direct precipitation of micron-size salbutamol …
TRANSCRIPT
DIRECT PRECIPITATION OF MICRON-SIZE
SALBUTAMOL SULPHATE BY ANTISOLVENT
CRYSTALLIZATION
XIE SHUYI
NATIONAL UNIVERSITY OF SINGAPORE
2010
DIRECT PRECIPITATION OF MICRON-SIZE
SALBUTAMOL SULPHATE BY ANTISOLVENT
CRYSTALLIZATION
XIE SHUYI
(M. ENG., NUS)
A THESIS SUBMITTED
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF CHEMICAL AND BIOMOLECULAR
ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2010
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
i
Acknowledgements
This thesis is the end of my journey in obtaining my Ph. D degree from Department
of Chemical and Biomolecular Engineering in National University of Singapore (NUS).
The journey would have taken far longer to complete without the help, support and
encouragement from many others over the last few years.
First of all, I would like to express my sincere gratitude to my supervisor, Prof.
Reginald Tan, for his invaluable guidance and thoughtful insight throughout the study. I
am very fortunate to have been his student again after two years study with him for my
master degree and I have learned a lot from him during almost nine years studying with
him. The completion of this thesis would be impossible without his patient supervision.
This work has been carried out at the Institute of Chemical and Engineering
Sciences (ICES). I am very grateful to my team leader in Crystallization and Particle
Sciences (CPS) group at ICES, Dr. Chow Pui Shan, for her invaluable help and moral
support throughout this work. She has always been available for technical discussions
and has given her time for proof reading my reports and manuscripts.
My thankfulness is extended to all the colleagues in CPS group at ICES for
technical assistance and useful advices during the research stage. Special thanks go to Dr.
Poornachary for helping in molecular modeling; Dr. Yu Zaiqun for helping with the
ATR-FTIR experiments; Dr. Aitipamula for solving the single crystal XRD structure; lab
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
ii
officers in CPS lab, especially Angeline Seo and Ng Junwei for their cheerful assistance
on crystallization experiments and crystalline product characterization.
Last but not least, I am deeply indebted to my family members for their selfless
love and affection over the past years. My husband Dou Jian, has given me enormous
support and encouragement so that I have strength and energy to go through this difficult
journey. My parents and parents-in-law have suffered a lot of time helping me to take
care of my child. My beloved children, Jiazheng and Jiarong, are the greatest source of
my strength and motivation.
Xie Shuyi
Singapore, July 2010
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
iii
Table of Contents
Acknowledgements ...............................................................................................................i
Table of Contents ............................................................................................................... iii
Summary ........................................................................................................................... vii
List of Figures ...................................................................................................................... x
List of Tables ..................................................................................................................... xv
Nomenclature ....................................................................................................................xvi
Chapter 1 Introduction...................................................................................................... 1
1.1 Challenges in producing fine particles in pharmaceutical industries ........................ 2
1.2 Antisolvent crystallization ........................................................................................ 5
1.3 The role of surfactants and polymers ........................................................................ 6
1.4 Objectives and approach ........................................................................................... 7
1.5 Thesis outline ............................................................................................................ 8
Chapter 2 Crystallization Fundamentals and the Role of Additives ............................ 9
2.1 Solubility, supersaturation and solvent selection ...................................................... 9
2.2 Nucleation ............................................................................................................... 13
2.2.1 Homogeneous nucleation ............................................................................ 14
2.2.2 Heterogeneous nucleation ........................................................................... 17
2.2.3 Secondary nucleation .................................................................................. 18
2.3 Crystal growth ......................................................................................................... 19
2.3.1 Diffusion controlled growth ....................................................................... 20
2.3.2 Integration controlled growth ..................................................................... 22
2.3.3 Diffusion-integration controlled growth ..................................................... 23
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
iv
2.4 Effect of additives ................................................................................................... 25
2.4.1 Solubility ..................................................................................................... 25
2.4.2 Crystal nucleation ....................................................................................... 27
2.4.3 Crystal growth ............................................................................................ 30
2.5 Molecular modeling ................................................................................................ 33
2.5.1 BFDH Theory ............................................................................................. 34
2.5.2 AE Theory .................................................................................................. 35
2.5.3 Equilibrium Morphology (SE model) ......................................................... 36
2.5.4 Impurity interactions: Binding energy ........................................................ 37
2.6 Closing remarks ...................................................................................................... 39
Chapter 3 Experimental – In Situ and Ex Situ Analytical Techniques ...................... 41
3.1 Ex situ analytical techniques ................................................................................... 41
3.1.1 High Performance Liquid Chromatography (HPLC) ................................. 41
3.1.2 Malvern MasterSizer .................................................................................. 43
3.1.3 X-ray Diffraction (XRD) ............................................................................ 46
3.1.4 Microscopy analysis ................................................................................... 47
3.1.5 Inverse Gas Chromatography (IGC) ........................................................... 48
3.1.6 Viscosity study ............................................................................................ 50
3.2 Process Analytical Technology ............................................................................... 51
3.2.1 FBRM ......................................................................................................... 52
3.2.2 ATR-FTIR .................................................................................................. 54
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization .......... 57
4.1 Introduction ............................................................................................................. 57
4.2 Experimental methods ............................................................................................. 59
4.2.1 Materials ..................................................................................................... 59
4.2.2 Solubility measurement .............................................................................. 60
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
v
4.2.3 Crystallization Experiments ........................................................................ 61
4.3 Results and Discussion ............................................................................................ 61
4.3.1 Solubility measurement .............................................................................. 61
4.3.2 Initial supersaturation ratio ......................................................................... 64
4.3.3 Crystal habit ................................................................................................ 67
4.3.4 Agglomeration ............................................................................................ 72
4.3.5 Particle size ................................................................................................. 73
4.3.6 PXRD analysis ............................................................................................ 75
4.3.7 Effect of mixed antisolvent ......................................................................... 77
4.4 Conclusions ............................................................................................................. 79
Chapter 5 Application of Surfactants and Polymeric Additives ................................. 82
5.1 Introduction ............................................................................................................. 82
5.2 Experimental ........................................................................................................... 85
5.2.1 Materials ..................................................................................................... 85
5.2.2 Crystallization Experiments ........................................................................ 85
5.3 Screening of polymers and surfactants ................................................................... 87
5.4 Crystal structure and growth morphology of SS ..................................................... 90
5.4.1 Crystal structure .......................................................................................... 91
5.4.2 Crystal habit prediction ............................................................................... 93
5.5 Mechanism of influence of the additives on SS crystal growth .............................. 97
5.5.1 Effects of PVP concentration .................................................................... 101
5.5.2 Effects of molecular weight of PVP ......................................................... 106
5.5.3 Variation of crystal surface energy ........................................................... 110
5.6 Conclusions ........................................................................................................... 112
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization ........................ 116
6.1 Introduction ........................................................................................................... 116
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
vi
6.2 Experimental ......................................................................................................... 118
6.2.1 Experimental setup ................................................................................... 118
6.2.2 Experiments .............................................................................................. 120
6.3 Results and discussion ........................................................................................... 123
6.3.1 Calibration of ATR-FTIR ......................................................................... 123
6.3.2 Solubility measurement ............................................................................ 128
6.3.3 Effects of feeding rate ............................................................................... 130
6.3.4 Effects of agitation speed .......................................................................... 135
6.3.5 Effect of addition of PVP ......................................................................... 139
6.4 Conclusion ............................................................................................................. 143
Chapter 7 Conclusions and Future Work ................................................................... 145
7.1 Main contributions and findings ........................................................................... 145
7.1.1 Screening of operational parameters in antisolvent crystallization of SS 145
7.1.2 Application of surfactant and polymers .................................................... 146
7.1.3 Process study of semi-batch antisolvent crystallization ........................... 147
7.2 Future work ........................................................................................................... 148
7.2.1 Supersaturation control via application of ATR-FTIR technique ............ 148
7.2.2 Scale-up of anti-solvent crystallization .................................................... 149
7.2.3 Solvent-influenced morphology prediction .............................................. 150
7.2.4 Molecular modeling on stabilization effect .............................................. 150
References ........................................................................................................................ 152
List of Publications .......................................................................................................... 173
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
vii
Summary
Particle size is a key factor for inhalation drug powder as it determines the dosing
efficiency, drug deposition in various regions, and the systemic absorption. Efficient
delivery of drugs to the deep lungs, necessitates, amongst other factors, that drug particles
are small enough to pass through the conducting airways and deposit in the deep lung, but
not so small that they are exhaled out. Apart from the conventional micronization method,
addition of surfactants into the crystallization solution has been recognized as an effective
approach to achieve micron-size drug particles. For this new approach, it is not only
important to explore the effect of the additives, but also understand the mechanisms
underlying their role on the crystal growth process. Only then a robust process can be
developed to obtain the inhalation drug particles with the desired solid-state attributes.
Having said this, the objective of this work is to obtain micron-size drug particles by
direct precipitation in the presence of proper additives, and to understand the mechanism
by which the additives inhibit crystal growth of the drug particles. Subsequently, it aims
to develop a semi-batch crystallization process to produce micron-sized drug particles.
In this work, salbutamol sulphate (SS), a drug for treatment of asthma, was used as
a model compound. Antisolvent process was used to access the possibility of generating
rapid crystallization. Several surfactants and polymeric additives including lecithin (from
soybean), Span 85, hydroxypropyl methyl cellulose (HPMC) and polyvinylpyrrolidone
were applied in the antisolvent crystallization process.
The effect of various antisolvents on crystal morphology, size and crystallinity of
SS was studied. It is observed that SS can precipitate at different induction times in the
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
viii
various antisolvent systems and exhibit two main crystal habits: plate-shape in aprotic
solvents and needle-shape in protic solvents, due to the different nature of the
antisolvents and the interaction between the antisolvent molecules and the specific facet
of SS. Antisolvents have impact on particle size measurement of SS by influencing both
crystal size and agglomeration degree. Agglomeration degree is higher in plate-shape
crystals than in needle-shape crystals because two dominating faces in plate-shape habit
enable crystals to collide and adhere easily as compared to the needle-shape crystals.
Based on the study in this chapter, ethanol was selected as the antisolvent in the
subsequent antisolvent crystallization experiments.
Antisolvent crystallization in the presence of several surfactants and polymeric
additives was carried out to directly produce SS particles. SS crystals obtained in the
presence of 2.5 mg/mL PVP K25 have the smallest particle size and remained stabilized
even after 24 hours. Studying the molecular interactions between the additives and SS
crystal surface showed that PVP adsorbs selectively on the crystal surface and inhibits
growth along the b-axis. Experimental data on the effect of PVP concentration and
molecular weight is also supportive of the proposed mechanism. Surface dispersive
energy measurement by IGC suggests that PVP is adsorbed on the crystal surface, which
agrees with the proposed mechanism.
In the semi-batch crystallization process, it is observed that both induction time and
particle size are dependent on the feeding rate at low stirring speed but independent on
the high feeding rate at high stirring speed. This suggests that the system is limited by
micromixing at high mixing intensity, whereas macromixing is the controlling regime
when mixing intensity is low. Increased agitation intensity enhances the formation of
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
ix
agglomeration, and hence leads to reduced particle size. However, further increase in
agitation intensity turned the system into growth control regime. Addition of PVP K25
retards nucleation rate and inhibits crystal growth. These results support the mechanism
proposed in the previous chapter that several types of hydrogen bonds exist between SS
molecules and PVP repeating units.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
x
List of Figures
Figure 2-1 The solubility/ supersaturation phase diagram.
Figure 2-2 Example solubility curves for drawing-out crystallizations.
Figure 2-3 Nucleation mechanisms (Randolph and Maurice, 1988).
Figure 2-4 Induction time vs. supersaturation for antisolvent crystallization of Abecarnil.
Figure 2-5 Adsorbed layer of solute on the surface of a growing crystal (□,
solute molecules; ○ and Δ solute ions) (Clontz and McCabe, 1971).
Figure 2-6 Crystal growth model. (a) Continuous growth, (b) birth and
spread, (c) screw dislocation. Figure 2-7 Concentration driving forces in diffusion-integration controlled
growth. Figure 2-8 Solubility curves of ideal (curve 1) and real (curve 2) solutions.
Curve 3 is the metastable curve. Figure 2-9 Adsorptions of impurities on (a) kink, (b) step and (c) ledge
(Coombes et al., 2002).. Figure 2-10 Comparison of predicted and observed morphology (Coombes et
al., 2002). Figure 2-11 Predicted morphology of a pure naphthalene morphology (left)
and a modified naphthalene morphology in the presence of biphenyl (Clydesdale et al., 2005).
Figure 3-1 High Performance Liquid Chromatography calibration for SS
water solution. Figure 3-2 Measurement of particle size at various pump speeds.
Figure 3-3 Measurement of particle size at various ultrasonic levels.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xi
Figure 3-4 Schematic illustration of Bragg's law.
Figure 3-5 The schematics of an Focused Beam Reflectance Measurement (FBRM) probe (left) and the internal signal processing (right).
Figure 3-6 Working principle of Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR).
Figure 4-1 Salbutamol sulphate structural formula.
Figure 4-2 Measured solubility of salbutamol sulpahte in binary solvents.
Figure 4-3 SEM images of salbutamol sulphate crystals in acetone-water system with volume ratio of acetone to water of (a) 3; (b) 5; (c) 7 and (d) 9.
Figure 4-4 SEM images of salbutamol sulphate crystals in acetonitrile-
water system with volume ratio of acetonitrile to water of (a) 5; (b) 7 and (c) 9.
Figure 4-5 SEM images of salbutamol sulphate crystals in ethanol-water
system with volume ratio of ethanol to water of (a) 5; (b) 7 and (c) 9.
Figure 4-6 SEM images of salbutamol sulphate crystals in 1-propanol-water
system with volume ratio of 1-propanol to water of (a) 7 and (b) 9.
Figure 4-7 SEM images of salbutamol sulphate crystals in 2-propanol-water
system with volume ratio of 2-propanol to water of (a) 3; (b); (c) 7 and (d) 9.
Figure 4-8 Volume mean size of salbutamol sulphate crystals against initial
supersaturation ratio at time interval of 1 hour. Figure 4-9 Powder X-ray diffraction patterns of (a) salbutamol sulphate raw
material and salbutamol sulphate crystals obtained from (b) acetone, (c) acetonitrile, (d) ethanol, (e) 1-propanol and (f) 2-propanol.
Figure 4-10 SEM images of salbutamol sulphate crystals formed from
multiple-antisolvent containing (a) 20% ethanol and 80% acetone; (b) 80% ethanol and 20% acetone.
Figure 5-1 Molecular structure.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xii
Figure 5-2 Salbutamol sulphate crystals obtained at 30 minutes after nucleation in the presence of (a) no additives, (b) lecithin, (c) Span 85, (d) Hydroxypropyl Methylcellulose (HPMC) and (e) polyvinylpyrrolidone (PVP) K25.
Figure 5-3 Comparison of Powder X-ray Diffraction patterns for salbutamol
sulphate obtained in the absence of additives and those in the presence of lecithin, Span 85, polyvinylpyrrolidone (PVP) K25 and Hydroxypropyl Methylcellulose (HPMC).
Figure 5-4 3-D view of the monoclinic crystal structure (unit cell) of
salbutamol sulphate. Color scale of atoms: C (gray), H (white), O (red), S (yellow) and N (blue). Hydrogen bonding is represented with cyan dotted lines.
Figure 5-5 Comparison of Powder X-ray Diffraction patterns for
experimental result and patterns simulated from Cambridge Structural Database and the current work.
Figure 5-6 Predicted morphology of salbutamol sulphate by (a) the BFDH
method and (b – c) the AE method using various force field potentials. (b) The DreidingQEq potential and (c) the COMPASS force field applied on the crystal structure respectively.
Figure 5-7 Salbutamol sulphate crystals obtained from evaporation
crystallization in the presence of 5.0mg/mL of polyvinylpyrrolidone (PVP) K25.
Figure 5-8 Molecular modeling of (a) interaction of polyvinylpyrrolidone
monomer unit with the {110} face of salbutamol sulphate crystals, (b) & (c) molecular topology of the {200} and {002} faces of salbutamol sulphate crystals respectively.
Figure 5-9 Molecular modeling of interaction of Hydroxypropyl
Methylcellulose (HPMC) monomer unit with the {110} face of salbutamol sulphate crystals.
Figure 5-10 Molecular modeling of interaction between lecithin and
salbutamol sulphate crystals. Figure 5-11 Salbutamol sulphate crystals obtained via antisolvent
crystallization at 24 hours after nucleation in the presence of (a) 1.0 mg/mL, (b) 2.5 mg/mL and (c) 5.0 mg/mL of polyvinylpyrrolidone (PVP) K25.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xiii
Figure 5- 12 Particle size distributions of SS crystals obtained via antisolvent crystallization in the presence of polyvinylpyrrolidone (PVP) K25 at (a) 30 min and (b) 24 hours after nucleation.
Figure 5-13 Comparison of particle size distribution obtained in the presence
of 2.5 mg/mL of polyvinylpyrrolidone (PVP) K25 at different times after nucleation.
Figure 5-14 Salbutamol sulphate crystals obtained from evaporation
crystallization in the presence of polyvinylpyrrolidone (PVP) K25 at (a) 0 mg/mL, (b) 1.0 mg/mL, (c) 2.5 mg/mL and (d) 5.0 mg/mL. The scale bar is 1000m.
Figure 5-15 Salbutamol sulphate crystals obtained in the presence of (a) no
additives; (b) 25 mg/mL of 1-Methyl-N-pyrrolidone; and 1.0 mg/mL of polyvinylpyrrolidone (PVP) (c) K10; (d) K25; (e) K40 and (f) K360.
Figure 6-1 Experimental set-up for semi-batch antisolvent crystallization. Figure 6-3 Residuals of a chemometric calibration model for salbutamol
sulphate concentration. Figure 6-4 Residuals of a chemometric calibration model for solvent
composition. Figure 6-5 Residuals of a chemometric calibration model for additive
composition. Figure 6-6 Solubility data measured by Attenuated Total Reflectance
Fourier Transform Infrared (ATR-FTIR). Figure 6-7 Evolutions of total chord counts at various feeding rates when
agitation speed is (a) 200 rpm and (b) 500 rpm. Figure 6-8 Chord length distributions at the time interval of 1 hr after onset
of nucleation at different feeding rate with stirring speed of 500 rpm.
Figure 6-9 Images of crystals obtained at different feeding rate of (a) 10
mL/min, (b) 20 mL/min and (3) 30 mL/min with stirring speed of 500 rpm.
Figure 6-10 Evolutions of total counts of chord length at various stirring
speed when feeding rate is 20 mL/min.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xiv
Figure 6-11 Chord length distributions at time interval of 4 hrs after onset of nucleation at various agitation speeds when feeding rate of 20 mL/min.
Figure 6-12 Images of crystals obtained at agitation speed of (a) 100 rpm and (b) 500 rpm.
Figure 6-13 Effect of addition of polyvinylpyrrolidone (PVP) K 25 on
evolutions of the chord length counts at feeding rate of 20 mL/min and 500 rpm.
Figure 6-14 Chord length distributions of crystals at time interval of 1 h after
onset of nucleation at feeding rate of 20 mL/min and 500 rpm. Figure 6-15 Effect of addition of polyvinylpyrrolidone (PVP) K 25 on
supersaturation ratio evolutions at feeding rate of 20 mL/min and 500 rpm.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xv
List of Tables
Table 4-1 Property table of solvents. Table 4-2 Initial supersaturation ratio and observation on nucleation at various
volume ratios of antisolvent/water. Table 4-3 Particle size distribution of salbutamol sulphate crystals obtained from
various antisolvents. Table 4-4 Comparison of particle size and induction time for different volumetric
ratios of ethanol-acetone mixtures. Table 5-1 Particle size distributions of salbutamol sulphate obtained in the
presence of additives at 30 minutes after nucleation. Table 5-2 Comparison of the unit cell dimensions between Cambridge Structural
Database and new determination of the crystal structure of salbutamol sulphate.
Table 5-3 The fraction of particle size less than 5 microns of salbutamol sulphate
particles obtained in the presence of polyvinylpyrrolidone (PVP) K25. Table 5-4 Particle size distributions of SS obtained in the presence of 1-Methyl-
N-pyrrolidone and polyvinylpyrrolidone of different molecular weight. Table 5-5 Dispersive surface energy values ( d
s ) of salbutamol sulphate and
polyvinylpyrrolidone K25 particles. Table 6-1 Solvent compositions, salbutamol sulphate and polyvinylpyrrolidone
concentrations used in calibration experiment Table 6-2 Experimental conditions for semi-batch antisolvent crystallization Table 6-3 Particle size distribution of particles obtained at different feeding rate
with stirring speed of 500 rpm. Table 6-4 Particle size distribution of particles obtained at different agitation
speed with feeding rate of 20 mL/min. Table 6-5 Particle size distribution of particles obtained with and without the
presence of polyvinylpyrrolidone.
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xvi
Nomenclature
a actual activity of the solute [mol L-1
]
a* activity of the solute in a saturated solution [mol L-1
]
A surface area of the crystal [m2]
A’ Kinetic factor affecting the crystal nucleation [-]
Δb differential binding energy [kJ mol-1]
c concentration [g/g or %]
ceq equilibrium concentration of solute [g/g or %]
Δc concentration difference [g/g or %]
d impeller diameter
Ecr lattice energy [kJ mol-1]
Esl, Esl’ slice energy [kJ mol-1], slice energy with impurity molecule [kJ mol-1]
Eatt, Esl’ Attachment energy [kJ mol-1], attachment energy with impurity molecule [kJ mol-1]
ΔG free energy change [J]
ΔHm , ΔHs, ΔHmix enthalpy of melting, solution and mixing [J]
J nucleation rate [μm-3
s-1
]
k Boltzmann constant [m2
kg s-2
K-1
]
kd rate constant for surface integration process [-]
kr coefficient of mass transfer [-]
kv, ks volume and area shape factor [-]
Direct Precipitation of Micron-size Salbutamol Sulphate by Antisolvent Crystallization
xvii
K mass transfer constant
L characteristic diameter of the crystal [m]
M mass of the crystal [kg]
Ms suspension density [kg m-3
]
N mixing rate [rpm]
r size of nuclei [μm]
r* critical size of nuclei [μm]
R gas constant [J mol-1
K-1
]
S a Supersaturation ratio [-]
T temperature [ºC]
Ts, Tref column temperature [ºC], reference temperature [ºC]
VN net retention volume [m-3]
ρ crystal density, [kg m-3
]
υ molecular volume of the crystal [m3
mol-1
]
υh volume of a growing unit [m-3]
Σef effective surface specific energy [J m-2]
γ interfacial energy between solid and liquid [J m-2
]
γsd dispersive components of surface free energy for solid [J m
-2]
γsl dispersive components of surface free energy for liquid [J m
-2]
μ chemical potential [J mol-1
]
ω Slurry voidage [-]
Chapter 1 Introduction
1
Chapter 1
Introduction
Inhalation of drug powder has been recognized as one of the major methods for
treatment of pulmonary diseases because the local drug delivery to the respiratory tract
offers several advantages over the systemic route in treating broncho-constrictive
conditions (Bennett et al., 2002). With increasing rate of different diseases, inhalation
therapy nowadays has been expanded from the conventional therapeutic areas such as
asthma and chronic obstructive pulmonary disease (COPD) to treatment of symptoms
such as diabetes, osteoporosis, cancer pain, erectile dysfunction and lung complications
due to AIDS (Montgomery, 1992; Edwards et al., 1997; Patton et al., 1999; Haynes et al.,
2003; Dudley et al., 2008). The rapid growth of inhalation therapy has sparked
considerable interest in developing more effective technologies to enhance efficiency of
drug delivery. This may be achieved through developing novel inhaler devices and/or
improving the existing inhalation formulations (Nagao et al., 2005).
The efficiency of the inhalation drug formulations has been limited by the difficulty
in making appropriate doses available to the lower respiratory tract of the patient
(Shekunov and York, 2000). It is found that deposition of inhalation drugs is only 5-15%
(Schurch et al., 1990) and it reduces further in patients with dysponea or mucus plugs in
Chapter 1 Introduction
2
the respiratory pathway. In particular, inhalation drug delivery systems are not much
effective for pediatric age group asthma and advanced COPD because for these patients
small airways is the predominant targeting area and drugs deposit poorly in them
(Bhavna et al., 2009).
It is known that the deposition of inhaled aerosol is significantly related to the
particle size. Drug particles with aerodynamic diameter greater than 10 µm tend to
deposit in the mouth and throat; only smaller particles can travel farther and settle in the
deeper and smaller airways (Weda et al., 2004). For example, particles between 5 and 10
µm tend to deposit in the trachea and bronchial regions; particles of approximately 3 µm
is inclined to deposit in the alveoli (although these will also deposit in mouth/throat and
tracheobronchial regions). However there appears to be a limit for reducing the particle
size for delivery efficiency enhancement because of the fact that submicron particles
(with size lesser than 1 μm) tend to be exhaled significantly and mismatch targeting
(Sung et al., 2007). Some studies revealed that ultrafine particles, i. e. nanoparticles, are
promising carrier systems for respiratory drug delivering to the alveoli due to the high
penetration ability, although their clinical roles in respiratory tract diseases are unclear
(Kreuter, 1991; Shekunov et al., 2007). Nevertheless, instability is a concern owing to
particle agglomeration and settling caused by strong particle–particle interactions of
nanoparticles, which could lead to poor functionality of the inhaler (Dailey et al., 2003).
1.1 Challenges in producing fine particles in pharmaceutical industries
Conventionally, micron-size particles are produced through milling of larger
crystals previously formed in crude crystallization processes (Rasenack and Muller,
Chapter 1 Introduction
3
2004). After crystallization, the suspension of crystalline drugs is filtered, dried and
subjected to comminution procedure. High energy input in such methods, however, may
result in undesired changes to the drug particles, for instance, changes in crystal
polymorphism (Chikhalia et al., 2006) and/or generation of amorphous regions on the
drug surface (Chow et al., 2007). The amorphous materials may then recrystallize in an
uncontrolled manner, form solid bridges between particles at high humidity, and
consequently, the powder dispersibility may deteriorate during storage (Berard et al.,
2002). In addition, the decrease in crystallinity could facilitate chemical degradation,
leading to a decrease in shelf-life of the final product. Milled particles have also been
reported to exhibit changes in particle size during storage (Ward and Schultz, 1995; Ng et
al., 2008). Therefore, developing alternative techniques that produce the drug with the
desired particle size, morphology and crystallinity without intensive milling is of
significant interest (Chan, 2006).
Spray drying is often used to produce fine particles of pharmaceutical compound as
a quick drying method in crystallization process (Moran and Buckton, 2007). It involves
evaporation of moisture from an atomized feed by mixing the spray and the drying
medium (Qian et al., 2009). However, the high temperatures involved in this application
restrict processing of thermally labile compounds and it typically produces amorphous
particles. In addition, organic liquids involved may necessitate high machine expenditure
and causes environmental problems. Supercritical fluid technology has been successfully
used to prepare anti-asthmatic compounds (Shekunov and York, 2000), in which a drug
that is insoluble in supercritical carbon dioxide is precipitated out of an organic solution
Chapter 1 Introduction
4
that is sprayed into the supercritical fluid. But the disadvantage of this technique is high
machine cost and limited industrial application (Charpentier et al., 2008).
Lower-micron sized particles can be obtained from direct precipitation for the
systems with very fast nucleation kinetics. In such systems mixing becomes a critical
factor because with insufficient mixing nucleation can happen before a homogeneous
supersaturation level is achieved in the reactor. The inhomogeneous field of
supersaturation causes different driving force for nucleation and particle growth, which
dramatically affects the quality of final product. Therefore some apparatus are designed
to enhance mixing in order to produce fine particles with narrow size distribution.
Impinging jet is one of such apparatus merging recently in the study of pharmaceutical
particles. It is a rapid process in which certain organic reactions or precipitations occurs
at high supersaturation through rapid mixing provided by jet mixers (Johnson and
Prud'homme, 2003). Another example, rotor-stator is a high speed mixing device where
the rotor potion is a blade and the stator is a container with opening through which the
stream can pass into an outer housing and then out of system (Padron et al., 2008). High
pressure homogenizer generates high speed mixing using a high pressure generator.
Reactive streams can mix rapidly before nucleation occurs after passing through a
pressure generator and then entering into a homogenizing valve (Floury et al., 2004).
Although these techniques can successfully produce fine particle with narrow size
distribution, there are few main disadvantages, i.e. low production rate, difficulty in
maintaining the alignment of the streams for impinging jet, special care for high pressure
equipment, prone to clogging and expensive to fabricate at large scale.
Chapter 1 Introduction
5
1.2 Antisolvent crystallization
Antisolvent crystallization is an alternative to form particles with considerably
small size. Unlike other conventional crystallization methods such as cooling and
evaporation, it can generate rapid precipitation through involving the addition of a second
liquid (antisolvent) to a first liquid comprising a solvent and a dissolved substance
(Karpinski and Wey, 2002 ). The two liquids are miscible and lead to a rapid generation
of supersaturation by lowering the solubility of the material to be crystallized in the
mixed solvents. Such rapid generation of supersaturation favors crystal nucleation more
than growth, resulting in formation of fine particles. Antisolvent crystallization has
unique advantages over other techniques mentioned above: it can be carried out at
ambient temperatures and in conventional crystallization vessels, which can be easily
applied in pharmaceutical industries. Therefore it is proposed in the current work to
achieve direct precipitation of salbutamol sulphate (SS) particles.
