Synthesis of hexanitrostilbene (HNS) using a kenics static mixer

Article abstract of DOI:10.1021/op1000644

Previous work on a Shipp-Kaplan type synthesis of hexanitrostilbene (HNS) using a Kenics static mixer (ref1, from Chemineer Ltd., Cranmer Road, West Meadows, Derby DE21 6XT, UK; www.chemineer.com) has been extended by using a larger mixer (9.5 mm vs 4.75 mm OD), enabling the scale to be increased from 2.5 g to 25 g of trinitrotoluene (TNT) for the same mixing time (<4 min). Various conditions for the after-reaction period (2 h) have been explored on the basis of previous work with a batch reactor, including (i) no pH control, (ii) pH control using aqueous H₂SO₄ and NaOH solutions, and (iii) pH control using aqueous RNH₃Cl and RNH₂ solutions. Other parameters that have been varied are the ratio of NaOCl to TNT (0.5-1.2) and the concentration of both reactants. The yields of crude HNS that have been achieved, whilst not outstanding, are an improvement over the conventional batch process. It has been demonstrated that the yield and selectivity of the HNS synthesis can be considerably increased if, during the after-reaction period, the reaction conditions are pH controlled. The yield and selectivity are also significantly enhanced by using more dilute reaction solutions.

Full text of DOI:10.1021/op1000644

Organic Process Research & Development 2010, 14, 632–639
Synthesis of Hexanitrostilbene (HNS) using a Kenics Static Mixer
Anthony J. Bellamy*
Cranfield UniVersity, Defence Academy of the United Kingdom, ShriVenham, Swindon SN6 8LA, U.K.
Scheme 1. Reaction scheme for NaOCl oxidation of TNT
Abstract:
Previous work on a Shipp-Kaplan type synthesis of hexani-
trostilbene (HNS) using a Kenics static mixer (ref 1, from
Chemineer Ltd., Cranmer Road, West Meadows, Derby DE21
6XT, UK; www.chemineer.com) has been extended by using a
larger mixer (9.5 mm vs 4.75 mm OD), enabling the scale to be
increased from 2.5 g to 25 g of trinitrotoluene (TNT) for the same
mixing time (<4 min). Various conditions for the after-reaction
period (2 h) have been explored on the basis of previous work
with a batch reactor, including (i) no pH control, (ii) pH control
using aqueous H2SO4 and NaOH solutions, and (iii) pH
control using aqueous RNH3Cl and RNH2 solutions. Other
parameters that have been varied are the ratio of NaOCl
to TNT (0.5-1.2) and the concentration of both reactants.
The yields of crude HNS that have been achieved, whilst
not outstanding, are an improvement over the conventional
batch process. It has been demonstrated that the yield and
selectivity of the HNS synthesis can be considerably in-
creased if, during the after-reaction period, the reaction
conditions are pH controlled. The yield and selectivity are
also significantly enhanced by using more dilute reaction
solutions.
1. Introduction
The synthesis of 2,2′,4,4′,6,6′-hexanitrostilbene (HNS) by
reacting a solution of trinitrotoluene (TNT) with sodium
hypochlorite (NaOCl) was first reported by Shipp and Kaplan2,3
in 1964 and has been widely studied since.4-7 As described by
Shipp and Kaplan,3 the reaction conditions may be varied so
that the isolated product is either HNS (by adding TNT to
NaOCl - yield normally 40-45%), hexanitrobibenzyl (HNBB,
by adding NaOCl to TNT - yield normally ∼80%), or
trinitrobenzyl chloride (TNBCl), the initially formed reaction
intermediate (by quenching the reaction mixture in acid a short
time after mixing - yield normally ∼85%). The likely reaction
scheme is shown in Scheme 1. One can envisage that HNS is
formed by self-condensation of TNBCl (reaction between
TNBCl and its conjugate base B), via R-chlorohexanitrobiben-
zyl7 (R-Cl·HNBB; C in Scheme 1), whilst HNBB is formed
by cross condensation of TNBCl with the conjugate base of
TNT (A), the mode of addition generating conditions which
favor one or the other of these two products. In less controlled
mixing regimes both products are likely to be formed.
* Author for correspondence. E-mail: banda.bellamy@btinternet.com.
(1) From Chemineer Ltd., Cranmer Road, West Meadows, Derby DE21
6XT, UK; www.chemineer.com.
(2) Shipp, K. G. Reactions of R-substituted polynitrotoluenes. I. Synthesis
of 2,2′,4,4′,6,6′-hexanitrostilbene. J. Org. Chem. 1964, 29, 2620–2623.
