Dual-Channel Ratiometric Colorimetric Sensor Array for Quantification and Discrimination of o-, m-, and p-Phenols

The determination of trace o-, m-, and p-phenols in food samples was of great significance, but it is still very challenging due to the low concertation and complex matrix. In comparison to traditional sensors based on the “lock-and-key” strategy, the sensor arrays that mimicked the human olfactory system exhibited excellent universality and efficiency in assays of structural analogs. Herein, we described a costless and robust colorimetric sensor array for quantification and discrimination of o-, m-, and p-phenols. The obtained Cu2(OH)3Cl and Ni-doped Cu2(OH)3Cl with peroxidase-like (POD) activity were served as nanozymes to accelerate the oxidation of 3,3′,5,5′-tetramethylbenzidine (TMB) with H2O2. UV–vis absorbance spectroscopy displayed remarkable enhancements both at 652 nm and 450 nm, individually attributable to the formation of single and dual electron-oxidized TMB (TMBox) substances. Then varied phenols regulated the absorbance via specific electron transfer processes. With the two peaks as dual signals, a ratiometric colorimetric sensor array based on nanozymes (i.e., Cu2(OH)3Cl and Ni-doped Cu2(OH)3Cl) was designed to monitor different phenols. The ratio (I652/I450) was recorded to quantify phenols sensitively, where internal reference ruled out interferences from the environment and equipment. Moreover, this array was also explored for simultaneous discrimination of o-, m-, and p-phenols. By integrating with colorimetric fingerprints and principal component analysis (PCA), each kind of phenol can be distinctly discriminated even at a very low concentration. This work demonstrated the reliability of a ratiometric sensor array for the recognition of multiple phenols.


Introduction
As the basic raw materials and intermediates of the organic chemistry industry, phenols were wildly applied in the petrochemical industry, food processing, and medicine. Nevertheless, the discharge of these pollutants into the environment received increasing attention due to their extreme toxicity and low biodegradability (Arfin et al. 2019). Those residual phenols in food contact materials might easily migrate to food as they were in the form of a physical combination with a polymer rather than a covalent bond (Du et al. 2016). Considering the negative effects on nervous urinary and digestive systems in humans (Michałowicz and Duda 2007), it is urgent to develop a simple, accurate, and cost-effective method to quantify and discriminate a series of phenols.
Until now, a series of analytical techniques such as chromatography (Chernonosov et al. 2017), fluorescence spectroscopy (Qi et al. 2018), immunoassay (Karim and Fakhruddin 2012), and electrochemical analysis (Cheng et al. 2017) were developed to detect the phenols with high sensitivity. These methods always required complex pretreatments, expensive equipment, and skilled operators, making them not suitable for rapid analysis. In contrast, colorimetric sensing has drawn special attention because of operational simplicity, easy readout, low cost, and high throughput. Notably, some nanozymes as colorimetric receptors have been explored to sensitively monitor the phenols via amplifying the signals (Tian et al. 2019;Ma et al. 2020). Yang et al. developed CeVO 4 as an oxidase mimic to determine hydroquinone from resorcinol and catechol . Niu et al. used rGO/Cu 8 S 5 /PPy as a POD mimic for the detection of phenol (Jiang et al. 2017). Lee's group prepared coral-like silver citrate to be a Laccase mimic to determine different phenols (Koyappayil et al. 2021). By looking through these works, it was noticed that the existing nanozyme-based sensors were almost based on the changes of absorption intensity at a fixed wavelength, which were easily interfered with by sensor concentration, environments, and instrumental factors . In contrast, a ratiometric sensing strategy could resolve these limits via providing double or multiple signal outputs. This strategy has been previously reported in fluorescence analysis, in which ratiometric fluorescence was used to quantify the targets (Zhang et al. 2019). One of the emissions served as an internal reference to amplify the ratio of signal to noise, realizing sensitive quantification. As far as we know, there were few reports about nanozymebased colorimetry for ratiometric detection.
The structure of phenols exhibited a high degree of heterogeneity due to different numbers and positions of substituted hydroxyl groups on the benzene ring. This caused their drastically different features in physical and chemical properties, resulting in divergent environment hazards and biological toxicity even at a very low concentration (Czaplicka 2004;Xu et al. 2020). The sensor array was considered to be an extremely practical technology for the discrimination of multiple analogs (Ghasemi et al. 2015). The matrix employed sensing receptor elements to generate distinguishable responses against various targets. With the aid of statistical analysis, these analytes were able to be clustered separately from each other. Compared with the "lock-and-key" recognition mold, it was not necessary to design many specific sensors for each target, drastically cutting down the cost and simplifying the assay procedure. In this regard, several colorimetric sensor arrays have been explored to distinguish ions (Najafzadeh et al. 2018), antioxidants , and phenols ), based on their specific influence on the catalytic kinetics of nanozymes.
In this study, a colorimetric sensor array based on Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl as the POD nanozymes has been proposed for quantification and discrimination of o-, m-, and p-phenols. The Cu 2 (OH) 3 Cl exhibited a layered structure composed of edge-sharing HO-Cu-Cl octahedral (Wei et al. 2013;Yang et al. 2016). This tribasic copper salt mainly occurs in three polymorphs, including atacamite (α-Cu 2 (OH) 3 Cl), paratacamite (γ-Cu 2 (OH) 3 Cl), and bollackite (Fleet 1975;Han et al. 2003). Among the polymorphs, γ-Cu 2 (OH) 3 Cl showed superior affinity to H 2 O 2 , which was more in favor of the enhancement of the POD activity (Wang et al. 2021a). In addition, the POD activity of Ni-doped γ-Cu 2 (OH) 3 Cl can be further progressed because of the enhancement in the number of oxygen vacancies.

