Raman as a PAT Tool

FDA's process analytical technology (PAT) initiative that asserts "quality cannot be tested into products, it should be built in or should be by design" has generated a large amount of interest in new technologies for pharmaceutical analysis.^1,2 A key aspect of FDA's PAT philosophy is that every stage of the manufacturing process should be monitored and understood so that the end product is right first time, every time. In practice, the implementation of this philosophy will require a range of analytical tools capable of operating throughout the manufacturing process: at-line, on-line and in-line.

These technologies must be capable of generating detailed, meaningful results in real time, which may be fed back into the control mechanisms of the process to ensure the final product quality is consistently maintained. Near infrared (NIR) spectroscopy has been widely discussed and applied in this role.³ Raman spectroscopy also has great potential in this application and has indeed formed part of FDA's PAT initiative education programme¹ and reports are now beginning to appear in the literature discussing Raman's application as a PAT-type tool.^4–8

Benefits for PAT Applications

The Raman effect was first observed in 1928 by C.V. Raman

⁹

and has been extensively studied and applied, generating an enormous body of scientific literature. This superior scientific foundation enables a deeper, molecular level understanding of the manufacturing process and gives confidence to both the manufacturer and the regulatory authorities in the value of this technique.

Well resolved, information rich spectra. The sharp features of the Raman spectra enable individual constituents in complex mixtures to be identified. Characteristic features of the chemical species of interest can often be identified in isolation from interfering features from other materials allowing simultaneous monitoring of a large number of chemical species and the use of simple, robust univariate prediction models. This is illustrated in Figure 1(a) where a simple model was generated to determine the relative concentrations of two solvents in a mixture. The Raman bands at 840 cm^-1 and 882 cm^-1 are characteristic of solvent A and solvent B respectively, and are sufficiently well resolved to be used without the need for multivariate analysis. The intensity of the Raman bands is expected to be linearly proportional to the concentration of the scattering species and this was found to be the case (Figure 1(b)).

Additionally, the large amount of chemical information present in a typical Raman spectrum can contribute to a fuller understanding of the process at a molecular level.

Figure 1: An example of the use of Raman spectroscopy to determine the component concentrations in a binary solvent mixture.

Aqueous samples, slurries, solids, liquids and gases. Water has a very strong absorption in NIR and mid-infrared spectroscopy, which often totally obscures the signal from the materials of interest in aqueous environments. The Raman spectrum of water is, however, very weak and generally does not interfere with the spectra of the materials in solution — this is one of Raman spectroscopy's key advantages when compared with infrared (IR) absorption methods. This is illustrated in Figure 2 where an IR absorption spectrum and a Raman spectrum from an aqueous suspension/solution of an active pharmaceutical ingredient (API) are compared.

Figure 2: Comparison of (a) IR absorption and (b) Raman spectra for an aqueous suspension/solution of an API.

Although for the IR measurement the optical path length was only ~10–20 microns, an extremely large feature because of water is observed at ~1640 cm^-1 as well as a number of sharper features because of atmospheric water vapour. By contrast, the Raman spectrum was obtained by focussing through the wall of a sealed cuvette ~0.5 mm into the sample; the Raman band of water, which also lies at ~1640 cm^-1, is hidden under the signal from the API.

Sampling flexibility. In principle, all that is required to obtain a Raman spectrum is to shine monochromatic laser light on a sample, and collect and spectrally analyse the resulting scattered light. There are many possible configurations to achieve this. Fibre optic delivery of the laser light and collection of the Raman light is particularly suitable for PAT applications. Figure 3 shows a schematic cut-away view of such a probe. Fibres several tens of metres long may be used to connect a central Raman instrument to remote parts of a production line. The sampling (distal) end of the fibre optic probe can be very compact and robust, facilitating easy integration with the production line. A number of commercial manufacturers supply a range of probes for assorted sampling arrangements.

Figure 3: A cut-away schematic of a fibre optic couple Raman probe system suitable for pharmaceutical PAT applications.

Commercially available Raman spectrometers designed for PAT-type applications are available with up to 60 fibre optic inputs allowing potentially 60 individual points on the production line to be monitored simultaneously. Use of a suitable optical 'relay' can potentially increase this degree of multiplexing further.

