NANO LETTERS
Patterning Graphene at the Nanometer Scale via Hydrogen Desorption
2009 Vol. 9, No. 12 4343-4347
Paolo Sessi,† Jeffrey R. Guest,‡ Matthias Bode,‡ and Nathan P. Guisinger*,‡ Center for Nanoscale Materials, Argonne National Laboratory, Argonne, Illinois 60439, and CNISM-Dipartimento di Fisica, Politecnico di Milano, I-20133 Milano, Italy Received August 10, 2009; Revised Manuscript Received October 7, 2009
ABSTRACT We have demonstrated the reversible and local modification of the electronic properties of graphene by hydrogen passivation and subsequent electron-stimulated hydrogen desorption with an scanning tunneling microscope tip. In addition to changing the morphology, we show that the hydrogen passivation is stable at room temperature and modifies the electronic properties of graphene, opening a gap in the local density of states. This insulating state is reversed by local desorption of the hydrogen, and the unaltered electronic properties of graphene are recovered. Using this mechanism, we have “written” graphene patterns on nanometer length scales. For patterned regions that are roughly 20 nm or greater, the inherent electronic properties of graphene are completely recovered. Below 20 nm we observe dramatic variations in the electronic properties of the graphene as a function of pattern size. This reversible and local mechanism for modifying the electronic properties of graphene has far-reaching implications for nanoscale circuitry fabricated from this revolutionary material.
Graphene is a nearly ideal two-dimensional conductor that is comprised of a single sheet of hexagonally packed carbon atoms.1 Since the first electrical measurements made on graphene,2-6 researchers have been trying to exploit the unique properties of this material for a variety of applications that span numerous scientific and engineering disciplines.7 In order fully realize the potential of graphene for novel electronic applications, the ability to control the electronic properties of this material on a nanometer length scale is a key challenge. To date this has been accomplished through electrostatic gating and chemical doping and/or modification.8-12 Recently researchers have found that the interaction of hydrogen with graphene can dramatically affect its electronic properties,13,14 rendering this nearly ideal twodimensional conductor insulating. Atomic hydrogen chemically binds to graphene in a pair configuration and when both sides are saturated the sp2 bonds of graphene are converted to sp3 bonded “graphane,” which is highly insulating.13 Here we report the utilization of ultrahigh vacuum (UHV) scanning tunneling microscopy (STM) to not only characterize but to pattern hydrogen saturated graphene. We have found that saturating just a single side of graphene is enough to dramatically alter its electronic properties. Through electron-stimulated desorption (ESD) of hydrogen from the surface, we are able to selectively pattern regions of graphene * Corresponding author’s current address: Argonne National Laboratory, Center for Nanoscale Materials, 9700 South Cass Ave., Bldg. 440, Argonne, IL 60439; e-mail,
[email protected]. † CNISM-Dipartimento di Fisica, Politecnico di Milano. ‡ Center for Nanoscale Materials, Argonne National Laboratory. 10.1021/nl902605t CCC: $40.75 Published on Web 11/02/2009
2009 American Chemical Society
in a controlled manner with nanometer scale resolution. With scanning tunneling spectroscopy (STS), we are able to extract electronic information and identify key signatures of epitaxial graphene within the patterned regions and verify that after patterning the unique structural and electronic properties of graphene can be fully recovered. We have also found that the electronic properties of the patterned graphene are dependent upon the size of the pattern. Experiments were performed on graphene thermally grown on the Si-terminated face of commercially purchased 4H: SiC(0001) following a series of in situ high-temperature flashes at 1250 °C. Room temperature UHV STM measurements were utilized for atomic-scale characterization of the surface, where Figure 1a is a topographic image of a cleanly prepared graphene surface. A (6 × 6) superstructure which is indicative of the underlying SiC surface reconstruction is clearly observed.15 The STM image of Figure 1B shows the disordered morphology of a hydrogen-saturated surface following passivation. The passivation process is performed by keeping the sample at room temperature, while an atomic-hydrogen source that utilizes an e-beam heated W capillary is positioned with line-of-sight to the sample.16 A high-flux source of atomic hydrogen is critical, as a standard W cracking filament does not generate the flux of atomic hydrogen needed for saturating the graphene. We found that Pt-Ir STM probes have a high tendency to strip off the adsorbed hydrogen under normal imaging conditions, displaying an almost catalytic behavior. Therefore, W probes were utilized and capable of imaging under normal conditions
