J. Am. Ceram. Soc., 96 [10] 3035–3037 (2013) DOI: 10.1111/jace.12562 © 2013 The American Ceramic Society
Journal
Rapid Communication
Calorimetric Signature of Structural Heterogeneity in a Ternary Silicate Glass Yanfei Zhang,‡,§ Guang Yang,¶ and Yuanzheng Yue‡,§,†,* ‡
Key Laboratory of Processing and Testing Technology of Glass & Functional Ceramics of Shandong Province, Qilu University of Technology, Jinan 250353, China §
Section of Chemistry, Aalborg University, Aalborg DK-9000, Denmark
¶
Electronic Materials Research Laboratory, Key Laboratory of the Ministry of Education & International Center for Dielectric Research, Xi’an Jiaotong University, Xi’an 710049, China Using the hyperquenched–annealing–DSC method, it has been found that the stepwise pattern of energy release with increasing annealing temperature is due to the structural heterogeneity.9 In this work, we show the detailed calorimetric signature of the structural heterogeneity in a hyperquenched glass, i.e., commercially available wool fibers (4–10 lm in diameter and 5–20 mm in length) with a composition of 20.8CaO–15.6MgO–62.2SiO2 (in wt%). This type of wool fibers was attenuated from the corresponding melt at temperatures around 1600°C by means of the centrifugal process. During this process, the fibers were hyperquenched at about 106 K/s,10 and thus are referred to as HQG. Part of the HQG samples were subjected to sub-Tg annealing, i.e., annealing below Tg.
We investigate the structural heterogeneity in a silicate glass by hyperquenching–annealing–calorimetry approach. The results show a striking phenomenon: two separated sub-Tg relaxation peaks appear on the calorimetric curve of the hyperquenched CaO–MgO–SiO2 glass, implying the existence of two distinct structural domains of higher and lower potential energies, respectively. The higher energy domains in nanoscale are so unstable that they become ordered during hyperquenching. This is verified by the high-resolution transmission electron microscopy image exhibiting nanoordered domains in the glass matrix. The higher energy domains relax similar to a strong glass phase, whereas the lower energy domains do similar to a fragile one.
I.
T
Introduction II.
glass transition and relaxation phenomena have been a long-standing subject of the condensed matter physics and material science.1 To understand these phenomena, the structural and dynamic heterogeneities have become one of the key issues to be investigated.2–5 The dynamic heterogeneity in a supercooled liquid implies the significant fluctuations of the dynamics during quenching as a result of the reduction in the number of independent regions. The dynamic heterogeneity is first believed to be seen through some dynamic variables.3,5 The recent findings show the direct evidence for dynamic heterogeneities from both calorimetric analysis and modeling of selenium glass relaxation during sub-Tg isothermal annealing.6 However, in contrast to the common belief, the features of dynamic heterogeneity are considered to be of structural origin. Extensive efforts have been devoted to clarifying what local structural properties influence the dynamic heterogeneity. These investigations focused on the form of clusters of low molecular mobility,3 crystal-like bond orientational order,4 or icosahedral order.7 Apart from the structural properties, the potential energy in glass is also correlated with the dynamics at certain length scale. It has been found that a molecule with lower potential energy is less mobile than that with higher potential energy.8 Previous studies of the enthalpy relaxation in the hyperquenched glass (HQG) have provided information on the energetic/structural heterogeneity in the supercooled liquids.9 HE
Experimental Procedure
The sub-Tg enthalpy relaxation of the studied HQG was detected using a differential scanning calorimeter (DSC) (NETZSCH STA 449C Jupiter, NETZSCH-Gerätebau GmbH, Selb, Germany). The fresh and sub-Tg annealed samples were up- and downscanned at 20 K/min in the DSC with argon atmosphere. The detailed procedure has been described elsewhere.9,11,12 The standard glass used here refers to the sample subjected to the standard cooling rate of 20 K/min. A high-resolution transmission electron microscope (HRTEM) (FEI Titan G2 60-300 microscope, FEI Company, Hillsboro, OR, equipped with image spherical aberration corrector operated at 300 kV) was used to directly visualize the heterogeneous structure in the studied HQG.
III.
