Dictra

University of Thessaly
School of Engineering
Department of Mechanical Engineering
Simulation and design

of the homogenization process
of 6xxx extrudable Aluminum alloys
Development of design rules for high extrudability
by
Panagiota I. Sarafoglou
Dipl. Chemical Engineering, NTUA
M.Sc. Materials, NTUA
Submitted
in partial fulfilment of the requirements of the degree of
DOCTOR OF PHILOSOPHY
Supervisor
Professor Gregory N. Haidemenopoulos
Volos 2016
Institutional Repository - Library & Information Centre - University of Thessaly

26/03/2018 03:27:43 EEST - 137.108.70.7
2

26/03/2018 03:27:43 EEST - 137.108.70.7
Thesis examination committee
1. Prof. G. N. Haidemenopoulos Department of Mechanical Engineering
(Thesis advisor)
2. Prof. N. Aravas
Department of Mechanical Engineering
National Technical University of Athens

3. Prof. D. Manolakos
School of Mechanical Engineering
University of Patras
4. Prof. S. Pantelakis Department of Mechanical Engineering
and Aeronautics
5. Prof. D. I. Pantelis School of Naval Architecture and Marine
Engineering
6. Prof. E. Pavlatou
School of Chemical Engineering
Drexel University
7. Prof. A. Zavaliangos Department of Materials Science and
Engineering

26/03/2018 03:27:43 EEST - 137.108.70.7
2

26/03/2018 03:27:43 EEST - 137.108.70.7
Simulation and design of the homogenization process of 6xxx
extrudable Aluminum alloys
Development of design rules for high extrudability
Abstract
New demanding light-weight applications in the automotive and aircraft sectors require the
development of high-strength Al-alloy extrusions based on the Al-Mg-Si system (6xxx series). The
increase of strength is possible with higher alloying with Mg and Si, in order to form higher
amounts of the strengthening phase Mg2Si (e.g. 6061 and 6082 alloys). However increased
alloying deteriorates the extrudability, leading to extremely low extrusion speeds for the high-
strength alloy systems. In the other end of the spectrum, i.e. low-alloy low-strength applications
(e.g. 6060 or 6063 alloys), the main industrial requirement is to increase the extrusion speed, in
order to increase production rates. In both cases, it appears that extrudability is the key property
that should be controlled and improved through a carefully designed homogenization process. So
far the problem has been tackled in a rather empirical way, involving intense trial-and-error
laboratory or even industrial experiments to figure out the effect of alloy chemistry and
homogenization process parameters such as the homogenization temperature and time and the
cooling rate following homogenization. In most cases only individual aspects of the process are
considered and the effect of prior processing is often neglected.
The design of new high-strength or high extrudability 6xxx alloys calls for an integrated
approach considering all parts of the process chain and their effect on extrudability. The aim of
this doctoral thesis has been to apply a simulation-based approach, validated by relevant
experimental data, in order to describe the effect of alloy chemistry and homogenization process
conditions on the material microstructure prior to extrusion. In this way it was possible to shape
design rules (in terms of composition and processing) for the development of either high-
strength or high extrudability alloys.
Computational alloy thermodynamics and kinetics were applied to predict the rate of phase
transformations, which shape the microstructure across the process chain. This approach is
extremely powerful and was used to simulate a wide range of different phenomena such as
solidification and microsegregation during casting, dissolution of Mg2Si, removal of
microsegregation and transformation of iron intermetallics during homogenization and finally re-
precipitation during cooling from the homogenization temperature.
More specifically computational alloy thermodynamics, based on the CALPHAD approach,
has been applied to perform Scheil-Gulliver solidification simulations. The resulting
microsegregation of elements and phases in the as-cast microstructure were calculated as a
function of alloy composition. The results of the simulations were confirmed experimentally by
quantitative image analysis for the measurement of phase fractions. The variation of the mole
fractions of the extrudability-limiting β-AlFeSi phase and the strengthening Mg2Si phase with
alloying elements has been mapped over the useful range (0-1.2 mass%) in the Mg-Si
composition space. The constructed maps indicate that low mole fractions of β-AlFeSi are
associated with lower Si and higher Mg compositions. On the other hand, high mole fractions of
Mg2Si are associated with both higher Si and Mg compositions, with Mg possessing a stronger

26/03/2018 03:27:43 EEST - 137.108.70.7
effect. Construction of maps for different levels of Mn has shown that addition of Mn could allow
for higher alloying with Mg and Si, in order to obtain higher amount of Mg2Si, without at the
same time increasing the β-AlFeSi intermetallic phase in the as-cast microstructure.
The Dual grain model (DGM) has been developed to treat the homogenization process in
multicomponent and multiphase Al-alloys exhibiting a large variability of the as-cast grain size.
With the model it has been possible to simulate the temporal and spatial evolution of phase
fractions and element concentrations during homogenization. Regarding the evolution of phase
fractions during homogenization, the predictions of the DGM model have been validated
experimentally with XRD analysis. The evolution of the β-AlFeSi→α-AlFeSi transformation is
predicted by the DGM and is confirmed by XRD in the same time scale. The DGM predictions
regarding the evolution of Fe and Mg concentrations with homogenization time are in excellent
agreement. There are only some discrepancies in the profiles of Si and Mn. The DGM predicts the
fast dissolution of Mg2Si during homogenization and its re-precipitation during cooling. The DGM
can describe the temporal and spatial evolution of the β-AlFeSi→α-AlFeSi transformation. The
spatial evolution exhibits an exact spatial correspondence. Both the Mg2Si dissolution and the β-
AlFeSi→α-AlFeSi transformation are faster in the smaller grain as predicted by the DGM. In
addition the concentration profiles of the elements homogenize faster in the smaller grain,
indicating that a fine grain size in the as-cast microstructure accelerates the homogenization
process. The DGM predicts correctly the effect of homogenization temperature. The rate of both,
Mg2Si dissolution and β-AlFeSi→α-AlFeSi transformation increase with the homogenization
temperature. A preliminary attempt to develop homogenization process maps has been
performed using the DGM. These maps are, at present, based on the β-AlFeSi→α-AlFeSi
transformation and can be used for the design of the homogenization heat treatment. The
quantification of the homogenization state was achieved by determine specific microstructural
indices, such as the aspect ratio and the circularity of the intermetallic phases. In this way it was
possible to quantify the shape evolution of the intermetallic compounds during homogenization.
Precipitation during cooling from the homogenization temperature was simulated with the
Kampman-Wagner Numerical (KWN) precipitation model. The relevant particle size distribution
(PSD), evolution of volume fraction and particle size during cooling was calculated. The results
were introduced in a suitable strength model to calculate the resulting hardness. The simulation
results are in satisfactory agreement with experimental data. The hardness of the homogenized
billets was investigated experimentally. Regarding the effect of homogenization cooling rate,
water quenching results in higher hardness relative to air cooling and forced air cooling. The
hardness difference increases with alloying in the order 6063→6005→6082. The effect is
attributed to the higher amounts of Mg and Si retained in solid solution during cooling. The effect
of excess Si on the precipitation of β΄-Mg2Si was determined using the KWN precipitation model.
The resulting shift in the continuous cooling precipitation (CCP) diagram was calculated. This
allows the design of a suitable cooling program in order to avoid the precipitation of β-Mg2Si and
allow the precipitation of β΄-Mg2Si. This allows the use of a lower preheating temperature prior
to extrusion and leads to the use of higher press loads in order to obtain higher extrusion speeds.
Specific design rules were formulated, regarding the as-cast microstructure, the
homogenization temperature and time as well as the cooling from the homogenization
temperature in order to obtain high extrudability. The microstructure developed during DC
casting is affected by the solidification conditions. Grain size and dendrite arm spacing (DAS) are
ii

26/03/2018 03:27:43 EEST - 137.108.70.7
the most important factors. They can be controlled by the casting speed (solidification rate) and
seeding for grain refining. A smaller grain size and DAS improves the strength and also shortens
the diffusion distances, leading to a more efficient homogenization for a given homogenization
cycle. Phase fraction mapping, developed in this thesis, can be used for the selection of alloy
compositions to minimize the undesirable β-AlFeSi intermetallic in the as-cast microstructure, in
order to achieve a good starting point for the homogenization to follow. In addition, it has been
shown that Mn reduces the amount of β-AlFeSi for a given Mg-Si-Fe combination. With the DGM
model developed in this thesis, it is possible to determine the required time for Mg2Si dissolution
and the transformation β→α-AlFeSi as a function of homogenization temperature and alloy
composition. The DGM model also indicated that a small as-cast grain size accelerates the
kinetics of homogenization. Microstructural indices, such as the aspect ratio and the circularity
can be used to characterize the homogenization stage quantitatively. A fully homogenized billet,
with the potential for high extrudability should have all β-AlFeSi transformed to α-AlFeSi, with
necklace morphology and with aspect ratio and circularity approaching unity. In addition all Mg2Si
should be dissolved and Mg and Si should be distributed as uniformly as possible in the grain
interiors. The cooling from the homogenization temperature should be designed so that
precipitation of β-Mg2Si is avoided and only precipitation of β΄-Mg2Si takes place. A CCP diagram
like the one developed in this thesis for a 6060 alloy can be used to design the cooling cycle.
Finally it is important to match the preheating conditions to the homogenization cooling
conditions in order to get the most benefit of homogenization. β-Mg2Si precipitates require
higher temperatures to dissolve (higher solvus) and have a detrimental effect on hot ductility. On
the other hand the β΄-Mg2Si ppts, are finer, semi – coherent with the matrix and have a lower
solvus temperature. Thus a lower preheating temperature can be employed in combination with
a high extrusion speed. This will raise the exit temperature from the extrusion press to a value
sufficient for complete dissolution of Mg2Si.
The integrated simulation of microsegregation during casting, dissolution of Mg2Si and
transformation of iron intermetallics during homogenization as well as re-precipitation of Mg2Si
during homogenization cooling, enables the computer-aided development of new alloys and the
design of the associated homogenization treatments for high extrudability at a fraction of time
and cost associated with empirical development.
iii

