Distributed optical fiber sensors (DOFS) have demonstrated significant values in the structural health inspection, biomedicine, and aerospace industries by providing continuous measurements of strain and temperature monitoring at more than a few thousand locations at low cost and small size along the length of an optical fiber [1,2]. Commercially available communication fibers have become the mainstream fiber used in DOFS because of the advantages of low cost and standardization, but the backscattering light of standard fibers is designed to be as low as possible, which seems to be contrary to the needs of Rayleigh scattering based distributed sensing. Therefore, recent research has focused on new fibers with higher backscattering to improve the resolution and accuracy of the sensing system [3,4].

Techniques to enhance Rayleigh backscattering can be divided into two categories: special fibers with composition changes and modifications to standard fibers. Doping (germanium) in fiber core is a straightforward way to correlate with Rayleigh scattering, and the doped components increase the inhomogeneity of the core material, which in turn enhances Rayleigh scattering [5]. Doping nanoparticle engineering (${\mathrm{ZrO}}_2$ nanoparticles [6], gold nanoparticles [7], MgO nanoparticles [8], Ca-based nanoparticles [9], and Sr-based nanoparticles [10]) in the fiber core has also been used to enhance backscattering. However, the introduction of dopant components in optical fibers requires an extremely high degree of process control to trade off loss and scattering enhancement, which limits the sensing distance. On the other hand, the post-processing of standard optical fibers appears to be a more flexible and promising technique. UV laser radiation of photosensitive fibers increases defects such as color centers in the core [11], or forms ultra-weak fiber Bragg grating (FBG) arrays [12] and random fiber gratings [13], which already have applications in biomedicine [11].

Femtosecond lasers can preserve the fiber coating to process the fiber and do not require photosensitivity enhancement. In particular, the permanent modifications formed by femtosecond lasers have higher temperature stability ($>800°\mathrm{C}$), which makes them especially suitable for measurements in harsh environments [4]. A variety of structures prepared by femtosecond lasers in fiber cores have been shown to enhance the performance of distributed sensing [14–18]. A parameter of interest is to obtain as low a loss as possible while significantly improving the optical signal-to-noise ratio [19]. By introducing continuous nano gratings in the fiber core through reel-to-reel fabrication, the Rayleigh backscattering signal can be increased by more than 35 dB, and the transmission loss is 0.01 dB/cm [15]. Highly localized refractive-index change points can be written in the fiber core using the point-by-point inscription technique, and the incident light is either scattered or reflected at these points, which enhances the intensity of the backward signal. Arrays of Fabry–Perot cavities consisting of scatterers exhibit high spatial resolution and strong multiplexing in distributed high-temperature measurements, with each cavity possessing a loss of 0.0009 dB [16]. 860 permanent scatterers were directly prepared in the core with a minimum loss per point of 0.0007 dB [17]. The insertion loss (IL) was 0.0001 dB per point when the permanent scatterer boosted the backward Rayleigh scattering by 17 dB [18]. A common feature of these point-type scatterers (or reflectors) is the fact that they are located in the fiber core. However, the refractive-index fluctuations of the core directly deplete the transmitted light, which causes the simultaneous presence of high scattering enhancement and low loss to be impractical.

Here we show a cladding-type Rayleigh scattering enhancement (cl-RSE) structure with ultra-low loss in the single-mode fiber (SMF). The structure is composed of refractive-index modulation points inscribed in the fiber cladding close to the core using the femtosecond laser point-by-point inscription method. Actually, the mode field diameter of the SMF is slightly larger than the geometric diameter of the core, which results in the vast majority of the transmitted light being concentrated in the core, with only a small portion of the energy in the cladding in the form of an evanescent field. The cl-RSE structure simply disrupts the wave-vector propagation direction of the evanescent field, and part of the light is recycled in the form of backward Rayleigh scattering. This process does not involve the transmitted light in the core, which results in extremely low IL values. The effects of the offset from the core edge, the pulse energy, and the interval between scatterers on the Rayleigh backscattering intensity are investigated. After optimizing the inscription parameters, the loss characteristics and the behavior at high temperatures of the cl-RSE structure are tested, and the cl-RSE structure has a 10 times lower loss than the core-RSE structure. Meanwhile, it exhibits superior stability at high temperatures. The cl-RSE structure with ultra-low loss may have application potential for improving the performance aspects of distributed sensors.

