Perspective pubs.acs.org/cm
Top-Down versus Bottom-Up Fabrication of Graphene-Based Electronics James M. Tour* Department of Chemistry, Department of Computer Science, Department of Mechanical Engineering and Materials Science, and the Smalley Institute for Nanoscale Science and Technology, Rice University, 6100 Main Street, Houston, Texas 77005, United States ABSTRACT: Graphene electronic devices can be made by top-down (TD) or bottom-up (BU) approaches. This Perspective defines and explains those two approaches and discusses the advantages and limitations of each, particularly in the context of graphene fabrication. It is further exemplified using graphene nanoribbons as the prototypical graphene structure that can be prepared using either a TD or BU approach. The TD approach is well-suited for placement of large arrays of devices on a chip using standard patterning tools. However, the TD approach severely compromises the edges of the graphene since present fabrication tools are coarse relative to the ∼0.1 nm definition of a C−C bond. The BU approach can afford exquisite control of the graphene edges; however, placing the structures, en mass, in the locations of interest is often impossible. Also, using the BU approach, it can be very difficult to make device structures long enough for integration with TDderived probe electrodes. Specific examples are shown, along with an outlook for optimization of future graphene devices in order to capitalize upon the advantages of both TD and BU fabrication methodologies. KEYWORDS: graphene-based electronics, top-down, bottom-up, fabrication, graphene
■
■
INTRODUCTION Single-layer graphene is a one-atom-thick two-dimensional material with exceedingly high carrier mobility and high conductivity resulting in it understandably becoming an exciting material for consideration in electronics platforms.1−4 Graphene can be prepared in bulk solution5,6 or as atomically thin films on a surface;7−10 it can be functionalized on the edges or basal planes11−13 and selectively patterned; and small sections can be synthesized using chemical methods.14 Graphene can be used as a transparent conducting electrode.9,15,16 It can be fashioned into ribbons that can have differing properties depending on the widths and edge configurations.17,18 Since 2004 when graphene’s extreme properties were realized,3 a plethora of fabrication, manipulation, and device embodiments have been disclosed, resulting in it becoming a favorite topic of research and targeted commercial innovation.1,14 Researchers are considering two primary methods for the fabrication of graphene-based electronic devices: a top-down (TD) and a bottom-up (BU) approach. First, in this Perspective, TD and BU will be defined using explanatory analogies. Second, the advantages and limitations of each of those device fabrication methods will be delineated. Third, specific graphene-based TD and BU approaches will be shown, particularly in the context of one common structural embodiment: graphene nanoribbons (GNRs). Finally, a future combined TD and BU approach will be suggested, in the context of GNRs, which can exploit the advantages of both approaches. © 2013 American Chemical Society
DEFINITION OF TOP-DOWN (TD) AND BOTTOM-UP (BU) FABRICATION19
TD fabrication is analogous to cutting down a tree and chiseling a statue from the tree trunk (Figure 1a). Intricate detail can be fashioned into the statue, and when complete, it might have little resemblance to the starting tree. But one often starts from something large and proceeds to something smaller through the removal of extraneous material. TD fabrication is the approach that is currently used in, for example, the silicon industry where one starts with ton-weighing, cylinder-shaped single-crystal-silicon ingots that are 300 to 450 mm in diameter.21 The ingots are sliced into millimeter-thin wafers, and then using resists (chemicals), light, ion implantations, and depositions, micro- and nanoscopic features such as wires, diodes and transistors are cut and built into those wafers. The wafers are then diced into chips that are a few centimeters on each edge. The chips are tested and used in a plethora of electronics applications. The silicon industry has had trillions of dollars and millions of person years invested into refining those TD fabrication tools and techniques, thus standing as the main driver of high technology processes for the past 50 years. Special Issue: Celebrating Twenty-Five Years of Chemistry of Materials Received: July 3, 2013 Revised: August 13, 2013 Published: August 23, 2013 163
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
■
Perspective
ADVANTAGES OF TD FABRICATION
The advantages of the TD approach are several-fold. The most significant advantage is the ability to address the placement or location of the desired entity. In the example of GNRs, using a predefined mask for the patterning, one knows precisely where the GNRs are located and where their ends are located relative to the mask’s alignment marks. This is the greatest advantage in a TD approach that will likely make graphene rapidly amenable to use as a complement to silicon because it uses the same methodologies that have been developed over 50 years of silicon fabrication for precise feature placement. In placing an entity at a specific location, there is also a route to having it interfaced with the surrounding network of electronic device structures. Since a silicon under-layer is patterned with similar techniques, the GNR is ready to have its ends treated with metals, for example, to be addressed with electrodes. Moreover, the GNR could have easily been fashioned to the desired length over a specific location on the chip that had been prefabricated with a bottom gate to now render the GNR as the active layer in a field-effect transistor (FET) or chemical sensor. Therefore, the TD approach permits precise placement of the graphene-based structure, precise length of the structure, in the location to be addressed, and it provides immediate integration of the graphene with the surrounding electronic host and substructures.