26/03/2018 03:27:43 EEST - 137.108.70.7
iv

26/03/2018 03:27:43 EEST - 137.108.70.7
Προσομοί ωση καί σχεδίασμο ς της κατεργασί ας ομογενοποί ησης
σε δίελα σίμα κρα ματα αλουμίνί ου της σείρα ς 6xxx.
Δίαμο ρφωση κανο νων σχεδίασμου γία υψηλη δίελασίμο τητα
Περίληψη
Οι σύγχρονες ελαφρές κατασκευές τόσο στην αυτοκινητοβιομηχανία όσο και την
αεροπορική βιομηχανία απαιτούν την ανάπτυξη διελάσιμων κραμάτων αλουμινίου υψηλής
αντοχής με βάση το σύστημα Al-Mg-Si (σειρά 6xxx). Η αύξηση της αντοχής είναι δυνατή με την
αύξηση του επιπέδου κραμάτωσης σε Mg και Si, ώστε να σχηματιστεί μεγαλύτερο ποσοστό από
τη φάση ισχυροποίησης Mg2Si (π.χ στα κράματα 6061 και 6082). Ωστόσο η αυξημένη
κραμάτωση προκαλεί υποβάθμιση της διελασιμότητας και οδηγεί σε εξαιρετικά χαμηλές
ταχύτητες διέλασης στα κράματα αυτά. Στην άλλη μεριά του φάσματος, δηλαδή σε εφαρμογές
όπου απαιτείται χαμηλή αντοχή (π.χ. κράματα 6060 και 6063), η βασική βιομηχανική απαίτηση
είναι η αύξηση της ταχύτητας διέλασης, η οποία με τη σειρά της οδηγεί σε αύξηση της
παραγωγικότητας. Και στις δύο περιπτώσεις φαίνεται ότι η διελασιμότητα αποτελεί την
«ιδιότητα-κλειδί» που πρέπει να ελεγχθεί και να βελτιωθεί μέσω ενός προσεκτικού σχεδιασμού
της κατεργασίας ομογενοποίησης. Μέχρι τώρα το πρόβλημα αυτό έχει αντιμετωπιστεί
περισσότερο εμπειρικά, με πειραματικές διαδικασίες σε εργαστηριακή αλλά και σε βιομηχανική
κλίμακα, σε μία προσπάθεια να διερευνηθούν τόσο η επίδραση της χημικής σύστασης όσο και οι
συνθήκες της κατεργασίας όπως η θερμοκρασία, ο χρόνος και ο ρυθμός ψύξης της
ομογενοποίησης. Στις περισσότερες περιπτώσεις έχουν μελετηθεί μεμονωμένες επιδράσεις ενώ
συνήθως δεν μελετάται η επιρροή του ιστορικού των κατεργασιών.
Ο σχεδιασμός νέων κραμάτων της σειράς 6xxx με υψηλή αντοχή ή υψηλή διελασιμότητα
απαιτεί την εφαρμογή μιας ολοκληρωμένης μεθοδολογίας, η οποία να περιλαμβάνει όλους τους
κρίκους της αλυσίδας κατεργασιών και την επίδρασή τους στη διελασιμότητα.
Ο στόχος της παρούσας Διδακτορικής Διατριβής ήταν η εφαρμογή μιας μεθοδολογίας
προσομοίωσης, επικυρωμένης με πειραματικά δεδομένα, για την περιγραφή της επίδρασης της
χημικής σύστασης και των συνθηκών ομογενοποίησης στην μικροδομή του κράματος πριν τη
διέλαση. Με τον τρόπο αυτό διαμορφώθηκαν κανόνες σχεδιασμού, που αφορούν στη σύσταση
και τη κατεργασία, για την ανάπτυξη κραμάτων υψηλής διελασιμότητας.
Οι τεχνικές της υπολογιστικής θερμοδυναμικής και κινητικής κραμάτων εφαρμόστηκαν για
την πρόβλεψη της ταχύτητας των μετασχηματισμών των φάσεων, που διαμορφώνουν την
μικροδομή κατα μήκος της αλυσίδας κατεργασιών στα διελάσιμα κράματα αλουμινίου. Αυτή η
μεθοδολογία είναι πολύ αποτελεσματική και εφαρμόστηκε για την προσομοίωση διεργασιών
όπως η στερεοποίηση και η διαμόρφωση του μικροδιαφορισμού κατά τη χύτευση, η απομείωση
του μικροδιαφορισμού, η διαλυτοποίηση του Mg2Si και ο μετασχηματισμός των ενδομεταλλικών
ενώσεων του σιδήρου κατά την ομογενοποίηση αλλά και η επανακαθίζηση του Mg2Si κατά την
ψύξη από τη θερμοκρασία ομογενοποίησης.
Πιο συγκεκριμένα, η υπολογιστική θερμοδυναμική, βασισμένη στη μεθοδολογία CALPHAD,
εφαρφόστηκε για την διενέργεια προσομοιώσεων στερεοποίησης τύπου Scheil-Gulliver. Με το
τρόπο αυτό υπολογίστηκε ο μικροδιαφορισμός των στοιχείων και των φάσεων στη χυτή
μικροδομή σε συνάρτηση με τη χημική σύσταση του κράματος. Τα αποτελέσματα των

26/03/2018 03:27:43 EEST - 137.108.70.7
προσομοιώσεων επικυρώθηκαν με πειραματικά δεδομένα ποσοτικής μεταλλογραφίας, που
αφορούσαν στη μέτρηση του ποσοστού των φάσεων. Έτσι διαμορφώθηκαν χάρτες που
απεικονίζουν τα ποσοστά της φάσης ισχυροποίησης Mg2Si και της φάσης που περιορίζει τη
διελασιμότητα, την β-AlFeSi, σε συνάρτηση με τα κραματικά στοιχεία Mg και Si. Οι χάρτες αυτοί
υποδεικνύουν ότι χαμηλή κραμάτωση σε Si και υψηλή κραμάτωση σε Mg οδηγεί σε χαμηλά
ποσοστά της ανεπιθύμητης φάσης β-AlFeSi και ενδεχόμενη αύξηση της διελασιμότητας. Από την
άλλη μεριά, υψηλότερη κραμάτωση τόσο σε Mg όσο και σε Si οδηγεί σε υψηλά ποσοστά της
φάσης ισχυροποίησης Mg2Si και ενδεχόμενη αύξηση της αντοχής. Η κατασκευή των χαρτών για
διαφορετικά ποσοστά Mn έδειξε ότι η προσθήκη Mn επιτρέπει την αύξηση της κραμάτωσης σε
Mg και Si για την αύξηση της αντοχής, χωρίς την ταυτόχρονη αύξηση της ανεπιθύμητης φάσης β-
AlFeSi στη χυτή μικροδομή.
Για την επίλυση του προβλήματος της ομογενοποίησης σε πολυσυστατικά και πολυφασικά
κραματικά συστήματα αλουμινίου, που παρουσιάζουν μεγάλη διασπορά στο μέγεθος του
κόκκου, ανπτύχθηκε το μοντέλο ομογενοποίησης διπλού κόκκου (Dual Grain Model, DGM). Με
το μοντέλο αυτό προσδιορίστηκε τόσο η χρονική όσο και η χωρική εξέλιξη του ποσοστού των
διαφορετικών φάσεων της μικροδομής καθώς και του μικροδιαφορισμού κατά τη διάρκεια της
ομογενοποίησης. Τα αποτελέσματα των προσομοιώσεων επικυρώθηκαν με αντίστοιχα
δεδομένα από πειραματικές τεχνικές XRD και SEM/EDX. Το DGM περιγράφει την χρονική και
χωρική εξέλιξη του μετασχηματισμού β-AlFeSi→α-AlFeSi. Η χωρική εξέλιξη παρουσιάζει μια
ακριβή αντιστοιχία μεταξύ των ποσοστών της α και β-AlFeSi. Επίσης το DGM προβλέπει την
ταυτόχρονη διαλυτοποίηση της φάσης Mg2Si και την επανακαθίζησή της κατά τη ψύξη. Όλοι οι
μετασχηματισμοί παρουσιάζουν μεγαλύτερες ταχύτητες στο μικρότερο κόκκο του DGM, γεγονός
που υποδεικνύει ότι μία εκλέπτυνση του κόκκου της χυτής μικροδομής μπορεί να επιταχύνει την
ομογενοποίηση. Επίσης το DGM προβλέπει σωστά την επίδραση της θερμοκρασίας
ομογενοποίησης. Οι ταχύτητες τόσο της διαλυτοποίησης του Mg2Si όσο και του
μετασχηματισμού β-AlFeSi→α-AlFeSi αυξάνονται με την αύξηση της θερμοκρασίας. Τέλος
πραγματοποιήθηκε μια αρχική προσπάθεια κατασκευής ενός χάρτη ομογενοποίησης
(homogenization process map) με τη χρήση του DGM. Ο χάρτης αυτός, προς το παρόν, βασίζεται
στο χρόνο περάτωσης του μετασχηματισμού β-AlFeSi→α-AlFeSi και μπορεί να χρησιμοποιηθεί
για το σχεδιασμό της κατεργασίας ομογενοποίησης. Ταυτόχρονα πραγματοποιήθηκε και η
ποσοτικοποίηση της κατάστασης ομογενοποίησης με τον προδσιορισμό μεταλλογραφικών
δεικτών όπως η αναλογία πλάτους/μήκος και η κυκλικότητα των ενδομεταλλικών ενώσεων. Με
τον τρόπο αυτό ποσοτικοποιήθηκε η εξέλιξη της μορφολογίας των ενδομεταλλικών ενώσεων
κατά τη διάρκεια της ομογενοποίησης.
Η καθίζηση κατά τη διάρκεια της ψύξης από τη θερμοκρασία ομογενοποίησης
προσομοιώθηκε με το μοντέλο Kampman-Wagner-Numerical (KWN). Έτσι προσδιορίστηκε η
κατανομή μεγέθους σωματιδίων και η εξέλιξη του κλάσματος όγκου και του μεγέθους των
σωματιδίων. Τα αποτελέσματα της προσομοίωσης επιβεβαιώθηκαν από πειραματικές μετρήσεις
της σκληρότητας.
Η επίδραση του πρόσθετου Si (Excess Si) στην διαμόρφωση του διαγράμματος καθίζησης
υπό συνεχή ψύξη (CCP diagram) που αφορά στη καθίζηση της μετασταθούς φάσης β΄- Mg2Si
πραγματοποιήθηκε με το KWN. Αυτό επέτρεψε τον σχεδιασμό ενός προγράμματος ψύξης για
την αποφυγή σχηματισμού της φάσης ισορροπίας β- Mg2Si και τον σχηματισμό της επιθυμητής
φάσης β΄- Mg2Si. Αυτό οδηγεί στην εφαρμογή χαμηλότερης θερμοκρασίας προθέρμανσης πριν
vi

