Transport in Amorphous Solid Water Films: Implications for Self

R. Scott Smith , Jesper Matthiesen , Jake Knox , and Bruce D. Kay. The Journal of ... Deepanjan Bhattacharya , Candace N. Payne , and Vlad Sadtchenko...
1 downloads 0 Views 2MB Size
Download PDF
J. Phys. Chem. B 2006, 110, 17987-17997

17987

Transport in Amorphous Solid Water Films: Implications for Self-Diffusivity Sean M. McClure, Evan T. Barlow, Minta C. Akin, Douglas J. Safarik, Thomas M. Truskett, and C. Buddie Mullins* Institute for Theoretical Chemistry and Texas Materials Institute, Department of Chemical Engineering, UniVersity of Texas at Austin, 1 UniVersity Station, C0400 Austin, Texas 78712-0321 ReceiVed: May 26, 2006; In Final Form: July 12, 2006

Thermal desorption spectroscopy is employed to examine transport mechanisms in structured, nanoscale films consisting of labeled amorphous solid water (ASW, H218O, H216O) and organic spacer layers (CCl4, CHCl3) prior to ASW crystallization (T ≈ 150-160 K). Self-transport is studied as a function of both the ASW layer and the organic spacer layer film thickness, and the effectiveness of these spacer layers as a bulk diffusion “barrier” is also investigated. Isothermal desorption measurements of structured films are combined with gas uptake measurements (CClF2H) to investigate water self-transport and changes in ASW film morphology during crystallization and annealing. CCl4 desorption is employed as a means to investigate the effects of ASW film thickness and heating schedule on vapor-phase transport. Combined, these results demonstrate that the interlayer mixing observed near T ≈ 150-160 K is inconsistent with a mechanism involving diffusion through a dense phase; rather, we propose that intermixing occurs via vapor-phase transport through an interconnected network of cracks/fractures created within the ASW film during crystallization. Consequently, the self-diffusivity of ASW prior to crystallization (T ≈ 150-160 K) is significantly smaller than that expected for a “fragile” liquid, indicating that water undergoes either a glass transition or a fragile-to-strong transition at a temperature above 160 K.

Introduction Determining the glass transition temperature, Tg, of water and its fragility at low temperatures (T < 230 K) remains key in developing a clear picture of its liquid state.1-5 Progress has been hampered by the difficulty in probing liquid water’s properties above the homogeneous nucleation temperature associated with heating the glass (T ≈ 160 K) and below that of supercooling the liquid (T ≈ 230 K), a virtual “no-man’s land”5 where water crystallizes rapidly on experimental time scales. Below 160 K, transport properties such as viscosity (µ) and self-diffusivity (D), used to characterize fragility in liquids,2-3,6-7 are too small to be probed experimentally using bulk samples. Thus, despite novel experimental and theoretical investigations,1-5 the nature of water below 230 K remains controversial. Water’s Tg is a particular point of contention, with many experimental studies (on various forms of glassy water: hyperquenched glassy water, vapor-deposited amorphous solid water (ASW),8 pressure amorphized water, and confined water) leading to differing conclusions. Some are consistent with Tg ≈ 136 K (calorimetry studies,9-17 blunt probe measurements,18 dielectric studies,19-23 extrapolation of binary solution data,3,24 diffusion studies,25,26 time-of-flight secondary ion mass spectrometry (TOF-SIMS) studies27,28), while others (dielectric studies,29-31 isotope exchange studies,32 differential scanning calorimetry (DSC) studies/scaling arguments24,33-36, soft-landed ions37) suggest Tg > 160-165 K.34,35 In the latter case, observation of Tg would be masked by crystallization upon heating near T ≈ 150 K. While the “conventional” assignment of water’s glass transition temperature is Tg ≈ 136 K, this assignment is not universally accepted. * To whom correspondence [email protected].

should

be

addressed.

E-mail:

If Tg ≈ 136 K, uncertainty still surrounds the fragility of the supercooled liquid prior to crystallization between T ≈ 150160 K. Fragility characterizes the temperature dependencies of relaxation processes (and hence, µ and D) in liquids2,3,6,7 and is believed to be related to the cooperative nature (configurational entropy38) of relaxation processes which occur upon heating and cooling. “Strong” liquids show Arrhenius temperature dependencies, and “fragile” liquids are non-Arrhenius.2,3,6,7 Strong liquids have short- and intermediate-range structures that are relatively insensitive to temperature upon heating through Tg, whereas fragile liquids have structures which quickly disappear above Tg.2,6,7 While supercooled liquid water is known to be one of the most fragile liquids at higher temperatures (T > 230 K) as suggested by dielectric,39 self-diffusivity,40-42 and thermodynamic arguments,43 the nature of ASW at lower temperatures is an open question. Arguments based on DSC13,43-44 and dielectric29,31 measurements, along with crystallization kinetics,45 have been interpreted as evidence that liquid water is “strong” at lower temperatures.24,44 Given liquid water’s known behavior above T ≈ 230 K, this scenario would require a fragile-to-strong transition43 (from high to low temperatures) between T ≈ 230-160 K. While fragile-to-strong behavior is not common for liquids, theoretical studies of water46-55 (and other network forming liquids (SiO2,56 BeF257)), are consistent with transitions from fragile dynamics at higher temperatures to strong dynamics at lower temperatures. Recent neutronscattering58,59 and dielectric60,61 studies of confined water are also consistent with this scenario. Conversely, temperature-programmed desorption (TPD) experiments by Smith and Kay25,26 have detected intermixing in thin, isotopically labeled (H216O, H218O) ASW films near T ≈ 150-160 K. In their study, layered films deposited at 77 K were heated and desorption monitored to measure interlayer

10.1021/jp063259y CCC: $33.50 © 2006 American Chemical Society Published on Web 08/22/2006

17988 J. Phys. Chem. B, Vol. 110, No. 36, 2006 mixing. The intermixing, which appears during crystallization (T ≈ 150-160 K), has been interpreted as bulk diffusion and evidence that ASW behaves as a “fragile” liquid above Tg ≈ 136 K. Simulations by Paschek et al.62 and dielectric studies of sequestered water63 are also consistent with this picture. Here, we focus on translational motion in vapor-deposited ASW with the hope of providing further insights regarding water self-diffusion near T ≈ 150-160 K. Earlier, we presented preliminary results64 indicating that mixing observed during TPD of nanoscale ASW films between (T ≈ 150-157 K) is inconsistent with bulk diffusion. Rather, mixing is primarily due to transport of H2O through an interconnected porous network of fractures created during crystallization. This suggests that self-diffusivity of ASW from 150 to 157 K is smaller than previously thought,25,26 making it unlikely that ASW is a fragile liquid prior to crystallization. Instead, water either remains a (i) glass upon heating to Tg ≈ 160 K or (ii) is a “strong” liquid with Tg ≈ 136 K, undergoing a fragile-to-strong transition between T ≈ 160-230 K in order to connect with liquid water diffusivities40-42 at higher temperatures. The new results presented here provide further experimental support for our earlier conclusions.64 In particular, we present here additional TPD mixing experiments with differing ASW film thicknesses, spacer layer thicknesses, and spacer layer materials (CHCl3 and CCl4). Isothermal desorption experiments are also presented, enabling comparison of self-transport with surface area measurements obtained from separate gas uptake measurements. These measurements provide an independent means (besides abrupt CCl4 desorption from beneath ASW65) of probing crystallizationinduced changes in ASW film morphology. Methanol is employed as a “probe” molecule to test the effectiveness of CCl4 diffusion “barrier” layers used in our experiments. Additionally, we attempt to investigate the effects of heating schedule and ASW film thickness on vapor phase transport, using abrupt CCl4 desorption65 from beneath the ASW film as a probe of fracture. These CCl4 desorption results are related qualitatively to the changes observed in H2O self-mixing as ASW film thickness and heating schedules are varied. Experimental Section Experiments were conducted in an ultrahigh vacuum (UHV)/ molecular beam apparatus containing two molecular beam lines, as described in detail previously.66 One set of molecular beam apertures (beam line B) creates a beam spot larger than the Ir(111) single-crystal substrate (a circular disk ∼9.1 mm in diameter, ∼1 mm thick) employed for deposition, such that the entire substrate is in the “umbra” region67 ensuring a uniform flux of molecules over the whole surface. The second molecular beam line (beam line A) creates a noncircular spot (diamond shaped; covering ∼70% of the surface) contained entirely within the sample area and allowing for growth of a small “pill” layer of molecules as well as minimizing exposure to other surfaces in the UHV chamber. Figure schematics indicate the beam line(s) used for each experiment. The UHV/surface analysis chamber (Pbase ≈ 1 × 10-10 Torr) is equipped with a quadrupole mass spectrometer (QMS), Auger electron spectroscopy (AES), and low-energy electron diffraction (LEED) optics. The substrate can be cooled to T ≈ 77 K via contact with a liquid nitrogen reservoir and resistively heated to T ≈ 1550 K. The ionization region of the QMS (∼9 in. from the sample) is unshielded. Films of ASW and organic layers are grown and desorbed on the sample side facing away from the QMS ionizer; hence, the collected TPD spectra are angle-integrated. The UHV scattering

