Devitrification Properties of Vapor-Deposited Ethylcyclohexane

Feb 18, 2015 - We constructed an adiabatic calorimeter adapted for the preparation and in situ thermal characterization of vapor-deposited glasses and...
0 downloads 16 Views 1MB Size
Article pubs.acs.org/JPCB

Devitrification Properties of Vapor-Deposited Ethylcyclohexane Glasses and Interpretation of the Molecular Mechanism for Formation of Vapor-Deposited Glasses Sergio Luis L. M. Ramos,*,† Atsuko. K. Chigira, and Masaharu Oguni Department of Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8551, Japan S Supporting Information *

ABSTRACT: We constructed an adiabatic calorimeter adapted for the preparation and in situ thermal characterization of vapor-deposited glasses and reported the investigation of the enthalpic states and dynamic properties of ethylcyclohexane (ECH) glasses prepared by vapor deposition in the temperature range of (0.71−0.96)Tg,liq; Tg,liq = (101 ± 1) K is the calorimetric glass transition temperature of the bulk liquid. It was verified that the ECH glasses deposited at temperatures immediately below Tg,liq were characterized by lower enthalpies and higher devitrification temperatures (Tdev), as compared to the glass obtained by supercooling the bulk liquid. The deposition temperature (TD) expected to yield experimentally the entity with the highest Tdev and the lowest enthalpic state was estimated to be 0.93Tg,liq. A model potentially elucidating the fundamental mechanism of formation and devitrification for the glasses prepared via the physical vapor deposition method as a function of TD was proposed. The fundamental point is that the glass is formed by deposition in a molecule-by-molecule fashion and the molecule deposited is frozen in a certain configuration determined by its being itself on the surface. For amorphous entities prepared at a TD much lower than Tg,liq, the surface molecule is frozen mostly as they are deposited. For the entity deposited at TD = 0.93Tg,liq in the case of ECH, the surface molecule is mobile immediately after the deposition to look for its stable configuration only on account of the intermolecular interactions with the molecules beneath and in the same surface layer as itself and freezes in a certain reasonably stable configuration; the molecules below the surface layer have already frozen in and get more stabilization energy through the additional interactions with the surface molecules. As a result, the intermolecular interaction of the molecules accumulated in such a way is stronger than that in the bulk liquid glass. It is argued that this is the fundamental reason why the glass formed immediately below Tg,liq has a lower enthalpy and a higher devitrification temperature than those of the liquid-cooled one.

1. INTRODUCTION

Recently, there is a growing interest in investigating and understanding the structural and dynamical properties surrounding glassy entities prepared via physical vapor deposition.3−19 Such interests have mainly been sparked given the results recently obtained through investigations on the thermal properties of these materials as a function of the vapor deposition conditions, such as deposition temperature TD and rate νD, on the thermal properties.8−15 It has been experimentally shown that a molecular glass prepared via the vapor deposition method can have higher kinetic stability and lower enthalpy, as compared with the corresponding glass prepared by the traditional liquid-cooling method,15 when TD is set at temperatures slightly below the glass transition temperature (Tg,liq) for the ordinary liquid-cooled (LC) glass. On the basis of these findings, there is an inevitable need to reconsider the molecular picture drawn previously for under-

Vitrification by vapor deposition constitutes a typical method for producing amorphous entities of fairly simple molecular substances. Yet, it fundamentally differs from the more traditional vitrification method of supercooling a liquid substance below its temperature Tfus of fusion, in view that it depicts a process in which a glassy structure is prepared in a molecule-by-molecule deposition fashion. Behind the idea of bringing molecules into the glassy state directly from the gaseous one, with skipping the liquid one, lie two fundamental notions of what the deposition process can achieve, (1) higher cooling rates to be actually attained in laboratory and (2) the avoidance of the liquid regime where unwanted crystallization could take place, thus, potentially allowing for the vitrification of yet unvitrified substances. The use of vapor deposition as a mean for vitrification or its attempt have been widely applied to substances that were once known to be more reluctant to glass formation. The method has allowed for the vitrification of, for example, water1 or propane,2 both of which are simple in their molecular structures. © 2015 American Chemical Society

Received: October 31, 2014 Revised: February 17, 2015 Published: February 18, 2015 4076

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B

apparatuses, (1) a standard adiabatic calorimeter previously reported22 (apparatus 1) and (2) the adiabatic calorimeter, of the filling-tube-type, described above and in the Supporting Information (apparatus 2). For the measurements in which apparatus 1 was employed, ECH with the mass of (11.766 ± 0.001) g was loaded in the calorimeter cell under an atmosphere of helium gas at ambient temperature, and the cell was hermetically sealed vacuum-tight with an indium gasket. Upon employing apparatus 2, the ECH vapor was deposited, at a constant average accumulation rate of (2 ± 0.5) nm s−1 (see the Supporting Information), into the calorimeter cell kept at the lower, rather constant temperatures to within ±0.1 K. The masses of vapor-deposited samples were determined by fitting the heat capacities, obtained in the crystalline state after the measurements in the glassy or supercooled liquid states, to the standard heat capacity values derived from the measurements with apparatus 1. Prior to its characterization, the crystal was thoroughly annealed at 140 K as to eliminate potential amorphous regions left. The final masses (ms) of the deposited ECH glasses were estimated to be 0.712, 0.721, 0.301, 0.569, and 0.162 g for the samples with the respective deposition temperatures TD of 97.0, 92.0, 87.0, 82.0, and 72.0 K. Upon utilizing these mass values, good agreement of the molar heat capacity values in the supercooled liquid region was automatically observed between measured samples and the standard data. The thermometry period for the heat capacity measurements of vapor-deposited samples was taken to be 24 min, while that for the measurements with apparatus 1 was 9 min. The reason is that rather a long time was required until the stainless steel pipes located between the filling tube and the calorimeter cell had established a quasi-equilibrium state after each temperature increment. Additionally, in the measurements using apparatus 1, fractional melting analysis was performed by experimentally determining Tobs, under strict thermodynamic equilibrium conditions, for relatively small enthalpy increments while maintaining the sample within the temperature interval of the fusion event. In this case, thermometry periods of over 1 h were used.

