Towards ultra-subwavelength optical latches Jianwei Mu, Zhaohong Han, Stefano Grillanda, Andrea Melloni, Jurgen Michel, L. C. Kimerling, and Anu Agarwal Citation: Applied Physics Letters 103, 043115 (2013); doi: 10.1063/1.4816755 View online: http://dx.doi.org/10.1063/1.4816755 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/103/4?ver=pdfcov Published by the AIP Publishing
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.113.126.132 On: Sun, 23 Feb 2014 03:42:59
APPLIED PHYSICS LETTERS 103, 043115 (2013)
Towards ultra-subwavelength optical latches Jianwei Mu,1,a),b) Zhaohong Han,1,a) Stefano Grillanda,1,2 Andrea Melloni,2 Jurgen Michel,1 L. C. Kimerling,1 and Anu Agarwal1
1 Microphotonics Center and Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA 2 Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, 20133 Milano, Italy
(Received 23 June 2013; accepted 10 July 2013; published online 24 July 2013) We propose an electrically driven nanometer scale plasmonic optical latch integrated with Ge2Sb2Te5 chalcogenide glasses. For an effective switching length of 167 nm, the extinction ratio between the ON and OFF states of the proposed switch is as large as 10 dB, with net operation energy as low as C 2013 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4816755] 30.4 pJ per cycle. V Photonic Integrated Circuits (PICs) have been considered as the critical technology for short reach communication systems, in which functional photonic devices with low power consumption and small footprint are highly desired.1 Functional devices based on surface plasmonic polaritons (SPPs) have been proposed and studied to overcome the diffraction limit and increase the integration density; however, only few have addressed the active metallic nano-scale device2 such as optical ON/OFF switches (latches). Generally, optical ON/OFF switches can be grouped by their working mechanism:3 optomechanical switches with low insertion loss and crosstalk but slow switching speed (millisecond scale); electro-optic switches where the refractive index of the material in the coupling region of a directional coupler is varied by applying a voltage; semiconductor optical switches that employ ring resonators or Mach-Zehnder interferometers (MZI) in which the refractive index of the optical path is changed by carrier injection mechanisms; switching achieved by thermo-optic effect, i.e., variation of the refractive index by changing the temperature and other switches based on liquid crystals or acoustic effects. Recently, optical switches based on Ge2Sb2Te5 (GST) chalcogenide glasses have been proposed4 and demonstrated.5–7 The optical properties of GSTs, i.e., refractive index and absorption, are changed by switching the material state between amorphous and crystalline, simply by applying a near band-gap optical field. Moreover, the index and absorption change remains in the material with no need of a continuous power supply, thus making these devices appealing for energy saving and optical memory applications. Ikuma et al. experimentally demonstrated an optically controlled GST switch with a 9.7 dB extinction ratio between the amorphous and crystalline states in the wavelength range from 1525 to 1600 nm.6 However, the control signal is applied through an external pulsed laser whose beam needs to be precisely aligned with the waveguide. Here, we propose an ultra-subwavelength ON/OFF optical switch based on the variation of the attenuation. It consists of a metal-dielectric-metal waveguide structure integrated with Ge2Sb2Te5 chalcogenide glasses. There are two important characteristics of GSTs that enable GSTassisted optical latches driven by electrical signals.8–12 First, a)
Jianwei Mu and Zhaohong Han contributed equally to this work. Corresponding author. Email:
[email protected].
b)
0003-6951/2013/103(4)/043115/4/$30.00
GST materials change their state from amorphous to crystalline by simple Joule heating.8 Second, accurate spatial location control is obtained by lithographically patterned electrical contacts, hence alignment is embedded. Our proposed structure overcomes the two main challenges for electrically driven schemes: (i) metal contacts demonstrating low absorption, low resistance, and high temperature stability and (ii) low loss optical wave guiding with very thin (tens of nanometers) GST films. Our nanometer scale latch exhibits a high extinction ratio between ON and OFF states. Compared to other nano-optical switches,13–15 our device is easy to fabricate with good spatial control and is amenable to integration. It is expected to find promising applications in optical signal routing and switching. The schematic of the proposed optical latch is shown in Fig. 1. The GST is sandwiched by two gold thin films. Gold (i) enables the formation of an optical waveguide by Plasmon polaritonic resonance effect and (ii) serves as an electrical contact. In this simulation, at 1550 nm, the refractive index of GST is set to 4.4 þ i0.098 for the amorphous state and 7.1 þ i0.78 for the crystalline state,6 while the refractive index of the gold is 0.55 þ i11.51.16 The entire structure is surrounded laterally by an electrical isolating SiO2 (refractive index 1.45) layer. We use finite element method based multi-physics software to guide the design and analysis of the structure. In the simulations, the following parameters are employed: a
FIG. 1. Cross section of an electrically driven GST optical latch: the GST material is sandwiched by two gold thin films and laterally surrounded by SiO2 for isolation. Gold (i) enables the formation of an optical waveguide by Plasmon polariton resonance effect and (ii) serves as an electrical contact.
