1
MULTI-SCALE CHARACTERIZATION OF ANTI-SOLVENT CRYSTALLIZATION
Des O’Grady
December 15th 2008
School of Chemical and Bio-Process Engineering
PhD Viva
2
Introduction
Utilize a relatively neglected model system where sound chemical
engineering principles can be applied to improve understanding
Focus on key stages in the pharmaceutical drug development
lifecycle
Combine existing crystallization theory with in situ analytics and
modern models to improve understanding
Develop simple and effective techniques that can be used to
improve industrial crystallization
Implement a common sense scale-up protocol based on extensive
laboratory-scale characterization
3
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
4
Chapter 3 – Solubility – Goals
Generate solubility data for model system
Investigate most suitable way to express anti-solvent solubility
Assess feasibility of different techniques and cross-validate each
method
Calibrate ATR-FTIR probe
Apply a theoretical model to measured solubility data
5
Simple Solubility Expression
0 10 20 30 40 50 60 70 800
0.1
0.2
0.3
0.4
0.5
0.6Solubility Data
Solute Concentration
anti-solvent %
g s
olu
te /
(g s
olv
ent
+ a
nti
-solv
ent)
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Solubility Data
Solute Concentration
g anti-solvent / g solvent
g s
olu
te /
g s
olv
ent
Traditional vs. “anti-solvent free” solubility expression
1. 2.
6
Gravimetric Analysis
Two Gravimetric methods were used to measure solubility Solid analysis – mass of solute left in filter cake after drying Liquid Analysis – mass of solute remaining after filtrate was
evaporated
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Solid Analysis
Liquid Analysis
(g water/ g ethanol)
(g b
enzo
ic a
cid/
g e
thanol)
7
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Water Concentration (g water/ g ethanol)
Solu
bilit
y (
g b
enzo
ic a
cid/
g
eth
anol)
FBRM Analysis
“Polythermal” method was used to generate solubility information Solubility temperature was measured at a range of anti-solvent concentrations Anti-solvent concentration vs. solubility plot was then interpolated at 25ºC This method opens possibility of 3-dimensional solubility plot
15 20 25 300.0
0.1
0.2
0.3
0.4
0.5
0.6
0.70.90 (g water/g ethanol)
1.37 (g water/g ethanol)
2.08 (g water/g ethanol)
3.37 (g water/g ethanol)
Temperature (°C)
Solu
bilit
y g
benzo
ic a
cid/g
eth
anol
8
ATR-FTIR Solubility Measurement
Serial additions of solvent to saturated suspension at 25ºC Solubility point taken after 30 minute hold period Equilibrium achieved very quickly – dissolution kinetics are fast Potential for slow, continuous solvent addition to generate continuous solubility
curve
0 2000 4000 6000 8000 10000 120000
0.5
1
1.5
2
2.5
3
0
0.1
0.2
0.3
0.4
0.5
0.6Water Concentration
Benzoic Acid Concentration
Time (s)
Wate
r C
once
ntr
ati
on (
g/g
)
Benzo
ic A
cid C
once
ntr
ati
on
(g/g
)
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.0
0.1
0.2
0.3
0.4
0.5
0.6
Water Concentration (g water/ g ethanol)
Solu
bilit
y (
g b
enzoic
acid
/ g
eth
anol)
Slight increase possibly due to temperature effect – heat of mixing
9
Overall Solubility Measurement
Good agreement between 3 methods Accuracy of generated solubility curve is validated Liquid Analysis systematically underestimated the solubility
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
ATR-FTIR Solubility
FBRM Solubiliy
Solid Analysis
Liquid Analysis
Water Concentration (g water / g ethanol)
g b
enzo
ic a
cid /
g e
thanol
10
UNIQUAC Model
UNIQUAC model follows observed solubility trends Predicts increase in solubility at low water concentrations
11
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
12
Chapter 3 – Solubility – Conclusions An accurate solubility curve has been generated and validated
Liquid analysis underestimated solubility
Increase in solubility observed at low water concentrations
‘Anti-solvent free’ solubility expression simplifies analysis
MSZW and supersaturation simple to understand and define
Analogous to cooling crystallization
ATR-FTIR probe is calibrated
Calibration model accuracy within 2%
Supersaturation can be monitored during subsequent experiments
UNIQUAC model generated
Theoretical model fits measured data adequately
13
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
14
Chapter 4 – MSZW & Mixing – Setup
Study the impact of key process
parameters on MSZW
Develop a model to account for differences
in the MSZW
Develop an agitation dependent
expression for the nucleation rate
Impact of Parameters on MSZW
16
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50.0
0.1
0.2
0.3
0.4
475 rpm - wall addition location
Addition Rate (g water/s )
MS
ZW
(g w
ate
r/g e
thanol)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5
-0.1
0.0
0.1
0.2
0.3
0.4
475 rpm - impeller addition loca-tion
Addition Rate (g water/s )
MS
ZW
(g w
ate
r/g e
thanol)
Impact of agitation, addition rate and addition location on MSZW Experiments were performed in triplicate – error bars are 95% confidence
