Quasi-Flat Plasmonic Bands in Twisted Bilayer Graphene - Nano

Oct 24, 2016 - Abstract: Twisted bilayer graphene (tBLG) with van Hove Singularity (VHS) has exhibited novel twist-angle-dependent chemical and physic...
3 downloads 4 Views 4MB Size
Letter pubs.acs.org/NanoLett

Quasi-Flat Plasmonic Bands in Twisted Bilayer Graphene Tobias Stauber*,† and Heinerich Kohler†,‡ †

Departamento de Teoría y Simulación de Materiales, Instituto de Ciencia de Materiales de Madrid, CSIC, 28049 Madrid, Spain Fakultät für Physik, Universität Duisburg-Essen, Lotharstrasse 1, 47048 Duisburg, Germany



S Supporting Information *

ABSTRACT: The charge susceptibility of twisted bilayer graphene is investigated in the Dirac cone, respectively, random-phase approximation. For small enough twist angles θ ≲ 2°, we find weakly Landau damped interband plasmons, that is, collective excitonic modes that exist in the undoped material with an almost constant energy dispersion. In this regime, the loss function can be described as a Fano resonance, and we argue that these excitations arise from the interaction of quasilocalized states with the incident light field. These predictions can be tested by nanoinfrared imaging and possible applications include a “perfect” lens without the need of lefthanded materials. KEYWORDS: Twisted bilayer graphene, plasmons, local field effects, Fano resonance, perfect lens

W

potential and, therefore, can be interpreted as interband plasmons or collective excitonic oscillations. As argued above, terahertz plasmons in graphene and/or superlattices are only present at finite chemical.20,21 But socalled π-plasmons can also be observed in neutral free-standing graphene at energies ωπp ≳ 4.5 eV.22,23 These are related to a van Hove singularity,24 and an obvious guess would be that there will be similar plasmonic excitations at lower energies due to the appearance of emerging van Hove singularities located between the two Dirac cones of the two twisted layers.25,26 However, the above plasmons are invoked by delocalized πelectrons, whereas the plasmons discussed here originate from quasi-localized states, reminiscent to a recent study on localized plasmons in disordered graphene27 and bilayer nanodisks.28 The emergence of interband plasmonic modes around the neutrality point is related to the deviation from Dirac Fermion behavior in the charge response χ, that is, the imaginary part Im χ has to decay faster than ω−1 for ω → ∞. Interband (out-ofphase) plasmons with a linear (sound-like) dispersion should therefore be hosted by topological insulators such as mercury telluride described by the BHZ-model which mixes Dirac with Schrödinger electrons.29,30 In the case of twisted bilayer graphene, we also find deviations from the typical Dirac response for small twist angles, however, here the emerging interband plasmons arise through the interaction of the incidence light with quasi-localized states, displaying an almost constant dispersion with energies ℏω ∼ 20−200 meV, tunable by the twist angle. Furthermore, they carry a dipole moment

ith the discovery of graphene and other two-dimensional (2D) crystals,1 the field of plasmonics has received renewed attention.2−7 In particular, single-layer graphene on hexagonal boron-nitride (h-BN) displays outstanding properties, hosting long-lived plasmons with life times of the order of 500 fs and offering the possibility of tuning the plasmonic resonances via an electrostatic gate.8 The plasmonic modes can also be modified when the 2D materials form Moiré patterns with the underlying substrate and emerging highenergy modes were observed for graphene on top of h-BN.9,10 Along these lines, twisted bilayer graphene offers new perspectives for tuning the electromagnetic response by changing the twist angle.11−15 Here, we will investigate the plasmonic spectrum of twisted bilayer graphene using the continuous model by Lopes-Santos et al.,16 that is, we extend previous results for the local conductivity to finite momentum transfer.17,18 For small enough twist angles θ ≲ 2°, we will find novel weakly Landau-damped interband plasmons, that is, collective excitonic modes that exist in the undoped material with an almost constant energy dispersion, arising from quasilocalized states. Plasmons are collective charge oscillations leading to nanoscale optical fields and thus they are linked to the existence of a plasma, that is, to a finite charge stiffness or Drude weight, D. Within a hydrodynamic model, the plasmon energy is related to the Drude weight as ωp ∼ √D which rules out the existence of plasmons for neutral systems for which D = 0. Nevertheless, here we will show that for a sufficiently small twist angle close to the magic angle at which the Fermi velocity becomes zero and flat bands develop,19 genuine collective modes emerge even in the case of zero chemical potential. These excitations prevail for not too large finite chemical © XXXX American Chemical Society

