A Sensitive Hydroquinone Amperometric Sensor Based on a Novel Palladium Nanoparticle/Porous Silicon/Polypyrrole-Carbon Black Nanocomposite

Exposure to hydroquinone (HQ) can cause various health hazards and negative impacts on the environment. Therefore, we developed an efficient electrochemical sensor to detect and quantify HQ based on palladium nanoparticles deposited in a porous silicon-polypyrrole-carbon black nanocomposite (Pd@PSi−PPy−C)-fabricated glassy carbon electrode. The structural and morphological characteristics of the newly fabricated Pd@PSi−PPy−C nanocomposite were investigated utilizing FESEM, TEM, EDS, XPS, XRD, and FTIR spectroscopy. The exceptionally higher sensitivity of 3.0156 μAμM−1 cm−2 and a low limit of detection (LOD) of 0.074 μM were achieved for this innovative electrochemical HQ sensor. Applying this novel modified electrode, we could detect wide-ranging HQ (1–450 μM) in neutral pH media. This newly fabricated HQ sensor showed satisfactory outcomes during the real sample investigations. During the analytical investigation, the Pd@PSi−PPy−C/GCE sensor demonstrated excellent reproducibility, repeatability, and stability. Hence, this work can be an effective method in developing a sensitive electrochemical sensor to detect harmful phenol derivatives for the green environment.


Introduction
Hydroquinone (HQ) is widely utilized in various industries including textiles, pharmaceuticals, dyes, oil refinery, cosmetics, and others [1]. Therefore, HQ certainly pollutes the environment and causes a critical hazard to human health [2,3]. Even moderate HQ contact can cause a nuisance, fatigue, haziness, vomiting, etc., [4]. Therefore, the HQ is designated as one of the common hazards in the environment by the US Environmental Protection Agency and the European Union [5]. Consequently, there is an urgent need to develop an effective technique to detect HQ from environmental samples. So far, several HQ determination methods have been reported such as the spectrophotometric method [6], fluorimetric method [7], HPLC [8], GC-MS [9], and electrochemical methods [10][11][12][13]. However, the interference effect from closely related pollutants was frequently observed in spectrophotometric detections. Furthermore, the HPLC and GC-MS require huge amounts of costly solvents, sophisticated instruments, and expert hands. Due to these various challenges and slowness, these techniques are unsuitable for on-site detection. Furthermore, fluorimetric determinations often fail to reproduce investigation results. However, the Powdered silicon (~40 µm), NaH 2 PO 4 , Na 2 HPO 4 , HF, HNO 3 , palladium chloride, polypyrrole-doped carbon black, and hydroquinone were purchased from Sigma Aldrich and used as received. We utilized double-distilled water for preparing all the solutions. The XPS for the Pd@PSi−PPy−C was achieved utilizing the MgKα spectrometer (JEOL, JPS 9200) under the following conditions: pass energy = 50 eV (wide-scan) and 30 eV (narrow-scan), voltage = 10 kV, and current = 20 mA. XRD spectra were recorded using the PANalytical X-ray diffractometer using Cu Kα 1/2 , λα 1 = 154.060 p.m., λα 2 = 154.439 p.m. radiation. A Perkin Elmer 100 spectrometer was used to record the FTIR spectra from the PSi and Pd@PSi−PPy−C nanocomposite. FE-SEM investigations were performed using an FE-scanning electron microanalyzer (JEOL-6300F, 5 kV). The elemental analysis of the as-grown Pd@PSi−PPy−C was performed by EDS (JEOL, Japan). TEM micrographs were taken at 200 kV using a JEOL JEM-2100F-UHR field emission instrument fitted out with a Gatan GIF 2001 energy filter and 1 k-CCD camera. Electrochemical investigations were performed utilizing a Zahner Zennium potentiostat (German).

