Anomalous Transformation of Vapor-Deposited Highly Stable Glasses

Letter pubs.acs.org/JPCL

Anomalous Transformation of Vapor-Deposited Highly Stable Glasses of Toluene into Mixed Glassy States by Annealing Above Tg A. Sepúlveda,† E. Leon-Gutierrez,† M. Gonzalez-Silveira,† M. T. Clavaguera-Mora,† and J. Rodríguez-Viejo*,†,‡ †

Nanomaterials and Microsystems Group, Physics Department, Universidad Autónoma de Barcelona, 08193 Bellaterra, Spain MATGAS Research Center, Campus UAB, 08193 Bellaterra, Spain

‡

S Supporting Information *

ABSTRACT: Vapor-deposited glasses have recently emerged as a remarkable new class of materials that can form much denser and stable glasses than those obtained by cooling the liquid. These new amorphous materials reach lower regions of the energy landscape and may impact important technologies that use vapor-deposition. Here, we report on the formation of a glass with two distinct glassy states obtained through the partial annealing of highly stable vapor-deposited glassy films of toluene. The resulting glass exhibits two clear heat capacity overshoots with different onset and fictive temperatures. The transformation times of the ultrastable glass are around 105 times slower than the structural relaxation time (τα) of supercooled liquid toluene. We show that the nature of the transformed glass depends on the annealing temperature above Tg. This finding suggests the formation of distinct supercooled liquids at temperatures slightly above Tg during the transformation of the highly stable glass. Our results are compatible with the existence of polyamorphism in toluene. SECTION: Macromolecules, Soft Matter

T

(naphthylbenzene) (TNB) and IMC glasses and found that the transformation above Tg is exceedingly slow, as much as 4500 times slower than the structural relaxation time of the liquid. A growth-front mechanism initiated at the surface and evolving at a constant velocity was considered to account for those results.15 The mechanism that transforms glass into a supercooled liquid upon heating may be directly inferred from scanning calorimetry, provided that there is a clear distinction of the glass stability levels prior to and after the transformation. Our investigation enables this direct analysis from heat capacity data. We showed previously that vapor-deposited glasses of toluene grown at around 0.80Tg (Tg = 117 K) have low enthalpies and large heat capacity overshoots; these are shifted to high temperatures with respect to samples slowly cooled from the liquid.18,19 In this Letter, we analyze the influence of isothermal treatments above Tg on the heat capacity of vapor-deposited ultrastable glassy films of toluene. We follow the kinetics of the transformation and identify the influence of the annealing temperature on the final properties of the glass. We discuss our results in relation to the possible existence of a LL transition in toluene. We deposited thin films of toluene and followed the annealing protocol described in the Experimental Section and

he glass transition observed in the laboratory is a kinetic event that depends on an experimental time scale relative to the time scales for molecular rearrangement.1−3 The presence of two resolved calorimetric glass transitions in a single-component molecular glass former is very unusual and was recently observed by Ediger et al.4 in highly stable indomethacin (IMC) glasses prepared by vapor deposition. The two calorimetric peaks were attributed to the coexistence of glasses having different stability as a result of different packing arrangements in the vapor-deposited glass.5 Today, it is generally accepted that a single-component liquid can exhibit different liquid phases. Probably, the best-known example of polyamorphism is water, where high-density amorphous water ice undergoes a first-order transition into the low-density form.6 Other materials such as silica, silicon, and yttria−alumina exhibit polyamorphism.7−9 The presence of a liquid−liquid (LL) transition has also been observed in organic compounds like triphenyl phospite (TPP),10,11 where the so-called glacial phase (glass II) is formed at the expenses of the supercooled liquid (liquid I) during an isothermal transformation at temperatures between TgI = 205 K and TgII = 230 K. However, the nature of the glacial phase is still a matter of debate.12,13 Ishii and co-workers14 have argued the presence of a LL transition in isopropylbenzene (IPB), which suggests the possibility of polyamorphism in benzene simple molecule glass formers. Another related intriguing aspect of highly stable glasses is their transformation into the liquid state. Ediger and coworkers15−17 conducted experiments on vapor-deposited tris© 2012 American Chemical Society