Figure 1. TD and BU fabrication compared (not to scale). (a) TD fabrication showing a wooden statue of an owl made from a tree. (b) BU fabrication where a tree is derived from an acorn. (c) BU fabrication where a seed might be programmed, via DNA, to directly form a wooden statue.20
Conversely, BU construction is done by starting with smaller entities and building them up to larger functional constructs, such as an acorn to a tree (Figure 1b). BU methodologies are natural in that all systems in nature are constructed BU. Unlike the TD fabrication of a statue from a tree, in nature, molecules with specific features assemble to form higher order structures. Further construction, some of it self-assembly and much of it enzyme-controlled assembly, can result in production of cells, and further into more complex life forms. The processes for this natural assembly, from the synthetic chemical perspective, while fascinatingly complex, are nevertheless ubiquitous on Earth. Self-assembly is a subset of BU construction where thermodynamically favorable processes ensue to form an organized aggregated structure. Enzyme-controlled construction is also a subset of BU assembly, but using nature’s nanomachines for specific nonregular formation of structures. While human beings often manufacture TD, nature almost exclusively assembles its structures BU. Using the tree example above, the tree was assembled BU and not constructed from a yet larger tree. A BU strategy to build a wooden statue might involve modification of DNA so that it codes for the enzymes that directly build the statue and not the tree (Figure 1c). It is important to note that complex function cannot arise from selfassembly alone; agents are needed, and in nature’s case enzymes are needed that work upon specified recognition elements to build nonregularly sequenced structures. Complex function can only arise from nonregular sequences of device patterns.22 A repeated pattern of units or devices, for example, ABAB or ABCABC, cannot show complex function. Complexity comes from nonregular patterns of structures or devices, such as in all biological systems as well as in complex computer chips. And the coding need not be envisioned in DNA only, and the construction need not be imagined with enzymes only. It could conceivably be done with nanomachines that are nonbiological in form, such as nanoconstruction machines bringing carbohydrates into place for coupling, until the statue is built to desired specifications. Indeed, such a scenario is science fiction-sounding, yet truly bioinspired; a natural method of manufacture and a methodology which is certain to influence much in the future, albeit the distant future.23
■
LIMITATIONS OF TD FABRICATION While TD fabrication has powerful advantages, there are several notable limitations of the TD approach to graphene devices. First, there is a frustrating resolution limitation. Single-crystal graphene is an extraordinarily well-defined matrix of hexagonal carbon arrangements that has angstrom-level precision in its atomic orientation.24 Following upon the GNR example above, the edges of the TD-fabricated GNR25−33 will be jagged and only as well-defined as are the mask and/or the cutting tool (ebeam, ion-beam, or even scanning probe tip) used to define the edges. In general, all regioregularity of the edge structure is lost, now comprising a mixture of zigzag, armchair, and other possible configurational structures. The dangling or radicalcontaining edges of the GNR will be capped by residual oxygen in the tool’s chamber or upon exposure to air, resulting in hydroperoxy, hydroxyl, and carbonyl moieties. This edge variability, the inability to attain a desired edge of interest, and the uncontrolled edge functionalities result in the GNR having far less utility than it otherwise could have had. This is especially true when the width of the GNR is sub-10nm.17,18,34−39 It has been shown experimentally and further explained theoretically, that mixtures of edge configurations in narrow GNRs will afford what might initially be thought of as a bandgap. However, it is not a true bandgap but rather a “mobility gap” or “transport gap” that has been induced due to carrier retardation by the irregular edges.36−39 These differences reveal themselves through variable-temperature current− voltage studies. A second and related resolution-based disadvantage of the TD approach can be seen in the width control of the TDfabricated GNRs. The width is again limited by the controlled width limitations in the mask and the resolution of the patterning or cutting tool used. For example, in e-beam lithography the resolution is determined by secondary electrons generated by the substrate and resist and thus results in the effective beam width being 15 to 20 nm. But even if one could 164
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
Figure 2. “Controlled layer-by-layer removal of graphene. (A) Schematic illustration of the method: (1) The bilayer graphene on top of a Si/SiO2 substrate. (2) A patterned layer of zinc metal is sputtered atop the graphene. (3) The zinc is removed by aqueous HCl (0.02 M) in 3−5 min with simultaneous removal of one graphene layer. (4) Patterning of a second zinc stripe. (5) HCl treatment removes the second stripe of zinc plus the underlying carbon layer. (B−D) SEM image of the same bilayer GO flake: (B) original, (C) after the first, and (D) after the second Zn/HCl treatment. (E) SEM image of a monolayer GO flake patterned in the image of an owl. (F and G) SEM images of a continuous GO film patterned with horizontal and vertical stripes in two consecutive Zn/HCl treatments. The lightest squares (an example is marked with “n-2”), where the horizontal and vertical stripes overlap, represent areas exposed to two treatments. Areas exposed to one treatment (examples are marked with “n-1”) are with a shade between the lightest and darkest squares. The darkest squares (examples are marked with “n”) represent the areas with the original untreated GO film.” Figure and caption from ref 40. Reprinted with permission from AAAS.
