3.1. Characterization of ChPBN
Morphological studies using SEM and TEM provided visual information about the size, shape and the distribution of PB nanoparticles within the chitosan matrix. The SEM image in
Figure 2a shows the morphology of the ChPBN thin film. Here, the SEM image clearly shows the tubular network structure of ChPBN. The tubular structure also allows the formation of pores of different magnitudes. The TEM image in
Figure 2b shows the drop-casted ChPBN. The TEM image also confirmed the network structure that allows porosity. During the preparation of the nanocomposite, acidic chitosan solution and ferricyanide solution were mixed. Due to the electrostatic attraction between cationic chitosan chains and anionic ferricyanide ions, the PB nanocube formation took place at pre-adhered [Fe(CN)
6]
−3 sites on the chitosan polymer networks. Considering that the surface energy of PB nanoparticles is typically higher than that of polymers, it is reasonable to assume that the chitosan matrix covers the PB nanoparticles in the inset of
Figure 2b. Therefore, ChPBN shows a highly porous network structure that allows the penetration of the PVC-based ion-selective membrane, which is the outer layer of the SS-Na
+ISE. The inter-penetration of the two layers contributes to the high sensitivity and the stability of the potential response.
Figure 2c shows the TEM image of PB nanocubes prepared without chitosan matrix. The sizes of the nanocubes were 40 ± 5 nm. Morphological studies notably conclude that there is the entrapment of the PB nanoparticle and thus the formation of a mesoporous composite with chitosan.
Figure 3a shows the FTIR spectra of the ChPBN, as well as chitosan (control) and PB (control). The PB (control) shows a strong and sharp peak for the stretching vibration of C≡N group at 2050 cm
−1. The two peaks at 697 and 589 cm
−1 are assigned to the stretching vibration of Fe–(C≡N) and Fe–C. All these peaks are the signature of metal hexacyanoferrates. In the FTIR spectrum of the chitosan sample, the strong peaks at 3246 cm
−1 and at 2887 cm
−1 of the chitosan (control) correspond to O–H and C–H stretching vibrations, respectively. The band at 1617 cm
−1 is attributed to N–H bonds from primary amines. The low-intensity peaks at 1411 cm
−1 and 1062 cm
−1 correspond to the bending of the C–H bonds of the methyl groups and the stretching vibrations of the C=O bonds, respectively. All the peaks are in accordance with [
58]. The FTIR spectrum of ChPBN shows signature peaks of both PB and chitosan, indicating the successful synthesis of the nanocomposite. For ChPBN, there was a significant peak around 1657 cm
−1 corresponding to the formation of an imine bond (C=N). This is Schiff’s base structure formed by the reaction between the amino groups of chitosan and the aldehyde groups of glutaraldehyde (the cross-linking agent) [
59]. The cross-linking actually occupies the amine groups of chitosan making the composite less susceptible to water intake. Only 2% swelling was observed for the thin film of ChPBN after 30 h of complete submergence in water and no further swelling was observed after seven days in water. The XPS spectra (
Figure 3b) of ChPBN and chitosan (control) films show that the main peaks of N1s, O1s and C1s exist for both samples, whereas the Fe 2p peak appears only in ChPBN. A detailed analysis of the Fe 2p area revealed two peaks with lower bond energy at 706.7 eV (Fe 2p
3/2) and 719.5 eV (Fe 2p
1/2). These two peaks correspond to the oxidation states of Fe(II) and Fe(III), respectively. The chemical formula of the insoluble form of PB can be assumed as Fe
4III[Fe
II(CN)
6]
3 [
60].
Figure 3c shows The EDS spectrum from ChPBN film. The EDS elemental analysis confirms the presence of Fe, C and N elemental signatures. The absence of the K elemental signal proves the successful removal of excess potassium from the ChPBN. The signature peaks for Au represent the gold surface coating for SEM imaging. All the above morphological and chemical characterization favors the claim of the successful synthesis of the nanocomposite.
