The core activity of CIPS is the development of a long-range vision for research and the development of integrated photonic devices and systems.  As an academic institution we can work openly with a variety of different organizations in developing and gathering input for our models. Whether it is performance data for new devices “in the lab,” yield data for existing manufacturing processes, planning documents, or first-hand observations of the corporate decision making process, CIPS researchers benefit greatly from the unique relationship between MIT and industry. The level of detail and intellectual rigor of the models being developed here is complemented by the high quality of data available to us. CIPS researchers are developing models of optical and electronic devices, the packages they are wrapped inside, the manufacturing processes that assemble them, the standards that define them, the market that buys them, and the policy processes that influence their deployment.

The Departments of Electrical Engineering and Computer Science, Materials Science and Engineering, Mechanical Engineering and Economics are consistently ranked as the top graduate programs in the country. Likewise, the Sloan School of Management has consistently ranked first in the nation in the areas of information technology, operations research, and supply chain management. CIPS leverages MIT’s strengths, by unifying the photonics researchers in these departments and laboratories to focus on technological developments in photonics. The combined volume of research funds in the photonics area at MIT exceeds $20 million dollars annually. The faculty and staff at MIT in photonics-related areas have included Claude Shannon (founder of information theory), Charles Townes (inventor of the laser), Robert Rediker (inventor of the semiconductor lasers), and Hermann Haus (inventor of the single-frequency semiconductor laser & ultrafast optical switch). CIPS-affiliated faculty and staff continue this tradition of excellence in areas ranging from optical network architectures, to novel optical devices, to novel photonic materials.

Collaborators

• H. Lee, RLE
• H. I. Smith, EECS
• F. X. Kaertner, EECS
• J. Hudgings, Mt. Holyoke College
• J. L. Hoyt, EECS
• V. Stojanovic, EECS
• K. Asanovic, UC Berkeley

Graduate Students

• R. Amatya, Res. Asst., EECS
• S. Goh, Res. Asst., DMSE
• E .Kapusta, Res. Asst, EECS
• K.S.K. Lee, Res. Asst., EECS
• J. Orcutt, Res. Asst., EECS
• P. Santhanam, Res. Asst., EECS

Support Staff

• G. Brew, Administrative Assistant II

Publications

Lee, K.S. and Ram, R.J., 2009, Plastic–PDMS bonding for high pressure hydrolytically stable active microfluidics, Lab on a Chip, 9, 1618-24, 2009.

M Farzaneh, K Maize, D Lüerßen, J A Summers, P M Mayer, P E Raad, K P Pipe, A Shakouri, R J Ram and J A Hudgings, 2009, CCD-based thermoreflectance microscopy: principles and applications, J. Phys. D: Appl. Phys., 42, 143001, 2009.

Santhanam, P.. and Ram, R.J., 2010, Self-Consistent Drift–Diffusion Transport in Thermoelectrics and Implications for Measuring the Scattering Parameter, Journal of Electronic Materials, on-line, 2010.

Amatya, R.. and Ram, R.J., 2010, Solar Thermoelectric Generator for Micropower Applications, Journal of Electronic Materials, on-line April, 2010.

Orcutt, J.. and Ram, R.J., 2010, Photonic Device Layout within the Foundry CMOS Design Environment, IEEE Photonics Technology Letters, 22, 544-546, 2010.

Summers, Joseph A. and Farzaneh, Maryam and Ram, Rajeev J. and Hudgings, Janice A.. 2010, Thermal and Optical Characterization of Photonic Integrated Circuits by Thermoreflectance Microscopy. IEEE Journal of Quantum Electronics, 46, 3-10, 2010

Surface plasmon polaritons (SPPs) are optical waves that propagate at the interface between a metal and a dielectric, and can provide high optical mode confinement for the realization of compact optical devices ^{[1] [2]} .  Because SPPs can support polarizations only where the electric field is normal to the metal-semiconductor interface, metals can be patterned onto purely dielectric waveguides to form optical polarizers, where either the TE or TM mode of the dielectric waveguide is resonantly coupled to the SPP ^{[3] [4]} .

