CO2 Offgas as a Mechanistic Probe and Scale-up Tool in N-Acylations Using Mixed Anhydrides from Amino Acids and Isobutyl Chloroformate

Article abstract of DOI:10.1021/op034085a

The reaction pathways of an N-acylation reaction, involving a mixed anhydride intermediate produced from reaction of an amino acid with isobutyl chloroformate, were analyzed using a novel technique involving the quantitation of CO₂ evolved during various parts of the process. Under the normal mode of addition, in which chloroformate is added to a carboxylate anion solution, the dominant mode of yield loss to undesired product was shown to be the formation of a symmetric anhydride, implying that byproduct urethane formation was almost exclusively due to reaction of unreacted chloroformate with amine. This result suggests that inferred high concentrations of carboxylate anion present with the normal addition mode should be minimized to decrease yield loss. This hypothesis was confirmed by demonstrating that symmetric anhydride formation is almost eliminated when the reverse addition (viz., addition of carboxylate anion solution to chloroformate) was carried out These results, in conjunction with the relative temperature insensitivity of the pathways occurring after amine addition, were used to deduce that the higher amounts of remaining starting acid observed on scale-up were due to slower mixing at larger scales, resulting in greater symmetric anhydride yields. The reverse addition, although forming much lower symmetric anhydride amounts, is nevertheless a rapid reaction whose selectivity could be impacted adversely on scale-up due to slower mixing. The offgas methodology presented here can be used to quantify reaction selectivity as a function of scale and thus serve as a diagnostic tool for assessing mixing efficiency at larger scales.

Full text of DOI:10.1021/op034085a

Organic Process Research & Development 2003, 7, 888−895
CO Offgas as a Mechanistic Probe and Scale-up Tool in N-Acylations Using
2
Mixed Anhydrides from Amino Acids and Isobutyl Chloroformate
Apurva Chaudhary, Michael J. Girgis,* Mahavir Prashad,* Bin Hu, Denis Har, Oljan Repicˇ, and Thomas J. Blacklock
NoVartis Institute for Biomedical Research, Chemical and Analytical DeVelopment, 1 Health Plaza,
East HanoVer, New Jersey 07936, U.S.A.
Scheme 1
Abstract:
The reaction pathways of an N-acylation reaction, involving a
mixed anhydride intermediate produced from reaction of an
amino acid with isobutyl chloroformate, were analyzed using
a novel technique involving the quantitation of CO2 evolved
during various parts of the process. Under the normal mode
of addition, in which chloroformate is added to a carboxylate
anion solution, the dominant mode of yield loss to undesired
product was shown to be the formation of a symmetric
anhydride, implying that byproduct urethane formation was
almost exclusively due to reaction of unreacted chloroformate
with amine. This result suggests that inferred high concentra-
tions of carboxylate anion present with the normal addition
mode should be minimized to decrease yield loss. This hypoth-
esis was confirmed by demonstrating that symmetric anhydride
formation is almost eliminated when the reverse addition (viz.,
addition of carboxylate anion solution to chloroformate) was
carried out. These results, in conjunction with the relative
temperature insensitivity of the pathways occurring after amine
addition, were used to deduce that the higher amounts of
remaining starting acid observed on scale-up were due to slower
mixing at larger scales, resulting in greater symmetric anhy-
dride yields. The reverse addition, although forming much lower
symmetric anhydride amounts, is nevertheless a rapid reaction
whose selectivity could be impacted adversely on scale-up due
to slower mixing. The offgas methodology presented here can
be used to quantify reaction selectivity as a function of scale
and thus serve as a diagnostic tool for assessing mixing efficiency
at larger scales.
chloroformate in this case, is added to the reaction mixture
and reacts with the carboxylate anion to form a mixed
carboxylic-carbonic, or asymmetric, anhydride (Figure 1);
this mode of addition is termed here the “normal addition
mode”.
Undesired pathway a occurs via reaction of asym-
anhydride with additional carboxylate anion to form a
symmetric anhydride and a carbonate anion. The latter
ultimately forms a carbonic acid, which decomposes to give
isobutyl alcohol and CO2. The formation of CO2 before
addition of 2 is thus indicative of symmetric anhydride
formation, if no additional pathways forming CO2 occur.
Although two such additional pathways are possible, it is
shown in the Supporting Information that their contributions
are negligible under the present conditions. The formation
of CO2 before addition of 2 is thus indicative solely of sym-
anhydride formation.
The desired reaction to form 3 involves reaction of asym-
anhydride with the amine at the carboxylic acid carbonyl.
