Anticorrosive PEO coatings on metallic cast heat enhancers for thermal energy storage

ABSTRACT The research aims to determine whether a protective oxide coating obtained by plasma electrolytic oxidation (PEO) will prevent unfavourable changes occurring on the surface of aluminium heat enhancers for thermal energy storage (TES) based on phase change materials (PCMs). For domestic purposes (short-term, solar energy), salt hydrates are widely utilised as PCMs. Their low thermal conductivity makes the application of metal enhancers necessary to improve heat transfer in the unit. Due to the chemically aggressive nature of MgCl2·H2O, aluminium enhancers can be negatively affected during working cycles. PEO was proposed to overcome corrosion issues in the units. Cast samples, coated with PEO for a short time in KOH-Na2SiO3 electrolyte, were subjected to a molten MgCl2·H2O environment. Mass change and the surface were studied via SEM, EDS, and XRD measurements. A thin layer of aluminium oxide prevented castings from changes occurring on the surface of the enhancer.


Introduction
Thermal energy storage (TES) for latent heat with the use of Phase Change Materials (PCMs) is becoming an increasingly recognised method of heat storage.Nevertheless, TES can be negatively affected by a number of different factors.One of the most popular groups of PCMs are salts and salt hydrates mixtures, especially when it comes to solar thermal energy storage.The main obstacles connected to their application are not only low thermal conductivity extending charging time and increasing the temperature gradient in the device, but high corrosivity as well.The first issue can be overcome with the use of metal enhancers of custom geometry, which improve heat transfer in the system.The most common shapes are fins, double pipes, honeycombs, or biomimetic shapes such as snowflakes or trees, symmetrical and asymmetrical [1][2][3][4][5][6].
One of the most crucial problems for PCM units is the issue of corrosion, not only due to the limited lifetime of the storage unit but also because corrosion products affect the thermophysical properties of PCM [7][8][9].Mass changes caused by corrosion in thermal storage applications were observed by [9].They presented the analysis of corrosion behaviours of different kinds of stainless steel and FeCrAl alloy in nitrate, chloride and carbonate salt melts.The authors reported that the most reasonable choice for a controlled quiescent corrosion process was the use of nitrate salts; in other cases, the rate of degradation was high and hard to predict.Aluminium, often exploited in TES units because of its ease of formation and low weight, is exceptionally prone to corrosion by deep-pitting in the harsh environment of phase change materials based on e.g.sodium chloride or sodium acetate [10,11].Its corrosion resistance is noticeably lower than the one exhibited by copper [10], and it decreases severely when in galvanic contact with other metals.Aluminium alloys working in neutral pH conditions usually are sufficiently protected against corrosion due to the creation of a thin oxide layer by passivation.Nevertheless, this natural coating does not withstand the harmful environment of acidic or alkaline pH [12] in which it dissolves, exposing the material beneath.Phase change materials based on chloride hydrates are extremely detrimental to aluminium and its alloys [13], especially for the ones with Si as the main alloying compound.Such alloys, called silumins, suffer from localised corrosion, which rate is notably affected by the Si content [14,15].Therefore mitigation measures such as the deposition of protective surface treatments for these materials are of general interest (e.g.anodising, plasma electrolytic oxidation (PEO), sol-gel).
PEO layer originates directly from the surface of the coated metal, and therefore, it provides an excellent adherence to the substrate, offering also enhanced corrosion resistance and wear performance, more beneficial than observed for regular anodising.PEO process is environmentally friendly as it does not require any toxic chemicals for its deposition, using only inert low-concentrated water-based electrolytes [16].This technique is conventionally applied for lightweight metals (i.e.Ti, Al, Mg) and allows for a relatively easy coating of complex shapes without any special pretreatment.Moreover, some of the properties which are desired for TES applications, such as favourable thermal conductivity, heat emissivity, or thermal diffusivity, can be tailored by the addition of various particles to the electrolyte and their incorporation in fabricated layers [17][18][19].Nevertheless, despite the fact that PEO coatings possess extraordinary properties, there are also some difficulties in their utilisation, such as the porosity of thus produced layers and proper coverage of large areas.
Silumins are widely used in industry due to their good mechanical properties and relatively low cost [20].Therefore, the impact of silicon amount in the aluminium alloys on the PEO coating's formation has been widely studied.It was shown in [21] that the addition of silicon in the alloy might be responsible for the lack of high voltage during the PEO process, but increasing the amount of Si could cause an increase in layer porosity.In [22] PEO coating was applied to an alloy containing 9% wt.Si using constant current mode and constant voltage mode, subsequently.The electrochemical impedance spectroscopy test proved that the sectionalised PEO coating procedure provides better results in corrosion protection in comparison to the results obtained for bare alloy.Proven corrosion protection was also depicted in [23], where aluminate-based electrolytes and the addition of K 2 TiF 6 were proposed for AC-46500 Al-Si (11% Si wt.) secondary cast alloy.The obtained layers were thick and dense and the corrosion resistance was reported to be improved by approximately ten times due to the use of the electrolyte solution based on KOH, Na 4 P 2 O 7 ·10H 2 O and Na 2 SiO 3 ·9H 2 O with the addition of 0.5 g/L TiC nanoparticles [24].Another direction, especially beneficial for TES-related applications, was to manufacture a highly thermally conductive coating by means of PEO on an aluminium substrate [25].The process was carried out on untreated samples, only degreased and cleaned, in the concentrated electrolytea solution of 40 g of sodium silicate in 1 L of water.The developed layers were approximately 10 µm thick and were characterised by thermal diffusivity nine times higher than observed for bulk Al 2 O 3 .
For now, no wide-scale studies and research regarding the use of PEO coating as an anticorrosive layer for thermal energy storage applications have been published.Only one investigation [26], conducted by the authors of this work, was published on the lab-scale performance of the PEO-coated cast 44200 aluminium alloy heat transfer enhancer.After several working cycles in a nitrate salt mixture as PCM, no signs of corrosion or negative effect on the charging and discharging rates were observed.In this paper, the application of a thin oxide layer by means of plasma electrolytic oxidation was proposed to improve the lifetime of heat enhancers in TES units.The PCM of choice was MgCl 2 ·H 2 O, known to be characterised by high corrosivity.Metal samples were cast by the combined methods of additive manufacturing for model preparation and investment casting.Anticorrosive PEO coating was deposited on the samples cut from the casting.The composition of the used electrolyte was inspired by other works, where the aluminium alloys with high content of silicon were studied [27][28][29][30], and the authors claimed to use electrolytes based mainly on sodium silicate and sodium or potassium hydroxide.Coated and as-cast samples were subjected to the molten salt environment and tested at elevated temperature for up to 20 days.The surface structure and the products obtained on the surface were observed for both the coated and uncoated samples, and the mass change during the process was considered.

