The polymer coating of an optical fiber protects it by reducing the impact of external effects such as nicks, scrapes, shocks, and moisture on the glass cladding. Without such coatings, an optical fiber is fragile to the point where a single microscopic nick in the cladding can cause a crack to propagate when it is under strain [1], leading to breakage of or loss within the fiber. Some types of fiber lasers even rely on the coating to confine pump light [2]. For high-power fiber laser applications, the coating quality in terms of uniformity and optical loss is very important. Minor coating defects must be spotted early and rectified so they do not lead to light-scattering and thermal effects that can lead to catastrophic fiber failure [3]. For the telecommunications industry, fibers are deployed underground for many years and coating defects must be rectified if the fibers are expected to last [4]. Hence, it is of great importance to detect microscopic defects during manufacturing. Although there are point-by-point measurement techniques for inspecting the quality of coatings [5,6], they operate by analyzing the scattered light from external illumination, which is not practical for long lengths of fibers. Simply guiding light along the fiber is insensitive to coating defects. Illuminating the fiber with visible light and checking for bright spots [7] is not feasible in outdoor environments or when the fiber has additional buffers and/or jackets that make them opaque to the outside. Recently, skew rays have been explored for enhancing fiber-optic sensors [8] and for extracting uniform fiber loss coefficients through the theoretical fitting of the experimental data on angle-resolved power [9]. Here, we demonstrate a highly sensitive interrogation technique for detecting microscopic defects within optical fibers and their coatings, based on the excitation and detection of skew rays. This enables the analysis of detecting microscopic defects within coatings.

Skew rays [10,11] are generated when meridional rays encounter the curved geometry of a cylindrical multimode fiber, as illustrated in Fig. 1(a). They propagate in a helical pattern, which results in a greater number of Fresnel reflections per unit length than meridional rays. Additionally, skew rays facilitate a circular coverage shown in Fig. 1(b) along the cladding-coating interface. The combination of these two factors makes skew rays significantly more sensitive to surface imperfections than meridional rays. The number of reflections per unit length can be expressed as follows [12]:

 

Fig. 1. Illustration of skew ray propagation with (a) axial and transverse angles and (b) the circular coverage of the fiber cross section.

Download Full Size | PDF

It is clear from Eq. (1) that higher-order skew rays with larger ${\theta }_{\phi }$ and a larger launch angle ($\theta$) (i.e., between the rays and the fiber axis leading to a smaller ${\theta }_z$) yield a larger number of reflections per unit length. This gives rise to a greater chance of detecting defects in close proximity to the cladding-coating interface.

Figure 2 shows the schematic of the interrogation technique. An unpolarized single-mode He-Ne laser source (i.e., Thorlabs HNLS008R-EC) was used to excite the fiber under test (FUT). The beam excites the FUT with a spot diameter much larger than the cladding diameter of the FUT (i.e., 5:1 ratio), which provides a uniform intensity distribution. The FUT was mounted on a motorized rotation stage (i.e., Thorlabs CR1/M-Z7) such that the end-face was positioned at the center of rotation. Although skew rays can also be generated from focused light, the contribution from low-order skew rays limits the sensitivity as it reduces the total number of reflections, in addition to being alignment critical and unstable. The transmitted rays were collected by an integrating sphere (i.e., Thorlabs S142C) connected to a power meter (i.e., Thorlabs PM100A). A reference thermopile (i.e., Newport 919P-003-10) connected to a power meter (i.e., Newport 843-R) collected a fraction of the output power from the He-Ne laser during each measurement to normalize the transmitted power in the presence of low-frequency laser intensity drift. A division by the cosine of $\theta$ was applied to compensate for the reduced input power from the Lambertian angle effect in order to achieve a flat-top angle-resolved power profile when the FUT has low loss. No compensation was needed for the angle-dependent reflectivity, because its contribution at $\theta <45\mathrm{deg}$ is negligible. These effects shape the profile but do not affect the attenuation associated with defects.

 

Fig. 2. Schematic of the interrogation system.

Download Full Size | PDF

The cross-sectional profile of the FUT (i.e., custom made) is shown in Fig. 3. Each strand is a silica fiber with a core diameter of 10 μm, a cladding diameter of 200 μm, a cladding-coating (i.e., Efiron PC-375 AP) diameter of 300 μm, and a fiber length of 1 m. Note that kilometers of fiber can be used as long as the received power is sufficient for detection.

