Photonic applications based on planar waveguide technology impose stringent requirements on properties such as optical propagation losses, light coupling to optical fibers, integration density, as well as on reliability and reproducibility. The latter is correlated to a high level of control of the refractive index and waveguide geometry. In this paper, we review a versatile dielectric waveguide platform, called TriPleX, which is based on alternating silicon nitride and silicon dioxide films. Fabrication with CMOS-compatible equipment based on low-pressure chemical vapor deposition enables the realization of stable material compositions being a prerequisite to the control of waveguide properties and modal shape. The transparency window of both materials allows for the realization of low-loss waveguides over a wide wavelength range (400 nm–2.35 μm). Propagation losses as low as 5×10-4 dB/cm are reported. Three basic geometries (box shell, double stripe, and filled box) can be distinguished. A specific tapering technology is developed for on-chip, low-loss (<0.1 dB) spotsize convertors, allowing for combining efficient fiber to chip coupling with high-contrast waveguides required for increased functional complexity as well as for hybrid integration with other photonic platforms such as InP and SOI. The functionality of the TriPleX platform is captured by verified basic building blocks. The corresponding library and associated design kit is available for multi-project wafer (MPW) runs. Several applications of this platform technology in communications, biomedicine, sensing, as well as a few special fields of photonics are treated in more detail.
Over the past decades integrated optics solutions for communications and sensing have rapidly developed and became increasingly important for the telecommunication market, space applications, in environmental control, as well as healthcare and life science [1–3].
In general, the applied photonics technology has to fulfill requirements such as low optical propagation losses, efficient light coupling to optical fibers, high integration density, low cost, manufacturability, and combining excellent performance of passive photonic circuitry with access to active functionality for light generation, detection, and modulation. However, the quantification of these demands in terms of transmission loss limits, applicable wavelengths ranges, cost price, and manufacturing scale are rather diverse for the different application fields.
For the telecommunication market, the ever-increasing demand for capacity can be fulfilled by the steadily developing optical fiber networks. Consequently, similar performance requirements rise for switching and routing functionalities leading to the replacements of electronics by optical components for signal processing [4–6]. Examples of integrated photonic solutions are demonstrated in microwave photonics (MWP) [7, 8], network components for fiber-to-the-home (FTTH) [9, 10], and datacom applications [11–13]. For compatibility with the existing fiber backbone, Telecom and FTTH applications require wavelength ranges at 1.3–1.55 μm, while for datacom applications, the availability of low-cost, high-speed modulating VCSELs is of importance. Operating wavelengths for the latter application field are either 850 nm (multimode fiber links in rack to rack connections) or 1.3 μm (single-mode fiber connections in large-scale datacenters). Besides performance and reliability, cost per bit of transmitted data and potential for large-scale manufacturability are key requirements of this market segment. To date, the breakthrough of photonic components is being hampered, however, by high cost due to large optical footprint (i.e., chip real estate) and costly packaging solutions.
In the field of optical sensing and biomedical applications, mainly the potential for miniaturization and sensitivity improvement compared to state-of-the art systems adds to the attractiveness of integrated photonic solutions [14, 15]. Nowadays, waveguide-based optical sensors reach the detection limit of 10-7/refractive index unit (RIU) , which covers most interesting applications. Current research focuses on creating sensor arrays and cost price reduction by further miniaturization . Recent progress in miniaturization of spectrometers for Raman detection and optical coherence tomography (OCT) [17–19] demonstrates the potential of photonics technology for point-of-care diagnostics. The requirements and technology needs for these application fields are, however, hard to generalize. The applicable wavelengths, for example, cover the spectrum of light ranging from UV-VIS for environmental monitoring and absorption spectroscopy, from VIS to near-IR for clinical investigation techniques such as Raman spectroscopy and OCT, up to the mid-IR for gas sensing and space exploration (wavelength ranges based on astronomy division and sensor response division schemes as addressed in  and references therein). Adding active functions (light sources, detectors, switches, etc.), microfluidic components, and/or nonreciprocal elements such as circulators increases the functional and technological complexity beyond the integration of passive circuitry. Therefore, one of the main challenges remains to be the access to cost-efficient (hybrid) assembly and packaging techniques.
Materials and technologies applied in integrated optics research and development vary over a wide range, e.g., doped silica, polymers, silicon-on-insulator, III-V semiconductors, chalcogenides, lithium niobate and dielectric materials such as Si3N4 or Al2O3. In Figure 1, a (non-exclusive) overview of known optical materials is given, showing their applicability as function of wavelength. Although many material systems have been successfully applied in some of the application fields, none of the technologies is capable to fulfill the complete set of requirements for all photonic applications. Therefore, the integration of complementary technologies is often pursued.
At present, three technology platforms are commercially available in multi-project wafer (MPW) runs: the III-V InP technology, silicon photonics, and the TriPleX platform. Customers have access to a library containing basic building blocks, allowing complex functional designs. Several design houses [22, 23] support these platforms. Fabrication facilities for these platforms are available: InP [24–27], silicon (Si) photonics [28–30], TriPleX [31–33]. The key property performances of these technologies are summarized in Figure 2.
As indicated in Figure 2, the InP technology is applied in the full range of active functionalities yielding a large variety of photonic applications, such as laser (array)s, (fast) modulators, and detectors. Si-photonics and TriPleX are both silicon based and can, therefore, rely on CMOS-compatible fabrication technology enabling low-cost volume manufacturing . Silicon photonics has combined its passive platform with significant progress in the realization of on-chip detectors and modulators. Most applications can be found in datacom . TriPleX has focused on passive performance (i.e., no on-chip gain media or high-frequency tuning) with ultralow loss over a wide wavelength range including the visible range down to near UV. TriPleX is widely applied for microwave photonics, sensors, and visible light applications [36, 37].
For all three platforms, tight curvature of waveguide bends enabling cascading of multiple structures and reduction of optical footprint is mandatory. Therefore, a large refractive index contrast between waveguide core and cladding is needed. On the other hand, very efficient coupling to low-contrast optical fibers has to be accomplished implying completely opposite requirements on the index contrast of the channel waveguide. For the TriPleX platform, a monolithic spotsize converter technology has been developed. This technology allows for on-chip modal field size conversion and, therefore, enables highly efficient fiber-to-chip coupling over the full accessible wavelength range while maintaining a large flexibility in waveguide geometry being inherent to the TriPleX photonic platform.
In this paper, we provide a comprehensive review on the TriPleX photonic waveguide platform. In Section 2, the properties, technological and design aspects will be discussed with emphasis on the different commercialized TriPleX geometries, integration aspects, fabrication concepts, and basic building blocks. Section 3 will focus on the applications and performance of TriPleX-based components for communications, biomedicine, and sensing as well as a few special fields of photonics.