Received: June 23, 2016 Revised: September 13, 2016 Published: October 24, 2016 A

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. (A) Real space image of twisted bilayer: Regions of AA-stacked graphene are surrounded by regions of AB-stacked graphene which form the Morié-pattern. (B) Band-structure with corresponding density of states and (C1)−(C6) the local density corresponding to the first six conduction bands (CB). Clearly seen are the localized states inside the AA-stacked regions, especially in the case of the lowest CB. In all cases, the twist angle corresponds to i = 20 (θ20 = 1.61°). (D) Sketch of the Brillouin zone of twisted bilayer for a large angle with i = 3, θ3 = 9.14°. Details are given in the Supporting Information.

(in-phase plasmons) and, therefore, it should be easier to observe genuine interband plasmons in twisted bilayer graphene with twist angles of θ ≲ 2° via, for example, nanoinfrared imaging or scattering-type scanning near-field optical microscopy (s-SNOM).31,32 Before we outline the explicit calculations, let us specify our definition of plasmonic excitations. Often, a peak in the electron energy loss function serves as criterion for a plasmon mode. Nevertheless, this is only an indication for an enhanced charge response and not for collective oscillations that are indicated by a pole in the two-particle Green’s function or, alternatively, a zero in the (real part of) the dielectric function. Within this definition, the modes found in refs 33−36 for undoped graphene or energies larger than twice the chemical potential on various substrates are not genuine plasmons as discussed in ref 4. Model. In order to describe twisted bilayer graphene, we follow refs 16 and 37 with the intra(inter)layer hopping amplitude t0 = 2.78 (t⊥ = −0.33) eV. A self-contained discussion on the model is given in the Supporting Information (SI).38 In the following, we focus on a discrete set of twist angles θi, which are labeled by only one integer i through 1 cos(θi) = 1 − 2A with Ai = 3i2 + 3i + 1. The unit cell of

single or bilayer graphene and the unit vector thus of size Ai a with a ≈ 2.46 Å. In Figure 1, (A) the real-space image, (B) is the band-structure, and (Cn) is the local density corresponding to the nth conduction band are shown for an twist angle θi=20 = 1.61°, as well as the extended Brillouin zone for θi=3 = 9.14° (D). For large angles, the spectrum can be approximately described by a single parameter α =

A i / 3 t⊥ 2π t 0

and perturbation

theory correctly predicts the emerging Dirac cone physics with renormalized Fermi velocity given by v*F ≈ vF(1−3α2)/(1 + 6α2) with ℏvF = √3at0/2.16,19 But for small twist angles with θ < 2°, a new regime occurs where the bands are very flat and the velocity tends to zero,39 see Figure 1 (C1). It is this new regime that will give rise to novel interband plasmons. Dynamical Charge Response. The dynamical density− density response function is defined as 1 χ (r, r′, t ) = iℏ Θ(t )⟨[n(̂ r, t ), n(̂ r′)]⟩ and for |q| ≪ |Gi| its

