8. Dealing with Fe in Secondary Al-Si Alloys Including Metal Melt Filtration

In pure Al–Si–Fe alloys, the hexagonal α_h-Al–Si–Fe phase (space group P6₃/mmc) with a composition given by Al7.1Fe2Si develops [4]. However, in presence of as small amounts as > 0.03 wt% Cr [5] or > 0.3 wt% Mn [5‐7] the cubic α_c-Al–Si–(Fe,Mn,Cr) phase forms instead of α_h with approximate compositions between Al₉Mn₂Si_1.8 [8], Al₂₀(Fe,Mn)₅Si₂ [9] and Al₁₅(Fe,Mn)₃Si₂ [10]. It has space group \(Im\overline{3 }\) at low ratios of Mn or Cr to Fe content or \(Pm\overline{3 }\) at high ratios of Mn or Cr to Fe content [9, 11].

Both the structures of α_h and α_c can be derived from periodic arrangements of Mackay icosahedra (MI), supplemented by further atoms (Fig. 8.1a,b). Thereby, the ideal icosahedral symmetry of such clusters is, of course, broken (Fig. 8.1c). In α_c the MI form a bcc arrangement. Additional sites of Al/Si atoms are located between the second-nearest neighbor clusters [9].

a. A ball and stick model of the icosahedral cluster composed of A l, F e, and S i and a threefold capped hexagonal prism cluster, along with unit cells. b. A ball and stick model of the icosahedral cluster composed of F e, M n, A l, and S i with a unit cell. c. A model of the near five-fold axis.

In α_h the MI form a primitive hexagonal arrangement, where ordered occupation of 1/2 of the trigonal prismatic interstices by further clusters leads to a doubling of the c axis. There P6₃/mmc symmetry results by the latter clusters occupying the (1/3,2/3,1/4) and (2/3,1/3,3/4) positions forming an AB stacking and the MI end up on the octahedral sites. It is worth to stress that the α_h and α_c phase are clearly distinct phases with different properties and should be clearly distinguished during microstructure analysis.

The α_h phase forms during primary crystallization from a Fe-containing Al–Si alloy melt under slow cooling rates in a shape of large polyhedra that can contain inclusions of Al [14, 15]. In co-dendritic and co-eutectic stages of solidification¹ and at high cooling rates the evolution of the α_h phase is suppressed [15]. The α_c phase forms in Fe-containing Al–Si alloys containing further transition metal elements especially Mn and Cr. Furthermore, experiments on the kinetics of intermetallic phase formation from ternary Fe-containing Al–Si alloys on different substrates reveals that under special conditions the α_c phase forms as metastable phase in the ternary Al–Si–Fe system. This might be attributed to the possible presence of icosahedral clusters already in the melt [31, 32]. Depending on the external conditions, e.g. elemental composition and cooling rate [18], the α_c phase evolves into a broad variety of morphologies (Chinese script, bulk facetted polyhedra etc.) [16, 18]. The morphologies of the α_c phase can be evaluated e. g. on the basis of EBSD data. Challenges arising from pseudosymmetry issues during indexing of EBSD patterns of the α_c phase are scrutinized in [29].

The structural characteristics of the β-Al–Si–Fe and δ-Al–Si–Fe phase in this section have been detailed in depth in [15, 22, 23]. The atomic structure of the β and δ phase consists of basic layers built from bicapped antiprisms with Al and Si atoms at the corners and with Fe atoms in the center (Fig. 8.2a). Within these layers, the bicapped square antiprisms share common edges, i.e. two (Al, Si) atoms. In the crystal structure of the β phase, pairs of these layers are condensed by sharing (Al, Si) atoms at the caps, thus, forming double layers (Fig. 8.2b,c). Stacking of these double layers by relative layer shifts of a/2 + c_d or b/2 + c_d allows four positions of the double layers, A, B, C, D, where c_d is a vector perpendicular to unit cell basis vectors a and b. |c_d|, thereby, is the distance between the double layers.

a to c. A schematic diagram of the ordered polytypes with the stacked layers in A B C D sequence with coordinates. d. A SEM image of B S E at 200 micrometers has a line-like structure, and a single line is magnified. e. A TEM image of the line at 500 nanometers. f. A H A A D F image of the lamella.

