The properties of amorphous solid water at and near the calorimetric glass transition temperature, Tg, of 136 K have been debated for years. One hypothesis is that water turns into a "true" liquid at Tg (i.e., it becomes ergodic) and exhibits all the characteristics of an ergodic liquid, including translational diffusion. A competing hypothesis is that only rotational motion becomes active at Tg, while the "real" glass transition in water is at a considerably higher temperature. To address this dispute, we have investigated the diffusive mixing in nanoscale water films, with thicknesses up to ∼100 nm, using infrared (IR) spectroscopy. The experiments used films that were composed of at least 90% H2O with D2O making up the balance and were conducted under conditions where H/D exchange was essentially eliminated. Because the IR spectra of multilayer D2O films (e.g., thicknesses of ∼3-6 nm) embedded within thick H2O films are distinct from the spectrum of isolated D2O molecules within H2O, the diffusive mixing of (initially) isotopically layered water films could be followed as a function of annealing time and temperature. The results show that water films with total thicknesses ranging from ∼20 to 100 nm diffusively mixed prior to crystallization for temperatures between 120 and 144 K. The translational diffusion had an Arrhenius temperature dependence with an activation energy of 40.8 ± 3.5 kJ/mol, which indicates that water at and near Tg is a strong liquid. The measured diffusion coefficient at 136 K is 6.25 ± 1.4 × 10-21 m2/s.

Isotopically layered amorphous solid water films were used to measure the diffusivity of deeply supercooled liquid water near the glass transition. The films, composed of separate H218O and H216O layers, were grown by vapor deposition at low temperature and then heated to observe the intermixing of the isotopic layers. Very slow heating rates (as low as 10-4 K/s) were used to decouple the diffusion and crystallization processes to ensure that the observed intermixing occurred at temperatures that were well-separated from the onset of crystallization. Numerical simulations of the desorption spectra were used to extract the translational diffusivities. The diffusivities obtained in this paper are consistent with translational liquid-like motion at temperatures near and above the proposed Tg of 136 K. These findings support the idea that the melt of amorphous water, above its glass transition temperature is thermodynamically continuous with normal supercooled liquid.

Glassy solid forms of water can be made by vapor deposition, by the rapid cooling of small liquid droplets or by squashing ice. For the glass to be connected to normal water by a reversible path at atmospheric pressure its residual excess entropy relative to ice needs to be effectively zero. Measurement of the rates of evaporation of glassy water and ice near 150 K gives a measure of their free energy difference which implies an entropy difference of ΔS = 1.3±1.7 J/(K·mol) at 150 K. There are two interpretations of the small value of ΔS, depending on whether the measured free energy of the glass reflects the calorimetric or the statistical mechanical value of the entropy of a glass. An experiment is proposed to test that issue.

The atomic-level characterization of active sites is essential to understanding the mechanisms behind catalytic reactions. In this study, using scanning tunneling microscopy and X-ray photoelectron spectroscopy, we follow the morphological changes of a model Rh catalyst supported on Fe3O4(001) as a function of temperature and Rh coverage. We identify the preparation conditions leading to model catalysts containing homotopic or nearly homotopic Rh species bound as adatoms, substitutional octahedral sites within the Fe3O4(001), and nanoparticles. Adsorbates such as CO and CO2 are subsequently used to characterize the properties of different Rh sites. Using temperature-programmed desorption, we demonstrate that adatoms and nanoparticles exhibit high-temperature CO desorption (250-600 K). Strong binding on such sites further allows for CO oxidation to CO2 via the Mars-van Krevelen mechanism. In contrast, CO2 was found to interact weakly with all Rh sites. Differences in desorption temperature enable the use of CO and CO2 as titration methods for nanoparticles and Fe3O4(001), respectively. A small quantity of CO2 was found to be reduced to CO on Rh adatoms and small nanoparticles.

