Eur. Phys. J. Appl. Phys. 38, 1–5 (2007) DOI: 10.1051/epjap:2007054
THE EUROPEAN PHYSICAL JOURNAL APPLIED PHYSICS
Phase separation phenomena in Sn-Sb-Se glassy semiconductors P. Kumar, J. Kumar, and R. Thangaraja Semiconductors Laboratory, Department of Applied Physics, Guru Nanak Dev University, 143 005 Amritsar, India Received: 9 June 2006 / Received in final form: 20 January 2007 / Accepted: 13 February 2007 c EDP Sciences Published online: 14 March 2007 – Abstract. Differential scanning calorimetric (DSC) and X-ray diffraction (XRD) measurements on Sn-SbSe glassy semiconductors have been performed. Phase separation has been revealed with the broadening of the exothermic peak for 11 at% of Sn. The X-ray diffraction studies of annealed samples show that the investigated system could be treated as a solid solution of Sb2 Se3 and SnSe2 phases. The deconvolution procedure was used to investigate the crystallization mechanism for the broad exothermic peaks in the thermal scans. From the heating rate dependence of the crystallization temperature, the crystallization parameters have been determined using Kissinger equation and Matusita equation. The investigated parameters indicate higher thermal stability and glass forming ability of Sn10 Sb20 Se70 chalcogenide glass. PACS. 61.43.Fs Glasses – 64.70.Pf Glass transitions – 81.05.Gc Amorphous semiconductors
1 Introduction There has been considerable interest in amorphous semiconductors because of their interesting physical properties as well as their wide technological applications including threshold and memory switching [1–3]. More recently, they have found application in digital X-ray imaging, programmable metallization cell memories and far infrared optical fiber [4]. Selenium based chalcogenide glasses and their physical properties have attracted much interest due to their potential device applications [5–9]. Phase separation produces usually pronounced changes in the physical properties including the lowering of the glass transition temperature, lowering of optical gap and an increase of molar volumes [10]. Phase separation can exist on a variety of length scales from nanometer to micrometer or millimeter. For the purpose of photonics applications, it is important that phase separation be eliminated or at least minimized to the length scales much smaller than that of the wavelength of light being used. Apart from the technical importance, the knowledge of crystallization kinetics is very important for a better understanding of the amorphous state of these glassy systems. The Ge-Sb-Se glasses are attractive candidates for the applications requiring low transmission losses, as they are transparent to IR radiation from 2 to 16 µm [11]. Adam et al. have investigated IV-V-VI ternary chalcogenide system in which heavy mass elements has been alloyed with Se [12]. The use of heavier elements results in the small glass-forming region as compared to the compositions with lighter atomic masses. The recent interest in Sn-Sb-Se chalcogenide glasses is due to their probable application as a
e-mail:
[email protected]
low loss carbon dioxide (10.6 µm) laser power applications. The use of Sn in place of Ge is an attempt to increase the transmission window with decreased optical loss for photonic applications. Recently, the transport properties have been discussed by considering the topological phase transitions [12–14], but no studies have been reported on the crystallization kinetics and phase separation phenomena in this system. The present work is concerned with phase separation effects by evaluation of crystallization parameters for Snx Sb20 Se80−x (x = 10, 11, 12.5) chalcogenide system using the non-isothermal method. The effect of Sn addition on the structure and phase separation has been elaborated by performing X-ray diffraction on the annealed samples.
2 Experimental procedure Bulk Snx Sb20 Se80−x (x = 10, 11, 12.5) was prepared by conventional melt quenching technique [14]. To prepare the precursor alloy, high purity Sn, Sb and Se (CDH, India) was taken in a quartz glass ampoule (length 10 cm and internal diameter 0.6 cm). The contents of the ampoule were sealed under a vacuum of 10−4 mbar and heated at 1023 k for 45 h. The ampoule was constantly inverted in order to ensure the homogeneous mixing of the constituents. The melt was quenched in ice water to obtain the glass. The material was separated from the ampoule by dissolving it in a solution of HF+H2 O2 for 48 h. Differential scanning calorimetric (DSC) studies were carried out at different heating rates (5, 10 and 20 K/min.) on METTLER TOLEDO Stare system. Approximately 10 mg of the powdered sample with empty reference pans
Article published by EDP Sciences and available at http://www.edpsciences.org/epjap or http://dx.doi.org/10.1051/epjap:2007054
2
The European Physical Journal Applied Physics
Table 1. Thermal parameters and apparent activation energy for glass transition (Ea ) for Snx Sb20 Se80−x system. x at% Z Tg (K) 10.0 2.40 424.1 11.0 2.42 426.2 12.5 2.45 438.5
T0 (K) ∆T (K) B 514.6 90.5 9.3 ± 0.7 498.9 72.7 5.8 ± 0.1 498.2 59.7 3.6 ± 0.6
Ea (kJ/mol) 104.2 ± 3.7 247.5 ± 2.5 320.6 ± 7.5
Fig. 1. Typical DSC traces for Snx Sb20 Se80−x (x 10, 11, 12.5) glassy semiconductors.