Synthesis of the PSi, PSi−PPy-C, and Pd@PSi−PPy-C Nanocomposite
First, 2.0 g of powdered silicon was disseminated in 20 mL 48% HF and 80 mL distilled water. Then, 10 mL 70% HNO 3 was added to the beaker dropwise under mild stirring at ambient conditions [35]. The production of nitrogen dioxide vapor indicated the end of stain etching method. We collected the PSi nanoparticles (NPs) by decantation followed by We used a facile sonication method to synthesize the PSi−PPy−C nanocomposite containing 5 wt% PPy−C polymer. For this, 0.4 g PSiNPs and 0.02 g PPy−C were carefully mixed and then dispersed in 80 mL distilled water with a mild sonication for~30 min. Lastly, the PSi−PPy−C nanocomposite was collected and dried at 60 • C.
Later, we deposited 1% PdNPs onto the PSi-PPy-C nanocomposite utilizing a photodeposition method. Herein, 0.2 g PSi-PPy-C was disseminated in 1% methanol solution (v/v) and stirred for 5 min. Then, 500 µL of a palladium solution comprising 0.004 g palladium was added dropwise to this 1% methanol solution. Finally, we irradiated light for 24 hrs in mildly stirred conditions using a UV source from a Philips Hg lamp (illumination intensity at 350 nm: 2.0 mWcm −2 ). We collected this 1%Pd@PSi−PPy−C nanocomposite via decantation and dried it at 60 • C. This as-grown 1% Pd@PSi−PPy−C is denoted as Pd@PSi−PPy−C in this work.

Modification of Glassy Carbon Working Electrode Using Pd@PSi−PPy−C Nanocomposite
Glassy carbon electrodes (GCEs) were cleaned, respectively, utilizing 1 µm diamond followed by 0.05 µm alumina. Later, GCEs were modified with the Pd@PSi-Ppy-C nanocomposite utilizing the Nafion solution. In the fabrication process, a 3.0 mg Pd@PSi-PPy-C was homogeneously mixed in 0.05 mL Nafion-0.45 mL propan-2-ol mixture and then optimized 1.5 µL suspension was cautiously transferred to clean, polished GCEs and dried at 60 • C for 20 min. Such modified GCEs are denoted as Pd@PSi−PPy−C/GCE. For the control experiments, PSi/GCE, and PSi-PPy-C/GCE were also fabricated by similar procedures. A standard 3-electrode electrochemical cell was used where a Pd@PSi−PPy−C/GCE, Ag/AgCl, and a platinum spiral were utilized as a working electrode, a reference electrode, and a counter electrode, respectively. The electrochemical investigations of HQ (1-700 µM) were carried out using cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and amperometry at room temperature utilizing 0.1 M PBS of pH 7.0.

Characterization of the Pd@PSi−PPy−C Nanocomposite
XPS was used to study the purity and structure of Pd@PSi−PPy−C. From the XPS study presented in Figure 1a-f, it is clear that this Pd@PSi−PPy−C nanocomposite was composed of Si, Pd, C, N, and O atoms only. The Pd-3d spectrum shows two well-resolved peaks ( Figure 1b) that appeared at 337.2 and 342.1 eV and that can be correlated to Pd3d 5/2 and Sb3d 3/2 , respectively. These are also consistent with the binding energies of Pd3d [36]. In the fine scan of the Si-2p XPS spectrum (Figure 1c), the peaks at 99.5 and 103.2 eV were correlated to Si2p 3/2 and Si2p 1/2 , respectively [37]. In the deconvoluted C-1s spectrum (Figure 1d), three peaks appeared at 284.1, 285.4, and 288.1 eV, and of these, the peaks at 284.1 and 285.4 eV could be consigned to C-C and C-O-H bonds, respectively [38,39], and the remaining peak at 288.1 eV was