The tilted fiber Bragg grating is a new kind of sensor that possesses all the advantages of well-established Bragg grating technology in addition to being able to excite cladding modes resonantly. This device opens up a multitude of opportunities for single-point sensing in hard-to-reach spaces.
Fiber Bragg gratings, or FBGs, are narrow pass-band and rejection-band filters that have been fabricated in ordinary optical fibers. These gratings, which have been widely deployed since the 1990s, are now used as in-fiber cavity mirrors in fiber lasers, as dispersion compensators in long-haul optical networks and as stabilizers for the pump laser diodes used in optical amplifiers. However, because the Bragg wavelength is sensitive to strain and temperature, the application of FBGs that has generated the most intense research activity has been in the area of sensing.
Now, a new twist on the configuration of FBGs has enabled a broader range of sensing capabilities than was previously possible, with applications in biochemistry, plasmonics and nonlinear optics. By angling the grating planes by a few degrees relative to the plane perpendicular to the fiber axis, researchers have now developed so-called tilted fiber Bragg gratings that can sense the magnitude and direction of a bend in a structure, detect chemicals or biochemical reactions near the fiber surface and control the polarization state of the light near the cladding boundary.
Over the past five years, my research group has been developing multimodal fiber grating sensors based on tilted FBGs in various shapes and forms, always keeping the device itself simple to fabricate and compatible with low-cost, standard FBG technology.
Limitations of fiber grating sensors
In conventional single-mode fiber FBGs, the only strong coupling permitted occurs between the forward and backward propagating core modes, for which the light is confined near the fiber axis and isolated from the surroundings by a relatively thick cladding. As a result, FBGs can only sense temperature and axial strain (as they both modify the effective indices and the period, to varying degrees), while they are insensitive to perturbations that do not reach the core. Although FBGs have a multitude of potential applications, especially for in situ and remote sensing, they are limited in that they cannot be used to detect pure bending and changing in the medium adjacent to the cladding boundary.
Hence, researchers have made many efforts to get core light to interact with the fiber surroundings. This is usually accomplished by etching or polishing the cladding away, tapering or using special microstructured fibers with holes or cladding slots, etc. In all the resulting devices, the goal is to make the effective index of the core mode depend on more than just the diameter of the core and the refractive indices of the core and cladding.
Using modes guided by the cladding of the fiber
Of course, many more sensing modalities become accessible if the grating can couple light to cladding modes. These modes are guided by the interface between the cladding and the outside medium, a structure that can support several hundred guided modes in conventional single-mode fiber. As it turns out, forward-coupling gratings, the so-called “long-period gratings” (LPGs), can be used to couple light into select cladding modes resonantly and efficiently. The change of sign in the phase-matching equation causes LPGs to benefit from two sensitivity enhancements: the much larger period (which multiplies any effective index change) and the fact that the difference in effective indices can be optimized to shift rapidly in response to certain perturbations by waveguide design.
However, while LPGs have been around for as long as FBGs, they have not found widespread use because they suffer from strong cross-sensitivities that prevent most applications, except for a few niche ones where high packaging costs do not matter.
Researchers have developed a new kind of fiber grating sensor that possesses all the advantages of the well-established FBG technology in addition to being able to excite cladding modes resonantly: the tilted FBG. The TFBG is essentially an FBG whose grating planes are angled by a few degrees relative to the plane perpendicular to the fiber axis. This grating tilt has the effect of locally breaking the cylindrical symmetry of the fiber in a way that allows strong coupling between the core guided light and a large number of cladding modes.
Tilted fiber Bragg grating transmission spectrum with labels indicating the effective indices of selected modes.
Apart from tilt, everything we know about FBGs, including the many techniques used to fabricate them, can transfer directly to TFBGs. Tilted gratings have also been studied for almost as long as FBGs, with larger tilt angles for extracting light from the fiber altogether. For example, TFBGs are used in some niche telecommunications applications as all-fiber polarizers, gain flattening filters for optical amplifiers and channel monitors. It is only recently, however, that they have been used in the sensing arena.
The figure on the right shows a typical transmission spectrum of a TFBG. Each dip in the spectrum corresponds to light that has been removed from the single-mode core and coupled to one or several backward propagating cladding modes. In most instances, the back propagating modes do not return all the way to the fiber input because of the large attenuation of jacketed fibers, bends and so on (similar to the forward-coupled cladding modes of LPGs).
