Gavin P. Andrews

Hot-melt extrusion (HME) is an established process that has been used since the early 1930s, predominately in the plastics manufacturing industry, but also in the food processing industry. Currently, more than half of all plastic products, including bags, sheets, and pipes, are manufactured using HME.

  Since the advent of plastics production, polymers have been melted and formed to different shapes for a variety of industrial and domestic applications. Extrusion can be simply defined as the process of forming a new material (the extrudate) by forcing it through an orifice or die under controlled conditions,

  such as temperature, mixing, feed-rate and pressure. 

At the most fundamental level, an extruder consists of a platform that supports a drive system, an extrusion barrel, a rotating screw arranged on a screw shaft and an extrusion die for defining product shape. Typically, process parameters are controlled via connection to a central electronic control unit. The extrusion drive system generally comprises motor, gearbox, linkage and thrust bearings, whereas the barrel and screw is commonly utilized in a modular configuration.

  Simple single screw arrangements consist of only a single rotating screw inside a stationary extruder barrel, whereas more advanced machines involve twin-screw systems utilizing either a corotating or counter-rotating screw configuration. It is common for the extrusion screw to be characterized by the length/diameter (L/D) ratio, which typically ranges from 20 to 40:1. Typical pilot plant extruders have diameters ranging 18–30 mm, whereas production machines are much larger with diameters typically exceeding 50 mm. Irrespective of the complexity of the machine, the extruder must be capable of rotating the screw(s) at a selected speed while compensating for the torque generated from the material being extruded. 

A simple single screw extrusion system comprises one rotating screw inside a stationary barrel that may be conveniently subdivided into three distinct zones: feed zone, compression zone and metering zone. The depth and/or pitch of the screw flights differ within each zone, generating variable pressure along the screw length (zone dependent). Because of the large screw flight depth and pitch, the pressure within the feed zone is very low, allowing for consistent feeding from the hopper and gentle mixing of API and excipients (Figure 1). The primary function of the subsequent compression zone is to melt, homogenize and compress the extrudate so that it reaches the metering zone in a form suitable for extrusion. Consequently, the compression zone must impart a high degree of mixing and compression to the material. This is achieved by decreasing the screw pitch and/or the flight depth, resulting in a gradual increase in pressure along the length of the compression zone.

  The final section, the metering zone stabilizes the pulsating flow of the matrix, thus ensuring the extruded product has a uniform thickness. Constant screw flight depth and pitch helps maintain continuous high pressure to ensure a uniform delivery rate of molten material through the extrusion die and, hence, a uniform product.  

At a minimum, a screw extruder consists of three distinct parts:  					

			a conveying system for material transport and mixing     

downstream auxiliary equipment for cooling, cutting and/or collecting the finished product.  

Individual components within the extruder are the feed hopper, temperature controlled barrel, rotating screw, a die and heating/cooling elements. Additional systems include mass flow feeders to accurately meter materials into the feed hopper, PAT to measure extrudate properties (near infra red systems and laser systems), liquid and solid side stuffers, vacuum pumps for degassing, pelletizers and calendaring equipment. Standard process control and monitoring devices include zone temperature and screw speed with optional monitoring of torque, drive amperage, melt pressure and melt viscosity. Temperatures are normally controlled by electrical heating bands and monitored by thermocouples. 

Single-screw extrusion is a fundamental operation for polymer processing. It is used to increase pressure within a polymer melt, allowing extrusion through a die or injection into a mould. Although a relatively simple process, singlescrew extrusion does not possess the mixing capability of a twin-screw machine and is, therefore, not the preferred approach for the production of pharmaceutical formulations. Moreover, the versatility of a twin-screw extruder (process manipulation and optimization) and the ability to accommodate various pharmaceutical formulations makes this set-up much more favourable. In relation to machine design, rotation of the screws inside the extruder barrel may either corotate (same direction) or counter-rotate (opposite direction); both directions being equivalent from a processing perspective. Another significant design variable is whether the two screws are intermeshing or nonintermeshing, the former being preferred because of the greater degree of conveying achievable and the shorter residence times. Additionally, the configuration of the screws themselves may be varied using forward conveying elements, reverse conveying elements, kneading blocks and other designs to achieve particular mixing characteristics.

Twin-screw extrusion offers the pharmaceutical formulator a rapid, continuous process that has much better mixing capability than singlescrew extrusion. Moreover, twinscrew extrusion provides a more stable melting process, shorter residence times and significantly greater output. Industrially, twinscrew extrusion has become extremely favourable because of process practicality and the ability to combine separate batch operations into a single continuous process, thus increasing manufacturing efficiency. In recent years, FDA has published guidance to the pharmaceutical industry to enhance and modernize the manufacture of pharmaceutical products to produce better quality products.

  The guidance facilitates the introduction of new technologies to improve the efficiency and effectiveness of manufacturing process design, control and quality assurance. Two key areas where this has been progressed are in the move toward continuous manufacturing processes and the implementation of PAT to monitor, control and understand manufacturing processes. HME is a process that lends itself well to continuous, lean manufacturing principles, and the implementation of in-line measurement and control technologies.      			

HME technology is being explored and used in the pharmaceutical industry and academia because of several advantages compared with traditional processing methods.

  Not only is it an efficient manufacturing process, it may also enhance the quality and efficacy of manufactured products.

  Of particular interest is the use of HME to disperse APIs in a matrix at the molecular level, thus forming solid solutions. HME is an approach utilized in the delivery of poorly water-soluble, class II compounds because of the increased dissolution achievable, and improved absorption and therapeutic efficacy.

  Whilst the formation of solid solutions may significantly enhance drug dissolution rate in vivo, the presence of a metastable state, high internal energy and specific volume of the amorphous state leads to a tendency during storage (thermal and/or humidity stress) towards recrystallization. Extruded solid solutions offer greater thermodynamic stability than those prepared by alternative processes such as spray drying, solvent evaporation and other hotmelt methods.