TriPleX waveguide technology is based on alternating well-defined and highly stable silicon oxide (SiO2) and silicon nitride (Si3N4) layers [38, 39]. The CMOS-compatible fabrication equipment based on batch processing by low-pressure chemical vapor deposition (LPCVD) enables volume production as well as good reproducibility [32, 40]. TriPleX waveguides are transparent for wavelengths from 0.4 up to 2.35 μm and exhibit very low optical propagation losses. This technology combines good integration potential with high design flexibility allowing for tailoring of the waveguide properties . We distinguish three commercially available basic waveguide geometries, which can be obtained by modification of individual steps in the generic fabrication flow . The geometries are categorized by their shape: box shell (I), double stripe (II), and filled box (III), whereas type II is subdivided into the symmetric double stripe (IIa), the asymmetric double stripe (IIc), and the single stripe (IIb) being a special case of IIc. The schematic layouts of the geometries and SEM micrographs of realized structures are depicted in Figure 3. The generic process flow covering the fabrication steps of all TriPleX types is given in Figure 4.
While the overall geometrical dimensions of the waveguide cores of these shapes are typically in the order of 1 μm2, their waveguide characteristics and potential application areas differ greatly. Some key characteristics are effective index of the waveguide mode Neff and group index Ng (for TE-polarized light), channel propagation loss αch [dB/cm], polarization-dependent loss PDL [dB], minimum bending radius Rb [μm], mode field diameter MFD [μm] (TE00 mode), fiber-chip coupling loss αf-c [dB/facet], and waveguide birefringence ΔNeff.
The box shell layout is highly suited for telecom applications: due to its symmetrical shape, the polarization dependence is largely reduced . The box shell is available in a low  and high  index contrast variant based on different ratios of the Si3N4 and SiO2 thicknesses in the core region. In the low contrast version, a 1×1-μm2 SiO2 core is surrounded by 50-nm-thick Si3N4, while the high contrast is achieved with SiO2 and Si3N4 core dimensions of 0.5×0.5 μm2 and 170 nm, respectively. The waveguide propagation loss of the high-contrast box shell geometry is below 0.2 dB/cm and reduces to 0.06 dB/cm for the low-contrast variant . The main difference in the characteristics of both versions can be found in the bending radius, which is designed to fulfill the 0.01 dB/cm bend loss criterion, and the mode field diameter of the circular mode shape: Rb=500 μm, MFD=3.6 μm (low contrast), and Rb=150 μm, MFD=1.4 μm (high contrast).
The symmetric double stripe layout is typically applied in components requiring tight bending radii and large polarization birefringence. The optimized geometry is composed of two 170-nm-thick and 1.2-μm-wide Si3N4 layers separated by a 500-nm-thick SiO2 interlayer. The effective index of the waveguide mode at 1.55 μm wavelength and the group index are 1.535 and 1.72, respectively, while the waveguide birefringence is 5.3×10-2. The waveguide attenuation is ≤0.1 dB/cm . The tight bending radius allows for functional complexity at VLSI level. As shown in Figure 5, waveguide propagation loss as low as 0.095 dB/cm was measured in optical ring resonator (ORR) structures with a bending radius down to 70 μm . The circular mode of this waveguide geometry has a mode field diameter of ∼1.5 μm. To enable low-loss coupling to standard single-mode fibers (SMF28), both Si3N4 stripes are adiabatically tapered in the vertical direction. The adiabaticity criterion is fulfilled if the angle of the MFD change is below 1° (i.e., geometrical taper angle ≤0.01°). In case of this TriPleX geometry, this results in typical taper lengths of 600 μm. At an optimized thickness of 35 nm for both tapered Si3N4 layers, fiber-chip coupling losses of <0.5 dB were demonstrated.
The a-symmetric double-stripe geometry is highly suitable for combining regions with low and high effective indices of the mode on a single chip. In a realized example of this geometry type, a 40-nm-thick low-contrast Si3N4 stripe is separated in the vertical direction by a 100-nm SiO2 interlayer from the 175-nm-thick high-contrast Si3N4 channel . An adiabatic transition is made between both contrast regions by tapering the thicker layer to zero. In the high-contrast region, the channel width is set to 1.5 μm, while the width in the low contrast area is variable. The high contrast waveguide exhibits an MFD of 1.7 μm and 1.3 μm in the in-plane and out-of-plane directions, respectively. The waveguide birefringence is in the order of 5×10-2. Demonstrated loss values for the low and high modal confinement areas on the same chip were 0.015 dB/cm and 0.15 dB/cm, respectively.
The single stripe geometry is applied when extremely low optical losses are required: as explained, the single stripe can be converted from the asymmetric double stripe by locally removing the high-contrast waveguide. A detailed study on propagation loss and bending radii as a function of core aspect ratio (width/thickness) of this low-contrast stripe was carried out over a wide range of layer thicknesses (20–100 nm) and channel widths (0.3–14 μm) . While high aspect ratios (channel width over thickness) are favorable for low propagation losses, the bending radius decreases at lower aspect ratios. Propagation losses as low as 0.03 dB/cm in a stripe geometry allowing for 2-mm bending radii were demonstrated upon light propagation at 1.5 μm wavelength. Analyzing measured loss values with models taking into account various loss mechanisms the stripe propagation losses could be attributed to residual roughness and layer imperfections originating from the fabrication processes. The waveguide birefringence of 100-nm-thick Si3N4 waveguides with channel widths varying from 1.4 to 3.5 μm was investigated . For core widths above 2 μm, the birefringence exceeds 10-2 and good polarization-maintaining performance can, therefore, be expected. TE-pass polarizers with cross-polarization as low as -75 dB (measurement limited) have been reported .
Significant improvement of the loss figure was achieved by optimization of the channel fabrication process and replacement of the upper PECVD SiO2 cladding layer by a bonded thermal SiO2 film . The propagation loss through a single mode waveguide channel with 50-nm×5.3-μm geometrical cross-section was reduced to 0.007 dB/cm . Record-low loss of below 0.0005 dB/cm was measured at 1.58 μm wavelength on 40-nm×13-μm and 50-nm×6.5-μm multimode stripe waveguides .
The challenge of minimizing the bend radius while maintaining low propagation losses was addressed by various approaches. In Figure 6, light propagation through a stripe geometry based on 40-nm-thick and 14-μm-wide Si3N4 core is demonstrated down to a bending radius of 165 μm. However, in that geometrical range, the channel is multimode, and an additional tapering section is required for adiabatic widening of the channel width .
