Letter pubs.acs.org/NanoLett
Transfer of Chemically Modified Graphene with Retention of Functionality for Surface Engineering Keith E. Whitener, Jr.,*,† Woo-Kyung Lee,† Nabil D. Bassim,‡ Rhonda M. Stroud,‡ Jeremy T. Robinson,§ and Paul E. Sheehan† †
Chemistry Division, ‡Materials Science and Technology Division, and §Electronic Science and Technology Division, U.S. Naval Research Laboratory, Washington, D.C. 20375, United States S Supporting Information *
ABSTRACT: Single-layer graphene chemically reduced by the Birch process delaminates from a Si/SiOx substrate when exposed to an ethanol/water mixture, enabling transfer of chemically functionalized graphene to arbitrary substrates such as metals, dielectrics, and polymers. Unlike in previous reports, the graphene retains hydrogen, methyl, and aryl functional groups during the transfer process. This enables one to functionalize the receiving substrate with the properties of the chemically modified graphene (CMG). For instance, magnetic force microscopy shows that the previously reported magnetic properties of partially hydrogenated graphene remain after transfer. We also transfer hydrogenated graphene from its copper growth substrate to a Si/SiOx wafer and thermally dehydrogenate it to demonstrate a polymer- and etchant-free graphene transfer for potential use in transmission electron microscopy. Finally, we show that the Birch reduction facilitates delamination of CMG by weakening van der Waals forces between graphene and its substrate. KEYWORDS: Surface functionalization, Birch reduction, magnetic graphene, functionality transfer surface property. Subsequently, the 2D “designer” surface would be transferred onto the desired substrate. This approach effectively decouples a material’s surface chemistry from its bulk properties, enabling the surface and the bulk to be optimized independently to suit a given application. Several properties of graphene make it ideal for this endeavor. As noted above, its large surface-to-volume ratio means that van der Waals forces can robustly pin it to most surfaces, obviating the need to develop surface chemistry for the sensor material. Second, graphene chemistry is a young but rapidly expanding field which has already seen the introduction of chemical functionality on graphene ranging from simple atoms such as hydrogen3,4 and fluorine5,6 to organic groups7−11 to polymers12 and biomolecules.13,14 In effect, by using graphene, one could bring the flexibility of organic chemistry to almost any surface. The difficulty lies in that while several methods exist for transferring pristine graphene from substrate to substrate,15−23 transferring chemically modified graphene (CMG) without losing functional groups has been much more challenging. The transfer process itself might be to blame: etching the metal growth substrate with an oxidizing agent can induce defects in
C
hemically modifying a surface to impart new functionality is a widespread scientific and technological endeavor. While burying a substrate with a thick coating such as paint is well-developed, the routine generation of ultrathin films that allow interaction between the substrate and the exterior is more challenging. Consider the fabrication of gate dielectrics on III− V materials by atomic layer deposition or the functionalization of the surface of a chemical or biological sensor. The challenge is that materials with desirable electronic properties often have poorly developed surface chemistries (and vice versa), requiring extensive development of the surface chemistry before use. A common solution has been self-assembled monolayers (SAMs), short chain molecules that have different functional groups on either end of the chainone to generate the desired surface chemistry and one to bind to the surface. Examples of SAMs include thiol chemistry on gold,1 silane chemistry on silicon oxide,2 and a proliferation of others in the literature. Unfortunately, robust SAM chemistries are available for only a few materials, leaving many laboratories to expend significant resources exploring the surface chemistry of each new potential platform material. A radically different approach is to transfer a complete surface chemistry onto the surface of a substrate using a 2D material. Since the 2D material adheres to the substrate via van der Waals forces, the technique is relatively indifferent to the choice of substrate. Ideally, the 2D material would be functionalized with a set of chemical groups that give a desired This article not subject to U.S. Copyright. Published XXXX by the American Chemical Society
Received: December 11, 2015 Revised: January 14, 2016
A
DOI: 10.1021/acs.nanolett.5b05073 Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
