In Situ Monitoring and Modeling of the Solvent-Mediated
Polymorphic Transformation of L-Glutamic Acid
Jochen Scho¨ll, Davide Bonalumi, Lars Vicum, and Marco Mazzotti*
ETH Swiss Federal Institute of Technology Zurich, Institute of Process Engineering,
Sonneggstrasse 5, CH-8092 Zurich, Switzerland
Martin Mu¨ller
NoVartis Pharma AG, Chemical & Analytical DeVelopment, CH-4002, Basel
ReceiVed July 18, 2005; ReVised Manuscript ReceiVed January 24, 2006
ABSTRACT: In this paper, we present the application of four different in situ analytical techniques to monitor the solvent-mediated
polymorphic transformation of L-glutamic acid. Focused beam reflectance measurement (FBRM) and particle vision and measurement
(PVM) have been used to track the chord length and morphology of the crystals over the course of the transformation. The polymorphic
forms present have been monitored using Raman spectroscopy, while attenuated total reflection Fourier transform infrared (ATR-
FTIR) spectroscopy has been used to measure the liquid-phase concentration profile. The combination of the different in situ data
was used to identify the fundamental phenomena of nucleation and growth that govern the process. Moreover, the measurement data
were combined with a mathematical model based on population balance equations and the fundamental equations describing the
kinetics of nucleation and growth of both polymorphs. This combination allowed for the estimation of the characteristic nucleation
and growth rates of the two polymorphic forms, while the dissolution process of the metastable polymorph was estimated using a
Sherwood correlation. Finally, the experimental results obtained with different initial conditions and their simulation allowed for the
validation of the population balance model and for a deeper understanding of the transformation process.
1. Introduction
Polymorphism is the ability of a substance to crystallize in
different crystal modifications, each of them having the same
chemical structure but different stacking of atoms or molecules
in the crystal lattice. Several physical properties are affected
by polymorphism, such as stability, solubility, melting temper-
ature, hygroscopicity, chemical reactivity, and rate of dissolution.
Because of differences in stability, the solid-state transformation
of one polymorph into another can occur.1 The solvent-mediated
polymorphic transformation is induced by the difference in
solubility of the polymorphs. In recent years, there has been a
growing interest in monitoring and controlling the solid-phase
transformation during crystallization and precipitation processes,
particularly in the pharmaceutical and fine chemical industries.2,3
Different offline analytical techniques have been used to
characterize the polymorphs obtained during crystallization,
namely, X-ray powder diffraction (XRPD), differential scanning
calorimetry (DSC), and thermal gravimetric analysis (TGA).
These techniques require sampling and are therefore not suitable
for the in situ monitoring of formation and transformation of
polymorphs during the process. Recently, Raman spectroscopy
has been successfully applied to monitor the polymorphic
transformation in situ by different research groups.4-6 Moreover,
in situ Raman spectroscopy has been combined with a popula-
tion balance model to estimate the polymorph transformation
kinetics of L-glutamic acid.7 In this work, we combined the
different in situ analytical techniques mentioned above to
monitor the liquid and the solid phase during the polymorphic
transformation of the metastable R form of L-glutamic acid into
the stable â form at 45 °C. The experimental data were then
used together with a population balance model accounting for
the fundamental phenomena to estimate the relevant parameters
in the relationships for crystal nucleation and growth. Finally,
the accuracy of the model was assessed.
2. Experimental Section
2.1. Materials and Methods. The stable â polymorph of L-glutamic
acid (>99%, Sigma-Aldrich, Buchs, Switzerland) and deionized water
were used for all experiments. To produce pure metastable R-form, an
aqueous L-glutamic acid solution with a concentration of 48 g/kg of
solvent was cooled from 80 to 45 °C at a rate of 1.5 K/min. Nucleation
started only after 45 °C had been reached, and soon after nucleation
the suspension was filtered, washed, and dried. The polymorphic form
was verified using X-ray powder diffraction and scanning electron
microscopy. SEM samples were sputtered with 2 nm of platinum in
high vacuum before being recorded with a Leo 1530 microscope (Zeiss/
LEO, Oberkochen, Germany). For both polymorphs, Figures 1 and 2
show the characteristic XRD patterns and scanning electron micro-
graphs, respectively. It is worth noting that crystals of the R form are
* To whom correspondence should be addressed. Phone: +41-44-
6322456. Fax: +41-44-6321141. E-mail: marco.mazzotti@ipe.mavt.ethz.ch.
Figure 1. X-ray powder diffraction patterns of the metastable R and
the stable â form of L-glutamic acid.
