Graphene photonics and optoelectronics

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

611

E

lectrons propagating through the bidimensional struc-ture of graphene have a linear relation between energy and

momentum, and thus behave as massless Dirac fermions 1–3. Consequently, graphene exhibits electronic properties for a two-dimensional (2D) gas of charged particles described by the relativistic Dirac equation, rather than the non-relativistic Schr?dinger equation with an eff ective mass 1,2, with carriers mim-icking particles with zero mass and an eff ective ‘speed of light’ of

around 106 m s –1

.

Graphene exhibits a variety of transport phenomena that are characteristic of 2D Dirac fermions, such as specifi c integer and

fractional quantum Hall eff ects 4,5

, a ‘minimum’ conductivity of ~4e 2

/h even when the carrier concentration tends to zero 1, and Shubnikov–de Haas oscillations with a π phase shift due to Berry’s

phase 1. Mobilities (μ) of up to 106 cm 2 V –1 s –1

are observed in sus-pended samples. Th is, combined with near-ballistic transport at room temperature, makes graphene a potential material for nano-electronics 6,7, particularly for high-frequency applications 8.

Graphene also shows remarkable optical properties. For exam-ple, it can be optically visualized, despite being only a single atom thick 9,10. Its transmittance (T ) can be expressed in terms of the fi ne-structure constant 11. Th e linear dispersion of the Dirac elec-trons makes broadband applications possible. Saturable absorption is observed as a consequence of Pauli blocking 12,13, and non-equilibrium carriers result in hot luminescence 14–17. Chemical and physical treatments can also lead to luminescence 18–21. Th ese prop-erties make it an ideal photonic and optoelectronic material.

E lectronic and optical properties Ele ctronic prope rtie s. Th e electronic structure of single-layer graphene (SLG) can be described using a tight-binding Hamiltonian 2,3

. Because the bonding and anti-bonding σ-bands are well separated in energy (>10 eV at the Brillouin zone centre Γ), they can be neglected in semi-empirical calculations, retaining only the two remaining π-bands 3. Th e electronic wavefunctions from diff erent atoms on the hexagonal lattice overlap. However, any such overlap between the p z (π) and the s or p x and p y orbit-als is strictly zero by symmetry. Consequently, the p z electrons, which form the π-bonds, can be treated independently from the other valence electrons. Within this π-band approximation it is

easy to describe the electronic spectrum of the total Hamiltonian and to obtain the dispersion relations E ± (k x , k y ) restricted to first-nearest-neighbour interactions only:

G raphene photonics and optoelectronics F. Bonaccorso, Z. Sun, T. Hasan and A. C. Ferrari*

The richness of optical and electronic properties of graphene attracts enormous interest. Graphene has high mobility and optical transparency, in ad d ition to ? exibility, robustness and environmental stability. So far, the main focus has been on fundamental physics and electronic devices. However, we believe its true potential lies in photonics and optoelectronics, where the combination of its unique optical and electronic properties can be fully exploited, even in the absence of a bandgap, and the linear dispersion of the Dirac electrons enables ultrawideband tunability. The rise of graphene in photonics and optoelectronics is shown by several recent results, ranging from solar cells and light-emitting devices to touch screens, photodetectors and ultrafast lasers. Here we review the state-of-the-art in this emerging ? eld.

E +–(k x ,k y

)=+?γ01 + 4 cos cos + 4 cos 23k x a 2√√k y a 2k y a 2

(1)

where a = √3 a cc (with a cc = 1.42 ? being the carbon–carbon

distance) and γ0 is the transfer integral between fi rst-neighbour π-orbitals (typical values for γ0 are 2.9–3.1 eV). Th e k = (k x , k y ) vec-tors in the fi rst Brillouin zone constitute the ensemble of available electronic momenta.

With one p z electron per atom in the π–π* model (the three other s , p x , p y electrons fi ll the low-lying σ-band), the (–) band (negative energy branch) in equation (1) is fully occupied, whereas the (+) branch is totally empty. Th ese occupied and unoc-cupied bands touch at the K points. Th e Fermi level E F is the zero-energy reference, and the Fermi surface is defi ned by K and K ′. Expanding equation (1) at K (K ′) yields the linear π- and π*-bands for Dirac fermions:

E ±(κ) = ±?ν

F |κ| (2)

where κ = k – K and νF is the electronic group velocity, which is given by νF = √3 γ0a /(2?) ≈ 106 m s –1.Th e linear dispersion given by equation (2) is the solution to the following eff ective Hamiltonian at the K (K ′) point H = ±?νF (σ ? κ), where κ = –i ? and σ are the pseudo-spin Pauli matrices operating in the space of the electron amplitude on the A–B sublattices of graphene 2,3.

L ine ar optical absorption. Th e optical image contrast can be

used to identify graphene on top of a Si/SiO 2 substrate (Fig. 1a)9. Th is scales with the number of layers and is the result of interfer-ence, with SiO 2 acting as a spacer. Th e contrast can be maximized by adjusting the spacer thickness or the light wavelength 9,10. Th e transmittance of a freestanding SLG can be derived by applying the Fresnel equations in the thin-fi

lm limit for a material with a fi xed universal optical conductance 22 G 0 = e 2/(4?) ≈ 6.08 × 10–5

Ω –1, to give: T = (1 + 0.5πα)–2 ≈ 1 – πα ≈ 97.7% (3)where α = e 2/(4πε0?c ) = G 0/(πε0c ) ≈ 1/137 is the fi ne-structure con-stant 11

. Graphene only refl ects <0.1% of the incident light in the

visible region 11

, rising to ~2% for ten layers 9. Th us, we can take

Department of Engineering, University of Cambridge, Cambridge CB3 0FA, UK. *e-mail: acf26@16a9398c83d049649b6658c7

? 2010

612

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

the optical absorption of graphene layers to be proportional to the number of layers, each absorbing A ≈ 1 – T ≈ πα ≈ 2.3% over the visible spectrum (Fig. 1b). In a few-layer graphene (FLG) sample, each sheet can be seen as a 2D electron gas, with little perturba-tion from the adjacent layers, making it optically equivalent to a superposition of almost non-interacting SLG 9. Th e absorption spectrum of SLG is quite fl at from 300 to 2,500 nm with a peak in the ultraviolet region (~270 nm), due to the exciton-shift ed van Hove singularity in the graphene density of states. In FLG, other absorption features can be seen at lower energies, associated with interband transitions 23,24.

Saturable

absorption. Interband excitation by ultrafast opti-cal pulses produces a non-equilibrium carrier population in the valence and conduction bands (Fig. 1c). In time-resolved experi-ments 25, two relaxation timescales are typically seen: a faster one of ~100 fs that is usually associated with carrier–carrier intraband collisions and phonon emission, and a slower one, on a picosec-ond timescale, which corresponds to electron interband relaxation and cooling of hot phonons 26,27.Th e linear dispersion of the Dirac electrons implies that for any excitation there will always be an electron–hole pair in resonance. A quantitative treatment of the electron–hole dynamics requires the solution of the kinetic equation for the electron and hole distri-bution functions, f e (p ) and f h (p ), p being the momentum counted from the Dirac point 13. If the relaxation times are shorter than the pulse duration, then during the pulse the electrons reach a station-ary state, and collisions put electrons and holes into thermal equi-librium at an eff ective temperature 13. Th

e populations determine electron and hole densities, total energy density and a reduction o

f photon absorption per layer, due to Pauli blocking, by a factor of

ΔA /A = [1 – f e (p )][1 – f h (p )] – 1. Assuming effi cient carrier–carrier relaxation (both intraband and interband) and effi cient cooling of the graphene phonons, the main bottleneck is energy transfer from electrons to phonons 13.

For linear dispersions near the Dirac point, pair–carrier colli-sions cannot lead to interband relaxation, thereby conserving the total number of electrons and holes separately 13,28. Interband relax-ation by phonon emission can occur only if the electron and hole energies are close to the Dirac point (within the phonon energy). Radiative recombination of the hot electron–hole population has also been suggested 14–17. For graphite fl akes, the dispersion is quadratic and pair–carrier collisions can lead to interband relaxa-tion. Th us, in principle, decoupled SLG can provide the highest saturable absorption for a given amount of material 13.

L umine sce nce . Graphene could be made luminescent by inducing a bandgap, following two main routes. One is by cutting it into ribbons and quantum dots, the other is by chemical or physical treatments, to reduce the connectivity of the π-electron network. Even though graphene nanoribbons have been produced with varying bandgaps 7, as yet no photoluminescence has been reported from them. However, bulk graphene oxide dispersions and solids do show a broad photo-luminescence 19–21,29. Inpidual graphene fl akes can be made brightly luminescent by mild oxygen plasma treatment 18. Th e resulting pho-toluminescence is uniform across large areas, as shown in Fig. 1d, in which a photoluminescence map and the corresponding elastic scat-tering image are compared. It is possible to make hybrid structures by etching just the top layer, while leaving underlying layers intact 18. Th is combination of photoluminescent and conductive layers could be used in sandwich light-emitting diodes. Luminescent graphene-based materials can now be routinely produced that cover the infra-red, visible and blue spectral ranges 18–21,29.

Even though some groups have ascribed photoluminescence in graphene oxide to bandgap emission from electron-confi ned sp 2 islands 19–21, this is more likely to arise from oxygen-related defect states 18. Whatever the origin, fl uorescent organic compounds are of importance to the development of low-cost optoelectronic devices 30. Blue photoluminescence from aromatic or olefi nic molecules is particularly important for display and lighting applications 31. Luminescent quantum dots are widely used for bio-labelling and bio-imaging. However, their toxicity and potential environmental hazard limit widespread use and in vivo applications. Fluorescent bio-compatible carbon-based nanomaterials may be a more suit-able alternative. Fluorescent species in the infrared and near-infrared are useful for biological applications, because cells and tissues show little auto-fl uorescence in this region 32. Sun et al. exploited photoluminescent graphene oxide for live cell imaging in the near-infrared with little background 20.

