Identification of ammonium chloride as an effective promoter of the asymmetric hydrogenation of a β-enamine amide

Article abstract of DOI:10.1021/op050232o

An investigation into the cause of substrate specific hydrogenation performance variability was conducted. A significant and unexpected correlation was observed between apparent pH of a solution of the substrate and rate of conversion and enantioselectivity. This observation led to the examination of low and variable levels of native ammonium chloride in different lots of substrate. The presence of ammonium chloride was found to have a positive effect on reaction rate and enantioselectivity when controlled within a relatively narrow range. Optimal performance was achieved with a mole ratio of 1:1 ammonium chloride to catalyst.

Full text of DOI:10.1021/op050232o

Organic Process Research & Development 2006, 10, 723−726
Identification of Ammonium Chloride as an Effective Promoter of the
Asymmetric Hydrogenation of a â-Enamine Amide
Andrew M. Clausen,* Brianne Dziadul, Kristine L. Cappuccio, Mahmoud Kaba, Cindy Starbuck, Yi Hsiao,† and
Thomas M. Dowling
Merck & Co., Inc., Process Research & DeVelopment (Process Research), P.O. Box 2000,
Rahway, New Jersey 07065, U.S.A.
Abstract:
This examination led us to conclude that the batches of
substrate performing at the upper end of the range contained
a species present in low concentration which promoted both
conversion and desired enantioselectivity. This species was
ultimately identified as ammonium chloride. The presence
of a substoichiometric amount of ammonium chloride relative
to the substrate had a tonic effect on the performance of the
hydrogenation. Similar dependence on halide salt additives
with the same and similar catalyst systems have been
observed before, yet in those cases, the halide salt species
were effective only when added in much higher concentra-
tions.4-6
An investigation into the cause of substrate specific hydrogena-
tion performance variability was conducted. A significant and
unexpected correlation was observed between apparent pH of
a solution of the substrate and rate of conversion and enanti-
oselectivity. This observation led to the examination of low and
variable levels of native ammonium chloride in different lots
of substrate. The presence of ammonium chloride was found
to have a positive effect on reaction rate and enantioselectivity
when controlled within a relatively narrow range. Optimal
performance was achieved with a mole ratio of 1:1 ammonium
chloride to catalyst.
Experimental Section
Hydrogenation of Enamine Amide. Methanol (230 mL)
is charged to a reactor with 25 g of enamine amide. The
resulting slurry is degassed followed by the addition of 0.003
mol equiv (0.046 g) of [(COD)RhCl]2 dimer and 0.0031 mol
equiv (0.104 g) of Josiphos SL-J002-1 ligand (Solvias). The
reaction mixture is heated to 50 °C and hydrogenated at 100
psig (or 115 psi). After 16 h at temperature and under
hydrogen pressure, the batch is cooled to 20 °C and sampled
to analyze for percent conversion and enantiomeric excess.
HPLC Method for Conversion. Percent conversion of
enamine amide to the freebase of sitagliptin phosphate was
analyzed by reverse phase HPLC on an Agilent 1100
according to the following conditions: column, Agilent
Extend C-18, 150 mm × 4.6 mm i.d., 5 µm particles; eluent
A, 1.21 g of TRIS (Sigma), 800 mL of water, 200 mL of
methanol (EM Science), 90 µL of concentrated hydrochloric
acid (Fisher); eluent B, 1.21 g of TRIS (Sigma), 200 mL of
water, 800 mL of methanol (EM Science), 90 µL of
concentrated hydrochloric acid (Fisher); gradient, eluent B,
45% at 0 min to 76% at 8 min held at 76% to 15 min,
reequilibrated at initial conditions for 5 min prior to next
injection.; flow rate 2.0 mL/min; UV detection at 215 nm;
injection volume 5 µL; temperature, 23 °C. Typical retention
times were: sitagliptin, 4.2 min; enamine amide, 5.9 min;
dimer-like impurity, 12.3 min. Sample preparation: pipet 1
mL of hydrogenation stream into a 50-mL volumetric flask
and dilute with eluent A.
Introduction
Sitagliptin phosphate is a dipeptidyl peptidase IV (DPP-
IV) inhibitor which is being studied by Merck & Co., Inc.
for the treatment of diabetes.1 Commercial manufacture of
the drug substance involves two isolated steps (see Scheme
1).2,3 Obtaining the desired compound in high yield relies
heavily on the novel enantioselective hydrogenation of an
unprotected â-enamine amide. Such syntheses, while power-
ful, have not been completely characterized, making predic-
tion of the effect of hydrogenation conditions and additives
difficult.
During the development of this process, it was found that
while individual lots of the starting material enamine amide
gave consistent hydrogenation performance (i.e., rate and
enantioselectivity) there was significant variability across lots.
Results for the hydrogenation ranged from 82% conversion
and 89% enantiomeric excess to 99% conversion and 95%
enantiomeric excess. Process capabilities could tolerate these
fluctuations, but such fluctuations would translate into yield
swings of up to at least 21%. Therefore, an exhaustive
investigation into the source of this variable performance
was initiated.
* To whom correspondence should be addressed. E-mail: andrew_clausen@
merck.com.
† Process Research.
(1) Kim, D.; Wang, L.; Beconi, M.; Eiermann, G. J.; Fisher, M. H.; He, H.;
Hickey, G. J.; Kowalchick, J. E.; Leiting, B.; Lyons, K.; Marsilio, F.;
McCann, M. E.; Patel, R. A.; Petrov, A.; Scapin, G.; Patel, S. B.; Roy, R.
S.; Wu, J. K.; Wyvratt, M. J.; Zhang, B. B.; Zhu, L.; Thornberry, N. A.;
Weber, A. E. J. Med. Chem. 2005, 48, 141-151.
(2) Hsiao, Y.; Rivera, N. R.; Rosner, T.; Krska, S. W.; Njolito, E.; Wang, F.;
Sun, Y.; Armstrong, J. D.; Grabowski, E. J. J.; Tillyer, R. D.; Spindler, F.;
Malan, C. J. Am. Chem. Soc. 2004, 126, 9918-9919.
HPLC Method for Enantiomeric Excess. A normal
phase chiral HPLC method was used to determine enantio-
(4) Fagnou, K.; Lautens, M. Angew. Chem., Int. Ed. 2002, 41, 26-47.
(5) Lautens, M.; Fagnou, K. J. Am. Chem. Soc. 2001, 123, 7170-7171.
(6) Leong, P.; Lautens, M. J. Org. Chem. 2004, 69, 2194-2196.
(3) Cypes, S. H.; Wenslow, R. M., Jr.; Thomas, S. M.; Chen, A. M.; Dorwart,
J. G.; Corte, J. R.; Kaba, M. Org. Process Res. DeV. 2004, 8, 576-582.
10.1021/op050232o CCC: $33.50 © 2006 American Chemical Society
Published on Web 05/17/2006
Vol. 10, No. 4, 2006 / Organic Process Research & Development
•
723

