High Performance Zinc Oxide Nanorod-Doped Ion Imprinted Polypyrrole for the Selective Electrosensing of Mercury II Ions

: A biomimetic, ion-imprinted polymer (IIP) was prepared by electropolymerization of pyrrole at the surface of gold electrodes decorated with vertically grown ZnO nanorods. The vertical growth of the nanorods was achieved via an ultrathin aryl monolayer grafted by reduction of diazonium salt precursor. Pyrrole was polymerized in the presence of L-cysteine as chelating agent and Hg 2 + (template). Hg 2 + -imprinted polypyrrole (PPy) was also prepared on a bare gold electrode in order to compare the two methods of sensor design (Au-ZnO-IIP vs. Au-IIP). Non-imprinted PPy was prepared in the same conditions but in the absence of any Hg 2 + template. The strategy combining diazonium salt modiﬁcation and ZnO nanorod decoration of gold electrodes permitted us to increase considerably the speciﬁc surface area and thus improve the sensor performance. The limit of detection (LOD) of the designed sensor was ~1 pM, the lowest value ever reported in the literature for gold electrode sensors. The dissociation constants between PPy and Hg 2 + were estimated at [K d1 = (7.89 ± 3.63) mM and K d2 = (38.10 ± 9.22) pM]. The sensitivity of the designed sensor was found to be 0.692 ± 0.034 µ A.pM − 1 . The Au-ZnO-IIP was found to be highly selective towards Hg 2 + compared to cadmium, lead and copper ions. This sensor design strategy could open up new horizons in monitoring toxic heavy metal ions in water and therefore contribute to enhancing environmental quality.


Introduction
One of the most serious problems affecting water is chemical pollution by organic or inorganic compounds such as pesticides and heavy metals, respectively. Among the inorganic pollutants, mercury (II) presents several health and ecosystem concerns [1][2][3][4]. Mercury is released into the environment in large quantities through waste from thermal power plants, cement kilns, chlor-alkali plants, gold mining and trash incinerators (hazardous waste, medical waste, regular garbage, etc.) [5][6][7][8]. Exposure to mercury ions, even at low doses, can lead to severe central nervous system problems and vital human organ damage [2,[8][9][10][11]. The discharge of mercury into the environment is forbidden offer to the surface scientist in the quest of constructing a robust electrochemical sensor to address environmental issues.
Herein, a polypyrrole-based IIP was prepared in the presence of Hg 2+ as a template and L-cysteine as a chelatant agent deposited on a gold electrode sequentially functionalized with diazonium salts and nanostructured with ZnO nanorods in order to have very high roughness for the selective removal of mercury ions in a sample medium (KCl solution). For comparison, an IIP on a bare gold electrode was prepared. A non-imprinted polymer was also prepared in the same conditions of IIP but in the absence of any Hg 2+ . Electropolymerization time, incubation time and number of cycles of electroreduction of diazonium salts were investigated and optimized. Electrode sensing materials were characterized by atomic force microscopy (AFM) in order to account for the morphological structures of the nanocomposites. Square wave voltammetry (SWV) experiments permitted us to determine the optimal parameters for sensing Hg 2+ , namely linear response range and the limit of detection of the ZnO-IIP sensor. Selectivity and competition between mercury and other ions was investigated by examining the electrochemical responses in the presence of copper, lead and cadmium taken as interfering metal ions.

Instruments and Characterization
All electrochemical measurements were carried out with a Bio-Logic portable potentiostat (model PG581). Conventional three-electrode cell was used. A silver chloride electrode (SSC, +0.222 V vs. SHE), a platinum grid and a gold substrate were the reference, counter and working electrodes, respectively.
Topographical characterization was performed by atomic force microscopy (AFM, MFP-3 Asylum Research). X-ray photoelectron spectroscopy (XPS) analyses were conducted using a K Alpha apparatus (Thermo Fisher Scientific, East Grinsted, UK) fitted with an Al monochromatic X-ray source (1486.6 eV; spot size = 400 µm). A flood gun was used for charge compensation.

Surface Functionalization with Diazonium Salt
In order to mitigate the impurities likely to be present on the surface of the gold electrode and which can subsequently alter the quality of the surface modification, the gold substrates were cleaned with acetone in 10 min and then in a mixture of piranha (98% H 2 SO 4 /30% H 2 O 2 2:1 V/V) for 20 and 10 min in ethanol. We then washed the gold electrodes thoroughly with distilled water and dried them at 60 • C.
The surface modification was carried out by electrografting of 4-carboxybenzenediazonium-tetrafluoroborate (previously prepared [66] and characterized by FTIR and carbon-13 NMR) on gold electrodes by cyclic voltammetry at scan rate of 100 mV/s between −1.0 and 0.0 V/SSC for 20 cycles, in acetonitrile solution, containing 1 mM of diazonium salt and 0.1 M of tetrabutylammonium tetrafluoroborate. The modified electrodes were then rinsed with ultrapure water and ethanol and dried. The number of cycles was optimized and controlled using ferrocyanide redox probe.

Synthesis, Deposition and Vertical Growth of ZnO
The ZnO grafting procedure was performed in three steps by the low temperature method (Mekki et al. [47], Pacholski et al. [60]; Chander and Raychaudhuri [61]): (i) synthesis of ZnO nanoparticles from zinc acetate [Zn(CH3COO)2] and sodium hydroxide (NaOH), (ii) deposition of these nanoparticles on the surface and (iii) the vertical growth of the ZnO nanorods. The sodium hydroxide dissolves in the solution to give Na + , which will bind to the acetate; the OH − group will precipitate with the Zn 2+ which is in the solution, before obtaining the zinc monoxide, which has a white color.

