Graphene Visualizes the First Water Adlayers on Mica at Ambient Conditions

See allHide authors and affiliations

Science  03 Sep 2010:
Vol. 329, Issue 5996, pp. 1188-1191
DOI: 10.1126/science.1192907


The dynamic nature of the first water adlayers on solid surfaces at room temperature has made the direct detection of their microscopic structure challenging. We used graphene as an atomically flat coating for atomic force microscopy to determine the structure of the water adlayers on mica at room temperature as a function of relative humidity. Water adlayers grew epitaxially on the mica substrate in a layer-by-layer fashion. Submonolayers form atomically flat, faceted islands of height 0.37 ± 0.02 nanometers, in agreement with the height of a monolayer of ice. The second adlayers, observed at higher relative humidity, also appear icelike, and thicker layers appear liquidlike. Our results also indicate nanometer-scale surface defects serve as nucleation centers for the formation of both the first and the second adlayers.

Water coats all hydrophilic surfaces under ambient conditions, and the first water adlayers on a solid often dominate the surface behavior (14). Although scanning tunneling microscopy (STM) and other ultrahigh vacuum surface characterization techniques have been extensively used to study water (ice) adlayers on solids at cryogenic temperatures (1, 2), such techniques are not applicable to room-temperature studies because of the high vapor pressure of water (2, 3). Various optical methods have been used at ambient conditions to probe the averaged properties of water adlayers over macroscopic areas (3, 57). Atomically resolved studies have remained challenging. For example, although thin ice layers have been studied with atomic force microscopy (AFM) below freezing temperatures (8, 9), reliable AFM imaging of water adlayers under ambient conditions is confounded by tip-sample interactions (2). For example, the capillary menisci formed between the tip and the sample strongly perturb the water adlayers on solids (10).

Scanning polarization force microscopy (SPFM) has been used to image water adlayers (2, 11, 12). For SPFM, the tip-sample distance is kept at tens of nanometers. By briefly contacting the tip on a mica surface to induce capillary condensation, metastable islandlike structures were observed in SPFM images. These islands were interpreted as a second adlayer on a monolayer of water (2, 12). However, the lateral resolution of SPFM is relatively low, and the measured apparent heights reflect local polarizability instead of actual heights. Furthermore, the structure of the first adlayer was not observed, likely because of the low lateral resolution and/or the dynamic nature of the first adlayer (12).

Here, we report on the use of monolayer graphene sheets as ultrathin coatings for enabling an AFM study of the first water adlayers on mica. Sputtered carbon is commonly used to coat biological systems, such as cells, for electron microscopy imaging. The carbon enables the imaging experiments by providing a protective (and conductive) coating. The graphene coating used here plays a somewhat similar role; graphene can tightly seal what are otherwise elusive adlayers and, stably “fix” the water adlayer structures, thus permitting the detection of the structure of the first water adlayers under ambient conditions. Humidity-dependent experiments further reveal how the structure of the water adlayers evolves at the nanometer and molecular scale.

Graphene sheets were deposited onto the (001) surface of muscovite mica through the standard method of mechanical exfoliation of Kish graphite flakes (13, 14). We readily identified graphene monolayers, as well as bilayers and multilayers, on mica through a back-illuminated optical microscope [the mica substrate is translucent, and the graphene layers absorb white light (15); see fig. S1 (16)]. The numbers of graphene layers were unambiguously confirmed through spatially resolved Raman spectroscopy (17, 18) (fig. S2).

In Fig. 1, we present typical AFM images of graphene deposited on mica at ambient conditions [room temperature; relative humidity (RH) ~ 40%]. In agreement with a recent study (14), we found that graphene sheets spread atomically flat on mica over areas of 100 to 200 nm on a side (fig. S6). Over larger areas, however, islandlike plateaus varying from a couple nanometers to a few micrometers in lateral size were observed across all the graphene samples (Fig. 1 and figs. S3 and S5). These plateaus appear atomically flat (fig. S6), and plateaus from different samples have the same height of 0.37 ± 0.02 nm (SE) (Fig. 1, E and G) regardless of the lateral dimensions. Although dotlike thicker features were occasionally present, no plateaus with heights smaller than ~0.35 nm were observed.

