## Abstract

Terahertz plasmons and magnetoplasmons propagating along electrically and chemically doped graphene p-n junctions are investigated. It is shown that such junctions support non-reciprocal magnetoplasmonic modes which get concentrated at the middle of the junction in one direction and split away from the middle of the junction in the other direction under the application of an external static magnetic field. This phenomenon follows from the combined effects of circular birefringence and carrier density non-uniformity. It can be exploited for the realization of plasmonic isolators.

© 2013 OSA

## 1. Introduction

The linear band structure, tunability and ambipolarity of graphene [1,2] have recently opened up new horizons in the area of plasmonics. These fundamental properties lead to unique plasmonic phenomena, such as the existence of both TE and TM plasmons [3–7] and voltage tunable plasmonic modes [8–11]. These modes have been recently investigated towards the realization of various enhanced plasmonic devices [3, 7, 10, 12–16].

Using electrical gating, one can modify and tune the charge profile to generate useful modes on graphene structures. For instance, one can obtain regions with opposite carrier types on a graphene strip by applying a tangential transverse electric field across it so as to create a p-n junction. In [9], it was theoretically shown at zero temperature that such a junction supports localized plasmonic and magnetoplasmonic modes with the magnetoplasmonic modes existing only for one direction of propagation (the modes in the other direction being cut off) in the limit of a very high magnetic bias field.

Allowing non-zero temperatures leads to the excitation of thermal carriers, which induce non-zero conductivity at the center of the junction and hence affects the electromagnetic field distribution across the strip. Moreover, allowing arbitrary magnetic bias fields leads to the existence of a circular birefringence regime where magnetoplasmonic modes exist in the two propagation directions but exhibit very different distributions, one being concentrated at the center of the junction and the other one being split away from it. This paper reveals these non-reciprocal phenomena and proposes isolator devices based on them.

The magnetoplasmonic modes are computed by numerically solving Maxwell equations for the graphene strip with exact non-uniform carrier densities in the Kubo conductivity, which properly captures the quantized Landau regime prevailing near the center of the junction. Two types of doping are considered and compared: doping by a transverse electric field and chemical doping, represented in Figs. 1(a) and 1(b), respectively. The corresponding structures are analyzed with the 2D finite difference frequency domain technique (FDFD) [17], where graphene is modeled as a zero-thickness conductive sheet with a conductivity tensor given by the Kubo formula [18, 19]. Adaptive mesh refinement is used around the edges and small features for better accuracy.

## 2. Magnetoplasmon energy concentration and splitting in graphene p-n junctions

#### 2.1. Electrically doped graphene

First consider the electrically doped graphene structure [Fig. 1(a)]. A transverse electric field applied tangentially to the graphene strip creates a non-uniform carrier density and therefore a non-uniform conductivity across the strip. The net charge profile can be found by solving the integral equation

*ρ*(

*x′*,

*y′*), where $G(x,y;{x}^{\prime},{y}^{\prime})=\frac{-1}{2\pi {\epsilon}_{0}}\text{ln}\sqrt{{({x}^{\prime}-x)}^{2}+{({y}^{\prime}-y)}^{2}}$ is the 2D free-space Green function for the Poisson equation and

*E*

_{0}is the applied electrostatic field. The resulting net carrier density,

*n*

_{net}=

*n*−

*p*, where

*n*and

*p*are the electron and hole densities, respectively, is plotted in Fig. 2 along with the electric potential for a graphene strip of width

*w*= 50

*μ*m. Different carrier types symmetrically appear at the opposite sides of the strip, which results in the formation of a p-n junction at the center of the strip.

The chemical potential (*μ _{c}*) and the electron and hole
densities are found from

*n*

_{net}by numerically solving the equation

*f*(

_{d}*E*,

*μ*) is the Fermi-Dirac distribution and $N(E)=\frac{2\left|E\right|}{\pi {\overline{h}}^{2}{v}_{f}^{2}}$ is the density of energy states. The integrals represent the electron and hole densities (

_{c}*n*and

*p*, respectively). The carrier densities are plotted in Fig. 3 while the corresponding chemical potential and Kubo conductivity are plotted in Fig. 4. Note that at the center of the strip, despite the zero net carrier density, a significant amount of thermally excited electrons and holes are present, which leads to the significant conductivity observed in Fig. 4. For simplicity, we assumed an energy independent scattering time of

*τ*= 0.1 ps. Moreover, we assumed that the graphene has less than 10

^{10}cm

^{−2}of unintentional doping fluctuations.

The graphene strip is next simulated as a conductive sheet in the FDFD resolution of Maxwell
equations in order to compute the plasmonic and magnetoplasmonic modes. The slow-wave factor and
loss for the plasmon modes propagating along the structure are shown in Fig. 5, where *k _{z}* and

*k*

_{0}are the propagation constant in the

*z*direction and the free space wave number, respectively. The structure supports a plasmon mode localized at the p-n junction, an infinite number of bulk modes (only the first four are shown here) and two edge modes (not shown here). The losses for the first few modes are shown in Fig. 5(b). Since the carrier density is low near the center of the graphene strip, the p-n junction mode has higher loss than the other modes. The farther the mode is from the center, the lower its loss is. Note that the p-n junction mode has a longitudinal current component that is maximal at the center. Since essentially no carriers are present at the center at zero temperature, this mode is not supported at zero temperature.

