Kinetics of the alkaline hydrolysis of high explosives RDX and HMX in aqueous solution and adsorbed to activated carbon

Article abstract of DOI:10.1021/es9504101

Aqueous alkaline hydrolysis of bulk quantities and wastewater contaminated with high explosives is a promising technology for treatment and disposal of the worldwide surplus of munitions. We investigated the hydrolysis kinetics of the high explosives hexahydro-1,3,5-trinitro-1,3,5-triazine (RDX) and octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX) for temperatures ranging from 50 to 80 °C and in the pH range from 10 to 12. The experimental data were described using a pseudo-first-order model with subsequent calculation of second-order rate constants from experiments with excess hydroxide concentration. The temperature dependency of the rate constants was evaluated using the Arrhenius model. The activation energies were determined to be E(RDX) = 99.9 ± 1.9 kJ mol^-1 and E(HMX) = 111.9 ± 0.8 kJ mol^-1. The rate of HMX hydrolysis is much slower than the rate of RDX hydrolysis and may become rate limiting in the treatment of RDX/HMX mixtures. The alkaline hydrolysis of RDX yields 1.6 M NO₂^-, 1.5 M HCOO^-, 0.1 M CH₃COO^-, 1.1 M HCHO (11), 0.9 M NH₃, 1.1 M N₂O, and 0.34 M N₂ per mol of RDX hydrolyzed. Acetate ion (CH₃COO^-) is a previously unknown end product of the alkaline hydrolysis of RDX. A mass balance showed a recovery of 94% carbon and 90% nitrogen. During GC/MS analysis of the end products, no further unknown products could be found. In batch desorption studies, it was also shown that RDX-laden activated carbon can be regenerated using alkaline hydrolysis (T = 80 °C, pH 12) and that the desorption of the hydrolysis products is complete.

Full text of DOI:10.1021/es9504101

Environ. Sci. Technol. 1996, 30, 1485-1492
Introduction
Kinetics of the Alkaline
Hydrolysis of High Explosives
RDX and HMX in Aqueous
Solution and Adsorbed to
Activated Carbon
The interest in military waste has recently increased due
to growing environmental concerns and the end of the Cold
War. The environmental fate and health impacts of high
explosives waste and their appropriate treatment tech-
nologies are yet to be identified. Byrd and Humphreys (1)
report a U.S. demilitarization inventory at the end of fiscal
year 1992 of 313 000 000 kg of high explosives (e.g., RDX,
hexahydro-1,3,5-trinitro-1,3,5-triazine; HMX, octahydro-
1
,3,5,7-tetranitro-1,3,5,7-tetrazocine; TNT, 2,4,6-trinitro-
†
H A R R O M . H E I L M A N N ,
toluene). The current nuclear weapons dismantling efforts
by the U.S. Department of Energy will generate another
‡
U D O W I E S M A N N , A N D
M I C H A E L K . S T E N S T R O M *
,
†
5
0 000 kg/ year of excess high explosives waste through the
Department of Civil and Environmental Engineering,
University of California, Los Angeles, California 90095-1600,
and Institut f u¨ r Verfahrenstechnik der Technischen Universit a¨ t
Berlin, D-10623 Berlin, Germany
year 2000. In addition to bulk high explosives, contami-
nated groundwater and process water from closed military
bases, former production plants, contaminated soil, and
weapon dismantling facilities will need to be treated.
Problems related to contaminated sites and demilitarization
activities also exist in Europe and the former Soviet Union,
with some of the sites dating back to World War II.
The heterocyclic nitroorganic compounds RDX (CAS
Registry No. 121-82-4) and HMX (CAS Registry No. 2691-
Aqueous alkaline hydrolysis of bulk quantities and
wastewater contaminated with high explosives is a
promising technology for treatment and disposal of
the worldwide surplus of munitions. We investigated
the hydrolysis kinetics of the high explosives hexahy-
dro-1,3,5-trinitro-1,3,5-triazine (RDX) and
4
1-0) are some of the most important high explosives. Both
are major components in nuclear and conventional weap-
ons. Today, they are widely used as part of high-
performance explosives in the United States, such as plastic
bonded explosives PBX-9011 (90% HMX), -9404 (94% HMX),
-9407 (94% RDX), and -9501 (95% HMX) (2) and LX-04-1
octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazocine (HMX)
for temperatures ranging from 50 to 80 °C and in
the pH range from 10 to 12. The experimental data
were described using a pseudo-first-order model
with subsequent calculation of second-order rate
constants from experiments with excess hydroxide
concentration. The temperature dependency of the
rate constants was evaluated using the Arrhenius
model. The activation energies were determined to
(
85% HMX), -07-2 (90% HMX), -09-0 (93% HMX), -10-0 (95%
HMX), -11-0 (80% HMX), and -14-0 (95.5% HMX) (3).
Recently, HMX has replaced its homologue RDX in im-
portance in the United States, and it is more prevalent in
recently manufactured explosive devices. This is due in
part to HMX’s greater yield and resistance to unwanted
detonation.
RDXand HMXhave important health and toxicityeffects.
RDXis a class C possible human carcinogen and has adverse
effects on the central nervous system in mammals (4). HMX
also has central nervous system effects but only in higher
concentrations (5). Generally, less is known about HMX,
and further chronic carcinogenicity and toxicity studies
are needed.
