Anal. Chem. 2010, 82, 1277–1282
Application of Raman Microscopy to Biodegradable Double-Walled Microspheres Effendi Widjaja,*,† Wei Li Lee,‡ and Say Chye Joachim Loo*,‡ Process Science and Modeling, Institute of Chemical and Engineering Sciences, Agency for Science, Technology and Research (ASTAR), 1 Pesek Rd, Jurong Island, Singapore 627833, and School of Materials Science and Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798 Raman mapping measurements were performed on the cross section of the ternary-phase biodegradable doublewalled microsphere (DWMS) of poly(D,L-lactide-co-glycolide) (50:50) (PLGA), poly(L-lactide) (PLLA), and poly(εcaprolactone) (PCL), which was fabricated by a one-step solvent evaporation method. The collected Raman spectra were subjected to a band-target entropy minimization (BTEM) algorithm in order to reconstruct the pure component spectra of the species observed in this sample. Seven pure component spectral estimates were recovered, and their spatial distributions within DWMS were determined. The first three spectral estimates were identified as PLLA, PLGA 50:50, and PCL, which were the main components in DWMS. The last four spectral estimates were identified as semicrystalline polyglycolic acid (PGA), dichloromethane (DCM), copper-phthalocyanine blue, and calcite, which were the minor components in DWMS. PGA was the decomposition product of PLGA. DCM was the solvent used in DWMS fabrication. Copper-phthalocyanine blue and calcite were the unexpected contaminants. The current result showed that combined Raman microscopy and BTEM analysis can provide a sensitive characterization tool to DWMS, as it can give more specific information on the chemical species present as well as the spatial distributions. This novel analytical method for microsphere characterization can serve as a complementary tool to other more established analytical techniques, suchasscanningelectronmicroscopyandopticalmicroscopy. In recent years, there has been an increasing interest to investigate the composite double-walled microspheres (DWMS), which comprise a core of one polymer and a shell of a second polymer.1 The interest has arisen since the double-walled microspheres can provide better enhanced control of drug releases compared to conventional single polymer microspheres.2,3 As drug delivery devices, DWMS has demonstrated its capabilities to overcome the limitations generally found in single polymer * To whom correspondence should be addressed. E-mail: effendi_widjaja@ ices.a-star.edu.sg (E.W.);
[email protected] (S.C.J.L.). † Agency for Science, Technology and Research (ASTAR). ‡ Nanyang Technological University. (1) Pekarek, K. J.; Jacob, J. S.; Mathiowitz, E. Nature 1994, 367, 256–260. (2) Lee, T. H.; Wang, J.; Wang, C. H. J. Controlled Release 2002, 83, 437–452. (3) Berkland, C.; Pollauf, E.; Pack, D. W.; Kim, K. J. Controlled Release 2004, 96, 101–111. 10.1021/ac9022549 2010 American Chemical Society Published on Web 12/17/2009
microspheres. Some of these limitations are high initial burst,4 inability to obtain zero order release,5 lack of controlled release in a sequential manner,5 and low encapsulation efficiency for highly water-soluble drugs.6 In the past years, several fabrication methods of DWMS have been developed. These methods include the coating technologies that use pan, fluidized beds, or spray drying;7,8 precision particle fabrication technology that employs a series of annular nozzles to create a compound jet;9,10 and emulsion solvent evaporation technique that involves the phase separation of a polymer mixture owing to solvent evaporation.11,12 Following fabrication, the characterization of DWMS is carried out mostly using scanning electron microscopy (SEM) and optical microscopy.2,13 The aims of these analyses are to observe the surface and the cross-sectional morphologies, to determine the chemical composition of the core and shell polymer and to determine the drug distribution within the core and shell microspheres. Although these types of characterizations are important since they could provide high magnification and high resolution micrographs of DWMS, they lack more detailed structural and chemical information. As such, more precise species identification of polymers, drugs, and impurities contained in microspheres and their spatial distributions cannot be directly obtained via these techniques. In order to overcome the limitation of SEM and optical microscopy analyses, a novel characterization approach based on Raman microscopy mapping is proposed in current study. Raman microscopy, which integrates spectroscopy with mapping and imaging technology, can give both spectral and spatial information of the sample being investigated. The spectral information provides detailed vibrational and rotational energies of molecular bonds that relate to molecular fingerprints, and the spatial information provides detailed chemical distribution of the observed species in two spatial dimensions. In addition, Raman microscopy (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Jalil, R.; Nixon, J. R. J. Microencapsulation 1990, 7, 53–66. Pekarek, K. J.; Jacob, J. S.; Mathiowitz, E. Adv. Mater. 1994, 6, 684–686. Bodmeier, R.; McGinity, J. W. Pharm. Res. 1987, 4, 465–471. Lee, H. K.; Park, J. H.; Kwon, K. C. J. Controlled Release 1997, 44, 283– 293. Wang, F. J.; Wang, C. H. J. Controlled Release 2002, 81, 263–280. Berkland, C.; Kim, K.; Pack, D. W. J. Controlled Release 2001, 73, 59–74. Berkland, C.; King, M.; Cox, A.; Kim, K.; Pack, D. W. J. Controlled Release 2002, 82, 137–147. Leach, K.; Noh, K.; Mathiowitz, E. J. Microencapsulation 1999, 16, 153– 167. Tan, E. C.; Lin, R.; Wang, C. H. J. Colloid Interface Sci. 2005, 291, 135– 143. Pollauf, E.; Pack, D. W. Biomaterials 2006, 27, 2898–2906.
Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
1277
is also a nondestructive technique and does not require any particular sample preparation.14 The fabricated double-walled microspheres can be easily mounted on the microscope stage and hundreds or even thousands of Raman spectra can be collected from one specific area of sample using the point-to-point mapping mode. It is also worth mentioning that another advantage of this characterization is that Raman scattering of the water molecule is very weak, and hence, it does not interfere with Raman signals corresponding to other molecules. Visual evaluation of all the collected spectra from one specific area of DWMS is certainly not practical. Therefore, a more sophisticated data analysis technique, such as chemometrics is certainly needed. Among various chemometric tools that are currently available for spectroscopic data analysis, the blind-source separation algorithm is one tool that could fit into this purpose. The blind source separation algorithm or self-modeling curve resolution (SMCR, a more well-known term in the chemometrics area) tries to reconstruct the pure component spectral estimates from a set of spectroscopic mixture data without recourse to any a priori information, particularly spectral libraries. Among various SMCR techniques that have been developed in the past years, band-target entropy minimization (BTEM) is one special technique that has been effectively used to recover pure component spectra of minor components from weak spectroscopic signals.15 It is initially developed to analyze complex infrared spectroscopic data obtained from both reactive and nonreactive systems;16,17 however, due to its practicality and generality, in these past few years, the BTEM technique has been applied to other vibrational spectroscopic data sets, such as Raman.18,19 In addition, the BTEM technique has been also employed to analyze hyperspectral image data obtained from biomedical,18 pharmaceutical,14 bioactive glass,20 and commercial samples.19 In this study, BTEM analysis was applied to recover the underlying pure component spectra from DWMS samples fabricated from three kinds of biopolymers, i.e., poly(L-lactide) (PLLA), poly(D,L-lactide-co-glycolide) (PLGA), and poly(ε-caprolactone) (PCL). Besides these three main species, pure component spectra of the biopolymer degradation product, the trapped solvent, and contaminants were also recovered and identified. The spatial distributions of all these observable species, i.e., biopolymers and impurities, were obtained accordingly. MATERIALS AND METHODS Materials. Poly(L-lactide) (PLLA, intrinsic viscosity/IV: 2.38, Bio Invigor), poly(D,L-lactide-co-glycolide 50:50) (PLGA, IV: 1.18, Bio Invigor), poly(ε-caprolactone) (PCL, MW 80 kDa, Aldrich) and poly(vinyl alcohol) (PVA, MW 30-70 kDa, Sigma-Aldrich) were used without further purification. High-performance liquid chromatography (HPLC) grade dichloromethane (DCM) (Tedia Company Inc.) was used as solvents, as received. (14) Widjaja, E.; Seah, R. K. H. J. Pharm. Biomed. Anal. 2008, 46, 274–281. (15) Widjaja, E. Development of band-target entropy minimization (BTEM) and associated software tools, PhD Thesis, National University of Singapore, Singapore, 2002. (16) Li, C. Z.; Widjaja, E.; Garland, M. J. Am. Chem. Soc. 2003, 125, 5540– 5548. (17) Widjaja, E.; Li, C. Z.; Chew, W.; Garland, M. Anal. Chem. 2003, 75, 4499– 4507. (18) Widjaja, E.; Crane, N.; Chen, T.-S.; Morris, M. D.; Ignelzi, M. A., Jr.; McCreadie, B. R. Appl. Spectrosc. 2003, 57, 1353–1362. (19) Widjaja, E.; Garland, M. Anal. Chem. 2008, 80, 729–733. (20) Seah, R. K. H.; Garland, M.; Loo, J. S. C.; Widjaja, E. Anal. Chem. 2009, 81, 1442–1449.
