Dielectric properties of pharmaceutical materials relevant to microwave processing: Effects of field frequency, material density, and moisture content

Dielectric Properties of Pharmaceutical Materials Relevant to Microwave Processing: Effects of Field Frequency, Material Density, and Moisture Content PAUL W.S. HENG,1 Z.H. LOH,1 CELINE V. LIEW,1 C.C. LEE2 1

Faculty of Science, Department of Pharmacy, 18 Science Drive 4, National University of Singapore, Singapore 117543, Singapore 2

Akzo Nobel Surface Chemistry Pte. Ltd., 41 Science Park Road #03-03, The Gemini, Singapore Science Park II, Singapore 117610, Singapore

Received 5 March 2009; revised 17 May 2009; accepted 9 June 2009 Published online 25 August 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21872

ABSTRACT: The rising popularity of microwaves for drying, material processing and quality sensing has fuelled the need for knowledge concerning dielectric properties of common pharmaceutical materials. This article represents one of the few reports on the density and moisture content dependence of the dielectric properties of primary pharmaceutical materials and their relevance to microwave-assisted processing. Dielectric constants ð"0 Þ and losses ð"00 Þ of 13 pharmaceutical materials were measured over a frequency range of 1 MHz–1 GHz at 23  18C using a parallel-electrode measurement system. Effects of field frequency, material density and moisture content on dielectric properties were studied. Material dielectric properties varied considerably with frequency. At microwave frequencies, linear relationships pffiffiffiffi pffiffiffiffiffi were established between cube-root functions of the dielectric parameters ( 3 "0 and 3 "00 ) and density which enabled dielectric properties of materials at various densities to be estimated by regression. Moisture content was the main factor that contributed to the disparities in dielectric properties and heating capabilities of the materials in a laboratory microwave oven. The effectiveness of a single frequency density-independent dielectric function for moisture sensing applications was explored and found to be suitable within low ranges of moisture contents for a model material. ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:941–957, 2010

Keywords: excipients; solid state; processing; compaction; preformulation; water in solids; physical characterization; powder technology; dielectric properties; microwave frequency

INTRODUCTION Dielectric properties are fundamental electrical characteristics of materials which dictate to a large extent, their behavior when subjected to high frequency fields for heating, drying or

material processing.1 In dielectric analysis, the characteristic response of a material under the influence of an alternating electric field, referred to as the complex permittivity, e, is measured over a wide frequency or temperature range.2 e is governed by the following equation:3 " ¼ "0  j"00

Correspondence to: Paul W.S. Heng (Telephone: 6565162930; Fax: 65-67752265; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 941–957 (2010) ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association

(1)

where "0 refers to the dielectric constant of the material and reflects its polarizability or ability to store electrical charge. "00 , the material dielectric

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

941

942

HENG ET AL.

loss, reflects the heating capability of the material on exposure to electromagnetic energy. The loss tangent, tan d, defined as "00 ="0 , represents the fraction of incoming energy that is dissipated as heat within the material. With the exception of materials that do not absorb energy from electromagnetic waves, dielectric properties of most materials vary considerably with frequency.4 In the microwave frequency range, this frequency dependence or dielectric dispersion as it is alternatively known, stems from the effects of polarization that arise predominantly from the orientational movements of dipolar molecules as they attempt to align themselves with the oscillating electric field. This response is critically dependent on the relaxation time of the molecular dipoles. Relaxation time refers to the time taken by the dipoles to revert to random orientation when the field is terminated and is influenced by their molecular weight and mobility. The latter governs the responsiveness of the dipoles to the applied field and affects the extent to which microwaves can be effectively coupled into the material for heat production. Classically, the frequency-dependent variation in dielectric constant and loss of a pure, polar molecule had been described mathematically by Debye5 (Fig. 1). For a pure polar liquid like water for instance, the polar molecules are randomly oriented in the absence of an electric field. When an electric field is applied at low frequencies, the time interval taken for the field to reverse its polarity would presumably be longer than the relaxation times of the polar molecules. As a result, ample time is available for the molecules to respond and orientate in accordance to the

Figure 1. Dielectric dispersion of a pure polar substance. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

direction of field changes. The partial neutralization of electrical charges imposed by the external field led to charge storage by the polar molecules as evidenced by a high dielectric constant (region A). As field frequency increases, the time interval between the next reversal in field polarity gradually becomes of a similar order to the relaxation times of the polar molecules. At this juncture, they retain their ability to respond to the changing fields albeit with an increasing lag time. This causes the dielectric constant to decline with the occurrence of a dielectric loss peak as a result of energy absorption and dissipation (region B). At very high frequencies, the time interval at which the field stays in the same direction is much shorter than the time taken for the polar molecules to relax. These molecules will cease to respond and remain in their random, steady state orientations with minimal charge storage and energy dissipation (region C). Thus when the applied field frequency is exceedingly high or low with respect to the relaxation frequencies of the polar molecules in a material, there is minimal energy absorption and the resultant heating effects will be negligible. At intermediate frequencies, heating effects are more pronounced with the most effective conversion of microwave energy to heat occurring at the frequency where the material exhibits its maximum dielectric loss. Small, noninteracting dipoles in a liquid system couple most readily with microwave fields and conform to the Debye model. The dielectric loss peak of pure liquid water in its unbound or ‘‘free’’ state occurs approximately at 17 GHz at 208C.2 This implies that the conversion of microwave energy into thermal energy for liquid water would be most efficient at this frequency. However, in many materials of practical interest, water rarely exists in its unbound state. Often, it is physically absorbed in material capillaries or cavities or chemically bound to other molecules in the material. Furthermore, depending on the structural properties of the material, various forms of bound water may exist which differ in their binding affinities.6 Due to their restricted mobility, bound forms of water possess longer relaxation times and undergo dielectric dispersion at lower frequencies,3,4,7 with loss peaks typically occurring in the range of 1 MHz–1 GHz.3,4,7,8 In the same vein, the extensive intermolecular bonding interactions prevalent in solid materials restrict dipolar orientations and impair the responsiveness of the dipoles to microwave fields. Polarization in solid materials originate primarily DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

from electron cloud distortions (electronic polarization) or the movements of oppositely charged ions and atoms relative to each other (ionic or atomic polarization).9,10 Polymers constitute a special category of solids in that interest lie in their macroscopic polarization involving the slow and hindered movements of the bulky polymer chains and their associated side groups. These phenomena are often investigated at lower frequency ranges. The dielectric susceptibilities of polymers are affected by the presence of crystalline and amorphous regions within the main chains as well as the sizes of their pendant side groups.11 Apart from frequency, the dielectric properties of materials are affected by their physicochemical characteristics. As air possesses a low dielectric constant and loss of 1 and 0, respectively,12 the mass per unit volume or bulk density of particulate materials exert a profound effect on their dielectric properties. Generally, materials with higher bulk densities and correspondingly smaller interparticulate air volumes possess higher dielectric constants and losses.1,4,6,12–14 On the other hand, the markedly higher dielectric constant of pure, liquid water ("0  78 at 1 GHz)3 relative to that of dry, organic materials ("0  2  5) at ambient conditions contributes to a heavy reliance of material dielectric properties on moisture content. This relationship has been exploited for the online moisture determination of particulate food materials such as cereal grains, wheat and coffee using microwave techniques.15,16 However, the concomitant effects of bulk density on dielectric properties constituted a major confounding factor that greatly limited the accuracy of these moisture-sensing devices. Measures have since been adopted to compensate for the effects of bulk density by the development of density-independent functions which depended exclusively on material moisture content under standard temperature conditions.17–21 Densityindependent functions may comprise 1 or 2 dielectric parameters ("0 and "00 ) determined at different frequencies. At a single frequency, both variables "0 and "00 , are usually measured simultaneously and combined in a single expression. These two variable density-independent functions have been purported to be universal in that they could be successfully applied at microwave frequencies regardless of the dielectric measurement technique.20 Comprehensive summaries on these functions have been provided by Shrestha et al.22 as well as Berbert and Stenning.23 DOI 10.1002/jps