  Moreover, HME offers many advantages compared with traditional processing techniques that have been typically used to produce oral solid dosage forms. In particular, solvents are not required, which makes the process more environmentally friendly and cost effective; fewer unit batch operations are required than in traditional dosage form manufacture, and improved content uniformity because of intense mixing and agitation may be achieved. Furthermore, the polymeric components used in the extrusion process may function as thermal binders, drug stabilizers, drug solubilizers and/or drug release controlling excipients with no compressibility requirements. Typical examples of pharmaceutically approved polymeric materials include vinyl polymers (polyvinylpyrrolidone [PVP]), PVPvinyl acetate [PVP-VA]), polyethylene oxide (PEO), Eudragit (acrylates), PE glycol (PEG) and cellulose derivatives (hydroxypropylcellulose [HPC], hydroxypropylemethylcellulose [HPMC], HPMC acetate succinate [HPMCAS], cellulose acetate [CA] and CA-phthalate [CAP]). While residence time within the extruder and highprocessing temperatures (required to melt the polymeric carrier) were initially projected as significant disadvantages of this technology, the ability to modify screw configuration and the high shear forces generated within the extruder allow for processing at lower temperatures, coupled with the use of plasticizing agents and the introduction of twinscrew extruders, removed such concerns. Typical plasticizing agents for HME include PEGs,

  However, APIs have also been shown to be effective plasticizers in many cases.

  These plasticizers occupy sites along the polymer chain, reduce polymer–polymer chain secondary bonding and provide more mobility for the macromolecules, resulting in a softer, more easily deformable mass,

Although initial investigations focused on the effects of formulation and processing variables on the properties of the final dosage form,

  more recent scientific articles have described the use of HME for the manufacture of solid dispersions and for the development of minimatrices. Early work by De Brabander 					

 described the preparation of matrix minitablets that minimize the risk of dose dumping, reduce inter- and intrasubject variability and provide highly dispersive formulations within the gastrointestinal tract. This was complemented by further investigations into the properties of sustained release minimatrices manufactured from ethyl cellulose, HPMC and ibuprofen.

  The potential of HME to produce a sustained release, matrix-in-cylinder device, was conducted by Mehuys 

  A fourfold increase in the bioavailability of propanolol was observed when comparing the HME formulation with a commercially available formulation (Inderal; AstraZeneca, Belgium). 

During the last 5 years, there has been a significant increase in the use of HME for the manufacture of drug delivery systems. It has been used to manufacture granules, pellets, immediate and modified release tablets, transmucosal/trandermal films and implantable reservoir devices;

  for example, with respect to oral drug delivery, molecular solid dispersions of nifedipine,

  have been successfully produced using HME technology. In a recent article by Chokshi 					

  stable, amorphous indomethacin dispersions have been manufactured using pharmaceutically acceptable hydrophilic polymers. Sustained release pellets have been successfully developed by Young 					

								 whereas HME floating tablets have been examined by Fukuda 					

  Additionally, targeted drug delivery systems, including enteric matrix tablets and capsules systems, have been extruded.

 have demonstrated the ability of HME to act as an efficient dispersive process for aggregated, fine engineering particles to enhance their wettability and improve dissolution rate properties. 

Current interest in HME technology is growing rapidly with in excess of 100 papers published in the scientific literature during the last decade. Germany and the US and Germany hold approximately more than half (56%) of all issued patents.

  Despite this increased interest, there are few commercialized HME pharmaceutical products currently marketed. 

Several companies specialize in the use of HME as a drug delivery technology including PharmaForm (TX, USA) and SOLIQS (Germany). SOLIQS has developed a proprietary Meltrex formulation and redeveloped a proteaseinhibitor combination product, Kaletra, for the treatment of human immunodeficiency virus (HIV) (Figure 2). The formulated HME product significantly enhanced the bioavailability of both APIs.

  Moreover, HME Kaletra tablets were shown to have significant advantages for the patient compared with the previous soft gel capsule formulation, such as reduced dosing frequency and improved stability, as evidenced by FDA's decision to fasttrack approval. In addition, SOLIQS has also developed a fast-onset ibuprofen system and a sustained release formulation of verapamil (Isoptin SRE) that was the first directly shaped HME product on the market.  

Similarly, PharmaForm, a service provider for HME R&D has published more than 40 peerreviewed publications on the topic and has developed significant intellectual property. One proprietary melt extrusionbased technology, PharmaForm Abuse Deterrent Technology (PADT), has been developed to deter and prevent drug abuse. This technology utilizes HME to produce a drug delivery platform that helps prevent rapid drug absorption and the desired euphoric effect. The drug delivery technology offered by the company provides a solution to this problem as the dosage forms   are very difficult to chew and/or crush. Additionally, PADT systems prevent alcoholinduced dose-dumping and alcohol extraction by maintaining sustained release characteristics in 40% alcohol, comparable to the release in water or normal dissolution media, for more than 3 h.  

The use of HME to improve oral bioavailability has been detailed in numerous scientific articles and patents. A recent study examined the performance of itraconazole melt extrudates prepared using HPMC and PVP-VA/Eudragit E polymeric platforms. In this specific example, HPMC was shown to significantly increase the area under the plasma concentration time curve compared with the other extrudates, despite having a slower dissolution rate.

  One conclusion drawn from this work is that dissolution rate may not always be the most important factor when screening such formulations; inhibition of precipitation from a supersaturated state may be more important. More recently, HPMCAS has been shown to be of particular benefit in the preparation of solid dispersions and for the production of enteric dosage forms.

  Industrially, Pfizer has used HPMCAS solid dispersions prepared by a spray drying method to significantly enhance AUC for a number of their poorly soluble drugs compared with conventional formulations.

  The acetyl and succinate groups present within HPMCAS provide sufficient capacity for API/polymer hydrogen bonding; secondary forces that have been shown to be highly significant in the formation and stabilization of solid dispersions.

The application of HME technology in the pharmaceutical industry has tended to focus on the development of bio-enhanced formulations to increase the efficacy of poorly watersoluble compounds. Whilst this will probably remain the major focus for the pharmaceutical industry, there has also been an increase in the application of HME for the development of controlled release formulations, in the form of pellets, beads or minimatrices, and as a means to facilitate the continuous processing of products to reduce the number of manufacturing unit operations.

  Moreover, there have been several articles investigating the application of HME technology for the production of transdermal or bioadhesive films,

  and as a compounding process for the production of a high-density powder for subsequent compression into a conventional oral dosage form.

  Recent studies have also demonstrated the production of biocompatible shape-memory polymers for use in biomedical applications, using HME as a manufacturing process.