Deposition of ZnO Nanoparticles on Aryl-Modified Gold Electrodes
After synthesis, ZnO nanoparticles were deposited on the modified gold electrode (Au-diazo) by spin-coating to serve as a germination layer for the nucleation process during the growth of ZnO nanorods. These electrodes were dried in the oven at 125 • C for one hour before proceeding to the growth step.

Vertical Growth of ZnO Nanorods
For the nanorod growth step, the ZnO NP-coated arylated gold substrates were immersed in a beaker containing a 30 mM solution of the zinc nitrate hexahydrate [Zn(NO 3 ) 2 . 6H 2 O] and 30 mM hexamethylenetetramine (HMT) dissolved in 200 mL of distilled water; the mixture was kept at 95 • C for 5 h with constant stirring. Since the particles act as a nucleation site, after the five hours of growth, the samples were removed from solution, rinsed with distilled water and dried (for 1 h at 125 • C).

Preparation of Au-IIP, Au-ZnO-IIP and Au-ZnO-NIP
Two different IIPs were prepared. The first one was made on a bare gold electrode and the second one was prepared on a gold electrode modified with diazonium salt and ZnO nanorods. These two configurations were carried out in two steps; for the first, a thin PPy barrier layer was deposited by chronoamperometry for 2 s at an applied potential of 1.05 V (vs Ag/AgCl) in a KCl solution containing 10 −2 M of purified pyrrole (monomer) and 10 −4 M of L-cysteine; the second step involved 8 s in a KCl solution containing 10 −2 M of purified pyrrole, 10 −4 M of L-cysteine and 10 −3 M of Hg 2+ (template). These different steps were carried out by chronoamperometry at an applied potential of Appl. Sci. 2020, 10, 7010 5 of 22 1.05 V (vs. Ag/AgCl). The non-imprinted polymer (Au-ZnO-NIP) was also prepared under the same conditions in the absence of any Hg 2+ ion template.

Choice and Role of the Chelating Agent
The formation of mercury receptor sites in the ZnO-IIP was achieved using L-cysteine, a chelator of mercury. This choice was driven by the thiophilicity of metal ions [68][69][70] and especially high affinity of L-Cys for Hg 2+ [71]. L-cysteine is an ideal compound, bearing electron-rich functional groups such as -NH 2 , -COOH and -SH, which allow it to have a Hg-cysteine complex [72][73][74]. The incorporation of L-cysteine/Hg 2+ into polypyrrole is likely to take place via electrostatic bonds between the carboxylic group of L-cys and the nitrogen of pyrrole (Scheme 1) [74].
Appl. Sci. 2020, 10, x 5 of 23 potential of 1.05 V (vs Ag/AgCl). The non-imprinted polymer (Au-ZnO-NIP) was also prepared under the same conditions in the absence of any Hg 2+ ion template.

Choice and Role of the Chelating Agent
The formation of mercury receptor sites in the ZnO-IIP was achieved using L-cysteine, a chelator of mercury. This choice was driven by the thiophilicity of metal ions [68][69][70] and especially high affinity of L-Cys for Hg 2+ [71]. L-cysteine is an ideal compound, bearing electron-rich functional groups such as -NH2, -COOH and -SH, which allow it to have a Hg-cysteine complex [72][73][74]. The incorporation of L-cysteine/Hg 2+ into polypyrrole is likely to take place via electrostatic bonds between the carboxylic group of L-cys and the nitrogen of pyrrole (Scheme 1) [74]. Scheme 1. Schematic representation of Hg 2+ complexation by L-cysteine in the polypyrrole layer.

Electrochemical Measurements
The comparison between the prepared configurations, the Hg 2+ detection tests, selectivity, competition between Hg 2+ and other ions was controlled by square wave voltammetry between −0.95 and 0.95 V under the following conditions: E-pulse: 0.01 V, E-step: 0.01 V, frequency: 25 Hz, equilibration time: 2 s. After each use, the IIP was extracted for 20 min in EDTA and then 10 min in ultra-pure water with gentle stirring. Once the IIP was extracted, it could be used another time.

Surface Functionalization by 4-Carboxybenzenediazonium Tetrafluoroborate and Quasi-Vertical Growth of ZnO
Scheme 2 displays the general three-step design of IIPs. To do so, we functionalized the surface of the gold electrode with carboxyphenyl groups from the corresponding diazonium salt (Scheme 2A, (1)) and then deposited (by spin-coating) ZnO nanoparticles for the vertical growth of the ZnO nanorods (Scheme 2A, (2)). The second step consisted of the formation of the ion-imprinted polypyrrole by electropolymerization of the pyrrole in the presence of L-cysteine (chelator) and Hg 2+ template (Scheme 2B).
The different steps followed for the formation of ionic cavities are illustrated in Scheme 2C [75]. Removal of the template by washing, adapted to the nature of the functional monomer-template bond, leaves artificial sites capable of selective rebinding of the template.

Electrochemical Measurements
The comparison between the prepared configurations, the Hg 2+ detection tests, selectivity, competition between Hg 2+ and other ions was controlled by square wave voltammetry between −0.95 and 0.95 V under the following conditions: E-pulse: 0.01 V, E-step: 0.01 V, frequency: 25 Hz, equilibration time: 2 s. After each use, the IIP was extracted for 20 min in EDTA and then 10 min in ultra-pure water with gentle stirring. Once the IIP was extracted, it could be used another time.

