Effect of Cu on Nitriding of α-Fe

Abstract

Nitriding of Fe-1 wt.% Cu and Fe-5 wt.% Cu alloys at 813 K leads to the formation of predominantly the γ′-iron nitride phase (γ′-Fe₄N) when using nitriding conditions, which lead to pronounced formation of ε-iron nitride phase (ε-Fe₃N_1+x) upon nitriding of pure α-Fe. Energy dispersive X-ray analysis reveals that the developing γ′ can attain a Cu content corresponding to that of the base material. In contrast, tiny amounts of ε-nitride that eventually develop contain considerably less Cu. The microstructure implies that the formation of the ε-nitride requires long-range substitutional interdiffusion to achieve the Cu partitioning. These observations were interpreted in terms of a significantly higher solubility of Cu in the γ′ phase than in the ε phase, which is explainable by the phases’ crystal structures. The observations were rationalized in terms of schematic Fe–Cu–N phase diagrams valid for 813 K.

1. Introduction

Nitriding (and nitrocarburizing) of steel and (e.g., cast) iron is a widely applied thermochemical heat treatment procedure, by which N (and C) from an external source, typically being a gas or a plasma, are introduced into the surface-near region of a workpiece by diffusion [1]. If applied to non-austenitic base materials, nitriding is usually performed below the eutectoid temperature of the Fe-base material. The treatment makes use of the moderate solubility of N and high diffusivity of N in α-Fe, leading to diffusion depths of up to several 100 μm. Generally, one distinguishes between a compound layer at the surface and a diffusion layer underneath. The compound layer is typically composed of the ε and/or the γ′ phase, being hcp- and fcc-based interstitial phases, both showing long-range ordering of N on the octahedral sites. See Figure 1 for the crystal structure of the γ′ phase. The term “diffusion layer” refers to the substrate underneath the compound layer being enriched in N. In the case of pure Fe, N is dissolved in solid solution and may precipitate during cooling. It is also the diffusion layer where characteristic alloying element-base nitride precipitates can already develop during nitriding, which can increase hardness in the diffusion zone and improve fatigue properties by inducing compressive stress.

The fundamental basis for the understanding of the phase transformations occurring upon nitriding iron-base alloys is the thermodynamics of the Fe–N system. Phase diagrams Fe–N are shown in Figure 2a,b in two forms. Figure 2a depicts what is typically shown as a Fe–N phase diagram. It reflects the solid-state phase equilibria between iron nitrides with N contents up to the ε phase, in the present version ignoring the ζ phase, Fe₂N, and higher-N content iron nitrides [2,3] which were not included in the thermodynamic modelling, being the basis for that diagram [4].

Both ε and γ′ nitride are metastable with respect to N-containing α/γ and N₂ gas at 1 atm, implying that Figure 2a shows actually a metastable phase diagram. In practice, these nitrides can be prepared by metastable equilibration of iron at some elevated temperature with NH₃ + H₂ containing atmospheres via the reaction

$${\mathrm{NH}}_3 \rightleftarrows [\mathrm{N}]+\frac{3}{2}{\mathrm{H}}_2,$$

(1)

where [N] represents N incorporated into a solid phase. The chemical potential of N in a solid equilibrated with an NH₃ + H₂ containing atmospheres can be shown to be related to the atmosphere’s nitriding potential r_N [5]:

$$r_{\mathrm{N}}=\frac{p_{{\mathrm{NH}}_3}}{p_{{\mathrm{H}}_2}^{3/2}},$$

(2)

with the partial pressures p_i of the gas-phase species i. In more detail, in the case of a gas–solid equilibrium according to Equation (1), the activity a_N of N at the surface of the solid increases linearly with r_N and the chemical potential increases with lnr_N. Note that the equilibrium N₂ pressure corresponding to the NH₃ + H₂ mixtures can easily amount to several 1000 atm.

Figure 1. Unit cell of the crystal structure of γ′-Fe₄N [6,7] based on an fcc arrangement of the Fe atoms and ordered occupation of ¼ of the octahedral sites by N. That ordering leads to two crystallographically distinct Fe(0) and Fe(II) atoms, with the Fe(0) being substituted by Cu. Note that this structure is frequently referred to as antiperowskite [8].