CRYSTAL
GROWTH
& DESIGN
2006
VOL.6,NO.4
881-891
10.1021/cg0503402 CCC: $33.50 © 2006 American Chemical Society
Published on Web 03/03/2006
Administrator
高亮
Administrator
高亮
prismatic, whereas those of the â form are needlelike. An alternative
way to form R crystals is by pH-shift precipitation as discussed
elsewhere.8
2.2. Batch Crystallizer Setup. A jacketed 2-L borosilicate glass
reactor with an inner diameter of 150 mm from LTS (Basel,
Switzerland) was used as a crystallizer in all experiments. The 4-blade
glass stirrer had 45° inclined blades (with a diameter of 70 mm), was
positioned 15 mm above the bottom of the reactor, and was operated
at 300 rpm. The temperature in the crystallizer was controlled using a
Pt 100 and a CC 240 WL-3 thermostat from Huber (Offenburg,
Germany). Figure 3 shows the experimental setup together with the
four in situ measurement instruments. To optimize the quality of the
data recorded in situ and to minimize clogging of the probe windows,
the position of the immersion probes was chosen in the zone of high
fluid velocities, i.e., close to the bottom and near the reactor walls.
2.3. Concentration Measurement using ATR-FTIR Spectroscopy.
Attenuated total reflection Fourier transform infrared (ATR-FTIR)
spectroscopy can be successfully applied to monitor the liquid phase
during crystallization processes.8-11 The ATR probe allows for the
acquisition of liquid-phase IR spectra in the presence of solid material
due to the low penetration depth of the IR beam, which is generally on
the order of 1 ím only.12 All ATR-FTIR measurements in this work
were carried out using a ReactIR 4000 system from Mettler-Toledo
(Schwerzenbach, Switzerland) equipped with a 11.75” DiComp im-
mersion probe and a diamond as ATR crystal.
2.3.1. ATR-FTIR Calibration. The law of Lambert-Beer is the
fundamental equation describing the relationship between the incident
and transmitted radiation intensities in vibrational spectroscopy. It can
be expressed as
where A is the absorbance, a is the absorption coefficient, b is the
effective path length, c is the sample concentration, I0 is the radiation
emitted by the spectrometer, and I is the transmitted radiation of the
sample.13 All IR spectra were recorded with the ReactIR 3.0 data
acquisition software, and each spectra consisted of 128 scans at a
resolution of 4 cm-1. To determine the concentration of L-glutamic
acid in water, several solutions of known concentration have been
prepared at 45 °C, and the height of the carboxylate stretching band at
1408 cm-1 from the baseline at 1380 cm-1 in the IR spectra has been
used to obtain the calibration as shown in Figure 4.
2.3.2. Solubility Measurement of R and â Polymorphs. The
solubility of both polymorphs in water has been determined in a range
from 20 to 60 °C in slurry experiments using ATR-FTIR spectroscopy.
Since IR spectra are temperature dependent, calibrations as described
in the previous section have been performed at different temperatures.
To determine the solubility of both polymorphs, crystals of the
corresponding polymorph have been added in excess to an undersatu-
rated solution, and the liquid-phase concentration has been measured
Figure 2. Scanning electron micrographs of the metastable R and the stable â form of L-glutamic acid.
Figure 3. Schematic of the 2-L batch crystallizer combining the
different in situ process analytical technologies of FBRM, PVM, and
ATR-FTIR and Raman spectroscopy.
Figure 4. Calibration of the ATR-FTIR absorbance of the carboxylate
stretching band at 1408 cm-1 against known L-glutamic acid concentra-
tions in water at 45 °C.
A ) abc ) log10
I0
I (1)
882 Crystal Growth & Design, Vol. 6, No. 4, 2006 Scho¨ll et al.
Administrator
高亮
until no further concentration change was observed. This procedure
was repeated at the temperatures where the system was calibrated
before. Figure 5 shows the measured aqueous solubilities of the two
polymorphs, together with solubility data found in the literature.14 The
experimental error was found to be below 1.5% for all measurements,
and the measured solubilities of R and â agree well with the literature
values. It is worth noting that above a temperature of 50 °C it was not
possible to obtain reliable solubility values for the metastable R form
since the transformation to the stable â was faster than the attainment
of equilibrium. The solubility of R is higher than the one of â over the
whole temperature range, thus indicating that the system is monotropic.
At 45 °C, the solubility of the R and â form in water is 21.7 and 17
g/kg of solvent, respectively; these values will be used in the model
equations.