the graphene sheet and leave behind metal ion contaminants, while complete removal of the polymer support is difficult if not impossible to achieve.24−26 For example, Lee et al. recently observed that hydrogenated graphene transferred from silicon oxide by first supporting the functionalized graphene with a polymer and then etching the oxide with strong base led to a complete loss of the hydrogen.27 In this paper, we present a simple method for retaining the chemical functionalities formed by the Birch reduction of graphene during transfer onto diverse substrates without supporting polymers or chemical etchantstwo prime sources of contamination. Related efforts by other groups have shown that pristine graphene adheres weakly to hydrogen-terminated silicon28 and germanium29 substrates, enabling its delamination under mild conditions and without chemical etchants. During our own research on Birch hydrogenated graphene (HG),3 we observed that HG spontaneously delaminates from Si/SiOx when exposed to water. We conjectured that hydrogenation of graphene was weakening the van der Waals adhesion forces between graphene and its substrate. If the observed delamination were controllable and repeatable, we reasoned that hydrogenation would provide a simple means to move CMG from one substrate to another. Herein we describe delamination of HG from a Si/SiOx substrate and subsequent transfer to arbitrary substrates. Critically, we show that functional groups such as methyl and diazonium-grafted aryl groups are retained after transfer and that even a film property as delicate as ferromagnetism in partially hydrogenated graphene30 (pHG) may be transferred. To take advantage of the clean thermal restoration of HG to pristine graphene,3,4 we also delaminate HG from its copper growth substrate and thermally anneal it to restore pristine graphene. The transferred pristine graphene is clean and uncontaminated by metal ions. The process requires trivial infrastructure (Figure 1a). The generation of HG from the Birch process is described in detail previously.3 A small dish containing an ethanol−water mixture is placed over a bright light source (in this case, the flashlight of a cell phone), and the room is darkened. A substrate (e.g., Si/ SiOx or Cu) with HG is dipped into the aqueous mixture, and the HG delaminates from the substrate immediately. The high light contrast at the surface of the water allows the free-floating HG to be just barely visible (Figure 1b). We note that pure water also delaminates HG but that adding a small amount of ethanol decreases surface tension and allows efficient relamination onto hydrophobic surfaces. We discuss this more at length in the Materials and Methods section. The free-floating HG can then be picked up from the solution with a wide range of substrates. Figure 1c shows an optical micrograph of an HG film (∼1 cm2) retrieved on a Si/ SiOx sample. A 500 μm × 665 μm Raman map of the integrated intensity of the defect (D) peak centered at 1325 cm−1 in Figure 1d indicates the extensive coverage of sp3 centers and shows that the graphene is contiguous across a large area, with some small holes, but few large tears or cracks, being introduced during the transfer process. More interesting is the straightforward transfer of the chemical functionality to dissimilar substrates. Figure 2 compares the Raman spectra of HG before and after transfer from Si/SiOx onto glass, aluminum foil, and poly(vinylidene fluoride) (PVDF). These substrates have a wide range of surface energies and range from very hydrophilic to very hydrophobic, illustrating the breadth of substrates that can receive delaminated HG. Transfer onto PVDF is particularly notable since PVDF is incompatible with
Figure 1. Hydrogen assisted graphene transfer. (a) Experimental setup: a shallow dish of ethanol/water solution sits atop a bright light source (in this case the flashlight on a cell phone). This image shows transfer of HG onto a PVDF wafer. (b) Close-up detail of HG on the solvent surface. The HG sheet itself is quite difficult to see, but the lighted edge (black arrow) shows up clearly. Dashed lines have been added around the approximate edges of the graphene sheet as a guide for the eye. (c) Optical image of HG transferred onto Si/SiOx (scale bar: 1 cm). In this case, the graphene has been thermally dehydrogenated to enhance optical contrast. (d) Raman map of D peak integrated intensity in transferred HG showing retention of functional groups. The dark spots are holes in the graphene, and the green and yellow areas are continuous graphene. The scale bar is 100 μm, and the resolution is 5 μm.
Figure 2. Table of Raman spectra of HG before and after transfer from Si/SiOx to glass, aluminum foil, and PVDF. Note: the sharp peaks present in the PVDF and glass spectra after the transfer are due to glass from either the substrate or, in the case of PVDF, the microscope slide, which appears due to the translucency of the polymer.