Figure 1. Unit cell of the crystal structure of γ′-Fe₄N [6,7] based on an fcc arrangement of the Fe atoms and ordered occupation of ¼ of the octahedral sites by N. That ordering leads to two crystallographically distinct Fe(0) and Fe(II) atoms, with the Fe(0) being substituted by Cu. Note that this structure is frequently referred to as antiperowskite [8].

Accordingly, exposing the Fe to an atmosphere exhibiting a certain value of r_N at a given temperature, a specific solid phase with specific composition should be in equilibrium with that atmosphere (if decomposition processes leading to N₂ can be ignored; if that is not the case, a somewhat lower N content will be attained; “close to equilibrium”). The phase forming is classically indicated by the so-called Lehrer diagram (see Figure 2b), corresponding to a potential phase diagram, noting that the chemical potential of N is the variable conjugate to the amount of N in the system. As also seen in the present work, nitriding of massive iron typically leads to (close to) equilibrium at the surface, while the inwards diffusion occurs at a finite rate. Hence, the N content will decrease with increasing distance to the surface. If the value of r_N suffices to produce an ε-iron nitride layer at the surface of the specimen, the decrease of the N content will lead under local equilibrium to an intermediate γ′ layer between the ε phase at the surface and the α substrate (ε/γ′ double layer).

The current work deals with the thermodynamics of nitriding in the presence of minor amounts of an alloying element M (considering M to be a metal or metalloid element, but not carbon). Estimating the results of nitriding of binary Fe-M alloys typically involves comparison of the stability of the nitrides in the M–N system with those in the Fe–N system. Higher stability of the nitrides in the M–N system typically allows formation of such M nitrides at r_N values much lower than required for the formation γ′ or ε iron nitrides. This explains the formation of alloying element nitrides like VN, CrN or AlN in the diffusion zone [9], occurring while depleting the matrix by the alloying element (referred to as internal nitriding). N uptake beyond that expected from the precipitation of such stable, e.g., stoichiometric MN nitrides and N saturated Fe (as implied by bulk thermodynamics) was reported for nitrided Fe-M alloys. This is often ascribed to so-called excess nitrogen caused by the interface precipitate matrix and due to the strain field in the matrix caused by the precipitates [10,11].

Considerations to estimate the amount of excess nitrogen often appear to neglect a possible uptake of Fe by the M–N nitrides. Predicting such an uptake is obstructed by limited knowledge of the equilibrium phase diagrams Fe–M–N at the relevant temperatures and by the usually nanoscopic nature of the nitrides developing (obstructing, in any case, the validity of consideration purely based on the equilibrium phase diagrams/bulk thermodynamics). There is, however, clear evidence for such a Fe uptake by nitrides, e.g., by CrN [12,13] and by amorphous Si₃N₄ (actually (Si₃N₄)_1−x(Fe₃N₂)_x [14]). N uptake by such nitrides is sometimes predicted in the course of thermodynamic calculations by the CALPHAD method, e.g., Fe uptake by the VN phase [15], which, however, does not necessarily imply that the thermodynamic description of the phase is really based on clear experimental evidence for such particular behaviour.

Figure 2. Binary phase diagrams relevant for the present work with arrows marking the treatment temperature of 813 K. (a,b) Metastable phase diagrams Fe–N calculated from a thermodynamic database [4] excluding N₂ gas (redrawn): (a) Temperature vs. molar fraction of N x_N and (b) complementary potential phase diagram in the form of the Lehrer diagram T vs. nitriding potential r_N. The colored lines in (b) mark the equilibrium existence of the corresponding phases at 813 K have also been used in following schematics and phase diagrams, with the points and numbers indicating the nitriding potentials applied experimentally in atm^−1/2. (c) Phase diagram Fe–Cu [16].

