European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
European Journal of Pharmaceutics and Biopharmaceutics
journal homepage: www.elsevier.com/locate/ejpb
Research paper
Moisture and drug solid-state monitoring during a continuous drying
process using empirical and mass balance models
Margot Fonteyne a,⇑, Delphine Gildemyn a, Elisabeth Peeters b, Séverine Thérèse F.C. Mortier c,
Jurgen Vercruysse b, Krist V. Gernaey d, Chris Vervaet b, Jean Paul Remon b, Ingmar Nopens c,
Thomas De Beer a
a
Laboratory of Pharmaceutical Process Analytical Technology, Ghent University, Ghent, Belgium
Laboratory of Pharmaceutical Technology, Ghent University, Ghent, Belgium
c
BIOMATH, Department of Mathematical Modelling, Statistics and Bioinformatics, Ghent University, Ghent, Belgium
d
Department of Chemical and Biochemical Engineering, Technical University of Denmark, Lyngby, Denmark
b
a r t i c l e
i n f o
Article history:
Received 21 December 2013
Accepted in revised form 27 February 2014
Available online xxxx
Keywords:
NIR spectroscopy
Raman spectroscopy
Real-time monitoring
End-point detection
Granule size fractions
Fluid bed drying
PAT
a b s t r a c t
Classically, the end point detection during fluid bed drying has been performed using indirect parameters, such as the product temperature or the humidity of the outlet drying air. This paper aims at
comparing those classic methods to both in-line moisture and solid-state determination by means of
Process Analytical Technology (PAT) tools (Raman and NIR spectroscopy) and a mass balance approach.
The six-segmented fluid bed drying system being part of a fully continuous from-powder-to-tablet
production line (ConsiGma™-25) was used for this study. A theophylline:lactose:PVP (30:67.5:2.5) blend
was chosen as model formulation. For the development of the NIR-based moisture determination model,
15 calibration experiments in the fluid bed dryer were performed. Six test experiments were conducted
afterwards, and the product was monitored in-line with NIR and Raman spectroscopy during drying. The
results (drying endpoint and residual moisture) obtained via the NIR-based moisture determination
model, the classical approach by means of indirect parameters and the mass balance model were then
compared. Our conclusion is that the PAT-based method is most suited for use in a production set-up.
Secondly, the different size fractions of the dried granules obtained during different experiments (fines,
yield and oversized granules) were compared separately, revealing differences in both solid state of
theophylline and moisture content between the different granule size fractions.
Ó 2014 Elsevier B.V. All rights reserved.
1. Introduction
About 80% of the currently available pharmaceuticals are formulated as solid dosage forms, the majority of them being tablets.
In many cases tabletting requires granulation of the starting materials (powders) prior to compaction. As a consequence industrial
tabletting is still a multi-step process nowadays. In case of wet
Abbreviations: API, Active Pharmaceutical Ingredient; DoE, Design of Experiments; KF, Karl Fischer; NCO, Non-contact optic; NIR, Near Infrared; PAT, Process
Analytical Technology; PC, Principal Component; PCA, Principal Component Analysis; PLS, Partial Least Squares; PVP, polyvinylpyrrolidone; RMSEP, Root Mean
Square Error of Prediction; rpm, rotations per minute; SNV, Standard Normal
Variate.
⇑ Corresponding author. Laboratory of Pharmaceutical Process Analytical Technology, Faculty of Pharmaceutical Sciences, Ghent University, Ghent, Belgium. Tel.:
+32 9 264 83 55; fax: +32 9 222 82 36.
E-mail addresses: Margot.Fonteyne@Ugent.be (M. Fonteyne), Thomas.DeBeer@
Ugent.be (T. De Beer).
granulation, a drying step follows the granulation step. Afterwards
the dry granules may be mixed with an external phase (i.e. lubricant, disintegrant). The final blend is then fed to the tabletting machine for compaction, which might be followed by coating, before
the tablets are blistered and packed. After each step of this production chain the critical (intermediate) product characteristics of random samples of the batch are generally evaluated by means of offline analyses in analytical laboratories. Batches will either proceed
to the next processing step or will be rejected in case of failure of
these analysis tests. Hence, traditional batch production is a timeconsuming and expensive production method. Partly due to the
increasing competition and decreasing profits in the pharmaceutical industry (i.e. generics, smaller pipelines, expiring patents, etc.),
innovative manufacturing models are more and more desired in
order to make the production processes faster, cheaper, more
efficient and hence more competitive. Therefore, continuous
production gains increasing interest in the pharmaceutical
industry, also in tabletting applications. Recent manufacturing
http://dx.doi.org/10.1016/j.ejpb.2014.02.015
0939-6411/Ó 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
2
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
technology advances have already shown the advantages of this
approach [1–6].
Recently, full continuous from-powder-to-tablet production
lines became commercially available [7–11]. The traditionally applied quality assessment approach based on off-line analysis is
not applicable in a continuous manufacturing environment, since
the process cannot be stopped and immediate and continuous process and product quality information is required. The need for offline tests would counterbalance the advantages of continuous
manufacturing. Therefore, continuous real-time quality control
should be ensured by means of in-process analysis methods as it
is advised in the Food and Drug Administration’s PAT guideline
[12].
Conventionally, the end point of a drying process – usually performed in a fluid bed dryer – after wet granulation is determined
by means of indirect parameters. The humidity and the temperature of the outgoing drying air, the product temperature and the
pressure difference over the fluidized bed can be evaluated in order
to control the fluidized bed drying of wet granules [13,14]. These
methods give an idea about the water evaporation progress. The
drying process is considered as finished when water evaporation
is no longer detected. However, the in-line monitoring of residual
moisture content during drying using Near Infrared (NIR) spectroscopy is beneficial for two reasons: (i) monitoring the moisture content allows end point detection of the drying process and makes
real-time decision making possible, hence preventing over-drying;
and (ii) information regarding possible structural changes of the
Active Pharmaceutical Ingredient (API) or/and excipients can be
obtained.
The utility of NIR spectroscopy for the in-line monitoring and
moisture assessment during fluid bed and spray drying processes
has been demonstrated extensively in the literature [15–23]. Furthermore, NIR spectroscopy has also recently been used for the
monitoring of a continuous granulation and drying process [10].
However, in this study the measurements were performed after
the drying unit and not during drying. Besides monitoring of the
drying process, NIR spectroscopy can also be used for the continuous evaluation of process induced solid-state transformations of
both APIs and excipients. Romer et al. [24] monitored the solidstate conversions of erythromycin dehydrate using an in-line NIR
spectrometer in a miniaturized fluid bed dryer. Aaltonen et al.
