Fluorescence-Based Optical Flow Sensor with Direction-Independent Response

Fig. 1 shows the structure and key specifications of the phosphor-attached pillars fabricated in this study. The pillars were made of EPDM rubber, with a fixed length of 15.0 mm and an inner diameter of 1.8 mm. The outer diameter was varied to 2.8, 3.8, and 4.8 mm for three different configurations, hereinafter referred to as the Ø2.8, Ø3.8, and Ø4.8 pillars, respectively.

Fig. 1.

Structure and dimensions of the phosphor-attached pillars. Each pillar was classified according to its outer diameter, namely (a) Ø2.8, (b) Ø3.8, and (c) Ø4.8, while maintaining the same height of 15.0 mm and inner diameter of 1.8 mm. The bottom images show the hollow structure observed from the top of each pillar and actual photographs of the phosphor attached to the upper surface.

A hollow core with a diameter of 1.8 mm was formed at the center of each pillar to function as an optical waveguide. In this structure, the excitation light exposed from a blue LED (470 ± 10 nm) was guided through the hollow channel to the phosphor on the upper surface of the pillar, and the red fluorescence emitted from the excited phosphor was then transmitted back through the same optical path to the camera. This optical path design minimizes scattering losses towards the filler exterior and enables stable detection of changes in fluorescence intensity.

A phosphor was attached to the upper surface of the pillar to visualize fluorescence responses under varying flow conditions. The phosphor consisted of a spherical bead made of soda-lime glass (HCMS-P-SLGS-FMR, Ø2 mm, Cospheric, Goleta, CA, USA) coated with a red fluorescent material. The excitation and emission wavelengths of the phosphor were 575 and 607 nm, respectively, and the phosphor could be effectively excited by the 470 nm blue LED source owing to its broad absorption characteristics.

The phosphor was attached to the upper central surface of the pillar using an adhesive to ensure alignment between the optical axis of the hollow channel and center of the phosphor. The attached phosphor was cured at room temperature for more than 24 h to ensure bonding stability. An instant-curing adhesive (Loctite 406, Henkel Loctite Co., Ltd., Shandong, China) with significant adhesion to rubber was used for this process.

Fig. 2 shows a schematic of the experimental setup used to evaluate the flow-response characteristics of the phosphor-attached pillar. The system consisted of a peristaltic pump, a flow test chamber, a fluid reservoir, and an optical observation system.

Fig. 2.

Schematic of the experimental setup for evaluating the flow-response characteristics of the phosphor-attached pillar.

A peristaltic pump (Masterflex I/P® Digital Process Drive, Masterflex, Vernon Hills, IL, USA) coupled with Puri-Flex® Tubing I/P 26 (Masterflex, USA) was employed for fluid circulation. This pump system was used to circulate the fluid, and the flow rate was controlled within the range of 0–3000 mL/min, considering the pump capability and inner diameter of the tubing.

Triple-distilled water was used as the circulating fluid to minimize optical interference caused by impurities. The pump was carefully configured such that the fluid flowed in the direction indicated by the dashed arrows, aligning the inflow nozzle center with the phosphor center to ensure that the flow penetrated the phosphor frontally. The phosphor-attached pillar was vertically fixed inside the flow test chamber, and an observation window was installed at the top to enable non-contact optical measurement of fluorescence responses.

The flow test chamber was constructed as an internal rectangular structure with dimensions of 130.0 mm × 73.5 mm × 53.7 mm. Inlets and outlets were configured on opposite sides of the chamber, allowing division of the flow conditions by the inflow direction. The frontal flow condition refers to the fluid entering through a nozzle on the short side (73.5 mm direction) and impinging the front pillar surface; in contrast, the lateral flow condition corresponds to fluid entering from a nozzle on the long side (130.0 mm direction), interacting with the pillar from a different direction. The distance between the nozzle and pillar, as well as the phosphor position, were maintained identical in both conditions so that the internal flow distribution was changed solely by the length difference of the flow path from the inlet to outlet.

All experiments were conducted using the same top-view observation, allowing for direct comparison of the fluorescence response characteristics under identical observation directions but different flow directions.

An optical observation system was constructed to measure the fluorescence response characteristics of the phosphor-attached pillars in a non-contact manner.

A USB camera (acA2440-75um, Basler, Ahrensburg, Germany) was used for fluorescence signal detection. The camera was equipped with a telecentric lens (CompactTL^TM, Edmund Optics, Barrington, NJ, USA) and a band-pass filter (center 607 nm, FWHM 36 nm; Edmund Optics, Barrington, NJ, USA). The telecentric lens minimized the focus dispersion caused by variations in the phosphor depth, reducing distortion of the observed signal, whereas the band-pass filter blocked the excitation light from the LED (470 ± 10 nm) while selectively transmitting only the fluorescence signal emitted from the phosphor (607 ± 36 nm).

