- 1. In Process Monitoring of Polymorphic Form Conversion by
Raman Spectroscopy and Turbidity Measurements Susan Barnes, Jason
Gillian, Ann Diederich, Delphi Burton, Darryl Ertl ForewordThe aim
of pharmaceutical development is to design superior active
pharmaceutical ingredients (APIs) with robust manufacturing
processes that consistently deliver product of pre-defined quality
and performance. Quality cannot be tested into products; it should
be built-in or should be by design.1 The information and knowledge
gained from research and development studies as well as
manufacturing experience, provide scientific understanding to
enable determination of the design space, product specifications,
and manufacturing controls.2 Process Analytical Technologies (PAT)
are systems for designing, analyzing, and controlling
pharmaceutical manufacturing processes allowing timely measurements
of critical quality and performance attributes of raw and
in-process materials. Incorporating PAT into quality by design
(QbD) encompasses the use of scientifically based process
optimization, appropriate sensor technologies, statistical tools
(chemometrics), feedback process control strategies and knowledge
management tools to ensure production of final high quality
products at manufacturing.IntroductionCrystallization of APIs is a
common unit operation used to stabilise and purify process
intermediates and finished products. This procedure allows firm
control of crystal size distribution, morphology and polymorphic
form, which are often critical quality attributes of a drug
substance. 3 Taking a QbD approach to producing the final particle
forming step is central to form control and critical to ensure
product stability, safety, efficacy and performance at secondary
manufacturing.Historically, form identification has been achieved
using off-line techniques such as XRPD, NMR, DSC and FTIR. 3, 4
Off-line methods provide no continuous information on the process,
often involve sampling delays and can alter the processing history.
More recently process analytical technologies (PAT) such as near
infrared (NIR) spectroscopy 5, 6and Raman spectroscopy7-9 have been
realized as techniques for characterization of crystal form and
conversion kinetics. Tools, such as FBRM, are also being utilized
for monitoring of particle size distribution. 10-12This article
details the application of fiber-optic Raman spectroscopy for
in-situ measurement of a form transformation during a final
crystallization and isolation step. Data were acquired in order to
develop an understanding of the design space for the process over a
range of operating conditions. In alignment with the goals of the
PAT framework and the quality by design tenet, this knowledge has
enabled scale-up of a robust crystallization that consistently
ensures a pre-defined form and product quality.
2. Case StudyThe desired form of the API is an anhydrate (Form
A). Research has been conducted to develop a cooled seeded
crystallization intended to isolate the anhydrate form of the API
in a methanol-water solvent system. During this work, a monohydrate
and two distinct methanol solvate forms (M1 and M2) were also
identified. Distinct differences in crystal size and morphology
were observed between the four forms (Figure 1). These differences
can affect processability at secondary manufacturing and product
bioavailability.Each crystal form was found to have a very
distinctive powder X-ray diffraction pattern and off-line FTIR and
Raman spectra. Preliminary work indicated that the final form
recovered from this process was governed by the methanol-water
solvent ratio, batch isolation temperature and hold time after
seeding. Induction of a form conversion from the anhydrate to the
methanol solvate was seen in batches with extended holds over a
range of isolation temperatures. Consequently, determination of the
maximum hold time before occurrence of an undesirable form
conversion was paramount to the design of a robust process. Figure
1: Optical microscope images of the monohydrous, anhydrous and
methanolsolvate forms of the API In-situ Raman spectroscopy was
utilized to develop a phase diagram for identification of the most
stable form over a wide range of operating conditions. Other
additional experiments involving in-situ monitoring were designed
to determine form and induction time as a function of isolation
temperature, cooling rate and solvent ratio. All in-situ data was
supported by optical microscopy, XRPD and DATR.One major advantage
of utilizing Raman for this application was the low sensitivity of
the technique to water. Spectra of slurries of all four crystal
forms showed very distinct features with good separation from the
solvent bands, making Raman an excellent qualitative technique for
form discrimination. In-situ data were acquired from slurries of
the API in methanol-water with an 18 long, diameter short focus
immersion optic, interfaced with a port in the top of a 1 L JLR.