The δ phase is structurally closely related to the β phase, but thermodynamically β and δ are clearly distinct phases [23]. The atomic structure of the δ phase is composed of fully condensed basic layers by sharing all cap atoms. The δ phase is described in space group I4/mcm or space group Pbcn when ordering of Al and Si atoms is considered (Fig. 8.2c) [27]. The β phase has an ideal chemical composition of Al_4.5FeSi [10, 20, 21, 23, 33, 34]. The δ phase is typically described with an ideal chemical composition Al₃FeSi₂ [27].

Several crystal structure variants for the β phase have been reported. All these structures can be described using near-tetragonal lattice parameters that can be grouped into monoclinic [10, 20, 33] e.g. with space group A12/a1 [20, 21], orthorhombic [24, 25] or tetragonal [24, 33] structures which provide lattice parameter c either around 20.8 Å [20, 21, 24, 25] or 41.6 Å [10, 24, 33]. Based on SEM, TEM including SAED and HAADF and DFT calculations, that series of structures was rationalized to a more complete image considering several hierarchical levels of the crystal structure [22].

From the stacking of the A, B, C and D double layers, ordered polytypes or disordered structures arise which were elaborated in [23]. The ordered polytypes that form from a periodic stacking of the layers by b/2 + c_d – b/2 + c_d or b/2 + c_d – a/2 + c_d – b/2 + c_d – a/2 + c_d result in an AB sequence or an ABCD sequence. The former ideally shows orthorhombic Aeam symmetry (with a pseudotetragonal unit cell shape) and the latter ideal tetragonal I4₁/acd symmetry (Fig. 8.2b). However, none of these ideal structures explains the clearly observed unit cell distortions away from a tetragonal or orthorhombic metric, nor some additional diffraction peaks hinting at further ordering effects. For the purpose of orientationally robust indexing e.g. of EBSD patterns of the β phase with locally varying stacking sequence including also disorderedly stacked regions, an artificial, partially disordered tetragonal crystal structure with I4/mmm symmetry has been devised and successfully applied in practice, e.g. upon indexing of EBSD data from δ + β containing microstructures [23, 35].

Reconciling experimental evidence from literature and own studies in combination with new theoretical evidence from DFT calculations allowed to address these unsolved microstructure interpretations in [22]. Differently Al versus Si ordered model structures have been subjected to DFT calculations. Analysis of the relaxed lattice parameters imply that unit cell distortions away from an orthorhombic or tetragonal metric are generally enforced by the symmetry induced by the atomic arrangement of Al versus Si atoms. Two monoclinic structures differing solely by the orientation of the ordering in the double layers with respect to the AB stacking provide the lowest-energy. These two structures can be described in space groups.

The monoclinic angle deviates by about 1.3–1.6° from 90°, which is compatible with results from powder-X-ray diffraction analysis. The ordering in these two lowest-energy variants can be transferred to the ABCD stacking of double layers, also leading to monoclinic symmetry. Thus, previously unexplained irreconcilable sets of additional diffraction peaks comply with the three ordered structures. Three different crystal structures coexisting simultaneously at constant composition of β-Al_4.5FeSi are unexpected under thermodynamic equilibrium and can be non-equilibrium effects.

Various disordered stacking and Al versus Si ordering variants of the β can be considered as defect structure. In characteristically plate-shaped particles (Fig. 8.2d, e, f) neighboring extended phase regions of double layer stacked β phase and full condensed basic layer stacking of δ phase exist. Furthermore, next to these regions, short successive additionally condensed basic layer sequences are frequently present in the β phase. Such δ-like stacking sequences are typical planar defects of the β phase. Their presence can be related to compensation of excess Si, as implied by the higher Si content of the δ phase (Al₃FeSi₂) [22].

