ARTICLE IN PRESS

Physica B ] (]]]]) ]]]–]]]

Contents lists available at ScienceDirect

Physica B
journal homepage: www.elsevier.com/locate/physb

Integration of plasmonics into nanoelectronic circuits
Ping Bai Ã, Hong Son Chu, Mingxia Gu, Oka Kurniawan, Erping Li
Computational Electronics and Photonics, Institute of High Performance Computing, 1 Fusionopolis Way #16-16 Connexis, Singapore 138632, Singapore

a r t i c l e in fo

abstract
Plasmonic waveguides are proposed to transmit information optically from one electronic component to another in integrated nanoelectronic circuits. A metal–insulator–metal (MIM) plasmonic waveguide, rather than a traditional dielectric waveguide or an electric wire, is used for data transmission. The MIM plasmonic waveguide confines plasmonic (electromagnetic) waves to a dielectric slot with a crosssection of 50 nm  50 nm, and propagates them close to the speed of light. Two nanorods that form a nanoantenna are employed to receive the optical power from the waveguide, and localize as well as concentrate the received power in the proximity around the two nanorods. The localized optical power is converted to electric signals by a nanoscale plasmonics-to-electronics converter, which has a nanosized active volume and a bandwidth of up to 1 THz. Both the plasmonic waveguide and the converter have nanoscale dimensions comparable to those of modern nanoelectronic counterparts. The proposed method could be used to improve the performance of existing nanoelectronic systems by exploiting the strength of optics. & 2010 Elsevier B.V. All rights reserved.

Keywords: Plasmonics Subwavelength waveguide Plasmonic converter Nanoantenna

1. Introduction Integration of optical and electronic circuits leads to remarkable benefit in data transmission and processing by combining the advantages of the large bandwidth of optics and the compactness of electronics [1]. Continuous reduction in size of electronic components has resulted in better, faster, and smaller electronic everyday products for society. However, interconnects on a chip have become a bottle-neck preventing further improvement of the performance of electronic products due to the increasing signal delay and power consumption associated with the interconnects [2]. Optical interconnects are superior to the electronic interconnects by virtue of their large operational bandwidth, and huge data transmission capability. However, the dimensions of a traditional optical device are fundamentally limited by the law of diffraction. The large mismatch in size prevents integrating optics with electronics for a better performance. Plasmonics allows manipulating the flow of light in a nanometer scale well below the diffraction limit, by exploiting the unique optical properties of metallic structures [3]. Much attention has been attracted to the study of plasmonics, from the fundamentals to applications in recent years [4–8], including nanoscale optical waveguides [4], perfect lenses, subwavelength lithography, highly sensitive biosensors and ultra-fast modulators [5]. Plasmonics could provide a new way for optical data transmission near light speed requiring a size well below the
à Corresponding author. Tel.: + 65 64191242; fax: + 65 64635176.

free-space wavelength, with efficient power delivery. Plasmonics will be the potential technology to achieve information transmission between nanoscale electronic devices at optical frequencies, and bridge the gap between photonics and electronics in the nanoscale regime [9]. We explore the convergence of optics and electronics by using plasmonic waveguides for on-chip nanoscale optical interconnects. To optically transmit data between electronic devices, we need to develop a number of plasmonic devices, including plasmonic sources, modulators, waveguides, switches, filters and detectors. As a first step, we focus on the waveguide and detector. We use plasmonic subwavelength waveguides as optical interconnects. The plasmonic waves carry information, and propagate along the waveguide close to the speed of light. At the end of the waveguide, the plasmonic waves are converted into electronic signals through a plasmonic converter. Fig. 1 shows a schematic diagram of a metal–insulator–metal subwavelength waveguide coupled to a plasmonic converter.

2. Subwavelength waveguides The major challenge for the on-chip optical data transmission is to confine the electromagnetic waves in a waveguide of nanometer size compatible with that of the electric components. Previous studies show that plasmonics could be confined in, and propagated along, a waveguide with dimensions below one-tenth of the free-space wavelength [4]. Different plasmonic waveguides have been studied, including metal–insulator–metal (MIM),

E-mail address: baiping@ihpc.a-star.edu.sg (P. Bai). 0921-4526/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.physb.2010.01.017

Please cite this article as: P. Bai, et al., Physica B (2010), doi:10.1016/j.physb.2010.01.017

