1. Introduction

Graphene, a one atom thick allotrope of carbon, has been a center of attention for materials research over the past two decades due to its extraordinary electrical, optical, structural, and thermal properties [1]. The reliable isolation of a new class of two-dimensional materials like graphene has enabled the creation of previously unrealizable devices using atomically-precise crystal stacking [2]. Shaping such devices laterally is typically done through conventional lithographic means. However, the thinness of these devices allows for another technique of patterning: laser ablation. Laser ablation is a fast, single step process, that allows for resist-free patterning observable under a microscope in real-time.

The ablation of graphene has attracted interest in recent years [3,4], mainly for the purpose of creating thin ribbons of graphene between two or more cuts, with Zhang et al. achieving 25 $\mu$m wide ribbons [5], Kalita et al. 5 $\mu$m wide ribbons [6], and Sahin et al. 400 nm wide ribbons [7]. Following Stohr et al. [8], who reported the creation of 35 nm graphene wires using ablation, our work presented here uses a simple Ti:sapphire oscillator instead of the more complex and expensive amplified systems of these earlier authors.

Optical microscopy and Raman microspectroscopy are common analytic techniques in the literature to evaluate the ablation of graphene. Mortazavi et al. [9] studied a large range of cut widths depending on the utilized laser fluence, and used the combination of optical microscopy and Raman spectroscopy to study potential defects, noting disorder primarily along the edges of ablated areas. Atomic force microscopy (AFM) is sometimes used [10] to characterize cut quality, but is uncommon in the literature. Additionally, in the work by Stohr et al., the transport properties of graphene nanowires patterned by ablation were measured, however the measurements were not used to directly verify that the ablation creates electrically insulating regions.

Building on the work of others, we have patterned graphene by laser ablation and evaluated the quality of cuts by optical microscopy, electrical measurements, and atomic force microscopy. While optical and electrical characterization of the cuts indicated that the graphene has been entirely removed from the channels, careful topographic characterization of surface roughness using an AFM revealed that ablation in air can create significant structural defects in channels cut in graphene. By modifying the ablation process via submerging the graphene in a droplet of water, we demonstrated that these defects are largely prevented. Below, we directly compare ablation in ambient air conditions and in water immersion and display the improved quality of cuts processed in water.

2. Experimental

2.1 Graphene exfoliation and flake identification

We exfoliated graphene flakes from highly ordered pyrolytic graphite with adhesive tape onto 1 cm $\times$ 1 cm SiO$_2$ on Si wafer chips. Before transferring the graphene from tape to substrate, we cleaned the silicon chips with O$_2$ plasma, as this technique has been demonstrated to remove ambient impurities from the surface of the substrate and additionally can lead to a larger total surface area of adhered graphene flakes when compared with traditional mechanical exfoliation techniques [11]. Following exfoliation, the chips were heated at 120$^\circ$ C for 10 minutes. We identified regions of varying thickness by optical contrast microscopy (Nikon LV-UDM) and later confirmed their layer count by PeakForce tapping atomic force microscopy (Bruker Dimension Edge) [12]. The transport and scanning probe characterization reported on here was primarily performed on larger area bilayer samples, though consistent scanning probe characteristics were also observed on monolayer samples.

2.2 Ambient-air ablation setup

We mounted samples on an inverted fluorescence microscope (IX-71, Olympus) coupled to a motorized X-Y translation stage (ProScan III, Prior Scientific) (Fig. 1). A mode-locked titanium:sapphire oscillator (KM Labs Collegiate) with a central wavelength of 820 nm, a pulse width of approximately 300 fs, and a repetition rate of 90 MHz was used for the ablation. The laser was spatially filtered and slightly under-filled the back aperture of a 40$\times$ 0.75 NA objective before focusing on the substrate. The typical average power used was 100 mW at the sample, which, for a 2 $\mu$m diameter spot size, corresponded to a fluence of 35 mJ/cm$^2$. The samples were illuminated using a 590 nm LED (Thorlabs) and were viewed in real time on a CMOS camera (Thorlabs DCC1645C). The shutter and motion pattern of the stage were controlled with custom software written in LabView (National Instruments). All cuts were patterned at a speed of 1 $\mu$m/s.

