The solubility and recrystallization of 1,3,5-triamino-2,4,6- trinitrobenzene in a 3-ethyl-1-methylimidazolium acetate-DMSO co-solvent system

Article abstract of DOI:10.1039/b810109d

Ionic liquids have previously been shown to dissolve strong inter- and intramolecular hydrogen-bonded solids, including natural fibers. Much of this solubility is attributed to the anions in ionic liquids, which can disrupt hydrogen bonding. We have studi

Full text of DOI:10.1039/b810109d

View Article Online / Journal Homepage / Table of Contents for this issue
New Journal of Chemistry
An international journal of the chemical sciences
Volume 33 | Number 1 | January 2009 | Pages 1–212
www.rsc.org/njc
ISSN 1144-0546
PAPER
T. Yong-Jin Han et al.
The solubility and recrystallization of
1,3,5-triamino-2,4,6-trinitrobenzene
in a 3-ethyl-1-methylimidazolium
1144-0546(2009)33:1;1-2
acetate–DMSO co-solvent system

PAPER
www.rsc.org/njc | New Journal of Chemistry
The solubility and recrystallization of 1,3,5-triamino-2,4,6-trinitrobenzene
in a 3-ethyl-1-methylimidazolium acetate–DMSO co-solvent system
T. Yong-Jin Han,* Philip F. Pagoria, Alexander E. Gash, Amitesh Maiti,
Christine A. Orme, Alexander R. Mitchell and Laurence E. Fried
Received (in Gainesville, FL, USA) 17th June 2008, Accepted 6th August 2008
First published as an Advance Article on the web 18th September 2008
DOI: 10.1039/b810109d
Ionic liquids have previously been shown to dissolve strong inter- and intramolecular hydrogen-
bonded solids, including natural fibers. Much of this solubility is attributed to the anions in ionic
liquids, which can disrupt hydrogen bonding. We have studied the solubility and recrystallization
of 1,3,5-triamino-2,4,6-trinitrobenzene (TATB), a very strong inter- and intramolecular hydrogen-
bonded solid, in various ionic liquid solvent systems. We discovered that acetate-based ionÀic
liquids were the best solvents for dissolving TATB, while other anions, such as ClÀ, HSO4 and
À
NO3 showed moderate improvements in the solubility compared to conventional organic
solvents. Ionic liquid–DMSO co-solvent systems were also investigated for dissolving and
recrystallizing TATB.
been a topic of intense discussion,8,9 since the original crystal
structure determined by Cady and Larson showned4 TATB to
1. Introduction
Ionic liquids (ILs) have recently been shown to be ideal
solvents for dissolving hydrogen-bonded solids, including
cellulose1 and natural fibers.2,3 The use of imidazolium-based
cations with halides as counter-anions has significantly im-
proved the solubilities of these natural products. Successful
dissolution of these highly hydrogen-bonded solids is largely
attributed to the ability of ILs’ anions to act as hydrogen bond
acceptors and disrupt the hydrogen bonds in these materials.1
With a graphite-like crystalline structure,4 1,3,5-triamino-
2,4,6-trinitrobenzene (TATB) is one of the most strongly
hydrogen-bonded solids known. Owing to its inter- and intra-
molecular hydrogen bonds, both in-plane and out-of-plane
(see Fig. 1), the solubility of TATB in conventional organic
solvents is minuscule. With a capacity to dissolve 70 ppm
(0.007% w/v) at room temperature,5 DMSO is the best con-
ventional organic solvent known to dissolve TATB. Super-
acids, such as concentrated sulfuric acid, have been shown to
dissolve up to ca. 240 000 ppm (24% w/v) at room tempera-
ture,5 but due to their highly corrosive nature are often
avoided. There is a need to find a desirable solvent for TATB
as it has become necessary to control the particle size, as well
as the morphology, of TATB crystals.
