Influence of the nitrogen content on the corrosion resistances of multicomponent AlCrNbYZrN coatings

In this study, the relationship between the nitrogen content and the corrosion resistances of non-equimolar multicomponent AlCrNbYZrN films (N = 13 – 49 at.%) is probed. While there was no linear relationship be-tween nitrogen content and corrosion resistance, the results clearly show that the corrosion resistances of the films were instead determined by their nitrogen-induced porosities i.e. the less porous the sample, the higher the corrosion resistance. The 23, 30 and 37 at.% N samples were denser while the 13 at.% N sample was porous and the 49 at.% N film had an underdense nanocrystalline columnar cross section permitting the ingress of electrolyte.


Introduction
The emergence of high entropy alloys has driven an upsurge in the development of alloys containing multiple metallic elements broadly known as multicomponent alloys [1].This has made it possible to design alloyed coatings with improved corrosion resistances, abrasion resistances and mechanical properties [2].The design of a truly exceptional coating must involve the simultaneous and scrupulous consideration of several factors which synergistically affect the materials properties.To obtain a high corrosion resistance the thermodynamic stabilities of the elements used in the alloy, the microstructure and porosity of the alloy, as well as the possibilities to form a passive layer on the surface of the alloy all need to be considered with respect to the application in question.
At present, most coatings in commercial use are binary or ternary metal nitrides.However, since the simultaneous realization of multicomponent alloys by Cantor et al. [1] and high entropy alloys by Yeh et al. [3] many new multicomponent materials with atomic level complexities, versatile functionalities and excellent combinations of mechanical and corrosion properties have been made [4][5][6].
Furthermore, the addition of a p-element such as nitrogen to form nitrides often gives even better material properties than that of the pure metallic alloy [7,8].Due to their high tolerance to heat, physical wear and tear and chemical inertness, metallic nitrides are front-runners for coatings in cutting and precision tools [9].As the surface of a material may be exposed to wear and tear, heat, corrosion, radiation, electrical/magnetic fields, the base material is then coated to obtain a composite surface that can withstand these aggressive conditions.The nitride coating on the substrate could be so thin (less than 10 μm), that it adds essentially nothing to the mass of the substrate while imbuing the surface with longevity, strength and increased functionality [10].For instance, Barshilia et al. [11,12], who deposited NbN, TiN coatings and composites thereof on tool steels, found that the corrosion current density (CCD) of the coated tool steels was reduced by two orders of magnitude in 0.5 M HCl.
Multicomponent alloys, which combine multiple alloying metals in almost equal concentrations, and their nitride analogues, are new classes of alloys being developed at breakneck pace for hard coating applications due to their versatile mechanical properties and corrosion resistances [7,13,14].
The rationalization for the combination of so many elements, as in a multicomponent alloy, is based on the thermodynamic definition of free energy change which determines the chemical stability of a certain phase [15].The mixing entropy (ΔS mix ) for a randomised solid solution is given by Where R is the ideal gas constant, N is the number of elements and c i is the atomic fraction of component i.The higher the number of alloying elements, the higher the entropy in the metal lattice [15].Additionally, alloys in which components are present in equal amounts have the highest entropy.Thus, a multicomponent alloy with five components in equal amounts has a mixing entropy of 1.61R, compared to that of a binary alloy with 0.69R [16,17].
However, the entropy parameter is not the sole determinant of the stability of a certain phase.The enthalpy of mixing (ΔH mix ) must also be considered.
Where N is the number of elements and c i/j is the atomic fraction of component i/j.If the Gibbs free energy (ΔG) for a random solid solution (RSS), which is determined by both S mix and H mix , is lower than the Gibbs free energy change for the formation of an intermetallic (IM) phase, the random solid solution is favored [15].
The above thermodynamic guidelines have driven the design of many hybrid multicomponent materials [4,13,18].This has given credence to the belief that the presence of a dominating single-phase solid solution (bcc/ccp) is central to the realization of improved corrosion properties [19][20][21].The presence of intermetallic phases has often been diabolized and linked to poor corrosion resistance with reasons such as elemental segregation, inclusions, heterogeneity and dealloying being cited to explain the lack of performance [19][20][21].
It is complex to determine the thermodynamic equilibria of a multicomponent matrix and the resultant material properties [13,22].While the thermodynamics report on the theoretical tendency of corrosion, kinetics dictate the corrosion rate that can be observed experimentally.This means that while thermodynamics can predict if a specific phase should corrode or not, it cannot predict the corrosion rate (and hence not the corrosion current density).The corrosion rate is typically governed by the rate of the electron transfer rate or the mass transport at the metal/solution interface which depend on a number of parameters including the morphology of the material and its content of metals forming passive oxide layers.As a result, the corrosion rates generally need to be determined experimentally.In an accelerated corrosion test such as a potentiodynamic polarization experiments, a corrosion-resistant material exhibits low current densities over a wide potential range [19].
At present a small number of studies have explored the relationship between the nitrogen content and the corrosion resistance of multicomponent nitrides.In one study this relationship was studied for coatings composed of (AlCrSiTiZr) 100 − x N x (with 0-53 at.%N) by means of an anodic polarization in 0.1 M H 2 SO 4 from -0.5 to +1.5 V vs. the open circuit potential (OCP) [8].The authors noted that the sample with the highest nitrogen content exhibited the lowest corrosion current and that the corrosion behaviors of the (AlCrSiTiZr) 100 − x N x coatings were highly dependent on the employed nitrogen flow rate ratio used in the deposition of the coatings.The latter was most likely a result of the nitrogen content affecting the composition and structure of the coating, as well as the bonding to the substrates [8].In another study Pei et al. [23] explored the effects of nitrogen alloying on the microstructure and pitting corrosion resistance of CrCoNiN medium-entropy alloys in NaCl.It was found that nitrogen addition impeded the initiation of metastable pitting, decreased the CCDs and increased the pitting and repassivation potentials [23].
While AlCrNbYZrN coatings have been studied based on their excellent mechanical properties [24,25], the influence of the nitrogen contents on the corrosion behavior of such coatings has not been previously explored.As part of larger study focused on the mechanical properties, von Fieandt et al. [25] performed a brief investigation of the corrosion resistances of AlCrNbYZr nitrides containing 46-48 at.%N. In this case, the microstructures of the films were varied by changing the bias and temperature.Although it was found that nitride films with lower porosities and surface roughnesses featured lower corrosion currents [25], the influence of the nitrogen content on the corrosion resistance was not studied.Al, Cr, Nb, Y and Zr were chosen due to their ability to form passivating oxide films, their high mutual solid solubility, and their ability to form stable nitrides which offer improved chemical resistance in chemically aggressive environments [7,26,27].Kinetic stabilization was provided by the fact that the coatings were made using magnetron sputtering [28].During the sputtering process, the deposition of solid films directly from the gas phase leads to very high quenching rates (>106 K/s) of the adsorbed atoms during film growth [29,30].The atoms impinging on the surface of the substrate lose energy very quickly, are immobilized, and become incorporated into the growing film [29,30].This should increase the probability of forming metastable single-phase solid solutions, even if that is not what is thermodynamically predicted under equilibrium conditions [29,30].Despite this kinetic 'loophole' afforded by magnetron sputtering, Al, Cr, Nb, Y, and Zr still readily form intermetallic phases with each other [31,32] which, as mentioned previously, is in discord with the general thermodynamic precepts for the design of a corrosion resistant multicomponent alloy.In another study by von Fieandt [24], it was also shown that the mechanical properties of AlCrNbYZrN films (with N = 15-51 at.%) depended significantly on their nitrogen content.As it is important to be able to manufacture films which combine good mechanical properties with high corrosion resistances there is clearly a need for a detailed study also of the influence of the nitrogen content on the corrosion resistance for this type of films.
The aim of the present work is to study the influence of the nitrogen content on the corrosion resistance of AlCrNbYZrN coatings for nitrogen contents between 13 and 49 at.% made with magnetron sputtering.In addition, this study aims to investigate if the formation of a random solid solution is truly cardinal to the attainment of a corrosion resistant multicomponent material.The influence of the morphology of the coatings on the corrosion resistance is likewise studied as the porosity of the films can be expected to heavily influence the CCD, the passive behavior as well as the overall electrochemical behavior of the films.A combination of electrochemical, analytical and spectroscopic techniques is used to understand how the addition of nitrogen impacts the corrosion behavior of AlCrNbYZr films.The results are discussed and compared with thermodynamic expectations using Pourbaix diagrams for the metals present in the coatings.