ARTICLE IN PRESS
2 P. Bai et al. / Physica B ] (]]]]) ]]]–]]]

insulator-metal-insulator, and dielectric-loaded structures. The MIM waveguide is most suitable for the on-chip data transmission as it provides a better trade-off between the lateral confinement and the propagation loss [10]. For the purpose of this paper we have considered a subwavelength MIM waveguide. Fig. 2a shows the schematics of the MIM waveguide, which consists of a dielectric slot between two thin metallic films. The width and the thickness of each metal film are 1 mm and 50 nm, respectively. The gap between the two metal films is 50 nm. Fig. 2b shows the electric field propagating along the Ag– air–Ag waveguide at a free-space wavelength of 1.55 mm, simulated with CST Microwave Studio [11]. Plasmonic waves are well confined within the nanoscale slot. The distribution of the electric field in the xz-plane across the center of the waveguide is shown in Fig. 2c. We notice that the magnitude of the electric field decreases along the waveguide. Fig. 3 shows the relationship between the propagation distance and the wavelength. It shows that the propagation distance increases when the free-space wavelength of the input field is increased. This can be attributed to the fact that the metal becomes less lossy at lower frequency. That is why the metal can be treated as a perfect conductor in microwave regime. Therefore, shorter freespace wavelength corresponds to a larger propagation loss.

The length of an on-chip interconnect can range from submicrometers (local interconnects) to centimeters (global interconnects) in an advanced electronic chip. However, the MIM plasmonic waveguide is very lossy due to high absorption of light in the optical frequency regime. The propagation distance in the MIM waveguide can only reach a distance of deca-micrometers with the field being reasonably confined. The performance of existing plasmonic waveguides is not good enough for on-chip interconnect applications yet. There is urgent need to increase the propagation distance of plasmonic waveguides while maintain a high confinement in order to integrate them into semiconductor-based electronic circuits. Fortunately, studies on novel materials, such as gain materials, plasmonic metamaterials and epsilon-near-zero materials, and on novel structures have shown a very promising future for the implementation of subwavelength waveguides with ultra-low propagation loss [8].

Plasmonic Waveguide

Plasmonic Converter SiO2

Y Z

Substrate X
Fig. 1. Schematic illustration shows plasmonic waves propagating along a metal– insulator–metal subwavelength waveguide and detected by a plasmonics-toelectronics converter. Fig. 3. Propagation distance of the Ag–air–Ag waveguide (slot cross-section of 50 nm by 50 nm) as a function of free-space wavelength.

Fig. 2. Plasmonic waves propagating along an MIM waveguide: (a) 2-D illustration of the MIM waveguide, (b) 3-D electric fields propagating along the Ag–air–Ag waveguide with a cross-section of 50 nm  50 nm and at a free-space wavelength of 1.55 mm and (c) the distribution of the electric fields in the xz-plane across the center of the waveguide.

Please cite this article as: P. Bai, et al., Physica B (2010), doi:10.1016/j.physb.2010.01.017

ARTICLE IN PRESS
P. Bai et al. / Physica B ] (]]]]) ]]]–]]] 3

3. Plasmonics–electronics conversion In order to integrate a plasmonic subwavelength waveguide into an electronic circuit, a novel plasmonic converter with a high bandwidth is required to convert plasmonic signals to electronic ones. Various photodetectors have been developed to detect electromagnetic waves propagating along a conventional dielectric waveguide at the micrometer scale [12]. The plasmonic converter, however, cannot be obtained by simply scaling down the traditional micrometer-scaled waveguide photodetector. Firstly, plasmonics, unlike traditional optics which is based on diffraction phenomena in dielectrics, confines electromagnetic fields along the interface between metal and dielectric. The plasmonic waves are generally very weak at the end of the waveguide as the plasmonic waveguide is very lossy compared with the dielectric one. Secondly, the plasmonic waves decay into surrounding dielectrics very rapidly at the open end of the plasmonic waveguide. Only a small fraction of the plasmonic energy can reach the active region of a plasmonic converter if the converter is simply attached to the end of the waveguide. Fig. 4a shows the electric field in the xy-plane, at 50 nm away from the end of an MIM waveguide. The field is too weak to be detected. Thirdly, there will also be a large reflection at the interface between the waveguide and the converter due to a large contrast of permittivities of the waveguide and the converter. Lastly, the nanoscale photodetector does not have enough space to form a resonant cavity, which needs a size of (n + 1) * l/2 where n is an integer. Directly scaling down of the traditional waveguide photodetector cannot provide enough photoresponsivity to detect the weak plasmonic waves. Therefore, the coupling between the plasmonic waveguide and the converter becomes a major challenge for the design of a plasmonics-to-electronics converter. In our study, metallic nanorods are used to couple plasmonic waves from the subwavelength waveguide to the plasmonic converter, and enhance the electromagnetic near-fields around the nanorods at the same time. The optical properties of the nanoscale metallic particles enable localized surface plasmon modes around the particles. The metallic nanorods can also be considered to form a dipole nanoantenna at optical frequencies analogue to that at radio or microwave frequencies [13]. We place two metallic nanorods near the end of the plasmonic waveguide as shown in Fig. 1. The two nanorods, which act as a dipole nanoantenna, receive the plasmonic waves from the waveguide. As a result, the electromagnetic fields are coupled from the

