Niklas Sandler

When developing active pharmaceutical ingredients (APIs) it is essential, first, to screen the most suitable polymorph for the formulation and, second, to be able to preserve its properties when the entire multistep manufacturing process is completed. However, it is well known that during processing, APIs and excipients can be subject to mechanical and/or thermal stress, and exposed to aqueous environments, which can cause transformations in the solid state.

  These process-induced transformations (PITs) cannot be neglected as they can lead to changes in processability, bioavailability and stability of the drug.

  The properties mentioned can change as a result of polymorphic interconversions, desolvation, alteration of the crystallinity or formation of hydrates. 

Solid state characterization with various complementary techniques is required to generate meaningful supporting data throughout the manufacturing process to assess concerns regarding PITs. Among a wide range of analytical techniques, Raman spectroscopy (RS) and near infrared (NIR) spectroscopy have attracted growing attention as they have great potential in process analysis and control. The recent developments in optics and probe design have made timely measurements easier in dynamic process conditions, such as moving powder beds and agitated suspensions, and the number of papers dealing with these issues is constantly growing. This paper lists some examples of PITs and how they can be monitored, with a focus on recent applications of Raman and NIR spectroscopy in different processing steps of pharmaceutical solids (Figure 1). 

The first unit operation during drug manufacture is often crystallization. To ensure the right solid form and avoid setbacks at later stages caused by undesirable solid forms, crystallization has to be performed in a highly controlled manner. As crystallization is a very complex process, several parameters have to be monitored to achieve a well-controlled process. 

Different kinds of PITs may occur during crystallization; for example, if the API being crystallized is enantiotropic (a system in which the stable polymorphic form is not the same above and below a specific transition temperature), temperature changes during the crystallization process can induce solid phase transformations. Transformations during crystallization may also occur because the first solid form to crystallize is often a metastable form that converts to a more stable form as the crystallization process proceeds. 

 used RS to simultaneously measure the desupersaturation profile (drug concentration in the solution) and polymorphic form of an enantiotropic API, flufenamic acid, during the crystallization process.

  Raman spectra were recorded in the study with a fibre-optic immersion probe in the crystallization vessel. Peaks characteristic of different phases of flufenamic acid (polymorphic forms I and III, and the solute) were identified in the spectra and quantitative information of the liquid and the solid phase was derived from the same spectra. Quantification of the solid phases was based on ratios of characteristic peak intensities. The Raman spectral analysis of drug concentration in the solution was in agreement with ultraviolet (UV) spectroscopic verification, and a temperature-induced interconversion between the polymorphic forms during the crystallization process was quantitatively monitored. 

 studied solid state transformations during crystallization and filtration processes with NIR spectroscopy.

  The study was performed with an API (SaC) appearing in two crystalline forms (SaC I and SaC II) and one amorphous form during processing. Crystallizations were performed in acetone solutions, and NIR spectra were recorded with a transreflectance probe immersed in the solution during crystallization. The downstream filtration process was monitored with a diffuse reflectance probe above the filter cake. Quantitative analysis was based on a multivariate method, principal component regression, using only a small number of calibration samples. Transformations from SaC I to SaC II (the stable form) were observed at different rates depending on various parameters (seeding, temperature and solvent composition). The transformations were successfully measured with NIR spectroscopy, and the coarse calibration with only few data points was found satisfactory for the measurement of polymorphic composition during processing. 

Milling or grinding is an important manufacturing process for solid dosage forms. A possible unwanted milling-induced phase transformation is the production of small levels of disorder or amorphous material in the milled crystalline material. This is usually found predominantly at the surfaces of crystals or powders, and these can lead to significant chemical and physical instability and change the processability of the formulation. However, a change from crystalline to amorphous can also be deliberately made by grinding when the aim is to produce an amorphous drug. 

 investigated the effect of morphology on milling induced changes.