943

The popularity of microwave technology for food and agricultural applications however, has not been followed as readily by the pharmaceutical industry. With the development of single pot processors and novel hybrid dryers, the use of microwaves has largely been confined to drying pharmaceutical powders and granules. Interest in microwave and related techniques for nondestructive, real time moisture determination of wet powders and granules in high shear and fluidizedbed processes has caught on only in recent years.24–26 On a smaller scale, microwaves have been employed for modulating drug and excipient properties via specific material–microwave interactions which are nonthermal in nature.27–32 These included physicochemical modifications of susceptible drug compounds to improve their solubilities and bioavailabilities as well as the enhancement of polymer cross-linking and interactions in matrices to sustain the release of encapsulated drugs. However, depending on the intended application of microwave technology, such specific material–microwave interactions may be undesirable as they threaten the therapeutic and safety profiles of pharmaceutical products manufactured using microwave technologies. More often than not, the insidious nature of these interactions go undetected till they manifest as altered physical properties of the product, or at a more microscopic level, transformations in the solid state structure, phase and surface areas of active constituents or excipients. Knowledge of the dielectric properties of pharmaceutical materials is thus essential for the avoidance of latent and potentially adverse effects of microwaves on pharmaceutical products. It is also useful for quick identification and selection of materials with dielectric responses suitable for the intended application of microwave technology. In drying applications for instance, dielectrically inert or microwave transparent materials are preferred as this ensures that microwaves interact selectively with water molecules. This minimizes undesirable heating and thermal degradation of the product particularly at the end stages of drying when its residual moisture content is low. Under circumstances where microwaves are employed for purposeful modification of material properties, materials with suitably high dielectric susceptibilities are required to ensure an optimal, targeted and efficient usage of microwave energy. Apart from some pioneering studies by McLoughlin et al.33,34 who examined the low frequency (102 to 106 Hz) dielectric properties of JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

944

HENG ET AL.

was, in turn, elucidated by examining the relationship between the moisture contents and dielectric properties of materials at their respective true densities. This eliminated confounding density influences. As a convenient and more economical alternative for the assessment of electromagnetic properties, the microwave-induced heating capabilities of materials were determined using a laboratory microwave oven.34,35 Since the dielectric analyzer and laboratory microwave oven differed in their test frequency ranges, sample preparation and handling methods, the results derived from the two modes of measurements were compared. Finally, the applicability and sensitivity of a single frequency densityindependent function as outlined by Powell et al.19 to moisture variation of a model material was explored.

several drugs and excipients and their impact on drying patterns of the wetted materials in a microwave oven dryer, there remains a paucity of information on the high frequency dielectric properties of primary pharmaceutical materials despite their greater relevance and significance to microwave processing. Hence, this study aimed to evaluate the high frequency dielectric properties of pharmaceutical materials in relation to their responses to microwaves under typical processing situations. Dielectric properties of 13 pharmaceutical materials comprising fillers, binders, disintegrants and active ingredients widely employed in the formulation of solid dosage forms were measured over a frequency range of 1 MHz–1 GHz at 23  18C using a parallel-electrode sample holder connected to an impedance analyzer. The effects of field frequency on material dielectric constants and losses were first evaluated. Dielectric measurements were also performed on dried forms of materials that possessed high initial moisture contents. Since typical pharmaceutical processes such as mixing, granulation or drying entail dynamically changing material densities and moisture contents, the effects of these changes on dielectric properties are important. At specific microwave frequencies within the range of 300 MHz–1 GHz, the effect of material density was investigated by measuring the dielectric properties of material samples of different densities. The effect of moisture content

MATERIALS AND METHODS Materials The materials used and their typical functions in solid dosage forms are shown in Table 1. a-Lactose monohydrate (Pharmatose 200M, DMV, Veghel, The Netherlands), starch (National 78–1551, pregelatinized corn starch NF, National Starch and Chemical, Bridgewater, NJ) and anhydrous dicalcium phosphate (Rhodia, Cranbury, NJ) are

Table 1. Function and Properties of Pharmaceutical Materials Used

Material

Function

Lactose Anhydrous dicalcium phosphate Starch Sodium alginate Polyvinylpyrrolidone C15 K25 K29/32 K90D Polyvinylpyrrolidone–vinyl acetate copolymer S630 Cross-linked polyvinylpyrrolidone XL XL10 Paracetamol Acetylsalicylic acid

Filler

a

Moisture Contenta (%w/w)

True Density (g/mL)

Bulk Densitya (g/mL)

0.236 (0.046) 0.225 (0.064)

1.535 2.842

0.425 (0.010) 0.701 (0.004)

9.402 (0.354) 9.744 (0.176)

1.483 1.720

0.478 (0.009) 0.867 (0.015)

Binder

6.866 6.766 8.492 17.113 4.675

(0.375) (0.475) (0.080) (1.067) (0.123)

1.196 1.179 1.140 1.200 1.197

0.449 0.387 0.333 0.349 0.245

(0.002) (0.002) (0.002) (0.003) (0.001)

11.727 13.302 0.221 0.156

(0.037) (0.669) (0.136) (0.041)

1.197 1.196 1.301 1.392

0.215 0.264 0.271 0.587

(0.002) (0.008) (0.002) (0.020)

Disintegrant

Drug

Standard deviations are indicated in parentheses.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

common tablet diluents. Polyvinylpyrrolidone (PVP) C15, K25, K29/32 and K90D (Plasdone C15, K25, K29/32 and K90D, ISP Corp., Wayne, NJ) are long chain synthetic water-soluble homopolymers of N-vinyl-2-pyrrolidone differing in their molecular weights and polymer chain lengths. Polyvinylpyrrolidone–vinyl acetate (PVP–VA) S630 copolymer (Plasdone S630, ISP Corp.) is a synthetic, linear 6:4 random copolymer of N-vinyl2-pyrrolidone and vinyl acetate. Sodium alginate (Manucol LB, ISP Corp.) is a naturally occurring polymer derived from brown algae. The polyvinylpyrrolidone polymers and sodium alginate are typically employed as binding agents. Crosslinked PVP XL and XL10 (Polyplasdone XL and XL10, ISP Corp.) differ in their mean particle sizes (31.97 and 121.33 mm, respectively) and are commonly used as tablet disintegrants. Paracetamol (Kangle, Wenzhou Pharm Factory, Wenzhou, China) and acetylsalicylic acid (Sintor, Bucharest, Romania) are model drugs chosen for this study.

Determination of Moisture Content and Physical Characteristics of Materials Moisture content was determined by thermogravimetric analysis (DTG 60H, Shimadzu, Kyoto, Japan). An aluminum pan was filled with a thin layer of material and heated from 28 to 1058C at a rate of 58C/min under a nitrogen environment. Depending on bulk density, the quantity of material used each time ranged from 6 to 10 mg. Moisture content (%w/w wet basis) was calculated from the loss in weight of the material upon heating and drying. Thermo-gravimetric analysis was also used to determine the critical moisture content, Mc, of each material. Mc was defined as the moisture content of the material that marked the end of the constant rate phase and start of the decreasing rate phase of drying.36,37 The drying profiles (moisture content vs. drying time) of the materials were first plotted following which the rate of moisture loss with time was derived by differentiation. For each material, the drying time at which the rate of moisture loss began to decrease was determined. The moisture content of the material at this particular drying time point was then estimated from the respective drying profile and defined as the Mc of the material. True density was determined using a pycnometer (Pentapycnometer, Quantachrome, Boynton Beach, FL) under helium purge. Apart from acetylsalicylic DOI 10.1002/jps

945

acid, materials were dried for 2 h in a convection oven (Modell 600, Memmert, Bavaria, Germany) maintained at 958C and cooled overnight in a dessicator prior to the test. For acetylsalicylic acid, drying was conducted overnight in a vacuum oven (Vacuum Oven, Gallenkamp, England, UK) set at 450 mbar and 608C. Bulk density was determined by sieving the material through a 1 mm aperture sieve such that it flowed freely, with the aid of a glass funnel, into a graduated cylinder cut exactly at the 25 mL mark. The weight of material occupying 25 mL was determined. Bulk density was defined as the quotient of weight and volume of material. A minimum of three replicated experiments was conducted for each test. The moisture contents and physical characteristics of materials are summarized in Table 1.