the Birch reduction. This emphasizes the point that a surface chemistry can be transferred in toto without consideration of chemistry of the substrate. Raman spectra confirmed that the graphene retains a high degree of functionalization after B
DOI: 10.1021/acs.nanolett.5b05073 Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
PVDF, again indicating robust retention of the functional groups throughout the transfer to a variety of surfaces. For aryl graphene, we first functionalized with nitrobenzene diazonium tetrafluoroborate and then applied the Birch reduction to hydrogenate the graphene. Incidentally, this step also reduces the nitro group on the aryl adsorbate to a primary amine, which could be useful for attaching biomolecules to graphene. An Xray photoelectron spectroscopy (XPS) mapping study was carried out on the transferred aryl-functionalized material. As shown in Figure 3, the transferred material retains a significant
transfer, regardless of whether the target substrate is dielectric, polymeric, or metallic in nature. The high-intensity D peak, overall peak broadening, and significant decrease in the intensity of overtone and combination bands are all qualitative signs of extensive sp3-bond formation (i.e., chemical functionalization) in the transferred graphene. An accurate measure of the molar amount of hydrogen on single-layer graphene is very difficult to obtain for two reasons. First, there is not enough hydrogen on a single layer to employ bulk methods such as thermogravimetric analysis, and second, surface characterization techniques are either insensitive to hydrogen (as is the case for X-ray photoelectron spectroscopy) or are easily contaminated with noisy signals (as is the case for infrared (IR) spectroscopy where hydrocarbons from the air often contaminate weak CH signals, and the deuterated stretch is expected to appear in the same region of the spectrum as CO2, a common IR contaminant). To overcome these difficulties and ensure that chemical functionality was not lost during transfer, Raman spectroscopy was used to track the extent of chemical functionalization at each stage. Specifically, a common and convenient quantitative measure of sp3 defect density is the ratio of the intensity of the Raman D peak at 1345 cm−1 to that of the G peak at 1585 cm−1, commonly called the D/G ratio.31 On the basis of this, we performed a modified analysis of defect density. Since HG n-dopes graphene3,32 and since doping shifts the Raman peak positions, the intensity at a single energy in the spectrum inaccurately measures the defect density of these samples.33 Moreover, as the Birch reduction introduces a very high degree of hydrogenation in the material, the defect-activated D′ peak at 1620 cm−1 and the G peak broaden into one another. Since the areal ratio of the D and D′ peaks is roughly constant (D′/D ≈ 1/13 for sp3 defects),34 we can use the integrated areas of the D and “G” (A“G” = AG + AD′) peaks to give a modified D/G ratio which follows roughly the same trend as the simpler ratio introduced by Lucchese et al. for their activation model of graphene defect density.31 In this paper, we denote fully hydrogenated graphene (HG) as having a D/G ratio of between 1 and 2, accompanied by a weak 2D peak. The original activation model in ref 31 suggests that our hydrogen density is >10 nm−2. We denote partially hydrogenated graphene (pHG) as having a D/G ratio indicating a lower density than this, accompanied by a strong 2D peak. We stress that this is only a rough approximation, as the activation model in ref 31 makes assumptions that may not apply to our system. Using our modified model for HG transferred from Si/SiOx to glass, the D/G ratio is 1.33 ± 0.05 before the transfer and 1.24 ± 0.02 after the transfer. For HG transferred from Si/SiOx to Al foil, the D/G ratio is 1.30 ± 0.01 before the transfer and 1.32 ± 0.07 afterward. Finally, for the transfer to PVDF, the D/G ratio is 1.30 ± 0.01 before the transfer and 1.30 ± 0.16 afterward. Consequently, all show robust retention of chemical functionality during transfer. The delamination technique is quite flexible and can transfer functionalities besides hydrogen. To illustrate this point, we prepared methyl- and aryl-functionalized graphenes for transfer. For methyl graphene, we quenched the standard Birch reduction with methyl iodide instead of ethanol. The Raman spectra show (see for example Figure S1 in the Supporting Information) a negligible D/G ratio change, from 1.38 ± 0.24 to 1.55 ± 0.19 for transfer from Si/SiOx to glass, from 1.56 ± 0.14 to 1.46 ± 0.30 for transfer from Si/SiOx to Al foil, and from 1.51 ± 0.33 to 1.66 ± 0.15 for transfer from Si/SiOx to
Figure 3. (a) XPS map of transferred aryl-functionalized graphene. The intensity of the N 1s peak at 399.6 eV is shown as a heat map. The bright yellow areas are aryl graphene, the orange area is physisorbed nitrogenous species on bare Si/SiOx, and the black area shows the edge of the wafer. The resolution is 200 μm. (b) N 1s region of XPS before (top) and after (bottom) transfer, showing retention of nitrogen-containing functional groups.
