Nanoscale Optical Dielectric Rod Antenna for On-Chip ... - IEEE Xplore

2624

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 59, NO. 10, OCTOBER 2011

Nanoscale Optical Dielectric Rod Antenna for On-Chip Interconnecting Networks Hongyu Zhou, Student Member, IEEE, Xi Chen, Student Member, IEEE, David S. Espinoza, Student Member, IEEE, Alan Mickelson, Senior Member, IEEE, and Dejan S. Filipovic, Senior Member, IEEE

Abstract—A nanoscale on-chip optical dielectric rod antenna is demonstrated in this paper. The antenna is designed and fabricated on a 200-mm silicon-on-insulator platform, using IMEC 193-nm-deep UV lithography. A 500-nm-thick polymer layer is designed and deposited to act as an asymmetric slab waveguide, confining the radiated wave within the layer. Full-wave analysis predicts antenna return loss above 25 dB, and end-fire gain greater than 9 dBi from 172 to 222 THz. Six antenna pairs with 1-, 3-, 5-, 7-, 12-, and 17- m separations are fabricated. Corresponding transmissions are measured from 190 to 200 THz. Two on-chip optical signal hubs composed of 16 and 32 antennas designed for core-to-core interconnection for the next-generation multicore microprocessors are also demonstrated. Good agreement between the modeling and measurement is obtained. Index Terms—Dielectric rod antenna, nanoantennas, nanointerconnect, on-chip antenna, optical antenna, silicon-on-insulator (SOI) wafer, silicon photonics.

I. INTRODUCTION

T

HE DENSITY of silicon complementary metal–oxide semiconducting (CMOS) transistors continues to increase at a fast pace. Heat dissipation has dictated that the electronic fabric within silicon microprocessors be segmented into cores where computation occurs and interconnection networks. However, the power consumption of the traditional electrical interconnect fabric may well exceed the projected on-chip communication power budget for future technology generations [1]. A new interconnection network that allows integrations of tens or even hundreds of cores becomes critical for further development of multicore microprocessors. The recent developments in silicon photonics enable a promising alternative, specifically a nanophotonic interconnect [2]. Using on-chip optical waveguides, rings, couplers, etc., a variety of interconnect components have already been developed. Substantial improvements over electrical interconnects in throughput, latency, and power efficiency have been demonstrated by simulations based on nanophotonic device research. However, even for nanophotonic interconnect, the Manuscript received January 13, 2011; revised April 19, 2011; accepted April 25, 2011. Date of publication June 09, 2011; date of current version October 12, 2011. This work was supported by the National Science Foundation (NSF) under Grant CCF-0829950 “EMT/NANO: Broadcast Optical Interconnects for Global Communication in Many-Core Chip-Multiprocessor.” The authors are with the Department of Electrical, Computer, and Energy Engineering, University of Colorado at Boulder, Boulder, CO 80309-0425 USA (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TMTT.2011.2156423

latency-critical coordination messages are not handled well [3]. Coordination messages that require broadcast or multicast employ multiple circuits and retransmissions for each coordination transaction, which increases not only signal latency, but also power consumption. To address this issue, a novel optical interconnect approach based on on-chip optical antennas is proposed in [3] and [4]. Dramatic improvement is predicted in cache miss latency and network power consumption compared to both electric and conventional point-to-point nanophotonic solutions. However, technology compatible and efficient optical antennas for on-chip signal transmissions have yet to be demonstrated. Even though microwave monolithic integrated antennas for intra-chip interconnects have been successfully developed [5]–[8], reported terahertz and optical on-chip antennas suffer from material loss and require unconventional feed approaches [9]–[12] to operate as effective on-chip transmitters or receivers. In this paper, 172- (1750 nm) to 222-THz (1350 nm) optical dielectric rod antennas for on-chip optical interconnection are demonstrated. Antennas are fabricated on a 200-mm silicon-on-insulator (SOI) platform based on IMEC 193-nmdeep UV lithography [13]. A 500-nm-thick polymer layer is deposited on the top of the wafer during post processing to function as an asymmetric slab waveguide, thus confining the antenna radiated field to the layer. The antenna is fed directly from an optical waveguide [14] that receives the input signal from a single-mode fiber (SMF) through a predefined grating coupler on the wafer [15], [16]. Within its bandwidth, the antenna achieves return loss and endfire gain greater than 25 dB and 9 dBi, respectively. For the measurements, the antennas are arranged in pairs facing each other with different separations. The transmission measurements are performed from 190 to 200 THz and are limited by the bandwidth of the spectrum analyzer and grating couplers. Good agreements with full wave modeling are obtained. Following that, two optical signal hubs with 16 and 32 optical dielectric rod antennas are demonstrated. The hubs operate as data distribution centers, broadcasting signals from an arbitrary antenna to all the others with minimum transmission imbalance, thus allowing low latency wireless links between wavelength-division multiplexing (WDM) channels. To construct the hub, antennas are compactly and symmetrically arranged in a circular ring with carefully designed separations. The proposed optical signal hubs operate over a 172–222-THz bandwidth, which fully satisfies the WDM requirements [3]. The test setup developed in the 190–200-THz band shows good agreement with numerical modeling. The fabricated 32-channel hub achieves 22.4-dB transmission minimum. This requires a

