PHYSICAL REVIEW B 86, 125450 (2012) Edge magnetoplasmons and the optical excitations in graphene disks Weihua Wang, * S. Peter Apell, and Jari M. Kinaret Department of Applied Physics, Chalmers University of Technology, SE-412 96 G¨ oteborg, Sweden (Received 5 July 2012; revised manuscript received 10 September 2012; published 28 September 2012) We discuss the edge magnetoplasmon properties in highly doped graphene disks, and the corresponding optical excitations. Edge magnetoplasmons with nonzero angular momentum (l = 0) have two branches corresponding to different edge current rotations with respect to the magnetic field. The resonance energies of one branch are blueshifted and the other redshifted relative to energies at B = 0, with the energy differences linearly proportional to the magnetic field. Recently, the l = 1 dipole mode has been investigated by two experiments using optical transmission spectroscopy [Crassee et al., Nano Lett. 12, 2470 (2012); Yan et al., ibid. 12, 3766 (2012)], and classical cyclotron resonances were found in highly doped graphene samples. These are determined by graphene magneto-optical conductivities, which behave like a conventional two-dimensional electron system in the high doping limit. DOI: 10.1103/PhysRevB.86.125450 PACS number(s): 73.20.Mf, 71.35.Ji, 78.67.Wj I. INTRODUCTION Plasmons are fundamental collective oscillations of elec- trons that have captured the interest of scientists for years, driving research in theory, experiment, and applications. Dirac plasmons of graphene, 1 which have also been identified recently, are predicted to have particularly interesting and unique properties, e.g., longer plasmon propagating length and higher field confinement. As an ideal two-dimensional electron system (2DES), the carrier concentration in graphene can be controlled through chemical doping or electrostatic gating. 25 Hence, one can tailor the local conductivity by a bias voltage, and it might provide exciting venues to achieve wave-guiding in simple graphene flakes without additional nanostructures, which have been suggested to per- form transformation optics and cloaking on a one-atom-thick surface. 6,7 Engineering plasmonic nanostructures in graphene, such as disks and ribbons, enables rich functionalities to be obtained, such as modulating the emitter radiation, 8 enhancing the light matter interaction, and realizing plasmon wave-guiding. 9,10 These kinds of structures have also been demonstrated theoretically to achieve a very high optical absorption. 11,12 Quantum effects are also very important in small-sized graphene nanostructures, 13 and very recently, atomically localized plasmon was observed in a graphene flake with a sub-nanoscale localized defect. 14 From microsize down to nanosize, graphene plasmonics has the potential to bridge the size gap between electronics and photonics, 15,16 and it may enable new functional optoelectronics devices to be implemented. II. MAGNETO-OPTICAL CONDUCTIVITIES OF GRAPHENE Active tunability is a key issue in plasmonics. In metal- based plasmonics, the designed nanostructures can usually only work at a specific frequency. However, in graphene plasmonics, the electron concentration can be easily tuned by electrostatic gating, so the plasmon frequency will be tuned as well (e.g., the dipolar plasmon frequencies depend on the chemical potential 11 ). Very recently, this idea was demonstrated experimentally by two groups using a scattering- type scanning near-field optical microscope. 17,18 In addition to this electric tuning method, graphene plasmons can also be tuned through magnetic fields. Graphene magnetoplasmons have been studied theoretically in infinite graphene sheets, 1922 semi-infinite sheets, 23,24 and finite structures. 25,26 As a linear and gapless energy spectrum, graphene has a different Landau level (LL) distribution compared with a usual 2DES, reading E n n¯ B , (1) where ω B = 2v F /l B , and the ± represent an electronlike (+) or holelike () LL index. Here v F 10 6 m/s is the Fermi velocity in graphene and l B = ¯ h/eB is the magnetic length. With this energy dispersion, the conductivities can be calculated using the Kubo formula, 27 σ xx (ω) = e 2 2π ¯ h n=0 B (ω + 1 ) [n F (E n ) n F (E n+1 )] + [n F (E n+1 ) n F (E n )] (ω + 1 ) 2 f 2 intra (n)ω 2 B f intra (n) + B (ω + 1 ) [n F (E n ) n F (E n+1 )] + [n F (E n+1 ) n F (E n )] (ω + 1 ) 2 f 2 inter (n)ω 2 B f inter (n) , (2a) σ xy (ω) = e 2 2π ¯ h n=0 {[n F (E n ) n F (E n+1 )] [n F (E n+1 ) n F (E n )]} × ω 2 B (ω + 1 ) 2 f 2 intra (n)ω 2 B + ω 2 B (ω + 1 ) 2 f 2 inter (n)ω 2 B , (2b) 125450-1 1098-0121/2012/86(12)/125450(5) ©2012 American Physical Society