  They showed that succinic acid with a plate-like morphology has more tendency to disorder than a needle-like morphology of the drug. It was also found that the type of mill influences the amount of change in crystallinity. Of the methods used, solution calorimetry, proved to be useful in quantifying small levels of amorphous material, which X-ray powder diffraction (XRPD) and disc-spinning calorimetry (DSC) could not detect. Mackin 

 used isothermal heat conduction microcalorimetry and dynamic vapour sorption to quantify low levels of amorphous material within a crystalline API after micronization.

  They found that amorphous material can be generated with extremely small changes in the operating conditions of the micronizer and were able to detect 0.5% levels of amorphous material with the above mentioned techniques.  

The aim is to monitor changes from crystalline to amorphous during processing, but so far very few examples of real-time measurements of phase transformations during milling exist in the cited literature. Only very recently Savolainen 

used on-line NIR and RS together with multivariate modelling tools to monitor process-induced changes from the crystalline to amorphous form for a crystalline starting material, lactose monohydrate.

  They concluded that spectroscopic techniques do not only provide a tool to monitor the changes in the amorphous content during milling, but they are also applicable as fast and uncomplicated methods to monitor amorphous content at different locations of the milling bowl. The variation in amorphous content in the bowl could also be used to describe the state of the process and to determine the end-point of milling. 

APIs are exposed to water in many pharmaceutical processes and the subsequent API–water interaction may induce phase transformations. As a typical example of this, wet granulation processes provide an ideal environment for the crystallization of new phases of the API. In fact, a number of PITs can occur during different process steps of wet granulation, which include wetting, mechanical stress and drying.  

Granulation or the manufacture of pellets using the extrusion–spheronization technique includes several process stages (blending of the dry mass, wet granulation of the mass, extrusion of the moist mass, rotation of the extrudate by spheronization and drying). The amount of wetting liquid in the powder mass in pelletization by extrusion–spheronization is relatively high compared with other granulation methods. Consequently, depending on the drug substance and excipients processed, solution-mediated polymorphic transformations are probable. Previously, Sandler 

have investigated phase transitions occurring in nitrofurantoin and theophylline formulations during pelletization by extrusion– spheronization.

  An at-line monitoring approach was used to increase the understanding of the solid-state behaviour of the APIs during the process. RS, NIR spectroscopy and XRPD were used in the characterization of polymorphic changes during the process. Samples were collected at the end of each processing stage (blending, granulation, extrusion, spheronization and drying). Water induced a hydrate formation in both model formulations during processing, and NIR spectroscopy gave valuable real-time data about the state of water in the system, but it was not able to detect the hydrate formation in the theophylline and nitrofurantoin formulations during the granulation, extrusion and spheronization stages because of saturation of the water signal. Raman and XRPD measurements confirmed the expected pseudopolymorphic changes of the APIs in the wet process stages. However, a relatively low Raman signal of theophylline (10% [w/w] in the formulation) complicated the interpretation of spectra. 

RS has been shown to be very helpful in controlling wet granulation processes. Jørgensen 

successfully used (off-line) charged coupled device (CCD) RS and NIR spectroscopy to investigate solid phase transitions of caffeine and theophylline during wet granulation.

  They concluded that the symmetric vibrations in the drug molecules could be effectively detected and analysed with RS. As Raman bands were much narrower than those in the NIR spectra (which has dominating OH bands) it was easier to observe the changes in the Raman spectra. Wikström 

successfully interfaced in-line RS to a high-shear granulation process and also monitored solid phase transformations of anhydrous theophylline.

  They also studied effects of several processing parameters, namely the mixing speed and monohydrate seeding. It was found that the mixing speed had the greatest effect on the phase transformation, showing that an increase in mixing speed shortened the onset time of conversion and increased the rate of transformation to the monohydrate. However, the seeding with monohydrate or a change in the incorporation of the binder to the system did have an effect on the transformation profile. 

