F. Yu,
A. C. Stoot,
P. Bøggild and
L. Camilli*
Technical University of Denmark, Department of Micro- and Nanotechnology, Kgs. Lyngby, DK-2800, Denmark. E-mail: lcam@nanotech.dtu.dk
First published on 17th February 2016
Being impermeable to all gases, graphene has been proposed as an effective ultrathin barrier film and protective coating. However, here it is shown how the gastight property of graphene-based coatings may indirectly lead to their catastrophic failure under certain conditions. When nickel coated with thick, high-quality chemical vapor deposited multilayered graphene is exposed to acidic solutions, a dramatic evolution of gas is observed at the coating–substrate interface. The gas bubbles grow and merge, eventually rupturing and delaminating the coating. This behavior, attributed to cathodic hydrogen evolution, can also occur spontaneously on a range of other technologically important metals and alloys based on iron, zinc, aluminum and manganese; this makes these findings relevant for practical applications of graphene-based coatings.
The hexagonal lattice of defect-free monolayer graphene has been proved both theoretically and experimentally to be impermeable to all liquids and gases including the smallest molecules, hydrogen and helium.2–4 This outstanding feature has led to an enormous interest in employing graphene as an anticorrosion coating for metals and alloys.5–7 Nevertheless, chemical vapour deposited (CVD) monolayer graphene generally exhibits defects through which molecules and radical species can diffuse, thus locally initiating metal corrosion.8–10 In addition, it has been reported that once these local corrosion processes begin to take place underneath a monolayer graphene cover, they are actually even accelerated by the presence of graphene itself.11 In this context, using a film made of several layers instead of a monolayer is one logical approach to improve performance of graphene-based protective coatings.12–14 In this work, however, we report a so far unnoticed issue related to coatings with low permeability and high structural integrity. Indeed, when nickel coated with a thick, high-quality multi-layered graphene is immersed in an acidic solution, the cathodic reaction spontaneously yields hydrogen at the nickel surface. The formed gas is not able to escape the high quality regions of the graphene film and therefore remains trapped and eventually forms bubbles, which can even lead to the delamination of the whole coating.
Transfer of graphene coating onto glass slide or ∼88 nm SiO2 wafer was carried out via chemical etching of nickel in 5% HCl and 30% H2O2 mixed solution for 24 hours.
Optical images were taken with a Nikon Eclipse L200N optical microscope, while Raman spectra were collected by a Thermo Fisher Scientific DXR Raman microscope (excitation wavelength 455 nm). Quanta 200 FEG environmental scanning electron microscope (SEM) and FEI Titan T20 G2 transmission electron microscope (TEM) were employed for the electron microscopy characterizations.
Nickel and its alloys are usually corrosion resistant in neutral, alkaline and diluted acidic media, while they deteriorate in aerated aggressive acidic environment. Hence, to investigate the corrosion performance of MLG coatings on nickel, we employ standard potentiodynamic polarization technique in 0.5 M HCl solution (pH = 0.3). With this method, by sweeping potential from negative to positive values with respect to the open circuit potential, the sample is electrochemically polarized, which allows us to obtain information about thermodynamics (i.e., corrosion potential) and kinetics (i.e., corrosion current density and corrosion rate) of the corrosion process.18 Here, we estimate that the corrosion rate for the MLG-coated nickel is less than half that of the bare nickel specimen (see ESI† for more details). This trend is also confirmed by electrochemical impedance spectroscopy analysis (see ESI†) and is in agreement with previous studies on coatings based on few-layer graphene on nickel.19 During the test, as expected, we could notice formation of hydrogen bubbles on the surface of both samples at the cathodic branch of the polarization curves (Fig. 2a), according to the reduction reaction:
2H3O+(aq) + 2e− → H2(g) + 2H2O(aq) | (1) |
We further investigated the specimen surface by means of scanning electron microscopy (SEM). In Fig. 2b we report an example of an area where the MLG coating was locally delaminated. The approximately circular shape of the hole in the coating suggests that it has been caused upon rupture of a hydrogen bubble. However, the outward direction of the remaining MLG flakes at the surface near the edges of the hole suggests that the bubble was not located on the surface of the coating, but rather at the interface between it and the nickel substrate.