103, 043115-1
C 2013 AIP Publishing LLC V
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.113.126.132 On: Sun, 23 Feb 2014 03:42:59
043115-2
Mu et al.
Appl. Phys. Lett. 103, 043115 (2013)
FIG. 2. Simulated electric field (normal to the surface) of the optical mode of the GST-assisted plasmonic latch when the GST is in the ON state (amorphous) and OFF state (crystalline), assuming a waveguide core width of 30 nm and thickness of 50 nm. At 1550 nm, refractive index of GST is 4.4 þ i0.098 and 7.1 þ i0.78, respectively, for amorphous and crystalline states, while the refractive index of gold is 0.55 þ i11.5. The effective indices are 4.184-i0.279 (ON) and 8.3359-i1.982 (OFF).
thickness of the GST layer of 50 nm, typical of electronic phase change materials (PCMs);18 a waveguide width w ¼ 30 nm; and thickness of the gold contacts of t ¼ 100 nm. The electric field of the optical mode of the proposed structure for the two states of the latch is represented in Fig. 2. The light is well confined in the waveguide due to the plasmonic resonance effect. The coupling between the GST and the metal is enhanced along the interface for the OFF states since the refractive index of GST in the crystalline state is larger than in the amorphous state. There are two important geometric parameters that affect the performance of the latch, namely, the width and the thickness of the GST film. We first investigate the impact of GST film thickness. Figure 3(a) represents the total insertion loss per unit of length of the latch as a function of the GST thickness from 10 to 50 nm (waveguide width is fixed to 30 nm). For both states, the insertion loss of the latch decreases with thicker GST films. However, the OFF state (crystalline GST) always experiences a higher level of loss than the ON state (amorphous GST) as the GST film thickness decreases, because SPP resonance between GST and the gold layer is enhanced for the thinner film, thus inducing a higher absorption loss. Figure 3(b) shows the variation of insertion loss per unit of length of the latch as a function of the waveguide core width from 20 to 100 nm (the thickness of the GST layer is fixed to 50 nm). In general, losses vary differently with waveguide width in the ON (amorphous GST) and OFF (crystalline GST) states. In the ON state, optical losses do not exhibit significant changes with waveguide width.
FIG. 3. Insertion loss per unit of length of the latch (a) as a function of the GST thickness when the waveguide width is fixed to 30 nm and (b) as a function of the waveguide width when the GST thickness is fixed to 50 nm. Black solid lines and square markers indicate the ON state (amorphous GST), while red lines and circle markers refer to the OFF state (crystalline GST). Insertion loss versus waveguide width is shifted by 60 dB in order to easily read the image.
However, in the OFF state, higher width dependence is seen as higher losses are exhibited because coupling with the lossy metal layer is stronger. Figure 3(b) also shows that the insertion loss for ON state changes slowly with the variation of waveguide width, while the insertion loss for OFF state is significantly increased with the waveguide width. This indicates that a larger waveguide width will result in a higher extinction ratio between ON and OFF states. On the other hand, more energy is required to drive the phase change of the material. Therefore, there exists an optimum waveguide width if both between energy cost and extinction ratio are considered, as we will show below. To find the energy cost and operation time of our latch structure, we build the Joule heat model that couples both electrical and thermal properties of the materials (listed in Table I). Although thinner GST layers (such as 30 or 40 nm) provide higher extinction ratios, we choose a thickness of 50 nm for the GST film because material parameters required for our electrical simulations are readily available for this thickness.17,18 We begin our analysis with the switching-off process, i.e., the transition of the GST material from the amorphous to crystalline state. The process is essentially a recrystallization under high temperature, yet below the melting point of the material. The resistivity of amorphous GST is so high (typically about 10 Xm) that there is almost no measurable current below the threshold voltage.8 However, amorphous GST has a threshold voltage, occurring at a certain current density, above which it becomes conductive without any phase change and consequently generates Joule heating that induces the phase change.9,10 The typical switching-off time (or “Set” process in PCM) of 100 ns is used.8 This is much longer than the typical switching-on time (or “Reset” process), which is around 1 ns,18 which indicates that the energy consumption of the switching-off process is dominant, therefore requiring only an optimization of the energy cost for the switching-off process. It should be pointed out that faster switching mechanisms of GSTs are hard to be found and lack detailed study, but it is expected that both the switching time and power cost will decrease due to the increased free carriers and structure defects.18 Figure 4 reports the power density and total energy of our latch. As shown in Fig. 4(a), the electrical power density required for the heating process decreases with increasing waveguide width because the effective surface increases,
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.113.126.132 On: Sun, 23 Feb 2014 03:42:59
043115-3
Mu et al.
