PERSPECTIVE PUBLISHED ONLINE: 23 MAY 2011 | DOI: 10.1038/NCHEM.1043 Blueprinting macromolecular electronics Carlos-Andres Palma† and Paolo Samorì* Recently, by mastering either top-down or bottom-up approaches, tailor-made macromolecular nano-objects with semiconducting properties have been fabricated. These engineered nanostructures for organic electronics are based on conjugated systems predominantly made up of sp2-hybridized carbon, such as graphene nanoribbons. Here, we describe developments in a selection of these nanofabrication techniques, which include graphene carving, stimulus-induced synthesis of conjugated polymers and surface-assisted synthesis. We also assess their potential to reproduce chemically and spatially precise molecular arrangements, that is, molecular blueprints. In a broad context, the engineering of a molecular blueprint represents the fabrication of an integrated all-organic macromolecular electronic circuit. In this Perspective, we suggest chemical routes, as well as convergent on-surface synthesis and microfabrication approaches, for the ultimate goal of bringing the field closer to technology. F rom a windmill to a circuit board, behind the crafting of every tool there is a strict layout of its constituent elements; its structural blueprint determines the very position of each component. The principle of creating a (macro- or supra-) molecule from a complex blueprint, that is, a diagram defining the arrangement of all atoms in space, is at the very heart of modern retro-synthetic chemistry1, transition-state-designed catalysis2 and supramolecular scaffolding3. Creating nanomaterials should be no exception to such blueprinting, but the crafting of them for electronics applications has been greatly limited by the lack of methods that enable the accurate positioning of elements (molecules) in space4–6. In this Perspective we highlight top-down and bottom-up approaches that are converging towards the engineering of singlemacromolecular devices for organic (nano)electronics. In particular, we explore both approaches and their limitations to accurately reproducing the desired atomic layout or ‘molecular blueprint’ needed for the efficient production and integration of molecular circuitry. The top-down approach cannot reproduce a molecular blueprint with chemical accuracy, whereas the bottom-up surface syntheses approaches still lack full thermodynamic and kinetic control over the desired blueprint product (Fig. 1). We discuss how to overcome these limitations with the ultimate goal of manufacturing fully integrated all-organic (nano)devices. Graphene nanoribbons as electronic components Despite promises of inexpensive logic applications with chemically tunable ‘plastic’ materials, active components of organic electronics such as molecules7, polymers8 and carbon nanotubes9, are not yet competitive with existing silicon technology. In miniaturization, organic materials are confronted with two major challenges: the inherent fabrication of a macromolecular blueprint and its interfacing with metal electrodes4. In this regard, graphene holds great promise as an organic component capable of replacing silicon technology10 in the Moore paradigm11 because it is a large two-dimensional crystal suitable for carving, with low contact resistance12. Although not a semiconductor13, graphene can be cut up into smaller objects such as graphene nanoribbons (GNRs), which opens up a bandgap14–16. Figure 2a shows how the theoretical bandgap in GNRs was found to depend critically on their atomic blueprint. The data is from refs 15 and 16, in which different levels of density functional theory were employed (namely the local spin density approximation and its GW correction16 and the generalized gradient approximation through the PBE (Perdew-Burke-Enzerhof) functional and the screened exchange hybrid density functional15). The inset depicts the corresponding atomic structure of a D2h symmetric armchair GNR, showing that the architecture’s width can be modulated approximately by multiples of the lattice constant of graphene, a = 2.47 Å (red arrows in Fig. 2a). Although an experimental bandgap in such sub-3-nm wide ribbons is yet to be measured, interesting semiconducting properties were already detected in larger GNRs. In 2008, preliminary devices based on GNRs17–19 showed transistor properties with on/off rations of 1 × 107 and single molecular transistor behaviour, already surpassing properties of previous organic20,21 and silicon nanowire22 devices. However, the lack of reproducibility and practical integration of graphene nanoribbons into devices is still a limiting factor. This can be exemplified in the number of theoretical versus experimental reports on graphene nanoribbons during the past five years, where