Scale-up and safety evaluation of a sandmeyer reaction

Article abstract of DOI:10.1021/op0498823

A scale-up of a new process for the production of 2-chloro-5- trifluoromethyl-benzonitrile by a Sandmeyer reaction of 2-Chloro-5- trifluoromethylaniline with copper cyanide/sodium cyanide is described. To ensure a safe implementation, a safety evaluation of the process was carried out, which is described in the following. The new process gives a better working environment and better productivity due to shorter reaction time.

Full text of DOI:10.1021/op0498823

Organic Process Research & Development 2004, 8, 1059−1064
Scale-Up and Safety Evaluation of a Sandmeyer Reaction
Martin Anker Nielsen,† Michael Kim Nielsen,‡ and Thomas Pittelkow*,‡
Chemical Production, H. Lundbeck A/S, 182 OddenVej, 4500 Nykøbing Sj, Denmark
Scheme 3. Formation of
2-chloro-5-trifluoromethyl-benzonitrile
Abstract:
A scale-up of a new process for the production of 2-chloro-5-
trifluoromethyl-benzonitrile by a Sandmeyer reaction of 2-Chlo-
ro-5-trifluoromethylaniline with copper cyanide/sodium cyanide
is described. To ensure a safe implementation, a safety evalu-
ation of the process was carried out, which is described in the
following. The new process gives a better working environment
and better productivity due to shorter reaction time.
In the current procedure, developed in the early 1970s,
four batches of the diazonium salt solution were produced
by mixing V (2.0 kg) with ice (4 L) and concentrated sulfuric
acid (1.3 L) in an open container (28 L). After stirring for
30 min, sodium nitrite (0.8 kg) dissolved in water was added
to the container, which was cooled to -2 to +2 °C by
addition of ice. The procedure was performed in a large-
scale hood.
Introduction
During the last two decades, Chemical Production at
Lundbeck have produced 2-chloro-5-trifluoromethyl-ben-
zonitrile (I) (Scheme 1). The product is a key intermediate
The four batches were filtered on a steel nutsche filter,
and the combined filtrates were added to sodium carbonate
(12 kg), copper cyanide (3.8 kg), and sodium cyanide (3.2
kg) suspended in water (30 L) in a 250-L reactor.
After stirring for 2 h, the heptane was added to the
mixture, and the phases were separated. A small volume of
interphase emulsion was filtered on a steel nutsche filter and
the recovered aqueous phase returned to the heptane phase.
This procedure was repeated five times (giving a total of
10 diazonium batches), and the combined heptane phases
were washed with hot water.
Scheme 1
in the production of the anti-psychotic drugs Fluanxol and
Fluanxol Depot (III and II) (Scheme 2).
After evaporation to dryness the product (I) was distilled
under reduced pressure.
Scheme 2
It took about 5 days to produce one batch to give a yield
of 25 kg/week.
Process Scale-Up
A number of issues had to be addressed to scale-up the
original process. The main focus was on optimization of the
unit operations to ensure a safe overall process.
Recently it was decided to reevaluate the process to
develop a procedure suitable for scale-up and moreover to
improve the personal safety for the operators performing the
reactions.
In connection with the scale-up, a process safety evalu-
ation was performed.
2-Chloro-5-trifluoromethyl-benzonitrile is produced by a
Sandmeyer reaction of 2-chloro-5-trifluoromethylaniline (V)
with sodium nitrite in aqueous sulfuric acid. The formed
diazonium salt (IV) is then reacted with a mixture of copper-
(I) cyanide and sodium cyanide in the presence of sodium
carbonate to afford the desired product (Scheme 3).