Preparation of Material Compacts for Dielectric Analysis Untreated Materials An accurately weighed sample of material was filled into a cylindrical stainless steel die of a universal testing machine (Autograph AG-100kNE, Shimadzu) and compacted between 2 flat-faced punches (14.93 mm in diameter) with the upper punch approaching a stationary lower punch at 2 mm/min. Each material was compacted under a range of pressures to produce compacts of different densities. Due to the inherent differences in their compactibility characteristics, the amounts of materials used and pressures applied ranged from 0.127 to 0.878 g and 6.9 to 542.6 MPa, respectively. These values were chosen such that the resultant compacts formed were mechanically strong and did not chip or break during handling. At the end of each compaction cycle, the lower punch was removed and the formed compact gently ejected from the die. Each compact was individually sealed in plastic bags and allowed to recover for a minimum of 3 days prior to use. Dried Materials Materials containing higher moisture contents, namely, starch, sodium alginate, the PVP binders and disintegrants were dried to constant weight in a convection oven (Modell 600, Memmert) maintained at 958C. Upon cooling, the dried materials were compacted following the procedure described in the previous section. Drying adversely affected JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

946

HENG ET AL.

the compactibility characteristics of the materials as negligible amounts of moisture remained to facilitate interparticulate bond formation. Therefore, significantly higher pressures as compared to those applied for the untreated materials were required for the preparation of compacts with reasonable strength. Resultant compacts were stored and equilibrated over silica gel for a minimum of 3 days prior to use. As dried sodium alginate could not be compacted within the allowable limits of the equipment load cell (100 kN), the dried form of sodium alginate was excluded from the study. Starch is a popular excipient in pharmaceutical dosage forms as it performs multi-functional roles. This material was selected to test the effectiveness of a density-independent function and its sensitivity to moisture variation. Several petri-dishes each containing an accurately weighed 2–4 g sample of starch were dried at 908C in the convection oven (Modell 600, Memmert). The drying profile of starch was determined gravimetrically by monitoring the weight loss of five samples of starch at predetermined time points of 1.5, 3, 4, 8, 12, 15, and 30 min into the drying process. At these selected time points, samples of starch were also retrieved from the remaining petri-dishes and stored in tightly sealed amber bottles. After a total of 90 min, drying was terminated and all the starch samples were cooled and equilibrated over silica gel for 24 h. The end point of drying was set at 90 min as it was found from preliminary trials that moisture loss from starch became negligible after 1 h. The dried and cooled starch samples were re-weighed and the values used for the computation of the initial moisture content (%w/w wet basis) of starch. Cooled starch samples previously stored in sealed amber bottles at the various drying time points were compacted in an identical manner as aforementioned. Five replicated compacts were prepared each time. Measurement of Compact Density All the compacts produced were accurately weighed (CP423S, Sartorius, Lower Saxony, Germany). Thickness measurements were performed with a micrometer screw gauge (Mitutoyo, Tokyo, Japan) at three predefined locations diametrically across each compact and averaged to obtain the mean thickness of each compact. The density of each compact was computed from the quotient of its weight and volume. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

Dielectric Analysis Material dielectric properties were determined over a frequency range of 1 MHz–1 GHz at 23  18C. Prior to the actual measurements, the equipment was calibrated using polytetrafluoroethylene discs. Polytetrafluoroethylene is non polar and possesses a dielectric constant of 2 over the range of frequencies investigated in this study.38,39 During the actual test measurements, each material compact was inserted between the plates of a parallel-electrode sample holder (16453A Dielectric material test fixture, Agilent, Santa Clara, CA) connected to an impedance analyzer (E4991A RF Impedance/material analyzer, Agilent). The upper electrode had an internal spring which allowed the compact to be fastened between the electrodes thereby ensuring intimate contact between the surfaces of the compact and electrodes. This minimized any air gaps existing between the samples and the electrodes which may act as a further capacitor in series.2 As an oscillating electric field of increasing frequency was applied to each material, the alternating positive and negative charges on the electrode plates resulted in polarization of the polar molecules and functional groups in the material which enabled the storage of electrical charge. From the mean thickness of each compact, the capacitance of the material was calculated and corresponding dielectric constant ð"0 Þ and loss ð"00 Þ derived. Resultant dielectric constant and loss spectra were obtained by scanning each compact repeatedly over the frequency range of interest. A minimum of three compacts was analyzed for each untreated and dried material.

Microwave-Induced Heating Capabilities of Materials in a Laboratory Microwave Oven The microwave-induced heating capabilities of materials were assessed using a laboratory microwave oven (NN-MX21WF, Panasonic, Osaka, Japan). Prior to the test, the initial temperature of a material, Ti (8C), was measured by means of a platinum 1000 V thermocouple connected to a temperature controller with a digital display (Model 89810-10, Cole-Parmer Instrument Co., Vernon Hills, IL). Each material was gently sieved through a 1 mm aperture sieve such that they packed uniformly, with the aid of a glass funnel, into a 10 mL measuring glass. The weight of each DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

947

material occupying a fixed 10 mL volume was recorded and their densities computed. With minimal disturbance, each material was then placed on the center of the microwave oven turntable and exposed to a microwave power output of 800 W for 60 s. The turntable minimized the effect of field variations within the oven thereby ensuring uniformity in microwave exposure of the material. The final temperature of the material after microwave exposure, Tf (8C), was measured immediately after irradiation. The thermocouple was fully inserted into the material each time its temperature was taken. The microwave-induced heating capability of the material was evaluated based on its temperature rise, DT (8C), obtained by calculating (Tf  Ti). All experiments were conducted at ambient conditions of 248C. A minimum of four replicated experiments was performed for each material.

RESULTS AND DISCUSSION Effect of Field Frequency on Material Dielectric Properties The variations in dielectric constants and losses of all materials plotted as functions of frequency expressed on a logarithmic scale are shown in Figures 2 and 3. Expectedly, linear relationships were observed between material dielectric constants and field frequency (Fig. 2). Apart from lactose, acetylsalicylic acid and paracetamol whose dielectric constants remained relatively consistent at 2.6–2.7 across the whole frequency range, dielectric constants of the remaining materials declined gradually as frequency increased. This indicated that the frequency range of 1 MHz–1 GHz fell within the region of dielectric dispersion of these materials and the decrease in their dielectric constants stemmed from the decreasing polarizabilities of the molecules with respect to increased field frequency. As the differences in density between the untreated and corresponding dried forms of materials were marginal (0.002–0.034 g/mL), direct comparisons were made between them. When plotted on the same scale, the decrease in dielectric constant was visibly more pronounced for the untreated materials which indicated that their dielectric dispersions were attributed primarily to water molecules inherent in their structures. At the frequency range studied, the dielectric contributions of the bulky polysaccharDOI 10.1002/jps

Figure 2. Frequency dependence of the dielectric constants of (a) anhydrous dicalcium phosphate, (b) starch, (c) sodium alginate, (d) PVP C15, (e) PVP K29/32, (f) PVP K90D, (g) cross-linked PVP XL10, (h) PVP K25, (i) PVP–VA S630, (j) cross-linked PVP XL, (k, l) lactose, acetyl salicylic acid, and (m) paracetamol.

ide and polymer chains per se were outweighed by the effects of water molecules. Energy dissipation of the water molecules contributed to the peaks observable in the dielectric loss spectra of materials containing high moisture contents such as the PVP binders (K90D and K29/32), disintegrants (XL and XL10) and starch (Fig. 3i). On the other hand, materials containing negligible amounts of moisture like paracetamol, acetylsalicylic acid and lactose as well as those subjected to drying as shown in Figure 3ii possessed unvarying and minimal dielectric losses at all frequencies. By virtue of their increased losses, untreated materials were likely to heat better on exposure to electromagnetic waves as compared to their dried counterparts, with the most effective conversion of electromagnetic energy to heat occurring at JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

948

HENG ET AL.