Fourier transform in space and time, χ(q,ω), is well-defined. Let Eτm(k) and |τ,m,k⟩ denote eigenvalues and eigenfunctions of the effective Hamiltonian given in the SI.38 Then χ(q,ω) can be expressed in the long wavelength limit as

i

twisted bilayer graphene is Ai times larger than the unit cell of B

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. Imaginary (A) and real (B) part of the charge response χ(q,ω) for different twist angles θ5 ≈ 6.01° (black), θ10 ≈ 3.14° (red), θ15 ≈ 2.13° (green), and θ20 ≈ 1.61° (blue). The in-plane momentum transfer is qa = 0.02 in the direction of ΔK and the dashed lines indicate the zero of the real part of eq 3 for an effective dielectric medium with ϵ = 1. Density plot for the loss function S(q,ω) = −Im ϵ−1(q,ω) of twisted bilayer graphene with a twist angle θ20 ≈ 1.61° with ϵ = 1 (C) and θ25 ≈ 1.30° with ϵ = 2.4 (D) as a function of frequency ω and in-plane momentum q. (E,F) Density plot for the loss function S = −Im ϵ−1(q,ω) of twisted bilayer graphene with a twist angle θ20 ≈ 1.61° at fixed wavenumber qa = 0.02 and ϵ = 1 as a function of frequency ω and chemical potential μ for two different scales. Also shown are vertical dashed lines at μ/t0 = 0.01,0.065 and horizontal dotted lines at ℏω/t0 = 0.03,0.075 as guide for the eye. χ (q, ω) =

gs V

Static Response. For large twist angle, the electronic spectrum of twisted bilayer graphene is characterized by two Dirac cones per valley and the static susceptibility thus scales linear with the wavenumber, that is, Re χ(q,0) ∼ |q| as seen in eq 2. Remarkably, for small twist angles θ ≲ θi=20, the static response becomes quadratic for small momenta, that is, χ(q,0) ∼ |q|2, see SI.38 This departure from Dirac cone physics is crucial to host genuine plasmons at zero doping.29,30 Numerical Results. In Figure 2A, the imaginary part of χ(q,ω) is shown for various angles at constant momentum transfer qa = 0.02 in the direction of ΔK = K − Kθ. With ω* denoting the frequency at which Im χ becomes maximal, Im χ (q, ω*) ω* is independent of i for i ≲ 15 consistent with a square-root divergency. Furthermore, we have ω* = v*F q for i ≲ 15 whereas for smaller angles with i ≳ 15, ω*>vF*q and Im χ (q, ω*) ω* is not constant anymore. We thus observe a departure from Dirac cone physics for small angles in the charge response. The crossover behavior around i ≈ 15 can also be observed in the real part of the susceptibility, Figure 2B. For i ≳15, Re χ becomes positive for certain frequencies and this opens up the possibility for the existence of genuine plasmons with energy ℏωp. This is indicated by the dashed line where the dielectric function within the random-phase approximation (RPA) becomes zero

∑∑ ∑ τ =± m , n k ∈ 1.BZ

|⟨τ , n , k + q|τ , m , k⟩|2

nF[Emτ (k)] − nF[Enτ (k + q)] ℏω − Enτ (k + q) + Emτ (k) + i 0

(1) β(x−μ)

−1

Here, nF[x] = (e + 1) is the Fermi function and gs = 2 is the spin degeneracy. The sum over k is over the first Brillouin zone of the supercell and m,n denotes the band indices. Note that eq 1 comprises eigenvalues and eigenstates of both inequivalent Dirac points τ, such that χ(q,ω) manifestly fulfils the usual symmetry relations of a response function,40 namely χ(q,ω) = χ*(q,−ω) and χ(q,ω) = χ(−q,−ω) due to time reversal invariance and a real response, respectively. For energies ℏω ≫ t⊥, the effects of the interlayer coupling become negligible and the result for two decoupled graphene monolayer at zero temperature and zero chemical potential41 χ0 (q, ω) =

−igl g v gs

q2

16ℏ

ω 2 − (vFq)2

(2)

must be recovered (gl = 2 and gv = 2 are the layer and the valley degeneracy). This result holds for vanishing coupling strength t⊥ or, more precisely, for vanishing α. For small α, the renormalization of the Fermi velocity is expected to be the main effect and the response function should be well described by χ*0 (q,ω), which is defined as χ0(q,ω) in eq 2 but with vF replaced by vF*. Obviously the locus of the singularity of χ* moves toward zero as the Fermi velocity decreases while its spectral weight increases with 1/ vF* . This can be seen in Figure 2A for large twist angles with i = 5,10.

