Research Article Received: 14 April 2009 Accepted: 6 July 2009 Published online in Wiley Interscience: 1 September 2009 (www.interscience.wiley.com) DOI 10.1002/jrs.2420 Use of thermo-Raman spectroscopy and chemometric analysis to identify dehydration steps of hydrated inorganic samples – application to copper sulfate pentahydrate Effendi Widjaja,∗ Hui Heng Chong and Martin Tjahjono In situ thermo-Raman spectroscopy (TRS) measurements were performed in order to investigate solid-phase transformation of the copper sulfate pentahydrate from room temperature up to 300 ◦ C. Band-target entropy minimization (BTEM), a blind-source separation algorithm, was employed in order to identify and reconstruct the pure component spectra of the species involved in the dehydration process. In spite of low signal-to-noise ratio and elevated baseline spectral data, BTEM was successfully utilized to identify and reconstruct four pure component spectra of copper sulfate pentahydrate, trihydrate, monohydrate, and anhydrate, which were formed during this thermally induced process. Subsequent mapping of these four pure component spectral estimates back onto the preprocessed spectra yielded the relative concentrations of each individual species. Finally, the transition temperatures of each dehydration step could be unambiguously deduced from the obtained concentration profile. The current study shows that combined thermo-Raman spectroscopy and chemometric analysis provides an effective tool to determine the dehydration temperatures as well as to identify the structures of each individual species involved in a c 2009 John Wiley & Sons, Ltd. solid-phase dehydration process. Copyright
Supporting information may be found in the online version of this article. Keywords: thermo-Raman spectroscopy; chemometric analysis; band-target entropy minimization; dehydration; pure component spectra Introduction J. Raman Spectrosc. 2010, 41, 181–186 ∗ Correspondence to: Effendi Widjaja, Process Science and Modeling, Institute of Chemical and Engineering Sciences, Agency for Science, Technology and Research (A∗ STAR), 1 Pesek Road, Jurong Island, Singapore 627833. E-mail: effendi widjaja@ices.a-star.edu.sg Process Science and Modeling, Institute of Chemical and Engineering Sciences, Agency for Science, Technology and Research (A∗ STAR), 1 Pesek Road, Jurong Island, Singapore 627833 c 2009 John Wiley & Sons, Ltd. Copyright
181 In the past years, thermo-Raman spectroscopy (TRS) has been developed in order to investigate thermally induced phase transformations in solids.[1] This spectroscopic technique has become increasingly popular since it can provide both qualitative and quantitative information related to the dynamic thermal process. The changes of Raman band positions, band intensities, and bandwidths as well as the appearance and disappearance of certain spectral bands can be used to identify the structural changes in the solid sample. Subsequent spectral data analysis can be performed in order to determine the composition and the phase of the sample. Obtaining such information is undoubtedly useful for the better understanding of the solid transformation processes. The more widely used techniques for studying solidphase transformation, namely thermogravimetric analysis (TGA) and differential thermal analysis (DTA), can only provide the bulk information associated with the thermal properties of the solid sample, but not the specific information related to the structures and their compositions.[2] The utility of TRS to investigate the solid structural transformation was first demonstrated in the study of the thermal decomposition of CaC2 O4 .