Search

The performance of silicon photonic components and integrated circuits has improved dramatically in recent years. As a key enabler, heterogeneous integration not only provides the optical gain which is absent from native Si substrates and enables complete photonic functionalities on chip, but also lays the foundation of versatile integrated photonic device performance engineering. This paper reviews recent progress of high-performance silicon photonics using heterogeneous integration, with emphasis on ultra-low-loss waveguides, single-wavelength lasers, comb lasers, and photonic integrated circuits including optical phased arrays for LiDAR and optical transceivers for datacenter interconnects.

Normal group velocity dispersion (GVD) microcombs offer high comb line power and high pumping efficiency compared to bright pulse microcombs. The recent demonstration of normal GVD microcombs using CMOS-foundry-produced microresonators is an important step towards scalable production. However, the chromatic dispersion of CMOS devices is large and impairs generation of broadband microcombs. Here, we report the development of a microresonator in which GVD is reduced due to a couple-ring resonator configuration. Operating in the turnkey self-injection-locking mode, the resonator is hybridly integrated with a semiconductor laser pump to produce high-power-efficiency combs spanning a bandwidth of 9.9 nm (1.22 THz) centered at 1560 nm, corresponding to 62 comb lines. Fast, linear optical sampling of the comb waveform is used to observe the rich set of near-zero GVD comb behaviors, including soliton molecules, switching waves (platicons) and their hybrids. Tuning of the 20 GHz repetition rate by electrical actuation enables servo locking to a microwave reference, which simultaneously stabilizes the comb repetition rate, offset frequency and temporal waveform. This hybridly integrated system could be used in coherent communications or for ultra-stable microwave signal generation by two-point optical frequency division., 8 pages, 4 figures

Silicon nitride (SiN) waveguides with ultra-low optical loss enable integrated photonic applications including low noise, narrow linewidth lasers, chip-scale nonlinear photonics, and microwave photonics. Lasers are key components to SiN photonic integrated circuits (PICs), but are difficult to fully integrate with low-index SiN waveguides due to their large mismatch with the high-index III-V gain materials. The recent demonstration of multilayer heterogeneous integration provides a practical solution and enabled the first-generation of lasers fully integrated with SiN waveguides. However, a laser with high device yield and high output power at telecommunication wavelengths, where photonics applications are clustered, is still missing, hindered by large mode transition loss, non-optimized cavity design, and a complicated fabrication process. Here, we report high-performance lasers on SiN with tens of milliwatts output power through the SiN waveguide and sub-kHz fundamental linewidth, addressing all the aforementioned issues. We also show Hertz-level fundamental linewidth lasers are achievable with the developed integration techniques. These lasers, together with high-Q SiN resonators, mark a milestone towards a fully integrated low-noise silicon nitride photonics platform. This laser should find potential applications in LIDAR, microwave photonics and coherent optical communications., Achieving high output power and low noise integrated lasers is a major challenge. Here the authors experimentally demonstrate integrated lasers from a Si/SiN heterogeneous platform that shows Hertz-level linewidth, paving the way toward fully integrating low-noise silicon nitride photonics in volume using real devices for lasing.

Lasers with hertz linewidths at time scales of seconds are critical for metrology, timekeeping, and manipulation of quantum systems. Such frequency stability relies on bulk-optic lasers and reference cavities, where increased size is leveraged to reduce noise but with the trade-off of cost, hand assembly, and limited applications. Alternatively, planar waveguide–based lasers enjoy complementary metal-oxide semiconductor scalability yet are fundamentally limited from achieving hertz linewidths by stochastic noise and thermal sensitivity. In this work, we demonstrate a laser system with a 1-s linewidth of 1.1 Hz and fractional frequency instability below 10 −14 to 1 s. This low-noise performance leverages integrated lasers together with an 8-ml vacuum-gap cavity using microfabricated mirrors. All critical components are lithographically defined on planar substrates, holding potential for high-volume manufacturing. Consequently, this work provides an important advance toward compact lasers with hertz linewidths for portable optical clocks, radio frequency photonic oscillators, and related communication and navigation systems.

