A self-healing coating based on facile pH-responsive nanocontainers for corrosion protection of magnesium alloy

Yuejun Ouyng ,Lin-Xin Li ,Zhi-Hui Xie ,Lili Tng ,Fuhui Wng ,Chun-Jin Zhong

a Hunan Engineering Laboratory for Preparation Technology of Polyvinyl Alcohol (PVA) Fiber Material,Huaihua University,Huaihua 418000,PR China

bChemical Synthesis and Pollution Control Key Laboratory of Sichuan Province/Institute of Synthesis and Application of Functional Materials,College of Chemistry and Chemical Engineering,China West Normal University,Nanchong 637002,PR China

c Shenyang National Laboratory for Materials Science,Northeastern University,3-11 Wenhua Road,Shenyang 110819,PR China

d Department of Chemistry,State University of New York at Binghamton,Binghamton,NY 13902,USA

Abstract The preparation of pH-responsive nanocontainers by typical silane modificatio of the mesoporous silica nanoparticle (MSN) surface is usually high-cost,complex,and time-consuming,which remains a great challenge for effective corrosion protection of magnesium alloy.Here,a new strategy to construct pH-responsive nanocontainers (MSN-MBT@LDH) is demonstrated.The nanocontainers consist of corrosion inhibitor (2-mercaptobenzothiazole,MBT) loaded MSN core and layered double hydroxide (LDH) nanosheet shell serving as gatekeepers.The successful loading of MBT and encapsulation by LDH nanosheets were confirme by a series of characterization such as scanning transmission electron microscopy coupled with energy dispersive X-ray spectroscopy (STEM-EDS) and N2 adsorption/desorption isotherms.The pH-responsive feature of the nanocontainers was demonstrated by determination of the MBT concentration in buffer solutions with different pH values.A smart corrosion protection system on Mg alloy is obtained by incorporating the synthesized nanocontainers into a self-assembled nanophase particle (SNAP) coating.The electrochemical tests and visual observations show that the hybrid coating has the best barrier properties and robustness in corrosion protection in NaCl corrosive solutions in comparison with the control coatings.The present method simplifie the synthesis processes of nanocontainers and eliminates the potential detrimental effect of excess gatekeepers on the coating.The finding provide new insights into the preparation of scalable nanocontainers.The self-healing coatings are expected to have widespread applications for corrosion protection of Mg alloy and other metals.

Keywords: Mg alloy;Coating;Corrosion;Layered double hydroxide;Nanocontainer.

1.Introduction

Coating is one of the most effective methods in preventing high chemically-reactive magnesium (Mg) alloy from corrosion[1,2].Pure coatings on Mg alloy with traditional methods such as electroless nickel[3,4],chemical conversion[5,6],and plasma electrolytic oxidation (PEO) [2,7-9],cannot provide continuous corrosion protection once the corrosive mediums penetrate the coating and reach the alloy surface.Two or even more technologies abovementioned are usually employed to form a composite coating for improve the corrosion protection[10,11].Corrosion inhibitors are added directly into a coating system to improve the corrosion resistance and prolong the lifetime of a coating [12-14],but it shows a drawback of premature leakage of inhibitor before onset of corrosion and thus decreases the effica y of corrosion inhibition.

Smart coating was proposed to address the uncontrollable release of inhibitor at the beginning of this century and has been attracted much attention in the corrosion community in the last decade [11,15-18].Smart coating for corrosion protection typically consists of organic/PEO coating and dispersed nanocontainers for storing inhibitor [11,19].To date,several types of containers including mesoporous silica[20],halloysite nanotubes [11,15],zeolites nanoparticles [16],graphene [21],lignin [17],and microcapsules [22],have been developed.Among these containers,microcapsules with diameters of dozens of micrometers are too big as the thicknesses of most coatings for corrosion protection are smaller than 100 μm [23,24].By comparison,mesoporous silica nanoparticle (MSN) is the best choice as a scaffold of inhibitor due to its very large specifi surface area (～1000 m2g-1) [20].To obtain good self-healing effect,nanocontainers for smart coating also require responsive sensitive to various external and internal stimuli,such as the pH changes at the defects,impacts of light and temperature in the environment surrounding the coating [25],rather than a slow release of inhibitors even before corrosion.The abundant silanol groups (Si-OH)on the pore surface of the mesoporous silica render it possible to control the release of inhibitor by various organic silane modification such as the installation of pseudorotaxane nanovalves [26].

The main shortcomings of silane modificatio of the pore surface are high-cost of reagents and time-consuming synthesis processes during the preparation of supramolecular,which is very unfavorable for the practical application of an intelligent coating.Furthermore,literature at present mainly focuses on the construction of stimuli-responsive nanoparticles,while the potential effects of the nanovalves on eco-environment and coating characteristic are neglected.It would be helpful for improving the properties of the coating if the nanovalves can provide both controlled release effect and additional protective functionality,which is rarely reported to date.

Layered double hydroxide (LDH) is also promising nanostructured inclusion in a smart coating and has attracted significan research interest in recent years [27-29].Application of LDHs in corrosion and protection area can be classifie into two main categories,i.e.,nanocontainer and coating,both of these take advantage of the ion exchangeability of LDHs to control the release of inhibitor and trap aggressive ions,such as Cl-[30,31].The former application consists of two steps including synthesis of inhibitor-loaded LDHs followed by incorporation of composite LDHs into an organic matrix coating [30].The latter one is direct formation of a protective LDHs fil on the bare or pretreated metal substrate surface,which is now mainly limited to several metals,such as Mg and Al,and types of LDHs,such as MgAl-LDHs and NiAl-LDHs [29,32-34].One defect of LDHs is the instability in an acidic solution,and the structure can be destroyed owing to the dissolution of hydroxide layers and carbonate ions[35,36],which is usually ignored in the previously reported study in the areas of corrosion and protection.