experimental works are outnumbered by a factor of four (out of 299 records matching the topic ‘graphene nanoribbons’ in ISI Web of Science updated 21/8/2010, 239 were found not to contain the patterns litho, etch, pattern, synthe in the title or abstract). In fact, Fig. 2a also serves as a bright example on how the ultimate goal of sp2-carbon electronics is to achieve atomic precision, not only for tuning the device properties but also for reproducibility of the electronic characteristics. More importantly, the design beyond complementary metal-oxide semiconductor technologies, such as valley valves23 and single-electron transistors19, relies on the possibility of atomic control over graphene. The question arises of how carbon electronics with precise molecular structures can be embedded into the device configuration of choice. More generally, how hundreds of moieties can be stitched together into a precise million kilodalton macromolecule or how small can a material be precisely cut into a well-defined macromolecule? To this end, we present recent works that have shed light on the cutting and stitching of organic molecules to reproduce graphene nanoribbon molecular blueprints for (nano)electronics. Blueprinting from the top The most straightforward manner to follow a blueprint in order to craft a single block of organic circuitry is through lithography or Nanochemistry Laboratory, ISIS – CNRS 7006, Université de Strasbourg, 8 allée Gaspard Monge, 67000 Strasbourg, France. †Present address: Max-Planck Institute for Polymer Research, 55128, Mainz, Germany. *e-mail: samori@unistra.fr NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved 431 PERSPECTIVE NATURE CHEMISTRY DOI: 10.1038/NCHEM.1043 a b 4 nm 2 nm m (i) - Pattern mask (ii) - Etch (iii) - Remove mask X X X X ~20 nm +X Substrate X X n X Cyclodehydrogenation X 2 nm X Side products Side products l Side products Figure 1 | Ideal and reality. Generation of a simple element of circuitry reflecting an exact molecular blueprint (images with blue background). a, A topdown strategy through the etching or carving63 of a single graphene sheet with a resolution of ~20 nm, which could be subsequently reduced to ~5nm (ref. 28). The uncertainty between the exact formation of the product and the desired molecular blueprint depends on the precise etching of single carbon atoms. Top-right, AFM image taken from ref. 28. b, A bottom-up approach through pre-programmed synthesis42. The uncertainty (due to entropy) between the products and the molecular blueprint in this case depends upon the (physico-)chemical level of pre-programming for reducing undesired isomers and side products. Top-right, STM image from ref. 42; scale bar, 3 nm. Note that in both cases a strict blueprint should also comprise the position and orientation of a substrate and other elements, if any. The X substituents represent labile atoms such as halogens. All carbons shown are sp2 hybridized. engraving methods, which can be accomplished at the nanoscale with the help of electromagnetic radiation and/or high-throughput sharp probes. In graphene, lithographic methods include masking and chemical etching19,24 or the thermal–electrochemical reduction of graphene or graphene oxide with sharp probes25–27. However, such methodologies have only produced uncertain macromolecular structures with widths not smaller than ~20 nm, and bandgaps (Eg) close to kT (~26 meV at 298 K)24. Scuseria et al.15 predicted in 2006 that for GNRs to be technologically competitive semiconductors, their width should be effectively controlled down to ~3 nm, or Eg = 0.5 – 0.3 eV. Just recently, chemical etching techniques have been developed in which patterning of graphene can be achieved with ~5 nm resolution. By using a gas mixture of ammonia and oxygen on pre-etched patterns, Dai and Wang28 could etch carbon nanoribbons with ~1-nm-per-minute control. As a result ~5-nm-wide GNRs could be produced with transistor properties showing on/off ratios as high as 1 × 104, the record among lithographically fabricated GNR devices. Likewise, Zhang and co-workers29 reported the production of ~8-nm-wide nanoribbons through the use of H2-plasma as an etching agent (Fig. 2b). They showed that the etching is anisotropic, working preferentially along the armchair edge, allowing the generation of zig-zag nanoribbons along the [21̄10] direction. Carving 2D materials is not the only route to accomplishing this goal — nanoscopic 3D materials have also been cut down to create elements of graphene circuitry30,31. Such top-down strategies may be ultimately applied to reproduce an exact molecular blueprint (Fig. 1a). Chemical etching has come near this goal because the H2-plasma technique allows the etching of single carbon atoms, that is, etching