[1] Subsequently, it has been used to investigate the dehydration processes of CuSO4 ·5H2 O,[3] CaSO4 ·2H2 O, CaSO4 ·0.5H2 O,[4] Na3 PO4 ·12H2 O,[5] NaH2 PO4 ·2H2 O,[6] Na4 P2 O7 ·10H2 O,[7] saleeite,[8] manasseite,[9] etc. TRS has also been employed to study the phase transformations of KNO3 ,[10] Na2 SO4 ,[11] MoO3 ,[12] Na4 P2 O7 ,[7] kintoreite,[13] titanium oxide nanotubes,[14] anatase and rutile,[15] etc. The TRS spectral data analysis is typically carried out by simply observing the changes of Raman intensities, band positions, and bandwidths, as well as the presence or absence of certain spectral bands, from the in situ thermal-dependence and/or isothermal time-dependent Raman spectra. Such spectral analysis has been proven useful; however, it may still pose some limitations, especially when highly overlapping in situ spectral bands are involved. This situation is often encountered when different species with different compositions are simultaneously formed during thermal transformation. The conventional band-fitting technique has been typically used to resolve this problem. However, the solutions offered from this technique are somewhat subjective depending on the number of reconstructed bands used and often relying on the available a priori spectral information. E. Widjaja, H. H. Chong and M. Tjahjono In the present study, a more advanced chemometric data analysis approach is utilized in order to analyze the spectral data collected from in situ TRS. This chemometric approach, namely band-target entropy minimization (BTEM), employs a blind source separation algorithm to reconstruct the pure component spectra of the species without recourse to any a priori information, particularly spectral libraries.[16] Initially, BTEM was developed for analyzing the large sets of multicomponent infrared spectroscopic data.[17,18] However, due to its generality, it can be applied to analyze spectral data generated from other spectroscopic measurements. These include NMR,[19] Raman spectroscopy,[20 – 22] mass spectrometry,[23] and XRD.[24] One of the particular advantages in using BTEM is its ability to recover the pure component spectra of the trace constituents (i.e. present at ppm concentrations or less), as demonstrated in several in situ spectroscopic studies of complex reactions,[25] mixed metal oxide characterizations,[26] pharmaceutical tablets,[27] biological samples,[28] etc. In this work, BTEM is used to identify and reconstruct the pure component of the species observed during solid thermally induced phase transformations. The results obtained from BTEM analysis are analyzed further to determine the compositions of these associated species as a function of temperature. This information can provide a more detailed picture of the corresponding solid transformation process. It is also worth mentioning that the analysis of TRS data itself is quite challenging since the TRS data consists of highly overlapping spectral bands with rather low signal-to-noise ratio. In this regard, copper(II) sulfate pentahydrate was selected in order to demonstrate the utility of the present methodology. Although the dehydration process of copper(II) sulfate pentahydrate has been well studied using various thermal instruments, such as TGA and DTA, there is not much information on the structural details of the associated hydrated species.[29 – 35] In the present study, we combine the use of TRS and BTEM analysis to identify the dehydration steps of copper(II) sulfate pentahydrate. Dehydration temperatures obtained from this combined method are also compared with the results obtained using TGA. In addition, this study also highlights the important factor, namely laser power, used in the determination of the dehydration temperatures using the present in situ TRS technique. Figure 1. A total of 275 preprocessed in situ Raman spectra measured at temperature 26–300 ◦ C from the first experimental setup. laser. The temperature of the stage was controlled by a temperature programmer TMS 94 (Linkam Scientific Instruments, UK). In situ Raman spectra were then collected as a function of temperature. The measurement of the sample was started at room temperature (26 ◦ C), which was subsequently heated to 300 ◦ C at a heating rate of 2 ◦ C/min. The stage was held at