Here,we plan to utilize the defect of the instability of LDHs to form MSN core@LDH shell nanohybrids(MSN-MBT@LDH) for construction of pH-dependent nanocontainers,which is shown schematically in Fig.1.The functionalization of nanocontainers is realized through a low cost and facile protocol,including MBT-loading and capping with LDH nanosheets using simple adsorption steps.As far as we know,taking advantage of adsorbed LDH nanosheets as gatekeepers on the MSNs surface to prepare inhibitorloaded nanocontainers and a smart coating for corrosion protection of Mg alloy have not been reported so far.The potential detrimental effects originating from nanovales to environment and coating are avoided effectively.Instead,the excess un-exfoliated LDH nanoplates were embedded in the coating along with the MSN-MBT@LDH nanocontainers and served as the second type of nanocontainers,which is beneficia to the coating resistance against corrosion during real application.

2.Experimental section

2.1.Materials and chemicals

Cetyltrimethylammonium bromide (CTAB,≥ 98%),tetraethylorthosilicate (TEOS,≥98%),ethyl acetate (EA,≥95%),3-glycidyloxypropylmethyl diethoxysilane (GPTMS,≥98%),triethylenetetramine (TETA,≥98%) were purchased from Sigma-Aldrich.The reagents used to synthesize MgAl-LDH were mainly acquired from Aladdin Industrial Inc.The deionized water (18.2 MΩcm at 298K) was obtained by a water purificatio system (UPT-II-10T).Mg alloy AZ31 was used as the substrate for coating.

2.2.Synthesis of MSN-MBT@LDH nanocontainers

The MgAl-LDHs sample was prepared by a coprecipitation method.Firstly,Mg(NO3)26H2O (3.82g) and Al(NO3)39H2O(2.70g)were dissolved into 240mL of boiled and cooled deionized water to yield a mixture.The mixture was transferred into a 1000mL round bottom three-necked flask followed by addition of about 240mL of sodium hydroxide solution (0.175mol/L) until the pH of the solution is close to 10.After that,the solution was reflu ed at 100°C for 16h under a N2atmosphere and was cooled down in the air to room temperature.Finally,MgAl-LDH powder was obtained by centrifugation,washing with distilled water,and drying of the precipitate overnight (～12h) under vacuum condition at 65°C.

The synthesis procedure of MSNs can be found in our recent publication [20].Briefl,3.5mL of NaOH solution(2mol/L) was added into a 480mL of CTAB (1.0g) aqueous solution with continuous stirring at 70°C for 30 min,followed by addition of 5.5mL of TEOS.One minute later,5mL of EA was added with vigorous stirring until a large amount of white precipitate emerged.The precipitate was fil tered and washed with ethanol and distilled water,and dried under vacuum at 50°C for 10h to obtain crude product.Afterwards,the crude product was reflu ed in a mixed solution of ethanol:hydrochloric acid=10:1 (v/v,170mL) to remove the CTAB template.Lastly,blank white MSN powder was obtained after filtration washing,and drying.

Fig.1.Schematic diagram of synthesis of a smart MSN-MBT@LDH nanocontainer with a MSN as core and acidic-responsive LDHs (only 3 layers of the nanosheets are drawn) as shell.

The MSN-MBT@LDH nanocontainer was prepared as follows:Firstly,1.0g of blank MSNs was dispersed ultrasonically in a 50mL ethanol solution (70%) containing 1.0g of MBT for 20min.The suspension was allowed to stand for 4h at room temperature after oscillating absorption for 12h under low vacuum condition by a platform shaker.The shaking step was repeated twice for loading more MBT and adsorption equilibrium.Secondly,dispersion solution of MgAl-LDH nanosheets,which has been exfoliated in advance by ultrasonically dispersing of MgAl-LDH powder (100mg) in a formamide solution (5mL,0.05%) for 1h,was added into the mixture above,followed by 4h of shaking.After centrifugation and drying at 65°C for 12h,MSN-MBT@LDH nanocontainer was obtained.

2.3.Preparation of MSN-MBT@LDH-Incorporated coating on Mg alloy

Acetic acid solution (4.5mL,0.05mol/L) was added dropwise into a mixture of TEOS (4.54mL) and GPTMS(13.48mL),followed by 3h of magnetically stirring and 4 days of aging in a sealed container to produce a self-assembled nanophase particles (SNAP) sol.MSNMBT@LDH nanocontainers (45mg) were added into the SNAP sol and dispersed ultrasonically for 10s,followed by addition of 0.55mL of TETA as a crosslinking agent.Then,the mixture was stirred continuously at room temperature for 10 min and coated on the surface of Mg alloy by a dipcoating technique using an automated dip-coater (SYDC-100,Shanghai SAN-YAN Technology Co.,Ltd.).The pretreated Mg alloy substrate was soaked in the mixture above for 180s,and then pulled up at a rising rate of 350μm s-1.The coating process was performed twice,and the coated sample was solidifie in air and dried at 100°C for 1h and then marked as SNAP-MM@L coating.For comparison,control experiments were carried out and some abbreviations are explained as follows:blank SNAP matrix without nanoparticles,SNAP matrix with blank MSNs,and SNAP matrix with MSN@LDH(MSNs capped with LDH nanosheets but without MBT loading)nanoparticles were noted by SNAP,SNAP-M,and SNAPM@L coatings,respectively.

2.4.Characterization and electrochemical tests

Scanning electron microscopy (SEM,SU8020,3kV),scanning transmission electron microscopy coupled with energy dispersive X-ray spectroscopy (STEM-EDS),and transmission electron microscopy (TEM,Tecnai G2 F20,2000kV) were used to observe and analyze the surface morphologies and compositions of the MSNs,MSN-MBT@LDH nanocontainers and thicknesses of different SNAP coatings.The characteristic diffraction peaks of the nanoparticles were obtained by X-ray diffractometer (XRD,D8 Advance,Cu target,incident wavelengthλ=0.154nm).A thermogravimetric analyzer (TGA/SDTA851e) was used to analyze the weight loss variations of different nanoparticles at elevated temperatures.The temperature of TG test ranges from 25 to 600°C at a heating rate of 10.0 °C min-1under N2atmosphere.The specifi surface area,average pore diameter,and pore volume of MSNs and MSN-MBT@LDH nanocontainers were measured using N2adsorption/desorption isotherms at 77K (Mike ASAP 2460,America).The Brunauer-Emmett-Teller (BET)method was used to calculate the surface area from isotherms in the low-pressure range (0.05≤P/Po≤0.2).The average pore size and pore volume were calculated based on Barret-Joyner-Halenda(BJH)method[37].Shimadzu UV-1800 UVvis spectrophotometer was used to determine the MBT concentration released from MSN-MBT@LDH nanocontainers.

