Designing for Compaction in Tablets with Spray Dried ... AWS
Understanding spray dried intermediates and their compaction in tablets using PAT, modeling and simulations November 2017 Marcio Temtem1, Katie Kleissas Haynes2, Ligia Bras3, João Henriques1, Kristin Ploeger4 Hovione – Drug Product Development R&D, Loures, Portugal Merck – MMD Process Analytical Technology, West Point, PA, USA 3 Hovione – Operational Excellence, Loures, Portugal 4 Merck – Pharmaceutical Commercialization Technology, West Point, PA, USA 1 2
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 2
Spray Drying Principles F_feed (P_feed / F_atom)
T_in, F_drying HX1
F1
Spray Dryer
Feed tank
F2 HX2
T_out Cyclone
Bag filter
HP-P
•
Flash drying
•
Mild process
•
Control of SDD attributes
Condenser
T_cond
(particle engineering) •
Scalable
•
Commercially demonstrated
•
Solvent based process
Solvent Spray Dried Dispersion
Fines
3
Spray Drying Principles 1. Thermodynamics
2
3
• Spray drying conditions determined through heat / mass balance, L-V equilibrium equations • Detect abnormal conditions of operation
2. Atomization • Droplet size estimation as a tool for scale-up • Achieve target particle size distribution
3. Drying Kinetics • Effect on powder properties: e.g. morphology and bulk density.
Fundamental Approach
1, 4
4. Computational Fluid Dynamics • Optimize process performance • Identify pitfalls
4
Spray Drying Principles •
Droplet and particle size are intrinsically connected, however particles can follow different formation pathways depending on drying kinetics and product properties
Fast drying
Twb
Tdrying gas
High T_out / Low RS_out / High HMT
heat
Slow drying Low T_out / High RS_out / Low HMT
dD
drying gas
mass RSdroplet
Inflated Particles
RSdrying gas
Fast Drying
Properties
Slow Drying
Spherical
Morphology
Shriveled
Low
Bulk density
High
Low
Residual solvent
High
Inflated (breakage may occur)
Particle size
Shrunken Particles
Decrease 5
Designing a Robust Spray Drying Process
Time in Development
Process Definition
Initial Process Scale-up • Thermodynamic tools inform space for process investigation
Process Development • DOEs and experimental data to inform process response
Process Commercialization • MVDA (and in-line PAT) monitors process robustness of a defined process over time
Thermodynamic Process Space Spray drying
feed flow rate
outlet temperature
Thermodynamic Space
inlet temperature
7
Thermodynamic Process Space Spray drying High T_in High RS_out
High F_feed
(high solvent content in the powder)
(equipment limitations)
(equipment limitations)
High T_out
feed flow rate
outlet temperature
(impact on powder properties)
Low RS_out (impact to powder properties)
Low F_feed (process is inefficient)
inlet temperature
8
Case Study Spray drying Main goals • Transfer the process from the PSD-2 (pilot scale) to the PSD-4 (commercial scale) • Identify adequate nozzle and spray drying conditions to produce SDD intermediate with pre-
defined properties
20
80
70
From the PSD2…
60
• RS_out (scale dependent) vs.
10
50
Tg, ºC
RS_out acetone, %
15
acetone content • Tg (scale independent) vs.
