The high demand for machine intelligence of doubling every three months is driving novel hardware... more The high demand for machine intelligence of doubling every three months is driving novel hardware solutions beyond charging of electrical wires given a resurrection to application specific integrated circuit (ASIC)-based accelerators. These innovations include photonic-based ASICs (P-ASIC) due to prospects of performing optical linear (and also nonlinear) operations, such as multiply-accumulate for vector matrix multiplications or convolutions, without iterative architectures. Such photonic linear algebra enables picosecond delay when photonic integrated circuits are utilized, via 'on-the-fly' mathematics. However, the neuron's full function includes providing a nonlinear activation function, knowns as thresholding, to enable decision making on inferred data. Many P-ASIC solutions performing this nonlinearity in the electronic domain, which brings challenges in terms of data throughput and delay, thus breaking the optical link and introducing increased system complexity via domain crossings. This work follows the notion of utilizing enhanced light-matter-interactions to provide efficient, compact, and engineerable electro-optic neuron nonlinearity. Here, we introduce and demonstrate a novel electro-optic device to engineer the shape of this optical nonlinearity to resemble a rectifying linear unit (ReLU)-the most-commonly used nonlinear activation function in neural networks. We combine the counter-directional transfer functions from heterostructures made out of two electro-optic materials to design a diode-like nonlinear response of the device. Integrating this nonlinearity into a photonic neural network, we show how the electrostatics of this thresholder's gating junction improves machine learning inference accuracy and the energy efficiency of the neural network. THE MANUSCRIPT I. INTRODUCTION The growing demands of neural network systems create an urgent need for the development of advanced devices to perform complex operations with fast throughput (operations/s) and lower power dissipation (J/operations), and compact footprint leading to high operation density (operations/s/mm 2) [1, 2]. Photonic integrated circuit based artificial neurons can pave the way for this specific challenge. One of the most significant benefits of photonics over electronics is that distinct signals can be straightforwardly and efficiently combined due to their wave-nature exploiting atto-Joule efficient electro-optic (EO) modulators [3-5], phase shifters, and combiners, simplifying essential operations such as weighted sum or addition, vector
Thanks to the unique molecular fingerprints in the mid-infrared spectral region, absorption spect... more Thanks to the unique molecular fingerprints in the mid-infrared spectral region, absorption spectroscopy in this regime has attracted widespread attention in recent years. Contrary to commercially available infrared spectrometers, which are limited by being bulky and cost-intensive, laboratory-on-chip infrared spectrometers can offer sensor advancements including raw sensing performance in addition to utilization such as enhanced portability. Several platforms have been proposed in the past for on-chip ethanol detection. However, selective sensing with high sensitivity at room temperature has remained a challenge. Here, we experimentally demonstrate an on-chip ethyl alcohol sensor based on a holey photonic crystal waveguide on silicon on insulator-based photonics sensing platform offering an enhanced photoabsorption thus improving sensitivity. This is achieved by designing and engineering an optical slow-light mode with a high group-index of n g = 73 and a strong localization of th...
Atomically thin 2D materials provide a wide range of basic building blocks with unique properties... more Atomically thin 2D materials provide a wide range of basic building blocks with unique properties, making them ideal for heterogeneous integration with a mature chip platform. An understanding the role of excitons in transition metal dichalcogenides in Silicon photonic platform is a prerequisite for advances in optical communication technology, signal processing, and possibly computing. Here we demonstrate passive tunable coupling by integrating few layers of MoTe2 on a micro-ring resonator. We find a TMD-to-rings circumference coverage length ratio to place the ring into critical coupling to be about 10% as determined from the variation of spectral resonance visibility and loss as a function of TMD coverage. Using this TMD ring heterostructure, we further demonstrate a semi-empirical method to determine the index of an unknown TMD material (nMoTe2 of 4.36+.011i) near for telecommunication-relevant wavelength.