(3) Shipp, K. G.; Kaplan, L. A. Reaction of R-substituted polynitrotolu-
enes. II. The generation and reactions of 2,4,6-trinitrobenzyl anion. J.
Org. Chem. 1966, 31, 857–861.
In the batch synthesis of HNS, the reaction occurs in
essentially two stages. In stage one (mixing - duration )
minutes), TNT reacts exothermically with NaOCl to form
TNBCl. In stage two (ageing - duration ≈ 2 h), the TNBCl is
allowed to react with the basic medium (NaOCl/NaOH)
generating HNS which precipitates as a solid. The best
selectivity in the reaction is achieved if the temperature on
mixing is kept below 15 °C and the pH during the ageing period
is maintained around 10.4,9 Shipp and Kaplan3 kept the mixing
temperature low by precooling both solutions to 0 °C, but did
(4) Golding, P.; Hayes, G. F. A parametric study of the synthesis of
2,2′,4,4′,6,6′-hexanitrostilbene from trinitrotoluene and sodium hy-
pochlorite. Propellants, Explos., Pyrotech. 1983, 8, 35–39.
(5) Singh, B.; Singh, H. Synthesis of 2,2′,4,4′,6,6′-hexanitrostilbene. Def.
Sci. J. 1982, 31, 305–308.
(6) Singh, B.; Malhotra, R. K. Hexanitrostilbene and its properties. Def.
Sci. J. 1983, 33, 165–176.
(7) Kayser, E. G. An investigation of the Shipp hexanitrostilbene (HNS)
process. J. Energ. Mater. 1983, 1, 325–348.
632
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development
10.1021/op1000644  2010 American Chemical Society
Published on Web 04/22/2010

not control the pH during the ageing period (initially 11-12,
reducing to <7 during the ageing period due to elimination of
HCl). Under these conditions the reaction temperature peaked
at 24 °C after 2 min (10 g scale; see Table 3, Run F).
A batch HNS process as it was operated at the Royal
Ordnance Factory, Bridgwater, UK, has been reported.9
This involved mixing a 10% solution of TNT in tetrahy-
drofuran (THF)/methanol and 5% aqueous sodium hy-
pochlorite at 13 °C before lowering the initial pH (11-12)
to ∼10 with carbonate/bicarbonate buffer. After an ageing
period of 2 h, the precipitated HNS was coagulated by
acidification with dilute HCl, filtered off and washed
sequentially with acetone and water. The yield of HNS
was ∼45% and could be produced at the rate of ∼20
tonnes per annum. Both THF and methanol from the
mother liquor and acetone from the washings were
recycled. The plant was closed for economic reasons
caused by environmental issues. At the present time HNS
is produced via a batch process by Eurenco Bofors in
Sweden and China.
In scaling up the batch process, one of the major problems
is controlling the temperature at the point of mixing. This is
also likely to be a problem if the reaction is performed as a
semi-continuous flow system. Any excess temperature rise on
mixing will cause HNS to form and precipitate prematurely in
the mixing tube and result in blockage. In order to minimise
this problem we have explored10 the use of a Kenics static
mixer1 immersed in ice-water in the hope that any heat
generated on mixing will be rapidly dissipated through the metal
walls of the mixer into the surrounding bath.
This preliminary study10 using a Kenics static mixer [37-
03-075, material 316SS, length 190 mm, OD 4.75 mm, with
27 mixing elements] investigated the following aspects:
1.1. HNS vs HNBB Formation. The yield and composition
of the product from a fixed amount of TNT (1.8 g) was
investigated as the amount of NaOCl was varied from 0.5 to
1.8 equiv, with no attempt being made to control the pH in the
ageing period. In theory only one equivalent of NaOCl is
required to convert all of the TNT to TNBCl and thence to
HNS. In practice, the highest material yield occurs in the region
of 1 equiv of NaOCl, but the product in that case contains a
high proportion of HNBB. The highest proportion of HNS
(93-96%) occurs when 1.2-1.4 equiv of NaOCl are used, but
by this stage the material yield has started to decrease. The
proportion of HNBB in the product steadily decreases from 86%
when 0.5 equiv of NaOCl is used and to ∼5% when >1.2 equiv
are used.
Figure 1. Kenics static mixer (schematic) showing three mixing
elements.
and the pH during the ageing period was controlled by addition
of aqueous H2SO4 and NaOH. Consistently higher material
yields were obtained with pH control, compared to the
uncontrolled reaction with 1.2 equiv of NaOCl. The material
yield and composition of the product was not particularly
sensitive to pH in the range 9.5-11.5, but slightly purer HNS
was obtained in the lower part of this range. This is in agreement
with Golding and Hayes’ observations4 from a series of batch
reactions.