The reaction byproduct is a carbonic acid which ultimately
decomposes to give isobutyl alcohol and CO2. Formation of
the latter after addition of 2 is thus indicative of attack by
the amine at the carboxylic acid carbonyl As this pathway
gives the desired product, it is termed attack at the “desired”
carbonyl in what follows.
Introduction
Amine 2 can also react at the carbonic acid carbonyl by
an analogous mechanism (undesired pathway b). In this case,
referred to here as attack at the “undesired” carbonyl, the
reaction products include a byproduct urethane and 1.
Although 1 formed via by the latter route can be converted
to 3 by using additional amounts of N,N,-dimethylbenzy-
lamine, isobutyl chloroformate, and 2, this represents an
inefficient mode of preparing 3, as it requires multiple
additions of reagents to consume the liberated 1.
Attack by 2 at either carbonyl carbon atom of the
symmetric anhydride (Figure 1) can also occur to give the
desired amide 3 and liberate 1. This pathway for 3 production
is also inefficient because 2 moles of 1 are consumed per
N-Acylation reactions, with mixed carboxylic-carbonic
anhydrides intermediates produced by reaction of amino acids
with isobutyl chloroformate, comprise key steps in the
synthesis of a series of selective antagonists of the human
NK-1 tachykinin receptor.1 The first step in the synthesis of
one such compound involves the coupling of BOC-L-3-(2-
naphthyl)alanine (1) with N-benzylmethylamine (2) in the
presence of isobutyl chloroformate and N,N,-dimethylben-
zylamine (Scheme 1).
The coupling is performed by first deprotonating amino
acid 1 with N,N-dimethylbenzylamine to form a carboxylate
anion. In typical preparations,2,3 the coupling agent, isobutyl
(1) Walpole, C. S. J.; Prashad, M.; Har, D.; U. S. Patent No. 6,107,293, 2000.
(2) Sole, N.; Torres, J. L.; Garcia Anton, J. M.; Valencia. G.; Reig, F.
Tetrahedron 1986, 42 (1), 193-198.
(3) Chen, F. M. F.; Lee, Y.; Steinauer, R.;Benoiton, N. L. Can. J. Chem. 1987,
65, 613-618.
888
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development
10.1021/op034085a CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/30/2003

Figure 1. Reaction scheme for entire process. Part I comprises formation of the carboxylate anion and reaction of the latter to
form the anhydrides. Part II occurs after addition of amine 2.
mole of 3 produced. The presence of 1 at the end of the
reaction can thus occur by either of two undesired pathways,
one of which (undesired path b) also forms a urethane
byproduct. The urethane can also form from reaction of 2
with isobutyl chloroformate remaining from high consump-
tion of 1 via undesired pathway a.
knowledge, to devise rational processing strategies to
improve both selectivity and process robustness with increas-
ing scale.
In this contribution, a novel methodology is presented in
detail which allows quantitative determination of the con-
tributions of the operative pathways; a preliminary account
is given elsewhere.5 This methodology comprises quantifying
the CO2 amount evolved during the various parts of the
process in conjunction with HPLC analyses of the starting
acid 1 and final product 3. From such data, one can (a)
determine the amount of sym- and asym-anhydride formed
during the reaction of 1 with chloroformate before 2 addition
and thus quantify the contribution of undesired path a and
(b) distinguish between attack by 2 at the carboxylic acid
carbonyl of the asym anhydride giving the desired product
vs attack at the carbonic acid carbonyl (undesired path b).
The contributions of the two urethane-forming pathways can
also be established and thus allow rational development of
conditions that give minimal byproduct yields. In addition,
it was hoped that such quantitative insights could allow
inference of the cause of the increasing amounts of 1
observed at the end of the reaction upon scale-up and thus
provide guidance for improving process robustness.
Although HPLC analysis can in principle be used to
quantify the amounts of sym- vs asym-anhydride formed, its
Previous investigations2,3 have reported formation of the
urethane byproduct and addressed key parameters impacting
its yield (e.g, selection of solvent, coupling agent, and
temperature); its formation pathways have not been delin-
eated. Earlier work4 has also shown that at larger scales,
higher amounts of starting acid were detected at the end of
the reaction and, by inference, lower amide yields. For
example, at a reaction scale corresponding to 15.7 g of 1,
less than 3% of the 1 added initially was detected at the end
of the reaction; in contrast, at a 400-g scale, 22% of the 1
added was detected. Additional amounts of reagents needed
to be added to convert the remaining 1 to 3. The reaction
thus became more inefficient as the scale was increased. (In
the latter work, N-methylmorpholine was used as the
deprotonating base instead of N,N-dimethylbenzylamine, and
ethyl acetate was the solvent, but neither change impacts
the conclusions cited.) It would thus be instructive and useful
to have a general means by which the contributions of the
various reaction pathways can be quantified and, from this
(4) Prashad, M.; Prasad, K.; Repicˇ, O.; Blacklock, T. J. Org. Process Res. DeV.
