- Email: [email protected]
Characterization of the thermal properties of microcrystalline cellulose by modulated temperature differential scanning calorimetry
Characterization of the Thermal Properties of Microcrystalline Cellulose by Modulated Temperature Differential Scanning Calorimetry KATHARINA M. PICKER,1 STEPHEN W. HOAG2 1
Martin-Luther-University Halle-Wittenberg, Institute of Pharmaceutical Technology and Biopharmacy, Wolfgang-Langenbeck-Str. 4, 06120 Halle/Saale, Germany 2
University of Maryland School of Pharmacy, 20 N. Pine Street, Baltimore, Maryland 21201
Received 6 August 2000; revised 20 August 2001; accepted 30 August 2001
ABSTRACT: The purpose of this study was to characterize the thermal properties of microcrystalline cellulose (MCC) and to investigate the in¯uence of water on these properties. Differential scanning calorimetry (DSC), modulated temperature differential scanning calorimetry (MTDSC), thermomechanical analysis (TMA), and scanning electron microscopy (SEM) were used to characterize MCC. Three reproducible step transitions were detected in the dry material at 132, 159, and 1848C; for these transitions the magnitude of the heat capacity change varied by a factor of two. Exposure of MCC to water lowers the transition temperature in a manner comparable to a glass transition. The effect of water was different for samples equilibrated to different atmospheric humidities versus water added by granulation. A change in the physical properties of MCC after granulation with high amounts of water was observed. In conclusion, it appears that MCC has glass transitions, which come in reproducible triplets, and these transitions are affected by the presence of water. Also, for the materials studied, the transition temperatures are not affected by particle size and pulp source. ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 91:342±349, 2002
Keywords: modulated temperature differential scanning calorimetry; glass transition temperature; microcrystalline cellulose; granulation; humidity
INTRODUCTION Microcrystalline cellulose (MCC) is a well-established and important tableting excipient.1 It is produced by severe acid hydrolysis of highly puri®ed a-celluloses, which have a low hemicellulose content. Different pulp sources are used to produce materials with different relative densities,2 and for each of these densities, different particle size materials are available. All the difCorrespondence to: Katharina M. Picker (Telephone: 49345-552 5138; Fax: 49-345-552 7022; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 91, 342±349 (2002) ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
342
ferent types of MCC contain both crystalline and amorphous regions. Using X-ray diffraction, several authors have determined the amount of amorphous material in MCC to be 30%,3±6 and it has been shown that there are no signi®cant differences in the degree of crystallinity among different commercial grades of MCC.3 The glass transition temperature (Tg) is the critical temperature at which the material properties of a polymer dramatically change and it is the characteristic temperature for amorphous and semicrystalline materials like MCC.7,8 Therefore, knowledge of the Tg is very important in the characterization of MCC. Stubberud et al.9 were not able to determine the Tg of MCC by differential scanning calorimetry (DSC) at standard
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
heating rates. Batzer and Kreibich10,11 found a Tg of 2308C for dry cellulose, which decreased with increasing amounts of water. Although cellulose has been extensively studied, at present, the nature of the glass transition of MCC has not been completely characterized, in part because the amorphous regions only comprise 30% of the total material. Modulated temperature differential scanning calorimetry (MTDSC) is a relatively new technique for characterizing the thermal behavior of materials. MTDSC makes it possible to separate different thermal events because this method superimposes a faster sinusoidal heating rate onto a slower underlying linear heating rate, thereby allowing the total heat ¯ow to be separated into the reversing heat ¯ow (heat capacity related events) and nonreversing heat ¯ow (timeand temperature-dependent events).12 In conventional DSC, a reversing event like the glass transition may be hidden by a nonreversing event, such as enthalpic relaxation. Thus, it is possible to detect hidden and weak glass transitions because the combination of fast instantaneous and slower underlying heating rates permits both high sensitivity and high resolution.13,14 In this context, a weak transition is a transition that can only be detected with a very sensitive method. Therefore, the aim of this study was to use the newer more sensitive method of MTDSC to determine the Tg of dry MCC and determine whether the Tg varies for different grades and lots of MCC. Standard DSC and thermomechanical analysis (TMA) were used to corroborate the MTDSC results. In addition, the in¯uence of water on the glass transition of MCC was studied; samples were either equilibrated at different relative humidities or granulated.
MATERIALS AND METHODS Materials The following grades and lots of microcrystalline cellulose were tested: Avicel PH 101 (lot # 1135, 1146, and 1605), Avicel PH 102 (lot # 2610 and 2710), Avicel PH 105 (lot # 5438), Avicel PH 200 (lot # M405C and M710C), Avicel PH 301 (lot # P509C), and Avicel PH 302 (lot # 2509C); all MCC samples were obtained from FMC Corp. (Princeton, NJ). Reagent grade potassium acetate and sodium chloride were obtained from Aldrich Che-
343
mical Company, Inc., (Milwaukee, WI) and Sigma Chemical Company (St. Louis, MO), respectively. Sample Preparation Dry MCC samples were obtained by oven drying at 1008C for 24 h. MCC samples were equilibrated in a constant humidity atmosphere for 1 week by placing a thin layer (< 0.5 cm) of the material in a dish inside a desiccator containing a saturated salt solution of CH3COOH for 23% relative humidity (RH), NaCl for 75% RH, or pure water for 100% RH. The granulation solvents used were water and pure isopropanol. For the granulations, 50, 80, 110, or 130% w/w water or 130% w/w isopropanol was mixed with MCC in a planetary mixer at 30-RPM (Erweka, Heusenstamm, Germany); the wet mass was then sieved through a 1.18-mm sieve. Differential Scanning Calorimetry Following sample preparation, the samples were immediately weighed, sealed in aluminum DSC pans, and analyzed. Loose powder was used to ensure that powder properties were not in¯uenced by compaction. The powder was equilibrated in desiccators of different humidities and then transferred as soon as possible into the pan, which was sealed immediately. Pinholed DSC pans were used. The samples were equilibrated in the DSC cell at 58C for 10 min and then heated to 2008C. In addition, the dry samples were initially equilibrated in the DSC cell at 1208C for 10 min to remove any residual moisture. For samples containing different amounts of water, hermetically sealed and nonhermatically sealed pans were tested. The hermetically sealed pans could not withstand the high pressures generated during heating, and the samples showed a sudden uncontrolled loss of water. Even special hermetic pans could not withstand the pressure. Furthermore, the size of the hermetic pans limited the sample size to 0.010 g, and it was found that at least 0.015 g was necessary for the MTDSC measurements. Therefore, nonhermatically sealed pans with a pinhole in the lid were used; for the non-anhydrous samples, these pans showed a continuous loss of water. This experimental protocol is in agreement with studies done by McPhillips et al.16,17 The sample size used was between 0.015 and 0.020 g; in this range, no in¯uence of sample size was observed. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
344
PICKER AND HOAG