Wang et al . have reported a gate-controlled, tunable gap up to 250 meV in bilayer graphene 23. Th is may make new photonic devices possible for far-infrared light generation, amplifi

cation and detection.

Broadband nonlinear photoluminescence is also possible follow-ing non-equilibrium excitation of untreated graphene layers (Fig. 1c), as recently reported by several groups 14–17. Emission occurs through-out the visible spectrum, for energies both higher and lower than the exciting one, in contrast with conventional photoluminescence proc-esses 14–17. Th is broadband nonlinear photoluminescence is thought to result from radiative recombination of a distribution of hot electrons and holes, generated by rapid scattering between photoexcited carri-ers aft er the optical excitation 14–17, their temperature being determined by interactions with strongly coupled optical phonons 15. It scales with the number of layers and can be used as a quantitative imaging tool, as well as to reveal the dynamics of the hot electron–hole plasma 14–17 (Fig. 1c). As for oxygen-induced luminescence, further work is neces-

sary to fully explain this hot luminescence.

R e c o m b i n a t i o n

Air

Bilayer

G r a p h e n e

2.3%

T r a n s m i t t a n c e (%)

Distance (μm)

Elastic scattering

1L

1L

1L

1L

1L

a b

c

d

F

igure 1 | The optical properties of graphene. a , Elastic light scattering (Rayleigh) image of a graphite ? ake with varying number of graphene

layers 9

. b , Transmittance for an increasing number of layers. Inset, sample design for the experiment of ref. 11, showing a thick metal support structure with several apertures, on top of which grahene ? akes are placed.

c , Schematic of photoexcite

d electron kinetics in graphene, with possibl

e relaxation mechanisms for the non-equilibrium electron population. d , Photoluminescence (top) and elastic scattering (bottom) images o

f an oxygen-treated ? ake 18. 1L indicates single-layer graphene. Figure reproduced with permission from: a , ref. 9, ? 2007 ACS; b , ref. 11, ? 2008 AAAS; c , ref. 13, ? 2010 ACS; d , ref. 18, ? 2009 ACS.

REVIEW ARTICLE

NATURE PHOTONICS DOI: 10.1038/NPHOTON.2010.186

? 2010

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

613

E lectroluminescence was also recently reported in pristine graphene 33. Although the power conversion effi ciency is lower than it is for carbon nanotubes (CNTs), this could lead to new light-emitting devices based entirely on graphene.

P roduction

Graphene was fi rst produced by micromechanical exfolia-tion of graphite 34

. Th is approach still gives the best samples in terms of purity, defects, mobility and optoelectronic proper-ties. However, it is clear that large-scale assembly is needed for the widespread application of this material. Several approaches have been developed to provide a steady supply of graphene in large areas and quantities, amenable for mass applications. Th ese include growth by chemical vapour deposition (CVD)35–39, seg-regation by heat treatment of carbon-containing substrates 40–42, and liquid phase exfoliation 43–47. In fact, most of these methods date back several decades. Th e current interest in graphene has pushed these early approaches to large yields, controlled growth and large areas, and made it possible in just six years to go from micrometre-sized fl akes to near-mass-production of layer-controlled samples.

Microme chanical

cle avage . Th is method involves peeling off a

piece of graphite by means of adhesive tape 34

. It has been optimized to produce SLG of up to millimetres in size, and with high struc-tural and electronic quality. Although this is the method of choice for fundamental research, with most key results on inpidual SLG being obtained on such fl akes, it has disadvantages in terms of yield and throughput, and is impractical for large-scale applications.L iquid-phase e xfoliation. Liquid-phase exfoliation (LPE) consists of chemical wet dispersion followed by ultrasonication, in both aque-ous 45 and non-aqueous solvents 44. Up to ~70% SLG can be achieved by mild sonication in water with sodium deoxycholate followed by sedimentation-based ultracentrifugation. Bile salt surfactants also allow the isolation of fl akes with controlled thickness, when com-bined with density gradient ultracentrifugation 48. E xfoliation of graphite-intercalated compounds 46 and expandable graphite 49 has also been reported.

LPE can also produce graphene nanoribbons with widths less than 10 nm (ref. 43) and off ers advantages of scalability and no require-ment for expensive growth substrates. Furthermore, it is an ideal means to produce fi lms and composites.

a

Graphene ITO

Arc discharge SWNTs

TiO 2/Ag/TiO 2

ZnO/Ag/ZnO

200

400600

800

20

40

60

80

100

T r a n s m i t t a n c e (%)

Wavelength (nm)

b

c d T r a n s m i t t a n c e (%)

40

60

80

100

T r a n s m i t t a n c e (%)

Thickness (nm)

S h e e t r e s i s t a n c e (Ω/ )

0.1

1

10

1001,000

Sheet resistance (Ω/ )

Sheet resistance (Ω/ )

F igure 2 | Graphene as transparent conductor. a , Transmittance for diff erent transparent conductors: GTCFs 39, single-walled carbon nanotubes (SWNTs)77, ITO 75, ZnO/Ag/ZnO (ref. 81) and TiO 2/Ag/TiO 2 (ref. 70). b , Thickness dependence of the sheet resistance. The blue rhombuses show roll-to-roll GTCFs based on CVD-grown graphene 39; red squares, ITO 75; grey dots, metal nanowires 75; green rhombuses, SWNTs 77. Two limiting lines for GTCFs are also plotted (enclosing the shaded area), calculated from equation (6) using typical values for n and μ. c , Transmittance versus sheet resistance for diff erent

transparent conductors: blue rhombuses, roll-to-roll GTCFs based on CVD-grown graphene 39; red line, ITO 75; grey dots, metal nanowires 75

; green triangles, SWNTs 77. Shaded area enclosed by limiting lines for GTCFs calculated using n and μ as in b . d , Transmittance versus sheet resistance for GTCFs grouped according to production strategies: triangles, CVD 37–39,97; blue rhombuses, micromechanical cleavage (MC)88; red rhombuses, organic synthesis from

polyaromatic hydrocarbons (PAHs)65; dots, liquid-phase exfoliation (LPE) of pristine graphene 44,45,48,88; and stars, reduced graphene oxide (RGO)52,54,83,84,96. A theoretical line as for equation (6) is also plotted for comparison.

REVIEW ARTICLE

NATURE PHOTONICS DOI: 10.1038/NPHOTON.2010.186

? 2010

614

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

G raphe ne oxide . Sonication of graphite oxide can be used to produce

graphene oxide 47

, following the 50-year-old Hummers method 50. Th e oxidation of graphite in the presence of acids and oxidants, proposed in the nineteenth century 51, disrupts the sp 2 network and introduces hydroxyl or epoxide groups 52,53 with carboxylic or carbo-nyl groups attached to the edges. Th ese make graphene oxide sheets readily dispersible in water and several other solvents. Although large fl akes can be obtained, these are intrinsically defective and electri-cally insulating. Despite several attempts 47,52, reduced graphene oxide does not fully regain the pristine graphene electrical conductivity 52,54. It is therefore important to distinguish between dispersion-processed graphene fl akes, which retain the electronic properties of graphene, and insulating graphene oxide layers.

C hemical vapour deposition. Th e CV

D of FLGs was reported more

than 40 years ago 35

. SLG and FLG can now be grown on various sub-strates by feeding hydrocarbons at a suitable temperature 35–39,55,56. Th e scale of progress in CVD growth is given by ref. 39, in which

samples of over 60 cm were reported. Plasma-enhanced CVD can be applied on substrates without a catalyst 56. Note that most as-grown CVD samples are multilayered. Even if their Raman spectrum seems similar 37,38 to that of ideal SLG 57

, this is just an indication of electronic decoupling of the layers, not defi nite proof of SLG growth.C arbon se gre gation. Graphene can also be produced through carbon segregation from silicon carbide 40,58,59 (SiC) or metal substrates 41,55,60–62 following high-temperature annealing. Acheson reported 58 a method of producing graphite from SiC in as early as 1896, and the segrega-tion of graphene from Ni(111) was investigated over 30 years ago 60

. High-quality layers can now be produced on SiC in an argon atmos-phere 42, and electronic decoupling from the underlying SiC substrate can be achieved by hydrogen treatment 63.C he mical synthe sis. Graphene or carbon nanosheets can be produced through chemical synthesis 64. Total organic synthesis yields graphene-like polyaromatic hydrocarbons 65. Th ese synthetic nanographenes can then be assembled to form larger layers, or to achieve atomically precise bottom-up fabrication of nanoribbons. Supramolecular inter-actions can be used to cover SLG with polyaromatic hydrocarbons, keeping the sp 2 network intact without compromising the transport properties. Nanographenes form ordered layers, with precise control of orientation and spacing 66

. Th ese interact with the graphene back-bone, making it possible, in principle, to control and tune its optoelec-tronic properties 66.D e te rministic place me nt. A fundamental step in the production of useful devices is the deterministic placement of graphene on pre-defi ned positions on a substrate of choice. Transfer processes are common in the semiconductor industry, and extensive experi-ence of transfer has been developed for CNTs. Reina et al. reported transfer of SLG and FLG from SiO 2/Si to other substrates 67. A layer of

poly(methyl methacrylate) (PMMA) was coated on graphene depos-ited on SiO 2, then subsequently detached by partial SiO 2 etching 67. Th e PMMA/graphene membrane was then placed over the target

substrate and PMMA dissolved with acetone 67. Kim et al . used a dry-method based on a polydimethylsiloxane stamp to transfer patterned fi lms 37. Bae et al . scaled the process to a roll-based layer-by-layer transfer onto plastic substrates 39.