in the inset of Figure 1C. The origin of this shift is attributed to graphene’s interaction with surface states of the underlying SiC surface reconstruction. Interestingly, the regions with a single layer of graphene (not shown here) do not exhibit the same spectra as the bilayer regions. In fact, the layer of graphene closest to the SiC surface is commonly referred to as a “buffer layer”.17 It is still unclear as to what extent the electronic properties of this layer behave like graphene or how they may be altered by its interaction with the SiC surface states. Structurally we observe no difference and the hydrogen adsorption shows no dependence upon graphene thickness. A characteristic dI/dU measurement taken over the hydrogen-saturated surface is plotted in Figure 1D. The starting surface is a mixture of both single and bilayer graphene, and following hydrogen saturation the spectra are identical on either region. These spectra are featureless at negative sample bias and do not show a sharp increase in the LDOS followed by a dip at the Dirac point, as observed over regions of clean bilayer graphene.
Figure 1. (A) This STM image (100 nm × 100 nm, sample bias 0.3 V, 0.1 nA) shows a cleanly prepared graphene surface that has been epitaxially grown on 4H:SiC(0001). (B) The surface is fully saturated with hydrogen at room temperature, as illustrated in this STM image (100 nm × 100 nm). (C) STS is utilized to measure this characteristic dI/dU curve over cleanly prepared bilayer graphene. The Dirac point of the bilayer graphene is shifted from the Fermi level to a value of -260 meV. This shift is inherent to epitaxial graphene on SiC and is illustrated by the inset, where graphene’s E(kx, ky) dispersion relation is plotted with an added plane representing the shifted Fermi level. (D) Similar dI/dU measurements were made on the hydrogen-saturated surface and do not show the same characteristic features indicating that the graphene’s electronic structure has been modified. (E) This image shows a patterned box of graphene. When the bias is increased to +4.5 V, the hydrogen easily desorbs from the surface in a controllable way leaving behind pristine graphene, as illustrated in the atomic resolution image of the inset.
(sample bias, +2 V; tunneling current, 50 pA) without altering the surface. With STS we are able to probe the electronic information of the surface that is proportional to the local density of states (LDOS). A dI/dU measurement utilizing a lock-in amplifier over bilayer graphene (Figure 1c) shows a distinct dip in the spectra at approximately -260 meV. This value is almost identical to the shift of the Dirac point relative to the Fermi level observed by angle-resolved photoemission spectroscopy taken on bilayer epitaxial graphene.8 Graphene epitaxially grown on SiC is known to have n-type behavior, with the Fermi level shifted into the conduction band relative to the Dirac point, as illustrated by the E(kx, ky) dispersion relation 4344
Keeping in mind that the STM is probing the LDOS of both the graphene and the underlying substrate, it is important to note that there are surface states in the SiC band gap due to its surface reconstruction. The STM image of Figure 1B shows the disordered morphology of a hydrogen-saturated surface following passivation. It has been discussed that the most stable configuration for “graphane” involves hydrogen adsorption on both sides of the graphene. It has also been shown that graphene is very impermeable to gas species,14,18 including hydrogen. In order to realistically produce an experimentally stable form of “graphane”, one would have to passivate both sides of a suspended sheet. In this report we have found that saturating just a single side of graphene with hydrogen is stable at room temperature and completely alters the electronic properties of the graphene. We will not label it “graphane” as it is not the ideal form. In our case, the hydrogen adsorption presumably opens a gap in the graphene leaving only substrate states