5 40
acetone content 0
30 1
2
3
4
5
acetone content, %
9
Case Study Spray drying BSV vs. RS_out across scales • Good predictive capacity of bulk specific volume (BSV = 1 / bulk density) • There is an offset between spray dryer scales
7 PSD2 scale
BSV_dry, mL/g
6
PSD4 scale
5 4 3 2 1 5
10
15
20
25
RS_out acetone, %
10
Case Study Spray drying BSV vs. HMT across scales • Confirm that Heat to Mass Transfer ratio (HMT) is a good predictor of bulk specific volume (BSV) across scales o Removes some of the offset from scale-densification 6
BSV_dry, mL/g
5 high HMT (PSD2 scale)
4
3 PSD2 scale PSD4 scale
2 0.2 low HMT (PSD2 scale)
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
HMT_acetone, -
11
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 12
Compactability as a Property • Compactability: the formation of material strength in response to an applied force • Powder characterization: – Mimic compaction forces & strain rate on a Huxley Bertram compaction simulator – Fit data empirically; Gompertz provides a good fit w/ 3 parameters – Interested in the slope of the curve (a*k), especially at 20~200 MPa (RC & tableting pressures)
Tensile Strength (MPa)
a
HQ00011.30 Sample A HQ00011.50 Sample B
5
4
slope: a*k
3
2
Gompertz Eqn:
a/e
SDI
A
B
a
4.8
3.4
xc
83
93
k
0.014
0.013
R2adj
0.988
0.996
1
0 0
xc 100
200
300
Compaction Pressure (MPa)
400
13
Tableting Process Development • 3 Unit operations impact material microstructure Process Parameters: HMT
Process Parameters: Roll Pressure
Process Parameters: Compression Force
(+ droplet size)
(+ gap, roll speed)
(+ press speed, precompression, feeder speed)
Spray Drying
Roller Compaction
SDD Attributes: BSV, PSD, compactability, porosity, flow
20 um
GRN Attributes: PSD, density, tensile strength, compactability
100 um
Tabletting
TAB Attributes: disintegration, dissolution, friability, hardness
100 um
14
Tableting Process Development • 3 Unit operations impact material microstructure Process Parameters: HMT
Process Parameters: Roll Pressure
Process Parameters: Compression Force
Spray Drying
Roller Compaction
Tabletting
• Study primary compactability factor in each step strength • Evaluate in a multifactorial DoE
• DoE goal: • Determine how much stress path affects final microstructure (porosity & strength) • Compaction = cohesion, rearrangement, fracture, adhesion, deformation 15
Designing SDI for Compactability • Bulk density vs. compactability – Both are governed by the same process parameters • Tout is the dominant effect – Tcond, gas flow, and liquid feed flow are not significant within a scale
• HMT provides a good prediction of compressibility within a scale Standardized Effects: bulk density Pareto Chart of the Standardized Effects (response is BDdry, Alpha = 0.05)
2.78
Tout Fgas Tcond Ffeed
T_out
Term
F_gas
T_cond
F_feed
0
2
4
6 8 10 Standardized Effect
12
14
16
Pareto Chart of the Standardized Effects
Standardized Effects: a*k (response is a*k, Alpha = 0.05)
2.776
Tout Tcond Ffeed Fgas
T_out
Term
T_cond
F_feed
F_gas
0
1
2 3 Standardized Effect
4
16
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 17
Process Commercialization • Primary goal of a commercial process: ROBUST! – Intra-batch stability – Inter-batch stability
• Numerous tools, including: – PAT monitoring – Process analytics
18
Process robustness PAT applications Solution preparation
Spray drying
Post drying
Feed tank Process Laser Diffraction
FBRM
Process Mass Spectroscopy Exhaust gas analysis
Spray Dryer Real-time particle size distribution
Turbidimetry
Cyclone Process NIR Drying end-point solid state characterization
Viscosimetry
30 25
Process NIR
20
Post dryer
15 10
Process NIR
5 0 0
10
20
30
40
50
60
70
Real-time physical/ chemical information
Process robustness Process laser diffraction
• Real-time, in-line particle size measurements • Four measurements per second, averaged to one minute • Process robustness and stability can be directly evaluated Spray-drier chamber
Cyclone
Product specific sampling system
Malvern Insitec X 20
Process robustness Process laser diffraction Spray-drier chamber
Stabilization
Stationary phase
Transmission
Dv(90) Dv(50)
Cyclone
11.3 50
7.5 3.8
0 1
10
100
Particle Diameter (µm)
0.00 1000
100
15.0 11.3
50
7.5 3.8
0 1
10
100
Volume Frequency (%)
15.0
Cumulative Volume (%)
100
Volume Frequency (%)
Product specific sampling system
Cumulative Volume (%)
Particle size profile
0.00 1000
Particle Diameter (µm)
Process laser diffraction provides real-time particle size distribution dynamics 21
Process Development vs. Commercialization Process laser diffraction Spray-drier chamber
DoE runs
Change in conditions
Change in conditions
Transmission Stationary phase
Stationary phase
Stationary phase
Dv90 Dv50 Dv10
Cyclone
Feeding stopped
Feeding started Stationary phases
Product specific sampling system
On-line average particle size distribution of spraydrying DoE runs targeting different particle sizes
Process laser diffraction provides real-time particle size distribution dynamics 22
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 23
MVDA Background and Benefits F_feed (P_feed / F_atom)
T_in, F_drying HX1
