A hetero-core structured optical fiber SPR sensor is shown schematically in Figure 1. The hetero-core optical fiber consisted of a transmission line multimode (GI) fiber and an inserted segment of a single-mode (SI) fiber, which works as a sensor region. Details of hetero-core SPR sensors were described in our previous works [16,17]. With this structure, since the core diameter of the sensing part was much smaller than that of transmission line fiber, the propagated light wave in the transmission line would largely leak into the cladding layer of the sensor region. When the leaked light is reflected at the boundary of the cladding layer and the surrounding environment under the condition of internal total reflection, such a light wave induces an optical evanescent wave at the surface of the cladding layer, hence SPR excitation by coating the cladding surface with a thin metal film, as shown in Figure 1, similar to the case of the conventional Kretschmann configuration sensor.

A light wave traveling through the sensor region could partially re-couple into the core of latter transmission line. Additionally, since light in the transmission line propagates in multimode forming numerous modes, many SPR spectra are induced corresponding to various incident angles. As a result, an output spectrum could be overlapped to be a broader spectrum due to many excited resonance wavelengths. This broader spectrum makes possible flexible usage in which the intensity mode of operation is easily made with a fixed wavelength of probing. No coating on the hetero-core produces only a transparent spectrum curve without any SPR resonance spectrum.

A sensor for the experiment was fabricated with a GI MMF with core and cladding diameters of 50 and 125 μm, respectively, and an inserted sensitive region of a 15-mm-long SI SMF with a core diameter of 3 μm and the same cladding diameter of 125 μm as the GI fiber. The fabricated hetero-core region was then cylindrically Ta₂O₅-coated with thicknesses of 0–60 nm over a 25-nm thick Au first layer by using an RF sputtering machine (CFS-4DS-231, Shibaura Mechatronics Corp.), in which a specially designed rotating mechanism was devised to symmetrically realize uniform deposition on the cladding surface. According to our previous work [17] for the Au-SPR sensor, the Au thickness around 20 nm offered us the larger SPR dip, but the smaller SPR shift due to the change of refractive index. The Au/Ta₂O₅ coating combination was properly selected with 25-nm Au to give a sufficient SPR dip according to a series of experiments. The experimental set-up of Ta₂O₅ thickness characteristics consisted of a halogen lamp, whose wavelength was in the range of 400–1,800 nm, and an optical spectrum analyzer (AQ6315A, Ando Electric Co., Ltd.), as shown in Figure 1. The output from the halogen light source was guided into a 100-m upstream transmission fiber and finally introduced to the spectrum analyzer after traveling in a downstream 100-m fiber through the sensing part. Considering the possible usage of this sensor in the outdoors full-scale experiment where several tens of fiber extension could be required, 100-m fibers are connected before and after the sensor element. The characteristics obtained in this work are reasonably reserved regardless the length of in- and out-transmission fiber as long as multi-mode is fully excited in a stable state, except that the offset level of transmission loss could be changed with the length. In order to avoid unfavorable fluctuation due to mode disturbance of MMF, the sensor region was firmly mounted in a straight line configuration with a stretching support. In the series of experiments, glycerin solutions were prepared as test liquids with concentrations of 0–50% by weight, corresponding to refractive indices ranged 1.333–1.398 RIU.

Figure 2 shows SPR spectra A, B and C experimentally obtained in the case of water for the sensors having Ta₂O₅ thicknesses of 0, 25 and 60 nm, respectively. These results are normalized with the spectrum for air. As the result of increasing the Ta₂O₅ thicknesses extending to 60 nm, the experiment shows that the resonance wavelength is largely sifted to the near-infrared wavelength region over 1,000 nm. Although each spectrum is somewhat uneven at the wavelength around 1,400–1,600 nm, this could be caused by an interference effect between longer wavelength modes when the light wave re-couples into the core downstream, since the same results can be appeared in the case of no coating hetero-core without water. Figure 3 shows theoretical SPR spectra based on multi-layered model [19,20] for three Ta₂O₅ thicknesses, assuming that there would be a multi-mode distribution [17,18] with the corresponding incident angles in the hetero-core structure. The refractive indices of each layer at 1,310 nm are given as Au: 0.413 RIU, Ta₂O₅: 1.888 RIU using the wavelength dependence previously reported [20,21].

