Water adsorption kinetics on graphene controlled by surface modification of supporting substrates

Sensing layers with an increased affinity for water molecules are essential for the development of highly sensitive humidity sensors. Graphene possesses superior electrical properties that make it suitable for the fabrication of low-noise miniaturized sensors. However, the enhancement of water affinity by introducing surface defects such as covalently attached hydrophilic groups reduces the electrical conductivity of graphene. In this study, we exploit the wetting transparency of graphene to increase its water affinity without introducing defects. Kinetic measurements using a Kelvin probe with a large-diameter tip showed that the rate constant of water adsorption was higher for graphene deposited on a hydrophilic substrate. These findings suggest that the wetting transparency of graphene can be exploited to reduce defect introduction into the graphene sensing layer, and has potential applications in sensor technologies.


Introduction
Graphene is a two-dimensional (2D) material composed of carbon atoms arranged in a hexagonal lattice [1].In addition to excellent mechanical strength and thermal conductivity [2], the electrical characteristics of this archetypal 2D material include an ultrahigh mobility of 200,000 cm 2 V −1 s −1 at an electron density of 2 × 10 11 cm −2 , far exceeding that of silicon [3].The monoatomic thickness of graphene results in an ultimately high surface-to-volume ratio, because of which electron transport is significantly affected by the surface adsorption of foreign molecules.Furthermore, graphene has been shown to produce low electrical noise [4,5], making it a promising candidate for use in high-sensitivity sensors.A wide variety of sensing targets have been reported thus far, including gases, such as NH 3 , CO, H 2 O, and NO 2 [4], and molecules present in living organisms, such as deoxyribonucleic acids [6], amino acids [7], and enzymes [8].The possibility of detecting H 2 O has led to its application as a humidity sensor.When water molecules are adsorbed onto a graphene surface, the resulting interactions modulate the Fermi level of graphene and alter its carrier density [9].Field-effect transistor sensors, which are based on the high electrical conductivity of graphene, can sense water adsorption as a change in the electrical conductivity of graphene [10,11].
The affinity of an adsorbate for a solid surface varies significantly according to the surface characteristics of the solid (presence or absence of chemical functional groups, roughness, etc.), which directly affects the sensitivity of the sensors.Graphene oxide (GO) and reduced graphene oxide (rGO) are frequently used as sensing layers because of the ease of their manufacture and the presence of hydrophilic functional groups on the surface (such as hydroxyl and carboxyl) [12][13][14].According to Anichini et al rGO that is chemically modified with hydrophilic groups showed higher sensitivity to humidity than unmodified rGO; conversely, when modified with hydrophobic aliphatic chains, the sensitivity to humidity decreased [12].Thus, the interaction between the sensor surface and water molecules plays an important role in humidity sensing.Interestingly, there are not many reports of pristine graphene-based sensors with relatively inert surfaces.A study of graphene synthesized by chemical vapor deposition (CVD) on copper foil and subsequently transferred to SiO 2 /Si substrates showed that the electrical changes upon exposure to humidity required tens of minutes to several hours to reach equilibrium [15], which is a much higher response time than that displayed by GO and rGO.
GO has much lower electrical conductivity than pristine graphene [16], while that of rGO is much closer.However, rGO still has many structural defects, including remnant oxygen groups arising from the imperfect reduction of GO [17].Structural defects impair the mechanical and electrical properties of graphene, and it is important to minimize these defects to improve the mechanical strength and sensitivity of sensors.In addition, if the high mobility of graphene can be maintained, the current density can be increased, which may contribute to device size reduction.These considerations indicate that there is a trade-off between the introduction of defects, which increases the affinity for water molecules, and high electrical performance.Therefore, it is preferable to achieve affinity enhancement without the introduction of defects.
In this study, we focused on the wetting transparency [18,19] of graphene on a supporting substrate, where the wettability of the graphene surface reflects that of the underlying substrate.In other words, graphene formed on a hydrophilic surface is expected to exhibit an increased affinity for water molecules.The effect of the wetting transparency on the kinetics of water adsorption was investigated by measuring the change in the surface potential upon water exposure over time.The adsorption kinetics are directly related to the response time, which determines sensor performance.Because graphene-related studies often use small graphene flakes obtained by mechanical exfoliation, their surface potential is measured using a local probe by Kelvin probe force microscopy (KFM) [20].However, KFM is based on a scanning probe and requires long measurement times that make it unsuitable for the measurement of kinetics.Therefore, in this study, we employed a Kelvin probe with a largediameter (4 mm) probe tip, which has a low signal-to-noise ratio and can track changes in the surface potential in a short time.For this purpose, we used a transferred film of CVD graphene with an area larger than the tip diameter.The results show that the adsorption rate of water molecules was higher for graphene formed on hydrophilic substrates.The positive effect of wetting transparency on the response time can be exploited to develop sensors with a high affinity for water molecules.