and has limitations in numbers of entities that can be madeat least when compared to chemical methods.
obtain an intrinsic 3-nm-resolution, that is still coarse relative to the sizes now desired. Recall that GNRs become especially interesting at sub-10-nm in width, and a 3 nm beam is very wide when considering that the C−C bond length in graphene is ∼0.1 nm. Thus, when it was described above that TD fabrication provides a method to control the “precise” length of the GNR, “precise” is a relative term and is only as good as the mask and cutting tool resolution. One could argue that when silicon-based fabrication tools are applied to graphene, at least as it concerns graphene edges, it is analogous to using a chainsaw to perform finishing work on a dining tableall edges will be painfully jagged. Hence, TD fabrication has limitations on resolutions that are set by the coarseness of the tools available for TD patterning. The third deficiency in TD fabrication is the inability to fabricate very large numbers of entities. Using photolithography, billions of devices can be made on a single wafer. That is wonderful-sounding for many applications, but when compared to 1023 prepared in a chemical process, billions is a small number. Therefore, TD fabrication, though tremendously advantageous for placement and integration, lacks the resolution that is often desirable in graphene device building
■
ADVANTAGES OF BU FABRICATION Turning to BU fabrication, there are notable advantages. BU assembly is often chemical-based, so the enormity of Avogadro’s number works to the advantage of the chemist: large numbers of chemical entities are possible when preparations are conducted in a flask. The second and probably the most compelling advantage for BU fabrication is the control of resolution. Since molecules are built from discrete atomic or small molecular entities, the fabrication can be done with atomic-level precision using synthetic chemical techniques that have been developed over the past century. At this level, in GNRs, for example, synthetic routes to zigzag versus armchair configurations would be entirely different and distinguishable. Here, angstrom-level differences can be controlled, albeit they are often slow, laborious, and expensive since organic synthesis is required with separations at intermediate steps. The chemical vapor deposition (CVD) growth of graphene on copper or nickel using methane or solid carbon sources 165
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
Figure 3. “MML fabrication. (A) MML GNR fabrication scheme. Formation of GNRs occurs at the edge of the lithography pattern. Blue arrows point to the meniscus of the adsorbed water which serves as the mask for the ultimate GNRs. (B−I) SEM images of individual GNRs at different magnifications. (B and F) ox-GNRs on Si/SiO2; (C and G) ox-GNRs on BN; (D and H) ar-GNRs on Si/SiO2; (E and I) ar-GNRs on BN. The scale bar for images (F−I) is 100 nm. (J) A Pt wire on Si/SiO2. (K) The letter R patterned with ar-GNRs. Inset: magnified image of a part of the ar-GNR comprising the letter R. The mean width is 9.2 ± 1.1 nm.” Figure and caption reprinted with permission from ref 41. Copyright 2013 American Chemical Society. ar = argon and ox = oxygen.
