A simple scanner-based colorimetric method for fluoride determination using Alfusone reagent Authors: Masahide Hagiri, Shion Nomura, Erika Fukasawa, Takeshi Kato, Atsushi Manaka Published: 2026 ABSTRACT ObjectiveThe quantitative determination of fluoride ions is important for water quality monitoring and public health. This study aimed to develop a simple and simultaneous multi-sample digital colorimetric method for fluoride determination using a 96-well plate and a flatbed scanner, without requiring conventional analytical instruments.ResultsFluoride solutions were color-developed using Alfusone reagent in a 96-well plate and scanned with a flatbed scanner. Color information was extracted from the scanned images and analyzed using red-green-blue (RGB), cyan-magenta-yellow-key (CMYK), hue-saturation-value (HSV), CIE XYZ, and CIELAB parameters. Among the evaluated parameters, the HSV H parameter showed the lowest limit of detection and limit of quantification of 0.086 and 0.29 mg/L, respectively. The CIELAB b* parameter showed the lowest values among the evaluated CIELAB parameters, with a limit of detection and limit of quantification of 0.19 and 0.64 mg/L, respectively. These values were below the World Health Organization guideline value for fluoride and the Japanese drinking water quality standard. Spike-and-recovery tests using bottled and tap water showed that the R and a* parameters provided recoveries closer to 100% than b*. The method offers a simple, attachment-free approach that enables simultaneous analysis of multiple samples and calibration standards using commercially available reagents and office automation equipment.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s13104-026-07926-z. Introduction Fluoride ion (F−) is an important analyte in drinking and environmental waters because its intake is closely related to dental health, whereas excessive exposure can cause adverse effects such as dental fluorosis [1–3]. Reliable monitoring of F− concentration is therefore required for water quality control. Conventional methods, including ion-selective electrodes and spectrophotometric methods based on metal–dye complexes such as lanthanum–alizarin complexone systems, are well established, but they generally require dedicated instruments and manual analytical procedures [4, 5]. Digital colorimetric analysis has recently attracted attention as a simple alternative for chemical sensing. In F− analysis, smartphone-based methods have been reported in which color information such as RGB values or hue is extracted from captured images and correlated with F− concentration [6–11]. However, image-based quantification using smartphones may be affected by ambient light, camera settings, automatic image processing, and device-to-device variation. Optical attachments or dedicated holders can improve reproducibility, but they add fabrication steps and may depend on the smartphone model [10, 11]. In addition, when multi-well plates are photographed with an ordinary camera, geometric factors such as viewing angle and non-uniform imaging conditions may introduce positional errors across the image field. Although flatbed scanner-based colorimetric analysis using multiwell plates has been reported for fluoride determination, such studies remain limited and have been reported primarily by a single research group [12]. In this study, we developed a digital colorimetric method for F− determination using a flatbed scanner and a 96-well plate. The scanner-based setup provides controlled imaging conditions without requiring a smartphone attachment, while the 96-well format enables simultaneous measurement of multiple samples and calibration standards in a single scan, thereby reducing the time required for repetitive individual measurements [13–15]. The method uses a commercially available Alfusone reagent, a powdered reagent containing a pre-formed lanthanum–alizarin complexone (ALC–La) system and buffer components, which simplifies reagent preparation to dissolution and mixing [16, 17]. The molecular structure of alizarin complexone is shown in Scheme S1. The color response of the ALC–La system to F− has generally been attributed to the formation of ALC–La–F–related colored species, although the detailed solution chemistry and color-development mechanism remain subject to discussion [18, 19]. Furthermore, the relationships between F− concentration and multiple color parameters, including RGB, CMYK, XYZ, HSV, and CIELAB values, were systematically examined to identify suitable descriptors for quantitative analysis. RGB directly represents red, green, and blue image intensities, whereas HSV and CIELAB describe color changes in terms of hue, saturation, brightness, and perceptual color coordinates [20–22]. Thus, these transformed color spaces may capture color changes that are not readily represented by individual