26/03/2018 03:27:43 EEST - 137.108.70.7
τη διέλαση και επιτρέπει την εφαρμογή μεγαλύτερων φορτίων στην πρέσσα διέλασης με
σημαντική αύξηση της ταχύτητας διέλασης.
Διαμορφώθηκαν κανόνες σχεδιασμού για την επίτευξη υψηλής διελασιμότητας. Οι
κανόνες αυτοί αφορούν στη κραμάτωση, τη χυτή μικροδομή, την ομογενοποίηση και την ψύξη
ομογενοποίησης. Μικρότερο μέγεθος κόκκου στη χυτή μκροδομή επιτρέπει την αύξηση της
κινητικής κατά την ομογενοποίηση οδηγώντας σε μικρότερους χρόνους ομογενοποίησης. Το
μέγεθος του κόκκου μπορεί να ελεγχθεί μέσω της ταχύτητας στερεοποίησης και του
εμβολιασμού του υγρού μετάλλου για ετερογενή πυρήνωση κατά τη στερεοποίηση. Οι χάρτες
του ποσοστού των φάσεων στη χυτή μικροδομή επιτρέπουν την επιλογή της κραμάτωσης σε Mg
και Si για την μείωση της ανεπιθύμητης β-AlFeSi. Αυτό αποτελεί ένα καλό αρχικό σημείο για την
ομγενοποίηση που ακολουθεί. Με το DGM είναι δυνατόν να υπολογιστεί ο απαιτούμενος
χρόνος ομογενοποίησης. Με την ποσοτικοποίηση της κατάστασης ομογενοποίησης μέσω
μεταλλογραφικών δεικτών φαίνεται ότι μια πλήρως ομογενοποιημένη κολώνα αλουμινίου για να
έχει υψηλή διελασιμότητα πρέπει οι ενδομεταλλικές φάσεις να είναι σφαιροποιημένες και να
παρουσιάζουν την μορφολογία neckless. Η ψύξη από τη θερμοκρασία ομογενοποίησης πρέπει
να σχεδιαστεί ώστε να αποφευχθεί ο σχηματισμός της β- Mg2Si και αντ’ αυτής να σχηματιστεί η
μετασταθής φάση β΄- Mg2Si. Αυτό οδηγεί σε χαμηλότερες θερμοκρασίες προθέρμανσης και
επιτρέπει την επιβολή μεγαλύτερων φορτίων στη πρέσσα για την επίτευξη υψηλής ταχύτητας
διέλασης.
Η ολοκληρωμένη προσομοίωση του μικροδιαφορισμού κατά τη στερεοποίηση, της
διαλυτοποίησης του Mg2Si και του μετασχηματισμού των ενδομεταλλικών ενώσεων του
σιδήρου β-AlFeSi→α-AlFeSi κατά την ομογενοποίηση καθώς και της επανακαθίζησης του Mg2Si
κατά την ψύξη επιτρέπει τον σχεδιασμό τόσο νέων διελάσιμων κραμάτων αλουμινίου αλλά και
της αντίστοιχης κατεργασίας ομογενοποίησης για την επίτευξη υψηλής διελασιμότητας, σε
κλάσμα μόνο του χρόνου και κόστους, που απαιτείται για τον αντίστοιχο εμπειρικό σχεδιασμό.
vii

26/03/2018 03:27:43 EEST - 137.108.70.7
viii

26/03/2018 03:27:43 EEST - 137.108.70.7
Acknowledgements
The completion of this dissertation would not have been possible without the help of
several people whom I would like to thank;
I am grateful to my advisor Prof. G.N. Haidemenopoulos for trusting me and providing me
this opportunity to work on his alloyneering concept and for his encouragement and guidance
both personally and professionally. I owe him many thanks for his friendship and teaching during
these years.
I am very grateful to Prof. N. Aravas for serving in my thesis advisory committee and Prof.
D. Manolakos for serving in my thesis advisory committee and for providing guidance during my
years at NTUA.
I would like to acknowledge Professors D. Pantelis, E. Pavlatou, S. Pantelakis and
A.Zavaliangos for serving in my thesis examination committee.
I would like to express my gratitude to the Aluminium of Greece for providing the materials
for my thesis and especially El. Grigoriou, A. Triantafyllou and K. Liapis for the fruitful discussions
regarding the thesis.
I would like to thank the staff of the Laboratory of Materials of the University of Thessaly
but especially Eleni Kamoutsi and Eleana Pappa. During my thesis I had a great support from
them.
I am also glad to have worked with undergraduate students but also friends Spyros –
Alexandros Tousias, Margianna Tzini, John Aristeidakis and John Fanikos. They have been great
partners for me and I wish the best for them.
I would like to thank Lin Shang and Andreas Stieben (RWTH- Aachen) as well as Helen Rosolymou
(Chemical Engineering, NTUA) for their assistance with XRD measurements and Dr. Ch.Sarafoglou
(Naval Architecture & Marine Engineering) for her assistance with SEM/EDX.
Living in Volos has been exciting and I would like to thank my special friends:
The very first person I met here and best friend since then, Dimitra for just being who she is.
My amazing road-trip friends, Latharo, Vagner and Apostolis for their long coffee hours. Their
unconditional support has been essential all these years in many ways. My dear lawyer and friend
Helen and Dimitris for his music. I am not sure I would have done my thesis without them.
My friends from my NTUA years, Antonis Stefanidis, Kostas Ntaskas, Giorgos Chortis, for
being by my side.
I am grateful to my unique in many ways big perfect family. They have cherished with me
every great moment and supported me whenever I needed it. And I need to thank them in Greek
now,
ix

26/03/2018 03:27:43 EEST - 137.108.70.7
Ευχαριστώ την οικογένειά μου. Πρώτα από όλους τους γονείς μου Ιωακείμ και Άννα
Σαράφογλου καθώς τους θαυμάζω ξεχωριστά σαν ανθρώπους αλλά και για την ανιδιοτελή τους
αγάπη και στήριξη των σπουδών μου. Υπήρξαν το μεγαλύτερο κίνητρο για μένα όλα αυτά τα
χρόνια. Ευχαριστώ την αδερφή μου Χαρά γιατί είναι αυτή που είναι και γιατί υπήρξε για μένα το
πρότυπο ανθρώπου και επιστήμονα και μου δίνει το κουράγιο να συνεχίζω. Το Σταύρο που χωρίς
αυτόν δεν θα ήμουν εδώ σήμερα καθώς πρώτος αυτός πίστεψε σε μένα. Τον αδερφό μου Κώστα
και την Αμαλία για την μουσική παιδεία που έχω σήμερα. Τον αδερφό μου Μπάμπη και τη Λέττα
για την μαθηματική μου περιέργεια. Τους ευχαριστώ για τα θαυμάσια ανίψια μου !
Την Άννα, τον Ιωακείμ, το Στράτο, τη Ζωή μου, το Στράτο και εσένα που έρχεσαι σύντομα!!
Αυτό που είμαι σήμερα οφείλεται στους γονείς και τα αδέρφια μου.
Last but not least my beloved Apostolis for all laughing, crying, travelling, fishing and finally
living…I would like to dedicate this thesis to him.
Gioula Sarafoglou
Volos 2016

26/03/2018 03:27:43 EEST - 137.108.70.7
Contents
1 Introduction ................................................................................................................................................... 9
1.1. Problem statement and aim of the thesis ................................................................................... 9
1.2. State-of-the-art ................................................................................................................................... 11
1.3. Methodology ........................................................................................................................................ 12
1.4. Innovative aspects of the thesis ................................................................................................... 16
1.5. Structure of the thesis ...................................................................................................................... 17
2. Microsegregation and Phase Fraction Mapping ............................................................................ 19
2.1. Introduction and literature review ............................................................................................. 19
2.2. Computational methodology ......................................................................................................... 21
2.2.1. Computational Alloy Thermodynamics ............................................................................ 21
2.2.2. Amendments in COST507 Al database ............................................................................. 27
2.2.3. Simulation of microsegregation .......................................................................................... 27
2.3. Experimental procedures for model validation ..................................................................... 31
2.4. Results and Discussion .................................................................................................................... 33
2.4.1. Equilibrium calculations......................................................................................................... 33
2.4.2. Scheil microsegregation simulations................................................................................. 35
2.4.3. Parametric study on 6060 alloy .......................................................................................... 37
2.4.4. Transformation of α to β-AlFeSi after solidification .................................................... 42
2.5. Experimental validation .................................................................................................................. 44
2.6. Effect of Mn on Mg2Si and β-AlFeSi phases .............................................................................. 53
2.7. Mapping of intermetallic phases .................................................................................................. 55
2.8. Conclusions .......................................................................................................................................... 58
2.9. References ............................................................................................................................................ 59
3. Simulation of Homogenization ............................................................................................................. 61
3.1. Introduction ......................................................................................................................................... 61
3.2. Literature review ............................................................................................................................... 62
3.3. Methodology ........................................................................................................................................ 66
3.3.1. Computational Method ........................................................................................................... 66
3.3.2. Experimental procedures ...................................................................................................... 72

26/03/2018 03:27:43 EEST - 137.108.70.7
3.4. Results and Discussion .................................................................................................................... 74
3.4.1. Dual Grain Model ....................................................................................................................... 74
3.4.2. Mg2Si dissolution ....................................................................................................................... 76
3.4.3. β-AlFeSi→α-AlFeSi transformation .................................................................................... 77
3.4.4. The effect of temperature and grain size ......................................................................... 79
3.4.5. Removal of the microsegregation of the alloying elements...................................... 81
3.4.6. Homogenization Process Map of extrudable 6xxx alloys .......................................... 82
3.4.7. Experimental validation ......................................................................................................... 83
3.4.8. Transformation β-AlFeSi → α-AlFeSi ................................................................................. 88
3.4.9. Evolution of elemental profiles during homogenization ........................................... 89
3.4.10. Profile and mapping for Mg .............................................................................................. 90
3.4.11. Profile and mapping for Si ................................................................................................. 91
3.4.12. Profile and mapping for Mn .............................................................................................. 92
3.5. Conclusions .......................................................................................................................................... 94
3.6. References ............................................................................................................................................ 95
4. Quantification of Homogenization state ........................................................................................... 97
4.1. Introduction ......................................................................................................................................... 97
4.2. Literature review ............................................................................................................................... 97
4.3. Methodology ........................................................................................................................................ 98
4.3.1. Quantitative metallography .................................................................................................. 99
4.3.2. Image analysis ......................................................................................................................... 100
4.3.3. Statistical Analysis ................................................................................................................. 102
4.4. Results and Discussion ................................................................................................................. 103
4.4.1. Metallographic Investigation ............................................................................................. 103
4.4.2. Characterization of the particles ...................................................................................... 104
4.5. Results and discussion .................................................................................................................. 106
4.6. Conclusions ....................................................................................................................................... 108
4.7. References ......................................................................................................................................... 108
5. Precipitation during Cooling from the Homogenization Temperature ............................. 109
5.1. Introduction ...................................................................................................................................... 109
5.2. Precipitation and hardening during cooling from the homogenization state ......... 111
2

