**3. Results**

#### *3.1. NMR Studies*

To verify the effective formation of a host–guest complex between β-CD and MPT we mainly employed the NMR spectroscopy, already widely used to study inclusion complexes of CD [42–44].

The 1H NMR spectrum in D2O of a sample containing an equimolar amount of β-CD and MPT showed induced chemical shifts (Δδ) for all the protons of the host as well as for most of the gues<sup>t</sup> protons (Figure 2). The complexation-induced shifts (CIS) observed for the internal protons of β-CD (H3, H5) were remarkably larger than those displayed by the external ones (H-1, H-2, and H-4) (Table 1), thus indicating that the gues<sup>t</sup> likely interacts with the inner cavity of β-CD.

**Figure 2.** The 1H NMR spectra in D2O for solutions of (**a**) β-CD; (**b**) MPT; (**c**) equimolar amount of β-CD and MPT.

**Table 1.** The 1H-NMR chemical shifts (δ, ppm) for H protons of β-CD alone, MPT alone, and complexation induced shifts (CIS = <sup>δ</sup>complex – δfree) of equimolar amounts of them in D2O at 23 ◦C.


a The center of the multiplet was taken into account; b overlapping signals.

The small CIS observed for the ortho-protons of the phenyl moiety of MPT (0.04 ppm, Table 1) was also particularly significant, indicating that a part of the phenyl ring of MPT entered into the cavity of β-CD. It is worthy of note that no new peaks appeared in the spectrum, signifying that the inclusion of MPT in β-CD is a fast exchange process that takes place on the NMR timescale.

In order to gather information on the sites of binding we carried out a series of monodimensional ROESY-1D experiments [45] that provided only a small nuclear Overhauser effect (NOE) on the inner H3 proton of β-CD when ortho-H phenyl protons of MPT were irradiated.

The experimental observations collected up to this point were compatible with the inclusion structure of Figure 3, which was in rapid equilibrium with the two separate molecules. In fact, the large CIS exhibited by inner protons H3 and H5 of β-CD upon the addition of MPT clearly indicated a deep insertion of the host into the hydrophobic cavity of the β-CD. However, the lack of dipolar interactions between the ortho- and meta-protons of the aromatic ring of MPT and the H5 proton of β-CD excluded the complete insertion of the phenyl moiety into the β-CD cavity. On the other hand, the weak rotating-frame Overhauser effect (ROE) existing between the ortho-protons of MPT and the H3 proton of β-CD suggested that a partial insertion of the phenyl moiety of MPT occurred at the wide rim of β-CD. Finally, the upfield shift of the H-6 protons of the β-CD (see Table 1) could be justified by the partial protrusion of the gues<sup>t</sup> (tetrazole moiety) from the narrow rim of the β-CD (Figure 3) [46].

**Figure 3.** Proposed geometry for the inclusion of MPT into β-CD as deduced from the CIS and ROESY-1D experiments.

To investigate the strength of complexation we carried out a 1H NMR titration of MPT with β-CD [36]. The procedure adopted in the titration allowed us to operate with a constant gues<sup>t</sup> concentration ([MPT] = constant) during the whole experiment.

The binding isotherm relative to the ortho-protons of the phenyl moiety of MPT is depicted in Figure 4. The value of the stability constant β as log *K*a of the β-CD–MPT inclusion complex was calculated by the curve fitting method [36], using the commercial HypNMR2008 [37] program (details are given in SI) and was found to be equal to 2.93 M−<sup>1</sup> (*K*a = 851 <sup>M</sup>−1).

The 1:1 stoichiometry of the complexation adequately described the binding data obtained from the NMR titration and, on the other hand, the physically unrealistic binding parameters (some negative *K*s) when the 1:2 or 2:1 models were applied to the NMR titration. This confirmed that the 1:1 stoichiometry was dominating in the investigated concentration range. The 1:1 stoichiometry of the complex was further confirmed by ESI-Mass Spectrometry (ESI-MS).

DOSY spectroscopy [37,42–47] experiments were also carried out to confirm, qualitatively and quantitatively, the intermolecular interactions between MPT and β-CD in solution. The DOSY technique allowed for the determination of the individual self-diffusion coefficients (D) in multicomponent systems that directly reflected the association behavior of the interacting species [48].