Fig. 1

Graphene visualizes the first water adlayer on mica surface at ambient conditions. (A) A schematic of how graphene locks the first water adlayer on mica into fixed patterns and serves as an ultrathin coating for AFM. (B) The structure of ordinary ice (ice Ih). Open balls represent O atoms, and smaller, solid balls represent H atoms. A single puckered bilayer is highlighted with red. Interlayer distance is c/2 = 0.369 nm when close to 0°C. Adapted from (22). (C) AFM image of a monolayer graphene sheet deposited on mica at ambient conditions. (D) A close-up of the blue square in (C). (E) Height profiles along the green line in (D) and from a different sample (fig. S3). The dashed line indicates z = 0.37 nm. (F) AFM image of another sample, where the edge of a monolayer graphene sheet is folded underneath itself. The arrow points to an island with multiple 120° corners. (G) The height profile along the red line in (F), crossing the folded region. Scale bars indicate 1 μm for (C) and 200 nm for (D) and (F). The same height scale (4 nm) is used for all images.

The observed plateaus are not flakes of graphene or mica layers; such structures were not observed on the exfoliated surfaces of graphite and mica or graphene deposited on SiO2 substrates (fig. S3). In addition, the ~0.37-nm height is not consistent with the layer thicknesses of graphene (0.335 nm) or mica (0.99 nm), and, as described below, the structures themselves depended on the RH of the experimental conditions. For the case in which the edge of a monolayer graphene sheet is folded underneath itself (Fig. 1, F and G), a ~0.34-nm step height was observed for the folded graphene, whereas the same ~0.37-nm height was observed for plateaus both in and out of the folded region. Plateaus with the same height were also observed in bilayer graphene sheets, and the plateaus appeared to be continuous across monolayer-bilayer borders (fig. S3). Our phase images further indicated that the plateau structures were under the graphene sheets (fig. S4). Plateaulike structures a few tenths of nanometers in height have been noticed for multilayer graphene on mica and were also identified as gases or moisture trapped under graphene (19). Indeed, albeit only one atom thick, monolayer graphene is known to be robust and impermeable to liquid and gas (20, 21).

The observation of atomically flat plateaus that have well-defined heights and depend on the RH indicates that the observed structures are not random gas molecules trapped between the graphene and mica surfaces but instead are ordered water adlayers (Fig. 1A). Previous SPFM studies have observed water layers on mica surfaces forming two-dimensional islands tens of nanometers to several micrometers in lateral size (2, 11, 12). The shapes and size distribution of those water islands are in good agreement with the trapped structures in Fig. 1, except that the presence of water islands smaller than a few tens of nanometers were not previously known, likely because of the ~10-nm lateral resolution of SPFM. The heights of the water islands were also not accurately determined with SPFM. The ~0.37-nm height we measured is in good agreement with the height of a monolayer of a “puckered bilayer” of ice (c/2 = 0.369 nm, where c is the lattice constant indicated in Fig. 1B) (22), a widely assumed model for how water molecules arrange in the first adlayer on a solid (1, 4, 23).

We emphasize that, although morphologically similar, the islands observed in the previous SPFM studies were the second water adlayer artificially induced on top of the first adlayer (2, 12). The nature of the first adlayer was largely unknown because of the high mobility of water molecules at room temperature (12). As will become more evident in the RH-dependent experiments (Figs. 2 and 3), the plateaus we observed on samples prepared at ambient conditions (Fig. 1) are the first water adlayer on the mica surface: The second adlayer appears only with high frequency at RH >~ 90%. For our case, graphene serves as an ultrathin coating that locks the first water adlayer into fixed patterns for AFM imaging. The fixed patterns are remarkably stable: Besides preventing any appreciable changes of morphology during the several hours of our AFM operation, we found that the patterns are stable for weeks under ambient conditions (fig. S5). However, the water adlayer can become mobile again when the mica substrate is subjected to extensive bending (fig. S5). Bending causes shear and displacement of graphene on the mica surface, thus releasing the locked water. The adlayer reorganizes accordingly, reflecting its dynamic nature.