26/03/2018 03:27:43 EEST - 137.108.70.7
5.2.1. Simulations with the KWN and Strength models ...................................................... 111
5.2.2. Experimental validation ...................................................................................................... 115
5.3. Hardening potential of homogenized billets ........................................................................ 115
5.3.1. Introduction ............................................................................................................................. 115
5.3.2. Experimental procedures ................................................................................................... 116
5.3.3. Results ........................................................................................................................................ 117
5.4. Effect of excess-Si on the CCP diagram ................................................................................... 127
5.4.1. Introduction ............................................................................................................................. 127
5.4.2. Methodology ............................................................................................................................ 128
5.4.3. Results ........................................................................................................................................ 130
5.5. Design rules for high extrudability .......................................................................................... 134
5.6. Conclusions ....................................................................................................................................... 137
5.7. References ......................................................................................................................................... 138
6. Conclusions ............................................................................................................................................... 139
7. Proposal for further research ............................................................................................................ 145

26/03/2018 03:27:43 EEST - 137.108.70.7
List of Figures
Figure 1.1: The process chain of extrudable Al-alloys ............................................................................ 10
Figure 2.1: The three stages during solidification of a binary alloy .................................................. 28
Figure 2.2: The composition of the solid at the interface Cs as a function of the fraction
solid fs that has formed, by considering a mass balance ................................................. 29
Figure 2.3: Transverse cross sections along the billet diameter ........................................................ 31
Figure 2.4:(a) Si isopleth for 0.45Mg, 0.03Mn and 0.2 Fe (mass%) and (b) Mg isopleth for
0.4Si, 0.03Mn and 0.2 Fe (L=liquid, F=FCC, α=α-AlFeSi, β=β-AlFeSi, M=Mg2Si) ..... 34
Figure 2.5: Isothermal section of Al-Mg-Si with 0.2Fe-0.03Mn (mass %) at 300oC,
(L=liquid, F=FCC, α=α-AlFeSi, β=β-AlFeSi, M=Mg2Si) ....................................................... 34
Figure 2.6: Typical solidification paths for a)6082, b)6063, c)6005, d)6060 ................................ 35
Figure 2.7: Calculated Solidus temperatures .............................................................................................. 36
Figure 2.8: Mole fraction of phases as a function of temperature during solidification of Al-
0.38Mg-0.4Si-0.2Fe-0.03Mn (mass%) .................................................................................... 36
Figure 2.9: Microsegregation of phases (a) and microsegregation of alloying elements (b)
of Al-0.38Mg-0.4Si-0.2Fe-0.03Mn (mass%) alloy during solidification .................... 37
Figure 2.10: Segregation of phases for 6060_1 and 6060_3 a) α-AlFeSi, b)β-AlFeSi, c)
Mg2Si, d) Si (Diamond) .................................................................................................................. 38
Figure 2.11: Segregation of phases in 6060-2 and 6060_4 alloys, a) α-AlFeSi, b)β-AlFeSi, c)
Mg2Si, d) Si (Diamond) .................................................................................................................. 39
Figure 2.12: Segregation of elements in 6060_1 and 6060_3 alloys a) Si, b)Mg, c)Fe, d)Mn ... 40
Figure 2.13: Segregation of elements in 6060_2 and 6060_4 alloys a) Si, b)Mg, c)Fe, d)Mn ... 41
Figure 2.14: The geometric model for the kinetic simulation using the DICTRA
computational kinetics software performed for the 6082 alloy................................... 43
Figure 2.15: The position of fcc/β-AlFeSi interface as a function of simulation (cooling)
time ....................................................................................................................................................... 43
Figure 2.16: The as-cast microstructure of 6082 alloy as observed in the SEM ........................... 44
Figure 2.17: Elemental area mapping (SEM/EDS) in 6082 alloy a) Al, b)Si, c)Mg, d)Fe,
e)Mn ..................................................................................................................................................... 46
Figure 2.18: Matrix composition adjacent to boundary intermetallics: (a) SEM micrograph,
(b) Scheil microsegregation profile and (c) EDS signal ................................................... 47
Figure 2.19: Typical microstructure of alloys (a) 6082, (b)6063 and (c)6060 ............................. 49
Figure 2.20: Typical morphology of intermetallic α-AlFeSi in 6082 known as “Chinese-
script” .................................................................................................................................................. 50
Figure 2.21: The (a) quaternary eutectic structure and spherical Mg2Si and eutectics of
6082 alloy in as-cast condition .................................................................................................. 50
Figure 2.22: Microstructures (left) and corresponding Image-J displays used for the
measurement of phase fractions ............................................................................................... 52
Figure 2.23: Mn-induced reduction of β-AlFeSi phase as a function of Si alloy content for
various levels of Mg in the alloy ................................................................................................ 54
Figure 2.24: Mn-induced reduction in Mg2Si phase as a function of Si alloy content for (a)
0.4-0.6 and (b) 0.8 1.2 Mg (mass %) in the alloy ................................................................ 55

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.25: Contour plots (maps) of fixed mole fractions of (a) β-AlFeSi and (b) Mg2Si
phases in the as-cast microstructure of Al-Mg-0.2Fe-Mn alloys (full lines 0.03
Mn, dotted lines 0.45 Mn, mass%) ........................................................................................... 56
Figure 2.26: Contour plots (maps) of fixed mole fractions β-AlFeSi (solid lines) and Mg2Si
(dotted lines) in as-cast microstructure of Al-Mg-Si-0.2Fe-Mn alloys for
(a) 0.03 Mn and (b) 0.45 Mn, mass%. Compositions for alloys 6060, 6063 and
6082 are indicated on the maps. 57
Figure 3.1: The Dual Grain Model considering the different grain sizes between
neighboring grains ......................................................................................................................... 67
Figure 3.2: Dispersed Phase model in DICTRA .......................................................................................... 68
Figure 3.3: The temperature – time profile of the homogenization process.................................. 72
Figure 3.4 Temporal evolution of phase transformations during the homogenization
process a) Isothermal holding at 540οC for 32 h and b) holding at 540ο C for
32h cooling down (500οC/2h), preheating 500οC/1h. ..................................................... 75
Figure 3.5: The spatial evolution of the Mg2Si phase from the grain boundaries and its re-
precipitation in the grain interiors .......................................................................................... 76
Figure 3.6: The spatial evolution of the β-AlFeSi→α-AlFeSi transformation a) the initial
profiles of β-AlFeSi and α-AlFeSi in the as-cast structure, b) the final profiles
after the completion of the transformation .......................................................................... 77
Figure 3.7: Spatial evolution of β→α-AlFeSi transformation during homogenization for times
a) 15 b) 45min c) 1h d)1 ½ h ......................................................................................................... 79
Figure 3.8: The weight fraction of a)Mg2Si phase and b)β-AlFeSi phase for 3 different
homogenization temperatures, 540, 560 and 580oC for homogenization time
t=2sec and 500sec respectively ................................................................................................. 80
Figure 3.9: The evolution of the concentration profiles of Mg, Si, Fe and Mn in the FCC
phase during homogenization at 540oC ................................................................................. 82
Figure 3.10: Homogenization process map of 6082 alloy based on the β-AlFeSi→α-AlFeSi
transformation ................................................................................................................................. 83
Figure3.11: The effect of the homogenization time of the 6082 aluminum alloy for 30min
2,4 and 8h ........................................................................................................................................... 85
Figure 3.12: SEM micrographs of 6082 alloy a) as- cast condition b) 0.5, c) 4h, d)8h, e)32h . 86
Figure 3.13: The homogenized microstructure of the 6060 aluminum ........................................... 87
Figure 3.14: a)Rounded α-AlFeSi particles and b) neckless morphology in SEM
micrographs ...................................................................................................................................... 87
Figure 3.15: XRD spectrum of the β-AlFeSi→α-AlFeSi transformation of 6082 alloy ................ 88
Figure 3.16: The XRD spectrum for the 6060 alloy homogenized at 560oC for 0.5h .................. 89
Figure 3.17: a) Simulation results and experimental for Fe element, b)profile mapping in
as – cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time ........................... 90
Figure 3.18: a) Simulation results and experimental for Mg element, b)profile mapping in
Figure 3.19: a) Simulation results and experimental for Si element, b)profile mapping in
as–cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time ............................. 92
Figure 3.20: a) Simulation results and experimental for Mn element, b)profile mapping in

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 4.1: SEM image used for the measurement of indices: (a) Low magnification image,
(b) high magnification isolation of the group of particles, (c) image J display
used for the measurement of the indices ........................................................................... 101
Figure 4.2: Metallographic images, (a) as-cast, (b) homogenized at 540oC, (c) 560oC and
(d) 580oC ......................................................................................................................................... 104
Figure 4.3: Images indicating the rounding of the edges of the particles during
homogenization ............................................................................................................................ 105
Figure 4.4: Imaged revealing the pinching process during homogenization .............................. 105
Figure 4.5: Images revealing the spherical shape at the third stage .............................................. 106
Figure 4.6: The values of indices for the as-cast and homogenized alloys: (a) aspect ratio,
(b) feret, (c) circularity and (d) circularity of edges ...................................................... 107
Figure 5.1: Measured cooling profiles used in simulations and in the experimental
validation, (a) natural cooling, (b) forced cooling ........................................................... 112
Figure 5.2: Simulation results for the 6063 alloy: (a) evolution of Mg2Si volume fraction
during cooling for three homogenization temperatures 540, 560 and 580oC
and (b) evolution of particle size during cooling for natural and forced cooling
from 580oC homogenization temperature ......................................................................... 113
Figure 5.3: Particle size distribution (PSD) predicted by the KWN model for the 6063 alloy
for the two cooling profiles, natural and forced cooling for the 580oC
homogenization temperature ................................................................................................. 114
Figure 5.4: Evolution of hardness during cooling of the 6063 alloy for natural and forced
cooling and for three homogenization temperatures 540, 560 and 580oC. .......... 114
Figure 5.5: Homogenization cooling profiles ........................................................................................... 117
Figure 5.6: Brinnel hardness of alloy 6082 immediately after cooling from the
Figure 5.9 Brinnel hardness immediately after cooling from 540oC homogenization
temperature ................................................................................................................................... 121
Figure 5.10: Brinnel hardness immediately after cooling from 560oC homogenization
temperature ................................................................................................................................... 121
temperature ................................................................................................................................... 121
Figure 5.12: Brinnel hardness after homogenization and natural aging for the 6082 alloy . 123
Figure 5.15: Hardening potential Hp for the 6082 alloy at (a) 540, (b) 560 and (c) 580oC
homogenization temperatures ............................................................................................... 124
homogenization temperatures ............................................................................................... 125
homogenization temperatures. .............................................................................................. 126
Figure 5.18: DSC spectra for 6005 alloy (0.57Mg-0.68Si) for cooling rates from 0.1 to 375
K/min [18] ...................................................................................................................................... 129

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.19: CCP diagram for 6060, 6063, 6005 and 6082 alloys [18] ......................................... 130
Figure 5.20: Arrhenius fitting of the data of Table 2 a) Fitting the β-nose, b) fitting β’ -
nose ................................................................................................................................................... 131
Figure 5.21: Effect of excess Si on the precipitation of β’-Μg2Si ...................................................... 133
Figure 5.22: CCP diagram of the 6060_3 alloy (red lines). The dotted β΄ line is shifted by
taking into account only the Mg effect. The β΄ nose (full line) is shifted by
taking into account the effect of excess Si in the 6060_3 alloy................................... 134