Sample preparation
Metal samples were prepared via the investment casting method.The shape of the honeycomb-like enhancer was designed and modelled in Autodesk Inventor Professional software.Based on the Gcode file generated by the PrusaSlicer tool, the model was 3D-printed with Prusa IMK3 printer using the fused deposition modelling (FDM) technique.Polylactide (PLA) filament was chosen for this process.The obtained model was immersed in a gypsum slurry and left to cure.When solidified, the gypsum moulds were burnt out at around 720°C and cast with molten aluminium alloy EN AC 44200 (composition of the alloy is given in Table 1).The specimens for experiments were cut off the structure and, thereafter, subjected to the PEO process to be covered with a protective oxide layer.The manufactured casting is shown in Figure 1(a) and (b).
For the roughness profile determination, a Mehr Surf PS 10 was used.

Development of PEO coating
For the coating process, samples cut from the casting (surface area approximately 30.1 cm 2 , Figure 1(c)) were cleaned and degreased with acetone in an ultrasonic cleaner for 10 min and dried with hot air.No grinding or polishing was applied to the samples so as not to affect the shape of the structure.The stainless steel and the specimen with a holder were used as a cathode and an anode, respectively.The electrolyte was composed of sodium silicate (Na 2 SiO 3 ) and potassium hydroxide (KOH) with a concentration of 10 g/L each in one case, and in the second, the concentration of Na 2 SiO 3 was duplicated and amounted to 20 g/L.The PEO process was conducted using a custommade PEO unit (Microarc, Poland).The current density was set at 33 A/dm 2 ; no inverse current was used.The duty cycle was 30% (t on : t off = 0.3 ms: 0.7 ms).The electrolyte was kept at a temperature below 40°C.The processing time in both cases was 530 s.The sample with a thicker PEO layer consisting of two phases, compact and porous, was chosen for further investigation.

