Ultrafast Electron Crystallography of Interfacial Water

See allHide authors and affiliations

Science  02 Apr 2004:
Vol. 304, Issue 5667, pp. 80-84
DOI: 10.1126/science.1094818


We report direct determination of the structures and dynamics of interfacial water on a hydrophilic surface with atomic-scale resolution using ultrafast electron crystallography. On the nanometer scale, we observed the coexistence of ordered surface water and crystallite-like ice structures, evident in the superposition of Bragg spots and Debye-Scherrer rings. The structures were determined to be dominantly cubic, but each undergoes different dynamics after the ultrafast substrate temperature jump. From changes in local bond distances (OH··O and O···O) with time, we elucidated the structural changes in the far-from-equilibrium regime at short times and near-equilibration at long times.

The nature of interfacial molecular assemblies of nanometer scale is of fundamental importance to chemical and biological phenomena (14). For water, the directional molecular features of hydrogen bonding (5, 6) and the different structures possible, from amorphous (7) to crystalline (8), make the interfacial (9) collective assembly on the mesoscopic (10) scale much less understood. Structurally, the nature of water on a substrate is determined by forces of orientation at the interface and by the net charge density, which establishes the hydrophilic or hydrophobic character of the substrate. However, the transformation from ordered to disordered structure and their coexistence critically depends on the time scales for the movements of atoms locally and at long range. Therefore, it is essential to elucidate the nature of these structures and the time scales for their equilibration.

Here, we report direct determination of the structures of interfacial water with atomic-scale resolution, using diffraction and the dynamics following ultrafast infrared (IR) laser-initiated temperature jump. Interfacial water is formed on a hydrophilic surface (silicon, chlorine-terminated) under controlled ultrahigh vacuum (UHV) conditions (Fig. 1). With these atomic-scale spatial, temporal, and energy resolutions, the evolution of nonequilibrium structures was monitored, their ordered or disordered nature was established, and the time scale for the breakage of long-range bonding and formation of new structures was determined. We identified the structured and ordered interfacial water from the Bragg diffraction and the layered crystallite structure from the Debye-Scherrer rings. The temporal evolution of interfacial water and layered ice after the temperature jump was studied with submonolayer sensitivity. We compared these results with those obtained on hydrophobic surfaces, such as hydrogen-terminated silicon or silver substrate.

Fig. 1.

Structured water at the hydrophilic interface. The chlorine termination on a Si(111) substrate forms a hydrophilic layer that orients the water bilayer. The closest packing distance (4.43 Å) between oxygen atoms in the bottom layer of water is similar to the distance (4.50 Å) between the on-top and interstitial sites of the chlorine layer, resulting in specific bilayer orientations (±30°) with respect to the silicon substrate. This ordered stacking persists for three to four bilayers (∼1 nm) before disorientation takes place and results in crystallite islands, forming the layered structure. The size of atoms is not to scale for the van der Waals radii.

Spectroscopic techniques, such as internal reflection (11) and nonlinear [second-harmonic generation (12) and sum-frequency generation (SFG) (13)] optical methods, are sensitive to surface molecular changes. For example, the presence of polar ordering of ice films on a Pt(111) surface was shown to have a decay length of 30 monolayers (14), and transient SFG response from D2O ice crystals on a CO/Pt(111) surface has indicated the presence of melting and recrystallization without desorption (15). Here, structures were determined using diffraction with ultrafast time resolution, providing a spatial resolution of 0.01 Å. We can monitor the change on selective internuclear distances of the hydrogen bonding network, e.g., those of OH··O and O···O at 2.75 Å and 4.5 Å, respectively. Unlike previous studies on ultrafast surface restructuring (16) and subnanosecond melting (17), we can probe supramolecular structural dynamics on surfaces and observe clear separation of the diffraction of the interfacial water from that of the substrate.