**Figure 4.** The 1H NMR titration of MPT with β-CD: (**a**) Chemical shift change of the ortho-protons of the phenyl moiety of MPT with increasing β-CD concentration. Positive values mean downfield shifts. The small complexation induced shift (CIS) observed for the ortho-protons of the phenyl moiety of MPT (0.04 ppm, see Table 1) was particularly significant, indicating that a part of the phenyl ring of MPT entered into the cavity of β-CD. (**b**) Curve-fitting analysis by the HypNMR2008 program.

The principle on which DOSY is based is very simple and can be summarized as follows: when the host and the gues<sup>t</sup> are in the free state, they have their own diffusion coefficient that depends on their molecular weight and their shape. However, when they interact tightly together to form a complex, they behave as a single molecular entity and therefore should have the same diffusion coefficient [48,49].

Taking into account the fact that we are studying a rapid equilibrium on the NMR time scale between bound and free gues<sup>t</sup> molecules, the observed (measured) diffusion coefficient (*D*obs) is the weighted average of the free and bound diffusion coefficients (*D*free and *D*bound, respectively) and can therefore be used to calculate the bound fraction *p* by using the following Equation (1):

$$D\_{\rm obs} = \, p \cdot D\_{\rm bound} + (1 - p) \cdot D\_{\rm free} \tag{1}$$

which can be rearranged to yield:

$$p = \frac{D\_{\text{frac}} - D\_{\text{obs}}}{D\_{\text{frac}} - D\_{\text{bvund}}} \tag{2}$$

where *p* is the fraction of complexed substrate molecules.

After binding of a small gues<sup>t</sup> molecule (MPT) to a large host molecule (β-CD) the diffusion coefficient of the host is not greatly perturbed, therefore, the diffusion coefficient of the host–guest complex can be assumed to be the same as that of the non-complexed host molecule [50].

Pseudo 2D DOSY spectra are shown in Figure 5. The f1 dimension represents the self-diffusion coefficient (*D*) and the f2 dimension reports the chemical shift. The f1 is specific for each molecule, so the protons belonging to the same molecule appear in the same f1 row. The diffusion coefficients (*D*) and the fraction of complexed MPT molecules (*p*) measured at 23 ◦C in D2O are reported in Table 2.

As expected, the *D* value of encapsulated MPT (4.87·10−<sup>6</sup> cm2·s<sup>−</sup>1) was lower than that of free MPT (6.205·10−<sup>6</sup> cm2·s<sup>−</sup>1) (Table 2, Figure 5) thus proving that MPT is included in the β-CD cavity and diffuses more slowly. Recalling that the association constant, *K*a, for a 1:1 host–guest equilibrium of the type H + G - HG is defined by:

$$K\_{\mathbf{a}} = \begin{array}{c} \text{[HG]}\\ \hline \text{[H]}[\text{G]} \end{array} \tag{3}$$

where [H], [G], and [HG] are the equilibrium concentrations of the free host, free guest, and complex, respectively, Equation (3) can be rewritten as a function of the molar fraction [48] as:

$$K\_\mathbf{a} = \frac{p}{(1-p)([\mathbf{H}]\_0 - p[\mathbf{G}]\_0)}\tag{4}$$

where [H0] and [G0] are the total concentrations of the host and guest, respectively.

Lastly, inserting the value of the molar fraction just obtained from the DOSY experiments in the above Equation (4) we can calculate *K*a by using the single-point procedure [51,52]. The association constant measured in this way was 654 M−<sup>1</sup> at 23 ◦C (Table 2), and although the value obtained by the single-point approximation method results in large uncertainty, it is consistent with the corresponding value of 851 M−<sup>1</sup> estimated via NMR titration.

**Figure 5.** Pseudo 2D DOSY spectra of MPT (**a**) 1.5 mM; (**b**) in the presence of β-CD 1.4 mM in D2O, at 23 ◦C. (See Supplementary Information, SI, for details).