Fabrication of Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl
The preparation of Cu 2 (OH) 3 Cl was as follows: 1.54 g of CuCl 2 ·2H 2 O was dissolved in 10-mL ultrapure water (defined as solution A), 2.46 g of 2-methylimidazole was dissolved in 12 mL ultrapure water (solution B), then A and B were fast transferred into 33 mL of isopropanol and stirred for 1 h, and the final mixture was transferred to 100 mL of Teflonlined stainless-steel autoclave and kept at 120 ℃ for 8 h. The final product was centrifuged and washed with ethanol several times, collected samples were dried at 40 ℃ for further applications. As the same as Cu 2 (OH) 3 Cl, the Ni-doped Cu 2 (OH) 3 Cl was also synthesized in which the molar ratio of Cu/Ni was fixed at 1:1.

POD Activity of Cu 2 (OH) 3 Cl
Typically, 0.1 mL of H 2 O 2 (100 mM), 0.1 mL of TMB (10 mM), and 0.8 mL of Cu 2 (OH) 3 Cl (0.1 mg/mL) were separately placed into 1 mL of NaAc-HAc buffer (0.1 M, pH 4.0) with the final volume of 1 mL. The above system was incubated for 30 min at room temperature, and the absorbance of final products (i.e., TMBox species) was recorded at the wavelength range of 300-700 nm. In order to understand the principle of the POD activity, we have investigated the cyclic voltammetry characteristics for each system (see the detail in ESI). Similarly, ABTS and OPD were also explored for the POD activity test, and UV-vis spectra of final products were determined with a wavelength from 300 to 700 nm.

Quantification and Discrimination of Phenols
In this experiment, four phenolic substances were selected as the target analytes, including p-Dihydroxybenzene (p-DHB), m-Dihydroxybenzene (m-DHB), o-Dihydroxybenzene (o-DHB), and phenol. First, 0.8 mL of the Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl (0.1 mg/mL) were separately added into two 4-mL microcentrifuge tubes. Afterward, varied concentrations of analytes, 0.1 mL of H 2 O 2 (100 mM), 0.1 mL of TMB (10 mM), and 980 μL of NaAc-HAc buffer (0.1 M, pH 4.0) were injected into the above systems. After incubation for 30 min at room temperature, the ratio (I 652 /I 450 ) for each sample was recorded by using UV-vis absorbance spectroscopy.