McClure et al. section is pumped by a diffusion pump equipped with a liquid nitrogen cooled cryotrap with an H2O pumping speed measured as ∼750 L/s. The Ir(111) substrate was initially cleaned and ordered via repeated Ar ion sputtering (1 keV; PAr ≈ 1 × 10-5 Torr), O2 anneal cycles (600 K, PO2 ≈ 5 × 10-7 Torr), and in vacuo anneal cycles (1400 K), as verified via LEED and AES. Further, the Ir(111) substrate was cleaned prior to each experiment with O2 adsorption/desorption cycles (and annealing cycles to order the substrate when necessary). Nanoscale structures of labeled (H218O/H216O) ASW films were grown on the substrate at 77 K at normal incidence using room temperature, quasi-effusive molecular beams of pure water vapor. These growth conditions have been shown to produce smooth, dense, nonporous films of ASW.68-72 ASW (H216O/ H218O) films were always grown with a beam flux of ∼0.17 bilayers/s (BL/s) (One BLH2O ≈ 1 × 10-15 molecules/cm2)73,74 employing “beam line B”, which provides a uniform flux across the substrate. TPD spectra have been corrected for mass fragmentation occurring during ionization in the QMS. Fragmentation patterns were determined by characterization of pure H216O and H218O beams using the QMS. Molecular beams of water were formed from distilled, deionized H216O and isotopically labeled H218O (Isotec, 95-98% 18O atom purity), which were thoroughly degassed prior to use. H2O dose rates were calibrated via the reflectivity technique of King and Wells.75 Films of carbon tetrachloride (CCl4) (beam flux ≈ 0.04 ML/ s), chloroform (CHCl3) (beam flux ≈ 0.06 ML/s), and methanol (MeOH) (beam flux ≈ 0.07 ML/s) were also utilized in these studies. Beam fluxes were calibrated by determining the exposure necessary to saturate the substrate (at T ≈ 140 K for CCl4 and CHCl3; T ≈ 150 K for MeOH) via the method of King and Wells.75 CHCl3, CCl4, and MeOH were thoroughly degassed prior to use without further purification. All CCl4, CHCl3, and MeOH films were deposited at normal incidence at 77 K. In all experiments, beam line A is used to deposit MeOH. Both beam lines A and B are used to deliver CCl4 to the sample, depending on the particular experiment. Figure captions and schematics clarify which beam lines are used for each experiment. Thermal desorption of chlorodifluoromethane, CClF2H, from ASW films can be used as a probe to monitor crystallization kinetics and surface area changes (due to surface roughening and porosity formation)76,77 during the transformation of ASW to crystalline ice (CI). Briefly, a solid water film is heated and held at T ≈ 86.5 K (temperature at which CClF2H multilayers are rapidly desorbed), and CClF2H is dosed via beam line A until the monolayer is saturated, as determined via the King and Wells technique.75 CClF2H TPD spectra are subsequently obtained, and the integrated TPD area determined. By repeating experiments for several isothermal anneal times changes in the relatiVe surface area during ASW crystallization can be monitored. [Surface area relative to the “standard” surface area of an ASW film annealed to 115 K,76,77 i.e., SArel ) (amount CClF2H uptake on annealed ASW film)/(amount of CClF2H uptake on an ASW film annealed to 115 K).] Finally, a simple 1-D bulk diffusion-desorption model is employed which numerically solves the diffusion equation taking into account ASW crystallization,78 ASW and CI desorption parameters,79 and extrapolated CI diffusion parameters.80 Diffusivities [DASW(T), DCI(T)] are modeled with an Arrhenius temperature dependence [D ≈ D0 exp (-Ea/RT)]. The overall diffusivity of the film is estimated to be a crystallized fraction-weighted average of crystalline and amorphous diffusivity values at a given temperature [Doverall(T) ) DASW(T)(1 -

Transport in ASW Films

Figure 1. ASW TPD mixing experiment and desorption ratio. Shown in Figure 1a is a TPD mixing experiment of a structured film composed of labeled ASW (H218O, H216O). The sample was constructed (see schematic) by first depositing 16 BL H218O (blue), followed by deposition of 16 BL H216O ASW (black), and finally heating at a rate of 0.6 K/s. Shown in Figure 1b is the desorption ratio (r18/r16) trace (open red circles) from the ASW TPD Mixing experiment shown in Figure 1a. Additionally, a series of desorption ratio traces (solid lines) calculated from a simple TPD desorption/diffusion model (see Experimental section) to illustrate mixing behavior observed in a bulk diffusion mechanism. The series of model desorption traces span a wide range of ASW bulk diffusion values (“high” diffusivity [Ea ) 220 kJ/mol; D0 ) 1.28 × 1062 cm2/s] to “low” diffusivity [Ea ) 120 kJ/mol; D0 ) 6.42 × 1024 cm2/s]). For example, these parameters produce ASW diffusion coefficients at T ≈ 155 K of DASW(155 K) ≈ 9 × 10-13 cm2/s (high diffusivity) and DASW(155 K) ≈ 2 × 10-16 cm2/s (low diffusivity). Inset: Example of CCl4 “molecular volcano65” experiment. Inset to Figure 1a shows an example of a CCl4 “molecular volcano65” experiment conducted in our laboratory. Spectra (i) shows a TPD spectra of a layered film (see schematic) constructed by first depositing 30 BL H216O ASW (black), followed by deposition of a ∼6 ML “pill” of CCl4 (green dashed). Spectra (ii) shows a TPD spectra of a sample constructed by first depositing a ∼6 ML CCl4 “pill” (purple dashed) (signal × 0.5), followed by deposition of a 30 BL H216O ASW layer (black). The TPD ramp rate for each spectra was 0.6 K/s. In depositing the ASW films (H216O), beam line B was employed to provided uniform coverage across the Ir(111) sample; for CCl4, beam line A was employed to deposit a small CCl4 ‘pill’ on the sample.

X(T))+DCI(T)X(T)].25,26 This model allows qualitative comparisons between experimental mixing behavior and that expected if mixing were due to bulk diffusion between two dense phases. Results and Discussion ASW Mixing Experiments. Shown in Figure 1a are TPD spectra of layered, labeled ASW (H218O/H216O) films similar to those shown previously64 except here the films are thinner. The sample shown here was constructed by first depositing 16 BL H218O ASW, followed by 16 BL H216O ASW in order to create a sample identical to that studied by Smith and Kay.26 This film was then heated at 0.6 K/s, and as the spectra illustrate, water begins to desorb near T ≈ 140 K and interlayer mixing (appearance of H218O) occurs between T ≈ 150-157 K (crystallization also occurs over this temperature range). The results shown in Figure 1a are in quantitative agreement with the data of Smith and Kay26 who interpreted the intermixing as due to “liquidlike” bulk diffusion between the ASW films prior to crystallization. Bulk diffusion after the film has crystallized (T > 160 K) is expected to be negligible based on estimates of crystalline ice self-diffusivity.73,80 Closer analysis of the intermixing between 150 and 160 K, as shown in Figure 1b, reveals that neither our data nor that of previous investigators25,26 can be explained via a bulk diffusion mechanism alone. This figure

