D4.2: Fabrication and performance evaluation of the WRR 2x2 DLSPP switch

ICT - Information and Communication Technologies Merging Plasmonic and Silicon Photonics Technology towards Tb/s routing in optical interconnects Collaborative Project Grant Agreement Number 249135 D4.2: Fabrication and performance evaluation of the WRR 2x2 DLSPP switch Due Date of Deliverable: 30/09/2011 Actual Submission Date: 07/11/2011 Revision: Draft Start date of project: January 1 st 2010 Duration: 36 months Organization name of lead contractor for this deliverable: SDU Author: S.I. Bozhevolnyi (SDU) Contributors: A. Kumar (SDU), V.S. Volkov (SDU), J. Gosciniak (SDU), S.I. Bozhevolnyi (SDU), K. Hassan (UB), L. Markey (UB), J.C. Weeber (UB), A.Dereux (UB), O. Tsilipakos (CERTH/ITI) A. Pitilakis (CERTH/ITI), E. Kriezis (CERTH/ITI), K. Vyrsokinos (CERTH/ITI), G. Kalfas (CERTH/ITI), S. Papaioannou (CERTH/ITI), N. Pleros (CERTH/ITI), D. Kalavrouziotis (ICCS/NTUA), G. Giannoulis (ICCS/NTUA), D. Apostolopoulos (ICCS/NTUA), H. Avramopoulos (ICCS/NTUA) November 7, 2011 FP7-249135 The PLATON Consortium Page 1 of 16

Project Information PROJECT Project name: Project acronym: Project start date: Project duration: Contract number: Project coordinator: Instrument: Activity: Merging Plasmonic and Silicon Photonics Technology towards Tb/s routing in optical interconnects PLATON 01/01/2010 36 months 249135 Nikos Pleros CERTH STREP THEME CHALLENGE 3: Components, Systems, Engineering DOCUMENT Document title: Document type: Deliverable number: Contractual date of delivery: Calendar date of delivery: Editor: Authors: Workpackage number: Workpackage title: Fabrication and performance evaluation of the WRR 2x2 DLSPP switch Report D4.2 30/09/2011 07/11/2011 S.I. Bozhevolnyi (SDU) A. Kumar, V.S. Volkov, J. Gosciniak, S.I. Bozhevolnyi, K. Hassan, L. Markey, J.C. Weeber, A.Dereux, O. Tsilipakos, A. Pitilakis, E. Kriezis, K. Vyrsokinos, G. Kalfas, S. Papaioannou, N. Pleros, D. Kalavrouziotis, G. Giannoulis, D. Apostolopoulos, H. Avramopoulos WP4 Development and system-evaluation of plasmonic switching elements Lead partner: Dissemination level: Date created: SDU CO 24/08/2011 Updated: Version: Total number of Pages: Document status: Final 16 Final November 7, 2011 FP7-249135 The PLATON Consortium Page 2 of 16

TABLE OF CONTENTS 1 EXECUTIVE SUMMARY... 4 2 INTRODUCTION... 5 3 DESIGN OF DUAL RESONATOR 2X2 DLSPP SWITCHES... 6 3.1 PMMA LOADED SWITCHES... 6 3.2 CYCLOMER LOADED SWITCHES... 9 4 EXPERIMENTAL EVALUATION OF PMMA AND CYCLOMER LOADED SOI DLSPP WRR SWITCHES... 11 4.1 EVALUATION OF AN ALL PASS PMMA LOADED WRR... 11 4.2 FABRICATION OF THE CYCLOMER LOADED DUAL RESONATOR 2X2 SWITCHES... 12 4.3 EXPERIMENTAL RESULTS... 12 5 CONCLUSION... 14 ABBREVIATIONS... 15 REFERENCES... 16 November 7, 2011 FP7-249135 The PLATON Consortium Page 3 of 16