0018-9480/$26.00 © 2011 IEEE

ZHOU et al.: NANOSCALE OPTICAL DIELECTRIC ROD ANTENNA FOR ON-CHIP INTERCONNECTING NETWORKS

2625

Fig. 2. FIT and FEM simulated antenna reflection coefficient and end-fire realized gain. To reduce running time and yet have enough data points for comparison, only seven discrete frequencies are computed with the FEM.

Fig. 1. Proposed optical dielectric rod antenna configuration. (top) Top and side views. (bottom) 3-D model (L1 = 4 m, L2 = 0:5 m, L3 = 3 m, W 1 = 0:45 m, W 2 = 0:75 m, W 3 = 0:13 m, T 1 = 0:22 m, T 2 = 2 m, T 3 = 0:5 m).

minimum input power of 37 mW to establish the communication between the cores [4], a number significantly more efficient than the electrical interconnect alternatives. This paper is organized as follows.Section II outlines the proposed optical dielectric rod antenna geometry. Two different full wave methods are employed to analyze and computationally validate antenna performance. Section III discusses the measurement setup and the obtained results. Section IV provides the layout and performance of two optical signal hubs. Discussion pertained to the function of the polymer coating is given in Section V. II. OPTICAL ANTENNA GEOMETRY AND MODELING A. Antenna Geometry The configuration of the proposed optical dielectric rod antenna is shown in Fig. 1. The antenna is designed on a 200-mm SOI wafer following IMEC 193-nm-deep UV lithography guidelines [13]. The SOI wafer features three layers, m top silicon layer , a a m thick buried oxide (BOX) layer , and a 700- m bulk silicon substrate . Note that all the designed structures are developed on the top silicon m by layer. The antenna is extruded from an m by m launcher and exponentially tapered down to a m tip over a m

m by m optical length. A standard m axial waveguide [14] is linearly tapered up over an nm polymer length and connected to the launcher. A is post developed on the top of the SOI wafer layer to confine the radiation, as depicted by the gray area in Fig. 1 (side view). For the purpose of modeling, the SOI wafer is modeled as a 12 m by 16 m rectangular cell with the antenna placed in the middle. The excitation is provided by a waveguide fed from the rectangular port at the cell boundary, as shown in Fig. 1 (3-D modeling). The bottom bulk silicon substrate is modeled as a 0.5- m-thick silicon layer, face terminated in a radiation boundary. This reduces the computational complexity while ensuring proper substrate functionality. Compared to a linearly tapered antenna, the exponentially tapered rod has smaller gain roll off at both band ends and 2–3 dB higher endfire gain. Reduced ripples in the antenna -plane radiation patterns are also observed. It is important to note that the antenna length and width affect the antenna gain and radiation patterns. The chosen values give a decent gain with 0-dBi -plane beamwidth above 100 and minimal pattern ripples, as shown in our previous study [17]. B. Antenna Full-Wave Analysis The time-domain finite integration technique (FIT) code CST [18] and frequency-domain finite-element method (FEM) code HFSS [19] are used in the numerical modeling and design. Excellent agreement between the two validates the theoretical performance of the antenna. From 172 to 222 THz, the designed antenna achieves return loss and endfire gain greater than 25 dB and 9 dBi, respectively, as shown in Fig. 2. Note that the singlewaveguide mode is from mode bandwidth of the excited 172 to 222 THz [14], thus directly determining the antenna operating range. The simulated radiation patterns at 175, 185, 205, and 220 THz are shown in Fig. 3. As seen, good consistency throughout the bandwidth is observed. Note that the FIT and FEM results are very similar, and only FIT results are shown for clarity. As seen, the co-polarized patterns in the -plane ( – plane) show good stability, while those in the -plane ( –

2626


Fig. 4. SEM image of the fabricated antenna.