Received: December 22, 2011 Accepted: March 13, 2012 Published: March 13, 2012 919

dx.doi.org/10.1021/jz201681v | J. Phys. Chem. Lett. 2012, 3, 919−923

The Journal of Physical Chemistry Letters

Letter

Figure 1. Specific heat versus temperature following annealing at T = (a) 122, (b) 124, and (c) 128 K of stable vapor-deposited films. The lines only provide guidance to the eye. The vertical arrow in (c) indicates the time sequence. AD refers to the as-deposited highly stable glass, and FC refers to the glass obtained after refreezing the supercooled liquid at a rate of ∼700 K/s.

Figure 2. (a) Enthalpy plot of AD (blue) and FC (red) samples. The dashed line indicates the supercooled liquid. Vertical arrows show the onset and fictive temperatures of AD and FC glasses. The onset temperature is determined at a heating rate of 35000 K/s. Extrapolation to lower heating rates, that is, 10 K/min (see Supporting Information) locates Ton(FC) and Ton(AD) at 115 and 125 K, respectively. Arrows schematically indicate the enthalpy path during annealing at 128 K. (b) Specific heat of AD glass (blue circles) and the glass obtained after refreezing the sample annealed for 1800 s at 128 K (red squares).

generated heat capacity curves using fast scans (Figure 1a−c). See the Supporting Information for more details. There are several remarkable characteristics of these heat capacity curves. (1) After annealing for a sufficient time at temperatures above Tg, the calorimetric traces show a double overshoot. This is reminiscent of the glassy state formed by a mixture of two amorphous solids with different onsets and even fictive temperatures. (2) There is a clear evolution of the relative area of both peaks, starting from the initial highly stable glass and progressing toward a less stable glass, with the annealing time at each temperature. (3) Interestingly, annealing at 122 K for 36600 s (Figure 1a) produces a glass (pentagon symbol and blue line) that is more stable (higher overshoot and higher onset temperature) than the glass obtained by fast cooling of the supercooled liquid (× symbol and violet line, referred to hereafter as FC glass). Similarly, annealing at 124 K for 7200 s (Figure 1b) produces a glass (pentagonal symbol and blue line) with enhanced stability with respect to the glass obtained by fast cooling of the supercooled liquid. Annealing at 128 K for 1800 s (Figure 1c) produces a glass that is calorimetrically very similar to our standard FC glass.

We start by discussing point (1) above in connection with the transformation kinetics of the as-deposited (AD) glass samples when annealed at various temperatures above Tg. In the nanocalorimetric experiments, we measured the heat rate needed to increase the enthalpy as the temperature increased through the glass transition temperature range. Figure 2a shows a schematic of the enthalpy path followed by the transformation process for a stable glass annealed at 128 K. The dotted line corresponds to the equilibrated supercooled liquid of Yamamuro et al.20 extrapolated to below Tg. The fictive and onset temperatures of the AD and FC glasses are also marked in the figure. Arrows mark the enthalpy path of the AD sample as it is transformed upon heating at 128 K, together with cooling at ∼700 K/s to 90 K. The cycle can be detailed as follows: After growth at a deposition temperature of Tdep = 93 K, the sample is immediately (within 1s) heated from Tdep to TANN. The enthalpy follows the initial path marked as 1 in the figure. If this trajectory (path 1) is followed to the higher temperatures, the total measured heat capacity will result in the blue curve of Figure 2b. At TANN = 128 K (>Tg), the sample transforms into a supercooled liquid (path 2). If the waiting 920

dx.doi.org/10.1021/jz201681v | J. Phys. Chem. Lett. 2012, 3, 919−923

The Journal of Physical Chemistry Letters

Letter

corresponds to the as-yet untransformed highly stable vapordeposited glass. We evaluated the transformation of the AD samples when annealed at various temperatures above Tg. The transformed fraction at various temperatures and times is shown in Figure 4.