Scheme 1
meric selectivity of the reactions. All analyses were per-
formed on an Agilent 1100 according to the following
conditions: column, Diacel AD-H, 250 mm × 4.6 mm i.d.,
5 µm particle size; eluent, 600 mL of ethanol (Pharmco),
400 mL of hexanes (EM Science), 0.1 mL of diethylamine
(Aldrich), 0.1 mL of water; flow rate, 0.8 mL/min; UV
detection at 268 nm; injection volume, 10 µL; temperature,
35 °C; diluent, 900 mL of methanol (EM Science) and 100
mL of water. Typical retention times were: minor enanti-
omer, 14.5 min; sitagliptin, 17.5 min. Samples were pre-
pared: pipet 1 mL of hydrogenation stream into a 25-mL
volumetric flask and dilute with diluent.
Apparent pH Measurement. Approximately 1 g of
enamine amide sample was dissolved in 10 mL of dry
dimethyl sulfoxide (EM Science). A combined pH electrode
(Fisher) was first calibrated in an aqueous phosphate buffer
at pH 7 (Fisher). After rinsing with water and dimethyl
sulfoxide, the electrode was immersed in the sample until
the reading stabilized. The electrode was washed with water
prior to re-calibration. Each sample was analyzed three times,
and the apparent pH of each solution was taken as the
average of these three analyses. The electrode was calibrated
prior to each measurement.
enamine amide (provided by Merck Process Research and
Merck Chemical Engineering Research and Development and
Lonza) were prepared by dissolving ∼50 mg of each sample
into 10 mL of methanol. The samples were vortexed
thoroughly and filtered through a 1 µm PTFE syringe filter
(Whatman) prior to analysis. All analyses were performed
on an Agilent capillary electrophoresis instrument.
Purification of Enamine Amide. Enamine amide may
be purified by adding approximately 190 mL of degassed
methanol to a 500-mL three-neck flask. To this 23 g of
enamine amide is added. The slurry is stirred at 25 °C for
60 min and then cooled to 2 °C over 1.25 h and held for
2 h at this temperature with continued stirring. The slurry is
then filtered and the cake washed with methanol cooled to
2 °C. Solids, liquors, and washes are collected.
Results and Discussion
Quite often, when process difficulties are encountered and
found to be substrate specific, the investigation will center
on identifying contaminants with negative effects on the
reaction, and initially, this examination followed this ap-
proach. Enamine amide is normally isolated at greater than
99.5% purity, and all lots described here met or exceeded
this purity. Exhaustive chemical and physical analysis of
numerous batches of enamine amide provided no obvious
or distinct differences that would account for the observed
differences in hydrogenation performance.
Capillary Electrophoresis Method for Ammonium. The
concentration of ammonium in enamine amide samples was
determined via a nonaqueous capillary electrophoresis method.7
Standards (0.0017, 0.0033, 0.0083, and 0.017 mg am-
monium/mL) were prepared by dissolving ammonium chlo-
ride (EM Science) in methanol (EM Science). Samples of
The first measurement that showed meaningful dif-
ferentiation among enamine amide lots and appeared to
correlate with hydrogenation performance was apparent pH
(pH*). Enamine amide is not readily soluble in water, and it
(7) Gong, X.; Shen, Y.; Mao, B. J. Liq. Chromatogr. Relat. Technol. 2004, 27,
661-675.
724
•
Vol. 10, No. 4, 2006 / Organic Process Research & Development