 

Fig. 1. Laser ablation schematic. The ultrafast laser is coupled into an inverted microscope equipped with a computer-controlled x-y stage. The sample is viewed in real time during the ablation using reflected illumination from an LED. The inset shows how the sample is placed underwater by sandwiching a drop of water between the sample and a coverslip; the coverslip and water are not used for ablation performed in air.

Download Full Size | PDF

2.3 Underwater ablation modification

Underwater ablation was nearly identical to ambient-air ablation, but modified by pipetting a droplet of deionized water on the sample and securing a #1 coverslip (Corning) over the drop so that the water was approximately 50 $\mu$m thick. The coverslip allowed us to continue using the 40$\times$ 0.75 NA objective without needing to immerse the lens in water as well. The objective did not have a correction collar and so the focus was not corrected in switching from air to water. We note that our measured fluence threshold for water ablation, 8 mJ/cm$^2$, was half the level needed for ablation in air.

2.4 Gold lead deposition

Gold leads were patterned by PMMA-resist electron beam lithography (Tescan Mira) and thermal evaporation of 1 nm/80 nm Cr/Au (Edwards 306).

2.5 Conductance measurements

After wire-bonding the leads to a conventional DIP circuit package, two-terminal conductance measurements were performed using a lock-in amplifier (SRS 830) outputting a low-frequency (17 Hz) $100 \mu {V}$ AC bias. The Fermi level of the graphene was independently modified using a DC bias gating voltage (Keithley 2400) connected between the graphene and the doped Si substrate, which were separated by the 285 nm insulating layer of SiO$_2$.

3. Results and discussion

Lithography via laser ablation is a fast method of selective lateral patterning by the removal of material. The technique involves focusing an ultrafast laser beam through a microscope objective onto a sample. Our process is performed through an inverted microscope so that it can be viewed and controlled in real time. Figure 2(a) shows a flake of graphene during ablation processing, with the glowing plasma spot at the focal point. One advantage laser ablation has over comparable lithographic techniques is that it requires fewer steps. A sample can be placed on a microscope and patterned in minutes without the need to spin coat or develop photoresists. Additionally, its flexibility affords the freedom to pattern electrical leads or other conventional lithographic structures onto the graphene before or after ablation, allowing for the measurements of electrical properties at distinct points in the life-cycle of a sample. To demonstrate this ability, Fig. 2(b) shows gold leads attached to closely-spaced (<2 $\mu$m center-to-center distance) submicron graphene ribbons after their fabrication, while Fig. 3(c) shows an ablation cut through graphene after leads are already attached to it.

 

Fig. 2. (a) A flake of graphene halfway through ablation processing. The arrow indicates the direction of the ablation path traced out by the bright laser spot. (b) The same flake of graphene post-ablation and after electron beam lithography and thermal evaporation were used to attach gold leads to the fabricated submicron-thin graphene wires.

Download Full Size | PDF

 

Fig. 3. (a) Unablated flake of graphene with gold leads used to measure two-terminal electrical transport; the source and drain are indicated on the image. (b) Before ablation, the room-temperature conductance measurement as a function of gating voltage shows graphene’s characteristic conductance minimum at its charge neutral point as its two-terminal resistance peaks at 33 k$\Omega$. (c) The same flake after ablation has been used to sever the connection between source and drain. (d) Following ablation the $\gtrsim {10}~ M\Omega$ measured resistance is indistinguishable from the the leakage impedance of our measurement setup, indicating the ablating cut is electrically complete.

Download Full Size | PDF

We used the processing flexibility of ablation to our advantage in studying the electrical completeness of ablated cuts using two-terminal transport measurements – to our knowledge the first transport test of ablation patterning. Measuring the conductivity of a graphene flake before and after bisecting it via ablation we found that its initial gate-dependent conductance in the 30-40 $\mu$S range was reduced, within our measurement error, to 0 nS (Fig. 3).