MTDSC was performed using a TA Instruments model 2920 MTDSC (New Castle, DE). The calorimeter was calibrated for temperature and cell constant using indium (melting point, 156.618C; enthalpy of fusion, 28.71 J/g). Baseline calibration was performed by heating the empty cell. Heat capacity calibration was performed with a standard sapphire sample. This sample shows a very slight change in heat capacity at 1618C; however, the intensity of this change was small enough to be neglected for these studies. Different underlying heating rates (0.1± 5.08C/min) and modulation amplitudes ( 0.5± 38C) were studied to ®nd the best conditions as determined by quality of the sine wave from a graph of the modulated temperature heat ¯ow versus temperature. Based on prior experience, the modulation period was ®xed at 60 s. The ®nal optimized conditions were an underlying heating rate of 58C/min, an amplitude of 28C, and a period of 60 s. The neccessity for valid results to have 4±6 regular modulations with a transition was met. From the sine wave modulation it became obvious that no heating gradiants appeared to be in the sample. These special conditions, which are at the edge of the normal usage conditions, appeared to be the most ®tting for the special problem of studying very weak transitions. The heat capacity (reversing heat ¯ow) component was used to determine the Tg. The Tg was calculated from (1) the peak of the derivative of the reversing heat ¯ow and (2) the in¯ection point of the step transition in the reversing heat ¯ow. Calculating the derivative of the reversing heat ¯ow15 allows a more accurate determination of the Tg because the calculated temperature will be a peak that is easier to be determined than an in¯ection point; however, the difference between the two methods was minor. For standard DSC measurements, samples were analyzed using a Netzsch DSC (DSC 200, Netzsch GeraÈtebau, Selb, Germany). The sample size was 5 mg, and the samples were analyzed in standard pans without a pinhole in the lid. The dry samples were initially equilibrated in the DSC cell at 1208C for 10 min to remove any residual moisture, then they were cooled to 208C and heated to 2708C at a heating rate of 608C/min. As described by Jaffe et al.,18 this rapid heating rate was used to see if standard DSC could detect a transition. However, at these high heating rates, the data must be carefully interpreted. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
Thermomechanical Analysis Tablets consisting of 200 mg of MCC were prepared on a calibrated, instrumented eccentric press (EK0/DMS, No. 1.0083.92, Korsch GmbH, Berlin, Germany) with 11-mm diameter, ¯atfaced punches at a low maximum peak pressure of 1.5 kN. Tablets were made from oven-dried material at room conditions. The tablets analyzed by TMA (Netzsch GeraÈtebau, Selb, Germany) were heated at a rate of 58C/min to 2008C, cooled to 208C, and reheated a second time at the same rate. The onset of the Tg was determined from the peak of the second derivative of the expansion curve. Powder TMA was done using TA Instruments model 2940 TMA (New Castle, DE). Powder samples using a dilatometer sample holder were equilibrated at 1208C for 2 h, and a ramp of 2.008C/min to 2508C was used. Scanning Electron Microscopy Vacuum-oven-dried samples were mounted on a sample holder and coated with gold/palladium. The samples were examined with a scanning electron microscope (model JEOL JSM T-200, Tokyo, Japan) at an accelerating voltage of 25 keV.
RESULTS Glass Transition Temperatures of Dry MCC To determine the thermal properties of dry MCC, traditional DSC, MTDSC, TMA using a powder sample, and TMA using tablets were used. A typical MTDSC analysis of dried MCC is shown in Figure 1. The Tg determinations for the six different grades of MCC are summarized in Table 1. With dry MCC, there were three step transitions of which two were highly reproducible. Double or triple transitions could not be found with other materials using the same conditions, even if one might expect them a function of heating rate. The mean temperature of the strongest Tg [in this context, strongest refers to largest change in heat capacity at constant pressure (DCp) at Tg] was 159.8 0.88C. Another slightly weaker Tg was observed at 184.1 1.08C. A third glass transition at 132.7 4.18C was also detected, but this transition was weaker and showed a higher standard deviation. The strongest transition at 1608C had a baseline shift that was twice as large as the one at 1848C (see Fig. 1). As indicated in Table 1, there were no differences in Tg between the different types and lots of
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
Figure 1. Typical example of the glass transition temperatures for dried MCC determined by MTDSC: [1] reversed heat ¯ow, [2] derivative of reversed heat ¯ow.
material. This result means that there are no differences in Tg for materials of different particle sizes, like PH 101, 102, 105, and 200 or PH 301 and 302. PH 101, 102, 105, and 200 and PH 301 and 302 also contain different amounts of hemicelluloses (8%)2 and are made from different density wood pulps (Avicel PH 301 and 302 are produced from hardwood, all the other types are produced from softwood). Therefore, for the materials studied, it appears that the Tg of MCC is independent of cellulose source and particle size. For the standard DSC (see Fig. 2), these curves show a glass transition. The average of eight replications was 173.8 11.98C. As is well known with DSC, the faster the heating rate, the greater the sensitivity but the lower the resolu-
345
tion. Thus, ®nding only one transition at this high heating rate is not unexpected. Even though, at this high heating rate, the standard deviation was relatively large, these data still show that there is a glass transition and it is < 2008C. The same result could be obtained by MTDSC at a heating rate of 2.58C. The triplet of transitions is hidden when sensitivity is less. TMA was performed on powders and tablets. Tablets were made at very low pressures (1.5 kN) to minimize the in¯uence of compression. In Figure 3, the results for a typical tablet produced from dried MCC are given. The results show a change in baseline occurring; the mean values and standard deviations of ®ve replicates are 157.3 2.4, 170.3 2.8, and 179.4 4.38C; these values correlate with the strong transitions determined by MTDSC for the dry material. For powders, the TMA results showed a single Tg at 173.188C. In summary, the determination of Tg is dependent on the kinetics of heating. Thus, it is not surprising that the results vary for the different methods. Glass Transition Temperature of Moisture Equilibrated MCC Analysis of MCC PH 102 (lot # 2710), which was equilibrated at relative humidities of 22, 75, and 100%, showed 2±4 weaker transitions between 40 and 1008C; these transitions were not observed in the dry material. The temperatures were similar for the different equilibration humidities. The Tg
Table 1. The Glass Transitions Temperatures for Dry MCC Type
Lot #
Temp. 1 (8C)
Temp. 2 (8C)
Temp. 3 (8C)
MCC 101
1135
136.88 130.86 122.62 130.12 7.16 132.50 133.25 126.32 137.06 133.54 136.40 130.12 137.32 133.73 131.41 134.55 133.23 1.63 133.34 132.66 4.07
160.36 159.34 159.40 159.68 0.59 159.98 160.65 159.73 159.93 160.26 159.84 159.80 159.36 159.16 159.25 159.88 159.43 0.39 160.44 159.83 0.83
185.90 184.53 186.93 185.79 1.20 186.33 185.79 183.55 184.76 185.76 184.33 185.27 185.25 185.42 183.44 184.18 184.35 1.00 184.92 184.09 0.97
Average SD MCC 102 MCC 105 MCC 200
1146 1605 2710 2610 M405c M710c
MCC 301
P509c
Average SD MCC 302 Total average SD
2509c
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
346
PICKER AND HOAG
Figure 2. Typical example of the glass transition temperatures for dried MCC determined by standard DSC, with three replicates showing a glass transition; for curve 1, Tg is 185.58C, for curve 2, Tg is 186.88C, and for curve 3, Tg is178.08C determined by the derivative of the curve.
