This PDF file contains the front matter associated with SPIE Proceedings Volume 9133 including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.

The lack of potent integrated light emitters is one of the bottlenecks that have so far hindered the silicon photonics platform from revolutionizing the communication market. Photonic circuits with integrated light sources have the potential to address a wide range of applications from short-distance data communication to long-haul optical transmission. Notably, the integration of lasers would allow saving large assembly costs and reduce the footprint of optoelectronic products by combining photonic and microelectronic functionalities on a single chip. Since silicon and germanium-based sources are still in their infancy, hybrid approaches using III-V semiconductor materials are currently pursued by several research laboratories in academia as well as in industry. In this paper we review recent developments of hybrid III-V/silicon lasers and discuss the advantages and drawbacks of several integration schemes. The integration approach followed in our laboratory makes use of wafer-bonded III-V material on structured silicon-on-insulator substrates and is based on adiabatic mode transfers between silicon and III-V waveguides. We will highlight some of the most interesting results from devices such as wavelength-tunable lasers and AWG lasers. The good performance demonstrates that an efficient mode transfer can be achieved between III-V and silicon waveguides and encourages further research efforts in this direction.

A promising approach to obtain light emission from Si-based materials is represented by doping with rare earths and in particular with Eu. In this paper the comparison of the performances of SiO₂ and SiOC layers as host matrices for optically active Eu ions is presented. A SiO₂ matrix allows to observe light emission from both Eu²⁺ and Eu³⁺ ions, owing to a proper tuning of the thermal annealing process used to optically activate the rare earth. However the photoluminescence efficiency of both ions remains relatively low and quite far from the requirements for technological applications, mainly due to the extensive formation of Eu-containing precipitates. A detailed study of the structural and optical properties of these layers allowed us to analyze and elucidate the clustering process and to find suitable strategies to minimize it. We found that the substitution of the SiO₂ matrix with a SiOC film allows to obtain a very bright light emission at about 440 nm from Eu²⁺ ions. In fact SiOC is able to efficiently promote the Eu reduction; furthermore, Eu ions are characterized by an enhanced mobility and solubility in this matrix and, as a consequence, Eu precipitation is strongly reduced. Finally, by taking advantage of the dependence of the photoluminescence peak position on the Eu concentration, an intense white emission is obtained at room temperature by combining two layers with different Eu contents. Since SiOC is a material fully compatible with standard Si technology, Eu-doped SiOC layers can be considered a highly interesting candidate for applications in photonics or in solid-state lighting.

We report the fabrication and optical characterization of long, spiral-shaped erbium-doped aluminum oxide (Al₂O₃:Er³⁺) channel waveguides for achieving high overall signal amplification on a small footprint. Al₂O₃:Er³⁺ films with Er³⁺ concentrations in the range between 0.44−3.1×10²⁰ cm^-3 were deposited by reactive co-sputtering onto standard, thermally oxidized silicon substrates. Spiral-shaped waveguides were designed and structured into the films by chlorinebased reactive ion etching. In the current design, each spiral waveguide occupies an area of 1 cm². Typical background propagation losses near 1500 nm are (0.2±0.1) dB/cm. A commercially available, pigtailed diode laser at 976 nm was employed as the pump source. The erbium-doped waveguide amplifiers were characterized in the small-signal-gain regime at the peak-gain wavelength (λ = 1532 nm) of Al₂O₃:Er³⁺. A maximum of 20 dB of internal net gain was measured for a 24.5-cm-long spiral waveguide with an Er³⁺ concentration of 0.95×10²⁰ cm^-3. Similar results were obtained for a shorter spiral with an Er³⁺ concentration about twice as high. Samples with lower concentration exhibited lower gain because of insufficient pump absorption, while samples with higher concentration showed less gain because of migration-accelerated energy transfer up-conversion and, more importantly, a fast quenching process.

In-situ doping of Tb³⁺ ions in silicon oxides and oxynitrides deposited by electron-cyclotron-resonance plasma enhanced chemical-vapour (ECR-PECVD) has been performed. Oxygen and nitrogen gas flow rates were changed to produce a gradual substitution of oxygen by nitrogen in the host matrix. Bright green luminescence from as-deposited layers is observed by the naked eye under daylight conditions. Tbdoped nitrogen-rich samples showed a considerable photoluminescence (PL) enhancement compared to Tb-doped silicon oxides. An optimum layer composition for efficient Tb³⁺ excitation under non-resonant optical pumping is obtained. The combination of a low temperature treatment with bright luminescence could be instrumental for the development of light emitting devices in other platforms with more restrictive temperature requirements.