  The production of multiparticulate dosage forms using HME has been investigated using hotmelt pelletization

  and, lately, the use of die face-cutting the polymer extrudate to produce HME pellets shows the continuing utilization of technology from the plastics industry for pharmaceutical manufacturing.

  The scope of the technology has also been broadened to expand the range of polymers and APIs that can be processed through application of supercritical fluid assisted HME.

  has shown that pressurized carbon dioxide can reduce the processing temperature of pamino salicylic acid, a thermally labile API, extruded in ethyl cellulose. This research demonstrates that using supercritical fluids to reduce the melt viscosity of polymers enables the HME process to operate at lower temperatures, thereby lowering the risk of API degradation. It has also been recognized that HME is a robust, solvent-free manufacturing process for the production of films, which may present future opportunities to develop gastroretentive films for prolonged drug delivery, and multilayer films to modulate drug release for oral and transdermal applications. A multilayer gastrointestinal mucoadhesive patch system (GI-MAPS) for the oral delivery of proteins has been reported by Eiamtrakarn 					

  (2002) and although the films were solventcast, a HME process would also offer a rapid, solventfree alternative manufacturing process for the polymers described.  

The growing market in medical devices, including those that incorporate drugs such as biodegradable stents and drugloaded catheters, will require HME manufacturing processes to be commercialized, and may lead to new areas of collaboration across pharmaceutical, medical device and biotechnology research. The breadth of research and published articles demonstrates that HME is a versatile processing technology for the pharmaceutical industry and has exciting prospects for the future.  

is a lecturer at the School of Pharmacy at Queen's University Belfast (UK). 

is a PhD student at Queen's University Belfast (UK). 

is the Chair of biomaterial science at the School of Pharmacy at Queen's University Belfast (UK). 

is a PhD student at the School of Pharmacy at Queen's University Belfast (UK). 			

Introduction to Polymer Science and Technology

 (John Wiley & Sons, New York, USA, 1977). 

European Journal of Pharmaceutics and Biopharmaceutics

The Dynisco Extrusion Processor Handbook, 1st Edition

 (London School of Polymer Technology, London, UK, 1996). 

4. R.J. Crawford, Plastics Engineering (Butterworth-Heinemann, Oxford, UK, 1998). 

5. J.L. White, Twin Screw Extrusion: Technology and Principles (Hanser/Gardner Publications Inc., Cincinnati, OH, USA, 1991). 

6. FDA Guidance for Industry — Pharmaceutical cGMPs for the 21st Century: A Risk-Based Approach, September 2004.  

8. I. Ghebre-Sellassie and C. Martin (Eds.), Pharmaceutical Extrusion Technology, (Marcel Dekker Inc., New York, USA, 2003). 

Pharmaceutical Development and Technology

13. N. Follonier, E. Doelker and E.T. Cole, 

14. C. Aitken-Nichol, F. Zhang and J.W. McGinity, 

19. J.P. Liu, F. Zhang and J.W. McGinity, 

European Journal of Pharmaceutics and Biopharmaceuticals

21. C. De Brabander, C. Vervaet and J.P. Remon, 

et al., European Journal of Pharmaceutics and Biopharmaceutics 

29. C.R. Young, J.J. Koleng and J.W. McGinity

30. M. Fukuda, N.A. Peppas and J.W. McGinity, 

31. E. Mehuys, J.P. Remon and C. Vervaet, 

Articles

Hot-melt extrusion offers many advantages compared with conventional solid dosage form manufacturing, and has consequently received considerable attention from both the pharmaceutical industry and academia as a novel drug delivery technology. The possibility of forming solid dispersions with improved bioavailability renders hot-melt extrusion an excellent alternative to other conventionally employed techniques.

Hot-melt extrusion: an emerging drug delivery technology

Fluidized hot melt granulation (FHMG) is an emerging technique combining the advantages of both dry and wet granulation methods, and represents an innovative continuous granulation process capable of mixing and agglomerating excipients and active pharmaceutical ingredients (APIs) to produce uniform blends of particles suitable for use in the manufacture of pharmaceutically elegant solid dosage forms.  

This study investigates the influence of process parameters and formulation variables on the FHMG of two model pharmaceutical systems using either polyethylene glycol (PEG) or a copolymer of polyoxyethylene/polyoxypropylene (Lutrol F68, BASF; Ludwigshafen, Germany). Granule growth within the FHMG process reached equilibration after a defined period for both model systems. The Lutrol F68 system reached equilibration after 2.5 min, whereas PEG required at least 5 min. The equilibration time significantly decreased as the molecular weight (Mw) of PEG increased from 6000 to 10000.  

Granule growth for these systems proceeded via a coalescence mechanism, which decreased significantly as the molecular weight of the binder increased.  

Increasing the binder content within the systems increased the mean particle diameter until a threshold value was reached (20% w/w PEG and 10% w/w Lutrol F68), after which the system defluidized and an overwetted mass was produced. Interestingly, the regularity of the granules was shown to be dependent upon the binder concentration within the fluidized system. Increased binder content resulted in irregular granules, whereas lower binder contents produced highly spherical granules. 

Orally administered drug delivery platforms, such as tablets and capsules, are considered the preferred and most patient-convenient dosage forms available today.

  This is primarily because of the ease of administration, convenience of handling and increased stability compared with their liquid counterparts.

  Typically, solid dosage forms administered orally are an intricate blend of excipients (diluents, binders, disintegrants, glidants, lubricants and flavours) and APIs. To successfully manufacture acceptable pharmaceutical products, these materials must be adequately mixed and/or granulated to ensure that the resultant agglomerates possess the required fluidity and compressibility, and avoid demixing during postgranulation processes.

Currently, the techniques in existence for the agglomeration and mixing of pharmaceutical powders involve either wet or dry methods.

  Although dry techniques lead to associated decreases in process time and the avoidance of wetting and drying processes, the inherent difficulties in compressing crystalline solids, the uneven and erratic flow properties of APIs, and the development costs associated with dry methods, culminates in wet granulation remaining the preferred and most widely accepted method for powder agglomeration. However, it is well accepted that there is an increasing need for alternative processes to dramatically improve particle processing to ensure acceptable and reproducible solid dosage forms.  