1 Executive Summary This document provides a detailed analysis of the design and the first experimental results obtained with 2x2 Waveguide Ring Resonators (WRRs) that are considered as possible switching element for adoption in PLATON s platform 4x4 switching matrix. Switching in WRRs is achieved by tuning its resonant frequency through heating of the plasmonic waveguide resulting a variation of its effective refractive index. The heating is controlled by the appliance of a voltage on the gold layer placed under the polymer. The maximum temperature variation and maximum wavelength shift is determined by the polymer type that is utilized in the plasmonic waveguide. Within PLATON, two possible materials are examined; PMMA with -1.05 x 10-4 1/ 0 C Thermo Optic Coefficient (TOC) for T=80 0 C and Cyclomer with TOC= -2.90 x 10-4 1/ 0 C [see D2.3] for the same temperature change. The design of the WRRs was performed for plasmonic waveguides loaded both with PMMA and Clyclomer polymers following the three-dimensional vector finite element method (3D-FEM). The design was based on the dual-resonator add-drop filter layout with perpendicular access waveguides, concluding to the optimal parameters in terms of: ring radius, racetrack section length, waveguide gaps, crossing dimensions. The Extinction Ratio (ER) performance for both output ports of the PMMA-loaded device was always found to be better than 5 db over a spectral window larger than 5nm, assuming an electrically induced temperature change of 80K. In the case of a Cyclomer-loaded dualresonator switch, a temperature shift of 80 K provided an extinction ratio of 6 db over a broad wavelength range of up to 6nm, ensuring the necessary multi-wavelength operation for PLATON s traffic format. This report includes also the first experimental results that were obtained using a DLSPP WRR device integrated on a SOI waveguide platform. An all pass PMMA-loaded WRR with R= 5.5um radius, L=0.8um and 0.35um gap placed on top of the Si board was characterized in terms of its thermo-optic switching performance, revealing a wavelength shift of 7nm for an applied electrical current of 50mA and required power of less than 3.5mW. These results indicate the suitability of the DLSPP-on-SOI platform for enabling high-quality and low energy thermooptic switching. Moreover, the thermo-optic switching performance of the new polymer loading DLSPP platform has been characterized, using a Cyclomer-loaded dual-resonator add-drop filter layout. A wavelength shift of 9nm was observed for an electrically induced temperature deviation of Τ=80K (HOT state). The measured ER values were higher than 6dB over a 6nm spectral range around 1558nm for both output ports, confirming the broadband switching characteristics of the Cyclomer-loaded switch that conforms with PLATON platform routing specifications described in detail in D2.4 November 7, 2011 FP7-249135 The PLATON Consortium Page 4 of 16

2 Introduction According to the work plan, this deliverable is related to WP4 objectives summarized as: a) To describe the results obtained from simulations regarding the optimum parameters for the dual-resonator add-drop filter design with: i. PMMA loading and untreated waveguide crossing. ii. iii. PMMA loading and treated (tapered) waveguide crossing. Cyclomer loading b) To describe the experimental results derived from an all pass WRR loaded with PMMA polymer c) To provide the experimental results that were received during the evaluation of the dual-resonator add-drop filter loaded with cyclomer polymer This deliverable is associated with Task 4.2 Fabrication, characterization and systemlevel evaluation of single element 2x2 thermo-optic waveguide-ring resonator plasmonic switch [M9-M21]. The WRRs switching elements are considered as a possible solution for adoption in PLATON s 2x2 and 4x4 switching matrix. November 7, 2011 FP7-249135 The PLATON Consortium Page 5 of 16