ꢀ
have a centrosymmetric structure with the space group P1,
Z = 2, which is incompatible with NLO activity. Some have
attributed the NLO activity of TATB to a small amount of a
second, presently unidentified, TATB polymorph mixed in
with the bulk centrosymmetric TATB crystals.10,11 Therefore,
identifying a suitable solvent system that can increase the
TATB is of particular interest in the energetic materials
(EM) community due to its extreme insensitivity to impact,
shock and heat, while providing a good detonation velocity.6
This combination of insensitivity with good performance
characteristics makes TATB an ideal insensitive high explosive
(IHE) in numerous applications. TATB has also attracted
researchers from the field of optics, due to its unexpectedly
strong secondary harmonic generation (SHG) efficiency.7 The
source of the nonlinear optical (NLO) property of TATB has
Chemistry, Materials, Earth and Life Sciences Directorate, Lawrence
Livermore National Laboratory, Livermore, CA 94551, USA.
E-mail: han5@llnl.gov
Fig. 1 The crystal structure of TATB, a centrosymmetric structure
ꢀ
with space group P1, Z = 2. (a) A–B plane view, (b) C plane view.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
50 | New J. Chem., 2009, 33, 50–56

solubility of TATB may also allow a new opportunity to
crystallize and isolate the potential second polymorph with
exceptional SHG efficiency.
cooled back down to room temperature without stirring.
Occasionally, an Omega (series 2010) programmable controller
was used to control the cooling rate.
Herein, we report the solubility of highly hydrogen-bonded
TATB in both commercially available and custom synthesized
imidazolium-based ILs. In particular, 3-ethyl-1-methylimida-
zolium acetate (EMImOAc) was extensively investigated in
its pure form, as well as in combination with DMSO as a
co-solvent.
For a typical anti-solvent crystallization, 20 mL of an 80 : 20
DMSO–EMImOAc solution was placed in a four-necked
100 mL round-bottomed flask equipped with an overhead
stirrer, drying tube, thermocouple and septum inlet. To this
was added TATB (0.5 g), and the mixture was stirred and
heated slightly (50 1C) until all of the TATB had dissolved and
a red-orange solution had formed. The temperature of the
sample was maintained at the desired temperature using a
J-KEM temperature controller. The mixture was stirred slowly
as a solution of acetic acid (4 g) in dry DMSO (40 mL) was
added via a syringe and long needle connected to a syringe
pump, set to deliver at 2 mL hÀ1. The resulting TATB was
collected by suction filtration, and washed with water (25 mL)
and MeOH (10 mL) to yield 0.46 g of a yellow microcrystalline
solid. Raman spectroscopy was used to determine the purity of
the recrystallized TATB.
2. Experimental
2.1 Materials
3-Ethyl-1-methylimidazolium chloride (EMImCl), 3-butyl-1-
methylimidazolium chloride (BMImCl), 3-ethyl-1-methyl-
imidazolium acetate (EMImOAc), 3-ethyl-1-methylimidazolium
nitrate (EMImNO3), 3-butyl-1-methylimidazolium hydrogen
sulfate (BMImHSO4) and DMSO were purchased from
Sigma-Aldrich, and used without purification unless otherwise
noted. Vacuum distillation was performed on EMImOAc to
remove any impurities prior to experiments.
3. Results
3-Allyl-1-methylimidazolium chloride (AllylMImCl) was
synthesized according to a published report.12 3-(Methoxy-
methyl)-1-methylimidazolium chloride (MeOMImCl) was
synthesized using a modification of the procedure reported
for the synthesis of AllylMImCl.
3.1 Solubility of TATB in ILs
There are certain advantages that ILs have over conventional
solvents that make them an attractive alternative for the
dissolution of TATB. ILs, because of their inherent low vapor
pressure and high-temperature stability, have reduced environ-
mental and safety concerns compared to conventional
organic solvents. Also, in theory, the IL is recoverable after
precipitation or distillation of impurities from it.
Synthesis of 3-(methoxymethyl)-1-methylimidazolium chloride
(MeOMImCl). Into a 500 mL round-bottomed flask equipped
with a stirrer bar, argon inlet and addition funnel, was
dissolved 1-methylimidazole (25 g, 0.31 mol) in trichloro-
ethylene (100 mL). With stirring, chloromethyl methyl ether
(35 g, 0.43 mol) was added dropwise over a 0.5 h period. The
mixture was warmed and a turbid, two-layer mixture formed.