The corrosion resistance of the SNAP coating was evaluated in a NaCl solution (3.5wt.%) at 25°C based on the electrochemical impedance spectroscopy (EIS) [38] and potentiodynamic polarization curve [39] acquired by a Gamry electrochemical workstation (Interface,1010E).The working electrode is an Mg alloy with a coating (1 cm2),the auxiliary electrode is a platinum electrode,and the reference electrode is a saturated calomel electrode (SCE).The electrochemical impedance tests were performed at a frequency range of 105-10-2Hz with an AC voltage amplitude of 30mV.Relative inhibition efficiencie (η) of different SNAP coatings with different exposure time were calculated by charge transfer resistance (Rct) values derived from EIS results using the following equation:

whereRct,tandRct,0are theRctvalues of the different SNAP coatings with different exposure time and the bare SNAP coating after 10 days of immersion in NaCl solution,respectively.Noting that bare SNAP coating rather than Mg alloy substrate is used as the control sample to evaluate the self-healing effi ciency of the coating after incorporation of stimuli-responsive nanocontainers.The different SNAP coatings were potentiodynamically polarized up to different potentials from -500 to +500mV vs.open circuit potential (OCP) with a sweep rate of 5mV s-1.The three-electrode system was soaked in a NaCl solution for about 30min to obtain a stable OCP and placed in a Faraday shielding box to eliminate external electromagnetic interference before acquisition of EIS and potentiodynamic polarization curves.All the electrochemical tests were carried out three times at least at each situation to ensure the repeatability of the experiment.The electrochemical data of both EIS and potentiodynamic polarization curves were analyzed by Gamry Echem Analyst software.

Fig.2.(a,b) SEM and (c,d) TEM images of MgAl-LDH nanoplates at different magnifications (a),the inset in (a),and (b) are 10k,50k,and 100k,respectively.(d) Enlarged morphology of the circled area in (c).

3.Results and discussion

3.1.Characterization of LDHs,MSNs,and MSN-MBT@LDH nanocontainers

The surface morphology of the MgAl-LDH white powder was characterized by SEM and TEM (Fig.2).Fig.2 shows clearly the typical nanoplate structure of the MgAl-LDH,and much longer dimension inab-face (length and width) compared to that in thickness (the size alongc-axis) originating from the preferred orientation inab-face of the MgAl-LDH crystallites.The sizes of the MgAl-LDH are near 20nm in thickness based on the SEM image (Fig.2b) and～75nm inab-face based on the TEM image (Fig.2d),respectively.Fig.2d also shows that most of the MgAl-LDH crystallites have randomly shape inab-plane,through several typical hexagonal MgAl-LDH nanoplates can be identified exhibiting the close growth rate in different directions alongab-plane for most MgAl-LDH crystallites prepared by the present coprecipitation method.

To observe the variation of surface morphology of the MSNs before and after modification SEM and TEM images of MSNs and MSN-MBT@LDH nanocontainers are given in Fig.3.As shown in Fig.3a,the blank MSNs show a uniform spherical shape with an average diameter of 80±5nm.The corresponding TEM image (inset) demonstrates the well-ordered hexagonal mesoporous structure with an average pore size of 2.3±0.1nm,which is supported later by BJH calculation result based on the N2adsorption/desorption test.This result was consistent with data in previous reports [20,40].The MSN-MBT@LDH nanocontainers keep uniform spherical shape but are coarser and have larger diameters (88±5nm) after MBT-loading and LDH nanosheets encapsulation (Fig.3b),indicating the formation of nanoshells with thicknesses of～4nm on the surface of MSNs by electrostatic interactions to block the nanopores.The layer number of the nanoshells is about eight,assuming a thickness of～0.5nm for a monolayer LDH nanosheet[41].Also,from the TEM image in Fig.3b,it can be seen that the original mesopores of the MSN blur,suggesting that the channels in the MSN are blocked by MBT inhibitor and the outer surface of the MSNS is wrapped well by the LDH nanosheets after the interaction.The white circle areas show that a small amount of LDHs is not exfoliated and remains in the form of nanoplates.STEM-EDS maps of different elements for a single MSN-MBT@LDH nanocontainer (Fig.3c)clearly shows that exception of Si element,Mg,Al,N,and S elements exist and distribute uniformly on the surface of the nanocontainer.The existence of N and S elements originates from thiazole ring and thiol of MBT,and Mg and Al elements stems from MgAl-LDH nanosheets,confirmin the successful loading of inhibitor in the mesopores and encapsulation of MBT-loaded MSNs with LDH nanosheets.

Fig.3.(a,b) SEM images of MSNs and MSN-MBT@LDH nanocontainers at the same magnificatio (100k).The insets are corresponding TEM images of a single MSN and MSN-MBT@LDH particle.(c) STEM images of a single MSN-MBT@LDH nanoparticle,and the overlapped and individual elemental maps of different elements including Si,C,Mg,Al,N,and S at the same position.

Fig.4.Nitrogen adsorption/desorption isotherms and corresponding pore size distributions (insets) of (a) MSNs and (b) MSN-MBT@LDH nanocontainers.

To quantitative measure the changes of specifi surface area and pore size of MSNs and MSN-MBT@LDH nanocontainers before and after encapsulation,N2adsorption/desorption isotherms were obtained (Fig.4)and the corresponding calculated results based on BET and BJH methods are listed in Table S1.Fig.4a shows a typical isotherm of type IV with a small adsorption step at the intermediate P/P0region,i.e.,0.3 -0.4.,originating from N2condensation in the mesoporous channels by capillarity,further corroborates the mesoporous structure of the MSN powders.The small hysteresis loop and narrow pore size distribution as given in inset of Fig.4a indicate the uniform cylindrical mesopores.The pore volume and pore diameter of the pure MSNs based on BJH method are 1.9 cm3g-1and 2.4nm,respectively (Table S1),which is consistent with the TEM results (Fig.3a).After being adsorbed with MBT and sealed with MgAl-LDH nanosheets,the specifi surface area decreases significantl from 925 to 64 m2g-1,and the pore volume decreases evidently to 0.14 cm3g-1,manifesting the successful impregnation of MBT and wrapping by MgAl-LDH nanosheets.As shown in inset of Fig.4b,several different peaks are observed in the pore size distribution curve of MSN-MBT@LDH nanocontainers,implying the formation of new types of nanopores during loading and wrapping processes.Considering the remarkable decrease in specifi surface area and pore volume after modification these nonuniform nanopores in the MSN-MBT@LDH nanocontainers are unlikely related to the mesopores in the blank MSNs but results from the gaps among the scattered LDH nanoplates and the nanocontainers,which is evidenced by the sudden increase in N2absorption at high pressure (P/Po=0.9-1.0) range [20].