These structural details, on the one hand, can obscure the microstructure characterization when the full complexity is not considered during the microstructure analysis, e. g. when additional superstructure reflections would remain unidentified or misleadingly suggest the presence of further phases. On the other hand, some characterization methods might not resolve all details, so that it is recommended to use artificially higher symmetries in the microstructure analysis. The method and the structural model should address the same level of resolution.

The results and conclusions highlighted in this section mainly reported in [14] outline the state of melt conditioning based on new results and literature. The current limitations and chances for dealing with Fe in Al–Si alloys by removal of Fe on the basis of melt conditioning and metal melt filtration are presented. Melt conditioning addresses the intended adjustment of the melt or semi-liquid state for a specific purpose. Here, the formation of primary, Fe-containing, intermetallic phases, the so-called sludge, in liquid aluminium is the precondition for removal of Fe by metal melt filtration. To maximally reduce the residual content of Fe in the Al melt by separating off those primary, Fe-containing, intermetallic particles, the melt has to be specifically conditioned for that purpose according to the parameters described in the following:

The primary particles form between the liquidus temperature T_L and the onset temperature of the (Al)-solid solution T_Al-s. These temperatures as well as the type(s) of the primary phase(s) depend on the alloy composition. Information about these temperatures and phases is required to define the conditioning parameters for the melt conditioning. Three approaches can be evaluated to access these parameters: the sludge factor approach, thermodynamic calculations and thermal analysis.

Sludge factor: Conditions for the formation of primary, intermetallic particles, the sludge, can be estimated based on the sludge factor SF [36‐40]w are the mass percentages of the elements in the index. The sludge factor can be used to estimate the effect of the transition-metal elements on the sludge formation predominantly on the basis of the weighted sum of the mass percentages of the metal elements Fe, Mn and Cr. The effect of Mn is twice and that of Cr three times as large as that of Fe. The critical temperature below which sludge forms was proposed to depend solely on the sludge factor:

Also another relationship for the critical temperature in dependence on the Fe content has been suggested:

According to [36] such dependences represent a kind of liquidus temperature and in case of equilibrium conditions the sludge temperature corresponds to the liquidus temperature. Therefore, in the following, the sludge temperature T_sludge is referred to as liquidus temperature T_L.

Thermodynamic calculations: A more advanced approach to access T_L and T_Al-s are thermodynamic calculations e.g. by using the CALPHAD method [52]. Additionally, these calculations provide information about the type(s) of primary phase(s) that form and provide information about the phase formation sequence during solidification. However, the current thermodynamic databases for thermodynamic calculations evidently require some fine-tuning to predict the effect of modifying elements as Mn and Cr correctly. That current limitation but also possible kinetic effects that cause a deviation from equilibrium conditions, make experimental investigations e.g. by thermal analysis necessary to achieve information on T_L, T_Al-s and evolving phases.

Thermal analysis: Thermal analysis data e.g. differential thermal analysis (DTA) curves allow experimentally evaluating reactions occurring during melting of alloys. Occurrence of such reactions is revealed by endothermic signals in DTA data. From data evaluation reaction temperatures e.g. T_L and T_Al-s can be extracted. The primary phases are identified based on their coarse morphology as compared to all other phases present in the microstructure. Thus, the formation of the primary phases can be assigned to the occurrence of the endothermic signal observed in the DTA signal corresponding to final melting of the alloy at T_L. T_Al-s can be derived from the high intensity signal in the DTA curve corresponding to melting of the main microstructural component Al in the alloy.

The three approaches have been applied to obtain the temperatures T_L and T_Al-s for the alloys of the compositions Al7.1Si(1.5-x_M)Fe(x_M)M with M = Mg, Mn, Cr and x_M = 0, 0.3, 0.6, 0.9, 1.2 and 1.5 at%. The numbers preceding the elements indicate molar fractions. The resulting temperatures and calculated or observed primary phases are shown in Fig. 8.3.