plasmonic waveguide to the nanorods, and the near-fields are formed and concentrated in the area around the nanorods [14]. The dimensions of the metallic nanorods have a significant effect on the enhancement of the near-field around the nanorods. In our study, the plasmonic waves in the MIM waveguide are excited with light at a free-space wavelength of 1.55 mm. Two Ag nanorods are used to collect the plasmonic waves. Each nanorod has a dimension of 150 Â 50 Â 50 nm3; and the gap between two nanorods is optimized to be 50 nm to give the highest electric fields in the gap region. The total length between two extreme ends of the nanorods is the most sensitive parameter affecting the enhancement of the near-field around the nanorods. This is analogue to the effective length of a half-wavelength dipole antenna in microwave applications. However, the optimal length of this nanorod antenna is found to be 350 nm, which is much shorter than one-half of 1.55 mm. The half-wave dipole resonance here might correspond to a shorter effective wavelength, which is much shorter than that in free space owing to the plasmonic effects of the metallic nanorods [15]. Fig. 4 shows the calculated electric field in the xy-plane across the center of the nanorods without and with the nanoantenna. Results show that, the near-field around the nanorods is greatly condensed by applying the nanoantenna. The electric fields in the gap region can be enhanced by more than 100 times with the optimized geometry of the nanoantenna. Once the optical power is concentrated in a nano-scaled volume, the plasmonic converter can be designed by using the structures of traditional photodetectors with its active region enclosing the volume. A PIN photodetector structure is used in our study as shown in Fig. 1. An absorption material InGaAs is filled in the volume around the two nanorods where the near-fields are locally enhanced. The P-I-N layered structure is formed in the ydirection with two electrodes located at the bottom and the top of the PIN layers, respectively. The photonic energy from the plasmonic waves, which can be treated as photons incident in the semiconductor material InGaAs, is absorbed by the electrons in the valence band. These high-energy electrons are excited to become free electrons in the conduction band, and leave holes in the valence band. As a result, electron–hole pairs are generated in the absorption material, and the electrons and holes drift to the two electrodes separately to form photocurrents under an external bias voltage. The electronic characteristics of the plasmonic converter are simulated based on the Silvaco TCAD software [16]. A twodimensional model is built in the xy-plane across the center of the nanorods. The photogeneration rate in the absorption material is modeled with G ¼ ZjEj2 where

la
2p‘ c

ð1Þ

a ¼ 4pk=l Z is the internal quantum efficiency; ‘ is Planck’s constant; l is the free space wavelength; c is the speed of light; a is the absorption coefficient; k is the imaginary part of the refraction index of the absorption material; and E is the integrated value of the electric field in the z-direction. Fig. 5 shows the calculated photogeneration in the active area of the converter. It can be seen that the density distribution of the excited carriers is proportional to the distribution of the electromagnetic field as shown in Fig. 4b. The calculated photocurrent of the converter as a function of the bias voltage is shown in Fig. 6. This proposed plasmonic converter would have very high responsivity and large bandwidth. The weak plasmonic waves at the end of the plasmonic waveguide have been coupled and

Fig. 4. Electric fields in the xy-plane 50 nm from the end of the waveguide: (a) without nanoantenna and (b) with the nanoantenna.

