Determination of chiral purity using vibrational circular dichroism (VCD)

1. Official Publication of the Society for Applied SpectroscopyOfficial Publication of the Society for Applied Spectroscopy
62/9
SEPTEMBER 2008
ISSN: 0003-7028
Determination of a Low-Level Percent Enantiomer of a
Compound with No Ultraviolet Chromophore Using
Vibrational Circular Dichroism (VCD): Enantiomeric Purity
by VCD of a Compound with Three Chiral Centers
Laila Kott,a*
Jelena Petrovic,a
Dean Phelps,a
Robert Roginski,b
Jared Schubertc
a
Takeda Pharmaceuticals International Co., Analytical Development Small Molecules, 40 Landsdowne St., Cambridge, MA 02139
USA
b
Eigenvector Research Inc., 3905 West Eaglerock Dr., Wenatchee, WA 98801 USA
c
Boston University, School of Law, 765 Commonwealth Ave., Boston, MA 02215 USA

2. Determination of a Low-Level Percent Enantiomer of a
Compound with No Ultraviolet Chromophore Using
Vibrational Circular Dichroism (VCD): Enantiomeric Purity
by VCD of a Compound with Three Chiral Centers
Laila Kott,a*
Jelena Petrovic,a
Dean Phelps,a
Robert Roginski,b
Jared Schubertc
a
Takeda Pharmaceuticals International Co., Analytical Development Small Molecules, 40 Landsdowne St., Cambridge, MA 02139
USA
b
Eigenvector Research Inc., 3905 West Eaglerock Dr., Wenatchee, WA 98801 USA
c
Boston University, School of Law, 765 Commonwealth Ave., Boston, MA 02215 USA
The chiral configuration of three of the four chiral centers in the
investigational drug MLN4924 is locked by an intermediate
(1S,2S,4R)-4-amino-2-(hydroxymethyl)cyclopentanol (designated
as INT1a). The intermediate INT1a is a key component to the
molecule, but its multiple chiral centers and lack of chromophore
make it challenging to analyze for chiral purity of the desired
enantiomer when it is contaminated with a small amount of its
undesired enantiomer. Vibrational circular dichroism (VCD) is a
technique that uses the infrared (IR) regions of the electromagnetic
spectrum and as INT1a contains IR active groups, we considered
using VCD to determine the chiral purity of INT1a. Since the VCD
spectra of enantiomers are of equal intensity and opposite in sign,
it was possible to construct calibration curves to detect the
presence of low levels of this compound in the presence of its
enantiomer. By normalizing the observed intensities of the VCD
signals with the observed IR spectra, a partial least squares model
was constructed having a root mean squared error of cross
validation of 0.46% absolute over a range of 97 to 99.9% pure
enantiomer (or 97–99.8% enantiomeric excess). This work demon-
strates that VCD can be used for the low-level detection of a
compound in the presence of its enantiomer and thus eliminates
the need for an ultraviolet chromophore and chromatographic
separation of the two enantiomers.
Index Headings: Multiple chiral centers; Mid-infrared; Quantitation;
Vibrational circular dichroism; VCD; Enantiomer; a-pinene; Partial
least squares; PLS.
INTRODUCTION
The investigational oncology drug MLN4924, whose
mechanism of action is the inhibition of the Nedd8
activating enzyme (NAE), is a complex molecule with
four chiral centers. Three of these centers are locked by
the intermediate INT1a. As seen in Fig. 1, INT1a has no
ultraviolet (UV) chromophore, which is the standard
detection used in liquid chromatographic (HPLC) sepa-
ration techniques for chiral purity assays. The molecule
is small enough and possibly volatile enough for chiral
gas chromatographic (GC) analysis, but amine functional
groups make choosing a column for GC analysis difficult
because these compounds tend to tail and interact with
GC columns in such a way as to make integration and
quantitation difficult.1
There are also fewer choices of
chiral columns for analysis by GC.