[14] used both NIR and Raman spectroscopy to monitor the solid-state changes of theophylline using the same mini-dryer. They
linked the in-line obtained spectra to the traditionally monitored
fluid bed parameters such as absolute humidity of the outlet air
and pressure difference over the fluidized bed. The solid-state
changes were quantified in real-time, which is impossible with
the traditional indirect parameters. Furthermore, the same micro
scale fluid bed dryer was used by Kogermann et al. [25] to quantify
the solid-state changes of piroxicam and carbamazepine in-line
using Raman spectroscopy.
The presented study aims at evaluating Raman and NIR spectroscopy for the in-line monitoring of the drying process and determination of the end point, the residual moisture content and the
product solid state during continuous drying in a six-segmented
continuous fluid bed drying unit, which is part of a fully continuous
from-powder-to-tablet manufacturing line (ConsiGma™-25, GEA
Pharma Systems nv., Collette™, Wommelgem, Belgium). Furthermore, data derived from the in-line acquired spectroscopic data
are compared with the conclusions obtained from the conventional
indirect approach using the logged univariate parameters such as
humidity of the outlet air and product temperature. Additionally,
the spectroscopic observations are compared with the residual
moisture content conclusions that can be derived from a mass balance model, which was recently developed for the six-segmented
continuous fluid bed dryer [26]. This mass balance model is based
on the physics governing the continuous drying process, hence
forcing fundamental process understanding. Mass balance modeling requires the definition of the composition of the physical inand outgoing gas (moisture content) and the liquid and solid
streams in the process. It is examined whether feeding the continuously logged process parameters (e.g., humidity and temperature
of inlet and outlet air, product temperature, etc.) into this mass
balance model allows visualizing the drying process progress and
calculating the end point of drying and the corresponding residual
moisture content. By comparing these results to the spectroscopic
results, the necessity of using spectroscopic monitoring during
drying is evaluated and discussed.
2. Materials and methods
2.1. Materials
Anhydrous theophylline (Farma-Quimica Sur S.L., Malaga,
Spain) (30%, w/w) was used as a model drug and granulated together with lactose monohydrate 200 M (Caldic Belgium NV,
Hemiksem, Belgium) as filler. Polyvinylpyrrolidone (Kollidon 30Ò,
BASF, Burgbernheim, Germany) was added as a binder to the dry
powder mixture in a concentration of 2.5% (w/w). Distilled water
was used as granulation liquid. Sodium lauryl sulfate (Fagron,
Waregem, Belgium) was added to the granulation liquid (0.5% w/
v) to improve the wettability of the dry powder mixture.
2.2. Continuous twin-screw granulation and fluid bed drying
Continuous granulation and drying was performed using the
ConsiGma™-25 unit (GEA Pharma Systems nv., Collette™, Wommelgem, Belgium), which consists of three major units: a continuous twin screw high shear granulator, a six-segmented fluid bed
dryer and a discharge system. The system has been extensively described elsewhere [7,11]. After discharging, a lubricant can be
added and blended into the dried granules, after which the final
blend can be compressed using an in-line tabletting machine.
One of the assets that ConsiGma™ offers is the continuous logging
and storage of numerous process parameters and outcomes in each
unit (i.e., temperature granulator barrel, torque on twin screws,
weight powder dosing unit, temperature of product in the dryer,
etc.).
2.3. NIR spectroscopy
A Fourier-Transform NIR spectrometer (Thermo Fisher Scientific, Zellik, Belgium, Nicolet Antaris II near-IR analyzer) equipped
with an InGaAs detector, a quartz halogen lamp and a fiber optic
contact probe was used. The probe was inserted in cell 5 of the
six-segmented fluid bed dryer by means of an in-house developed
accessory (Fig. 1). Each spectrum was collected in the 10,000–
4500 cm 1 spectral region with a resolution of 16 cm 1 and was
averaged over 16 scans. Spectra were recorded continuously during drying and a spectrum was collected approximately each
10 s. The same fiber optic contact probe and spectrometer settings
were used for off-line measurements. Spectra were mean centered
and Standard Normal Variate (SNV)-corrected prior to multivariate
data analysis. Data collection and data transfer were done using
Thermo Fisher Scientific’s Result Software.
2.4. Raman spectroscopy
A RamanRxn1 spectrometer (Kaiser Optical Systems, Inc., Ann
Arbor, Michigan, US) equipped with an air-cooled CCD detector
(back-illuminated deep depletion design) was used. For the in-line
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
3
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
Fig. 1. NIR fiber optic contact probe mounted in cell 5 of the dryer by means of an
in-house made accessory.
measurements, an immersion optic (IO), connected to an MR Probe
(Kaiser Optical Systems, Ann Arbor, Michigan) was inserted in cell
2 of the six-segmented continuous fluid bed dryer by means of an
in-house developed accessory. An IO with a working distance of
3 mm was used (i.e. ‘‘long’’). The off-line spectra were collected
with a non-contact optic (NCO), being connected to an MR probe
(Kaiser Optical Systems, Inc., Ann Arbor, Michigan, US). All spectra
were recorded with a resolution of 4 cm 1 and an exposure time of
10 s, using a laser power of 400 mW. In-line spectra were taken
continuously during the 20 min of drying. The system was not
paused in between two spectra, resulting in a spectrum each 10 s
approximately. Spectra were mean centered and SNV-corrected
before data analysis in the spectral region from 200 to
1800 cm 1. Data collection and data transfer were automated
using the HoloGRAMS™ data collection software, the HoloREACT™
reaction analysis and profiling software and the Matlab software
(version 7.1, The MathWorks Inc., Natick, Massachusetts, US).
to dry inside the first cell for a user defined drying time after which
they are unloaded and discharged to the discharge module. Every
cell repeats this cycle of loading, drying and discharging, hence enabling the continuous drying of the continuously produced wet
granules. Two of the dryer cells (cell 2 and cell 5) contain an interfacing opening for PAT tools. In-house designed accessories were
used to insert the Raman (cell 2) and NIR (cell 5) spectroscopic
probes in these interfacing places (Fig. 1).
One of the aims of this study is to evaluate the applicability of
in-line NIR spectroscopy for the in-process monitoring of the drying progress and for the granule moisture content prediction at the
end of drying. To develop an NIR calibration model for residual
moisture content, granules with different residual moisture contents were produced. Therefore, wet granules were manufactured
using a powder feed rate of 10 kg/h, a screw speed of 950 rpm,
9.94% (w/w) granulation liquid and a granulator barrel temperature of 25 °C. During the consecutive drying step, the inlet airflow
and temperature of the drying air were varied as well as the filling
and drying time, hence generating granules with different residual
moisture contents (Table 1). Immediately after drying, the granules
were collected and twenty NIR spectra were measured off-line.
Karl Fischer moisture determination was also performed directly
after the NIR measurements of each calibration experiment. These
data (i.e., the off-line collected NIR spectra (X) and the corresponding Karl Fischer determined residual moisture contents (Y)) were
used for the construction of a Partial Least Squares (PLS) model.
