Precipitation of Iron-Containing Intermetallic Phases from Aluminum Alloys by Metal Melt Filtration

Abstract

Iron (Fe) provides a non-reactive dissolved impurity in aluminum (Al) alloys, which forms a coarse, plate-shaped intermetallic β-phase during solidification. This β-phase is detrimental to the mechanical and casting properties. Therefore, the reduction of Fe by binding in Fe-containing intermetal-lics (sludge phase) was realized via a two-stage procedure, which consisted of conditioning of the melt by manganese (Mn) and chromium (Cr) with subsequent-ly applied metal melt filtration. For this purpose, the formation characteristics of the Fe-rich intermetallic phases were investigated regarding the temperature, time, and initial chemical composition to separate these intermetallics from the residual melt. To evaluate the different process parameters of Fe removal for a potential implementation in lightweight metal foundries, a process technology on an indus-trial scale was developed in cooperation with an industrial partner. The examina-tion of samples in optical microscopy (OM) using image analysis were conducted to determine the area fractions of Fe-rich intermetallics. In addition, optical emis-sion spectrometer (OES) measurements were performed. Complementary investi-gations were achieved by scanning electron microscopy (SEM), with energy dis-persive spectroscopy (EDS), and electron backscatter diffraction (EBSD) to measure the partial chemical composition and for phase identification. The for-mation characteristics of the Fe-containing phases were investigated using DSC cooling curves and selective sampling in quenching experiments. In the experi-mental trials, a maximum reduction of iron of ≈50% was revealed compared to the unfiltered sample, whereby greater influence on the formation of α-intermetallics was inferred by temperature than by time. Moreover, the elements Mn and Cr were reduced by about 66% and 86% at 620 °C, respectively, thus, the element contents in the filtered samples approached the chemical composition of the standard alloy (EN-AC-AlSi9Cu3(Fe)).

31.1 Introduction

Iron (Fe) is considered as an accompanying impurity in aluminum (Al) casting alloys. However, due to the formation of the pre-eutectic, Fe-containing β-AlFeSi phase, which exhibits a plate-like morphology, its presence is detrimental to the mechanical and casting properties [1, 2]. Since an increasing Fe content in the melt resulted in the diffusion-related formation of coarse plates, the effective feeding with residual melt is hindered [3, 4]. In addition, shrinkage porosity and notching stresses are increased, which significantly reduce the ductility of the castings as well [4,5,6]. Fe impurities become particularly enriched in secondary Al alloys due to insufficient scrap separation in the global aluminum recycling process [7,8,9]. The usage of secondary Al alloys for castings is much more beneficial from an economic and environmental aspect than the production of primary aluminum. Compared to the primary production (in the fused-salt electrolysis process), secondary alloys require only 5% of the energy and CO₂ emissions [10]. Moreover, this contributes to the energy-intensive electrolytic reduction of Al₂O₃. Currently, the reduction of Fe in secondary Al alloys is performed by dilution with high-cost pure aluminum (≥ 50%). Therefore, investigations are conducted to reduce the Fe content in a commercial secondary AlSi9Cu3 alloy using a two-step procedure.

First, Fe-rich intermetallics are generated in the melt using melt conditioning by adding alloying elements such as manganese (Mn), chromium (Cr), nickel (Ni), beryllium (Be), cobalt (Co), etc., resulting in a modification of the morphology into a compact and filterable α-intermetallic (or also called sludge phase) [9, 11,12,13]. With increasing the initial Fe, Mn, and Cr content, the formation temperature of the Fe-rich intermetallics shifts to higher temperatures, forming a primary phase, i.e., before the formation of the dendrite network (α-Al solid solution) [14,15,16,17]. Consequently, the melt temperature must be set below the formation temperature of the Fe-rich intermetallic phase and above the formation temperature of Al-dendrites in order to obtain the Fe-containing intermetallics. As a result, the residual melt exhibits an Fe-reduced content. In the second step, the intermetallic phases can be finally separated by an appropriate filtration technique using multifunctional and/or specially coated ceramic foam filters (CFF), which are developed in cooperation with other subprojects in the Collaborative Research Center 920 (CRC 920).

In order to implement the two-step procedure for a potential filtration technology, the influences on the formation of the Fe-rich intermetallics has to be investigated. In particular, the formation characteristics of intermetallics depend on the temperature, time, cooling rate, and initial chemical composition [18,19,20,21]. For this purpose, differential scanning calorimetry (DSC) cooling curves are analyzed to examine the different influences (e.g., temperature, chemical composition, etc.) on the formation of intermetallic phases. Based on these insights, the operating range can be defined to initiate the filtration process. Furthermore, the potential for Fe removal is to be evaluated via experimental filtration trials using a specially developed laboratory filtration apparatus. The efficiency of Fe removal is determined using variations in time and temperature. Additionally, the influence of multifunctional filter coatings and materials is also considered. Besides different filter porosities (20, 30 pores per inch - ppi), foam ceramic filters (CFF) made of alumina, C-bonded alumina filters, and alumina CFF with rough alumina coating are used. Finally, the transfer project T03 will implement an appropriate filtration process technology for an industrial application in the light metal foundry. Hence, the setup and evaluation in industrial scale trials are presented. Industrially manufactured CFF as well as appropriate filter structures from the CRC 920 will be tested.

31.2 Experimental Section

31.2.1 Castings and Materials

Commercial secondary AlSi9Cu3(Fe) Al alloy (EN AC-46000, VAR 226 D) from Nemak (Dillingen, Germany), master alloys AlMn20¹ and AlCr20¹ (Technologica, Germany), and steel scrap (Schönheider Guss, Germany) were used to obtain various alloy compositions. Table 31.1 shows the element contents of the used base materials.