Appl. Phys. Lett. 103, 043115 (2013)
TABLE I. Material parameters used in the electrical and thermal simulations. Material GST (crystalline) GST (amorphous) GST (amorphousabove threshold voltage) Au
Conductivity (X1 m1)
Heat capacity (J Kg1 K1)
Thermal conductivity (W m1K1)
2.77 103 (Ref. 19) 0.1 (Ref. 19) 2.77 103 (Ref. 19) 4.52 107 (Ref. 22)
210 (Ref. 20) 210 (Ref. 20) 210 (Ref. 20) 129 (Ref. 22)
0.24 (Ref. 21) 0.28 (Ref. 21) 0.28 (Ref. 21) 318 (Ref. 22)
thus making the heat transfer into the substrate faster. However, the energy cost per unit of length of the latch required by the switching-off operation increases with increasing waveguide width because of the larger volume of GST that needs to be crystallized. On the other hand, as shown previously, larger width also means larger extinction ratio per unit of length of the device. Figure 4(b) instead reports the total energy required for the switching-off process given a desired absorption change of 10 dB between the ON and OFF states. The optimal dimension for a 50 nm thick GST is 30 nm (width). With a length of 167 nm (length), this device gives an insertion loss of 1.64 (On) and 11.64 dB (Off), with operation energy of only 30 pJ. It is noted that there are several ways to increase the extinction ratio between ON/OFF states, such as using longer device length, thinner GST layers, or larger waveguide width. For example, the extinction ratio will increase to 20 dB if we extend the device length from 167 nm to 334 nm. For the switching-on process, the crystalline GST melts into liquid and then is rapidly quenched, typically within a few nanoseconds,17 which results in an amorphization of the structure.18 Assuming that the conductivity of the crystalline GST does not change before melting and applying a voltage of 3.5 V,17 we calculate a current density of 1.94 104 kA/cm2. After 57 ps, the maximum temperature reaches the melting temperature (883 K).19 With further heating, GST will melt and the latent heat (1.37 105 J/kg) will consume most of the input power,23 pinning the temperature to the melting point. By treating the conductivity of GST as constant and assuming that all the heating energy goes into latent heat after 57 ps, the total phase transition time is only 0.12 ns, which is a little shorter than other reports.18 The energy consumption required for the phase transition is 0.4 pJ, which is much smaller than that of the
switching-off process. So the energy consumption per cycle (including both the switching-on and switching-off processes) is 30.4 pJ. In terms of footprint and electrical power consumption, using a figure-of-merit (FOM) defined as power times the footprint of the device, the FOM of our switch scales down with operation frequency, thanks to the intrinsic self-holding nature of the switch. The operation time for our switch is 100 ns. We estimate that the overall energy cost of our device (including packaging) will be similar to PCM devices, which is 290 pJ per cycle. By comparing our device with other switches based on different schemes such as optical Microelectromechanical systems (MEMS),24 silica-based Mach-Zehnder interferometer (MZI),25 thermooptical,26 electroholographic,27 Semiconductor Optical Amplifier (SOA)based,28 acousto-optic,29 and microdisks,30 it is clear that the FOM of this device within the frequency cut-off is higher than most of other optical switches, opening the way to high component density within certain power limits and area constraints. In conclusion, we propose an electrically driven plasmonic optical latch integrated with Ge2Sb2Te5 chalcogenide glass. The latch is characterized by a nanometer scale size and a high extinction ratio. For a 167 nm long latch, the total energy required is 30 pJ for switching-off and 0.4 pJ for switching-on processes, while the extinction ratio between amorphous and crystalline states is 10 dB at the wavelength of 1550 nm. J. Mu acknowledges the financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) through a postdoctoral fellowship at MIT. S. Grillanda acknowledges the financial support of the Progetto Roberto Rocca Fellowship. 1
FIG. 4. (a) Electrical power density (black solid line and square markers) and energy cost per unit of length (red solid line and square markers) required by the switching-off process as a function of the waveguide width (thickness is fixed to 50 nm) and (b) total energy required by the switchingoff operation as a function of the waveguide width, showing an optimal waveguide width of 30 nm.