Experimental
Highly n-doped Si wafers (dopant: Sb; resistivity < 0.02 Ω cm) with 285 nm SiO 2 formed by dry oxidation were used as substrates.The substrate surface was cleaned by ultrasonication with acetone and 2-propanol and treated with an oxygen plasma cleaner (PDC-32G, Harrick Plasma).The wettability to water was modulated by the chemical modification of the substrate surface after cleaning.The hydrophilic treatment was performed using a self-assembled monolayer (SAM) of 3-aminopropyltriethoxysilane [21].Specifically, the cleaned substrates were immersed for 24 h in a mixture of 10 ml ethanol, 7.5 μl acetic acid, and 0.17 ml 3-aminopropyltriethoxysilane (APTS; Nacalai Tesque, purity >98%).The substrates were sonicated in pure ethanol to remove the excess adhering molecules.For hydrophobic treatment, hexamethyldisilazane (HMDS; Sigma-Aldrich, purity 99%) or 1H,1H,2H,2H-perfluorooctyltriethoxysilane (FOTS; Tokyo Chemical Industry, purity >95%) were used as the SAM molecules [22,23].For HMDS, the cleaned substrates were immersed in a mixture of hexane (9 ml) and HMDS (1 ml) for 10 h.For FOTS, the substrates were immersed in a 2 wt% hexane solution for 20 h.The substrates were cleaned using pure hexane in an ultrasonic bath.Large-area monolayer graphene layers, larger than the tip diameter of the Kelvin probe (4 mm), were deposited on the SAM-modified substrates from CVDgrown single-layer graphene (Sigma-Aldrich) on Cu via a wet-transfer process using a PMMA support layer.They were scooped onto hydrophobically treated substrates by mixing ethanol with ultrapure water (see Supplementary material for details).A 1 nm thick Cr adhesive layer and 30 nm thick Au layer were vacuumdeposited onto the graphene layer by resistive heating through a shadow mask.It was ensured that when viewed from above, the Cr/Au electrode on the graphene layer was not covered by the Kelvin probe tip placed above the sample.
Surface potentials were measured using a Kelvin probe (UHVKP020, KP Technology) in a stainless-steel vessel with a flow of pure nitrogen or water-containing nitrogen.The nitrogen flow rate was maintained at 200 ml min −1 using a mass flow controller and the water molecule concentration was maintained at 411 ppm using a permeator (PD-1B, GASTEC).The total gas flow rate was maintained at 200 ml min −1 during surface potential measurements to avoid the influence of pressure changes in the vessel.The wettability was quantified by measuring the water contact angle in air using a contact angle meter (SImage Standard 100, Excimer).The size of the water droplets used for the contact angle measurements was approximately 5 μl.To avoid the possible effects of residual water molecules on the surface potential measurements, contact angle measurements were performed last.To obtain information on the structure of graphene, Raman scattering spectra were recorded in air using a laser Raman microscope (RAMAN-11, Nanophoton) at an excitation wavelength of 532 nm.Raman spectra were acquired on an area of 380 μm 2 per measurement point to eliminate the influence of micro-scale position dependence.Raman scattering spectroscopy confirmed that the thickness of the transferred graphene layers resembles that of a monolayer.All the above measurements were performed at room temperature (∼25 °C).
The adhesion of graphene to the supporting surface is improved by heating [24].Heating also desorbs the water molecules that would remain on the graphene surface after the water adsorption experiments, which is essential for resetting the measurement conditions.Therefore, each sample was heated at 200 °C for 2 h in argon.Each measurement was repeated after 2 h of heating in Ar.For several samples, the heating-measurement procedures were repeated under the same conditions (total heating time was 4 h).