Table 31.1 Chemical compositions (weight%) of the base materials in the optical emission spectrometer (OES) SPECTROMAXx (Ametek, USA), where the element contents are within the tolerances of the standardizations (DIN EN 1706 and DIN EN 10268) [22]

Table 31.2 Chemical compositions (weight%) of various series of an AlSi9Cu3(Fe) alloy with weight fraction ratio of w_Mn/w_Fe and sludge factor (SF) according to Eq. 31.1 in optical emission spectrometer (OES) SPECTROMAXx (Ametek, USA) [22]

The different chemical compositions are constituted to investigate the formation of primary Fe-rich intermetallics (sludge phases). They are shown in Table 31.2. Their preparation was realized in an electric resistance furnace (Nabertherm, Germany) with a clay-bonded ISO graphite crucible size A 60 (Mammut-Wetro, Germany) and a capacity of ≈25 kg Al. The AlSi9Cu3(Fe) base alloy was melted first at 820 ℃. Then, the Fe, Mn, and Cr contents were adjusted by adding steel scrap (HC180P) and the master alloys (AlMn20 un AlCr20) by liquid metal transfer. Due to their higher melting temperature, the master alloys were externally melted to temperatures above 1000 ℃, as reported in previous studies [21, 22]. For each chemical composition, the chemical analysis was examined first. After that, samples were poured into a preheated metallic die, from which specimens were taken for metallography and DSC measurements. Finally, the prepared melt was transferred to a specially designed laboratory filtration apparatus for experimental filtration trials using a ladle with ≈3 kg capacity.

31.2.2 Analytical Methods

The chemical composition values correspond to the average of three individual sparks in the optical emission spectrometer (OES) SPECTROMAXx (Ametek, USA). The microstructural examinations were conducted on polished cross-sections prepared for the final stage with colloidal silica suspension OP-U using a semiautomatic grinding and polishing machine TegraPol-31 (Struers, Germany). For obtaining a better contrast of Fe-rich intermetallics, a hot sulfuric acid solution (80 ℃, 1 min, 30% H₂SO₄) was used, leading to a dark visualization of these phases compared to the less-affected silicon [22]. As a result, the area fractions of intermetallic phases were considerably easier to determine using image analysis software Stream Motion (Olympus, Japan). Based on a polyhedral and Chinese script-like morphology, detection examples using image analysis software and hot sulfuric acid etching are shown in Fig. 31.1.

Fig. 31.1

Illustration to determine the area fractions of Fe-rich intermetallic phases (representative of a polyhedral –left– and Chinese script-like morphology –right–) based on exemplary cross-section in light microscopy: a, b original polished micrographs, c, d after etching (80 ℃, 1 min, 30% H₂SO₄) to obtain a better contrast, and e, f detected Fe-rich intermetallics according to a gray scale value in red [22]

For experimental filtration trials, panoramic images were taken first on the polished cross-sections and then after etching by a Keyence VHX 2000 digital microscope (Keyence, Japan). Afterwards, the area fractions were obtained via the software Stream Motion. Supplementary microstructural investigations were performed by scanning electron microscope (SEM; VEGA3 W-REM, Tescan, Germany) using secondary electron (SE) and backscattered electron (BSE) images. Finally, measurements (mappings and spot analysis) for elemental distribution were transacted in SEM by energy-dispersive X-ray spectroscopy (EDS; XFlash 610 M detector, Bruker Nano, Germany).

For determining the formation temperature of Fe-rich intermetallics and solidification temperature of Al-dendrites, the differential scanning calorimetry (DSC) measurements were realized in an STA 449 F5 Jupiter heat flow DSC with a Proteus software (Netzsch, Germany). The DSC device operates with three individually adjustable mass flow controllers (MFC), where a constant flow rate of argon gas of 40 mL min⁻¹ was applied for the DSC cooling curves. As a reference, an empty crucible was used for each DSC curve. The investigated specimens (slices: diameter of 3.0 mm, height of 2.0 mm, and weight of ≈40 mg) were always subjected to the same temperature regime: heating with a constant rate of 20 K min⁻¹, holding the temperature of 800 ℃ for 30 min, and then varying the cooling rate between 0.5 and 40 K min⁻¹ [22]. Consequently, the peak temperature and an extrapolated onset temperature of the Fe-rich intermetallic phases were determined from the DSC cooling curves. Thereby, the onset temperature was extrapolated using the tangent intersection method taken on the inflection point and the interpolated baseline. The baseline is the curve section where no sample reaction occurred before and after the detecting peak. Accordingly, the DSC cooling curves must be fitted horizontally and smoothed (elimination of background noise). The described procedure has been used in preliminary studies [14, 22].

31.2.3 Setup of the Laboratory Filtration Apparatus

The experimental trials were conducted on a specially designed laboratory filtration apparatus (Fig. 31.2). Predefined thermal conditions provided by the heating element (CeraSys 1300; Berghütten, Germany) allow high repeatability and good comparability between the batches in the filtration process. The heating module has an electrical power of 2.1 kW, which is limited due to the dimensions of the ferritic heat conductor (CrFeAl139) and the output voltage (≈30 V) of the used converter (transformer). The base material (secondary Al alloy) can be melted or retained liquid in an A 3 graphite crucible (Mammut-Wetro, Germany), which is inserted into the heating element and placed on a base plate with a pedestal made of refractory material. The graphite crucible is fixed between the filter box half (stainless steel) and the base plate via four turnbuckles. A gasket made of Alsitra KP refractory paper (Rath, Germany) is integrated between the filter box half and the graphite crucible. Thereafter, a 30 ppi ceramic foam filter (CFF) with dimensions 50 × 50 × 22 mm (Hofmann Ceramic, Germany) is located in the cavity or chamber of the filter box half. To prevent bypass formation during filtration, another gasket (refractory paper) is added between the two halves of the filter box. The other half is mounted inversely with hexagonal cylinder head screws and has a channel with a receptacle for a screw-in flange to which a vacuum pump is connected. The vacuum can be applied before and/or during filtration and achieve a pressure of ≈500 mbar. The filtration process is initiated by tilting the apparatus (180°) with the aid of pivot bolts and structural abutments, transferring the conditioned melt from one graphite crucible to another. The opposite graphite crucible A 3 is clamped using a closing lever push rod clamp FL-121/45.

