nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene concepts is thus fundamental to reduce the electrode. A hybrid poly(3,4 overall costs and increase efficiency. ethylenedioxythiophene):poly(styrenesulphonat e) (PEDOT:PSS) graphene oxide composite was Graphene can fulfil multiple functions in used as counter-electrode, to obtain η = 4.5%, photovoltaic devices: TC window, antireflective comparable to the 6.3% for a Pt counter- layer, photoactive material, cannel for charge electrode tested under the same conditions transport and catalyst [104]. GTCFs can be used [239] but now with a cheaper material. More as window electrodes in inorganic [231], importantly, it was recently demonstrated organic [232] and dye-sensitized solar cells that graphene can be used as counter- (DSSCs) [223], see Fig. 22 a,b,c respectively. The electrode material to replace simultaneously best performance achieved to date has η≈ 1.2% both Pt as catalyst and the TC oxide as using CVD graphene as the TC, with Rs values conductive electrode [240]. This is a of 230Ω/□ and T=72% [233]. However, further fundamental step towards cost reduction and optimization is certainly possible, considering large scale integration of DSSCs. that GTCFs with Rs=30Ω/□ and T=90% have already been demonstrated [3] and graphene- Current solar cell technologies use only a hybrids have been reported with even better small part of the solar spectrum [212], due to results (Rs 20Ω/□, T=90%) [221]. their intrinsic band gap limiting the maximum detectable wavelength. The absence of a GO dispersions were also used in bulk band-gap in graphene translates into the heterojunction photovoltaic devices, as absence of this detectable wavelength limit. electron-acceptors achieving η≈1.4% [234]. This means that solar radiation over a much Theoretically η~12% should be possible with wider spectral range could be converted to graphene as photoactive material [235]. energy. Graphene can cover an even larger number of The combination of graphene with plasmonic functions in DSSCs. Indeed, other than as TC nanostructures can also be used to increase window [223], graphene can be incorporated the light harvesting properties in solar cells into the nanostructured TiO2 photoanode to [241]. enhance the charge transport rate, preventing recombination, thus improving the 8.3 Organic Light Emitting Diodes internal photocurrent efficiency [236]. An efficiency of ~7%, higher than conventional Organic light-emitting diodes (OLEDs) are a nanocrystalline TiO2 photoanodes in the same class of optoelectronic devices that can take experimental conditions, was demonstrated advantage of graphene. Low power in Ref. [236]. Graphene quantum dots with consumption and ultra-thin OLEDs have been tuneable absorption have been designed, developed more than 20 years ago [242], and produced and demonstrated as promising are now applied in ultra-thin televisions and photoactive materials in DSSCs [237]. Further other displays, such as those on digital optimization is required for the optimum cameras and mobile phones. adsorption of these molecules with TiO2 nanoparticles by covalently attaching binding OLED have an electroluminescent layer groups to the quantum dots in order to between two charge- injecting electrodes, at improve charge injection. Another option is to least one of which transparent. In these use graphene, with its high surface area [238], diodes, holes are injected into the highest as substitute for the platinum (Pt) counter- occupied molecular orbital (HOMO) of the polymer from the anode, and electrons into nanoICT Strategic Research Agenda 49

Annex 1 nanoICT working groups position papers Graphene the lowest unoccupied molecular orbital and where keyboard and mouse do not allow (LUMO) from the cathode. For efficient a satisfactory, intuitive, quick, or accurate injection, the anode and cathode work interaction with the display content. functions should match the HOMO and LUMO of the light-emitting polymer. Traditionally, Resistive and capacitive (see Fig. 22e) touch ITO is used as the transparent conductive panels are the most common. A resistive film. However, it has a number of touch panel comprises a conductive disadvantages. First, ITO may be too substrate, a LCD front-panel, and a TCF [245]. expensive for use in OLEDs for lighting When pressed by a finger or pen, the front- because of the increasing cost and the low panel film comes into contact with the throughput deposition process. Second, metal bottom TC and the coordinates of the contact oxides such as ITO are brittle and therefore of point are calculated on the basis of their limited use on flexible substrates. Third, In is resistance values. The TC requirements for known to diffuse into the active layers of resistive screens are Rs ~500−2000Ω/□ and OLEDs, which leads to a degradation of T>90% at 550nm [245]. Favourable performance over time. There is a clear need mechanical properties, including brittleness for alternative TCEs with optical and electrical and wear resistance, high chemical durability, performance similar to ITO but without its no toxicity, and low production costs are also drawbacks. Graphene has a work function of important. GTCFs can satisfy the 4.5 eV, similar to ITO. This, combined with its requirements for resistive touch screens in promise as a flexible and cheap transparent terms of T and Rs, when combined with large conductor, makes it an ideal candidate for an area uniformity. Ref. [3] reported a graphene- OLED anode (Fig. 22d), while also eliminating based touch panel by screen-printing a CVD issues related to In diffusion. sample. Considering the Rs and T required by analogue resistive screens, GTCF or GOTCF Ref [243] developed an OLED with a few produced via LPE also offer a viable nanometers of graphene as transparent alternative, and further cost reduction. conductor. Ref [244] reported a flexible OLED based on a modified graphene anode having a Capacitive touch screens are emerging as the high work function and low Rs. The high-end version, especially since the launch of performance (37.2 lm W–1 in fluorescent Apple’s iPhone. These consist of an insulator OLEDs, 102.7 lm W–1 in phosphorescent such as glass, coated with ITO [245]. As the OLEDs) outperforms optimized devices with human body is also a conductor, touching the an ITO anode (24.1 lm W–1 in fluorescent surface of the screen results in electrostatic OLEDs, 85.6 lm W–1 in phosphorescent OLEDs) field distortion, measurable as a change in [244]. capacitance. Although capacitive touch screens do not work by poking with a pen, thus These results pave the way for inexpensive mechanical stresses are lower with respect to OLED mass production on large-area low-cost resistive ones; the use of GTCFs can improve flexible plastic substrates, which could be performance and reduce costs. rolled up like wallpaper and applied to any substrate However, these solutions do not yet provide 100% satisfaction in terms of user experience, 8.4 Touch screens as touch screens tend to be inert in the way they interact with a user. Also, the proliferation Touch panels are used in a wide range of of icons, virtual keys and densely packed applications, such as cell phones and cameras, 50 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene browsing menus on mobile touch screen photon energy into electrical current. They are widely used in a range of devices [250], displays require increasing cognitive efforts such as integrated optoelectronic circuits, televisions, DVD players, biomedical imaging, from the user in order to be located, remote sensing and control, optical communications, and quantum information distinguished and manipulated. Solutions for technology. Most exploit the internal photoeffect, in which the absorption of low-cognitive effort user interfaces (UI), such as photons results in carriers excited from the valence to the conduction band, outputting vibration enabled tactile feedback, are an electric current. The spectral bandwidth is typically limited by the absorption [250]. currently gaining market momentum by Graphene absorbs from the UV to THz [24,251,252,253]. As a result, graphene-based improving usability, interaction photodetectors (GPD) (Fig.23) could work over a much broader wavelength range. The interoperability, and user acceptance. response time is ruled by the carrier mobility. Graphene has huge mobilities, so it can be Thus far, most active tactile feedback ultrafast [251]. Graphene’s suitability for high- solutions have been implemented through speed photodetection was demonstrated in simple monolithic vibrations of the entire an comms link operating at 10Gbit s−1 [254]. device driven by a single or very few vibrating actuators, typically electromechanical or Figure 23. Graphene-based photodetector [104]. piezoelectric [246]. The types of tactile feedback that can be provided by such The photoelectrical response of graphene has traditional techniques are limited to relatively been investigated both experimentally and basic feedback patterns which are only partially theoretically [254,255,256,257,258]. Although correlated to finger position, perceived audio- the exact mechanism for light to current visual information and actions. Such solutions conversion is still debated [256,259,260], a p–n do not yet provide complete satisfaction in junction is usually required to separate the terms of user experience. Key to this is the photo-generated e–h pairs. Such p–n junctions inability of monolithic vibrations to provide are often created close to the contacts, localized tactile feedback associated with visual because of the difference in the work functions images, and this is related to the difficulty in of metal and graphene [148,261]. Responses implementing tactile feedback directly from a at wavelengths of 0.514, 0.633, 1.5 and 2.4μm display surface. To address the problem of providing localized tactile feedback directly from the display of a device, a flexible and optically transparent graphene-based programmable electrostatic tactile (ET) system was developed [247] capable of delivering localized tactile information to the user’s skin, directly from the display’s surface and in accordance with the displayed visual information. The device is based on the electrovibration a phenomenon [248] that can be explained through electrostatic interaction between the touch surface and the user finger [249]. 8.5 Graphene photodetectors Photodetectors measure photon flux or optical power by converting the absorbed nanoICT Strategic Research Agenda 51

Annex 1 nanoICT working groups position papers Graphene have been reported [255]. Much broader photocurrent for the condition of uniform spectral detection is expected because of the flood illumination on both contacts of the graphene ultrawideband absorption [258]. device. Unless the contacts are made of The operating bandwidth of GPDs is mainly different materials, the voltage/current limited by their time constant resulting from produced at both contacts will be of opposite the device resistance, R, and capacitance, C. polarity for symmetry reasons, resulting in An RC-limited bandwidth of about 640 GHz zero net signal [254,256,264]. was reported for graphene [258], comparable to traditional photodetectors [262]. However, The optimization of the contacts needs to be the maximum possible operating bandwidth is pursed both theoretically and experimentally. typically restricted by their transit time, the Other possible ways of overcoming these finite duration of the photogenerated current restrictions is to utilize plasmonic [250]. The transit-time-limited bandwidth of nanostructures placed near the contacts as graphene photodetectors could be over 1,500 recently demonstrated [241]. Such a field GHz [258]. enhancement, exactly in the area of the p–n junction formed in graphene, can result in a Although an external electric field can significant performance improvement. produce efficient photocurrent generation with an electron–hole separation The significant photothermoelectric efficiency>30% [256], zero source–drain bias contribution to the photocurrent in graphene and dark current operations could be p-n junctions was recently pointed out [260]. achieved by using the internal electric field This regime, which features a long-lived and formed near the metal electrode–graphene spatially distributed hot carrier population, interfaces [254,223]. However, the small may offer a path to hot carrier–assisted effective area of the internal electric field thermoelectric technologies for efficient solar could decrease the detection efficiency energy harvesting. [214,262], as most of the generated electron– hole pairs would be out of the electric field, Recently, a novel hybrid graphene-quantum thus recombining, rather than being dot phototransistor with ultrahigh gain has separated. The internal photocurrent been demonstrated [265]. The proposed efficiencies (15−30% [256,257] and external device takes advantage of the strong light responsivities (generated electric current for a absorption in quantum dots and the two- given input optical power) of ~6.1 mA/W, dimensionality and high mobility [20] of much higher than the so far reported [254] graphene to merge these materials into a for GPDs are relatively low compared with hybrid system for photodetection with current photodetectors [250]. This is mainly extremely high sensitivity [265]. The due to limited optical absorption when only graphene-quantum dot phototransistor has one SLG is used, short photocarrier lifetimes shown ultrahigh gain of 108 and ten orders of and small effective areas (~200 nm [258]). magnitude larger responsivity with respect to pristine graphene photodetectors [265]. Future work will target the low light Moreover, the hybrid graphene-quantum dot absorption of graphene (2.3% of normal phototransistors exhibit spectral selectivity incident light [24,263]), the difficulty of from infrared to visible, gate-tunable extracting photoelectrons (only a small area sensitivity, and can be integrated with current of the p–n junction contributes to current circuit technologies [265]. generation); and the absence of a 52 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene Graphene was also demonstrated to be ideal confinement of graphene enables a new class for the enclosure within a planar λ/2 optical of functional devices as, for example, microcavity [266]. The latter is a photonic spectrally selective and highly directional light structure that confines optical fields between emitters, detectors, and modulators. two highly reflecting mirrors with a spacing of only one half wavelength of light [267]. The 8.6 Graphene saturable aborbers optical confinement provide a powerful means to control both the otherwise Materials with nonlinear optical and electro- featureless optical absorption [24] and the optical properties are needed in most spectrally broad thermal emission [268,269] photonic applications. Laser sources of graphene. The monolithic integration of a producing nano- to sub-picosecond pulses are graphene transistor with a planar optical key components in the portfolio of leading microcavity permitted the control on laser manufacturers. Regardless of photocurrent generation. Tuning the wavelength, the majority of ultrafast lasers excitation wavelength on resonance with the use a mode-locking technique, where a optical microcavity, an enhancement of ~20 in nonlinear optical element, called a saturable photocurrent was measured with respect to absorber, turns the continuous-wave output the out of resonance excitation [266]. In the into a train of ultrafast optical pulses [270]. same condition (spectral interval) a non- The key requirements are fast response time, confined graphene transistor has shown a strong nonlinearity, broad wavelength range, photocurrent increase of only 2. This low optical losses, high power handling, low demonstrates that the mocrocavity-controlled power consumption, low cost and ease of graphene transistor acts as light detector with integration into an optical system. Currently, spectral selectivity [266]. the dominant technology is based on semiconductor saturable absorber mirrors Moreover, electrically excited, thermal light (SESAM) [270,271]. However, these have a emission of graphene can be controlled by the narrow tuning range, and require complex spectral properties of the microcavity. Indeed, fabrication and packaging [270,272]. The the thermal emission spectrum of a linear dispersion of the Dirac electrons in microcavity-controlled graphene transistor graphene offers an ideal solution: for any displays a single, narrow peak at excitation there is always an electron–hole λcavity= 925nm having a fullwidth-at-half- pair in resonance. The ultrafast carrier maximum of 50 nm, providing a 140-fold dynamics [273,274] combined with large spectral narrowing with respect to the absorption [24,75] and Pauli blocking, make simulated free-space thermal spectrum at graphene an ideal ultrabroadband, fast T = 650 K [266]. saturable absorber [253,272]. Unlike SESAM and CNTs, there is no need for bandgap Moreover, the integrated graphene transistor engineering or chirality/diameter control electrical transport characteristic is [253,272]. profoundly modifies by the optical confinement of graphene by the optical Since its first demonstration in 2009 [272], microcavity [266]. The modifications of the the performance of ultrafast lasers mode- electrical transport can be related to the locked by graphene has improved significantly microcavity-induced enhancement or (Fig.24). For example, the average output inhibition of spontaneous emission of thermal power has increased from a few mW [272] to photons [266]. The concept of optical 1 W [275]. Some of the aforementioned nanoICT Strategic Research Agenda 53

Annex 1 nanoICT working groups position papers Graphene production strategies (e.g. LPE [101,253, [289]. GSAs have also been demonstrated to 272,276,277,278], CVD [279,280], carbon mode-lock solid-state lasers [284,285,290]. In segregation [281], mechanical exfoliation this case, CVD graphene (>1cm2) has been [282,283]) have been used for graphene directly transferred to quartz for solid-state saturable absorber (GSA) fabrication. So far, laser mode-locking [285]. Ref. [285] reported GSAs have been demonstrated for pulse 94fs pulses with 230mW output power. generation at 1µm [275,284], 1.2µm [285], Another approach for GSA fabrication relies in 1.5µm [253,272,279, 280,282,286] and 2µm spin-coating LPE graphene either on quartz [287]. The most common wavelength so far is substrates or high-reflectivity mirrors. GSA ~1.5μm, not due to GSAs wavelength can then be inserted into a solid-state cavity restriction, but because this is the standard for ultrafast pulse generation achieving wavelength of optical telecommunications. average power up to 1W using a solid-state Ref. [288] reported a widely tunable fiber Nd: YVO4 laser [275]. The output wavelength laser mode-locked with a GSA demonstrating is ~1 µm with power energy of ~14nJ. its “full-band” operation performance. For fiber lasers, the simplest and most 8.7 Optical limiters and frequency converters economical approach to GSA integration is sandwiching it between two fiber connectors Optical limiters are devices that have high (Fig.24) [253, 272, 279, 280, 282, 283, 288]. transmittance for low incident intensity and Other GSA integration approaches (e.g. low transmittance for high intensity [291]. evanescent-wave based integration [287]) There is a great interest in these for optical have been demonstrated for high-power sensors and human eye protection, as retinal generation. Sub-200fs pulses have been damage can occur when intensities exceed a achieved using a stretched-pulse design, certain threshold [291]. Passive optical limiters, where the cavity dispersion is balanced to which use a nonlinear optical material, have stretch the pulse for the limitation of the potential to be simple, compact and cheap nonlinear effects [276]. [291]. However, so far no passive optical limiters have been able to protect eyes and Figure 24. Graphene mode-locked fiber. WDM, other common sensors over the entire visible wavelength division multiplexer; PC, polarization and near-infrared range [291]. Typical materials controller; EDF, erbium-doped fiber; ISO, isolator [77]. include semiconductors (i.e. ZnSe, InSb), organic molecules (i.e. phthalocyanines), liquid Solid-state lasers are typically used for high- crystals and carbon-based materials (i.e. power output, as alternative to fiber lasers carbon-black dispersions, CNTs and fullerenes) [291,292]. In graphene-based optical limiters the absorbed light energy converts into heat, creating bubbles and microplasmas [292], which results in reduced transmission. Graphene dispersions can be used as wideband optical limiters covering visible and near- infrared. Broad optical limiting (at 532 and 1,064 nm) by LPE graphene was reported for nanosecond pulses [292]. It was also shown [293] that functionalized graphene dispersions could outperform C60 as an optical limiter. 54 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene Optical frequency converters are used to integration of memory and computation expand the wavelength accessibility of lasers functions on the same substrate. (for example, frequency doubling, parametric This section overviews the current status of amplification and oscillation, and four-wave graphene transistors as potential supplement mixing) [291]. Calculations suggest that to CMOS technology. nonlinear frequency generation in graphene (harmonics of input light, for example) should Figure 25. A futuristic graphene integrated circuit (not be possible for sufficiently high external electric to scale), wherein the desirable properties of various fields (>100 V cm–1) [294]. thicknesses of graphene layers are utilized along with strategic oxides (SiO2, ferroelectric, ferromagnetic, Second-harmonic generation from a 150fs multiferroic, etc.) in response to various external laser at 800nm has been reported for a stimuli, such as electric or magnetic fields. In the graphene film [295]. In addition, four-wave present illustration, the device structure is fabricated mixing to generate near-infrared wavelength from a very thin single-crystal graphite sheet after tunable light has been demonstrated using subsequent patterning/selective ablation. The SLG and FLG [296]. Graphene’s third-order remaining graphite acts as a good ohmic contact and susceptibility |χ3| was measured to be ~10−7 interconnection between the top Al metallization e.s.u. [296]—up to one order of magnitude (which also acts as a self-aligned mask, protecting the larger than CNTs [296]. Other features of underlying graphite) and the variable-thickness graphene, such as the possibility of tuning the graphene-based devices. Extract from [297]. nonlinearity by changing the number of layers, and wavelength-independent 9.1 Band gap opening in graphene nonlinear susceptibility [296] could be used Creating a band gap in graphene is one of the for various applications (optical imaging, for major challenges for employing graphene in example). conventional device circuits, both for analog and digital applications. For digital 9. Graphene transistors and applications a band gap is mandatory as electronics applications on/off ratios larger 104 are required [298]. For analog applications however, a band gap is Figure 25 shows an artist drawing of a future not required per-se, but the lack of drain graphene integrated circuit, in which, after initial wafer scale integration, further lithography, chemical treatments, would be used to engineer active and passive electronic, phononic or spintronics components. This illustrates graphene as a suitable platform to crosslink the nanoscale to conventional microelectronics. One can envision integrating on the same chip advanced functionalities including chemical sensing, nano-electromechanical resonators, thermal management, and electronic functions. Spintronics will also offer a co- nanoICT Strategic Research Agenda 55