• Powerful, yet simple, fault detection in real time – Benefit: Can detect and possibly correct process faults before the fault causes the batch to be discarded
F1
Spray Dryer
Feed tank
F2 HX2
T_out Cyclone
Bag filter
HP-P Condenser
T_cond
Solvent Spray Dried Dispersion
Fines
• Batch-level models offer another tool in which to summarize large amounts of data into easily-understood plots – Benefit: Good tool for a retrospective analysis of a process and process fingerprinting
• Multiple variables and their interactions are continuously evaluated – Benefit: Accurate diagnosis and simplify process visualization
Process Monitoring • Historical batches were used to create a multivariate model during a Spray Drying process, allowing for an exploration into the variables and batches • This plot shows the scores run chart with the typical trajectory of a steady-state process
MVDA Evaluation of the Process on New Equipment • Goal: To evaluate the impact on the process due to a change – Here, a transfer to another “equivalent” spray dryer
• Analysis: Predicted on the batch using the model built from historical batches on the original spray dryer • Conclusion: The process on the new dryer is well within the normal variation of the historical process Score Plot for the Second Spray Dryer Batch
new process
historic expected variation
historic average
Predictive Capability of Process Variables for Determining Material Properties • Goal: To determine if process variables can predict downstream properties • Analysis: Developed an OPLS model to correlate the average process variable values to key material properties
• Conclusion: The best correlations were to tap and bulk density, consistent with prior knowledge
Conclusions • Scale-up – First-principle models of the spray drying process can constrain the process space which should be explored during development – Scale-independent correlations can be used to design spray dried powder properties and reduce number of trials for scale-up
• Process space – Process development of tablets containing spray dried intermediate must be aware of interdependency of unit ops – Compactability is another material attribute to consider in development of a spray dried powder
• Process robustness: PAT & MVDA – Process robustness can be monitored by several different approaches – PAT tools provide real-time information of material attributes and process stability – MVDA allows simple visualization of multiple variables for early-fault diagnosis and batch-to-batch comparison 28
Recognition Special thanks to the many people who have supported this work: • Dave Lavrich
• Paula Cordeiro
• Jerry Klinzing
• Susana Campos
• Chris Neu
• Luís Rato
• Aditya Tatavarti • Melissa Curley
and many others…
• Bend Research
29
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 2
Spray Drying Principles F_feed (P_feed / F_atom)
T_in, F_drying HX1
F1
Spray Dryer
Feed tank
F2 HX2
T_out Cyclone
Bag filter
HP-P
•
Flash drying
•
Mild process
•
Control of SDD attributes
Condenser
T_cond
(particle engineering) •
Scalable
•
Commercially demonstrated
•
Solvent based process
Solvent Spray Dried Dispersion
Fines
3
Spray Drying Principles 1. Thermodynamics
2
3
• Spray drying conditions determined through heat / mass balance, L-V equilibrium equations • Detect abnormal conditions of operation
2. Atomization • Droplet size estimation as a tool for scale-up • Achieve target particle size distribution
3. Drying Kinetics • Effect on powder properties: e.g. morphology and bulk density.
Fundamental Approach
1, 4
4. Computational Fluid Dynamics • Optimize process performance • Identify pitfalls
4
Spray Drying Principles •
Droplet and particle size are intrinsically connected, however particles can follow different formation pathways depending on drying kinetics and product properties
Fast drying
Twb
Tdrying gas
High T_out / Low RS_out / High HMT
heat
Slow drying Low T_out / High RS_out / Low HMT
dD
drying gas
mass RSdroplet
Inflated Particles
RSdrying gas
Fast Drying
Properties
Slow Drying
Spherical
Morphology
Shriveled
Low
Bulk density
High
Low
Residual solvent
High
Inflated (breakage may occur)
Particle size
Shrunken Particles
Decrease 5
Designing a Robust Spray Drying Process
Time in Development
Process Definition
Initial Process Scale-up • Thermodynamic tools inform space for process investigation
Process Development • DOEs and experimental data to inform process response
Process Commercialization • MVDA (and in-line PAT) monitors process robustness of a defined process over time
Thermodynamic Process Space Spray drying
feed flow rate
outlet temperature
Thermodynamic Space
inlet temperature
7
Thermodynamic Process Space Spray drying High T_in High RS_out
High F_feed
(high solvent content in the powder)
(equipment limitations)
(equipment limitations)
High T_out
feed flow rate
outlet temperature
(impact on powder properties)
Low RS_out (impact to powder properties)
Low F_feed (process is inefficient)
inlet temperature
8
Case Study Spray drying Main goals • Transfer the process from the PSD-2 (pilot scale) to the PSD-4 (commercial scale) • Identify adequate nozzle and spray drying conditions to produce SDD intermediate with pre-
defined properties
20
80
70
From the PSD2…
60
• RS_out (scale dependent) vs.