of the equilibrated samples cannot be precisely quantitated because non-hermetic pans were used; thus, the sample moisture content is not accurately known during analysis due to water loss. The lowering of Tg by water can be estimated by the Fox equation:19 1 w1 w2 Tg Tg1 Tg2
1
where, Tg, Tg1, and Tg2 are the glass transition temperatures of the polymer/water, polymer, and water, respectively; and w1 and w2 are the weight fractions of the polymer and water, respectively. The predicted results indicate that there should be three Tg temperatures between 2 and 808C for a 30% amorphous material equilibrated between 75 and 100% RH, which is in agreement with our
Figure 3. Typical example of the glass transition temperature of dried MCC determined by TMA: [1] relative change in length of the compact (Ð), [2] second derivative versus temperature of change of the length of the compact (D2dL - -). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
®ndings. The fourth transition exhibited by some samples could be explained by considering that the same material could have undergone a second transition, due to the loss of water, and hence a raising of the Tg to the analysis temperature. The existence of these lower temperature transitions in equilibrated samples, which were not present in dry samples, indicates these transitions behave like glass transitions. In addition, these results are in general agreement with other authors who also found that water lowered the Tg by 1008C for poly(vinyl pyrrolidone) (PVP)20 and other cellulosic polymers.21,22 Glass Transition Temperatures of Granulated MCC Several glass transitions were observed in the granulated samples of Avicel PH 102; all granulation samples were run in triplicate. Granulating with 50 or 80% w/w water shows results similar to the equilibrated samples already described [140.69 1.21, 161.31 1.34, and 183.92 1.318C for 50 % (w/w); 148.88 2.97, 162.14 2.37, and 180.84 1.328C for 80% (w/w)]. Additionally, a weak Tg was observed at 115.4 4.28C. Based on a rough estimate of water loss, this glass transition is occurring in a material that has lost about half of the initial water content, which means the samples had a water content of 25± 40% w/w. The granules granulated with 110 or 130% w/w water also show transitions below 1008C. The transitions of granules after drying were not as pronounced, however the transition temperatures were still the same [151.03 2.83, 162.89 1.82, and 181.73 1.07 for 110% w/w; 149.67 2.75, 162.31 0.96, and 183.39 2.66 for 130% w/w]. In addition, other transitions were detected between 100 and 1308C [114.24 3.778C for 110% w/w and 112.67 1.00, 123.61 2.46, and 130.22 1.278C for 130% w/w]. These transitions occur when about half of the water is lost, which means that the total material contains 55 or 65% w/w water. For granules produced with isopropanol and for granules dried prior to analysis, only transitions at 133, 160, and 1848C were observed, and there were no glass transitions for these granules at lower temperatures. Scanning Electron Microscopy Representative SEM photographs of neat and granulated MCC are shown in Figures 4a±d. When compared with ungranulated material (Fig. 4a), the 80% w/w material (Fig. 4b) shows a
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
347
slight texture change whereas the 130% w/w material (Fig. 4c) shows larger structural changes that are due to the interactions with water. The material granulated with 130% w/w isopropanol (Fig. 4d) appears to be similar to ungranulated MCC. It is interesting to note that after drying, the granules produced with water and isopropanol and the dry MCC all have the same original Tg. Thus, these physical changes induced by granulation (see Fig. 4) do not appear to be responsible for the glass transitions of MCC.
DISCUSSION The classic de®nition of a phase is a homogenous distinct region of a system separated by a boundary. However, a more contemporary de®nition of a phase is based on the degree or amplitude of molecular motions a particular system is permitted in a given equilibrium or nonequilibrium state.23 Also, the glass transition is now considered much more complicated than previously thought. Going from the glassy state to the rubbery state was once considered a monolithic event involving a global mobility change, but more recent research has shown the transition to be much more complicated. For example, instead of a complete change in polymer mobility, now there are long- and short-range mobility changes and different degrees of cooperative and noncooperative motion.24 Thus, the transition has gone from being considered a single event to one of several possible relaxation events involving, for example, d-, g-, b-, and Tl±l transitions.25 In addition, instead of a single rubbery state, there are several possible mesophases into which the glass can transform,23 and some of these mesophases have been found in cellulosic systems.26 MCC appears to have glass transitions that come in triplets and at temperatures lower than that of cellulose (Tg of dry cellulose is 2308C10,11). Given the classic de®nition of a phase transition, these multiple transitions are dif®cult to explain; however, when considering all the transitions and molecular motions possible in MCC, these results are feasible. The authors believe the multiple transitions at lower temperatures are a result of Figure 4. Scanning electron micrographs of MCC for (a) the ungranulated material, (b) a material granulated with 80% w/w water, (c) a material granulated with 130% w/w water, and (d) a material granulated with 130% w/w isopropanol. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
348
PICKER AND HOAG
structural changes that occur during the MCC manufacturing process. To produce MCC, the cellulose ®bers undergo a controlled hydrolysis in a strong acid solution. This process preferentially reduces the molecular weight of the amorphous regions to a degree of depolymerization of 100±200.27 This depolymerization of the amorphous regions increases the number of end groups in these amorphous regions, which is associated with an increase in free volume. Thus, the authors believe the lowering of molecular weight via acid hydrolysis is responsible for lowering Tg. Discussed next are some possible explanations for the multiple transitions, which the authors separate into two hypotheses: (1) amorphous structural heterogeneity and secondary structural rearrangements and (2) nano- and mesophase transitions. First, structural heterogeneity, in which regions of the MCC have different thermal properties, could account for the multiple transitions observed. For lower molecular weights, a key determinant of Tg is polymer molecular weight. It is possible that during the acid hydrolysis of cellulose, certain regions may be preferentially hydrolyzed. Hence, regions of different molecular weight could have been formed, which give rise to different transition temperatures. In addition to differences in molecular weight, there could be differences in chemical structure. For example, in MCC, carboxyl groups from processing can be found;28 for one microcrystal, 6600 carboxyl groups were found via titration with ammonium cations, most of which are located at the C6 position of the glucose unit.28 Also, it is very likely that aldehyde groups form during processing with acid.29 These carboxyl and aldehyde groups are probably located in the amorphous regions of MCC where the cellulose chains are more susceptible to acid hydrolysis, and for cellulose ethers different substituted groups are known to in¯uence Tg. It is possible that the different substituents are not uniformly distributed throughout the amorphous regions, which could result in heterogeneous regions each with their own Tg. In addition, native and some processed cellulose ®bers are known to have an extended-chain ®brous structure, and these microscopic, highly oriented ®bers are called micro®brils.26 It is possible that these micro®brils could rearrange on heating, which could account for some of the observed transitions. Secondly, Chen and Wunderlich24 discuss the existence of different nano- and mesophases and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
how these different phases can result in multiple transitions. Beiner et al.30 found multiple Tgs in methacrylate-based homopolymers. For MCC, the three glass transitions could exist because of nanophase separation or because the semicrystalline nature of the MCC may create mesophases between the crystalline and amorphous regions, which could have a different Tg values. The Tg of an amorphous material is lowered by water.20±22 This effect has been shown for PVP,20 cellulose ethers,22 and other amorphous materials.21 Water can exist in different thermodynamic states in which it may be bound differently.31,32 Thus, it can in¯uence the molecular mobility; especially the mobility of ends groups and side chain groups, as described by Ahlgren.33 The fact that the Tgs were lowered for the materials exposed to water, coupled with the TMA data, indicates that the transitions were glass transitions. Granules made with isopropanol seem to be unaffected by the solvent when compared with granules made with water. Whereas water clearly interacts with MCC, isopropanol does not appear to because of its polarity and bonding ability. Thus, the fact that the material does not change is a good control for the effect of a solvent on the transitions in MCC. In conclusion, the transitions in MCC are numerous, complex, and subtle. Most of them are weak transitions resulting in a minor baseline shift. Although the shifts are reproducible and measurable, their weak character is open to multiple interpretation. Turi and Chartoff34 present the current state of data on the subject of weak transitions, discussing the existence of transitions lower than Tg. Because the primary factors affecting Tg for semi-crystalline polymeric materials are the chemical composition of a polymer molecule and molecular weight of the amorphous regions, we need to know more about the microstructure of MCC before de®nitive conclusions can be made.
