Utilizing on- and off-line monitoring tools to follow a kinetic resolution step during flow synthesis

Article abstract of DOI:10.1002/mrc.4494

In situ reaction monitoring tools offer the ability to track the progress of a synthetic reaction in real time to facilitate reaction optimization and provide kinetic/mechanistic insight. Herein, we report the utilization of flow NMR, flow IR, and other off-line spectroscopy tools to monitor the progress of a flow chemistry reaction. The on-line and off-line tools were selected to facilitate the stereoselective kinetic resolution of a key racemic monomer, which lacked a chromophore, making conventional reaction monitoring difficult. Copyright

Full text of DOI:10.1002/mrc.4494

Special issue research article
Received: 15 June 2016
Revised: 12 July 2016
Accepted: 24 July 2016
Published online in Wiley Online Library
(wileyonlinelibrary.com) DOI 10.1002/mrc.4494
Utilizing on- and off-line monitoring tools to
follow a kinetic resolution step during
flow synthesis
Kathleen A. Farley,*† Usa Reilly,† Dennis P. Anderson, Brian P. Boscoe,
Mark W. Bundesmann, David A. Foley, Manjinder S. Lall, Chao Li,
Matthew R. Reese and Jiangli Yan
ABSTRACT In situ reaction monitoring tools offer the ability to track the progress of a synthetic reaction in real time to facilitate
reaction optimization and provide kinetic/mechanistic insight. Herein, we report the utilization of flow NMR, flow IR, and other
off-line spectroscopy tools to monitor the progress of a flow chemistry reaction. The on-line and off-line tools were selected to
facilitate the stereoselective kinetic resolution of a key racemic monomer, which lacked a chromophore, making conventional
reaction monitoring difficult. Copyright © 2016 John Wiley & Sons, Ltd.
Keywords: flow NMR; flow IR; reaction monitoring; 1H NMR; flow synthesis; enzyme; kinetic resolution
probe volumes and have shorter distances from the reaction
mixture to the NMR instrument, thus achieving much lower overall
Introduction
The ability to monitor a chemical reaction effectively is an impor-
tant aspect in developing a robust synthetic process. Discerning
unwanted or competitive side reactions are crucial to understand-
ing the mechanism of the intended reaction. Critical information
can be obtained through real-time reaction analysis or off-line sam-
pling, and these data can be used to determine reaction kinetics or
monitor reaction progress. If the components are stable to the sam-
pling technique employed, then analysis by off-line NMR or mass
spectroscopy offers a straightforward way to study the reaction. In
situ reaction monitoring provides real-time information of the reac-
tion progress, and its non-invasive quality is better suited to track
transient reaction intermediate species, handle fast kinetics, or pro-
vide more comprehensive insight regarding reaction mechanism.
Several on-line reaction monitoring tools are available to help un-
derstand reaction kinetics, speciation, yield, and monitor putative
intermediate formation.[1–3] Each technology has its unique advan-
tages and limitations. Infrared and Raman spectroscopy are popular
process analytical technology (PAT) tools for on-line qualitative or
quantitative measurements.[4,5] Recently, in situ Fourier transform
infrared (FTIR)[6] spectroscopy and in situ Raman spectroscopy[7]
have been developed to achieve high resolution and sensitivity.
On-line or flow NMR spectroscopy is an attractive option for PAT
because it provides detailed structural information of each of the
reaction components. Historically, large reaction volumes were
required with a high field NMR instrument because of the long
transit from the reaction vessel[8] to the instrument and the volume
required in the NMR probe. Conversely, reactions that were minia-
turized in an NMR tube for ease in reaction monitoring often did
not have the ability to acquire realistic stirred batch-like kinetic
information.[9] Recently, several different vendors have introduced
low field benchtop NMR instruments that can be adapted for flow
reaction monitoring.[10,11] These instruments accommodate smaller
volumes when compared with their high field counterparts.
Small chiral functional building blocks are pharmaceutically
interesting monomers that can be incorporated into drug like
molecules. These building blocks often possess poor physical
properties (poor ionization, no chromophore, volatile, etc.) which
makes them more challenging to monitor in real time or by tradi-
tional means. Herein, we describe the recent application of a low
field benchtop NMR instrument[10,11] employed in real time to
monitor a stereoselective esterification of a racemic cis chiral
alcohol precursor. This experimental design was devised to accom-
modate smaller, post flow reaction volumes at a shorter distance
from the flow equipment to the analytical benchtop instruments,
monitor the longer term performance of the immobilized enzyme
used in the flow reaction for agreement with off-line data, and
thereby introduce more holistic analytical method development
approaches to aid larger scale synthesis.
Synthesis and discussion
The synthesis of optically pure pyranone building blocks from read-
ily available starting materials has been published.[12] The authors
describe a Prins cyclization reaction between readily available acet-
aldehyde and 3-buten-1-ol to generate predominantly racemic cis
*
Correspondence to: Kathleen Farley, Medicinal Sciences, Pfizer Worldwide
Research and Development, 224 Eastern Point Road, MS 8220-4448, Groton, CT
06340-5002, USA. E-mail: kathleen.a.farley@pfizer.com
†These a uthors contributed equally to this work
Medicinal Sciences, Pfizer Worldwide Research and Development, Groton, CT,
United States
Magn. Reson. Chem. 2016
Copyright © 2016 John Wiley & Sons, Ltd.