Fig.5.(a) Wide-angle XRD patterns of MSNs,MgAl-LDH,and MSN-MBT@LDH powders.(b) TG results of MSN,MSN@LDH,and MSN-MBT@LDH nanoparticles.

The successful formation of nanoscale LDH shells and loading of MBT on the MSN surface are corroborated by the XRD patterns and TG curves (Fig.5).Fig.5a clearly shows several strong and sharp characteristic peaks located at 11.33°,22.80°,and 34.74° for pure LDH powder,which are assigned to the typical (003),(006),and (012) reflection of an LDH phase,respectively,evidencing the layered structure of LDH nanoplates (JCPDS NO:35-0965) [42,43].Compared with the blank MSNs that exhibit only a broad peak at 21.98°deriving from the amorphous structure of silica (JCPDS NO:29-0085),a new peak at the same peak position of(003)plane of a LDH crystal is clearly identifie in the XRD pattern of MSN-MBT@LDH nanocontainer,demonstrating further the existence of LDH layer on the MSNs surface.The peaks that correspond to(006)and(012)planes and especially other crystal planes of the LDH at higher 2θvalues are hard to be identifie in the XRD pattern of MSN-MBT@LDH nanocontainer,which is ascribed to the loss of the original crystal plane structure during ultrasonic exfoliation and sealing processes,extreme low content of LDH nanosheets in comparison with MSNs,and the broad peak at 21.98° for MSNs.TG curves show that weight losses of all nanoparticles are only less than 2.2% when the temperature is lower than 100 °C (Fig.5b),which is mainly attributed to the evaporation of physically adsorbed water molecules on the surface of the nanoparticles.The TG curves of MSN,MSN@LDH,and MSN-MBT@LDH display slight difference until the temperature is elevated to about 200 °C.The fina weight loss of blank MSNs is 8.23% when the temperature increases to 600 °C,which is due to the residual organic substances such as CTAB template and EA during preparation.Compared with MSNs,the higher value of fina weight loss of MSN@LDH sample (14.73%) is owing to the loss of water absorbed on the surface,structural interlayer water,and evolution of H2O vapor and CO2gas by dehydroxylation and decomposition of carbonate (CO32-) in the LDHs,respectively [44].As for MSN-MBT@LDH nanocontainer,the weight loss starts to decrease sharply when the temperature is higher than～200 °C,and shows further decrease until fina value reaching 19.6%at 600 °C owing to decomposition of MBT inhibitor.

3.2.Controlled release of MBT from MSN-MBT@LDH nanocontainers

Fig.S1 depicts the ultraviolet visible spectra to investigate the MBT release profil of MSN-MBT@LDH nanocontainers at buffer solutions with different acidities.A strong and maximum absorption wavelength of MBT (λmax) appears at 308nm (Fig.S1a).The absorbance value atλmaxagainst MBT concentration in the range of 1.25 to 7.5μg mL-1is plotted as a standard curve which shows aR-squared value(R2=0.99971,≥0.999) extremely close to 1 and good linear relationship (Figure S1b),indicating the validity of the test for determination of MBT concentration.

Based on the ultraviolet visible spectra,the dependence of the MBT released from MSN-MBT@LDH nanocontainers with release time (t) is shown in Fig.6.Fig.6a,which exhibits obvious difference at various pH values,though all the release curves except control curve (pH=7,washed) can be divided into two stages at least.One is the fast release stage within 7.5h and the other is the sustained release stage in the remained investigated time.After 30h of releasing,for the samples without washing at the last step during preparation of MSN-MBT@LDH nanocontainers,the MBT concentration is 5.40μg mL-1when pH=2,while the value is 3.42μg mL-1when pH=7.MBT release profil for MSNMBT@LDH nanocontainers that were washed thoroughly by deionized water at the last step during synthesis is illustrated in Fig.6a (curve marked as washed).The variation of MBT concentration vs.time indicates that only a small amount of MBT is released in a neutral environment.To further analyze the release profil at different pH values,release percentage of MBT is plotted as a function of time (Fig.6b).The MBT release percentage for the buffer solution with a pH value of 2.0 is about 92% at the end of the fast release stage (t=7.5h).At the same moment,the release percentages are about 40% and 27% when the solution with pHs of 5.0 and 7.0,respectively,manifesting a clear difference in release percentage of the MSN-MBT@LDH nanocontainers under different acidities.

Fig.6.(a) MBT release profil of MSN-MBT@LDH nanocontainers as a function of time at different pHs calculated by UV absorption data.(b) Variations of release percentage after deduction of the difference between washed and not washed samples at pH 7.0,assuming that the MBT concentration at t=30h when pH=2.0 is define as the total quantity of MBT that can be released.

