❖

❖
❖

❖
Sr. No. Stage Critical Parameter Available PAT Tools

1 API manufacturing Identification of molecule and estimation of Real time Raman process
concentration spectrometer

Crystallization end point – particle size Focussed beam

reflectance measurement
(FRBRM),

Laser diffraction method

2 Powder blending Water content NIR tool

Blend uniformity

3 Fluid bed process Water content Microwave moisture

(fluid bed analyser,
granulation and
pellet coating) NIR moisture analyser
tools
Residual solvents Process mass
spectrometer (e.g.,
Promaxion)
Particle growth rate/end point Malvern’s Parsum/Eyecon
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

% Assay prediction using Inline NIR during Blending

120
100
%Assay NIR

80
60
40
20
0
-20
1 21 41 61 81 101 121 141 161 181 201 221 241 261 281

No of revolution
 Feasibility Risk Method Method Life cycle
assessment assessment development Validation management

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
Sr. No. Stage Critical Parameter Available PAT Tools

1 Power consumption or Motor power consumption Relatively less sensitive method,

ammeter readings because motor power is
influenced by multiple factors
such as age, wear and tear, etc.

2 Indirect torque (calculated Load on motor due to impeller load Relatively less sensitive method
based on motor ammeter since the torque is calculated
reading) from motor power consumption.
It is difficult to discriminate for
every small fluid addition or
kneading.

3 Focused Beam Reflectance Particle size of the granules through laser light beam End point based on particle size
Measurement (FBRM) distribution. It is suitable if
sensor dissolution is sensitive to particle
size (not much impacted by
density/porosity or wet mass
consistency).

4 Drag Force Flow (DFF sensor Drag force of granules on the sensor probe, End point is based on the wet
influenced by particle size, density, and wet mass mass consistency and particle
consistency of the blend size/density. Thus, it is more
sensitive and precise.

5 Impeller torque sensor Rotational force required to rotate the impeller, It directly reflects the load on the
(direct torque) influenced by the wet mass consistency impeller. Thus, it is more
sensitive, accurate and precise

6 Near Infra-Red (NIR) sensor Shift in baseline NIR spectra happens based on End point is based on base line
density and particle size of granule. Water band spectral shift overlapped with
gives information on free water/bound water with reference batch.
excipients available for granulation
Three different binder levels (blue,
green & red): dry mixing 3 min (-180
to 0 seconds) followed by fluid
addition 3 min (0-180 sec) &
kneading 180-360 sec)
Four similar formulations with
difference in kneading time, 0, 1, 3
and 5 min after fluid addition 180
seconds (starting from zero sec)
 Impeller Chopper

Torque
Sensor

Sensor location
 End point

(A) (B)
 Granulation Time duration Batch I Batch II
Stages (min)

Amp Torque (Nm) Torque/ Amp Amp

ratio

Dry mixing 5 14.62 63.74 4.36 14.37

Water 1.5 L 5 15.11 118.19 7.82 14.86

Kneading-1 2 15.36 152.14 9.91 15.42

Additional Water 1 15.67 175.09 11.17 15.68

1.5 L

Kneading-2 2 15.64 187.06 11.96 15.43

NIR spectra: Dry mixing (1st spectrum from the
bottom); water addition (2nd to 7th spectra from
bottom); kneading (top 3)
❖

❖
❖

❖
Company/Vendor Name of Model Type Wavelength range

Foss XDS Rapid Content AnalyzerTM Dispersive grating 400-2500nm

Bruker MPA FT 12800-3600cm-1

Buchi N-500 FT-NIR FT 800-2500nm

ThermoFisher AntarisTM II FT 833-2630nm

Perten DA 7250 NIR analyzer Diode array detector 950-1650nm

PerkinElmer Frontier NIR Reflectance FT 700-2500nm

System

ABB MB3600 FT 11,000-3,900cm-1 (909-2564)