Afterwards, six test experiments were performed, where NIR and
Raman spectra were collected in-line during drying, for both moisture and drug solid-state monitoring. These test experiments were
performed using a powder feed rate of 17.5 kg/h, a liquid rate of
9.94% (w/w), a granulator barrel temperature of 25 °C and a screw
speed of 950 rpm. The filling time and the drying time were 5 and
20 min respectively. An airflow of 360 m3/h was blown through the
bottom plate of the dryer. The test experiments differed regarding
the applied drying air temperature varying from 30 °C to 80 °C in
steps of 10 °C (Table 1). No other cells, besides 2 and 5 were filled
with granules. Directly after each in-line measurement, the moisture content of the dried granules of cell 2 and cell 5 was measured
in triplicate by means of Karl Fisher titration.
2.5. In-process moisture and drug solid-state monitoring
This study focuses on the continuous fluid bed dryer of the
ConsiGma™25-system, which consists of six parallel cells. Wet
granules are loaded into the first cell during a user defined time
period, after which wet granules are loaded into the second cell,
etc. While the second cell is being filled, wet granules continue
2.6. Moisture and drug solid-state variability between the different
granule size fractions
In order to obtain granules with diverse characteristics (i.e.
granule size distribution, residual moisture content, solid state of
Table 1
Experiments for the development of the NIR moisture calibration model and the test experiments.
Exp
Air flow drying air (m3/h)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Wet granules
360
360
360
360
200
360
360
360
360
360
360
360
360
360
Test experiments
TX
360
Temperature drying air (°C)
Filling time (s)
Drying time (s)
Moisture content (%)
45
45
45
45
45
45
45
45
45
45
60
60
60
60
75
45
30
30
75
60
60
60
120
300
150
60
300
300
75
45
30
30
75
90
150
380
600
900
900
900
900
1500
11.54
5.17
6.24
5.34
6.05
7.08
6.16
5.75
5.03
3.95
4.26
3.62
3.78
3.53
3.6
X = 30, 40, 50, 60, 70, 80
300
1200
Varied
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
4
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
and methods and Table 1). Twenty NIR spectra were collected
per calibration sample, resulting in 20 15 = 300 spectra. A PLS
model was developed by regressing these NIR spectra versus the
residual moisture contents of the calibration samples, assessed
via Karl Fischer titration. Subsequently, six test experiments were
performed hereby using in-line NIR and Raman spectroscopic monitoring in order to assess moisture content during drying (using the
developed NIR moisture PLS model) as well as the drug solid state,
respectively. The optimal number of PLS components for the NIR
moisture model was evaluated according to the cumulative fraction of the variation of the y-variable explained (R2[Y]), the fraction
of the total variation that can be predicted estimated by cross-validation (Q2) and the Root Mean Square Error of Prediction (RMSEP)
for the predictions in the test set. The RMSEP was determined in
two ways: (i) comparing the moisture content predicted from the
last five collected in-line NIR spectra of each test experiment. Then
the average of the five obtained predictions was compared to the
actual moisture content determined via Karl Fischer titration
immediately after each experiment; (ii) comparing the moisture
content predicted from only the last NIR spectrum of each test
experiment to the same KF results. Applying the spectral range of
10,000–4500 cm 1 of the 300 calibration spectra resulted in a
three PLS-components model, with an R2[X] of 0.936, an R2[Y] of
0.94 and a Q2 of 0.916. Applying this model for the moisture prediction based on the last 5 in-line collected NIR spectra of the 6 test
experiments (6 5) resulted in an RMSEP of 0.5659%. When only
the last spectrum (6 1) of each test experiment was used, a similar RMSEP of 0.5676% was obtained. The model was optimized by
removing calibration runs 0, 7 and 8 (these runs had a high standard deviation on the Karl Fischer measurements, hence creating
doubt on the correctness of the reference measurements for these
experiments). The R2[X] was 0.994 and both the R2[Y] and the Q2 of
this model were 0.962. The final model consisted out of four PLScomponents and an RMSEP (6 5) of 0.5608% and (6 1)
0.5649% was obtained. This model, including the spectral range
from 10,000 to 4500 cm 1, but excluding calibration experiment
0, 7 and 8 was used for further calculations (Fig. 2).
The six in-line test experiments were performed at six different
drying temperatures varying from 30 °C to 80 °C in steps of 10 °C
per experiment (Table 1). The end point of the drying cycle could
be determined by means of real-time NIR monitoring of the granule’s moisture content. The actual moisture content values, determined with Karl Fischer at the end of the drying cycle of each test
experiment are shown in Fig. 2 (red dashed line). The granules are
considered as dry when a moisture content of 3.6% is reached, since
this is the moisture content of the premix used to produce the
granules. The in-line NIR predicted moisture contents during the
entire drying process of each test experiment are also plotted in
Fig. 2 (blue full lines). During the first five drying minutes of each
test experiment, the cell was loaded with granules and the NIR
probe was not embedded in granules. The probe was measuring
the API), an experimental design (DoE) was performed on the
ConsiGma™-25 unit. A 4-factor full factorial design with high
and low levels and three centerpoints, resulting in 19 experiments,
was performed (Table 2). The temperature of the granulator barrel
was varied between 25 °C and 40 °C and powder feed rates of 10
and 25 kg/h were used. The screw speed was kept constant at
950 rpm and the granulation liquid was added at 9.16% (w/w).
For the drying unit the drying air temperature was varied between
35 °C and 75 °C and an air flow between 350 and 450 m3/h was applied. Two of the 19 experiments, namely experiment 8 and 12,
could practically not be performed, since the produced granules
consisted mainly of fines and the high air flow (450 m3/h) blew
the fines in the filters, leading to blockage of the filters. The
remaining 17 experiments were performed in randomized order.
An amount of 1.25 kg was dried in each cell during 10 min and
each DoE experiment was done in duplicate.
After each DoE experiment, the moisture content and the solidstate of the API were determined for separate sieve fractions: fines
(<150 lm), yield (151 lm–1400 lm) and oversized granules
(>1400 lm). The granules were separated in these three fractions
by means of a sieve tower. For each experiment and its duplicate,
5 Raman spectra were taken from the total granule load and from
the three sieve fractions, resulting in 40 spectra per DoE experiment. Furthermore, the residual moisture content was determined
for both the total fractions of granules and the three sieve fractions
individually by means of Karl Fischer titration (n = 2).
2.7. Karl Fischer moisture determination
The residual moisture of samples was determined by volumetric Karl Fischer titration (KF) using a V30 volumetric Karl Fischer
titrator (Mettler Toledo, Zaventem, Belgium). Methanol (Hydranal,
Sigma Aldrich, Germany) was used as a solvent. Before titration,
granules were stirred and dissolved during 5 min.
2.8. Data analysis
The analysis of the spectra and the development of the Principal
Component Analysis (PCA) and Partial Least Squares (PLS) models
was done using the Simca P+ 12.0.1 software (Umetrics AB,
Umeå, Sweden).
