Optimization and scale-up of a novel process for 2-aminoindan hydrochloride production

Article abstract of DOI:10.1021/op000222h

The need for an economical process for producing 2-aminoindan hydrochloride, a key starting material in manufacturing novel bioactive molecules, motivated development of a novel synthetic route using an inexpensive reactant, ninhydrin. The synthesis, involving oximation of ninhydrin followed by catalytic reduction of the resulting oxime intermediate to give 2-aminoindan, was demonstrated successfully, and a product purification scheme was developed to isolate 2-aminoindan as the hydrochloride salt form. Subsequent process development optimized the reduction step by identifying regimes of fast and slow reaction (corresponding to reduction of oxime and diketone functions, respectively), and tailoring reaction conditions to use mild conditions during the fast exothermic regime to ensure process safety followed by more severe conditions for the slower reaction. The process was successfully scaled up 100-fold in a pilot plant, with excellent yield and product quality agreement between laboratory and pilot plant.

Full text of DOI:10.1021/op000222h

Organic Process Research & Development 2001, 5, 167−175
Optimization and Scale-Up of a Novel Process for 2-Aminoindan Hydrochloride
Production
Didier Roche, David Sans, Michael J. Girgis,* Kapa Prasad,* Oljan Repic, and Thomas J. Blacklock
Process R & D, Chemical and Analytical DeVelopment, NoVartis Institute for Biomedical Research,
59 Route 10, East HanoVer, New Jersey 07936, U.S.A.
Scheme 1. Literature preparations of 2-aminoindan HCl
salt using either 1- or 2-indanone
Abstract:
The need for an economical process for producing 2-aminoin-
dan hydrochloride, a key starting material in manufacturing
novel bioactive molecules, motivated development of a novel
synthetic route using an inexpensive reactant, ninhydrin. The
synthesis, involving oximation of ninhydrin followed by catalytic
reduction of the resulting oxime intermediate to give 2-ami-
noindan, was demonstrated successfully, and a product puri-
fication scheme was developed to isolate 2-aminoindan as the
hydrochloride salt form. Subsequent process development
optimized the reduction step by identifying regimes of fast and
slow reaction (corresponding to reduction of oxime and diketone
functions, respectively), and tailoring reaction conditions to use
mild conditions during the fast exothermic regime to ensure
process safety followed by more severe conditions for the slower
reaction. The process was successfully scaled up 100-fold in a
pilot plant, with excellent yield and product quality agreement
between laboratory and pilot plant.
Scheme 2. Proposed 2-aminoindan hydrochloride
preparation from ninhydrin
of preparing 2-aminoindan hydrochloride from ninhydrin, (b)
develop and optimize a process for scale-up in a pilot plant
and commercial facility, and (c) demonstrate the process on
a large scale and thus prove the synthetic utility of Scheme
2 for bulk-scale 2-aminoindan hydrochloride production.
Introduction
The difficulty in rapidly obtaining bulk amounts of
2-aminoindan hydrochloride, a key starting material in
producing bioactive molecules, prompted an in-house evalu-
ation and development of synthetic routes for its economical
large-scale production. Literature preparations of the 2-ami-
noindan involve oximation of either 1-indanone1 or 2-in-
danone,2 followed by reduction of the oxime and ketone
functions (Scheme 1). However, the relatively high costs of
both 1-indanone and 2-indanone preclude their large-scale
use as starting materials.
The structural similarity of ninhydrin to 1- or 2-indanone,
its availability in bulk amounts (owing to its use as an
analytical reagent for detecting primary amines3), and its low
cost relative to that of 1-indanone and 2-indanone prompted
evaluation of an analogous oximation/reduction sequence
(Scheme 2) using ninhydrin. Preparation of the oxime by
reaction of ninhydrin with hydroxylamine hydrochloride in
ethanol has been reported previously.4 The objectives of this
work were thus to (a) demonstrate the synthetic feasibility
Experimental Section
Apparatus and Materials. Process development experi-
ments were performed using a Mettler-Toledo RC1 reaction
calorimeter equipped with a 1-L jacketed glass vessel rated
for 10-bar operation, a Hastelloy C pitched-blade agitator, a
hydrogen gas reservoir, a pressure regulator for setting the
reactor pressure, and valves for venting and purging (Figure
1). Calorimeter operating principles are discussed elsewhere.5
Sources and purities of materials used are given in Table
1. All materials were used as received. Although we used
hydroxylamine in the salt form, the free base could undergo
a runaway reaction leading to an explosion.6 Two dry 10%
Pd/C catalysts and two wet 10% Pd/C catalysts were
investigated.