ASD LabSpec 4 Bench Benchtop Silicon array detector 350-2500nm

Unity Scientific 2600 XT-R Predispersive scanning 680-2600nm

monochromator
❖

❖
❖

❖
 1. Sample Preparation of
different conc. 6. Addition of 3 final scale batch
samples into model. Includes the
variability

Hardness & Weight Variation

2. Spectral Acquisition
Diff. lot of raw materials etc

Tablets for CU Analysis 7. Model Optimization

Spectral File

3. HPLC Analysis of each

tablet (assay)
8. Model
CU value from HPLC
Internal
Validation
4. Model Preparation
Not
External Satisfied
Validation

5. Calibration performance Satisfactor

Not y Results
Satisfied
9. Final model release for use

If Needed
Satisfactory
Results
❖

❖
Examples of Baseline Offsets Pictorial Representation Cause Of Offsets

Vertical offset baseline Observed mainly due to instruments

effects or luminescence.

Slope offset Observed mainly due to instruments.

Curved slope offset

Observed usually due to fluorescence from
impurities or the analyte properties.
When fluorescence impacts, it may result
in to smaller peaks instead of tall sharper
peaks.
❖

❖
Pictorial representation of spectra Method applied Pictorial representation of spectra
before baseline correction after baseline correction

Detrending & Asymmetric least squares

smoothing (AsLS) method

1st order derivative method

2nd order derivative method

❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖

❖
❖
 Model Updating Methods Description

Slope and bias adjustments Simple post-processing method involving adjustments to the
slope and bias of predictions. Effective under limited
circumstances, such as fixed concentration changes in analytes or
alterations in optical properties. Ineffective for correcting new
variations in data.

Adding samples to existing calibration model Expansion of the calibration set by adding samples exhibiting new
variations. Suitable when a new analyte is introduced or a
previously fixed analyte starts to vary. Multiple samples may be
needed, and methods for up weighting new samples exist.

Instrument standardization/calibration transfer At times, there could be a need to transfer models between
procedures instrument of the same make or of a different make. Transfer
learning in such scenarios requires to measure samples in both
instruments and develop the estimates for the target instrument.
This utilizes methods like Direct Standardization (DS), Piecewise
Direct Standardization (PDS), Spectral Space Transformation (SST)
variants, and more. Methods may reduce net analyte signal but
preserve common features. Challenging in regulatory
environment.

Complete recalibration Considered a last resort, complete recalibration is performed

when the original model lacks relevant information. Typically
done when the model has essentially no useful data. Rarely
recommended due to its drastic nature.
 Fluid Bed Processes Typical Process Quality
Unit Operations Parameters Attributes

❖ % Weight gain
❖ Spray nozzle size
❖ Moisture content
Granulation ❖ Spray nozzle location
❖ Solvent content
(top spray) ❖ Inlet temperature
❖ Particle Size Distribution (PSD)
❖ Product/exhaust temperature
❖ Rheology

❖ Air flow ❖ Loss On Drying (LOD)

Drying ❖ Inlet temperature ❖ Moisture content

❖ Product temperature ❖ Solvent content

❖ Spray rate

❖ Atomization pressure

❖ Air flow ❖ % Weight gain

❖ Gun nozzle size ❖ Moisture content

Coating
❖ Inlet temperature ❖ Solvent content

❖ Product temperature ❖ PSD

❖ Column height

❖ Air distribution plate type

 Fluid Bed Processes Typical Process Quality
Unit Operations Parameters Attributes

Direct imaging particle size analyser

Direct imaging particle size analyser (Eyecon) (Eyecon)

Particle size Distribution Focussed Bed Reflectance Measurement (FBRM) Focussed Bed Reflectance Measurement
(FBRM)
Laser diffraction PS analyser (Insitec system, Parsum)
Laser diffraction PS analyser (Malvern)

❖ Moisture analyser
Moisture content
loss on drying ❖ Near Infra-Red (NIR) ❖ LOD moisture analyser
solvent content
❖ Mass spectrometer (MS)