After wet granulation or pelletization, the granules/pellets are usually dried to a certain moisture content. The drying can be performed, for example, in an oven on a tray or in a fluidized bed dryer. Either way, heat is brought to the system to evaporate the excess moisture from the particles, but in fluidized bed drying moisture removal is made more effective by blowing an air stream through the fluidized particles. The drying processes can cause phase transformations, in particular dehydration of hydrate forms. 

used in-line NIR and RS to quantitatively monitor the solid phase of an API during fluidized bed drying.

  Theophylline monohydrate granules were fluidized in a microscale fluid bed dryer and NIR and Raman spectra were recorded through a quartz sight window in the dryer wall (Figure 2). The quantification of the solid phase was based on a multivariate method, partial least squares regression. During fluidization, theophylline monohydrate lost its water from crystallization. The loss of water was clearly visible in the NIR spectra and quantification of monohydrate content was accurate. However, the real-time Raman spectra of the fluidized granules exhibited apparent features not present in Raman spectra of pure monohydrate or anhydrate form. These features were linked to the intermediate metastable anhydrate form upon dehydration of theophylline monohydrate to theophylline anhydrate. It was concluded that as a method sensitive to water, NIR spectroscopy detects the dehydration, and as a complementary method sensitive to alterations in the crystal structure, RS detects the anhydrate formation. 

Even though dissolution is not a manufacturing process, similar transformations can occur during dissolution of solid material. When a dosage form or an API is introduced to the gastrointestinal tract, or dissolution media mimicking it, a transformation from a metastable to a more stable form with lower solubility and bioavailability is possible. Such transformations have traditionally been detected as dissolution rate changes during intrinsic dissolution rate tests. However, the solid phase transformations during dissolution involve several overlapping factors affecting the dissolution rate that can hinder the detection of the transformation. For example, changes in the crystal habit and specific surface area as well as the concentration of the dissolution medium have an effect on the transformation kinetics and the overall dissolution rate. 

combined intrinsic dissolution rate testing with simultaneous 

 solid phase analysis of the dissolving sample to directly measure solid phase transformations during dissolution testing.

  The solid phase was quantitatively analysed by RS through a quartz window and the quantification was performed using ratios of characteristic peak intensities of each solid form in the Raman spectra. Dissolution behaviour of two hydrate-forming APIs (theophylline anhydrate and nitrofurantoin anhydrate) was investigated. Both APIs transformed to monohydrate forms during dissolution, but their dissolution behaviours were very different. Figure 3 shows results of an intrinsic dissolution rate test of theophylline anhydrate and, for comparison, theophylline monohydrate. The dissolution rate of theophylline anhydrate decreased clearly as the solid phase transformation to monohydrate form proceeded. The transformation was also very fast (~7 min). Dissolution rate changes of nitrofurantoin were more complicated and the transformation took much longer (~130 min). The incorporation of real time solid phase analysis to dissolution testing enables extensive knowledge of dissolution behaviour of drugs and is an improvement to prevailing practice, as it provides accurate information of the whole dissolution process, and not only the drug concentration in the dissolution medium. 

This overview shows that process-induced transformations are likely to occur during manufacture and end-product analysis of solid dosage forms. RS and NIR spectroscopy often provide suitable solutions for process monitoring, but the choice of the process analytical technique has to be made carefully to obtain meaningful data from the process. Developments in analytical methods and the immense activity in the field will hopefully result in a large amount of new examples of PIT monitoring during processing of solid dosage forms in the near future.  

is a postdoctoral researcher at the Division of Pharmaceutical Technology, University of Helsinki (Finland). His research area is solid phase analysis, with a special interest on the development of spectroscopic tools for real-time process analysis. 

is a senior scientist at AstraZeneca R&D (UK) in the area of material science supporting product development projects. He has experience in applications of image analysis in solid dosage forms process monitoring and solid state characterization with spectroscopic tools. 

10. H. Wikström, P.J. Marsac and L.S. Taylor, 

Articles

When a dosage form or an API is introduced to the gastrointestinal tract, or dissolution media mimicking it, a transformation from a metastable to a more stable form with lower solubility and bioavailability is possible.

