Professional Documents
Culture Documents
Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 10:54:13 UTC from IEEE Xplore. Restrictions apply.
4 (c t ) 2 n 1/ 2 4c t additionally explicit calculations of (6) for the magnetized
J zn 1 Jz J xn 1/ 2 (6b) graphene. Because we use the CN scheme for (2) and Jx and Jz
4 (c t ) 2
4 (c t ) 2
are synchronized with Ex and Ez in time, the stability condition
On the other hand, we also decompose (1) into two sub-time of the proposed method is the same as that for electrostatic
steps i.e., for n∆t –(n+1/2)∆t field biased graphene and consequently the same as that for
1 E 1 H the conventional HIE-FDTD [7].
H J , E (7a, b)
2 t 2 t
and for n∆t (n+1/2)∆t–(n+1)∆t III. NUMERICAL RESULTS AND DISCUSSIONS
To check the stability of the proposed method, a perfectly
1 E 1 H
0, 0 (8a, b) electric conductor (PEC) cavity loaded with a graphene sheet
2 t 2 t
is studied as depicted in Fig. 1. The domain size has 50×50×50
Similar as the method provided in [4] [5], by applying the HIE cells. In the simulations, we set T=300 K, ν=0.2 THz, μc=0.1
scheme to (7), we have eV, B0=5 T, fmax=1 THz, ∆x=∆z=10∆y=c0/(20fmax), c0 being the
t H n H zn speed of light in vacuum. The maximum time step of the
E yn 1/ 2 E yn ( x ) (7a)
z x conventional HIE-FDTD is ∆tmax=1/[c0(1/∆x2+1/∆z2)1/2]. The
t Ezn Exn graphene patch is placed at the center of the domain in the y-
H yn 1/ 2 H yn ( ) (7b) direction. A line current source Ez(t)=exp[-4π(t-t0)2/τ2] along
x z
n 1/ 2
the z-direction and one cell away from the cavity center is
t ( H zn 1/ 2 H zn ) H y J xn 1/ 2 J xn used to excited the field, where t0=τ=50∆tmax. The simulation
Exn 1/ 2 Exn ( ) (7c)
2y z 2 runs up to 104∆tmax. The observation point is set one cell away
n 1/ 2 from the center of the right surface.
t H y ( H xn 1/ 2 H xn ) J zn 1/ 2 J zn
Ezn 1/ 2 Ezn ( ) (7d)
x 2y 2
Graphene Line source
n 1/ 2
t E y ( Ezn 1/ 2 Ezn ) z
H xn 1/ 2 H xn ( ) (7e) 1cell
z 2y o
t ( Exn 1/ 2 Exn ) E y
n 1/ 2 y
H zn 1/ 2 H zn ( ) (7f) x
2y x B0, E0
From (8), we have
Obs.
E n 1 E n 1/ 2 , H n 1 H n 1/ 2 (8a, b)
n+1/2
Further, to compute Ex , we substitute the x-component of 50 cells
(4) and (7f) into (7c) to obtain
(a)
t 2 2 ta2 n 1/ 2 t 2 2 ta2 n
(1 ) Ex (1 ) Ex
4 y 2
2 4 y 2 2
n 1/ 2 2 n 1/ 2 (9)
t H n H y t 2 E y t (a1 1) n
( z ) Jx
y z 2 yx 2 6 6.15
Proposed t=tmax
where a1=(2-ν∆t)/(2+ν∆t) and a2=σ0ν∆t/ (2+ν∆t)h. To compute 6.10
4 Ref. [8] t=0.7tmax
En+1/2
z , we substitute the z-component of (4) and (7e) into (7d): Ref. [8] t=0.8tmax
Recorded Ez (10-7 V/m)
and Hn+1/2
z by (7e) and (7f); explicitly update Jn+1/2 x and Jn+1/2
z by Fig. 1. Recorded Ez computed by the proposed HIE-FDTD and the method
(4); explicitly update Jn+1 x and Jn+1z by (6). Compared with the provided in [8] with different values of ∆t.
method for isotropic graphene sheet [6] [7], there are only two
Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 10:54:13 UTC from IEEE Xplore. Restrictions apply.