related to COOH [40]. A previous report recommended that the C-1s peak appearing at 285.4 eV was also correlated to the C-N bonds of PPy [41]. A deconvolution plot of the N-1s spectrum ( Figure 1e) displayed a peak at 399.9 eV that was related to the C-N bond of PPy moiety [42]. Figure 1f shows two peaks that appeared at 533.1 and 533.7 eV from the deconvoluted O1s spectrum that could be correlated to Si-O and C-O bonds, respectively [43]. In the XRD patterns (Figure 2a), the diffraction bands that appeared at 2θ = 28.3°, 47.3°, 56.0°, 69.1°, and 76.3° were related to (2 2 0), (3 1 1), (4 0 0), (3 3 1), and (4 2 2) lattice planes for Si (JCPDS # 27-1402), respectively [44]. The carbon-related peak of carbon black present in the Pd@PSi-PPy-C often appeared at 2θ = 24.3°, which was correlated to (0 0 2) plane [21], which is not properly visible in Figure 2a due to its sluggish intensity. Because of the low palladium content (1%) in Pd@PSi-PPy-C, the PdNPs peaks were not visible in the XRD patterns; however, the presence of PdNPs in the Pd@PSi-PPy-C was confirmed by XPS, EDS, SEM, and TEM. Figure 2b displays the FTIR investigation results of the Pd@PSi-PPy-C nanocomposite. Characteristic PSi NPs' vibrational bands appearing at 2287 and 2095 cm −1 could be correlated to Si-H2 and Si-H stretching modes of vibration, respectively [45]. Distinct FTIR peaks that appeared at 1075 and 559 cm −1 were attributed to asymmetric Si-O stretching modes of vibrations [46,47].   [44]. The carbon-related peak of carbon black present in the Pd@PSi-PPy-C often appeared at 2θ = 24.3 • , which was correlated to (0 0 2) plane [21], which is not properly visible in Figure 2a due to its sluggish intensity. Because of the low palladium content (1%) in Pd@PSi-PPy-C, the PdNPs peaks were not visible in the XRD patterns; however, the presence of PdNPs in the Pd@PSi-PPy-C was confirmed by XPS, EDS, SEM, and TEM. Figure 2b displays the FTIR investigation results of the Pd@PSi-PPy-C nanocomposite. Characteristic PSi NPs' vibrational bands appearing at 2287 and 2095 cm −1 could be correlated to Si-H 2 and Si-H stretching modes of vibration, respectively [45]. Distinct FTIR peaks that appeared at 1075 and 559 cm −1 were attributed to asymmetric Si-O stretching modes of vibrations [46,47]. In the XRD patterns (Figure 2a), the diffraction bands that appeared at 2θ = 28.3°, 47.3°, 56.0°, 69.1°, and 76.3° were related to (2 2 0), (3 1 1), (4 0 0), (3 3 1), and (4 2 2) lattice planes for Si (JCPDS # 27-1402), respectively [44]. The carbon-related peak of carbon black present in the Pd@PSi-PPy-C often appeared at 2θ = 24.3°, which was correlated to (0 0 2) plane [21], which is not properly visible in Figure 2a due to its sluggish intensity. Because of the low palladium content (1%) in Pd@PSi-PPy-C, the PdNPs peaks were not visible in the XRD patterns; however, the presence of PdNPs in the Pd@PSi-PPy-C was confirmed by XPS, EDS, SEM, and TEM. Figure 2b displays the FTIR investigation results of the Pd@PSi-PPy-C nanocomposite. Characteristic PSi NPs' vibrational bands appearing at 2287 and 2095 cm −1 could be correlated to Si-H2 and Si-H stretching modes of vibration, respectively [45]. Distinct FTIR peaks that appeared at 1075 and 559 cm −1 were attributed to asymmetric Si-O stretching modes of vibrations [46,47].