The experiments were performed using a newly developed instrument for ultrafast electron crystallography (UEC) (16). Briefly, water was prepared on a single crystal Si(111) surface terminated chemically with chlorine atoms (18) to make the hydrophilic interface (19). The crystal was mounted on a goniometer (angular precision, 0.005°) in a UHV environment (16). The layer preparation on the surface (20) requires characterization of the substrate by low-energy electron diffraction and Auger spectroscopy, as well as in situ monitoring of layer growth by reflective high energy electron diffraction. In our case, the electron flux (∼1 pA/mm2) was relatively small, so there was no damage and no charging of the molecular layers. The IR pulses (typically ∼1 mJ, 800 nm, and 120 fs at a 1 kHz repetition rate) were directed at a 30° angle into the scattering chamber and focused on the substrate to initiate the temperature jump. A weaker beam was split from the IR beam, frequency tripled (∼10 nJ at 266 nm), and focused onto a back-illuminated silver photocathode after an adjustable time delay to generate the electron pulses via the photoelectric effect. These pulses (21) have a de Broglie wavelength λ = 0.07 Å at 30 keV. A series of deflectors and apertures were used to guide the electron beam for an incidence to the surface of θi < 1°, which is required for high sensitivity on nanometer-scale surface assemblies. The arrival of the electron pulses was controlled to define a sequence of images that were recorded with a low-noise, image-intensified, charge-coupled device (CCD) camera assembly capable of single-electron detection.

We first characterized the diffraction of the Si substrate before dosing with water. By rotating the crystal, we obtained the rocking curves and the dependence of diffraction pattern on θi. These diffraction patterns represent the intersection between Ewald's sphere and the reciprocal lattice defined by the substrate. Thus, the momentum transfer coordinates (s), defined by Math where θ the scattering angle, can be mapped out for any given diffraction image at the incidence angle θi; in situ zero-of-time was also determined using the substrate surface temperature jump. As interfacial ice was forming on the surface at 110 K (20), the diffraction images showed the transition from the Bragg spots of the substrate to the new spots and rings characteristic of water. After the substrate temperature jump, we used different sequencing of electron pulses in order to separately image evolving structures of water. This diffraction difference method (22) allows for the isolation of the only transient structure involved because the reference time (tref) can be chosen before or after the arrival of the initiating pulse, or different times during the change.

We show the diffraction images obtained with ultrashort electron pulses (Fig. 2, A to G), but without the initiating laser pulse (equivalent to tref at negative time). The diffraction pattern is composed of rings, spots, and streaks. The disappearance of substrate diffraction and the appearance of surface water diffraction as monolayers of ice formed after annealing are evident (Fig. 2, C to F). The observed Bragg spots indicate that water molecules are oriented by the substrate with a long-range order. The rings coincide with the spots in the reciprocal space (s space), and these rings indicate the emergence of crystallite structure: islands of ordered water, but disordered in orientation. The rings are sharp enough to define a crystallite that is not amorphous. The structural evolution of the interfacial water layers as a function of temperature was calibrated at near-equilibrium condition (temperature ramp, ∼2°/min). We found that at this nanometer scale the crystallization of the initially deposited amorphous ice begins near 140 K and reaches the saturation at ∼150 K with the highest degree of long-range order. The sublimation of the water layers occurs at 157 ± 1 K. From both Bragg spots and the rings, we can determine the structure by comparing these diffraction images with those predicted by the symmetry of ice lattices.

Fig. 2.