3 Design of dual-resonator 2x2 DLSPP switches In the process of designing resonator-based 2x2 switching elements, several configurations have been examined. Specifically, the performance of components with parallel or perpendicular access waveguides and one or two resonators has been thoroughly investigated [1]. All simulations were conducted by utilizing the threedimensional vector finite element method (3D-FEM). 3.1 PMMA-loaded switches Initially, the polymer of choice was PMMA. In this case, the only component capable of delivering high extinction ratios for both output ports (i.e., exhibiting 2x2 switching capabilities) over a sufficiently large wavelength range is the dual-resonator add-drop filter with perpendicular access waveguides [1]. Two variants of the aforementioned component, i.e., with an untreated or a treated waveguide crossing, are depicted in Fig. 3.1. The presence of two resonators renders the structure symmetric, meaning that one can find two equivalent input ports even when the coupling conditions between resonator and input or output waveguide are different (i.e., g 1 g 2 or L 0). Needless to say that for this to hold the input waveguide-resonator and output waveguideresonator coupling conditions should be identical for both resonators (see insets of Fig. 3.1). In addition, the presence of the second resonator introduces an extra route to the drop port. We name this path Q-route due to the shape of its trace (see insets of Fig. 3.1). This means that in both cases (untreated or treated crossing), the drop port transmission is a result of interfering waves. Figure 3.1: Schematic of a DLSPP-based add-drop filter with perpendicular access waveguides and an (a) untreated or (b) treated by means of elliptical tapering (L X = 6 um, W X = 1.6 um) waveguide crossing. Arrows indicate the route of the control current in the event of heating. The insets depict the main geometrical parameters and the interfering waves shaping the drop port transmission. Furthermore, in the untreated-crossing case cross talk (XT), i.e., power coupled to the perpendicular waveguide while traversing the crossing, contributes to interference as well. Specifically, for an operating wavelength of 1.55 um, cross talk is ~ 15 db. This is actually not negligible. If an input wave of unit amplitude impinges on the crossing, the wave traveling in the perpendicular waveguide (either direction) has an amplitude of approximately 0.2. By observing the dominant electric field component (E z ) distribution in such a waveguide crossing (Fig. 3.2(a)), one can verify that some light does indeed couple to the perpendicular waveguide while some is lost via scattering. The insertion loss (IL) due to both cross talk and scattering is 0.5 db. November 7, 2011 FP7-249135 The PLATON Consortium Page 6 of 16

Figure 3.2: Distribution of the dominant electric field component (E z ) in an (a) untreated and (b) treated by means of elliptical tapering (L X = 6 um, W X = 1.6 um) DLSPP waveguide crossing. The real part of the E z component is plotted at an xy-plane located 10 nm above the metal surface. On the other hand, treatment of a waveguide crossing suppresses cross talk, since the level of cross talk is proportional to the level of mode confinement. This can be explained by the following reasoning: During the extent of the crossing the mode tends to spread out laterally due to the momentary absence of waveguide walls. The tighter the field confinement in the waveguide, the stronger the diffraction effects in the intersection, making the spreading more pronounced. Clearly, wider spreading results in easier coupling to the perpendicular waveguide, i.e., higher cross talk. Thus, cross talk can be suppressed by relaxing the mode confinement just before it arrives at the waveguide intersection. This can be done by treating the waveguide crossing, i.e., providing some kind of tapering which would be responsible for expanding the guided mode. Both linear and elliptical tapering shapes were considered having several values of tapering widths (W X ) and lengths (L X ). The linear tapering scheme consists of two identical rhombi perpendicular to each other, centered at the waveguide intersection. The short diagonal is equal to W X and the large diagonal equal to L X. In the same way, the elliptical tapering scheme consists of two identical ellipses centered at the waveguide intersection. This time, the minor axis is equal to W X and the large axis equal to L X. A parametric study was performed in order to determine the optimum tapering dimensions for minimizing crosstalk. We found that the tapering length of 6 um is a favorable compromise between smooth mode expansion and compactness. We therefore fix L X to 6 um and vary the tapering width W X. In general, as W X increases cross talk is suppressed for both tapering shapes. However, elliptical tapering was found to provide the best balance between low XT and small ILs. Given the above, for the purpose of minimizing cross talk, we conclude to an elliptic tapering shape with L X = 6 um and W X = 1.6 um. For an operating wavelength of 1.55 um these dimensions result in a cross talk of ~ 37dB and an insertion loss of ~ 0.8 db, only slightly larger than that of the untreated crossing (0.5 db). The distribution of the dominant electric field component in this case is depicted in Fig. 3.2(b). As can be seen, cross talk is greatly suppressed. On the other hand, scattering losses have increased. Untreated waveguide crossing variant Having examined the behavior of the waveguide crossing we turn to the performance of the entire switch. Let us first focus on the untreated-crossing variant (Fig. 3.1(a)). The drop port transmission in this case is the result of three interfering waves: the one via the resonator, the one coming from the crossing, and the one following the Q-route (inset of Fig. 3.1(a)). It is exactly these interference effects that can be advantageously exploited leading to high ERs for the drop port. As an example, we examine a filter of this type with the following geometrical parameters: R = 5.5 um, L = 0, g 1 = 0.3 um, and g 2 = 0.5 um. All parameter values are carefully selected so as to ensure high performance in terms of ER. Specifically, the 5.5 um radius value is a favorable compromise between November 7, 2011 FP7-249135 The PLATON Consortium Page 7 of 16