time exceeds the time required for equilibration, the sample will completely transform into the equilibrated supercooled liquid. Upon passive cooling at ∼700 K/s, the sample follows the supercooled liquid (dashed line) until it crosses the glass line and freezes into a less-stable glass that is characteristic of the previous thermal history, that is, the cooling rate (path 4). Further fast heating of this sample to high temperatures results in the heat capacity curve (red squares) of Figure 2b. Careful examination of the results presented in Figure 1 reveals that for sufficiently long annealing times, the AD sample transforms completely into a less-stable glass through the intervention of the supercooled liquid. While Figure 2 schematically shows the expected behavior upon annealing a glassy sample above Tg, the data in Figure 1 illustrate the intriguing behavior stated above in point 3, that is, the final lessstable glass produced by annealing the vapor-deposited glass above Tg depends on the annealing temperature. Figure 3

Figure 4. Transformed fraction versus time of the AD glass into the partially equilibrated supercooled liquid (filled symbols) at the three annealing temperatures. The full line represents the fit with the kinetic equation, whereas the dashed line corresponds to a linear growth model. The inset shows the values of v(T)/S and the viscosity of the supercooled liquid versus temperature.

This is estimated from multiple fits similar to the ones shown in the Supporting Information. For sufficiently long annealing times, the AD glass completely changes into a less-stable glass through the intervention of the isothermal glass → liquid transformation and the subsequent quenching. The experimental data are tentatively fitted with the kinetic equation x = 1 − exp[−K(T)t], where x is the AD-transformed fraction and K(T) = K0 exp(−E/RT), and with a nearly linear transformation equation x = (v(T)/S )t, where v is the growth rate and S may be interpreted to be the film thickness in a surfaceinitiated growth process. Values K0 = (0.5−30) × 1043 s−1 and E = (110 ± 2) kJ/mol are found for the transformation of the AD glass into the less-stable glass. The corresponding fitted curves are shown in Figure 4 as full lines. Slight variations in the fitting procedure do not result in significant changes to the value of the activation energy; this is reflected in the low associated uncertainty. The values of v(T)/S versus the annealing temperature are shown in the inset of Figure 4, and the dashed lines in that figure show the corresponding transformation curves. The onset of the transformation for samples annealed at 128 and 124 K is compatible with the growth-front-type mechanism postulated recently by Ediger et al.15 However, the establishment of this transformation mechanism is delayed for samples annealed at 122 K. In agreement with previous findings,15,16 the temperature dependence of the growth velocity is similar to that of the inverse of the viscosity in the equilibrium supercooled liquid (see the inset in Figure 4a). Both tentative fittings are clearly connected because for K(T)t ≪ 1, the equation x = 1 − exp[−K(T)t] becomes x = K(T)t, that is, K(T) ≈ v(T)/S . Nevertheless, the slowing of the transformation at long annealing times still requires elucidation. We now compare the transformation times with the structural relaxation times for supercooled toluene. As shown previously, there is a dramatic difference between a conven-

Figure 3. (a) Specific heat versus temperature after long annealing of AD glasses at Tmax.

compares the calorimetric traces from the FC and the longannealed AD glasses. Clearly, at TANN = 128 K, the molten phase may be considered an equilibrated supercooled liquid; however, at TANN = 122−124 K, it should be considered a partially equilibrated liquid. That is, the glassy state obtained by refreezing the liquid depends markedly on the relaxational properties of the liquid at the annealing temperature. The transformation of the high-density glass (equivalent to our AD state) of ethylbenzene (EB) and IPB has been reported by Ishi and co-workers.14,21 Their results suggest that multiple nonequilibrium supercooled liquid states can be accessed through transformation of the high-density vapor-deposited glass. We have verified that the change in the onset temperature of the refrozen glasses as a function of annealing temperature is not due to the effect of the cooling rate variation in the sample. See the Supporting Information for more details. Thus, the double overshoot observed in the calorimetric traces in Figure 1 is interpreted to indicate a sample with two glasses of different kinetic stability, a remnant AD glass and a (less-stable) glass fast cooled from the supercooled liquid. This behavior is very unusual, although it bears some resemblance to recent findings by Ediger et al.4 In their work, several heat capacity peaks appeared in ultrastable as-grown vapor-deposited glassy films, which were attributed to different packing arrangements within the stable glass. In contrast, our glass recovered by partial annealing is a mixture of two parts; the first has been transformed into a supercooled liquid before being refrozen to a low temperature to form the less-stable glass, and the second 921