The reaction pathways of glycerol on TiO2(110) have been studied by molecular beam dosing and temperature-programmed desorption. The majority of adsorbed glycerol undergoes reactions to yield water, hydrogen, and carbon-containing products. For both water and hydrogen, we identify two distinct reaction channels resulting from the cleavage of O-H and C-H bonds. Quantification of the desorption yields reveals that the C-H bond scission channel, observed at temperatures higher than for the O-H bond scission, is a dominant reaction channel for hydrogen evolution. For the carbon-containing products, we identify gaseous allyl radical as the major product. Various minor carbon-containing products, including propylene, acrolein, allyl alcohol, acetone, formaldehyde, and ethylene, are also observed and quantified. The different surface chemistries observed between glycerol and simpler alcohols and glycols on TiO2(110), suggest that smaller oxygenates cannot serve as models for larger oxygenates in probing their reaction pathways on oxide surfaces.

The adsorption of the C2 hydrocarbons, including ethane, ethene, and ethyne, are studied on magnetite Fe3O4(001) by a combination of molecular beam dosing, temperature programmed desorption, and X-ray photoelectron spectroscopy. The ethane desorption profile has a single temperature invariant peak at 100 K, while ethene and ethyne exhibit additional peaks at similar to 120 -and similar to 135 K. An inversion analysis is used to extract coverage-dependent desorption energies as well as coverage-averaged prefactors for each molecule. Ethene and ethyne exhibit moderate coverage-dependent desorption energies decreasing from similar to 80 to similar to 30 kJ/mol at saturation, while ethane shows a relatively coverage invariant desorption energy similar to 28 kJ/mol. The desorption energies of the C2 hydrocarbons increase in the order of increasing bond order. This is likely due to the enhanced interaction of the ethene and ethyne pi system to the coordinatively unsaturated octahedral Fe sites (Fe-oct) on the oxide surface. The saturation coverages of each C2 hydrocarbon at 82 K were also determined. These match well with the 2-dimensional area determined from their liquid densities.

We measure the isothermal crystallization kinetics of amorphous acetonitrile films using molecular beam dosing and reflection adsorption infrared spectroscopy techniques. Experiments on a graphene covered Pt(111) substrate revealed that the crystallization rate slows dramatically during long time periods and that the overall kinetics cannot be described by a simple application of the Avrami equation. The crystallization kinetics also have a thickness dependence with the thinner films crystallizing much slower than the thicker ones. Additional experiments showed that decane layers at both the substrate and vacuum interfaces can also affect the crystallization rates. A comparison of the crystallization rates for CH3CN and CD3CN films showed only an isotope effect of ∼1.09. When amorphous films were deposited on a crystalline film, the crystalline layer did not act as a template for the formation of a crystalline growth front. These overall results suggest that the crystallization kinetics are complicated, indicating the possibility of multiple nucleation and growth mechanisms.

Crystalline acetonitrile has two polymorphs, a high-temperature (HT) phase that is stable between 217 K and its melting point at 229 K and a low-temperature (LT) phase that is stable below 217 K. Solid acetonitrile films can be prepared by vapor deposition in an ultrahigh vacuum chamber. To prevent sublimation of the film, temperatures are often kept below 150 K. While the LT phase is thermodynamically favored at these low temperatures, such preparation usually results in the formation of the metastable HT polymorph. In this work we use reflection adsorption infrared spectroscopy (RAIRS) and temperature-programmed desorption (TPD) experiments to investigate the effects of the deposition temperature and underlying substrate on the morphology of acetonitrile films prepared with molecular beam deposition. We obtained the elusive LT phase when dosing at 120 K on a graphene substrate and on a crystalline decane layer. Dosing acetonitrile on other surfaces produced the HT phase, as did annealing of amorphous films. We used TPD experiments to determine the Gibbs energy difference between the HT and the LT phases. Our ΔG values agree with extrapolation of equilibrium calorimetry data. We also observed that acetonitrile films were amorphous when dosed at temperatures ≤ 60 K and porous for temperatures ≤ 50 K.