Figure 3. Frequency dependence of the dielectric losses of (a) starch, (b) sodium alginate, (c) PVP C15, (d) PVP K29/32, (e) PVP K90D, (f) cross-linked PVP XL10, (g) anhydrous dicalcium phosphate, (h) PVP K25, (i) cross-linked PVP XL, (j) PVP–VA S630, and (k, l, and m) lactose, acetylsalicylic acid, and paracetamol.

frequency bands at which the materials exhibited their maximum dielectric losses. However, the electromagnetic frequencies available for industrial processing are often displaced from these maxima. Although microwaves span a wide frequency range of 300 MHz to 300 GHz of the electromagnetic spectrum, 2.45 GHz had been designated for industrial uses to avoid potential interferences with frequencies employed for telecommunication, defense and maritime applications. Since the maximum attainable microwave frequency in this study is 1 GHz, subsequent discussions would focus on the effects of material density and moisture content on dielectric properties of materials at the lower and upper limits of the range of microwave frequencies available (300 MHz–1 GHz). Effect of Material Density on Dielectric Properties The effects of density on the microwave dielectric constants and losses of the materials are shown in Figure 4. The values were extracted from the JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

Figure 4. Density dependence of the dielectric properties of ( ) lactose, (~) anhydrous dicalcium phosphate, (!) starch, (*) sodium alginate, (&) PVP C15, (^) PVP K25, (5) PVP K29/32, (^) PVP K90D, (*) PVP–VA S630, (~) cross-linked PVP XL, ( ) crosslinked PVP XL10, (&) paracetamol and ( ) acetylsalicylic acid at LF 8.5 (i–ii) and 9 (iii–iv). DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

spectra of the untreated materials at specific logarithmic frequencies (LF) of 8.5 and 9 which corresponded, respectively, to the lower (300 MHz) and upper limit (1 GHz) of the microwave frequency range examined. It could be observed that the dielectric constants and losses of the majority of the materials increased significantly with density at both microwave frequencies. The effects of density were comparatively less pronounced for lactose, acetylsalicylic acid and paracetamol whose dielectric constants and losses remained low at all density levels. Taking the dielectric constant and loss of air (density ¼ 0 g/mL) to be 140 and 0,8 respectively, linear relationships were observed between cubepffiffiffiroot functions of the dielectric ffi pffiffiffiffiffi parameters ( 3 "0 and 3 "00 ) and density for all materials apart from PVP–VA S630. For the latter, a linear relationship was observed directly between its dielectric parameters ð"0 ; "00 Þ and density. The cubic relationships observed between the dielectric properties of the majority of materials and their densities were consistent with the Landau and Lifshitz,41 Looyenga42 dielectric mixture equation which describes the relationship between the dielectric properties of a true solid material and its corresponding air–particle mixture. This equation defines the effective dielectric properties of solid mixtures as the addition of the cube roots of the dielectric properties of its constituents taken in proportion of their volume fractions: ð"Þ1=3 ¼ v1 ð"1 Þ1=3 þ v2 ð"2 Þ1=3

(2)

e represents the complex permittivity of the mixture and e1, the permittivity of the medium in which particles of permittivity e2 are dispersed. v1 and v2 are volume fractions of the respective constituents, where v1 þ v2 ¼ 1. This equation has been routinely employed in the food, agricultural and mining industries for the calculation of the dielectric properties of pure solid materials when only particulate forms of the materials were available.13 Its relevance to pharmaceutical materials and formulations has not been extensively studied. Attempts were also made in this study to fit the experimental data to the next commonly encountered complex refractive index mixture equation which expresses the dielectric parameters of materials as quadratic functions of their densities: ð"Þ1=2 ¼ v1 ð"1 Þ1=2 þ v2 ð"2 Þ1=2 DOI 10.1002/jps

(3)

949

These two equations were selected as they had been found to be most suitable for estimating the solid material permittivities of minerals, plastics and food materials.13 However, based on the goodness-of-fit (R2) of the experimental data obtained, it appeared that the majority of the materials in this study conformed preferentially to the Landau and Lifshitz, Looyenga mixture equation (Eq. 2). Unlike the complex refractive index mixture equation which was found to be inappropriate for selected materials due to poor fitting, the Landau and Lifshitz, Looyenga equation was applicable to the entire range of materials studied. The linear relationships and R2 values are summarized in Table 2. Microwave-Induced Heating Capabilities of Materials in a Laboratory Microwave Oven To assess the relationship between the measured dielectric properties of materials and their responses to microwaves at 2.45 GHz, the equations in Table 2 were used for the computation of material dielectric losses at densities equivalent to those in the heating experiments performed using the laboratory microwave oven. Results obtained from the two methods of dielectric assessment were compared. Figure 5 presents the heating capabilities of the materials. At their respective densities during testing, DT of the materials differed significantly with starch and PVP K90D exhibiting the highest heating responses of 48 and 348C, respectively. These were followed by sodium alginate, cross-linked PVP XL10, PVP K29/32 and cross-linked PVP XL with relatively high DT values in the range of 17.8– 26.98C. The microwave-induced heating capabilities of PVP C15, anhydrous dicalcium phosphate, PVP K25 and PVP–VA S630 were in the lower range of 10.5–14.98C. Lactose, paracetamol and acetylsalicylic acid heated at the slowest rates, with mere temperature increments of 4.1–7.78C. The high dielectric susceptibility of starch as reflected by its dielectric profiles and DT value suggested an increased likelihood of microwave– starch interaction. Indeed, literature reports have shown that starches exposed to microwave irradiation experienced several physicochemical modifications as exemplified by changes in their solubilities, crystalline structures, swelling characteristics and morphologies. Furthermore, the extents of the microwave-induced changes were found to be strongly correlated to the crystal structures, amylose content and botanical origins JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

950

HENG ET AL.

Table 2. Equations Governing the Relationships between Microwave Dielectric Properties and Densities of Materials

Material Lactose Anhydrous dicalcium phosphate Starch Sodium alginate Polyvinylpyrrolidone C15 K25 K29/32 K90D Polyvinylpyrrolidone–vinyl acetate copolymer S630 Cross-linked polyvinylpyrrolidone XL XL10 Paracetamol Acetylsalicylic acid

Eqn (LF 8.5) Eqn (LF 8.5) Eqn (LF 9) Eqn (LF 9) fð"0 Þ ¼ kr þ 1 fð"00 Þ ¼ kr fð"0 Þ ¼ kr þ 1 fð"00 Þ ¼ kr R2 R2 R2 R2 p pffiffiffiffiffi pffiffiffiffi pffiffiffiffiffi ffiffiffiffi 3 0 " ¼ 0.313r þ 1 0.999 3 "00 ¼ 0.180r 0.993 3 "0 ¼ 0.308r þ 1 0.999 3 "00 ¼ 0.167r 0.993 ffiffiffi ffi p p ffiffiffiffi ffi p ffiffiffi ffi pffiffiffiffiffi 3 0 " ¼ 0.551r þ 1 0.991 3 "00 ¼ 0.273r 0.997 3 "0 ¼ 0.545r þ 1 0.992 3 "00 ¼ 0.329r 0.996 ffiffiffiffi p pffiffiffiffiffi pffiffiffiffi pffiffiffiffiffi 3 0 " ¼ 0.542r þ 1 0.947 3 "00 ¼ 0.745r 0.876 3 "0 ¼ 0.489r þ 1 0.951 3 "00 ¼ 0.690r 0.882 ffiffiffiffi p pffiffiffiffiffi pffiffiffiffi pffiffiffiffiffi 3 0 " ¼ 0.459r þ 1 0.974 3 "00 ¼ 0.428r 0.993 3 "0 ¼ 0.439r þ 1 0.975 3 "00 ¼ 0.418r 0.993 ffiffiffiffi p 3 0 " ¼ 0.543r þ 1 ffiffiffiffi p 3 0 " ¼ 0.524r þ 1 ffiffiffiffi p 3 0 " ¼ 0.553r þ 1 ffiffiffiffi p 3 0 " ¼ 0.583r þ 1 "0 ¼ 1.869r þ 1