The mixture was refluxed for 2 h, cooled and poured into a
separating funnel. The organic layer was separated, filtered and
the solvent removed under vacuum at 45 1C to yield a
tan-beige viscous liquid (52 g).
The solubility of TATB was first investigated in BMImCl,
since BMImCl has previously been shown to dissolve
cellulose,1 Bombyx mori silk fibers2 and wool keratin fibers3
in relatively high concentrations (10, 13.2 and 4 wt%, respec-
tively at 100 1C). A solution of 0.5 wt% of TATB in BMImCl
was stirred rapidly at 100 1C for 20 h. However, at the end of
the 20th hour, there were still TATB particles present in the
flask. The color of the solution was only slightly yellow
(the color of the original TATB powder), signifying that only
a small amount of TATB had dissolved. Similar results were
observed for other ILs with ClÀ anions, including EMImCl
and custom synthesized AllylMImCl (see Table 1). As noted
previously, short chain-substituted imidazolium-based ILs
with ClÀ anions have been effective in dissolving natural
polymers. The hydrogen bond-accepting ClÀ anion is thought
to be the crucial component in disrupting hydrogen bonding in
the biopolymer.1 However, for TATB the hydrogen bond
disruption caused by the ClÀ ions was not strong enough to
significantly dissolve TATB particles.
2.2 Solubility measurements
Small scale solubility tests (o10 mg) of TATB in ILs
were monitored with a Nikon optical microscope equipped
with a temperature controlled heating stage, under cross-
polarized light.
Large scale solubility measurements were performed using a
three-necked round-bottomed flask in a silicone oil bath
at a constant temperature of 100 1C. Due to its high density
(1.93 g cmÀ3) and bright yellow color, visual inspection of
TATB particles in solutions was easily achieved with the aid of
a hand-held flashlight.
In an attempt to improve the solubility of TATB, a new
imidazolium-based cation, 3-methoxymethyl-1-methylimida-
zolium chloride (MeOMImCl), was synthesized. Unlike the
BMIm and EMIm cations, which may have limited, if any,
hydrogen bond-accepting capability, the ether side chain of
MeOMImCl is a hydrogen bond acceptor13 and may assist in
disrupting the strong hydrogen bonding of TATB. However,
when 0.5 wt% of TATB was added to MeOMImCl and heated
with stirring at 100 1C for 20 h, no significant quantity of
2.3 Crystallization
A non-agitated cooling crystallization method was employed
to grow TATB crystals from a DMSO–EMImOAc co-solvent
system. Typically, in a 250 mL round-bottomed flask equipped
with a drying tube and a thermocouple, TATB (4 g) was
added, along with 100 g of DMSO–EMImOAc (80 : 20 w/w).
The solution was slowly heated to 90 1C with constant stirring.
Once all of the TATB had dissolved, the solution was slowly
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
New J. Chem., 2009, 33, 50–56 | 51

Table 1 The solubility of TATB in various IL systems at 100 1C
IL
R
XÀ
TATB solubility (wt%)
BMImCl
EMImCl
CH3CH2CH2CH2
CH3CH2
CH2QCH
CH3OCH2
CH3CH2CH2CH2
CH3CH2
CH3CH2
ClÀ
o0.5
o0.5
o0.5
o0.5
o0.5
o0.5
10
ClÀ
AllylMImCl
MeOMImCl
BMImHSO4
EMImNO3
EMImOAc
ClÀ
ClÀ
HSO4
À
À
NO3
CH3COOÀ
TATB dissolved, similar to BMImCl and EMImCl, which was
visually evident. This result suggests that even with an active
cation, the overall hydrogen bond-accepting capacity of
MeOMImCl is not significant enough to disturb the hydrogen
bonding network of TATB crystals.
The effects of anions other than ClÀ in dissolving TATB
were also examined. For this study, BMImHSO4, EMImNO3,
and EMImOAc were selected and tested. These ILs were
chosen for their anions’ hydrogen bond-accepting capability.