the same temperature during the collection of Raman spectra at each particular temperature. Two different in situ Raman experimental setups were used in order to observe the effect of laser power on the dehydration temperatures of copper(II) sulfate pentahydrate. The first experimental setup employed a 20× microscope objective lens, ∼1.5 mW laser power, ∼30 s scanning time, and the measurement spectral range from 200 to 1500 cm−1 . The second experimental setup employed a 10× microscope objective lens, ∼0.1 mW laser power, ∼30 s scanning time, and the measurement range from 700 to 1300 cm−1 . Thermogravimetric analysis Experimental Copper(II) sulfate pentahydrate from Riedel-deHaen (99–100%) was used without further purification. Thermal dehydration of the copper(II) sulfate pentahydrate was performed in a TA Instruments thermogravimetric analyzer (SDT2960) with a flowing nitrogen atmosphere. Approximately 20.2 mg copper(II) sulfate pentahydrate was heated at a ramping rate of 2 ◦ C/min from room temperature up to 500 ◦ C. Thermo-Raman spectroscopy Spectral analysis Copper(II) sulfate pentahydrate was exposed in air for approximately 24 h. This air exposure was important in order to confirm that the tested sample had absorbed enough water from air to become a stable pentahydrate form. A small amount of sample was then put into a glass crucible, which was placed in an enclosed heating stage (Linkam Scientific Instruments, UK) situated directly under the Raman microscope (InVia Reflex, Renishaw, UK). The Raman microscope was equipped with near-infrared-enhanced deep-depleted thermoelectrically Peltier-cooled CCD array detector (576 × 384 pixels) and a high-grade Leica microscope. The Raman scattering was excited with a 785-nm near-infrared diode The collected Raman spectra from dehydration experiments were first smoothed using adjacent three-point averaging smoothing and de-spiked to remove spectral artifacts due to cosmic rays. The preprocessed spectra were then analyzed using BTEM[16] to reconstruct the pure component spectra of the observable species during thermal process or dehydration without the need of any spectral libraries or other a priori information. Subsequent modeling included a least-squares fit of the pure component spectra to the individual experimental Raman spectra taken at particular temperature so that the relative concentration profiles versus each temperature can be generated. Materials 182 www.interscience.wiley.com/journal/jrs c 2009 John Wiley & Sons, Ltd. Copyright
J. Raman Spectrosc. 2010, 41, 181–186 Thermo-Raman spectroscopy and chemometric analysis of hydrated inorganic samples Figure 2. The pure component spectral estimates obtained from BTEM analysis of the first experimental data set: (a) copper sulfate pentahydrate, (b) copper sulfate trihydrate, (c) copper sulfate monohydrate, (d) copper sulfate anhydrate. Results and Discussion Thermo-Raman spectroscopy First measurement setup, 20× microscope objective lens, 1.5 mW laser power J. Raman Spectrosc. 2010, 41, 181–186 c 2009 John Wiley & Sons, Ltd. Copyright
www.interscience.wiley.com/journal/jrs 183 A total of 275 in situ Raman spectra measured as a function of temperature were first preprocessed and subsequently subjected to BTEM analysis. The results of the preprocessed spectra are shown in Fig. 1. Four distinct spectral bands observed from right singular vectors obtained from singular value decomposition of the preprocessed spectra, i.e. 985, 1009, 1019, and 1056 cm−1 , were used as the band-targets for BTEM analysis. These four pure component spectral estimates were successfully reconstructed, and they are presented in Fig. 2. The reconstructed Raman bands of hydrates and anhydrous copper sulfate observed in Fig. 2 show significant peaks at 400–1200 cm−1 , which belong to the four vibrational modes ν 1 , ν 2 , ν 3 , and ν 4 of SO4 2− . These reconstructed Raman bands are in agreement with those reported in the previous work.