Mass change
Rectangular samples cut out of the cast structure, both coated and uncoated, were weighed on a MEDICAT LTD 160 M scale (with a precision of 0.001 g) and placed in conical flasks, then covered with magnesium chloride salt hydrate (MgCl 2 ·H 2 O) purchased from Biomus sp.z o.o., 50 g each.The flasks were sealed tightly, placed in the furnace preheated to 130°C, and kept at elevated temperature for 480 h (20 days).The mass change was measured after 2, 4, 19, 25, 40, 70, 113, 120, 360, and 480 h.After that time, the flasks were opened, metal sheets were taken out, gently rinsed in water, dried, and weighed again to determine the changes in mass.

Microstructure characterisation
The microstructure and surface morphology of the samples and the layer thickness were observed and assessed with a Hitachi TM3000 scanning electron microscope equipped with an EDS/EDX analyser.Additionally, XRD measurements were conducted with the use of an Ultima IV X-ray diffractometer (Rigaku, Japan, 40 kV/40 mA, step size 0.05).

Results and discussion
Aluminium samples prepared through investment casting are characterised by highly developed surfaces.This is connected not only to the complex, layershaped polymer model but to the porous nature of castings as well.As mentioned before, no pre-treatment was applied to the surface of the specimens.Nevertheless, it has been reported in the literature that proper surface treatment can significantly  improve the layer growth rate and reduce energy consumption.For instance, Li et al. [20] pointed out that surface pre-treatment, such as chemical etching of samples made of Al-Si alloy using HNO 3 and HF, increased the growth rate by almost 40%.
Figure 2 shows the roughness profile of the manufactured castings.The average value of Ra (arithmetic mean deviation) reached 13.27 µm, and Rz (maximum height) -56.25 µm.The hill-like structure replicates the pattern of layers created during the printing of the polymer model.

The process of coating growth
Figure 3 shows the variation of voltage in time during the development of the PEO coating for both cases.The number of gas bubbles arising from the sample surface increased with increased voltage in the experiments that use PEO in electrolyte of lower concentration.At first, the voltage grew significantly but steadily.The first part of the graph corresponds to anodic oxidation [30] when the voltage exceeds 150 V, the temporary growth stop is visible.This might be connected to the high silicon content and porosity of the treated alloy.In [32] it was reported that in the case of porous materials prepared by additive manufacturing, the alloy, especially near the pores, can be dissolved by the electrolyte, therefore impairing the formation of an oxide layer.A similar potential drop was reported in [23] where an aluminate electrolyte was used.The authors claim that the use of such electrolytes may cause the development of a film preventing from a process occurring directly on the metal surface.
After that, the voltage rose gradually to about 330 V and yellow arc discharges began to occur on the treated surface.The breakdown voltage was approximately 325 V.In approximately the 160th second, the process stabilised and ran at about 400 V, slowly increasing the voltage value.In the last part of the oxidation, the sparkling connected to the microarc discharges was uniform and intense, while the maximum voltage value reached 430 V.According to Wang et al. [33], the first stage of the coating growth is associated with anodic oxidation and the formation of Al and Si oxides on the alloy's surface.Then, in the second stage, the discharges appeared at the interphase between silicon and aluminium.The third part of the process described by the authors, related to the increase in voltage, has not been reported in this case.
For PEO with a higher electrolyte concentration, the shape of the curve is rather comparable.It is noteworthy, nonetheless, that the time needed to overcome each step of the process is definitely longer and the maximum voltage values are lower.The potential does not remain constant, single drops appear.It seems unlikely that the PEO process occurred correctly since the voltage did not rise above 300 V, higher values are expected for aluminium-silicon alloys.At such low voltage, anodising rather than plasma oxide generation is suspected.