K. A. Farley et al.
starting materials (SMs), 2R,4R, and 2S,4S pyranol isomers (SM1 and
SM2, respectively).[13] This is followed by a subsequent kinetic reso-
lution key step as shown in Scheme 1. The authors devised a batch
cell followed by a benchtop low field 42.5 MHz flow NMR using
commercial 1/16 in. PFA (Norell, Inc. Morganton, N.C., USA) tubing
(Fig. 1). Anhydrous tetrahyrdrofuran (THF) solvent was pumped
through the priming port and pressure transducer of the flow
system at 0.2 ml/min rate to equilibrate the on-line analytical instru-
ments and allow for temperature equilibration prior to the reaction.
A 1 M stock solution containing a mixture of pyranal cis alcohols
[SM1: SM2 (1:1)] and vinyl butyrate (0.45 eq.) in THF was pumped
through the packed enzyme at specified rates (0.25–0.75 ml/min)
and 1H NMR and IR spectra were recorded and processed on-line.
Samples were also collected as 2 ml fractions for off-line analysis
using a high field 400 MHz flow 1H NMR and chiral gas chromato-
graphic (GC) analysis.
resolution process with a commercially available immobilized lipase
™
enzyme (Novozyme
435, Novozymes North America, Inc.
Franklinton, NC, USA) supported on a resin as one option toward
a large scale synthesis. During any kinetic resolution of a racemic
mixture, one desires the relative kinetic rate of acylation by the
enzyme to favor transesterification of one isomer over the other
during the reaction period (i.e. kSM1> kSM2). In practice, up to half
an equivalent of vinyl butyrate (SM3) is available and is fully con-
sumed as an acyl transfer reagent to furnish a mixture of optically
pure cis- methylpyranylbutyrate (P1), unreacted cis alcohol enantio-
mer (SM2), and volatile acetaldehyde (P3) by-product. The isolation
process takes advantage of the differences between alcohol and es-
ter properties such as aqueous/organic solvent solubility, and the
unreacted alcohol can be purged via extraction, or other isolation
methods such as chromatography. Upon batch scale up, concerns
around conversion or chiral purity can quickly arise because of
prolonged enzyme contact or ‘residence time’ resulting in unde-
sired products (P2, SM1). Additional side products may form via
chiral erosion, e.g. hydrolysis, or transesterification of SM2. Other
issues such as enzyme stability, enzyme deactivation, or other scal-
ing parameter effects can also be problematic. For instance, erosion
of enantiomeric excess (e.e.) was observed in earlier batches when
the recommended 2 h batch conditions were extended to 22 h
(97% e.e. to 86% e.e.), suggesting larger scale delivery would have
to be carefully controlled. One option to mitigate scaling risk is to
devise a flow process where all of the parameters (flow rate, concen-
tration, ‘residence time’, temperature, etc.) can be optimized and
fixed on a small scale. This leaves the integrity of the enzyme over
time as the predominant variable, which can impact the chiral purity
or yield of a larger scale continuous flow production. In view of this
monitoring requirement, a proof of concept portable flow system
was designed to track this key step in real time and vary those
parameters that mimic enzyme performance changes over time.
Reaction Monitoring
Benchtop and high field flow NMR
To determine viability of the low field 42.5MHz benchtop flow
NMR instrument to monitor the reaction, a 400 MHz Agilent
NMR using a Protasis microflow probe with 10 μl flow cell was also
connected to the flow stream. For this setup, a 1H-NMR was
recorded every 15s for comparison with the low field data. The
instrument was prepared using THF-d8 to establish an adequate
shim set using the deuterium lock signal, followed by 1H FID shim-
ming of fully protonated THF. Because the reaction concentration
was high enough, no solvent suppression was needed to observe
key reactant and product signals in spectra using a single scan
(Fig. 2A). Flow rates ranging from 50 to 150 μl/min through the
flow probe were tested and signal-to-noise ratios began to dimin-
ish in the 90–100 μl/min range. For the spectra in Fig. 2B, a flow
rate of 100 μl/min was used. As the reaction volume in the flow
cell was being completely replaced in the 15 s timeframe, a 90
degree pulse was utilized for maximum signal-to-noise without
concern for signal saturation.[14] To achieve the optimum 0.5ml/
min flow rate for the enzyme column, a splitter was employed
to link the output from the Omnifit (Diba Industries Inc, Danbury,
CT, USA) column directly to the flow NMR probe. The spectra
collected on the high field instrument not only generated better
resolution than the benchtop NMR but also gave us confidence
that we were not missing any intermediates that the low field
instrument was unable to detect.
Experimental setup
This process was executed using a Vapourtec R2/R4 flow setup with
™
a simple glass Omnifit column packed with Novozyme 435 en-
zyme [248 mg (0.68 ml resin bed)]. The flow synthesis system was
configured with an on-line benchtop IR equipped with a microflow
Scheme 1. Kinetic resolution of racemic cis pyranols.
wileyonlinelibrary.com/journal/mrc
Copyright © 2016 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2016