One of the main scale-up problems with the original
process was the filtration of the potentially explosive
diazonium salt. The filtration was performed due to problems
with an interphase emulsion, but we found that addition of
toluene could break the emulsion down. We found that the
best procedure was to add toluene prior to the addition of
sodium nitrite to practically eliminate the foam formation
and thereby avoid its reaching the overhead condenser. Other
issues were problems with the formation of NOx gases and
hot spots during the mixing of the sodium nitrite solution
with V. This problem was solved by addition of the nitrite
solution through a dip pipe.
After the diazotation process, the phases were allowed to
separate, the toluene phase was discharged, and the water
phase (containing the diazonium ions) was further processed.
* Author for correspondence. E-mail: thpi@lundbeck.com.
† Process Development, H. Lundbeck A/S.
‡ Chemical Implementation & Production, H. Lundbeck A/S.
10.1021/op0498823 CCC: $27.50 © 2004 American Chemical Society
Published on Web 10/26/2004
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
1059

Table 1. Some general stability features regarding aromatic diazonium ions
destabilising effects of aromatic diazonium ions
stabilising effects on aromatic diazonium ions
(a) oxidizing substituents on the aromatic moiety5
(d) sulphate and chloride anions are reported as having stabilizing effects
on the diazonium ion1
(b) solid aromatic diazonium ions
are destabilized by oxidizing anions5
(e) the explosive character is diminished by higher molecular weight and
large and complex anions (with no oxidizing properties)5
(c) electron-withdrawing substituents and potential leaving
groups favor elimination leading to a highly reactive benzyne
intermediate and possible isomerisation by addition6
(f) enhanced stability against nucleophilic attack (i.e. phenol formation
in aqueous solution) and electrophilic substitution by
electron-withdrawing substituents on the aromatic moiety5
This easily enabled scaling-up the diazotation step by a
factor of 20, allowing further scale-up if required.
runaway. Some general features that influence the stability
of aromatic diazonium ions are listed in Table 1:
The overall reaction concentration in connection with the
diazotation was reduced from 1:13 to 1:6 by using external
cooling instead of ice as used in the original procedure.
The aqueous phase was then transferred to a slurry of
sodium carbonate, copper cyanide, and sodium cyanide at
80 °C, and after reacting for an additional 2 h the product
was extracted with heptane. The subsequent heptane and
water phases were separated, and the water phase was
removed. A small intermediate heptane/water emulsion was
filtered on a steel nutsche filter and the recovered aqueous
phase added back to the heptane phase.
The heptane phase was evaporated to dryness and the
residue distilled under reduced pressure to give the product.
A further advance of the new revised procedure is that it
is now possible to perform the process in conventional
production vessels. This is a major improvement in relation
to the working environment for the operators performing the
process due to the fact that the procedure is now is conducted
in closed vessels and not in open containers.
Comparing to the actual case in which 2-chloro-5-
trifluoromethyl-benzenediazonium ion (IV) is used, we see
that the negative counterion is well chosen since the sulphate
group is relatively large and non-oxidizing (cf. d and e in
Table 1). With both chlorine and the trifluoromethyl group
present, there is a possibility of elimination of hydrogen
chloride leading to a highly reactive benzyne intermediate
followed by a possible addition and isomerization7 (cf. c in
Table 1). This reaction happens in basic environment and is
therefore not expected to happen to any greater extent as
the diazonium ion is dissolved in aqueous sulfuric acid. By
addition to the Sandmeyer solution, which is basic (pH ≈
11), the benzyne formation can occur, but as it then competes
with the Sandmeyer reaction, it is not expected to happen in
large proportions.
The lack of electron-donating substituents and the pres-
ence of trifluoromethyl group diminishes the affinity towards
nucleophilic attack, i.e., the hydrolysis in aqueous solutions
yielding phenols and molecular nitrogen (cf. f). This should
limit formation of side products.
Safety Investigations
The above structural analysis of IV indicates that the
compound is probably not too unstable for scale-up and
changed handling. A structural analysis is of course not made
to replace physical testing of the materials but merely to
achieve some knowledge of what to expect in the physical
testing. To ensure that this process was also safe for scale-
up, in reality a thorough experimental testing of the synthesis
and the relevant compounds was needed.