Electrochemical Investigation of Pd@PSi−PPy−C/GCE
We explored the electrochemical activities of fabricated electrodes by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Figure 4a displays a weak CV output resulting from the bare GCE with 350 µM HQ at +0.61 V. PSi/GCE and PSi−PPy−C/GCE electrodes produced an improved CV response in the presence of 350 µM HQ at +0.52 V and +0.49 V, respectively vs. Ag/AgCl; however, the significantly improved CV result at +0.34 V was attained for Pd@PSi−PPy−C/GCE. Therefore, it was confirmed that the Pd@PSi−PPy−C/GCE electrode showed the best electrocatalytic activities during the HQ detection compared to other electrodes specified in Figure 4a. Hence, we nominated the Pd@PSi−PPy−C/GCE electrode as the HQ sensor in this work. Additionally, for the Pd@PSi−PPy−C/GCE electrode, a distinctive CV peak was achieved with 350 µM HQ, but, in the absence of HQ, no response was observed (Figure 4b

Electrochemical Investigation of Pd@PSi−PPy−C/GCE
We explored the electrochemical activities of fabricated electrodes by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). Figure 4a displays a weak CV output resulting from the bare GCE with 350 µM HQ at +0.61 V. PSi/GCE and PSi−PPy−C/GCE electrodes produced an improved CV response in the presence of 350 µM HQ at +0.52 V and +0.49 V, respectively vs. Ag/AgCl; however, the significantly improved CV result at +0.34 V was attained for Pd@PSi−PPy−C/GCE. Therefore, it was confirmed that the Pd@PSi−PPy−C/GCE electrode showed the best electrocatalytic activities during the HQ detection compared to other electrodes specified in Figure 4a. Hence, we nominated the Pd@PSi−PPy−C/GCE electrode as the HQ sensor in this work. Additionally, for the Pd@PSi−PPy−C/GCE electrode, a distinctive CV peak was achieved with 350 µM HQ, but, in the absence of HQ, no response was observed (Figure 4b) that further established the effective electrochemical properties of the Pd@PSi−PPy−C/GCE HQ sensor. Figure 4c shows the EIS Nyquist plots for the bare GCE, PSi/GCE, PSi−PPy−C/GCE, and Pd@PSi−PPy−C/GCE and the corresponding equivalent circuit is given in the inset. A low semicircular diameter obtained for the Pd@PSi−PPy−C/GCE suggested a lowered R ct value (32 kΩ) of this electrode compared to the bare GCE (78 kΩ), PSi/GCE (47 kΩ), and PSi−PPy−C/GCE (39 kΩ) [48,49]. Therefore, we conclude that the Pd@PSi−PPy−C/GCE electrode exhibited enhanced electron transfer capability in comparison to other electrodes examined in Figure 4c.  To explore the electrochemical oxidation of HQ, we studied the pH effect in the range of 6.0-8.0 in the presence of 350 µM HQ. Figure 5a,b show that the Ipa value gradually increased for pH 6.0-7.0, whereas a decreasing trend for pH 7.0-8.0 was detected, and thus the optimal Ipa was observed at pH ~ 7.0 in Figure 5b. Consequently, pH 7.0 was fixed for the rest of the experiments in this work. Figure 5c displays a straight line plot of Epa vs. pH having the regression Equation (1). To explore the electrochemical oxidation of HQ, we studied the pH effect in the range of 6.0-8.0 in the presence of 350 µM HQ. Figure 5a,b show that the I pa value gradually increased for pH 6.0-7.0, whereas a decreasing trend for pH 7.0-8.0 was detected, and thus the optimal I pa was observed at pH~7.0 in Figure 5b. Consequently, pH 7.0 was fixed for the rest of the experiments in this work. Figure 5c displays a straight line plot of E pa vs. pH having the regression Equation (1). Figure 5c shows that for the examined pH range of 6.0-8.0, a gradient −52 mV per pH unit was very close to the theoretical value −59, confirming that the transferred electron and proton numbers associated with this HQ oxidation were equivalent [16,24].
The scan rate (υ) study in Figure 6a shows the CVs of 350 µM HQ recorded for varying scan rates (40-380 mVs −1 ) utilizing the Pd@PSi−PPy−C/GCE electrode. The I pa value in Figure 6a increased with increasing υ, while the E pa values were only slightly shifted in the positive directions. Figure 6b exhibits the nonlinear variation of I pa vs. υ indicating that the HQ oxidation did not follow a surface-controlled process [50,51]. Moreover, for υ > 0.04 Vs −1 , a linear I pa vs. υ 1/2 curve was also obtained in Figure 6c, confirming a diffusion-controlled process [5,[52][53][54] according to the following Equation (2).  Figure 5c shows that for the examined pH range of 6.0 -8.0, a gradient −52 mV per pH unit was very close to the theoretical value −59, confirming that the transferred electron and proton numbers associated with this HQ oxidation were equivalent [16,24].