26/03/2018 03:27:43 EEST - 137.108.70.7
List of Tables
Table 1.1: Methodology of the present Thesis ........................................................................................... 15
Table 2.1: Chemical composition of alloys 6082, 6063, 6005, 6060 (in mass %) ........................ 31
Table 2.2: Metallographic procedure for 6060, 6082 and 6063 alloys ............................................ 32
Table 2.3: Segregation index for the 6060 alloy variants ...................................................................... 42
Table 2.4: Chemical composition of α-AlFeSi phase (mass %) ............................................................ 45
Table 2.5: Comparison between calculated and measured (EDS) matrix composition
adjacent to boundary intermetallics ....................................................................................... 48
Table 2.6: Comparison between calculated and measured phase fractions in the as-cast
6082 alloy .......................................................................................................................................... 53
Table 3.1: Eutectic melting temperatures in Al-Mg-Si alloys [10]...................................................... 64
Table 4.1 Homogenization heat treatments ................................................................................................ 99
Table 4.2: Number of images analyzed and number of particles measured per billet for the
quantification of the homogenization state .......................................................................... 99
Table 4.3: The indices employed for the quantification of the homogenization state ............ 100
Table 4.4: Respective measurements for each particle referring to Figure 4.1 ......................... 102
Table 5.1: Microhardness (HV0.3) as a function of cooling profile ................................................... 115
Table 5.2 Position of β and β noses from the CCP diagrams of Fig.2 .............................................. 131
Table 5.3: Calculation of excess Si of the alloys (compositions in wt%) ...................................... 132

26/03/2018 03:27:43 EEST - 137.108.70.7
Introduction
1.1. Problem statement and aim of the thesis
Extrudable aluminum alloys of the 6xxx series are heat-treatable alloys based on the
Al-Mg-Si system. These alloys find significant applications in the building, construction and
transportation sectors due to the excellent strength/ductility properties combined with
reduced weight and oxidation resistance. Large scale ingots or billets, suitable for hot
extrusion, are produced with the direct-chill casting method. The as-cast material exhibits
low formability due to microstructural inhomogeneities, arising from the solidification
process and the alloy chemistry. Major inhomogeneities in the as-cast material include:
elemental microsegregation in the level of the secondary dendrite arms, grain boundary
segregation and the formation of several low-melting eutectics and intermetallic
compounds such as the iron-containing α-AlFeSi and β-AlFeSi, which degrade the
extrudablity of the as-cast billet.
These effects can be partially or completely eliminated with the application of a
homogenization treatment of the as-cast billets. The benefits of this treatment include the
following: removal of elemental microsegregation, dissolution of low melting eutectics,
transformation of the iron-containing intermetallic compounds, shape control (round-off)
of hard particles with sharp edges, dissolution of the grain-boundary Mg2Si phase and re-
precipitation (during cooling) with a more homogeneous in-grain distribution. All these
effects improve the extrudability and increase the response of the material to natural or
artificial aging.
New demanding light-weight applications in the automotive and aircraft sectors
require the development of high-strength alloy extrusions. Such applications include pillar
and chassis structural parts in vehicles and wing or fuselage reinforcement parts in
aircrafts. The increase of strength is possible with higher alloying with Mg and Si, in order
to form higher amounts of the strengthening phase Mg2Si (e.g. 6061 and 6082 alloys).
However increased alloying deteriorates the extrudability, leading to extremely low
extrusion speeds for the high-strength alloy systems. In the other end of the spectrum, i.e.
low-alloy low-strength applications (e.g. 6060 or 6063 alloys), the main industrial

26/03/2018 03:27:43 EEST - 137.108.70.7
requirement is to increase the extrusion speed, in order to increase production rates. In
both cases, it appears that extrudability is the key property that should be controlled and
improved through a carefully designed homogenization process.
So far the problem has been tackled in a rather empirical way, involving intense trial-
and-error laboratory or even industrial experiments to determine the effect of alloy
chemistry and homogenization process parameters such as the homogenization
temperature and time and the cooling rate following homogenization. In most cases only
individual aspects of the process are considered and the effect of prior processing is often
neglected.
The process chain of extrudable Al-alloys is depicted in Figure 1.1. . The design of new
high-strength 6xxx alloys, exhibiting adequate extrudability, as well as the design of low-
strength alloys with high extrudability, calls for an integrated approach considering all parts
of the process chain and their effect on extrudability.
Figure 2.1: The process chain of extrudable Al-alloys
By this it is meant that not only processing steps prior to extrusion are important but
steps following the extrusion process should be considered as well, since they set important
microstructural requirements to be achieved in the preceding process steps. For example,
the homogenization process is carried out in order to eliminate the microsegregation
10

26/03/2018 03:27:43 EEST - 137.108.70.7
established in the previous process steps, however it should also aim to prepare the
material for best response to the aging treatment, which follows the extrusion and sets the
final properties of the extruded profiles.
An even more ambitious approach is to combine the thermodynamic calculations with
kinetic models, e.g. diffusion calculations and thereby predict the rate of phase
transformations, which shape the microstructure across the process chain. This approach is
extremely powerful and may be used to simulate a wide range of different phenomena such
as solidification and microsegregation during casting, dissolution of Mg2Si, removal of
microsegregation and transformation of iron intermetallics during homogenization and
finally re-precipitation during cooling from the homogenization temperature.
The aim of this doctoral thesis is to apply a simulation-based approach, validated by
relevant experimental data, in order to describe the effect of alloy chemistry and
homogenization process conditions on the material microstructure prior to extrusion. In
this way it will be possible to shape design rules (in terms of composition and processing)
for the development of either high-strength alloys with considerable extrudability or low-
strength alloys with high extrudability.
1.2. State-of-the-art
As it will be shown by the literature review in the next sections, the problem of the
design of alloy compositions and homogenization treatments for enhanced extrudability
has been considered by experimental methods either in the laboratory or by industrial
trials. In most cases, parts of the process chain have been considered individually. Most
works consider the effect of composition or process parameters such as the
homogenization time and temperature or the effect of the cooling rate individually on the
outcome of the homogenization process. In this way, ‘best-practice’ homogenization recipes
have been established for certain alloys. For the design of a new alloy composition or the
design of a new homogenization treatment for enhanced extrudability, the traditional trial-
and-error methodology is still followed.
11

26/03/2018 03:27:43 EEST - 137.108.70.7
In terms of process simulation, the work performed so far is rather limited to
modeling specific aspects of the process and to the author’s knowledge there is no
integrated simulation approach that eventually leads to the development of design rules for
alloy composition and homogenization treatments of extrudable aluminum alloys.
A complete literature review is presented in the following sections of the thesis.
1.3. Methodology
The approach followed in this thesis has both computational and experimental
aspects. A flow chart of the methodology is depicted in Table 1.1.
 Computational
Computational alloy thermodynamics, based on the CALPHAD approach, has been
applied for the calculation of equilibrium phase constitution as a function of alloy
composition and temperature. Isopleth phase diagrams in multicomponent alloys have
been calculated permitting the determination of the stability limits of the various phases in
the system. The relevant thermochemical database COST-507 for light alloys has been
modified to incorporate Manganese (Mn) in the α-AlFeSi iron intermetallic. The relevant
software used was Thermo-Calc. The program was also applied to carry out Scheil-Gulliver
solidification simulations and calculate the resulting microsegregation of elements and
phases in the as-cast microstructure. Segregation profiles for elements and phases across
the grain were determined as a function of alloy composition. The results were confirmed
experimentally.
Computational kinetics, based on the solution of the diffusion equations in
multicomponent systems (Onsager principle) was applied in order to calculate the rate of
diffusional phase transformations during homogenization. In this way the dissolution of the
Mg2Si phase, the transformation of iron intermetallics (β-AlFeSi to α-AlFeSi) as well as the
reprecipitation during cooling was simulated. This included the development of a dual-
grain model (DGM) to take into account the different grain sizes between neighboring
grains. The relevant software was DICTRA, which incorporates the so-called sharp interface
model for treating diffusion-controlled transformations. The interdiffusion coefficients
12

26/03/2018 03:27:43 EEST - 137.108.70.7
were calculated by the program with data from the mobility database MOB2, which
contains all the information on the temperature dependence of mobility coefficients. The
relevant thermodynamic factors, taking into account the compositional dependence, were
calculated by a link to Thermo-Calc. Initial composition profiles of elements and phases
were the direct results of the Scheil simulations of the previous step. In this way it was
possible to connect the solidification step with the homogenization step.
The Kampmann-Wagner numerical model (KWN), with which precipitation in
isothermal conditions can be calculated, was applied under continuous cooling conditions
(by a suitable matlab code) in order to evaluate the effect of excess-Si on the precipitation
of metastable β’-Mg2Si during cooling from the homogenization temperature. This led to the
modification of existing CCP diagrams and permits the design of cooling programs to
achieve precipitation of the desirable β’ instead of the equilibrium phase β-Mg2Si. In
addition the KWN model was combined with a suitable strength model, involving
precipitation strengthening, in order to calculate the strength (hardness) of the alloy
following homogenization cooling.
 Experimental
Standard metallographic techniques including optical microscopy and scanning

electron microscopy (SEM) were applied to reveal the microstructure of the as-cast and
homogenized materials. Phase fractions were measured by applying quantitative image
analysis.
Segregation profiles and the evolution of elemental profiles during homogenization
was performed by SEM-EDX measurements
Evolution of phase transformation during homogenization was performed with XRD
Characterization of the homogenization state was performed with the measurement of
microstructural indices (aspect ratio, feret, circularity, circularity of edges) in a large
number of SEM images.
13

26/03/2018 03:27:43 EEST - 137.108.70.7
Laboratory homogenization treatments were performed in selected alloys in order to
study the progress of homogenization. The hardness after cooling from the homogenization
temperature was also determined.
14

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 2.1: Methodology of the present Thesis
Mg-Si map of phase fractions in

Scheil – Gulliver as-cast structure
Microsegregation
Casting (CALPHAD – based)
Elements/phases Selection of compositions with
Ch.2
high Mg2Si and low β-AlFeSi
Dual Grain Model (DGM)
Multicomponent Diffusion in Mg2Si dissolution
1st process map based on time
multiphase, variable grain β→α – AlFeSi
for β→α - AlFeSi transformation
system transformation
Homogenization Ch.3
Quantitative metallography Index – based
Selection of homogenization
Measurements of indices: quantification of the
temperature/time to achieve
aspect ratio and circularity homogenization
specific index values
Ch.4 state
Evolution of PSD of
Kampmann – Wagner
Mg2Si during
Numerical Model (KWN)
cooling
Homogenization CCP diagram for the design of
Hardness of
cooling homogenization cooling
Strength Model homogenized billets
Ch.5 Effects on CCP
diagrams
15