Results and Discussion
Figs. 1(a) and (b) show, respectively, the DSC output and the heat capacity (Cp) of the fresh wool fibers (i.e., HQG) determined in DSC during upscanning in argon. In Fig. 1(a), we can see a sub-Tg exothermic relaxation peak between 550 and 1000 K, followed by an endothermic peak responsible for glass transition. Furthermore, the DSC output curve also shows a tiny exothermic peak right above the glass transition overshoot, followed by a big, sharp exothermic peak, indicating the formation of one minor and one major crystalline phase. Here, we take a close look at the sub-Tg relaxation [marked by the box of Fig. 1 (a)]. First, a striking phenomenon can be observed, i.e., the exothermic relaxation peak is composed of two well-separated peaks [see the Cp1 curve in Fig. 1 (b)] rather than only one with a shoulder [see the inset of Fig. 1(b)].9 The two well-separated relaxation peaks appeared on the first upscan heat capacity curve (Cp1) obtained at 20 K/min. This implies that two different structural domains coexist in the HQG, and hence, in the supercooled liquid at a temperature
J. C. Mauro—contributing editor
Manuscript No. 33481. Received July 9, 2013; approved July 18, 2013. *Member, The American Ceramic Society. † Author to whom correspondence should be addressed. e-mail:
[email protected]
3035
3036
1.6
Ta=823 K
0
-1
-1
Cp (Jg K )
1.4 1.5
-30 -1
-1
Cp (Jg K )
-1 -1
DSC output (Js g )
Vol. 96, No. 10
Rapid Communications of the American Ceramic Society
1.2
-60
600
900
1200
(a)
T (K)
600
800
1000
1200
0.8
1400
(a)
400
600
800
0.9 400
600
-1 1000
T (K)
Cp2
0.1
0.0 400
0.1
600
800
1000
T (K)
HQ glass
1d
6h 1h
4d
0.0
Cp1 600
0.2
-1 -1 C (Jg K ) p
800
-1
-1
-1
Cp (Jg K )
400
1.2
Cp,excess (Jg K )
-1
Cp1
0.8
-1
Cp (Jg K )
Cp2
1.0
Ta=823 K
Basaltic HQ Glass
0.2
1.4 1.2
1000
T (K)
T (K) 1.6
1.5
4d
1h 6h HQ glass 1d
1.0
0.9
400
standard glass
1.2
(b)
800
1000
T (K) Fig. 1. (a) The differential scanning calorimeter (DSC) output of the hyperquenched glass (HQG) versus temperature (T). Inset: the Cp curve taken from the box of the DSC curve. (b) Isobaric heat capacity (Cp) versus temperature (T) for the HQG measured during the first upscan (Cp1) and second upscan (Cp2) at 20 K/min in argon. Inset: the Cp curves of the basaltic HQG for comparison.11
(i.e., the fictive temperature Tf) well above Tg.10 Each of both domains has its own structural/energetic distribution expressed by a certain stretching exponent b.13 The b value for the structural domain corresponding to the low-temperature relaxation peak is calculated to be 0.51, whereas the b value for the structural domain corresponding to the high-temperature relaxation peak is calculated to be a smaller value, i.e., 0.36, indicating a more stretched structural relaxation. In contrast, an ordinary HQG, e.g., a basaltic HQG [inset of Fig. 1 (b)] exhibits a broad relaxation peak with a weak shoulder at lower temperature.9,11 Cp2 in Fig. 1(b) refers to the heat capacity measured during the second DSC upscan (i.e., the standard curve) at 20 K/min. To obtain the Cp2 curve of the pure glass, crystallization of the HQG must be avoided during the first upscan, and thus the maximum temperature of the first upscan is set to be 1033 K. Fig. 2(a) shows the evolution of the two relaxation peaks with the annealing time (ta) at the temperature of 823 K, which reflects the dynamic features of both structural domains. It can be seen that both relaxation peaks become smaller with increasing ta, but in a different manner. The structural domain corresponding to the lower temperature relaxation peak must possess higher potential energy so that low kinetic energy can induce the rearrangement of the structural units to attain the energy state of the standard glass.9 On the contrary, the local structural domains corresponding to the higher temperature relaxation peak must have a low potential energy. This is indicative of the structural heterogeneity in the studied glass.
(b)
400
600
800
1000
T (K) Fig. 2. (a) Influence of the annealing time (ta) on the two sub-Tg relaxation peaks of the Cp curve of the hyperquenched glass (HQG) annealed at 823 K. (b) Influence of ta on the excess heat capacity Cp, Cp1) of the HQG annealed at 823 K. Inset: the Cp,excess excess (Cp2 of the basaltic HQG11 for comparison.