1.3. Quenched reactions. One of the main objectives of
our former study10 was to prepare TNBCl using a flow reactor,
quenching the reaction with HCl almost immediately after
mixing. Since the highest selectivity for HNS formation most
probably reflects efficient conversion of TNT to TNBCl, we
chose those reactant proportions that gave the purest HNS, viz
1.2-1.4 equiv of NaOCl, for testing in quenched reactions.
Reactions using 1.4 equiv of NaOCl gave high selectivity
(∼95%) and high material yield (∼95%) of TNBCl. Indeed,
these results are similar to those one obtains using the Shipp
and Kaplan procedure,3 viz ∼98% purity and 87% material
yield.
1.2. pH Control. All experiments were conducted using 1.2
equiv of NaOCl (the amount that gave a low proportion of
HNBB but without reducing the material yield excessively),
Our earlier study has now been extended by using a larger
Kenics static mixer.1
(8) Bellamy, A. J. Identification of R-chloro-2,2′,4,4′,6,6′-hexanitrobibenzyl
as an impurity in hexanitrostilbene (HNS). Proceedings of the 12th
Seminar; New Trends in Research of Energetic Materials, Pardubice,
Czech Republic, 2009, pp 433-441. J. Energ. Mater. 2010, 28, 1-
16.
2. Experimental Section
Whilst both TNT and HNS are classed as relatiWely
insensitiWe explosiWes, they should not be subjected to excessiWe
impact, friction, or electrostatic discharge.
2.1. Mixing System. The mixer-reaction system employed
consisted of a Kenics Static Tube Mixer 37-06-1101 [material
316SS, length 356 mm, OD 9.5 mm, ID 8.1 mm,] with 27
mixing elements (see Figure 1 - only 3 of the 27 mixing
(9) Gallo, A. E.; Tench, N. Commisioning of a production plant for
hexanitrostilbene. J. Hazard. Mater. 1984, 9, 5–11.
(10) Bellamy, A. J., Lomax, V. Read, K. The use of a Kenics static mixer
for the reaction of TNT with NaOCl. Proceedings of the 9th Seminar;
New Trends in Research of Energetic Materials, Pardubice, Czech
Republic, 2006, pp 491-499.
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
633

Figure 3. Top of static mixer and inlet system.
Figure 4. Complete system in ice-water bath.
Figure 5. Mixing system in ice-water bath.
Figure 2. Intercoiled inlet tubes (left), coiled exit tube (9 m,
right), static mixer (vertical in centre).
elements are shown), oriented vertically with the inlet at the
top. The reactant solutions [TNT in THF/MeOH/water and
aqueous NaOCl] were delivered by two peristaltic pumps
[Watson Marlow model 302S, butyl tubing], via PTFE tubing
[ID 3.2 mm, OD 4.8 mm] to the mixing tee [Swagelok SS-
600-3, ID 7.1 mm, internal volume after mixing point ∼0.55
cm3] (or cross-piece [Swagelok SS-600-4] when it was required
to monitor the internal temperature at the point of mixing). The
mixing tee was directly attached to the top end of the static
mixer. The outflow from the bottom of the static mixer was
conveyed by PTFE tubing [ID 3.2 mm, OD 4.8 mm, variable
length 0.8-9 m], also immersed in the ice-water bath, to the
ageing or quenching vessel. Various parts of the system are
shown in Figures 2-5. Before commencing a reaction,
the whole of the system [inlet tubes (to ensure that the
reactant solutions were precooled), mixing tee, static
mixer, outlet tube] was immersed in ice-water contained
in a stirred vessel (Figure 4, 5), and the inlet tubes were
primed to within 2 cm of the mixing tee (Figure 6). The
PTFE delivery tubes were partially coiled (diam ≈ 6 cm,
total length of immersed coiled tube ≈ 2 m). This allowed
precooling of the solutions from ambient to 1-2 °C before
mixing. The exit tube was mainly in the form of a coil
[diam ≈ 6 cm, total length of immersed coiled tube )
3-9 m].
that reported in a UK patent.11 In the present work the TNT
was dissolved in THF/MeOH/H2O (60:30:10% by volume) -
see Table 1 (all processes scaled to 25 g of TNT), since in a
2.2. Choice of Reaction Conditions (Table 1). The original
Shipp-Kaplan procedure3 (10-g scale) used TNT in THF/
MeOH (150 mL; 2:1 by volume) and 1.53 equiv of 5 wt %
NaOCl (volume equal to that of THF). The procedure adopted
at Bridgwater9 appears to mimic that of Shipp and Kaplan and
(11) Jones, R. H.; Pryde, A. W. H. Production of hexanitrostilbene, GB
1570569, 1980.