Table 1. Results from “crossover” experiment
purified
lot 1
lot 10 with liquors
lot 10 from lot 1 slurry wash
lot 1
% conversion
% ee
95
95
91
91
91
91
95
95
Figure 1. Apparent pH of 10 mg/mL solutions of enamine
amide in dimethyl sulfoxide. Error bars are the standard
deviation obtained from triplicate analyses.
Figure 3. Results of hydrogenation of enamine amide lots
containing various amounts of ammonium chloride. Data points
marked by (b) correspond to percent conversion as measured
by LC (100 × (area of the peak corresponding to enamine
amide)/(sum of the areas for the peaks corresponding to
enamine amide and sitagliptin freebase). Data points marked
by (9) correspond to percent enantiomeric excess as measured
by LC (100 × (area of the peak corresponding to enamine amide
- area of the peak corresponding to the sitagliptin freebase)/
(sum of the areas for the peaks corresponding to the S-
enantiomer of sitagliptin freebase and sitagliptin freebase).
Curves are added solely to illustrate trends and do not represent
a fit of the data.
Figure 2. Apparent pH of enamine amide lots versus the
percent enantiomeric excess resulting from asymmetric hydro-
genation. The solid line is a linear fit of the data. Error bars
are the standard deviation obtained from triplicate analyses.
reaction was due to a poison or a promoter. A portion of
Lot 1 which displayed desirable performance characteristics
was purified via a slurry wash with methanol. The enamine
amide solids were collected from the slurry wash and
hydrogenated. The liquors and washes were collected and
used as the solvent for the hydrogenation of Lot 10 which
in its native state performed poorly in the hydrogenation.
Results of these hydrogenations appear in Table 1. As
expected, the performance of Lot 1 in the hydrogenation
degraded upon purification, but the performance of Lot 10
improved to that of unpurified Lot 1. These data provided
evidence that lack of an unidentified promoter was the cause
of the poorer conversion and enantioselectivity of Lots 10-
18.
With this in mind, we revisited data which we had
collected earlier during in depth chemical and physical
examination of all lots of enamine amide studied. Our
approach in developing this chemistry had been to supply
the hydrogenation step with the most pure enamine amide
feasible. As such, we had monitored, among many other
characteristics, the residual ammonium chloride which had
hydrolyzes, making an aqueous pH measurement impossible;
therefore, measurement of pH* of a rather concentrated
solution of the substrate was made in dimethyl sulfoxide.
Inspection of Figure 1 shows a tenuous but observable
distinction between lots. Lots 1-9 produced solutions with
pH* less than 12. More distinctive is when pH* of a subset
(those batches which were hydrogenated alone, not part of
a mixture) of these data is plotted versus percent enantiomeric
excess (% ee). Figure 2 clearly shows that those batches
which gave a lower solution pH* performed better in this
hydrogenation.
Here we note an important feature of Figure 1. Lots 10-
18 had all been subjected to a purification operation in the
form of a slurry wash, recrystallization, or recrystallization
and carbon treatment. The significance of an additional
purification step led us to two possible conclusions: either
a poison was being concentrated by the purification, or a
substance which promoted the hydrogenation was being
washed out.
A simple “crossover” experiment was completed to assess
whether the stochastic nature of the performance of the
Vol. 10, No. 4, 2006 / Organic Process Research & Development
•
725

