*2.3. Fabrication of Paper-Based Colorimetric Device for Fe2*<sup>+</sup> *and Cu2*<sup>+</sup>

#### 2.3.1. Reagents

Iron chloride tetrahydrate (FeCl2·4H2O), HCl, copper sulfate pentahydrate (CuSO4·5H2O), potassium ferricyanide (K3[Fe(CN)6]), and potassium iodide (KI) were purchased from Merck.

2.3.2. Iron and Copper Calibration Curve Standard Solutions Preparation

FeCl2·4H2O was dissolved in HCl (0.5 M) in order to achieve 1000 <sup>μ</sup>g/mL of Fe2<sup>+</sup> standard stock solution whereas CuSO4·5H2O was used to prepare 1000 <sup>μ</sup>g/mL Cu2<sup>+</sup> standard stock solution in Milli-Q water. The series of four standard solutions (25, 50, 100 and 200 μg/mL) of Fe2<sup>+</sup> and Cu2<sup>+</sup> were prepared by diluting the standard stock solutions with different volumes of Milli-Q water. After these steps, K3[Fe(CN)6] (5 mM) and KI (0.4 M) solutions were prepared for Fe<sup>2</sup><sup>+</sup> and Cu2<sup>+</sup> detection, respectively.

#### 2.3.3. Fabrication of the Paper Analytical Device (PAD)

The fabrication of PAD was developed as follows:


### 2.3.4. Assay Procedure

A small volume (5 μL) of Fe2<sup>+</sup> and Cu2<sup>+</sup> assay reagents ((K3[Fe(CN)6]) and KI) was spotted by drop-casting on paper circular dots using a micropipette and allowed to dry at the room temperature for 3 h. Five microliters of each standard solution was added to the corresponding labeled spots of the PAD. The Fe2<sup>+</sup> of the standard solutions reacted with the K3[Fe(CN)6] generating blue colored complex in the detection zones. Instead, the Cu2+, reacting with KI, produced a red-brown compound. The intensity of the color was proportional to the standard solution concentration. A schematic representation of the described process was represented in Figure 2.

**Figure 2. Schematic Fe2**<sup>+</sup> **and Cu2**<sup>+</sup> **colorimetric assay procedure**: (**a**) The four spots were achieved by wax pen in order to create hydrophobic barriers after heating using a hot plate. (**b**,**c**) Five microliters of each standard solution were added by drop-casting to the corresponding labeled spot. (**d**) On the Fe(II) paper-based analytical device (PAD), a blue complex was formed after the reaction between the Fe2<sup>+</sup> and ((K3[Fe(CN)6]); the blue color intensity directly correlated with the Fe<sup>2</sup><sup>+</sup> concentration (**e**) On the Cu(II) PAD a red-brown compound was developed, generating by Cu2<sup>+</sup> and KI reaction, whose color intensity was dependent on Cu2<sup>+</sup> concentration.

#### *2.4. Quantitative Image Processing by ImageJ 1.47 Software*

Once the color changes were achieved due to the chemical interaction with the different hazardous compounds, the corresponding PADs images were captured using scanner Samsung SCX-3400 with a resolution of 300 dpi. Then, the images were stored in JPEG format and analyzed in RGB format with the open-source software, ImageJ [46]. An adjustment of the color threshold was applied to each image to filter out all colors that were not correlated to the colored complex to be detected during the analysis. For instance, the Fe2<sup>+</sup> color adjustment was applied to delete all colors which was not in the blue range from the analysis spectrum. The color adjustment was set as follows:


3. The hue was adjusted by moving the sliders directly below the "Hue" spectrum until only the color of interest was visible. The hue threshold ranges set for each metal were fixed as follows: NH3 (244–255), C2H4O (38–240), Fe2<sup>+</sup> (171–197), Cu2<sup>+</sup> (37–255).

The images were then converted to an 8-bit grayscale ("Image" → "Type" → "8-bit") and inverted ("Edit" → "Invert"). The intensity measurements yielded a positive slope when plotted versus metal amounts. Mean Gray Value (MGV) was measured for each RGB channel (red, blue and green, "Image" → "Color" → "Merge Channel") by first selecting "mean gray value" and "limit to threshold" in the "Set measurements window," found from the ImageJ menu by selecting "Analyze" → "Set measurements". Each area was selected using the wand tool, which automatically found the edge of an object and traced its shape. The gray intensity of the outlined area was measured by selecting "Analyze" → "Measure." Then, the RGB channel was selected with the highest sensitivity for the metal detection according to Yu et al. [47]. The blue channel was selected for both metal cations¸ the red channel for NH3 and green channel for C2H4O were selected. Data were then imported into Microsoft Excel 2019 in order to obtain the different calibration curves for the NH3, C2H4O, Fe2<sup>+</sup>, Cu2<sup>+</sup> concentrations.