ϵ(q, ωp) = 1 − vqχ (q, ωp) = 0

(3)

Above, we defined the Coulomb potential vq =

e2 2ϵ0ϵ |q|

for an

effective dielectric medium with static dielectric constant ϵ. C

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

collective excitonic oscillations. In fact, also for larger interlayer bias Δ ≳ ℏωmp , the resonances persist supported by the local gap present in the AB-stacked regions, see SI.38 Disorder can be qualitatively modeled by introducing a finite damping term in eq 1. Numerically, we first obtain Imχ and then Reχ, invoking the Kramers−Kronig relation. A moderate broadening in Imχ and consequently in Reχ does not alter our general predictions. The same holds for finite room-temperature, see SI.38 Real Space Interpretation. The novel plasmon modes consist of collective interband transitions and, therefore, the corresponding electron and hole densities are equal. In an extended systems, electron and hole densities must move out of phase in order to generate a restoring force which maintains the charge oscillations. If the system is partially confined due to an external potential, electron and hole densities can also move inphase making them susceptible to dipole coupling to an incident light field. Assuming the confinement to be harmonic, the spectrum is given by equally spaced energy levels. Moreover, the center-ofmass equation of motion is linear and all Fourier-components move with the same frequency, that is, the dispersion is constant and independent of q. Both features are reflected by the loss function of twisted bilayer graphene that has to be contrasted to the case of interlayer plasmons in mercury telluride that shows a linear dispersion in accordance to out-ofphase oscillations that do not couple to light.29,30 A simple model describing these quasi-confined regions is discussed in the SI.38 Exciting the system by s-SNOM, a particle-hole or excitonic density is created, oscillating within several adjacent AA-stacked regions of quasi-localized wave functions. Moreover, there is a linear shift in the resonant plasmon energy for different twist angles with ωmp ∼ 1/R, where R ∼ Ai ∼ i denotes the radius of the localized AA-stacked region, which is approximately linear for large i ≳ 15, see SI.38 Applications. A plasmonic resonance with almost constant dispersion at ℏω0 opens up several possible applications. Let us highlight here a device with two twisted bilayer graphene layers on top and on the bottom of a dielectric ϵ of width d. Following ref 44, we find exponential amplification of the near-field modes at constant energies ω ex p,1 = ω 0 + O(e − 2qd ) and ϵ−1 ωexp,2 = ϵ + 1 ω0 + O(e−2qd). A “perfect” lens in the spirit of Pendry45 can thus be designed without the need of left-handed materials. Also extraordinary absorption of propagating light at ℏω0 is expected due to coupling to the reciprocal vector of the Moirésuperlattice. For a polarization in direction of G1 and twist angle θi=25, we have |G1|a ≈ 0.16 and peaks in the loss function correspond to enhanced absorption. Summary and Discussion. We have predicted novel interband-plasmons in undoped twisted bilayer graphene for small twist angles. Moreover, we showed that the plasmonic excitations are connected to the deviations of Dirac cone physics and consequently to quasi-localized states giving rise to Fano resonances. This makes twisted bilayer graphene an exciting new metamaterial with extraordinary properties leading to enhanced absorption and exponential amplification at constant energy giving rise to the possibility of a perfect lens without the need of left-handed materials. The new interband plasmonic modes can be interpreted as collective excitonic in-phase oscillations in a periodic but quasi-

