Process improvement in the production of a pharmaceutical intermediate using a reaction calorimeter for studies on the reaction kinetics of amination of a bromopropyl compound

Article abstract of DOI:10.1021/op970055u

A synthetic process to produce a pharmaceutical intermediate, [3-(dimethylamino)propyl]triphenylphosphonium bromide (1), was established in the laboratory. Process safety evaluations did not show any severe problems. In order to scale up to plant scale, w

Full text of DOI:10.1021/op970055u

Organic Process Research & Development 1998, 2, 169−174
Process Improvement in the Production of a Pharmaceutical Intermediate
Using a Reaction Calorimeter for Studies on the Reaction Kinetics of Amination
of a Bromopropyl Compound
Takahiro Sano,* Toru Sugaya, and Masaji Kasai
Sakai Research Laboratories, Kyowa Hakko Kogyo Co., Ltd., 1-1-53 Takasu-Cho, Sakai-City, Osaka, 590, Japan
Abstract:
A synthetic process to produce a pharmaceutical intermediate,
[3-(dimethylamino)propyl]triphenylphosphonium bromide (1),
was established in the laboratory. Process safety evaluations
did not show any severe problems. In order to scale up to plant
scale, we simulated the reaction for both heat production and
byproduct production profiles with kinetic parameters. The
kinetic parameters were obtained by laboratory experiments
and reaction calorimeter measurements. Under the established
conditions, the production operated in a safe manner with high
quality.
Figure 1. Synthesis of [3-(dimethylamino)propyl]triphen-
ylphosphonium bromide (1).
Introduction
A pharmaceutical intermediate, [3-(dimethylamino)pro-
pyl]triphenylphosphonium bromide (1), has been synthesized
from (3-bromopropyl)triphenylphosphonium bromide (2) and
1
dimethylamine (Figure 1) . The process flow diagram is
shown in Figure 2. The solid of 2 in the solvent (methanol)
gradually dissolves during the addition of aqueous dimethyl-
amine and is completely dissolved at the end of the addition.
The conversion of the starting material 2 was quantitative,
and the isolated yield of 1 was over 90%; however, a small
amount (about 1% at 30 °C) of dimeric byproduct 3 was
produced (Figure 3), which was difficult to remove in the
purification step.
Figure 2. Process flow diagram of synthesis 1.
For the commercial production of 1, it was planned to
3
carry out the process on a 2 m scale, so this process could
be assessed from a scale-up point of view. We evaluated it
for thermal safety as well as control of byproduct formation,
and we studied the reaction kinetics in order to determine
the best conditions for plant scale production.
Figure 3. Subreaction to produce the dimeric byproduct 3.
Results and Discussion
position, oxygen balance, and so on) of all of the compounds
were ranked “low”.
Safety Evaluation of the Process. From the MSDS
(
material safety data sheets), there were no toxicity and
The results of the differential scanning calorimeter (DSC)
analysis showed that there was no exothermic decomposition
till 200 °C with the starting materials 2 and 3, and the
quantities of heat were less than 400 J/g. From the results
of the accelerating rate calorimeter (ARC) analysis, the
observed onset temperatures of 1 and 2 were also over 200
°C although thermal inertia factors (PHI factors) were less
than 5 (1.7-3.9). These materials were therefore unlikely
to present thermal instability hazards at the proposed plant
operating temperature.
explosiveness problems with regard to the starting materials
and the product. On the basis of the CHETAH calculation,
the plosive hazard classifications (maximum heat of decom-
2
†
Corresponding author. Phone: +81-722-23-5545. Fax: +81-722-27-7214.
E-Mail: takahiro.sano@kyowa.co.jp.
(
1) (a) Oshima, E.; Kumazawa, T.; Ohtaki, S.; Obase, H.; Ohmori, K.; Ishii,
H.; Manabe, H.; Tamura, T.; Shuto, K. Eur. Patent EP 235796, 1987. (b)
Corey, E. J.; Desai, C. Tetrahedron Lett. 1985, 26, 5747. (c) Satzinger, G.;
Fritschi, E.; Herrman, M. Ger. Patent DE 3248094, 1984. (d) Tretter, J. R.
U.S. Patent US 3509175, 1969.