Fig. 31.2

Laboratory filtration apparatus in a isometric sectional view with silicon nitride protection tube to record the bath temperature, b isometric illustration of the complete device, and c dimetric sectional illustration including markings (numbers 1–8) of the component and material designation [22]

Pos.	Designation of components
1	Pivot bolt: M16 with hexagon head screw (stainless steel)
2	Graphite crucible of form A 3 for 1 kg Aluminum
3	Filter box halves with 240 x 238 x 30 mm (stainless steel)
4	Turnbuckles GN832-55-Ni (stainless steel)
5	Closing lever: Push rod clamp FL-121/45
6	Mounting for screw-in flange M16 DN 16 KF (stainless steel)
7	Heating element: CeraSys™ 1300 with an electric power of 2.1 kW
8	Base-plate of refractory material with the dimension of 230 x 228 x 40 mm

The operating principle and heat balance assessment was already described in preliminary study [22]. Accordingly, excessive heat loss was reduced using an appropriate pressure-resistant, fiber-reinforced insulation material (OxOx, Schunk Carbon Technology, Germany), resulting in a decrease of preheating temperature within the filter box halves to the range of 220–260 ℃. The temperature range corresponds to the preheating used in permanent die casting. The fiber-reinforced aluminum oxide ceramic was inserted as a crucible cover in a machined groove of the metallic filter box half above the A 3 crucible. The thermal balance of the device was monitored by four thermocouples placed at different positions as described in the literature [22]. The melt measurements were obtained using a thermocouple in a protective tube made of silicon nitride (StarCeram N 7000; Kyocera Fineceramics, Germany). Silicon nitride has a comparatively high thermal conductivity of ≈20 W mK⁻¹, allowing relatively instantaneous measurement of the temperature signatures in the melt bath. The predefined thermal conditions simultaneously led to a predetermined cooling rate during the shutdown of the apparatus. The cooling rate was recorded in the molten range (730–600 ℃), i.e., before the formation of the Al dendrite network, and consisted of 3.8 K min⁻¹ [22]. Hence, the influences on forming primary solidifying Fe-rich intermetallics are investigated in DSC cooling curves at 3.8 K min⁻¹ in the following for subsequent filtration trials.

31.3 Microstructure and DSC Cooling Curves

The formation temperature of the primary Fe-rich intermetallic phases is examined in DSC cooling curves with different cooling rates in two series based on an AlSi9Cu3(Fe) alloy with varying initial Fe, Mn, and Cr contents. In particular, Mn and Cr lead to the phase transition from a harmful plate-like β-Al_4.5FeSi morphology into a less detrimental cubic α-intermetallic [1, 13, 19, 20]. These α-intermetallics exhibit a Chinese-script, a star-like, and a coarse polyhedral morphology [18,19,20]. The literature widely describes the effect of phase transformation of plate-shaped β-Al_4.5FeSi [1, 20, 21, 23, 24]. Mn and Cr, inter alia (such as Ni, Be, Co, etc.), represent the most effective elements for influencing the morphology of Fe-rich intermetallics [13, 17, 19, 25, 26]. Thus, the formation of Fe-rich α-intermetallic (sludge phase) can be calculated via an empirical equation as sludge factor (SF) regarding the initial weight fraction w_M of the elements M = Fe, Mn, and Cr [12, 14, 17, 20, 25].

$${\text{SF = 100 }}\left( {{\text{w}}_{{\text{Fe }}} {{ + 2w}}_{{{{\text{Mn}}}}} {\text{ + 3w}}_{{{\text{Cr}}}} } \right)$$

(31.1)

The different weight fractions of elements Fe, Mn, and Cr are represented as the variation of the initial chemical composition by the sludge factor (SF). For this purpose, microstructural investigations are conducted starting from the first series of alloy compositions (Leg A–E, Table 31.2) in the as-cast state. The compositions are partly beyond the tolerances of the technical standard DIN EN 1706, resulting in initial Fe contents referring to previous studies by Wagner et al. [1]. Since Mackay [27] investigated the influence of an increasing Fe content on the formation temperature of the plate-like β-Al_4.5FeSi phase, a pre-dendritic formation of the Fe-rich β-intermetallics was detected starting from ≈1.2 wt% Fe via thermal analysis cooling curves (at 6 and 30 K min⁻¹) in an A356 alloy (corresponds to AlSi7Mg0.3) and A413 alloy (corresponds to AlSi12(Fe)). Based on these findings, the initial Fe content was successively increased in the experimental series to ≈1.6 wt% (Leg A–E) and ≈1.2 wt% Fe (Leg I–X), respectively. A similar correlation of primary formation of Fe-rich intermetallics between initial Fe content and formation temperature was established by Shabestari [28] on an A413 alloy. The empirical correlation of formation temperature of primary solidifying intermetallics (sludge) and initial weight fraction of Fe is consequently provided [22, 28].

$${\text{T}}\,_{{{\text{Sludge}}}} \, \, = \,\left( {645.7 + 34.2 \, \left( {100\,w_{{{\text{Fe}}}} } \right)^{2} } \right)^\circ {\text{C}}$$

(31.2)

According to Eq. 31.2, an increase in Fe content from 0.95 to 1.31 wt.% (Leg A to B) would raise the formation temperature from 676.6 to 704.4 ℃. However, the objective is to create sludge particles (α-intermetallics) as a primary solidifying phase in a filterable size (>100 µm²), thus, the elements Mn and Cr are also added to equal proportions to Fe to modify the morphology. Therefore, the filtration efficiency for Fe removal would be improved, and the operating range can be extended to initiate an appropriate filtration process.

Low contents of Cr (>0.04 wt%) and Mn (>0.3 wt%) lead to a transformation of the β-intermetallic into a cubic α-intermetallic [18]. Consequently, in the first series (Leg A–E), the Mn content is successively increased from 0.1 wt% to ≈1.5 wt%. Then, the Cr weight fraction increases slightly to ≈0.2 wt% and remains constant. Due to the increasing Fe, Mn, and Cr contents, the w_Mn/w_Fe ratio and sludge factor SF also increase, corresponding to Eq. 31.1. In the AlSi9Cu3(Fe) base alloy (Leg A), their values amount 0.11 (w_Mn/w_Fe) and 1.28 (SF). From Leg C to Leg E, the ratio reaches 0.39 and 1.0, with a rising SF of 3.30, and finally 5.23. The cross-sectional micrographs of these specimens in gravity die casting are shown in Fig. 31.3.