standing the similarity and dissimilarity between the properties of vapor-deposited and LC glasses. In an attempt to potentially clarify the universal features of these glasses, Chen and Richert19 investigated some of the physical properties of the supercooled liquid state for some organic compounds that have been reported to show kinetic enhancement when prepared by vapor deposition at temperatures immediately below Tg,liq. On the basis of their analyses, assumptions, and understanding, they concluded that compounds such as ethylcyclohexane (ECH) potentially form no glasses with the enhanced kinetic properties when vapordeposited. Nakayama et al.,5 on the other hand, estimated the temperature dependence of molar volumes from reflected light intensity measurements for vapor-deposited ECH. Opposing the view of Chen and Richert, their results and conclusion supported that the formation of glasses denser than the LC one does indeed occur for vapor-deposited ECH prepared at TD = (97.7−100.1) K. Considering the importance of identifying potential groups of compounds that cannot be kinetically enhanced by vapor deposition near Tg,liq, in this work, we investigated by means of adiabatic calorimetry the thermal properties of ECH prepared by vapor depositing at temperatures immediately below Tg,liq. In addition to this purpose, we also present the details of the construction of an adiabatic calorimeter adapted for the preparation and enthalpy measurements of in situ vapordeposited samples in the Supporting Information. The apparatus presented here is similar to those reported previously in the literature by Hikawa et al.20 and Sugisaski et al.21 Notably, their apparatuses were designed primarily aiming to circumvent sample crystallization through the preparation of vapor-deposited samples at relatively low deposition temperatures. The apparatus presented here, on the other hand, has been specifically designed to allow sample preparation (vapor deposition) at relatively high deposition temperatures and at desired rates under optimal temperature control and samplesteady intake conditions. Calorimetric data for vapor-deposited glasses of ethylbenzene (EB) prepared and measured with the present apparatus were previously reported elsewhere.11 In the present article, the heat capacities of the crystalline ECH were used to estimate the precision of the present calorimeter for convenience, which involves the need for reasonable vapor pressure for enabling the sample preparation via vapor deposition to occur.

3. RESULTS Figure 1 shows the molar heat capacities Cp,m, obtained with the ordinary adiabatic calorimeter reported previously (hereafter designated as apparatus 1),22 of ECH in the crystalline and liquid states at temperatures below 180 K. The fusion enthalpy of the crystal was determined to be ΔfusHm = (8.449 ± 0.016) kJ mol−1 as the average value from two distinct enthalpy measurements. The details of the procedure were similar to those reported for sec-butylcyclohexane in ref 23. The inset of Figure 1 shows the result of thermodynamic analysis, by the fractional fusion method, applied to the determination of the fusion temperature Tfus and sample purity; they were determined to be (161.94 ± 0.01) K and 99.92%, respectively. Figure 2 shows the imprecision of the heat capacities measured for the vapor-deposited samples with the abovestated calorimeter for the vapor-deposited sample (apparatus 2) as the deviation of the data from a smoothed polynomial fitting curve. The imprecision was estimated to be within ±0.3% above ∼75 K and within ±1.0% below ∼75 K. Figure 3 displays the temperature dependence of the rates of spontaneous heat release/absorption observed upon intermittently heating the ECH glasses prepared both by liquid cooling with apparatus 1 and by vapor deposition with apparatus 2.

2. EXPERIMENTAL SECTION 2.1. Construction of an Adiabatic Calorimeter Adapted for the Preparation and In Situ Thermal Characterization of Vapor-Deposited Glasses. Details are provided in the Supporting Information. The essential point of the difference between the newly constructed calorimeter and the similar one reported previously20 is that an exhaustion pipe as well as a sample vapor inlet pipe was installed as attached to the calorimeter cell lid. With the presence of this pipe, the vapor condensation could be carried out under a high-vacuum condition, leading to a rather constant condensation rate. 2.2. Sample Preparation and Heat Capacity Measurements. Commercial reagent ECH was purchased from SigmaAldrich Co. Ltd. with a minimum chemical assay of 99% and was further purified by fractional distillation under vacuum conditions prior to its use. Heat capacity measurements on ECH were carried out with two different high-precision 4077

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B

Figure 1. Molar heat capacities for crystalline and liquid ECH obtained with an ordinary adiabatic calorimeter (apparatus 1). The inset shows the equilibrium fusion temperature as a function of the reciprocal of the liquid fraction f within the sample. The solid line represents a linear fit to the experimental values; the fitting parameters are written within the inset. The intercept value β yielded the Tfus for the pure sample, and by means of the inclination value α, the molar fraction of impurity x was estimated as −(αΔfusHm)/(RTfus2), where R is the gas constant.