We have developed a procedure for deterministic placement, fol-lowing transfer. Th is exploits a water layer between the PMMA/graph-ene foil and the substrate, which enables the PMMA to move. Th is allows us to place graphene layers on any substrate in any pre-defi ned location, prepare ‘artifi cial’ multilayers and create sandwich structures with other materials (such as BN or MoS 2). We will show an example of this technique in the section entitled ‘Saturable absorbers and ultra-fast lasers’, by placing graphene on the core of an optical fi 16a9398c83d049649b6658c7rge-scale placement of LPE samples can be achieved by spin- coating and Langmuir–Blodgett techniques 49. Surface modifi cation by self-assembled monolayers can enable targeted deposition of graphene fl akes on the large scale. Di-electrophoresis allows control-led placement of inpidual graphene fl akes between pre-patterned

electrodes 68

. Inkjet printing is another attractive technique 69, and could directly ‘write’ optoelectronic devices or thin-fi lm transistors.

P hotonics and optoelectronics applications T ranspare nt conductors. Optoelectronic devices such as displays, touch screens, light-emitting diodes and solar cells require materi-als with low sheet resistance R s and high transparency. In a thin fi lm, R s = ρ/t , where t is the fi lm thickness and ρ = 1/σ is the resistivity, σ being the d.c. conductivity. For a rectangle of length L and width W , the resistance R is:

R = × = R s × ρt L W L W Th e term L /W can be seen as the number of squares of side W that can be superimposed on the resistor without overlapping. Th us, even

if R s has units of ohms (as R does), it is historically quoted in ‘ohms per square’ (Ω/).

Current transparent conductors are semiconductor-based 70: doped

indium oxide (In 2O 3)71, zinc oxide (ZnO)72 or tin oxide (SnO 2)70, as well as ternary compounds based on their combinations 70,72,73. Th e dominant material is indium tin oxide (ITO), a doped n-type semi-conductor composed of ~90% In 2O 3 and ~10% SnO 2 (ref. 70). Th e

electrical and optical properties of ITO are strongly aff ected by impu-rities 70. Tin atoms function as n-type donors 70

. ITO has strong absorp-tion above 4 eV due to interband transitions 70, with other features at lower energy related to scattering of free electrons by tin atoms or

grain boundaries 70. ITO is commercially available with T ≈ 80% and

R s as low as 10 Ω/ on glass 72, and ~60?300 Ω/ on polyethylene terephthalate 73. Note that T is typically quoted at 550 nm, as this is where the spectral response of the human eye is highest 70.ITO suff ers severe limitations: an ever-increasing cost due to indium

scarcity 70

, processing requirements, diffi culties in patterning 70,73 and a sensitivity to both acidic and basic environments. Moreover, it is brit-tle and can easily wear out or crack when used in applications involv-ing bending, such as touch screens and fl exible displays 74. Th is means

that new transparent conductor materials are needed with improved performance. Metal grids 75, metallic nanowires 76 or other metal oxides 73 have been explored as alternatives. Nanotubes and graphene

also show great promise. In particular, graphene fi lms have a higher

T over a wider wavelength range than single-walled carbon nanotube

(SWNT) fi

lms 77–79, thin metallic fi lms 75,76 and ITO 70,72 (Fig. 2a).We now present a relation between T and R s for FLG fi lms of vary-ing doping levels. From equation (3), T depends on the optical con-ductivity G 0:

T = 1+ N

G 00

–2

(4)where N is the number of layers. Th e sheet resistance R s is linked to the bidimensional d.c. conductivity σ2D by:

R s = (σ2D N )–1

(5)Combining equations (4) and (5) and eliminating N gives:T

= 1 (6)where Z 0 = 1/ε0c = 377 Ω is the free-space impedance, ε0 is the

REVIEW ARTICLE

NATURE PHOTONICS

DOI: 10.1038/NPHOTON.2010.186

? 2010

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

615

free-space electric constant and c is the speed of light. In graphene 1 we can take σ2D = nμe , where n is the number of charge carriers. Note that for n ≈ 0, σ2D does not go to zero, but assumes a constant value 1 of approximately 4e 2/h , giving R s ≈ 6 kΩ/ for an ideal intrin-sic SLG with T ≈ 97.7%. Th us, an ideal intrinsic SLG would beat the best ITO only in terms of T , not R s . However, real samples depos-ited on substrates, in thin fi lms or embedded in polymers are never intrinsic. Exfoliated SLG typically has n ≥ 1012 cm ?2 (see ref. 80, for example), and much smaller R s . Th e range of T and R s that can be realistically achieved for graphene layers of varying thickness can be estimated by taking n = 1012–1013 cm ?2 and μ = 1,000–20,000 cm 2 V –1 s –1, which is typical for fi lms grown by CVD. Figure 2b,c shows that graphene can achieve the same R s as ITO, ZnO/Ag/ZnO (ref. 81), TiO 2/Ag/TiO 2 and SWNTs with a similar or even higher T . Figure 2c plots T versus R s for ITO (ref. 75), Ag nanowires 75, SWNTs 77 and the best graphene-based transparent conductive fi lms (TCFs) reported so far 39, again showing that the latter is superior. For instance, taking n = 3.4 × 1012 cm ?2 and μ = 2 × 104 cm 2 V –1 s –1, we get T = 90% and R s = 20 Ω/.

Note that equation (6) is intended as a guideline for TCF design and optimization — not as a statement on the transport physics of graphene. For TCF design, empirical expressions of σ2D as a func-tion of carrier concentration and doping are enough, whatever the origin and precise quantifi cation of the minimal conductivity, and of the dependence of R s on doping, defects, electron–hole puddles and so on.Diff erent strategies were explored to prepare graphene-based TCFs (GTCFs): spraying 82, dip 83 and spin coating 84, vacuum fi ltra-tion 54 and roll-to-roll processing 39

. Considerable progress has been made since the fi rst attempts to produce graphene-oxide-based TCFs (GOTCFs). Because graphene oxide is insulating, it must be reduced to improve R s (ref. 47). Gilje et al .82 decreased R s from 40 GΩ/ to 4 MΩ/ following reduction with dimethylhydrazine. Graphitization 84, hydrazine exposure and low-temperature anneal-ing 54, or high-temperature vacuum annealing 85 further decreased R s down to 800 Ω/ for T = 82% (ref. 85).

Dispersions of graphite-intercalated compounds 86 and hybrid nanocomposites (graphene oxide sheets mixed with silica sols or CNTs 87) were also attempted, with a minimum R s = 240 Ω/ for T = 86% (ref. 87). Graphene fi lms produced by chemical synthesis currently show R s = 1.6 kΩ/ for T = 55% (ref. 65).

Blake et al .88 have reported the best GTCF so far, from LPE of graphite. Th is was fabricated by vacuum fi ltration followed by annealing, achieving R s = 5 kΩ/ and T ≈ 90%. Th e high R s is most

likely due to the small fl ake size and lack of percolation 48,88

. Th e role of percolation can be seen in ref. 48, where R s and T went from 6 kΩ/ and ~75%, to 2 kΩ/ and ~77% with increasing fl ake size.A key strategy to improving performance is stable chemical dop-ing. Blake et al .88 prepared GTCFs, produced by micromechanical cleavage, with T ≈ 98% and R s = 400 Ω/, using a layer of polyvinyl alcohol to induce n-type doping. Bae et al .39 achieved R s ≈ 30 Ω/ and T ≈ 90% by nitric acid treatment of GTCFs derived from CVD-grown fl akes — one order of magnitude lower in terms of R s than previous GTCFs from the wet transfer of CVD fi lms 37.

Figure 2d is an overview of current GTCFs and GOTCFs. It shows that GTCFs derived from CVD fl akes, combined with doping, could outperform ITO, metal wires and SWNTs. Note that GTCFs and GOTCFs produced by other methods, such as LPE, although currently having higher R s at T = 90%, have already been tested in organic light emitters 85,89 and solar cells 83,90. Th ese are cheaper and easier to scale than micromechanical cleavage or CVD fi lms, and must be consid-ered for applications in which cost reduction is crucial.

P hotovoltaic de vice s. A photovoltaic cell converts light to elec-tricity 91

. Th e energy conversion effi ciency is η = P max /P inc , where P max = V OC × I SC × FF and P inc is the incident power. Here, I SC is the maximum short-circuit current, V OC is the maximum open-cir-cuit voltage and FF is the fi ll factor, defi ned as FF = (V max × I max )/

(V OC × I SC ), where V max and I max are the maximum voltage and cur-rent, respectively. Th e fraction of absorbed photons converted to current defi nes the internal photocurrent effi ciency.