Published warnings of the use of diazonium ions are stated
on the basis of experience from previous accidents. The
incident reports found1-3 are based on the conclusions that
a sensitive diazonium salt precipitates in valves or in a reactor
and is hence detonated by friction during cleaning/disposal
of the equipment. In a single case a decomposition of
dissolved diazonium ion is reported.4
The diazonium ion is known to exhibit unstable charac-
teristics that can cause some salts to be explosive. The
reactivity of each individual diazonium compound should
be investigated to determine their explosive potential.
The main reason for the instability of the diazonium
compound lies in the properties of molecular nitrogen, which
is an extremely good leaving group. The reactivity of the
diazonium compounds is therefore very high, which in many
cases leads to impurities forming side reactions during
synthesis/handling. More problematic, however, is the pro-
pensity for diazonium compounds to undergo rapid decom-
position with low activation energy leading to thermal
Thermal Screenings
Equipment: Mettler-Toledo DSC 821e using standard
aluminum crucibles (40 µL) and high-pressure gold-plated
crucibles (50 µL). Setaram C80 II calorimeter using a high-
pressure Hastelloy C276 vessel with pressure transducer.
Results
More than 20 thermal screenings of the diazonium ion
were conducted at varying concentrations and ramp rates.
Both open and sealed crucibles were used during the DSC
(5) Schank, K., Patai, S., Eds. In The Chemistry of Diazonium and Diazo
Groups; Part 2: Preparation of Diazonium Groups; Wiley and Sons: New
York, 1978; pp 646-647.
(6) Wulfman, D. S., Patai S., Eds. In The Chemistry of Diazonium and Diazo
Groups; Part 2: Synthetic Applications of Diazonium Ions; Wiley and
Sons: New York, 1978; pp 251.
(1) Ullrich, R.; Grewer, Th. Thermochim. Acta 1993, 225, 201-211.
(2) Urben, P. G. Bretherick’s Handbook of ReactiVe Chemical Hazards, 6th
ed.; Butterworth Heinemann: Oxford, 1999; Vol. 2, pp 96-97.
(3) Anon.; Sichere Chemiearbeit 1993, 45, 8.
(4) 6th International Symposium, Loss PreVention and Safety Promotion in
the Process Industries, Oslo, Norway, June 19-22, 1989; Vol. 2, pp 37.l1-
37.l5.
(7) Carey, F. A.; Sundberg R. J. AdVanced Organic Chemistry, 3rd ed.; Plenum
Press: New York, 1990; Part B, p 599.
1060
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development

Table 2. Values of thermal screenings by DSCa
sample
heating rate (K/min)
onset (°C)
peak (°C)
energy (J/g)
>10 samples of a 5-9% diazonium ion soln
5-10
>64
ca. 110
ca. 170
>72
ca. 120
ca. 175
<119
<334
<357
DSC of ca. 9% diazonium ion soln: oxidized in air 24 h
5
55.1
159.1
225.5
66.1
163.3
234.3
105
195
45
a First entry reveals the minimum onset and peak values and maximum energy release values measured in more than 10 samples. Every sample was run in both
standard aluminum crucibles with pierced lid and sealed high-pressure gold-plated crucibles. Second entry shows the data from the experiment showing the lowest
onset. This experiment was run in a standard aluminum crucible with pierced lid.
Table 3. Values of thermal screenings by Setaram C80a
sample
ramp and heating rate onset (°C) peak (°C) onset, gas evolution (°C) energy (J/g)
9% diazonium ion soln
30-100 °C
0.03 °C/min
50 °C
22 h
45 °C
>63.4
N.D.
>79.3
N.D.
>63
N.D.
N.D.
<179
N.D.
N.D.
5% diazonium ion soln: isothermal study
5% diazonium ion soln: isothermal study
N.D.