Acetonitrile thin films were prepared on Pt(111) and on graphene on Pt(111) using molecular beam techniques. Temperature programed desorption (TPD) experiments of acetonitrile on Pt(111) displayed first-order kinetics for monolayer desorption and zero-order kinetics for multilayer desorption. We used reflection adsorption infrared spectroscopy (RAIRS) to characterize the orientation of the monolayer on Pt(111). Acetonitrile molecules have been shown to adopt a bridging structure where the C N bond breaks to form two single bonds to underlying platinum atoms. When T-deposition >= 120 K, our RAIRS spectra were consistent with this bridging configuration. However, when T-deposition was lowered to 60 K, molecules retained the C N bond and adopted a vertical orientation with the nitrogen closest to the platinum. On graphene, the TPD experiments show that both the monolayer and multilayer desorb with zero-order kinetics. Our RAIRS spectra suggest that the monolayer consists of acetonitrile molecules oriented with their molecular axes parallel to the graphene substrate. This orientation did not change when T(deposition)( )was varied from 60 to 120 K. Our results provide a more complete picture of how acetonitrile molecules interact with Pt(111) and with graphene surfaces.

The interaction of methanol with iron oxide surfaces is of interest due to its potential in hydrogen storage and from a fundamental perspective as a chemical probe of reactivity. We present here a study examining the adsorption and reaction of methanol on magnetite Fe3O4(001) at cryogenic temperatures using a combination of temperature programmed desorption, x-ray photoelectron spectroscopy, and scanning tunneling microscopy. The methanol desorption profile from Fe3O4(001) is complex, exhibiting peaks at 140 K, 173 K, 230 K, and 268 K, corresponding to the desorption of intact methanol, as well as peaks at 341 K and 495 K due to the reaction of methoxy intermediates. The saturation of a monolayer of methanol corresponds to ∼5 molecules/unit cell (u.c.), which is slightly higher than the number of surface octahedral iron atoms of 4/u.c. We probe the kinetics and thermodynamics of the desorption of molecular methanol using inversion analysis. The deconvolution of the complex desorption profile into individual peaks allows for calculations of both the desorption energy and the prefactor of each feature. The initial 0.7 methanol/u.c. reacts to form methoxy and hydroxy intermediates at 180 K, which remain on the surface above room temperature after intact methanol has desorbed. The methoxy species react via one of two channels, a recombination reaction with surface hydroxyls to form additional methanol at ∼350 K and a disproportionation reaction to form methanol and formaldehyde at ∼500 K. Only 20% of the methoxy species undergo the disproportionation reaction, with most of them reacting via the 350 K pathway.

The crystallization kinetics of transiently heated, nanoscale water films were investigated for 188 K < Tpulse < 230 K, where Tpulse is the maximum temperature obtained during a heat pulse. The water films, which had thicknesses ranging from approximately 15-30 nm, were adsorbed on a Pt(111) single crystal and heated with ∼10 ns laser pulses, which produced heating and cooling rates of ∼109-1010 K/s in the adsorbed water films. Because the ice growth rates have been measured independently, the ice nucleation rates could be determined by modeling the observed crystallization kinetics. The experiments show that the nucleation rate goes through a maximum at T = 216 K ± 4 K, and the rate at the maximum is 1029±1 m-3 s-1. The maximum nucleation rate reported here for flat, thin water films is consistent with recent measurements of the nucleation rate in nanometer-sized water drops at comparable temperatures. However, the nucleation rate drops rapidly at lower temperatures, which is different from the nearly temperature-independent rates observed for the nanometer-sized drops. At T ∼ 189 K, the nucleation rate for the current experiments is a factor of ∼104-5 smaller than the rate at the maximum. The nucleation rate also decreases for Tpulse > 220 K, but the transiently heated water films are not very sensitive to the smaller nucleation rates at higher temperatures. The crystallization kinetics are consistent with a "classical" nucleation and growth mechanism indicating that there is an energetic barrier for deeply supercooled water to convert to ice.