The figure on the right shows how core guided light is extracted from the core by the grating and excites different cladding modes, selected according to wavelength (color) and polarization. Since the cladding modes each have a unique mode field shape and effective index (angle of incidence at the cladding boundary, as shown), they react differently to perturbations inside and outside of the fiber.
A transmission spectrum such as the one shown in the figure thus contains an incredible wealth of information that begs to be “read,” provided that the transducing mechanisms are well understood. In 2002, two independent research groups (Ferdinand et al. in France and Lee et al. in Korea) first recognized that the strong cladding mode resonances in the transmission spectrum of TFBGs could be used to sense refractive indices and bending, respectively.
Thermometers that are also insensitive to temperature
Standard FBGs can sense two things very well—temperature and strain—but only those two things. For anything else, a transducing mechanism must be provided to transform the required measurand into a change in temperature or strain. This added mechanism most often involves packaging the FBG into a structure that itself must respond to the perturbation by straining or heating the fiber. This is not easy to do for biochemical sensing applications. Furthermore, the FBG responds the same way to temperature and strain (a wavelength shift of a narrow resonance) so it is normally impossible to distinguish between the two perturbations with a single device.
Bending sensors. (Top) Differential TFBG transmission spectrum due to 8-cm diameter bend. (Bottom) Vector inclinometer structure and response (φ is the orientation of the bend relative to the tilt plane).
The TFBG circumvents these two problems, thanks to the simultaneous presence of a large number of resonances from a single device. Since all resonances have the same temperature dependence (roughly 10 pm/°C or one part in 105), the temperature sensitivity of any other sensing modality can be factored out by using relative wavelengths instead of absolute ones. On the other hand, the strain dependence of high order cladding modes is quite different from that of the core mode. Therefore, a temperature desensitized axial strain sensor is obtained from a single TFBG. Of course, it is also possible to use the temperature sensitivity of any one of these modes as an in situ thermometer and use other resonances to sense more parameters.
Bending and vibration
Each cladding mode “samples” the fiber cladding cross-section and the medium adjacent to it differently: For instance, the large nonuniform strains occurring in bent fibers strongly influence a very distinct subset of resonances but not at all the Bragg resonance, for the same reason that FBGs are insensitive to bending. The peculiar response of TFBGs to bending has been used to develop a very sensitive accelerometer as well as a vector inclinometer (i.e., a single grating device that can detect both the magnitude and direction of a bend in a structure by comparing the power returned in two distinct passbands).
In both cases, some of the backward-propagating cladding modes are recoupled into the fiber core by an upstream, broadband, core-to-cladding-mode coupler, and detection of the vibration or bend does not require precise wavelength measurements with a spectrometer but one or two bandpass-filtered power measurements.
The other important fiber-sensing application is refractometry. That’s because TFBGs allow for the detection of chemicals or biochemical reactions near the fiber surface, provided the protective jacket of the fiber has been removed. Refractive index changes in the medium immediately adjacent to the fiber also produce very distinctive spectral signatures in TFBGs. As in all kinds of waveguide sensors, it is the penetration of the evanescent wave of the guided modes into the external medium that is critical in determining the sensitivity. The maximum sensitivity of a mode usually occurs when the light wavelength or the index difference that is responsible for waveguiding are such that the mode in question approaches its cutoff point, just before the guided light escapes radiatively.
Why FBGs work as sensors
As shown in the following equation, a grating perturbation of period Λ (with or without a tilt angle φ) couples an incident core mode to another mode of the structure (labeled [m,n]) at a very specific wavelength (λmn) that is uniquely determined by the effective indices of the modes in question. This is called "phase matching." The plus and minus signs apply to backward and forward coupling, respectively, with the consequence that the phase matching grating period is shorter than the light wavelength for backward coupling but longer by one or two orders of magnitude for forward coupling:
λmn = (Neffm,n ± Neffcore )Λ/cos (φ) .
As a result, the resonance wavelengths of fiber gratings depend very precisely on any perturbation imposed on the fiber that changes one of the parameters of the equation. This property has been used to make very sophisticated and practical sensor systems for a number of important applications in structural engineering, resource extraction and, increasingly, in the biomedical field.