3.2. Electrochemical Characterization of Electrodes with ChPBN Solid Contact
Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) elucidated the electrochemical properties of the electrodes that employ ChPBN as the solid contact. CV of the SPE/ChPBN sample in the 0.1 M NaCl solution within a potential ranging from −1 V–+1 V at the scan rate of 100 mV/s revealed the redox-capacitive behavior of the nanocomposite layer. Control samples include bare SPE, the SPE coated with PB nanoparticles without chitosan (SPE/PB) and the SPE coated with a neat chitosan layer (SPE/Ch) (
Figure 4a). The voltammogram of SPE/Ch did not have any redox peak, whereas SPE/PB showed redox couples at 141 mV and 826 mV. These two redox couples correspond to the conversion from Prussian white to Prussian blue and from Prussian blue to Berlin green, respectively (with increasing voltage) [
61]. For SPE/ChPBN, a shift in redox peaks towards negative potential (−567 mV; from Prussian white to Prussian blue) was observed with a prominent peak separation for reduction and oxidation compared to SPE/PB. In addition, very faint redox peaks for Prussian blue and Berlin Green conversion were observed in the experimental condition. Here, the chitosan layer on PB can cause the unequal distribution of ions compared to the case that all PB nanoparticles are exposed to the solution [
62]. The chitosan polymer network and PB nanoparticles acted as a cationic matrix and anionic sites in ChPBN, respectively. The high capacitive current of chitosan eventually masked the redox peaks of Prussian blue and Berlin Green conversion of PB. The generation of Donnan potential in the ChPBN can explain this phenomenon.
Figure 4a shows that the capacitive current of the bare electrode is negligible. SPE/ChPBN has higher capacitive current than SPE/PB. The current outputs of SPE/ChPBN and SPE/Ch are very similar in the capacitive region (the flat regions between −0.2 V and 1 V). These indicate that chitosan contributes to the amount of capacitive charge in the nanocomposite film. The scan rate modulation was performed to predict the electron transfer mechanism for the modified electrode.
Figure 4b shows the effect of scan rate on the CV characteristics of the SPE/ChPBN electrode. Here, the anodic (
Ipa) and cathodic (
Ipc) peak currents increased with increasing scan rate. The anodic peaks shifted by +87.6 mV when the scan rate increased from 10–100 mV/s towards positive potential. The increase in the current response with the increasing scan rate represents the reversible redox behavior of ChPBN composite film. It indicates that the chitosan coating is not affecting the electron transfer or no such chemical reaction is associated with the redox couple of PB. The plots of the Ip vs. v
1/2 (
Figure S1; Supplementary Materials) for both the anodic and cathodic half showed linearity with scan rate. This indicates fast electron transfer at the ChPBN-electrode interface. The solid contact coating on the electrode is not hindering the electron transfer mechanism in any case. The EIS studies were performed to characterize the resistance and capacitance contribution of ChPBN towards the modified electrode.
Figure 4c shows Nyquist plots from the EIS spectrum (complex plane plots of −
Z” vs.
Z’) in 0.1 M NaCl solution at scanning frequencies from 100 kHz–100 mHz with a modulation amplitude of 0.01 V. The fitting based on the equivalent circuit models (NOVA 2.0.2 software; Metrohm Autolab B.V. Utrecht, The Netherlands) was used to extract capacitance and interface resistance values. Here, the impedance spectra for bare SPE, SPE/PB, SPE/CH and SPE/ChPBN are following a nearly vertical (90°) line at lower frequencies (capacitive line). The absence of the high frequency semicircle (the inset of
Figure 4c) indicates very fast ion and electron transfers at the ChPBN/solution interfaces [
11]. Following Boback’s protocol [
11], we assumed the electrochemical electrode as an equivalent circuit with the interfacial resistance and the capacitance connected as a series circuit. The capacitance (
C) can be determined from the
Z” value of the imaginary part of impedance for the lowest frequency using Formula (1):
where
f is frequency and
Z” is the value of the imaginary part of impedance.
Table 1 represents the capacitance values determined from the four electrodes. The effect of film thickness on the capacitance values of the SPE/ChPBN electrodes was also studied and represented in the Nyquist plots in
Figure 4d. Here, the loading volume to make the drop-casted electrodes varied; larger loading generated thicker film. After applying Boback’s protocol, the capacitance values decreased about 20 ± 2 µF when the loading increased from 10–30 μL. The variance in the capacitance corresponds to 13% of the droplet capacitance from the 10 µL sample. Here, we used 10 µL of loading volume in the remainder of the article.
