Y.Fu ,G.G.Wng ,A.Hu ,Y.Li ,K.B.Thcker ,J.P.Weiler ,H.Hu,*

aDepartment of Mechanical,Automotive and Materials Engineering,University of Windsor,401 Sunset Ave,Windsor,ON N9B 3P4,Canada

b Global Technology Centre,Meridian Lightweight Technologies,25 MacNab Ave,Strathroy,ON N7G 4H6,Canada

Abstract Sludge consisting of heavy element phases and oxides is often generated during the casting operation of aluminum (Al) and magnesium(Mg) alloys.With the help of the well-established Sludge Factor (SF) formula,it is relatively easy to control the sludge generation in aluminum alloys.But formation mechanisms and characteristics of sludge in die casting magnesium alloys are still unclear.To ensure the production of high quality die cast components at a low cost,a full understanding of sludge in die casting Mg alloys and its proper control measures need to be developed,since excessive sludge formation affects deleteriously material and operation cost,and casting performance.In the present report,the formation,characteristics and control of Mg die-casting sludge,based on the established knowledge of sludge formation and sludge factor in Al die casting alloys,are reviewed.Previous work on characterization and assessment of sludge in die cast Mg alloys are reviewed.Metallurgical principles for control of sludge in ingot production in association with die casting of Mg alloys are discussed.Rapid assessment of Mg oxide and intermetallics relevant to sludge formation in Mg alloys are highlighted.

Keywords: Sludge;Magnesium alloys;Die casting;Al-Mn(-Fe) intermetallics;Magnesium oxide;Microstructure characterization.

1.Introduction

Magnesium alloys as the lightest structural metals which have been increasingly desired hugely in the automotive related manufacturing industry,because they possess high strength-to-weight ratios,good ductility,low density and excellent corrosion resistance [1,2].Magnesium alloys have high fluidit and minimal affinit for tool steel.These properties,combined with its low density advantage,make them very suitable for the high pressure die casting process.They can be cast into near-net shape castings with thin walls and very complex geometry,with superior mechanical properties in as-cast state.Magnesium die castings are therefore widely used in automotive applications [3]

Currently,the most widely used commercial magnesium alloys,such as AM60B and AZ91D,are based on Mg-Al system.All the current magnesium die casting alloys are based upon aluminum as the main alloying element.In the Mg-Al alloys,Al8Mn5intermetallic particles form during solidification which are important for ensuring adequate corrosion resistance.However,the presence of the excessive Al8Mn5particles interacting with magnesium oxide inclusions above theα-Mg liquidus temperature leads to sludge formation and sedimentation.The interaction between the Al8Mn5particles and oxide film generates deleterious casting defects [4,5].The sludge from the die casting operation often experiences a buildup of solid heavy-element compounds at the bottom of the melting and holding crucibles.The sludge buildup can cause damage to the crucibles,the “hard-spot” inclusions in castings and the restriction of metal fl w fillin the die cavity.The sludge contributes a significan amount of metal loss in the Mg melting and casting process.In addition,the formation of excessive sludge formed on the bottom of the crucibles acting as a thermal barrier could result in significan melt temperature fluctuatio in the die casting crucibles.The melt temperature fluctuatio impacts normal casting operations.A very low melt temperature could stop the casting productions,while the melt picks up easily iron from the steel crucible at the high temperature to generate additional sludge.Therefore,sludge formation in the die casting crucible is a very important subject,but very often is overlooked by academia,customers,primary ingot producers and even die casters.It is essential to understand Mg sludge formation and minimize it as much as possible.

Compared with that in the Mg alloys,the sludge formation in Al die casting alloys has been well understood.A Sludge Factor relating to metal chemistry is proposed and proven to be useful to determine sludge formation tendency and quantity of a specifi alloy chemistry.The work had been performed to relate sludge formation with the Sludge Factor and the Al melt holding temperature [6-12].However,similar studies on the formation and control of sludge in die casting Mg alloys are limited.

This report intends to give an overview of formation,characteristics and control of Mg die-casting sludge,based on the established knowledge of sludge formation and factor in Al die casting alloys.Previous work on characterization and assessment of sludge in die cast Mg alloys are reviewed.Metallurgical principles for control of sludge in ingot production in association with die casting of Mg alloys are discussed.Rapid assessment of Mg oxide and intermetallics relevant to sludge formation in Mg alloys are highlighted.

2.Sludge formation in aluminum alloys

In-depth studies on metallurgical aspects of aluminum alloys,in particular,mechanisms of sludge formation,have been extensively carried out.This is because aluminum foundries,especially aluminum die casting operations,usually experience a buildup at the base of the melting and holding crucibles,commonly called “sludge”.Sludge is made of primary crystals that contain aluminum and silicon,and are also rich in iron (Fe),manganese (Mn),and chromium (Cr).These crystals have high melting points and high specifi gravity,which cause them to settle to the bottom of the melt.The precipitation of sludge crystals often occurs only in the melt having sufficientl large amounts of iron,manganese,and/or chromium in relation to the furnace operation temperature[10].

2.1.Sludge factor

Sludge formation depends on not only the chemical composition of Al alloys,but also the process parameters such as melting and holding temperatures,and holding time of the melt.To reveal the effect of the chemical composition on the sludge formation,Jorstad [6] and Groteke [7] define a Sludge Factor (SF) for die casting Al alloy A380.This factor is calculated from the Fe,Mn,and Cr contents of the alloys as follows:

Fig.1.Calculation of the sludge factor with the SF formula [6].