(
2) ASTM Subcommittee. The ASTM CHETAH Version 7.0; The Computer
Program For Chemical Thermodynamic and Energy Release Evaluation;
ASTM Data Series DS51B: Philadelphia, 1994.
Next, the process safety was evaluated using a reaction
calorimeter (RC1). The heat of the reaction was measured
S1083-6160(97)00055-8 CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 04/21/1998
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
169

Chart 1. Kinetic equations of this process (serial-parallel
reaction)a
dC2
)
-kmainC C
- ksubC C2
2
HNMe2 1
dt
dCHNMe2
dt
)
-kmainC C
2 HNMe2
dC1
dt
)
kmainC C
- ksubC C2
2
HNMe2
1
dC3
)
ksubC C2
1
dt
Figure 4. Heat flow of the reaction: bold line (s), heat flow;
normal line (s), dimethylamine addition; dashed line (- - -),
temperature of the reactor.
a
C is the concentration of the material (the subscript indicates the compound),
t is the reaction time, kmain is the reaction constant of the main reaction (product
1), and ksub is that of the subreaction (product 3).
to be 93.6 kJ/mol during the addition of dimethylamine (at
30 °C) and 56.4 kJ/mol after the addition (Figure 4). The
heat flow curve dropped under zero base line during and
after the temperature ramp. We thought that this phenom-
3
enon was caused by heat loss from the cover of the reactor,
and we measured the heat flow with mixed methanol and
aqueous dimethylamine (without 1) under the same temper-
ature conditions. Comparing these results, there was no heat
during both the heating up and refluxing periods. The entire
heat of reaction was 150.0 kJ/mol, and the adiabatic
temperature rise (∆Tad) was calculated as about 40 °C on
the basis of the reaction mass and the heat capacity of the
reactants.
Figure 5. Arrhenius plot of the rate constants obtained from
isothermal reaction conditions: kmain shows the main reaction,
and ksub shows the subreaction.
Even when this reaction was performed with the reaction
calorimeter at a higher temperature (55 °C) or under adiabatic
conditions, the heat of the reaction did not increase (almost
Table 1. Arrhenius regression analysis results of kmain and
k
sub
param
value
correlation coeff
0.9886
150 kJ/mol) and the maximum temperature increase under
k
main
E
A
E
A
a
0
a
0
132.5 kJ/mol
adiabatic conditions (from 30 °C) was 37.0 °C (data not
shown). The maximum temperature was 67 °C (30 + 37 )
15
4.68 × 10 L/mol‚h
k
sub
101.6 kJ/mol
0.9837
67 °C), but the boiling point of the reactant was 72 °C
10
1.00 × 10 L/mol‚h
(aqueous methanol), and reflux did not occurr. From these
results, this process was judged to be safe, and it was thought
that no runaway reaction would occur even in the case of
an incorrect reaction temperature or a cooling failure. But
at higher temperatures (including adiabatic conditions),
byproduct 3 production was increased more than under
normal conditions (0.6% at 25 °C and 7.3% at 55 °C).
Kinetics Studies. The addition period and the addition/
reaction temperature affected the yield and quality of the
product 1, so we studied the reaction kinetics.
This reaction contains a main reaction (produces 1) and
a subreaction (produces 3). The substrate 2 reacts through
the main reaction and the subreaction (parallel reaction), and
the product 1 further reacts through the subreaction (serial
reaction). This type of reaction is a so-called serial-parallel
reaction (Chart 1).
Arrhenius plot. The subreaction rate constants and param-
eters were then obtained by the reaction of 1 and 2 under
isothermal conditions (Figure 5; Table 1). The Arrhenius
2
expressions fitted very well with the observed data (r >
0
.98).
Modification of the Rate Parameters with Heat Flow.
Generally, considering that the heat production is propor-
tional to the conversion, the heat production profile could
be calculated using the reaction kinetic parameters. In this
process, the subreaction is 20-70 times slower than the main
reaction (kmain . ksub) as shown in Figure 5, and the heat
production would be estimated only by considering the main
reaction.
In scale-up production, it seems difficult to keep the
reaction isothermal during the rapid addition of aqueous
dimethylamine, which would need about a 2 h addition
period (to be mentioned later). At first, we studied the
reaction under 30 °C isothermal conditions with a 1 h
addition period, and we compared the heat production profile
simulated with kinetic parameters and the actual heat
production measured by the reaction calorimeter. Obviously,
there was a difference, as shown in Figure 6.