Figure 2. Binary phase diagrams relevant for the present work with arrows marking the treatment temperature of 813 K. (a,b) Metastable phase diagrams Fe–N calculated from a thermodynamic database [4] excluding N₂ gas (redrawn): (a) Temperature vs. molar fraction of N x_N and (b) complementary potential phase diagram in the form of the Lehrer diagram T vs. nitriding potential r_N. The colored lines in (b) mark the equilibrium existence of the corresponding phases at 813 K have also been used in following schematics and phase diagrams, with the points and numbers indicating the nitriding potentials applied experimentally in atm^−1/2. (c) Phase diagram Fe–Cu [16].

Much less is known about the effect of elements M forming less-stable nitrides than Fe. As it concerns the series of 3d metals, stability of transition metal nitrides generally decreases from Ti towards Cu [17]. For Ti…Ni this has also been shown quantitatively on the basis of the Gibbs energies of dissolution of N into terminal transition metal-based solid solution [18]. In the whole row Ti…Cu, M = Cu should develop the least stable nitrides. The best-known binary Cu nitride is Cu₃N, which is, however, not accessible from elemental Cu upon interaction with NH₃ + H₂ containing atmospheres, but requires non-metallic precursors to be treated with NH₃ [19,20,21]. Cu₃N decomposes irreversibly above (620–720) K under kinetic control to Cu + N₂ [19,21]. This nitride is definitely not expected to develop upon nitriding Fe–Cu alloy using NH₃ + H₂ atmospheres typically used to nitride Fe-base alloys. Instead, only iron nitrides are expected to form, which may or may not be affected by the presence of Cu.

Below the eutectoid temperature, Cu has a very low solubility in iron, as visible from the binary Fe–Cu phase diagram [16] shown in Figure 2c. In steel, minor amounts of Cu may be used to improve corrosion properties and to improve strength [22,23]. However, it is usually regarded as an unwanted tramp element usually incorporated during recycling [24,25], and too high contents may lead to hot cracking [26,27,28]. As it concerns cast iron, Cu is occasionally deliberately added (alongside with other elements) to promote the formation of pearlite [29].

It was a study on nitriding of Cu-containing, white-solidified cast iron [30] which suggested that Cu promotes formation of the γ′ nitride against that of ε, which was attributed to different solubilities of these elements in the two iron-nitride phases. To confirm this idea, in the present work the effect of Cu in iron on the competition between the γ′ with ε iron nitrides is studied using binary Fe–Cu alloys to exclude complications due to Si and C, as presented in Ref. [30].

2. Materials and Methods

2.1. Alloy Preparation

Fe was supplied as cold-rolled plates (“α-Fe”; thickness 1 mm; 99.99%, Goodfellow) which were encapsulated in fused silica tubes under Ar. For recrystallization, the tubes were heated for 2 h at 973 K and slowly cooled.

Fe–Cu Alloys with nominal compositions of 1 wt.% Cu (“Fe-1Cu”) and 5 wt.% Cu (“Fe-5Cu”) were prepared by arc melting in an Arc Melter AM 200 (Edmund Bühler GmbH, Bodelshausen, Germany) from Fe granules (99.98%, Alfa Aesar) and Cu metal (99.9995%, VEB Spurenmetalle Freiberg, Germany) with batch sizes of about 10 g. Material loss upon preparation was very small, such that the nominal composition corresponding to molar fractions of x_Cu = 0.009 for Fe-1Cu and x_Cu = 0.044 were adopted to be valid. The alloy batches were hammered to a thickness of about 4 mm and cut into smaller, plate-like specimens. These were encapsulated as above and heat treated for 20 min at 1383 K. These heat treatments were terminated by quenching in water with a temperature of about 300 K, including crushing the fused silica tubes. This should ensure rapid occurrence of the γ → α transition in order to avoid long-range Cu partitioning.

Prior to nitriding, the α-Fe, Fe-1Cu and Fe-5Cu plate specimens were ground and polished (final stage 1 µm diamond paste, performed directly prior to nitriding), washed with water and ethanol and then dried. Nitriding in a NH₃ + H₂ gas mixture at 813 K was performed in a vertical fused-silica tube furnace (tube diameter of 28 mm) described in some more detail in Ref. [5]. The overall gas-flow rate was adjusted to 500 mL/min pertaining to room temperature. The gas was composed from NH₃ and H₂ to result in r_N = 1, 2.5 or 4 atm^−1/2 according to Equation (2). The nitriding treatment was terminated by quenching in water having a temperature of about 300 K, which was previously flushed with N₂.