Reaction Monitoring and Product Analysis. For reac-
tion monitoring, samples were analyzed by HPLC using a
Hewlett-Packard Series 1000 HPLC equipped with a UV
detector. Method details are given in Table 2.
In process development experiments, reaction calorimetry
was also used to monitor reaction progress. During the
(1) Paul, H. J. Prakt. Chem. 1965, 28, 297; Nichols, D. E. J. Med. Chem. 1974,
17, 161-165; Cannon, J. G. J. Med. Chem. 1982, 25, 1442-1446; Johnson,
M. P. J. Med. Chem. 1991, 34, 1662-1668.
(2) Levin, J. Org. Chem. 1944, 380-386; Rosen, J. Org. Chem. 1963, 28,
2797-2801; Paleo, M. Tetrahedron Lett. 1994, 11, 3627-3638.
(3) The Merck Index, 12th ed.; Budavari, S., Ed.; Merck & Co.: Whitehouse
Station, NJ, 1996; p. 1126.
(5) Girgis, M. J.; Kiss, K.; Ziltener, C. A.; Prashad, M.; Har, D.; Yoskowitz,
R. S.; Basso, B.; Repic, O.; Blacklock, T. J.; Landau, R. N. Org. Process
Res. DeV. 1997, 1, 339-349.
(6) Herman, S.; Zhenglong, X.; Sundareswaran, p. C.; Wang, Z. 218th ACS
National Meeting, Book of Abstracts; American Chemical Society: Wash-
ington, DC, Aug. 22-26, 1999.
(4) Paria, P. K., et al. J. Indian Chem. Soc. 1990, 67, 532.
10.1021/op000222h CCC: $20.00 © 2001 American Chemical Society and The Royal Society of Chemistry
Published on Web 02/02/2001
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
167

Figure 1. Schematic diagram of RC1 calorimeter equipped for hydrogenation reactions.
Table 1. Sources and purities of materials
material
molecular formula/CAS no.
source
purity %
ninhydrin
C9H6O4, [485-47-2]
Aldrich
Aldrich
Aldrich
Aldrich
Aldrich
Precious Metals Corp.
Precious Metals Corp.
Mallinckrodt
97
99
99
hydroxylamine hydrochloride
hydroxylamine sulfate
(A) dry 10% Pd/C
H2NOH‚HCl, [5470-11-1]
(H2NOH)2‚H2SO4, [10039-54-0]
Pd [7440-05-3, 7440-06-4]
Pd [7440-05-3, 7440-06-4]
Pd [7440-05-3, 7440-06-4]
Pd [7440-05-3, 7440-06-4]
H2SO4 [7664-93-9]
-
(B) dry 10% Pd/C
-
(C) 60% wet 10% Pd/C
(D) 30% wet 10% Pd/C
concentrated sulfuric acid
acetic acid glacial
-
-
96.1
99.7
99.999
CH3CO2H [64-19-7]
H2 [1333-74-0]
Mallinckrodt
AGL
hydrogen
Table 2. HPLC method for reaction monitoring
Waters Symmetry C-18, 15 cm,
mixture because of a decomposition process starting at 84
°C. Cool to 25 °C and add a slurry of 2.64 g of dry 10% Pd
on activated carbon premixed with 77.40 g of glacial acetic
acid (caution: Dry Pd/C is flammable). Close the reactor and
pressurize to 40 psig with N2 and then depressurize. Repeat
the cycle of 40 psig pressurization/depressurization with N2
twice. Turn the stirrer off and then pressurize with hydrogen
to 40 psig and depressurize. Repeat the cycle of 40 psig
pressurization/depressurization with H2 twice. Pressurize with
hydrogen to 50 psig and test for leaks. If no leaks are found,
depressurize to 20 psig. With the stirrer still off, open the
hydrogen supply valve to the reactor. With the pressure on
the reactor at 20 psig, turn the stirrer on (800 rpm). A
maximum heat release rate of 35 W/kg is measured during
the first 45 min. After 1 h at 25 °C and 20 psig, increase the
pressure to 40 psig. Wait for 15 min and increase the batch
temperature from 25 to 35 °C. HPLC analysis during the
first 5 h reveals that the oxime intermediate is totally
consumed after the first 3 h. The reaction is monitored by
calorimetry and H2 consumption. The heat flow attains a
near-zero value, and H2 consumption stops after about 8 h.