One such emerging technique is FHMG, a nonambient temperature process conducted at elevated temperatures to change the physical nature of one or more of the components. FHMG is an innovative process involving meltable binders to agglomerate fluidized dry powders.

  FHMG avoids using solvents — negating the problems associated with in-process hydrolysis and solvent removal. FHMG is a simple and rapid technique that does not require powder blends to possess high levels of fluidity and compressibility, and reflects the simplicity of dry techniques in that it may be performed in one step. This is in contrast to conventional wet techniques whereby transfer from the granulator to the drying equipment is usually necessary — commonly involving losses in transfer, contamination of manufacturing equipment, increased processing and operator time, and increased dust levels: issues that are particularly pertinent when manufacturing dosage forms containing potent drugs.  

Although FHMG is being highlighted as a next generation method for producing pharmaceutical granules, very little research in this area has been reported in scientific literature. The majority of work that has been described involves processes whereby molten binder is sprayed onto a bed of fluidized particles. While not identical to FHMG, investigations in this respect have identified key parameters affecting granule growth within the fluid bed. Abberger 

. have previously shown that granule growth mechanisms are dependent upon the ratio of binder/powder particle size, whereas Kidokoro 

. have shown that the molecular weight of the binder is significant in controlling the physical properties of compressed tablets.

 have previously reported the influence of process parameters on FHMG and the characteristics of resultant tablets pressed from pharmaceutical powders.

This investigation reports the use of FHMG for the agglomeration of model pharmaceutical powders and discusses the effect of binder type, molecular weight, concentration and granulation time on both granule growth mechanisms and granule characteristics. Moreover, we identify the significant potential of the process and provide a fundamental understanding of the growth mechanisms within the fluidized bed using both a model pharmaceutical excipient (lactose) and standard Ballotini (glass) beads as inert fillers. We also highlight the diversity of the process in relation to the choice of meltable binder using two different examples, PEG and Lutrol F68. 

Lactose (alpha-D-lactose monohydrate), polyvinylpyrrolidone (PVP) (Mw 29000) was supplied by Sigma-Aldrich Chemical Company (Gillingham, UK). Standard Ballotini beads (100–250 μm) were supplied by Ulster Anaesthetics (County Down, UK). PEG (Mw 6000, 10000 and 20000) were supplied by Fluka Chemie AG (Buchs, Switzerland) and Lutrol F68 (Mw ~7000, mean average particle size [Dp] 430 μm), a copolymer of polyoxyethylene-polyoxypropylene was supplied by BASF plc (Cheadle, UK). In this study, the PEG flakes were finely ground and sieved between 250 μm and 106 μm mesh to give a mass mean particle size of 178 μm. 

Pharmaceutical powders were granulated in a Sherwood Scientific (Mark II) fluidized bed dryer (Cambridge, UK) consisting of a 5 L container with a fine mesh nylon gauze air distributor, stainless steel support gauze and a filter bag at the top of the unit. FHMG was performed at a temperature of 100 °C ±1 °C using a constant air velocity of either 1.0 (PEG/lactose blends) or 0.5 ms

  (Lutrol/PEG/Ballotini blends) for predetermined time periods (5–20 min).  

Given that all FMHG processes were in excess of the melt temperature of the binders, consolidation of the agglomerated particles was achieved by fluidizing the granulate using ambient air for 30 s at the end of each granulation run.

  The granules that were produced were sized using standard sieving techniques to determine the effect binder viscosity, concentration, molecular weight and filler type had on the particle size distribution of the powder bed. The viscosity of the binders (PEG and Lutrol) were determined at 100 °C using an AR2000 rotational rheometer (TA Instruments; Crawley, UK) in continuous shear mode. Granule shape, uniformity and dimensions were determined using a Nikon Eclipse ME600 microscope using ×50 magnification and Lucia G software (both Nikon Instruments; Badhoevedorp, The Netherlands). 

Typically, the characteristics of granulated powder blends (particle size, particle shape, uniformity of powder blend and powder compressibility) are major contributors to the performance of orally administered dosage forms post-manufacture and, therefore, the ability to control such factors is pertinent to their successful manufacture. Despite the vast array of current granulation methods, they all inherently have problems. The process described in this article is a highly novel and exciting combination of both wet and dry techniques.  

FHMG is an innovative method of preparing pharmaceutical granules of a suitable size and compression profile for subsequent processing into high-performance solid dosage forms. In FHMG, a molten binder, API and other functional excipients are heated together in a fluidized bed system; when the granules are cooled to the solid state (also within the fluidized bed) it is found that the molten binder acts to form solid bridges resulting in a consolidated granule.  

FHMG has significant potential for the preparation of granules for tablet pressing, but it has not been widely applied because the detailed mechanism of granule growth and the relationship between the growth parameters and the properties of the granules that will ultimately determine the performance of the product are poorly understood.  

The effect of binder viscosity on granule growth within melt granulation processes has been reported by many authors,

  and it has been suggested by Iveson and Lister that variation in the binder viscosity may change the dominant granule growth mechanism within the powder bed.

  Figure 1 illustrates the effect of PEG molecular weight (6000, 10000 and 20000) on the mass mean diameter at various granulation times (5–15 min) using lactose as an inert filler. The extent of granulation was significantly higher for PEG 6000 compared with 10000 and 20000 at granulation times exceeding 5 min (10 and 15 min). 

Notably, the mass mean diameter of PEG 6000 increased significantly (P<0.05) as a function of granulation time, whereas granule growth did not significantly change (P>0.05) as a function of granulation time for 10000 and 20000, suggesting that agglomerates within these systems had reached an equilibrium mass after 5 min. PEG 6000 has a viscosity of 500 mPa.s, whereas 10000 and 20000 have significantly higher molten viscosities (3500 and 19000 mPa.s, respectively). The increased molecular weight, and hence, chain length within PEG 10000 and 20000 significantly reduces polymer chain mobility and deformation.

  Given the shear forces operating within the fluidized bed are significantly lower than alternative granulating systems, the extremely low granule deformation within this process and the extent of polymer chain mobility for PEG 10000 and 20000 is insufficient to force the high viscosity binders to the outer layers of agglomerating particles during granule–granule collision within the fluidized bed, and, therefore, granules fail to grow in mass after initial powder layering and coalescence has occurred.