Composition and microstructure of oxide layers
Figure 4 shows cross-sections of samples coated with PEO according to the previously described process and embedded in the resin.Figure 4(a) shows a layer grown in an electrolyte containing a low concentration of Na 2 SiO 3 .The coating consisted of two thin partsa more compact film near the alloy domain of the sample, and a highly porous one.The average thickness of the layer is approximately 6-8 µm.In Figure 4(b), a higher concentration of Na 2 SiO 3 (20 g/L) was used, resulting in a thinner coating (not exceeding 5 µm) consisting of only one layer that appears to be non-continuous after 530 s of the process.
Thus, for the next test, a sample processed in the lower concentration electrolyte was chosen.
The morphology of the sample cross-section after coating is shown in Figure 5(a).The characteristic layered, bulge-shaped structure of the samples is related to the manufacturing method (layers are formed during the model printing phase).Silicon crystals are visible, occurring as homogeneously distributed fractions throughout the sample's volume, but their position near the surface may have a direct effect on the oxidation process at a particular point.
The pores are distributed evenly over the depicted area, with varying diameters not exceeding 10 µm (see Figure 5(b)).
The SEM-EDS results for the coating obtained on AC 44200 alloy are given in Table 2, and the EDS spectrum is shown in Figure 6.The presence of potassium and sodium in small amounts is related to the composition of the electrolyte in which the process was conducted, the same was observed in EDX measurement in [34].Several atoms were incorporated into the structure of the coating during the discharge on the sample's area.For the cross-section of the sample, the element map is presented in Figure 7.The approximate composition of the elements is clearly seen.The distribution of silicon is strongly related to the crystal structure of the alloy and coincides with the layout of dendrites formed during the casting hardening.

Mass change
The mass of the samples was measured before and after exposure to high temperatures in an aggressive solar salt environment.Figure 8 presents the changes in the mass in time.As-cast probes, cleaned with acetone, were left in molten magnesium chloride salt at 130°C for up to 20 days.At the beginning of the process, the weight gain was relatively rapid, reaching a maximum oxidation rate of 0.062 mg/cm 2 , then the oxidation stabilised within 10 days at approximately 0.046 mg/cm 2 .The reason for this can be found in the formation of an oxide layer on the surface of the samples, which separated the alloy from the aggressive environment of the molten magnesium chloride salt.Presumably, at the beginning of the process, various types of micro protrusions and roughness are subjected to oxidation, which may fall off the surface over time, resulting in a decrease in weight gain.Finally, after 20 days, the specimen maintained the same mass-gain value.
Meanwhile, the metal sheets covered with the PEO layer were measured.The samples cut from the coated casting were secured with protective varnish on the uncoated parts to avoid the reaction of the alloy  material with the molten salt.Slight differences in mass are associated with the degradation of the varnish layer.The PEO-coated specimen did not react with the molten MgCl 2 ·H 2 O, which proves that even a thin oxide layer protects the system from damage and uncontrolled changes during prolonged exposure to high temperatures in an aggressive environment.
Surface after exposure to salt environment After 5 days of exposure to molten magnesium salt, the as-cast sample was examined.EDS linear analysis of the created oxide film was conducted (see Figure 9).The results showed that the darker layer formed on the surface of the casting consisted of significantly more oxygen and magnesium the lighter part of the sample.Nevertheless, it is believed the oxide layer was formed uniformly on the alloy surface, but as it grew, part of the product detached from the alloy surface, forming conglomerates, as can be seen in Figure 10.
The surfaces of each sample were observed after exposure to molten MgCl 2 ·H 2 O.The differences between as-cast and the coated ones are evident.