1999, 3, 409-415.
(5) Chaudhary, A.; Girgis, M. J.; Prashad, M.; Hu, B.; Har, D.; Repicˇ, O.;
Blacklock, T. J. Tetrahedron Lett. 2003, 44, 5543-5546.
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
889

with a 5-µL sampling loop, an Agilent GasPro capillary
column (60-m length, 0.25 mm i.d.) and a thermal conduc-
tivity detector. GC operating parameters are given in the
Supporting Information.
Offgas volume was determined using a Ritter model TG05
wet test meter, which operates on the principle of gas
displacement. The wet test meter was filled with a paraffinic
white oil (Aldrich Mineral Oil, white, light, catalogue no.
33,077-9) in which N2 and CO2 are very sparingly soluble
at atmospheric pressure; for example, at atmospheric pres-
sure, the equilibrium mole fraction of N2 and CO2 in
n-hexadecane are 0.0013 at 25 °C and 0.014 at 27 °C,
respectively.8,9 The wet test meter was placed downstream
of the GC sampling loop, and the cumulative volume was
recorded using a data-logging system.
HPLC analyses were performed on selected samples using
a Hewlett-Packard series 1100 HPLC equipped with a diode
array detector. Method details are given in the Supporting
Information. All materials were used as received. Sources
and purities of all materials are given in the Supporting
Information.
Figure 2. Schematic diagram of apparatus.
application is complicated by the possible instability of the
anhydrides under HPLC analysis conditions and the need to
synthesize and purify authentic samples of the intermediate
anhydrides to serve as calibration standards for quantitative
analysis. Using the CO2 offgas technique, analysis of
intermediates is obviated, and only HPLC analyses of starting
acid 1 and product 3, which are stable, are needed. This
methodology also allows quantitative application to similar
coupling reactions without the need to synthesize intermedi-
ates.
In situ Fourier transform infrared spectroscopy6 could also
be used to quantify the anhydride amounts. However,
application of this method, which was tried at first, also
requires synthesizing authentic samples of the anhydrides
to identify their respective spectroscopic bands and to serve
as calibration standards. Additional disadvantages include
the complexity of the spectra due to the various carbonyl
species present and the tedious calibrations required for
quantitative analysis.
Offgas Flow Volume and Flow Rate Determinations.
During a gas-forming reaction, the wet test meter measures
the cumulative volume of total gas flowing through it at
ambient temperature and pressure, including N2 purge gas
as well as chemical offgas (i.e., the gas produced in the
chemical reaction). To quantify the chemical offgas volume,
it is necessary to decouple its flow rate from that of the purge
gas. This decoupling was done as follows. First, after
charging starting materials but before the start of the reaction,
the flow rate of N2 purge gas was set to a constant value
(nominally 50 cm3/min) with the mass flow controller. The
cumulative volume was then determined with the wet test
meter for at least 30 min. The actual purge gas flow rate
was determined by numerical differentiation of the cumula-
tive volume during this period. The numerical differentiation
was carried out by determining the slope of a line, fitted by
the method of least squares,10 to at least 50 consecutive
volume readings, each taken every 5 s.
Experimental Section
Apparatus and Materials. Reaction experiments were
performed using the apparatus shown schematically in Figure
2. The apparatus comprises a Mettler-Toledo RC1 reaction
calorimeter equipped with a 1-L glass MP-10 vessel. The
vessel inserts included a Hastelloy pitched-blade agitator and
Hastelloy temperature and calibration probes. The jacket
temperature was also monitored. Calorimeter operating
principles are discussed elsewhere.7 A dosing pump (Promi-
nent G4) was used to meter liquids into the reaction vessel.
Nitrogen gas was introduced either in the headspace or by
sparging into the liquid with a dip tube fitted with a sintered
metal filter element attached at the end of the tube to give
fine bubbles. The nitrogen gas flow was metered with a
Brooks model 5850i mass flow controller.