Surface Functionalization by 4-Carboxybenzenediazonium Tetrafluoroborate and Quasi-Vertical Growth of ZnO
Scheme 2 displays the general three-step design of IIPs. To do so, we functionalized the surface of the gold electrode with carboxyphenyl groups from the corresponding diazonium salt (Scheme 2A, (1)) and then deposited (by spin-coating) ZnO nanoparticles for the vertical growth of the ZnO nanorods (Scheme 2A, (2)). The second step consisted of the formation of the ion-imprinted polypyrrole by electropolymerization of the pyrrole in the presence of L-cysteine (chelator) and Hg 2+ template (Scheme 2B).
The different steps followed for the formation of ionic cavities are illustrated in Scheme 2C [75]. Removal of the template by washing, adapted to the nature of the functional monomer-template bond, leaves artificial sites capable of selective rebinding of the template. The electroreduction of the diazonium salt carried out by cyclic voltammetry in 20 cycles in the range −1.0 to 0.0 V (vs Ag/AgCl) gave a virtually zero reduction current ( Figure 1A) resulting from a passivation of the electrode following the grafting of diazonium groups (formation of gold-diazo covalent bonds). The electroreduction peak is centered at −0.2 V (vs Ag/AgCl). This step of surface functionalization by the diazonium salt makes it possible to tether ZnO NPs to the surface on the one hand and enables vertical growth of ZnO NRs on the other hand. We have shown in a previous work that the growth of ZnO NRs on a surface not modified by diazonium salts gives an orientation of the NRs in all directions and in a random manner [47].
Passivation of the electrode was followed by cyclic voltammetry in the presence of 0.01 M of K3[Fe(CN)6] and 0.01 M of K4[Fe(CN)6]. Results presented in Figure 1B indicate that the electrode was passivated while keeping its conductive character ( Figure 1B, red). The diazonium salt step makes it possible to achieve an almost homogeneous distribution of quasi-vertical growth of ZnO nanorods [47]. The electroreduction of the diazonium salt carried out by cyclic voltammetry in 20 cycles in the range −1.0 to 0.0 V (vs. Ag/AgCl) gave a virtually zero reduction current ( Figure 1A) resulting from a passivation of the electrode following the grafting of diazonium groups (formation of gold-diazo covalent bonds). The electroreduction peak is centered at −0.2 V (vs. Ag/AgCl). This step of surface functionalization by the diazonium salt makes it possible to tether ZnO NPs to the surface on the one hand and enables vertical growth of ZnO NRs on the other hand. We have shown in a previous work that the growth of ZnO NRs on a surface not modified by diazonium salts gives an orientation of the NRs in all directions and in a random manner [47].

Preparation of IIP and NIP-Based Electrodes
Mercury oxidation on gold electrodes yields two peaks: the first peak is attributed to the oxidation reaction of Hg(0) to Hg(I) (in the form Hg2 2+ ) while the second peak is due to the  6 ]. Results presented in Figure 1B indicate that the electrode was passivated while keeping its conductive character ( Figure 1B, red). The diazonium salt step makes it possible to achieve an almost homogeneous distribution of quasi-vertical growth of ZnO nanorods [47].

Preparation of IIP and NIP-Based Electrodes
Mercury oxidation on gold electrodes yields two peaks: the first peak is attributed to the oxidation reaction of Hg(0) to Hg(I) (in the form Hg 2 2+ ) while the second peak is due to the transformation of Hg(I) into Hg 2+ [68,76,77]: Before preparing the films on gold electrodes, we first characterized the signature of mercury with a gold disk working electrode in a solution containing mercury and KCl 0.1 M. The response obtained by SWV shows that the oxidation of mercury on gold gives two peaks at 0.061 and 0.71 V (vs. Ag/AgCl) ( Figure 2). The position of the two peaks depends on the pH of the solution, the nature of the working electrode, the electrolyte and the layer or film deposited on gold. In the following, we will consider the most intense peak located at 0.71 V (vs. Ag/AgCl), which corresponds to Hg 2+ . This dominant peak means that most of the mercury is converted to Hg 2+ .