Fiber Bragg Grating or FBG sensors are gaining more and more interest in structural health monitoring of composite materials. Often, the weakest point in such a system is the ingress point of the fiber sensing chain into the composite material. For this reason we have developed a strongly miniaturized FBG interrogator unit with wireless power and data transmission, which can be incorporated in the composite structure. The interrogator is based on an arrayed waveguide grating (AWG) filter fabricated in a SOI technology, which is tailored in such a way to give large cross-talk between neighboring channels. The AWG signals are read by a linear 128 pixel InGaAs array flip-chipped on top of the Photonic Circuit (PIC). The spectrometer unit is completed with a ROIC mounted on the same substrate. The SLED and remaining electronics are integrated on a small and thin substrate and surrounded by the wireless antenna. The interrogator has an overall dimension of 100 mm diameter by max 7 mm height. The power dissipation of the electronics unit is limited to 1.5 W. The unit is capable of measuring strain values as low as 5 micro-strain.

Photonic integrated circuits on silicon substrates are fully compatible with CMOS processes. The measurements show that light coupling with insertion loss of ~2 dB between grating waveguides and optical fibers, 60 Gbps optical modulation with Vpp of less than 4 V, and 4-port optical switches with cross-talk of < -12 dB and extinction ratio of > 15 dB. Another work shows that a depletion mode PN junction can be operated as a 40 Gbps photodetector under a special reverse bias of 7 - 8 V.

Cost-effective packaging of silicon photonic devices presents a significant bottleneck to commercialization of the technology. One way of addressing this packaging challenge is to use techniques that have been developed by the electronics industry and which also benefit from the use of advanced electronics assembly equipment. Even packaging processes such as fiber coupling can benefit from this approach, along with the hybrid integration of devices such as electronic components (e.g. modulator driver integrated circuits). In this paper, we will present developments made by our group towards achieving scalable fiber and electronic packaging processes that rely on electronic assembly techniques such as flip-chip assembly. We will also provide an overview of packaged prototypes being developed within our group for telecom and sensing applications and how these packaging technologies are now being made available to users through the ePIXfab foundry service.

We have experimentally demonstrated the ability to couple an arbitrary polarization state from a fiber to the TE-mode of a single waveguide in an integrated silicon photonics circuit with an extinction ratio larger than 31 dB, measured between the output ports of the integrated photonic circuit. To achieve this we combined a 2D- grating coupler and a Mach-Zehnder Interferometer (MZI). After accounting for setup and coupling losses, for a 1 mW input into the 2D coupler, we obtain an average output power of 0.98 mW at the desired waveguide, with less than 1.2 dB variation across all input polarization states. The experiments were performed at a wavelength of 1.55 μm.

We present a new approach to planar photonic interconnects based on spatial adiabatic passage between thin ridge silicon waveguides. Our approach provides robust coupling between arbitrary pairs of well-separated waveguides across a single chip, potentially bypassing intermediate waveguides and structures. This new technique presents opportunities for waveguide routing and device topologies that cannot be achieved using traditional evanescent coupling, while remaining compatible with conventional CMOS fabrication techniques.

We demonstrate enhanced conversion efficiency (CE) and parametric amplification of optical pulses via quasiphase- matched four-wave-mixing (FWM) in long-period Bragg waveguides made of silicon. Our study is based on a rigorous theoretical model that describes optical pulse dynamics in a periodically, adiabatically modulated silicon photonic waveguide and a comprehensive set of numerical simulations of pulse interaction in such gratings. More specifically, our theoretical model takes into account all of the relevant linear and nonlinear optical effects, including free-carriers generation, two-photon absorption, and self-phase modulation, as well as modal frequency dispersion up to the fourth-order. Due to its relevance to practical applications, a key issue investigated in our work is the dependence of the efficiency of the FWM process on the waveguide parameters and the operating wavelength. In particular, our analysis suggests that by varying the waveguide width by just a few tens of nanometers the wavelengths of the phase-matched waves can be shifted by hundreds of nanometers. Our numerical simulations show also that, in the anomalous group-velocity dispersion regime, a CE enhancement of more than 20 dB, as compared to the case of a waveguide with constant width, can be easily achieved.