0.992 0.982 0.980 0.967 0.989

p ffiffiffiffi 3 0 " ¼ 0.511r þ 1 ffiffiffiffi p 3 0 " ¼ 0.583r þ 1 ffiffiffi ffi p 3 0 " ¼ 0.354r þ 1 ffiffiffiffi p 3 0 " ¼ 0.307r þ 1

pffiffiffiffiffi 0.989 3 "00 ¼ 0.599r pffiffiffiffiffi 0.993 3 "00 ¼ 0.708r pffiffiffiffiffi 0.996 3 "00 ¼ 0.079r pffiffiffiffiffi 0.995 3 "00 ¼ 0.128r

p ffiffiffiffiffi 3 00 " ¼ 0.631r p ffiffiffiffiffi 3 00 " ¼ 0.586r p ffiffiffiffiffi 3 00 " ¼ 0.664r p ffiffiffiffiffi 3 00 " ¼ 0.663r "00 ¼ 0.048r

0.982 0.934 0.972 0.969 0.955

p ffiffiffiffi 3 0 " ¼ 0.513r þ 1 p ffiffiffiffi 3 0 " ¼ 0.500r þ 1 p ffiffiffiffi 3 0 " ¼ 0.520r þ 1 p ffiffiffiffi 3 0 " ¼ 0.542r þ 1 "0 ¼ 1.801r þ 1

pffiffiffiffi 0.815 3 "0 ¼ 0.485r þ 1 pffiffiffiffi 0.956 3 "0 ¼ 0.545r þ 1 pffiffiffiffi 0.620 3 "0 ¼ 0.350r þ 1 pffiffiffiffi 0.981 3 "0 ¼ 0.304r þ 1

p ffiffiffiffiffi 3 00 " ¼ 0.603r p ffiffiffiffiffi 3 00 " ¼ 0.579r p ffiffiffiffiffi 3 00 " ¼ 0.660r p ffiffiffiffiffi 3 00 " ¼ 0.660r "00 ¼ 0.048r

0.981 0.927 0.973 0.965 0.952

pffiffiffiffiffi 0.990 3 "00 ¼ 0.605r pffiffiffiffiffi 0.994 3 "00 ¼ 0.717r pffiffiffiffiffi 0.996 3 "00 ¼ 0.115r pffiffiffiffiffi 0.995 3 "00 ¼ 0.147r

0.828 0.958 0.935 0.989

0.992 0.981 0.983 0.971 0.990

k is a constant and "0 , "00 and r refer to the dielectric constant, loss and density (g/mL) of a material, respectively.

of the irradiated starch samples.43–45 Moreover, recent comparisons between potato and corn starches revealed that the structural differences between the two types of starches contributed to

Figure 5. Micro wave-induced heating capabilities (DT) of materials at their respective densities during testing (standard deviations arc parenthesized). Lac, lactose; DCP, anhydrous dicalcium phosphate; Sta, starch; Alg, sodium alginate; C15, PVP C15; K25, PVP K25; K29/32, PVP K29/32; K90D, PVP K90D; S630, PVP–VA S630; XL, cross-linked PVP XL; XL10, cross-linked PVP XL10; Para, paracetamol and ASA, acetylsalicylic acid. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

their markedly dissimilar microwave susceptibilities.46 The crystallinity, water retention capability, swelling power and capacity of potato starch were significantly affected by microwave irradiation whereas corn starch remained unchanged under identical conditions. It was postulated from these findings that corn starch was more suitable for use in product formulations subjected to microwave-assisted drying or processing. The specific interaction of microwaves with potato starch on the other hand, provides a possible avenue for purposeful material design and modification using microwave technology. Practically, materials should possess dielectric losses of greater than 0.01 for effective heating under the influence of microwaves.3,47 It was apparent from Figure 4ii and iv that lactose, paracetamol and acetylsalicylic acid barely fulfilled this criteria at all density levels and this could possibly account for their poor heating capabilities relative to the remaining materials. These results echoed earlier observations by several authors. The dielectrically inert character of lactose had been reported by McLoughlin et al.34 as well as Kudra et al.48 and was postulated to contribute to its physical stability DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

during microwave-assisted processing. This was verified in a recent study by Szepes et al.49 in which 20 min of continuous microwave irradiation of a-lactose monohydrate was found not to cause any significant alterations to its crystallinity, thermal properties and particle size. The low dielectric susceptibilities of acetylsalicylic acid and paracetamol had been similarly observed by McLoughlin et al.34 Based on a similar argument, these active ingredients may potentially be safeguarded against possible microwave-induced transformations during processing.

Correlation between Material Dielectric Properties and their Microwave-Induced Heating Capabilities The heating responses and calculated microwave dielectric losses of the materials correlated significantly with each other (Pearson correlation coefficient R ¼ 0.782 and 0.761 at LF 8.5 and 9, respectively, p < 0.01), which implied a close association between the two modes of dielectric assessment. It could be inferred from these results that the capacitance-based measurement of dielectric properties in which compacted forms of materials were exposed to rapidly alternating electric fields within the confines of two charged parallel electrodes may be likened to the microwave processing of material particulates in a cavity or enclosure. The latter functioned similarly as the pair of electrodes by providing boundaries for the electric field in which materials were interrogated or processed. Evidently, for successful development and optimization of microwave-assisted pharmaceutical processes, systematic dielectric evaluations of pharmaceutical materials are needed at the preformulation stages of product development. By performing dielectric measurements on small quantities of materials, their amenability to microwave processing may be readily assessed. Moreover, regression techniques could be employed in tracking the rapid shifts in dielectric properties of materials as they undergo density changes during processes such as mixing or granulation. Under circumstances where the high costs of instrumentation preclude the use of high frequency measurements, the laboratory microwave oven provides a reliable alternative for assessing the susceptibilities of materials to microwave energy. DOI 10.1002/jps

951

Relationship between the Moisture Contents and Dielectric Properties of Materials The enhanced dielectric susceptibility of moistened materials provides the fundamental basis for the use of microwaves in pharmaceutical drying operations. Drying performance is governed by the heating ability of the material which in turn is dependent on its dielectric loss at microwave frequencies. Figure 6 depicts qualitatively, the classical relationship between dielectric loss and moisture content of a material.3,8 Distinct inflexion points in the profile demarcate the transition between the changing states of water in the material. At low moisture contents, material dielectric loss is negligible as moisture in the material exists primarily in bound form on the solid surface and possesses limited mobility in the presence of electromagnetic waves. As its moisture content increases and attains the critical moisture content or Mc, all the available binding sites for water molecules become saturated. Further additions of water beyond Mc result in a population of water molecules bound to a lesser extent and which couple more readily with microwaves due to their greater rotational mobility. As the fraction of mobile water molecules increases further, material dielectric loss increases proportionately and may plateau off as it approaches that of free, bulk water at moisture contents of 20–30%. The dielectric loss versus moisture content profile reflects the drying behavior of a material subjected to microwave-assisted drying. At the

Figure 6. Classical relationship between moisture content and dielectric loss of a material. Mc refers to the critical moisture content of the material. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

952

HENG ET AL.