26/03/2018 03:27:43 EEST - 137.108.70.7
1.4. Innovative aspects of the thesis
The thesis provides results leading to progress beyond the state-of-the-art in the area
of the computational-based design of the homogenization process of extrudable aluminum
alloys. Such innovative aspects are:
 The application of a computational thermodynamics and kinetics approach, for the
integrated simulation of the process chain including, solidification-induced
microsegregation and subsequent homogenization in multicomponent and
multiphase aluminum alloys.
 The development of a homogenization dual-grain model (DGM), for the evaluation of
the effect of different grain size between neighboring grains.
 The description of temporal and spatial correspondence, of the β-AlFeSi to α-AlFeSi
transformation during homogenization.
 The evaluation of the effect of excess-Si on the precipitation of metastable β’-Mg2Si
during cooling from the homogenization temperature by the application of the KWN
precipitation model.
 The application of microstructural indices for the characterization of
homogenization state of extrudable aluminum alloys by quantitative metallography.
 The development of design rules for the as-cast microstructure in terms of mapping
phase fractions of desirable and undesirable phases with respect to alloy
composition.
 The development of design rules for selection of alloy compositions and processing
conditions for homogenization.
16

26/03/2018 03:27:43 EEST - 137.108.70.7
1.5. Structure of the thesis
The structure of the thesis follows the process chain, of the extrudable Al-alloys.
Chapter 2 deals with the evolution of the as-cast microstructure and more specifically
with the microsegregation of elements and phases as a result of solidification. The mapping
of phase fraction in the Mg-Si composition space is also presented in this chapter.
The simulation of homogenization with the DGM is described in Chapter 3. The DGM
deals with the temporal and spatial evolution of the β to α – AlFeSi transformation. A
process map based on the time for the completion of the β to α-AlFeSi transformation is
presented.
The index–based quantification of the homogenization state is presented in Chapter 4.
Precipitation of the Mg2Si during homogenization cooling whith the KWN model is
presented in Chapter 5. The development of design rules for high extrudability is also
presented in Chapter 5.
The conclusions of this thesis and proposal for further research are presented in Chapter
6 and 7 respectively.
17

26/03/2018 03:27:43 EEST - 137.108.70.7
18

26/03/2018 03:27:43 EEST - 137.108.70.7
2. Microsegregation and Phase Fraction
Mapping
2.1. Introduction and literature review
Extrudable aluminum alloys of the 6xxx series, based on the Al-Mg-Si system, are
subjected to a rather complicated process chain involving melt treatment, casting,
homogenization, preheating, extrusion, and ageing. The process chain begins with the direct
casting (DC casting) of large cylindrical billets. The as-cast microstructure, which depends
on composition and casting conditions, influences in a large degree the properties of
extrudable 6xxx Al–alloys. The as-cast material exhibits low formability due to
microstructural inhomogeneities, arising from the solidification process and the alloy
chemistry. Major inhomogeneities in the as-cast material, which degrade the extrudability
of the as cast billet include:
 elemental microsegregation in the level of the secondary dendrite arms
 grain boundary segregation and formation of low-melting eutectics
 formation of intermetallic compounds
The research work described in this section focuses on the composition and
microstructural evolution of the as–cast billets since these key factors influence not only the
extrudability but the product properties of Al-alloys after extrusion has been completed.
The 6xxx series alloys, which are based on the ternary Al-Mg-Si alloy system, contain
magnesium (Mg) and silicon (Si) as major alloying elements for the formation of Mg2Si,
which is the major strengthening phase, providing precipitation hardening after aging.
Besides the intentional additions Mg and Si, transition metals and impurities such as
Fe, Mn and Cr are present. Even a small amount of these impurities causes the formation of
new intermetallic compounds (IMCs). The composition of the alloy as well as casting
conditions will influence the exact type and volume fraction of intermetallic phases that will
form during casting [1].
Iron (Fe), is the main impurity in these alloys forming intermetallic compounds in
combination with Si such as β-Al5FeSi and α–Al12Fe3Si. In addition, Mn can substitute for
Fe, and promote the formation of α-Al12(FeMn)3Si. These phases, from now on called β-
19

26/03/2018 03:27:43 EEST - 137.108.70.7
AlFeSi and α-AlFeSi respectively, are mostly located at the grain boundaries in the as-cast
microstructure. The β-AlFeSi phase has a monoclinic crystal structure and plate–like
morphology, while α–AlFeSi has a cubic crystal structure and a globular morphology. The
plate–like morphology of the β-AlFeSi phase limits the exrtudability of the alloy since it
induces local cracking and surface defects in the extruded material, while the presence of
the more rounded α–AlFeSi phase has a lesser impact on the extrudability of the alloy [2].
The disadvantage of the high Fe content is the enhanced probability of forming large
particles of β-AlFeSi, with a detrimental impact on the extrudability of the alloy [3]. On the
other hand, Si has an important role together with magnesium (Mg) in forming Mg2Si, or β-
phase, the transition phase of which, β΄΄, is the main hardening precipitate in the alloy [4].
Cr may assist in the formation of the AlFeCrSi phase, which has a very important role in the
grain size control [5]. Furthermore microsegregation takes place during solidification in the
scale of the secondary dendrite arm spacing (SDAS) due to the difference of the
thermodynamic equilibrium solubility of alloying elements between the solid and liquid
phases. The influence of alloy composition on the formation of intermetallic compounds has
been generally studied experimentally [2-4]. Considerably less work has been performed
on the formation of intermetallic phases during casting of Al-billets. The main objective of
these works has been the study of the morphology and dispersion of IMC’s in the as-cast
microstructure. The effect of cooling rate on as-cast microstructure formation [6-8] as well
as the formation of Mg2Si and other intermetallic phases in Al-Mg-Si alloys [9, 10] have been
studied experimentally. Finally a thermodynamic description of the Al-Fe-Mg-Mn-Si system
and investigations on the microsegregation during directional solidification of alloys based
on that system has been presented [11].
Despite the above works, a systematic analysis of the effect of alloying elements on
microsegregation and IMC formation in Al-Mg-Si-Fe-Mn alloys does not exist. Such a study
would assist the selection of alloy compositions for the minimization of the undesirable β-
AlFeSi phase in the as-cast microstructure. With such an optimized as-cast microstructure it
would be easier to reach the homogenization target and increase the extrudability without
sacrificing strength in these alloys.
20

26/03/2018 03:27:43 EEST - 137.108.70.7
2.2. Computational methodology
The present work is an effort to present in detail the effect of the major alloying
elements Mg and Si on microsegregation and the formation of Mg2Si and Fe intermetallics
during solidification of extrudable 6xxx Al-alloys. Mn is added in these alloys in order to
speed up the β→α transformation during homogenization. However the effect of Mn on IMC
formation in the as-cast microstructure has not been assessed. Therefore in this work the
effect of Mn is also quantified. In the first part, a computational thermodynamics-based
analysis of equilibrium phase constitution is presented, in order to identify the major
phases and their stability limits in the alloys. Solidification of the Al-billets is then modelled
by applying Scheil simulations in order to compute the mole fractions of all phases as well
as their segregation in the as-cast structure. Computed fractions are validated against
experimented measurements. The mole fraction of Mg2Si and the β-AlFeSi phase is then
mapped over the useful (0-1.2 mass %) Mg-Si composition space for 6xxx extrudable alloys.
2.2.1. Computational Alloy Thermodynamics
All necessary thermodynamics calculations for the description of the Gibbs free
energy of the various phases in the system were performed with the Thermo-Calc software
[12], which is based on the CALPHAD approach [13]. This is a method that overcomes the
problem of the multi-dimensionality posed by a system with many components. A
thermodynamic description of a system requires assignment of thermodynamic functions
for each phase. The CALPHAD method employs a variety of models to describe the
temperature, pressure and concentration dependence of the free energy functions of the
various phases. The description refers to pure metals, stoichiometric and non-
stoichiometric phases and intermetallic compounds as well as substitutional and interstitial
solid solutions. It is based on the minimization of the total free energy of the system,
comprising of all phases, which take part in equilibrium of a multicomponent system.
The total free energy is the sum of the free energies of individual phases
21

26/03/2018 03:27:43 EEST - 137.108.70.7
p
G   niGi (1)
i 1
where,
ni is the number of moles of phase i
G is the Gibbs free energy of phase i
For multicomponent systems the Gibbs free energy of a phase is the sum of four
contributions:
Gi  G o  Gid  G xs  G mag (2)
where
G o is the contribution of the pure components

G id corresponds to the ideal or entropic term
G xs is the excess term corresponding to the deviation from ideality and
G mag is the magnetic contribution to the Gibbs free energy due to magnetic ordering.
The major phases and stability limits for the case of Al-Mg-Si-Fe system are described
in the COST507 light alloys database [14]. This is a public domain thermodynamic database
created in the course of the European concerted action program “Development of a
thermochemical and thermos-physical database for light alloys”. The Laboratory of
Materials at the University of Thessaly was a member of this consortium and contributed to
the development of the database. The description of the relevant phases of interest in the
COST507 database is given in the following sections.
 FCC and Liquid
The free energy of the FCC substitutional solid solution phase (α-Al) as well as the
liquid phase are expressed according to equation (2) as:
22

26/03/2018 03:27:43 EEST - 137.108.70.7
G   X i G oi  RT  X i ln X i   X i X j  L(ijp ) ( X i  X j ) p   XX i j X k  L(ijkp )
i i i i j p i i  j k p
(3)
where,
R is the universal gas constant
T is the temperature (in Kelvin)
X i are the mole fractions of the elements (i =Al, Mg, Si, Fe, Mn)
G oi are the free energies of the elements
The free energy contribution of the pure elements is
Go, LIQ  X Al GAlo, LIQ  X Mg GMg

o , LIQ
 X SiGSio, LIQ  X FeGFe
o , LIQ
 X MnGMn
o , LIQ
(4)
Where, for example, GMg

o , LIQ
is the free energy of Mg, which for 298.15<Τ<923Κ is
o , LIQ
GMg  H Mg (298.15K )  8202.24  8.83T  8.01x1020 T 7  GMg
SER
(T )
(5)
and for 923<Τ<6000Κ the free energy term is
o , LIQ
GMg  H Mg (298.15K )  8690.31  9.39T  1.03x1028 T 9  GMg
SER
(T )
(6)
SER
GMg (T )
is the free energy with respect to the reference state (Stable Element Reference,
SER) defined as the enthalpy of the pure metal at 298.15K.
For the Liquid phase, the ideal term Gid can be described as
Gid ,LIQ  RT ( X Al ln X Al  X Mg ln X Mg  X Si ln X Si  X Fe ln X Fe  X Mn ln X Mn )
(7)
As can be seen from equation (7) the ideal term corresponds to the entropic contribution to
free energy.
23