1.3 Model compound – salbutamol sulphate
Salbutamol is a short-acting β2-adrenergic receptor agonist. It is used for treatment
of breathing problems in patients who have asthma or chronic obstructive pulmonary
disease (COPD). Usually used in the form of sulphate salt, salbutamol was developed
because its molecule is similar to adrenaline, a natural bronchodilator produced by human
body. During an asthma or COPD attack, the bronchioles that feed air deep into the lungs
become inflamed and narrowed due to muscles in their walls contracting, which is very
dangerous and can be life-threatening. In this case body’s natural hormone, adrenaline
which has a specific shape that helps the molecule fit into active sites on the cells in the
Chapter 1 Introduction
6
muscle walls, will bring about muscles relaxation, thus cause the airways to dilate. This
process is called bronchodilation. Salbutamol, along with other modern bronchodilators,
was therefore developed based on the mechanism that adrenaline works on the
bronchioles.
Salbutamol (as sulphate) became available in the United Kingdom in 1969 and in
the United States in 1980 under the trade name “Ventolin”. It is usually given by the
inhaled route for direct effect on bronchial smooth muscle through a metered dose inhaler
(MDI), nebuliser or other proprietary delivery devices (e.g. Rotahaler or Autohaler).
1.4 The role of surfactants and polymers
It is reported that addition of surfactants or polymers into a growth medium can
have a significant impact on the processes of nucleation of crystallizing phases and
subsequently the growth of the nucleated solids (Rasenack and Muller, 2002; Sangwal,
2007).
Additives influence different processes involved during crystallization, therefore the
understanding the mechanism between additives and the crystallizing phase is of key
importance. In one of the earliest literatures, Buckley (1951) surveyed the effect of
addition of additives on the external appearance of the crystal - crystal growth habit or
morphology. From 1950 onwards, the efforts were diverted to understanding the basic
mechanisms responsible for changes in the growth habit of crystals caused by impurities
on the microscopic level, with topics covering habit modification, kinetic data on the face
growth rates and layer displacement rates, nucleation and precipitation kinetics,
observation of the presence of the dead supersaturation zones at low impurity
Chapter 1 Introduction
7
concentrations, segregation coefficient of impurities, and chemistry of adsorption in
solution growth (Simon and Boistelle, 1981; Kubota, 2001; Weaver et al., 2007;
Poornachary et al., 2008).
Although several studies have been carried out to investigate the mechanism of the
effect of additives, most of the reported studies have focused on small additive molecules.
However, it is known that some additives used for pharmaceutical industrial applications
indeed are large molecules (long chain compounds). Additives having small molecular
dimensions are capable of moving over the surface or can remain immobile after being
adsorbed on the surface, depending on the affinity between additive molecules and the
surface molecules; whereas for larger additives, interaction between additive molecules
and surface molecules are more complex due to the higher possibility of formation of
multiple hydrogen bonds, rending them less mobile. Therefore, a better understanding on
the mechanism of large molecule additives affecting the crystal product is needed.
1.5 Objectives and approach
The objective of this work is to propose a direct approach to produce drug particles
of desired size (between 1 and 5 μm) through antisolvent crystallization. It also aims to
provide a new insight of surfactant or polymeric additives in the formation of lower-
micron crystals, and attempts to understand the mechanism behind it via molecular
modeling. The thesis work will comprise the following milestones:
Determine the solvent and antisolvent by examining solubility, nucleation
time, crystal size and habit for different solvent candidates.
Chapter 1 Introduction
8
Study the effect of surfactants and polymeric additives on SS crystal habit
and size by performing batch antisolvent crystallization experiments in the
presence and absence of additives.
Explain the effect of additives on crystal habit modification by modeling the
interactions between additive molecules and SS molecules at the host crystal
surfaces.
Investigate the effects of operational parameters on particle size distribution
in semi-batch stirred vessel by applying in-situ monitoring tools to the
antisolvent crystallization process with surfactant addition.
1.6 Thesis outline
Chapter 2 introduces the fundamental knowledge of crystallization and the role of
the additives in crystallization process in both experimental and theoretical view.
Chapter 3 describes the model system, experimental setup and experimental
procedures. It also reviews characterization techniques which include off-line
measurement methods and Process Analytical Technology (PAT) used in semi-batch
crystallization experiments. Calibration results of Attenuated Total Reflectance Fourier
Transform Infrared (ATR-FTIR) are also presented in Chapter 3.
Chapter 4 reports the results on screening antisolvents via studying particle
morphology, particle size and agglomeration of crystal products. It also compares
solubility, nucleation time and product yield among candidate solvents in order to
determine the proper antisolvent used in the subsequent experiments.
Chapter 1 Introduction
9
Chapter 5 describes the effect of surfactants and polymeric additives on SS particle
size and habit. It reports the growth inhibition effect of polyvinylpyrrolidone (PVP), the
most effective additive among those screened during antisolvent crystallization of SS.
Further investigation on the effect of concentration and molecular weight are also
discussed. The Mechanism by which additive molecules interact with SS crystal surfaces
are proposed on the basis of the molecular modeling results.
Chapter 6 is devoted to an in-situ investigation of an unseeded semi-batch
antisolvent crystallization of SS. It screens key operating factors in terms of particle size
and particle size distribution (PSD) of crystal products. Effect of the addition of PVP on
crystallization kinetics is discussed.
Chapter 7 consists of conclusions and suggestions for future work.
Chapter 2 Crystallization Fundamentals and the Role of Additives
9
Chapter 2
Crystallization Fundamentals and the
Role of Additives
As a main separation and purification process in production of bulk commodity
chemicals, specialty chemicals and pharmaceuticals, crystallization influence the
attributes of the product crystals via various operating conditions. Well understanding
mechanisms of nucleation and crystal growth, the effect of additives, and the influence of
the various operating conditions thus is indispensable before the research work starts.
This chapter reviews these fundamental aspects and the role of the additives in
crystallization in the context of the current work.
2.1 Solubility, supersaturation and solvent selection
Solubility and supersaturation can be demonstrated using a typical phase diagram
shown in Fig. 2-1 (Mullin, 2001). Solubility is an equilibrated status on which a solution
exhibits a constant concentration after prolonged contact with a solid solute in a system at
constant temperature and pressure. Such a solution is said to be saturated. When the
solute concentration exceeds its solubility limit, the solution is supersaturated. A solution
may remain its supersaturation over a concentration range for a certain period without the
Chapter 2 Crystallization Fundamentals and the Role of Additives
10
formation of a secondary phase. This region is called the metastable region. Metastable
boundary is defined as the point at which the supersaturation reaches to a certain level
that nucleation occurs spontaneously.
Figure 2-1 The solubility/ supersaturation phase diagram.
It is worth noting that, first, not all the solubility of a solute in a solvent increases
with temperature. Exceptions can be found in solubility of sodium chloride which
increases only slightly with an increase in temperature (Mullin, 2001). Such a system is
obviously not suitable for cooling crystallization. Second, not all solubility curves are
smooth. A sudden change in the solubility curve indicates a phage change. There are
many forms used to express solubility, for example, kg solute/ kg solution, kg solute /kg
solvent, kg solute/ m3 solution, kg solute/ m3 solvent or mol solute/mol solution etc.
Solubility in this work is reported on solvent basis, that is, the mass of solute in unit mass
or volume of solvent, g/kg or mg/mL.
As the driving force for a crystallization process, supersaturation is defined
thermodynamically as the difference in free Gibbs energy between the actual equilibrium
conditions of the system:
Chapter 2 Crystallization Fundamentals and the Role of Additives
11
µ µ µ 2 1
where and are the chemical potentials in solution and in the bulk of the crystal
phase. When ∆ 0, the system is supersaturated and nucleation could occur either
homogeneously or heterogeneously. It can be written as:
∆ 2 2
where a and a* are activity of solute in actual and saturated solution respectively, R the
gas constant, T the absolute temperature and Sa the supersaturation (Davey and Garside,
2000).
Since it is practically difficult to measure activity, supersaturatoin is usually
expressed by the comparison of concentrations in the following ways:
1. Concentration driving force: Δc c c , where and c and ceq represent the
actual and equilibrium concentration of the solute;
2. Supersaturation ratio: . For an ideal system, this expression is
equivalent to the Eq. 2-2.
3. Relative supersaturation: ∆ 1.
Solvent plays an important role in the crystallization process and its selection could
very well determine the success or failure of the operation (Wankat, 1990; Myerson,
2002). Solubility is usually considered as the criteria for solvent selection in
crystallization process. Nass (1994) described a strategy for choosing crystallization
solvents based on equilibrium limits. In this method, activity coefficient in given solvent
system at the saturation point was predicted by a group-contribution method (UNIFAC)
Chapter 2 Crystallization Fundamentals and the Role of Additives
12
and subsequently used to predict the solubility, which determines the maximum
theoretical yield for the process. These quantities can be used to rank order solvents
and/or their mixtures relative to one another according to their solvent power and
potential process yield. Frank et al. (1999) reviewed strategies for solvent selection for
various types of crystallization processes such as cooling crystallization and drowning
out crystallization. An example (shown in Fig. 2-2) was given in identifying potential
solvent/antisolvent pairs using the solubility estimation methods of Extended Hansen
Model (Modarresi et al., 2008) by paring the best and poorest candidate solvents.
Figure 2-2 Example solubility curves for drowing-out crystallization.
Figure 2-2 Example solubility curves for drawing-out crystallizations.
Certain solvents may affect crystal morphology due to favorable molecular interaction at
certain faces of crystals. Undesired crystal morphology caused by this can reduce the
efficiency of the filtration step, alter purity of product, and create problems during drying,
packaging, handling and storage. Chen et al. (2008) reported morphology changes of
hydrocortisone (HC) molecule when using methanol, pyridine, acetone, dichloromethane,
and isopropanol as solvents. It is founded that the solvent molecules could enter the
crystal lattice and form HC solvates, thus HC grow into different habits depending on the
Chapter 2 Crystallization Fundamentals and the Role of Additives
13
certain solvent that entered the inner crystal lattice as well as the external crystallization
conditions. Solvent composition may also affect polymorphic stability and transformation
rate. Kitamura and Nakamura (2002) observed that the relative stability of various
polymorphs of a thiazole derivative pharmaceutical (BPT) changed with solvent
composition, as indicated by their solubility and transformation behavior at different
solvent compositions. Moreover, safety and toxicity characteristics of the solvents are
also very crucial.
In pharmaceutical industry, trial and error decision process is still common when
selecting solvents for a new process. This method is costly and time-consuming.
Therefore there is a need to develop an effective knowledge-based alternative to trial and
error approach for finding optimal solvent, by considering factors such as solvent effect
on crystal morphology, solvent ability to solubilize impurities, solvent inflammability,
solvent toxicity, etc., as the criterion. One approach (Karunanithi et al., 2006) is the
computer-aided molecular design (CAMD), which is a computational framework for the
design and/or selection of chemicals with desired properties. The result for promising
solvents obtained through CAMD can serve as a guide to experimentation.
2.2 Nucleation
Two major stages are involved in the crystallization process - nucleation and crystal
growth. Nucleation is the first step in which the crystal nuclei or clusters from the
supersaturated solution generates. Subsequently, the stable crystal nucleus grows in size.
Nucleation can take place through a variety of mechanisms (Fig. 2-3), i.e. primary
nucleation from homogeneous or heterogeneous solutions, or secondary nucleation.
Chapter 2 Crystallization Fundamentals and the Role of Additives
14
Some of these are driven essentially only by free energy considerations; while some are
heavily dependent on imposed conditions (Randolph and Maurice, 1988; Davey and
Garside, 2000).
Figure 2-3 Nucleation mechanisms (Randolph and Maurice, 1988).
2.2.1 Homogeneous nucleation
The classic nucleation theory developed by Gibbs, Volmer, Nielsen and others
(Mullin, 2001) assumes that during the cluster formation process, the free energy is a
balance between that for formation of the nucleus surface (a positive value, ΔGs) and that
for phase transformation (a negative value ΔGv). A stable critical cluster is formed when
the intermolecular force accompanying the addition of new molecules competes with the
force required to extend the surface area of the aggregate. If the hypothetical nucleus is
smaller than the critical size (denoted by r*), the energy that would be released by
forming its volume is not enough to create its surface. Therefore there is high possibility
Nucleation
Secondary
Primary
Homogenous
Heterogeneous
Contact
Needle
Attrition
Fracture
Shear
Chapter 2 Crystallization Fundamentals and the Role of Additives
15
that nucleation does not proceed. However, when the nucleus’ size r is equal to r*, the
nucleation proceeds. This can be expressed as:
ΔG 4πr γ43πr G 2 3
where the first term stands for the energy loss due to surface tension of the new interface
and the second term shows the energy gain of creating a new volume, is the interfacial
energy between solid and liquid. The critical clusters are formed when
0 2 4
where
2 2 5
and free energy reaches its maximum value:
∆16
3 ∆4
3 2 6
Gibbs-Thomson equation describes the relationship between supersaturation and a
stable nucleus size:
2 2 7
where is the Boltzmann constant and υ is the molar volume of the crystal. Substituting
Eq. (2-7) into Eq. (2-6),
∆16 υ
3 2 8
Chapter 2 Crystallization Fundamentals and the Role of Additives
16
The rate of homogeneous nucleation of a sphere, defined as the number of nuclei
formed per unit volume per unit time, can be expressed in the form of Arrhenius reaction
rate equation:
∆ 16 υ
3 2 9
where A is a prefactor containing solute diffusivity, supersaturation, molecular diameter
and volume, equilibrium solute concentration, and Avogadro’s number. In one of the
earliest systematic studies on crystallization kinetics, Nielsen (1964) theoretically
predicted the value of A (~10 ).
The classical nucleation theory has a few shortcomings because of the
oversimplifications of crystallization system by making few assumptions. Firstly, the
classic thermodynamic considerations may not applicable to small nucleating clusters as
it is based on averages of larger groupings of molecules. Secondly, the theory assumes
that one molecule adds on the growing cluster at one time, which is not true considering
the agglomeration and rearrangement of the clusters while they are growing into the
critical size. For organic molecules with complex lattice held together with by relatively
weak forces, the molecules might not be able to orient themselves properly before the
addition of molecules to the forming solid phase, leading to an amorphous crystal
structure. The slow surface integration rate might be the explanation for the experimental
results obtained by Mahajan and Kirwan (1996), who found that their result of A value is
orders of magnitude lower than Nielsen’s.
Chapter 2 Crystallization Fundamentals and the Role of Additives
17
2.2.2 Heterogeneous nucleation
Foreign particles or undissolved solute presented in a solution may act as “catalysts”
of nucleation. This phenomenon is called heterogeneous nucleation. According to Söhnel
and Garside (1992), a liquid from which physical impurities have been removed still can
contain from 1013 to 1023 particulate contaminants per 1 m3. These particles in
supersaturated solutions reduce the energy required for nucleation, therefore in Eq. (8-
2) , and thus . Reduction of critical free energy in
heterogeneous nucleation leads to a greater tendency towards as compared to
homogeneous nucleation at the same supersaturation when there are favorable surface
interactions between the solute cluster and the substrate. The wettability of the surface of
a foreign solid may have effect on the ability of the clusters to form on this foreign
particle. Theoretically a contact wetting angle of 0° should result in spontaneous
nucleation, although the occurrence is impossible in reality.
Fig. 2-4 depicts typical nucleation regions for a fast precipitation system where
abecarnil crystallized from isopropyl acetate by addition of hexane as the antisolvent
(Beckmann, 1999). The addition and mixing time was small as compared to the induction
time. By studying relationship between induction time and supersaturation, two regions
of nucleation were discerned, homogeneous nucleation at high supersaturation levels and
heterogeneous nucleation at low supersaturation levels. The distinct change in slopes
indicated a change in the mechanism of nucleation from heterogeneous at low to
homogeneous at high supersaturation levels.
Chapter 2 Crystallization Fundamentals and the Role of Additives
18
Figure 2-4 Induction time vs. supersaturation for antisolvent crystallization of Abecarnil.
The knowledge of the heterogeneous nucleation can be utilized to control
polymorphism when nucleation of a certain polymorph is favored against others provided
the substrate topography and lattice parameters fit in with the structure of its pre-
nucleation aggregates.
2.2.3 Secondary nucleation
Nucleation of new crystals which are induced only because of the prior presence of
crystals of the material is termed secondary nucleation. The prior crystals could be
already presented or deliberately added into the system, acting as catalysis and thus result
in the much lower degree of supersaturation required than the homogeneous nucleation or
even heterogeneous nucleation. A number of different mechanisms can be used to explain
secondary nucleation (Myerson, 2002).
Initial or dust breeding are generated from the crystals formed on parent crystals
during the growth period or broken off in storage. These crystallites are able to grow into
new crystals once their size exceeds the critical size. If this type of nucleation is not
desirable, i. e. for seeded batch crystallization, an appropriate pre-treatment is needed to
Chapter 2 Crystallization Fundamentals and the Role of Additives
19
remove the nuclei. Needle breeding occurs when the dendrite crystals formed at higher
levels of supersaturation fragment in the solution and serve as nucleation sites.
Microabrasion of crystals at high stirring speeds can produce fragments that serve as
nucleation sites. It is usually called collision or attrition breeding. Since it has a clear
mechanism, it can often be controlled with suitable choices of mixing impellers and other
parameters. Contact nucleation occurs due to the crystal-crystal, crystal-stirrer and
crystal-crystallizer collisions. It is the most common type of secondary nucleation. It can
be expressed by the following equation:
∆ 2 10
where J is the nucleation rate in number per unit volume per second, Δc the concentration
difference, Ms the suspension density, N the mixing rate in rpm. j, k, b are the exponents
for suspension density, agitation rate and concentration driving force respectively.
2.3 Crystal growth
Crystal growth occurs after the formation of the stable nuclei in the system.
Compared to the nucleation, the real growth process is much faster since the crystals
contain dislocations which provide the necessary growth points.
Mullin (2001) categorized crystal growth mechanism into surface energy theories,
adsorption layer theories, kinematic theories and diffusion-reaction theory. The surface
energy theories are based on the assumption that crystals always grow into the shape with
minimum surface energy. This approach is debatable because there is lack of quantitative
evidence to support it and it is failed to explain the effects of supersaturation and solution
movement on the crystal grow rate. Kinematic theories were developed based on the
Chapter 2 Crystallization Fundamentals and the Role of Additives
20
adsorption layer theories. It considers the movement of macrosteps of unequal distance
apart by assuming the generation of steps at some source on the crystal face is followed
by the movement of layers across the face. The other two approaches which have gained
more common acceptance will be discussed in details in the following sections. This
thesis reviews the nucleation mechanism by sorting them into three parts: diffusion
controlled, integration controlled and diffusion-integration controlled growth.
2.3.1 Diffusion controlled growth
Before solute is integrated into the crystal lattice, it must be desolvated and the
solvent must diffuse away from the surface. For a system in which both sparingly soluble
material and highly soluble material exist, however, it is hard to determine the
dominating manner in which the diffusion or integration takes places. Adsorption layer
theories assume that when crystallizing species arrive at the crystal surface, they are not
immediately integrated into the lattice, but merely loose one degree of freedom and freely
move over the surface. This loose absorbed layer of the integrating units at the interface
acts as a “third phase”, playing an important role in crystal growth and secondary
nucleation (Fig. 2-5). Mullin (2001) estimated the thickness of the adsorption layer of
around 100Å.
Chapter 2 Crystallization Fundamentals and the Role of Additives
21
Figure 2-5 Adsorbed layer of solute on the surface of a growing crystal (□, solute molecules; ○ and Δ solute ions) (Clontz and McCabe, 1971).
If the crystal growth rate is limited by the rate of diffusion through the laminar lay
of solution, the mechanism of crystal growth is diffusion controlled. The growth rate can
be expressed by the simple mass transfer equation
2 11
where K is a mass transfer coefficient, A is the surface area of the crystal, c and cs are the
actual and saturated concentration respectively. M is the mass of the crystal, which can be
replaced by ρkvL3, where ρ is crystal density, kv is volume shape factor and L is a
characteristic diameter of the crystal. Crystal surface can be expressed by kaL2, where ka
is an area shape factor. Using these expressions, Eq. 2-11 then becomes
3 2 12
or
2 13
Diffusion
Growing crystal
Supersaturated bulk solution
Chapter 2 Crystallization Fundamentals and the Role of Additives
22
2.3.2 Integration controlled growth
When a molecule or cluster arrives at a surface, there are three possible occurrences
of the attachment. It can attach at the surface with (1) only one attachment, (2) two
attachments (both the surface and a growing step), and (3) three attachments (a “kink”
site). The latest case is the most favourable bonding type for crystal growth. There are
many theories describing the integration of molecules occurred on the crystal surfaces.
These theories can be categorized as:
Continuous growth model. The continuous growth model (Fig. 2-6a) assumes that a
flat crystal surface may contain one or more kinks on the moving layers on which solutes
(atoms, molecules or ions) are loosely adsorbed. These kinks may easily “attract” solutes
while moving along the step and the face is eventually completed, followed with
generation of fresh step by surface nucleation. Under these conditions, the growth rate is
linearly proportional to the supersaturation.
Birth and spread model. For the majority of the substance, there may not be enough
roughness on the surface of the crystal as described in the continuous growth models.
Therefore in this case steps must be created to accept the incoming growth units. A two-
dimensional nucleation mechanism, the birth and spread model, was developed based on
crystal surface nucleation and followed by the spread of the monolayer. As shown in Fig.
2-6b, the nucleation occurs at the faces or edges of a crystal and growth develops on the
surface two-dimensionally. Further nuclei might occur on the monolayer nuclei when
they spread across the surface of the crystal.
Screw dislocation model. For the cases of low supersaturation in practice, the above
two models which assume two-dimensional surface nucleation underestimate the result as
Chapter 2 Crystallization Fundamentals and the Role of Additives
23
compared with the observation. The problem was solved by Frank (1949) who postulated
that most crystals contain dislocations which cause steps to be formed on the faces and
promote growth. The screw dislocation is important in crystal growth because once it is
formed, the crystal face can grow perpetually without necessities of fulfilling the ideal
layer-by-layer growth manner. The growth will continue uninterruptedly at near the
maximum theoretical rate for the given level of supersaturation (Fig. 2-6c).
Figure 2-6 Crystal growth model. (a) Continuous growth, (b) birth and spread, (c) screw dislocation.
2.3.3 Diffusion-integration controlled growth
For the system where both diffusion and integration affect the overall growth rate,
solute molecules are firstly transported from the bulk solution to the solid surface, and
then integrate themselves into the crystal lattice (Mullin, 2001). Assuming that the
surface reaction is linear, these two stages with concentration difference with driving
force can be represented conceptually as shown in Fig. 2-7. They can also be expressed in
the equations below:
Surface nucleus
Chapter 2 Crystallization Fundamentals and the Role of Additives
24
diffusion 2 14
and
reaction 2 15
where kd is a coefficient of mass transfer by diffusion and kr a rate constant for the surface
integration process. ci is the solute concentration at the crystal-solution interface. Since it
is difficult to measure ci in practice, the above equations can be combined into a overall
expression
2 16
where KG is an overall crystal growth coefficient. The exponent g is usually referred to as
the order of the overall crystal growth process. It is worth nothing
that unlike the one in the chemical kinetics, this exponent here has no fundamental
significance and does not give any information of the number of the elementary species
involved in the process.
Figure 2-7 Concentration driving forces in diffusion-integration controlled growth.
Crystal
Adsorption layer
Driving force for diffusion
Driving force for reaction
Crystal-solution interface
Bulk solution
C
Ci
C*
Con
cent
ratio
n
Chapter 2 Crystallization Fundamentals and the Role of Additives
25
Taking into account other operating factors, an empirical power-law relationship
expressing the overall specific rate of mass deposition is found to be most useful:
11
∆ 2 17
where ω is slurry voidage. Slurry voidage and agitation intensity may influence the
turbulence and the relative solution velocity in the suspension. The effect of temperature
can also be incorporated into the expression by an Arrhenius relation if necessary.
The value of exponent g in Eq. 2-17 has been found experimentally to be smaller
than that of b in Eq. 2-10 for nucleation rate, showing that nucleation rate is more
sensitive to supersaturation than crystal growth. This sensitivity difference to
supersaturation level between nucleation rate and growth rate is the basis of PSD
engineering.
2.4 Effect of additives
Additives affect different processes involved during crystallization. Therefore
understanding mechanisms by which additives presented in solution affect solubility,
nucleation and crystal growth is of essential importance.
2.4.1 Solubility
The effect of additives on solubility of the compound has received less attention as
compared with that on crystal nucleation and growth. However, the change in solubility
may play a significant role affecting the parameters such as supersaturation, and thus
affecting crystal growth kinetics.
Chapter 2 Crystallization Fundamentals and the Role of Additives
26
The solubility of a solute in a solvent is temperature dependent. For an ideal
solution, the relation between solubility and temperature can be expressed as follows:
∆1 2 18
where ceq is the solubility expressed in mole fraction. ΔHm is the enthalpy of melting
which depends on the solvents used and Tm is the melting point of the solute. The
solubility of a real solution bears a similar expression where one obtains the enthalpy of
solution ΔHs instead of ΔHm and
∆ ∆ ∆ 2 19
where ΔHmix is the enthalpy of mixing, attributing to deviations from ideal solubility
behavior due to the nature of the interactions between solute ions and solvent molecules.
For real solutions ∆ ∆ and ∆ 0.
Based on the above knowledge, changes in solubility due to the presence of
additives can be interpreted in terms of changes in the values of and ∆ . Fig. 2-8
represents typical solubility curves for ideal and real solutions depending on temperature
according to Eq. 2-18. It can be seen that changes of the value of Tm by 5K can result in
up to 10% changes of solubility values.
Several other factors are also known to have effect on solubility of different
compounds, e. g. common-ion effect, effect of foreign ions, effect of pH and formation of
complexes. These effects can also be attributed to the changes of the enthalpy of mixing
∆ that is determined by the bond energy.
Chapter 2 Crystallization Fundamentals and the Role of Additives
27
Figure 2-8 Solubility curves of ideal (curve 1) and real (curve 2) solutions. Curve 3 is the metastable curve.
Different concentration of additives can also result in different solubility value. In
general, the solubility of a compound increases with increase in additive concentration.
For example, the presence of bi- and trivalent cationic additives could affect solubility of
ammonium oxalate in this manner (Sangwal, 1996). However, the solubility can also be
reduced with the increase of the concentration of additives, depending the additive
properties and solution temperature. In a study of solubility of guaifenesin in water in the
presence of salts, sugars, and cosolvents as additives, the solubility of drug was found to
reduce with increasing concentrations of salts and sugars, where the reduction extent
depended on the type of the additive used (Mani et al., 2003).
2.4.2 Crystal nucleation
The presence of additives in the solution may affect the crystal nucleation. In
general, the change of the nucleation rate is attributed to the changes in solid-liquid
interfacial energy (Sangwal, 2007). Additive molecules adsorbed on the nucleus surface
decrease the interfacial energy, therefore leading to an increase in nucleation rate.
Tm=545
Tm=550K
Chapter 2 Crystallization Fundamentals and the Role of Additives
28
However, this mechanism is limited by the extremely short lifetime of nuclei. Adsorption
may not be happened as there are not enough time for additive molecule to attach on the
surface of the near-critical nuclei clusters. In this case, the nucleation rate may not be
affected.
According the Langmuir adsorption isotherm, a three-dimensional nucleation rate
corresponding to the additive concentration can be expressed as:
′/ 1 exp 2 20
where is the concentration of additives, ′ is the kinetic factor that accounts for the
mechanism of attachment of growing units to the nucleus and K is the Langmuir constant
given by exp . is the gas constant and is the differentical heat of
adsorption corresponding to the impurity coverage θ of the available adsorption sites
(Sangwal and Mielniczek-Brzoska, 2004). Apart from interfacial energy, kinetic factor
may be another changing factor when additives are applied in the solution. An additive
molecule that physically block the existing active sites for the attachment of growth units
can cause the change of ′, leading to the change of nucleation rate.
Kashchiev and Firoozabad (2002) proposed a more specific model explaining how the
changes in supersaturation (∆ ), effective surface specific energy ( and kinetic factor
( ′) affect the crystal nucleation:
′ ∆exp 4 /27 ∆ 2 21
where c is a numerical shape factor for spherical clusters and υ is the volume of a
growing unit. When additive concentration is sufficiently low, additive exerts
practically no effect on the chemical potentials solute molecules in the solution and/or
Chapter 2 Crystallization Fundamentals and the Role of Additives
29
crystal. Therefore ∆μ in Eq. (2-21) remains unchanged and as a result, the additive
cannot change the nucleation rate J through the supersaturation.
However, unlike the supersaturation, the effective specific surface energy σ can be
affected by the additive in solution through adsorption of the additive at the
nucleus/solution, nucleus/substrate, solution/substrate interfaces. The surface energy can
be considerably reduced due to the adsorption, leading to an increasing nucleation rate. In
this case the additive is used as a kinetic promoter instead of an inhibitor. This is
supported by a recent observation on the effect of additives on nucleation rate (Gosavi et
al., 2009). It should be noted that for some systems with very short nucleation time,
where additives may not able to reach onto the nucleus surface by diffusion, the additive
adsorption may not happen, causing no effect of the value of σ . However in this case,
the additive could still act as a nucleation promoter because it may provide new
nucleation sites in the solution (van der Leeden et al., 1993) and therefore enhance
secondary nucleation by initiating crack propagation at the adsorbed defect sites on the
crystal surface fostering the nucleation event (Kashchiev, 2000). In some cases,
impurities may enhance the degree of cluster formation in a supersaturated solution, and
therefore, promote crystal nucleation (Myerson and Lo, 1991; Ginde and Myerson, 1993).
As for kinetic factor, in principle, it can be reduced because when additives adsorb
at the solute and solution interface or onto the surfaces of the micro-particles and the
solid substrates in the solution, the surface area or the number of active sites available for
nucleation is reduced. Thus in this case, the additive acts as a kinetic inhibitor retarding
nucleation process. Based on the above argument, it is possible to quantitatively evaluate
the overall effect of additive on nucleation rate.