specific places of interest. Likewise, laminar flow techniques, although interesting in singular laboratory demonstrations, have proven of little use in providing the precision needed for arrangement of a large numbers of nanodevices.19 When one is making a film where the precise arrangement of nanostructures is not required, then even simple spin-coated, dip-coated, or shear-induced ordering techniques can be useful. But if seeking discrete entity placement, then BU methods are of limited use. When considering chemical synthesis as a fabrication method, building structures of precise width is done exceeding well and with angstrom-level control. But making the structures of precise length, in the size regions necessary for attachment to electrodes, is often difficult to attain. Moreover, even if precise lengths are not needed, the structures still have to be large enough to address with electrodes that are likely made TD. Chemical coupling reactions are often insufficient to afford the degrees of polymerization needed to generate the requisite lengths. In GNRs, while precise width and edge configurations are accessible, the desired length of the GNRs would often be unattainable at >100-nm. This is due to the fact that the number of needed coupling reactions is too great, based on the yields of the individual coupling reactions, to afford a nanoconstruct of addressable size. Therefore, placement and length problems often render BU methods insufficient for use in device embodiments where significant numbers of devices,
under a hydrogen atmosphere is fundamentally a BU process.7−10 Here, organic compounds are broken down at high temperatures atop or within metals to afford atomic carbon structures. The carbon atoms are then assembled, either at high temperature as in the case of copper surfaces or upon cooling to afford surface precipitation as in the case of nickel substrates. This represents a self-assembly of carbon atoms into the desired graphene lattice structure which is the thermodynamically preferred structure and which is energetically accessible at the higher temperatures. Here again, there is angstrom-level-controlled ordering in this BU assembly.
■
LIMITATIONS OF BU FABRICATION The most significant deficiency in BU fabrication is the limitation in placement of the device structure. Continuing with GNRs as an example, how could one go from a flask full of 1016 nanosized GNRs to getting 1000 of them in the precise locations of interest upon a chip? This problem has plagued carbon nanotube device development, and it is now being seen as well in BU-fabricated graphene devices. Researchers have tried self-assembly techniques, but often the substrate itself has defects or unhelpful surface patterns that render a poorly assembled structure or one that only has small domains of order.19 Langmuir−Blodgett films have been tried, but being nonthermodynamic, this technique has proven to be of little practical utility in the precise ordering of nanostructures at 166
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
Figure 4. Synthesis and characterization of BU-synthesized GNRs. (a) Synthesis scheme from small molecules, to surface polymerization, and cyclodehydrogenation. (b) STM image before cyclodehydrogenation. (c) STM image after cyclodehydrogenation. (d) Raman spectrum of the GNRs. (e) High resolution STM of the final GNRs. Figure reprinted by permission from Macmillan Publishers Ltd., Nature,42 copyright 2010.
these GNRs are extremely jagged with hydroxyl and carbonyl termini due to O2 quenching of the radicals after argon- or oxygen-ion etching. Nonetheless, for making GNRs that are as long as desired and sub-10-nm in width, this is likely the most satisfactory method presently available.
such as 100 or more, are needed to afford useful integrated architectures.
■
EXAMPLES OF TD FABRICATION “The patterning of graphene is useful in fabricating electronic devices, but until recently methods did not allow control of the number of layers of graphene that are removed. We showed that sputter-coating graphene and graphene-like materials with zinc and dissolving the latter with dilute acid removes one graphene layer and leaves the lower layers intact (Figure 2). The method works with the four different types of graphene and graphene-like materials: graphene oxide (GO), chemically converted graphene (CCG), CVD graphene, and micromechanically cleaved (“adhesive-tape”) graphene. The top graphene layer is damaged by the sputtering process and the acid treatment removes the damaged layer of carbon.”40 As with all TD methods, the lateral resolution of the technique is limited to the definition of the mask and then the metalmigration near the edge of the formation. However, in the vertical dimension, “this method can be viewed as lithography that etches the sample with single-atomic-layer resolution.”40 We recently developed a technique that is “a new planar TD method for the fabrication of precisely positioned very narrow (sub-10 nm), high aspect ratio (>2000) GNRs from graphene sheets which we call meniscus-mask lithography (MML). The method does not require demanding high resolution lithography tools. The mechanism involves masking by atmospheric water adsorbed at the edge of the lithography pattern written on top of the target material (Figure 3). The GNR electronic properties depend on the graphene etching method with argon reactive ion etching yielding remarkably consistent results.”41 But still, at the atomic level, the edges of
■
EXAMPLES OF BU FABRICATION Turning to examples of BU construction in graphene-based systems, the most exciting demonstration has come from the team of Mullen and Fasel et al.42 Small dibromo molecular systems were synthesized, Ullman-like polymerized atop a metallic substrate, and then oxidatively cyclized affording ultranarrow arm-chair-edged GNRs (Figure 4). This is BU from the standpoint of the chemical synthesis to make the dibromo-monomers, and then BU again to have the assembly into the polymers and oxidatively cyclized. This is a magnificent display of the power of BU construction to control the edge states of GNRs. But as all BU-derived structures, the GNRs are beset with only crude length control, and the structures were too short to be conveniently addressed with probes. Furthermore, there is no suitable method to precisely set many of these GNRs into specific locations for addressing. However, even with these deficiencies, which are slowing the testing of these GNRs in individual devices, it is a remarkable demonstration of the power of BU chemical synthesis to control the angstrom-level structure of the edges. One can also consider BU fabrication starting from existing nanosized entities that were themselves grown BU. Multiwalled carbon nanotubes (MWCNTs) can be longitudinally split to form GNRs using both oxidative and reductive conditions (Figure 5).43−47 The variety of such methods was highlighted in a recent review,1 but here we underscore the KMnO4 route to 167
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
and then patterning metallic leads across their two ends. Therefore, though these flasks of GNRs are fine for making conductive displays by spray-coating,48 for example, the BU method does not permit the precise placement in large arrays as desired. Furthermore, the GNRs in this case have jagged edges resulting from the longitudinal unzipping process.