1278
Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
Preparation of Double-Walled Microspheres. The PLLA/ PLGA/PCL composite microspheres were prepared using (O/ W) the emulsion solvent evaporation method. Briefly, 0.1 g of PCL, 0.2 g of PLGA, and 0.3 g of PLLA were dissolved in 10.5 mL of DCM. The resultant polymer solution was added into 500 mL of the water containing PVA at a concentration of 0.5% (w/v) and emulsified using an overhead stirrer (Calframo BDC1850-220) at a rate of 300 rpm for 4 h. The evaporation of DCM will give rise to phase separation of PLLA, PLGA, and PCL, followed by the formation of ternary-phase composite microspheres. Next, the microspheres produced were filtered, rinsed with deionized water, lyophilized, and stored in a desiccator. The microspheres to be examined were first cross-sectioned using a razor blade. The microspheres were deliberately fabricated to be relatively large (∼300 µm), to allow for easy cross sectioning at the microspheres’ centerline with sufficient accuracy with the razor blade. Scanning Electron Microscopy (SEM) Characterization. The first analysis was performed to observe the morphology of the cross-sectional microspheres. This analysis was carried out by scanning electron microscopy (SEM, Jeol, JEM-6360A), which was operated at a voltage of 5.0 kV. Before analysis, the samples were first mounted onto metal stubs and then coated with gold using a sputter coater model SPI-Module. Raman Mapping Characterization. The cross-sectioned double-walled microsphere was placed under the microscope objective, and laser power up to ca. 20 mW was shone onto the surface of the sample. Raman point-by-point mapping measurements were performed on the area of 280 µm × 100 µm with a step size of 5 µm in both the x and y directions using a Raman microscope (InVia Reflex, Renishaw) equipped with a near-infrared enhanced deep-depleted thermoelectrically Peltier cooled CCD array detector (576 × 384 pixels) and a high grade Leica microscope. The sample was irradiated with a 785 nm nearinfrared diode laser, and a 50× objective lens was used to collect the backscattered light. The diameter of laser beam is approximately 2 µm. Measurement scans were collected using a static 1800 groove per mm dispersive grating in a spectral window from 300 to 1900 cm-1, and the acquisition time for each spectrum was around 35 s. The final Raman mapping data collected from this experiment had a dimension of 57 × 21 × 1740. Spectral preprocessing that includes spike removal due to cosmic rays were carried out first before the data was further analyzed using the band-target entropy minimization (BTEM) algorithm. Spectral Analysis by Band-Target Entropy Minimization (BTEM) Algorithm. As mentioned above, the BTEM algorithm was developed in order to reconstruct the pure component spectra of underlying constituents from a set of mixture spectra without using any spectral libraries or any a priori knowledge. When all normalized pure component spectra of all underlying constituents have been reconstructed, the relative contributions of each constituent can be calculated by projecting them back onto the baseline-corrected and normalized data set. The spatial distribution of each underlying constituent can then be generated. For a more detailed description of this algorithm, readers are referred to refs 15 and 17.
Figure 1. SEM cross-sectional view of PLLA/PLGA/PCL composite microsphere.
Figure 3. First top eight vectors associated with the first eight right singular vectors and the last two bottom vectors associated with the 15th and 16th right singular vectors obtained from SVD on Raman mapping spectra measured from the DWMS sample.
Figure 2. 1197 despiked Raman mapping spectra measured from the DWMS sample.