Finally, the filled-box geometry was developed for applications requiring ultrahigh confinement of the modal field in the core layer. In this geometry, core widths of 0.8–1.0 μm and thicknesses varying from 0.8 to 1.2 μm were realized . The waveguide channel is multimodal; up to three modes exist. For the 1.55-μm wavelength and TE-polarized light, the corresponding effective index of the mode and MFD of the fundamental mode (TE00) are around 1.79 and 1 μm, respectively. A minimum bending radius of 12.5 μm was calculated for the TE00 mode taking the 0.01 dB/cm loss criterion into account. As stoichiometric LPCVD Si3N4 films exhibit high tensile stress, the layer thickness growth is limited to <500 nm. Therefore, trench etching (into the thermal oxide layer) and trench filling (by LPCVD Si3N4 deposition) in combination with chemical mechanical polishing (CMP) was applied for the fabrication of crack-free waveguides (see also Figure 4). The optical propagation loss of these waveguides at 1.55 μm wavelength was 0.4 dB/cm .
The high design flexibility provided by the TriPleX platform is prerequisite to boost photonic integration. The availability of low- and high-contrast regions on a single chip easily enables interposer functionality in photonic assemblies. The large freedom in pitch and spotsize conversion is, for example, utilized in low-loss connections between optical fiber arrays and high-confinement waveguide technologies such as InP or silicon on insulator (SOI) . In return, active functionality (light generation, detection, and modulation) available on those platforms is efficiently joined with the TriPleX circuitry. Various complementary functionalities such as high-Q resonators, arrayed waveguide gratings, tunable delay lines, and devices for polarization control available in the ultralow loss silicon nitride waveguide technology and in silicon photonics are reviewed [55, 56]. On the SOI side among others, modules based on hybrid integration with the III-V InP technology and magneto-optical materials add to the functionality portfolio. The proposed integration concepts for silicon nitride waveguides with the silicon photonic platform are based on a hybrid bonding process utilizing an SiO2 interface layer with a controlled thickness [57, 58]. The integration concept and an SEM micrograph of a fabricated device are shown in Figure 7.
Currently, a novel assembly concept for integrating InP and TriPleX chips on a carrier platform is being developed by a European consortium: PHASTFlex . Besides addressing low-cost, high-volume assembly, the approach focuses on bridging the gap between typical flip-chip placement accuracies (≤1 μm) and the sub-100-nm alignment precision required by high-contrast coupling interfaces by monolithically integrated MEMS-based fine tuning.
On-chip adiabatic tapering sections are key elements for many integration aspects. Therefore, the optimization of the taper design and technology was emphasized. Vertical tapers are usually fabricated by standard lithography and isotropic etching of the silicon nitride layers. Measurements of propagation losses through such tapering sections resulted in ≤0.5 dB/taper . The most critical part of the tapering path is at the starting position of the taper (thin side), where the propagation of the weakly confined mode is extremely sensitive to geometrical changes. Therefore, LioniX has developed a novel ultrataper procedure, in which the shape of the tapering section can be controlled by design rather than solely by the process-specific etch profile . The preliminary results based on insertion loss measurements of waveguides with multiple tapering sections indicate a taper loss reduction by about one order of magnitude.
Last but not the least, the TriPleX platform is also well suited for lab-on-a-chip applications. Manufacturable concepts (Figure 8) for the integration of microfluidic systems as well as the assembly of VCSEL arrays and detectors have been developed [42, 61].
Low-threshold access to the TriPleX platform is, among others, enabled through MPW services . The MPW approach provides a solution for researchers and entrepreneurs to prototype their design at affordable cost and time levels. The designs of several participants are integrated on a single mask and fabricated in a shuttle run. Circuitry simulation and mask design can be carried out by PhoeniX software tools . A prerequisite to the successful implementation of such an approach is the availability of well-defined and verified design elements. Therefore, a library of standard optical components was established and implemented as basic building blocks (BBBs) in the licensed design kit of the PhoeniX software. In the past MPW services, components for the box shell geometry were applied, while in currently active MPW calls, the double-stripe geometry is offered. The BBB libraries include the following main elements: waveguide straight, bent waveguide, Y-junction, directional coupler, spotsize converter, and heater section. This allows for the easy design of integrated functions such as interposers, microring resonators (MRRs), several types of interferometers (multimode, Michelson, Young, Mach-Zehnder), arrayed waveguide gratings, multiplexers, frequency, and modal filters, mode and polarization converters as well as implementation of thermo-optic tuning or sensing windows. The library kit for the asymmetric stripe geometry is under preparation.
TriPleX services as well as its combination with other photonic platforms and technologies are also available through the European initiative ACTPHAST , which has created a unique one-stop-shop European access center for photonics innovation solutions and technology support.
In this section, we will review the performance of TriPleX photonics technology reported for a wide range of application fields. Several components for communications will be addressed first. Then details on state of the art in biomedical and sensing applications will be given, followed by several special areas such as nonlinear optics, light generation, and components for space.
In many cases, photonic functions used in communication applications like true time delay networks, wavelength routing, data packet buffering, narrowband filtering, timing references, or narrow linewidth light sources are based on long propagation distances or resonators with high-quality factor and require, therefore, extremely low propagation loss (<0.10 dB/cm) . A range of TriPleX-based photonic functions and building blocks applicable in telecommunication components including integration with the III-V or silicon platforms was recently investigated by researchers of UCSB .
State-of-the-art quality factors exceeding 4×108 have been demonstrated in silica microtoroids on a chip . Light coupling to the toroidal cavities is usually realized through tapered fibers because on-chip integration of access waveguides is challenging. The integration of high-Q ring resonators in the ultralow-loss TriPleX platform is an attractive approach, as access waveguides, cavities, and directional couplers can be monolithically integrated. In the TriPleX single stripe geometry (40 nm×11 μm), add-drop ring resonators, with a 9.65-mm bending radius and a 3.3-GHz FSR at 1580 nm wavelength, have been integrated [65, 66]. Several directional coupler concepts (symmetric, straight, pulley, and tapered) have been implemented to assess and minimize the coupler excess loss. An intrinsic Q-factor of 0.81×108 was measured with a weakly tapered directional coupler design. Round trip and propagation losses of 0.019 dB and 0.0032 dB/cm were extracted, respectively. Upon further coupler optimization and full exploitation of the low-loss TriPleX platform (i.e., αch<0.0005 dB/cm) Q-factors as high as 6×108 are predicted.
A range of ring resonators suited for application over a wide wavelength range (1060, 1310 and 1550 nm) was studied in . Based on single-mode TE propagation in two different TriPleX single stripe geometries (50 nm×5.3 μm and 80 nm×2.8 μm with bending radii of 5 and 2 mm, respectively), Q-factors of 19, 28, 7×106 were demonstrated at 1060, 1310, and 1550 nm wavelength, respectively. Analysis of the filter characteristics resulted in 16, 38, 300 MHz bandwidths for the 1060, 1310, and 1550 nm operations, respectively. The performance, as well as the wide wavelength range, makes these building blocks attractive for many applications ranging from ultra-narrowband MWP filters to highly sensitive biosensors.