During the experiment, which was always performed with
N2 purge gas flowing, the cumulative total gas volume was
measured with the wet test meter. The flow rate of total
offgas was determined by numerical differentiation as
described above. The flow rate of the chemical offgas was
determined by subtracting the purge gas flow rate determined
earlier from the total offgas rate during the experiment. The
cumulative offgas volume was subsequently determined by
numerical integration of the offgas flow rate using the
trapezoidal rule.10 To allow rapid computation of these
quantities, the procedure was automated by first inputting
the purge and total offgas cumulative volumes into a
relational database, and subsequently performing the calcula-
tions using a Microsoft Visual Basic computer program
which read the input data from the database.
The composition of the offgas was determined by gas
chromatography using a Hewlett-Packard 5890 GC equipped
(6) Landau, R. N.; McKenzie, P. F.; Forman, A. L.; Dauer, R. R.; Futran, M.;
Epstein, A. D. Process Control Qual. 1995, 7, 133-142.
(7) Girgis, M. J.; Kiss, K.; Ziltener, C. A.; Prashad, M.; Har, D.; Yoskowitz,
R. S.; Basso, B.; Repicˇ, O.; Blacklock, T. J.; Landau, R. N. Org. Process
Res. DeV. 1997, 1, 339-349.
(8) Tremper, K. K.; Prausnitz, J. M. J. Chem. Eng. Data 1976, 21, 295-299.
(9) Chai, C. P.; Paulaitis, M. E. J. Chem. Eng. Data 1981, 26, 277-279.
(10) Press, W. H.; Flannery, B. P.; Teukolsky, S. A.; Vettering, W. T. Numerical
Recipes The Art of Scientific Computing; Cambridge University Press: New
York, 1986; Chapter 14.
890
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

In the experiments performed here, the gas evolution
occurred during periods in which liquid was added to the
reactor. To prevent incorrectly counting the gas displaced
by the liquid as evolved chemical offgas, it was necessary
to subtract the volume of liquid added from the cumulative
chemical offgas evolved. The reliability of the offgas
measurement was demonstrated using a reaction forming a
known amount of gas during addition of a liquid, namely,
the neutralization of sodium bicarbonate with sulfuric acid;
details are given in the Supporting Information.
The last reaction is due to unreacted IBCF as a consequence
of 1 consumption to form sym-anhydride via undesired
pathway a (Figure 1).
The contributions of the four reactions above can be
determined from mass balances. Thus, the number of moles
of 3 formed via attack at the desired carbonyl atom is given
by:
nd3esired carbonyl ) nII
(5)
CO2
The contribution of the third pathway, representing the
number of moles of 1 or 3 formed by reaction of sym-
anhydride with 2, can be calculated from the CO2 amount
formed in Part I:
Results and Discussion
Reaction Pathway Analysis. Part I reactions can be
represented by the following chemical equations:
nf1rom sym-anhydride ) nf3rom sym-anhydride ) nICO
(6)
1 + DMBA + IBCF f asym-anhydride + DMBA-HCl
1 + asym-anhydride f sym-anhydride + i-BuOH + CO2
2
The contribution of the fourth pathway can be determined
as follows. The number of moles of unreacted IBCF, and
where DMBA, DMBA-HCl, and IBCF, denote, respectively,
N,N-dimethylbenzylamine, its hydrochloride salt, and isobu-
tyl chloroformate. The chemical equations above indicate
that 1 mol of CO2 is produced per mol of sym-anhydride
formed:
sym
thus the moles of urethane n
formed due to con-
urethane
sumption of 1 to form sym-anhydride, can be calculated from
the moles of sym-anhydride formed. Note, however, that the
A2rxn
urethane
total amount of urethane n
formed by reaction of 2
and IBCF includes the amount formed due to the stoichio-
metric excess of chloroformate reacting with 2:
nsym ) nICO
(1)
2
A2rxn
urethane
sym
urethane
0
0
n
) n
+ (nIBCF0 - n1 ) ) nCI O + (nIBCF0 - n1 )
where ni denotes the moles of component i and the
superscript refers to Part I. The stoichiometry also implies
that the molar amount of asym-anhydride formed equals the
difference between the moles of 1 charged and the moles of
1 consumed to form sym-anhydride; because 2 mol of 1 are
required to form 1 mol of sym-anhydride, the mass balance
shows:
2
(7)
Equations 5-7 indicate that the CO2 amounts formed, in
conjunction with the stoichiometry, suffice to determine the
contributions of three of the four pathways above.