[3] Since these Raman band characteristics have been addressed in detail in the previous work,[3] they will be discussed only briefly in present study. As CuSO4 ·5H2 O transformed to CuSO4 ·3H2 O, the ν 1 mode of SO4 2− shifted from 985 to 1009 cm−1 . As CuSO4 ·3H2 O transformed to CuSO4 ·H2 O, the ν 1 mode of SO4 2− split into a doublet at 1019 and 1044 cm−1 . Similar splitting of the ν 1 symmetry band into a doublet had been observed in previous work. This doublet suggests that the sulfate ion of monohydrate occupies two different symmetry sites.[3,36] In the final dehydration step, when it was further dehydrated, the doublet of the ν 1 mode of SO4 2− re-merged and shifted to 1056 cm−1 . The symmetric stretching ν 1 mode of SO4 2− ion has been systematically shifted to higher wavenumber during the dehydration process (i.e. from 985 cm−1 for the pentahydrate to 1056 cm−1 for the anhydrate). This shift is due to the combined effects of the change in intermolecular H-bonding (becoming less extensive) and the simultaneous increase in coordination bonding between the sulfate group and the copper cation.[37] Spectral changes were also seen for the ν 2 mode of SO4 2− when the dehydration progresses from pentahydrate to anhydrate. The peak at 468 cm−1 (pentahydrate) split into a doublet at 432 and 482 cm−1 (trihydrate) and then shifted to 426 and 512 cm−1 (monohydrate) and later re-merged and further shifted to 459 and 495 cm−1 . The spectral band of the ν 3 mode of SO4 2− also shifted from 1143 cm−1 (pentahydrate) to 1126 cm−1 (trihydrate), and then split into a doublet at 1097 and 1206 cm−1 (monohydrate) and finally re-emerged to be a narrower doublet at 1157 and 1192 cm−1 (anhydrate). The spectral band of the ν 4 mode of SO4 2− also split from a singlet centered at 612 cm−1 (pentahydrate) into a doublet centered at 590 and 622 cm−1 (trihydrate), and then shifted to 622 and 670 cm−1 , and finally changed to be a narrower doublet at 584 and 605 cm−1 (anhydrate). The Raman bands seen below 400 cm−1 are much weaker and the band shapes are a bit broader. These Raman bands essentially belong to the lattice modes. However, spectral shifting and band splitting are also observed. As seen in Fig. 2, the band at 280 cm−1 (pentahydrate) shifted to 252 cm−1 (trihydrate), and later it split into a doublet at 246 and 272 cm−1 (monohydrate) and then finally re-merged and shifted to 262 cm−1 . The four pure component spectral estimates were mapped back onto the preprocessed spectra to obtain the relative concentration of each individual component. The calculated relative concentrations are plotted as a function of temperature in Fig. 3. On careful observation of Fig. 3, it can be seen that the changes of the pentahydrate to the trihydrate has started at temperature ∼39 ◦ C and completed at ∼41 ◦ C. The changes from the trihydrate to the monohydrate has started at a temperature ∼44 ◦ C and completed at ∼58 ◦ C. In the last dehydration step, the changes from the monohydrate to the anhydrate started at a temperature ∼153 ◦ C and completed at ∼195 ◦ C. Slight fluctuations are observed in the relative concentrations in Fig. 3. These fluctuations are due to the changes of the spectral intensities, which are caused by slight defocusing of the Raman microscope objective during in situ measurements. The transition temperatures for each dehydration step obtained from Raman analysis (in Fig. 3) are much lower than the dehydration temperatures obtained from TGA analysis (see Fig. 7). This discrepancy can be due to the tunable laser power used in the present Raman measurements (i.e. ∼1.5 mW laser power), which might be too strong for the hydrated copper sulfate. The continuous laser irradiation has been shown to induce some localized heating at the surface of the sample,[38,39] which can subsequently accelerate the dehydration process. In order to verify that this discrepancy is not in part due to the quality of the spectral reconstructions obtained, the