This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final Accepted for publication - to be published in 2015 publication. Citation information: DOI 10.1109/LAWP.2014.2378174, IEEE Antennas and Wireless Propagation Letters, 2014. IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, APRIL 2014 (REVIEWED ON SEPTEMBER 23, 2014) 1 FDTD Formulation for Graphene Modelling Based on Piecewise Linear Recursive Convolution and Thin Material Sheets Techniques Rodrigo M. S. de Oliveira∗ , Nilton R. N. M. Rodrigues† and Victor Dmitriev‡ Federal University of Pará (UFPA) - Institute of Technology (ITEC) Belém, Pará, PO Box 8619, Zip Code 66073-900, Brazil. E-mails: ∗ [email protected], † [email protected], ‡ [email protected] Abstract—A finite-difference time-domain formulation based on piecewise linear recursive convolution method and on thin material sheets technique is developed for modelling terahertz graphene antennas and some other photonic components. The graphene sheets are modelled by specific recursive equations obtained for tangential electric field components allowing one to apply easily voltage or current sources between the sheets. The effective conductivity of graphene sheets in Yee’s threedimensional lattice is calculated and used in simulations. A bowtie-like geometry is investigated, aiming at resonance tuning. The developed numerical formulation is validated by comparison of results with data published in literature. Keywords—Graphene, Nanoantennas, FDTD method, piecewise linear recursive convolution, thin material sheets, terahertz devices. I. I NTRODUCTION RAPHENE consists of carbon atoms arranged in a two-dimensional honeycomb hexagonal lattice forming a sheet with the thickness of a single atom [1]. Owing to its potential applications this material recently has attracted tremendous research interest in many fields of technology [2] including nanophotonics and in particular in terahertz antennas [3]–[6]. The main attractive characteristics of graphene are a possibility of dynamic spectral reconfigurability (by modifying chemical potential) [5], relatively low losses in terahertz band (in comparison with metals) [7] and a possibility to miniaturize antennas (up to ∼ 6% of the wavelength in free space) due to plasmonic effects [3]. We consider a small-signal regime where the influence of electromagnetic field of surface plasmon polariton (SPP) modes on graphene chemical potential is negligible. This way, graphene conductivity is modelled usually by the Kubo formalism which can include both intraband and interband components [8]. One of the challenges in the numerical calculations of such structures is different scales of dimensions of the real devices. The length and width of the devices can be 3-4 orders higher than their thickness [3]–[5], [9], [10]. In several recently published works on the graphene antennas and related topics, calculations were performed by some commercial simulators [3]–[5]. Such calculations require a considerable amount of computational resources because of high discretization levels which are necessary to represent graphene sheets. Besides, those simulators are usually based on frequency domain techniques where one simulation is required for each frequency of interest. In contrast, with time domain G techniques the impulsive response of a given structure can be obtained with a single simulation run [11]. Recently, Transmission-Line Modeling (TLM) [12] and analytical time-domain models [13] have been applied for modelling graphene devices. A widely known, powerful and yet relatively simple method in computational electrodynamics is the Finite-Difference Time-Domain (FDTD) method [11]. Recently, FDTD has been used for modelling graphene-based