Open Access
Share with Facebook
Tweet This
Post on reddit
Share with LinkedIn
Add to CiteULike
Add to Mendeley
Add to BibSonomyAbstract
A complete Mach-Zehnder interferometer monolithically integrated on silicon is presented and employed as a refractive index and bio-chemical sensor. The device consists of broad-band light sources optically coupled to photodetectors through monomodal waveguides forming arrays of Mach-Zehnder interferometers, all components being monolithically integrated on silicon through mainstream silicon technology. The interferometer is photonically engineered in a way that the phase difference of light travelling through the sensing and reference arms is approximately wavelength independent. Consequently, upon effective medium changes, it becomes feasible even with a broad-band source to induce sinusoidal-type of detector photocurrents similar to the classical monochromatic counterparts. The device is completed with its fluidic and interconnect components so that on chip interferometric measurements can be performed. Examples of refractive index and protein sensing are presented to establish the potential of the proposed device for real-time in situ monitoring applications. This is the only silicon device that has achieved complete on-chip interferometry.
© 2014 Optical Society of America
1. Introduction
Photonic probing is an elegant and sensitive option in detection technology since, as a contactless method, isolates galvanically the sensing element from the excitation and detection electronics and allows a harmonious interplay between the detection and fluidic components [1–3]. If label free detection is desired, then guided optics interferometry is the way of choice due to its extreme sensitivity to effective index changes of the sensing waveguide [4]. Such changes result either from cover medium refractive index (RI) variations or from adlayer formation on top of the sensing waveguide. In integrated optical devices a constant matter of concern is how to couple light to the waveguides in a way that is effective, reliable, cost efficient and allows for small size and portability. Monolithic optocoupler based interferometric devices would provide to the light coupling issue a robust solution compared to hybrid integration approaches. In the later, external light sources have to be coupled to the extremely thin and narrow waveguides through bulk optics, such as lenses, prism couplers, collimators [5] etc. raising accuracy, reliability and cost issues, especially in portable systems. If, on the other hand, the monolithic interferometer is realized on silicon there are obvious benefits in terms of cost, foundry service availability and ability to integrate mainstream readout electronics. Such interferometric devices can find their place as the core elements in point of need devices in decentralized sensing [6]. For truly monolithic silicon interferometry, the light source has to be integrated on chip. Although silicon is an indirect band gap material, light emission is possible in properly nanoengineered silicon diodes that incorporate either silicon nanocrystals imbedded in a silicon dioxide insulator [7], or silicon quantum wires [8,9]. Another electroluminescent device is the silicon avalanche p/n diode which emits light when biased beyond its breakdown voltage [10]. All of the above diodes emit a broad spectrum that includes the visible and the near IR. Provided that standard interferometers, like ring resonators or Mach-Zehnder devices are based on monochromatic light sources, any monolithic silicon optocoupler devices must deal with the broad-band nature of the silicon based light sources. In this respect, we present a unique Mach-Zehnder interferometer photonically engineered to accommodate broad-band light through the well-known sinusoidal expression for the detected intensity as a function of the sensing arm effective index, as in the case of Mach-Zehnder devices that employ monochromatic light sources. The sensing and reference arm core thicknesses are chosen so that the phase difference of the light travelling through them is more or less wavelength independent in the emission spectral range of the light emitting diodes. Based on such an optical design an interferometric monolithic silicon device was built and, along with itsfluidic and electrical interface, was tested as a refractive index and multianalyte label-free protein sensor.
2. Monolithic device
The monolithic optocoupler in Fig. 1(a) integrates silicon avalanche diode LEDs along with silicon nitride waveguides and p/n junction detectors through mainstream silicon technology. The photonic path diagram and the waveguided broad spectrum of the optocoupler-interferometer is illustrated in Figs. 1(b) and 1(c). The quantum efficiency of the avalanche LED is not as high as in the nanocrystal case [7]. In our case, however, light emission is confined in a very narrow region which makes possible high coupling efficiencies in such optocouplers in association with self-alignment techniques [10–13] that place the emitting junction right under the waveguide front edge, Fig. 1(a). So far, these optocouplers employed multimode waveguides and, as such, the relevant optics was non-interferometric while detection was enabled through extinction labels [11,13] or cantilever structures [12]. Here, monomodal rib waveguides form the Mach-Zehnder interferometer (MZI) and the p/n junction detector detects only the fundamental mode while higher order modes diffuse away before reaching the detector. The sensing arm is exposed to the cover media or adlayers while the reference arm lies under a thick overcladding oxide.