Synthesis of thin films
A sputter system (Qprep500i from Mantis Deposition Ltd.) with an ultra-high vacuum chamber containing three magnetrons in a sputterdown configuration and a target-to-substrate distance of 13 cm was used to deposit the films.The base pressure of the chamber was 3.7⋅10 − 10 Torr (4.9⋅10 -8 Pa) at room temperature and never above ~9⋅10 -7 Torr (~1.2⋅10 -4 Pa) during deposition at 500 • C. Two of the three magnetrons consisted of segmented targets (3 ′′ diameter): Al/Cr (1:1) and Nb/Zr (1:1), while the third magnetron (2 ′′ diameter) only contained the Y target.The claimed purities were ≥99.9 % for the Al, Cr, Nb and Y targets, while the purity grade for the Zr target was 702.The substrates used were silicon(100) for scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS), and α-Al 2 O 3 (001) for corrosion analyses and X-ray diffraction (XRD).The substrates were cleaned in acetone and ethanol in an ultrasonic bath for five minutes and dried in flowing nitrogen before introduction into the deposition chamber.The substrates were preheated for at least one hour to avoid the formation of temperature gradients.The temperature was monitored with a thermocouple calibrated using a dual beam pyrometer on the substrates.
Prior to film growth, the substrates were cleaned by Ar + ion bombardment for two minutes, using an rf substrate bias of − 150 V.The targets were also cleaned prior to deposition by using the same target powers as used in the depositions.The working pressure in the chamber during film growth was 7.0 mTorr (i.e.0.93 Pa).A rotating substrate holder was used to ensure homogenous composition of the resulting films and a substrate rf bias of − 50 V was used for all depositions.The target powers were constant for all depositions and set to 130 W for Al/ Cr, 250 W for Nb/Zr and 50 W for Y.The films had thicknesses ranging from 0.8 to 1.1 μm.The gas mixture consisted of Ar and N 2 (99.9999 % and 99.9995 % purity, respectively), and the total flow rate was held constant at 60 sccm, although the Ar and N 2 flows rates were changed in order to achieve different N 2 flow rate ratios (R N = N 2 /(N 2 +Ar), between 0 and 60 %.By increasing the N 2 ratio, films with different nitrogen contents could thus be obtained.

X-ray photoelectron spectroscopy (XPS)
The elemental compositions of the as-deposited films were probed by X-ray photoelectron spectroscopy (XPS) using a Ulvac-Phi Quantera II scanning XPS microprobe with monochromatic Al Kα radiation and an X-ray spot size of 100 μm.Sputter depth profiles were employed with an area of 1 × 1 mm 2 and Ar + ion etch energy of 200 eV.The sensitivity factors, used for the bulk composition determinations, were obtained from reference measurements by time-of-flight elastic recoil detection analysis (ToF-ERDA) and XPS measurements on a metallic bulk reference sample with a composition of 10 at.%Al, 30 at.% Cr, 30 at.% Nb, 10 at.%Y and 10 at.%Zr.
The surface chemical compositions of selected samples before and after 72 h in 1.0 M HCl were studied using high-resolution measurements of the Al 2p, Cr 2p, Nb 3d, Y 3d and Zr 3d core levels.The surface elemental compositions were determined using sensitivity factors provided by Ulvac-Phi.To separate the metal and/or nitride from the oxide components, peak fittings were performed.The metal and nitride core level peaks were fitted using Doniac-Sunjic-profiles whereas the oxide core level peaks were fitted using Voigt profiles [33].Lorentzian broadenings according to reported natural widths were used for all elements [34] as well as reported spin-orbit splitting's [35,36].The background was modelled by a Shirley-type function.

X-Ray diffraction (XRD)
X-ray diffraction (XRD) measurements were performed on the films with a Philips MRD X'Pert diffractometer with Cu Kα radiation.The optical setup for the X-ray beam involved a parallel beam geometry, consisting of a Göbel mirror as primary optics and parallel plate collimator with a 0.27 • divergence as secondary optics.Grazing incidence (GI) scans with an incident angle of 2 • were conducted.The cell parameters of the nitrogen containing samples were refined in the software Unitcell using a non-linear least square method.

