Thomas Rades

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, 

Articles

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.

Applications of Raman spectroscopy in aqueous environments

Since initial research on terahertz radiation began to gain momentum nearly 20 years ago, this technology has evolved from being a pure research tool to one that is used for pharmaceutical applications. More specifically, techniques using pulsed terahertz radiation such as terahertz pulsed spectroscopy (TPS) and terahertz pulsed imaging (TPI) are now increasingly employed to analyse solid dosage forms. So far, TPS has been predominantly used to acquire physical information of solid-state properties of active pharmaceutical ingredients (APIs), particularly in the field of polymorph recognition, characterization and quantification. 

TPI can be used to nondestructively analyse tablet matrix physicochemical composition, coatings and buried interface properties in solid oral dosage forms because of its pulsed and coherent nature of the radiation together with its frequency-dependent unique penetration characteristics. In addition, both two- and three-dimensional (2D, 3D) imaging techniques are currently being developed.

The term 'terahertz radiation' corresponds to the far-infrared region of the electromagnetic spectrum (3–333 cm

  or 100 GHz–10 THz), or parts thereof, residing between microwave and mid-infrared radiation.

  Many modern TPS devices typically utilize the frequency range from 0.1–4 THz (3.3–130 cm

). Rather than representing information originating from intramolecular vibrations in the mid-infrared region of the spectrum, terahertz radiation induces intermolecular noncovalent bond vibrations and translations in solids. As terahertz radiation directly probes interactions between molecules, it is intrinsically sensitive to changes in crystalline structure (crystalline phonon vibrations).

  Raman and mid-infrared spectroscopy, however, are classic intramolecular spectroscopic techniques that are most sensitive to chemical information.

Both TPS and TPI are built around the same core terahertz technology. The source of radiation is a photoconductive semiconductor antenna that generates broadband pulses of terahertz light by injecting charge carriers into the substrate with a femtosecond laser. After transmission through the sample material, the radiation is detected as a time domain waveform of the terahertz pulse in a reverse process using a time-gated probe beam for the detection process (Figure 1).  

Figure 1 Setup of the TPS system (left) and the TPI system (right).			

In contrast to conventional far-infrared spectroscopy, terahertz radiation is unique in that its generation by the photoconductive antenna is coherent and pulsed.

  Moreover, compared with most other spectroscopic techniques, in TPS and TPI the photo-induced electrical field is measured directly rather than just its signal intensity. Using the information on the amplitude and phase of the signal both the absorption coefficient and the refractive index of the sample can be measured directly.

After performing a Fourier transformation of both sample and reference waveform into the frequency domain, the terahertz absorbance spectrum is calculated. For the structural image generation in TPI, the measured time delay originating from reflections at interfaces within the sample in the time domain waveform is used to resolve the depth dimension. Spatial information (horizontal and vertical dimensions) is obtained by fully automated point mapping of the sample. This TPI waveform data for each pixel can also be Fourier transformed to yield a fourth dimension with spectral information of the sample. Using this approach, both structural information and chemical composition in the sample can be described using TPI.

Polymorphism is a solid-state phenomenon where, though possessing the same chemical structure, pharmaceutical solids may exhibit different arrangements of their molecules in the unit cell, leading to subsequent physicochemical differences for these modifications for properties such as bioavailability, processibility and stability. The prevalence of polymorphism in pharmaceutical solids and the paramount importance in its identification are consistently highlighted in the literature.

 TPS has been used to distinguish amorphous, crystalline, hydrate, solvate and liquid crystalline solid states in a number of drug molecules. It is now commonly accepted in the literature that terahertz spectra provide sufficient information to distinguish subtle differences in condensed matter properties. 

Examples of drugs studied by TPS include different polymorphic modifications of carbamazepine, enalapril maleate, ranitidine hydrochloride and sulfathiazole.

  All compounds showed significant spectral peak differences between the polymorphs in the terahertz region (Figure 2). Amorphous samples of glucose, fructose, sucrose and indomethacin all exhibit no apparent peaks;

  instead, they show smooth, featureless and increasing terahertz absorption spectra. This is largely because of diminished long-range order in the amorphous state.  