■
UNION OF TD AND BU FABRICATION METHODS
“As low dimensional materials, graphene and single-walled carbon nanotubes (SWCNTs) exhibit exceptional properties such as high carrier mobility, high electrical and thermal conductivity and large specific surface area. The combination of nanotube carpets and graphene is an approach to extend those properties into three dimensions. However, high quality, singleand few-walled carbon nanotube carpets with large specific surface area are usually grown on alumina, an insulator that immobilizes the catalyst particles and supports efficient growth. In contrast, high quality graphene is usually produced on metal substrates such as copper or nickel. These two different growth conditions hamper efforts to covalently combine the two materials during growth. As a result, attempts to grow graphene and carbon nanotube (CNT) carpet hybrid materials have failed to match the predicted values on specific surface area (SSA), electrical connection or both.” Most importantly, it was not until recently that the “actual connection between the CNTs and graphene was” imaged, and this constitutes “a method to prepare high quality graphene with few-walled (1− 3-walled) CNT carpets seamlessly connected via covalent bonds. Without sacrificing their stand-alone properties, the ohmic interconnected graphene and CNT carpet hybrid can be produced in a high surface area material” (Figure 8).49 “The atomic scale aberration corrected STEM analysis gives the detailed information of the conjugated covalent bonds in the junction region, which is in agreement with previous simulation studies.”49 The seamless conjoining of the CNTs with graphene
Figure 5. Schematic view of the longitudinal opening of MWCNTs to form GNRs. Figure reprinted by permission from Macmillan Publishers Ltd., Nature,43 copyright 2009.43−47
generate graphene oxide nanoribbons (GONRs), which could subsequently be reduced to GNRs (Figure 6).43,44 Also shown are the direct reductive splitting of MWCNTs using K vapor45 or Na−K alloy in solution (Figure 7).46,47 These methods conveniently render a flask full of GNRs, and the processes are being industrially scaled. Though the nanotube unzipping method is superb for rendering large quantities of GNRs, the few GNRs that were individually tested were aligned and ordered, painstakingly, by dispersing a solution of the GNRs on a surface, searching by SEM for flat structures of appropriate size that were not folded,
Figure 6. MWCNTs to GONRs. (a) TEM images of the MWCNTs (left) before the reaction and the GNRs after the reaction (right). (b) Sonication-cut GONRs imaged by AFM. (c) SEM image of a GONR. Figure reprinted with permission from Nature.43 Copyright 2009 Macmillan Publishers Ltd. 168
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
Figure 7. “Scheme for the in-situ intercalation replacement and selective functionalization of GNRs. (a) Intercalation of potassium between the walls of MWCNTs. (b) Splitting process of MWCNTs and formation of active carboanionic edges is shown (M = K+ or Na+). (c) In situ functionalization and intercalation of GNRs with alkyl groups. (d) Deintercalation of functionalized GNRs.” Figure and caption reprinted with permission.46 Copyright 2012 American Chemical Society.
(Figure 9).51 Therefore, the union of a BU synthesis of the graphene, TD patterning of the graphene into defined configurations,51 and then the BU vertical growth of CNTs on the patterned graphene demonstrate a marriage of the TD and BU methodologies.