Initial data analysis was conducted by simple integration of bands
associated with each form, allowing mapping of the transformation
kinetics and end point determination. 3. Figure 2: MCR analysis of
in-situ Raman data acquired from conversion of the anhydrate to the
methanol solvate form of an API in methanol at 25 o CAB80
8000070Anhydrate, 25 C70000 PC 1 PC 2 Peak area60000 6050000
Integrated peal area (arb) 5040000 Integrated area (au) PC (arb) 40
arb units30000 CMethanolate 1, 25C 30 20000 1510 10000 205 0
010-10000 -50 -20000-100500 1000 1500 2000
-10-30000-1511001150120012501300Time (minutes) Time (minutes) Raman
Shift (1/cm)Figure 2A shows a waterfall plot of in-situ Raman data
acquired from the anhydrate slurried in methanol at 25 o C. The
objective of the experiment was to determine an induction time for
conversion of the batch to the solvate during an extended hold at a
25 o C isolation temperature.The waterfall plot, presented as a
function of hold time, shows notable changes in the spectral
features in the fingerprint region (1250 1080 cm-1). The integrated
area of the feature at 1148 cm-1, ascribed to the first methanolate
form (M1), was used to map the form conversion over time (Figure
2B). Data were acquired over a 2 day period and showed a form
conversion induction time of 13.5 hours and a total conversion time
of 10 hours. A subsequent approach to analyzing the spectral data,
which moves away from relying on the presence of isolated bands to
profile kinetics, is the application of Multivariate Curve
Resolution (MCR).13, 14 Figure 2B is an overlay of the integrated
peak area plot with the profiles of first two MCR components
calculated from analysis of the spectral region (1250 1080 cm-1).
Data were pre-processed by baseline correction and calculation of
the 1st derivative. Excellent agreement can be observed between the
peak area profile and the component profiles. Analysis of the
estimated spectral profiles (Figure 2C) showed strong similarities
with the 1st derivative spectra of the anhydrate and M1
respectively which gives confidence that the correct profiles were
extracted from the data.A further range of experiments were
conducted to determine the effect of isolation temperature and hold
time on final form. Figure 3 presents results from four such
experiments in a 4:1 methanol-water solvent system. The anhydrate
was slurried and held for extended periods of time Methanol-water
at four different isolation temperatures (0, 2, 10 and 20 o C).
Figure 3: Form conversion induction time for as a function of
isolation temperature 4. Raman data from each experiment were used
to identify form, induction time and total conversion time (Shown
in Table). The data showed that lowering the temperature had the
effect of significantly reducing the time at which the slurries
could be held before isolation. Raman data showed that the first
methanolate form was the more stable form at 20 o C, whereas M2 was
more stable between 0-10 o C. In-situ monitoring proved to be an
invaluable tool for acquiring continuous process data and removed
the need for laborious off-line testing over long periods of time
(up to 25 hours). Real-time identification of form by Raman was
confirmed by off-line analysis of the isolated material. Figure 4:
Phase diagram showing thermodynamic stability of each form of the
API as afunction of solvent composition and slurry temperature
Figure 4 is an experimental phase diagram showing the thermodynamic
stability of each form of the API as a function of solvent
composition and slurry temperature. Full development of the phase
diagram led to selection of an operating region where the desired
form is the most stable allowing development of a cooled seeded
crystallization to reproducibly produce the desired form. Combined
with kinetic data on the form change induction times over a range
of temperatures, the data set allowed selection of an acceptable
temperature for material to be isolated following the
crystallization. This information is invaluable, especially at
scale-up, where hold times before material isolation can become
much longer than those typically seen in the laboratory. Scale-up
to Pilot Plant 5. Although shown to be a key tool for in-situ
analysis of form on the laboratory scale, Raman spectroscopy is
currently not available for analysis of the crystallization process
at the pilot plant used for scale up of this particular process.