Figure 2 Terahertz spectra of the five polymorphic forms of sulfathiazole.			

As terahertz spectra are reflecting intermolecular vibrations, TPS is an excellent tool for differentiating amorphous systems from their respective crystalline counterparts.

  Super-cooled liquid crystals are also solid-state systems with limited long-range order and it is speculated that this is the main reason for their featureless terahertz spectrum, which is quite similar to that of amorphous systems. Fenoprofen calcium in its quench cooled liquid crystalline state has been used to demonstrate the ability of TPS to differentiate liquid crystalline from the crystalline form. However, further work must be performed to confirm the main causes of the diffused terahertz spectrum.

As in many other polymorphic characterization studies, powder samples of the above drugs were used for the studies, raising the question whether TPS is also able to identify polymorphs in commercial tablets. In 2002, Taday 

 used commercial tablets of ranitidine hydrochloride (Zantac and apo-ranitidine) to demonstrate the capability of TPS for polymorph recognition. Pronounced differences were found in the terahertz transmission spectra of the different tablet samples and it was clearly shown that Zantac contained polymorph II while apo-ranitidine consisted of polymorph I.

TPS is not only a very valuable tool in polymorph recognition, it is also effective for solid-state quantification. The clear differences generated between different crystalline forms in the terahertz spectra can be exploited for quantitative analysis. Binary mixtures of a number of different forms of polymorphic drug materials (carbamazepine and enalapril maleate), crystalline in amorphous form (indomethacin) and crystalline in liquid crystalline form (fenoprofen calcium) were quantified by Strachan 

 In combination with multivariate analysis using partial least squares (PLS) algorithm, a limit of detection of one polymorphic form in the other as low as 1.80% and crystallinity limit of detection as low as 1.05% were observed (Figure 3). Furthermore, paracetamol has been quantified in the presence of common excipients using the same technique and has shown promising results.

  In addition to these proof-of-principle studies, work is in progress on an exhaustive validation study investigating the quantitative capabilities of TPS.  

Figure 3 (a) Terahertz spectra of binary mixtures of form I and III of carbamazepine; (b) PLS regression used for the quantification of low levels of carbamazepine in binary mixtures of form I in form III.			

As illustrated previously, TPS is a nondestructive technique that requires minimal sample preparation and operates at long wavelength and low power densities, minimizing any solid phase changes or thermal strains in the sample. Polymorphic recognition by TPS is very fast with acquisition times as low as 30 ms. Because most plastic packaging materials used for blistering pharmaceutical solid dosage forms are transparent within the terahertz range, this allows for possible applications as an in-line process analytical tool (PAT).

Applications of terahertz imaging in pharmaceutical sciences have only been reported in the last year.

  While Fourier transform infrared and Raman spectroscopy imaging techniques are extensively developed and are now widely employed for chemical imaging in the near- and mid-infrared regions, the far-infrared region has not been exploited for imaging. This is largely because of the difficult detection process. TPI, however, fills the gap of far-infrared imaging, capable of operating at room temperature without the need for cryogenic cooling.

2D chemical maps are acquired from 3D data sets with two spatial dimensions (vertical and horizontal) and the spectral frequency as the third dimension. This imaging technique can either be performed in transmission or reflection mode depending on the sample investigated and the information needed.

  When operating in reflection mode, rapid scans of the sample can be achieved (20 pixels/s) with a spectral resolution of 1 cm

. Pellets featuring spatially resolved domains of lactose, sucrose and mixtures of the two were successfully scanned. Each measurement could map out the spatial distribution of the respective chemical component.

  The spatial resolution for such mapping experiments is diffraction limited and in the order of a few hundred micrometers for a broadband TPI system. 

While operated in reflection mode, TPI lends itself towards on-line analytical applications because of its precise, rapid scan capability; it is also possible to operate the TPI in transmission mode. Pharmaceutical packaging material such as plastic or paper are transparent at terahertz frequencies, thus a terahertz wave is able to detect and subsequently construct chemical images after penetrating through such materials.