-1
be ERDX ) 99.9 ( 1.9 kJ mol and EHMX ) 111.9 (
-
1
0
.8 kJ mol . The rate of HMX hydrolysis is much
slower than the rate of RDX hydrolysis and may
become rate limiting in the treatment of RDX/HMX
mixtures. The alkaline hydrolysis of RDX yields 1.6 M
-
-
-
NO
11), 0.9 M NH
RDX hydrolyzed. Acetate ion (CH
2
, 1.5 M HCOO , 0.1 M CH
3
COO , 1.1 M HCHO
per mol of
COO ) is a previ-
Currently high explosives are being destroyed by open
burning/ open detonation (1, 6). In fiscal year 1992, 80%
ofthe U.S. Department ofDefense’s annualdemilitarization
tonnage of 56 000 000 kg was destroyed this way (1);
however, due to environmental and legal concerns, this
technology will soon be unavailable. Combustion of high
explosives can lead to various unwanted toxic products.
Holl and Schneider (7) documented a pyrolytic reaction
under criticalconditions (delayed ignition;low ignition heat;
accidental smolder fire) that leads to the formation of large
amounts of hydrocyanic acid (HCN) during RDX combus-
tion.
(
3
, 1.1 M N O, and 0.34 M N
2
2
-
3
ously unknown end product of the alkaline hydrolysis
of RDX. A mass balance showed a recovery of 94%
carbon and 90% nitrogen. During GC/MS analysis of
the end products, no further unknown products
could be found. In batch desorption studies, it was
also shown that RDX-laden activated carbon can
be regenerated using alkaline hydrolysis (T ) 80 °C,
pH 12) and that the desorption of the hydrolysis
products is complete.
Process waters or groundwaters contaminated with high
explosives from dismantling activities or contaminated sites
generally have been treated by adsorption onto activated
carbon (8). The laden carbon cannot be thermally regen-
erated because of explosion risk and must be disposed as
hazardous waste. Burning of the spent carbon has now
been banned for the same reasons as mentioned for open
*
Corresponding author telephone: (310)825-1408; fax: (310)206-
5
476; e-mail address: stenstro@seas.ucla.edu.
†
‡
University of California.
Institut f u¨ r Verfahrenstechnik der Technischen Universit a¨ t.
0
013-936X/96/0930-1485$12.00/0

1996 Am erican Chem ical Society
VOL. 30, NO. 5, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1 4 8 5

burning/ open detonation, and other disposal methods are
expected to become more difficult and more expensive due
to more stringent regulations.
The U.S. Departments of Energy and of Defense are now
combining their efforts to identify possible future tech-
nologies for demilitarization. Alkaline hydrolysis has been
identified as an alternative treatment technology; however,
additional basic research is necessary before commercial
or industrial facilities can be successfully constructed.
Several researchers have investigated the kinetics of the
alkaline hydrolysis of RDX (9-12). Hoffsommer et al. (11)
identified end products and a rate-determ ining E2-
elimination mechanism as the initial step of the alkaline
hydrolysis ofRDX. Identified end products and by products
-
-
are NO2 , N2, NH3, N2O, HCOO , CH2O, and H2. The
formaldehyde is subject to a Canizarro reaction at the
elevated pH and produces formate (13). The reaction was
also confirmed to be second-order with respect to RDX
FIGURE 1. Treatment scheme for waters contaminated with high
explosives.
carbon adsorption with solvent and biological carbon
regeneration.
-
and OH concentrations. Second-order rate constants were
calculated from pseudo-first-order rate data (for excess
OH ).
In this paper, we investigate the chemical kinetics of the
alkaline hydrolysis of HMX and RDX under aqueous
homogeneous conditions. Because the rates of the alkaline
hydrolysis increase at higher temperatures and pressures,
and these conditions seem more feasible for industrial
operation, we investigate the kinetics of the alkaline
hydrolysis of HMX and RDX at elevated temperatures
ranging from 50 to 80 °C. The pH varied on the range
-
Less research has been performed on the alkaline
hydrolysis ofHMX;however, we anticipate some similarities
to the alkaline hydrolysis of RDX. Epstein et al. (9)
concluded that the alkaline hydrolysis of HMX is second-
order. Croce et al. (12), in their study on rate enhancement
effects in the presence of cationic micelles, proposed that
the alkaline hydrolysis of HMX follows the same basic
mechanism as RDX with regard to the almost similar
activation parameters for HMX and RDX that were obtained
in these and the Hoffsommer et al. (11) rate studies.
Nevertheless, Croce et al. (12) report aqueous standard
solutions containing 6, 9, and 12 mg of HMX/ L, which are
above the solubility limit of 5 mg of HMX/ L (14), which we
confirmed. Possible heterogeneous effects in their study
need special interpretation. All these investigations were
at temperatures ranging from 0 to 45 °C.
1
0-12. The results are meaningful for treatment of
explosives-contaminated waters and bulk high explosives.
We also investigated the principal viability of activated
carbon regeneration by alkaline hydrolysis with batch
studies. Our research on a continuous process (Figure 1)
is ongoing.