3 scatter plots of temperature versus x F e, x M g, x M n, x C r, and S F. a and c. 2 plots are distributed on each side of the 4 increasing fit lines between 550 and 900. b. 2 plots are distributed on each side of the decreasing lines.

According to the results from the DTA experiments on the Al7.1Si(1.5-x_M)Fe(x_M)M alloys, solidification of the (Al)-solid solution starts closely below 620 °C. The thermodynamic calculations and DTA experiments show good agreement for T_Al-s. Thus, the process window for the melt-conditioning temperature can be reasonably well predicted by the CALPHAD method [52] using the TCAL4 database [53]. Hence, holding the melts at 620 °C, as applied here for melt conditioning, is the reasonable lowest possible for potential development of primary, intermetallic phases without solidification of the (Al)-solid solution during melt conditioning.

In contrast to T_Al-s, the liquidus temperatures T_L and primary phases obtained from DTA experiments, thermodynamic calculations or the sludge factor approach agree to only some limited extend.

In the alloys with Mg, the deviation is less than 8 K for all Mg-containing alloys including the alloys with x_Mg ≥ 0.6 at% and with x_Mg ≤ 0.3 at%. Note that, in the former, the (Al)-solid solution is the primary phase. In the latter, the β phase is the primary phase in the experiment while the α_h phase results from thermodynamic calculations. The discrepancy of the formed primary intermetallic phase can be understood in relation to the retarded nucleation of the α_h phase if the cooling rates are sufficiently high, as obviously the 0.5 K/s applied for cooling in the DTA experiments. Note that, after the actual melt-conditioning treatment of 2 h at 620 °C, the α_h phase is observed as primary phase (Fig. 8.4).

6 SEM slash B S E images of the alloy at 250 micrometers. The morphologies of the M g, M n, and C r based alloys have several phase distributions, which are represented by the arrows.

For Mn- and Cr-containing alloys, the observed type(s) of primary phase(s) after DTA experiments and melt conditioning agree, however, partly differ from CALPHAD based predictions. The appearance of the primary phases in the alloys quenched after melt conditioning is illustrated in Fig. 8.4.

In alloys with Mn, the difference of the thermodynamically predicted and observed liquidus temperatures is remarkable (>50 K) although the primary phase is the α_c phase in both cases.

In alloys with Cr, the difference of the liquidus temperatures is less pronounced compared to Mn-containing alloys. However, the predicted phases for all compositions differ from the observed phases. The α_c phase is observed as the only primary phase for x_Cr = 0.3, 0.6, 0.9 at%. For x_Cr = 1.2, 1.5 at%, the (Al,Si)₂Cr, the Al₁₃Cr₄Si₄ and the α_c phase occur before the onset of solidification of (Al)-solid solution. The α_c phase is not implemented in the description of the Al–Si–Fe–Cr system in the TCAL4 database [53] while the (Al,Si)₂Cr is implemented but not predicted for the present alloy compositions. Furthermore, the observed (Al,Si)₂Cr phase should generally dissolve during solidification and should not be present together with Al in the solid state [54]. Instead, experimentally not observed formation of Al₅Cr and Al₄₅Cr₇ is predicted by thermodynamic calculations employing the TCAL4 database [53].

It can be concluded that the databases for prediction of the formation of primary intermetallic phases by CALPHAD generally need some further fine-tuning e.g. by implementation of the α_c phase in the Al–Si–Fe–Cr system which is a true quaternary phase in this system not existing in any of the ternary subsystems.