RGB components. Thus, the present method aims to combine simple reagent preparation, high-throughput plate-based operation, and reproducible digital color acquisition for practical F− determination in water samples. The specific value of this protocol lies in its combination of a ready-to-use commercial reagent, attachment-free scanner imaging, simultaneous plate-based operation, and systematic comparison of RGB, CMYK, HSV, XYZ, and CIELAB parameters. Main text Materials and methods Materials A fluoride ion standard solution (1000 mg/L; Kishida Chemical Co., Ltd.), Alfusone reagent (Dojindo Laboratories), and acetone (analytical grade; FUJIFILM Wako Pure Chemical Corp.) were used as received. SPADNS (1,8-dihydroxy-2-(4-sulfophenylazo)naphthalene-3,6-disulfonic acid trisodium salt; Sigma-Aldrich Co. LLC), zirconyl chloride octahydrate (ZrOCl₂·8H₂O; FUJIFILM Wako Pure Chemical Corp.), and hydrochloric acid (FUJIFILM Wako Pure Chemical Corp.) were used for the SPADNS method. The Alfusone reagent is a commercially available powder composed of a pre-formed ALC–La complex and buffer components, designed to provide optimal conditions for F− determination upon simple dissolution in water. Distilled water was used throughout all experiments. For spike-and-recovery tests, commercially available bottled waters and tap water were used as real samples. Soft water was I-LOHAS (Coca-Cola Company, hardness: 27 mg/L), and hard water was Evian (Ito En, Ltd., hardness: 304 mg/L). Tap water was collected in Numata City, Gunma Prefecture, Japan, and supplied from the Numata municipal water treatment plant (hardness: 29 mg/L). Color development by the Alfusone method and absorbance measurement Color development based on the Alfusone method was carried out to evaluate the response to F− ions. In each well of a 96-well plate, 150 µL of F− standard solution, 30 µL of 5% Alfusone solution, and 120 µL of acetone were added sequentially. The F− standard solutions were prepared in the concentration range of 0–1.6 mg/L, and eight wells were used for each concentration. After mixing, the color change corresponding to the F− concentration was visually confirmed after 30 min of reaction. For comparison with the scanner-based analysis, absorbance at 620 nm was measured using a microplate reader (SH-1300Lab, Corona Electric Co., Ltd.) under the same conditions. The absorbance data were used as reference values to evaluate the validity of the image-based quantification. Scanner-based colorimetric analysis and calibration The 96-well plates prepared as described above were imaged using a flatbed scanner (DocuCentre-VII C5573, FUJIFILM Business Innovation Corp.) 30 min after mixing. Scanning was performed at 600 dpi in full-color mode. The acquired images were analyzed using ImageJ (version 1.54g). For each well, only the liquid region inside the well was selected as the region of interest. A circular area with identical position and area was defined within each well to ensure consistent analysis. The RGB values were extracted from each region of interest and further converted into CMYK, HSV, XYZ, and CIELAB color parameters. In this analysis, RGB values were expressed on a scale from 0 to 255; CMYK, S, and V values on a 0–1 scale; H values in degrees from 0 to 360; XYZ values as normalized tristimulus values; and CIELAB values as L*, a*, and b* coordinates, with a* and b* treated on an output scale from − 128 to + 127 in the present analysis. Calibration curves were constructed by plotting each color parameter obtained from the blank and F− standard solutions against the F− concentration. For each color parameter, linear regression was performed over the concentration range used for quantification, and the slope of the calibration curve was used to evaluate analytical sensitivity. The limit of detection (LOD) and limit of quantification (LOQ) were determined using the 3σ and 10σ criteria, respectively. The standard deviation (σ) was calculated independently for each color parameter from eight replicate signal values obtained for the blank solution (0 mg/L F−, n = 8). For each color parameter, the LOD and LOQ were calculated as 3σ/S and 10σ/S, respectively, where S is the slope of the corresponding calibration curve. Spike-and-recovery test To evaluate the applicability of the method to real water samples, spike-and-recovery tests were performed using the samples described in the Materials section. A F− standard solution was added to each sample to obtain a final F− concentration of 0.8 mg/L. Specifically, 8 mL of a 10 mg/L F− standard solution was added, and the total volume was adjusted to 100 mL with each water sample. The prepared samples were analyzed using the same procedure described above, and the F− concentrations were determined from the calibration curves. The unspiked water samples were also analyzed to determine the