ACKNOWLEDGMENTS Katharina M. Picker thanks the Martin-LutherUniversity Halle-Wittenberg, and especially Prof. Dr. habil. G. Zessin for providing the time to spend at the University of Maryland, School of Pharmacy. In addition, the authors thank Nasser N. Nyamweya for his technical and intellectual assistance during this project.
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
REFERENCES 1. Reier GE, Shangraw RF. 1966. Microcrystalline cellulose in tableting. J Pharm Sci 55:510±514. 2. Reier GE. Personal communication, 1997. 3. Rowe RC, McKillopp AG, Bray D. 1994. The effect of batch and source variation on the crystallinity of microcrystalline cellulose. Int J Pharm 101:169± 172. 4. Kleinebudde P. 1997. The crystallit-gel-model for microcrystalline cellulose in wet-granulation, extrusion and spheronization. Pharm Res 14:804±809. 5. Nakai Y, Fukuoka E, Nakajima S, Hasegawa J. 1977. Crystallinity and physical characteristics of microcrystalline cellulose. Chem Pharm Bull 25:96±101. 6. Nakai Y, Fukuoka E, Nakajima S, Yamamoto K. 1977. Crystallinity and physical characteristics of microcrystalline cellulose. II. Fine structure of ground microcrystalline cellulose. Chem Pharm Bull 25:2490±2496. 7. Hancock BC, Shamblin SL, Zogra® G. 1995. Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures. Pharm Res 12:799±806. 8. Hancock BC, Zogra® G. 1997. Characteristics and signi®cance of the amorphous state in pharmaceutical systems. J Pharm Sci 86:1±12. 9. Stubberud L, Arwidson HG, Larsson A, Graffner C. 1996. Water solid interactions. II. Effect of moisture sorption and glass transition temperature on compactibility of microcrystalline cellulose alone or in binary mixtures with polyvinyl pyrrolidone. Int J Pharm 134:79±88. 10. Batzer H, Kreibich UT. 1981. In¯uence of water on thermal transitions in natural polymers and synthetic polyamides. Polym Bull 5:585±590. 11. Salmen NL, Back EL. 1977. The in¯uence of water on the glass transition temperature of cellulose. TAPPI J 60:137. 12. Wunderlich B, Okazaki I. 1997. Modulated differential scanning calorimetry in the glass transition region. VI. Model calculations based on poly(ethylene terephthalate). J Therm Anal 49:57±70. 13. Reading M, Elliott D, Hill VL. 1993. A new approach to the calorimetric investigation of physical and chemical transitions. J Therm Anal 40:949±955. 14. Wunderlich B, Boller A, Okazaki I, Kreitmeier S. 1996. Modulated differential scanning calorimetry in the glass transition region. II. The mathematical treatment of the kinetics of the glass transition. J Therm Anal 47:1013±1026. 15. Hourston DJ, Song M, Pollock HM, Hammiche A. 1997. Modulated differential scanning calorimetry. III. Applications to measurements of the glass transition temperature and increment of heat capacity. J Therm Anal 49:209±218. 16. McPhillips H, Craig DQM, Royall PG, Hill VL. 1999. Characterisation of the glass transition of
17. 18. 19. 20.
21.
22.
23. 24. 25. 26. 27.
28. 29. 30.
31. 32. 33. 34.
349
HPMC using modulated temperature differential scanning calorimetry. Int J Pharm 180:83±90. Hill VL, Craig DQM, Feely LC. 1999. The effects of experimental parameters and calibration on MTDSC data. Int J Pharm 192:21±32. Jaffe M, Menczel JD, Bessey WE. 1997. In: Turi EA, editor. Fibers. San Diego: Academic Press, pp. 1767±1954. Fox TG. 1956. In¯uence of diluent and of copolymer composition on the glass transition temperature of a polymer system. Bull Am Phys Soc 1:123. Oksanen CA, Zogra® G. 1990. The relationship between the glass transition temperature and water vapor absorption by poly(vinylpyrrolidone). Pharm Res 7:654±657. Hancock BC, Zogra® G. 1994. The relationship between glass transition temperature and the water content of amorphous solids. Pharm Res 11:471±477. Picker KM, Mielck JB. 1998. Effect of relative humidity during tabletting on matrix formation of hydrocolloids: Densi®cation behavior of cellulose ethers. Pharm Technol Dev 3:31±41. Wunderlich B. 1999. A classi®cation of molecules, phases, and transitions as recognized by thermal analysis. Thermochim Acta 340-1:37±52. Chen W, Wunderlich B. 1999. Nanophase separation of small and large molecules. Macromol Chem Phys 200:283±311. Egorov VM. 1994. Differential scanning calorimetry of polymers, physics, chemistry, analysis, technology. New York: Ellis Horwood, p. 253. Gilbert RD, Kadla JF. 1998. In: Kaplan DL, editor. Biopolymers from renewable sources. Berlin: Springer, pp. 47±95. Doelker E, Gurny R, Schurz J, Janosi A, Matin N. 1987. Degrees of crystallinity and polymerization of modi®ed cellulose powders for direct tableting. Powder Technol 52:207±213. Edelson MR, Hermans J Jr. 1963. Flow of gels of cellulose microcrystals. II. Effect of added electrolyte. J Polym Sci Part C 2:145±152. Fleck E. Personal communication, 1997. Beiner M, SchroÈter K, Hempel E, Reissig S, Donth E. 1999. Multiple glass transition and nanophase separation in poly(n-alkyl methacrylate) homopolymers. Macromolecules 32:6278±6282. Zogra® G. 1988. States of water associated with solids. Drug Dev Ind Pharm 14:1905±1926. Zogra® G, Kontny MJ. 1986. The interactions of water with cellulose- and starch-derived pharmaceutical excipients. Pharm Res 3:187±194. Ahlgren P. 1995. The application of microcrystalline cellulose in pharmaceutical tablet-making. Nordic Pulp Paper Res J 1:12±16. Chartoff RP. 1997. In: Turi EA, editor. Thermoplastic polymers. San Diego: Academic Press, pp. 483±743.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
Martin-Luther-University Halle-Wittenberg, Institute of Pharmaceutical Technology and Biopharmacy, Wolfgang-Langenbeck-Str. 4, 06120 Halle/Saale, Germany 2
University of Maryland School of Pharmacy, 20 N. Pine Street, Baltimore, Maryland 21201
Received 6 August 2000; revised 20 August 2001; accepted 30 August 2001
ABSTRACT: The purpose of this study was to characterize the thermal properties of microcrystalline cellulose (MCC) and to investigate the in¯uence of water on these properties. Differential scanning calorimetry (DSC), modulated temperature differential scanning calorimetry (MTDSC), thermomechanical analysis (TMA), and scanning electron microscopy (SEM) were used to characterize MCC. Three reproducible step transitions were detected in the dry material at 132, 159, and 1848C; for these transitions the magnitude of the heat capacity change varied by a factor of two. Exposure of MCC to water lowers the transition temperature in a manner comparable to a glass transition. The effect of water was different for samples equilibrated to different atmospheric humidities versus water added by granulation. A change in the physical properties of MCC after granulation with high amounts of water was observed. In conclusion, it appears that MCC has glass transitions, which come in reproducible triplets, and these transitions are affected by the presence of water. Also, for the materials studied, the transition temperatures are not affected by particle size and pulp source. ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association J Pharm Sci 91:342±349, 2002