Annex 1 nanoICT working groups position papers Graphene current saturation in graphene based FETs Figure 27. a) Drain current versus back gate voltage of and the resulting low voltage gain are major a graphene FET. The change of the sweep direction obstacles. A small band-gap~100meV would results in considerable hysteresis of 22 V. b) Mobility improve the situation. versus electric field in graphene FETs. Covering graphene with a gate insulator leads to mobility 9.2 Macroscopic graphene field effect reduction. Contacts have a considerable influence on transistors graphene FETs [299]. Universal mobility of silicon included as reference after Takagi [303]. The most straightforward device application of graphene may seem to be as a replacement The best current modulation reported to date channel material for silicon MOSFETs. Fig. 26a has been about 40, measured at room shows a schematic of such a graphene FET, temperature for water gated graphene [95]. including a top gate electrode, gate dielectric Furthermore, in conjunction with randomly and source and drain metal contacts. Fig. 26b distributed oxide charges the zero band gap shows a top view optical micrograph of a leads to a finite minimum charge density even macroscopic graphene FET on SiO2 [299]. The without any applied gate voltage [304]. fabrication of graphene FETs follows standard Consequently, macroscopic graphene Si process technology once the graphene is transistors conduct substantial current even deposited and identified. This includes the at their point of minimum conductance (also use of photo- or e-beam lithography, reactive referred to as Dirac point or charge neutrality ion etching and thin film deposition for the point), preventing their application as Si gate insulators and contacts. Details of typical MOSFET replacement. Fig. 27a further shows fabrication processes are described in Refs. hysteresis as the gate voltage is swept from [300,301,302]. negative to positive and vice versa. This typical behaviour occurs despite measuring in Figure 26. a) Schematic cross section and b) optical vacuum conditions (P=5x10-3 mbar) and is a top-view micrograph of a graphene field effect strong indicator of charge traps near the transistor [299]. graphene/insulator interface. While suspended graphene measured in ultra high The transfer characteristics (here: drain vacuum has been shown to have mobilities current Id vs. back gate voltage Vbg) of a exceeding 1000000 cm2/Vs [20], realistic typical graphene transistor is shown in Fig. 27. graphene FETs are limited by substrates and It reveals a major drawback of macroscopic top gates. graphene MOSFETs: the absence of a band gap severely limits the current modulation in Nonetheless, the mobilities in top gated the graphene FET and, in addition, leads to devices exceed those of Si and are typically on ambipolar behaviour. the order of several hundred to a 56 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene thousand cm2/Vs, even though graphene/ results for any form of GNRs should be insulator interfaces have not at all been regarded with care, as they typically share an optimized yet [299,300,301,302,305,306]. Fig. optimistic assumption of well controlled 27b shows electron and hole mobilities termination of dangling bonds. In reality, extracted from several top gated devices, there is likely a great variety of chemical both in 2-point and by 4-point probe groups terminating the GNR edges. A first configuration. detailed discussion has been recently published to address these issues [320], but it 9.3 Graphene nanoribbon transistors is probably reasonable to consider the nature of “real life” zig-zag GNRs still an open A potential method to create a band gap in question. The predicted presence of a band graphene is to cut it into narrow ribbons of gap in specific GNRs has been experimentally less than a few tens of nanometers. confirmed. GNRs can be armchair or zig-zag edge Figure 28. a) Simulated DOS vs. energy of hydrogen terminated. In armchair GNRs, the transition terminated armchair GNRs for various numbers of from 2D graphene to 1D GNRs leads to electrons N across the ribbon [14]. b) Experimental quantum confinement and a bandgap that is Ion/Ioff ratios versus GNR width taken from literature. roughly inversely proportional to the GNR None of the GNRs thus far has shown metallic width (Eg ~ 1/W) [307,308]. The precise value behavior. The lower colored part of the graph of the band gap is further predicted to indicates Ion/Ioff values of typical macroscopic depend on the number N of carbon atoms graphene FETs; From [14]. across the ribbon [307,308,309,310]. This is demonstrated in Fig. 28a, where the simulated density of states (DOS) versus energy for three different hydrogen- terminated armchair GNRs with N = 11, 12 and 13 is shown [14,311]. While the GNR with N = 11 is semi-metallic, the ribbons with 12 and 13 atoms are semiconducting (generally armchair ribbons are semi-metallic at N = 3m− 1, where m is an integer [312]). In hydrogen-terminated zig-zag GNRs, however, the situation is more complicated. Ref. [307] predicted that localized edge states near the Fermi level lead to semi-metallic behaviour, regardless of the number of carbon atoms. On the other hand, Ref. 309 calculated ab initio that edge magnetization causes a staggered sublattice potential on the graphene lattice that induces a band gap. Finally, GNRs with other chiral orientation have been considered, including a mix of edges along a ribbon, adding to the complexity of this option [313,314,315, 316,317,318,319]. In summary, the simulated nanoICT Strategic Research Agenda 57

Annex 1 nanoICT working groups position papers Graphene First evidence was reported by Refs. control as well as edge state definition remain [321,322], where GNRs were structured by e- tremendous challenges. To this end, a beam lithography and etched in oxygen recently developed technique, helium ion plasma with minimum widths~20 nm. The beam microscopy, has been shown to have band gaps of these GNRs where in the range potential for precise nano-patterning of ~30 meV and resulted in field effect [326,327]. transistors with Ion/Ioff ratios of about 3 orders of magnitude at low temperatures (1.7–4 K), 9.4 Bilayer graphene reduced to~ 10 at room temperature. These investigations support theoretical predictions Another viable approach to obtain a band gap that sub-10 nm GNRs are required for true field in graphene is to break its symmetry. Ref. effect transistor action at room temperature. [328] proposed that macroscopic BLG would More importantly, the experiments revealed a display a suitable band gap if a transverse band gap regardless of the chiral orientation of electric field was applied to break the layer the GNRs [321]. This latter result was symmetry. Ref. [328] predicts a roughly linear attributed to a strong influence of edge states dependence of the gap on applied [319,321,323], which dominate over the displacement field, with each 0.1V/nm chirality of the band structure. To date, two adding~10 meV and a subsequent saturation examples of sub-10 nm GNRs have been at about 300meV. shown experimentally. Ref. 324 fabricated GNRs with a minimum width~1 nm and a This prediction was experimentally confirmed band gap of about 500meV using e-beam by Ref. 329 for BLG on SiC through angle- lithography and repeated, careful over- resolved photoemission spectroscopy etching. The resulting transistors (ARPES). In their work, they used potassium consequently switched off at room doping to modify the carrier density, which temperature to “no measurable conductance” lead to changes in the electronic band gap [324]. An alternative fabrication process for [329]. The size of the created band gap was GNRs was presented by Ref. [109]. Here, also confirmed by means of infrared GNRs were solution derived from graphite by spectroscopy [330,331].Ref. [332] took this thermal exfoliation, ultrasonication and approach a step further by applying an ultracentrifugation. The resulting dispersions electrostatic field through a double gate were deposited onto substrates and GNRs configuration with individually controlled back were identified with AFM. The devices and top-gate electrodes. However the band exhibited well behaved transistor action at gap in their devices could not be utilized to room temperature with Ion/Ioff>106 [109,325]. increase the on/off ratio at room temperature due to the presence of intergap states [332]. The experimental Ion/Ioff ratios reported to Only at low temperatures they observed an date are summarized in Fig. 28b. While they insulating behaviour in BLG with a clearly support theoretical predictions and perpendicular applied electric displacement show promise for GNR electronics, they also field. By optimizing the process technology Refs show an urgent need for further research in [258,333] reduced the intergap states in BLG- this field: Statistical data is scarce and the FETs significantly so that they were able to discrepancies between theory and increase the on/off ratio up~100 at room experiment need to be addressed. The temperature by applying an electrical necessity of controllable sub-10nm feature displacement field. This is comparable to the sizes and great uncertainties in chirality on/off ratio in small band-gap III/V transistors 58 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene and sufficient for most analog applications. The Using BLG with a perpendicular electric field channel resistance of a double-gated BLG FET is Ref.335 realized logic gates. By combining a p- depicted in Fig. 29 [333]. type and an n-type BLG FET with an induced band gap, Ref. 335 realized NAND and NOR gates, fundamental building blocks for digital logic circuits. The operation of NOR gates and the resulting operation of the BLG based NOR gate are depicted in Fig. 30 Figure 29. Top:SEM image (false color) of a double- Figure. 30. Top: Device wiring used for realizing a NOR gated bilayer graphene FET [333]. Bottom:Channel gate; corresponding circuit diagram and the truth Resistance of a double-gated bilayer FET as a function table of the operation (taken from [194]). Bottom: The of the applied top-gate voltage for different back-gate operation at a drive voltage 0.7 V and 1.0 V, voltages [333]. respectively (taken from [335]). In this device the resistance modulation is 8 at 9.5 Summary band gap engineering in zero applied back-gate and increases to 80 at graphene an applied back-gate of 40V. However, using a double gate structure, with two individually Over the past few years different approaches controlled gates, increases complexity, which is were explored for creating a band gap in not desired in applications. Recently it has been graphene, whereof confinement to GNRs and shown that the static gate can be replaced by BLG are the most promising. Using graphene adsorbate doping [334]. In this work a band in switches for digital operation requires a gap could be obtained and utilized to increase band gap>400meV, hence confining graphene the on/off ratio with only one gate electrode to GNRs is necessary. Using bottom-up [334]. synthesis on/off ratios larger 106 are achievable using GNRs [325]. nanoICT Strategic Research Agenda 59

Annex 1 nanoICT working groups position papers Graphene The challenges is to develop industrially compatible top- down methods for the fabrication of such small structures. In contrast, in BLGs a band gap could be realized using standard top- down fabrication, but as the theoretical limits of the band gap in BLG are~300 meV an application to digital circuits will be challenging. However analog applications can strongly benefit from this band gap, which is tuneable by electric means. 9.6 RF Transistors Figure 31. Summary of published graphene RF transistor data (Courtesy of Yu Min Lin, IBM, USA). Numerous communication systems rely on electromechanical devices, such as filters, both using exfoliated [306,338] and epitaxial resonators, and RF switches. Their materials [339,340]. Graphene transistors miniaturization will strongly affect the with a 240nm gate operating at frequencies development of future communication up to 100 GHz were demonstrated in early systems. The ultimate limit to this 2010 [341]. This cut-off frequency is already miniaturization is represented by graphene higher than those achieved with the best electromechanical devices. silicon MOSFET having similar gate lengths [342]. Graphene combines exceptional electronic properties with excellent mechanical Higher fT can be expected for optimized devices properties. Its ambipolar transport properties, with shorter gate lengths as demonstrated by ultra thin and flexible, and electrostatic the progress achieved by the IBM group that in doping offer a new degree of freedom for the 2 years improved the performances of their development of advanced electronic devices Graphene-based RF transistor passing from with many potential applications in 26 GHz[306] to 155 GHz[343] (Fig. 31). The communications and RF electronics. latter result was achieved with CVD graphene was grown on copper film and transferred to a For RF analog applications a high on-off ratio wafer of diamond-like carbon. This once more is desirable but not mandatory. Instead, most emphasizes the need for graphene/insulator relevant for good RF performance is a FET interface engineering. channel with excellent carrier transport properties (high mobility and maximum Cut-off frequency over 300 GHz were velocity) [336], combined with a small scale demonstrated with a 140 nm channel length length, which improves strongly as the [344], comparable with the very top high- channel material thickness is reduced [337]. electron-mobility transistors (HEMTs) with Graphene RF FETs have been investigated, similar gate length [345], see Fig.32. These 60 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene results are impressive, considering that RF and fabrication technology of these new graphene research has been done for a fraction devices matures, their integration with of the time devoted thus far to conventional conventional Si electronics, and/or flexible devices. As shown in Fig. 32, the progress in and transparent substrates has the potential graphene-based high frequency devices has to transform communications. Thus graphene been impressive. This progress opens new can be seen, rather than as a replacement of horizons for quick development of plethora of Si technology, as a complement to it in a applications in RF communications. “More than Moore” [346] perspective. Figure 32. Comparing cut-off frequency versus gate 9.7 Non-conventional graphene switches length for graphene MOSFETs, nanotube FETs and three types of radiofrequency FET (Adapted from A number of concepts for (non-volatile) ref. 321). graphene switches have emerged based on mechanisms other than the classic field effect. Even though a thorough review is beyond the scope of this leaflet, they are briefly introduced in this section. A first concept are graphene/GO Schottky barrier MOSFETs [347], where semiconducting GO acts as the transistor channel. Another approach relies on creating nanoscale gaps by electric fields [348]. These are reversibly opened and closed by breaking and re-forming the carbon atomic chains [348]. Compared to conventional Si and III-V Figure 33. Possible applications of graphene RF materials, ambipolar graphene electronics has transistor (Courtesy of Tomas Palacios, MIT, USA). many advantages. The higher mobility allows higher operating frequencies for frequency Chemical surface modification affects strongly doubling and mixing. In addition, ambipolar the electronic band structure of graphene devices can significantly reduce the number of [349]. The electrostatical control of drain transistors needed. Simpler circuits mean less current can be reversibly modified in a power consumption and smaller chip area. Graphene is also an ideal material for flexible electronics integration. To take advantage of the full potential of graphene devices, basic research needs to be combined with improved material growth and device technology. A better understanding of parameters such as breakdown voltage, electron velocity, and saturation current is needed to allow a complete benchmark and evaluation of this material. Once the growth nanoICT Strategic Research Agenda 61

Annex 1 nanoICT working groups position papers Graphene graphene FET by controlled chemisorption current in the off-state is too large for digital [350]. Ferroelectric gating has been shown to applications, when applying drain-to-source electrostatically dope graphene and change the voltages complying with the International drain currents in a non-volatile way [351]. Technology Roadmap for Semiconductors While these early concepts are far from (ITRS) [298], i.e. approximately 0.5 V. mature, they nevertheless demonstrate the potential of graphene for nanoelectronics The large tunnelling component, instead of applications that might not be anticipated being detrimental like in thermionic devices, today. can be turned into an advantage in tunnel FETs [353], whose main appeal is represented by a 9.8 Atomistic simulations of transistors sub-threshold swing well below those achievable in thermionic devices (<60mV/dec) Due to the novelty of the graphene research [354]. This would allow reducing the supply field, device design principles have not yet voltage for digital logic applications, and been extensively elaborated, and different consequently the power consumption. It is design options must be explored, evaluated indeed well known that power dissipation is and optimized. From this perspective, nowadays the most limiting factor in integrated numerical simulations can greatly help. circuits, so that all future approaches need be directed towards the design of low power When dealing with the investigation of device devices, the so-called “Green Transistors”[355]. electrical behaviour, atomistic features of the From this perspective, BLG tunnel FET (see material, such as for example the lattice Fig.34) represents a viable option, because of potential, disappear behind a series of its large Ion/Ioff ratios (large than 103), even synthetic parameters like the effective mass when applying ultra-low power supply (0.1 V) tensor. The price to pay is however the lack of [353]. An alternative FET, based on graphene accuracy at the atomic level, which can be heterostructures with atomically thin boron obtained through ab-initio simulations, at the nitride acting as a tunnel barrier, was recently expense of huge computational requirements, demonstrated [356]. especially when the number of atoms is of the order of few thousands like in realistic FETs. The strategy is then to find a trade-off, leveraging on atomistic simulations based on the Non-Equilibrium Green’s Function (NEGF) formalism, which can provide physical insight at the atomic level, but with fewer computational burden as compared to DFT calculations. Efforts towards the understanding of BLG for Figure 34. Sketch of the bilayer graphene tunnel FET next-generation devices have been first proposed in [353]. addressed in [352], where a double gate FET was simulated by means of the self-consistent solution of the 2D Poisson and Schrödinger equation with open-boundary conditions, within NEGF. Strong band-to-band tunnelling heavily limits device performance, since the 62 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene temperature switching ratios of ~50. However, this value can be improved by optimizing the device structure and up to 105 was recently reported by using MoS2 instead of BN [357]. Lateral confinement, as in GNRs, is another approach widely addressed in the last year to induce an energy gap. Simulations have provided relevant information regarding the potential and the limits of this approach [358,359]. In particular, GNRs smaller than 3 nm are needed to comply with ITRS, which is very difficult to achieve with state-of-the-art electron beam lithography [360]. Things get even worse when considering sources of non- idealities, such as line-edge roughness or single vacancy defects, which strongly limit electron mobility in GNRs, another important figure of merit for digital applications [361]. However, even when dealing with perfect edges, mobility in narrow devices (W<2nm) is strongly limited by electron-phonon coupling, leading to similar mobility as the Si counterpart [362]. Figure 35. (a) Schematic structure of the graphene Chemical functionalization is a viable route heterostructure FET. The corresponding band toward bandgap engineering. Ref. [363] structure with (b) no gate voltage and (c) finite gate exposed graphene to a stream of hydrogen voltage Vg and zero bias Vb applied. (d) Band atoms leading to a material close to the ideal structure when both Vg and Vb are finite. The cones graphane. DFT suggests that [364] graphane illustrate graphene’s Dirac-like spectrum and, for would have a 5.4 eV gap, while it reduces to simplicity, the tunnel barrier for electrons is 3.2 eV in 50% hydrogenated graphene. A considered [adapted from ref. 322]. recent study (Fig. 36) [365] suggested that such devices can provide large currents as The operation of the device (see Fig. 35) relies well as Ion/Ioff ratios, and can represent a on the voltage tunability of the tunnelling promising option for future technology nodes. density of states and of the effective height of the tunnel barrier adjacent to the graphene Probably more than towards digital applications, electrode [356]. The device exhibits room research should be focus on analog electronics, where the lack of a bandgap does not represent an issue, and high performance in terms of cutoff frequency in RF applications can be achieved [366], and negative differential resistance [367,368] can be exploited in a range of applications, such as oscillators, fast switching logic, and low power amplifiers. nanoICT Strategic Research Agenda 63