Narrow-linewidth lasers are important to many applications spanning precision metrology to sensing systems. Characterization of these lasers requires precise measurements of their frequency noise spectra. Here we demonstrate a correlated self-heterodyne (COSH) method capable of measuring frequency noise as low as 0.01 Hz$^2$/Hz at 1 MHz offset frequency. The measurement setup is characterized by both commercial and lab-built lasers, and features low optical power requirements, fast acquisition time and high intensity noise rejection., 10 pages, 6 figures

Driven by narrow-linewidth bench-top lasers, coherent optical systems spanning optical communications, metrology and sensing provide unrivalled performance. To transfer these capabilities from the laboratory to the real world, a key missing ingredient is a mass-produced integrated laser with superior coherence. Here, we bridge conventional semiconductor lasers and coherent optical systems using CMOS-foundry-fabricated microresonators with record high $Q$ factor over 260 million and finesse over 42,000. Five orders-of-magnitude noise reduction in the pump laser is demonstrated, and for the first time, fundamental noise below 1 Hz$^2$ Hz$^{-1}$ is achieved in an electrically-pumped integrated laser. Moreover, the same configuration is shown to relieve dispersion requirements for microcomb generation that have handicapped certain nonlinear platforms. The simultaneous realization of record-high $Q$ factor, highly coherent lasers and frequency combs using foundry-based technologies paves the way for volume manufacturing of a wide range of coherent optical systems., Comment: 19 pages, 11 figures

A semiconductor laser is self-injection-locked to a spiral, ultra-high-Q silicon-nitride resonator. The laser attains record-low frequency noise noise floor (0.006 Hz2/Hz or 40 mHz white-noise-linewidth) with stability at low-offset frequencies comparable to high-performance fiber lasers.

We demonstrated mode-hop-free tuning over 225 GHz (1.8 nm) in an integrated external-cavity diode laser operating around 1550 nm and with an intrinsic linewidth of 4.7 kHz.

A conventional semiconductor DFB laser is self-injection-locked to a CMOS-foundry- fabricated ultra-high Q silicon nitride microresonator, suppressing high-offset frequency noise to 0.2 Hz2 Hz-1 and yielding instantaneous linewidth of 1.2 Hz.

A Correction to this paper has been published: https://doi.org/10.1038/s41467-021-22031-4

We propose a model for understanding dark solitons formation dynamics in self-injection-locking laser-resonator systems using dynamical instabilities and domain walls. Snapshots of solitons are captured with dual-comb imaging techniques and validate the model.

The use of silicon nitride in integrated photonics has rapidly progressed in recent decades. Ultra-low-loss waveguides based on silicon nitride are a favorable platform for the research of nonlinear and microwave photonics and their application to a wide variety of fields, including precision metrology, communications, sensing, imaging, navigation, computation, and quantum physics. In recent years, the integration of Si and III-V materials has enabled new large-scale, advanced silicon nitride-based photonic integrated circuits with versatile functionality. In this perspective article, we review current trends and the state-of-the-art in silicon nitride-based photonic devices and circuits. We highlight the hybrid and heterogeneous integration of III-V with silicon nitride for electrically pumped soliton microcomb generation and ultra-low-noise lasers with fundamental linewidths in the tens of mHz range. We also discuss several ultimate limits and challenges of silicon nitride-based photonic device performance and provide routes and prospects for future development.