At first, the reaction was performed under isothermal
conditions (25-50 °C) as described in the Experimental
Section and the main reaction rate constants (kmain) were
obtained. The kinetic parameters were calculated from the
(
3) The heat flow curve dropped under the zero base line during and after the
temperature ramp. This phenomenon was caused by heat loss from the upper
cover of the reactor and not by the endotherm: Operating instructions of
Reflux and Distillation Set, Mettler Toledo Instrument, 1989.
170
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

Figure 6. Heat flow measured with the reaction calorimeter.
The measured heat flow under 30 °C isothermal conditions with
a 1 h addition period is represented by the solid line (s), and
the calculated heat flow under the same conditions is repre-
sented by boxes (0). The total heat of the reaction (about 150
kJ/mol) is distributed on the basis of the reaction rate.
Figure 8. Heat flow curves measured for various stirring
speeds at 30 °C with a 2 h period of addition of aqueous
dimethylamine.
Figure 7. Heat flow measured after the dissolution of 2: solid
line (s), measured heat flow; boxes (0), calculated heat flow.
Figure 9. Heat flow curves measured for various addition
periods at 30 °C with a 200 rpm stirring speed.
In this reaction, 2 did not dissolve during the early stage,
but it dissolved gradually during the addition period. This
dissolution mode of 2 was thought to be one reason for the
difference that appeared in Figure 6. We then measured the
heat profile with the reaction calorimeter after dissolving 2
content increased). If the dissolution of 2 depended on the
solvent components, the time for dissolution should be same.
The addition periods were changed in the range 1-3 h
although the times for dissolution after the end of addition
were apparently different from each other (Figure 9). From
these results, the dissolution of 2 was not affected by physical
properties like agitation or solvent composition.
For all of the conditions shown in Figures 8 and 9, the
conversion of 2 was about 55% at the dissolution point. From
these results, the dissolution of 2 was affected by the reaction
progress; that is, the reaction rate up to the conversion of
4
with more methanol. The actual measurement result showed
a good agreement with the simulated profile based on the
reaction rate (Figure 7). In this case, the evaluation of the
heat production calculation using the parameters listed in
Table 1 showed high reliability.
On the other hand, these results also proved that the
dissolution of 2 was important for considering the heat
production, and during the early period of this reaction, the
rate-determining step seemed to be the dissolution of 2. If
so, the measured heat flow curve showed the dissolution rate
and we intended to clarify the entire reaction from these
aspects.
If there would be some mechanical factors influencing
the dissolution of 2, the time to dissolution might change
with stirring speed. The stirring speed of the calorimeter
was changed in the range 100-300 rpm, and heat flows were
measured, but they gave almost the same pattern (Figure 8).
Thus, the dissolution was not affected by agitation. By the
way, the composition of the reaction solvent changed during
the addition of aqueous dimethylamine (initially, the solvent
was only methanol; however, during the addition, the water
55% seemed to be equal to the rate of dissolution of 2 which
was expressed by the Arrhenius equation of the measured
heat flow curve. Therefore, the dissolution of 2 did not
change with scale-up.
The reaction calorimeter was operated under isothermal
conditions at different temperatures in the range 20-30 °C,
and from the measured heat flow curves the rate constants
were obtained by the Arrhenius plot (kdissolve, Figures 10 and
1
1; Table 2). From these results, the heat rate could be
expressed as in the equation in Chart 2.
The calculated heat production profile is as follows. The
heat production rate is proportional to the dissolution rate
until the conversion of 2 is 55%, after which the heat rate is
proportional to the main reaction rate as shown in Chart 2.
The dissolution rate and the main reaction rate were
calculated using the kinetic parameters listed in Tables 1 and
2. We checked the reliability of these kinetic parameters
by comparison of the actual and the calculated heat flow,
and the result showed good agreement (Figure 12), so we
(
4) Compound 2 is not very soluble in water (<0.005 g/mL at 20 °C) or
methanol (0.5 g/mL at 20 °C), so the reaction was significantly influenced
by the dissolution rate of 2. To use more solvent volume in order to dissolve
2
was not good for production efficiency; using less solvent to increase
efficiency caused the byproduct 3 to increase significantly because of high
concentration.