To minimize the sludge formation,a small SF needs to be maintained for the A380 alloy,when the furnace holding temperature is low.Shabestari,and Gruzleski [8] studied the influenc of chemical composition,holding temperature and cooling rate on sludge formation in Al-12.7 wt.% silicon(Si) alloys.They found that the Fe,Mn and Cr contents of the alloy as well as the cooling rate significantl affected the morphology,quantity,and size of the sludge particles.Sludge was unable to form until a specifi temperature was reached for a given Fe content,and the sludge forming temperature depended on the Fe content of the alloy.The following relationship was proposed to describe the dependence of sludge forming temperature on the iron content:

It was found that the sludge crystals consisted mainly of Fe,Mn and Cr-rich compounds.

2.2.Role of iron,manganese,and chromium

To prevent molten Al alloys from soldering die steel,iron is introduced into die casting aluminum alloys as a desirable element.Most casting Al alloys contain about 0.8 wt.% Fe.The Al alloys containing the Fe content above this amount exhibit almost no tendency to dissolve die steel,while the two materials are in intimate contact.For this reason,most aluminum die casters desire that their alloys contain between 0.8% and 1.1% Fe [6].Often Manganese and chromium present in the aluminum alloys as secondary alloying elements,which are beneficia to their mechanical properties.Jorstad [6] found that Mn and Cr changed the morphology of the Fe-rich phase from acicular to cubic form.Consequently,the ductility and strength of cast components were improved.

2.3.Determination of sludge factor

Fig.2.Sludge factor vs.Furnace temperature (°C=(°F -32) × 5/9) [6].

Fig.3.An example showing the selection of the holding furnace temperature(660-663 °C),i.e.,(1200-1250°F),for an A380 melt with a SF of 1.8 [6].

Fig.1 illustrates the calculation of the sludge factor which determines the tolerable limits by using the simple and straightforward formula.The SF value is 1.83 for the melt containing 0.89 wt.% Fe,0.35% Mn and 0.08 wt.% Cr.Fig.2 presents the tolerable sludge factor verses temperature chart,which can be used to determine a minimum holding furnace temperature for a given melt of the A380 alloy to avoid sludge formation during the die casting operation.The holding furnace temperature should increase with the SF values for the A380 melt.If the temperature for the melt with the high SF was insufficien high to prevent all the heavy-element phase precipitating from solution,the sludge buildup could take place.For instance,as shown in Fig.3,the furnace temperature of 660-663 °C (1200-1250°F) needs to be selected for the A380 melt with a SF of 1.8.

2.4.Effect of holding temperature on sludge formation

Despite the fact that many studies were conducted on the mechanism of sludge formation in Al-based alloys,published results seem inconsistent.Jorstad [6] and Groteke [7] (Fig.4)predicted that,for a hypoeutectic Al-Si alloy containing 1%Fe,sludge could form when the holding temperature was below 600 °C.But Shabestari [8] predicted that for the same iron content,sludge could be found only as the melt was held at a high temperature of 680 °C (Fig.4).

Fig.4.Temperature vs sludge factor for the studied alloys [10].

The influenc of holding temperature and time and alloy chemistry on sludge formation in Al-Si-Cu alloys was investigated by Flores et al.[9].It was found that,in Al-Si-Cu alloys with the composition exceeding 0.60% Fe,0.50% Mn and 8%Si,sludge in the form of Al(Fe,Mn)Si formed at temperatures in the range 610-660 °C.As the melt of a 7.5%Si,1.2%Fe,3.53%Cu,0.60%Mn,0%Cr alloy was held at a selected temperature for 50 min,the area percentage of Al(Fe,Mn)Si type sludge formed at a location 14 cm from the melt surface was approximately 8.2%,2.4%,and 1.1% for holding temperatures 630 °C,640 °C,and 660 °C,respectively.In addition,at 640 °C,the area percent of the sludge increased significantl when the Cr content of the alloy was increased from 0% to 0.2%.Flores et al.[9] also reported that the average size of the sludge particles depended on the holding temperature and time.At 640 °C,the average size of the sludge particles was over 40 μm after holding for one to two minutes,while at 630 °C and 660 °C,it was about 5 μm for the same holding time.These finding suggest that the Al(Fe,Mn)Si type sludge forms most readily at about 640 °C.

Makhlouf and Apelian[10]investigated the effects of holding temperatures and chemical composition of fi e hypoeutectic and hypereutectic Al-Si experimental alloys with different contents of alloying and impurity elements,Si,Cu,Mg,Fe,Mn and Cr on sludge formation,in comparison with the commercial die casting alloy A380 (Table 1) .The SF values of the tested alloys are listed in Table 2.The experiment alloys were prepared and melted in an electric resistance furnace at 850 °C and held for 30 min from pure aluminum and Al-Si,Al-Fe,Al-Mg,master alloys.To evaluate the effect of holding temperatures,the melt temperature then was lowered to 720 °C or 670 °C for holding 3 hrs.Upon the completion of holding,the melt was moved out of the furnace and solidifie quiescently in air for slow cooling,or was poured into a copper wedge mold for fast cooling.

Table 1 Chemical composition of Al-Si-Cu alloys tested by Makhlouf and Apelian [10].

Table 2 Sludge factors for the alloys tested by Makhlouf and Apelian [10].