J. Phys. Chem. B, Vol. 110, No. 36, 2006 17989 shows the ratio of desorption rates of H218O and H216O, “r18/ r16” [r18/r16 ) (desorption rate of H218O)/(desorption rate of H216O)], determined from the TPD data of Figure 1a, and a family of desorption ratios (solid lines) calculated from our simple diffusion model using various bulk diffusion parameters for ASW. The quantity r18/r16 can be viewed as an instantaneous measure of the relative surface concentrations of the labeled water molecules. If bulk diffusion were the dominant mode of transport, r18/r16 would, by necessity, increase in a monotonic fashion upon heating (as illustrated by the model calculations shown in Figure 1b), reflecting the increase in the H218O surface concentration under such conditions. The family of model calculation curves in Figure 1b shows this characteristic behavior regardless of the magnitude of the self-diffusion coefficient used. In contrast, the experimental desorption ratio r18/r16 (open circles) increases sharply at T ≈ 153-157 K, peaks at T ≈ 157 K, decreases, and then increases (slightly) prior to complete desorption of the water film at T ≈ 166 K. Hence, the observed intermixing appears to be inconsistent with a bulk diffusion mechanism, suggesting that another mode of transport is at play. Fracture Formation in ASW Films. As mentioned earlier, nanoscale ASW films are known to fracture during crystallization. This can be observed via the novel CCl4 “molecular volcano” experiment of Smith et al.65 as shown in the Figure 1a inset: in (i) 30 BL H216O ASW is dosed onto the substrate at 77 K, followed by a ∼6 ML CCl4 “pill” (beam line A). Upon heating, CCl4 multilayers desorb from T ≈ 120-142 K, followed by desorption of the H2O. Here, crystallization of the ASW film can be visualized (as “bump”) in the water desorption spectra. Since ASW has a higher desorption rate than CI,79,81 a decrease in the desorption rate (bump) is observed (T ≈ 154155 K) as the film is crystallized. In Figure 1a inset (ii), the order of CCl4 and ASW deposition has been reversed; first a ∼6 ML CCl4 “pill” is deposited followed by growth of the ASW overlayer. As the film is heated, CCl4 remains trapped below the ASW overlayer until the water film begins to crystallize and fracture.65 Once complete, an interconnected pathway of fractures has been created from the CCl4 underlayer to the top of the ASW film, CCl4 abruptly escapes [“molecular volcano65”] since at these temperatures it is quite volatile. The two water desorption spectra shown in insets i and ii of Figure 1a are virtually identical, regardless of CCl4 placement. A variety of molecules (CCl4, O2, N2, CH4, Ar)65,82,83 trapped beneath ASW have been shown to abruptly escape upon crystallization (and fracture) illustrating the ability of these pathways to provide a means of vapor phase transport for desorbing species of various character. There is additional evidence suggesting that ASW films undergo structural changes upon crystallization. Microscopy studies (AFM,84 TEM85) demonstrate glassy water films undergo changes in morphology upon annealing and crystallization. Gas uptake measurements performed on thin, annealed ASW films also suggest roughening and fracture occur during crystallization.77,78 Increases in diffuse reflectance in optical interferometry measurements have been observed during crystallization of ASW films, consistent with fracturing.86,87 Recently, Souda et al.27,28 has conducted TOF-SIMS measurements on thin (∼50 BL) ASW films, which suggest a morphological change upon heating between T ∼ 135-160 K and interpreted this as due to dewetting. Studies by Kimmel et al.88 also suggest that nanoscale ASW films dewet when annealed and crystallized at higher temperatures (T > 140 K), presumably due to the hydrophobic nature of the metal-bound, H2O monolayer. How this dewetting phenomena and film

17990 J. Phys. Chem. B, Vol. 110, No. 36, 2006

McClure et al.

Figure 2. ASW TPD mixing experiments: 30 BL ASW thicknesses. ASW TPD mixing experiments with different structured films (see schematics) composed of labeled ASW and CCl4 are displayed. For Figure 2a, 30 BL of H218O (blue) deposited first followed by 30 BL of H216O (black); Figure 2b, 30 BL of H218O (blue) deposited first, then 5 ML of CCl4 (red), and finally 30 BL of H216O (black); Figure 2c, 30 BL of H218O (blue) deposited first, followed by 30 ML of CCl4 (red), and then 30 BL of H216O (black). For films in parts a-c of Figure 2, molecular beam line B was employed to provided uniform films of H218O, H216O, and CCl4 to the Ir(111) sample. All TPD spectra collected at a temperature ramp rate of 0.6 K/s. Inset: H218O/H216O desorption ratio. Shown in the Figure 2a inset is the desorption ratio (r18/r16) trace (open circles) from the ASW TPD mixing experiment shown in Figure 2a. Additionally, a series of desorption ratio traces (solid lines) calculated from a simple TPD desorption/diffusion model to illustrate mixing behavior observed in a bulk diffusion mechanism. The series of model desorption traces span a range of bulk diffusion parameters (“high” diffusivity [Ea ) 220 kJ/mol; D0 ) 1.28 × 1062 cm2/s] to “low” diffusivity [Ea ) 120 kJ/mol; D0 ) 6.42 × 1024 cm2/s]). For example, these parameters produce ASW diffusion coefficients at T ≈ 155 K of DASW(155 K) ≈ 9 × 10-13 cm2/s (high diffusivity) and DASW(155 K) ≈ 2 × 10-16 cm2/s (low diffusivity).

fracture are related to one another remains an open question. Our focus here is examining the implications of these crystallization-induced transport pathways with regard to the observed water transport occurring near crystallization. Inspection of Figure 1 reveals that porosity formation is concurrent with the onset of isotopic mixing, suggesting a relationship between these two phenomena. Water has an appreciable desorption rate [∼0.5 BL/s (see Figure 1a inset)] during film fracture (T ≈ 154-155 K); thus, like CCl4, underlying H2O molecules may be transported through the porous network. The remainder of the paper will focus on experimental results that strongly suggest that porosity-mediated transport, not bulk diffusion through a dense phase, is the predominant mode of transport in nanoscale ASW films between T ≈ 150-160 K. CCl4 Diffusion Barrier Experiments. Shown in parts a-c of Figure 2 are spectra from structured films similar to Figure 1a, constructed from labeled ASW (H216O, H218O); however, various CCl4 diffusion “barrier” layers (0, 5, and 30 ML) have been placed between the water layers. These hydrophobic, immiscible CCl4 layers serve two purposes: (1) to provide a hindrance to bulk diffusion between the two ASW layers prior to porosity formation in the film and (2) to serve as a “marker” for the onset of porosity formation within ASW. Figure 2a shows TPD spectra of a structured film prepared by deposition of 30 BL H218O ASW followed by 30 BL H216O ASW. The sample is then heated at 0.6 K/s, resulting in intermixing behavior similar to that shown in Figure 1a (16 BL H216O on 16 BL H218O), i.e., the onset of intermixing occurs during crystallization of the ASW film (T ≈ 154-160 K). Shown as an inset to Figure 2a is a plot of the desorption ratio (r18/r16) of the same experiment (open circles), illustrating nonmonotonic behavior which is inconsistent with a bulk diffusion mechanism (solid lines). Figure 2b, constructed by depositing 30 BL H218O, followed by 5 ML CCl4, followed by 30 BL H216O, exhibits remarkably similar behavior to Figure 2a with no CCl4 barrier layer. Likewise, Figure 2c, an equivalent TPD experiment with a much thicker CCl4 barrier layer (30 ML) exhibits mixing similar to parts a and b of Figure 2 (with only slight differences after 157 K). The similarity between the

