Melt Protection of Mg-Al Based Alloys

Abstract

This paper reports the current status of Mg melt protection in view to identify near-future challenges, but also opportunities, for Mg melt protection of Mg-Al based alloys. The goal is to design and manufacture sustainable Mg alloys for resource efficiency, recycling and minimising waste. Among alternative cover gas technologies for Mg melt protection other than SF₆: commercially available technologies containing―HFC-134a, fluorinated ketone and dilute SO₂―and developed technologies containing solid CO₂, BF₃ and SO₂F₂, can potentially produce toxic and/or corrosive by-products. On the other hand, additions of alkaline earth metal oxides to Mg and its alloys have developed a strong comparative advantage in the field of Mg melt protection. The near-future challenges and opportunities for Mg-Al based alloys include optimising and using CO₂ gas as feedstock for both melt protection and grain refinement and TiO₂ additions for melt protection.

1. Introduction

The design and manufacturer [1] of sustainable Mg alloys and the need to reduce environmental impacts are the main factors that advance the development of new processes such as the replacement of SF₆ gas [2,3,4,5]; and industrial furnace design, including lid design, addressing energy efficiency and metal losses in new and existing furnaces. In comparison to structural materials such as Al and steel, Mg components for automotive applications have better greenhouse gas emission performance (GHGEP) based on regular lifecycle assessments. However, this GHGEP advantage of Mg components can be significantly reduced depending on the production technology adopted, i.e., electrolytic versus thermal process, in both cases assuming replacement of SF₆ cover gas [6].

Since SF₆ has been identified as a GHG [5,7,8], the protective films formed on molten magnesium and its alloys under gas protection have been an active subject of research, with the aim of finding suitable industrial alternatives to SF₆ gas for melt protection [7,8,9,10,11,12,13,14,15,16,17,18,19,20,21]. The use of SF₆ gas will be banned from the European Union from 2018 [22] (as cited in Reference [23]), in which case, replacement of SF₆ as a cover gas in the Mg industry would no longer be eligible as Clean Development Mechanism (CDM) project or Joint Implementation (JI) project, nor sell the reduced emission as ‘carbon credits’ [24]. The International Magnesium Association (IMA) has recognised the need to identify melt protection alternatives with both low toxicity and GHG emissions [25,26]. In this paper, taking as a starting point a brief overview of the work on Mg melt protection, the near-future challenges, but also opportunities, for melt protection of Mg-Al based alloys are identified.

2. Reactive Element Effect

It has been seventy-nine years since the ‘rare earth element effect’ was first proposed [27]. The original premise was that it was possible to obtain Ni-20Cr alloys in order to improve resistance to high temperature oxidation by additions of rare earth elements. A few years later the original concept of rare earth element effect was adapted and extended to other elements with high affinity to oxygen, to increase the corrosion resistance of refractory alloys without impairing their creep resistance. At this time, the ‘active elements’ suggested by Pfeil’s patent were Sc, Y, La, Ti, Zr, Hf, Nb, Ta, Al, Si, Ce, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu and Th [28] (as cited in Reference [29]). The topic became more commonly known as the ‘Reactive Element Effect’ (REE). A detailed review of the REE on high temperature oxidation of Mg was recently published by Czerwinski [27], including Ca, Be, Sr and Ti, which readily oxidise when exposed to oxygen, give the magnesium alloy improved resistance to oxidation during melting. The high affinity to oxygen of those and all other rare earth elements are the basis of the REE [30,31,32]. Some examples of Ca, Be, Sr and Ti additions to molten Mg are given below. An Ellingham diagram [33,34] for their oxides including Al and Mg oxides is depicted in Figure 1, from which it can be seen that their affinity to oxygen until ~1000 K (726.85 °C) decreases in the following order Ca > Be > Mg > Sr > Al > Ti.