Microstructure examination of the samples either slowly or fast cooled to resemble die-casing showed that the holding temperature (670 °C vs.720 °C) appeared little influenc on sludge formation in the experimental alloys with the SF ranging from 1.32 to 2.51.However,a few small star-like Ferich particles were found in samples from alloy A380 meltsthat were held at 670 °C for 3 h.This phase was not detected in A380 melts that were held at 720 °C for 3 h.The particles were small and might not contribute to sludge.This phenomenon became more important when the cooling rate became lower.It was also observed that,for fast cooling in a copper wedge mold,more and larger star-like Fe-rich particles were observed in the casting whose melt was held at 670 °C than in the casting whose melt was held at 720 °C,as shown in Figs.5 and 6.It was explained that ingots and other charge material used to produce A380 alloy might contain intermetallic particles of high melting temperatures,which could dissolve in the melt at the high holding temperature,but not completely dissolve at the lower temperature.These intermetallic particles and their residues could act as nuclei for the sludge phases.

Fig.5.Optical microstructure of alloy A380,melt was held at 670 °C for 3 hrs,cast in a copper wedge mold,at wall thickness of 0.36′′ (fast cooling).T-Star-like,B-Blocky particle [10].

2.5.Effect of cooling rate on sludge formation

Fig.6.Optical microstructure of alloy A380,melt was held at 720 °C for 3 hrs,cast in a copper wedge mold,at wall thickness of 0.36′′ (fast cooling).T-Star-like,B-Blocky particle [10].

In the open literature,various opinions about the role of cooling rate in sludge formation are present.Most studies indicated that the morphology of the sludge is significantl affected by the cooling rate.The study by Ghomashchi[11]and Gustafsson et al.[12] showed that the polyhedral and Chinese script morphologies of the Al(Fe,Mn)Si type sludge were independent of cooling rate.Jorstad [6] failed to mention the cooling condition used to develop the SF formula (Eg.1)

To reveal cooling rates (slow,medium,and fast) playing an important role in determining the amount,size and morphology of sludge,fi e hypoeutectic and hypereutectic Al-Si experimental alloys with different contents of alloying and impurity elements,Si,Cu,Mg,Fe,Mn and Cr and the high(2.54)and low (1.27) SFs on sludge formation,in comparison with the commercial die casting alloy A380 (SF=1.7) were studied by Makhlouf and Apelian [10].They found that,as the melt was cooled slowly in the crucible after holding,the sludge formed in the alloys with both the high and low SFs in the form of large Chinese script,and polyhedral and blocky particles (Figs.7 and 8).The size of the sludge particles and the volume fraction of sludge decreased as the cooling rate increased.In the alloys with the low SFs,the Fe-rich phases form in the interdendritic regions,and they became so small which made no contribution to sludge.When the fast cooling was applied,almost no sludge was observed in the alloys.For alloy A380,it appeared that the holding temperatures had more influenc on sludge formation than the cooling rate.More and larger sludge particles were present in the cast alloys,which were solidifie under both the slow and fast cooling conditions after holding at a relatively low temperature (670 °C) (Fig.5).There were fewer sludge particles in alloy A380 solidifie after holding at a relatively high temperature (720 °C) as shown in Fig.6.

Fig.7.Fe-rich needle and polyhedral particle surrounded by the primary Si in alloy#3(slow cooling).S-Primary Si,P-Polyhedral,N-Needle(or Platelet)[10].

Fig.8.Microstructure of alloy #1,melt was held at 670 °C for 3 hrs,cast in a copper mold,at wall thickness of 4.3 mm (fast cooling).T-Star-like,N-Needle (or Platelet),S-Primary Si [10].

3.Control,characterization and assessment of sludge in die cast Mg alloys

3.1.Metallurgical principles for control of sludge in ingot production and die casting of Mg alloys

Presently,the most popular automotive magnesium alloys,such as AM60B and AZ91D,are based on the Mg-Al-Mn-Zn system,and contain primaryα-Mg and eutecticα-Mg/β-Mg17Al12as the major constituent phases [1,2].The corrosion resistance of the automotive Mg alloys depends signifi cantly on the iron (Fe) content.In the absence of manganese(Mn),Fe and Al can form precipitates,which act as effective micro-cathodes in the primaryα-Mg matrix to form galvanic corrosion.To maximize corrosion resistance,0.3 wt.% Mn,which could vary among commercial Mg alloys,is usually added to wrap the impurity Fe in manganese aluminides,a particle of iron embedded in a particle of manganese aluminide is less detrimental to magnesium because the galvanic activity between manganese aluminide andα-Mg is less than that between Mg and Fe [13].The beneficia effect of Mn in reducing the Fe content needs to be counterbalanced by an increased quantity of Al-Mn(-Fe) intermetallic compounds in the melt during die casting practice.With the normal level of Mn addition,the intermetallic compounds form at a temperature above theα-Mg liquidus temperature,and precipitate by gravitational sedimentation to the bottom of a crucible holding the alloy melt.However,excessive usage of Mn and improper temperature control might generate a large amount of sludge in the crucible,which could cause a deleterious effect on die casting operation and product quality [4,5].