For the integration (de)multiplexing functionality in the 1310- and 1550-nm wavelength ranges, 16-channel arrayed waveguide gratings (AWG) with a 200-GHz channel spacing were realized in TriPleX single stripe geometry (50 nm×5.5 μm) on a chip footprint of 15 mm×20 mm . The component layout is depicted in Figure 9A. For an operation at 1.3 μm wavelength, an FSR of 18.8 nm (2350 GHz), adjacent channel crosstalk of -30 dB and non-adjacent channel crosstalk of -40 dB were measured. The measured performance of the AWG in the 1.5-μm wavelengths range is shown in Figure 9B. Furthermore, the spectral response to a change in temperature (Δλ/ΔT) was measured to be 1.1×10-2 nm/°C. The implementation of AWGs in a 400-Gb/s WDM receiver unit with hybrid integrated detectors was demonstrated as well .
Another highly interesting area in communications is MWP being an interdisciplinary field investigating the techniques for generation, transmission, processing, and analyzing of RF/microwave signals using photonic devices or subsystems . The properties of photonic components, such as large instantaneous bandwidth, low loss, small size and weight, immunity from electromagnetic interference (EMI), as well as easy tunability and reconfigurability, are leveraged to realize novel RF/microwave functionalities with unprecedented features .
In recent years, a number of RF/microwave functionalities have been demonstrated using on-chip MWP signal processing in TriPleX technology. Examples range from tunable microwave filters, phase shifters, frequency converters, ultra-wideband microwave signal generators to tunable true time delay (TTD) elements for optical beam-forming networks (OBFN) [69–73]. A detailed discussion of the basic building blocks applied in integrated OBFNs can be found in .
The optical components are based on optical MRRs. Tuning and bandwidth reconfiguration in the order of a few GHz can be easily implemented, and squint-free performance and seamless beam steering for broadband applications are enabled. For the utilization of OBFNs in satellite-tracking phased array antenna system operating in the Ku-band (10.7–12.75 GHz) delay lengths in the order of tens of centimeter are required. Consequently, low loss and compact waveguide configurations are essential for meeting performance and cost requirements. Therefore, the double stripe TriPleX geometry enabling MRRs with 125 μm radius is chosen.
To date, three different types of optical beam formers have been implemented with our technology: single wavelength, multi-wavelength, and multi-beam . A realized chip of a 16×1 single wavelength OBFN component is shown in Figure 10A. On the 7-mm×22-mm chip area, a TTD unit with 40 MRRs, splitters/combiners, and an optical sideband filter are included [33, 76]. Delays of 650 ps over a 4.5-GHz bandwidth were measured.
By introducing a novel hardware architecture  based on parallelism, a 16×16 multi-wavelength OBFN could be realized on chip real estate as small as 36×8 mm2. Besides delay units and combiners, this chip (Figure 10B) contains optical phase shifters, carrier tuners, and (de)multiplexers. A continuously tunable delay of up to 140 ps was measured over an instantaneous bandwidth from 2 to 10 GHz.
A multi-beam layout based on a Nolen matrix has gone through the design phase. For a 255-antenna×36-beam Nolen matrix based on tunable Mach-Zehnder coupler unit cells, a chip size around 90 cm2 is expected. This footprint represents a reduction of three orders of magnitude compared to conventional RF-BFN techniques .
A fully packaged optical TTD component for wide bandwidth signals and an insertion loss of 5.3 dB was reported by Morton et al. [77, 78]. Delays of 535 ps and 632 ps were measured for 10 and 20 GHz wide signals, respectively.
Another approach for integrated 4-bit tunable temporal delay lines has been demonstrated by switching spiral-shaped waveguides based on the single stripe TriPleX geometry  over a chip area of 4.5×8.5 cm2. Delays of up to 12.35 ns with 0.85 ns resolution were measured. The insertion loss measured for all 16 states varied from -12 to -14 dB.
Further developments toward a fully integrated and low-cost MWP beamformer system exploit the large-scale integration potential of the TriPleX technology. Simultaneous implementation of components operating with phase modulation and direct detection  is reported, and novel fully tunable RF filters are implemented . The various stages of the MPW beamformer system – from the filter schematic to the fully packaged commercial product – are depicted in Figure 11A–D.
As mentioned Section 1, integrated photonics solutions become increasingly important for biomedical applications and started to enter fields like OCT, Raman spectroscopy, flow cytometry, etc. TriPleX-based photonics technology has been applied for a range of novel, miniaturized biomedical components.
Optical coherence tomography (OCT) is a widely applied noninvasive medical imaging technique for 3-D biological tissue investigation. The technique is based on low-coherence interferometry from scattering media. Operation in the 800- to 1300-nm wavelength range enables up to several millimeter penetration depth in the tissue. Utilization of wide spectral ranges (>100 nm) result in submicrometer resolution. In spectral domain OCT, we distinguish two basic data acquisition modes: swept-source and spectrometer-based OCT.
An integrated design for swept-source OCT operating with an external swept source at 1300 nm was developed in cooperation with the AMC in Amsterdam (Figure 12) . The TriPleX chip functionality is designed in single-stripe TriPleX geometry (50 nm×3.4 μm) and has a chip footprint of 4 mm×18 mm (Figure 12B). Component characterization resulted in 21±1 μm lateral resolution, -80 dB sensitivity, and a maximum imaging depth of 5.09 mm. The axial resolution of 12.7±0.5 μm is in agreement with the bandwidth-limited resolution.
Exploiting the TriPleX box shell geometry, an integrated interferometer chip for spectrometer-based OCT was developed by Ghent University in one of the MPW runs . The design supports both polarizations and combines several Y-branches with a 190-mm-long reference arm at a footprint of only 10 mm×33 mm (Figure 13). The axial resolution of the system was 14 μm. The sensitivity measured at 0.25 mm depth and 0.1 mW optical powers on the sample was 65 dB. Comparison with the response of a fiber-based system showed good agreement in performance. As dispersion control is an important aspect in wide-range spectral systems, the reduction of group velocity dispersion is investigated, and its potential implementation devices based on TriPleX geometry is discussed.
The integration of TriPleX box shell geometry waveguides with microfluidic channels was applied in the demonstration of optical particle trapping in combination with Raman spectroscopy, which was realized in one of the MPW runs . The optical trap was created by splitting the 785-nm optical signal by a 3-dB Y-junction over two excitation branches facing toward each other at the 5-μm-wide trapping and excitation region of the fluidic channel (Figure 14). Chip dimensions were as small as 8 mm×16 mm. Strong trapping forces as well as measurements of Raman spectra over integration times as short as 0.25 s were demonstrated.
A highly useful component for many biomedical and sensing applications is an integrated optical version of a laser beam combiner (LBC). The component developed by XiO Photonics combines the emission from four to eight laser sources in the 400- to 850-nm wavelength range into one single-mode fiber output [37, 84]. On-chip polarization maintaining filtering functions based on o.a. MRRs in TriPleX technology are applied. The resulting fully packaged component, as shown in Figure 15, is highly compact.
