**Control of Polydimethylsiloxane Surface Hydrophobicity by Plasma Polymerized Hexamethyldisilazane Deposition**

#### **Virginija Jankauskaite˙ 1,\*, Pranas Narmontas <sup>2</sup> and Algirdas Lazauskas <sup>2</sup>**


Received: 30 November 2018; Accepted: 10 January 2019; Published: 11 January 2019

**Abstract:** The properties of a polydimethylsiloxane (PDMS) surface were modified by a one-step deposition of plasma polymerized hexamethyldisilazane (pp-HMDS) by the arc discharge method. Scanning electron microscopy, atomic force microscopy, and Fourier-transform infrared spectroscopy analytical techniques were employed for morphological, structural, and chemical characterization of the pp-HMDS modified PDMS surface. The changes in PDMS substrate wetting properties were evaluated by means of contact angle measurements. The unmodified PDMS surface is hydrophobic with a contact angle of 122◦, while, after pp-HMDS film deposition, a dual-scale roughness PDMS surface with contact angle values as high as 170◦ was obtained. It was found that the value of the contact angle depends on the plasma processing time. Chemically, the pp-HMDS presents methyl moieties, rendering it hydrophobic and making it an attractive material for creating a superhydrophobic surface, and eliminating the need for complex chemical routes. The presented approach may open up new avenues in design and fabrication of superhydrophobic and flexible organosilicon materials with a self-cleaning function.

**Keywords:** polydimethylsiloxane; superhydrophobicity; hexamethyldisilazane; plasma polymerization

#### **1. Introduction**

Polydimethylsiloxanes (PMDS) are the most widely used silicon-based organic polymers, commonly referred to as silicones. Because of easy fabrication, non-toxicity, biocompatibility and biodurability they have found potential applications in various fields. The surface of PDMS is naturally hydrophobic, but a number of efforts have been made to modify PDMS and further enhance its hydrophobicity [1,2]. PDMS hydrophobicity plays an important role in diverse applications e.g., self-cleaning surfaces [3], microfluidics [4], microelectromechanical systems [5], and biomedical applications [6].

Superhydrophobic PDMS surfaces can be fabricated by pulsed laser irradiation resulting in surface modification with a static contact angle (CA) value of 170◦ [7]. However, the whole irradiation procedure is highly time-consuming, thus limiting the scalability of this method. A more reliable and effective practice includes the deposition/formation of a thin film on the surface of the material to obtain the desired functionality. Plasma treatment is attractive as the processing time is short, the process involves low temperature, and procedures are relatively simple. Importantly, a single-step technique is desired for obtaining superhydrophobic and self-cleaning surface functionalities.

In this contribution, we fabricated a superhydrophobic PDMS surface via plasma polymerized hexamethyldisilazane (pp-HMDS) thin film deposition by arc discharge. To the best of our knowledge, the technique adopted here has been not reported for the fabrication of superhydrophobic PDMS surface using hexamethyldisilazane monomer as a precursor. HMDS is well known as being widely used for hydrophobic coatings on various hydroxyl-bearing surfaces [8,9]. HMDS chemical activity derives from the presence of a highly reactive nitrogen atom within the compound. The presented one step deposition of in situ polymerized hexamethyldisilazane is simple and scalable, and thus can provide a new strategy for the large scale fabrication of superhydrophobic surfaces with a self-cleaning function on flexible substrates.

#### **2. Materials and Methods**

The addition-curing silicone rubber Elastosil RT 601 A/B with a viscosity of 3500 mPa·s at 23 ◦C (Wacker Chemie AG, Múnich, Germany) was used as received for flexible films fabrication. HMDS of analytical grade (≥99%, Sigma-Aldrich, Saint Louis, MO, USA) was used as received.

The experimental setup of arc plasma reactor and technological conditions have been reported previously [10]. Briefly, a rod-shaped graphite anode and cathode were placed at the center of the discharge chamber. A quartz cuvette containing HMDS solution was positioned 20 mm from the anode, and PDMS substrate was placed at a distance of 15 mm from the electrodes. The chamber was connected to a vacuum line backed by a rotary pump. Arc-discharge was generated between anode and cathode using a DC transferred arc process using ~4.3 mA current and ~25 kV voltage. The deposition time was varied up to 60 s.

A FEI Quanta 200 scanning electron microscope (SEM, Thermo Fisher Scientific, Waltham, MA, USA) was used to collect micrographs of the investigated surface. The samples were imaged at an accelerating voltage of 30 kV. Atomic force microscopy (AFM) experiments were carried out with NT-206 (Microtestmachines, New Taipei City, Taiwan) in air at room temperature (22 ± 1 ◦C) using a V-shaped silicon cantilever operating in contact mode. The surface morphology of the resulting films was evaluated based on the AFM surface topography images and roughness parameters. Vertex 70 Fourier transform infrared (FTIR) spectrometer (Bruker Optics Inc., Billerica, MA, USA) equipped with a 30Spec (Pike Technologies, Madison, IA, USA) specular reflectance accessory having a fixed 30◦ angle of incidence was used for the chemical characterization of the modified PMDS surface.

CA measurements were performed at room temperature using the sessile drop method. A droplet of deionized water (5 μL) was deposited onto the investigated surface. Optical images of the droplet were recorded with a PC-connected digital camera after 10 s of dropping and CA measurements were carried out using an active contour method based on B-spline snakes (active contours) [11]. The contact angle hysteresis was measured as the difference between the advancing and receding contact angle of a sliding droplet. The test was performed by setting a droplet on a sample, which was placed on a horizontal plate. The plate was tilted slowly until the water drop began to slide along the surface; at this point the camera shutter was activated. The advancing and receding contact angles where then measured.

#### **3. Results**

The morphology of unmodified and plasma polymerized HMDS (pp-HMDS) modified flexible PDMS substrate and water droplets on the PDMS surface before and after pp-HMDS film deposition at different times are compared in Figure 1. Cured PDMS is produced spontaneously forming wavy structures on the surface with micro-scale amplitude and periodicity of 128 nm. SEM images of the resulting pp-HDMS thin film surface for deposition times of 30 and 60 s are presented in Figure 1b,c, respectively. The deposition resulted in a highly branched and crosslinked pp-HMDS structures composed of quasi-spherical nanoparticles with size in the range of 15–60 nm. Growth in three-dimensional assembles and formation of large nanoparticles aggregates were observed as the deposition time increased.

These morphological alterations change the wetting properties of the PDMS surface (Figure 1). The unmodified PDMS surface exhibits hydrophobic behavior with a static CA value not higher than 122◦. A considerable improvement in non-wetting characteristics of pp-HMDS film functionalized surfaces was observed. After 30 s of deposition, the nanostructured pp-HMDS film exhibited superhydrophobic properties with static CA values of 169◦–170◦. In this case the low value of CA hysteresis (2◦), defined as the difference between the CA at the front of the droplet (advancing CA) and at the back of the droplet (receding CA), was obtained. The increase of deposition times up to 60 s results in lower CA values, i.e., CA = 159◦–161◦.

**Figure 1.** SEM image of unmodified (**a**) and pp-HMDS film modified PDMS surfaces (**b**,**c**) and water droplets on PDMS surface before and after pp-HMDS film deposition at different times: (**a**) 0; (**b**) 30 s; (**c**) 60 s.

The presence of nanostructures in the form of quasi-spherical nanoparticles and interconnection caused by the formation of large aggregates at longer pp-HMDS deposition time (60 s) was confirmed using characteristic AFM topographical and surface profile images shown in Figure 2. In this case dual-scale roughness of the surface was maintained, the pp-HMDS film surface was found to be rough with the root-mean square roughness having a value of 96.11 nm. However, the spiky surface morphology changes into a bumpy one and the negative surface skewness parameter value (−0.2) indicates predominance of valleys.

As can be seen from Figure 3a, in the FTIR absorbance spectrum of unmodified PDMS the bands at 2965 and 2906 cm−<sup>1</sup> are assigned to asymmetric and symmetric stretching of CH3 groups, respectively [12]. The asymmetric and symmetric bending vibrations of CH3 groups are also observed at 1410 and 1258 cm−1, respectively. The bands at 1072 and 1007 cm−<sup>1</sup> are characteristic of Si–O–Si asymmetric and symmetric stretching vibrations, respectively. Asymmetric rocking at 864 cm−<sup>1</sup> and stretching at 785 cm−<sup>1</sup> vibrations can be attributed to the Si–CH3 group [12].

**Figure 2.** AFM image of 3D (**a**) and 2D (**b**) topography with normalized *Z* (nm scale), and profilogram of pp-HMDS film at deposition time of 60 s (**c**).

The deposition of polymerized pp-HMDS film leads to the obvious PDMS surface functional group changes (Figure 3b). The broad band between 3400 and 3700 cm−<sup>1</sup> is related to O–H stretching in Si–OH bonds of hydrophilic silanol groups [13]. The absorbance at 3350 cm−<sup>1</sup> is characteristic for stretching of the N–H bond, while the doublet at 2350 cm−<sup>1</sup> is attributed to CO2 species [10]. As in the case of unmodified PDMS, the presence of methyl moieties in the modified surface is confirmed by an absorption band at 1410 cm−1, related to CH3 asymmetric bending in Si–CH3 bonds, and 2965 and 2906 cm−<sup>1</sup> bands, which are characteristic for asymmetric stretching and symmetric stretching of the CH3 group, respectively [14,15]. A low intensity band at 1454 cm−<sup>1</sup> is assigned to the asymmetric bending vibrations of the CH2 group in the Si–CH2–CH2–Si link that play a substantial role in the cross-linking process during HMDS polymerization [15]. The band located at 2250 cm−<sup>1</sup> corresponds to Si–H stretching vibration [16], while the band at 1629 cm−<sup>1</sup> can be assigned to stretching of C=O [17]. Some oxygen related functional groups could arise from free radical (possibly trapped in the film structure) reaction with the atmosphere, when the samples are removed from the reactor [18].

**Figure 3.** FTIR absorbance spectra of flexible PDMS substrate (**a**) and pp-HMDS (**b**) with functional groups assigned and schematic diagram of probable deposition mechanism (**c**); HMDS monomer and pp-HMDS network fragment are shown in van der Waal's-based representation.

#### **4. Discussions**

Generally, the hydrophobic properties of films are determined by the kind and amount of grafted hydrophobic groups and surface roughness parameters. One step coating by in situ HMDS deposition and polymerization is an easy and rapid method to impart non-wetting properties to the PDMS surface. The plasma polymerization process of HMDS monomers resulted in highly branched and crosslinked structures composed of quasi-spherical nanoparticles. After pp-HMDS film deposition, the PDMS surface shows superhydrophobic characteristics with a CA value close to 170◦. High hydrophobicity of pp-HMDS originates from the high amount of CH3 species and specific film surface morphology. The pp-HMDS film functionalized surfaces exhibited Cassie–Baxter state with a "lotus effect" observable and a low CA hysteresis of 2◦, suggesting that a water droplet is not able to wet the spaces between surface morphological features allowing air pockets to remain at the interface. The increase of pp-HMDS film deposition time influences the decrease of CA value. It can be attributed to a higher solid fraction of surface morphological features in contact with the water droplet, which decreases the concentration of air pockets trapped at the interface of pp-HMDS with the droplet.

Based on the SEM, FTIR data, and surface wetting studies, it is suggested that the HDMS monomer molecules passing to the arc plasma region during the operational process are fragmented with partial retention and formation of new chemical bonds. The corresponding repetition of fragmentation and recombination reactions of HMDS monomers in arc plasma leads to the deposition of a randomly crosslinked network structure of pp-HMDS (Figure 3c) and forms a heterogeneous surface with a high fraction of methyl moieties retained, thus providing superhydrophobic characteristics with a self-cleaning function.

Thus, FTIR investigations reveal multiple non-covalent interaction achieved by in situ HMDS polymerization with physical anchoring on the polymer surface [19]. Such an interaction can be recognized as the driving force for constructing and fabrication of superhydrophobic and flexible organosilicon materials with a self-cleaning function.

#### **5. Conclusions**

Herein, we successfully enhanced the non-wetting properties of a flexible polydimethylsiloxane substrate surface via plasma polymerized hexamethyldisilazane thin film deposition by the arc discharge method. Such a film is composed of quasi-spherical nanoparticles stacked together, which form a branched network. The deposited nanostructured plasma polymerized hexamethyldisilazane film exhibits superhydrophobic properties with static contact angle values as high as 170◦ and a low contact angle hysteresis of 2◦. The PDMS surface undergoes self-cleaning and non-wetting behavior due to the multiple non-covalent interactions attended by the incorporation in the surface layer of methyl groups and a nano-rough surface formation. This is a facile and effective method that can provide a new strategy for the large scale fabrication of superhydrophobic surfaces with a self-cleaning function on flexible substrates.