After 10 h, close the hydrogen supply valve, stop agitation
and vent to the atmosphere. Purge three times with nitrogen
and open the reactor. Remove the slurry from the reactor.
Charge the 9-cm Bu¨chner filter with 20 g of Celite. Filter
the reaction mixture. Wash the cake with 70 g of acetic acid
(do not let the cake dry). Evaporate 550 mL of acetic acid
column
4.6 µm particles
mobile phase
water (with 0.1%
trifluoroacetic acid)
and acetonitrile, with
water/acetonitrile ratio
initially 95:5
changing to 10:90 over 15 min
mobile phase flow rate, mL/min 1.5
wavelength used, nm
column temperature, °C
250
30
catalytic reduction, hydrogen consumption was determined
by measuring the H2 gas pressure inside the hydrogen
reservoir, which had a known volume (3.785 cm3) and was
kept at room temperature (typically 25 °C).
Detailed Procedure. Charge the 1-L RC1 MP10 reactor
with 33.00 g (0.185 mol) of ninhydrin and 554.70 g of glacial
acetic acid. The reaction mixture is blue. Seal the reactor
under an N2 atmosphere. Add at 25 °C 54.42 g of sulfuric
acid1 with a dosing rate of 1 mL/min and 31.63 g of
hydroxylamine sulfate. Keep the reactor under an N2
atmosphere. Mix the reactor content using a stirring speed
of 800 rpm. Heat the reaction mixture to 55 °C. Hold at 55
°C for 30 min. The process is endothermic. The reaction
mixture turns from blue to yellow. The intermediate oxime
precipitates. It is important not to overheat the reaction
168
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development

Table 3. Reaction scheme and conditions used to demonstrate synthesis feasibility
under vacuum (20 mbar, 50-55 °C). Add 250 g of xylene
and evaporate 300 mL of the xylene/acetic acid azeotropic
mixture under vacuum (20 mbar, 50-55 °C). Add 170 g of
xylene. Add at 25 °C under stirring 367 g of a 20% NaOH
solution (exothermic, maintain the temperature below 55 °C).
Measure the pH of the aqueous layer (pH paper). The pH
must be above 11. If not, continue the basification with 20%
NaOH until this pH value is reached. Separate the bottom
aqueous layer for proper disposal. Filter the top organic layer.
Add to the filtrate 51 g of a 4 N HCl solution in 1-pentanol
(exothermic, maintain the temperature below 30 °C). Cool
the resulting suspension to 0 °C and maintain this temperature
for 1 h. Filter off the solids through a Bu¨chner funnel
equipped with a polypropylene pad and wash the cake once
with 100 g of heptane. Dry the filter cake at 45-50 °C (100
( 5 mbar) for 16 h to afford 19.14 g of 2-aminoindane
hydrochloride (identical in all respects with the authentic
sample) as a white solid, mp 241-242 °C (decomposition).
Yield: 66%.
Results and Discussion
Process Development: Ninhydrin to 2-Aminoindan
Hydrochloride. Demonstration of Synthetic Route. At the
outset the aim was to produce 2-aminoindan hydrochloride
via a one-pot synthesis, involving oximation of ninhydrin
with hydroxylamine hydrochloride in glacial acetic solvent
followed by reduction of the oxime and ketone functions.
An initial attempt at conducting the reduction concurrently
with the oximation by mixing ninhydrin and hydroxylamine
hydrochloride in acetic acid, adding Pd/C catalyst and heating
to 55 °C for 10 h under 50 psig of H2 did not yield the desired
product (Table 3, entry 1).
In subsequent experiments, the oximation was performed
before the reduction step, with the oximation step carried
out by heating ninhydrin and hydroxylamine hydrochloride
at 55 °C in glacial acetic acid for 30 min, followed by cooling
to room temperature. During this process the 2-oximino
intermediate precipitated from the solution.
Using 1 equiv of hydroxylamine hydrochloride followed
by reduction in the presence of 1.6 equiv of sulfuric acid as
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
169

Scheme 3. Proposed reaction pathways for reduction of ninhydrin oxime to 2-aminoindan
an additive (Table 3, entry 2) gave 2-aminoindan hydro-
chloride in 16% yield along with oxime diketone intermediate
and 2-aminoindanol. A dramatic improvement was observed
when 3 equiv of sulfuric acid were used (Table 3, entry 3),
with the desired product formed in 52% yield after 6 h along
with some 1-indanone byproduct. These conditions were
found to be reproducible (Table 3, entry 4) and were used
in subsequent process development.