As anticipated, BMImHSO4 and EMImNO3 had very similar
results compared to BMImCl, each showing less than 0.5 wt%
solubility of TATB at 100 1C. However, EMImOAc showed a
surprisingly good solubility of TATB. At 100 1C in a large
scale test, EMImOAc was able to dissolve up to 10 Æ 1.0 wt%
of TATB, confirmed by visual inspection. There was, however,
a noticeable difference in the color of the acetate solution when
compared to the solutions from the other anions (Fig. 2A
inset). When TATB was added to EMImOAc and heated, the
entire mixture turned a dark blood-red color, whereas in the
other ILs, no color change was observed (at times, some
turned slightly yellow, the original color of the TATB crys-
tals). The source of this color change was investigated by
UV-vis spectrophotometry. As seen in Fig. 2A, TATB added
to EMImOAc (diluted 100 times with DMSO) clearly shows a
pronounced peak at lmax = 409 nm, whereas TATB dissolved
in DMSO only shows a TATB absorbance at lmax = 357 nm.
The origin of the peak at lmax = 409 nm can be assigned to a
s-complex. We also carried out a Raman spectroscopy measure-
ment of the s-complex (Fig. 2B). The peaks observed at
807 and 1149 cmÀ1 correspond well to a previously reported
signature of a s-complex.14 The formation of a s-complex
with TATB was described during the synthesis of TATB by the
vicarious nucleophilic substitution of hydrogen.15 Selig also
assigned the absorption band at 409 nm to a s-complex
between strong bases and TATB.5
Fig. 2 (A) UV-vis spectra of TATB dissolved in DMSO and the
DMSO–EMImOAc system. The inset shows a photograph of TATB
dissolved in DMSO (left) and in EMImOAc (right). (B) Raman
spectra of (a) TATB, (b) EMImOAc in DMSO, (c) TATB dissolved
in EMImOAc–DMSO solution; the peaks at 807 and 1149 cmÀ1 are
signatures of a s-complex.
3.2 Crystallization of TATB
The primary need for a solvent system that will readily dissolve
TATB is to improve overall processability. More specifically,
high solubility is necessary to produce high quality crystals
from a supersaturated solution of TATB. Therefore, crystal-
lization experiments were performed using EMImOAc via a
non-agitated cooling method. An optical microscope fitted
with a heating stage was used to study the recrystallization of
TATB in EMImOAc. A few drops of TATB particles in
EMImOAc (4 wt% solution) were placed on a cover slip on
a heating stage. The mixture was heated to 100 1C and kept at
this temperature until all of the particles had dissolved (Fig. 3a
and b). The homogeneous solution was cooled slowly by
natural convection. From the saturated solution, crystals
started to emerge when the temperature reached 70 1C. Over
time, single, well-faceted crystals of TATB appeared and grew
larger (Fig. 3c and d). These crystals were far better than the
starting TATB crystals, which were almost all aggregates with
few, if any, well defined facets.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
52 | New J. Chem., 2009, 33, 50–56

Fig. 5 SEM micrographs of TATB (a) recrystallized from EMImOAc,
(b) starting material and (c) H2O-precipitated material.
Fig. 3 Optical images of TATB dissolving and recrystallizing in
EMImOAc: (a) at room temperature, (b) at 100 1C, (c) at 70 1C and
(d) cross-polarized light view of image (c).
micrographs of the recrystallized TATB crystals showed good
crystal morphology (Fig. 5a) compared to the starting crystals
(Fig. 5b). The crystal sizes of the recrystallized TATB ranged
from 10–50 mm. On the other hand, the water crash-precipi-
tated crystals showed an irregular crystal morphology, with
crystal sizes that ranged from sub-500 nm–5 mm (Fig. 5c). The
Raman spectrum of the recrystallized TATB crystals con-
firmed that the structure of the recrystallized TATB matched
well with that of the starting material (Fig. 6).
Similar recrystallization experiments were performed on a
larger scale (100 mL volume). However, due to the viscosity of
EMImOAc at room temperature, it was difficult to filter and
isolate the recrystallized TATB. In order to lower the viscosity
of EMImOAc, DMSO was added as a co-solvent. When
DMSO was added, the solubility of TATB decreased linearly
with respect to the DMSO concentration (Fig. 4).