competing resistive and radiation losses. It leads to minimum overall losses, and consequently maximum quality factors, for the resonator resonances. This has been verified by eigenvalue simulations of the uncoupled resonator. The g 1 gap is set to 0.3 um since smaller values can be challenging from a fabrication standpoint. We do not employ a racetrack resonator since it is not the through port ER that limits the filter performance. The g 2 gap is rather large. If g 2 is too small the amplitude of the wave via the ring is substantially higher than the other two, meaning that no cancellation of the interfering waves is to be expected. The transmission for both ports and states is depicted in Fig. 3.3(a). The irregular shape of the drop port transmission curve demonstrates the presence of interference effects. Due to this complex curve shape, the operating wavelengths leading to high ERs simultaneously for both ports are hard to identify. To this end, we plot in Fig. 3.3(b) the ER for both ports in the entire wavelength range. We are interested in continuous wavelength regions that can provide high extinction ratios for both ports simultaneously. As can be seen, an ER better than 5 db can be provided by two wavelength regions with an aggregate size of 10 nm. This wavelength span can accommodate twelve WDM (wavelength-division multiplexing) channels with 100 GHz (0.8 nm) spacing. Alternatively, if we define the ER threshold at 8 db we find a single wavelength region of 3.2 nm (four 100-GHz spaced WDM channels) that can provide it. The optimum performance is achieved at the operating wavelength of 1.537 um, where the ERs are 10.8 and 9 db for the through and drop ports, respectively. Figure 3.3: (a) Transmission vs wavelength for both output ports and both heated and unheated states. The geometrical parameters of the dual-resonator add-drop filter with perpendicular access waveguides and an untreated crossing are: R = 5.5 um, L = 0, g 1 = 0.3 um, and g 2 = 0.5 um. (b) Extinction ratio vs wavelength for both ports. Treated waveguide crossing variant Filters with an untreated waveguide crossing base their performance on the cross talk level of the waveguide crossing. If for some reason the level of cross talk in an actually implemented filter is somewhat different from that predicted by simulations, e.g., because the 90-degree corners of the polymer ridge are not well defined, then the designs will not provide their nominal extinction ratio. It is therefore expedient to examine the performance of a dual-resonator add-drop filter with a treated crossing. The cross talk in this case is so small (~37 db) that any fabrication inaccuracies will not change the fact that its contribution to the interference is negligible. This time, the drop port transmission is a result of two interfering waves: the one via the resonator and the one following the Q-route. As the wave following the Q-route suffers extra propagation losses, the interference effects are bound to be relatively weak. However, we are interested in seeing whether this variant can still provide wavelength regions with high ERs for both ports. We focus on a filter with the following geometrical parameters: R = 5.7 um, L = 0, g 1 = 0.3 um, and g 2 = 0.48 um. The transmission for both ports and states is depicted in Fig. 3.4(a). The corresponding extinction ratios can be found in Fig. 3.4(b). As can be seen, extinction ratios higher than 5 db can be provided by a wavelength range of 5 nm, November 7, 2011 FP7-249135 The PLATON Consortium Page 8 of 16

