Blending End Point Monitoring By Near Infrared Spectrometer Biology Essay

Published:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

The objective of this project was to apply multivariate methods of chemometrics with the spectral data obtained from the Near Infrared Spectrometer to find out the end point of blending process. Here, α-Lactose monohydrate and Kaolin are used as the binary mixture and are blended using a Turbula blender. This operation is carried out for ten minutes for each batch and eleven batches are run overall. For every 40 seconds a sample is collected for each batch. By using the Near Infrared Spectrometer, with single solid reflectance aplication the samples are scanned. The obtained data is analysed using the NIRWare, NIRCal and Unscrambler Softwares by transforming to secondary derivatives and Principal Component Analysis and Cluster analysis were carried out to determine the blending end-point.

Another method Moving-Window-pooled Standard Deviation is also applied.

The obtained results prove that PCA and Cluster analysis as potent methods for monitoring the blending end point. While the Moving-Window-Pooled Standard deviation showed negative results.

INTRODUCTION

Blending process has a significant role in the process of manufacturing of pharmaceutical dosage forms. According to many authors, blend uniformity of a binary mixture is achieved when all the components in the mixture have an equal probability to obtain from any part of the blender. If it doesn't takes place it can be termed as non- uniform mixing. In most of the cases the mixing is because of gravity and not because of physical interaction and that leads to segregation of the active ingredients in pharmaceutical dosage forms. Many other factors lead to non-uniformity like surface charges, cohesiveness, spreading coefficients, etc. Generally these parameters are not routinely used for predicting the blending end point as they are not categorized as major factors affecting blending apart from particle size and flow properties, which are considered as major factors. Blending process mainly depends on physical properties of sample, the degree of mixing carried out and conditions at which the process takes place. The extensive work of Derjaguin explains the force acting between two spheres on basis of energy per unit area.

In this study, we mainly investigate the time (end point) at which uniform blend is obtained considering some factors affecting the blend homogeneity and by applying the chemometric methods.

Near Infra Red Spectroscopy (NIRS) is a potential device for evaluating the homogeneity of powder blends. In this particular context of laboratory analysis, I am taking a mixture of two components with an objective to determine end point of blending process. I evaluated 11 batches of the same samples in same proportions to check the robustness of the homogenous mixing of α-Lactose monohydrate and Kaolin.

Blending process is generally evaluated by many conventional methods; the mostly applied method among them is using a volumetric sample thief where the powder is withdrawn by inserting the probe. Application of this method is very effective so far because of its reproducibility. Other methods are all sample destructive and so non-invasively methods are preferred mostly. Also time taken for analysing the end point is more [1] . At this context there is a demand for fast and non-destructive method to determine blend homogeneity like NIR Spectrometry. Here off-line procedure is followed to explain the uniformity using NIRS and some statistical methods.

Chemometrics are appplied for the quantitative determination using method Moving-Window-pooled-Standard Deviation (MWPSD), Principal Component Analysis, Cluster Analysis, etc. Even though some authors claim that these techniques underestimate blending end point, all that claims are subsided by applying these chemometric methods [2] .

AIMS AND OBJECTIVES

Because of the effects of the blending process, it plays a key role in the preparation of pharmaceutical products.

To assess the appropriate time at which the uniform homogeneity is obtained.

To monitor the variations caused in the blending process and to check the presence of impurities by verifying the graphs obtained from the spectra.

To understand the application of statistical techniques and to define a new method for monitoring the blending ends point.

To check whether the results obtained are reliable and reproducible.

BACKGROUND

BLENDING PROCESS:

Blending is method in which two are more different samples are mixed using blenders or other mixing instruments. It is an essential step in the preparation of dosage formulations, as if demixing takes place this may result in reducing the lifetime of any formulation. It is the major factor considered in drug stability assessments. In this process after certain time, all the samples in a mixing container mix physically, without contaminating the sample. Hence, the evaluation of exact time taken for this process and achievement of a homogenous mixture has much significance. This project helps in showing the correct methods to monitor the blending end point using NIRS.

NEAR INFRARED SPECTROSCOPY:

Near Infrared Spectroscopy is a potential technique, which has many applications in Pharmaceutical, Cosmetic and Food processing sectors. It works on the principle of over toning and combination vibrations of molecules. When energy transitions take place in atoms they don't follow electric dipole method and emit a spectral line. The spectra attained by this overtone and combination vibrations are very complex and are with broad bands [3] .

NIR spectrometer was used for the first time in 1950s [4] , but it was used in combinations with other devices like UV, Vis, MIR spectrometers, etc. and was later used as a single unit. The beginning of light-fibre optics and monochromatic-detector made NIR a powerful tool for scientific research. NIRS is applied in the analysis of foodstuffs, pharmaceuticals, medical tool and a branch of astronomical spectroscopy.