Scanning electron microscopy (SEM)
The morphologies of the films were studied by high-resolution scanning electron microscopy (HR-SEM) using a Zeiss Merlin instrument.Fractured cross-sections for the pristine films were studied for microstructure and thickness estimation.

Potentiodynamic polarization
Potentiodynamic polarization experiments were carried out in 1.0 M HCl from -0.7 to 1.4 V vs. Ag/AgCl at a scan rate of 1 mV/s using a SP-300 potentiostat (Biologic Instruments) and the EC-lab Version 11.10 Software.Prior to these experiments, the films were left in the electrolyte for two hours to equilibrate under open circuit conditions.A threeelectrode cell was used in which the sample served as the working electrode while a platinum wire and a Ag/AgCl electrode (containing 3 M KCl) were employed as the counter and reference electrodes, respectively.The corrosion current densities, j corr, and corrosion potentials, E corr, were calculated using a Tafel fitting [37] in the EC-lab software.

Open circuit measurements and electrochemical impedance spectroscopy
The open circuit potential was measured at 30 min and at 72 h of film exposure to 1.0 M HCl.The ac impedance responses of the films were recorded at the open circuit potential after 30 min and after 72 h in 1.0 M HCl using the three-electrode cell and potentiostat indicated above.The frequency range spanned from 200 kHz to 0.1 Hz and the ac amplitude was 10 mV.Each measurement was conducted three times.

Inductively coupled plasmamass spectrometry (ICP-MS)
The ICP-MS measurements of the metal ion concentrations were performed on electrolyte (i.e.1.0 M HCl) that had been exposed to the thin film samples for two hours under OCP conditions.A NexION 300D ICP-MS (Perkin Elmer, USA), using the Syngistix™ 1.0 software and equipped with nickel cones, a cyclonic spray chamber and an ES-2040 PFA-ST MicroFlow nebulizer was used for the determinations of 27 Al, 53 Cr, 89 Y, 90 Zr, and 93 Nb while 103 Rh was used as an internal standard.
All isotopes were measured in the kinetic energy discrimination (KED) mode to reduce the polyatomic interferences.The instrument was stabilized for 45 min before optimizing the Ar gas flow to the nebulizer and mass calibrating by measuring Li, Mg, In, and U, as a part of the optimization procedure.The MS-method involved three replicate readings of 30 sweeps over the analyte mass-range with an integration time of 75 ms for each mass per sweep.The sample aspiration rate was 0.3 mL/ min with a sample pre-flush of 45 s before the analysis of each sample.After each sample the system was washed with 1% HNO 3 for 90 s.When running the instrument in the KED mode, a He flow rate of 4.5 mL/min was used.
All chemicals were of analytical or supra pure grade and the solutions were diluted with Milli-Q purified water (Millipore, Billerica, MA, USA), henceforth referred to as MQ.Hydrochloric acid (37 %, EMSURE® ISO, Merck, Germany) was used to matrix match the calibration standards.Multi-element stock solutions were prepared by dilution of appropriate single element stock solutions (Teknolab AS, Norway) in 10 %(v/v) HCl.All other standard solutions were prepared from the appropriate stock solutions.An in-house prepared optimization solution containing 1 μg/L Be, Ce, Fe, In, Li, Mg, Pb, and U in 1%(v/v) HNO 3 , was used in the daily ICP-MS optimization procedure.

X-ray photoelectron spectroscopy
The nitrogen contents of the AlCrNbYZrN films were 0, 13, 23, 30, 37, and 49 at.%.The full elemental composition and nitrogen flow ratios (R N ) of all the films are presented in Table 1.The target powers for the metals were kept the same in every deposition and only the flow rate of nitrogen was changed.The nitrogen content increased from 0 to 49 at.% with increasing R N .Increasing the flow rate increased the incorporation of nitrogen in the film and lead to a concomitant decrease in the metal content.However, the ratio between the individual metallic elements remained mostly consistent.The content of oxygen, a contaminant, was between 3-5 at.%. Stoichiometric nitride films were obtained at R N = 60 % because at this flow rate the N content approached 50 at.%.
The film with 0 at.%N was totally metallic while the 13, 23, 30 and 37 at.%N films were partly nitridic and partly metallic.The 49 at.%N film was, on the other hand a full nitride.A prior study on the mechanical properties of AlCrNbYZrN x films showed that the N1s peak positions ranged between 397.0-397.2eV, close to the reference values of AlN, CrN, NbN and ZrN, indicating that the nitrogen was present in a nitride state for all nitrogen concentrations [24,38].The core level shifts in the metal spectra indicated that the nitrogen in the films affected all the metals except yttrium [24].It was thus assumed that yttrium, at least partly, was present in its metallic form.

X-ray diffraction
X-ray diffraction was used to evaluate the phase content and crystallinity of the films.An increase in the crystallinity was seen when increasing the nitrogen content.This structural evolution is shown in Fig. 1 and is demarcated by three different regions A, B and C.
Region A corresponds to the purely metallic film.In the diffractogram for the metallic sample, there are two distinct peaks accompanied by broader features.These most likely correspond to one (or several) intermetallic phase(s), although an exact identification is challenging due to the lack of sharp peaks and the vast number of possible phases due to the many elements included in the film.What is clear is that the purely metallic sample had low crystallinity and a high degree of amorphousness, indicating a nanocomposite structure.
Region B shows the diffractograms for the films with 13, 23, 30 and 37 at.%N which show a nanocomposite mixture of weak peaks and broad features, but with increasing crystallinity.The observed peaks for these films match a cubic structure with an fcc lattice.Peak shifts toward higher 2Ɵ angles, as indicated by the tilted dashed lines in Fig. 2, occurred due to a decrease in the lattice parameter from 4.85 to 4.69 Å.These lattice parameters are relatively large and are closest to the lattice parameters of YN (a = 4.89 Å) and ZrN (a = 4.58 Å) [39,40].The crystalline phase could thus be related to a solid solution mostly consisting of the elements with larger atomic radii i.e.Y and Zr.
Region C shows the diffractogram for the film with 49 at.% N. It shows distinct peaks matching a NaCl-type structure.The relatively weak intensity of the observed peaks is due to the strong <100> texture in this film.This was determined from theta-2theta XRD scans made on the same sample, shown in the supplementary information and it shows that the (200) peak intensity of the sample is very strong.The cell parameter was estimated to be 4.41 ± 0.0005 Å.This value is relatively close to the theoretical value of 4.38 Å, calculated using the rule of mixture and the elemental composition and cell parameters of the relevant binary nitrides.As this indicates that Vegard's rule was followed, a solid solution of all constituent elements was most likely formed [41].The small discrepancy between the experimental and theoretical value could be due to residual stresses in the film [24,25].
In conclusion, a nanocomposite structure with increasing crystallinity was maintained up to 37 at.%N while the nitride film containing 49 at.%N was a single phase fcc solid solution.The latter crystalline phase can be attributed to a NaCl-type AlCrNbYZrN solid solution.