In view of the sludge factor approach, the agreement of the liquidus temperatures T_L is surprisingly good for the temperature estimation from Eq. 8.3 [36] with experimental values as long as an intermetallic phase is the primary phase (Fig. 8.3). The sludge factor approach was initially introduced for Cu-containing, secondary Al–Si alloys with similar Si content. The agreement to the temperature estimations from Eqs. 8.2 and 8.4 [37, 51] is less good which might be related to a remarkably higher Si content in the alloys in [51] compared to the present alloy compositions. Furthermore, it has been observed that Cu has a minor and Mg has no effect on the temperature of sludge formation [36] leading to the good agreement of the sludge factor approach in the present Cu-free Al–Si alloys. From this observation, it can be further assumed that the sludge factor approach can be applied to commercial Al–Si alloys in presence of alloying elements that do not take part in the sludge formation like Zn, Ni, Pb, Sn and Ti [36, 38] or when the amount of the element is negligible. However, in case of elements like Sc, B and V that promote the α_c-phase formation [1], or if the effect of elements is unknown like Be, Sr, Co, Mo and S the sludge factor approach in form of Eqs. 8.1–8.4 should be used with care.

The solubility of the elements Fe, Mn, Cr and Mg in the liquid Al at 620 °C determines the remaining minimum amount of these elements in the alloy after separation of the sludge from the liquid metal. These amounts, determined by EDS area measurements, are summarized in the quasi-ternary (Al,Si)–Fe–M projection of the Al-rich corner of the corresponding quaternary phase diagram (Fig. 8.5a) for the alloy series Al7.1Si(1.5-x_M)Fe(x_M)M with M = Mg, Mn, Cr and Al7.1Si1.5Fex_MM with M = Mn, Cr where x_M = 0, 0.3, 0.6, 0.9, 1.2 and 1.5. at% in both alloy series.

A triangular phase diagram of the A l S i F e M alloy. The length of the triangle ranges from 0 to 1.5. The plots for residual melt composition in alloy and initial melt composition are distributed on each side of the fit inside the triangle. b. 6 SEM slash B S E images of M g, M n, and C r samples.

For the evaluation of the success of Fe removal often relative Fe reductions are specified. However, those values depend on the initial Fe content. Consequently, it is recommended to use absolute elemental contents e. g. of the remaining melt in preference or in addition to the relative reductions. That facilitates the comparison of the success of melt conditioning for Fe removal with other studies.

In the Mg-containing alloys with x_Mg ≥ 0.6 at%, no solid phase was present at 620 °C and the chemical composition after melt conditioning equals the initial chemical alloy composition. In the other alloys, the formation of primary, Fe-containing, Mg-free intermetallic particles reduces the Fe content in the remaining material. In the ternary Al7.1Si1.5Fe alloy, 0.88 at% Fe remain dissolved in the liquid metal. The phase responsible for binding Fe in the alloys with x_Mg ≤ 0.3 at% is the α_h phase. Note that Mg as typical additional alloying element of Al–Si cast alloys is not bound by sludge particles and remains in the melt.

In case of alloys with Mn and Cr, the Fe content and the total transition-metal content in the remaining melt decreases with increasing x_Mn/(x_Mn + x_Fe) and x_Cr/(x_Cr + x_Fe) ratios in the alloys (Fig. 8.5a). That is attributed to the decreased ability of the melt to dissolve Mn and Cr. In presence of Mn and Cr, the α_c phase is responsible for binding Fe and also Mn and Cr in agreement with [14, 41, 44‐46]. Furthermore, a smaller amount of Cr than Mn for the same initial Mn and Cr content remains in the melt related to additionally binding Cr in the (Si,Al)₂Cr and the Al₁₃Cr₄Si₄ phase which, however, do not bind Fe in relevant amounts.

The optimum Mn and Cr content for Fe reduction sensitively depends on the actual alloy composition [41‐43]. Next to the experimentally determined residual melt composition, the chemical composition of the melt at 620 °C from thermodynamic calculations and from the sludge factor approach are illustrated in Fig. 8.5a. In the latter case, the minimum content of Fe in the melt is reached according to [36] when the sludge factor equals 1. In agreement with experimental observation, own and literature, reported in [14, 41‐45], the solubility of Fe decreases with increasing x_Mn/(x_Fe + x_Mn) ratio and x_Cr/(x_Fe + x_Cr) ratio. However, the optimum ratio with respect to the residual transition-metal content is controversial. While [43] achieved minimum x_Fe + x_Mn contents as low as 0.3 at% for an initial x_Mn:x_Fe ratio of 3:1, in the present investigation, a continuous decrease of the transition-metal content towards the ternary Al–Si–Mn and Al–Si–Cr alloy composition was observed. In addition to the chemical composition, the residual Fe, Mn and Cr content in the melt decreases with decreasing conditioning temperature [41, 46, 47] and approaches the lowest achievable content after a critical holding time. Then, the maximum amount of Fe and Mn are bound in primary, intermetallic particles [42‐44, 46] and their maximum volume fraction is reached. Consequently, a lower residual Fe and Mn content in the present study was achieved compared to similar alloys if these were conditioned at higher temperatures [41, 42, 46] or shorter times [42, 46].