The scan rate () study in Figure 6a shows the CVs of 350 µM HQ recorded for varying scan rates (40-380 mVs −1 ) utilizing the Pd@PSi−PPy−C/GCE electrode. The Ipa value in Figure 6a increased with increasing , while the Epa values were only slightly shifted in the positive directions. Figure 6b exhibits the nonlinear variation of Ipa vs.  indicating that the HQ oxidation did not follow a surface-controlled process [50,51]. Moreover, for  > 0.04 Vs −1 , a linear Ipa vs.  1/2 curve was also obtained in Figure 6c, confirming a diffusioncontrolled process [5,[52][53][54] according to the following Equation (2). Ipa(µA) = 99.6886  1/2 (V 1/2 s −1/2 ) + 1.2246 (R 2 = 0.9998) (2)   Figure 5c shows that for the examined pH range of 6.0 -8.0, a gradient −52 mV per pH unit was very close to the theoretical value −59, confirming that the transferred electron and proton numbers associated with this HQ oxidation were equivalent [16,24].
The scan rate () study in Figure 6a shows the CVs of 350 µM HQ recorded for varying scan rates (40-380 mVs −1 ) utilizing the Pd@PSi−PPy−C/GCE electrode. The Ipa value in Figure 6a increased with increasing , while the Epa values were only slightly shifted in the positive directions. Figure 6b exhibits the nonlinear variation of Ipa vs.  indicating that the HQ oxidation did not follow a surface-controlled process [50,51]. Moreover, for  > 0.04 Vs −1 , a linear Ipa vs.  1/2 curve was also obtained in Figure 6c, confirming a diffusioncontrolled process [5,[52][53][54] according to the following Equation (2). Ipa(µA) = 99.6886  1/2 (V 1/2 s −1/2 ) + 1.2246 (R 2 = 0.9998) (2) Again, Figure 6d displays a linear relation between log(I pa ) and log(υ) with the following Equation (3) confirming the diffusion-controlled process [11]. Furthermore, in Figure 6e, another linear plot of E pa vs. log(υ) was achieved with the following Equation (4). E pa (V) = 0.0356 log[υ (Vs −1 )] + 0.3879 (R 2 = 0.9874) (4) Figure 6a exhibits that for υ < 150 mVs −1 , [E pa -E pc ]/2 stayed nearly the same as 48.3 mV. Therefore, at the 50 mVs −1 scan rate, [E pa -E pc ]/2 can be taken as 90.5/n α mV [1], and thus the number of transferred electrons (n α ) was calculated as 1.87 ≈ 2. Thus, we decided that the HQ oxidation at the Pd@PSi-PPy-C/GCE sensor was involved in transferring two electrons. Therefore, the scan rate and the pH studies established that the HQ oxidation at the Pd@PSi-PPy-C/GCE sensor was a combination of two-electron plus two-protons, which is in line with the literature [1].

Determination of Sensor Parameters for the Pd@PSi-PPy-C/GCE Sensor
We used the amperometric technique to explore the sensor performance of the Pd@PSi-PPy-C/GCE sensor. The amperometric (i-t) response was acquired at the optimized potential of +0.35 V with successive HQ additions with varying concentrations (1-700 µM). Figure 7a displays the amperometric i-t curve accomplished with HQ using the Pd@PSi−PPy −C/GCE assembly. Here, for each HQ addition, the current response reached~96% of its highest current in only 6 s. Figure 7b displays two linear segments in the calibration plot: for the lower concentration region, 1-13 µM HQ, and for the higher concentration part, 13-450 µM HQ plotted based on the amperometric responses having the following Equations (5) and (6) Figure 6a exhibits that for  < 150 mVs −1 , [Epa-Epc]/2 stayed nearly the same as 48.3 mV. Therefore, at the 50 mVs −1 scan rate, [Epa-Epc]/2 can be taken as 90.5/ n mV [1], and thus the number of transferred electrons (n) was calculated as 1.87 ≈ 2. Thus, we decided that the HQ oxidation at the Pd@PSi-PPy-C/GCE sensor was involved in transferring two electrons. Therefore, the scan rate and the pH studies established that the HQ oxidation at the Pd@PSi-PPy-C/GCE sensor was a combination of two-electron plus two-protons, which is in line with the literature [1].