interlayer mixing exhibited in parts a-c of Figures 2 is inconsistent with bulk diffusion since the CCl4 spacer layer should noticeably hinder mixing of the labeled layers. Thus, self-transport of H2O via bulk diffusion prior to desorption of the CCl4 layer (154-155 K) must be very small on these time (seconds) and length (10-100 nm) scales. Although these data (Figure 2) have been presented previously,64 we include these data and discussion for direct comparison with our new results, which were conducted on comparably structured ASW films. Shown in parts a-c of Figure 3 are TPD mixing experiments, similar to Figure 2, of structured films of labeled ASW (H216O, H218O) of 16 BL thickness, separated by varying amounts of CCl4 (0, 5, and 30 ML, respectively) which we expect to hinder bulk diffusion of ASW. Again, the results demonstrate remarkably similar mixing between TPD spectra of structures containing no barrier [Figure 3a] and structures containing CCl4 layers [parts b and c of Figure 3]. Similar to the results of parts a-c of Figure 2, slight differences are observed (after T ≈ 155 K) when a barrier with a thickness of 30 ML CCl4 is present. Additional experiments using a different hydrophobic layer (CHCl3) were conducted to test the effect on intermixing. If H218O could readily diffuse through these layers, one might expect different mixing depending on barrier material. CHCl3 exhibits properties similar to CCl4, i.e., it remains “trapped” below ASW overlayers prior to crystallization. Parts a-c of Figure S1 shows mixing experiments, similar to those of parts a-c of Figure 2, except with layers of CHCl3 (5-30 ML). Similar ASW mixing behavior is observed suggesting that the results are not highly specific to barrier material. This is consistent with negligible bulk diffusion occurring between the labeled ASW layers prior to film fracture. TPD Experiments with Different Temperature Ramp Rates. Changing the TPD temperature ramp rate enables alteration of the time scale over which ASW films crystallize (since crystallization is a function of both time and temperature), and thus, alteration of the onset of porosity within ASW films also occurs. Parts a-c of Figure S2 show TPD mixing spectra of identical films composed by first depositing 60 BL H218O, then a ∼7 ML CCl4 “pill”, followed by 60 BL H216O heated at different rates (0.05, 0.6, and 2 K/s). The ASW films employed

Transport in ASW Films

J. Phys. Chem. B, Vol. 110, No. 36, 2006 17991

Figure 3. ASW TPD mixing experiments: 16 BL ASW thicknesses. ASW TPD mixing experiments with different structured films (see schematics) composed of labeled ASW and CCl4 are displayed in Figure 3. For Figure 3a 16 BL of H218O (blue) deposited first followed by 16 BL of H216O (black); Figure 3b, 16 BL of H218O (blue) deposited first, then 5 ML of CCl4 (red), and finally 16 BL of H216O (black); Figure 3c, 16 BL of H218O (blue) deposited first, followed by 30 ML of CCl4 (red; signal × 0.3) and then 16 BL of H216O (black). For films in parts a-c of Figure 3, molecular beam line B was employed to deposit uniform films of H218O, H216O, and CCl4 across the entire Ir(111) sample. All TPD spectra collected at a temperature ramp rate of 0.6 K/s.

here are twice as thick as those previously presented64 and thus exhibit slightly different H2O mixing behavior and CCl4 desorption temperatures. The CCl4 “pill” between the ASW layers serves as a marker for the onset of crystallization induced porosity. As parts a-c of Figure S2 illustrates, varying the TPD ramp rate changes the abrupt CCl4 desorption temperature from T ≈ 149 K at a ramp rate of 0.05 K/s to T ≈ 160 K at 2 K/s. If transport of water within these films is occurring via porosity, altering the TPD ramp rate should shift the onset of water mixing similarly. Indeed, this is observed in the spectra of parts a-c of Figure S2, with substantial mixing at all ramp rates, and with the onset of H218O desorption occurring during porosity formation. If ASW were a fragile liquid with a Tg ≈ 136 K prior to crystallization, its self-diffusivity will likely be highly activated near T ≈ 150-160 K, to maintain a smooth connection with known fragile behavior40-42 at higher temperatures (T > 230 K). To illustrate, the fragile ASW bulk diffusion parameters calculated by Smith and Kay25,26 predict that the bulk diffusion coefficient D of ASW should change by over 3 orders of magnitude from T ≈ 149 K to T ≈ 160 K. A simple calculation indicates that altering the TPD ramp rate of mixing experiments in this manner would change the effective length scale (L) of mixing due to bulk diffusion by roughly an order of magnitude (using D ≈ L2/t; where t is time). If the observed mixing were solely due to bulk diffusion, we would expect (and observe in model calculations not shown) striking differences in the experimentally observed intermixing as a function of TPD ramp rate. However, parts a-c of Figure S2 show similarly extensive mixing for each ramp rate; behavior consistent with porositymediated transport. Analysis of labeled mixing experiments with ASW layers of different thicknesses (30 BL on 30 BL,64 100 BL on 100 BL) with varying TPD ramp rates show similar qualitative behavior to those of parts a-c of Figure S2, i.e., intermixing concurrent with film fracture. Shown in parts a-c of Figure 4 are TPD measurements conducted with ASW film thicknesses of 30, 60, and 100 BL, conducted at the single ramp rate 0.6 K/s. As shown, the onset of mixing occurs during crystallization concurrent with the onset of film fracture as evidenced by abrupt CCl4 desorption. Desorption ratio (r18/r16) behavior of these experiments is qualitatively similar (nonmonotonic) to that displayed in the insets of Figures 1b and 2a. However, the interlayer mixing observed later in the TPD spectra of these thicker films appears to be less complete than in thinner films, in agreement with previous investigators.25,26 Closer inspection of parts a-c of Figure 4 shows that the time for CCl4 desorption (porosity formation) is slightly delayed for thicker films (from

Figure 4. TPD mixing experiments with varying ASW thickness. Parts a-c of Figure 4 show ASW TPD mixing experiments with varying thickness of ASW layers (30 BL on 30 BL, 60 BL on 60 BL, 100 BL on 100 BL, respectively). Each film was constructed by first dosing H218O ASW (using beam line B), followed by deposition of a ∼6 ML CCl4 “pill” (using beam line A) followed by dosing of H216O ASW (using beam line B, see figure schematic). All TPD spectra collected at a temperature ramp rate of 0.6 K/s.

T ≈ 155 K to T ≈ 158 K), suggesting an apparent thickness dependence for porosity mediated transport. As discussed later, film thickness is an important variable with regard to porositymediated transport in thin ASW films, as shown by CCl4 volcano desorption65 experiments.

17992 J. Phys. Chem. B, Vol. 110, No. 36, 2006

Figure 5. Isothermal Desorption Pure and Structured ASW Films. Figure 5b shows isothermal desorption of 60 BL H216O ASW film (black). The heating schedule shown in Figure 5a was used to heat the ASW film to T ≈ 150 K (heating between t ) 0-148 s; ramp rate ≈ 0.5 K/s) and held there while desorbing species were monitored. Displayed in Figure 5c are relative surface area measurements (solid circles) obtained from 60 BL H216O films annealed at T ≈ 146 K. Shown in Figure 5d is an isothermal desorption spectra of a structured film, prepared by deposition of 30 BL H218O ASW (blue), followed by deposition of ∼6 ML “pill” of CCl4 (red; signal × 0.25), and finally deposition of 30 BL H216O ASW (black). This structured film was heated to T ≈ 150 K with an identical heating schedule as Figure 5b, thus allowing for direct comparison of the two spectra. ASW films (H218O, H216O) deposited using beam line B; for CCl4, beam line A was employed to deposit a small CCl4 “pill” on the sample.

Isothermal ASW Mixing Experiments and Surface Area Adsorption Measurements. Gas-uptake measurements serve as a useful probe of the increases in apparent surface area due to crystallization-induced fracture in ASW. Previous uptake measurements on thicker glassy water films (CClF2H on 1501050 BL ASW films77 and N2 on 150 BL ASW films78) have displayed increases in apparent surface area upon crystallization. Uptake and desorption measurements of CClF2H monolayers76,77 reveal information regarding the phase (ASW,CI) and relative surface area changes of ASW films as they are crystallized. Here CClF2H gas uptake measurements are utilized to investigate changes in surface area as films are annealed isothermally (T between 146 and 154 K). Combining these measurements with isothermal mixing experiments of layered ASW films enables comparison of mixing behavior with the changing surface area. If transport is linked to film fracture, increases in film surface area should be observed coincident with the onset of interlayer mixing. Before discussing the surface area/isothermal mixing measurements, we first examine isothermal desorption features of a pure ASW film. Shown in Figure 5 are two isothermal desorption (T ≈ 150 K; desorption rate vs time) experiments;