original spectral data are compared to the calculated ones. Figure 4 shows the comparison of the original spectra data measured at 40, 45, 125, and 275 ◦ C to the spectral data reconstructed from the pure components reported in Fig. 2. As can be seen, the quality of the least-squares fit is good. The fitting error is only ∼2% and this confirms that no physically relevant spectral information is inadvertently lost during data processing. E. Widjaja, H. H. Chong and M. Tjahjono Figure 3. The relative concentration profiles as a function of temperature: (a) copper sulfate pentahydrate, (b) copper sulfate trihydrate, (c) copper sulfate monohydrate, (d) copper sulfate anhydrate. Figure 4. Comparison of original spectral data with the calculated spectral data to demonstrate that physically meaningful spectral features are retained during data processing. Solid line: original spectra; dashed-line: calculated spectra. Second measurement setup, 10× microscope objective lens, 0.1 mW laser power 184 In this second measurement setup, a lower magnification microscope objective with 10 times lower laser power was used to collect the Raman spectra of the copper sulfate sample as a function of temperature. A total of 275 in situ Raman spectra were collected and then preprocessed. The preprocessed Raman spectra can be seen in Supporting Information. The signal-to-noise ratio of the present spectra is much lower than those collected from the previous setup (Fig. 1). The highest Raman intensities were www.interscience.wiley.com/journal/jrs Figure 5. The pure component spectral estimates obtained from BTEM analysis of the second experimental data set: (a) copper sulfate pentahydrate, (b) copper sulfate trihydrate, (c) copper sulfate monohydrate, (d) copper sulfate anhydrate. only ∼300 counts. The preprocessed spectra were then subjected to singular value decomposition and BTEM analysis. Again, four distinct spectral features were observed, which were subsequently used as the band-targets for BTEM analysis. Four pure component spectral estimates were successfully reconstructed, and they are shown in Fig. 5. As shown in Fig. 5, the reconstructed Raman bands of these four spectral estimates are well resolved and the peak positions of the bands correspond well with the spectral estimates presented in Fig.2. Although signal qualities of the collected Raman spectra are quite poor due to the use of lower laser power and to the lower magnification of microscope objective, BTEM analysis is still able to recover rather good quality of the pure component spectral estimates. These four pure spectral estimates were subsequently mapped back onto the preprocessed spectra to obtain the relative concentration of each individual component as shown in Fig. 6. As expected, the concentration profiles obtained from the present measurement setup are noisier compared to those obtained from the first measurement setup. This is not surprising since lower signal-to-noise Raman spectra were collected from the second measurement setup. However, the transition temperatures for each dehydration steps can still be clearly observed from this profile. The dehydration temperatures of pentahydrate to trihydrate, trihydrate to monohydrate, and monohydrate to anhydrate are ∼57–61 ◦ C, 85–92 ◦ C, and 228–235 ◦ C, respectively. These dehydration temperatures are certainly higher compared to those indicated from the first Raman measurement setup. Thermogravimetric analysis The percentage of mass and its derivative as well as the heat flow during thermal dehydration measured from 22 to 500 ◦ C by TGA is shown in Fig. 7. The TG thermogram revealed a loss of ∼15% weight in the first step, a loss of ∼15% weight in the second step, and finally a loss of ∼7% weight in the last step. The derivatives of mass show three endothermic peaks at 67, 92, and 221 ◦ C, which c 2009 John Wiley & Sons, Ltd. Copyright