The growth rate of crystalline ice (CI) in amorphous solid water (ASW) films was investigated using reflection absorption infrared spectroscopy. Two different experiments were set up to measure rates of the crystallization front propagation from the underlying crystalline template upward and from the vacuum interface downward. In one set of experiments, layers of ASW (5% D2O in H2O) were grown on a CI template and capped with a decane layer. In isothermal experiments from 140 to 150 K, crystallization was observed from the onset (no induction time) and the extent of crystallization increased linearly with time. In a second set of experiments, uncapped ASW films without a CI template were studied. The films were created by placing a 100 ML isotopic layer (5% D2O in H2O) at various positions in a 1000 ML ASW (H2O) film. The CI growth rates obtained from the two configurations (capped films with a CI template and uncapped films without a CI template) are in quantitative agreement. The results support the idea that for ASW films in a vacuum, a crystalline layer forms at the surface that then acts as a CI template for a growth front that moves downward into the film.

The adsorption of cyclohexane, benzene, and alkyl-substituted benzene derivatives is studied on rutile TiO2(110) by a combination of molecular beam dosing, temperature-programmed desorption, and density functional theory (DFT). An inversion analysis is used to extract the coverage-dependent desorption energies from TiO2(110). The values of the suitable prefactors are derived from simple statistical mechanical models assuming different limits in the adsorbate mobility on the surface. The prefactor values determined using the vibrational frequencies from DFT calculations corroborate this analysis and show that the adsorbates are mobile in one or two dimensions on a corrugated TiO2(110) surface. The adsorption of benzene derivatives is found to be dominated by the dative Lewis acid-base interactions of the pi system with the surface Ti ions. While the desorption energy generally increases with increasing the length and the number of substituents, the difference between the desorption energies decreases as the number and length of substituents are increased. This is a consequence of the destabilization of the optimum bonding configuration of the benzene ring and the alkyl groups with their increasing length and number. The absolute saturation coverages of uncompressed layers correspond approximately to one molecule per three Ti-5c sites and decrease slightly with increasing molecule size, in good agreement with van der Waals sizes of the molecules.

Single-atom catalysis has been a topic of increasing interest due to the potential for improved selectivity, reactivity, and catalyst cost. However, single-atom catalysts are still difficult to characterize under realistic reaction conditions, leading to controversy regarding the capabilities of single atoms and a need for model studies. Herein, we examine the reaction of methanol on single Pd atoms supported on Fe3O4(001) under ultrahigh vacuum conditions. On Pd-free Fe3O4(001), a small fraction of methanol is converted to formaldehyde through a methoxy intermediate at 516 K. The addition of single Pd atoms lowers the barrier to C-H bond cleavage by a factor of 2, resulting in formaldehyde desorption by 290 K. However, Pd atoms begin to sinter by 300 K in the presence of methanol, and Pd clusters do not exhibit the same chemistry. Single atoms significantly lower the barrier to the oxidation of methanol, although their stability remains an issue.

The interaction of carbon dioxide (CO2) with a graphene-covered Pt(111) surface was investigated using temperature-programmed desorption (TPD) and reflection absorption infrared spectroscopy (RAIRS). The TPD spectra show monolayer and multilayer desorption peaks; however, the multilayer peak is not well-separated from the monolayer peak. The TPD spectra for submonolayer and multilayer coverages align on separate common leading edges. This alignment is a signature of zero-order desorption kinetics. The RAIRS spectra for submonolayer coverages have a relatively sharp peak at ∼2350 cm-1, which is assigned to the ν3 asymmetric stretch. The peak is observed at the onset of CO2 adsorption, and the area of the peak increases linearly with coverage. This suggests that CO2 does not lie flat on the surface but instead has a component of its bond axis perpendicular to the graphene surface.

Isotopic exchange reactions in mixed D2O and H2O amorphous solid water (ASW) films were investigated using reflection absorption infrared spectroscopy. Nanoscale films composed of 5% D2O in H2O were deposited on Pt(111) and graphene covered Pt(111) substrates. At 130 K, we find that the reaction is strongly dependent on the substrate with the H/D exchange being significantly more rapid on the Pt(111) surface than on graphene. At 140 K, the films eventually crystallize with the final products on the two substrates being primarily HOD molecule on Pt(111) and a mixture of HOD and unreacted D2O on graphene. We demonstrate by pre-dosing H2 and O2 on Pt(111) that the observed differences in reactivity on the two substrates are likely due to the formation of hydrogen ions at the Pt(111) surface that are not formed on graphene. Once formed the mobile protons move through the ASW overlayer to initiate the H/D exchange reaction.