Chapter 2 Crystallization Fundamentals and the Role of Additives
30
The effect of additives on crystal nucleation is often reflected by changes in the
measured values of metastable zone widths and induction periods (Arunmozhi et al.,
1997; Dhanaraj et al., 2008). However, it has not been possible to attempt a general
explanation for the effects on the nucleation step. Many of the proposed hypotheses are
based on an adsorption-based mechanism discussed earlier, at which the adsorbed
additive can act as crystal nuclei and block the active growth sites, leading an increase in
the supersaturation barrier for homogeneous nucleation, or, it can increase the interfacial
tension. In both these cases, the metastable zone widths (or the induction periods)
increase in solutions doped with the additives.
2.4.3 Crystal growth
Additives presented in a growth medium can influence crystal growth. Even trace
amount of impurities may have profound effect on the growth rate of some industrially
important compounds. In general, the effect of foreign substances on crystal growth can
be concluded as kinetic effect and thermodynamic effect.
Adsorption of additive substances can change the properties of the crystal-medium
interface such that the interfacial energy γ in the growth models is changed. The change
in interfacial energy results in a change in nucleation rate in the two-dimensional
nucleation models, and the radius of critical two-dimensional nucleus and hence the
spacing between the spiral steps in the BCF theory. Adsorption of additive substance can
either increase or decrease crystal growth rate. The majority of the experimental
observations on the effect of the additives on the crystal growth are that the additives
retard the crystal growth. In this case the interfacial energy is assumed to be increased by
Chapter 2 Crystallization Fundamentals and the Role of Additives
31
the adsorption of the foreign substances on the crystal surface (Sangwal, 2009). However
since the adsorption process involves adsorption isotherms and reversible adsorption
equilibrium, adsorption of additives may also decrease the interfacial energy value,
leading to an increase in growth rate (Ohara and Reid, 1973; Davey, 1979).
Adsorption of additives can change the kinetic parameters associated with the
density of kinks and obstructions provided in the movement of steps on the surface by
impurity particles (Sangwal, 2007) . It is assumed that additives (ions, atoms or
molecules) wandering over the growing surface can preferentially attach on three sites
(Fig. 2-9) and disrupt the flow of growth layers across the faces, viz. at a kink, at a step or
on a ledge between steps. This can retard the velocity of motion of free energy by
adsorbing on the growing surface, and hence affect the growth rate of the crystals.
Figure 2-9 Adsorptions of impurities on (a) kink, (b) step and (c) ledge (Tung et al., 2009).
A number of research works relate to the adsorption of impurities or additives on
the distinct sites. Changes in step morphology and growth mode on the (100) face of an
adipic acid in the presence of octanoic acid were observed through AFM imaging (Keel
et al., 2004). In the presence of relatively low levels of the additives, growth of the face
(100) was not influenced even though there seemed to be an increased level of stress
within the lattice. For this case, single molecular steps dominated the crystal surface, with
Chapter 2 Crystallization Fundamentals and the Role of Additives
32
the macrosteps becoming less. Therefore impurity concentration was not sufficient to
completely stop tangential growth, although the step edges moved noticeably slowly.
This is probably due to the strongly adsorbed water layer rejecting the majority of the
more hydrophobic octanoic acid molecules. When the additive concentration was
increased, the development of the (100) face was changed in terms of step morphology
and growth mode, as a result of the impurity binding to the surface and pinning the
monomolecular growth steps. Fig. 2-10 demonstrates the comparison of adsorption of
impurity between low and high impurity concentrations.
Another study on crystallization of lysozyme in commercial and purified solutions
in the presence of impurities shows that impurity effects on the (101) surface as against
the (110) surface (Nakada et al., 1999). Several typical residual impurities contained in
commercial lyzozyme were examined and only covalently bound lysozyme dimer was
found to affect the (101) step morphology. Mechanism at which the impurities might
reduce the step velocity was deducted through measuring separation distances of
adsorbed impurities on the surface and estimating the critical nucleation size.
The effect of structurally-related additives (acetanilide and metacetamol) on
development of paracetamol crystals were studied topographically (Thompson et al.
2004). The corrugated steps on the (001) face formed as compared to the surface
obtained from pure solution indicated adsorption of impurity molecules onto step or
terrace regions which caused pinning of the growth steps. For the case of acetanilide as
the impurity, it was found that holes emerged on the surface and subsequently deepened
towards the crystal core. They also spread out laterally as a consequence of dissolution of
the face indicating that the impurity caused an increase in the solubility of paracetamol.
Chapter 2 Crystallization Fundamentals and the Role of Additives
33
However in the presence of metacetamol, the growth steps on the (001) face acquired a
pointed appearance due to pinning and resulted in decreased step velocity.
2.5 Molecular modeling
Crystal habit is the shape adopted by crystal according to the proportions occupied
by each crystal face (Jones, 2002). On crystal surfaces two-dimensional molecules are
highly ordered and arrayed, with functional groups exposing onto them. The specific
functional groups dominates surface properties, and, influences bulk properties including
wettability, dissolution rate, flowability and tablettability of the crystals (Muster and
Prestidge, 2002; Abendan and Swift, 2005). The effect of additives on the crystal habit is
in fact the interaction of the additive molecules with crystal surface. Therefore, it is very
important to learn the knowledge of the surface structure at a molecular level in order to
understand mechanism of additive effect on crystal habit, and furthermore, potential
controlling of pharmaceutically important bulk properties for crystalline pharmaceutical
solids.
Molecular modeling is a powerful approach which involves the prediction of
growth morphology of crystals from its molecular crystal structure (Brunsteiner and Price,
2001). It is also feasible to study habit modification in the presence of additives by
modeling the interaction of additives with the crystal faces of the growth morphology,
and subsequently, by calculating their interaction energies within the host crystal. Two
methods are usually used to predict the crystal growth morphology of an organic material
- BFDH (Bravais-Freidel-Donnay-Harker) method and the Attachment Energy method.
Both methods require two of the basic information derived from its molecular crystal
Chapter 2 Crystallization Fundamentals and the Role of Additives
34
structure. The first is the symmetry constraints (the point group) in the asymmetric unit,
and the second, the inter-atomic distances between the molecules. Both methods assume
that a (hkl) crystal surface grows by the addition of a complete molecular layer of
thickness dhkl (i.e. the interplanar spacing along [hkl] direction) having the same
orientation and structure as the bulk crystal structure. In addition, the energy released due
to the addition of the layer (i.e. attachment energy) is directly proportional to the growth
rate of a given face (Bennema, 1995).
2.5.1 Bravais-Friedel-Donnay-Harker (BFDH)Theory
Initial work on predicting crystal morphologies was based on the assumption of the
lowest growth rates occurring at the faces with the largest interplanar spacing (Friedel,
1907; Bravais, 1913). Crystal symmetry was then subsequently taken into account by
reducing the interplanar growth slices at the faces containing a screw axis or glide plane
as the slice at these faces can be reduced into identical subslices (Donnay and Harker,
1937). BFDH theory, a purely structural approach, assumes that the relative growth rate
(or interaction energy) of faces (hkl) is inversely proportional to the interplanar spacing,
and thus their morphology importance, i.e. those faces with the slowest growth rates (or
lowest interaction energy) are the most morphologically important.
However, this method takes into account only cell parameters while ignoring the
nature of the chemical interactions between the molecules in the lattice. This is probably
the reason of inconsistence of the predicted and experimental crystal morphologies. For
example, the morphology of urea crystals predicted using this method is significantly
different as compared with the one from vapor (Bisker-Leib and Doherty, 2001). The
Chapter 2 Crystallization Fundamentals and the Role of Additives
35
predicted shape is elongated along the b direction, exhibiting faces {110}, {101}, {001},
and {011}. In contrast, the observed morphology is prismatic, elongated along the c
direction and bounded by faces {110}, {001}, and {111}.
2.5.2 Attachment Energy (AE) Theory
The model was proposed by Hartman and Perdok (1955), who considered the role
of intermolecular forces in crystal growth morphology. In Hartman-Perdok theory, crystal
faces were classified into three types: F (flat) face, S (step) face, or K (kink) face,
depending on the number of periodic bond chains (PBCs) (i.e. uninterrupted chains of
“strong” bonds), in the growth slice. F faces refer to the slices with two or more PBCs,
and S faces are those with one PBC, while the rest are K faces.
The theory developed the concepts of the “slice” and “attachment” energies. The
slice energy ( ) is the energy released on the formation of a stochiometric growth slice,
and the attachment energy ( ) is the energy released when the slice is attached to the
surface of the growing crystal. The relation between the crystallization energy Ecr (or
lattice energy), slice energy and attachment energy is given by
2 22
where the crystallization energy equals the sublimation enthalpy of the substance
corrected for the difference between the gas-phase enthalpy and the solid-state enthalpy
(Bisker-Leib and Doherty, 2001).
The interaction energies
between the individual non-bonded atoms of the
molecules that constitute the crystal are usually calculated in terms of van der Waals and
electrostatic contributions. It is worth noting that the attachment energy predictions are
Chapter 2 Crystallization Fundamentals and the Role of Additives
36
sensitive to choice of force field potentials. Brunsteiner and Price (2001) showed that
variations in both the repulsion−dispersion parameters and in the accuracy of the
electrostatic model (including the atomic point charges) also affected the predictions.
Therefore, the accuracy of an intermolecular potential to model morphologies of organic
crystals should be verified by comparing the computed values of the lattice energies with
the experimental values of sublimation enthalpies.
2.5.3 Equilibrium Morphology (SE model)
SE model is an alternative method to predict crystal morphology. It involves the
calculation of surface energy, the difference in energy of the surface ions compared with
those in the bulk per unit surface area. The surface energy is defined as
⁄ 2 23
where n is the number o unit cells on the surface, A(hkl) is the surface area, Etotal is the
total energy of the system, Eboundary is the difference between the surface atoms and the
bulk atoms. Since the surface free energy must be the minimum for a crystal of give
volume in equilibrium with its surroundings, the morphological importance of a face is
inversely proportional to the surface energy.
Coombes et al. (2002) compared the aforesaid three prediction modeling techniques
by studying morphologies of the three polymorphs of a pharmaceutical compound
BRL61063. Their results suggested that the growth morphology models (BFDH and AE)
were more accurate in representing the observed morphologies as compared with the SE
model, which tended to overestimate the importance of the higher index faces. Fig. 2-10
demonstrates the example of the comparison of the predicted and observed morphology
Chapter 2 Crystallization Fundamentals and the Role of Additives
37
of form A. It should be noted that in this work, the effect of the solvent interactions on
the surfaces of the crystals were not taken into account, which might be limited use for
certain solution crystallizations in which solute-solvent interaction is strong.
Figure 2-10 Comparison of predicted and observed morphology (Coombes et al.,
2002).
Berkovitchyellin (1985) considered solvent effects in calculating the habit of
organic crystals (benzamide, benzoic acid, α-glycine, cinnamide and succinic acid) using
the attachment energy method. Calculated morphologies were in good agreement with
observations for crystals grown by sublimation. Solvent effects on the habit of crystals
were explained through preferential adsorption of solvent molecules on specific crystal
faces, whose relative polarities were comprehended from the electrostatic potential maps
at closest approach distances.
2.5.4 Impurity interactions: Binding energy
A factor called “differential binding energy” is introduced in order to assess the
likelihood of impurity incorporation into the host crystal lattice. In doing so the effects of
additives (or impurities) on the growth morphology can be studied through the above-
mentioned methods for calculating surface attachment energies. The differential binding
Chapter 2 Crystallization Fundamentals and the Role of Additives
38
energy is defined as the difference between the incorporation energies of solute (Eb) and
the impurity molecule (Eb’):
∆ ′ � ′ ′ 2 24
where Esl’ is the slice energy calculated with an impurity molecule at the centre of the
slice and Eatt’ is the attachment energy of a growth slice containing an impurity onto a
pure surface. From Eq. 2-34, it can be seen that the impurities are likely to incorporate on
crystal faces where there is a minimum change in the binding energy. For the case of
differential binding energy strongly depending upon crystal orientation, the incorporation
is specific to one crystal face and vice versa. A direct measure of growth rate of a crystal
face “poisoned” with an additive molecule can be made through introduction of a further
parameter Eatt’’, which reflects the energy released on the addition of a pure growth slice
onto a surface on which an impurity molecule has adsorbed. This enables calculation of
the crystal morphology in the presence of additives. Low value of Δb indicates a
favorable additive adsorption, thus Eatt’’ being used as the measure of growth rate for that
face. In contrast, if Δb is unfavorable, meaning the additive cannot bind here and the host
Eatt value is used instead (Clydesdale et al., 2005).
Clydesdale et al. (2005) have applied the differential binding energy method to
predict habit modification in a number of organic crystals including naphthalene
(biphenyl as the additive), phenanthrene (biphenyl as the additive) and caprolactam
(cyclohexane, cyclohexanol, cyclohexanone and caprolactim as additives). The molecular
modeling procedures included an optimization of the position and orientation of the
adsorbed additive within the host lattice, thus simulating the crystal chemistry of the
hetero-species within the host lattice. As an illustration, the predicted modified
Chapter 2 Crystallization Fundamentals and the Role of Additives
39
morphologies of naphthalene in the presence of biphenyl are shown in Fig. 2-11. The
predicted morphologies suggested that the {110} faces with low differential binding
energy values were the most favorable faces for incorporation of the additive.
Figure 2-11 Predicted morphology of a pure naphthalene morphology (left) and a modified naphthalene morphology in the presence of biphenyl (Clydesdale et al., 2005).
The binding energy model has been used to explain the unidirectional crystal
growth phenomenon. Srinivasan and Sherwood (2005) reported that R-resorcinol crystals
grow unidirectionally at the faces in the vapor phase and suggested that this
phenomenon of unidirectional growth is intrinsic to polar crystals. Weissbuch et al. (2006)
performed binding energy calculations to elucidate a “self-poisoning” mechanism
explaining the unidirectional growth of α-resorcinol in the vapor phase. It was shown that
a molecular misorientation (of the resorcinol molecule) at the growing crystal interface
can act as a tailor-made impurity, and therefore affect growth inhibition.
2.6 Closing remarks
In recent decades, the needs for fine particles have been expanded from ordinary
uses to electronics, biology and pharmacology etc. In pharmaceutical manufacturing, fine
particles can be obtained by choosing proper crystallization techniques, changing
Chapter 2 Crystallization Fundamentals and the Role of Additives
40
operational conditions, and/or applying structure related additives during the
crystallization process. Therefore, understanding of the mechanisms of such procedures is
a requisite so as to design and control robust crystallization processes. Experimental
investigations combined with molecular modeling studies may provide a promising tool
toward a better understanding of interactions between crystal surface and the additive
molecules, among which the knowledge of polymer-crystal interactions is a particularly
useful guild to design low-temperature technologies for biomedical applications.
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
41
Chapter 3
Experimental – In Situ and Ex Situ
Analytical Techniques
In this chapter, offline analytical techniques used in this work to characterize crystal
size, morphology, surface and rheological properties are discussed. Two Process
Analytical Technology (PAT) devices, FBRM and ATR-FTIR, for on-line investigation
are briefly reviewed.
3.1 Ex situ analytical techniques
3.1.1 High Performance Liquid Chromatography (HPLC)
The concentration of SS solution was measured by an Agilent 1100 HPLC system
consisting of an HPLC Agilent 1100 series quaternary pump with degasser and variable
wavelength detector (VWD). The reverse-phase column used was Zorbax® SB-C18 (4.6
mm × 150 mm with a pore size of 3.5m). The mobile phases consisting of 80% v/v
methanol and 20% v/v 0.1M phosphate buffer with pH of 7 were chosen to achieve
maximum separation and sensitivity. The mobile phases were filtered and degassed
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
42
before use. The flow rate was 1 mL/min. Samples were detected at 227 nm using variable
wavelength detector. The system was controlled and data analysis was performed with
Agilent ChemStation. All the calculations concerning the quantitative analysis were
performed with external standardization by measurement of peak areas. Results were
expressed as the mean of three determinations.
Five standard solutions were prepared and applied to HPLC. Firstly, 2.5 mg of
standard SS was dissolved in water in a 25 mL volumetric flask to obtain Solution 1 with
concentration of 1mg/mL. Solution 2 was made by withdrawing 8 mL of solution 1 and
then topping up with water in a 10 mL volumetric flask. Its concentration is 0.8mg/mL.
Solution 3, 4 and 5 with concentration of 0.4 mg/mL, 0.2mg/mL and 0.1 mg/mL
respectively were prepared using the same method except that the dilution times were
twice.
The accuracy of HPLC methods can be inferred from the analytical parameters
used in assay validation such as linearity and precision. In order to establish the linear
detection range for SS, standard stock solutions were analyzed to determine method
linearity. Calibration ranges of 0.05 ~ 1 mg/mL SS solution were prepared. Triplicate 5
µL injections were made for each standard solution to see the reproducibility of the
detector response at each concentration level. The peak area of each drug was plotted
against the concentration to obtain the calibration graph. The six concentrations were
subjected to regression analysis to calculate calibration equation and correlation
coefficients (Fig. 3-1). The equation of the regression line formula is:
y 7453.2x 46.286 3 1
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
43
with correlation coefficient 9999.02 r . The regression analysis indicated that there was
an excellent linearity between the peak area and SS concentrations.
Figure 3-1 High Performance Liquid Chromatography calibration for SS water solution.
The interday precision is defined as relative standard deviation (RSD) calculated
from the values measured from three samples at SS concentrations of 0.1, 0.2, 0.4, 0.8
and 1 mg/mL, respectively, in three different days. The calculated interday precision
(RSD) values ranged from 2.56% to 6.6% indicated that the HPLC method is accurate for
measurement of SS concentrations.
3.1.2 Malvern MasterSizer
The Malvern MasterSizer 2000 (Malvern Instruments Ltd, Worcestershire, UK) is
one of the most widely used particle sizing instruments. It is categorized as a non-
imaging optical system because sizing is accomplished without forming an image of the
particle on a detector. The hydro µP dispersion unit allows small amount of samples to be
used for the characterization. A SOPs (Standard Operating Procedures) driven software is
0
1000
2000
3000
4000
5000
6000
7000
8000
0 0.2 0.4 0.6 0.8 1 1.2
Peak area
Concentration (mg/ml)
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
44
0
20
40
60
80
100
120
140
160
180
0 1000 2000 3000 4000 5000
Particle size (µm)
Pump speed (rpm)
d(0.1)
d(0.5)
d(0.9)
used to control and standardize the complete measurement including analyzing the data
of particle size.
A well dispersed sample is always desirable before measurement is conducted. In
order to obtain a well dispersed sample, a stirring pump and ultrasonics provided in hydro
µP dispersion unit can be utilized. Fig. 3-2 and Fig. 3-3 determine the pump stirring rate
and ultrasonic intensity in measuring particle size in this work respectively. SS used in
the experiment was obtained from same batch of the raw product. Ethanol was used as
dispersants. Each sample was measured at least 3 times, and the average result was used
for further discussion. Particle size distribution is characterized by the d (0.1) (10%
below this size), the d (0.5) and d (0.9) value.
Figure 3-2 Measurement of particle size at various pump speeds.
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
45
0
200
400
600
800
1000
1200
1400
1600
1800
0 10 20 30 40 50 60 70 90 100
Particle size (µm)
Ultrasonics level (%)
d(0.1)
d(0.5)
d(0.9)
Figure 3-3 Measurement of particle size at various ultrasonic levels.
It is found that particle size (d (0.9)) dramatically decreases from 168 to 83 µm
when pump stirring speed increases from 500 to 1000 rpm (Fig. 3-2), being the result of
the effective dispersion of the agglomerated particles. Particle size is then remained
unchanged until the pump stirring speed reaches 3500 rpm, at which the particle size
starts to drop. This is probably due to the breakage the of the needle-shape particles. SEM
measurement was utilized as a reference to facilitate the particle size measurement and
resultant analysis, showing the sample size mainly located between 50-100 µm. The
particle size (d (0.9)) value of 168 µm at ultrasonics intensity of zero indicates that
dispersion is insufficient without ultrasound applied (Fig. 3-4). Ultrasonics level of 30%
seems to be sufficient for a good dispersion. When the ultrasonics level reaches 80%, the
particle size increases dramatically, this is due to the occurrence of the air bubbles at such
a high intensity. Therefore stirring speed of 2500 rpm and ultrasonication level of 30%
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
46
were used in the present work for characterization of particle size using Malvern
Mastersizer 2000.
3.1.3 X-ray Diffraction (XRD)
A well formed crystal is composed of regularly arranged atoms locating on planes
separated by a defined distance d. When a monochromatic X-ray beam (wavelength is λ)
is projected onto the crystal at an angle θ, it scatters into many different directions.
However, diffraction only occurs when the differences in the travel path is equal to
integer multiples of the wavelength. The phenomenon is explained by Bragg’s law (Fig.
3-4):
2 sin 3 2
The Bragg's Law can be solved by varying the angle θ to obtain different d spacings.
When it occurs, it can be defined as the nth diffraction for that particular plane. Therefore
one will observe an x-ray scattering response for every plan defined by a unique miller
index (h, k, l).
Figure 3-4 Schematic illustration of Bragg's law.
Single crystal XRD data was used to determine the unit cell of SS crystal, including
cell dimensions and positions of atoms within the lattice. In this work, the morphology
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
47
of SS can be predicted based on such data. A single crystal having approximate
dimensions of 0.11 x 0.22 x 0.44 mm obtained from evaporation crystallization
experiment was placed on a fibre needle which was then mounted on the goniometer of
the X-ray diffractometer. The crystal was purged with a cooled nitrogen gas stream at
110 K throughout the data collection. X-Ray reflections were collected on a Rigaku
Saturn CCD area detector with graphite monochromated Mo Kα radiation (λ = 0.71073
Å).
In this work, powder X-ray diffraction was applied mainly to examine different
crystal forms and the crystallinity. Powder X-ray diffraction patterns were recorded after
samples were spread uniformly over the sample holder using a D8 Advance powder X-
ray diffractometer with Cu Kα1
radiation of λ = 1.54 Å. (Bruker AXS GmbH, Germany).
The voltage and current applied were 40 kV and 40 mA respectively. The sample powder
was packed into the rotation sample holder and scanned in the 2θ range 10° to 40°, at the
rate of 0.02°/ second. SS morphology was identified by comparing the characteristic 2θ
peaks (“fingerprints”) of the XRD pattern.
3.1.4 Microscopy analysis
The morphology of SS particles was investigated by a scanning electron
microscopy (SEM). The scanning electron microscope (SEM) is a type of electron
microscope that images the sample surface by scanning it with a high-energy beam of
electrons in a raster scan pattern. The electrons interact with the atoms that make up the
sample producing signals that contain information about the sample's surface topography,
composition and other properties such as electrical conductivity. The signals resulting
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
48
from interactions of the electron beam with atoms at or near the surface of the sample can
produce very high-resolution three-dimensional images of a sample, revealing
morphology and surface properties for the crystals.
Samples for SEM analysis must be electrically conductive in order to avoid the
accumulation of electrostatic charge at the surface during electron irradiation. Samples
are therefore usually coated with an ultrathin coating of electrically-conducting material
such as gold or platinum using a coating sputter. In this work, SS samples were mounted
on an aluminium stub on which a double-sided adhesive tape was first placed and after
removing the protective covering, a few particles were scattered on the tape. The particles
were then coated with platinum for 1 minute using an ion sputter coater (Cressington®
208HR, Ted Pella Inc., USA) under a vacuum of 0.09 mbar and a current of 40 mA.
Images were then acquired by a JSM-6700 scanning electron microscope (JEOL, Japan).
The SS crystals obtained from evaporation crystallization experiments were
examined using an Olympus BX51 polarized light microscope. The images were
recorded using Soft Imaging System’s Analysis image capture software.
3.1.5 Inverse Gas Chromatography (IGC)
Inverse gas chromatography (IGC) is a gas phase technique developed for studying
the surface and bulk properties of particulate and fibrous materials. Nowadays it has been
increasingly used throughout the pharmaceutical industry in which increasing
sophistication of pharmaceutical materials has necessitated the use for more sensitive,
thermodynamic based techniques for materials characterization. The applications can be
found in polymorph characterization (Tong et al., 2002), effect of processing steps (Heng
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
49
et al., 2006), and drug-carrier interactions for dry powder formulations (Feeley et al.,
1998), and differentiation of crystalline and amorphous phases (Ahfat et al., 2000).
Unlike traditional analytical Gas Chromatography (GC) in which a standard column
is used to separate and characterize several gases and/or vapors, the stationary phase is
the sample under investigation while a single gas or vapor is injected as a probe molecule
in IGC. Probe molecules can be injected using pulse or frontal techniques. In a pulse
experiment a certain amount of the probe molecule is injected and the pulse is transported
by the mobile phase (carrier gas) through the system to the column. Subsequently,
adsorption and desorption occurs and the result is a peak in the chromatogram. In a
frontal experiment, the probe molecule is added continuously to the carrier gas and the
chromatogram shows a breakthrough curve. In general, pulse technique is more
commonly used since it is faster, easier to control and more accurate, especially if
interactions between probe and solid are rather weak (Thielmann, 2004).
The primary information in IGC measurement is retention time, more specifically,
gross retention time (tR) and dead retention time (t0). The former is defined as the time
period from occurrence of adsorption to that of desorption of the probe molecule; while
the latter is the time the probe molecule would require to travel through the system
without any interaction. The retention time is measured by typical chromatographic
detection methods such as flame ionisation (FID) or thermal conductivity (TCD) detector.
After the determination of tR and t0, the net retention volume VN can then be calculated as:
VNj
mω · tR t ·
TT
3 3
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
50
where T and T are the column temperature and reference temperature for the flow rate
determination respectively; m is sample mass; j is the James–Martin correction, ω is the
exit flow rate at 1 atm and the reference temperature. IGC uses the net retention volume
of the adsorbates to calculate the dispersive component of surface energy of the sample.
Because the probes are injected at infinite dilution, they will only interact with the most
energetic sites of the solid. The solid surface energy ( ds ) can be obtained from:
RTlnVN a γ / · 2N γ / C 3 4
where a is the molecular area of adsorbate; N is the Avogadro’s Number; ds and d
l are
dispersive components of surface free energy of the solid and liquid probe respectively. C
is a constant (Djordjevic et al., 1992) .
In this work, dispersive surface energy of SS was analyzed using a SMS-iGC 2000
inverse gas chromatography instrument (Surface Measurement Systems, London, UK).
SS sample was packed into a pre-silanized glass column (300 × 4 mm ID) with silanized
glass wool packing at each end to prevent powder movement. The SS samples were pre-
treated with helium for 16 hours at 303K to remove surface moisture. Adsorption
isotherms were measured with a series of purely dispersive n-alkane vapor probes
(undecane, decane, nonane and octane) injected at infinite dilution at 303.15K. The
dispersive surface energy of SS was calculated from net retention volumes using the
SMS-iGC Standard Analysis Suite based on a peak maximum analysis.
3.1.6 Viscosity study
Viscosity of the solution can play an important role in crystallization process. High
viscosity may influence diffusivity of the solute molecules in the solution, and thus affect
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
51
the nucleation rate and growth rate. In addition, it may affect the processability because a
high viscosity liquid requires more power to pump than a low viscosity one (Baldyga and
Bourne, 1995). Murnane et al. (2008) showed that in a stirred-tank, shear rate may
influence the viscosity of the fluids and hence affect mixing characteristics and mass
transfer phenomena. Therefore in this work, a continuous shear rheology is used for
determination of dynamic viscosity of the crystallization medium.
Viscosity of PVP solutions were studied using an Anton Paar Physica Modular
Compact Rheometer (MCR-501) using the cylindrical double-gap configuration (DG-
26.7). A series of PVP solutions in ethanol-water mixture (90%, v/v) at concentration of
1.0 mg/mL was prepared in the molecular weight range (10K–360K). The measurement
was performed at 25°C controlled by a circulator (LAUDA RE 104). The measuring
mode is shear rate controlled, and steps in the shear rate are run for the duration of 15 s.
The shear rate ranges were from 0 to 5000 1/s.
3.2 Process Analytical Technology
Sample handling in industrial crystallization can be undesirable due to difficulty in
sampling, safety concerns, process efficiency, and lengthy offline analysis times. Process
Analytical Technology (PAT) which offers virtually real time data of critical quality and
performance attributes of in-process materials and products without sample handling can
successfully solve the problems. Including scientifically based process design,
appropriate measurement devices, statistical and information technology tools, and
feedback process control strategies, the PAT system can be used for design, analysis, and
control of manufacturing processes in crystallization industry (Yu et al., 2004)
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
52
3.2.1 Focused Beam Reflectance Measurement (FBRM)
Real-time monitoring of PSD in dense slurries is often desired in industrial
crystallization for tighter control of the process. Most traditional sizing methods require
withdrawing samples from the slurry or diverting a side stream through their sensing
compartment, raising difficulty in obtaining representative samples from the slurries. In
addition, sample handling procedures introduce uncertainty to the sizing result. Such
problems can be solved by using Focused Beam Reflectance Measurement (FBRM), a
widely used inline PSD measurement technique in the past decade.
The measurement principle, as shown is Fig. 3-5, is based on backscattering of laser
beam by particles. An infrared laser beam is focused by a set of rotating optics fitted in
the probe and then projected into the particle system through a sapphire window sealed at
the end of the probe. The focused beam rotates at the same frequency as the optics,
sweeping the particle system along a circular path at linear speeds of 2-6 ms-1. Different
transparencies or spot light reflections on the particle surface cause a varying light
scattering intensity, as visualized in right of Fig. 3-5. Therefore, the original scattering
signal is filtered. As long as the filtered signal surpasses the discrimination threshold, a
trigger signal is set. The time difference between the two trigger events (t2 - t1) is
recorded and multiplied by the laser’s velocity Vlas, the resulting distance being known as
a chord length and each detecting event of chord lengths called a “count”. Typically,
thousands of chord lengths can be registered in a second and sorted into channels to yield
a chord length distribution (CLD). There are 1324 hardware channels and each channel
represents a different range of chord length and records the counts falling in its range.