Preparation of IIP and NIP-Based Electrodes
Mercury oxidation on gold electrodes yields two peaks: the first peak is attributed to the oxidation reaction of Hg(0) to Hg(I) (in the form Hg2 2+ ) while the second peak is due to the transformation of Hg(I) into Hg 2+ [68,76,77]: Before preparing the films on gold electrodes, we first characterized the signature of mercury with a gold disk working electrode in a solution containing mercury and KCl 0.1 M. The response obtained by SWV shows that the oxidation of mercury on gold gives two peaks at 61 and 710 mV (vs Ag/AgCl) ( Figure 2). The position of the two peaks depends on the pH of the solution, the nature of the working electrode, the electrolyte and the layer or film deposited on gold. In the following, we will consider the most intense peak located at 0.71 V (vs. Ag/AgCl), which corresponds to the generation of Hg 2+ . This dominant peak means that most of the mercury is converted to Hg 2+ . A thin PPy barrier layer was deposited by chronoamperometry (CA) for 2 s before Au-ZnO-IIP electro-formation to prevent the complexation of mercury by Au-ZnO electrodes ( Figure 3A, black curve). Both ZnO-IIP and ZnO-NIP were prepared by the same technique ( Figure 3A, red curve) and then compared using SWV ( Figure 3B). On the one hand, one can notice the obvious difference between the responses of ZnO-IIP and ZnO-NIP, i.e., mainly for the peak current corresponding to a potential of around 0.62 V (vs. Ag/AgCl) in the case of ZnO-IIP will be considered for quantitative monitoring of mercury. On the other hand, the removal of Hg 2+ from ZnO-IIP by EDTA leads to a flat SWV response, close to that of ZnO-NIP ( Figure 3B). This means that all receptor sites have been released and thus can be reused for rebinding the target ions.
between the responses of ZnO-IIP and ZnO-NIP, i.e., mainly for the peak current corresponding to a potential of around 0.62 V (vs Ag/AgCl) in the case of ZnO-IIP will be considered for quantitative monitoring of mercury. On the other hand, the removal of Hg 2+ from ZnO-IIP by EDTA leads to a flat SWV response, close to that of ZnO-NIP ( Figure 3B). This means that all receptor sites have been released and thus can be reused for rebinding the target ions. In order to showcase the central role of ZnO nanorods, we prepared two different IIPs: the first was prepared directly on a bare gold electrode and the second IIP was prepared on Au-ZnO. After extraction of the metal ions from these two IIPs in EDTA solution, they were incubated in 10 −3 M mercury solution for 20 min for the rebinding test.
From the results displayed in Figure 4, one can notice the important role of ZnO NRs thanks to the spaces between the NRs. ZnO NRs will be considered as a substrate for depositing the polymer matrix instead of depositing it directly on the gold electrode.  In order to showcase the central role of ZnO nanorods, we prepared two different IIPs: the first was prepared directly on a bare gold electrode and the second IIP was prepared on Au-ZnO. After extraction of the metal ions from these two IIPs in EDTA solution, they were incubated in 10 −3 M mercury solution for 20 min for the rebinding test.
From the results displayed in Figure 4, one can notice the important role of ZnO NRs thanks to the spaces between the NRs. ZnO NRs will be considered as a substrate for depositing the polymer matrix instead of depositing it directly on the gold electrode.
potential of around 0.62 V (vs Ag/AgCl) in the case of ZnO-IIP will be considered for quantitative monitoring of mercury. On the other hand, the removal of Hg 2+ from ZnO-IIP by EDTA leads to a flat SWV response, close to that of ZnO-NIP ( Figure 3B). This means that all receptor sites have been released and thus can be reused for rebinding the target ions. In order to showcase the central role of ZnO nanorods, we prepared two different IIPs: the first was prepared directly on a bare gold electrode and the second IIP was prepared on Au-ZnO. After extraction of the metal ions from these two IIPs in EDTA solution, they were incubated in 10 −3 M mercury solution for 20 min for the rebinding test.
From the results displayed in Figure 4, one can notice the important role of ZnO NRs thanks to the spaces between the NRs. ZnO NRs will be considered as a substrate for depositing the polymer matrix instead of depositing it directly on the gold electrode.  Square wave voltammograms presented in Figure 4 show that the current intensity of the IIP prepared on ZnO nanorods is twice as high as that prepared on bare gold electrodes. Given that the surface area used on the bare gold electrode to form Au/Hg-IIP is identical to that used to form Au-ZnO/Hg-IIP, this can be related to the increase in the surface area thanks to ZnO nanorods, which makes it possible to deposit more IIP and thus a large number of artificial sites to capture more Hg 2+ ions. Figure 5 shows the survey spectra for the bare gold electrode, Au/diazo-COOH, Au-diazo-ZnO nanoparticles, Au-diazo-ZnO nanorods, ZnO-IIP and ZnO-NIP. Figure 5 shows the survey spectra for the bare gold electrode, Au/diazo-COOH, Au-diazo-ZnO nanoparticles, Au-diazo-ZnO nanorods, ZnO-IIP and ZnO-NIP. Firstly, comparison of the XPS spectrum of bare gold electrode with Au/diazo-COOH confirms the effective reductive electrografting reaction of the diazonium salt on the gold electrode. Indeed, it is noticed the presence of characteristic signals of the diazo-COOH, namely C1s and N1s peaks centered at 285 eV and 400 eV, respectively ( Figure 5). It should be noted that the surface-grafted layer seems to be relatively thin (less than 10 nm); this was also confirmed by the atomic percentages shown in Table 1. The atomic percentage of gold decreased from 41.6% for the bare gold electrode to 35.8% after electrografting of the diazonium salt, with a small increase in the percentage of carbon (grafting of the diazonium salt while keeping the conductive activity of the electrode).  Firstly, comparison of the XPS spectrum of bare gold electrode with Au/diazo-COOH confirms the effective reductive electrografting reaction of the diazonium salt on the gold electrode. Indeed, it is noticed the presence of characteristic signals of the diazo-COOH, namely C1s and N1s peaks centered at 285 eV and 400 eV, respectively ( Figure 5). It should be noted that the surface-grafted layer seems to be relatively thin (less than 10 nm); this was also confirmed by the atomic percentages shown in Table 1. The atomic percentage of gold decreased from 41.6% for the bare gold electrode to 35.8% after electrografting of the diazonium salt, with a small increase in the percentage of carbon (grafting of the diazonium salt while keeping the conductive activity of the electrode). After the deposition of ZnO nanoparticles by spin-coating, we have the appearance of a peak at 1022 eV which corresponds to the Zn2p 3/2 of ZnO ( Figure 5), with an atomic percentage of 10.7%. After the ZnO growth step, the atomic percentages of zinc and oxygen increase (27.4% and 38.5%, respectively). We can also note a decrease in the peaks corresponding to the Au4f 7/2 and C1s signals ( Figure 5), which confirms the success of the ZnO nanorods growth step.