Fig. 1 Monolithic silicon optocoupler-interferometer. (a). Optocoupler including the avalanche LED (1), the silicon nitride waveguide and the Mach-Zehnder interferometer (2), the silicon p/n junction detector (3) and the SiO2 bending spacers (4). The field oxide is 3 microns thick. The spacers provide for the smooth bending of the waveguides to reduce associated losses when the waveguide goes from vertical to horizontal [11]. The metallurgical p/n junction of the avalanche diode is self-aligned to the up-going segment of the waveguide by implanting the P++ emitter through the nitride core [11] to assure high coupling efficiency. On top of the chip the fluidic compartment is shown. More details on the fluidics and interconnects are provided later. (b). Optical interconnect diagram. For reasons relevant to achieving maximum self-alignment, the waveguide at the LED starts as a two micron multimode strip waveguide. The LED is in contact with the strip waveguide (dark colored) which in turn connects to the rib waveguide (light colored) through two back to back tapers. The two tapers are introduced to reduce the strip-rib conversion losses. The rib waveguide is 1.25 microns wide and a 4 nm etch depth. The core thickness is 167 nm for the reference arm, the same as the nitride slab thickness. The sensing arm is 162 nm thick. The reference-sensing arm core thickness differential was obtained by over-etching the nitride layer at the sensing window. The interferometer is placed between the second taper and the detector, where the rib waveguide is monomodal. (c). Spectra of the TE and TM mode in the single mode rib waveguide. The spectrum is recorded at the emitting edge of a semi-integrated chip obtained by cleaving off the detector. The inset shows the total emitted mode profile at the cleaved edge of a semi-integrated chip. The nitride core is also shown as a thin horizontal slab.
When a broad-band source excites the interferometer input, the output spectrum will be modulated by the difference in the phase shifts experienced by light propagating through the two arms. Let Nr and Ns be the effective indices of the reference and the sensing arms, respectively. Then, the transfer function of the MZI with a sensing arm length L is:
where Ιin, Ιout are the input and output power at a specific wavelength λ, ΔNrs is the effective index difference between the two arms,, and φ(λ) is the phase difference,. Here we assume that the overcladding oxide refractive index (nox) is bigger than the cover medium one (nc) and, consequently, ΔNrs(λ)is positive, to relax this assumption later. The frequency response of the transfer function (Ιout/Ιin) can be tailored to the desired shape by choosing the reference and sensing arm thickness [14,15]. Here, the desired shape in the emission spectral range of the LED is a flat transfer function, and the next paragraph shows how this is possible for the fundamental TE mode.The phase difference, φ(λ), in Eq. (1) is an increasing function of the wavelength in the blue since for diminishing wavelengths ΔNrs(λ)/λ becomes proportional to λ due to the core confinement of the wavefunctions [14]. In the infrared, on the other hand, and near cut-off the wavefunctions mainly lie in the cladding and the two arm effective indices tend to nox. Consequently, ΔNrs(λ), as well as ΔNrs(λ)/λ, are decreasing functions of λ. Between the increasing and decreasing phase spectral regions, a relatively flat maximum exists giving rise to a bell shaped function, as shown in Fig. 2. At the maximum, λm, the wavelength derivative of the phase difference goes to zero, or equivalently
The position of λm, as determined by Eq. (2), and the spectral location of the flat region around λm, are functions of the reference and sensing arms relative thickness: wavelength λm shifts to the blue as the waveguide cores get thinner and, especially, as we introduce a core thickness differential by thinning the sensing waveguide relative to the reference one [14,15].Fig. 2 Simulated phase φ(λ), (a), and output-to-input power ratio, (b), for the TE polarization and water solutions as cover media. Here L = 600 μm. (a) The phase has a maximum in the 700-800 nm region and drops towards the green and the near IR. As the cover medium RI increases from 1.335 to 1.338 and then to 1.34 the phase curve merely moves down by δNs/λ which is more or less wavelength independent [15]. At 750 nm the simulated phase sensitivity on nc is 592 rads/RIU. (b) The intensity ratios of Eq. (1). In the middle region, 650-900 nm, the curve goes monotonically from the maximum to the minimum when nc changes from 1.335 to 1.34. The mode effective indices Nr, Ns were obtained from the FemSIM software package (SYNOPSYS). The wavelength dispersions of the nitride core and claddings were taken into account.