(A to F) The processes of in situ growth of ordered ice is illustrated through vapor deposition of water on a cold (110 K) silicon substrate (20). The adsorption of water on the substrate is seen from the disappearance of 111 Bragg spot of silicon (A) and the formation of 111 Bragg spot of crystalline ice, together with the diffraction rings of amorphous ice (B). Annealing promotes the formation of long-range crystalline structure, as shown in the increase of the brightness of the spots and the sharpening of the rings (C and D). The structures stablize as the diffraction shows nearly no change in the rings, spots, and streak (E and F). (G to I) Experimentally observed diffraction and simulations of spots from a nanometer-thick substrate, orientated cubic (Ic) and hexagonal (Ih) structures. Experimental diffraction rings when radially averaged in s space produce one-dimensional diffraction intensity curves (J). Theoretical diffraction intensity curves are also given (K and L), with the peaks identified with Bragg reflections. The clear distinction of different orderings and the early appearance of spots (not rings) in the annealing suggest heterogeneous layers on the surface with the crystallites layered by the ordered interfacial water. Moreover, the diffraction spots are sharp, which indicates that the interfacial water posses well-defined orientation; in contrast, the diffraction rings are circular, which is completely in agreement with randomly orientated crystallites.

The intensity of diffraction rings when plotted against s gives the corresponding peaks of Bragg reflections. The theoretical diffraction patterns for both cubic and hexagonal structures are shown in Fig. 2, K and L. These were obtained by summing the phases of a three-dimensional (3D) crystallite (a cell of a dimension of 5 nm in each direction) and averaged over all orientations. The peaks in Fig. 2J agree well with reflections from the 111, 220, and 311, etc., planes for cubic ice (Fig. 2K), which is a dominant structure [for comparison, see hexagonal ice diffraction (Fig. 2L)]. From the s values in the Fig. 2J, we determined the interplanar distances to be 3.80 ± 0.23 Å, 2.27 ± 0.15 Å, and 1.93 ± 0.07 Å for the (111), (220), and (311) planes, respectively, consistent with reported powder diffractions of cubic structure (23). The uncertainties are governed by the observed width of the diffraction peaks.

The surface-oriented water is a well-defined crystalline structure (epitaxially grown from the substrate), and this structure is responsible for the sharp Bragg spots. We reproduced theoretically the position and relative intensities of Bragg spots (Fig. 2H) by summing phases of long-range ordered interfacial water (10 nm by 10 nm wide; 1.5-nm-thick layer). For this hydrophilic substrate, we found that the closest packing distance between oxygen atoms in the bottom layer of water (4.5 Å) is similar to that between on-top and interstitial sites (4.43 Å) of the chlorine layer. This makes possible the long-range packing of the water bilayer (9) on the surface and leads to the unique 30° rotations with respect to the substrate layer. Thus the two-dimensional (2D) surface unit cell of water can be described as a superlattice Math (Fig. 1). This assignment was derived directly from the symmetries and positions of Bragg spots of the substrate and those of water. Inspection of the Bragg spots of ice identifies the two domains (±30° rotations) in the satellites of the main Bragg (i.e., 111) peak; one set is formed by 022, 111, 311; the other is by 202, 111, 131 (Fig. 2H). Water interacts with the substrate at two sites, likely through its oxygen in the interstitial site with sp3 hybridized orbitals overlapping with three chlorine atoms, or through hydrogen sitting on top of the chlorine atom.

From the diffraction results, we established that water is structured on the hydrophilic surface mainly as cubic [see the comparison between theory for orientated cubic (Ic) and orientated hexagonal (Ih), and the experiment (Fig. 2)] and is different from structures (hexagonal) found on Pt(111) (24) substrate. Our theoretical modeling of the position of the 111 Bragg spot gives an interlayer spacing of 3.66 ± 0.26 Å, entirely consistent with the value obtained from the rings (3.80 ± 0.23 Å). The apparent brightness and width of Bragg spots are an inherent reflection of the size of interferences, and from them we obtained a thickness of nanometer scale, which is also consistent with theory.