This has been set equal to one (vacuum) in the corresponding dashed curve and we will also discuss results for a finite value with ϵ = 2.4, corresponding to a twisted bilayer graphene on top of SiO2. Loss Function. Undamped plasmons only exist if ϵ(q,ω) = 0. Nevertheless, plasmons with frequency ωp can also be defined for Im χ ≠ 0 by the condition Re ϵ(q,ωp) = 0 as long as the loss function S(q,ω) = −Im ε−1(q,ω) is peaked around ωp with width γ ≪ ωp. This condition allows the plasmon to oscillate sufficiently long before decaying through Landau damping into the particle−hole continuum and renders it detectable by, for example, nanoinfrared imaging.31,32 In Figure 2C, the density plot of the loss function for a surrounding medium with ϵ = 1 is shown as a function of momentum and of frequency for a twist angle θ20 ≈ 1.61°. The in-plane momentum vector points into the direction of ΔK but the results hardly depend on the direction of q. In Figure 2D, the loss function is shown for θ25 ≈ 1.30° with ϵ = 2.4. For this angle, even with an effective dielectric medium up to ϵ = 4, genuine plasmons with Re ϵ(q,ωp) = 0 are present at T = 0. For both angles, we observe several almost equally spaced plasmon branches extending to large q-values. These quasi-localized plasmons emerge from quasi-localized eigenstates as we will argue below and are the main observation of this work. For smaller momentum transfer qa ≈ 0.02, the two lowest resonances show an asymmetric line shape that can be well fitted by a Fano resonance, see SI.38 This is because the loss function is directly related to the extinction spectrum for which Fano resonances are well-known for confined plasmonic systems and that occur when localized states interact with a continuum.42 The asymmetry increases with decreasing angle and also the peak position shifts to lower energies, see SI.38 These results go in line with the stronger localization for twist angles close to the magic angle at θi=31 ≈ 1.05° as well as the increasing dot size given by the AA-stacked island proportional to Ai ∝ i . Local Field Effects. For small angles and/or large wavenumber, local field effects have to be taken into account because the wavenumber q becomes comparable to the length of the first reciprocal lattice vector |Gi| = |G0| / Ai /3 with |G0| a = 4π/√3. In the SI, we analyze the local field effects on the loss function and conclude that there are no significant changes, that is, the plasmonic resonances are only slightly shifted but prevail.38 Finite Doping, Interlayer Bias, Disorder, and Temperature. For finite chemical potential with μ ≲ ℏωmp with m counting the plasmonic resonances, the peaks with energy ωmp prevail. This supports our interpretation of the collective excitonic excitations due to interband transitions, that is, novel interband plasmons due to the hybridization of the localized states with the incident light field. Plasmons can thus be tuned and quenched/enhanced by changing the twist angle and chemical potential, respectively, as seen in Figure 2E,F, which shows the loss function as a function of frequency and chemical potential at fixed in-plane momentum qa = 0.02 for two different scales. By applying an interlayer bias Δ, a gap is opened in the spectrum of Bernal(AB)-stacked graphene bilayer,43 but for twisted bilayer graphene only the energy levels of the two Dirac points of one valley are shifted to positive and negative energies, respectively. Again, the localized plasmon modes are preserved for Δ ≲ ℏωmp , showing the robustness of these D