Analysis of this intermediate is possible by more
traditional approaches, such as derivatization with
Marfey’s Reagent2
(1-fluoro-2,4-dinitrophenyl-5-L-ala-
nine amide, FDAA), followed by HPLC analysis. How-
ever, as often happens with derivatization reactions, the
validation of such methods is challenging due to
complicated sample preparation and proving complete
derivatization.3,4
Since this compound is nonchromo-
phoric, chiral, and is a nitrogen-containing compound,
HPLC analysis using a polarimeter, a circular dichroism
(CD) detector, and a chemiluminescence nitrogen
detector (HPLC-CLND) were tried. Chromatography was
possible with the CLND, but no peaks were observed
using the two chiral detectors which covered a wave-
length range of 230 to 700 nm. In an effort to push to
lower limits, an alternative technique was considered.
A search for other possible methods for determining
the chiral purity of INT1a led to a lesser-known
spectroscopic technique based on CD in the infrared
region. This technique is known as vibrational circular
dichroism (VCD) and most commonly makes use of the
mid-IR region from 2000 to 800 cmÀ1
(5000 to 12 500
nm).5,6
Note: The entire IR region could theoretically be
used and spans 14 000 to 800 cmÀ1
($715 to 12 500 nm).
To employ VCD, several experimental parameters had
to be considered. A suitable solvent that is transparent in
the region of interest is required, and to ensure that the
VCD peaks are detectable above the noise, the IR bands
of the solution to be analyzed should be at a concentra-
tion that will result in absorbance between 0.2 to 0.8.7
Similar to other spectroscopic techniques, an increase in
collection time yields higher signal-to-noise ratios and a
more robust dataset.8
Determination of the optimal
spectral region and collection time that is required for
each compound to obtain the desired spectral quality
depends on the functional groups and the flexibility of the
molecule. In general, the less flexible the molecule, the
shorter the collection time that is required for a
satisfactory signal-to-noise ratio. Therefore, due to its
ring structure and limited conformational freedom, INT1a
was a good candidate for VCD analysis.
Primer on Vibrational Circular Dichroism (VCD).
Vibrational circular dichroism, like the CD measure-
ments in the UV (also known as electronic CD or ECD), is
Received 18 April 2013; accepted 4 April 2014.
* Author to whom correspondence should be sent. E-mail: Laila.Kott@
takeda.com.
DOI: 10.1366/13-07112
1108 Volume 68, Number 10, 2014 APPLIED SPECTROSCOPY0003-7028/14/6810-1108/0
Q 2014 Society for Applied Spectroscopy

3. a spectroscopic technique which detects differences in
absorption of left and right circularly polarized IR and
near-IR light, where vibrational transitions in molecules
are observed.9
Only chiral molecules display CD/VCD
activity.9
DA ¼ AL À AR ð1Þ
Equation 1 shows how VCD is a measure of the
difference in absorption (DA) of left (AL) and right (AR)
circularly polarized IR light. The observed signals are
typically very small, roughly four orders of magnitude
smaller than the parent IR intensities.8
The current principal use of VCD in the pharmaceutical
industry is for the determination of absolute configura-
tion of chiral active pharmaceutical ingredients, which is
accomplished by comparison of a VCD measurement to
an ab initio calculation of a selected enantiomer.8
However, VCD is an absorption measurement, and its
intensity is linearly proportional to concentration and
path-length in accordance to Beer’s law, but unlike its
parent IR, the VCD intensity is also linearly proportional
to enantiomeric purity of the sample. This feature of VCD
allows direct measurement of the percentage of desired
enantiomer without a need for separation. Simulations of
real-time reaction monitoring have shown that it is
possible to follow and quantitate the conversion of one
enantiomer to the other.10
Since VCD is fundamentally an IR measurement, with
its abundance of bands, it lends itself to a chemometrics
analysis. Chemometrics allows for the use of a defined
wavelength range of interest, takes into consideration
many more data points, and incorporates these points
along with the variability (error) in the data into the
analysis. Chemometric analysis is especially useful
during manufacturing (for in-process control measure-
ments) as it can account for cell variability, analyst
variability, temperature fluctuations, and instrument
variability. Chemometrics use, therefore, leads to a
more robust and sensitive approach to analysis of
complex VCD datasets.