The SMF used in the experiments was a standard commercial silica-core fiber (Corning, SMF-28 Ultra, 8.2/125 μm) without any pre-processing, and the fiber was fixed on a high-precision three-dimensional displacement stage (Newport, XMS100). A femtosecond laser with a wavelength of 800 nm and a pulse duration of 50 fs was used, and the laser repetition frequency was fixed at 200 Hz. The Gaussian femtosecond laser beam was focused by a microscope objective ($40\times$, $\mathrm{NA}=0.75$), and the laser focus was ensured to be located at the optical fiber cladding by means of a confocal device behind the objective. During the preparation process, the interval between each scatterer can be controlled by a pre-set moving speed. Meanwhile, one end of the fiber was connected to a commercial optical backscattering reflectometer (LUNA, OBR 4600) to detect the Rayleigh backscattering signal. The schematic diagram of the preparation process is shown in Fig. 1(a). We set the parameters of the preparation as follows: the scatterer interval is 10 μm (corresponding to a moving speed of $2\mathrm{mm}{\mathrm{s}}^{-1}$), the offset from the core edge is 2 μm, and the single pulse energy is 285 nJ. Figure 1(b) shows the backscattering profile of a typical cl-RSE structure of a length of 100 mm, which clearly exhibits more than 20 dB of scattering enhancement.

Images of the sides of the cl-RSE structure were recorded by optical microscopy, as shown in Fig. 1(c). The cl-RSE structure was set with an interval of 10 μm between the scatterers and 2 μm offset from the core edge, and no additional modulation points appeared in the fiber core. Then, the cl-RSE fiber section was cleaved and the morphology of its cross section was observed by SEM. Figure 1(d) shows the SEM image of the eccentric refractive-index modulation point, and it can be observed that the modulation point is composed of sub-micrometer period nano gratings and that the cladding nano gratings are similar to the structure of femtosecond laser-induced self-organized nano gratings [15,20].

In order to demonstrate the scattering operation of the cl-RSE structure in the fiber, we simulated the electric field mode distribution on the side of the fiber using the finite element simulation method, as shown in Figs. 2(a) and 2(b). The parameters of a standard SMF-28 Ultra fiber ($n_{\mathrm{core}}=1.4682$, $r_{\mathrm{core}}=8.2\mathrm{\mu m}$, $\mathrm{NA}=0.14$) were used to build a 2D fiber planar model, with the fiber diameter set to 30 μm to reduce the computational effort. In modeling, the refractive-index modulation point was simplified to a void-shell structure, with diameters of 0.5 μm and 1 μm for the void and shell, respectively [21,22]. The average effect of the void and shell was further taken into account, with refractive-index variations averaged over $\mathrm{\Delta }{n}_{\mathrm{void}}=-0.02$ and $\mathrm{\Delta }{n}_{\mathrm{shell}}=+0.02$ [23]. The periodic refractive-index variations in the cladding are shown in Fig. 2(b), with an interval of 10 μm and an offset of 2 μm. The simulation results for 100 cladding scatterers are shown in Fig. 2(c), which clearly show that because of the existence of the cl-RSE structure, there are more chaotic energy distributions in the cladding, and in particular the energy distributions on both sides of the scatterers are more intensive. In the diffraction characteristics of FBG, the diffraction of a specific wavelength into the cladding will have several specific angles [24,25]. Unlike FBG, the larger interval of the cl-RSE structure eliminates this angular correlation. In other words, the larger interval (similar to the period of FBG) allows the angle of diffraction to increase to an indistinguishable angle. The electric field strength distribution on the flanks was monitored, as shown in Fig. 2(d). The scattering intensity is highest in the region where the cl-RSE structure is present, and the effect of scattering gradually decreases until zero as it continues to propagate forward. It indicates that the transmission properties in the overall fiber are not disrupted by the cl-RSE structure, which provides for a scattering-enhanced structure with low transmission loss. To further illustrate the phenomenon of recycling cladding-transmitted light by the cl-RSE structure, we compare the electric field energy distributions in the radius direction of the ordinary SMF and the cl-RSE SMF. Compared to the smoothly transmitted electric field in the ordinary SMF, with the addition of scattering points, the scattered light is not directional because the refractive-index modulation points take on an irregular ellipsoidal shape, and the mode field distributions in the two cases are shown in Figs. 2(e) and 2(f). As shown in Fig. 2(g), the intensity distributions were monitored near the input port ($z=10\mathrm{\mu m}$), where it can be seen that the electric field energy at the core position of the fiber is almost unchanged, but the cladding energy intensity of the cl-RSE fiber (blue solid line) is significantly higher than that of the ordinary SMF (red dashed line), and these extra parts are obviously scattered by the cl-RSE structure appearance in terms of backward scattering.