Within the context of globalization, the development of renewable energy is critical for attaining sustainable development, and the digital economy is also a critical driver for achieving it. This article incorporates globalization, renewable energy development, and the digital economy into its research framework, investigates the relationship between globalization and renewable energy development, and explores the moderating effect of the digital economy, using panel data from countries along the Belt and Road (B&R) from 2001 to 2018. It is found that globalization facilitates the development of renewable energy. The 1% increase in globalization results in a 1.06% increase in renewable energy development; the level of globalization has a significant effect on renewable energy development in high-income countries, upper-middle-income countries, and low-income countries, but not in lower-middle-income countries; the digital economy has a moderating effect on the impact of globalization on renewable energy development in countries along the B&R. Simultaneously, the effect of globalization on renewable energy development in B&R countries is influenced by the digital economy’s single threshold effect, and the effect of globalization on renewable energy development is more pronounced when the level of digital economy development is less than the threshold of 0.061. The conclusions of this article have significant implications for the B&R countries’ sustainable development in the contexts of globalization and the digital economy.

Publisher's Note: This paper, originally published on 2 July 2019, was replaced with a corrected/revised version on 3 August 2020. If you downloaded the original PDF but are unable to access the revision, please contact SPIE Digital Library Customer Service for assistance.

We demonstrate the first heterogeneously integrated laser with a Si3N4 external cavity. Through a multilayer heterogeneous integration with InP and Si, the laser shows narrow linewidth and high temperature stability expected from a Si3N4 cavity.

We observe stimulated Brillouin scattering (SBS) in AlGaAs-on-insulator integrated waveguides. A guided transverse acoustic mode has a 12.3 GHz Brillouin shift, and a full-width half-maximum of 25 MHz.

We demonstrate an efficient RF photonics filter based on AlGaAs-on-insulator integrated comb. With a record low on-chip pumping power of —20 mW, the reconfigurable filter achieves a main-to-secondary sidelobe ratio of > 25 dB.

Mainstream silicon photonic integrated circuits are based on compact and low-loss silicon-on-insulator (SOI) waveguide platforms. However, monolithic SOI-based photonics provides only a limited number of functional device types. Here, to extend the on-chip capabilities, we propose a general heterogeneous integration approach to embed highly nonlinear III-V (AlGaAs) photonics into the SOI platform. We develop low-loss AlGaAs-on-SOI photonic circuits with integrated Si waveguides and showcase sub-milliwatt-threshold ( ∼ 0.25 mW ) Kerr frequency comb generation in ultrahigh- Q AlGaAs microrings ( Q over 10 6 ) at the telecom bands. Our demonstration complements existing mature Si photonics technology with efficient nonlinear functionalities provided by III-V and propels conventional Si photonics into emerging nonlinear photonic applications towards fully chip-based nonlinear engines.

We self-injection-lock a diode laser to a 1.41 m long, ultra-high Q integrated resonator. The hybrid integrated laser reaches a frequency noise floor of 0.006 H z 2 / H z at 4 MHz offset, corresponding to a Lorentzian linewidth below 40 mHz—a record among semiconductor lasers. It also exhibits exceptional stability at low-offset frequencies, with frequency noise of 200 H z 2 / H z at 100 Hz offset. Such performance, realized in a system comprised entirely of integrated photonic chips, marks a milestone in the development of integrated photonics; and, for the first time, to the best of our knowledge, exceeds the frequency noise performance of commercially available, high-performance fiber lasers.

Silicon photonics is a truly disruptive technology that has grown tremendously in the past two decades. It has now become a recognized leader in the next generation of consumer products. These future products require realizing complex optical functionalities—chief among which is an on-chip laser. Low noise and wide tuning range are the two main important characteristics of an on-chip laser as the components required to achieve these characteristics constitute the bulk of current benchtop lasers. In this paper, we review the impact of silicon photonic chips on dramatically improving the tuning range and the noise characteristics. Two different types of technologies are reviewed— heterogeneous integration by wafer bonding and hybrid integration by butt-coupling. We cover the two most important wavelength bands for optical communications—O-band for data communication applications and C-band for telecom applications.