MATERIALS AND METHODS
Infrared (IR) and Vibrational Circular Dichroism
(VCD) Measurements for a-Pinene Study. The IR and
VCD spectra of a-pinene were measured on a BioTools
ChiralIR-2X spectrometer (Jupiter, FL) equipped with
DualPEM (two photoelastic modulators) and a Syncrocell
rotating sample cell holder. The liquid R- and S-pinene
samples (Sigma-Aldrich, St. Louis, MO) used were of
several different assay and enantiomeric purities and
were mixed volumetrically at different R : S ratios, up to
12.5% R. These solutions were placed in a BaF2 cell with
a path length of 50 lm and run both holding the sample
stage stationary and rotating it counterclockwise at
3 RPM. Data were acquired at a resolution of 4 cmÀ1
,
for 20 min. No baseline corrections were performed, and
chemometric analysis was carried out using SOLO
software (Eigenvector Research Inc., Wenatchee, WA).
Different calibration curves prepared on different days
by separate analysts were collected and all the data
combined so as to have a sufficiently large calibration
set and to account for day-to-day and person-to-person
variability within the chemometric model.
Ab Initio Calculations for Theoretical Spectra of
INT1a, its Enantiomer, and Diastereomers. For all of
the stereoisomers, an initial search of conformational
space was performed with Molecular Operating Environ-
ment (Molecular Operating Environment (MOE) 2012.10,
Chemical Computing Group) software using molecular
mechanics (MM), stochastic search, and a 7 kcal/mol
FIG. 1. Structure of (a) MLN4924, (b) the intermediate INT1a where the chirality is locked, and (c) R- and S-a-pinene.
APPLIED SPECTROSCOPY 1109

4. energy window. All the conformers output from MM were
submitted for density functional theory (DFT) minimiza-
tion. After DFT minimization, Boltzmann distribution, and
redundancy check, four conformers were within 2 kcal/
mol from the lowest energy conformer for INT1a, while
two conformers were within 2 kcal/mol from the lowest
energy conformer for INT1c, INT1e, and INT1g.
The output from MOE was inputted into Gaussian09 to
perform spectral calculations for each conformer using
B3LYP/6-31G basis set. The conformers’ spectra were
then combined into a final spectrum using Boltzmann
distributions to generate a theoretical VCD and IR
spectrum for each enantiomer and diastereomer. Com-
pare VOA version 1.1 by BioTools was used to generate
the final spectra. The calculations were performed in
vacuo.
Infrared (IR) and Vibrational Circular Dichroism
(VCD) Measurements for INT1a Study. The IR and
VCD spectra were measured on a BioTools ChiralIR-2X
spectrometers equipped with DualPEM. The samples
were placed in a BaF2 cell with a path length of 100 lm.
For the first calibration set, data were acquired at a
resolution of 8 cmÀ1
, for 10 h. In order to determine the
limit of detection for the undesired enantiomer of INT1a,
a calibration curve was constructed at 100 mg/mL total
compound, ranging from 97.0:3.0 to 99.9:0.1 (desired :
undesired) enantiomer ratio of INT1a in methanol-d4
(Sigma-Aldrich, St. Louis, MO). Chemometric analysis
was carried out using PLS_Toolbox (Eigenvector Re-
search Inc.) running on Matlab. Two sets of samples (ee1
and ee2) were made on two separate days to account for
variability in the technique.
A second curve was constructed at 125 mg/mL total
compound, ranging from 80:20 to 100:0 (desired : unde-
sired) enantiomer ratio of INT1a in methanol-d4 using a
different instrument to show reproducibility. Data were
acquired at a resolution of 4 cmÀ1
, for 24 h. Data were
normalized against their molar concentrations, and all
chemometric analyses were carried out using SOLO
software (Eigenvector Research Inc.).
RESULTS AND DISCUSSION
Considering that CD spectra have been used for the
quantitation of enantiomers in the UV,11
it was thought
that the same could be applied to VCD for the
determination of the amount of an enantiomeric impurity.
The VCD spectra of enantiomers are opposite in sign;
therefore, when added together, the mixture of enantio-
mers result in a spectrum that has a lower intensity than
the spectrum of a single pure enantiomer. In the case of
a racemic mixture, the equal and opposite-signed
enantiomer spectra cancel each other out, resulting in
no VCD spectrum.12
Providing the total concentration of
all the compounds remains the same; however, the
resulting IR spectrum should show constant intensities
for the mixture of enantiomers.