Utilizing on- and off-line monitoring tools during flow synthesis
Figure 1. Schematic view of the flow synthesis system.
When the analysis was conducted with the benchtop Spinsolve
(Magritek, San Diego, CA, USA) NMR spectrometer, the disappear-
ance of the methyl doublet from SM1/SM2 was monitored as well
as the three doublets of doublets from the vinyl protons of SM3
(Fig. 3). As the reaction progressed, the appearance of a slightly
shifted methyl (P1) was monitored as well as the aldehyde side
product (P3). The kinetic resolution reaction was monitored using
Results
The maximum flow rate applied (without losing enzyme perfor-
mance) was studied in real time to optimize the flow rate and resi-
dence time of the SM stock solution in contact with the enzyme. As
the flow rate was gradually increased to 0.5ml/min, the peak height
of the products continued to increase and then leveled off (Fig. 4).
Deviation from the optimized flow process was apparent when the
flow rate increased from 0.5 to 0.75 ml/min. The percentage conver-
sion dropped, as indicated by the dip in the peak height for the on-
line flow NMR (Fig. 4) and presence of starting material (SM3). A
similar effect was also observed in the flow IR. When the reaction
flow rate was returned to 0.5 ml/min, the conversion % increased
to the previous level, and the appropriate starting materials (SM1
and SM3) were completely consumed. Finally, off-line monitoring
via sampling the reaction mixture and analysis by 1H-NMR and GC
also support this conclusion (Table 1). Overall, the on-line NMR
and IR are in good agreement with the more sensitive off-line
NMR results and could be used as a standalone technique to
monitor this transformation. Following these results, the reaction
was run at 0.5ml/min.
1
a single scan with a 90° H pulse in protonated THF every 15 s. A
56 scan 1H NMR spectrum was collected several times during the
first few hours to monitor for additional side reactions. Diagnosti-
cally, for SM1 and SM2, the doublet at 0.9 ppm was monitored
and in SM3 three doublet of doublets were monitored at 4.3 and
7.0ppm. For P1 (and potentially P2 products), the shifted methyl
at 0.8ppm was monitored, whereas for P3, the aldehyde resonance
at 9.3ppm was observed.
The products were monitored by the appearance of a slightly
shifted methyl and the formation of an aldehyde (Fig. 3D). The peak
heights of the methyl and aldehyde resonances were plotted ver-
sus time and are shown in Fig. 4.
In order to understand the stability of the enzyme, the reaction
mixture was monitored for a total of 8.5h: 5.0h the first day and
3.5 h the following day. At the end of the first day, the Novozyme
Flow IR
The flow IR probe uses ATR-FTIR technology to detect IR absorption
of organic molecules in solution in the mid-IR region (4000 to
650 cmÀ1). The C = C bond of vinyl butyrate at 1648 cm-1 and the
carbonyl regions of the ester at 1735 cm-1 for P1 were monitored
(Fig. 5).
™
435 column was flushed with stock THF for 16h to keep it
solvated. The second day on-line and off-line data indicated that
the packed enzyme is stable for over 24h, as shown in Day 2 results
in Table 1.
The maximum flow rate was determined to be 0.5ml/min under
this experimental set up and was shown not to lose catalytic
efficiency over the 24 h period. Overall, 27 g racemic alcohol along
with 0.45 eq. of the vinyl ester in THF was processed through the
catalyst, and afforded 12.8 g of the desired butyrate after isolation
with high e.e.% (98.5%). The low field benchtop flow NMR and flow
IR were effective PATs of the kinetic resolution reaction at the 1 M
concentration. Using these tools for reactions at lower concentra-
tions would need an increased number of scans. The low field
Off-line GC analysis
A chiral GC method was developed and used to measure the ste-
reochemical outcome of the process as well as the conversion of
the desired product relative to the staring material (Fig. 6). The peak
areas were converted to concentration by calibration against each
of the pure reaction components for final percentage conversion
determination.
Magn. Reson. Chem. 2016
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