McClure et al. (1) desorption of a 60 BL H216O ASW film (Figure 5b) and (2) desorption of a structured film (Figure 5d), constructed by dosing 30 BL H218O ASW, followed by a ∼6 ML CCl4 “pill”, and then 30 BL H216O ASW. Once deposited, the film is heated and held at T ≈ 150 K according to the schedule shown in Figure 5a and desorbing species are monitored via the QMS. In the case of Figure 5b, when the 60 BL H216O film reaches T ≈ 150 K (around t ≈ 148 s) it is desorbing as ASW (desorption rate ≈ 0.18 BL/s). During the anneal at T ≈ 150 K, the ASW begins to convert to CI (which has a lower desorption rate than ASW79,81) and is manifested in the decrease in desorption rate between t ≈ 152 s and t ≈ 182 s. This decrease in rate is the same behavior that gives rise to the “bump” in the TPD experiment77,89,90 of the Figure 1a inset. Once the film is completely transformed, it proceeds to desorb as CI (desorption rate of ∼0.10 BL/s) until the entire multilayer film is desorbed (t ≈ 750 s). Shown in Figure 5c are relative surface area measurements of a 60 BL ASW film (such as that shown in Figure 5b) as it is annealed at T ≈ 150 K. The heating schedule is that of Figure 5a; hence, these results can be compared directly to those of Figure 5b. As the data illustrate, the relative surface area of the film increases as the ASW begins to crystallize; after crystallization, decreases in surface area are observed upon further annealing. This behavior suggests the crystallization induced cracks are dynamic in nature, opening and sintering closed during annealing.76,77 Displayed in Figure 5d is the isothermal desorption spectra of the layered ASW film. This sample has been heated to T ≈ 150 K identically to Figure 5b and since the films have the same total thickness (60 BL), the two desorption spectra and the surface area measurements of Figure 5c can be compared. Readily apparent are the coincidence of (i) intermixing between the isotopically labeled ASW layers, (ii) the onset of porosity (evidenced by CCl4 desorption), and (iii) the onset of apparent surface area increases with crystallization. Parts a-c of Figure 6 show isothermal desorption measurements conducted at 146, 152, and 154 K, respectively. Displayed above each spectra are relative surface area measurements obtained from separate CClF2H desorption measurements conducted on 60 BL ASW (H216O) films. As the spectra illustrate, the mixing is similar to those seen in TPD experiments, namely, the underlying H218O layer undergoes significant desorption during crystallization, and hence during fracturing. This behavior is apparent over the T ≈ 146-154 K temperature range of the experiments. (Note: multilayer desorption times vary for isothermal experiments since desorption rates will be different at each temperature.) Similar to the behavior of Figure 5c, the relative surface area (for all anneal temperatures T ≈ 146-154 K) is observed to increase during crystallization, then decrease after crystallization as the transformed film is further annealed. This sintering process may play a key role in the mixing behavior observed in our TPD spectra, eliminating these porous pathways from contributing to further mixing after crystallization and possibly trapping water molecules within the densifying film. CCl4 Barrier Layer Effectiveness: Methanol Probe Experiments. CCl4 diffusion barrier experiments (Figures 2-3) suggest that negligible bulk diffusion is occurring within ASW films prior to the onset of crystallization-induced fracture. The presence of a hydrophobic, immiscible CCl4 barrier layer (covering the entire H216O/H218O interface) would be expected to hinder bulk diffusion between labeled ASW layers prior to crystallization-induced film fracture. The additional spacing

Transport in ASW Films

J. Phys. Chem. B, Vol. 110, No. 36, 2006 17993

Figure 7. CCl4 desorption experiments: increasing ASW overlayer thickness. Shown in parts a-f of Figure 7 are TPD spectra of TPD desorption experiments of films constructed by first depositing ∼6 ML CCl4 “pill” (black) on the Ir(111) surface, followed by deposition of varying amounts (15-500 BL) of ASW H216O. All TPD spectra were conducted at a temperature ramp rate of 0.6 K/s. H2O desorption spectra are not shown in the figure due to their large scale. Inset: The inset to Figure 7 is a plot of “trapped CCl4” as a function of ASW film thickness. ASW films (H216O) deposited using beam line B; for CCl4, beam line A was employed to deposit a small CCl4 “pill” on the sample.

Figure 6. Isothermal ASW mixing experiments with CClF2H Uptake Measurements. Shown in parts a-c of Figure 6 are isothermal anneals (146, 152, and 154 K, respectively) of structured ASW films grown by dosing 30 BL H218O ASW (blue), followed by 30 BL H216O ASW (black) (see figure schematics). Films are then heated (using heating schedules similar to Figure 5a) to the desired anneal temperature and the films are allowed to desorb. Shown above parts a-c of Figure 6 are relative surface area measurements (SArel) (solid circles) obtained from 60 BL H216O ASW films annealed to 146, 152, and 154 K. Black dotted lines shown merely to guide the eye.

between the ASW films should also contribute to this hindering effect. However, issues such as CCl4 barrier layer roughness and wetability (how uniformly CCl4 covers the entire H216O/ H218O interface) will influence the barrier effectiveness and thus the validity of our broader conclusions. We have attempted to test the effectiveness of the CCl4 barrier layer using methanol (MeOH) as a “probe” molecule. The experimental results of this study (Figures S3 and S4) and accompanying discussion are included as online Supporting Information. Consistent with previous studies,28 we find MeOH to readily diffuse through dense ASW films (Figure S3). By the study of MeOH mobility in structured nanoscale films of ASW and CCl4, it appears that CCl4 layers do indeed hinder transport of MeOH. In fact, CCl4 barrier layers with thicknesses exceeding 15 ML appear to completely block MeOH transport prior to film fracture (Fig

S4) suggesting that they might also hinder interlayer diffusion of water, a molecule that could be considered similar to MeOH. Transport through Fractures in ASW. Transport of vaporphase water through the interconnected network created in crystallizing ASW films is a function of crack/fracture propagation and sintering kinetics as well as the desorption rate of water. Additionally, transport in porous media is often a complicated combination of Knudsen diffusion, viscous flow, ordinary diffusion, and surface diffusion.91,92 Parameters key to defining the relevant transport conditions within a porous ASW film, such as pore size distributions, geometry, and the nature of the vapor phase within the pores (viscous, molecular flow) are not known. Thus, developing a complete, quantitatiVe model of porous transport within crystallizing ASW films remains a formidable task. Despite these difficulties, we can obtain important qualitatiVe information regarding porosity creation and transport during ASW crystallization. Here we probe the effects of film thickness and heating schedule on fracture and transport in ASW using CCl4 as a “marker” for porosity formation. ASW film thickness plays a role in the transport behavior observed in labeled, structured films (e.g., parts a-c of Figure 4) with overall mixing becoming less complete with increasing thickness. Shown in Figure 7 are measurements of abrupt CCl4 desorption (∼6 ML of CCl4) from ASW overlayers with increasing thickness. As the overlying ASW film thickness is increased from 15 to 90 BL, abrupt CCl4 desorption is shifted

17994 J. Phys. Chem. B, Vol. 110, No. 36, 2006 from T ≈ 155 K to T ≈ 158 K. This is consistent with previous observations of increases in abrupt CCl4 desorption temperature with increasing D2O ASW overlayer thickness (from 30 to 60 BL).65 In parts a-c of Figure 4, this same delay in the abrupt CCl4 desorption (and onset of isotopic mixing) is observed as the thickness of the film is increased. As the ASW overlayer thickness is further increased to 180-500 BL, a second CCl4 desorption feature becomes apparent. This feature appears near the completion of ASW overlayer desorption and is due to CCl4 “trapped” below the crystallized ASW, either unable to escape through cracks prior to sintering (due to increased pore length) and/or that are not sufficiently connected to the ASW surface. Figure 7 shows the portion of “trapped” CCl4 as a function of ASW film thickness; complete “trapping” of the CCl4 occurs at a thickness of ∼500 BL. These results give an estimate of the ASW film thickness (360-500 BL) over which crack/ fracture in ASW is relevant with regard to CCl4 transport. All mixing experiments conducted in our study have a thickness less than 360 BL; hence, we expect at least some fracture pathways to span the entire film thickness. The decrease in apparent mixing we observe in thicker films (parts a-c of Figure 4) is consistent with the findings of Figure 7; as film thickness is increased, interlayer mixing is lessened, presumably by the reduction in the number of cracks/fractures that span the entire water film and/or kinetic competition with film densification. The delay in the first CCl4 desorption feature (concurrent with crystallization) with increasing overlayer thickness can be related to a number of factors. As shown in previous studies,90 for thin ASW films ( 160 K; Figures 2-3). This closing of fracture pathways may give rise to the nonmonotonic ratio (r18/ r16) behavior (Figure 2 inset) observed near crystallization. Surface diffusion occurring along pore surfaces likely only plays a minor role in interlayer mixing. While recent measurements93 of vertical H2O/D2O diffusion near ASW surfaces (T ≈ 100-140 K, 1-5 BL thick films) by Kang et al. suggest that water diffusivity is more rapid near ASW surfaces than in the bulk, extrapolations of these diffusivities to temperatures relevant to our experiments yield diffusion coefficients (Dsurface,equiv(155 K) ≈ 2.5 × 10-16 cm2/s93) too small to describe our observed mixing. Additionally, the lifetime of a water