Annex 1 nanoICT working groups position papers Graphene Figure 36. Sketch of the functionalized graphene exploiting the uniqueness of graphene and based transistor related inorganic 2d crystals, rather than just try to displace existing and established 10. Conclusion technologies). Graphene has already demonstrated high Graphene may allow high-speed, compact- potential to impact most areas of electronic footprint electro-optical modulators, switches information technology, ranging from top end and photodetectors integrated with high performance applications in ultrafast waveguides or plasmonic circuits. Moreover, information processing, to consumer the mechanical flexibility of graphene will also applications using transparent or flexible enable integration with bendable substrates electronic structures. This is testified by the and plastic waveguides. increasing number of chip-makers now active in graphene research. Graphene can detect light beyond the current limit set by the band gap of traditional The combination of its unique optical and semiconductors, opening new applications in electronic properties, in addition to flexibility, the far-infrared (terahertz, THz) and mid- robustness and environmental stability, make infrared regimes (e.g. bolometers and graphene an extremely interesting material cameras), and provides the potential for for future photonic and optoelectronic ultrafast pixelated detection with ballistic devices. transport of generated n. This could enable portable terahertz sensors for remote Graphene can successfully replace many detection of dangerous agents, materials (i.e. ITO) in several existing environmental monitoring or wireless applications, but certainly, the combination of communication links with transmission rates its unique properties will inspire completely above 100 Gbit/s. Moreover, it can be new applications and is where future research exploited for photocurrent generation by should focus (i.e. creating new technologies providing a gain mechanism where multiple carriers are created from one incident photon. The compatibility of graphene with standard CMOS processes at wafer scale makes it a promising candidate for future electronics, particularly for high data-rate (inter- and intra-chip) interconnects. While macroscopic graphene transistors are not suitable for logic applications, due to the lack of band gap, graphene RF transistors seem promising and feasible. Graphene nanoribbons, on the other hand, show promise as a CMOS compatible approach, but their extreme sensitivity at an atomic level to both geometric and edge termination variations may well render their practical use extremely challenging. Non- classic switching mechanisms may eventually 64 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene lead to a co-integration of graphene into [5] Lin, Y-M et al., Wafer-Scale Graphene Integrated silicon technology, even though details are Circuit Science 332, 1294 (2011). not yet clear today. In addition to these device related issues, there is the imperative [6] Torrisi, F., et al., Ink-Jet Printed Graphene need of a large area, CMOS compatible Electronics. arXiv:1111.4970v1 (2011). deposition or placement technique. As CVD and related methods are being explored, [7] Hernandez, Y., et al. High-yield production of many device related questions are being graphene by liquid-phase exfoliation of graphite. addressed using the existing manufacturing Nature Nanotech. 3, 563 (2008). methods. These insights will be transferable as soon as industrially relevant production [8] Tannock, Q., Exploiting carbon flatland, Nat. and placement technologies will become Mater. 11, 2 (2012). available. [9] Harrison, W.A., ed. Electronic Structure and Acknowledgements Properties of Solids: the Physics of the Chemical Bond. 1989, Dover Publications. CPE thanks COST MP0901 “NanoTP” for support, as well as ANR NANOSIM-GRAPHENE. [10] Charlier, J.C., et al. Electron and phonon JC acknowledges financial support from the properties of graphene: Their relationship with Alexander von Humboldt Foundation, the carbon nanotubes Topics. Appl. Physics 111, 673 French ANR (NMGEM project) and the EC (2008). (GRENADA project). A. C. F. acknowledges funding from EU grants GRAPHENE-CA, [11] Wallace, P. R., The Band Theory of Graphite Phys. NANOPOTS, GENIUS, RODIN, EPSRC grants Rev. 71, 622 (1947). EP/GO30480/1 and EP/F00897X/1, and a Royal Society Wofson Research Merit Award. [12] Geim, A.K. and K.S. Novoselov, The rise of FB and ACF acknowledge Paul Beecher, graphene. Nature Mater.6, 183 (2007). Tawfique Hasan, Zoran Radivojevic and Zhipei Sun for useful discussion. [13] Ando, T., The electronic properties of graphene and carbon nanotubes. NPG Asia Mater. 1, 17 (2009). 11. References [14] Lemme M., Current status in graphene [1] Novoselov, K. S., Castro Neto, A. New directions in transistors. Solid State Phenom., 499 156 (2010). science and technology: two-dimensional crystals Rep. Prog. Phys. 74 082501 (2011) [15] Novoselov, K.S., et al., Two-dimensional gas of [2]http://www.nobelprize.org/nobel_prizes/physics/l massless Dirac fermions in graphene. Nature 438, 197 aureates/ 2010/press.html. (2005). [3] Bae, S. et al. Roll-to-roll production of 30-inch graphene films for transparent electrodes Nature [16] Berger, C., et al., Electronic confinement and Nanotech. 5, 1 (2010). coherence in patterned epitaxial graphene. Science, [4] http://research.nokia.com/morph 312 1191 (2006). [17] Bolotin, K.I., et al., Ultrahigh electron mobility in suspended graphene. Solid State Communications 146, 351 (2008). [18] Morozov, S.V., et al., Giant Intrinsic Carrier Mobilities in Graphene and Its Bilayer. Phys. Rev. Lett., 100, 016602 (2008). [19] Du, X., et al., Approaching ballistic transport in suspended graphene. Nature Nanotech. 3, 491 (2008). [20] Castro, E. V., et al., Limits on Charge Carrier Mobility in Suspended Graphene due to Flexural Phonons. Phys. Rev. Lett. 105, 266601 (2010). [21] Moser, J., Barreiro, A., Bachtold, A. Current- induced cleaning of graphene. Appl. Phys. Lett. 91, 163513 (2007). nanoICT Strategic Research Agenda 65

Annex 1 nanoICT working groups position papers Graphene [22] Balandin, A.A., et al., Superior Thermal [37] Matsuo, Y., et al. Silylation of graphite oxide, Conductivity of Single-Layer Graphene. Nano Lett. 8, Carbon, 42, 2117 (2004). 902 (2008). [38] Englert, J.M., et al. Covalent bulk [23] Ghosh, S., et al., Extremely high thermal functionalization of graphene Nature Chem. 3, 279 conductivity of graphene: Prospects for thermal (2011). management applications in nanoelectronic circuits. Appl. Phys. Lett. 92, 151911 (2008). [39] Backes, C., et al., Nanotube surfactant design: the versatility of water-soluble perylene bisimides. Adv. [24] Nair, R. R., et al. Fine structure constant defi nes Mater., 22, 788 (2010). transparency of graphene. Science 320, 1308 (2008). [40] Wang, Q.H. and Hersam, M.C. Room- [25] Lee, C., et al. Measurement of the Elastic temperature molecular-resolution characterization of Properties and Intrinsic Strength of Monolayer self-assembled organic monolayers on epitaxial Graphene Science 321, 385, (2008). graphene. Nature Chem. 1, 206 (2009). [26] Booth, T.J., et al., Macroscopic Graphene [41] Mattevi, C. et al. Evolution of electrical, chemical, Membranes and Their Extraordinary Stiffness. Nano and structural properties of transparent and Letters 8, 2442 (2008). conducting chemically derived graphene thin films. Adv. Funct. Mater. 19, 2577 (2009). [27] Baughman, R. H., Zakhidov, A. A., de Heer, W. A. Carbon Nanotubes—the Route Toward Applications [42] Erickson, K., R. et al., Determination of the local Science 297, 787 (2002). chemical structure of graphene oxide and reduced graphene oxide Adv. Mater, 22, 4467 (2010). [28] Cai, J., P. et al., Atomically precise bottom-up fabrication of graphene nanoribbons Nature 466, 470 [43] Eda, G., Chhowalla, M., Chemically Derived (2010). Graphene Oxide: Towards Large-Area Thin-Film Electronics and Optoelectronics. Adv. Mater. 22, 2392 [29] Brodie, B.C., Sur le poids atomique du graphite (2010). Ann. Chim. Phys., 55, 466 (1860). [44] Becerril, H.A., et al., Evaluation of solution- [30] Hummers, W.S. and R.E. Offeman, Preparation of processed reduced graphene oxide films as graphite oxide J. Am. Chem. Soc. 80, 1339 (1958). transparent conductors. ACS Nano, 2, 463 (2008). [31] Staudenmaier, L., Ber. Deut. Chem. Ges., 31, [45] Stankovich, S., et al., Synthesis of graphene- 1481 (1898). based nanosheets via chemical reduction of exfoliated graphite oxide. Carbon, 2007, 45, 1558- [32] Titelman, G. I., et al. Characteristics and 1565. microstructure of aqueous colloidal dispersions of graphite oxide Carbon 43, 641 (2005). [46] Gilje, S., et al., A chemical route to graphene for device applications. Nano Lett., 7, 3394 (2007). [33] Liscio, A., et al., Charge transport in graphene– polythiophene blends as studied by Kelvin Probe [47] Zhou, M., et al., Electrochemically Reduced Force Microscopy and transistor characterization J. Graphene Oxide Films. Chemistry-a European Journal, Mater. Chem. 21, 2924 (2011). 15, 6116 (2009). [34] Treossi, E., M. et al., High-contrast visualization of [48] Yao, P.P., et al., Electric Current Induced graphene oxide on dye-sensitized glass, quartz and Reduction of Graphene Oxide and Its Application as silicon by fluorescence quenching J. Am. Chem. Soc., Gap Electrodes in Organic Photoswitching Devices. 131, 15576 (2009). Adv. Mater. 22, 5008 (2010). [35] Quintana, M., A., et al., Selective organic [49] Bagri, A., et al., Structural evolution during the functionalization of graphene bulk or graphene edges reduction of chemically derived graphene oxide, Chem. Comm. 47, 9330 (2011). Nature Chem. 2, 581 (2010). [36] Melucci, M., E. et al., Facile covalent [50] Pang, S.P., et al. Patterned Graphene Electrodes functionalization of graphene oxide using from Solution-Processed Graphite Oxide Films for microwaves: bottom-up development of functional Organic Field-Effect Transistors Adv. Mater. 21, 3488 graphitic materials J. Mater. Chem., 20, 9052 (2010). (2009). 66 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene [51] Eda, G. and Chhowalla, M. Graphene-based Stat. Solidi B-Basic Solid State Physics 245, 2064 Composite Thin Films for Electronics. Nano Letters, 9, (2008). 814 (2009). [66 Zhang, Y.H., et al. Tuning the electronic structure [52] Wu, J.B., et al., Organic solar cells with solution- and transport properties of graphene by noncovalent processed graphene transparent electrodes. Appl. functionalization: effects of organic donor, acceptor Phys. Lett. 92, 263302 (2008). and metal atoms. Nanotechnology 21, (2010). [53] Yang, N.L., et al., TiO2 Nanocrystals Grown on [67] Matte, H., et al., Quenching of fluorescence of Graphene as Advanced Photocatalytic Hybrid aromatic molecules by graphene due to electron Materials. ACS Nano, 4, 887 (2010). transfer. Chem. Phys. Lett., 506, 260 (2011). [54] Zhu, X.J., et al., Nanostructured reduced [68] Swathi, R. S. and Sebastiana K. L. Resonance graphene oxide/Fe2O3 composite as a high- energy transfer from a dye molecule to graphene J. performance anode material for lithium ion batteries. Chem. Phys. 129, 054703 (2008). ACS Nano, 5, 3333 (2011). [69] Treossi, E., et al., High-Contrast Visualization of [55] Wei, Z.Q., et al., Nanoscale Tunable Reduction of Graphene Oxide on Dye-Sensitized Glass, Quartz, and Graphene Oxide for Graphene Electronics. Science, Silicon by Fluorescence Quenching. J. Am. Chem. Soc. 328, 1373 (2010). 131, 15576 (2009). [56 Alaboson, J.M.P., et al., Conductive Atomic Force [70] Wang, Y.B., et al., Fluorescence Quenching in Microscope Nanopatterning of Epitaxial Graphene on Conjugated Polymers Blended with Reduced Graphitic SiC(0001) in Ambient Conditions. Adv. Mater., 23, Oxide. J. Phys. Chem. C, , 114, 4153 (2010). 2181 (2011). [71 Kim, J., et al. Graphene Oxide Sheets at Interfaces. [57] Mativetsky, J.M., et al., Local Current Mapping J. Am. Chem. Soc. 132, 260 (2010). and Patterning of Reduced Graphene Oxide. J. Am. Chem. Soc., 132, 14130 (2010). [72] Loh, K.P., et al., Graphene oxide as a chemically tunable platform for optical applications. Nature [58] He, S. J., et al., A Graphene Nanoprobe for Rapid, Chem. 2, 1015 (2010). Sensitive, and Multicolor Fluorescent DNA Analysis. Adv. Funct. Mater. 20, 453 (2010). [73] Rao, C.N.R. and Voggu, R. Charge-transfer with graphene and nanotubes. Materials Today 13, 34 [59] Merchant, C. A., et al., DNA Translocation (2010). through Graphene Nanopores. Nano Lett. 10, 2915 (2010). [74] Novoselov, K.S., et al., Electric Field Effect in Atomically Thin Carbon Films. Science 306, 666 [60] Schneider, G.F., et al., DNA Translocation through (2004). Graphene Nanopores, Nano Lett. 10, 3163 (2010). [75] Casiraghi, C., et al., Rayleigh Imaging of Graphene [61] Rochefort, A. and Wuest, J.D. Interaction of and Graphene Layers. Nano Lett. 7, 2711 (2007). Substituted Aromatic Compounds with Graphene Langmuir, 2009, 25, 210-215. [76] Blake, P., et al., Making graphene visible. Appl. Phys. Lett. 91, 063124 (2007.). [62] Roos, M., et al., Hierarchical Interactions and Their Influence upon the Adsorption of Organic [77] Abergel, D.S.L., A. Russell, and V.I. Fal'ko, Visibility Molecules on a Graphene Film. J. Am. Chem. Soc. 133, of graphene flakes on a dielectric substrate. Appl. 9208 (2011). Phys. Lett. 91, 063125 (2007). [63 Gierz, I., et al. Atomic Hole Doping of Graphene. [78] Ni, Z.H., et al., Graphene Thickness Nano Lett. 8, 4603 (2008). Determination Using Reflection and Contrast Spectroscopy. Nano Lett. 7, 2758 (2007). [64] Wehling, T.O., et al., Molecular doping of graphene. Nano Lett. 8, 173 (2008). [79] Ferrari, A.C., et al., Raman Spectrum of Graphene and Graphene Layers. Phys. Rev.Lett. 97, 187401 [65] Lauffer, P., et al., Molecular and electronic (2006). structure of PTCDA on bilayer graphene on SiC(0001) studied with scanning tunneling microscopy. Phys. [80] Ferrari, A.C., Raman spectroscopy of graphene and graphite: Disorder, electron-phonon coupling, nanoICT Strategic Research Agenda 67