=
were taken for DSC runs. The DSC equipment was calibrated with standard materials prior to start of the experiment. All the DSC scans were performed under N2 atmosphere. The crystallized fraction (X) was calculated using partial area analysis. A best fit for the results was calculated by the least square method. The deconvolution of the broad exothermic peaks was evaluated by using the peakFit v4.12 (SeaSolve, USA) software program. The DSC data for the exothermic peak was well fitted by using the Savitzky-Golay algorithm. The best-fitted parameters for the deconvoluted peaks are summarized in Table 1 for the two samples. Annealing of the powdered samples has been performed in the sealed quartz ampoules (∼10−5 mbar) at peak crystallization temperature for 1 h. A Philips X-ray diffractometer type-1710 was used to identify the amorphous/crystalline phases in the as-prepared and annealed glassy samples. The patterns were recorded with Cu Kα line.
3 Results and discussions Calorimetric studies can be performed in two ways viz. isothermal and non-isothermal conditions. In isothermal method, the sample is brought quickly to a temperature above the glass transition temperature and heat evolved during the crystallization process is recorded as a function of time. In the second case, the sample is heated at a fixed rate and heat evolved is again recorded as a function of temperature or time. The present experiment was performed in the non-isothermal way. Figure 1 shows the DSC curves for Snx Sb20 Se80−x (x = 10, 11, 12.5) glass, at 10 K/min. heating rate. Two characteristic phenom-
ena (endothermic and exothermic peaks) are evident in the DSC curves in the temperature range of study. The glass transition temperature (Tg ) and onset of crystallization temperature (To ) was taken as the temperature corresponding to the intersection of the two linear portions adjoining the transition elbow for endothermic and exothermic peaks respectively, while the peak of the crystallization exotherm was used for the determination of peak crystallization temperature (Tp ). Table 1 shows an increase in Tg and decrease in To with the increase in the Sn concentration. The increase in Tg with Sn concentration (x > 10) can be explained as follows. The addition of Sn is at the expense of Se concentration, which enters into the (Se)n chains connecting the existing pyramidal Sb(Se1/2 )3 and tetrahedral (SnSe1/2 )4 and may also cross-link the Se chains. The cross-linking by the addition of Sn further strengthens the bond structure and therefore accounts for the increase of Tg with the increase in Sn content. This increase of Tg may also be due to the increase in the average coordination number (Z) with the increase in the Sn content. The coordination state of Sn, Sb and Se as 4, 3 and 2 respectively was taken for calculating the average coordination number of Snx Sby Sez glass system as: Z = (4x + 3y + 2z)/(x + y + z). Thus, the variation in Tg with Z could be due to the increase in the structural connectivity of the glass network [13,14]. A single Tg in the thermal scans shows that these glasses are compositionally homogeneous while two or more crystallization peaks for x = 11.0, 12.5 suggest the occurrence of phase separation in the present system. A single exothermic peak was observed for x = 10 glass sample. The composition dependence of thermal scans shows that the area of crystallization exotherm broadens as the Sn concentration increases. This shows the crystallization of different phases occurs as the Sn content increases. The phase separation effects have been observed in the range of composition with average coordination number between 2.40 and 2.67 [4,10,14]. The phase separation effects have been observed beyond Z = 2.40 in the present system. The difference ∆T = To − Tg , which is an indication of the thermal stability of glasses [15], decreases as the Sn content increases in the glass. The glass, thus, shows an increased tendency towards crystallization and results in appearance of additional exothermic peak for higher Sn content. This could be accounted by the phase separation effects at the higher Sn concentration [10]. Thus, the kinetic resistance to crystallization is decreased by increasing Sn content, which results in higher thermal stability for x = 10 in the present glassy system. The broad exothermic peak for x = 11.0, 12.5 on deconvolution gives two onset temperatures (To ) and two peak crystallization temperatures (Tp ) corresponding to the first and second peaks respectively. The deconvoluted exothermic peaks are shown in Figure 2 and the peak fit parameters are shown in Table 2. These parameters suggest that as the content of Sn increases, the area under the second peak decreases with the decrease in the amplitude of the deconvoluted peak. This shows that the addition of
P. Kumar et al.: Phase separation phenomena in Sn-Sb-Se chalcogenide glassy
Fig. 2. Exothermic peaks before and after deconvolution peak for (a) Sn11 Sb20 Se69 and (b) Sn12.5 Sb20 Se67.5 glassy samples. The upper peak shows the best theoretical fit with the simulated base line for the peak while the two deconvoluted exotherms are shown in the lower part. The peak positions at the respective temperatures were also shown in the figure.