Keywords: modulated temperature differential scanning calorimetry; glass transition temperature; microcrystalline cellulose; granulation; humidity
INTRODUCTION Microcrystalline cellulose (MCC) is a well-established and important tableting excipient.1 It is produced by severe acid hydrolysis of highly puri®ed a-celluloses, which have a low hemicellulose content. Different pulp sources are used to produce materials with different relative densities,2 and for each of these densities, different particle size materials are available. All the difCorrespondence to: Katharina M. Picker (Telephone: 49345-552 5138; Fax: 49-345-552 7022; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 91, 342±349 (2002) ß 2002 Wiley-Liss, Inc. and the American Pharmaceutical Association
342
ferent types of MCC contain both crystalline and amorphous regions. Using X-ray diffraction, several authors have determined the amount of amorphous material in MCC to be 30%,3±6 and it has been shown that there are no signi®cant differences in the degree of crystallinity among different commercial grades of MCC.3 The glass transition temperature (Tg) is the critical temperature at which the material properties of a polymer dramatically change and it is the characteristic temperature for amorphous and semicrystalline materials like MCC.7,8 Therefore, knowledge of the Tg is very important in the characterization of MCC. Stubberud et al.9 were not able to determine the Tg of MCC by differential scanning calorimetry (DSC) at standard
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
heating rates. Batzer and Kreibich10,11 found a Tg of 2308C for dry cellulose, which decreased with increasing amounts of water. Although cellulose has been extensively studied, at present, the nature of the glass transition of MCC has not been completely characterized, in part because the amorphous regions only comprise 30% of the total material. Modulated temperature differential scanning calorimetry (MTDSC) is a relatively new technique for characterizing the thermal behavior of materials. MTDSC makes it possible to separate different thermal events because this method superimposes a faster sinusoidal heating rate onto a slower underlying linear heating rate, thereby allowing the total heat ¯ow to be separated into the reversing heat ¯ow (heat capacity related events) and nonreversing heat ¯ow (timeand temperature-dependent events).12 In conventional DSC, a reversing event like the glass transition may be hidden by a nonreversing event, such as enthalpic relaxation. Thus, it is possible to detect hidden and weak glass transitions because the combination of fast instantaneous and slower underlying heating rates permits both high sensitivity and high resolution.13,14 In this context, a weak transition is a transition that can only be detected with a very sensitive method. Therefore, the aim of this study was to use the newer more sensitive method of MTDSC to determine the Tg of dry MCC and determine whether the Tg varies for different grades and lots of MCC. Standard DSC and thermomechanical analysis (TMA) were used to corroborate the MTDSC results. In addition, the in¯uence of water on the glass transition of MCC was studied; samples were either equilibrated at different relative humidities or granulated.
MATERIALS AND METHODS Materials The following grades and lots of microcrystalline cellulose were tested: Avicel PH 101 (lot # 1135, 1146, and 1605), Avicel PH 102 (lot # 2610 and 2710), Avicel PH 105 (lot # 5438), Avicel PH 200 (lot # M405C and M710C), Avicel PH 301 (lot # P509C), and Avicel PH 302 (lot # 2509C); all MCC samples were obtained from FMC Corp. (Princeton, NJ). Reagent grade potassium acetate and sodium chloride were obtained from Aldrich Che-
343
mical Company, Inc., (Milwaukee, WI) and Sigma Chemical Company (St. Louis, MO), respectively. Sample Preparation Dry MCC samples were obtained by oven drying at 1008C for 24 h. MCC samples were equilibrated in a constant humidity atmosphere for 1 week by placing a thin layer (< 0.5 cm) of the material in a dish inside a desiccator containing a saturated salt solution of CH3COOH for 23% relative humidity (RH), NaCl for 75% RH, or pure water for 100% RH. The granulation solvents used were water and pure isopropanol. For the granulations, 50, 80, 110, or 130% w/w water or 130% w/w isopropanol was mixed with MCC in a planetary mixer at 30-RPM (Erweka, Heusenstamm, Germany); the wet mass was then sieved through a 1.18-mm sieve. Differential Scanning Calorimetry Following sample preparation, the samples were immediately weighed, sealed in aluminum DSC pans, and analyzed. Loose powder was used to ensure that powder properties were not in¯uenced by compaction. The powder was equilibrated in desiccators of different humidities and then transferred as soon as possible into the pan, which was sealed immediately. Pinholed DSC pans were used. The samples were equilibrated in the DSC cell at 58C for 10 min and then heated to 2008C. In addition, the dry samples were initially equilibrated in the DSC cell at 1208C for 10 min to remove any residual moisture. For samples containing different amounts of water, hermetically sealed and nonhermatically sealed pans were tested. The hermetically sealed pans could not withstand the high pressures generated during heating, and the samples showed a sudden uncontrolled loss of water. Even special hermetic pans could not withstand the pressure. Furthermore, the size of the hermetic pans limited the sample size to 0.010 g, and it was found that at least 0.015 g was necessary for the MTDSC measurements. Therefore, nonhermatically sealed pans with a pinhole in the lid were used; for the non-anhydrous samples, these pans showed a continuous loss of water. This experimental protocol is in agreement with studies done by McPhillips et al.16,17 The sample size used was between 0.015 and 0.020 g; in this range, no in¯uence of sample size was observed. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
344
PICKER AND HOAG