Here, we describe the design of a compact SPP-based TE polarizer that can be easily integrated with a standard InP-based ridge waveguide (Figure 1).  The InP ridge waveguide consists of a 610-nm-thick InGaAsP core (Q=1.44), with InP upper and lower claddings, and the 8-mm-long TE polarizer is created by coating the top and sides of the ridge with a layer of Ag, allowing an SPP mode to be supported at the top Ag-InP interface.

The polarizer works by removing TM polarized light from the semiconductor ridge via resonant coupling to the TM mode of the SPP.  After light is fully coupled to the SPP, the Ag layer is abruptly terminated, allowing the light to radiate into unguided modes.  Due to the large absorption loss of the SPP-mode (~0.8dB/mm), resonant coupling from the semiconductor ridge mode is sensitive to the coupling strength and the relative propagation losses of the symmetric and antisymmetric coupled waveguide supermodes.  If the InP upper cladding thickness is varied, the coupling strength can be adjusted to achieve critical coupling, where the propagation losses of the ridge-SPP supermodes are equal.  For this waveguide geometry, critical coupling and optimal power transfer to the SPP mode occurs at an upper cladding thickness of 280 nm, enabling a compact device length of 8 mm, 20 dB polarization extinction, and 2 dB insertion loss at 1550 nm (Figure 2).

1. W. L. Barnes, A. Dereux, and T. W. Ebbesen, “Surface plasmon subwavelength optics,” Nature, vol. 424, pp. 824-830, Aug. 2003.
2. S. A. Maier and H. A. Atwater, “Plasmonics: Localization and guiding of electromagnetic energy in metal/dielectric structures,” Journal of Applied Physics, vol. 98 no. 011101, July 2005.
3. C.-H. Chen and L. Wang, “Design of finite-length metal-clad optical waveguide polarizer,” IEEE Journal of Quantum Electronics, vol. 34, no. 7, pp. 1089-1097, July 1998.
4. W. Johnstone, G. Stewart, T. Hart, and B. Culshaw, “Surface plasmon polaritons in thin metal films and their role in fiber optic polarizing devices,” IEEE Journal of Lightwave Technology, vol. 8, no. 4, pp. 538-544, Apr. 1990.

There exists great interest in the utilization of the human body as a possible power supply for running such electronics as medical monitoring devices and sensors. This project details the effective utilization of a new type of form-fitting, wearable, wristband thermoelectric heat sink, Figure 1, which relies on 2D heat-spreading for dissipating heat from the cold side of the TEG for the purposes of energy harvesting. The performance of the TEG/heat sink systems is limited by (1) the small working temperature differential between the body and ambient temperature; (2) the desire to use natural air convection cooling on the cold-side of the generator; and (3) the requirement for thin, light-weight systems that are comfortable for long-term use. The 2D heat-spreading wristbands produce 285 μW from a Bi₂Te₃ TEG module (256 thermocouples, each is 2 mm in thickness and 1 mm² in area). The large available surface area of the wristband as well as the highly thermally conductive nature of copper ensures effective cooling of the cold side and allows heat to spread and dissipate throughout the wristband, thereby enhancing the effective heat transfer coefficient. A 75-minute field test was conducted that measured the voltage output from the TEG as well as the temperature difference across the module when worn on the upper forearm. In addition, the output of the TEG was connected to a commercially available DC-DC converter, which boosted the voltage to a steady 4.15 V, shown in Figure 2, allowing for coupling with external applications such as EKG chips and pulse oximeters.

We are investigating thermoelectric transport in light-emitting diodes (LEDs). Electro-luminescent cooling is a phenomenon in which the optical output power from an LED exceeds its electrical input power, resulting in a net cooling of the emitter. The physical process is outlined in Figure 1. Proposed in the 1950s ^[1] , electro-luminescent cooling has been a topic of recent theoretical interest ^{[2] [3]} , but there are no known reports that it has been experimentally observed. Additionally, this investigation could lead to practical light emitters operating with efficiencies approaching or even surpassing unity.