Fig. 1 shows the recorded Ez computed by the proposed TABLE I
STABILITY OF THE PROPOSED METHOD (OUTSIDE BRACKET) AND THAT OF [8]
method with ∆t=∆tmax. The results computed by the method (INSIDE BRACKETS) FOR VARIOUS μc (eV) AND B0 (T).
provided in [8] with ∆t=0.7∆tmax and 0.8∆tmax are also given ‘Y’: STABLE. ‘N’: UNSTABLE.
for comparison. All the results show good agreement at early μc
0.01 0.02 0.04 0.06 0.08 0.1 0.2 0.4 0.6 0.8 1.0
time. However, the recorded Ez computed by the proposed B0
0 Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y)
method remains stable at later times while the result computed 0.5 Y(N) Y(Y) Y(N) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y)
1 Y(N) Y(N) Y(Y) Y(N) Y(N) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y)
by the method provided in [8] with ∆t=0.8∆tmax is unstable. 2 Y(N) Y(N) Y(N) Y(N) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y) Y(Y)
Therefore, for the parameters selected, the present method 4
6
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(N)
Y(Y)
Y(N)
Y(Y)
Y(N)
Y(Y)
Y(Y)
Y(Y)
Y(Y)
Y(Y)
Y(Y)
retains the same stability condition as the conventional HIE- 8 Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(Y) Y(Y)
10 Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(Y) Y(Y)
FDTD while the method provided in [8] must use a reduced 20 Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N) Y(N)
time step.
6 6.15
4 6.10
Proposed t=tmax
Ref. [8] t=0.7tmax PML
Ref. [8] t=0.8tmax
Recorded Ez (10-7 V/m)
2
6.05
TF/SF
Obs.
180 cells
6.00
2.05 2.10 2.15
0 B0
y Graphene E0
-2
z x Obs. PBC
-4
PML
-6
10 cells
0 5 10 15 20 25 30 35 40
Time (ps) (a)
Fig. 1. Recorded Ez computed by the proposed HIE-FDTD and the method 80
provided in [8] with different values of ∆t.
60
To further check the stability condition of the proposed 40
Farady Rotation angle (Deg)
by the method provided in [8] with ∆t=∆tmax are also given in -80
e. Anal. B0=4 T
e. Anal. B0=8 T
the brackets. It is seen that the proposed method is always Imag. Anal. B0=1 T
-100 Imag. Anal. B0=4 T
stable while the method provided in [8] is not. Moreover, the Imag. Anal. B0=8 T
-120
method in [8] becomes unstable with decreasing μc or 109 1010 1011 1012 1013
increasing B0. This is because the conductivity of the graphene Frequency (Hz)
decreases with decreasing μc and the coupling effect becomes (b)
stronger with increasing B0. We conjecture that one of the
reasons for the more restrictive stability condition for the Fig. 2. Real and imaginary parts of Faraday rotation angle computed by the
method in [8] might be that the lack of synchronicity of J and proposed method with different values of B0.
E in time and a strong coupling would exacerbate this issue.
We have run the proposed method for a wide range of To check the accuracy of the proposed method, the Faraday
parameter choices and found it always stable with ∆t=∆tmax. and Kerr rotations of graphene sheets are studied. In our
Therefore, we may draw a conclusion that the stability simulations, we set fmax=10 THz, μc=0.4 eV, other parameters
condition is verified numerically. We note a rigorous are set the same as those in Fig. 1. The total computational
analytical stability analysis [9] of the proposed method is very domain is 10×180×10 cells including 15-layer perfectly
challenging to the complexity of the model equations. matched layers (PMLs) [10] in the both end of y-direction as
depicted in Fig. 2. Total field/scatted field (TF/SF) is used to
introduce an x-polarized incident pulse. A periodic boundary
Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 10:54:13 UTC from IEEE Xplore. Restrictions apply.