Electrochemical behaviour was studied to explore the capacitance and resistance of the ChPBN solid contact of SPE/ChPBN/Na
+ISE. The efficiency of charge exchange was evaluated by chronopotentiometry (CP) and EIS measurements. CV could not be used due to the high resistance of the PVC-based ion-selective membrane. CP was performed by consecutive application of anodic and cathodic currents (
Figure 5a). For both anodic and cathodic polarization, flipping caused the instant potential jumps, followed by temporary potential drift. From the slope of the linear part of the potential drift (immediately after each potential jump), the redox capacitance can be estimated by:
where
I is applied current and
C is capacitance [
10]. It is notable that near zero currents (±1 nA) were too small to monitor polarization effects. Here, CP was performed at relatively high current (±100 nA) for 60 s of anodic and cathodic cycles to trigger potential jumps. Curves i and ii in
Figure 5a compare the potential jumps and drifts for the SPE/Na
+ISE and SPE/ChPBN/Na
+ISE electrodes during the periodic polarization cycles. Considering that the bulk resistance of the PVC-based membrane dominates the total resistance of the electrode, the membrane resistance (
Rm) can be estimated by:
where
E represent the amount of the instant potential change due to the applied current
I [
10]. According to Equation (3), the membrane resistance was 372.31 kΩ for SPE/Na
+ISE and 89.41 kΩ for SPE/ChPBN/Na
+ISE. The amount of potential jumps and ohmic drops decreased when the ChPBN layer is incorporated in the ISE. In addition, the effect of conditioning the SPE/ChPBN/Na
+ISE electrode in the 0.1 M NaCl solution for 1 h is shown in Curves iii (before) and iv (after conditioning) of
Figure 5a. Here, alternating near zero currents (±1 nA) were applied with a 60-s interval, whereas the current could not cause potential drop. The conditioning clearly improved the potential stability.
EIS spectra were also recorded for Na
+ISE with and without the ChPBN solid contact for a frequency range from 1 mHz–1 mHz (in 0.1 M NaCl solution; 0.01 V modulation amplitude).
Figure 5b shows that all EIS spectra have semi-circular high frequency patterns and linear low frequency output. Here, the broadening of the semicircles in the Nyquist plots reflects the increase in resistance by double layer depositions. The higher resistance hinders efficient charge transfer. More explicitly, the SPE/ChPBN/Na
+ISE electrode with the solid contact can be represented by an equivalent circuit model. The experimental data fit into a simple Randles circuit to extract the components [
63]. The Randles circuit (
Figure 5b, inset) represents the solution resistance (
Rs), charge transfer resistance (
Rct) at the membrane/solution interface, the double layer capacitance (
Cdl) and the finite-length Warburg diffusion impedance (
Zw).
Table 2 summarizes the estimated potential drift
and capacitance (
C) values from SPE/Na
+ISE and SPE/ChPBN/Na
+ISE under various operating conditions. The capacity of the ChPBN layer was calculated to be 154.5 µF, which in turn contributed to the increased capacitance from 457 µF (SPE/Na
+ISE) to 737 µF (SPE/ChPBN/Na
+ISE). The excess capacity may come from the PVC-based membrane of Na
+ISE. The capacitance values are comparable to conducting polymer solid contacts, such as PEDOT(PSS) of 162 µF [
12], whereas the value is higher than graphene, carbon black, fullerene and carbon nanotubes. The calculated capacity values are better than those measured for TCNQ of 154 µF [
32], carbon black of 51 µF [
64], SWCNT of 59 µF [
17], graphene sheet of 91 µF [
22] solid contact electrodes. The high capacitive nature of ChPBN provides long-term potential stability and thus makes it suitable as a solid contact for all-solid-state ion-selective electrodes.
Table S1 in the Supporting Information summarizes extracted parameters from 21 different systems from previous literature, clearly showing the comparison.
3.3. Potential Response and Stability of SPE/ChPBN/Na+ISE
The solid-state sodium selective electrodes (SPE/ChPBN/Na
+ISE) were tested for Na
+-selective response.
Figure 6a shows the electrode potential of SPE/ChPBN/Na
+ISE with respect to [Na
+] ranging from 10
−7 M–1 M. The sodium solutions were supplemented with 80 mM ionic strength adjuster (ISA) solution. The slope of the fitting to the linear range (10
−4–1 M) was 52.4 ± 0.4 mV/decade. The ChPBN layer was effective in achieving stable potential, as evidenced by the time-dependent measurement in
Figure 6b. Here, the output potential significantly stabilized for SPE/ChPBN/Na
+ISE when compared to that of SPE/Na
+ISE. The potential stability was also monitored at a fixed primary ion concentration ([Na
+] = 130 mM) for 20 h (10 cycles of 2-h measurements; no interval between the cycles).