While Moore’s law predicted shrinking transistors would enable exponential scaling of electronic circuits, the footprint of photonic components is limited by the wavelength of light. Thus, future high-complexity photonic integrated circuits (PICs) such as petabit-per-second transceivers, thousand-channel switches, and photonic quantum computers will require more area than a single reticle provides. In our novel approach, we overlay and widen waveguides in adjacent reticles to stitch a smooth transition between misaligned exposures. In SiN waveguides, we measure ultralow loss of 0.0004 dB per stitch, and produce a stitched delay line 23 m in length. We extend the design to silicon channel waveguides, and predict 50-fold lower loss or 50-fold smaller footprint versus a multimode-waveguide-based method. Our approach enables large-scale PICs to scale seamlessly beyond the single-reticle limit.

Photonic network-on-chip (PNoC) is a promising candidate to replace traditional electrical NoC in manycore systems that require substantial bandwidths. The photonic links in the PNoC comprise laser sources, optical ring resonators, passive waveguides, and photodetectors. Reliable link operation requires laser sources and ring resonators to have matching optical frequencies. However, inherent thermal sensitivity of photonic devices and manufacturing process variations can lead to a frequency mismatch. To avoid this mismatch, micro-heaters are used for thermal trimming and tuning, which can dissipate a significant amount of power. This paper proposes a novel FreqAlign workload allocation policy, accompanying an adaptive frequency tuning ( AFT ) policy, that is capable of reducing thermal tuning power of PNoC. FreqAlign uses thread allocation and thread migration to control temperature for matching the optical frequencies of ring resonators in each photonic link. The AFT policy reduces the remaining optical frequency difference among ring resonators and corresponding on-chip laser sources by hardware tuning methods. We use a full modeling stack of a PNoC that includes a performance simulator, a power simulator, and a thermal simulator with a temperature-dependent laser source power model to design and evaluate our proposed policies. Our experimental results demonstrate that FreqAlign reduces the resonant frequency gradient between ring resonators by 50%–60% when compared to existing workload allocation policies. Coupled with AFT , FreqAlign reduces localized thermal tuning power by 19.28 W on average, and is capable of saving up to 34.57 W when running realistic loads in a 256-core system without any performance degradation.

In this paper, we propose a novel data representation-classification model learning algorithm. The model is a convolutional neural network (CNN), and we learn its parameters to achieve A-optimality. The input multi-instance data are represented by a CNN model, and then classified by a linear classification model. The A-optimality of a classification model is measured by the trace of the covariance matrix of the model parameter vector. To achieve the A-optimality of the CNN model, we minimize the classification errors and a regularization term to present the classification model parameter as a function of the CNN filter parameters, and minimize its trace of the covariance matrix. We show that the minimization problem can be solved easily by transferring it to another coupled minimization problem. In the experiments over benchmark data sets of molecular, image, and seismic waveform, we show the advantages of the proposed A-optimal CNN model.

Silicon nitride (Si3N4), as a complementary metal-oxide-semiconductor (CMOS) material, finds wide use in modern integrated circuit (IC) technology. The past decade has witnessed tremendous development of Si3N4 in photonic areas, with innovations in nonlinear photonics, optical sensing, etc. However, the lack of an integrated laser with high performance prohibits the large-scale integration of Si3N4 waveguides into complex photonic integrated circuits (PICs). Here, we demonstrate a novel III-V/Si/Si3N4 structure to enable efficient electrically pumped lasing in a Si3N4 based laser external cavity. The laser shows superior temperature stability and low phase noise compared with lasers purely dependent on semiconductors. Beyond this, the demonstrated multilayer heterogeneous integration provides a practical path to incorporate efficient optical gain with various low-refractive-index materials. Multilayer heterogeneous integration could extend the capabilities of semiconductor lasers to improve performance and enable a new class of devices such as integrated optical clocks and optical gyroscopes., 14 pages, 3 figures