Materials and Methods
Analytical Section. High-Perform ance Liquid Chrom a-
tography (HPLC). Dissolved HMX and RDX were analyzed
using high-performance liquid chromatography with a
variable wavelength UV-detector (HPLC/ UV, Hewlett-
Packard Series 1050) set to 236 nm. A mobile phase
consisting of40%water, 30%methanol, and 30%acetonitrile
Current research is under wayat the Los Alamos National
Laboratory. The emphasis in these investigations is on the
heterogeneous alkaline hydrolysis of solid HMX and other
bulk high explosives (15). Spontarelli et al. demonstrated
that many propellants, explosives, and pyrotechnics can
be base hydrolyzed at temperatures ranging from 60 to 150
(
vol %) was used at a flow rate of 1 mL/ min. (All solvents
were HPLC grade, Fisher Scientific, Springfield, NJ.) An
Adsorbosphere-C18 10-µm reversed-phase column (Alltech,
Deerfield, IL) with prefilter element and guard column (C18
5-µm, Alltech) was used. The injection volume was set to
20 µL using the autosampler. Peaks were detected at
retention times between 3.5 and 3.7 min (HMX) and
between 4.1 and 4.3 min (RDX). The peak area was a linear
function of the concentration between 0.1 and 5 mg/ L
(HMX) and between 0.1 and 40 mg/ L (RDX). The amount
of HMX impurity contained in the RDX was quantified. We
found 0.1 mg/ L to be the detection limit with this method.
For the external calibration, at least three replications were
performed for each standard concentration. The mean was
then used for the calibration curve. Allsamples were filtered
through sterile Acrodisc-13 0.2-µm syringe microfilters
(Gelman Sciences, Ann Arbor, MI) before injection.
Ion Chrom atographysForm ate and Nitrite. The an-
ions NO2- (nitrite) and HCOO (formate) were analyzed
using a Dionex ion chromatograph (basic chromatography
module CMB-2, gradient pump GPM-1; Dionex, Sunnyvale,
CA) with suppressed conductivity detection (conductivity
detector CDM-1). An Ion Pac AS9-SC column (4 mm i.d.)
°
C.
To develop an economical treatment scheme for HMX-
and RDX-contaminated waters, additional basic research
is necessary. HMX and RDX are sparingly soluble in water
(
solubility limits at 25 °C: ∼5 mg of HMX/ L [14] and ∼50
mg of RDX/ L [2]). Direct treatment of the wastewater with
base hydrolysis is not economical. An alternative treatment
process using granular activated carbon (GAC) adsorption
of HMX and RDX and subsequent off-line regeneration of
the laden GAC through alkaline hydrolysis is a potentially
viable process (Figure 1). The contaminated water is first
treated through HMX and RDX adsorption onto GAC, which
concentrates the contaminants on the carbon and reduces
the volume ofmaterialto be treated with alkaline hydrolysis.
The laden carbon is regenerated with alkaline hydrolysis
and can be reused to treat another charge of contaminated
water. The end products are thought to be low molecular
weight compounds that can easily be degraded in regular
wastewater treatment plants with nitrification/ denitrifi-
cation stages. This treatment scheme is complementary
to the earlier proposal of Hesselmann et al. (16) that used
-
1
4 8 6
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 5, 1996

and a suppressor column were applied. The mobile phase
consisted of 0.75 mM NaHCO3 and 2 mM Na2CO3 dis-
solved/ LofMilliq water at a flow rate of 2 mL/ min. Samples
were manually injected into a 50-µL sample loop. Peaks
were detected at retention times between 1 and 1.1 min
h prior to the start of an experiment. From a stock solution
(approximately 4 mg of HMX/ L or 36 mg of RDX/ L,
respectively) 1000 mL was placed into a reaction vessel
(1000-mL Erlenmeyer) and heated in the water bath until
constant temperature was reached. The mixture was stirred
by an overhead motor stirring unit with a stainless steel
-
-
(
HCOO ) and between 1.7 and 1.8 min (NO2 ), respectively.
-
The peak area was a linear function of the concentration
mixer. The appropriate amounts of OH in order to reach
-
between 0.33 and 13.24 mg/ L (HCOO ) and between 0.33
a pH of 10-12 were added using a Eppendorf adjustable
pipettor or a common pipet (23 mM NaOH with HMX); 1
M NaOH was used for the experiments with HMX (0.22,
-
and 13.33 mg/ L (NO2 ). We found 0.01 mg/ L to be the
detection limit with this method at neutral pH. For the
external calibration, at least three data points were gathered
for each standard concentration. The mean was then used
for the calibration curve. All samples were filtered through
sterile Acrodisc-13 0.2-µm syringe microfilters (Gelman
Sciences) before injection. We found an influence of the
-
2.1, 23 mM OH ); 1 M and 10 M NaOH was used with RDX
(1.5, 2.1, 10, 20 mmol); the volume change was evaluated
for 23 mmol. A precision scientific timer was set simul-
taneously to the start of an experiment. Samples of 1 mL
were taken at predefined times using 1-mL microsyringes
and attached needles; sample collection time was 2 s;
samples were immediately filtered into cooled (0 °C) HPLC
vials. The pH was monitored and did not vary more than
(1%.
Am m onia Experim ents. The ammonia concentration
could not be determined in the experiments above because
of the volatile character of ammonia (particularly at the
elevated pH and temperature). To determine the ammonia
production, separate experiments were performed: 40 mL
of an aqueous RDX solution was put into a vial and 80 µL
of 10 M NaOH (pH 12) or 80 µL of 1 M NaOH (pH 11),
respectively, were added. The vials were immediatelysealed
with a Teflon-lined cap and put into a water bath at constant
temperature. Selected vials were then taken from the water
bath at precise time intervals and immediately cooled in
an ice-bath (T ) 0 °C). The ammonia concentration was
then measured using an ammonia electrode.
Gas Analysis. A 5-g sample of RDX was put into a gas-
tight stainless steel reactor (Parr-Instruments, Moline, IL);
the total reactor volume was 465 mL. The system was then
sealed and evacuated with a vacuum system. Subsequently,
50 mL of 1.5 M NaOH were added into the sealed system.