outset, it provides an indication of the moisture content required in the material before effective microwave-assisted moisture removal could occur. Above the Mc of the material, drying proceeds at a rapid rate due to the enhanced thermal properties of unbound moisture. The concentration of microwave energy in wetter regions of the material load enables rapid moisture leveling in the initial stages of drying. Towards the end of the process, drying ceases due to diminished interactions between microwaves and the low concentrations of bound moisture in the material. With these profiles, the changing dielectric properties and drying patterns of materials may be monitored in real time allowing for adaptive process control such as the gradual tuning down of microwave output towards the end of drying when the Mc of the material is attained. This minimizes unwarranted exposure of the material to microwave energy. The contribution of moisture content was analyzed with respect to the microwave dielectric losses of the materials in view of its greater relevance to microwave-assisted drying. As dielectric losses measured in this study were affected by simultaneous variations in the densities and moisture contents of materials, the contribution of moisture content was elucidated by examining the relationship between the moisture contents and dielectric properties of materials in their true solid states, where the effects of intra- and interparticulate air inclusions on dielectric properties could be eliminated. Figure 7 shows the relationship between the moisture contents and microwave dielectric losses of the different materials extrapolated to their respective true densities. In spite of the diverse compositions, structures and moisture-binding affinities of the materials, the effect of moisture content on microwave dielectric property was apparent in that materials with higher moisture contents generally exhibited higher dielectric losses at both microwave frequencies. Two distinct outliers, namely, anhydrous dicalcium phosphate and starch could be observed at the outset. They possessed unexpectedly higher dielectric losses than predicted based on their moisture contents. Materials with inherently low moisture contents such as lactose, paracetamol and acetylsalicylic acid possessed minimal dielectric losses close to 0. The subsequent increase in microwave dielectric losses assumed by the different materials with increasingly higher moisture contents followed a somewhat sigmoidal JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

Figure 7. Relationship between the moisture contents and microwave dielectric tosses of materials at their respective true densities at LF 8.5 (&) and 9 (*). Lac: lactose, DCP, anhydrous dicalcium phosphate; Sta, starch; Alg, sodium alginate; C15, PVP C15; K25, PVP K25; K29/32, PVP K29/32; K90D, PVP K90D; S630, PVP–VA S630; XL, cross-linked PVP XL; XL10, crosslinked PVP XL10; Para, paracetamol and ASA, acetylsalicylic acid.

profile (indicated by dotted lines) similar to that shown earlier in Figure 6. From the profiles, two groups of materials with different Mc values could be distinguished. The Mc values for PVP–VA S630, PVP K25, C15, K29/32, and sodium alginate were similar and approximated at 4–5% (w/w). For materials with higher moisture contents such as the cross-linked PVP(s) XL, XL10 and PVP K90D, their Mc values were higher and ranged from 9% to 10% (w/w). At this point, it may seem invalid to categorize and analyze materials of differing identities collectively. However, further experimental evidence provided the necessary justification for this approach. It had been previously shown in Figure 3ii that as field frequency increased to the microwave range (LF 8.5–9), the dielectric losses of the dried materials, namely, PVP–VA S630, PVP K25, C15, K29/32, K90D, cross-linked PVP(s) XL and XL10 approached each other and assumed values of 0.05 or less. This meant that at negligible moisture levels, the microwave dielectric losses of these materials were similar to each other and that of lactose, acetylsalicylic acid and paracetamol. Hence, their sigmoidal profiles as depicted in Figure 7 were likely to share similar points of origin. Figure 8 lent further support to the argument. It shows the relationship between the moisture contents and microwave-induced heating capabilities (DT) of the different materials DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

Figure 8. Relationship between the moisture contents and microwave-induced heating capabilities (DT) of the materials at their respective densities during testing. Lac, lactose; DCP, anhydrous dicalcium phosphate; Sta, starch; Alg, sodium alginate; C15, PVP C15; K25, PVP K25; K29/32, PVP K29/32; K90D, PVP K90D; S630, PVP–VA S630, XL, cross-linked PVP XL; XL10, cross-linked PVP XL10; Para, paracetamol and ASA, acetylsalicylic acid.

at their respective densities during testing. In spite of the differences in material physical properties, frequency and mode of measurement, the relationship that emerged resembled that of Figure 7. In general, materials with higher moisture contents exhibited improved microwave-induced heating capabilities. Anhydrous dicalcium phosphate and starch showed slight deviations from the trend and exhibited moderately higher heating capabilities than predicted based on their moisture contents. The remaining materials may similarly be classified into two groups with different Mc values. The Mc values of PVP–VA S630, PVP K25, C15, K29/32, and sodium alginate fell within 3–4% (w/w). The corresponding values for cross-linked PVP XL, XL10, and PVP K90D were in a higher range of 7–8% (w/w). As the assignment of Mc values to the different materials remained subjective at this point, the Mc of each material was verified using thermogravimetry. The specific Mc values are summarized in Figure 9. As expected, two distinct groups of materials could be identified, classified based on their different ranges of Mc. PVP–VA S630, PVP K25, C15, K29/32, and sodium alginate possessed lower Mc values ranging from 2.1% to 3.5% (w/w). DOI 10.1002/jps

953

Figure 9. Critical moisture contents (Mc) of materials as determined by thermo-gravimetric analysis. S630, PVP–VA S630; K25, PVP K25; C15, PVP C15; K29/32, PVP K29/32; Alg, sodium alginate; XL, cross linked PVP XL; XL10, cross-linked PVP XL10 and K90D, PVP K90D.

The Mc values of cross-linked PVP XL, XL10 and PVP K90D were in a higher range of 5.3–6.7% (w/ w). Generally, the trends observed from the dielectric measurements, heating experiments and thermo-gravimetric analyses corroborated with each other apart from some minor discrepancies. It appeared that the Mc values estimated from the dielectric measurements (Fig. 7) were slightly higher than those derived from the heating experiments (Fig. 8) and thermo-gravimetric analyses (Fig. 9). These disparities stemmed from the differences in physical properties of the material samples. Since water molecules partake in interparticulate bonding during compaction, compacted forms of materials such as those used in the dielectric studies would be expected to contain higher levels of bound moisture and exhibit higher Mc values as compared to their corresponding bulk forms employed in both the heating experiments and thermogravimetric analyses. The Mc estimates from the heating experiments were therefore closer approximations of those determined from thermogravimetric analyses. The disparities in bound moisture concentrations in different physical forms of an identical material may have affected the sensitivity of dielectric responses to moisture content. In the dielectric studies where moisture in the materials became bound to a greater extent due to compaction, a lower proportion of free moisture was available to interact with the applied microwave fields leading to a rapid tapering of material dielectric responses at higher moisture contents. On the other hand, for the corresponding materials tested in their bulk forms JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

954

HENG ET AL.

in the heating experiments, a comparatively lower concentration of bound moisture translated to increased availability of free moisture which interacted far more readily with microwaves, leading to an almost proportionate increase in dielectric response with moisture content upon saturation. From the results obtained thus far, it appeared that between material density and moisture content, the latter contributed more significantly to the variation exhibited in the microwave dielectric losses of lactose, sodium alginate, paracetamol, acetylsalicylic acid, the PVP binders and disintegrants. Specifically, it was the way moisture apportioned into bound and unbound forms in these materials that was responsible for such differences. In their dry states, these materials were likely to possess similar dielectric responses at microwave frequencies. On the other hand, it may be inferred that regardless of bulk density or moisture content, the innate molecular properties of anhydrous dicalcium phosphate and starch were conducive for interacting with and absorbing microwave energy. Literature abounds with examples of microwave–starch interactions. Less information however, is available as yet on the microwave susceptibility of anhydrous dicalcium phosphate. Anhydrous dicalcium phosphate comprises permanent dipoles bound in a crystalline lattice. From the results, it appeared that a large part of its dielectric susceptibility arose from its polarizable nature as exemplified by its high dielectric constant (Fig. 2i). As molecular orientations were unlikely due to its structural rigidity, the higher loss of anhydrous dicalcium phosphate arose from electronic and ionic polarization of the dipoles. The former occurred when electron clouds surrounding the nucleus were displaced with respect to the positive center under the influence of an external field. Alternatively, the positive and negative ions may have been displaced from their equilibrium positions by the external field which resulted in changes to their effective dipole moments. These mechanisms are known to dominate the dielectric permittivity of solids within the frequency range of 1 MHz–1 GHz.10