Pharmaceutical granulation encompasses a large number of processes designed to agglomerate a powder mass together and is a classical example of particle design, whereby the characteristics of the resultant agglomerates are carefully tailored using a unique combination of formulation design and process design.

  One of the most important processes during the granulation of suspended solid particles, is the formation of new primary particles (wetting and nucleation). Therefore, an adequate quantity of binder is required to ensure that this process continues; however, excessive binder may result in a slurry as a result of an overwetted powder mass.

  Thus, the concentration of binder within the powder bed is fundamental to the growth and stability of the granulation process.  

The concentration of meltable binder within the fluidized bed is extremely important to the successful manufacture of particles possessing the required characteristics. In this investigation, granule growth was examined as a function of binder concentration (10%, 12% and 14% w/w) and granulation time (5, 10, 15 and 20 min). There was shown to be no significant granule growth occurring after a period of 10 min (mass mean Dp ~0.70). Additionally, granule growth also appeared to be largely independent of binder content after initial nucleation and coalescence had occurred. When granulation was attempted with a PEG content of 20% (w/w), an overwetted slurry was produced and granulation moved to the overwet massing regime (i.e., the system was overwetted and defluidized). The oversaturation at 20% w/w PEG content induced a liquid-like agglomerate that was unable to maintain its identity under the shearing forces acting within the FHMG process.

The second system investigated in this study involved the use of glass beads as a model pharmaceutical filler and Lutrol F68 as a model meltable binder. The effect of varying binder concentration on the particle size distribution of the powder bed is shown in Figure 2. Lutrol F68 has a relatively low viscosity and, therefore, when in the molten state has a high degree of mobility resulting in diffusion to the outer surface of the growing agglomerates.

Consequently, the Ballotini/Lutrol F68 system overwetted at a much lower concentration (8–10% w/w) than any previous PEG systems that have been investigated.

  Analysis of the granule shape (×50 magnification) illustrated a distinct difference in granule shape as a function of binder concentration (Figure 3). As the concentration of binder increased within the powder bed, the resultant agglomerate was more irregular in shape, whereas those agglomerates formed with lower concentrations of binder were more uniform. The shape of the granules supports the visual observation of overwetting occurring during the 10% and 12% w/w binder content. At such high concentrations of binder, the binder diffuses to the outer surface of the agglomerates. During granulation, two colliding agglomerates fuse together, a coalescence granule growth mechanism dominates and the resultant particle is highly irregular. Conversely, at low binder concentrations (3%, 5% and 8% w/w) the mechanism of granule growth is that of layering, where the binder is able to distribute freely through out the bed to form spherical granules.

The effect of granulation time on the mean particle diameter is shown in Figure 4. The mean particle diameter does not change significantly after 2.5 min suggesting that granule growth has reached equilibrium after this time period. As previously discussed, Lutrol F68 is a low molecular weight binder with a high degree of mobility. After 2.5 min the binder has diffused throughout the entire fluidized bed of Ballotini and formed liquid–liquid bridges. These liquid-bridged ballotini form a flocculated system, which continues to grow, thus forming defined agglomerates. 

The viscosity of Lutrol F68 is such that it supports the rapid diffusion of the binder, and consequently the process is effectively complete within a short time period (2.5 min). However, increasing the binder concentration from 3% to 5% w/w significantly increased the mean particle diameter. Granule growth will only continue to occur when binder is available at the granule surface to form pendular bonds between colliding granules.

  Therefore, the system containing 5% w/w Lutrol F68 has a higher mass of binder available and hence, more pendular bonds maybe formed, increasing the probability of granular growth during collisions, explaining the observed difference in mean particle diameter between powders blends containing 3% and 5% w/w Lutrol F68. 

Continuous granulation within the pharmaceutical industry is very much in its infancy, with it being more typical for pharmaceutical solid dosage forms to be batch-processed. However, more recently, many factors have led to an increased interest in continuous processes such as fluidized bed agglomeration, spray-drying, extrusion, instant agglomeration and roller compaction.

  The technique described in this article (FHMG) is an innovative example of continuous granulation, combining the advantages of both conventional wet and dry processes.  

While FHMG has significant potential, the process has been bottlenecked by the lack of understanding in relation to granule growth within the fluidized system when employing a meltable binder and the relationship between process parameters and granule properties. The research described in this paper addresses this deficit through the use of two model pharmaceutical systems. The first consisted of PEG as a meltable binder and lactose as a model inert diluent. The experimental data obtained for this system suggested that binder content, molecular weight and granulation time were significant factors in controlling the granule growth during processing. After initial nucleation and coalescence, binder content was shown to have no effect on granule growth until 20% w/w, at which point an overwetted slurry was produced and the powder bed defluidized. Binder viscosity was shown to significantly influence granule growth, and it may be concluded that low viscosity, highly mobile molten polymers will produce a system that grows via coalescence and layering, whereas higher viscosity binders will result in a system that fails to grow after powder layering has occurred because of the inability of the binder to migrate to the outer surface of the colliding granules.  

In the second system containing glass Ballotini and Lutrol F68, granule growth was significantly influenced by binder concentration and granulation time. Within this system, the process produced an overwetted powder mass at binder concentrations greater than 8% w/w. This may be a result of the nondeformable, nonpermeable nature of the Ballotini compared with the lactose that results in a much lower tolerability of binder within the powder bed (8% versus 20% w/w). Moreover, the granule shape was shown to be more irregular at higher binder concentrations, evident of granule fusion and coalescence; at lower binder concentrations (3% and 5% w/w), granules were highly regular, typical of a layering process in which the binder can distribute freely throughout the fluidized system. For both systems, granulation time was shown to be extremely important with both exhibiting a threshold value, after which no further granule growth occurred.  