XRD measurement results
Figure 12 shows the XRD results for coated and as-cast samples before and after treatment with aggressive molten salts.In the case of Figure 12(a), the aluminium (Al) and silicon (Si) peaks shown were detected from the Al-Si alloy substrates.After exposure to molten MgCl 2 , new peaks indicating magnesium aluminium hydroxide hydrate, magnesium dichloride monohydrate, and aluminium chloride hexahydrate occurred, which the authors observed as a peeling layer on the sample surface.Ushak et al. [35] studied the corrosion behaviour of 1100 aluminium alloy in MgCl 2 •6H 2 O environment.After 1500 h of exposure, the samples were subjected to XRD tests and the following corrosion products were observed: aluminium chloride hydroxide hydrate, magnesium chloride hydrate, and magnesium aluminium hydroxide hydrate.Peaks corresponding to pure aluminium were also found.These findings imply that both cations and anions (respectively Mg 2+ and Cl − ) participate in the reaction with Al substrate and contribute to the formation of new phases.In [13], Zheng et al. performed the corrosion test for aluminium alloy containing 1.8% of silicon in chloride solutionswater solutions of NaCl in quantities: 3.5%, 2% and 5%, by weight.Corrosion in the form of pitting was observed on the parts of the sample not covered with a natural oxide layergrain boundaries.They also described the formation of insoluble hydroxide Al(OH)Cl 2 − .In the case of the  alloy studied by the authors, the products formed during the corrosion are hydroxides as well, the difference in the composition might be connected to the different salt used, thus magnesium can be present in the produced sludge.Figure 12(b) shows the specimens coated with PEO.Compared with the XRD of as-cast sample before molten salt treatment, only γ-Al 2 O 3 was additionally identified in the coating.The remaining Al and Si peaks correspond to the alloy composition and possible cavities in the coating.This composition may be related to the duration of the process.Wang and Nie [36] reported that for the 319 and 390 aluminium alloys coated in an electrolyte consisting of 4 g/L of Na 2 SiO 3 and under the density current of 0.05 A/ cm 2 , the layers obtained after the first five to ten minutes consisted of only Al 2 O 3 , Si and Al.
However, no additional substrates corresponding to the corrosion process were registered after treatment in molten MgCl 2 .This fact may confirm that PEO layers, even with a thickness of a few micrometres, are a sufficient barrier to separate the metallic material of the casting from the aggressive salt environment.
J. Li and X. Tan [14] performed a series of corrosion tests in two different aluminium alloys.Each of the alloys contained different quantities of silicon.The corrosion test was conducted using MgCl 2 solution.Drops of the solution were placed on the surface of the samples, and the test was carried out using Kelvin probe.They observed that the content of Si in the alloy can significantly affect the corrosion form and rate.The higher Si concentration in the alloy, the easier it is for corrosion to form on the sample surface.Moreover, in the case of a sample containing a higher value of Si (excess of silicon particles), meta-stable pitting corrosion was observed, and the morphology of the sample changed around Si crystals on the alloy surface.It was concluded that the excess of silicon particles was more significant to the process than the initial concentration of the salt solution.In view of the proposed application of EN-44200 alloy, the key is to isolate the interface between silicon crystals (visible close to the surface, as in Figure 5(a)) and molten chloride salts.According to the presented test results, is achieved adequately by PEO coatings.

Corrosion mechanism
Basing on conducted research and literature review, the corrosion mechanism is proposed as follows.During melting, water built in the structure of MgCl 2- •6H 2 O undergoes reaction of hydrolysis and as the effect, MgO and HCl are produced.This leads to the reaction between HCl and metal immersed in the salt, causing the production of magnesium oxides and, in the studied case, aluminium chlorides and chloride hydrates.The process was described in ref. [37], where authors studied hot corrosion behaviour for application in thermal energy storage.Alloys were tested, among others, in the mix of chloride alts: MgCl 2 , NaCl, and CaCl 2 .The corrosion reaction of siliconmagnesiumaluminium alloy in chloride salt solutions was as well described by Zheng et al. [13].According to their work, it is believed that the corrosion process starts at grain boundaries, which are not covered with a protective, natural passivation layer.Ions of Cl -and OH -present in the electrolyte, as well as ions of Al 3+ formed during the anodic reaction of the metal substrate and the solution of the system, form together aluminium hydroxide hydrate, which is insoluble and visible also on the graphs in Figure 12.