2.1. Ca Additions

Sakamoto et al. [35] reported that the thin oxide layer formed on the surface of molten Mg―1.5Ca alloy was protective and consisted of an outer layer of CaO and an inner layer composed of a mixture of CaO and MgO. In a subsequent study by Wiese et al. [23] at higher additions of Ca, it was found that the thin oxide layer formed on the surface of molten Mg―7.2%CaO alloy was also protective after exposure at 720 °C for 5 min in both air and SF₆/Ar and by using SEM and TEM techniques, identified the formation of Mg/Mg₂Ca eutectic phase at the inner surface layer and of a mixture of MgO/CaO at the outer surface layer, but also Mg/Mg₂Ca eutectic phase at the outer surface layer, suggesting that Mg₂Ca phase was protective. It is interesting to note that Mg₂F was not found at the surface layer despite using an SF₆/Ar gas mixture in the latter case. This was attributed to be due to the possible protective character of the MgO/CaO mixture, preventing the reaction of SF₆ gas at the surface of the Mg melt.

In practice, additions of CaO to Mg and Mg-Al based alloys can lead to the formation of Mg₂Ca according to reactions (1) and (2), respectively [4]. Note the solubility of Ca in liquid Mg. According to the Mg-rich region of the equilibrium phase diagram of eutectic type for the Mg-Ca system [36], the invariant point is 10.5 at.% Ca at 516.5 °C. Therefore, the presence of Mg₂Ca phase is to be expected for example at 700 °C in high Ca-containing Mg alloys only.

10Mg_(l) + 4CaO_(s) = 4Mg₂Ca_(s) + [O₂ + 2MgO (dross)]

(1)

Mg [Al, other alloying elements]_(l) + CaO_(s) = Mg₂Ca_(s) + Al₂Ca_(s) + [(Mg, Al, other alloying elements)₂Ca_(s)] + [O₂ + MgO (dross)]

(2)

Indeed, recent work by Wiese et al. [37,38] confirms the reaction of CaO and the formation of Mg₂Ca laves phase in the bulk of Mg-16Ca + 6CaO and Mg-10CaO (in wt.%) by in situ synchrotron measurements of the reaction between molten Mg and CaO. Mg₂Ca and Al₂Ca are Laves phases with crystal structures and melting points of: hexagonal, 714 °C (Mg₂Ca) and cubic, 1079 °C (Al₂Ca), respectively [39]. The additions of alkaline earth metal oxides to Mg and its alloys have developed a strong comparative advantage in the field of Mg melt protection.

2.2. Be Additions

Be has a higher affinity to oxygen than Mg. Trace additions of beryllium in the range between 5 and 30 ppm are made to Mg alloys [40], in particular AZ91D, AM60, AM50 and AZ31 Mg alloys may contain 5–10 ppm Be [27].

2.3. Sr Additions

Rare or alkaline earth metals additions are commonly made to Mg-Al based alloys for creep-resistance applications, e.g., Mg-Al-RE and Mg-Al-Sr alloys [41]. The presence of SrO/MgO in AJ62 alloy would be initially expected according to reaction (3) [42]. Note the solubility of Sr in liquid Mg. At high concentrations of Sr, it could react with molten Mg to form intermetallics such as Mg₁₇Sr₂, Mg₃₈Sr₉, Mg₂₃Sr₆ and Mg₂Sr phases [43], but also SrO could react according to reaction (4) [4].

2Mg_(l) + 2Sr_(l) + 4/3Al₂O_3(s) = 2MgO_(s) + 2SrO_(s) + 8/3Al

(3)

at 690 °C, ΔG°₃ = − 947,620 J/mol.

Mg [Al, other alloying elements]_(l) + SrO_(s) = Mg₂Sr_(s) + Al₂Sr_(s) + [(Mg, Al, other alloying elements)₂Sr_(s)] + [O₂ + MgO (dross)]

(4)

From previous studies [42,44], Sr-bearing AJ62 alloy held at 690 for 10 s exhibited higher resistance to oxidation than that of Sr-free AZ91D alloy held at 650 for 10 s, as shown from macroscopic observations of the reaction products formed on the surface in Figure 2, from which it can be seen two distinctive regions of non-nodular growth (in a layer manner) and of nodular growth, indicating two different stages of the surface reaction. Samples from these regions are outlined by the white squares marked 1―non-nodular growth―and 2―nodular growth―in Figure 2. FEG-SEM revealed that all the morphologies, i.e., sponge-like, nodular and layer, were porous and hence non-protective with respect to the evaporation and oxidation of Mg, Zn and Sr from the corresponding surface of the AZ91D and AJ62 alloy melt, as shown in Figure 3. XRD results from the surface of AZ91D, AZ31, AM60 and AJ61 alloys indicated that, in all these regions, MgO was the main oxidation product along with traces of AlN. Additionally, traces of ZnO reflected the Zn levels of the Mg magnesium alloys and the oxidation conditions investigated.

