Near-Infrared Spectroscopy as a Process Analytical Tool

Near-infrared (NIR) spectroscopy has been used for qualitative and quantitative measurements in the agricultural, food, chemical and pharmaceutical industries for several decades. It has been used effectively from the inspection of incoming raw materials to the final testing of manufactured products. This article discusses NIR implementation strategies for process applications in the pharmaceutical industry, and provides examples of the use of NIR spectroscopy for measurements such as moisture and reaction monitoring, and polymorph determinations.

Improving product quality and lowering costs are the key factors behind the decisions made in many industries. Ensuring product quality throughout the manufacturing process can be time-consuming, with materials and products 'quarantined' until test results are generated. Rapid testing by near-infrared (NIR) spectroscopy at all stages of the manufacturing process can reduce production time and provide assurances at each step of the process that product quality is being maintained.

Figure 1 Illustration of places throughout manufacturing that NIR spectroscopy can be implemented.

NIR spectroscopy

The NIR region of the electromagnetic spectrum covers the 800-2500 nm range and is characterized by broad, overlapping bands that arise from combination and overtones of the fundamental vibrations found in the mid-infrared region of the spectrum. NIR absorption bands are typically one to two orders of magnitude weaker than the fundamental bands, making them ideal for direct analysis of samples without the need for dilutions or very narrow pathlengths. The ability to use longer pathlengths and to carry NIR light through low OH-silica fibre optics makes NIR spectroscopy a very good technique for in-process use. Additionally, because an NIR spectrum can be measured on a sample without any preparation (in less than a minute), NIR spectroscopy is a rapid analysis method. The NIR spectral region carries chemical information, with the strongest absorbances being those related to the CH, OH and NH functional groups. This allows for quantitative measurement of chemical concentrations in a matrix.

^1,2

The NIR spectrum also carries information related to the physical characteristic of samples that are created by scattering differences, which cause baseline offsets. This allows the NIR spectrum to be used to estimate average particle size³ and per cent biomass in fermentations.⁴ Other physical parameters have also been measured by NIR, including quantitation of polymorphs and the degree of crystallinity.^5-7

Figure 2 Reflectance NIR spectra of common excipient materials including mannitol and microcrystalline cellulose.

NIR spectroscopy, combined with chemometric tools, can be employed to develop calibration models for the qualitative and quantitative analysis of materials measured in transmission or reflectance. It has also been widely used for applications such as the rapid identification/testing of materials,^8-10 assay of intact tablets¹¹ and process monitoring.^12-14 Figure 1 provides an illustration of applications of NIR throughout the manufacturing process.

Process analytical chemistry

Process analytical chemistry refers to measurements made that provide analytical information to increase process control. The measurements can be made in a laboratory, a materials dispensary or directly in the process stream. Analysis that is performed with direct interface of the analyser to the process is called in-line, whereas online analysis entails the use of a recirculation loop or a sample transport system that automatically removes the sample from the process and brings it to the analyser. Analyses that are performed on samples removed from the process may be off-line in a remote or centralized laboratory, or at-line, where an analyser is located in close proximity to the process.

^15,16

NIR spectroscopy is one of many tools used to help ensure product quality by measurement throughout a process, from incoming raw materials to final product. It is important that process analytical projects be problem driven and not technology driven, ensuring that the tool that provides the required data and information is used. The analysis must be made in such a manner that the data provide information that can be used to control the process. The analyser must be able to detect changes in the process as and when they occur.

Figure 3 Transmittance NIR spectrum of water.

NIR process analytical applications can be used to follow trends in quantitative analysis for the determination of end-points, control of specification limits or to determine if a process is following similar reaction trends as previous runs of the same process. Qualitative analysis allows identification and qualification of raw materials, and ensures that final product spectra compare well with previous "good" batches. By using NIR as a process analytical tool, information regarding the process can be generated in near real-time.

Overview of NIR applications

Raw materials testing.

The application of NIR spectroscopy as an identification method provides a rapid means of testing materials without the need for sample preparation. It can be used to distinguish between different materials such as excipients (Figure 2) or materials that vary in their state of hydration.

¹⁷

In the pharmaceutical industry, methods of identifying incoming raw materials by NIR have been validated

¹⁰

and used for testing large volumes of material without the need for extensive testing by wet chemical or chromatographic techniques. An NIR raw material library is then developed by scanning numerous lots of raw materials and discriminate analysis is used to qualitatively identify the materials. As a routine test procedure, the spectrum of a new sample is scanned and its spectrum compared with the mean spectrum of each product in the library to obtain a rapid identification. Qualification (determination of good/bad quality) can be verified using tolerances from spectra of previous "good" quality materials, which is an additional benefit of the NIR method.