**Author Contributions:** Conceptualization, A.L. and V.J.; Methodology, A.L. and P.N.; Validation, V.J. and P.N.; Formal Analysis, V.J.; Investigation, A.L. and P.N.; Resources, A.L. and V.J.; Writing—Original Draft Preparation, V.J. and A.L.; Writing—Review & Editing, V.J. and A.L.; Visualization, A.L. and V.J.; Supervision, V.J.; Project Administration, A.L.; Funding Acquisition, V.J.

**Funding:** This research was funded by the European Social Fund under "Development of Competences of Scientists, other Researchers and Students through Practical Research Activities" measure (No. 09.3.3-LMT-K-712-01-0074).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Properties of Post-Consumer Polyethylene Terephthalate Coating Mechanically Deposited on Mild Steels**

#### **Elisângela Silva 1,\*, Michele Fedel 2, Flavio Deflorian 2, Fernando Cotting <sup>1</sup> and Vanessa Lins <sup>1</sup>**


Received: 16 October 2018; Accepted: 28 December 2018; Published: 5 January 2019

**Abstract:** An anticorrosive coating of post-consumer polyethylene terephthalate (PET) was applied on carbon steel by using an industrial press. The PET layer showed a good adhesion on the substrate, evaluated by using a pull off test, when compared with the traditional organic coatings. In addition, scanning electron microscopy (SEM) analysis showed that the PET layer was uniform, homogeneous, and free of cracks or defects. The Fourier-transform infrared spectroscopy (FTIR) and differential scanning calorimetry (DSC) proved that the PET properties were not affected by the deposition process. The PET organic coating is a promising coating, due to its corrosion resistance evaluated by using salt spray tests, even though the applied thickness of 65 μm was considered thin for a high-performance coating. The electrochemical impedance spectroscopy (EIS) showed that the PET coating has a capacitive effect and its electrochemical behavior was not affected as the exposure time increased, resulting in an impedance modulus value of 10<sup>10</sup> <sup>Ω</sup>·cm2, after 576 h of immersion in an aqueous solution of NaCl 3.0 wt %.

**Keywords:** corrosion; electrochemical impedance spectroscopy (EIS); organic coatings; polyethylene terephthalate (PET)

#### **1. Introduction**

An alternative contribution to solve the problem of urban solid waste (USW) is to use solid polymeric waste to develop coatings for steels [1]. Carbon steel has excellent mechanical properties and low cost but has some disadvantages such as low wear and corrosion resistance in various media [2]. The most employed method to protect the carbon steel against corrosion process has been the organic coatings. Polymeric layers are applied on the carbon steel to avoid the contact between metallic surface and the aggressive environment. The barrier protection conferred by the polymeric coatings can be associated to their water uptake property [3–10].

The organic coatings based on polyester, epoxy, and polyurethane resins, which present a lower water uptake value compared to alkyd resins, are widely used. Nevertheless, for long exposure time, the absorbed water reaches the metal and the corrosion process is established. In order to retard this process, thick films (from tens of μm to a few mm) of these organic coatings are applied on the substrate, thus causing an increase in the total cost of this project [8–15].

Lins et al. [1] demonstrated that post-consumer commingled polymer (PCCP) coatings on carbon steels can be produced by thermal spraying and investigating their erosion behavior. Literature reports the use of post-consumer tires to develop sustainable polymeric composite materials which are characterized by good mechanical and functional properties [3]. Çinar and Kar [4] mixed PET waste particles in a screwed extruder with marble dust to produce a composite material.

Municipal solid waste management has received a great deal of attention as countries such as India, which produces an estimated quantity of 50–600 million tons of urban solid waste annually [5]. The municipal solid waste is mainly composed of metals, glass and plastics. The municipal solid waste of Bangalore, India, contains 6.23% of plastics [5]. In 2016, 27.1 million tons of plastic waste was collected through official schemes in the EU28 plus Norway and Switzerland to be treated, and for the first time, more plastic waste was recycled than landfilled [6].

One of the main components of plastic fraction of USW is the polyethylene terephthalate (PET) which is one of the most versatile polymers available, which makes it the most produced and used polymer worldwide [7,8]. PET has characteristics such as flexibility, transparency, good adhesion, low permeability to liquids and gases, high thermal resistance and low cost, all of which enable PET to be an excellent candidate for a protective coating with low thickness [8–11]. Literature reports PET deposition on steels by using low-velocity flame spray technology [9].

In this work, post-consumer PET was deposited on carbon steels by using mechanical deposition. Mechanical PET deposition presents simplicity, lower cost in relation to the thermal spraying, and was not found in literature to the best of our knowledge. The process of comminution of PET from the post-consumer bottles until a powder of the desired granulometry is not a simple process; one reason for this relies on the fact that PET is hygroscopic. Based on the obtained results, a procedure was developed for the comminution of PET from the post-consumer bottles, which has proved to be very efficient. The effect of the mechanical deposition on the PET structural and thermal properties was investigated by FT-IR spectroscopy and differential scanning calorimetry (DSC) measurements. The protection properties provided by the PET coatings were evaluated by using electrochemical impedance spectroscopy (EIS) analysis and exposure in the neutral salt spray chamber (NSST, according to the ASTM B117 [12] and ISO 9227-2017 standards [13]). The obtained results have been compared with literature data of pristine paints employed for corrosion protection purposes. In addition, adhesion tests were performed by the pull-off test, in order to characterize the interaction between PET coating and steel substrate.

#### **2. Materials and Methods**

#### *2.1. Materials*

The post-consumer PET bottles were collected from household waste; the labels and the lid were removed and discarded. The bottles were washed in soapy water. The grinding process allowed a yield of 85% of the overall use of the bottle; only the top and bottom of the bottles were not used.

The substrate used was Q-Panel type R steel with dimensions 80 mm × 40 mm × 1 mm. The composition of this carbon steel was 0.15 C, 0.60 Mn, 0.030 P, 0.035 S (wt %). Prior to deposition, a blast pre-treatment of the steel samples was used, creating a roughness profile on the carbon steel surface (*R*z) of approximately 20 μm. An SBC 350 sandblasting machine (Nova, Rehovot, Israel) was used with Garnet filter sand. After this process, the plate was washed with acetone in an ultrasonic bath for 2 min and dried with clean air.

#### *2.2. Deposition Process*

The amount of 0.8 g of PET ground with particle size of approximately 0.07 mm was deposited on carbon steel. The set was placed in the press for heating at 260 ◦C for 5 min without pressure; there after a pressure of 0.5 ton was applied for 2 min. The material remained in the press and was cooled for 5 min. The coated steel was then removed from the press and conditioned at room temperature for 24 h. The PET coating thickness was 65 ± 5 μm.

#### *2.3. Coating Characterization*

The morphology of the coating was examined by using scanning electron microscopy (SEM), with a FEG-Quanta 200 FEI equipment (FEI Company, Fremont, CA, USA). The coated steel was also characterized by using Fourier transform infrared spectroscopy employing an attenuated total reflection (ATR) geometry, by means of Varian 4100 Excalibur Series equipment (Santa Clara, CA, USA). The wavelength range was 500–4000 cm−<sup>1</sup> and the resolution of 4 cm<sup>−</sup>1. The powder and PET coating are from the same batch before and after pressing in Caver laboratory press, model 2699, Ser. No. 2699-12748, Fred S. carver Inc., Wabash, IN, USA.

The differential scanning calorimetry (DSC) test was performed using the Mettler DSC30 (Mettler Toledo, New York, NY, USA) equipment in three cycles, with heating from 0 to 300 ◦C under a nitrogen flow of 10 mL·min<sup>−</sup>1, followed by cooling and heating from 0 to 300 ◦C, at a heating rate of 10 ◦C min<sup>−</sup>1. Thermal analysis provided the glass transition temperature (*T*g), crystallization temperature (*T*c) and melting temperature (*T*m). The crystallinity value (χc) was calculated by using the values of the endothermic melting peak using the Equation (1):

$$\chi\_{\rm c} = \frac{\Delta H\_{\rm m}}{\Delta H\_{\rm m}^{0}} \tag{1}$$

where Δ*H*<sup>m</sup> is melting enthalpy, and Δ*H*<sup>0</sup> <sup>m</sup> the melting enthalpy for polyethylene terephthalate, considered 140 J·g−<sup>1</sup> [14].

The adhesion of the coating was evaluated by using the pull-off test according to the ASTM D4541 standard [15]. Dollies of 20 mm were glued on the plate coated with Huntsman Araldite 2000 glue; the cure occurred in a period of 24 h. The equipment used was a De Felsko PosiTest AT-M (Ogdensburd, NY, USA). The tests were performed in triplicate.

Coating thickness was measured by using a digital layer thickness gauge on a ferrous base Digi-Derm Mitutoyo (Aurora, CO, USA). Ten measurements were collected on the coated steel.

The size distribution of the PET particles was performed by laser diffraction in the Mastersizer 3000 equipment (Worcestershire, UK). The obtained values were *D*(4,3), which represents the average size distribution; *D*v(10), which represents that only 10% of the particles were smaller than the displayed value; *D*v(50), showing that 50% of the particles were larger, and 50% smaller than the determined value; and *D*v(90), which reveals that 90% of the particles were below this value.

#### *2.4. Corrosion Resistance Evaluation*

Corrosion tests were performed using an Equilam salt spray chamber (Diadema, Brazil) in compliance with the ASTM B117 standard [12] for 500 h. The solution used was 5.0 wt % NaCl with pH between 6.5 and 7.2. A linear scratch was made on the coated steel with a tungsten carbide tool at an angle of 60◦.

Electrochemical impedance spectroscopy was performed by using Metrohm AUTOLAB 302 N potentiostat/FRA (Utrecht, The Netherlands) at room temperature. The measurements were performed using a three-electrode electrochemical cell; a platinum foil was used as a counter electrode, a Ag/AgCl(sat) reference electrode was used, and the coated samples were employed as working electrodes. A polyvinylchloride (PVC) tube was affixed with silicone glue on the coated plate, delimiting the exposure area to 6.25 cm2, where the 3.0 wt % NaCl solution was poured. The open circuit potential was measured for one hour or until stabilization. The EIS measurements were collected at the OCP after 1, 24, 192, 360, 480, and 576 h of immersion. The potential signal amplitude applied was 10 mV (rms), and the frequency range analyzed was 105–10−<sup>2</sup> Hz. The data collected were analyzed by using the ZView2 software.

Evolution of the capacitance of the film was employed to measure the phenomena of water absorption since the presence of moisture modifies the dielectric constant of the polymer [16,17].

The ratio of the dielectric constant (ε) of the coating was calculated using Equation (2) [18]:

$$
\varepsilon = \frac{\mathcal{C}\_{\rm c}}{\varepsilon\_0 A} L \tag{2}
$$

<sup>ε</sup><sup>0</sup> is the dielectric constant of free space (8.854 × <sup>10</sup>−<sup>12</sup> F m−1), *<sup>A</sup>* is the coating area and *<sup>L</sup>* is the coating thickness.

The fraction of water volume of coating at saturation (Ø∞) was calculated by using the Brasher and Kingsbury (BK) formula:

$$\mathcal{Q}\_{\infty} = \frac{\log \, \mathbb{C}\_{\infty} / \mathbb{C}\_{0}}{\log \mathbb{C}\_{t}} \tag{3}$$

where *C*<sup>∞</sup> is the coating capacitance at immersion time *t* (determined from EIS data); *C*<sup>0</sup> is the initial coating capacitance and *Ct* is the relative permittivity of the water. The parameters *C*<sup>∞</sup> and Ø are dependent on the immersion time [19].

#### **3. Results and Discussion**

#### *3.1. Characterization Results*

Figure 1 shows the PET powder produced from post-consumer bottles. The particles are heterogeneous in shape and size. More elongated and other equiaxial particles can be identified. Particle sizes have been calculated from measurement of particle areas using the IMAGEJ-win64 software. The PET powder presented a particle area between 430 and 51,200 μm2, and 90% of PET particles showed the size of 635 μm.

**Figure 1.** PET powder granulometric distribution.

Figure 2 presents the cross-sectional section of the coated steel obtained by using SEM analysis. The film is uniform and free of cracks and voids.