Monitoring of phase transformations during processing of solid dosage forms

Different spectroscopic techniques are widely used to help understand the physical solid-state characteristics of drug substances and formulations — and Raman spectroscopy has become a commonly used technique that possesses many advantages over other analytical techniques.  

The principles and applications of Raman spectroscopy have been recently well-addressed in this journal.

  Benefits of the technique are, for instance, that sample preparation is unnecessary or minimal; data acquisition is rapid and nondestructive; and measurements in aqueous environments are possible because water is a weak Raman scatterer (Figure 1). These benefits also allow bulk and final products to be tested directly in their packaging, such as glass bottles.

Figure 1 Examples of Raman spectra of Indomethacin (a) and lactose monohydrate (b) measured from a powder and suspension. For comparison, a spectra of only water is also shown.			

Numerous papers in literature emphasize that Raman spectroscopy is a very attractive characterization tool, specifically because it enables measurements in presence of water, but, surprisingly, few studies have applied Raman spectroscopy in an aqueous environment. This paper focuses on recent uses of Raman spectroscopy to characterize pharmaceutical materials and processes in water. 

Probing critical solid-state processing steps and tablet dissolution 

One of the earliest studies in a pharmaceutical setting demonstrating the use of Raman spectroscopy in crystallization was performed by Findlay and Bugay.

  They studied crystallization dynamics by measuring supersaturated menthol solutions with variable temperature (VT) Fourier Transform (FT)-Raman spectroscopy. Elkordy 

				 reported the measurement of crystallized and spray-dried lysozyme, a model protein, by FT-Raman spectroscopy in 8% (w/v) solutions. They used a relatively high concentration to detect well-resolved Raman bands. After a water background had been subtracted, aqueous state FT-Raman spectra revealed protein conformation changes.  

Raman spectroscopy has been particularly valuable for controlling wet granulation processes. Jørgensen 

				 successfully employed charged coupled device (CCD) Raman and near infrared (NIR) spectroscopy to monitor polymorphic transitions of caffeine and theophylline during wet granulation. They concluded that the symmetric vibrations in the drug molecules could be readily detected and analysed with Raman spectroscopy. As Raman bands were much narrower than those in the NIR spectra (which has dominating OH bands) it was easier to observe the changes in the Raman spectra.  

				 successfully interfaced in-line Raman spectroscopy to a high-shear granulation process and also monitored pseudopolymorphic conversions of anhydrous theophylline. In a pelletization study, Raman spectroscopy, NIR spectroscopy and X-ray powder diffraction (XRPD) were used to characterize process-induced polymorphic changes.

  Samples were collected at the end of each processing stage (blending, granulation, extrusion, spheronization and drying). Raman and XRPD measurements confirmed the expected pseudopolymorphic changes of the active pharmaceutical ingredients (APIs) in the wet process stages. However, a relatively low Raman signal of theophylline (5% [w/w] in the formulation) complicated the interpretation.  

				 Raman spectroscopy was elegantly used to measure the solid state properties 

 during intrinsic dissolution measurements. They showed that the solid-state transformation from anhydrate to hydrate can be quantitatively measured for both theophylline and nitrofurantoin tablets using a Raman probe (Figure 2).  

Figure 2 A schematic of the experimental setup used in a solid-state dissolution study. (Figure modified from reference 9.)			

They concluded that by combining dissolved drug concentration with solid-state measurements, a deeper understanding of the solvent-mediated phase transformation phenomena during dissolution is achieved. 

Controlling the properties of emulsions and suspensions

				 demonstrated the use of Raman to monitor a bench-scale formulation process of topical formulations. They heated the aqueous gel and mixed it with an oil phase to produce an emulsion, showing that a Raman shift of a peak at 925 cm

  was linked to an ionic interaction between Carbopol and triethanol amine. The height of the 925 cm

  increased with greater neutralization, as measured by the pH.  