The linear and nonlinear optical response of SiGe waveguides in the mid-infrared are experimentally measured. By cutback measurements we find the linear losses to be less than 1.5dB/cm between 3μm and 5μm, with a record low loss of 0.5dB/cm at a wavelength of 4.75μm. By launching picosecond pulses between 3.25μm and 4.75μm into the waveguides and measuring both their self-phase modulation and nonlinear transmission we find that nonlinear losses can be significant in this wavelength range due to free-carrier absorption induced by multi-photon absorption. This should be considered when engineering SiGe photonic devices for nonlinear applications in the mid-IR.

A novel method to significantly decrease power consumption in a silicon switch based on an asymmetric Mach-Zehnder interferometer (MZI) structure is proposed and experimentally demonstrated. A radical power consumption reduction up to 50% is achieved for switching digital data at bit rates from 10 to 30Gbps with respect to a conventional switch based on a symmetric MZI. Furthermore, the broadband performance of the proposed silicon MZI comb switch is also demonstrated by transmitting a 120 Gbps DWDM data stream.

We report on the developments of Ge/SiGe quantum well (QW) waveguide modulators operating at 1.3 μm. First we studied QW structures grown on a 13-μm SiGe buffer on bulk silicon. Light was directly coupled and propagated in the active region. Using a 3-μm wide and 50-μm long modulator, an extinction ratio larger than 4 dB was obtained for a drive voltage lower than 5 V in a 15 nm wavelength range. Then simulations were performed to evaluate the performances of an integrated modulator on silicon on insulator (SOI) platform. An eigenmode expension method was used to model the vertical optical coupling between SOI waveguide and Ge/SiGe devices. It is shown that a reduction of the thickness of the buffer leads to a significant improvement in the performances (extinction ratio, insertion loss) and footprint of the waveguide-integrated devices.

We introduce a photonics-electronics convergence system and a silicon optical interposer in this system. We developed the optical components for the silicon optical interposer and achieved a high-bandwidth-density silicon optical interposer of 30 Tbps/cm² with a channel line rate of 20 Gbps.

An integrated high-resolution ratio-metric wavelength monitor (RMWM) is demonstrated on SOI platform. The device consists of a reconfigurable demultiplexing filter based on cascaded thermally tunable microring resonators (MRRs) and Ge-Si photodetectors integrated with each drop port of the MRRs. The MRRs are supposed to achieve specific resonant wavelength spacing to form the “X-type” spectral response between adjacent channels. The ratio of the two drop power between adjacent channels varies linearly with the wavelength in the “X-type” spectral range, thus the wavelength can be monitored by investigating the drop power ratio between two pre-configured resonant channels. The functional wavelength range and monitor resolution can be adjusted flexibly by thermally tuning the resonant wavelength spacing between adjacent rings, and an ultra-high resolution of 5 pm or higher is achieved while the resonant spacing is tuned to 1.2nm. By tuning the resonant wavelength of the two MRRs synchronously, the monitor can cover the whole 9.6nm free spectral range (FSR) with only two ring channels. The power consumption is as small as 8 mW/nm. We also demonstrate the multi-channel monitor that can measure multi-wavelength-channel simultaneously and cover the whole FSR by presetting the resonant wavelengths of every MRR without any additional power consumption. The improvements to increase the resolution are also discussed.

High Energy Physics experiments make extensive use of micro-strip silicon sensors for tracking purposes. However, the high granularity of the modern detectors makes the connection between the segmented sensor channels and the readout electronics very complex. Enhancing the complexity, a direct connection is not possible in most of the cases due to the mismatch between the detector pad pitch and the electronics. A new method based on laser technology is presented for the fabrication of pitch adapters. In this new method the high-density metal traces are manufactured by means of laser ablation of the metal layer deposited on top of a substrate. Glass, Kapton and Silicon substrates were metal coated and tested for the fabrication of pitch adapters. Finally, a metal-on-glass prototype has been successfully manufactured and tested for electrical conductivity, bondability and metrology. Detectors have been assembled using this pitch adapters design and tested in particle beams at CERN.