Electrooptic modulation performs the conversion between the electrical and optical domain with ap... more Electrooptic modulation performs the conversion between the electrical and optical domain with applications in data communication for optical interconnects, but also for novel optical compute algorithms such as providing nonlinearity at the output stage of optical perceptrons in neuromorphic analogue optical computing. Since the clock frequency for photonics on chip has a power overhead sweet slot around 10s of GHz, ultrafast modulation may only be required in long distance communication, but not for short onchip links. Here we show a roadmap towards atto Joule per bit efficient modulators on chip as well as some experimental demonstrations of novel plasmon modulators with sub 1fJ per bit efficiencies. We then discuss the first experimental demonstration of a photon plasmon-hybrid Graphene-based electroabsorption modulator on silicon. The device shows a sub 1V steep switching enabled by near ideal electrostatics delivering a high 0.05dB per V um performance requiring only 110 aJ per...
Electro-optic modulators are utilized ubiquitously ranging from applications in data communicatio... more Electro-optic modulators are utilized ubiquitously ranging from applications in data communication to photonic neural networks. While tremendous progress has been made over the years, efficient phase-shifting modulators are challenged with fundamental tradeoffs, such as voltage-length, index change-losses or energy-bandwidth, and no single solution available checks all boxes. While voltage-driven phase modulators, such as based on lithium niobate, offer low loss and high speed operation, their footprint of 10's of cm-scale is prohibitively large, especially for density-critical applications, for example in photonic neural networks. Ignoring modulators for quantum applications, where loss is critical, here we distinguish between current versus voltage-driven modulators. We focus on the former, since current-based schemes of emerging thin electro-optical materials have shown unity-strong index modulation suitable for heterogeneous integration into foundry waveguides. Here, we prov...
The high demand for machine intelligence of doubling every three months is driving novel hardware... more The high demand for machine intelligence of doubling every three months is driving novel hardware solutions beyond charging of electrical wires given a resurrection to application specific integrated circuit (ASIC)-based accelerators. These innovations include photonic-based ASICs (P-ASIC) due to prospects of performing optical linear (and also nonlinear) operations, such as multiply-accumulate for vector matrix multiplications or convolutions, without iterative architectures. Such photonic linear algebra enables picosecond delay when photonic integrated circuits are utilized, via 'on-the-fly' mathematics. However, the neuron's full function includes providing a nonlinear activation function, knowns as thresholding, to enable decision making on inferred data. Many P-ASIC solutions performing this nonlinearity in the electronic domain, which brings challenges in terms of data throughput and delay, thus breaking the optical link and introducing increased system complexity via domain crossings. This work follows the notion of utilizing enhanced light-matter-interactions to provide efficient, compact, and engineerable electro-optic neuron nonlinearity. Here, we introduce and demonstrate a novel electro-optic device to engineer the shape of this optical nonlinearity to resemble a leaky rectifying linear unit (ReLU)-the most-commonly used nonlinear activation function in neural networks. We combine the counter-directional transfer functions from heterostructures made out of two electro-optic materials to design a diode-like nonlinear response of the device. Integrating this nonlinearity into a photonic neural network, we show how the electrostatics of this thresholder's gating junction improves machine learning inference accuracy and the energy efficiency of the neural network. THE MANUSCRIPT I. INTRODUCTION The growing demands of neural network systems create an urgent need for the development of advanced devices to perform complex operations with fast throughput (operations/s) and lower power dissipation (J/operations), and compact footprint leading to high operation density (operations/s/mm 2) [1, 2]. Photonic integrated circuit based artificial neurons can pave the way for this specific challenge. One of the most significant benefits of photonics over electronics is that distinct signals can be straightforwardly and efficiently combined due to their wave-nature exploiting atto-Joule efficient electro-optic (EO) modulators, phase shifters, and combiners [3-5], simplifying essential operations such as weighted sum or addition, vector
Photodetectors are key optoelectronic building blocks performing the essential optical-to-electri... more Photodetectors are key optoelectronic building blocks performing the essential optical-to-electrical signal conversion, and unlike solar cells, operate at a specific wavelength and at high signal or sensory speeds. Towards achieving high detector performance, device physics, however, places a fundamental limit of the achievable detector sensitivity, such as responsivity and gain, when simultaneously aimed to increasing the detector’s temporal response (speed) known as the gain-bandwidth product (GBP). While detector’s GBP has been increasing in recent years, the average GBP is still relatively modest (∼106-109 Hz-A/W). Here we discuss photoconductor-based detector performance limits and opportunities based on arguments from scaling length theory relating photocarrier channel length, mobility, electrical resistance with optical waveguide mode constrains. We show that short-channel detectors are synergistic with slot-waveguide approaches, and when combined, offer a high-degree of dete...