Figure 3 shows the FT-IR spectra of PET powder and coating, and Table 1 shows characteristic absorptions of functional groups present in PET powder and PET coating. The PET absorption band at 1715 cm−<sup>1</sup> is attributed to vibrations of the carbonyl group of saturated esters; the bands at 724 and 871 cm−<sup>1</sup> are due to the interaction of polar ester groups and benzene rings, vibration =C–H out of the plane. The 1097 cm−<sup>1</sup> band is associated with the stretching vibration mode of C–O bonds and bands at 1242 cm−<sup>1</sup> are associated to the specific absorption of the terephthalate group (OOC–C6H4–COO). Bands related to the asymmetric deformation in CH2, at 1407 and 1018 cm<sup>−</sup>1, were identified [20–22]. At 2960 cm−1, absorption associated to the symmetrical stretch of the C–H bond is observed, with a higher intensity for the PET powder than PET coating. Compared to the coating, the PET powder showed a broad band around 3600 cm−<sup>1</sup> probably due to the water absorption. Notice that the FT-IR spectra of the powder and of the coating are very similar. The coating deposition treatment does not seem to affect the chemical structure of the polymer.

**Figure 2.** Cross-sectional view of PET coating.

**Figure 3.** FTIR spectra of PET powder and PET coating.

**Table 1.** Characteristic absorptions of functional groups present in PET powder and PET coating.


Figure 4 shows DSC results for PET powder and PET coating (removed from the mild steel substrate to carry out the analysis). Table 2 summarizes the thermal analysis results for PET powder and coating. Analyzing the *T*<sup>g</sup> values, the onset values for PET powder and PET coating are very close being 73.8 and 72.8 ◦C, respectively, within the margin of error of the measurement that is 2 ◦C. The glass transition temperature is the temperature at which the carbon–carbon bonds become more flexible. The PET coating is rigid at room temperature, since it is below *T*g.

The similar *T*<sup>g</sup> values of powder and coating do not indicate significant damage in the reorganization of the polymer chain of the coating during deposition. As far as thermally sprayed PET on mild steels are concerned, Duarte et al. [9] reported that the *T*<sup>g</sup> of coating was lower than that of the powder, which indicated a thermal degradation of the PET during the thermal deposition, decreasing its molar mass. In [9], the *T*<sup>g</sup> value for the PET powder was 79 ◦C, about 5 ◦C above the value obtained in this work. The glass transition temperature depends on the heating rate which is the same in this work and in reference [8]. In the case of the thermal spray technique, the polymer is subjected to high temperatures above 1000 ◦C, above the degradation temperature of the PET which

is about 450 ◦C. Demirel [23] studied mound surface temperature in injection stretch blow molding (ISBM) of polyethylene terephthalate (PET) bottles for carbonated soft drinks (CSD) storage. The cited author reported the effect of mound surface temperature on *T*g of PET bottles which varied between 58.3 and 59.4 ◦C.

**Figure 4.** DSC results of PET powder and PET coating on mild steel.

**Table 2.** Glass transition and melting temperatures, and crystallinity of PET powder and PET coatings.


The crystallization temperature of the coating showed an increase of 6.7% in relation to the powder, as seen in Figure 4 and Table 1. This phenomenon may have occurred due to the PET melting at 260 ◦C for application on the steel, and due to the slow cooling process, which provides time to recrystallization of part of the material. The χ<sup>c</sup> values also confirm the crystallinity increase, being 25.5% for the powder and 31.7% for the coating. Takeshita et al. [24] studied the influence of cooling time on the physical properties of the polyester powder coating. They concluded that the cooling time influences the crystallinity values of the polymer and that the ratio between the cooling time and the degree of crystallinity is almost linear. In this sense, the increase in crystallinity of the PET coating was due to the cooling process of the coating deposition. Demirel [23] showed values of crystallinity of PET bottles similar to that found in this work, in the range from 21% up to 29%. Melting temperatures for PET powder and coating were similar (233 and 235 ◦C, respectively). Literature reports values of PET melting temperature in the range of 242–260 ◦C [22,25].

PET degradation occurred at temperatures of 454.7 and 455.0 ◦C for powder and coating, respectively. As the PET reached 260 ◦C in the deposition process, degradation did not occur. Duarte et al. [9] reported degradation temperatures of 444 ◦C for PET powder and 437–446 ◦C for thermally sprayed PET coating.

Dry adhesion test was carried out by means of pull-off test. The detachment of the dolly occurred at the interface between the glue and the coating, at about 5 MPa. Based on these findings, we can assume that the adhesion strength is greater than 5 MPa. The value found for PET coating was compared with literature values for polyester powder coatings. As a result, we can observe in Table 3 that the adhesion values for the polyester family is very close to the value found in this work.


**Table 3.** Thickness and adhesion for polyester and polyethylene terephtahalate.

#### *3.2. Corrosion Resistance of PET Coated Steel*

3.2.1. Results of Exposure of the Coating on Scratched Samples after 480 h in Salt Spray Chamber

Salt spray results are shown in Figure 5 for up to 480 h of exposure in chamber.

The salt spray tests showed a reduced extent of coating delamination resulting from the scratch. No blisters where observed during the 480 h of immersion. As the test time increased, a marked increase in the corrosion of the carbon steel in the scratch area was observed. However, the cathodic front did not advance underneath the coating. No coating detachment was observed during exposure time of up to 480 h.

**Figure 5.** Photographs of PET coated steel after exposure for (**a**) 72, (**b**) 240, and (**c**) 480 h in a salt spray chamber.

#### 3.2.2. Electrochemical Impedance Spectroscopy (EIS)

Figure 6 shows Bode diagram for PET coated steel in an aqueous solution of NaCl for immersion up to 576 h. The phase angle curves are sometimes scattered, in particular between 10 and 100 Hz and around 0.1 Hz, probably due to the relatively high impedance of the coating during immersion time. According to the impedance modulus diagrams (Figure 6a) the impedance values in the low frequency range (at about 0.01 Hz, |*Z*|0.01) was high and almost stable as it ranged from 1 × 1010 to <sup>3</sup> × 1010 <sup>Ω</sup> cm2 throughout immersion time. A straight line with the slope close to −1 was observed in the Bode diagram, thus suggesting an almost capacitive behavior over a wide range of frequencies. A single time relaxation process was observed in the middle frequency range.

After 360 h of immersion, a decrease in impedance in the low frequency domain was observed, however, there is a trend for stabilization (Figure 7). The low frequency impedance was in the order of 1 × <sup>10</sup>10–3 × 1010 <sup>Ω</sup> cm2 during all immersion time, thus suggesting that the PET applied coating provided steel corrosion protection. The PET coating seems to be able to retard the direct contact between the corrosive medium and metal substrate. To better investigate the properties of the recycled PET coating, EIS data were fitted employing the equivalent circuit shown in Figure 7. According

to [28,29], a *R*e(*Q*c*R*p) was employed to fit the experimental spectra. *R*<sup>e</sup> is the electrolyte resistance. The constant phase element, *Q*c, and the resistance, *R*p, where attributed to the dielectric properties and the pore resistance of the PET coating, respectively.

**Figure 6.** (**a**) Bode diagram of impedance modulus versus frequency for the PET coated steel in a saline environment for 1, 24, 192, 360, 480, and 576 h, with amplification of low frequency region; (**b**) Bode diagram of phase angle versus frequency for the PET coated steel in a saline environment for 1, 24, 192, 360, 480, and 576 h.

**Figure 7.** Equivalent circuit for EIS data of PET coated steel in NaCl solution.

Considering the results reported in Table 4, as the values of the '*n*' exponent is about 0.98 during the whole immersion time, the pre-factor of the CPE named *Q*c can be assumed as the coating capacitance. For this reason, from now on, the pre-exponential factors of the CPE will be considered throughout the paper as an acceptable approximation of the capacitance values during immersion time. Figure 8 shows the evolution of the normalized coating capacitance (respect to the initial capacitance value, *Ct* <sup>=</sup> *<sup>x</sup>*/*Ct* = 0) during the first 24 h of immersion. Initially, a sharp increase in the normalized capacitance was observed, probably due to micropores formed in the coating which facilitated water permeation. After the first hours of immersion, the capacitance reached a sort of saturation, remaining stable at a value 1.12/1.14 times the initial capacitance. As far as the pore resistance is concerned, from Table <sup>4</sup> it is possible to observe that it decreased from 1.82 × 1010 to 0.81 × 1010 <sup>Ω</sup> cm<sup>2</sup> after 576 h due to continuous immersion in the electrolyte, thus suggesting that the recycled PET coatings have excellent barrier properties.

In fact, the pore resistance *R*p values of the investigated coatings are comparable or, in particular cases, highly compared to previously reported literature data (Table 5) related to polyester-based paints (derived from pristine materials).


**Table 4.** Electrochemical parameters for PET coated steel in a saline medium.

**Figure 8.** Capacitance versus immersion time of PET coated steel.

**Table 5.** *R*p values for polyester from the literature in electrolytes of 3.0–3.5 wt % of NaCl.


Relative permittivity of dry coating ε<sup>d</sup> and fraction of water volume Ø were calculated by Equations (2) and (3). The value of ε<sup>d</sup> for PET coating was 5.0: this value is in accordance with relative permittivity reported in the literature to dry coating 3–8 [29,32,33]. The coating to be considered protective and water resistant must show the value of Ø between 2.0% and 15.98%, approximately [29,34]. The value of fraction of water volume estimated for the PET coating was 3.2% after 24 h of immersion in 3.0 wt % NaCl solution.

#### **4. Conclusions**

The DSC results indicated no significant damage to the polymer produced by the mechanical deposition (press recycled).

The PET organic coating with a thin layer of 65 μm presented good adhesion to the substrate, superior to 5 MPa, evaluated by using the pull-off test, as well as a high corrosion protection for carbon steel during long immersion times in the NaCl 3.0 wt % electrolyte; both of which are important characteristics for an anticorrosive organic coating. The polarization resistance of PET coated steel in a saline solution was 8.1 × 109 <sup>Ω</sup> cm2 after 576 h of immersion. The value of fraction of water volume absorbed was 3.2% indicating the excellent protective action of coating.

The PET organic coatings can be considered as an alternative, both for the PET recycling and for a new anticorrosive coating.

Wear evaluation of PET coating will be performed in further investigation.

**Author Contributions:** Conceptualization, E.S., V.L., F.D. and M.F.; Methodology, E.S.; Validation, E.S., F.D., M.F. and F.C.; Formal Analysis, E.S., M.F. and F.C.; Investigation, E.S. and M.F.; Writing–Original Draft Preparation, E.S. and V.L.; Writing–Review & Editing, E.S., M.F., F.D., F.C. and V.L.; Supervision, E.S.; Project Administration, E.S., F.D. and M.F.

**Funding:** This research was funded by Foundation CAPES (Coordination for higher Education Staff Development) (No. 88881.135023/2016-01).

**Acknowledgments:** The authors would like to sincerely thank the polymers laboratory of University of Trento.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

## **Studies on Synthesis and Characterization of Aqueous Hybrid Silicone-Acrylic and Acrylic-Silicone Dispersions and Coatings. Part I**

**Janusz Kozakiewicz 1,\*, Joanna Trzaskowska 1, Wojciech Domanowski 1, Anna Kieplin 2, Izabela Ofat-Kawalec 1, Jarosław Przybylski 1, Monika Wo ´zniak 2, Dariusz Witwicki 2,† and Krystyna Sylwestrzak <sup>1</sup>**


Received: 15 November 2018; Accepted: 23 December 2018; Published: 3 January 2019

**Abstract:** The objective of the study was to investigate the effect of the method of synthesis on properties of aqueous hybrid silicone-acrylic (SIL-ACR) and acrylic-silicone (ACR-SIL) dispersions. SIL-ACR dispersions were obtained by emulsion polymerization of mixtures of acrylic and styrene monomers (butyl acrylate, styrene, acrylic acid and methacrylamide) of two different compositions in aqueous dispersions of silicone resins synthesized from mixtures of silicone monomers (octamethylcyclotetrasiloxane, vinyltriethoxysilane and methyltriethoxysilane) of two different compositions. ACR-SIL dispersions were obtained by emulsion polymerization of mixtures of the same silicone monomers in aqueous dispersions of acrylic/styrene copolymers synthesized from the same mixtures of acrylic and styrene monomers, so the compositions of ACR and SIL parts in corresponding ACR-SIL and SIL-ACR hybrid dispersions were the same. Examination of the properties of hybrid dispersions (particle size, particle structure, minimum film forming temperature, *T*<sup>g</sup> of dispersion solids) as well as of corresponding coatings (contact angle, water resistance, water vapour permeability, impact resistance, elasticity) and films (tensile strength, elongation at break, % swell in toluene), revealed that they depended on the method of dispersion synthesis that led to different dispersion particle structures and on composition of ACR and SIL part. Generally, coatings produced from hybrid dispersions showed much better properties than coatings made from starting acrylic/styrene copolymer dispersions.