portion of its nitrogen content, with an N/C atomic ratio of 0.0254 ± 0.0014 before transfer and a ratio of 0.0221 ± 0.0039 after transfer from Si/SiOx to glass and an N/C ratio of 0.0222 ± 0.0010 before transfer and 0.0228 ± 0.0026 after transfer from Si/SiOx to Al foil. The relevant data for transfer to PVDF can be found in the Supporting Information. Thus, the transfer process boasts a retention of 87% of the chemical functional groups to a SiOx substrate and complete retention of functional groups to aluminum foil. Any groups that were lost during the transfer process were likely only tenuously physisorbed to the graphene in the first place, whereas the majority of the groups were strongly chemisorbed and therefore retained. Thin films are generally deposited to generate new surface properties.2,35,36 Several groups have demonstrated or predicted magnetism in CMGs,37−40 and our own group has shown previously that pHG is ferromagnetic using magnetic force microscopy (MFM).30 Here, we show that a ferromagnetic film prepared on one SiO2 substrate may be transferred onto another substrate while retaining its functionality. Figures 4a and 4b show AFM height measurements and the MFM magnetic response of pHG before transfer to the second substrate. Prior to imaging, we locally removed the pHG film with a razor blade to create a boundary between magnetic pHG and nonmagnetic SiOx. In Figure 4b, the SiOx area (left) displayed a negative phase shift (darker) against the pHG (right). This phase shift indicates that pHG responded to the cantilever’s magnetic field, but the exposed SiOx did not. We then confirmed the magnetic response after transferring the same film onto a second SiOx substrate, shown in Figure 4d. (Note that Figures 4b and 4d are not the same location of the sample.) Our results indicate that pHG remains magnetic after the transfer process. The contrast between magnetic and C
DOI: 10.1021/acs.nanolett.5b05073 Nano Lett. XXXX, XXX, XXX−XXX
Nano Letters
Letter
Figure 5. (a) Raman spectrum of transferred HG before thermal annealing. (b) Raman spectrum of transferred graphene after annealing. (c) Optical micrograph of annealed transferred graphene on SiO2 (scale bar: 10 μm). (d) Fourier-filtered annular dark field STEM image of annealed transferred graphene (scale bar: 1 nm) showing an area of graphene free from contamination. The raw image is in the Supporting Information.
Figure 4. (a, b) AFM height and MFM (-B polarized) response data for pHG on Si/SiOx before transfer. The left edge of the image was swiped with a razor blade to show the MFM contrast between magnetic pHG and nonmagnetic SiOx. (c, d) Height and magnetic response data for the same sample after transfer to a new Si/SiOx wafer. Prominent wrinkles are visible in the transferred samples in (c) and (d).
this sample before annealing was 1.33 ± 0.08 and after annealing was 0.45 ± 0.16, confirming that the hydrogen sp3 defect centers were significantly diminished on the graphene. It is worth noting that this ratio is highly sample dependent and can be somewhat higher, depending on the number of tears and holes introduced during the transfer process. For confirmation of cleanliness, we performed aberration-corrected scanning transmission electron microscopy (STEM) on samples transferred onto grids. STEM images prior to electron beam irradiation show small islands (tens of nanometers or smaller in diameter, 2 or 3 atoms thick) of surface hydrocarbon contamination, likely due to handling of the specimens in air during transfer to the TEM grids and loading into the microscope. Critically and unlike graphene transferred with a polymer support, exposure to the electron beam rapidly removes this surface contamination to leave atomically clean graphene monolayers. Moreover, the surface was free of the metal ion contaminants often seen in samples generated by etching the substrate. Indeed, residual impurity atoms on the graphene of any type were rare, and the few observed were all Si atoms as determined by energy-dispersive X-ray spectroscopy (EDS). The annealed graphene was further examined with optical microscopy (Figure 5c), STEM (Figure 5d), and AFM (Figure S2), with all techniques revealing a clean graphene surface which was continuous over length scales of ∼100 μm. Experiments are underway to transfer larger area films as well as to use STEM to further characterize the spatial extent of surface contamination on these samples. These experiments suggest that hydrogenating graphene considerably weakens its adhesion to the substrate. There are at least two mechanisms at play: first, the lithium and hydrogen atoms in the Birch reduction could interfere with the graphene/ substrate interface, and second, the van der Waals forces binding the graphene to the substrate might weaken upon graphene hydrogenation. We consider the first mechanism in the Supporting Information (see “Discussion of Mechanism”) and examine the second, more likely mechanism here in detail. One measure of van der Waals attraction between two substances is the Hamaker constant.41 Notably, the Hamaker
nonmagnetic regions is somewhat lower after transfer, but the signal is persistent. The sample clearly shows retention both of the Raman characteristics and of the ferromagnetism. Thus, our transfer technique represents a powerful way to render arbitrary surfaces ferromagnetic. Finally, a significant advantage of hydrogenated graphene over other common CMGs such as fluorinated graphene or graphene oxide is that the hydrogen may be removed cleanly by thermal annealing to restore essentially pristine graphene. Across 0.5 mm, the transferred and restored graphene has a resistance of 5.74 ± 1.40 kΩ vs 0.88 ± 0.35 kΩ across the same distance, so that electrical characteristics can be restored in transferred graphene to within an order of magnitude of the original material. Note that during transfer graphene is fully hydrogenated and so is an insulator. The slight increase in resistance after transfer and anneal is largely due to transferinduced defects in the graphene sheet. This thermal reversibility3,4 suggests a method to transfer graphene without a chemical etchant, such as ammonium persulfate, that often contaminates the sample with residual Cu2+ ions. To this end, we hydrogenated graphene on its copper growth substrate and transferred it onto a Si/SiOx wafer as described above. It was found that pHG does not delaminate from copper. Delaminating HG from copper was successful but more challenging than from Si/SiOx, yielding relatively small (