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
53
Figure 3-5 The schematics of an Focused Beam Reflectance Measurement (FBRM) probe
(left) and the internal signal processing (right). Based on the working principle described above, different chord lengths are
measured even for the same particle, depending on the position of the laser. Therefore
FBRM data are often used in a relative sense for monitoring product quality, processing
characteristics of particles and process states. In other words, process information is
extracted from the change in their values, instead of their absolute values. Conversion of
a measured CLD into its corresponding PSD always lack accuracy because all
reconstruction works involve many assumptions, e.g. zero particle velocity with respect
to the probe window, ideal backscattering of the particles, dimensionless laser spot, and
insignificant effect of slurry concentration (Tadayyon and Rohani, 1998; Ruf et al., 2000;
Hukkanen and Braatz, 2003; Worlitschek et al., 2005; Kail et al., 2009). Some
researchers attempt to establish empirical correlations of FBRM data with the results of
traditional sizing techniques (Law et al., 1997; Phillips and Walling, 1998; Heath et al.,
2002). The main problems for these correlations are overestimation of small particles
and/or underestimation of larger ones.
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
54
In this work, a D600L FBRM probe (Lasentec) was used to obtain inline
information of particle number and size distribution. The total count of chord lengths was
used to indicate the changes in particle number and the square-weighted CLD to
demonstrate the growth of crystals.
3.2.2 Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR)
Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR) spectrometer,
first introduced by Dunuwila and his colleagues in 1994 (Dunuwila et al., 1994), has been
used widely for the measurement of solute concentration in slurry. The application is
based on the theory of the Beer-Lambert law correlating the molar concentration of the
solute (c) to the absorbance of the solution (A) by A = εcl, where ε is the molar
absorptivity (constant for the solute causing absorbance) and l is the path length of the
sample medium (Kemp, 1987). According to this principle, quantitative determination of
solute concentration can be made through solution absorbance measurements using IR.
The technique of ATR-FTIR depends on spectral data in the mid-IR range where
greater chemical selectivity is gained compared with near-IR. Fig. 3-6 shows its working
principle. A laser beam generated in the FTIR machine is directed by optics to the
prismatic ATR crystal (ZnSe) that is in contact with the slurry. Part of the beam is
reflected at the ATR crystal/liquid interface. Another part propagates into the liquid
phase where absorbance takes place. After being reflected back into the probe it is
directed to the detector.
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
55
Figure 3-6 Working principle of Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR).
The depth of penetration of the infrared energy field (dp) into the solution is
significantly short comparing with the liquid phase barrier between the probe and solid
crystal particles. According to Eq. 3-5:
2 / 3 5
the value of dp is as small as 2 μm, provided that θ=45º, λ=1000 cm-1, and the refractive
index of ZnSe crystal (n1) and water (n2) are 2.4 and 1.5 respectively. Such a depth is
shorter than the boundary layer of laminar flow surrounding the ATR crystal, and few
solid particles in the slurry can enter this layer to interfere with infrared light absorption.
Therefore when the ATR probe is inserted into crystal slurry, the absorbance recorded is
from the liquid phase in immediate contact with ATR crystal only and there is negligible
interference from the solid crystals. The spectra can be correlated with the solution
concentrations of more than one compound since every compound in the solution will
absorb the electromagnetic wave independently.
I0, incident beam
dp, depth of penetration
IR, reflected beam
ATR crystal with refractive index n1
IAR, absorbed reflected beam
solution sample with refractive index n2
Chapter 3Experimental – In Situ and Ex Situ Analytical Techniques
56
ATR-FTIR technique has many advantages over other measurement techniques. For
example, it requires the sample directly contact with the ATR crystal and the result is not
affected by the surrounding solids (Dunuwila and Berglund, 1997). Moreover, it can
provide the information of many species in the solution simultaneously (Togkalidou et al.,
2002). Recently the technique has also been used in feedback control and optimization
(Yu et al., 2006; Lindenberg et al., 2009).
In this work, absorbance spectra were collected on a Nicolet Nexus 4700
spectrophotometer (Nicolet Instrument Co.) which was equipped with a Dipper-210
Axiom analytical attenuated total reflectance (ATR) immersion probe. Every spectrum
was the average of 64 scans in the range of 600-4000 cm-1. The spectra of ultra pure
water at 23 °C were used as background. The FTIR spectrometer was purged
continuously by purge gas that was supplied by a FTIR purge-gas generator (Parker
Balston, model 75-52-12VDC).
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
57
Chapter 4
Screening of Operating Parameters in
Antisolvent Crystallization
In this chapter, screening results of operating parameters in antisolvent
crystallization of SS are presented. The solubility of SS in several candidate antisolvents
including acetone, acetonitrile, methanol, ethanol, 1-propanol and 2-proponal were
studied in order to investigate their capability to generate supersaturation and their
influence on induction time. The antisolvent crystallization experiments were carried out
using water as the solvent at different solvent/antisolvent volumetric ratios. Particle size,
crystal morphology and agglomeration phenomenon were examined in order to determine
the appropriate antisolvent and volume ratio of solvent and antisolvent used in the
following studies.
4.1 Introduction
Cooling crystallization, as the one of the earliest and simplest crystallization
methods, is not necessarily the best option for some systems even if solubility proves to
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
58
be sufficiently temperature sensitive. The solute may decompose or react with the solvent
at the required starting temperature, resulting in a yield loss and impurity introduction.
The impurities formed may adversely affect crystal habits for the purity or color of the
dry product (Kundig et al., 2001).
In such situations, antisolvent crystallization which can be operated at ambient
environment using a solvent/antisolvent pair might be an attractive alternative.
Antisolvent crystallization supersaturates solutions by lowering the solubilization ability
of the media. As compared with cooling and evaporative crystallization, addition of a
second solvent (antisolvent) into the solute solution can generate higher supersaturation,
therefore produce smaller crystals. Solvent and antisolvent should be miscible with each
other. Usually there is one solvent and one antisolvent used in the crystallization process.
However, sometimes multiple solvents or antisolvents are used for the purpose of
producing a sharper solubility curve or a more desirable crystal shape or polymorph
(Nagy et al., 2008). By adding in antisolvent(s), the activity of the mixed solvents can be
significantly changed. Therefore this approach can have more profound effect on the
crystal morphology or polymorphic form than in the case of cooling crystallization
(Roelands et al., 2006).
Antisolvent has a strong influence on the habit and PSD of crystalline product;
however, the role played by solvent-surface interactions in enhancing or inhibiting crystal
growth is still not completely revealed (Davey et al., 1988). To date, an antisolvent is still
selected primarily through empirical screening by considering the effect of the
antisolvent on the outcome of the process (Myerson, 2002).
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
59
Firstly, the antisolvent must be miscible with the solvent in the proportions used.
Secondly, the solubility of SS in the antisolvent should be <100 mg/mL to minimize the
loss of drug at the end of the crystallization process. Thirdly, the antisolvent should not
yield different polymorphic forms during crystallization. Previous studies have reported
the use of antisolvents such as isoproponal (Chiou et al., 2007), ethanol (Larhrib et al.,
2003) and acetone (Nocent et al., 2001) etc. for SS crystallization in which no
polymorphic forms of SS were observed. Based on the above criteria, this work screened
several solvents (acetone, acetonitrile, methanol, ethanol, 1-propanol and 2-propanol) in
an attempt to find out the appropriate antisolvent for subsequent crystallization
experiments using different additives.
The dependence of solubility of SS on temperature was investigated, followed by
measurement of SS solubility in various binary solvent systems. Initial supersaturation
therefore was determined on the basis of the resulting solubility data. Operating
parameters including initial supersaturation, antisolvent type and composition were
studied to examine their effects on crystal habit, crystal size and size distribution and
crystallinity of SS. The determined type of the antisolvent and the solvent/antisolvent
volumetric ratio will be employed in the subsequent stages of this work.
4.2 Experimental methods
4.2.1 Materials
Salbutamol sulphate, abbreviated to SS in this work, has the chemical name α-[(tert-
Butylamino)methyl]-4-hydroxy-m-xylene-α,α′-diol. It has the following structural
formula (Fig. 4-1) and a molecular weight of 288.36 g/mole.
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
60
Figure 4-1 Salbutamol sulphate structural formula.
In this work, two different sources of SS were used. Salbutamol sulphate (99.9%,
CAS 51022-70-9) purchased from Sigma-Aldrich Co. (USA) was used to prepare HPLC
standard solution. Salbutamol sulphate (99.0%, 031007) used in the other experiments
was purchased from Junda Pharm Chem Plant Co., Ltd (Jiangsu, China).
SS is freely soluble in water and sparingly soluble in organic solvents. Therefore in
this work water was used as solvent and organic solvents as antisolvents. Antisolvents
studied in this work include acetone, acetonitrile, methanol, ethanol, 1-propanol and 2-
propanol.
All the solvents used were of analytical grade. The water used was ultrapure water
purified by Milli-Q Gradient A10 system (Millipore, USA).
4.2.2 Solubility measurement
Solubility of SS in various solvent/antisolvent mixtures at room temperature (~23°C)
was determined using HPLC. Antisolvents being examined include ethanol, acetone,
acetonitrile, methanol, 1-proponal and 2-proponal.
SS was added in excess to a series of 100 mL antisolvent-water mixtures varying
from 0 to 100% (v/v) of antisolvent composition. The suspensions were continuously
stirred overnight, and then filtered using 0.2 m Millipore® membrane filter. Saturated
SS solution was diluted to an appropriate level with water and concentration measured
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
61
using an Agilent 1100 HPLC system equipped with a quaternary pump, degasser and a
Variable Wavelength Detector (VWD). 5 µL of samples (in triplicate) were injected into
a reverse-phase Zorbax® SB-C18 column for concentration analysis.
4.2.3 Crystallization Experiments
Crystallization experiments were conducted in a 100ml beaker at room temperature
(~ 23 °C). SS solution was prepared by dissolving 2 g of SS in 10 mL of water. Various
antisolvents with different volume ratios to water (3, 5, 7 and 9) were added at flow rate
of 10ml/min into the drug solution with continuous stirring using a magnetic stirrer at 500
rpm. The onset of nucleation was monitored by naked eyes and the slurry was filtered 10
minutes after that. Crystals were washed with 20 mL of antisolvent, dried overnight in a
vacuum oven (55 C and 100 mbar), and stored in a desiccators for further
characterization.
4.3 Results and Discussion
4.3.1 Solubility measurement
Solvents can be broadly classified into polar (hydrophilic) and non-polar
(lipophilic). The dielectric constant є of the solvents can be used as an index of its
polarity, determining what type of compounds it is able to dissolve and with what other
solvents it is miscible with. As a rule of thumb, polar solvents dissolve polar compounds
best and non-polar solvents dissolve non-polar compounds best: "like dissolves like".
Polar solvents can be further subdivided into polar protic solvents and polar aprotic
solvents, depending if or not the solvent has a hydrogen atom attached to a strongly
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
62
electronegative element (e.g. oxygen) that forms hydrogen bonds. Table 4-1 lists
properties of the solvents (water, acetone, acetonitirl, methanol, ethanol, 1-propanol and
2-propanol) used in this work (Dean, 1992).
Table 4-1 Property table of solvents.
SS is highly soluble in water, thus water is used as the solvent in the current work.
Acetone, acetonitrile, methanol, ethanol, 1-propanol and 2-propanol are solvents widely
used in pharmaceutical industries. They are all polar solvents and fully miscible with
water. These solvents are selected as candidate antisolvents in preparation of micron-size
SS particles.
As a salicyl alcohol derivative, SS was reported to decompose in aqueous solutions
at elevated temperatures. Mälkki and Tammilehto (1990) studied the influence of
temperature on the decomposition rate of SS in the range 55-85°C at pH 8.8. The results
showed that SS started to decompose at 55°C and the decomposition obeyed apparent
first-order kinetics with respect to SS. Therefore in order to fully ensure the integrity of
Solvent Chemical formula Dielectric constant є
Solvent type
water H=O=H 80.1 polar protic
acetone CH3-C(=O)-CH3 20.7 polar aprotic
acetonitrile CH3-C≡N 37 polar aprotic
methanol CH3-OH 33.6 polar protic
ethanol CH3-CH2-OH 25 polar protic
1-proponal CH3-CH2-CH2-OH 20 polar protic
2-proponal CH3-CH(-OH)-CH3 18 polar protic
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
63
SS compound in our experiments, this work measured solubility of SS in water between
the maximum temperature 37°C (body temperature) and the minimum temperature 10°C.
It is found that SS is highly soluble in water and the solubility reduces linearly with the
decrease in temperature, changing from 316.8 mg/mL at 37°C to 281 mg/mL at 10 °C.
Considering the insignificant reduction in solubility values within the temperature range
and the instability of SS at higher temperature, it can be concluded that antisolvent
crystallization is a more suitable method to precipitate SS over cooling crystallization.
The experimental results over the solubility of SS in water-antisolvent mixtures at
room temperature are shown in Fig. 4-2. The solubility, Ce, is given as mg of SS/mL of
solution and represents the average of n samples from the same solution. It is observed
that the solubility of SS dramatically decrease with the increase of the content of
antisolvents. Moreover, SS has extremely low solubility in pure antisolvents, (e.g. 0.07
mg/mL in ethanol) except in methanol (15.5 mg/mL).
Figure 4-2 Measured solubility of salbutamol sulphate in binary solvents.
0
50
100
150
200
250
300
0 10 20 30 40 50 60 70 80 90 100 110
Solubility (mg/ml)
Volume percentage of antisolvent (%)
Acetone
Acetonitril
Methanol
Ethanol
1‐propanol
2‐propanol
0
50
100
80 85 90 95 100
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
64
Related to the polarity of the solvent, the dielectric constant can be regarded as a
measure of the ability of the solvent to separate ionic charges. In general, a high dielectric
constant is required to dissolve ionic compounds, and ionic compounds are more soluble
in polar solvents having higher dielectric constants (Lowery and Richardson, 1987). This
gives explanation on why SS are highly soluble in water which has the highest є value of
80.1, and sparingly soluble in antisolvents with low є values (Table 4-1). However, it is
also observed in Fig. 4-3 that methanol displays the highest solubilization ability for SS
among the candidate antisolvents at higher antisolvent percentage (80~100); while
acetonitrile which has higher dielectric constant than methanol dissolves less amount of
SS. This is because the dielectric constant is not the only characteristic of an effective ion
solvent; hydrogen bonding also influences the solubility of one substance in another.
Protic solvents have higher tendency to form hydrogen bond with the anions of the solute
so that the solubility of the solute can be enhanced as compared with those aportic
solvents (Klymchenko et al., 2007). Therefore the solubility of SS in candidate
antisolvents is dependent on both the polarity and hydrogen-bonding capability between
solute and solvent molecules.
4.3.2 Initial supersaturation ratio
When antisolvent is added into the SS aqueous solution, sudden reduction of the
solubility of SS in the liquid medium leads to generation of high supersaturation, which
as mentioned in the previous section, is the driving force for precipitation of the
crystalline product. Depending on the solution/antisolvent ratios, series of initial
supersaturation can be created. Table 4-2 summarizes the calculated initial
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
65
supersaturation ratios of SS and the observation results on nucleation occurrence. The
solubility data of SS were taken from Fig. 4-2.
Table 4-2 Initial supersaturation ratio and observation on nucleation at various
volume ratios of antisolvent/water.
Antisolvent Volume ratio, x
(antisolvent: water) Solubility (mg/mL)
Initial supersaturation
ratio, Si
Crystallization in one hour
acetone
3 18.6 2.57 yes
5 4.10 7.75 yes
7 2.30 10.4 yes
9 1.98 9.63 yes
acetonitrile
3 36.9 1.29 no
5 13.6 2.34 yes
7 6.30 3.78 yes
9 2.50 7.64 yes
methanol
3 71.7 0.66 no
5 41.6 0.76 no
7 31.0 0.77 no
9 25.0 0.76 no
ethanol
3 35.0 1.36 no
5 13.4 2.36 yes
7 6.90 3.45 yes
9 2.91 6.56 yes
1-proponal
3 54.9 0.87 no
5 25.4 1.25 no
7 15.0 1.59 yes
9 8.50 2.25 yes
2-proponal
3 27.3 1.74 yes
5 9.04 3.51 yes
7 5.31 4.49 yes
9 3.50 5.46 Yes
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
66
The initial supersaturation ratios generally increased with the addition of
antisolvents for acetone, acetonitrile, ethanol, 1-propanol and 2-propanol. When acetone
was used as the antisolvent, crystallization occurred rapidly at any antisolvent volume
ratio. The maximum initial supersaturation ratio Si was found to be up to 10 at
antisolvent volume value of 7. Such a high supersaturation favours occurrence of
homogeneous nucleation and thus the generation of plenty of nuclei, resulting in
formation of micron-crystals.
The experiments with methanol as the antisolvent show remarkable differences as
compared with other antisolvents. It generated much lower initial supersaturation ratios
(< 1) and the value of Si remained unchanged when x was increased. The supersaturation
build up in this experiment are not sufficient to induce the formation of nuclei, therefore,
for any methanol addition ratio, there was no crystal observed even after 24 hours.
Observation on occurrence of nucleation shows that induction time strongly
depended on the initial superstation ratio. When initial supersaturation ratio is higher than
1.5, nucleation was ensured; when solution is undersaturated, low initial supersaturation
ratio (< 1.0) did not lead to any occurrence of crystallization. When initial supersaturation
ratio is between 1.0 and 1.5, the induction time exceeded 1 hour.
Therefore based on the above investigations, it can be concluded that addition of
high ratio of acetone, acetonitrile, ethanol, 1-propanol and 2-propanol into SS aqueous
solution is the prerequisite of generating high supersaturation.
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
67
4.3.3 Crystal habit
SEM images of SS crystals obtained in acetone-water system are shown in Fig. 4-3.
It is observed that SS crystals obtained using acetone as the antisolvent are very thin and
flaky with unsmooth surfaces and irregular edges. With the increasing acetone volume
ratio, the roughness of the surface and irregularity of the edges become more significant
and agglomeration become more serious. Rough surfaces are generally not desired
Figure 4-3 SEM images of salbutamol sulphate crystals in acetone-water system with
volume ratio of acetone to water of (a) 3; (b) 5; (c) 7 and (d) 9.
(a) (b)
(c) (d)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
68
because increasing surface smoothness of drug particles by means of crystal engineering
has been reported to increase the percentage of fine particle fraction (FPF) (Zeng et al.,
2000). It was also found that increase in the surface smoothness of crystals improves drug
delivery efficiency for inhalation drug (Larhrib et al., 2003). Therefore, SS precipitated
from acetone-water system are obviously not the desirable product for inhalation
administration.
As compared with those obtained from acetone, SS crystals precipitated from
acetonitrile display similar plate-like shape but greater thickness, and crystal surfaces are
much smoother and more regular as shown in Fig. 4-4. This type of the primary plate-
Figure 4-4 SEM images of salbutamol sulphate crystals in acetonitrile-water system with volume ratio of acetonitrile to water of (a) 5; (b) 7 and (c) 9.
(a) (b)
(c)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
69
shape micro-crystals tends to aggregate tightly and further form secondary particles,
leading to larger particle size. The agglomeration degree tended to increase when volume
ratio of antisovlents reached up to 7 and 9 for both acetone and acetonitrile systems.
SS crystals precipitated from alcoholic solvents, ethanol, 1-propanol and 2-propanol,
are shown in Fig. 4-5, 4-6 and 4-7respectively. These protic solvents as the antisolvents
produced needle-like crystals with the long dimension of tens of microns. SS crystals
obtained from the ethanol-water medium had uniform and regular crystal shape and
Figure 4-5 SEM images of salbutamol sulphate crystals in ethanol-water system with
volume ratio of ethanol to water of (a) 5; (b) 7 and (c) 9.
(a)
(c)
(b)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
70
(a) (b)
Figure 4-6 SEM images of salbutamol sulphate crystals in 1-propanol-water system with
volume ratio of 1-propanol to water of (a) 7 and (b) 9.
Figure 4-7 SEM images of salbutamol sulphate crystals in 2-propanol-water system with volume ratio of 2-propanol to water of (a) 3; (b) 5; (c) 7 and (d) 9.
(c) (d)
(a) (b)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
71
length, especially those at higher ethanol composition, whereas those obtained from 1-
propanol and 2-propanol were found to have higher proportion of fine particles. For
example, at higher composition of 2-propanol (Fig. 4-7c,d), considerable amount of small
particles with broaden size distribution were observed.
It has been well known that solvent has a strong influence on habit of crystals;
however the role played by solvent on crystal growth and morphology is still not
completely revealed. One theory suggested that favourable interactions between solute
and solvent on the specific faces can reduce the interfacial tension, thus causing a
transition from a smooth to a rough interface and a concomitant faster surface growth
(Elwenspoek et al., 1987; Bennema, 1992). Another approach assumes that solvent
molecules act as “tailor-made” additives, hindering regular deposition of oncoming solute
molecular layers on the specific face and hence inhibit crystal growth on that face
(Weissbuch et al., 1991). However, solvent-surface geometry cannot be as well defined
as that of “tailor-made” additives; therefore it is difficult to analyze solvent-surface
interactions using this approach. Moreover, it is required that “tailor-made” additives
have similar structure as host molecules such that they can replace host molecules. The
interactions between antisolvent molecules and SS crystal surfaces in the current work
apparently do not belong to this category.
The different roles played by protic (ethanol, 1-propanol and 2-propanol)
antisolvents from aprotic (acetone and acetonitrile) solvents on SS crystal habit might be
due to the presence of hydroxyl group in solvent molecules. It is suggested that the
hydroxyl group can be favorablely adsorbed on the specific face of SS by forming
hydrogen bond with polar groups of SS molecules on that face, and such hydrogen bonds
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
72
dominate the solvent-surface interaction. As a result SS crystals displayed different habit
from those in aprotic antisolvent systems.
4.3.4 Agglomeration
Agglomeration is essentially an outcome of collision of two or more crystals in the
suspension. The agglomeration rate depends on the crystal morphology and the attractive
forces between the crystals for all possible contacts (Brunsteiner et al., 2005). Plate-shape
crystals observed in acetone-water and acetonitrile-water systems have two dominating
crystal surfaces which occupy most of the crystal surface area and determine crystal
surface properties. In suspensions the possibility of these dominating crystal surfaces
approaching and colliding with each other is much higher compared to crystals with other
morphology, which could partially explain the serious agglomeration phenomena found
in the acetone-water and acetonitrile-water medium in the current work.
Apart from acetone and acetonitrile systems, agglomeration was also observed at
higher antisolvent ratio for needle-like crystals, for example, in Fig. 4-7d. This might be
due to the high supersaturation generated by the addition of high ratio of antisolvents.
Rapid addition of antisolvents causes formation of small particles with high surface
energy. Collision of the particles will lower the free energy by contracting the total
surface of them. Therefore, these rapidly formed crystals tend to aggregate intensively,
leading to large particle sizes and broad particle size distributions (Karpinski and Wey,
2002 ). Furthermore, agglomeration is increased when attractive inter-particle forces,
such as Van der Waals forces, are strong. In the presence of high supersaturation,
aggregates will be cemented together more quickly.
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
73
4.3.5 Particle size
Particle size of SS crystals obtained from various types of antisovlents at different
volume ratios 10 mins after the onset of nucleation are measured as shown in Table 4-3.
It can be seen that d(0.1) at all experimental conditions ranges from 0.9 μm to 13.9 μm,
d(0.5) from 6.0 μm to 48 μm, and d(0.9) from 30.1 μm to 115.3 μm. SS crystals obtained
Table 4-3 Particle size distribution of salbutamol sulphate crystals obtained from various antisolvents.
Antisolvent Volume ratio (antisolvent to
water)
Particle size distribution (μm)
D(0.1) D(0.5) D(0.9)
acetone
3 6.1 31.5 105.0
5 5.2 18.6 87.5
7 7.1 26.0 82.3
9 6.7 35.5 106.9
acetonitrile
5 13.9 42.8 80.0
7 12.2 48.0 96.2
9 11.8 39.4 94.5
ethanol
5 4.0 20.0 61.0
7 2.8 30.8 88.3
9 2.2 27.9 82.8
1-proponal 7 2.5 11.5 50.9
9 3.4 21.2 115.3
2-proponal
3 1.8 7.3 73.0
5 0.9 6.0 40.1
7 1.6 7.9 30.1
9 2.6 31.6 57.6
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
74
from ethanol and 2-propanol displayed relatively smaller size as compared with those
from other antisolvents. For each antisolvent, it appeared that there was no correlation
between particle size and volume ratio of antisolvent.
Fig. 4-8 shows measured volume mean diameters of SS crystals against initial
supersaturation ratios. The volume mean diameter ( ) was calculated as:
∑∑
4 1
where n is the number of particles of equivalent diameter d. In general, the particle size
should decrease with an increase in supersaturation ratio according to Gibbs-Thomson
equation (Mullin, 2001):
2 4 2
where S is the supersaturation ratio, γ is the surface energy, V is the molecular volume, k
is the Boltzmann constant, T is the absolute temperature and r is the radius of the particle.
Figure 4-8 Volume mean size of salbutamol sulphate crystals against initial supersaturation ratio at time interval of 1 hour.
0
10
20
30
40
50
60
0 2 4 6 8 10 12
dv(μm)
Initial supersaturation ratio (‐)
Acetone
Acetonitrile
Ethanol
1‐propanol
2‐propanol
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
75
However the opposite trend has been observed for all antisolvents used. Such
deviation from the Gibbs-Thomson equation can be explained by agglomeration. At
lower supersaturation, the phenomenon of agglomeration is insignificant. The particle
size decreases with increasing supersaturation obeying the Gibbs-Thomson equation;
whereas higher supersaturation ratio results in faster cementation and stronger bonding
among primary particles, enabling agglomeration of particles and thus higher particle size
measurement (Yu et al., 2005). From SEM images shown in the previous section (Fig. 4-
3~Fig. 4-7), agglomeration was observed for both plate-shape crystals obtained from
aprotic solvents and needle-shape crystals obtained from protic solvents at higher initial
supersaturation ratios. Therefore the complexity in particle size vs. supersaturation shown
in Fig 4-8 may attribute to the competing effect of primary nucleation and agglomeration.
4.3.6 Powder X-ray Diffraction analysis
PXRD was performed to examine the crystallinity of SS crystals obtained from
different antisolvent candidates (Fig. 4-9). The PXRD patterns of SS obtained from
different antisolvents had the same characteristic peaks with the raw material provided by
the manufacture (Fig. 4-9a), indicating that the solid form of SS from all the sources is
the same. It is concluded that solvents did not change the crystal structure in the
antisolvent crystallization of SS, although they modified the morphology of SS crystals
markedly.
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
76
Figure 4-9 Powder X-ray diffraction patterns of (a) salbutamol sulphate raw material and salbutamol sulphate crystals obtained from (b) acetone, (c) acetonitrile, (d) ethanol, (e) 1-
propanol and (f) 2-propanol.
0
50
100
150
200
250
300
5 10 15 20 25 30 35 40
Intensity
2θ
(a)(a)
0
200
400
600
800
1000
1200
1400
5 10 15 20 25 30 35 40
Intensity
2θ
(b)
0
50
100
150
200
250
300
350
5 10 15 20 25 30 35 40
Intensity
2θ
(d)
0
50
100
150
200
250
300
350
400
5 10 15 20 25 30 35 40
Intensity
2θ
(c)
050100150200250300350400450
5 10 15 20 25 30 35 40
Intensity
2θ
(e)
0
50
100
150
200
250
300
350
5 10 15 20 25 30 35 40
Intensity
2θ
(f)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
77
4.3.7 Effect of mixed antisolvent
It has been reported that mixed antisolvents may have considerable effect on the
final product attributes. From previous experiments, it is observed that the induction time
for crystallization of SS using ethanol as antisolvent is very long, despite the fine, needle-
like crystals produced. On the other hand, although the induction time for crystallization
of SS using acetone is very short, the crystals formed are flake-like crystals with
aggregates, which is not the desired morphology. Hence a test using multiple-antisolvent
composed of ethanol and acetone (solvent/antisolvent volumetric ratio is 1:9) was carried
out to determine its effects on the crystallization of SS.
Comparisons of particle size and induction times for different volumetric ratios of
mixed antisovlents are shown in Table 4-4. The induction time was markedly reduced
from 360 s at 80% of ethanol to 30 s at 20% of ethanol, indicating that the nucleation is
sensitive to the composition of the antisolvent. However, particle size did not display
large difference between the two compositions of mixed antisolvent.
Table 4-4 Comparison of particle size and induction time for different volumetric ratios of ethanol-acetone mixtures.
Volume percentage of ethanol
Particle size (µm) Induction time (s)
d(0.1) d(0.5) d(0.9) dv
80 2.03 13.2 71.9 25.3 360
20 8.51 15.7 71.5 29.0 30
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
78
SEM images (Fig. 4-10) show that SS product obtained from mixed antisolvent
with 20% of ethanol consists of mainly flake-like crystals and small part of needle-like
crystals (Fig. 4-10a). The agglomeration phenomenon is significant which is similar to
the case of pure acetone as the antisolvent. In contrast, the SS crystals obtained from
multiple-antisolvent with 80% of ethanol (Fig. 4-10b) are all needle-like which is close to
the case of pure ethanol used as the antisolvent. However, due to the presence of acetone,
more aggregates are observed in the current experiment as compared with that from pure
ethanol-water system.
Figure 4-10 SEM images of salbutamol sulphate crystals formed from multiple-antisolvent
containing (a) 20% ethanol and 80% acetone; (b) 80% ethanol and 20% acetone.