RESULTS AND DISCUSSION Scanning Electron Microscopy (SEM). Figure 1 shows the SEM cross-sectional view of the fabricated composite microsphere of PLLA/PLGA/PCL. The size of the microsphere ca. 300 µm can be determined from this micrograph. The double wall of the microsphere is clearly seen with a distinct segregation between core and shell. It is also apparent that the shell of the microsphere is less dense than its core. Raman Microscopy and BTEM Analysis. Raman microscopy measurements were subsequently performed on the surface of the cross-sectioned DWMS sample to further identify its contents. A total of 1197 Raman spectra was collected, and the despiked Raman spectra are shown in Figure 2. As can be seen, a relatively good signal-to-noise ratio of Raman spectra could be obtained despite large variation of backgrounds due to baseline changes.
The despiked Raman mapping spectra were subjected to singular value decomposition (SVD), and the resulted first few right singular vectors (commonly named as basis eigenvectors) are shown in Figure 3. These basis vectors contain abstract representations on the pure component spectra of the observable components present in the system and are ordered according to their significant contribution to the total variance in the observations. Spectral information associated with major components observed in the system is usually found in the first few vectors. On the other hand, the spectral information concerning minor components can be observed in the higher basis vectors. For instance, in the present study, the spectral information of calcite (one of the minor components identified in the present sample) is only observed from the 16th vector and above. Exhaustive band targeting for all important observable spectral features seen in Figure 3 was performed. A superset of reconstructed pure component spectra was obtained, and this set was further reduced to eliminate redundancies. Finally, after reduction, seven pure component spectra were reconstructed, which resulted from seven band-targeted spectral features at 705, 848, 874, 999, 1087, 1110, and 1529 cm-1. The full spectral range pure component spectral estimates are presented in Figures 4b and 5b. The relative contributions from each component were then obtained by projecting the spectral estimates back onto the preprocessed Raman mapping spectra (preprocessing involves despiking, baseline correction, and normalization to maximum height). The purpose of spectral normalization is to overcome the problem of pixel-to-pixel Raman intensity variation during mapping measurements. The obtained relative contributions were then normalized and reorganized into a map or spatial distribution for Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
1279
Figure 4. (a) 10× Optical microscope image. (b) Major pure component Raman spectral estimates from BTEM. (c) Score images for each pure component spectra that represent its spatial distribution.
Figure 5. (a) 10× Optical microscope image. (b) Minor pure component Raman spectral estimates from BTEM. (c) Score images for each pure component spectra that represent its spatial distribution.
each component, which are shown in Figures 4c and 5c. These score images provide semiquantitative content for the seven observed species in the present sample. The summation of the intensity scale of all species at each pixel is equal to unity. It is worth noting that the axes for spatial distribution are in pixel number, which can be directly converted to distance by multiplying by 5 µm for each pixel. Combined altogether, the seven reconstructed pure component spectra have accounted for more than 99% of the measured mixture spectral signals. The pure component spectral estimates obtained from BTEM analysis (without any a priori spectral information) were compared to known spectral libraries. It was found that the first three spectral estimates as shown in Figure 4 were the three biopolymers used to fabricate the DWMS, i.e., PLLA, PLGA, and PCL. The Raman bands of these three spectral estimates were very well 1280
Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
resolved with high signal-to-noise ratio and were in good agreement with the literature.21–24 The significant Raman peaks of these spectral estimates and their corresponding peak assignments are shown in Table 1. As can be seen in Figure 4, Raman mapping and BTEM analysis results show that PLGA formed the core and PLLA formed the shell of the present fabricated DWMS. PCL, however, did not form any new layer within the microsphere, but it was (21) Kister, G.; Cassanas, G.; Vert, M.; Pauvert, B.; Terol, A. J. Raman Spectrosc. 1995, 26, 307–311. (22) Geze, A.; Chourpa, I.; Boury, F.; Benoit, J. P.; Dubois, P. Analyst 1999, 124, 37–42. (23) Kister, G.; Cassanas, G.; Bergounhon, M.; Hoarau, D.; Vert, M. Polymer 2000, 41, 925–932. (24) Misra, R. M.; Agarwal, R.; Tandon, P.; Gupta, V. D. Eur. Polym. J. 2004, 40, 1787–1798.