Surface Analysis by XPS
Comparison of the XPS spectra of ZnO-IIP and ZnO-NIP ( Figure 5) indicates the grafting of a polypyrrole film on the surface of the electrode. However, mercury is present only in the case of the IIP material, which indicates that the template has been well trapped in the polymeric matrix. The atomic percentage of Zn for NIP and IIP decreased due to screening ZnO with polypyrrole ( Table 1).
As per the thickness of the aryl layer, it was determined using the Beer-Lambert exponential decay of the intensity of bare gold: where I is the peak intensity of Au4f 7/2 core level signal after coating, I • is the intensity for the bare gold, d is the aryl thickness, λ the mean free path of Au4f 7/2 in the aryl layer and θ the analysis angle (here, θ = 0, and cosθ = 1). The thickness is estimated to be 0.43 nm by considering λ to be around 4.1 nm, which, for a monolayer of aryl groups given the fact that the size of length of benzoic acid, is around 0.49 nm.
As will be shown below by AFM, ZnO NPs are large compared to the sampling depth (~2-10 nm) in XPS. For this reason, no gold is detected for Au-diazo-ZnO NPs. After growth of the NRs, Au4f doublet is detected again, which signifies that deposition of the NRs probably induces partial leaching of ZnO NPs, hence the escape of Au4f electrons between the NRs and their detection in XPS.

Topographical Characterization
The atomic force microscopy (AFM) measurements have been carried out in order to monitor the topographic changes after each surface functionalization step. The evidence of nanorod growth is not the purpose herein, that being clearly shown in a previous work [47]. Therefore, we can use standard tapping mode AFM probes instead of high aspect ratio ones. The robustness of the former allows comparative investigation of several samples' surfaces without changing the probe. Two-dimensional AFM images of (5 × 5) µm 2 have thus been achieved on the same sample, before (bare gold substrate) and after each functionalization step (diazonium salt, ZnO nanoparticles, ZnO nanorods, then IIP) and finally after extraction of the Hg 2+ template ( Figure 6). After the deposition of ZnO nanoparticles by spin-coating, we have the appearance of a peak at 1022 eV which corresponds to the Zn2p3/2 of ZnO ( Figure 5), with an atomic percentage of 10.7%. After the ZnO growth step, the atomic percentages of zinc and oxygen increase (27.4% and 38.5%, respectively). We can also note a decrease in the peaks corresponding to the Au4f7/2 and C1s signals ( Figure 5), which confirms the success of the ZnO nanorods growth step.
Comparison of the XPS spectra of ZnO-IIP and ZnO-NIP ( Figure 5) indicates the grafting of a polypyrrole film on the surface of the electrode. However, mercury is present only in the case of the IIP material, which indicates that the template has been well trapped in the polymeric matrix. The atomic percentage of Zn for NIP and IIP decreased due to screening ZnO with polypyrrole ( Table 1).
As per the thickness of the aryl layer, it was determined using the Beer-Lambert exponential decay of the intensity of bare gold: where I is the peak intensity of Au4f7/2 core level signal after coating, I° is the intensity for the bare gold, d is the aryl thickness, λ the mean free path of Au4f7/2 in the aryl layer and θ the analysis angle (here, θ = 0, and cosθ = 1). The thickness is estimated to be 0.43 nm by considering λ to be around 4.1 nm, which, for a monolayer of aryl groups given the fact that the size of length of benzoic acid, is around 0.49 nm.
As will be shown below by AFM, ZnO NPs are large compared to the sampling depth (~2-10 nm) in XPS. For this reason, no gold is detected for Au-diazo-ZnO NPs. After growth of the NRs, Au4f doublet is detected again, which signifies that deposition of the NRs probably induces partial leaching of ZnO NPs, hence the escape of Au4f electrons between the NRs and their detection in XPS.

