Sie sind auf Seite 1von 6


6, NOVEMBER 2013 1031

A Hybrid Plasmonic Waveguide for the Propagation

of Surface Plasmon Polariton at 1.55 μm
on SOI Substrate
As’ad Amirhosseini and Reza Safian, Member, IEEE

Abstract—A hybrid plasmonic waveguide (HPW) has been nu- ever, the tradeoff is that it also increases the mode size [5], [6].
merically investigated, based on silicon-on-insulator (SOI) struc- Therefore, to solve this problem to some extent, a hybrid waveg-
ture near the telecom wavelength λ = 1550 nm using the finite- uiding mechanism was proposed recently [7], [8]. In these struc-
element method. The HPW consists of metal cap placed on SOI
rib, which has a trench with a semicircle interface at the bottom of tures, the mechanism consists of a high-index-contrast dielectric
the trench. With the proper choice of waveguide sizes, our proposed waveguide placed close to a plasmonic structure. In this regard,
HPW can support hybrid long-range surface plasmon which have one of the important structures that attracted a lot of attention is
strong mode confinement Ae ff /A0 = ∼ 0.0035–0.035 and long prop-
the hybrid waveguide consisting of a high-index medium sepa-
agation length (Le ff > 525 μm) simultaneously. In addition, for rated from a metal surface by a low-index spacer [7]–[10]. It has
the proposed HPW, the calculated coupling length for two parallel
identical waveguides which is separated by distance d = 250 nm is been shown that this waveguide can have a good confinement
approximately 1.3 μm, which is very helpful in realizing photonic and propagation distance at the same time [7]. Other kinds of
integration circuits with ultrahigh integration density. hybrid plasmonic waveguide for TM polarization can be created
Index Terms—Coupling length, figure of merit (FOM), surface by placing a metal cap [8] or metal plate [11] on top of a silicon-
plasmon polariton (SPP), silicon-on-insulator (SOI) substrate. on-insulator (SOI) rib waveguide with SiO2 on the surface, or
by incorporating air nanotrench at the interface of metal and
silicon layer [12].
I. INTRODUCTION In this paper, we present a novel hybrid plasmonic waveguide,
N the recent years, nanoscale optical waveguides based on which consists of metal cap placed on the SOI rib, with a hole in
I surface plasmon polariton have become a promising solu-
tion to achieve photonic integrated circuits (PICs) with a high-
silicon layer under the metal, which has a semicircle interface at
its bottom. With this idea, we basically incorporate both ideas
integration density. Due to their ability to break the diffraction of using circular geometry [13] to have very tight confinement
limit, they have the ability of true nanoscale waveguiding and and long propagation length and SOI rib platform to ensure
confinement of light. Till now, a variety of plasmonic structures compatibility to CMOS technology, simultaneously. By intro-
have been investigated, including long-range surface plasmon ducing this new low-index region, the present hybrid plasmonic
polariton (LRSPP) [1], [2], metal–insulator–metal (MIM) [3], waveguide provides a very high power density in the region of
dielectric-loaded SPP (DLSPP) [4]. However, due to the large low-index hole, due to the field-enhancement resulting from the
ohmic loss of metals, all these waveguides suffer from a tradeoff effects of electric field discontinuity and surface plasmonic. The
between the confinement ability and the propagation length. In simulation results show a very small effective mode area in the
general, the SPP confinement is achieved primarily by decreas- proposed structure. It will be shown that this novel structure also
ing the SPP spatial extension into the dielectric. On the other has very small decoupled separation for two parallel identical
hand, this increases the portion of SPP power being absorbed waveguides, which is helpful in realizing high-density PICs.
by the metal, which increases the loss. Hence, the choice of
optimum guiding configuration is subject to tradeoff with many II. WAVEGUIDE STRUCTURE
intricate issues. Fig. 1 shows the cross section of our proposed hybrid plas-
It has been shown that covering the metal strip with two monic waveguide. The main feature of this geometry is the
thin layers of lower index material can be a way to reduce the semicircle silicon interface at the bottom of the trench. The
loss and increasing the propagation length significantly. How- wavelength is set at λ = 1550 nm, and permittivities of SiO2 ,
Si, and Ag are Sio 2 = 2.25, and Si = 12.25, and Ag = −129 +
3.3i [14], respectively. The minimum distance between top of
Manuscript received August 14, 2012; revised December 23, 2012; accepted the semicircle and the metallic layer is h. This helps to confine
January 5, 2013. Date of publication May 17, 2013; date of current version electromagnetic field in a very small area, which leads to a very
November 6, 2013. The review of this paper was arranged by Associate Editor small effective mode area. By proper choice of the waveguide
A. Ghosh.
The authors are with the Department of Electrical and Computer Engineer- dimensions, it also provides a large propagation length at the
ing, Isfahan University of Technology, Isfahan, Iran (e-mail: sa.amirhoseyni@ same time. In our simulations through this paper, we use the; waveguide parameters such as diameter of the semicircle (D),
Color versions of one or more of the figures in this paper are available online
at silicon height (H), distance between the tip of the semicircle
Digital Object Identifier 10.1109/TNANO.2013.2263987 part of the trench to the conductor (h), and the width of the
1536-125X © 2013 IEEE