2.2. Microstructure Analysis

X-ray diffraction (XRD) measurements were performed on the surface of the samples in a diffraction angle 2θ range of 30°–125° with a Bruker D8 ADVANCE (Bruker AXS, Karlsruhe Germany) diffractometer being equipped with a Co tube, a quartz Johannsson monochromator in the primary beam selecting the CoK_α1 component of the fluorescence radiation (wave length 1.78897 Å). The diffracted intensity was recorded using a Lynxeye position sensitive detector (Bruker AXS, Karlsruhe, Germany). The diffraction patterns were evaluated using the Bruker-AXS TOPAS software (Version 5, Bruker AXS, Karlsruhe, Germany) [31] based on Rietveld and Pawley fits, considering the crystal structures of α-Fe, γ′-Fe₄N and ε-Fe₃N_1+x (e.g., [2,7,32]).

Scanning electron microscopy (SEM) was performed on polished metallographic cross-sections. To prepare these, the nitrided specimens were coated electrochemically by a nickel layer to reduce edge effects upon polishing the cross sections. The coated specimens were cut, embedded, ground and polished (final stage colloidal silica). The SEM investigations were conducted using a JEOL JSM-7800 F (JEOL, Tokyo, Japan) equipped with EDAX Octane Elite EDS system (EDS: Energy dispersive electron spectroscopy) and EDAX Hikari Super EBSD system (EBSD: Electron backscatter diffraction), both Ametek, Weiterstadt, Germany. An atomic number sensitive backscatter electron (BSE) was used for imaging. EBSD was measured with a step-width of 250 nm and 50 nm for overview and detailed measurements, respectively. EBSD patterns were indexed based on the crystal structure data from [2,18,32].

3. Results

The compositions of the Fe-1Cu and Fe-5Cu alloys are located within the γ solid solution region at 1383 K [16], see Figure 2c. SEM-based EDS investigations of metallographic cross-sections of the as-quenched alloys did not indicate a macroscopically inhomogeneous distribution of Cu. While the microstructure of the Fe-1Cu was ferritic, that of Fe-5Cu showed a more complicated microstructure, suggesting a massive or martensitic transformation mechanism, in agreement with Ref. [33].

In reasonable agreement with [33,34], the alloys water-quenched from the austenite region showed (room-temperature) lattice parameters of 2.869 Å (Fe-1Cu) and 2.872 Å (Fe-5Cu), which are significantly increased as compared to the lattice parameter of pure α-Fe (2.866 Å). Heat treatments at 813 K, the temperature also selected for nitriding, revealed a decrease in the lattice parameters with increasing annealing time, asymptotically approaching the value of α-Fe. Comparison with, e.g., Ref. [33] implies the formation of finely distributed nanoscopic Cu-rich precipitates, which is also confirmed by more in-depth analysis of the nitrided specimens, see what follows.

Exemplary cross-sectional microstructures of the compound layers, formed upon nitriding of α-Fe, Fe-1Cu and Fe-5Cu alloys at 813 K and varying nitriding potentials, are depicted in Figure 3. EBSD analysis, thereby, allowed straightforward identification of the phases having developed as the compound layer. X-ray diffraction data, depicted in the Supplementary Material (Figures S1–S3), always show, apart from reflections of the α-substrate material, reflections due to nitride phases identified by SEM-based investigations.

Nitriding of pure α-Fe at r_N = 1 atm^−1/2 for 4 h leads to the formation of a pure γ′ layer (Figure 3a). Upon increasing the nitriding potential to 2.5 atm^−1/2 and keeping the treatment time at 4 h, an ε/γ′ double layer develops on pure α-Fe (Figure 3b). The microstructure is similar for r_N = 4 atm^−1/2, only leading to a thicker double layer (not shown).

