Plasmonic sensors based on tilted Bragg gratings in multicore optical fibers

: Bare and gold-coated tilted fiber Bragg gratings (TFBGs) can nowadays be considered as a mature technology for volume and surface refractometric sensing, respectively. As for other technologies, a continuous effort is made towards the production of even more sensitive sensors, thereby enabling a high-resolution screening of the surroundings and the possible detection of rare events. To this aim, we study in this work the development of TFBG refractometers in 4-core fibers. In particular, we show that the refractometric sensitivity of the cut-off mode can reach 100 nm/RIU for a bare grating. Using another demodulation method, a tenfold sensitivity increase is obtained when tracking the extremum of the SPR (surface plasmon resonance) envelope for a gold-coated TFBG configuration. Immobilization of DNA probes was performed as a proof-of-concept to assess the high surface sensitivity of the device.


Introduction
Optical fiber sensors are widely studied and have been intensively used for biosensing applications over the last decade. They bring particular assets such as passiveness, low loss and high flexibility, especially for remote sensing applications. All these advantages have caught the attention of the scientific community for the development of novel types of biosensors [1][2][3][4][5]. One of the new demands for biosensing nowadays is point-of-care (POC) diagnosis, which is based on the detection of any target in real time [6][7][8]. In that context, optical fibers represent a step forward to perform portable and miniaturized devices capable of early detection.
In order to get the best performances from optical fiber-based biosensors, many structures are developed and studied such as U-shaped and side-polished fibers as well as Fabry-Perot interferometers [9][10][11][12]. Among such configurations, attractive assets arise from optical diffractionbased structures. These are performed by modifying the refractive index of the core of the optical fiber causing a periodical structure. By means of this technique, fiber Bragg gratings (FBGs) sensors have been developed in the last decades [3,13]. For uniform FBGs, light remains confined into the core, allowing their use for physical measurements of temperature, strain or pressure. For biosensing purposes, an interaction with the surrounding medium is needed and is usually obtained from an excitation of the cladding modes. Long period fiber gratings (LPFGs) and tilted fiber Bragg gratings (TFBGs) are mainly used for this purpose [14][15][16][17][18][19]. The latter, which is the technology used in this paper, consists in tilting by a few degrees the refractive index modulation pattern of the core with respect to the perpendicular direction to the optical fiber axis and thus, to excite the cladding modes of the fiber favoring the interaction with the external medium.
Another important milestone in the biosensing field was the implementation of plasmonic excitations on optical devices [20,21]. The specific combination of TFBGs with Surface Plasmon Resonance (SPR) improves the sensitivity of optical fiber-based biosensors [22,23]. To this aim, a metal film (usually gold) is coated around the fiber grating area. Such configurations have been successfully used for several applications, including glucose detection and cancer biomarkers sensing [24][25][26][27][28].
While the vast majority of developments on TFBGs refractometers and biosensors implies telecommunication-grade single-mode optical fiber (SMF), it has been shown that other fiber types such as multimode optical fiber (MMF) and polarization-maintaining fiber (PMF) bring relevant practical assets [29]. Indeed, for the same fiber diameter, an increase of the fiber core radius, which implies a reduction of the core-surrounding medium distance, causes an improvement in the sensitivity. In this vein, multicore fibers (MCF) appear very promising [30,31]. Such fibers consist in multiple cores embedded in a common cladding. Normally, the cores are localized far enough from each other to avoid any electromagnetic interaction in between. Thus, the cores are decoupled and the coupling from one core to other is therefore around -60 dBm/m. Such features make MCFs very interesting for telecommunication purposes, because in the same optical fiber, each core acts as a single channel, considerably increasing the throughput in the same physical medium [32]. MCFs have been used for sensing applications too [33][34]. In particular, bending and 3D-shape sensors have been demonstrated through the inscription of FBGs in each core [35,36], which is very useful for the development of optical endoscopes, for instance. More recently, MCFs have been used to inscribe tilted Bragg gratings. D. Barrera et al. demonstrated that it is possible to use TFBGs in MCFs for bending, temperature and refractive index sensing [37].
In this paper, we demonstrate both experimentally and through numerical simulations using a finite-element mode solver that gold-coated TFBGs inscribed in a 4-MCF provide an increase in refractometric sensitivity compared to their counterparts in SMF. Through numerical simulations using a finite-element mode solver, we show a good agreement with the experimental results. In addition, we highlight the impact of the core-surrounding medium distance parameter on the sensitivity. For this purpose, two 4-MCFs with different core distributions have been used as refractometers and compared. Finally, we show the possibility to monitor DNA-receptors immobilization on the fiber surface using this original sensing platform. At the expenses of a more complex optical connection, we believe that our findings are relevant in the course towards higher performances for biosensors, paving the way to the subsequent detection of rare events.

