10734
J. Phys. Chem. B 2010, 114, 10734–10739
Roles of Individual and Cooperative Motions of Molecules in Glass-Liquid Transition and Crystallization of Toluene Ryutaro Souda* International Center for Materials Nanoarchitectonics, National Institute for Materials Science, 1-1 Namiki, Tsukuba, Ibaraki 305-0044, Japan ReceiVed: May 18, 2010; ReVised Manuscript ReceiVed: July 21, 2010
Deeply supercooled fragile liquid is known to be dynamically heterogeneous, where super Arrhenius behavior of shear viscosity and R and β relaxation processes have been observed. To clarify origins of these behaviors, we have investigated correlations between microscopic molecular diffusion and macroscopic hydrodynamics of vapor-deposited toluene films by using time-of-flight secondary ion mass spectrometry. The molecules are intermixed gradually at around 70 K on the film deposited at 15 K, which is followed by an abrupt film morphology change at around the calorimetric glass-transition temperature (Tg) of 117 K. For the film deposited at 100 K, intermixing of the molecules occurs at temperatures over 100-130 K, but no abrupt increase in diffusivity is recognizable at Tg. This result can be explained as dynamical heterogeneity or decoupling between translational diffusion and viscosity. The self-diffusion is thought to occur in a sub-Tg region as a precursor state of the glass-liquid transition; that is, individual motion in the solid-like state evolves into cooperative motion of liquid-like state at Tg. The supercooled liquid nucleates at 147 K, although the crystal can grow even at 117 K provided that nuclei preexist. The origins of unusual glass transition behaviors of amorphous solid water are also discussed in comparison with this standard. 1. Introduction The glass-liquid transition has recently been an area of intense experimental and theoretical interest. The molecular dynamics in glass-forming liquids slows dramatically as the temperature is lowered toward Tg, but its origin remains a major challenge of modern physical chemistry. For understanding general features of glass transition phenomena, it is useful to classify glass-forming liquids using a “strong-fragile” concept.1 The covalently bonded network oxides, such as SiO2 and GeO2, are prototypes of strong liquids; their viscosity behaves in a nearly Arrhenius fashion and their heat capacity changes very little across Tg. By contrast, organic materials, such as orthoterphenyl (OTP), salol, and toluene, are typical examples of fragile liquids that are characterized by a strong non-Arrhenius behavior in viscosity and a substantial drop in heat capacity at Tg. The dynamics of a supercooled fragile liquid is known to be spatially heterogeneous,2 as manifested by decoupling of translational diffusion and viscosity in deeply supercooled regions, T < 1.2Tg. Decoupling is related to the sequential decay of two types of particle motion: R relaxation exhibits nonArrhenius behavior before it disappears at Tg and secondary β relaxation (the Johari-Goldstein process3) displays Arrhenius behavior and continues even below Tg. The microscopic origins of these two relaxation processes are not fully established. Experimentally, Tg has been determined from the onset temperature of the heat capacity jump in differential scanning calorimetry (DSC). Another definition of Tg is provided by the temperature at which shear viscosity reaches 1013 poise or the dielectric relaxation time becomes 100 s. The values of Tg obtained using these methods are generally comparable among the majority of glass-forming liquids. However, difficulties are often encountered in assignments of calorimetric Tg of poorly glass-forming materials. A typical example is amorphous solid * E-mail:
[email protected].
Figure 1. Schematic views of DSC scans. No appreciable glasstransition endotherm is observed for water (a), although good glass formers like toluene (b) exhibit glass-transition endotherm (Tg) prior to crystallization exotherm (Tc). Because of decoupling, the translational molecular diffusion might occur at Td prior to the glass transition at Tg, but the former is difficult to be identified using DSC.