2.4. Ti Additions

Some notable successes in Mg and Mg-9Al alloy have been an increase of over 30% in strength and elongation with the introduction of O interstitial atoms, 50% increase in formability, double creep resistance and a 10-fold increase in corrosion resistance [45].

First TiO₂ nano-particles having high chemical potential energy are introduced into the Mg melt, and then they decompose to O, Ti and TiO. Subsequently, Ti and TiO having high specific gravity sink down to the bottom of the crucible. Finally, oxygen atoms in the Mg melt will remain interstitial after solidification. This was due to an increase in the activation energy for the formation of MgO.

Ti has been found to inhibit Mg corrosion in aqueous solutions [27], and might also impact greatly on the field of high temperature oxidation of Mg alloys. TiO₂ particles could react at the surface of Mg-Al based alloys. Thermodynamic analysis in conjunction with experimental evidence showed that the formation of TiAl can be favoured at the expense of other Ti_xAl_y compounds [46]. This is because Mg has a higher affinity to oxygen than Ti and Ti reacts preferentially with Al to form TiAl (reaction (7)). TiAl has a high oxidation limit of 900 °C and a low density of 4.1–4.7 gr/cm³ [46,47]. In contrast, the oxidation resistance of Al₃Ti is much better than those of TiAl and Ti₃Al [47].

3TiO_2(s) + 7Al_(l) = 2Al₂O_3(s)+ 3TiAl

(5)

at 700 °C, ΔG°₅ = −648,733 J/mol [46]

3Mg_(l) + Al₂O_3(s) = 3MgO_(s) + 2Al_(l)

(6)

ΔG°₆ = −145,810 + 26.32T [48], where ΔG° is in J/mol and T is in K.

Reactions (5) and (6) can be combined to obtain the standard Gibbs energy change of reaction (7) at 700 °C.

3TiO_2(s) + 3Al_(l) + 6Mg_(l) = 6MgO_(s) + 3TiAl_(s)

(7)

ΔG°₇ = ΔG°₅ + 2ΔG°₆. At 700 °C, ΔG°₇ = −889,126 J/mol, hence, reaction (7) is shifted to the right to form MgO and TiAl.

3. Flux Additions

Typical salt fluxes are chloride KCl, NaCl and MgCl₂ and fluoride CaF₂. MgCl₂ can be used in Mg alloys containing Al, Zn and Mn alloying elements such as AZ91D, AM60, AM50 and AZ31 representing the majority of commercial Mg alloys. Flux entrapment, release of corrosive gas and melt losses are the major causes for the shift from flux to gas protective atmospheres [49].

4. Alternatives to SF₆ for Mg Melt Protection

Fruehling, investigated protective atmospheres for molten magnesium [50]. Mirak et al. [16] summarised the significant findings on protective gases in stagnant melts held isothermally under a mixture of a protective gas (e.g., SF₆, SO₂ or 1,1,1,2-tetrafluoroethane (HFC-134a)) and a carrier gas (e.g., dry air or N₂), based on characterisation of the reaction layer formed on the melt surface by the use of various techniques, including SEM, TEM, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS) and thermogravimetric analysis (TGA) and assessment of the corresponding oxidation behaviour. Mg melt protection in the context of the nature, origin and control of inclusions with regard to reactions with air, fluxes and protective gases during melt treatment, casting and alloying was later reviewed by Lun et al. [49].