accommodating six 100-GHz-spaced WDM channels. To summarize, dual-resonator filters with perpendicular access waveguides and an untreated crossing can provide superior performance in terms of ER compared to treated-crossing ones, due to the stronger interference effects associated with the drop port transmission. However, filters of the second category can still provide high ERs for both ports over fairly broad wavelength regions. In addition, such filters do not rely on the cross talk level of the waveguide crossing, therefore resulting in sounder designs. Figure 3.4: (a) Transmission vs wavelength for both output ports and both heated and unheated states. The geometrical parameters of the dual-resonator add-drop filter with perpendicular access waveguides and a treated crossing (elliptic tapering with L X = 6 um, W X = 1.6 um) are: R = 5.7 um, L = 0, g 1 = 0.3 um, and g 2 = 0.48 um. (b) Extinction ratio vs wavelength for both ports. 3.2 Cyclomer-loaded switches Following the design of PMMA-loaded WRR 2x2 switches, another design round based on Cyclomer was conducted. Cyclomer possesses a slightly higher refractive index compared to PMMA. Specifically the refractive index value at room temperature is approximately 1.52. More importantly, the thermo-optic coefficient (TOC) is almost three times larger compared to PMMA: -2.9 10-4 K -1 instead of -1.05 10-4 K -1. As far as resonator-based components are concerned this means that the thermo-optic shift is enhanced. In fact, a temperature difference of 100 K permits a shift equal to 25 nm, i.e. half the free spectral range (FSR) of a 5-um ring resonator (Fig. 3.5). Thus, the entire transmission variation is captured, leading to higher extinction ratios (Fig. 3.5). In other words, the actual extinction ratio is the maximum possible. This means, that the performance of resonatorbased components with Cyclomer as the loading material is boosted. Designs such as the classic add-drop filter with parallel access waveguides that exhibited poor performance when implemented with PMMA [1] can now offer ERs higher than 5 db over broad wavelength ranges (Fig. 3.6). Figure 3.5: Schematic and transmission of a Cyclomer-loaded all-pass microring resonator filter. The geometrical parameters are R = 5.7 um, L = 0.7 um, g 1 = 0.3 um. The thermo-optic shift is 25 nm and corresponds to a temperature difference of 100 K. November 7, 2011 FP7-249135 The PLATON Consortium Page 9 of 16

Figure 3.6: (a) Transmission vs wavelength for both output ports and both heated and unheated states. The geometrical parameters of the dual-resonator add-drop filter with perpendicular access waveguides and an untreated crossing are: R = 5.2 um, L = 0.5 um, g 1 = 0.3 um, and g 2 = 0.6 um. (b) Extinction ratio vs wavelength for both ports. However, the best performance is provided again by dual-resonator filters with perpendicular access waveguides. The large thermo-optic shift can in general provide higher extinction ratios over larger bandwidths, compared to PMMA-based designs. Note, however, that due to the irregular shape of the drop port transmission it is not always advantageous to utilize the entire temperature increase sustained by the polymer, i.e., the maximum shift does not necessarily lead to optimum performance. For example, for a filter with the following geometrical parameters: R = 5.2 um, L = 0.5 um, g 1 = 0.3 um, and g 2 = 0.5 um, a temperature difference of 100 K (25 nm thermo-optic shift) provides an extinction ratio of 8 db over a broad wavelength range (Fig. 3.7). On the other hand, a smaller temperature increase of 55 K (13 nm thermo-optic shift) provides a higher extinction ratio of 10 db, albeit over a shorter bandwidth (Fig. 3.8). Figure 3.7: (a) Transmission vs wavelength for both output ports and both heated and unheated states. The temperature difference between the two states is 100 K. The geometrical parameters of the dual-resonator add-drop filter with perpendicular access waveguides and an untreated crossing are: R = 5.2 um, L = 0.5 um, g 1 = 0.3 um, and g 2 = 0.5 um. (b) Extinction ratio vs wavelength for both ports. Figure 3.8: (a) Transmission vs wavelength for both output ports and both heated and unheated states. The temperature difference between the two states is 55 K. The geometrical parameters of the dual-resonator add-drop filter with perpendicular access waveguides and an untreated crossing are: R = 5.2 um, L = 0.5 um, g 1 = 0.3 um, and g 2 = 0.5 um. (b) Extinction ratio vs wavelength for both ports. November 7, 2011 FP7-249135 The PLATON Consortium Page 10 of 16

