Influence of Ni in high Fe containing recyclable Al-Si cast alloys

Research in recycling Al-alloy is necessary for a sustainable industrial development. Iron present in the recycled Al-alloys deteriorates its mechanical properties. The challenge, therefore, is to tackle the iron impurity using different methodologies. The present study focuses on a strategy by which iron containing beta phase could be destabilized with the addition of Ni. A large number of microstructural image, lattice parameter data and mechanical properties have been obtained using optical microscopy with state-of-the-art image analysis, FESEM with EDS and EBSD, XRD, Vickers microhardness and universal tensile testing. Based on these results, the present work provide necessary insight about the effect of Ni addition in the recycled Al-Si cast alloys containing as high as 2wt% Fe. Finally it was concluded that upto 4wt% Ni addition could be beneficial for Al-Si alloys Si content limited to 9wt%.


Introduction
Energy requirement for the production of aluminium could be reduced by ten times by recycling the aluminium products rather than extracting the same from the bauxite ore 12 [1] [2]. However, the recycled aluminium accumulates different metallic impurities due to the diversified source of the scrap and also with the increasing number of recycling processes. Presence of these metallic impurities, either in solid solution or as a separate phase, poses a great challenge to obtain the same physico-mechanical properties as that of the virgin aluminium 34 [3,4]. Therefore, the aim of the recycling could be shifted from making the commercially pure Al to developing a new castable Al-alloy with a different chemical composition. Such method of developing of new castable Al-alloy through recycling process enjoys a huge economical benefit 5 [5]. Despite the substantial energy saving and economic viability such newly developed alloy suffers from the accumulated metallic impurities, and amongst all metallic impurities Fe is known for its notoriety. In cast Al-alloys where Si is ubiquitous, Fe promotes the formation of the so-called β-A C C E P T E D M A N U S C R I P T phase (Al 9 Fe 2 Si 2 ) 67 [6,7]. It is now established that presence of Fe reduces the castability by reducing the fluidity of the melt and by increasing the porosity in the cast 89 [8,9]. The network of β-phase, which appears as needle like feature in the micrograph, is directly responsible to reduce the strength and ductility of the alloy 1011 [10,11].
There exist different philosophies on how to tackle iron impurity in the recycled cast Al-alloy.
Merits and disadvantages of each school of thoughts have been discussed elsewhere in details 5 [5]. It suffices here to briefly state the methodologies of tackling β-phase for the sake of the completeness; as mentioned belowi) Removal of β-phase (and thereby iron impurity) by gravity segregation or by the filtration of the liquid metal 12 [12] ii) Modification of the β-phase morphology by suitable heat treatment 7 [7] iii) Destabilizing the β-phase by changing the chemical composition and promoting some other phase with or without further heat-treatment 613 [6,13] The last method attracted many research works in the recent past; including the current one. For example, Mn was the first element identified as the modifier that promotes the formation of the polygonal Al 15 (Fe, Mn) 3 Si 2 phase rather than the β-phase, however, its effect vanishes beyond a Fe concentration of 1.2wt% in the Al-alloy 6 [6]. It was recently demonstrated that Cu can destabilized the β-phase by promoting the formation of ω-phase (Al 7 Cu 2 Fe) 13 [13]. The present work demonstrates Ni as the potential element to destabilize the β-phase and investigates its effect in the microstructure and the tensile properties of the modified Al-Si-Fe alloys with Si of 6wt%, 9wt% and 12wt% having a fixed Fe content of 2wt%.