water (ASW)4 that exhibits no appreciable endotherm during a heating scan prior to the crystallization temperature, Tc, as schematically shown in Figure 1a, providing a striking contrast to good glass formers like OTP and toluene (Figure 1b). All these materials are fragile liquids that are expected to exhibit a
10.1021/jp104520f 2010 American Chemical Society Published on Web 08/02/2010
Motions of Molecules in Glass-Liquid Transition large heat capacity jump at Tg. The unusual behavior of water might be explained as a specific “fragile-to-strong transition” in supercooled water near 228 K,5 although its existence appears to be incompatible with the result of diffusivity measurements.6 An alternative possibility is that water’s Tg is higher than Tc.7,8 Crystallization might be induced by self-diffusion of molecules at temperatures even below Tg because of decoupling. Therefore, it is quite important to determine the onset temperature of selfdiffusion of molecules, Td, and reveal its roles in the glass-liquid transition and crystallization. So far, we have discussed the nature of the glass-liquid transition of water using time-offlight secondary ion mass spectrometry (TOF-SIMS).9,10 The advantage of TOF-SIMS relative to the calorimetric study is that it can reveal both microscopic molecular diffusion (at Td) and macroscopic hydrodynamics (Tg). In this paper, TOF-SIMS is used for the analysis of the glass-liquid transition dynamics of vapor-deposited toluene films. Toluene is one of the typical fragile liquids, and its Tg has been determined calorimetrically.11,12 Both R and β relaxation times of toluene have also been measured by dielectric spectroscopy.13 The β-process is intrinsic to a solid-like state, originating from restricted reorientation of molecules in a cage, whereas the R-process represents relaxation of the cage itself. By using TOF-SIMS, the cage collapse can be identified straightforwardly via the morphological change of thin films. From the comparison with toluene, the origin of the unusual glass-transition behaviors of water is highlighted. Attention is focused on the kinetic and thermodynamic aspects of the glass-liquid transition. The roles of self-diffusion and hydrodynamic flow in nucleation and growth of crystals are also discussed. 2. Experimental Section Experiments were performed in an ultrahigh vacuum (UHV) chamber with base pressure of less than 1 × 10-8 Pa. For the TOF-SIMS measurements, a He+ beam with kinetic energy of 2.0 keV was generated in a differentially pumped electronimpact-type ion source and was chopped into pulses of 40 ns in width and 15 kHz in repetition rate using electrostatic deflection plates. The pulsed He+ beam (ca. 5 pA; 4 mm in spot size) was incident onto a sample surface that was floated with a bias voltage of +500 V. Positive secondary ions were extracted from the surface by a grounded stainless steel mesh placed immediately in front of the sample. They were pulsecounted using a channel plate after traveling through a fieldfree linear TOF tube. The TOF-SIMS spectrum was created using a multichannel scaler. The substrate was a Ni(111) surface that was spot-welded to a Ta holder using Ni strips. The Ni(111) substrate was heated to ca. 1200 K by electron bombardment from behind. After the sample was cleaned via several cycles of heating and sputtering, it was cooled to 15 K by means of a closed-cycle helium refrigerator. The temperature was controlled using a digital temperature controller by monitoring temperature of a coldfinger close to the sample position using Au(Fe)-chromel thermocouples. The toluene vapor was admitted into the UHV chamber through high-precision leak valves; thin films were deposited onto the cooled Ni(111) surface by backfilling the UHV chamber. The toluene’s coverage was estimated from the evolutions of secondary ion intensities as a function of exposure. According to the literature,14-17 crystallization of the glassy film can be identified from a change in the desorption rate of molecules. To determine Tc, therefore, temperature programmed desorption (TPD) spectra of toluene were taken using a differentially pumped quadrupole mass analyzer. The TOFSIMS and TPD spectra were recorded at the same ramping speed
J. Phys. Chem. B, Vol. 114, No. 33, 2010 10735
Figure 2. (a) Typical TOF-SIMS spectra from the toluene-deposited Ni(111) surface. The results between 1- and 10-ML toluene films are compared. (b) Evolutions of typical secondary ion intensities as a function of coverage of toluene deposited on the Ni(111) surface at 15 K.