**Keywords:** aqueous polymer dispersions; silicone-acrylic; acrylic-silicone; hybrid particle structure; coatings

#### **1. Introduction**

Aqueous polymer dispersions are currently produced in quantities exceeding globally 20 million tons per annum [1] and are commonly used, inter alia, as binders for organic coatings, especially for aqueous dispersion-based architectural paints. The reason for a great increase in research and business interest in that specific group of products is not only the fact that they are environmentally friendly, but also the possibility of tailoring the composition and structure of dispersion particle in order to achieve desired performance characteristics of the final coating. If the particles have a hybrid (It is worth to note that generally a "hybrid material" is a material that is composed of at least two components mixed at molecular scale [2] and although this term is normally used for polymer-inorganic structure composites [3], it can be also applied to polymer-polymer systems) structure (i.e., are composed of at least two different polymers) and their diameter is less than 100 nm they may be called "dispersion nanoparticles with hybrid structure" within which the occurrence of specific interactions between these polymers optionally leading to synergistic effect may be expected. Then, due to a synergistic effect, new and sometimes quite unexpected features of coatings or films made using such hybrid dispersions as binders may be found—see Figure 1.

**Figure 1.** Differences between mixture of two aqueous dispersions of different polymers (polymer 1—blue color and polymer 2—red color) and aqueous dispersion with hybrid particle structure composed of the same two different polymers. In the mixture of two dispersions (**a**) synergistic effect is much less probable than in dispersion with hybrid particle structure (**b**) [4].

Although some authors reported that certain specific properties of coatings could be improved by using blends of dispersions of different polymers (e.g., dirt resistance could be enhanced this way [5]), clear advantages of application of dispersions with hybrid particle structure as coating binders have been confirmed in the literature, e.g., [1,4,6] and descriptions of methods applied for synthesis of such dispersions can be found in books and review papers, e.g., [7–10]. The particle morphology that is most frequently referred to in the research articles is a "core-shell" structure, but other morphologies, like core-double shell, gradient, eye-ball-like, raspberry-like, fruit cake or embedded sphere can also be obtained [11,12]. It has been proved [8] that not only the hybrid dispersion particle size and chemical composition, but also its morphology can significantly affect the properties of both dispersions and coatings produced from such dispersions. Therefore, investigation of the hybrid dispersion systems is of great interest to many researchers.

According to [7], the following general approaches can be applied to synthesis of hybrid aqueous dispersions in the emulsion polymerization process:


As is clear from the literature, e.g., [4,13] the aforementioned methods of hybrid aqueous dispersion synthesis can be successfully applied to obtain different dispersion particle morphologies, depending on the selection of the starting materials (polymers, monomers, surfactants, initiators etc.). It can be expected that if methods (1) or (2) are applied, the "core-shell" morphology will be the most probable one supposing that certain conditions are fulfilled: "core" polymer should be more hydrophobic than "shell" polymer and formation of separate particles of polymer X in the course of

polymerization in dispersion of polymer Y resulting from homolytic nucleation is retarded, e.g., by diminishing the monomer and surfactant concentration in the reaction mixture. The mechanism of hybrid particle formation in the emulsion polymerization process has been described in detail, e.g., in [4,13–15] and factors that determine creation of specific particle morphology have been identified. A review of fundamental theoretical aspects of the formation of dispersion particles with a hybrid structure has been published [16].

One of the hybrid aqueous dispersion systems that is most interesting from the point of view of practical application, especially as coating binders, is dispersion with particles containing organic polymer (usually polyacrylate) and silicone. This is because silicone-containing polymer systems allow for achieving specific features of coating surface like e.g., water repellency without affecting its general performance [17]. A comprehensive review of synthesis and characterization of such hybrid silicone-acrylic dispersions as well as of coatings and films or powders produced from them has been published in 2015 [11]. It has been proved in a number of research papers both referred to in that review paper and published later that the unique properties of coatings like e.g., high surface hydrophobicity and water resistance combined with enhanced water vapour permeability and good mechanical properties can be achieved by applying methods (1) to (3) described above to synthesize hybrid dispersions containing silicones where monomer X is acrylic monomer and monomer Y is silicone monomer—see e.g., [18–21] for method (1), [22–24] for method (2) or [24–26] for method (3). If fluoroacrylic monomer was used as monomer X [27–30], the surface hydrophobicity of coatings could be improved even more. It was also proved in our earlier studies [31–33] that the application of method (1) to emulsion polymerization of methyl methacrylate in aqueous silicone resin dispersions led to stable "core-shell" silicone-poly(methyl methacrylate) hybrid dispersions which, after drying, produced corresponding "nanopowders" that were later used as very effective impact modifiers for powder coatings.

In the present study we investigated the effect of the approach to synthesis (method (1) or method (2) as defined above) on the properties of hybrid aqueous dispersions and corresponding coatings. Two different silicone resin dispersions and two different acrylic/styrene copolymer dispersions were used as starting media in which emulsion polymerization of acrylic and styrene monomers or silicone monomers respectively was conducted. The mass ratio equal to 1/3 of silicone part (SIL 1 or SIL 2) to acrylic/styrene part (ACR A or ACR B) in the synthesis was selected based on the assumption supported by the literature [11] that this proportion would be sufficient to observe the influence of the presence of silicone in the dispersion particle on the properties of hybrid dispersions and coatings. Further studies on the effect of SIL/ACR ratio on the properties of hybrid dispersions and coatings are ongoing and will be published soon.

#### **2. Materials and Methods**

#### *2.1. Starting Materials*

Octamethylcyclotetrasiloxane (D4) was obtained from Momentive (Waterford, NY, USA). Other silicone monomers: vinyltriethoxysilane (VTES) and methyltriethoxysilane (MTES)) were obtained from Evonik (Essen, Germany) under trade names Dynasylan® VTEO and Dynasylan® MTES. Surfactants dodecylbenzenesulphonic acid (DBSA) and Rokanol T18 (nonionic surfactant based on ethoxylated C16–C18 alcohols) were obtained from PCC Exol (Brzeg Dolny, Poland). Emulgator E30 (anionic surfactant based on C15 alkylsulfonate) was obtained from LeunaTenside GmbH (Leuna, Germany). Other standard ingredients used in the synthesis of dispersions (sodium acetate, sodium hydrocarbonate, potassium persulphate and aqueous ammonia solution) were obtained from Standard Lublin (Poland) as pure reagents. Biocide used to protect dispersions from infestation was Acticide MBS obtained from THOR (Wincham, UK). Starting acrylic/styrene copolymer dispersions (ACR A and ACR B) characterized by different *T*gs were supplied by Dispersions & Resins (D&R, Włocławek, Poland). Monomers applied in synthesis of ACR A and ACR B dispersions by D&R were butyl

acrylate (BA) obtained from ECEM, Arkema, Indianapolis, IN, USA, styrene (ST) obtained from KH Chemicals, Helm, Zwijndrecht, The Netherlands, acrylic acid (AA) obtained from Prochema, BASF, Wien, Austria, and methacrylamide (MA) obtained from ECEM, Arkema. Acrylic and styrene monomers were used as received as mixtures designated as A and B with compositions corresponding to compositions of monomers applied to synthesize dispersions ACR A and ACR B, respectively. Exact compositions could not be revealed due to commercial secret, but were appropriately designed to get *T*<sup>g</sup> of dispersion solids at a level of ca. +15 ◦C (ACR A) and of ca. +30 ◦C (ACR B). For structures of acrylic monomers-see Figure 2.

**Figure 2.** Structures of silicone monomers used in synthesis of silicone resin dispersions SIL 1 and SIL 2 and acrylic/styrene polymer dispersions ACR A and ACR B.

ACR A and ACR B dispersions were not neutralized after polymerization had been completed in order to ensure the low pH value (ca. 3) that was needed to conduct polymerization of silicone monomers in the process of synthesis of hybrid acrylic-silicone dispersions.

#### *2.2. Synthesis of Silicone Resin Dispersions and Hybrid Silicone-Acrylic and Acrylic-Silicone Dispersions*

Silicone resin dispersions (SIL 1 and SIL 2) were synthesized according to the procedure described in [31], although different functional silanes were used along with D4 as silicone monomers—see Figure 2 for the structures of these silicone monomers.

The compositions (wt.%) of mixtures of silicone monomers used in synthesis of SIL 1 and SIL 2 were as follows:


DBSA was used as surfactant and D4 ring-opening catalyst. The reaction that proceeded in the process of SIL 1 and SIL 2 synthesis was simultaneous hydrolysis of trifunctional ethoxysilanes and breaking of Si-O bond in D4 leading to the formation of partially crosslinked poly(dimethylsiloxane) containing unsaturated bonds originating from VTES (see Figure 3).

After distillation of ethanol under vacuum no free VTES or MTES were detected by gas chromatography (GC) in the resulting SIL dispersions, although small amounts of D4 (ca. 0.8%) and ethanol (ca. 0.2%) were still present.

**Figure 3.** Simplified structure of partially crosslinked silicone resin obtained in synthesis of SIL 1 and SIL 2 dispersions.

Hybrid silicone-acrylic dispersions SIL-ACR 1-A and SIL-ACR 1-B were synthesized by emulsion polymerization of mixtures of acrylic and styrene monomers A and B, respectively, in silicone resin dispersion SIL 1. Hybrid silicone-acrylic dispersions SIL-ACR 2-A and SIL-ACR 2-B were synthesized by emulsion polymerization of mixtures of acrylic and styrene monomers A and B, respectively, in silicone resin dispersion SIL 2. Compositions of acrylic and styrene monomers mixtures A and B corresponded to compositions of acrylic and styrene monomers in starting acrylic/styrene copolymer dispersions ACR A and ACR B. Polymerization was carried out at 78–79 ◦C for 5 h. After cooling to room temperature, dispersions were neutralized with 25% aqueous NH3 solution to reach pH ca. 6.0–6.5, then 0.15% of biocide was added and dispersions were filtered through 190 mesh net. Free acrylic and styrene monomers contents as tested by GC were <0.01%. No free VTES or MTES were detected by GC, although small amounts of D4 (ca. 0.4%) and ethanol (ca. 0.1%) were still present. Hybrid acrylic-silicone dispersions ACR-SIL A-1 and ACR-SIL A-2 were synthesized by emulsion polymerization of mixtures of silicone monomers 1 and 2, respectively, in acrylic/styrene copolymer dispersion ACR A. Hybrid acrylic-silicone dispersions (ACR-SIL B-1 and ACR-SIL B-2 were synthesized by emulsion polymerization of mixtures of silicone monomers 1 and 2, respectively, in acrylic/styrene copolymer dispersion ACR B. Compositions of silicone monomers mixtures 1 and 2 corresponded to compositions of silicone monomers in starting silicone resin dispersions SIL 1 and SIL 2. Polymerization was carried out at 88–89 ◦C for 4 h and then ethanol that was formed in hydrolysis of VTES and MTES was distilled off under vacuum for 3 h. After cooling to room temperature, dispersions were neutralized with 25% NH3 solution to reach pH ca. 6.0–6.5, then 0.15% of biocide was added and dispersions were filtered through 190 mesh net. No free VTES or MTES or acrylic and styrene monomers were detected by GC in the resulting dispersions, though small amounts of D4 (ca. 0.4%) and ethanol (ca. 0.1%) were still present.

It was essential that the composition and concentration of surfactants remained the same in SIL-ACR and ACR-SIL dispersions, so their properties (and properties of coatings obtained from them) could be compared.

#### *2.3. Characterization of Dispersions*

All dispersions were characterized by:


#### *2.4. Characterization of Coatings*

Coatings were produced from dispersions by applying them on glass (for testing contact angle, hardness, adhesion or water resistance), aluminium plates (for testing elasticity) or on steel plates (for testing impact resistance and cupping) using 120 μm applicator. Drying was carried out for 30 min at 50 ◦C and then the coatings were seasoned in a climatic chamber at 23 ◦C and 55% relative humidity (R.H.) for 72 h. Since no continuous coating could be obtained in this procedure for SIL-ACR 1-B and SIL-ACR 2-B, the relevant dispersions were dried at 8 ◦C for 2 h and then seasoned as above. The resulting coatings were characterized by:


#### *2.5. Characterization of Films*


#### **3. Results and Discussion**

#### *3.1. Properties of Dispersions*

Properties of hybrid silicone-acrylic (SIL-ACR) and acrylic-silicone (ACR-SIL) dispersions prepared with SIL/ACR *w*/*w* 1/3 ratio, starting silicone resin dispersions (SIL 1 and SIL 2) and starting acrylic/styrene copolymer dispersions (ACR A and ACR B), are presented in Table 1. All hybrid dispersions were mechanically stable, slightly opalescent white liquids with low viscosity, pH in the range 5.8–6.3 and solids contents close to 42%. Coagulum content was at a very low level –0.04%–0.38%. Blends of starting SIL and ACR dispersions at *w*/*w* 1/3 ratio were also made, but the resulting dispersions were not mechanically stable and did not produced continuous coatings at room temperature.