J. Raman Spectrosc. 2010, 41, 181–186 Thermo-Raman spectroscopy and chemometric analysis of hydrated inorganic samples Table 1. Comparison of the dehydration temperatures obtained from Raman microscope, TGA, and the literature Dehydration 1 2 3 (A), ◦ C (B), ◦ C TGA, ◦ C Literature,[31] ◦ C 39–41 44–58 153–195 57–61 85–92 228–235 66–67 91–93 220–223 65 90 220 Dehydration (1) refers to sample losing first two water molecules; (2) refers to loss of second two water molecules; (3) refers to loss of the fifth water molecule; (A) first measurement setup; (B) second measurement setup. Figure 6. The relative concentration profiles as a function of temperature: (a) copper sulfate pentahydrate, (b) copper sulfate trihydrate, (c) copper sulfate monohydrate, (d) copper sulfate anhydrate. Figure 7. Thermogravimetric analysis of CuSO4 ·5H2 O with a heating rate of 2 ◦ C/min from room temperature up to 500 ◦ C in a flowing nitrogen atmosphere. correspond to the dehydration temperatures for each dehydration step. The numbers of water molecules lost in these dehydration steps were 2, 2, and 1, respectively. Comparison among results obtained from Raman microscope, TGA, and the literature J. Raman Spectrosc. 2010, 41, 181–186 Conclusion The present work demonstrates the combined use of thermoRaman spectroscopy and BTEM analysis to determine the dehydration temperatures of hydrated inorganic samples as well as to identify each individual species involved in the dehydration process. This combined approach was successfully utilized for investigating the solid phase transformations of copper sulfate pentahydrate. The Raman spectral measurements performed with a low laser power could provide good estimates for the dehydration temperatures, which were in agreement with those independently determined using TGA measurements. The proposed method is rather general and certainly can be extended to investigate the dehydration and/or solid phase transformations of other hydrated inorganic solid compounds. c 2009 John Wiley & Sons, Ltd. Copyright
www.interscience.wiley.com/journal/jrs 185 Table 1 shows the comparison of the dehydration temperatures for copper sulfate pentahydrate obtained from Raman microscope, TGA analysis, and the literature. It is found that the dehydration temperatures obtained from the Raman microscope with a lower laser power (second measurement setup) and those from TGA analysis are in good agreement with the literature values.[34] Meanwhile, those dehydration temperatures determined using a higher laser lower (first measurement setup) are not comparable. These dehydration temperatures are much lower than those obtained in other studies and the reported data in the literature. This comparison strongly suggests that the higher laser power used in the first measurement setup has caused some local heating on the measured spot and has increased the heat being released to the sample, which can significantly accelerate the dehydration process. Although this situation is not applicable to all hydrated inorganic samples, some attention has to be given to the laser power used for Raman measurements. Inappropriate choice of laser power used for Raman measurements can lead to misleading interpretation of the dehydration and/or phase transformation temperatures. TGA measurement provides information on the total mass changes when the sample loses water molecules due to the dehydration process. On the other hand, Raman microscopy gives the information about the spectral changes associated with the structural changes of the dehydrated samples. These two analytical methods can be used effectively in tandem to determine the dehydration temperatures as well as to identify the structure of individual species. The use of Raman microscopy alone to determine the dehydration step is always possible as long as the laser power used for the measurement is properly chosen. In addition, the use of BTEM analysis has helped to solve the difficult analysis problems due to highly overlapping spectra and the nonexistence of some spectral