 min, if neither azo nor azoxy compounds are of inter-
est, the methanol concentration can be increased over 1 min.
Very good reproducibility and performance was obtained
under all three conditions, however a slightly increased vari-
ation in the retention times was observed at 52.5 ◦C (range
of R.S.D.: 0.02–0.27%; n = 3). The retention time factors
were less than 12.4 and the resolution was better than 1.2
for all compounds with baseline resolution (Rs ≥ 1.5) for
most of the investigated compounds (Table 1).
Since sorption of an ion-pair reagent to the column pack-
ing material can change the characteristics of a column, the
reproducibility of the presented results was tested by re-
confirming the quality of separation with a new Supelcosil
LC-8 column. The new column had never been exposed to
ion-pair reagent. The retention times of the 14 EPA 8330
compounds at 50.5 ◦C changed by less than 2.2% between
the columns and the resolution was of the same quality as
with the LC-8 column that had been exposed to ion-pair
reagent. The slight change in tr may be due to column aging
rather than sorption of the ion-pair reagent.
In order to test the developed method in a real case sce-
nario, samples were taken from a TNT (bio)transformation
experiment ([33], and unpublished data) and analyzed us-
ing the developed gradient method at 37 ◦C. No significant
retention time shift was observed over a period of 86
days after analysis of more than 500 samples from var-
ious TNT degradation studies using the same Supelcosil
LC-8 chromatography column. The two chromatograms in
Fig. 5 demonstrate the degradation of TNT by strain ES6
in the presence of hydrous ferric oxide and the electron
shuttling compound 9,10-anthraquinone-2,6-disulfonate
90 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
Fig. 5. Randomly selected chromatograms from the analysis of aqueous samples of a TNT (bio)degradation experiment [33]. Test tubes contained
synthetic ground water [38], carbonate buffer (pH 7), hydrous ferric oxide, 9,10-anthraquinone-2,6-disulfonate (AQDS), sucrose, and strain ES6, tentatively
identified as a Cellulomonas sp. [33]. The test tubes were injected with 52
M of TNT after 14 days of inoculation and TNT transformation and
metabolite patterns were observed over time. The solid line and the dotted line show TNT and its metabolites after 3 days and 86 days of incubation,
respectively. The chromatograms are shown at 230 nm and were obtained using the developed gradient elution method at 37 ◦C. M: microbially related
metabolites, U: unidentified peaks, I: sample inherent peaks.
(AQDS). TNT was completely transformed via the HAD-
NTs and ADNTs metabolites to 2,4-DANT and TAT after
86 days.
3.2. Analysis of EPA Method 8330 explosives
Fourteen explosives, octahydro-1,3,5,7-tetranitro-1,3,5,7-
tetrazocine (HMX), hexahydro-1,3,5-trinitro-1,3,5-triazine
(RDX), 1,3,5-trinitrobenzene (TNB), 1,3-dinitrobenzene
(DNB), nitrobenzene (NB), TNT, 2,4-dinitrotoluene (2,4-
DNT), 2,6-dinitrotoluene (2,6-DNT), 2,4,6-trinitrophenyl-
methylnitramin (Tetryl), 2-ADNT, 4-ADNT, 2-nitrotoluene
(2-NT), 3-nitrotoluene (3-NT), and 4-nitrotoluene (4-NT)
are included in the EPA Method 8330 [34].
The main difficulty with the isocratic EPA Method 8330 is
the reported co-elution of the DNTs and ADNTs when em-
ploying the recommended primary C-18 (250 mm×4.6 mm,
5
m) RP-HPLC column [3,27,34]. Consequently, a second
confirming run is needed for the separation of the DNTs
and ADNTs, which is commonly performed by utilizing a
CN (250 mm × 4.6 mm, 5
m) RP-HPLC column [27,34].
The co-elution problem of DNTs and ADNTs was solved in
a recent study by use of a two-phase approach in a single
RP-HPLC run. Lang and Burns [3] used a CN guard column
in series with a C-18 column, while keeping the EPA Method
8330 specifications. Lang and Burns [3] demonstrated that,
although, the two-phase approach resulted in a total run time
of approximately 32 min compared to approximately 24 min
for the EPA Method 8330, it only required one HPLC run
since the co-elution of the DNTs and ADNTs was avoided
[3,34].
The goal of this part of the study was to develop an al-
ternative method, that would decrease the total run time, re-
duce the consumption of solvents (mobile phases) compared
to the discussed methods, and still baseline separate all 14
EPA Method 8330 compounds.