rates corresponding to 2.7 ± 0.5 Å min–1 (ref. 32). However, the mean edge roughness of the GNR in the H2-plasma approach amounts to 2 nm (ref. 29), and introduces random topological factors. It is worth pointing out that, so far, the maximum resolution of chemically precise patterning has been obtained by scanning tunnelling tip-induced reactions, showing ~2-nm-resolution patterning of organic self-assembled monolayers into nanowires33 (Fig. 2c). Nonetheless, scanning tunnelling tip-induced reactions have only been demonstrated on particular diacetylene self-assembled monolayer systems, in which the polymerization occurs perpendicular to the molecular axis34,35. In contrast, organic thin films of semiconducting polymers can be patterned freely with record resolutions below 28 nm (ref. 36). Unfortunately, also in this case the as-written blueprint is not molecularly precise. 432 Because of the factors outlined above, a molecular blueprint is still not ready to be reproduced solely through top-down approaches, although the development of new high-resolution etching agents, such as nanoparticles37, and the design of new ordered monolayers may one day allow the direct top-down patterning of defined macromolecules. Blueprinting through synthetic programming The need for such a high degree of atomic (thus bandgap and energy level) precision in the crafting of molecular devices can be accomplished by engineering protocols relying on the covalent or non-covalent38 tethering of small units to form the designed molecular blueprints. By mastering principles of heterogeneous reactions and catalysis39–41, on-surface macromolecular synthesis seems feasible. In this light, it is natural to identify the critical steps for developing chemistry compatible with different surfaces. In addition to catalysis and reaction design, surface-supported chemistry allows the reduction of the six degrees of rotational and translational freedom to (ideally) only one. This can be effectively thought of as a reduction of the absolute translational and rotational entropy, thus immediately constituting one level of thermodynamic programming. Combined with computer-simulated mechanistic studies, de novo design of the precursors yielding highly defined macromolecular adducts may become reality, thus molecular engineering of accurate organic electronic blueprints may also become a reasonable target. In the next section, we put forward evidence to show how early computer simulations and empirical observations have identified preliminary rules for the synthesis of extended architectures at surfaces. Although still far from predicting precise protocols for the synthesis of electronic circuitry, the examples demonstrate how graphene materials are just the tip of the iceberg with respect to new materials for electronics. In particular for the early examples of radical additions42–47, two strategies may already be put forward for the synthesis of extended macromolecular architectures: those in which the reaction rate is diffusion-limited and those where the rate is limited by the reaction probability. Diffusion-limited on-surface synthesis In diffusion-limited radical reactions, the macromolecular rate of growth is dominated by the passive transport of the elements to the reaction centre of the molecular blueprint, each with a hopping rate NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved PERSPECTIVE NATURE CHEMISTRY DOI: 10.1038/NCHEM.1043 a 5 1 Theoretical bandgap (eV) 4 2 3 Width in lattice constants (a) 4 5 6 Ref. 16 (LSDA) Ref. 16 (GW) Ref. 15 (PBE) Ref. 15 (HSE) 7 8 9 a 10 c 2 nm 11 a ~ 2.47 Å 1.65 nm 3 M n; n>>M 2 b 1.8 eV 0.8 eV 0.8 eV Ag (111) 0.3 eV 2.2 eV 0 0 1 2 3 4 5 6 7 8 b 2.0 eV 1 nm Cu (111) 9 Width in repeat units (M) 2.3 eV HH c 0.2 eV 0.1 eV 0.9 eV 0.2 eV 1 0.7 eV 1.7 eV HH H H H H H H H H H H H H + ~8 nm ~8 nm HH Diffusion ~12 nm ~22 nm m m 6n 6n 250 nm 500 nm Figure 2 | Cutting a macromolecule into functional electronic elements. a, The theoretical bandgap for the D2h symmetry armchair nanoribbon, Dh-AGNR (inset); hydrogens are implicit. Red arrows depict the lattice constant. M represents the number of lattice constants in the (01̄10) direction, starting at a minimum of 2, and the ribbon is assumed to be much larger in the perpendicular direction n (data from refs 15,16). b, Cutting graphene into ~8–22-nm-long ribbons. Image reproduced with permission from ref. 29, © 2010 Wiley. c, Stitching top-down strategy using a tunnelling microscopy tip to regio-polymerize one row of di-acetylene monolayers into a molecular wire (left) and an adjacent row (right)33. ~exp(–Ediff/kT); where for small molecules the energy of diffusion, Ediff, is nearly 20% of the adsorption energy48. Such a dependence of the diffusion rate on the adsorption