10
50
Tg, ºC
RS_out acetone, %
15
acetone content • Tg (scale independent) vs.
5 40
acetone content 0
30 1
2
3
4
5
acetone content, %
9
Case Study Spray drying BSV vs. RS_out across scales • Good predictive capacity of bulk specific volume (BSV = 1 / bulk density) • There is an offset between spray dryer scales
7 PSD2 scale
BSV_dry, mL/g
6
PSD4 scale
5 4 3 2 1 5
10
15
20
25
RS_out acetone, %
10
Case Study Spray drying BSV vs. HMT across scales • Confirm that Heat to Mass Transfer ratio (HMT) is a good predictor of bulk specific volume (BSV) across scales o Removes some of the offset from scale-densification 6
BSV_dry, mL/g
5 high HMT (PSD2 scale)
4
3 PSD2 scale PSD4 scale
2 0.2 low HMT (PSD2 scale)
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
HMT_acetone, -
11
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 12
Compactability as a Property • Compactability: the formation of material strength in response to an applied force • Powder characterization: – Mimic compaction forces & strain rate on a Huxley Bertram compaction simulator – Fit data empirically; Gompertz provides a good fit w/ 3 parameters – Interested in the slope of the curve (a*k), especially at 20~200 MPa (RC & tableting pressures)
Tensile Strength (MPa)
a
HQ00011.30 Sample A HQ00011.50 Sample B
5
4
slope: a*k
3
2
Gompertz Eqn:
a/e
SDI
A
B
a
4.8
3.4
xc
83
93
k
0.014
0.013
R2adj
0.988
0.996
1
0 0
xc 100
200
300
Compaction Pressure (MPa)
400
13
Tableting Process Development • 3 Unit operations impact material microstructure Process Parameters: HMT
Process Parameters: Roll Pressure
Process Parameters: Compression Force
(+ droplet size)
(+ gap, roll speed)
(+ press speed, precompression, feeder speed)
Spray Drying
Roller Compaction
SDD Attributes: BSV, PSD, compactability, porosity, flow
20 um
GRN Attributes: PSD, density, tensile strength, compactability
100 um
Tabletting
TAB Attributes: disintegration, dissolution, friability, hardness
100 um
14
Tableting Process Development • 3 Unit operations impact material microstructure Process Parameters: HMT
Process Parameters: Roll Pressure
Process Parameters: Compression Force
Spray Drying
Roller Compaction
Tabletting
• Study primary compactability factor in each step strength • Evaluate in a multifactorial DoE
• DoE goal: • Determine how much stress path affects final microstructure (porosity & strength) • Compaction = cohesion, rearrangement, fracture, adhesion, deformation 15
Designing SDI for Compactability • Bulk density vs. compactability – Both are governed by the same process parameters • Tout is the dominant effect – Tcond, gas flow, and liquid feed flow are not significant within a scale
• HMT provides a good prediction of compressibility within a scale Standardized Effects: bulk density Pareto Chart of the Standardized Effects (response is BDdry, Alpha = 0.05)
2.78
Tout Fgas Tcond Ffeed
T_out
Term
F_gas
T_cond
F_feed
0
2
4
6 8 10 Standardized Effect
12
14
16
Pareto Chart of the Standardized Effects
Standardized Effects: a*k (response is a*k, Alpha = 0.05)
2.776
Tout Tcond Ffeed Fgas
T_out
Term
T_cond
F_feed
F_gas
0
1
2 3 Standardized Effect
4
16
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 17
Process Commercialization • Primary goal of a commercial process: ROBUST! – Intra-batch stability – Inter-batch stability
• Numerous tools, including: – PAT monitoring – Process analytics
18
Process robustness PAT applications Solution preparation
Spray drying
Post drying
Feed tank Process Laser Diffraction
FBRM
Process Mass Spectroscopy Exhaust gas analysis
Spray Dryer Real-time particle size distribution
Turbidimetry
Cyclone Process NIR Drying end-point solid state characterization
Viscosimetry
30 25
Process NIR
20
Post dryer
15 10
Process NIR
5 0 0
10
20
30
40
50
60
70
Real-time physical/ chemical information
Process robustness Process laser diffraction
• Real-time, in-line particle size measurements • Four measurements per second, averaged to one minute • Process robustness and stability can be directly evaluated Spray-drier chamber
Cyclone
Product specific sampling system
Malvern Insitec X 20