Fig. 31.3

Optical micrographs of polished cross-sections of two exemplary alloys in the as-cast state for a Leg A and b Leg E (compared to Table 31.2). The marked arrows with numbers 1 and 3 represent the different morphology of Fe-rich intermetallics according to the literature [22]

The secondary AlSi9Cu3(Fe) alloy (Leg A) consists the typical microstructure of α-Al dendrites, eutectic Al-Si phase, a compact Cu-containing phase (Al₂Cu), and the intermetallic Fe-containing phase (β-Al_4.5FeSi, marked in Fig. 31.3a). By increasing the initial Mn content (to ≈1.5 wt%, Leg E), the phase transition occurred as described, leading to a primary solidified Fe-rich α-intermetallic (marked with number 1 in Fig. 31.3b). Further phase components are also observed, wherein the eutectic Al-Si phase gradually recedes into the interdendritic regions of the α-Al solid solution due to the higher volume-related formation of Fe-rich intermetallics. These α-intermetallics exhibit the mentioned star-like and/or coarse polyhedral morphology.

The reason for this is the increasing Mn content, which equally increases the w_Mn/w_Fe ratio. In Leg E, no needle-like or plate-like β-morphology of intermetallic phases is observed. Therefore, Fig. 31.4 shows SEM images in backscattered electron (BSE) contrast for selected intermetallic morphologies. In sample Leg C, the evidence of plate-like β-intermetallics can be seen adjacent to star-like α-intermetallics (Fig. 31.4a). However, this indicates an incomplete phase transition in Leg C resulting in a w_Mn/w_Fe ratio of 0.39. In Leg D, no plate-like (or needle-like in depth etching) β-intermetallics occur in optical micrographs or SEM images, confirming the critical ratio (w_Mn/w_Fe of 0.5) referred to the literature [11, 22]. In addition, a further particular morphology, i.e., an altered Chinese script-like morphology, is significant in Leg C (Fig. 31.4b). Theoretically, the coral-like morphology (as Chinese script-like intermetallics) corresponds to a pre-eutectic or post-dendritic signature in cooling curves. Hence, their volume-related extension is limited within the interdendritic regions, resulting in a mainly branched form of these particles. Similar morphology has been reported by Fabrizi and Timelli [25], indicating an abrupt transition in the growth mechanism initiated by a strong supercooling effect or local concentration difference. According to Cao and Campbell [29, 30], the influence of strontium (Sr) also affects the growth mechanism of the Fe-rich intermetallics, thus, a coral-like morphology may be formed instead of a finely branched Chinese script-like structure. Due to the chemical composition of secondary commercial alloys, this influence is distinctive, as they are mainly provided in the permanently refined state (Sr ≥ 200 ppm) [22, 25, 29,30,31].

Fig. 31.4

SEM images in BSE contrast (a–c) from polished cross-sections of exemplary alloy compositions of the first series (Leg A–E) in the as-cast state, representing the Fe-rich intermetallics in plate-like β-Al_4.5FeSi morphology (no. 3), Chinese-script or coral-like, as well as the star-like morphology of α-intermetallics (no. 1). SEM images of characteristically morphological structures of Fe-rich intermetallics (d–f) are represented after deep-etching with hot sodium hydroxide solution (65 ℃, 1 min, NaOH)

Figure 31.4f depicts a section of primary solidifying Fe-rich α-intermetallics after deep etching with hot sodium hydroxide solution. Thereby, the growth mechanism of the α-intermetallic particles can be demonstrated. This is similar to the growth mechanism of Al dendrites. Starting from centric nuclei, which are usually formed as regular hollowed dodecahedrons [17, 18, 23, 25], the continuous growth of primary dendrite arms occurs in the {110} plane along the < 100 > direction. Subsequently, secondary and ternary dendrite arms form on the primary arms. In the 2D optical images, the coalescing ternary arms of the Fe-rich intermetallic particles appear as a window shaped (hollow cube) structure. As diffusion progresses, closed coarse polyhedral intermetallic particles are finally obtained. From Fig. 31.4f, the progressive growth of the secondary arms, which emerge from the center, can be observed. According to the literature [17,18,19, 25], the cooling rate and growth mechanism are directly related. However, despite that, a correlation between initial chemical composition and the growth mechanism and/or morphology can be established in the following from DSC cooling curves.

The formation temperature of stable and metastable phase compounds was examined based on the table shown in Fig. 31.5. In addition to the distinctive phase compounds of Al dendrites, eutectic Al-Si phase, and Cu-containing complexes (peaks 2–7), which are typical for this type of alloy, the formation of the primary α-intermetallics (sludge, peak 1) was also observed [14,15,16]. However, all reaction peaks except the Mg₂Si phase (peak 5) have been detected in the DSC cooling curves. Regarding the formation temperature, a tendency to higher temperatures of the Fe-rich α-intermetallics with increasing initial content of Fe, Mn, and Cr can be observed in the cooling curves at a constant cooling rate of 3.8 K min⁻¹ (Fig. 31.5). This tendency to form primary Fe-rich intermetallics (onset temperature) is observed for any cooling rate in all alloy compositions according to the literature data [14,15,16, 22]. However, no formation temperature of primary solidified α-intermetallics is detectable in the base alloy (Leg A). Starting from Leg B, the formation temperature increases from 677.2 to 695.3 ℃ (Leg D) and reaches a maximum of 704.7 ℃ with the further addition of Fe, Mn, and Cr (Leg E). The peak temperatures for the α-Al dendrites and the Al-Si eutectic remained nearly constant at ≈585 ℃ and 565 ℃ in all alloys, respectively. The Cu-containing phases (Al₂Cu and Al₅Mg₈Si₂Cu₂) behaved similarly at ≈495 ℃. Due to measurement noise, the distinction of these phases was complicated, hence they are simply summarized under the designation 6 - 7.

Fig. 31.5

DSC cooling curves with the first derivative (DDSC) of the first series of alloys (Leg A–E) at constant cooling rate (3.8 K min⁻¹). The average of three individual measurements for maximum peak (1) and onset temperature (1*) is presented, including the corresponding standard deviation (in parentheses). The designations 1–7 correlate with the peak reactions and/or phase formation of the AlSi9Cu3(Fe) alloy type, according to the literature data [22]

In contrast, the β-Al_4.5FeSi phase was revealed in the base Al alloy and is expected to form a pre-eutectic phase at about 583.8 ℃ [14, 22]. However, the micrographs also showed their evidence in other alloys (including Leg C). Furthermore, the pre-eutectic β-Al_4.5FeSi phase is not clearly displayed as individual peaks in the DSC curves. The β-intermetallics can be merely detected as a gradient alteration in the first derivative of the curve. Likewise, this is evident as an inflexion point in peak 4 within the DSC cooling curves. However, this implies that no onset or peak temperature can be determined for the β-Al_4.5FeSi phase. Consequently, the pre-eutectic β-phase and the eutectic Al-Si phase are being consolidated as peak 4 [22].