It can be seen from the above study that the usage of mixed antisolvents (acetone
and ethanol) did not appear to be able to improve resultant SS crystal products in terms of
particle size and crystal morphology. It is determined therefore that ethanol will be used
as the single antisolvent in the subsequent experiments.
(a) (b)
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
79
4.4 Conclusions
The effect of antisolvents (acetone, acetonitrile, methanol, ethanol, 1-propanol and
2-propanol) on crystal morphology, size and crystallinity of SS has been studied. SS
precipitated at different induction time in the various antisolvent systems, except that no
crystals could be achieved in methanol during the experiment. Two main crystal habits
were obtained: plate-shape in two aprotic solvents acetone and acetonitrile and needle-
shape in protic solvents including ethanol, 1-propanol and 2-propanol. The different
crystal habits could be due to the different nature of the antisolvents and the interaction
between the antisolvent molecules and the specific facet of SS. A further study at
molecular view is desired in order to understand the mechanism by which antisolvents
affect crystal shape. However, this exceeds the scope of the current study. Resultant
particle size measurement was found to attribute to both crystal size and agglomeration
degree. Antisolvents were found to have no effect on crystal structure even though the
morphology was markedly affected.
Agglomeration phenomenon was observed in the experiments harvesting plate-
shape crystals when acetone and acetonitrile were used as the antisolvents. Two
dominating faces in such a habit enable crystals to collide and adhere easily as compared
with the needle-shape crystals. Agglomeration was also observed in the experiments
using ethanol, 1-propanol and 2-propanol at high initial supersaturation. This might be
due to the sudden increase of the surface energy resulted from rapid addition of the
antisolvent. Both SEM observations and size measurement indicated the occurrence of
agglomeration phenomena in the above antisolvent systems.
Chapter 4 Screening of Operating Parameters in Antisolvent Crystallization
80
A study of multiple antisolvents (acetone and ethanol) was carried out by changing
volume ratio of two antisolvents. The resultant particle size and crystal morphology show
that there is no noticeable effect on the final product of SS by applying multiple
antisovlents.
Based on the experimental screening results for various antisolvents, ethanol
appeared to offer the best antisolvent because it produces regular shape of crystals, small
crystal size and insignificant agglomeration. Therefore ethanol will be used as the
antisolvent in the subsequent antisolvent crystallization experiments.
Chapter5 Application of Surfactants and Polymeric Additives
82
Chapter 5
Application of Surfactants and Polymeric
Additives
In chapter 4, it was found that precipitation of small SS crystals can be achieved by
using ethanol as the antisolvent with the volume ratio of ethanol to water of 9. However,
the particle size and PSD of SS crystals obtained via such a manner still exceeds the
requirement for efficient inhalation delivery. In this chapter, the inhibition effect of
surfactants and polymeric additives on crystal growth in the antisolvent crystallization of
SS was reported. The inhibition mechanism for the selected polymers and surfactants
were further studied based on molecular interaction of SS and additives using molecular
modeling.
5.1 Introduction
Control of particle size can be realized via using “tailor-made” additives. Addition
of additives can have a significant impact on the processes of nucleation of crystallizing
phases and subsequently the growth of the nucleated solids. Three different types of
Chapter5 Application of Surfactants and Polymeric Additives
83
additives have been widely reported for this purpose: low molecular-weight inorganic
compounds, low molecular-weight organic compounds, or substances similar in structure
to crystallizing solute (Kumar et al., 2008). Polymeric materials and surfactants, which
have no similar structure to the crystal surface, can also influence crystallization process.
However, the investigation on this type of additives is relatively less explored.
Polymers or surfactants can be used as additives to control crystal growth and
suppress agglomeration (Femioyewo and Spring, 1994; Taylor and Zografi, 1997;
Murnane et al., 2008). It is thought that polymers and surfactants inhibit crystal growth
and agglomeration by adsorbing on the crystal surfaces. An early study on sulfathiazole
crystal growth in the presence of PVP suggested that the polymer forms a net-like
structure on the surface, and subsequent growth occurs between the netted pores resulting
in formation of finger-like protrusions and a rough surface (Simonelli et al., 1970). The
higher curvatures of these protrusions eventually increase the minimum supersaturation
required for crystal growth. Furthermore, diffusional processes relative rates of
diffusion of the polymer as compared to the solute from the bulk to the crystal surface
controlled the effective concentration of PVP required for growth inhibition. More
recently, the influence of polymers on the antisolvent crystallization of hydrocortisone
acetate was studied (Raghavan et al., 2001). The mechanism of nucleation retardation by
the polymers was explained in terms of association of the solute molecules with the
polymer in solution through hydrogen bonding. Growth inhibition was explained based
on the presence of a hydrodynamic boundary layer surrounding the crystal and
anisotropic adsorption of the polymer on the growing crystal faces. This in turn, is
Chapter5 Application of Surfactants and Polymeric Additives
84
dependent on the hydrogen bonding functional groups that are exposed at each facet of
the crystal.
Investigation of the influence of PVP on crystallization of paracetamol (Garekani et
al., 2000) found that PVP of higher molecular weights (K10 and K50) were more
effective additives than lower molecular weight PVP (K2). With increasing molecular
weight and/or concentration of PVP in the solution, the induction time for crystallization
of paracetamol increased, the amount of PVP adsorbed onto the crystallized particles
increased, the percentage recovery of paracetamol decreased, and, at the same time, very
thin and flaky crystals were produced. Furthermore, differential scanning calorimetric
analysis showed that the onsets of melting point and enthalpies of fusion of paracetamol
crystallized in the presence of PVP decreased as compared to the untreated samples.
These reductions were attributed to interaction between paracetamol and PVP in the
crystals, possibly via hydrogen bonding.
In this work, polymers and surfactants (Lecithin, Span 85, PVP and HMPC) were
used as additives in antisolvent crystallization of SS. Lecithin, Span 85, PVP and HMPC
are the materials which have been frequently used in pharmaceutical industry as
surfactants and are included in the GRAS (Generally Regarded As Safe) list. The
additives used in the current work were selected to represent different types of surfactants.
For example, lecithin is an ionic surfactant and Span 85 is a neutral surfactant. PVP and
HMPC are the polymers with different types of functional groups. These additives were
chosen to gain a molecular level understanding of the mechanism between additives and
the crystallizing phase.
Chapter5 Application of Surfactants and Polymeric Additives
85
The effect of these additives on the crystal size and habit of SS will be studied
experimentally by comparing results obtained in the absence and presence of the
additives, and theoretically by exploring the inhibition mechanism for the selected
polymers and surfactants based on molecular interaction of SS using the method of
molecular modeling.
5.2 Experimental
5.2.1 Materials
SS (99.0+% purity) was purchased from Junda Pharm. Chem. Co Ltd. (China). The
molecular structure of SS is shown in Fig. 4-1. Hydroxypropyl methyl cellulose (HPMC,
K10), polyvinylpyrrolidone (PVP K10, K25, K40 and K360), lecithin (asolectin from
soybean) and Span 85 were supplied by Sigma-Aldrich. 1-Mythl-N-pyrrolidone was
obtained from Merck. Molecular structures of the above additives are shown in Fig. 5-1.
5.2.2 Crystallization Experiments
Antisolvent crystallization experiments were conducted in a 100 mL beaker
(diameter = 5.5 cm) at room temperature (~ 23 °C). The SS solution was prepared by
dissolving 2 g of SS in 10 mL of water. Appropriate amounts of surfactant or polymer
were dissolved in either the drug solution or in ethanol (antisolvent) depending on its
solubility in either of the solvents. 90 mL of ethanol was added rapidly into the drug
solution with continuous stirring using a magnetic stirrer (length = 2 cm) at 500 rpm. SS
crystals were obtained after an induction time varying from 0 - 15 min depending on the
type of additive used. The slurry was filtered using filter paper. The crystals were washed
Chapter5 Application of Surfactants and Polymeric Additives
86
Figure 5-1 Molecular structures.
N O
HC CH2
n
PVP 1-Methyl-N-pyrrolidone
HPMC
Span 85
Lecithin
Chapter5 Application of Surfactants and Polymeric Additives
87
with 20 mL of ethanol, dried overnight in a vacuum oven (55 ºC and 100 mbar), and
stored in a desiccator for further characterization.
SS single crystals with distinguishable faces were obtained via slow evaporation of
aqueous SS saturated solution at room temperature. PVP was dissolved into water at
different concentrations (0 – 5.0 mg/mL) prior to the addition of SS. The resulting
solutions were allowed to evaporate from glass crystallization dishes sealed with a
parafilm containing small holes.
5.3 Screening of polymers and surfactants
Lecithin, Span 85, PVP K25 and HPMC were screened as additives during
antisolvent crystallization of SS in an attempt to inhibit crystal growth and alleviate
agglomeration. Lecithin and HPMC (2.0 mg/mL each), PVP K25 (2.5 mg/mL) were
dissolved in the SS aqueous solutions, while Span 85 (2.5 mg/mL) was dissolved in
ethanol because of its poor solubility in water. The concentration of additive was
calculated based on the amount of additive dissolved in the total volume of the solvent
(water + ethanol) in the final mixed solution.
Fig. 5-2a shows the morphologies of SS crystals obtained in the absence of the
additives. SS crystals obtained are needle-like and formed aggregates. SS crystals
obtained in the presence of Lecithin, Span 85 and HPMC also grow as elongated needles
(Fig. 5-2c,d). In contrast, SS crystals obtained in the presence of PVP K25 (Fig.5-2e) are
shorter in length, with aspect ratio of about 3 to 5. This is significantly smaller than the
aspect ratio obtained in the presence of other additives (ca. 8-10). Powder X-ray
diffraction patterns as shown in Fig. 5-3 confirmed that the SS fine particles obtained in
Chapter5 Application of Surfactants and Polymeric Additives
88
the presence of polymers and surfactants are crystalline and are of the same crystal
structure as that of SS from pure solution.
Figure 5-2 Salbutamol sulphate crystals obtained at 30 minutes after nucleation in the
presence of (a) no additives, (b)lecithin, (c) Span 85, (d) Hydroxypropyl Methylcellulose (HPMC) and (e) polyvinylpyrrolidone (PVP) K25.
(a (b
(c
(e)
(d
Chapter5 Application of Surfactants and Polymeric Additives
89
Figure 5-3 Comparison of Powder X-ray Diffraction patterns for salbutamol sulphate obtained in the absence of additives and those in the presence of lecithin, Span 85,
polyvinylpyrrolidone (PVP) K25 and Hydroxypropyl Methylcellulose (HPMC).
Table 5-1 shows the particle size distributions of SS crystals obtained in the
presence of polymers and surfactants at 30 minutes after nucleation has occurred. The
difference in the performance of different additives is obvious. The d(0.5) value in the
presence of Span 85 is approximately 30 m which is even higher than that obtained in
the absence of additives, demonstrating that Span 85 cannot stabilize the newly formed
crystals and prevent crystal growth. In the presence of HPMC, the d(0.5) value is about
4.3 m, which is much lower than that obtained in the absence of additive, but the d(0.9)
value is 30.3 m which is still very high. This shows that although addition of HPMC
controls size of SS crystals to some extent, the particle size still does not meet the needs
for inhalation drugs. In the presence of lecithin, d(0.1), d(0.5) and d(0.9) values are 0.96,
2.84 and 10.29 m respectively, and are much lower than those obtained in the of
additives. The particle size of SS crystals obtained in the presence of PVP K25 is the
5 10 15 20 25 30 35 40
Intensity
2θ (°)
no additives
lecithin
Span 85
PVP K25
HPMC
Chapter5 Application of Surfactants and Polymeric Additives
90
smallest with the value of d(0.1), d(0.5) and d(0.9) at 0.31, 0.63, and 5.99m respectively.
From the observations, it could be concluded that PVP K25 is the most effective additive
among the four additives tested for inhibition of crystal growth and agglomeration, and
PVP was selected for further detailed investigation.
Table 5-1. Particle size distributions of salbutamol sulphate obtained in the presence of additives at 30 minutes after nucleation.
Additives Particle size distribution (μm)
d(0.1) d(0.5) d(0.9)
No additives 2.05
(±0.05) 13.71
(±8.64) 80.36
(±29.78)
Lecithin 0.98
(±0.04) 2.91
(±0.15) 10.68
(±0.63)
Span 85 1.94
(±0.32) 29.10
(±6.48) 65.78
(±4.29)
PVP K25 0.32
(±0.26) 0.63
(±0.01) 6.00
(±0.02)
HPMC 0.84
(±0.19) 2.20
(±1.89) 30.26
(±7.39)
5.4 Crystal structure and growth morphology of SS
Crystal structure of SS is studied in order to facilitate a better understanding of the
molecular interaction between the additive molecules and SS crystal surfaces. The growth
morphology of SS is predicted based on the crystal structures using theoretical methods
and is subsequently compared to the experimental crystal habit.
Chapter5 Application of Surfactants and Polymeric Additives
91
5.4.1 Crystal structure
It was initially intended to use the crystal structure of SS (Refcode SALBUT)
obtained from the Cambridge Structural Database in order to predict its theoretical
growth morphology. However, it was found that in the reported crystal structure, the
hydrogen atoms were not located for two of the oxygen atoms in the asymmetric unit, and
one of the molecules contained an additional oxygen atom. Besides, the crystal structure
was solved incorrectly in the Cc space group.
Therefore in this work, single crystals of SS were prepared by slow evaporative
crystallization in pure water solution in order to determine the crystal structure of SS. It
was found that the crystal structure belongs to monoclinic, C2/c space group. The
primary interaction that holds the SS molecules are hydrogen bonds via O—H...O and
N—H...O as shown in Fig. 5-4.
Figure 5-4 3-D view of the monoclinic crystal structure (unit cell) of salbutamol sulphate. Color scale of atoms: C (gray), H (white), O (red), S (yellow) and N (blue). Hydrogen
bonding is represented with cyan dotted lines.
Chapter5 Application of Surfactants and Polymeric Additives
92
The crystallographic data between reported and current work are compared in Table
5-2. The integrity of crystal structure determination for both CSD and the current work
were examined using a checkCIF program developed by the International Union of
Crystallography (IUCr), showing that the crystal structure obtained in the current work is
valid and the one CSD reported is invalid. It is noted that the R factor in this work is
relatively high, which might be due to the twinning crystal obtained from the slow
evaporative crystallization experiment. The accuracy of the predicted crystal structure
was also verified by comparison of simulating PXRD with the experimental pattern (Fig.
5-5).
Table 5-2 Comparison of the unit cell dimensions between Cambridge Structural Database and new determination of the crystal structure of salbutamol sulphate.
CSD (Leger et al., 1978) This work
Crystal system Monoclinic Monoclinic
Space group C c C2/c
Lattice Parameters
a = 28.069 Å a = 27.890 Å
b = 6.183 Å b = 6.1857 Å
c = 16.914 Å c = 16.760 Å
α = 90.00° α = 90.00°
= 81.19° = 98.79°
γ = 90.00° γ = 90.00°
Z 8 4
V(Å3) 2900.8 2857.5
R factor 6.7 9.14
Chapter5 Application of Surfactants and Polymeric Additives
93
Figure 5-5 Comparison of Powder X-ray Diffraction patterns for experimental result and patterns simulated from Cambridge Structural Database and the current work.
5.4.2 Crystal habit prediction
The growth morphology of SS was predicted from its crystal structure via
molecular modeling-based simulation techniques such as the Bravais-Friedel-Donnay-
Harker (BFDH) method (Bennema, 1995) and the attachment energy (AE) model
(Clydesdale et al., 1996) as implemented in Accelrys Materials Studio (Elwenspoek et al.,
1987).
An introduction to the BFDH method and the AE method can be found in Chapter 2.
It is worth noting that in the BFDH theory, neither the intermolecular interactions nor the
thermodynamics of crystal interfaces in the crystal structure are taken into account
(Bennema, 1995). It only takes into consideration the unit cell parameters and interplanar
spacing in the prediction of crystal habit. However, despite its low accuracy, the BFDH
5 10 15 20 25 30 35 40
Intensity
2θ (°)
Experimental
Simulated from CSDSimulated from current work
Chapter5 Application of Surfactants and Polymeric Additives
94
method is still often used as a fast initial approach to predict the crystal shape and
importance faces of a growing crystal.
Within the framework of the Hartman-Perdok theory (Hartman and Perdok, 1955),
the AE model is based on the assumption that the growth rate of a crystal face (hkl) is
proportional to the ‘‘attachment energy’. On the basis of this theory, crystal morphology
is bounded by the slow growing facets that have lower attachment energy values.
Modeling the growth morphology of a crystal involves the selection of an appropriate
force field and a description of charge set that can define the intermolecular interactions
in a molecular crystal precisely (Bisker-Leib and Doherty, 2003). For a potential set to be
considered accurate, it should mimic the lattice energy of the system satisfactorily. In this
work, non-bonded intermolecular interactions within the crystal lattice were calculated
using an atom-atom summation method (Clydesdale et al., 1996). To this end, two
different force fields were used: (1) the Dreiding force field (Mayo et al., 1990) with
atomic point charges assigned for an SS molecule using the charge equilibration (QEq)
method (Rappe and Goddard, 1991); (2) the COMPASS force field with force field-
assigned atomic point charges (Maiti and Gee, 2009). A detailed description on the
selection of force field potential for simulation of crystal morphology is reported
elsewhere (Poornachary et al., 2008).
The Dreiding-QEq potential set when applied to the crystal structure of SS obtained
from CSD produced a net negative charge, possibly due to the impropriate location of the
hydrogen atoms for two of the oxygen atoms in the asymmetric unit as mentioned earlier.
As a result, growth morphology of SS could not be predicted using this potential. To
circumvent this problem, the crystal structure of SS was determined experimentally as
Chapter5 Application of Surfactants and Polymeric Additives
95
shown in Table 5-2. Consequently, the growth morphology of SS crystal was predicted
from our crystal structure using the BFDH model and the AE model using DreidingQEq
potential (Fig. 5-6a,b). The AE model, and, to a lesser degree, the BFDH model both
predicted a higher b/c aspect ratio for SS crystal, indicating that the b axis is intrinsically
the fast-growing direction. The predicted AE model compared well with the experimental
crystal habit obtained from evaporative crystallization, which is bounded by the {200},
{002}, {110} and {111} facets (Fig. 5-7).
In order to verify the above result, SS crystal growth morphology was also
predicted from the reported crystal structure (SALBUT) using the COMPASS force field.
The predicted habit (Fig. 5-6c) was needle-like and elongated along the b-axis, in good
agreement with the experimental morphology obtained from the antisolvent
crystallization in the absence of additives.
(a)
Chapter5 Application of Surfactants and Polymeric Additives
96
(b)
(c)
Figure 5-6 Predicted morphology of salbutamol sulphate by (a) the BFDH method and (b – c) the AE method using various force field potentials. (b) The DreidingQEq potential and
(c) the COMPASS force field applied on the crystal structure respectively.
Figure 5-7 Salbutamol sulphate crystals obtained from evaporation crystallization in the presence of 5.0mg/mL of polyvinylpyrrolidone (PVP) K25.
Chapter5 Application of Surfactants and Polymeric Additives
97
5.5 Mechanism of influence of the additives on SS crystal growth
The mechanism by which PVP inhibits growth of SS crystal was examined using
molecular modeling. Due to the uncertainty in the tertiary structure of the polymer, an
energy minimization step should be carried out in order to find out the optimized
configuration. This is computationally intensive and out of the scope of the current work.
Therefore this work investigated the potential intermolecular interactions between an
additive and SS molecules on the dominant crystal faces. To this end, the additive
molecule (recurring units of the polymer (PVP and HPMC) and a lecithin molecule) was
allowed to approach the SS molecules on the crystal surface. At any particular site on the
crystal surface, several orientations of the additive with respect to the surface were tried
with the aim to find the orientation that would result in the optimum interaction between
the additive and SS crystal surface. However, it should be noted that this approach was
based purely on visual representation and a more rigorous modeling would involve
calculation of binding energies for the adsorption of additives onto the crystal surface.
The {110} and {111} faces primarily expose the hydroxyl functional groups
bonded to the phenyl ring of SS molecule and oriented normal to the facet. A PVP
monomer molecule approaching the {110} face (or {111} face) could interact with a SS
molecule on the surface through its pyrrolidone fragment forming hydrogen bonds of
three different possibilities (Fig. 5-8a), namely, OHsalbut…Opyrrole hydrogen bond (site 1),
OHsalbut…Npyrrole hydrogen bond (site 2) and NHsalbut
…Opyrrole hydrogen bond (site 3). Of
the three hydrogen bonding interactions, the carbonyl group of PVP is considered as the
most favorable site for interactions due to steric constraints on the nitrogen
Chapter5 Application of Surfactants and Polymeric Additives
98
Figure 5-8 Molecular modeling of (a) interaction of polyvinylpyrrolidone monomer unit with the {110} face of salbutamol sulphate crystals, (b) & (c) molecular topology of the {200}
and {002} faces of salbutamol sulphate crystals respectively.
Site 1 Site 3
(a)
Site 2 (110) face
(002) face
(c)
(b)
(200) face
Chapter5 Application of Surfactants and Polymeric Additives
99
(Taylor and Zografi, 1997). Hence intermolecular interactions due to OHsalbut…Opyrrole
hydrogen bond could contribute to strong binding affinity for PVP on the {110} faces.
In contrast, on the {200} and {002} surfaces (Fig. 5-8b,c), the methyl groups of SS
are primarily exposed normal to the facet, and therefore, lack potential sites to form
hydrogen bonds with the pyrrolidone fragment of PVP. At the same time, on the {200}
faces, the HSO4 ions of SS molecule could undergo strong interactions with the solvent
(ethanol and water), and consequently, solvation could limit its interactions with PVP. On
the {002} faces, the sites exposing hydroxyl groups of salbutamol molecule are
submerged in between the ridge-sites exposing the methyl groups, and hence, the surface
topology may only favor weaker interactions with PVP as compared to the {110} faces.
Consequently, PVP will be preferentially adsorbed on the {110} and {111} faces, and
affect further attachment of solute molecules leading to growth inhibition along the b
direction.
Applying a similar methodology, intermolecular interactions between a monomer
unit of HPMC and the SS crystal was analyzed as shown in Fig. 5-9. Considering its
molecular structure, HPMC has high levels of hydroxyl and methoxy functional groups
capable of hydrogen bonding with the hydroxyl and amine functional groups of SS.
However, on the {110} face the likelihood of hydrogen bonding between the OHsalbut and
CH3OHPMC functional groups is less probable due to steric hindrance caused by the
bulkier methyl (and propyl) groups of the polymer additive. Hence, it is postulated that
HPMC would have weaker interactions with any of SS crystal faces, and therefore, may
not affect its growth significantly.
Chapter5 Application of Surfactants and Polymeric Additives
100
Figure 5-9 Molecular modeling of interaction of Hydroxypropyl Methylcellulose (HPMC) monomer unit with the {110} face of salbutamol sulphate crystals.
Lecithin, like PVP, exhibited considerable growth inhibition in SS crystals as
reflected from the PSD. In order to understand this effect, interactions between a lecithin
molecule and SS molecules on the crystal surface were probed. As shown in Fig. 5-10,
the lecithin molecule can interact with the hydroxyl group of SS via the phosphonic acid
(PO3) group (site 1) or carbonyl (CO) group (site 2) potentially forming
OHsalbutOlecithin hydrogen bonds. These types of interactions are specific to the {110}
face, and therefore, lecithin could have a stronger binding affinity to that face. Upon
adsorption of the polar head groups of lecithin onto the SS crystal surface, the saturated
fatty acyl groups protrude from the surface and hence provide barrier to further crystal
growth along the b axis.
{110} face
Chapter5 Application of Surfactants and Polymeric Additives
101
Figure 5-10 Molecular modeling of interaction between lecithin and salbutamol sulphate crystals.
Further experimental data were obtained in order to further substantiate the
mechanism of SS crystal growth inhibition by PVP. This involved studying the effects of
PVP concentration and molecular weight on SS crystallization and is discussed below.
5.5.1 Effects of polyvinylpyrrolidone (PVP) concentration
The effects of PVP concentration on resulting SS crystals were examined through
two methods – antisolvent crystallization and slow evaporative crystallization. As shown
in Fig. 5-11, the SS crystals produced via antisolvent crystallization at different
concentration levels of PVP K25 in the crystallization medium exhibited a needle-shaped
morphology. However, the b/c aspect ratio gradually decreased with an increase in PVP
concentration, producing chunkier crystals. As the PVP concentration was increased from
1.0 to 2.5 mg/mL, the SS crystals exhibited a significant reduction in the b/c aspect ratio
along with a change in crystal habit from needle-like to block-like shape. On further
Chapter5 Application of Surfactants and Polymeric Additives
102
increasing the PVP concentration to 5.0 mg/mL, however, no significant reduction in the
aspect ratio could be observed.
Figure 5-11 Salbutamol sulphate crystals obtained via antisolvent crystallization at 24 hours after nucleation in the presence of (a) 1.0 mg/mL, (b) 2.5 mg/mL and (c) 5.0 mg/mL of
polyvinylpyrrolidone (PVP) K25.
A similar effect of PVP K25 concentration on antisolvent crystallization of SS was
also observed from the resulting particle size distribution (Fig. 5-12). The size of SS
crystals obtained at 30 min and 24 h after nucleation decreased when the concentration of
PVP was increased from 0 to 2.5 mg/mL. However, the particle size of SS crystals did
not decrease any further when the concentration of PVP was increased from 2.5 to 5.0
mg/mL. Besides, SS crystals obtained in the presence of 2.5 mg/mL PVP did not show an
(a)
(c
(b)
Chapter5 Application of Surfactants and Polymeric Additives
103
increase in particle size even after 24 hr, indicating that crystals prepared under this
condition were stable against growth and/or agglomeration (Fig. 5-13).
0
10
20
30
40
50
60
70
80
0 1 2 3 4 5
Particle size (µm)
Concentration of PVP (mg/ml)
d(0.1)
d(0.5)
d(0.9)
0
10
20
30
40
50
60
70
80
90
100
0 1 2 3 4 5
Particle size (µm)
Concentration of PVP (mg/ml)
d(0.1)
d(0.5)
d(0.9)
Figure 5- 12 Particle size distributions of SS crystals obtained via antisolvent crystallization in the presence of polyvinylpyrrolidone (PVP) K25 at (a) 30
min and (b) 24 hours after nucleation. .
(b)
(a)
Chapter5 Application of Surfactants and Polymeric Additives
104
0
1
2
3
4
5
6
7
1 2
Particle size (µm)
d(0.1)
d(0.5)
d(0.9)
30min 24 hr
Figure 5-13 Comparison of particle size distribution obtained in the presence of 2.5 mg/mL of polyvinylpyrrolidone (PVP) K25 at different times after
nucleation.
The fraction of particle size less than 5 microns of SS obtained in the presence of
PVP K25 at various additive concentrations generally reduced after growing for 24 h
comparing with that growing for 30 min (Table 5-3). For both 30 min and 24 h growth
time, the fraction of particle size less than 5 microns increased with the increase of PVP
concentration changing from 0 to 2.5 mg/mL. Nevertheless, it did not increase further
when PVP concentration increased from 2.5 to 5.0 mg/mL.
Table 5-3 The fraction of particle size less than 5 microns of salbutamol sulphate particles obtained in the presence of polyvinylpyrrolidone (PVP) K25.
Concentration of PVP (mg/mL)
Fraction < 5μm (%) 30 min 24 h
0 3.5 3.76
1.0 73.7 40.1
2.5 87.5 87.3
5.0 80.2 72.4
Chapter5 Application of Surfactants and Polymeric Additives
105
Slow evaporative crystallization was performed in order to better understand the
mechanism by which PVP K25 influences the crystallization of SS. The crystals display
the same morphology but larger size as compared with those in antisolvent crystallization
(Fig. 5-14). The difference in crystal size is due to the different driving force in the two
processes, which can be expressed by the degree of supersaturation. For evaporative
crystallization, it is hard to quantify this parameter because of the difficulty in measuring
solute concentration during evaporation (Martins et al., 2005). High degree of
supersaturation generated in the antisolvent crystallization of SS brought about short
Figure 5-14 Salbutamol sulphate crystals obtained from evaporation crystallization in the presence of polyvinylpyrrolidone (PVP) K25 at (a) 0 mg/mL, (b) 1.0 mg/mL, (c) 2.5 mg/mL
and (d) 5.0 mg/mL. The scale bar is 1000m.
(c)
(b) (a)
(d)
Chapter5 Application of Surfactants and Polymeric Additives
106
induction time (less than 10 mins) and small crystals. While in evaporative crystallization,
nucleation occurred after much longer time (24 hours). Such slow induction favors
crystal growth more than nucleation, resulting in large crystal size and good morphology.
SS crystals obtained from the water solution in the absence of PVP K25 display a
needle-like shape (Fig. 5-14a). As PVP K25 concentration increases from 0 to 2.5
mg/mL, the aspect ratio of the crystal decreases with the crystal changing from needle-
shape to chunk shape. However, the aspect ratio does not reduce markedly when the
concentration of PVP K25 increases from 2.5 to 5.0 mg/mL (Fig. 5-14b ~ d). Herein, it is
emphasized that the levels of supersaturation generated in the antisolvent crystallization
is significantly higher than the evaporation method, which influenced the induction time
and number of crystals nucleated. Nonetheless, the effect of PVP concentration on SS
crystal growth and morphology was consistent between these two methods.
A direct inference from these morphological changes occurred in both antisolvent
and slow evaporative crystallization is that growth inhibition occurs primarily along the
b-axis in the presence of PVP. Another additional inference could also be drawn. The
magnitude of growth inhibition initially increases with PVP concentration and reaches a
maximum beyond which growth inhibition is less significant. This implies that an
equilibrium process, controlled by the adsorption of PVP, could be in action.