Recent advances in nonlinear optics have revolutionized integrated photonics, providing on-chip solutions to a wide range of new applications. Currently, state of the art integrated nonlinear photonic devices are mainly based on dielectric material platforms, such as Si3N4 and SiO2. While semiconductor materials feature much higher nonlinear coefficients and convenience in active integration, they have suffered from high waveguide losses that prevent the realization of efficient nonlinear processes on-chip. Here, we challenge this status quo and demonstrate a low loss AlGaAs-on-insulator platform with anomalous dispersion and quality (Q) factors beyond 1.5 × 106. Such a high quality factor, combined with high nonlinear coefficient and small mode volume, enabled us to demonstrate a Kerr frequency comb threshold of only ∼36 µW in a resonator with a 1 THz free spectral range, ∼100 times lower compared to that in previous semiconductor platforms. Moreover, combs with broad spans (>250 nm) have been generated with a pump power of ∼300 µW, which is lower than the threshold power of state-of the-art dielectric micro combs. A soliton-step transition has also been observed for the first time in an AlGaAs resonator., Despite larger nonlinear coefficients, waveguide losses have prevented using semiconductors instead of dielectric materials for on-chip frequency-comb sources. By significantly reducing waveguide loss, ultra-low-threshold Kerr comb generation is demonstrated in a high-Q AlGaAs-on-insulator microresonator system.

In this Letter, we demonstrate a low loss gallium arsenide and aluminum gallium arsenide on an insulator platform by heterogenous integration. The resonators on this platform exhibit record high quality factors up to 1.5×106, corresponding to a propagation loss ∼0.4 dB/cm. For the first time, to the best of our knowledge, the loss of integrated III–V semiconductor on insulator waveguides becomes comparable with that of the silicon-on-insulator waveguides. This Letter should have a significant impact on photonic integrated circuits (PICs) and become an essential building block for the evolving nonlinear PICs and integrated quantum photonic systems in the future.

We have developed three light-based educational outreach activities targeted towards pre-university students, emphasizing experiential hands-on components for core learning via challenges the students must complete. These activities leverage photonics concepts from two active research areas at the Univ. of California Santa Barbara: integrated optics and solid-state lighting. The activities center on (1) building a free-space optical link, (2) subtractive and additive color mixing, and (3) guiding light using gelatin “waveguides”. These activities are self-contained that is, the necessary background and intuition are introduced and built, respectively, before culminating in the main demonstration. The color-mixing and gelatin waveguide activities were designed for middle school students (ages 10-13) and their families, while the free-space optical link activity was designed for high school students (ages 14-18). Graduate students, not necessarily in photonics or optics, typically lead these activities for groups of 20-30 students after an initial training. We have found that we are able to garner considerable excitement from students when activities culminate in a challenge, especially if it has a competitive nature. This allows leaders to emphasize important practices in scientific research, such as: using the success of others’ experiments to one’s benefit, making informed hypotheses and testing them, persistence in understanding and solving a problem, and finally, the desire to improve upon a working solution.

Incomplete data are quite common which can deteriorate statistical inference, often affecting evidence-based policymaking. A typical example is the Business Longitudinal Analysis Data Environment (BLADE), an Australian Government’s national data asset. In this paper, motivated by helping BLADE practitioners select and implement advanced imputation methods with a solid understanding of the impact different methods will have on data accuracy and reliability, we implement and examine performance of data imputation techniques based on 12 machine learning algorithms. They range from linear regression to neural networks. We compare the performance of these algorithms and assess the impact of various settings, including the number of input features and the length of time spans. To examine generalisability, we also impute two features with distinct characteristics. Experimental results show that three ensemble algorithms: extra trees regressor, bagging regressor and random forest consistently maintain high imputation performance over the benchmark linear regression across a range of performance metrics. Among them, we would recommend the extra trees regressor for its accuracy and computational efficiency.

We demonstrate a CMOS-foundry-based S i 3 N 4 photonic platform at blue and violet wavelengths that exhibits record-high intrinsic Qs of around 6 M at 453 nm and < 1 d B / c m waveguide propagation loss of around 405 nm.