In preliminary studies, the deferral in peak and onset temperatures was also observed depending on different cooling rates (0.5–40 K min⁻¹), particularly for the first series of alloy compositions (Leg A–E). Thereby, the onset temperature dropped by about 26 K from ≈687 ℃ towards 661 ℃ in Leg C. However, the formation temperature of the Fe-rich intermetallics decreased simultaneously in Leg D and E by about 24 K and 8 K, respectively, as reported in the literature [14, 16, 22].

Since the specific enthalpy ΔH theoretically reflects the area fractions of the respective phase formation in the microstructure, microstructural investigations were examined on the half-width cross-sections of the DSC samples. The microstructure was significantly affected by the cooling rate. With increasing cooling rate from 3.8 to 40 K min⁻¹ the Al dendrites, the eutectic Al-Si phase, as well as the lamellar and/or complex microstructure of Cu-containing phases became considerably finer dispersed [14, 15, 22].

Similarly, the primary solidifying Fe-rich α-intermetallics are affected regarding the growth mechanism by the cooling rate. As the cooling rate increases from 0.5 over 3.8 to 40 K min⁻¹, the morphology is impaired from coarse, polyhedral α-intermetallic particles towards an open, star-like, and/or window-like morphology, which has been described previously [14, 15, 22]. Moreover, the number of primary α-particles increases with increasing cooling rate due to a significantly larger supercooling during solidification. Consequently, the time interval in the solidification range of the Fe-rich α-intermetallics is shorter, interrupting the growth mechanism towards closed intermetallic particles and deferring it to the post-dendritic and/or pre-eutectic solidification range. This leads to a Chinese script-like or coral-like manifestation of the α-intermetallics in the interdendritic regions due to a strong supercooling effect.

The cooling rate and thus the occurring morphology of Fe-rich intermetallics is predetermined depending on the casting process (e.g., ≈10 K min⁻¹, representing gravity sand casting conditions [21]). However, due to the thermal stablility of the filtration apparatus used, the cooling rate was predefined as 3.8 K min⁻¹. In preliminary study, compact polyhedral α-intermetallics have been detected by EDS measurements in scanning electron microscopy (SEM) at this cooling rate [22]. Therefore, a second series with increasing weight fractions of Cr (Leg I–X, Table 31.2) is conducted to determine the formation range of primary solidifying α-intermetallics by further DSC measurements. Herein, the successive addition of Cr to equal proportions as Fe and Mn (to ≈1.2 wt%) has been carried out. According to Eq. 31.1, the sludge factor (SF) also increases, enabling the determination of the formation range of Fe-rich intermetallic particles by the extrapolated onset temperatures in the DSC cooling curves for an applicable filtration process.

Figure 31.6 depicts the formation (or operating) range by two regression curves (dashed lines) obtained for the Fe-rich α-intermetallics (sludge phase) and the Al dendrites. Corresponding to the chemical compositions of the second series (Leg I–X), the increase of Fe content from initially 0.8 to ≈1.2 wt% occurs first. Equation 31.2 indicates that the formation temperature would thereby rise from ≈669 ℃ to 693 ℃. However, the empirical calculation is not consistent with the onset temperatures of the DSC cooling curves. Consequently, increasing the initial Fe content merely resulted in an insufficient widening of the formation range for Fe-rich α-intermetallics. Similarly, the operating range for potential removal of the Fe-rich phases would be constrained. Further increasing of the initial Mn content from initially ≈0.3 to 1.2 wt% while maintaining a constant Cr (≈0.04 wt%) extended the formation range (by about 50 K). Finally, a subsequent increase of the Cr content to also 1.2 wt% significantly promoted on the formation temperature of the Fe-rich α-intermetallics to above 700 ℃. Since, according to Eq. 31.1, Cr has the greatest impact (factor 3) on sludge formation, the formation range is widened to initiate an appropriate filtration process.

Fig. 31.6

Determination of the formation range (operating range) via two regression lines (incl. equations) obtained from the onset temperature in DSC cooling curves for the Fe-rich α-intermetallics (sludge phase) and the Al dendrites at a cooling rate of 3.8 K min⁻¹. The chemical compositions of the experimental series are designated as Leg A–E and Leg I–X

The formation temperature values from the second series (Leg I–X) are also compared to those from the first series (designated as Leg A–E). These values coincide with the regression curves (dashed lines). Strikingly, the higher effect of Cr on the formation temperature can be emphasized by comparing samples from Leg B and III. The averaged formation temperature for Fe-rich α-intermetallics is ≈677 ℃ (Leg B) and is considerably higher than that of Leg III, despite that the chemical compositions merely differ by 0.1 wt% Fe and 0.2 wt% Cr. Hence, Cr and Mn are recommended to add in equal proportions in the conditioning step to ensure a reliable operating range for a metal melt filtration process.

31.4 Experimental Filtration Trials in a Laboratory Filtration Apparatus

Samples before and after filtration were examined by image analysis using the previously described laboratory filtration apparatus to appraise the removal potential of Fe-rich intermetallic phases. For this purpose, an examination field was established for a predefined alloy composition (exemplary on Leg D) with the variation of time and temperature (620 ℃, 655 ℃, and 685 ℃). After finalizing a predetermined time and temperature, the filtration process was initiated by tilting the apparatus (180°). Consequently, the residual Fe-depleted melt can be purified from the Fe-rich particles by a 30 ppi foam ceramic filter (CFF). The parameters for investigation were preselected based on the literature [14, 15, 22]. Ashtari et al. [2], e.g., reported on the sludge formation after short times of retention of ≈15 min in an A356 alloy. Therefore, the minimum dwell time was selected for 20 min. Conversely, studies on the time-dependent formation of Fe-rich intermetallics (sludge) by Dietrich et al. [21] were conducted for 4 h, 24 h, and 96 h. However, the sludge formation stagnated after a certain time, and thus a time exceeding of 4 h is not required for the formation of coarse polyhedral α-intermetallics. Accordingly, the maximum time was set at 120 min, considering that extended dwell times will not promote further phase growth [21, 22]. Hence, the examination was conducted for 20, 70, and 120 min.