Determination of Sensor Parameters for the Pd@PSi-PPy-C/GCE Sensor
We used the amperometric technique to explore the sensor performance of the Pd@PSi-PPy-C/GCE sensor. The amperometric (i-t) response was acquired at the optimized potential of +0.35 V with successive HQ additions with varying concentrations (1-700 µM). Figure 7a displays the amperometric i-t curve accomplished with HQ using the Pd@PSi−PPy−C/GCE assembly. Here, for each HQ addition, the current response reached ⁓96% of its highest current in only 6 s. Figure 7b displays two linear segments in the calibration plot: for the lower concentration region, 1-13 µM HQ, and for the higher concentration part, 13-450 µM HQ plotted based on the amperometric responses having the following Equations (5) and (6)   Thus, the linear dynamic range (LDR) for the Pd@PSi-PPy-C/GCE sensor was obtained as 1-450 µM. Furthermore, the estimated sensitivity value for the Pd@PSi−PPy−C/ GCE sensor was achieved as 3.0156 µAµM -1 cm −2 , the LOD calculated as~0.074 µM (S/N = 3), and the limit of quantification (LOQ) as 0.227 µM. For the sensitivity calculation, we used the equation sensitivity = S/A eff [11], where A eff denotes the surface area of the active electrode (0.0512 cm 2 ) as presented in the electronic supplementary materials [55]. The LOD and LOQ were estimated using equations LOD = 3.3(S b /S) and LOQ = 10(S b /S), respectively [56,57]; herein, the S b (0.0035) stands for RSD of five blank runs and S represents the slope of the calibration curve.
In electrochemical kinetics, electrocatalytic activities depend on two factors: (i) the intensification of the current responses and (ii) the reduction of overpotential during the electrooxidation. Consequently, we tried to enhance the electrochemical property of the active electrode by fabricating the working GCEs with the active Pd@PSi−PPy−C nanocomposite. The obtained results confirmed that this Pd@PSi−PPy−C/GCE sensor effectively fulfills both of the above-mentioned factors. The above Figure 4a shows a more substantial negative shift of E pa and massive I pa increase from the Pd@PSi−PPy−C/GCE sensor than other electrodes employed here. Actually, we attained~3 times the I pa value than the bare GCE during the HQ oxidation at the Pd@PSi−PPy−C/GCE.

Selectivity, Repeatability, Reproducibility, and Stability of Modified Electrodes
For the verification of selectivity of the Pd@PSi−PPy−C/GCE sensor using common interfering chemicals such as catechol (CC), 4-nitrophenol (4-NP), 2-nitrophenol, (2-NP), 4-acetamidophenol (AcP), and Cl − ions, the amperometric response was noted with 20 µM HQ with the same quantity of each interfering chemical (Figure 8a). Herein, the HQ addition produced a current response, but for the interfering chemicals, an insignificant response was observed. Thus, we confirmed the selectivity of the Pd@PSi−PPy−C/GCE sensor in HQ determination. Additionally, other sensor parameters of Pd@PSi−PPy−C/GCE were also studied utilizing CV using 350 µM HQ at the 0.04 Vs −1 scan rate. Figure 8b displays the repeatability study, in which the newly modified Pd@PSi−PPy−C/GCE electrode was utilized in determining 350 µM HQ. An almost-identical CV response was achieved in five runs having 3.8% RSD establishing excellent repeatability. Figure 8c shows a reproducibility investigation for the Pd@PSi−PPy−C/GCE sensor, in which five freshly fabricated Pd@PSi−PPy−C/GCE electrodes were utilized. The CV results showed a 4.2% RSD for I pa variations, confirming excellent reproducibility. Following the fabrication of the Pd@PSi−PPy−C/GCE sensor, we collected CVs every fifth day in a row to test the stability and kept it at ambient conditions. A bar graph of the stability analysis is shown in Figure 8d. It reveals that after 20 days of storage at ambient conditions, the CV response remained at, or near, 85% of its initial value, and the Pd@PSi−PPy−C/GCE surface remained undamaged.