Experimental
In these Al-Si-Fe-Ni alloys, three Si content were chosen, namely, 6wt%, 9wt% and 12wt% and for each Si content Ni was varied from 0wt% to 8wt% with an interval of 2wt% and Fe content was kept at a constant 2wt% for all these alloys. For the sake of brevity an alloy designated as 9Si6Ni would indicate an Al-alloy with a composition of 9wt% Si, 6wt% Ni, 2wt% Fe and the rest is Al. This style of designation will be used throughout in the rest of this article to refer a particular alloy composition.  14 [14]. Apart from the cast samples, homogenized specimens were made by cutting the cast samples pieces and isothermally holding at 823K for 48 hrs. followed by water quenching.
Standard metallographic technique with a final finish with 0.025µm colloidal silica polish was used for microstructural evaluation using both optical microscope and field emission scanning electron microscope (FESEM) fitted with energy dispersive spectroscope (EDS) and electron back scatter diffraction (EBSD) camera. Chemical composition of any particular phase, determined using EDS, reported here is based on the average value of 10 different measurements made in spot mode. Image analysis was carried out on the optical photomicrographs using ImageJ computer code 15 [15]. X-ray diffraction (XRD) traces were recorded using Bruker D8 machine that uses θ-θ goniometer with sample rotation; 1.6 kW Cu-K α radiation with a step size of 0.01 o and 1 s dwell time at each step was used. Rietveld analysis of the XRD traces was carried out with GSAS II software 1617 [16,17].Vickers microhardness testing were carried out as per ASTM E384-17 standard using 1 kg load and a dwell time of 10 s 18 [18]. Rectangular shaped tensile samples were cut for tensile testing. Round and rectangular tensile specimens with some selected 9Si and 12Si alloy with varying Ni content upto 4wt% were made using high pressure die casting (HPDC). Sample dimensions and tensile test parameters conform to ASTM B557-10 standard 19 [19].

Results
Image analysis is a useful tool for determining the area fraction of multiple phases which might or might not appear in similar grey-scale level on lightly shaded matrix; thus the area fraction of matrix phase can also be determined. This data also corroborates the phase fraction (by volume) obtained from the XRD analysis. It is pertinent here to discuss about the technique deployed for  Table 1 and compared with that of homogenized samples in supplementary table S1.
Lattice parameters and the phase fractions obtained from the Rietveld analysis along with the matrix area fractions calculated from the image analysis are presented in Table 1. Whole pattern fitting of the XRD traces for Rietveld analysis for the as-cast samples could be found in the supplementary fig. S4 (a-l). It is worthwhile to note that area fraction and volume fraction would never match even for the uniformly distributed equal sized spherical particle as a second phase; nonetheless, area fraction and volume fraction has a one to one correspondence between them 20 [20].
To respectively. SEM micrograph of as-cast 12Si2Ni sample is presented in fig. 6 (a) and corresponding elemental mapping of Si-K α , Fe-K α and Ni-K α .using EDS are presented in fig. 6 (b-d) respectively. Al K α , Si K α , Ni K α and Fe K α x-ray respectively.