Figure 3. Temperature dependence of the spontaneous heat-release or -absorption rates observed in the intermittent heating processes under adiabatic conditions for the LC glass (a) and for the glasses (b−f) prepared by vapor deposition at different temperatures TD. In the case of (a), the filled and open circles represent the results for the rapidly or slowly precooled samples, respectively. The vertical dashed line marks the glass transition temperature observed for the LC glass, that is, Tg,liq.

Tg,liq for the glass prepared in such a manner was determined as (101 ± 1) K with applying an empirical relation that, when the relaxation time τ becomes 1 ks, the largest absolute value in the heat absorption effect is obtained for the slowly precooled sample and the change from the heat release to heat absorption effect occurs for the rapidly precooled sample by adiabatic calorimetry.22,25 Figure 3b−f gives the spontaneous heat release/absorption effects for the ECH glasses vapor-deposited at several distinct temperatures. These were mainly characterized by single endothermic peaks at slightly higher temperatures than Tg,liq. The temperature dependence shown in Figure 3b, nevertheless, revealed an exothermic effect at lower temperatures in addition to the endothermic one. These exothermic/endothermic effects characterize the structural changes during the devitrification process of the glass prepared at each TD. Notably, the sizes, shapes, and peak-top positions of the endothermic effects changed with TD. Taking the temperatures characterizing the peak tops as representative of the temperature of devitrification (hereafter designated as the devitrification temperature, Tdev), it is possible to characterize its variation as a function of TD. The temperature dependencies of the enthalpy variations for all glasses prepared are shown in Figure 4. The enthalpy curve (black circles and black line in the figure) for the LC glass was calculated from the data, obtained by using apparatus 1, for the slowly precooled sample by simple integration of Cp,m. The enthalpy changes for the other samples prepared by vapor deposition were evaluated as explained in the Supporting Information. The enthalpy curves are representative of the behaviors of the structural relaxation. Within lower temperatures than TD, all of the glasses exhibited enthalpy curves

Figure 2. Imprecision for heat capacity measurements employing the filling-tube-type calorimeter (apparatus 2) as estimated from crystallized samples vapor-deposited at TD = (97 ± 0.1) K. The overall imprecision was within ±1.0% in the range of 30 < T/K < 75 and within ±0.3% in 75 < T/K < 150. The different marks represent the results in different series of measurements.

These rates were derived, from the experimentally observed temperature drift rates, by subtracting the natural drift rates (see the Supporting Information). Open and filled circles in Figure 3a represent the results for the ECH samples precooled rapidly and slowly, respectively, from their supercooled liquid state at 110 K. Notably, whereas the rapidly precooled sample revealed a large heat release effect followed by a small heat absorption one, the slowly precooled one displayed only a large heat absorption effect. This hysteretic effect is indicative of a glass transition phenomenon.22,24 The transition temperature 4078

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B

Figure 4. Enthalpic paths followed by the differently vapor-deposited and LC ECH glasses; black open circles, slowly precooled LC glass; black filled circles, rapidly precooled LC glass; pink circles, TD = 72 K; blue circles, TD = 82 K; deep blue circles, TD = 87 K; green circles, TD = 92 K; orange circles, TD = 97 K glass. A dashed line represents the temperature dependence of the enthalpies for the supercooled liquid under equilibrium, as estimated from the linear extrapolation of the respective data. The short-dashed line depicts approximately the expected enthalpic path followed by the glass prepared at TD = 72 K in the absence of structural relaxation, that is, an exothermic effect.

Figure 5. Enthalpic states for ECH vapor-deposited glasses at their respective TD’s (open circles). The open double-triangle mark stands for the enthalpy of the LC glass at Tfic. The black thick line represents the equilibrium enthalpies as linearly extrapolated from the enthalpy data in the supercooled liquid temperature region, and the dashed one depicts the enthalpy variation for the LC glass. The inset shows the dependencies of Tfic (left ordinate; filled circles and filled doubletriangle mark) and Tg,dev (right ordinate; open circles and open double-triangle mark) as a function of TD or Tfic; Tfic, instead of TD, and the double-triangle marks were used to represent the behaviors for the LC glass, the arrows stand for indication of the respective ordinates expressing the data, and the solid, dashed, and short-dashed lines are linear guidelines for the eyes. The vertical dotted line marks the characteristic temperature 0.93Tg,liq, referred to within the text.