When the HQ molecule touched the Pd@PSi−PPy−C surface, an electrooxidation reaction occurred. Due to the reducing properties of HQ molecules, the electron donation from HQ to the conduction band of the Pd@PSi−PPy−C nanocomposite can occur, which ultimately enhances the Pd@PSi−PPy−C/GCE sensor's conductivity. Thus, the CV responses are improved. The current Pd@PSi−PPy−C/GCE sensor showed an extreme sensitivity during HQ detection compared to existing HQ sensors as shown in Table 1 [4,[58][59][60][61][62][63][64].
Therefore, it is concluded that the Pd@PSi−PPy−C nanocomposite is exceptionally efficient to oxidize HQ. The Pd@PSi−PPy−C nanocomposite provides an encouraging nano-environment in HQ detection. The PSi, PPy−C, and combined PSi−PPy−C nanocomposites are p-type semiconductors [24]. Hence, a combined PdNPs and PSi−PPy−C may generate metal-semiconductor (MS) junctions that provide synergistic effects [65]. A low R ct value of Pd@PSi−PPy−C/GCE as achieved from the EIS spectrum (Figure 4c) further confirms the synergistic effects between PdNPs and PSi−PPy−C. Such a combination might produce an electron donor-acceptor pair resulting in a potential development at the MS junction, which may lead to a decrease in the energy barrier in the oxidation process [35]. Such a Pd@PSi−PPy−C nanocomposite-HQ interaction may be the main reason that makes the Pd@PSi−PPy−C/GCE appropriate in HQ detection. During the HQ oxidation, the dispersed PdNPs onto the PSi-PPy-C surface expedite HQ molecules' congregation at the electrode/solution interface, and this increases the electrode sensitivity in HQ sensing [66]. The effective Pd@PSi−PPy−C nanocomposite-HQ interactions enable the Pd@PSi−PPy−C/GCE sensor appropriate for HQ determination. Furthermore, the enhanced performance of the currently developed Pd@PSi−PPy−C/GCE-based HQ sensor is also likely related to the efficient attachment of the active nanocomposite onto the GCE surface, which provides a 56% higher effective surface area than the bare, unmodified GCE, facilitating a rapid electron transfer during the electrooxidation of HQ. Scheme 1 denotes the electrooxidation of HQ at the Pd@PSi−PPy−C/GCE assembly. When the HQ molecule touched the Pd@PSi−PPy−C surface, an electrooxidation reaction occurred. Due to the reducing properties of HQ molecules, the electron donation from HQ to the conduction band of the Pd@PSi−PPy−C nanocomposite can occur, which ultimately enhances the Pd@PSi−PPy−C/GCE sensor's conductivity. Thus, the CV responses are improved. The current Pd@PSi−PPy−C/GCE sensor showed an extreme sensitivity during HQ detection compared to existing HQ sensors as shown in Table 1 [4,[58][59][60][61][62][63][64]. Pd@PSi   Therefore, it is concluded that the Pd@PSi−PPy−C nanocomposite is exceptionally efficient to oxidize HQ. The Pd@PSi−PPy−C nanocomposite provides an encouraging nanoenvironment in HQ detection. The PSi, PPy−C, and combined PSi−PPy−C nanocomposites are p-type semiconductors [24]. Hence, a combined PdNPs and PSi−PPy−C may generate metal-semiconductor (MS) junctions that provide synergistic effects [65]. A low Rct value of Pd@PSi−PPy−C/GCE as achieved from the EIS spectrum ( Figure 4c) further confirms the synergistic effects between PdNPs and PSi−PPy−C. Such a combination might produce an electron donor-acceptor pair resulting in a potential development at the MS junction, which may lead to a decrease in the energy barrier in the oxidation process [35]. Such a Pd@PSi−PPy−C nanocomposite-HQ interaction may be the main reason that makes the Pd@PSi−PPy−C/GCE appropriate in HQ detection. During the HQ oxidation, the dispersed PdNPs onto the PSi-PPy-C surface expedite HQ molecules' congregation at the electrode/solution interface, and this increases the electrode sensitivity in HQ sensing [66]. The effective Pd@PSi−PPy−C nanocomposite-HQ interactions enable the Pd@PSi−PPy−C/GCE sensor appropriate for HQ determination. Furthermore, the enhanced performance of the currently developed Pd@PSi−PPy−C/GCE-based HQ sensor is also likely related to the efficient attachment of the active nanocomposite onto the GCE surface, which provides a 56% higher effective surface area than the bare, unmodified GCE, facilitating a rapid electron transfer during the electrooxidation of HQ. Scheme 1 denotes the electrooxidation of HQ at the Pd@PSi−PPy−C/GCE assembly. Scheme 1. Schematic representation of electrochemical oxidation of HQ at Pd@PSi−PPy−C/GCE sensor electrode.