To further identify the structural heterogeneity, we trace the enthalpy evolution with ta by plotting the excess heat capacity (Cp,excess = Cp2 Cp1) against temperature for the samples annealed for different ta at 823 K as shown in Fig. 2(b). For the studied HQG, the Cp,excess curve exhibits two distinct peaks at low and high temperatures, respectively. In contrast, for an ordinary silicate HQG, there is a broad peak with a shoulder on the low-temperature side [see the inset of Fig. 2(b)]. According to Yue et al,11 the existence of the shoulder implies the overlapping of two relaxing structures below Tg. Moreover, the shoulder disappears more rapidly with the annealing degree. In Fig. 2(b), the lowtemperature peak vertically diminishes with increasing ta, and this is similar to the relaxation behavior in strong glass formers.14 In contrast, the high-temperature peak horizontally decreases with increasing ta. In addition, as ta is further extended, e.g., to 4 days, an endothermic prepeak occurs just prior to the high-temperature exothermic peak as shown in Ref. [11]. The occurrence of the prepeak indicates the structural heterogeneity in glass. The structural domain corresponding to the higher temperature exothermic peak relaxes in a similar manner (horizontally) as does a fragile glass former.11 This is further substantiated by the fact that a more fragile system exhibits a smaller stretching exponent b.15 These findings imply that the two exothermic peaks below Tg are associated with two different local structures trapped by hyperquenching: one behaves as a strong glass phase, whereas another does as a fragile one. This clearly indicates that the structure of the studied HQG is rather heterogeneous.
October 2013
Rapid Communications of the American Ceramic Society
3037
studied HQG. Each domain itself is also structurally heterogeneous and this is confirmed by the features of the sub-Tg relaxation. The structural domain associated with the low-temperature peak relaxes similar to a strong glass former when increasing the annealing time. In contrast, the domain associated with the high-temperature peak behaves similar to a fragile glass former. By means of HRTEM, the highly ordered nanodomains are discovered in the HQG sample. This suggests that the formation of the ordered domains cannot be avoided even by hyperquenching, implying that some structural domains are extremely unstable.
Acknowledgment
Fig. 3. The HRTEM image of the studied hyperquenched glass. The marked circles indicate the ordered structures, i.e., nanocrystals.
Figure 3 is a HRTEM image of the studied HQG, which shows the existence of the nanoordered domains (or nanocrystals) in the glass matrix (see the area in the circles). Similar crystalline features are seen in all observed area. It should be noted that no indication of crystalline formation due to electron beam irradiation was observed during TEM analysis in the glass samples. Thus, the formation of the nanocrystals must occur during the hyperquenching process. This implies that the structural heterogeneity exists in the supercooled liquid, which facilitates the formation of nanocrystals. In terms of the potential energy landscape, the domain with the ordered structure has lower potential energy, i.e., higher activation barrier, than the domain in the amorphous state,1,4 and thus has higher relaxation time,16 correspondingly results in lower mobility.8 Therefore, during sub-Tg annealing, the different local structural domains (i.e., the ordered/disordered domains and their boundary regimes) must have different mobilities. It has been reported that the crystal growth from the melt is generally along the crystal–melt interface as the free energy of the domains at the crystal–melt interface are higher than those in both crystal and amorphous phases.17 Considering that there are two different local structural domains in the studied HQG, we can infer that the structural domains corresponding to the crystal–melt interface have higher potential energy than those corresponding to both crystal and amorphous phases. The nanocrystals first originate from the higher potential energy domains. This further verifies that a large extent of structural heterogeneity exists in the studied glass. The main implication of both the attained enthalpy relaxation patterns and the HRTEM image of the studied HQG is that structural heterogeneity already exists in the supercooled liquid. Such an implication is helpful for understanding the relation between structural heterogeneity and glass-forming ability as evidenced by our recent study (Y. F. Zhang, L. N. Hu, S. J. Liu, C. F. Zhu and Y. Z. Yue, under review). According to both this work and previous studies,18,19 the structural heterogeneity could be a universal phenomenon of glass-forming liquids.
IV.
Conclusions
In summary, two structural domains with higher and lower potential energies, respectively, are verified to exist in the
This work is financially supported by the Danish Research Council under Grant No. 09-064216 and the Taishan Scholar Fund of Shandong Province, China. GY acknowledge the funding from the National Natural Science Foundation of China (Grant No 51202180) and the Fundamental Research Funds for the Central Universities in China.