3. Results and discussion
3.1. In-line moisture calibration model and mass balance approach
Fifteen different loads of granules (i.e., calibration set) were
produced for the development of the moisture calibration model.
The process parameters of these 15 runs were varied to obtain
granules with different residual moisture contents (see materials
Table 2
Overview of the DoE parameters. Average moisture content of the total granule load and its granule size fractions.
Run
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
Temperature granulator barrel
(°C)
Powder feed rate (kg/h)
Temperature drying air (°C)
Air flow drying air (m3/h)
40
40
40
40
32.5
25
25
25
32.5
32.5
25
25
25
25
40
40
40
40
25
10
75
350
25
75
450
10
35
450
25
35
350
17.5
55
400
25
75
350
10
35
350
10
75
450
17.5
55
400
17.5
55
400
25
35
450
10
35
450
10
75
350
25
35
350
25
35
450
10
75
450
10
35
350
25
75
350
25
75
450
4.00
4.37
4.17
4.24
3.68
4.07
4.06
4.21
6.51
6.74
6.75
5.81
7.34
7.64
7.10
6.07
4.96
5.79
4.85
4.60
3.88
4.05
3.85
3.84
6.39
6.71
6.80
5.99
5.40
5.87
5.40
4.72
4.26
5.25
4.57
4.05
5.76
6.68
6.32
4.76
3.79
3.97
3.95
3.76
5.48
6.39
5.99
4.95
6.46
6.81
6.39
5.52
3.87
4.05
4.03
4.03
5.13
6.21
5.70
4.62
3.85
3.99
4.10
3.99
3.73
4.06
3.85
3.85
Moisture content (w/w%) of
Total granule load
Oversized granules
Yield
Fines
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
5
Fig. 2. Moisture content (%) in function of time (min). Prediction of moisture content according to in-line NIR measurements (full blue line) and residual moisture content as
obtained by means of Karl Fischer after each test experiment (red dashed line). (For interpretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)
air and the resulting noisy spectra were therefore excluded. As drying proceeds, the moisture content, measured by means of NIR,
reaches a steady state for all six test experiments. One can notice
certain ‘‘plateaus’’ of the blue lines in the graphs (during the first
part of drying), where for different time points exactly the same
predicted moisture contents were obtained. The plateaus were
caused by fouling of the NIR probe. When wet particles were stuck
on the probe window, similar spectra were captured for several
time points. This phenomenon can clearly be seen at the start of
the cycles with low drying temperatures (T 30 and T 40). For the
granule loads that were dry after the drying cycle (T 60, T 70 and
T 80), the drying process end points can be clearly detected and
are indicated in Fig. 2.
Conventionally, the end point of drying is estimated on the basis of a measurement of the humidity of the outlet air or by evaluating the temperature of the drying product. These data are plotted
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
6
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
10
75
9
70
11:43
65
8
7
60
55
13:38
6
50
5
45
Humidity Drying air oultet (%RH)
Product Tempearture (°C)
80
4
40
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20
Time (min)
Fig. 3. Product temperature (yellow line) and humidity of the outlet air (black line) during the test experiments dried at a. 60 °C, b. 70 °C and c. 80 °C. The vertical blue line
indicates the end point as determined by the NIR-based moisture determination model. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)
in Fig. 3 for the test experiments, which resulted in a dry product
(i.e., test experiments which were dried at 60, 70 and 80 °C). The
time points when the humidity of the outlet air and the product
temperature reach steady state are indicated as well as the end
point predicted by the in-line NIR measurements (vertical blue
line). Regarding the humidity of the outlet air, no steady state
can be observed for the experiment conducted at 60 °C. The
humidity of the outlet air reaches a steady state for the experi-
ments run at 70 °C and 80 °C, but this is at a later time instant compared to the drying process end point as indicated by the NIR
moisture determination model. The product temperature seems
to be a more reliable parameter for the drying endpoint determination of the six-segmented fluid bed dryer since each cell has its
own temperature sensor installed, and the sensor is really
embedded in the product. The determined product temperature
steady states are indicated as well in Fig. 3.
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
The evolution of the granules’ residual moisture content during
drying can also be calculated using a mass balance, which was recently developed by our group for the six-segmented fluid bed drying system of the ConsiGma™-25 system [26]. Using this mass
balance, univariate data (e.g., temperature of inlet and outlet air,
product temperature during drying, humidity of inlet and outlet
air), which are automatically and continuously logged by the ConsiGma™25-system are processed to calculate the amount of water,
which evaporates from the granules during drying. Since Karl
Fischer titration measures both free and bound water and since
the mass balance model only determines the free water, the moisture content of the granules in the dryer determined via the mass
balance needs to be corrected for the bound water fraction. Therefore, the water content of the dry premix – being 3.6% – has to be
added to the moisture content values predicted via the mass balance model.
The drying curves obtained through the mass balance model are
plotted in Fig. 4. At the end of the drying cycle, the granules dried
at 60, 70 and 80 °C are considered to be dry according to the model
(they reach 0%). This is confirmed by both the off-line Karl Fischer
measurements and the NIR-based drying curves (Fig. 2). The mass
balance model predicted moisture content values at the end of the
drying cycle when drying at 30, 40 and 50 °C are comparable to
Karl Fischer findings (6.69%, 5.93% and 4.3% respectively). The typical shape of the drying curves in Fig. 4 can unfortunately not be
compared with the curve obtained via NIR predictions (Fig. 2),
since it was only possible to collect good spectra after five drying
minutes in the used experimental set-up (cfr. supra). When comparing the NIR model and mass balance model based obtained
moisture content values after 5 min of drying, mass balance model
predictions are approximately 2% (absolute value) higher than the
values obtained by means of NIR prediction, when granules are
dried at 30, 40 and 50 °C. When granules are dried at 60, 70 an
80 °C the predicted moisture value by the NIR method is comparable to the moisture value calculated via the mass balance model. As
an example, in Fig. 5, the moisture contents as predicted by the NIR
calibration model and the mass balance model for the experiment
performed at 70 °C are presented in an overlay plot. A value of 3.6%
(i.e. moisture content of the dry premix) is added to the mass balance model predictions for better comparison.
7
Fig. 5. Experiment at 70 °C: Overlay of the predicted moisture content via the mass
balance model (MBM, dotted black line) and the NIR calibration model (NIR, full
blue line). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Even though the granules are not dry when a temperature of
30 °C or 40 °C is applied a steady state in the drying curves can
be observed in Fig. 4. The same steady state can be seen in the
NIR prediction curves, which suggests that at lower temperatures
after a given time no extra water evaporates from the granules,
even though they are not dry yet and still fluidizing. The experiment ran at 50 °C does not show this steady state, but a slow decrease in water moisture content during the last 5 min in both
the moisture curves obtained by the NIR model as the mass balance model can be detected. This means the results of the mass
balance model and the NIR-based model are in line.
