Stability of suspended monolayer graphene membranes in alkaline environment

ABSTRACT We report on the structural, chemical and electrical compatibility of graphene in an alkaline environment, namely diluted KOH solutions as those used in standard nanofabrication methods. Electron microscopy and Raman spectroscopy indicate that graphene preserves its structural properties during the process. The electrical transport properties such as conductance and shape of the Dirac point, appear degraded after KOH treatment, but can be fully recovered by current annealing. Electron beam-induced etching can provide further nanofabrication subsequent to KOH processing. These results demonstrate that, with the due precautions, graphene-based structures can be integrated in standard fabrication processes involving KOH. GRAPHICAL ABSTRACT IMPACT STATEMENT To be implemented in novel (nano) devices, graphene must be compatible with standard fabrication techniques. We show how graphene’s intrinsic properties can be restored after exposure to an alkaline environment.

A relevant aspect for the application of both supported and suspended graphene as a component in modern electronic devices or M/NEMS is the possibility to integrate it with both well-established technologies, namely Si-technology, and more recently developed ones such as flexible electronics [22,23]. As  to assess the chemical and physical compatibility of graphene to the fabrication steps involved in these technologies. Concerning chemical processes, for instance, graphene has shown very high stability after immersion in both pure and buffered HF (BOE), 1 while, as a pure layer of carbon, it is damaged by immersion in sulfuric acid, or so-called piranha etchants.
It is thus of interest to study the interaction of KOH with graphene in order to assess the compatibility of KOH-based processes with graphene-based devices. Previous studies on KOH interaction with graphene were limited to its role as an electrolyte [30][31][32] as well as an agent for chemical modification of graphene [33] for supercapacitive applications, or as a processing step to remove graphene from SiO 2 [34,35]. However a comprehensive study on the interaction of KOH processing on the structural and electrical properties of micronsized single layer flakes is missing. In this paper we present an investigation of the modification of the structural and electrical properties of suspended graphene microribbons after immersion in a KOH solution and successive rinsing in water.
Commercially purchased chemical vapor deposition (CVD) graphene 2 deposited on a 300 nm Si/SiO 2 substrate was patterned into microribbons (2.5 μm × 20 μm) using optical lithography, followed by O 2 plasma ashing. The microribbons were contacted using standard electron beam lithography and metal deposition (Ti/Au 5/120 nm) followed by lift-off in acetone [36]. A second electron beam lithography step was used to create a rectangular opening in PMMA around the patterned graphene. This opening serves as a mask for underetching the SiO 2 and thus create an electrically contacted, suspended strip of graphene [37,38].
The under-etching was achieved by immersing the sample in buffered HF, followed by rinsing in water at room temperature (RT). The samples were then immersed in a 5-20% water-diluted KOH solution at RT to 60°C, for 1-3 minutes. In order to remove the PMMA mask, the samples were transferred in semiconductor grade acetone for 5-10 minutes, and isopropyl alcohol (IPA) for 5 minutes at the same temperature of the KOH solution. After a final rinsing for 5 to 10 minutes in deionized water at the same temperature as previous solvents, the samples were finally transferred to hot IPA (60°C). This solution is allowed to evaporate, and the low-surface tension of warm isopropanol ensures that capillary forces do not crack the suspended graphene devices during evaporation.
SEM pictures of graphene microribbons, suspended over Si (Figure 1(a)) and SiO 2 ( Figure 1(b)) after processing in KOH show that in both cases the overall process does not cause visible structural damages or deformation in the graphene flakes. Figure 1(c,d) 3 show schematically the top and crosssectional view of the suspended graphene device we fabricated: W, and L are the width and length of the suspended graphene between the two contacts, and h is the thickness of the removed SiO 2 after exposure to buffered HF, and d is the thickness of graphene [5] All devices exhibit similar results in the investigated range of parameters (temperature, KOH concentration, immersion time . . . ).
To investigate further the structural modification of samples during KOH treatment, Raman spectroscopy was performed on the sample using a commercial set-up at the intermediate stage of the process, i.e. immediately  after suspension, before immersion in KOH, and at the final stage, after exposure to KOH. The Raman spectra of suspended microribbons with and without KOH treatment are shown in Figure 2. The spectra were taken at the same nominal conditions and on the same position of a suspended graphene microribbon using a laser at a wavelength of 532 nm as a source and a spectrometer with 2400 l/mm. The characteristic Raman peaks [39] D, G and 2D (also called G ) of graphene are clearly visible in both the spectra. For a better evaluation of the figure, the intensity of the spectra has been normalized so that the intensity of the G peak is equal to I(G) = 1000 in both spectra.  Table 1 shows the numerical values of the principal peaks of the spectra shown in Figure 2. The position, full width half maximum (FWHM) and intensity of the peaks were calculated by fitting the experimental data with single Lorentzian curves. The ratio between the intensity of the 2D and G peak I(2D)/I(G) is about 1.81 ± 002, indicating that graphene conserves its structural properties [39] after immersion in KOH. Both the position of the G and 2D peaks show a modest blueshift of 1.1 cm −1 after KOH treatment, which suggests a slight residual pdoping [40] as also confirmed by the reduction of the FWHM of the G peak after immersion.
The ratio between the intensities G and D peak is around I(G)/I(D) = 5.5 before and I(G)/I(D) = 6.1 after KOH, while the FWHM of the D peak increases by 42% after KOH treatment. By applying a recent version of the Tuinstra and Koenig model [41], we can estimate the crystallite graphene size as L(nm) = (2.4 × 10 −10 )λ 4 laser (I G /I D ) to be 107 and 118 nm, respectively. We explain the increase in crystallite size by the fact that the D peak is highly edge sensitive [42], and residual KOH might fill vacancies or gaps in the defective structure of suspended graphene, but further studies would be necessary for a better understanding of the phenomenon.