Melt protection alternatives to both flux-based and SF₆, consisting of active gases: fluorine-based blended gas (HFC-134a), fluorinated ketones (FK) and dilute SO₂ (1.5%) mixed with a carrier gas, generally N₂, CO₂ or dry air, are commercially available [20,25,26]. The work of Ha and Kim [18] showed that hydrofluorocarbon CF₃CH₂F (HFC-134a) possessed better protection properties than do SF₆ and SO₂, using air as a carrier gas. However, since HF can form in a mixture of HFC-134a/air and is highly corrosive, it was recommended the use of N₂ or CO₂ carrier gases instead of air. The reaction products between Mg melt and SF₆, HFC-134a and FK are MgO and MgF₂. Mg melt can react with SO₂ to form MgSO₄, MgO or MgS [49].

On the other hand, several alternative melt protection technologies that are not currently commercialized or readily available have been developed [20,25,26]. Those alternative techniques generate solid CO₂, a small amount of BF₃ gas and SO₂F₂ gas for Mg melt protection. The use of solid CO₂ can generate toxic CO [51]. BF₃ is not a GHG, however, is highly toxic, expensive and require the use of special storage conditions [49]. It should be noted that nowadays the use of SO₂ gas for Mg melt protection might require a SO₂ scrubber system (i.e., flue gas desulfurization (FGD) technology to comply with the environmental legislation) as well as additional costs associated with corrosion of equipment handling dilute SO₂. On the other hand, the Global Warming Potential (100 years) and atmospheric lifetime (years) values are 23,900, 3200 (SF₆); 1300, 14.6 (HFC-134a); ~1, 0.014 (Novec™ 612) and ~1,—(SO₂F₂) compared with the value of 1, 100–150 for CO₂, respectively [25]. As concerns about global climate change grow, search for alternatives exist to reduce GHG emissions including HFCs.

Aarstad [9] tested a mixture of 1% SF₆ in air, N₂, Ar and CO₂ and 1% SO₂ in air, N₂ and CO₂ to determine whether a protective effect was also achieved with other carried gases than air. It was found that when using inert gases N₂ and Ar as carrier gases, no protective film was formed on the surface of the Mg melt. Furthermore, SF₆ in CO₂ was not a successful combination, conversely when air was added to a mixture of SF₆ and CO₂, a protective film was formed on the melt surface. This is because air is necessary for the development of a protective film on the surface of the Mg melt [9]. In a parallel effort, the authors determined the solubility of fluorine in molten Mg in the range between 700 °C and 950 °C [52] and found that it did not appear to be sufficient for direct dissolution of fluorine into the melt to be an alternative to SF₆.

Small amounts of spinel phase can form on the surface oxide layer of molten AZ91 alloy after exposure to air [53] and two plausible mechanisms for its formation were suggested by the authors. The first mechanism [53] was based on a kinetic study of the surface oxidation of liquid Al―3%Mg and Al―3%Mg―3%Si alloys by Salas et al. [54], who proposed that oxidation of Mg to MgO reaction (8), can directly expose MgO to subsurface melt regions enriched in Al, depleted in Mg compared to the bulk of the melt. This coupled with the exothermic reaction (8) at a rate sufficient to cause local superheating of the melt may result in the formation of Al₂O₃ and MgAl₂O₄ through the endothermic reactions (6) (shifted to the left) and (9), respectively. Spinel would hence form at the alloy/MgO interface.

Mg_(g) + 1/2O_2(g) = MgO_(s)

(8)

ΔG°_MgO = −612,955 + 128.08T [14,15,19]

4MgO_(s) + 2Al_(l) = MgAl₂O_4(s) + 3Mg_(l)

(9)

ΔG°₇ = 110,210 − 28.41T [48], where ΔG° is in J/mol and T is in K.

The second mechanism [53] was based on an immediate oxidation to Al₂O₃ and MgO and their solid state reaction [55,56,57] (reaction (10)) to form spinel through-thickness.

MgO_(s) + Al₂O_3(s) = MgAl₂O_4(s)

(10)

at 800 °C, ΔG°_MgAl2O4 = −46,700 J/mol [58].