The pH-dependent MBT release can be explained in terms of the interactions among MSNs,MBT molecules and LDH nanosheets.The MBT molecules are usually loaded in the nanopores and interiors of the MSNs by weak non-covalent interactions,such as physical adsorption,electrostatic interaction,hydrogen bonding,andπ-πstacking [45].Under neutral environments,the MSNs surface carry negative charge because of abundant Si-OH,while the exfoliated LDH nanosheets have positive surface charge [46].The LDH nanosheets will be adsorbed on the MSN surface to seal the MBT and mesopores by electrostatic interactions when LDH nanosheets were added into the MBT-loaded MSNs sol.Meanwhile,the free MBT molecules without adsorption in the mesopores will interact with excess LDH nanoplates that were un-exfoliated and were not adsorbed on the MSNs surface.The interaction between MBT molecules and LDH nanoplates involves two forms including physical surface adsorption and confine intercalation by ion exchange[31,47].The physically adsorbed MBT molecules are not confine and can diffuse freely when existence of concentration gradient even without acidic stimuli.In other words,the obtained MBT concentration during release experiment for the MSN-MBT@LDH nanocontainers without washing contains a certain quantity of inhibitor that was not released from the mesopores of the MSNs.However,the physically adsorbed MBT on the LDH nanoplates probably be washed out by deionized water,showing a much lower MBT release rate for the washed MSNMBT@LDH nanocontainers compared to the samples without washing at the same pH value (pH=7.0,Fig.6a).The intercalated MBT in the LDH nanoplates can be released by ion exchange because the phosphate anions existing in the buffer solution have a higher affinit for MgAl-LDH intercalation compared to MBT [47-49],which accounts for a certain percentage of MBT release even under a neutral environment after exclusion of physically adsorbed MBT (pH=7.0,Fig.6b).In a weakly acidic environment (pH=5.0),the LDH nanosheets on the MSNs surface are dissolved slowly,and both MSN and MBT carry positive surface charge,causing a larger electrostatic repulsion force between MSN and MBT,thus faster release of MBT from the nanopores of the MSN is observed [45,46].When the pH decreases to 2.0,both the LDH nanoshells on the MSNs surface and the excess unexfoliated LDH nanoplates with stacked structure can be dissolved rapidly [35,36],resulting in double release percentage in comparison with that obtained at pH 5.0.The pHdependent release rate of MBT demonstrates the validation of using LDH nanosheet as nanovalves by virtue of the different dissolve rate under different acidic environments.

It is worth to note that zero release is usually required at physiological pH (7.4) in a drug delivery system for protecting the healthy tissues from the toxic drugs and the decomposition/denaturing of the drugs before reaching the targeted cells [50],while the release of the unconfine MBT molecules in a corrosion resistant coating system is desirable in comparison with the case that inhibitor is only stored in the nanopores of the nanoreservoirs,because even direct incorporation of inhibitors without scaffolds in an organic coating is also favorable for improving the corrosion resistance [51].That is to say,washing should be avoided intentionally at the fina step when preparation of MSN-MBT@LDH nanocontainers to achieve high effectiveness of corrosion protection in a coating system.Therefore,MSN-MBT@LDH nanocontainers without washing were dispersed in the SNAP coating to evaluate the corrosion resistance capability in the following study.

3.3.Corrosion resistance comparison for different SNAP coatings

The thickness of different coatings was observed by SEM before electrochemical test as the important role in affecting corrosion resistance(Fig.7)[3].Fig.7 displays that the crosssectional morphologies of the four different coatings are very similar,though the thicknesses of SNAP-M,SNAP-M@L,and SNAP-MM@L coatings decrease slightly after introduction of silica nanoparticles.The thicknesses are 237,220,226,and 222μm for SNAP,SNAP-M,SNAP-M@L,and SNAPMM@L coatings,respectively,indicating that incorporation of silica nanoparticles is detrimental to the corrosion resistance from the aspect of coating thickness.The difference in thicknesses of the three coatings with different silica nanoparticles is very small and can be regarded as approximately equal.

Fig.7.SEM images for the cross-sectional morphologies and thicknesses of (a) SNAP,(b) SNAP-M,(c) SNAP-M@L,and (d) SNAP-MM@L coatings.

Fig.8.Potentiodynamic polarization curves of Mg alloy,SNAP,SNAPM,SNAP-M@L,and SNAP-MM@L coatings in 3.5wt.% NaCl medium at 298K.

Fig.8 shows potentiodynamic polarization curves of different coatings tested in NaCl solution at 25°C.Considering the abnormal large anodic Tafel slope during Tafel extrapolation,only cathodic Tafel region of the polarization curve is extrapolated back to the OCP to give corrosion current density (icorr) [52],and the fitte results of the polarization curve are listed in Table S2.For comparison,the electrochemical parameters of substrate acquired in our previous study are also given [31].Compared to Mg alloy,the corrosion potential (Ecorr) andiorrof all the samples coated with different SNAP coatings shift obviously towards more positive potential and decrease significantl by more than three orders of magnitude,respectively,suggesting the obvious improvement of corrosion resistance from the aspects of both thermodynamics and kinetics.The SNAP-MM@L coating displays the most positiveEcorrvalue (-0.39V vs.SCE) in comparison to that of the pure coating (-0.73V) and control coatings with MSN (-0.63V) and MSN@LDH (-0.65V) nanoparticles.The positive shift ofEcorrvalue for SNAP-MM@L coating stems from the MBT which is a mixed-type corrosion inhibitor but predominantly controls the anodic reaction.Theicorrvalues of pure SNAP,SNAP-M,and SNAP-M@L coatings are 1.13,0.98,and 1.02 nA cm-2,respectively.The extremely lowicorrdown to nanoscale of the SNAP-based coating is attributed to the good insulation of organic layer prepared by sol-gel technology,which is also observed in similar corrosion protection systems by other researchers [53,54].The close values oficorrfor the three different coatings may be ascribed to the opposite contributions of the coating thickness and compactness to the corrosion resistance.It is no doubt that the decrease of coating thickness,as shown in Fig.7,is detrimental to corrosion resistance,but compact crack-free structural composite coatings tend to be formed and a positive impact of corrosion protection will be observed when moderate concentration of nanoparticles are incorporated [4,55].In other words,the weakened effect on corrosion resistance owing to thinner thickness can be compensated through higher compactness after incorporation of MSN and MSN@LDH nanoparticles,and even be overwhelmed and showing a lowericorrvalue (0.74 nA cm-2) when inhibitor-loaded nanoparticles,i.e.,MSN-MBT@LDH nanocontainers,were introduced.Overall,the differences oficorrvalues based on potentiodynamic polarization curves for the different pristine coatings,especially before introduction of inhibitor,are very small.