interval Some negative MSZW were observed – nucleation prior to saturation point Changing the addition location has the most significant impact
17
Chapter 4 – Mixing and MSZW – Interim Conclusions
Wall addition location results in significant variability
Variability worse at low agitation and fast addition rate
Negative MSZWs observed at low agitation
Typical widening of MSZW at high supersaturation generation rates not
observed
Impeller addition location results in little or no variability
MSZW widens with increasing supersaturation generation rate
Reducing agitation intensity results in a wider MSZW
Rate and degree of anti-solvent incorporation needs to be studied
Inadequate anti-solvent incorporation may result in locally high
supersaturation
A CFD model may help understand the differences between addition locations
18
CFD Model
CFD models show velocity in z-direction (up-down) at liquid surface Close to the wall the velocity is upward – hindering anti-solvent incorporation Close to the impeller the velocity is downward – facilitating anti-solvent incorporation These models provide a model for the observed results Nucleation is hydrodynamically limited when anti-solvent is added at the wall
325rpm
475rpm
impeller
impeller
wall wallimpeller
impeller
wall wall
19
Nucleation kinetics
Classical nucleation kinetics are modified for an anti-solvent system Only experimental data gathered at the impeller location is considered A non-linear regression is used to estimate kinetic parameters – including an
agitation parameter
-2.4 -2.2 -2 -1.8 -1.6 -1.4 -1.2 -1-9.0
-8.0
-7.0
-6.0
-5.0
-4.0
475rpm325rpm
ln (ΔAmax)
ln (
R)
kn = 1.9x10-3
n = 2.5 = 1.1
3.
4.
5.
xy
6.
20
Predicting MSZW
Good agreement between measured MSZW and predicted values Clearly impeller speed plays a critical role in nucleation kinetics This kinetic model allows MSZWs to be predicted under certain
hydrodynamic conditions
0 0.0010.0020.0030.0040.0050.0060.0070.0080.0
0.1
0.2
0.3
0.4
475 rpm
Addition Rate (g g-1 s-1)
Meta
sta
ble
Zone W
idth
(g g
-1)
kn = 1.9x10-3
n = 2.5 = 1.1
7.
8.
21
Chapter 4 – Mixing and MSZW - Conclusions
Agitation, addition location, and addition rate all impact the MSZW
A minor change in the feed location resulted in significant variability in the MSZW
To ensure process robustness a slow addition rate, high agitation and optimal feed
location should be chosen
The CFD model indicates that anti-solvent incorporation is the limiting step
Flow patterns in the vessel either facilitate or hinder anti-solvent incorporation
Possible to model larger vessels and locate optimal feed location upon scale-up
An agitation dependent nucleation model was developed
Model predictions fitted the experimental data adequately
Potential to predict MSZW under different mixing conditions at different scales
22
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
23
Chapter 4 – Growth and Consistency - Goals
Identify conditions under which FBRM and
ATR-FTIR trends are repeatable
Understand crystallization mechanism with
FBRM and PVM
Combine ATR-FTIR supersaturation with
FBRM growth rate to elicit growth kinetics
Studying Process Consistency
24
ATR-FTIR measures the liquid phase concentration over time By combining concentration trends with know solubility values –
supersaturation may be monitored
0 600 1200 1800 2400 3000 36000.5
1.0
1.5
2.0
2.5
0.1
0.2
0.3
0.4
Water Run 1 Water Run 2Water Run 3 Benzoic Run 1
Time(s)
Wate
r C
once
ntr
ati
on (
g/g
)
Benzo
ic A
cid C
once
ntr
ati
on (
g/g
)
0 600 1200 1800 2400 3000 3600
-0.12
-0.09
-0.06
-0.03
0.00
0.03
0.06
run 1
run 2
run 3
Time (s)
Supers
atu
rati
on (
g/g
)
Studying Process Consistency
25
FBRM monitors the rate and degree of change to the particle count and dimension Crystal growth and nucleation is consistent across experiments FBRM trends indicate an initial primary nucleation event followed
by growth
0 600 1200 1800 2400 3000 36000
200
400
600
800
1000
1200
1400run 1
run 2
run 3
Time (s)
#/s
(1-1
0 m
icro
ns)
0 600 1200 1800 2400 3000 36000
100
200
300
400
500
600
run 1
run 2
run 3
Time (s)
#/s
(100-1
000 m
icro
ns)
Point of nucleation
Inflection point: primary nucleation complete
Nucleation zone
Growth zone
Understanding crystallization mechanism
PVM in situ images provide instant size and morphology information
Few fine crystals and little agglomeration after 30mins – growth dominates
Mechanism of primary nucleation followed by growth is somewhat validated
Needle width appears to be 20-80µm; Needle length appears to be 150- 500µm
Correlating FBRM and PVM
FBRM distributions taken after 30mins again highlight consistency Unweighted distribution may be a good track of crystal width –
mean~50µm
Square weighted distribution may be a good track of crystal length – mean~180µm
1 10 100 10000
1
2
3
4
5
6
7
8
run1 t = 30 mins
run2 t = 30 mins
Chord Length (microns)
#/s
square
weig
hte
d
1 10 100 1000 100000
10
20
30
40
50
60
70
80
90
100run1 t = 30 minsrun2 t = 30 minsrun3 t = 30 mins
Chord Length (microns)
#/s
Crystal width? Crystal length?