Integrated optical circuits are increasingly applied for the detection of chemical and biological substances , as they enable high measurement resolution and high sensitivity over a wide dynamic range. Compact integration of relevant optical functions, such as mode filters and interferometers, can be realized at a low cost. TriPleX technology can be applied over the relevant wavelength range of this application field and provides, moreover, a chemically stable detection platform .
Sensing principle: In these optical chips, the chemical parameter to be sensed (i.e., the measurand) influences the propagation properties of a guided light signal that propagates through an on-chip sensing region. The relevant interactions occur primarily in the evanescent field region just above the interface of the waveguide core layer, which can be accessed through a sensing window in the cover layer (see Figure 16). Generally, selectivity is provided by the chemo-optical transduction layer. This layer contains receptor units with selective affinity to the chemical entities of the measurand. Using identical optical chips and electronics, the concentration of a large variety of measurands can be determined by applying different transduction layers, such as antibody proteins as receptor molecules for immunosensors  or silicates containing functional organic groups for gas sensors . We distinguish between refractive index, absorption, and fluorescence sensing.
Refractive index sensing: The sensor design can be either based on MRRs [89, 90] or on a Mach-Zehnder interferometer (MZI) [87, 91–93] configuration. The state of the art in MZI-based sensing platforms – being widely applied by LioniX’ customers – will be subject to future publication. In this review, we focus on the reported sensor performance based on the MRR approach.
The layout of the MRR component is based on a ring-shaped resonator waveguide with two directional couplers having a length LCD and a gap g. For the design of the MRR sensor, the 850-nm wavelength is chosen enabling the use of low-cost VCSEL light sources. The MRR performance is optimized for the 100-nm-thick Si3N4 single stripe waveguides previously shown to result in highly sensitive performance in evanescent wave sensing. Based on a waveguide channel width of 1.0 μm, MRR with radii of 50 and 75 μm are designed. For g=0.75 μm and LDC varying from 0 to 250 μm, FSR values ranging from 360 to 880 pm are feasible . To realize low-loss coupling to the pig-tailed VCSEL, spotsize conversion is utilized by tapering the Si3N4 thickness to 25 nm.
The schematic layout of the MRR sensor chip is illustrated in Figure 17A. The spotsize conversion area with reduced waveguide thickness is indicated in green. The chip contains three MRRs, of which one serves as reference resonator (top cladding not removed), and two MRRs are equipped with a sensing window (indicated in blue). A microscope image of the fabricated chip is shown in Figure 17B. The response of a microring with a 50-μm radius and LDC=200 μm was measured with a tunable 850-nm VCSEL (Figure 17C). The measured FRS of 560 pm is in good agreement with the simulation results.
The lab-on-a-chip application potential of this sensor chip was investigated after integration with a microfluidic flow cell. Refractive index measurements in the fluidic channel resulted in a sensor response of 110±22 nm/RIU being an equivalent to a measurement resolution in the order of 10-6 RIU. Label-free biochemical surface reactions have been demonstrated with bovine serum albumin (bSA). The response curve to bSA concentrations ranging from 1×10-7 to 2.2×10-4 M is shown in Figure 17D. Based on the theoretical biochemical surface sensitivity shift for bSA, the detection limit of our sensor was estimated to be smaller than 1 pm of an average bSA layer growth. The feasibility of gas sensing was shown in a preliminary experiment based on the interaction between acetone and a Teflon transduction film applied in the sensing window. A quick sensor response was observed upon the presence of acetone in a beaker being in the proximity of the sensor device.
The feasibility of multichannel sensing was assessed for an eightfold MRR array designed in the asymmetric stripe TriPleX geometry with 35 nm and 85 nm lower and upper silicon nitride layer thickness, respectively . Challenges with respect to power budget aspects in multichannel operations are addressed by a grating-based vertical VCSEL coupling concept in combination with an optimized 1×8 multimode interference (MMI) coupler.
Absorption sensing: TriPleX-based chips have been designed and realized for a range of absorption-based applications in fluidic environments.
The development of processes in a microreactor largely benefits from integrated UV/VIS spectroscopic detection allowing for real-time analysis under segmented flow. An optofluidic chip optimized for evanescent field sensing based on a 65-nm-thick Si3N4 waveguide layer and an operational wavelength range from 488 to 632 nm was developed for the monitoring of gold nanoparticle production . Segmented flow detection at integration times as low as 2 ms was demonstrated.
Fluorescence sensing: A sensitive and low-cost detection scheme for molecular biomarkers has been demonstrated [95, 96] in TriPleX chips of 10 mm×10 mm. One of the applied chip layouts yielding a matrix of 100 sensing wells is depicted in Figure 18. Light of 642-nm wavelength is coupled from a low-cost diode laser array into the excitation waveguides of the chip. In each sensing well, analyte molecules can be captured. Excitation of the analytes is realized in the sensing wells, where the excitation light penetrates to a depth of about 70 nm. Fluorescence emission from analyte molecules is back-coupled into the collection waveguide and measured by a detector array.
The main advantages of such a microarray-based approach can be found in short acquisition times (typically 5–10 min), very high sensitivity (1 fM or smaller), and applicability for multispecies sensing in complex matrices such as blood, whole serum, or urine. To the particularly interesting application fields belongs early stage cancer detection, screening for infectious diseases, or asymptomatic individuals testing following a biohazard event. Moreover, the systems fulfill requirements for point of care applications as they are low-cost, portable, battery operated, and allow for wireless connectivity.
For another customer, a multichannel fluorescence-based sensing system with a 10×10 marker array for life detection on the ExoMars mission was developed . The pigtailed module operating at 635 nm wavelength was realized in LioniX’ optofluidic platform. Between the fiber input and the assay chamber containing the marker array, the optical signal is split by a 1×11 MMI. The optical and fluidic modules are connected by a bonding process and diced to a chip size of 52 mm×16 mm.
Besides the above-discussed conventional application fields, the technology platform has also entered emerging areas of photonics and special applications such as integrated light sources, nonlinear optic devices, components for space, and quantum computing.
The active/passive integration potential of the silicon nitride technology was, for example, demonstrated by the monolithic implementation of optically pumped DFB lasers (Figure 19). For this purpose, side wall gratings with 520-nm period have been etched by stepper lithography into the 4.6-μm-wide silicon nitride channel waveguides . The grating coupling constant can be controlled between 13 and 310 cm-1 by adjusting the aspect ratio between the original channel width and the sidewall gratings. High grating reflectivity was measured in the 1550-nm wavelength range. On top of the passive Si3N4 platform an erbium (Er3+)-doped aluminum oxide film with an Er3+ concentration of 1.3×1020 cm-3 was integrated by reactive co-sputtering . The laser performance of several DBR and DFB grating configurations was measured applying 974-nm pumping. Pump thresholds, laser output powers, conversion efficiencies, and side mode suppressions of 11 mW, 2.1 mW, 5.2%, and 50dB, respectively, were achieved.