TheEcorrandicorrvalues for different fresh SNAP coatings based on potentiodynamic polarization curves manifest the notable corrosion protection to the substrate,however,these values cannot recognize the robustness in corrosion resistance during long-term exposure.To demonstrate the superior corrosion resistance performance of the new designed SNAP-MM@L coating,EIS data of different coatings after exposure to NaCl solutions with different days were acquired(Fig.9).According to recent literature,the impedance moduli at low frequencies (|Z|?=0.01Hz),as illustrated in Fig.10a and Table S3,can be used to qualitatively evaluate the corrosion protection of a coating system [24,56].Figs.9a and 10a show that the initial |Z|?=0.01Hzof a blank SNAP coating is much higher than that of Mg alloy,and the |Z|?=0.01Hzvalues of the rest coatings increase further after introduction of silica nanoparticles.The |Z|?=0.01Hzvalue is～574Ωcm2only for substrate,increases forty thousand times to 23.8 MΩcm2for pure SNAP coating,and further rises to 55.8,80.8,and 103.3 MΩcm2for SNAP-M,SNAP-M@L and SNAP-MM@L coatings,respectively.The initial values of |Z|?=0.01Hzfor different coatings attest to the following order of corrosion resistance:SNAP＜SNAP-M ≈SNAPM@L＜SNAP-MM@L coatings.The enhancement in corrosion resistance is also confirme by the phase angles of Bode plots at low frequencies.Bare Mg alloy shows a positive phase angle (Fig.9b),indicating the existence of inductive loop resulting from relaxation processes of adsorbed intermediates such as Mg(OH)+ads[3,32,57].With increasing days of exposure to NaCl solutions,the |Z|?=0.01Hzvalues for SNAP,SNAP-M,and SNAP-M@L coatings show gradual decrease and finall reach to a very low level,approach or even lower than 106Ωcm2,when the immersion time increases up to 10 days (Fig.10a,c,e,and j,and Table S3),indicating the gradual degradation of the coating during immersion.As for SNAP-MM@L coating,the |Z|?=0.01Hzvalue increases more than two times to a very high value,259.8 MΩcm2,after one day of exposure,then drops back to the initial level,103.9 MΩcm2,after four days of exposure,and even increases slightly,rather than decrease,to 107.2 MΩcm2after ten days of exposure to NaCl solution,signifying the success release of MBT and formation of healing protection fil by corrosion inhibition at the defects [15,23,58].

Fig.9.Bode plots acquired for (a,b) pristine coatings and coatings after immersion in a NaCl solution for (c,d) one,(e,f) four,and (g,h) ten days and depicts as Bode plots exhibiting the change in (a,c,e,g) absolute impedance and (b,d,f,h) phase angle as a function of frequency for blank SNAP coating and coatings with different types of MSNs nanoparticles.

The robustness in corrosion protection of the different coatings can also be ascertained combing with the variations of Bode phase angle spectra (Fig.9b,d,f,and h,and Fig.10b).Phase differences between current and voltage,i.e.,phase angles at high frequencies (f ＞104Hz) in the EIS plots,reflec the current pathway in a parallel capacitor-resistor circuit [59].A coating with large resistance and/or capacitance is capacitive and current prefers to pass through a capacitor,resulting in high phase angles near -90°.Otherwise,a coating with small resistance and/or capacitance is resistive and current prefers to pass through a resistor,resulting in phase angles near 0° [59,60].Fig.9b displays the obvious shift of phase angles at high frequencies from near 0° for Mg alloy to near -90° for different SNAP coatings,indicating the capacitive behavior and high resistance of the SNAP coatings compared to the substrate that has only porous oxide layer with barely corrosion protection in a neutral environment.The variations of phage angles at high frequencies (Φf=10kHz) for different SNAP coatings with increasing exposure time is shown in Fig.10b.Initially,conductive pass way is not formed,thereby phase angles for all fresh coatings are higher than 82° (absolute value).Generally,with increasing exposure time,corrosive mediums gradually permeate into the coating,giving rise to decreased coating resistance and increased coating capacitance [61,62].According to the aforementioned analysis regarding the relationship betweenΦf=10kHzand resistance/capacitance,it can be seen that the decreased coating resistance and increased coating capacitance will cause opposite variations of phase angles,i.e.,decreased resistance leads to a lower phase angle (absolute value) while increased capacitance leads to a higher phase angle.However,the decrease of resistance generally takes place more dramatically than the increase of capacitance at the same time.Thus,the shift direction of the phase angle is primarily dominated by the variation of resistance,showing a gradual decrease of phase angle when the coating system is gradually degraded [59,61].As shown in Fig.10b,the phage angle drops sharply from 84.1° to 45.6° for pure SNAP coating,and the values for SNAP-M and SNAP-M@L coatings decrease obviously from 87.5° to 75.4° and 82.1°to 68.1°,respectively,after ten days of exposure,while the phage angle for MSN-MBT@LDH coating fluctuate between 84.6° to 88.1° during the long-term exposure in a corrosive environment,further suggesting the high robustness of the asprepared coating in a neutral corrosive environment.

Fig.10.Immersion time dependence of (a) impedance moduli at low frequencies (|Z|?=0.01Hz) and (b) phase angles at high frequencies (Φf=10kHz) based on the Bode plots of the SNAP,SNAP-M,SNAP-M@L,and SNAP-MM@L coatings.

Fig.11.Nyquist plots for the SNAP,SNAP-M,SNAP-M@L,and SNAP-MM@L SNAP coatings after 10 days of exposure in NaCl solution.(a),(b),and inset in (a) are presented with different axis scales for observing.