5.5.2 Effects of molecular weight of polyvinylpyrrolidone (PVP)
In order to validate the interaction mechanism between PVP and SS crystal,
antisolvent crystallization of SS was also performed in the presence of PVP of different
molecular weight. 1-methyl-N-pyrrolidone was chosen as a model for the recurring unit
Chapter5 Application of Surfactants and Polymeric Additives
107
of PVP as it has similar molecular structure as the repeating unit of PVP. SS crystals
obtained from solutions containing 25.0 mg/mL of 1-methyl-N-pyrrolidone (10 times
higher concentration than that of polymer additive) displayed an elongated needle-like
habit and a d(0.5) of 7.25 μm (Fig. 5-15b and Table 5-4). In comparison with SS crystals
obtained in the absence of any additive (d(0.5) = 13.71 µm), this observation indicates
that crystal growth was inhibited in the presence of the monomer. The viscosity of the
ethanol-water solution containing monomer at the same concentration (25 mg/mL) were
1.60 and 2.74 mPa·s at shear rates of 128 and 5000 s-1, respectively. This shows only a
slight increase as compared to the viscosity of ethanolwater mixture (1.50 and 2.66
mPa·s at shear rates of 128 and 5000 s-1 respectively), suggesting that mass transfer of the
solute molecules to the crystal surface is not the rate-limiting step for growth inhibition of
SS crystals. On the other hand, specific interactions between the monomer and SS
molecules at the crystal surface, possibly via hydrogen bonding, could lead to a retarding
effect on crystal growth. However, the extent of growth inhibition by the monomer is not
strong enough to produce significant habit change as observed with PVP K25. This could
be because the repeating units of PVP can provide greater H-bonding propensity with the
SS crystal surface in comparison with the monomer unit. Moreover, as proposed by
Simonelli et al. (1970), a net-like structure at the crystalsolution interface could be
formed with PVP and thereby result in an extra barrier for crystal growth.
The growth inhibition mechanism was further investigated to include the effect of
molecular weight of PVP. Fig. 5-15c~f show the resulting crystals from different
experiments using 1.0 mg/mL of PVP of molecular weights K10, K25, K40 and K360
respectively and their respective PSD are shown in Table 5-4. At 1.0 mg/mL
Chapter5 Application of Surfactants and Polymeric Additives
108
Figure 5-15 Salbutamol sulphate crystals obtained in the presence of (a) no additives; (b) 25 mg/mL of 1-Methyl-N-pyrrolidone; and 1.0 mg/mL of polyvinylpyrrolidone (PVP) (c) K10;
(d) K25; (e) K40 and (f) K360.
concentration, PVP K10 produced smaller crystals with size along the b direction
measured around 5 µm. At the same concentration of PVP K25, SS crystals obtained
were measured to be less than 5 µm along the b axis. Addition of PVP K40 further
reduced the crystal size to approximately 1 µm length, which is not suitable for inhalation
applications. Interestingly, the highest molecular weight (PVP K360) did not follow the
(a)(b)
(f)
(d) (c)
(e)
Chapter5 Application of Surfactants and Polymeric Additives
109
trend, but produced crystals of approximately 10 µm length. These results show a direct
relationship between PVP molecular weight and the extent of crystal growth inhibition.
Thus, for the range of molecular weights K10K40, a higher molecular weight of PVP
can provide higher concentration of the recurring vinyl pyrrolidone segments interacting
with the crystal surface, and therefore, a greater adsorbing tendency and, subsequently
resulting in stronger growth inhibition effect. On the other hand, for PVP K360, the rate
of PVP diffusion to the surface may be the rate determining step for the polymercrystal
interaction.
Table 5-4 Particle size distributions of salbutamol sulphate obtained in the presence of 1-Methyl-N-pyrrolidone and polyvinylpyrrolidone of different molecular weight
Additives Particle size distribution (μm)
d(0.1) d(0.5) d(0.9)
No additives 2.05
(±0.05) 13.71
(±8.64) 80.36
(±29.78)
1-methyl-N-pyrrolidone
1.51 (±0.12)
7.25 (±0.33)
72.48 (±13.83)
PVP K10 1.15
(±0.07) 5.21
(±0.56) 35.04
(±5.30)
PVP K25 1.02
(±0.04) 4.19
(±0.17) 21.32
(±1.70)
PVP K40 0.78
(±0.09) 2.75
(±0.24) 12.75
(±1.89)
PVP K360 2.45
(±0.19) 13.56
(±4.32) 81.52
(±5.99)
This argument was indeed supported by the measured viscosity of the crystallizing
medium in the presence PVP of different molecular weights. For PVP K10 to K40, the
solution viscosity ranges from 1.51 to 1.54 mPa·s at a minimum shear rate of 128 s-1, and
2.69 to 2.72 mPa·s at a maximum shear rate of 5000 s-1, showing only 1~2% increase
Chapter5 Application of Surfactants and Polymeric Additives
110
when the molecular weight increases from K10 to K40; however, for PVP K360, the
solution viscosity is 1.84 mPa·s at 128 s-1, which is 19% higher compared to that of PVP
K40, and 3.03 mPa·s at 5000 s-1, corresponding to 11% increase compared to that of PVP
K40.
The following conclusions can be drawn from the above results and discussion:
with PVP of a lower molecular weight (K10 and K40), inhibition of SS crystal growth is
controlled by the adsorption of the polymer on the crystal surface; whereas for the higher
molecular weight PVP (K360), the crystal growth inhibition mechanism is diffusion
controlled. This latter observation is consistent with Simonelli et al. (1970), wherein the
authors noted that the inhibition of crystal growth (sulfathiazole) was only controlled by
the relative rates of diffusion of the polymer (PVP) to the crystal surface.
5.5.3 Variation of crystal surface energy
Surface energy is defined as the excess energy at the surface of a material compared
to the bulk. It is analogous to surface tension of a liquid and directly related to the
thermodynamic work of adhesion between two materials. Surface energy can be divided
into two main components, dispersive (or non-polar) and polar components. The previous
crystallization experimental results and the molecular analysis suggest that when PVP is
adsorbed on the SS crystal surface, the non-polar functional groups of PVP should be
exposed. This would in turn be reflected by a change in the dispersive surface energy.
Based on this assumption, IGC analysis was performed with non-polar probes to
determine surface energy of SS crystallized from solutions added with PVP K25 at
different concentrations. IGC uses the net retention volume ( NV ) of the adsorbates to
Chapter5 Application of Surfactants and Polymeric Additives
111
calculate the dispersive component of surface energy of the sample. Because the probes
are injected at infinite dilution, they will only interact with the most energetic sites of the
solid. The solid surface energy ( ds ) can be obtained from:
RTlnVN a γ / 2N γ / 5 1
where a is the molecular area of adsorbate; N is the Avogadro’s Number; ds and d
l are
dispersive components of surface free energy of the solid and liquid probe respectively. C
is a constant (Djordjevic et al., 1992). By plotting NVRT ln vs. 2/1)( dlVa for n-alkane
non-polar adsorbates, the slope of the straight line gives the dispersive component of
surface energy ( ).
Table 5-5 shows measured results of dispersive surface energy values ( ds ) of SS and
PVP K25 particles. It can be seen that the values of dispersive surface energy of SS
crystals decreased with the increase in PVP K25 concentration in solution. Although IGC
does not provide face-specific information, the measured values of dispersive surface
energy reflects the total amount of non-polar functional groups exposed on the SS crystal
surfaces. Taking into consideration of SS crystal habit changes in the presence of the
polymer additive, simultaneous reduction in the values of dispersive surface energy could
be reasoned in two different ways. First, it could be due to the deposition of PVP on the
surface of SS crystals along the b direction through favorable interaction between the
vinyl pyrrolidone groups of PVP with the polar faces of SS crystals. As shown in Table
5-5, PVP K25 has a lower dispersive surface energy as compared to SS. Therefore, as the
concentration of PVP K25 increases and SS crystals are increasingly being covered by
the additive, the dispersive surface energy of SS crystal decreases. Second, the shortening
Chapter5 Application of Surfactants and Polymeric Additives
112
of SS crystals along the b direction with the addition of PVP leads to a higher proportion
of polar faces (i.e. {110} and {111}) being exposed in the crystal habit, then resulting in
a decrease of dispersive surface energy. By both means, reduction of dispersive surface
energy is in agreement with our proposed mechanism that underpins the effect of PVP on
crystallization of SS.
Table 5-5 Dispersive surface energy values ( ds ) of salbutamol sulphate and
polyvinylpyrrolidone (PVP) K25 particles
Samples ds (mJ/m2)
SS (unprocessed) 117.5
SS/PVP (1.0 mg/mL) 104.9
SS/PVP (2.5 mg/mL) 86.0
SS/PVP (5.0 mg/mL) 73.1
PVP 54.3
5.6 Conclusions
Antisolvent crystallization in the presence of several surfactants and polymeric
additives was carried out to directly produce SS particles of size distribution suitable for
inhalation drug delivery. Among the additives investigated, PVP K25 was found to be
able to produce fine SS crystals in the crystal size and size distribution suitable for
administration of inhalation drugs. The SS crystals obtained in the presence of 2.5
mg/mL PVP K25 were shown to have the smallest particle size and remained stabilized
even after 24 hours.
Chapter5 Application of Surfactants and Polymeric Additives
113
By studying the molecular interactions between the additives and SS crystal surface,
it is proposed that PVP adsorbs selectively on the crystal surface and inhibits growth
along the b-axis. Experimental data on the effect of PVP concentration and molecular
weight have also been obtained and found to be supportive of the proposed mechanism.
Surface dispersive energy measurement by IGC suggests that PVP is adsorbed on the
crystal surface, which agrees with the proposed mechanism.
It can be seen that our modeling analysis gives a good qualitative picture of the
intermolecular interactions between the polymer and the crystal surface. The proposed
mechanism provides a reasonable explanation of the effects of concentration, molecular
weight and surface adsorption of the polymer additive in antisolvent crystallization.
However, a more rigorous approach would involve performing molecular dynamics
calculations by considering the adsorption profile of the monomer unit (or a larger
fragment of polymer chain) with the crystal face as a function of simulation time (Konkel
and Myerson, 2008), and this may be considered as an extension to this work.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
116
Chapter 6
Process Study of Semi-batch Antisolvent
Crystallization
Based upon experimental observations on the habit and size modification in SS
crystals and study on intermolecular interaction between additives and SS crystal surface
(cf. chapter 5), it is proposed that additives adsorb selectively on the crystal surface based
on the H-bonding ability of the additive and the crystal surfaces. In this chapter, a process
study of semi-batch antisolvent crystallization in the presence of additive PVP is reported.
The influence of addition of PVP on nucleation and crystal growth were investigated at
the predetermined operating conditions.
6.1 Introduction
Semi-batch crystallization is a widely used operation mode in pharmaceutical
industry. Various experimental studies of antisolvent crystallization in an agitated semi-
batch vessel indicate that the crystal size distribution (CSD) depends strongly on the
operating conditions, such as agitation rate, mode of addition (direct or reverse), feeding
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
117
rate, solvent composition, and size of the crystallizer (Charmolue and Rousseau, 1991;
Mydlarz and Jones, 1991; Kaneko et al., 2002; Borissova et al., 2004). Most variations in
the operating conditions have a direct influence on the mixing of the antisolvent and the
solution, which affects the localized supersaturation and, thus, the crystal product.
Agitation driven by an impellor plays a key role in mixing of the solution and
antisolvent in a semi-batch crystallization vessel. Increasing agitation intensity can
enhance bulk circulation in the tank (macromixing), increase fluid velocity passing the
feeding pipe to improve disintegration of feed plume (mesomixing), and speed up
turbulent energy dissipation rate (micromixing). The energy dissipation rate near the
agitator is much higher than in other positions, for example, it can be higher by two
orders than that near liquid surface. This indicates a shorter micromixing time near the
agitator than in other positions (Assirelli et al., 2002). In this work, the addition position
of the antisolvent will be chosen near the agitator in order to minimize local
supersaturation profile.
Feeding configuration might have considerable influence on the final crystalline
product because different feeding configurations have different temporal and spatial
supersaturation profiles. In a system where micro-mixing time is much shorter than
induction time, nucleation takes place throughout the crystallizer and single-jet
configuration could meet general control demands (Karpinski and Wey, 2002 ). Yu (2006)
estimated the micro-mixing times at various stirrer speeds at the feeding position of near
impeller blade using the same semi-batch crystallization setup with the current work,
finding that the resultant micro-mixing time ranged from 0.024 s at 200 rpm to 0.0063 s
at 600 rpm. The mixing times are much shorter than the induction time of antisolvent
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
118
crystallization of SS where ethanol is used as the antisolvent, ruling out the necessity of
complex feeding configurations (double-jet or with premixing device). Therefore in this
work, single-jet will be chosen as the feeding configuration for studying of semi-batch
antisolvent crystallization in the current work.
Feeding rate of feeding streams also plays significant roles on controlling resultant
crystal quality because it can change the environment of local supersaturation or average
global supersaturation, manipulate crystallization kinetics, and thus influence physical
characteristics of the final product (Chang et al., 2006; Nowee et al., 2008).
Based on the above analysis, it can be seen that different operating variables act on
different aspects of mixing and crystallization, some of which are more influential than
others, and their importance depends on the properties of specific crystallizations. For
example in the current work, there was no difference observed in resulting crystal size for
SS system when the antisolvent addition mode was switched from normal to reverse.
Therefore as a preliminary study, only primary variables such as agitation intensity and
antisolvent feeding rate will be included. The effect of addition of PVP K25 on the semi-
batch crystallization process of SS will be also discussed.
6.2 Experimental
6.2.1 Experimental setup
For the semi-batch crystallization, SS (99.0%) supplied by Junda Pharm Chem
Plant Co., Ltd (Jiangsu, China) was used without further purification. Analytical grade
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
119
ethanol supplied by Sigma-Aldrich (Singapore) was used as the antisolvent for
precipitation of SS.
The experimental set-up for semi-batch antisolvent crystallization studies is shown
in Fig. 6-1 schematically. A flat-bottomed glass crystallizer vessel with volume of 1-liter
was used in the study. The vessel was fitted with four glass baffles on the inner wall and
with inner diameter of 100 mm. A 4-Pitched Blade Turbine impeller (45⁰ pitch, stainless
steel) with a diameter of 40 mm driven by a variable speed overhead stirrer motor was
used to generate agitation. The gap between the impeller and the crystallizer bottom was
about one third of the final working height of the suspension. The temperature in the
crystallizer was controlled by a heating and refrigerated circulator (Julabo FP50-HL). It
has a PID temperature controller and was connected to the control computer via a RS232
cable to receive temperature set points and send status message. A variable-speed
peristaltic pump (MasterFlex 7550) was employed to add anti-solvent. The injection point
was 5 mm above the impeller tip to accelerate dispersion. A FBRM probe (Model D600L,
Lasentec) was inserted into the turbulent zone of the suspension to obtain CLD
information. A Dipper-210 ATR-FTIR immersion probe (Axiom Analytical Inc.)
equipped with a ZnSe crystal was inserted into the vessel at the same level as the FBRM
probe.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
120
Figure 6-1 Experimental set-up for semi-batch antisolvent crystallization.
6.2.2 Experiments
Calibration of ATR-FTIR. Calibration of the IR spectra for SS solutions was
conducted in a 500-mL jacketed round bottom flask. Stirring was generated by a
magnetic stirrer (IKA RCT basic) which was operating at 500 rpm to ensure homogeneity
of its contents. The temperature in the crystallizer was controlled by a heating and
cooling circulator (Julabo, model FP50-HL). At the beginning of each calibration run, the
temperature was elevated to ~10 °C above the saturation point and maintained there for
20 min to ensure that all crystals have dissolved. The crystallizer then was cooled at a
rate of 0.5 °C/min while spectral data were collected every 2 min.
Thermocouple ATR-FTIR
Pum
FBRM
Circulato
Impelle
Overhead motor
Crystallization vessel
Computer
Baffle
Antisolvent
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
121
Temperature, SS concentration, solvent composition and additive concentration
need to be altered systematically to cover all the possible regions encountered during
antisolvent crystallization in the space spanned by t, csu, csv and cadd. Three SS
concentrations were prepared at every solvent composition and additive concentration.
Temperature was changed in the following ranges: 5-20 ºC, 10-30 ºC and 15-40 ºC.
Detailed experimental conditions are listed in Table 6-1.
Table 6-1 Solvent compositions, salbutamol sulphate and polyvinylpyrrolidone (PVP) concentrations used in calibration experiment
c (%)
c (g/g)
c (mg/g) 20-5(°C) 30-10(°C) 40-15(°C)
0 0.3492 0.3689 0.3936 0.5010
20 0.2851 0.3183 0.3556 1.0570
40 0.2210 0.2527 0.2865 11.07
60 0.1121 0.1288 0.1478 0.5740
80 0.02539 0.03121 0.03339 9.109
100 0.003276 0.003276 0.003276 0.1080
Solubility measurement via ATR-FTIR. Solubility measurements of SS with and
without the presence of PVP were performed in the same 500-mL jacketed round bottom
flask for calibration. Stirring was generated by a magnetic stirrer (IKA RCT basic) which
was operating at 500 rpm to ensure homogeneity of its contents. The temperature in the
crystallizer was controlled at 23C by a heating and cooling circulator (Julabo, model
FP50-HL). An excessive amount of SS was added in the ethanol-water mixture of
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
122
specific weight composition. For the measurement of solubility in the presence of PVP,
appropriate amount of PVP K25 was added into the water to ensure the final
concentration of PVP in the mixed solvent is 2.5 mg/mL. The ATR probe was inserted
into the slurry to collect spectra at regular interval. The calibration models obtained in
last section were integrated in a VB program which received spectral data from the ATR-
FTIR machine. The calculated solution concentration was recorded and displayed on the
computer screen every time a spectrum was collected. After the measured concentration
had stabilized for 30 minutes, appropriate amounts of ethanol and SS powders were
added into the crystallizer adjust the composition to the next desired value. Repeat the
process in the ethanol concentration range of 0-100 wt%.
Semi-batch crystallization. Saturated aqueous solutions at 23 °C were prepared by
adding excessive amount of SS into water and then stabilizing over night to ensure that
SS solutions reach equilibrium. After that the slurry was filtered by vacuum filtration and
then 100 mL of the solution was transferred to the crystallization vessel. 900 mL of
ethanol was added into the vessel by the peristaltic pump which was activated at a given
flow rate. The record of FBRM and ATR-FTIR was initiated simultaneously after
addition of ethanol started. Crystallization was allowed to continue for around 250 min
and then the products were isolated from the slurry by vacuum filtration followed by
drying in an oven at 40 °C for 24 hours.
Three series of experiments have been conducted, with one for the effect of feeding
rate of antisolvent (Series 1), one for the effects of agitation speed (Series 2), and the
other for the effects of addition of PVP (Sires 3). Table 6-2 lists experimental conditions
for 3 series of experiments.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
123
6-2 Experimental conditions for semi-batch antisolvent crystallization
Series Feeding rate
(mg/mL) Stirring Speed
(rpm) Addition of PVP
K25
1
10
200 500
- 20
30
2 20
100
-
300
500
800
3 20 500 2.5 mg/mL
6.3 Results and discussion
6.3.1 Calibration of Attenuated Total Reflectance Fourier Transform Infrared
(ATR-FTIR)
The application of ATR-FTIR spectroscopy for in-situ monitoring the concentration
and supersaturation in crystallization systems have been reported previously (Groen and
Roberts, 2004; Yu et al., 2007). In this chapter, solubility, real-time concentration and
supersaturation of SS solution were determined using this methodology. Before
measurement is carried out, correlation between IR spectra and solution composition
must be established. Calibration of ATR-FTIR is based on the Lambert-Beer law (Cornel
et al., 2008), according to which the change in intensity (absorbance or transmittance) of
the IR spectral peaks specific for the solute, solvent and antisolvent is proportional to
their concentration in the solution.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
124
A proper calibration model should be selected for calibration being carried out.
Accuracy is the first and foremost consideration. Secondly, criteria such as ease to
implement and burden on experimental data should also be considered. Some calibration
models were developed using peak intensity at characteristic wavelengths. For example,
Dunuwila et al. (1994) proposed an exponential calibration model correlating the relative
transmittance of citric acid solution at a certain wave length with the solute and solvent
concentrations; Grön et al. (2003) adopted a power form to correlate spectra with the
concentration of monosodium glutamate in water. There are also few models were
developed based on peak area in narrow ranges of wavelengths. Feng and Berglund (2002)
correlated the ratio of peak areas with the concentration of succinic acid in water. It gave
better results in terms of reproducibility and accuracy than the calibration models based
on peak intensity at specific wavelengths. In spite of the fact that only a small number of
experimental data are required to determine the limited number of model parameters, it is
difficult to obtain a proper function form by manual selection of characteristic
wavelengths for the abovementioned calibration models. In addition, overlapping of the
characteristic peaks for different compounds may turn up as an obstacle to the selection
of characteristic wavelengths.
Chemometric models can solve these problems. It enables correlations of
absorbance in a wide range of wavelengths with any variation in solution composition.
There is no need to manually select characteristic peaks in chemometric calibration
models. A glimpse of the spectra graph suffices to identify the range of wavelengths that
reflects the absorbance feature of the systems under consideration. Then the spectra data
in this range can be fed into proper chemometrics software. It has been shown that by this
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
125
type of calibration models, higher accuracy can be obtained as compared with those
models replying on one or two wavelength points (Togkalidou et al., 2001; Liotta and
Sabesan, 2004).
For SS system, it is observed that characteristic absorbance peaks are located in the
range from of 800-1800 cm-1. Therefore, absorbance in this range (including 519 discrete
wavelengths) and temperature were used as predictor variables. The 644 sets of data
collected during experiment were ordered randomly. Then they were divided into two
groups: one is calibration group and the other is validation group. The randomization of
data order is necessary to provide sufficient coverage of experimental space for the
establishment and validation of calibration model. Three components, solvent (ethanol)
composition, solute (SS) concentration and additive (PVP K25) concentration were
calibrated. The calibration results were calculated via chemometrics. This method has
been proven to produce the mean width of prediction interval (MWPI) as low as 3.54
×10-4 g/g for paracetamol in water/acetone system (Yu, 2006).
The prediction residuals were obtained with MWPI value of ± 2.49×10-4 g/ g for SS
concentration (Fig. 6-3), ±6.77×10-5 mg/g for PVP concentration (Fig. 6-4) and
±2.29×10-4 for solvent composition (Fig. 6-5). The accuracy of prediction intervals of SS
concentration, PVP concentration and solvent composition are sufficient for solubility
and supersaturation measurement.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
126
Figure 6-3 Residuals of a chemometric calibration model for salbutamol sulphate concentration.
‐1.5
‐1
‐0.5
0
0.5
1
1.5
5 10 15 20 25 30 35 40
Residule * 10³ (g/g)
Temperature (°C)
‐1.5
‐1
‐0.5
0
0.5
1
1.5
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Residule * 10³ (g/g)
SS concentration (g/g)
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
127
Figure 6-4 Residuals of a chemometric calibration model for solvent composition.
‐0.4
‐0.3
‐0.2
‐0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
5 10 15 20 25 30 35 40
Residule * 10³ (g/g)
Temperature (°C)
‐0.4
‐0.3
‐0.2
‐0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0 1 2 3 4 5 6 7 8 9 10
Residule * 10³ (m
g/g)
PVP concentration (mg/g)
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
128
Figure 6-5 Residuals of a chemometric calibration model for additive composition.
6.3.2 Solubility measurement
The measured solubility data via ATR-FTIR is shown in Fig. 6-6. A high
antisolvent ratio is prerequisite in attaining a high supersaturation level and hence
producing small particles in a nucleation-dominated antisolvent crystallization process.
‐1.5
‐1
‐0.5
0
0.5
1
1.5
5 10 15 20 25 30 35 40
Residule * 10³ (g/g)
Temperature (°C)
‐1.5
‐1
‐0.5
0
0.5
1
1.5
0 10 20 30 40 50 60 70 80 90 100
Residule * 10³ (%
)
Solvent composition (%)
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
129
Therefore in this chapter, the solubility of SS in the presence of PVP K25 (2.5 mg/mL)
was measured only at high ethanol compositions (70~100%).
Figure 6-6 Solubility data measured by Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR).
It has been reported that when polymers are added into the solution, the solubility
of the solute may increase, resulting in a decrease in supersaturation. The decrease in
supersaturation therefore leads to reduction of growth rate, causing the crystal size and
habit change (Loftsson et al., 1996). Rodriguez-Hornedo and Murphy (1999) also
suggested that polymeric additives do not inhibit nucleation but just increase solubility
and thus decrease the supersaturation. However in this work, it is found that the solubility
of SS in the absence of PVP K25 approximates that in the presence of 2.5 mg/mL PVP
K25 at higher ethanol composition (Fig. 6-6). Therefore, the inhibitive effect of PVP on
the crystal size and habit observed in this work could not be due to the reduction of the
solubility caused by addition of polymer. Similar observation has been reported by
Iervolino et al. (2000) in which addition of 1% of HPMC did not change the solubility of
0.0000
0.0500
0.1000
0.1500
0.2000
0.2500
0.3000
0.3500
0 20 40 60 80 100 120
Solubility (g/g)
Ethanol composition (%)
without PVP
with PVP
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
130
Ibuprofen and Alexanian el al. (2008) in which PVP had minor effects on the aqueous
solubility of nimesulide.
6.3.3 Effects of feeding rate
Effects of feeding rate at different agitation speeds (Series 1 in Table 6-2) were
investigated with the aid of FBRM. The development of total chord counts at various
feeding rate for the agitation rate of 200 and 500 rpm are shown in Fig. 6-7a and Fig. 6-
7b respectively.
Different phases of a crystallization process can be identified in both figures:
induction, nucleation and crystal growth. The induction time started from the addition of
anti-solvent and ended when counts began to grow significantly. The counts of chord
lengths remained very low and almost constant during this period. Nucleation period
refers to the interval during which counts rose sharply. Most of the nuclei were created
during this stage. Nucleation period was followed by crystal growth period where total
counts of chord lengths were kept relatively constant.
At agitation speed of 200 rpm, the onset of nucleation appeared at around 120, 70
and 30 min at feeding rate of 10, 20 and 30 mL/min respectively (Fig. 6-7a). For feeding
rate of 10 and 20 mL/min, nucleation was detected after completion of antisolvent
addition, while for feeding rate of 30 mL/min, nucleation was observed before
completion of antisolvent addition. A shorter induction time resulted from a higher
feeding rate indicates a higher rate of supersaturation generation, and thus an increased
nucleation rate. As a result, larger numbers of chord counts were observed.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
131
(a)
(b)
Figure 6-7 Evolutions of total chord counts at various feeding rates when agitation speed is (a) 200 rpm and (b) 500 rpm.
0
5000
10000
15000
20000
25000
30000
35000
0 50 100 150 200 250
Total chord counts (s‐1)
Time (min)
10 ml/min
20 ml/min
30 ml/min
0
5000
10000
15000
20000
25000
30000
35000
40000
0 50 100 150 200 250
Counts of chord lengths (s
‐1)
Time (min)
10 ml/min
20 ml/min
30 ml/min
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
132
At agitation speed of 500 rpm, the induction time was 102 min at feeding rate of 10
mL/min, and decreased rapidly to 30 min at 20 mL/min of feeding rate. However in
contrast to 200 rpm, the induction time did not decrease when the feeding rate was
increased to 30 mL/min. The difference in induction time and total chord counts can be
neglected for feeding rates of 20 and 30 mL/min.
The comparison of the induction time at different stirring speeds suggests that the
system is limited by mixing. Mixing in a crystallization vessel mainly occurs by three
mechanisms. Micromixing is the intimate mixing that occurs at molecular and near
molecular length scales. It is a diffusional process. Macromixing is the mixing driven by
the largest scales motion in the fluid. It is characterized by the blend time in a batch
system. Mesomixing is the mixing occurring at intermediate length scales (Paul et al.,
2004). In general, at high stirring speed, fast macromixing and micromixing bring
solution and antisolvent in contact rapidly and results in more nuclei formed. For high
supersaturation rates (>20 mL/min), the system is independent on feeding rate of the
system. This shows that micromixing is the limiting step in the progress of fast nucleation,
because micromixing dramatically accelerates the rate of production of interfacial area
available for diffusion. However at low mixing speeds, the system is limited by
macromixing. Macromixing limits the dispersion of the antisolvent throughout the vessel,
thus the nucleation is affected by the feeding rate (Myerson, 2002).
This fact coincides with the CLD measured at different feeding rates with the
stirring speed of 500 rpm (Fig. 6-8). It can be seen that with increasing feeding rate from
10 to 20 mL/min, the average chord length decreased and the total chord counts increased.
When feeding rate was increased from 20 to 30 mL/min, the chord count showed
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
133
negligible difference and chord length, however, increased slightly. Offline PSD
measurement (Table 6-3) is consistent with the CLD obtained with FBRM.
Figure 6-8 Chord length distributions at the time interval of 1 hr after onset of nucleation at different feeding rate with stirring speed of 500 rpm.
Table 6-3 Particle Size Distribution obtained at different feeding rate with stirring speed of 500 rpm.