Thorvaldsen et al.[14] investigated the generation of sludge and dross during the melting and handling of Mg alloys,AZ91D,AM60 and AS41,in Mg die casting operations.The two main by-products of the melting and holding process were sludge and dross,sludge settled to the bottom of crucibles,while dross floate on the melt surface.The results revealed that the amount of sludge removed from the melting crucibles varied between 2 and 4% of the total metal input.There were three major constituents in the tested sludge,which were oxides (29% ± 14%),intermetallics (0.8% ±0.8%) and entrapped metal (70% ± 14%).The oxide content was affected by process parameters,the agitation of melt surface,the melt temperature,and gas protection system,rather than to the low oxide level in the primary ingots.The low agitation level of the melt surface could reduce the extent of oxidation by careful charge of ingots into the crucible and using automatic transfer systems instead of manual discharging.The low melt temperature near the liquidus temperature reduced the oxidation rate of molten Mg alloys.When applying the gas-protection system,the balance between the melt loss due to oxidation and the cost of the protective gasses should be taken into consideration.Intermetallic particles (AlxMny)originated from the primary ingots played an important role as precipitation sites for Fe in lowering the level of dissolved Fe and increasing the overall corrosion resistance of the castings.It was suggested that a reduction in the sludge formation by a factor of 10 might be achievable,while the oxidation of molten Mg alloys in the crucible was reduced and the Mn content of the primary ingot was kept at a low level.

To establish the optimum level of Mn in the Mg alloys AZ91,AM60 and AM50(Table 3),Holta et al.[15]and Holta and Westengen [16] determined the temperature-dependent mutual solubility of Fe and Mn for each alloy as shown in Fig.9,which gave the production routes of the primary ingots for various Mn additions.Each curve in Fig.9 represented a specifi manganese addition and each data point gave the measured equilibrium contents of Fe and Mn.By fittin the data,the polynomial equations were determined,which were used to attain the desired iron by controlling the Mn con-tent and the holding temperature of the melt.The empirical equations for AZ91,AM60 and AM50 are given below [15,16]:

Table 3 Chemical compositions of ASTM B93 and B94 standards for automotive Mg alloys AZ91D,AM60B and AM50A for ingots and die castings,respectively[17,18].

Table 4 Die casting process,Mg alloys and casting temperatures for sludge generation[19].

where T is the casting temperature (°C),and the element contents,[%Fe],[%Mn] and [%Al] are in weight percentage.

Also,it can be seen from Fig.9 that the specifie maximum limit of Fe in the ingots could be fulfille for different combination of casting temperatures and Mn additions.For instance,at a casting temperature of 660 °C,the addition of slightly more than 0.3 wt.% Mn produced an alloy of AZ91 with approximately 0.2 wt.% Mn and 0.004 wt.% Fe as indicated by point A in Fig.9(a).If the casting temperature of the ingots increased to 680 °C,and the Fe (0.004 wt.%)content in the ingot remained,the addition of Mn needed to be raised to 0.6 wt.%,i.e.,twice as much as that for ingot production at 660 °C,as illustrated by point B in Fig.9(a).

The Mn solubility depended also on the aluminum content of the alloys (AZ91,AM60 and AM50) as illustrated in Fig.10.At a fi ed casting temperature,the solubility of Mn increased with decreasing aluminum content.In the real casting operation,the high casting temperature is employed for the alloys low in Al content (AM60 and AM50) due to their high liquidus temperatures.The high Mn content is necessary to be used in these AM alloys.

By combining the phase identificatio data of AZ91 and AM series obtained by the X-ray and electron microscopy techniques,Holta et al.[15] established the ternary Mg-Al-Mn phase diagram (Fig.11) with an assumption of negligible Zn effect.Fig.11 shows the Mg-rich corners of Mg-Al-Mn phase diagrams at temperatures of (a) 660 °C and (b) 700 °C.The strong influenc of 0.005 wt.% Fe in the melt on phase development at the two temperatures was demonstrated.

To minimize sludge formation and improve corrosion resistance,Holta et al.[15] provided the basic metallurgical principles for the production of the primary ingots and the die cast parts.The provided principles are listed below.

■The maximum iron content of the ingots should be lower than the limit for die casting (0.004 wt.%) to allow for minor Fe pick-up during re-melting in the die casting operation;

■The maximum iron content of the finishe die cast parts should not exceed 0.005 wt.% to ensure acceptable corrosion performance of the parts;

■The minimum Mn content of the ingots should reflec the liquid solubility of the alloy at the anticipated minimum casting temperature.If high casting temperatures were used,the Mn content should be raised accordingly;

■The minimum Mn content of the ingots should be low to ensure freedom in selection of low die casting temperatures.

Fig.9.Temperature-dependent mutual solubilities of Fe and Mn in Mg alloys(a) AZ91,(b) AM60 and (c) AM50 [15].

Fig.10.Solubility of Mn in Mg-Al alloys,AZ91,AM60 and AM50 [16].

Fig.11.Magnesium-rich corner of the Mg-Al-Mn phase diagram,(a) 660 °C and (b) 700 °C [15].

After characterizing the die-casting Mg sludge from different companies,Corby et al.[19] indicated that it was necessary to have a residual quantity of Mn present in the primary ingot to protect the melt from iron pick-up upon re-melting in die casting shops.Ideally the melt should remain just under the Mn saturation point,so that any significan pick-up of iron would precipitate out of the melt as an intermetallic.In other words,the relationship between the primary producer’s ingot pouring temperature (Tp) and the die-casters furnace temperature (Td) had an impact on the amount of sludge generated in the die casting operation.If the ingots were re-melted at a lower furnace temperature (Td＜Tp),this would cause a solubility difference and force the precipitation of intermetallics,which formed as sludge.If the furnace temperature is higher,(Td＞Tp),then the metal was left susceptible to iron-pick up.The ideal situation to protect the melt from iron pickup and minimize intermetallic sludge was to keep the melt temperature consistently just below Tp[19].Practically,this was difficul to achieve when there were constant temperature fluctu ations caused by charging ingots.Temperature cycling within the furnace during normal operation could cause both problems (Td＜Tpand Td＞Tp) to occur.As new ingots were added,the drop in temperature resulted in the melt generating intermetallic particles.However,as the temperature returned to normal,the melt was above the Mn saturation point and susceptible to iron pick-up.This might explain why some particles in AZ91D handled at a high holding temperature,compared to those in AM60 and AM50 holding at low temperatures,had high levels of iron,when the majority were generally low.Alternatively,it could be due to iron variations within the melt [19].