To quantitative analyze the EIS data,in combination with Nyquist (Fig.11) and Bode plots,two different equivalent circuit models,as shown in Figure S2,are used to fi the EIS spectra of bare Mg alloy and different SNAP coatings,respectively.The equivalent circuit model for bare Mg alloy consists of solution resistance (Rs),Rct,inductance (L),inductance-related resistance(RL),and constant phase element for the electrical double layer (CPEdl)to replace ideal capacitance owing to deviation of phase angles from -90° (Fig.S2a) [63].In addition toRs,Rct,andCPEdl,the equivalent circuit for coatings has three new electrical components,i.e.,coating resistance (Rcoating),coating capacitance (CPEcoating),and Warburg impedance (Zw) based on the linear region in the range of low frequency in the Nyquist plots (Fig.11),as illustrated in Figure S2b.TheZwelement is related to the diffusion-controlled mass transport of the dissolved oxygen to the metal surface or the transportation of dissolved magnesium cations from the surface of substrate to the bulk solution,which was also observed in previous research when using MBT as inhibitor to protect copper plate from corrosion [64].The existence ofZwwill be confirme later by the Nyquist diagram which shows a straight line at low frequency region.The fitte electrochemical parameters based on the equivalent circuit model for the four coatings are listed in Table S3.All the fresh coatings show very highRcoatingandRctvalues.Values ofRcoatingandRctfor SNAP-M,and SNAP-M@L coatings are higher than that of pure SNAP coating,indicating again the positive impact of MSNs on the corrosion resistance.With the introduction of MSN-MBT@LDH nanocontainers,the initial values ofRcoatingandRctrespective increase significantl from 2.60 and 18.89 MΩcm2for pure SNAP coating to 57.40 and 149.70 MΩcm2for SNAP-MM@L coating.The high values ofRcoatingandRctmanifest the good corrosion protection of pure SNAP coating and reinforcement in corrosion resistance after introduction of MSNs and MSN@LDH nanoparticles,especially MSN-MBT@LDH nanocontainers.With the increase of immersion days,all the values ofRcoatingandRctfor the four different coatings decrease obviously,but the decrements are quite different.For instance,theRcoof bare SNAP coating decreases sharply by a factor of～520 from 2.6 MΩcm2to only～5000Ωcm2after ten days of exposure.According to the previous study,a coating system with impedance below 106Ωcm2generally cannot provides good corrosion protection [61].The impedance of bare SNAP coating is lower than 31 kΩcm2after four days of exposure in the corrosive environment,implying that the coating system has deteriorated excessively and the lifetime of effective corrosion protection is shorter than four days.Compared with bare SNAP coating,both SNAP-M and SNAP-M@L coatings show similar variation tendencies with increasing immersion time but the values ofRcoatingandRctare much higher and still larger than 106Ω·cm2after four days of immersion,indicating that incorporation of silica nanoparticles is favorable for improving the coating corrosion protection.After ten days of soaking,theRctandRcoatingvalues for both SNAP-M and SNAP-M@L coatings are near or even less than 106Ωcm2,indicating the coating almost loses their effectiveness in corrosion protection.As for SNAP-MM@L coating,theRcoatingandRctonly reduce by one-half to 27.6 MΩcm2and less than two-thirds to 55.7 MΩcm2,respectively,even after ten days of immersion.BothRcoatingandRctof the SNAP-MM@L coating after ten days of immersion are much higher than that of the control coatings even before exposure.

The values ofCPEcoatingandCPEdlare directly related to the amounts of penetrated water in the coating and the substrate area exposed to the corrosive electrolyte,respectively [62].The higher water uptake of the coating and larger exposed area of metal lead to higher values ofCPEcoatingandCPEdl,respectively.As shown in Table S3,the values ofCPEcoatingandCPEdlfor all the coatings except SNAPMM@L sample increase evidently after ten days of exposure,manifesting the gradual degradation of the three coatings.By contrast,for the SNAP-MM@L coating,theCPEcovalue is essentially constant and the increment ofCPEdlvalue is very small after long-term exposure.Besides,the Nyquist diagrams as shown in Fig.11a clearly show that the inductive loop of SNAP-MM@L SNAP coating is so large that the inductive loops of the other coatings are overwhelmed totally.The impedance of SNAP-MM@L SNAP coating is much higher than that of others and maintains at a very high level ever after 10 days of exposure,illustrating the superiority and robustness in corrosion resistance of the new designed self-healing coating.

Fig.12.Digital pictures of(a)SNAP,(b)SNAP-M,(c)SNAP-M@L,and(d)SNAP-MM@L coatings before and after 10 days of exposure to 3.5% NaCl solutions at 298K.Red dotted circles indicate the visual observed evident corroded areas and corrosion pits of the coatings after exposure.

The good corrosion protection of the self-healing coating system to Mg alloy is also exhibited by theηvalues over different days of immersion (Table S3).All the coatings show a very high initialηvalues above 99.40%,exhibiting the good corrosion protection of these fresh coatings to substrate.Theηvalue for bare SNAP coating keeps relatively stable after 1 day of immersion but sharply decreases to 85.62% when the immersion time extends to 4 days,exhibiting the fast degradation of the SNAP coating without nanoparticles.After 10 days of exposure,theηvalues for both SNAP-M and SNAPM@L coatings also decrease obviously,while theηvalue for SNAP-MM@L coating decreases slightly to 99.80% from the initial value,99.92%,demonstrating the excellent corrosion protection of the self-healing coating system even after long-term exposure.

The digital pictures of the four different types of SNAP coatings are shown in Fig.12.After 10 days of exposure to NaCl solutions,obvious corroded areas and corrosion pits can be found at the surface of pure SNAP coating,as indicated by the red circles (Fig.12a).The superiority of SNAP-M coating is slight because several corrosion areas can also be found after long-term exposure (Fig.12b),implying the very limited benefi by the embedding of unmodifie MSNs in the coating.By comparison,only two corrosion spots can be identifie for the SNAP-M@L coating (Fig.12c),suggesting a certain degree of contribution of the unexfoliated LDHs nanoplates to corrosion protection.Especially,for the SNAPMM@L coating (Fig.12d),corrosion pit cannot be observed visually within the investigated time,confirmin the excellent barrier properties and robustness of the new prepared SNAP coating in corrosion protection.

3.4.Summary of corrosion protection mechanism

Fig.13.Schematic diagram for representation of the corrosion protection mechanism of the MSN-MBT@LDH nanocontainers incorporated SNAP coating on Mg alloy.(a) nanocontainers incorporated hybrid coating with local damage,(b,c) healing process including desorption of physically adsorbed MBT,release of intercalated MBT by ion exchange,and leakage of MBT from MSNs surface by the dissolution of LDH nanoshells,and (d)SNAP coating after self-healing.