initial F− concentration in each sample. The initial F− concentration in each original water sample and the F− concentration after spiking were also confirmed using the Alfusone-based absorbance method at 620 nm. The recovery was calculated as Cdetected / (Cadded + Cinitial) × 100, where Cdetected is the F− concentration obtained from the colorimetric analysis for the spiked sample, Cinitial is the F− concentration determined for the corresponding unspiked sample, and Cadded is the added F− concentration. After the spike-and-recovery evaluation, a representative soft bottled water sample spiked with 0.8 mg/L F− was additionally analyzed using the SPADNS method as a limited cross-check with an independent colorimetric method. The SPADNS reagent was prepared by mixing equal volumes of SPADNS solution and zirconyl chloride/hydrochloric acid solution, and absorbance was measured at 570 nm after 3 min of reaction. Results and discussion Color development by the Alfusone method and absorbance measurement The solutions were dispensed into an 8 × 8 array of wells as described above, and color development of F− solutions was carried out using the Alfusone method. In the absence of F−, the solution exhibited a reddish-purple color. With increasing F− concentration, the color gradually shifted toward a bluish-purple. A representative scanned image of the 96-well plate showing this concentration-dependent color change is provided in the Supplementary Information (Fig. S1). As discussed in detail below, this color change was reflected in the RGB values, where the R component was relatively dominant at low F− concentrations, while the B component increased at higher concentrations. The systematic variation in color with F− concentration indicated the feasibility of quantitative analysis based on colorimetric parameters. For validation, the absorbance at 620 nm was measured using a microplate reader. The calibration curve showed good linearity, with a slope of 0.316, an intercept of 0.447, and a coefficient of determination (R2) of 0.997 (Fig. S2). Based on the standard deviation of the blank signal at 0 mg/L F− (σ = 0.00705), the LOD and LOQ were calculated to be 0.067 and 0.22 mg/L, respectively. These results confirm the reliability of the Alfusone color reaction for quantitative F− analysis under the present conditions. Scanner-based colorimetric analysis and calibration The same plate was scanned using a flatbed scanner, and the color information of each well was extracted from the scanned image. The RGB values obtained from the defined circular regions within the wells were used directly or converted into other color parameters, including CMYK, HSV, XYZ, and CIELAB values. Calibration curves were then constructed by plotting each color parameter against the F− concentration. The resulting calibration parameters are summarized in Table 1. Among the RGB parameters, the R value showed a negative correlation with F− concentration, consistent with the visual decrease in the reddish-purple component. In contrast, the G and B values showed poorer linearity. Several converted color parameters also exhibited linearity, particularly the H, a*, and b* values, indicating that scanner-derived color information can be used for semiquantitative or quantitative F− analysis. The LOD and LOQ values were calculated from the slope of each calibration curve and the standard deviation of the blank signal. Table 1Calibration parameters (slope, intercept, and R2), standard deviation of blank signal, and detection limits (LOD and LOQ) for F− determination based on various color parametersColor parameterSlopeIntercept R 2 Standard deviation (σ)LOD [mg/L]LOQ [mg/L]R-16.760.20.9833.580.642.1G2.0314.00.4331.65-†-B1.8642.20.5642.57--C0.130-0.02700.7940--M-0.1030.7650.7940.01620.471.6Ye‡-0.2130.2880.8980.01360.190.64K0.04230.7690.9150.01401.03.3H-30.53260.9830.8770.0860.29S-0.1160.7660.8870.01620.421.4V-0.04750.2320.7810.01400.893.0X-0.007300.02430.9350.002911.24.0Y-0.003350.01440.8590.001791.65.3Z0.001570.02350.4450.00273--L*-2.4512.20.8711.101.44.5a*-6.7625.90.9770.9900.441.5b*-5.41-7.070.9800.3450.190.64† “–” indicates that LOD and LOQ were not calculated because the corresponding calibration curve was not suitable for quantitative analysis‡ “Ye” denotes the yellow component in the CMYK color space, used to distinguish it from the Y value in the XYZ color space Table 1Calibration parameters (slope, intercept, and R2), standard deviation of blank signal, and detection limits (LOD and LOQ) for F− determination based on various color parametersColor parameterSlopeIntercept R 2 Standard deviation (σ)LOD [mg/L]LOQ [mg/L]R-16.760.20.9833.580.642.1G2.0314.00.4331.65-†-B1.8642.20.5642.57--C0.130-0.02700.7940--M-0.1030.7650.7940.01620.471.6Ye‡-0.2130.2880.8980.01360.190.64K0.04230.7690.9150.01401.03.3H-30.53260.9830.8770.0860.29S-0.1160.7660.8870.01620.421.4V-0.04750.2320.7810.01400.893.0X-0.007300.02430.9350.002911.24.0Y-0.003350.01440.8590.001791.65.3Z0.001570.02350.4450.00273--L*-2.4512.20.8711.101.44.5a*-6.7625.90.9770.9900.441.5b*-5.41-7.070.9800.3450.190.64† “–” indicates that LOD and LOQ were not calculated because the corresponding calibration curve was not suitable for quantitative analysis‡ “Ye” denotes the yellow component in the CMYK color space, used to distinguish it from the Y value in the XYZ color space Calibration parameters (slope, intercept, and R2), standard deviation of blank signal, and detection limits (LOD and LOQ) for F− determination based on various color parameters † “–” indicates that LOD and LOQ were not calculated because the corresponding calibration curve was not suitable for quantitative analysis ‡ “Ye” denotes the yellow component in the CMYK color space, used to distinguish it from the Y value in the XYZ color space Among the evaluated parameters, the R, H, a*, and b* values showed particularly low LOD and LOQ values (Table 1), suggesting that transformed color-space parameters may provide higher analytical sensitivity than the original RGB values. Representative calibration curves obtained from the scanner-based colorimetric analysis are shown in Fig. 1. The residual plots for representative calibration models are provided in the Supplementary Information (Fig. S3) and showed no obvious systematic deviation from linearity within the tested concentration range. Furthermore, additional calibration curves for the H and a* parameters are provided in the Supplementary Information (Fig. S4). Fig. 1Representative calibration curves for scanner-based fluoride determination using (a) R values and (b) b* values extracted from scanned images of the 96-well plate. Error bars indicate the standard deviation (n = 8), and solid lines represent linear regression fits Representative calibration curves for scanner-based fluoride determination using (a) R values and (b) b* values extracted from scanned images of the 96-well plate. Error bars indicate the standard deviation (n = 8), and solid lines represent linear regression fits The R value (Fig. 1a) decreased linearly with increasing F− concentration, reflecting the gradual decrease in the reddish component of the solution color. Similarly, the b* value (Fig. 1b), which represents the yellow–blue axis in the CIELAB color space, also showed a strong negative correlation with F− concentration, reflecting the gradual shift of the solution color toward a bluish-purple tone. Notably, the H parameter exhibited the lowest LOD and LOQ values, which were calculated to be 0.086 and 0.29 mg/L, respectively. The b* parameter also showed low LOD and LOQ values (0.19 and 0.64 mg/L, respectively). The LOQ values obtained for both H and b* were lower than the WHO guideline value for F− in drinking water (1.5 mg/L) as well as the Japanese drinking water quality standard (0.8 mg/L), suggesting that the present scanner-based colorimetric method may be applicable to the screening and quantitative evaluation of F− concentrations relevant to drinking-water quality management. In contrast, the R value showed higher LOD and LOQ values (0.64 and 2.1 mg/L, respectively). Although the H parameter exhibited the highest analytical sensitivity, the H calibration plot showed slight deviations from ideal linear behavior (Fig. S4), whereas the b* parameter provided a comparatively simple concentration response. In addition, a multiple linear regression model based on color parameters was evaluated for comparison and is presented in the Supplementary Information (Fig. S5). Although several digital colorimetric F− assays have reported lower LOD values, they often require dedicated optical attachments, controlled illumination, or specially prepared sensing materials. For example, SPADNS- or nanoparticle-based smartphone methods have achieved LOD values of approximately 0.06 mg/L, while Wen et al. reported an LOD of 0.08 mg/L using the ALC–La reaction; however, these methods used smartphone attachments or controlled optical setups [7, 10, 11]. Saranchina et al. achieved an LOD of 0.03 mg/L using a poly(methacrylate) optode containing an immobilized ALC–Zr complex [6]. In contrast, the present method uses only commercially available reagents, a standard 96-well plate, and a flatbed scanner. Thus, although its sensitivity is not the highest among reported methods, the method offers practical advantages in simplicity, accessibility, and simultaneous multi-sample analysis. The use of CIELAB b* as a quantitative descriptor also differentiates this approach from conventional RGB- or hue-based analyses [6, 7]. To examine scanner-to-scanner reproducibility, calibration curves based on the R parameter were additionally obtained using two other commercially available flatbed scanners. As shown in Fig. S6 in