characterized with positive slopes, bearing some resemblance to that of the supercooled liquid glass; at temperatures in the vicinity of Tg,liq and above, the appearance of endothermic structural relaxations, as the result of the devitrification processes depicted in Figure 3, manifested with the curves abruptly converging to the supercooled liquid equilibrium line. Exceptionally, the glass prepared at TD = 72 K exhibited additional structural relaxation at temperatures just above its TD, as evidenced by the negative slope of the enthalpy curve at around (72−85) K, in correspondence with the exothermic effect in Figure 3b. Fictive temperature (Tfic), defined as the temperature at which each enthalpy curve intersects the supercooled liquid equilibrium line, was extracted as a representative parameter of the measure of enthalpy for a given glassy structure. Figure 5 shows the TD (Tfic for the LC glass) dependence of the enthalpic states at formation, that is, the enthalpies at the respective TD’s. A gradual departure from the extrapolated supercooled liquid equilibrium line, depicted by a thick solid line, was observed to occur for ECH glasses vapor-deposited in between 92 and 97 K. Below this temperature, the departure from the equilibrium line further increased with decreasing TD. Interestingly, the glasses prepared with TD in between ∼80 K and Tg,liq were observed to have smaller enthalpies at the TD’s than the respective values of the LC glass represented by the dashed curve. In opposition, the glass prepared at 72 K had a larger enthalpy than that of the LC glass. The inset of Figure 5 displays the variation of Tfic and Tdev as a function of TD (Tfic for the LC glass). It is noted that the behaviors of Tfic and Tdev were, in a sense, inversely proportional to each other at TD > 80 K because Tdev increased as Tfic decreased and vice versa. For TD’s near Tg,liq, Tfic appeared to follow the relation Tfic = TD. Assuming regions of quasi-linear variations for Tdev and Tfic with TD, a maximum and a minimum for Tdev and Tfic, respectively, were estimated to occur in the vicinity of TD ≈ 94 K (=0.93Tg,liq), as indicated by the vertical dotted line in the inset of Figure 5. Apparently, the

data varied smoothly with the exception of the sample prepared at 72 K. The molar heat capacities Cp,m for the LC and vapordeposited glasses within the devitrification temperature region are presented in Figure 6. The Cp,m profiles represented by lines were derived by first fitting the measured enthalpy data of the glasses to a smooth function according to the same procedure as that employed previously11 and, subsequently, calculating the derivatives with regard to temperature. Notably, the Cp,m profiles for the vapor-deposited glasses were characterized by large overshoots. The temperatures characterizing the Cp,m maxima of the overshoots varied systematically with TD, qualitatively corresponding with the variation of Tdev with TD. It was further observed that the glass prepared with TD = 0.96Tg,liq, which corresponds to the temperature region where TD ≥ 0.93Tg,liq, exhibited an exceptionally sharp peak shape in the Cp,m overshoot, whereas the glasses prepared with TD < 0.93Tg,liq featured a relatively broad peak shape in the Cp,m overshoots. The solid and short-dashed black lines in Figure 6 represent the Cp,m for, respectively, the slowly and rapidly precooled LC glass measured with apparatus 1. The devitrification profiles, as well as the Cp,m values characterizing the devitification process, for LC glasses were different according to the manner in which they were precooled. The inset of Figure 6, which shows the same Cp,m profiles within the temperature region of 80 ≤ T/K 4079

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B

enhanced kinetic properties in agreement with the results of Nakayama et al.5 and contrary to the prediction of Chen and Richert.19 Kearns et al. have reported the heat capacities, measured by means of AC nanocalorimetry, for indomethacin (IMC) vapordeposited at 0.84Tg,liq to be 4−5% lower than that of the LC glass in the temperature range of 0.94−1.02Tg,liq using a modulation frequency of 20 Hz.12 Recently, Ahrenberg et al. and Whitaker et al. respectively reported similar AC nanocalorimetric results for EB and toluene (TL) with TD > 0.8Tg,liq in the temperature range of 1.00−1.06Tg,liq9 and for cis-decalin with TD = 0.73Tg,liq in the temperature range of 0.74− 1.00Tg,liq,8 employing the same modulation frequency. In principle, AC nanocalorimetry below the Tg,dev differs from adiabatic calorimetry in that the latter involves the thermal character stemming from any irreversible process, including enthalpy relaxation effects due to glass transitions, in addition to vibrational contributions. It is difficult to compare the heat capacity contribution stemming only from the vibrational degree of freedom in the present work, although a similar result was obtained above. Previously, the devitrification properties of kinetically enhanced vapor-deposited glasses were characterized to potentially depend on the film thickness h of the sample. Kearns et al.13 investigated the transformation time for vapordeposited IMC from the glassy state into the supercooled liquid one by means of AC nanocalorimetry and characterized this transformation time to vary with h for films within the range of 75−600 nm of thickness. This dependence, however, was not observed to occur for thicknesses above 1.4 μm. Considering our present results on ECH, from the vapor-deposited sample masses ms and the substrate area, as well as assuming a globular size of ∼0.7 nm for the ECH molecule, the thicknesses of our prepared films are estimated to be on the order of 100 μm. Although some film inhomogeneity in h is expected to characterize the ECH films, such inhomogeneity is, intuitively, not expected to fall out of the micron scale. Therefore, considering the thickness regime for the ECH samples, the characterized kinetic properties are understood to be representative of relatively thick h-independent samples, and accordingly, the comparison between film properties with different TD’s can be assured to be valid. In passing, the possibility of co-deposition of undesired impurities, such as water molecules for example, during the preparation of the ECH films cannot indeed be excluded. Naturally, the cell, tube pipe, and glass tube inner walls potentially contained, at some points, adsorbed water molecules that were only gradually released with time; even despite the exhaustion by heating these walls under vacuum conditions, some impurities can be foreseen to be co-deposited with ECH, especially given the fact that our experimental setup did not reach ultrahigh vacuum conditions during sample deposition. Nevertheless, because we could measure crystalline ECH Cp,m correctly with apparatus 2 (Figure 2), co-deposition was not seen as being a substantial issue 4.2. Potential Picture of the Formation Process and the Formed Structures of Vapor-Deposited Glasses. In speculating as to why glasses with the described enthalpic states are formed, it seems clear that the resultant structures stem from the molecular processes and structures involved in and during the vapor deposition process. Up to the present, a few ideas and concepts have been proposed in attempts to explain the similar kinetic enhancement effects in other compounds