energy was first reported qualitatively by Hecht, Grill and co-workers49. By growing 2D clusters of tetrakis-bromophenyl porphyrins on Au(111), they found that if the porphyrin precursors (radicals) were formed before the deposition on Au, the size of the formed clusters was smaller (Fig. 3a), probably due to stronger chemisorption of the tetra-radical porphyrins. This demonstrated that hindering random nucleation is crucial to allow ordered growth, and relies on the fact that small clusters or islands adsorb more strongly and will not diffuse nor fuse to form single-crystalline structures. On the other hand, when engineering of small macromolecular clusters is sought, then nucleation should be promoted over growth50. In either case, kinetic and thermodynamic control over the plethora of subproducts that may also be formed is key for synthesis of the desired material blueprint with nanoscale precision. A quantitative illustration for promoting ordered growth over random nucleation has been recently put forward by Bieri, Fasel and co-workers51. By using a 2D porous-graphene precursor45, a hexaido cyclohexa-m-phenylene, they showed that crystalline domains are formed preferentially on Ag(111) and Au(111) whereas only random networks are observed on Cu(111) — on annealing H H H H HH Reaction 4 nm Figure 3 | Stitching molecules together into macromolecular architectures. a, A small organic nanostructure made from eight covalently bound porphyrins at the Au interface49. b, Potential energy landscape of two reacting cyclohexa-m-phenylene radicals on Ag (top) and Cu (bottom) interfaces. The combination of smaller diffusion barriers with high reactions barriers in Ag results in the formation of crystalline domains (top right) whereas the opposite results in the formation of amorphous networks (bottom right). c, Epitaxial synthesis of EDOT trimer along one commensurate crystallographic direction on Cu(110). Insets: DFT atomic structure (top; blue bar indicates 1.65-nm scale bar shown in main image) and a ~12 unit oligomer (bottom). Images in b and c reproduced with permission from: b, (top) ref. 45, © 2009 RSC; b (bottom) ref. 51, © 2010 ACS; c, ref. 43, © 2010, NAS, USA. above 525, 575 and 475 K, respectively, for each metal surface (Fig. 3b). By using Monte-Carlo simulations, the authors were able to reproduce the exact experimental observations. Assigning high reaction probabilities causes the simulations to match the Cu porous graphene-grown structures, whereas lower probabilities match the more crystalline Ag and Au structures. Further, they compared the reaction probabilities with DFT transition-state studies: at closing reaction geometries molecules have a 2.2 eV diffusion barrier on Cu(111) whereas they have a 0.8 eV barrier on Ag(111). Precursors on copper readily react after the diffusive process (Fig. 3b, bottom) but they present an additional 1.8 eV barrier on silver substrates (Fig. 3b, top). The previous barriers are translated into diffusionlimited rates on copper and reaction-limited rates on silver. It can be generalized that diffusion-limited polymerizations may lead to formation of short 1D polymer chains and disordered networks. One strategy to avoid the problem of random coupling on account of small diffusion coefficients is to have molecules undergoing chemisorption onto specific binding sites and subsequently reacting with adjacent molecules. This condition is reminiscent of epitaxial growth52. The strategy, however, requires both engineering of the reaction and the substrate, increasing the programming level of complexity; Rosei, Perepichka and co-workers demonstrated the successful polymerization of 3,4-ethylenedioxythiophene (EDOT) monomers forming uniaxially aligned polymers (PEDOT)43, the doped form of the latter being among the most employed organic conductive materials. The polymerization reaction occurred on heating the EDOT-coated Cu(110) surface at 500 K, leading to NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved 433 PERSPECTIVE NATURE CHEMISTRY DOI: 10.1038/NCHEM.1043 a b 2 nm c 20 nm 2 nm d 3 nm Figure 4 | Programming surface reactions into extended macromolecular architectures. a, A ~100-nm-long surface-synthetized polyfluorene chain. Arrows indicate the covalent bonds on the chain with respect to the chemical structure. b, A large-area 2D surface covalent polygonal network. Inset: higher magnification image; 56 × 56 nm2. c, Atomically precise graphene nanoribbons from small molecule precursors42. d, A covalent sp2-carbon macromolecule representing the versatility of a programmed synthetic approach42. Images in a and b reproduced with permission from: a, ref 47, © 2009 AAAS; b, ref. 54, © 2010 RSC. PEDOT polymer sizes of up to ~6 nm in length, depending on the surface