The solution was kept at a constant temperature (T ) 90
°C) and continuously stirred. Gas samples were taken every
30 min until equilibrium was reached in the gas phase. The
system pressure was measured. The total head space
volume was 0.412 L. The gas samples were analyzed with
a CEC 21-104 mass spectrometer (CEC, now Du Pont
Corporation, Chicago, IL).
-
-
pH value in the determination of HCOO and NO2
concentrations with the described method. Therefore,
standard curves were obtained for neutral pH and also pH
1
1 and pH 12.
Ion Chrom atographysAcetate. Acetate ion (CH3COO-)
analysis was performed on a Dionex 2110i ion chromato-
graph with an AS10 column and an AG10 guard column.
The eluent was 70 mM NaOH at 1 mL/ min. A suppressed
conductivity detector was applied; sample size was 50 µL.
The retention time for acetate ion with this method was 4.3
min. Standard curves were obtained for 2, 4, 6, 8, 12, 16
mg/ L. The pH had no influence on the detection with this
method in the observed pH range from neutral to pH 12.
Gas Chrom atography/Mass Spectrom etry (GC/MS). A
Finnigan GC/ MS system, consisting of a Finnigan gas
chromatograph Model 9610 equipped with a Grob-type
splitless injector and a Finnigan quadrupole mass spec-
trometer Model 4000 with an INCOS Model 2300 data
system, was applied (Finnigan, Sunnyvale, CA). The GC/
MS was operated with an electron energy of 70 eV, a source
-
1
temperature of 240 °C, and a scan speed of 1 s scan from
5
0 to 550 amu. A 30-m DB5-MS (0.25 mm i.d., 25 µm film
thickness, J&W Scientific, Folsom, CA) fused silica column
was programmed for 4 min at 30 °C, then up to 300 °C at
6
°C/ min, and hold at 300 °C for 60 min. The carrier gas
was helium at a flow rate of 35 cm/ s.
Liquid-Liquid-Extraction (LLE). Prior to GC/ MS analy-
sis, samples were extracted and concentrated with dichlo-
romethane (Optima-grade, Fisher Scientific). Atotalof1000
mL of the sample was contacted with 100 mL of dichlo-
romethane in a Pyrex accelerated one-step extractor
concentrator (Pyrex, Corning, NY). The concentrator tube
was kept in a water bath at 80 °C. Recondensation of the
recirculating dichloromethane was achieved with a con-
denser on top of the extractor body that was operated at
Regeneration Kinetics. A 0.13-g sample of dried (24 h
at 105 °C) Filtrasorb-400 activated carbon (Calgon Cor-
poration, Pittsburgh, PA) was contacted with 1 L of RDX
solution (36 mg of RDX/ L) in an Erlenmeyer flask until
equilibrium was reached; the equilibrium solid phase
concentration was qe )187 mg ofRDX/ g ofactivated carbon.
The bulk liquid was then discharged, and the carbon was
dried at ambient temperature. The dry loaded carbon was
then added to an Erlenmeyer flask containing 1 L of a
2
-4 °C. After 5.5 h, the extraction was complete, and the
bottom was concentrated to 1-5 mL. The sample volume
was then further reduced by evaporation to 500 µL with
-
9
9.999% helium. The concentrated sample was transferred
regeneration solution (20 mmol of OH / L; pH 12) that was
into an autosampler vial using a gas-tight syringe with
attached luer-tip needle. The vial was closed and sealed
with a Teflon-lined cap. The final sample concentration
was 1:2000 with this method.
kept constant at T ) 80 °C for at least 1 h before the
experiment. Timing commenced simultaneously. Samples
-
-
were taken and analyzed for RDX, HCOO , and NO2
content.
Am m onia. The ammonia concentration was measured
with a Orion-95-12 (Orion Research, Boston, MA) ammonia
electrode. The electrode was calibrated using the Orion-
Kinetic Models. In accordance with previous research-
ers (11), a second-order rate equation was fit to the
experimental data. The second-order rate law can be
reduced to a pseudo-first-order rate equation for constant
9
51006 standard solution (0.1 M NH4Cl).
-
Experim ental Section. Hom ogeneous Kinetics. Ex-
OH concentration provided through excess base. Pseudo-
periments were performed in a water bath at 50, 60, 70, 80
C, which was held at constant temperature for at least 1
first-order rate constants were obtained through a linear
least-squares fit of the sample data to the pseudo-first-
°
VOL. 30, NO. 5, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1 4 8 7

FIGURE 3. Pseudo-first-order reaction rate coefficient for RDX and
HMX versus hydroxide concentration at various temperatures.
FIGURE2. RDXandHMXhydrolysis atpH11forvarious temperatures.
negligible for the respective process conditions. We
obtained standard errors for the pseudo-first-order rate
constants varying between (0.4% (kHMX,1 for 50 °C and 23
order model, as follows:
c(t)
-
-
ln
2 OH
-
) -k t ) -k c t
(1)
mM OH / L) and ( 3.1% (kRDX,1 for 80 °C and 1.5 mM OH /
L) of the rate constant with the majority of the error limits
below 2% or 1% (Table 1). We found the subsequent
calculation of the second-order rate constants from the
pseudo-first-order rates to be precise (Table 2). The rates
of the alkaline hydrolysis of RDX are approximately 11.5
times greater than the rates for the alkaline hydrolysis of
HMXat similar temperatures. With increasing temperature
however, the difference in rates is reduced and the isokinetic
temperature is approximately 510 °C.