Manufacturing process
In this work, two different 4-MCFs were used with the aim to study the effect of the distance between the isolated cores and the surrounding medium. Cross sections of both fibers are shown in Fig. 1(a). The 4-MCF-1 presents cores closer to the center of the fiber than the 4-MCF-2. It is considered that the localization of the cores near to the fiber edge will strengthen the interaction of the cladding modes with the surrounding medium, therefore contributing to an increase in sensitivity. Both fibers were purchased from Fibercore [SM-4C1500(8.0/125)/001]. Each core features similarities with the one of an SMF, with a diameter of 8 µm and a numerical aperture in the range of 0.14-0.17. Cores are also embedded in a typical silica cladding.
The measurement setup works in reflection. It consists of a LUNA optical vector analyzer (OVA CTe) as optical interrogator connected to a dedicated fan in/out (Fibercore). Such a coupler relies on 4 independent SMFs in one leg and a single 4-MCF in the other leg. The multicore fiber was connected mechanically to the 4-MCF sensor through a rotating system in order to easily change the channel. Light is reflected from the cleaved end face of the fiber and guided backwards to the interrogator. The interrogation setup is sketched in Fig. 1 TFBG inscription in the aforementioned MCFs was conducted using the Noria inscription setup (NorthLab Photonics), which implies to operate with a static exposure. It appears clear that the most suited method to produce gratings in each core would be to rotate the fiber during the inscription process but this was not possible to implement in the Noria system. Therefore, as discussed hereafter, we have evaluated alternative solutions to produce a maximum of well-defined amplitude spectra, to enable high-quality refractometric measurements. Due to the spatial distribution of the cores, the inscription presents a strong dependence with respect to the orientation of the fiber, as reported in [38] and this was evaluated in our experiments.
First, we studied the difference of the TFBG inscription in the two multicore fibers. Prior to the inscription of the gratings, the cleaved end of the fiber was connected to a red laser pointer and a Fiberscope to ensure the required orientation. First of all, the inscription was performed in the same conditions for the two fibers with an orientation of the 4 cores forming a square with respect to the laser incidence, as depicted in Fig. 2. In these conditions, our observations show that 4-MCF_2 allows to produce TFBG amplitude spectra of better quality than the 4-MCF_1. The peak-to-peak amplitude of the cladding mode resonances can reach close to 30 dB while it is more limited and noisier for the 4-MCF_1. Obviously, gratings are written in the two cores facing the UV incidence. The two other cores remain almost unaltered since they are in the shadow of the upper cores.
Then, keeping MCF_2 as a substrate to produce high quality gratings, another intuitive inscription orientation was performed to study the dependence and determine the optimum one: a diamond form of the cores with respect to the laser incidence was selected, as depicted in Fig. 3. As expected, the diamond orientation enables a stronger inscription in the core facing the laser incidence (more than 30 dB of peak-to-peak amplitude for the highest cladding mode resonances). The latter will therefore be selected for further experiments. The bottom core remains almost unaltered, again due to a shadowing effect. The two other cores show an interesting spectrum. The black curve is of less intensity and noisier, certainly because the corresponding core receives less power density due to a non-optimal alignment of the laser beam with respect to the two lateral cores. TFBGs were then used as refractometers in order to obtain the sensitivity of both fibers to changes in the external medium.