MTDSC was performed using a TA Instruments model 2920 MTDSC (New Castle, DE). The calorimeter was calibrated for temperature and cell constant using indium (melting point, 156.618C; enthalpy of fusion, 28.71 J/g). Baseline calibration was performed by heating the empty cell. Heat capacity calibration was performed with a standard sapphire sample. This sample shows a very slight change in heat capacity at 1618C; however, the intensity of this change was small enough to be neglected for these studies. Different underlying heating rates (0.1± 5.08C/min) and modulation amplitudes ( 0.5± 38C) were studied to ®nd the best conditions as determined by quality of the sine wave from a graph of the modulated temperature heat ¯ow versus temperature. Based on prior experience, the modulation period was ®xed at 60 s. The ®nal optimized conditions were an underlying heating rate of 58C/min, an amplitude of 28C, and a period of 60 s. The neccessity for valid results to have 4±6 regular modulations with a transition was met. From the sine wave modulation it became obvious that no heating gradiants appeared to be in the sample. These special conditions, which are at the edge of the normal usage conditions, appeared to be the most ®tting for the special problem of studying very weak transitions. The heat capacity (reversing heat ¯ow) component was used to determine the Tg. The Tg was calculated from (1) the peak of the derivative of the reversing heat ¯ow and (2) the in¯ection point of the step transition in the reversing heat ¯ow. Calculating the derivative of the reversing heat ¯ow15 allows a more accurate determination of the Tg because the calculated temperature will be a peak that is easier to be determined than an in¯ection point; however, the difference between the two methods was minor. For standard DSC measurements, samples were analyzed using a Netzsch DSC (DSC 200, Netzsch GeraÈtebau, Selb, Germany). The sample size was 5 mg, and the samples were analyzed in standard pans without a pinhole in the lid. The dry samples were initially equilibrated in the DSC cell at 1208C for 10 min to remove any residual moisture, then they were cooled to 208C and heated to 2708C at a heating rate of 608C/min. As described by Jaffe et al.,18 this rapid heating rate was used to see if standard DSC could detect a transition. However, at these high heating rates, the data must be carefully interpreted. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
Thermomechanical Analysis Tablets consisting of 200 mg of MCC were prepared on a calibrated, instrumented eccentric press (EK0/DMS, No. 1.0083.92, Korsch GmbH, Berlin, Germany) with 11-mm diameter, ¯atfaced punches at a low maximum peak pressure of 1.5 kN. Tablets were made from oven-dried material at room conditions. The tablets analyzed by TMA (Netzsch GeraÈtebau, Selb, Germany) were heated at a rate of 58C/min to 2008C, cooled to 208C, and reheated a second time at the same rate. The onset of the Tg was determined from the peak of the second derivative of the expansion curve. Powder TMA was done using TA Instruments model 2940 TMA (New Castle, DE). Powder samples using a dilatometer sample holder were equilibrated at 1208C for 2 h, and a ramp of 2.008C/min to 2508C was used. Scanning Electron Microscopy Vacuum-oven-dried samples were mounted on a sample holder and coated with gold/palladium. The samples were examined with a scanning electron microscope (model JEOL JSM T-200, Tokyo, Japan) at an accelerating voltage of 25 keV.
RESULTS Glass Transition Temperatures of Dry MCC To determine the thermal properties of dry MCC, traditional DSC, MTDSC, TMA using a powder sample, and TMA using tablets were used. A typical MTDSC analysis of dried MCC is shown in Figure 1. The Tg determinations for the six different grades of MCC are summarized in Table 1. With dry MCC, there were three step transitions of which two were highly reproducible. Double or triple transitions could not be found with other materials using the same conditions, even if one might expect them a function of heating rate. The mean temperature of the strongest Tg [in this context, strongest refers to largest change in heat capacity at constant pressure (DCp) at Tg] was 159.8 0.88C. Another slightly weaker Tg was observed at 184.1 1.08C. A third glass transition at 132.7 4.18C was also detected, but this transition was weaker and showed a higher standard deviation. The strongest transition at 1608C had a baseline shift that was twice as large as the one at 1848C (see Fig. 1). As indicated in Table 1, there were no differences in Tg between the different types and lots of
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
Figure 1. Typical example of the glass transition temperatures for dried MCC determined by MTDSC: [1] reversed heat ¯ow, [2] derivative of reversed heat ¯ow.
material. This result means that there are no differences in Tg for materials of different particle sizes, like PH 101, 102, 105, and 200 or PH 301 and 302. PH 101, 102, 105, and 200 and PH 301 and 302 also contain different amounts of hemicelluloses (8%)2 and are made from different density wood pulps (Avicel PH 301 and 302 are produced from hardwood, all the other types are produced from softwood). Therefore, for the materials studied, it appears that the Tg of MCC is independent of cellulose source and particle size. For the standard DSC (see Fig. 2), these curves show a glass transition. The average of eight replications was 173.8 11.98C. As is well known with DSC, the faster the heating rate, the greater the sensitivity but the lower the resolu-
345
tion. Thus, ®nding only one transition at this high heating rate is not unexpected. Even though, at this high heating rate, the standard deviation was relatively large, these data still show that there is a glass transition and it is < 2008C. The same result could be obtained by MTDSC at a heating rate of 2.58C. The triplet of transitions is hidden when sensitivity is less. TMA was performed on powders and tablets. Tablets were made at very low pressures (1.5 kN) to minimize the in¯uence of compression. In Figure 3, the results for a typical tablet produced from dried MCC are given. The results show a change in baseline occurring; the mean values and standard deviations of ®ve replicates are 157.3 2.4, 170.3 2.8, and 179.4 4.38C; these values correlate with the strong transitions determined by MTDSC for the dry material. For powders, the TMA results showed a single Tg at 173.188C. In summary, the determination of Tg is dependent on the kinetics of heating. Thus, it is not surprising that the results vary for the different methods. Glass Transition Temperature of Moisture Equilibrated MCC Analysis of MCC PH 102 (lot # 2710), which was equilibrated at relative humidities of 22, 75, and 100%, showed 2±4 weaker transitions between 40 and 1008C; these transitions were not observed in the dry material. The temperatures were similar for the different equilibration humidities. The Tg
Table 1. The Glass Transitions Temperatures for Dry MCC Type
Lot #
Temp. 1 (8C)
Temp. 2 (8C)
Temp. 3 (8C)
MCC 101
1135
136.88 130.86 122.62 130.12 7.16 132.50 133.25 126.32 137.06 133.54 136.40 130.12 137.32 133.73 131.41 134.55 133.23 1.63 133.34 132.66 4.07
160.36 159.34 159.40 159.68 0.59 159.98 160.65 159.73 159.93 160.26 159.84 159.80 159.36 159.16 159.25 159.88 159.43 0.39 160.44 159.83 0.83
185.90 184.53 186.93 185.79 1.20 186.33 185.79 183.55 184.76 185.76 184.33 185.27 185.25 185.42 183.44 184.18 184.35 1.00 184.92 184.09 0.97
Average SD MCC 102 MCC 105 MCC 200
1146 1605 2710 2610 M405c M710c
MCC 301
P509c
Average SD MCC 302 Total average SD
2509c
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
346
PICKER AND HOAG
Figure 2. Typical example of the glass transition temperatures for dried MCC determined by standard DSC, with three replicates showing a glass transition; for curve 1, Tg is 185.58C, for curve 2, Tg is 186.88C, and for curve 3, Tg is178.08C determined by the derivative of the curve.