dx.doi.org/10.1021/jz201681v | J. Phys. Chem. Lett. 2012, 3, 919−923

The Journal of Physical Chemistry Letters

Letter

In summary, the presence of two well-defined heat capacity overshoots after the partial annealing of highly stable glasses above Tg indicates the presence of two distinct glassy states in the same sample. The relative change of the peak area is used to determine the kinetics of the transformation from the highly stable glass to the supercooled liquid. It is found that the ultrastable glass has a much slower transformation time compared to the structural relaxation time of the supercooled liquid. The transformation follows an Avrami law with the n exponent close to 1. The new glass formed during annealing above Tg and the subsequent cooling of the supercooled liquid retains memory of the annealing temperature at which it was created, implying access to several distinctive nonequilibrium supercooled liquids at different temperatures. Our results are compatible with a scenario of a LL transition occurring between the highly stable vapor-deposited glass and the supercooled LI.

tional glass cooled from a liquid at 10 K/min and the ultrastable glass grown from a vapor at 0.80Tg. This difference appears in the remarkable variation of relaxation times compared to data obtained by various spectroscopies.22 The time needed to transform glass at temperatures above Tg (Figure 4) is much larger than that typically observed in conventional glasses obtained from a liquid. Similar behavior was previously reported by Ediger and co-workers15, who showed that the time needed to transform an ultrastable film of IMC was dependent on the film thickness for thicknesses below 1 μm and was also higher than the structural relaxation time. In Table 1, we compare the time required to transform 50% of the Table 1. Values of the Times Required to Transform 50% of the Ultrastable Glass at the Different Temperatures and Comparison to the Data of Hinze et al. (ref 23) temperature (K) 122 124 128

τ (s) time to 50% liquid 2700 450 15

τα (s) Hinze et al.23 −2

3.3 × 10 4.5 × 10−3 1.6 × 10−4

■

EXPERIMENTAL SECTION Materials. Toluene (purity 99.5%) was purchased from Sigma Aldrich and placed in a Pyrex container within a high-vacuum prechamber. Final purification was achieved by several sequences of freeze−pump−thaw cycles and distillation with molecular sieves prior to evaporation into the chamber. Growth Process. Samples were deposited onto the SiNx membrane of the nanocalorimeter chip. The transfer to the UHV chamber was performed by carefully opening a highprecision leak valve connecting the prechamber with the main chamber. The vapor was injected through a microcapillary glass connected to the leak valve and in close proximity to the nanocalorimetric chip. This system allows the deposition of organic films at very slow growth rates on cold substrates. A more detailed explanation of this configuration has been previously reported.18,19,26 Films of toluene are grown from the vapor at a temperature of 0.80Tg, that is, 93 K, and a growth rate of 0.10 ± 0.015 nm/s. These conditions have been previously shown to produce very stable glasses with enhanced onset and reduced fictive temperatures, which indicate higher kinetic and thermodynamic stability compared to glasses obtained from the liquid.19 The thickness of all of the glassy films analyzed in this report is around 100 nm, which enables size effects to be neglected because they behave calorimetrically as bulk glasses.26 Nanocalorimetry. The heat capacity curve of the various samples is measured by raising the temperature at very fast rates, ∼35000 K/s, using the methodology developed by Allen et al.27−29 (see Supporting Information for details). Thermal Treatments. The thermal history of the samples undergoing the transformation is (i) growth from the vapor at 93 K, (ii) fast jump to TANN = 122, 124, or 128 K (Tg + 5, Tg + 8, and Tg + 11 K) above the nominal Tg = 117 K of conventional glass20,30,31 and annealing for times ranging from 1 × 104 to 3.6 × 104 s, (iii) cooling to 90 K by passive cooling, that is, ∼700 K/s, (iv) fast scanning to record the calorimetric trace in the glass transition region, that is, the calorimetric traces reflect the state of the glass after partial transformation at TANN and fast cooling to 90 K.