The effect of different cl-RSE structures on the enhancement of backscattered signals is shown in Fig. 3, where we focused on three distinct parameters: the offset from the core edge, the pulse energy, and the interval between the scatterers. Micrographs of representative samples at different parameters are shown in Fig. 3(a). The offsets of the modulation points were set to vary from 1 to 5 μm in 1-μm steps. At the same time, the Rayleigh scattering profiles in the fiber were recorded with the OBR, and the enhancement of the backscattered signal strengths in the inscribed portion was accounted for in Fig. 3(b). No significant RSE was observed in the OBR beyond 5 μm. In addition, the RSE effect was more pronounced the closer the cl-RSE structure was to the core, which was due to the Gaussian distribution of the optical field in the SMF, where the intensity in the cladding had a sharp attenuation along the radius to zero compared to the vast majority of the energy present in the core, and this phenomenon was also reflected in the enhancement of the Rayleigh scattering at the scatterer of the cladding.

Six cl-RSE structures with an interval of 10 μm and an offset of 2 μm were prepared by increasing the laser single pulse energy from 248 to 310 nJ, and the corresponding RSEs were recorded in Fig. 3(b). As the energy increased, the color of the modulation points in the micrographs was deepened and there was a slight increase in the size of the modulation points, as shown in Fig. 3(a), where the enhancement of the backward signal strength increased with the increase in pulse energy. However, when the energy continues to increase, the modulation region in the micrograph shows a significant outward expansion, probably because the laser energy reaches the threshold of the void-type modulation [26]. The expansion of the refractive-index modulation region into the fiber core brings additional loss. Therefore, the laser energy was kept at a moderate 285 nJ in the subsequent experiments to obtain a more stable scattering enhancement.

We prepared cl-RSE structures with different intervals, and the repetition frequency of the pulsed laser was 200 Hz to change the moving speed of the fiber ($\mathrm{interval}=\mathrm{speed}/200\mathrm{Hz}$), and cl-RSE fibers with 0.5, 1, 2, 5, 10, 20, and 2000 μm intervals were prepared. The enhancement of the backward signal intensity for 0.5–20 μm is shown in Fig. 3(c), and, when the intervals are $<1\mathrm{\mu m}$, the modulation points show a continuous shape, and it should be noted that the diameter of each modulation point is about 1 μm, and the extremely high-density nano grating contributes the most to the scattered signal [27]. When the interval is 0.5 μm, multiple pulses may cause the modulation effect to be erased and the enhancement effect to be reduced. For separated modulation points, there is no significant difference in the enhancement of the backward signal.

The modulation points with an interval of 2000 μm were prepared in a length of 20 mm, and Fig. 3(d) shows the 11 measured modulation points, which can clearly distinguish the position of each modulation point and facilitate the analysis of the action of individual modulation points. The results show that even only one refractive-index modulation point can provide more than 20 dB enhancement of the backscattered signal, while there is a 9 dB difference in the enhancement due to the dependence of the cl-RSE on the offset and laser energy instability. To balance this discrepancy, the interval of the cl-RSE fiber used for subsequent characterization was chosen to be 10 μm.

The wavelength dependence of the cl-RSE structure was characterized. The transmission spectra were measured using a supercontinuum light source (NKT, Superk Compact) and an optical spectrum analyzer (Yokogawa, AQ6370D). Figure 4(b) shows the transmission spectra of a 100 mm long fiber sample in the range of 700–1700 nm and no significant loss of wavelength in the 1000 nm range. Transparent transmission in the NIR band promotes the application of cl-RSE fiber in the integration of communication and sensing.