In integrated photonics, specific wavelengths are preferred such as 1550 nm due to low-loss trans... more In integrated photonics, specific wavelengths are preferred such as 1550 nm due to low-loss transmission and the availability of optical gain in this spectral region. For chip-based photodetectors, layered two-dimensional (2D) materials bear scientific and technologicallyrelevant properties such as electrostatic tunability and strong light-matter interactions. However, no efficient photodetector in the telecommunication C-band has been realized with 2D transition metal dichalcogenide (TMDCs) materials due to their large optical bandgap. Here, we demonstrate a MoTe2-based photodetector featuring strong photoresponse (responsivity = 0.5 A/W) operating at 1550 nm on silicon micro ring resonator enabled by strain engineering of the transition-metal-dichalcogenide film. We show that an induced tensile strain of ~4% reduces the bandgap of MoTe2, resulting in large photo-response in the telecommunication wavelength, in otherwise photo-inactive medium when unstrained. Unlike Graphene-based photodetectors relying on a gapless band structure, this semiconductor-2D material detector shows a ~100X improved dark current enabling an efficient noise-equivalent power of just 90 pW/Hz 0.5. Such strain-engineered integrated photodetector provides new opportunities for integrated optoelectronic systems.
Here, we experimentally demonstrate an Indium Tin Oxide (ITO) Mach-Zehnder interferometer heterog... more Here, we experimentally demonstrate an Indium Tin Oxide (ITO) Mach-Zehnder interferometer heterogeneously integrated in silicon photonics. The phase shifter section is realized in a novel lateral MOS configuration, which, due to favorable electrostatic overlap, leads to efficient modulation (VπL = 63 V.µm). This is achieved by (i) selecting a strong index changing material (ITO) and (ii) improving the field-overlap as verified by the electrostatic field lines. Furthermore, we show that this platform serves as a building block in an end-fire silicon photonics optical phased array (OPA) with a half-wavelength pitch within the waveguides with anticipated performance, including narrow main beam lobe (<3°) and >10 dB suppression of the side lobes, while electrostatically steering the emission profile up to ±80°, and if further engineered, can lead not only towards nanosecond-fast beam steering capabilities in LiDAR systems but also in holographic display, free-space optical communications, and optical switches.
Atomically thin 2D materials such as transition metal dichalcogenides (TMDs) provide a wide range... more Atomically thin 2D materials such as transition metal dichalcogenides (TMDs) provide a wide range of basic building blocks with unique properties, making them ideal for heterogeneous integration with a mature chip platform for advances in optical communication technology. The control and understanding of the precise value of the optical index of these materials, however, is challenging, as the standard metrology techniques such as the millimeter-large ellipsometry is often not usable due the small lateral 2D material flake dimension. Here, we demonstrate an approach of passive tunable coupling by integrating few layers of MoTe2 onto a microring resonator connected to a waveguide bus. We find the TMD-to-ring circumference coverage length ratio required to precisely place the ring into a critical coupling condition to be about 10% as determined from the variation of spectral resonance visibility and loss as a function of TMD coverage. Using this TMD-ring heterostructure, we further de...
Layered two-dimensional (2D) materials provide a wide range of unique properties as compared to t... more Layered two-dimensional (2D) materials provide a wide range of unique properties as compared to their bulk counterpart, making them ideal for heterogeneous integration for on-chip interconnects. Hence, a detailed understanding of the loss and index change on Si integrated platform is a prerequisite for advances in opto-electronic devices impacting optical communication technology, signal processing, and possibly photonic-based computing. Here, we present an experimental guide to characterize transition metal dichalcogenides (TMDs), once monolithically integrated into the Silicon photonic platform at 1.55 µm wavelength. We describe the passive tunable coupling effect of the resonator in terms of loss induced as a function of 2D material layer coverage length and thickness. Further, we demonstrate a TMD-ring based hybrid platform as a refractive index sensor where resonance shift has been mapped out as a function of flakes thickness which correlates well with our simulated data. These experimental findings on passive TMD-Si hybrid platform open up a new dimension by controlling the effective change in loss and index, which may lead to the potential application of 2D material based active on chip photonics.