Table 1. Peak Positions (cm-1) and Corresponding Vibrational Assignments of BTEM Spectral Estimates For PLLA, PLGA 50/50, PCL, and PGAa semicrystalline PLLA peak position (cm-1)
assignment
303 398 411 714 738 874 1043 1093 1128 1295 1388 1454 1752 1768 1775
COC deformation δCCO δCCO γCdO δCdO νC-COO νC-CH3 νsCOC rasCH3 δ2CH δsCH3 δasCH3 + δCH2 νCdO νCdO νCdO
a
PLGA 50/50 peak position (cm-1)
assignment
706 748 848 874 892 1033 1048 1093 1130 1275 1303 1351 1387 1427 1454 1771
γCdO δCdO rCH2 νC-COO νC-C + rCH2 νsCOC νC-CH3 νsCOC rasCH3 twCH2 δ2CH δ1CH + δsCH3 δsCH3 δCH2 δasCH3 + δCH2 νCdO
semicrystalline PCL
semicrystalline PGA
peak position (cm-1)
assignment
peak position (cm-1)
assignment
915 960 1040 1066 1110 1285 1307 1418 1443 1470 1724
νC-COO νC-C νC-CH3 νC-C + νsCOC νC-C + νasCOC twCH2 twCH2 δCH2 δCH2 δCH2 νCdO
288 317 605 875 998 1091 1166 1249 1406 1439 1780
δCOCR δOCRC γCdO νC-C + rCH2 νC-C + rCH2 νsCOC νasCOC twCH2 wCH2 δCH2 νCdO
δ ) bending; γ ) scissoring; ν ) stretching; s ) symmetric; as ) asymmetric; tw ) twisting; r ) rocking.
dispersed in the PLLA shell. Its presence in the core was almost unobservable. The DWMS fabricated in the present study was, therefore, PLGA-cored with a PLLA-shell impregnated with PCL particulates. The last four BTEM pure component spectral estimates shown in Figure 5 are rather unexpected, and they need to be further examined. Comparison with known spectral libraries verified that these four spectral estimates are similar to the pure component spectra of semicrystalline polyglycolic acid (PGA), dichloromethane (DCM), copper-phthalocyanine blue, and calcite, respectively. The first BTEM estimate shows significant Raman peaks at 288, 317, 605, 875, 998, 1091, 1166, 1249, 1406, 1439, and 1780 cm-1. The peak positions and peak relative intensities are very consistent with the PGA spectrum reported in previous work.25 The peak assignments for the PGA spectrum are shown in Table 1.25 An additional peak at 704 cm-1 was also found in this estimate. It does not belong to PGA, but it belongs to PLGA. The appearance of this additional peak might be due to imperfect spectral reconstruction via BTEM analysis. Such an artifact can be observed, especially when BTEM is used to recover pure component spectra of minor components, whose signals are just above noise signals.14 The second BTEM estimate shows significant Raman peaks at 287, 705, 739, 1163, and 1424 cm-1, which are consistent with DCM.26 Additional small peaks at 369, 669, 848, 873, 893, 914, 1033, 1449, 1527, 1773, and 1786 cm-1 are also observed, and these are mainly associated with the Raman peaks of PLGA and copperphthalocyanine blue (see the third estimate below). Again, such artifacts are present due to the minute level of DCM signals in the present study. Nevertheless, such imperfection did not hinder the species identification of DCM in general since all significant Raman peaks were well resolved. The third BTEM estimate in Figure 5 shows significant Raman peaks at 485, 596, 682, 749, 955, 1009, 1110, 1145, 1185, 1220, 1309, 1342, 1365, 1452, (25) Kister, G.; Cassanas, G.; Vert, M. Spectrochim. Acta, Part A 1997, 53, 1399– 1403. (26) www.sigmaaldrich.com.