of the equilibrated samples cannot be precisely quantitated because non-hermetic pans were used; thus, the sample moisture content is not accurately known during analysis due to water loss. The lowering of Tg by water can be estimated by the Fox equation:19 1 w1 w2 Tg Tg1 Tg2
1
where, Tg, Tg1, and Tg2 are the glass transition temperatures of the polymer/water, polymer, and water, respectively; and w1 and w2 are the weight fractions of the polymer and water, respectively. The predicted results indicate that there should be three Tg temperatures between 2 and 808C for a 30% amorphous material equilibrated between 75 and 100% RH, which is in agreement with our
Figure 3. Typical example of the glass transition temperature of dried MCC determined by TMA: [1] relative change in length of the compact (Ð), [2] second derivative versus temperature of change of the length of the compact (D2dL - -). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
®ndings. The fourth transition exhibited by some samples could be explained by considering that the same material could have undergone a second transition, due to the loss of water, and hence a raising of the Tg to the analysis temperature. The existence of these lower temperature transitions in equilibrated samples, which were not present in dry samples, indicates these transitions behave like glass transitions. In addition, these results are in general agreement with other authors who also found that water lowered the Tg by 1008C for poly(vinyl pyrrolidone) (PVP)20 and other cellulosic polymers.21,22 Glass Transition Temperatures of Granulated MCC Several glass transitions were observed in the granulated samples of Avicel PH 102; all granulation samples were run in triplicate. Granulating with 50 or 80% w/w water shows results similar to the equilibrated samples already described [140.69 1.21, 161.31 1.34, and 183.92 1.318C for 50 % (w/w); 148.88 2.97, 162.14 2.37, and 180.84 1.328C for 80% (w/w)]. Additionally, a weak Tg was observed at 115.4 4.28C. Based on a rough estimate of water loss, this glass transition is occurring in a material that has lost about half of the initial water content, which means the samples had a water content of 25± 40% w/w. The granules granulated with 110 or 130% w/w water also show transitions below 1008C. The transitions of granules after drying were not as pronounced, however the transition temperatures were still the same [151.03 2.83, 162.89 1.82, and 181.73 1.07 for 110% w/w; 149.67 2.75, 162.31 0.96, and 183.39 2.66 for 130% w/w]. In addition, other transitions were detected between 100 and 1308C [114.24 3.778C for 110% w/w and 112.67 1.00, 123.61 2.46, and 130.22 1.278C for 130% w/w]. These transitions occur when about half of the water is lost, which means that the total material contains 55 or 65% w/w water. For granules produced with isopropanol and for granules dried prior to analysis, only transitions at 133, 160, and 1848C were observed, and there were no glass transitions for these granules at lower temperatures. Scanning Electron Microscopy Representative SEM photographs of neat and granulated MCC are shown in Figures 4a±d. When compared with ungranulated material (Fig. 4a), the 80% w/w material (Fig. 4b) shows a
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
347
slight texture change whereas the 130% w/w material (Fig. 4c) shows larger structural changes that are due to the interactions with water. The material granulated with 130% w/w isopropanol (Fig. 4d) appears to be similar to ungranulated MCC. It is interesting to note that after drying, the granules produced with water and isopropanol and the dry MCC all have the same original Tg. Thus, these physical changes induced by granulation (see Fig. 4) do not appear to be responsible for the glass transitions of MCC.
DISCUSSION The classic de®nition of a phase is a homogenous distinct region of a system separated by a boundary. However, a more contemporary de®nition of a phase is based on the degree or amplitude of molecular motions a particular system is permitted in a given equilibrium or nonequilibrium state.23 Also, the glass transition is now considered much more complicated than previously thought. Going from the glassy state to the rubbery state was once considered a monolithic event involving a global mobility change, but more recent research has shown the transition to be much more complicated. For example, instead of a complete change in polymer mobility, now there are long- and short-range mobility changes and different degrees of cooperative and noncooperative motion.24 Thus, the transition has gone from being considered a single event to one of several possible relaxation events involving, for example, d-, g-, b-, and Tl±l transitions.25 In addition, instead of a single rubbery state, there are several possible mesophases into which the glass can transform,23 and some of these mesophases have been found in cellulosic systems.26 MCC appears to have glass transitions that come in triplets and at temperatures lower than that of cellulose (Tg of dry cellulose is 2308C10,11). Given the classic de®nition of a phase transition, these multiple transitions are dif®cult to explain; however, when considering all the transitions and molecular motions possible in MCC, these results are feasible. The authors believe the multiple transitions at lower temperatures are a result of Figure 4. Scanning electron micrographs of MCC for (a) the ungranulated material, (b) a material granulated with 80% w/w water, (c) a material granulated with 130% w/w water, and (d) a material granulated with 130% w/w isopropanol. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
348
PICKER AND HOAG
structural changes that occur during the MCC manufacturing process. To produce MCC, the cellulose ®bers undergo a controlled hydrolysis in a strong acid solution. This process preferentially reduces the molecular weight of the amorphous regions to a degree of depolymerization of 100±200.27 This depolymerization of the amorphous regions increases the number of end groups in these amorphous regions, which is associated with an increase in free volume. Thus, the authors believe the lowering of molecular weight via acid hydrolysis is responsible for lowering Tg. Discussed next are some possible explanations for the multiple transitions, which the authors separate into two hypotheses: (1) amorphous structural heterogeneity and secondary structural rearrangements and (2) nano- and mesophase transitions. First, structural heterogeneity, in which regions of the MCC have different thermal properties, could account for the multiple transitions observed. For lower molecular weights, a key determinant of Tg is polymer molecular weight. It is possible that during the acid hydrolysis of cellulose, certain regions may be preferentially hydrolyzed. Hence, regions of different molecular weight could have been formed, which give rise to different transition temperatures. In addition to differences in molecular weight, there could be differences in chemical structure. For example, in MCC, carboxyl groups from processing can be found;28 for one microcrystal, 6600 carboxyl groups were found via titration with ammonium cations, most of which are located at the C6 position of the glucose unit.28 Also, it is very likely that aldehyde groups form during processing with acid.29 These carboxyl and aldehyde groups are probably located in the amorphous regions of MCC where the cellulose chains are more susceptible to acid hydrolysis, and for cellulose ethers different substituted groups are known to in¯uence Tg. It is possible that the different substituents are not uniformly distributed throughout the amorphous regions, which could result in heterogeneous regions each with their own Tg. In addition, native and some processed cellulose ®bers are known to have an extended-chain ®brous structure, and these microscopic, highly oriented ®bers are called micro®brils.26 It is possible that these micro®brils could rearrange on heating, which could account for some of the observed transitions. Secondly, Chen and Wunderlich24 discuss the existence of different nano- and mesophases and JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002
how these different phases can result in multiple transitions. Beiner et al.30 found multiple Tgs in methacrylate-based homopolymers. For MCC, the three glass transitions could exist because of nanophase separation or because the semicrystalline nature of the MCC may create mesophases between the crystalline and amorphous regions, which could have a different Tg values. The Tg of an amorphous material is lowered by water.20±22 This effect has been shown for PVP,20 cellulose ethers,22 and other amorphous materials.21 Water can exist in different thermodynamic states in which it may be bound differently.31,32 Thus, it can in¯uence the molecular mobility; especially the mobility of ends groups and side chain groups, as described by Ahlgren.33 The fact that the Tgs were lowered for the materials exposed to water, coupled with the TMA data, indicates that the transitions were glass transitions. Granules made with isopropanol seem to be unaffected by the solvent when compared with granules made with water. Whereas water clearly interacts with MCC, isopropanol does not appear to because of its polarity and bonding ability. Thus, the fact that the material does not change is a good control for the effect of a solvent on the transitions in MCC. In conclusion, the transitions in MCC are numerous, complex, and subtle. Most of them are weak transitions resulting in a minor baseline shift. Although the shifts are reproducible and measurable, their weak character is open to multiple interpretation. Turi and Chartoff34 present the current state of data on the subject of weak transitions, discussing the existence of transitions lower than Tg. Because the primary factors affecting Tg for semi-crystalline polymeric materials are the chemical composition of a polymer molecule and molecular weight of the amorphous regions, we need to know more about the microstructure of MCC before de®nitive conclusions can be made.