The colorimetric detection limits of NH3, C2H4O, Fe<sup>2</sup><sup>+</sup> and Cu2<sup>+</sup> were estimated based on 3SB/S according to IUPAC rules, where SB and S are standard deviation and slope, respectively [48,49].

#### *2.5. Interference Studies*

The selectivity of PAD to Cu2<sup>+</sup> and Fe2<sup>+</sup> was evaluated by interferences assessment exposing the functionalized PAD to several metal ions solutions containing Na+, K<sup>+</sup>, Mg2<sup>+</sup>, Ca2<sup>+</sup>, Al3<sup>+</sup>, Mn2<sup>+</sup>, Fe3+, Co2+, Ni2+, Zn2+, Cd2<sup>+</sup> and Pb2<sup>+</sup> at a concentration of 100 μg/mL. The same procedure was used to assess the specificity of PAD to NH3 and C2H4O using methylamine, ethylamine, triethylamine, benzene, toluene, ethyl benzene, formaldehyde and ethanol at a concentration of 500 ppm.

#### **3. Results and Discussion**

In recent years, the environmental pollution has been at the center of many debates, due to the progressive and intense industrialization; the scientific community has thus focused its attention on the potentially toxic effects of certain substances on the living organisms [50]. Several people are exposed to different kinds of substances owing to the contamination of several environments in particular water, atmosphere and soil [51]. Among these, the most widespread are certainly the transition metals, heavy metals and gaseous substances, that are produced by intense processing activities especially in the agrifood sector [52]. These chemicals are generally released into the atmosphere and they can reach the groundwater as well as lakes and sea reaching living organisms with subsequent collateral effects [3,53]. In this scenario, environmental monitoring is a fundamental objective to prevent and to know at what doses an organism was exposed. The conventional analytical techniques (gas chromatography–mass spectrometry, high-performance liquid chromatography–mass spectrometry, atomic absorption spectroscopy) are sophisticated systems that require high energy consumption and expensive laboratory systems. Paradoxically, in fact, the environment analysis by the use of these instruments induces in turn pollution (energy, consumables, toxic reagents). Starting from these assumptions, we have developed a PAD that can be used without the need for trained operators to monitor some hazardous materials such as NH3, C2H4O, Fe2<sup>+</sup> and Cu2+. For gaseous substances, namely NH3 and C2H4O, we performed two different techniques to functionalize the filter paper. In particular, for NH3 detection, we used a PANI film functionalization following the polymerization of aniline directly on paper substrate. The PANI film obtained was in the form of green emeraldine salt due to the protonation of the backbone induced by HCl. We selected four doses of NH3 on the basis of toxicological results obtained in literature, as reported in the Introduction section (Section 1). When NH3 molecules reached the functionalized paper, the deprotonation of PANI chains and, consequently, the transformation of them into a blue emeraldine base occurred. In addition, this dye shows peculiar chemical properties consisting of the reversible doping/dedoping nature. The dye reacted with the

NH3 determining the color change; when the analyte was removed, it can be reverted to its initial chemical state. Due to the reversible nature of the process, the functionalized PAD can be reused many times (ca. 30 times) before its discard. After the exposure to different concentrations of NH3 vapor (100, 300, 500, 1000 ppm) the color appeared in a few minutes. Immediately, a digital scanner was used to freeze the specific color. The scanner acquired the image in JPEG format, allowing the next analysis by ImageJ software. As shown in Figure 3, the paper assumed a specific coloration that can be visualized with the naked eye. The color switch from light green to blue at the higher concentration tested. By The JPEG images were analyzed after setting the specific parameters (Hue adjustment section of the Threshold Color window) described in detail in the Materials and Methods section (Section 2). The assay reproducibility was evaluated for three identical test zones.

**Figure 3.** Hue adjustment section of the Threshold Color window in ImageJ analysis software of the NH3 PAD.

The functionalization of PAD for the detection of C2H4O was achieved by the use of methyl red. The latter is determined by the concentration of acidic (red) and basic (yellow) forms. The colorimetric sensor was designed to show a selective response based on a chemical reaction, such as the nucleophile addition. Using an excess of hydroxide ions, the C2H4O underwent the nucleophile addition reaction, resulting in the sensor alkalinity changes and consequently in a color change, from yellow to red. The color change was almost instantaneous and it was stable for several days after drying. After the color response, the scanner was used to acquire the image and color intensity. The latter was analyzed for the second step of the experimental session using ImageJ analysis by the Hue adjustment section of the Threshold Color (Figure 4). The reproducibility was evaluated for three identical test zones.