3.2. In-line solid-state monitoring
3.2.1. In-line Raman spectroscopy
The theophylline solid-state changes were monitored using inline Raman spectroscopy during the six test experiments (Fig. 6a–
f). Some spectral differences can be noted according to the applied
temperature of the drying air. To evaluate the solid state of theophylline, one should focus on the spectral region from 1650 to
1750 cm 1, since theophylline monohydrate shows a band at
Fig. 4. Drying curves of the six different test experiments as calculated with the mass balance model. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
8
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
(d)
0,5
0,5
0,0
0,0
Intensity
Intensity
(a)
-0,5
-0,5
-1,0
-1,0
-1,5
Raman Shift (1/cm)
1718
1706
1712
1700
1694
1688
1682
1676
Raman Shift (1/cm)
(e)
(b)
Intensity
0,5
Intensity
1670
1664
1658
1652
1640
1646
1718
1712
1706
1700
1694
1688
1682
1676
1670
1664
1658
1652
1646
1640
-1,5
0,0
-0,5
0,5
0,0
-0,5
-1,0
-1,0
-1,5
(f)
0,5
1718
1712
1706
1700
1694
1688
1682
0,5
0,0
0,0
Intensity
-0,5
-0,5
-1,0
-1,0
1717
1711
1705
1699
1693
1687
1681
1675
1669
1663
1657
1651
1717
1711
1705
1699
1693
1687
1681
1675
1669
1663
1657
1651
1645
1640
Raman Shift (1/cm)
1645
-1,5
-1,5
1640
Intensity
1676
Raman Shift (1/cm)
Raman Shift (1/cm)
(c)
1670
1664
1658
1652
1646
1640
1718
1712
1700
1706
1694
1688
1682
1676
1670
1664
1658
1652
1646
1640
-1,5
Raman Shift (1/cm)
Fig. 6. In-line Raman spectra collected during the six test experiments. The spectra are colored from light blue to black according to drying time. The first spectrum is light
blue while the last collected spectrum is black. Test experiments dried at: a. 30 °C, b. 40 °C, c. 50 °C, d. 60 °C, e. 70 °C, f. 80 °C. (For interpretation of the references to color in
this figure legend, the reader is referred to the web version of this article.)
1687.5 cm 1, whereas a spectrum of anhydrous theophylline
shows two peaks, one at 1665 cm 1 and one at 1707 cm 1 [7]. Furthermore, a metastable form of theophylline exists, which shows a
typical Raman band at 1692 cm 1 [27]. When granules were dried
at 30 °C, no spectral changes (i.e. changes in solid state) can be noticed (Fig. 6a). Only a band at 1687.5 cm 1 can be seen indicating
that theophylline remains in the monohydrate form from the start
till the end of the drying cycle at 30 °C. In the spectra, collected
during drying at 50 °C (Fig. 6c), two extra peaks can be noticed at
1665 cm 1 and 1707 cm 1, indicating that an amount of theophylline is dehydrated to the anhydrous form. This can also be noted,
when granules are dried at 40 °C albeit to a much lesser extent
(Fig. 6b). The test experiments with drying temperatures of 60 °C
and up (Fig. 6d, e and f) show a shift of the Raman band at
1687.5 cm 1–1692 cm 1. This shift denotes a polymorphic change
from theophylline monohydrate to metastable theophylline. When
comparing Fig. 6d to Fig. 6e and f, the color code of the spectra
shows that the higher the temperature of the incoming air is, the
faster this conversion takes place. Furthermore, additional peaks
at 1665 cm 1 and 1707 cm 1 appear during these test experi-
ments, indicating the formation of anhydrous theophylline. This
means that the dried granules of these test experiments contain
both metastable and anhydrous theophylline.
The findings regarding the solid state of theophylline correspond with the detected steady states in the drying curves (Figs. 2
and 4) of the test experiments conducted at 30 and 40 °C. From a
certain time point, no more water is evaporating during these
experiments, furthermore theophylline stays in the monohydrate
form. These findings suggest that, when drying at 30 or 40 °C the
theophylline hydrate water will stay bound to the theophylline
molecule and hence no extra water will evaporate. For the test
experiment conducted at 50 °C a slow conversion from theophylline monohydrate to anhydrous theophylline can be noted. This
can be correlated with the slightly decreasing slope in the NIR predicted drying curve (minute 13 till end) and the slope in the mass
balance calculated curve (minute 10 till end).
The spectra collected during the drying test experiment at
60 °C were analyzed using Principal Component Analysis (PCA).
The spectral region from 200 cm 1 to 1800 cm 1 was centered
and SNV-corrected. The resulting principal component 1 versus
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
9
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
Fig. 7. PC1 versus PC2 scores plot of the Raman spectra of the test experiment at 60 °C. (For interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)
principal component 2 scatter scores plot (Fig. 7) shows the transition from theophylline monohydrate towards a mixture of metastable and anhydrous theophylline. The first principal component
represented 84.65% of the spectral variance, the second 8.25%.
The spectra taken in the first 12 min only show a band at
1687.5 cm 1 indicating that theophylline remained in the monohydrate state. After 15 min the band has clearly shifted to
1692 cm 1, and bands at 1665 cm 1 and 1707 cm 1 appear as well.
From minute 12 to minute 15, one can observe a transition period
in the plotted data and the Raman spectra. When drying at 70 °C it
took 11m30s to convert theophylline monohydrate to a mixture of
metastable and anhydrous theophylline and the transition period
started after 8 min. A mixture of metastable theophylline and
anhydrous theophylline could be detected after 11 min of drying
at 80 °C. A transition period from the eighth till the eleventh minute could be observed. These findings are different compared to the
results obtained by others, since some found the metastable theophylline at lower drying temperatures (40 °C) [28] when heating
theophylline monohydrate powder in a variable temperature X-ray
powder diffractometer. Morris et al. [29] stated that the metastable
form of theophylline most likely occurs at low temperatures (<
60 °C) when fluid bed drying is applied after wet granulation.
These findings were confirmed when theophylline was wetmassed, followed by drying at 60 °C during 50 min [30]. Airaksinen
et al. [31] already found metastable theophylline when drying at a
temperature as low as 30 °C and using dry inlet air. When using
ambient air for drying the highest relative amount of metastable
theophylline was found when drying at 40 °C. The differences in
observations compared to this study, where metastable theophylline could only be found at 60 °C, might be due to the fact that all
other authors that were mentioned used pure theophylline or pure
theophylline granules and did not use any excipients.
3.2.2. In-line NIR spectroscopy
When evaluating the in-line recorded NIR spectra, not only
the disappearing water bands attract attention. As for Raman
2,0
1,5
Intensity
1,0
0,5
0,0
-0,5
-1,0
-1,5
5901
5908
5916
5924
5931
5939
5947
5955
5962
5970
5978
5985
5993
6001
6009
6016
6024
6032
6039
6047
6055
6063
6070
6078
6086
6093
6101
-2,0
Wavenumber (1/cm)
Fig. 8. In-line collected NIR spectra, SNV-corrected in the spectral range from 6100 to 5900 cm 1. The earliest spectrum is light blue changing to black for the last spectrum.