3
Fig. 3. X-ray diffraction pattern for (a) Sn11 Sb20 Se69 annealed at 510 K, (b) and (c) Sn12.5 Sb20 Se67.5 annealed at 515 and 550 K respectively for 1h. The symbols () and () are used for the crystallization fractions of c-Sb2 Se3 and c-SnSe2 phases respectively in the diffraction patterns.
Table 2. Deconvolution parameters of the investigated two glass compositions. x at.% Peak T0 (K) Tp (K) %area Amplitude R2 | 498.2 508.5 54.3 7.33 11.0 0.9945 508.3 518.9 45.7 6.17 | 496.6 506.4 67.8 8.23 12.5 0.9835 509.3 519.1 32.2 3.92
Sn leads to the separation of some phase with lower crystallization temperatures. The higher temperature peak is the same as that of the x = 10 sample. This shows that the phase responsible for the crystallization in x = 10 decreases and higher values of x results in the appearance of new phase. The diffraction patterns (XRD) of the annealed samples are shown in Figure 3. This figure tells us that crystallization of Sb2 Se3 and SnSe2 takes place. Thus the present glassy alloy could be treated as a solid solution of some Sb-rich and Sn-rich phases. 3.1 The glass transition The dependence of Tg on the heating rate has been studied by two different formulations. The first one corresponds to the empirical relation of the form [16], Tg = A + B ln(α) where A and B are constants for a given glass composition and α is the heating rate. It is found that this equation holds good for the studied samples. Figure 4 shows the validity of this relation for Snx Sb20 Se80−x glass system. The value of B is found to decrease with the increase of Sn content (Tab. 1). A change in the value of B indicates that structural changes are taking place in the samples with the change in composition [16]. In the second approach, the heating rate dependence of the glass transition temperature in chalcogenide glasses [5–9,15] is interpreted in terms of the glass transition activation energy. The Kissinger peak shift method
Fig. 4. Dependence of glass transition temperature (Tg ) on the heating rate (α) for Snx Sb20 Se80−x (x = 10, 11, 12.5) glassy semiconductors.
was used to calculate activation energy for glass transition, in spite of the fact that this equation was used for the evaluation of crystallization activation energy. The apparent activation energy for glass transition Ea can be calculated from the Tg dependence on heating rate, α by using the Kissinger equation [17]: ln (T2g /α) + const. = Ea /RTg
(1)
where Ea is the apparent activation energy for glass transition. The plots of ln(α/T2g ) against 1/Tg for all the samples show a linear behavior as seen in Figure 5. The values of Ea obtained are listed in Table 1. The apparent activation energy for glass transition increases with the increase in Sn content. This indicates that the network dimensionality plays a major role as compared to the phase separation effects in explaining the glass transition temperature for the present glass system.
4
The European Physical Journal Applied Physics
Fig. 5. Plot of ln (α/Tg2 ) versus 1000/Tg for Snx Sb20 Se80−x (x = 10, 11, 12.5) glassy semiconductors.
Fig. 6. Plot of ln(α) versus 1000/Tp for the deconvoluted peak of Sn11 Sb20 Se69 glassy semiconductor.