634
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

Table 1. Reagent proportions for the NaOCl oxidation of TNT
TNT,
g
THF,
mL
MeOH,
mL
H2O,
mL
NaOCl,
mL
NaOCl,
% wt/vol
NaOCl,
equiv
comments
Batch Processesa
Shipp-Kaplanb
25
25
25
250
213
180
125
107
90
0
0
250
225
90
5.0
1.53
batch process for HNS
(10 g scale)
GB 1570569c
present work
4.4
1.2
batch process for HNS
(500 g scale)
30
8.2-11.8
0.9-1.3
optimisation of a batch recipe
(50 g scale)
static mixer
present work
present work
present work
25
25
25
180
180
180
90
90
90
30
30
30
200
200
200
11.6
11.6
2.5
1.32
1.37
0.86
HNS synthesis (example)
TNBCl synthesis (example)
HNBB synthesis; blocked tube
3
11
a Scaled to 25 g TNT. b Reference c Reference
Figure 6. (a) Tap positions for priming of inlet tubes (b) Tap positions for mixing (arrows indicate flow direction).
commercial process the solvent is recovered (as a mixture by
distillation) from a previous run and often contains 10-12 wt
% H2O.
The combined flow rate of the two solutions lies within the
turbulent flow regime of the Kenics static mixer.1
(1) Weigh both reagent solutions.
The reactant proportions adopted for the flow mixer system,
both in previous work10 and the present work, are based on a
commercial recipe (see Table 1).
(2) Both T-taps set for vertical flow (see Figure 6a), pump
until the reagents are just above the taps, then switch pumps
off.
2.3. Standard Procedure. TNT solution: TNT (25 g) was
dissolved in THF (180 mL) and then diluted with MeOH (90
mL) and H2O (30 mL). Total volume after dissolution ∼320
mL (0.078 g/mL, 0.344 mmol/mL). Pump: 4.8 mm ID butyl
tubing, pump operating at 40% of its maximum flow i.e. ∼1.20
mL/s.
NaOCl solution: the appropriate volume of concentrated
NaOCl solution (14.89 g/100 mL) was diluted with H2O to 200
mL to give the required equivalents of NaOCl. For example,
for 1.2 equiv of NaOCl, 137 mL of the original NaOCl solution
(0.1489 g/mL) diluted to 200 mL (0.102 g/mL, 1.37 mmol/
mL). Pump: 3.2 mm ID butyl tubing, pump operating at 27.5%
of its maximum flow i.e. ∼0.36 mL/s.
(3) Turn both T-taps to allow flow into the mixer (see Figure
6b).
(4) Switch on both pumps simultaneously. Collect the
reaction solution leaving the exit tube in a magnetically stirred,
uncooled flask. Switch off both pumps when the last of the
TNT solution or NaOCl solution leaves the reservoir. Cease
collection at this point (normally <4 min).
(5)Turn both T-taps back to vertical flow.
(6) Reverse pumps and draw back reagents in the tubes. Then
reweigh both reagent solutions.
(7) Flush system, via the TNT inlet system, with MeOH/
THF/H2O (50 mL).
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
635

Table 2. Data sheet for each experiment
run: 10
experiment: standard, 9 m exit tube
conditions:
TNT (25 g) in THF (180 mL) +
MeOH (90 mL) + H2O (30 mL); pump: 4.8 mm i..d.; 40%
maximum flow
NaOCl (19.95 g) in H2O (200 mL);
1.19 equiv; pump: 3.2 mm ID; 27.5% maximum flow
mixing time:
4 min 00 s (240 s)
mass (g)
TNT solution
NaOCl solution
total
initial + flask
flask
443.1
158.4
284.7
195.1
158.4
36.7
343.1
109.6
233.5
243.2
109.6
133.6
Figure 7. 1H NMR spectrum (DMSO-d6) of crude HNS
(illustrative example). Data: TNT 2.56 (s, CH3), 9.03 ppm (s,
2H, Ar); TNBCl 5.00 (s, CH2Cl), 9.09 ppm (s, 2H, Ar); HNS
7.14 (s, 2H, CHdCH), 9.11 ppm (s, 4H, Ar); HNBB 3.35 (s,
4H, CH2), 9.08 ppm (s, 4H, Ar); r-Cl·HNBB8 4.20 (d, 2H, J )
6.9 Hz), 5.90 ppm (t, 1H, J ) 6.8 Hz); due to the low abundance
of this product, the aromatic peaks were not visible.
initial
final + flask
flask
final
introduced (used)
248.0
+
99.9
) 347.9
collected + flask
flask
collected
455.2
197.2
258.0
collection vessel, it would have increased the mass of the
product proportionately.