Percent Enantiomer Study with a-Pinene. Since
investigational compounds such as INT1a and its
enantiomer are not readily available in large quantities,
a-pinene (see Fig. 1) was chosen for the first study, as
both the R and the S are readily available at known
purities. As this compound was a liquid, samples were
prepared by making volumetric dilutions of the R
enantiomer with the S enantiomer. Each concentration
was analyzed both while the sample was stationary and
while rotating. Three sets of samples were made fresh
daily and analyzed on different days (using both the
stationary and rotating cell configurations) by different
analysts, to ensure that there was sufficient statistical
variability in the data for a successful chemometric
analysis. All concentrations were corrected for overall
and entiomeric purity, for both the R- and the S-pinene.
To construct the chemometric model, six a-pinene
datasets were combined (three measured with a
stationary cell, three with a rotating cell). The IR peaks
of a-pinene at 1265 and 1285 cmÀ1
showed a change in
intensity with increasing R enantiomer. This is indicative
of a non-chiral impurity since it is apparent in the IR
spectrum, and as such, the corresponding peak in the
VCD spectrum was removed prior to chemometric
analysis. The data range used for chemometric analysis
was from 1345 to 940 cmÀ1
, with the peak from 1285 to
1235 cmÀ1
excluded (see Fig. 2). For a-pinene, mean
centering preprocessing was used. No further process-
ing was required as the pinene samples were neat
liquids with the same physical characteristics, and since
there was no interference from water vapor in the IR
spectra, there was no need to correct for it during the
data analysis of these experiments.
A partial least squares (PLS) model was built with a
total of two latent variables. Venetian blind (with seven
data splits) cross validation was performed (48 samples
in the dataset). The root mean squared error of
calibration (RMSEC) was 0.62%, while the root mean
squared error of cross validation (RMSECV) was 0.69%.
This indicates that the enantiomeric purity of S-a-pinene
could be determined to 60.69%. Figure 3 shows the
results of the PLS model for predicted-to-measured
(RMSEC) and for cross-validation predicted-to-measured
(RMSECV) enantiomeric purity.
The success of this model can be shown by using it to
predict values of unknowns. We collected two more sets
of data and used them as a prediction set. We applied
FIG. 2. The VCD spectral range used for modeling the a-pinene data
(1345 to 940 cmÀ1
)
1110 Volume 68, Number 10, 2014

5. FIG. 3. Results of the PLS model to predict a-pinene enantiomeric purity. The top plot (a) shows predicted versus measured, and the bottom (b)
cross validated prediction versus measured.
FIG. 4. A plot showing the predicted values of a test dataset as determined from the chemometric model for a-pinene.
APPLIED SPECTROSCOPY 1111

6. our model to them, and Fig. 4 shows that the test values
(i.e. the unknowns) can be predicted using the model
developed above.
These results indicated that quantitation to low
enantiomeric levels by VCD is possible, and the work
shifted from our model compound to 1NT1a.
Theoretical Spectra of INT1a, its Enantiomer, and
Diastereomers. Prior to starting the enantiomeric purity
study, we had to verify that we could differentiate
between the enantiomers and diastereomers; therefore,
the theoretical infrared (IR) and VCD spectra were
generated. A comparison of the VCD spectra of the
diastereomers is shown in Fig. 5. In theory, we could
choose bands or a small spectral range, specific to the
enantiomeric pair in question to determine chiral purity.
We could also quantify any other diastereomers off of
their unique bands. However, in this case, since we were
dealing with known standards, we simply verified that
the measured VCD spectrum matched the theoretical to
ensure that we had the proper components.
The success of developing a model for a-pinene and
confirmation that the enantiomers and diastereomers
are spectroscopically unique led to the investigation to
determine the lower limit to enantiomeric detectability
for INT1a by VCD. A chemometric analysis approach was
also applied to the INT1a data. If the levels of detection
were found to be low enough, theoretically this non-
chromatographic method could be used for either
FIG. 5. Theoretical IR and VCD spectra of INT1a, INT1c (RRR), INT1e (RSR), and INT1g (SRR). The IR spectra show that they are all the same;
however, each displays a unique VCD spectra.