Please cite this article as: P. Bai, et al., Physica B (2010), doi:10.1016/j.physb.2010.01.017

ARTICLE IN PRESS
4 P. Bai et al. / Physica B ] (]]]]) ]]]–]]]

-0.4 -0.3 -0.2 Y (Microns) -0.1 0 0.1 0.2 0.3 0.4 0.5
Photogeneration 1.87e+21 1.5e+21 1.12e+21 7.48e+20 3.74e+20 0

p+

200 nm. The transit time for the excited carriers to travel to the two electrodes can be o1 ps. This not only greatly reduces the chances for the recombination of electron–hole pairs, leading to high responsivity, but also allows the converter to operate at a very high speed. The rough estimation suggests that the converter is able to operate in THz frequency range.

4. Conclusion We have proposed a method to enhance the performance of electronic circuits by using subwavelength plasmonic waveguides as on-chip optical interconnects. We have transmitted data by using MIM waveguides that can propagate plasmonic waves in a cross-section much smaller than the free-space wavelength. A plasmonics-to-electronics converter has been designed to convert the plasmonic waves into electric signals. Two nanorods, which form a dipole nanoantenna, are used to couple the plasmonic waves from the waveguide to the converter. The electromagnetic near-field around the nanorods, which are inside the active area of the converter, is greatly enhanced through the localized surface plasmon effects. The converter has ultra-compact dimensions, high responsivity, potential THz bandwidth, and very low power dispersion. The nanoscale features of the plasmonic waveguide and the converter, comparable to those of the nanoelectronic counterparts, are paving the way for the integration of optical and electronic components on the same chip; and hence a tremendous synergy could be reached by combining the advantages of both photonics and nanoelectronics. References
[1] R. Kirchain, L. Kimerling, Nat. Photonics 1 (2007) 303. [2] R.G. Beausoleil, P.J. Kuekes, G.S. Snider, S.Y. Wang, R.S. Williams, Proc. IEEE 96 (2008) 230. [3] H.A. Atwater, Sci. Am. 296 (2007) 56. [4] R.F. Oulton, V.J. Sorger, D.A. Genov, D.F.P. Pile, X. Zhang, Nat. Photonics 2 (2008) 496. ´ [5] K.F. MacDonald, Z.L. Samson, M.I. Stockman, N.I. Zheludev, Nat. Photonics 3 (2009) 55. [6] M. Silveirinha, N. Engheta, Phys. Rev. Lett. 97 (2006) 157403. [7] J. Seidel, S. Grafstrom, L. Eng, Phys. Rev. Lett. 94 (2005) 177401. [8] Z. Yu, G. Veronis, S. Fan, Appl. Phys. Lett. 92 (2008) 041117. [9] E. Ozbay, Science 311 (2006) 189. [10] G. Veronis, S. Fan., J. Lightwave Tech. 25 (2007) 2511. [11] /http://www.cst.com/Content/Products/MWS/Overview.aspxS. ´ ´ [12] L. Vivien, J. Osmond, J.M. Fedeli, D. Marris-Morini, P. Crozat, J.F. Damlencourt, E. Cassan, Y. Lecunff, S. Laval, Opt. Express 17 (2009) 6252.  [13] A. Alu, N. Engheta, Nat. Photonics 2 (2008) 307. [14] L. Tang, S.E. Kocabas, S. Latif, A.K. Okyay, D.S. Ly-Gagnon, K.C. Saraswat, D.A.B. Miller, Nat. Photonics 2 (2008) 226. [15] P. Muhlschlegel, H.J. Eisler, O.J.F. Martin, B. Hecht, D.W. Pohl, Science 308 (2005) 1607. [16] /http://www.silvaco.com/products/TCAD.htmlS.

n-

-0.2

-0.1

0 X (Microns)

0.1

0.2

Fig. 5. Photogeneration in the cross section via the nanorods in the plasmonic converter. The value has been integrated in the z-direction.

Cathode Current (A) 1e-09 8e-10 6e-10 4e-10 2e-10 0 0 0.1 0.2 0.3 0.4 0.5

Cathode Voltage (V)
Fig. 6. Photocurrent of the plasmonic converter as a function of the bias.

highly enhanced within the small volume around the nanorods. This leads to a high responsivity of the plasmonic waves. The active area of the plasmonic converter is at a nanometer scale as the field is concentrated in a volume approximately 500 nm  200 nm  100 nm. At such a small size, the capacitance of the converter is about 15af as predicted by a simple parallel-plate model. This allows relatively large external resistance. Note that the distance between the two electrodes can be as small as

Please cite this article as: P. Bai, et al., Physica B (2010), doi:10.1016/j.physb.2010.01.017