The Second Law of Thermodynamics places limits on the power density achievable in these high-efficiency devices. Nevertheless, LEDs designed to exploit thermoelectric transport may find practical application in lighting, spectroscopy, and communication.

Since such devices would see improved efficiency at higher ambient temperatures, this work could lead to LED designs that do not suffer from efficiency droop, a major hurdle for state-of-the-art LEDs ^[4] . If LED packaging was designed for the concentration of heat, just as in an incandescent filament, self-heating due to non-radiative recombination could raise the achievable power density at visible wavelengths to practical levels.

Remote and portable spectroscopy applications like pulse oximetry, infrared oilfield spectroscopy, and automotive emissions monitoring may benefit from such highly efficient photon sources. Drastically reduced power demands may increase the lifespan of such remote platforms, or even permit the use of ambient energy-harvesting power sources in place of batteries. The fundamental limitations on the power density of this technology do not impede this case, as the less-constricting power requirements for spectroscopic sources are set by the noise-equivalent power of the detector and the required signal-to-noise ratio. Similarly, since the power requirements for infrared optical communication are set by ambient blackbody radiation, the quality of the detector, and the desired bitrate, dim but efficient sources may prove practical.

1. J. Tauc, “The share of thermal energy taken from the surroundings in the electro-luminescent energy radiated from a p-n junction,” Czechoslovak Journal of Physics, vol. 7, no. 3, pp. 275-276, Jan. 1957.
2. O. Heikkilä, J. Oksanen, and J. Tulkki, “Ultimate limit and temperature dependency of light-emitting diode efficiency,” Journal of Applied Physics, vol. 105, pp. 093119:1-9, May 2009.
3. P. Han, K.-J. Jin, Y.-L. Zhou, H.-B. Lu, and G.-H. Yang, “Numerical designing of semiconductor structure for optothermionic refrigeration,” Journal of Applied Physics, vol. 101, pp. 014506:1-4, Jan. 2007.
4. J. Piprek, “Efficiency droop in nitride-based light-emitting diodes,” Phys. Status Solidi A, vol. 207, no. 10, pp. 2217-2225, July 2010.

Figure 1: (a) Propagation loss as a function of waveguide width for wafers with and without thermal processing representative of the full electronics process. Solid fit lines estimating the confinement factor scaling for the indicated bulk material losses overlay measured data points. Inset of electric field contours show confinement change from core to cladding material for narrow waveguide widths. (b) Propagation loss as a function of wavelength and waveguide width for the thermally processed wafer.

Low propagation loss deposited waveguides are an important component to enable photonic integration in the majority of high-volume electronics processes. Polycrystalline silicon waveguides are desirable in this role as propagation losses below 10 dB/cm are achievable ^{[1] [2] [3]} using materials already common in such processes. However, previous demonstrations of low-loss poly-Si waveguides have utilized layer thicknesses of 200 nm or greater and reduced index-contrast oxynitride claddings to achieve such results. Further, the low-loss performance was not verified to withstand the high-temperature steps present in electronics processes.

To provide a suitable process integration test platform, 120-nm thick poly-Si waveguides were deposited and patterned on 300- mm wafers within the surrounding dielectric stack-up of a state-of-the-art memory process. To measure the impact of the thermal processing, the wafers were split with and without the all of the high-temperature anneals present in the full process for comparison. To minimize the impact of this thermal processing on top surface roughness, the waveguide core was first deposited at low-temperature by LPCVD to form an amorphous film and crystallized to form poly-Si with a ~950 °C 20-second anneal.

To measure the resulting waveguide loss, “paperclip” structures were patterned in which straight waveguide sections of varying propagation lengths are connected by identical input and output bends and vertical grating couplers. The resulting losses at 1550 nm as a function of waveguide width for wafers with and without thermal processing are shown in Figure 1a and are shown as a function of wavelength for the thermally processed wafers in Figure 1b. Although the waveguide loss increases slightly after thermal processing, both the as-deposited and end-of-line waveguide losses are below 10 dB/cm for 350-nm waveguide widths at 1550 nm. The narrow waveguides enable long-distance, on-chip routing to be enabled with 6.2 dB/cm loss.