Femtosecond laser-prepared FBGs [14], Rayleigh scattering centers [15], and micro-cavity arrays [16] have shown excellent reliability in measurements at high temperatures. In the test, the spatial resolution of the OBR was set to 0.1 mm, and the wavelength scanning range was 1530.000–1571.713 nm. A 100 mm long cl-RSE fiber was used to test high-temperature characteristics. The temperature was set to cycle from room temperature to 800°C [Fig. 5(a)], with the temperature increasing by 50°C every 10 min and remaining stable for 20 min. The frequency shift in the three temperature cycles is shown in Fig. 5(b), where the temperature response of the cl-RSE fiber is about 1.7–1.8 $\mathrm{GHz}°{\mathrm{C}}^{-1}$. The sample fiber shows good consistency during the temperature cycles.

The long-term stability of cl-RSE fiber is shown in Figs. 5(c1)–5(c3). We prepared three 100 mm long cl-RSE fibers with the same parameters and kept them at 800°C, 900°C, and 1000°C for 10 h. Finally, they were naturally cooling down to room temperature, and the Rayleigh scattering profiles in the fibers were monitored every half an hour during the high-temperature holding. The results showed that the Rayleigh scattering profile at 800°C did not change compared to that at room temperature. Meanwhile, some additional scatterers appeared in the Rayleigh scattering profile when the sample fiber was heated at 900°C for 6 h. At 1000°C, more scatterers appeared after only 2.5 h. The appearance of these scatterers may be due to additional defects above 800°C when the silica starts to soften. However, the Rayleigh scattering intensity enhancement caused by the cl-RSE structure is still clearly visible, indicating that the cladding scatterers can survive at 1000°C [29,30]. These results suggest that the fiber used in the experiments softens above 800°C and that the fragile fiber itself is a major constraint on the sensing performance after softening.

An exciting point is that the cl-RSE fiber demonstrated higher stability than the unmodulated fiber. Temperature stability measurements were made by first measuring the Rayleigh scattering characteristics once at room temperature, and this data was stored as a shift reference. Keeping the fiber position and temperature, etc., constant, measurement was taken once after 10 min, and the amount of temperature change was recorded. The results in Fig. 5(d) show that the temperature fluctuation of a 100-mm portion of the cl-RSE fiber is 0.0262°C, which is a factor of 2.3 lower than that of the unmodulated fiber. After annealing at 800°C as described above, we tested the temperature fluctuation in the same way and found that the temperature fluctuation of the cl-RSE fiber portion is only 0.00273°C, which is nearly 10 times lower compared to that before annealing and 9.1 times lower than that of the unmodulated fiber portion [Fig. 5(e)]. The annealing process eliminates defects such as stress residues at the modulation point. This result shows that the cl-RSE structure enhances the backscattered intensity by about 20 dB, which improves the temperature measurement accuracy by about 10 times. Using fibers with cl-RSE structure is expected to realize ultra-high accuracy temperature measurement at high temperatures.

In summary, we proposed and demonstrated an ultra-low loss Rayleigh scattering enhancement structure in SMF by recycling light in cladding. The cl-RSE structure is inscribed into the cladding of SMF by the femtosecond laser point-by-point direct-writing technology. The simulation results show that the cl-RSE structure is able to enhance scattering in the fibers efficiently without affecting the forward transmitted light in the core. In addition, the scattering enhancement capability of the cl-RSE structure is shown to be flexibly controlled by offset distance, interval, and pulse energy, providing at least 20 dB of scattering signal enhancement per scatterer under suitable parameters. We tested the corresponding IL and demonstrated an IL as low as 0.00001 dB per scatterer, which is 10 times lower than that of the scattering enhancement structure in the fiber core. Finally, the cl-RSE fiber was characterized at high temperatures with a temperature response of about 1.7–1.8 $\mathrm{GHz}°{\mathrm{C}}^{-1}$. After annealing, the temperature measurement of the cl-RSE fiber fluctuated by only 0.00273°C. Transferring the region of the scattering center from the fiber core to the cladding provides a new inspiration for femtosecond laser preparation of scattering-enhanced fibers, in which scattering enhancement is achieved by recycling the cladding light, significantly reducing the impact on insertion loss.