K. A. Farley et al.
Figure 2. Off-line high field flow 1H NMR: (A) The spectrum of the reaction mixture as recorded with the Protasis microflow probe at 400 MHz with
characteristic downfield signals of SM3, P1/P2, and P3 signals labeled. (B) The progression of single scan spectra recorded every 15 s to track spectral changes.
Spinsolve instrument was also not equipped for controlling the
temperature, so development of temperature capabilities would
be required for reactions that require monitoring outside of ambi-
ent temperature.
be reliably performed in flow, (ii) monitor the efficiency/stability of
the enzyme over time, and (iii) compare on and off-line monitoring
for agreement across several complementary analytical tools. Real
time analysis of the reaction mixture stream enabled the efficient
optimization of the flow rate required for optimal product conver-
sion and enzyme stability over the reaction period. This outcome
could be achieved from a single experiment, with real time data fa-
cilitating decisions and adjustments to be made without the need
of post experimental data analysis.
Summary
Flow reaction monitoring tools can offer real time and quality con-
trol insight to enable smoother scale up processes. We have dem-
onstrated the ability to utilize flow NMR technology with on- and
off-line capabilities to monitor a multigram kinetic resolution of a
racemic precursor lacking a chromophore using flow technology
in non-deuterated solvents. The experiment objectives were de-
signed to (i) demonstrate an immobilized enzymatic process could
Experimental
1
On-line H NMR spectra were recorded on a Magritek (Magritek,
San Diego, CA, USA) 42.5MHz Spinsolve instrument equipped with
wileyonlinelibrary.com/journal/mrc
Copyright © 2016 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2016

Utilizing on- and off-line monitoring tools during flow synthesis
Figure 3. The 42.5 MHz 1H spectra of (A) non-deuterated THF solvent, (B) SM1/SM2, (C) SM3, and (D) P1/SM2/P3.
a 1 ml flow cell using non-deuterated THF as a solvent. Spectra of
each starting material at a concentration of 1 M was collected in
non-deuterated THF using 64 complex points and 32–64 scans prior
to the start of the reaction. The instrument was shimmed on water
at the beginning of each day. The reaction monitoring was
conducted using a single scan spectrum with an acquisition time
of 15 s and without solvent suppression at room temperature.
Spectral data were processed using MestreNova Mnova (Mestrelab
Research Escondido, CA, USA) reaction monitoring software.
Flow NMR spectra were also recorded on an Agilent 400MHz
VNMRS (Agilent Technologies Santa Clara, CA, USA) spectrometer
using a 10 μl Protasis flow probe at 298K. A flow rate of
100 μl/min was used to allow for total volume replacement in the
flow cell, and to minimize signal-to-noise loss. Single scan 1H spec-
tra, using 90° pulses and 32K complex points, were recorded every
15s without solvent suppression in protonated THF. The beginning
concentration of reagents was approximately 1 M.
Off-line NMR spectra to determine conversion were recorded on
a Bruker 600MHz (Bruker BioSpin Corporation, Billerica, MA, USA)
spectrometer using a 5 mm TCI cryoprobe. Reaction product
Figure 4. Real time on-line monitoring of product conversion using low
field bench-top flow NMR (42.5 MHz, Magritek).
Figure 5. IR spectra and on-line monitoring of product conversion at steady state (real time).
Magn. Reson. Chem. 2016
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