Estimating Growth Rate - FBRM
Crystal growth is tracked by trending the square weighted mean Considering the rate of change of this statistic a growth rate can be
calculated Growth is initially fast but slows over time as supersaturation is
consumed
1300 1800 2300 2800 3300 3800
-0.1
0.0
0.1
0.2
0.3run 1
run 2
run 3
Time (s)
Square
Weig
hte
d M
ean G
row
th
Rate
(m
icro
ns s
-1)
0 600 1200 1800 2400 3000 36000
50
100
150
200
250
300
350
400run 1
run 2
run 3
Time (s)
Square
Weig
hte
d M
ean C
hord
Length
(m
icro
ns)
Averaging over Three Runs
Supersaturation and growth rates are averaged over three runs Consistent smooth trends are generated These can be combined to generate growth rate kinetics
0 600 1200 1800 2400 3000 3600
-0.12
-0.10
-0.08
-0.06
-0.04
-0.02
0.00
0.02
0.04
Time (s)
Avera
ged S
upers
atu
rati
on (
g/g
)
0 600 1200 1800 2400 3000 3600
-0.1
-0.05
0
0.05
0.1
0.15
0.2
Time (s)
Avera
gaed G
row
th R
ate
(m
icro
ns
s-1
)
Overall Growth Rate Kinetics
Supersaturation is plotted against the FBRM growth rate to estimate kinetics The growth order of 1.1 is typical of organic systems
-0.005 0.000 0.005 0.010 0.015 0.020 0.025 0.0300
0.05
0.1
0.15
0.2
0.25
Experimental
Model
Supersaturation (g/g)
Gro
wth
Rate
MS
QW
( m
icro
ns s
-1) G = 7.8ΔC1.1
31
Chapter 5 – Monitoring and Growth - Conclusions
Suitable parameters were chosen to ensure a consistent process
ATR-FTIR showed consistency in desupersaturation
FBRM showed consistency in crystal nucleation and growth
Crystallization mechanisms were revealed using FBRM and validated by
PVM
FBRM shows an initial primary nucleation followed by slow crystal growth
PVM images after 30 mins confirm large well formed crystals with few fines
The unweighted and square weighted distributions track crystal width and length
respectively
In situ growth rate kinetics were estimated using ATR-FTIR and FBRM
Potential to implement feedback control using this method
Possible to include breakage, agglomeration, secondary nucleation terms in
model
32
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
33
Chapter 6 – Scale-up – Setup
1. Impeller Addition Location 2. Wall Addition Location 3. PVM 4. FBRM 5.
Overflow
Choose suitable process parameters for large
scale batches to mimic lab scale results
Conserve particle size
Maintain a short cycle time
Achieve similar yield
Identify optimal addition locations using CFD
model
Scale-up vessel is geometrically dissimilar
Also has baffles and a different impeller
Study process consistency at scale using in
situ tools
How do FBRM trends compare between batches?
How doe FBRM trends compare between scales?
