Various light sensors are being developed and utilized to measure light intensity in different wavelength bands such as ultraviolet, visible, and infrared. A photo-diode based sensor, an electronic sensor, is the most representative and is used in various fields[1]. Recently, smart farms, smart livestock, and smart aquaculture are aiding the growth of agricultural, livestock, and marine products using light emitting diodes (LEDs)[2-4]. Because of the nature of smart farms, the light intensity must be measured over a wide area and the light intensity must be appropriately adjusted for the target object[5]. Point-measurement-based electronic sensors require a power supply and electrical signal measurement using wires in humid environments, making installation complex. When a malfunction occurs, frequent replacements result in significant time and economic losses. To address these issues, research on optical visible light sensors is necessary. Fiber Bragg grating (FBG) sensors, renowned in the optical sensing domain, are predominantly deployed for temperature and strain measurements at multiple points[6, 7]. Notably, these sensors demonstrate exceptional corrosion resistance in moisture-laden environments and remain impervious to external electromagnetic interferences[8]. Harnessing the intrinsic merits of FBG sensors and pioneering the development of fiber optic-based sensors for visible light detection can potentially lead to a new spectrum of applications.

Recently, Shivananju et al. introduced an innovative broadband optical detector that encompasses the visible light spectrum by integrating graphene onto the FBG surface[9]. This design is based on the principle that graphene absorbs incoming light and generates a photothermal effect. This thermal response subtly modulates the refractive index within the FBG grating, culminating in a discernible shift in its central wavelength. This detector has significant potential as an adaptable fiber-optic sensor for visible light detection. However, the synthesis and subsequent transfer of graphene onto the FBG surface can be complicated. Notably, transferring graphene onto a curved optical fiber surface poses significant technical challenges, potentially increasing the sensor production costs.

Carbon nanotubes (CNTs) are allotropes of carbon characterized by their cylindrical nanostructures[10]. A CNT possesses various properties; notably, when light is incident on a CNT, its electrical conductivity increases, and it bends in the same direction as the propagation of light[11]. Previous studies indicated that structural distortion caused by surface van der Waals forces could also modify the electronic structure of CNTs[12, 13]. The elastic response of CNT bundles to visible light was investigated for many actuator applications such as micro-cantilevers, nano-tweezers, and nano-robots[11, 14, 15]. They observed the elastic movement of the filaments of single-walled carbon nanotube (SWCNT) bundles under visible light illumination. This elastic behavior was explained by the electrostatic interaction of the SWCNT bundles as a result of a photovoltaic or light-induced thermoelectric effect that was physically related to the modification of the electronic structure during bundle formation. This deformation caused both filaments to move away from the light source. It is noted that Shivananju et al. also reported photomechanical optical modulator based on an FBG coated directly with vertically aligned multi-walled CNTs using chemical vapor deposition[16]. It is a good approach of light sensor, but the CNT coating process is still complicated for manufacture of fiber optics sensors.

In this study, we introduced a fiber-optic sensor for visible light detection by integrating a CNT and an UV-cured elastic polymer composite as a coating on an FBG. The deformation of the CNT bundle triggered by incident visible light subtly modifies the grating period in the FBG. This causes a corresponding shift in the central wavelength of the FBG, that, when measured, provides an indication of the intensity of the incoming visible light. To evaluate this principle, we prepared samples by coating the FBG with an SWCNT-inclusive polymer and then assessed the photo-response within the 400–700 nm visible light range. To enhance the detection sensitivity, we fabricated samples with an FBG diameter reduced to 80 μm; the 80 μm FBG sensor exhibited approximately double the sensitivity of its 125 μm counterpart. Given its superior responsiveness to visible light, combined with minimal measurement errors, this newly proposed sensor demonstrates promising potential for optical sensing applications.