Electro-optic modulation is a key function in optical data communication and possible future opti... more Electro-optic modulation is a key function in optical data communication and possible future optical compute engines. The performance of modulators intricately depends on the interaction between the actively modulated material and the propagating waveguide mode. While a variety of high-performance modulators have been demonstrated, no comprehensive picture of what factors are most responsible for high performance has emerged so far. Here we report the first systematic and comprehensive analytical and computational investigation for high-performance compact on-chip electro-optic modulators by considering emerging active materials, model considerations and cavity feedback at the nanoscale. We discover that the delicate interplay between the material characteristics and the optical mode properties plays a key role in defining the modulator performance. Based on physical tradeoffs between index modulation, loss, optical confinement factors and slow-light effects, we find that there exis...
Tunable optical material properties enable novel applications in both versatile metamaterials and... more Tunable optical material properties enable novel applications in both versatile metamaterials and photonic components including optical sources and modulators. Transparent conductive oxides (TCOs) are able to highly tune their optical properties with applied bias via altering their free carrier concentration and hence plasma dispersion. The TCO material indium tin oxide (ITO) exhibits unity-strong index changes, and epsilon-near-zero behavior. However, with such tuning the corresponding high optical losses, originating from the fundamental Kramers-Kronig relations, result in low cavity finesse. However, achieving efficient tuning in ITO-cavities without using light matter interaction enhancement techniques such as polaritonic modes, which are inherently lossy, is a challenge. Here we discuss a novel one-dimensional photonic crystal nanobeam cavity to deliver a cavity system offering a wide range of resonance tuning range, while preserving physical compact footprints. We show that a vertical Silicon-slot waveguide incorporating an actively gated-ITO layer delivers ~3.4 nm of tuning. By deploying distributed feedback, we are able to keep the Q-factor moderately high with tuning. Combining this with the sub-diffraction limited mode volume (0.1 (λ/2n) 3) from the photonic (non-plasmonic) slot waveguide, facilitates a high Purcell factor exceeding one thousand. This strong light-matter-interaction shows that reducing the mode volume of a cavity outweighs reducing the losses in diffraction limited modal cavities such as those from bulk Si 3 N 4. These tunable cavities enable future modulators and optical sources such as tunable lasers.
An efficient silicon-based light source presents an unreached goal in the field of photonics, due... more An efficient silicon-based light source presents an unreached goal in the field of photonics, due to Silicon's indirect electronic band structure preventing direct carrier recombination and subsequent photon emission. Here we utilize inelastically tunneling electrons to demonstrate an electrically-driven light emitting silicon-based tunnel junction operating at room temperature. We show that such a junction is a source for plasmons driven by the electrical tunnel current. We find that the emission spectrum is not given by the quantum condition where the emission frequency would be proportional to the applied voltage, but the spectrum is determined by the spectral overlap between the energy-dependent tunnel current and the modal dispersion of the plasmon. Experimentally we find the highest light outcoupling efficiency corresponding to the skin-depth of the metallic contact of this metalinsulator-semiconductor junction. Distinct from LEDs, the temporal response of this tunnel source is not governed by nanosecond carrier lifetimes known to semiconductors, but rather by the tunnel event itself and Heisenberg's uncertainty principle.
IEEE Journal of Selected Topics in Quantum Electronics, 2017
The ability to modulate light using 2-dimensional (2D) materials is fundamentally challenged by t... more The ability to modulate light using 2-dimensional (2D) materials is fundamentally challenged by their small optical crosssection leading to miniscule modal confinements in diffractionlimited photonics despite intrinsically high electro-optic absorption modulation (EAM) potential given by their strong exciton binding energies. However the inherent polarization anisotropy in 2D materials and device tradeoffs lead to additional requirements with respect to electric field directions and modal confinement. A detailed relationship between modal confinement factor and obtainable modulation strength including definitions on bounding limits are outstanding. Here, we show that the modal confinement factor is a key parameter determining both the modulation strength and the modulator extinction ratio-to-insertion loss metric. We show that the modal confinement and hence the modulation strength of a single-layer modulated 2D material in a plasmonically confined mode is able to improve by more than 10× compared to diffraction-limited modes. Combined with the strong-index modulation of graphene, the modulation strength can be more than 2-orders of magnitude higher compared to Silicon-based EAMs. Furthermore, modal confinement was found to be synergistic with performance optimization via enhanced light-matter-interactions. These results show that there is room for scaling 2D-material EAMs with respect to modal engineering toward realizing synergistic designs leading to high-performance modulators.