1
c(t ) 0)
where c is the concentration of the respective compound
and k1 is the pseudo-first-order rate constant. The second-
order rate constant can then be evaluated from the pseudo-
first-order rate constants using the equation k2 ) k1/cOH-
-
and performing several experiments at different OH
concentrations. The temperature dependencyofthe kinetic
constants was further evaluated using the Arrhenius equa-
tion:
An increase in temperature of10 °C produced an increase
in the rate of 3.6-2.9-fold for the alkaline hydrolysis of
HMX and 3.3-2.6-fold for RDX over the investigated
temperature range. The evaluation of the temperature
dependency of the calculated second-order rate constants
using eq 2 is shown in an Arrhenius plot (Figure 4). The
activation energy E over the range from 50 to 80 °C is
constant as shown bythe excellent correspondence between
experimental and model data. We obtained values for the
E
RT
ln k (T) ) ln A -
(2)
2
where A is the empirical Arrhenius factor or pre-exponential
factor; E is the empirical activation energy; R is the gas
constant; and T is the absolute temperature. The Arrhenius
factor A and the constant E/R can be determined from a
linear least-squares fit of the logarithm of the specific rate
constants against the reciprocalofthe absolute temperature
-
1
activation energies of ERDX ) 99.9 ( 1.9 kJ mol and EHMX
-1
(
17). All error determination limits were obtained using
) 111.9 ( 0.8 kJ mol , respectively (Table 3). The precision
-
1
standard error of the mean.
is much less than the (4 kJ mol that is acceptable in
common practice (18). We also found the Arrhenius
-
1
-1
Results and Discussion
parameter for RDX (ln ARDX ) 39.09 ( 0.67 L mol min
)
Figure 2 shows a subset of the kinetic experimental results
for HMX and RDX hydrolysis, and they are typical of all
results. At 50 °C and pH 12, alkaline hydrolysis of HMX is
very slow, and after more than 20 h less than 10% of the
initial HMX was hydrolyzed. For higher pHs and temper-
atures, hydrolysis is much more rapid, and at pH 12 and
0 °C, alkaline hydrolysis is more than 99.9% complete after
00 min. RDX hydrolysis is much more rapid. At pH 12
and 80 °C, the hydrolysis is essentially complete after only
a few minutes.
to be very close to the one of HMX (ln AHMX ) 40.93 ( 0.27
-1
-1
Lmol min ). This suggests a similar reaction mechanism
for the hydrolysis of RDX and HMX. Kinetic coefficients at
50 and 80 °C, extrapolated from previous research over the
range of 25-45 °C (11, 12), are 20-22% lower for RDX and
3-27% lower for HMX than measured in this investigation.
-
-
8
1
The kinetics of NO2 and HCOO formation are shown
in Figure 5. Equilibrium is reached in approximately 6 h
only for T ) 80 °C and pH 12 (Figure 5d). In all other cases,
-
the formation of HCOO was incomplete after 360 min
All experimental data could be fit accurately using the
pseudo-first-order model (eq 1) and the linear least-squares
fit method. Figure 3 shows the pseudo-first-order rate
constants calculated from all results as a function of
hydroxide concentration. Therefore, we consider the model
(Figure 5a-c). Therefore, all experiments were continued
until a final equilibrium was reached after several days or
weeks. The equilibrium concentrations are shown in Table
-
4. The molar yields are approximately 1.5 M HCOO and
-
1.5 M NO2 formed/ M RDX hydrolyzed. These results are
-
applicable and deviations from constant OH concentration
significantly higher than those reported in previous re-
1
4 8 8
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 5, 1996

TABLE 1
-
Pseudo-First-Order Rate Constants of Alkaline Hydrolysis of HMX and RDX for Different OH Concentrations
kHMX.1 (10- min-1
3
)
-
-
-
T (°C)
0.22 mmol of OH /L
2.1 mmol of OH /L
23 mmol of OH /L
5
6
7
8
0
0
0
0
0.089 ( 0.002
0.35 ( 0.009
1.19 ( 0.022
3.62 ( 0.055
0.99 ( 0.014
3.28 ( 0.027
11.87 ( 0.16
39.64 ( 0.25
10.82 ( 0.04
38.45 ( 0.72
126.5 ( 1.26
374.9 ( 4.73
kRDX.1 (10- min-1
3
)
-
-
-
-
T (°C)
1.5 mmol of OH /L
2.1 mmol of OH /L
10 mmol of OH /L
20 mmol of OH /L
5
6
7
8
0
0
0
0
9.25 ( 0.075
29.52 ( 0.23
72.97 ( 1.75
190.9 ( 5.89
12.98 ( 0.09
41.45 ( 0.38
122.8 ( 1.45
288.5 ( 5.83
58.2 ( 0.77
204.3 ( 1.33
607.9 ( 5.65
1484 ( 26.0
127.2 ( 1.21
412.8 ( 3.31
1162 ( 16.7
2977 ( 43.4
TABLE 2
TABLE 3
Calculated Second-Order Rate Constants of Alkaline
Hydrolysis of HMX and RDX
Temperature Dependency of Second-Order Rate
Constants of Alkaline Hydrolysis of HMX and RDX:
Arrhenius Parameter and Activation Energy
1
kHMX.2 (L mol-1 min-1
kRDX.2 (L mol-1 min-1
T (°C)
)
)
ln A (L mol- min-1
)
E (kJ mol-1
)
-
-
4
5
6
7
8
0
0
0
0
0.47 ( 4.2 × 10
1.68 ( 8.7 × 10
5.50 ( 0.0137
16.19 ( 0.221
6.35 ( 0.216
20.73 ( 0.053
58.71 ( 1.332
150.62 ( 0.482
3
HMX
RDX
40.93 ( 0.27
39.09 ( 0.67
111.9 ( 0.76
99.92 ( 1.87
mencement of the experiment. A total of 5 g of RDX (0.0225
M) in 50 mL of 1.5 M NaOH at T ) 90 °C produced a system
pressure of 240.1 kPa at equilibrium. The gas analysis
showed 75% N2O and 23% N2. The remaining 2% included
small amounts of argon, oxygen, ammonia (1.5%), water,
and methanol (0.1%). Using the ideal gas law, the total
moles are calculated to 0.033 M. If the molar yield results
from the gas analysis are rounded to 0.025 M N2O and
0
0
.0075 M N2, the normalized molar yield is 1.1 M N2O and
.34 M N2/ M RDX hydrolyzed, respectively (Table 5). Using
the previously discussed results, a mass balance for carbon
and nitrogen can be calculated (Table 6). The recovery
was close to 100%: 94% of the carbon and 90% of the
nitrogen from RDX were found in the described products.