Figure 1(a) shows the solution prepared by mixing SWCNT powder (900711, Sigma-Aldrich Inc.) with deionized water in a small bottle and stirring for 30 min with a stirrer (HZ1, LABTron Inc.) to loosen the clumped SWCNT powder. The stirred SWCNT solution was transferred onto a glass slide, as shown in Fig. 1(b). It was then dried on a heating plate (HSD 150, Misung Scientific Inc.) at approximately 80 °C for 3 h. A microscale image of the dried SWCNT bundle sample was captured using a scanning electron microscope (SEM, SNE-4500M, SEC Inc.), as shown in Fig. 1(c). The dried SWCNT bundles were peeled off the glass slide and mixed with an ultraviolet (UV)-curable acrylic resin using a magnetic stirrer for 24 h. It was then uniformly coated to a thickness of 900 μm using a UV recoater (MiniCoater 2, Nyfors Inc.). Two types of optical fibers were used for this purpose: an FBG with a diameter of 80 μm to enhance the device sensitivity, and another FBG with a diameter of 125 μm for comparison. We wrote the FBG on a commercially available optical fiber with an 80 μm diameter (SM980G80, Thorlabs Inc.). The grating length of both FBGs was 5 mm, and their center wavelengths were located around the 1550 nm wavelength bands. Figure 1(d) shows the visible light FBG sensors coated with the SWCNT-containing polymer on both FBGs with a coating length of approximately 5.3 cm.

Figure 1. (a) Stirred sample after mixing CNT power and DI water, (b) Distributed SWCNT bundles dried on the heating plate, (c) SEM image of the sample, (d) 80 μm FBG and 125 μm FBG coated by the SWCNT-containing polymer with a coating diameter of 900 μm.

Figure 2(a) shows a schematic of the response of the SWCNT-coated FBG sensor when exposed to visible light emitted by an LED. When visible light from the LED reached the SWCNT, the SWCNT bent toward the direction of light propagation[11]. This reversible bending of the SWCNT induced deformation in the acrylate polymer as the coating material on the FBGs, that in turn, lead to a change in the grating period of the FBG. Consequently, the central wavelength of the FBG reflection spectrum shifted toward longer wavelengths. Figure 2(b) shows the wavelength spectrum of the LED (rear flash LED, iPhone 13 Pro, Apple Inc.) used in the experiment, measured using a compact CCD spectrometer (Thorlabs, CCS100). The LED had a bandwidth ranging from approximately 450 to 700 nm.

Figure 2. (Color online) (a) Conceptual representation of the operating mechanism of the proposed SWCNT-coated FBG visible light sensor and (b) optical spectrum of the visible LED employed in the study.

Figure 3 shows the center wavelength shift of the SWCNT-coated FBGs with diameters of 125 μm and 80 μm in response to the various optical powers of the LED. The wavelengths of both FBG sensors shifted toward longer wavelengths as the LED power increased. When the light was turned off, it returned to its starting wavelength. Notably, the visible light sensitivity of the diameter-downsized FBG improved compared with that of the FBG with 125 μm diameter. By observing the shift in the central wavelength of the FBG owing to the externally introduced visible light, the intensity of the incoming visible light could be measured.

Figure 3. (Color online) Center wavelength shift of SWCNT-coated FBGs in response to visible LED power ranging from 0 to 25 mW: (a) FBG with 125 μm diameter and (b) FBG with 80 μm diameter.

We set up our experiment as depicted in Fig. 2(a) and subjected both sensors, as shown in Fig. 1(d), to visible light. Figure 4 displays the data captured every 2 s using an optical spectrum analyzer (OSA, MS9710C, Anritsu Inc.) with a wavelength resolution of 0.02 nm that traces the central wavelength shifts of the SWCNT-coated FBG sensor. Using the peak detection data from the sensors, as shown in Fig. 3, we examined the manner in which the wavelength varied over time owing to visible light exposure. The LED used in the experiment had an output power of 25 mW. Initially, we activated the LED by exposing the sensor to visible light for 60 s. After this interval, the LED was deactivated, and we noted changes in the central wavelength. When the LED was powered on, a notable spike in wavelength was evident; however, near the 60-s mark, changes slowed, hinting at saturation. The subdued responsiveness of the sensor was not primarily a result of the inherent response rate of the SWCNTs, but rather the dampening effect of the acrylate polymer. With further research on refining the composition of the acylate polymer, we anticipate that the response time of this sensor will be enhanced. Concerning the 80 μm FBG, a maximum wavelength shift of 0.58 nm was observed post-60 s, whereas it was 0.32 nm for the 125 μm FBG. The findings confirmed that the SWCNT optical sensor crafted using the 80 μm FBG exhibited a responsiveness rate 1.8 times superior to that of the 125 μm FBG. This higher responsiveness could be attributed to the reduced thickness of the 80 μm FBG that efficiently conveyed force as the CNT-infused acrylate polymer elongated. After deactivating the LED after 60 s, we noticed a swift central wavelength transition akin to the shift upon LED activation, with the central wavelength of the optical sensor reverting to its original value before the 60-s period concluded. The rapid center wavelength adjustments in response to the LED on/off cycles underscored the impressive sensitivity of the optical sensor. For a comparative evaluation of the role of SWCNTs, we applied only an acrylate polymer without an SWCNT onto the 125 μm FBG and conducted an identical test. As shown in Fig. 4, the sensor was nonresponsive to visible light.