N.D.
22 h
a First entry reveals the minimum onset and peak values and maximum energy release values measured in five experiments. An Antoine plot was made from the
pressure data obtained during the experiments to find the onset of gas evolution.
experiments to establish whether a dramatic lowering of the
onset was found in either case. No general trend was found,
however, apart from the fact that the decomposition of the
diazonium ion was very unpredictable in the sense that only
in a minority of the DSC-runs was the first onset reported
to be below 80 °C.
As seen in Table 2, the decomposition of the diazonium
ion is liable to form three distinct peaks in solution. Attempts
to crystallize the diazonium salt have thus far been unsuc-
cessful, and all runs have therefore been performed on the
reaction mixture with a diazonium ion concentration of
5-9%. Overall the lowest onset is seen at 55.1 °C in a
sample that was oxidized in air for 24 h. Attempts to
reproduce such low onset by oxidizing other samples were
not possible. The overall energy emitted was in the range of
∼350-800 J/g.
samples representative to the process conditions was 63 °C.
Applying the usual safety margin of 50 °C to the onset of
large-scale thermal screenings in the absence of adiabatic
test results, the temperature limit for this process reaches 13
°C. However, on the basis of the 20-year process experience
with this reaction and handling of the diazonium solution, it
has been decided to set the temperature limit of the
diazonium ion solution at 15 °C, which is found reasonable.
Reaction Calorimetry
To ensure that a scale-up would not result in an
uncontrolled temperature rise during the synthesis, which
would make it impossible to keep the temperature below the
limits, a series of reaction calorimetric tests was performed
in a Mettler-Toledo RC-1 system. This was made by
performing six independent experiments with three measure-
ments of the diazonium ion synthesis and three measurements
of the Sandmeyer reaction. In each of the experiments,
measurements on the diazonium salt formation and subse-
quent Sandmeyer reaction were made using the following
procedures.
Equipment. Mettler-Toledo RC-1 reaction calorimeter
equipped with a 1-L conical shaped reactor (SV01) with a
metal/PTFE cover (MT01) and a custom-made (Fauske),
glass-lined paddle stirrer.
Diazotation. After forming the sulphate salt of V, the
resulting slurry was cooled to 2 °C. To this solution a 44.4%
sodium nitrite solution held at ca. 10 °C was added. The
resulting solution was divided into three equally sized
portions and then added to either heptane (first two portions)
or toluene (last portion) and then charged to the SV01-vessel
prepared for the following Sandmeyer reaction.
Since DSC analysis is known to give results with a relative
high uncertainty regarding the decomposition onset due to
the small sample size and high heating rates, it was important
to carry out screening experiments in a larger sample size.
This was done in a Setaram C80II equipped with a high-
pressure vessel. As seen in Table 3, however, a drop in the
decomposition onset was not observed in comparison to the
DSC-datasdespite the larger sample size (approximately 1
g) and a very low heating rate. During the isothermal study
at 50 °C referred to in Table 3, the sample had clearly
undergone some degree of decomposition according to the
visual inspection after the test (charring of sample), but no
thermal effects were measured. At 45 °C, the solution
remained yellow as characteristic of the ion.
Concluding the thermal screenings, no thermal effects
were monitored by the decomposition happening during the
isothermal tests performed at 50 °C in the Setaram C80
(decomposition visually detected on sample after run). The
lowest onset measured in C80- and DSC-screenings of
The Sandmeyer Reaction. The Sandmeyer solution
(NaCN, CuCN, and sodium carbonate dissolved in water)
was heated to 65 or 80 °C (portions 1 and 2 and portion 3,
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
1061

Figure 1. Reaction calorimetric profile of one of the diazonium ion formations performed in the RC-1. 3rd axis refers to both the
heat flow effect and the temperature of the reaction mixture.