Thus, the yield of product ) mass of product + % lost in
mixer and exit tube (see Table 2).
lost in system
347.9
-
258.0
) 89.9 (-25.8%)
Percentage yield (assuming all product is HNS) ) 100 ×
[mass of product isolated + % lost in mixer and exit tube] ×
[2 × molecular mass of TNT]/[mass of TNT consumed] ×
[molecular mass of HNS].
(8) Flush system, via the NaOCl inlet system, with H2O (100
mL).
(9) Flush system, via the NaOCl inlet system, with NMP
(50 mL).
(10) Flush system, via the NaOCl inlet system, with H2O
(100 mL).
(11) Flush system, via the TNT inlet system, with MeOH
(50 mL).
(12) Leave both reagent systems empty (by reversing the
pumps) and remove tubing from pumps.
(13) Weigh the contents of the receiver before filtering off
the product.
At the end of the reaction period the product was filtered
off on a preweighed sintered glass funnel (No. 4, 6.5 cm diam),
applying suction until the filtrate flow had almost ceased. It
was then washed with (i) methanol (50 mL) - the funnel walls
were washed down 3× before adding the bulk of the methanol
- and (ii) water (50 mL). The product was dried in a vacuum
desiccator over silica gel. The results are tabulated as shown in
Table 2 (typical entry).
2.4. Product Analysis. 1H NMR (10 mg/1.0 mL DMSO-
d6) was used to confirm product identity and distribution. An
illustrative NMR spectrum is shown in Figure 7.
HPLC may also be used to analyse the product mixture but
was not used in the present work. Conditions: µ-Bondapak C-18
10 µm; 3.9 mm × 150 mm; mobile phase: 52.5% water/47.5%
acetonitrile; flow rate 1.0 mL/min. The retention times were:
TNT 5.6 min, TNBCl 6.4 min, HNS 11.6 min, HNBB 13.8
min, R-Cl·HNBB 16.4 min.
2.5. pH-Controlled Reactions. The mixing stage was as
above. When mixing was complete, the pH of the reaction
mixture was reduced from 12-13 to the desired pH e.g. 10,
by the addition of 25 wt % H2SO4 (or to pH 9.7 by the addition
of aqueous 4 M aqueous ethanolamine hydrochloride12), and
the pH was then controlled at the desired pH by automatic
addition of 6 wt % NaOH (alternatively neat ethanolamine12).
Control equipment: combination glass electrode (Thermo
Scientific CW757/A ) 300 mm), Radiometer Copenhagen
PMH 82 standard pH meter with a TTT titrator). The filtered
product was washed with MeOH (50 mL) and water (2 × 50
mL).
Calculation of conditions actually used (using data in Table
2):
284.7 g of TNT solution contains 25 g of TNT ∴ the 248.0
g of TNT solution used contains 21.8 g (96 mmol) TNT.
233.5 g of NaOCl solution contains 19.95 g of NaOCl ∴
the 99.9 g of NaOCl solution used contains 8.54 g (115 mmol)
of NaOCl.
Ratio of reactants (NaOCl/TNT) ) 115/96 ) 1.19 equiv.
The material remaining in the mixer and exit tube when the
pumps were switched off (this material could neither be pumped
into the receiver vessel nor withdrawn into the reagent storage
flasks due to the geometry of the system) was calculated by
subtracting the mass of the collected solution (measured at the
end of the reaction period) from the sum of the TNT solution
and the NaOCl solution delivered into the mixing chamber. It
was assumed that, had this portion been pumped into the
2.6. Quenched Reactions. The mixing stage was as above.
However, the mixture leaving the exit tube was added to a
solution of conc HCl (25.4 mL) in H2O (1250 mL). This caused
precipitation of the products. The resultant mixture was stirred
at ambient temperature during 1 h before the product was
filtered off and washed with water.
2.7. Shipp-Kaplan Conditions (Run F, Table 3). A
solution of TNT (10 g) in THF (100 mL) + MeOH (50
(12) Sollott, G. P. Conversion of 2,4,6-trinitrobenzyl chloride to 2,2′,4,4′,6,6′-
hexanitrostilbene by nitrogen bases. J. Org. Chem. 1982, 47, 2471–
2474.