Annex 1 nanoICT working groups position papers Graphene doping and nonadiabatic effects. Solid State Comm. [95] Das, A.,et al., Monitoring dopants by Raman 143, 47 (2007). scattering in an electrochemically top-gated graphene transistor. Nature Nanotech. 3, 210 (2008). [81] Tuinstra, F. and Koenig, J. L. Raman Spectrum of Graphite J. Chem. Phys. 53, 1126 (1970). [96] Casiraghi, C,. et al., Raman spectroscopy of graphene edges. Nano Lett. 9, 1433 (2009). [82] Ferrari, A.C. and Robertson, J., Raman of amorphous, nanostructured, diamond-like carbon [97] Ferrari A. C. et al., Resonant Raman spectroscopy and nano-diamond. Philos. Trans. R. Soc. Ser. A 362, of disordered, amorphous, and diamondlike carbon. 2477 (2004). Phys. Rev. B 64, 075414 (2001). [83] J. Yan, et al., Electric Field Effect Tuning of [98] Lotya, M. et al. Liquid phase production of Electron-Phonon Coupling in Graphene. Phys. Rev. graphene by exfoliation of graphite in Lett. 98, 166802 (2007). surfactant/water solutions. J. Am. Chem. Soc. 131, 3611 [84] Ando, T. , Magnetic Oscillation of Optical Phonon in Graphene J. Phys. Soc. Jpn. 76, 024712 (2007). [99] Green, A. A. & Hersam, M. C. Solution phase production of graphene with controlled thickness via [85] Goerbig, M.O. et al., Filling-Factor-Dependent density diff erentiation. Nano Lett. 9, 4031 (2009). Magnetophonon Resonance in Graphene Phys. Rev. Lett. 99, 087402 (2007). [100] Maragò, O. M., et al., Brownian motion of graphene. ACS Nano 4, 7515 (2010). [86] Basko, D.M., et al., Electron-electron interactions and doping dependence of the two-phonon Raman [101] Hasan, T., et al. Solution-phase exfoliation of intensity in graphene. Phys. Rev. B 80, 165413 (2009). graphite for ultrafast photonics. Phys. Status Solidi B 247, 2953 (2010) [87] Das, A., et al., Monitoring dopants by Raman scattering in an electrochemically top-gated graphene [102] Khan, U., et al. High-Concentration Solvent transistor. Nature Nanotechnology 3, 210 - 215 Exfoliation of Graphene.Small 6, 864 (2010). (2008). [103] Nuvoli, D. et al. High concentration few-layer [88] Malard, L. M. et al., Probing the electronic graphene sheets obtained by liquid phase exfoliation structure of bilayer graphene by Raman scattering. of graphite in ionic liquid. J.Mater. Chem. 21, 3428 Phys. Rev. B 76, 201401 (2007). (2011). [89] Pisana, S. et al., Breakdown of the adiabatic [104] Bonaccorso F., et al., Graphene photonics and Born–Oppenheimer approximation in graphene. optoelectronics. Nature Photon. 4, 611 (2010). Nature Mater. 6, 198 (2007). [105] Lee, J. H., et al., One-Step Exfoliation Synthesis [90] Yan, J., et al., Observation of Magnetophonon of Easily Soluble Graphite and Transparent Resonance of Dirac Fermions in Graphite. Phys. Rev. Conducting Graphene Sheets. Adv. Mater. 21, 4383 Lett. 105, 227401 (2010). (2009). [91] Faugeras, C., et al., Magneto-Raman Scattering of [106] Valles, C., et al., Solutions of negatively charged Graphene on Graphite: Electronic and Phonon graphene sheets and ribbons. J. Am. Chem. Soc. 130, Excitations. Phys. Rev. Lett. 107, 036807(2011). 15802 (2008). [92] Mohiuddin, M. G., et al., Uniaxial strain in [107] Li, D. et al. Processable aqueous dispersions of graphene by Raman spectroscopy: G peak splitting, graphene nanosheets. Nature Nanotech. 3, 101 Grüneisen parameters, and sample orientation. Phys. (2008). Rev. B 79, 205433 (2009). [108] An, X. et al. Stable Aqueous Dispersions of [93] Ferralis,N. Maboudian, R. and Carraro, C., Noncovalently Functionalized Graphene from Evidence of Structural Strain in Epitaxial Graphene Graphite and their Multifunctional High-Performance Layers on 6H-SiC(0001). Phys. Rev. Lett. 101, 156801 Applications. Nano Lett. 10, 4295 (2010). (2008). [94] Casiraghi, C., et al., Raman Fingerprint of Charged Impurities in Graphene. Appl. Phys. Lett. 91, 233108 (2007). 68 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene [109] Li, X., et., Chemically Derived, Ultrasmooth [124] Coraux, J., et al., Structural Coherency of Graphene Nanoribbon Semiconductors. Science 319, Graphene on Ir(111). Nano Lett. 8, 565 (2008). 1229 (2008). [125] N'Diaye, A.T., et al., Structure of epitaxial [110] Coleman, J. N., et al. , Two-Dimensional graphene on Ir(111). New Journal of Physics 10, Nanosheets Produced by Liquid Exfoliation of Layered 043033 (2008). Materials. Science 331, 568 (2011). [126] Li, X., et al., Large-Area Synthesis of High-Quality [111] Acheson, E. G., Production of artifi cial and Uniform Graphene Films on Copper Foils. Science crystalline carbonaceous materials; article of 324, 1312 (2009). carborundum and process of the manufacture thereof carborundum. US patent 615,648 (1896). [127] Rummeli, M. H., et al. Direct Low-Temperature Nanographene CVD Synthesis over a Dielectric [112] Berger, C., et al., Ultrathin epitaxial graphite: 2D Insulator. ACS nano 4, 4206 (2010). electron gas properties and a route toward graphene- based nanoelectronics. J. Phys. Chem. B 108, 19912 [128] Hwang, J., et al., Epitaxial growth of graphitic (2004). carbon on C-face SiC and Sapphire by chemical vapor deposition (CVD). J. Crystal Growth 312, 3219 (2010). [113] Berger, C., et al., Electronic confinement and coherence in patterned epitaxial graphene. Science [129] Michon, A. , et al., Direct growth of few-layer 312, 1191 (2006). graphene on 6H-SiC and 3C-SiC/Si via propane chemical vapor deposition. Appl. Phys. Lett. (2010) 97, [114] De Heer, W.A., et al., Epitaxial graphene. Solid 171909. State Communications. 143, 92 (2007). [130] Strupinski, W., et al., Graphene Epitaxy by [115] Emtsev, K. V., et al., Towards wafer-size Chemical Vapor Deposition on SiC. Nano Lett. 11, graphene layers by atmospheric pressure 1786 (2011). graphitization of silicon carbide. Nature Mater. 8, 203 (2009). [131] Fanton, M. A., et al., Characterization of Graphene Films and Transistors Grown on Sapphire [116] Riedl, C. et al. Quasi-free-standing epitaxial by Metal-Free Chemical Vapor Deposition. ACS Nano graphene on SiC obtained by hydrogen intercalation. 5, 8062 (2011). Phys. Rev. Lett. 103, 246804 (2009). [132] Sun, J., et al., Large-area uniform graphene-like [117] Kedzierski, J., et al., Epitaxial Graphene thin films grown by chemical vapor deposition directly Transistors on SiC Substrates. Electron Devices, IEEE on silicon nitride. Appl. Phys. Lett. 98, 252107 (2011). Transactions. 55, 2078 (2008). [133] Scott, A., et al., The catalytic potential of high-κ [118] Rohrl, J., et al., Raman spectra of epitaxial dielectrics for graphene formation.Appl. Phys. Lett. graphene on SiC(0001). Appl. Phys. Lett. 92, 201918 98, 073110 (2011). (2008). [134] Dato, A. et al., Substrate-Free Gas-Phase [119] Obraztsov, A. N., et al., Chemical vapor Synthesis of Graphene Sheets Nano Lett. 8, 2012 deposition of thin graphite films of nanometer (2008). thickness. Carbon 45, 2017 (2007). [135] Al-Temimy, A., et al., Low temperature growth [120] Yu, Q., et al., Graphene segregated on Ni of epitaxial graphene on SiC induced by carbon surfaces and transferred to insulators. Appl. Phys. evaporation. Appl. Phys. Lett. 95, 231907 (2009). Lett. 93, 113103 (2008). [136] Bertoni, G., et al., First-principles calculation of [121] Reina, A., et al., Large Area, Few-Layer the electronic structure and EELS spectra at the Graphene Films on Arbitrary Substrates by Chemical graphene/Ni(111) interface. Phys. Rev. B 71, 075402 Vapor Deposition. Nano Lett. 9, 30 (2009). (2004) [122] Kim, K.S., et al., Large-scale pattern growth of [137] Land, T. A., et al., STM investigation of single graphene films for stretchable transparent electrodes. layer graphite structures produced on Pt(111) by Nature 457, 706 (2009). hydrocarbon decomposition Surf. Sci. 264, 261 (1992). [123] Sutter, P.W., J.-I. Flege, and E.A. Sutter, Epitaxial graphene on ruthenium. Nature Mater. 7, 406 (2008). nanoICT Strategic Research Agenda 69

Annex 1 nanoICT working groups position papers Graphene [138] N'Diaye, A. T., et al., Two-Dimensional Ir Cluster [153] Stradi, D., et al., Role of Dispersion Forces in the Lattice on a Graphene Moiré on Ir(111). Phys. Rev. Structure of Graphene Monolayers on Ru Surfaces. Lett. 97, 215501 (2006). Phys. Rev. Lett. 106, 186102 (2011). [139] Klusek, Z., et al., Local electronic edge states of [154] Ugeda, M. M., et al., Missing Atom as a Source graphene layer deposited on Ir(1 1 1) surface studied of Carbon Magnetism. Phys. Rev. Lett. 104, 096804 by STM/CITS. Appl. Surf. Sci. 252, 1221 (2005). (2010). [140] Marchini, S., Günther, S., Wintterlin, J., Scanning [155] N'Diaye, A. T., et al., In situ observation of stress tunneling microscopy of graphene on Ru(0001). Phys. relaxation in epitaxial graphene. New J. Phys. 11, Rev. B 76, 075429 (2007). 113056 (2009) [141] Vázquez de Parga, A. L., et al., Periodically [156] Rusponi, S., et al., Highly Anisotropic Dirac Rippled Graphene: Growth and Spatially Resolved Cones in Epitaxial Graphene Modulated by an Island Electronic Structure. Phys. Rev. Lett. 100, 056807 Superlattice. Phys. Rev. Lett. 105, 246803 (2010) (2007). [157] Gyamfi, M., et al., Inhomogeneous electronic [142] Rosei, R., et al., Structure of graphitic carbon on properties of monolayer graphene on Ru(0001) Phys. Ni(111): A surface extended energy-loss fi ne- Rev. B 83, 153418 (2011) structure study. Phys. Rev. B 28, 1161–1164 (1983). [158] Pletikosić, I., et al., Dirac Cones and Minigaps for [143] Dedkov, Y. S., Fonin, M., Laubschat, C., A Graphene on Ir(111) Phys. Rev. Lett. 102, 056808 possible source of spin-polarized electrons: The inert (2009). graphene/Ni(111) system. Appl. Phys. Lett. 92, 052506 (2008). [159] Varykhalov, A., et al., Electronic and Magnetic Properties of Quasifreestanding Graphene on Ni. [144] Dedkov Y. S., et al., Rashba Effect in the Phys. Rev. Lett. 101, 157601 (2008). Graphene/Ni(111) System. Phys. Rev. Lett. 100, 107602 (2008). [160] Balog, R., et al., Bandgap opening in graphene induced by patterned hydrogen adsorption. Nature [145] Martoccia, D., et al., Graphene on Ru(0001): A Mater. 9, 315 (2010). 25×25 Supercell. Phys. Rev. Lett. 101, 126102 (2008). [161] Haberer, D., et al., Tunable Band Gap in [146] Blanc, N., et al., submitted Hydrogenated Quasi-Free-Standing Graphene. Nano Lett. 10, 3360 (2010). [147] Busse, C., et al., Graphene on Ir(111): Physisorption with Chemical Modulation. Phys. Rev. [162] Rader, O., et al., Is There a Rashba Effect in Lett., 107, 036101 (2011). Graphene on 3d Ferromagnets? Phys. Rev. Lett. 102, 057602 (2009). [148] Giovannetti, G., et al. Doping Graphene with Metal Contacts Phys. Rev. Lett.101, 026803 (2008). [163] N'Diaye, A. T., et al., A versatile fabrication method for cluster Superlattices. New J. Phys. 11, [149] Wang, B., et al., Chemical origin of a graphene 103045 (2009). moiré overlayer on Ru(0001). Phys. Chem. Chem. Phys. 10, 3530 (2008). [164] Pozzo, M., et al., Thermal Expansion of Supported and Freestanding Graphene: Lattice [150] Morits, W., et al., Structure Determination of Constant versus Interatomic Distance Phys. Rev. Lett. the Coincidence Phase of Graphene on Ru(0001) 106, 135501 (2011). Phys. Rev. Lett. 104, 136102 (2010). [165] Sun, Z., et al., Topographic and electronic [151] Borca, B., et al., Electronic and geometric contrast of the graphene moiré on Ir(111) probed by corrugation of periodically rippled, self- scanning tunneling microscopy and noncontact nanostructured graphene epitaxially grown on atomic force microscopy Phys. Rev. B 83, 081415(R) Ru(0001) New J. Phys. 12, 093018 (2010). (2011). [152] Vanin, M., et al., Graphene on metals: A van der [166] Altenburg, S. J. , et al., Graphene on Ru(0001): Waals density functional study. Phys. Rev. B 81, Contact Formation and Chemical Reactivity on the 081408(R) (2010). Atomic Scale Phys. Rev. Lett. 105, 236101 (2010). 70 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene [167] Hill, E.W., et al., Graphene Spin Valve Devices. [182] Vo-Van, C., et al., Epitaxial graphene prepared IEEE TRANSACTIONS ON MAGNETICS, VOL. 42, 2694 by chemical vapor deposition on single crystal thin (2006). iridium films on sapphire Appl. Phys. Lett. 98, 181903 (2011). [168] Weser, M., et al., Electronic structure and magnetic properties of the graphene/Fe/Ni(111) [183] Park, H. J., et al., Growth and properties of few- intercalation-like system Phys. Chem. Chem. Phys. 13, layer graphene prepared by chemical vapor 7534 (2011). deposition. Carbon 48, 1088 (2009). [169] Vo-Van, C., et al., Ultrathin epitaxial cobalt films [184] Amara, H., Bichara, C. , Ducastelle, F., on graphene for spintronic investigations and Formation of carbon nanostructures on nickel applications New J. Phys. 12, 103040 (2010). surfaces: A tight-binding grand canonical Monte Carlo study. Phys. Rev. B 73, 113404 (2006). [170] Yazyev, O. V., Pasquarello, A., Magnetoresistive junctions based on epitaxial graphene and hexagonal [185] Amara H., et al., Tight-binding potential for boron nitride. Phys. Rev. B 80, 035408 (2009). atomistic simulations of carbon interacting with transition metals: Application to the Ni-C system. [171] Sicot, M., et al., Nucleation and growth of Phys. Rev. B 79, 014109 (2009). nickel nanoclusters on graphene Moiré on Rh(111) Appl. Phys. Lett. 96, 093115 (2010). [186] Karoui, S., et al., Nickel-Assisted Healing of Defective Graphene. [172] Donner, K. , Jakob, P., Structural properties and ACS Nano 4, 6114 (2010). site specific interactions of Pt with the graphene/Ru(0001) moiré overlayer. J. Chem. Phys. [187] Thiel, S., et al., Engineering polycrystalline Ni 131, 164701 (2010). films to improve thickness uniformity of the chemical- [173] Pollard, A. J., et al., Supramolecular Assemblies Formed on an Epitaxial Graphene Superstructure. vapor-deposition-grown graphene films. Angew. Chem. Int. Ed. 49, 1794 (2010). Nanotechnology 21, 015601 (2010). [174] Tontegode, A. Y. , Carbon on transition metal surfaces Prog. Surf. Sci. 38, 201 (1991). [188] Baraton, L. et al., Europhys. Lett., in press [175] Coraux, J., et al., Growth of graphene on Ir(111) [189] Dervishi, E., et al., The role of hydrocarbon New J. Phys. 11, 023006 (2009). concentration on the synthesis of large area few to multi-layer graphene structures. Chem. Phys. Lett. [176] Gao, M., et al., Tunable interfacial properties of 501, 390 (2011). epitaxial graphene on metal substrates Appl. Phys. Lett. 96, 053109 (2010). [190] Hesjelda, T., Continuous roll-to-roll growth of graphene films by chemical vapor deposition. Appl. [177] van Gastel, R., et al., Selecting a single Phys. Lett. 98, 133106 (2011). orientation for millimeter sized graphene sheets. Appl. Phys. Lett. 95, 121901 (2009). [191] Mattevi, C., Kim, H., Chhowalla, M., A review of chemical vapour deposition of graphene on copper. J. [178] Günther, S., et al., Single Terrace Growth of Mater. Chem. 21, 3324 (2011). Graphene on a Metal Surface. Nano Lett. 11, 1895 (2011). [192] Oznuler, T., et al., Synthesis of graphene on gold. Appl. Phys. Lett. 98, 183101 (2011). [179] Hattab, H., et al., Growth temperature dependent graphene alignment on Ir(111) Appl. Phys. [193] Robertson, A. W., Warner, J. H., Hexagonal Lett. 98, 141903 (2011). Single Crystal Domains of Few-Layer Graphene on Copper Foils. Nano Lett. 11, 1182 (2011). [180] Dong, G. et al. (in press) [194] Wu, Y. A., et al., Aligned Rectangular Few-Layer [181] Iwasaki, T., et al., Long-Range Ordered Single- Graphene Domains on Copper Surfaces. Chem. Crystal Graphene on High-Quality Heteroepitaxial Ni Mater. 23, 4544 (2011). Thin Films Grown on MgO(111). Nano Lett. 11, 79 (2011) nanoICT Strategic Research Agenda 71