Transport in ASW Films molecule on the pore surface at relevant temperatures (ASW desorption rate at 155 K ≈ 0.5 BL/s) is very short. Implications for the Glass Transition and Water Fragility. Our data suggest that intermixing in ASW films near crystallization is largely driven by transport via crystallization induced fracture pathways, suggesting that bulk self-diffusion in ASW is lower than previously thought.25,26 Figure 9 shows bulk diffusion coefficients for supercooled liquid water40-42 (open squares), estimates of ASW self-diffusivity (Ea ) 170 kJ/mol; D0 ) 2.9 × 1043 cm2/s, calculated neglecting any porosity mediated transport) obtained by Smith et al.25,26 (open circles), and some currently debated proposals ((i) fragile-to-strong transition with decreasing temperature (green dashed line); (ii) Tg g 160 K (dotted line)) regarding water’s diffusivity. As shown by Smith et al., the VFT equation (using one set of E, T0 parameters) can describe both higher-temperature supercooled liquid water diffusivities and their estimates of ASW selfdiffusivity25,26 near T ≈ 150-157 K. This has been cited as evidence for fragility between T ≈ 150-157 K and a Tg near 136 K.25,26 (The empirical VFT equation [D ) D0 exp(E/(T T0))] can often describe the temperature dependence of diffusion in fragile supercooled liquids.2) The characteristic relaxation time at the glass transition is typically defined to be near τ ≈ 100 s.2,7 Thus, for molecular scale mobility [translation on the order of angstroms (1 Å ) 10-8 cm)] corresponding to this relaxation time, one expects water’s diffusivity (D ≈ L2/ τ) to lie near D ≈ 10-18 cm2/s at Tg. Thus, ASW’s Tg (whether (i) Tg ≈ 136 K (solid triangle) or (ii) Tg g 160 K (solid circle)) and the supercooled liquid water diffusivity data40-42 can provide constraints for ASW diffusivity between T ≈ 160-230 K. In the case of (ii) (Tg > 160 K), diffusivity values near T ≈ 150-160 K would be too small to contribute to mixing on the length and time scales of our TPD experiments. This scenario can be visualized by the dotted line in Figure 9a, which shows a VFT equation fit to higher temperature supercooled liquid data and D ) 10-18 cm2/s at Tg ) 160 K. The curve shows that if Tg g 160 K, diffusivities of ASW would be many orders of magnitude smaller than those expected for a liquid between T ≈ 150-160 K. This picture is one possibility consistent with our measurements. In the case of scenario (i) (Tg ≈ 136 K), the picture is more complicated. In theory, the ASW self-diffusivity can take on a range of values near T ≈ 150-160 K consistent with D ) 10-18 cm2/s at Tg ≈ 136 K. While some of these self-diffusivity values would be of sufficient magnitude to contribute to bulk diffusive intermixing on the length and time scales of our experiments (e.g., values estimated in ref 25), smaller values would prohibit measurable bulk diffusion (as is suggested by Figures 2, 3, and S1). Despite the absence of a porous transport model we can gain additional qualitative insights into ASW self-diffusivity prior to crystallization by using a simple TPD bulk diffusion model to estimate values of ASW self-diffusivity in which bulk diffusive mixing would be detected in simulated TPD spectra. These values are an “upper limit” to the actual ASW selfdiffusivity (recall that the data of parts a-c of Figure 2 show no evidence of self-diffusion prior to crystallization). Here we alter the ASW self-diffusivity parameters (activation energy (Ea) and pre-exponential (D0)) in our simple bulk diffusion model (maintaining D ) 10-18 cm2/s at Tg ≈ 136 K) and monitor the simulated TPD spectra for interlayer bulk diffusive mixing (parts b and inset of Figure 9). The ASW self-diffusivity values (Ea ) 70 kJ/mol; D0 ) 7.7 × 108 cm2/s) obtained from this exercise are shown as the red line in Figure 9b inset. When these values

J. Phys. Chem. B, Vol. 110, No. 36, 2006 17995

Figure 9. Implications for the ASW self-diffusivity. Shown in Figure 9a is an Arrhenius plot [log(D) vs 1000/T] with literature values of supercooled liquid water diffusivities40-42 (open squares), previous ASW self-diffusivity estimates [Ea ) 170 kJ/mol; D0 ) 2.9 × 1043 cm2/s] of Smith and Kay25,26 (open circles, calculated neglecting porositymediated transport) and a VFT fit to both sets of data (bold solid line). The solid circle and solid triangle, represent D ) 10-18 cm2/s at temperatures of T ≈ 160 and 136 K, respectively. The black dotted line in Figure 9a is a VFT fit of the supercooled liquid data with D ) 10-18 cm2/s at Tg ≈ 160 K. Green dashed line illustrates the behavior which might be expected if water diffusivity exhibits a change from “fragile” to “strong” behavior (with decreasing temperature) between T ≈ 231 and 160 K. (Note: this line is not a calculation or data, it is solely for illustrative purposes.) Shown as the red solid line are values (see Figure 9b caption) of the ASW diffusivity predicted by a simple diffusion/desorption model to produce noticeable interlayer mixing. (b) and inset: Displayed in Figures 9b and inset is a demonstration of how the red line shown in Figure 9a was determined using our simple bulk diffusion/desorption model, a model which neglects porosity mediated transport (see Experimental section). To obtain this estimate, ASW self-diffusion parameters [DASW ) D0 exp (-Ea/RT)] were varied, as shown in the Figure 9b inset, maintaining D ) 10-18 cm2/s at Tg ≈ 136 K. These parameters were used to calculate simulated mixing spectra (for 30 BL H216O on 30 BL H218O at 0.6 K/s) using our simple bulk diffusion TPD model, monitoring the extent of mixing in the simulated TPD results. Shown in Figure 9b are simulated TPD spectra obtained by employing D ) 10-18 cm2/s (black and gray lines) and Arrhenius parameters D0 ) 7.7 × 108 cm2/s, Ea ) 70 kJ/mol (red and dark red lines). As Figure 9b demonstrates, as the ASW diffusivity parameters are increased [from D ) 10-18 cm2/s; (black line in Figure 9b inset) to D0 ) 7.7 × 108 cm2/s, Ea ) 70 kJ/mol (red line in Figure 9b inset)], increased diffusive mixing is observed in the corresponding simulated TPD spectra. This increased diffusive mixing amounts to 1.9 BL H218O, as determined by integration of the simulated TPD spectra (shaded gray area).