The contribution of the pathway pertaining to attack at
the undesired carbonyl can be determined two ways. First,
the molar amount of 1 formed by this pathway, which is the
same as the molar amount of urethane formed by this
pathway, is equal to the difference between the total molar
amount of 1 obtained at the end of the reaction (which can
be determined by HPLC) and the amount of 1 formed via
sym-anhydride conversion. As the latter amount equals the
molar amount of CO2 formed in Part I, it follows that:
nasym ) n10 - 2nsym ) n10 - 2nICO
(2)
2
where the 0 superscript denotes initial amount.
The contributions of the reaction pathways forming asym-
and sym-anhydride can be expressed on a scale-independent
basis as percentage yields:
I
n
nsym
CO2
nu1ndesired carbonyl ) nundesired carbonyl
)
ysym
)
)
× 100 )
× 100
(3)
(4)
urethane
0
0
n1
n1
n1 - n1from sym-anhydride ) n1 - nICO (8)
2
n10 - 2nI
nasym
CO2
yasym
× 100 )
× 100
The undesired carbonyl pathway can also be expressed in
terms of 3 molar amounts because 1 mol of 3 is formed per
mol of 1 in the reaction of sym-anhydride with 2. The
contribution of sym-anhydride conversion can be expressed
as the difference between the total molar amount of 3 formed
and the molar amount of 3 formed by attack at the desired
carbonyl. Thus, the contribution of the undesired carbonyl
pathway can also be expressed as follows:
0
0
n1
n1
Because 2 mol of 1 are required to form 1 mol of
sym-anhydride, the sum of the percentage yields above do
not sum to 100. Instead, twice the sym yield added to the
asym yield gives 100.
The following chemical equations follow, respectively,
from the pathways for Part II:
nu1ndesired carbonyl ) nurethane
undesired carbonyl ) n1 - (n3 - nd3esired carbonyl) )
asym-anhydride + 2 f 3 + CO2 + i-BuOH
II
n1 - (n3 - n ) (9)
CO2
asym-anhydride + 2 f 1 + urethane
sym-anhydride + 2 f 3 + 1
The Part II results can also be expressed in terms of scale-
independent percentage yields:
IBCF + 2 f urethane + HCl
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
891

II
CO2
close to 100%, it is inferred that the contribution of the
pathway pertaining to attack at the undesired carbonyl is
negligible. Calculation of the yield of 1 formed by the latter
pathway using eqs 8 and 10 gives a value of 6.0%; a very
small value (0.4%) is obtained if the calculation is done using
eqs 9 and 10. These relatively small values and their large
differences indicate that accurate calculation of the pathway
involving 2 attack at the undesired carbonyl is difficult,
primarily because it involves subtracting quantities that are
similar in magnitude (e.g., nA1 and nICO ). Since the com-
bined contributions of the other pathways2are close to 100%,
the main conclusion is that the contribution of the undesired
carbonyl pathway is very small, and as a consequence, the
amount of urethane formed due to attack at the undesired
carbonyl is also small. Thus, the dominant pathway for
urethane formation under the conditions used here is due to
reaction of 2 with chloroformate, unconsumed due to sym-
anhydride formation.
The foregoing results also imply that the main cause of
yield loss with the normal addition mode at the scale
employed here is the formation of sym-anhydride in Part I,
and not due to attack at the undesired carbonyl by 2. This
finding suggests that high local 1 concentrations, which favor
reaction of asym-anhydride with 1 anion to form sym-
anhydride, should be avoided to obtain greater 3 yields. This
hypothesis prompted us to change the addition mode to a
reverse addition in Part I (viz., by dosing IBCF into 1 anion
solution) with the expectation that the lower 1 anion amounts
in the vessel should give less sym-anhydride formation.
Reverse Addition: Dosing of 1 Anion Solution into
IBCF. A 500-mL Erlenmeyer flask was charged with 51.23
g (0.162 mol) of 1 in 112.95 g (131 mL) of toluene. N,N,-
dimethylbenzylamine (23.8 g, 0.176 mol, 1.08 equiv) was
added dropwise using an addition funnel, which was
subsequently rinsed with 5.0 g (5.8 mL) of toluene. The
MP10 vessel was charged with isobutyl chloroformate (23.55
g, 0.172 mol, 1.06 equiv) and toluene (337 g, 390 mL), and
the mixture was cooled to -12 °C. The 1 anion solution
(191.7 g) was then added at that temperature over a 40-min
period, and the transfer line was rinsed with 10 g (11.6 mL)
of toluene (Figure 4). CO2 was again the sole offgas
component detected, but as shown in Figure 3 and Table 1,
the volume of CO2 evolved was lower by almost a factor of
about 9 in comparison to results from the experiment utilizing
the normal addition mode.
n
y3desired carbonyl
)
× 100
(10)
0
n1
n1undesired carbonyl
yu1ndesired carbonyl ) yundesired carbonyl
)
× 100 (11)
urethane
0
n1
where n1undesired carbonyl is given by either eq 8 or eq 9.
