control of size and shape distribution in continuous crystallisation
TRANSCRIPT
1
McMaster University
Control of Size and Shape Distribution in Continuous
Crystallisation Systems
Zoltan K. NAGY1,2, Aniruddha MAJUMDER2, Ali SALEEMI2
1School of Chemical Engineering , Purdue University,
West Lafayette, IN, USA2Chemical Engineering Department, Loughborough University, UK
September 13, 2012
EPSRC Centre in Continuous Manufacturing and Crystallisation Open Day
University of Strathclyde, Glasgow
2
Acknowledgments
Dr. Aniruddha Majumder
Dr. Ali Saleemi
European Research Council Grant No. 280106-CrySys
CrySys
2
3
Introduction
Composite-PAT array (CPA) and CryPRINS
Continuous real-time monitoring of crystallization processes
CSD control via tailored dissolution in batch crystallisation (model-based and model-free)
Model-based control of CSD in continuous crystallizers
Control of shape distribution using growth modifiers
Conclusions
Overview
4
Monitoring and Control of CrystallisationProcesses - Motivation
Many technology and economic drivers
70% of all solid products & 90% of APIs involve a crystallization step
Control of crystal properties important Efficient downstream operations (filtration, drying)
Product effectiveness (tablet stability, bio-availability)
Strict regulatory requirements related to variation of quality
High economic penalty of producing off-spec product (£1-2 million/batch)
Quality-by-design, fast scale up and product consistency
3
5
Continuous Crystallisation
Has been identified as a key paradigm shift with high potential of improving pharmaceutical production [1]
Continuous processing has the advantages:
– Consistency in product quality
– Reduction of cost by asset utilization
– Shorter down time
– Ease of scale up
– Achieve operating conditions unattainable in batch processes
Continuous processing is impossible without suitable control strategies
[1] Chen et al., Cryst. Growth Des., 2011, 11, 887-895
6
New generation of integrated, intensified & intelligent crystallization systems with drastically improved flexibility, predictability, stability & controllability.
New generation of integrated, intensified & intelligent crystallization systems with drastically improved flexibility, predictability, stability & controllability.
• Seed addition
• Cooling profile
• Antisolvent
• Growth/nucleation modifiers
Manipulated inputs:
Key role of monitoring and control in process intensification and integration
Plant-wide PRINS
Integrated, intensified & reconfigurable plant
Batch versus continuous manufacturingNagy&Braatz, Handbook of Ind. Cryst., 2012; Nagy&Braatz, Annu. Rev. Chem. Biomol. Eng., 2012
CrySys
4
7
Crystallization product engineering via real-time feedback control
. . .
Composite PAT array
...
Continuous real-time monitoring
Full characterisation of crystallisation state in the phase diagram using CPA
Complementarity & redundancy in measurements robust control
Automated/adaptive operation to design particles with tailored made properties
Model-based and model-free crystallization design & control approaches
Automated Intelligent Decision Support and
Control System
Real-time control
CryPRINS
RAMANFBRM
PVMATR-
UV/Vis, FTIR
8
Introduction
Composite-PAT array (CPA) and CryPRINS
Continuous real-time monitoring of crystallization processes
CSD control via tailored dissolution in batch crystallisation (model-based and model-free)
Model-based control of CSD in continuous crystallizers
Control of shape distribution using growth modifiers
Conclusions
Overview
5
9
60 80 100 120 140 1600
20
40
60
80
Temperature (C)
Hea
t F
low
(W
/g) Melting point
60 80 100 120 140 1600
20
40
60
80
Temperature (C)
Hea
t F
low
(W
/g) Melting point
Form 1 to form 2 transformation
60 80 100 120 140 1600
20
40
60
80
Temperature (C)
He
at F
low
(W
/g)
Form 1 to form 2 transformation
Melting point
FBRM for detection of OABA polymorphism
9
60 80 100 120 140 1600
20
40
60
80
Temperature (C)
Hea
t F
low
(W
/g) Melting point
Form 1 to form 2 transformation
60 80 100 120 140 1600
20
40
60
80
100
Temperature (C)
Hea
t F
low
(W
/g)
Form 1 to form 2 transformation
Melting point
20
30
40
50
60
Tem
per
atu
re (C
)
0.02
0.04
0.06
0.08
Ab
sorb
ance
0 200 400 6000
5000
10000
15000
Time (min)
To
tal C
ou
nts
/s
Total Counts/sTemperature (C)Absorbance
Nucleation
1
2
Polymorphictransformation
3
4
5
Form 1 (raw) Form 2 (nucleation) Form 2 + 1 (starts) Form 2 + 1 Form 1
Surface roughness due to polymorphic transformation
Solubility difference between polymorphs
1010
Image analysis (IA) for monitoring polymorphic transformation of OABA