It is also worth noting that Raman spectra may readily be obtained through glass windows, or from samples inside sealed-glass containers. Optics are available with a working distance of typically a few centimetres (although the acquisition of Raman spectra at much greater stand off distances has also been demonstrated). The flexibility with which Raman spectroscopy can be implemented also increases the potential for integration with other analytical techniques to perform multiparametric measurements and, therefore, increases the amount of information obtained from the sample.

Challenges In Applying Raman Spectroscopy

The practical implementation of Raman spectroscopy as a PAT tool also presents a number of technical challenges, many of which stem from the weakness of the Raman effect.

Low signal level and long acquisition times. For dilute samples, simply obtaining a good signal-to-noise spectrum in a reasonable time may present a challenge; long signal accumulation times may be required and even then, in some circumstances, it may not be possible to acquire a satisfactory spectrum. This limitation can be overcome to some extent by employing a more powerful excitation source and a more sensitive detection technology, but both of these options can be costly. It should be noted, however, that in favourable circumstances excellent quality spectra may be readily acquired in fractions of a second.

Cost and safety of excitation source. The cost of these will depend to a large extent on their performance (e.g., the narrowness and stability of the wavelength emitted and the power output). The required performance will very much depend on the specific application. For example, many PAT-type applications may only require a relatively low specification laser. The cost and performance of the laser source is also highly dependant on the exciting wavelength required. The optimum wavelength will again be application specific. The Raman scattering efficiency is inversely proportional to the fourth power of the laser wavelength. A combination of this fact and the spectral response of silicon detectors generally used for Raman spectroscopy means that blue, green or red laser excitation is generally preferred. For many samples, however, problems with sample fluorescence dictates the use of near infrared excitation wavelengths (see later). Obviously, wherever lasers are employed eye safety is of primary concern and this will affect how the technique is integrated within the production line.

Cost of detectors. Sensitive, low-noise detection of the Raman scattered light is required with potential cost implications. As with the lasers discussed earlier, the specification of the detector should match that required by the application. Where plenty of signal is available and only modest spectral resolution is required, relatively inexpensive detectors may be employed. In more challenging applications a high specification cooled charged couple device (CCD) may be required. Such a device can represent a significant proportion of the overall cost of an instrument, although its suitability for simultaneous multiplex measurements (e.g., from 60 or more fibre optics from various parts of the plant) and its potential to improve sensitivity, resolution and speed of measurement may justify this cost for many applications.

Sample fluorescence and ambient light. A particular problem often encountered with 'real world' applications of Raman spectroscopy is that even relatively small amounts of auto-fluorescence from the sample can totally swamp the Raman signal. This problem can be reduced somewhat by the use of NIR excitation. Traditionally, this has been done by employing 1064 nm Nd:YAG lasers together with Fourier Transform Raman spectrometers. However, the relatively low sensitivity of such systems means consequent increases in required laser power and data accumulation times. In recent years, instruments employing ~780 nm excitation have become popular since at this wavelength fluorescence problems are reduced for many applications, yet sensitive multiplex detectors may still be used.

Alternatively the use of far ultraviolet (UV) excitation also reduces problems because of sample fluorescence. This is demonstrated in Figure 4 where four different excitation wavelengths ranging from the far UV to the NIR have been used to obtain spectra from the same API.

Figure 4: An example of the effect of different exciting wavelengths on the fluorescence background generated when recording Raman spectra from an API.

It can be seen that when 488 nm excitation is used a fluorescent background is generated which obscures much of the detail in the Raman spectrum. However, when 782 nm NIR excitation is used, the fluorescent background is reduced revealing a Raman spectrum rich in detail. When 244 nm far UV excitation is used the fluorescence background is once again absent and resonance effects enhance the intensity of certain Raman bands simplifying the spectra and aiding interpretation.

The optical design of in-line implementations also needs to ensure ambient light from the plant does not enter the Raman spectrometer since even low levels of stray light can obscure the Raman signal.

In Practice

This article has outlined some of the general strengths and weaknesses of Raman spectroscopy as a PAT tool. When considering this technique in a particular application these strengths and weaknesses must be taken into account to ensure that Raman spectroscopy is a viable technique. If this is the case then the detailed requirements of the application must be determined so that appropriate instrumentation is used.