1. L. Liao et al., “Optical transmission losses in polycrystalline silicon strip waveguides: effects of waveguide dimensions, thermal treatment, hydrogen passivation, and wavelength,” J. Electron. Mater. vol. 29, pp. 1380-1386, 2000.
2. Q. Fang et al., “Low loss (~6.45 dB/cm) sub-micron polycrystalline silicon waveguide integrated with efficient SiON waveguide coupler,” Opt. Express, vol. 16, pp. 6425-6432, 2008.
3. C. W. Holzwarth et al., “Localized substrate removal technique enabling strong-confinement microphotonics in bulk Si CMOS processes,” in Conference on Lasers and Electro-Optics, Technical Digest (Optical Society of America, 2008), paper CThKK5.

Figure 1: (a) Through transmission of 40 µm radius ring with 200 nm coupling gap for four different in-waveguide input powers. (b) Measured through transmissions (circles) on resonance normalized to off-resonance transmission, and fits (lines) for 20 µm rings, and (c) for 40 µm rings with various coupling gaps g; along with the gap, the fitted β is labeled for each sample.

Nonlinear optics comprises a rich body of physics that could enable a number of interesting functionalities in integrated silicon photonics ^[1] . We are currently interested in the phenomenon of two-photon absorption (TPA) as a means to achieving sub-bandgap photodetection in silicon. Though TPA is a relatively weak absorption mechanism in crystalline silicon, the fact that absorbed power scales quadratically with intensity implies that its strength can be increased for device applications by concentrating a large amount of light energy in a small mode-volume resonator, for example in photonic crystal microcavities ^{[2] [3]} .

Further enhancement of TPA may be expected when polycrystalline Si is used instead of crystalline Si as the detector material, due to a larger TPA coefficient (β) resulting from mid-gap electronic states at grain boundaries. Though β has been fairly well characterized for crystalline Si ^[4] , an understanding of two-photon absorption in the polycrystalline phase is lacking. We have therefore measured β in polycrystalline Si waveguide devices by monitoring the peak through transmission dip in ring resonator devices as a function of input power and fitting the results to a model with β as a fitting parameter; data from such measurements are shown in Figure 1. These measurements indicate β = 310 ± 70 cm/GW, a value over two orders of magnitude larger than that in crystalline Si at 1550 nm and on the same order of magnitude as the value observed in amorphous silicon ^[5] . As polysilicon is widely available in advanced CMOS processes, these results suggest that high-density, integrated devices utilizing nonlinear absorption at relatively low powers should be achievable.

1. J. Leuthold, C. Koos, and W. Freude, “Nonlinear silicon photonics.” Nature Photonics, vol. 4, pp. 535-544, July 2010.
2. J. Bravo-Abad, E.P. Ippen, and M. Soljacic. “Ultrafast photodetection in an all-silicon chip enabled by two-photon absorption.” Appl. Phys. Lett., vol. 94, no. 241103, pp. 1-3, June 2009.
3. T. Tanabe, H. Sumikura, H. Taniyama, A. Shinya, and M. Notomi. “All-silicon sub-Gb/s telecom detector with low dark current and high quantum efficiency on chip,” Appl. Phys. Lett., vol. 96, no. 101103, pp. 1-3, March 2010.
4. M. Dinu, F. Quochi, and H. Garcia. “Third-order nonlinearities in silicon at telecom wavelengths,” Appl. Phys. Lett., vol. 82, pp. 2954-2956, May 2003.
5. K. Ikeda,Y. Shen and Y. Fainman. “Enhanced optical nonlinearity in amorphous silicon and its application to waveguide devices,” Opt. Express, vol. 15, 17761-17771, December 2003.