The principle involves two vibrational laws, namely Hookes law and the Franck Condon principle.

HOOKE'S LAW:

á¹¼(in cm-1) = 1/2Ï€c

This law is represented by the following equation, for a simple two body harmonic oscillator:

Whereas,

á¹¼= Frequency of vibration

m1, m2 = Masses

c = Speed of light

k = Force constant

In mid IR, the fundamental vibrations of diatomic molecule can be calculated by using Hooke's law. As, NIR contains a combination of bands and overtones, these combinations of mid IR fundamental can be used to calculate NIR bands.

FRANCK-CONDON PRINCIPLE:

"When a molecule vibrates, the probability of finding a given atom at a certain period is inversely proportional to its velocity when it is at that point. When the nuclear configurations are same in both exited and ground state and when the nuclear kinetic energies are small, transitions takes place between vibrational levels. The vigorosity of a vibration is directly proportional to anharmonisity.

By taking water as an example, the principle fundamental vibrations,

It is thought that because of the asymmetrical stretch at 3756cm-1 and scissoring bend at 1596cm-1 combination band of water appears at 5150 cm-1 , that means calculated value 3756 + 1596 = 5352 cm-1.

Hydrogen bonding in bulk water is the main cause for larger difference between the calculated and derived. The estimated fundamental for bulk water was found at 3500 cm-1 and 1645 cm-1 and the calculated was 5145 cm-1 C-H stretch, N-H stretch, C=O and C-O-H can also be calculated as above . Then above 4 stretches contains the bulk of NIR active vibrated bands along with O-H stretch.

ADVANTAGES OF NIR OVER OTHER METHODS:

OVER WET-LABORATORY PROCEDURES:

It can handle several factors concurrently and very efficient.

It generates reports very rapid.

This method is very easy to handle without any complexity.

Operating expenses are very less and satisfactory.

Reactions are very safe and leave no chemical wastes.

OVER OTHER ANALYTICAL INSTRUMENTS:

NIRFlex N-500 covers a wide range of spectrum from 1000 to 2500 nm, where as other techniques can serve at specific wavelengths with small range.

The resolution provided is optimum to record the samples.

Operation is very quick and all frequencies are measured at same time and signals strike the detector simultaneously.

Signal- to-noise ratio is very efficient and reduces the signals caused by external atmosphere.

Near infrared rays can travel very deep into the sample than that of Mid Infrared rays.

NIRS is not a sensitive technique and can handle bulk materials with minute samples.

NIRS can be used to measure different varieties of samples.

CHEMO METRICS:

Chemometrics are the methods applied for sorting data from analytical instruments and uses statistical methods for developing the results. They deal with the correlated variables as well as un-related variables and then reduce the spectral data. Among these methods, multivariate statistical techniques are the mostly adopted methods for NIR spectrometric analysis. Here it is applied for the prediction of the samples and determine the end point when the blend uniformity of their mixture is acquired.

Chemometric techniques are majorly used with NIRS and also with other analytical instruments like NMR, IR and UV industrially and academically to extract the results from instrumental data.

PRINCIPAL COMPONENT ANALYSIS:

Principal component analysis (PCA) is a multivariate statistical technique which is valid only for correlated variables and data it functions by reducing the whole data. It finds the principal components that are linear combinations of the original variables describing the fluorescence intensities at the given wavelengths and then creates new variables. The new variables created by using the coefficients are not correlated and then two principal components with largest variation are selected. These principal components obtained from the measure of the joint variance of two variables are called as Covariance matrix. PCA is also applied in the fields of multiple regressions.

Principal Component Analysis is a mathematical method that reduces data from NIRS in order to reveal differences between samples and classifies them.

FIGURE: miller and miller

CLUSTER ANALYSIS:

Cluster analysis is a multivariate statistical method which involves searching for groups, dividing the groups into objects and assembling all the analogous objects in one new group forming a cluster. In the whole variable space, it searches and identifies closely related objects and categorizes them into different classes, but no assumptions are carried about the distribution of those variables. This method also decides whether to measure all the data on the same scale or on a different scale. Measuring on a same scale is called as standardization and it helps in preventing domination among different variables. Cluster analysis method involves forming new clusters with each object of equal size and comparison of distances between those clusters.

The steps involve formation of a new cluster by joining the two closest objects together. Again from these clusters the close clusters will join to form another new cluster comparing the distances between the two clusters. This step repeats and if there is more than one member in distance between two clusters then the distance between the neighbours is considered which is called as single linkage method.

FIGURE:

SECONDARY DERIVATIVE METHOD:

The secondary derivative methods are used to sort the spectra and remove unwanted data points. At same wavelength it has a minimum band (negative) and zero order has a maximum band. The main band has two satellite bands on its either sides. The number of bands can be known by adding one to its corresponding order.