Current photovoltaic technology is dominated by silicon cells 91, with η up to ~25% (ref. 92). Organic photovoltaic cells rely on poly-mers for light absorption and charge transport 93. Th ey can be manu-factured economically compared with silicon cells, for example by a roll-to-roll process 94

, even though they have lower η. An organic photovoltaic cell consists of a transparent conductor, a photoac-tive layer and the electrode 93. Dye-sensitized solar cells use a liquid electrolyte as a charge-transport medium 95. Th is type of solar cell consists of a high-porosity nanocrystalline photoanode, comprising TiO 2 and dye molecules, both deposited on a transparent conduc-tor 95. When illuminated, the dye molecules capture the incident photon, generating electron–hole pairs. Th e electrons are injected into the conduction band of the TiO 2 and are then transported to the counter-electrode 95. Dye molecules are regenerated by capturing electrons from a liquid electrolyte. At present, ITO is the most com-mon material for use both as a photoanode and cathode, the latter with a platinum coating.Graphene can fulfi l multiple functions in photovoltaic devices: as the transparent conductor window, photoactive material, channel for charge transport, and catalyst. GTCFs can be used as window elec-trodes in inorganic (Fig. 3a), organic (Fig. 3b) and dye-sensitized solar cell devices (Fig. 3c). Wang et al . used GTCFs produced by chemical synthesis, reporting η ≈ 0.3% (ref. 65). A higher η of ~0.4% was achieved using reduced graphene oxide 96, with R s = 1.6 kΩ/ instead of 5 kΩ/ (ref. 65), despite a lower T (55% instead of 80%). De Arco et al . achieved better performance (η ≈ 1.2%) using CVD graphene as the transparent conductor, with R s = 230 Ω/ and T = 72% (ref. 97). Further optimization is certainly possible, considering the

performance of the best GTCF so far 39.

Graphene oxide dispersions were also used in bulk-heterojunction photovoltaic devices, as electron-acceptors with poly(3-hexylthiophene) and poly(3-octylthiophene) as donors, achieving η ≈ 1.4% (ref. 90). Y ong et al . claim that η > 12% should be possible with graphene as photoactive material 98.

Graphene can cover an even larger number of functions in dye-sensitized solar cells. Wang et al . reported a solid-state solar cell based on the organic compound spiro-OMeTAD1 (as the hole transport material) and porous TiO 2 (for electron transport) using a GTCF anode, with η ≈ 0.26% (ref. 83). Graphene can be incorporated into the nanostructured TiO 2 photoanode to enhance the charge transport rate, preventing recombination, thus improving the internal photo-current effi ciency 99. Y ang et al. used graphene as a TiO 2 bridge, achiev-ing faster electron transport and lower recombination, and leading to η ≈ 7%, which is higher than they achieved with conventional nanoc-rystalline TiO 2 photoanodes in the same experimental conditions 99. Another option is to use graphene, with its high specifi c surface area, to substitute for the platinum counter-electrode. A hybrid poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate) (P E

DOT:PSS)/graphene oxide composite was used as counter-electrode, to obtain η = 4.5%, comparable to the 6.3% for a platinum counter-electrode tested under the same conditions 100 but now with a cheaper material.

L ight-e mitting de vice s . Organic light-emitting diodes (OLE Ds) have an electroluminescent layer between two charge-injecting elec-trodes, at least one of which is transparent 101. In these diodes, holes are injected into the highest occupied molecular orbital (HOMO) of the polymer from the anode, and electrons are injected into the low-est unoccupied molecular orbital (LUMO) from the cathode. For effi

cient injection, the anode and cathode work functions should match the HOMO and LUMO of the light-emitting polymer 101. Because of their high image quality, low power consumption and

REVIEW ARTICLE

NATURE PHOTONICS DOI: 10.1038/NPHOTON.2010.186

? 2010

616

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

ultrathin device structure, OLEDs fi nd applications in ultrathin tel-evisions and other display screens such as computer monitors, dig-ital cameras and mobile phones. Traditionally, ITO, with its work function of 4.4–4.5 eV , is used as the transparent conductive fi lm. However, besides cost issues, ITO is brittle and limited as a fl exible

substrate 73

. In addition, indium tends to diff use into the active OLED layers, which reduces device performance over time 70. Th us, there is a need for alternative TCFs with optical and electrical performance similar to ITO, but without its drawbacks. Graphene has a work function of 4.5 eV , similar to ITO. Th is, combined with its promise as a fl exible and cheap TCF, makes it an ideal candidate for an OLED anode (Fig. 3d), while also eliminating the issues related to indium diff usion. GTCFs anodes enable an out-coupling effi ciency compa-rable to ITO 85. Considering that the R s and T of ref. 85 were 800 Ω/ and 82% at 550 nm, it is reasonable to expect that further optimiza-tion will improve performance.

Matyba et al .89 used a GOTCF in a light-emitting electrochemical cell. Similar to an OLED, this is a device in which the light-emitting polymer is blended with an electrolyte 102. Th e mobile ions in the electrolyte rearrange when a potential is applied between the elec-trodes, forming layers with high charge density at each electrode interface, which allows effi cient and balanced injection of electrons and holes, regardless of the work function of the electrodes 102. Usually, the cells have at least one metal electrode. Electrochemical side-reactions, involving the electrode materials, can cause prob-lems in terms of operational lifetime and effi ciency 89. Th is also hinders the development of fl exible devices. Graphene is the ideal material to overcome these problems. Matyba et al .89 demonstrated a light-emitting electrochemical cell based solely on dispersion-processable carbon-based materials, paving the way towards totally organic low-voltage, inexpensive and effi cient LEDs.P hotode te ctors. Photodetectors measure photon fl ux or optical power by converting the absorbed photon energy into electrical cur-rent. Th ey are widely used in a range of common devices 103, such as remote controls, televisions and DVD players. Most exploit the internal photoeff ect, in which the absorption of photons results in carriers excited from the valence to the conduction band, outputting

an electric current. Th e spectral bandwidth is typically limited by the material’s absorption 103. For example, photodetectors based on iv and iii–v semiconductors suff er from the ‘long-wavelength limit’, as these become transparent when the incident energy is smaller than the bandgap 103. Graphene absorbs from the ultraviolet to terahertz range 11,13,104,105. As a result, graphene-based photodetectors (GPDs; see Fig. 3e) could work over a much broader wavelength range. Th e

response time is ruled by the carrier mobility 103

. Graphene has huge mobilities, so GPDs can be ultrafast.Th e photoelectrical response of graphene has been widely investi-gated both experimentally and theoretically 106–110. Responses at wave-lengths of 0.514, 0.633, 1.5 and 2.4 μm have been reported 110. Much broader spectral detection is expected because of the graphene ultraw-ideband absorption. Xia et al . demonstrated a GPD with a photore-sponse of up to 40 GHz (ref. 109). Th e operating bandwidth of GPDs is mainly limited by their time constant resulting from the device resist-ance, R , a nd c apacitance, C . X ia e t a l . r eported a n R C -limited b andwidth of about 640 GHz (ref. 109), which is comparable to traditional pho-todetectors 111. However, the maximum possible operating bandwidth of photodetectors is typically restricted by their transit time, the fi nite

duration of the photogenerated current 103

. Th e transit-time-limited bandwidth of a GPD could be well over 1,500 GHz (ref. 109), surpassing state-of-the-art photodetectors.Although an external electric fi eld can produce effi cient photocur-rent generation with an electron–hole separation effi ciency of over 30% (ref. 107), zero source–drain bias and dark current operations could be achieved by using the internal electric fi eld formed near the

metal electrode–graphene interfaces 109,110

. However, the small eff ec-tive area of the internal electric fi eld could decrease the detection

effi ciency 109,110

, as most of the generated electron–hole pairs would be out of the electric fi eld, thus recombining, rather than being sepa-rated. Th e internal photocurrent effi ciencies (15?30%; refs 107,108) and external responsivity (generated electric current for a given input optical power) of ~6.1 mA per watt so far reported 110 for GPDs are relatively low compared with current photodetectors 103. Th is is mainly due to limited optical absorption when only one SLG is used, short photocarrier lifetimes and small eff ective photodetection areas (~200 nm in ref. 109).

Light

Transparent

graphene electrode

Back re?ector

electrode

p layer

Intrinsic layer

n layer

Transparent

graphene electrode

Electrode

Electron blocking

layer Polymer/

graphene active layer

Si Output

Substrate

Graphene Organic light-emitting layers

Cathode a

d

b

c

e

Graphene SiO 2

Light

Metal contact

Light

Light

Transparent graphene electrode

Graphene counter-electrode

I –

I 3

–Graphene bridge structure

F igure 3 | Graphene-based optoelectronics. a –c , Schematics of inorganic (a ), organic (b ) and dye-sensitized (c ) solar cells. I ? and I ?3 are iodide and tri-iodide, respectively. The I ? and I ?

3 ions transfer electrons to the oxidized dye molecules, thus completing the internal electrochemical circuit between the photoanode and the counter-electrode. d ,e , Schematics of an organic LED (d ) and a photodetector (e ). The cylinder in d represents an applied voltage.

REVIEW ARTICLE

NATURE PHOTONICS DOI: 10.1038/NPHOTON.2010.186

? 2010

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

617

Th e photothermoelectric eff ect, which exploits the conversion of photon energy into heat and then electric signal 103, may play an important part in photocurrent generation in graphene devices 107,112. Th us photothermoelectric GPDs may be possible.

T ouch scre e ns. Touch screens are visual outputs that can detect the presence and location of a touch within the display area, permitting physical interaction with what is shown on the display itself 113. Touch panels are currently used in a wide range of applications such as cel-lular phones and digital cameras because they allow quick, intuitive and accurate interaction by the user with the display content.

Resistive and capacitive touch panels are the most common (Fig. 4a). A resistive touch panel comprises a conductive substrate, a liquid-crystal device front panel and a TCF 113. When pressed, the front-panel fi lm comes into contact with the bottom TCF , and the coordinates of the contact point are calculated on the basis of their resistance values. Th ere are two categories of resistive touch screens: matrix and analogue 113. Th e matrix has striped electrodes, whereas the analogue has a non-patterned transparent conductive electrode with lower production costs. Th e TCF requirements for resistive screens are R s ≈ 500?2,000 Ω/ and T > 90% at 550 nm (ref. 113). Favourable mechanical properties, including brittleness and wear resistance, high chemical durability, no toxicity and low production costs, are also important. Cost, brittleness, wear resistance and chemi-cal durability are the main limitations of ITO 70,73, which cannot with-stand the repeated fl exing and poking involved with this type of application. Th us, for resistive touch screens there is an eff ort to fi nd an alternative transparent conductor.