In this paper, a new method is proposed to provide an inter- and intra-chip optical interconnect at the standard telecommunication wavelength of 1550 nm. The proposed optical interconnect consists of two optical leaky- wave nano-antennas as transmitter and receiver. The leaky-wave antennas are fed through a hybrid plasmonic waveguide with low propagation loss. Since the propagation length of plasmonic waveguides is not so long, each plasmonic waveguide is coupled to a silicon waveguide through an optical coupler. In comparison with previously proposed method of optical interconnect, the most important advantage of our method is its planar structure which makes it fully integrable with the photonic integrated circuits (PIC). The Fluguet theorem and the theory of surface plasmons are used to obtain an analytical model for design purposes, and the accuracy of the proposed method is verified by a 3-dimensional full-wave numerical analysis.

Based on side-port multimode interference coupler, a novel design of 1.31/1.55-μm wavelength multiplexer/demutiplexer on SOI platform with conventional channel waveguides is proposed and analyzed by using wide-angle beam propagation method. With a 25.9μm long ultra-short MMI section, nearly an order of magnitude shorter than that of the previously reported 1.31/1.55-μm wavelength MMI splitters on SOI, simulation results exhibit contrasts of 28dB and 25dB at wavelength 1.31 and 1.55 μm, respectively, and the insertion losses are both below 0.55dB. Meanwhile, the analysis shows that the proposed structure has larger fabrication tolerances than restricted MMI based structures and the present design methodology also applies to split other wavelengths and in different material platforms, such as InP, GaAs and PLC guides, etc.

Broad adoption of silicon photonics technology for photonic integrated circuits requires standardized design flows that are similar to what is available for analog and mixed signal electrical circuit design. We have developed a design flow that combines mature electronic design automation (EDA) software with optical simulation software. An essential component of any design flow, whether electrical or photonic, is the ability to accurately simulate largescale circuits. This is particularly important when the behavior of the circuit is not trivially related to the individual component performance. While this is clearly the case for electronic circuits consisting of hundreds to billions of transistors, it is already becoming important in photonic circuits such as WDM transmitters, where signal cross talk needs to be considered, as well as optical cross-connect switches. In addition, optical routing to connect different components requires the introduction of additional waveguide sections, waveguide bends, and waveguide crossings, which affect the overall circuit performance. Manufacturing variability can also have dramatic circuit-level consequences that need to be simulated. Circuit simulations must rely on compact models that can accurately represent the behavior of each component, and the compact model parameters must be extracted from physical level simulation and experimental results. We show how large scale circuits can be simulated in both the time and frequency domains, including the effects of bidirectional and, where appropriate, multimode and multichannel photonic waveguides. We also show how active, passive and nonlinear individual components such as grating couplers, waveguides, splitters, filters, electro-optical modulators and detectors can be simulated using a combination of electrical and optical algorithms, and good agreement with experimental results can be obtained. We then show how parameters, with inclusion of fabrication process variations, can be extracted for use in the circuit level simulations. Ultimately, we show how a multi-channel WDM transceiver can be created, from schematic design to tapeout, using key features of EDA design flows such as schematic driven layout, design rule checking and layout versus schematic.

With the advancements in design processes down to the sub 7nm levels, the Electronic Design Automation industry appears to be coming to an end of advancements, as the size of the silicon atom becomes the limiting factor. Or is it? The commercial viability of mass-producing silicon photonics is bringing about the Optoelectronic Design Automation (OEDA) industry. With the science of photonics in its infancy, adding these circuits to ever-increasing complex electronic designs, will allow for new generations of advancements. Learning from the past 50 years of the EDA industry’s mistakes and missed opportunities, the photonics industry is starting with electronic standards and extending them to become photonically aware. Adapting the use of pre-existing standards into this relatively new industry will allow for easier integration into the present infrastructure and faster time to market.

In this work, the modeling of phase shifters based in PN interleaved junctions is analyzed. Three different models based on different approximations are presented in details. Comparisons with previous published experimental data are presented, as well as a comparison and a discussion on the different models.

We propose the Fourier Modal Method (FMM) as a convenient numerical tool for the design and analysis of nonlinear optical waveguides. The scope of this work includes the design of a polarization-independent nonlinear cross-slot waveguide for telecommunication applications at the wavelength of 1550 nm. The FMM method has been implemented, obeying the proper Fourier factorization rules, within a MATLAB^TM environment. The influence of the modal field intensity on the transverse refractive index distribution due to the optical Kerr effect is modeled with FMM for a propagation invariant scheme of the waveguide. The waveguide is geometrically optimized for an enhanced nonlinear light matter interaction. A silicon-inorganic hybrid material platform based on hydrogenated amorphous silicon (a-Si:H) and amorphous titanium dioxide (TiO₂) is considered for the mentioned waveguide. With the optimized design of the waveguide, the achieved value of the nonlinear waveguide parameter (γ) is 4.678 × 10⁴ W^-1Km^-1.