This is where the TFBG in standard silica glass fibers shines because it provides a very fine comb of mode resonances with effective indices ranging from well below 1.3 to near 1.44. Thus, for all media with refractive indices in this range, the spectrum has at least some resonances near cutoff and hence at their maximum sensitivity, as well as other “insensitive” resonances that can be used as wavelength and power references. Furthermore, the refractometric sensitivity of TFBGs can be improved by more than one order of magnitude by using plasmonic effects (described below).
Playing with polarization
One of the questions I am frequently asked is how the input light polarization affects the performance of TFBGs. For FBGs, the polarization dependence of the response is usually negligible, and this remains true for the core mode resonance of TFBGs because of the relatively small tilt angles we use and the fact that the core mode is weakly guiding and thereby supports linearly polarized (LP) modes. On the other hand, cladding modes are guided by a larger step discontinuity in refractive index, such that their vectorial nature cannot be ignored. In fiber-optic terms, the “LP mode” approximation no longer applies. Experimentally, we observe that high-order cladding modes—whose resonances occur at short wavelengths that are far from the Bragg resonance—do appear to be very strongly polarization-dependent. This is because higher-order cladding mode resonances actually come in pairs made up of nearly degenerate orthogonally polarized modes (the EH and HE vector modes of the cladding) and the TFBG automatically selects one or the other polarization when core mode light is linearly polarized either in the plane of the tilt (P-polarization) or perpendicular to it (S-polarization), respectively.
Polarization dependence of high-order mode resonances. Polarization dependence of high-order mode resonances in a TFBG transmission spectrum measured in air. S- and P-polarization have their usual meaning relative to the tilted grating planes.
Simulated x- and y-components of the electric fields. S resonances have predominantly tangential electric fields around the fiber cladding boundary. In other words, the x component of the field is maximum at the top and bottom boundaries, and the y component is maximum at the left-right boundaries. P resonances have electric fields that are predominantly radial at the boundary. Therefore, P resonances can transfer energy to a plasmon wave on the metal surface (bright fringe around the fiber) but S resonances cannot.
Apart from an obvious application as a narrowband, high-extinction-ratio polarizer, the polarization dependency of high-order modes in TFBGs has an important consequence for sensing. Polarization control allows the user to select only HE or EH modes across large portions of the spectrum. The electric field of high-order EH modes is oriented predominantly radially at the cladding boundary, while that of HE modes is mostly tangential. Therefore, any sensing modality that depends strongly on the polarization state of the light near the cladding boundary can be controlled very effectively with TFBGs.
Plasmonics on fibers
One of the most important consequences of this EH-HE discrimination is that TFBGs can be used to efficiently excite quasi-cylindrical surface plasmons on metal-coated optical fibers. Surface plasmons are highly confined electromagnetic waves that are propagating along a metal-dielectric interface with a well-defined propagation constant (completely determined by the permittivities of the metal and dielectric) and an electric field oriented perpendicular to the interface. Spectral or angular interrogation of such interfaces with light beams—using the so-called surface plasmon resonance, or SPR, instruments—is now done to detect minute changes in the properties of the metal or of the dielectric, and hence to sense chemical or biochemical reactions.
Researchers have pursued various fiber-based SPR techniques to miniaturize the sensing probe and to facilitate the input-output coupling of the light. The more successful fiber SPR techniques rely on measuring the transmission spectrum of etched cladding or tapered multimode fibers where the SPR shows up as a loss band that occurs only for modes whose propagation constant is equal to that of the plasmon wave for the external medium of the fiber. This requires accurate and stable control of the distribution of the light among all of the possible modes of the fiber, and polarization must either be scrambled or linearly polarized at the input of the fiber, thereby losing a great deal of the signal-to-noise ratio in either case because plasmons can only be excited by radially polarized light along the fiber boundary.
Both constraints also mean that the fiber must be kept relatively straight and cannot be moved during experiments without causing changes in the shape or position of the SPR signature.