devices [9] [10]. In [9], a graphene conductivity formulation based on multiple complex conjugate pole-residue pairs is presented. In [10], a precise surface boundary condition technique is proposed for modelling graphene sheets laying at magnetic field planes of Yee’s cell (exploiting the TM plasmonic mode ~ components can make in graphene). However, the use of H it difficult to establish controlled voltage or current between two (or more) graphene sheets due to the staggered nature of Yee’s cell [11]. In this work, we develop a FDTD formulation based on piecewise linear recursive convolution (PLRC) [14] and on thin material sheets [15] techniques for including graphene on planes of Yee’s cells with electric field components tangential to the sheets. This is possible because graphene plasmonic mode is also characterized by tangential current densities [4], [8]. The proposed formulation has two main advantages: 1) it is mathematically very simple and precise, and 2) correct numerical solutions are obtained for graphene sheets placed ~ field planes in Yee’s cells, simplifying FDTD on transverse E modelling photonic devices fed by voltage or current sources. II. M ATHEMATICAL F ORMULATION Graphene’s intraband dispersive behavior can be modelled by the surface conductivity σ∗ 1 σ̃(ω) = , (1) d jω + 2Γ which is an approximation of Kubo’s formula [8]. In (1), Γ = 1/(2τ0 ), d is the graphene sheet’s thickness [5] and q2 k T µc −µc /kB T σ ∗ = eπh̄B2 , where τ0 is the kB T + 2 ln 1 + e transport relaxation time, qe is the electron’s charge, kB is the Boltzmann constant, T is temperature, h̄ is the reduced Plank’s constant and µc is the chemical potential [8]. As demonstrated in [11], [15], embedding a thin conductive sheet in FDTD 3D-lattice (in free space) produces an effective conductivity σ̃eff = (d/∆)σ̃, in which ∆ is the edge of a cubic Yee cell. By 2 IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, APRIL 2014 (REVIEWED ON SEPTEMBER 23, 2014) calculating the inverse Fourier transform of σ̃eff , one obtains the effective conductivity σ(t) = GC e−2Γt , t > 0, becomes (2) ∂Dy + Jy ∂t in which GC = σ /∆. In Maxwell’s equations (3) ~ ~ ~ ~ with D(t) = ǫE(t) and B(t) = µH(t), for current J~ evaluation one can use the conductivity (2) since the convolution ~ = J(t) Z t ~ − τ )σ(τ )dτ E(t (4) 0 is calculated for tangential current densities on graphene. For the problem at hand, Ampère’s law in (3) becomes ǫ ~ ∂E + ∂t Z t ~ − τ )σ(τ )dτ E(t 0 ~ = ∇ × H. (5) Taking into account the exponential behaviour of σ(t) in (2), it is possible to evaluate the convolution in (5) by a recursive procedure using the PLRC technique described in [14]. In order to simplify demonstration of the method, we shall treat the problem for a single spatial dimension and the time. We use Ey and Hz as field components for a wave propagating along the x direction. Taking into account that τ = m∆t and ~ and H ~ are calculated respectively at n and n+ 1 FDTD that E 2 ∂D discrete instants, we obtain for the term ∂ty + Jy in (3) and (5) the following expression: =ǫ n+ 21 ! ! Jyn+1 + Jyn Eyn+1 − Eyn + = ≈ǫ ∆t 2 ! Z Eyn+1 − Eyn 1 (n+1)∆t (n+1)−m + Ey σ(τ )dτ + ∆t 2 0 Z 1 n∆t n−m Ey σ(τ )dτ ), (6) + 2 0 ∂Dy + Jy ∂t where t = n∆t and the average current density 1 n+1 + Jyn is used in (6) to provide the time synchro2 Jy nization between Jy and Hz in (5). The function Ey can be considered constant during each individual FDTD time step. Therefore we can write the piecewise linear approximation Jyn ≈ n−1 X m=0 Jyn Eyn−m Z Jyn+1 , (m+1)∆t σ(τ )dτ. (7) m∆t Applying (7) to and extracting the first term (m = 0) from Jyn+1 and grouping the remaining summations, (6) + ≈ǫ Eyn+1 − Eyn ∆t ! + Z ∆t 1 + Eyn+1 σ(τ )dτ + 2 0 ! Z (m+2)∆t Z (m+1)∆t σ(τ )dτ + σ(τ )dτ . ∗ ~ ~ ∂D ~ ∂ B = −∇ × E ~ + J~ = ∇ × H, ∂t ∂t n+ 12 n−1 1 X n−m E 2 m=0 y (m+1)∆t m∆t (8) To simplify notation in (8), we define the functions Z (m+1)∆t σm (i) = σ(τ, i)dτ (9) m∆t and Sm (i) = σm (i) + σm+1 (i), (10) where x = i∆x. As a result, from Ampère’s law and (8)-(10), we see that 1 n n ~ + ǫ(i) ∇×H ∆t Ey (i) − 2 Ψy (i) y n+1 , (11) Ey (i) = ǫ(i) 1 ∆t + 2 σ0 (i) ~ and Ψny (i) is ~ is the y-component of ∇ × H where ∇ × H y the following convolution for the ith FDTD cell: Ψny (i) = n−1 X Eyn−m (i)Sm (i). (12) m=0 At this point, one should notice that the calculation of (11) is computationally prohibitive because of the present form of (12). Therefore, the aim here is to obtain a recursive FDTD equation for Ψny in (12). Substituting (2) in (9), one obtains Z (m+1)∆t σm (i) = GC e−2Γt dτ (13) m∆t and after integration GC −2Γm∆t (14) e − e−2Γ(m+1)∆t . 2Γ i From (14), one can notice that GC −2Γ(m+1)∆t e − e−2Γ(m+2)∆t = σm+1 (i) = 2Γ i = e−2Γ∆t σm (i). (15) σm (i) = Additionally, from (10) and (15), one has the recursive relation Sm+1 (i) = e−2Γ∆t Sm (i). (16) As a consequence, expressions (12) and (16) lead to the following simple recursive FDTD updating equation for (12): Ψny (i) = Eyn (i)S0 (i) + e−2Γ∆t Ψyn−1 (i), (17) with Ψ0y (i) = 0. Once (17) is calculated, Ey can be updated −2Γ∆t with (11). Observe GC /(2Γ) and that σ0 = 1 − e −4Γ∆t S0 = 1 − e GC /(2Γ) are obtained from (14) and (10). DE OLIVEIRA et al.: A FDTD FORMULATION FOR GRAPHENE SHEETS BASED ON PLRC TECHNIQUE 3 Fig. 3. Transient signals: (a) total currents flowing into electrode B, (b) gap voltages. Fig. 1. Graphene antenna [3] modelled for validation of the proposed method. Current Source Graphene sheet . Graphene z y x P.E.C. Electrode B Ey (i,j,k+1) Sheet . Current Source Ex P.E.C. Electrode A . (i,j,k) . Graphene sheet x Ex (i+1,j,k) Ey (a) (b) y Fig. 2. Graphene modelled in FDTD space: (a) a part of sheet in 3D space and (b) two graphene sheets, the P.E.C. electrodes A and B and current source. Ex and Ey are electric field components on graphene sheets. III. R ESULTS A. Validation of the proposed numerical formulation In order to validate the FDTD methodology developed in this work, two graphene-based antennas proposed in [3] have been modelled with our FDTD solver. Both antennas consist of two coplanar graphene sheets placed on a semi-infinite glass substrate with ǫr = 3.8 and σ = 0 as illustrated by Fig.1. The sheets are separated by a gap Sg = 3µm for both antennas. Antenna 1 is described by the parameters L = 17µm, W = 10µm and µc = 0.13eV and antenna 2 is characterized by L = 23µm, W = 20µm and µc = 0.25eV. Fig.2(a) shows a part of a graphene sheet in 3D FDTD ~ lattice. The E-field components Ex and Ey on the graphene sheets (shown in Figs.2 (a) and (b)) were calculated by using the proposed technique. These fields produce the current (4). Other field components were calculated by their corresponding original FDTD equations. In order to avoid numerical rounding problems when evaluating (11) and (17), σ0 , S0 , Gc and the coefficient e−2Γ∆t were calculated by using the open source Multiple Precision library GMP/MPFR from GNU. As long as the results in [3] were obtained by using the commercial frequency-domain FEM solver HFSS, we conceived the excitation structure of Fig.2(b) which consists of current source placed on the antenna’s plane between two perfect conducting (PEC) electrodes A and B. The metallic electrodes were included in our FDTD model exclusively because they are used in HFSS excitation port. The width of the electrodes is 500 nm. Soft current source [11] was implemented by forcing current densities Jx at the antenna gap to follow a monocycle pulse. The most part of energy