Nevertheless, the hydrolysate was further analyzed using
liquid-liquid extraction (LLE) and GC/ MS analysis. Any
previously unknown organic intermediates, if present and
extractable by dichloromethane and chromatographicable,
will be detected with this method. The samples were
concentrated from 1000 mLto 500 µL. In order to determine
the origin of compounds in the analytical protocol, a
comprehensive set of blank samples was extracted with
LLE and analyzed with GC/ MS. LLE system blanks (100
mL of dichloromethane with no sample recirculated for 5.5
h) as well as GC/ MS system blanks (GC/ MS injection of
pure Optima-grade dichloromethane, Fisher Scientific) were
run regularly to determine any system changes or con-
tamination. Deionized water and HPLC-grade water blanks
were extracted (LLE) and analyzed on the GC/ MS to
determine any contamination originating from the water.
Furthermore, possible contamination from the base was
determined with a set of base blanks (2 mL of 10 M NaOH
in 1000 mL of deionized water or HPLC-grade water,
respectively; NaOH and HPLC-grade water: Fisher Scien-
tific). Subsequently, blanks of unhydrolyzed RDXand HMX
in deionized and HPLC-grade water, respectively, were
prepared and analyzed to determine impurities in RDX and
HMX. Eventually, RDX and HMX hydrolysis samples were
prepared with deionized and HPLC-grade water (40 mg of
FIGURE 4. Arrhenius plot of second-order rate constants of the
alkaline hydrolysis of HMX and RDX.
search. Hoffsommer et al. (11) reported molar yields of 1.1
-
-
M NO2 / M RDX and 0.7 M HCOO / M RDX. However, they
conducted their study at lower temperatures, and the results
indicate that the normalized molar yields may have been
calculated before complete equilibrium was reached. We
could not find a determinable variation trend with pH or
temperature in the molar yield. It is also evident for all
-
four conditions (Figure 5) that NO2 is a direct product
from RDX formed during the first reaction step in timely
accordance with RDXdegradation and thus follows the same
rate law.
For all other compounds, only equilibrium concentra-
tions were determined (Tables 4 and 5). The normalized
molar yield of NH3 is 0.9 M/ M RDX hydrolyzed for pH 12
at 50 and 80 °C as well as pH 11 and 80 °C. We also
-
discovered acetate ion (CH3HCOO ) as an additional and
previously unknown reaction product (9-12). The yield of
-
0
.1 M CH3HCOO / M RDX, however, is comparatively small.
The analysis for gaseous products was performed in a 465-
mL gas-tight reactor that was evacuated before the com-
VOL. 30, NO. 5, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1 4 8 9

FIGURE 5. Kinetics of the alkaline hydrolysis of RDX: nitrite and formate formation (normalized concentration C ) M product formed/M
RDX hydrolyzed, t ) 0 as a function of time).
TABLE 4
TABLE 6
Normalized Molar Yields for Products of Alkaline
Carbon and Nitrogen Balance for Alkaline
Hydrolysis of RDX (C H O N )
3 6 6 6
a
Hydrolysis of RDX (Results from Liquid Analysis)
pH 11
pH 12
HCOO- CH3COO- HCHOa
sum RDX
(M) (M) (M)
δ
recovery
(%)
(M)
(M)
(M)
50 °C
80 °C
50 °C
80 °C
carbon
1.5
0.2
1.1
2.8
3
-0.2
94
NO2-
HCOO
CH3COO
NH3
1.5
1.3
1.5
1.5
1.3
1.5
1.6
1.5
0.1
0.9
-
NO2- NH3 N O N2 sum RDX
(M) (M) (M) (M) (M) (M)
2
δ
(M)
recovery
(%)
-
0.9
0.9
nitrogen
1.6
0.9 2.2 0.7 5.4
6
-0.6
90
a
M product form ed/M RDX hydrolyzed.
a
Data from ref 11.
TABLE 5
Normalized Molar Yields For Products of Alkaline
a
Hydrolysis of RDX (Results from Gas Analysis)
normalized molar yield
N2O
N2
1.1
0.34
a
M/M RDX hydrolyzed.
RDX or 4 mg of HMX, respectively, was stirred for 6 h in
000 mL of water with 2 mL of 10 M NaOH at 80 °C). No
1
further intermediates or end products could be discovered
by GC/ MS analysis from the alkaline hydrolysis of RDX and
HMX.