 While this study describes the use of two model pharmaceutical systems, both have shown that the properties and growth of granules within FHMG may be carefully controlled through the manipulation of formulation variables (binder type, concentration and molecular weight) and process variables (granulation time).

is a lecturer in pharmaceutics and academic member of the MPRI. More recently, he has worked as a visiting scientist within Professor McGinity's research group at the University of Texas at Austin.  

holds the Chair of Biomaterial Science, School of Pharmacy, Queen's University Belfast (UK). Professor Jones is a director of the Drug Delivery and Biomaterials Group and team leader of the Biomaterials Group within the Medical Polymers research Institute (MPRI). His particular interests are in the design, synthesis and physicochemical characterization of biomaterials as medical devices and drug delivery systems. 

is a senior lecturer in the School of Chemistry & Chemical Engineering at Queen's University Belfast (UK). Dr Walker's particular interests involve mathematical aspects of materials technology and granulation processing. 

is a senior lecturer in the School of Chemistry & Chemical Engineering at Queen's University Belfast (UK). His research group are working on the development and use of laser-based spectroscopic methods to study problems of fundamental interest, such the nature of structural changes that occur in ultra fast chemical reactions, and on the nondestructive characterization of complex real-life chemical systems, ranging from foodstuffs to forensic evidence. 

Fluidized hot melt granulation (FHMG) is an emerging technique combining the advantages of both dry and wet granulation methods, and represents an innovative continuous granulation process capable of mixing and agglomerating excipients and active pharmaceutical ingredients (APIs) to produce uniform blends of particles suitable for use in the manufacture of pharmaceutically elegant solid dosage forms.

Particle design using fluidized hot melt granulation

	Pharmaceutical dosage forms contain both pharmacologically-active compounds and excipients, which have been added to aid the formulation and manufacture of the dosage form. The properties of the final dosage form are highly dependent on the excipients chosen, their concentration and interaction with both the active compound and each other. Excipients cannot be regarded as inert or inactive ingredients and a detailed understanding of their physiochemical properties, safety and handling, and regulatory status is essential for pharmaceutical scientists.

	There has been increasing interest in the use of bioadhesive polymers as excipients in the design of drug delivery systems. One of the advantages of using these materials is that they can maintain contact with mucosal surfaces for longer periods than nonbioadhesive polymers. Because polymers possessing bioadhesive properties can retain drugs in close proximity to membranes rich in underlying vasculature, they may offer a solution to the poor bioavailability of some drugs and a method to avoid enzymatic degradation of others.

	Film formulations based on poly(methylvinylether/maleic anhydride [PMVE/MA]) are known to possess moisture-activated bioadhesive properties.

 Bioadhesive films cast from aqueous blends of PMVE/MA have diverse uses, such as a means of establishing an electrically conducting interface for bioelectrodes and as an adhesive drug delivery matrix.

 In structural terms, the five-membered anhydride ring of PMVE/MA contributes two carbon atoms to the polymer backbone and, therefore, confers rigidity on the system. On hydrolysing this anhydride moiety to the corresponding free acid form (PMVE/MAC [Figure 1]), a reduction in glass transition temperature (T

) of only 10 °C from the relatively high T

 of the dry powder (151 °C) is observed because of the increased flexibility of the free acid structure.

 Films cast from aqueous blends of the free acid are, consequently, very brittle and of little use in formulating films for drug delivery purposes. Therefore, these systems require the inclusion of a second excipient to act as a plasticizer in the film casting blend.

	Plasticizers used for PMVE/MAC film formulations are, typically, water-miscible, polyhydric alcohols, such as glycerol. These plasticizers are suspected of cross-linking PMVE/MAC on heating or on prolonged standing at room temperature.

 This study investigates the influence of two different plasticizers on the properties of films cast from aqueous blends of PMVE/MA and potential mechanisms by which the polymer may cross-link on storage. The aim is to produce a stable film that maintains its adhesion and conformability properties on storage, rendering it suitable for use as a topical drug delivery system.

 Gantrez AN-139, a copolymer of MVE and MA was provided by ISP Co. Ltd (UK). Glycerol, tripropyleneglycol methyl ether (TPM [Dowanol]) and methylene blue were obtained from Sigma Aldrich (UK). All other chemicals used were of analytical reagent quality.

 Aqueous polymer blends were prepared as previously described using the required weight of PMVE/MA, which was added to ice-cooled water (reagent grade 1) and stirred vigorously.

 The mixture was then heated and maintained between 95 and 100 °C until a clear solution was formed. Upon cooling, the required amount of plasticizer was added, the pH adjusted to 4.5 using 10 M NaOH and the casting blend adjusted to final weight with water. Bioadhesive films were prepared by slowly pouring the aqueous blend (30 g) into a mould consisting of a release liner (FL2000 PET 75 μ 1S; Rexam Release BV, The Netherlands), siliconized side-up, secured to a Perspex base plate using a stainless steel clamp. Once assembled, the internal dimensions available for casting were 100 mm by 100 mm (Figure 2). The mould was placed on a levelled surface to allow the blend to spread evenly across the area of the mould. The cast blend was dried under a constant airflow at 25 °C.

	Films cast from blends containing PMVE/MA alone, PMVE/MA with glycerol and PMVE/MA with TPM were stored at 25 °C in a sealed vessel with a relative humidity of 65% maintained by the presence of a saturated solution of sodium nitrite. Films were removed from storage at 7-day intervals and their bioadhesive, tensile and swelling characteristics investigated.

	The bioadhesive properties of all films were evaluated quantitatively using a TA-XT2 texture analyser (Stable Microsystems, UK) in tensile mode. Full thickness, shaved, neonate porcine skin was attached with cyanoacrylate adhesive to a lower platform. Film segments (1 cm

) were attached to the probe of the texture analyser using double-sided adhesive tape. Adhesion was initiated by adding a defined amount of water (10 μL) over an exposed skin sample (1 cm

) and immediately lowering the probe with attached film.

	Upon contact, a force of 5 N for 30 s was applied before the probe was moved upwards at a speed of 0.1 mm s

. Adhesion was recorded as the force required to detach the sample from the surface of the excised skin. The distance to separation of a test film from the skin substrate, that is, the normal displacement from the skin surface that the probe had travelled at the instant the film and substrate lost contact with each other, was also recorded to provide some measure of the cohesion within the film sample. Results were reported as the mean (± standard deviation [SD]) of five replicates.

	The tensile strength and percentage elongation at break of films were determined using the texture analyser. Film strips of 5 mm width were grasped using an upper and lower flat-faced metal grip laminated with a smooth rubber grip. The distance between the grips was set at 20 mm and this distance, therefore, represented the length of film under stress. A cross-head speed of 6 mm s

 was used for all measurements. The resultant force–time profiles were analysed using propriety software (Dimension 3.7E; Stable Microsystems, UK). Only results from films that were observed to break in the middle region of the test strip during testing were used. The percentage elongation at break, E

, of tested films was determined using Equation 1.