Density-Independent Function for Moisture Sensing Applications Process analytical technology has become the recent catchphrase in pharmaceutical processing. The ability to control and monitor pharmaceutical JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

processes online has turned the emerging concept of ‘‘quality by design’’ into a reality. Recent additions to the list of techniques for the control and monitoring of granulation and drying processes involve the application of microwaves for in situ moisture determination of wetted powders and granules during high shear granulation26 and fluidized bed drying,25 respectively. In the former, changes to the impeller speed generated varying material density patterns in the mixer. The agglomeration of particles inevitably reduced the effective area of contact between the probe and material with the consequence of diminished probe sensitivity. Particle fluidization in the fluidized-bed dryer also created complex and rapidly shifting material density patterns which affected dielectric property measurements. These confounding effects of density led to compromises on the accuracies of moisture measurement. This justified the need for density-independent functions to eliminate the effects of density on moisture determination. As a preliminary step necessary in guiding further use of density-independent functions in pharmaceutical processes, the effectiveness and applicability of a single density-indepffiffiffiffiffi pfrequency ffiffiffiffi pendent function "00 =ð 3 "0  1Þ as described by Powell et al.19 was evaluated. Compared to another commonly used single frequency expression "00 =ð"0  1Þ defined by Meyer and Schilz,18 this equation was selected on the basis that it provided better results in terms of density independence when applied to all materials testedp inffiffiffiffithis ffi pstudy. ffiffiffiffi For all materials, the magnitudes of "00 =ð 3 "0  1Þ were computed at both microwave frequencies and plotted against density (Fig. 10). In contrast to Figure 4, it could be observed that the plotted points were mostly parallel to the density axis at both frequencies, indicating that the chosen function possessed density-independent character. Slightly better results were obtained for materials such as starch, anhydrous dicalcium phosphate, sodium alginate and PVP K90D. Starch was selectedpas model material to test ffiffiffiffiffi a p ffiffiffiffi the sensitivity of "00 =ð 3 "0  1Þ to moisture variation. The drying curve, obtained by drying starch in a convection oven at 908C, is shown in Figure 11. The initial moisture content of starch was calculated based on the weight losses of the completely dried and cooled starch samples as described in the methods section. From Figure 11, the Mc of starch was determined and it referred to the moisture content at which the rate of moisture loss from starch began to decrease. The initial DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

Figure 10. character of the pffiffiffiffiffi Density-independent pffiffiffiffi function "00 =ð 3 "0  1Þ as applied to ( ) lactose, (~) anhydrous dicalcium phosphate, (!) starch, (*) sodium alginate, (&) PVP C15, (^) PVP K25, (5) PVP K29/32, (^) PVP K90D, (*) PVP–VA S630, (~) cross-linked PVP XL, ( ) cross-linked PVP XLIO, (&) paracetamol and ( ) acetylsalicylic acid at both LF 8.5 (i) and 9 (ii).

955

(moisture content of starch ¼ 9.40%, w/w, Tab. 1; Mc ¼ 3.1%, w/w). Physical characterization of the compacts prepared from starch sampled at the various drying time points revealed that their average densities spanned 0.9–1.2 g/mL. These values were within density ranges where pffiffiffiffiffi of pstarch ffiffiffiffi density-independence of "00 =ð 3 "0  1Þ was achieved (Fig. 10). starch sample, the pffiffiffiffiffiFor pffiffiffieach ffi average value of "00 =ð 3 "0  1Þ was calculated and plotted with respect to moisture content at both microwave frequencies (Fig. 12). At ffi pthe pffiffiffiffi ffiffiffiffi outset, the relationship between "00 =ð 3 "0  1Þ and moisture content seemed nonlinear. On closer examination however, it appeared that the association between these variables differed depending on the moisture content of starch. Below the Mc, a strong p linear ffiffiffiffi ffiffiffiffiffi prelationship could be observed between "00 =ð 3 "0  1Þ and moisture content with R2 values of 0.973 and 0.978 at LF8.5 and 9, respectively. When the moisture content of starch exceeded the Mc, a different relationship predominated which could not be accurately defined at this point of study. From these investigations, it appeared that the relevance of the selected density-independent equation for moisture sensing was confined within certain limits of material moisture content as dictated by the state of binding and resultant mobility of the associated water molecules. This is however not within the scope of the present study.

CONCLUSIONS

moisture content and Mc of starch were found to be 9.51% (w/w) and approximately 3% (w/w), respectively. These were consistent with the data obtained from thermo-gravimetric analysis

Dielectric properties described in terms of the dielectric constants and losses of 13 pharmaceutical materials comprising fillers, binders and active ingredients widely employed in the formulation of pharmaceutical dosage forms were

Figure 11. Drying curve obtained by drying starch at 908C in a convective oven.

Figure 12. Moisture content of the denpffiffiffiffiffi dependence pffiffiffiffi sity-independent function "00 =ð 3 "0  1Þ at LF 8.5 (&) and 9 (&) as applied to starch.

DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

956

HENG ET AL.

measured over a frequency range of 1 MHz–1 GHz at 23  18C using a parallel-electrode sample holder connected to an impedance analyzer. Dielectric constants of the materials decreased linearly but their losses varied considerably with field frequency. At microwave frequencies, linear relationships were established between cube-root pffiffiffiffi functions of both dielectric parameters ( 3 "0 and ffiffiffiffi ffi p 3 00 " ) and density, conforming to the Landau and Lifshitz, Looyenga mixture equation. This enabled the microwave dielectric properties of materials at varying densities to be estimated by regression. At equivalent density, the microwave dielectric losses of the materials corroborated with their heating abilities in a laboratory microwave oven. This indicated that the measured dielectric properties of materials provided accurate representations of their responses to microwaves under typical processing situations. Between material density and moisture content, the latter was the main factor responsible for the disparities in dielectric losses and heating capabilities of lactose, paracetamol, acetylsalicylic acid, sodium alginate, the PVP binders and disintegrants at microwave frequencies. Materials with higher contents of unbound moisture possessed improved microwave susceptibilities up to a certain limit depending on the mode of measurement employed. Starch and anhydrous dicalcium phosphate surfaced as potential ‘‘microwave absorbers’’ with innate molecular properties favorable for coupling with microwave energy. As a preliminary step towards the development of microwave techniques for moisture sensing in pharmaceutical processes, the density-independent character of the dielectric function ffiffiffiffiffi p p ffiffiffiffi "00 =ð 3 "0  1Þ was tested and found to be applicable to all materials tested in the study. Based on preliminary data obtained on starch, it appeared that use of the function was restricted to a narrow range of moisture content that was dependent on the specific moisture-binding abilities of the material.

ACKNOWLEDGMENTS The authors are deeply grateful to Professor Lim Hock and Dr. Kong Ling Bing of Temasek Laboratories, National University of Singapore, for their kind assistance and use of equipment for dielectric analysis as well as the funding support from the National University of Singapore Academic Research Fund (R-148-000-076-112). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010