In order to determine if alkaline hydrolysis is a feasible
regeneration process for high “explosives-laden” activated
carbon, we compared the batch homogeneous kinetics at
FIGURE 6. Comparison of the kinetics of nitrite and formate formation
during aqueous homogeneous alkaline hydrolysis of RDX and
regeneration of RDX-laden activated carbon.
formation from the ring cleavage and the fourth-order
-
8
0 °C and pH 12 with batch desorption kinetics during
Canizarro reaction of formaldehyde. The ratios of NO2
-
alkaline hydrolysis regeneration of activated carbon under
the same conditions. However, because the adsorbed RDX
is rapidly destroyed during desorption at elevated pH and
temperature, its concentration cannot be used to determine
and HCOO produced per mole of RDX hydrolyzed can
then be compared to solid-phase hydrolysis and desorption
as well as to aqueous homogeneous conditions.
The results are shown in Figure 6. In the homogeneous
-
-
-
desorption kinetics. Therefore, NO2 and HCOO were
chosen as appropriate parameters to indicate desorption
experiment, 20 mM OH (pH 12) was added to an aqueous
RDX solution (35 mg/ L) that was stirred and heated (T )
80 °C). The regeneration experiment was carried out with
0.131 g of loaded activated carbon (qe ) 187 mg of RDX/ g
of activated carbon Filtrasorb-400), which was stirred in a
-
and hydrolysis rates during regeneration. NO2 is the first
product formed during the alkaline hydrolysis of RDX, while
HCOO- is more slowly produced because of its dual
1
4 9 0
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 5, 1996

-
-
regeneration solution at 80 °C (20 mM OH or pH 12). NO2
is evidently produced much faster under homogeneous
conditions than in desorption, which suggests limitation
-
through diffusion. However, the HCOO concentration
during regeneration seems to be diffusion limited only in
the beginning of the experiment. At the end of the
experiment, the curves of the regeneration and the ho-
mogeneous experiments are almost the same, which
indicates chemical limitation at this stage of the process.
RDX could not be detected in the liquid phase at any time
during the regeneration experiment.
Most importantly, however, the final yields are virtually
the same for homogeneous conditions and regeneration
-
conditions. The homogeneous reaction yields 1.6 M NO2
-
and 1.5 M HCOO / mol of RDX. The regeneration experi-
ments yielded 1.6 M NO2 and 1.6 M HCOO / mol of RDX.
The slight difference for HCOO is likelydue to experimental
-
-
-
error. The regenerated carbon was furthermore contacted
with 10 mL of acetonitrile in a 25-mL vial on a shaker table.
No residual adsorbed RDX could be found in the bulk liquid
after several days and weeks, respectively. This indicates
that the alkaline hydrolysis ofadsorbed RDXand subsequent
desorption of hydrolysis products is complete.
We found excellent pseudo-first-order and second-order
rate dependencies for the alkaline hydrolysis of HMX and
RDX (Tables 1 and 2). For elevated temperatures ranging
from 50 to 80 °C the reaction is second-order, and
preliminary studies on the rate and reaction order of the
alkaline hydrolysis of HMX and RDX at temperatures
ranging from 0 to 45 °C (9-11) are verified for the higher
temperature range. The temperature dependency of the
rate constants is accurately predicted using the Arrhenius
model (Figure 4; Table 3). The almost similar activation
parameters for the alkaline hydrolysis of RDX and HMX
also suggest that the reaction follows the same basic reaction
mechanism.
FIGURE 7. Proposed reaction mechanism for the alkaline hydrolysis
of RDX.
complex intermediates are persistent but are completely
reacted to the low molecular weight compounds as shown
in Figure 7. Compared to previous research, the mass
balance for carbon and nitrogen is improved. Hoffsommer
et al. (11) recovered 77% nitrogen and 60% carbon. We
recovered 90% of the nitrogen and 94% of the carbon,
respectively.
Our objective in this study, which is part of our
investigations ofa novelphysicochemicaltreatment scheme
for explosives-contaminated waters (Figure 1), was to
determine basic data for the chemical stage of the scheme
(
activated carbon regeneration by alkaline hydrolysis). The
alkaline hydrolysis of RDX and HMX was rapid under
homogeneous conditions. The application of the process
for activated carbon regeneration led to 100% desorption
of hydrolysis products, and no residual RDX could be
extracted from regenerated activated carbon. Therefore,
we consider the alkaline hydrolysis of HMX and RDX in
presence ofintermediate alkaline concentrations (2-25 mM
The alkaline hydrolysis of RDX in the presence of
hydroxide ions is approximately 10 times faster than the
alkaline hydrolysis of HMX. Since HMX is a frequent
contaminant in RDX, the compounds almost always occur
together in explosives-contaminated wastewater and at
contaminated sites. Therefore, the rate of the alkaline
hydrolysis of HMX may be rate limiting in the design of a
treatment system using alkaline hydrolysis; however, our
results show that an increase in temperature has a higher
influence on the rate of the alkaline hydrolysis of HMX
than RDX. Higher temperatures (80 °C) may be preferable
to further increased hydroxide concentrations (>25 mmol/
L) for treatment of HMX using alkaline hydrolysis. The
isokinetic temperature for the alkaline hydrolysis of RDX
and HMX is approximately 510 °C.