Fig.12.Typical sludge as seen (a) and (b) in AZ91D,(c) AM60B,and (d) AM50A [19].

3.2.Characterization of Mg sludge

3.2.1.Chemical compositions of Mg die-casting alloys and sludge

Corby et al.[19] analyzed the sludge collected in two North American companies,which was taken from the bottom of a 400 series stainless steel standard size crucible and heated by gas.The process,Mg alloys and casting temperatures employed in the die-casting process are listed in Table 4.The top shape of the furnace is approximately an oblong areawith a total length of 2300 mm,which has a 420 mm radius at each end.The sidewalls are tapered at approximately 5°.The depth of the crucible is approximately 610 mm.The sludge from Company A (A) had a residence time of 2-4 h prior to cleaning and Company B (B) had a residence time of 6-8 h.

Table 5 ICP results for melt compositions (wt.%) taken from spectrometer discs [19].

Tables 5 and 6 list the chemical compositions of the die casting alloys and the corresponding sludges,which were analyzed using Inductively Coupled Plasma-Atomic Emission Spectroscopy (ICP-AES).The chemical analysis results indicate that all of the furnaces had the alloys within the ASTM specificatio except AM60B,which was high in Fe.Other than Mg,the main elements in the metallic portion of the sludge were Al and Mn.With the exception of AM60B,Femade up a very low percentage of the melt composition.Mn made up a much larger proportion of the AM50A sludge compared to AZ91D and AM60,because of a higher Mn content in the melt or a lower furnace temperature,encouraging the precipitation of Mn rich particles.

Table 6 Metallic sludge composition [19].

3.2.2.Characterization of die-casting sludge

3.2.2.1.Morphology and sizes of intermetallics in die-casting sludge.The characteristics of the die-casting sludge were analyzed using a Scanning Electron Microscope (SEM)equipped with Electron Dispersive Spectroscopy (EDS) by Corby et al.[19].

Initial investigation through SEM showed the majority of the sludge was entrapped magnesium,but it also contained some magnesium oxide (black) and Al-Mn-Fe intermetallic(white) as shown in Fig.12.The intermetallic particles tended to form in clusters,as demonstrated in Fig.12(a) and by the lack of them in Fig.12(b),but they were similar in appearance regardless of the melt composition as seen in Fig.12(c).Many of the particles display faceted edges,as illustrated in Fig.12(d).Particles ranged in size around 5-400 μm,however the majority of the particles were below 50 μm.They also found other impurity in the AZ91D that were Al-Fe-Cr,Mg,O and Cl.Al-Fe-Cr phase was similar to that observed with interactions between the mild steel of the crucible and molten Mg-Al alloys.The Mg,O and Cl may be leftover from MgCl2flu to extinguish the burning dross [19].Fig.12(d)shows the largest particle appeared in the die cast AM50A alloy.The particle coarsening could result from the high Mn content in the alloy melt (Table 5),and/or the low furnace temperature (Table 4) promoting the precipitation of Mn-rich particles.

Table 7 Fe contents of intermetallic particles in sludge [18].

3.2.2.2.Chemical composition of intermetallics in die-casting sludge.The composition of the particles determined using a JOEL electron probe is shown in Fig.13.The results were generally within the atomic proportions for an Al8(Mn,Fe)5phase,assuming that Fe does not greatly affect the stability of Al8Mn5in the Al-Mn system,although the compositions of the particles [19] varied more than that previously reported by Holta et al.[15].The possible compositions forβ-Mn are marked,as is the composition forα-AlMnFe reportedly found in AM50 [15].

Fig.13.Probe results indicating that the majority of the particles could be Al8(MnFe)5 [15,19].

Fig.14.EDS mapping of an AZ91D intermetallic.Elements appeared evenly distributed [19].

The outlying particles with higher Mn from the AM50A alloy were all greater than 100 μm in size,although their morphology under SEM was indistinguishable from other particles,which could be a different type of Al-Mn-Fe phase.

The Fe compositions of the particles are given in Table 7.The Fe contents in the AM50A and AM60B alloys were generally lower than that found in the AZ91D samples.This was probably due to the lower furnace temperatures,which reduced the Fe pick-up from the steel of the crucible.

Table 8 Oxygen and MgO contents in primary AZ and AM ingots [35].

Table 9 Contents of MgO and Al-Mn intermetallic compounds in recycled AZ91 Ingots [35].

EDS mapping of an AZ91D intermetallic particle in Fig.14 suggested that they were homogeneous in composition.It was difficul to distinguish between theα-Fe andβ-Fe for low concentrations of Fe as the peak overlap took place.