17996 J. Phys. Chem. B, Vol. 110, No. 36, 2006 are employed in our model, noticeable transport and intermixing of H218O is observed ((∼1.9 BL) as determined by integration of the simulated TPD spectra in Figure 9b (red and dark red lines)). Displayed in Figure 9a (red line) are the same ASW diffusivities (Ea ) 70 kJ/mol; D0 ) 7.7 × 108 cm2/s) shown in the Figure 9b inset that are a rough “upper limit” of the actual values. This suggests that the value of the actual bulk diffusivity of ASW is smaller than that characteristic of a fragile liquid (open circles) between T ≈ 150-157 K. As a consequence, to maintain a “smooth” connection with supercooled liquid water diffusivity data (∼242-298 K) and maintain Tg ≈ 136 K, the ASW diffusion coefficient would necessitate a transition from fragile to strong behavior, between T ≈ 231 K and T ≈ 160 K as depicted by the green dashed line in Figure 9a. A “fragileto-strong” transition in water diffusivity behavior has been suggested in recent theoretical work.51,52,54,55 Thus, if water’s Tg ≈ 136 K, our experimental results appear more consistent with a fragile-to-strong transition in water rather than continuous fragile behavior, as water is cooled from higher temperatures (T > 231 K). We have checked the sensitivity of these qualitative arguments using various values of Tg and diffusivity at the glass transition. After performing the same analysis employing a different value of the glass transition temperature (Tg ≈ 141 K) or using a different value of the diffusivity at Tg ≈ 136 K (D ) 10-20 cm2/s) we come to similar conclusions, i.e., ASW diffusivity values that are much smaller than those characteristic of a “fragile” liquid prior to crystallization. The results of this qualitative analysis are consistent with ASW diffusivities being characteristic of a strong liquid with Tg ≈ 136 K (undergoing a fragile-to-strong transition between T ≈ 160-230 K) or a glass prior to crystallization (with Tg > 160 K). Which of these two possibilities is correct remains an open question, since both scenarios are consistent with our experimental data. Answering this question will require obtaining more precise, quantitative values of the ASW self-diffusion coefficient prior to crystallization. Souda et al. have shown the utility of TOF-SIMS in studying morphology changes in thin glassy water films27 and intermixing of thin MeOH/ASW films.28 This technique could prove useful in studying mixing between isotopically labeled, structured films of ASW with similar thicknesses as employed here. Structured films could be annealed for time scales/temperatures such that ASW crystallization/fracture does not occur, cooled rapidly to quench further water diffusion, and sputtered to obtain concentration profiles from which diffusivity values could be calculated. Ideally, sputtering techniques could be combined with desorption methods to both (1) determine concentration profiles in annealed, structured ASW (H218O, H218O) films which have not fractured and (2) verify no film fracture or morphology changes have occurred in the structured film due to the anneal treatment, via abrupt CCl4 desorption,65 surface area measurements,76,77 and/ or Kr desorption measurements.88 Conclusions In summary, we have studied transport processes in structured, nanoscale ASW films to gain insight into the nature of water’s glass transition and fragility at temperatures prior to transformation to CI (T ≈ 150-160 K). The effects of ASW film thickness, diffusion barrier layer thickness and type (CCl4, CCl3H), and heating schedule on water self-transport were examined. Isothermal mixing experiments were also conducted, along with surface area measurements, which provide an

McClure et al. additional measure (besides abrupt CCl4 desorption) of film fracture and morphology changes. Structured films containing MeOH were studied in an attempt to probe the effectiveness of CCl4 layers in hindering diffusive transport. Carbon tetrachloride (CCl4) was used as a “marker” in an attempt to gain qualitative insights into the thickness and temperature dependence of a porosity-mediated transport mechanism. We find that: (1) The intermixing of isotopically labeled, structured films occurs similarly with and without hydrophobic spacer layers (Figures 2, 3, and S1). (2) The nonmonotonic behavior of the desorption ratio (r18/r16) during desorption of structured ASW films is fundamentally inconsistent with a bulk diffusion mechanism (insets of Figures 1b and 2a). (3) The onset of intermixing is concurrent with porosity formation (abrupt CCl4 desorption) regardless of the heating schedule or ASW film thickness (Figures 4, 5, 6, 8, and S2). (4) Isothermal experiments exhibit intermixing concurrent with increases in the surface area due to crack/fracture of the ASW film (Figures 5 and 6) from T ≈ 146-154 K. (5) Experiments (Figures S3, S4) employing MeOH as a “probe” molecule exhibit behavior suggesting that CCl4 and CCl3H barrier layers can hinder translational motion in these films. (6) Carbon tetrachloride (CCl4) was used as a “marker” to gain qualitative insights into the thickness and temperature dependence of a porosity-mediated transport mechanism (Figures 7, 8, and S5). The qualitative trends that are observed in such data mirror behavior exhibited in layered ASW mixing experiments with variable ASW film thicknesses (Figure 4). These findings further support the idea that the interlayer mixing observed in structured ASW films near T ≈ 150-160 K occurs via vapor-phase transport through an interconnected network of cracks/fractures created within the film during crystallization. Consequently, the self-diffusivity of ASW prior to crystallization (T ≈ 150-160 K) is significantly smaller than that expected for a “fragile” liquid, indicating that water undergoes either a glass transition or a fragile-to-strong transition at a temperature above 160 K. Acknowledgment. The authors would like to thank Bruce D. Kay, Zdenek Dohna´lek and R. Scott Smith for useful discussions. T.M.T. acknowledges financial support from the National Science Foundation (CAREER CTS-0448721), the David and Lucile Packard Foundation, and the Alfred P. Sloan Foundation. C.B.M. acknowledges both the Welch Foundation (F-1436) and the Department of Energy (DE-FG02-04ER15587) for financial support. Supporting Information Available: Measurements regarding transport properties of structured ASW films employing different spacer materials, temperature ramp rates, methanol as a probe of barrier effectiveness, and ASW film thicknesses are included as supplemental information. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Debenedetti, P. G. J. Phys. Condens. Matter 2003, 15, R1669. (2) Debenedetti, P. G. Metastable Liquids: Concepts and Principles; Princeton University Press: Princeton, NJ, 1996. (3) Angell, C. A. Chem. ReV. 2002, 102, 2627. (4) Angell, C. A. Annu. ReV. Phys. Chem. 2004, 55, 559. (5) Mishima, O.; Stanley, H. E. Nature 1998, 396, 329. (6) Angell, C. A. Science 1995, 267, 1924. (7) Debenedetti, P. G.; Stillinger, F. H. Nature 2001, 410, 259. (8) Burton, E. F.; Oliver, W. F. Proc. R. Soc. London A. 1935, 153, 166. (9) McMillan, J. A.; Los, S. C. Nature 1965, 206, 806. (10) Sugisaki, M.; Suga, H.; Seki, S. J. Chem. Soc. Jpn. 1968, 41, 2591.