Regarding the temperature, they were obtained from the onset and peak temperatures of the respective DSC cooling curves of the alloy. The peak temperature was preferred due to the reliable formation of intermetallic particles. According to Fig. 31.6, the formation temperature of the Al dendrites increases moderately along the regression line and equation. Consequently, 620 ℃ was selected as the lower filterable limit to ensure the range of 20–30 K above their formation temperature as a safety margin. Moreover, CalPhaD simulation results (JMatPro) revealed the highest filterable quantity of α-intermetallics at this temperature [22]. This is evident in the microscopic images of the half-widths taken from selective samples in quenching experiments.

Figure 31.7 displays the micrographs for the base AlSi9Cu3(Fe) alloy (Leg A), the conditioned alloy (Leg D) used for the examination field, as well as the dimensions of the quenched sample stick. Following the sampling principle using glass tubes and a Peleus ball, the molten metal with expected primarily solidified particles was pulled into the cavity of the glass tube via negative pressure. The thin glass tubes had an inner diameter of ≈5 mm. The extracted material was then quenched in water bath (at room temperature) in order to preserve the microstructure state. The manifestation of the Fe-rich intermetallic particles in terms of size and distribution during extraction at 620 ℃ in Leg D provides evidence for their formation as a primary solidifying phase. Etching (hot sulfuric acid solution) leads to higher contrast (blackening) of the intermetallics compared to the remaining microstructure. The remaining microstructure (including Al dendrites, Al-Si eutectic, etc.) is finely formed and dispersed due to the relatively high quenching cooling rate (see Leg A). In contrast, the intermetallic α-particles are present in the desired size (>100 µm) and occurred predominantly in section A/3 of the stick specimen due to the suction effect. Consequently, the coarse polyhedral α-intermetallics must have formed previously during temperature conduction in the laboratory apparatus.

Fig. 31.7

Optical micrographs of etched cross-sections of the half-widths of sampling sticks from section A/3 taken from the sampling scheme presented above, exemplified on two alloy designations (Leg A and E). The specimens were extracted from the laboratory apparatus at 620 ℃ and at height of ≈20 mm above the crucible bottom

Concluding from the evidence for the primary solidification of Fe-rich intermetallics, evaluation of samples before and after filtration by optical microscopy was performed for each parameter combination pursuant to the literature [22]. For each section, the porosity was determined first on the polished samples and then compared to the total fraction, consisting of pore and etched Fe-rich intermetallics fractions. Samples for the determination of these fractions are provided in Fig. 31.8.

Fig. 31.8

Examination of exemplary samples to determine the porosity fraction (polished samples) and the total fraction (etched samples: porosity and Fe-rich phases) based on section III for the parameters a 685 ℃ and 120 min before filtration, b after filtration, and c 620 ℃ and 20 min after filtration. Values correspond to the literature data [22]

Figure 31.8 presents exemplarily micrographs of section III before filtration (685 ℃ and 120 min) and after filtration (685 ℃ and 120 min, as well as 620 ℃ and 20 min). The division of sections and the determination of area fractions have been reported in preliminary investigations [21, 22]. Porosity localization differs between the samples before and after filtration due to procedural reasons via 180° tilting. The area fraction of the Fe-rich intermetallics is calculated by the difference of the total minus porosity fractions. They amount to 10.9% before, as well as 5.26%, and 1.78% after filtration, respectively. Therein, the inhomogeneous distribution of the Fe-containing intermetallics is striking. Before filtration, the α-intermetallics are distributed along the edges of the crucible wall (Fig. 31.8a), while after filtration, the intermetallic particles are more homogeneous and finely distributed. The manifestation of particle distribution is explained by the pronounced primary solidification of the intermetallic phases, as already shown in the DSC cooling curves (Fig. 31.5). In this case, the α-intermetallic preferentially forms heterogeneously on pre-existing surfaces [22, 29, 31, 32]. Consequently, the samples exhibit an exogenous, shell-like solidification type before the filtration process is initiated. The samples after filtration show isolated compact intermetallic particles (with an area size of ≈10.6 µm²). During further cooling in the solidification range, the particles may have formed depending on their filtration or process temperature. This evidence was also shown in preliminary investigations based on the summary of the detected area fractions of the sections as an upper (I and II), as well as lower segment (III and IV) [22]. The reduction in area fraction in samples after filtration from about 5.0% (685 ℃) to finally ≈1.5% (620 ℃) was due to the temperature-dependent solubility of Fe in molten aluminum [33].

The comparison of Fig. 31.8b to c shows notably reduced and detected intermetallic particles after etching at 620 ℃, although the dwell time was shortened by 100 min prior to the initiation of the filtration process. The quantitative consideration (area fractions) is consistent with the qualitative results of chemical composition after the filtration (Table 31.3), showing no further reduction in α-intermetallic particles regardless of the dwell times (20, 70, and 120 min) [22]. Accordingly, the greater influence on the formation of α-intermetallics is inferred by temperature than by time. These assertions are consistent with the literature, indicating that intermetallic particles can form continuously throughout the entire solidification range until the onset of the formation of the α-Al dendrites [17, 23, 24].

Table 31.3 Experimentally determined chemical compositions (weight%) after remelting of the sample material in the examination field with a variation of time and temperature on an exemplary predefined alloy composition (Leg D) in the optical emission spectrometer OES SPECTROMAXx (Ametek, USA). Values correspond to the literature data [22]

Fe reduction was initially achieved on the remaining crucible material, which had not been used for microstructural examinations. However, the results of the optical emission spectrometer (OES) measurements required to be more differentiated due to the distribution of Fe-rich intermetallic particles. In consequence, the sample materials were subsequently remelted in the crucible induction furnace at above 800 ℃ to provide a reliable evaluation of the chemical composition. The results of Fe, Mn, and Cr examination in the OES of the remelted sample material before and after filtration according to the scope of this study are listed in Table 31.3 [22].