#### 3.1.1. Particle Size and Particle Size Distribution

For hybrid dispersions, particle size distribution was monomodal and rather narrow, although in most cases wider than that for starting SIL and ACR dispersions. Zeta potentials were all very low (i.e., very negative) which indicated good dispersion stability that was confirmed by mechanical stability tests.

The average particle size was distinctly higher for hybrid dispersions SIL-ACR than for starting SIL dispersion and almost the same for ACR-SIL than for starting ACR dispersion (see Figure 4) what could indicate formation of shell on starting SIL dispersion core particles during polymerization of ACR monomers and lack of formation of core-shell particle structure in the case of polymerization of SIL monomers in starting ACR dispersion. The comparison of particle size distribution patterns confirmed that assumption for ACR-SIL dispersions—see Figure 5a. As it can be seen in Figure 5b, in synthesis of SIL-ACR dispersions acrylic/styrene copolymer particles smaller than particles of starting SIL dispersion were probably formed along with core-shell SIL-ACR particles.

In general, average particle size was significantly higher for SIL-ACR dispersions than for ACR-SIL dispersions of the same composition of SIL and ACR parts—see Figure 6 where the particle size distribution of one of the SIL-ACR dispersions (SIL-ACR 2-B) and of the corresponding ACR-SIL dispersion (ACR-SIL B-2) is shown. The reason for that was higher particle size of starting SIL dispersions than for starting ACR dispersions.


#### *Coatings* **2019** , *9*, 25

**Table 1.** all properties were determined

Properties of hybrid

silicone-acrylic

 for dispersions

 neutralized

 after

polymerization

 while ACR dispersions

 before

neutralization

 (with pH ca. 3.0) were used in synthesis

 (SIL-ACR) and ACR-SIL dispersions

 and of starting SIL and ACR dispersions.

 In the case of starting ACR dispersions

**Figure 4.** Comparison of average particle size of SIL and ACR dispersions and hybrid SIL-ACR and ACR-SIL dispersions.

**Figure 5.** Comparison of particle size distribution patterns of hybrid ACR-SIL A-2 dispersion and starting ACR A dispersion (**a**) and of hybrid SIL-ACR 2-A dispersion and starting SIL 2 dispersion (**b**). X-axis is logarithmic.

**Figure 6.** Comparison of particle size distribution patterns for SIL-ACR and ACR-SIL dispersions of the same composition of SIL and ACR parts. X-axis is logarithmic.

#### 3.1.2. Particle Structure

In Figure 7 the structure of hybrid dispersion particles of SIL-ACR and ACR-SIL dispersions determined by TEM is shown. As can be seen in Figure 7a, in the case of SIL-ACR dispersion coalescence of particles proceeded during testing, so the TEM image shows a tiny piece of film rather than the single particle, but it is clear that well defined silicone resin particles (lighter shade) are surrounded by acrylic/styrene copolymer phase (darker shade). Individual particles can be identified better in Figure 7b where lower magnification was used and it can be concluded that a kind of "fruit cake" particle structure where a few "cores" made of one polymer are surrounded by continuous mass of the other polymer was formed during polymerization of ACR monomers in SIL dispersion. In the case of ACR-SIL, dispersion coalescence of particles during testing also proceeded. While both individual particles and aggregates of silicone resin particles and acrylic/styrene copolymer particles were present, it was also possible to identify in TEM images abundant single particles of specific structure shown in Figure 7c. In this structure kinds of spheres made of silicone resin (lighter shade) were embedded in the mass of acrylic/styrene copolymer (darker shade). It can be anticipated that in the course of synthesis of ACR-SIL hybrid dispersions silicone monomers penetrated into acrylic/styrene copolymer particles and after completion of polymerization a kind of sphere of silicone resin was formed because of lack of compatibility of acrylic/styrene copolymer and silicone resin. Such a particle structure called an "embedded sphere" has been found also earlier in polyurethane-acrylic/styrene hybrid dispersions [4].

**Figure 7.** Structure of hybrid dispersion particles settled on the micromesh net as determined by transmission electron microscopy (TEM): (**a**) SIL-ACR dispersion, higher magnification, (**b**) SIL-ACR dispersion, lower magnification, (**c**) ACR-SIL dispersion, higher magnification. Lighter shade represents silicone resin and darker shade–acrylic/styrene copolymer.

Lack of formation of core-shell ACR-SIL hybrid particles in the course of polymerization of silicone monomers in acrylic/styrene copolymer dispersion could have been expected since it was clear from the review of available literature on that subject [11] that only if special approaches were applied to synthesis (e.g., functionalization of acrylic particle surface with silane and hydrolysis of alkoxysilane groups prior to polymerization [22]) the particles with acrylic polymer core and silicone shell could be obtained.

#### 3.1.3. Minimum Film-Forming Temperature (MFFT)

As it can be seen in Figure 8 MFFT values determined for ACR-SIL hybrid dispersions were much lower than for starting ACR dispersion and lower than for SIL-ACR dispersions of the same SIL and ACR parts composition what can be explained by the fact that only a fraction of particles of ACR-SIL dispersion hybrid structure exhibited a hybrid morphology shown in Figure 6b and the presence of separate silicone resin particles resulted in lower MFFT.

**Figure 8.** Comparison of MFFT determined for hybrid dispersions SIL-ACR and ACR-SIL. MFFT of starting ACR dispersion used for synthesis of ACR-SIL dispersions is also shown.

#### 3.1.4. Glass Transition Temperature (*T*g)

DSC results showed that hybrid dispersion solids usually exhibited two *T*gs: one corresponding to SIL part at c.a. −120 ◦C and the other corresponding to ACR part in the range of ca. 15–30 ◦C, depending on the *T*<sup>g</sup> of starting acrylic/styrene copolymer—see Figure 9 where DSC patterns determined for starting SIL and ACR dispersions and for SIL-ACR and ACR-SIL dispersions having the same composition of ACR and SIL parts are presented.

**Figure 9.** Differential scanning calorimetry (DSC) patterns determined for starting SIL and ACR dispersions and for SIL-ACR and ACR-SIL dispersions having the same composition of ACR and SIL parts.

Only for two dispersions (ACR-SIL A-2 and ACR-SIL B-2) just one *T*g was detected at around 16 and 20 ◦C, respectively, what suggested that in the case of these two dispersions the particle structure was rather uniform and no separate silicone resin particles were formed. This phenomenon can be explained by the fact that in these two dispersions silicone monomers (D4 + ethoxy-functional silane) mixture that was polymerized in acrylic/styrene copolymer dispersion contained much more VTES (more polar) and did not contain MTES (less polar), so penetration into acrylic/styrene copolymer particles was easier and grafting of VTES on acrylic/styrene copolymer and formation of "embedded sphere" structures shown in Figure 7c were much more probable.

It was also interesting that *T*gs of SIL and ACR parts of all hybrid dispersion solids where two glass transitions were detected were significantly lower than *T*gs of starting SIL and ACR dispersions solids. Decrease in *T*g of ACR part can be explained by plasticizing effect of modification with silicone resin. However, in order to clarify why decrease in *T*g of SIL partly occurred, more insight is needed to the processes which took place in the course of both silicone monomers polymerization in acrylic/styrene copolymer dispersion and acrylic/styrene monomers polymerization in silicone resin dispersion. The key assumption (confirmed by the hybrid particle structures) is that in hybrid dispersion particles silicone resin particles are "trapped" within a mass of acrylic/styrene copolymer, so D4 and higher oligodimethylsiloxane cycles (e.g., D5) which are always present in SIL dispersions [43] and are also formed in synthesis of hybrid ACR-SIL dispersions are also "trapped" and therefore cannot be released during drying and may plasticize the silicone resin contained in dispersion solids. That "trapping" of silicone resin in acrylic/styrene copolymer part of hybrid dispersion particles should be more distinct if the SIL part contained more VTES because of possibility of grafting the decrease in *T*g should be more distinct for hybrid dispersions ACR-SIL 1-A and ACR-SIL 1-B than for hybrid dispersions ACR-SIL 2-A and ACR-SIL 2-B. Comparison of the relevant *T*g values in Table 1 confirmed that this was actually the case.

#### *3.2. Properties of Coatings and Films*

Properties of coatings and films obtained from hybrid silicone-acrylic (SIL-ACR) and acrylic-silicone (ACR-SIL) dispersions prepared with SIL/ACR *w*/*w* 1/3 ratio, starting silicone resin dispersions (SIL 1 and SIL 2) and starting acrylic/styrene copolymer dispersions (ACR A and ACR B) are presented in Table 2. Some hybrid dispersions and starting silicone resin dispersions did not form mechanically strong continuous coatings or films, but certain properties like e.g., contact angle or % swell could be determined by casting layers which, after drying, formed mechanically weak coatings or films.

It is essential that for all hybrid dispersions the key coating properties that were expected to improve as compared to acrylic/styrene copolymer dispersions (contact angle, water vapor permeability and water resistance) actually did improve significantly. Mechanical properties of coatings (e.g., impact resistance or elasticity) also improved, but hardness decreased what could be expected. The same trend was reflected in film properties—increase in elongation at break was accompanied by a decrease in tensile strength.

#### 3.2.1. Surface Properties

The high contact angle of coatings is important since it means high surface hydrophobicity and, consequently, lower water uptake and lower dirt deposition [5]. As can be seen in Table 2, all coatings obtained from hybrid SIL-ACR and ACR-SIL dispersions showed high contact angles in the range of 80–90◦ while contact angles recorded for films obtained from starting ACR dispersions were quite low (ca. 30◦). It is worth to note that contact angles recorded for coatings produced from ACR-SIL dispersions were generally higher than those recorded for coatings produced from SIL-ACR dispersions (see Figure 10) what indicates that in the former case more silicone migrated to the coating surface.

Migration of silicone to the coating surface observed for coatings containing silicones was described in the earlier papers, e.g., [32,44,45] and was fully confirmed by XPS also for coatings obtained from SIL-ACR and ACR-SIL hybrid dispersions. In Figure 11 the percentage of Si in the layers close to coating surface as determined by XPS for hybrid SIL-ACR and ACR-SIL dispersions is plotted against distance from the surface. It is clear from Figure 11 that in the coatings obtained from hybrid dispersions silicone migrated to coating surface and that migration was different for coatings obtained from ACR-SIL dispersions than for those obtained from SIL-ACR dispersions, most probably due to "trapping" of silicone resin in acrylic/styrene copolymer particles in the latter coating.


*Coatings* **2019**, *9*, 25

**Table 2.**

Properties of coatings and films made from hybrid SIL-ACR and ACR/SIL dispersions

 and of starting SIL and ACR dispersions.

 N.A. = Not Applicable

ACR-SIL

B-2

82

 28.0

0(S0)

Medium

3.9

 19.6

 11.0

 passed

 0.058

 5

 31

 991

 3.2

 947

whitening

**Figure 10.** Comparison of contact angle values determined for coatings obtained from starting SIL and ACR dispersions and corresponding hybrid SIL-ACR and ACR-SIL dispersions having the same composition of ACR and SIL parts. Contact angle determined for starting acrylic/styrene copolymer dispersion (ACR B) is also shown.

**Figure 11.** Decrease in Si content with distance from coating surface determined by XPS for coatings obtained from SIL-ACR and ACR-SIL dispersions.

#### 3.2.2. Water Resistance

Good water resistance of architectural paints is crucial since it ensures longer life of the paint and better comfort of the building walls (lack of water uptake) if combined with high water vapour permeability. Therefore, determination of the water resistance of coatings produced from dispersions which are intended to be applied as binders for architectural paints seems to be very important test. In our investigations we measured water resistance of coatings obtained from starting ACR dispersions and from SIL-ACR and ACR-SIL dispersions using our own method partly described in Section 2.4 and the results were assessed based on EN ISO 4628-2 [40]. All coatings made from hybrid dispersions exhibited better water resistance than those produced from starting ACR dispersions and it was significantly better for coatings obtained from ACR-SIL dispersions than from SIL-ACR dispersions—see Figure 12 where photos of coatings produced from different dispersions and left under water for 6 days are shown.