Fig. 1. Cross section of our proposed hybrid plasmonic waveguide with a

nanohole at the interface of metal and silicon layer.

trench walls (W ) to optimize the characteristic properties of Fig. 2. Real part of the effective mode index for different h and W parameters
our proposed waveguide. as D varies. (a) h = 5 nm. (b) h = 20 nm.
The main characteristic parameters of plasmonic waveguides
are effective mode index, Neff , propagation length Leff , and
effective mode area Aeff . The modal properties are investi-
gated by means of the finite-element method (FEM) using
COMSOL software. The eigenmode solver is used with the
scattering boundary condition. Convergence tests are done to
ensure that the numerical boundaries and meshing do not in-
terfere with the solutions. In addition, to make sure there is no
disturbance from the boundaries of the computational domain,
we choose the calculation domain to be large enough, (−2 μm <
y < 2 μm and −2 μm < x < 2 μm). Throughout this paper, we
set H = D/2, H1 = D/2 + h, and H2 = 100 nm and we use
other parameters of the waveguide in the optimization process.
To analyze the structure, we have to define the partial differ-
ential equation derived from Maxwell’s equations that describes
the light propagation in this waveguide which is
∇n2 (x, y) × Et 2π
∇2 Et + ∇ + n2 (x, y) Et = β 2 Et
n2 (x, y) λ
(1) Fig. 3. Propagation length for different values of W, as D varies when h =
5, 20 nm.
β = β − jβ (2)
where Et is the transverse electric field, λ is the wavelength, Another characteristic of the plasmonic waveguide, which
n(x,y) is the refractive index distribution, and β is the complex plays an important role in the design process, is propagation
propagation constant (eigenvalue of the wave equation). Instead length Leff , which is defined as the distance that the energy
of using β itself usually the parameter Neff which is defined (intensity) of the field attenuates to 1/e, i.e.,
as β/k0 [k0 is the wavenumber in vacuum (k0 = 2π/λ)] is used.
Leff = λ/ (4π {Neff }) . (3)
In Fig. 2, we have plotted the real part of effective mode index
{Neff } versus circle diameter D, for different h parameter In Fig. 3, we have plotted propagation length versus D as W
values as D and W vary. varies for h = 5 nm and h = 20 nm, (logarithmic scale on the
Normally, it is desirable to design the waveguide with high y-axis). According to Fig. 3, longer propagation length in the
real part of the effective mode index, which leads to higher range of millimeter can be achieved by increasing the diameter
concentration of the light. It is clear that the increasing the trench D and distance h and decreasing the value of W in the substrate.
aperture increases {Neff }. Also with thicker trench walls (W ) However, longer propagation length usually causes a problem
and smaller distance between the semicircle dielectric and the in plasmonic waveguides, which is reduction in the effective
conductor (h), we can have larger real part of the effective mode mode index of the waveguide. To see that, we have plotted the
index. normalized effective mode area (Aeff /A0 ), for h = 5 nm and

Fig. 5. Propagation length for h = 5 nm and h = 20 nm as H varies for fixed

values of D = 200 nm and W = 10 nm.

Fig. 4. Normalized effective mode area (A e ff /A 0 ) for h = 5 nm and 20 nm,

based on two definitions (a) h = 5 nm and (b) h = 20 nm.