Figure 6. Temperature dependence of the molar heat capacities Cp,m for vapor-deposited and LC ECH glasses as extracted from the enthalpy curves; black solid line, slowly precooled LC glass; black short-dashed line, rapidly precooled LC glass; pink line, TD = 72 K glass; blue line, TD = 82 K; deep blue line, TD = 87 K; green line, TD = 92 K; orange line, TD = 97 K. The inset shows Cp,m at lower temperatures.

≤ 100, indicated little difference in Cp,m at around 0.79Tg,liq (=80 K) between the vapor-deposited glasses with TD ≥ 0.81Tg,liq and the LC ones. Nevertheless, in the range of 0.94Tg,liq (=95.3 K) − 0.99Tg,liq (=100 K), the Cp,m value for the vapor-deposited glass with TD = 0.96Tg,liq (=97 K) was smaller than that of slowly precooled LC glass by ∼2−10%.

4. DISCUSSION 4.1. Characteristics of the Vapor-Deposited ECH Glassy Films. The characters identified in the results presented above are similar to the ones previously reported on EB,11 as well as other molecules prepared in the similar manner under similar conditions;8,10,14,15 namely, the glasses prepared with 0.81 ≤ TD < Tg,liq are characterized by relatively low enthalpic states and high Tdev values. The dependence of Tdev on TD, with a Tdev maximum appearing within the TD range of 0.84− 0.92Tg,liq, is a well-established behavior for kinetically enhanced vapor-deposited glassy systems.8,11,15 In the case of vapordeposited ECH glasses, our estimated value of 0.93Tg,liq for the maximum is intuitively appropriate as it falls in the vicinity of this expected range. We previously investigated the calorimetric properties of EB and argued the change of peak characters from broad to sharp Cp,m overshoots as a function of TD to be a general feature of vapor-deposited glasses.11 The characterized behaviors of the Cp,m overshoots, as a function of TD, for the vapor-deposited ECH glasses are in good agreement with the published results on EB and, consequently, further support the potential generality of these characters for vapor-deposited glasses. In conclusion, given the identification of the features akin to kinetically enhanced glasses and the observed increase of Tdev relative to Tg,liq for all glasses prepared here by vapor deposition, it seems clear that ECH does form a glass with 4080

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B prepared by vapor deposition.4,7,10,14−18 Kearns et al. reported a double-peak feature in the DSC curves for the devitrification process of vapor-deposited IMC at 0.84Tg,liq and considered the possibility of distinct local packing structures within the IMC glass.14 Sepulveda et al. observed as well a double-peak feature in the Cp curve for devitrification of the glassy TL vapordeposited at 0.80Tg,liq, as measured by nanocalorimetry.10 They investigated the dependence of the overshoots on annealing temperature and concluded that the results were compatible with the existence of polyamorphism and a liquid−liquid (LL) transformation scenario in TL. Ishii et al. investigated the volumetric properties of alkylbenzenes and have argued for a potential LL transformation process for low-density vapordeposited glasses to high-density ones.7 They considered the high-density packing as resulting from molecular dimerization. Thus, concerning the molecular structure of the glasses vapordeposited immediately below Tg,liq, two views have been proposed. One apparently claims recognizing them as entities intertwined with the realization of polyamorphism. Opposite to this, the other view claims that the vapor deposition process, at temperatures immediately below Tg,liq, produces highly aged glasses.4 In the latter, the molecular structures were regarded to correspond to those expected for the liquid in its equilibrium state. Whether it is polyamorphism or highly aged structures, the ideas for correlating the glasses vapor-deposited near Tg,liq with the potential molecular structures have been forwarded in the form of a molecular mechanism that considers formation of these glasses in terms of an accumulation process of the surface molecules, with enhanced mobility, that organize the amorphous structure as to nearly achieve the equilibrium one.14,17,18 According to this view, the vapor-deposited surface molecules undergo rearrangements and, while sampling various packing arrangements, search for local packing configurations that are energetically lower in the potential energy landscape; increasing TD is portrayed to favor the surface equilibration process to proceed more rapidly and completely.17 Swallen et al. argued for mobility of molecules within a few nanometers of the surface.18 In this framework, the appearance of polyamorphism is explained in terms of yet unidentified liquid equilibrium states, which have been imagined to enthalpically underlie at temperatures in between Tg,liq and the Kauzmann temperature TK and are assumed to be partially accessible by vapor depositing below Tg,liq.16 In considering the structures that were potentially formed in the vapor-deposited ECH glasses, we propose below a potential concept that for some interval of TD, the resultant structures are a bulk manifestation of surface-specific molecular structures or structures based on surface-specific molecular interactions. It is pictured that the vapor deposition proceeds in a molecule-bymolecule fashion, and with TD below Tg,liq, it is noticed that the rearrangement of the molecule deposited is frozen in a certain configuration as the relaxation time elongates with development of the configurational ordering among the molecules on the surface or just beneath the surface. Namely, the configuration that the deposited molecule has frozen in depends on the TD, determining the dynamical property of the surface molecule, and the features of the intermolecular interactions of the surface molecules remain in the configuration of molecules frozen as explained in the following. The most important point in the present view is that the rearrangement as an elementary process is that of each molecule but not the cooperative rearrangement of plural