SAVITZKY AND GOLAY DERIVATIVE METHOD:

It is a very competent and forms basis of the derivatization algorithm in many analytical instruments. N number of data points is collected and fixed with a polynomial to analyse derivative.

Whereas,

a0...al are coefficients at each wavelength and are multiplied by order.

This method gives a smooth data. When data points are more than the order, all data points are not covered by polynomial and generate a smooth estimate to the original data points. By using this property the deprivation of signal-to-noise ratio can be countered.

Features Of Savitzky And Golay Secondary Derivative Method:

Graphics

The increase in complexity by addition of bands, in the higher derivative spectra is useful for characterising and identifying the samples.

Resolution enhancement

The resolution is increased by abolishing the background effects.

Background elimination

Unwanted baseline shifts are totally eliminated, that are caused by mistake in sample handling.

Discrimination

This method helps in suppressing broad bands to narrow bands and increases amplitude with increase in derivative order.

Reduction in scattering

This also helps in reducing the scattering errors.

Matrix suppression

The absorbing background helps in spotting and categorizing tiny components.

Signal-to-noise ratio

The decrease in signal-to-noise ratio at higher orders of derivatives is countered by this method.

SOFTWARES USED:

UNSCRAMBLER SOFTWARE:

This is statistical software used for the data analysis of large data with multiple variables. Its applications are helpful in predicting the samples in analytical tests and used along with many instruments including near infrared spectroscopy. This software was first developed by Harald Martens and was later acquired by CAMO Software Inc. (New jersey, USA ).

Unscrambler is majorly used software because of its wide range of tasks like transformation, analysis and prediction of the unknown samples. Using unscramble we can perform several transformations like smoothing, normalizing and derivitization.

The samples can be analysed using this software by the methods Principal component analysis, cluster analysis, multiple linear regression, principal component regression and partial least squares method. The data obtained are reproducible and the software is very user friendly.

The methods applied in this experiment are Savitzky-Golay derivative method, Standard Normal Variate method, Principal Component Analysis and Cluster Analysis using the Unscambler software.

The spectral data obtained from the NIRS contains multiple variables and so, Multivariate Data Analysis is carried out to identify the samples. It is difficult to analyse this large data and hence reduced into small components representing the whole variables using the derivitization and smoothing techniques.

NIRWARE SOFTWARE SUITE:

NIRWare and NIRCal are the two software tools that manage wide range of applications and meet operational and calibration demands of NIR Spectrometer.

NIRWare software suite mainly helps in managing spectral information of samples, tailoring user interface and designing applications.

It comprises tools for import and export of spectral information, security tools and backup tools to recover the data whenever needed, apart from calibrations which are carried out by NIRCal 5 Chemometric Software. Its main applications are NIRWare Management Console and NIRWare Operator.

NIRWare Management Console:

NIRWare Management Console holds all snap-in modules and is configured for several user collections.it includes

Application Designer: This guides about handling procedure for analysis, defining instrument, application, calibrations and estimating properties.

Sample Management: This helps in checking the previous data, and monitors in managing samples and properties.

Administrative Tools: it includes

Audit Trail / System Logger: A part of FDA's 21 CFR Part 11 regulations.

Database Maintenance: Backup files.

Import / Export: Easy data exchange.

Security Designer: It helps in protecting files from unauthorized access. Account Policy helps in defining user names and passwords

Library Designer: It is useful in applying chemometrics and also spectral comparison with other files in library.

NIRWare Operator:

The operator helps in handling measurements and results are displayed as reports. It is user friendly and guides the users in tailoring interface and in applications.

NIRCAL 5 CHEMOMETRIC SOFTWARE:

NIR Cal software is efficient software for handling large spectra. This suite is well known for its speed, range of wavelength selection and accurate and reproducible resulting data. Its beneficial features are

Customization can be done to export and import data for the user interface and making it easily accessible.

Data Exchange aids in converting the data into any other file formats like MATLAB, GRAMS, EXCEL, etc.

Data Visualization is possible by plotting the graphs or 1D, 2D, 3D imaging and other visualization tools.

NIRCal Toolbox is the main feature that helps in pretreatments, transformations and exchange of algorithms.

Advantages of NIRCal:

It works together with NIRWare and collects data from it for calibration.

Calibration is carried out on knowledge basis.

All spectra and calibrations are carried out with the help of NIR Explorer.

NIRCal Chemometrics:

In helps in performing both Qualitative and Quantitative analysis.

Statistical methods like PLS, PCR, MLR, CLUSTER and SIMCA are performed.

Multivariate tasks like normalization, SNV, Derivativization, Smoothing and Transformations are available.

LITERATURE REVIEW

Many efforts were carried out in search of a proper and efficient method for determining the appropriate method to monitor the blending homogeneity.

O. Berntsson et al. in year 2001 by using Fourier transform ŽFT.-NIR, carried out a profound study on chemical composition of powder blends. It is a quantitative estimation done using reflection fibre-optic probe and using simple multivariate method for calibrations.By relating process time and chemical composition positive results are obtained, helping in assessment of blend homogeneity and average content distribution.

Osama S et al. in 2000 carried out their study on charecterization of granule blending. They mainly focused on the fill level and mixing time, by taking four fill levels and changing loading patterns. By using the first order mixing model they found that the axial mixing resulted in limiting the blend homogeneity. This work is carried out only on granules and it differs a lot from the current powder blending. Moreover there is no wider application of chemometrics in deriving the results.

In 2005, ARWA S. EL-HAGRASY and JAMES K. DRENNEN III by the initiation of Process Analytical Technique and FDA worked on blend homogeneity estimation. They applied the real-time procedure using NIRS and performed quantitative statistical analysis using Principal Component Regression, Partial Least Square and Multi-linear Regression methods. The predicted results showed positive when correlated with UV reference method. At single wavelength linear regression showed positive results.

zhenqi Shia et al. in 2007 used in-line method for charecterizing the powder blending. They used NIRS and applied multivariate statistical methods like Root mean square values, T-tests and used Plackett-Burman design to determine the blending end point. They used a ternary mixture of acetaminophen, lactose and MCC with different blending limits and compositions. But they didn't use technical software and transformations.

By using this Mean and Standard deviation of Root Mean square values they determined the variability and robustness of the blending end point.

In 2008, I. Storme-Paris et al. worked with NIRS to estimate homogeneity of complex blends of colimycin, tobramycin and some excepients. They carried out a small scale procedure as part of a treatment for children. They estimated 39 different blend with 14 formulations changing the proportions of ingredients. They referred their methods using HPLC and Gravimetry. For qualitative evaluations they used Moving Block Standard Deviation Partial Least Square Discriminate Analysis and Partial Least Square Cross-Validated method for quantitative evaluation. The end point resulted at 16 minutes process of blending. This experiment showed the efficiency of NIRS and Chemometrics for evaluating blend homogeneity.

Most recently in 2010, A.D.Karande, with his group gone through a case study on powder blends. They used a multi component mixture of chlorpheniramine maleate, lactose, microcrystalline cellulose and magnesium stearate to study the spectral acquisitions for laboratory and IBC mixings. They performed 24 blending runs and used Partial Least Square calibration method for the determination of the mixing homogeneity. They attained good results for PLS models and IBC mixed blends with dynamic acquisition. But they got poor results in static acquisition models.

Apart from all the previous research carried out in monitoring blending-end point, this project mainly focus on the use of other analytical-statistic methods such as Principal Component Analysis (PCA), Cluster Analysis and Moving-Window-pooled Standard Deviation. These are carried out by applying transformations like secondary derivative method and Standard Normal Variate method.

EXPERIMENTAL PROCEDURES

MATERIALS AND EQUIPMENT:

BINARY MIXTURE:

The binary mixture of α-Lactose monohydrate (Sigma-Aldrich, UK) and Kaolin (Sigma-Aldrich, UK) is used as samples for the blending end point monitoring.

Several properties and favouring characters of lactose made it compatible as filler in many pharmaceutical formulations like tablets and capsules. The properties like low hygroscopicity, and less compatibility with Pharmaceutical Active Ingredients (PAI) and other excipients and its physical and chemical stability are the most favourable properties of lactose. Here crystalline a-lactose monohydrate is used for blending as it is the majorly used form of lactose.

Even though kaolin is not much pharmaceutical preparations because of its less compatible nature it suits for this experiment to monitor blending point. Kaolin is used as a tablet excipient, serves as an emollient (topically) and controls diarrhoea when ingested.

Blend uniformity is achieved easily with good flow properties like regular shape (particularly spherical) and narrow particle size distributions. The flow properties of α-Lactose monohydrate are good compared with Kaolin as the later one shows some gritty nature but these are negligible. Powder density also plays a good role in this process as the density increases the rate of sedimentation of the particles increases and this result in a non-uniform mixture i.e.., segregation of the denser particles fast in the pharmaceutical formulations. The density of the lactose is good and in acceptable range for the pharmaceutical preparations.