of 5 K min-1. All TOF-SIMS spectra were taken with fluence less than 1012 ions/cm2 to avoid damaging the sample surface. 3. Experimental Results Figure 2a shows TOF-SIMS spectra from the toluenedeposited Ni(111) surface at 15 K; the toluene films with coverage of 1 and 10 monolayers (ML) are compared. For the 1-ML film, the Ni+ and its adducts, Ni+(C7H8)n, are sputtered intensively from the surface, in addition to the fragment ions from toluene, such as H+, C2H3+, C3H3+, and C7Hx+ (x ) 7, 8). The former (the latter) disappears completely (decreases considerably) for the 10-ML film. This tendency is clearly recognizable in Figure 2b, where the intensities of typical secondary ions are plotted as a function of coverage. The Ni+ intensity decreases almost exponentially following a peak at small coverage. Such is observed commonly for molecular-solid films deposited on the Ni(111) substrate.10,18 The Ni+ intensity tends to peak when adspecies form a monolayer. Therefore, the film thickness was estimated from the exposure relative to the peak position of the Ni+ intensity. The fragment ion also peaks at around 1 ML, whereas the H+ intensity increases monotonically. The enhancement of the C7Hx+ intensity is noticeable at small coverage. This phenomenon might be explained as the gas-phase dissociation of the sputtered Ni+(C7H8)n ions, as inferred from quite similar evolutions of the Ni+ and C7Hx+ ions for thin films ( 117 K). An aggregate of fine crystallites grows at around 110 K on the surface of crystalline toluene that was formed accidentally during the quenching process. This phenomenon has been explained in terms of molecular rearrangement in the β-process because the formation of a crystal embryo requires the local structural fluctuation. In these studies, it is assumed that molecular diffusion is fundamentally quenched in the glassy state, but crystallization might be induced by uncorrelated motions of individual molecules in the sub-Tg region. Our seeding experiment (Figure 5b) showed no indication that the crystal grows at T < Tg, but it might occur in a longer time scale (hours or days). In any case, it is natural to consider that the cooperatively moving molecules formed at T > Tg play a more important role in crystallization because local structural transformation of the molecules on the crystal surface is facilitated. In contrast to toluene, the determination of water’s calorimetric Tg is controversial. Johari et al.4 observed an extremely small endotherm at 136 K in DSC scans only after special annealing procedures and assigned it as water’s Tg. However, this assignment was claimed by Angell and co-workers7,8 because the observed endtherm is only 1/14th of the expected value. They concluded that ASW remains a glass up until crystallization.39 On the other hand, the glass-transition behaviors of ASW have been observed clearly using TOF-SIMS.9,10 The self-diffusion of the water molecules commences at Td ) 130-140 K, which is followed by the film morphology change at 160-165 K. The evolution of supercooled liquid (Tg) is not observed separately from the growth of crystal grains (Tc), as far as the film dewetting behaviors are concerned. The TOFSIMS result of ASW is quite similar to that of seeded toluene, as shown in Figure 5b. Therefore, the experimental results of ASW can be interpreted as that nucleation and growth occur immediately after supercooled liquid water evolves (Tg ) Tc). This behavior is compatible with a poor glass-forming nature of supercooled liquid water that nucleates spontaneously at T < 235 K. In this context, therefore, identification of the glasstransition endotherm using DSC is fundamentally impossible because it is hindered by the huge crystallization exotherm. The relatively weak signal of the β-process is also very difficult to observe using DSC, but what was observed by Johari et al. after the special annealing4 might be considered as the β-process. To date, water’s Tg has been assigned to 136 K from their observation, but it must be Td rather than Tg, as revealed from the comparison of the TOF-SIMS and DSC results between water and toluene. Angell and co-workers5 suggested that the unusually small glass-transition endotherm of water is a characteristic of strong liquid. According to this idea, the glass-liquid transition