TABLE I ANTENNA GEOMETRY VARIATIONS DUE TO FABRICATION

Fig. 3. FIT simulated antenna radiation patterns at: (a) 175, (b) 185, (c) 205, and (d) 220 THz.

plane) broaden some as frequency is increased. The cross-polarization in the -plane cannot be seen on the presented scale, while in the -plane, it increases slightly with frequency and reaches 25 dB around 220 THz. III. ANTENNA MEASUREMENTS A. Fabrication and Measurement Setup The designed antennas are fabricated on a 200-mm-diameter SOI wafer with an IMEC 193-nm-deep UV lithography process [13]. A 500-nm polymer coating is deposited after the wafer is fabricated. The polymer used is a photo-definable polyimide (HD-8820) from MicroSystems [20]. An scanning electron microscope (SEM) image of the fabricated antenna is shown in Fig. 4. Due to process tolerance, the structural parameters of the fabricated devices are different from the design values, both shown in Table I. The fabricated length and width are 8.7% and 10.7% different than designed, respectively. The antenna tip is rounded and about 30.8% narrower than the designed value. Note that the minimum allowed linewidth in the process is 120 nm, about 10% lower than our designed value. The width of the standard 220 nm 450 nm optical waveguide is also narrowed to 380 nm. Six antenna pairs with 1-, 3-, 5-, 7-, 12-, and 17- m separations are fabricated. Each pair has the antennas facing each other so that on wafer transmission can be inferred. An SEM image of the antenna pair with 1- m separation is shown in Fig. 5. All

Fig. 5. SEM image of an antenna pair with 1-m separation.

antennas are directly connected to the 220 nm 450 nm optical waveguides, which are connected to the predefined grating couplers on the two sides of the chip [15]. The grating coupler acts as an interface between the optical waveguide and the SMF, coupling in/out light signals from the SMF. The SMF is connected to the spectrum analyzer for measurements. Note that both the spectrum analyzer and the grating coupler have narrower bandwidth than the proposed antenna, thus the measurements are limited from 190 to 200 THz.


Fig. 6. Photograph of the measurement setup: sample is placed on the L-shape holder (black piece) at the bottom; the SMF fiber (yellow wires (in online version) in the image) movements are controlled by 3-D micro-positioning system; visible camera [blue cube (in online version)] together with an infinity-corrected lens is used to monitor fiber movement above sample surface.

Fig. 7. Measured and simulated transmissions for six antenna pairs (uncoated) with different separations at: (top) 190 THz, (middle) 193 THz, and (bottom) 200 THz.

The calibration is performed using a bare waveguide directly connecting two grating couplers on each side of the chip. The reference transmission level is obtained from this device. To measure the true antenna transmission, (by calibrating out the losses from the grating couplers [16] and the optical waveguides [14]), the reference transmission level is subtracted from the antenna measurements. A super-luminescent light emitting diode (EXALOS 10-dB bandwidth of 120-nm EXS1510-2111) that has wavelength is employed as the broadband light source input. An optical spectrum analyzer, HP 71450A, is used for char-

2627

Fig. 8. Measured transmissions from 190 to 200 THz of the six antenna pairs with different separations after polymer coating.

Fig. 9. Measured and simulated transmissions for six antenna pairs (coated) with different separations at: (top) 190 THz, (middle) 193 THz, and (bottom) 200 THz.

acterizing the transmission spectrum of these silicon photonic devices. The submicrometer step size and alignments of the input and output SMF are precisely controlled by two 3-D micropositioning controllers (Newport ESP301 and motors), as shown in Fig. 6. B. Measurements The transmissions of the six antenna pairs without the 500-nm polymer coating are measured first. Fig. 7 shows

2628


TABLE II GEOMETRY VARIATIONS FOR 16-CHANNEL SIGNAL HUB. GEOMETRY VARIATIONS FOR 32-CHANNEL SIGNAL HUB.

increases the transmission between the antenna pairs roughly 20 dB compared to the noncoated case. IV. OPTICAL ANTENNA BASED ON-CHIP SIGNAL HUBS A. Configuration and Mechanism