coverage. Figure 3c depicts EDOT oligomers perfectly aligned commensurably with the substrate43. The growth takes place through the adsorption of the radical precursor, which was theoretically estimated to amount to an energy of 5.92 eV. Such high adsorption energy may be responsible for negligible molecular diffusion at a temperature of about 500 K, thus such reactions may also be regarded as diffusion-limited, depending only on the rate of surface coverage. Reaction-limited on-surface synthesis From the previous section, it seems that most reactions on surfaces are diffusion-limited to some extent, given diffusion barriers will always be present. Besides the magisterial example of Bieri and Fasel51 in tuning the molecule–substrate interactions to minimize reactivity, inherent molecular design can be also programmed to allow fast mobilities. Towards this goal, in a subsequent experiment47, Hecht, Grill and co-workers designed a laterally methylated fluorene building block for the synthesis of polyfluorene. The methyl groups prevent the molecule from adsorbing flat onto the surface, thus the fast mobility of the molecules allowed the formation of long polyfluorene chains, greater than 100 nm in length, on polymerization at 520 K on Au(111) (Fig. 4a). The strategy of achieving high surfacediffusion rates through a non-planar precursor can be also used in the formation of 2D networks. Introduced by Gutzler, Lackinger and co-workers53, radical homolysis and subsequent polymerization of tris(4-bromophenyl)benzene molecules was found to yield small 2D polygonal domains on Cu(111). Blunt, Beton and co-workers54 showed that larger domains may be formed on Au(111), which may also accommodate fullerenes (Fig. 4b). One higher level of chemical programming has been recently demonstrated specifically for the synthesis of long well-defined graphene nanoribbons. Fasel, Müllen and co-workers42 employed surface-activated synthesis of molecular building blocks to create armchair nanoribbons with lengths up to ~40 nm (Fig. 4c). By changing the chemical structure of the reactant, that is, the starting molecular building block, they could tailor the structure 434 of the nanoribbons. For instance, a bisanthracene halide building block yields the straight ribbon depicted in Fig. 4c, whereas using an asymmetric hexaphenylbenzene halide results in a chevrontype ribbon. Moreover, multiprecursors were successfully used to finally reproduce a molecular blueprint (Fig. 1b) yielding a complex macromolecular component of electronic importance (Fig. 4d). In light of such a proof of principle, new avenues of exploration can be undertaken for extending the rules governing reactions in a variety of elements and substrates, that is, a grammar of synthesis on surfaces. The reaction is also a bright example of a programmed reaction for fast diffusion rates: this is because the starting molecular building blocks have a 3D conformation and as such their clusters may polymerize swiftly at 500 K on Au(111) through radical formation. Once the final macromolecule length is reached, a cyclodehydrogenation55 reaction is induced with 700 K heating of the sample, thus increasing the adsorption energy and consolidating the graphene nanoribbons. The biggest challenge after reaction programming and surface diffusion optimization, as described in the above examples, will be the reduction of isothermal entropy. Every possible degree of freedom that can be populated at the reacting temperature may give a macromolecular product, and only one among the various populated states may correspond to the desired molecular blueprint (Fig. 1b). Furthermore, there are a few disadvantages limiting the practical syntheses of prototype electronic devices from molecular blueprints made by using surface synthesis protocols. The most important is the use of the conducting substrates Au(111) and Cu(111) as 2D supports of the macromolecular device. Secondly, the Ullmann-like coupling35,44,56,57 mechanism characteristic of the previous radical additions, has also the drawback of contaminating the surface with the halogen side-products43,45, unless sufficiently high temperatures are employed42. In general, until a quantitative tool for engineering covalent synthesis is established, reaction products may have to be mechanically processed for their immediate technological application. Supermolecule resists or nanoreactors? Having pointed out the state-of-the-art in atomically cutting and stitching new materials, solutions for leveraging the fundamental science presented above to a real technology of complex macromolecular blueprints are required. Such solutions should consist of large-area and atomically precise patterns of active electronic components and interconnect paths wired