For the dynamics, we followed the diffraction as a function of time after the temperature jump of the substrate (Fig. 3). When tref is at negative time (–70 ps), the image at –30 ps (referenced to the –70 ps, Fig. 3A) shows no diffraction difference intensity, as expected. At positive time (Fig. 3, B and I), both the Bragg spots and rings emerge, but with striking structural changes (note the displacement of rings and spots). The different panels in Fig. 3 display the evolution with clear indication of the disappearance of the “old” structure and the appearance of a “new” structure. However, the behavior is similar in appearance to that of a “phase transition”: at short times [10 ps (Fig. 3D) and 20 ps (Fig. 3E)], we observed depletion of old structure, whereas at intermediate times (Fig. 3, F and G), there appeared a region of coexistence of disordered and crystal-like water. At very long time (Fig. 3H), the system reverted to the original structure, but with some difference in bond distances. [In a separate cryocooling experiment, we changed the substrate temperature by ramping at near–thermodynamic equilibrium condition, just below sublimation temperature (157 K), and found the ice structure to be in agreement with those obtained after restructuring at long times.] This behavior for the crystallites away from the surface (rings) contrasts that of the structured, crystal-like water on the surface (spots). We show the evolution of one spot (Fig. 3I) with time that exhibits the same trend— depletion and restructuring—but the dynamics are very different.

Fig. 3.

(A and B) Diffraction difference images at negative (–30 ps) and positive (100 ps) times. The tref is at –70 ps. (C to H) The radially averaged diffraction difference intensity curves at several delay times [(C) –30 ps, (D) 10 ps, (E) 20 ps, (F) 100 ps, (G) 530 ps, (H) 1130 ps] show the structural dynamics for layered crystallites at substrate energy fluence of 22 mJ/cm2. Note the depletion (negative difference) and the increase (positive difference) at well defined s values. (I) The diffraction difference images for the 111 Bragg spot. The vertical axis is s in the reciprocal space, and the horizontal axis is the azimuthal scattering angle.

By examining the rate of change at different temperatures, which is controlled by changing the fluence of the heating pulse of the substrate, we found (Fig. 4) that gating in the 111 reflection region shows a depletion of the old peak (Fig. 4, A and C) and a corresponding buildup (Fig. 4, B and D) of a new peak. This behavior mirrors the breakage of old bonds (by melting) and the formation of a new structure. The melting of the layered crystallites follows the laser excitation at the substrate surface within 5 ps. In contrast, from the same data, images gated on the 111 Bragg spot had a delay time of 36 ± 3 ps for interfacial water at the same energy fluence (22 mJ/cm2). Because water does not absorb light directly at 800 nm, the relatively prompt response of layered crystallites signifies the high efficiency of heating by nondiffusive vibrational couplings on this ultrashort time scale (25, 26); if diffusional, the layered ice should melt after the interfacial one. As the fluence increases, the delay decreased for interfacial water, but the time constants for depletion remained similar at 37 ± 5 ps. The results suggest the presence of a higher energy (friction) barrier for the structured water caused by the long-range order and dipolar orientation force of the hydrophilic substrate.

Fig. 4.

Temporal evolution of diffraction gated in the 111 reflection region. (A) The early-time (≤100 ps) depletion of old 111 diffraction spot and ring at several energy fluences. (B) The formation of new 111 diffraction spot and ring, also at early times. Note the apparent delays with respect to the zero-of-time, which is independently determined with uncertainty of 3 ps. The corresponding changes at longer times are shown in (C) and (D).

These observations indicate that at the highest temperature of the substrate (fluence, 42 mJ/cm2), the interfacial water continues to lose long-range order and that only when the maximum change is reached will restructuring begin (Fig. 4, A and B). However, near the maximum change, the new structure begins to form in the region of coexistence, followed by restructuring (Fig. 4, C and D). In the restructuring at long times, the dissipation of energy (cooling) occurs through redistribution and heat diffusion. In this regime, from the solution of the heat diffusion equation (27), we estimate a surface temperature of ∼150 K at 1 ns; the temperature at maximum change is ∼370 K (28). Over the entire time scale of structural change, thermal desorption was found to be insignificant because we observed the recovery of diffraction rings and spots back to nearly their original intensities, as shown from the constancy of the baselines at negative times (Fig. 4A) for all fluences. The lack of effective desorption, as also found for ice on CO/Pt surface (15), reflects the difference between desorption at near-equilibrium or far-from-equilibrium temperatures. The appearance of the plateau region at the higher fluence (42 mJ/cm2) for the depletion of old structure and the making of new structure, as well as the coexistence of old and new structures (10), suggests the involvement of collective modes in restructuring, analogous to phase transitions.