for tunneling, which gives rise to the observed LDOS in measurements over the hydrogen-saturated graphene. Because hydrogen is reacting to the graphene, the disappearance of the peak and Dirac point in the spectrum clearly distinguish these features as related to the electronic structure of the graphene. In addition to characterizing the topography and probing the electronic structure, the STM can be utilized as a patterning tool. When the sample bias is increased to +4.5 V, the hydrogen is easily desorbed from the surface in a controlled manner, as illustrated by the patterned square in Figure 1E. The inset is a zoomed in region of this area showing an atomic-resolution image of the patterned graphene. All of our patterning has resulted in pristine regions of graphene with no additional defects, such as lattice vacancies or point defects which are easily observed with STM.19 This illustrates that the hydrogen adsorption is completely reversible and can be controlled through electron stimulated desorption (ESD), providing the ability to pattern graphene on demand. Nano Lett., Vol. 9, No. 12, 2009
Figure 2. (A) This STM image shows an increasing graphene pattern width as a function of increasing positive sample bias. The tip velocity and set point current were held constant. (B) To illustrate the level of control over the graphene writing, we patterned our institutional logo and initials with a line width of 5 nm.
In order to fine tune the patterning procedure, a series of graphene lines were written as a function of sample bias (Figure 2a), while the set point current (50 pA) and tip velocity (75 nm/s) were held constant. At a patterning bias of +5.0 V the line width is on average between 15 and 20 nm, while at +4.0 V the width ranges between 5 and 10 nm. It was found that the threshold for observable graphene lines is around +3.75 to +4.0 V. Only a few desorption events have occurred at +3.5 V with no desorption observed at +3.0 V. The utilization of the STM as a graphene patterning tool is analogous to and inspired by previous work involving ESD of hydrogen from silicon.20 From this work, techniques such as feedback-controlled lithography have been developed and allow for controlled desorption of single hydrogen atoms enabling true atomic-scale patterning;21 this capability should also extend to hydrogenated graphene. To demonstrate the potential for highly controlled graphene patterns, we have written our institutional logo and initials, as illustrated in Figure 2B. The width of the lettering is 5 nm. Following the ESD of hydrogen, STS was utilized to extract electronic information of the patterned graphene. The STM image of Figure 3A shows a fairly wide stripe (∼40 nm) that was intentionally patterned over the transition from bilayer to single layer graphene as indicated by a yellow arrow (the region to the bottom right is the single layer graphene). A numbered line has been overlaid on the bilayer region to indicate the spatial position where a row of dI/dU spectra were measured (Figure 3B). The measured dI/dU values were taken on the region of patterned bilayer graphene and show a characteristic peak at -110 meV followed by a dip at the Dirac point, as was observed on the cleanly prepared sample. As the position of the point spectra moves into the hydrogen-saturated region, the peak disappears and the characteristic spectra for hydrogen passivated graphene is recovered (spectrum 9 in Figure 3B). Although not shown, Nano Lett., Vol. 9, No. 12, 2009
Figure 3. (A) STM image showing a graphene pattern that transitions from bilayer to single layer, as identified by a yellow arrow (the lower right-hand portion is the single layer region). A numbered line has been overlaid to show the spatial position for STS measurements on the bilayer graphene. (B) Individual dI/dV spectra were measured over the patterned bilayer graphene and extend into the hydrogen-saturated region. The spectra show the characteristic features of a peak followed by a dip at the Dirac point on the bilayer region. This peak disappears, and the featureless wide-gap spectrum of the hydrogen-saturated region is observed for spectrum number 9. (C) A spatial conductance map was concurrently measured over the patterned graphene at a sample energy of -110 meV. This value corresponds to the peak position observed in the spectra. Notice the effect of isolated hydrogen adsorption sites (white arrows) on the bilayer region and the dramatic difference between single and bilayer (yellow arrow is transition).