Fig. 2

AFM images of graphene deposited on mica at RH ~ 2%, revealing the influence of surface defects on water adlayer nucleation. (A) A representative sample. ML indicates monolayer graphene; 2L, bilayer graphene. A dotlike defect is highlighted in the height profile across the cyan line. (B) Image of monolayer graphene deposited on a mica surface with high density of surface defects. A height profile is given for the pink line. The dash line indicates z = 0.37 nm. The same height scale (4 nm) is used for both images.

Fig. 3

AFM images of graphene deposited on mica at RH ~ 90%, revealing the structure of the second water adlayer. (A) A representative sample. ML, monolayer graphene; 2L, bilayer graphene. (B) A close-up of the graphene edge, corresponding to the blue square at the bottom left of (A). A height profile is given for the red line. The first step (~0.7 nm in height) corresponds to monolayer graphene on bare mica surface. The second step (~0.37 nm) corresponds to the first water adlayer on mica, which had been sealed by the graphene. (C) A close-up of the pinholes, corresponding to the yellow square in (A). A height profile is given for the green line. (D) Image of monolayer graphene deposited on mica with a high density of surface defects. (E) A close-up of the second adlayer islands, corresponding to the orange square in (D). Height profiles are given for the pink and cyan lines. The dash line indicates z = 0.38 nm. Scale bars, 1 μm for (A) and (D) and 200 nm for other images. The same height scale (4 nm) is used for all images.

The boundaries of the islands formed by the first water adlayer often exhibited fascinating polygonal shapes with preferred angles of ~120°. For example, the arrow in Fig. 1F points to an island with multiple 120° corners. This geometry suggests that, at ambient conditions, the first water adlayer has an icelike structure that grew epitaxially on the substrate, similar to what was previously observed for the second water adlayer (2, 11). We also note that all the islands had the same height as a single layer of ice. These results are consistent with previous sum-frequency-generation spectroscopy results obtained over large areas (6). By contrast, we also imaged adlayers of tetrahydrofuran on mica by using graphene templating, and we observed monolayer island structures that did not exhibit polygonal shapes (fig. S7). For water, the observed sub-monolayer coverage at ambient conditions is also consistent with previous macroscopic optical studies (3, 5, 7), which indicated one statistical monolayer coverage at RH ~75% and a surface coverage of ~50% at RH ~40%. In stark contrast to mica, graphitic surfaces (including graphene) are highly hydrophobic (24, 25), and water is known to only adsorb on graphitic surfaces below ~150 K (26). Thus, in the sandwich structure (Fig. 1A) no water is expected to come from the graphene side. The occasionally observed dotlike thicker features are possibly caused by surface defects that attract water, as discussed below.

To investigate how the water adlayers evolved as the environmental humidity varies, we deposited graphene onto mica under controlled RH (16) and characterized the samples with AFM at ambient conditions. These studies also permitted investigations into the role that surface defects play in the initial formation of water adlayers. Figure 2A presents an AFM image of graphene deposited on mica under dry conditions (RH ~ 2%). No islandlike structures are observed for most samples prepared in this way: Graphene lies atomically flat (14) (fig. S6) without observable features, except for sporadic dotlike structures ~2 nm in height, which are likely due to surface defects. This result agrees with previous optical studies (5, 7), which indicated no reliably detectable water adsorption on mica surfaces at RH ~ 2%.