of the pulse is contained in the spectral range 0.2-1.8 THz. The antennas’ impedances were ˜ calculated by using the Fourier P transforms Ṽ (f ) and I(f ) of the transient voltage V (t) = Ex ∆x produced between the electrodes and of the total transient current I(t) flowing into electrode B (Fig.2(b)). The obtained voltages V (t) and total currents I(t) for both antennas are shown by Fig.3. For modelling graphene antennas by the FDTD method, we have used cubic Yee cells with ∆ = 500nm (spatial step) and set a 96 × 90 × 71 grid. Antenna 1 was also simulated with ∆ = 250nm and ∆ = 125nm. The computational domain was truncated by using the CPML formulation [11] and numerical stability was observed when ∆t was set to 99% of Courant’s limit [11]. Fig.4 shows a comparison among impedance curves obtained with the PLRC-FDTD formulation proposed in this paper and the results of [3] for the same antennas. It can be observed that for ∆ = 500nm the results agree well in terms of resonance frequencies (where imaginary part of impedance is zero). Small deviations are observed for the peaks of real part of the impedance for antenna 1 (about 4%) and antenna 2 (approximately 1.5%). Fig.4 also shows impedance curves obtained for antenna 1 with higher discretization levels (∆ = 250nm and ∆ = 125nm). It is observed that 1) all resonance frequencies are still close to those given in [3] and very small shifts are observed, as expected, and 2) peaks of real part of impedance converges to the values given by [3]. This means that the proposed formulation can model the graphene thickness d independently of ∆, as described in [15], since ∆ > d/2 and ∆ ≤ λmin /10. Modifications in results are simply due to improved precision in numerical evaluation of derivatives in Maxwell’s equations and the corresponding reduction of the numerical dispersion [11]. Finally, Fig.5 shows the spatial current distribution for antenna 1 at the resonance frequency of 1.02 THz, obtained via discrete Fourier transforms. This result also fully agrees with the current diagram in [3]. B. Tuning resonance frequency In this section, we use our technique for analysis of one example of graphene antennas shown in Fig.1. It consists of gradual increasing of the width of the graphene sheets, such as illustrated by Fig.6(a). Thus, we deal with a bow-tie antenna. For the antenna in Fig.6, the width from antenna 1 is progressively incremented starting from the metallic electrodes with 10µm to 16µm at the end. The length of the original device (antenna 1) was preserved. This geometric modification 4 IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, APRIL 2014 (REVIEWED ON SEPTEMBER 23, 2014) D = 500 nm Antenna 1 – FDTD: (RE) - (D = 0.25 mm) (IM) - (D = 0.25 mm) (RE) - (D = 0.125 mm) (IM) - (D = 0.125 mm) 1.02 1.17 1.35 1.53 Resonances electric field components are present. This allows one to easily excite the antennas by voltage or current sources coplanarly placed to the sheets. Two rectangular antennas with graphene sheets placed on a dielectric substrate were modelled and full agreement with results published in literature was observed. A bow-tie geometry was analyzed and it was shown that in such antenna some tuning of the device’s resonances is possible. Finally, notice that the presented formulation can be used not only for analysis and design of graphene antennas and but also for modelling other photonic devices. R EFERENCES [1] Fig. 4. Validation of the proposed method: impedances of antennas 1 and 2. [2] [3] [4] [5] Fig. 5. Current distribution on graphene sheets for antenna 1, f = 1.02THz. [6] [7] [8] [9] Fig. 6. Bow-tie graphene antenna: (a) geometry and current distribution, f = 0.90THz, (b) impedance Z. [10] of the usual