2.4. Polymorph Characterization Using in Situ Raman Spec-
troscopy. Several analytical methods have been used to characterize
and quantify polymorphic crystalline material offline, i.e., X-ray
diffraction, solid-state NMR, vibrational spectroscopy, and thermal
analysis.15 However, for this purpose only two techniques have been
applied so far in situ during crystallization: X-ray diffraction and Raman
spectroscopy.4,16 During this work, we used a RA 400 Raman
spectrometer (Mettler-Toledo, Greifensee, Switzerland), equipped with
a 250 mW frequency stabilized laser diode at 785 nm and a
thermoelectrically cooled Raman detector. The spectrometer is con-
nected via a fiber optic to a 5/8” ball type immersion Reaction
RamanProbe from Inphotonics (Norwood, USA) with a wetted length
of 330 mm that allows for spectra acquisition within a spectral range
from 200 to 3900 cm-1 Stokes at a resolution of 3.6 cm-1. To record
spectra with a satisfying peak-to-noise ratio, the sample exposure time
was set to 45 s, and 10 scans were averaged to give one Raman spectra.
Raman powder spectra of pure R and â l-glutamic acid crystals are
shown in Figure 6, where similarities and differences can be seen.
2.4.1. Polymorphic Content. Methods for the measurement of the
polymorphic content during crystallization have been proposed in the
literature.4,5,17 All methods are based on ex situ calibration using
prepared polymorph mixtures, either in suspension4,17 or as dry powder.5
No influence of the crystal size on the calibrated signal has been noted
or reported.
In this work, we have tried to calibrate the polymorphic content of
L-glutamic acid in a similar way. In a first step, a series of calibration
samples with different mass ratios of solid R and â particles were
suspended in a saturated solution of L-glutamic acid at 45 °C. Then,
the polymorphic composition was calculated by using one characteristic
peak for R and another one for â and by employing the following
equation:
where xR is the mass fraction of suspended R crystals, and AR and Aâ
are the baseline integrated area of the corresponding peaks in the Raman
spectrum. Since not only the polymorphic concentration but also the
crystal size of the suspended crystals changes during the course of the
transformation process, we have repeated the calibration with the same
â fraction, but a second population constituted coarser R crystals. Figure
7 shows the results for the two calibration sets where 1004 and 941
cm-1 were chosen as characteristic peaks for R and â, respectively, as
well as the volume-weighted chord length distribution of the R
populations measured by focused beam reflectance measurement
(FBRM). Although the trends are similar, the differences between the
two calibrations are as large as 10 wt %. Similar results have been
obtained whatever the choice of the characteristic peak for R and â in
the spectrum. Therefore, these results indicate that the measured Raman
signal is a function not only of the polymorphic content but also of the
crystal size and of the crystal size distribution. As a consequence, the
Raman data measured in this work using the same characteristic peaks
as in Figure 7 were not calibrated but scaled to fulfill the material
balance as discussed in section 3.1 and therefore used only qualitatively.
2.5. Monitoring the Particle Size Using FBRM. FBRM allows
for in situ measurements of the chord length distribution (CLD) of the
particle population even at high solid concentrations. It has been used
for various purposes in the field of crystallization, such as the
measurement of the solubility and of the metastable zone width,18 the
estimation of crystallization kinetics,8,19 or the control of fines.20 The
principle of the measurement technique is described in detail else-
where.21 In all experiments, we used a laboratory scale FBRM 600L
from Lasentec (Redmond, USA).
2.6. Optical Image Acquisition. In situ high-resolution images have
been taken throughout the processes using a Lasentec PVM 800 probe
(Redmond, USA). Six independent laser sources inside the PVM probe
illuminate a fixed area at the probe tip. The backscattered light is
focused on a CCD camera producing an image of 1760 � 1320 ím
with a resolution of approximately 10 ím. The images yield time-
resolved qualitative information about the average particle size and
Figure 5. Solubility of R and â L-glutamic acid in water as a function
of temperature. b and [ represent the solubility of R and â measured
by ATR-FTIR spectroscopy in this work, respectively. The lines are
guides for the eye. For comparison, the gravimetrically determined
solubility data of R and â L-glutamic acid reported by Sakata14 is shown
as O and ], respectively.
xR ) f ( ARAR + Aâ) (2)
Figure 6. Raman powder spectra of the metastable R and the stable â
form of L-glutamic acid.
Figure 7. Experimental results for the calibration of two different R
populations in suspension with â crystals. The volume-weighted CLD
of both R populations is shown in the upper left corner. The population
of â crystals was the same for both calibrations.