Full exploitation of the unique properties of high quality factor micro-optical Whispering Gallery Mode (WGM) resonators requires a controllable and robust coupling of the light to the cavity, either for fundamental investigations or even more for practical applications. Fiber tapers are ideal phase-and-mode-matched couplers and are typically used for lab demonstrations in silica based micro-resonators or in low-index crystalline disks. Prism-based coupling basically adapts to any material and offers improved robustness and reliability for the implementation of devices based on larger resonators. We present the results of our studies on alternative methods based on integrated waveguides with specific reference to the coupling to lithium niobate disk resonators. We also demonstrate efficient coupling from fiber tapers to higher order azimuthal modes in coated microspheres and for third harmonic generation in silica microspheres. We finally propose a new method based on fiber gratings for improved robustness in biosensing applications.

We report on waveguide lasers at 1064.5 nm in femtosecond laser written double cladding waveguides in Nd:GdVO₄ crystals. The core waveguide guides both TE and TM polarized modes with considerably symmetric single modal profiles, and show good transmission property (propagation loss as low as 1.0 dB/cm). Under the optical pumping at 808 nm, maximum output power up to 0.43 W of the continuous wave waveguide laser with a slope efficiency of 52.3% have been achieved, which is 21.6% and 23% higher than those of the laser generated from single depressed cladding waveguide, respectively. Furthermore, the maximum output power of the waveguide laser is 72% higher than that of the double-line waveguide.

We present a concept for a circulator that has the same bandwidth efficiency as a photonic crystal circulator but which relies on a ring resonator and thereby is experimentally much easier to realize. We achieve this by side coupling three waveguides to the ring resonator. The desired standing wave pattern which recreates the photonic crystal type circulator spectrum is realized by exciting both the clockwise and counter-clockwise traveling wave through a Bragg reflector.

This paper presents a new alignment concept for the alignment of multichannel photonic intergrated circuits (PICs) using flexible photonic waveguides on one of the PICs that are positionable by integrated micro electro mechanical system (MEMS) actuators. The concept aims for high precision and high degree of assembly process automation. The proposed concept includes pre-alignment of both PICs on a common substrate followed by fine-alignment using the on-chip flexible waveguides and MEMS functionality. This paper introduces the alignment approach and reports on the development and fabrication of suspended and mechanically flexible photonic waveguides. Single suspended waveguide beams and suspended arrays with two and four coupled parallel waveguide beams of different lengths (250 μm to 1000 μm) and different widths (18 μm to 34 μm) are designed and fabricated. After fabrication, waveguide beam fracturing is observed. The fabrication process has been extended by an additional under-etching step in order to reduce beam fracturing. The static out-of-plane deflection of the fabricated devices follows a specific profile with a dominating upward curvature resulting in a measured maximum out-of-plane deflection of 2% of the length. The beam stiffness of the fabricated devices is measured and proves to be within the available force of microactuators.

Semiconductor ring lasers are promising sources in photonic integrated circuits because they do not require cleaved facets or mirrors to form a laser cavity. In this work, we characterize the wavelength switching speed of a tunable semiconductor ring lasers using filtered optical feedback. The filtered optical feedback is realized by employing two arrayed waveguide gratings to split/recombine light into different wavelength channels. Semiconductor optical amplifiers are placed in the feedback loop in order to control the feedback of each wavelength channel independently. The wavelength switching is achieved by changing the currents injected in the semiconductor optical amplifier gates. Experimentally, we observe a wavelength transition time of 5 ns. However, we also noticed a non-negligible delay in the switching process. [ Khoder et al, IEEE Photon. Technol. Lett. 26, 520{523, 2014]. We numerically reproduce the experimental results using rate equations taking into account the effect of spontaneous emission. The simulations further elaborate on the effect of the noise strength on the wavelength transition time and the delay time.