PRODUCT

DENSITY

Lactose Fine Crystals

0.73 (Bulk )

0.86 (Tapped )

Kaolin

2.6 (Relative with water =1)

Properties of Lactose:

LACTOSE

HOSOGAWA FLOW INDEX

FLOWABILITY-QUALIFICATION

Coarse crystals

Crystals

Fine crystals

79

75

75

good

good

good

Many other factors are also considered in blending process to avoid parallax errors. Chemically lactose is very stable; the low hygroscopicity of lactose supports its inertness and prevents from Maillard reaction. It has no tendency to attract moisture and the water of crystallisation is bound tightly in the crystal lattice and is lost only form 100°C to 140°C. This nature of lactose helps in preventing unwanted reactions and contamination of the active ingredients. Both kaolin and lactose are incompatible and cannot bind with each other in the middle of the process and it results in a uniform mixture.

α-Lactose monohydrate

(β-D-Gal-(1→4)-α-D-Glc, Milk sugar, 4-O-β-D-Galactopyranosyl-α-D-glucose)

Kaolin

(Aluminum silicate hydroxide, Bolus, Hydrated aluminum silicate)

MW

360.31

258

Formula

C12H22O11 · H2O

~Al2Si2O5(OH)4

Impurities

0.01% Glucose

≤0.0025% heavy metals (as Pb)

≤1% soluble in acid (as SO4)

Kaolin has surface charges but they do not affect the blending in regular basis. If the '-' ve or '+'ve charge of kaolin raise they develop the electrostatic forces which results in colloidal instability and interactions with low molecular polyions.

In order to achieve ideal blending the components should not interact chemically and lactose and kaolin both satisfy that feature.

Equipment:

The equipment used in this experiment is a spatula to withdraw the samples, 200 pcs. of 4ml flat bottomed sample vials (15 x 45 mm) (Waters, UK) for handling samples to use on NIRS.

A stop watch is used to count the equal intervals of time at which samples are collected, a 200 ml plastic container two hold the binary mixture and 2 beakers, a sample vial holder, a scientific calculator, weighing boats and digital weighing balance.

INSTRUMENTATION:

BUCHI NIRFLEX N-500 FT-NIR SPECTROMETER:

The instrument used is Buchi NIRFlex N-500 FT-NIR spectrometer (Buchi UK Ltd, Oldham, UK). It works on the principle of over toning and combination vibrations of molecules. When energy transitions take place in atoms they don't follow electric dipole method and emit a spectral line. The spectra attained by this overtone and combination vibrations are very complex and are with broad bands.

NIR spectrometer was used for the first time in 1950s, but it was used in combinations with other devices like UV, Vis, MIR spectrometers, etc. and was later used as a single unit. The beginning of light-fibre optics and monochromatic-detector made NIR a powerful tool for scientific research. NIRS is applied in the analysis of foodstuffs, pharmaceuticals, medical tool and a branch of astronomical spectroscopy.

BUCHI NIRFLEX N-500

FIGURE :http://www.buchi.com/NIRFlex-N-500.465.0.html

It is equipped with a light source and a thermo-electrically cooled InGaAs detector (with Indium gallium arsenide semiconductor), dispersive element and polarization interferometer and a wide spectral range of 800 - 2500 nm (12,500 - 4000 cm -1). In order to reduce the mechanical distortions and to manage difference of spatial movements and the optical path shifts a rugged crystal polarization interferometer is used which is of superior performance. The polarization is carried out through the crystals with very high refractive index. The principle involved is, the incident light is split to two when it strikes the crystal, and they are polarized in right angles and traverse the crystal at different velocities. Then they shift to different phases because of the moving prisms and hence polarization of combined beam is changed.

It also helps in obtaining with an optimum resolution of 8cm-1. For solid samples because of broad absorption bands, higher resolutions are not fruitful for NIR applications. Also, higher resolution gives unnecessary bulk data sets and leads to a poor signal-to-noise ratio, where it is 10,000: 1 by using rugged crystal interferometer. This NIRS collects an interferogram and transforms it to a new single-beam frequency-domain spectrum. Then it subtracts the reference spectrum that is acquired in the starting from this transformed spectrum to giving a new reflectance spectrum.

Spectral range

800-2500 nm

12500-4000 cm-1

Resolution

8 cm-1 (with boxcar apodization)

Type of interferometer

Polarisation interferometer with TeO2 wedges

Wavenumber accuracy

± 0.2 cm-1

Signal-to-noise ratio

10,000 : 1

Number of scans

2-4/sec

Analog digital converter

24 bit

Ambient temperature

5-35 °C

Type of lamp/lifetime lamp (MTBF)

Tungsten halogen lamp / 12 000 h (2x 6000 h)

Type of laser

12 VDC HeNe, wavelength at 632.992 nm

Photometric dyn. range

2AU

FIGURE: SPECIFICATIONS OF ------ Prospekt_NIRFlex_en_0810

The spectrum is measured by single sample solid reflectance angle.

The data obtained is then used for extracting the chemical information using the multivariate calibration techniques like PCA and Cluster Analysis. For all near infrared techniques develop the calibration samples and are reduced and calibrated using multivariate techniques.