The simulated phase for the fundamental TE mode shown in Fig. 2(a) corresponds to an arm thickness of 167 nm (reference) and 162 nm (sensing). The two thicknesses were chosen so that λm lies in the middle of the waveguided spectrum. The 5 nm core thickness differential is obtained by an extra etching step on the exposed sensing arm. When nc increases by δnc, Νs increases by δΝs, the difference ΔNrs(λ) drops by δΝs and the bell shaped plot in Fig. 2(a) moves downwards by 2πLδΝs/λ. Such a move is nearly a solid translation along the vertical axis as a result of the δΝs/λ term being nearly independent of the wavelength [15].
Consequently, the increasing segment (blue-green region) makes a red shift while the decreasing one (IR region) makes a blue shift, Fig. 2(a). At the same time, the middle region with the almost flat phase,, behaves as a solid spectral band and moves only up and down upon cover medium changes. The spectral extent of the flat region arround λm is inversely proportional to the quare root of L according to:
where c is the second derivative of the ΔNrs(λ)/λ ratio at λm. The above equation results from Eq. (2) as the first derivate of the ΔNrs(λ)/λ disappears at λm. The δΝs/λ ratio determines the phase semsitivity to cover medium changes asThe same spectral shift trends appear in the transfer function (Ιout/Ιin) shown in Fig. 2(b): With increasing nc, red spectral shifts are expected below 650 nm while above 900 nm blue shifts are expected. In the middle 650-900 nm region the behaviour is that of a solid spectral band following sinusoidally the phase changes as in Eq. (1). Through Eq. (3), the selection of L (600 μm) was such that this middle region matches the waveguided spectrum, Fig. 1(c). The 600 μm value for L will be used in the experimental results that follow. Consequently, monochromatic type Mach-Zehnder interferometry is expected, despite the broad-band source. Now, the nearly wavelength independent parameter δΝs/λ must change by 1/L for a full swing to occur. The above analysis holds even in case nox<nc, but now the flat region will be a minimum and the shifts change sign.
3. Fluidic and electrical interfacing
To demonstrate the analytical potential of the monolithic device and its suitability in point of care testing, a proper fluidic and electrical interfacing strategy was chosen to assure small size, chip re-usability and inexpensive packaging. That includes an on-chip fluidic cell and an associated probe head for the electrical-fluidic interface. A shallow fluidic chamber was mounted on top of the chip, Figs. 3(a) and 3(b), to supply liquids through the inlet-outlet conical holes. The fluidic chamber is made by the milling of 800µm thick Polymethylmethacrylate sheets and achieves a water proof seal with the chip surface through a photo-patterned sealing ring, Fig. 3(b). The ring is a 75µm thick dry film photoresist laminated onto the milled PMMA sheet and patterned to the desired shape that excludes sharp corners to avoid bubble trapping. The top side of the photoresist ring is sticky so that by pressing the cover against the aligned die the fluidic cover is attached and seals the chip. The contact pads of the optocoupler were contacted through spring loaded pins mounted on the probe head, Fig. 3(c). This way the electrical interfacing is removable (no bonding wires required) and the chips become reusable. The same probe head is equipped with two tapered steel cannulas, Figs. 3(c) and 3(d), that secure the fluidic interface at the conical holes that provide for auto-alignment. The probe head is aligned on top of the chip with the help of rods protruding out of the brass chuck where the chip is placed, Figs. 3(c)–3(e). As shown in Figs. 3(a) and 3(b), the 10 integrated MZIs share the same detector and are either multiplexed by biasing one LED at a time so that all 10 interferometers can be interrogated, or just one LED is biased and the photocurrent of the photodetector is continuously monitored.
Fig. 3 Packaged chips (a,b) and probe-fluidic head (c,d,e). (a) Interferometric chip with the fluidic cover on top. The pads on the left are the 10 LED emitter contacts and grounds. The ten interferometers converge on the same detector (bottom right contact). The two conical (45° slant to auto-align the cannula to the 300µm cylindrical via hole) holes placed diagonally on the fluidic cover are the inlet and outlet ports that hermetically seal with the spring loaded cannulas of the probe head. (b) Top view schematic of the chip in (a) showing the 10 MZIs converging on the same detector (right). The exposed arms are numbered and shown as the dark straight line segments on the waveguides. The curved green strip around the exposed arms is the sealing ring. The blue diagonal circles are the inlet-outlet holes on the fluidic cover. The yellow lines are the metal interconnects. The three exposed arm subsets (a,b,c) indicate the three regions to be spotted by different molecules when running the biosensing experiments. The chip dimensions are 4X9 mm2. (c) Probe-fluidic head. The packaged chip is sitting in a recess on a brass chuck while the protruding rods serve as indexing pins for the alignment of the probe-fluidic head coming from top. The spring loaded pins, 1, (aligned to the metal pads) and the cannulas, 2, (aligned to the conical fluidic cover holes) are visible. (d) Fluidic head close-up emphasizing the spring loaded pins and the cannula. (e). Vertical assembly of the upper (head) and the lower part (chip) shown in (c).