Scanning electron microscopy
SEM images of the fractured cross-sections and the top surfaces of the films are shown in Figs. 2 and 3.The cross-section of the pure metallic film (0 at.% N) reveals a loosely packed and granular microstructure, while the films with 13, 23 and 30 at.% N all show an increasingly dense, fine-grained and near featureless microstructure.The 37 at.%N film also appears to be densely packed but with the faint traces of column-like morphology whereas the nitride sample with 49 at.%N has a distinct columnar morphology which ran all the way from the substrate to the interface.
The top views of the films also appear to follow a similar trend as a function of the nitrogen content.The increase in nitrogen gave rise to fine grained films up to a content of 30 at.% N while distinct grains are seen for the 37 at.%N film, which appears densely packed.The 49 at.%N film also exhibits distinct grains but the surface appears less densely packed than that of the 37 at.%N sample.The increase in the nitrogen content hence appeared to guide a clear evolution in the microstructure.An increase in the nitrogen content thus gave denser, more fine-grained films until a columnar structure was attained close to the 1:1 stoichiometry.An increased compaction of the microstructure should give coatings which better protect the substrate from their environment [29,42].A more compact coating is also expected to have a higher corrosion resistance due to the higher likelihood of the formation of a well-functioning passive layer.
Together, the diffractograms and SEM images indicate that three different types of structures were obtained when increasing the nitrogen content from 0 to 49 at.%N.This is interesting as this evolution in the microstructure should be reflected in the corrosion resistances of the films.Porous, loosely packed films might allow the electrolyte to permeate the pores, thus facilitating the process of metal dissolution through the entirety of the film [43].The 13 at.%N sample appeared denser, but did still not have an impervious structure.It is therefore likely that a strongly corrosive acid and/or a high potential could give rise to corrosion at weak points of the grains on the interface.The 23, 30 and 37 at.%N samples are expected to have higher corrosion resistances due to their smooth and compact microstructures.While the 49 at.%N sample could be expected to have good chemical inertness due to the presence of metal nitrides [14], it appears less dense than the other coatings and its columnar structure may provide a pathway for the electrolyte (akin to pores) to the substrate.

Corrosion resistances of the films -Short time domain studies 3.2.1. Potentiodynamic polarization
During potentiodynamic polarization measurements, the current densities of the test material is recorded as a function of the potential in a pre-determined potential window using a given electrolyte over a  relative short time period [44,45].The scanning of the potential in the positive (i.e.anodic) direction provides an external driving force for the corrosion reactions making it an accelerated corrosion test.As the experiment is not carried out under equilibrium conditions, a (time-dependent) change in the electroactive electrode area as a result of a change in the porosity of the films can also be expected to influence the results.In addition to the latter kinetic effect, thermodynamic effects based on the formation of metal nitrides with higher oxidation potentials than the corresponding metals would also be expected to be present.
The potentiodynamic polarization curves obtained for the AlCrN-bYZrN films in 1.0 M HCl are shown in Fig. 4. Missing from the electrochemical analyses is the 0 at.%N sample which dissolved on immediate exposure to 1.0 M HCl, i.e. before the polarization curve could be recorded, most likely due to its porous structure.The reactions that were expected to occur during the recording of the polarization curves for the AlCrNbYZrN films are shown in Table 2. Based on the thermodynamic data presented in the Supplementary Information, it is clear that all of the metal and metal nitride films should form an oxide surface layer upon their exposure to air, which is also confirmed by the  XPS studies in the forthcoming section.While the metal nitrides are more stable towards oxidation than the metals, the metal nitrides should still undergo oxidation (to form oxides) at potentials more positive than about -0.4 V vs. Ag/AgCl.The surfaces of all of the metal nitrides of interest here would thus have had a layer of a metal oxide protecting the metal nitride underneath.This native oxide has been left intact (i.e not removed prior to the polarisation) because this condition most closely resembles the intended application where the native oxide would be found on the coating in a harsh environment.
A comprehensive evaluation of the corrosion resistances of the films exhibiting active-passive behaviors can be attained based on the amalgamated assessment of: i) the corrosion current densities and corrosion potentials acquired from Tafel fittings, ii) the gradients and relative magnitudes of the current densities in the passive region, iii) the currentpotential relationships in the transpassive region and iv) the overall shapes of the polarization curves within the entire potential range.These points will therefore be discussed one by one below.

The corrosion current density and corrosion potential.
Two of the hallmarks of a corrosion resistant coating are a low corrosion current density (j corr ) and a high corrosion potential (E corr ) [46,47].These values, which are typically determined from the intersection of fits of the anodic and cathodic branches of the log (current density) vs. potential plot to the Tafel equation [37], are tabulated in the lower right-hand corner of Fig. 3.
The 13 at.%N sample had the highest corrosion current density (i.e.2.1 × 10 − 4 A/cm 2 ), indicating the highest rate of corrosion.This could be explained by its loosely packed granular microstructure which allowed the electrolyte to permeate into the film, resulting in a large electroactive surface area and a high corrosion rate.As a result, the corrosion potential, i.e.− 0.39 V vs. Ag/AgCl, was the lowest of all the films.Nevertheless, the 13 at.%N film presented a significant improvement in corrosion resistance when compared to the 0 at.%N sample as the latter immediately dissolved in a drop of 1.0 M HCl.One reason for the improved corrosion resistance for the 13 at.%N over the 0 at.%N sample was the more compact microstructure obtained with 13 at.%N. The formation of nitrides in the 13 % sample is another factor which may have improved the corrosion resistance of the film.However, as the 13 at.%N film was mainly metallic, the influence of this nitride effect should have been small.Increasing the nitrogen content to 23 at.%N resulted in a decrease in j corr by more than an order of magnitude.This was likely due to the grain refining effect of nitrogen which gives impervious, and denser microstructures [8,24,27], as evidenced by the SEM images of the cross-sections of the films.Further increases in the N content to 30, 37 and 49 at.%decreased j corr further and also shifted E corr to more positive values.As the 49 at.%N sample had the lowest j corr and highest E corr value, despite its nanocolumnar morphology, it is clear that the j corr and E corr values must have depended both on the microstructure and the amounts of metal nitrides in the films.The metal nitrides likely provided some degree of chemical inertness which lowered the corrosion current density and gave a forward shift in corrosion potential.Had this film been denser, its corrosion current density would have been even lower.The latter means that the results were influenced by both kinetic and thermodynamic factors.As will be discussed below, this is important to keep in mind when interpreting the polarization curves.