References 1
F. H. Stillinger, “A Topographic View of Supercooled Liquids and Glass Formation,” Science, 267, 1935–9 (1995). 2 K. D. Vargheese, A. Tandia, and J. C. Mauro, “Origin of dynamical heterogeneities in calcium aluminosilicate liquids”, J. Chem. Phys., 132, 194501, 9 pp (2010). 3 G. S. Matharoo, M. S. G. Razul, and P. H. Poole, “Structural and Dynamical Heterogeneity in a Glass-Forming Liquid”, Phys. Rev. E, 74, 050502, 4 pp (2006). 4 T. Kawasaki, T. Araki, and H. Tanaka, “Correlation Between Dynamic Heterogeneity and Medium-Range Order in Two-Dimensional Glass-Forming Liquids,” Phys. Rev. Lett., 99, 215701, 4 pp (2007). 5 I. Ladadwa and H. Teichler, “Low-Frequency Dynamical Heterogeneity in Simulated Amorphous Ni0.5Zr0.5 Below its Glass Temperature: Correlations with Cage Volume and Local Order Fluctuations,” Phys. Rev. E, 73, 031501, 11 pp (2006). 6 O. Gulbiten, J. C. Mauro, and P. Lucas, “Relaxation of Enthalpy Fluctuations During sub-Tg Annealing of Glassy Selenium”, J. Chem. Phys., 138, 244504, 7 pp (2013). 7 F. Sausset and G. Tarjus, “Growing Static and Dynamic Length Scales in a Glass-Forming Liquid”, Phys. Rev. Lett., 104, 065701, 4 pp (2010). 8 C. Donati, S. C. Glotzer, P. H. Poole, W. Kob, and S. J. Plimpton, “Spatial Correlations of Mobility and Immobility in a Glass-Forming LennardJones Liquid,” Phys. Rev. E, 60, 3107–19 (1999). 9 Y. Z. Yue, “Influence of Physical Ageing on the Excessive Heat Capacity of Hyperquenched Silicate Glass Fibers,” J. Non-Cryst. Solids, 348, 72–7 (2004). 10 Y. Z. Yue, R. von der Ohe, and S. L. Jensen, “Fictive Temperature, Cooling Rate, and Viscosity of Glasses,” J. Chem. Phys., 120, 8053–9 (2004). 11 Y. Z. Yue, S. L. Jensen, and J. deC. Christiansen, “Physical Aging in a Hyperquenched Glass,” Appl. Phys.Lett., 81 2983–5 (2002). 12 M. Ya, J. Deubener, and Y. Z. Yue, “Enthalpy and Anisotropy Relaxation of Glass Fibers,” J. Am. Ceram. Soc., 91, 745–52 (2008). 13 G. Williams and D. C. Watts, “Non-Symmetrical Dielectric Relaxation Behaviour Arising from a Simple Empirical Decay Function,” Trans. Faraday Soc., 66, 80–5 (1970). 14 L. N. Hu and Y. Z. Yue, “Secondary Relaxation Behavior in a Strong Glass,” J. Phys. Chem. B, 112, 9053–7 (2008). 15 G. J. Fan, J. F. Loffler, R. K. Wunderlich, and H. J. Fecht, “Thermodynamics, Enthalpy Relaxation and Fragility of the Bulk Metallic Glass-Forming Liquid Pd43Ni10Cu27P20,” Acta Mater., 52, 667–74 (2004). 16 J. C. Mauro, S. S. Uzun, W. Bras, and S. Sen, “Nonmonotonic Evolution of Density Fluctuations during Glass Relaxation”, Phys. Rev. Lett., 102, 155506, 4 pp (2009). 17 K. A. Jackson, D. R. Uhlmann, and J. D. Hunt, “On the Nature of Crystal Growth from the Melt,” J. Cryst. Growth, 1, 1–36 (1967). 18 H. F. Yuan, T. Xia, M. Plazanet, B. Deme, and M. Orrit, “Communication: Crystallite Nucleation in Supercooled Glycerol Near the Glass Transition”, J. Chem. Phys., 136, 041102, 4 pp (2012). 19 Y. Z. Yue, “Experimental Evidence for the Existence of an Ordered Structure in a Silicate Liquid Above its Liquidus Temperature,” J. Non-Cryst. Solids, 345–346 523–7 (2004). h