a similar row of spectra was taken solely on the patterned graphene and bridged the transition from bilayer to single layer graphene. These spectra showed a similar trend where the peak observed on bilayer region gradually diminished and ultimately disappeared as the position of the spectra went deeper into the single layer region. We spatially mapped the LDOS at the bilayer peak position of -110 meV. The bilayer region of the patterned graphene shows uniformly higher conductance, in stark 4345
contrast to the hydrogen-saturated region and the lower conductance single layer graphene. It is interesting to note that there is no abrupt electronic change at the boundary between bilayer and single layer graphene (yellow arrow in Figure 3C) but rather a gradual transition. The growth mode of graphene on SiC involves subsequent layers growing below the previous layers. So the top graphene sheet of the bilayer region makes a continuous transition and also comprises the single layer graphene. It is possible that this layer is slightly decoupled from the interface near the transition, which may result in the gradual change of the electronic structure. It is also observed that individual hydrogen adsorption sites (white arrows in Figure 3C) can have a dramatic effect on the electronic properties of the graphene. The conductance of the patterned bilayer graphene is clearly reduced over a radius that is delocalized and extends up to 5 nm away from the adsorption site. Not all of the adsorbed hydrogen exhibits this behavior at this particular energy. In fact, the LDOS over the hydrogen-saturated region is somewhat inhomogeneous at -110 meV. This inhomogeneity persists across the surface regardless of the underlying graphene thickness and may reflect a variation in the charge distribution of the chemically adsorbed hydrogen itself. The hydrogen-saturated regions do become clearly homogeneous at higher energy. In addition to the strong contrast in the LDOS between patterned and unpatterned regions of bilayer graphene, the width of the pattern can also affect the electronic properties. Three different sized patterns of bilayer graphene are illustrated in Figure 4A, where pattern iii was created by desorbing the hydrogen between patterns i and ii. Spectroscopic dI/dU measurements were taken in each region of patterned graphene, as plotted in Figure 4B. For line widths of roughly 10 nm or less (pattern i) the exposed graphene has the same electronic structure as the hydrogen-saturated region. Essentially there is no electronic change, which may be a result of the long-range delocalized effect that the adsorbed hydrogen has on the LDOS, as previously discussed. When the line width is expanded to roughly 15 nm (pattern ii) there is a clear change in the LDOS of the graphene. This graphene is in an intermediate state showing an increase in the dI/dU at negative bias indicating that its conduction band states have started to fill with electrons, but we observe a shoulder rather than a sharp dip at the Dirac point. Bridging these two patterns results in a graphene region that is roughly 20 nm high by 40 nm long (pattern iii). At this point, the graphene exhibits the characteristic bilayer spectra, which is now also observed in the regions that use to make up patterns i and ii. A lot of attention is currently focused on graphene nanoribbons,22 where the width of the ribbon alters the electronic properties, specifically the size of the band gap that forms. At the moment it is unclear whether or not similar affects can be observed by patterning graphene through ESD of hydrogen. The boundary conditions between a graphene edge and adsorbed hydrogen are completely different, and in this case the substrate strongly influences the electronic properties of the graphene. However, we cannot rule out that 4346
Figure 4. (A) These gradient-enhanced STM images show three different sized graphene patterns on bilayer graphene, where pattern iii was created by desorbing the hydrogen between patterns i and ii. (b) STS measurements were taken on each graphene pattern and the dI/dU are plotted (note: these curves are offset for clarity). For graphene patterns that are less than 10 nm, pattern i, we observe almost no change from the background hydrogen-saturated regions. At rougly 15 nm, pattern ii, we start to see a change in the electronic properties. Patterns greater than 20 nm, pattern iii, fully recover the characteristic bilayer graphene spectra even in the regions that had been patterns i and ii.