Comparing the experimental result (Figure 2) with the numerical calculation (Figure 3), it is indicated that the wavelength-shifts are found to be similar for the increases of Ta₂O₅ thickness. In this work, the coating of 25/60 (Au/Ta₂O₅) shows sufficient resonant absorption at the wavelength around 1,310 nm, although slight differences are seen between the experimental and the theoretical results. The Au/Ta₂O₅ coating combination was properly selected to give a sufficient SPR dip according to a series of experiments. The thicker the Ta₂O₅ coating is, the broader and more shallow the SPR spectrum becomes. This work used only Ta₂O₅ coating as one of high refractive index materials since the purpose of this scheme is firstly to show the capability as a liquid detector rather than to attempt possible other materials previously reported [22]. Figure 4 shows the SPR spectrum change with various refractive indices of liquids for the 25/60 coating, under the condition that the resonant wavelength appeared around 1,300 nm because of the use of 60 nm-Ta₂O₅ coating.

These results were normalized with the spectrum for air. The resonant absorption in the wavelength range of 1,000–1,400 nm decreases with the increase of refractive index, showing that the same interference effect as in Figure 2 appears in the 1,400–1,600 nm range. As can be seen from Figure 4, probably because of such interference effect, the intensity change in the wavelength region longer than 1,300 nm is heavily disturbed in a complicated manner, hence with an ambiguous resonant shift. The sensitivity can be derived as 1.3 dB per 0.065 RIU at 1,310 nm. The sensitivity obtained is comparable to that in the previous works [10,11,17,22]. The data accordingly shows that the wavelength-shifted hetero-core scheme can be used more favorably as liquid detection, rather than as a refractive index (RI) measurement. The obtained experimental results indicate that an attractive liquid detection sensor could be realized for liquids with refractive indexes of 1.333–1.398 RIU such as water or ethanol.

The absorption loss of the hetero-core SPR sensor can be reproducibly adjusted by changing the insertion length [23]. Since SPR interaction occurs effectively with longer insertion lengths, the sensitivity of the sensor is increased. However, the insertion loss is also increased. The insertion length thus shows a trade-off between sensitivity and insertion loss. Taking advantage of the flexible change of SPR loss, authors have developed a multi-point liquid detection system using three different sensors tandem connected in a single line, in which the three sensors can be identified. In order to separately interrogate the state of each sensor, specified losses were given to each sensor by tuning the insertion length. Sensor detection succeeded only with a combination of a light source of LED and a photo diode. Since this multi-point detection system uses no OTDR, hence no data averaging which is a time consuming process, the detection can be made on a real-time base. The experimental setup for multi-point liquid detection was arranged as shown in Figure 5, and consisted of three hetero-core SPR sensors connected in tandem, an LED (−19.5 dBm) whose wavelength is 1,310 nm as a light source and an optical power meter (PD: OP710, Opto Test Co.).

The output from the LED was guided into a transmission fiber and finally introduced to the PD through the three sensors. The insertion lengths (IL) of the three sensors, denoted as sensor 1, 2 and 3, are 2, 5, and 15 mm, respectively. In the experiments, seven detection patterns were tested as shown in Figure 6. Depending on the sensor combination, extinguished losses are indicated. The losses for each combination were measured as the optical intensity for 30-second, then all sensors were exposed to air for 30-second before the next test. In this way, the losses for all patterns were continuously changed in the order of loss increase with the sampling frequency of 1.67 Hz. Figure 6 shows the real-time response of optical loss change in all patterns. As can be seen, sensor combination is found to be shown as the different levels of loss. In order to evaluate the repeatability of the system, the average of optical loss of each pattern over five trials is shown in Figure 7, which shows that the loss detection is reproducible enough to find the sensor combination. As explained above, the feasibility of a multi-point liquid detection system based on LED-PD has been demonstrated by employing the hetero-core optical fiber SPR sensor coated with Au/Ta₂O₅ thin metal. In addition, improvement of this multi-point detection system would be expected by giving the loss as the n^th power-of-two (a × 2⁽ⁿ⁻¹⁾) interval to an n^th sensors.

A water-level gauge can be also devised by employing a number of vertically arranged sensors which are connected in tandem, as shown in Figure 8. An identical loss is given to all sensors so that the sum of losses indicates the water level increase. Resolution of the water-level can be given as 0.2 m for this case, and be selected by changing the number of sensors per unit length which are placed in a stepwise form. The measurement system consists of six hetero-core SPR sensors connected in tandem and the same LED-PD combination as seen in Figure 5.

Figure 9 shows the average of optical loss as a function of the number of sensing points immersed in water, in which these losses were obtained by five consecutive experiments.

Optical losses of each water-level were measured after each sensing head was completely immersed. As can be seen, it is found that the optical loss monotonically increases with the water-level raise. The minimum loss step per sensing head immersed in water is 0.22 dB. The error bars shown in Figure 9 correspond to a reproducibility less than 0.06 m.