-
OH / L) and elevated temperatures (above 50 °C) to be a
feasible chemical regeneration process for high explosives-
laden activated carbon as part of an off-line regeneration
treatment scheme (Figure 1). Concentrating the high
explosives on the activated carbon can make alkaline
hydrolysis an economically feasible treatment method. The
organic end products of the hydrolysis and byproducts
should be treatable in conventional biological treatment
processes. The high salinity that results from both the
sodium hydroxide used for hydrolysis, and the acid needed
for neutralization may require special handling and condi-
tions
-
The formation of HCOO is obviously slow in compari-
Acknowledgments
This research was supported in part by the following
organizations: Germ an Governm ent/ State of Berlin
son with RDX degradation and NO2- formation. This is
due to the production offormate from both the ring cleavage
and subsequently from the fourth-order Canizarro reaction
of formaldehyde (HCHO) at the elevated pH (11), the latter
one being slow (13, 19, 20). The carbon and nitrogen
balance over the reaction time shows that RDX-h5 does
not accumulate and is followed by the formation ofcomplex
intermediates from ring cleavage. The further mineraliza-
tion of these intermediates is slow and rate-limiting. The
reaction mechanism should be rewritten as shown in Figure
(
“NaF o¨ G” postgraduate scholarship); German Academic
Exchange Service (DAADsforeign country upgrade schol-
arship); Lawrence Livermore National Laboratory; and the
Pantex Plant. The authors are grateful for the assistance
and suggestions of J. I. Daniels, W. Desmare, P. Goodfellow,
J. P. Knezovich, C. Lee, and E. Ruth.
Literature Cited
7
. However, we could not prove the existence of any
(
1) Byrd, J., Jr.; Humphreys, J. Proceedings of the 12th International
Incineration Conference; Knoxville, TN, May 3-7, 1993; pp 621-
624.
relatively nonpolar organic compounds by GC/ MS analysis
at equilibrium. We therefore conclude that none of these
VOL. 30, NO. 5, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1 4 9 1

(
(
2) LASL Explosive Property Data; Gibbs, Terry R., Popolato, Al-
phonse, Eds.; University of California Press: Berkeley, CA, 1980.
3) Dobratz, B. M. LLNL Explosives Handbook; Technical Report
UCRL-52997; Lawrence Livermore National Laboratory: Liv-
ermore, CA, 1981.
(13) Sykes, Peter. A Guidebook to Mechanism in Organic Chemistry,
5th ed.; Longman: London, U.K., and New York, 1981.
(14) Rosenblatt, David H.; Burrows, Elizabeth P.; Mitchell, Wayne R.;
Parmer, David L. In The Handbook of Environmental Chemistry;
Springer: Berlin, Germany, 1991; Vol. 3, part G; pp 195-235.
(
(
4) McLellan, W.; Hartley, W. R.; Brower, M. Health advisory for
hexahydro-1,3,5-tetranitro-1,3,5-triazine;TechnicalReport PB90-
(
15) Spontarelli, T.; Buntain, G. A.; Sanchez, J. A.; Benziger, T. M.
Proceedings of the 12th International Incineration Conference;
Knoxville, TN, May 3-7, 1993; pp 787-791.
2
73533; Office of Drinking Water, U.S. Environmental Protection
Agency: Washington, DC, 1988.
(
16) Hesselmann, R. P. X.; Alatriste-Mondragon, F.; Babcock, R. W.;
Knezovich, J. P.; Daniels, J. I.; Mah, R. A.; Stenstrom, M. K.
Presented at the 1992 Waste Management and Environmental
Sciences Conference, San Juan, Puerto Rico, April 9-11.
5) McLellan, W.; Hartley, W. R.; Brower, M. Health advisory for
octahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazine; Technical Report
PB90-273525; Office of Drinking Water, U.S. Environmental
Protection Agency: Washington, DC, 1988.
(
(
17) Benson, Sidney W. The Foundations of Chemical Kinetics;
Krieger: Malabar, FL, 1982; pp 66-68.
18) Baerns, M.; Hofmann, H.; Renken, A. Chemische Reaktionstech-
nik, 2nd ed.; Thieme: Stuttgart, Germany, 1992; pp 205-207.
(
(
(
6) Patterson, W. E.;Phelan, P. F. Proceedingsofthe12th International
Incineration Conference; Knoxville, TN, May 3-7, 1993; pp 625-
6
29.
7) Holl, G.; Schneider, M. In Contaminated Soil ‘93; Arendt, F.,
Annokkee, G. J., Bosman, R., van den Brink, W. J., Eds.; Kluwer:
Dordrecht, The Netherlands, 1993; pp 941-942.
8) Patterson, J.; Shapira, N. I.; Brown, J.; Duckert, W.; Polson, J.
State of the Art Military Explosives and Propellants Industry
Volume III. Wastewater Treatment; Technical Report PB-265042;
U.S. Environmental Protection Agency: Washington, DC, 1976;
p 92.
(19) Birstein, G.; Lobanow, N. Z. Anorg. Allg. Chem. 1927, 160, 377-
386.
(20) Euler, H. V.; L o¨ vgren, Thor. Z. Anorg. Allg. Chem. 1925, 147,
123-134.
Received for review June 13, 1995. Revised manuscript re-
X
ceived December 6, 1995. Accepted December 17, 1995.
(
9) Epstein, S.; Winkler, C. A. Can. J. Chem. 1951, 29, 731-733.
(
(
10) Jones, W. H. J. Am. Chem. Soc. 1953, 76, 829-835.
11) Hoffsommer, J. C.; Kubose, D. A.; Glover, D. J. J. Phys. Chem.
ES9504101
1
977, 81, 380-385.
X
(
12) Croce, M.; Okamoto, Y. J. Org. Chem. 1979, 44, 2100-2103.
Abstract published in Advance ACS Abstracts, March 1, 1996.
1
4 9 2
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 5, 1996