4, entry 3). This approach was not so successful because
azeotropic removal of acetic acid with toluene is not efficient
(i.e., the toluene/acetic acid azeotrope contains 28% acetic
acid), giving 2-aminoindan hydrochloride in only 44% yield.
Using xylene instead of toluene gave a more favorable
azeotrope (71% acetic acid), and thus more thorough removal
of acetic acid, resulting in a clean separation between the
aqueous and organic layers upon basification. After separa-
tion of the layers, 2-aminoindan hydrochloride was precipi-
tated using a 4 N HCl in ethanol solution (Table 4, entry 4).
This solution led to a layering and agglomeration of particles,
rendering the filtration cumbersome. The use of HCl solution
in 1-pentanol removed this problem (Table 4, entry 5).
During the basification step, 20% NaOH appeared to be the
optimum for saturating the aqueous layer. With these
optimized conditions, 2-aminoindan hydrochloride was iso-
lated in 66% yield (99% purity PAN, 96.69% ES).
Thus, a novel one-pot synthesis of 2-aminoindan hydro-
chloride from ninhydrin was demonstrated. This success
motivated further process optimization with emphasis on
investigating the impacts of H2 pressure, temperature, amount
of sulfuric acid, and Pd catalyst type during the reduction.
Process Optimization: Oxime Diketone Reduction.
Selection of Agitation Rate in Laboratory Experiments. To
obtain similar conversion rates between laboratory and pilot
plant reactors, gas-liquid reactions must be performed in
the same gas-liquid mass transport regime on both scales,
especially if gas-liquid mass transfer limits the conversion
rate. Gas-liquid mass transfer in the pilot plant vessel in
which scale-up was planned was quantified by measurement
of the volumetric mass transfer coefficient kLa at a typical
agitation rate (150 rpm) as described elsewhere8 and
determined to be about 0.1 s-1. The agitation rate in the RC1
vessel (800 rpm) was then selected to give a 0.1 s-1 kLa value.
The same agitation rate was also employed for ninhydrin
oximation, as it was sufficient to keep the precipitated oxime
diketone intermediate suspended.
The probable reduction reaction pathway (Scheme 3) was
inferred on the basis of all intermediates identified by LC/
MS in several experiments. Although intermediates in
brackets were not detected, their presence is presumed, as
they are necessary for understanding the formation of
2-aminoindan from ninhydrin oxime. The requirement for
excess sulfuric acid is understandable if the two dehydration
steps (MW 165.19 f 147.18 and 149.19 f 131.18), are
assumed to be rate-determining. Hydrogenation without
intermediate olefin formation may also occur from the
intermediacy of aziridine (neighbouring group participation).
As basic nitrogen compounds can inhibit reactions oc-
curring on metal catalysts by adsorbing strongly and blocking
catalytic sites,7 greater amounts of sulfuric acid equivalents
could improve reaction performance by protonating the
nitrogen compounds to decrease their extent of adsorption.
Product Isolation Optimization. As sulfuric acid is already
present in the reaction mixture, the sulfuric acid salt of
2-aminoindan (Table 4, entry 1) was first isolated. Although
isolation of the sulfuric acid salt gave good yield, it could
not be used in the subsequent synthetic step. Attempts to
isolate the free base led to degradations (Table 4, entry 2).
Subsequent product purification attempts focused on
isolating the HCl salt. The main difficulty encountered was
the formation of a strong emulsion upon basification with
sodium hydroxide. To avoid this problem, acetic acid was
first removed by azeotropic distillation with toluene (Table
(7) Emmett, P. H. Catalysis; Reinhold Publishing Corporation: New York,
1954; Vol. I.
170
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development

Table 4. Product workup optimization
react.
procedurea
comments
yield (%)
67
purity (%)
99
1
-azeotropic concentration
(1-pentanol) until precipitation
-sulfuric acid salt
contaminated with
inorganic salts.
-filtration/drying
-poured into ice
-basification (50% NaOH)
(PAN)b
n. d.
2
3
-emulsion due to NaOAc
-rapid degradation of
2-aminoindan as free base
n. d.