Figure 8.5b depicts the intermetallic phases formed with the Fe, Mn and Cr which have remained in the melt despite optimum melt-conditioning at 620 °C and quenching. These intermetallic phases have formed during the co-dendritic and co-eutectic stage and are located in the Al–Si eutectic regions. Corresponding to the remaining contents of Fe, Mn and Cr, the volume fraction of intermetallic particles is lower in alloys with initially high x_Mn/(x_Mn + x_Fe) and x_Cr/(x_Cr + x_Fe) ratios. In case of the Mg-containing alloys, the reduced amount of the Fe-containing, intermetallic particles reflects the initially lower Fe content with increasing x_Mg/(x_Mg + x_Fe) ratio, although the Fe content is not (x_Mg ≥ 0.6 at%) or not remarkably (x_Mg = 0.3 at%) changed in presence of Mg. In Mn- and Cr-containing alloys, the reduced amount of the Fe-containing, intermetallic phases is attributed to the remarkably reduced Mn and Cr level, simultaneously to the reduced Fe level remaining in the melt after melt conditioning. Furthermore, although depending to some extent on the cooling rate, the morphology of the Mn- and Cr-containing, intermetallic particles is irregular and roundish compared to the plate-shaped particles in the eutectic region in alloys with low initial x_Mn/(x_Mn + x_Fe) and x_Cr/(x_Cr + x_Fe) ratios.

In this section results and conclusions from [15] about the microstructures that have formed under different cooling rates and x_Mn/(x_Mn + x_Fe) ratios in Al7.1Si(1.5-x_Mn)Fe(x_Mn)Mn with x_Mn = 0, 0.3, 0.375, 0.6 and 0.75 at% and a focus on the Fe-containing, intermetallic phases is presented. Mn is the most frequently used element for modification of the microstructure. Depending on the casting technique, relevant cooling rates range from 1–10 K/s in permanent mold casting to 10–100 K/s in high pressure die casting [55]. Modification of intermetallic particles into harmless microstructural components in order to deal with Fe impurities in secondary Al–Si casting alloys requires detailed understanding of the solidification path in presence of modifying elements and non-equilibrium solidification. The present data is further analyzed to provide a systematic understanding on the solidification path for a broad range of Fe- and Mn-containing Al–Si casting alloys. Thus, it is shown how seemingly controversial results of intermetallic phases in the microstructures of related alloys observed in other studies well integrate to give a consistent picture when considering the effect of cooling rate and presence of further elements especially Mn. Solid-state phase transformations that could occur during subsequent heat treatments are beyond the scope of the presented investigations. No signs of such transformation have been observed in [15], except of marginal transformation of the α_h phase into the β phase.

The microstructures of the Al7.1Si(1.5-x_Mn)Fe(x_Mn)Mn alloys solidified with cooling rates of 0.05, 1.4, 11.4 and 200 K/s are shown in Fig. 8.6. The type of phases corresponding to the intermetallic particles with different morphology and location within the microstructure and within the sample are indicated. The volume fractions of the intermetallic particles in the solidified microstructures are presented in Fig. 8.7. The intermetallic particles have been quantitatively analyzed of in view of specific morphologies and type of the phases.