molecules; the neighboring molecules surrounding the rearranging molecule just slightly displace as to lower the potential barrier of the rearranging molecule in association with the relevant rearrangement of the focused molecule but are never located at any activated state for any rearrangement.26 The structure of deposited glasses may be conceived for simplicity in terms of molecular monolayers, even though the molecules may not necessarily constitute a layer structure in the really obtained glasses. Figure 7 conveys this picture schemati-

Figure 7. Conceptual picture of the intermolecular interactions: (a) the intermolecular interaction force within a bulk molecular environment; (b) a stronger intermolecular interaction force working for a molecule at the vapor deposition front; (c) vapor deposition front with the uppermost surface monolayer fully covered; (d) vapor deposition front with the uppermost monolayer partially covered. The lightestshaded molecules, (1) in (d), represent the surface ones that may undergo rearrangement motions with relative ease in view of no restriction by molecular embedment. The gray-shaded layers represent the surface molecules for which the molecular rearrangements are restricted to the least degree by molecular embedment. The blackcolored layers represent the molecules for which rearrangement motion is restricted by the presence of the upper surface layers.

cally. First, let us consider the intermolecular interactions working regarding the focused molecule illustrated with gray color as drawn in Figure 7a and b and the associated mobility of the molecule. In the bulk liquid, each molecule is surrounded by many other molecules, as illustrated in Figure 7a, and the focused molecule finds a stable configuration after having wholly taken into account the interactions with all of the surrounding molecules. Therefore, the molecule makes a relatively weak interaction with each neighboring molecule but is hard to change its configuration because of the large interaction energy obtained on the whole. On the other hand, in the situation shown in Figure 7b, the focused gray-colored molecule representing a surface molecule interacts only with the molecules being beneath itself and finds its configuration by taking into account only the interactions of a smaller number than that for the molecule in bulk. Because of the smallness in the number, the molecule makes stronger interactions with each of the molecules located beneath but is more mobile due to the weak interaction on the whole than the bulk one. It corresponds to the same chemical phenomenon that gives rise to the property of surface tension. We hypothesize that in the ECH case, TD = 0.93Tg,liq yields expectedly the Tdev maximum and corresponds to a condition where only the molecules constituting the uppermost surface layer are thermally able to undergo rearrangement motions and have frozen in when having filled the surface area. Therefore, at this temperature, the molecular rearrangement in the underlayers, that is, the second and third surface layers, remains arrested. The intermolecular interaction formed between the uppermost layer molecule and each of the others is expected to be rather stronger than that formed in the bulk, as explained above. The uppermost layer, in which the molecular configuration was frozen in, then becomes the second surface layer when the next vapor molecules have been deposited as forming a new uppermost surface layer. The molecules in the 4081

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

Article

The Journal of Physical Chemistry B

Figure 8. Conceptual pictures representing the liquid or vitreous structures near the surface region for ECH-like elliptical molecules prepared at various conditions: (a) equilibrium liquid state, characterized by isotropic arrangement of molecules on the average and a rather flattened configuration of the uppermost surface molecules; (b) vitreous state prepared by vapor deposition at TD ≪ Tg,liq, characterized as accumulation rather of an as-deposited molecular configuration; (c) vitreous state prepared by vapor deposition at TD ≈ 0.93Tg,liq, characterized as accumulation of elliptical surface molecules taking their flat configuration and the resulting clearly anisotropic structure.