libraries, since BTEM is able to reconstruct the pure component spectral estimates of underlying species of TRS data without any a priori information. The obtained pure component spectral estimates could facilitate further evaluation of the relative concentration of each individual species as a function of temperature, which, in turn, is very useful to determine the dehydration temperatures. E. Widjaja, H. H. Chong and M. Tjahjono Acknowledgements 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. Supporting information Supporting information may be found in the online version of this article. References [1] H. Chang, P. J. Huang, Anal. Chem. 1997, 69, 1485. [2] W. W. Wendlandt, Thermal Analysis (3rd edn), Wiley: New York, 1986. [3] H. Chang, P. J. Huang, J. Chin. Chem. Soc. 1998, 45, 59. [4] H. Chang, P. J. Huang, S. C. Hou, Mater. Chem. Phys. 1999, 58, 12. [5] A. Ghule, R. Murugan, H. Chang, Thermochim. Acta 2001, 371, 127. [6] A. Ghule, R. Murugan, H. Chang, Inorg. Chem. 2001, 40, 5917. [7] C. Bhongale, A. Ghule, R. Murugan, H. Chang, J. Therm. Anal. Calor. 2001, 65, 891. [8] R. L. Frost, M. L. Weier, J. Raman Spectrosc. 2004, 35, 299. [9] R. L. Frost, B. J. Reddy, Chem. Phys. Lett. 2003, 381, 262. [10] R. Murugan, A. Ghule, H. Chang, J. Appl. Phys. 1999, 86, 6779. [11] R. Murugan, A. Ghule, H. Chang, J. Phys.: Condensed Matt. 2000, 12, 677. [12] R. Murugan, A. Ghule, C. Bhongale, H. Chang, J. Mater. Chem. 2000, 10, 2157. [13] R. L. Frost, M. L. Weier, W. Martens, S. Mills, Spectrochim. Acta Part A. 2006, 63, 282. [14] M. A. Corte’s-Ja’come, G. Ferrat-Torres, L. F. F. Ortiz, C. AngelesCha’vez, E. Lo’pez-Salinas, J. Escobar, M. L. Mosqueira, J. A. ToledoAntonio, Catal. Today 2007, 126, 248. [15] H. Chang, P. J. Huang, J. Raman Spectrosc. 1998, 29, 97. [16] E. Widjaja, Development of band-target entropy minimization (BTEM) and associated software tools. PhD Thesis. National University of Singapore, Singapore, 2002. [17] E. Widjaja, C. Z. Li, M. Garland, Organometallics 2002, 21, 1991. [18] W. Chew, E. Widjaja, M. Garland, Organometallics 2002, 21, 1982. [19] L. F. Guo, A. Wiesmath, P. Sprenger, M. Garland, Anal. Chem. 2005, 77, 1655. [20] L. R. Ong, E. Widjaja, R. Stanforth, M. Garland, J. Raman Spectrosc. 2003, 34, 282. [21] S. Y. Sin, E. Widjaja, L. E. Yu, M. Garland, J. Raman Spectrosc. 2003, 34, 795. [22] C. Srilakshmi, E. Widjaja, M. Garland, B. G. Anderson, J. Raman Spectrosc. 2007, 38, 349. [23] H. J. Zhang, M. Garland, Y. Z. Zeng, P. Wu, J. Am. Soc. Mass Spec. 2003, 14, 1295. [24] L. F. Guo, F. Kooli, M. Garland, Anal. Chim. Acta 2004, 517, 229. [25] C. Z. Li, E. Widjaja, W. Chew, M. Garland, Angew. Chemie. 2002, 41, 3784. [26] E. Widjaja, J. T. Sampanthar, X. D. Han, E. Goh, Catal. Today 2008, 131, 21. [27] E. Widjaja, R. K. H. Seah, J. Pharma. Biomed. Anal. 2008, 46, 274. [28] E. Widjaja, N. Crane, T. S. Chen, M. D. Morris, M. A. Ignelzi Jr, B. R. McCreadie, Appl. Spectrosc. 2003, 57, 1353. [29] T. I. Taylor, H. P. Klug, J. Chem. Phys. 1936, 4, 601. [30] A. Reisman, J. Karlak, J. Am. Chem. Soc. 1958, 80, 6500. [31] W. W. Wendlandt, Anal. Chim. Acta 1962, 27, 309. [32] P. N. Nandi, D. A. Deshpande, V. G. Kher, Proc. Indian Acad. Sci. 1979, 88A, 113. [33] W. L. Ng, C. C. Ho, S. K. Ng, J. Inorg. Nucl. Chem. 1978, 34, 459. [34] S. EL-Houte, M. E. Ali, O. T. Sorensen, Thermochim. Acta 1989, 138, 107. [35] S. De Castelnuovo, J. B. Harness, I. J. McColm, J. Therm. Anal. Calor. 2001, 63, 233. [36] V. M. Petrusevski, V. Ivanovski, B. Soptrajanov, M. Zugik, J. Molecular Struct. 2001, 563-564, 329. [37] C. H. Chio, S. K. Sharma, D. W. Muenow, Spectrochim Acta Part A 2005, 61, 2428. [38] S. W. Song, K. S. Han, H. Fujita, M. Yoshimura, Chem. Phys. Lett. 2001, 344, 299. [39] A. Paolone, A. Sacchetti, T. Corridoni, P. Postorino, R. Cantelli, G. Rousse, C. Masquelier, Solid State Ion. 2004, 170, 135. 186 www.interscience.wiley.com/journal/jrs c 2009 John Wiley & Sons, Ltd. Copyright
J. Raman Spectrosc. 2010, 41, 181–186