Polymorphic Transformation of L-Glutamic Acid Crystal Growth & Design, Vol. 6, No. 4, 2006 883
Administrator
高亮
polymorph content since the two polymorphs of L-glutamic acid exhibit
different shapes (see Figure 2).
2.7. Unseeded Experiments. At the beginning of all unseeded
experiments, L-glutamic acid was dissolved in deionized water at 80
°C to yield the desired initial concentration c0. After complete
dissolution of the solid material, the solution was cooled at a rate of
1.5 K/min to a temperature of 45 °C, which was then held constant
throughout the process. Upon reaching 45 °C, the data acquisition of
all in situ monitoring tools was started. It is worth noting that no
nucleation could be observed during the cooling phase; hence, the
crystallization process that follows can be considered to be isothermal.
2.8. Seeded Experiments. Seeded experiments have been performed
by preparing a saturated solution with respect to the R polymorph at
45 °C, i.e., at a concentration of 21.7 g/kg of solvent. A sufficient
amount of R crystals has been produced as described in section 2. Two
different seed populations have been prepared: a fine population of
seed crystals obtained by milling R crystals that passed through a 250
ím sieve, and a coarser population consisting of the sieve fraction
between 250 and 500 ím. The mass of seed crystals was chosen to be
26.3 g/kg of solvent, thus corresponding to the mass of R crystals
produced in the first phase of an unseeded experiment with an initial
concentration of 48 g/kg of solvent. Upon addition of the R seed
crystals, the data acquisition of all in situ monitoring tools was started.
It is worth noting that no nucleation or growth of R crystals could occur
during the seeded experiments since the added R seeds were suspended
in a saturated solution. Yet, the suspension was supersaturated with
respect to the stable â form that could therefore nucleate and grow,
thereby consuming supersaturation and eventually triggering the
dissolution of the R crystals.
3. Experimental Results
In this section, we report two series of experiments, i.e.,
unseeded (section 3.2) and seeded experiments (section 3.3).
Before that, in section 3.1 a single reference case is taken as
example of the whole set of experiments to describe and discuss
in detail the results obtained and the information we can extract
from the data collected from the different instruments used to
monitor the process.
3.1. Unseeded Experiments: Reference Case. Let us
consider the unseeded experiment carried out with an initial
concentration of 43 g/kg of solvent. Information obtained during
the course of this experiment (20 h) from ATR-FTIR spectros-
copy, particle vision and measurement (PVM), Raman spec-
troscopy, and FBRM are reported in Figures 8-11, respectively.
Figure 8 shows the time evolution of the L-glutamic acid
concentration measured from ATR-FTIR spectra using the
calibration illustrated in Figure 4 during the formation of R
crystals and their following transformation into â crystals. These
phenomena are clearly recognized when considering the PVM
images in Figure 9, where the prismatic R crystals present at
the beginning of the experiment are replaced by â needles while
the experiment proceeds. Also shown in Figure 8 is the IR
absorbance at 1120 cm-1 (uncalibrated signal shown in arbitrary
units) which indicates the presence of solid particles on the
ATR-FTIR probe window. It can be readily observed that
whenever the latter signal indicates the presence of particles
the main signal used to calculate concentration loses accuracy.
The measurement could be continued only after mechanical
cleaning of the window; this was done five times during the
first 5 h of this experimental run. It is worth noting that ATR-
FTIR was the only in situ technique where probe clogging had
significant effects on the recorded data. Although the ATR probe
tip was in a reactor region with high liquid velocities, particle
formation on the ATR crystal during the early stages of the
experiment could not be completely prevented. This is probably
due to the recessed position of the ATR crystal at the probe
tip, which is more prone to solid formation than the planar
probes used by the other instruments. With reference to the
concentration profile in Figure 8, it is worth noting that apart
from the small bumps before probe cleaning, its regularity and
accuracy are satisfactory.
Figure 8. Solute concentration profile during the transformation
process at 45 °C with an initial concentration of 43 g of L-glutamic
acid/kg of solvent. Below the concentration profile the scaled peak
height data is shown to indicate probe clogging during the beginning
of the process. Independently measured solubility of R and â poly-
morphs are shown by horizonal solid and dashed lines, respectively.
Figure 9. In situ PVM images of the solvent-mediated transformation
process at 45 °C with an initial concentration of 43 g of L-glutamic
acid/kg of solvent: after 1 h the prismatic R crystals prevail, after 6 h
both polymorphs are present, and after 16 h only needlelike â crystals
are visible.