4. Experimental results and discussion
The response upon cover medium changes is shown in Fig. 4. Here the response includes the spectral output of the semi-integrated chips in addition to the photocurrent transient of the fully integrated chips. The semi-integrated chips are obtained by cleaving the integrated detector so that the waveguided light can be captured by coupling the chip emitting edge to a spectrometer (Maya Pro 2000, Ocean Optics) through an external fiber. The light spectra recorded are shown in Figs. 4(a) and 4(b). The TE polarization alone, Fig. 4(a), obtained with the use of a polarizer between the chip edge and the external fiber, shows the spectrum going up and down as a Gaussian band following the cover medium RI changes, as predicted by the theory outlined above. The evolution in time of the TE spectra is shown in Media 1. In addition to the TE, the TM mode is also excited at theLED-waveguide interface [15]. If the total (TE + TM) light emitted is monitored (no polarizer) then the combined spectrum, Fig. 4(b), also swings up and down while the TM fringes at a higher frequency move to the blue as the medium index increases. In fact, the TM bell shaped region lies in the blue [14]. The evolution in time of the combined spectra is shown in Media 2. In total, the TM photon sum is practically independent from the cover medium index since several periods are squeezed in the emission band. By summing up all the photon counts measured by the spectrometer we obtain the sinusoidal curve of Fig. 4(c). The spectrometer photon sum plot compares well with the actual photodetector response of the fully integrated device shown in Fig. 4(d) and obtained for the same RI swing. The small deviations are due to the different spectral responses of the integrated detector and the external spectrometer. The results shown in Fig. 4 demonstrate in a direct way that it is possible to photonically engineer Mach-Zehnder interferometers that behave as band-pass filters of entire spectral regions in the visible and near IR. The transmission coefficient of such this filter is a function of the cover medium and a function of any adlayers on top of the sensing arm.
Fig. 4 Experimental results for the flat region operation in the case of the semi-integrated (a,b,c) and integrated version (d) for the MZI described above (ts = 162nm, tr = 167nm, L = 600 μm). The cover medium RI change was 1.2x10−2 RIU: water (initial) to isopropanol solution (16,66% solution) and then back to water (final). (a). TE, only, spectra from semi-integrated chips at initial-water (0), lowest intensity (1), highest intensity (2) and propanol position (3). The full transient is shown in Media 1. (b). Spectra from semi-integrated chips of the total mode output (TE + TM) at the same points as in (a). The full transient is shown in Media 2. (c). Plot of the sum of the TE + TM counts in (b) as a function of time during the cover medium transition: water (1) - isopropanol solution (2) - water (1). A 2.16π phase oscillation is measured. (d). Integrated chip photocurrent for the same RI transition as in (c). A 2.22π phase oscillation is measured. Curves 1 and 2 in (a) and (b) correspond, respectively, to the lowest and highest points in (c) and (d). Here, one LED was biased and the ouput photocurrent was monitored through a KEITHEY 6517A femto-amperometer with a rate of 1Hz.
The sensitivity plot of the photocurrent phase, Φ, as a function of δnc is obtained by repeating cover medium transitions from water to isopropanol solutions with increasing degrees of dilution covering a range for δnc from 1.2x10−2 to 3.5x10−5 RIU. The phase is calculated from Eq. (1) as:
where Ispp, Idpp are the semi-sum, (Imax + Imin)/2, and semi-difference, (Imax-Imin)/2, of the highest (Imax) and the lowest values (Imin) of the recorded photocurrent (Iph). The introduction of the semi-sum and semi-difference photocurrent values are necassary to isolate the dc component of the photocurrent transient. The sensitivity plot is shown in Fig. 5 and the phase sensitivity, δΦ/δnc from Eq. (4), is measured at SΦ = 581 rad/RIU. Such a value is in excellent agreement with the simulated phase sensitivity (592 rad/RIU) shown in Fig. 2. It is also in good agreement with phase shift sensitivities (603 rad/RIU) obtained from the spectral shifts of other semi-integrated devices [15] for the same L value (600 μm) but with slightly thinner sensing arm core (150 nm). The limit of detection depends on the photocurrent noise which was measured at Inrms = 19 fA (rms). From Eq. (5) and with Idpp measured at 9 pA we obtain a phase noise, Φnrms, equal to 2.11x10−3 rads. Consequently, the limit of detection (LOD) as obtained [16] from Eq. (6)is calculated at LOD = 1.09x10−5 RIU. Such a an LOD agrees with the high dilution results in Fig. 5 and provides a measure for the range of applications of the RI sensor. The above LOD value is about 4 times the one (2.75x10−6 RIU) obtained from the spectral shift analysis of the semi-integrated devices [15] with an L value at 2 mm. Such an LOD ratio basically reflects the L ratio (2/0.6 = 3.33) of the two devices in accordance with Eq. (6).Fig. 5 Sensitivity plot of the photocurrent phase as a function of the cover medium RI change. The volume dilution starts at 1/6 (δnc = 1.2x10−2 RIU) and ends up at 1/2000 (δnc = 3.6x10−5 RIU).
The same integrated interferometer can be employed as a multianalyte biosensor provided the sensing arm set is spotted with the appropriate panel of probe molecules. The binding reaction will cause again a transition region swing and a sinusoidal signal at the output photocurrent. Following the reasoning preceeding Eq. (4), biomolecular binding causes Ns to increase, and an averaging change of δΝs/λ will cause an 2πLδΝs/λ change in the ouput signal phase. To demonstrate biosensing, the three sensing arm regions shown in Fig. 3(b) where spotted with mouse IgG (waveguides 1,4,7,10, region a), b-BSA (waveguides 2,5,8, region b) and plain BSA (waveguides 3,6,9, region c). The spotted chips had their fluidic cover placed on top and then were placed on the docking station shown in Fig. 3(c). The fluidic interface supplied the reagent sequence while the multiplexed excitation and detection electronics interrogated all 10 waveguides. The demultiplexed photodetector transient, Fig. 6, shows a fast response of the b-BSA spotted waveguides upon intoduction of the streptavidin solution and a tempered response of the mouse IgG spotted waveguides to the introduction of the antibody. Either reaction lasts for 20 minutes. On the other hand, the BSA spotted waveguides (3,6,9) remain insensitive to either streptavidin or anti-Mouse IgG. The response for the 1 nm streptavidin-biotin reaction was 3.33, 3.54 and 3.75 radians for interferometers 8, 5 and 2, respectively. The response for the 10 nm antibody-antigen reaction was 1.266, 1.1, 0.95, and 1.03 radians for interferometers 10, 7, 4, and 1, respectively. The response variation within a single molecule reflects the different flow rates at various sites. For example, waveguide 10 is closest to the inlet hole and has the largest response between the antigen spotted waveguides. The variation between molecules reflects the high binding rate constant of the biotin-streptavidin reaction compared to the antigen-antibody one.
Fig. 6 Demultiplexed response of the spotted chip showing the normalized photocurrent for all 10 interferometers which are numbered as in Fig. 3(b). The multiplexer interrogates all 10 waveguides every 1 second and supplies the photocurrent to a readout chain that provides amplification and signal conditioning. The values shown are normalized with respect to a unit photocurrent of 25 pA. The fluidic interface supplies the buffer and reagent sequence; assay buffer (1% BSA in PBS solution), 1 nM streptavidin solution in assay buffer (20 min), assay buffer (5 min), 10 nM anti-mouse IgG in assay buffer, (20 minutes), assay buffer. The signal noise appears higher compared to the one in Fig. 4 because here the signal integration time was 0.1 s compared to 1 s in Fig. 4.
5. Conclusions
In summary, a monolithic Mach-Zehnder interferometer integrated on silicon was presented and demonstrated as a refractive index and bio-chemical sensor. The interferometer is photonically engineered to handle the type of broad-band light sources integrated on silicon. The sensing and reference arm core thicknesses are chosen so that the phase difference of the light travelling through them is more or less wavelength independent in the emission spectral range of the integrated avalanche diode emitters. As a result, sinusoidal type of detector photocurrents are measured with cover medium and adlayer changes similar to the monochromatic MZIs. A limit of detection at 10−5 RIU was demonstrated. This is the only silicon device with a complete on chip interferometry and is proposed as the core element in point of need portable analytical devices.