20 nm as W and D vary, in Fig. 4(a) and (b). In each plot, we

have calculated the effective mode area based on two different
definitions: one according to statistical measurement definition
and one according to energy density definition. Based on the
statistical definition of effective mode area,
 2 2
E (r) dA
Aeff =  4 (4)
E (r) dA
where E(r) is the electric field [15], and in the energy density

W (r) dA
Aeff = (5)
MAX {W (r)}
where in this definition, W (r) is the energy density [15]. To
compare these definitions, we have normalized both of them to Fig. 6. Normalized effective mode area A e ff versus H for h = 5 nm and h =
20 nm, based on statistical and energy density definitions.
A0 , the diffraction-limited mode area defined as λ2 /4.
It is clear that as we increase the value of D, confinement be-
comes worse, which is not desirable for the purposes of compact ment simultaneously. In Fig. 5, a plot of propagation length for
integration circuits. A good design must consider both Aeff and h = 5 nm and 20 nm versus silicon height H is shown for D =
Leff at the same time, which will be a tradeoff between them. 200 nm and W = 10 nm. From Fig. 5, it is obvious that for larger
For h = 5 nm based on the statistical definition, we can see that H, longer propagation length can be achieved (logarithmic scale
for all values of D between 100 and 400 nm or even higher val- on the y-axis).
ues, we have a very good confinement (less than 0.035) which To see the effect of the rib heights H on the effective
ensures a high localization of field in area between top of sil- mode area, we have plotted the normalized effective mode area
icon semicircle and metallic layer. Also based on the energy Aeff /A0 versus H for different values of h using statistical and
density definition, we obtained Aeff /A0 below 0.08, which is energy density definition in Fig. 6. Based on the results, effec-
also very good, compared to other plasmonic waveguide struc- tive mode area grows monotonically with H; therefore, better
tures. In Fig. 4(b), by increasing D, the effective mode area confinement is achieved with lower values of Si rib height. It
increases with a fast rate up to the point D = 225 nm, and for should be mentioned that although we have used both the sta-
D > 225 nm, it increases with slower rate. The reason is that tistical and energy density definitions to calculate the effective
around D = 225 nm, the structure behavior changes from hybrid mode area, the energy density definition is not proper for our
structure to cylindrical mode waveguide mode [7]. proposed structure because of two main reasons. First, since we
Another parameter that can be used in the optimization pro- have a hole in the Si layer, the peak of energy density distri-
cess is the silicon rib height H. By proper choice of H, we bution on side walls becomes so large, which leads to inaccu-
can achieve large propagation length with good mode confine- rate calculated results [15]. Second, it has been shown that the

Fig. 7. Quasi-TM fundamental mode |E y | of the proposed hybrid plasmonic Fig. 9. |E y | distribution in the y-direction along the dashed lines in Fig. 7.
waveguide with a semicircle hole at the interface of metal and Si (H = 350 nm,
H 2 = 100 nm, W = 10 nm, D = 200 nm, and h = 5 nm).

Fig. 10. Cross section of our proposed directional coupler for a coupling
system with two parallel identical hybrid plasmonic waveguides. The separation
distance is d.

Fig. 8. |E y | distribution in the x-direction along the dashed lines in Fig. 7.

Fig. 11 shows the calculated coupling length Lc of two paral-
lel hybrid plasmonic waveguides as the distance between them,
hybrid plasmonic waveguide is good for high energy density d, varies from 250 to 700 nm for given separations h = 5 nm
rather than real energy confinement [16]. Hence, the best choice and h = 20 nm (logarithmic scale on the y-axis). The other
for analyzing Aeff and designing an optimum structure is using parameters are D = 200 nm, H2 = 100 nm, W = 10 nm, and
the statistical definition. For the final design, according to the H = 300 nm. The coupling length is given by
presented results and explanations and based on the statistical
Lc = π/(βo − βe ) (6)
definition, the optimum structure should have the parameter val-
ues H = 350 nm, H2 = 100 nm, W = 10 nm, D = 200 nm, where βo and βe are the propagation constants of the odd and
and h = 5 nm. The quasi-TM fundamental mode |Ey | for these even super modes of the two parallel waveguides, respectively.
sizes is shown in Fig. 7. We also have plotted the mode profile They can be calculated using a full-vector FEM mode-solver.
along x- and y-axis in Figs. 8 and 9, respectively. In Fig. 8, the According to Fig. 11, the coupling length almost increases
lateral confinement is almost 100 nm, which is more proper than exponentially as the separation d increases, which is similar
500 nm that was proposed in [13]. to the conventional dielectric optical waveguides. When the
separation is decreased to 250 nm, the coupling length is as small
as Lc = 1.30 μm. On the other hand, the coupling length is only
III. DIRECTIONAL COUPLER BASED ON THE PROPOSED several micrometers, when a small separation is chosen. This is
very useful for an ultracompact directional coupler, which is a
In Fig. 10, the cross section of our proposed directional cou- basic element for nanophotonic integration circuits.
pler for a coupling system with two parallel identical hybrid In order to highlight the limits of the proposed device in
plasmonic waveguides is shown. The distance between two terms of fabrication process and fabrication tolerances, we have
waveguides is d. simulated our structure based on situations when an accurate

