Search for: All records

X-ray scattering has been used to characterize the columnar packing and the π stacking in a glass-forming discotic liquid crystal. In the equilibrium liquid state, the intensities of the scattering peaks for π stacking and columnar packing are proportional to each other, indicating concurrent development of the two orders. Upon cooling into the glassy state, the π–π distance shows a kinetic arrest with a change in the thermal expansion coefficient (TEC) from 321 to 109 ppm/K, while the intercolumnar spacing exhibits a constant TEC of 113 ppm/K. By changing the cooling rate, it is possible to prepare glasses with a wide range of columnar and π stacking orders, including zero order. For each glass, the columnar order and the π stacking order correspond to a much hotter liquid than its enthalpy and π–π distance, with the difference between the two internal (fictive) temperatures exceeding 100 K. By comparison with the relaxation map obtained by dielectric spectroscopy, we find that the δ mode (disk tumbling within a column) controls the columnar order and the π stacking order trapped in the glass, while the α mode (disk spinning about its axis) controls the enthalpy and the π–π spacing. Our finding is relevant for controlling the different structural features of a molecular glass to optimize its properties. 

Molecules at a liquid/vapor interface have different organizations and mobilities from those in the bulk. These differences potentially influence the rate of crystal nucleation, but the effect remains imperfectly understood. We have measured the crystal nucleation rates at the surface and in the bulk of amorphous poscaconazole, a rod-like molecule known to have a preferred interfacial orientation. We find that surface nucleation is vastly enhanced over bulk nucleation, by ∼9 orders of magnitude, and selects a different polymorph (II) from bulk nucleation (I). This phenomenon mirrors the recently reported case of D-arabitol and stems from the similarity of anisotropic surface molecular packing to the structure of the surface-nucleating polymorph. In contrast to these two systems, the surface enhancement of nucleation is weaker (though still significant) in acetaminophen and in water and does not select a different polymorph. Together, the systems investigated to date all feature surface enhancement, not suppression, of crystal nucleation, and those showing a polymorphic change feature (1) structural reconstruction at the surface relative to the bulk and (2) existence of a different polymorph that can take advantage of the surface environment to nucleate. These results help predict the effect of a liquid/vapor interface on crystal nucleation and polymorph selection, especially in systems with a large surface/volume ratio, such as atmospheric water and amorphous particles.

 

Surface diffusion has been measured in the glass of an organic semiconductor, MTDATA, using the method of surface grating decay. The decay rate was measured as a function of temperature and grating wavelength, and the results indicate that the decay mechanism is viscous flow at high temperatures and surface diffusion at low temperatures. Surface diffusion in MTDATA is enhanced by 4 orders of magnitude relative to bulk diffusion when compared at the glass transition temperature T g . The result on MTDATA has been analyzed along with the results on other molecular glasses without extensive hydrogen bonds. In total, these systems cover a wide range of molecular geometries from rod-like to quasi-spherical to discotic and their surface diffusion coefficients vary by 9 orders of magnitude. We find that the variation is well explained by the existence of a steep surface mobility gradient and the anchoring of surface molecules at different depths. Quantitative analysis of these results supports a recently proposed double-exponential form for the mobility gradient: log  D( T, z) = log  D v ( T) + [log  D 0 − log  D v ( T)]exp(− z/ξ), where D( T, z) is the depth-dependent diffusion coefficient, D v ( T) is the bulk diffusion coefficient, D 0 ≈ 10 −8  m 2 /s, and ξ ≈ 1.5 nm. Assuming representative bulk diffusion coefficients for these fragile glass formers, the model reproduces the presently known surface diffusion rates within 0.6 decade. Our result provides a general way to predict the surface diffusion rates in molecular glasses.