The electrochemical behavior in the passive region.
The passive region, which is the potential range over which the films (should) maintain a relatively constant current density, is situated between the corrosion potential and the onset of the transpassive region [48,49].It is indicated by a horizontal arrow in Fig. 4.
In general, the films show three types of passive behaviors.The 13 at.% N sample maintained a high and relatively constant current density (indicating a high and approximately constant metal dissolution rate) up to about 0.4 V vs. Ag/AgCl.The 23, 30 and 37 at.%N films all exhibited stable passive current densities (~10 − 5 A/cm 2 ) from their respective corrosion potentials up to the transpassive region in which a large transpassive peak was seen.For the 30 and 37 at.%N films, the passive region was up to ~0.40 V wider than that seen for the 23 at.%N film.A long passive region with an unchanged current density is a significant indicator of improved corrosion resistance as it shows that the film remains relatively inactive even at increasingly oxidizing potentials.The present results thus demonstrate that an increased nitrogen content in the films shifted the transpassive region to higher potentials for nitrogen contents up to 37 at.%N.
In contrast, the 49 at.%N sample, mainly containing metal nitrides, did not display a stable passive current density as the current density increased until a peak at 1.04 V vs. Ag/AgCl was seen.This indicates that the film became increasingly active during the potential scan, most likely due to a continuously increasing electroactive area resulting from the corrosion of the columnar film.The latter prevented the film from acquiring a passivity similar to that seen for the 37 at.%N film.As the presence of metal nitrides in the 49 at.%N sample did not result in an improved corrosion resistance it is hypothesised that the shift in the onset of the transpassive region mainly was due to kinetic effects associated with the inability to form a proper passive layer.It is conjectured that the inability to passivate properly arose as a result of the porosity of the 49 at.% film, however it is not possible to directly detect the issue of porosity from a polarisation measurement.This however shows that the presence of a single phase fcc solid solution was hence no guarantee for the attainment of a high corrosion resistance.
The Pourbaix diagrams for Al, Cr, Y, and Zr suggest that these elements should be present as solvated ions i.e.Al 3+ , Cr 3+ , Y 3+ and Zr 4+ in the passive potential region at pH 0 [50].The chemical reactions for the formation of the metal ions from the oxide or nitride are presented in Table 2.While the Pourbaix diagrams predict that a film containing Al, Cr, Y and Zr should undergo rapid corrosion, Nb 2 O 5 should be stable at pH 0 at potentials above -0.4V vs. Ag/AgCl [50].It can therefore be assumed that the observed passivity of the 23, 30 and 37 at.%N films stemmed from the presence of Nb 2 O 5 on the surface of the samples.If so, the Nb 2 O 5 would have been enriched on the surfaces of the samples thus suppressing the corrosion of the other underlying metals.Here it should, however, be recalled that oxide films should have been present on the surfaces of the films prior to the recording of the polarization curve.As the metal dissolution rates would depend on the solubility of the different metal oxides it is then possible that the dissolution of e.g.Cr 2 O 3 (see below) was incomplete due to its low solubility.Another reason could be that the local pH at the electrode surface should have been higher than in the bulk of the electrolyte due to the presence of hydrogen evolution during the initial part of the anodic scan.It is therefore likely that the passive layer was composed of a mixture of Nb 2 O 5 and Cr 2 O 3 .For nitrogen contents up to 37 at.%, the formation of a passive layer was facilitated by the films becoming denser during deposition when the nitrogen content was increased.
As mentioned above, a passive behavior was, however, not seen for the 49 at.%N film.It is suspected that the nanoporosity of the sample may have impeded the formation of a passive layer when the passive film depends on the enrichment of one or two out of several metals.

The electrochemical behavior in the transpassive region.
As shown in Fig. 4, a transpassive region with rapidly increasing current densities could be seen for all films except the 13 at.%N film.The latter only showed a relatively small oxidation peak at about 0.5 V vs. Ag/AgCl, most likely as a consequence of the already high current densities seen at the lower potentials.For all the films, except the 49 at.%N film, the transpassive behavior resulted in a complete loss of the film.The transpassive current density was then determined by the corrosion of the Al 2 O 3 substrate.While the presence of a transpassive region should be expected based on the oxidation of Cr 2 O 3 yielding soluble Cr(VI) species (e.g.Cr 2 O 3 + 4 H 2 O = Cr 2 O 7 2− + 8 H + + 6 e), it is not immediately clear why the onset of the transpassive region depended on the nitrogen content in the films.For the 23, 30 and 37 at.%N films, a transpassive peak was hence seen at about 0.8, 1.1 and 1.15 V vs. Ag/AgCl, respectively, while a very broad peak on a sloping background was seen for the 49 at.%N film.An increased content of metal nitrides in the films cannot explain this shift in the onset of the transpassive region as both the exposed nitrides and the metals would be oxidized at these high potentials as shown in the SI.This means that the effect must be due to mass transport rather than thermodynamic effects, most likely as a result of the influence of the porosity on the current density discussed above.The onset of the transpassive region can then be explained by the rate at which the passive layer was broken down as a result of the oxidation of Cr 2 O 3 to Cr(VI).This rate should clearly be higher for a porous film than for a nonporous film because of the higher electroactive area.Since the potential axis of a polarization curve is also a time axis, the onset of the transpassive region would then be shifted to higher potentials (i.e.longer times) as the films become denser.In Fig. 4, a decrease in the current density is seen at potentials beyond the transpassive peak potential for all films except the 49 at.%N film.As mentioned above, this can be explained by the loss of the films resulting in corrosion of the Al 2 O 3 substrate.The different behavior seen for the 49 at.%N film can be ascribed to the simultaneous corrosion of the film and the Al 2 O 3 substrate exposed as a result of the nanocolumnar-based corrosion (see below).It should also be noted that the loss of the passive oxide layer would result in an oxidation of the exposed metals and metal nitrides as described in Table 2.The fact that the oxidation of the nitrides would give rise to nitrogen evolution should hence also increase the corrosion rate of the film in the transpassive region.
A. Srinath et al. 3.2.1.4.The overall forms of the polarization curves.A material that is corrosion resistant must display a combination of the following characteristics: a low corrosion current density, a high corrosion potential, a stable and low passive current density, and a high transpassive potential.In the short time domain, these requirements were best fulfilled by the 37 at.%N film which was a densely packed, very fine-grained nanocomposite film with a high nitride content.The only nanocrystalline film was the 49 at.%film, which, however, showed unstable behavior at high potentials despite its promising j corr and E corr values.These results clearly show that the overall electrochemical performance of the sample should be considered, not only the j corr and E corr values.A high corrosion resistance is unlikely to be obtained if the microstructure of a material is not uniform and dense.Although the widely accepted guideline is that multicomponent corrosion resistant materials should be single phase (to avoid selective corrosion problems associated with the corrosion of a specific phase), it is clearly also very important that the material has a compact microstructure which facilitates the formation of a passive layer.
The present data shows that increasing the nitrogen content in AlCrNbYZrN films (N = 13-49 at.%) has several apparent effects.First, and most importantly, it affects the microstructure of the films, with amounts up to 37 at.%N giving increasingly fine-grained films.This decreases the porosity of the films which decreases the corrosion current densities.Secondly, the formation of metal-nitride bonds shifts the corrosion potential to more positive values.Thirdly, a high nitrogen content (up to 37 at.%N) increases the width of the passive regions and delays transpassive dissolution by up to ~0.40 V which increase the corrosion resistance of the films.The latter effect was, however, also a result of the microstructure changes mentioned above.