4 Experimental evaluation of PMMA- and Cyclomer- loaded SOI-DLSPP WRRR switches 4.1 Evaluation of an all-pass PMMA loaded WRRR The switching performance of PMMA loaded SOI DLSPP WRR switches was evaluated experimentally with an all-pass ring that was incorporated in the PLA08-F02 chip according to the 1 st PLATON mask. Fig. 4.1(a) illustrates a SEM image of this WRR. The all passs ring resonator had 5.5um radius, 0.35um gap and 0.8um interaction length, followed by 20um straight DLSPP waveguides. The hosting areas of the plasmonic structures on each motherboard have been etched forming 200nm-deep cavities in the Silicon Dioxide substrate. The cavities were covered with 65nm-thick and 3um wide gold film, on top of which a PMMA strip with a cross-section of 600x500nm 2 was placed, forming Dielectric Loaded Surface Plasmon Polariton (DLSPP) waveguide configuration on silicon dioxide substrate. Since Surface Plasmon Polaritons (SPP) support only TM propagating mode, the SOI motherboard was equipped with TM grating couplers as interfaces for incoming/outgoing signal and 340x400nm 2 silicon rib waveguides with 50nm-thick slab that were able to support also TM light propagation. The spectral responses of the WRR were received with a continuous wave (CW) signal that was generated by a tunable laser with a tuning range from 1500 to 1580nm with 10pm step resolution. The injected light was adjusted for TM-polarization ensured by a polarization controller beforee entry into the hybrid Silicon Plasmonic ring. Fig. 4.1(b) illustrates the respective experimental results. The blue line represents the spectral response of the device for 0mA injected current (COOL state) while the red line corresponds to the spectral response that is obtained for 50mA injected current (HOT state). The simultaneous shifting of the two resonances suggests thermo-optic tuning of the device by 7nm towards smaller wavelengths, owing to the negative sign of the thermo-optic coefficient of PMMA. This wavelength shift yields an extinction ratio of 8dB at 1561nm between the HOT and COOL states of the device, corresponding to an induced temperature change of 61K and consuming only 3.3mW of electrical power. Returning, however, to the initial state of the ring resonator, after switching off current, was not achieved, implying that the reversibility of the device was lost due to heating it above the maximum servicee temperature of the PMMA strip. The Q-value of the resonator was measured around 98 at COOL state and 142 at HOT state for the 1565nm and 1561nm resonant wavelengths, respectively, while its Free Spectral Range (FSR) value was 41.7nm when unheated. More details regarding the testing of the PMMA-loaded SOI-DLSPP WRR structure are presented in D2.4. Figure 4.1: (a) SEM image of the Si-Plasmonic WRR switch, (b) Spectral Response of the All- Pass Ring Resonator for 0mA and 50mA current. November 7, 2011 FP7-249135 The PLATON Consortium Page 11 of 16