itself can be considered as a “strongto-fragile liquid transition”: strong liquid evolves at Td < T < Tg for both toluene and water prior to the evolution of fragile liquid (i.e., supercooled liquid). The formation of supercooled liquid water at 160-165 K has been inferred from the drastic change in hydration properties.40 The glass-transition kinetics of water is more serious than toluene; the isothermal TOF-SIMS measurement revealed that the morphology of the ASW film changes at Td ) 136 K after a much longer aging time (10-20 min).41 This phenomenon might be related to polyamorphism of water:42 ASW evolves into low density liquid (LDL) that is distinct from normal water or high density liquid (HDL). Therefore, strong liquid formed at Td < T < Tg can be regarded as LDL. The significant aging properties for water are ascribable to the phase transition from LDL to supercooled liquid. It is hypothesized that the polyamorphic phase transition occurs only
Motions of Molecules in Glass-Liquid Transition in the high pressure region,42 but these experimental results strongly suggest that liquid-liquid phase transition occurs even at ambient pressure immediately before crystallization. On the other hand, the properties of liquid-like water at higher temperatures have been investigated using nanoconfined water,43 as well as a water-ice mixture (i.e., premelting layer formed on crystallites).44 Because of dynamical heterogeneity, the Stokes-Einstein and Debye-Stokes-Einstein relations are violated at around 225 K, suggesting the coexistence of HDL and LDL for nanoconfined water43 and that of ice, HDL, and LDL for the supercooled water-ice mixture.44 The glass is not always regarded as a thermodynamic extension of the liquid as in the case of water, but the glass-liquid transition of toluene is by no means the phase transition. The difference in the glass-transition behaviors between water and toluene might be attributed to the nature of caging that is intimately related to the intermolecular interaction. Because of the nondirectional force, the toluene molecules can be caged without formation of any specific structures, thereby resulting in the convertibility between glass and liquid. In contrast, the water molecules are caged by directional hydrogen bonds; the tetrahedrally coordinated structure of glassy water is apparently distinct from more disordered structures of liquid water.45 The network oxides like SiO2 are fundamentally classified as strong liquids. They exhibit neither the liquid-liquid phase transition nor the strong-to-fragile liquid transition at ambient pressure because of the strong covalency in Si-O-Si bonds. Therefore, the liquid of network oxides can be regarded as an aggregate of strongly bonded domains although fluidized liquid, if any, might exist at the domain boundaries with a small amount, resulting in a relatively slow evolution of fluidity with increasing temperature. 5. Conclusion The microscopic origins of fragility and the R and β relaxations of deeply supercooled liquids were investigated by using TOF-SIMS. The toluene molecules move individually in the sub-Tg region (Td ≈ 100 K). The film deposited at 15 K is characterized by low density and exhibits lower Td (∼70 K). The structural relaxation is thought to be related to hopping of the molecule in a cage, so that the relaxation time (or experimental Td) is thought to be dependent on the cage structure (or density). The film morphology changes at Tg ) 117 K because supercooled liquid emerges, but no apparent change in diffusivity of the molecules is recognizable across Tg. This phenomenon is thought to be an experimental evidence of dynamical heterogeneity or decoupling between diffusivity and viscosity. The solid-like state in which the molecules can move individually is considered to be a precursor state for the glass-liquid transition, although it can be regarded as strong liquid if self-diffusion is considered to be a characteristic of liquid. Therefore, the glass-liquid transition is a two-step process: strong liquid is formed at Td and supercooled (or fragile) liquid is created at Tg. The glass-liquid transition of water is also a two-step process, but the creation of supercooled liquid water requires the polyamorphic phase transition of strong liquid (LDL). The calorimetric assignment of water’s Tg is fundamentally impossible because supercooled liquid water crystallizes immediately. The supercooled liquid toluene nucleates spontaneously at Tc ) 147 K, where the growth of crystal grains results in a more significant change in the film morphology than that observed at Tg. The crystal grows at Tg ) 117 K when nuclei