Besides the non-agitated cooling method of TATB crystal-
lization, we also investigated an anti-solvent crystallization
method. As seen from the non-agitated crystallization method
above, a s-complex, which forms when TATB is dissolved in
EMImOAc, requires a proton source to fully convert it back to
TATB at room temperature. Thus, we employed acetic acid as
an anti-solvent to provide the necessary proton for the reac-
tion. Acetic acid, a weak acid, was chosen to limit the rate of
reaction to avoid the crash precipitation of TATB. Preliminary
experiments showed that the concentration and rate of addi-
tion of acetic acid is critical in controlling the overall size of
the recrystallized TATB. The overall morphology of the
TATB crystals formed by this method showed a significant
improvement compared to the starting materials. By carefully
controlling the rate of addition, significantly large TATB
crystals (o500 mm) could be formed via this method.
However, even with the DMSO concentration as high as
80 wt%, the solubility of TATB was still significantly high,
approximately 4 wt%. Recrystallization experiments were
performed using an 80 : 20 DMSO–EMImOAc solution
(80 wt% DMSO, 20 wt% EMImOAc). Upon heating this
solution with TATB to 90 1C, a color change was once again
observed, signifying the formation of a s-complex. Once the
added amount of TATB had fully dissolved, the solution was
allowed to cool to room temperature by natural convection,
allowing crystals to form. The recrystallized TATB was re-
covered by simple vacuum filtration. It was visually apparent
that the filtrate contained a significant amount of the s-complex.
Therefore, the addition of excess water or another proton
donor (i.e. acetic acid) was necessary in order to fully recover
the remaining TATB dissolved (ca. 1 wt%) in the filtrate. SEM
Fig. 4 The solubility curve of TATB in the DMSO–EMImAOc
Fig. 6 Raman spectra of (a) starting TATB and (b) recrystallized
system.
TATB.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
New J. Chem., 2009, 33, 50–56 | 53

the nitro groups, thus stabilizing the s-complex. Most studies
on the formation of s-complexes investigated 1-substituted-
2,4-dinitro- or 1-substituted-2,4,6-trinitrobenzenes, and very
few s-complexes of fully-substituted aromatic rings are
known. In the TATB molecule, because of its high-symmetry,
there are three equivalent positions at which the acetate anion
may react to produce the s-complex. The s-complex would be
stabilized by the presence of the three nitro groups at the 2-, 4-
and 6-positions relative to the tetrahedral carbon containing
the acetate group. In addition, the stability of the s-complex
relative to TATB would be enhanced by relieving steric
crowding on the TATB ring of the adjacent amino and nitro
groups upon forming the tetrahedral carbon center at the site
of base addition.
4. Discussion
In order to determine the ineffectiveness of ILs in solubilizing
TATB compared to cellulose, the cohesive energy density
(CED) of the two systems was compared. CED is often
expressed in its square-root form, known as the solubility
parameter, d. For cellulose, a variety of measurements, in-
cluding mechanical and surface free energy measurements,
1
2
16
suggest a value of d B 25 MPa . For TATB, the heat
of sublimation (i.e. cohesive energy per molecule) is
B40.2 kcal molÀ1 17
.