The combination of optical inspection and transport measurements would seem to confirm that the graphene was cleanly removed. However, careful topographical AFM images of these ambient-air-ablated samples did not, in fact, appear complete. An example of this effect is shown in Figs. 4(a), 4(d), and 4(g), where there are edge defects higher than the graphene height as well as trenches higher than the substrate-level SiO$_2$ in the ablated regions, despite optically clean cuts. Hypothesizing that ablation in an aqueous environment would mitigate heat damage or deposition, we modified our technique so that the ablation could be carried out while the samples were immersed in water. The use of water resulted in cuts that appeared optically identical to the air cuts, but now topographically showed channels milled down to the substrate level (Figs. 4(b), 4(e), and 4(h)). To directly compare ablation in air and water, we performed each type of cut on the same sample, confirming a significant difference in the surface roughness of the channels produced by each method. Figures 4(c), 4(f), and 4(i) show a flake with optically identical perpendicular cuts (vertical in air, and horizontal in water), while the topographic scan along a channel demonstrates a disparity between them. The height data in Fig. 4(i) along the water-ablated channel indicates a step between the floor of the channel processed in water (in this case, at the SiO$_2$ level) and the floor of the channel processed by a cut in ambient air. The defects following ablation in air are both higher than the SiO$_2$ surface but lower than the graphene, indicating that it is atomic-scale defects that are mitigated by the use of water. We also note that the roughness is persistent at the intersection of the air and water cuts, even though the water cuts were performed after the air cuts. This implies that the roughness is not simply due to partial graphene ablation, since any remaining graphene should be removed by the ablation in water; rather, the surface has been altered in such a way that further processing by laser ablation is insufficient to restore complete channels. Trying to remove remnants from incomplete cuts in air with solvents and thermal annealing also proved unsuccessful.

 

Fig. 4. Optical images were taken of a serpentine pattern processed in (a) ambient air and (b) water, as well as (c) perpendicular air/water channels patterned on the same sample (red arrows point along air channels, blue arrows point along water channels). (d-f) These patterns were subsequently scanned across their full areas with an AFM. (g-i) Representative height data is shown with the colored dotted boxes in (d-f) indicating the scanned area corresponding to the similarly colored data. (g) Air cuts reveal highly disordered ablation trenches when compared to (h) water cuts. (i) Scans across air and water trenches show different post-ablation substrate heights. The water cut (blue scan box) produced a trench at the unexposed SiO$_2$ level, while the air cut trench (red scan box) is above it. A scan within the water trench intersecting an air cut (purple scan box) highlights the $\sim$0.5 nm difference in trench height. The insets of (e) (of the boxed orange region) and (f) are scans of the same samples performed with the AFM probe moving at 90$^\circ$ relative to the main image to cross-check for height artifacts due to scan direction.

Download Full Size | PDF

The improvement of ablation processes performed in water, both in terms of higher efficiency and reduced debris, has been observed in other materials both in bulk and thin-film forms [13–16]. Multiple mechanisms for water-assisted improvement have been studied, and may be acting individually or collectively in our experiment: At sub-ns time scales the dominant thermal transport mechanism for conductive materials is electronic rather than phononic [17]. In our case the graphene is between electrically insulating air and SiO$_2$, which will lead to less efficient dissipation of heat and a higher probability of thermally-driven reactions at the surface level. The presence of water thus provides an additional conduction channel to reduce these effects. At ns time scales the plasma of ablated material that forms and expands above the surface is altered and contained in microbubbles [18,19], which helps protect the freshly ablated region from charge and heat. On the $\mu$s time scales of ion recombination, the low reactivity and high polarity of water molecules allows radicals and reaction products [20] to be carried away from the ablated surface. We performed additional tests ablating graphene in other liquids with comparable conductive properties but varying viscosities and polarities (methanol, acetone, propylene glycol, hexanes), which did not show marked improvement over ablation in air, suggesting that it may be multiple mechanisms acting in concert that advantages water.

4. Conclusion

We have demonstrated the ability to quickly pattern graphene by ablation using a fairly inexpensive setup and to characterize the quality of the cuts made in air and in water optically, electrically, and topographically. We found that ablation in air removed the graphene and produced an insulating channel, but left behind a roughened substrate, only revealed by AFM imaging. Ablation performed under water prevented this roughening and resulted in cleaner samples. The potential for ablating in alternate environments may be useful for patterning complex nanostructures out of both graphene and the many other 2-D materials that are sensitive to oxygen environments.