GTCFs can satisfy the requirements for resistive touch screens in terms of T and R s , as well as exhibiting large-area uniformity. Bae et al .39 recently produced a graphene-based touch panel display by screen-printing a CVD-grown sample (Fig. 4b). Considering the R s and T required by analogue resistive screens, GTCFs or GOTCFs produced by LPE are also viable alternatives, further reducing costs.

Capacitive touch screens are emerging as the high-end technol-ogy, especially since the launch of Apple’s iPhone. Th ese consist of an

insulator such as glass, coated with ITO 113

. As the human body is also a conductor, touching the surface of the screen results in an electro-static fi eld distortion, which is measurable as a change in capacitance. Although capacitive touch screens do not work by poking with a pen (making mechanical stresses lower than for resistive screens), the use of GTCFs can still improve performance and reduce costs.

F

lexible smart windows and bistable displays. Polymer-dispersed liquid-crystal (PDLC) devices were introduced in the early 1980s 114. Th ese consist of thin fi lms of optically transparent polymers with micrometre-sized liquid-crystal droplets contained within pores of the polymer. Light passing through the liquid-crystal/polymer is strongly scattered, producing a milky fi lm. If the liquid crystal’s ordinary refractive index is close to that of the host polymer, apply-ing an electric fi eld results in a transparent state 74. In principle, any type of thermotropic liquid crystal may be used in PDLC devices for applications not requiring high switching speeds. In particular, the ability to switch from translucent to opaque makes them attrac-tive for electrically switchable ‘smart windows’ that can be activated when privacy is required. Conventionally, ITO on glass is used as the conductive layer to apply the electric fi eld across the PDLC. However, one of the reasons behind the limited market penetration of smart windows is the high cost of ITO. Furthermore, fl exibility is hindered when using ITO, reducing potential applications such as PDLC fl exible displays 74. Transparent or coloured/tinted smart win-dows generally require T to be 60–90% or higher and R s to be 100–1,000 Ω/, depending on production cost, application and manufacturer. In addition to fl exibility, the electrodes need to be as large as the window itself and must have long-term physical and chemical stability, as well as being compatible with the roll-to-roll PDLC production process. Liquid crystals could also be used for next-generation zero-power monochromatic and coloured fl exible bistable displays, which can retain an image with no power con-sumption. Th ese are attractive for signs and advertisements or for e-readers, and require a transparent fl exible conductor for switch-ing the image. Th e present ITO devices are not ideal for this appli-cation, owing to the limitations discussed above.All these defi ciencies of ITO electrodes can be overcome by GTCFs. Figure 4c,d shows their working principle, and Fig. 4e shows a prototype of a fl exible smart window with polyethylene terephtha-late used as a substrate.

S aturable absorbe rs and ultrafast lase rs. Materials with nonlin-ear optical and electro-optical properties are needed in most phot-onic applications. Laser sources producing nano- to subpicosecond pulses are a key component in the portfolio of leading laser manu-facturers. Solid-state lasers have so far been the short-pulse source of choice, being deployed in applications ranging from basic research to materials processing, from eye surgery and printed circuit-board manufacturing to metrology and the trimming of electronic

Graphene-based transparent electrode

F

igure 4 | Graphene touch screen and smart window. a , Schematic of a capacitive touch screen. b , Resistive graphene-based touch screen. c , Schematic of a PDLC smart window using a GTCF. d , With no voltage, the liquid-crystal molecules are not aligned, making the window opaque. e , Graphene/nanotube-based smart window in either an off (left) or on (right) state. Image in b reproduced with permission from ref. 39, ? 2010 NPG.

? 2010

618

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

components such as resistors and capacitors. Regardless of wave-length, the majority of ultrafast laser systems use a mode-locking technique, whereby a nonlinear optical element, called a saturable absorber, turns the continuous-wave output into a train of ultrafast optical pulses 115. Th e key requirements for nonlinear materials are fast response time, strong nonlinearity, broad wavelength range, low optical loss, high power handling, low power consumption, low cost and ease of integration into an optical system. Currently, the dominant technology is based on semiconductor saturable absorber mirrors 115. However, these have a narrow tuning range, and require complex fabrication and packaging 12,115. A simple, cost-eff ective

alternative is to use SWNTs 12,116

, in which the diameter controls the gap and thus the operating wavelength. Broadband tunability is pos-sible using SWNTs with a wide diameter distribution 12,116. However, when operating at a particular wavelength, SWNTs not in resonance are not used and contribute unwanted losses.

As discussed above, the linear dispersion of the Dirac electrons in graphene off ers an ideal solution: for any excitation there is always an

Figure 6b plots T as a function of average pump power for six wave-lengths. Saturable absorption is evident from the T increase with power at all wavelengths.

Various strategies have been proposed to integrate graphene satu-rable absorbers (GSAs) in laser cavities for ultrafast pulse generation. Th e most common is to sandwich a GSA between two fi bre con-nectors with a fi bre adaptor, as shown schematically in Fig. 5d 12,13,118. Graphene on a side-polished fi bre has also been reported, aimed at high power generation by evanescent fi eld interaction 123. A quartz substrate coated with graphene has been used for free-space solid-state lasers 124.Th e most common wavelength of generated ultrafast pulses so far is ~1.5 μm, not because GSAs have any preference for a particular wavelength, but because this is the standard wavelength of optical telecommunications. A solid-state laser mode-locked by graphene has been reported at ~1 μm (ref. 124). Figure 6c shows a GSA mode-locked laser, made from erbium-doped fi bre and tunable from 1,526 to 1,559 nm, with the tuning range mainly limited by the tunable fi lter, not the GSA 118. Figure 6d,e shows the pulse from a graphene-oxide-based saturable absorber. Th e possibility of tuning the GSA properties by functionalization or by using diff erent layers or composite concen-trations off ers considerable design freedom. Table 1 gives a perform-ance comparison of graphene-based ultrafast lasers and the main carbon-nanotube-based devices 125,126.

O ptical limite rs. Optical limiters are devices that have high trans-mittance for low incident light intensity, and low transmittance for high intensity 127. Th ere is a great interest in these for optical sensors and human eye protection, as retinal damage can occur when intensi-ties exceed a certain threshold 127. Passive optical limiters, which use a nonlinear optical material, have the potential to be simple, compact and cheap 127. However, so far no passive optical limiters have been able to protect eyes and other common sensors over the entire visible and near-infrared range 127. Typical materials include semiconductors (for example ZnSe, InSb), organic molecules (for example phthalo-cyanines), liquid crystals and carbon-based materials (for example, carbon-black dispersions, CNTs and fullerenes)127,128. Fullerenes and their derivatives 129,130 and CNT dispersions 130 have good optical lim-iting performance, in particular for nanosecond pulses at 532 and 1,064 nm (ref. 130).

In graphene-based optical limiters the absorbed light energy con-verts into heat, creating bubbles and microplasmas 128, which results in reduced transmission. Graphene dispersions can be used as

Graphene

EDF

WDM

LD

PC

ISO

d

Graphene SA

Fibre connectors

Output coupler

F igure 5 | Graphene integration in ? bre lasers. a , An optical ? bre is mounted onto a holder. Once detached from the original substrate, a polymer/graphene membrane is slid and aligned with the ? bre core. b , Flake originally deposited on SiO 2/Si. c , The same ? ake after

deterministic placement and dissolution of the polymer layer. d , Graphene mode-locked ultrafast laser: a graphene saturable absorber (SA) is inserted between two ? bre connectors. An erbium-doped ? bre (EDF) is the gain medium, pumped by a laser diode (LD) with a wavelength-pision multiplexer (WDM). An isolator (ISO) maintains unidirectional operation, and a polarization controller (PC) optimizes mode-locking.

? 2010

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics 619

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 128. It has also been shown 131 that functionalized graphene dispersions could outperform C 60 as an optical limiter. O ptical frequency converters. Optical frequency converters are used to expand the wavelength accessibility of lasers (for example, frequency doubling, parametric amplifi cation and oscillation, and four-wave

mixing)127

. Calculations suggest that nonlinear frequency genera-tion in graphene (harmonics of input light, for example) should be possible for suffi ciently high external electric fi elds (>100 V cm –1)132. Second-harmonic generation from a 150 fs laser at 800 nm has been reported for a graphene fi lm 133. In addition, four-wave mixing to gen-erate near-infrared wavelength tunable light has been demonstrated using SLG and FLG 134. Graphene’s third-order susceptibility |χ3| was measured to be ~10?7 e.s.u. (ref. 134) — up to one order of magnitude larger than that reported so far for similar measurements on CNTs 134. However, photon-counting electronics is typically needed to measure the output 133, indicating a low conversion effi ciency. Other features of graphene, such as the possibility of tuning the nonlinearity by chang-ing the number of layers 134, and wavelength-independent nonlinear susceptibility 134, still could be potentially used for various photonic applications (optical imaging 134, for example).

Te rahe rtz de vice s. Radiation in the 0.3–10 THz range (30 μm to 1 mm) is attractive for biomedical imaging, security, remote sens-ing and spectroscopy 135. Much unexplored territory still remains for terahertz technology, mainly owing to a lack of aff ordable

and effi cient sources and detectors 135

. Th e frequency of graphene

plasma waves 136

lies in the terahertz range, as well as the gap of graphene nanoribbons, and the bilayer graphene tunable band-gap, making graphene appealing for terahertz generation and detection. Various terahertz sources have been suggested based on electrical 136 or optical 136 pumping of graphene devices. Recent experimental observations of terahertz emission 137 and amplifi ca-tion 138

in optically pumped graphene have shown the feasibility of graphene-based terahertz generation. Twisted multilayers, retain-ing the electronic properties of SLG, could also be very interesting for such applications.