This, coupled with a molar volume of
1
2
3
4
221.2 A in the crystal phase, yields a value of d B 35.5 MPa .
In other words, the CED (= d2) of TATB is approximately
2 times that of cellulose, explaining why the former is more
difficult to dissolve than the latter. The above-obtained solu-
bility parameter can be used as an approximate tool to screen
solvents for TATB. As an example, we note that the conven-
TATB added to EMImCl does not form a s-complex. In
order to understand the difference between the action of the
acetate and chloride anions on TATB, we carried out DFT-
based investigations using a state-of-the-art quantum chemical
conductor-like screening model (COSMO)25 and its extension
to real solvents (COSMO-RS).26 This model computes the
chemical potential of a solute in its own environment and in
solvent environments. From the difference between these
chemical potentials, one can estimate the solubility of the
solute in the solvent. Details of the procedure are described
elsewhere.27 Table 2 (column 2) displays the computed solu-
bility of TATB in EMImCl and EMImOAc, respectively. The
computed solubility in EMImCl is in line with the observed
low solubility (i.e. o0.5 wt%) in this solvent. In contrast, the
computed solubility in EMImOAc is 250 times smaller than
the experimentally observed value of 10 wt% at 100 1C. This
result, in conjunction with color changes observed in the
acetate IL solution, indicates that while TATB dissolves in
its pure form in EMImCl, it undergoes some chemical reaction
during its dissolution in EMImOAc. We have also computed
the possibility of a deprotonation mechanism for the observed
solubility in EMImOAc. To do this, we have investigated one
of the simplest reactions, i.e. an NH2 group of TATB loses a
proton to a neighboring anion of the IL (thereby forming an
acetic acid molecule in the acetate IL or a HCl molecule in the
chloride IL), while the unpaired cation of the IL binds to the
ortho position of the deprotonated TATB anion. Column 3 of
Table 2 lists the computed heat of reaction for such a chemical
process using the program Dmol3.20–22 The reaction is highly
endothermic and clearly prohibitive in the chloride IL, but
exothermic and likely to occur in the acetate IL. The above
results can be qualitatively explained based on the stronger
basicity of an OAcÀ group compared to ClÀ. The computed
UV-vis spectrum (using the semi-empirical program
tional water-soluble IL, BMIm+BF4À, has a solubility para-
meter near to d = 26.5 MPa .18 Given that TATB has a much
1
2
higher value of d, we expect the solubility of TATB to be poor
in such prototypical solvents in the absence of chemical
modification. In order to understand the reason for the higher
CED in TATB, we employed first-principles density functional
theory (DFT) using the program DMol3 19–22 to compute the
interlayer van der Waals binding and the intralayer (inter-
molecular) hydrogen bond contribution to the total cohesive
energy. We found that these two energies were comparable,
with the van der Waals contribution being almost 90% that of
the hydrogen bond contribution. The hydrogen bonds them-
selves have an average strength of B3.5 kcal molÀ1, slightly
stronger than the hydrogen bonds in cellulose.23
The remarkable solubility of TATB in EMImOAc com-
pared to other ILs was initially very puzzling. Since the cation
of the ILs doesn’t seem to affect the solubility of TATB, we
concluded that the acetate moiety plays a key role in solubiliz-
ing TATB. TATB dissolved in EMImOAc produces a deep red
color, observed at lmax = 409 nm. The origin of this peak can
be assigned to a s-complex. There are many reviews on the
mechanism of the formation of s-complexes in intermolecular
nucleophilic displacement reactions involving electrophilic,
nitro-activated aromatic substrates.24 The mechanism gener-
ally involves the addition of a nucleophile to a position on the
electrophilic aromatic ring that results in the stabilization of
the negative charge by an ortho- and/or para-substituted nitro
group. The structure of the s-complex is shown in Fig. 7 and
consists of a cyclohexadienyl anion, in which the carbon center
that undergoes substitution is converted to an sp3-hybridized
center. The resulting negative charge may be delocalized into
Table 2 COSMO-RS results for TATB dissolution in EMImOAc
and EMImCl
TATB solubility
at 100 1C (wt%)a
Solvent
DEdeprotonation/kcal molÀ1b
EMImOAc
EMImCl
0.04
0.10
À1.6
+22.0
a
b
1 wt% = 1 g 100 mLÀ1 = 10 g LÀ1
.
Energy of the reaction: TATB +
]
EMIm+AnionÀ - EMIm+[TATBdeprotonated
positive (negative) for an endothermic (exothermic) reaction.
+ H-Anion; DE is
À
Fig. 7 Schematic of s-complex formation from TATB.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
54 | New J. Chem., 2009, 33, 50–56

acid, were also found to be effective. A limiting factor in the
ability of these inorganic salts to act as an efficient anti-solvent
was their solubility in DMSO. Some had limited solubility,
rendering them less effective in precipitating TATB in good
yields. The addition of gaseous CO2 to the solvent mixture also
precipitated TATB in excellent yields. It is not clear whether
the CO2 acted to neutralize the mixture, making it less basic,
or actually acted as a counter-solvent to precipitate the TATB.
The use of gaseous CO2 is attractive because of its cost and
availability, but it was hard to control the particle size of the
recrystallized TATB when adding CO2 in gaseous form.
The anti-solvent of choice was acetic acid because it was
reasoned that it could be removed from the IL to yield
recovered EMImOAc without contamination from other salts.