26/03/2018 03:27:43 EEST - 137.108.70.7
The third and fourth terms on the right hand side of equation (3) represent the excess
term in the form of a Redlich-Kister-Muggianu (RKM) expansion, which takes into account
the composition dependence of the interaction between elements, a procedure followed in
L(ijp ) L(ijkp )
the CALPHAD approach. The interaction parameters and are the binary and
ternary interaction parameters of order p and are temperature dependent. With the use of
the interaction parameters, the excess term can be described as follows:
XS ,LIQ (0) (1) (2)

G X X [L L (X  X )L (X  X )2 ] 
Al Mg Al , Mg Al , Mg Al Mg Al , Mg Al Mg
(0) (1) (2)
X X [L L (X  X )  L ( X  X )2 ] 
Al Si Al , Si Al , Si Al Si Al , Si Al Si
(0) (1) (2)
X X [L L (X  X )  L ( X  X )2 ] 
Al Fe Al , Fe Al , Fe Al Fe Al , Fe Al Fe
(0) (1) (2)
X X [L L (X  X )  L ( X  X )2 ] 
Al Mn Al , Mn Al , Mn Al Mn Al , Mn Al Mn
(0) (1) (0) (1)
X X [L L (X  X )]  X X [ L L ( X  X )] 
Fe Mg Fe, Mg Fe, Mg Fe Mg Fe Mn Fe, Mn Fe, Mn Fe Mn
(0) (1) (2) (3) (0)

X X [L L (X  X )  L ( X  X )2  L ( X  X )3 ]  X X L
Fe Si Fe, Si Fe, Si Fe Si Fe, Si Fe Si Fe, Si Fe Si Mg Mn Mg , Mn
(0) (1) (2) (3) (4)

X X [L L (X  X )L (X  X )2  L (X  X )3  L (X  X )4 ] 
Mg Si Mg , Si Mg , Si Mg Si Mg , Si Mg Si Mg , Si Mg Si Mg , Si Mg Si
(0) (1) (2) (3)

X X [L L (X  X )L (X  X )2  L (X  X )3 ] 
Mn Si Mn, Si Mn, Si Mn Si Mn, Si Mn Si Mn, Si Mn Si
(0) (1) (2) (0)
X X X [L L L ] X X X L
Al Mg Si Al , Mg , Si Al , Mg , Si Al , Mg , Si Al Mn Si Al , Mn, Si (8)
All values of the relevant interaction parameters are stored in the COST507 database.
As mentioned above, an additional term expressing the magnetic contribution to the
Gibbs free energy has to be added to equation (1). This term is expressed according to
Inden [15] as
G mag  RT ln(  1) f ( ) (9)
where,
24

26/03/2018 03:27:43 EEST - 137.108.70.7
 is the magnetic moment in Bohr magnetons. It is composition-dependent and is given by
   X i io   X i X j [ 
( p)
( X i  X j )( p ) ] (10)
 i i j p
where
 is a normalized temperature,
with
  T /Tc and Tc being the Curie temperature
which is composition-dependent and is given by
Tc   X iTcio   X i X j [Tc
( p)
( X i  X j )( p ) ] (11)
 i i j p
Tc
The values of  and are stored for the FCC phase in the COST 507 database.
 β-AlFeSi phase
The β-AlFeSi phase is treated as a stoichiometric Al14Fe3Si3 intermetallic compound.

The Gibbs free energy is then
GAl14 Fe3Si3 (T ) 14H Al (298.15)  3H Fe (298.15)  3H Si (298.15)  GAlf 14 Fe3Si3 (T )  14GAlSER (T )  3GFeSER (T )  3GSiSER (T )
(12)
As mentioned above, the left hand side in (12) is the free energy relative to the
reference state, called stable element reference (SER), with the enthalpies of the elements
GAlf 14 Fe3Si3
evaluated at room temperature (298.15K). The term is the temperature-
dependent energy of formation of the intermetallic compound and is stored together with
the free energies of the pure elements

GiSER (T ) (i=Al, Fe, Si) in the COST 507 database.
 α-AlFeSi phase
25

26/03/2018 03:27:43 EEST - 137.108.70.7
The α-AlFeSi phase is treated as a non-stoichiometric intermetallic compound. The
free energy description is based on the sublattice model, developed by Sundman and Agren
[16]. Four sub lattices are used for the phase description as (Al):(Fe):(Si):(Al,Si) with sites
0.6612:0.19:0.0496:0.0992 respectively. The first three sub lattices are exclusively occupied
by a single element while in the fourth sublattice there is partial occupation by Al and Si.
The site occupation in the fourth sublattice is expressed by the site fraction
yi(4) where
i=Al,Si. Considering exclusive occupation by Al or Si in the fourth sublattice, two terminal

stoichiometric compounds are defined, Al0.7604Fe0.19Si0.496 and Al0.6612Fe0.19
Si0.1488 called G1 and G2 respectively from now on for clarity. Then the free energy of the
α-AlFeSi phase can be described in terms of the free energies of the terminal compounds as
Ga AlFeSi  y Al
(4)
G1  ySi(4)G2  0.0992RT ( y (4)
Al ln y Al  ySi ln ySi )
(4) (4) (4)
(13)
where the first two terms correspond to the free energies of the terminal stoichiometric
compounds and the third is the entropic term.
The free energies of the terminal compounds are
G1 (T )  0.7604 H Al (298.15)  0.19 H Fe (298.15)  0.0496 H Si (298.15) 

 G1f  0.7604GAlSER (T )  0.19GFe
SER
(T )  0.0496GSiSER (T )
(14)
and
G2 (T )  0.6612 H Al (298.15)  0.19 H Fe (298.15)  0.1488H Si (298.15) 

 G2f  0.6612GAlSER (T )  0.19GFe
SER
(T )  0.1488GSiSER (T )
(15)
The free energies of formation of

G1f and G2f of the terminal compounds are stored in
the COST 507 database.
26

26/03/2018 03:27:43 EEST - 137.108.70.7
 Mg2Si strengthening phase
The Mg2Si phase is also treated as a stoichiometric phase. The Gibbs free energy is
then
GMg2Si (T )  2H Mg  298.15   H Si  298.15   GMg

f
(T )  2GMg
SER
(T )  GSiSER (T )
2 Si (16)
2.2.2. Amendments in COST507 Al database
In the present study the thermodynamic database for light non-ferrous alloys COST
507 was used [14]. COST-507 database contains thermodynamic data for the 19 elements
most commonly encountered in Al alloys (Al, B, C, Ce, Cr, Cu, Fe, Li, Mg, Mn, N, Nd, Ni, Si, Sn,
V, Y, Zn, Zr). According to the thermodynamic description in the COST507 database Mn does
not enter in the description of the, α-AlFeSi. Therefore this database was amended by
recent thermodynamic descriptions, where a higher number of ternary and quaternary
systems have been assessed [17]. In the present work the description of the α-AlFeSi phase
was amended in order to introduce Mn in the iron sublattice. This was accomplished by the
introduction of the relevant thermodynamic quantities in the GES module of Thermo-Calc,
where the composition and temperature dependence of the various thermodynamic
parameters are stored.
2.2.3. Simulation of microsegregation
Microsegregation during the complex solidification process of Al-Mg-Si alloys is

usually simulated by employing the Scheil-Gulliver model [18, 19] with the main
assumption that the diffusion coefficients of the components in the liquid phase are
infinitively fast whereas in the solid are zero. The assumption of negligible diffusion in the
solid is acceptable due to the short local solidification time encountered in industrial direct-
chill casting conditions. Along each step in the cooling process during solidification, local
equilibrium is established at the solid/liquid interface and the composition of the solid and
27

26/03/2018 03:27:43 EEST - 137.108.70.7
liquid phase is given by the phase diagram. Also along each step the solid phase formed
retains its composition and the liquid composition is always homogeneous. The basic
derivation of the Scheil-Gulliver equation is presented below.
As mentioned above, the cooling during solidification is so fast that the solid
composition cannot be homogenized by diffusion. The opposite is true for the liquid, where
the composition is always homogeneous during the course of solidification. Consider Figure
2.1, which depicts three stages during solidification of a binary alloy. At the beginning,
solidification starts at the liquidus temperature TL and the first solid forms with a
composition kCo, where k is the partition coefficient (k=Cs/CL). At an intermediate
temperature, the compositions of the solid and liquid phases at the interface are provided
by the phase diagram as Cs and CL respectively. While the liquid composition is
homogenized to the value CL, the solid composition varies from the initial kCo to the value
Cs at the interface.
Figure 2.1: The three stages during solidification of a binary alloy
Due to the limited diffusion in the solid the average composition is lower than the
composition it would have, if solidification proceeded under global thermodynamic
equilibrium. It is possible to calculate the composition of the solid at the interface Cs as a
function of the fraction solid fs that has formed, by considering a mass balance in Figure 2.2.
28

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.2: The composition of the solid at the interface Cs as a function of the fraction solid
fs that has formed, by considering a mass balance
When a small quantity of solid dfs forms, solute is rejected into the liquid and the
liquid composition increases by dCL. The mass balance gives
(CL  CS )df S  (1  f S )dCL (17)
Rearranging terms and integrating we get
fs C L
df L dC
0 (1  f S )  C CL (1 L k )
o (18)
which leads to the Scheil-Gulliver equation

Cs  kCo (1  f s )k 1 (19)
As mentioned above, equation (19) predicts a lower composition in the solid than the
equilibrium prediction, at the same time it predicts a higher fraction of remaining liquid to
be solidified. Towards the end of solidification a rapid increase in the solute concentration
of the liquid as well as an increase in the concentration of the solid to the value Csa results in
the formation of a non-equilibrium eutectic [20]. Therefore the Scheil-Gulliver equation
29

26/03/2018 03:27:43 EEST - 137.108.70.7
predicts a certain amount of eutectic even when the alloy solidifies as a single-phase alloy
under thermodynamic equilibrium. This results in an especially complex as-cast
microstructure in the 6xxx series alloys, because of the different sets of phases that can
form in different areas of the billet due to differences in the cooling rate [21].
The Scheil-Gulliver equation (19) can be generalized to higher order multicomponent
systems. The problem is treated with a special module (called the SCHEIL module) in the
Thermo-Calc software. A simulation is conducted step by step along the cooling process, by
decreasing the temperature (normally by 1oC in each step) below liquidus, and after each
step the new liquid composition is used as the local overall composition at the next step.
The simulation can handle cases of concurrent precipitation of two or more solid phases,
such as eutectic reactions. It should be mentioned that other computational methods, which
take into account macroscopic transport phenomena (melt flow and transport of crystals)
have been developed for ternary alloys [22-23]. However these methods use locally
linearized phase diagrams and do not benefit, at the moment, from the available
thermodynamic databases.
With the SCHEIL module it is possible to calculate:
 the solidification path, i.e. the sequence of formation of solid phases during
solidification
 the composition of solid as a fraction of solid formed during solidification,
corresponding to the microsegregation of elements
 the fraction of solid phases vs fraction solid formed during solidification,
corresponding to the microsegregation of phases
 the exact liquidus and solidus temperatures of the alloy
The development of microsegregation has been studied in four different variants of
6060 as well as in 6082, 6063 and 6005 alloys. Emphasis has been given on the effect of Mg
and Si. These elements influence the strength properties and extrudability, which are
important for the final performance. The chemical compositions of the alloys are listed in
Table 2.1
30