Annex 1 nanoICT working groups position papers Graphene [195] Lherbier, A., et al., Charge Transport in [209] Russo V. et al. Raman spectroscopy of Bi-Te thin Chemically Doped 2D Graphene. Phys. Rev. Lett. 101, films. J. Raman Spectrosc. 39, 205 (2008) 036808 (2008). [210] Gordon, R. J., Criteria for Choosing Transparent [196] Bieri, M., et. al. Porous graphenes: two- Conductors. MRS Bulletin 8, 53 (2000). dimensional polymer synthesis with atomic precision. Chem Commun. 2009 , 6919-6921. [211] Hamberg, I. and Granqvist, C. G., Evaporated Sn- doped In2O3 films: basic optical properties and [197] Booth, T., et al, Discrete Dynamics of applications to energy-efficient windows. J. Appl. Nanoparticle Channelling in Suspended Graphene. Phys. 60, R123 (1986). Nano Lett. 11, 2689 (2011). [212] Granqvist, C. G., Transparent conductors as [198] Tapaszto, L., et al., Tailoring the atomic solar energy materials: a panoramic review. Sol. structure of graphene nanoribbons by scanning Energy Mater. Sol. Cells 91, 1529–1598 (2007). tunnelling microscope lithography. Nature Nanotech. 3, 397-401 (2008). [213] Sheraw, C. D., et al., Organic thin-fi lm transistor-driven polymer dispersed liquid crystal [199] Wagner, P., Bezanilla, A., Roche, S., Ivanovskaya, displays on fl exible polymeric substrates. Appl. Phys. V. V., Rayson, M. J., Ewels, C. P. in preparation (2011) Lett. 80, 1088 (2002). [200] Cervantes-Sodi, F., et al., Edge-functionalized [214] Lee, J. Y., et al., Solution-processed metal and substitutionally doped graphene nanoribbons: nanowire mesh transparent electrodes. Nano Lett. 8, Electronic and spin properties. Phys. Rev. B. 77, 689 (2008). 165427 (2008). [215] De, S., et al. Silver nanowire networks as fl [201] Ivanovskaya, V. V., et al, Low-energy exible, transparent, conducting films: extremely high termination of graphene edges via the formation of dc to optical conductivity ratios. ACS Nano 3, 1767 narrow nanotubes. Phys. Rev. Lett., 107, 065502 (2009). (2011). [216] Geng, H. Z., et al., Effect of acid treatment on [202] Wagner, P., et al, Ripple edge engineering of carbon nanotube-based flexible transparent graphene nanoribbons. arXiv:1107.1977v1 (2011). conducting films. J. Am. Chem. Soc. 129, 7758 (2007). [203] Cocchi, C., et al., Designing All-Graphene [217] Wu, Z., et al., Transparent, conductive carbon Nanojunctions by Covalent FunctionalizationJ. Phys. nanotube films. Science 305, 1273 (2004). Chem. C, 115, 2969 (2011). [218] De, S. and Coleman, J. N,. Are there [204] Biel, B., et al., Anomalous Doping Effects on fundamental limitations on the sheet resistanceand Charge Transport in Graphene Nanoribbons. Phys. transmittance of thin graphene films? ACS Nano 4, Rev. Lett. 102, 096803 (2009). 2713 (2010). [205] Banhart, F., Kotakoski, J. Krasheninnikov, A. V., [219] Casiraghi, C., et al., Raman fingerprint of Structural Defects in Graphene. ACS Nano 5, 26 charged impurities in graphene Appl. Phys. Lett. 91, (2011). 233108 (2007) [206] Ferrari, A. C. and Robertson, J. (eds), Raman [220] Sahu, D. R., Lin, S. Y. and Huang, J. L., spectroscopy in carbons: from nanotubes to diamond, ZnO/Ag/ZnO multilayer films for the application of a Theme Issue, Phil. Trans. Roy. Soc. A 362, 2267-2565 very low resistance transparent electrode. Appl. Surf. (2004). Sci. 252, 7509 (2006). [207] Nemanich, R.J., Solin, S. A., Martin, R.M., Light- [221] Zhu, Y., et al., Rational Design of Hybrid scattering study of boron nitride microcrystals, Phys. Graphene Films for High-Performance Transparent Rev. B 23, 6348 (1981). Electrodes. Acs Nano 5, 6472 (2011). [208] Reich, S., et al., Resonant Raman scattering in [222] Tung, V. C., et al,. Low-temperature solution cubic and hexagonal boron nitride. Phys. Rev B 71, processing of graphene-carbon nanotube hybrid 205201 (2005). materials for high-performance transparent conductors. Nano Lett. 9, 1949 (2009). 72 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene [223] Wang, X., Zhi, L., and Mullen, K., Transparent, [238] Stoller, M. D., et al., Graphene-based conductive graphene electrodes for dye-sensitized ultracapacitor. Nano Lett., vol. 8, 3498 (2008). solar cells. Nano Lett. 8, 323 (2007). [239] Hong, W., et al., Transparent graphene/PEDOT- [224] Eda, G., Fanchini, G. & Chhowalla, M., Large- PSS composite fi lms as counter electrodes of dye area ultrathin films of reduced graphene oxide as a sensitized solar cells” Electrochem. Commun. 10, transparent and flexible electronic material. Nature 1555 (2008). Nanotech. 3, 270 (2008). [240] Bonaccorso, F., et al., Graphene-based Natural [225] King, P. J., et al.,Improvement of Transparent Dye-Sensitized Solar Cells” Graphene 2011, Conducting Nanotube Films by Addition of Small Imaginenano 11-14 April, Bilbao, (Spain). Quantities of Graphene. ACS Nano 4, 4238 (2010). [241] Echtermeyer, T.J., et al. Strong plasmonic [226] Matyba, P., et al., Graphene and mobile ions: enhancement of photovoltage in graphene. Nature the key to all-plastic, solution processed light-emitting Comm. 2, 458 (2011). devices. ACS Nano 4, 637 (2010). [242] Burroughes, J. H., et al. Light-emitting diodes [227] Bonaccorso, F., Graphene photonics and based on conjugated polymers. Nature 347, 539–541 optoelectronic. ArXiv1006.4854v1. (1990). [228] Chapin, D. M., Fuller, C. S. and Pearson, G. L., A [243] Wu, J., et al. Organic light-emitting diodes on new silicon p-n junction photocell for converting solar solution-processed graphene transparent electrodes. radiation into electrical power. J. Appl. Phys. 25, 676 ACS Nano 4, 43–48 (2009). (1954). [244] Han, T-H., et al., Extremely efficient flexible [229] Green, M. A., et al., Solar cell efficiency tables. organic light emitting diodes with modified graphene Prog. Photovolt. Res. Appl. 7, 321 (1999). anode. Nature Photon. 6, 105 (2012). [230] Peter, L. M., Towards sustainable photovoltaics: [245] Pickering, J. A., Touch-sensitive screens: the the search for new material. Phil. Trans. R. Soc. A technologies and their applications. Int. J. Man. Mach. 369, 1840 (2011). Stud. 25, 249 (1986). [231] Li, X., et al., Graphene-On-Silicon Schottky [246] Silfverberg, M. 5th International Symposium, Junction Solar Cells. Adv. Mater. 22, 2743,(2010). Mobile HCI 2003, Udine, Italy, 2003. [232] Wang, X., et al., Transparent carbon films as [247] Beecher, P., et al., Material Choices For Haptic electrodes in organic solar cells. Angew. Chem. 47, Interfaces. Material Research Society 2011 Spring 2990 (2008). Meeting April 26-29, 2011 San Francisco, California (USA). [233] De Arco, L. G., et al., Continuous, highly flexible, and transparent graphene films by chemical vapor [248] E. Mallinckrodt, A. L. Hughes, W., Sleator deposition for organic photovoltaics” ACS Nano 4, Science, 118, 277 (1953) 2865 (2010). [249] Kaczmarek, K. A. et al., Polarity Effect in [234] Liu, Z., et al., Organic photovoltaic devices Electrovibration for Tactile Display,\" Biomedical based on a novel acceptor material: Graphene. Adv. Engineering. IEEE Transactions on, 53, 2047 (2006). Mater. 20, 3924 (2008). [250] Saleh, B. E. A. & Teich, M. C. Fundamentals of [235] Yong, Y., Tour, J. M., Theoretical efficiency of Photonics Ch. 18, 784–803 (Wiley, 2007). nanostructured graphene-based photovoltaics. Small 6, 313 (2009). [251] Dawlaty, J. M., et al. Measurement of the optical absorption spectra of epitaxial graphene from [236] Yang, N., et al., Two-dimensional graphene terahertz to visible. Appl. Phys. Lett. 93, 131905 bridges enhanced photoinduced charge transport in (2008). dye-sensitized solar cells ACS Nano 4, 887 (2010). [252] Wright, A. R., Cao, J. C. and Zhang, C., Enhanced [237] Yan, X., et al., “Large, Solution-Processable optical conductivity of bilayer graphene nanoribbons Graphene Quantum Dots as Light Absorbers for in the terahertz regime. Phys. Rev. Lett. 103, 207401 Photovoltaics” Nano Lett., 10, 1869, (2010). (2009). [253] Sun, Z., et al. Graphene mode-locked ultrafast laser. ACS Nano 4, 803–810 (2010). nanoICT Strategic Research Agenda 73

Annex 1 nanoICT working groups position papers Graphene [254] Mueller, T., Xia, F. and Avouris, P., Graphene [270] Keller, U., Recent developments in compact photodetectors for high-speed optical ultrafast lasers. Nature 424, 831 (2003). communications. Nature Photon. 4, 297 (2010). [271] Okhotnikov, O., Grudinin, A., Pessa, M., Ultra- [255] Vasko, F. T., and Ryzhii, V., Photoconductivity of fast fibre laser systems based on SESAM technology: intrinsic graphene. Phys. Rev. B 77, 195433 (2008). new horizons and applications. New J. Phys. 6, 177 (2004) [256] Park, J., Ahn, Y. H. and Ruiz-Vargas, C., Imaging of photocurrent generation and collection in single- [272] Hasan, T., et al. Nanotube–polymer composites layer graphene. Nano Lett. 9, 1742 (2009). for ultrafast photonics. Adv. Mater. 21, 3874 (2009). [257] Xia, F. N., et al. Photocurrent imaging and effi [273] Breusing, M., Ropers, C. and Elsaesser, T., cient photon detection in a graphene transistor. Nano Ultrafast carrier dynamics in graphite. Phys. Rev. Lett. Lett. 9, 1039 (2009). 102, 086809 (2009). [258] Xia, F., et al., Ultrafast graphene photodetector. [274] Sun, D., et al., Ultrafast Relaxation of Excited Nature Nanotech. 4, 839 (2009). Dirac Fermions in Epitaxial Graphene Using Optical Differential Transmission Spectroscopy. Phys. Rev. [259] Xu, X. D., et al., Photo-thermoelectric eff ect at a Lett. 101, 157402, (2008). graphene interface junction. Nano Lett. 10, 562 (2010). [275] Sun, Z., et al., High-power ultrafast solid-state laser using graphene based saturable absorber, in The [260] Gabor, N. M., et al., Hot Carrier–Assisted Conference on Lasers and Electro-Optics(Baltimore, Intrinsic Photoresponse in Graphene. Science 334, US, 2011), JWA79. 648 (2011) [276] Popa, D., et al., Sub 200fs pulse generation from [261] Blake, P., et al. Influence of metal contacts and a graphene mode-locked fiber laser. Appl. Phys. Lett. charge inhomogeneity on transport properties of 97, 203106 (2010) graphene near the neutrality point. Solid State Commun.149, 1068 (2009) [277] Popa, D., et al., Graphene Q-switched, tunable fiber laser”. Appl. Phys. Lett. 98, 073106 (2011) [262] Kang, Y. M. , et al. Monolithic germanium/silicon avalanche photodiodes with [278] Popa, D., et al., Sub 200fs pulse generation from 340 GHz gain–bandwidth product. Nature Photon. 3, a graphene mode-loched fiber laser. Appl. Phys. Lett. 59 (2009) 97, 203106 (2010). [263] Kuzmenko, A. B., et al., Universal optical [279] Zhang, H., et al. Vector dissipative solitons in conductance of graphite. Phys. Rev. Lett. 100, 117401 graphene mode locked fiber lasers. Opt. Commun. (2008). 283, 3334 (2010). [264] Lee, E. J. H., et al. Contact and edge effects in [280] Zhang, H., et al. Compact graphene mode- graphene devices. Nat. Nanotechnol. 3, 486 (2008) locked wavelength-tunable erbium-doped fiber lasers: from all anomalous dispersion to all normal [265] Konstantatos, G., et al., Hybrid graphene- dispersion. Laser Phys. Lett. 7, 591 (2010). quantum dot phototransistors with ultrahigh gain. ArXiv. 1112.4730v1 [281] Yu, H., et al. Large Energy Pulse Generation Modulated by Graphene Epitaxially Grown on Silicon [266] Engel, M., et al.,Light-matter interaction in a Carbide. ACS Nano 4, 7582 (2010). microcavity-controlled graphene transistor. arXiv:1112.1380v1. [282] Chang, Y. M.,et al., Multilayered graphene efficiently formed by mechanical exfoliation for non [267] Vahala, K. J., Optical microcavities. Nature 424, linear saturable absorbers in fiber mode locked lasers. 839 (2003). Appl. Phys. Lett. 97, 211102 (2010). [268] Freitag, M., et al., Thermal infrared emission [283] Martinez, A., Fuse, K. and Yamashita, S., from biased graphene. Nat. Nanotechnol. 5, 497 Mechanical exfoliation of graphene for the passive (2010). mode-locking of fiber lasers. 99, 121107 (2011). [269] Berciaud, S., et al., Electron and Optical Phonon [284] Tan, W. D., et al., Mode locking of ceramic Temperatures in Electrically Biased Graphene. Phys. Nd:yttrium aluminum garnet with graphene as a Rev. Lett. 104, 227401 (2010). 74 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene saturable absorber. Appl. Phys. Lett. 96, 031106 [301] Echtermeyer, T. J., et al., Graphene field-effect (2010). devices. European Physical Journal-Special Topics. 148, 19 (2007) [285] Cho, W. B. et al., High-quality, large-area monolayer graphene for efficient bulk laser mode- [302] Williams, J.R., Di Carlo, L. and Marcus, C.M., locking near 1.25 um. Opt. Lett., 36, 4089-4091 (2011) Quantum Hall Effect in a Gate-Controlled p-n Junction of Graphene. Science 317, 638 (2007). [286] Martinez, A., et al., Optical deposition of graphene and carbon nanotubes in a fiber ferrule for [303] Takagi, S., et al., On the universality of inversion passive mode-locked lasing. Opt. Express 18, 23054 layer mobility in Si MOSFET's: Part I-effects of (2010). substrate impurity concentration. Electron Devices, IEEE Transactions 41, 2357 (1994). [287] Wang, F. ,et al., ~2um Graphene Q-switched Thulium Fiber Laser. Submitted (2011). [304] Martin, J., et al., Observation of electron-hole puddles in graphene using a scanning single-electron [288] Sun, Z., et al., A stable, wideband tunable, near transistor. Nature Phys. 4, 144 (2008). transform-limited, graphene-mode-locked, ultrafast laser. Nano Res. 3, 404 (2010). [305] Lemme, M.C., et al., Mobility in graphene double gate field effect transistors. Solid-State [289] W. Koechner, Solid-State Laser Engineering Electronic 52, 514 (2008). (Springer, New York, 2006). [306] Lin, Y.-M., et al., Operation of Graphene [290] Lee, C.-C. et al., Ultra-Short Optical Pulse Transistors at Gigahertz Frequencies. Nano Lett. 9, Generation with Single-Layer Graphene. J. Nonlinear 422 (2009). Opt. Phys. Mater. 19, 767-771 (2010). [307] Nakada, K., et al., Edge state in graphene [291] Bass, M., Li, G. and Stryland, E. V. Handbook of ribbons: Nanometer size effect and edge shape Optics IV (McGraw-Hill, 2001). dependence. Physical Review B. 54, 17954 (1996). [292] Wang, J. et al., Broadband nonlinear optical [308] Barone, V., Hod, O., and Scuseria G.E., response of graphene dispersions. Adv. Mater. 21, Electronic Structure and Stability of Semiconducting 2430 (2009). Graphene Nanoribbons. Nano Lett. 6, 2748 (2006). [293] Xu, Y., et al. A graphene hybrid material [309] Son, Y.-W., Cohen, M.L. and Louie S.G., Energy covalently functionalized with porphyrin: synthesis Gaps in Graphene Nanoribbons. Phys. Rev. Lett. 97, and optical limiting property. Adv. Mater. 21, 1275 216803 (2006). (2009). [310] Obradovic, B., et al., Analysis of graphene [294] Mikhailov, S. A. Non-linear electromagnetic nanoribbons as a channel material for field-effect response of graphene. Europhys. Lett. 79, 27002 transistors. Appl. Phys. Lett. 88, 142102 (2006). (2007). [311] Yang, L., et al., Band-gap change of carbon [295] Dean, J. J. and van Driel, H. M., Second nanotubes: Effect of small uniaxial and torsional harmonic generation from graphene and graphitic strain. Phys. Rev. B, 60, 13874 (1999). films. Appl. Phys. Lett. 95, 261910 (2009). [312] Fujita, M., et al., Peculiar Localized State at [296] Hendry, E., et al., Strong nonlinear optical Zigzag Graphite Edge. J. Phys. Soc. Jpn. 65, 1920 response of graphene flakes measured by four-wave (1996). mixing. Phys. Rev. Lett. 105, 097401 (2010). [313] Huang, B., et al., Making a field effect transistor [297] Dhar, S. et al., A new route to graphene layers on a single graphene nanoribbon by selective doping. by selective laser ablation AIP ADVANCES 1, 022109 Appl. Phys. Lett.. 91, 253122 (2007). (2011). [314] Yan, Q., et al., Intrinsic Current-Voltage [298] ITRS roadmap 2009, www.itrs.net (2011). Characteristics of Graphene Nanoribbon Transistors and Effect of Edge Doping. Nano Lett. 7, 1469 (2007). [299] Lemme, M.C., et al., Towards Graphene Field Effect Transistors. ECS Transactions, 11, 413 (2007). [315] Kan, E.-J., et al., Will zigzag graphene nanoribbon turn to half metal under electric field? [300] Lemme, M.C., et al., A Graphene Field-Effect Appl. Phys. Lett. 91, 243116 (2007). Device. Electron Device Letters, IEEE, 28, 282 (2007). nanoICT Strategic Research Agenda 75

Annex 1 nanoICT working groups position papers Graphene [316] Hod, O., et al., Enhanced Half-Metallicity in [331] Zhang, Y. T. et al., Direct observation of a widely Edge-Oxidized Zigzag Graphene Nanoribbons. Nano tunable bandgap in bilayer graphene. Nature 459, 820 Lett. 7, 2295 (2007). (2009). [317] Kudin, K.N., Zigzag Graphene Nanoribbons with [332] Oostinga, J.B., et al., Gate-induced insulating Saturated Edges. ACS Nano 2, 516 (2008). state in bilayer graphene devices. Nature Mater. 7, 151 (2008). [318] Yoon, Y. and Guo, J. Effect of edge roughness in graphene nanoribbon transistors. Appl. Phys. Lett. 91, [333] Szafranek, B. N. et al., Electrical observation of a 073103 (2007). tunable band gap in bilayer graphene nanoribbons at room temperature Appl. Phys. Lett. 96, 112103 [319] Sols, F., Guinea, F. and Neto A.H.C., Coulomb (2010). Blockade in Graphene Nanoribbons. Phys. Rev. Lett. 99, 166803 (2007). [334] Szafranek, B. N. et al., High On/Off Ratios in Bilayer Graphene Field Effect Transistors Realized by [320] Cervantes-Sodi, F., et al., Edge-functionalized Surface Dopants. Nano Lett. 11, 2640 (2011). and substitutionally doped graphene nanoribbons: Electronic and spin properties. Phys. Rev. B 77, [335] Li, S-L., et al., Complementary-like Graphene 165427 (2008). Logic Gates Controlled by Electrostatic Doping. Small 7, 1552 (2011). [321] Han, M.Y., et al., Energy Band-Gap Engineering of Graphene Nanoribbons. Phys. Rev. Lett. 98 206805 [336] Schwierz, F. and Liou, J.J., RF transistors: Recent (2007). developments and roadmap toward terahertz applications. Solid-State Electronics. 51, 1079 (2007). [322] Chen, Z., et al., Graphene nano-ribbon electronics. Physica E: Low-dimensional Systems and [337] Frank, D.J., Taur, Y. and Wong, H.S.P., Nanostructures 40, 228 (2007). Generalized scale length for two-dimensional effects in MOSFETs. Electron Device Letters, IEEE. 19, 385 [323] Han, M.Y., et al., Electron Transport in (1998). Disordered Graphene Nanoribbons Phys. Rev. Lett.104, 056801 (2010). [338] Meric, I., et al. RF performance of top-gated, zero-bandgap graphene field-effect transistors. in [324] Ponomarenko, L.A., et al., Chaotic Dirac Billiard Electron Devices Meeting, 2008. IEDM 2008. IEEE in Graphene Quantum Dots. Science 320, 356 (2008). International. 2008. [325] Wang, X., et al., Room-Temperature All- [339] Moon, J.S., et al., Epitaxial-Graphene RF Field- Semiconducting Sub-10-nm Graphene Nanoribbon Effect Transistors on Si-Face 6H-SiC Substrates. Field-Effect Transistors. Phys. Rev. Lett. 100, 206803 Electron Device Letters, IEEE. 30, 650 (2009). (2008). [340] Palacios, T., Hsu, A., and Wang H. Applications [326] Lemme, M.C., et al., Etching of Graphene of graphene devices in RF communications. Devices with a Helium Ion Beam. ACS Nano, 3, 2674 Communications Magazine, IEEE 48, 122 (2010). (2009). [341] Lin, Y-M., et al., 100-GHz Transistors from [327] Bell, D.C., et al., Precision Cutting and Patterning Wafer-Scale Epitaxial Graphene. Science 327, 662 of Graphene with He Ions. Nanotechnology 20, (2010). 455301 (2009). [342] Schwierz, F. and Liou, J. J. Modern Microwave [328] McCann, E., Asymmetry gap in the electronic Transistors – Theory, Design, and Performance (Wiley, band structure of bilayer graphene. Physical Review B 2003). (Condensed Matter and Materials Physics) 74, 161403 (2006). [343] Wu, Y., et al., High-frequency, scaled graphene transistors on diamond-like carbon, Nature 472, 74 [329] Ohta, T., et al., Controlling the Electronic (2011). Structure of Bilayer Graphene. Science 313, 951 (2006). [344] Liao, L., et al.High-speed graphene transistors with a self-aligned nanowire gate. Nature 467, 305 [330] Mak K. F., et al., Observation of an Electric-Field- (2010). Induced Band Gap in Bilayer Graphene by Infrared Spectroscopy Phys. Rev. Lett. 102, 256405 (2009). 76 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Graphene [345] Schwierz, F., Graphene transistors. Nature [360] Grassi, R., et al., Hierarchical modeling of carbon Nanotech. 5, 487 (2010) nanoribbon devices for CNR-FETs engineering, Device Research Conference, 2008. [346] http://www.itrs.net/Links/2010ITRS/IRC-ITRS- MtM-v2%203.pdf [361] Betti, A., Fiori, G., Iannaccone, G., Atomistic Investigation of Low-Field Mobility in Graphene [347] Wu, X., et al., Epitaxial-Graphene/Graphene- Nanoribbons ,IEEE Trans. Electr. Dev., available at: Oxide Junction: An Essential Step towards Epitaxial ieeexplore.ieee.org Graphene Electronics. Phys. Rev. Lett. 101, 026801 (2008). [362] Ferry, D. K., Semiconductor transport London: Taylor & Francis, 2000. Technology & Engineering [348] Standley, B., et al., Graphene-Based Atomic- ISBN 0-7484-0865-7. Scale Switches. Nano Letters, 2008. 8(10): p. 3345- 3349. [363] Elias, D. , et al., Control of Graphene's Properties by Reversible Hydrogenation: Evidence for [349] Boukhvalov, D.W. and Katsnelson, M.I. , Tuning Graphane. Science 323, 610 (2009). the gap in bilayer graphene using chemical functionalization: Density functional calculations. [364] Lebègue, S., et al., Accurate electronic band Phys. Rev. B 78, 085413 (2008). gap of pure and functionalized graphane from GW calculations. Phys. Rev. B 79, 245117 (2009). [350] Echtermeyer, T.J., et al., Nonvolatile Switching in Graphene Field-Effect Devices. Electron Device [365] Fiori, G., et al., Simulation of hydrogenated Letters, IEEE. 29, 952 (2008). graphene field-effect transistors through a multiscale approach. Phys. Rev. B, 82, 153404 (2010). [351] Zheng, Y., et al., Gate-controlled nonvolatile graphene-ferroelectric memory. Appl. Phys. Lett. 94, [366] Chauhan, J., and Guo, J., Assessment of High- 163505 (2009). Frequency Performance Limits of Graphene Field- Effect Transistors. Nano Res. 4, 571 (2011). [352] Fiori, G., Iannaccone, G., On the possibility of tunable-gap bilayer graphene FET, IEEE Electron [367] Nam Do, V., et al., Electronic transport and spin- Device Letters, Vol. 30, pp. 261-264, 2009. polarization effects of relativisticlike particles in mesoscopic graphene structures, J. Appl. Phys. 104, [353] Fiori, G., Iannaccone, G., Ultralow-Voltage 063708 (2008). Bilayer graphene tunnel FET, IEEE Electron Device Letters, Vol. 30, pp.1096-1098, 2009. [368] Hung Nguyen, V., et al., Quantum transport of dirac fermions in graphene field effect transistors, [354] Chang, L., et al., Extremely Scaled Silicon Nano- SISPAD 2010, p.9. CMOS Device. Proceedings of the IEEE, 91, 1860 (2003). [355] Hu, C., et al., Green Transistor - A VDD Scaling Path for Future Low Power ICs. IEEE VLSI Technology, Systems and Applications, 2008. VLSI-TSA 2008. [356] Britnell, L., et al., Field-effect tunneling transistor based on vertical graphene heterostructures. ArXiv1112.4999 [357] Geim, A.K., QED in a Pencil Trace KITP Graphene Workshop 01-09-07. [358] Fiori, G., Iannaccone, G. Simulation of Graphene Nanoribbon Field-Effect Transistors. IEEE, Electron Device Letters 28, 760 (2007). [359] Yoon, Y., et al., Performance Comparison of Graphene Nanoribbon FETs With Schottky Contacts and Doped Reservoirs. IEEE Transaction on Electron Devices 55, 2314 (2008). nanoICT Strategic Research Agenda 77