On the other hand, with a TFBG in single-mode fiber, the light propagates as a single mode of the core everywhere but at the sensor itself. In other words, the fiber can be moved, and the light source and interrogation system can even be located many kilometers away from the sensor. In addition, cladding modes with the correct (radial) polarization and propagation constant to excite the plasmon can be automatically selected by the TFBG.
Along with the other nice features of the TFBG device itself, and especially the temperature insensitivity, this fiber-SPR device is extremely robust, inherently low loss, easy to interrogate with widely available standard telecommunication test and measurement instruments, and yet as simple to fabricate as conventional FBGs. (The only added step is the metal coating that can be performed in batches using either gas phase or liquid phase deposition processes.)
Biochemical detection with tilted grating SPR sensors
Recently, we have demonstrated that TFBG-SPR devices have wavelength shift sensitivities to refractive index change that is similar to all other SPR implementations (bulk or fiber)—i.e., around 500 nm/refractive index unit—and that they provide absolute refractive index measurements in the range between 1.3 and 1.4. This is not surprising, since all such devices are based on the same underlying physics. However, we have also discovered that differential polarization measurements of TFBG-SPRs reveal narrowband resonant features that have full widths under 100 pm (at least two orders of magnitude smaller that “regular” SPR resonances) and wavelength shift sensitivities above 300 nm/refractive index units.
Because it is a lot easier to measure wavelength shifts of narrower resonances, we have been able to demonstrate reliable and repeatable resolution of refractive index variations of 1x10–5, and coating thickness changes of 1 nm for both metals and dielectrics. This increased sensitivity has allowed us to demonstrate the biomolecular recognition of micromolar concentrations of proteins in solution by attaching synthetic DNA sequences on a gold-coated TFBG-SPR sensor.
Where is this technology going?
TFBG-SPR refractometer. The difference between the S- and P-polarized transmission spectra of metal coated fibers (top) yields a PDL spectrum with a strong notch (near 1,550 nm, bottom) whose position provides an absolute value of the refractive index of the external medium with an accuracy of 10–3. Shifts of the narrow feature at the bottom of the notch further provide a relative refractive index resolution of 10–5.
There are other ways to generate spectrally narrow resonances that have high sensitivity to various perturbations, and optical resonators such as rings and Fabry-Pérot cavities can be designed and fabricated with very high quality factors, including narrow resonances and wide spectral ranges. They can also be multiplexed in large numbers on photonic waveguide chips, and they can have higher sensitivities than the TFBG. However, unlike TFBGs, they only provide relative measurements within their free spectral range—the measurement of the absolute value of the refractive index of an unknown substance is difficult, for instance.
In addition, they can only measure one parameter, and they suffer from temperature cross-sensitivities that need careful design and packaging to avoid. Clearly, different optical sensing technologies have advantages and disadvantages that must be balanced against the requirements of a particular sensing problem.
In the case of the TFBG, the device provides a multitude of sensing opportunities for single-point sensing in hard-to-reach spaces, with very controllable cross-sensitivities, absolute and relative measurements of various parameters and an extreme but controllable sensitivity to metal particles and coatings. These characteristics should open up new avenues of research in plasmonics and nonlinear optics. In fact, the controlled excitation of cladding modes has been recently used to demonstrate ultrafast switching of picosecond pulses and four-wave mixing in a TFBG coated with carbon nanotubes, without requiring the fiber to be etched or tapered in order to get core guided light to interact with a nonlinear material external to the fiber.
I would like to acknowledge all my graduate students and collaborators who actually did the work described in this article. This work is funded by NSERC and LxData Inc.
Jacques Albert holds the Canada Research Chair in Advanced Photonic Components in the department of electronics at Carleton University in Ontario, Canada.
References and Resources
>> C. Chan et al. Appl. Opt. 46, 1142 (2007).
>> R. Kashyap. Fiber Bragg Gratings, 2nd ed., Academic Press 2009.
>> Y. Shevchenko et al. Opt. Lett. 35, 637 (2010).
>> C. Caucheteur et al. Opt. Express 19, 1656 (2011).
>> G. E. Villanueva et al. Opt. Lett. 36, 2104 (2011).
>> Y. Shevchenko et al. Anal. Chem. 83 (2011); doi: 10.1021/ac201641n.
>> An exhaustive list of references on our work can be found at: http://photonics.carleton.ca/publist.docx.