3.2 Crystallization kinetics The crystallization process is generally understood when the activation energy for crystallization (Ec ), the order of crystallization mechanism (n) and frequency factor (Ko ) are known. The activation energy for crystallization (Ec ) has been obtained using the modified Kissinger equation [19,20] ln (αn /Tp2) = −mEc /RTp + ln k
(2)
where k is a constant, containing factors depending on the thermal history of the samples, n and m are the constants having values between 1 and 4 depending on the morphology of growth. The plots of ln α vs. 1000/Tp are straight lines for all the samples. Figure 6 shows a plot for the two deconvoluted peaks in Sn11 Sb20 Se69 glassy sample. From the slope of the curves, the value of mEc /n was determined and are listed in Table 3. According to Matusita et al. [21] for non-isothermal crystallization ln [− ln (1 − X)] = −n ln α − 1.052mEc/RT + const. (3) where X is the volume fraction of crystals precipitated in the glass heated at uniform rate, Ec the activation energy for crystallization and R is the gas constant. The
value of mEc is obtained from the straight-line portions in plots of ln [− ln (1 − X)] against 1/T as shown in Figure 7 and values are summarized in Table 3. There is a break in the straight-line portions, which is attributed to the saturation of nucleation sites in the final stages of crystallization [22] or to the restriction of the crystal growth in the small size of the particles [23]. The analysis is restricted to the initial linear region extending over a large range. The value of n is obtained from the slopes of ln[− ln(1−X)] against ln(α) curve. Figure 8 shows the plot of ln[− ln(1 − X)] versus ln(α) at different temperatures for two different crystallization peaks for Sn11 Sb20 Se69 chalcogenide glass. The value of m is taken to be n−1 since no prior heat treatment was given to the samples before the thermal analysis runs [3]. For the second and higher order exothermic peaks, the nucleation centers were formed due to the prior crystallized fraction and hence, the value of m was taken equal to n for those peaks. The values of n and m are shown in Table 3. The first peak in the deconvoluted exotherm is due to bulk nucleation with threedimensional growth. However, for the second and higher order peaks in the thermal scans for glasses with x = 11, 12.5 sample, the value of m was taken equal to n, since crystallization of previously crystallized part acts as the nucleation centers for the considered peak [15]. The result shows that all the phases crystallized in the present system are by bulk nucleation with three-dimensional growth. The values of Ec evaluated using these values of n and m are given in Table 3. The value of Ec for the first peak increases with the increase in the Sn content as evident from Table 3. Since the activation energy is an indication of the speed of crystallization [15], one can argue that the composition with x = 12.5 has the maximum rate of crystallization. In our previous report [14], we showed that the addition of Sn decreases the crystalline nature of Sb20 Se80 and amorphous samples could be obtained for 8 x 18 while further increase in Sn content gives crystalline samples. The crystallization activation energy and thermal stability results also support the previous results. The values obtained for Ec are shown in Table 3. The comparison of the Ec values obtained using two different formulations shows that they give different values of activation energies for a given sample while the trend remains the same. In the Kissinger method, the variation of peak crystallization temperature with heating rate was considered for evaluating the activation energy. While the Matusita method takes into account the process of nucleation and growth for the crystallization transformation for calculating activation energy, the Kissinger method does not. This probably results in the difference in the activation energies given in Table 3. A similar discrepancy in the activation energies calculated by using different formalisms in other chalcogenide glasses has also been observed [8,15,18].
4 Conclusions Calorimetric studies and phase separation effects have been elaborated in Sn-Sb-Se glassy system by
P. Kumar et al.: Phase separation phenomena in Sn-Sb-Se chalcogenide glassy
5
Table 3. Parameters determined from heating rate dependence of Tp in Snx Sb20 Se80−x (x = 10, 11, 12.5) glassy semiconductor. In the table, KM and MM stands for Kissinger method and Matusitas method respectively. x at.% 10.0 11.0 12.5
Peak | | | |
mEo /n(kJ/mol) 233.1 ± 0.1 149.3 ± 3.4 131.3 ± 2.7 227.4 ± 0.2 198.3 ± 1.1 127.9 ± 6.1
n 4.1 ± 0.1 4.3 ± 0.4 4.9 ± 0.4 4.1 ± 0.2 3.9 ± 0.5 2.9 ± 0.1
m 3 3 3 3 3 3
Ec (KM )(kJ/mol) 170.5 217.5 214.9 312.3 191.7 123.6
Fig. 7. The ln [−ln(1 − X)] versus 1000/T plots for Sn11 Sb20 Se69 glassy semiconductor. The dark and open symbols show the analysis deconvoluted first and second peaks for the crystallization exotherm respectively at different heating rates.