confinement effects are at play in the spectra observed for intermediate pattern widths. In summary, we have utilized the UHV STM to pattern regions of graphene through the ESD of hydrogen from a passivated surface of epitaxial graphene grown on 4H: SiC(0001). The hydrogen adsorption occurs on one side of the graphene, is stable at room temperature, and dramatically alters the electronic properties of the graphene. STM based ESD of hydrogen from the surface is easily achieved at sample biases greater than +3.75 V. The graphene patterns are physically pristine. For large regions of patterned graphene we are able to determine, through spectroscopic measurements, that it recovers its inherent electronic properties. The electronic properties are found to vary dramatically as a function of pattern size below 20 nm. Ultimately, we are altering the electronic properties of the graphene substrate through hydrogenation on a global scale, while controlling this effect through ESD at the nanometer scale. With this capability, we have demonstrated a means for generating arbitrary graphene patterns on demand. Acknowledgment. The use of the Center for Nanoscale Materials at Argonne National Laboratory was supported by Nano Lett., Vol. 9, No. 12, 2009
the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC0206CH11357. The authors would like to thank H. Zheng and J. F. Mitchell for assistance in sample preparation and B. L. Fisher for his technical assistance. References (1) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183. (2) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306, 666. (3) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 10451. (4) Berger, C.; Song, Z. M.; Li, T. B.; Li, X. B.; Ogbazghi, A. Y.; Feng, R.; Dai, Z. T.; Marchenkov, A. N.; Conrad, E. H.; First, P. N.; de Heer, W. A. J. Phys. Chem. B 2004, 108, 19912. (5) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Katsnelson, M. I.; Grigorieva, I. V.; Dubonos, S. V.; Firsov, A. A. Nature 2005, 438, 197. (6) Zhang, Y. B.; Tan, Y. W.; Stormer, H. L.; Kim, P. Nature 2005, 438, 201. (7) Geim, A. K. Science 2009, 324, 1530.
Nano Lett., Vol. 9, No. 12, 2009
(8) Ohta, T.; Bostwick, A.; Seyller, T.; Horn, K.; Rotenberg, E. Science 2006, 313, 951. (9) Zhou, S. Y.; Gweon, G. H.; Fedorov, A. V.; First, P. N.; De Heer, W. A.; Lee, D. H.; Guinea, F.; Neto, A. H. C.; Lanzara, A. Nat. Mater. 2007, 6, 770. (10) Chen, J. H.; Jang, C.; Xiao, S. D.; Ishigami, M.; Fuhrer, M. S. Nat. Nanotechnol. 2008, 3, 206. (11) Oostinga, J. B.; Heersche, H. B.; Liu, X. L.; Morpurgo, A. F.; Vandersypen, L. M. K. Nat. Mater. 2008, 7, 151. (12) Chen, J. H.; Jang, C.; Adam, S.; Fuhrer, M. S.; Williams, E. D.; Ishigami, M. Nat. Phys. 2008, 4, 377. (13) Elias, D. C. Science 2009, 323, 610. (14) Guisinger, N. P.; Rutter, G. M.; Crain, J. N.; First, P. N.; Stroscio, J. A. Nano Lett. 2009, 9, 1462. (15) Rutter, G. M. Phys. ReV. B 2007, 76, 235416. (16) Balog, R. J. Am. Chem. Soc. 2009, 131, 8744. (17) Varchon, F. Phys. ReV. Lett. 2007, 99, 126805. (18) Bunch, J. S. Nano Lett. 2008, 8, 2548. (19) Rutter, G. M.; Crain, J. N.; Guisinger, N. P.; Li, T.; First, P. N.; Stroscio, J. A. Science 2007, 317, 5835. (20) Shen, T.-C. Science 1995, 268, 1590. (21) Hersam, M. C.; Guisinger, N. P.; Lyding, J. W. Nanotechnology 2000, 11, 70. (22) Li, X.; Wang, X.; Zhang, L.; Lee, S.; Dai, H. Science 2008, 319, 1229.
NL902605T
4347