  Terahertz 2D images of lactose, aspirin, sucrose and tartaric acid pellets were acquired through layers of paper cover,

  while illicit drugs were identified through polyethylene bags.

Unlike Raman and infrared spectroscopy, where chemical imaging is restricted to spectral information from just below or on the sample surface, typical pulses of terahertz radiation used in TPI have a solid-state penetration depth of up to 3 mm depending on the material. Figure 4 demonstrates the capability of TPI in 3D imaging; here, a trilayered tablet 3D model is reconstructed from the terahertz signal. Various spectroscopic and imaging techniques have been employed to investigate tablet coating thickness and uniformity. Techniques such as confocal laser scanning microscopy, infrared spectroscopy, X-ray microtomography and electron paramagnetic resonance all show promising results.

  However, at least one of the following analytical constraints is limiting their wider application: long data acquisition time, destructive measuring method or the fact that the generated images are limited to information from the surface of the tablet coating.  

Figure 4 (a) and (c) 3D models reconstructed from the terahertz signal of a trilayered tablet (b) highlighting the interfaces between the three layers.			

TPI is a nondestructive, high throughput (the acquisition of a single waveform takes less than 20 ms), 3D imaging technique.

  It can be used to acquire single point measurements for very fast tablet coating thickness measurements at one point, or using fully automated scans over the whole surface of a solid dosage form to produce coating uniformity maps. For measurements of solid oral dosage forms terahertz images are recorded in reflection mode. Here, the difference in refractive indices at the air/coating interface and the coating/core interface as a function of time is recorded. The time delay between the signals from the different interfaces is used to calculate the thickness of the coating layer(s). Using two different sugar-coated brands of ibuprofen tablets as illustrations, Fitzgerald 

 were able to convey the potential of TPI for investigating coating thickness, using changes in the refractive indices at interfaces within the tablet matrix. Furthermore, the coating uniformity can be visualized by virtual cross sections through the tablet. These scans revealed that the two different brands of ibuprofen exhibit very different coating structures and coating thickness (Figure 5).

Figure 5 Coating layer thickness (a) and quality (b) of an enteric coated tablet.			

In addition to the structural images obtained by sample mapping, a spectral dimension can be introduced to the existing dataset. Thus a 3D terahertz image is constructed from a 4D (vertical, horizontal, axial and spectral frequency) dataset. Consequently, not only chemical information, but also the precise spatial location of each of the chemicals, in a heterogeneous sample matrix such as most pharmaceutical solid dosage forms is obtained. For example, a polyethylene tablet containing lactose and tartaric acid 'impurities' was imaged with TPI, employing the 3D imaging technique. As a result, the exact location and the chemical composition of these 'impurities' were mapped out.

  However, measurements such as this are still far from routine and more research is necessary before such images are readily available. 

Compared with Raman and infrared spectroscopy, terahertz technology is an exciting and novel technique with a remarkable capability for the physical characterization of solid dosage forms. Spectroscopy and both chemical and structural imaging are possible. With high throughput and nondestructive characteristics, terahertz technology holds potential for application as a solid-state PAT tool. Current terahertz research in the area of solid-state applications is very active and new applications of the technology are still being explored. Emerging results of current applications will pave the way for its future use in pharmaceuticals.

is chief research scientist in the Physical Pharmacy, Drug Discovery and Development Technology Centre, Faculty of Pharmacy, University of Helsinki (Finland).  

are PhD students at the School of Pharmacy, University of Otago (New Zealand) and Cavedish Laboratory, University of Cambridge (UK). 

is a professor in the department of pharmaceutics and analytical chemistry at the Danish University of Pharmaceutical Sciences (Denmark).			

holds a chair of pharmaceutical sciences at New Zealand's National School of Pharmacy and 

is associate professor at the department of chemistry, both at the University of Otago (New Zealand). 