Next generation on-chip light sources require high modulation bandwidth, compact footprint, and e... more Next generation on-chip light sources require high modulation bandwidth, compact footprint, and efficient power consumption. Plasmon-based sources are able to address the footprint challenge set by both the diffraction limited of light and internal laser physics such as plasmon utilization. However, the high losses, large plasmonic-momentum of these sources hinder efficient light coupling to on-chip waveguides, thus, questioning their usefulness. Here we show that plasmon light sources can be useful devices; they can deliver efficient outcoupling power to on-chip waveguides and are able to surpass modulation speeds set by gain-compression. We find that waveguide-integrated plasmon nanocavity sources allow to transfer about ∼60% of their emission into planar on-chip waveguides, while sustaining a physical small footprint of ∼0.06 μm 2. These sources are able to provide output powers of tens of microwatts for microamp-low injection currents and reach milliwatts for higher pump rates. Moreover, the direct modulation bandwidth exceeds that of classical, gain compression-limited on-chip sources by more than 200%. Furthermore, these novel sources feature high power efficiencies (∼1 fJ/bit) enabled by both minuscule electrical capacitance and efficient internal photon utilization. Such strong light−matter interaction devices might allow redesigning photonic circuits that only demand microwatts of signal power in the future.
Photonic signal processing requires efficient on-chip light sources with higher modulation bandwi... more Photonic signal processing requires efficient on-chip light sources with higher modulation bandwidths. Today's conventional fastest semiconductor diode lasers exhibit modulation speeds only on the order of a few tens of GHz due to gain compression effects and parasitic electrical capacitances. Here we theoretically show an electrically-driven carbon nanotube (CNT)-based laser utilizing strong light-matter-interaction via monolithic integration into Silicon photonic crystal nanobeam (PCNB) cavities. The laser is formed by single-walled CNTs inside a combo-cavity consisting of both a plasmonic metal-oxide-semiconductor hybrid mode embedded in the one dimensional PCNB cavity. The emission originates from interband recombinations of electrostatically-doped nanotubes depending on the tubes' chirality towards matching the C-band. Our simulation results show that the laser operates at telecom frequencies resulting in a power output > 3 (100) µW and > 100 (1000)'s GHz modulation speed at 1 × (10 ×) threshold. Such monolithic integration schemes provide an alternative promising approach for light source in future photonic integrated circuits.
No further publication processing will occur until we receive your response to this proof. Please... more No further publication processing will occur until we receive your response to this proof. Please respond within 48 hours of receipt of this message.
2015 IEEE Industry Applications Society Annual Meeting, 2015
This paper presents a methodology for optimizing investment in data center battery storage capaci... more This paper presents a methodology for optimizing investment in data center battery storage capacity. Utility grid managers spend significant resources towards predicting and matching available power generation capacity to demand in real time. It is therefore essential for the success of the power industry that economic dispatch, energy efficiency, and grid security be maintained as power requirements change. This is especially challenging for microgrids during periods of peak demand due to limited available capacity. Data centers possess a unique requirement for short-term battery power supply where cost savings, emissions reduction, and reliability enhancement can be achieved through investment in additional battery capacity. To maximize these benefits, an optimization methodology is presented through a case study for an existing data center and microgrid. Here we discuss a case study demonstrating the effectiveness of the proposed approach. For the selected mid-size data center our results indicate monetize monthly savings of up to ten thousand dollars and 0.5 percent reduction in loss of load probability while simultaneously reducing carbon footprints. The results of this work are directed towards large data centers at university and corporate campuses, microgrids, and military installations.
Uploads
Papers by Volker Sorger