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
171

Figure 10. Heat flows at various temperatures and their slopes
during 2 h addition with a 200 rpm stirring speed.
Figure 12. Predicted and actual heat flow of the reaction at
30 °C with a 1 h addition period: solid line (s), actual heat
flow measured by reaction calorimeter; dotted line (‚‚‚), heat
flow calculated with eq 2 (considering dissolution); dashed line
(- - -), heat flow calculated without considering dissolution.
Figure 11. Arrhenius plot of the dissolution rate constants
(ksub) obtained from the slopes of heat flows measured by
reaction calorimeter (Figure 10).
Figure 13. Predicted yield of 3 for various temperatures and
addition periods.
Table 2. Arrhenius regression analysis results on the rate of
dissolution of 2
param
value
correlation coeff
0.9886
k
dissolve
E
A
a
141.3 kJ/mol
14
0
2.34 × 10 L/mol‚h
Chart 2. Equations for calculating the rate of heat
production considering dissolution of 2
1
5
10880/T
rate of heat production = (4.68 × 10 )e
C C
2 HNMe2
(
conversion of 2 >55%)
1
4
11580/T
=
(2.34 × 10 )e
C C
2 HNMe2
Figure 14. Predicted reaction time for various temperatures
with a 1 h addition of aqueous dimethylamine.
(
conversion of 2 > 55%)
could use those for further simulation studies. When the
slurry of 2 in methanol without aqueous dimethylamine was
heated up to 50 °C and dissolved, of course, no reaction
occurred and there was no exotherm or endotherm during
the operation. Thus the heat production profile could be
calculated without considering the dissolution itself.
Simulation Studies on the Optimization of the Reac-
tion. In the simulation studies, the addition of dimethylamine
was assumed to be operated under isothermal conditions and
then the temperature was also to be maintained constant. The
reaction was then judged to end when the conversion of 2
became less than 0.1%/h.
At first, we simulated the reaction from a quality point
of view. On the basis of the kinetic parameters (Table 1),
the yield of byproduct 3 was calculated at various temper-
atures and addition periods (Figure 13). When the operating
temperature was under 25 °C, 3 was expected to be formed
at almost 1% in each addition period. Even when the
temperature was 40 °C and the addition period was shorter
than 1 h, the yield of 3 was expected to be less than 2%. At
temperatures higher than 35 °C and addition periods greater
than 2 h, the byproduct 3 would increase markedly.
The yield of 3 seemed to be reduced at low temperature,
but the productivity of the product 1 became worse because
of the slow reaction (Figure 14). On the basis of productivity
and quality considerations, the reaction should be finished
within 12 h and the formation of 3 should be less than 2%.5
The best temperature was thought to be about 25 °C. If
the temperature was higher than 25 °C, the byproduct 3
would increase, and if the temperature was lower than 25
°
C, the reaction would not end in 8 h.
(5) In the subsequent synthetic steps, the byproduct 3 is effectively removed
when the yield of 3 is less than 3%.
172
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development

Figure 15. Predicted shortest addition period at various
temperatures of actual plant reactor.
Figure 16. Maximum temperature of the reaction in case of a
cooling failure (Tcf).
Chart 3. Equations for calculating the Damk o1 hler number
Table 3. Scale-up results of this reaction
a
and Tcf equations
yield, %
2
π
1
Da )
) kC2initialt
2
(
1 - x)
scale addn period, h temp range, °C
1
3
2
lab
plant
∼1
1.3
∼30
25-27
90-95 0.3-0.8
93.4 0.8
T ) T +
∆Tad
cf
r
x
π(Da)
a
Da is the Damk o¨ hler number, x is the conversion of starting material 2, k
is the rate constant of the main reaction, C2initial is the concentration of 2 at the
start of reaction, t is the reaction time, Tcf is the maximum temperature in the
is the reaction temperature, ∆Tad is the adiabatic
temperature rise of this reaction (about 40 °C), and π is the circular constant.
case of a cooling failure, T
r
the cooling ability of the reactor was worse than the estimated
cooling ability and/or the addition period became longer than
2
h, the yield of 3 was calculated to be 0.9% and the Tcf was
In order to decrease the byproduct, the addition period
should be short (Figure 13), but the minimum addition period
depended on the cooling ability of the actual reactor. By
considering the cooling ability, the shortest addition period
calculated as 57.2 °C. When the temperature was 30 °C for
the same addition period (1.7 h), the yield of 3 would be
0
.9% and the Tcf would be 56.8 °C. These differences in
the yield of 3 and the Tcf seemed not to be a severe problem
in this process. Therefore, the temperature for the addition
of dimethylamine could be between 25 and 30 °C.