COMPARISON OF OUR FINAL RESULT (D = 200 nm, H = 350 nm, W =
200 nm, H = 350 nm, AND W = 10 nm) AND [12]
(W = 220 nm AND W a = 200 nm)

Fig. 11. Coupling length L c of two parallel hybrid plasmonic waveguides for
h = 5 nm and h = 20 nm (D = 200 nm, H 2 = 100 nm, W = 10 nm, and H =
300 nm).
COMPARISON OF OUR FINAL RESULT (D = 200 nm, H = 350 nm, W =
200 nm, H = 350 nm, AND W = 10 nm) AND [12]
(W = 220 nm AND W a = 200 nm)

Fig. 12. Proposed hybrid plasmonic waveguide without an accurate semicircle

on silicon rib (H = 300 nm, W = 10 nm, D = 200 nm, h = 5 nm, and W =
10 nm). In Tables II and III, we have compared our final results for
TABLE I h = 5 nm and h = 20 nm, with two other structures in [7]
COMPARISON OF OUR FINAL RESULT FOR ACCURATE AND INACCURATE with parameters (D = 200 nm, H = 350 nm, W = 10 nm)
FABRICATION PROCESS (H = 300 nm) and [12] with parameters (W = 220 nm and Wa = 200 nm). As
mentioned before, statistical definition is more suitable for our
structure. For h = 20 nm, it provides a large effective length and
the lowest effective mode area compared to other structures.

We have studied a Si-based hybrid plasmonic waveguide with
semicircle hole at the interface of metal and silicon layer for
nanoscale light confinement as well as relatively long propaga-
tion distance (in the range of millimeter). The present hybrid
plasmonic waveguide has been shown, that has a relatively long
propagation length and nanoscale optical confinement at the
and fine semicircle cannot be fabricated. For this condition, in same time, by adjusting the waveguide parameters. Due to the
Fig. 12, we have assumed that during the fabrication process, field enhancement in the low-index region (which is air in our
the semicircle edges on the rib part associate with unexpected design) for TM polarization, an ultrasmall normalized effec-
tolerance. Table I will show that a very fine and accurate design tive mode area was achieved (e.g., 0.0035 at λ = 1550 nm). In
does not require getting a good result: (D = 200 nm, H = addition, when the distance between metallic layer and semi-
300 nm, W = 10 nm and h = 5 nm). circle (h) decreases, the field is more confined in the low-index
Table I reveals that even with inaccurate fabrication process region. The coupling length between two adjacent waveguides
(inaccurate semicircle edge), the results almost does not change. has also been calculated. The simulation results show very small