Powder rheology
-- FT4
(bulk, flow & shear properties)
❖

❖
❖

❖
 Measurement type Particle size

Measurement range 0.1 to 2500µm

Measurement principle Laser diffraction

Optical models Mie theory

Accuracy ±2% on Dv(50) reported using the verification reticle

Power 100/240V

Enclosure rating IP65

Operating platforms 10 bar (g)

RTSizer (for instrument control) Malvern Link II (for

Software
system automation and data link)

Maximum distance from instrument to PC 500m (up to 2 km using fiber optics)

Temperature 10°C – 70°C

Humidity 35% - 80% (non-condensing)

❖

❖
 ❖

900
Methanol content by GC Analysis
800
Methanol content, ppm

700 ATPMS-PBO-NT-003
ATPMS-PBO-NT-009
600 ATPMS-PBO-NT-010
500
400
300
200
100
0
0 10 20 30 40 50 60 70
Time, min
 Methanol Content, PPM

Time, Min ATPMS-PBONT-011

0 517

10 289

20 229

30 231

40 165

50 164
References :

1. Control Engineering Europe, In-line particle sizer proves its PAT credentials (2013).
In-line particle sizer proves its PAT credentials (controlengeurope.com)

2. Jun Huang, Chimanlall Goolcharran, Julia Utz, Pedro Hernandez-Abad, Krishnendu Ghosh, Arwinder
Nagi. J Pharm Innov (2010) 5:58–68. A PAT Approach to Enhance Process Understanding of Fluid Bed
Granulation Using In-line Particle Size Characterization and Multivariate Analysis. Published online: 18
June 2010

3. Interphex 2010, 20 - 22 April, New York, US. Malvern exhibits proven PAT solutions for real-time
particle sizing (manufacturingchemist.com, 2010). Viewed on 4th March 2024

4. Tim Freeman. The FT4 Powder Rheometer - Measuring Bulk, Flow and Shear Properties of Powders -
Freeman Technology, Welland, Worcs, UK, [email protected] (2019). Viewed on 4th March 2024

5. Avila, C.R., Ferré, J., de Oliveira, R.R (2020). Process Monitoring of Moisture Content and Mass Transfer
Rate in a Fluidised Bed with a Low Cost Inline MEMS NIR Sensor. Pharm Res 37, 84.

6. European Pharmaceutical Review (2013). Real-time monitoring of the moisture content in different
type of dryers by NIR spectroscopy as a PAT tool.

7. Kim EJ, Kim JH, Kim MS, Jeong SH, Choi DH. Process Analytical Technology Tools for Monitoring
Pharmaceutical Unit Operations: A Control Strategy for Continuous Process Verification.
Pharmaceutics. 2021 Jun 21;13(6):919

8. Pro Maxion: Supplier Data Sheet. https://pdf.directindustry.com/pdf/ametek-process-

instruments/promaxion-mass-process-spectrometer/14269-577550.html. Viewed on 4th March 2024
 Laser Diffraction FBRM

❖ Utilizes diffraction patterns of a laser beam passed through ❖ Primarily used as a tool for development of crystallization
any object, ranging from nanometers to millimeters in size, to processes and real time analysis.
quickly measure geometrical dimensions of a particle.

❖ A probe-based instrument is inserted directly into processes

❖ Uses Mie theory of light scattering to calculate particle size to track changing particle size and count in real time at full
distribution, assuming a volume equivalent sphere model. process concentrations.

❖ Measures the chord length distribution, which in turn tracks

how particle size and count change from the beginning until
the end of a process.
❖

❖
❖

❖
❖
❖

❖
❖

❖
❖

❖
❖

❖
Samples preparation
of different DOE

Identification of
Data generation with required results
FBRM Method or end point -
Interpolation crystallization
of both data end point

Data generation with

conventional method

Model preparation
Dissatisfaction?
Change instrument
parameters & Re-
calibrate
Calibration
performance

Application of the model to

plant scale & validation
 Laser diffraction FBRM

Result: Result:
Dv (90): 210.669 μm Dv (90): 208.38 μm

❖
Seed charged
❖