Preliminary results show that crystal size and shape are
strongly dependent on the rate of addition and the concentra-
tion of the anti-solvent. Detailed information regarding the
crystallization of TATB via the anti-solvent method will be
published elsewhere.
Fig. 8 An optical micrograph of recrystallized TATB prepared by the
slow cooling method.
ZINDO)28 of the deprotonated TATB displayed a sharp peak
at B410 cmÀ1, which is also in excellent agreement with our
experimental observations. Therefore, we cannot at this point
discount the possibility of alternative chemical pathways of
s-complex formation and the deprotonation mechanism.
We have also attempted to dissolve cellulose in EMImOAc.
Although a recent report29 showed a high solubility of cellu-
lose (Avicel PH-101) in EMImOAc (15 wt%) at 110 1C, the
solubility of our cellulose (Eastman Kodak) in EMImOAc at
100 1C was less than 0.5 wt%. This discrepancy can be
attributed to different cellulose sources and therefore the level
of recalcitrance of the cellulose tested. Our cellulose experi-
ment result supports the idea that the mechanism of solubility
for TATB in EMImOAc is indeed by chemical modification
(rather than by hydrogen bond disruption). It is important to
note that ILs can potentially modify the solute, as with TATB,
to increase its solubility. Therefore, one must be very cautious
in determining the solubility of various solutes in ILs to make
sure that chemical modification of the solute is not the cause of
the observed increase in solubility.
5. Conclusion
ILs previously shown to dissolve highly hydrogen-bonded
solids were ineffective with TATB. This may be due to the
cohesive energy of TATB, which is almost 2 times that of
cellulose. The remarkable solubility of TATB in EMImOAc
was attributed to the formation of a s-complex or to the
deprotonation of TATB. Owing to the basicity and nucleo-
philicity of the OAcÀ anion, the s-complex could easily form
in EMImOAc, but not in the presence other anions such as
ClÀ. Therefore, the solubility of TATB in EMImOAc is
proportional to the concentration of EMImOAc. The dis-
solved s-complex can revert back to TATB, either via cooling
or by the addition of an anti-solvent. The recrystallized TATB
shows a much improved crystal morphology compared to the
starting material. We are currently performing experiments to
further control the crystallization of TATB by the anti-solvent
crystallization method.
Crystallization of TATB via the non-agitated cooling
method in EMImOAc resulted in an improved morphology
of the TATB crystals obtained compared to the starting
materials. Cooling by natural convection yielded crystals in
the size range of 10–50 mm (Fig. 5a). However, when the
cooling rate was controlled, larger crystals were obtained. As
seen in Fig. 8, when the crystallization solution was cooling at
a rate of 1 1C minÀ1, 200–500 mm sized crystals were obtained.
One drawback of the cooling crystallization method is that at
room temperature, the amount of recovered TATB is less than
the original amount of TATB added, due to dissolved TATB
in the 80 : 20 solution; the addition of a protic solvent is
necessary to recover the remaining TATB. Thus, an anti-
solvent crystallization scheme was employed as an alternative.
As mentioned above, during our initial attempts to recrystal-
lize of TATB from DMSO–EMImOAc solution, the recovery
of the TATB was rather poor because it is soluble in the
mixture at a 1 wt% level at room temperature. The most
effective anti-solvents of all those tested were hydroxylic
compounds such as organic alcohols and acids. These solvents
provided a proton source that released the acetate from the
s-complex and precipitated the TATB. In addition, some
inorganic acids, such as sodium hydrogen phosphate and boric
Acknowledgements
This work performed under the auspices of the US Depart-
ment of Energy by Lawrence Livermore National Laboratory
under Contract DE-AC52-07NA27344. The project 06-SI-005
was funded by the Laboratory Directed Research and Develop-
ment Program.
References
1 R. P. Swatloski, S. K. Spear, J. D. Holbrey and R. D. Rogers,
J. Am. Chem. Soc., 2002, 124, 4974.
2 D. M. Phillips, L. F. Drummy, D. G. Conrady, D. M. Fox,
R. R. Naik, M. O. Stone, P. C. Trulove, H. C. De Long and
R. A. Mantz, J. Am. Chem. Soc., 2004, 126, 14350.
3 H. B. Xie, S. H. Li and S. B. Zhang, Green Chem., 2005, 7, 606.
4 H. H. Cady and A. C. Larson, Acta Crystallogr., 1965, 18, 485.
5 W. Selig, Lawrence Livermore National Laboratory Report
UCID-17412, 1977.