Bare TFBGs in 4-MCF
First, we studied the sensing performance of a bare TFBG by modifying the surrounding refractive index (SRI). The results for both MCFs are shown in Fig. 4. The read-out technique is based on the tracking of the wavelength shift of the so-called cut-off mode, i.e. the cladding mode resonance whose effective refractive index is the closest to the one of the surrounding medium. In practice, all spectra are aligned on the Bragg resonance, both in wavelength and in amplitude, so as to compensate for unwanted temperature variations and optical power fluctuations. The wavelength shift of the most sensitive mode is then tracked as a function of the measurand.  Experiments were performed as follows: a salt solution with a refractive index around 1.34 (stock solution) was first prepared. Then, 6 different solutions were prepared by dilution with a step of 2 × 10 −4 RIU. The refractive index value of each solution was computed with a calibrated refractometer (Reichert RI Chek). Optical fibers were then immersed in each solution, to obtain an increase of the SRI. After each measurement, the fiber was rinsed out with pure water to remove any rest of salt. Figure 5 depicts the calibration curves for both fibers. A linear regression of the raw data shows that the refractometric sensitivity of MCF_1 is 93 nm/RIU while the one of MCF_2 is 101 nm/RIU. This sensitivity increase is consistent with the distribution of the cores that are closed to the surrounding medium interface for MCF_2. In addition, both fibers show a sensitivity ∼3 times higher compared with the one of SMF [29]. Our observations prove that the distance between the core and the outside medium plays a crucial role in the intrinsic sensitivity.

Gold-coated TFBGs in 4-MCF
We then focused on gold-coated TFBGs. To that end, we covered the grating area with a nano film of gold through a sputtering method (Leica EM SCD500), using the same recipe as the one we use for standard TFBGs in SMF. The fabrication of the probe was achieved as follows: due to the features of the sputtering equipment, a layer of gold was first deposited on one side of the optical fiber, and subsequently, the fiber was turned around 180 deg to cover the other side. The two depositions were made each time with a gold thickness of 23 nm measured by a Quartz microbalance placed in the sputtering chamber. Considering the calibration of our system, this results in a quite homogeneous coating all around the fiber cross-section with a mean thickness of ∼40 nm. The obtained spectrum for the 4-MCF_2 is shown in Fig. 6, showing the typical SPR signature when the P-polarization state is selected. As reported in previous works [39], the S-polarization state cannot tunnel inside the gold layer.
We studied the refractometric response of the sensor, following the same experimental procedure as for bare gratings. The recorded spectra in each solution for the fiber core facing the laser inscription are shown in Fig. 7 for both fibers while the calibration curves are depicted in Fig. 8. The latter was obtained by tracking the most sensitive cladding mode resonance located on the right-hand shoulder of the SPR envelope, as done in previous works. The results of Fig. 8 show once again that the sensitivity of 4-MCF_2 is higher than the one of 4-MCF_1. Also, the sensitivity of both fibers is higher than the one previously computed for bare configurations. This being said, the sensitivity increase is limited to ∼15% when the read-out technique is based on the wavelength tracking of a single resonance among the spectral comb. The presence of the gold-coating allows to use another demodulation technique. It has been recently shown that an interrogation based on the envelope of the SPR signature yields a higher sensitivity, as it better  reflects the behavior of the actual SPR mode [39]. This demodulation process was also applied to the gold-coated TFBG in 4-MCF_2 and the results are shown in Fig. 9.   Figure 9(a) depicts a zoom around the SPR signature area for a gold-coated TFBG in 4-MCF_2 immersed in liquids with different RI values. It shows the computation of the upper and lower envelopes of the cladding mode resonances spectrum. For demodulation purposes, the focus is then made on the minimum of the upper envelope and on the maximum of the lower envelope, whose wavelength positions are tracked as a function of the SRI value. Results are depicted in Figs. 9(b) and 9(c). The sensitivity achieved by this method is around 1200 nm/RIU, which one order of magnitude over the one is obtained by the mode tacking method. According to our knowledge, this is the highest sensitivity reported so far for a TFBG-based refractometer. It results from the good spectrum quality and the proximity of the core to the surrounding medium. In order to confirm the tendency through statistics, we have repeated the experiments three times and computed the standard deviation, depicted as error bars in Figs. 9(b) and 9(c).
In the following section, we will compare these experimental results with numerical simulations performed using a commercial software.

