Ultrastable glasses

Since 2008 GNaM is devoted to the study of ultrastable glass formation via vapor-deposition at temperatures in the vicinity of the glass transition temperature. The field started in 2007 with the discovery by Ediger’s group of ultrastability in various pharmaceutical compounds grown from the vapor. This subject is rapidly gaining interest in the glass community and ultrastable glass formation has been extended to many molecules and materials, including small and medium size organic molecules, polymers and metallic glasses. The increased of stability is traced-back to enhanced molecular mobility at the surface. Molecules grown from the vapor have time to explore lower positions of the energy landscape before being buried by other molecules.

Ultrastable organic glasses from small-molecules

Our first attempts to obtain vapor-deposited molecular glasses with enhanced stability date back to 2008. At that time we developed an ultra-high vacuum chamber with evaporation capabilities and in-situ nanocalorimetry that enabled us to monitor the heat capacity of the as-grown thin film glasses over a wide temperature interval. See JCP2008 and SSA2008 for details. We demonstrated that small molecular glasses such as toluene and/or ethylbenzene could indeed be grown as ultrastable glasses at temperatures below Tg, similarly to previous reports on larger molecules of indomethacin and this-naphtylbenzene. Toluene and ethylbenzene films grown at around 0.8Tg showed enhanced kinetic and thermodynamic stability (see PCCP2010). This work also evidenced that the growth rate of the films, a very important parameter in larger molecules, did not have a large influence on stability. The reason being that smaller molecules have larger mobilities and can efficiently sample lower positions of the PEL in shorter time scales.

Our ability to monitor the heat capacity of very thin films using in-situ quasi-adiabatic nanocalorimetry permitted us to explore the influence of the thickness of the films on the onset of the glass transition and on the fictive temperature of glasses of different stabilities. Since we did observe a clear variation of the onset of the glass transition as a function of thickness in our specific heat data, our analysis concluded that indeed vapor-deposited glasses have a size-dependent glass transition. The changes were made evident at sizes below 20-30 nm (see JPCL2010). Later measurements in indomethacin have shown that the transformation mechanism has a tremendous impact in the specific heat evaluation and therefore the observed variation in toluene glasses could be a consequence of the heterogeneous transformation from the surface. New experiments are under way to confirm this fact.

Ultrastable glasses of pharmaceutical drugs

Our latest work refers to larger molecules, such as indomethacin, the archetypical ultrastable glass former, or celecoxib.

New molecules

We have recently demonstrated that celecoxib, a poor glass former with a strong tendency towards crystallization, could be grown as an ultrastable glass at temperatures in the vicinity of 0.85 Tg. Vapor-deposited ultrastable glasses of Celecoxib show an onset temperature of the glass transition 20 K above its conventional value and a high thermodynamic stability. The time required for the loss of molecular anisotropy is 3 orders of magnitude higher than the time scales required for the alpha relaxation. We also report the influence of the stability of the glass on the surface crystallization at temperatures below the glass transition temperature. The growth of surface crystals is 30% faster in conventional glasses prepared from the liquid than in highly stable glasses.

CXIB crystal growth

(a) Optical microscopy picture of a surface crystal after 10 h at 318 K and 50% of relative humidity. (b) Square root of the area of a crystal (normalized to the initial area) as a function of time for three different batches of samples. The slow cooled samples have been obtained by heating an as-deposited UG sample above the glass transition temperature and cooling it down at 10 K/min. The same crystalline phase appeared on all samples, independently on the batch or being as-deposited or slow cooled.

See JNCS2014.

Low-T properties

A remarkable feature of glasses is that, independently of their nature and composition, they exhibit universal properties in the low-temperature range. Of interest here, the specific heat is characterized by a linear term below 1 K, ascribed to quantum tunneling between two states of similar energy. We have investigated if this ubiquitous behavior also applies to so-called 'ultrastable glasses', directly synthesized from the vapor phase into low-energy positions of the potential-energy landscape. Interestingly, we find a full suppression of the linear term in the specific heat, which questions the current view of the popular tunneling model and sheds light on the microscopic origin of two-level systems in glasses.

Low-T specific heat of IMC

Specific-heat data for USGs of IMC 50 μm- (USG-1) and 80 μm (USG-2) thin films, compared with the crystalline phase (Debye extrapolated at lower temperatures) and the conventional glass. A degraded USG (Materials and Methods) has also been measured and is presented. Dashed lines show the corresponding linear fits CP = cTLS·T + cD·T3 for experimental data below 2K. (A) Debye-reduced CP/T3 versus T representation

See PNAS2014.

Transformation fronts

Ultrastable thin film glasses transform into supercooled liquid via propagating fronts starting from the surface and/or interfaces. In this paper, we analyze the consequences of this mechanism in the interpretation of specific heat curves of ultrastable glasses of indomethacin for samples with varying thickness from 20 nm up to several microns. We demonstrate that ultrastable films above 20 nm have identical fictive temperatures and that the apparent change of onset temperature in the specific heat curves originates from the mechanism of transformation and the normalization procedure. An ad hoc surface normalization of the heat capacity yields curves which collapse into a single one irrespective of their thickness. Furthermore, we fit the surface-normalized specific heat curves with a heterogeneous transformation model to evaluate the velocity of the growth front over a much wider temperature interval than previously reported. Our data expands previous values up to Tg + 75 K, covering 12 orders of magnitude in relaxation times. The results are consistent with preceding experimental and theoretical studies. Interestingly, the mobility of the supercooled liquid in the region behind the transformation front remains constant throughout the thickness of the layers.

Transformation speed of growth front of IMC

In the first figure there is the surface-normalized heat capacity of highly stable IMC thin film glasses, the collapse of all curves to a common onset of the transformation can be described by a parallel growth front mechanism. In the second figure there's the transformation speed as a function of relaxation time (via VFT) for films with thicknesses from 20 nm to 4 μm evaluated from nanocalorimetry and DSC data using eq 3. Continuous lines: this work. Symbols: blue circles, data from SIMS; 9 violet circles, data from dielectric spectroscopy; 10 red square, data from ac nanocalorimetry.11 The value in the graph indicates the initial thickness (from the surface) at which the velocity of the transformation is evaluated. The black dashed line corresponds to the function vgr = 0.1·τ-0.78, where τ = τ0 exp(DT0/(T - T0)) is the VFT fit of the relaxation time using the bulk values for IMC.

See JPCB2014.

Campus d'excel·lència internacional U A B