Real Sample Investigation
To confirm the sensor electrode appropriateness, we detected HQ from spiked tap water utilizing the Pd@PSi−PPy−C/GCE sensor via a standard addition method. Herein, equal volumes of HQ solutions (varying concentrations) and tap water were mixed individually, and we recorded the CV responses in PBS utilizing the Pd@PSi−PPy−C/GCE sen-Scheme 1. Schematic representation of electrochemical oxidation of HQ at Pd@PSi−PPy−C/GCE sensor electrode.

Real Sample Investigation
To confirm the sensor electrode appropriateness, we detected HQ from spiked tap water utilizing the Pd@PSi−PPy−C/GCE sensor via a standard addition method. Herein, equal volumes of HQ solutions (varying concentrations) and tap water were mixed individually, and we recorded the CV responses in PBS utilizing the Pd@PSi−PPy−C/GCE sensor. Table 2 summarizes the results achieved, which demonstrated that the Pd@PSi−PPy−C/GCE sensor exhibited~100% recovery of HQ. Consequently, we concluded that the Pd@PSi−PPy−C/ GCE assembly is suitable, precise, and consistent for the determination of HQ from a real sample. Herein, during the HQ detection, the electrochemical response progressively increased with the increasing HQ concentrations. Thus, HQ became oxidized at the Pd@PSi−PPy−C/ GCE sensor surface by losing two electrons to the conduction band of the Pd@PSi−PPy−C nanocomposite that originated the electrochemical response [1]. Therefore, the overall senor activity of the Pd@PSi−PPy−C/GCE can be presented as in Scheme 2.

Conclusions
We successfully synthesized and systematically characterized a new ternary Pd@PSi−PPy−C nanocomposite via facile stain etching, sonication, and photodeposition procedures. An electrochemical hydroquinone sensor with extremely high sensitivity was designed using the Pd@PSi−PPy−C−fabricated GCE. A significantly high sensitivity value indicated the suitability of the Pd@PSi−PPy−C/GCE sensor in determining the wide-ranging HQ. Exceptional promising features of the Pd@PSi−PPy−C/GCE sensor including outstanding stability and lower LOD established the prospective of the ternary Pd@PSi−PPy−C nanocomposite during HQ detection and quantification. The reliability of this newly designed HQ sensor was confirmed utilizing spiked tap water investigations with promising analytical results. Therefore, the ternary Pd@PSi−PPy−C nanocompositebased electrochemical sensor design method paves a new route to develop efficient electrochemical sensors to detect and quantify pollutants.

Conclusions
We successfully synthesized and systematically characterized a new ternary Pd@PSi −PPy−C nanocomposite via facile stain etching, sonication, and photodeposition procedures. An electrochemical hydroquinone sensor with extremely high sensitivity was designed using the Pd@PSi−PPy−C−fabricated GCE. A significantly high sensitivity value indicated the suitability of the Pd@PSi−PPy−C/GCE sensor in determining the wide-ranging HQ. Exceptional promising features of the Pd@PSi−PPy−C/GCE sensor including outstanding stability and lower LOD established the prospective of the ternary Pd@PSi−PPy−C nanocomposite during HQ detection and quantification. The reliability of this newly designed HQ sensor was confirmed utilizing spiked tap water investigations with promising analytical results. Therefore, the ternary Pd@PSi−PPy−C nanocompositebased electrochemical sensor design method paves a new route to develop efficient electrochemical sensors to detect and quantify pollutants.