 

Fig. 3. Labeled cross section of the double-clad FUT.

Download Full Size | PDF

To demonstrate the effectiveness of the interrogation technique at detecting microscopic nicks in the coating, a surgical blade was used to gently slice the coating, creating nicks that are roughly 20 μm wide, 80–120 μm long, and 50 μm deep, as shown in Fig. 4(a).

 

Fig. 4. Photograph of (a) a nick in the coating, (b) the rubber-coated section, and (c) the microsphere inside the cladding.

Download Full Size | PDF

When the FUT was excited with red light at $\theta =0\mathrm{deg}$, light scattering was not visible to the naked eye. With $\theta >5\mathrm{deg}$, the refractive-index (RI) perturbation induced by the nicks scatter light guided within the cladding, and these defects become visible as bright spots along the FUT, as shown in Fig. 5(a). Finally, the blade was used to scrape a 25 mm patch off the coating to link the nicks, in order to compare its impact with that of individual nicks. Figure 5(b) shows that the affected section glows under red light excitation (i.e., appears bright from $\theta$ and long exposure time) due to the nonuniform RI change of the debris that scatters light. The inset photographs taken with a microscope show that if the scrape is cleaned with alcohol, the removal of debris reduces scattering.

 

Fig. 5. Photograph of (a) the nicks scattering light; (b) the scrape scattering light, inset: cleaning effect; (c) the high-index recoating absorbing light; (d) the spray-on rubber recoating absorbing and scattering light; and (e) the internal microsphere scattering light when excited with angled red light.

Download Full Size | PDF

The interrogation system was configured such that $\theta$ could be varied from 50 deg to $-50\mathrm{deg}$. The collected data is angle resolved in Fig. 6. $\theta$ is converted from degrees into numerical aperture $(\mathrm{NA}=\sin (\theta ))$ units to help identify the NA of the FUT. The angle-resolved power profile is not flat at the top within the cutoff NA of meridional rays because the low-RI PC-375 coating of the double-clad FUT is lossy and thus attenuates light based on its number of reflections. The symmetrical tails at high $\theta$ are attributed to the gradual leak of skew rays from low order to high order. Upon close inspection, the profile reveals that at $\theta =0\mathrm{deg}$, the impact of the nicks is negligible, as might be expected because rays at this angle travel through the FUT without exposure to its surface. However, by increasing $\theta$ to excite high-order skew rays, there is observable attenuation from the extra reflections, leading to a higher chance that the rays will encounter any defect located on the surface of the FUT. As a result, even a single microscopic nick is noticeable in Fig. 7, with the difference amplified from near zero to at least 0.10 dB for a single nick with increasing $\theta$. Local maximum and minimum attenuation arise from the skew rays encountering a defect fully, partially, or not at all. With a larger number of nicks, the difference is even more visible. The trend at $\theta =7\mathrm{deg}$ (i.e., $\mathrm{NA}=0.12$) is $4.7\times {10}^{-2}\mathrm{dB}/\text{nick}$ (i.e., each nick is roughly 20 μm wide, 100 μm long, and 50 μm deep). Similarly, the attenuation of the scrape is boosted from 0.5 to 6.8 dB (i.e., the linear factor of 4.3) when $\theta$ is incremented from 0 to 37 deg (i.e., $\mathrm{NA}=0.60$). It is worth noting that in reality each defect is likely to attenuate light differently due to a combination of varying shapes, sizes, and locations.

 

Fig. 6. Comparison of the angle-resolved power for a coating with an increasing number of nicks, and then a scrape that engulfs them all.

Download Full Size | PDF

 

Fig. 7. Selected region of Fig. 6 for the analysis of attenuation.

Download Full Size | PDF

Similarly, the interrogation technique was used to compare two different low-quality recoatings of coating-stripped FUTs. First, a high-RI ultraviolet (UV) polymer was chosen as the recoating material. A surgical blade was used to window strip a section of coating before it was cleaned thoroughly with isopropanol. Then, the bare fiber section was raised and fixed on a microscope slide. Lastly, polymer NOA-63 (i.e., RI is 1.56 at 635 nm) was used to fully immerse the bare fiber section, and it was UV cured for high robustness. Figure 5(c) shows that angled red-light excitation leading to severe optical leakage arising from the higher RI of the recoating than that of the silica cladding.