Fig. 10. SEM image of the fabricated signal hubs with: (top) 16 antennas and (bottom) 32 antennas.

the obtained transmission results at 190, 193, and 200 THz. Both the designed and fabricated antenna configurations are modeled. As seen, excellent agreement between measurements and simulations is obtained. The fabrication variations on the antenna structural parameters lower the transmissions about 3–7 dB for 7- and 12- m separations, while for 1- m separation, the transmission is better for the fabricated model. Note that the antenna is designed to give the best performance with the 500-nm polymer coating. The measured transmissions of the polymer coated antenna pairs are shown in Fig. 8. The comparisons between the simulations and measurements are shown in Fig. 9, for 190, 193, and 200 THz. The actual (fabricated) antenna distance is 0.5 m larger than designed, which lowers the antenna transmission about 1.5 dB, as predicted by the corresponding simulations in Fig. 9. Due to the polymer coating, the light wave radiated by the antenna is confined within the polymer layer and transmitted as a 2-D guided wave, which, in turn, offers better signal transmission compared to the noncoated case. As seen in Fig. 9, when the antenna separation is larger than 7 m, the polymer coating

To study the applicability of the proposed optical dielectric rod antenna for system level interconnections [3], optical antenna based signal hubs acting as data distribution centers are investigated. As a multicore broadcast interconnect fabric, the hub should be highly efficient, with good output balance and channel symmetry. The antennas constituting the hub should be identical and bidirectional in that they should operate in both transmitting or receiving modes. Given these constraints, multiple antennas are specifically reshaped and arranged. Two signal hubs composed of 16 and 32 antennas are designed, fabricated and measured. To form the hub, the antennas are compactly arranged in a circular ring, as shown in the SEM images in Fig. 10. The minimum/maximum antenna distance is designed to be m m and m m for 16- and 32-channel hubs, respectively. The antenna geometry is designed to vary a little bit for each hub to give the best performance. For the 16-channel hub, W2 is shortened to be 0.45 m, which means the antenna is directly extruded from the standard optical waveguide rather than the launcher section. The antenna length L3 is shortened to be 1.8 m. For the 32-channel hub, W2 is also shortened to be 0.45 m, and L3 is equal to 1.5 m. For both hubs, the 500-nm-thick polymer layer is post developed to enhance the transmissions. The operation of the antenna-based signal hubs relies on both near field coupling and far-field radiation [4]. In that regard, the transmitting antenna will couple the output signal to the nearby antennas lying within its near-field region, and radiate the signal to the antennas outside this region. Good hub output balance can


Fig. 11. Measured and FIT simulated transmissions for 16-channel hub at: (top) 190 THz, (middle) 193 THz, and (bottom) 200 THz. Results are normalized to the transmission maximum.

Fig. 12. Measured and FIT simulated efficiency, transmission minimum, and transmission imbalance for 16-channel hub from 190 to 200 THz.

be achieved by carefully designing the antenna length L3 and hub radius D. High overall efficiency can be obtained by using the 500-nm-thick polymer coating, and shrinking the antenna minimum distance . Due to the fabrication process, the dimensions of the hubs also change, as shown in Table II. For the 16-channel hub, the critical parameters, waveguide/antenna width W1, minimum antenna distance , and maximum antenna distance or hub radius D vary more than 10%, which will inevitably impact the hub performance. For the 32-channel hub, the fabrication variations are smaller than 5.5%. It is expected that further process maturation will lead to the fabricated subsystems much closer to the designed models. B. Signal Hub Measurements To measure the transmission performance of the on-chip optical signal hub, an arbitrary antenna is chosen as the transmitter

2629

Fig. 13. Measured and FIT simulated transmissions for 32-channel hub at: (top) 190 THz, (middle) 193 THz, and (bottom) 200 THz. Results are normalized to the transmission maximum.