Table 4. Results of reaction calorimetric investigations
heat capacity of
reaction mass
maximum heat flow
(W/L)
heat released
(kJ/mol)
∆TAD
(K)
reaction
(J/kg‚K)
diazonium ion formation
(three expts)
Sandmeyer reaction
(three expts)
3.12-3.39
3.62-3.99
315
(short-lived)
135
44-66
7-9
(of V)
200-280
(of diazonium ion)
9-16
respectively), and to this mixture the diazonium solution was
added slowly. After the addition of the diazonium solution,
heptane/toluene (portions 1 and 2 and portion 3, respectively)
was added, and the mixture in solutions 1 and 2 was then
left stirring at 65/80 °C for 2 h. During the course of the
experiments, the proposed process changed, and this is the
reason for the changes in the procedure.8
small quantity is dosed (33 g in total), making the mixing
less continuous; additionally, a short pump failure is seen
during the second addition where the heat flow drops
markedly. Despite the irregularities, it is seen in Figure 1
that no accumulation happens to any large extent. Table 4
shows the values measured in both the diazonium ion
formation and the Sandmeyer reaction, and it is seen that
the diazonium ion formation in general emits a fairly low
amount of energy in all three experiments. The adiabatic
temperature rise in the three solutions is calculated to be
7-9 K based on the relation (eq 2):
Results and Discussion of Reaction Calorimetric Data9
As seen in Figure 1, the addition of sodium nitrite solution
was divided into three portions of equal size to get the best
possible idea of any possible accumulation of reactants.
Irregularities in the heat flow are due to the fact that only a
∆Tadiabatic ) ∆H/mCp
(2)
(8) No solvent was added in experiment 3: crude I was tapped from the bottom
valve after stirring at 65 °C for 2 h.
(9) Reaction Calorimetric Theory: The energy balance of the investigations is
made on the basis of the following eq 1:
The peak emission is relatively high with an emission of
about 315 W/L. The emission by peak rate is short-lived,
though, and the total energy emitted is therefore low.
The calorimetric profile of the Sandmeyer reaction is
shown in Figure 2. The addition of the diazonium ion
solution to the Sandmeyer mixture was divided into three
portions of unequal size, and evidently there is no accumula-
tion seen in either of the dosings. By a relatively high dosing
rate, the emission of heat is averaged to about 100 W/L with
a peak emission of about 135 W/L. Corrections are made to
Qr ) Qflow + Qaccum + Qdose
(1)
where: Qr ) heat of reaction expressed in watts (J/s). Qflow ) the heat
flow through the jacket of the vessel which equals UïA(Tr - Ta). Tr is the
reaction temperature, Ta is the average jacket temperature, and Uï is the
overall heat transfer coefficient, and A is the heat transfer area of the virtual
volume of the reaction mass. Qaccum ) the combined heat flow accumulation
of the reaction mass and the inserts. This equals (mrCpr + miCpi) dTr/dt
where mCp is referring to the mass and heat capacity of the reaction mass
(mrCpr) and of the inserts (miCpi).
1062
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development

Figure 2. Reaction calorimetric profile of one of the Sandmeyer reactions measured in the RC-1.
compensate for the cooling effect coming from the addition
of a cold liquid to a warm, but the overall heat flow results
in a decrease in the reaction temperature during addition.
The combined energy emitted during the Sandmeyer reaction
is moderately high (200-280 kJ/mol diazonium ion), but
due to the dilution and the high heat capacity of the solution
the ∆Tadiabatic is only 9-16 °C (Table 4).
Provided that precautions are taken against the low onset
of the decomposition of the diazonium ion, the data presented
does not reveal any serious threat seen from a process safety
viewpoint.
Given the unpredictable nature of the diazonium func-
tionality, conservative standards have been set as a frame of
the large-scale production of I. In this context it should be
emphasized that the process temperature must be kept at ca.
2 °C, and that the temperature of any bulk diazonium ion
solution never exceeds 15 °C.
of the valve. Also, the number of valves is minimized to
further reduce this problem.