τ/τα 8 × 10 105 9 × 104 4

vapor-deposited sample into the supercooled liquid with the relaxation times obtained by deuteron spin−lattice relaxation by Hinze et al.23 In all cases, τ ≈ 9 ± 1 × 104 τα. The above results show strong evidence that the transformation from the highly stable glass to the supercooled liquid in toluene is heterogeneous and proceeds through a nucleation and growth (NG) mechanism. This is the first time that such results are reported in low Tg materials and extends the possibility of growth-front transformation as a general characteristic of these highly stable materials. Two coexisting distinct glasses are identified from the results of Figure 1. Do they represent the existence of two underlying distinct liquid phases in toluene? A first-order transition between the low-temperature liquid with a low concentration of excitations and the high-temperature liquid with a high concentration of excitations at a temperature of TLL = 113 K has been proposed for toluene by Matyushov and Angell.24,25 In our case, the low-temperature liquid has become an amorphous solid, the highly stable AD glass, whereas the high-temperature liquid is the normal supercooled liquid. In this view, for toluene (a fragile liquid), the high entropy of the supercooled liquid in the range of Tg ≤ T ≤ Tm would be finally lost in a first-order transition to a low-entropy liquid state at the temperature TLL, which, according to fits to experimental data, lies between the Kauzmann temperature TK and Tg.24 The transformation observed for toluene at 128 K bears similarities with the LL transition in TPP. In TPP, an isothermal treatment of supercooled liquid I (LI, the low-density phase) between 215 and 230 K produces by NG liquid II (LII, the high-density liquid) in the form of the glacial phase, glass II. In our case, the isothermal treatment at T = 128 K, above Tg of the AD glass (Tg ≈ 125 at 10 K/min; see Supporting Information), transforms glass II (the highly stable, high-density AD phase) by a NG mechanism into LI (the low-density liquid associated with the FC glass). This could be an indication that the highly stable glassy phase, AD, should be associated with a new highdensity liquid state, LII. Unfortunately, the presence of LII cannot be directly inferred from our heat capacity measurements.

■

ASSOCIATED CONTENT

S Supporting Information *

Description of the methodology employed in the nanocalorimetric scans and the fitting procedure used to extract the kinetics of the transformation from heat capacity curves. Cooling rate variation of the onset and fictive temperatures. 922