A set of 12 SEM slash B S E images of A l, S i, F e, and M n alloys at different magnifications. The morphology of the alloy has a line structure with the formation of coarse polyhedral shapes at different sizes.

5 graphs, a to e, plot volume fraction versus cooling rate. a plots 2 decreasing trends. b and c plot 2 increasing trends and a decreasing trend. d and e plot an increasing trend and 2 decreasing trends. f. It exhibits the phases, morphology, and legends given in the graphs.

In the Mn-free alloy, at the edge of the slowest cooled samples, where the alloy is in contact with the surrounding oxide film, coarse polyhedral particles of the α_h phase are present. Plate-shaped particles appear associated with the Al–Si eutectic, i.e. interdendritic, region, but partly reach into the Al-dendrites. All other higher cooling rates have led to microstructures without any α_h-phase particles which is attributable to retarded nucleation of the α_h phase. Only plate-shaped particles occur. The plate-shaped particles in samples solidified with intermediate cooling rates appear within the Al-dendrites. In samples solidified with the highest cooling rate, the plate-shaped particles are located within the Al–Si eutectic region. These plate-shaped particles consist mainly of the β phase after solidification with low cooling rates and of the δ phase after solidification with high cooling rates. The observed suppression of the α_h phase and the formation of the δ phase under high cooling rates contradict thermodynamically calculated solidification paths for equilibrium and Scheil solidification conditions which predict the α_h and β phase or solely β phase.

In Mn-containing alloys, at the edge of the samples cooled up to intermediate cooling rates, coarse polyhedral and coarse dendritic particles of the α_c phase form during solidification. After solidification with low cooling rates plate-shaped particles accompany the coarse particles located in association with the Al–Si eutectic region. Under the highest cooling rates, plate-shaped particles also form, however, with smaller size and within the Al–Si eutectic. At intermediate cooling rates Chinese-script particles have formed being located within the Al-dendrites. The α_c-phase particles form as the only intermetallic microstructural component above a Mn-content dependent cooling rate. The relevant range of cooling rates for the solely α_c-phase formation increases with increasing Mn content. In presence of Mn, the discrepancy to the thermodynamically calculated solidification paths is less pronounced. The equilibrium calculation by trend agrees with the solidification at low cooling rates. Scheil solidification calculations are in accordance with the solidification at low to intermediate cooling rates. However, the formation of the δ phase is not predicted by the thermodynamic calculations similar to the Mn-free Al–Si alloys.

These observed deviations from the calculated equilibrium and Scheil solidification path substantiate the reported importance of kinetic effects during solidification of secondary Al–Si alloys [19, 56‐58]. These kinetic effects can include:

The influence of such kinetic effects during non-equilibrium solidification cannot be fully accounted for by Scheil solidification conditions or alternatively by setting a certain phase or several phases dormant. Therefore, it is suggested to construct so-called apparent liquidus projections (Fig. 8.8) which provide a tool to estimate the phase formation during solidification. Their construction is based on the experimental microstructures, which are analyzed according to the present phases and their relative location in the microstructures and in the sample.

6 phase diagrams of the A l F e S i system for thermodynamic calculation, intermediate cooling rate, high cooling rate, and low cooling rate. Each system has different phases for theta, alpha, gamma, beta, and delta, A l, and S i.

As obvious from the apparent liquidus projections, the kinetic effects on the microstructure formation in secondary Al–Si alloys are pronounced. It can be assumed that two aims are followed by addition of further elements and Mn and solidification with a high cooling rate: avoiding sludge formation and suppressing the formation of plate-shaped particles. However, some trade-offs must be considered when the parameters are optimized.

The general alloy composition including the Si content and the total transition metal content, as well as the x_Mn/(x_Mn + x_Fe) ratio, determine the potentially forming phases. Aiming at an optimized microstructure with a low intermetallic phase fraction, without plate-shaped particles and without sludge particles, Mn should be present, but the x_Mn/(x_Mn + x_Fe) ratio and total transition metal content should be as low as required and the minimum cooling rate should be chosen according to the chemical composition of the alloy.