1112 Volume 68, Number 10, 2014

7. (offline) in-process testing or release testing of the INT1a
intermediate without any derivatization.
The compounds in this series were salts and were,
therefore, insoluble in most of the typical VCD solvents
like CCl4 and CDCl3 and had only a 0.2 mg/mL solubility
in methylene chloride. To determine a low level of
quantitation for INT1a, we chose methanol as our solvent
for two reasons. Firstly, it was the solvent that our
compound was most soluble in, and it was available in
deuterated form at high purity. Secondly, it removed the
exchangeable protons from the spectrum, thereby
simplifying the final spectrum. Since all solutions were
made up in methanol-d4, any possible solvent effects
would be present in all solutions, normalizing any
effects. As the purpose of these experiments was to
push the level of quantitation as low as possible, not to
characterize the molecule, the criteria for choice of
solvent were (1) high concentration and (2) having an
appropriate section of the VCD spectrum that would be
amenable to quantitation by chemometrics.
In the first round, two sets of calibration curves (ee1
and ee2) were made and analyzed on two different days
to ensure that there was sufficient statistical variability in
the data for a successful chemometric analysis. The
process of analysis involved the following steps:
(1) Finding the best way to compensate for atmospheric
water vapor in the IR spectra.
(2) Learning and accounting for the baseline fluctua-
tions in VCD spectra.
(3) Evaluating the best spectral regions for analysis.
(4) Normalizing the IR and VCD spectra for molar
concentration and path length fluctuations. Vibra-
tional circular dichroism is a product of molar
concentration multiplied by the percentage of
enantiomer and path length, so correction is a
critical step.13
A unique feature of this VCD instrument was the
collection of IR spectra simultaneously with VCD spectra.
This allows for the VCD spectra to be corrected for minor
changes in concentration.
Based on the spectra collected, the best region for IR
normalization was 1234–1585 cmÀ1
and for VCD analysis
was 1234–1474 cmÀ1
. This region corresponds to IR
bands associated with some C–H, C–O, and C–N modes.
Since peaks in this region show up in the VCD spectra as
well, these modes are considered to be representative of
the bonds connected to the chiral centers of interest,
thereby making this region appropriate for analyses. The
FIG. 6. (a) IR spectra (1430–1630 cmÀ1
) and (b) VCD spectra (1240–1590
cmÀ1
) for datasets ee1 and ee2. The vertical dashed line in both plots is
coincident with the artifact observed in some of the VCD spectra at 1512
cmÀ1
. Note that this feature is absent in the IR spectra. The sharp
features in the IR spectra between 1480–1580 cmÀ1
are attributed to
water vapor; negative features in the ee2 set indicate that the
background measurement contained more water vapor than the
subsequent spectral measurements.
FIG. 7. Results of the PLS model to predict INT1a enantiomeric purity
for the first calibration set. The top plot (a) shows predicted versus
measured, and the bottom (b) cross validated prediction versus
measured.
APPLIED SPECTROSCOPY 1113

8. region for VCD is smaller due to the fact that the
instruments used exhibited an artifact in the noise
spectra (even when no cell is present), which then
manifested itself as small peak in the VCD spectra.
Figure 6 shows that the artifact peak in the VCD spectrum
has no equivalent in the IR spectrum. If it was a true peak
due to the sample, it would be visible in both.14
After removing water vapor absorption lines in the IR
spectrum and applying baseline correction in the VCD
spectra using Savitzky–Golay derivatives, a partial least
squares (PLS) model was built with a total of two latent
variables. The root mean squared error of calibration
(RMSEC) was 0.24%, while the root mean squared error
of cross validation (RMSECV) was 0.46%. Leave-one-out
cross-validation was performed, which is appropriate
given the small size of the dataset (10 samples in the
dataset). Figure 7 shows the results of the PLS model to
predict enantiomeric purity for both the RMSEC and
RMSECV.