Hybrid integration of prefabricated III-V laser diodes with sub-micrometric silicon photonic waveguides suffers from a tradeoff between alignment tolerance and coupling efficiency. In this work, we demonstrate integrated coupling devices that substantially alleviate this problem by means of a balanced distribution of the laser power between two on-chip single mode SOI waveguides. With the reported coupling devices, a horizontal misalignment of the laser is converted in a variation of the relative phase of the light coupled into the two waveguides, allowing to satisfy the reciprocity principle while maintaining a high total coupling efficiency and a balanced power splitting. The relaxed alignment tolerances facilitate passive assembly of the lasers with pick-and-place tools. The balanced splitting of the power between waveguides is particularly well suited for optical interconnects with parallel transmitters. Here, the device design is discussed for both edge couplers and grating couplers relying on similar design principles. Furthermore, experimental characterization of edge-coupling structures with a lensed fiber and a Fabry-Pérot laser is presented. These devices have been fabricated with 193nm DUV optical lithography and are compatible with mainstream CMOS technology. The edge couplers with the best horizontal misalignment exhibits an excellent 1 dB loss horizontal misalignment range of 3.8 μm with excess insertion losses below 3.1 dB (in addition to the 3dB splitting). The back-reflection induced by the device has been assessed to be below -20 dB and measured relative intensity noise is better than measured from the same laser coupled to a lensed fiber.

We describe a hybrid III-V on Silicon laser designed for low noise class-A dynamics. The laser is based on an InP active region and a passive silicon region integrated in a long laser cavity. High-Q ring resonators are used as optical filters in order to achieve single frequency operation. A fiber-coupled output power of 4.6 mW and a 55 dB side mode suppression ratio are obtained. For a pumping rate of 5.2, the hybrid laser exhibits a Relative Intensity Noise below -145 dB/Hz over a wide frequency bandwidth, from 100 MHz to 40 GHz but still suffers from some noise excess due to relaxation oscillations phenomena and side modes noise. The optimization of the laser cavity design is discussed in order to reach class-A dynamics while reducing residual noise excess.

The electrical and electroluminescence (EL) properties of Si-rich oxynitride (SRON)/SiO₂ superlattices are studied for different silicon excess and layer thicknesses. The precipitation and crystallization of the Si excess present within the SRON layers is induced by a post-deposition annealing treatment, in order to form Si nanocrystals (Si-NCs). The electrical characterization performed in dark conditions allowed for deducing the charge transport mechanism through the superlattice structure, found to follow the Poole-Frenkel law. In addition, the EL investigation revealed the correlation between EL excitation and transport mechanisms, suggesting that impact ionization of high-energy conduction electrons dominates the whole frame. The reduction of the SiO₂ barrier thickness and the increase in the Si excess were found to enhance the carrier transport through the superlattices due to the reduction of the electrons mean free path, which, in turn, modifies the EL properties.

We present a method to characterize the temperature dynamics of miniaturized thermal IR sources. The method circumvents the limitations of current IR photodetectors, by relying only on an electrical measurement rather than on optical detection. Thus, it enables the characterization of the light emission of IR sources over their full operation frequency range. Moreover, we develop a model of thermal IR sources allowing simulations of their thermal and electrical behavior. By combining measurements and modeling, we achieve a comprehensive characterization of a Pt nanowire IR source: the reference resistance R₀ = 17.7Ω, the TCR α = 2.0 × 10^-3 K^-1, the thermal mass C = 2.7 × 10^-14 J/K, and the thermal conductance G = 1.3 × 10^-6 W/K. The thermal time constant could not be measured, because of the frequency limitation of our setup. However, the operation of the source has been tested and proved to function up to 1 MHz, indicating that the thermal time constant of the source is smaller than 1 μs.

Low temperature PECVD silicon nitride photonic waveguides have been fabricated by both electron beam lithography and 200 mm DUV lithography. Propagation losses and bend losses were both measured at 532 and 900 nm wavelength, revealing sub 1dB/cm propagation losses for cladded waveguides at both wavelengths for single mode operation. Without cladding, propagation losses were measured to be in the 1-3 dB range for 532 nm and remain below 1 dB/cm for 900 nm for single mode waveguides. Bend losses were measured for 532 nm and were well below 0.1 dB per 90 degree bend for radii larger than 10 μm.

A fully polymer slot Young interferometer operating at 633 nm wavelength was fabricated by using nanoimprint molding method. The phase response of the interference pattern was measured with several concentrations of glucose-water solutions, utilizing both TE and TM polarization states. The sensor was experimentally found to detect a bulk refractive index change of 6.4×10^-6 RIU. Temperature dependency of silicon slot waveguide has been demonstrated to be reduced with composite slot waveguide structure. The slot filled with thermally stable polymer having negative thermo-optic coefficient showed nearly an athermal operation of silicon slot waveguide. Experimental results show that the slot waveguide geometry covered with Ormocomp has thermo-optical coefficient of 6 pm/K.