■
OUTLOOK: TD JOINED WITH BU Recall that a TD procedure permits structures that are defined in precise locations with resolutions commensurate with the masking and cutting technique. But only the BU method will permit well-defined edge structures. An attractive option for GNRs would be to build a GNR TD in the place of interest, for example, as described above using MML technique, but then use BU chemical techniques to remove the oxygen functionality from the edges. While mixed zigzag and armchair edges would be retained, there would be little oxygen termini affecting the transport properties. This might be done by a sequence of reduction of the carbonyls (ketones and carboxylic acids), followed by treatment under phenol reduction conditions to afford the hydrogen-atom terminated aryl groups. Furthermore, since the edges are generally more reactive than the basal planes,12 it might be possible to effect Diels−Alder edge caps that would homogenize the edge configurations. It is conceded that there are few details disclosed here for this sequence, hence the outlook to the future. In order to maintain the TD-placed graphene structure, chemistry would have to be executed on the surface. This would likely retard SN2 reactions where backside displacement reactions are required, so an entire arm of synthetic chemistry might be inaccessible. But this does not inhibit single-electron reduction reactions or higher temperature homolytic cleavage chemistries, both of which would have to be done with precise control of structure. This will provide
Figure 8. “Scheme for the synthesis of CNT carpets directly from graphene. (a) Copper foil substrate. (b) Graphene is formed on the copper foil by CVD or solid carbon source growth. (c) Iron and alumina are deposited on the graphene covered copper foil by using ebeam evaporation. (d) A CNT carpet is directly grown from graphene surface. The iron catalyst and alumina protective layer are lifted up by the CNT carpet as it grows.” Figure and caption reprinted with permission from Macmillan Publishers, Ltd., Nature Communications,49 copyright 2012.
has permitted the formation of very higher performance supercapacitors,49,50 field emitters,50 and microsupercapacitors 169
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
Perspective
Figure 9. Design of microsupercapacitors and material characterizations. (a) Schematic of the structure of microsupercapacitor. Inset: enlarged scheme of pillar structure; (b) SEM image of a fabricated microsupercapacitor; (c−e) TEM images of individual single-, double-, and few-wall CNTs; (f−h) cross-sectioned SEM images of CNT carpet grown for 1, 2.5, and 5 min. Image reprinted with permission.51 Copyright 2013 American Chemical Society.
fertile research fields for organic chemists as they consider surface chemistry restrictions but press ahead with a battery of new methodology developments for surface-confined carbon structures. So the future is richest for the synthetic chemist as it applies to BU modification of TD-derived structures.52 While the specifics need to be more precisely developed, outlined here is the conceptual approach to capitalize upon the advantages of both the TD and BU routes for further optimization of GNRs or graphene devices in general.
progress in graphene, then it will not be long before exquisite control in length, width, and design of graphene devices are realized, probably via a union of TD and BU approaches.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
Biography
CONCLUSION AND THE FUTURE The main methods for graphene device construction involve the TD and BU approaches. The TD approach has the advantages of facile device placement and device addressing while suffering from a resolution limitation that is governed by the masking and cutting-tool (such as ion- or e-beam) techniques. The result can be jagged-edged structures that are not suitably controlled for the desired use. BU methods provide routes to large numbers of devices that can have angstromlevel-defined edges. However, methods of the placement of BUmade devices, and their addressing, are severely limited. Hybrid approaches based on TD with BU methods are known, and they deserve further exploitation to render precisely placed angstrom-level-defined graphene devices of interest. This Perspective and its final proposal underscore the need for continued development in graphene materials chemistry. If these developments move as quickly as the first decade of
James M. Tour, a synthetic organic chemist, received his Bachelor of Science degree from Syracuse University, his Ph.D. from Purdue University with Ei-ichi Negishi, and postdoctoral training at the University of Wisconsin and Stanford University with B. M. Trost. He is presently the T. T. and W. F. Chao Professor of Chemistry, Professor of Computer Science, and Professor of Mechanical Engineering and Materials Science at Rice University. He has over 500 research publications and over 60 patents.