Graphene devices can be used for terahertz detection and fre-quency conversion. Th e possibility of tuning the electronic and optical properties by external means (for example through electric or magnetic fi elds, or using an optical pump) makes SLG and FLG suitable for infrared and terahertz radiation manipulation. Th e possible devices include modulators, fi lters, switches, beamsplitters and polarizers.

P erspective

Graphene fi lms and composites have ideal electronic and optical properties for photonics and optoelectronics. Graphene is an attrac-tive replacement for ITO and other transparent conductors. In many cases (touch screens or OLEDs, for example), this increases fabrica-tion fl exibility, in addition to having economic advantages. Present

PVA

Graphene-PVA

-6

-4

-2

2

4

6

1,559 nm

1,553 nm

1,547 nm

1,541 nm

1,534 nm

1,526 nm

20 μm

~1.3%

500

1,0001,5002,000Measurement Sech 2

?t

0.0

0.40.81.2

1.6

64.565.000.20.40.60.81.00

0.20.40.60.81.0Time delay (ps)

1,520

1,540

1,560

1,580

Wavelength (nm)1,530

1,5451,5601,575

N o r m a l i z e d s p e c t r a l i n t e n s i t y

N o r m a l i z e d i n t e n s i t y

N o r m a l i z e d i n t e n s i t y

A b s o r b a n c e

T r a n s m i t t a n c e (%)

Wavelength (nm)a

b

e

c

d

Wavelength (nm)

Average pump power (mW)

F

igure 6 | Graphene mode-locked laser performance. a , Absorption of graphene–PVA (polyvinyl alcohol) composite and reference PVA. Inset: micrograph of the composite. b , Typical transmittance as a function of pump power at six diff erent wavelengths. Transmittance increases with power. c , Tunable (>30 nm) ? bre laser mode-locked by graphene.

d ,

e , Autocorrelation (d ) and spectrum (e ) o

f output pulses of a graphene oxide mode-locked laser, with a ~743 fs pulse duration. Figure a ,b reproduced with permission from ref. 13, ? 2010 ACS.

REVIEW ARTICLE

NATURE PHOTONICS DOI: 10.1038/NPHOTON.2010.186

? 2010

620

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

liquid-crystal-based devices face high fabrication costs associated with the requirement for large transparent electrodes. Th e move to a graphene-based technology could make them more viable. New forms of graphene-based transparent electrodes on fl exible sub-strates for solar cells can add value and a level of operational fl exibil-ity that is not possible with current transparent conductors and rigid glass substrates. Recent progress in growth and dispersion process-ing of graphene have defi nitely made this material ‘come of age’, thus encouraging industrial applications. Deterministic placement of graphene layers on arbitrary substrates, and the creation of multilay-ers by the inpidual assembly of monolayers at given angles, are now possible. Future eff orts in the fi eld of nonlinear optical devices will focus on demonstrators at diff erent wavelengths to make full use of graphene’s ultrawide broadband capability. Th is could include high-speed, transparent and fl exible photosensitive systems, which could be further functionalized to enable chemical sensing. Ultrafast and tunable lasers have become a reality, with an ever-growing number of groups entering this fi eld. Th e combination of graphene photonics with plasmonics could lead to a wide range of advanced devices.

References

1.

Geim, A. K. & Novoselov, K. S. Th e rise of graphene. Nature Mater. 6, 183–191 (2007).

2. Charlier, J. C., Eklund, P . C., Zhu, J. & Ferrari, A. C. Electron and phonon

properties of graphene: Th eir relationship with carbon nanotubes. Top. Appl. Phys. 111, 673–709 (2008).3. Wallace, P . R. Th e band theory of graphite. Phys. Rev. 71, 622–634 (1947).4. Zhang, Y ., Tan, Y .-W ., Stormer, H. L. & Kim, P . Experimental observation of the

quantum Hall eff ect and Berry’s phase in graphene. Nature 438, 201–204 (2005).5. Du, X. I. et al. Fractional quantum Hall eff ect and insulating phase of Dirac

electrons in graphene. Nature 462, 192–195 (2009).6. Lemme, M. C., Echtermeyer, T. J., Baus, M. & Kurz, H. A graphene fi eld-eff ect

device. IEEE Electr. Device Lett. 28, 282–284 (2007).

7. Han, M. Y., Ozyilmaz, B., Zhang, Y. & Kim, P . Energy band-gap engineering of

graphene nanoribbons. Phys. Rev. Lett. 98, 206805 (2007).8. Lin, Y.-M. et al. 100-GHz transistors from wafer-scale epitaxial graphene.

Science 327, 662 (2010).9. Casiraghi, C. et al. Rayleigh imaging of graphene and graphene layers.

Nano Lett. 7, 2711–2717 (2007).10. Blake, P . et al. Making graphene visible. Appl. Phys. Lett. 91, 063124 (2007).11. Nair, R. R. et al. Fine structure constant defi nes transparency of graphene.

Science 320, 1308–1308 (2008).12. Hasan, T. et al. Nanotube–polymer composites for ultrafast photonics.

Adv. Mater. 21, 3874–3899 (2009).13. Sun, Z. et al. Graphene mode-locked ultrafast laser. ACS Nano 4,

803–810 (2010).

14. Stoehr, R. J., Kolesov, R., Pfl aum, J. & Wrachtrup, J. Fluorescence of laser

created electron–hole plasma in graphene. Preprint at <16a9398c83d049649b6658c7/abs/1006.5434v1> (2010).

15. Liu, C. H., Mak, K. F., Shan, J. & Heinz, T. F. Ultrafast photoluminescence

from graphene. Preprint at <16a9398c83d049649b6658c7/abs/1006.5769v1> (2010).16. Wu, S. et al. Nonlinear photoluminescence from graphene. Abstract number:

BAPS.2010.MAR.Z22.11, APS March Meeting, Portland, Oregon (2010). 17. Hartschuh, A. et al. Excited state energies and decay dynamics in carbon

nanotubes and graphene. E-MRS Spring Meeting (2010).18. Gokus, T. et al. Making graphene luminescent by oxygen plasma treatment.

ACS Nano 3, 3963–3968 (2009).19. da, G. et al. Blue photoluminescence from chemically derived graphene

oxide. Adv. Mater. 22, 505–509 (2009).20. Sun, X. et al. Nano-graphene oxide for cellular imaging and drug delivery.

Nano Res. 1, 203–212 (2008).

21. Luo, Z., Vora, P . M., Mele, E. J., Johnson, A. T. & Kikkawa, J. M.

Photoluminescence and band gap modulation in graphene oxide. Appl. Phys. Lett. 94, 111909 (2009).

22. Kuzmenko, A. B., van Heumen, E., Carbone, F. & van der Marel, D. Universal

optical conductance of graphite. Phys. Rev. Lett. 100, 117401 (2008).23. Wang, F. et al. Gate-variable optical transitions in graphene. Science 320,

206–209 (2008).

24. Mak, K. F., Shan, J. & Heinz, T. F. Electronic structure of few-layer graphene:

experimental demonstration of strong dependence on stacking sequence. Phys. Rev. Lett. 104, 176404 (2009).

25. Breusing, M., Ropers, C. & Elsaesser, T. Ultrafast carrier dynamics in graphite.

Phys. Rev. Lett. 102, 086809 (2009).

26. Kampfrath, T., Perfetti, L., Schapper, F ., Frischkorn, C. & Wolf, M. Strongly

coupled optical phonons in the ultrafast dynamics of the electronic energy and current relaxation in graphite. Phys. Rev. Lett. 95, 187403 (2005).

27. Lazzeri, M., Piscanec, S., Mauri, F., Ferrari, A. C. & Robertson, J. Electronic

transport and hot phonons in carbon nanotubes. Phys. Rev. Lett. 95, 236802 (2005).

28. González, J., Guinea, F. & Vozmediano, M. A. H. Unconventional

quasiparticle lifetime in graphite. Phys. Rev. Lett. 77, 3589–3592 (1996).29. Lu, J. et al. One-pot synthesis of fl uorescent carbon nanoribbons,

nanoparticles, and graphene by the exfoliation of graphite in ionic liquids. ACS Nano 3, 2367–2375 (2009).

30. Sheats, J. R. et al. Organic electroluminescent devices. Science 273,

884–888 (1996).

31. Rothberg, L. J. & Lovinger, A. J. Status of and prospects for organic

electroluminescence. J. Mater. Res. 11, 3174–3187 (1996).32. Frangioni, J. V . In vivo near-infrared fl uorescence imaging.

Curr. Opin. Chem. Biol. 7, 626–634 (2003).33. ssig, S. et al. Phonon-assisted electroluminescence from metallic carbon

nanotubes and graphene. Nano Lett. 10, 1589–1594 (2010).

34. Novoselov, K. S. et al. Two-dimensional atomic crystals. Proc. Natl Acad. Sci.

USA 102, 10451–10453 (2005).

35. Karu, A. E. & Beer, M. Pyrolytic formation of highly crystalline graphite fi lms.

J. Appl. Phys. 37, 2179–2181 (1966).

36. Obraztsov, A. N., Obraztsova, E. A., Tyurnina, A. V . & Zolotukhin, A. A.

Chemical vapor deposition of thin graphite fi lms of nanometer thickness. Carbon 45, 2017–2021 (2007).