The authors, therefore, concluded that these peak height ratios can be used to control the amount of neutralizer used during manufacture. Raman spectroscopy in combination with stepwise linear regression has also been successfully used to confirm the polymorphic stability of benzimidazole in a complex suspension emulsion formulation.

  The formulation consisted of active drug suspended in oil droplets, which are emulsified in an aqueous matrix. No polymorphic transition was noted during the duration of the study.  

However, a small sampling area within the heterogenous matrix caused some measurement variability. Thus, when measuring heterogenous samples multiple spectra should be recorded with a sufficiently large sample area to minimize variable and unrepresentative sample analysis. 

The polymorphic composition of L-glutamic acid crystals has been monitored during batch cooling of saturated solutions using Raman spectroscopy 

  By using peak height ratios, the α and β forms were quantified and at 25 °C in aqueous solution, the α form dominated. However, a solvent mediated transformation to the β polymorph followed, with the conversion accelerated at higher temperatures. This 

 quantification allows the optimization of both crystallization kinetics and polymorphic purity.  

Raman spectroscopy offers an attractive possibility for direct and nondestructive quantitative measurements in an aqueous environment. Raman spectroscopy has been successfully used to determine the amount of medroxyprogesterone acetate in an aqueous suspension.

  Besides the fact that water does not interfere, an advantage of Raman spectroscopy is that it allows direct measurement of suspended drug concentrations, without the need to know the exact volume or density of the sample suspension, as would be the case with HPLC. Thus compared with HPLC, Raman spectroscopy provided a more reliable quantitation method for the amount of medroxyprogesterone acetate in a suspension.  

Raman spectroscopy can also be used to detect changes in dissolved compounds. Li 

				 used FT-Raman spectroscopy to examine the acid-base characteristics of various citrate buffer systems. The results indicated that the ratio of ionized and un-ionised species remained the same in the solution, the frozen solution and the amorphous state. Quantitation was based on the FT-Raman peak intensity ratio of the -COO to -COOH carbonyl stretching vibrations at 1400 cm

, respectively. The signal to noise ratio was largest in the frozen samples, which was probably because of the shorter acquisition times that were necessary to prevent the samples melting. 

Studying water sorption and hydration phenomena 

Raman spectroscopy has been widely used to detect changes in the solid-state of a drug because of water sorption into the material. Airaksinen 

				 used Raman spectroscopy to determine the effect of excipients on hydrate formation of theophylline during wet granulation. Despite the excipients in the formulation, Raman spectroscopy could still detect the theophylline anhydrate to monohydrate conversion without excipient interference.  

A blue shift of the O=C-N bending band at 550 cm

  and changes in the C=O stretch region at about 1700–1650 cm

  were used to detect the hydrate formation. Raman spectroscopy has also been used in-line for quantitation of changes in the amorphous content of spray-dried amorphous lactose and trehalose that were stored in high relative humidity (RH [Figure 3]).

  The quantitation was based on multivariate analysis. Both of the amorphous sugars recrystallized during storage at 85% RH. Amorphous materials are very hygroscopic and absorb moisture when stored in high RH. Unlike with NIR, the absorbed water did not interfere with the Raman spectra and thus Raman spectroscopy was more suitable than NIR spectroscopy for the crystallinity quantitation.  

Figure 3 Following the conversion of amorphoustrehalose to trehalose dihydrate at high relative humidity (85% RH) using in-line Raman spectroscopy; (a) illustrates the Raman spectra and (b) shows the increase in crystallinity determined by a PLS model. (Data modified from reference 16.)			

Raman spectroscopy has been used to study the gelatinization and liquefaction (hydrolysis) process of starch.

  An aqueous starch suspension was heated above the gelatinization temperature and spectra recorded in-line. During gelatinization, the uptake of water into the crystalline starch caused inter- and intramolecular hydrogen bond breakage and new hydrogen bonds were formed with water molecules.  