REFERENCES 1. Venkatesh MS, Raghavan GSV. 2004. An overview of microwave processing and dielectric properties of agri-food materials. Biosyst Eng 88:1–18. 2. Craig DQM, editor. 1995. Dielectric analysis of pharmaceutical systems. London: Taylor and Francis. 3. Metaxas AC, Meredith RJ. 1983. Dielectric properties. In: Metaxas AC, Meredith RJ, editors. Industrial microwave heating. London: Peter Peregrinus Ltd. pp. 26–69. 4. I˙ c¸ier F, Baysal T. 2004. Dielectrical properties of food materials. 1. Factors affecting and industrial uses. Crit Rev Food Sci 44:465–471. 5. Debye PJW, editor. 1929. Polar molecules. New York: The Chemical Catalogue Co. 6. Nelson SO. 1994. Measurement of microwave dielectric properties of particulate materials. J Food Eng 21:365–384. 7. De Loor GP. 1968. Dielectric properties of heterogeneous mixtures containing water. J Microwave Power 3:67–73. 8. Schiffmann RF. 1995. Microwave and dielectric drying. In: Mujumdar AS, editor. Handbook of industrial drying. New York: Marcel Dekker. pp. 345–372. 9. Sirdeshmukh DB, Sirdeshmukh L, Subhadra KG. 2006. Dielectric and electrical properties of solids. In: Sirdeshmukh DB, Sirdeshmukh L, Subhadra KG, editors. Micro- and macro-properties of solids—Thermal, mechanical and dielectric properties. New York: Springer. pp. 199–255. 10. Frederikse HPR. 2008. Section 12 Polarizabilities of atoms and ions in solids. In: Lide DR, editor. Handbook of chemistry and physics. Ohio: CRC Press. pp. 13–14. 11. Parker TG. 1972. Dielectric properties of polymers II. In: Jenkins AD, editor. Polymer science. A materials science handbook. Amsterdam: North-Holland Publishing Company. pp. 1297–1327. 12. Schiffmann RF. 1986. Food product development for microwave processing. Food Technol 40:94–98. 13. Nelson SO. 1992. Estimation of permittivities of solids from measurements on pulverized or granular materials. In: Priou A, editor. Dielectric properties of heterogeneous materials. New York: Elsevier Science Publishing Co. pp. 231–267. 14. Nelson SO. 1992. Measurement and applications of dielectric properties of agricultural products. IEEE T Instrum Meas 41:116–122. 15. Nelson SO, Trabelsi S. 2004. Principles for microwave moisture and density measurement in grain and seed. J Microwave Power EE 39:107–117. 16. Kraszewski A. 1980. Microwave aquametry—A review. J Microwave Power 15:209–220. 17. Kent M, Meyer W. 1982. A density-independent microwave moisture meter for heterogeneous foodstuffs. J Food Eng 1:31–42. DOI 10.1002/jps

DIELECTRIC PROPERTIES OF PHARMACEUTICAL MATERIALS

18. Meyer W, Schilz W. 1980. A microwave method for density-independent determination of the moisture content of solids. J Phys D Appl Phys 13:1823–1830. 19. Powell SD, McLendon BD, Nelson SO, Kraszewski A, Allison JM. 1988. Use of a density-independent function and microwave measurement system for grain moisture measurement. Trans ASAE 31: 1875–1880. 20. Trabelsi S, Nelson SO. 1998. Density independent functions for on-line microwave moisture meters: A general discussion. Meas Sci Technol 9:570–578. 21. Kraszewski A, Trabelsi S, Nelson SO. 1998. Comparison of density-independent expressions for moisture content determination in wheat at microwave frequencies. J Agric Eng Res 71:227–237. 22. Shrestha BL, Wood HC, Sokhansanj S. 2005. Prediction of moisture content of Alfalfa using densityindependent functions of microwave dielectric properties. Meas Sci Technol 16:1179–1185. 23. Berbert PA, Stenning BC. 1996. Analysis of densityindependent equations for determination of moisture content of wheat in the radiofrequency range. J Agric Eng Res 65:275–286. 24. Chaplin G, Pugsley T. 2005. Application of electrical capacitance tomography to the fluidized bed drying of pharmaceutical granule. Chem Eng Sci 60:7022–7033. 25. Buschmu¨ ller C, Wiedey W, Do¨ scher C, Dressler J, Breitkreutz J. 2007. In-line monitoring of granule moisture in fluidized–bed dryers using microwave resonance technology. Eur J Pharm Biopharm 69: 380–387. 26. Gradinarsky L, Brage H, Lagerholm B, Bjo¨ rn IN, Folestad S. 2006. In situ monitoring and control of moisture content in pharmaceutical powder processes using an open-ended coaxial probe. Meas Sci Technol 17:1847–1853. 27. Bergese P, Colombo I, Gervasoni D, Depero LE. 2003. Microwave generated nanocomposites for making insoluble drugs soluble. Mater Sci Eng C 23:791–795. 28. Kercˇ J, Srcˇ icˇ S, Kofler B. 1998. Alternative solventfree preparation methods for felodipine surface solid dispersions. Drug Dev Ind Pharm 24:359–363. 29. Nurjaya S, Wong TW. 2005. Effects of microwave on drug release properties of matrices of pectin. Carbohyd Polym 62:245–257. 30. Vandelli MA, Romagnoli M, Monti A, Gozzi M, Guerra P, Rivasi F, Forni F. 2004. Microwave-treated gelatin microspheres as drug delivery system. J Control Release 96:67–84. 31. Wong TW, Chan LW, Kho SB, Heng PWS. 2002. Design of controlled release solid dosage forms of alginate and chitosan using microwave. J Control Release 84:99–114. 32. Wong TW, Chan LW, Kho SB, Heng PWS. 2005. Aging and microwave effects on alginate/chitosan matrices. J Control Release 104:461–475. DOI 10.1002/jps

957

33. McLoughlin CM, McMinn WAM, Magee TRA. 2000. Microwave drying of pharmaceutical powders. Trans IChemE 78:90–96. 34. McLoughlin CM, McMinn WAM, Magee TRA. 2003. Physical and dielectric properties of pharmaceutical powders. Powder Technol 134:40–51. 35. Gabriel C, Gabriel S, Grant EH, Halstead BSJ, Mingos DMP. 1998. Dielectric parameters relevant to microwave dielectric heating. Chem Soc Rev 27: 213–223. 36. Van Scoik KG, Zoglio MA, Carstensen JT. 1991. In: Swarbrick J, Boylan JC, editors. Encyclopedia of pharmaceutical technology, Vol. 4. New York: Marcel Dekker. pp. 485–515. 37. Heng PWS, Liew CV, Soh JLP. 2004. Pre-formulation studies on moisture absorption in microcrystalline cellulose using differential thermo-gravimetric analysis. Chem Pharm Bull 52:384–390. 38. Kinney GF, editor. 1957. Engineering properties and applications of plastics. New York: John Wiley and Sons Inc. 39. Mathes KN. 1988. Electrical properties. In: Kroschwitz JI, editor. Electrical and electronic properties of polymers: A state-of-the-art compendium. New York: John Wiley and Sons Inc. pp. 101–181. 40. Hedrick JL, Miller RD, Hawker CJ, Carter KR, Volksen W, Yoon DY, Trollsa˚ s M. 1998. Templating nanoporosity in thin-film dielectric insulators. Adv Mater 10:1049–1053. 41. Landau LD, Lifshitz EM, editors. 1960. Electrodynamics of continuous media. Oxford: Pergamon Press. 42. Looyenga H. 1965. Dielectric constants of heterogeneous mixtures. Physica 31:401–406. 43. Lewandowicz G, Fornal J, Walkowski A. 1997. Effect of microwave radiation on potato and tapioca starches. Carbohyd Polym 34:213–220. 44. Lewandowicz G, Jankowski T, Fornal J. 2000. Effect of microwave radiation on physico-chemical properties and structure of cereal starches. Carbohyd Polym 42:193–199. 45. Szepes A, Hasznos-Nezdei M, Kova´ cs J, Funke Z, Ulrich J, Szabo´ -Re´ ve´ sz P. 2005. Microwave processing of natural biopolymers—Studies on the properties of different starches. Int J Pharm 302:166–171. 46. Szepes A, Szabo´ -Re´ ve´ sz P. 2007. Water sorption behavior and swelling characteristics of starches subjected to dielectric heating. Pharm Dev Technol 12:555–561. 47. Orfeuil M, editor. 1987. Electric process heating: Technologies, equipments, applications. Ohio: Battelle Press. 48. Kudra T, Raghavan GSV, Akyel C, Bosisio R, Van De Foort FR. 1992. Electromagnetic properties of milk and its constituents at 2.45 GHz. J Microwave Power EE 27:199–204. 49. Szepes A, Fiebig A, Ulrich J, Szabo´ -Re´ ve´ sz P. 2007. Structural study of a-lactose monohydrate subjected to microwave irradiation. J Therm Anal Calorim 89:757–760. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010