to the outside world. At the beginning of this Perspective, we pointed out how cutting a 2D polymer such as graphene with atomic precision may be used to reproduce chemically precise blueprints. Indeed it is not too daunting to foresee inorganic nanowires acting as nanoprobes for chemically cutting groups of carbon atoms precisely. However, for efficient circuitry patterning to be possible in the future, more challenging strategies should be considered. One possibility relies on the development of highly ordered organic (mono) layers for polymerization using local external stimuli such as thermal gradients, high-energy photons or low-energy electrons. To polymerize or crosslink individual molecules into a precise macromolecule, the incident high-resolution reaction stimulus should only affect one molecule at a time. Given that directwriting nanofabrication methods have demonstrated spatial precision in the 10–30 nm length scale, the size of the single molecular building block to be successively crosslinked should be ~10 nm, thus allowing the precise formation of a macromolecule on radiative or thermal reaction. The previous strategy translates into the synthesis of monomers containing less than 1,500 atoms or consisting of giant macrocycles, with a defined polymerization chemistry and hence the term ‘supermolecule resists’. The ultimate achievement would be to crosslink supermolecule resists through stimulated topological reactions58. NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved PERSPECTIVE NATURE CHEMISTRY DOI: 10.1038/NCHEM.1043 An approach expected to be developed in the immediate future involves template control over the on-surface reactions described previously. This time, the circuit’s source and drain electrodes can be pre-patterned alongside metallic interconnects on insulated surfaces. When such electrodes are made of catalytic materials for surface-assisted chemistry they may be regarded as 2D nanoreactors. In such a scenario, the prior engineering of the geometry of the nanoreactors will make it possible to polymerize in situ macromolecules that extend over insulated areas and bridge the nanoelectrodes, forming a macromolecular transistor channel. An early example of how on-surface-synthesized molecular components extend over non-catalytic dielectric areas was reported by Grill, Hecht and co-workers59, who showed that the polyfluorene synthesis also occurs in the presence of NaCl clusters. This time, the on-surface-synthesized polyfluorene chains indeed extend randomly over insulating NaCl islands. Essentially, the suggestions we put forward critically depend on advancements in two fields, reaction engineering, and preparation of (single crystal) catalytic nanostructures. Reaction engineering requires fast development of computational chemistry, for example quantum chemistry and molecular dynamics, both of which depend on contemporary computational limits. On the other hand, patterning single crystals and geometries is subject to advances in the rich field of inorganic epitaxial nanofabrication. Addressing the previous issues alongside expanding empirical contributions to the fields is fundamental for the translation of macromolecules into technology. Summary The engineering of structurally well-defined individual electroactive macromolecules can be accomplished by mastering and combining the top-down and the bottom-up approaches. The endless combinations of chemistry60 will, in the years to come, offer an exciting and versatile — yet winding — avenue to molecular electronics. Indeed bottom-up chemistry already yields GNRs61 and 2D (nano)materials62. Although structurally well-defined electronic elements can now be accurately reproduced — like long, monodisperse synthetic nanoribbons — cumbersome stamping or transferring techniques would be required to manufacture devices; techniques that can’t be envisaged as strategies for nanofabrication. We may foresee that one day a single well-defined covalent macro- or mega-molecule will become, in itself, a logical computing element. Such a view may only become reality when experimental and theoretical efforts give access to a library of surface-programmed reactions, rendering synthetic chemistry at surfaces a predictive science. Reaction engineering may converge and blur the horizons of top-down and bottom-up methods into those of integrated circuitry based, for example, on supermolecule resists or nanoreactor patterning. In this fascinating view, we see a joint future effort of experimental and theoretical chemists and physicists to a priori design sp2-carbon (graphene-like) molecular blueprints. Here, theory will not only offer the design and modelling of molecular blueprints but will help actively in the programming of the chemistry and thermodynamics of reactions at surfaces. In 1965, G. E. Moore described quite accurately that “The future of integrated electronics is the future of electronics itself ”. Extending this analogy, the future of macro- or mega-molecular blueprinting is the future of molecular electronics itself. References 1. Corey, E. J. Retrosynthetic thinking – essentials and examples. Chem. Soc. Rev. 17, 111–133 (1988). 