Results and discussion
The controllability of the wettability was confirmed by measuring the water contact angle.Measurements were recorded at three or four different positions in order to eliminate any possible position dependence.Figures 1(a) and (b) show the water contact angles of the hydrophobic and hydrophilic substrate samples, respectively, measured after heating in Ar for 2 h.The plot of measured contact angles against heating time (figure 1(c)), showed a large variation for the hydrophilic sample before annealing.This is likely due to the inhomogeneity of graphene adhesion, which causes microscale wrinkles and local hydrophobicity.Upon heating, the degree of adhesion to the substrate surface improved and the variation decreased.The initial decrease in the contact angle after annealing for 2 h was primarily attributable to the improved adhesion of the graphene layers.In addition, the change in the water contact angle due to heating was small, indicating that the degradation of the SAMs due to heat treatment was limited.Although the cause of the slight increase in the contact angle after annealing for 4 h remains unknown, it is possible that the partial degradation of the SAMs may explain this phenomenon.Graphene showed a larger water contact angle on the hydrophobically treated substrate than on the hydrophilic substrate, indicating that controlling the wettability of the substrate surface can control the wettability of graphene.
The hydrophobic and hydrophilic samples were subjected to time-resolved measurements of the surface potential using a Kelvin probe (figure 2(a)).Figure 2(b) (left) shows the surface potential change of as-transferred graphene on an HMDS-modified substrate upon the introduction of water vapor.The surface potential (work function) increased sharply with the introduction of water vapor and decreased slowly when the water vapor stopped, indicating that the adsorption of water molecules was a fast process, whereas desorption was a slow process.An enlargement of the potential increase (figure 2(b), right) shows a two-tiered change, attributed to the inhomogeneous adhesion of graphene to the substrate surface.The parts in firm contact with the substrate reflect the wettability of the substrate surface, whereas the improperly adhered portions remain unaffected by the substrate surface, resulting in inhomogeneous wettability.Upon heating the substrate at 200 °C in Ar for 2 h, the surface potential exhibited a one-tiered change (figure 2(c), left).Heat treatment improves adhesion to the substrate [24], and the wettability of the substrate surface is reflected over the entire graphene layer.Because the water contact angle is a macroscopically averaged quantity, the following discussion focuses on the one-tiered change in the surface potential.In addition, considering that the potential decrease after stopping the water vapor is very gradual and that the effect of baseline fluctuations cannot be completely eliminated, a meaningful discussion on water desorption is impractical with the current dataset.Therefore, this study focuses on the increase in surface potential associated with water adsorption.
The surface potential increase (figure 2(c)) exhibits a shape that can be expressed using the following equation: where ( ) j t is the surface potential over time, ∆j is the variation in surface potential, λ is the rate constant, and t 0 is the central time of change.t 0 is determined by the start time of the introduction of water vapor; thus, ( ) j t is regulated by ∆j and λ. Figure 3(a) illustrates the shape of this equation.The two parameters defining the surface potential change were extracted by fitting the experimentally obtained changes, as indicated by the solid line in the right panel of figure 2(c).The obtained rate constants were plotted against the water contact angle of each sample (figure 3(b)).The rate of change in the surface potential tended to increase for samples with smaller water contact angles, that is, the more hydrophilic samples.This implies that the rate of water adsorption on the graphene formed on the substrate can be modulated by controlling the wettability of the underlying substrate surface.
In addition to surface wettability, defects in graphene are known to increase the adsorption energy of water molecules (figure 4(a)) [25,26].Therefore, the degree of defectivity of the graphene layers was examined using Raman scattering spectroscopy (figure 4(b)).The presence of defects in the graphene samples was confirmed by  the appearance of a peak called the D-band [27], the intensity of which did not increase significantly upon annealing.A plot of the rate constant against the ratio of the D-and G-band peak intensities, D/G, was used to quantify the degree of defectivity (figure 4(c)).Data points with high rate constants were observed at both low and high D/G ratios, suggesting that the rate constant decreased at moderate defect densities.However, even in the region of high defect density, the rate constant was small for the hydrophobic surfaces.In addition, less defective graphene is known to induce a lower binding energy for water; however, this is not the case for graphene on hydrophilic surfaces, as the rate constant is high even at low D/G ratios.Therefore, the differences in the adsorption rates obtained in this study cannot be attributed to the defects in the graphene layers.
Strong charge-transfer interactions between the adsorbed molecules and graphene can also increase the rate of adsorption [28].Charge transfer occurs because of the relative energies of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of the adsorbate and the Fermi level of the graphene.If the increase in surface potential observed herein is due to charge transfer, it can be considered as a deepening of the Fermi level of graphene owing to the transfer of electrons to the adsorbed molecule (figure 5(a)).Graphene has a small density of electronic states near the charge-neutral point, and the Fermi level shift can be caused by charge transfer phenomena [1]. Figure 5(b) shows a plot of the rate constant versus the surface potential of graphene immediately before the introduction of water vapor.A higher rate constant was observed for the hydrophilic sample, even when the surface potential is high and the amount of electron transfer to the adsorbates is expected to be small.Therefore, the difference in the rate constants obtained in this study cannot be understood based on charge-transfer interactions.Additionally, because this experiment was conducted under nitrogen flow, the possibility of water-oxygen redox couples [29] could be eliminated, making it necessary to consider the possibility of charge transfer to water molecules alone.Charge transfer between graphene and water is unlikely because water clusters do not have molecular orbitals close to the Fermi level of graphene [30], which is consistent with the results of this study.Of the two parameters extracted by fitting, we focused on the rate constant in the discussion above.Plots of the other parameter, the variation in surface potential, against the water contact angle, D/G ratio, and surface potential before water vapor introduction in figures 6(a)-(c), respectively, did not reveal any clear dependence.The surface potential measures the work function, which is defined as the difference between the Fermi and vacuum levels.The Fermi level is affected by charge transfer because the density of states of graphene around the charge-neutrality point is sufficiently low to cause a Fermi level shift upon charge transfer.The vacuum level is affected by the presence of surface dipoles, which form an electric double layer, causing a potential shift.As discussed in the previous paragraph, there is little direct charge transfer between graphene and water molecules.Therefore, a possible cause of the surface potential change is the net electric double layer formed by the adsorbed water molecules on graphene.Computational studies have shown that the most stable configuration for the  adsorption of isolated water molecules orients the hydrogen atoms toward the graphene surface and the oxygen atoms outward [31][32][33][34] and is applicable to the adsorption on Stone-Wales defects [33].This orientation results in the formation of an electric double layer that increases the work function, which is consistent with the present results.The absence of dependence on each of the quantities on the horizontal axes in figure 6 suggests that the orientation of the adsorbed water molecules was not affected by variations in the experimental conditions of the present study.