In the presence of a magnetic biasing field, the lowest magnetoplasmon mode exhibits particularly
interesting non-reciprocal properties. If the field is sufficiently high, it propagates only in one
direction, as shown in the dispersion diagram of Fig. 6 for
*B*_{0} = 0.1 T. As the magnetic field is increased, the forward mode
concentrates at the center, whereas its energy splits away from the center in the backward
direction, as shown in Fig. 6. Therefore, if the strip is
excited at its center, in the forward direction, the mode whose energy is localized at the junction
is excited, whereas no modes are excited in the backward direction, as there is no mode with energy
at the center. However, the source beam will need to be highly confined to avoid exciting the
backward mode. Note that the modes exhibit non-commensurate field patterns, with energy being
squeezed near the center, as a result of the non-uniform conductivity. In the chemically doped p-n
junction, to be studied next, it will be shown that the higher order modes exhibit commensurate
resonances with higher contrast between the field patterns of the p-n junction mode in the forward
and backward direction.

#### 2.2. Chemically doped graphene

The chemically doped p-n junction is shown in Fig. 1(b).
This structure is composed of two chemically doped graphene strips with opposite polarities forming
a p-n junction and separated by a nano-gap isolating electron and hole carriers. As the structure of
Fig. 1(a), this structure supports a plasmonic mode localized
at the middle of the p-n junction. The dispersion curves for the non-biased structure is shown in
Fig. 7. The structure supports two edge modes, an infinite
number of bulk modes, and the p-n junction mode, plotted in red. The unbiased structure has
symmetric dispersion for the forward and backward propagation directions. However, as the magnetic
bias is applied, time reversal symmetry is broken and the p-n junction mode exhibits different
properties for the forward and backward directions. The dispersion curves for the structure of Fig. 1(b) under magnetic bias is plotted in Fig. 8 for a magnetic bias of *B*_{0} = 1 T. It
is seen that the mode propagating at the junction exhibits very different properties for the forward
and backward directions. In the forward direction, this mode is concentrated at the center, whereas
in the backward direction it has very little energy at the center.

It should be noted that the electrically doped and chemically doped graphene junction structures presented in this section are isolators in the sense that propagation is supported along one direction only but not in the sense of conventional isolators [20]. In the latter case, the power sent in the stop direction is dissipated in the devices whereas in the former case it is reflected by the device. Another point to notice is that, although we are considering modal excitation of a multimode structure, the time reversal symmetry breaking is inherent to our structure in the presence of all the modes, due to the odd nature of the magnetic biasing field under time reversal symmetry. Therefore, the proposed isolator represents a true optical isolator in the sense of [21].

## 3. Phenomenological explanation

The non-reciprocal property of the magnetically biased structures of Fig. 1(a) and Fig. 1(b), results from
the circular birefringence for the two propagation directions. The electric field pattern of the p-n
junction mode in the plane of graphene is shown in Fig. 9 for
the unbiased structure of Fig. 1(b). As the wave propagates
along the strip in the forward direction, point *R* on the right strip sees a
clock-wise rotating electric field, and point L on the left strip sees a counter clock-wise rotating
electric field. In magnetically biased graphene, which is characterized by a conductivity tensor
*σ̿* =
*σ _{d}*(

**x̂x̂**+

**ẑẑ**) +

*σ*(

_{o}**x̂ẑ**−

**ẑx̂**), where

*σ*and

_{d}*σ*are the diagonal and off-diagonal conductivities, respectively, the right and left-hand circularly polarized waves see different scalar conductivities,

_{o}*σ*+

_{d}*jσ*and

_{o}*σ*−

_{d}*jσ*, respectively. On the other hand, when a magnetic field is applied, the p-doped and n-doped strips have off-diagonal conductivities with opposite signs. As a result, for the forward propagation both strips see the conductivity

_{o}*σ*−

_{d}*jσ*while for the backward direction the conductivity seen by the wave is

_{o}*σ*+

_{d}*jσ*. Therefore, the mode sees different media for different propagation directions, corresponding to different dispersions, as observed in Fig. 8. In the forward direction, the p-n junction mode sees a conductivity with a slightly higher imaginary part and is therefore more concentrated. In the opposite direction, the imaginary part of the conductivity is slightly decreased and the mode becomes less localized. The evolution of mode 1 with increasing magnetic field is plotted in Fig. 10 for the forward and backward directions. As the magnetic field is increased, in the forward direction the mode gradually becomes more concentrated on the p-n junction, whereas in the backward direction it moves away from the center.

_{o}## 4. Conclusions

The energy concentration and splitting effect allows the realization of non-reciprocal plasmonic
devices such as isolators. If the structure is excited at the center, in the forward direction the
p-n junction mode is excited, whereas in the backward direction there is only negligible coupling to
the backward mode which has little energy at the center. The electrically doped isolator will be
very lossy however, as the carrier density is low at the center (the loss is shown in Fig. 6(b)). In a chemically doped graphene p-n junctions the loss
can be mitigated by appropriate doping (the loss is shown in Fig.
8(b) for *n* = *p* = 10^{13}).