NIRS technique surmounts other established techniques by its unique properties

TURBULA T 2 F MIXER (WAB, MUTTENZ 1, SWITZERLAND):

FIGURE:

TURBULA T 2 F was used in various fields like Cosmetic, Pharmaceutical and Chemical industries while handling sample load up to 10 Kg. Turbula mixer is very efficient as it can mix particles with different size and densities, can handle both dry and wet samples and can generates a homogenous mixture within a short period. It consists of a basket wind with elastic belt to hold container and gears.

The efficient homogeneity is reached by the 3-Dimensional mixing following the Paul Schatz theory of geometry. The basket with container runs in rhythmic movements with rotation, translation and inversion motions. His first practical application of inversion motion was applied in Turbula.

The other instrument used for mixing samples industrially and in laboratories is INVERSINA (Bioengineering AG, CH - 8636 Wald). But because of many additional features of the Turbula it is used in determining the blending homogeneity.

It is highly versatile and can handle containers up to 55 litres depending on the requirement.

3-Dimensional motions as per Schatz.

Belt driven and geared basket which is adjustable.

Rotational speed is 23 per minute. And can be adjustable up to 101 rpm.

Frequency converter to adjust frequency.

Process is sterile avoiding exposure of sample contents to external environment.

STEPS INVOLVED:

SAMPLE PREPARATION :

11 batches of binary mixtures are prepared: α-Lactose monohydrate (Sigma-Aldrich, UK) with Kaolin (Sigma-Aldrich, UK). These are taken in 2:5 proportion respectively to carry out the analysis.

42.85 gm / 0.042 kg of a-Lactose monohydrate is weighed using a digital weighing balance. This powder sample is transferred into a clean and dry container and closed tightly.

107.14 gm / 0.107 kg of Kaolin is weighed and transferred carefully into another clean, dry container without any agitation.

Both the chemicals (150 gm) are transferred carefully (without mixing) into a dry plastic cylindrical container of 200 ml volume and the lid is closed tightly.

All this process is carried out carefully without any agitation or tilting to prevent the false prediction of the blending end point.

BLENDING OPERATION :

Blending operation is carried out using a Turbula T 2 F mixer (WAB, Muttenz 1, Switzerland) blender.

The required frequency and speed are adjusted to obtain the optimum reproducible results.

The geared basket is adjusted using a driver so as to give enough space for the plastic container.

Then chemical filled sample container is then fastened inside the basket, in order to avoid any misplacement of the container in the middle of the process.

The blender is operated and the sample is collected into labelled sample vials using spatula for every 40 seconds up to 10 minutes.

The sample is collected without creating any disturbance to the rest of the sample.

PREPARING NIRS INSTRUMENT :

The NIRWare software and NIRCal software (BUCHI, UK) are installed in the NIRFlex N-500 (BUCHI, UK) instrument to perform the analysis of samples.

The application required for blending end point monitoring is Single Solid Reflectance. This application is created using a reference for identity check of incoming substances.

System Suitability test is performed by clicking on "advanced", then "perform system suitability test".

External reference is measured carried with a white block.

Reference measurements are carried out to make the spectrometer sustainable. External reference helps to measure by dividing sample spectrum by reference spectrum. Internal referencing helps in reducing the need for external reference.

ACQUISITION OF NIR SPECTRA :

The sample vials are cleaned at the bottom, and are placed on the cell positions and run.

Enter the sample id and after that click on "result and spectra".

These spectra are saved as separate documents in XPS Format.

DATA ANALYSIS:

Data analysis is carried out using NIRCal software. The steps involved here are

Get in to "File", then click on "databases", then "search" and click on " import spectra " and then select the files corresponding to required spectra.

Then go to "table" , then "spectra", then "original", and right click and export table.

Then save all these files as separate documents in the XLS Format.

Each data matrix obtained gives 1501 wave numbers in rows and their corresponding readings.

METHODOLOGY:

Further analysis is carried out by using this data with unscrambler applying chemometrics.

The data obtained in the XLS file format is edited by removing the irrelevant data in columns and rows. Then these data is transposed, rows into columns and vice versa. Then these are uploaded into the unscrambler and chemometric methods are applied.

The spectral data is uploaded to data matrix and transformations are made using Savitzky-Golay method secondary derivative and Standard Normal Variate method.

Two line graphs are plot using these new transformations.

Using the secondary derivative method Principal Component Analysis and Cluster analysis is carried out.

Graphs are plot individually for the Scores and loadings in PCA and for Clusters formed.

MOVING WINDOW POOLED STANDARD DEVIATION METHOD:

In this method all the variables are collected in an Excel sheet. Then these values are transposed, rows to columns and columns to rows. Then Standard deviation is performed for all the data and these standard deviations are pooled. Then taking time interval on X-axis and pooled standard deviation values on Y-axis a graph is plot.