The higher value of RMSECV compared to RMSEC
directly indicates that the latter is biased by the presence
of all samples in the calculation of the error. Given that
this analysis is multivariate in nature, and there are no
well-established standard methodologies to extract
traditional univariate figures of merit such as limit of
detection (LoD) and limit of quantitation (LoQ) from this
method, we will focus our attention on the best available
metric for future performance: RMSECV. In this scenario,
the value of RMSECV can be thought of as a degree of
uncertainty around the predicted analytical value, a type
of one standard deviation (1r).15
Using the root mean squared error of cross validation
approach, we have an uncertainty of 0.46% on any future
values determined from this calibration set. Since
mixtures with low concentrations of the undesired
enantiomer were used to develop this model (from 3%
down to 0.1%), we could theoretically expect that we
could detect down to 0.1% undesired enantiomer, with an
error of 60.46%, or more accurately a purity of 99.9%
with an error of 60.46%. A level of detection with that
confidence is appropriate for in-process testing.
Applying what was learned from the first INT1a
calibration set, a second round of data was collected.
To improve on the signal-to-noise in the VCD spectra, the
data was collected at slightly higher concentrations at a
resolution of 4 cmÀ1
and for a total of 24 h. Based on the
spectra collected, the best region for VCD analysis was
1275–1500 cmÀ1
.
Since the cell compartment is open to the environ-
ment, instead of using the IR spectra for normalization
and correcting for atmospheric vapor, normalization was
done against the molar concentrations, as these solu-
tions were prepared in the testing laboratory and the
concentrations were known. This was done though
preprocessing in SOLO. Once corrected, a partial least
squares (PLS) model was built with a total of two latent
variables. The RMSEC was 1.1% while the RMSECV was
2.0%. Leave-one-out cross-validation was performed,
which is appropriate given the small size of the dataset
(eight samples in the dataset).
Due to the availability of material in the second round
of experiments with INT1a, only a few datasets were
collected, and as such, the model was not as well
defined or as rugged as the first set. The trend, however,
is the same that, as we generate more data, the model
can be pushed to be more reliable at lower concentra-
tions of enantiomer.
CONCLUSIONS
This study was an initial attempt to estimate the
sensitivity of VCD for determination of the percentage of
undesired enantiomer. For a-pinene, our model showed
that we could differentiate and predict levels of enantio-
meric impurities down to about 0.69%. When considering
that some of the S-pinene used in this study had 2.7% R
enantiomer in it, a level of uncertainty of 0.69% appears
sufficient for this compound. It should be noted that one
of the biggest challenges during these experiments was
obtaining pure pinene and accurate chiral purity data.
This became a serious consideration while running the
experiments and also highlighted another possible use
for this type of analysis—raw material testing.
For the first experiments with INT1a, the RMSECV
obtained was 0.46% for the PLS model (two latent
variables), and for the second set of experiments the
RMSECV was 2.0% for the PLS model (two latent
variables); however, our dataset was smaller than the
first set of experiments, and as such, the model
generated could not be as robust. These values are
comparable to an adequate sensitivity level for an in-
process test (target limit was 0.5 to 1.0%); however, to
replace the chromatographic testing, it would be
desirable to get closer to a level of 0.05 to 0.1%. If we
can get down to these lower levels, this technique could
also be used as a release test, replacing chromatogra-
phy completely. Given that the scope of this study did not
include much in the way of parameter optimization, it is
not unreasonable to expect an improvement in perfor-
mance that would make this measurement more com-
parable to the gold standard.
The data from all the INT1a experiments could theoret-
ically be combined, but as we were interested in the
reproducibility from site to site, we did not combine them
at this time. However, for this technique to work outside of
the research lab, calibration models must be transferable,
and the differences in instrumentation have to be negated.
This is a topic we will be researching further.
A final note on instrumentation: The current VCD
optical configuration is open to the environment, and the
nitrogen sweep is not sufficient to remove any atmo-
spheric effects (i.e., water vapor). It would be our
recommendation that, if instrument manufacturers are
interested in their equipment being used for quantitation,
as presented here, that an option for a closed purged cell
be available.
ACKNOWLEDGMENTS
The authors would like to thank Ashley McCarron and Ian Armitage
for synthesizing the compounds and Fred Hicks and Beth Piro for
sharing their experiences with these molecules. The authors would
also like to thank Larry Nafie, Rina Dukor, and the team at BioTools for
their help in executing the experiments and introducing us to the
technique.