Feeding rate (mL/min)
Particle size distribution (μm)
d(0.1) d(0.5) d(0.9)
10 30.86 44.21 61.25
20 1.82 10.79 49.21
30 2.54 21.68 52.29
This can be explained by the effect of agglomeration. It is known that increased
feeding rate can lead to an increased rate of supersaturation generation, which is expected
to result in stronger nucleation and hence in a product where the crystal size is smaller
0
2000
4000
6000
8000
10000
12000
14000
1 10 100 1000
Total chord counts (s‐1)
Chord length (μm)
10 ml/min
20 ml/min
30 ml/min
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
134
and crystal number is larger, and this is in accordance with our experimental results under
20 mL/min. Nonetheless, higher feeding rate also results in higher degree of
agglomeration (Yu, 2006), and the presence of more fines promotes agglomeration
further (Zumstein and Rousseau, 1989). Therefore for feeding rate of 30 mL/min,
agglomeration may counteract the decrease in particle size by increasing feeding rate,
thus showing no obvious size reduction as compared to 20 mL/min. As shown in Fig. 6-9
and Table 6-3, particles obtained at low feeding rate (10 mL/min) display less degree
Figure 6-9 Images of crystals obtained at different feeding rate of (a) 10 mL/min, (b) 20 mL/min and (3) 30 mL/min with stirring speed of 500 rpm.
agglomeration and narrower particle size distribution as compared to those obtained at
higher feeding rates (20 and 30 mL/min). It should also be noted that for all feeding rates
(a)
(b) (c)
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
135
at 500 rpm, the particle size is bigger than the target particle size for inhalation
administration.
6.3.4 Effects of agitation speed
Effects of agitation speed on the induction time was investigated (Series 2 in Table
6-2) with the aid of FBRM. The development of total chord counts at various stirring
rates is shown in Fig. 6-10.
Figure 6-10 Evolutions of total counts of chord length at various stirring speed when feeding rate is 20 mL/min.
It is observed that the agitation rate has a strong influence upon the induction time.
With the increase in stirring speed from 100 to 500 rpm, the induction time decreased and
the chord counts notably increased (Fig. 6-10). However when agitation speed is further
increased to 800 rpm, only a slight reduction in induction time and enhancement in
0
5000
10000
15000
20000
25000
30000
35000
40000
0 100 200 300
Total chord counts (s‐1)
Time (min)
100 rpm
300 rpm
500 rpm
800 rpm
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
136
particle population were observed. As described in chapter 2, two steps have been
considered to involve in the nucleation process: the diffusional transport step, followed
by integration of molecules into the crystal lattice (Mullin and Raven, 1961). According
to Mullin (2001), stronger stirring has greater intensity to form nuclei as compared to a
gentle stirring. This is consistent with the current work where the increasing agitation
speeds ranged from 100 to 500 rpm resulted in decreasing induction time. However,
when agitation reaches a particular level, solute diffusion is maximized, and the effect of
agitation upon the nucleation rate of solutions is insignificant (Bonsdorff-Nikander et al.,
2003). This is in accordance with our experimental results where imperceptible difference
in nucleation rate was found between stirring speed of 500 and 800 rpm. It also from
another perspective verified our hypothesis that the crystallization at high stirring speed is
micromixing controlled.
The above observation on the total chord counts coincides with chord length
distributions measured in the experiments in Series 2. As shown in Fig. 6-11, the peak of
chord length distribution shifts toward the lower value when agitation speed is increased
from 100 rpm to 500 rpm, whereas there was no further reduction in chord length when
agitation speed increased from 500 to 800 rpm. Stronger agitation also leads to an
increase in the total count of chord lengths with the agitation speed changing from 100 to
500 rpm, and there was no obvious difference in total count of chord lengths between 500
and 800 rpm.
Crystal size distribution of the particles obtained at 4 hrs after the onset of
nucleation (Table 6-4) shows consistent result with the data obtained from FBRM.
Crystal size is primarily determined by the number of crystals that shares the total
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
137
Figure 6-11 Chord length distributions at time interval of 4 hrs after onset of nucleation at various agitation speeds when feeding rate of 20 mL/min.
crystallized mass. Therefore an increasing stirring speed from 100 to 500 rpm induced
larger number of crystals and hence generated smaller particle size. No large difference
in particle counts was observed for stirring rate changed from 500 to 800 rpm, as a result
no further size reduction was detected for the particles obtained from the above stirring
speeds.
Table 6-4 Particle Size Distribution of particles obtained at different agitation speed with feeding rate of 20 mL/min.
Stirring speed (rpm)
Particle size distribution (μm)
d(0.1) d(0.5) d(0.9)
100 5.62(±0.94) 29.83(±2.74) 286.83(±55.77)
300 2.93(±0.43) 20.83(±2.36) 132.55(±28.73)
500 1.22(±0.14) 9.96(±0.84) 59.80(±17.28)
800 1.65(±0.62) 9.67(±1.75) 57.49(±17.05)
0
2000
4000
6000
8000
10000
12000
14000
1 10 100 1000
Counts of chord lengths (s‐1)
Chold length (μm)
100 rpm
300 rpm
500 rpm
800 rpm
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
138
High degree of agglomeration was observed at low stirring intensity such as 100
and 300 rpm, which can be evidenced by the shoulder at upper chord length for 300 rpm
shown in Fig. 6-11 and the SEM image for 100 rpm shown in Fig. 6-12a. Agglomeration
can be attributed to the larger particle size measurement at lower agitation rate. With a
low agitation speed, antisolvent fed into the crystallizer did not blend quickly with the
solution, forming many pockets of fluids rich in antisolvent. Longer macromixing time
allowed these pockets to exist for longer time, which induced excessive nucleation due to
higher local supersaturation and led to a large amount of fines. The increasing number of
fine particles coupled with higher local supersaturation promoted agglomeration. At low
agitation intensity, agglomeration predominated over nucleation and crystal growth in the
course. With increasing stirrer speeds up to 500 rpm, anti-solvent was dispersed more and
more rapidly, as a result the appearance of high-supersaturation pockets was eliminated
and localized nucleation was suppressed. The crystallization process switched to
nucleation-dominated regime as evidenced by the great increase in the chord counts from
300 to 500 rpm (Fig. 6-10 and Fig. 6-11). Therefore particles obtained at high agitation
rate displayed less degree of agglomeration and small particle size as shown in Fig. 6-12b.
Insignificant enlargement in particle counts with further increase in agitation above 500
rpm indicated the crystallization process was shifted from nucleation-dominated regime
to growth-dominated regime. Simultaneously at high agitation intensity, more
agglomerates were fractured by shear force or direct mechanical impact from crystallizer
inner fittings, offsetting the particle enlargement. This might explain the reason why there
was no obvious variation in particle size when agitation increased from 500 to 800 rpm.
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
139
Figure 6-12 Images of crystals obtained at agitation speed of (a) 100 rpm and (b) 500 rpm.
6.3.5 Effect of addition of polyvinylpyrrolidone (PVP)
The effect of addition of PVP K25 in the semi-batch crystallization was
investigated (Series 3 in Table 6-2). As shown in Fig. 6-13, addition of PVP retarded the
nucleation of SS from the solution. For the system in the absence of PVP, the first
particle was detected by FBRM probe at 40 minutes, whereas for the solution in the
presence of PVP, nucleation started after 75 minutes. The induction times are related to
the effect of additive on the crystal nucleation rate. The longer the induction time for
crystallization, the greater is the inhibitory effect of the polymer additive on nucleation
kinetics (Nielsen and Sohnel, 1971).
(a) (b)
(a) (b)
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
140
Figure 6-13 Effect of addition of polyvinylpyrrolidone (PVP) K 25 on evolutions of the chord length counts at feeding rate of 20 mL/min and 500 rpm.
The increase in induction time in the presence of polymers has been observed in a
number of previous works (Ma et al., 1996; Usui et al., 1997; Raghavan et al., 2001;
Jensen et al., 2008; Tian et al., 2009). Raghavan et al. (2001) reported the influence of
polymers on the antisolvent crystallization of hydrocortisone acetate and explained the
mechanism of nucleation retardation by the polymers in terms of association of the solute
molecules with the polymer in solution through hydrogen bonding. This is indeed
consistent with the molecular modeling analysis presented in the previous chapter in this
work, where the PVP recurring units were found to interact with SS molecules via
hydrogen bonding. For nucleation to occur, these hydrogen bonds have to be broken for
the SS molecules to diffuse and form a critical nucleus. Therefore the interactions
between SS molecules and PVP may inhibit or retard the formation of nuclei through
collision.
0
10000
20000
30000
40000
50000
60000
0 50 100 150 200 250
Counts of hord lengths (s
Time (min)
without PVP
with PVP
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
141
Fig. 6-14 compares the CLD for the systems with and without the presence of
additive. The peak of CLD shifted towards lower value with the addition of PVP,
indicating the inhibitive effect by using PVP as the stabilizing agent. This is validated by
the result of particle size measured by Malvern Mastersizer shown at Table 6-5. The
d(0.5) value was reduced from 10 to 4 μm when PVP was added into the crystallization
vessel.
Figure 6-14 Chord length distributions of crystals at time interval of 1 h after onset of nucleation at feeding rate of 20 mL/min and 500 rpm.
Supersaturation measured by ATR-FTIR in situ provides a clear picture of the
driving force behind the phenomena observed in Fig 6-13 and 6-14. The supersaturation
ratio in the course of crystallization of SS with and without the presence of PVP are
shown in Fig. 6-15. For instance, the peak position shifted towards the right with addition
of PVP, representing a longer induction time which is in agreement with the results
0
200
400
600
800
1000
1200
1400
1600
1800
2000
1 10 100 1000
Counts of chord lengths (s‐1)
Chord length (μm)
without PVP
with PVP
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
142
obtained from FBRM. The maximum supersaturation value built up with the presence of
PVP is remarkly higher as compared to the crystallization stystem in the absence of PVP.
High superstauration favors formation of nuclei and thus leads to small size of the
crystalline product. This is consistent with the result of FBRM and Malvern mastersizer.
Table 6-5 Particle Size Distribution of particles obtained with and without the presence of polyvinylpyrrolidone (PVP)
Particle size distribution (μm)
d(0.1) d(0.5) d(0.9)
SS 1.22(±0.14) 9.96(±0.84) 59.80(±17.28)
SS/PVP 1.01(±0.08) 4.07(±0.27) 20.58(±2.53)
Figure 6-15 Effect of addition of polyvinylpyrrolidone (PVP) K 25 on supersaturation ratio evolutions at feeding rate of 20 mL/min and 500 rpm.
0
2
4
6
8
10
12
14
16
18
0 50 100 150 200 250
Relative supersaturation ratio (‐)
time (min)
without PVP
with PVP
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
143
6.4 Conclusion
A semi-batch crystallization process with the addition of PVP has been developed
to obtain micron-size salbutamol sulphate particles.
For the crystallization process without the presence of PVP, both induction time and
particle size were found to be independent on the high feeding rate at high stirring speed,
whereas this phenomenon was not observed for the low feeding rate. It suggests that the
system was governed by different mixing regimes with different feeding rates. The
system is limited by micromixing at high mixing intensity, whereas macromixing is the
controlling regime when mixing intensity is low.
Intensive agitation led to an enhanced nucleation rate, increased particle population
and reduced particle size. Agglomeration was found to be attributed to the larger size
measurement at low agitation speeds. Increased agitation intensity sped up the dispersion
of anti-solvent and reduced localized nucleation, switching crystallization system from
agglomeration control regime to nucleation control regime, leading to reduction of
particle size. Further increase in agitation intensity turned the system into growth control
regime. At the same time, breakage of agglomerates caused by high stirring speed pulled
down the average particle size, offsetting the particle size measurement results.
Addition of PVP K25 (2.5 mg/mL) was found to be able to retard nucleation rate,
and inhibit crystal growth. These results support the mechanism proposed in the previous
chapter that several types of hydrogen bonds exist between SS molecules and PVP
repeating units. In the liquid phase, the difficulty in breaking of the hydrogen bonds due
to the interaction prevents solute molecules from diffusing to form the nuclei, therefore
Chapter 6 Process Study of Semi-batch Antisolvent Crystallization
144
resulting in a prolonged induction time. On the other hand, these hydrogen bonds formed
on the SS surfaces hinder the accumulation of solute molecules on the surfaces, leading
to an inhibitive effect on crystal growth.
Chapter 7 Conclusions and Future Work
145
Chapter 7
Conclusions and Future Work
Addition of surfactants and polymeric additives to crystallization processes can
have great impact on resultant crystal size. This is particularly important to inhalation
drug production for which small particle size and tight particle size distribution are of key
importance for achieving high drug delivery efficiency. A good knowledge of the role of
the additives on crystallization processes perhaps could provide useful guidelines for
pharmaceutical industries to design novel technologies to optimize solid-state attributes
of inhalation products. With this objective, this work aims at the application of
surfactants and polymeric additives to direct precipitation of SS and understanding the
role of these additives on crystal growth. The main findings are briefly reviewed herein.
7.1 Main contributions and findings
7.1.1 Screening of operational parameters in antisolvent crystallization of SS
Screening of operating parameters was conducted in antisolvent crystallization
process. Various antisolvents were found to have different effects on induction time,
Chapter 7 Conclusions and Future Work
146
crystal morphology, size and crystallinity of SS. SS can precipitate out rapidly from two
aprotic solvents namely acetone and acetonitrile; whereas the induction time in the protic
antisolvents are much longer. Aprotic antisolvents produce plate-shape crystals, whereas
protic solvents produce needle-shape crystals. This could be explained by the different
nature of the antisolvents and the interaction between the antisolvent molecules and the
specific facet of SS. Resultant particle size measurement was found to depend on
supersaturation generated by various antisolvents with different solvent/antisovlent
volumetric ratios. Aggregation was observed for aprotic solvents and protic solvents at
high supersaturation, which in turn, affect the particle size measurement. The observation
on aggregation is due to the sudden increase of the surface energy resulted from rapid
addition of the antisolvent. Based on the experimental screening results for various
antisolvents, ethanol was selected as the antisolvent in the subsequent antisolvent
crystallization experiments with volume ratio of antisolvent to solvent of 9.
7.1.2 Application of surfactant and polymers
The experimental investigation has demonstrated that addition of several types of
surfactants and polymers has different effect on particle habit and size. Among them PVP
K25 was found to be able to produce fine SS crystals in the crystal size and size
distribution suitable for administration of inhalation drugs. Molecular modeling was
employed to study the mechanism by which the abovementioned additives impact the
resultant particle size. The growth morphology of SS was predicted from the crystal
structure provided by CSD database and obtained from the evaporative crystallization
experiments in the current work via molecular modeling-based simulation techniques
Chapter 7 Conclusions and Future Work
147
such as the BFDH and the attachment energy methods. All resultant morphologies
exhibited the main crystal faces were {200}, {002} and {110} and growth direction was
along the b-axis, which is consistent with our experimental morphology. It has been
proposed that additives adsorb selectively on the crystal surface based on the H-bonding
ability of the additive and the crystal surface, leading to different inhibitive effect on the
crystal growth. This is in accordance with the experimental data on the effect of PVP
concentration and molecular weight and the surface dispersive energy measurement on
the crystals obtained in the presence of various amount of PVP K25. Therefore our
proposed mechanism enables a good qualitative understanding of the intermolecular
interactions between the additive and the crystal surface and thereby providing a useful
guideline for designing of novel technologies to optimize solid-state attributes of
inhalation products.
7.1.3 Process study of semi-batch antisolvent crystallization
The process study of semi-batch antisolvent crystallization of SS has shown that the
final size of the crystalline product was sensitive to the operating parameters such as
feeding rate of antisolvent, stirring speed, and addition of PVP. Induction time and
particle size were highly dependent on the feeding rate at low stirring speed, however
independent on the high feeding rate at high stirring speed, suggesting that the system
was governed by different mixing regimes with different addition rates. In addition,
induction time and particle size were reduced and particle population was increased with
the increase in stirring speed. Addition of PVP in the semi-batch antisolvent
crystallization process showed the same inhibitive effect on the crystal size as the
Chapter 7 Conclusions and Future Work
148
previous beaker-scale study. Furthermore, it was found that nucleation rate was retarded
and particle population was enhanced in the presence of PVP in comparison those in the
absence of PVP. These findings verified our previous proposed mechanism that hydrogen
bonds exist between SS molecules and PVP repeating units.
Thus summarizing, this thesis addresses the application of surfactants and
polymeric additives to antisolvent crystallization in production of micron-size particles of
SS. The proposed mechanism enables a good qualitative understanding of the
intermolecular interactions between the additive and the crystal surface and thereby
providing a useful guideline for designing of novel technologies to optimize solid-state
attributes of inhalation products. In addition, direct precipitation of SS in semi-batch
crystallization provides a potential protocol in producing inhalation drugs in large scales
without the need for intensive milling.
7.2 Future work
7.2.1 Supersaturation control via application of ATR-FTIR technique
Current study has shown that PSD is depended on supersaturation level in the solute
solution. It has also been reported that control on the generation rate of supersaturation in
order for it to match with the growing surface area can yield a better PSD (Hojjati et al.,
2007). For semi-batch crystallization, a controllable generation rate of supersaturation
can be achieved by a predetermined time-varying flow rate. However, predetermining
such a flow-rate profile requires prior knowledge of the exact crystallization kinetics,
which is difficult and time-consuming to obtain for industrial conditions. In particular,
Chapter 7 Conclusions and Future Work
149
the presence of impurity content often causes inconsistency in the results and is difficult
to be tracked precisely from batch to batch, rendering the predetermined profile not
optimal anymore. ATR-FTIR as a real-time measurement technique can be utilized in
crystallization control without the requirement of kinetic knowledge. In the future work,
it can serve as a feedback tool, by which the flow rate of antisolvent containing an
additive can be manipulated accordingly.
7.2.2 Scale-up of anti-solvent crystallization
It is challenging to reproduce key solid-state attributes of the resultant crystals when
switching bench crystallization to a scaling-up commercial application. It often results in
significant changes to crystal size distribution (CSD), purity and morphology, which are
key determinants of product quality and has implications for downstream operations such
as filtration. Due to the different dependence of dimensionless numbers on scales, it is
impractical to achieve the same mixing status in commercial crystallizers as in bench
crystallizers. Up to date only some rules of thumb appear in open literature for scale-up
of crystallization and there seems to be lack of knowledge of know-how to success. In the
future work, the performance of optimal crystallization protocol obtained from bench
experiments can be investigated in a larger scale crystallizer and interplay between
inherent properties of crystals. Operating conditions can also be explored to generalize
the outcome of scale-up study.
Chapter 7 Conclusions and Future Work
150
7.2.3 Solvent-influenced morphology prediction
Molecular modeling has increasingly been recognized as a powerful tool to predict
the morphology of an organic compound. A proper morphology prediction of a
compound crystallizing from a solution should take into account of the solvent properties
because the solvent can specifically interacts with certain crystal faces (Lahav and
Leiserowitz, 2001). With such a solvent-influenced morphology prediction, it is possible
to screen and select solvent for crystallization of organic compounds from solution, when
the shape of the crystals is of large importance. The current work has demonstrated that
different solvents can impact product morphology of SS in the antisolvent crystallization
process. In the molecular modeling, interfacial surface energies for a variety of solvent
molecules will be computed and combined with the calculations involving solution-
effected attachment energies (Hammond et al., 2006). The resultant morphology of SS
crystals grown from these solvents in the previous crystallization experiments will be
compared to validate the modeling results.
7.2.4 Molecular modeling on stabilization effect
The stabilization of crystalline material is indeed non-trivial as the number of
pharmaceutically acceptable surfactants is limited. However, it is very time consuming to
screen each new drug compound empirically with the various surfactant combinations to
find the most stable formulation. Molecular modeling approach can be used for rapid
surfactant screening, which could identify stabilizing surfactant systems for a particular
drug, based on its crystal structure (Konkel and Myerson, 2008). This work has
successfully probed the stabilization effect of addition of PVP in the antisolvent
Chapter 7 Conclusions and Future Work
151
crystallization process by studying the interaction between the additive and SS molecules.
To further understand the mechanism behind and hence apply this technique in surfactant
screening, a more rigorous approach involving performing molecular dynamics
calculations would be needed by considering the adsorption profile of the monomer unit
(or a larger fragment of polymer chain) with the crystal face as a function of simulation
time.
Direct Precipitation of Micron-size Salbutamol Sulphate
152
References
Abendan, R.S., Swift, J.A., 2005. Dissolution on Cholesterol Monohydrate Single-Crystal
Surfaces Monitored by in Situ Atomic Force Microscopy. Crystal Growth & Design 5, 2146-2153.
Ahfat, N.M., Buckton, G., Burrows, R., Ticehurst, M.D., 2000. An Exploration of Inter-Relationships between Contact Angle, Inverse Phase Gas Chromatography and Triboelectric Charging Data. European Journal of Pharmaceutical Sciences 9, 271-276.
Alexanian, C., Papademou, H., Vertzoni, M., Archontaki, H., Valsami, G., 2008. Effect of Ph and Water-Soluble Polymers on the Aqueous Solubility of Nimesulide in the Absence and Presence of Beta-Cyclodextrin Derivatives. Journal of Pharmacy and Pharmacology 60, 1433-1439.
Arunmozhi, G., Jayavel, R., Subramanian, C., 1997. Experimental Determination of Metastable Zone Width, Induction Period and Interfacial Energy of Lap Family Crystals. Journal of Crystal Growth 178, 387-392.
Assirelli, M., Bujalski, W., Eaglesham, A., Nienow, A.W., 2002. Study of Micromixing in a Stirred Tank Using a Rushton Turbine - Comparison of Feed Positions and Other Mixing Devices. Chemical Engineering Research & Design 80, 855-863.
Baldyga, J., Bourne, J.R., 1995. Interpretation of Turbulent Mixing Using, Fractals and Multifractals. Chemical Engineering Science 50, 381-400.
Direct Precipitation of Micron-size Salbutamol Sulphate
153
Beckmann, W., 1999. Nucleation Phenomena During the Crystallisation and Precipitation of Abecarnil. Journal of Crystal Growth 198, 1307-1314.
Bennema, P., 1992. Theory of Growth and Morphology Applied to Organic Crystals: Possible Application to Protein Crystals. Journal of Crystal Growth, 110-119.
Bennema, P. 1995. In Science and Technology of Crystal Growth, edited by J. P. van der Eerder and O. S. L. Bruinsma: Kluwer Academic Publishers.
Bennema, P., 1995. Morphology of Crystals: Past and Future. In: Van Der Eerder, J.P., Bruinsma, O.S.L. (Eds.), Science and Technology of Crystal Growth. Kluwer Academic Publishers.
Bennett, W.D., Brown, J.S., Zeman, K.L., Hu, S.C., Scheuch, G., Sommerer, K., 2002. Targeting Delivery of Aerosols to Different Lung Regions. Journal of Aerosol Medicine-Deposition Clearance and Effects in the Lung 15, 179-188.
Berard, V., Lesniewska, E., Andres, C., Pertuy, D., Laroche, C., Pourcelot, Y., 2002. Affinity Scale between a Carrier and a Drug in Dpi Studied by Atomic Force Microscopy. International Journal of Pharmaceutics 247, 127-137.
Berkovitchyellin, Z., 1985. Toward an Ab Initio Derivation of Crystal Morphology. Journal of the American Chemical Society 107, 8239-8253.
Bhavna, Ahmad, F.J., Mittal, G., Jain, G.K., Malhotra, G., Khar, R.K., Bhatnagar, A., 2009. Nano-Salbutamol Dry Powder Inhalation: A New Approach for Treating Broncho-Constrictive Conditions. European Journal of Pharmaceutics and Biopharmaceutics 71, 282-291.
Bisker-Leib, V., Doherty, M.F., 2001. Modeling the Crystal Shape of Polar Organic Materials: Prediction of Urea Crystals Grown from Polar and Nonpolar Solvents. Crystal Growth & Design 1, 455-461.
Direct Precipitation of Micron-size Salbutamol Sulphate
154
Bisker-Leib, V., Doherty, M.F., 2003. Modeling Crystal Shape of Polar Organic Materials: Applications to Amino Acids. Crystal Growth & Design 3, 221-237.
Bonsdorff-Nikander, A., Rantanen, J., Christiansen, L., Yliruusi, J., 2003. Optimizing the Crystal Size and Habit of Beta-Sitosterol in Suspension. AAPS PharmSciTech 4, 116-123.
Borissova, A., Dashova, Z., Lai, X., Roberts, K.J., 2004. Examination of the Semi-Batch Crystallization of Benzophenone from Saturated Methanol Solution Via Aqueous Antisolvent Drowning-out as Monitored in-Process Using Atr Ftir Spectroscopy. Crystal Growth & Design 4, 1053-1060.
Bravais, A., 1913. Etudes Crystallographiquies. Academie des Sciences. Paris.
Brunsteiner, M., Jones, A.G., Pratola, F., Price, S.L., Simons, S.J.R., 2005. Toward a Molecular Understanding of Crystal Agglomeration. Crystal Growth & Design 5, 3-16.
Brunsteiner, M., Price, S.L., 2001. Morphologies of Organic Crystals: Sensitivity of Attachment Energy Predictions to the Model Intermolecular Potential. Crystal Growth & Design 1, 447-453.
Buckley, H.E., 1951. Crystal Growth. John Wiley & Sons. New York.
Chan, H.K., 2006. Dry Powder Aerosol Drug Delivery - Opportunities for Colloid and Surface Scientists. Colloids Surf. A 284, 50-55.
Chang, S.M., Kim, J.M., Kim, I.H., Shin, D.M., Kim, W.S., 2006. Agglomeration Control of L-Ornithine Aspartate Crystals by Operating Variables in Drowning-out Crystallization. Industrial & Engineering Chemistry Research 45, 1631-1635.
Direct Precipitation of Micron-size Salbutamol Sulphate
155
Charmolue, H., Rousseau, R.W., 1991. L-Serine Obtained by Methanol Addition in Batch Crystallization. Aiche Journal 37, 1121-1128.
Charpentier, P.A., Jia, M., Lucky, R.A., 2008. Study of the Ress Process for Producing Beclomethasone-17,21-Dipropionate Particles Suitable for Pulmonary Delivery. AAPS PHARMSCITECH 9, 39-46.
Chen, J.X., Wang, J.K., Ulrich, J., Yin, Q.X., Xue, L.Z., 2008. Effect of Solvent on the Crystal Structure and Habit of Hydrocortisone. Crystal Growth & Design 8, 1490-1494.
Chikhalia, V., Forbes, R.T., Storey, R.A., Ticehurst, M., 2006. The Effect of Crystal Morphology and Mill Type on Milling Induced Crystal Disorder. European Journal of Pharmaceutical Sciences 27, 19-26.
Chiou, H., Li, L., Hu, T.T., Chan, H.K., Chen, J.F., Yun, J., 2007. Production of Salbutamol Sulfate for Inhalation by High-Gravity Controlled Antisolvent Precipitation. International Journal of Pharmaceutics 331, 93-98.
Chow, A.H.L., Tong, H.H.Y., Chattopadhyay, P., Shekunov, B.Y., 2007. Particle Engineering for Pulmonary Drug Delivery. Pharmaceutical Research 24, 411-437.
Clontz, N.A., McCabe, W.L., 1971. Contact Nucleation of Magnesium Sulfate Heptahydrate. Chemical Engineering Progress Symposium Series 110, 6-17.
Clydesdale, G., Hammond, R.B., Ramachandran, V., Roberts, K.J., 2005. Molecular Modelling of the Morphology of Organic Crystals in the Presence of Impurity Species: Recent Applications to Naphthalene, Phenanthrene, and Caprolactam Crystals. Molecular Crystals and Liquid Crystals 440, 235-257.
Clydesdale, G., Roberts, K.J., Docherty, R., 1996. Habit95 - a Program for Predicting the Morphology of Molecular Crystals as a Function of the Growth Environment. Journal of Crystal Growth 166, 78-83.
Direct Precipitation of Micron-size Salbutamol Sulphate
156
Clydesdale, G., Thomson, G.B., Walker, E.M., Roberts, K.J., Meenan, P., Docherty, R., 2005. A Molecular Modeling Study of the Crystal Morphology of Adipic Acid and Its Habit Modification by Homologous Impurities. Crystal Growth & Design 5, 2154-2163.
Coombes, D.S., Catlow, C.R.A., Gale, J.D., Hardy, M.J., Saunders, M.R., 2002. Theoretical and Experimental Investigations on the Morphology and Pharmaceutical Crystals. Journal of Pharmaceutical Sciences 91, 1652-1658.
Cornel, J., Lindenberg, C., Mazzotti, M., 2008. Quantitative Application of in Situ Atr-Ftir and Raman Spectroscopy in Crystallization Processes. Industrial & Engineering Chemistry Research 47, 4870-4882.
Dailey, L.A., Schmehl, T., Gessler, T., Wittmar, M., Grimminger, F., Seeger, W., Kissel, T., 2003. Nebulization of Biodegradable Nanoparticles: Impact of Nebulizer Technology and Nanoparticle Characteristics on Aerosol Features. Journal of Controlled Release 86, 131-144.
Davey, R.J., 1979. In: De Jong E.J. And Jancic, S.J. (Eds), Industrial Crystallization. North-Halland. Amsterdam.
Davey, R.J., Garside, J., 2000. From Molecules to Crystallizers: An Introduction to Crystallization. Oxford University Press, UK.
Davey, R.J., Milisavljevic, B., Bourne, J.R., 1988. Solvent Interactions at Crystal-Surfaces - the Kinetic Story of Alpha-Resorcinol. Journal of Physical Chemistry 92, 2032-2036.
Dean, J.A., 1992. Lange's Handbook of Chemistry, Mcgraw-Hill, New York.
Direct Precipitation of Micron-size Salbutamol Sulphate
157
Dhanaraj, P.V., Bhagavannarayana, G., Rajesh, N.P., 2008. Effect of Amino Acid Additives on Crystal Growth Parameters and Properties of Ammonium Dihydrogen Orthophosphate Crystals. Materials Chemistry and Physics 112, 490-495.
Djordjevic, N.M., Rohr, G., Hinterleitner, M., Schreiber, B., 1992. Adsorption of Water on Cyclosporin a, from Zero to Finite Surface Coverage. International Journal of Pharmaceutics 81, 21-29.