Figure 4. (Color online) Visible light-induced wavelength shifts of the SWCNT-coated FBGs with 125 μm and 80 μm diameter (obtained from peak-detection in the FBG reflection spectra of Fig. 3(a) and (b)). In addition, 125 μm-diameter FBG coating only acrylate polymer is tested.

To verify the reproducibility of the experiments using the fabricated SWCNT-coated FBG sensors, repeated tests were conducted for each sensor. The experiment was repeated five times, turning the LED power on and off at intervals of 60 s between the ON and OFF states. Figure 5(a) displays the graph for five repeated tests on the 125 μm FBG coated with SWCNT, whereas Fig. 5(b) presents the graph for five repetitions on the optical sensor coated with SWCNT on the 80 μm FBG. As observed from the graphs, the 125 μm FBG SWCNT optical sensor exhibited a center wavelength shift of 0.24–0.26 nm, and the 80 μm FBG SWCNT optical sensor showed a shift of 0.48–0.5 nm. Furthermore, after multiple repetitions, the central wavelength returned to its initial value, indicating that the hysteresis of the SWCNT-coated FBG sensor was low. The consistency in the amount of change in the central wavelength across the tests suggested that the sensor exhibited good reproducibility.

Figure 5. (Color online) Responses in wavelength induced by visible light over five repetitions for (a) 125 μm FBG and (b) 80 μm FBG coated with a SWCNT-containing polymer.

Figure 6 presents a graph of the wavelength shift in response to the LED power. The LED light intensity was incrementally increased by 5% from 0 to 25 mW, and its power was measured using a power meter (1936-R, Newport Inc.). The SWCNT-coated FBG sensor was positioned at the same height as the power meter when measuring the LED power, and the wavelength shift corresponding to the LED power was recorded. After repeating the experiment five times, the averages and standard deviations are as shown in Fig. 6. The maximum standard deviation was 0.01 nm that was smaller than the resolution of 0.02 nm set to the OSA, indicating that both sensors had good reactivity to visible light. The graph aligned well with the second-degree polynomial fitting that appeared to reflect the response characteristics of the SWCNT to visible light. The nonlinear responses of the FBG sensors to visible light were caused by the physical interlinks between the electrostatic, elastic, and thermal effects in nanotube composite actuators[11]. The measurement sensitivity of the 125 μm FBG SWCNT sensor was 0.0128 nm/mW, whereas that of the 80 μm FBG sensor was 0.0240 nm/mW.

Figure 6. (Color online) Wavelength shifts in SWCNT-coated FBGs with diameters of 80 μm and 125 μm in response to the power of visible light.

We investigated the shift in the center wavelength of FBG sensors with diameters of 80 μm and 125 μm when coated with UV-curable polymer mixed with SWCNTs , in response to the intensity of visible light. The underlying principle of the experiment was that the SWCNT coated on the FBG bent in the direction of incoming visible light. This bending effect was subsequently transmitted to the elastomer. The force transferred to the elastomer influenced the spacing of the FBG grating pattern, leading to a change in its center wavelength. Using an FBG with a diameter of 125 μm, we observed the response of the FBG optical sensor, coated with both an SWCNT and elastomer, to visible light. Notably, employing an 80 μm diameter FBG induced a more significant shift in the center wavelength under the same light intensity. Specifically, the 80 μm FBG exhibited a shift almost twice as large as that of the 125 μm FBG. We ascertained that the FBG optical sensors coated with SWCNTs were effective in detecting the intensity of light within the visible range. The results of this study highlight the potential utility of FBG optical sensors, particularly in the domain of visible light detection.

This research was supported by “Regional Innovation Strategy (RIS)" (2021RIS-002) through the National Research Foundation of Korea(NRF) funded by the Ministry of Education(MOE) and Collabo R&D project (RS-2023-00225059) between Industry, University, and Research Institute funded by Korea Ministry of SMEs and Startups in 2024-0025.

The authors acknowledge Prof. Min-Ki Kwon (Dep. Photonic Engineering, Chosun University, Korea) for fruitful help in CNT material preparation.