The measured height of monolayer graphene on bare mica surface was sensitive to the specific settings of AFM and could vary from 0.4 to 0.9 nm. We attributed this observation and similar height variations observed for monolayer graphene on SiO2 (0.5 to 1 nm) (17, 27) to the large chemical contrast between graphene and the substrate (17). This is why Raman spectroscopy provides such a useful probe for distinguishing graphene monolayers from bilayers and thicker films. The heights of the water islands in this study, however, can be accurately determined: The AFM tip always interacted with the same material (graphene) that uniformly coats the underlying sample (Fig. 1A); variations in tip-sample interactions were avoided.

Patchy islands were occasionally observed for graphene deposited, at 2% RH, on mica surfaces that were characterized by a high density of surface defects (Fig. 2B). The same height of ~0.37 nm is again measured for those islands, indicating a single adlayer of water. Interestingly, most islands connect nearby defects, suggesting the importance of defects for water adlayer nucleation. The adlayer boundaries appear round near the defect sites but resume the 120° polygonal shape away from the defects, indicating a competition between capillary interactions and the epitaxial interactions with the substrate.

When graphene was deposited on mica at high humidity (RH ~ 90%), the samples typically appear flat over large areas (Fig. 3A). However, a closer look at the edge of the graphene sheets revealed that the graphene rides on top of a near-complete monolayer of water adlayer (Fig. 3B). At about 10 nm from the edge of the graphene-water-mica sandwich structure, water evaporated away and graphene came into direct contact with the mica surface, sealing and preserving the remaining water adlayers. The ~0.37-nm height (Fig. 3B) indicates that the trapped water is a single adlayer. Polygonal pinholes ~10 nm in lateral size and ~0.37 nm in depth were also observed on the overall continuous adlayer (Fig. 3C), indicating that the monolayer is not 100% complete.

Different results were obtained for graphene deposited, at 90% RH, on a mica characterized by a high density of surface defects (Fig. 3D). Besides a completed (no pinholes) first adlayer of water that is missing only at the graphene sheet edge, islands of various lateral sizes were observed on top of the first adlayer, often surrounding or connecting local defect sites. These islands were atomically flat (fig. S6) and were 0.38 ± 0.02 nm in height over the first adlayer (Fig. 3E), again in agreement with the height of a single puckered bilayer of ice (0.369 nm). The observed ~120° polygonal shapes of these islands agree with previous SPFM results on tip-induced second water adlayers (2, 11, 12). Thus, the islands observed in Fig. 3, D and E, are the second water adlayer, which also has an icelike structure at room temperature and is epitaxial to the first adlayer. Bulgelike features a few nanometers in height were also observed but appeared to be liquidlike (roundish) and have varying heights. No icelike islands or plateaus were observed beyond the second adlayer. Previous optical studies (3, 5, 7) indicated that statistically only a few adlayers on the mica surface exist at RH ~ 90%, but with large sample-to-sample variations, a result that is consistent with the observations reported here.

Under ambient conditions, water adlayers grow epitaxially on mica in a strictly layer-by-layer fashion: The second adlayer forms only after the first adlayer is fully completed. In the submonolayer regime, two-dimensional islands form because of interactions between adsorbed molecules, possibly akin to the Frank–van der Merwe growth mechanism in heteroepitaxy (28). This result is consistent with previous studies that indicated the absence of dangling O-H bonds (6) and a minimum in entropy (7) at one statistical monolayer coverage. It also explains why water adsorption isotherms cannot be modeled with theories based on continuum models (5). Our findings also highlight the role that surface defects play in water adsorption: Defects apparently serve as nucleation centers for the formation of both the first and second adlayers. The importance of surface defects helps explain the large sample-to-sample variations previously reported in isotherm measurements (3, 5). The use of STM (2931) to characterize the atomic structures of graphene on water adlayers represents an exciting future challenge.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank G. Rossman for generous assistance in using the micro-Raman spectrometer and L. Qin and W. Li for helpful discussions. This work was supported by the U.S. Department of Energy, Basic Energy Sciences (grant no. DE-FG02-04ER46175).
View Abstract

Stay Connected to Science

Navigate This Article