Topographical Characterization
The atomic force microscopy (AFM) measurements have been carried out in order to monitor the topographic changes after each surface functionalization step. The evidence of nanorod growth is not the purpose herein, that being clearly shown in a previous work [47]. Therefore, we can use standard tapping mode AFM probes instead of high aspect ratio ones. The robustness of the former allows comparative investigation of several samples' surfaces without changing the probe. Two-dimensional AFM images of (5 × 5) μm 2 have thus been achieved on the same sample, before (bare gold substrate) and after each functionalization step (diazonium salt, ZnO nanoparticles, ZnO nanorods, then IIP) and finally after extraction of the Hg 2+ template ( Figure 6). The AFM images ( Figure 6A-F) exhibit a granular-like aspect and seem different enough in both shape and size. The grafting of diazonium salt ( Figure 6B) leads to the formation of a layer veiling the granular appearance. The latter is again visible ( Figure 6C) after grafting the ZnO nanoparticles, with higher contrast than for the bare gold surface ( Figure 6A). Figure 6D, which corresponds to ZnO nanorod growth, exhibits the largest size of surface singularities, which seems reduced by the electropolymerization of IIP ( Figure 6E). The extraction of template from the IIP film ( Figure 6F) leads to a different aspect that is probably due to the local rearrangement of the IIP. The AFM images ( Figure 6A-F) exhibit a granular-like aspect and seem different enough in both shape and size. The grafting of diazonium salt ( Figure 6B) leads to the formation of a layer veiling the granular appearance. The latter is again visible ( Figure 6C) after grafting the ZnO nanoparticles, with higher contrast than for the bare gold surface ( Figure 6A). Figure 6D, which corresponds to ZnO nanorod growth, exhibits the largest size of surface singularities, which seems reduced by the electropolymerization of IIP ( Figure 6E). The extraction of template from the IIP film ( Figure 6F) leads to a different aspect that is probably due to the local rearrangement of the IIP.
Despite the usefulness of AFM images, qualitative description remains limited due to the high contrast that may exist in the presence of multiple defects with high amplitudes. Measurements should therefore be considered from a quantitative point of view by considering relevant roughness statistical parameters, such as rms roughness (Sq) and peak-valley height (Sp-Sv). Values of these two parameters are presented in Figure 7. The AFM images ( Figure 6A-F) exhibit a granular-like aspect and seem different enough in both shape and size. The grafting of diazonium salt ( Figure 6B) leads to the formation of a layer veiling the granular appearance. The latter is again visible ( Figure 6C) after grafting the ZnO nanoparticles, with higher contrast than for the bare gold surface ( Figure 6A). Figure 6D, which corresponds to ZnO nanorod growth, exhibits the largest size of surface singularities, which seems reduced by the electropolymerization of IIP ( Figure 6E). The extraction of template from the IIP film ( Figure 6F) leads to a different aspect that is probably due to the local rearrangement of the IIP.
Despite the usefulness of AFM images, qualitative description remains limited due to the high contrast that may exist in the presence of multiple defects with high amplitudes. Measurements should therefore be considered from a quantitative point of view by considering relevant roughness statistical parameters, such as rms roughness (Sq) and peak-valley height (Sp-Sv). Values of these two parameters are presented in Figure 7. As expected, each surface modification leads to a noticeable change in roughness parameters. The most important one occurs after diazonium salt grafting. This first step of functionalization led to an increase in (Sq) and (Sp-Sv) parameter values by a factor of 5.7 and 7.8, respectively. The increase in roughness continued with the grafting of ZnO NPs, but to a lesser extent, by a factor of As expected, each surface modification leads to a noticeable change in roughness parameters. The most important one occurs after diazonium salt grafting. This first step of functionalization led to an increase in (Sq) and (Sp-Sv) parameter values by a factor of 5.7 and 7.8, respectively. The increase in roughness continued with the grafting of ZnO NPs, but to a lesser extent, by a factor of 1.7 for (Sq), while the (Sp-Sv) value remains relatively stable. The growth of the nanorods generated a variation in the two considered parameters by a factor of 2.9 and 2.4, respectively. The electro-polymerization of IIP produced a coating effect resulting in a significant decrease in the values of the two roughness parameters by a factor of 4.7 for (Sq) and 3.3 for (Sp-Sv). These two parameters decreased by a factor of around 1.5 only, after the analyte extraction step. This slight decrease is probably due to the local rearrangements which occur once the Hg 2+ ions are extracted from the polymer matrix.
The corresponding 2D images of ZnO NPs (Figure 8a) and ZnO NRs (Figure 8b), show the dimensions of the ZnO NPs and NRs. The diameter of the NPs is between 70 and 80 nm, while that of the obtained NRs is in the range of 300-350 nm (the same diameter was obtained in our previous work [47]). Figure 8c,d represents 2D and 3D images for the same analysis area on a surface of (80 × 80 µm 2 ). The growth of NRs is generally homogeneous, and the height after growth of NRs reaches 325 nm (Figure 8d).

Extraction Time in EDTA and Incubation Time in Mercury Solutions
Ethylenedinitrilo tetraacetic acid (EDTA) has been used in this work to extract mercury ions from the ZnO-IIP sensor. EDTA is an acid that is difficult to biodegrade. It is a stable chelator used in several applications, such as the medical field and the nuclear industry, but it is also used to adsorb heavy ions through functional groups, avoiding their precipitation [78][79][80]. Experiments (results not shown here) indicate that the optimum extraction was obtained by incubating the IIP in EDTA 0.1 M, for 20 min, and by then rinsing the films for 10 min in ultra-pure water.
The film obtained after extraction was incubated in a mercury solution (10 −4 M) for different periods of time. Results presented in Figure 9 indicate that the number of Hg 2+ ions captured by the sensor increased only slightly between 20 and 40 min, compared to the sharp increase between 5 and 20 min of incubation. For practical reasons, the incubation time was set to 20 min.

Extraction Time in EDTA and Incubation Time in Mercury Solutions
Ethylenedinitrilo tetraacetic acid (EDTA) has been used in this work to extract mercury ions from the ZnO-IIP sensor. EDTA is an acid that is difficult to biodegrade. It is a stable chelator used in several applications, such as the medical field and the nuclear industry, but it is also used to adsorb heavy ions through functional groups, avoiding their precipitation [78][79][80]. Experiments (results not shown here) indicate that the optimum extraction was obtained by incubating the IIP in EDTA 0.1 M, for 20 min, and by then rinsing the films for 10 min in ultra-pure water.
The film obtained after extraction was incubated in a mercury solution (10 −4 M) for different periods of time. Results presented in Figure 9 indicate that the number of Hg 2+ ions captured by the sensor increased only slightly between 20 and 40 min, compared to the sharp increase between 5 and 20 min of incubation. For practical reasons, the incubation time was set to 20 min. Appl. Sci. 2020, 10, x 14 of 23

Electrochemical Sensing of Mercury
SWV was used to investigate the electrochemical response of the ZnO/Hg 2+ -IIP sensor towards mercury in solution at different concentrations. Sensor response was recorded, via the characteristic peak current, for concentrations between 10 −12 and 10 −3 M (Figure 10).
The sensor limit of detection (LOD) was determined from the lowest concentration that gave a visible signal [81,82], instead of the method based on the signal to noise ratio of 3, as usually done. Here, the obtained LOD of 1 pM is much lower than the WHO maximum permissible limit in drinking water of (1 μg.L -1 ).  To confirm the performance of the sensor, this Hg 2+ detection step was conducted on three different ZnO/Hg 2+ -IIP, and the average calibration curve is shown in Figure 11 (variations of output signal of electrochemical IIP-based sensor versus Hg 2+ concentrations).