Summary. We have simulated this process with high
quality, high productivity, and safety in the plant by studying
the reaction kinetics. After these studies, we did scale
testing. For temperature control, we used partial brine flow
6
for keeping the reaction at 25 °C was calculated to be 1.7 h
(Figure 15). So we thought that the best set of conditions
was addition of dimethylamine as rapidly as possible with
the temperature kept at about 25 °C (1.7 h).
Next, we studied the reaction from a safety point of view.
When the reaction was terminated, the time needed after the
addition of dimethylamine and the heat accumulation was
not clear (Figure 4). Therefore, we had to establish the
conditions under which the heat accumulation would be
lower. When the reaction temperature is low, the heat
accumulation becomes large because of the slow reaction,
and the maximum temperature in the case of a cooling failure
(-5 °C) during and after the dimethylamine addition. The
cooling ability of the reactor was better than estimated, and
the addition period was shorter than 1.7 h, which were good
for quality and productivity. The results are shown in Table
3
.
In this case, we have studied only isothermal conditions,
(Tcf) will be high. On the other hand, when the reaction
but expanding to nonisothermal conditions will make it
possible to find an easier way to control the operation with
high-quality product. We intend as the next step to simulate
the process under nonisothermal conditions.
temperature is high, the heat accumulation becomes small,
but it means that the starting temperature is high, so the Tcf
also becomes high. In these cases, the Tcf can be calculated
using the Damk o¨ her number (Da), which is the dimensionless
number for the reaction rate (Chart 3).7 According to the
equations in Chart 3, the Tcf of this process could be
calculated for various temperatures and addition periods
Experimental Section
All materials were commercially available. (3-Bromopro-
pyl)triphenylphosphonium bromide (2) was purchased from
Manac Incorporated (>97% purity), and MeOH and 50%
aqueous dimethylamine were purchased from Ishizu Seiyaku,
Ltd. For an analytical sample, the product, [3-(dimethyl-
amino)propyl]triphenylphosphonium bromide (1), was pur-
chased from Aldrich Co.
(Figure 16). It was thought that the lower the Tcf is, the
easier it is to control the reaction temperature if a cooling
failure occurrs, which provides more safety.
Using the best conditions from a quality point of view of
“25 °C, 1.7 h addition” (yield of byproduct 3 was calculated
to be 0.8% in this case), the Tcf was estimated as 59.6 °C. If
The analysis of the reaction was carried out using the
1
HPLC internal standard method as described below. H and
(
6) The shortest addition period was calculated from the balance of the total
heat production and the cooling capacity of the actual reactor in which this
reaction was operated.
1
3
C NMR spectra were recorded on a Bruker AC300
spectrometer with TMS as the internal standard. IR spectra
were recorded on a Shimadzu FTIR-4300.
(7) (a) Hugo, P.; Steinbach, J.; Stoessel, F. Chem. Eng. Sci. 1988, 43, 2147.
(b) Stoessel, F. Chem. Eng. Prog. 1995, 9, 46.
Vol. 2, No. 3, 1998 / Organic Process Research & Development
•
173

HPLC Analysis. Column: YMC-Pack ODS-AM 312.