J. Phys. Chem. B, Vol. 114, No. 33, 2010 10739 preexist. The film morphology changes gradually in the sub-Tg region after some aging time because of the uncorrelated motion of molecules, which might also play a role in a slower crystal growth in the sub-Tg region. Both nucleation and growth require structural fluctuation of many molecules, so that they are induced fundamentally by cooperatively moving molecules emerging at T > T g. References and Notes (1) Angell, C. A. J. Non-Cryst. Solids 1991, 131-133, 13. (2) Debenedetti, P. G.; Stillinger, F. H. Nature (London) 2001, 410, 259. (3) Johari, G. P.; Goldstein, M. J. Chem. Phys. 1970, 53, 2372. (4) Johari, G. P.; Hallbrucker, A.; Mayer, E. Nature (London) 1987, 330, 552. (5) Ito, K.; Moynihan, C. T.; Angell, C. A. Nature (London) 1999, 398, 492. (6) Smith, R. S.; Kay, B. D. Nature (London) 1999, 398, 788. (7) Velikov, V.; Borick, S.; Angell, C. A. Science 2001, 294, 2335. (8) Yue, Y.; Angell, C. A. Nature (London) 2004, 427, 717. (9) Souda, R. Phys. ReV. Lett. 2004, 93, 235502. (10) Souda, R. Phys. ReV. B 2004, 70, 165412. (11) Alba, C.; Busse, L. E.; List, D. J.; Angell, C. A. J. Chem. Phys. 1990, 92, 617. (12) Yamamoto, O.; Tsukushi, I.; Lindqvist, A.; Takahara, S.; Ishikawa, M.; Matsuo, T. J. Phys. Chem. B 1998, 102, 1605. (13) Kudlik, A.; Tschirwitz, C.; Blochowicz, T.; Benkhof, S.; Rossler, E. J. Non-Cryst. Solids 1998, 235-237, 406. (14) Kouchi, A. Nature (London) 1987, 330, 550. (15) Lofgren, P.; Ahlstrom, P.; Chakarov, D. V.; Lausmaa, J.; Kasemo, B. Surf. Sci. 1996, 367, L19. (16) Smith, R. S.; Huang, C.; Wong, E. K. L.; Kay, B. D. Surf. Sci. 1996, 367, L13. (17) Dohnalek, Z.; Ciolli, R. L.; Kimmel, G. A.; Stevenson, K. P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 1999, 110, 5489. (18) Souda, R. J. Phys. Chem. B 2008, 112, 15349. (19) Swallen, S. F.; Kearns, K. L.; Mapes, M. K.; Kim, Y. S.; McMahon, R. J.; Ediger, M. D.; Wu, T.; Yu, L.; Satija, S. Science 2007, 315, 353. (20) Ishii, K.; Nakayama, H.; Hirabayashi, S.; Moriyama, R. Chem. Phys. Lett. 2008, 459, 109. (21) Souda, R., in preparation. (22) Adam, G.; Gibbs, J. H. J. Chem. Phys. 1965, 43, 139. (23) Donth, E. J. Non-Cryst. Solids 1982, 53, 325. (24) Stillinger, F. H. J. Chem. Phys. 1988, 89, 6461. (25) Kirkpatrick, T. R.; Thirumalai, D.; Wolynes, P. G. Phys. ReV. A 1989, 40, 1045. (26) Cohen, M. H.; Grest, G. S. Phys. ReV. B 1981, 24, 4091. (27) Hodgdon, J. A.; Stillinger, F. H. Phys. ReV. E 1993, 48, 207. (28) Stillinger, F. H.; Hodgdon, J. A. Phys. ReV. E 1994, 50, 2064. (29) Kob, W.; Donati, C.; Plimpton, S. J.; Poole, P. H.; Glotzer, S. C. Phys. ReV. Lett. 1997, 79, 2827. (30) Donati, C.; Douglas, J. F.; Kob, W.; Plimpton, S. J.; Poole, P. H.; Glotzer, S. C. Phys. ReV. Lett. 1998, 80, 2338. (31) Yamamoto, R.; Onuki, A. Phys. ReV. Lett. 1998, 81, 4915. (32) Fujara, F.; Geil, B.; Sillescu, H.; Fleischer, G. Z. Phys. B 1992, 88, 195. (33) Kind, R.; Liechti, O.; Korner, N.; Hulliger, J.; Donlinsek, J.; Blink, R. Phys. ReV. B 1992, 45, 7697. (34) Cicerone, M. T.; Ediger, M. D. J. Chem. Phys. 1996, 104, 7210. (35) Sillescu, H. J. Non-Cryst. Solids 1999, 243, 81. (36) Hatase, M.; Hanaya, M.; Hikima, T.; Oguni, M. J. Non-Cryst. Solids 2002, 307-310, 257. (37) Sun, Y.; Xi, H.; Ediger, M. D.; Yu, L. J. Phys. Chem. 2008, 112, 661. (38) Tanaka, H. Phys. ReV. E 2003, 68, 011505. (39) Minoguchi, A.; Richert, R.; Angell, A. C. Phys. ReV. Lett. 2004, 93, 215703. (40) Souda, R. J. Chem. Phys. 2007, 127, 214505. (41) Souda, R. Chem. Phys. Lett. 2005, 415, 146. (42) Mishima, O.; Stanley, H. E. Nature (London) 1998, 396, 329. (43) Chen, S.-H.; Mallamace, F.; Mou, C.-Y.; Broccio, M.; Corsaro, C.; Faraone, A.; Liu, L. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 12974. (44) Banerjee, D.; Bhat, S. N.; Bhat, S. V.; Leporini, D. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 11448. (45) Finney, J. L.; Hallbrucker, A.; Kohl, I.; Soper, A. K.; Bowron, D. T. Phys. ReV. Lett. 2002, 88, 225503.
JP104520F