■
ACKNOWLEDGMENTS Our program in graphene-based electronic devices is funded by the AFOSR (FA9550-09-1-0581), ONR-MURI Graphene Program (00006766, N00014-09-1-1066), AFOSR-MURI Program (FA9550-12-1-0035), Sandia National Laboratory, AZ Electronic Materials and Lockheed Martin through the LANCER Program. 170
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
Chemistry of Materials
■
Perspective
(37) Oostinga, J. B.; Sacépé, B.; Craciun, M. F.; Morpurgo, A. F. Phys. Rev. B 2010, 81, 193408. (38) Han, M. Y.; Brant, J. C.; Kim, P. Phys. Rev. Lett. 2010, 104, 056801. (39) Sols, F.; Guinea, F.; Neto, A. H. C. Phys. Rev. Lett. 2007, 99, 166803. (40) Dimiev, A.; Kosynkin, D. V.; Sinitskii, A.; Slesarev, A.; Sun, Z.; Tour, J. M. Science 2011, 331, 1168−1172. (41) Abramova, V.; Slesarev, A.; Tour, J. M. ACS Nano 2013, 7, 6894−6898. (42) Cai, J.; Ruffieux, P.; Jaafar, R.; Bieri, M.; Braun, T.; Blankenburg, S.; Muoth, M.; Seitsonen, A. P.; Saleh, M.; Feng, X.; Mullen, K.; Fasel, R. Nature 2010, 466, 470−473. (43) Kosynkin, D. V.; Higginbotham, A. L.; Sinitskii, A.; Lomeda, J. R.; Dimiev, A.; Price, B. K.; Tour, J. M. Nature 2009, 458, 872−826. (44) Higginbotham, A. L.; Kosynkin, D. V.; Sinitskii, A.; Sun, Z.; Tour, J. M. ACS Nano 2010, 4, 2059−2069. (45) Kosynkin, D. V.; Lu, W.; Sinitskii, A.; Pera, G.; Sun, Z.; Tour, J. M. ACS Nano 2011, 5, 968−974. (46) Genorio, B.; Peng, Z.; Lu, W.; Hoelscher, B. K. P.; Novosel, B.; Tour, J. M. ACS Nano 2012, 6, 10396−10404. (47) Lu, W.; Ruan, G.; Genorio, B.; Zhu, Y.; Novosel, B.; Tour, J. M. ACS Nano 2013, 7, 2669−2675. (48) Zhu, Y.; Lu, W.; Sun, Z.; Kosynkin, D. V.; Yao, J.; Tour, J. M. Chem. Mater. 2011, 23, 935−939. (49) Zhu, Y.; Li, L.; Zhang, C.; Casillas, G.; Sun, Z.; Yan, Z.; Ruan, G.; Peng, Z.; Raji, A.-R. O.; Kittrell, C.; Hauge, R. H.; Tour, J. M. Nat. Commun. 2012, 3, 1225. (50) Yan, Y.; Ma, L.; Zhu, Y.; Lahiri, I.; Hahm, M. G.; Liu, L.; Yang, S.; Xiang, C.; Lu, W.; Peng, Z.; Sun, Z.; Kittrell, C.; Lou, J.; Choi, W.; Ajayan, P. M.; Tour, J. M. ACS Nano 2013, 7, 58−64. (51) Lin, J.; Zhang, C.; Yan, Z.; Zhu, Y.; Peng, Z.; Hauge, R. H.; Natelson, D.; Tour, J. M. Nano Lett. 2013, 13, 72−78. (52) Tour, J. M. Mater. Today 2011, 14, 454.
REFERENCES
(1) Zhu, Y.; James, D. K.; Tour, J. M. Adv. Mater. 2012, 24, 4924− 4955. (2) Geim, A. K; Novoselov, K. S. Nat. Mater. 2007, 6, 183−191. (3) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsob, A. A. Science 2004, 306, 666−669. (4) Ruoff, R. S. Nat. Nanotechnol. 2008, 3, 10−11. (5) Hamilton, C. E.; Lomeda, J. R.; Sun, Z.; Tour, J. M.; Barron, A. R. Nano Lett. 2009, 9, 3460−3462. (6) Hummers, W. S.; Offeman, R. E. J. Am. Chem. Soc. 1958, 80, 1339−1339. (7) Reina, A.; Jia, X.; Ho, J.; Nezich, D.; Son, H.; Bulovic, V.; Dresselhaus, M. S.; Kong, J. Nano Lett. 2009, 9, 30−35. (8) Li, X.; Cai, W.; An, J.; Kim, S.; Nah, J.; Yang, D.; Piner, R.; Velamakanni, A.; Jung, I.; Tutuc, E.; Banerjee, S. K.; Colombo, L.; Ruoff, R. S. Science 2009, 324, 1312−1314. (9) Kim, K. S.; Zhao, Y.; Jang, H.; Lee, S. Y.; Kim, J. M.; Kim, K. S.; Ahn, J.