I
n
CO2
ys1ym ) ys3ym
)
× 100
(12)
(13)
0
n1
nICO + (n0IBCF - n01)
2
A2 rxn
urethane
y
)
× 100
n10
Normal Addition Mode: Addition of IBCF to 1 Anion
Solution. The pathways were first characterized quantita-
tively using the normal addition mode. 1 (51.22 g, 0.162
mol) and toluene (450.9 g, or 520 mL) were added to the
MP10 vessel. At an agitation rate of 400 rpm, N,N,-
dimethylbenzylamine (23.8 g, 0.176 mol, 1.08 equiv) was
pumped into the vessel over 20 min at a 20 °C batch
temperature. The transfer line (between the pump and vessel)
was subsequently rinsed with toluene (3.9 g, 4.5 mL), which
was added to the batch. The reaction mixture was cooled to
-12 °C at 1 °C/min, and held at that temperature for 85
min.
Isobutyl chloroformate (23.5 g, 0.172 mol, 1.06 equiv)
was subsequently added at a -12 °C batch temperature over
a 40-min period, with gas evolution observed during the
addition (Figure 3). The online GC analysis indicated that
CO2 was the sole reaction offgas component. The CO2 rate
and cumulative volume could therefore be determined by
subtracting the contribution of the N2 purge gas from the
total flow. The CO2 volume during Part I evolved was 688
cm3, corresponding to 0.0285 mol at the 22 °C wet test meter
temperature. Based on equations (3) and (4), the sym- and
asym-anhydride yields are calculated to be 17.5% and 65.0%,
respectively. Thus, a significant amount of sym-anhydride
was formed using the normal addition mode illustrated in
Figure 3.
The reaction mixture was held at -12 °C for 45 min,
after which 37.7 g of a solution of 2 in toluene (prepared by
adding 36.4 g of 2 to 20.2 g of toluene, giving 64.3 wt. %
of 2) was added over a 30-min period. The 2 added (0.200
mol) corresponds to 1.23 equiv. Additional CO2 was evolved
during this addition, the subsequent 30-min hold, the warm-
up of the reaction mixture to 21 °C at 1 °C/min, and the
H2SO4 addition (128 g of solution comprised of 7 g of
concentrated H2SO4 in 121 g of water added over 12 min)
to quench any remaining amines (Figure 3). Note that any
HCl formed due to reaction between 2 and isobutyl chloro-
formate would be quenched by excess 2 in solution and thus
not evolve as a gas.
After holding the reaction mixture for 40 min at -12 °C,
37.73 g of a solution comprising 36.4 g (0.200 mol, 1.2
equiv) of N-benzylmethylamine in 20.2 g of toluene was
added over a 30-min period, with the transfer line subse-
quently rinsed with 9 g (10 mL) of toluene. After holding
for 60 min, the reaction mixture was heated to 20 °C at 1
°C/min. Once the reaction mixture attained 20 °C and
following a 10-min hold, a solution (128.2 g) comprised of
7 g of concentrated sulfuric acid in 121.2 g of water was
added over 12 min, and the mixture was held for 1.5 h. CO2
evolution was monitored throughout the experiment. As in
the previous experiment, any HCl formed due to reaction
Using eqs 10 and 12, the yields of 3 formed by sym-
anhydride attack at the desired carbonyl and from reaction
of sym-anhydride are calculated to be 66.7% and 17.5%,
respectively. As twice the latter quantity added to 66.7 is
892
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 3. Temporal profile of cumulative offgas volume and reaction mixture mass with reaction performed using the normal
addition mode.
Figure 4. Temporal profiles of cumulative CO2 volume and reaction mixture in experiment with reverse addition.
between 2 and isobutyl chloroformate would be quenched
by excess 2 in solution.
yield loss under the normal addition mode (viz., sym-
anhydride formation), is thus almost completely eliminated
by the reverse addition. In fact, with the reverse addition,
the last two lines of Table 1 imply that most of the urethane
product is formed because of the stoichiometric excess of
IBCF added.