6 C. M. Tarver, J. W. Kury and R. D. Breithaupt, J. Appl. Phys.,
1997, 82, 3771.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
New J. Chem., 2009, 33, 50–56 | 55

7 S. F. Son, B. W. Asay, B. F. Henson, R. K. Sander, A. N. Ali,
P. M. Zielinski, D. S. Phillips, R. B. Schwarz and C. B. Skidmore,
J. Phys. Chem. B, 1999, 103, 5434.
8 J. L. Bredas, F. Meyers, B. M. Pierce and J. Zyss, J. Am. Chem.
Soc., 1992, 114, 4928.
9 G. Filippini and A. Gavezzotti, Chem. Phys. Lett., 1994, 231, 86.
10 I. G. Voigt-Martin, G. Li, A. Yakimanski, G. Schulz and
J. J. Wolff, J. Am. Chem. Soc., 1996, 118, 12830.
total cohesive energy was computed by taking the difference of
the total energy of a TATB unit cell (normalized per TATB
molecule) from that of an isolated TATB molecule. Next, a
vacuum slab normal to the basal plane was created in order to
isolate the hydrogen-bonded layers of TATB from each other. The
total energy of such a cell yielded a contribution to the total CED
coming from just the intralayer hydrogen bond interactions.
20 B. Delley, J. Chem. Phys., 1990, 92, 508.
11 I. G. Voigt-Martin, G. Li, A. A. Yakimanski, J. J. Wolff and
H. Gross, J. Phys. Chem. A, 1997, 101, 7265.
12 H. Zhang, J. Wu, J. Zhang and J. S. He, Macromolecules, 2005, 38,
8272.
13 M. Palusiak and S. J. Grabowski, J. Mol. Struct., 2002, 642, 97.
14 P. S. Santos and N. S. Goncalves, J. Raman Spectrosc., 1989, 20,
551.
15 A. R. Mitchell, M. D. Coburn, R. D. Schmidt, P. F. Pagoria and
G. S. Lee, Thermochim. Acta, 2002, 384, 205.
16 R. J. Roberts and R. C. Rowe, Int. J. Pharm., 1993, 99, 157.
17 J. M. Rosen and C. Dickinson, J. Chem. Eng. Data, 1969, 14, 120.
18 H. Jin, B. O’Hare, J. Dong, S. Arzhantsev, G. A. Baker,
J. F. Wishart, A. J. Benesi and M. Maroncelli, J. Phys. Chem. B,
2008, 112, 81.
19 We performed all-electron calculations employing the double
numeric polarized (DNP) basis set on a fine integration grid and
the gradient-corrected PBE exchange correlation function. The
21 B. Delley, J. Chem. Phys., 2000, 113, 7756.
22 B. Delley, Phys. Rev. B: Condens. Matter Mater. Phys., 2002, 66,
155125.
23 B. E. Dale and G. T. Tsao, J. Appl. Polym. Sci., 1982, 27, 1233.
24 G. A. Artamkina, M. P. Egorov and I. P. Beletskaya, Chem. Rev.,
1982, 82, 427.
25 A. Klamt and G. Schuurmann, J. Chem. Soc., Perkin Trans. 2,
1993, 799.
26 A. Klamt, COSMO-RS: From Quantum Chemistry to Fluid Phase
Thermodynamics and Drug Design, Elsevier, Amsterdam, 2005.
27 A. Maiti, P. F. Pagoria, A. E. Gash, T. Y. J. Han, C. A. Orme,
R. H. Gee and L. E. Fried, Phys. Chem. Chem. Phys., 2008, 10,
5050.
28 M. Zerner, Reviews in Computational Chemistry, VCH, New York,
vol. 2, 1991.
29 H. Zhao, G. A. Baker, Z. Y. Song, O. Olubajo, T. Crittle and
D. Peters, Green Chem., 2008, 10, 696.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2009
56 | New J. Chem., 2009, 33, 50–56