dx.doi.org/10.1021/jz201681v | J. Phys. Chem. Lett. 2012, 3, 919−923

The Journal of Physical Chemistry Letters

Letter

(17) Kearns, K. L.; Whitaker, K. R.; Ediger, M. D.; Huth, H.; Schick, C. Observation of Low Heat Capacities for Vapor-Deposited Glasses of Indomethacin as Determined by AC Nanocalorimetry. J. Chem. Phys. 2010, 133, 014702. (18) Leon Gutierrez, E.; Garcia, G.; Lopeandia, A. F.; Fraxedas, J.; Clavaguera-Mora, M. T. In Situ Nanocalorimetry of Thin Glassy Organic Films. J. Chem. Phys. 2008, 129, 181101. (19) Leon-Gutierrez, E.; Sepulveda, A.; Garcia, G.; Teresa Clavaguera-Mora, M.; Rodriguez-Viejo, J. Stability of Thin Film Glasses of Toluene and Ethylbenzene Formed by Vapor Deposition: An In Situ Nanocalorimetric Study. Phys. Chem. Chem. Phys. 2010, 12, 14693−14698. (20) Yamamuro, O.; Tsukushi, I.; Lindqvist, A.; Takahara, S.; Ishikawa, M.; Matsuo, T. Calorimetric Study of Glassy and Liquid Toluene and Ethylbenzene: Thermodynamic Approach to Spatial Heterogeneity in Glass-Forming Molecular Liquids. J. Phys. Chem. B 1998, 102, 1605−1609. (21) Ishii, K.; Yokoyama, Y.; Moriyama, R.; Nakayama, H. Liquid− Liquid Relaxation in the Supercooled Liquid State of Ethylbenzene: Thermal Studies Using a Prototype DTA Sensor for the Study of Vapor-Deposited Samples. Chem. Lett. 2010, 39, 958−960. (22) Leon-Gutierrez, E.; Garcia, G.; Clavaguera-Mora, M. T.; Rodriguez-Viejo, J. Glass Transition in Vapor Deposited Thin films of Toluene. Thermochim. Acta 2009, 492, 51−54. (23) Hinze, G.; Sillescu, H. Nuclear Magnetic Resonance Study of Supercooled Toluene: Slow and Fast Processes above and below the Glass Transition. J. Chem. Phys. 1996, 104, 314−319. (24) Matyushov, D. V.; Angell, C. A. Two-Gaussian Excitations Model for the Glass Transition. J. Chem. Phys. 2005, 123, 034506. (25) Matyushov, D. V.; Angell, C. A. Gaussian Excitations Model for Glass-Former Dynamics and Thermodynamics. J. Chem. Phys. 2007, 126, 094501. (26) Leon-Gutierrez, E.; Garcia, G.; Lopeandia, A. F.; ClavagueraMora, M. T.; Rodriguez-Viejo, J. Size Effects and Extraordinary Stability of Ultrathin Vapor Deposited Glassy Films of Toluene. J. Phys. Chem. Lett. 2010, 1, 341−345. (27) Efremov, M. Y.; Olson, E. A.; Zhang, M.; Lai, S. L.; Schiettekatte, F.; Zhang, Z. S.; Allen, L. H. Thin-Film Differential Scanning Nanocalorimetry: Heat Capacity Analysis. Thermochim. Acta 2004, 412, 13−23. (28) Efremov, M. Y.; Olson, E. A.; Zhang, M.; Schiettekatte, F.; Zhang, Z. S.; Allen, L. H. Ultrasensitive, Fast, Thin-Film Differential Scanning Calorimeter. Rev. Sci. Instrum. 2004, 75, 179−191. (29) Efremov, M. Y.; Olson, E. A.; Zhang, M.; Zhang, Z. S.; Allen, L. H. Probing Glass Transition of Ultrathin Polymer Films at a Time Scale of Seconds Using Fast Differential Scanning Calorimetry. Macromolecules 2004, 37, 4607−4616. (30) Alba, C.; Busse, L. E.; List, D. J.; Angell, C. A. Thermodynamic Aspects of the Vitrification of Toluene, and Xylene Isomers, and the Fragility of Liquid Hydrocarbons. J. Chem. Phys. 1990, 92, 617−624. (31) Sepulveda, A.; Leon-Gutierrez, E.; Gonzalez-Silveira, M.; Rodriguez-Tinoco, C.; Clavaguera-Mora, M. T.; Rodriguez-Viejo, J. Accelerated Aging in Ultrathin Films of a Molecular Glass Former. Phys. Rev. Lett. 2011, 107, 025901.

Heating rate variation of the onset temperature of the glass transition. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work is supported by MICINN MAT2010-15202 and by the Generalitat de Catalunya through SGR2009-01225. A.S. thanks CONACYT for a predoctoral fellowship. We thank F. J. Muñoz Pascual for support in the microfabrication of the calorimetric chips.