Test experiment at 60 °C. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
10
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
spectroscopy, the changes in solid state of the API can also be
identified in the NIR spectra. Two extra bands appear during dehydration, which can be attributed to appearance of anhydrous theophylline: one at 5962 cm 1 and one, less obvious at 6009 cm 1
[7]. Near infrared spectroscopy is incapable of differentiating between stable and metastable anhydrous theophylline [14]. In
Fig. 8, the spectral range from 5900 to 6100 cm 1 of the in-line collected spectra (T = 60 °C) was selected and was corrected using
SNV. PCA was applied on these corrected spectra. Fig. 9 shows
the resulting PC 1 versus PC 2 scores plot for this test experiment.
The PC 1 represented 95.1% of the variance and PC 2 represented
3.91%. Three clusters can be distinguished in the scores plot: (i)
the spectra of the first 12m30s, (ii) the spectra collected between
12m30s and 15m45s (iii) the spectra collected after 15m45s
minutes of drying. During the first 12m30s no bands can be seen
at 5962 cm 1 and 6009 cm 1, indicating that all theophylline
remained in the monohydrate form. After 15m45s, the two extra
bands appear clearly, indicating the transformation to anhydrous
theophylline. Between 12m30s and 15m45s a mixture of theophylline anhydrate and monohydrate was found. For the test experiment conducted at 70 °C, anhydrous theophylline could be found
after 11m15s. When granules were dried at 80 °C, the anhydrous
theophylline bands appeared after 9 min.
3.2.3. End-point detection during drying and solid-state monitoring:
evaluation of different techniques
Four different methods to detect the end point of a drying cycle
have been presented: (i) by means of an NIR-based PLS model, (ii)
by logging of the humidity of the outlet air, (iii) by logging of the
product temperature and (iv) by means of a mass balance model.
Secondly, NIR and Raman proved to be able to monitor the
Fig. 10. Residual moisture content of the different granule size fractions as
determined by means of Karl Fischer titration. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this
article.)
solid-state changes of the API in real-time during drying. Table 3
shows an overview of the time points (end of drying or solid-state
change) as calculated using the different methods. For the indirect
methods (ii) and (iii), a clear steady state can be found for the product temperature, but not for the humidity of the outlet drying air.
Both the NIR prediction model and the mass balance model indicate the drying process end point sooner than the two indirect
methods. The dehydration of theophylline monohydrate was monitored with both NIR and Raman spectroscopy. The time points of
1,5
1,0
0,5
PC 2
145
146
143
144
142
141
140
139
138
137
47
48
56
49
50
61
60
41
42
43
44
45
46
57
58
59
66
52
53
54
55
68
69
67 51
62
63
64
65
77
135
134
136
133
132
131
78
81
80
79
76
71
84
82
85
83
72
70
87
86
75
73
74
88
89
0,0
130
129
127
128
126
124
125
122
123
121
90
120
119
118
117
-0,5
9192
94
93
97
96
95
98
99
100
101
102
103
104
106
105
107
108110
109
-1,0
116
115
114
111112
113
-1,5
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
PC 1
Fig. 9. PCA of In-line collected NIR spectra, SNV-corrected in the spectral range from 6100 to 5900 cm 1. PC1 versus PC 2 scores plot, test experiment at 60 °C. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
Table 3
Summary and comparison between the different detected drying process end points and solid-state changes from theophylline monohydrate towards metastable and anhydrous
theophylline.
Applied Temperature (°C)
60
70
80
Drying endpoint according to
Time point of solid-state shift according to
NIR PLS model
Mass balance model
Humidity of the outlet air
Product temperature
Raman
NIR
16:08
11:37
10:00
13:30
11:15
9:00
/
15:46
13:38
16:27
14:49
11:43
15:00
11:30
11:00
15:45
11:15
9:00
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
11
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
conversion from theophylline monohydrate to anhydrous theophylline, derived from the Raman and NIR spectra respectively, are
comparable. It is important to emphasize that both measurements
were conducted in different cells of the dryer, and thus a small
deviation between the results obtained from both spectroscopic
methods is understandable. These time points are also comparable
with the drying process end points calculated using the NIR-based
PLS model. It can be concluded that the in-line monitoring of moisture content by means of PAT tools and mass balance models is
beneficial. The indirect methods seem to overestimate the needed
drying time. It should be remarked that, unlike a conventional fluid
bed dryer where only one load is dried at a time, the Consigma™25 consists of six parallel drying cells, and that only one humidity
sensor is placed on top of the dryer. For this reason, the NIR
method is to be preferred over the mass balance approach in a
production set-up, since the measured humidity of the outlet air
will correspond to the average value of the six cells when all drying
cells are filled with product. Secondly, NIR spectroscopy does not
only provide information regarding residual moisture, but also
regarding the solid state of the API (and excipients). Raman spec-
troscopy will give information regarding the solid state, and is able
to differentiate between anhydrous theophylline and metastable
theophylline. Challenges for a successful integration of in-line
NIR and Raman spectroscopy are (i) the avoidance of fouling of
the probe; and, (ii) an adequate method to mount the probe in
the dryer should be found as well, so that spectra can be collected
from the very start of the drying cycle. Furthermore software packages with high performances for the fast collection and processing
of in-line data would be beneficial.
3.3. Particle size and polymorphism of the API
Differences in polymorphism of excipients depending on the
particle size of spray-dried particles have been reported previously
[32]. The in-line Raman and NIR spectroscopic measurements give
information about the total granule load, but do not provide
granule information related to size fractions within that total
granule load. This section of the paper focuses on the differences
in moisture content and solid state that can be observed between
(a)
8
6
4
PC 2
2
0
-2
-4
-6
-8
-10
-24
-22
-20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6
8
10
2
4
6
8
10
12
14
16
18
20
22
24
PC 1
(b)
15
10
PC 2
5
0
-5
-10
-15
-24 -22 -20 -18 -16 -14 -12 -10
-8
-6
-4
-2
0
12
14
16
18
20
22
24
PC 1
Fig. 11. PCA of the Raman spectra of the different granule size fractions: PC 1 versus PC 2 scores plot. Spectra of fines (blue triangles), yield (red dots), oversized granules
(turquoise blue squares) and total granule load (black diamonds) a. Experiment 4 and b. Experiment 19. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
12
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
fines (<150 lm), yield (151–1400 lm) and oversized granules
(>1400 lm).