**Figure 12.** Comparison of water resistance of starting acrylic/styrene copolymer dispersion. ACR A (**a**), acrylic-silicone dispersion (ACR-SIL A-1) (**b**) and corresponding silicone-acrylic dispersion SIL-ACR (**c**). Samples were kept under water for 6 days. ACR A—deterioration of coating occurred, ACR-SIL A-1—coating did not change except for light whitening, SIL-ACR 1-A—coating changed significantly– numerous small bubbles.

#### 3.2.3. Swell in Water and in Toluene

As can be concluded from Table 2 percent of swell in water was very similar for all films (despite of differences in water resistance of coatings) and was quite low (ca. 20%) while swell in toluene that can be considered as a measure of crosslinking density (higher swell means lower crosslinking density) was much higher for films made from ACR dispersions than for films made from SIL dispersions, and also much higher for films made from hybrid ACR-SIL dispersions than for films made from SIL-ACR dispersions—see the relevant comparison in Figure 13.

**Figure 13.** Comparison of % swell in toluene determined for starting silicone resin dispersion (SIL-1), starting acrylic/styrene copolymer dispersion (ACR B) and hybrid dispersions ACR-SIL B-1 and SIL-ACR 1-B having the same composition of SIL and ACR parts.

The difference between crosslinking density of films (i.e., also for coatings) made from ACR-SIL and SIL-ACR dispersions having the same composition of ACR and SIL parts can be explained by a higher possibility of grafting of acrylic/styrene monomers on silicone resin than of grafting VTES on acrylic/styrene copolymer. Another reason can be a higher possibility of trapping of partly crosslinked silicone resin inside particles made of acrylic/styrene copolymer in the case of films made from SIL-ACR dispersions than in the case of films made from ACR-SIL dispersions—see the discussion of hybrid dispersions particle structures contained in Section 3.1.2.

#### 3.2.4. Water Vapour Permeability

As has already been pointed out in Section 3.2.2, good architectural paint should exhibit not only good water resistance, but also good water vapour permeability. This positive combination of properties can be achieved in practice only for paints based on silicone-acrylic binders because silicone polymers are characterized by good permeability of gases due to high mobility of poly(dimethylsiloxane) chains. It was proved in our study that coatings produced from hybrid SIL-ACR and ACR-SIL dispersions showed higher water vapour permeability than those produced from starting ACR dispersions—see Figure 14.

**Figure 14.** Comparison of water vapour permeability determined for starting acrylic/styrene copolymer dispersion (ACR A) and hybrid dispersions ACR-SIL A-1 and SIL-ACR 1-A having the same composition of SIL and ACR parts.

It can be noted from the results presented in Figure 14 that water vapour permeability was better for coatings obtained from SIL-ACR dispersions than from ACR-SIL dispersions, probably because of differences in coating structure that resulted from differences in dispersion particle structure.

#### 3.2.5. Mechanical Properties

If the results of testing the mechanical properties of coatings and films produced from hybrid ACR-SIL and SIL-ACR dispersions presented in Table 2 are compared with mechanical properties of coatings produced from starting ACR dispersions, it is clear that modification with silicone led generally to less brittle coatings, especially in the case of starting dispersion ACR A. The most spectacular difference was in the (direct) impact resistance of coatings—see Figure 15.

For coatings and films produced from starting dispersion ACR B and hybrid coatings and films where ACR B composition of monomers was applied in synthesis of the relevant dispersions, the results of mechanical tests were much less convincing, presumably because *T*<sup>g</sup> of ACR B was quite high (over 30 ◦C). Cupping test results were good for all coatings and in direct elasticity measurements, only coatings produced from starting dispersion ACR B failed. Elongation at break increased for some films made from hybrid dispersions as compared to films made from starting ACR dispersions and decreased for some others (specifically for these produced from hybrid dispersions with particles having ACR B composition of ACR part) and tensile strength decreased for all films where this could be expected taking into account plasticizing effect of silicone resin. Much higher elongation at break and much lower tensile strength observed for films made from ACR-SIL dispersions than from SIL-ACR dispersions can be explained by a different supramolecular structure of films that results from different morphology of hybrid dispersion particles (see Figure 7) that coalesce to produce these films in the process of air-drying of dispersions.

**Figure 15.** Comparison of impact resistance (direct) determined for coatings obtained from starting acrylic/styrene copolymer dispersion (ACR A) and hybrid dispersions ACR-SIL A-1 and SIL-ACR 1-A having the same composition of SIL and ACR parts.

#### **4. Conclusions**

Simultaneous synthesis of aqueous silicone-acrylic and acrylic-silicone hybrid dispersions (SIL-ACR and ACR-SIL) by (1) emulsion polymerization of acrylic/styrene monomers (BA, ST, KA and MA) mixtures of different composition (ACR A and ACR B) in aqueous dispersions of silicone resins of different composition (SIL 1 and SIL 2) and (2) emulsion polymerization of silicone monomers (D4, VTES and MTES) mixtures of different composition (SIL 1 and SIL 2) in aqueous dispersions of acrylic/styrene copolymers (ACR A and ACR B) was successfully conducted. Hybrid dispersions had good mechanical stability, low minimum film-forming temperature and particle size in the range of 100–150 nm, narrow particle size distribution, and contained very little of coagulate. TEM investigation of hybrid dispersions particle structure revealed that particles of SIL-ACR dispersions exhibited "fruit cake" structure while particles of ACR-SIL dispersions showed "embedded sphere" structure. For most of the dispersions two separate *T*gs of dispersion solids (one for SIL part and the other for ACR part) that were detected by DSC were lower than *T*gs of corresponding starting SIL and ACR dispersions while single *T*<sup>g</sup> was detected for two of them. These differences were explained by differences in dispersion particle structure.

Most of the hybrid dispersions formed mechanically strong continuous coatings and films. As compared to coatings obtained from starting ACR dispersions, those obtained from hybrid dispersions showed much higher contact angles, much better water resistance and water vapour permeability and exhibited much better impact resistance. Different coating properties were observed when coatings were produced from SIL-ACR and ACR-SIL dispersions having the same composition of ACR and SIL parts, which most probably resulted from different structure of dispersions particles. Films produced from hybrid dispersions were less brittle than those produced from starting ACR dispersions. Determinations of % swell in toluene measured for films produced from hybrid dispersions revealed the difference between crosslinking density of films (i.e., also for coatings) made from ACR-SIL and SIL-ACR dispersions having the same composition of ACR and SIL parts, which was explained by higher possibility of grafting of acrylic/styrene monomers on silicone resin than of grafting VTES on acrylic/styrene copolymer. The authors believe that the selected hybrid dispersions described in this paper can be applied as binders in the formulation of architectural paints that will be characterized by high water resistance and high surface hydrophobicity combined with high water vapour permeability.

**Author Contributions:** Conceptualization, J.K. and J.T.; Methodology, J.T., W.D., I.O.-K., A.K., K.S. and J.P.; Investigation, J.T., W.D. and A.K.; Writing-Original Draft Preparation, J.K.; Writing-Review & Editing, J.T. and W.D.; Visualization, J.P., W.D. and J.T.; Supervision, J.K., D.W. and M.W.; Project Administration and Funding Aquisition, J.K.

**Funding:** This research was funded by Polish State R&D Centre (NCBiR, No. PBS/B1/8/2015).

**Acknowledgments:** The authors wish to thank Piotr Bazarnik from Warsaw University for conducting TEM studies, and Janusz Sobczak from Polish Academy of Sciences for conducting XPS studies. The assistance of Colleagues from the Industrial Chemistry Research Institute in testing mechanical properties of films, water vapour permeability and dispersion particle size distribution, is also acknowledged.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Assessing of New Coatings for Iron Artifacts Conservation by Recurrence Plots Analysis**

#### **Paola Roncagliolo Barrera 1,\*, Francisco Javier Rodríguez Gómez <sup>1</sup> and Esteban García Ochoa <sup>2</sup>**


Received: 16 November 2018; Accepted: 23 December 2018; Published: 26 December 2018 -

**Abstract:** Cast iron has stood for centuries of invention. It is a very versatile and durable material. Coating systems are a low-maintenance protection method. The purpose of this research is to increase the Paraloid coating's resistance when applied to iron in high humidity atmospheres, with the addition of caffeine (1,3,7-dimethylxanthine) and nicotine (S)-3-(1-methylpyrrolidin-2-yl) pyridine as corrosion inhibitors; the resistance of protection versus exposure time will be evaluated by using electrochemical noise. A statistical analysis of the electrochemical noise signals was carried out. Recurrence plots were used as a powerful tool in the analysis to complement the data obtained and they predicted the evaluation of coatings behaviors performance versus time. The outcomes show that the addition of inhibitors increases and improves the performance as a temporary protection of Paraloid and that protection in high relative humidity was improved. Recurrence plots and parameter quantification show the variances in the surface corrosion dynamics.

**Keywords:** cast iron; Paraloid; natural inhibitor; electrochemical noise; recurrence plots

#### **1. Introduction**

Temporary protection systems based on coatings are usually used in iron objects preserved in museums. The protective layer insulates metal surfaces from moisture, air pollutants, acids, etc. This method provides a passive protection against corrosion [1]. Acrylic systems, such as Paraloid B-72, a copolymer of ethyl methacrylate and methyl methacrylate, have been extensively used for more than twenty years in restoration work as an adhesive and in conservation as a protective film. One of its most notable properties is that, according to Feller [2], Paraloid B-72 is one of the few polymers that has an expected duration of 100 years, and that, under average museum conditions, it can be stored without changes in its transparency. Furthermore, its original solvent is soluble for more than 200 years (according to a projection based on accelerated aging studies). It is a fairly stable resin, nevertheless, its polymer chains deteriorate quickly if exposed to ultraviolet radiation, and its high water permeability decreases its useful life, which is well-known in the industrial field [3].

Another method employed for anticorrosive protection is the use of corrosion inhibitors, which contrast to industrial applications, since, in metal preservation, the addition of the inhibitor to the electrolyte is not mainly used. By definition, inhibitors are used in closed or controlled environments, while objects are exposed to atmospheric conditions that are difficult to control [4]. On the contrary, inhibitors are applied directly on the surface to produce modifications or they are mixed with the varnish. This change in concept and application is important for research on the use of inhibitors in the field of cultural heritage [5–7].

Some inhibitors are being widely used in the preservation and restoration treatments of copper, iron, and silver alloys [8]. For iron alloys, tannic acid and benzotriazole (BTA) are used, tannins being the most used [9]. However, the appearance of a complex of black tannate on the surface is a disadvantage, which is why the use of natural-origin corrosion inhibitors has been proposed [10].

Most of these compounds contain nitrogen, sulfur, and oxygen with a free pair of electrons; additionally, they have aromatic systems [11–13]. These compounds can act on the metal surface through adsorption, by blocking the active sites or by forming a protective layer that can reduce the corrosion rate. N, O, and S are atoms present in the heterocyclic compounds, and are oriented towards the sites where the adsorption probably occurs, because of the availability of a free electron pair [14]. Existing data show that most organic inhibitors act by adsorption at the metal/solution interface, specifically by the displacement of water molecules, which forms a compact film that works as a barrier.

Coating modifications used in the conservation of metallic artifacts are proposed, adding caffeine and nicotine as alternative inhibitors to increase the performance of Paraloid B-72 in high-humidity environments. Improvements are discovered in the mechanism of the metal coating interface through inhibition. Cast iron was selected for evaluation since it shows significant modifications in the corrosion mechanism when atmospheric conditions change. The evaluation of the coatings was carried out through electrochemical noise (EN) with atmospheric corrosion monitors in conditions of 40% and 98% relative humidity (RH), which allows obtaining results in relatively short exposure times due to the high sensitivity of this technique [15–17].

The electrochemical noise technique (the study of the spontaneous fluctuations in voltage and current of the electrode in function of the time) has been employed to determine the anticorrosive resistance in coated substrates [18–22]. This technique allows discerning the corrosion mechanisms that occur through the film, as well as the resistance that it presents. This is why it has proven to be an effective method for comparing coatings performance [23–25]. Time series can be analyzed by statistical and spectral methods, to estimate characteristics that generally describe the behavior of the coated metal [26]. The phenomena that occur on the metal surface in atmospheric conditions are not precisely stationary systems where the dynamics do not change as a function of time. Accordingly, the study of chaotic systems is justified by the existence of many phenomena that have a temporal evolution governed by perfectly deterministic models. An alternative to the statistical analysis of the phenomena that behave in a complex or non-regular way is the recurrence of plots method which allows evaluating the dynamic behavior when identifying the changes in the signals obtained.

The main aim of this research is the study of the current electrochemical noise signal to characterize the dynamics of the corrosion process of the coated iron when relative humidity conditions are modified. Qualitative and quantitative analysis is proposed by recurrence plots (RP) as a tool that promotes the use of the technique to assess coatings, optimizing the information obtained.