## References and links

**1. **K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva, and A. A. Firsov, “Electric field effect in atomically thin carbon
films,” Science 22**306**, 666–669 (2004). [CrossRef]

**2. **A. K. Geim and K. S. Novoselov, “The rise of graphene,”
Nature Materials **6**, 183–191 (2007). [CrossRef] [PubMed]

**3. **A. N. Grigorenko, M. Polini, and K. S. Novoselov, “Graphene plasmonics,” Nat.
Photon. **6**, 7490 758 (2012). [CrossRef]

**4. **S. A. Mikhailov and K. Ziegler, “New Electromagnetic Mode in
Graphene,” Phys. Rev. Lett. **99**, 016 803 (2007). [CrossRef]

**5. **G. W. Hanson, “Dyadic Greens functions and guided surface waves for a
surface conductivity model of graphene,” J. Appl. Phys. **103**, 064 302 (2008). [CrossRef]

**6. **I. Crassee, M. Orlita, M. Potemski, A. L. Walter, M. Ostler, T. Seyller, I. Gaponenko, J. Chen, and A. B. Kuzmenko, “Intrinsic terahertz plasmons and magnetoplasmons in large
scale monolayer graphene,” Nano Letters **12**, 2470–2474 (2012). [CrossRef] [PubMed]

**7. **A. Vakil and N. Engheta, “Transformation Optics Using
Graphene,” Science **332**, 1291–1294 (2011). [CrossRef] [PubMed]

**8. **P. G. Silvestrov and K. B. Efetov, “Charge accumulation at the boundaries of a graphene strip
induced by a gate voltage: Electrostatic approach,” Phys. Rev.
B **77**, 155 436 (2008). [CrossRef]

**9. **E. G. Mishchenko, A. V. Shytov, and P. G. Silvestrov, “Guided Plasmons in Graphene p-n
Junctions,” Phys. Rev. Lett. **104**, 156 806 (2010). [CrossRef]

**10. **S. Thongrattanasiri, I. Silveiro, and F. J. G. de Abajo, “Plasmons in electrostatically doped
graphene,” Appl. Phys. Lett. **100**, 201 105 (2012). [CrossRef]

**11. **I. Petković, F. I. B. Williams, K. Bennaceur, F. Portier, P. Roche, and D. C. Glattli, “Carrier Drift Velocity and Edge Magnetoplasmons in
Graphene,” Phys. Rev. Lett. **110**, 016 801 (2013). [CrossRef]

**12. **T. Echtermeyer, L. Britnell, P. Jasnos, A. Lombardo, R. Gorbachev, A. Grigorenko, A. Geim, A. Ferrari, and K. Novoselov, “Strong plasmonic enhancement of photovoltage in
graphene,” Nat. Commun. **2**, 458 (2011). [CrossRef] [PubMed]

**13. **T. Mueller, F. Xia, and P. Avouris, “Graphene photodetectors for high-speed optical
communications,” Nat. Photon. **4**, 297–301 (2010). [CrossRef]

**14. **N. M. Gabor, J. C. W. Song, Q. Ma, N. L. Nair, T. Taychatanapat, K. Watanabe, T. Taniguchi, L. S. Levitov, and P. Jarillo-Herrero, “Hot carrierassisted intrinsic photoresponse in
graphene,” Science **334**, 648–652 (2011). [CrossRef] [PubMed]

**15. **N. Chamanara, D. Sounas, and C. Caloz, “Non-reciprocal magnetoplasmon graphene
coupler,” Opt. Express **21**, 11248–11256 (2013). [CrossRef] [PubMed]

**16. **D. L. Sounas and C. Caloz, “Edge surface modes in magnetically biased chemically doped
graphene strips,” Appl. Phys. Lett. **99**, 231 902:13 (2011). [CrossRef]

**17. **Y. Zhao, K. Wu, and K. M. Cheng, “A Compact 2-D Full-Wave Finite-Difference Frequency-Domain
Method for General Guided Wave Structures,” IEEE Trans. Microwave
Theory Tech. **50**, 1844–1848 (2002). [CrossRef]

**18. **V. P. Gusynin, S. G. Sharapov, and J. P. Carbotte, “Magneto-optical conductivity in
graphene,” Journal of Physics: Condensed Matter **19**, 026 222 (2007). [CrossRef]

**19. **V. P. Gusynin, S. G. Sharapov, and J. P. Carbotte, “On the universal ac optical background in
graphene,” New Journal of Physics **11**, 095 013 (2009). [CrossRef]

**20. **D. M. Pozar, *Microwave engineering*(Danvers, MA:
Wiley, 2005), 3rd
edn.

**21. **D. Jalas, A. Petrov, M. Eich, W. Freude, S. Fan, Z. Yu, and H. Renner, “What is and what is not an optical
isolator,” Nature Photonics **7**, 579–582 (2013). [CrossRef]