The formation of hydrogen bubbles at the interface between coating and substrate in aggressive acid environment is a known phenomenon in corrosion science.20,21 Also, it is worth reminding that, though under different conditions, the evolution of hydrogen bubbles at the interface between graphene and metal substrate during the cathodic reaction is a method largely used for graphene transfer.22
To investigate whether the hydrogen bubbles are being generated both at the surface of the MLG film and underneath, we applied a fixed negative potential (−0.6 V vs. Ag/AgCl) to the coated nickel sample in 0.5 M HCl solution for only a short period of time (3 min). In this way, the bubbles can be observed at the interface before they become large enough to burst. As soon as the potential was applied, numerous hydrogen bubbles appeared at the surface of the coating. After this test, the sample was rinsed with water and gently blow-dried with nitrogen. Under the optical microscope, a few nearly circular protrusions of approximately 20 μm in diameter were found at seemingly un-correlated locations under the graphitic film. To prove that these protrusions are actually blisters, i.e. trapped gas underneath the graphene film and not particles or contaminants present on the surface, we characterized these structures by scanning electron microscopy. Fig. 2c displays two such protrusions that indeed seem to be blisters of the coating. Notably, the characteristic graphene wrinkles23 at the base of the blisters gradually unfold near the center of the blister, as would be expected. To demonstrate that these blisters can become large enough to burst and locally delaminate the coating, we apply a fixed negative potential (−0.6 V vs. Ag/AgCl) to the coated nickel sample in 0.5 M HCl solution for 30 min. As displayed in Fig. 2d, SEM analysis shows that in many places, the coating is no longer covering the metal substrate, similar to what was found after the potentiodynamic scan experiment, which is another indication that the film has now ruptured and delaminated. Raman spectroscopy was also used to prove that no graphene is present in the delaminated areas (see ESI†).
Raman spectroscopy is also known to provide information regarding mechanical stress in the graphene lattice. A significant blue-shift of all the peaks in the spectrum has been indeed reported for the case of micro-balloons made of single and bi-layer graphene on a oxidized silicon substrate.24 However, within our samples, we do not observe any remarkable shift of the peaks in the Raman spectra either recorded at the center of the blister, at its base, or away from the blister on a flat graphene region (see ESI†). This can be understood looking at Fig. 2c, where the wrinkles seem to unfold at the center of the blister, thereby relieving any stress caused by the trapped hydrogen. The wrinkles are naturally formed upon cooling after the growth process as a consequence of the different thermal expansion coefficients of graphene and nickel;23 likewise the unfolding of wrinkles is a way to release the mechanical stress caused by the formation of the blisters.
To gain more insights into the formation of the hydrogen bubbles below the MLG film, we set up two experiments for real-time monitoring of the process. The first experiment consists of placing a droplet of 0.5 M HCl solution in the center of the MLG-coated Ni foil under an optical microscope. Already after 5 min, few blisters can be seen randomly distributed over the imaged area, initially with diameters ranging from 10 μm to 20 μm (Fig. 3a). These blisters grow over time due to a build-up of hydrogen at the interface between nickel and the coating until they eventually start to merge. This behavior continues as long as the HCl droplet is present on top of the MLG film. After two hours the blisters can be larger than 100 μm and locally lift up the coating (Fig. 3a). It is worth noticing that the time scale for hydrogen bubble formation actually varies depending on the sample under study. Interestingly, once the acid droplet is removed, the region of the sample under the droplet does not seem to be visibly corroded, i.e. there is no noticeable change in color or appearance. On the other hand, the region of the sample surrounding the droplet, which is thus not in immediate contact with the acid, exhibits remarkable signs of degradation (see Fig. 3b and c, respectively). If the acid is distributed all over the sample area, the blisters eventually cause delamination of the whole coating (see ESI†).
In the second experiment, SEM is used in order to study the hydrogen bubble formation in real-time with higher spatial resolution. A droplet of 0.5 M HCl solution is placed in the center of the MLG-coated nickel foil, and gently removed with a tissue. The sample is mounted inside the scanning electron microscope without prior rinsing or blow-drying. When the electron beam is focused on a flat region of the sample, a dramatic hydrogen evolution takes place below the MLG film (Fig. 4).
At first, a single micrometer-size blister is observed, then, other smaller blisters are formed all over the scanned area (Fig. 4b and c, respectively). At lower magnification, it is possible to see that the blister-like protrusions are mainly found in the area that was initially irradiated by the scanning electron beam (see ESI†). This might be explained by a monolayer of the acidic solution still being adsorbed on the surface25 as well as within defects and crevices of the MLG film after the sample has being placed inside the microscope and evacuated. Then, the local heating and the creation of defects26 induced by the electron beam trigger and accelerate the diffusion of water through the MLG coating thus giving rise to the hydrogen bubble formation. This experiment points out that (i) the fact that such gas evolution only occurs with samples that have been exposed to HCl solution rules out that this phenomenon is caused by the presence of residual gas trapped either between graphene and nickel or into the nickel itself during the CVD process; and (ii) the trapping of gas at the graphene–metal interface in vacuum conditions inside the SEM chamber indicates that the overall quality and integrity of the film is high, as also corroborated by Raman spectroscopy investigation (see Fig. 1c).