#### **2. Materials and Methods**

#### *2.1. Materials*

Class 25 gray iron was used as working electrode, the chemical composition was set to ensure that variations in composition and microstructure did not alter the electrochemical results. To perform the analysis, the sample was fixed with two parallel surface areas of 5 cm2. After a suitable preparation, an analysis process was carried out by arc/spark optical emission spectroscopy (OES) analyzer. The equipment used was a SpectroLab Spectrometer brand model LAVWA18B analytical instrument (Mahwah, NJ, USA). The chemical composition obtained is shown in Table 1.

**Table 1.** Chemical composition determined by an arc/spark OES.


The caffeine and nicotine used were Sigma-Aldrich brand (Saint Louis, MO, USA) and were analytical grade reagents with ≥99% purity. Paraloid B72 was provided by Carl Roth (Schoemperlenstraße, Karlsruhe, Germany) and the acetone used was provided by J. T. Baker RA (Center Valley, PA, USA). Paraloid B72 was dissolved in acetone 5% (*w*/*v*) and applied with Kolinsky Brush No. 7, Winsor and Newton (Glasgow, Scotland, UK). To set the concentration on the surface, the Paraloid and the compounds were mixed. For each 100 mL of coating, 5 g of Paraloid B72 and 50 mg of the compound were dissolved in acetone. A total of 0.1 mL of the mixture was applied per square centimeter of area. The inhibitor concentration on the surface was set at 50 <sup>μ</sup><sup>g</sup> · cm−2. Both compounds were added in the preparation to control the amount of inhibitor on the surface, and not on the surface as a pretreatment (as they are most commonly applied). Only one coating layer was applied and placed into a desiccator for 24 h; five linear measurements on the dry film were performed, and the typical standard deviation was calculated. The dry film thickness was set at 30 μm with a deviation of ±5 μm carried out by an Elcometer coating thickness gauge. Measurements were carried out under the ASTM B499-09 Standard [27].

#### *2.2. Sensor Construction (Atmospheric Corrosion Monitor, ACM)*

An electrochemical monitor (ACM) was assembled to assess the performance of the coatings in atmospheric conditions. The design consists of an arrangement of three identical electrodes: two electrodes as working electrodes (W1, W2) and a central electrode functions as a "pseudo" reference electrode (Ref). The metal plates were cut to 2.5 × 2.5 cm2 with a thickness of 0.4 cm. To avoid the short-circuit of the metal electrodes, a 0.1 mm thick Teflon plate was placed between each of them, as shown in Figure 1.

**Figure 1.** ACM monitors construction.

Each of the plaques was welded to a vulcanized cable, to ensure the continuity of the signal. To maintain the union of the set of plates, Teflon adjusted was used. This arrangement was isolated with epoxy glass resin resistant to chemical attack, as shown in Figure 2.

**Figure 2.** ACM monitor assembled.

Once the monitor is assembled, only a transverse surface was exposed to corrosion. The exposed surface is rough and was polished to a mirror finish. Then it is degreased with acetone and air-dried. This type of corrosion monitor has been applied frequently according to the literature [28,29].

#### *2.3. Relative Humidity*

The monitor was placed in a glass desiccator, as shown in Figure 3. The relative humidity was modified with different saline solutions supersaturated under the ASTM E-104 standard [30]. The humidity measures selected during the experimentation phase were 40% RH using potassium carbonate (K2CO3) and 98% RH using potassium sulfate (K2SO4).

**Figure 3.** An illustrative diagram of ACM into the desiccator.

The humidity was monitored with a thermo-hygrometer by Instruments, and a thermostat was used to maintain a constant temperature of 25 ◦C. Two humidity conditions were selected: 40% to evaluate a lower thickness of water on the surface—a percentage of humidity that museums around the world have reported suitable in their exhibitions for the optimal durability of Paraloid and, consequently, the conservation of metal—and 98% relative humidity to have a thickness of water that could be considered as electrochemical corrosion and is the extreme condition.

#### *2.4. Electrochemical Noise Measurements*

Voltage and current of electrochemical noise (EN) were measured using a sampling frequency of 0.5 spots per second with 2048 measurements, and a frequency of 1 Hz in order of 0.25 MHz; a measurement was conducted every two hours for 48 h. The measurements were posted out with a zero resistance ammeter (ZRA) by Gill AC 1123 (ACM Instruments, Cumbria, UK); a Faraday cage was used to keep out external interference (static and electromagnetic influences). The resolution was 1 pA for current and 1 μV for voltage measurements. Trend removal was done by subtracting polynomial method of the raw data. The time series were obtained from the response of the system, which allows not only the analysis of the behavior of the inhibition layer but also the determination of its kinetic mechanism. The experiments for each condition were conducted in triplicate to guarantee its reproducibility and reliability.

#### **3. Results**

#### *3.1. Electrochemical Noise Analysis*

Electrochemical noise signals are registers of current and potential over time. According to this, the random or deterministic behavior of temporal records is conditioned by the possibility of establishing a relationship between the different parameters that govern the corrosion process that is being studied. The comparative analyses of the electrochemical noise signals obtained from a series of potential or current in function of the acquisition time are shown below.

Figure 4 shows the time series of potential and current with the removal of the trend after 24 h of exposure in relative humidity of 40% and 98%. No current density has been shown since the area factor of the electrode is 1 cm2. To perform a quantitative comparison, the standard deviation of the removed trend values is contained. The blank metal (Figure 4a) to 40% relative humidity presents current and potential lower values (1.8 × <sup>10</sup>−<sup>9</sup> mA, 0.02 mV), compared to the bare metal (Figure 4b) at 98% relative humidity (5.33 × <sup>10</sup>−<sup>7</sup> mA, 0.74 mV). Changes are observed in the fluctuations in both signals obtained as a function of the increase in relative humidity, as a consequence of increasing the thickness of the water film and consequently, the interaction present in the microcells becomes more intense. When the surface is covered with Paraloid (Figure 4c) at 40%, the current and potential changes are not apparently large (1.43 × <sup>10</sup>−<sup>8</sup> mA, 0.23 mV) compared to the bare metal. But for Paraloid (Figure 4d) at 98%, current output decreases (4.43 × <sup>10</sup>−<sup>7</sup> mA) and changes in the amplitude of the potential signal (1.55 mV) are observed, as a consequence of presenting localized active sites where the Paraloid loses protection by absorbing water and the iron corrosion process may occur. When the inhibitors are added to the Paraloid films, modifications are observed as shown in Figure 5.

**Figure 4.** Cast iron's time series in current and potential for (**a**) blank at 40% RH, (**b**) blank at 98% RH, (**c**) Paraloid B72 at 40% RH, and (**d**) Paraloid B72 at 98% RH.

When caffeine is added to Paraloid (Figure 5a) at 40% RH, both current and potential responses decrease (5.06 × <sup>10</sup>−<sup>9</sup> mA, 0.36 mV) and (Figure 5b) for 98% RH (1.75 × <sup>10</sup>−<sup>8</sup> mA, 1.03 mV) almost an order of magnitude, while nicotine films (Figure 5c,d) have smaller values in current and potential at 40% RH (1.69 × <sup>10</sup>−<sup>9</sup> mA, 0.34 mV) as at 98% RH (1.68 × <sup>10</sup>−<sup>9</sup> mA, 0.93 mV) compared to Paraloid without inhibitor. This decrease in the amplitude and response in current and potential reveals the activity of these molecules on the superficial phenomenon of corrosion.

In general, the noise signal apparently has certain changes depending on the conditions, but performing the visual analysis only for a time series is not enough to measure the gain in the impedance of the coating or its performance as a mapping of time.

**Figure 5.** Cast iron's time series in current and potential for (**a**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> with caffeine at 40% RH, (**b**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> with caffeine at 98% RH, (**c**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> with nicotine at 40% of RH, and (**d**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> with nicotine at 98% RH.

This behavior is much more evident when calculating the value of noise resistance (*R*n), which is inversely proportional to the intensity of the corrosive attack [31]. Therefore, it is necessary to establish a statistical treatment of the signal to be able to recognize the changes that are submitted in an objective direction and the effect the compounds have along with the Paraloid.

First, the most common statistical parameters, which are the mean and the standard deviation of the signals both in current and in potential, are obtained. Founded on these two statistical parameters, as described in the literature, noise resistance (*R*n) is limited. In many investigations, it has been associated with polarization resistance, which is known to be inversely proportional to the corrosion rate when the mechanism is controlled by charge transfer [32]. *R*<sup>n</sup> is defined based on the following Equation (1) [33–35]:

$$R\_{\rm II} = \frac{\sigma\_{\rm E}}{\sigma\_{\rm I}} \tag{1}$$

where σ<sup>I</sup> is the standard deviation in current, whereas σ<sup>E</sup> is the standard deviation in potential. The second parameter that can be obtained is the so-called localization index (LI), which indicates changes in the corrosion mechanisms, whether homogeneous, mixed or localized, and shows how localized is the attack to the surface of the metal in the medium. This parameter can only take values between 0 and 1, with 0 being a homogeneous attack and 1 being an extremely localized attack [36,37]. The LI is defined based on Equation (2) and only the current noise signal is necessary to determine it. As shown, (*X*I) is the value of the mean in current *y* (σI) the standard deviation in current [38]:

$$\text{LI} = \frac{\sigma\_{\text{I}}}{\sqrt{\sigma\_{\text{I}}^2 + X\_{\text{I}}^2}} \tag{2}$$

Figure 6 shows summarizes the statistical analysis of the noise signals of both relative humidity, where both the noise resistance *R*n and LI are determined.

**Figure 6.** 1/*R*n (proportional to the corrosion rate) and the location index (LI) according to the times for different coatings in 40% and 98% relative humidity.

Figure 6 shows the statistical analysis of the noise signals in both relative humidities in the function of the exposure time in hours. It is important to note that the logarithm of the reciprocal of *R*n is shown to easily appreciate the changes that are generated on the surface with respect to the rate of corrosion. Both *R*n and LI are plotted on a logarithmic scale, which works well for comparison purposes, since the differences between the conditions evaluated are very large. In the first place, it is evident that, for both relative humidity, the bare material presents the greatest possible corrosion, as expected. In the relative humidity of 40%, a value of 1/*R*<sup>n</sup> is presented in the order of 10−<sup>7</sup> (Ω·cm2) <sup>−</sup>1, while for the relative humidity of 98%, it is 10−<sup>6</sup> (Ω·cm2) <sup>−</sup>1, which is a higher order of magnitude. This result is logical because, in this relative humidity, a continuous water film is formed over the entire surface, resulting in a corrosion mechanism controlled by water and dissolved oxygen, while at 40% there is an ohmic type of control. This clearly shows how humidity is a critical factor in relation to the deterioration of gray iron metal parts. Subsequently, the presence of a physical barrier—the Paraloid B72—clearly decreases the values of 1/*R*n, in the order 10−<sup>9</sup> (Ω·cm2) <sup>−</sup><sup>1</sup> for 40% RH, where it is known that the Paraloid has good performance. However, the Paraloid does not present good protection against high relative humidity, because it tends to allow oxygen and water/moisture, resulting in a value of 1/*R*n that diminishes two orders of magnitude, up to values of 10−<sup>7</sup> (Ω·cm2) <sup>−</sup><sup>1</sup> for a 98% relative humidity.

This is an important component of this statistical analysis, where it is possible to watch the performance of caffeine and nicotine as corrosion inhibitors. Caffeine has an approximate value of 10−9.5 (Ω·cm2) <sup>−</sup><sup>1</sup> for 40% relative humidity and 10−8.5 (Ω·cm2) <sup>−</sup><sup>1</sup> to 98% relative humidity, while nicotine causes much lower value, of the order of 10−<sup>10</sup> (Ω·cm2) <sup>−</sup><sup>1</sup> to 40% relative humidity and 10−9.5 (Ω·cm2) <sup>−</sup><sup>1</sup> to 98% relative humidity. It is evident how nicotine has a better efficiency as a corrosion inhibitor in both relative humidity.

The location index (LI) also shows notable changes in the different conditions assessed. For a 40% humidity, the average location index of the substrate is 10−1.22 and for the Paraloid it is 10−1.15. The LI values in the presence of inhibitor are modified and reach 10−1.76 with caffeine and 10−2.52 with nicotine. For a relative humidity of 98%, the trend is modified for bare metal and Paraloid, where the bare substrate has an LI of 10−1.03 (mixed mechanism), with changes after 32 h at an LI of 10−0.76, which is identified as a localized mechanism. The change of mechanism is affected by the presence of corrosion products on the surface, which locally changes the reactions. The Paraloid presents an intermediate LI value of 10<sup>−</sup>1.31, which represents a mixed mechanism that tends to be localized by the hydration presented by such film. With the addition of caffeine, the LI is approximately 10−1.52 and with nicotine, it is 10<sup>−</sup>2.04.