	Where E is the extension to break of the film and L

 is its original length. The break strength, B, of tested films was determined using Equation 2.

	Where F is the break force of the film and AR is its cross-sectional area. Results were reported as the mean (±SD) of five replicates.

	The swelling behaviour of films was investigated by immersing dry, preweighed film segments (1 cm

) in distilled water (30 mL) at 25 ºC. The segments were allowed to swell to equilibrium and then reweighed following careful blotting with absorbent paper to remove surface water. The equilibrium weight swelling index (SI) of the films was defined as the ratio of the weight of the swollen film (W

	Results were reported as the mean (±SD) of five replicates. Where appropriate, results were analysed using a one-way analysis of variance (ANOVA). Post-hoc comparisons were made using Fisher's protected least significant difference (PLSD) test where p <0.05 was taken to represent a statistically significant difference.

	Films cast from aqueous blends of PMVE/MA with no plasticizer were brittle and unsuitable for use as the basis of a conformable, bioadhesive delivery system. Addition of TPM caused a significant reduction in T

 of cast films, as determined using differential scanning calorimetry (DSC [DSC 2920 with refrigerated cooling system; TA Instruments, UK, data not shown]). T

 significantly decreased as TPM contents increased.

Tensile strengths of films cast from aqueous blends containing 10% w/w PMVE/MA decreased significantly (p <0.0001) with the addition of TPM to the casting blend, while the percentage elongations at break showed significant (p <0.0001) increases (Table 1). Glycerol addition to the casting blend produced similar, significant (p <0.0001), reductions in tensile strengths and similar, significant (p <0.0001) increases in percentage elongations at break.

	Neither glycerol (p=0.5964) nor TPM addition (p=0.0516) had a significant effect on the bioadhesion of freshly prepared films to shaved neonate porcine skin (Table 2). Initially, freshly prepared films plasticized with glycerol showed significant (p <0.0001) increases in distance to separation relative to unplasticized films, whereas similar changes were not observed in films plasticized with TPM.

	Films cast from blends containing 5% w/w glycerol lost their bioadhesive strength over time (Figure 3). There were significant decreases in adhesion after only 7 days' storage (p <0.0001). By contrast, the bioadhesive properties of films cast from blends containing 5% w/w TPM remained unaltered over a similar time frame. Films cast from blends containing 10% w/w glycerol also showed significant reductions in adhesion after 6 weeks' storage (p <0.0001), an effect not seen for films cast from blends containing 10% w/w TPM or unplasticized films.

	All films containing glycerol showed pronounced decreases in distance to separation over time, with significant decreases being noted in each case after 7 days' storage. For example, for films cast from blends containing 10% w/w glycerol, the mean distance to separation had already decreased significantly (p <0.0001) after only 7 days. Further significant decreases in distance to separation occurred with increasing storage times. Unplasticized films did not display this effect, while only slight decreases were noted for films containing TPM (data not shown).

	Storage was found to have a profound effect on the tensile properties of films containing glycerol in particular. Films cast from blends containing 5% w/w glycerol showed increases (p <0.0001) in break strength over time, significant changes being observed after only 7 days' storage. Films cast from blends containing 10% w/w glycerol also showed significant increases in break strength over time. After 42 days' storage these films had a mean break strength of 0.424×106 N m

. The original mean break strength of these films was 0.04×106 N m

. Films cast from blends containing 5% w/w TPM showed significant increases in break strength after 6 weeks' storage (p <0.0001). Importantly, films cast from blends containing 10% w/w TPM showed unaltered tensile strengths after 6 weeks' storage (p=0.9368).

An examination of the elongation at break for films cast from blends containing 10% w/w glycerol showed significant decreases on storage (Figure 4). For example, after 6 weeks' storage, the mean percentage elongation at break of films cast from blends containing 10% w/w glycerol had decreased significantly (p <0.0001). A similar pattern was observed for films cast from blends containing 5% w/w glycerol. By contrast, films cast from blends containing 10% w/w TPM showed no significant decreases in percentage elongation at break after 6 weeks' storage (p=0.4455).

	Films cast from blends containing 5% w/w TPM did, however, show significantly decreased percentage elongations at break after 6 weeks' storage, which had mean values of zero (p <0.0001). The original mean percentage elongation at break of these films was 592.2%.

Films cast from aqueous blends of PMVE/MA, with and without plasticizer, dissolved in water initially. Films cast from blends containing 5% w/w glycerol became insoluble in water after 7 days' storage, showing significant increases in equilibrium weight swelling indices (p <0.0001). The swelling indices (Figure 5) then decreased progressively, and films cast from blends containing 10% w/w glycerol displayed a similar pattern. However, the initial rise in the swelling index was much more pronounced in the latter case. No such effect was observed for films plasticized with TPM.

	Plasticized bioadhesive films based on hydrolysed PMVE/MAC, are potentially useful drug delivery platforms for a range of topical applications, including local delivery to moist mucosal epithelial tissue. Such films have excellent initial adhesion and conformability properties.

 However, this study demonstrates the loss of adhesion and concomitant changes in mechanical properties that occurs during storage. Consequently, a detailed investigation has been made of the fundamental behaviour of the plasticized PMVE/MAC film system to allow drug delivery films with stable adhesion and mechanical properties to be formulated.

	Polyhydric alcohols, such as propylene glycol, glycerol, and poly(ethylene glycol) 400, have previously been shown to act as plasticizers of PMVE/MAC films.

 Glycerol is most typically used as a plasticizer for PMVE/MAC.

	However, films cast from blends plasticized with glycerol were observed to lose their adhesion progressively on storage (Figure 3). Distance to separation during tensile mode adhesion testing also decreased. The flexibility of glycerol-plasticized films was reduced concomitantly over time, with films becoming insoluble and showing a marked reduction on water swelling indices after only 7 days' storage (Figure 5). These observations are all consistent with a progressive increase in film cross-linking density during storage.