This could be observed in the Raman spectra because the bands at 1633 cm

  increased while the others decreased. The band at 3213 cm

  was attributed to hydrogen bonding with water and the band at 1633 cm

  to decreased starch crystallinity. After gelatinization, the liquefaction was initiated by addition of α-amylase. Significant changes in the spectra were used to monitor the liquefaction. The formation of free hydroxyl groups because of the hydrolysis reaction was noted by the increase of the bands in the region of 1057 cm

In addition to solid-state transitions in aqueous environments, Raman spectroscopy can be used to study the molecular interaction between water and different substances. Raman spectroscopy has been used to determine the changes in the strength of hydrogen bonding between water and different pharmaceutical polymers.

  Raman spectroscopy could be used to study the hydrogen bonding as the studied polymers, PVP, PVAc and PVP/VA were unable to form intra- or intermolecular hydrogen bonds because of the lack of acidic protons. Therefore, any changes in the spectra of these polymers, when in solution, were as a result of hydrogen bonding with water molecules. However, the spectral changes were because of not only the changes in the hydrogen bonding, but also the plasticizing effect of water, which increased the molecular mobility. The increased molecular mobility affected the C-H and C-C vibrations. These groups form the backbone of the polymers. 

In studies on nanostructured lipid carriers it has been shown that Raman spectra of lipids are sensitive to conformational, packing and dynamical changes involving hydrocarbon chains.

  Therefore, this technique can be applied to provide more detailed information on the microstructure of colloidal lipid dispersions such as nanoemulsions (NE), solid lipid nanoparticles (SLN) and nanostructured lipid carriers (NLC). Investigations on SLN based on glycerol behenate and NLC (mixtures of glycerol behenate and medium chain triglcyceride [MCT]) resulted in the conclusion that MCT loading did not lead to changes in order or packing behaviour of the lipid chains, meaning that oil incorporation could not be confirmed.

When investigating the same systems, but using cetyl palmitate as solid lipid and MCT as oil component, Raman spectroscopy revealed that the NE was characterized by random coiled chains, while SLN and NLC formulations clearly showed sharp bands at 1128 cm

  indicating the high conformational order of the acyl chains.  

To determine whether the NLC-formulation spectrum is a simple mixture of spectra of SLN-formulation and nanoemulsion, a spectrum was calculated as a linear combination of 0.9 times the pure SLN spectrum plus 0.1 times the pure nanoemulsion spectrum. This calculated spectrum was compared with the NLC spectrum. The main peak positions did not change and subtraction of the normalized spectra revealed no major differences between the experimental and calculated spectra, suggesting that the SLN in the NLC particles is completely unperturbed by the presence of the oil.

This paper has demonstrated the versatile use of Raman spectroscopy in pharmaceutical analysis. The technique offers a major advantage that spectra can be obtained noninvasively in the presence of water. However, the full potential of Raman spectroscopy analysis in aqueous environments is still to be realized, and it is envisaged that the number of application and studies in this area will grow rapidly in the near future.  

is a senior scientist at AstraZeneca R&D (UK).			

 is a research scientist at the Division of Pharmaceutical Technology, Faculty of Pharmacy, University of Helsinki (Finland). 			

 is a research scientist at the School of Pharmacy, University of Otago (New Zealand). 			

is chief research scientist at the Division of Pharmaceutical Technology, Faculty of Pharmacy, University of Helsinki (Finland). 			

is a Professor at the School of Pharmacy, University of Otago (New Zealand).			

5. A.A. Elkordy, R.T. Forbes and B.W. Barry, 

7. H. Wikström, P.J. Marsac and L.S. Taylor, 

Raman spectroscopy has become a commonly used technique for physicochemical analysis that possesses many advantages over other analytical techniques. It is a very attractive characterization tool, not least because it enables measurements in water. However, very few examples of its application in an aqueous environment exist in literature. This paper provides some recent applications of Raman spectroscopy in pharmaceutical material and process characterization when water is present.