Characterization of Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl
The morphology and microstructure of the Cu 2 (OH) 3 Cl were further investigated by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The SEM image (Fig. 2a) showed a hierarchical structure of Cu 2 (OH) 3 Cl sheets with an overall diameter of 20-30 μm. The HRTEM image (Fig. 2b) displayed two lattice fringes with a spacing of 0.24 nm and 0.22 nm, corresponding to (202) and (023) planes of γ-Cu 2 (OH) 3 Cl. The elemental mappings (Fig. 2c-f) revealed that Cu, O, and Cl were homogeneously distributed on the surface of γ-Cu 2 (OH) 3 Cl (Fig. 2d), and the atomic ratio of O to Cl reached 2.6:1. The lower oxygen composition in Cu 2 (OH) 3 Cl was possible due to existence of oxygen vacancies (O V ), which was generally common in oxides and hydroxides (Ye et al. 2019;Al-Hashem et al. 2019). In the EDS spectrum of Ni-doped Cu 2 (OH) 3 Cl (Fig. S1), this ratio was further reduced to 1.76:1, indicative of enhancement in the number of oxygen vacancies). XPS analyses were performed to investigate the chemical composition and oxidation states presented in Cu 2 (OH) 3 Cl. The whole XPS spectrum (Fig. 3a) confirmed the presence of Cu, O, and Cl without any impurities. Their atomic percentages obtained were 35.54% (Cu), 45.02% (O), and 19.44% (Cl), respectively. This result was almost consistent with the EDS measurements, further indicative of oxygen deficiency in Cu 2 (OH) 3 Cl. As depicted in Fig. 3b, the binding energies of Cu 2p 3/2 , and Cu 2p 1/2 were maxima at 934.7 eV and 954.7 eV, respectively. Obviously, Cu 2+ and Cu + coexisted in Cu 2 (OH) 3 Cl due to the weak stability in lattices (Lu et al. 2021). Besides, the O 1 s XPS spectrum was fitted into three peaks (Fig. 3c), and they were individually assigned to lattice oxygen (O L )/hydroxyl (O OH ) at 530.7 eV, the oxygen atoms in the vicinity of an oxygen vacancy (O V ) at 531.4 eV and chemisorbed water (O W ) at 532.1 eV (Tang et al. 2020). Being an effective tool for the determination of unpaired electrons in materials, electron paramagnetic resonance (EPR) was further employed to evidence the O V in Cu 2 (OH) 3 Cl (Zhang et al. 2016;Hao et al. 2018) (Fig. 3d). The obtained Cu 2 (OH) 3 Cl represented an obvious EPR signal arising from the electrons trapped on oxygen vacancies. Figure S2 showed the O 1 s XPS spectrum of Ni-doped Cu 2 (OH) 3 Cl, in which we could see an energy shift from 531.2 to 531.4 eV in peak maximum. This result further confirmed the population of oxygen vacancies in Ni-doped Cu 2 (OH) 3 Cl increased. Basically, the catalytic activities of oxygen-contained nanozymes were strongly related to the amounts of O V in the material (Lu et al. 2020;Wang et al. 2022). They served as active sites for adsorption and activation of O 2 and H 2 O 2 (Wang et al. 2021b).
The FTIR spectrum of Cu 2 (OH) 3 Cl (Fig. S3) displayed three typical peaks at 3309 cm −1 , 3359 cm −1 , and 3447 cm −1 , corresponding to typical O-H stretches with different atomic distances (d O-H) and hydrogen bond angles (θ O-H⋅⋅⋅Cl) (Wang et al. 2021a). And the absorption peak located at 1634 cm −1 should be attributed to the O-H bending vibration. These results demonstrated that the obtained Cu 2 (OH) 3 Cl was functionalized by amounts of hydroxyl groups. In the fingerprint region, besides, the peak was clearly observed at 912 cm −1 , attributable to Cu-O-H of copper hydroxychlorides (Zhao et al. 2020). Besides, Fig. S4 depicts TG and DTG curves. The decomposition of Cu 2 (OH) 3 Cl basically displayed two major steps: Initially, 13% mass loss was observed in the range of 187 to 330 °C accompanied by an endothermal peak centered at 263 °C. This step was caused by the dehydration process. The second step was displayed between 330 and 604 °C with an endothermal peak at 542 °C. It was possibly attributed to the loss of halogen and the formation of CuO, indicative of its strong thermostability (Bhatta et al. 2021).