Ultra-narrow linewidth laser components have been realized by hybrid integration of III-V laser diodes with a TriPleX cavity in a so-called waveguide-based external cavity semiconductor laser (WECSL) configuration . The design is based on the box shell geometry and exploits the response of two integrated MRRs acting as external mirror and having radii of 50 and 55 μm, respectively. The layout of the frequency-selective mirror component realized in TriPleX technology is shown in Figure 20. A laser diode equipped with only one mirror is connected to the input port (IN) of the chip. To monitor the laser and mirror performance in greater detail, three output ports (OUT) are added to the design, whereas OUT3 provides the actual response of the WECSL. The measured FWHM laser linewidth and side mode suppression ratio of the WECSL are 25 kHz and 50 dB, respectively. Wavelength tuning over the full telecommunications C-band (1530–1565 nm) is demonstrated.
For nonlinear optic applications, high modal confinement in the photonic waveguide is of importance. Therefore, the filled box geometry of the TriPleX platform with core cross sections of 1.2 μm×0.8 μm is highly attractive. In ongoing research, this technology is implemented for the design and realization of components for supercontinuum generation with over 700 nm spectral bandwidth  and four-wave mixing in the 800–1500 nm wavelength range .
The specific requirements of several applications for space research can be met by the TriPleX technology. In general, components for space missions need to be highly compact, robust, lightweight, and low on power consumption. One example of a realized module for the ExoMars mission was discussed in the fluorescence sensing section of this paper. Another application is addressed by the development of spectroscopic techniques on earth-orbiting satellites where an in-flight calibration module for the 2.3- to 2.4-μm wavelength range was designed with thermally tunable MRR building blocks based on the asymmetric stripe geometry [21, 103]. The packaged module (Figure 21) has an insertion loss as low as 2.4 dB.
LioniX’ TriPleX technology is a versatile photonics platform suited for applications over a wide wavelength range covering the transparency window from 0.4 to 2.35 μm. Ultralow-loss channel waveguides with propagation losses down to 0.0005 dB/cm have been demonstrated. The monolithic integration of low- and high-contrast waveguides on the same chip enables the realization of low-loss spotsize converters. Therefore, TriPleX chips are suitable joints between different high-index platforms such as III-V or SOI and standard optical fibers giving access to the ‘outside world’. Currently, substantial effort is put into the development of low-cost integration and assembly techniques for adding complementary functionalities of the different photonic platforms into a single package.
In this paper, we have reviewed a wide range of application examples that have been realized in or enabled by TriPleX technology. In the area of communications, components for telecommunication networks and microwave photonics operating in the 1.3- and 1.5-μm wavelength window are discussed. High-performance filtering and true time delay concepts are surveyed, and components with high functional complexity for optical beam-forming networks are introduced. Milestones in the development of miniaturized biomedical devices can be mainly found in the field of optical coherence tomography and integrated Raman spectroscopy. The results addressed in this paper cover functional devices operating over a broad spectral range, from VIS to 1.3 μm. Although results in this field are still preliminary, the potential of integrated optics solutions becomes clear. Finally, sensor devices and subsystems are addressed. Main application fields are environmental control, food/water safety, and security. The spectral range of interest is at a visible wavelength and at 850 nm due to access to low-cost VCSEL light sources. State-of-the-art sensing concepts developed on the TriPleX platform allow for detection limits around 10-7 RIU.
The functionality of the TriPleX platform is captured by verified basic building blocks and incorporated in the library of a design kit developed by LioniX and PhoeniX Software. The availability of this library in combination with affordable manufacturing through MPW shuttle runs lowers the threshold for researchers and entrepreneurs to enter and exploit photonics technology.
The authors would like to thank R. Duer (iNDX Lifecare Inc.), C. Roeloffzen (SatraX BV), and J. van Kerkhof (XiO Photonics BV) for providing photographs of their commercial products based on TriPleX chip technology. Co-workers of the following research groups and companies are acknowledged for the fruitful discussions and collaborations: Department of Electrical and Computer Engineering at the University of California Santa Barbara, Politecnico di Milano, SatraX BV, Telecommunication Engineering Group at the University of Twente, Morton Photonics Inc., Photonics Research Group at Ghent University, Biomedical Engineering and Physics Group at Amsterdam Medical Center, Department of Imaging Physics at Delft University of Technology, TNO, iNDX Lifecare Inc., Division of Information Transmission Systems and Material Technology at National Technical University of Athens, XiO Photonics BV, Laser Physics and Nonlinear Optics Group at the University of Twente, PhoeniX BV.
 W. Sohler and R. De La Rue, Eds., Laser Photonics Rev. 6, Special Issue on Integrated Optics: From Single Photon Sources to Complex Photonic Circuits, (2012) pp. 1–143. http://onlinelibrary.wiley.com/doi/10.1002/lpor.v6.1/issuetoc. Search in Google Scholar