The desorption of benzene, 1,3,5-trifluorobenzene (TFB), and hexafluorobenzene (HFB) from a graphene covered Pt(111) substrate was investigated using temperature-programmed desorption (TPD). All three species have well-resolved monolayer and second-layer desorption peaks. The desorption spectra for submonolayer coverages of benzene and HFB are consistent with first-order desorption kinetics. In contrast, the submonolayer TPD spectra for TFB align on a common leading-edge, which is indicative of zero-order desorption kinetics. The desorption behavior of the three molecules can be correlated with the strength of the quadrupole moments. Calculations (second-order Møller-Plesset perturbation and density functional theory) show that the potential minimum for coplanar TFB dimers is more than a factor of 2 greater than that for either benzene or HFB dimers. The calculations support the interpretation that benzene and HFB are less likely to form the two-dimensional islands that are needed for submonolayer zero-order desorption kinetics.

The desorption kinetics for benzene and cyclohexane from a graphene covered Pt(111) surface were investigated using temperature-programmed desorption (TPD). The benzene desorption spectra show well-resolved monolayer and multilayer desorption peaks. The benzene monolayer and submonolayer TPD spectra for coverages greater than ∼0.1 ML have nearly the same desorption peak temperature and have line shapes which are consistent with first-order desorption kinetics. For benzene coverages greater than 1 ML, the TPD spectra align on a common leading edge which is consistent with zero-order desorption. An "inversion" procedure in which the prefactor is varied to find the value that best reproduces the entire set of experimental desorption spectra was used to analyze the benzene data. The inversion analysis of the benzene TPD spectra yielded a desorption activation energy of 54 ± 3 kJ/mol with a prefactor of 1017±1 s-1. The TPD spectra for cyclohexane also have well-resolved monolayer and multilayer desorption features. The desorption leading edges for the monolayer and the multilayer TPD spectra are aligned indicating zero-order desorption kinetics in both cases. An Arrhenius analysis of the monolayer cyclohexane TPD spectra yielded a desorption activation energy of 53.5 ± 2 kJ/mol with a prefactor of 1016±1 ML s-1.

The crystallization of amorphous solid water (ASW) nanoscale films was investigated using reflection absorption infrared spectroscopy. Two ASW film configurations were studied. In one case the ASW film was deposited on top of and capped with a decane layer ("sandwich" configuration). In the other case, the ASW film was deposited on top of a decane layer and not capped ("no cap" configuration). Crystallization of ASW films in the "sandwich" configuration is about eight times slower than in the "no cap." Selective placement of an isotopic layer (5% D2O in H2O) at various positions in an ASW (H2O) film was used to determine the crystallization mechanism. In the "sandwich" configuration, the crystallization kinetics were independent of the isotopic layer placement whereas in the "no cap" configuration the closer the isotopic layer was to the vacuum interface, the earlier the isotopic layer crystallized. These results are consistent with a mechanism whereby the decane overlayer suppresses surface nucleation and provide evidence that the observed ASW crystallization in "sandwich" films is the result of uniform bulk nucleation.

We have investigated the nucleation and growth of crystalline ice in 0.24 μm thick, supercooled water films adsorbed on Pt(111). The films were transiently heated with ∼10 ns infrared laser pulses, which produced typical heating and cooling rates of ∼109-1010 K/s. The crystallization of these water films was monitored with infrared spectroscopy. The experimental conditions were chosen to suppress ice nucleation at both the water/metal and water/vacuum interfaces. Furthermore, internal pressure increases due to curvature effects are precluded in these flat films. Therefore, the experiments were sensitive to the homogeneous ice nucleation rate from ∼210 to 225 K. The experiments show that Jmax, the maximum for the homogeneous ice nucleation rate, J(T), needs to be ≥1026 m-3 s-1 and is likely to be ∼1029±2 m-3 s-1. We argue that such large nucleation rates are consistent with experiments on hyperquenched glassy water, which typically have crystalline fractions of ∼1% or more.