Recently, the photonics community has a renewed attention for silicon nitride.^1-3 When deposited at temperatures below 650K with plasma-enhanced chemical vapor deposition (PECVD),⁴ it enables photonic circuits fabricated on-top of standard complementary metaloxidesemiconductor (CMOS) electronics. Silicon nitride is moreover transparent to wavelengths that are visible to the human eye and detectable with available silicon detectors, thus offering a photonics platform for a range of applications that is not accessible with the popular silicon-on-insulator platform. However, first-time-right design of large-scale circuits for demanding specifications requires reliable models of the basic photonic building blocks, like evanescent couplers (Figure 1), components that couple power between multiple waveguides. While these models typically exist for the silicon-on-insulator platform, they still lack maturity for the emerging silicon nitride platform. Therefore, we meticulously studied silicon nitride-based evanescent couplers fabricated in our 200mm-wafer facility. We produced the structures in a silicon nitride film deposited with low-temperature PECVD, and patterned it using optical lithography at a wavelength of 193nm and reactive ion etching. We measured the performance of as much as 250 different designs at 532nm wavelength, a central wavelength in the visible range for which laser sources are widespread. For each design, we measured the progressive transmission of up-to 10 cascaded identical couplers (Figure 2(a)), yielding very accurate figures for the coupling factor (Figure 2(b)). This paper presents the trends extracted from this vast data set (Figure 3), and elaborates on the impact of the couplers bend radius and gap on its coupling factors (Figure 4 and Figure 5). We think that the large- scale characterization of evanescent couplers presented in this paper, in excellent agreement with the simulated performance of the devices, forms the basis for a component library that enables accurate design of silicon nitride-based photonic circuitry.

It is well known that the main problem in the Y-branch splitting approach is the processing of the branching point where two waveguides start to separate. This is technologically very difficult; leading generally to an asymmetric splitting ratio causing non-uniformity of the split power over all the output waveguides. In this work we show that not only processing of branching points influences strongly splitting properties of the device but also the used waveguide structure itself. The standard low index waveguides have usually size of 6 μm x 6 μm ensuring on one side small coupling loses between fibers and waveguides and on the other side supporting mainly the single mode light propagation. However, our simulations showed that in the standard 6 μm x 6 μm waveguides is the presence of the first mode already so strong that it causes additional asymmetric splitting of the optical signals. To suppress the presence of the first mode we reduced the waveguide core size from 6 μm x 6 μm to 5.5 μm x 5.5 μm and 5 μm x 5 μm and this way were able to improve the uniformity of the split power over all the output waveguides by factor 3. Additionally, based on these results we were also able to reduce the size of the designed Y-branch to the half.

Several applications in integrated optics require an equal distribution of power from a single input port among many photonic components, whether they be projection components or sensors. One method of achieving such a system is through using progressively more tightly coupled evanescent couplers to route power from a single feeding line [1]. While very compact, this approach requires careful design and characterization of evanescent couplers, and is vulnerable to process variations as the ratio of coupling has a non-linear relation to the couplers’ gap size. Fractals, widely present in nature, are recursive objects where each section is geometrically similar to its parent. They find applications in various fields [2], including RF antenna design and feeding [3]. In this paper we propose to use the fractal approach for spreading power evenly over an area using micro-machined photonic waveguides. In the fractal routing demonstrated in this work, an 1×2 multimode interference (MMI) coupler splits the power at each fractal stage. This provides several advantages. First, only one power splitter design is needed. Second, MMI couplers are well known, and more robust to process tolerances than evanescent couplers [3]. Third, they are symmetrical, and therefore provide a theoretically perfect power distribution independent of the fractal depth. We therefore demonstrate that a fractal routing provides a way to evenly and efficiently distribute power over a large area.

An efficient split-step time-domain modeling method is developed for the simulation of pulse propagation characteristics through all-pass and add/drop filters made of ring resonators. The bus and ring waveguides are divided into several sections of equal size and the phase and coupling is updated during each time step corresponding to the propagation time over a section. The pulse propagation dependence on the coupling ratio of the all-pass and add/drop filters is investigated. It is shown that very large scale ring resonator devices such as cascaded add/drop filters can be very efficiently modeled in time domain.