Figure 6c shows the drift of the output potential of the SS-Na
+ISE (SPE/ChPBN/Na
+ISE). The potential drift of SPE/ChPBN/Na
+ISE was 3.3 µV/h over the first 4 h (inset), whereas the drift in the later time was noticeably lower. The average potential drift over 20 h was 1.3 µV/h (3.61 × 10
−4 µV/s), which is better than the potential drift reported for the solid contact electrodes based on porous carbon of 212 µV/s [
23], graphene sheet of 55 µV/s [
22], reduced graphene oxide of 12.8 µV/s [
20], SWCNT of 85 µV/s [
17] and TCNQ of 9.2−11.1 µV/s [
32]. The reduced potential drift for ChPBN is very much comparable to polypyrrole film of 9.23 × 10
−3 µV/s [
8], polyaniline doped-POT of 9.23 × 10
−3 µV/s [
10], three-dimensionally-ordered macroporous (3DOM) carbon of 3.25 × 10
−3 µV/s [
16], colloid-imprinted mesoporous (CIM) carbon of 3.61 × 10
−4 µV/s [
24] and carbon black 4.19 × 10
−3 µV/s [
65] (see
Table S1 in the Supplementary Materials). The enhanced potential stability also supports the ChPBN as a suitable material as a solid contact in ISEs.
The [Na
+] detectability of SPE/ChPBN/Na
+ISE in the presence of background ions was tested in sodium phosphate buffer (NaPB, 10 mM, pH 7.4) with added NaCl with the concentration 10–160 mM and potential response measured at zero current.
Figure 7 shows the performance of SPE/ChPBN/Na
+ISE in sodium phosphate buffer. Here, we could confirm the near-Nernstian slope of 58 mV/(log
10[Na
+]). The standard curve in the
Figure 6a shows the current response of the sensor only against the known concentration of added NaCl. The electrolyte background was adjusted with ammonium chloride/ammonium hydroxide ion strength adjuster solution. Therefore, the responses were detected only from Na ion activity. Furthermore, in
Figure 6a, the standard curve showed two different slopes within a broad range of NaCl from 10
−7–1 M. The lower concentration range (10
−7–10
−4 M) showed low potential difference with a lower slope value (sub-Nernstian response). For the higher concentration range (10
−4–1 M), the sensor showed a near-Nernstian slope. In
Figure 7, the current response was more prominent with a single slope of 58 mV. Here, the responses were measured in the presence of 10 mM sodium phosphate buffer as a background electrolyte. The NaPB of 10 mM strength actually retains 15 mM of Na
+ ions (10 mM Na
+ from dibasic and 5 mM Na
+ from monobasic salts of sodium), which is additionally calculated for each supplemented solution. Each supplemented sodium concentration (10–160 mM) had added sodium ions from the background buffer. The inset of
Figure 7 shows the actual measured values with the calculated sodium ion concentration.
3.4. Water Layer Test and Selectivity Test
When an SS-ISE is exposed to aqueous solutions for a long period of time, a thin water layer may form between the ISE and the solid state electrode. The water layer is essentially an isolated reservoir because the equilibration of ions with respect to the surrounding environment can be delayed by the diffusion. In the presence of interfering ions, the equilibrium can be further delayed. For example, when the water layer contains K
+ ions while the surrounding has Na
+ ions, it takes time to equilibrate by replacing the K
+ ions with the Na
+. This causes a drift and delayed response of the Na
+ISE. Based on this principle, a water layer test is devised [
64].
Figure 8a shows the time-dependent potentiometric response when SPE/Na
+ISE and SPE/ChPBN/Na
+ISE were submerged in a series of ionic solutions: (i) the primary ion solution ([Na
+] = 100 mM, [K
+] = 0 mM; initially submerged for 15 days), (ii) the interfering ion (K
+) solution ([Na
+] = 0 mM, [K
+] = 100 mM) and (iii) back to the primary ion solution. The ChPBN solid contact clearly improved the Na
+ISE’s response. This indicates that the water layer formation was reduced with the presence of the ChPBN layer.
The selectivity of SPE/ChPBN/Na
+ISE towards the primary ion (i.e., Na
+) was examined by a separate solution method where the selectivity coefficient (
) were obtained from the response potentials in the presence of different interfering ions [
66,
67]. The (
) can be defined as the direct function of the differences of the individual potentials for the primary ion, I, and the interfering ion, j, of the same ionic activity. The alkali metal or alkali earth metal chloride solutions were used in the separate solution method keeping the same ion activity as the primary ion (Na
+). In
Figure 8b, (
) represents the selectivity coefficients of interfering ions w.r.t. the Na
+ ion and were found to be 10
3–10
4-times lower in the case of interfering ions, which supports the selectivity towards the primary ion (i.e., Na
+).