3.3.Al-Mn(-Fe) intermetallic formation and thermodynamic assessment

3.3.1.Al-Mn(-Fe) intermetallic formation and interaction with oxide

It was pointed out [14,15,19] that the composition of the sludge was roughly 60-80%metallic and 20-40%oxides,and intermetallics were estimated to be a low proportion,typically around 0.8 ± 0.8%.According to Holta [15],particles vary in Fe and Mn composition and are typically Al8(Mn,Fe)5in AM60 and AZ91,andα-AlMnFe in AM50.To minimize the build-up of die-casting sludge and prevent large Al-Mn(-Fe)clusters from entering castings,Peng et al.[4] investigated mechanisms of Fe pick-up,and settling of Al-Mn(-Fe) particles in association with oxide films

Two forms of casting experiments with alloy AZ91 were performed

(1) to mimic Fe pick-up in industrial melts,2 g of AZ91 was melted and held at 700 °C for 4 h in uncoated Fe-0.2%C cylindrical crucibles with dimensions of inner diameter of 12 mm and inner height of 18 mm,within sealed quartz tubes backfille with Ar;

(2) a similar procedure involving an Al2O3crucible in quartz tubes was used to generate equivalent microstructures without Fe pick-up.

The casting samples were then solidifie by placing the 700 °C quartz tubes in the vertical cylindrical hole of a steel mold at room temperature with a cooling rate of 4 K/s in the range 700-650 °C.

In real-time radiography experiments,the AZ91 specimen and cell were heated and melted at a constant rate of 0.5 K/s and subsequently cooled at a constant rate of either or 0.083 or 0.5 K/s.During heating and cooling,transmitted x-ray images were recorded at a rate of 2.5 frames per second.

They found that Fe pick-up from mild steel crucibles held at 700 °C caused the formation of a B2-Al(Mn,Fe) compound,resulting in two-phase Mn-bearing intermetallic particles consisting of a B2 core and a D810-Al8Mn5shell.Zeng et al.[5] studied the nucleation and growth crystallography of Al8Mn5on B2-Al(Mn,Fe) in Mg alloy AZ91,and indicated that Al8Mn5nucleated on B2-Al(Mn,Fe) particles and an incomplete peritectic transformation resulted in a Fe-rich B2-Al(Mn,Fe) core enveloped by a low-Fe Al8Mn5shell.

Fig.15.Typical cross-sections of primary Al-Mn(-Fe)particles in AZ91 after 4 h isothermal holding at 700 °C in (a) an Al2O3 crucible and (b) a mild steel crucible.In (a) and (b),the top are BSE-SEM images,the middle are corresponding EBSD phase maps and the bottom are IPF-X maps.(c) Pole figure for two families of planes from the sample in (a) showing cyclic twinning [4].

At low Fe content (＜10 ppm),the particles were mostly D810-Al8Mn5as shown in Fig.15.For both low-Fe and high-Fe AZ91,primary Al8Mn5particles were cyclic twinned and contained up to four Al8Mn5orientations similar to Ref.5 (Fig.15).The particles had equiaxed polyhedral morphology with multiple facets and often contained internal liquid channels.In the sludge prepared in the Al2O3crucible containing entrained oxide,the attachment of Al8Mn5particles to the oxides was revealed by SEM.The direct observation on Al8Mn5particle settling and sludge formation by the real-time synchrotron x-ray radiography confirme that Al8Mn5particles appeared to nucleate on entrained oxides(Fig.16).After numerous Al8Mn5particles became trapped and/or nucleated on entrained oxides,they continue to grow on cooling,leading to a large cluster of intermetallics.Also,the entrained oxides acted as filter to Al8Mn5particles,trapping them as they settle.The settling data were in reasonable agreement with Stokes’ law once correction factors for the thin sample geometry,the non-spherical particles,and the internal liquid channels are accounted for.

3.3.2.Thermodynamic assessment of Mg-Al-Mn system

Fig.16.Typical 3D morphologies of primary Al8Mn5 particles nucleating on the oxide in AZ91 [4].

To thermodynamically understand the formation of Al-Mn(-Fe) intermetallics in Mg Alloys AZ91,AM60 and AM50,the phase equilibria and solidificatio process for Mgrich Mg-Al-Mn-Zn alloys were analyzed based on a combination of computational thermochemistry and thermal analysis with differential scanning calorimetry (DSC) measurements[20,21].They found that the primary precipitate was Al8Mn5intermetallic phase at high Mn compositions (＞0.2 wt.%)in the Mg-rich Mg-Al-Mn-Zn alloys as shown in Fig.17.The computed results and DSC analysis indicated that,during the solidificatio of AZ91 alloy,the precipitation of the Al8Mn5intermetallic phase took place at 642 °C.That was much higher than the temperature of 600 °C at which the initial formation ofα-Mg solid solution occurred.

Shukla and Pelton [22] assessed thermodynamically the Mg-Al-Mn system with the FactSage thermochemical software.The solubility of Mn and the stability of Al8Mn5precipitates in the Mg-Al alloys in the temperature range of 650-750 °C were predicted.With the Mn content of 0.4 wt.%,the precipitation of the Al8Mn5intermetallic phase in the Mg-Al-Mn alloys containing 5.0 and 6.0 wt.% would begin at 660 and 670 °C,respectively.Meanwhile,the Al8Mn5intermetallic particles could precipitate at a high temperature of 700 °C in the Mg-9.0 wt.% Al-0.4 wt.% Mn alloy,which was similar to AZ91 alloy.Ye and Liu [23] studied the in situ formation behaviors of Al8Mn5particles in Mg-Al-Mn alloys by introducing 10 wt.% Al and 2.5 wt.% Mn into AM60 alloy in a steel crucible inside an induction vacuum furnace for 60 min at 750,800,and 850 °C,respectively.It was that,with the absence of oxygen,Al8Mn5was a stable phase in the Mg-Al-Mn system,and was formed in the liquid phase (even at 850 °C) during the melt processing.Unlike many compounds formed in liquid phase,the Al8Mn5particles resisted coarsening in the melt at high temperature even at a very high concentration.An increase in Mn content and/or melting temperature promoted the in situ formation of Al8Mn5particles in the Mg-Al-Mn alloys.Fig.18 shows the massive presence of the Al8Mn5particles in the Mg-10 wt.%Al-2.5 wt.%Mn alloy processed at 850 °C.