37. Kim, K. S. et al. Large-scale pattern growth of graphene fi lms for stretchable

transparent electrodes. Nature 457, 706–710 (2009).38. Reina, A. et al. Large area, few-layer graphene fi lms on arbitrary substrates by

chemical vapor deposition. Nano Lett. 9, 30–35 (2009).39. Bae, S. et al. Roll-to-roll production of 30-inch graphene fi lms for

transparent electrodes. Nature Nanotech. 4, 574–578 (2010).40. Berger, C. et al. Ultrathin epitaxial graphite: 2D electron gas properties

and a route toward graphene-based nanoelectronics. J. Phys. Chem. B 108, 19912–19916 (2004).

41. Sutter, P . W ., Flege, J.-I. & Sutter, E. A. Epitaxial graphene on ruthenium.

Nature Mater. 7, 406–411 (2008).

42. Emtsev, K. V . et al. Towards wafer-size graphene layers by atmospheric

pressure graphitization of silicon carbide. Nature Mater. 8, 203–207 (2009).43. Li, X., Wang, X., Zhang, L., Lee, S. & Dai, H. Chemically derived, ultrasmooth

graphene nanoribbon semiconductors. Science 319, 1229–1232 (2008).44. Hernandez, Y. et al. High-yield production of graphene by liquid-phase

exfoliation of graphite. Nature Nanotech. 3, 563–568 (2008).45. Lotya, M. et al. Liquid phase production of graphene by exfoliation of graphite

in surfactant/water solutions. J. Am. Chem. Soc. 131, 3611–3620 (2009).46. Valles, C. et al. Solutions of negatively charged graphene sheets and ribbons.

J. Am. Chem. Soc. 130, 15802–15804 (2008).47. Stankovich, S. et al. Graphene-based composite materials. Nature 442,

282–286 (2006).

48. Green, A. A. & Hersam, M. C. Solution phase production of graphene with

controlled thickness via density diff erentiation. Nano Lett. 9, 4031–4036 (2009).49. Li, X. et al. Highly conducting graphene sheets and Langmuir–Blodgett fi lms.

Nature Nanotech. 3, 538–542 (2008).50. Hummers, W . S. & Off eman, R. E. Preparation of graphite oxide.

J. Am. Chem. Soc. 80, 1339–1339 (1958).

51. Brodie, B. C. Sur le poids atomique du graphite. Ann. Chim. Phys. 59,

466–472 (1860).52. Mattevi, C. et al. Evolution of electrical, chemical, and structural properties of

transparent and conducting chemically derived graphene thin fi lms. Adv. Funct. Mater. 19, 2577–2583 (2009).53. Cai, W . et al. Synthesis and solid-state NMR structural characterization of

13

C-labeled graphite oxide. Science 321, 1815–1817 (2008).54. Eda, G., Fanchini, G. & Chhowalla, M. Large-area ultrathin fi lms of reduced

graphene oxide as a transparent and fl exible electronic material. Nature Nanotech. 3, 270–274 (2008).

55. Oshima, C. & Nagashima, A. Ultra-thin epitaxial fi lms of graphite and

hexagonal boron nitride on solid surfaces. J. Phys. Condens. Mat. 9, 1–20 (1997).56. Wang, J. et al. Synthesis of carbon nanosheets by inductively coupled

radio-frequency plasma enhanced chemical vapor deposition. Carbon 42, 2867–2872 (2004).

57. Ferrari, A. C. et al. Raman spectrum of graphene and graphene layers.

Phys. Rev. Lett. 97, 187401 (2006).58. Acheson, E. G. Production of artifi cial crystalline carbonaceous materials;

article of carborundum and process of the manufacture thereof carborundum. US patent 615,648 (1896).

REVIEW ARTICLE

NATURE PHOTONICS

DOI: 10.1038/NPHOTON.2010.186

? 2010

NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

621

59. Badami, D. V . Graphitization of α-silicon carbide. Nature 193, 569–570 (1962).60. Isett, L. C. & Blakely, J. M. Segregation isosteres for carbon at the (100) surface

of nickel. Surf. Sci. 58, 397–414 (1976).

61. Gamo, Y ., Nagashima, A., Wakabayashi, M., Terai, M. & Oshima, C. Atomic

structure of monolayer graphite formed on Ni(111). Surf. Sci. 374, 61–64 (1997).62. Rosei, R. et al. Structure of graphitic carbon on Ni(111): A surface extended-energy-loss fi ne-structure study. Phys. Rev. B 28, 1161–1164 (1983).63. Riedl, C. et al. Quasi-free-standing epitaxial graphene on SiC obtained by

hydrogen intercalation. Phys. Rev. Lett. 103, 246804 (2009).64. Choucair, M., Th ordarson, P . & Stride, J. A. Gram-scale production of

graphene based on solvothermal synthesis and sonication. Nature Nanotech. 4, 30–33 (2009).65. Wang, X. et al. Transparent carbon fi lms as electrodes in organic solar cells.

Angew. Chem. 47, 2990–2992 (2008).

66. Wu, J., Pisula, W . & Mullen, K. Graphenes as potential material for electronics.

Chem. Rev. 107, 718–747 (2007).67. Reina, A. et al. Transferring and identifi cation of single-and few-layer

graphene on arbitrary substrates. J. Phys. Chem. C 112, 17741–17744 (2008).68. Vijayaraghavan, A. et al. Dielectrophoretic assembly of high-density arrays

of inpidual graphene devices for rapid screening. ACS Nano 3, 1729–1734 (2009).69. Beecher, P . et al. Ink-jet printing of carbon nanotube thin fi lm transistors.

J. Appl. Phys. 102, 043710 (2007).

70. Hamberg, I. & Granqvist, C. G. Evaporated Sn-doped In 2O 3 fi

lms: basic optical properties and applications to energy-effi cient windows. J. Appl. Phys. 60, R123–R160 (1986).71. Holland, L. & Siddall, G. Th e properties of some reactively sputtered metal

oxide fi lms. Vacuum 3, 375–391 (1953).

72. Minami, T. Transparent conducting oxide semiconductors for transparent

electrodes. Semicond. Sci. Technol. 20, S35–S44 (2005).

73. Granqvist, C. G. Transparent conductors as solar energy materials: a

panoramic review. Sol. Energy Mater. Sol. Cells 91, 1529–1598 (2007).74. Sheraw, C. D. et al. Organic thin-fi lm transistor-driven polymer dispersed

liquid crystal displays on fl exible polymeric substrates. Appl. Phys. Lett. 80, 1088–1090 (2002).

75. Lee, J. Y., Connor, S. T., Cui, Y. & Peumans, P . Solution-processed metal

nanowire mesh transparent electrodes. Nano Lett. 8, 689–692 (2008).76. De, S. et al. Silver nanowire networks as fl exible, transparent, conducting

fi lms: extremely high dc to optical conductivity ratios. ACS Nano 3, 1767–1774 (2009).77. Geng, H. Z. et al. Eff ect of acid treatment on carbon nanotube-based fl exible

transparent conducting fi lms. J. Am. Chem. Soc. 129, 7758–7759 (2007).78. Wu, Z. et al. Transparent, conductive carbon nanotube fi lms. Science 305,

1273–1276 (2004).

79. De, S. & Coleman, J. N. Are there fundamental limitations on the sheet resistance

and transmittance of thin graphene fi lms? ACS Nano 4, 2713–2720 (2010).

80. Casiraghi, C., Pisana, S., Novoselov, K. S., Geim, A. K. & Ferrari, A. C. Raman

fi ngerprint of charged impurities in graphene. Appl. Phys. Lett. 91, 233108 (2007).

81. Sahu, D. R., Lin, S. Y. & Huang, J. L. ZnO/Ag/ZnO multilayer fi lms for the

application of a very low resistance transparent electrode. Appl. Surf. Sci. 252, 7509–7514 (2006).

82. Gilje, S., Han, S., Wang, M., Wang, K. L. & Kaner, R. B. A chemical route to

graphene for device applications. Nano Lett. 7, 3394–3398 (2007).

83. Wang, X., Zhi, L. & Mullen, K. Transparent, conductive graphene electrodes

for dye-sensitized solar cells. Nano Lett. 8, 323–327 (2007).

84. Becerril, H. A. et al. Evaluation of solution-processed reduced graphene oxide

fi lms as transparent conductors. ACS Nano 2, 463–470 (2008).85. Wu, J. et al. Organic light-emitting diodes on solution-processed graphene

transparent electrodes. ACS Nano 4, 43–48 (2009).

86. Biswas, S. & Drzal, L. T. A novel approach to create a highly ordered

monolayer fi lm of graphene nanosheets at the liquid–liquid interface. Nano Lett. 9, 167–172 (2008).

87. Tung, V . C. et al. Low-temperature solution processing of graphene-carbon

nanotube hybrid materials for high-performance transparent conductors. Nano Lett. 9, 1949–1955 (2009).88. Blake, P . et al. Graphene-based liquid crystal device. Nano Lett. 8,

1704–1708 (2008).89. Matyba, P . et al. Graphene and mobile ions: the key to all-plastic, solution-processed light-emitting devices. ACS Nano 4, 637–642 (2010).90. Liu, Z. et al. Organic photovoltaic devices based on a novel acceptor material:

graphene. Adv. Mater. 20, 3924–3930 (2008).

91. Chapin, D. M., Fuller, C. S. & Pearson, G. L. A new silicon p-n junction

photocell for converting solar radiation into electrical power. J. Appl. Phys. 25, 676–677 (1954).

92. Green, M. A., Emery, K., Bücher, K., King, D. L. & Igari, S. Solar cell effi ciency

tables. Prog. Photovolt. Res. Appl. 7, 321–326 (1999).