Process robustness Process laser diffraction Spray-drier chamber
Stabilization
Stationary phase
Transmission
Dv(90) Dv(50)
Cyclone
11.3 50
7.5 3.8
0 1
10
100
Particle Diameter (µm)
0.00 1000
100
15.0 11.3
50
7.5 3.8
0 1
10
100
Volume Frequency (%)
15.0
Cumulative Volume (%)
100
Volume Frequency (%)
Product specific sampling system
Cumulative Volume (%)
Particle size profile
0.00 1000
Particle Diameter (µm)
Process laser diffraction provides real-time particle size distribution dynamics 21
Process Development vs. Commercialization Process laser diffraction Spray-drier chamber
DoE runs
Change in conditions
Change in conditions
Transmission Stationary phase
Stationary phase
Stationary phase
Dv90 Dv50 Dv10
Cyclone
Feeding stopped
Feeding started Stationary phases
Product specific sampling system
On-line average particle size distribution of spraydrying DoE runs targeting different particle sizes
Process laser diffraction provides real-time particle size distribution dynamics 22
Agenda 1. Spray drying principles a. Initial process scoping: Thermodynamic models b. Scale-independent correlations
2. Compaction simulation a. Downstream impact (compaction simulation) b. Experimental design (DoEs) 3. PAT monitoring 4. Chemometrics / MVDA a. Process monitoring b. Process Understanding 23
MVDA Background and Benefits F_feed (P_feed / F_atom)
T_in, F_drying HX1
• Powerful, yet simple, fault detection in real time – Benefit: Can detect and possibly correct process faults before the fault causes the batch to be discarded
F1
Spray Dryer
Feed tank
F2 HX2
T_out Cyclone
Bag filter
HP-P Condenser
T_cond
Solvent Spray Dried Dispersion
Fines
• Batch-level models offer another tool in which to summarize large amounts of data into easily-understood plots – Benefit: Good tool for a retrospective analysis of a process and process fingerprinting
• Multiple variables and their interactions are continuously evaluated – Benefit: Accurate diagnosis and simplify process visualization
Process Monitoring • Historical batches were used to create a multivariate model during a Spray Drying process, allowing for an exploration into the variables and batches • This plot shows the scores run chart with the typical trajectory of a steady-state process
MVDA Evaluation of the Process on New Equipment • Goal: To evaluate the impact on the process due to a change – Here, a transfer to another “equivalent” spray dryer
• Analysis: Predicted on the batch using the model built from historical batches on the original spray dryer • Conclusion: The process on the new dryer is well within the normal variation of the historical process Score Plot for the Second Spray Dryer Batch
new process
historic expected variation
historic average
Predictive Capability of Process Variables for Determining Material Properties • Goal: To determine if process variables can predict downstream properties • Analysis: Developed an OPLS model to correlate the average process variable values to key material properties
• Conclusion: The best correlations were to tap and bulk density, consistent with prior knowledge
Conclusions • Scale-up – First-principle models of the spray drying process can constrain the process space which should be explored during development – Scale-independent correlations can be used to design spray dried powder properties and reduce number of trials for scale-up
• Process space – Process development of tablets containing spray dried intermediate must be aware of interdependency of unit ops – Compactability is another material attribute to consider in development of a spray dried powder
• Process robustness: PAT & MVDA – Process robustness can be monitored by several different approaches – PAT tools provide real-time information of material attributes and process stability – MVDA allows simple visualization of multiple variables for early-fault diagnosis and batch-to-batch comparison 28
Recognition Special thanks to the many people who have supported this work: • Dave Lavrich
• Paula Cordeiro
• Jerry Klinzing
• Susana Campos
• Chris Neu
• Luís Rato
• Aditya Tatavarti • Melissa Curley
and many others…
• Bend Research
29