3.4.Rapid assessment of Mg oxide and intermetallic in Mg alloys

As the magnesium oxide served as the nucleation site for the Al-Mn(-Fe) intermetallics and played as major role (20-40%) in sludge,the techniques for the assessment of both the non-metallic inclusions (oxides) and metallic inclusions(intermetallics) were essential for controlling and minimizing sludge formation [24-36].The common techniques could be classifie into four main groups:spectroscopy,fracture test,infiltration and hybrid.The firs three techniques were focused on the detection of non-metallic inclusions,while the last one was used to the simultaneous measurements of both the non-metallic inclusions (oxides) and metallic inclusions(intermetallics).

3.4.1.Spectroscopic technique

Fast neutron activation analysis (FNAA),glow discharge mass spectroscopy(GD-MS)and glow discharge atomic emission spectroscopy (GD-AES) are considered the methods of physical measurements for the determination of the chemical composition of Mg melts [24,25].The FNAA method employed the samples were irradiated with 14.8 MeV neutrons and the reaction product (16N) was detected.The number of16N atoms detected directly corresponded to the number of oxygen atoms in the sample [25].In glow discharge source techniques,the sample was exposed to an argon plasma which uniformly eroded material from the sample surface.The sputtered atoms were ionized in this plasma and extracted into a mass spectrometer for separation and detection in GD-MS.In the case of GD-AES,a spectrometer was used to measure the wavelength and intensity of the light emitted by the sputtered atoms when they returned to the ground electronic state.FNAA,GD-MS and GD-AES had low detection limits (0.1-10 ppm) and high accuracy (5-20%).But,they all involved extensive sample preparation,and required sophisticated and expensive instrumentation.At present time,the FNAA method is the only commonly accepted method for oxide (oxygen)measurement by the magnesium industry [26-28].However,the FNAA method needed a neutron source such as a nuclear reactor,which resulted in the accessibility in question.Furthermore,these methods provided little information on the oxide size and morphology [24,25].

Fig.18.Optical micrographs showing massive presence of Al8Mn5 particles in Mg-10 wt.% Al-2.5 wt.%Mn alloy processed at 850 °C,(a) low and (b) high magnification [23].

Fig.19.K-mold method [24,29].

3.4.2.Fracture test

Fracture tests have long been recognized and are extensively employed as an inexpensive and rapid shop-floo means to evaluate the melt cleanness in foundries.It could be subdivided into two classes:K-Mold and light reflectance

3.4.2.1.K-Mold.The K-Mold technique employed a fla plate with four notches cast into its cope surface (Fig.19) [24,29].These notches served as fracture points.The design of the knife edges in the mold enhanced the efficien y of capturing the oxides on the fracture faces through the effect of some eddy occurring during the mold filling In one test,a number of sampling plates were cast in preheated molds using the molten metal to be evaluated.The cast plates were then fractured immediately by operators.The fracture surfaces were examined visually for oxides by naked eyes.The inclusion level was expressed as the number of defects seen per number of fracture surfaces examined.

The main advantages of the K-mold were result quickness,very low cost,simple preparation of samples,and high sampling fl xibility.But,the technique was less quantitative and gave inaccurate results in comparison with the other techniques.Difficultie were encountered in detecting small oxide inclusions (＜100 μm) and in assessing molten alloy in which the level of inclusions was somewhat lower [24,29].

3.4.2.2.Light reflectanc .In generally,the magnesium oxides were the most common inclusions present in magnesium alloys,and appeared darker than the relatively bright alloy metal as revealed by the optical microscope (Fig.20).

Instead of using the unaided eye,an optical technique,i.e.,a brightimeter,was developed based on the differences in optical characteristics between Mg and magnesium oxide for the evaluation of the fracture surface [30,31].In this optical technique,a conical cast sample was fractured and the fracture surface was placed in the aperture of a brightimeter.The sample was illuminated at a 45° angle and the intensity of the reflecte light is measured (Fig.21).If oxide inclusions were present in the material,the incident light was scattered at the surface of the specimen due to multiple reflection and refractions,which was called diffuse reflectance In Mg alloys,the oxide inclusions absorbed more light than the matrix,and consequently the reflectanc of the specimen was reduced.As such,the content of the Mg oxide was correlated to light reflectance However,this technique exhibited high detection limits (～2%).Its accuracy and consistency in detecting the oxide content of the melt and its capability of identifying other types of inclusions were questionable.

3.4.3.Filtration

The hydro magnesium inclusion assessment method(HMIAM),an inclusion assessment systems that was developed by Norsk Hydro for molten magnesium in the early 1990s was a typical application of the vacuum filtratio technique [32].The application of this technique to molten magnesium was well demonstrated.Fig.22 illustrates a HMIAM vacuum filterin system which consisted of a filte,a filte cup,a tapered plug and a vacuum container (steel tube) [33,34].In this technique,molten metal was drawn through a sampling filte by vacuum behind the filte.Several procedures were involved in an entire test.First,a tapered plug was used to protect the filte from contamination before immersion.Next,the entire unit was immersed and kept in molten metal bath for a period of time,which allowed the unit to be preheated.Then,the tapered plug was pulled out and vacuum starts to suck the molten metal through the filte.After sampling a certain amount of melt,the unit was lifted out of the bath for cooling.Upon the solidificatio and cooling of the unit,the filte was removed from the filte cup and sectioned along the diameter perpendicular to the filte surface.The oxide inclusion and intermetallic particle concentrations were determined by the volume of particles per unit weight of metal drawn through the filte.