Transport in ASW Films (11) Hallbrucker, A.; Mayer, E.; Johari, G. P. J. Phys. Chem. 1989, 93, 4986. (12) Johari, G. P.; Hallbrucker, A.; Mayer, E. Nature 1987, 330, 552. (13) Kohl, I.; Bachmann, L.; Hallbrucker, A.; Mayer, E.; Loerting, T. Phys. Chem. Chem. Phys. 2005, 7, 3210. (14) Johari, G. P. J. Chem. Phys. 2003, 119, 2935. (15) Johari, G. P. J. Chem. Phys. B 2003, 107, 9063. (16) Kohl, I.; Bachmann, L.; Mayer, E.; Hallbrucker, A.; Loerting, T. Nature 2005, 435, E1. (17) Johari, G. P. J. Chem. Phys. 2002, 116, 8067. (18) Johari, G. P. J. Phys. Chem. B 1998, 102, 4711. (19) Johari, G. P. J. Chem. Phys. 2005, 123, 016102. (20) Johari, G. P. J. Chem. Phys. 2005, 122, 144508. (21) Johari, G. P.; Hallbrucker, A.; Mayer, E. J. Chem. Phys. 1992, 97, 5851. (22) Johari, G. P.; Hallbrucker, A.; Mayer, E. J. Chem. Phys. 1991, 95, 2955. (23) Johari, G. P. Phys. Chem. Chem. Phys. 2005, 7, 1091. (24) MacFarlane, D. R.; Angell, C. A. J. Phys. Chem. 1984, 88, 759. (25) Smith, R. S.; Kay, B. D. Nature 1999, 398, 788. (26) Smith, R. S.; Dohna´lek, Z.; Kimmel, G. A.; Stevenson, K. P.; Kay, B. D. Chem. Phys. 2000, 258, 291. (27) Souda, R. Chem. Phys. Lett. 2005, 415, 146. (28) Souda, R. Phys. ReV. Lett. 2004, 93, 235502/1. (29) Minoguchi, A.; Richert, R.; Angell, C. A. J. Phys. Chem. B. 2004, 108, 19825. (30) Cerveny, S.; Schwartz, G. A.; Bergman, R.; Swenson, J. Phys. ReV. Lett. 2004, 93, 245702/1. (31) Minoguchi, A.; Richert, R.; Angell, C. A. Phys. ReV. Lett. 2004, 93, 215703-1. (32) Fisher, M.; Devlin, J. P. J. Phys. Chem. 1995, 99, 11584. (33) Ghormley, J. A. J. Chem. Phys. 1968, 48, 503. (34) Yue, Y.; Angell, C. A. Nature 2004, 427, 717. (35) Velikov, V.; Borick, S.; Angell, C. A. Science 2001, 294, 2335. (36) Yue, Y.; Angell, C. A. Nature 2005, 435, E2. (37) Tsekouras, A. A.; Iedema, M. J.; Cowin, J. P. Phys. ReV. Lett. 1998, 80, 5798. (38) Adam, G.; Gibbs, J. H. J. Chem. Phys. 1965, 43, 139. (39) Bertolini, D.; Cassettari, M.; Salvetti, G. J. Chem. Phys. 1982, 76, 3285. (40) Gillen, K. T.; Douglass, D. C.; Hoch, M. J. R. J. Chem. Phys. 1972, 57, 5117. (41) Price, W. S.; Ide, H.; Arata, Y.; So¨derman, O. J. Phys. Chem. B 2000, 104, 5874. (42) Pruppacher, H. R. J. Chem. Phys. 1972, 56, 101. (43) Ito, K.; Moynihan, C. T.; Angell, C. A. Nature 1999, 398, 492. (44) Angell, C. A. J. Phys. Chem. 1993, 97, 6339. (45) Jenniskens, P.; Blake, D. F. Astro. J. 1996, 473, 1104. (46) Sciortino, F.; Gallo, P.; Tartaglia, P.; Chen, H.-S. Phys. ReV. E 1996, 54, 6331. (47) Sciortino, F.; Fabbian, L.; Chen, H.-S.; Tartaglia, P. Phys. ReV. E 1997, 56, 5937. (48) Gallo, P.; Sciortino, F.; Tartaglia, P.; Chen, S-.H. Phys. ReV. Lett. 1996, 76, 2730. (49) Starr, F. W.; Harrington, S.; Sciortino, F.; Stanley, H. E. Phys. ReV. Lett. 1999, 82, 3629. (50) Starr, F. W.; Sciortino, F.; Stanley, H. E. Phys. ReV. E 1999, 60, 6757. (51) Starr, F. W.; Angell, C. A.; Stanley, H. E. Physica A 2003, 323, 51. (52) Starr, F. W.; Angell, C. A.; La Nave, E.; Sastry, S.; Sciortino, F.; Stanley, H. E. Biophys. Chem. 2003, 105, 573. (53) Xu, L.; Kumar, P.; Buldyrev, S. V.; Chen, S.-H.; Poole, P. H.; Sciortino, F.; Stanley, H. E. Proc. Nat. Acad. Sci. 2005, 102, 16558. (54) Truskett, T. M.; Dill, K. A. J. Phys. Chem. B 2002, 106, 11829. (55) Truskett, T. M.; Dill, K. A. Biophys. Chem. 2003, 105, 447. (56) Saika-Voivod, I.; Poole, P. H.; Sciortino, F. Nature 2001, 412, 514. (57) Hemmati, M.; Moynihan, C. T.; Angell, C. A. J. Chem. Phys. 2001, 115, 6663.

J. Phys. Chem. B, Vol. 110, No. 36, 2006 17997 (58) Faraone, A.; Liu, L.; Mou, C.-Y.; Yen, C.-W.; Chen, S.-H. J. Chem. Phys. 2004, 121, 10843. (59) Liu, L.; Chen, S.-H.; Faraone, A.; Yen, C.-W.; Mou, C.-Y. Phys. ReV. Lett. 2005, 95, 117802/1. (60) Bergman, R.; Swenson, J. Nature 2000, 403, 283. (61) Swenson, J. J. Phys. Condens. Matter 2004, 16, S5317. (62) Patschek, D.; Geiger, A. J. Phys. Chem. B 1999, 103, 4139. (63) Johari, G. P. J. Chem. Phys. 1996, 105, 7079. (64) McClure, S. M.; Safarik, D. J.; Truskett, T. M.; Mullins, C. B. J. Phys. Chem. B 2006, 110, 11033. (65) Smith, R. S.; Huang, C.; Wong, E. K. L.; Kay, B. D. Phys. ReV. Lett. 1997, 79, 909. (66) Davis, J. E.; Karseboom, S. G.; Nolan, P. D.; Mullins, C. B. J. Chem. Phys. 1996, 105, 8362. (67) Ramsey, N. F. Molecular Beams: Cambridge University Press: 1960. (68) Dohnalek, Z.; Kimmel, G. A.; Ayotte, P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2003, 118, 364. (69) Kimmel, G. A.; Stevenson, K. P.; Dohna´lek, Z.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2001, 114, 5284. (70) Dohna´lek, Z.; Kimmel, G. A.; Ayotte, P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2003, 118, 364. (71) Stevenson, K. P.; Kimmel, G. A.; Dohna´lek, Z.; Smith, R. S.; Kay, B. D. Science 1999, 283, 1505. (72) Kimmel, G. A.; Dohna´lek, Z.; Stevenson, K. P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2001, 114, 5295. (73) Brown, D. E.; George, S. M. J. Phys. Chem. 1996, 100, 15460. (74) Thiel, P. A.; Madey, T. E. Surf. Sci. Rep. 1987, 7, 211. (75) King, D. A.; Wells, M. G. Proc. R. Soc. London Ser. A. 1974, 339, 245. (76) Safarik, D. J.; Meyer, R. J.; Mullins, C. B. J. Vac. Sci. Technol. A. 2001, 19, 1537. (77) Safarik, D. J.; Meyer, R. J.; Mullins, C. B. J. Chem. Phys. 2003, 118, 4660. (78) Dohna´lek, Z.; Kimmel, G. A.; Ciolli, R. L.; Stevenson, K. P.; Smith, R. S.; Kay, B. D. J. Chem. Phys. 2000, 112, 5932. (79) Speedy, R. J.; Debenedetti, P. G.; Smith, R. S.; Huang, C. Kay, B. D. J. Chem. Phys. 105, 240. (80) Goto, K.; Hondoh, T.; Higashi, A. Jpn J. Appl. Phys. 1986, 25, 351. (81) Sack, N. J.; Baragiola, R. A. Phys. ReV. B.: Condens. Matter 1993, 48, 9973. (82) Ayotte, P.; Smith, R. S.; Stevenson, K. P.; Kimmel, G. A.; Kay, B. D. J. Geo. Res. 2001, 106, E12 33387. (83) Blanchard, J. L.; Roberts, J. T. Langmuir 1994, 10, 3303. (84) Donev, J. M. K.; Yu, Q.; Long, B. R.; Bollinger, R. K.; Fain, S. C., Jr. J. Chem. Phys. 2005, 123, 044706/1. (85) Jenniskens, P.; Banham, S. F.; Blake, D. F.; Mccoustra, M. R. S. J. Chem. Phys. 1997, 107, 1232. (86) Ghormley, J. A.; Hochanadel, C. J. Science 1971, 171, 62. (87) Westley, M. S.; Baratta, G. A.; Baragiola, R. A. J. Chem. Phys. 1998, 108, 3321. (88) Kimmel, G. A.; Petrik, N. G.; Dohna´lek, Z.; Kay, B. D. Phys. ReV. Lett. 2005, 95, 166102. (89) Lo¨fgren, P.; Ahlstro¨m, P.; Chakarov, D. V.; Lausmaa, J.; Kasemo, B. Surf. Sci. 1996, 367, L19. (90) Smith, R. S.; Huang, C.; Wong, E. K. L.; Kay, B. D. Surf. Sci. 1996, 367, L13. (91) Adler, P. Porous Media: Geometry and Transports; ButterworthHeinemann: Stoneham, MA, 1992. (92) Mason, E. A.; Malinauskas, A. P. Gas Transport in Porous Media: The Dusty-Gas Model; Elsevier Science Publishers: Amsterdam, The Netherlands, 1983. (93) Jung, K.-H.; Park, S.-C.; Kim, J.-H.; Kang, H. J. Chem. Phys. 2004, 121, 2758. (94) Livingston, F. E.; Smith, J. A. J. Phys. Chem. A. 2002, 106, 6309. (95) Miller, G. A.; Carpenter, D. A. J. Chem. Eng. Data 1964, 9, 371.