Addition of 2 at Higher Temperature. In an attempt to
induce reaction at the undesired carbonyl, an experiment was
performed using the reverse addition but with subsequent 2
solution addition carried out at -2 °C, i.e., 10 °C higher
than in previous experiments. As this higher temperature
represents one-third of the adiabatic temperature rise of the
Substantially more CO2 was evolved upon addition of 2
in solution and subsequent warming and quenching with H2-
SO4. Table 1 also shows that the contribution of the undesired
carbonyl pathway was again very small. This result is
plausible, because Part II was carried out in the same way
in both cases, and thus, the contribution of this pathway
would not be expected to change. Furthermore, with the
reverse addition, the contribution of the pathway forming 1
and 3 via the sym-anhydride is small as expected, since so
little sym-anhydride is formed in Part I. The main source of
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
893

the end of the reaction at larger scales. Bourne11 has identified
three mixing stages for single-phase fluids: (a) distributive,
or macromixing, in which large fluid eddies exchange
positions and mass, but with compositional uniformity
occurring only at a scale greater than the eddy size; (b)
dispersive, or mesomixing, in which larger eddies are reduced
in size due to turbulent shear and finer-grained mixture forms,
but with the mixture remaining highly segregated at the
molecular scale; (c) diffusive, or micromixing, in which
molecular diffusion occurs over short distances between
finely dispersed structures to give a mixture that is random-
ized on the molecular scale. Good micromixing is thus a
prerequisite to avoiding compositional inhomogeneities on
the molecular scale (i.e., high local concentrations) that can
impact selectivity of chemical reactions.
Table 1. Comparison of pathway contributions using two
modes of addition
normal
addition
reverse
addition
Part I
y
sym (eq 3)
17.5%
65.0%
1.89%
96.2%
yasym (eq 4)
Part II
yd3esired carbonyl (eq 10)
66.7%
94.7%
<2%
yu3ndesired carbonyl ) yundesired carbonyl (eq 11) <2%
urethane
y1sym ) ys3ym ) ysuyrmethane (eq 12)
y2urrextnhane (eq 13)
17.5%
23.7%
1.89%
8.74%
The macromixing time typically increases with increasing
scale.11 For example, when scaling up at constant power per
unit volume in geometrically similar vessels, the macromix-
ing time varies as the 2/3 power of the vessel diameter.11 At
constant power per unit volume, however, the micromixing
time is not impacted significantly.11 Thus, given that the
stages occur mostly in series,11 one should expect a longer
time to achieve uniformity at the molecular scale with
increasing scale (i.e. macromixing becomes the limiting
mixing process).
Table 2. Comparison of process performance with reverse
addition with 2 added at different temperatures
temperature of 2 addition, °C
-12 °C
-2 °C
yd3esired carbonyl (eq 10)
94.7% >99%
yu3ndesired carbonyl ) yundesired carbonyl (eq 11) <2%
<3%
2.41%
urethane
y1sym ) ys3ym ) ysuyrmethane (eq 12)
1.89%
We therefore infer that the yield loss of 3 occurring upon
scale-up using the normal addition mode4 is due to slower
macromixing of isobutyl chloroformate with increasing scale.
The slower macromixing is hypothesized to result in high
local 1 concentrations over longer time periods, resulting in
faster sym-anhydride formation, and thus giving greater yields
of the latter as the scale is increased. The small contributions
in all cases of the pathway involving attack of the wrong
carbonyl, coupled with the yield insensitivity when adding
2 at -2 °C to simulate hot spot formation due to poor mixing,
imply that the undesired carbonyl pathway is not a major
cause of 3 yield loss on scale-up.
Although the reverse addition reduced high local 1 anion
concentrations in the experiments performed on our scale,
the temperature profiles during the reverse addition (Figure
5) are indicative of a rapid, addition-controlled reaction
whose selectivity could be sensitive to the mixing rate,
suggesting that yield loss could occur on scale-up with the
reverse addition. The offgas methodology presented here
could then be used to quantify the impact of the slower
mixing expected upon scale-up on selectivity by measuring
the CO2 volume evolved during Part I as a function of scale.
If it is found, for example, that higher CO2 yields result at
larger scales, slower mixing at the latter scale can be
implicated. Furthermore, measures can be implemented to
counter the impact of slower mixing, such as decreasing the
addition rate, changing the feed location, adding at multiple
points, and adding more dilute solutions. The effectiveness
of these methods could be evaluated by measuring the CO2
volume evolved during Part I. The offgas methodology could
thus serve as valuable diagnostic tool that quantifies the
reaction, it represents a reasonable estimate of the extent of
local heating (i.e., a hot-spot formation) that could arise
because of slow dispersion of 2 into the solution containing
the anhydrides prepared in Part I.