2. Brady, P. A. & Sanders, J. K. M. Selection approaches to catalytic systems. Chem. Soc. Rev. 26, 327–336 (1997). 3. Lehn, J.-M. Supramolecular Chemistry: Concepts and Perspectives (Wiley-VCH, 1995). 4. Moth-Poulsen, K. & Bjornholm, T. Molecular electronics with single molecules in solid-state devices. Nature Nanotech. 4, 551–556 (2009). 5. Geim, A. K. Graphene: status and prospects. Science 324, 1530–1534 (2009). 6. Metzger, R. M. Unimolecular electronics. J. Mater. Chem. 18, 4364–4396 (2008). 7. Joachim, C,. Gimzewski, J. K. & Aviram, A. Electronics using hybrid-molecular and mono-molecular devices. Nature 408, 541–548 (2000). 8. Arias, A. C,. MacKenzie, J. D,. McCulloch, I,. Rivnay, J. & Salleo, A. Materials and applications for large area electronics: solution-based approaches. Chem. Rev. 110, 3–24 (2010). 9. Baughman, R. H,. Zakhidov, A. A. & de Heer, W. A. Carbon nanotubes - the route toward applications. Science 297, 787–792 (2002). 10.Schwierz, F. Graphene transistors. Nature Nanotech. 5, 487–496 (2010). 11.Moore, G. E. Cramming more components onto integrated circuits. Proc. IEEE 86, 82–85 (1998). 12.Xia, F,. Perebeinos, V,. Lin, Y.-M,. Wu, Y. & Avouris, P. The origins and limits of metal-graphene junction resistance. Nature Nanotech. 6, 179–184 (2011). 13.Novoselov, K. Mind the gap. Nature Mater. 6, 720–721 (2007). 14.Nakada, K,. Fujita, M,. Dresselhaus, G. & Dresselhaus, M. S. Edge state in graphene ribbons: Nanometer size effect and edge shape dependence. Phys. Rev. B 54, 17954–17961 (1996). 15.Barone, V,. Hod, O. & Scuseria, G. E. Electronic structure and stability of semiconducting graphene nanoribbons. Nano Lett. 6, 2748–2754 (2006). 16.Yang, L,. Park, C.-H,. Son, Y.-W,. Cohen, M. L. & Louie, S. G. Quasiparticle energies and band gaps in graphene nanoribbons. Phys. Rev. Lett. 99, 186801 (2007). 17.Li, X. L,. Wang, X. R,. Zhang, L,. Lee, S. W. & Dai, H. J. Chemically derived, ultrasmooth graphene nanoribbon semiconductors. Science 319, 1229–1232 (2008). 18.Avouris, P,. Chen, Z. H. & Perebeinos, V. Carbon-based electronics. Nature Nanotech. 2, 605–615 (2007). 19.Ponomarenko, L. A. et al. Chaotic dirac billiard in graphene quantum dots. Science 320, 356–358 (2008). 20.Murphy, A. R. & Fréchet, J. M. J. Organic semiconducting oligomers for use in thin film transistors. Chem. Rev. 107, 1066–1096 (2007). 21.Braga, D. & Horowitz, G. High-performance organic field-effect transistors. Adv. Mater. 21, 1473–1486 (2009). 22.Huang, Y. et al. Logic gates and computation from assembled nanowire building blocks. Science 294, 1313–1317 (2001). 23.Rycerz, A,. Tworzydlo, J. & Beenakker, C. W. J. Valley filter and valley valve in graphene. Nature Phys. 3, 172–175 (2007). 24.Han, M. Y,. Ozyilmaz, B,. Zhang, Y. B. & Kim, P. Energy band-gap engineering of graphene nanoribbons. Phys. Rev. Lett. 98, 206805 (2007). 25.Mativetsky, J. M. et al. Local current mapping and patterning of reduced graphene oxide. J. Am. Chem. Soc. 132, 14130–14136 (2010). 26.Wei, Z. Q. et al. Nanoscale tunable reduction of graphene oxide for graphene electronics. Science 328, 1373–1376 (2010). 27.Tapasztó, L,. Dobrik, G,. Lambin, P. & Biró, L. P. Tailoring the atomic structure of graphene nanoribbons by scanning tunnelling microscope lithography. Nature Nanotech. 3, 397–401 (2008). 28.Wang, X. R. & Dai, H. J. Etching and narrowing of graphene from the edges. Nature Chem. 2, 661–665 (2010). 29.Yang, R. et al. An Anisotropic etching effect in the graphene basal plane. Adv. Mater. 22, 4014–4019 (2010). 30.Kosynkin, D. V. et al. Longitudinal unzipping of carbon nanotubes to form graphene nanoribbons. Nature 458, 872–876 (2009). 31.Jiao, L,. Zhang, L,. Wang, X. R,. Diankov, G. & Dai, H. J. Narrow graphene nanoribbons from carbon nanotubes. Nature 458, 877–880 (2009). 32.Xie, L,. Jiao, L. & Dai, H. J. Selective etching of graphene edges by hydrogen plasma. J. Am. Chem. Soc. 132, 14751–14753 (2010). 33.Okawa, Y. & Aono, M. Nanoscale control of chain polymerization. Nature 409, 683–684 (2001). 34. Akai-Kasaya, M. et al. Electronic structure of a polydiacetylene nanowire fabricated on highly ordered pyrolytic graphite. Phys. Rev. Lett. 91, 255501 (2003). 35.Hla, S.-W,. Bartels, L,. Meyer, G. & Rieder, K.-H. Inducing all steps of a chemical reaction with the scanning tunneling microscope tip: towards single molecule engineering. Phys. Rev. Lett. 85, 2777 (2000). 