We are investigating the performance of low-noise, high-power optoelectronic oscillators (OEO) using all slab-coupled optical waveguide (SCOW) components. The optoelectronic oscillator (OEO) demonstrates the capability to generate a pristine tone through down-conversion of a modulated optical carrier into the microwave regime ^{[1] [2]} . The two primary advantages for using an optical carrier are low loss (high quality factor (Q)) and large bandwidth (50+ GHz modulation and detection). Other notable advantages of the OEO are its light weight, immunity to electromagnetic interference, and ability to generate both optical and microwave clock outputs. A schematic diagram of the SCOW based-OEO structure is shown in Figure 1. The phase noise performance of an OEO depends largely on the relative intensity noise (RIN) properties of the laser source. Figure 2 compares the SCOW external cavity laser (SCOWECL) ^[3] to a commercial external cavity semiconductor laser (RIO Orion). The SCOWECL RIN is 10-15 dB lower than the RIO Orion RIN.

High performance optoelectronic oscillators are important as low-noise sources for driving low noise modelocked lasers and as stable local oscillators (LO) for RADAR and communication applications. In modelocked lasers, the noise of the RF oscillator directly transfers to the laser through jitter present during modulation of the optical pulse. In RADAR, the reflected signal after the Doppler shift must overcome the phase noise of the master oscillator reflected off background clutter. Finally, in global positioning system (GPS) applications, the master oscillator’s stability is essential during triangulation of the receiver position. Clock phase noise results in range errors that limit the accuracy of the computed receiver position. In addition to requiring high performance, all of these applications can benefit from a low-size, -weight, and -power compact master oscillator. This is especially true during flight as SWaP becomes critically important for airborne systems compared to ground-based systems.

1. X. S. Yao and L. Maleki, “Optoelectronic microwave oscillator,” J. Opt. Soc. Am. B, vol. 13, pp. 1725-1735, 1996.
2. D. Eliyahu and L. Maleki, “Low phase noise and spurious level in multi-loop opto-electronic oscillators,” in Proc. IEEE International Frequency Control Symposium, 2003, pp. 405-410.
3. W. Loh, F. J. O’Donnell, J. J. Plant, M. A. Brattain, L. J. Missaggia, and P. W. Juodawlkis, “Packaged, high-power, narrow-linewidth slab-coupled optical waveguide external cavity laser (SCOWECL),” accepted for publication in IEEE Photon. Technol. Lett., 2011.

Carbon nanotubes (CNTs) are excellent heat spreaders for packaging electronic/photonic devices due to their high thermal conductivity and good mechanical strength ^{[1] [2]} . They have been used for packaging LED chips ^{[3] [4] [5]} . However, poor adhesion between CNTs and adjacent materials may impact device yield and reliability. We proposed a platform for optoelectronic device packaging by integrating vertical aligned CNTs (VACNTs) onto patterned silicon U-grooves for achieving enhanced thermal conduction and stable mechanical properties.

Figure 1 shows the schematic of the CNT-based platform. CNTs are grown on a silicon trench to provide heat spreading for the bonded device chip. A thin dielectric layer between CNTs and silicon substrate is usually needed to eliminate current leakage to the substrate. The device chip is bonded to the platform by the bonding pads on chip surface. Figure 2 shows the calculated thermal resistance of the CNT platform by using the Finite Element Method. A silicon monoxide (SiO) film of 500-nm thick is used as the dielectric layer. The planar silicon platform with the same thickness of SiO is also simulated for comparisons. When the top open area of the U-groove covers about 87% of the device chip and the depth is about 30% of the 200-mm thick silicon substrate, the CNT platform can provide about 39% reduction in thermal resistance if the thermal conductivity of CNTs is 600 W/mK. The reduction can reach about 50% by using a deeper U-groove with taller CNTs. The enhancement on the thermal conduction by the CNT platform results not only from the use of high-thermal-conductivity VACNTs but also from the increased surface area of the low-thermal-conductance dielectric layer.

The CNT platforms were demonstrated for packaging red LEDs to have high saturation current and high manufacturing yield ^[6] .