rectangular antenna can be used for controlling the resonance frequency. Fig.6(b) shows a comparison of impedance curves for antenna 1 and the bow-tie antenna. It can be seen that the resonances were displaced from 1.35 THz to 1.237 THz and from 1.02 THz to 0.90 THz. This effect can be understood by inspecting Figs. 5 and 6(a), where we see that the bow-tie geometry forces the most part of the current to flow over longer trajectories in comparison with the rectangular geometry. [11] [12] [13] [14] IV. F INAL R EMARKS In this paper, we presented a PLRC-FDTD formulation for including the effective conductivity of graphene sheets in FDTD lattices. With the formulation, it is possible to model graphene sheets at planes in Yee’s grid on which tangential [15] A. K. Geim and K. S. Novoselov, “The rise of graphene,” Nature Materials, vol. 6, no. 3, pp. 183–191, 2007. K. S. Novoselov, V. I. Fal’ko, L. Colombo, P. R. Gellert, M. G. Schwab, and K. Kim, “A roadmap for graphene,” Nature, vol. 490, no. 7419, pp. 192–200, 2012. M. Tamagnone, J. S. Gomez-Diaz, J. R. Mosig, and J. PerruisseauCarrier, “Analysis and design of terahertz antennas based on plasmonic resonant graphene sheets,” Journal of Applied Physics, vol. 112, no. 11, pp. 1–4, 2012. M. Tamagnone and J. Perruisseau-Carrier, “Predicting input impedance and efficiency of graphene reconfigurable dipoles using a simple circuit model,” Antennas and Wireless Propagation Letters, IEEE, vol. 13, pp. 313–316, 2014. I. Llatser, C. Kremers, A. Cabellos-Aparicio, J. M. Jornet, E. Alarco, and D. N. Chigrin, “Graphene-based nano-patch antenna for terahertz radiation,” Photonics and Nanostructures - Fundamentals and Applications, vol. 10, no. 4, pp. 353–358, 2012. Z. Zhu, S. Joshi, S. Grover1, and G. Moddel, “Graphene geometric diodes for terahertz rectennas,” Journal of Physics D: Applied Physics, vol. 46, no. 18, pp. 1–6, 2013. J. Christensen, A. Manjavacas, S. Thongrattanasiri, F. H. L. Koppens, and F. J. G. de Abajo, “Graphene plasmon waveguiding and hybridization in individual and paired nanoribbons,” ACS Nano, vol. 6, no. 1, pp. 431–440, 2012. G. W. Hanson, “Dyadic Green’s functions and guided surface waves for a surface conductivity model of graphene,” J. Appl. Phys., vol. 103, no. 064302, 2008. H. Lin, M. F. Pantoja, L. D. Angulo, J. Alvarez, R. G. Martin, and S. G. Garcia, “FDTD Modeling of Graphene Devices Using Complex Conjugate Dispersion Material Model,” Microwave and Wireless Components Letters, IEEE, vol. 22, no. 12, pp. 612–614, Dec 2012. V. Nayyeri, M. Soleimani, and O. Ramahi, “Modeling graphene in the finite-difference time-domain method using a surface boundary condition,” IEEE Transactions on Antennas and Propagation, vol. 61, no. 8, pp. 4176–4182, Aug 2013. A. Taflove and S. C. Hagness, Computational Electrodynamics, The Finite-Difference Time-Domain Method, 3rd ed. Artech House, 2005. S. H. Nasiri, M. K. M.-Farshi, and R. Faez, “Time Domain Analysis of Graphene Nanoribbon Interconnects Based on Transmission Line Model,” Iranian Journal of Electrical & Electronic Engineering, vol. 8, no. 1, pp. 37–44, 2012. A. Nishad and R. Sharma, “Analytical Time-Domain Models for Performance Optimization of Multilayer GNR Interconnects,” Selected Topics in Quantum Electronics, IEEE Journal of, vol. 20, no. 1, pp. 17–24, Jan 2014. D. Kelley and R. Luebbers, “Piecewise linear recursive convolution for dispersive media using FDTD,” IEEE Transactions on Antennas and Propagation, vol. 44, no. 6, pp. 792–797, Jun 1996. J. Maloney and G. Smith, “The efficient modeling of thin material sheets in the finite-difference time-domain (FDTD) method,” IEEE Transactions on Antennas and Propagation, vol. 40, no. 3, pp. 323– 330, Mar 1992.
© Copyright 2025 Paperzz