coupling length for small separation distance (d), (e.g., Lc = [14] P. B. Johnson and R. W. Christy, “Optical constants of the noble confine-
1.30 μm) which is very helpful to realize photonic integration ment and long-range propagation,” Nat. Photon. Metals, Phys. Rev. B,
vol. 6, pp. 4370–4379, 1972.
circuits with ultrahigh integration density. [15] R. F. Oulton, G. Bartal, D. F. Pile, and X. Zhang, “Confinement and prop-
agation characteristics of subwavelength plasmonic modes,” N. J. Phys.,
REFERENCES vol. 10, no. 10, p. 105018, 2008.
[16] Z. Ruixi, Z. Yuan, and H. Sailing, “Energy intensity analysis of modes
[1] R. Charbonneau, N. Lahoud, G. Mattiussi, and P. Berini, “Demonstra- in hybrid plasmonic waveguide,” Frontiers Optoelectronics China, vol. 5,
tion of integrated optics elements based on long-ranging surface plasmon pp. 68–72, 2012.
polaritons,” Opt. Exp., vol. 13, no. 3, pp. 977–984, 2005.
[2] P. Berini, “Plasmon-polariton modes guided by a metal film of finite width
bounded by different dielectrics,” Opt. Exp., vol. 7, no. 10, pp. 329–335,
[3] J. Dionne, L. Sweatlock, H. Atwater, and A. Polman, “Plasmon slot
waveguides: Towards chip-scale propagation with subwavelength-scale
localization,” Phys. Rev. B, vol. 73, no. 3, p. 035407, 2006. As’ad Amirhosseini was born in Gachsaran, Iran,
[4] T. Holmgaard and S. Bozhevolnyi, “Theoretical analysis of dielectric- in 1987. He received the B.Sc. degree from the Per-
loaded surface plasmon-polariton waveguides,” Phys. Rev. B, vol. 75, sian Gulf University, Boushehr, Iran, in 2010 and the
no. 24, p. 245405, 2007. M.Sc. degree from the Isfahan University of Tech-
[5] J. P. Guo and R. Adato, “Extended long range plasmon waves in finite nology, Isfahan, Iran, in 2012, both in electrical and
thickness metal film and layered dielectric materials,” Opt. Exp., vol. 14, computer engineering. He has been working toward
no. 25, pp. 12409–12418, 2006. the Ph.D. degree at Isfahan University of Technology,
[6] J. P. Guo and R. Adato, “Control of 2D plasmon-polariton mode with Isfahan, Iran, since 2012.
dielectric nanolayers,” Opt. Exp., vol. 16, no. 2, pp. 1232–1237, 2008. His research activities are focused on surface plas-
[7] R. F. Oulton, V. J. Sorger, D. A. Genov, D. F. P. Pile, and X. Zhang, “A mon waves, nano antennas, and plasmonic wave-
hybrid plasmonic waveguide for subwavelength confinement and long- guide.
range propagation,” Nat. Photon, vol. 2, no. 8, pp. 496–500, 2008.
[8] D. Dai and S. He, “A silicon-based hybrid plasmonic waveguide with a
metal cap for a nano-scale light confinement,” Opt. Exp., vol. 17, no. 19,
pp. 16646–16653, 2009.
[9] M. Fujii, J. Leuthold, and W. Freude, “Dispersion relation and loss of sub-
wavelength confined mode of metal-dielectric-gap optical waveguides,”
IEEE Photon. Technol. Lett., vol. 21, no. 6, pp. 362–364, Mar. 2009.
[10] M. Z. Alam, J. Meier, J. S. Aitchison, and M. Mojahedi. (2007). “Super Reza Safian (S’04–M’09) received the B.Sc. degree
mode propagation in low index medium,” in Proc CLEO/QELS Conf. [On- in electrical engineering from Isfahan University of
line]. Available: Technology, Isfahan, Iran, in 1999, the M.A.Sc. de-
2007-JThD112 gree in electrical engineering from McMaster Univer-
[11] Z. Wang, D. Dai, Y. Shi, G. Somesfalean, P. Holmstrom, L. Thylen, S. He, sity, Hamilton, ON, Canada, in 2003, and the Ph.D.
and L. Wosinski, “Experimental realization of a low-loss nano-scale Si degree in electrical engineering from the University
hybrid plasmonic waveguide,” presented at the Opt. Fiber Commun. Conf., of Toronto, Toronto, ON, Canada, in 2008.
Los Angeles, CA, USA, 2011, Paper JThA017. From 1999 to 2002, he was an RF Design Engineer
[12] Y. Su, Z. Zheng, Y. Bian, Y. Liu, J. Liu, J. Zhu, and T. Zhou, “Low-loss with the Electrical and Computer Engineering Re-
silicon-based hybrid plasmonic waveguide with an air nano trench for search Center (ECERC), Isfahan University of Tech-
sub-wavelength mode confinement,” Micro Nano Lett., IET, vol. 6, no. 8, nology. In 2008, he joined the faculty of the Isfahan,
pp. 643–645, Aug. 2011. where he is currently an Assistant Professor with the Electrical and Computer
[13] I. Avrutsky, R. Soref, and W. Buchwald, “Sub-wavelength plasmonic Engineering Department. His technical interests include basic electromagnetic
modes in a conductor-gap-dielectric system with a nanoscale gap,” Opt. theory, microwave circuits, wave propagation in dispersive materials, and mi-
Exp., vol. 18, pp. 348–363, 2010. crowave imaging.