3.2.1. Influence of temperature and ion-pair reagent
Baseline resolution was achieved for all compounds
(except for 4-NT; Rs = 1.3) by optimizing the column tem-
perature of the described gradient elution method. Fig. 6
illustrates how the increased temperature sensitivity of the
ADNTs can be used to separate them from the DNTs.
4-ADNT and 2,4-DNT co-eluted at 42 ◦C but were com-
pletely separated at 50 ◦C. The higher temperature sensitiv-
ity of the ADNTs might be due to the ionizable character of
these compounds [25]. An optimum temperature of 50.5 ◦C
was found to rapidly (18 min), reliably (R.S.D. < 0.2%),
and fully separate all the EPA Method 8330 compounds
(Fig. 7a and Table 1).
The flow rate in this study was 1 ml/min as compared to
1.5 ml/min used in the EPA Method 8330 and the method
proposed by Lang and Burns [3,34]. Additionally, the total
run time in this study was reduced by approximately 7 min
(≈22%) compared to Lang and Burns [3], based on a 7 min
column conditioning time following the elution of 3-NT. The
combined decrease in total run time and flow rate resulted in
an approximately 48% reduction in solvent consumption as
compared to Lang and Burns and a 30% reduction compared
to the EPA Method 8330 [3,34].
Optimal chromatographic separation in the presence of
an ion-pair reagent was obtained at 41.5 ◦C (Fig. 7b). The
T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94 91
Tc = 50 °C 
Tc = 42°C 
Tc = 47 °C 
Fig. 6. Influence of temperature (42, 47, and 50 ◦C) on the separation of selected EPA Method 8330 chemicals in the absence of an ion-pair reagent.
Note the co-elution of 4-ADNT and 2,4-DNT at 42 ◦C and the improved separation with increased temperature.
Fig. 7. (a) Baseline separation of EPA Method 8330 compounds at the optimized temperature of 50.5 ◦C (no ion-pair reagent added). (b) Separation of
the fourteen EPA Method 8330 chemicals in the presence of an ion-pair reagent at the optimized temperature of 41.5 ◦C. Note how the ion-pair reagent
prevents baseline separation of several compounds.
92 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
Fig. 8. (a) Separation of the 23 TNT metabolites and EPA Method 8330 compounds at 50.5 ◦C in the absence of an ion-pair reagent. (b) Separation at
41.5 ◦C in the presence of an ion-pair reagent. Note the slightly improved separation of 2,6-DANT and HMX in the presence of an ion-pair reagent but
the overall decreased chromatographic resolution. “A” is a possible transformation product (probably 2,4′,6,6′-tetranitro-2′,4-azoxytoluene, see text for
discussion) and “IMP” is an inherent impurity contained in the 4-HADNT standard.
temperature selectivity of the ADNTs relative to the DNTs
was decreased in the presence of an ion-pair reagent, ex-
emplified by the reverse elution order of 4-ADNT and
2,4-DNT (Fig. 7a and b). The affinity of the ion-pair
reagent for the ADNTs resulted in increased retention times
and consequently a poor separation of 2-ADNT, 2,4-DNT
(Rs = 0.83), and 4-ADNT (Rs = 0.63) (Fig. 7b). Hence,
it is not recommendable to use an ion-pair reagent for the
separation of compounds included in the EPA Method 8330.
3.3. Combined analysis of TNT metabolites and EPA
Method 8330 compounds
Soils and water contaminated with TNT and its metabo-
lites often contain co-contaminants such as the compounds
included in EPA Method 8330 [18,27]. Thus, it is impor-
tant to minimize the potential for co-elution of chemicals
to prevent false positive results when analyzing complex
environmental samples. It was therefore tested whether the
T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94 93
developed gradient method would be capable of separating
all 23 chemicals investigated in this study ranging from the
very polar TAT to the less polar nitrotoluene, azoxy, and azo
compounds in a single run.
Various temperatures were investigated (data not shown)
and it was found that the same temperature used for the op-
timal separation of EPA Method 8330 compounds (50.5 ◦C)
gave the most satisfying chromatographic separation (Fig. 8a
and Table 1). 2,6-DANT and HMX co-eluted and 2-HADNT
and TNT had a peak resolution of less than 1. Neverthe-
less, reliable identification was obtained for the majority of
the investigated compounds including the 7 pairs of isomers
contained in the analyte mixture (Table 1).
Decreasing the steepness of the proposed initial methanol
gradient can prevent the co-elution of 2,6-DANT and
HMX. However, this will result in a poor separation of
compounds with longer retention times (data not shown).
The use of an ion-pair reagent in the aqueous mobile
phase at 41.5 ◦C allowed for a slightly better separation
of 2,6-DANT and HMX (Rs ∼ 0.7), however, it also re-
sulted in an overall decreased chromatographic resolution
(Fig. 8b).