Microstructure
It has been shown earlier that, for a given concentration of Fe, the formation of β-phase (Al 9 Fe 2 Si 2 ) requires a minimum level of Si concentration in Al-Fe-Si system 7 [7]. In the present alloys Fe was kept constant at 2wt%, as a minimum Si content of 1.5wt% is required for the formation of β-Al 9 Fe 2 Si 2 phase 7 [7]. However, juxtaposing fig.1  with certainty that addition of Ni tends to destabilize the β-Al 9 Fe 2 Si 2 and favors the formation of Al 9 FeNi phase; also, this destabilization effect is strong at low Si content. Table 1 also ascertains that formation of Al 3 Ni phase requires a minimum amount of Ni addition for a given amount of Fe content. Hao et. al. has also shown earlier that in Al-8wt%Si-1wt%Fe alloy addition of even 1wt% Ni destabilized β-Al 9 Fe 2 Si 2 phase formation 21 [21].
It is important to note that β-phase forms in needle like morphology, as shown in 12Si samples Effect of homogenization on the Si and β-phase are well known and the probable mechanisms of morphological changes are also proposed for the Al-Fe-Si system 7 [7]. Addition of Ni does not play any significant role in changing the morphology of Si. The morphological changes of phases in 12Si2Ni and 12Si4Ni sample is rather apparent as presented in the fig. 7 (a-b) and (cd), respectively; where the blocky Si particles in as-cast microstructure tend to be spherodized, the β-Al 9 Fe 2 Si 2 phase and Al 9 FeNi phase gets fragmented in sausage like morphology. The compositional changes of different phases will be discussed in the next section.
In case of higher Ni content alloys the Al 3 Ni phase usually forms onto the matrix-Al 9 FeNi interface as presented in the fig. 9 (a-c). The phase map image generated by indexing Kikuchi patterns, in fig. 9 (c), is particularly important to identify the Al 9 FeNi phase (blue colored) and Al 3 Ni phase (red colored) distinctly. It was observed during Rietveld fitting that both Al 9 FeNi phase and Al 3 Ni phase required higher order of spherical harmonics (often more than 6 th order), signifying highly textured phase. The same could be reconfirmed through orientation mapping (Euler angle) as presented in fig. 9 (d); it shows identical coloring for Al 9 FeNi phase disjoint across the spatial scale indicating that this phase posses high degree of crystallographic preferred orientation. Similar comment can also be made for the Al 3 Ni phase, in fact texture index for this phase is even higher than the Al 9 FeNi phase, as revealed during Rietveld analysis. It needs to mention here that Al and Si phase was not considered during the indexing of Kikuchi pattern since the phase map and orientation map would look extremely cluttered and the very purpose of showing Al 9 FeNi and Al 3 Ni phase would go in vein. Fig. 9 (e-h) is the elemental mapping corroborating the presence of these phases. It is important to note here that even after homogenization treatment no major change was observed in terms of the phase fraction, except for the fact the more Al 9 FeNi phase forms with the expense of β-Al 9 Fe 2 Si 2 phase. The homogenization treatment does not alter the crystallographic orientation of Al 9 FeNi and Al 3 Ni phase as well, as could be observed from the fig. 9 (d).

Phase composition
It is known that β-Al 9 Fe 2 Si 2 phase is highly faulted structure and because of that it has considerable capacity in accommodating other elements (e.g. Cu etc.) into its lattice by suitable site substitution 22 [22]. In 12Si2Ni alloy, EDS analysis reveals that Ni can replace the Fe atoms in the β-Al 9 Fe 2 Si 2 phase; in as-cast condition about 13% of Fe sites were occupied by the Ni atom and upon homogenization Ni occupies about half of the Fe sites in the β-Al 9 Fe 2 Si 2 phase; however, no alteration in Si content was observed in these two cases. Therefore, under equilibrium condition the β-Al 9 Fe 2 Si 2 phase in 6Si2Ni alloy can be expressed as Al 9 (Fe 0.5 Ni 0.5 ) 2 Si 2 . Similar site occupancy of β-Al 9 Fe 2 Si 2 phase gives better fitting during the Rietveld analysis of XRD data.
EDS analysis shows that in as-cast 6Si2Ni sample the Al 9 FeNi phase contains 2at%Si, more than 13at% Fe and about 8at% Ni against the theoretical value of about 9.1 at% for Fe and Ni. Upon homogenization it was observed that while Fe decreases slightly to reach 11at%, the Si content increases in Al 9 FeNi phase from 2at% to upto 6.5 at%; and Ni too approaches 6.5at% from 8at%.
However, it was observed that the sum of Si content and Ni content remains nearly equal to Fe content. It needs to be emphasized here that in Al 9 FeNi phase both Fe and Ni occupy the same site since Al 9 FeNi phase is crystallographically equivalent to Al 9 Co 2 prototype 23 [23]. Therefore, in case of 6Si2Ni alloy the equilibrium Al 9 FeNi phase can actually be expressed as Al 9 Fe x (Si 0.5 ,Ni 0.5 ) 2-x , where x is slightly less than 1 and solely depends on the alloy composition.
Similarly, in case of homogenized 12Si8Ni alloy Al 9 FeNi phase can be approximated as Al 9 (Fe 0.67 Si 0.33 )Ni, since Fe was about 6at% and approximately twice than that of Si content, however, sum of Si and Fe content remains nearly equal to the Ni content. Therefore, it could be said that in the low-Si low-Ni end in 2wt% Fe isopleths of Al-Fe-Si-Ni system (e.g. 6Si2Ni) Fe:Si:Ni is equals to 2:1:1 in Al 9 FeNi phase; whereas in high-Si high-Ni end (e.g. 12Si4Ni and 12Si8Ni) the ratio was found to be 2:1:3 as sum of Fe and Si content equals to the Ni content.
Interestingly, in 12Si2Ni sample the chemical composition of Al 9 FeNi phase reveals the ratio of Fe:Si:Ni as 1:1:1.