**Figure 4.** Hue adjustment section of the Threshold Color window in ImageJ of C2H4O PAD.

The test zones were used to create the calibration curve. Figure 5 shows the calibration curve for NH3 detection using the color change after the exposure to the four concentrations. In detail, in Figure 5a we reported the pieces of devices related to the functionalized and unexposed PAD (top circle) and the PAD exposed to pure NH3 (28%, bottom circle) with the relative MGV values extracted from the ImageJ analysis that were 10.3 ± 1.5 and 75 ± 4.5, respectively. In Figure 5b, the pieces of PAD after exposure to 100, 300, 500 and 1000 ppm of NH3 were represented. Observing the pictures, it was possible to visualize a color trend with the naked eye, from the lightest to the darkest as the concentration increased. The successive ImageJ analysis performed on the scanner acquisitions correlated with the concentration with a specific MGV obtaining a calibration curve with R2 = 0.99. The limit of detection (LOD) value was 7.64 ppm. The values were obtained by repeating the experiment three times. In order to understand if the device actually worked even with intermediate concentrations, we exposed the PAD to average concentrations calculated between the first and second (200 ppm) and third and fourth doses (750 ppm). Additionally, in this case, PANI film was able to efficiently induce the color response; it was possible to find the concentration simply by interpolating the MGV data on the straight line as shown in Figure 5c.

The same procedure was applied to the paper functionalized with methyl red, capable of detecting the C2H4O vapor. (Figure 6). Figure 6a shows the as-prepared paper device (yellow) and after exposure to pure C2H4O (≥99.5%, dark red) with the corresponding MGV values that were 20.3 ± 2.7 and 155 ± 7.5, respectively. In Figure 6b, the progression from yellow to red was observed when the tested concentrations increased. We used 100, 300, 500 and 1000 ppm concentrations and we built the calibration curve obtaining an R2 value of 0.98 (Figure 6c). The LOD value was 11.09 ppm. The effectiveness of this colorimetric response was verified using two average concentrations: 200 and 750 ppm. Additionally, in this case, the MGV values were interpolated with the curve that exactly corresponded to the tested concentrations.

**Figure 5.** (**a**) Mean Gray Value (MGV) values of PAD as prepared and after the exposure to pure NH3 (28%). (**b**) color change after NH3 exposure. (**c**) Interpolation of NH3 intermediate values (200 and 750 ppm). Data reported were the average of three independent experiments ± SD. The difference between the as-prepared paper and colored papers was considered statistically significant performing a Student's *t*-test with *p* < 0.05 (<0.05 \*).

**Figure 6.** (**a**) MGV values of PAD as-prepared and after the exposure to pure C2H4O (>99%). (**b**) color change after C2H4O exposure. (**c**) Interpolation of C2H4O intermediate values (200 and 750 ppm). Data reported were the average of three independent experiments ± SD. The difference between the as-prepared paper and colored papers was considered statistically significant performing a Student's *t*-test with *p* < 0.05 (<0.05 \*).

After the analysis of gaseous molecules, we used the PAD to detect Cu2<sup>+</sup> and Fe2+, which are the most common metals released in the environment [1]. Then, we moved to the detection of these metals in water at low concentrations. Firstly, we designed four circle spots using a wax pen in order to achieve hydrophobic barriers without the use of wax printing, inkjet printing and screen-printing technologies. Once the heat produced by the hot plate allowed the penetration of the wax into the cellulose porous, the specific chemical analytes, K3[Fe(CN)6] for Fe<sup>2</sup><sup>+</sup> and KI for Cu2<sup>+</sup>, were deposited in the spot's center by drop-casting. The wax channels prevented the typical diffusion phenomenon of the liquid substances deposited on the paper. Five microliters of FeCl2.4H2O and CuSO4.5H2O (25, 50, 100, 200 μg/mL) were used. Chelation (1) and redox (2) chemical reactions produced a blue and brown color, respectively. The two reactions were the following:

(1) Fe2<sup>+</sup> <sup>+</sup> Fe(CN)3−<sup>6</sup> <sup>→</sup> Fe3[Fe(CN)6] 2;

(2) Cu2<sup>+</sup> <sup>+</sup> 2I<sup>−</sup> <sup>→</sup>CuI2 <sup>→</sup> 2CuI2 <sup>=</sup> 2CuI <sup>+</sup> I2.