ACKNOWLEDGMENTS Katharina M. Picker thanks the Martin-LutherUniversity Halle-Wittenberg, and especially Prof. Dr. habil. G. Zessin for providing the time to spend at the University of Maryland, School of Pharmacy. In addition, the authors thank Nasser N. Nyamweya for his technical and intellectual assistance during this project.
GLASS TRANSITION OF MICROCRYSTALLINE CELLULOSE
REFERENCES 1. Reier GE, Shangraw RF. 1966. Microcrystalline cellulose in tableting. J Pharm Sci 55:510±514. 2. Reier GE. Personal communication, 1997. 3. Rowe RC, McKillopp AG, Bray D. 1994. The effect of batch and source variation on the crystallinity of microcrystalline cellulose. Int J Pharm 101:169± 172. 4. Kleinebudde P. 1997. The crystallit-gel-model for microcrystalline cellulose in wet-granulation, extrusion and spheronization. Pharm Res 14:804±809. 5. Nakai Y, Fukuoka E, Nakajima S, Hasegawa J. 1977. Crystallinity and physical characteristics of microcrystalline cellulose. Chem Pharm Bull 25:96±101. 6. Nakai Y, Fukuoka E, Nakajima S, Yamamoto K. 1977. Crystallinity and physical characteristics of microcrystalline cellulose. II. Fine structure of ground microcrystalline cellulose. Chem Pharm Bull 25:2490±2496. 7. Hancock BC, Shamblin SL, Zogra® G. 1995. Molecular mobility of amorphous pharmaceutical solids below their glass transition temperatures. Pharm Res 12:799±806. 8. Hancock BC, Zogra® G. 1997. Characteristics and signi®cance of the amorphous state in pharmaceutical systems. J Pharm Sci 86:1±12. 9. Stubberud L, Arwidson HG, Larsson A, Graffner C. 1996. Water solid interactions. II. Effect of moisture sorption and glass transition temperature on compactibility of microcrystalline cellulose alone or in binary mixtures with polyvinyl pyrrolidone. Int J Pharm 134:79±88. 10. Batzer H, Kreibich UT. 1981. In¯uence of water on thermal transitions in natural polymers and synthetic polyamides. Polym Bull 5:585±590. 11. Salmen NL, Back EL. 1977. The in¯uence of water on the glass transition temperature of cellulose. TAPPI J 60:137. 12. Wunderlich B, Okazaki I. 1997. Modulated differential scanning calorimetry in the glass transition region. VI. Model calculations based on poly(ethylene terephthalate). J Therm Anal 49:57±70. 13. Reading M, Elliott D, Hill VL. 1993. A new approach to the calorimetric investigation of physical and chemical transitions. J Therm Anal 40:949±955. 14. Wunderlich B, Boller A, Okazaki I, Kreitmeier S. 1996. Modulated differential scanning calorimetry in the glass transition region. II. The mathematical treatment of the kinetics of the glass transition. J Therm Anal 47:1013±1026. 15. Hourston DJ, Song M, Pollock HM, Hammiche A. 1997. Modulated differential scanning calorimetry. III. Applications to measurements of the glass transition temperature and increment of heat capacity. J Therm Anal 49:209±218. 16. McPhillips H, Craig DQM, Royall PG, Hill VL. 1999. Characterisation of the glass transition of
17. 18. 19. 20.
21.
22.
23. 24. 25. 26. 27.
28. 29. 30.
31. 32. 33. 34.
349
HPMC using modulated temperature differential scanning calorimetry. Int J Pharm 180:83±90. Hill VL, Craig DQM, Feely LC. 1999. The effects of experimental parameters and calibration on MTDSC data. Int J Pharm 192:21±32. Jaffe M, Menczel JD, Bessey WE. 1997. In: Turi EA, editor. Fibers. San Diego: Academic Press, pp. 1767±1954. Fox TG. 1956. In¯uence of diluent and of copolymer composition on the glass transition temperature of a polymer system. Bull Am Phys Soc 1:123. Oksanen CA, Zogra® G. 1990. The relationship between the glass transition temperature and water vapor absorption by poly(vinylpyrrolidone). Pharm Res 7:654±657. Hancock BC, Zogra® G. 1994. The relationship between glass transition temperature and the water content of amorphous solids. Pharm Res 11:471±477. Picker KM, Mielck JB. 1998. Effect of relative humidity during tabletting on matrix formation of hydrocolloids: Densi®cation behavior of cellulose ethers. Pharm Technol Dev 3:31±41. Wunderlich B. 1999. A classi®cation of molecules, phases, and transitions as recognized by thermal analysis. Thermochim Acta 340-1:37±52. Chen W, Wunderlich B. 1999. Nanophase separation of small and large molecules. Macromol Chem Phys 200:283±311. Egorov VM. 1994. Differential scanning calorimetry of polymers, physics, chemistry, analysis, technology. New York: Ellis Horwood, p. 253. Gilbert RD, Kadla JF. 1998. In: Kaplan DL, editor. Biopolymers from renewable sources. Berlin: Springer, pp. 47±95. Doelker E, Gurny R, Schurz J, Janosi A, Matin N. 1987. Degrees of crystallinity and polymerization of modi®ed cellulose powders for direct tableting. Powder Technol 52:207±213. Edelson MR, Hermans J Jr. 1963. Flow of gels of cellulose microcrystals. II. Effect of added electrolyte. J Polym Sci Part C 2:145±152. Fleck E. Personal communication, 1997. Beiner M, SchroÈter K, Hempel E, Reissig S, Donth E. 1999. Multiple glass transition and nanophase separation in poly(n-alkyl methacrylate) homopolymers. Macromolecules 32:6278±6282. Zogra® G. 1988. States of water associated with solids. Drug Dev Ind Pharm 14:1905±1926. Zogra® G, Kontny MJ. 1986. The interactions of water with cellulose- and starch-derived pharmaceutical excipients. Pharm Res 3:187±194. Ahlgren P. 1995. The application of microcrystalline cellulose in pharmaceutical tablet-making. Nordic Pulp Paper Res J 1:12±16. Chartoff RP. 1997. In: Turi EA, editor. Thermoplastic polymers. San Diego: Academic Press, pp. 483±743.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 91, NO. 2, FEBRUARY 2002