1. L. Kott, H.M. Chen. ‘‘Experimental Considerations in Headspace
Gas Chromatography’’. Pharm. Tech. 2010. 34(5): 76-79.
1114 Volume 68, Number 10, 2014

9. 2. R. Bhushan, H. Bru¨ ckner. ‘‘Marfey’s Reagent for Chiral Amino Acid
Analysis: A Review’’. Amino Acids. 2004. 27(3-4): 231-247.
3. G. Lunn, L.C. Hellwig. Handbook of Derivatization Reactions for
HPLC. New York: John Wiley and Sons, 1998. Pp. 639-644.
4. N.D. Danielson, P.A. Gallagher, J.J. Bao. ‘‘Chemical Reagents and
Derivatization Procedures in Drug Analysis’’. In: R.A. Meyers,
editor. Encyclopedia of Analytical Chemistry. Chichester: John
Wiley and Sons, 2000.
5. L.A. Nafie. Vibrational Optical Activity: Principles and Applications.
Chichester: John Wiley and Sons, 2011. Pp. 234-236.
6. S. Abbate, G. Longhi, E. Castiglioni. ‘‘Near-Infrared Vibrational
Circular Dichroism: NIR-VCD’’. In: N. Berova, P.L. Polavarapu, K.
Nakanishi, R.W. Woody, editors. Comprehensive Chiroptical Spec-
troscopy: Instrumentation, Methodologies, and Theoretical Simula-
tions, Volume 1. New York: John Wiley and Sons, 2012. Pp. 247-273.
7. Y. He, B. Wang, R.K. Dukor, L.A. Nafie. ‘‘Determination of Absolute
Configuration of Chiral Molecules Using Vibrational Optical
Activity: A Review’’. Appl. Spectrosc. 2011. 65(7): 699-723.
8. R.K. Dukor, L.A. Nafie. ‘‘Vibrational Optical Activity of Pharmaceu-
ticals and Biomolecules’’. In: Encyclopedia of Analytical Chemistry,
Online. John Wiley and Sons, 2006.
9. P.J. Stephens, F.J. Devlin, J.R. Cheeseman. VCD Spectroscopy for
Organic Chemists. Boca Raton: CRC Press, 2012. Pp. 20-24.
10. C. Guo, R.D. Shah, R.K. Dukor, X. Cao, T.B. Freedman, L. Nafie.
‘‘Determination of Enantiomeric Excess in Samples of Chiral
Molecules Using Fourier Transform Vibrational Circular Dichroism
Spectroscopy: Simulation of Real-Time Reaction Monitoring’’. Anal.
Chem. 2004. 76(23): 6956-6966.
11. H.G. Brittain. ‘‘Applications of Chiroptical Spectroscopy in the
Characterization of Compounds Having Pharmaceutical Impor-
tance’’. In: N. Berova, K. Nakanishi, R.W. Woody, editors. Circular
Dichroism: Principles and Applications. New York: John Wiley and
Sons;, 2000. Pp. 819–844.
12. G.K. Webster, L. Kott. ‘‘Method Development for Pharmaceutical
Chiral Chromatography’’. In: S. Ahuja, S. Scypiniski, editors.
Handbook of Modern Pharmaceutical Analysis. Burlington, MA:
Academic Press, 2011. P. 276.
13. R.K. Dukor, C. Guo, L.A. Nafie. ‘‘Reaction Monitoring of Chiral
Molecules Using Fourier Transform Infrared Vibrational Circular
Dichroism Spectroscopy’’. US Patent: US 7,378,283 B2. Filed 2003.
Issued 2008.
14. G. Yang, Y. Xu. ‘‘Vibrational Circular Dichroism Spectroscopy of
Chiral Molecules’’. Top. Curr. Chem. 2011. 298: 189-236.
15. K. Kjeldahl, R. Bro. ‘‘Some Common Misunderstandings in Chemo-
metrics’’. J. Chemom. 2010. 24(7–8): 558-564.
APPLIED SPECTROSCOPY 1115