Since in-situ analysis of this process on scale was desired because
of concerns over the scale-up effect on induction time, an
alternative PAT technique was assessed and pursued.Figure 5 shows
results from the application of Raman spectroscopy and turbidity
measurements to monitor a cooled, seeded crystallization of the
anhydrate in 4:1, methanol-water (Figure 5A). Further data were
collected to monitor the form change of the anhydrate to the
methanol solvate on extended hold at an isolation temperature of 0
o C. Whilst Raman is sensitive to the change in the molecular
structure of the API on form transformation, turbidity data is
sensitive to alteration in light scattering properties of the
slurry as result of the change in particle size, morphology and
distribution during conversion. The profiles from both data sets
determined that after an induction time of 9 hours the second
methanolate was formed with a total conversion time of 8.5 hours.
In the case of the Raman data, the integrated area of feature at
1200 cm-1 associated with the structure of M2 (Figure 5B) was used
to profile the conversion kinetics. Excellent agreement was seen
between the profiles of both data sets. Final form and complete
conversion was confirmed by XRPD, FTIR and optical microscopy
(Figure 5C). Figure 5: In-situ Raman and Turbidity data from
analysis of a seeded cooledcrystallization and subsequent form
conversion on hold at isolation temperature Although not a
molecular specific technique for form identification the
sensitivity of the turbidity measurement makes it a powerful
technique for in-situ detection of form transformation, especially
when coupled with offline spectroscopic analysis. In-situ turbidity
measurements were implemented as part of a 50 L pilot plant
campaign to follow the crystallization of the API in the selected
solvent system and to monitor for form transformation before
isolation. Several batches on 50 L scale were run to test the edges
of the design space using extended hold periods to determine the
maximum induction times on scale. Induction times at isolation
temperature were shown to be comparable to those determined at
laboratory scale which provided promising data for further scale up
of the process to pilot plant. 6. SummaryIn summary, in-situ Raman
spectroscopy is an effective technique for identification of
polymorphic form and form conversion kinetics. In-line Raman was
used extensively on the laboratory scale for development of the
phase diagram the most stable form of the API as a function of
solvent composition and slurry temperature. In-situ data
acquisition reduced the requirement for off-line sampling which was
time consuming, required sample preparation and ran the risk of
altering the processing history of the material. On- scale, in-situ
turbidity measurements were an effective tool for detection of form
transformation during hold at batch isolation temperature.
Turbidity is a technique which is simple and cost effective to
implement on scale. Although turbidity is an inferential
measurement, combined with one off-line sampling for confirmation
of final form, it has been demonstrated to be a powerful technology
for detection of form transformation in this
system.AcknowledgementsThe Authors would like to acknowledge our
colleagues Duncan Thompson and Thomas Thurston for their assistance
with the MCR data analysis aspect of the work presented here. Thank
you also to Charles Goss, Gregory Gervasio and the staff of the
Process Engineering group at Upper providence for assistance with
implementation of in-situ turbidity for measurement for form
conversion on scale-up.References 1.Pharmaceutical cGMPs for the
21st Century A Risk Based Approach, U.S Department ofHealth and
Human Services, U.S Food and Drug Administration, September
2004.2.Guidance for Industry Q8 Pharmaceutical Development, U.S
Department of Health andHuman Services, U.S Food and Drug
Administration, May 2006.3.Braatz, R, D., Advanced Control of
Crystallization Processes, Annual Reviews in Control26 (2002)
87-994.Yu, L, X., Lionberger R, A., Raw, A, S., D'Costa, R., Wu,
H., and Hussain A, S.,Applications of Process Analytical Technology
to Crystallization Processes, Adv. DrugDelivery Rev. Vol. 56,
349369.5.Fevotte, G., Calas, J., Puel, F., and Hoff, C.,
Applications of NIR Spectroscopy toMonitoring and Analyzing the
Solid State during Industrial Crystallization
Processes,International Journal of Pharmaceutics, Vol. 273,
159-169, 20046.Kobayashi, R., FujiMaki, Y., Ukita, T., and Hiyama,
Y., Monitoring of Solvent-MediatedPolymorphic Transitions Using
in-Situ Analysis Tools, Organic Process Research andDevelopment,
Vol. 10, 1219-1226, 20067.Wang, F., Watcher, J.A., Antosz, F.J.,
and Berglund, K.A., "An Investigation of Solvent-Mediated
Polymorphic Transformation of Progesterone Using In Situ Raman
Spectroscopy,"Organic Process Research & Development, Vol. 4,
No. 5, 20008.Agarwal, P., and Berglund, K. A.,In-situ Monitoring of
Calcium Carbonate Polymorphsduring Batch Crystallization in the
Presence of Polymeric Additives Using RamanSpectroscopy. Crystal
Growth and Design, Vol. 3, No. 5, 941-946, 2003 7. 9. OBrien, L.