4.2 Fabrication of the Cyclomer-loaded dual-resonator 2x2 switches As anticipated in previous studies reported in former deliverables, the utilization of DLSPP modules in chip-scale router circuits requires a loading material featuring ~ 3 times higher TOC than PMMA. Cyclomer was identified as a suitable candidate on the basis of its TOC measurement in a thin film configuration [D2.3]. In the present report, Cyclomer is tested as the loading material of a thermo-optic dual-resonator DLSPP layout in order to assess if the thermo-optic wavelength tuning properties is sufficient for systemqualified switching performance. Figure 4.2(a) shows a Scanning Electron Microscope (SEM) image of the DLSSP switch fabricated at UB, illustrating the dual-resonator layout that comprises two perpendicularly intersecting DLSPP waveguides along with two diagonally positioned DLSPP racetrack resonators, having R=5μm radius and two straight waveguide region lengths of L1=2μm and L2=0μm respectively. Figure 4.2: (a) SEM image of the dual-resonator Cyclomer-loaded DLSPP switch, (b) Transmission Spectra for Through output port in Cool and Hot state, (c) Transmission Spectra for Drop output port in Cool and Hot state. The image s inset reveals the adequate quality of the DLSPP structures with the Cyclomer-loaded waveguides exhibiting a nearly square 600x500nm2 cross-section, allowing for a gap resolution of 300nm between the intersecting waveguides and the racetrack resonators. The switch was fabricated onto a 3mm wide gold electrode, which was connected to a current source capable of delivering a maximum current of 400mA. Increasing the current from 0mA to its maximum 400mA value the plasmonic waveguide reaches a temperature of 90 C measured with a thermocouple. 4.3 Experimental results The transmission spectra of the Cyclomer-loaded dual-resonator switch for both the Through and Drop output ports are shown in Figure 4.2(b) and (c), respectively, when plasmons are excited at In#1 port according to Fig. 4.2(a). These spectra were recorded using Radiation Leakage Microscopy (RLM) at room temperature ( COOL ) and when the switch was heated at 100 C ( HOT state). A clear resonant behavior can be observed with the Free Spectral Range (FSR) of the Through-port resonances being equal to 38nm. When operating in unheated conditions, the resonant dips at the Through-port have an Extinction Ratio (ER) of more than 35dB, while the corresponding ER value for the Dropport resonant peak is close to 10dB in the 1560-1580nm spectral region. Insertion losses (IL) for the Through and Drop ports are -10dB and -8dB, respectively. The quality factor of the switch was calculated ~394 for the Through and ~82 for the Drop port at 1568nm and 1572nm resonance wavelengths respectively. The propagation loss factor of the DLSPP waveguide was measured also to be 0.1dB/μm. By raising the temperature of the November 7, 2011 FP7-249135 The PLATON Consortium Page 12 of 16

Cyclomer strip to 100 C, a wavelength shift of 9nm is observed at both Through and Drop-port resonances as a result of the thermo-optically induced phase shift experienced during propagation in the dual-resonator setup. Figure 4.3(a) presents the ER values obtained between COOL and HOT switching states at both output ports and over the complete 1500-1600nm spectral window, revealing clearly high-quality single channel switching with 23dB and 10dB ER at Through and Drop-outputs, respectively, at 1558nm. Extinction ratio values higher than 6dB are obtained over a 6nm spectral range around 1558nm for both output ports, confirming the feasibility of simultaneous switching eight 100 GHz-spaced channels. The 6dB over 6nm range ER values are similar to the performance of silicon-based switches employed in WDM on-chip routing platforms [2], however the DLSPP layout requires a significantly lower footprint while providing this ER over multiple 100GHz-spaced channels. Figure 4.3(b)-(c) illustrate the respective RLM images obtained at 1558nm, showing that the signal is exiting the circuit through its Through-port when it is operated in COOL state, while the optical power emerges at the Drop-port when turning to HOT state. It should be noted that when the same level of temperature change was applied to a PMMA-loaded dual-resonator switch (see D2.3), a small wavelength shift of only 4nm could be obtained yielding to poor ER values, never exceeding 4dB. Figure 4.3: (a) Extinction Ratio values between Cool and Hot switching states at Through and Drop outputs, (b) RLM image in Cool state at λ = 1558nm, (c) RLM image in Hot state at λ = 1558nm November 7, 2011 FP7-249135 The PLATON Consortium Page 13 of 16