Inductively coupled plasma mass spectrometry
The concentrations of the metal ions in the 1.0 M HCl electrolyte were measured after two hours of film exposure under open circuit conditions.As a native layer containing a mixture of metal oxides should have been present on the surfaces of the samples prior to their immersion in the 1.0 M HCl electrolyte, the results of this experiment were not only affected by the thermodynamic stabilities of the oxides [50] but also by the dissolution rates of the less stable oxides.The data revealed which elements in the film initiated the corrosion of the film, thus increasing the understanding of how the corrosion of the AlCrNbYZrN coatings takes place.
The concentrations of the ions found in the electrolyte are given in ppb alongside the OCP values at the two-hour mark in Table 3.All concentrations were above the limit of detection, calculated from the corresponding calibration curve using three times the error of signal divided by the slope.The concentrations were also above the limit of quantification (calculated as ten times the error of signal divided by the slope), except for Nb in some samples.
As already mentioned, the OCP values were given by the mixed potentials determined by the rates of the redox processes occurring at the metal-electrolyte interface.Note that the OCP values became increasingly anodic when increasing the nitrogen content in the films in excellent agreement with the results discussed above.
Table 3 shows that the concentrations found in the electrolyte varied from low ppb levels (for Nb) to ppm levels (for Al).The low Cr and Nb concentrations in the samples indicate that these elements were passive whereas Al and Zr were active (i.e. that these latter underwent dissolution).It should be noted that predictions regarding the relative activities of the metals based on the Pourbaix diagrams [50] are difficult since the latter do not consider the formation of chloride complexes.However, the results clearly show that the Al underwent rapid dissolution in 1.0 M HCl independent of the nitrogen concentrations in the films.This correlates well with the expected behavior based on the Pourbaix diagram for Al.The results also show that aluminum was the most active of all the elements.
To facilitate comparison of the results for the different metals, the concentrations are also presented as relative normalized concentrations in Table 4 after having normalized the concentrations with respect to the compositions of the samples (see Table 1).This showed that the corrosion mainly involved Al as more than 90 % of the ions in the solution were Al 3+ ions.A high corrosion rate for Al was, however, not unexpected as Al 2 O 3 should undergo rapid dissolution at pH 0, especially in 1.0 M HCl where Al 3+ and Cl − readily form soluble complexes.After Al, the results indicate that Zr was the most electroactive element.Although the Pourbaix diagram for Zr suggests that as Zr 4+ should be formed in a pH 0 solution not containing chloride, it is not clear what the speciation would be in 1.0 M HCl.Based on the results it, nevertheless, appears as if the ZrO 2 dissolution rate was significantly lower than that for Al 2 O 3 during the two hours at the OCP.The low normalized concentrations for Nb and Cr indicate that these two elements were responsible for the passivity of the films via the presence of Nb 2 O 5 and Cr 2 O 3 , respectively.While Nb 2 O 5 should be passive at the employed potential and pH used, Cr 2 O 3 would in fact be expected to undergo dissolution.Although these results explain why the relative corrosion rate for Nb was the lowest for all the elements, the results also show that the dissolution rates must be considered for the other elements.The low Cr concentrations found in the electrolyte can therefore be ascribed to a low Cr 2 O 3 dissolution rate.The low concentrations seen for Y, which should be present as Y 3+ at pH 0, also indicate that the dissolution rate of Y 2 O 3 was relatively low.It can thus be concluded that the ICP-MS data indicate the presence of a passive layer mainly composed of Nb 2 O 5 and Cr 2 O 3 in good agreement with the results of the polarization curve experiments.