Silicon nitride (Si3N4) is a widely used complementary metal–oxide–semiconductor (CMOS) compatible material in modern IC industry and recent years have witnessed tremendous development of Si3N4 in photonics. The benefits of Si3N4 in photonics include low loss, low thermo optic coefficient and negligible nonlinear absorption at telecom wavelengths. Devices based on Si3N4 waveguides innovates in the field of nonlinear photonics, sensing, microwave photonics and so on. However, they are all dependent on the excellent passive properties of Si3N4. There are no successful demonstrations of fully integrated laser with Si3N4 yet. For an external cavity laser design, it is beneficial to have a low loss waveguide. A low loss waveguide platform can enable a long optical cavity which can greatly reduce the laser phase noise. Butt-coupled hybrid integrated lasers are extensively studied using Si3N4 cavity, but they are based on separate chips and are not scalable for large volume integration. Heterogeneous integration of III-V material and Si has achieved great success in the past years, offering the silicon photonics industry with low-cost, high-performance silicon photonic integrated circuits fully integrated with lasers. For the heterogeneous integration of III-V material with Si3N4, the biggest challenge is the large index mismatch. Typical III-V epi used for heterogeneous lasers is about 2 μm thick and extreme tapering cannot facilitate a high coupling efficiency between the III-V layer and Si3N4 layer. In this work, we used an intermediate Si layer to bridge the index mismatch between the III-V and Si3N4 layer. By using multilayer heterogeneous integration shown in Fig.1 (a), we enable a III-V/Si/Si3N4 structure to efficiently couple the mode between III-V/Si and Si/Si3N4. Based on this structure, we designed the III-V/Si/Si3N4 laser using an external Si3N4 spiral grating as the feedback cavity. This spiral grating provides the narrow-band optical feedback and determines the laser wavelength. The laser configuration is shown in Fig. 1(b). The laser performance is summarized in Fig. 1 (c)-(e). The lasing wavelength is at 1546 nm and the laser has over 58 dB side mode suppression ratio (SMSR). This shows the excellent mode selectivity from the 20 mm long Si3N4 spiral grating. We measured the frequency noise of the laser. With optimized loop mirror reflectivity, the laser fundamental linewidth is about 4 KHz. This linewidth can be further improved by lowering the Si3N4 waveguide loss. We anticipate the laser fundamental linewidth can be reduced to below 100 Hz using this design. Another advantage of integrating Si3N4 into the laser cavity is to have much higher temperature stability. By comparison, III-V/Si/Si3N4 laser wavelength shifts by only 0.47 nm for 45 oC change, while for a III-V/Si DBR laser this wavelength shift is 3.3 nm. This is over 7x difference and is a result of the low thermo-optic coefficient of Si3N4. When combined with Si3N4 based arrayed waveguide gratings, our laser can be potentially used in dense wavelength-division-multiplexing (DWDM) systems. The successful demonstration of this multilayer heterogeneous integration will enable a whole new class of devices that require low loss waveguides and are presently not integrated. This is a key step in fully exploiting the capabilities of wafer-bonding technologies. We anticipate this platform can take narrow-linewidth semiconductor laser performance to a new level. Figure 1

High-Q Si ring resonators play an important role in the development of widely tunable heterogeneously integrated lasers. However, while a high Q-factor (Q > 1 million) is important for ring resonators in a laser cavity, the parasitic high-power density in a Si resonator can deteriorate the laser performance at high power levels due to nonlinear loss. Here, we experimentally show that this detrimental effect can happen at moderate power levels (a few milliwatts) where typical heterogeneously integrated lasers work. We further compare different ring resonators, including extended Si ring resonators and Si3N4 ring resonators and provide practical approaches to minimize this effect. Our results provide explanations and guidelines for high-Q ring resonator designs in heterogeneously integrated tunable lasers, and they are also applicable for hybrid integrated butt-coupled lasers.

Ultra-low-loss waveguide fabrication typically requires high-temperature annealing beyond 1000°C to reduce the hydrogen content in deposited dielectric films. However, realizing the full potential of an ultra-low loss will require the integration of active materials that cannot tolerate high temperature. Uniting ultra-low-loss waveguides with on-chip sources, modulators, and detectors will require a low-temperature, low-loss dielectric to serve as a passivation and spacer layers for complex fabrication processes. We report a 250°C deuterated silicon dioxide film for top cladding in ultra-low-loss waveguides. Using multiple techniques, we measure propagation loss below 12 dB/m for the entire 1200–1650 nm range and top-cladding material absorption below 1 dB/m in the S, C, and L bands.