0.5
0.55
0.6
Cir
cula
rity
0.58
0.6
0.62
0.64
0.66
0.68
0.7
Rec
tan
gu
lari
ty
1.7
1.8
1.9
2
2.1
Fer
et A
spec
t R
atio
0 100 200 3000
1000
2000
3000
4000
Time (min)
To
tal C
ou
nts
/s
Total Counts/sCircularityRectangularityFeret Aspect Ratio
Phase 1 Phase 2 Phase 3
a
b
(a/b
)
Decrease in aspect ratio
Increase in circularity
Increase in rectangularity agglomeration
Form 1 Form 2
Form 2 nucleates Mixture of forms 2+1 Form 1
6
11
200 400 600 800 1000 1200 1400 1600 18000
5
10x 104
Raman Shift (1/cm)
Inte
nsi
ty
200 400 600 800 1000 1200 1400 1600 18000
5
10x 107
Raman Shift (1/cm)
Inte
nsi
ty
200 400 600 800 1000 1200 1400 1600 18000
5
10x 107
Raman Shift (1/cm)
Inte
nsi
ty
Monitoring polymorphic transformation OABA using process Raman Spectroscopy
11
Form 1 and 2 monitored using fingerprint regions
Nucleation of form 2 and transformation into form 1 is monitored
25
30
35
40
45
50
55
60
Tem
per
atu
re (C
)
2.6
2.8
3
3.2
3.4
x 104
Arb
itra
ry U
nit
s
0 150 300 450 600 750
0
5
10
15x 10
4
Time (min)
Arb
itra
ry U
nit
s
Form 1
Temperature(C)
Form 2
Nucleation of form 2
Transformation of form 2 in form 1
12
Introduction
Composite-PAT array (CPA) and CryPRINS
Continuous real-time monitoring of crystallization processes
CSD control via tailored dissolution in batch crystallisation (model-based and model-free)
Model-based control of CSD in continuous crystallizers
Control of shape distribution using growth modifiers
Conclusions
Overview
7
13
In situ fine removal via controlled dissolution
15
20
25
30
35
40
Te
mp
era
ture
(C
)0 50 100
-4
-2
0
2
4
6
8x 10
-3
time (min)
Sup
ersa
tura
tion
(g
/g)
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 0 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
Seed CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 12 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 32 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 39 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 59 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 65 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 74 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 90 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
0 200 400 6000
0.005
0.01
0.015
0.02
0.025
t = 100 min
Particle size (m)
Vol
ume
(m
-1)
Target CSDSimulated CSD
Growth & nucleation
Dissolution
Optimal Temp ProfileOptimal Temp Profile
SupersaturationSupersaturation
0 200 400 600 8000
1
2
3
4
5
6
7
x 10-3
Particle Size (m)V
olum
e pd
f (m
-1)
CSD withoutDissolutionCSD withDissolution
15 20 25 30 35 400.04
0.05
0.06
0.07
0.08
0.09
0.1
Temperature (oC)
Co
ncen
tratio
n (w
t fra
ctio
n)
OptimisedTemperatureProfileSolubility Curve
Fines dissolutionFines dissolutionNucleationNucleation
“In situ dissolution loop”“In situ dissolution loop”
Nagy et al., CGD 2011; Nagy, CACE 2009
Uni-modal CSD achievable ONLY by using
controlled dissolution
Uni-modal CSD achievable ONLY by using
controlled dissolution
Optimal temperature includes heating stage
Use of controlled dissolution for programmed fine removal
Fine removal via control rather than equipment design
Suitable control strategy can eliminate design limitations
Model-free approach: DNC
14
Introduction
Composite-PAT array (CPA) and CryPRINS
Continuous real-time monitoring of crystallization processes
CSD control via tailored dissolution in batch crystallisation (model-based and model-free)
Model-based control of CSD in continuous crystallizers
Control of shape distribution using growth modifiers
Conclusions
Overview
8
15
Continuous plug flow crystallizer with controlled dissolution segments
T
Length
Spatially distributed operating profile (heating/cooling or antisolvent/solvent)
Alternating nucleation-growth-dissolution segments
16
PBM of the Continuous Plug Flow Crystallizer
Steady state model equations Population balance equation (PBE) is used
to model the crystallization process
Steady state PBE for growth and nucleation
; 0
PBE for dissolution
0; 0
Kinetics
1 ;
;
; Mass balance
3
16
High resolution technique for solving PBEs• Efficient, accurate and captures discontinuity
• Define the cell average in each cell
1Δ
;
• The PBE becomes a set of ODEs
1Δ
;
/ / /
• Flux reconstruction in cell boundary
12
• van Leer flux limiter to remove oscillation near discontinuity
1;
/
9
17
Optimization Problem: CSD shaping in continuous crystallisation
Temperature profile along the crystallizer with dissolution for fine removal
Objective function to be minimized:
min J , ,2 ;
Gradient based methods are often caught in local minima due to high nonlinearity of the problem