Donnay, J.D.H., Harker, D., 1937. A New Law of Crystal Morphology Extending the Law of Bravais. Am. Mineralogist 22, 446-447.
Dudley, M.N., Loutit, J., Griffith, D.C., 2008. Aerosol Antibiotics: Considerations in Pharmacological and Clinical Evaluation. Current Opinion in Biotechnology 19, 637-643.
Dunuwila, D.D., Berglund, K.A., 1997. Atr Ftir Spectroscopy for in Situ Measurement of Supersaturation. Journal of Crystal Growth 179, 185-193.
Dunuwila, D.D., Carroll, L.B., Berglund, K.A., 1994. An Investigation of the Applicability of Attenuated Total Reflection Infrared Spectroscopy for Measurement of Solubility and Supersaturation of Aqueous Citric Acid Solutions Journal of Crystal Growth 137, 561-568.
Dunuwila, D.D., Carroll, L.B., Berglund, K.A., 1994. An Investigation of the Applicability of Attenuated Total Reflection Infrared Spectroscopy for Measurement of Solubility and Supersaturation of Aqueous Citric Acid Solutions. Journal of Crystal Growth 137, 561-568.
Edwards, D.A., Hanes, J., Caponetti, G., Hrkach, J., BenJebria, A., Eskew, M.L., Mintzes, J., Deaver, D., Lotan, N., Langer, R., 1997. Large Porous Particles for Pulmonary Drug Delivery. Science 276, 1868-1871.
Direct Precipitation of Micron-size Salbutamol Sulphate
158
Elwenspoek, M., Bennema, P., Vandereerden, J.P., 1987. Orientational Order in Naphthalene Crystal Solution Interfaces. Journal of Crystal Growth 83, 297-305.
Feeley, J.C., York, P., Sumby, B.S., Dicks, H., 1998. Determination of Surface Properties and How Characteristics of Salbutamol Sulphate, before and after Micronisation. International Journal of Pharmaceutics 172, 89-96.
Femioyewo, M.N., Spring, M.S., 1994. Studies on Paracetamol Crystals Produced by Growth in Aqueous Solutions. International Journal of Pharmaceutics 112, 17-28.
Feng, L.L., Berglund, K.A., 2002. Atr-Ftir for Determining Optimal Cooling Curves for Batch Crystallization of Succinic Acid. Crystal Growth & Design 2, 449-452.
Floury, J., Bellettre, J., Legrand, J., Desrumaux, A., 2004. Analysis of a New Type of High Pressure Homogeniser. A Study of the Flow Pattern. Chemical Engineering Science 59, 843-853.
Frank, F.C., 1949. The Influence of Dislocations on Crystal Growth. Faraday Discussions 5, 33.
Frank, T.C., Downey, J.R., Gupta, S.K., 1999. Quickly Screen Solvents for Organic Solids. Chemical Engineering Progress 95, 41-61.
Friedel, G., 1907. Etudes Sur La Loi De Bravais. Bulletin de la Societe Francaise Mineralogique 20, 326-455.
Garekani, H.A., Ford, J.L., Rubinstein, M.H., Rajabi-Siahboomi, A.R., 2000. Highly Compressible Paracetamol: I: Crystallization and Characterization. International Journal of Pharmaceutics 208, 87-99.
Direct Precipitation of Micron-size Salbutamol Sulphate
159
Ginde, R.M., Myerson, A.S., 1993. Effect of Impurity on Cluster Growth and Nucleation. Journal of Crystal Growth 126, 216-222.
Gosavi, R.A., Bhamidi, V., Varanasi, S., Schall, C.A., 2009. Beneficial Effect of Solubility Enhancers on Protein Crystal Nucleation and Growth. Langmuir 25, 4579-4587.
Groen, H., Roberts, K.J., 2004. An Examination of the Crystallization of Urea from Supersaturated Aqueous and Aqueous-Methanol Solutions as Monitored in-Process Using Atr Ftir Spectroscopy. Crystal Growth and Design 4, 929-936.
Gron, H., Mougin, P., Thomas, A., White, G., Wilkinson, D., Hammond, R.B., Lai, X.J., Roberts, K.J., 2003. Dynamic in-Process Examination of Particle Size and Crystallographic Form under Defined Conditions of Reactant Supersaturation as Associated with the Batch Crystallization of Monosodium Glutamate from Aqueous Solution. Industrial & Engineering Chemistry Research 42, 4888-4898.
Hammond, R.B., Pencheva, K., Roberts, K.J., 2006. A Structural-Kinetic Approach to Model Face-Specific Solution/Crystal Surface Energy Associated with the Crystallization of Acetyl Salicylic Acid from Supersaturated Aqueous/Ethanol Solution. Crystal Growth & Design 6, 1324-1334.
Hartman, P., Perdok, W.G., 1955. On the Relations between Structure and Morphology of Crystals. Acta Crystallographica 8, 49-52.
Haynes, A., Shaik, M.S., Chatterjee, A., Singh, M., 2003. Evaluation of an Aerosolized Selective Cox-2 Inhibitor as a Potentiator of Doxorubicin in a Non-Small-Cell Lung Cancer Cell Line. Pharmaceutical Research 20, 1485-1495.
Heath, A.R., Fawell, P.D., Bahri, P.A., Swift, J.D., 2002. Estimating Average Particle Size by Focused Beam Reflectance Measurement (Fbrm). Particle & Particle Systems Characterization 19, 84-95.
Direct Precipitation of Micron-size Salbutamol Sulphate
160
Heng, J.Y.Y., Thielmann, F., Williams, D.R., 2006. The Effects of Milling on the Surface Properties of Form I Paracetamol Crystals. Pharmaceutical Research 23, 1918-1927.
Hojjati, H., Sheikhzadeh, M., Rohani, S., 2007. Control of Supersaturation in a Semibatch Antisolvent Crystallization Process Using a Fuzzy Logic Controller. Industrial & Engineering Chemistry Research 46, 1232-1240.
Hukkanen, E.J., Braatz, R.D., 2003. Measurement of Particle Size Distribution in Suspension Polymerization Using in Situ Laser Backscattering. Sensors and Actuators B-Chemical 96, 451-459.
Iervolino, M., Raghavan, S.L., Hadgraft, J., 2000. Membrane Penetration Enhancement of Ibuprofen Using Supersaturation. International Journal of Pharmaceutics 198, 229-238.
Jensen, L., Thomsen, K., von Solms, N., 2008. Propane Hydrate Nucleation: Experimental Investigation and Correlation. Chemical Engineering Science 63, 3069-3080.
Johnson, B.K., Prud'homme, R.K., 2003. Chemical Processing and Micromixing in Confined Impinging Jets. Aiche Journal 49, 2264-2282.
Jones, A.G., 2002. Crystallization Process Systems. Butterworth-Heinemann, Oxford.
Kail, N., Marquardt, W., Briesen, H., 2009. Process Analysis by Means of Focused Beam Reflectance Measurements. Industrial & Engineering Chemistry Research 48, 2936-2946.
Kaneko, S., Yamagami, Y., Tochihara, H., Hirasawa, I., 2002. Effect of Supersaturation on Crystal Size and Number of Crystals Produced in Antisolvent Crystallization. Journal of Chemical Engineering of Japan 35, 1219-1223.
Direct Precipitation of Micron-size Salbutamol Sulphate
161
Karpinski, P.H., Wey, J.S., 2002 Precipitation Processes, In: Myerson, A. S (Ed.) Handbook of Industrial Crystallization. Boston, Butterworth Heinemann: 231-348.
Karpinski, P.H., Wey, J.S., 2002 Precipitation Processes. A. S. Myerson. Handbook of Industrial Crystallization. Boston, Butterworth Heinemann: 231-348.
Karunanithi, A.T., Achenie, L.E.K., Gani, R., 2006. A Computer-Aided Molecular Design Framework for Crystallization Solvent Design. Chemical Engineering Science 61, 1247-1260.
Kashchiev, D., 2000. Nucleation − Basic Theory with Applications Butterworth- Heinemann, Oxford
Kashchiev, D., Firoozabadi, A., 2002. Nucleation of Gas Hydrates. Journal of Crystal Growth 243, 476-489.
Keel, T.R., Thompson, C., Davies, M.C., Tendler, U.B., Roberts, C.J., 2004. Afm Studies of the Crystallization and Habit Modification of an Excipient Material, Adipic Acid. International Journal of Pharmaceutics 280, 185-198.
Kemp, W., 1987. Organic Spectroscopy. 2nd Ed. English Language Book Society Macmillan.
Kitamura, M., Nakamura, K., 2002. Dependence of Polymorphic Transformation on Anti-Solvent Composition and Crystallization Behavior of Thiazole-Derivative Pharmaceutical. Journal of Chemical Engineering of Japan 35, 1116-1122.
Klymchenko, A.S., Kenfack, C., Duportail, G., Mely, Y., 2007. Effects of Polar Protic Solvents on Dual Emissions of 3-Hydroxychromones. Journal of Chemical Sciences 119, 83-89.
Direct Precipitation of Micron-size Salbutamol Sulphate
162
Konkel, J.T., Myerson, A.S., 2008. Empirical Molecular Modelling of Suspension Stabilisation with Polysorbate 80. Molecular Simulation 34, 1353-1357.
Kreuter, J., 1991. Nanoparticle-Based Dmg Delivery Systems. Journal of Controlled Release 16, 169-176.
Kubota, N., 2001. Effect of Impurities on the Growth Kinetics of Crystals. Crystal Research and Technology 36, 749-769.
Kumar, S., Chawla, G., Bansal, A.K., 2008. Role of Additives Like Polymers and Surfactants in the Crystallization of Mebendazole. Yakugaku Zasshi-Journal of the Pharmaceutical Society of Japan 128, 281-289.
Kundig, A.A., Loffler, J.F., Johnson, W.L., Uggowitzer, P.J., Thiyagarajan, P., 2001. Influence of Decomposition on the Thermal Stability of Undercooled Zr-Ti-Cu-Ni-Al Alloys. Scripta Materialia 44, 1269-1273.
Lahav, M., Leiserowitz, L., 2001. The Effect of Solvent on Crystal Growth and Morphology. Chemical Engineering Science 56, 2245-2253.
Larhrib, H., Martin, G.P., Marriott, C., Prime, D., 2003. The Influence of Carrier and Drug Morphology on Drug Delivery from Dry Powder Formulations. International Journal of Pharmaceutics 257, 283-296.
Law, D.J., Bale, A.J., Jones, S.E., 1997. Adaptation of Focused Beam Reflectance Measurement to in-Situ Particle Sizing in Estuaries and Coastal Waters. Marine Geology 140, 47-59.
Leger, J.M., Goursolle, M., Gadret, M., Carpy, A., 1978. Crystal Structure of Salbutamol Sulfate [2-Tert-Butylamino-1-(4-Hydroxy-3-Hydroxymethylphenyl)Ethanol.1/2h2so4]. Acta Crystallographica. Section B: Structural Crystallography and Crystal Chemistry 34, 1203-1208.
Direct Precipitation of Micron-size Salbutamol Sulphate
163
Lindenberg, C., Krattli, M., Cornel, J., Mazzotti, M., Brozio, J., 2009. Design and Optimization of a Combined Cooling/Antisolvent Crystallization Process. Crystal Growth & Design 9, 1124-1136.
Liotta, V., Sabesan, V., 2004. Monitoring and Feedback Control of Supersaturation Using Atr-Ftir to Produce an Active Pharmaceutical Ingredient of a Desired Crystal Size. Organic Process Research & Development 8, 488-494.
Loftsson, T., Fridriksdottir, H., Gudmundsdottir, K., 1996. The Effect of Water-Soluble Polymers on Aqueous, Solubility of Drugs. International Journal of Pharmaceutics 127, 293-296.
Lowery, T.H., Richardson, K.S., 1987. Mechanism and Theory in Organic Chemistry. 3rd Ed. Harper Collins Publishers.
Ma, X.G., Taw, J., Chiang, C.M., 1996. Control of Drug Crystallization in Transdermal Matrix System. International Journal of Pharmaceutics 142, 115-119.
Mahajan, A.J., Kirwan, D.J., 1996. Micromixing Effects in a Two-Impinging-Jets Precipitator. Aiche Journal 42, 1801-1814.
Maiti, A., Gee, R.H., 2009. Modeling Growth, Surface Kinetics, and Morphology Evolution in Petn. Propellants Explosives Pyrotechnics 34, 489-497.
Malkki, L., Tammilehto, S., 1990. Decomposition of Salbutamol in Aqueous Solutions. I. The Effect of Ph, Temperature and Drug Concentration. International Journal of Pharmaceutics 63, 17-22.
Mani, N., Jun, H.W., Beach, J.W., Nerurkar, J., 2003. Solubility of Guaifenesin in the Presence of Common Pharmaceutical Additives. Pharmaceutical Development and Technology 8, 385-396.
Direct Precipitation of Micron-size Salbutamol Sulphate
164
Martins, P.M., Rocha, F.A., Rein, P., 2005. Modeling Sucrose Evaporative Crystallization. Part 1. Vacuum Pan Monitoring by Mass Balance and Image Analysis Methods. Industrial & Engineering Chemistry Research 44, 8858-8864.
Mayo, S.L., Olafson, B.D., Goddard, W.A., 1990. Dreiding: A Generic Force Field for Molecular Simulations. Journal of Physical Chemistry 94, 8897-8909.
Modarresi, H., Conte, E., Abildskov, J., Gani, R., Crafts, P., 2008. Model-Based Calculation of Solid Solubility for Solvent Selection - a Review. Industrial & Engineering Chemistry Research 47, 5234-5242.
Montgomery, A.B., 1992. In: Hickey, A. J., Ed. Pharmaceutical Inhalation Aerosol Technology. Marcel Dekker: New York.
Moran, A., Buckton, G., 2007. Adjusting and Understanding the Properties and Crystallisation Behaviour of Amorphous Trehalose as a Function of Spray Drying Feed Concentration. International Journal of Pharmaceutics 343, 12-17.
Mullin, J.W., 2001. Crystallization.4th Ed. Butterworth-Heinemann. Oxford.
Mullin, J.W., Raven, K.D., 1961. Nucleation in Agitated Solutions. Nature, 190-251.
Murnane, D., Marriott, C., Martin, G.P., 2008. Comparison of Salmeterol Xinafoate Microparticle Production by Conventional and Novel Antisolvent Crystallization. European Journal of Pharmaceutics and Biopharmaceutics 69, 94-105.
Murnane, D., Marriott, C., Martin, G.P., 2008. In Situ and Ex Situ Analysis of Salmeterol Xinafoate Microcrystal Formation from Poly(Ethylene Glycol) 400 - Water Cosolvent Mixtures. Crystal Growth and Design 8, 1855-1862.
Direct Precipitation of Micron-size Salbutamol Sulphate
165
Muster, T.H., Prestidge, C.A., 2002. Face Specific Surface Properties of Pharmaceutical Crystals. Journal of Pharmaceutical Sciences 91, 1432-1444.
Mydlarz, J., Jones, A.G., 1991. Crystallization and Agglomeration Kinetics During the Batch Drowning-out Precipitation of Potash Alum with Aqueous Acetone. Powder Technology 65, 187-194.
Myerson, A.S., 2002. Handbook of Industrial Crystallization. Butterworth-Heinemann, Oxford.
Myerson, A.S., Lo, P.Y., 1991. Cluster Formation and Diffusion in Supersaturated Binary and Ternary Amino-Acid Solutions. Journal of Crystal Growth 110, 26-33.
Nagao, L.M., Lyapustina, S., Munos, M.K., Capizzi, M.D., 2005. Aspects of Particle Science and Regulation in Pharmaceutical Inhalation Drug Products. Crystal Growth & Design 5, 2261-2267.
Nagy, Z.K., Fujiwara, M., Braatz, R.D., 2008. Modelling and Control of Combined Cooling and Antisolvent Crystallization Processes. Journal of Process Control 18, 856-864.
Nakada, T., Sazaki, G., Miyashita, S., Durbin, S.D., Komatsu, H., 1999. Direct Afm Observations of Impurity Effects on a Lysozyme Crystal. Journal of Crystal Growth 196, 503-510.
Nass, K.K., 1994. Rational Solvent Seletion for Cooling Crystallizations. Industrial & Engineering Chemistry Research 33, 1580-1584.
Ng, W.K., Kwek, J.W., Tan, R.B.H., 2008. Anomalous Particle Size Shift During Post-Milling Storage. Pharmaceutical Research 25, 1175-1185.
Direct Precipitation of Micron-size Salbutamol Sulphate
166
Nielsen, A.E., 1964. Kinetics of Precipitation. Pergamon Press. New York.
Nielsen, A.E., Sohnel, O., 1971. Interfacial Tension Electrolyte Crystal-Aqueous Solution, from Nucleation Data. Journal of Crystal Growth 11, 233-242.
Nocent, M., Bertocchi, L., Espitalier, F., Baron, M., Couarraze, G., 2001. Definition of a Solvent System for Spherical Crystallization of Salbutamol Sulfate by Quasi-Emulsion Solvent Diffusion (Qesd) Method. Journal of Pharmaceutical Sciences 90, 1620-1627.
Nowee, S.M., Abbas, A., Romagnoli, J.A., 2008. Antisolvent Crystallization: Model Identification, Experimental Validation and Dynamic Simulation. Chemical Engineering Science 63, 5457-5467.
Ohara, M., Reid, R.C., 1973. Modeling Crystal Growth Rates from Solutions. Prentice-Hall. Englewood Cliffs, NJ.
Padron, G., Eagles, W.P., Ozcan-Taskin, G.N., McLeod, G., Xie, L., 2008. Effect of Particle Properties on the Break up of Nanoparticle Clusters Using an in-Line Rotor-Stator. Journal of Dispersion Science and Technology 29, 580-586.
Patton, J.S., Bukar, J., Nagarajan, S., 1999. Inhaled Insulin. Advanced Drug Delivery Reviews 35, 235-247.
Paul, E.L., Atiemo-Obeng, V.A., Kresta, S.M., 2004. Handbook of Industrial Mixing: Science and Practice, Hoboken, N. J, Wiley & Sons.
Phillips, J.M., Walling, D.E., 1998. Calibration of a Par-Tec 200 Laser Back-Scatter Probe for in Situ Sizing of Fluvial Suspended Sediment. Hydrological Processes 12, 221-231.
Direct Precipitation of Micron-size Salbutamol Sulphate
167
Poornachary, S.K., Chow, P.S., Tan, R.B.H., 2008. Effect of Solution Speciation of Impurities on Alpha-Glycine Crystal Habit: A Molecular Modeling Study. Journal of Crystal Growth 310, 3034-3041.
Poornachary, S.K., Chow, P.S., Tan, R.B.H., 2008. Impurity Effects on the Growth of Molecular Crystals: Experiments and Modeling. Advanced Powder Technology 19, 459-473.
Qian, F., Mathias, N., Moench, P., Chi, C., Desikan, S., Hussain, M., Smith, R.L., 2009. Pulmonary Delivery of a Glp-1 Receptor Agonist, Bms-686117. International Journal of Pharmaceutics 366, 218-220.
Raghavan, S.L., Trividic, A., Davis, A.F., Hadgraft, J., 2001. Crystallization of Hydrocortisone Acetate: Influence of Polymers. International Journal of Pharmaceutics 212, 213-221.
Randolph, A.D., Maurice, A.L., 1988. Theory of Particulate Processes: Analysis and Theniques of Continuous Crystallization, 2nd Ed. Academic Press. London, UK.
Rappe, A.K., Goddard, W.A., 1991. Charge Equilibration for Molecular Dynamics Simulations. Journal of Physical Chemistry 95, 3358-3363.
Rasenack, N., Muller, B.W., 2002. Crystal Habit and Tableting Behavior. International Journal of Pharmaceutics 244, 45-57.
Rasenack, N., Muller, B.W., 2004. Micron-Size Drug Particles: Common and Novel Micronization Techniques. Pharmaceutical Development and Technology 9, 1-13.
Rodriguez-Hornedo, N., Murphy, D., 1999. Significance of Controlling Crystallization Mechanisms and Kinetics in Pharmaceutical Systems. Journal of Pharmaceutical Sciences 88, 651-660.
Direct Precipitation of Micron-size Salbutamol Sulphate
168
Roelands, C.P.M., Jiang, S.F., Kitamura, M., ter Horst, J.H., Kramer, H.J.M., Jansens, P.J., 2006. Antisolvent Crystallization of the Polymorphs of L-Histidine as a Function of Supersaturation Ratio and of Solvent Composition. Crystal Growth & Design 6, 955-963.
Ruf, A., Worlitschek, J., Mazzotti, M., 2000. Modeling and Experimental Analysis of Psd Measurements through Fbrm. Particle & Particle Systems Characterization 17, 167-179.
Sangwal, K., 1996. Effects of Impurities on Crystal Growth Processes. Progress in Crystal Growth and Characterization of Materials 32, 3-43.
Sangwal, K., 2007. Additives and Crystallization Processes: From Fundamentals to Applications. John Wiley and Sons. England.
Sangwal, K., 2009. Effect of Impurities on the Metastable Zone Width of Solute-Solvent Systems. Journal of Crystal Growth 311, 4050-4061.
Sangwal, K., Mielniczek-Brzoska, E., 2004. Effect of Impurities on Metastable Zone Width for the Growth of Ammonium Oxalate Monohydrate Crystals from Aqueous Solutions. Journal of Crystal Growth 267, 662-675.
Schurch, S., Gehr, P., Hof, V.I., Geiser, M., Green, F., 1990. Surfactant Displaces Particles toward the Epithelium in Airways and Alveoli. Respiration Physiology 80, 17-32.
Shekunov, B.Y., Chattopadhyay, P., Tong, H.H.Y., Chow, A.H.L., 2007. Particle Size Analysis in Pharmaceutics: Principles, Methods and Applications. Pharmaceutical Research 24, 203-227.
Shekunov, B.Y., York, P., 2000. Crystallization Processes in Pharmaceutical Technology and Drug Delivery Design. Journal of Crystal Growth 211, 122-136.
Direct Precipitation of Micron-size Salbutamol Sulphate
169
Simon, B., Boistelle, R., 1981. Crystal-Growth from Low-Temperature Solutions. Journal of Crystal Growth 52, 779-788.
Simonelli, S.P., Mehta, S.C., Higuchi, W.I., 1970. Inhibition of Sulfathiazole Crystal Growth by Polyvinylpyrrolidone. Journal of Pharmaceutical Sciences 59, 633-637.
Sohnel, O., Garside, J., 1992. Precipitation. Butterworth-Heinemann, Oxford.
Srinivasan, K., Sherwood, J.N., 2005. Asymmetric Growth of Alpha-Resorcinol Crystals: Comparison of Growth from the Vapor Phase and from Aqueous Solution. Crystal Growth & Design 5, 1359-1370.
Sung, J.C., Pulliam, B.L., Edwards, D.A., 2007. Nanoparticles for Drug Delivery to the Lungs. Trends in Biotechnology 25, 563-570.
Tadayyon, A., Rohani, S., 1998. Determination of Particle Size Distribution by Par-Tec (R) 100: Modeling and Experimental Results. Particle & Particle Systems Characterization 15, 127-135.
Taylor, L.S., Zografi, G., 1997. Spectroscopic Characterization of Interactions between Pvp and Indomethacin in Amorphous Molecular Dispersions. Pharmaceutical Research 14, 1691-1698.
Thielmann, F., 2004. Introduction into the Characterisation of Porous Materials by Inverse Gas Chromatography. Journal of Chromatography A 1037, 115-123.
Tian, F., Baldursdottir, S., Rantanen, J., 2009. Effects of Polymer Additives on the Crystallization of Hydrates: A Molecular-Level Modulation. Molecular Pharmaceutics 6, 202-210.
Direct Precipitation of Micron-size Salbutamol Sulphate
170
Togkalidou, T., Fujiwara, M., Patel, S., Braatz, R.D., 2001. Solute Concentration Prediction Using Chemometrics and Atr-Ftir Spectroscopy. Journal of Crystal Growth 231, 534-543.
Togkalidou, T., Tung, H.H., Sun, Y.K., Andrews, A., Braatz, R.D., 2002. Solution Concentration Prediction for Pharmaceutical Crystallization Processes Using Robust Chemometrics and Atr Ftir Spectroscopy. Organic Process Research & Development 6, 317-322.
Tong, H.H.Y., Shekunov, B.Y., York, P., Chow, A.H.L., 2002. Influence of Polymorphism on the Surface Energetics of Salmeterol Xinafoate Crystallized from Supercritical Fluids. Pharmaceutical Research 19, 640-648.
Tung, H., Paul, E.L., Midler, M., McCauley, J.A., 2009. Crystallization of Organic Compounds: An Industrial Perspective. John Wiley & Sons, New Jersey.
Usui, F., Maeda, K., Kusai, A., Nishimura, K., Yamamoto, K., 1997. Inhibitory Effects of Water-Soluble Polymers on Precipitation of Rs-8359. International Journal of Pharmaceutics 154, 59-66.
van der Leeden, M.C., Kashchiev, D., Vanrosmalen, G.M., 1993. Effect of Additives on Nucleation Rate, Crystal Growth Rate and Induction Time in Precipitation. Journal of Crystal Growth 130, 221-232.
Wankat, P.C., 1990. Rate Controlled Separations, Blackie, Glasgwo.
Ward, G.H., Schultz, R.K., 1995. Process-Induced Crystallinity Changes in Albuterol Sulfate and Its Effect on Powder Physical Stability. Pharmaceutical Research 12, 773-779.
Weaver, M.L., Qiu, S.R., Hoyer, J.R., Casey, W.H., Nancollas, G.H., De Yoreo, J.J., 2007. Inhibition of Calcium Oxalate Monohydrate Growth by Citrate and the Effect of the Background Electrolyte. Journal of Crystal Growth 306, 135-145.
Direct Precipitation of Micron-size Salbutamol Sulphate
171
Weda, M., Zanen, P., de Boer, A.H., Barends, D.M., Frijlink, H.W., 2004. An Investigation into the Predictive Value of Cascade Impactor Results for Side Effects of Inhaled Salbutamol. International Journal of Pharmaceutics 287, 79-87.
Weissbuch, I., Addadi, L., Lahav, M., Leiserowitz, L., 1991. Molecular Recognition at Crystal Interfaces. Science 253, 637-645.
Weissbuch, I., Leiserowitz, L., Lahav, M., 2006. Self-Poisoning at {011} Faces of Alpha-Resorcinol Crystals May Explain Its Unidirectional Growth in the Vapor Phase: A Molecular Modeling Study. Crystal Growth & Design 6, 625-628.
Worlitschek, J., Hocker, T., Mazzotti, M., 2005. Restoration of Psd from Chord Length Distribution Data Using the Method of Projections onto Convex Sets. Particle & Particle Systems Characterization 22, 81-98.
Yu, L.X., Lionberger, R.A., Raw, A.S., D'Costa, R., Wu, H.Q., Hussain, A.S., 2004. Applicatlons of Process Analytical Technology to Crystallization Processes. Advanced Drug Delivery Reviews 56, 349-369.
Yu, Z.Q. 2006. Particle Engineering in Anti-Solvent Crystallization. Ph.D Thesis, National University of Singapore, Singapore.
Yu, Z.Q., Chew, J.W., Chow, P.S., Tan, R.B.H., 2007. Recent Advances in Crystallization Control - an Industrial Perspective. Chemical Engineering Research & Design 85, 893-905.
Yu, Z.Q., Chow, P.S., Tan, R.B.H., 2006. Application of Attenuated Total Reflectance-Fourier Transform Infrared (Atr-Ftir) Technique in the Monitoring and Control of Anti-Solvent Crystallization. Industrial & Engineering Chemistry Research 45, 438-444.
Direct Precipitation of Micron-size Salbutamol Sulphate
172
Yu, Z.Q., Tan, R.B.H., Chow, P.S., 2005. Effects of Operating Conditions on Agglomeration and Habit of Paracetamol Crystals in Anti-Solvent Crystallization. Journal of Crystal Growth 279, 477-488.
Zeng, X.M., Martin, G.P., Marriott, C., Pritchard, J., 2000. The Influence of Carrier Morphology on Drug Delivery by Dry Powder Inhalers. International Journal of Pharmaceutics 200, 93-106.
Zumstein, R.C., Rousseau, R.W., 1989. Agglomeration of Copper-Sulfate Pentahydrate Crystals within Well-Mixed Crystallizers. Chemical Engineering Science 44, 2149-2155.
Direct Precipitation of Micron-size Salbutamol Sulphate
173
List of Publications
[1] Xie, S., Poornachary, S. K., Chow, P. S., Tan, R. B. H. Direct Precipitation of Micron-Size Salbutamol Sulfate: New Insights into the Action of Surfactants and Polymeric Additives, Crystal Growth & Design, Article ASAP. DOI: 10.1021/cg901270x. Publication Date (Web): June 28, 2010
[2] Xie, S., Chow, P. S., Tan, R. B. H. Effects of Antisolvents on Crystallization of
Salbutamol Sulphate, in preparation. [3] Xie, S., Chow, P. S., Tan, R. B. H. Effects of Polymeric Additives on Semi-batch
Crystallization of Salbutamol Sulphate, in preparation.
Filename: Thesis revised version.docx Directory: D:\shuyi Template: C:\Documents and Settings\shuyi\Application Data\Microsoft\Templates\Normal.dotm Title: Direct precipitation of micron-sized salbutamol sulphate by antisolvent crystallization Subject: Author: shuyi Keywords: Comments: Creation Date: 4/25/2011 3:47:00 PM Change Number: 22 Last Saved On: 8/2/2011 5:23:00 PM Last Saved By: Total Editing Time: 5,684 Minutes Last Printed On: 8/2/2011 5:24:00 PM As of Last Complete Printing Number of Pages: 190 Number of Words: 63,467 (approx.) Number of Characters: 361,764 (approx.)