	TPM was shown to be an effective plasticizer for PMVE/MA-based films intended for drug administration purposes, progressively lowering the T

 of the bioadhesive polymer in direct proportion to the plasticizer concentration in the casting blend. Films cast from PMVE/MA blends containing 5% w/w of TPM maintained their bioadhesive properties (Figure 3) and remained water-soluble after 6 weeks' storage. Elongation at break was also maintained in stored films plasticized with TPM (Figure 4). Solid-state nuclear magnetic resonance (NMR) studies revealed the formation of ester bonds between the two excipients, PMVE/MAC and glycerol (Varian UNITY Inova spectrometer with a 7.05 T Oxford Instruments magnet and a 7 mm standard MAS probe; Doty Scientific Instruments, Columbia, SC, USA, data not shown).

	In contrast to previously reported plasticizers of PMVE/MAC

 TPM is monohydric, possessing only a terminal hydroxyl group. It was chosen as an alternative plasticizer in this study because it could not cross-link PMVE/MAC chains by means of an interaction with the carboxylic acid groups on the bioadhesive polymer, for example, by forming ester linkages. Films plasticized with 5% w/w TPM displayed a significant loss of flexibility on storage. Notwithstanding this, films remained sufficiently pliant for topical application. Whether this reduction in flexibility was a result of enhanced hydrogen bonding, a progressive increase in physical entanglements or to limited ester formation leading to increased interchain cohesion was not determined. However, films cast with higher levels of TPM (10% w/w) displayed mechanical properties similar to those found immediately after drying, the increased excipient content presumably aiding retention of water within the film structure.

	The time-dependent decrease in adhesion and distance to separation seen with glycerol-plasticized films are indicative of a polymer system that is becoming more restricted in movement by the formation of a tighter network of cross-links.

 Hence, film cohesion increases and distance to separation falls. In addition, the loss of functional groups, such as -COOH, which are capable of hydrogen bonding to a biological substrate, may contribute to the observed reduction in bioadhesion.

 In contrast, films containing TPM showed decreased distances to separation, but maintained their bioadhesive strength, suggesting an increased interchain cohesive attraction with preservation of chemical residues capable of hydrogen bonding.

	The swelling indices (Figure 5) of 5% w/w glycerol-plasticized films decreased progressively, suggesting a system that is becoming increasingly organized and of tighter disposition and can, therefore, accommodate less water within its network.

 A similar pattern was found in films cast from blends containing 10% w/w glycerol, but the initial rise in swelling index was much more pronounced. This may arise from the high glycerol content leading to greater interchain separation and formation of a looser network. All unplasticized films and those containing 5% and 10% w/w TPM remained water-soluble throughout the study period.

	The swelling characteristics observed in films cast from blends containing glycerol, however, together with their gradual insolubilization in water, clearly suggest that there is a slow chemical alteration, typically occurring over about 7 days, in the glycerol-plasticized PMVE/MAC film system. It is possible that glycerol is participating in a cross-linking reaction as a result of ester formation with the free carboxylic acid moieties formed on hydrolysis of PMVE/MA. Because glycerol is polyhydric, attachment could occur at more than one point on the bioadhesive polymer chain, thereby contributing to formation of a three-dimensional PMVE/MAC network.

	Hydrogen bonding to the hydroxyl groups on the free acid form of PMVE/MAC is unlikely to be the predominant cross-linking mechanism, however, because TPM has four equivalent oxygen atoms capable of hydrogen bonding in a similar way to those found in glycerol. The formation of a physically cross-linked system because of entanglements between neighbouring chains is, again, unlikely, as this would be more feasible in unplasticized systems because of the reduced interchain separation.

 This may also explain the increasing rigidity of the TPM plasticized polymer films, which are now becoming internally, rather than externally plasticized, because internal plasticization is known to be less efficient than external plasticization.

	As a result of the observed negative impact of excipient–excipient interactions between glycerol and PMVE/MAC on film properties, we have subsequently used TPM as the plasticizing excipient in all bioadhesive films prepared. These films have been employed as topical drug delivery systems for 5-aminolevulinic acid used in photodynamic therapy (PDT),

 which is a technique whereby a combination of a photosensitizing drug and visible light causes destruction of selected cells via singlet oxygen production.

	We have also made use of the observed excipient–excipient interaction to produce PMVE/MAC films cross-linked with poly(ethylene glycol [PEG] 6000). We postulated that, as cross-linking is known to retard drug release,

 an extensively cross-linked system should be able to immobilize a drug compound within the film matrix. Immobilization of a photosensitizer within a film surface coating should inhibit microbial growth on that surface under ambient lighting by generation of toxic singlet oxygen.

We prepared films as previously described from aqueous blends of PMVE/MA (20% w/w) and PEG 6000 (10% w/w). The final film contained 10 mg cm

 of the antimicrobial photosensitizer methylene blue, which had been added in appropriate amounts to the aqueous blend. Films were stored for 72 h in a desiccation chamber at 60 °C to force the cross-linking reaction. When drug release experiments (modified Franz cell; Crown Glass Co. Inc., Sommerville, NJ, USA) were performed immediately upon film preparation, approximately 90% of the methylene blue loading was released after 3 h. However, following cross-linking, the amount of drug released over a similar period was virtually undetectable (Figure 6). Discs of the crosslinked films incubated in 96-well plates under ambient lighting with 10

 MRSA 180 completely resisted biofilm growth.

	In summary, we have identified a detrimental excipient–excipient interaction between copolymer and plasticizer in pharmaceutical systems intended for topical application. As a result, we have employed an alternative plasticizer in films used subsequently in the clinic. However, discovery of the nature of the interaction has been exploited to produce photoactive materials containing immobilized photosensitizer with potential as anti-infective surface coatings for use in hospitals.

	1. R.C. Rowe, P.J. Sheskey and P.J. Weller, Eds., 

Handbook of Pharmaceutical Excipients, 4th Edition

 (Pharmaceutical Press, London, UK, 2003).

	4. A.D. Woolfson, D.F. McCafferty and G.P. Moss, 

	5. K.H. Chung, C.S. Wu and E.G. Malawer, 

	6. Gantrez AN-139 production bulletin,ISP Corp, Wayne, NJ, USA (1995).

Bioadhesive films cast from aqueous blends of PMVE/MA have diverse uses, such as a means of establishing an electrically conducting interface for bioelectrodes and as an adhesive drug delivery matrix.