With another FUT, a similar window of the coating was stripped and the bare fiber section was spray coated at a distance of 5 cm with white rubber (i.e., $n>1.45$ at 635 nm), shown in Fig. 4(b), forming a recoating thickness of roughly 5 μm.

When the FUT was excited with red light, the rough texture of the spray-on rubber created a nonuniform interface between the cladding and recoating, leading to significant scattering of light as well as absorption due to the high RI. This is evident in Fig. 5(d) as a bright spot where light first encountered the rubber recoating. The angle-resolved data plotted in Fig. 8 show considerable attenuation of 4.6 and 5.6 dB at $\theta =0\mathrm{deg}$ for the rubber and NOA-63 recoatings respectively. The RI of NOA-63 is slightly higher than that of rubber, which explains the higher loss. When $\theta$ exceeds 5 deg (i.e., $\mathrm{NA}=0.09$), both recoatings strongly attenuate the guided light inside the fiber.

 

Fig. 8. Comparison of the angle-resolved power for different recoatings.

Download Full Size | PDF

Owing to the same principles, this interrogation technique can identify defects that can occur inside a fiber due to fabrication errors rather than external effects. If the defect is located between the core and cladding, the angle-resolved power between the ideal FUT and the defective FUT should match in a specific range of $\theta$. These angles correspond to radial regions in the FUT that do not encounter the defect and, thus, provide an approximation to the distance of the defect from the core center. To demonstrate this, the interrogation technique was used to probe the FUT for internal defects. Laser micro-machining the FUT can produce an internal defect such as a microsphere with an altered RI to its surroundings. If this is detectable, air bubbles imparted during fiber drawing are even more visible, as higher-RI contrast scatters light more strongly.

An in-house laser micro-machining system (i.e., 524 nm wavelength, 250 fs pulse width, 5 MHz repetition rate, 80 nJ pulse energy, 1.25 NA objective lens, $\times 100$ magnification) described by Fig. 9 was used. A high-repetition rate enabled a thermal process in which laser modifications are spherical. The variable attenuation controlled the pulse energy. The beam was frequency doubled from 1047 to 524 nm to reduce the focal spot size after the objective lens in order to fabricate an internal microsphere that fits inside the cladding. Automated 6-axis air-bearing translation stages of 4 nm encoder resolution were used to position the FUT. To align the focal spot within the FUT, a visible light source was mounted facing down on top of the objective lens, and an integration sphere monitored the power of the scattered light collected by the FUT. Finally, the beam was focused into the FUT to create an internal microsphere of increased RI and $\sim 13\mathrm{\mu m}$ diameter inside the cladding, as shown in Fig. 4(c).

 

Fig. 9. Schematic of the femtosecond-laser writing system.

Download Full Size | PDF

As expected, when excited with red light the internal microsphere scattered skew rays more than meridional rays shown in Fig. 5(e) because of their larger cross-sectional coverage during propagation. Angle-resolving the collected data plotted in Fig. 10 shows the highest attenuation of 0.52 dB at $\theta =12\mathrm{deg}$. Lower $\theta$ yielded lower attenuation, such as 0.24 dB at $\theta =0\mathrm{deg}$, while higher $\theta$ produced even less. This was anticipated as the internal microsphere was close to the core. As a result, medium-order skew rays and meridional rays interacted with the microsphere more than low-order skew rays and meridional rays, while high-order skew rays (i.e., meridional rays are cut off at high $\theta$) could have no overlap with the microsphere.

 

Fig. 10. Comparison of the angle-resolved power for a cladding with and without an internal microsphere.

Download Full Size | PDF

In conclusion, we have demonstrated a highly sensitive interrogation technique that is capable of detecting microscopic defects inside optical fibers and their coatings. One example is a coating nick with a width of 20 μm and a length of 120 μm. Other defects tested are scrapes, low-quality recoatings, and internal defects. This technique is highly valuable for the industries relying on high-quality optical fibers and requiring a simple way to test their structural integrity.