Nitriding of the Fe–Cu alloys, however, appears to significantly suppress the development of the ε phase. As illustrated in Figure 3c,d, nitriding of Fe-1Cu and Fe-5Cu at r_N = 2.5 atm^−1/2 for 4 h leads to the formation of only little (Figure 3c) or no (Figure 3d) ε-iron nitride. Upon nitriding at r_N = 4 atm^−1/2 ε forms in both Fe-1Cu and Fe-5Cu, alongside some porosity (see Figure 3e). Nevertheless, the difference to the corresponding treatment of pure α-Fe remains considerable. Note that the small amounts of ε phase preferentially appear to occur close to the surface, where the microstructure is highly irregular, including the presence of pores which must have formed during nitriding. Apart from the ε-suppressing effect, the Cu in the substrate also promotes the formation of γ′-iron nitride in some depths of the substrate (see Figure 3e), which is not typical for pure α-Fe quenched after nitriding.

SEM imaging at a higher resolution reveals the presence of regularly distributed features appearing brighter than the α-Fe matrix in BSE contrast images. These bright features occur just beneath the compound layer (Figure 4a) but also in a larger depth of about 800 μm (Figure 4b), i.e., much below the diffusion depth of the N in the diffusion zone. In line with a large body of literature, these particles should correspond to Cu-rich precipitates developing in the α-Fe due to the thermal load independent of the nitriding [34,35,36]. Strikingly, these particles appear to be absent (apart from possible tiny remainders) in the γ′-iron nitride (compare Figure 4a).

The Cu content in the iron nitrides was analysed using EDS, see Figure 5. The Cu content in the γ′ phase appears to correspond to the u_Cu = x_Cu/(x_Fe + x_Cu) content in the substrate of u_Cu ≈ 0.044. In contrast, the Cu content is much lower in the ε particles. We refrain from stating a value due to the tiny character of the particles and mutual detection of Ni fluorescence radiation from the adjacent protective Ni layer.

4. Discussion

The outcome of the interaction of pure α-Fe with a NH₃ + H₂ atmosphere can be predicted by the potential phase diagram/Lehrer diagram depicted in Figure 2b. As explained in the introduction, the surface is (close to) equilibrated with the atmosphere, and due to the relatively slow inwards diffusion of the N, an N-content depth profile into the α-Fe develops. Accordingly (see Figure 3a,b), formation of γ′ phase for r_N = 1 atm^−1/2 and ε phase for r_N = 2.5 atm^−1/2 and 4 atm^−1/2 at the surface of the nitrided pure α-Fe is in good agreement with the prediction from Figure 2b. All this is in accordance with a large body of investigations on the nitriding behaviour of α-Fe at similar temperatures [1].

As expected from the low solubility of Cu in α-Fe at the nitriding temperature of 813 K [33,34] (see also Figure 2c), the Fe-1Cu and Fe-5Cu alloys develop towards two-phase assemblage under the thermal load, consisting in first approximation of pure α-Fe and pure Cu. On the basis of this, one may suppose that these two metals interact independently with the NH₃ + H₂ gas mixture. A “Lehrer diagram for Cu” is unknown, and the low thermodynamic stability of the Cu₃N nitride implies that a high, probably unattainable minimum nitriding potential r_N would be required to produce Cu₃N (e.g., by nitriding in close-to-pure NH₃). In fact, experimental evidence suggests rapid decomposition of once formed Cu₃N into Cu + N₂ at the presently applied temperature [19,20]. The minimum nitriding potential r_N to produce Cu₃N from pure Cu should be definitely higher than the minimum r_N required to produce the ε-iron nitride phase from α-Fe. Accordingly, nitriding of a two-phase mixture α-Fe and Cu should ideally lead to formation of those iron nitrides which are also observed upon nitriding pure iron. These iron nitrides should then coexist with pure Cu. That situation has been illustrated at metal fraction values of u_Cu = x_Cu/(x_Fe + x_Cu) = 0 and 1 in Figure 6a. This diagram will, however, be “filled” at 0 < u_Cu < 1 after the following considerations.