Interpretation is carried out using graphs from all these methods and the results obtained were discussed.

RESULTS AND DISCUSSIONS

RESULTS FOR BATCH-1

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

STANDARD NORMAL VARIATE:

C) PRINCIPAL COMPONENT ANALYSIS:

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R8, R13, R11 and R15 is very small and they are nearby. When these are correlated with Clusters the variables R8, R11 and R13 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R8 to R13 i.e., between 5.20 minutes to 8.40 minutes from the beginning of the experiment.

BATCH-2:

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

STANDARD NORMAL VARIATE:

PRINCIPAL COMPONENT ANALYSIS:

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R11, R8, R6 is very small and they are nearby. When these are correlated with Clusters the variables R11, R12, R8, R10 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R8 to R11 i.e., between 5.20 minutes to 7.20 minutes from the beginning of the experiment.

BATCH-3:

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

STANDARD NORMAL VARIATE:

PRINCIPAL COMPONENT ANALYSIS:

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R6, R8, R10 is very small and they are nearby. When these are correlated with Clusters the variables R9, R6, R8, are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R6 to R10 i.e., between 4 minutes to 6.40 minutes from the beginning of the experiment.

BATCH-4:

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

STANDARD NORMAL VARIATE:

PRINCIPAL COMPONENT ANALYSIS:

D) CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R13, R8, R11is very small and they are nearby. When these are correlated with Clusters the variables R10, R13, R8, R2are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R8 to R13 i.e., between 5.20 minutes to 8.40 minutes from the beginning of the experiment.

BATCH-5:

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

B) STANDARD NORMAL VARIATE:

PRINCIPAL COMPONENT ANALYSIS:

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R5, R9, R4 is very small and they are nearby. When these are correlated with Clusters the variables R5, R6, R9 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R5 to R9 i.e., between 3.20 minutes to 6.00 minutes from the beginning of the experiment.

BATCH-6:

STANDARD NORMAL VARIATE:

SECONDARY DERIVATIVE (SAVITZKY-GOLAY):

Principal Component Analysis:

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R7, R8, R12, R9 is very small and they are nearby. When these are correlated with Clusters the variables R7, R8, R11 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R7 to R11 i.e., between 4.40 minutes to 7.20 minutes from the beginning of the experiment.

BATCH-7:

Principal Component Analysis:

B) Standard Normal Variate;

Secondary derivative (Savitzky-Golay):

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R5, R9, R15 is very small and they are nearby. When these are correlated with Clusters the variables R5, R9 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R5 to R9 i.e., between 3.20 minutes to 6.00 minutes from the beginning of the experiment.

BATCH:8

Principal Component Analysis:

Standard Normal Variate;

Secondary derivative (Savitzky-Golay):

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variablesR10, R5, R14, is very small and they are nearby. When these are correlated with Clusters the variables R10, R14, R12 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R10 to R14 i.e., between 6.40 minutes to 9.20 minutes from the beginning of the experiment.

BATCH-9:

Principal Component Analysis:

B) Standard Normal Variate:

Secondary derivative (Savitzky-Golay):

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R8, R9 is very small and they are nearby. When these are correlated with Clusters the variables R7, R9 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R7 to R9 i.e., between 4.40 minutes to 6.00 minutes from the beginning of the experiment.

BATCH-10:

Principal Component Analysis:

Standard Normal Variate:

Secondary derivative (Savitzky-Golay):

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R10, R9, R11, R14 is very small and they are nearby. When these are correlated with Clusters the variables R14, R11, R9,R10 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R9 TO R14 i.e., between 6.00 minutes to 9.20 minutes from the beginning of the experiment.

BATCH-11:

Principal Component Analysis:

Standard Normal Variate:

Secondary derivative (Savitzky-Golay):

CLUSTER ANALYSIS:

GRAPHS: A, B, C, D

DISCUSSION:

By comparing line plots obtained from the secondary derivative and Standard Normal Variate, the differences can be observed which means there is elimination and discrimination of some variables in secondary derivative method, and increase in complexity can be observed in higher order.

From the scores of the PCAnalysis, it can be noticed that the relative distance between variables R11, R7, R15, R5 is very small and they are nearby. When these are correlated with Clusters the variables R7, R11, R14 are very close to each other.

By these we can analyse that, homogenous mixture is obtained between R7 to R11 i.e., between 4.40 minutes to 7.20 minutes from the beginning of the experiment.

CONCLUSION

From all the derived values we can draw a conclusion that to form a homogenous blend of α-Lactose monohydrate and Kaolin, it takes time between 5.20 minutes to 7.20 minutes.