L. C. Kimerling, L. D. Negro, S. Saini, Y. Yi, D. Ahn, S. Akiyama, D. Cannon, J. Liu, J. G. Sandland, D. Sparacin, J. Michel, K. Wada, and M. R. Watts, Top. Appl. Phys. 94, 89 (2004). 2 T. Liang and D. A. B. Miller, J. Nanophotonics 3, 030302 (2009). 3 G. I. Papadimitriou, C. Papazoglou, and A. S. Pomportsis, J. Lightwave Technol. 21, 384 (2003). 4 W. H. P. Pernice and H. Bhaskaran, Appl. Phys. Lett. 101, 171101 (2012). 5 Y. Ikuma, T. Saiki, and H. Tsuda, IEICE Electron. Express 5, 442 (2008). 6 Y. Ikuma, Y. Shoji, M. Kuwahara, X. Wang, K. Kintaka, H. Kawashima, D. Tanaka, and H. Tsuda, Electron. Lett. 46, 1460 (2010). 7 Y. Ikuma, Y. Shoji, M. Kuwahara, X. Wang, K. Kintaka, H. Kawashima, D. Tanaka, and H. Tsuda, Electron. Lett. 46, 368 (2010). 8 M. H. R. Lankhorst, B. W. S. M. M. Ketelaars, and R. A. M. Wolters, Nature Mater. 4, 347 (2005). 9 M. Wuttig and N. Yamada, Nature Mater. 6, 824 (2007). 10 A. Pirovano, A. L. Lacaita, F. Pellizzer, S. A. Kostylev, A. Benvenuti, and R. Bez, IEEE Trans. Electron Devices 51, 714 (2004). 11 D. Ielmini, A. L. Lacaita, A. Pirovano, F. Pellizzer, and R. Bez, IEEE Electron Device Lett. 25, 507 (2004).
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.113.126.132 On: Sun, 23 Feb 2014 03:42:59
043115-4 12
Mu et al.
H. Lv, P. Zhou, Y. Lin, T. Tang, B. Qiao, Y. Lai, J. Feng, B. Cai, and B. Chen, Microelectron. J. 37, 982 (2006). 13 G. Wang, H. Lu, X. Liu, and Y. Gong, Nanotechnology 23, 444009 (2012). 14 C. Min, P. Wang, C. Chen, Y. Deng, Y. Lu, H. Ming, T. Ning, Y. Zhou, and G. Yang, Opt. Lett. 33, 869 (2008). 15 B. Piccione, C. Cho, L. K. V. Vugt, and R. Agarwal, Nat. Nanotechnol. 7, 640 (2012). 16 E. D. Palik, Handbook of Optical Constants of Solids I (Academic, 1985). 17 D. Q. Huang, X. S. Miao, Z. Li, J. J. Sheng, J. J. Sun, J. H. Peng, J. H. Wang, Y. Chen, and X. M. Long, Appl. Phys. Lett. 98, 242106 (2011). 18 W. J. Wang, L. P. Shi, R. Zhao, K. G. Lim, H. K. Lee, T. C. Chong, and Y. H. Wu, Appl. Phys. Lett. 93, 043121 (2008). 19 T. Zhang, Z. Song, Y. Gong, Y. Lin, C. Xu, Y. Chen, B. Liu, and S. Feng, Appl. Phys. Lett. 92, 113503 (2008). 20 D. H. Kang, I. H. Kim, J. Jeong, B. Cheong, D. H. Ahn, D. Lee, H. M. Kim, K. B. Kim, and S. Y. Kim, J. Appl. Phys. 100, 054506 (2006). 21 V. Giraud, J. Cluzel, V. Sousa, A. Jacquot, A. Dauscher, B. Lenoir, H. Scherrrer, and S. Romer, J. Appl. Phys. 98, 013520 (2005).
Appl. Phys. Lett. 103, 043115 (2013) 22
R. A. Serway, Principles of Physics (Saunders College Publishing, 1998). S. Okamine, S. Hirasawa, M. Terao, and Y. Miyauchi, Opt. Data Storage 1316, 315 (1992). 24 P. De Dobbelaere, K. Falta, L. Fan, S. Gloeckner, and S. Patra, IEEE Commun. Mag. 40, 88 (2002). 25 M. R. Watts, J. Sun, C. DeRose, D. C. Trotter, R. W. Young, and G. N. Nielson, Opt. Lett. 38, 733 (2013). 26 M. H. Ibrahim, M. Yaacob, N. M. Kassim, A. S. M. Supa’at, and A. B. Mohammad, Optik (Jena) 10, 1016 (2013). 27 J. Agranat, “Electroholographic wavelength selective crossconnect,” in LEOS Summer Topical Meetings (1999), pp. 61–62. 28 V. Eramo, E. Miucci, A. Cianfrani, A. Germoni, and M. Listanti, Selected Topics on Optical Amplifiers in Present Scenario, edited by Sisir Kumar Garai, (InTech, 2012), Chap. 4. 29 X. Ma and G. Kuo, “Optical switching technology comparison: Optical MEMS vs. other technologies,” IEEE Commun. Mag. 41, S16 (2003). 30 M. R. Watts, W. A. Zortman, D. C. Trotter, R. W. Young, and A. L. Lentine, Opt. Express 19, 21989 (2011). 23
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 130.113.126.132 On: Sun, 23 Feb 2014 03:42:59