Factors such as the nature of the material to be studied, its physical state, how the instrumentation will be integrated in the plant and, importantly, what information the spectroscopy is expected to provide play a crucial part in determining the required performance specification of the instrument. This performance specification will determine the complexity and price of the instrument.

Applying Raman spectroscopy in a manufacturing environment or a regulatory environment places additional demands on the instrumentation over and above those required for typical laboratory-based instrumentation. Other considerations include:

• Environmental sealing to protect the delicate optical components from dust and moisture in the plant and general robustness/resistance to shock and vibration.

• Ease of use by production line staff or automated operation.

• Long-term stability (e.g., optical alignment, spectral calibration and intensity calibration).

• Built-in calibration and validation routines (with reference to American Society of Testing Materials [ASTM] and the National Institute of Standards and Technology [NIST] standards, for example).

• Fibre optic coupling to multiple remote probes.

• Overall performance must be sufficient to satisfy regulatory requirements.

• Data handled and stored in suitable fashion (e.g., 21 CFR Part 11 compliance).

• Low maintenance, short downtime.

Raman spectroscopy is not yet widely used as an at-line, on-line or in-line analytical technique in pharmaceutical manufacturing. It has, however, been used for some time in a manufacturing support role in other non-regulated industries such as semi-conductors and synthetic polymers, and a range of proprietary instrumentation tailored towards particular PAT-type applications is available.

Increasing interest in applying the technique in an industrial environment has lead to the instrument manufacturers and standards organizations working towards developing suitable performance verification standards. Indeed the ASTM standards organization in the USA has a committee dedicated to Raman spectroscopy and has a standard for the spectral calibration of Raman instruments. NIST, also in the USA, currently supplies Raman intensity standards and are developing frequency standards.¹⁰

Conclusions

The practical features of Raman spectroscopy offer great potential for pharmaceutical PAT applications. At the time of writing, however, its actual use is confined to a few niche applications and, generally, it has yet to make a significant transition from the laboratory to the production environment. This has probably been because of, at least in part, the industry's reluctance to adopt the new technology because of a lack of familiarity, concerns about the willingness of the regulatory authorities to accept new technologies, and the traditionally relatively high cost and limited performance of Raman instrumentation.

This situation, however, is now changing. FDA's PAT initiative seeks to generate an environment where the implementation of 'new' analytical technologies is positively encouraged (with the obvious proviso that their implementation is scientifically sound). There is also the accompanying prospect of future changes in the regulatory environment that will require far greater application of analytical techniques throughout the manufacturing process. The implications for the pharmaceutical industry are potentially enormous.

Raman technology has advanced greatly in the last few years giving vastly improved performance, ease of use and general applicability to 'real' world problems. Encouraged by the opportunities that will arise as a result of the PAT initiative, the Raman instrument manufacturers and scientific community are now putting considerable effort to develop the technique as a PAT tool. Indeed, a range of proprietary Raman PAT instrumentation is already available. This increased interest has accelerated the development of Raman technology, increasing performance and reducing costs.

While not necessarily the best choice for every application in the pharmaceutical industry, a critical appraisal reveals that in certain circumstances Raman spectroscopy possesses a number of compelling practical advantages. This has been demonstrated to good effect, for example in the analysis of pharmaceutical polymorphs. But perhaps Raman spectroscopy's greatest single advantage is its flexibility of implementation and it is this that will surely lead to it becoming an integral tool in the pharmaceutical industry's analytical armoury.

References

1.

www.fda.gov/cder/OPS/PAT.htm

.

2. G. Clark, Pharm. Technol. Eur. 16(10), 34–36 (2004).

3. K.A. Bakeev, Pharm. Technol. Eur. 15(9), 27–32 (2003).

4. S. Folestad and J.Johansson, Eur. Pharm. Rev. 4, 36 (2003).

5. M.T. Islam et al., Pharm. Res. 21(10), 1844–1851 (2004).

6. I. Clegg and N. Everall, Eur. Pharm. Rev. 3, 56 (2003).

7. C. Starbuck et al., Cryst. Growth Des. 2, 515 (2002).

8. G. J. Vergote et al., Eur. J. Pharm. Sci. 21(4), 479 (2004).

9. C.V. Raman and K.S. Krishnan, Nature 121, 501 (1928).

10. ASTM Standard: E1840-96 Standard Guide for Raman Shift Standards for Spectrometer Calibration (2002).