Conclusion
In conclusion, the adsorption rate of water molecules on graphene surfaces can be tuned by controlling the wettability of the supporting substrate surface through wetting transparency.The kinetics of the change in the surface potential upon water adsorption were measured using a Kelvin probe with a large-diameter probe tip; the rate constant of the surface potential change was found to be larger for graphene formed on hydrophilic substrates than that on hydrophobic substrates.However, no clear correlation was observed between the rate constant and the Raman D/G ratio, indicating that the effect of wettability control was more significant than the differences in defect density.Furthermore, the position of the Fermi level before water adsorption showed no correlation with the rate constant, indicating that the effect of the charge-transfer interactions between the water molecules and graphene was negligible.This implies a negligible number of water-oxygen redox couples, a wellknown source of hole doping, because the Kelvin probe measurements were conducted under a nitrogen atmosphere.Therefore, a possible mechanism for the surface potential change is not the Fermi level shift of graphene due to charge transfer, but the net electric double layer formed by the adsorbed water molecules on graphene.
The adsorption rate of water molecules is directly related to the improvement in the sensor response rate.The introduction of defects such as hydrophilic groups onto the graphene surface has proven to be effective in increasing the adsorption rate, which in turn impairs the electrical conductivity.The present results indicate that the loss of electrical performance can be mitigated by exploiting wetting transparency, which is capable of increasing the adsorption rate.This finding provides a step forward in the realization of humidity sensors possessing a high affinity for water molecules, which take advantage of the high electrical conductivity of graphene.

Figure 1 .
Figure 1.Control of the water contact angle of graphene by substrate surface treatment.Representative results of water contact angle measurements for (a) hydrophobic and (b) hydrophilic samples.(c) Change in the average water contact angle upon heat treatment in Ar.Error bars represent standard deviation.

Figure 2 .
Figure 2. Kinetic measurements of the change in the surface potential upon introduction of water vapor.(a) Schematic representation of the measurement and a photograph of the fabricated sample.(b,c) Surface potential change of graphene on an HMDS-modified substrate.Results (b) before and (c) after annealing the sample in Ar for 2 h.Nitrogen flow was maintained at 200 ml min −1 during the measurement, and water vapor was added at 411 ppm during the time range between the dashed lines.

Figure 3 .
Figure 3. (a) Fitting function simulation of the surface potential change.(b) Dependence of the extracted rate constant on the water contact angle.The legends consist of the abbreviated names of modification molecule, sample number, and total annealing time in parentheses.

Figure 4 .
Figure 4. Effect of defects in graphene on the water adsorption rate.(a) Schematic of the increase in water adsorption energy due to defects.(b) Raman scattering spectrum of graphene on the APTS-modified substrate.The spectra were normalized to the G-peak intensity and shifted for clarity.The D/G ratios, which are a measure of the defect density, are also shown.(c) Plot of the extracted rate constant against the D/G ratio.The legends consist of the abbreviated names of modification molecule, sample number, and total annealing time in parentheses.

Figure 5 .
Figure 5.Effect of graphene work function on the water adsorption rate.(a) Schematic of charge-transfer interactions with adsorbed molecules.The surface potential increase observed in this study can be explained by the transfer of electrons from graphene to water.(b) Plot of the rate constant against the surface potential before water vapor introduction.The legends consist of the abbreviated names of modification molecule, sample number, and total annealing time in parentheses.

Figure 6 .
Figure 6.Plot of the variation in surface potential against the (a) water contact angle, (b) D/G ratio, and (c) surface potential before water vapor introduction.The legends consist of the abbreviated names of modification molecule, sample number, and total annealing time in parentheses.