Fig. 14. Measured and FIT simulated efficiency, transmission minimum, and transmission imbalance for 32-channel hub from 190 to 200 THz.

and labeled as antenna 1. The receiving antennas are labeled as 2, 3, 4, etc., accordingly, as shown in Fig. 10. The measured and FIT simulated transmissions of the 16-channel hub at 190, 193, and 200 THz are shown Fig. 11. As seen, the hub transmission maximum and minimum occur at antenna 9 and antennas 4 and 14, respectively. Antenna 9 is right in the endfire direction of the transmitting antenna, where the maximum transmission occurs. Antennas 4 and 14 are located 68 away from the transmitting antenna endfire direction, where the gain is 0 dBi, as predicted in the antenna study. Within this region, the antenna near-field coupling starts to dominate, as the signal transmission is increasing from these two points. The hub transmission minimum, transmission imbalance (the difference between transmission maximum and minimum) and the overall transmission efficiency against frequency are shown in Fig. 12. In the design, the transmission minimum is maintained above 18 dB,

2630


Fig. 15. Simulated Poynting vectors of the on-chip optical antennas: (top) with and (bottom) without polymer coating.

and the imbalance is controlled to be around 5 dB. Due to the dimensional changes in the fabrication process, the measured transmission minimum is about 4 dB lower, and the imbalance is about 5 dB larger than designed. The overall efficiency is defined as the ratio of the total output power collected by all receiving antennas to the input power of the transmitting antenna. Within the measured frequency region, the fabricated hub has 40%–60% overall efficiency. Compared to the designed performance, the measured efficiency is up to 20% lower (195.5 THz). The measured and FIT simulated antenna transmissions of the 32-channel hub at 190, 193, and 200 THz are shown in Fig. 13. As seen, the transmission patterns from antenna 11 to antenna 23 show close resemblance to the -plane antenna radiation patterns, verifying the near-field coupling and far-field radiation mechanisms are separated around antennas 11 and 23. The transmission minimum, imbalance, and overall efficiency for the 32-channel hub are shown in Fig. 14. Since the structural variations in the fabrication are lesser, the transmission degradation is not as large as that of the 16-channel hub. The transmission minimum is only about 2 dB smaller than designed. The transmission imbalance is maintained 5–6 dB throughout the measured bandwidth. However, the measured overall transmission efficiency is up to 30% smaller than designed, indicating the transmission to each receiving antenna is lower for about 2 dB. The reason is that the bare waveguide used for the calibration is from a different chip fabricated in a different IMEC run. The lithography in the two runs may have produced small geometrical parameter differences on the same device such as the bare waveguide and the grating coupler. This, in turn, would cause the reference transmission level used in the calibration higher than the true level for the chip with the optical hubs, resulting in the measured transmissions about 2 dB lower than designed.

to the surface of the wafer. When the wafer is coated with a polymer layer with proper thickness and refractive index, the slab TE modes can be excited within the layer. For an asymmetric slab waveguide, the cutoff frequencies of modes can be derived from [21] the

V. DISCUSSION

VI. CONCLUSIONS

The transmission level comparison between antennas with and without polymer coating indicates the significance of applying an additional polymer layer in enhancing the on-chip antenna transmissions. For the dielectric rod antenna excited by mode of the feeding waveguide, the -plane is parallel the

Nanoscale on-chip dielectric rod antennas are demonstrated in this paper. The designed optical antennas have return loss above 25 dB, gain greater than 9 dBi, and consistent radiation patterns from 172 to 222 THz. Six optical antenna pairs are fabricated and on-chip antenna transmissions are measured from

where , , and are the reflective indices of the slab core, lower substrate, and upper cladding, respectively. is the mode is the normalized cutoff frequency, which is number, and given by

where is the corresponding cutoff wavenumber and is the slab thickness. For our case, the cutoff frequencies of the and slab modes are 99.4 and 312.3 THz, respectively, indicating within the antenna bandwidth region, the slab waveguide mode. operates in a single To better illustrate the power flows for the antennas with and without the polymer layer, the FIT simulated Poynting vectors in the antenna -plane are shown in Fig. 15. With the 500-nmthick polymer coating, most of the antenna radiated power is confined and nicely guided within the polymer layer. Only a small amount of power is radiated into the BOX layer. Without the polymer layer, however, majority of the radiation is directed into the BOX layer, bouncing back and forth there and eventually scattering into the silicon substrate. No power is confined on the surface of the wafer. The antenna transmission only depends on the back reflected wave from the silicon substrate in certain locations, making the transmission unstable with respect to the antenna separations, as shown in Fig. 7.