To explain why hydrogen is formed at the interface, we suggest that, even if few, local defects and inhomogeneities are intrinsically present in the coating. Notably, such inhomogeneities are areas with higher defect density, grain boundaries,27,28 and sometimes also fewer layers (see ESI†). These areas may originate from local impurities on the nickel surface, temperature gradients during the synthesis process or different catalytic activity of the nickel grains, as already reported in the literature for CVD grown multilayer graphene.29 In this scenario, as the coating is placed in contact with the acidic solution, water fills these defects and inhomogeneities thus forming a bridge which allows protons to shuttle from outside the coating to the nickel surface, through the hydrogen-bonding network (Fig. 5), similar to the case of proton transport through the channel of carbon nanotubes.30,31 As water reaches the nickel, the cathodic reaction locally takes place. In fact, as described by the potential-pH diagram of nickel/water system, also known as Pourbaix diagram,32 the evolution of hydrogen at the nickel–water interface is spontaneous at the low pH level used in this study. As the MLG coating on nickel shifts the open circuit potential of nickel by only −25 mV (see ESI†), the hydrogen evolution process for the MLG-coated nickel is still spontaneous, in accordance with our observations. Once hydrogen is formed at the coating-metal interface, we can assume it diffuses on the nickel surface, as observed in similar systems,33 and merge, thus giving rise to the observed blisters. Since the majority of the coating is made of regions with very small apparent density of defects, a build-up of hydrogen occurs under the MLG film, produces bubbles that grow and merge, and eventually delaminate the coating. The formed hydrogen is not likely to pass through the coating via local defects and inhomogeneities, since these are clogged by water molecules, as previously suggested for helium-leak-tight coatings made of graphene oxide.34 The fact that the formation of hydrogen bubbles takes place on a different time scale depending on the sample under study could be explained by varying density of inhomogeneous areas within the coating, which is easily accountable by variations of the nickel foil substrates or inevitable differences in the exact growth conditions commonly experienced for CVD graphene growth. Here, it is worth pointing out that proton transport through a defect-free single layer of graphene has already been observed, but not through a bilayer.35 In this study we report on the performance of a coating comprised of hundreds of graphene layers, and therefore cannot immediately adopt the explanation proposed by Hu and co-workers,35 where a tunnel mechanism was invoked.
The dramatic degradation of the sample in the regions surrounding the acid droplets can be ascribed to the formation of a galvanic cell between nickel and graphene. Every electrochemical reaction consists of a cathodic and an anodic part. In the case of the acid droplet deposited on the middle of the MLG-coated sample, the cathodic reaction (i.e., reduction of protons to hydrogen) occurs under the droplet due to the abundance of protons, while the anodic one (i.e., oxidation of the nickel) takes place away from the droplet. While reducing protons to hydrogen, electrons are being continuously depleted in the region under the droplets. In this picture, electrons will move from the region surrounding the droplet, to the region below the droplet in compensation. Here, as already reported in the case of single layer graphene grown on copper foil,11 the electrons may migrate through the graphene film due to its high conductivity, rather than through the passivated metal surface, thus giving rise to formation of a galvanic cell which spreads the oxidation of nickel to all regions in contact with graphene.
Ultimately, in order to verify that the formation of hydrogen beneath a graphene coating is a reaction occurring spontaneously in the system nickel/water at low pH values, regardless of the anions that are present in solution, we repeat the experiment of the droplet placed at the center of the coated substrate using different acids. At this aim, a MLG-coated nickel foil is cut into three pieces. The first piece is tested with HCl, the second one with H2SO4 whereas the third one with HNO3, all solutions being at the same concentration (namely 0.5 M). As displayed in Fig. 6, after 240 min of exposure, the coating results in being lifted up in several areas by hydrogen in all the three cases, regardless of the acids that was used.
On the other hand, however, it is also worth pointing out that (i) the possibility of spontaneously producing and effectively trapping hydrogen (or other gases) calls for further investigations of such MLG film in fields such as gas storage and production; and (ii) the capability of such membranes of being selectively permeable to water might be of great interest for particular environmental applications.36
Footnote |
† Electronic supplementary information (ESI) available: Polarization curves, electrochemical impedance spectroscopy, further Raman characterizations of the coating and the video of the real-time microscope experiments. See DOI: 10.1039/c6ra01556e |
This journal is © The Royal Society of Chemistry 2016 |