44
-extraction (EtOAc)/concentration
-azeotropic concentration (toluene × 3)
-dilution (H2O)/
basification (50% NaOH)
-extraction (toluene)
-precip. (2.5 N HCl/EtOH)
-filtration/drying
-emulsion due to NaOAc
95 (PAN)
87 (PAN)
4
5
-azeotropic concentration (xylene × 2)
-clear separation of aq./org. layers
-large amount of NaOH required
70
66
-dilution (H2O)/
basification (NaOH 10%)
-extraction (xylene)
-layering and formation of
agglomerates during concentration
-darkening of material upon drying
-precip. (4 N HCl/EtOH)
-concentration/filtration/drying
-conc/azeotropic conc (xylene × 1)
-basification (20% NaOH)
-extraction (xylene)
-clear separation of aq./org. layers
-aq. saturated but homogeneous
-clean precipitation of
2-aminoindan hydrochloride with
homogeneous particle size (no layering)
99 (PAN)
95.7 (ESc)
-precip. (4 N HCl/1-pentanol)/
cooling to 0 °C
-filtration/drying
a Reaction mixture in all cases was first filtered over Celite. b PAN signifies “Peak Area Normalization.” c ES signifies “External Standard.”
Table 5. Reaction conditions and product yield in initial process development experiments
2-aminoindan
hydrochloride
% yield
oximation
reduction
reduction
pressure
[psig]
composition
ninhydrin:hydroxylamine:
H2SO4 [mol]
loading
ninhydrin/solvent
[g]
temperature temperature
catalysta type
run
[°C]
[°C]
w/w %
after workup
1
2
3
4
55
55
55
55
80
20
60
20
40
40
40
40
wet type C
100%
wet type D
40%
dry type B
8%
dry type B
8%
1:1 (NH2OH‚HCl):3
33/630
33/630
33/630
33/630
0
1:1 (NH2OH‚HCl):3
53
n.d.
70
1:1.05 [(NH2OH)2‚H2SO4]:3
1:1.05 [(NH2OH)2‚H2SO4]:3
a Catalyst designations are given in Table 1.
Concerns about the scale-up of hydrogenation of oxime
diketone with at least some of the compound in the solid
phase was of concern. This concern was resolved a posteriori
by observing that the initial rate depended strongly on
reaction conditions (e.g., temperature and pressure) and with
the good agreement between laboratory and pilot plant results
on scale-up, indicating that the solids dissolution rate was
sufficiently large so as not to impact the overall conversion
rate.
Pd/C catalysts for the reduction. As a reaction time shorter
than 8 h was also sought, higher reaction temperatures (vs
room temperature) were also evaluated. Conditions for the
initial process development experiments are shown in Table
5. Hydroxylamine hydrochloride was used in only the first
two experiments, being replaced in later experiments with
hydroxylamine sulfate because of corrosion concerns. In all
cases, the mole ratio of ninhydrin:hydroxylamine salt was
1:1; thus, with the hydroxylamine sulfate, the number of
hydroxylamine equivalents was doubled. In all cases, the
oximation was performed by adding the hydroxylamine
hydrochloride or hydrogen sulfate to a mixture of glacial
acetic acid, ninhydrin, and sulfuric acid at 50 °C and
maintaining at this temperature. Although the reaction is
Impact of Catalyst Water Content and Reaction Temper-
ature. Safety concerns about pilot plant use of dry Pd/C
catalyst, which can ignite easily, prompted evaluation of wet
(8) Delmling, A.; Karandikar, B. M.; Shah, Y. T.; Carr, N. L. Chem. Eng. J.
1982, 29, 127-140.
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
171

Table 6. Presumed structures of some detected by-products
Figure 2. Calorimetric and H2 consumption profiles using wet
Pd/C at 80 °C (run 1, Table 5).
previous experiment, which was performed at higher tem-
perature and which gave essentially no aminoindan forma-
tion, the calorimetric profile in this case is characterized by
a high initial heat release rate, followed by a region of lower
heat release involving at least one autocatalytic reaction.
Since wet Pd/C catalysts of different water content used
either near or above room temperature did not give the
desired product in good yield, subsequent experiments
focused on using dry Pd/C catalysts. Even with a dry Pd/C
catalyst at 60 °C, a low yield of 2-aminoindan hydrogen
sulfate was obtained, with 2-aminoindanol being the domi-
nant product (Figure 4, Table 6). The incomplete reduction
was consistent with a H2 consumption below 6 mol H2/mol
ninhydrin. The calorimetric profile was also different than
in the other runs.