1494, 1530, and 1604 cm-1, which are consistent with the Raman spectrum of copper-phthalocyanine blue found in literature.19 The fourth estimate has significant Raman peaks at 281, 714, 1088, and 1448 cm-1, which are in agreement with calcite.27 These peaks are corresponding to rotational (Eg) mode, ν4 inplane bending, ν1 symmetric stretching, and ν3 asymmetric stretching of carbonate ions, respectively. The presence of copper-phthalocyanine blue and calcite in this microsphere was certainly surprising, since these two components were not parts of materials used for preparing the microsphere. Further investigation revealed that these two components were from the blue-colored paper, where the microsphere was laid during cross-sectioning using a razor blade. Copper-phthalocyanine blue is a frequently used blue dye in inkjet printers, paints, coloring paper, etc,28 and calcite is one of the filling materials used in paper-making for higher brightness and better printability. Apparently, a small portion of this paper containing these two components had contaminated the surface of the cross-sectioned microsphere. The presence of PGA in the microsphere indicated that some portions of PLGA had degraded to PGA.29,30 PLGA(50:50), used as the polymer in the core of the microspheres, is known to have a relatively fast rate of hydrolysis.29 The presence of DCM in the microsphere was rather not so surprising since DCM was used as the solvent in the microsphere fabrication; however, its presence was still unexpected. A small portion of DCM was possibly trapped in the microsphere due to incomplete evaporation, and its presence was clearly identified via BTEM analysis. The detection and identification of impurities may cause some modifications in microsphere fabrication procedures. This modification is particularly needed if the presence of the unexpected (27) Burgio, L.; Clark, R. J. H. Spectrochim. Acta, Part A 2001, 57, 1491–1521. (28) Chaplin, T. D.; Clark, R. J. H.; Beech, D. R. J. Raman Spectrosc. 2002, 33, 424–428. (29) Loo, S. C. J.; Ooi, C. P.; Boey, Y. C. F. Biomaterials 2005, 26, 3809–3817. (30) Wu, X. S. Synthesis and properties of biodegradable lactic/glycolic acid polymers. In Encyclopedic handbook of biomaterials and bioengineering; Wise, D. L., Trantolo, D. J., Altobelli, D. E., Yaszemski, M. J., Gresser, J. D., Schwartz, E. R., Eds. Marcel Dekker: New York, 1995; pp 1015-1054.
Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
1281
species can bring harm to DWMS to be used as a drug delivery device. Figure 5 further shows that PGA is mostly dispersed in the core, DCM is dispersed subtly in the core and the shell, copperphthalocyanine blue is dispersed mostly in the shell and in the border between the core and the shell, and calcite is only found in a couple of random spots. PGA, DCM, and calcite were certainly minor components in this DWMS sample, as indicated by their spatial distribution and their score intensities in Figure 5. In general, the present results show that BTEM analysis is undoubtedly a very useful tool to identify and to provide spatial distributions of minor constituents. Although the present study was carried out using a rather simple example of a single and quite large DWMS (diameter of 300 µm, ca. 5-8 times larger than what is currently done for pharmaceutical applications), the extension of the present analytical approach to the real interest of pharmaceutical samples is very straightforward. With the current development of Raman microscopy technology, it is now possible to perform Raman mapping measurements at submicrometer lateral resolution.31 As such, hundreds or thousands of Raman spectra can be acquired with a finer lateral resolution from a much smaller DWMS, and BTEM analysis is readily used to reconstruct the pure component spectra of observable species and their spatial distributions. In the case of well-known chemical compositions of the DWMS, the reconstruction of the pure component spectra via BTEM analysis can be very useful. In particular, the analysis can be performed to investigate any spectral changes between known spectral libraries and the recovered pure component spectra estimates. The spectral changes may reflect the changes of crystal structures, amorphization, and degradation of both polymer and (31) Belu, A.; Mahoney, C.; Wormuth, K. J. Controlled Release 2008, 126, 111– 121.
1282
Analytical Chemistry, Vol. 82, No. 4, February 15, 2010
active-ingredient due to exposure to the environment and/or polymer-polymer and polymer-drug interactions. As for the unknown system, this novel approach for characterizing DWMS particles can facilitate the species identification not only for the polymers and active-ingredient but also for the impurities and contaminants inserted and/or present in DWMS during and/or after fabrications. CONCLUSIONS The present contribution demonstrates that the combined use of Raman microscopy and BTEM analysis could identify the observable species contained in DWMS as well as determine their spatial distributions. As this approach could directly provide more detailed structural and chemical information, this approach certainly can be a good complementary to SEM, a more established analytical technique for microsphere morphology characterizations. Since the proposed method appears to be rather general, it can be readily extended to investigate the drug distribution and in situ drug release studies in DWMS. The ability to identify the degradation product, trapped solvent, and contaminants in DWMS may also facilitate the new possibilities to study the DWMS stability via the Raman microscopic technique. ACKNOWLEDGMENT This work was supported by the Agency for Science, Technology and Research (A*STAR), Singapore, under the Advanced Reaction Engineering, Process Analytics, and Chemometrics program of ICES. Authors also would like to thank Dr. Martin Tjahjono for valuable input and fruitful discussions. Received for review October 6, 2009. Accepted December 3, 2009. AC9022549