In supramolecular systems, such as the one discussed here, the local structures within a unit cell, in addition to the long-range order, can be examined by inverting (Fourier transform) the diffraction curves to the real space. Such obtained radial distribution functions [f(r)] (29) (Fig. 5, A to E) reveal the densities of internuclear distances. For cubic ice, the second nearest O···O distance (4.50 Å) correlates with the hydrogen bond OH··O distance (2.75 Å) in a diamond tetrahedron. The temporal changes of the density of these two distances thus provide the dynamics at the local molecular level of the restructuring of the hydrogen bond network. At negative time (–5 ps, Fig. 5B), no change of density is observed, as expected. At 10 ps (Fig. 5C), depletion of O···O peak is observed, although the change for the OH··O peak is insignificant. This depletion indicates the rupture of the network (to amorphous) in 10 ps. At 150-ps delay (Fig. 5D), significant but shifted depletions coexist with emerging new distances. At the longest delay, 1130 ps (Fig. 5E), the original cubic-like structure is recovered, as evidenced in the reduction in the diffraction difference, but the new structure is still slightly “hot” due to slow rate of diffusion (ns to μs). For such structures, we determined the interplanar distances of 4.22 ± 0.37 Å (111 plane), 2.42 ± 0.20 Å (220 plane), and 1.97 ± 0.05 Å (311 plane). A structural picture of local molecular changes is depicted in Fig. 5F.

Fig. 5.

(A) The radial distribution function (RDF) for Ic is shown with the internuclear distance density. The local distances at 2.75 Å (OH··O) and 4.5 Å (O···O) in a diamond tetrahedron unit are marked for the comparison with data (B to E). The experimental difference RDF curves were obtained by sine transform from the corresponding difference intensity curves (29) (Fig. 3). Changes of distance densities are evident at the three positive times, whereas no change is observed at negative time. The light gray curve in the panels is the Ic curve, scaled and superimposed for comparison. The change in the RDF clearly shows the depletion of the old structure and formation of the new structure (as in Fig. 3) but here identifies the bonds involved. (F) The corresponding structural changes at –5, +10, and +1130 ps.

Interfacial water on the hydrophilic surface substrate has distinctive structures and dynamics compared with those of water layers on hydrophobic surfaces. The time scale for the breakage of long-range order of the interfacial layer (37 ps) is an order of magnitude longer than that for breaking hydrogen bonds in bulk liquid water (30, 31), and the local OH··O and O···O bond distances from diffraction are directly involved in the change. These results suggest that the time scale for energy flow in the assembled water structure is much shorter than that of energy localization for desorption of individual molecules. (The maximum transient temperature is 370 K; the equilibrium desorption temperature is 157 K). Moreover, the restructuring time involving long-range order is longer than the time for amorphization, a process in which the O···O correlation is lost before the OH··O correlation. Perhaps it is not accidental that the time scale for losing the hydrogen bond network (37 ps) is similar to that reported for interfacial water near hydrophilic protein surfaces (20 to 50 ps) (31). In separate experiments, we also studied hydrophobic surfaces, such as hydrogen-terminated and silver-coated silicon substrates, and found that interfacial ordering has changed into a distribution around the (110) orientation for the former and is absent for the latter substrate (32).

With the unprecedented resolutions and sensitivity, these studies of nanometer-scale supramolecular structures, along with those of amorphous and crystalline solids (16, 33), demonstrate diverse applications of UEC, especially for the probing of interfaces and surfaces with atomic-scale resolution (33).

References and Notes

View Abstract

Navigate This Article