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

(11) Brihuega, I.; Mallet, P.; González-Herrero, H.; Trambly de Laissardière, G.; Ugeda, M. M.; Magaud, L.; Gómez-Rodríguez, J. M.; Ynduráin, F.; Veuillen, J.-Y. Phys. Rev. Lett. 2012, 109, 196802. (12) Havener, R. W.; Liang, Y.; Brown, L.; Yang, L.; Park, J. Nano Lett. 2014, 14, 3353−3357. (13) Schmidt, H.; Rode, J. C.; Smirnov, D.; Haug, R. J. Nat. Commun. 2014, 5, 5742. (14) Patel, H.; Havener, R. W.; Brown, L.; Liang, Y.; Yang, L.; Park, J.; Graham, M. W. Nano Lett. 2015, 15, 5932−5937. (15) Yin, J.; Wang, H.; Peng, H.; Tan, Z.; Liao, L.; Lin, L.; Sun, X.; Koh, A. L.; Chen, Y.; Peng, H.; Liu, Z. Nat. Commun. 2016, 7, 10699. (16) Lopes dos Santos, J. M. B.; Peres, N. M. R.; Castro Neto, A. H. Phys. Rev. Lett. 2007, 99, 256802. (17) Moon, P.; Koshino, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 87, 205404. (18) Stauber, T.; San-Jose, P.; Brey, L. New J. Phys. 2013, 15, 113050. (19) Bistritzer, R.; MacDonald, A. H. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 12233−12237. (20) Shung, K. W. K. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 34, 979−993. (21) Hawrylak, P.; Wu, J.-W.; Quinn, J. J. Phys. Rev. B: Condens. Matter Mater. Phys. 1985, 31, 7855−7858. (22) Eberlein, T.; Bangert, U.; Nair, R. R.; Jones, R.; Gass, M.; Bleloch, A. L.; Novoselov, K. S.; Geim, A.; Briddon, P. R. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 77, 233406. (23) Kinyanjui, M. K.; Kramberger, C.; Pichler, T.; Meyer, J. C.; Wachsmuth, P.; Benner, G.; Kaiser, U. EPL (Europhysics Letters) 2012, 97, 57005. (24) Stauber, T.; Schliemann, J.; Peres, N. M. R. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81, 085409. (25) Li, G.; Luican, A.; Lopes dos Santos, J. M. B.; Castro Neto, A. H.; Reina, A.; Kong, J.; Andrei, E. Y. Nat. Phys. 2009, 6, 109−113. (26) Luican, A.; Li, G.; Reina, A.; Kong, J.; Nair, R. R.; Novoselov, K. S.; Geim, A. K.; Andrei, E. Y. Phys. Rev. Lett. 2011, 106, 126802. (27) Muniz, R. A.; Dahal, H. P.; Balatsky, A. V.; Haas, S. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 081411. (28) Wang, W.; Xiao, S.; Mortensen, N. A. Phys. Rev. B: Condens. Matter Mater. Phys. 2016, 93, 165407. (29) Juergens, S.; Michetti, P.; Trauzettel, B. Phys. Rev. Lett. 2014, 112, 076804. (30) Juergens, S.; Michetti, P.; Trauzettel, B. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 115425. (31) Chen, J.; Badioli, M.; Alonso-Gonzalez, P.; Thongrattanasiri, S.; Huth, F.; Osmond, J.; Spasenovic, M.; Centeno, A.; Pesquera, A.; Godignon, P.; Zurutuza Elorza, A.; Camara, N.; de Abajo, F. J. G.; Hillenbrand, R.; Koppens, F. H. L. Nature 2012, 487, 77. (32) Fei, Z.; Rodin, A. S.; Andreev, G. O.; Bao, W.; McLeod, A. S.; Wagner, M.; Zhang, L. M.; Zhao, Z.; Thiemens, M.; Dominguez, G.; Fogler, M. M.; Neto, A. H. C.; Lau, C. N.; Keilmann, F.; Basov, D. N. Nature 2012, 487, 82. (33) Liu, Y.; Willis, R. F.; Emtsev, K. V.; Seyller, T. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 78, 201403. (34) Tegenkamp, C.; Pfnür, H.; Langer, T.; Baringhaus, J.; Schumacher, H. W. J. Phys.: Condens. Matter 2011, 23, 012001. (35) Politano, A.; Marino, A. R.; Formoso, V.; Farías, D.; Miranda, R.; Chiarello, G. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 84, 033401. (36) Langer, T.; Förster, D. F.; Busse, C.; Michely, T.; Pfnür, H.; Tegenkamp, C. New J. Phys. 2011, 13, 053006. (37) Mele, E. J. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81, 161405. (38) See Supporting Information (SI). (39) Trambly de Laissardière, G. T.; Mayou, D.; Magaud, L. Nano Lett. 2010, 10, 804−808. (40) Giuliani, G.; Vignale, G. Quantum Theory of the Electron Liquid; Cambridge Cambridge University Press: Cambridge, 2005. (41) Gonzalez, J.; Guinea, F.; Vozmediano, M. Nucl. Phys. B 1994, 424, 595−618.