A C C E P T E D M A N U S C R I P T
It is rather clear from the EDS line scanning that during homogenization part of β-Al 9 Fe 2 Si 2 phase gets converted to Al 9 FeNi phase at the β-matrix interface, as evident from the fig. 8 (a-b).
This observation highlights the fact that Ni tends to destabilize β-Al 9 Fe 2 Si 2 phase and promotes the formation of the Al 9 FeNi phase. Despite the usual limitation of the EDS technique in terms of the certainty in the chemical composition, the relative changes of Si, Fe and Ni in Al 9 FeNi phase clearly indicates large solubility range of these elements in that intermetallic phase; however, effect of chemical composition is not very apparent in their lattice parameters when ascast samples are considered.

Phase diagram and solidification
At this point it is imperative to compare the phase fractions obtained from the XRD data with the same obtained from the thermodynamic calculations. Therefore, to obtain the yield of different intermetallic phases both Scheil solidification model and equilibrium solidification model were considered. It is known that for the substitutional alloys Scheil solidification model works with reasonable accuracy 24 [24]. MatCalc software was used alongwith Hao's database for assessing the solidification and yield of the intermetallic phases 21 [21]. Fig. 4 (a), (b) and (c) represent yield (in wt%) of β-Al 9 Fe 2 Si 2 , Al 9 FeNi and Al 3 Ni phase as obtained from the Scheil solidification calculation. Fig. 4 (a) clearly shows the efficacy of Ni in destabilizing the Al 9 F 2 Si 2 phase in low-Si alloys; e.g. in 6Si alloy only 2wt%Ni is enough to destabilize β-Al 9 F 2 Si 2 phase, whereas in 12Si alloy the same could be achieved with 4wt%Ni. On the other hand fig. 4 (b) and (c) depicts increasing yield of Al 9 FeNi and Al 3 Ni phases, respectively, with increase in Ni content in an alloy for a given Si content. This trend is similar to that observed from the XRD results as well (see Table 1). Fig. 4 (d) represents the yield of Al 9 FeNi, the only intermetallic phase obtained as per the equilibrium solidification. It is interesting to note that the phase fraction and the trend pertaining to the yield of Al 9 FeNi phase obtained from the XRD analysis match fairly well to the same obtained from the equilibrium solidification calculation ( fig. 4-d).
Supplementary fig. S5 represents the isopleths calculated by MatCalc with their Al-database (top row) and the same using PandaT software with Hao's database 21 [21]. Isopleths obtained from MatCalc databse does not show the Al 9 FeNi as the stable phase at the room temperature; whereas Hao's database does not show Al 3 Ni as the stable phase. It is clear from the fig. S5 that these isopleths differ considerably in terms of temperature, phase stability and the phase boundary.

A C C E P T E D M A N U S C R I P T
In Al 9 FeNi crystal structure, the Fe and Ni sites are equivalent and therefore it was seen that Si can replace both Fe and Ni depending on the Si and Ni content of the alloy; therefore Al 9 FeNi phase can take up a large amount of Si; as discussed in detail in the previous section with the help of EDS result. On the other hand, as discussed earlier, Ni can replace as high as half of the Fe sites in Al 9 Fe 2 Si 2 phase at higher temperature and rejects Ni at lower temperature. Therefore, it seems reasonable to assume that both Si and Ni play a significant role in deciding the relative stability of the Al 9 FeNi and Al 9 Fe 2 Si 2 phase, respectively.