36.Fenwick, O. et al. Thermochemical nanopatterning of organic semiconductors. Nature Nanotech. 4, 664–668 (2009). 37.Campos, L. C,. Manfrinato, V. R,. Sanchez-Yamagishi, J. D,. Kong, J. & Jarillo-Herrero, P. Anisotropic etching and nanoribbon formation in singlelayer graphene. Nano Lett. 9, 2600–2604 (2009). 38.Ciesielski, A,. Palma, C. A,. Bonini, M. & Samorì, P. Towards supramolecular engineering of functional nanomaterials: pre-programming multi-component 2D self-assembly at solid-liquid interfaces. Adv. Mater. 22, 3506–3520 (2010). 39.Bent, B. E. Mimicking aspects of heterogeneous catalysis: generating, isolating, and reacting proposed surface intermediates on single crystals in vacuum. Chem. Rev. 96, 1361–1390 (1996). 40.Ma, Z. & Zaera, F. Organic chemistry on solid surfaces. Surf. Sci. Rep. 61, 229–281 (2006). NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved 435 PERSPECTIVE NATURE CHEMISTRY DOI: 10.1038/NCHEM.1043 41.Masel, R. L. Principles of Adsorption and Reaction on Solid Surfaces (Wiley, 1996). 42.Cai, J. M. et al. Atomically precise bottom-up fabrication of graphene nanoribbons. Nature 466, 470–473 (2010). 43.Lipton-Duffin, J. A. et al. Step-by-step growth of epitaxially aligned polythiophene by surface-confined reaction. Proc. Natl. Acad. Sci. USA 107, 11200–11204 (2010). 44.Szulczewski, G. J. & White, J. M. Thermal and photon-stimulated reactions of iodobenzene on Ag(111). Surf. Sci. 399, 305–315 (1998). 45.Bieri, M. et al. Porous graphenes: two-dimensional polymer synthesis with atomic precision. Chem. Commun. 6919–6921 (2009). 46.Lipton-Duffin, J. A,. Ivasenko, O,. Perepichka, D. F. & Rosei, F. Synthesis of polyphenylene molecular wires by surface-confined polymerization. Small 5, 592–597 (2009). 47.Lafferentz, L. et al. Conductance of a single conjugated polymer as a continuous function of its length. Science 323, 1193–1197 (2009). 48.Ertl, G. Reactions at Solid Surfaces (Wiley, 2009). 49.Grill, L. et al. Nano-architectures by covalent assembly of molecular building blocks. Nature Nanotech. 2, 687–691 (2007). 50.Lu, J,. Yeo, P. S. E,. Gan, C. K,. Wu, P. & Loh, K. P. Transforming C60 molecules into graphene quantum dots. Nature Nanotech. 6, 247–252 (2011). 51.Bieri, M. et al. Two-dimensional polymer formation on surfaces: insight into the roles of precursor mobility and reactivity. J. Am. Chem. Soc. 132, 16669–16676 (2010). 52.Shchukin, V. A,. Ledentsov, N. N. & Bimberg, D. Epitaxy of Nanostructures (Springer, 2004). 53.Gutzler, R. et al. Surface mediated synthesis of 2D covalent organic frameworks: 1,3,5-tris(4-bromophenyl)benzene on graphite(001), Cu(111), and Ag(110). Chem. Commun. 4456–4456 (2009). 54.Blunt, M. O,. Russell, J. C,. Champness, N. R. & Beton, P. H. Templating molecular adsorption using a covalent organic framework. Chem. Commun. 46, 7157–7159 (2010). 436 55.Treier, M. et al. Surface-assisted cyclodehydrogenation provides a synthetic route towards easily processable and chemically tailored nanographenes. Nature Chem. 3, 61–67 (2010). 56.Ullmann, F. & Bielecki, J. Ueber Synthesen in der Biphenylreihe. Ber. Dtsch. Chem. Ges. 34, 2174–2185 (1901). 57.Xi, M. & Bent, B. E. Mechanisms of the Ullmann coupling reaction in adsorbed monolayers. J. Am. Chem. Soc. 115, 7426–7433 (1993). 58. Sakamoto, J,. van Heijst, J,. Lukin, O. & Schlüter, A. D. Two-dimensional polymers: just a dream of synthetic chemists? Angew. Chem. Int. Ed. 48, 1030–1069 (2009). 59.Bombis, C. et al. Single molecular wires connecting metallic and insulating surface areas. Angew. Chem. Int. Ed. 48, 9966–9970 (2009). 60.Grzybowski, B. A,. Bishop, K. J. M,. Kowalczyk, B. & Wilmer, C. E. The ‘wired’ universe of organic chemistry. Nature Chem. 1, 31–36 (2009). 61.Hod, O,. Barone, V,. Peralta, J. E. & Scuseria, G. E. Enhanced half-metallicity in edge-oxidized zigzag graphene nanoribbons. Nano Lett. 7, 2295–2299 (2007). 62.Perepichka, D. F. & Rosei, F. Extending polymer conjugation into the second dimension. Science 323, 216–217 (2009). 63.Geim, A. K. & Novoselov, K. S. The rise of graphene. Nature Mater. 6, 183–191 (2007). Acknowledgements We thank Matthias Treier and Veronica Barone for sharing helpful data and for useful discussions. This work was financially supported by the EC MarieCurie RTNs PRAIRIES (MRTN-CT-2006-035810) and THREADMILL (MRTN-CT-2006-036040) as well as the ITNs SUPERIOR (PITN-GA-2009-238177) and GENIUS (PITN-GA-2010-264694), the EC FP7 ONE-P large-scale project no. 212311, the NanoSci-E+ project SENSORS and the International Center for Frontier Research in Chemistry (FRC, Strasbourg). Additional information The authors declare no competing financial interests. NATURE CHEMISTRY | VOL 3 | JUNE 2011 | www.nature.com/naturechemistry © 2011 Macmillan Publishers Limited. All rights reserved