190 to 200 THz. The obtained results show good agreement with numerical modeling. Significantly improved transmissions are observed when a 500-nm-thick polymer layer is deposited on the wafer to confine the antenna radiation. Two on-chip optical hubs consist of 16 and 32 antennas are also designed, fabricated, and measured. 22-dB transmission minimum, 5-dB transmission imbalance, and over 40% overall efficiency are measured for the 32-channel hub. Measured results also indicate that the proposed optical hub fully satisfy the WDM requirement in the core-to-core interconnections for the future multicore microprocessors.

2631

[15] D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourhout, P. Bienstman, and R. Baets, “Grating couplers for coupling between optical fibers and nanophotonic waveguides,” Jpn. J. Appl. Phys., vol. 45(8A), pp. 6071–6077, 2006. [16] W. Bogaerts, S. K. Selvaraja, and P. Dumon, “Technology paper IMEC_193_01,” ePIXfab, Gent, Belgium, Sep. 2008. [17] H. Zhou and D. S. Filipovic, “Optical dielectric rod antenna for on-chip communications,” in Antennas Propag. Soc. Int. Symp., Toronto, ON, Canada, Jul. 11–17, 2010, pp. 1–4. [18] “CST Microwave Studio User Manual, Version 2009,” CST GmbH, Darmstadt, Germany, 2009. [19] “HFSS v12.0 User Manual,” Ansoft Corporation, Pittsburgh, PA, 2009. [20] Hd8820 Aqueous Positive Polyimide. Process Guide. Parlin, NJ: MicroSystems, 2008. [21] M. J. Adams, Introduction to Optical Waveguides. New York: Wiley, 1981.

ACKNOWLEDGMENT The authors would like to acknowledge Prof. M. Vachharajani, Prof. W. Park, Prof. L. Shang, Dr. B. Schwartz, Dr. Z. Li, M. Mohamed, E. Dudley, K. Gemp, and K. Zekis, all with the University of Colorado at Boulder, for useful discussions. REFERENCES [1] J. D. Owens, W. J. Dally, R. Ho, D. N. Jayasimha, S. W. Keckler, and P. Li-Shiuan, “Research challenges for on-chip interconnection networks,” IEEE Micro, vol. 27, no. 5, pp. 96–108, Sep.–Oct. 2007. [2] R. G. Beausoleil, P. J. Kuekes, G. S. Snider, W. Shih-Yuan, and R. S. Williams, “Nanoelectronic and nanophotonic interconnect,” Proc. IEEE, vol. 96, no. 2, pp. 230–247, Feb. 2008. [3] Z. Li, M. Mohamed, H. Zhou, L. Shang, A. R. Mickelson, D. S. Filipovic, M. Vachharajani, X. Chen, W. Park, and Y. Sun, “Global on-chip coordination at light speed,” IEEE Design Test Comput., vol. 27, no. 4, pp. 54–67, Jul./Aug. 2010. [4] H. Zhou, Z. Li, L. Shang, A. Mickelson, and D. S. Filipovic, “Onchip wireless optical broadcast interconnection network,” J. Lightw. Technol., vol. 28, no. 24, pp. 3569–3577, Dec. 2010. [5] K. K. O, K. Kim, B. A. Floyd, J. L. Mehta, H. Yoon, C.-M. Hung, D. Bravo, T. O. Dickson, X. Guo, R. Li, N. Trichy, J. Caserta, W. R. Bomstad, II, J. Branch, D.-J. Yang, J. Bohorquez, E. Seok, L. Gao, A. Sugavanam, J. J. Lin, J. Chen, and J. E. Brewer, “On-chip antennas in silicon ICs and their application,” IEEE Trans. Electron Devices, vol. 52, no. 7, pp. 1312–1323, Jul. 2005. [6] Z. M. Chen and Y. Zhang, “Inter-chip wireless communication channel: Measurement, characterization, and modeling,” IEEE Trans. Antennas Propag., vol. 55, no. 3, pp. 978–986, Mar. 2007. [7] M. Sun, Y. P. Zhang, G. X. Zheng, and W. Yin, “Performance of intrachip wireless interconnect using on-chip antennas and UWB radios,” IEEE Trans. Antennas Propag., vol. 57, no. 9, pp. 2756–2762, Sep. 2009. [8] H. Yordanov and P. Russer, “Integrated on-chip antennas using CMOS ground planes,” in Proc. 10th Silicon Monolithic Integr. Circuits RF Syst. Topical Meeting, New Orleans, LA, Jan. 2010, pp. 53–56. [9] J. W. Bowen, S. Hadjiloucas, B. M. Towlson, L. S. Karatzas, S. T. G. Wootton, N. J. Cronin, S. R. Davies, C. E. McIntosh, J. M. Chamberlain, R. E. Miles, and R. D. Pollard, “Micromachined waveguide antennas for 1.6 THz,” Electron. Lett., vol. 42, no. 15, pp. 842–843, Jul. 2006. [10] A. Semenov, H. Richter, H. Hubers, B. Gunther, A. Smirnov, K. Ll’in, M. Siegel, and P. Karamarkovic, “Terahertz performance of integrated lens antennas with a hot-electron bolometer,” IEEE Trans. Microw. Theory Tech., vol. 55, no. 2, pp. 239–247, Feb. 2007. [11] N. Yu, E. Cubukcu, L. Diehl, M. A. Belkin, K. B. Crozier, F. Capasso, D. Bour, S. Corzine, and G. Hofler, “Plasmonic quantum cascade laser antenna,” Appl. Phys. Lett., vol. 91, pp. 173113-1–173113-3, 2007. [12] J. N. Farahani, H.-J. Eisler, D. W. Pohl, M. Pavius, P. Flückiger, G. Philippe, and B. Hecht, “Bow-tie optical antenna probes for singleemitter scanning near-field optical microscopy,” Nanotechnology, vol. 18, pp. 125–506, 2007. [13] “Silicon Photonics Platform, Building a Vertical Fibre Coupling Setup, Version 1.0,” ePIXfab, Gent, Belgium, 2006. [14] Y. Vlasov and S. McNab, “Losses in single-mode silicon-on-insulator strip waveguides and bends,” Opt. Exp., vol. 12, no. 8, pp. 1622–1631, Apr. 2004.