rearrange to rather approach the structure realized in the bulk liquid through the three-dimensionally determined intermolecular interactions. Figure 8 schematically displays conceptual illustrations on the molecular structures of ECH for the extreme cases as pictured according to our model. Figure 8a depicts the liquid in the bulk equilibrium state where the ECH molecules assumed to be elliptical make intermolecular interactions predominantly through dispersion force, resulting in an isotropic arrangement of molecules on the average. The uppermost molecules, nevertheless, tend to flatten as to make stronger interactions with the molecules beneath themselves, leaving their anisotropic nature of the relaxed surface structure. Figure 8b portrays the amorphous structure for the extreme case where molecules were deposited at TD ≪ Tg,liq. Here, the structure is highly disordered, and the molecules have been frozen in mostly as deposited. The surface molecules are lacking in establishment of their most stable configuration, and the resulting glass structure is characterized by explicit voids present among molecules. The illustration in Figure 8c, in turn, depicts the structure for the ECH glass deposited at 0.93Tg,liq. This structure is characterized by accumulation of flattening elliptical molecules, representing the structure of the energetically enough stabilized uppermost surface molecules. As explained above, an increase of TD beyond 0.93Tg,liq allows the flattening structure, leading to the structure of Figure 8c, to gradually relax, resulting in gradual resemblance to the bulk one depicted in Figure 8a. Recently, Dawson et al. reported conclusive evidence for the presence of molecular stacking and formation of an anisotropic structure in the vapor-deposited IMC glasses prepared at 0.84Tg,liq and investigated by means of 1D- and 2D-WAXS experiments.6 Accordingly, we understand the manifestation of this IMC anisotropic structure as a result of the strong intermolecular interactions at the deposition front as proposed by our molecular accumulation mechanism. Finally, we consider the low-Cp,m character of vapordeposited glasses. To our knowledge, no general explanation for this experimental fact has been provided yet. Reasoning within the framework of our proposed model, we can argue that the low heat capacities of the vapor-deposited glasses may potentially originate from the enhanced intermolecular interactions, in accordance with the low enthalpies of the vapor-deposited glasses. In a simplistic fashion, the intermolecular vibrations may be viewed as a sum of harmonic oscillators. The enhancement of intermolecular interactions ordinarily causes the force constant for the intermolecular vibrations to increase and therefore causes the splitting between

second layer, in the meantime, acquire new and strong intermolecular interactions with the new uppermost layer molecules. The accumulation of such intermolecular interactions is considered, as a total, to stabilize more the vapordeposited glass than that of bulk supercooled liquid glass where the molecular configuration is determined through threedimensionally determined intermolecular interactions. While the structure formed through this accumulation process is considered to be unstable from the entropic viewpoint as compared with the supercooled liquid, it is in the frozen-in state, resulting kinetically in the ineffectiveness of the entropic term for molecules to undergo any rearrangement toward the stable state from the viewpoint of Gibbs energy below Tdev. The glass structure characteristic of the surface layer molecular configuration must have an attribute of anisotropic nature in any sense. Because the sum of the intermolecular interactions of each molecule is larger than that in the bulk liquid glass, the Tf and Tdev of the vapor-deposited glass at TD = 0.93Tg,liq are lower and higher, respectively, than the values of the supercooled liquid glass. When the TD is below 0.93Tg,liq, the molecule constituting the uppermost surface monolayer undergoes, similarly to the case at TD = 0.93Tg,liq, its rearrangement in order to find more stable configurations than the one formed at the immediate instant of its deposition event. Nevertheless, the population of molecules within the uppermost surface monolayer that attains the stable configuration of TD = 0.93Tg,liq would be reduced according to the reduction in TD, namely, to the increase in the immobility of the surface molecules. The extent of the reduction would become striking as the TD is further lowered. Therefore, below TD = 0.93Tg,liq, these molecules are perceived to be arrested and frozen in a more unstable state than that attained at TD = 0.93Tg,liq. In opposition to this, increasing TD beyond 0.93Tg,liq, for example, to 0.95Tg,liq, is perceived to allow the surface molecules to rearrange and, to some extent, allow even the molecules immediately below the surface molecular layer to rearrange due to the easiness in the replacement of the surface layer molecules. Here, we hypothesize that a 0.02Tg,liq increment for TD from 0.93Tg,liq will be sufficient to enable molecular rearrangement motion for the majority of the molecules within the second surface monolayer to occur but insufficient to promote rearrangement motion for the molecules constituting the third surface monolayer given their degree of embedment. Potentially, at a higher TD, such as 0.97Tg,liq, restriction against the molecular rearrangement motion from embedment in the third surface layer might cease. Under such a condition, it is considered that the molecules constituting the second and third surface layers 4082

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083

The Journal of Physical Chemistry B



the energy levels of the harmonic oscillator to increase, resulting in a smaller contribution of the intermolecular vibrations to the heat capacity. The intermolecular vibrational frequency band of molecular liquids such as substituted benzene/cyclohexane compounds is located in the lowfrequency region below 150 cm−1.27 Provided that the Einstein’s or Debye’s model is applicable to the expression of the heat capacity due to the intermolecular vibrations, the heat capacity decrease of, say, 4% corresponds only to the increase in the frequency from 150 to 153 cm−1. Such a situation is in a range of plausibility for our postulated intermolecular interaction enhancement in the glasses vapor-deposited in the vicinity of Tg,liq.