To achieve optimum reaction performance, the level of
ammonium chloride in the reaction must be maintained
between 500 and 1500 ppm (0.38-1.14 mol % relative to
enamine amide). This is done by controlling the amount of
ammonium chloride in the enamine amide raw material at a
maximum of 0.1 wt % and adding 0.05 wt % to the reaction.
Such treatment rendered the reaction very reproducible and
no longer dependent on the source of substrate.
Conclusions
We have demonstrated that the performance and consis-
tency of the asymmetric hydrogenation of an unprotected
â-enamine amide was significantly affected by the apparent
pH of the reaction solution. Hsiao et al.2 observed deuterium
was integrated only in the â-position when a model
compound related to the enamine amide discussed here was
hydrogenated with the same catalyst and solvent system in
a deuterium atmosphere. The authors note that this may imply
that tautomerization plays a role in this reaction.2,9,10
The correlation of apparent pH and reaction performance
may be due to positive perturbations of the enamine-imine
tautomerization such that the equilibrium is shifted in a subtle
but positive way. The addition of substoichiometric quantities
of ammonium chloride adjusts the apparent pH of the
reaction solution to a range that gives optimum conversion
and enantioselectivity without over-promoting the formation
of a dimer-like side product. Studies are ongoing to better
understand the mechanism by which ammonium chloride
fosters the desired reaction and to elucidate the effect of
changing the protic source and salt anion.
Figure 4. Dimer levels observed with increasing concentration
of ammonium chloride present in the hydrogenation solution.
Area percent as measured by LC (100 × (area of the peak
corresponding to dimer-like impurity)/(sum of all peaks detected
in the chromatogram which were not present in a blank
injection). The line represents a linear fit to the data.
carried through during enamine amide synthesis.8 Am-
monium chloride had been controlled at maximum 0.1 wt
%; yet since this was the most probable source of a weak
acid “contaminant” in the substrate and thus related to pH*,
we more closely evaluated its level in the individual batches.
Using the nonaqueous capillary electrophoresis method
described above, we measured the level of ammonium in
enamine amide. Assuming that all ammonium was in the
form of the chloride salt, the native presence of this
compound ranged from 60 to 1400 ppm. These values of
native ammonium chloride along with a spike of 0.5 wt %
ammonium chloride into a hydrogenation were then plotted
versus percent conversion and percent enantiomeric excess
(see Figure 3). A very strong correlation exists between the
concentration of ammonium chloride until approximately 500
ppm at which time diminishing improvement is observed.
In fact, Figure 4 shows that if the amount of ammonium
chloride present in the reaction is too high the formation of
a dimer-like side product between enamine amide and
sitagliptin reaches a level at which yield is curtailed.
Overall, the addition of ammonium chloride has improved
yield, cycle time, and consistency of this novel asymmetric
hydrogenation. In a broader scope, there may be some
processes whose reactant-dependent performance, still un-
explained, may be revaluated in the light of identifying an
as yet unknown promoter.
Acknowledgment
We thank Dr. Xiaoyi Gong for his many helpful discus-
sions concerning quantitation of ammonium via nonaqueous
capillary electrophoresis and Dr. Thorsten Rosner for many
helpful discussions concerning the hydrogenation.
(8) Xu, F.; Armstrong, J. D.; Zhou, G. X.; Simmons, B.; Hughes, D.; Ge, Z.;
Grabowski, E. J. J. J. Am. Chem. Soc. 2004, 126, 13002-13009.
(9) Catalytic Asymmetric Synthesis; Wiley-VCH: New York, 2000.
(10) Noyori, R. Asymmetric Catalysis in Organic Synthesis; John Wiley &
Sons: New York, 1994.
Received for review December 1, 2005.
OP050232O
726
•
Vol. 10, No. 4, 2006 / Organic Process Research & Development