A solution of deprotonated 1 in toluene was prepared as
in the previous experiment and then added to the MP-10
vessel containing isobutyl chloroformate (23.56 g, 1.06
equiv) in toluene (335.3 g, 390 mL) at a -12 °C batch
temperature. After rinsing the transfer line with toluene (11
g, 13 mL) and holding for 40 min, the batch was heated to
-2 °C at 1 °C/min. A solution (37.73 g) of 2 in toluene
(36.4 g of N-benzylmethylamine in 20.2 g of toluene) was
added over 30 min (1.2 equiv of 2). After a 1-h hold, the
mixture was heated to 20 °C at 1 °C/min and quenched with
H2SO4 as described above after a 30-min hold.
As shown in Table 2, the product yields were similar to
those obtained with addition at -12 °C, suggesting that a
-2 °C addition temperature is insufficiently high to cause
attack by 2 at the undesired carbonyl.
Mechanistic and Scale-up Insights. The foregoing
experiments thus indicate that the CO2 offgas technique can
delineate quantitatively the formation pathways for urethane
byproduct formation. Although the pathway delineations
were carried out for one particular N-acylation, the meth-
odology is applicable to other reactions of the same type,2,3
and should be especially useful to probe the impact of key
process variables (e.g., temperature, solvent type) and other
properties (e.g., steric hindrance of the coupling agent) on
the pathways for byproduct formation.
(11) Bourne, J. R. Mixing in Single-phase Chemical Reactors. In Mixing in the
Process Industries, 2nd ed.; Harnby, N., Edwards, M. F., Nienow, A. W.,
Eds.; Butterworth Heinman: Woburn, MA, 1992.
The aforementioned experiments are also useful for
inferring the cause of the higher amounts of 1 observed at
894
•
Vol. 7, No. 6, 2003 / Organic Process Research & Development

Figure 5. Temporal temperature profiles in experiment with reverse addition.
impact of mixing indirectly. Advantages of this methodology
include its noninvasive nature (since it involves online
analysis of a waste stream) and ease of implementation, since
the integral nature of the offgas measurement reading merely
requires reading the wet test meter at certain time intervals.
These results were used to infer that the 3 yield losses
observed on scale-up in earlier work were due to the slower
macromixing on the larger scale, giving high local 1
concentrations that result in greater yields of sym-anhydride,
and not to 2 attack at the undesired carbonyl.
The temperature profile during the reverse addition is
indicative of a rapid, addition-controlled reaction whose
selectivity could be sensitive to the slower mixing expected
upon scale-up. The offgas methodology can be implemented
easily to quantify the CO2 yield formed in Part I as a function
of scale and thus serve as a quantitative diagnostic for mixing
rate.
Conclusions
A methodology for quantifying the offgas produced during
chemical reactions was demonstrated and applied to quantify
the CO2 evolved during various parts of an N-acylation
reaction having a mixed anhydride intermediate produced
from reaction of an amino acid with isobutyl chloroformate.
The dominant pathway of yield loss under the normal
addition mode at the conditions employed here was shown
to involve sym-anhydride formation before 2 addition instead
of attack of 2 at the undesired carbonyl of the asym-
anhydride. Furthermore, the relatively high percentage yield
(viz., 17.5%) of sym-anhydride formed in Part I, coupled
with the reaction pathway analysis, suggests that high local
1 concentrations were present and that minimizing these is
the key to decreasing yield loss.
This suggestion was corroborated in an experiment in
which 3 yield loss was almost eliminated by reversing the
mode of addition, viz., by adding carboxylate anion solution
to isobutyl chloroformate to minimize the inferred high 1
local concentrations. With the reverse addition, the contribu-
tion to yield loss due to attack on the undesired carbonyl
was also negligible, even when the 2 addition was performed
at a 10 °C higher temperature.
Acknowledgment
We thank Drs. Wen Shieh, Prasad Kapa, and Hong-Yong
Kim for helpful discussions. We are also grateful to Professor
R. K. Boeckman (University of Rochester) for helpful
suggestions. We especially thank Mr. Lee Alden for his
assistance with electronic interfacing of the wet test meter
to the data acquisition system.
Supporting Information Available
GC method, HPLC method, sources and purities of
material used, validation of methodology, and check on
methodology. This material is available free of charge via
the Internet at http://pubs.acs.org.
Received for review June 26, 2003.
OP034085A
Vol. 7, No. 6, 2003 / Organic Process Research & Development
•
895