Fig.20.Optical micrographs showing as-polished (a) a lacy oxide network in cast AM60,and (b) a snaky oxide fil associated with a surface defect in cast AM60 [25].

Fig.21.Schematic diagram showing the principle of a brightimeter [30,31].

Fig.22.Norsk Hydro’s magnesium inclusion assessment method (HMIAM)[24,32-34].

This technique had some advantages.Filtration could be carried out directly in the bulk molten metal at any designated location.A relatively large volume of the melt was filtere in a test,which improved the accuracy of inclusion assessment.However,the shortcomings of this technique were acknowledged in practice.Upon completion of sucking,back fl w ofmolten metal from the container to the filte cup might take place if the metal head was not appropriately balanced in the container by a corresponding vacuum.Also,after lifting the unit out of the molten metal bath,if the bottom of the fil ter cup was not chilled first the inclusions in the filte cup would be drawn through the filte during the solidificatio of the melt.The leakage might occur during the sampling if any one of joints is not sealed tightly.In addition,the quantity and morphology of oxides in ingots or die cast components might be different from those observed on the filte due to different solidificatio conditions (e.g.melt temperature) [24].

3.4.4.Hybrid

Bronfi et al.[35,36] developed a hybrid assessment method,named‘‘MagOxide’’,which combined the wet chemistry with Inductively Coupled Plasma Emission Spectrometry(ICP).The hybrid technique aimed at evaluating simultaneously the MgO and Al-Mn-Fe intermetallics in the primary and recycled Mg alloy ingot.The experimental procedures used in this technique included:

1 theα-Mg-matrix and eutectic phases such asβ-Mg17Al12phase of 2-3 g sample were dissolved in a mixture of organic solutions.But,the MgO particles and Al-Mn(-Fe)intermetallics remained in soluble in liquid solution.The dissolution process was conducted in Argon inert atmo-sphere at 30-40 °C to eliminate the reaction between organic solvents and water.

2 The solution was filtere and residues were rinsed with boric acid and distilled water.

3 The MgO and Al-Mn(-Fe)residues were dissolved in warm choric acid and filtered The filtrat is then analyzed by ICP.

The Mg concentration obtained by ICP was converted to the concentration of MgO or oxygen assuming that the magnesium remained after dissolving in organic solution was only bound to oxygen.The technique also allowed to quantify a weight fraction of Al-Mn-Fe intermetallics and their phase composition.

Bronfi et al.[35] used this hybrid technique to analyze both the primary and recycled Mg ingots of AZ and AM series produced by different companies.The results showed the acceptable correlation in the MgO contents in the primary ingot detected by the FNAA and the MagOxide techniques(Table 8).The technique was capable of providing simultaneously the measurements of the MgO and Al-Mn intermetallic compounds in the AZ91 ingots recycled by the different companies,as given in Table 9.

The MagOxide method was considered as an inexpensive and reliable technique for cleanliness evaluation of primary and recycled Mg ingots.However,due to the fact that,in this method,the small samples (2-3 g) were used in each test,the correct sampling and good statistics were needed.The technique required the tedious analytical procedure,and the massive sampling and analyses.

Summary

The establishment of the SF formula makes the control of sludge generation in casting of aluminum alloys easily achievable.The fully established sludge factors for aluminum alloys showed that the sludge formation was influence by not only the chemical composition but also process parameters such as holding temperatures and times.The idea and approach for producing Al sludge factors can be implemented in creating sludge factors for magnesium alloys.

Understanding of sludge formation in die casting of magnesium alloys is still immature,despite the fact that a number of studies have been reported on proper control of sludge generation in the primary ingot production.Information on the effects of melt holding and casting temperatures on sludge formation for die casting processes is inadequate,as the available relationships between the alloy chemistry,i.e.,iron and manganese contents,and the casting temperatures are only applied to the primary Mg ingot production.There is a lack of general knowledge about the kinetics of Mg sludge formation during casting of Mg alloys.Therefore,it is essential to investigate systematically and methodically the simultaneous effects of casting and holding times and temperatures as well as the alloy chemistry on the sludge formation in die casting of magnesium alloys.

Characterization of Mg die casting sludge has been focused only on the identificatio of phase constituents such as Al-Mn(-Fe) intermetallics and Mg oxides.Interaction between the intermetallic phase and oxide should be explored in practical die casting processes,because the previous investigation was performed only in a lab environment and on one alloy AZ91.Systematic work on nucleation,growth,and interaction of intermetallics with oxides in other automotive alloys such as AM60 and AM50 should be explored.Given the important role of oxides in the Mg sludge formation,the hybrid assessment techniques capable of simultaneously detecting different phase constituents in the Mg sludge need to be developed.

Acknowledgments

The authors would like to thank the Meridian Lightweight Technologies Inc.,Strathroy,Ontario Canada,and the University of Windsor,Windsor,Ontario,Canada for supporting this work.This work is part of a large project funded by Meridian Lightweight Technologies,Inc.