Open circuit potential measurements
In these experiments, the AlCrNbYZrN (13-49 at.%N) thin films were first exposed to 1.0 M HCl for 72 h under OCP conditions with the OCP and the impedances of the films being measured after 30 min and 72 h.The OCP values are presented in Table 5 while the Nyquist plots obtained after 30 min and 72 h are presented in Fig. 5.
After 30 min, more positive OCP values were seen for an increasing nitrogen content in the films, from -0.273 V vs. Ag/AgCl for the 13 at.%N sample to 0.318 V vs. Ag/AgCl for the 49 at.%N sample.This indicates that the film with 49 at.%N had the highest oxidation potential, in good agreement with the results in Fig. 4. The fact that the OCP values were more positive than the E corr values can be explained by the different experimental conditions (e.g.time-domains) of the measurements.While the OCP values obtained after 30 min corresponded to The numbers within the parentheses denote the random concentration errors [51].* Concentration below the limit of quantification.steady state mixed potentials, the E corr values indicated the potentials at which the anodic and cathodic currents driven by the applied potential were equal.Since a more positive potential should give a larger oxidation current during the scan, the E corr values should then be more negative than the OCP values obtained after 30 min.More negative E corr values should also be expected when increasing the scan rate.The potential at which the scan was initiated is also important.If the native oxide layer is not fully reduced prior to the start of the scan (as in the present case), a lower oxidative current would be expected which should give rise to a more positive E corr value compared to if the oxide were reduced at the start of the scan.Care should therefore be taken when comparing E corr values obtained under different experimental conditions.
After 72 h including the impedance measurements after 30 min, the variations in the OCPs were, however, large and there was also no clear trend with respect to the nitrogen contents in the films.While the values for the 13, 23 and 30 at.% N films were 0.16, 0.57 and 0.41 V more positive than the corresponding 30-minute values, the OCP values of the 37 and 49 at.%N samples were, on the other hand, 0.31 and 0.09 V more negative.This implies that the OCP values after 72 h were affected by the impedance measurements carried out after 30 min and/or the 72 -h period at the OCP.Considering the relative short duration of the impedance experiment (i.e.2.5 min), the latter seems more likely.The OCP values obtained after 72 h consequently indicate that the 23 and 30 at.% N films exhibited the highest corrosion resistances after 72 h.This finding, which demonstrates that the corrosion resistance of the films depended differently on the OCP time, is further supported by the impedance results presented in the next section.

Electrochemical impedance spectroscopy
As indicated above, the impedances of the AlCrNbYZrN (13-49 at.%N) thin films were measured after 30 min and 72 h at the OCP in 1.0 M HCl.All the measurements were conducted at the OCP using a sinusoidal ac perturbation of 10 mV and a frequency range of 200 kHz to 0.1 Hz.
The Nyquist plots of the films after 30 min and 72 h are shown in Fig. 5.At 30 min, the samples with higher nitrogen contents featured steeper gradients and tended more closely to the imaginary axis than the films with lower nitrogen contents.The film with 37 at.%N was the tallest and nearly straight indicating that it had the most capacitancelike behavior.While a perfect corrosion resistant coating should deliver a vertical line in a Nyquist plot, most metal-based electrodes would deviate from an ideal capacitive behavior in a corrosive environment due to the occurrence of corrosion reactions facilitated by the presence of compositional heterogeneity and porosity.After 30 min, all the films, barring the 49 at.%N sample which had a low impedance, followed a pattern with a higher nitrogen content yielding a higher impedance.
After 72 h, lower impedances were seen compared to after 30 min, except for the 23 at.%N film.The latter in fact had an impedance that was higher after 72 h than after 30 min.This is in good agreement with the increase in the OCP value seen after 72 h for the 23 at.%N.After 72 h, the 23 at.%N film hence had the highest impedance followed by the 37, 30 and the 13 at.%N samples.In addition, a temporary increase in the impedance was seen for the 13 at.%N film after 24 h (see Fig. S1 in the Supplementary Information).However, the impedance measurement at 72 h showed that the 23 at.% N coating exhibited the largest capacitive loop, followed by the 37 at.% N coating.In contrast, the open circuit potential values at 72 h shows that 30 at. % N coating had the highest OCP, followed by the 23 at.% N coating.The reason for this inconsistency is unknown.The supplementary information shows the equivalent circuit fits and the R ct values at 30 min and 72 h.Several factors affect the size of the capacitive loop and the magnitude of R ct , and the OCP is one of them.Among others are electroactive area and corrosion current.The results thus suggest a complex dependence of the impedance on the OCP corrosion time and nitrogen content in the films with different films performing best after different times.
Based on the OCP results after 30 min, the 49 at.%N film would have been expected to have the highest impedance after 30 min.This was,  A. Srinath et al.
however, not the case as the impedances of the 49 at.%N film were low both after 30 min and 72 h.The latter values are, on the other hand, in good agreement with the conclusions based on the polarization curve experiments, indicating that it was difficult to passivate the nanocolumnar 49 at.%N film.This clearly shows that corrosion resistance predictions based on OCP values alone can be misleading.It is likewise evident that an increased nitrogen concentration (yielding nitrides) does not necessarily yield an improved corrosion resistance.The results in fact demonstrate that the corrosion resistances were mainly determined by the (nitrogen-dependent) microstructures of the films.

SEMs of pitted films
Fig. 6 shows SEMs of the AlCrNbYZrN films after 72 h at the OCP in 1.0 M HCl.The images show the different ways in which the corrosion (e.g.pitting corrosion) manifested itself.Pitting corrosion has three main stages: localized pit initiation, pit growth, and extensive pitting leading to general wastage/film dissolution.The SEMs showed that the pitting corrosion became less pronounced with an increasing nitrogen content in the film.
The 13 at.%N sample exhibited all the signs of severe corrosion.The film was etched, in some places the film had delaminated after pitting and large pits with redeposited corrosion product and pit covers (dried salt) were also seen.The 23 at.%N film also displayed pits and parts of the film had dissolved whereas the 30 at.% N sample exhibited small pits that appeared to be in the process of merging with one another.The 37 at.%N sample featured both large and small pits where some of the pits had repassivated while a few had traversed down to the substrate surface.There was almost no etching, hardly any film dissolution and the corrosion was localized.The 49 at.%N sample, on the other hand, showed no evidence of any pitting after 72 h in 1.0 M HCl although the grain structure appeared more nanoporous but wholly intact.Only mild surface corrosion was seen in the form of localized etching.The corrosion could thus proceed without the formation of pits which supports the hypothesis that the low impedance of the 49 at.%N sample was due to its nanocolumnar structure.