Hongyu Zhou (S’07) received the Dipl. Eng. degree in electrical engineering from the Harbin Institute of Technology, Harbin, China, in 2007, the M.S.E.E. degree from the University of Colorado at Boulder, in 2009, and is currently working toward the Ph.D. degree at the University of Colorado at Boulder. His research interests include RF/microwave, millimeter-wave and optical components, and broadband antennas.

Xi Chen (S’10) received the B.S. degree in microelectronics from Peking University, Beijing, China, in 2006, the M.S.E.E degree from the University of Colorado at Boulder, in 2009, and is currently working toward the Ph.D. degree at the University of Colorado at Boulder. His research interests include nanophotonics material and devices.

David S. Espinoza (S’10) received the Bachelor of Science (B.S.) degree in electrical engineering from the Pontifical Catholic University of Peru, Lima, Peru, in 2002, and the Master of Science (M.S.) degree in telecommunications from the University of Colorado at Boulder, in 2011. He is currently with the Guided Waves Optics Laboratory, University of Colorado at Boulder. His research interests include wireless telecommunications and polymer application on SOI nanophotonic devices.

Alan Mickelson (S’72–M’78–SM’92) received the M.S. and Ph.D. degrees from the California Institute of Technology, Pasadena, in 1974 and 1978, respectively. Following a postdoctoral period with the California Institute of Technology in 1980, he joined the ElectronicsResearch Laboratory, Norwegian Institute of Technology, Trondheim, Norway, initially as an NTNF Postdoctoral Fellow, and then as a Staff Scientist, during which time his research primarily concerned characterization of optical fibers and fiber compatible components and devices. Since 1984, his research has included nonlinear optical devices in inorganic and organic materials, semiconductor laser stabilization, electro-optic sampling of microwave signals, RF photonics, and optical characterization of biological materials. His current research involves silicon photonics, plasmonic metamaterials, and information and communcation technology for development.

2632


Dejan S. Filipovic (S’97–M’02–SM’08) received the Dipl. Eng. degree in electrical engineering from the University of Niˇs, Niˇs, Serbia, in 1994, and the M.S.E.E. and Ph.D. degrees from The University of Michigan at Ann Arbor, in 1999 and 2002, respectively. He is currently an Associate Professor with the University of Colorado at Boulder. His research interests are antenna theory and design, modeling and design of passive millimeter-wave components and systems, as well as computational and applied electromagnetics.

Mr. Filipovic was the recipient of the Nikola Tesla Award and Provost’s Faculty Achievement Award. He and his students were corecipients of the Best Paper Award presented at the IEEE Antennas and Propagation Society (AP-S)/ URSI and Antenna Application Symposium Conferences.