Electrochemical Sensing of Mercury
SWV was used to investigate the electrochemical response of the ZnO/Hg 2+ -IIP sensor towards mercury in solution at different concentrations. Sensor response was recorded, via the characteristic peak current, for concentrations between 10 −12 and 10 −3 M (Figure 10).
The sensor limit of detection (LOD) was determined from the lowest concentration that gave a visible signal [81,82], instead of the method based on the signal to noise ratio of 3, as usually done. Here, the obtained LOD of 1 pM is much lower than the WHO maximum permissible limit in drinking water of (1 µg.L −1 ).
To confirm the performance of the sensor, this Hg 2+ detection step was conducted on three different ZnO/Hg 2+ -IIP, and the average calibration curve is shown in Figure 11 (variations of output signal of electrochemical IIP-based sensor versus Hg 2+ concentrations).

Electrochemical Sensing of Mercury
SWV was used to investigate the electrochemical response of the ZnO/Hg 2+ -IIP sensor towards mercury in solution at different concentrations. Sensor response was recorded, via the characteristic peak current, for concentrations between 10 −12 and 10 −3 M ( Figure 10).
The sensor limit of detection (LOD) was determined from the lowest concentration that gave a visible signal [81,82], instead of the method based on the signal to noise ratio of 3, as usually done. Here, the obtained LOD of 1 pM is much lower than the WHO maximum permissible limit in drinking water of (1 μg.L -1 ). To confirm the performance of the sensor, this Hg 2+ detection step was conducted on three different ZnO/Hg 2+ -IIP, and the average calibration curve is shown in Figure 11 (variations of output signal of electrochemical IIP-based sensor versus Hg 2+ concentrations).  The dissociation constant Kd is an important parameter that determines the degree of affinity between the polymer and the template. Different analytical models were tested to fit the experimental data of the electrochemical measurements. The most appropriate one was a combination of the one site model and Hill model ( Figure 11). The choice of this model makes it possible to take into consideration the mechanical restructuring of polymers at the nanometric scale [83].
is the output sensor's signal response (current) for a given concentration (C) of Hg 2+ . Kd1, and Kd2 are the first and second dissociation constants, A1 and A2 are empiric constants, and α is an empiric ponderation exponent. The values obtained for Kd1 and Kd2 were in the order of (7.9 ± 3.6) mM and (38.1 ± 9.2) pM, respectively. The value of Kd2 is consistent with that linked to the binding of Hg 2+ to the Cys4 site in CP-CCCC peptide, which shows that the IIP can imitate biological entities in terms of affinity for heavy ions [71,83]. The α value of 0.06 ± 0.002, indicates positive cooperative interactions between Hg 2+ ions and other binding sites [71].
The sensor sensitivity, computed from the slope at the origin of current/concentration curves (i.e., at lower concentrations), was in the order of 0.692 ± 0.034 μA.pM -1 . This value shows that this LCys-PPy sensor prepared on Au-ZnO to capture mercury is very sensitive compared to several sensors intended to detect heavy metals [84][85][86]. In addition to these interesting metrological features, we also tested the selectivity of the sensor by comparing the recognition of the targeted Hg 2+ to lead, cadmium and copper ions. The rationale for this set of ions is the following: Cd 2+ has an ionic radius close to that of Hg 2+ ; Cu 2+ has a lower ionic radius than Hg 2+ , whereas Pb 2+ is a larger ion than the targeted Hg 2+ .
We first incubated the extracted film in a solution containing an ion of the chosen ions with a concentration of 10 −4 M ( Figure 12A). The second step was to mix several ions with Hg 2+ with the same concentration of each ion ( Figure 12B, black curve) and conduct another test in a solution containing Pb 2+ and Hg 2+ with a concentration of Pb 2+ greater than 100 times that of Hg 2+ ( Figure 12B, red curve). The dissociation constant K d is an important parameter that determines the degree of affinity between the polymer and the template. Different analytical models were tested to fit the experimental data of the electrochemical measurements. The most appropriate one was a combination of the one site model and Hill model ( Figure 11). The choice of this model makes it possible to take into consideration the mechanical restructuring of polymers at the nanometric scale [83].
where y (C) is the output sensor's signal response (current) for a given concentration (C) of Hg 2+ . K d1 , and K d2 are the first and second dissociation constants, A 1 and A 2 are empiric constants, and α is an empiric ponderation exponent. The values obtained for K d1 and K d2 were in the order of (7.9 ± 3.6) mM and (38.1 ± 9.2) pM, respectively. The value of K d2 is consistent with that linked to the binding of Hg 2+ to the Cys4 site in CP-CCCC peptide, which shows that the IIP can imitate biological entities in terms of affinity for heavy ions [71,83]. The α value of 0.06 ± 0.002, indicates positive cooperative interactions between Hg 2+ ions and other binding sites [71]. The sensor sensitivity, computed from the slope at the origin of current/concentration curves (i.e., at lower concentrations), was in the order of 0.692 ± 0.034 µA.pM −1 . This value shows that this LCys-PPy sensor prepared on Au-ZnO to capture mercury is very sensitive compared to several sensors intended to detect heavy metals [84][85][86]. In addition to these interesting metrological features, we also tested the selectivity of the sensor by comparing the recognition of the targeted Hg 2+ to lead, cadmium and copper ions. The rationale for this set of ions is the following: Cd 2+ has an ionic radius close to that of Hg 2+ ; Cu 2+ has a lower ionic radius than Hg 2+ , whereas Pb 2+ is a larger ion than the targeted Hg 2+ .
We first incubated the extracted film in a solution containing an ion of the chosen ions with a concentration of 10 −4 M ( Figure 12A). The second step was to mix several ions with Hg 2+ with the same concentration of each ion ( Figure 12B, black curve) and conduct another test in a solution containing Pb 2+ and Hg 2+ with a concentration of Pb 2+ greater than 100 times that of Hg 2+ ( Figure 12B Figure  12A) shows that ZnO-IIP is specific for Hg 2+ .
The competition tests ( Figure 12B) were carried out by incubating the extracted film in a solution containing equal concentrations (10 −4 M) of Cu 2+ , Hg 2+ and Pb 2+ ions. The tests have shown that in the presence of Cu 2+ and/or Pb 2+ , the extracted film detects Hg 2+ highly selectively. One can note the existence of some competition, since in Figure 12A, the height of the Hg 2+ peak is 106 µ A but slightly decreases to 94 µ A in the presence of both Cu 2+ and Pb 2+ at the same concentration of 10 −4 M ( Figure 12B). If the Hg2+ concentration is decreased to 10 −6 M, one can note a decrease in the height of the peak to 77 µ A in the presence of 10 -4 M Pb 2+ ( Figure 12B). Quantitatively speaking, there is a clear selectivity of the sensor to Hg 2+ ; indeed, competition of other ions to the artificial receptors occurs but remains marginal. This is very important to prove, because in most fundamental studies, selectivity tests are usually conducted in single metal ion solutions and not in mixtures [87].
In a mixture containing 10 −4 M of Pb 2+ and 10 −6 M of Hg 2+ , we note that despite the low concentration of mercury with the presence of other ions, the sensor remains specific to Hg 2+ , hence the importance of shaping the receptor sites with this template. This also shows that there is no competitive chemisorption of Pb 2+ which hampers the response of the sensor to the Hg 2+ ions.
The performance of the IIP sensors designed so far is compared to that of a number of other sensors described in the literature (Table 2). Embedding ZnO nanorods in the IIP film provided a unique LOD, as low as 1 pM.  Figure 12A) shows that ZnO-IIP is specific for Hg 2+ .
The competition tests ( Figure 12B) were carried out by incubating the extracted film in a solution containing equal concentrations (10 −4 M) of Cu 2+ , Hg 2+ and Pb 2+ ions. The tests have shown that in the presence of Cu 2+ and/or Pb 2+ , the extracted film detects Hg 2+ highly selectively. One can note the existence of some competition, since in Figure 12A, the height of the Hg 2+ peak is 106 µA but slightly decreases to 94 µA in the presence of both Cu 2+ and Pb 2+ at the same concentration of 10 −4 M ( Figure 12B). If the Hg 2+ concentration is decreased to 10 −6 M, one can note a decrease in the height of the peak to 77 µA in the presence of 10 −4 M Pb 2+ ( Figure 12B). Quantitatively speaking, there is a clear selectivity of the sensor to Hg 2+ ; indeed, competition of other ions to the artificial receptors occurs but remains marginal. This is very important to prove, because in most fundamental studies, selectivity tests are usually conducted in single metal ion solutions and not in mixtures [87].
In a mixture containing 10 −4 M of Pb 2+ and 10 −6 M of Hg 2+ , we note that despite the low concentration of mercury with the presence of other ions, the sensor remains specific to Hg 2+ , hence the importance of shaping the receptor sites with this template. This also shows that there is no competitive chemisorption of Pb 2+ which hampers the response of the sensor to the Hg 2+ ions.
The performance of the IIP sensors designed so far is compared to that of a number of other sensors described in the literature (Table 2). Embedding ZnO nanorods in the IIP film provided a unique LOD, as low as 1 pM.