Conclusions
In the present study, PEO coating was successfully developed on an Al-Si alloy surface.The structure of the coating and its anti-corrosion properties during long-term exposure to molten magnesium chloride salts at elevated temperature have been tested.Based on the above results, the following conclusions can be drawn: . A thin oxide layer on the surface of Al-Si alloy was obtained through short-term exposure to PEO treatment in an electrolyte consisting of 10 g/L KOH and 10 g/L Na 2 SiO 3 .After 530 s of the process, a layer approximately 6-8 µm thick was developed.The layer was composed of 50% Al and almost 41% oxygen, with residuals of K, Na and Si.In the case of the sample processed in the electrolyte with a higher concentration of Na 2 SiO 3 , it was not possible to obtain high voltage values (400 V or more) in the same time range, and it is assumed that no plasma treatment occurred.Additionally, the layer obtained was thinner (not exceeding 5 µm) and not uniform, which could be ineffective in protecting the surface against salt corrosion. .The presence of a large amount of silicon in the alloy (up to 13.5%) and the highly developed surface of the casting caused drops in voltage during the coating process. .After the first 5 days, an oxide layer was formed on the surface of the uncoated sample, which under continuous exposure to the salt environment, exfoliated and formed an uneven oxide film.In the case of PEO-secured metal sheets, no changes were observed after 20 days.The coating prevented further uncontrolled oxidation of the alloy. .In the first days of exposure to an aggressive environment of molten magnesium chloride hydrate at elevated temperature, the as-cast sample gained approximately 0.062 mg/cm 2 , and after 10 days, the mass gain decreased and stabilised at 0.46 mg/cm 2 .The reason for this rapid weight gain is considered to be the presence of impurities and their oxidisation on the sample surface.The subsequent stabilisation indicates the formation of insoluble precipitates on the sample surface, the presence of which appears to be confirmed by XRD analysis.According to the literature, there is a possibility of pitting and increased corrosion activity at grain boundaries or silicon crystals near the sample surface.In the case of the PEO-coated sample, XRD analysis showed no additional products, as well as no mass gain was observed over 480 h.This leads to the conclusion that an oxide PEO coating deposited on the surface of an aluminium-silicon alloy results in a reduced likelihood of undesired corrosion and deterioration of metal components embedded in the molten salt deposits used in thermal energy storage accumulators.

Figure 1 .
Figure 1.Cast structure before cutting: (a) vertically, (b) horizontally, and (c) sample cut for the PEO processing.

Figure 3 .
Figure 3. Voltage evolution during the PEO process for both prepared samples.

Figure 4 .
Figure 4.The cross-section of PEO-coated samples processed in electrolytes consisting of: (a) 10 g/L of KOH and Na 2 SiO 3 ; (b) 10 g/L of KOH and 20 g/L of Na 2 SiO 3 .

Figure 5 .
Figure 5. PEO layer formed on the sample: (a) cross-section with a visible pattern of the casting and silicon crystals; (b) the porous surface of the coated sample with uniformly deposited pores.

Figure 6 .
Figure 6.EDS spectrum recorded on the top part of the PEO layer.

Figure 7 .
Figure 7. Element mapping on the specimen's cross-section: leftmicroscopic picture from the analysis; rightmaps of oxygen and silicon deposition.

Figure 11 (
Figure 11(a) and (b) represent sheets without PEO coating, after exposure to MgCl 2 ·H 2 O for 5 and 20 days, respectively.Figure 11(c) and (d) correspond to coated specimens, with the same testing times as for Figure 11(a) and (b), respectively.In Figure 11(a) (5 days of salt exposure), one can see the formation of a relatively uniform oxide layer over the whole surface of the sample.The brighter spots signal areas where the layer had started to exfoliate, as described above (Figures 9 and 10).The horizontal lines visible

Figure 8 .
Figure 8. Oxidation kinetics of as-cast aluminium and PEO coated specimens.

Figure 10 .
Figure 10.Exfoliating oxide layer forming an agglomerate at the interface between the pure alloy and the oxide layer.

Figure 12 .
Figure 12.XRD patterns with identified phases of: (a) as-cast sample before and after 20 days of exposure to molten salts; b) PEO coated sample before and after 20 days of exposure to molten salts.

Table 2 .
Composition of PEO coating obtained by EDS analysis.