The experimental observations made in Section 3 deviate from what is proposed above as ideal nitriding behavior of a two-phase assemblage α-Fe + Cu. Instead, the Cu seems to strongly affect the outcome of the nitriding by promoting the formation of the γ′ phase at the cost of the ε phase. That can be rationalized by assuming solubility of Cu in γ′ but not in the ε phase. This is supported by the following observations:

• The developing γ′ appears to dissolve the Cu precipitates present in the α-Fe matrix (see Figure 4a), corresponding to a reaction α + Cu → γ′, with the latter γ′ having the u_Cu fraction values pertaining to the Fe-1Cu and Fe-5Cu substrates (u_Cu = 0.009 and 0.044, see also Figure 5).
• The eventually developing ε phase contains significantly less Cu than the γ′ phase, see Figure 5.

Incorporation of Cu into γ′-Fe₄N requires Fe/Cu interdiffusion on a length scale of a few tens of nanometers upon dissolving the particles into the γ′ phase. That dissolution is compatible with experimental and theoretical evidence in the literature. γ′-Fe_4(1−uCu)Cu_4uCuN has been prepared by ball-milling and subsequent annealing [37]. In that work, preferential substitution of the Fe atoms on the Fe(0) site by Cu atoms has been concluded from Mössbauer spectroscopy, as illustrated in Figure 1a. Thereby, the Cu having a lower affinity towards N than Fe (see the low stability of Cu₃N) avoids close contact with the N atoms in the crystal structure (see Figure 1a). In that work, Cu uptake up to Fe_3.4Cu_0.6N corresponding to u_Cu = 0.15 (or substitution of 60 % of the Fe(0) sites) was found, with higher Cu contents leading to the formation of residual elemental Cu in the powder mixture. Further works on powder and thin film material [38,39] confirm the picture obtained in Ref. [37]. Some works have claimed the preparation of stoichiometric γ′-Fe₃CuN (u_Cu = 0.25) [40], where proof may be missing that indeed a higher Cu content than in [37,38,39] had been attained in the γ′ phase. High-throughput computational screening of magnetic antiperovskite compounds with various elements [41] revealed that the energy of γ′-Fe₃CuN is only 36 meV/atom above the convex hull of the Fe–Cu–N system (static structures at 0 K) devised in that work, such that γ′-Fe₃CuN might be an accessible material. Note that γ′-Fe₄N itself is located (10–20) meV/atom of higher energy than α-Fe + Fe₃N, whereby γ′-Fe₄N gets vibrationally stabilised at T > 0 K [42,43]. Nevertheless, in the considerations to follow, it will be assumed that the above-mentioned Fe_3.4Cu_0.6N corresponding to u_Cu = 0.15 constitute the maximum attainable Cu content on the γ′ phase.

Incorporation of Cu into the ε-iron nitride phase has not yet been reported or theoretically considered. The ideal crystal structure of ε-Fe₃N contains only one sort of Fe atom, and this sort has two N atoms in its first coordination sphere [44], similar to the Fe(II) atoms in γ′-Fe₄N. As the latter sites are apparently not occupied by Cu, it is also expected that similar sites in ε-iron nitride are inaccessible for Cu atoms.

The phase diagrams Fe–Cu–N at 813 K shown in Figure 6 were constructed based on

(a)

The Fe–N binary phase diagrams shown in Figure 2, in particular considering the N solubility in the ε phase, ranging approximately from 25 at.% to about 33 at.% (see also [2]) with a negligible solubility for Cu.

(b)

The low thermodynamic stability of Cu₃N requiring yet unknown high r_N values at 813 K, which are likely even higher than values derived for the Ni/Ni₃N equilibrium [18], as discussed in Section 1. This implies the existence of two-phase regions of the γ′ and ε nitrides with Cu. For simplicity it is, however, assumed that at high r_N values an equilibrium ε + Cu₃N exists.

(c)

Continuous substitution of Fe (Fe(0); see above) by Cu up to a composition corresponding to Fe_3.4Cu_0.6N [37], while minor deviations from 20 at.% N content are ignored.