X-ray photoelectron spectroscopy
To study the composition of the surfaces of two samples with different corrosion resistances, XPS analyses were performed on the 37 and 49 at.%N samples before and after their exposure to 1.0 M HCl for 72 h.These measurements served as a comparison to the OCP and impedance measurements as the corrosion time scale was the same.The XPS measurements should not, however, be compared to the polarization curves which demonstrate corrosion on a shorter time scale.XPS was consequently used to ascertain if the different impedance behaviors could be linked to differences in the surface layer compositions.
Table 6 shows the relative surface abundances of the metallic elements before and after exposure to 1.0 M HCl for 72 h.We note that compared to the bulk composition shown in Table 1, these numbers are quite different.This is because the nitrogen contents of 37 or 49 at.%N were omitted in Table 6 and as slightly different sensitivity factors were used in Tables 1 and 6 (see the Experimental section).For these reasons the numbers in Tables 1 and 6 should not be directly compared.However, the trends in the data can still be compared.The bulk composition in Table 1 shows that for the 37 at.%N sample, the relative abundance of the metallic elements decreased in the order Nb => Zr => Cr => Y => Al.At the surface of the pristine sample this order was instead Zr => Cr => Nb => Y => Al and after 72 h in 1.0 M HCl, the relative abundance decreased in the order of Nb => Cr => Zr => Y => Al as seen in Table 6.Although, the difference in surface abundance before and after corrosion was relatively small for the 37 at.%N sample, the results thus indicate the formation of a passive film enriched in Nb and Cr during the OCP.For the nanocolumnar 49 at.%N sample, the bulk composition shows a decreasing abundance in the order of Nb => Zr => Cr => Al => Y (Table 1), whereas for the pristine surface this order was Nb => Cr => Al => Zr => Y and after 72 h in 1.0 M HCl the order was Cr => Nb => Zr => Al => Y (Table 6).For the 49 at.%N sample, there was hence a notable increase in the Cr abundance after the corrosion.This clearly shows that the elemental composition at the surface of the pristine samples was different to the bulk composition and also that the surface composition changed during the exposure to 1.0 M HCl especially for the 49 at.%N sample.For both samples, the corrosion resistances were linked to the formation of stable Cr and Nb oxides.For the 49 at.%N sample, the relative abundance of especially Cr also increased during the corrosion so that Cr was the most abundant element on the surface after 72 h in 1.0 M HCl.A similar enrichment of Cr was, however, not seen for the 37 at.%N sample.This difference in the behavior is also in agreement with the changes seen in the OCP values after 72 h in Table 5.For both films, a more negative OCP value was seen after 72 h, but this effect was largest for the 37 at.%sample indicating a decrease in the corrosion resistance.
The degree of oxidation of the different elements is shown in Table 7 as the ratio between the oxidized components and the metal and/or nitride components (see the Supplementary Information).The numbers resulted from peak fittings of the metal core level spectra, which can be found Fig. S2 in the Supplementary Information.In general, the ratios were larger after 72 h in 1.0 M HCl suggesting the presence of a thicker oxide layer.After the corrosion, the 37 at.%N sample was also more oxidized than the 49 at.%N sample (this was found to be the case for all metallic elements).Based on the OCP data the 49 at.%N sample appeared to behave better than the 37 at.%N sample after 72 h in 1.0 M HCl.The impedance results, on the other hand, clearly show that the corrosion resistance of the 37 at.%N sample was higher than for the 49 at.%N sample.The latter is in good agreement with the formation of a thicker oxide containing Nb and Cr on the surface of the 37 at.%N sample, and the more porous structure of the 49 at.%N sample.

Conclusions
The effect of the N content on the electrochemical performance of non-equimolar AlCrNbYZrN films has been investigated in 1.0 M HCl.The addition of nitrogen to the films was found to alter the corrosion resistances of the films mainly via changes in the microstructures of the films.Denser nanocomposite films with better corrosion resistances were seen for nitrogen contents from 13 to 37 at.%N.For 49 at.%N, a nanocrystalline nitride film with underdense columnar microstructure and a significantly lower corrosion resistance was, however seen.The results therefore indicate that the corrosion resistances of the films mainly depended on the porosities of the films.The formation of a single-phase nanocrystalline nitride film did thus not give rise to an increased corrosion resistance as could be expected based on thermodynamic considerations.It can hence be concluded that a nanocrystalline material would need to have a densely packed microstructure to be highly corrosion resistant.A nanocomposite multicomponent material with a uniformly dense microstructure hence outperforms a single-phase multicomponent material with porous structure.
The results also indicate that the corrosion resistance depended on which method (e.g.OCP, polarization curve or impedance spectroscopy) that was used as well as the time domain of the measurements.The electrochemical results combined with the ICP-MS and XPS data indicate that the passivation of the films was due to the formation of an oxide layer enriched in Nb and Cr.In the transpassive region, the films were lost as a result of the formation of soluble Cr(VI) species and the oxidation of the exposed nitrides yielding nitrogen evolution.As Al was found to be the most active element it should be possible to increase the corrosion resistance of the films significantly by replacing Al with a more passive element.The design of multicomponent nitrides with extreme corrosion resistance hence requires the ability to manufacture highly dense coatings which facilitate the formation of a stable passive layer.For the present films it was found that at least two passive elements were needed, i.e.Nb and Cr.By replacing Cr with another more passive element it should also be possible to obtain films which are

Fig. 1 .
Fig. 1.Grazing incidence diffractograms for all AlCrNbYZrN films.The star (*) markers correspond to a crystalline phase with an fcc lattice whereas the square (◼) markers correspond to a (AlCrNbYZr)N solid solution phase with NaCl-type structure.The dashed lines are merely intended as guides for the eye.

Fig. 2 .
Fig. 2. SEM fractured cross-sectional images of the AlCrNbYZrN films.The N content is denoted in right corner of each image.The scale is valid for all images.

Fig. 3 .
Fig. 3. SEM top-view images of the AlCrNbYZrN films showing the change in surface microstructure when increasing their nitrogen contents.The scale bar is for 100 nm and applies to all samples.

Fig. 4 .
Fig. 4. Potentiodynamic polarization curves recorded at a scan rate of 1 mV/s in 1.0 M HCl for AlCrNbYZrN films containing 13 to 49 at.%N, as well as the calculated j corr and E corr values.

Fig. 5 .
Fig. 5. Electrochemical impedance measurements of the AlCrNbYZrN films with N contents between 13 and 49 at.%.The impedances were measured at the OCP after 30 min and 72 h.The frequency range was 100 000 Hz to 0.1 Hz.

A
. Srinath et al.

A
. Srinath et al.

Table 1
Average bulk compositions for the AlCrNbYZrN samples in at.% for the different nitrogen flow rate ratios, R N (%).

Table 3
Concentrations in ppb of the elements found in the electrolyte after two hours at the indicated OCP values.

Table 5
OCP values (vs.Ag/AgCl) for the AlCrNbZrN films after 30 min and 72 h, respectively.