Mechanical properties
The micro-hardness value of the both as-cast and homogenized alloys monotonously increases with increasing Ni, as shown in fig. 10 (a-b). The overall increase in hardness with increasing Si and Ni is due to the higher yield of the intermetallic phases, e.g. Al 9 FeNi and Al 3 Ni. However, homogenization treatment decreases the hardness and the extent of decrease is more at higher Si and Ni content. Such decrease is expected due to several reasons, e.g. chemical equilibration between matrix phase and intermetallic phases; the spheroidization of different phases that eliminates the stress concentration to a larger extent, as evident in fig. 7 (b) and (d).
The 0.2% yield stress and UTS are found higher in high pressure die cast (HPDC) sample than corresponding gravity mold cast sample, as can be seen from fig. 11 (a-b); however, the ductility of HPDC sample deteriorates beyond about 4wt%Ni. Overall it is apparent that irrespective of the Si content and the casting technique the alloy composition beyond 4wt%Ni reduces the ductility. It can be assumed safely that the ductility is not dependent on the size of the dendrites as the ductility does not vary much from gravity cast (GC) to HPDC samples for a given composition, see fig. 11 (c), despite huge difference in the cooling rate that determines the interdendritic spacing. Therefore, in the present case the ductility is chiefly a function of the fraction of the intermetallic phase. On the other hand due to higher cooling rate in HPDC the nucleation rate was also higher and therefore the intermetallic phases were finely distributed over the casting this result into the significant rise in the yield strength and UTS.

ACCEPTED MANUSCRIPT
A C C E P T E D M A N U S C R I P T

Conclusions
i. β-Al 9 Fe 2 Si 2 phase exhibits solubility for Ni, where Ni occupies Fe sites.
ii. Al 9 FeNi phase shows appreciable solubility for Si and Si can replace both Fe and Ni depending on the Si and Ni content of the alloy.
iii. Present thermodynamic description is indecisive about the stability of Al 9 FeNi and Al 3 Ni phase below 550 o C.
iv. Ni destabilizes β-Al 9 Fe 2 Si 2 phase and favors formation of Al 9 FeNi phase in Al-Si-Fe-Ni alloy system; the effect of destabilization is strong in high Fe containing Al-alloys with Si content less than 9wt%.
v. Ni addition is beneficial in high-Fe Al-Si cast alloy with low-Si content of below 9wt%; however, Ni content of more than 4wt% deteriorates the mechanical properties of the alloys.
A C C E P T E D M A N U S C R I P T     A C C E P T E D M A N U S C R I P T Fig. 1. From top row to bottom row, 6Si0Ni, 6Si2Ni, 6Si4Ni, 6Si6Ni and 6Si8Ni as-cast samples. As-obtained micrographs are in the first column; second column is FFT filtered and contrast adjusted and the last one is thresholded micrographs for calculating area fraction.

Fig. 2.
From top row to bottom row, 9Si0Ni, 9Si2Ni, 9Si4Ni, 9Si6Ni and 9Si8Ni as-cast samples. As-obtained micrographs are in the first column; second column is FFT filtered and contrast adjusted and the last one is thresholded micrographs for calculating area fraction. Fig. 3. From top row to bottom row, 12Si0Ni, 12Si2Ni, 12Si4Ni, 12Si6Ni and 12Si8Ni as-cast samples. As-obtained micrographs are in the first column; second column is FFT filtered and contrast adjusted and the last one is thresholded micrographs for calculating area fraction.        Si can replace both Fe and Ni sites in Al 9 FeNi phase.
 Destabilization of β-Al 9 Fe 2 Si 2 phase due to Ni is more prominent when Si content is less than 9wt%.
 Ni content of more than 4wt% deteriorates the mechanical properties of the alloys.