E., Timmins, P., Williams, A.C., and York, P., Use of in situ
Ft-Raman Spectroscopy to Study the Kinetics of the Transformation
of Carbamazepine Polymorphs, Journal of Pharmaceutical and
Biomedical Analysis, Vol. 36, No. 2, 335-340, 200410.Doki, N.,
Seki, H., Takano, K., Asatani, H., Yokota, M., and Kubota, N.,
Process Control of Seeded Batch Cooling Crystallization of the
Metastable -Form Glycine Using an In-Situ ATR-FTIR Spectrometer and
an In-Situ FBRM Particle Counter, Crystal Growth & Design, Vol.
4, 5, 949 -953, 200411.OSullivan, B., Barrett, P., Hsiao, G., Carr,
A., and Glennon, B., In-situ Monitoring of Polymorphic Transitions
Organic Process Research and Development, Vol. 7, 977-982,
200312.Barrett, P., Smith, B., Worlitschek, J., Bracken, V.,
O'Sullivan, B., and O'Grady, D., A Review of the Use of Process
Analytical Technology for the Understanding and Optimization of
Production Batch Crystallization Processes, Organic Process
Research and Development, Vol. 9, 3, 348 -355, 200513.Cuesta
Sanchez, F., Khots, M.S., and Massart. D.L., Algorithm for the
assessment of peak purity in liquid chromatography with
photodiode-array detection, Anal. Chim. Acta, Vol. 285, 181192,
199414.Tauler, R., and Casassas, E., Spectroscopic resolution of
macromolecular complexes using factor analysis: Cu(II)
-polyethyleneimine system, Chemom. Intell. Lab. Syst.,Vol. 14, 305
317, 1992Author BiographiesDr Susan Barnes is a Principal Scientist
at GlaxoSmithKline within the Process Analytical Technologies and
chemometrics group. She received her BS in Chemistry and PhD in
Mechanical Engineering from the University of Bradford in the UK.
Prior to Joining GSK, she was a Guest Research Scientist in the
Combinatorial Methods Group in the Polymers Division of the
National Institute of Standards and Technologies. Her research
interests include the implementation of in-situ spectroscopic
techniques for process understanding and control. She has authored
over 15 peer reviewed publications and conference proceedings.Dr
Jason Gillian is a Principal Scientist at GlaxoSmithKline within
the Particle Sciences Group in Chemical Development. Prior to
joining GSK, he was a Senior Engineer within Merck Manufacturing
Division for the commissioning, control and optimization of new
processes in organic process development. Dr Gillian received his
BS in Chemical Engineering from Virginia Tech, and was awarded his
MS and PhD in Chemical Engineering from University of Virginia.Ann
Diederich is an Investigator at GlaxoSmithKline within the Particle
Sciences group in Chemical Development. She received her Bachelor
degree in Chemistry from Ohio State University and her MS degree in
Organic Chemistry from Texas A&M University. After spending the
initial 9 years with the company in the Synthetic Chemistry, with
an emphasis on process development and scale-up, Ann has 8. been
specializing in Particle Sciences for the last 8 years. Ann has 6
papers and 6 patents accredited to her name.Delphilia Burton is an
Engineer at GlaxoSmithKline within the Process Engineering Group in
Chemical Development. She received her BS and MS in Chemical and
Biochemical Engineering from University of Maryland, Baltimore
County, where her research primarily focused on small scale
bioseparations, and down stream protein processing. Darryl Ertl is
the manager of the Process Analytical Technology and Chemometrics
group at GlaxoSmithKline in Chemical Development. Prior to joining
GSK he worked at Bristol-Myers Squibb where he was responsible for
implementing a world wide initiative for raw material
identification using NIR spectroscopy and at Eastman Kodak where he
applied numerous in-situ technologies to manufacturing processes
for process control. Darryl received his BS degree in Chemistry
from the University of Brockport.