Modeling

Position Paper on Modeling M. Macucci1, S. Roche2, A. Correia3, J. Greer4, This is a further update of the position paper X. Bouju5, M. Brandbyge6, J. J. Saenz7, M. for the Theory and Modelling Working Group, Bescond8, D. Rideau9, P. Blaise10, D. Sanchez- including the results of the last two meetings, Portal11, J. Iñiguez12, X. Oriols13, G. Cuniberti14, one held in Pisa during the IWCE conference, and H. Sevincli14 in October 2010, and the other held in Bilbao, during the IMAGINENANO conference, in April 1 Dipartimento di Ingegneria dell’informazione, 2011. Università di Pisa, Pisa, Italy 2 CIN2 (ICN-CSIC) and UAB, Catalan Institute of While the previous update (in 2009) of the Nanotechnology, Barcelona, Spain original 2008 document was substantially 3 Phantoms Foundation, Madrid, Spain focused on the main issues on which 4 Tyndall National Institute, Lee Maltings, Cork, collaboration within the modelling Ireland community is needed and on an analysis of 5 Centre d’Élaboration de Matériaux et d’Études the situation in Europe in comparison with Structurales (CEMES), Nanosciences Group, that in the rest of the world, the present Toulouse, France update primarily deals with the actual needs 6 DTU Nanotech, Department of Micro and of the industry in terms of modelling, with the Nanotechnology, Technical University of Denmark, relevance of physics-based modeling, with Kongens Lyngby, Denmark time-dependent analysis, with new 7 Universidad Autónoma de Madrid, Departamento computational approaches, and with issues de Física de la Materia Condensada, Madrid, Spain rising in the simulation of wide-bandgap 8 IM2NP, UMR CNRS 6242, Marseille, France semiconductors. 9 STMicroelectronics, Crolles, France 10 CEA, LETI, MINATEC, Grenoble, France 1. Introduction 11 Centro de Fisica de Materiales CSIC-UPV/EHU, Materials Physics Center MPC, San Sebastián, Spain We are presently witnessing the final phase of 12 Institut de Ciència de Materials de Barcelona the downscaling of MOS technology and, at (ICMAB-CSIC), Bellaterra, Spain the same time, the rise of a multiplicity of 13 Dept. of Electron. Eng., Univ. Autonoma de novel device concepts based on properties of Barcelona, Barcelona, Spain matter at the nanoscale and even at the 14 Institute for Materials Science and Max Bergmann molecular scale. Center of Biomaterials, Dresden University of Technology, Dresden, Germany Ultra-scaled MOS devices and nanodevices relying on new physical principles share the reduced dimensionality and, as a result, many nanoICT Strategic Research Agenda 79

Annex 1 nanoICT working groups position papers Modeling of the modelling challenges. In addition, new GPU-based general purpose computing, and materials and process steps are being the very recent development of software included into MOS technology at each new environments, such as OPENACC that allow an node, to be able to achieve the objectives of almost automated porting of legacy codes to the Roadmap; these changes make traditional hybrid GPU-CPU platforms) make the time simulation approaches inadequate for reliable ripe for a real leap forward in the scope and predictions. So far modelling at the nanoscale performance of computational approaches has been mainly aimed at supporting research for nanotechnology and nanosciences. and at explaining the origin of observed phenomena. 2. Current Status of MOS simulation and industrial needs In order to meet the needs of the MOS industry and to make practical exploitation of The continuous downscaling of MOSFET new device and solid-state or molecular critical dimensions, such as the gate length material concepts possible, a new integrated and the gate oxide thickness, has been a very approach to modelling at the nanoscale is successful process in current manufacturing, needed, as we will detail in the following. A as testified, e.g., by the ITRS requirements. hierarchy of multi-scale tools must be set up, However, conventional scaling down of in analogy with what already exists for MOSFET channel length is declining as the microelectronics, although with a more physical and economic limits of such an complex structure resulting from the more approach are coming closer. Novel solutions intricate physical nature of the devices and are increasingly being used in MOSFET the intrinsically multi-physics nature of the channel engineering within the industry. problems to be solved. Among the new technological features of very A coordinated effort in the field of modelling advanced devices, high-k dielectrics, the is apparent in the United States, where archetype of which is hafnium oxide, can significant funding has been awarded to the significantly reduce gate leakage. Mechanical Network for Computational Nanotechnology, strain applied in the channel and substrate which is coordinating efforts for the orientation can also significantly improve development of simulation tools for carrier mobility. Moreover, alternative nanotechnology of interest both for the geometries, such as double-gated devices, in academia and for the industry. which the channel doping level is relatively low, must be evaluated within the perspective Although the required integrated platforms of an industrial integration. In particular, the need to be developed, the efforts made in the subsequent effects of the high-k gate last few years by the modelling community dielectric and of the double-gate geometry on have yielded significant advances in terms of channel mobility must be clearly quantified. quantitatively reliable simulation and of ab- initio capability, which represent a solid basis Technology Computer-Aided Design (TCAD) on which a true multi-scale, multi-physics refers to the use of computer simulations to hierarchy can be built. The combination of develop and optimize semiconductor devices. these new advanced software tools and the State-of-the-art commercial TCAD device availability of an unprecedented simulators are currently working using the computational power (in particular Drift-Diffusion (DD) approximation of the considering the recent advances in terms of 80 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling Boltzmann transport equation. Quantum cannot be of great help when trying to effects are accounted for using the Density determine what effect a change of Gradient approximation, that works well for parameters will have in a new technology. traditional bulk devices, but that can be unreliable for advanced devices such as the It becomes urgent to develop new physically- double-gated-MOS structure or for new based models with a view of integrating them materials. Moreover, emerging materials also into a standardized simulation platform that significantly challenge the conventional DD- can be efficiently used in an industrial based tools, mostly due to a lack of environment. For this purpose, tight appropriate models and parameters. collaborations between world-class universities and research institutions, CAD It is indeed important to understand what vendors and industrial partners must be would make large semiconductor companies established. Within the framework of these really interested in advanced modeling tools: collaborations, there will be the best chances an answer recurrent from people with of success, both in terms of academic model experience in the industry is that a simulation development and theoretical achievements, code should be able to provide insight into but also in terms of concrete implementations why a given advanced device does not work, and benchmarks of new models in TCAD an answer that often cannot be obtained tools. Innovative concepts based on nano- from existing commercial codes. In particular, materials or molecular devices, new models it would be important to have a model that, and simulation tools would provide our ICT even though not quantitatively reliable, could industries a competitive advantage for device tell when things are not going according to development and optimization in terms of intuition. This would be far more important time-cycle and wafer-costs. for industry developers than having a very complex code that shows that there is just a 3. Commercial vs. academic small correction with respect to the behavior quantum-transport solvers of the device that could be predicted just by intuition or by means of standard simulators. In response to the industrial need of new simulation tools, a class of quantum and If the real added value in a sophisticated transport solvers is emerging. These model is to predict a behavior that differs commercial state-of-the-art solvers can be from what intuition would tell us, rather than divided into two categories. In the first providing quantitatively exact results, the category, one can find the quantum-transport question arises whether drift-diffusion is all solvers, such as those based on the Non- the industry needs, because with small Equilibrium Green Function Method, in which changes it has been adapted to successive carrier transport is treated using the full generations of devices, always yielding quantum Green function formalism. In the acceptable results. In reality, this is not true, second category one can include the Monte because adaptation of drift-diffusion have Carlo Solvers, that model carrier transport via often been made \"a posteriori,\" once the the Boltzmann equation. This equation is actual behavior of the device was known from solved in a stochastic way, using a classical measurements or from more sophisticated description of the free fly of the electrons but models. Very simple models, such as drift- a quantum description of the interactions. diffusion, although quite efficient when The currently available high-level NEGF[1] and dealing with a well established technology, nanoICT Strategic Research Agenda 81

Annex 1 nanoICT working groups position papers Modeling MC solutions [2] are still in the development be ``customized,'' in order to make fast and phase, and no ready-to-use industrial accurate simulations of advanced devices solutions are available so far to meet the possible. For instance, the effect of the high-k requirements of the 22 nm node and beyond. gate dielectric stack on device performance must be addressed with a particular attention From the point of view of technology to its impact on carrier transport properties. development support, the Monte Carlo This is definitely one of the most challenging simulators should be able to provide reliable issues in semiconductor industry at present. electrical results on a regular basis for 22 nm MOS devices. However the need for full-band Efficient modelling tools, as well as accurate Monte-Carlo codes together with physics highlights, would certainly bring a bandstructure solvers that account for strain significant competitive advantage for the and are capable of dealing with new materials development and the optimization of the must be highlighted. Indeed, some 22 nm CMOS technology and for future commercial 3D Schroedinger solvers[1] technologies, including molecular devices. combined with NEGF solvers start being available. These solvers can be used to model 4. Importance of modelling variability ballistic quantum transport in advanced devices with strong transverse confinement. Near the end of the current edition of the However, they do not include any inelastic International Technology Roadmap for scattering mechanism, and thus are not Semiconductors (ITRS) in 2018, transistors will suitable for the calculation of transport reach sub-10 nm dimensions [5]. In order to properties in the 22 nm node devices and maintain a good control of the electrical near-future nodes. characteristics, new transistor architectures have to be developed. It is widely recognized On the other hand, high-level device that quantum effects and intrinsic fluctuations simulation tools are at an early stage of introduced by the discreteness of electronic development in universities and research charge and atoms will be major factors institutions. These codes generally include affecting the scaling and integration of such advanced physical models, such as strain- devices as they approach few-nanometer dependent bandstructure and scattering dimensions [6-11]. mechanisms, and should provide accurate predictions in complex nano-systems. For instance, in conventional one-gate However, such simulation tools are in general nanotransistors, variations in the number and difficult to use in an industrial environment, in position of dopant atoms in the active and particular because of a lack of source/drain regions make each nano- documentation, support and graphical user transistor microscopically different from its interface, although an increasing number of neighbors [12-16]. In nanowire MOS academic codes are now including graphic transistors the trapping of one single electron tools [3,4]. Taking advantage of these ongoing in the channel region can change the current research projects, it should be possible to by over 90% [17,18]. Interface roughness of integrate such high-level codes into industrial the order of 1-2 atomic layers introduces TCAD tools or to use them to obtain variations in gate tunneling, quantum calibrated TCAD models useful for the confinement and surface/bulk mobility from industry. Concerning this latter point, the device to device. The inclusion of new quantum drift-diffusion-based solution must 82 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling materials such as SiGe will induce additional new nanodevices. Strategies to decrease the sources of fluctuations associated with amount of naturally occurring disorder or to random variations in the structure, defects, cope with it need to be devised as emerging strain and inelastic scattering [19,20]. These devices are developed into new technologies intrinsic fluctuations will have an important aiming at the limits of the downscaling process. impact on the functionality and reliability of the corresponding circuits at a time when 5. Integration between material fluctuation margins are shrinking due to and device simulation continuous reductions in supply voltage and increased transistor count per chip [7,8]. Both for decananometric MOSFETs and for most emerging devices, the distinction The problem of fluctuations and disorder is between material and device simulation is actually more general and affects getting increasingly blurred, because at low fundamental aspects of information storage dimensional scales the properties of the and processing as device size is scaled down. material sharply diverge from those of the The presence of disorder limits the capability bulk or of a thin film and become strongly of patterning by introducing a spatial dependent on the detailed device geometry. variance: when the pattern size approaches Such a convergence should start being the spatial variance, patterns are unavoidably reflected also in research funding, because, at lost. An analogous problem exists as a result the dimensional scale on which research is of time fluctuations (shot noise) associated currently focused, a project cannot possibly with the granularity of charge: as current take into consideration only one of these two levels are reduced, the signal power aspects. This was not the case until a few decreases faster (quadratically with current) years ago, when a material could be than the shot noise power (linear with investigated within the field of chemistry or current), leading to a progressive degradation material science and then parameters were of the signal-to-noise power ratio. passed on to those active in the field of device physics and design, who would include them Disorder has demonstrated all of its disruptive in their simulation tools. power on nanodevices in the case of single- electron transistors: as a result of their Fortunately, the nanoelectronics simulation extreme charge sensitivity, stray charges, community is not starting from scratch in randomly located in the substrate, are terms of atomic scale materials computations. sufficient to completely disrupt their Computational physics and quantum operation. chemistry researchers have been developing sophisticated programs, some with on the Fluctuations associated with the granularity of order of millions of lines of source code, to charge and spatial disorder are fundamental explicitly calculate the quantum mechanics of roadblocks that affect any effort towards solids and molecules from first principles. handling information on an increasingly small Since quantum mechanics determines the scale. charge distributions within materials, all electrical, optical, thermal and mechanical It is thus of strategic importance to develop properties, in fact any physical or chemical device simulation tools that are capable of property can in principle be deduced from efficiently exploring the extremely large these calculations. parameter space induced by such variability and evaluate the actual performance limits of nanoICT Strategic Research Agenda 83