Conclusions
A new electrochemical sensor based on a biomimetic ion-imprinted polymer for the detection of mercury (II) traces was prepared on a gold electrode with a strategy based on the combination of the surface modification with a diazonium salt and the growth of ZnO nanorods followed by electropolymerization of the pyrrole in the presence of the template (Hg 2+ ) and L-cysteine as crosslinker. These two steps (diazonium salt and ZnO nanorods) permitted us to increase the performance of the final sensor (Au-diazo-ZnO/IIP) in terms of selectivity and limit of detection (LOD), which is in the picomolar range. This is proven by investigating the performance of an IIP prepared directly on a bare gold electrode, without any surface modification and in the absence of any ZnO nanorods.
The fit of the electrochemical results obtained by the one site/Hill model allowed us to determine the dissociation constants [K d1 = (7.89 ± 3.63) mM and K d2 = (38.10 ± 9.22) pM] as well as the sensitivity, which was equal to 0.692 ± 0.034 µA.pM −1 .
The competitivity and selectivity tests were undertaken using several ions such as Cu 2+ , Pb 2+ and Cd 2+ , taking into account the ionic radius and their concentration in an aqueous medium. IIP showed excellent ability to distinguish between mercury and different competitive ions even at low concentrations. Au-diazo-ZnO/IIP can therefore be used for environmental applications to detect mercury in the laboratory or for field analysis using a portable potensiostat.