In other words, when the surface is bare or with Paraloid, the attack mechanism tends to be from mixed to localized as a function of time, while the presence of the inhibitor is manifested by a mixed mechanism for caffeine and a homogeneous one for nicotine. The resistance of Paraloid depends on the relative humidity, which leads to a mixed mechanism for caffeine, presuming that the places where it was not present would be more active due to the water absorption of Paraloid.

It should be noticed that the bare metal has the highest localization index and that the presence of caffeine and nicotine as inhibitors, mainly, reduced the LI value, showing the formation of a homogeneous film that prevents a localized approach. It can also be understood that the proposed used of caffeine and nicotine considerably increases the level of protection and modifies the morphology of the approach.

The purpose of the statistical analysis of the time series is to demonstrate the nature of the corrosion phenomenon. These analysis models are traditionally linear. The ease of the tools is the main line of reasoning in favor of linearity. Yet, theoretically, it is hardly justifiable that this phenomenon of atmospheric corrosion presents a linear behavior. The study of non-linearity is complex, but it is necessary for this type of complex systems. Thus, it is essential to introduce new concepts and instruments referred to non-linearity. Therefore, it is very interesting to be able to define the dynamic execution of this coating, so that in a study of non-linear dynamics the corrosion process is borne out through recursive graphics.

#### *3.2. Recurrence Plots Analysis*

Given all the evidence obtained about the chaotic or non-linear behavior, in the series the complete dynamics of the current mechanism cannot be captured through statistical analysis, so the implementation of visual recurrence analysis in complex performances is proposed, so that predictions are allowed for brief periods of time [39]. This method is founded on the immersion theorem of Takens, which establishes that, under certain conditions, it will be possible to receive an estimation that is topologically equivalent, and that it will, therefore, allow extracting all the relevant data about the underlying dynamic system and the unknown that generates the time series. Recurrence plots (RP) are an excellent tool to represent non-linear dynamics and find the repetition of a pattern, although the process is not periodic in the strict sense it is possible to show repetitive or "recurrent" behaviors to differentiate chaotic variability or detect changes in the state of the evolution of a system. Dynamic systems are represented in a phase space that is no more than a vector space, which can have several values. One of the ways to characterize this recurrence is to compare the difference between all the states of the path that describes the evolution of the system that requires obtaining information about the immersion dimension or the dimension of the vector space in which it is possible to represent the

dynamics of the system. After choosing a reconstruction dimension and obtaining the vectors, the RP represents a lot of stages in a square of dimension *M* × *M*, where *M*, the axes or the sides of the second power, represent the chronological succession of the vectors in the remodeled space. This analysis was based on the methodology proposed by Eckmann et al. [40]. It consists of specifying when the levels in the reconstructed phase space are infinitesimally close because they have applied a very small dimension. The progressive increase of the dimension until the false infinitesimals disappear provides a criterion of the necessary dimension for the reconstruction.

Subsequent developments have allowed quantifying the amount of recurrence present in the graph. Zbilut and Webber [41] propose the RQA (recurrence quantification analysis) as defined by the following characteristic indices:

The percentage of recurrence (%Rec) consists of the percentage of points that are in the value threshold, or the lowest value in the total points in the recursive graph. This parameter is associated with the periodicity of the signal. Described in Equation (3) as:

$$\% \text{Rec} = \frac{\text{NREURS} / M(M-1)}{2} \cdot 100 \tag{3}$$

NRECURS is the entire number of recurring periods in the superior triangle of the graph without counting the peaks of the bisector. The denominator of Equation (3) is the number of points in the upper triangle of the graph eliminated those of the bisector.

The percentage of determinism (%Det) is the percentage of stops that constitute a line parallel to the main diagonal. This parameter is a measurement of how much the events in the past affect those in the future. Described in Equation (4) as:

$$\% \text{Det} = \frac{\text{DRECTORS}}{\text{NRECTKS}} \cdot 100 \tag{4}$$

DRECURS is the number of points that are part of line segments parallel to the bisector of the square. A line segment is specified as two or more adjacent points.

The maximum line (LM) is the duration of the most recurrent segment and is reciprocally proportional to the maximum Lyapunov exponent, which, definitively, tells us about the sensibility of our system to the initial conditions. Finally, the entropy of information or Shannon Entropy corresponds to the probabilistic distribution of the diagonal lines of determinism and is related to the complexity of the determinant structure in the system [42]. Described in Equation (5) as:

$$\text{Entropy} = -\sum\_{i=0}^{n} p\_i \log p\_i \tag{5}$$

where *pi* is the relative frequency of the length of the recurring segments. If the series is random, this measure is 0.

The analysis mutual information (AMI) function was used to establish the delay time, and the false nearest neighbors (FNN) method was used for the immersion dimension to obtain the reconstruction parameters through the method proposed by Dämming and Mitschke [43]. The analysis of recurrence values was performed by the method proposed by Garcia-Ochoa et al. [44].

The RP obtained for the 24-h exposure time series are represented below in all the conditions assessed for the current time series, since they are directly linked to the localization index of the statistical analysis. As mentioned, the most important changes in signal amplitude are shown.

Figure 7 shows the graphs of recurrence for all the conditions assessed. Recurrence plots represent the dynamics in the time series in a two-dimensional space whose axes represent the number of data in the series (2048 data). The presence of regularities obtained in the dynamics of the time series appears in the graph. Time-series are sets of swings that have some periodicity and therefore, RP presented lines and correlation structures, which is suggestive of a stochastic behavior [45,46].

**Figure 7.** Recurrence plots of (**a**) Blank at 40% RH, (**b**) blank at 98% RH, (**c**) Paraloid B72 at 40% RH, (**d**) Paraloid B72 at 98% RH, (**e**) Paraloid B72 + <sup>μ</sup>g·cm−<sup>2</sup> of caffeine at 40% RH, (**f**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> of caffeine 98% at RH, (**g**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> of nicotine at 40% RH, and (**h**) Paraloid B72 + 50 <sup>μ</sup>g·cm−<sup>2</sup> of nicotine at 98% RH.

In the case of chaotic dynamics, short lines parallel to the main diagonal appear, while in the case of a random dynamics, the plot shows a uniform representation of points indicating that there is some structure in the data. In 40% relative humidity, the signals observed present more organized graphs, which shows a weakly consolidated portion of parallel structures, indicating periodicity in comparison to signals in 98% relative humidity. Furthermore, it was noted how the dynamics of the system contracts in some spots and then expands: this is a feature of systems with chaotic dynamics.

When the metal is coated with Paraloid, the formation of equidistant geometric structures was observed, which is presumed to be a more deterministic behavior. However, in this immersion condition, the recurrence plot shows a portion of reinforced parallel structures. The presence of caffeine and nicotine in two humidity conditions assessed features more structured graphics in comparison to the Paraloid, which will be reflected on signals with a higher degree of recurrence and will manifest themselves in a greater number of yellow and reddish spots. To this point, the analysis of graphs has been qualitative, and presumptive behavior may be viewed. To perform a more detailed analysis, we calculated the percentage of recurrence, the percentage of determinism, entropy and maximum line depending on weather for all series in the conditions evaluated at 40% and 98% relative humidity, set by the Euclidean method. The results are shown below in Figure 8.

**Figure 8.** Recurrence time as a function of time in 40% and 98% relative humidity.

The percentage of recurrence (%Rec) presents greater sensitivity, showing drastic changes as a mapping of relative humidity for uncoated iron, showing a lower value for bare metal at 98% relative humidity. This confirmation that iron modifies its periodic behavior when in contact with water, so the mechanisms that have been identified in the literature are corroborated through the recurrence of the signal. The Paraloid also presents modifications with respect to humidity, since in a relative humidity of 98% the system begins to be more random: this means that the system variables change in function of time, and then that the Paraloid, when in the presence of water, loses its protective capacities and limits its use because it will fail randomly.

The presence of nicotine in both humidity conditions reaches the highest recurrence values revealing a more periodic signal, this being a first evidence of the effect on system dynamics when adding the compound. The quantity of interaction between the different sites where the corrosion takes place, which corresponds to the percentage of determinism (%Det), is shown below. This parameter is remarkably sensitive to the surface conditions and the interaction of the microcells. First, %Det is minor for bare metal. This implies great randomness in the interactions of the electrochemical cells, since, for the case of a 40% relative humidity, an average value of 40% Det is initiated, while at a humidity of 98%, it is 40% Det during the first 10 h to values of 60% Det after 12 h, which describes how the interactions are modified when the surface presents corrosion products. This converts the response into a more synchronized system.

When the material is covered by the Paraloid, %Det is increased substantially to values of 80% at 40% relative humidity, and 50% at 98% relative humidity, but it changes with the same tendency as the percentage of recurrence, which decreases in function of time, showing again how the Paraloid loses protection in function of time, which decreases the synchronization and increases the interaction areas, allowing the surface to present arbitrary responses. When adding caffeine or nicotine, the remarkable growth of the %Det degree is quite noticeable in both humidity, this being more remarkable for the relative humidity of 98%, which, as cited above, implies the shaping of a water film on the airfoil. It can be seen that the chemical construction of both substances intervenes remarkably in the kinetics of the electrochemical microcells, being nicotine both the one that reported a greater degree of determinism and the one that reported a lower level of corrosion and localization.

It can be noted that the value of the entropy of information increases remarkably with the bearing of both caffeine and nicotine, the latter giving the highest value reported for both relative humidity. This is indicative of the increment in the complexity of the corrosion process dynamics that takes place, showing that by increasing the level of determinism, the level of complexity increases too, hence yielding a higher degree of security for both uniform and localized corrosion. Then it could be stated that there is a procedure of self-formation that results in greater protection.

Finally, the maximum line, which is an index of the level of sensibility to the initial conditions, once again demonstrates that nicotine and caffeine cause the participating sites to interact so that the system shows greater synchronization and, as a consequence, the most positive coefficient of Lyapunov is smaller.

#### **4. Discussion**

Significant changes are noted in the percentage of determinism since this is modified regarding the protection applied and the percentage of humidity assessed. The metal in different humidities has a less synchronized behavior. The best performance in anticorrosive protection is in Paraloid with nicotine, followed by Paraloid with caffeine. A relation based on the signals obtained and, on the parameters, analyzed shows that %Det indicates protection and continuity of the inhibition of the coatings over time. The dynamics of the systems have been confirmed as chaotic for both the uncoated metal and Paraloid B72 in both relative humidities since it was found that the entropy of both systems decreases, which shows that the systems have a sensitive dependence to the initial conditions.

Overall, it could be said that, based on the analysis of recursive graphs, the presence of caffeine and nicotine decreases the possibility of Paraloid to capture water and, thus, the metallic material will have fewer interactions. The inhibitors are taken up locally at these sites, which is in complete accord with the statistical analysis of the electrochemical noise signal. In this mode, the behavior of the coating is modified by adding both caffeine and nicotine, modifying and increasing the resistance when conferring protection, because when these compounds are added, the kinetics of the corrosion process are modified. This allows for the introduction of a system that is self-organized as a strategy to use the Paraloid, as the temporary security for its useful life is guaranteed in conditions where the relative humidity cannot be manipulated.

#### **5. Conclusions**

Electrochemical noise is capable to assess with great sensitivity the increase in corrosion resistance performance conferred by the inhibitors added to Paraloid; even the electrolytic resistance changes corrosion mechanism.

The addition of nicotine and caffeine to temporary protection is contemplated in a higher protection efficacy regardless of the relative humidity assessed. Nicotine presented a high degree of protection, superior to the caffeine protection added to Paraloid coatings.

The nonlinear analysis of the electrochemical noise signal by recurrence plots shows the variances in the surface corrosion dynamics, where a more synchronized process resulting in greater protection.

**Author Contributions:** This paper was written by all authors. P.R.B. and F.-J.R.G. conceived and designed the experiments; P.R.B. conducted a research and investigation process, specifically performed the experiments; P.R.B., F.-J.R.G. and E.G.O. discussed the results and revised the paper; P.R.B. wrote the paper.

**Funding:** Thanks to CONACYT for the financial support for the development of this research through the basic science project 239938.

**Acknowledgments:** Thanks to CONACYT for the scholarship granted to Paola Roncagliolo Barrera with (Scholarship CVU number: 332740) to develop her PhD research. The authors wish to thank to Carlos Rodríguez Rivera for the experimental assistance provided as technical supervisor throughout this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