After each batch, the valves are inspected to ensure that
solids are not deposited in the death lock of the valves.
Conclusions
A scalable Sandmeyer process has been developed. The
process is conducted in closed but not pressurized vessels
to ensure a better working environment for the operators
performing the reaction.
The process has been safety evaluated and proven safe
for scale-up.
Due to a much shorter batch time (0.75 days) this new
improved procedure is capable of producing 70-90 kg/week.
This procedure is scalable, and we plan to scale up this
procedure further by a factor of 10 to about 800 kg/week
Experimental Section
Process Implementation
General. All reagents and solvents were obtained from
commercial suppliers and used without further purification.
The GC analysis data is reported in area %, not adjusted
to weight %.
2-Chloro-5-trifluoromethyl-benzonitrile (1). Water (80
L) and concentrated sulfuric acid (29 kg, 245 mol) are added
to a 250-L apparatus. The solution is cooled to <14 °C and
2-chloro-5-trifluoromethylaniline (24 kg, 123 mol) is added,
keeping the temperature below 15 °C. After stirring for
approximately 30 min, toluene (50 L) is added. After cooling
the mixture to 3 °C, a solution of sodium nitrite (9.6 kg,
139 mol) in water (20 L) is added, keeping the temperature
below 3 °C.
To implement the modified process, we had to take into
consideration the potential danger in handling the diazonium
salt solution. Due to the hazardous nature of diazonium
compounds, the process equipment should be designed and
constructed in such a manner that the risk of depositing small
amounts of diazonium compoundswhich may later dry up
and thereby become a safety hazardsis reduced to a
minimum.
After the formation of the diazonium salt, the solution is
transferred to a second reactor containing the slurry of
cyanides. To avoid any mechanical stress to the diazonium
salt and the risk of having diazonium salt deposits in the
pump chambers, the transfer of the diazonium salt solution
is performed using vacuum.
The reaction mixture is stirred for 45 min, the phases are
separated, and the toluene phase is discharged.
The water phase is added to a hot (80 °C) slurry of sodium
carbonate (36 kg, 340 mol), copper(I) cyanide (11.3 kg, 402
mol), and sodium cyanide (9.6 kg, 196 mol) in water (87 L)
All the valves in contact with the diazonium salt solution
are diaphragm valves, whichsdue to their designseliminate/
reduce the risk of depositing diazonium salt in the death lock
Vol. 8, No. 6, 2004 / Organic Process Research & Development
•
1063

in a 800-L apparatus. The temperature during the addition
is almost constant.
2-Chloro-5-trifluoromethylaniline 16.1 min, 2-chloro-5-tri-
fluoromethyl-benzonitrile 9.6 min.
After stirring for 2 h (at 65 °C), heptane (175 L) is added,
and the phases are separated. A small volume of interphase
emulsion is filtered on a steel nutsche filter, and the recovered
aqueous phase is returned to the heptane phase.
The heptane phase is evaporated to dryness and the
residue (I) is distilled under reduced pressure. Yield: 12.5
kg (50%); bp 94-98 °C (1.3-2.6 Pa). Purity 94.4% by GC
(HP Innowax, capillary GC column, 30 m/0.32 mm), nitrogen
gas operating at a flow rate of 2 mL/min; inlet temperature
250 °C; column temperature ramped from 100 to 160 °C
during 15 min, maintained for 12 min. Retention times:
Acknowledgment
We thank Pernille Kofoed for the laboratory work
performed prior to the implementation and Per Thomsen for
organizing the implementation in the plant. Also, we thank
Jimi Byrsø Dan and Stefan Uwe Hagen for analytical support.
Finally, we thank H. Lundbeck A/S for allowing us to publish
this work.
Received for review June 15, 2004.
OP0498823
1064
•
Vol. 8, No. 6, 2004 / Organic Process Research & Development