Considering the ternary extension of the γ′ homogeneity region implied by point (b), the two-phase regions α + γ′ and ε + γ′ will also extend into the ternary compositional region. The arrangements of the α + Cu and α + γ′ two-phase regions require that the three-phase triangle α + γ′ + Cu in Figure 6b implied by point (b) has a corner at the low-N edge of the γ′ homogeneity region. That corner has been chosen to be at the Fe_3.4Cu_0.6N endpoint of the γ′ region. A further γ′ + ε + Cu triangle has a corner at the ε phase and can connect all three phases only if the triangle touches the γ′ region at the Fe_3.4Cu_0.6N endpoint. Hence, the γ′ + Cu two-phase region degenerates to a line. Adjacent to the γ′ + ε + Cu triangle, there is a two-phase region ε + Cu, which, in view of point (b), is followed by an experimentally unsupported ε + Cu + Cu₃N three-phase region, which is shown in Figure 6b to end at a ε + Cu₃N two-phase region. Higher N-contents than corresponding to the low-N end of this region are not considered here.

The diagram constructed in Figure 6b is a basis also for the diagram in Figure 6a. It was assumed that, at a given molar fraction of N, the r_N required to produce a given phase increases with increasing u_Cu, in agreement with the general trend of a lower stability of Cu nitrides as compared to Fe nitrides. Thereby, an extended γ′ field is obtained. Similarly, extended fields are shown for the other phases, exaggerating the homogeneity ranges with some artistic freedom in order to depict the three-phase reactions occurring with increasing r_N, which are invariant at the adopted constant pressure of 1 atm. In view of the γ′ + ε microstructures with Cu being depleted in the ε phase encountered at the surface of the Fe-1Cu alloy at r_N = 2.5 atm^−1/2 and Fe-1Cu and Fe-5Cu alloys at r_N = 4 atm^−1/2, it is concluded that the value of r_N corresponding to the invariant reaction γ′ ⇄ ε + Cu is not yet attained and is higher than 4 atm^−1/2. It appears, however, likely that the results are strongly affected by sluggish partitioning of the substitutional elements, i.e., enrichment of the Cu in the γ′. For instance, the non-observation of ε nitride in the Fe-5Cu alloy after treatment at r_N = 2.5 atm^−1/2 might be the consequence of a state of paraequilibrium. Note that, upon the formation of ε, Cu/Fe interdiffusion has to occur over much larger distances than upon dissolution of the Cu precipitates upon γ′ formation (compare the size of the ε grains in Figure 5 as compared to the distance between the Cu precipitates in Figure 4a). The microstructures developing, in any case, are strongly affected by pore formation, which is usually attributed to N₂ bubbles produced inside the solid [45]. Possibly, surface or grain-boundary diffusion contributes to the Fe/Cu partitioning necessary for ε formation, thus also contributing to the irregular microstructure shown in Figure 5.

The experiments of the current study have been conducted to reveal the basic influence of Cu in the Fe substrate on the outcome of nitriding. A more in-depth study is required in order to arrive at quantitative conclusions on the phase equilibria, e.g., to allow for thermodynamic modelling of the Fe–Cu–N system. This can only be achieved in future research.

5. Conclusions

Nitriding of Fe-1Cu and Fe-5Cu at 813 K at various nitriding potentials revealed marked differences as compared to nitriding of pure α-Fe at similar conditions. The presence of copper appears to stabilise the γ′-Fe₄N as compared to the ε-Fe₃N_1+x iron nitride. This experimental observation can be rationalized in terms of much higher solubility of Cu in the γ′ than in the ε phase. The high solubility of Cu in the γ′ can be explained by the phase’s crystal structure providing a particular atomic Fe-site which has no close N atoms in its neighbourhood. Such a site can preferentially be occupied by Cu atoms as already shown in previous works, whereas a corresponding site is missing in the crystal structure of the ε-iron nitride. On the basis of the data, a schematic phase diagram of the Fe–Cu–N system has been sketched for the temperature of 813 K. The considerations in the present work can serve as a model case for investigation of other elements M on the corresponding Fe–M–N systems.