A ring-type LED light source (SpecBright^TM; ProPhotonix, Boston, MA, USA) was used as the excitation source. The LED was positioned vertically on the upper surface of the chamber to deliver the excitation light uniformly along the inner channels of the pillars. The optical axis of the LED was aligned with the central axis of the pillar to maintain a consistent excitation intensity of the light impinging on the phosphor. The pillar was fixed onto a glass slide, and the red fluorescence emitted from the phosphor was passed through the band-pass filter to be captured by the camera.

The optical system parameters for the fluorescence response measurements were maintained identical throughout all experiments. The LED light source was operated in the constant-current mode (24 V, 350 mA), and the camera settings were fixed at a gain of 36, an exposure time of 70 ms, a frame rate of 10 fps, and gamma of 0.8. These parameters were selected to prevent fluorescence signal saturation while ensuring a sufficient signal-to-noise ratio.

The flow experiments started with recording the reference fluorescence intensity under a flow rate of 0 mL/min, followed by gradually increasing the flow rate. Fluorescence signals were acquired at 1000, 2000, and 3000 mL/min. Data were collected for 20 s from each flow rate, and the experiments were repeated under identical conditions for all three pillar types (Ø2.8, Ø3.8, and Ø4.8).

The acquired fluorescence images were processed using the ImageJ software (version 1.54k, National Institutes of Health, USA). The phosphor region was designated as the region of interest (ROI), and both the mean grayscale intensity and maximum intensity of each frame were calculated. To avoid signal saturation, the system was adjusted so that the maximum pixel brightness did not exceed 255 on an 8-bit scale. Using these calibrated conditions, the characteristics of the fluorescence intensity response to varying flow velocities were quantitatively analyzed and the differences in the responses among the pillar geometries were systematically compared.

Fig. 3 shows the quantitative analysis results of the brightness variations in the fluorescence images.

Fig. 3.

Quantitative analysis of fluorescence intensity under varying flow conditions. (a) Representative fluorescence images acquired at flow rates of 0, 1000, 2000, and 3000 mL/min. The yellow circular mark in each image indicates the region of interest (ROI) used for analysis. (b) Mean and maximum intensity values measured from the ROI at each flow rate.

Fig. 3(a) shows representative fluorescence images obtained at flow rates of 0, 1000, 2000, and 3000 mL/min. The closed yellow curve in each image indicates the ROI, which was consistently defined along the projected circumference of the phosphor. By maintaining a constant ROI, the analysis results were prevented from being influenced by focus shifts or differences in the observation field. The images were acquired in an 8-bit format, and the maximum pixel value was adjusted so as not to exceed 255 to prevent fluorescence signal saturation.

Fig. 3(b) presents the quantitative results obtained from the ROI specified in Fig. 3(a). The ROI area was fixed at 30,525 pixels, and as the flow rate increased, the mean fluorescence intensity (a.u.) decreased as 80.196 → 75.494 → 67.004 → 56.702, while the maximum intensity (a.u.) gradually reduced as 138 → 135 → 131 → 119. At 3000 mL/min, the mean and maximum intensities were reduced by 29.3% and 13.8%, respectively, compared with the initial condition of 0 mL/min. This trend was attributed to the slight tilting of the pillar top at higher flow velocities, which caused a portion of the emitted light to deviate from the detection axis. This implies that flow-induced changes in the emission path directly influence the detection intensity, demonstrating that micro-deformation of the pillar can be detected via fluorescence signal variations.

Fig. 4 shows a comparison of the fluorescence response characteristics with respect to the outer diameters of the pillars. All experiments were performed under identical optical conditions (LED: 24 V, exposure: 70 ms, gain: 36, gamma: 0.8, and frame rate: 10 fps), and the mean intensity within the ROI was calculated for each condition.

Fig. 4.

Comparison of fluorescence response according to pillar diameter (Ø2.8, Ø3.8, and Ø4.8 mm). (a) Fluorescence intensity variation under frontal flow conditions. (b) Fluorescence intensity variation under lateral flow conditions.

Fig. 4(a) shows the results under frontal flow conditions. As the flow rate increased, the fluorescence intensities of all three pillars gradually decreased. In particular, the Ø2.8 pillar showed the largest attenuation, with a decrease of approximately 30% at 3000 mL/min compared with the initial state. This behavior indicates that smaller-diameter pillars experience larger deformation under the same flow, leading to a greater deviation in the emitted light from the detection axis owing to tilting of the phosphor. In contrast, the Ø4.8 pillar with relatively higher stiffness exhibited smaller variations in fluorescence intensity.

Fig. 4(b) shows the results under lateral flow conditions. Despite the identical experimental setup and flow rates to those used for frontal flow, the degree of fluorescence attenuation was even greater. At a flow rate of 3000 mL/min, the mean intensity decreased to approximately 45 a.u., indicating that a stronger fluidic force was transmitted during lateral flow according to the different internal flow distributions and relative distances to the walls. Consequently, greater pillar deformation occurred, leading to further deviation of the emission path and a larger decrease in the fluorescence intensity.

These results confirm that smaller pillar diameters yield higher flow-detection sensitivity and that the fluorescence intensity consistently decreases, regardless of changes in flow direction. Therefore, the pillar diameter was identified as a key design parameter that directly affects the sensitivity of fluorescence-based flow detection.