K. A. Farley et al.
Figure 6. (A) GC Method development showing resolution of racemic SM1, SM2 and racemic P1,P2 and (B) offline reaction monitoring of the auto scaled
chiral GC chromatograms of each fraction collected. SM1 was completely consumed and not observed.
The Chiral GC/FID analysis was performed using an Agilent
Table 1. In situ monitoring of conversion and e.e.%
6890 N GC with FID, equipped with the Restek Rt-bDEXcst (Restek
Reaction
time (h)
Flow rate
(ml/min)
Conversion %
e.e.%
Corporation, U.S., Bellefonte, PA, USA) 30 m x 0.25 mm x 0.25 μm
column. The carrier gas was helium at a flow rate of 1.5 ml/min.
The injector temperature was 250 °C using an injection split ratio ra-
tio of 50:1. The FID detector used a temperature of 300 °C, a hydro-
gen flow of 40 ml/min, and an air flow of 400 ml/min with a helium
makeup gas at 30 ml/min.
On-line On-line IR Off-line Off-line
NMR
NMR
GC
Day 1 0.1
0.25
0.5
100
100
92
100
97
100
99
93
94
92
83
96
98.5
98.4
98.8
98.5
98.6
98.6
97.7
0.7
0.9
1.1
4.4
4.6
5.0
0.65
0.5
91
100
88
90
0.65
0.75
0.75
87
Acknowledgements
88
86
We are grateful to acknowledge Andrew Coy and John Trail from
Magritek for their technical support with the Spinsolve NMR
instrument and for useful discussions on data processing. We
would also like to thank the Biocatalysis Center of Emphasis at Pfizer
who provided useful suggestions and input on the use of
immobilized enzymes.
89
NA*
Column flush for overnight (16 h)
Day 2 0.1
0.5
0.5
0.5
0.5
90
100
95
NA*
99
92
100
98
NA*
98.8
98.7
NA*
0.3
0.4
3.5
95
90
92
92
*Solution was too dilute to measure
References
[1] M. A. Bernstein, M. Stefinovic, C. J. Sleigh. Magn. Reson. Chem. 2007, 45,
564–571.
[2] D. N. Patel, L. Li, C. L. Kee, X. Ge, M. Y. Low, H. L. Koh. J. Pharm. Biomed.
Anal. 2014, 87, 176–190.
samples were collected at different flow rates, dried, and dissolved
in d8-THF for analysis.
A ReactIR iC10 (Mettler Toledo, Mettler Toledo Columbus, OH,
USA) equipped with a DiComp ATR DS microflow cell (50 ml internal
volume) was incorporated into the flow stream post the
immobilized enzyme column to monitor the continuous flow
synthesis outcome. The data was acquired and processed with iC
IR 4.3 software (Mettler Toledo Columbus, OH, USA). The IR data
was acquired using the following parameters: selected spectral
region of 400 to 650 wavenumbers, a resolution of 8 wavenumbers,
and a scans rate of 256 scans/min. The peaks at 1648 (C = C from
vinyl butyrate) and 1735 cmÀ1 (C= O from the desired
methylpyranylbutyrate) were used to generate the reaction trends.
The absorbance was converted to the concentration via calibration
with the pure reaction components.
[3] E. Skibsted, S. B. Engelsen. Encycl. Spectrosc. Spectrom. 2010, 3 (2nd ed),
2651–2661.
[4] J. G. Rosas, M. Blanco, J. M. Gonzalez, M. Alcala. J. Pharm. Sci. 2011, 100
(10), 4442–4451.
[5] E. Tamburini, M. G. Marchetti, P. Pedrini. Sensors 2014, 14(10),
18941–18959.
[6] A. Pintar, J. Batista, J. Levec. Analyst 2002, 127(11), 1535–1540.
[7] T. A. Hamlin, N. E. Leadbeater. Beilstein J. Org. Chem. 2013, 9,
1843–1852.
[8] D. A. Foley, E. Bez, A. Codina, K. L. Colson, M. Fey, R. Krull, D. Pirola,
M. Zell, B. L. Marquez. Anal. Chem. 2014, 86(24), 12008–12013.
[9] D. A. Foley, A. L. Dunn, M. T. Zell. Magn. Reson. Chem. 2016, 54(6),
451–456.
[10] M. V. Silva Elipe, R. R. Milburn. Magn. Reson. Chem. 2016, 54(6),
437–443.
wileyonlinelibrary.com/journal/mrc
Copyright © 2016 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2016

Utilizing on- and off-line monitoring tools during flow synthesis
[11] E. Danieli, J. Perlo, A. L. L. Duchateau, G. K. M. Verzijl, V. M. Litvinov,
B. Blumich, F. Casanova. Chemphyschem 2014, 15(14), 3060–3066.
[12] K. R. Anderson, S. L. G. Atkinson, T. Fujiwara, M. E. Giles, T. Matsumoto,
E. Merifield, J. T. Singleton, T. Saito, T. Sotoguchi, J. A. Tornos, E. L. Way.
OPRD 2010, 14, 58–71.
[13] J. S. Yadav, S. Reddy, B. V. Subba Reddy, G. G. K. S Narayana Kumar,
S. Aravind. Synthesis 2008, 3, 395–400.
[14] F. Dalitz, M. Maiwald, G. Guthausen. Chem. Sci. Eng. 2012, 75, 318–326.
Magn. Reson. Chem. 2016
Copyright © 2016 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc