- Email: [email protected]
A compact refractometric sensor based on grated silicon photonic wires
Sensors and Actuators B 139 (2009) 194–198
Contents lists available at ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
A compact refractometric sensor based on grated silicon photonic wires L.J. Kauppinen ∗ , H.J.W.M. Hoekstra, R.M. de Ridder Integrated Optical MicroSystems Group, MESA+ Research Institute for Nanotechnology, University of Twente, 7500 AE Enschede, The Netherlands
a r t i c l e
i n f o
Article history: Available online 24 December 2008 Keywords: Grating Waveguide Sensor Silicon Photonic wire
a b s t r a c t Grated silicon photonic wires for refractometric applications have been fabricated using a 248-nm deep UV lithography. It is shown experimentally, that a device with length of only 180 m has an index sensitivity of 10−6 assuming a detector power resolution of 1%. It is also demonstrated that the device is suitable to monitor index changes in a liquid cladding, which could be used to monitor on chip chemical reactions. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Sensitive optical refractometric sensors, for example those based on Mach-Zehnder interferometers [1], are relatively large with length dimensions in the order of millimetres. In order to reduce the size of such sensors, the interaction between the optical field and the environment to be monitored should be increased. This can be achieved by employing waveguide structures exhibiting strong refractive index contrasts and multiple interference. Silicon photonic wires are particularly interesting for evanescent field sensors, since such waveguides may have a high sensitivity to cladding index changes [2], and the potential compactness of such devices enable a dense functional integration. Apart from pure bulk index sensing, Si photonic wires can be functionalized to versatile biosensor by immobilizing biointerface on the silicon surface [3]. In such a sensor, the sensitivity on the upper cladding index change and compact size are key factors. It is well known that a partly grated waveguide defines a cavity in the grated region for wavelengths outside the photonic stop band. For wavelengths near the stop band edge, particularly sharp spectral features can be observed, which are related to the strong modal dispersion and to the increased modal reflectivity in that wavelength region [4]. These sharp spectral phenomena exhibit a strong sensitivity to changes of the ambient refractive index, which can be exploited in sensor applications [5].
∗ Corresponding author at: Faculty of Electrical Engineering, Mathematics, and Computer Science, Integrated Optical MicroSystems, P.O. Box 217, 7500 AE Enschede; The Netherlands. Tel.: +31 53 4893857. E-mail address: [email protected] (L.J. Kauppinen). 0925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2008.12.023
We have investigated the refractometric sensing properties of grated silicon photonic wires (GSPW) fabricated in the 248-nm deep UV lithography that originally has been developed for the fabrication of CMOS integrated electronic circuits. Although very expensive processing equipment is needed to fabricate state of the art silicon CMOS devices, the investments have been justified by the economy of mass production which has proven to bring down the price per device to a sufficiently low level so as to open markets to the devices. The ability to use the same standardized process to produce a wide variety of devices can be claimed to be at least partly the reason behind the CMOS glory path. It has already been demonstrated that the tools that are used for the modern CMOS technology can be used to fabricate micro- and nanophotonic devices as well [6]. The research described in this paper was partly motivated by the desire to explore the interesting possibility to fabricate state of the art photonic devices with potentially low cost by sharing the equipment with electronics industry and research, and by applying the same standardized processes.
2. Fabrication and design of grated silicon photonic wires The GSPWs have been fabricated using the standard 248-nm deep-UV lithography process for photonic crystals and photonic wires established at IMEC, Leuven [7]. In this process the photonic devices are patterned in the 220 nm thick top layer of a silicon on insulator (SOI) wafer, using inductively coupled plasma (ICP) etching. An additional etch step is used to make grating couplers, that enable easy and low loss light coupling in to the Si waveguides [8]. The design of the optical structures is constrained mainly by the used waveguide material silicon which is transparent at wave-
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
Fig. 1. Lay-out of a grated silicon photonic wire. Waveguide widths w and d are of the order of 400–300 nm, = 450 nm and the waveguide thickness t is 220 nm.
lengths >1100 nm with a refractive index of 3.45 in the near infrared, as well as by the process limitations requiring a grating period >400 nm, silicon feature size >200 nm, and air feature size (gap between two Si features) >150 nm. We have realized the grating structure by periodically varying the waveguide width as shown in Fig. 1. In order to have the first order photonic stop band in the 1.5 m wavelength region of the available equipment, the dimensions for the wide (w) and narrow (d) segments need to be carefully chosen. Although only timeconsuming three-dimensional calculations can provide full and accurate information on the transmission properties of the structure [9], we have chosen to use a highly efficient two-dimensional method. The stop band location is given by the well-known equation: B = 2nav ,
(1)
where B is the Bragg wavelength, the period of the grating, and nav the average modal index of the grating, found by averaging modal indices of the narrow (nd ) and wide (nw ) waveguide section over one grating period. If the length ratio of the narrow to the wide waveguide sections is (duty cycle), the average index can be calculated as nav = nd + (1 − )nw . A 2D modesolver was used to calculate the allowable range for w and d, such that the waveguiding sections are single mode. The same mode solver also produced the corresponding effective refractive indices nw and nd . Based on the modal field calculations and taking into account the aforementioned process constraints, we have chosen the structural dimensions d and w to be between 300 and 400 nm, the grating period = 450 nm, and the duty cycle = 0.45. The difference w − d is varied from 20 nm to 90 nm. In the fabrication process an exposure dose sweep is performed across the wafer. The sweep will give additional variety of the fabricated gratings, as the size of small features on a wafer depends on the exposure dose. As the difference w − d increases, the index difference nw − nd increases as well. This leads, next to a wider stop band, to a higher modal reflectance at the onsets of the grating, but also to higher losses on reflection at these onsets. The higher reflectance leads to sharper spectral features on the one hand, which are obscured on the other hand by the increased losses [9]. The slope of the stop band edge is one of the factors that determine the sensitivity of the GSPW to refractive index changes as will be discussed in Section 3. The optimum GSPW for the envisioned sensor experiment has been selected via an experimental study of the stop band spectral features. Due to the fundamental limitations of the photolithographic process, sharp corners will not be faithfully reproduced, resulting in smoothened corners which become especially obvious in narrow structures. This smoothening effect is clearly visible in Fig. 2,
195
Fig. 2. SEM picture of 350 nm wide grated photonic wire. Lithography smoothens the rectangular shape and a sinusoidal-like grating is obtained.
showing that the grating shape, which had been designed to be rectangular, has deteriorated to a sinusoidal-like one. Nevertheless, despite the change from a rectangular to a sinusoidal grating, the stop band positions remained within the desired wavelength range (1550 ± 50 nm). 3. Bulk index sensing with grated silicon photonic wires The change in power T transmitted through the grating is in firstorder approximation directly proportional to the change in cladding index, nclad . The constant of proportionality is given by the product of the slope of the stop band edge, ∂T/∂, and the sensitivity of the Bragg wavelength B to nclad , ∂B /∂nclad . From this, the resolution in determining cladding index changes is given by [10] nclad =
∂nclad ∂ T ∂B ∂T
(2)
where T is the minimum detectable change of optical power, which depends on, among many things, laser power fluctuation and dark current noise of the photodetector. The sensitivity ∂B /∂nclad is composed of the dependence of B on the average modal index nav – as given in (1) – and the relation between nav and nclad , which depends on the geometry of the structure, and the material properties. The slope of the stop band edge depends on the length and strength (i.e. index contrast between high- and low-index regions) of the grating [5]. In order to get an estimate of the index resolution, the stop band position shift and the band edge slope were measured. The stop band shift of a 400 period grating was determined by dropping a water droplet on the sample area (10 mm × 5 mm). The water droplet (nclad = 1.32) provides an index shift of 0.32 in the upper cladding. The measured stop band positions before and after depositing the water droplet are shown in Fig. 3. It can be seen that the shift is roughly 60 nm. The slope of the stop band edge at a measured power level (1 nW) is 20 nW/nm. Assuming that at this power level the noise level enables a measurement resolution T of 10 pW, a resolution of nclad = 10−6 can be obtained. For highsensitivity measurements one should also consider the effect of the thermo-optic coefficient of the materials. For a silicon-based sensor the thermo-optic coefficient of silicon (16 × 10−5 K−1 ) is limiting the index resolution unless temperature is stabilized. A reference sensor can be used to compensate the effect of temperature fluctuation on refractive index and geometry of the sensor structure, as well as the refractive index of the main solvent for the species to be measured. The corner smoothening effect, discussed above, reduces the index contrast between high- and low-index segments. This effect
196
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
Fig. 3. Stop band positions with air (left) and water (right) claddings.
Fig. 5. Transmission spectra, measured at the beginning and at the end of the experiment.
likely reduces the obtainable sensitivity as a higher index contrast would provide a steeper band edge slope, although the effect of increased losses due to higher contrast should be considered [9]. The suitability to use the GSPW to measure small refractive index changes at a fixed wavelength, and therefore exploiting the spectral sharpness of the stop band edge, was evaluated by using a water-isopropanol solution. Isopropanol evaporates faster than water in ambient temperature and pressure conditions (air at 20 ◦ C and 1 atm pressure). Therefore the refractive index of a water-isopropanol droplet will decrease with time at some rate. Consequently, when a droplet of water-isopropanol solution is applied to a grating, a gradual stop band shift to lower wavelengths can be observed. To demonstrate this capability, a droplet of approximately 9 wt.% isopropanol solution was applied to the grating, and transmitted power at the stop band edge was recorded as a function of time, as shown in Fig. 4. The wavelength was fixed at 1550 nm. The data gap between 400 and 500 s was caused by a limitation of the measurement set-up. Fig. 5 shows two transmission spectra, one taken immediately after applying the solution and the other after 14 min had passed. The time required for a spectral scan is about 2 min. The
experiment was repeated by dropping a new droplet of the same solution. The initial and the final stop band positions of the first and the second experiment are nearly equal, as shown in Fig. 6. The refractive index of the droplet was measured with an Abbe refractometer before and after the experiment. The index of the solution had changed by 5 × 10−3 over the 16 min duration of the entire experiment. The measured index change agrees with the index change that can be calculated from the stop band shift: the stop band shift shown in Fig. 5 is roughly 1 nm, while from Fig. 3 it can be deduced that a shift of 1 nm corresponds to a refractive index change of 5.3 × 10−3 in the upper cladding material. Since the relation between power transmission and wavelength may be nonlinear, corresponding to non-constant ∂T/∂, a more accurate estimate of the sensitivity over a smaller refractive-index range has been made. First, ∂/∂n was estimated by measuring the wavelength shift of the long-wavelength band edge for a refractive index change n = 1.9 × 10−3 ± 25 × 10−5 (pure water, n = 1.3331, and water/isopropanol, n = 1.3350 ± 25 × 10−5 , as measured using an Abbe refractometer). From the transmission spectra shown in the inset of Fig. 7, the wavelength shift has been
Fig. 4. Transmitted power measured at the stop band edge (1550 nm) as a function of time during the evaporation of isopropanol from solution.
Fig. 6. Repeated measurement of the stop band position shift due to the evaporation of isopropanol. A fresh droplet of solution was used in the second experiment. Solid lines correspond to the first experiment.
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
197
Finally we note that low-loss light coupling into silicon waveguides needs careful attention. Especially when considering a packaged chip that houses arrays of sensor waveguides, the optical power reaching the detector should remain well above the noise level of the detector. Robust and efficient fiber-chip coupling solutions exist, for example involving an intermediate waveguide [11], which solve the stability problem mentioned above, and which reduce possible noise level problems. 4. Conclusions It was experimentally shown that using the near stop band edge phenomena of the GSPW, a resolution of the order of 10−6 to bulk index change is feasible assuming 1% noise level. It was also demonstrated that the device is suitable to measure the index change of a liquid cladding in real time, which could be used to monitor on chip chemical reactions. The compact size of the device enables the use of <1 pl sample volumes. The CMOS compatible fabrication process, mass production potential and compact size make these devices a very attractive sensor platform. Fig. 7. Experimentally obtained change of the cladding index versus the transmitted power measured at a wavelength of 1549.2 nm (see text). Inset: transmission spectra of the stop band edge of water and reference mixture: the indicated indices have been measured with an Abbe refractometor.
estimated to be = 0.30 ± 0.005 nm, leading to an estimated
∂/∂n = 160 ± 25 nm per RIU (refractive index units). The spectrum
for n = 1.3350 ± 25 × 10−5 was stored to be used as a reference for the next step. As a second step, a water/isopropanol mixture having a slightly higher index, n = 1.3365 ± 25 × 10−5 , as measured with an Abbe refractometer, was put on the sensor, and the transmission spectrum was continuously scanned and stored while the isopropanol slowly evaporated. This process was stopped after 5 scans, when the measured spectrum closely matched the previously measured reference spectrum for n = 1.3350 ± 25 × 10−5 . For each of the scans the average cladding index was estimated from the overall shift of the measured transmission spectrum, using ∂/∂n = 160 ± 25 nm/RIU as calculated in step 1. The transmitted power at = 1549.2 nm was recorded and attributed to the average cladding index for that scan. Finally, for each scan the calculated refractive index change and its uncertainty has been plotted versus the transmitted power at = 1549.2 nm, leading to the nonlinear response curve n(T) for real-time refractive-index monitoring, shown in Fig. 7. From the slope of the curve we obtain the estimated intrinsic sensitivity, ∂nclad /∂T, of the sensor. From Fig. 7 we find that ∂nclad /∂T is in the range of 1–2 × 10−4 RIU/nW. The index resolution of the sensor is simply a multiplication of the intrinsic sensitivity and the noise level T. For example, an index resolution of the order of 10−6 RIU can be obtained if T = 10−2 nW. The described monitoring of the band edge shift may be an attractive method to follow the progress of on-chip chemical or physical processes, provided that these processes cause a change of the refractive index. Since the index sensing part of the considered GSPW is only 180 m long and sub-micron wide, a very small sample volume (e.g. 180 m3 ) is sufficient for a measurement. The power levels of the transmission spectra measured at the beginning and at the end of the isopropanol evaporation experiment are different; see Figs. 5 and 6. This is due to alignment drift of the measurement set-up. A stabilization period before the experiment can be used to reduce the drift. However, in our end-fire set-up, the coupling conditions change when the droplet is applied to the sample, and due to evaporation of the isopropanol such a stabilization period can not be applied in this experiment.
Acknowledgements We thank Anton Hollink and Meindert Dijkstra their technical help. This research was supported by the Dutch Technology Foundation STW, applied science division of NWO and the Technology Program of the Ministry of Economic Affairs, and by the European Network of Excellence ePIXnet. References [1] R.G. Heideman, 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 (1999) 100–127. [2] A. Densmore, D.-X. Xu, P. Waldron, S. Janz, P. Cheben, J. Lapointe, A. Delâge, B. Lamontagne, J.H. Schmid, E. Post, A silicon-on-insulator photonic wire based evanescent field sensor, IEEE Photon. Technol. Lett. 18 (2006) 2520–2522. [3] K. De Vos, I. Bartolozzi, E. Schacht, P. Bienstman, R. Baets, Silicon-on-Insulator microring resonator for sensitive and label-free biosensing, Opt. Express 15 (2007) 7610–7615. [4] H.J.W.M. Hoekstra, W.C.L. Hopman, J. Kautz, R. Dekker, R.M. de Ridder, A simple coupled mode model for near band-edge phenomena in grated waveguides, Opt. Quantum Electron. 38 (2006) 799–813. [5] W.C.L. Hopman, P. Pottier, D. Yudistira, J. van Lith, P.V. Lambeck, R.M. De La Rue, A. Driessen, H.J.W.M. Hoekstra, R.M. de Ridder, Quasi-one-dimensional photonic crystal as a compact building-block for refractometric optical sensors, IEEE J. Select. Topics Quantum Electron. 11 (2005) 11–16. [6] W. Bogaerts, R. Baets, P. Dumon, V. Wiaux, S. Beckx, D. Taillaert, B. Luyssaert, J. Van Campenhout, P. Bienstman, D. Van Thourhout, Nanophotonic waveguides in silicon-on-insulator fabricated with CMOS technology, J. Lightwave Technol. 23 (2005) 401–412. [7] Silicon Photonics Platform, http://www.siliconphotonics.eu/. [8] D. Taillaert, F. van Laere, M. Ayre, W. Bogaerts, D. van Thourhout, P. Bienstman, R. Baets, Grating couplers for coupling between optical fibers and nanophotonic waveguides, Jpn. J. Appl. Phys. 45 (2006) 6071–6077. [9] M. Gnan, G. Bellanca, H.M.H. Chong, P. Bassi, R.M. de la Rue, Modelling of photonic wire Bragg gratings, Opt. Quantum Electron. 38 (2006) 133–148. [10] W.C.L. Hopman, H.J.W.M. Hoekstra, R. Dekker, L. Zhuang, R.M. de Ridder, Far-field scattering microscopy applied to analysis of slow light, power enhancement, and delay times in uniform Bragg waveguide gratings, Opt. Express 15 (2007) 1851–1870. [11] R. Costa, G. Cusmai, A. Melloni, M. Martinelli, TE-TM coupling of a standard fiber to a Si-wire waveguide, in: Proceedings of the 13th European Conference on Integrated Optics (ECIO’07), Copenhagen, Denmark, 2007, ThC3.
Biographies Lasse J. Kauppinen received his M.Sc. degree in electrical engineering from the Helsinki University of Technology in 2006. He is currently working toward the Ph.D. degree in integrated optics at the University of Twente. His research interests are slow-light devices for applications in optical switching and sensing. Hugo J.W.M. Hoekstra received the M.Sc. degree in experimental physics from the University of Amsterdam, Amsterdam, the Netherlands in 1977 and the Ph.D. degree
198
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
on magneto-optical effects in solids from the University of Groningen, Groningen, The Netherlands in 1984. Prior to receiving the Ph.D. degree, he was a high school teacher. From 1984 to 1988 he was a Postdoc-fellow in the field of surface science at both the University of Nijmegen, the Netherlands and the University of Groningen. In 1988, he joined the Integrated Optical Micro Systems group, MESA+ research institute, University of Twente, Enschede, the Netherlands. His current research interests include optical waveguide theory and integrated optical sensors. He is (co-) author of about 70 refereed scientific papers. René M. de Ridder was born 1950 in Amsterdam, the Netherlands. He received the M.S. and Ph.D. degrees in electrical engineering from the University of Twente,
Enschede, The Netherlands, in 1978 and 1988, respectively. His thesis subject was on sensitive thin-film magnetic-field sensors. Since 1981, he has been a Lecturer and Research Scientist with the University of Twente, where he is currently a member of the Integrated Optical MicroSystems Group in the MESA+ Institute for Nanotechnology. In 1988, his research focus turned to integrated optical devices. A basis for his work in this field was laid during a half year of sabbatical leave at (then) AT&T Bell Labs, Holmdel, NJ, in 1989. His current research interests include photonic crystal structures and wavelength-selective devices for applications in sensing and communication.
Contents lists available at ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
A compact refractometric sensor based on grated silicon photonic wires L.J. Kauppinen ∗ , H.J.W.M. Hoekstra, R.M. de Ridder Integrated Optical MicroSystems Group, MESA+ Research Institute for Nanotechnology, University of Twente, 7500 AE Enschede, The Netherlands
a r t i c l e
i n f o
Article history: Available online 24 December 2008 Keywords: Grating Waveguide Sensor Silicon Photonic wire
a b s t r a c t Grated silicon photonic wires for refractometric applications have been fabricated using a 248-nm deep UV lithography. It is shown experimentally, that a device with length of only 180 m has an index sensitivity of 10−6 assuming a detector power resolution of 1%. It is also demonstrated that the device is suitable to monitor index changes in a liquid cladding, which could be used to monitor on chip chemical reactions. © 2008 Elsevier B.V. All rights reserved.
1. Introduction Sensitive optical refractometric sensors, for example those based on Mach-Zehnder interferometers [1], are relatively large with length dimensions in the order of millimetres. In order to reduce the size of such sensors, the interaction between the optical field and the environment to be monitored should be increased. This can be achieved by employing waveguide structures exhibiting strong refractive index contrasts and multiple interference. Silicon photonic wires are particularly interesting for evanescent field sensors, since such waveguides may have a high sensitivity to cladding index changes [2], and the potential compactness of such devices enable a dense functional integration. Apart from pure bulk index sensing, Si photonic wires can be functionalized to versatile biosensor by immobilizing biointerface on the silicon surface [3]. In such a sensor, the sensitivity on the upper cladding index change and compact size are key factors. It is well known that a partly grated waveguide defines a cavity in the grated region for wavelengths outside the photonic stop band. For wavelengths near the stop band edge, particularly sharp spectral features can be observed, which are related to the strong modal dispersion and to the increased modal reflectivity in that wavelength region [4]. These sharp spectral phenomena exhibit a strong sensitivity to changes of the ambient refractive index, which can be exploited in sensor applications [5].
∗ Corresponding author at: Faculty of Electrical Engineering, Mathematics, and Computer Science, Integrated Optical MicroSystems, P.O. Box 217, 7500 AE Enschede; The Netherlands. Tel.: +31 53 4893857. E-mail address: [email protected] (L.J. Kauppinen). 0925-4005/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2008.12.023
We have investigated the refractometric sensing properties of grated silicon photonic wires (GSPW) fabricated in the 248-nm deep UV lithography that originally has been developed for the fabrication of CMOS integrated electronic circuits. Although very expensive processing equipment is needed to fabricate state of the art silicon CMOS devices, the investments have been justified by the economy of mass production which has proven to bring down the price per device to a sufficiently low level so as to open markets to the devices. The ability to use the same standardized process to produce a wide variety of devices can be claimed to be at least partly the reason behind the CMOS glory path. It has already been demonstrated that the tools that are used for the modern CMOS technology can be used to fabricate micro- and nanophotonic devices as well [6]. The research described in this paper was partly motivated by the desire to explore the interesting possibility to fabricate state of the art photonic devices with potentially low cost by sharing the equipment with electronics industry and research, and by applying the same standardized processes.
2. Fabrication and design of grated silicon photonic wires The GSPWs have been fabricated using the standard 248-nm deep-UV lithography process for photonic crystals and photonic wires established at IMEC, Leuven [7]. In this process the photonic devices are patterned in the 220 nm thick top layer of a silicon on insulator (SOI) wafer, using inductively coupled plasma (ICP) etching. An additional etch step is used to make grating couplers, that enable easy and low loss light coupling in to the Si waveguides [8]. The design of the optical structures is constrained mainly by the used waveguide material silicon which is transparent at wave-
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
Fig. 1. Lay-out of a grated silicon photonic wire. Waveguide widths w and d are of the order of 400–300 nm, = 450 nm and the waveguide thickness t is 220 nm.
lengths >1100 nm with a refractive index of 3.45 in the near infrared, as well as by the process limitations requiring a grating period >400 nm, silicon feature size >200 nm, and air feature size (gap between two Si features) >150 nm. We have realized the grating structure by periodically varying the waveguide width as shown in Fig. 1. In order to have the first order photonic stop band in the 1.5 m wavelength region of the available equipment, the dimensions for the wide (w) and narrow (d) segments need to be carefully chosen. Although only timeconsuming three-dimensional calculations can provide full and accurate information on the transmission properties of the structure [9], we have chosen to use a highly efficient two-dimensional method. The stop band location is given by the well-known equation: B = 2nav ,
(1)
where B is the Bragg wavelength, the period of the grating, and nav the average modal index of the grating, found by averaging modal indices of the narrow (nd ) and wide (nw ) waveguide section over one grating period. If the length ratio of the narrow to the wide waveguide sections is (duty cycle), the average index can be calculated as nav = nd + (1 − )nw . A 2D modesolver was used to calculate the allowable range for w and d, such that the waveguiding sections are single mode. The same mode solver also produced the corresponding effective refractive indices nw and nd . Based on the modal field calculations and taking into account the aforementioned process constraints, we have chosen the structural dimensions d and w to be between 300 and 400 nm, the grating period = 450 nm, and the duty cycle = 0.45. The difference w − d is varied from 20 nm to 90 nm. In the fabrication process an exposure dose sweep is performed across the wafer. The sweep will give additional variety of the fabricated gratings, as the size of small features on a wafer depends on the exposure dose. As the difference w − d increases, the index difference nw − nd increases as well. This leads, next to a wider stop band, to a higher modal reflectance at the onsets of the grating, but also to higher losses on reflection at these onsets. The higher reflectance leads to sharper spectral features on the one hand, which are obscured on the other hand by the increased losses [9]. The slope of the stop band edge is one of the factors that determine the sensitivity of the GSPW to refractive index changes as will be discussed in Section 3. The optimum GSPW for the envisioned sensor experiment has been selected via an experimental study of the stop band spectral features. Due to the fundamental limitations of the photolithographic process, sharp corners will not be faithfully reproduced, resulting in smoothened corners which become especially obvious in narrow structures. This smoothening effect is clearly visible in Fig. 2,
195
Fig. 2. SEM picture of 350 nm wide grated photonic wire. Lithography smoothens the rectangular shape and a sinusoidal-like grating is obtained.
showing that the grating shape, which had been designed to be rectangular, has deteriorated to a sinusoidal-like one. Nevertheless, despite the change from a rectangular to a sinusoidal grating, the stop band positions remained within the desired wavelength range (1550 ± 50 nm). 3. Bulk index sensing with grated silicon photonic wires The change in power T transmitted through the grating is in firstorder approximation directly proportional to the change in cladding index, nclad . The constant of proportionality is given by the product of the slope of the stop band edge, ∂T/∂, and the sensitivity of the Bragg wavelength B to nclad , ∂B /∂nclad . From this, the resolution in determining cladding index changes is given by [10] nclad =
∂nclad ∂ T ∂B ∂T
(2)
where T is the minimum detectable change of optical power, which depends on, among many things, laser power fluctuation and dark current noise of the photodetector. The sensitivity ∂B /∂nclad is composed of the dependence of B on the average modal index nav – as given in (1) – and the relation between nav and nclad , which depends on the geometry of the structure, and the material properties. The slope of the stop band edge depends on the length and strength (i.e. index contrast between high- and low-index regions) of the grating [5]. In order to get an estimate of the index resolution, the stop band position shift and the band edge slope were measured. The stop band shift of a 400 period grating was determined by dropping a water droplet on the sample area (10 mm × 5 mm). The water droplet (nclad = 1.32) provides an index shift of 0.32 in the upper cladding. The measured stop band positions before and after depositing the water droplet are shown in Fig. 3. It can be seen that the shift is roughly 60 nm. The slope of the stop band edge at a measured power level (1 nW) is 20 nW/nm. Assuming that at this power level the noise level enables a measurement resolution T of 10 pW, a resolution of nclad = 10−6 can be obtained. For highsensitivity measurements one should also consider the effect of the thermo-optic coefficient of the materials. For a silicon-based sensor the thermo-optic coefficient of silicon (16 × 10−5 K−1 ) is limiting the index resolution unless temperature is stabilized. A reference sensor can be used to compensate the effect of temperature fluctuation on refractive index and geometry of the sensor structure, as well as the refractive index of the main solvent for the species to be measured. The corner smoothening effect, discussed above, reduces the index contrast between high- and low-index segments. This effect
196
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
Fig. 3. Stop band positions with air (left) and water (right) claddings.
Fig. 5. Transmission spectra, measured at the beginning and at the end of the experiment.
likely reduces the obtainable sensitivity as a higher index contrast would provide a steeper band edge slope, although the effect of increased losses due to higher contrast should be considered [9]. The suitability to use the GSPW to measure small refractive index changes at a fixed wavelength, and therefore exploiting the spectral sharpness of the stop band edge, was evaluated by using a water-isopropanol solution. Isopropanol evaporates faster than water in ambient temperature and pressure conditions (air at 20 ◦ C and 1 atm pressure). Therefore the refractive index of a water-isopropanol droplet will decrease with time at some rate. Consequently, when a droplet of water-isopropanol solution is applied to a grating, a gradual stop band shift to lower wavelengths can be observed. To demonstrate this capability, a droplet of approximately 9 wt.% isopropanol solution was applied to the grating, and transmitted power at the stop band edge was recorded as a function of time, as shown in Fig. 4. The wavelength was fixed at 1550 nm. The data gap between 400 and 500 s was caused by a limitation of the measurement set-up. Fig. 5 shows two transmission spectra, one taken immediately after applying the solution and the other after 14 min had passed. The time required for a spectral scan is about 2 min. The
experiment was repeated by dropping a new droplet of the same solution. The initial and the final stop band positions of the first and the second experiment are nearly equal, as shown in Fig. 6. The refractive index of the droplet was measured with an Abbe refractometer before and after the experiment. The index of the solution had changed by 5 × 10−3 over the 16 min duration of the entire experiment. The measured index change agrees with the index change that can be calculated from the stop band shift: the stop band shift shown in Fig. 5 is roughly 1 nm, while from Fig. 3 it can be deduced that a shift of 1 nm corresponds to a refractive index change of 5.3 × 10−3 in the upper cladding material. Since the relation between power transmission and wavelength may be nonlinear, corresponding to non-constant ∂T/∂, a more accurate estimate of the sensitivity over a smaller refractive-index range has been made. First, ∂/∂n was estimated by measuring the wavelength shift of the long-wavelength band edge for a refractive index change n = 1.9 × 10−3 ± 25 × 10−5 (pure water, n = 1.3331, and water/isopropanol, n = 1.3350 ± 25 × 10−5 , as measured using an Abbe refractometer). From the transmission spectra shown in the inset of Fig. 7, the wavelength shift has been
Fig. 4. Transmitted power measured at the stop band edge (1550 nm) as a function of time during the evaporation of isopropanol from solution.
Fig. 6. Repeated measurement of the stop band position shift due to the evaporation of isopropanol. A fresh droplet of solution was used in the second experiment. Solid lines correspond to the first experiment.
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
197
Finally we note that low-loss light coupling into silicon waveguides needs careful attention. Especially when considering a packaged chip that houses arrays of sensor waveguides, the optical power reaching the detector should remain well above the noise level of the detector. Robust and efficient fiber-chip coupling solutions exist, for example involving an intermediate waveguide [11], which solve the stability problem mentioned above, and which reduce possible noise level problems. 4. Conclusions It was experimentally shown that using the near stop band edge phenomena of the GSPW, a resolution of the order of 10−6 to bulk index change is feasible assuming 1% noise level. It was also demonstrated that the device is suitable to measure the index change of a liquid cladding in real time, which could be used to monitor on chip chemical reactions. The compact size of the device enables the use of <1 pl sample volumes. The CMOS compatible fabrication process, mass production potential and compact size make these devices a very attractive sensor platform. Fig. 7. Experimentally obtained change of the cladding index versus the transmitted power measured at a wavelength of 1549.2 nm (see text). Inset: transmission spectra of the stop band edge of water and reference mixture: the indicated indices have been measured with an Abbe refractometor.
estimated to be = 0.30 ± 0.005 nm, leading to an estimated
∂/∂n = 160 ± 25 nm per RIU (refractive index units). The spectrum
for n = 1.3350 ± 25 × 10−5 was stored to be used as a reference for the next step. As a second step, a water/isopropanol mixture having a slightly higher index, n = 1.3365 ± 25 × 10−5 , as measured with an Abbe refractometer, was put on the sensor, and the transmission spectrum was continuously scanned and stored while the isopropanol slowly evaporated. This process was stopped after 5 scans, when the measured spectrum closely matched the previously measured reference spectrum for n = 1.3350 ± 25 × 10−5 . For each of the scans the average cladding index was estimated from the overall shift of the measured transmission spectrum, using ∂/∂n = 160 ± 25 nm/RIU as calculated in step 1. The transmitted power at = 1549.2 nm was recorded and attributed to the average cladding index for that scan. Finally, for each scan the calculated refractive index change and its uncertainty has been plotted versus the transmitted power at = 1549.2 nm, leading to the nonlinear response curve n(T) for real-time refractive-index monitoring, shown in Fig. 7. From the slope of the curve we obtain the estimated intrinsic sensitivity, ∂nclad /∂T, of the sensor. From Fig. 7 we find that ∂nclad /∂T is in the range of 1–2 × 10−4 RIU/nW. The index resolution of the sensor is simply a multiplication of the intrinsic sensitivity and the noise level T. For example, an index resolution of the order of 10−6 RIU can be obtained if T = 10−2 nW. The described monitoring of the band edge shift may be an attractive method to follow the progress of on-chip chemical or physical processes, provided that these processes cause a change of the refractive index. Since the index sensing part of the considered GSPW is only 180 m long and sub-micron wide, a very small sample volume (e.g. 180 m3 ) is sufficient for a measurement. The power levels of the transmission spectra measured at the beginning and at the end of the isopropanol evaporation experiment are different; see Figs. 5 and 6. This is due to alignment drift of the measurement set-up. A stabilization period before the experiment can be used to reduce the drift. However, in our end-fire set-up, the coupling conditions change when the droplet is applied to the sample, and due to evaporation of the isopropanol such a stabilization period can not be applied in this experiment.