Simulation
Simulations were performed with the FimmWave (PhotonDesign) finite-difference tool that provides a complex solver of modes in structures such as the ones shown in this paper. Our aim was to confirm that the experimental data were in agreement with the theory, and for that purpose, the two MCFs were simulated. We divided the simulations in two parts. First, we studied the cladding modes that appear when the fiber is bare. Then we added to the simulation a layer of gold in order to trigger the SPR effect.
We used the features of the optical fibers that were shown in section 2, the geometry and the refractive index of the core and cladding for all cases. The solver provides all the cladding modes that our fibers are able to support, but we have to compute only the modes that are excited due to the presence of the TFBG [37], which are the modes that are totally radially-polarized and azimuthally-polarized. Such modes correspond to the TM 0n -EH 1n and TE 0n -HE 1n families, respectively [26]. It is important to point out that the solver gives the complex refractive index of each mode, which means that the real part corresponds to the refractive index of the mode and the imaginary part is related to the losses of such mode. To correlate the refractive index of the cladding excited mode with the corresponding wavelength, we have to use the following equation: λ clad,i = (︁ n eff,core + n eff,clad, i )︁ Λ/cos θ where n eff,core is the core effective refractive index, λ clad,i represents the resonance corresponding to n eff,clad,i , Λ is the grating period and θ is the tilt of the angle. Then, we performed the same simulation with a gold layer around the MCFs and a thickness similar to the estimated value of the gold-coated TFBGs used in the previous section. Gold is characterized by a complex refractive index of 0.56-11.4i. Figure 10 shows that the refractometric sensitivity of 4-MCF_2 is higher than the one of 4-MCF_1 and that the sensitivity further increases with the addition of a gold layer. The simulation results differ from the experimental data, certainly because of the uncertainty on the actual gold layer thickness. Anyway, the simulations confirm the previously observed experimental trends.

Biosensing experiments and prospects
Gold-coated TFBGs in 4-MCF_2 were successively immersed into a phosphate buffer saline buffer (PBS) at pH 7.2 and into a solution of DNA molecules, namely thiolated aptamers. The fibers were rinsed again in PBS after the deposition to assess the anchoring of a molecular layer. Aptamers are small synthetic molecules that are selected to bind to specific targets. In this work, we have performed a biosensing experiment using thiolated single-strand DNA aptamers, specifically targeting thrombin proteins. The aim of this sensing experiment was to ensure a strong binding and detection of DNA receptors that can be of interest for further biosensors development. Compared to the small blue-shift drift occurring in PBS, a pronounced shift of the sensitive spectral resonances reaching more than 30 pm of variation is observed during the immobilization of thrombin aptamers (Fig. 11). This experiment paves the way through the design of multiple core fiber biosensors for the simultaneous detection of different analytes using the same optical fiber, in real time. The exploitation of both cores in real time for sensing is a great challenge but will indeed lead to interesting prospects in that field of research.

Conclusions
In this paper, we have studied the development of TFBG refractometers in two 4-core fibers from the same manufacturer, with two slightly different geometries. Our objective was to ensure the production of a high-quality grating spectrum in at least one of the cores and to this aim, we have confirmed that the fiber orientation with one core facing the inscription laser is the best solution in practice. We have then experimentally demonstrated an increase of the sensitivity to the surrounding refractive index change when the core-outside medium distance decreases. Following the wavelength shift of the cut-off resonance, a sensitivity close to 100 nm/RIU has been reported for bare configurations, which is a threefold improvement compared to TFBGs in standard single-mode optical fibers. Using a thin gold coating around the fiber, we have demonstrated the successful excitation of a surface plasmon resonance and have reported a sensitivity close to 1000 nm/RIU when tracking the extremum of the envelope of the cladding mode resonances around the SPR signature. This value is the highest ever reported for a TFBG-based refractometric sensor. Finally, a bioassay was conducted to assess the sensitivity of the technique to surface-located biological events, through the monitoring of DNA immobilization, opening the way for all-in multiplexed biosensors. Disclosures. The authors declare that there are no conflicts of interest related to this article. Data availability. Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.