condition (PBC) is used to model an infinitely wide graphene Tianjin University of Technology and Education under Grant
sheet. It is seen that the computed results agrees well with the of KJ1739.
analytical ones [1]. The imaginary part is close to zero in the
microwave band and increases in the THz band. This means REFERENCES
the transmitted wave changes from linear to elliptical [1] D. L. Sounas and C. Caloz, “Gyrotropy and nonreciprocity of graphene
polarized with increasing frequency. The same result can also for microwave applications,” IEEE Trans. Microw. Theory Tech., vol.
60, no. 4, pp. 901-914, Apr. 2012.
be obtained for the Kerr rotation as shown in Fig. 3.
[2] X. H. Wang, W. Y. Yin, and Z. Chen, “Matrix exponential FDTD
modeling of magnetized graphene sheet,” IEEE Antennas Wireless
40
Kerr Rotation angle (Deg)
e. Pro. B0=1 T, B0=4T, B0=8 T Propag. Lett., vol.12, pp. 1129-1132, Sep. 2013.
20 Imag. Pro. B0=1 T, B0=4T, B0=8 T [3] A. Taflove and S. C. Hagness, Computational Electrodynamics: The
0 Finite-Difference Time-Domain Method, 3nd Ed., Norwood, MA:
Artech House, 2005.
-20
[4] B. K. Huang, G. Wang, Y. S. Jiang, and W. B. Wang, “A hybrid implicit
-40 explicit FDTD scheme with weakly conditional stability,” Microw. Opt.
Technol. Lett., vol. 39, pp. 97-101, Oct. 2003.
-60
[5] J. Chen and J. Wang, “A 3D hybrid implicit explicit FDTD scheme with
109 1010 1011 1012 1013 weakly conditional stability,” Microw. Opt. Technol. Lett., vol. 48, pp.
Frequency (Hz) 2291-2294, Nov. 2006.
[6] J. Chen, J. Li, and Q. Liu, “Designing graphene-based absorber by using
Fig. 3. Real and imaginary parts of Kerr rotation angle computed by the HIE-FDTD method,” IEEE Trans. Antennas Propag, vol. 65, no.4,
proposed method with different values of B0. pp.1896-1902, Apr. 2017.
[7] M. L. Zhai, H. L. Peng, X. H. Wang, X. Wang, Z. Chen, and W. Y. Yin,
CONCLUDING REMARKS “The conformal HIE-FDTD method for simulating tunable graphene-
based couplers for THz applications,” IEEE Trans. THz Sci. Technol.,
A stability-improved HIE-FDTD for magnetostatic field vol. 5, no. 3, pp. 368-376, May 2015.
biased graphene has been introduced. Numerical results have [8] C. Tian and J. Chen, “Analysis of magnetically biased graphene
shown that it can be used to accurately simulate graphene- absorber using anisotropic HIE-FDTD method,” Journal of
Electromagnetic Waves and Applications, vol. 33, no.6, pp.722-733,
based structures with retaining the same stability condition as 2019.
the conventional HIE-FDTD. [9] S. Wang and F. L. Teixeira, ‘Some remarks on the stability of time-
domain electromagnetic simulations,’ IEEE Trans. Antennas
Propagation, vol. 52, no. 3, pp. 895-898, 2004.
ACKNOWLEDGEMENT [10] Z. Liu, Y. Chen, X. Sun, and Y. Liu, “Implementation of CFS-PML for
HIE-FDTD method,” IEEE Antennas Wireless Propag. Lett., vol.11, pp.
This work was supported by the the Science Foundation of 381-384, 2012
Authorized licensed use limited to: Carleton University. Downloaded on May 27,2021 at 10:54:13 UTC from IEEE Xplore. Restrictions apply.