A novel piezoelectric actuator is fabricated on a suspended silicon nitride ring resonator to tune the optical resonance via mechanical deformation. Fabricated devices exhibit tuning across a full free spectral range by applying 16 V across the piezoelectric.

A 3D integrated hybrid silicon photonic external cavity laser was realized experimentally. The laser was characterized for thermal performance and demonstrates promising characteristics for high-temperature and reliable operation.

A 3D integrated hybrid silicon photonic external cavity laser was realized experimentally. Continuous wave lasing was demonstrated with a side mode suppression ratio of 30 dB and optical output power up to 1.4 mW.

Typical integrated optical phase tuners alter the effective index. In this paper, we explore tuning by geometric deformation. We show that tuning efficiency, Vπ L, improves as the device size shrinks down to the optimal bend radius, contrary to conventional index-shift based approaches where Vπ L remains constant. We demonstrate that this approach is capable of ultra-low power tuning across a full FSR in a low-confinement silicon nitride based ring resonator of 580 μm radius. We demonstrate record performance with VFSR = 16 V, Vπ L = 3.6 V dB, Vπ Lα = 1.1 V dB, tuning current below 10 nA, and unattenuated tuning response up to 1 MHz. We also present optimized designs for high confinement silicon nitride and silicon based platforms with radius down to 80 μm and 45 μm, respectively, with performance well beyond current state-of-the-art. Applications include narrow-linewidth tunable diode lasers for spectroscopy and non-linear optics, optical phased array beamforming networks for RF antennas and LIDAR, and optical filters for WDM telecommunication links.

Nanoscale circuits operating at sub-threshold voltages are affected by growing impact of random telegraph signal (RTS) and thermal noise. Given the low operational voltages and subsequently lower noise margins, these noise phenomena are capable of changing the value of some of the nodes in the circuit, compromising the reliability of the computation. We propose a method for improving noise-tolerance by selectively applying feed-forward reinforcement to circuits based on use of existing invariant relationships. As reinforcement mechanism, we used a modification of the standard CMOS gates based on the Schmitt trigger circuit. SPICE simulations show our solution offers better noise immunity than both standard CMOS and fully reinforced circuits, with limited area and power overhead.

Many-core chip architectures are now feasible, but the power consumption of electrical networks-on-chip does not scale well. Silicon photonic NoCs (PNoCs) are more scalable and power efficient, but floorplan optimization is challenging. Prior work optimizes PNoC floorplans through simultaneous place and route, but does not address cross-layer effects that span optical and electrical boundaries, chip thermal profiles, or effects of job scheduling policies. This paper proposes a more comprehensive, cross-layer optimization of the silicon PNoC and core cluster floorplan. Our simultaneous placement (locations of router groups and core clusters) and routing (waveguide layout) considers scheduling policy, thermal tuning, and heterogeneity in chip power profiles. The core of our optimizer is a mixed-integer linear programming formulation that minimizes NoC power, including (1) laser source power due to propagation, bend and crossing losses; (2) electrical and electrical-optical-electrical conversion power; and (3) thermal tuning power. Our experiments vary numbers of cores, optical data rate per wavelength, number of waveguides and other parameters to investigate scalability and tradeoffs through a large design space. We demonstrate how the optimal floorplan changes with cross-layer awareness: metrics of interest such as optimal waveguide length or thermal tuning power change significantly (up to 4X) based on power and utilization levels of cores, chip and cluster aspect ratio, and laser source sharing mechanism. Exploration of a large solution space is achieved with reasonable runtimes, and is perfectly parallelizable. Our optimizer thus affords designers with more accurate, cross-layer chip planning decision support to accelerate adoption of PNoC-based solutions.