To obtain granules with diverse characteristics, a 4-factor full
factorial design (19 experiments) was performed (see materials
and methods, Table 2). Based on the residual moisture content of
the granules of the 19 experiments, they can be classified in three
groups (Table 2): (i) granules, which are ‘dry’ after drying, with a
moisture content of 4% or lower (experiments 1, 2, 6, 13, 16, 18
and 19); (ii) granules with a residual moisture content higher than
5% (experiments 3, 4, 7, 9, 11, 14, 15 and 17); and (iii), granules
having a residual moisture content that is in between 4% and 5%
(experiments 5 and 10). These three groups, except experiment
9, correspond to the applied drying air temperature. The ones with
residual moisture contents higher than 5% were dried at 35 °C,
while the granules having residual moisture contents below 4%
were dried at 75 °C. The latter group does not show significant differences in residual moisture content between the oversized fraction, the yield and the fines (Fig. 10). The residual moisture
content of the granules dried at 55 °C and 35 °C increases with
increasing granule size (Fig. 10). The fines fraction has a lower
residual moisture content compared to the yield and oversized
fraction. This is because fine particles can dry much faster due to
a higher surface to volume ratio. Furthermore, water in the core
of the granule will reach the surface of the granule faster in small
granules compared to larger granules.
Since the moisture content of each size fraction is different,
there might be a difference in solid state of the API as well. Experiment 4 shows the largest differences in residual moisture content
according to granule size. The Raman spectral region from
200 cm 1 to 1800 cm 1 was selected to perform PCA on the 40 collected and SNV preprocessed Raman spectra resulting from experiment 4. PC 1 explained 79.65% of the variance, whereas PC 2
represented 12.01%. Fig. 11a shows the resulting PC 1 versus PC
2 scores plot. The Raman spectra corresponding to the fine fraction
are clearly isolated along PC 1 from the spectra corresponding to
the other size fractions. The information regarding the solid state
of theophylline is reflected in the Raman spectral region from
520 cm 1 to 600 cm 1 [33] and from 1650 cm 1 to 1750 cm 1
[27,34]. The loadings plot of the first principal component
(Fig. 12) in these regions show three maxima: at 554 cm 1,
1685 cm 1 and 1707 cm 1, which indicates that the fraction of
fines contains anhydrous theophylline. The minima at 573 cm 1
and 1687 cm 1 on the other hand, imply that the other size fractions mainly contain theophylline monohydrate. Similar solidstate conclusions were obtained for experiment 3, 5, 7, 9, 10, 11,
14, 15 and 17 (fine fraction = anhydrate; other fractions = mainly
monohydrate).
Although the ‘dry’ runs (experiment 1, 2, 6, 13, 16, 18 and 19) do
not show differences in residual moisture content between the different granule size fractions, similar solid-state observations were
made (Fig. 11b). The PCA of the spectra obtained after experiment
19, resulted in PC 1 representing 61.28% and PC 2 representing
23.84% of the spectral variance, respectively. Again, fines were
mostly clustered in the positive part along the PC 1-axis, whereas
oversized granules are located in the negative part. The spectra
of the yield fraction can be found centrally. The loadings plot of
the first principal component of experiment 19 (Fig. 12) shows
again three maxima, attributable to theophylline anhydrate, and
indicating that the fines consist of anhydrous theophylline. The
minimum in the PC 1 loadings plot on the other hand has shifted
from 1687 cm 1 to 1692 cm 1, indicating that the oversized granules from the runs dried at higher temperatures contain metastable theophylline.
These experiments showed that the overall determined moisture content and the solid-state information derived from the total
granule load differ from moisture content and solid state of the different granule size fractions. Fines tend to dry faster and will contain theophylline anhydrate, whereas the oversized granules will
consist of theophylline monohydrate or metastable theophylline.
4. Conclusion
Raman and NIR spectroscopy were used successfully for the
real-time monitoring of a fluid bed drying process. The drying process end point could be predicted by means of an NIR-based PLS
model. NIR spectroscopy proved to be superior to conventional
end-point determination by means of indirect process parameters
or mass balance modeling in a production set-up. NIR and Raman
spectroscopy proved to be capable of in-line solid-state monitoring
as well. The time point of conversion to the dehydrated state of the
API was comparable to the drying process end point. Furthermore
care should be taken when evaluating the average moisture content and solid-state values obtained on the total granule load, since
significant differences between the different sieve fractions of the
granules were observed.
Acknowledgements
The authors express thanks to Lien Saerens and Anneleen Burggraeve (Ghent University at that time) for their technical assistance. Fund for Scientific Research Flanders (FWO Vlaanderen –
aspirantmandaat Séverine Thérèse F.C. Mortier) is gratefully
acknowledged.
References
Fig. 12. Loadings plot of the first PC of the PCA of the Raman spectra of the different
granule size fractions for Experiment 4 and 19. (For interpretation of the references
to color in this figure legend, the reader is referred to the web version of this
article.)
[1] C. Vervaet, J.P. Remon, Continuous granulation in the pharmaceutical industry,
Chem. Eng. Sci. 60 (2005) 3949–3957.
[2] G. Betz, P. Junker-Burgin, H. Leuenberger, Batch and continuous processing in
the production of pharmaceutical granules, Pharm. Dev. Technol. 8 (2003)
289–297.
[3] H. Leuenberger, New trends in the production of pharmaceutical granules:
batch versus continuous processing, Eur. J. Pharm. Biopharm. 52 (2001) 289–
296.
[4] H. Leuenberger, Scale-up in the 4th dimension in the field of granulation and
drying or how to avoid classical scale-up, Powder Technol. 130 (2003) 225–
230.
[5] H. Leuenberger, New trends in the production of pharmaceutical granules: the
classical batch concept and the problem of scale-up, Eur. J. Pharm. Biopharm.
52 (2001) 279–288.
[6] J. Werani, M. Grunberg, C. Ober, H. Leuenberger, Semicontinuous granulation –
the process of choice for the production of pharmaceutical granules?, Powder
Technol 140 (2004) 163–168.
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015
M. Fonteyne et al. / European Journal of Pharmaceutics and Biopharmaceutics xxx (2014) xxx–xxx
[7] M. Fonteyne, J. Vercruysse, D.C. Díaz, D. Gildemyn, C. Vervaet, J.P. Remon, T.D.
Beer, Real-time assessment of critical quality attributes of a continuous
granulation process, Pharm. Dev. Technol. 18 (2013) 85–97.
[8] M. Fonteyne, S. Soares, J. Vercruysse, E. Peeters, A. Burggraeve, C. Vervaet, J.P.
Remon, N. Sandler, T. De Beer, Prediction of quality attributes of continuously
produced granules using complementary pat tools, Eur. J. Pharm. Biopharm. 82
(2012) 429–436.
[9] J. Vercruysse, D. Córdoba Díaz, E. Peeters, M. Fonteyne, U. Delaet, I. Van Assche,
T. De Beer, J.P. Remon, C. Vervaet, Continuous twin screw granulation:
influence of process variables on granule and tablet quality, Eur. J. Pharm.