investigating the crystallization parameters. The Ea and Ec increase with the increase in Sn content. The crystalline transformation has been found to occur with growth in three-dimensions. The XRD pattern of the annealed samples shows that the present system could be considered as a solid solution of Sn-rich and Sb-rich phases. The authors are thankful to Mr. Jagtar Singh, Sophisticated Analytical Instrumentation Facility (SAIF), Punjab University, Chandigarh and Mr. Nripendra Singh, Central Instrumentation Center (CIL), NIPER, Chandigarh for their kind help in taking the XRD and DSC studies respectively.
References 1. S.R. Elliot, Physics of amorphous materials (Longman, London, 1991) 2. R.M. Mehra, R. Kumar, P.C. Mathur, Thin Solid Films 170, 15 (1989) 3. M.S. Kamboj, R. Thangaraj, Eur. Phys. J. Appl. Phys. 24, 33 (2003) 4. Fei Wang, P Boolchand, in Non-Crystalline Materials for Optoelectronics, edited by G Lucovsky, M. Popescu (INOE, 2004), Vol. 1 5. K. Ramesh, S. Asokan, K.S. Sangunni, E.S.R. Gopal, Eur. Phys. J. B 6, 207 (1998)
mEc (kJ/mol) 290.8 ± 3.5 348.3 ± 4.1 385.5 ± 3.7 358.6 ± 2.8 408.5 ± 3.5 436.2 ± 7.3
Ec (M M )(kJ/mol) 96.9 116.1 128.5 179.3 136.2 145.4
Fig. 8. Plot of ln [− ln (1 − X)] against 1000/T for the Sn11 Sb20 Se69 glass at different temperatures. The first (a) and second (b) deconvoluted peaks are shown in the figure.
6. N.A. Bhat, K.S. Sangunni, A. Kumar, Phys. Chem. Glasses 44, 296 (2003) 7. M.A. Abdel Rahim, Physica B 239, 238 (1997) 8. N. Mehta, A. Kumar, Mater. Chem. Phys. 96, 73 (2006) 9. M.M. Wakked, E. Kh. Shokr, S.H. Mohamed, Phys. Stat. Sol. (a) 183, 399 (2001). 10. P. Boolchand, D.G. Georgiev, T. Qu, F. Wang, L. Chai, S. Chakravarty, C. R. Chem. 5, 713 (2002) 11. P. Klocek, M. Roth, R.D. Rock, Opt. Eng. 26, 88 (1987) 12. A.B. Adam, S. Sakrani, Y. Wahab, J. Mater. Sci. 40, 1571 (2005) 13. P. Kumar, K.S. Bindra, N. Suri, R. Thangaraj, J. Phys. D: Appl. Phys. 39, 642 (2006) 14. P. Kumar, R. Thangaraj, J. Non-Cryst. Solids 352, 2288 (2006) 15. S. Mahadevan, A. Giridhar, A.K. Singh, J. Non-Cryst. Solids 88, 11 (1986) 16. M. Lasocka, Mater. Sci. Engng. 23, 173 (1976) 17. H.E. Kissinger, J. Res. Mat. Bur. Stand. 57, 217 (1956) 18. M.M.A. Imran, D. Bhandari, N.S. Saxena, Physica B 293, 394 (2001) 19. K. Matusita, S. Sakka, Phys. Chem. Glasses 20, 81 (1979) 20. D.R. Macfarlane, M. Matecki, M. Poulain, J. Non-Cryst. Solids 64, 291 (1984) 21. K. Matusita, T. Konatsu, R. Yokata, J. Mater. Sci. 19, 291 (1984) 22. J. Colemenero, J.M. Bandarian, J. Non-Cryst. Solids 30, 263 (1978) 23. R.F. Speyer, S.H. Risbud, Phys. Chem. Glasses 24, 26 (1983)