The healing mechanism of the nanocontainers-incorporated SNAP coating is schematically shown in Fig.13.The excess LDHs nanoplates,which were not exfoliated thoroughly,were incorporated into the SNAP-based coating along with the MSN-MBT@LDH nanocontainers,resulting in a hybrid SNAP coating involving MSN-MBT@LDH as primary nanocontainers and LDHs nanoplates as secondary nanocontainers (Fig.13a).The amount of the un-exfoliated LDHs in the coating heavily depends on exfoliation condition and quantity of MgAl-LDH powder in the formamide solution(see Section 2.2).A very low quantity of LDHs during adsorption process will lead to the incomplete wrapping of the nanopores of the MSNs and premature leakage of MBT from MSN-MBT@LDH nanocontainers and decrease the long-term corrosion protection of the smart coating;a very high quantity will increase the thickness of the LDH nanoshell on the MSN surface and the concentration of LDH nanoplates in the SNAP coating.The increased thickness of LDH shell may result in a very slow release rate of MBT[46],though the dispersed un-exfoliated LDH nanoplates can serve as nanoreservoir of MBT to increase the coating corrosion resistance.The optimization of the LDHs concentration during the encapsulation process is beyond the scope of this work and is not discussed here.

When the hybrid coating is exposed to NaCl solution with enough long time,a corrosive pathway will be formed,as depicted in Fig.13a.The aggressive Cl-may encounter dispersed LDH nanoplates during its diffusion from outside environment toward the coating/metal interface along the pathway,resulting in entrapment of Cl-and release of MBT by desorption of the physical adsorbed molecules and ion exchange to decrease the concentration of aggressive ions in the local environment and weaken the corrosion attack to Mg alloy (Fig.13b,c),which accounts for the higher |Z|?=0.01Hzvalues of SNAP-M@L coating than SNAP-M coating.However,because of the relatively low concentration of dispersed LDH nanoplates in the SNAP coating in this work,the superiority of SNAP-M@L coating compared to SNAP-M coating in corrosion resistance is slight.When the corrosive Cl-ions along with water molecules reach coating/metal interface after breakthrough the resistance of dispersed LDHs nanoplates,hydrogen evolution and local alkalization with local pH values of above 10-11 occur in the micro-cathodic areas owing to formation of hydroxide ions originating from the water or dissolved oxygen reduction reactions [14,65],

while the pH changes of the micro-anodic area heavily depend on the compositions of the metal and corrosive electrolyte in the microdefects.The main anodic reaction for pure Mg in a corrosive media without corrosion inhibitor is,

which also produces H2and hydroxyl,and thus leads to slight alkalization even at the anodic sites [14,19,66].However,pH in micro-anodic regions of corroding AZ31 Mg alloy was found to be slight decrease down to 5.3 due to the co-dissolution and hydrolysis of alloying Al (ca.2.5-3.5%)[66,67]:

Furthermore,if an appropriate corrosion inhibitor was introduced into the NaCl corrosive media,the anodic dissolution of the AZ31 Mg alloy can be significantl suppressed and the pH values can be kept in rather narrow acidic range,5.3±0.5,even after several days of exposure [14].

Thus,the corrosion protection mechanism can be briefl summarized as follows:a certain amount of MBT is firstl released by dissolution of the LDH nanoshells of the nanocontainers owing to the decreased pH value (～5.6) arising from dissolution of CO2from atmosphere in the water and formation of carbonic acid [68].The MBT molecules released from MSN-MBT@LDH nanocontainers can be adsorbed and deposited immediately at the defects of the Mg alloy by strong covalent Mg-S group to form a stable fil to slow down both anodic and cathodic reactions [69,70],ensuring the pH values in the defect are close to neutral rather than quick alkalization of the solution to relatively high pH values.Mg and Al will be dissolved if the concentration of released MBT is not high enough to effectively suppress the corrosion processes.With the hydrolysis of dissolved Mg and Al cations,especially for the latter,the pH value in the anodic zones decreases further,possibly down to 4.8 [14].The decreased pH accelerates the dissolution of LDH nanoshells and facilitates release of more MBT molecules,thus provides excellent and long-term corrosion protection properties to substrate in NaCl environments(Fig.13d).

Previous study demonstrates that the guest molecules are hard to be released from the nanocontainer with a weight ratio of 8:1 (guest-loaded MSNs:LDHs) within dozens of hours in a pH 5.0 buffer solution.Considering that the un-exfoliated LDH nanoplates are much thicker than the nanoshells on the MSNs surface [41] and the pH value of the micro-anodic area cannot decrease to a very low level by hydrolysis of Mg and Al cations,thus the intercalated MBT cannot be released considerably by the dissolution of stacked LDH nanoplates.In other words,the released MBT for corrosion inhibition mostly comes from the MSN-MBT@LDH nanocontainers by dissolution of the LDH nanoshells.

4.Conclusion

We were able to demonstrate the fabrication of a smart SNAP coating on Mg alloy with highly enhanced corrosion resistance by incorporation of pH-controlled MSNMBT@LDH nanocontainers consisting of MBT-loaded MSN core and LDH nanosheets shell.The mechanized MSNMBT@LDH nanocontainers are obtained easily via a simple two-step adsorption process and show a pH-dependent release of MBT from MSN under acidic conditions.Compared to pure SNAP coating,the MSN-MBT@LDH nanocontainers embedded SNAP coating exhibits significan enhancement and better robustness in corrosion resistance in NaCl solution.This is attributed mainly to the corrosion inhibition of MBT to the defects occurred at the coating-substrate interface and partially to the entrapment of aggressive Cl-ions taking advantage of the excess LDH nanoplates that were not adsorbed on the MSN surface.In view of the low cost and simplicity in construction of pH-responsive MSN-based and MBT-loaded nanocontainers in this work,we believe the proposed strategy will pave the way to exciting construction of inhibitor loaded nanoreservoir and its extensive application for preparation of smart anticorrosive coatings in the fiel of corrosion science.

Declaration of Competing Interest

The authors declare that they have no known competing financia interests or personal relationships that could have appeared to influenc the work reported in this paper.

Acknowledgments

This work has been supported by Sichuan Science and Technology Program (2018JY0483),Young Elite Scientists Sponsorship Program by China Association for Science and Technology (YESS,2018QNRC001),Natural Science Foundation of Hunan Province (2020JJ4073),Open Project of Chemical Synthesis and Pollution Control Key Laboratory of Sichuan Province (CSPC202004),Foundation of Huaihua University Double First-rate Applied Characteristic Discipline Construction Projects of Materials Science and Engineering (19CKA002),and the Fundamental Research Funds of China West Normal University (CXTD2020-1).CJZ acknowledges the support from the National Science Foundation(CHE 1566283).

Supplementary materials

Supplementary material associated with this article can be found,in the online version,at doi:10.1016/j.jma.2020.11.007.