Recent work is also indicating that there may be approaches to Mg alloy process routes without protective gas other than SF₆ gas. Mg alloy melting can be carried out lowering the oxygen content inside the enclosed area [3] to prevent catastrophic oxidation, ignition and burning of the melt [35,59]. Kim et al. [2] demonstrated successfully the development of AZ31 Mg alloy wrought process route without protective gas, instead, by the addition of 0.42 wt.% CaO to AZ31 alloy melt being protected by N₂ gas in the temperature range from 640 °C to 660 °C. This reveals alternative approaches to Mg melting and demonstrates the need for further research into the entire topic.

Elimination [60] or reduction [61] of protective SF₆ gas has been achieved in the development of innovative processes such as semi-solid injection moulding. However, just as important to process considerations are the addition of inoculants such as MgO [62] and Al₄C₃ [63,64] to produce a grain-refined microstructure. AZ91D chips were mixed with carbon black using a semi-solid injection moulding technique at 630 °C, the in situ reaction between aluminium and carbon occurred to form Al₄C₃ particles [64]. The authors found an optimal carbon black addition of 0.23 wt.% to produce a grain-refined microstructure giving improved properties. Carbon’s reaction kinetics are critical to the conversion efficiency, which in turn can affect the effectiveness of corrosion resistance.

Pure CO₂ [50] and mixtures up to 20 vol.% air/CO₂ can protect molten magnesium [9,65,66]. Equally important, however, are carbon’s reaction kinetics. A thermodynamic analysis by Aarstad [9] predicted that traces of CO (1.0 × 10⁻⁷ moles) and O₂ (5.1 × 10⁻⁸ moles) were the decomposition products of 1 mole of CO₂ at 700 °C and 1 bar pressure, and that MgO and C were the reaction products between CO₂ and molten Mg. In parallel, Shih et al. [67,68] reported the possible reactions that can occur between Mg and, O₂, CO₂ and CO (reactions (8), (11)–(13)). A diagram of the standard Gibbs energy change versus temperature for the reactions (11)–(13) is given in Reference [67], from which it can be seen that, reactions (9)–(11) can proceed spontaniously until about 3725 °C, 2475 °C and 1975 °C, respectively. In particular, at 720 °C, ΔG°₁₂ = −592,700 J/mol [66].

Mg_(g) + CO_2(g) = MgO_(s) + CO_(g)

(11)

Mg_{(l), (g)} + 1/2CO_2(g) = MgO_(s) + 1/2C_(s)

(12)

Mg_{(l), (g)} + CO_(g) = MgO_(s) + C_(s)

(13)

Thermo-oxidative experiments of pure Mg in the liquid state [65,66] have been performed by isothermal TGA in pure CO₂ and mixtures up to 20 vol.% air/CO₂ to monitor the surface oxidation kinetics at high temperatures (670 °C, 720 °C and 770 °C) and to relate those to their morphological and structural characteristics at different oxidation stages (2, 3 and 7 min) as determined by SEM. The TGA and SEM results by Emami and Sohn [65,66] indicated that the surface had a double-layer structure. The external layer consisted of non-protective MgO, whereas the internal layer was dense and protective, and composed of a mixture of MgO/C.

Note the solubility of C in liquid Mg. According to the Mg-C equilibrium phase diagram which was proposed in Reference [69], the value of the solubility of C in Mg is approximately 10 at ppm (5 wt. ppm) at 700 °C. The solubility of C in liquid Mg increases with increasing temperature up to 1093.6 °C at 1 bar pressure. All of these results suggest that both melt protection and grain refinement should be considered in developing practical applications for using CO₂ in Mg-Al based alloys.

5. Conclusions

This paper has examined the topic of Mg melt protection. Among alternative cover gas technologies for Mg melt protection other than SF₆: Commercially available technologies containing―HFC-134a, fluorinated ketone and dilute SO₂―and developed technologies containing solid CO₂, BF₃ and SO₂F₂, can potentially produce toxic and/or corrosive by-products. Additions of alkaline earth metal oxides to Mg and its alloys have developed a strong comparative advantage in the field of Mg melt protection.

The near-future challenges and opportunities for Mg-Al based alloys include, optimising and using CO₂ gas as feedstock for both melt protection and grain refinement, and TiO₂ additions for melt protection, in order to design and manufacture sustainable Mg alloys for resource efficiency, recycling and minimising waste.