34
CFD Model – 70L
CFD models show velocity in z-direction (up-down) at liquid surface The optimal feed location changes depending on the liquid volume At low volume the optimal feed location is close to the wall At high volume the optimal feed location is close the impeller
32 L 64 L
wall
impeller
impeller
wall
Averaging over Three Runs
Comparison of FBRM stats across batches Fines trends prove somewhat difficult to interpret at the start of each
run Probe coating was evident at the start of Runs 2 & 3 Inadequate mixing and significant segregation was evident during Run
1
0 500 1000 1500 2000 2500 3000 3500 40000
500
1000
1500
2000
2500
3000
Run 1
Run 2
Run 3
Time (s)
#/s
(1-1
0µ
m)
Probe coating
Averaging over Three Runs
Comparison of stats across batches The coarse counts and mean square weight are both similar Growth kinetics are similar across all batches
0 500 10001500200025003000350040000
100
200
300
400
500
600
Run 1
Run 2
Run 3
Time (s)
#/s
(100-1
000 µ
m)
0 500 10001500200025003000350040000
50
100
150
200
250
300
350
Run 1
Time (s)
Mean S
quare
Weig
ht
(µm
)
Averaging over Three Runs
FBRM endpoint comparison On the unweighted and square weighted distribution – endpoints
are comparable Run 1 appears to be slightly different – possibly due reduced
agitation intensity
1 10 100 10000
10
20
30
40
50
60
70
80
90
100
Run 1
Run 2
Run 3
Chord Length (µm)
#/s
1 10 100 10000
2
4
6
8
10
12
14
16
18Run 1
Run 2
Run 3
Chord Length (µm)
#/s
Averaging over Three Runs
FBRM endpoint comparison FBRM trends indicate that the growth and nucleation kinetics are
similar However nucleation occurs earlier at the 70L scale
0 500 1000 1500 2000 2500 3000 3500 40000
100
200
300
400
500
600
Pilot-ScaleLab-Scale
Time (s)
#/s
(100-1
000)
0 500 1000 1500 2000 2500 3000 3500 40000
200
400
600
800
1000
1200
Pilot-ScaleLab-Scale
Time (s)
#/s
(1-1
0)
Averaging over Three Runs
FBRM endpoint comparison On the unweighted and square weighted distribution – endpoints
are comparable Run 1 appears to be slightly different – possibly due reduced
agitation intensity
1 10 100 10000
5
10
15
20
25
30
500mL
70 L
Chord Length µm
#/s
(square
weig
hte
d)
1 10 100 10000
0.5
1
1.5
2
2.5
3
500mL
70 L
Chord Lenth µm
#/s
(unw
eig
hte
d)
Averaging over Three Runs
PVM endpoint comparison PVM images at both scales are comparable
70L 500 mL
41
Introduction 1
Chapter 1 – Introduction
Chapter 2 – Literature Review
Chapter 3 – Solubility
Chapter 4 – Mixing and MSZW
Chapter 5 – Monitoring and Growth
Chapter 6 – Scale-up
CONCLUSIONS
42
Conclusion
Chapter 3 – Solubility Measurement for an Anti-
Solvent System Using Gravimetric Analysis, ATR-FTIR
and FBRM
Simple Solubility ExpressionNovel FBRM Solubility MethodCalibration of ATR-FTIR probe
Solubility Measurement using ATR-FTIR
Gravimetric Solubility MeasurementUNIQAC Solubility Model
Chapter 4 – The Effect of Mixing on the Metastable Zone
Width in Anti-solvent Crystallization
MSZW using FBRM and ATR-FTIRImpact of Process Conditions on
MSZWMSZW Robustness and
RepeatabilityCFD Model for Process
UnderstandingModified Nucleation Kinetics
Solubility Measurement for an Anti-solvent Crystallization System Using Gravimetric Analysis, ATR-FTIR and FBRM, Crystal Growth and Design, in press
The Effect of Mixing on the Metastable Zone Width and Nucleation Kinetics In the Anti-solvent Crystallization of Benzoic Acid, Chemical Engineering Research and Design, Transactions IChemE part A, 85 (7) 945-953, 2007
43
Overview
Chapter 5 – The Use of FBRM and ATR-FTIR to Monitor Anti-
solvent Crystallization and Estimate Growth Kinetics
Process Repeatability StudyChoosing Optimal Process
ParametersIn situ Monitoring of
SupersaturationIn Situ Monitoring of Growth Rate
In Situ Estimation of Growth Kinetics
Chapter 6 – Scale-Up of Anti-solvent Crystallization
Choose Optimal Process ConditionsCFD Modeling to Reduce
ExperimentsOptimization based on Initial
ResultsProcess Similarity Achieved
Between ScalesTo Be Submitted – Journal of Crystal Growth
To Be Submitted – Chemical Engineering Science