This is because when we take the average of all those values mostly the end point is between R8-R11.

By this we understand that by using PCA and Cluster analysis the data are mostly reproducible and so, this methods are applicable for the blending end-point determination.

The results obtained by Moving-Window­-Pooled-Standard deviation are not reliable.

These results are not reproducible. Even though when we take average of all this results the are near R8 variable. But there is much variance between each batch.

REFERENCES

S.S. Sekulic, J. Wakeman, P. Doherty, P.A. Hailey. (1998). Automated system for the online monitoring of powder blending processes using near-infrared spectroscopy. J. Pharm. Biomed. Anal. 17 (2), 1285 - 1309

A.S. El-Hagrasy, M. Delgado-Lopez, J.K. Drennen. (2006). , A process analytical technology approach to near-infrared process control of pharmaceutical powder blending. J. Pharm. Sci. 95 (2), 407 -430

Antohny J. Owen. (1995). uses of derivative spectroscopy. Available: http://www.agilent.com/chem. Last accessed 28th aug 2010

Principal Component Analysis. Available: http://www.camo.com/resources/principal-component-analysis.html. Last accessed 26th aug 2010.

Buchi lboratories. (2007). NIRWare 1.2 Software Manual. Available: http://www.buchi.com/NIRFlex-N-500.465.0.html?&no_cache=1&file=353&uid=14363. Last accessed 24th July 20#

Rakoto malala. (2002). web projects. Available: http://www.chm.bris.ac.uk/webprojects2002/rakotomalala/maillard.htm. Last accessed 10th july 2010

Hubbard (2006). Encyclopedia of Surface and Colloid Science. 2nd ed. USA: Taylor & Francis . 546-548.

Anthony J. Owen . (1995). Agilent Technologies. Available: http://www.agilent.com/chem. Last accessed 25th aug

ARWA S. EL-HAGRASY. (2006). A Process Analytical Technology Approach to Near-Infrared. journal of pharmaceutical sciences. 95

Chee-Kong Lai, David Holt, James C. Leung, and Charles L. Cooney. (2001). Real Time and Noninvasive Monitoring of. AIChE Journal.

K. J. KAFFKA. (2002). PQS (POLAR QUALIFICATION SYSTEM) THE NEW DATA. Acta Alimentaria. 31(1) (31), 3-20

James N Miller (2005). Statistics and Chemometrics for Analytical Chemistry. 5th ed. England: Pearson Education Limited.

Pharmaceutical and analytical R&D, astra Zeneca R&D mo indal, se-431,Sweden(2001)

Osama S. Sudah. (2002). Quantitative characterization of mixing of free-flowing granular. Powder Technology. 126 (1), 191-2002

W.plugge. (1996). Near infrared spectroscopy as a tool to improve quality. pharmaceutical and biomedical analysis. 14 (01), 891-899

P.A. Hailey. (1996). Automated system for the on-line monitoring of powder. PHARMACEUTICAL AND BIOMEDICAL ANALYSIS. 14 (01), 551-559.

Ulf G. Indahl. (1999). Multivariate strategies for classification based on. chemometrics and intilligent laboratory systems. 49 (01), 19-31.

Zhenqi Shia. (2008). Process characterization of powder blending by near-infrared spectroscopy:. Journal of Pharmaceutical and Biomedical Analysis. 47 (01), 738-745

Fridrun Podczeck,. (1996). The influence of particle size and shape of components of binary powder mixtures on the maximum volume reduction due to packing. international journal pharmaceutics. 137-145.

Writing Services

Essay Writing
Service

Find out how the very best essay writing service can help you accomplish more and achieve higher marks today.

Assignment Writing Service

From complicated assignments to tricky tasks, our experts can tackle virtually any question thrown at them.

Dissertation Writing Service

A dissertation (also known as a thesis or research project) is probably the most important piece of work for any student! From full dissertations to individual chapters, we’re on hand to support you.

Coursework Writing Service

Our expert qualified writers can help you get your coursework right first time, every time.

Dissertation Proposal Service

The first step to completing a dissertation is to create a proposal that talks about what you wish to do. Our experts can design suitable methodologies - perfect to help you get started with a dissertation.

Report Writing
Service

Reports for any audience. Perfectly structured, professionally written, and tailored to suit your exact requirements.

Essay Skeleton Answer Service

If you’re just looking for some help to get started on an essay, our outline service provides you with a perfect essay plan.

Marking & Proofreading Service

Not sure if your work is hitting the mark? Struggling to get feedback from your lecturer? Our premium marking service was created just for you - get the feedback you deserve now.

Exam Revision
Service

Exams can be one of the most stressful experiences you’ll ever have! Revision is key, and we’re here to help. With custom created revision notes and exam answers, you’ll never feel underprepared again.