Acknowledgements We thank Anton Hollink and Meindert Dijkstra their technical help. This research was supported by the Dutch Technology Foundation STW, applied science division of NWO and the Technology Program of the Ministry of Economic Affairs, and by the European Network of Excellence ePIXnet. References [1] R.G. Heideman, 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 (1999) 100–127. [2] A. Densmore, D.-X. Xu, P. Waldron, S. Janz, P. Cheben, J. Lapointe, A. Delâge, B. Lamontagne, J.H. Schmid, E. Post, A silicon-on-insulator photonic wire based evanescent field sensor, IEEE Photon. Technol. Lett. 18 (2006) 2520–2522. [3] K. De Vos, I. Bartolozzi, E. Schacht, P. Bienstman, R. Baets, Silicon-on-Insulator microring resonator for sensitive and label-free biosensing, Opt. Express 15 (2007) 7610–7615. [4] H.J.W.M. Hoekstra, W.C.L. Hopman, J. Kautz, R. Dekker, R.M. de Ridder, A simple coupled mode model for near band-edge phenomena in grated waveguides, Opt. Quantum Electron. 38 (2006) 799–813. [5] W.C.L. Hopman, P. Pottier, D. Yudistira, J. van Lith, P.V. Lambeck, R.M. De La Rue, A. Driessen, H.J.W.M. Hoekstra, R.M. de Ridder, Quasi-one-dimensional photonic crystal as a compact building-block for refractometric optical sensors, IEEE J. Select. Topics Quantum Electron. 11 (2005) 11–16. [6] W. Bogaerts, R. Baets, P. Dumon, V. Wiaux, S. Beckx, D. Taillaert, B. Luyssaert, J. Van Campenhout, P. Bienstman, D. Van Thourhout, Nanophotonic waveguides in silicon-on-insulator fabricated with CMOS technology, J. Lightwave Technol. 23 (2005) 401–412. [7] Silicon Photonics Platform, http://www.siliconphotonics.eu/. [8] D. Taillaert, F. van Laere, M. Ayre, W. Bogaerts, D. van Thourhout, P. Bienstman, R. Baets, Grating couplers for coupling between optical fibers and nanophotonic waveguides, Jpn. J. Appl. Phys. 45 (2006) 6071–6077. [9] M. Gnan, G. Bellanca, H.M.H. Chong, P. Bassi, R.M. de la Rue, Modelling of photonic wire Bragg gratings, Opt. Quantum Electron. 38 (2006) 133–148. [10] W.C.L. Hopman, H.J.W.M. Hoekstra, R. Dekker, L. Zhuang, R.M. de Ridder, Far-field scattering microscopy applied to analysis of slow light, power enhancement, and delay times in uniform Bragg waveguide gratings, Opt. Express 15 (2007) 1851–1870. [11] R. Costa, G. Cusmai, A. Melloni, M. Martinelli, TE-TM coupling of a standard fiber to a Si-wire waveguide, in: Proceedings of the 13th European Conference on Integrated Optics (ECIO’07), Copenhagen, Denmark, 2007, ThC3.
Biographies Lasse J. Kauppinen received his M.Sc. degree in electrical engineering from the Helsinki University of Technology in 2006. He is currently working toward the Ph.D. degree in integrated optics at the University of Twente. His research interests are slow-light devices for applications in optical switching and sensing. Hugo J.W.M. Hoekstra received the M.Sc. degree in experimental physics from the University of Amsterdam, Amsterdam, the Netherlands in 1977 and the Ph.D. degree
198
L.J. Kauppinen et al. / Sensors and Actuators B 139 (2009) 194–198
on magneto-optical effects in solids from the University of Groningen, Groningen, The Netherlands in 1984. Prior to receiving the Ph.D. degree, he was a high school teacher. From 1984 to 1988 he was a Postdoc-fellow in the field of surface science at both the University of Nijmegen, the Netherlands and the University of Groningen. In 1988, he joined the Integrated Optical Micro Systems group, MESA+ research institute, University of Twente, Enschede, the Netherlands. His current research interests include optical waveguide theory and integrated optical sensors. He is (co-) author of about 70 refereed scientific papers. René M. de Ridder was born 1950 in Amsterdam, the Netherlands. He received the M.S. and Ph.D. degrees in electrical engineering from the University of Twente,
Enschede, The Netherlands, in 1978 and 1988, respectively. His thesis subject was on sensitive thin-film magnetic-field sensors. Since 1981, he has been a Lecturer and Research Scientist with the University of Twente, where he is currently a member of the Integrated Optical MicroSystems Group in the MESA+ Institute for Nanotechnology. In 1988, his research focus turned to integrated optical devices. A basis for his work in this field was laid during a half year of sabbatical leave at (then) AT&T Bell Labs, Holmdel, NJ, in 1989. His current research interests include photonic crystal structures and wavelength-selective devices for applications in sensing and communication.