Moisture measurement. It is important for pharmaceutical manufacturers to be able to monitor moisture levels. Water is a very strong absorber in the NIR region, making it a good candidate for quantitative analysis by NIR (Figure 3). The strong combination OH band at approximately 1930 nm and first overtone OH stretching from approximately 1450 nm can be used to measure moisture. In 1965, determination of moisture in agricultural products was one of the earliest applications of NIR.¹⁸ Figure 4 shows how the NIR spectrum of a product changes with moisture level variation. The moisture level can be determined when calibration models have been developed against a reference method, such as Karl Fischer titration.

Figure 4 Reflectance NIR spectra of compounds with varying moisture content.

NIR has been used to monitor moisture in lyophilized products without the need to remove samples from the lyophilization vials.¹⁹ It has also been used to measure the total water, as well as the surface and bound water in drug substances during drying processes.²⁰ Moisture has also been measured by NIR in wet granulation operations to determine the end-point of a granulation.³

Reaction monitoring. Online NIR application allows a reaction's progress to be monitored as it proceeds. It provides information such as the concentration of reactants, formation of by-products and the end-point. Multiple measurements can be made simultaneously within a reaction (Figure 5). By directly interfacing the analyser to the reactor, safety hazards that may arise from the need to remove a sample from a reactor are minimized. NIR has also been used to determine the end-point of a hydrogenation reaction,¹³ monitor fermentation reactions⁴ and for the process analysis of fluidized bed granulation.^3,21

Polymorph determination. Identifying the polymorphic forms of a pharmaceutical compound is important because of the differences in their efficacy and behaviour during processing. NIR is sensitive to different polymorphs because of the different molecular arrangement and the concomitant change in hydrogen bonding. The NIR spectra of two different polymorphs of a substance are given in Figure 6, which shows that polymorphism can lead to substantial differences in the spectrum. Rapid and sensitive methods to determine polymorphic quality using NIR and pattern recognition methods have been developed.⁵ Qualitative analysis has also been used to discriminate between desired and other polymorphic forms. Additionally, low levels of polymorphs in binary mixtures have been quantified using reflectance NIR spectroscopy.⁶

Figure 5 Trend chart of concentration of five components simultaneously being monitored by NIR.

Testing of solid dosage forms.

Final product testing of intact tablets can be performed using NIR reflectance.^11,22 The use of NIR on the solid tablet maintains sample integrity. The NIR spectra of tablets have been used to develop qualitative identification methods for tablets, as well as quantitative methods to measure dosage strength.

Implementation strategies

When applying NIR as a process analytical tool, careful consideration must be given regarding which part of the process to monitor. Reasons for monitoring a process include improving product quality and reducing product variability; reducing cycle time and manufacturing costs; and reducing hazards, including the potential dangers to operators who currently need to sample a process. The analyser should be chosen based on its ability to solve problems and provide the required information.

Implementing a process analyser requires a multifunctional team, particularly if the project involves technology transfer from the research and development operation to a manufacturing site. A successful team may include

• a strong leader with support from senior management

• an analytical chemist

• a chemometrician

• a process development chemist

• a control engineer

• a production engineer

• someone to oversee documentation and validation.

The project must be well defined and consideration must be given to all the steps involved in implementation (Table I). Some organizations have a process analytical group with responsibility for all analysers throughout the company. The collective experience of the personnel in such a group aids implementation.

Once a process has been identified for NIR analysis, there is a choice to use either reflectance or transmission measurement. For solids and highly scattering slurries, reflectance measurements can be made directly with a fibre optic probe. For liquid systems, transmission or transflectance measurements can be used. The optimal pathlength for the measurement must be determined. For batch reactions, measurements can be made online using a pair of transmission probes (one for sample illumination and the other for the collection of transmitted light) or in-line with a transflectance probe (Figure 7). For in-line measurement, the probe pair can be interfaced to a recirculation loop.

Figure 6 NIR spectra of two different polymorphs of a substance.

One of the biggest challenges of online and in-line analysis is the sampling system.^15,23 It is best if a probe can be interfaced directly into a process, but if this is not possible there are several other options. The sample that is being measured must be representative of the process being measured. It is also important to determine how the sample will be presented to the analyser. The placement of the analyser, probes and computer must be determined based on site and measurement requirements. How the probes interface with the sample must also be considered - will the measurement be online with a sidestream, conditioned sample slipstream or in-line? Care should also be taken to minimize interface cleaning during the process.