Good (70%) 2-aminoindan yield was obtained by using
dry Pd/C at 20 °C (run 4, Table 5, Figure 5). As the previous
experiment which gave at least some 2-aminoindan formation
(e.g., Figure 3) showed that a rapid and exothermic reaction
occurred initially followed by a less exothermic reaction, run
4 was performed at a lower pressure (20 psig) initially with
the pressure increased to 40 psig at 0.75 h, after the initial
exothermic reaction ceased. The lower initial pressure slowed
the initial rapid exothermic reaction to give a heat evolution
rate of 35 W/kg, and thus within the estimated 60 W/kg heat
removal rate of a commercial vessel;9 the subsequent higher
pressure gave a greater reaction rate during the slower part
of the process. Subsequent HPLC and LC/MS analyses
revealed that during the rapid initial reaction, most of the
oxime was converted and that during the slow reaction
regime (1-18 h), most of the ketone reduction occurred, with
autocatalytic behavior occurring at about 5.2 h. At these
conditions, good yields were obtained (70% isolated yield,
>95% yield in the reaction mixture based on external
standard) and with the expected H2 consumption being
slightly higher than the expected value of 6 (Figure 5).
The initial process development thus showed that using
either wet or dry Pd/C catalyst, the reduction could not be
Figure 3. Calorimetric and H2 consumption profiles using wet
Pd/C at 20 °C (run 2, Table 5).
endothermic, causing a heat absorption of +120 kJ/kg, the
short duration of the endothermic process (<1 min) did not
cause temperature control problems upon scale-up.
A reduction temperature of 80 °C with a wet Pd/C catalyst
gave a final hydrogen consumption of 5 mol H2/mol
ninhydrin (Figure 2). As complete reduction of the oxime
diketone requires 6 mol H2/mol ninhydrin (Scheme 2),
incomplete reduction and formation of other products was
inferred. HPLC analysis of the final reaction sample con-
firmed this inference, as it showed no formation of 2-ami-
noindan, with several other products being formed instead.
The calorimetric profile shows a maximum at about 1.1 h,
indicating an autocatalytic reaction.
Reduction at 20 °C using a wet Pd/C catalyst with a lower
water content gave a final hydrogen consumption of about
6 mol H2/mol ninhydrin (Figure 3). The 2-aminoindan
hydrochloride yield after product isolation and purification,
however, was only 53%. The low yield was consistent with
detection of several byproducts in the final reaction mixture
chromatogram. Examination of the proposed structures based
on LC/MS of some of these (Table 6) indicates incomplete
reduction, even though approximately 6 mol of hydrogen
were consumed per mole of ninhydrin, and the reaction was
stopped only after heat flow and hydrogen consumption both
ceased (Figure 3). Evidently, unknown side reactions involv-
ing additional hydrogen consumption occurred. Unlike the
(9) Stoessel, F. Course on Thermal Safety; Summit, NJ, April 1998.
172
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development

Figure 4. Calorimetric and H2 consumption profiles with dry
Pd/C catalyst at 60 °C (run 3, Table 5).
Figure 6. Calorimetric and H2 consumption profiles using dry
Pd/C catalyst at 25 °C with a 20-40 psig pressure ramp (run
5, Table 7).
Figure 5. Calorimetric and H2 consumption profiles using dry
Pd/C catalyst at 20 °C with 20-40 psig pressure ramp (run 4,
Table 5).
Figure 7. Calorimetric and H2 consumption profiles at
optimized conditions. (run 6, Table 7).
performed at higher temperatures (e.g., >60 °C). The best
results were obtained using a dry Pd/C catalyst at 20 °C (i.e.,
near room temperature), although some desired product was
also obtained using a wet Pd/C catalyst at 20 °C. In the two
cases where 2-aminoindan was formed, similar calorimetric
profiles were obtained, suggesting that the shape of calori-
metric profile can give a rapid qualitative indication for at
least some 2-aminoindan formation.
Reduction Optimization. Reaction optimization (Table 7)
focused on shortening the reaction time by modifying the
reaction pressure and temperature programs using the dry
Pd/C catalyst available for scale-up in the pilot plant (Catalyst
A, Table 1).