In order to establish the effect of immersion of KOH on the electrical properties of suspended graphene, we bonded a suspended sample after full treatment in KOH solution to a chip carrier and inserted it in an SEM with external electrical connections. This system has the advantage that SEM pictures can be taken in situ at any time before, during or after measurement while keeping the sample at the measurement conditions. The electrical measurements took place at RT and at a pressure of P = 5 × 10 −7 mBar. A constant potential difference (source drain voltage) of V SD = 100 mV was applied between the two contacts holding the suspended graphene device, while a backgate voltage applied to the conductive Si substrate was swept in the range −15V ≤ V BG ≤ +15V. The black curve in Figure 3(a) shows the resistance as a function of the backgate voltage of the KOH-treated sample. The sheet resistance is high, and the Dirac neutrality point is not visible in the explored backgate voltage range. This is due to the presence of residuals and impurities on the surface of graphene, which act as scatterers and degrade the conductance [43,44]. In order to improve the electrical properties of graphene, we performed current annealing [45], that is the application of a controlled large current to recover its electrical properties. In our case, we applied an increasing source drain voltage in steps of V SD = 100 mV and monitored the evolution of the device resistance: after each step we waited for the resistance to stabilize and then increased the applied voltage; the annealing process was completed, when successive steps did not show a significant change in resistance. For this device, we increased the voltage up to V SD = 3.2V, after which we noticed that the resistance did not decrease any further, a hint that current annealing was complete. This voltage corresponds to a current of I = 590 μA and a power P = 1.9 mW.
The electrical measurements of the resistance of the same sample after current annealing are shown as the red curve in Figure 3(a). Now a well-centered Dirac point with a sheet resistance of the same order of magnitude as in pristine graphene [3] is observable. This is due to the fact that current annealing removes effectively the KOH residues and other impurities from the surface of suspended graphene, as shown by the SEM images of the measured flake before (Figure 3(b)) and after ( Figure  3(c)) current annealing. In Figure 3(c) it can be observed that current annealing was pushed a bit too much and the graphene layer started to crack at the edges [46].
Characteristic electronic parameters of the cleaned graphene device, such as contact resistance R cont , mobility μ, and residual carrier concentration n 0 can be extrapolated by fitting [47] the gate-dependent resistance data with the formula: where R chan is the resistance of the graphene channel region, N sq the length-to-width ratio (number of squares) of the graphene device, e the charge of electron, and n tot the carrier density in the graphene channel region. The latter can be expressed as n tot = n 2 0 + n 2 BG that is the combination of n 0 , which is the residual carrier densities generated by defects and charge impurities, which are present in non-ideal graphene, and the voltage dependent carrier density given by n BG = (C eq /e)(V BG − V Dirac ). The equivalent capacitance C eq is calculated as the series of the two capacitors C eq = (h/ 0 + (d SiO 2 − h)/ 0 SiO 2 ) −1 , representing the etched layer (thickness h) and the remaining SiO 2 layer (thickness d SiO 2 ) between graphene and the Si backgate. Applying the fitting to the electrical measurements of the cleaned graphene device, for which h = 115 nm, and N sq = 0.73, we extrapolate μ = 3800 cm 2 /Vs, R cont = 4600 , and n 0 = 4.1 × 10 11 cm −2 .
These values are comparable with those of reference samples, which were fabricated from the same CVD graphene and with the same fabrication and annealing conditions, but without the immersion in KOH. Thus, the low value of mobility and the high value of impurity carriers compared to the best suspended graphene samples [5], should be attributed to the overall quality of processed CVD graphene, and not to the immersion in KOH.
Finally, we show that suspended graphene, after being treated after KOH processing, can still be directly patterned using resist free nanofabrication techniques such as water assisted electron beam-induced etching (EBIE) [37]. EBIE is a technique based on the selective etching of graphene by selectively irradiating parts of the sample with a focused electron beam in the presence of a water vapor environment. The combined effect of the electron beam and vapor adsorbate on graphene induces a chemical reaction which etches the irradiated parts of suspended graphene with a resolution of 7 nm [37]. To pattern the samples using this technique, we used an acceleration voltage V = 30 kV, a current I = 0.69 nA and a dose D = 10 μC/μm2. This dose is higher than the one needed for pristine suspended graphene, (4 μC/μm2), because KOH residue forms a protective layer on the surface of graphene. Figure 4 shows the successful patterning of nanoscopic structures such as nanoribbons (a), asymmetric constrictions (b), and quantum dots (c). Interestingly, after immersion in KOH, graphene acquires a remarkable stiffness, and after the cut, graphene remains suspended, even when attached to only one contact (see Figure 4(a)).
In summary, we have shown that suspended graphene is structurally stable after immersion in a KOH solution and subsequent rinsing in water. The Raman spectra before and after immersion show that the structural properties of graphene are preserved, apart from a limited residual doping. Residual contamination of KOH degrades the electrical properties of graphene with respect to pristine conditions, causing a high-sheet resistance, and disappearance of the Dirac point, however, intrinsic conditions can be fully recovered by current annealing. Furthermore, suspended graphene covered with a residual layer of KOH can be efficiently patterned by EBIE with nanometric precision, providing an additional possibility to subsequent nanofabrication.