-H.; Kim, P.; Choi, J.-Y.; Hong, B. H. Nature 2009, 457, 706. (10) Sun, Z.; Yan, Z.; Yao, J.; Beitler, E.; Zhu, Y.; Tour, J. M. Nature 2010, 468, 549−552. (11) Savchenko, A. Science 2009, 323, 589−590. (12) Sun, Z.; Kohama, S.; Zhang, Z.; Lomeda, J. R.; Tour, J. M. Nano Res. 2010, 3, 117−125. (13) Liu, L. H.; Yan, M. J. Mater. Chem. 2011, 21, 3273−3276. (14) Sun, Z.; James, D. K.; Tour, J. M. J. Phys. Chem. Lett. 2011, 2, 2425−2432. (15) Eda, G.; Lin, Y.-Y.; Miller, S.; Chen, C.-W.; Su, W.-F.; Chhowalla, M. Appl. Phys. Lett. 2008, 92, 233305-1. (16) Bae, S.; Kim, H.; Lee, Y.; Xu, X.; Park, J.-S.; Zheng, Y.; Balakrishnan, J.; Lei, T.; Kim, H. R.; Song, Y. I.; Kim, Y.-J.; Kim, K. S.; Ö zyilmaz, B.; Ahn, J.-H.; Hong, B. H.; Iijima, S. Nat. Nanotechnol. 2010, 5, 574−578. (17) Son, Y.-W.; Cohen, M. L.; Louie, S. G. Phys. Rev. Lett. 2006, 97, 216803. (18) Yang, L.; Park, C.-H.; Son, Y.-W.; Cohen, M. L.; Louie, S. G. Phys. Rev. Lett. 2007, 99, 186801. (19) Tour, J. M. “Molecular Electronics: Commercial Insights, Chemistry, Devices, Architecture and Programming; World Scientific: Teaneck, NJ, 2003 (ISBN 981-238-296-0 and 981-238-341-7 (pbk)). (20) Owl statue courtesy Woodson Research Center, Rice University. (21) http://en.wikipedia.org/wiki/Wafer_(electronics) (accessed July 1, 2013). (22) https://en.wikipedia.org/wiki/Pattern (accessed July 1, 2013). (23) Vives, G.; Tour, J. M. Acc. Chem. Res. 2009, 42, 473−487. (24) Yan, Z.; Lin, J.; Peng, Z.; Sun, Z.; Zhu, Y.; Li, L.; Xiang, C.; Samuel, E. L.; Kittrell, C.; Tour, J. M. ACS Nano 2012, 6, 9110−9117. (25) Fasoli, A.; Colli, A.; Lombardo, A.; Ferrari, A. C. Phys. Status Solidi B 2009, 246, 2514−2517. (26) Sinitskii, A.; Tour, J. M. Appl. Phys. Lett. 2012, 100, 103106. (27) Han, M. Y.; Ö zyilmaz, B.; Zhang, Y.; Kim, P. Phys. Rev. Lett. 2007, 98, 206805. (28) Masubuchi, S.; Ono, M.; Yoshida, K.; Hirakawa, K.; Machida, T. Appl. Phys. Lett. 2009, 94, 082107. (29) Tapasztó, L.; Dobrik, G.; Lambin, P.; Biró, L. P. Nat. Nanotechnol. 2008, 3, 397−401. (30) Bai, J.; Duan, X.; Huang, Y. Nano Lett. 2009, 9, 2083−2087. (31) Shi, Z.; Yang, R.; Liu, D.; Yang, W.; Cheng, M.; Wang, D.; Shi, D.; Zhang, G. Nano Lett. 2012, 12, 4642−4646. (32) Sprinkle, M.; Ruan, M.; Hu, Y.; Hankinson, J.; Rubio-Roy, M.; Zhang, B.; Wu, X.; Berger, C.; de Heer, W. A. Nat. Nanotechnol. 2010, 5, 727−731. (33) Kato, T.; Hatakeyama, R. Nat. Nanotechnol. 2012, 7, 651−656. (34) Yazyev, O. V.; Capaz, R. B.; Louie, S. G. Phys. Rev. B 2011, 84, 115406. (35) Li, X. L.; Wang, X. R.; Zhang, L.; Lee, S. W.; Dai, H. J. Science 2008, 319, 1229−1232. (36) Stampfer, C.; Güttinger, J.; Hellmüller, S.; Molitor, F.; Ensslin, K.; Ihn, T. Phys. Rev. Lett. 2009, 102, 056403. 171
dx.doi.org/10.1021/cm402179h | Chem. Mater. 2014, 26, 163−171
