The central idea of our experimental approach is conceptually simple, and represents a transformative modification and fusion of two traditional analytical methods: differential canning calorimetry (DSC), and temperature programmed desorption (TPD) spectroscopy. As illustrated in Fig.1, the core of the experimental apparatus is a "hair-thin" wolfram filament, which is 10 µm in diameter. The filament is attached to supports inside in a high vacuum chamber, and cooled to temperatures between 90 and 180 K. Under such conditions, a variety of amorphous, polycrystalline, or viscous liquid films with thicknesses from a few nanometers to a few micrometers can be grown on the filament by physical vapor deposition (PVD). In short, the first stage of our experiments is identical to the high vacuum surface science TPD methods, which have been used in classical surface science for many decades. Thus, our technique retains numerous advantages related to the material samples preparation method. For example, our micrometer thick films can be stratified with nanoscale layers of reactive "guest" chemical species, or saturated with dopants well beyond there solubility limits.
Similar to a TPD experiment, the filament (substrate) is heated to higher temperatures after deposition of a film. However, unlike the traditional TPD or DSC instruments, our system is capable of heating the filament with rates up to 1,000,000 Kelvins per second. The "ultrafast" heating capability of the apparatus instantly yields serval crucial advantages. For example, the rapid heating ensures that the filament reaches near ambient temperatures before evaporation of a significant fraction of a film into surrounding vacuum. This particular feature is especially important in studies of environmentally relevant dynamics and reactions in ice. Furthermore, the high heating rates yield an unprecedented sensitivity during our calorimetric measurements. In fact, a single fast colorimetric scan makes it possible to detect and characterize glass softening transitions in amorphous films as thin as 100 nm 1. Thus, by averaging the data from a few dozens of repeated scans, we can gain a variety of insight into molecular dynamics and phase transitions in materials confined to nanoscale dimensions. Yet, as we explain below, the most important advantage of the experimental approach arises from the synergy of two analytical techniques.
Figure 2 illustrates typical FSC-FTDR MS data obtained after completion of the three experimental stages: film deposition at constant temperature, its rapid heating and vaporization (Fig. 2a). The data in the left panel of Fig. 2b is the FSC thermogram of amorphous solid water (ASW), which is vapor-deposited at 95 K and doped with D2O. The thermogram, which is the overall heat capacity of the film plotted as a function of temperature, contains three discernible exotherms due to pore collapse, formation of nuclei of crystalline ice, and their growth2,3. Thus, the calorimetric data can be used to estimate the porosity of the original amorphous film, and the extent of its crystallinity. For example, repeating the experiment with an H2O film vapor-deposited at temperatures above 160 K yields a fully crystallized sample, which is devoid of any exotherms (see Fig. 2b). Both thermograms facilitate accurate determination of the deposited film's mass by comparing the overall heat capacity of the sample to known specific heat capacity of ASW or crystalline ice. The endothermic upswing in the apparent heat capacity at temperatures above - 35 oC is due to onset of rapid vaporization of the crystallized film: as the temperature is approaching ice melting point, the heat generated resistively by filament is carried away by sublimation of the micrometer scale film. As a result, the film-filament system enters a near-isothermal stage (see right panels of Fig. 2a and b).
During the isothermal stage of a FSC-FTDR MS experiment, the flux of particular reaction-vaporization products can be monitored with a quadrupole mass spectrometer as a function of time. The right panel of Fig.2 b shows FTDR MS spectra from a "sandwich-like" H2O/D2O/H2O film, in which the thickness of the D2O film is approximately 50 nm 4,5. Positioning the D2O layers at a known "depth" inside the H2O film defines the time of deuterium-hydrogen reaction, which is diffusion controlled at temperatures near 0 oC. Thus, by integrating TDR MS spectra of the reactant (D2O) and the product (HDO), we can infer water self-diffusion coefficient at grain boundaries of polycrystalline ice at a chosen temperature near ice melting point4,5.