Annex 1 nanoICT working groups position papers Modeling However, these programs have not been [21]. In particular, so-called order-N or quasi- written with nanoelectronic TCAD needs in order-N methodologies have been developed mind, and substantial theoretical and over the last two decades that, using the computational problems remain before their advantages of such local descriptions, allow application in process and device modeling for the calculation of the electronic structure reaches maturity. However, the coupling of of very large systems with a computational electronic structure theory programs to cost that scales linearly or close to linearly information technology simulations is with the size of the system [22,23]. This is in occurring now, and there is nothing to contrast with traditional methods that suggest this trend will not continue unabated. typically show a cubic scaling with the number of electrons in the system. Order-N schemes Quantum electronic structure codes come in are particularly powerful and robust for essentially two flavors: those using plane insulators and large biomolecules. However, wave expansions (or real-space grids) and the design of efficient and reliable order-N those using basis sets of atomic orbitals to methods for metallic systems is still a span the electronic wave functions. Plane challenge. wave codes are suitable for solid state calculations and have been mainly developed The most widely used codes for ab initio within the Solid State Physics community. simulations of solids and extended systems Codes using atomic orbitals were initially rely on the use of the Density Functional developed within the Quantum Chemistry Theory, rather than on Quantum Chemistry community, although recently have also methods. Many of them have been developed become popular within the Materials Physics in Europe, and some of them are commercial, community. Quantum Chemistry codes rely although their use is mostly limited to the heavily on the expansion of the atomic academic community. Among the commercial orbitals in terms of Gaussian functions. This is code, we can cite the plane-wave codes VASP mainly due to the fact that, with the use of [24] and CASTEP [25], and the Hartree- Gaussians, the four-center-integrals Fock/DFT code CRYSTAL [26] that utilizes associated with Coulomb and exchange Gaussian basis functions. Other widely used interactions become analytic and easy to plane-wave codes are the public domain CP2K calculate. However, within the framework of [27] and Abinit [28]. Abinit is distributed Density Functional Theory, due to non-linear under GNU license and has become a very dependence of the exchange-correlation complete code with a rapidly growing energy and potential with the density, the community of developers. Among the most evaluation of such contributions to the energy popular DFT codes using local atomic orbitals and Hamiltonian has to be necessarily as a basis set we can mention the order-N performed numerically and the advantage of code SIESTA [21,29], which uses a basis set of using Gaussian functions is mainly lost. This numerical atomic orbitals, and Quickstep [30], has opened the route for the use of other that uses a basis set of Gaussian functions. types of localized basis sets optimized to increase the efficiency of the calculations. For One of the reasons why codes using basis sets example, localized basis orbital can be of atomic orbitals have recently become very defined to minimize the range of the popular is that they provide the ideal interactions and, therefore, to increase the language to couple with transport codes efficiency of the calculation, the storage, and based on Non-Equilibrium-Green’s functions the solution of the electronic Hamiltonian (NEGFs) to study transport properties in 84 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling molecular junctions and similar systems. with different degrees of accuracy and detail. Using the local language implicit behind the An interesting intermediate stage between use of atomic orbitals (with tight-binding-like DFT calculations, which take into account the Hamiltonians) is trivial to partition the system full complexity of the materials, and empirical in different regions that can be treated on models, which disregard most of the different footings. For example, it becomes chemistry and structural details of the system, relatively simple to combine information from is provided by the tight-binding approaches. a bulk calculation to describe the electrodes Here a minimal basis of atomic orbitals is used with information obtained from a simulation to describe the electronic structure of the that explicitly considers the active part of the system. As a consequence, only the valence device. Again Europe has taken the lead along and lowest lying conduction bands can be this path. Two of the most popular simulation accurately treated. Traditionally, the hopping tools for ballistic transport using NEGFs and overlap matrix elements were considered combined with DFT have been developed empirical parameters that were adjusted to based on the SIESTA code: tranSIESTA [31] describe the electronic band structure of the and Smeagol [32]. In particular, tranSIESTA material and its variation with strain. This has was developed by a collaboration of Danish proven a quite powerful approach to describe and Spanish research groups and, although complex system like, for example, quantum there is a public domain version that will be dots containing thousands or millions of distributed with the latest version of the atoms [34, 35]. A very interesting variation of SIESTA code (siesta.3.0), it has also given rise these methods has been developed in recent to a commercial simulation package called years: the so-called DFT-tight-binding. Here QuantumWise [33]. the tight-binding parameters (hoppings, overlaps and short-ranged interatomic Electronic structure theory represents the repulsive potentials) are not obtained by lowest and most sophisticated level of fitting to empirical information but they are computation in our simulation hierarchy. At obtained from DFT calculations for simple this level there are many different degrees of systems (mainly diatomic molecules) [36]. The rigor and associated errors in the parameters obtained in this way have proven computations. The most accurate results are to be transferable enough to provide a provided by post-Hartree-Fock Quantum reasonable description of systems like large Chemical methods. Unfortunately, they are molecules and even solids. extremely demanding and not well suited for simulations of solids and condensed matter The ability to treat varying length and time systems. As already mentioned, the most scales, within varying degrees of popular approach to study such systems is approximation, leads to the already Density Functional Theory, which provides a mentioned requirement for a multi-scale good balance between the computational approach to coupled materials/device cost and the accuracy of the results. However, simulation. even at the DFT level, ab initio calculations are computationally too demanding to Although a multi-scale approach is more than perform realistic simulations of devices. ever needed at this stage, parameter Therefore, it is necessary to develop more extraction cannot be performed for a generic approximate methods and, finally, to combine material, but must be targeted for the them in the so-called multi-scale approaches, particular device structure being considered, in which different length scales are described especially for single-molecule transistors. nanoICT Strategic Research Agenda 85

Annex 1 nanoICT working groups position papers Modeling There has to be a closed loop between the a larger portion of material described using atomistic portion of the simulation and the empirical interatomic potentials, and higher-level parts, guaranteeing a seamless everything embedded in an even much larger integration. This convergence between region described using continuum elastic material and device studies also implies that a theory. Although still at its infancy, this kind much more interdisciplinary approach than in of multi-scale simulations has already been the past is needed, with close integration applied very successfully to the study of quite between chemistry, physics, engineering, and, different systems, like biological molecules, in a growing number of cases, biology. To crack propagation in mechanical engineering make an example, let us consider the or combustion processes [37, 38]. This is simulation of a silicon nanowire transistor: certainly one of the most promising routes atomistic calculations are needed to towards the simulation of realistic devices. determine the specific electronic structure for the cross-section of the device being Another example where multi-scale and investigated, then this information can be multi-physics simulations become essential is used in a full-band solver for transport or represented by the effort to merge parameters can be extracted for a simpler and electronics with nanophotonics. The faster transport analysis neglecting interband integration of CMOS circuits and tunneling; then the obtained device nanophotonic devices on the same chip opens characteristics can be used for the definition new perspectives for optical interconnections of a higher-level model useful for circuit (higher band widths, lower-latency links analysis. It is apparent that, for example, the compare with copper wires) as well as for the atomistic simulation is directly dependent on possible role of photons in the data the device geometry, and that, therefore, processing. These involve the modelling of work on the different parts of the simulation “standard” passive components, such as hierarchy has to be performed by the same waveguides, turning mirrors, splitters and group or by groups that are in close input and output couplers, as well as active collaboration. Indeed, the dependence elements, such as modulators and optical between results at different length scales switches. Modelling of ultra-scaled sometimes requires the combination of optoelectronic components involve the different techniques in the same simulation, design of nanometer scale optical not just the use of information from more architectures with new properties that may microscopic descriptions to provide differ considerably from those of their parameters for mesoscopic or macroscopic macroscopic counterparts. models. Following the previous example, the electronic properties of a silicon nanowire are This requires the development of new largely determined by its geometry. However, numerical tools able to describe the geometry of the wire is determined by the electromagnetic interactions and light strain present in the region where it grows, propagation at different length scales. They which in turn is a function of the whole should be able to describe the geometry of the device, its temperature and electromagnetic field from the scale of a few the mechanical constraints applied to it. Thus, light wavelengths (already of the order of the a reliable simulation of such device can whole micro-device) down to the nanometer require, for example, a quantum mechanical scale elements. These new tools should description of the nanowire itself, coupled to include a realistic description of the optical properties including electro- and magneto- 86 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling optical active nanostructures and plasmonic matching of interfacial resonance states. elements which are expected to be key These results triggered experimental activity ingredients of a new generation of active aimed at improving the quality of the growth optoelectronic components. process for the dielectrics and to the actual measurement of the predicted high TMR A mayor challenge of the “multi-physical” values [40,41]. The resulting magnetic tunnel modelling will be to simulate a full nano- junctions have found a very important field of device where electronics, mechanics and application in the readout heads for hard photonics meet at the nanoscale. For disks, allowing an unprecedented increase in instance, the coupling between mechanical the information density on a magnetic vibrations and quantum conductance of single substrate and represent the building block of nanotubes has been recently observed. The magnetic memories, which are likely to be interaction of an optical field with a device one of the main technologies of the future in takes place not only through the the market of non-volatile storage. This electromagnetic properties, but also through further demonstrates the importance of the mechanical response (radiation pressure simulation codes that can provide forces). The physical mechanisms and quantitatively reliable estimates of the possible applications of optical cooling of parameters of interest, instead of mechanical resonators are being explored. approximate models. Modelling Nano-Electro-Mechano-Optical (NEMO) devices is going to play a key (and 6. Carbon-based electronics and fascinating) role in the development and spintronics optimization of new transducers and devices. Amongst the most promising materials for the There are several examples, in the recent development of beyond CMOS history of electron devices, in which physics- nanoelectronics, Carbon Nanotubes & based modeling has played an essential role Graphene-based materials and devices in pointing out new avenues for research or in deserve some particular consideration. explaining unexpected phenomena. One Indeed, first, their unusual electronic and interesting example, which has had also structural physical properties promote carbon important important consequences in terms nanomaterials as promising candidates for a of applications, is that of the prediction of wide range of nanoscience and much improved tunneling magnetoresistance nanotechnology applications. Carbon is (TMR) of magnetic tunnel junctions. Up to a unique in possessing allotropes of each decade ago, the achieved values of TMR in possible dimensionality and, thus, has the structures with ferromagnet-dielectric- potential versatility of materials exhibiting ferromagnet were at most around 70%, different physical and chemical properties. because of the amorphous nature of the Diamond (3D), fullerenes (0D), nanotubes dielectric layer. Accurate first-principles (1D-CNTs), 2D graphene and graphene calculations [39] made it possible to predict ribbons are selected examples. Because of that with crystalline dielectric layers the their remarkable electronic properties, CNTs tunneling process was much more or graphene-based materials should certainly complicated than the previously used simple play a key role in future nanoscale electronics. barrier model, because of the relevance of Not only metallic nanotubes and graphene the symmetry of Bloch states at the Fermi energy and the effect on wave function nanoICT Strategic Research Agenda 87

Annex 1 nanoICT working groups position papers Modeling offer unprecedented ballistic transport ability, proximity effect [47] and also to control the but they are also mechanically very stable and spin polarization of current by a gate voltage strong, suggesting that they would make ideal [48, 49]. This configuration does not make use interconnects in nanosized devices. Further, of any ferromagnetic metallic contact to inject the intrinsic semiconducting character of spin-polarized electrons. Thus, it could be a either nanotubes or graphene nanoribbons, way to circumvent the problem of as controlled by their topology, allows us to “conductivity mismatch” [50-52] which build logic devices at the nanometer scale, as possibly limits the current spin injection already demonstrated in many laboratories. efficiency into a conventional semiconductor In particular, the combination of 2D graphene from a ferromagnetic metal. These for interconnects (charge mobilities in phenomena and the corresponding devices graphene layers as large as 400.000 cm2V-1s-1 need to be investigated using the appropriate have been reported close to the charge models of relativistic-like electron transport in neutrality point) together with graphene 2D graphene structures. Additionally, the nanoribbons for active field effect transistor presence of spin states at the edges of zigzag devices could allow the implementation of GNRS has also been demonstrated [53-55], completely carbon-made nanoelectronics. and may be exploited for spintronics. Using first-principles calculations, very large values Besides the potential of 2D graphene and of magnetoresistance have been predicted in GNRs for electronic device applications, GNR-based spin valves [56,57]. transport functionalities involving the spin of the carriers have very recently received a Additionally, the potential for 3D based particularly strong attention. First, although organic spintronics has been recently spin injection through ferromagnetic suggested by experimental studies [58]. metal/semiconductor interfaces remains a Organic spin valves have shown spin great challenge, the spectacular advances relaxation times in the order of the made in 2007 converting the spin information microsecond and spin tunnel junctions with into large electrical signals using carbon organic barriers have recently shown nanotubes [42] has opened a promising magnetoresistance values in the same order avenue for future carbon-based spintronic of magnitude as that of inorganic junctions applications. The further demonstration of based on Al2O3. However, spin transport in spin injection in graphene [43] and the these materials is basically unknown, and observation of long spin relaxation times and many groups are trying to decipher the lengths [44] have suggested that graphene impressive experimental complexity of such could add some novelty to carbon-based devices. Organic materials, either small spintronics. For instance, taking advantage of molecules or polymers, will definitely allow the long electronic mean free path and large scale and low cost production of negligible spin-orbit coupling, the concept of alternative non volatile memory technologies, a spin field-effect transistor with a 2D with reduced power consumption. These new graphene channel has been proposed with an materials have therefore the potential to expectation of near-ballistic spin transport create an entirely new generation of operation [45]. A gate-tunable spin valve has spintronics devices, and the diverse forms of been experimentally demonstrated [46]. carbon-based materials open novel horizons Finally, a ferromagnetic insulator, such as for further hybridization strategies and all- EuO, may be used to induce ferromagnetic carbon spintronics circuits, including logic and properties into graphene, through the memory devices. 88 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling The performance of these spintronic devices fabrication need to be observed from relies heavily on the efficient transfer of spin different perspectives. Indeed, recent polarization across different layers and experiments in 13C nanotubes reveal interfaces. This complex transfer process surprisingly strong nuclear spin effects that, if depends on individual material properties and properly harnessed, could provide a also, most importantly, on the structural and mechanism for manipulation and storage of electronic properties of the interfaces quantum information. between the different materials and defects that are common in real devices. Knowledge This may help to overcome the performance of these factors is especially important for the limitations of conventional materials and of relatively new field of carbon based the conventional technology for spin valve spintronics, which is affected by a severe lack devices. The real potential of graphene-based of suitable experimental techniques capable materials for FET and related spintronics of yielding depth-resolved information about applications thus requires advanced the spin polarization of charge carriers within modelling methods, including ab initio buried layers of real devices. techniques, and a precise description of spin degree of freedom. In that perspective, it is noteworthy to remark that the fantastic development of first To date, the development of nanotubes and principles non-equilibrium transport methods is progressively allowing for more and more graphene science have been strongly driven realistic assessment and anticipation on the true spintronics potential of carbon-based by theory and quantum simulation [60,61]. structures. This aspect also stands as an essential point for providing guidance and The great advantage of carbon-based interpretation schemes to experimental groups. As a matter of illustration, a few years materials and devices is that, in contrast to ago it has been theoretically shown that organic spin valves, obtained by sandwiching their silicon-based counterparts, their an organic molecule between magnetic contacts, could manifest large bias-dependent quantum simulation can be handled up to a magnetoresistance, provided a suited choice of molecules and anchoring groups was made, very high level of accuracy for realistic device which is now confirmed by experiments [59]. structures. The complete understanding and Finally one also notes that in addition to the potential for GMR in carbon-based materials, further versatile monitoring of novel forms of the spin manipulation and the realization of spin Qubits deserves a genuine consideration. chemically-modified nanotubes and graphene Recent theoretical proposals have shown that spin Qubits in graphene could be coupled will however lead to an increasing demand for over very long distances, as a direct consequence of the so-called Klein paradox more sophisticated computational inherent to the description of charge excitations in terms of massless Dirac approaches, combining first principles results fermions. The related challenges for device with advanced order N schemes to tackle material complexity and device features, as developed in some recent literature [62]. 7. The molecular scale and the end of the road Molecular electronics research continues to explore the use of single molecules as electron devices or for even more complex functions such as logic gates [63]. Experimentally and theoretically the majority of research work focuses on single molecules nanoICT Strategic Research Agenda 89

Annex 1 nanoICT working groups position papers Modeling between two metallic electrodes or on screening can be included using a simple non- molecular tunnel junctions. local self-energy model that basically contains image charge interactions that affect Reproducibility of the measurements and differently the HOMO and LUMO levels [70]. accurate predictions for currents across single molecule tunnel junctions remains a Another interesting issue is that of the challenging task, although the results of coupling of the electrons with structural theory and experiment are converging [64]. deformations and vibrational modes. Within To achieve the goal of using molecular the framework of NEGF+DFT calculations components for computing or storage, or inelastic effects associated with the excitation indeed for novel functions, requires of localized vibration within the contact refinement of the theoretical techniques to region have been successfully included in better mimic the conditions under which recent years. Most of the approaches rely on most experiments are performed and to more different perturbative expansions or on the accurately describe the electronic structure of so-called self-consistent Born approximation the molecules connected to the leads. [71]. This allows for accurate simulations of the inelastic signal in molecular junctions and, Most simulations of transport in molecular therefore, the characterization of the path junctions to date are based on the followed by the electrons by identifying which combination of the non-equilibrium Green’s vibrations are excited during the conduction functions techniques with DFT calculations process. Unfortunately, the self-consistent [65,66]. Although this approach has proven Born approximation does not allow quite powerful, it also presents important accounting for polaronic effects that are shortcomings. In particular, the transport crucial to understand the electronic transport calculation is based on the Kohn-Sham in flexible organic molecules [72]. In these spectrum as calculated using standard local or cases more sophisticated methods are semi-local exchange-correlation functionals. necessary, the coupling of which to ab initio These functionals usually give a reasonable DFT simulations is still an open question. description of the electronic spectrum for the normal metals that constitute the leads. Hence much of the work to date has focused However, they are known to be much less on the underlying physical mechanisms of reliable to predict the energy spectrum of charge transport across molecules, whereas small molecules. This is extremely important, very little is understood in terms of the use of since the relative position of the molecular molecular components in complex, or even levels and the Fermi energy of the leads is simple, circuits. This research area could also crucial to determine the transport be categorized as in its infancy in that very characteristic. Some of the deficiencies of DFT little is known about time-dependent or AC to describe localized levels can be corrected responses of molecules in tunnel junctions or by including the so-called self-interaction other circuit environments. To exploit corrections [67]. Still, the position of the molecules in information processing, the use molecular levels is also strongly influenced by of multi-scale tools [73] as described the dynamical screening induced by the previously are needed to embed molecular metallic leads [68,69]. An accurate description scale components between what are of all these effects requires more elaborate essentially classical objects: leads, drivers, and theoretical treatments, beyond standard DFT circuits. Further development of the calculations. Fortunately, part of the effects of simulations is needed to accurately describe 90 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling the transport processes in molecular junctions dependent response of molecules to external and its coupling to structural degrees of voltages and their interaction with light. As freedom and, in particular, to molecular the size of molecules considered as a vibrations, also the time-dependent response component remains quite large, adapted of the molecules to external voltages and methods using a scattering approach seem to their interaction with light needs to be be more relevant than high-level NEGF or studied. other sophisticated methods [75,76]. The concept of a molecular logic gate has In single-molecule transport treatment of the emerged recently and specific approaches leads is not well-established and difference of including quantum Hamiltonian computing orders of magnitude in the current can be [77] will be used for a numerical integration. found among different methods. Full Configuration-Interaction is often considered 8. Nano-bio-electronics to be the reference, but its application to practical devices is not currently possible, due A further fundamental research line which to the almost combinatorial growth of the emerged in parallel to the development of size of the problem with the number of molecular electronics is bio-electronics. It particles present in the system. mainly aims at (i) mimicking of biological and biophysical processes via molecular electronic In order to achieve reliable models all circuits, and (ii) the exploitation of self- quantities of interest should be computed in a assembling and self-recognition properties of single consistent framework, otherwise there many biomolecules as scaffolds in the is a risk of having quite different physical integration of nano- and sub-nanoscale descriptions of the various regions of a devices, and (iii) exploring the potential of device. arrays of biomolecules (like DNA and its artificial modifications) to serve as molecular Also assuming to be able to treat each domain wires interconnecting different parts of with an appropriate model, a significant molecular electronic devices. Obviously, great problem is represented by the description of challenges from both the experimental and the interface between different domains and from the theoretical side exist on the road to proper treatment of the coupling. The real achieve such goals. The theoretical difficulty is about boundaries and about how understanding of the bio/inorganic interface to handle them fast. is in its infancy, due to the large complexity of the systems and the variety of different Furthermore, there are several fundamental physical interactions playing a dominant role. problems that are still open: for example, the Further, state of the art simulation techniques integration between approaches such as DFT for large biomolecular systems are to a large (which is a ground state, equilibrium theory) degree still based on classical physics with non-equilibrium Green's functions. approaches (classical molecular dynamics, classical statistical physics); while this can still Finally, to position a single molecule at the provide valuable insight into many right place, experimental equipment has to be thermodynamical and dynamical properties of developed with accurate manipulation biomolecular systems, a crucial point is capabilities, as well as precision electrical nevertheless missing: the possibility to obtain probes for four-terminal measurements [74]. Further development of the simulation techniques is needed to describe the time- nanoICT Strategic Research Agenda 91