Analysis of POD Activity
A typical chromogenic substrate (TMB), was selected to validate the POD activity of Cu 2 (OH) 3 Cl. As shown in Fig. 4a, the TMB, H 2 O 2 , and Cu 2 (OH) 3 Cl were mixed together, which represented three absorption peaks located at about 370 nm, 450 nm, and 652 nm, respectively. Whereas, negligible absorbance was acquired in the absence of the Cu 2 (OH) 3 Cl or H 2 O 2 , indicative of superior peroxidaselike property for the Cu 2 (OH) 3 Cl rather than an oxidaselike activity (Fig. 4a). As illustrated in Fig. S5, there were two kinds of charge-transfer species during the whole peroxidation of TMB, namely single-electron reducing product (TMB-TMB ++ ) and double-electron reducing product (TMB diamine). The TMB-TMB ++ represented a blue color with absorbance at 370 nm and 652 nm, and TMB diimine showed a yellow color with a peak located at 450 nm ). However, this phenomenon has always been neglected in previous studies, and investigations were mainly focused on the peak variations at 652 nm (Jiang et al. 2017). As depicted in Fig. 4a, the Cu 2 (OH) 3 Cl-catalyzed product should be a mixture containing the above two species. Additionally, some other chromogenic substrates were used to evaluate its catalytic capabilities, such as 2,2′-azinobis (3-ethylbenzthiazoline-6-sulfonic acid) (ABTS) and o-phenylenediamine (OPD). Interestingly, these two indicators exhibited no responses in the presence of H 2 O 2 and Cu 2 (OH) 3 Cl.
In terms of the catalytic mechanism, the reaction pathways of POD were generally based on the principles of hydroxyl radicals and/or electron transfer (Zheng et al. 2016). To interpret the principle, we first used terephthalic acid (TA) as a fluorescence indicator to study if hydroxyl radicals were produced during the Cu 2 (OH) 3 Cl-catalyzed POD process. Figure 4b shows that there was a weak fluorescence emission at 440 nm (indicative of radical reaction) in the mixture of TA, H 2 O 2 , and Cu 2 (OH) 3 Cl. However, there were no obvious changes in the POD activity when the reaction system was mixed with isopropanol (IPA, a hydroxyl radical quencher) (Fig. 4c). The contradictory results indicated that hydroxyl radicals were indeed generated, but they were not the core factor to drive the catalytic process. Besides, Fig. 4d displays the cyclic voltammograms of Cu 2 (OH) 3 Cl with and without H 2 O 2 , of which a remarkable reduction response occurred in the presence of H 2 O 2 . Obviously, a distinct electron transfer has been triggered during the POD reaction of TMB, which was accelerated by Cu 2 (OH) 3 Cl as a mediator.

Ratiometric Sensor for Determination of Phenols
In the presence of Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl, varied regulations of different phenols on dual absorbance of TMBox mixture are discovered in Fig. 5 and Fig. S6. Inspired by the previous works (Wang et al. 2021a), we speculated the interaction between Cu 2 (OH) 3 Cl and phenols Fig. 4 a UV-vis absorbance profiles of reaction systems in the presence of Cu 2 (OH) 3 Cl and TMB (black line), H 2 O 2 and TMB (red line), Cu 2 (OH) 3 Cl and H 2 O 2 (blue line), as well as the mixture of the above three species (pink line). b Fluorescence spectra of Cu 2 (OH) 3 Cl + H 2 O 2 mixture before and after addition of TA. c Catalytic kinetics analysis of reaction system with and without IPA. d Cyclic voltammograms of obtained Cu 2 (OH) 3 Cl before and after the addition of H 2 O 2 are as follows (Scheme 1): (1) a σ-Cu-ligand complex was formed in which the surface Cu 2+ of Cu 2 (OH) 3 Cl combined with the hydroxyls of phenols. (2) Electrons of the phenols were transferred to Cu 2+ through the complex. (3) In the presence of H 2 O 2 , the σ-Cu-ligand complex would be cleaved resulting in both the reduction of Cu 2+ to Cu + and the hydroxylation of the aromatic ring, as well as producing hydroxyl radicals during a reaction. Integrating with the interaction principles among Cu 2 (OH) 3 Cl, TMBox, and phenols, it was reasonable to speculate that electron transfer might occur across these three compounds via a redox process (Scheme 1).
As depicted in Fig. S7, compared with the previous mixture of Cu 2 (OH) 3 Cl, H 2 O 2 , and TMB, a new oxidation peak was obviously seen in the cyclic voltammograms after the addition of o-THB. Furthermore, the oxidation peak for o-THB was significantly stronger than that of m-DHB (Fig. S8). It is indicated that the electron transfer was triggered among Cu 2 (OH) 3 Cl, phenols, and TMBox species. In addition, because of varied steric hindrances and reducing ability, different electron transfer pathways occurred for each phenol, resulting in the distinguished composition of TMBox species and colorimetric signal outputs. The difference in electrochemical features between o-THB and m-DHB confirmed this speculation. With this consideration, the POD-like Cu 2 (OH) 3 Cl could be employed as a receptor element in the colorimetric sensor array for the discrimination of various phenols.