50 × 6 mm (YMC Co., Ltd.). Column temperature: 25
dimethylamine (2.8 equiv) which was preheated to each
reaction temperature. The compound 2 dissolved after a few
minutes. Every 20 min, 1 mL of the resultant solution was
taken, to which was added 1 mL of 0.05 N HCl for
quenching the reaction. After each sample was diluted to
10 mL with MeOH and mixed with the internal standard,
the concentrations of the starting material 2, the product 1,
and the byproduct 3 were analyzed by HPLC. The main
reaction rate constant was obtained with correlation of the
reaction time and the concentrations of the product 1 as
follows: Considering first-order reaction, time was plotted
for (x axis) as a function of the natural logarithm of the
reciprocal of the remainder ratio of 1 (ln[1/(1 - x)]) (y axis);
the relation of these points was a linear correlation across
1
°
2 3
C. Mobile phase: 1.5 L of H O, 1.0 L of CH CN, and 0.5
L of MeOH with 10.25 g of potassium dihydrogen phosphate
and 6.5 g of sodium 1-octanesulfonate, pH 3.0 adjusted with
5% H PO . Flow rate: 1 mL/min. Detection: UV 225
3 4
nm. Internal standard: butyl p-hydroxybenzoate.
CHETAH Calculation. The CHETAH calculation was
run with Version 7.0. CHETAH has no group value of salts
such as 1-3, so these compounds were calculated as free
bases.
8
DSC Measurements. A MAC Science Co., Ltd.,
DSC3100 with sa tainless steel cell was used under 2 kg/
2
cm pressure (air). The heating rate was 10 °C /min from
2
5
0 to 500 °C.
the origin (correlation coefficient r > 0.98), and the slope
ARC Measurements. A Columbia Scientific Industries
of the line showed the main reaction rate constant kmain.
ARC with a titanium bomb was used as follows. Sample
mass: 1-3 g. Thermal inertia factors (PHI factors): 1.7-
As stated as above, the subreaction rate constant was
obtained; thus 4.3 g (10 mmol) of 1 and 4.6 g (10 mmol) of
2 in 10 mL of MeOH (dissolved) were reacted at each
reaction temperature. After the reaction was quenched, the
concentrations of 1-3 were analyzed by HPLC.
3
.9. Start temperature: 40 °C. End temperature: 405 °C.
Slope sensitivity: 0.02 °C/min. Heat step temperature: 5
C. Calculation step temperature: 0.2 °C. Wait time: 10
min.
Synthesis of N,N-Dimethyl-N,N-bis[3-(triphenylphos-
phonio)propyl]ammonium Tribromide (3). A mixture of
40.0 g (0.244 mol) of 2 and 96.0 g (0.244 mol) of 1 in 288
°
Measurement of Heats of Reaction. Mettler’s RC1
reaction calorimeter outfitted with an AP01 batch reactor (2
L glassware) was used. The starting material 2 (500 g) and
MeOH (700 mL) were charged in the reactor, and the
operating conditions (temperature, stirring speed) were
determined. After the temperature was constant, 50%
aqueous dimethylamine (2.8 equiv) at ambient temperature
was proportionally added with a dosing pump over a set
period. The measurement was continued till heat production
ceased and the end of the reaction was determined by HPLC
analysis. Calibrations for determining the heat capacity and
the heat transfer coefficient were performed before and after
the reaction. Values of heat production were cited per mole
of 2. Heat flow calculation terms included Qflow (heat
production), Qaccum (heat accumulation), Qdos (heat flow due
to dosing), and Qloss (heat loss through the reactor cover).
1
mL of MeOH was heated to reflux. After 14 h, the solvent
was evaporated and the residue was crystallized from 400
mL of toluene. The crude crystal of 3 (192.3 g) was
recrystallized from 500 mL of CH
3
CN. Then 109.8 g of 3
was obtained with 98% purity (50% yield).
1
H NMR (CDCl
m), 2.32 (6H, s), 3.75 (4H, t, J ) 15.8 Hz), 4.25 (4H, t, J )
.7 Hz), 7.65-7.96 (30 H, m).
3
-CD OD, 300 MHz): δ 2.16-2.33 (4H,
3
7
13
C NMR (CDCl
3
-CD OD, 75 MHz): δ 19.6 (d), 17.1
3
(
(
s), 50.7 (s), 63.7 (d), 117.4 (d), 130.7 (d), 134.1 (d), 135.2
d).
IR (KBr, cm ): 3420, 1420, 1440, 1180, 1000, 740, 720,
90.
-1
6
Determination of Reaction Constants. To a slurry of
(20.0 g, 43 mmol) in 28 mL of MeOH stirred at each
reaction temperature was quickly added 50% aqueous
Received for review October 28, 1997.
OP970055U
2
174
•
Vol. 2, No. 3, 1998 / Organic Process Research & Development