Annex 1 nanoICT working groups position papers Modeling information about the electronic structure of makes comparison with results for ideal the biomolecules, an issue which is essential materials difficult. in order to explore the efficiency of such systems to provide charge migration Furthermore large energy ranges are pathways. Moreover, due to the highly involved, for the particular applications of dynamical character of biomolecular systems these materials, which makes calculation of -seen e.g. in the presence of multiple time many parameters difficult: for example, it is scales in the atomic dynamics- the electronic quite hard to evaluate electron-phonon structure is strongly entangled with structural scattering at 10 eV. In addition, a fluctuations. We are thus confronting the semiclassical model is not easily applicable, problem of dealing with the interaction of due to the complexity of the electronic strongly fluctuating complex molecules with structure: this is similar to problems that are inorganic systems (substrates, etc). As a becoming relevant also in silicon, in the study result, multi-scale simulation techniques are of nanowires. Because of the type of crystal, urgently required, which should be able to the phonon model is much more complex combine quantum-mechanical approaches to than in silicon. Also the application of the electronic structure with molecular standard approaches to quantum transport is dynamical simulation methodologies dealing rather challenging. with the complex conformational dynamics of biological objects. Conventional simulation The typical concentration of dislocations is of tools of semiconducting microdevices can 108 per cm2 (a value that can in principle be obviously not deal with such situations. reduced, but most probably not below 106 per cm2) which makes it look like a miracle that 9. Simulation of wide-bandgap devices based on such materials actually semiconductors work. It is important to understand the role of extended defects in term of the optical The market for wide-bandgap materials is not properties and, since the defects cannot be as large as that for silicon, but they are eliminated, it is necessary to design around becoming important in the growing LED them, somehow including their effect into market, as demonstrated by the significant simulation tools. interest expressed by Applied Materials, as well as in high-power, high-frequency In terms of the development of devices, one applications. of the unsolved issues is to understand why p- type GaN is so hard to achieve. There is a Simulation of wide-bandgap semiconductors concrete opportunity to introduce much of is a task characterized by challenges of its what the ab-initio community has been own, which are often quite different from developing, in particular in terms of large- those typical of modeling silicon, due to the scale atomistic models. different crystal structure, the possibility of engineering the bandgap, the variability of Another relevant problem is the power material properties with orientation. It is also density: GaN HFETs exhibit a power density difficult to validate the results of the that is much larger than that of conventional simulations because such materials are often devices, therefore cooling techniques are an characterized by a high defectivity, which important field of research. 92 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling 10. Thermoelectric energy conversion affected from surface roughness in these wires. Similar behavior is also reported for The importance of research on thermoelectric graphene nanoribbons where edge disorder energy conversion is growing in parallel with reduces lattice thermal conductivity while the need for alternative sources of energy. electrons in the first conduction plateau stay With the recent developments in the field, almost intact [84]. thermoelectric energy generators have become a commercial product in the market At the sub-band edges, however, the and their efficiencies are improving electronic mean free path is discontinuous constantly, but the commercially available which yields a large increase of the Seebeck products did not take the advantage of nano- coefficient and ZT. Another criteria for high ZT technology yet. In fact, thermoelectricity is is to have a narrow distribution of the energy one of the areas in which nano-scale of the electrons participating in the transport fabrication techniques offer a breakthrough in process [85]. device performances. In molecular junctions, it is shown that the It was predicted theoretically that, lowering Seebeck coefficient is maximized when the the device dimensions, it is possible to electrode-molecule coupling is weak and the overcome the Wiedemann-Franz Law and HOMO level is close to the Fermi energy, enhance the device performances these fulfill Mahan's criterion in the zero- significantly [78]. dimensional case [86]. The weak coupling condition is also in favor of the need of Quasi one-dimensional quantum wires [79], reduced vibrational thermal conduction. engineered molecular junctions [80,81], Therefore self-assembled layers of molecules superlattices of quantum dots [82] are the between semiconducting surfaces are other possible routes proposed for achieving supposed to yield high ZT values. a high thermoelectric figure of merit, ZT. These developments offer new challenges for The thermoelectric figure of merit includes increasing device performances and also for three properties of the material, namely the the modelling of new devices. An accurate electrical conductivity σ, Seebeck coefficient modelling of the disordered structures or of S, the thermal conductivity κ=κel +κph with the layered structures of self-assembled electronic and phononic contributions as well molecules between surfaces requires a lot of as the temperature T, ZT =σ S2 T/ κ . improvements both from the methodological and from the computational point of view. In order to optimize ZT, an electron-crystal Simulation of these systems entails together with a phonon-glass behavior is consideration of very large number of atoms, required [83]. of the order of 106 to 108. Therefore fully parallelized order-N methods for both Indeed, it has recently been shown that Si electronic and phononic computations are nano-wires with rough surfaces can serve as needed. Though important progress has been high performance thermoelectric materials, achieved in order-N electronic calculations, since the edge roughness suppresses phonon such methods are still lacking a proper transport by a few orders of magnitude description for the phonons. Adoption of the whereas the electronic transport is weakly methods already developed for electrons to phonons is a first goal to be achieved. nanoICT Strategic Research Agenda 93

Annex 1 nanoICT working groups position papers Modeling 11. Multifunctional oxides example, we still lack a robust room- temperature magnetoelectric multiferroic Multifunctional oxides, ranging from system that can be integrated in real devices, piezoelectrics to magnetoelectric multiferroics, we still have to understand the main factors offer a wide range of physical effects that can controlling the performance of a ferroelectric be used to our advantage in the design of novel tunnel junction, etc. Finally, it should be nanodevices. For example, these materials stressed that the field is rich in opportunities make it possible to implement a variety of for the emergence of novel effects and tunable and/or switchable field effect devices concepts that go beyond the current at the nanoscale. Thus, a magnetoelectric prospects in the general area of electronics, multiferroic can be used to control the spin the recent discovery of high electronic polarization of the current through a magnetic mobilities in all-oxide heterostructures (i.e., at tunnel junction by merely applying a voltage; the interfaces between LaAlO3 and SrTiO3) or a piezoelectric layer can be used to exert being a remarkable example. very well controlled epitaxial-like pressures on the adjacent layers of a multilayered Quantum-mechanical simulation is playing a heterostructure, which e.g. can in turn trigger a key role in the progress in multifunctional magnetostructural response. These are just oxides. This has been historically the case, two examples of many novel applications that with many key contributions from the first- add up to the more traditional ones - as principles community to the understanding of sensors, actuators, memories, highly-tunable ferroelectric, magnetic and magnetoelectric dielectrics, etc. - that can now be scaled down bulk oxides [87,88]. This trend is just getting to nanometric sizes by means of modern stronger in this era of nanomaterias, for two deposition techniques. In fact, nanostructuring reasons: (i) there is greater need for theory to of different types, ranging from the explain the novel physical mechanisms at construction of oxide nanotubes to the more work in systems that are usually very difficult traditional multi-layered systems, is opening to characterize experimentally, and (ii) the door to endless possibilities for the modern deposition techniques offer the engineering/combination of the properties of unique chance to realize in the laboratory the this type of compounds, which are strongly most promising theoretical predictions for dependent on the system's size and (electrical, new materials. Thus, the importance of first- mechanical) boundary conditions. principles simulation for fundamental research in functional oxides is beyond doubt, The current challenges in the field are plenty, and very recent developments, e.g. for the from the more technical to the more first-principles study of the basic physics of fundamental. At the level of the applications, magnetoelectrics [89], ferroelectric tunnel the outstanding problems include the junctions [90] or novel oxide superlattices integration of these materials with silicon or [91], clearly prove it. The contribution from the identification of efficient and scalable simulations to resolve more applied problems growth techniques for complex oxide (e.g, that of the integration with silicon) is, on heterostructures. At a more fundamental the other hand, just starting, and its progress level, there is a pressing need to identify new will critically depend on our ability to develop systems and/or physical mechanisms that can novel multi-scale simulation methods that can materialize some of the most promising tackle the kinetics of specific growth concepts for the design of devices; for processes (from pulse-laser deposition to sputtering methods under various oxygen 94 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling atmospheres) directly. This is a major to provide accurate approximations for the challenge that will certainly generate a lot of many body problem and the computation of activity in the coming couple of decades. higher moments in time-dependent scenarios [101,102,103]. Unfortunately, because of 12. Simulation of the time-dependent their computational burden, these proposals behavior have been only used to predict noise and time-dependent behavior of very simple and Although most of the present quantum idealized quantum devices. simulators provide very accurate information about the DC behavior, they are unable to As happened for the microelectronic industry reliably predict the transient, AC or noise some time ago, in the next future, the properties of quantum devices in most nanoelectronic industry will ask for realistic practical operating conditions. There are predictions about the time-dependent and mainly two reasons that explain why such noise behavior of these novel quantum predictions are so difficult. First, at high- devices. Thus, an important effort must be frequency, one has to deal with the role of the done by the scientific community to develop “displacement current” that requires a accurate and versatile quantum simulators reasonable approximation for the (time- providing information beyond the DC dependent or frequency-dependent) Coulomb predictions that are interesting enough for interaction among electrons [92,93]. Second, in the nanoelectronic industry, in terms of order to compute correlations (the second or accuracy, versatility and computational higher moments of the electrical current for resources. instance) one has to measure at least at two different times. One has to take into account In this respect, multiscale modeling will be that a first measurement “collapses” the essential, because the time-dependent evolution of the quantum system and, behavior of a device or circuit involves the consequently, it will affect the outcome of a interaction of a large number of particles second (or further) measurement [94]. (charge carriers), but at the same time is influenced by detailed quantum interactions. There are some elegant theoretical proposals Therefore it requires a treatment based on a in the literature that show the path to include hierarchy of models, with a proper time-dependence and correlations into microscropic description of particle practical quantum simulators. Among others, correlations, a semiclassical transport model we cite the work of Büttiker and coworkers, within the potential landscape of the device, who generalized the successful (DC) Landauer and a higher-level description for a circuit [95,96] model towards current fluctuations made up of more than one device. and AC conductances [97,98,99]. The work of Levitov, Lesovik and others in the application 13. New computational approaches of Full Counting Statistics in mesoscopic devices for the computation of higher Although the number of transistors per unit moments of the electrical current [92,93,100]. area still follows Moore’s law, i.e. it doubles Recently, a novel proposal that treats every 18 months, and individual transistor quantum transport with quantum (Bohmian) dimensions are scaled down accordingly, trajectories has also demonstrated its ability thereby decreasing the switching delays, it is not any more possible to make classical nanoICT Strategic Research Agenda 95

Annex 1 nanoICT working groups position papers Modeling general-purpose processors with clock only of single-precision operations, but more frequency scaling in proportion, due to the recent generations of GPUs, such as the limitations in power dissipation and heat FireStream 9170 by AMD, can handle double removal. precision in hardware, although at a somewhat reduced rate (possibly by a factor This has halted the progressive improvement of 5). Another impressive feature of GPU in single-CPU performance that was typical of computation is the extremely high energy the 1980’s and 1990’s, leading to a change of efficiency, of the order of 5 Gigaflops/W in paradigm in the quest for ever increasing single precision or 1 Gigaflop/W in double computational power, with the introduction precision, an aspect of growing importance of multi-core chips. considering the costs for supplying power and air conditioning to computing installations. This type of parallelism reaches its most extreme expression Graphic Processor Units The latest GPU hardware opens really new (GPUs), which are special purpose processors, perspectives for simulation of nanodevices, originally designed for performing graphic also in \"production environments,\" because tasks. Typically, GPUs offer a peak GPU based systems could be easily performance that is an order of magnitude standardized and provided to end-users along larger than that of general purpose with the simulation software. processors. However, not all algorithms can use efficiently this additional performance, The latest supercomputers are based on a because GPUs are designed for massive well-balanced mix of GPUs and CPUs, as in the parallel vector operations. case of the new Cray hybrid supercomputer, the XK6, which will be based on a Indeed GPUs offer large-scale parallelism at combination of multi-core AMD CPUs and the low cost afforded by mass production; NVIDIA GPUs, scalable up to 500,000 cores. their processing units are specifically designed OPENACC, which is being implemented into to perform dense matrix operations several compilers, will be operable on (fundamental for 3D graphic rendering), multiple platforms and should allow a very which do represent a very significant portion straightforward acceleration of legacy codes. within many modelling codes. This means that large speedups are to be expected, and some 14. Major current deficiencies of TCAD initial tests have demonstrated factors up to 40-50. A clear gap, which has in part been addressed in the previous sections, has formed in the Modern Graphic Processing Units (GPU) last few years between what is available in approach a peak performance of a Teraflop the TCAD market and what would actually be (1012) floating point operations per second) needed by those working on the development thanks to a highly parallel structure and to an of advanced nanoscale devices. As already architecture focusing specifically on data mentioned, classical TCAD tools have not processing rather than on caching or flow been upgraded with realistic quantum control. This is the reason why GPUs excel in transport models yet, suitable for the current applications for which floating point 22 nm node in CMOS technology or for performance is paramount while memory emerging technologies, and, in addition, there bandwidth in not a primary issue. are some fields of increasing strategic Initially the main disadvantage was represented by the availability, in hardware, 96 nanoICT Strategic Research Agenda

Annex 1 nanoICT working groups position papers Modeling importance, such as the design of 15. Overview of networking for photovoltaic cells, for which no well- modelling in Europe and the established TCAD platforms exist. United States Bottom-up approaches, which, if successful, In the United States, the network for could provide a solution to one of the major computational nanotechnology (NCN) is a six- bottlenecks on the horizon, i.e. skyrocketing university initiative established in 2002 to fabrication costs, are not supported by any connect those who develop simulation tools type of TCAD tools as of now. This may be due with the potential users, including those in to the fact that bottom-up approaches are academia, and in industries. The NCN has still in their infancy and have not been received a funding of several million dollars demonstrated in any large-scale application, for 5 years of activity. One of the main tasks but the existence of suitable process of NCN is the consolidation of the simulation tools could, nevertheless, facilitate nanoHUB.org simulation gateway, which is their development into actual production currently providing access to computational techniques. Sophisticated tools that have codes and resources to the academic been developed within research projects are community. According to an NCN survey, the available on the web, mainly on academic total number of users of nanoHUB.org sites, but they are usually focused on specific reached almost 70,000 in March 2008, with problems and with a complex and non- more than 6,000 users having taken standardized user interface. advantage of the online simulation materials. The growth of the NCN is likely to attract Atomistic simulations are becoming a relevant increasing attention to the US computational part of device modelling, but there exists a nanotechnology platform from all over the discrepancy between what the industry world, from students, as well as from actually needs and what is being provided by academic and, more recently, industrials the academia. The industry would like to have researchers. In Europe an initiative similar to fast models, that quickly provide answers, the nanoHUB, but on a much smaller scale, while the academia has developed very was started within the Phantoms network of advanced models, which are on the average excellence (http://vonbiber.iet.unipi.it) and slower than previous, less sophisticated, has been active for several years; it is tools, and require much more CPU power. A currently being revived with some funding partial solution to this specific problem is within the NanoICT coordinated action. represented by the new hardware which has become available, such as that based on In a context in which the role of simulation GPUs, which in the near future could make might become strategically relevant for the advanced models practically usable on development of nanotechnologies, molecular desktop machines. nanosciences, nanoelectronics, nanomaterial science and nanobiotechnologies, it seems An effort would be needed to coordinate the urgent for Europe to set up a computational research groups working on the development platform infrastructure similar to NCN, in of the most advanced simulation approaches, order to ensure its positioning within the the TCAD companies and the final users, in international competition. The needs are order to define a common platform and create manifold. First, a detailed identification of the basis for multi-scale tools suitable to support the development of nanoelectronics in the next decade. nanoICT Strategic Research Agenda 97

Annex 1 nanoICT working groups position papers Modeling European initiatives and networks must be tools such as TCAD models that are of crucial performed, and de-fragmentation of such importance in software companies. Many fields activities undertaken. A pioneer initiative has such as organic electronics, spintronics, beyond been developed in Spain through the M4NANO CMOS nanoelectronics, nanoelectromechanical database (www.m4nano.com) gathering all devices, nanosensors, nanophotonics devices nanotechnology-related research activities in definitely lack standardized and enabling tools modelling at the national level. This Spanish that are however mandatory to assess the initiative could serve as a starting point to potential of new concepts, or to adapt extend the database to the European level. processes and architectures to achieve the Second, clear incentives need to be launched desire functionalities. The European excellence within the European Framework programmes in these fields is well known and in many to encourage and sustain networking and aspects overcomes that of the US or of Asian excellence in the field of computational countries. Within the framework of a new nanotechnology and nanosciences. To date, no initiative, specific targets should be addressed structure such as a Network of Excellence in relation with the modelling needs reported exists within the ICT programme, although the by small and medium sized software programme NMP supported a NANOQUANTA companies active in the development of NoE in FP6, and infrastructural funding has commercial simulation tools, such as been provided to the newly established ETSF QUANTUM WISE (www.quantumwise.com), (European Theoretical Spectroscopy Facility, SYNOPSYS (www.synopsys.com), NANOTIMES www.etsf.eu). This network mainly addresses (www.nanotimes-corp.com), SILVACO (www. optical characterization of nanomaterials, and silvaco.com), NEXTNANO3 (www. nextnano. provides an open platform for European users, de), TIBERCAD (www.tibercad.org). that can benefit from the gathered excellence and expertise, as well as standardized Similarly, larger companies such as computational tools. There is also a STMicroelectronics, Philips, THALES, IBM, coordinated initiative focused on the specific INTEL make extensive usage of commercial topic of electronic structure calculations, the simulation tools to design their technological Psi-k network (www.psi-k.org). processes, devices and packaging. The sustainable development of the An initiative similar to the American NCN would computational simulation software industry, be needed in Europe, within the ICT including innovative materials (carbon programme that encompasses the broad fields nanotubes, graphene, semiconducting of devices and applications or, better, in nanowires, molecular assemblies, organics, conjunction between the ICT and the NMP magnetic material) and novel applications programme, since the full scope from materials (spintronics, nanophotonics, beyond CMOS to devices and circuits should be addressed. nanoelectronics), could therefore be crucial to foster industrial innovation in the next 16. Past, present and future decade. European advances in computational approaches 17. Conclusions This novel initiative should be able to bridge Recent advances in nanoscale device advanced ab-initio/atomistic computational technology have made traditional simulation approaches to ultimate high-level simulation approaches obsolete from several points of 98 nanoICT Strategic Research Agenda