636
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

Table 3. Results of reaction of TNT with NaOCl
1
product composition by H NMR (mol %)
TNT
consumed, yield, yield,
NaOCl,
equiv
runa
g
g
%
HNS HNBB TNBCl TNT R-Cl·HNBB THF
conditions
A
B
C
50.0
50.0
50.0
18.0
21.9
28.2
36.3
44.2
57.0
1.36
1.00
1.00
93.5
87.2
83.8
1.5
6.4
3.2
0.8
1.3
1.9
0.0
0.0
1.8
4.2
5.1
9.4
3.9 batch reactor
5.8 batch reactor
3.7 batch reactor; pH 10.2 controlled
with H2SO4 and NaOH
D
E
F
50.0
50.0
10.0
30.8
29.6
3.5
62.1
59.7
35.3
1.00
1.36
1.55
85.6
94.2
93.8
2.0
0.9
0.0
1.3
1.4
1.0
1.7
0.0
0.0
9.4
3.5
5.2
3.6 batch reactor; pH 10.4 controlled
with H2SO4 and NaOH
3.8 batch reactor; pH 9.7 controlled
with ethanolamine buffer
6.0 Shipp-Kaplan synthesisbmaxi-
mum temp. 24 °C
1
2
3
17.6
21.9
21.9
7.16 41.0
9.20 42.4
30.15
1.22
1.13
1.26
88.1
78.9
6.6
14.6
1.3
1.7
70.1
0.0
0.0
29.9
4.0
4.8
6.2 standard; 80 cm exit tube
7.6 standard; 300 cm exit tube
standard; 300 cm exit tube;
quenched after mixing
4
5
6
7
8
(0.56)
1.66
1.15
0.79
0.86
standard, but NaOCl equiv halved;
300 cm exit tube; blocked
5.4 standard, but NaOCl undiluted;
300 cm exit tube
6.3 standard, but both concentrations
halved; 300 cm exit tube
2.3 standard, but pumps at half speed;
300 cm exit tube
1.2 standard, but NaOCl halved and
both concentrations halved; 300
cm exit tube; blocked
15.5
21.7
22.0
4.18 27.2
9.57 44.5
95.3
90.0
26.6
12.7
0.0
4.4
0.8
1.1
0.0
0.0
0.0
0.0
6.1
9.2
4.0
4.5
0.0
0.0
10.3
1.25
47.2
67.3
78.1
7.33
9
10
11
16.3
21.8
21.4
7.15 44.3
8.54 39.5
1.02
1.19
1.37
74.6
73.0
19.8
20.2
1.3
1.7
94.5
0.0
0.0
5.5
4.3
5.2
6.7 standard; 300 cm exit tube
7.2 standard; 900 cm exit tube
standard; 900 cm exit tube;
quenched after mixing
20.1
81.5
12
13
21.5
23.1
7.93 37.2
9.77 42.7
1.32
1.27
86.1
93.9
7.3
0.0
1.5
1.2
0.0
0.0
5.1
4.9
6.5 standard; 900 cm exit tube
5.6 standard, but both concentrations
halved; 900 cm exit tube
14
23.1
23.3
87.6
1.25
87.4
12.6
standard, but both concentrations
halved; 900 cm exit tube;
quenched after mixing
15
16
21.6
21.4
11.8
8.6
55.1
40.5
1.21
1.20
81.4
74.7
9.2
2.6
1.6
0.0
0.0
6.8
4.8
5.2 standard; 900 cm exit tube; pH
10.0
7.6 standard, but both concs halved
(as run 15; but no pH control);
900 cm exit tube
18.9
17
18
19
21.4
21.3
22.9
10.4
10.7
11.7
49.0
50.7
51.5
1.24
1.02
1.03
89.1
64.6
74.2
1.5
21.6
13.9
1.9
2.6
2.4
0.0
3.2
1.6
7.5
8.1
7.9
5.6 standard; 900 cm exit tube; pH
10.0,
5.2 standard; 900 cm exit tube; pH
10.2
5.9 standard, but both concentrations
halved; 900 cm exit tube; pH
10.2
20
21
22
21.3
23.1
20.9
11.8
13.3
5.9
55.9
58.1
28.5
1.16
1.19
1.55
76.9
89.6
94.2
12.0
3.2
2.7
1.9
0.9
2.2
0.7
0.0
6.3
4.7
4.9
5.7 standard; 900 cm exit tube; pH
9.7 controlled by ethanolamine
buffer
5.7 standard, but both concs halved;
900 cm exit tube; pH 9.7 by
ethanolamine buffer
0.0
6.3 standard; Shipp-Kaplan propor-
tions; 900 cm exit tube
a A-E: batch reactions in a stirred reactor (50 g scale); F: Shipp-Kaplan batch reaction (10 g scale); 1-22: static mixer reactions (25 g scale). b Reference 3.
mL) was cooled to ∼0 °C and was then added with rapid
mixing to 5% NaOCl (100 mL; also precooled to ∼0 °C).
The ensuing exothermic reaction caused the temperature
to rise to a maximum temperature of 24 °C within 2 min.