■

REFERENCES

(1) Goldstein, M. Viscous Liquids and Glass Transition  A Potential Energy Barrier Picture. J. Chem. Phys. 1969, 51, 3728−&. (2) Jackle, J. Models of the Glass-Transition. Rep. Prog. Phys. 1986, 49, 171−231. (3) Angell, C. A.; Ngai, K. L.; McKenna, G. B.; McMillan, P. F.; Martin, S. W. Relaxation in Glassforming Liquids and Amorphous Solids. J. Appl. Phys. 2000, 88, 3113−3157. (4) Kearns, K. L.; Swallen, S. F.; Ediger, M. D.; Sun, Y.; Yu, L. Calorimetric Evidence for Two Distinct Molecular Packing Arrangements in Stable Glasses of Indomethacin. J. Phys. Chem. B 2009, 113, 1579−1586. (5) Dawson, K. J.; Zhu, L.; Yu, L.; Ediger, M. D. Anisotropic Structure and Transformation Kinetics of Vapor-Deposited Indomethacin Glasses. J. Phys. Chem. B 2011, 115, 455−463. (6) Mishima, O.; Stanley, H. E. Decompression-Induced Melting of Ice IV and the Liquid−Liquid Transition in Water. Nature 1998, 392, 164−168. (7) Wilding, M. C.; McMillan, P. F. Polyamorphic Transitions in Yttria-Alumina Liquids. J. Non-Cryst. Solids 2001, 293, 357−365. (8) Senker, J.; Rossler, E. Triphenyl Phosphite: A Candidate for Liquid Polyamorphism. Chem. Geol. 2001, 174, 143−156. (9) Deb, S. K.; Wilding, M.; Somayazulu, M.; McMillan, P. F. Pressure-Induced Amorphization and an Amorphous−Amorphous Transition in Densified Porous Silicon. Nature 2001, 414, 528−530. (10) Cohen, I.; Ha, A.; Zhao, X. L.; Lee, M.; Fischer, T.; Strouse, M. J.; Kivelson, D. A Low-Temperature Amorphous Phase in a Fragile Glass-Forming Substance. J. Phys. Chem. 1996, 100, 8518−8526. (11) Tanaka, H.; Kurita, R.; Mataki, H. Liquid−Liquid Transition in the Molecular Liquid Triphenyl Phosphite. Phys. Rev. Lett. 2004, 92, 025701. (12) Hedoux, A.; Hernandez, O.; Lefebvre, J.; Guinet, Y.; Descamps, M. Mesoscopic Description of the Glacial State in Triphenyl Phosphite from an X-ray Diffraction Experiment. Phys. Rev. B 1999, 60, 9390− 9395. (13) Hassaine, M.; Jimenez-Rioboo, R. J.; Sharapova, I. V.; Korolyuk, O. A.; Krivchikov, A. I.; Ramos, M. A. Thermal Properties and Brillouin-Scattering Study of Glass, Crystal, and “Glacial” States in nButanol. J. Chem. Phys. 2009, 131, 174508. (14) Ishii, K.; Nakayama, H.; Moriyama, R. Nonequilibrium and Relaxation in Deeply Supercooled Liquid of Isopropylbenzene Obtained through Glass Transition from Vapor-Deposited Glass. J. Phys. Chem. B 2012, 116, 935−942. (15) Swallen, S. F.; Windsor, K.; McMahon, R. J.; Ediger, M. D.; Mates, T. E. Transformation of Stable Glasses into Supercooled Liquids: Growth Fronts and Anomalously Fast Liquid Diffusion. J. Phys. Chem. B 2010, 114, 2635−2643. (16) Kearns, K. L.; Ediger, M. D.; Huth, H.; Schick, C. One Micrometer Length Scale Controls Kinetic Stability of Low-Energy Glasses. J. Phys. Chem. Lett. 2010, 1, 388−392. 923

dx.doi.org/10.1021/jz201681v | J. Phys. Chem. Lett. 2012, 3, 919−923