After color formation, we acquired the images by a digital scanner to perform ImageJ analysis using the adjustment of Threshold Color both for Fe2<sup>+</sup> (Figure 7) and Cu2<sup>+</sup> (Figure 8). The analysis was repeated in three identical test zones.

**Figure 7.** Top: Image acquisition of PAD after FeCl2.4H2O deposition at different concentrations. Down: threshold analysis, saturation and brightness adjustment by ImageJ software.

**Figure 8.** Top: Image acquisition of PAD after CuSO4.5H2O deposition at different concentrations. Down: threshold analysis, saturation and brightness adjustment by ImageJ software.

As shown in Figure 9a, the color changed from light blue to dark blue, proportionally to the Fe2<sup>+</sup> concentration increase. The corresponding calibration curve was obtained plotting the MGV values analyzed by ImageJ analysis after standards solution deposition, showing an R2 value of 0.98 (Figure 9b).

**Figure 9.** (**a**) Color change of filter paper after exposure to Fe(II) at different concentrations. (**b**) The yellow rhombuses represented the interpolation of Fe2<sup>+</sup> intermediate concentrations (37 and 150 μg/mL). Data reported were the average of three independent experiments ± SD. The difference between as-prepared paper and colored papers was considered statistically significant performing a Student's *t*-test with *p* < 0.05 (<0.05 \*).

A similar R2 value was reported for the Cu2<sup>+</sup> calibration curve; in the latter case, the color changed from light brown to dark brown (Figure 10a). As demonstrated for NH3 and C2H4O we used two average concentrations between 25 and 50 μg/mL and between 100 and 200 μg/mL (37 and 150 μg/mL) to test the device reliability. The MGV values acquisitions revealed that the corresponding concentrations were on the calibration curve thus confirming the effectiveness and stability of the PAD (Figure 10b). The LOD for Fe2<sup>+</sup> was 3.8 μg/mL and 3.2 μg/mL for Cu2<sup>+</sup>. For both metals, the values were greatly below the maximum acceptable concentrations in drinking water stipulated by the World Health Organization (WHO) [54].

The LOD values of each device are summarized in Table 1.

**PADs Concentrations Range Limit of Detection (LOD)** NH3 100–1000 ppm 7.64 ppm C2H4O 100–1000 ppm 11.08 ppm Fe2<sup>+</sup> 25–200 μg/mL 3.8 μg/mL Cu2<sup>+</sup> 25–200 μg/mL 3.2 μg/mL

**Table 1.** LOD values of NH3, C2H4O, Fe2<sup>+</sup> and Cu2<sup>+</sup> PADs.

**Figure 10.** (**a**) Color change of filter paper after exposure to Cu(II) at different concentrations. (**b**) The yellow rhombuses represented the interpolation of Cu2<sup>+</sup> intermediate concentrations (37 and 150 μg/mL). Data reported were the average of three independent experiments ± SD. The difference between as-prepared paper and colored papers was considered statistically significant performing a Student's *t*-test with *p* < 0.05 (<0.05 \*).

In order to test the selectivity of the different PADs used in this study, several metal and gaseous solutions at 100 μg/mL and 100 ppm, respectively, were used. No significant visual color change had been observed in all the tested cases. For gaseous molecules, the PAD was exposed to methylamine, ethylamine, triethylamine, benzene, toluene, ethyl benzene, formaldehyde and ethanol at 100 ppm for ca. 15 min. Any noticeable effects on filter paper were recorded. This suggested the high selectivity of PAD to the NH3 and C2H4O only (Figure 11a,b). Similar results were obtained analyzing the interferences of different metal ions after PAD exposure for 15 min. It was observed that 100 μg/mL of Na+, K<sup>+</sup>, Mg2<sup>+</sup>, Ca2<sup>+</sup>, Al3<sup>+</sup>, Mn2<sup>+</sup>, Fe3<sup>+</sup>, Co2<sup>+</sup>, Ni2<sup>+</sup>, Zn2<sup>+</sup>, Cd2<sup>+</sup>, and Pb2<sup>+</sup> highlighted negligible colorimetric effects on the PAD due to the small affinity with the analytes deposited on filter paper (Figure 11c,d).

**Figure 11.** Interferences assay for NH3 and C2H4O (**a**,**b**) and Fe(II) and Cu (II) (**c**,**d**). The values were expressed as MGV. Data reported were the average of three independent experiments ± SD.