Biopharm. 82 (2012) 205–211.
[10] L. Chablani, T.M. K, M. Amit, R. Patrick, S.W. C, Inline real-time near-infrared
granule moisture measurements of a continuous granulation-drying-milling
process, AAPS PharmsciTech 12 (2011) 1050–1055.
[11] J. Vercruysse, U. Delaet, I. Van Assche, P. Cappuyns, F. Arata, G. Caporicci, T. De
Beer, J.P. Remon, C. Vervaet, Stability and repeatability of a continuous twin
screw granulation and drying system, Eur. J. Pharm. Biopharm. 85 (2013)
1031–1038.
[12] FDA-Administration, Guidance for Industry – PAT – A Framework for
Innovative Pharmaceutical Development, Manufacturing, and Quality
Assurance, in, 2004.
[13] K. Wöstheinrich, P.C. Schmidt, Evaluation and validation of a fully
instrumented Hüttlin HKC 05-TJ laboratory-scale fluidized bed granulator,
Drug Dev. Ind. Pharm. 26 (2000) 621–633.
[14] J. Aaltonen, K. Kogermann, C.J. Strachan, J. Rantanen, In-line monitoring of
solid-state transitions during fluidisation, Chem. Eng. Sci. 62 (2007) 408–415.
[15] P. Frake, D. Greenhalgh, S.M. Grierson, J.M. Hempenstall, D.R. Rudd, Process
control and end-point determination of a fluid bed granulation by application
of near infra-red spectroscopy, Int. J. Pharm. 151 (1997) 75–80.
[16] R.L. Green, G. Thurau, N.C. Pixley, A. Mateos, R.A. Reed, J.P. Higgins, In-line
monitoring of moisture content in fluid bed dryers using near-IR spectroscopy
with consideration of sampling effects on method accuracy, Anal. Chem. 77
(2005) 4515–4522.
[17] A. Hartung, M. Knoell, U. Schmidt, P. Langguth, Role of continuous moisture
profile monitoring by inline NIR spectroscopy during fluid bed granulation of
an Enalapril formulation, Drug Dev. Ind. Pharm. 37 (2011) 274–280.
[18] F.J.S. Nieuwmeyer, M. Damen, A. Gerich, F. Rusmini, K. van der Voort, H.
Vromans, Granule characterization during fluid bed drying by development of
a near infrared method to determine water content and median granule size,
Pharm. Res. 24 (2007) 1854–1861.
[19] A. Peinado, J. Hammond, A. Scott, Development, validation and transfer of a
Near Infrared method to determine in-line the end point of a fluidised drying
process for commercial production batches of an approved oral solid dose
pharmaceutical product, J. Pharm. Biomed. Anal. 54 (2011) 13–20.
[20] J. Rantanen, S. Lehtola, P. Ramet, J.P. Mannermaa, J. Yliruusi, On-line
monitoring of moisture content in an instrumented fluidized bed granulator
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
13
with a multi-channel NIR moisture sensor, Powder Technol. 99 (1998) 163–
170.
J. Rantanen, O. Antikainen, J.P. Mannermaa, J. Yliruusi, Use of the near-infrared
reflectance method for measurement of moisture content during granulation,
Pharm. Dev. Technol. 5 (2000) 209–217.
J. Rantanen, A. Jorgensen, E. Rasanen, P. Luukkonen, S. Airaksinen, J. Raiman, K.
Hänninen, O. Antikainen, J. Yliruusi, Process analysis of fluidized bed
granulation, AAPS PharmsciTech 2 (2001). article 21.
E. Rasanen, J. Rantanen, J.P. Mannermaa, J. Yliruusi, H. Vuorela, Dehydration
studies using a novel multichamber microscale fluid bed dryer with in-line
near-infrared measurement, J. Pharm. Sci. 92 (2003) 2074–2081.
M. Romer, J. Heinamaki, I. Miroshnyk, N. Kivikero, N. Sandler, J. Rantanen, J.
Yliruusi, Phase transformation of erythromycin a dihydrate during fluid bed
drying, J. Pharm. Sci. 97 (2008) 4020–4029.
K. Kogermann, J.A. Zeitler, J. Rantanen, T. Rades, P.F. Taday, M. Pepper, J.
Heinamaki, C.J. Strachan, Investigating dehydration from compacts using
terahertz pulsed, Raman, and near-infrared spectroscopy, Appl. Spectrosc. 61
(2007) 1265–1274.
S.T.F.C. Mortier, K.V. Gernaey, T. De Beer, I. Nopens, Analysing drying unit
performance in a continuous pharmaceutical manufacturing line by means of
mass – energy balances, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/
10.1016/j.ejpb.2013.12.014 (Epub ahead of print).
A. Dukic-Ott, T. De Beer, J.P. Remon, W. Baeyens, P. Foreman, C. Vervaet, Invitro and in-vivo evaluation of enteric-coated starch-based pellets prepared
via extrusion/spheronisation, Eur. J. Pharm. Biopharm. 70 (2008) 302–
312.
N.V. Phadnis, R. Suryanarayanan, Polymorphism in anhydrous theophylline –
implications on the dissolution rate of theophylline tablets, J. Pharm. Sci. 86
(1997) 1256–1263.
K.R. Morris, U.J. Griesser, C.J. Eckhardt, J.G. Stowell, Theoretical approaches to
physical transformations of active pharmaceutical ingredients during
manufacturing processes, Adv. Drug Deliv. Rev. 48 (2001) 91–114.
M. Alleso, F. Tian, C. Cornett, J. Rantanen, Towards effective solid form
screening, J. Pharm. Sci. 99 (2010) 3711–3718.
S. Airaksinen, M. Karjalainen, E. Rasanen, J. Rantanen, J. Yliruusi, Comparison of
the effects of two drying methods on polymorphism of theophylline, Int. J.
Pharm. 276 (2004) 129–141.
Y.Y. Lee, J.X. Wu, M.S. Yang, P.M. Young, F. van den Berg, J. Rantanen, Particle
size dependence of polymorphism in spray-dried mannitol, Eur. J. Pharm. Sci.
44 (2011) 41–48.
A. Jorgensen, J. Rantanen, M. Karjalainen, L. Khriachtchev, E. Rasanen, J.
Yliruusi, Hydrate formation during wet granulation studied by spectroscopic
methods and multivariate analysis, Pharm. Res. 19 (2002) 1285–1291.
H. Wikstrom, P.J. Marsac, L.S. Taylor, In-line monitoring of hydrate formation
during wet granulation using Raman spectroscopy, J. Pharm. Sci. 94 (2005)
209–219.
Please cite this article in press as: M. Fonteyne et al., Moisture and drug solid-state monitoring during a continuous drying process using empirical and
mass balance models, Eur. J. Pharm. Biopharm. (2014), http://dx.doi.org/10.1016/j.ejpb.2014.02.015