5 Conclusion This deliverable presents numerical and experimental results for 2x2 WRR switching structures, both with PMMA and Cyclomer loading. The simulations have been carried out by means of the 3D-FEM method and have concluded to the optimal designs for WRRbased 2x2 switching operation. A dual-resonator 2x2 switch configuration using PMMA loading has been found to provide 5dB ER over a 5nm wavelength region for an electrically induced temperature shift of 80K, while the same layout with Cyclomer loading can yield up to 8dB ER over a broad wavelength range of more than 6nm. The first experimental results from a PMMA-loaded SPP-on-SOI WRR structure are also reported. An all pass racetrack structure with R= 5.5um radius has been evaluated with respect to its thermooptic response and was found to provide a wavelength shift of 7nm for 50mA of applied current, corresponding to power requirements of less than 3.5mWs and to an induced temperature deviation of apprx. 60K. A Cyclomer-loaded dual-resonator switch configuration has been also fabricated and the first thermo-optic response characterization results are presented in this deliverable. This configuration provides 6dB ER over a spectral window of 6nm, confirming the improved switching credentials of the Cyclomer-loaded compared to the PMMA-loaded SPP platform. The next development steps incorporate the fabrication of the cross shaped dualresonator layout with the proper configuration; the two racetrack resonators will be in perpendicular directions ensuring that signals propagate through the same path until reaching the Add-Drop output ports. The symmetry in the new layout is expected to provide enhanced switching performance in terms of ER over the required 6nm wavelength region. Towards this task, new UV masks have been ordered by UB with the first experimental results expected in the near future. November 7, 2011 FP7-249135 The PLATON Consortium Page 14 of 16

Abbreviations WRR Waveguide Ring Resonators TOC 3D-FEM R L g ER XT IL W x L x FSR Thermo Optic Coefficient Three-Dimensional Vector Finite Element Method WRR radius Length of racetrack section Gap between WRR and straight waveguide Extinction Ratio Cross Talk Insertion Loss Tapering Width Tapering Length Free Spectral Range DLSPP SPP Dielectric-Loaded Surface Plasmon Polariton Surface Plasmon Polaritons DLSPPW CW SEM RLM Dielectric-Loaded Surface Plasmon Polariton Waveguide Continuous Wave Scanning Electron Microscope Radiation Leakage Microscopy November 7, 2011 FP7-249135 The PLATON Consortium Page 15 of 16

References [1] O. Tsilipakos, E. E. Kriezis, and S. I. Bozhevolnyi, "Thermo-optic microring resonator switching elements made of dielectric-loaded plasmonic waveguides," J. Appl. Phys., 109(7), 073111, (2011). [2] B.G. Lee, A. Biberman, J. Chan, and K. Bergman, High-Performance Modulators and Switches for Silicon Photonic Networks-on-Chip, IEEE J. of Sel. Topics in Quantum Electron., Vol. 16, No. 1, pp. 6-22, Jan./Feb. 2010 [3] N. Pleros, E.E. Kriezis, K. Vyrsokinos, "Optical Interconnects using plasmonics and Si-photonics", Invited review article, IEEE Photonics Journal, Special issue on Breakthroughs in Photonics 2010, Vol. 3, No. 2, pp. 296-301, April 2011 [4] D. Kalavrouziotis, G. Giannoulis, D. Apostolopoulos, S. Papaioannou, A. Kumar, S.I. Bozhevolnyi, L. Markey, K. Hassan, J.-C. Weeber, A. Dereux, M. Baus, M. Karl, T. Tekin, O. Tsilipakos, A. Pitilakis, E.E. Kriezis, H. Avramopoulos, K. Vyrsokinos and N. Pleros, "10 Gb/s Transmission and Thermo-Optic Resonance Tuning in Silicon-Plasmonic Waveguide Platform", in Proceedings of European Conference on Optical Communication (ECOC 2011), We.10.P1.27, Geneva, Switzerland, 18-22 September, 2011. November 7, 2011 FP7-249135 The PLATON Consortium Page 16 of 16