1. M.C. Beard, G.M. Turner and C.A. Schmuttenmaer, 

Philosophical Transactions of the Royal Society of London, Series A: Mathematical, Physical and Engineering Sciences, 

 (Applications of Spectroscopy to Biomedical Problems), 255–263 (2003). 

Innovations in Pharmaceutical Technology, 

 3D Chemical Mapping Using Terahertz Pulsed Imaging, Proceedings of SPIE, 5727(Terahertz and Gigahertz Electronics and Photonics IV), 24–31 (2005). 

This article investigates pharmaceutical applications of terahertz technology, specifically using techniques for solid dosage form analysis such as pulsed spectroscopy (to generate physical information and detect API changes) and pulsed imaging (to locate formulation impurities, and regulate tablet coating quality and thickness).

Terahertz applications for the analysis of solid dosage forms

Poorly water-soluble compounds with dissolution rate-limited low oral bioavailability present one of the major challenges in pharmaceutical development. There are many ways to increase the aqueous solubility of such compounds, including micronization, salt formation and formulation of the drug as a solid dispersion. For many compounds, however, decreasing the particle size may not lead to a significant or adequate increase in bioavailability. Salt formation may also be problematic, particularly with neutral compounds and weak acids. Solid dispersions, in which the drug may be present in the amorphous state, offer an attractive means of increasing the solubility and, therefore, potentially increasing the oral bioavailability of these 'problem' compounds (Figure 1). 	

The application of solid dispersions for increasing drug bioavailability is by no means a new field of pharmaceutical research. In their early paper on the use of solid dispersions, Chiou and Riegelman1 observed that: "It is believed that this relatively new field of pharmaceutical techniques and principles will play an important role in increasing dissolution, absorption and therapeutic efficacy of drugs in future dosage forms."	

Since then, a great deal of research has been done on solid dispersions but only a few have been marketed; for example, griseofulvin in polyethylene glycol (Grispeg, Sandoz) and nabilone in polyvinylpyrrolidone (Cesamet, Lilly). The difficulties associated with these systems limiting the production of usable dosage forms have been well documented.2-5 This paper introduces the application of melt extrusion for preparing solid dispersions and discusses how best to select drugs and excipients suitable for this process.	

Figure 1:  Drug bioavailability in dogs (crystalline versus amorphous).	

A solid dispersion is a homogeneous mixture of one or more active ingredients in a pharmacologically inert matrix (carrier) in the solid state. Table I shows the various classes of solid dispersion that can be produced and the type and number of solid phases present.				

Determination of the solid dispersion type and number of phases present is crucial to understanding the product. In this review, the discussion will be restricted to glass solutions, which are potentially the most advantageous for increasing oral bioavailability. 	

There are three main reasons why formulation of a drug as a glass solution can lead to an increase in the solubility of the active component:	

unlike the crystalline state there is no lattice energy to overcome in the amorphous state, which reduces the thermodynamic barrier to dissolution

the particle size of the components is maximally reduced to the molecular level

the intimate presence of a hydrophilic carrier leads to an increase in the wetting of the active, and possibly an increase in the solubility of the active in the diffusion layer surrounding the dissolving particle. 

The preparation of solid dispersions, particularly on a scale suitable for production has been discussed in a number of recent reviews.5,6 There are three general approaches to preparing solid dispersions. The first of these is solvent evaporation (including spray- and freeze drying), and this has been the preferred technique for use with compounds that decompose upon melting. Consequently, the solvent method allows the routine use of amorphous polymers such as polyvinylpyrrolidone (PVP) and hydroxymethylcellulose, which either do not melt, or degrade before melting. There are some disadvantages to solvent-evaporation that limit its application. Finding a suitable co-solvent with negligible toxicity for a hydrophobic drug and hydrophilic excipient can be problematic; large quantities of solvent often have to be used, which increases both cost and environmental impact. Achieving adequate drying is also problematic, as any residual solvent can plasticize dispersions, altering their physicochemical properties,1,2 and may have toxicity issues.	

Table I:  Classification of solid dispersions (A 5 amorphous, C 5 crystalline).