The very dark-coloured solution was then cooled to 15
°C and left for 2 h, during which time the temperature
rose slowly to 20 °C. The precipitated HNS was isolated
as described above. Yield of crude HNS 3.5 g (35%): see
Table 3, Run F.
2.8. Reactions with <1 equiv of NaOCl. Attempts were
made to synthesise hexanitrobibenzyl (HNBB) by reducing the
equivalents of NaOCl to ∼0.5, but in general these failed due
to unreacted TNT separating from the reaction mixture (due to
the increased water content after mixing).
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
637

3. Results and Discussion
gave a 56% yield with reduced selectivity. However a similar
run (Run 21) with halved concentrations gave 58% yield with
high selectivity.
The complete set of results for reactions (runs 1-22) utilising
the Kenics static mixer system as described above (arranged
chronologically), and comparable batch reactions (runs A-F)
mixed by adding the NaOCl solution to the TNT solution in a
stirred reactor, are presented in Table 3.
Runs 1-11 were essentially exploratory in nature to ascertain
the proper functioning of the system and to make any adjust-
ments that were deemed necessary.
From previous work10,13 it is known that the byproduct
R-chloro-2,2′,4,4′,6,6′-hexanitrobibenzyl (R-chloro·HNBB) in
crude HNS is converted by recrystallisation from NMP into
2-chloro-2′,4,4′,6,6′-pentanitrostilbene, which remains in the
recrystallised HNS. It is therefore important to reduce the
production of this byproduct as much as possible, even at the
expense of increasing the HNBB content since the latter is lost
on recrystallisation. In the present flow runs, the isolated HNS
contained 4-8 mol % of this byproduct. Other results from
batch experiments10,13 indicate that a higher pH level during
the after-reaction period favors the formation of R-chloro·HNBB.
Attempts to synthesise hexanitrobibenzyl (HNBB), by
reducing the equivalents of NaOCl to ∼0.5, failed due to
unreacted TNT separating from the reaction mixture (due to
the increased water content after mixing). This blocked the static
mixer (Runs 4 and 8).
Runs A and B were normal batch reactions (essentially
commercial conditions) with different equivalents of NaOCl
(1.36 and 1.00 equiv, respectively). The yields of crude HNS
were 36 and 44%, respectively, the HNS contents were 93 and
87%, respectively, and the HNBB contents were 1.5 and 6.4%,
respectively. As observed previously (see Introduction 1.1) the
proportion of HNBB decreases as the number of NaOCl
equivalents is increased.
Runs 3, 11, and 14 were quenched reactions performed to
measure the extent of reaction before the reaction mixture left
the exit tube. In Run 3 the exit tube was ∼3 m long; in Runs
11 and 14 it was ∼9 m long. From the amount of TNT
remaining in the quenched product, it was clear that the 9 m
exit tube was necessary for an almost complete reaction. With
only a 3 m exit tube, 30% TNT remained unreacted. If a
significant proportion of the TNT was still present at this point,
its continued reaction with NaOCl would generate heat in the
ageing vessel, giving less control over the after-reaction
conditions. This was observed as a more rapid temperature rise
(22 °C vs 13.5 °C after 30 min, respectively) in the collection
vessel in later runs. Run 11 indicated that moderately pure
TNBCl could be synthesised by this method, the only impurity
being TNT.
Run F was a standard (10-g scale) Shipp-Kaplan HNS
synthesis,3 and Run 22 was a static mixer run utilising the same
reactant concentrations and proportions as those in the
Shipp-Kaplan synthesis. Whilst the selectivity was high in both
cases, the yield of crude HNS was lower in the flow system
(28 vs 35%).
Runs 12 and 13 used the same number of equivalents of
NaOCl (∼1.3), but Run 13 used halved concentrations of both
reactants. The yield (37 vs 43%, respectively) and selectivity
(86 vs 94 mol % HNS, respectively) were better in the diluted
system.
Further observations relate to the observed temperatures, at
or after mixing. With water being pumped through both inlet
systems the observed temperature at the mixing point was 3.5 °C
(pump settings 40% and 27.5%). Thus, even though both liquids
had passed through 2 m of PTFE tubing immersed in ice-water,
the temperature of the liquid had only been reduced from
ambient to 3.5 °C. It is suggested that SS tubing, with its better
thermal conductivity, should be substituted for PTFE in the inlet
tubing coils in order to achieve better cooling of the inlet
solutions. That this temperature is the same as that observed
during an actual HNS run suggests that the present positioning
of the temperature sensor does not record the temperature
generated by the exothermic reaction on mixing. A temperature
sensor positioned at the bottom of the static mixing tube should
be explored.