Particularly, the chemical compositions of all samples before filtration are similar to the reference alloy Leg D (1.6 wt% Fe, 1.0 wt% Mn, and 0.26 wt% Cr). Otherwise, the element contents of Fe, Mn, and Cr of the filtered samples frequently decline below the limiting contents of the standard AlSi9Cu3(Fe) alloy composition (DIN EN 1706). Comparing the parameter combinations of 685 ℃ and 120 min with 620 ℃ and 120 min, the Fe content decreases from 1.130% to 0.808% and remains at ≈0.8 wt% Fe, even when the holding time is shortened to 70 or 20 min. Thus, the Fe content reaches the lowest values of 0.855%, 0.785% and 0.808 wt%, which represent the maximum reduction to about 50% in average (at 620 ℃) [22].

With increasing the process or holding temperature, the Fe reduction decreases significantly to about 35% (655 ℃) and 30% (685 ℃), respectively. The highest reduction of Mn and Cr is likewise achieved at 620 ℃ with ≈66% and 86% [22]. Consequently, the two-step process consisting of conditioning and filtration allows a noticeable reduction of the Fe, Mn, and Cr elements. A relatively high proportion of Fe can be bound due to the formation of intermetallics and separated via the laboratory filtration apparatus. Moreover, the maximum Fe reduction at 620 ℃ correlates to the highest quantity of α-intermetallics in the two-phase range, according to the JMatPro® simulation results. Based on these findings from the examination field, additional filtration trials were performed in the second experimental series (Leg I–X) regarding the most evidentially effective parameter combination (620 ℃ and 20 min) for Fe removal.

Figure 31.9 depicts the remaining elemental contents of Fe, Mn, and Cr after filtration using a 30 ppi alumina CFF in the second series (Leg I–X, Table 31.2). The initial concentrations are proportionated up to 1.2 wt%. The remaining sample materials were also remelted at 800 ℃ to analyze the influence of the varying initial chemical composition in OES on Fe removal. The respective addition of elements is marked in Fig. 31.9. With increasing the Fe, Mn, and Cr content, the initial sludge factor varied simultaneously, which explains that the residual contents in the filtered samples were plotted over the sludge factor. According to Eq. 31.1, the increase of initial Fe resulted in a rising SF from 1.64 to 1.97. Consequently, the Fe concentration in the residual content also ascended to over 1.0 wt%.

Fig. 31.9

Determination of elemental contents of Fe, Mn, and Cr after filtration of the Fe-rich intermetallics (sludge) based on the two-step procedure for the alloy compositions of the second series (Leg I–X, Table 31.2). The successively increasing addition of the respective element to the initial chemical composition is labeled with respect to the residual element contents

The further addition of Mn (to 1.2 wt%) increases the SF from 2.66 to 3.72. Moreover, the solidification range extends to form Fe-rich intermetallics (Fig. 31.6). The residual Fe content in the filtered samples is thus reduced from 0.879 to 0.677 wt%. However, the residual Mn content in the filtrate increases similarly from 0.421 to 0.532 wt% and exceeds the tolerance level specified in the standard. With the further proportional addition of Cr (up to 1.2 wt%), the content of both elements decreased. As a result, the Fe content declines to below 0.4 wt%, and the Mn content falls to less than 0.35 wt% below the tolerance limits. Logically, the addition of Cr leads to an increase in the residual Cr content, which still remains within the standard’s limits (DIN EN 1706).

Therefore, the promising results obtained in the laboratory filtration apparatus including the findings regarding the formation characteristics of Fe-rich intermetallics, need to be implemented for an industrial application. For this purpose, additional industrial scale filtration technology is presented in the following.

31.5 Development of a Filtration Process Technology on an Industrial Scale

During the operation of the new filtration process technology, the temperature management was adjusted according to the most effective parameter combination for a crucible induction furnace with a maximum capacity of ≈150 kg of liquid aluminum. For this purpose, the chemical composition was selected according to the optimum operating range determined previously. Accordingly, the conditioned melt is maintained at 620 ℃ for 20 min and then superheated to ≈700 ℃ in the crucible induction furnace with a heating rate of 12 K min⁻¹. Exceeding the superheating temperature should be avoided in order to achieve the desired formation range of Fe-rich α-intermetallics (Fig. 31.6).

The modular design of the newly developed industrial filtration apparatus consists of the modules inlet, outlet, and filtration unit (filter box chamber) and comprises a total capacity of about 48 kg. The inlet and outlet modules can be demounted at any time, and the unit can be integrated into an existing continuous launder system. However, the principle schematic shown in Fig. 31.10a enables a batch operation mode. Considering a compromise of using both institutional filters, produced in cooperation with the subprojects A01, A02, and A07, and industrially manufactured CFF (trapezoidal for continuous casting), the smallest 7-inch filter chamber size was elected. Thus, alumina-based CFF (e.g., Sivex filters; Pyrotek, Czech Republic) [34], C-bonded alumina filters, and alternative filter materials and coatings can be used. In this context, different filter materials and coatings were evaluated in advance using the laboratory filtration apparatus in order to evaluate suitable filter types under industrial conditions with higher melt volume (up to 100 kg). For this purpose, the C-bonded Al₂O₃ CFFs are used, which were developed in the third funding period of the Collaborative Research Center 920. They are manufactured in dimensions 178 × 178 × 50 mm (trapezoidal) by the Institute of Ceramics, Refractories, and Composite Materials and tested on an industrial scale. The fabricated CFF is thus inserted into the filtration unit’s filter chamber, including the gasket. Predefined filter positions of 3° ensure improved flow conditions on the inlet side. Both, standard 6.5 mm thick fiber and 3.5 mm thick expandable material are used as gasket types. The inlet and outlet modules, as well as the completely assembled filtration device are presented in Fig. 31.10.