Recognition of Phenols
To explore the recognition capability of our sensor array, three phenols (i.e., o-DHB, m-DHB, p-DHB) were chosen as models. In the presence of 20 μm of each phenol, the ratiometric absorbance data (I 652 /I 450 ) were significantly different between Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl due to their distinguished catalytic capability (Fig. 5 and Fig. S6). Doping with Ni could significantly improve the POD activity of Cu 2 (OH) 3 Cl because it can increase the number of oxygen vacancies. As depicted in Fig. 6a, the I 652 /I 450 changes of the Cu 2 (OH) 3 Cl and Ni-doped Cu 2 (OH) 3 Cl were used as the colorimetric response fingerprints of the analytes. To exhibit the fingerprints more explicitly, the training matrix pattern (two sensing receptors × three phenols × three repetitions) has been presented by fingerprints, canonical score, and box plots. Figure 6a shows that diverse I 652 /I 450 responses can be obtained from various targets, which indicated that each phenol has a distinctive colorimetric response to the Cu 2 (OH) 3 Cl and its Ni-doped Cu 2 (OH) 3 Cl. The PCA results also further confirmed that the three phenols were different from each other. As depicted in Fig. 6b, three phenols were divided into three distinguished clusters with a classification accuracy of 100%. Each cluster was well separated without any connection, which indicated that the sensors had a strong recognition ability for these phenols. Moreover, the box plot (Fig. 6c) showed that the amplitude of variation at I 652 /I 450 of Ni-doped Cu 2 (OH) 3 Cl was larger than the pure one, indicating that doping with Ni was in favor of increasing the POD activity and identification ability of Cu 2 (OH) 3 Cl.

Quantitative Analysis of a Single Phenols
To further quantify the analytes based on the array's response (i.e., Cu 2 (OH) 3 Cl and its Ni-doped species), the ratiometric colorimetric signals to a single phenol with different concentrations were recorded. The PCA analysis was used to form factors, and further regression or correlation analysis was carried out. Factor 1 and factor 2 are the factor scores calculated by the PCA analysis. Since factor 1 is the main influence factor of the array, it can be used to correlate the concentrations of 3 target analytes.
Scheme 1 Possible catalytic mechanism among Cu 2 (OH) 3 Cl, TMBox species, and phenols Fig. 6 a Fingerprints, b canonical score plots, and c box plots for ratiometric array's colorimetric signal patterns acquired against diverse levels of 20 μm phenols As illustrated in the PCA plots (Fig. 7), factor 1 changed linearly with the concentrations of the same phenol, which were clearly separated without any overlaps. Three linear relationships between the phenol levels and factor 1 were well fitted, in which the linear correlation coefficients (R 2 ) of o-, m-, and p-phenols were 0.99, 0.99, and 0.95, respectively.

Discrimination of Phenol Mixtures
The sensor array was applied to distinguish binary (o-DHB/ m-DHB, m-DHB/p-DHB, o-DHB/p-DHB) and ternary (o-DHB/ m-DHB/ p-DHB) mixtures with molar ratios of 1:1 and 1:1:1, respectively. As shown in Fig. 8, each phenol sample has its own separated cluster in PCA plots, suggesting that the array was able to differentiate complex phenols components.

Antijamming Assay
The antijamming assay was also performed based on the sensor array. In this test, a series of interfering substances such as amino acids and ions were used (Fig. 9). Although interfering biomolecules at 20 μm produce colorimetric signals, they are obviously separated from the three phenols without any overlap. This demonstrated that the sensor array exhibited excellent anti-interference performance for the targets.

Unknown Sample Recognition Ability
In order to examine the availability for sensing unknown samples via the cross-reactive sensor array, we randomly test 9 unknown samples. In the measurements, all unknown samples are blinded. As depicted in the PCA plot (Fig. 10), we compared the Mahalanobis squared distance from blind samples to the center of the specific phenol cluster. We can tell the phenol type it belonged to, judging from the shortest distance. Obviously, all 9 blind samples have been correctly recognized based on this method, indicating its powerful identifying capability for different phenols.

Real Sample Test in Juice
Real sample tests were carried out to examine the practicability of this array for the detection of these three phenols. In the diluted juice samples, three kinds of phenols at 5 and 20 μm levels were individually monitored (Fig. 11). After that, the ratiometric absorbance signals have been recorded and analyzed by a PCA clustering method. The final results indicated that the blank and spiked juice samples with two different concentrations of phenols produced unique response patterns, which were segregated from each other. In addition, the positions with lower concentration (5 μm) were all closer to the blank (Fig. 11), which suggested that the sensors had an outstanding ability in identifying.

Conclusion
In summary, we have developed a colorimetric sensor array for ratiometric determination of o-, m-, and p-phenols. This sensor array was carried out via nanozyme-mediated regulation of TMBox species by phenols. Diverse phenols triggered different changes in absorbance at 652 nm and 450 nm, resulting in variations in I 652 /I 450 and the color of the solution. Through the linear relationship between I 652 /I 450 and the concentration of each phenol, the quantification of targets has been successfully performed. In addition, PCA plots obtained from fingerprints were carried out for discrimination of o-, m-, and p-phenols. Moreover, this strategy exhibited excellent robustness in the recognition of blind samples. In contrast with previous assays, this sensor array showed higher accuracy and selectivity to phenols, and there is no need to prepare substantial specific recognition receptors for targets.