 W. Fritzsche and J. Popp, Eds., ‘Optical Nano- and Microsystems for Bioanalytics’, (Springer, Heidelberg, 2012) pp. 103–324. Search in Google Scholar
 B. E. Little, S. T. Chu, W. Pan and Y. Kokubun, IEEE Photonic Tech. L. 12, 323 (2000). Search in Google Scholar
 C. K. Madsen, J. Lightwave Technol. 21, 2412 (2003). Search in Google Scholar
 J. Capmany and D. Novak, Nat. Photonics 1, 319 (2007). Search in Google Scholar
 D. Marpaung, C. Roeloffzen, R. Heideman, A. Leinse, S. Sales, et al., Laser Photonics Rev. 7, 506 (2013). Search in Google Scholar
 K. Okamoto, in ‘Proc. Asia Optical Fiber Communication Conf.’, (IEEE, Shanghai, October 17–19, 2007) pp. 307–309. Search in Google Scholar
 K. Wada, S. Park and Y. Ishikawa, Proc. IEEE 97, 1329 (2009). Search in Google Scholar
 M. Asghari, in ‘Proc. National Fiber Optic Engineers Conf.’, (IEEE, San Diego, February 24–28, 2008), paper NThA4. Search in Google Scholar
 L. Chen, K. Preston, S. Manipatruni and M. Lipson, Opt. Express 17, 15248 (2009). Search in Google Scholar
 J. Klamkin, F. Gambini, S. Faralli, A. Malacarne, G. Meloni, et al., Opt. Express 22, 2150 (2014). Search in Google Scholar
 V. M. N. Passaro, C. de Tullio, B. Troia, M. La Notte, G. Giannoccaro, et al., Sensors 12, 15558 (2012). Search in Google Scholar
 P. Kozma, F. Kehl, E. Ehrentreich-Förster, Ch. Stamm and F. F. Bier, Biosens. Bioelectron. 58, 287 (2014). Search in Google Scholar
 S. Janz, D. X. Xu, M. Vachon, N. Sabourin, P. Cheben, et al., Opt. Express 21, 4623 (2013). Search in Google Scholar
 N. Ismail, L.-P. Choo-Smith, K. Wörhoff, A. Driessen, A. C. Baclig, et al., Opt. Lett. 36, 4629 (2011). Search in Google Scholar
 V. D. Nguyen, B. I. Akca, K. Wörhoff, R. M. de Ridder, M. Pollnau, et al., Opt. Lett. 36, 1293 (2011). Search in Google Scholar
 G. Yurtsever, N. Weiss, J. Kalkman, T. G. van Leeuwen and R. Baets, Opt. Lett. 39, 5228 (2014). Search in Google Scholar
 M. Hoekman, J. Dingjan, P. J. Harmsma, R. P. Ebeling, D. M. R. Lo Cascio, et al., Poster in ‘ESA Round Table on Micro and Nano Technologies’ (2014). Search in Google Scholar
 R. G. Heideman, A. Leinse, W. Hoving, R. Dekker, D. H. Geuzebroek, et al., in ‘Proc. SPIE Vol. 7221’, (2009) pp. 72210R1. Search in Google Scholar
 C. G. H. Roeloffzen, L. Zhuang, C. Taddei, A. Leinse, R. G. Heideman, et al., Opt. Express 21, 22937 (2013). Search in Google Scholar
 A. Leinse, R. G. Heideman, M. Hoekman, F. Schreuder, F. Falke, et al., in ‘Proc. SPIE Vol. 8767’, (SPIE, May 22, 2013) pp. 87670E-1. Search in Google Scholar
 R. Dekker, E. J. Klein and D. H. Geuzebroek, in ‘Proc. IEEE Photonics Conf.’, (IEEE, Burlingame, CA, 23–27 Sept. 2012) pp. 286–287. Search in Google Scholar
 R. G. Heideman and J. A. Walker, in ‘Proc. SPIE Vol. 6125’, (SPIE, Photonics West, San Jose, 2006) paper 021. Search in Google Scholar
 F. Morichetti, A. Melloni, M. Martinelli, R. G. Heideman, A. Leinse, et al., J. Lightwave Technol. 25, 2579 (2007). Search in Google Scholar
 R. G. Heideman, D. Geuzebroek, A. Leinse, A. Melloni, F. Morichetti, et al., in ‘Proc. ECIO’, (Copenhagen, April 25–27, 2007) invited paper WB0. Search in Google Scholar
 W. Hoving, D. Geuzebroek and R. Heideman, SPIE Newsroom 2008. http://www.spie.org/x25303.xml?pf=true&highlight=x2414. Search in Google Scholar
 R. G. Heideman, M. Hoekman and E. Schreuder, IEEE J. Sel. Top. Quant. 18, 1583 (2012). Search in Google Scholar
 R. G. Heideman, A. Melloni, M. Hoekman, A. Borreman, A. Leinse, et al., in ‘Proc. IEEE/LEOS Benelux Ann. Symp.’, (IEEE, Mons, December 1–2, 2005), pp. 71–74. Search in Google Scholar
 D. A. I. Marpaung, C. G. H. Roeloffzen, A. Leinse and M. Hoekman, Opt. Express 18, 27359 (2010). Search in Google Scholar
 G. Yurtsever, B. Považay, A. Alex, B. Zabihian, W. Drexler, et al., Biomedical Opt. Express 5, 1050 (2014). Search in Google Scholar
 L. Zhuang, D. Marpaung, M. Burla, W. Beeker, A. Leinse, et al., Opt. Express 19, 23162 (2011). Search in Google Scholar
 D. T. Spencer, M. J. R. Heck, R. Moreira, J. Bovington, J. E. Bowers, et al., in ‘Proc. OFC’, (OSA, San Francisco, March 9–14, 2014) paper Th1A. Search in Google Scholar
 J. F. Bauters, M. J. R. Heck, D. John, D. Dai, M. C. Tien, et al., Opt. Express 19, 3163 (2011). Search in Google Scholar
 D. Dai, Z. Wang, J. F. Bauters, M. C. Tien, M. Heck, et al. Bowers, in ‘Proc. GFP’, (IEEE, Bejing, September 1–3, 2010) , P2.13. Search in Google Scholar
 J. F. Bauters, M. J. R. Heck, D. Dai, J. S. Barton, D. J. Blumenthal, et al., IEEE Photonics J. 5, 6600207 (2013). Search in Google Scholar
 J. F. Bauters, M. J. R. Heck, D. D. John, J. S. Barton, C. M. Bruinink, et al., Opt. Express 19, 24090 (2011). Search in Google Scholar
 J. F. Bauters, M. J. R. Heck, D. D. John, M.-C. Tien, W. Li, et al., in ‘Proc. ECOC’, (OSA, Geneva, September 18–22, 2011) paper Th12.3. Search in Google Scholar
 J. P. Epping, M. Hoekman, R. Mateman, A. Leinse, R. G. Heideman, et al., Opt. Express 23, 643 (2015). Search in Google Scholar
 M. J. Wale, Roadmap for InP and TriPleX-based Photonic Integration, presented at 5th European Photonic Integration Forum, ECOC Workshop: Low-Cost Access to Photonic ICs (London, 22nd September 2013). Search in Google Scholar
 M. J. R. Heck, J. F. Bauters, M. L. Davenport, D. T. Spencer and J. E. Bowers, Laser Photonics Rev. 8, 667 (2014). Search in Google Scholar
 D. Dai, J. Bauters and J. E. Bowers, Light Sci. Appl. 1, 1 (2012). Search in Google Scholar
 M. Piels, J. F. Bauters, M. L. Davenport, M. J. R. Heck and J. E. Bowers, J. Lightwave Technol. 32, 817 (2014). Search in Google Scholar