the Supporting Information, the absolute Red values differed among devices, whereas the overall concentration dependence was preserved. These results indicate that device-specific calibration is important when applying the method across different scanners. Based on these results, the b* value was selected as the principal color parameter for subsequent evaluation of the method using real water samples. Spike-and-recovery tests were therefore performed to assess the practical applicability of the scanner-based method. Spike-and-recovery test Spike-and-recovery tests were performed using commercially available bottled waters and tap water to evaluate the applicability of the scanner-based method to real samples. The recovery results obtained using selected color parameters are summarized in Table 2. Table 2Recoveries of fluoride ion from real water samples determined using selected color parameters (n = 8)Color parameterRecovery / %I-LOHAS, soft bottled waterEvian, hard bottled waterTap waterR10299.193.5H10885.0105a*93.5109106b*13994.1125 Table 2Recoveries of fluoride ion from real water samples determined using selected color parameters (n = 8)Color parameterRecovery / %I-LOHAS, soft bottled waterEvian, hard bottled waterTap waterR10299.193.5H10885.0105a*93.5109106b*13994.1125 Recoveries of fluoride ion from real water samples determined using selected color parameters (n = 8) The R and a* parameters showed recoveries relatively close to 100% for all tested samples, indicating reasonable applicability for practical F− screening. In contrast, larger deviations from 100% were observed for the b* parameter, despite its low LOD and LOQ values. This tendency may be associated with the slight nonlinearity observed in the b*-based calibration curve. Although transformed color-space parameters such as CIELAB b* can improve analytical sensitivity, the relationship between color parameters and analyte concentration is not necessarily strictly linear over the entire concentration range. In addition, highly sensitive color parameters may be more susceptible to matrix effects arising from differences in water composition, such as hardness and dissolved ionic species. Therefore, while the b* parameter is advantageous for sensitive F− detection, the R and a* parameters may provide better robustness for practical screening applications using real water samples. As a limited cross-check with an independent colorimetric method, a representative soft bottled water sample spiked with 0.8 mg/L F− was also analyzed using the SPADNS method [23]. When the SPADNS result was used as the reference value, the relative recoveries obtained by the scanner-based method were 101%, 85.1%, 109%, and 86.9% for the R, H, a*, and b* parameters, respectively. These results indicate that the scanner-based method provided fluoride concentrations generally comparable to those obtained using the SPADNS method. Overall, the present scanner-based approach provides a simple and accessible method for F− screening using commercially available reagents and office automation equipment, while allowing simultaneous analysis of multiple samples and calibration standards in a single scan. Limitations This study has several limitations. First, although the CIELAB b* parameter showed low LOD and LOQ values and relatively good linearity, its recoveries in real water samples deviated more from 100% than those obtained using the R and a* parameters. Therefore, b* may be useful for sensitive screening, but it should not be regarded as the most robust parameter for all sample matrices. Second, the calibration curves were evaluated by linear regression, although some color parameters showed slight nonlinearity over the tested concentration range. This may limit the accuracy of quantification depending on the selected concentration range. Third, the method was validated using only three types of water samples and a limited number of scanner models under fixed scanning conditions. Interference from coexisting ions, colored or turbid matrix components, scanner-to-scanner variation, and broader applicability to more complex environmental samples remain to be examined. Because the proposed scanner-based method relies on the Alfusone color reaction, its chemical selectivity is fundamentally governed by the applicability and limitations of the Alfusone method itself. In addition, the present study evaluated a scanner-based approach using the Alfusone color reaction and compared it with conventional absorbance measurements based on the same chemistry. Although a limited cross-check using the SPADNS method was performed for one representative spiked sample, more extensive validation using independent analytical methods and a broader range of real samples would be valuable in future studies. Supplementary Information Supplementary Material 1. Supplementary Material 1.