A second approach to manufacture is by mechanical activation such as co-milling. This can lead to disruption of the crystalline lattice and may induce local melting and formation of amorphous material.7,8 Disadvantages to this approach are ensuring complete conversion to the amorphous state and developing a reproducible/robust methodology.5 	

The third approach to solid dispersion manufacture is co-melting. 	

The co-melt method has not been particularly successful in preparing glass solutions, as the majority of products formed have contained the drug in a crystalline state (Table II). The formation of amorphous dispersions is favoured by rapid cooling, which can entrap the solute (drug) molecules in the solvent (excipient) matrix by almost instantaneous solidification.8–10 Degradation of components upon heating and miscibility in the melt are the limiting factors in many systems.		

Table II:  Drugs prepared as solid dispersions by the melt method (A 5 amorphous dispersion, C 5 drug present in crystalline form).	

Melt extrusion. Melt extrusion has been used in the polymer industry since the turn of the century and the technique is employed in the manufacture of 50% of all plastic goods.11 Extruders are available in different configurations and designs for various processing applications. For melt extrusion within the pharmaceutical industry, screw extruders are predominately used because they were designed to	

incorporate and homogenize additives without exceeding degradation temperatures

generate high shear stress for dispersion of components/additives

homogenize two or more materials with different melt viscosities

provide a uniform shear stress and heat history

allow precise control of the mixing process.12

The recent resurgence of interest in melt extrusion in the pharmaceutical industry results from its potential advantages in solid dispersion production. During melt extrusion, the drug is incorporated into a polymer by melting or plasticizing both the drug and excipient using either one or two screws inside a heated barrel. The molten material is then cooled, typically on a chilled stainless steel conveyor belt. This approach may lead to formation of a glass solution of drug and polymer. For example, GrÃ¼nhagen13 commented that following the processing of several dozen drugs using melt extrusion, approximately one third were prepared as glass solutions. 	

Equation 1 and Figure 2:  DSC/TGA of piroxicam, showing onset of degradation before melting. The upper trace shows TGA and degradation onset at 198 ÃÂC and the lower trace shows the DSC of the drug melting at 205 ÃÂC.

Melt extrusion is essentially a combination of melting and mechanical preparation methods, but has several advantages compared with both. Melt extrusion is cost-effective, continuous and reproducible - and allows for a high-throughput.11 Drug degradation is decreased compared with co-melt methods because of the increased input of mechanical energy and the use of a closed barrel that can be flushed with nitrogen to decrease oxidation reactions. Melt extrusion does not require solvents and is not, therefore, associated with the environmental, toxicological and financial implications of the use of large solvent volumes. Importantly, melt extrusion for the formation of glass solutions allows the use of thermoplastic polymers that do not melt. PVP is an example of a thermoplastic polymer, as its viscosity decreases with increasing temperature and the presence of plasticizers. Polymers such as PVP confer increased physical and chemical stability to amorphous systems. 	

The major disadvantage of the technology has been that even when a lab-scale extruder is used, a large quantity of drug material is required, which may not be available at an early drug development stage. This problem has been partially resolved with the development of small-scale, benchtop extruders such as that produced by ThermoHaake,14 which can process approximately 5-10 g of input material. However, even gram quantities of drug still remain problematic in early development, particularly as companies seek to incorporate potential formulation strategies in candidate selection. The remainder of this article addresses two key questions:	

1.	How should compounds that are suitable for melt extrusion be selected at an early stage of drug development using only milligram quantities of drug substance?	

2.	How should a suitable excipient matrix for melt extrusion with the compound be selected to ensure a glass solution can be manufactured?	

For a drug to be suitable for melt extrusion, the glass solution produced must show an increase in dissolution rate compared with the crystalline drug and the product must be physically and chemically stable. Glass solutions are metastable and poor physical stability is the major factor limiting their use in drug delivery.5 For drug candidate selection to be successful, an indication of the likely physical stability of the product must be given. 		

Figure 3:  Chemical stability of beaker melts and extrudates as measured by HPLC (drug/PVP melt, drug/PVP extrudate). Reproduced with permission from Drug. Dev. Ind. Pharm and Figure 4:  X-ray powder diffractograms of amorphous (red) and crystalline indomethacin (black).				