 M. Davenport, J. Bauters, M. Piels, A. Chen, A. Fang, et al., in ‘Proc. NFOEC’, (OSA, Anaheim, March 17–21, 2013) paper PDP2C.5. Search in Google Scholar
 R. G. Heideman and M. Hoekman, Two-dimensional tapered coupler (nonlinear taper), U.S. Patent Application No.: 14/270,014. Search in Google Scholar
 J. Yue, F. H. Falke, J. C. Schouten and T. A. Nijhuis, Lab Chip 13, 4855 (2013). Search in Google Scholar
 T. J. Kippenberg, S. M. Spillane and K. J. Vahala, Appl. Phys. Lett. 85, 6113 (2004). Search in Google Scholar
 D. T. Spencer, Y. Tang, J. F. Bauters, M. J. R. Heck and J. E. Bowers, in ‘Proc. IEEE Photonics Conf.’, (IEEE, Burlingame, 2012) pp. 141–142. Search in Google Scholar
 D. T. Spencer, J. F. Bauters, M. J. R. Heck and J. E. Bowers, Optica 1, 153 (2014). Search in Google Scholar
 M. C. Tien, J. F. Bauters, M. J. R. Heck, D. T. Spencer, D. J. Blumenthal, et al., Opt. Express 19, 13551 (2011). Search in Google Scholar
 D. Dai, Z. Wang, J. F. Bauters, M. C. Tien, M. J. R. Heck, et al., Opt. Express 19, 14130 (2011). Search in Google Scholar
 L. Zhuang, W. P. Beeker, A. Leinse, R. G. Heideman and C. G. H. Roeloffzen, Opt. Express 21, 3114 (2013). Search in Google Scholar
 L. Zhuang, M. R. Khan, W. P. Beeker, A. Leinse, R. G. Heideman, et al., Opt. Express 20, 26499 (2012). Search in Google Scholar
 M. Burla, D. A. I. Marpaung, L. Zhuang, C. G. H. Roeloffzen, M. R. Khan, et al., Opt. Express 19, 21476 (2011). Search in Google Scholar
 D. A. I. Marpaung, L. Chevalier, M. Burla and C. G. H. Roeloffzen, Opt. Express 19, 24839 (2011). Search in Google Scholar
 D. Marpaung, B. Morrison, R. Pant, C. Roeloffzen, A. Leinse, et al., Opt. Express 21, 23286 (2013). Search in Google Scholar
 C. G. H. Roeloffzen, R. M. Oldenbeuving, R. B. Timens, P. W. L. van Dijk, C. Taddei, et al., in ‘Proc. Optical Fiber Communication Conf.’, (OSA, Los Angeles, March 22–26, 2015), invited paper Tu3F.4. Search in Google Scholar
 M. Burla, D. Marpaung, L. Zhuang, M. Khan, A. Leinse, et al., J. Lightwave Technol. 32, 3509 (2014). Search in Google Scholar
 L. Zhuang, C. G. H. Roeloffzen, A. Meijerink, M. Burla, D. A. I. Marpaung, et al., J. Lightwave Technol. 28, 19 (2010). Search in Google Scholar
 P. A. Morton, J. B. Khurgin, Z. Mizrahi and S. J. Morton, in ‘Proc. CLEO’, (OSA, San Jose, June 8–13, 2014) paper AW3P.6. Search in Google Scholar
 P. A. Morton, J. B. Khurgin, Z. Mizrahi and S. J. Morton, in ‘Proc. Avionics, Fiber-Optics and Photonics Technology Conf.’, (IEEE, Atlanta, GA, USA, 2014), pp. 27–28. Search in Google Scholar
 R. L. Moreira, J. Garcia, W. Li, J. Bauters, J. S. Barton, et al., IEEE Phot. Tech. L. 25, 1165 (2013). Search in Google Scholar
 L. Zhuang, M. Hoekman, C. Taddei, A. Leinse, R. G. Heideman, et al., Opt. Express 22, 17079 (2014). Search in Google Scholar
 C. Taddei, L. Zhuang, M. Hoekman, C. Roeloffzen, R. Oldenbeuving, et al., in ‘Proc. International Topical meeting on Microwave Photonics/The 9th Asia-Pacific Microwave Photonics Conf.’, (IEEE, Sapporo, Japan, 20–23 October 2014) paper TuC-4. Search in Google Scholar
 V. D. Nguyen, N. Weiss, W. Beeker, M. Hoekman, A. Leinse, et al., Opt. Lett. 37, 4820 (2012). Search in Google Scholar
 M. Boerkamp, T. van Leest, J. Heldens, A. Leinse, M. Hoekman, et al., Opt. Express 22, 30528 (2014). Search in Google Scholar
 P. V. Lambeck, Meas. Sci. Technol. 17, R93 (2006). Search in Google Scholar
 R. G. Heideman, R. P. H. Kooyman and J. Greve, Sensor Actuat B-Chem 10, 209 (1993). Search in Google Scholar
 A. Brandenburg, R. Edelhäuser and F. Hutter, Sensor Actuat B-Chem 11, 361 (1993). Search in Google Scholar
 K. De Vos, I. Bartolozzi, E. Schacht, P. Bienstman and R. Baets, Opt. Express 15, 7610 (2007). Search in Google Scholar
 D. X. Xu, A. Densmore, A. Delâge, P. Waldron, R. McKinnon, et al., Opt. Express 16, 15137 (2008). Search in Google Scholar
 K. Misiakos, I. Raptis, A. Salapatas, E. Makarona, A. Botsialas, et al., Opt. Express 22, 8856 (2014). Search in Google Scholar
 B. J. Luff, J. S. Wilkinson, J. Piehler, U. Hollenbach, J. Ingenhoff et al., J. Lightwave Technol. 16, 583 (1998). Search in Google Scholar
 A. Densmore, D. X. Xu, P. Waldron, S. Janz, P. Cheben, et al., IEEE Phot. Tech. L. 18, 2520 (2006). Search in Google Scholar
 L. Gounaridis, P. Groumas, E. Schreuder, R. Heideman, V. Katopodis, et al., Sensor Actuat. B-Chem., 209, 1057 (2015). Search in Google Scholar
 R. Duer, R. Lund, R. Tanaka, D. A. Christensen and J. N. Herron, Anal. Chem. 82, 8856 (2010). Search in Google Scholar
 A. Prak, H. Leeuwis, R. G. Heideman, A. Leinse and G. Borst, in ‘Proc. SPIE Vol. 7928’, (SPIE, San Francisco, January 22, 2011) paper L-1. Search in Google Scholar
 M. Belt, J. Bovington, R. Moreira, J. F. Bauters, M. J. R. Heck, et al., Opt. Express 21, 1181 (2013). Search in Google Scholar
 M. Belt and D.J. Blumenthal, Opt. Express 22, 10655 (2014). Search in Google Scholar
 R. M. Oldenbeuving, E. J. Klein, H. L. Offerhaus, C. J. Lee, H. Song, et al., Phys. Lett. 10, 015804 (2013). Search in Google Scholar
 J. P. Epping, T. Hellwig, R. Mateman, A. van Rees, M. Hoekman, et al., in ‘Proc. Of 6th EPS-QEOD Europhoton Conf.’, (OSA, Neuchâtel, Switzerland, August 24–29, 2014). Search in Google Scholar
 J. P. Epping, M. Kues, P. J. M. van der Slot, C. J. Lee, C. Fallnich, et al., Opt. Express 21, 32123 (2013). Search in Google Scholar
 R. P. Ebeling, P. J. Harmsma, D. M. R. Lo Cascio, M. Hoekman, W. P. Beeker, et al., in ‘Proc. IEEE/LEOS Benelux Ann. Symp.’, (IEEE, Mons, November 29–30, 2012) pp. 109–112. Search in Google Scholar