The analyser must be calibrated; therefore, there must be defined acceptance criteria for the calibrations. A procedure for testing the calibration must also be established and performed. Guidelines regarding quantitative multivariate analysis of spectrometers have previously been published;²⁴ there are also guidelines regarding the validation of process spectrometers.²⁵ Development of robust calibration models that cover expected variations of a process reduce the necessity of making updates to the calibration in the long run. It is also important that the laboratory analysis be representative of the sample that the NIR scans. Frequently, this means collecting samples during the NIR scan, or as near to that time as possible, and also in close proximity to where the NIR operates.

Figure 7 Probes that can be used in for transmission measurements on a process. (a) Transflectance probe. (b) Transmission pair probes and spacers.

Conclusions

NIR is a superior tool for use throughout the manufacturing process. It has been widely used for raw materials testing, and for quantitative measurements off-line and online. This article has discussed examples of applications of NIR in the pharmaceutical industry, from incoming raw material testing through the process to final product. The important aspects to consider when implementing NIR for process analysis have also been examined.

Table I The project team's view of a process analytical implementation plan.

References

1. P. Williams and K. Norris, Near-Infrared Technology in the Agricultural and Food Industries, 2nd Edition (American Association of Cereal Chemists, Inc., St Paul, Minnesota, USA, 2001).

2. D.A. Burns and E.W. Ciurczak, Handbook of Near-Infrared Analysis, 2nd Edition (Marcel Dekker, Inc., New York, New York, USA, 2001).

3. P. Frake et al., Int. J. Pharm. 151, 75-80 (1997).

4. S.A. Arnold et al., BioPharm Int. 15(11), 26-34 (2002).

5. P.K. Aldridge et al., Anal. Chem. 68(6), 997-1002 (1996).

6. A.D. Patel, P.E. Luner and M.S. Kemper, J. Pharm. Sci. 90(3), 360-370 (2001).

7. J.J. Seyer, P.E. Luner and M.S. Kemper, J. Pharm. Sci. 89(10), 1305-1316 (2000).

8. W.L. Yoon, R.D. Jee and A.C. Moffat, Analyst 125, 1817-1822 (2000).

9. A. Candolfi et al., J. Pharm. Biomed. Anal. 19, 923-935 (1999).

10. C. Anderson, Am. Pharm. Rev. 3(1), 30-42 (2000).

11. M. Blanco et al., J. Pharm. Biomed. Anal. 22, 139-148 (2000).

12. P.A. Hailey, Eur. Pharm. Rev. 1(2) (1996).

13. H.W. Ward et al., Appl. Spectrosc. 52(1), 17-21 (1998).

14. H.H. Lousberg et al., J. Appl. Polym. Sci. 84, 90-98 (2002).

15. D.C. Hassell and E.M. Bowman, Appl. Spectrosc. 52(1), 18A-29A (1998).

16. F. McLennan and B.R. Kowalski, Process Analytical Chemistry (Blackie Academic and Professional, New York, New York, USA, 1996).

17. S.H. Hogan and G. Buckton, Int. J. Pharm. 227, 57-69 (2001).

18. K.H. Norris and J.R. Hart, "Direct Spectrophotometric Determination of Moisture Content of Grain and Seeds," in A. Wexler, Ed., Principles and Methods of Measuring Moisture in Liquids and Solids, Vol. 4 (Reinhold, New York, New York, USA, 1965) p 19.

19. R.M. Leasure and M.K. Gangwer, Am. Pharm. Rev. 5, 103-109 (2002).

20. G.X. Zhou et al., J. Pharm. Sci. 92(5), 1058-1065 (2003).

21. J. Rantanen et al., AAPS PharmSciTech 2(4), 21 (2001).

22. R.A. Lodder and G.M. Heiftje, Appl. Spectrosc. 42(4), 556 (1988).

23. J. Maestro, "Sampling and Engineering Considerations," in J.M. Chalmers, Ed., Spectroscopy in Process Analysis (Sheffield Academic Press, Sheffield, UK, 2000).

24. ASTM E1655-00, Standard Practice for Infrared Quantitative Analysis (ASTM International, West Conshohocken, Pennsylvania 19428, USA).

25. ASTM D6122-99, Standard Practice for Validation of Multivariate Process Infrared Spectrophotometers (ASTM International, West Conshohocken, Pennsylvania 19428, USA).