confining potential surrounding the AA-stacked regions. We thus expect these modes to also emerge in other systems with electronic (quasi-)confinement and/or commensurate structure. Methods. The continuous model for twisted bilayer graphene was used to discuss the charge response. We first calculated the imaginary part of the Lindhard function by approximating the delta-function by a Gaussian/Lorentzian with appropriate half-width. We then employed the Kramers− Kronig relation in order to obtain the real part. For this, we approximated the response for large energies by the one of two uncoupled, undoped graphene sheets. The same procedure was used to discuss the local field effects by decomposing the real and imaginary parts accordingly. The local density of states of the circularly quasi-confined graphene dot by a finite massboundary was obtained by wave function matching.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b02587. Introduction to the continuous model for twisted bilayer graphene, discussion on the static response, angle dependence of the loss function, local field effects, finite doping and bias, and real space interpretation. (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Luis Brey and T. S. Guillermo GómezSantos for useful discussions. This work has been supported by Spain’s MINECO under Grant FIS2014-57432-P and by the Comunidad de Madrid under Grant S2013/MIT-3007 MAD2D-CM.



REFERENCES

(1) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 10451−10453. (2) Koppens, F. H. L.; Chang, D. E.; Garcia de Abajo, F. J. Nano Lett. 2011, 11, 3370−3377. (3) Grigorenko, A. N.; Polini, M.; Novoselov, K. S. Nat. Photonics 2012, 6, 749−758. (4) Stauber, T. J. Phys.: Condens. Matter 2014, 26, 123201. (5) Garcia de Abajo, F. J. G. ACS Photonics 2014, 1, 135−152. (6) Low, T.; Avouris, P. ACS Nano 2014, 8, 1086−1101. (7) Gonçalves, P. A. D.; Peres, N. M. R. An Introduction to Graphene Plasmonics; World Scientific: Singapore, 2016. (8) Woessner, A.; Lundeberg, M. B.; Gao, Y.; Principi, A.; AlonsoGonzález, P.; Carrega, M.; Watanabe, K.; Taniguchi, T.; Vignale, G.; Polini, M.; Hone, J.; Hillenbrand, R.; Koppens, F. H. L. Nat. Mater. 2014, 14, 421−425. (9) Tomadin, A.; Guinea, F.; Polini, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 90, 161406. (10) Ni, G. X.; Wang, H.; Wu, J. S.; Fei, Z.; Goldflam, M. D.; Keilmann, F.; Ozyilmaz, B.; Castro Neto, A. H.; Xie, X. M.; Fogler, M. M.; Basov, D. N. Nat. Mater. 2015, 14, 1217−1222. E

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters (42) Giannini, V.; Francescato, Y.; Amrania, H.; Phillips, C. C.; Maier, S. A. Nano Lett. 2011, 11, 2835−2840. (43) McCann, E. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 161403. (44) Stauber, T.; Gómez-Santos, G. Phys. Rev. B: Condens. Matter Mater. Phys. 2012, 85, 075410. (45) Pendry, J. B. Phys. Rev. Lett. 2000, 85, 3966−3969.

F

DOI: 10.1021/acs.nanolett.6b02587 Nano Lett. XXXX, XXX, XXX−XXX