1. S. Berber, Y. K. Kwon, and D. Tomanek, “Unusually high thermal conductivity of carbon nanotubes,” Phys. Rev. Lett., vol. 84, pp. 4613-4616, 2000.
2. P. Kim, L. Shi, A. Majumdar, and P. L. McEuen, “Thermal transport measurements of individual multiwalled nanotubes,” Phys. Rev. Lett., vol. 87, pp. 215502-215505, 2001.
3. K. Zhang, G. W. Xiao, C. K. Y. Wong, H. W. Gu, M. M. F. Yuen, P. C. H. Chan, and B. Xu, “Study on thermal interface material with carbon nanotubes and carbon black in high-brightness LED packaging with flip-chip technology,” Proc. 55^th IEEE Electronic Components and Technology Conference, pp. 60-65, Lake Buena Vista, FL, May 2005.
4. K. Zhang, Y. Chai, M. M. F. Yuen, D. G. W. Xiao, and P. C. H. Chan, “Carbon nanotube thermal interface material for high-brightness light-emitting diode cooling,” Nanotechnology, vol. 19, pp. 215706-215714, 2008.
5. M. Arik, S. E. Weaver, Jr., J. C. Carnahan, C. A.  Becker,  W. D. Gerstler, “Electronic devices and methods for making same using nanotube regions to assist in thermal heat-sinking,” US Patent, 6864571, Mar. 8,  2005.
6. S.-C. Chen, S.-L. Lee, H. Lo, Y.-J. Hung, K.-Y., Lee, C.-G. Tu, Y.-T. Pan, and R. J. Ram, “VACNT-on-silicon platform for improving heat conduction in optoelectronic packaging,” in Proc. Conference on Lasers and Electro-Optics (CLEO), Baltimore, MD., May  2011.

For systems biology, the models are more often limited by the absence of reliable experimental data than by available computational resources. Unfortunately, there is still great difficulty in making the leap from genetic and biochemical analysis to accurate verification with conventional culture growth experiments due to variations in culture conditions. Measurements of metabolic activity through substrate and product interactions or cellular activity through fluorescent interactions are highly dependent on environmental conditions and cellular metabolic state. For such experiments to be feasible, continuous cultures ^{[1] [2]} utilizing control strategies must be developed to measure chemical concentrations, introduce chemical inputs, and remove waste. An integrated microreactor system with built-in fluid metering will enable environmental control and programmable experiments capable of generating reproducible data.

The chip shown in Figure 1 is fabricated out of a rigid plastic polycarbonate, utilizing PDMS membranes for actuation and pumping ^[3] . The fabrication process for bonding plastic-PDMS hybrid devices has been described previously ^[4] . Mixing and oxygen delivery are performed through membranes between the fluidic and actuation layers of the growth chamber sections. A growth volume of 1 mL ensures the ability to couple sampled volume to offline chemical analysis. Culture experiments are performed using E. coli strain FB21591 grown on defined media. The glucose input is separated to provide input control. As shown in Figure 2, various metabolic states are observable through continuous flow control. Cell density is directly dependent on glucose input, and acid production is proportional to cell density in chemostat mode. In turbidostat mode, cell density can be kept constant and glucose utilization can be observed, demonstrating the direct observation of overflow metabolism.

1. J. Monod, “La technique de culture continue, theorie et applications,” Annales de I’Institut Pasteur, vol. 79, pp. 390-410, 1950.
2. A. Novick and L. Szilard, “Description of the chemostat,” Science, vol. 112, pp. 715-716, 1950.
3. V. Studer, G. Hang, A. Pandolfi, M. Ortiz, W. F. Anderson, and S. R. Quake, “Scaling properties of a low-actuation pressure microfluidic valve,” J. Applied Physics, vol. 95, pp. 393-398, 2004.
4. K. S. Lee and R. J. Ram, “Plastic-PDMS bonding for high pressure hydrolytically stable active microfluidics,” Lab on a Chip, no. 9, pp. 1618-1624, 2009.