A stochastic technique genetic algorithm (GA) is used to solve the optimization problem
Allows also to optimize number of segments needed MINLP
Combined design and control of crystallisation processes
Predicted CSDTarget CSD
18
Spatially distributed control of product CSD with fines removal
0 50 100 150 200 250 3000
0.005
0.01
0.015
0.02
0.025
Size, L [m]
Volume Den
sity, f v
Seed
Target
Optimal T profile
Linear T profile
Linear temperature profile produces large amount of fines
The optimal profile alternates between growth and dissolution eliminate fines
Three growth and 2/3 dissolution clusters
Optimal profile also needed within clusters
‐0.04
‐0.03
‐0.02
‐0.01
0
0.01
0.02
0.03
S [g/g of solvent]
0 5 10 15 20 25 3010
15
20
25
30
35
40
45
No. of Segments
T [
o C]
OptimalLinearOptimal supersaturation
Dissolution
Growth
10
19
Spatial evolution of CSD
010
2030
0
100
200
3000
0.01
0.02
0.03
No. of SegmentsSize, L[m]Volume density, fv [#/m]
0
10
20
30
0
100
200
3000
0.01
0.02
0.03
No. of SegmentsSize, L[m]
Volume density, f v [#/m]
No dissolution(Linear cooling)
Linear temperature profile produces lots of fines
The optimal profile alternates between growth and dissolution to eliminate fines
Optimal temperature profile with controlled dissolution
20
Spatial operating profile in the phase diagram
10 15 20 25 30 35 40 450.04
0.05
0.06
0.07
0.08
0.09
0.1
0.11
T [o C]
C [g/g of solvent]
Solubility
Operating curve
Spatial optimal profile can be represented in the phased diagram
Optimal operating profile both in the supersaturated and under saturated region
Dissolution rate is also manipulated
For poorly soluble compounds large under-saturation needed
11
21
Introduction
Composite-PAT array (CPA) and CryPRINS
Continuous real-time monitoring of crystallization processes
CSD control via tailored dissolution in batch crystallisation (model-based and model-free)
Model-based control of CSD in continuous crystallizers
Control of shape distribution using growth modifiers
Conclusions
Overview
22
Effect of additives on nucleation and crystal shape of Paracetamol
Significant increase in aspect ratio
Influence on both CSD and shape
22
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 1 2 3 4 5
Ind
uct
ion
tim
e (
ho
urs
)
Amount of metacetamol (mol %)
0% Metacetamol (pure) 4% Metacetamol2% Metacetamol
Solubility
0% MACM
2% MACM
Induction time increases with additive concentration
12
23
Multi-dimensional PBM with the effect of impurities/additives (growth modifiers)
"Poisoning effect" of impurities growth reduction (unsteady-state adsorption)
KDP grown in the presence of Fe3+ (Terry et al. Nature, 1999)
Step velocity decreases with time leading to degradation of crystal surface
0( ) ( )(1 (1 exp( )))
1i
i
KC tG t G t
KCa
t= - -
+Use additives to achieve & control product
properties (shape, polymorph, etc.) Use additives to achieve & control product
properties (shape, polymorph, etc.)
80 s80 s
0 s 0 sNo GMNo GM With GMWith GM
24
Hybrid Continuous-batch crystallisation setup for shape distribution control using additives (growth modifiers)
PVM(or on-line imaging
system)
Thermocouple
Controller
Thermostat
Growth Modifier
Fresh Solution
Solution with additive
Control Mean Aspect Ratio (MAR)
2D CSD measured in-situ or on-line
Manipulate additive concentration
Flow through system with constant volume
Continuous operation for liquid phase (can dynamically manipulate GM concentration)
Batch operation for solidFilter
13
25
Dynamic GM feedback control of mean aspect ratio
Mean aspect ratio=2.1(No control)
0 5000 10000 150000
1020304050
GM
flo
wra
te [
ml/
s]
0 5000 10000 150001
1.5
2
2.5
Time (s)
Asp
ect
rati
o
0 5000 10000 150000
10
20
30
40
GM
Co
nc
[mg
/l]
No control
GM Control (P=-200)
Set pointMean aspect ratio=1.5
(With GM control)
Low GM conc. at end
Hybrid operation allows dynamic manipulation of GM concentration
Mean aspect ration significantly reduced (30%)
GM concentration high during middle of batch but very low at the end (no contamination of product)
26
Summary
Many economic drivers for better control of CSD and shape
Concept of composite sensor array (CSA) and CryPRINS allow continuous and real-time full characterisation of the crystallisation
Model based optimal control and design can significantly improve the pharmaceutical production process
Novel control approach for continuous crystallisation with spatially distributed controlled dissolution for fine removal
Shape distribution control using GM and hybrid continuous-batch crystallisation operation
Thank You