93. Hoppe, H. & Saricift ci, N. S. Organic solar cells: an overview. MRS Bull . 19,

1924–1945 (2004).

94. Krebs, F. C. All solution roll-to-roll processed polymer solar cells free from

indium-tin-oxide and vacuum coating steps. Org. Electron . 10, 761–768 (2009).

95. O’Regan, B. & Gratzel, M. A low-cost, high-effi ciency solar cell based on dye-sensitized colloidal TiO 2 fi lms. Nature 353, 737–740 (1991).96. Wu, J. et al. Organic solar cells with solution-processed graphene transparent

electrodes. Appl. Phys. Lett. 92, 263302 (2008).97. De Arco, L. G. et al. Continuous, highly fl exible, and transparent graphene

fi lms by chemical vapor deposition for organic photovoltaics. ACS Nano 4, 2865–2873 (2010).

98. Y ong, V . & Tour, J. M. Th eoretical effi ciency of nanostructured graphene-based photovoltaics. Small 6, 313–318 (2009).

99. Yang, N., Zhai, J., Wang, D., Chen, Y. & Jiang, L. Two-dimensional graphene

bridges enhanced photoinduced charge transport in dye-sensitized solar cells. ACS Nano 4, 887–894 (2010).

100. Hong, W ., Xu, Y., Lu, G., Li, C. & Shi, G. Transparent graphene/PEDOT-PSS

composite fi lms as counter electrodes of dye sensitized solar cells. Electrochem. Commun. 10, 1555–1558 (2008).

101. Burroughes, J. H. et al. Light-emitting diodes based on conjugated polymers.

Nature 347, 539–541 (1990).

102. Pei, Q. & Heeger, A. J. Operating mechanism of light-emitting electrochemical

cells. Nature Mater. 7, 167 (2008).

103. Saleh, B. E. A. & Teich, M. C. Fundamentals of Photonics Ch. 18,

784–803 (Wiley, 2007).

104. Dawlaty, J. M. et al. Measurement of the optical absorption spectra of epitaxial

graphene from terahertz to visible. Appl. Phys. Lett. 93, 131905 (2008).

105. Wright, A. R., Cao, J. C. & Zhang, C. Enhanced optical conductivity of bilayer

graphene nanoribbons in the terahertz regime. Phys. Rev. Lett. 103, 207401 (2009).

106. Vasko, F. T. & Ryzhii, V . Photoconductivity of intrinsic graphene. Phys. Rev. B

77, 195433 (2008).

107. Park, J., Ahn, Y. H. & Ruiz-Vargas, C. Imaging of photocurrent generation and

collection in single-layer graphene. Nano Lett. 9, 1742–1746 (2009).108. Xia, F. N. et al. Photocurrent imaging and effi cient photon detection in a

graphene transistor. Nano Lett. 9, 1039–1044 (2009).

109. Xia, F., Mueller, T., Lin, Y.-M., Valdes-Garcia, A. & Avouris, P . Ultrafast

graphene photodetector. Nature Nanotech. 4, 839–843 (2009).

110. Mueller, T., Xia, F. & Avouris, P . Graphene photodetectors for high-speed

optical communications. Nature Photon. 4, 297–301(2010).

111. Kang, Y. M. et al. Monolithic germanium/silicon avalanche photodiodes with

340 GHz gain-bandwidth product. Nature Photon. 3, 59–63 (2009).

112. Xu, X. D., Gabor, N. M., Alden, J. S., van der Zande, A. M. & McEuen, P . L.

Photo-thermoelectric eff ect at a graphene interface junction. Nano Lett. 10, 562 (2010).

113. Pickering, J. A. Touch-sensitive screens: the technologies and their

applications. Int. J. Man. Mach. Stud. 25, 249–269 (1986).

114. Craighead, H. G., Cheng, J. & Hackwood, S. New display based on electrically

induced index-matching in an inhomogeneous medium. Appl. Phys. Lett. 40, 22–24 (1982).

115. Keller, U. Recent developments in compact ultrafast lasers. Nature 424,

831–838 (2003).

116. Wang, F. et al. Wideband-tuneable, nanotube mode-locked, fi bre laser.

Nature Nanotech. 3, 738–742 (2008).

117. Sun, D. et al. Ultrafast relaxation of excited Dirac fermions in epitaxial

graphene using optical diff erential transmission spectroscopy. Phys. Rev. Lett. 101, 157402 (2008).

118. Sun, Z. et al. Wideband tunable, graphene-mode locked, ultrafast laser.

Preprint at <16a9398c83d049649b6658c7/abs/1003.4714> (2010).

119. Bao, Q. et al. Atomic-layer graphene as a saturable absorber for ultrafast

pulsed lasers. Adv. Funct. Mater. 19, 3077–3083 (2010).

120. Zhang, H., Bao, Q. L., Tang, D. Y., Zhao, L. M. & Loh, K. Large energy soliton

erbium-doped fi ber laser with a graphene-polymer composite mode locker. Appl. Phys. Lett. 95, 141103 (2009).

121. Zhang, H., Tang, D. Y., Zhao, L. M., Bao, Q. L. & Loh, K. P . Large energy mode

locking of an erbium-doped fi ber laser with atomic layer graphene. Opt. Express 17, 17630–17635 (2009).

122. Zhang, H. et al. Graphene mode locked, wavelength-tunable, dissipative

soliton fi ber laser. Appl. Phys. Lett. 96, 111112 (2010).

123. Song, Y. W ., Jang, S. Y., Han, W . S. & Bae, M. K. Graphene mode-lockers for

fi ber lasers functioned with evanescent fi eld interaction. Appl. Phys. Lett. 96, 051122 (2010).

124. Tan, W . D. et al. Mode locking of ceramic Nd:yttrium aluminum garnet with

graphene as a saturable absorber. Appl. Phys. Lett. 96, 031106 (2010).

125. Scardaci, V . et al. Carbon nanotube polycarbonate composites for ultrafast

lasers. Adv. Mater. 20, 4040–4043 (2008).

REVIEW ARTICLE

NATURE PHOTONICS

DOI: 10.1038/NPHOTON.2010.186

? 2010

622 NATURE PHOTONICS | VOL 4 | SEPTEMBER 2010 | 16a9398c83d049649b6658c7/naturephotonics

126. Sun, Z. et al. A compact, high power, ultrafast laser mode-locked by carbon

nanotubes. Appl. Phys. Lett. 95, 253102 (2009).

127. Bass, M., Li, G. & Stryland, E. V . Handbook of Optics IV (McGraw-Hill, 2001).128. Wang, J., Hernandez, Y., Lotya, M., Coleman, J. N. & Blau, W . J. Broadband

nonlinear optical response of graphene dispersions. Adv. Mater. 21, 2430–2435 (2009).

129. Tutt, L. W . & Kost, A. Optical limiting performance of C 60 and C 70 solutions.

Nature 356, 225–226 (1992).

130. Wang, J., Chen, Y. & Blau, W . J. Carbon nanotubes and nanotube composites

for nonlinear optical devices. J. Mater. Chem. 19, 7425–7443 (2009).131. Xu, Y. et al. A graphene hybrid material covalently functionalized with

porphyrin: synthesis and optical limiting property. Adv. Mater. 21, 1275–1279 (2009).

132. Mikhailov, S. A. Non-linear electromagnetic response of graphene. Europhys.

Lett. 79, 27002 (2007).

133. Dean, J. J. & van Driel, H. M. Second harmonic generation from graphene and

graphitic fi lms. Appl. Phys. Lett. 95, 261910 (2009).

134. Hendry, E., Hale, P . J., Moger, J. J., Savchenko, A. K. & Mikhailov, S. A. Strong

nonlinear optical response of graphene fl akes measured by four-wave mixing. Preprint at <16a9398c83d049649b6658c7/abs/0912.5321v1> (2009).

135. Zhang, X.-C. & Xu, J. Introduction to THz Wave Photonics (Springer, 2010).

136. Rana, F. Graphene terahertz plasmon oscillators. IEEE Trans. Nanotechnol. 7,

91–99 (2008).

137. Sun, D. et al. Coherent control of ballistic photocurrents in multilayer epitaxial

graphene using quantum interference. Nano Lett. 10, 1293–1296 (2010).138. Otsuji, T. et al. Observation of amplifi ed stimulated terahertz emission from

optically pumped epitaxial graphene heterostructures. Preprint at <16a9398c83d049649b6658c7/abs/1001.5075v1> (2010).

Acknowledgements

We thank S. A. Awan, D. M. Basko, E. Lidorikis, A. Hartschuh, J. Coleman,

A. Dyadyusha, D. P . Chu, T. Etchermeyer, T. Kulmala, A. Lombardo, D. Popa, G. Privitera, F. Torrisi, O. Trushkevych, F. Wang, T. Seyller,

B. H. Hong, K. S. Novoselov and

A. K. Geim for discussions. We acknowledge funding from EPSRC grants EP/G042357/1 and EP/G030480/1, ERC grant NANOPOTS, a Royal Society Brian Mercer Award for Innovation, the Cambridge Integrated Knowledge Centre in Advanced Manufacturing Technology for Photonics and Electronics, and Cambridge Nokia Research Centre. F.

B. acknowledges funding from a Newton International Fellowship and T.H. from King’s College, Cambridge. A.

C.F. is a Royal Society Wolfson Research Merit Award holder.

Additional information

Th e authors declare no competing fi nancial interests.

REVIEW ARTICLE

NATURE PHOTONICS

DOI: 10.1038/NPHOTON.2010.186

? 2010

本文来源：https://www.bwwdw.com/article/a3sq.html