Fig. 31.10

The modular setup of the newly developed filtration technology for investigations on metal melt filtration of Fe-rich intermetallics on an industrial scale. a Schematic illustration of the setup, including b isometric projection of the outlet module and c inlet module, ensuring a consistent metallostatic level (priming height) of ≈160 mm above the 7ʺ CFF by modifying a pouring lip instead of tapping plug. d dimetric projection of the filtration unit (7-inch filter box chamber) and e illustration of the completely assembled filtration apparatus with a capacity of ≈50 kg of molten aluminum

The attachment modules of the inlet and outlet shown in Figs. 31.10b and c were manufactured as negative models from a milled part made of 24 mm thick plywood (birch) in ribbed construction. Castable hydraulic-setting concrete refractory (Rath, Germany) was used for the modular parts. In accordance with the manufacturer's instructions, fabrication was conducted with the aid of a vibratory table and a prescribed drying process. As insulation material, microporous insulation panels WDS (Rath, Germany) with a thickness of 50 mm were used and integrated with the modular parts into a welded steel structure with flanged surfaces (S235JR). The first modification occurred on the outlet side, where a pouring lip was implemented instead of plug construction. In order to ensure a continuous and complete wetting of the filter surface, this was necessary to provide a consistent metallostatic height above the filter position during the filtration process. The casting lip operates as overflow with a constructive height of 130 mm above the launder bottom and additionally 30 mm height to the upper filter chamber rim, resulting in excess pouring metal (>48 kg) being collected in cavities at the outlet side. The height of 160 mm is based on the priming height of the manufacturer's recommendation: Pyrotek filtration best practices [35], allowing filter grades or ppi for CFF to be used up to 30 ppi. Figure 31.10d presents the dimetric projection of the filtration unit. Therein, a side-mounted draining stone is seen, which allows to empty the entire filtration device after the successful filtration process. In addition, the completely assembled filtration apparatus is demonstrated in Fig. 31.10e. In particular, the heater (Leister LE 10000HT) mounted on the lid of the filter box requires an electrical output power of ≈15 kW, which provides an output temperature of ≈780 ℃ to preheat the filter chamber and CFF to over 500 ℃. However, the essential modification of preheating is particularly formative due to the C-bonded Al₂O₃ CFF since conventionally used gas burners would not be compatible with this filter material.

Figure 31.11a constitutes a representative cross-section of the 20 ppi C-bonded Al₂O₃ CFF used for a conditioned AlSi9Cu3 alloy with 0.866 wt% Fe, 0.781 wt% Mn, and 0.718 wt% Cr (SF: 4.58). This reveals a distinctive filter cake of intermetallic particles over the entire filter area. After weighing the CFF before and after the filtration process, i.e., the unloaded and loaded filter with particles, the result indicates a weight increase from the initial 920 g to 4032 g. Consequently, the CFF separated a significantly large amount of particles through the filter cake. The pouring temperature in the ladle was 690 ℃. Measuring points in the launder system at the inlet and outlet using the silicon nitride protection tubes provided a temperature of ≈671 ℃ in the filtration device. This can be attributed to the transfilling process. The measurements during filtration imply that the process temperature for SF of 4.58 must have been below the required formation temperature of ≈678 ℃, according to Fig. 31.6. Therefore, distinct loading or filter cake was observed on the filter surface with compact polyhedral intermetallic particles of  >100 µm (Fig. 31.11b). Furthermore, additional oxide films can be detected, which are retained within the filter cake (Fig. 31.11c). The elemental distribution in the EDS mapping from a section of Fig. 31.11b indicates the manifestation of Fe-rich intermetallics.

Fig. 31.11

Optical micrographs of the 20 ppi C-bonded Al₂O₃ filter of the conditioned Al alloy poured at 670 ℃ with a panoramic, and b individual images at 100×, and c 1000× magnification. The intermetallic particles are visible over the entire area in the filter cake

Figure 31.12 displays the SEM images in BSE contrast, as well as the distribution of the elements Al, Si, C, O, Fe, Mn, Cr, and Cu in the investigated section. Herein, Al, Si, and Cu represent the origin AlSi9Cu3 matrix alloy. The Fe-rich intermetallics retained in the filter cake are evident by the Fe, Mn, and Cr elements. The elemental distribution analysis includes only 1.96 wt%, 0.62 wt%, and 0.41 wt% of these elements, respectively, related to the entire section. The elements C and O are characterized by higher weight percentages, namely 39.79 and 7.33 wt%, and represent the filter material used for the CFF. Due to the new filtration process technology, a considerably high portion of Fe was separated, resulting from forming primary solidifying Fe-rich intermetallics. The elemental contents for Fe, Mn, and Cr were reduced in consequence of the filtration process by about 41%, 45%, and 61%, to values of 0.514 wt% (Fe), 0.432 wt% (Mn), and 0.283 wt% (Cr), respectively.

Fig. 31.12

SEM image in BSE contrasts: a the polished section from Fig. 31.11b, and b illustration of the elemental distribution in the cross-section including the detected elements, atomic number, weight%, and atomic% listed in the right table. c Individual element distribution (EDS mapping) of Al, Si, C, O, Fe, Mn, Cr, and Cu. SEM images were acquired with a magnification of 100x, a working distance (WD) of 14.9 mm, and a scale of 300 µm

31.6 Conclusions

In the transfer project T03, the Fe content of recycled and secondary aluminum alloys should be reduced to less than 0.4 wt% for an industrial application at Nemak Dillingen. Based on findings from the second funding period, the formation characteristics of Fe-rich intermetallics were investigated in collaboration with other subprojects. Subsequently, DSC cooling curves were analyzed to determine the formation temperatures of Fe-rich intermetallics under semi-technical and practical conditions. Consequently, the operating range for a corresponding industrial filtration process was defined by regression lines for the formation of Fe-rich intermetallics. The microstructural scope of this study was adopted to determine the most effective filtration parameters for achieving the greatest potential for separation of Fe-rich intermetallic particles. Subsequent investigation with the most effective parameter combination and variation of the initial chemical composition resulted in the highest Fe reduction to below 0.4 wt% (project target value). Furthermore, an appropriate filtration process technology has been developed, to separate Fe-rich particles on an industrial scale. However, due to the Fe reduction of ≈40–50% to 0.514 wt%, additional optimization of the filtration technology is required. Nevertheless, the CFF analysis indicates significant deposition of Fe-rich intermetallics via cake filtration, where the evidence of Fe-rich particles was revealed in EDS measurements. Further optimization options include external commissioning of auxiliary heating elements (launder heater: CeraSys 1150) with further adjustment of the filtration temperature to a lower level, as well as additional investigations of multifunctional filter coatings and materials with respect to Fe removal efficiencies.