The temperature in the collecting vessel during an actual
HNS run was never lower than 5.5 °C. This suggests that the
exothermic reaction may not be complete when the reaction
solution leaves the exit tube. This is supported by the observa-
tion in Runs 11 and 14 that some TNT remains unreacted at
this stage. Replacement of the PTFE exit tubing by SS tubing
should enhance the efficiency of cooling the reaction mixture.
This, of course, would reduce the rate of reaction further,
necessitating an extension (or broadening) of the exit tube in
order to achieve the same or increased extent of reaction at
this stage. Both changes would certainly be required if the pump
rate were to be increased in order to increase the scale of the
reaction.
Runs C and D were batch reactions run with pH control
(H2SO4, NaOH) during the after-reaction period, Run C at pH
10.2 and Run D at pH 10.4. Both used 1.0 equiv of NaOCl.
The yields of crude HNS were 57 and 62%, respectively. Flow
mixing, Run 17 (1.24 equiv of NaOCl) and Run 18 (1.02 equiv
of NaOCl), followed by control (H2SO4, NaOH) at pH 10.0
and 10.2, respectively, gave yields of ∼50% in both cases, but
selectivity was much higher with the greater proportion of
NaOCl (Run 17). A repeat of Run 18 with halved concentrations
(Run 19), gave essentially the same yield, but slightly better
selectivity.
In the present work, PTFE cooling coils were used both
before and after the mixer because they were more flexible to
work with. It was appreciated at the outset that SS coils would
give more efficient cooling. However, without experimental data
using the PTFE coils, it was not possible to say whether they
would have been of sufficient length to achieve the necessary
cooling.
Run E was a batch reaction run with pH control (pH 9.70)
using an ethanolamine buffer. The yield of crude HNS was 60%
with high selectivity. A corresponding static mixer run (Run
20), but with lower NaOCl (1.16 equiv; Run E used 1.36 equiv),
(13) Bellamy A. J. Unpublished work.
638
•
Vol. 14, No. 3, 2010 / Organic Process Research & Development

4. Conclusions
of the HNS synthesis can be considerably increased if, during
the after-reaction period, the reaction conditions are pH
controlled. The yield and selectivity are also significantly
enhanced by using more dilute reaction solutions.
The same mixer should be capable of mixing on a larger
scale by increasing the flow rate of the two solutions, but this
would necessitate an extension (or broadening) of the exit tube
to give sufficient reaction time before the reaction solution
leaves the system. Both the inlet PTFE cooling coils and the
exit PTFE cooling coil should be replaced by SS coils in order
to achieve more efficient cooling.
The use of larger, preferably jacketed Kenics static mixers1
for contacting the NaOCl and TNT solutions would enable the
temperature during the short period after contact, during which
the exothermic reaction to give trinitrobenzyl chloride occurs,
to be better controlled. This should lead to better selectivity in
the process. The control of conditions during the after-reaction
period merits further investigation.
Previous work10 on a Shipp-Kaplan type synthesis of
hexanitrostilbene (HNS) using a small Kenics static mixer1 has
been extended to the use of a larger mixer (9.5 mm vs 4.75
mm OD) enabling the scale to be increased from 2.5 g to 25 g
of TNT for the same mixing time (<4 min). In order to allow
almost complete initial reaction between the TNT and the
NaOCl before exiting the system (conversion to trinitrobenzyl
chloride TNBCl) the exit tube connected to the end of the mixer,
also cooled in ice-water, had to be extended from 3 m to >9
m. This was studied by quenching the reaction on exiting the
tube and measuring the consumption of TNT. The PTFE inlet
tubes to the mixer (length ∼2 m, also cooled in ice-water)
and the PTFE exit tube were configured as coils (6 cm diam).
Various conditions for the after-reaction period (2 h) have been
explored on the basis of previous work with a batch reactor
(50-g scale of TNT), including (i) no pH control, (ii) pH control
using aqueous H2SO4 and NaOH solutions, and (iii) pH control
using aqueous RNH3Cl and RNH2 solutions. Other parameters
that have been varied are the ratio of NaOCl to TNT (0.5-1.2)
and the concentration of both reactants. Attempts to synthesise
HNBB by using a low NaOCl/TNT ratio failed due to blockage
in the mixer from precipitation of TNT.
Acknowledgment
Financial support from UK Energetics (Contract No FTS1/
RAOWPE/01)andCranfieldUniversityisgratefullyacknowledged.
The yields of crude HNS that have been achieved, whilst
not outstanding, are an improvement over the conventional batch
process. It has been demonstrated that the yield and selectivity
Received for review March 6, 2010.
OP1000644
Vol. 14, No. 3, 2010 / Organic Process Research & Development
•
639