The peak labeled IMP was an inherent impurity (IMP)
originating from the 4-HADNT standard. The peak labeled
A in the chromatogram is a possible transformation prod-
uct, which accumulated over time. It is believed, based on
previous studies [5,35], that the degradation product (A)
originated from the spontaneous dimerization of the oxy-
gen sensitive HADNTs potentially producing an additional
azoxy-isomer (e.g. 2,4′,6,6′-tetranitro-2′,4-azoxytoluene or
2′,4,6,6′-tetranitro-2,4′-azoxytoluene). This was supported
by an abiotic experiment investigating the resulting trans-
formation product after oxygen-purging a standard mixture
containing only 2- and 4-HADNT by comparison of the
DAD spectra (data not shown).
4. Conclusion
This study demonstrates for the first time the importance
of optimizing the temperature for the improved separa-
tion of complex mixtures containing explosives-related
compounds. The findings herein evoke that column tem-
perature should not always be used as the last parameter
to optimize RP-HPLC methods. Additionally, this work
is a supplement to research illustrating the importance of
column temperature for the improved separation of com-
pounds like chlorophylls, herbicides, peptides, and drugs
(e.g. anticancer agents) [25,30,36,37].
The developed gradient method is unique because the
same method at different temperatures can be used to com-
pletely separate TNT and 12 of its reduced metabolites as
well as the compounds targeted in the EPA Method 8330.
The TNT metabolites included 2,4,6-triaminotoluene (TAT),
the DANTs, ADNTs, HADNTs, 2,4-DHANT, tetranitroa-
zoxytoluenes, and tetranitroazotoluenes.
The proposed chromatographic method for the EPA
Method 8330 compounds does not only provide improved
separation of the 14 target compounds but also a significant
reduction of total run time and solvent consumption when
compared to previous studies [3,27,34].
The use of an ion-pair reagent for increased selectivity
was also investigated. However, the use of the costly ion-pair
reagent could be totally avoided by optimizing the column
temperature.
Finally, the gradient method proved to be capable of sep-
arating all 23 explosives-related compounds, including the
12 reduced TNT metabolites tested, and the EPA Method
8330 compounds in a single run except for 2,4-DANT and
HMX, which co-eluted. Thus, the gradient elution method
described here can become a valuable tool for the fast and
reliable analysis of complex samples containing mixtures of
nitroaromatics, aminoaromatics, and nitramines.
Acknowledgements
The authors thank the US Department of Defense, Army
Research Office, Grant No. DAAD19-99-1-0092 for their
support. Partial financial support was provided by the Inland
Northwest Research Alliance. The authors thank J. Neu-
man, J. Harwood, and C. Sedlak (Montana State University)
for their assistance, J. Hawari and L. Paquet (Canadian
Biotechnology Research Institute) for valuable discussions
of chromatography issues, J. Hughes (Rice University) for
providing the 2,4-DHANT, S. Sporring (Lund University),
B. Svensmark (University of Copenhagen) for insightful
comments and the two anonymous reviewers for their valu-
able suggestions.
References
[1] J.C. Spain, J.B. Hughes, H.-J. Knackmuss, Biotransformation of
Nitroaromatic Compounds and Explosives, Lewis Publishers, 2000.
[2] J.C. Pennington, J.M. Brannon, Thermochim. Acta 384 (2002) 163.
[3] M.J. Lang, S.E. Burns, J. Chromatogr. A 849 (1999) 381.
[4] C.A. Groom, S. Beaudet, A. Halasz, L. Paquet, J. Hawari, J. Chro-
matogr. A 909 (2001) 53.
[5] C.Y. Wang, D. Zheng, J.B. Hughes, Biotechnol. Lett. 22 (2000) 15.
[6] A. Schmidt, W. Butte, Chemosphere 38 (1999) 1293.
[7] M.G. Musolino, G. Neri, C. Milone, S. Minico, S. Galvagno, J.
Chromatogr. A 818 (1998) 123.
[8] N. Cenas, A. Nemeikaite-Ceniene, E. Sergediene, H. Nivinskas, Z.
Anusevicius, J. Sarlauskas, Biochim. Biophys. Acta 1528 (2001) 31.
[9] B. Lachance, P.Y. Robidoux, J. Hawari, G. Ampleman, S. Thiboutot,
G.I. Sunahara, Mutat. Res. 444 (1999) 25.
[10] D.H. Rosenblatt, E.P. Burrows, W.R. Mitchell, D.L. Parmer, in:
O. Hutzinger (Ed.), The Handbook of Environmental Chemistry,
Springer-Verlag, Berlin, 1991, pp. 195–234.