REFERENCES

(1) MacMillan, J. A.; Los, S. C. J. Chem. Phys. 1965, 42, 829. (2) Takeda, K.; Oguni, M.; Suga, H. Thermochim. Acta 1990, 158, 195. (3) Ediger, M. D.; Harrowell, P. J. Chem. Phys. 2012, 137, 080901. (4) Singh, S.; Ediger, M. D.; de Pablo, J. J. Nat. Mater. 2013, 12, 139−144. (5) Nakayama, H.; Omori, K.; Ino-u-e, K.; Ishii, K. J. Phys. Chem. B 2013, 117, 10311−10319. (6) Dawson, K. J.; Zhu, L.; Yu, L.; Ediger, M. D. J. Phys. Chem. B 2011, 115, 455−463. (7) Ishii, K.; Nakayama, H.; Moriyama, R.; Yokoyama, Y. Bull. Chem. Soc. Jpn. 2009, 82 (10), 1240−1247. (8) Whitaker, K. R.; Scifo, D. J.; Ediger, M. D.; Ahrenberg, M.; Schick, C. J. Phys. Chem. B 2013, 117 (42), 12724−12733. (9) Ahrenberg, M.; Chua, Y. Z.; Whitaker, K. R.; Huth, H.; Ediger, M. D.; Schick, C. J. Chem. Phys. 2013, 138, 024501. (10) Sepulveda, A.; Leon-Gutierrez, E.; Gonzalez-Silveira, M.; Clavaguera-Mora, M. T.; Rodríguez-Viejo, J. J. Phys. Chem. Lett. 2012, 3, 919−923. (11) L. M. Ramos, S. L.; Oguni, M.; Ishii, K.; Nakayama, H. J. Phys. Chem. B 2011, 115 (49), 14327. (12) Kearns, K. L.; Whitaker, K. R.; Ediger, M. D.; Huth, H.; Schick, C. J. Chem. Phys. 2010, 133, 014702. (13) Kearns, K. L.; Ediger, M. D.; Huth, H.; Schick, C. J. Phys. Chem. Lett. 2010, 1 (1), 388−392. (14) Kearns, K. L.; Swallen, S. F.; Ediger, M. D.; Sun, Y.; Yu, L. J. Phys. Chem. B 2009, 113, 1579−1586. (15) Kearns, K. L.; Swallen, S. F.; Ediger, M. D.; Wu, T.; Yu, L. J. Chem. Phys. 2007, 127, 154702. (16) Dawson, K. J.; Kearns, K. L.; Yu, L.; Steffen, W.; Ediger, M. D. Proc. Natl. Acad. Sci. U.S.A 2009, 106, 15165−15170. (17) Kearns, K. L.; Swallen, S. F.; Ediger, M. D.; Wu, T.; Yu, L. J. Phys. Chem. B 2008, 112, 4934−4942. (18) Swallen, S. F.; Kearns, K. L.; Satija, S.; Traynor, K.; McMahon, R. J.; Ediger, M. D. J. Chem. Phys. 2008, 128, 214514. (19) Chen, Z.; Richert, R. J. Chem. Phys. 2011, 135, 124515/1− 124515/6. (20) Hikawa, H.; Oguni, M.; Suga, H. J. Non-Cryst. Solids 1988, 101, 90−100. (21) Sugisaki, M.; Suga, H.; Seki, S. Bull. Chem. Soc. Jpn. 1968, 41, 2591−2599. (22) Fujimori, H.; Oguni, M. J. Phys. Chem. Solids 1993, 54, 271− 280. (23) L. M. Ramos, S. L.; Ogino, M.; Oguni, M. J. Therm. Anal. Calorim. 2014, 115, 1353−1358. (24) Suga, H.; Seki, S. Faraday Discuss. 1980, 69, 221. (25) Matsuo, T.; Oguni, M.; Suga, H.; Seki, S. Bull. Chem. Soc. Jpn. 1974, 47, 57. Oguni, M.; Matsuo, T.; Suga, H.; Seki, S. Bull. Chem. Soc. Jpn. 1977, 50, 825. (26) Oguni, M. J. Non-Cryst. Solids 1997, 210, 171−177. (27) Chang, Y. J.; Castner, E. W., Jr. J. Phys. Chem. 1996, 100, 3330− 3343.

5. CONCLUSION A homemade high-precision adiabatic calorimeter adapted for the preparation of vapor-deposited samples was presented, and with it, the thermal properties for the ECH glasses vapordeposited at diverse temperatures immediately below Tg,liq were investigated. The imprecision of our calorimeter for vapordeposited glasses was ±0.3% within the temperature interval in which the vapor deposition temperature TD was set. It was shown that ECH undoubtedly forms a kinetically enhanced glass, as compared to the glass when prepared by the conventional method of cooling the liquid or supercooled across Tg,liq, when deposited with 0.80Tg,liq < TD < Tg,liq. In attempting to elucidate the origins behind the kinetic enhancement, a molecular picture and mechanism of formation for the deposited structures were proposed. It was argued that the formation of the glasses revealing peculiar characters originates especially from the molecule-by-molecule deposition manner in the vapor deposition process and depends mostly on the mobility/immobility of the uppermost surface molecules at TD. The vapor-deposited sample was argued to yield a glassy structure different from the equilibrium liquid or the LC glass structures, in view that the rearrangement of the surface molecules proceeds under a different molecular environment from that in the bulk. This aspect, in turn, not only enables the deposited molecule to favor other types of intermolecular interaction over conventionally favored ones in the bulk environments but also allows the molecules to interact more strongly with each of the other molecules. The deposition temperature at which the most anisotropic glass structure can be experimentally formed was predicted to be near 0.93Tg,liq for ECH.



Article

ASSOCIATED CONTENT

S Supporting Information *

Apparatus and technical details of an adiabatic calorimeter with filling tubes and procedures for sample preparation and calorimetric measurements. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address †

S.L.L.M.R.: Department of Applied Physics, Institute of Physics, University of São Paulo, Cidade Universitária, São Paulo 05508-090, Brazil. Notes

The authors declare no competing financial interest. 4083

DOI: 10.1021/jp5109174 J. Phys. Chem. B 2015, 119, 4076−4083