Run 5 (Figure 6) was designed to repeat the experimental
conditions from run 4 (Figure 5), viz., using a 20 to 40 psig
pressure increase after the initial exothermic reaction was
completed, but at a slightly higher reaction temperature (25
vs 20 °C). The catalyst was evidently less active than that
for run 4, as the rapid reaction terminated at 1.3 h (vs 0.75
before at a 5 °C lower temperature). Nevertheless, 2-ami-
noindan was produced in 91.3% yield based on external
standard at a 20-h reaction time, with good 2-aminoindan
hydrochloride yield (64%) obtained after workup (Table 8).
It was desirable to shorten the reaction time to under 16
h to allow performing the reduction in two pilot plant shifts.
The ketone reduction (i.e., in the slow reaction regime) was
thus performed at a higher temperature (35 °C) in the
subsequent experiment (run 6, Table 7). Thus, a pressure
increase from 20 psig to 40 psig was followed by a
temperature increase over 20 min from 25 to 35 °C. With
this modification, the reaction time was shortened to under
12 h (Figure 7).
Figure 7 shows that by performing the ketone reduction
at 35 °C, the heat evolution rate arising from the autocatalytic
reaction identified earlier increased significantly, with the
maximum heat evolution rate at about 2.5 h being close to
30 kJ/kg. The autocatalytic process can also be detected by
the inflection point (point of maximum slope) in the hydrogen
consumption profile, which coincides with the heat evolution
spike at ∼2.5 h. However, the small width of the profile
due to the autocatalytic reaction results in an adiabatic
temperature rise of 22 °C. This temperature rise does not
pose a significant safety concern if loss of cooling occurred.
A temperature rise due to the autocatalytic reaction would
also not impact product quality significantly on the basis of
process research experimental results, where temperature
excursions as high as 40 °C occurred without significant
product degradation. Final reaction conditions for the reduc-
tion were selected on the basis of run 6 (Figure 7), which
gave an isolated 2-aminoindan hydrochloride yield of 66%.
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
173

Table 7. Reaction conditions for reduction optimization
composition
ninhydrin:
loading
ninhydrin/
solvent
[g]
2-aminoindan
hydrochloride
% yield
oximation
temperature
[°C]
reduction
temperature
[°C]
reduction
pressure
[psig]
catalysta
type
w/w %
hydroxylamine sulfate:
H2SO4
run no.
[mol]
after workup
5
6
7
55
55
55
25
20-40
20-40
20-40
dry type A
8%
dry type A
8%
dry type A
8%
1/1.05/3
1/1.05/3
1/1.05/3
33/630
33/630
33/630
64%
66%
74%
25-35
25-35
a Catalyst designations are given in Table 1.
Table 8. Results of final optimization runs
performing an energy balance on the pilot plant reactor
contents:
% yield of
% yield of
reduction
reaction 2-aminoindan 2-aminoindan‚
run temperature catalysta
time
[h]
before
HCl after
workup
no.
[°C]
w/w %
work-up
5
6
7
25
dry type A
8%
dry type A
8%
dry type A
8%
20
12
8
91.3
93.7
94.1
64
66
74
The rate of energy loss to the surroundings is small and can
be neglected because during the first part of the reaction,
the 25 °C batch temperature is very close, if not equal, to
room temperature.
Thus, the rate of energy accumulation will equal the rate
of energy generation by chemical reaction less the rate of
energy loss to the jacket through the reactor wall:
25-35
25-35
a Catalyst designations are given in Table 1.
dT
mCp r ) Qrxn - UA(Tr - Tj)
dt
where:
m ) mass of reactor contents
Cp ) specific heat of reactor contents
Tr, Tj ) reactor and jacket temperatures, respectively
U ) overall heat transfer coefficient between reactor and
batch
A ) heat transfer area
Figure 8. Reproducibility of hydrogen consumption profiles
at optimized conditions runs 6 and 7, Table 7).
A further simplification can be made if it is assumed that
the rate of energy accumulation during the first part of the
reaction is small (i.e., the reactor is nearly isothermal).
Solving for the jacket temperature then gives:
A subsequent experiment (run 7) was performed to
evaluate reproducibility and to determine if the reaction could
be terminated after 8 instead of 10 h. Good reproducibility
can be seen based on the superimposable hydrogen con-
sumption profiles obtained (Figure 8), including the timing
of the autocatalytic process. Analysis of the reaction mixture
after 8 h in run 7 indicated a nearly identical yield as that
obtained with run 6 after a 12-h reaction time (Table 8).
These results indicate that there was a 2-h window (8 h after
the start of the reduction) for stopping the reduction without
significant yield loss.