[11] J. Hawari, A. Halasz, S. Beaudet, L. Paquet, G. Ampleman, S.
Thiboutot, Appl. Environ. Microbiol. 65 (1999) 2977.
[12] S.G. Pavlostathis, G.H. Jackson, Water Res. 36 (2002) 1699.
[13] J. Hawari, A. Halasz, L. Paquet, E. Zhou, B. Spencer, G. Ampleman,
S. Thiboutot, Appl. Environ. Microbiol. 64 (1998) 2200.
94 T. Borch, R. Gerlach / Journal of Chromatography A, 1022 (2004) 83–94
[14] P.M. Gates, E.T. Furlong, T.F. Dorsey, M.R. Burkhardt, Trends Anal.
Chem. 15 (1996) 319.
[15] T.C. Schmidt, D. Meinzer, L. Kaminski, E.v. Löw, G. Stork, Chro-
matographia 46 (1997) 501.
[16] M.E. Walsh, Talanta 54 (2001) 427.
[17] M. Markuszewski, J.D. Krass, T. Hippe, B. Jastorff, R. Kaliszan,
Chemosphere 37 (1998) 559.
[18] A. Halasz, C. Groom, E. Zhou, L. Paquet, C. Beaulieu, S. Deschamps,
A. Corriveau, S. Thiboutot, G. Ampleman, C. Dubois, J. Hawari, J.
Chromatogr. A 963 (2002) 411.
[19] U. Lewin, J. Efer, W. Engewald, J. Chromatogr. A 730 (1996) 161.
[20] X. Zhao, J. Yinon, J. Chromatogr. A 946 (2002) 125.
[21] F. Ahmad, D.J. Roberts, J. Chromatogr. A 693 (1995) 167.
[22] G. Daun, H. Lenke, M. Reuss, H.-J. Knackmuss, Environ. Sci.
Technol. 32 (1998) 1956.
[23] R. Bhadra, D.G. Wayment, J.B. Hughes, J.V. Shanks, Environ. Sci.
Technol. 33 (1999) 446.
[24] C. Groom, S. Beaudet, A. Halasz, L. Paquet, J. Hawari, Environ.
Sci. Technol. 34 (2000) 2330.
[25] J.W. Dolan, J. Chromatogr. A 965 (2002) 195.
[26] R.G. Wolcott, J.W. Dolan, L.R. Snyder, S.R. Bakalyar, M.A. Arnold,
J.A. Nichols, J. Chromatogr. A 869 (2000) 211.
[27] T.F. Jenkins, M.E. Walsh, P.W. Schumacher, C.F. Bauer, C.L. Grant,
J. Assoc. Off. Anal. Chem. 72 (1989) 890.
[28] W. Melander, D.E. Campbell, Cs. Horváth, J. Chromatogr. 158 (1978)
215.
[29] P.L. Zhu, L.R. Snyder, J.W. Dolan, N.M. Djordjevic, D.W. Hill, L.C.
Sander, T.J. Waeghe, J. Chromatogr. A 756 (1996) 21.
[30] P.L. Zhu, J.W. Dolan, L.R. Snyder, D.W. Hill, L. Van Heukelem,
T.J. Waeghe, J. Chromatogr. A 756 (1996) 51.
[31] J.W. Dolan, L.R. Snyder, N.M. Djordjevic, D.W. Hill, D.L. Saunders,
L. Van Heukelem, T.J. Waeghe, J. Chromatogr. A 803 (1998) 1.
[32] K.A. Thorn, K.R. Kennedy, Environ. Sci. Technol. 36 (2002)
3787.
[33] T. Borch, R. Gerlach, A.B. Cunningham, B.M. Peyton, W.A. Apel,
Influence of Biogenically Produced Fe(II) and Humic Acid Analogs
on the Fate of 2,4,6-Trinitrotoluene (TNT). Eos Trans. AGU 83[47],
B22E-11, 2002.
[34] EPA, SW 846 Method 8330: Nitroaromatics and Nitramines by
High Performance Liquid Chromatography (HPLC), Environmental
Protection Agency 8330, Washington, DC, 1994, pp. 1–21.
[35] A. Haidour, J.L. Ramos, Environ. Sci. Technol. 30 (1996) 2365.
[36] L.R. Snyder, J. Chromatogr., B: Biomed. Sci. Appl. 689 (1997)
105.
[37] H. Rosing, E. Doyle, J.H. Beijnen, J. Pharm. Biomed. Anal. 15
(1996) 279.
[38] J.N. Petersen, R.S. Skeen, K.M. Amos, B.S. Hooker, Biotechnol.
Bioeng. 43 (1994) 521.