One method of predicting the physical stability of a glass solution is determination of the glass transition temperature (Tg).2,15,16 To increase the chances of forming a stable product, the Tg should be well above the storage temperature - with some authors recommending that the Tg should be approximately 50 Â°C above the storage temperature.15 At temperatures above the Tg, the amorphous material will change from a hard, brittle material into a less viscous rubbery state, which is more prone to recrystallization17 and also difficult to process.5 The Tg of a glass solution can be determined by differential scanning calorimetry (DSC) or predicted using the Gordon-Taylor equation (Equation 1).	

The Tg of an amorphous drug can be determined experimentally using DSC or it can be estimated from the melting point based on the empirical relationship Tg5melting point30.7 (Kelvin). Depending on the quantity of drug available and the physical stability of the amorphous drug, the true density can either be measured or estimated. True density values for amorphous low molecular weight pharmaceuticals are typically in the range of 0.86-2.0 gcm-3.15	

To begin the drug selection process for melt extrusion, it is necessary to examine the drug in combination with a suitable carrier. The hydrophilic polymer PVP has been used successfully to form glass solutions with a variety of poorly water-soluble drugs13 and is a suitable excipient for preliminary studies on drug selection. The major limitation to the use of PVP is that it begins to degrade at temperatures in excess of 150 Â°C.18 Previous studies, however, indicate that as long as temperatures do not exceed 190 Â°C for long periods, the thermal stability of PVP should be acceptable.18 For drug selection, this means that compounds  with a melting point higher than 200 Â°C are not suitable for extrusion with PVP alone. Additionally, the majority of useful matrix excipients are prone to degradation above these temperatures and, therefore, additional plasticizing and/or solubilizing agents will be required, which may effect the stability and dissolution rate of the glass solution formed.	

Figure 5:  Glass transition temperature of indomethacin/PVP/VA (1:1) glass solutions determined by MTDSC (u) and predicted by the Gordon-Taylor equation (--).	

Small-scale preparation of glass solutions/solid dispersions. DSC is an ideal means of preparing solid dispersions on a 5-10 mg scale. DSC is routinely available to most pharmaceutical companies, allows excellent control of temperature (heating/cooling rate) and provides information regarding the solid state properties of the material. In addition, the use of a nitrogen purge gas will allow compounds prone to oxidation at elevated temperatures to be investigated. Hot stage microscopy (HSM) has similar advantages of scale and will be discussed further as a preparative technique. 	

A second option is to perform small-scale stainless steel beaker co-melts, incorporating limited shear stress in the form of stirring. Cooling of the co-melt can be easily achieved by partially submerging the beaker into ice cold water or pouring the melt onto a cold stainless steel surface. For analysis of the material, it is normally necessary to lightly grind the material into a powder, which can then be sieved to select an appropriate size fraction. Care should be taken when applying any mechanical processing not to change the solid state of the product. As described earlier, it should also be noted that co-melting often does not lead to glass solution formation. The authors found that beaker melts often resulted in incomplete melting of the drug and the production of some residual crystallinity in the solid systems. However, it was also found that the majority of the beaker melt product was produced as a glass solution and therefore, given the dearth of alternative approaches, stainless steel beaker co-melts are a useful preparation technique.18 The use of beaker co-melts is somewhat less attractive now because of the availability of small-scale melt extruders. However, these devices are comparatively expensive and are likely to be time-consuming to use. 				

Figure 6:  Drug selection protocol for melt extrusion (1 5 melting point 

Using melt extrusion to prepare glass solutions of poorly water-soluble drugs with hydrophilic excipients offers an exciting and advantageous alternative to existing formulation methods such as spray-drying and co-melting. Investigating potential methods to increase water solubility begins early in drug development. Techniques described in this paper show how only a small quantity of drug can be used to determine its suitability for melt extrusion, allowing the method to be considered at the same time as salt screening and particle size reduction work, and could speed up the formulation process.