Pilot Plant Scale-Up. Specification of Jacket Temperature
Profile. Because of the sluggish nature of the jacket
temperature control system in the pilot plant reactor and the
exothermic reaction occurring at early reaction time (Figure
7), cooling the jacket of the pilot plant reactor was needed
to prevent batch temperature excursions. A recommended
jacket temperature profile was calculated on the basis of the
heat release rate measured with the RC1, Figure 7, by
Qrxn
Tj ) Tr -
UA
The heat evolution rate Qrxn and the batch temperature are
determined as functions of time from calorimetric data
(Figure 7). The overall heat transfer coefficient and heat
transfer area corresponding to the pilot plant reactor were
measured, their product being equal to 225 W/°C. Using
these results, a jacket pre-cooling temperature of 10 °C was
estimated to allow nearly complete heat removal due to the
exothermic oxime reduction.
We emphasize that the approach outlined above is
applicable strictly for small batch temperature excursions.
For large-batch temperature excursions, the energy ac-
cumulation must be considered along with changes in
reaction mixture and reactor wall heat conductivity which
will impact the heat transfer coefficient.
174
•
Vol. 5, No. 2, 2001 / Organic Process Research & Development

2 h. The higher hydrogen consumption in the pilot plant
reactor (ca. 5%) could be due to H2 cylinder temperature
variations, as the cylinder temperature was not monitored.
Agreement in overall yield of 2-aminoindan hydrochloride
between laboratory (66%) and plant (68%) was excellent,
as was agreement in product purity (>97% in both cases).
Thus, successful scale-up was achieved.
Conclusions
A novel synthetic route was demonstrated for producing
2-aminoindan hydrochloride, a key starting material for a
drug substance, using a relatively inexpensive reactant,
ninhydrin. A product isolation procedure was also devised
to give the stable hydrochloride salt in good yield.
Subsequent process development work, focusing on
reduction of the oxime diketone intermediate, demonstrated
that the reduction could be achieved using a dry Pd/C catalyst
at 20-25 °C to obtain the desired product in good yield.
Using reaction calorimetry, two reaction regimes were
identified: a rapid exothermic reaction involving oxime
reduction, followed by a slower and much less exothermic
reaction in which reduction of the diketone occurred. The
delineation of these reaction regimes was used to optimize
the reduction process by performing the exothermic oxime
reduction at milder conditions (20 psig, 25 °C), with the
slower diketone reduction carried out at higher temperature
(35 °C) and pressure (40 psig) to obtain a faster reaction
and reduce pilot plant time.
Figure 9. Comparison of pilot plant and laboratory hydrogen
consumption profiles.
Scale-Up Results. Scale-up of Scheme 2 was performed
in a 114-L reactor, representing a 100-fold scale-up from
the RC1 experiments. In total, eight batches were performed
to produce 17.2 kg of product. Representative results (Figure
9) indicate that good qualitative agreement was obtained
between the laboratory and pilot plant hydrogen consumption
profiles. A discrepancy occurred starting at about 1 h of
reaction time, with the pilot plant hydrogen consumption
being higher. Close examination of the batch and jacket
temperature profiles revealed that after the jacket temperature
was increased from 10 to 20 °C (as batch temperature control
in the pilot plant was achieved by setting jacket temperature),
the batch temperature underwent a temperature excursion to
30 °C. This could explain the greater hydrogen consumption
in the pilot plant reactor after about 1 h of reaction due to a
greater reaction rate than that in the laboratory. A more
gradual jacket temperature increase would have resulted in
a smaller batch temperature excursion.
The optimized laboratory process was successfully scaled
up 100-fold in the pilot plant, giving excellent agreement
between both product yield (66% in the laboratory vs 68%
in pilot plant) and product purity (>97% in all cases).
Acknowledgment
We thank Tao He for performing the LC/MS analyses,
After the greater H2 consumption in the pilot plant reactor
after about 1 h of reaction time, the hydrogen consumption
profiles are almost parallel, indicating very good kinetic
agreement between laboratory and pilot plant reactors. In
particular, the autocatalytic process occurring after about 2.5
h was observed in the pilot plant, as evidenced by the
presence of an inflection point in the data occurring at about
Guy Yowell and Apurva Chaudhary for analytical results,
and are grateful to Richard Surma for sharing the pilot plant
results.
Received for review October 18, 2000.
OP000222H
Vol. 5, No. 2, 2001 / Organic Process Research & Development
•
175