**Table 2.** Diffusion coefficients (*D*, 10−<sup>6</sup> cm2·s<sup>−</sup>1) related to the itemized protons of MPTfree (MPT alone 1.5 mM), MPT(+ β-CD) (MPT 1.5 mM with β-CD 1.4 mM), and β-CD(+ MPT) (MPT 1.5 mM with β-CD 1.4 mM), fraction of complexed MPT and β-CD molecules (*p*) and association constant for the β-CD–MPT complex (*K*a).


in D2O at 23 ◦C; b estimated errors <5%; c overlapped signals.

#### *3.2. ESI Mass Spectra*

a

In order to provide further confirmation of the formation of the β-CD–MPT inclusion complex, some ESI-MS experiments were conducted in aqueous solutions containing MPT and β-CD in the ratio 1:1, 1:2, and 2:1. In all cases, a base peak at m/z 1311 corresponding to a 1:1 host–guest complex was detected (Figure 6).

**Figure 6.** Electrospray negative-ion mass spectrum (ESI-MS) of a 1:1 β-CD–MPT aqueous solution that revealed a base peak corresponding to the 1:1 host–guest complex at m/z 1311 and a peak at m/z 1133 (50%) relative to uncomplexed β-CD.

#### *3.3. FTIR Spectra*

The diffuse reflectance FTIR spectrum of the solid β-CD–MPT complex precipitated from equimolar β-CD and MPT aqueous solutions was recorded and compared to those obtained on pure MPT and β-CD. Figure 7 clearly shows that the spectrum of the complex almost completely overlapped that of β-CD, but two extra peaks at 1492 cm<sup>−</sup><sup>1</sup> and 1593 cm<sup>−</sup><sup>1</sup> occurred (as evidenced in the enlarged inset), which corresponded to intense bands of the MPT molecule and did not appear in the β-CD spectrum. According to X.R. Ye et al. [53], both peaks were connected to the C–C stretching of the phenyl ring in MPT and the former also corresponded to N–H bending. These spectra further confirmed the complex formation.

**Figure 7.** FTIR spectra of solid β-CD–MPT, MPT, and β-CD.

#### *3.4. Electrochemical Tests*

These tests were performed on dip coated bronze electrodes in ARX10, because the association of the obtained thin coatings and the higher aggressiveness of the concentrated solution allowed for the di fferentiation of the coating protectiveness within the 20 day immersion period.

Figure 8 collects the time evolution of *<sup>R</sup>*p and *E*cor values in this environment at 30 ◦C. For bare electrodes, high initial *<sup>R</sup>*p values (9.1 kohm·cm2) were obtained, which decreased quickly to about 1 kohm·cm<sup>2</sup> for immersions longer than 1 h and then increased again up to 4.6 kohm·cm<sup>2</sup> towards the end of the immersion period (Figure 8a). PropS-SH coatings showed much higher initial *<sup>R</sup>*p values (almost 700 kohm·cm2), which decreased by about 1 order of magnitude during the 20 days of immersion due to the slow penetration of the aggressive solution through the silane network. The addition of MPT to the silane solution determined rather low and constant *<sup>R</sup>*p values, close to 20 kohm·cm2. This behavior was not investigated but it is plausible that MPT interfered with the coating reticulation and/or a surface competitive adsorption between free MPT molecules on one side and the silanol and thiol groups of silane coatings on the other occurred, so impairing the coating adherence and performance. Instead, some improvements were achieved with the addition of β-CD, which due to its hydroxyl groups was likely capable of reacting with the silanol groups, so contributing to the silane network formation. Finally, a clear progression was observed after β-CD–MPT complex addition with high and rather constant *<sup>R</sup>*p values (around 1 Mohm·cm2). This suggests that beside the positive e ffect of β-CD, the release of MPT molecules from the β-CD cavity could also play an important role in corrosion inhibition.

**Figure 8.** Polarization resistance ( *R*p) (**a**) and corrosion potentials (*E*cor) values (**b**) measured during 20 days of immersion of bare and coated bronze electrodes in concentrated acid rain (ARX10) at 30 ◦C.