Acknowledgments
This work was supported by the EU-funded Projects “PYTHIA” (FP7-ICT-224030) www.pythia-project.eu and “FOODSNIFFER” (FP7-ICT-318319) www.foodsniffer.eu.
References and links
1. A. P. F. Turner, “Biosensors - sense and sensitivity,” Science 290(5495), 1315–1317 (2000). [CrossRef] [PubMed]
2. X. Fan and I. M. White, “Optofluidic microsystems for chemical andbiological analysis,” Nat. Photon. 5(10), 591–597 (2011). [CrossRef] [PubMed]
3. C. Monat, P. Domachuk, and B. J. Eggleton, “Integrated optofluidics: A new river of light,” Nat. Photon. 1(2), 106–114 (2007). [CrossRef]
4. R. G. Heideman and P. V. Lambeck, “Remote opto-chemical sensing with extreme sensitivity: design, fabrication and performance of a pigtailed integrated optical phase-modulated Mach–Zehnder interferometer system,” Sens. Actuat. B 61(1–3), 100–127 (1999). [CrossRef]
5. P. Kozma, F. Kehl, E. Ehrentreich-Förster, C. Stamm, and F. F. Bier, “Integrated planar optical waveguide interferometer biosensors: A comparative review,” Biosens. Bioelectron. 58, 287–307 (2014). [CrossRef] [PubMed]
6. P. Preechaburana, M. C. Gonzalez, A. Suska, and D. Filippini, “Surface plasmon resonance chemical sensing on cell phones,” Angew. Chem. Int. Ed. 51(46), 11585–11588 (2012). [CrossRef]
7. A. Anopchenko, A. Marconi, F. Sgrignuoli, L. Cattoni, A. Tengattini, G. Pucker, Y. Jestin, and L. Pavesi, “Electroluminescent devices based on nanosilicon multilayer structures,” Phys. Status Solidi A 210(8), 1525–1531 (2013). [CrossRef]
8. A. G. Nassiopoulos, S. Grigoropoulos, L. Canham, A. Halimaoui, A. Berbezier, I. Gogolides, and E. Papadimitriou, “Submicrometer luminescent porous silicon structures using lithographically patterned substrates,” Thin Solid Films 255(1–2), 329–333 (1995). [CrossRef]
9. A. G. Nassiopoulos, S. Grigoropoulos, E. Gogolides, and E. Papadimitriou, “Visible luminescence from one-dimensional and 2-dimensional silicon structures produced by conventional lithographic and reactive ion etching techniques,” Appl. Phys. Lett. 66(9), 1114–1116 (1995). [CrossRef]
10. A. Chynoweth and K. Mckay, “Photon emission from avalanche breakdown in silicon,” Phys. Rev. 102(2), 369–376 (1956). [CrossRef]
11. K. Misiakos, S. E. Kakabakos, P. S. Petrou, and H. H. Ruf, “A monolithic silicon optoelectronic transducer as a real-time affinity biosensor,” Anal. Chem. 76(5), 1366–1373 (2004). [CrossRef] [PubMed]
12. K. Misiakos, I. Raptis, A. Gerardino, H. Contopanagos, and M. Kitsara, “A monolithic photonic microcantilever device for in situ monitoring of volatile compounds,” Lab Chip 9(9), 1261–1266 (2009). [CrossRef] [PubMed]
13. E. Mavrogiannopoulou, P. S. Petrou, S. E. Kakabakos, and K. Misiakos, “Real-time detection of BRCA1 gene mutations using a monolithic silicon optocoupler array,” Biosens. Bioelectron. 24(5), 1341–1347 (2009). [CrossRef] [PubMed]
14. M. Kitsara, K. Misiakos, I. Raptis, and E. Makarona, “Integrated optical frequency-resolved Mach-Zehnder interferometers for label-free affinity sensing,” Opt. Express 18(8), 8193–8206 (2010). [CrossRef] [PubMed]
15. K. Misiakos, I. Raptis, A. Salapatas, E. Makarona, A. Botsialas, M. Hoekman, R. Stoffer, and G. Jobst, “Broad-band Mach-Zehnder interferometers as high performance refractive index sensors: Theory and monolithic implementation,” Opt. Express 22(8), 8856–8870 (2014). [CrossRef] [PubMed]
16. I. M. White and X. Fan, “On the performance quantification of resonant refractive index sensors,” Opt. Express 16(2), 1020–1028 (2008). [CrossRef] [PubMed]