The *E*cor values on bare electrodes evolved from −0.100 VSCE after 1 h immersion to about +0.043 VSCE after 20 days (Figure 8b). From previous research [54], it was found that this trend was the consequence of the degradation of the protective surface air-formed oxide film during the first half of the immersion period, which stimulated the cathodic reaction, with a consequent *E*cor shift towards nobler values. Then, in the second half of the immersion, the progressive accumulation of surface corrosion products induced a slight inhibition of the anodic process and further consequent *E*cor ennoblement. No significant *E*cor di fferences were detected in the presence of the coatings. In all cases, *E*cor increased to a certain extent during the initial 2 or 3 days of immersion, likely due to the evolution in the coating reticulation [39]. Then, they reached values in the range from 0.011 to 0.030 VSCE, independently of the corresponding *<sup>R</sup>*p values.

The polarization curves recorded at the end of the 20 days of immersion on bare and coated electrodes are shown in Figure 9, while Table 3 reports the electrochemical parameters derived from these curves.

**Figure 9.** Polarization curves recorded on bare and silane (PropS-SH)-coated bronze specimens in the absence and in the presence of β-CD, MPT, or β-CD–MPT complex, after 20 days of immersion in ARX10 at 30 ◦C.

**Table 3.** Corrosion potentials (*E*cor), corrosion current densities (*i*cor), and anodic Tafel slopes (*b*a) obtained on bare and coated bronze specimens after 20 days of immersion in ARX10 solution. Protection efficiencies (η) of silane coatings are also reported.


In particular, Table 3 collects the *E*cor and corrosion current (*i*cor) values, the anodic Tafel slopes, ba, and the protection efficiency (η) of the coatings, evaluated by the formula:

$$
\eta = \frac{i\_{\text{cor,b}} - i\_{\text{cor,c}}}{i\_{\text{cor,b}}} \times 100 \tag{5}
$$

where *i*cor,b and *i*cor,c are the corrosion currents evaluated on bare and coated electrodes, respectively. The cathodic Tafel slopes, *b*c, which were not reported in Table 3, were generally much higher than *b*a and close to infinity.

Figure 9 and Table 3 evidence that the coatings protect the underlying alloy from corrosion at different degrees. The plain PropS-SH coating mainly hindered the cathodic reaction, so determining a slight reactivation of *E*cor in comparison to those of the bare electrodes, and afforded a final η of about 84%. In agreemen<sup>t</sup> with the *<sup>R</sup>*p results, the addition of MPT to PropS-SH was detrimental to the coating protectiveness because it stimulated both the anodic and the cathodic reactions, suggesting a lower barrier effect of the coating and/or a lower surface adherence. In this case, the coating η value decreased down to 66% (Table 3). Conversely, the addition of β-CD or, even more, the β-CD–MPT complex in the coatings determined a decrease in the anodic and cathodic currents and induced η values of 92% and 98%, respectively. In contrast with MPT, the complex proved to be beneficial to the

coating performances, likely due to a higher compatibility of the external β-CD surface with the silane network and to the specific inhibition a fforded by released inhibitor molecules as evidenced by the following test.

#### *3.5. Cyclic AR Spray Test*

The self-healing capability of the silane coatings connected to the inhibitor release and adsorption at corrosion sites was evaluated during four weeks of exposure of the coated specimens with cross cut scratches to cyclic AR spray.

At the end of the test (Figure 10), the bronze coupons evidenced that the PropS-SH coating was rather protective at a distance from the scratches, but obviously did not avoid the substrate corrosion in scratched areas from which in fact the underfilm corrosion propagated. This corrosion form and coating delamination were less evident in the presence of β-CD, suggesting that this substance also increased the coating adherence, so improving the overall coating protectiveness. However, again, it did not prevent corrosion in the scratches. The addition of MPT in PropS-SH could not avoid the spread of corrosion attacks from the scratches and induced a significant surface color change (a brightening). Only β-CD–MPT complex addition significantly enhanced the substrate corrosion resistance and completely suppressed corrosion in the scratches and underfilm. This behavior suggested a self-healing capability of this coating type in the case of coating defects and mechanical damages, so prolonging the coating e ffectiveness.

**Figure 10.** Surface aspect of cross-cut coupons at the end of a 4 week exposure to a cyclic acid rain (AR) spray test; only the PropS-SH coating containing the β-CD–MPT complex prevented the development of a corrosion attack starting from the scratches.