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 2.1: Chemical composition of alloys 6082, 6063, 6005, 6060 (in mass %)
Alloy Al Mg Si Fe Mn
6082 Balance 0.63 0.9 0.2 0.45
6063 Balance 0.51 0.55 0.2 0.03
6005 Balance 0.48 0.68 0.15 0.2
6060_1 Balance 0.38 0.4 0.2 0.03
6060_2 Balance 0.48 0.4 0.2 0.03
6060_3 Balance 0.38 0.47 0.2 0.03
6060_4 Balance 0.48 0.47 0.2 0.03
2.3. Experimental procedures for model validation
The experimental procedures employed for the validation of simulation predictions

are discussed in this section. Due to the difficulties encountered when measuring low
volume fractions, the work was performed mainly on alloy 6082, which because of the
higher alloy content, possessed a higher volume fraction of intermetallic phases. However
metallographic results are presented for the 6060 and 6063 alloys as well.
The materials were received in the form of as-cast billets prepared with DC casting by
the Aluminum of Greece (AoG). The diameter of the billet was 200mm and the solidification
rate was 2mm s-1. The received billets were sectioned in transverse slices as shown in
Figure 2.3. Then the samples of 10mm thickness were cut from approximately the middle of
the billet’s slice radius in order to avoid edge inhomogenities and other casting defects. The
surface in a direction transverse to the billet axis was prepared for metallography.
Figure 2.3: Transverse cross sections along the billet diameter
31

26/03/2018 03:27:43 EEST - 137.108.70.7
Specimen preparation included cutting and grinding with SiC papers rating 800, 1000,
2400 and 4000 grit for alloy 6063 and grinding on 500 grit SiC water – cooled paper, polish
with 3μm and 1μm diamond paste and finally polished with 0.05μm Colloidal Silica for
alloys 6060 and 6082. Polishing was performed with 3 and 1μm diamond paste followed by
electro-polishing with Barkers reagent consisting of 10 millilitre fluoroboric acid (35 %)
and 200 millilitre water. The microstructure was revealed after etching in Keller’s solution,
consisting of 0.5% HF in 50ml H2O in order to reveal and identify the intermetallic phases in
the as-cast microstructure. Table 2.2 summarizes the procedure followed for
metallographic preparation.
Table 2.2: Metallographic procedure for 6060, 6082 and 6063 alloys
Stages Alloy
6060 6082 6063
Grinding
SiC Papers (grit)
320 x x x
500 x x
1000 x x x
2400 x x x
Polishing
Diamond Paste
3μm x x x
1μm x x x
Coloidal Silica x x x
Electro Polishing
stainless steel for
cathode;specimen is anode.
Anodize 40-80 s at approximately
x x
0.2 A/cm2 (1.3 A/in.2, or about 20
V dc). Check results on microscope
with crossed polarizers
Chemical etching
Keller's x
Modified Poultons x
The volume fractions of the Mg2Si and α+β AlFeSi phases have been measured,
through quantitative methods based on point counting, using 1000 grit on 10 sections
32

26/03/2018 03:27:43 EEST - 137.108.70.7
according to ASTM E-562 standard [24]. Processing has been performed using Photoshop
and Image J image analysis software. The results were then compared with the
corresponding results of the Scheil simulation of the solidification process of 6082 alloy.
The following methodology was applied: first the phases were identified using SEM
combined with EDS analysis and then a correspondence between SEM and optical
microscopy was established in order to identify the phases in optical micrographs. This
methodology enabled the use of a point counting method in order to determine the volume
fractions of phases, using a high number of metallographic sections at the lowest
magnification possible in order to get the most representative microstructure. Solute
microsegregation has been also experimentally determined by SEM/EDX analysis.
2.4. Results and Discussion

2.4.1. Equilibrium calculations
Equilibrium calculations are presented in this section, which aim at identifying the
major phases and their stability limits in the alloys. For the 6060 alloy, the Si isopleth
section for composition 0.48Mg, 0.03Mn and 0.2Fe (in mass %), is depicted in Figure 2.4a.
Similarly the Mg isopleth for 0.4Si, 0.03Mn and 0.2Fe (in mass %) is depicted in Figure 4b.
The Mg2Si phase is stable bellow 500οC. The α-AlFeSi phase is the first phase stabilized
bellow the liquidus temperature. The range of stability of α-AlFeSi extends to lower
temperatures depending on Mg content. On the contrary Si stabilizes the β-AlFeSi phase.
Additionally the isopleths of Figure 2.4 indicate that transformation of α to β-AlFeSi takes
place as the temperature drops below the solidus. For this reason as mentioned above, a
kinetic calculation, described in section 4.4.4, has been performed in order to determine the
amount of β-AlFeSi phase, which forms after solidification.
33

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
Figure 2.4:(a) Si isopleth for 0.45Mg, 0.03Mn and 0.2 Fe (mass%) and (b) Mg isopleth for
0.4Si, 0.03Mn and 0.2 Fe (L=liquid, F=FCC, α=α-AlFeSi, β=β-AlFeSi, M=Mg2Si)
An isothermal section of the Al-Mg-Si-Fe-Mn system at 300οC, and for 0.2Fe-0.03Mn

(mass %) is shown in Figure 2.5. In agreement to the above observations, Mg stabilizes the
α-AlFeSi phase up to 0.8 mass % Si and Si stabilizes the β-AlFeSi phase, which forms even
for low concentrations of Mg and Si.
Figure 2.5: Isothermal section of Al-Mg-Si with 0.2Fe-0.03Mn (mass %) at 300oC,

(L=liquid, F=FCC, α=α-AlFeSi, β=β-AlFeSi, M=Mg2Si)
34

26/03/2018 03:27:43 EEST - 137.108.70.7
2.4.2. Scheil microsegregation simulations
Typical Scheil solidification paths, for the studied alloys are shown in Figure 2.6a-d as
temperature vs mass fraction solid. The phase sequence during solidification is FCC→α-
AlFeSi→β-AlFeSi→Mg2Si→Si (diamond). The arrows depict the temperature and the mass
fraction of solid where each phase starts to form. Solidus temperatures for these alloys
were calculated and results are depicted in Figure 2.7.
(a) (b)
(c) (d)
Figure 2.6: Typical solidification paths for a)6082, b)6063, c)6005, d)6060
35

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.7: Calculated Solidus temperatures
The mole fraction of phases, formed during solidification, was calculated as a function
of temperature and is shown in Figure 2.8 for 6082 alloy. Formation of α-AlFeSi starts at
622οC while further formation of α-AlFeSi stops, when the β-AlFeSi starts to form at 588οC.
The Mg2Si phase starts to form at 585οC and continuous to grow simultaneously with the β-
AlFeSi phase until the quaternary eutectic temperature (554oC) is reached and the
remaining liquid transforms to a mixture of FCC, β-AlFeSi, Mg2Si and Si.
The phase fractions in the as-cast microstructure are 4.14x10-3 α-AlFeSi, 7.16 x10-4 β-
AlFeSi, 5.32x10-3 Mg2Si and 9.18 x10-4 Si-diamond.
Figure 2.8: Mole fraction of phases as a function of temperature during solidification of Al-
0.38Mg-0.4Si-0.2Fe-0.03Mn (mass%)
36

26/03/2018 03:27:43 EEST - 137.108.70.7
The above phases are not distributed uniformly in the microstructure but segregate to
regions where the liquid solidifies last, i.e. close to grain boundaries. This phase segregation
is depicted in Figure 2.9a, which gives the mole fraction of all phases vs fraction solid (FCC
matrix is excluded). It is seen that the α-AlFeSi phase forms above 85 % solidification while
all other phases form above 95 % solidification. This means that the intermetallic phases
form at the secondary dendrite boundaries towards the end of solidification. Accordingly
the microsegregation of the alloying elements in the FCC matrix follows a similar trend, i.e.
increase towards the end of solidification as shown in Figure 2.9b. The drop in the
concentration profile of Mg and Fe is attributed to the formation of Mg2Si and iron
intermetallics respectively.
(a) (b)
Figure 2.9: Microsegregation of phases (a) and microsegregation of alloying elements (b)
of Al-0.38Mg-0.4Si-0.2Fe-0.03Mn (mass%) alloy during solidification
2.4.3. Parametric study on 6060 alloy
A parametric study was performed for the four variants of alloy 6060, with the
compositions given in Table 1. Both variants, 6060_1 and 6060_3, contain 0.38%Mg but
with 0.4 and 0.47% Si respectively, while both variants 6060_2 and 6060_4 contain
0.48%Mg with 0.4 and 0.47%Si respectively.
37

26/03/2018 03:27:43 EEST - 137.108.70.7
Segregation of phases for 6060_1 and 6060_3 is depicted in Figure 2.10, while for the
6060_2 and 6060_4 in Figure 2.11. The phases depicted are α-AlFeSi in (a), β-AlFeSi in (b),
Mg2Si in (c) and Si-diamond in (d).
It is evident that the larger differences appear in the fraction of the β- AlFeSi phase,
where a higher fraction is formed in the alloys containing a higher amount of Si, i.e. in alloys
6060_3 and 6060_4. This is due to the stabilizing effect of Si on the β-AlFeSi phase. A lower
amount of β-AlFeSi is formed in the high Mg alloys 6060_2 and 6060_4. At the same time the
amount of the strengthening phase is higher in the high Mg alloys 6060_2 and 6060_4.
Therefore it is concluded that Mg has a destabilizing effect on the extrudability-limiting
phase β-AlFeSi, while it increases the amount of the strengthening phase Mg2Si.
(a) (b)
(c) (d)
Figure 2.10: Segregation of phases for 6060_1 and 6060_3 a) α-AlFeSi, b)β-AlFeSi, c) Mg2Si,
d) Si (Diamond)
38

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
(c) (d)
Figure 2.11: Segregation of phases in 6060-2 and 6060_4 alloys, a) α-AlFeSi, b)β-AlFeSi, c)
Mg2Si, d) Si (Diamond)
39

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
(c) (d)
Figure 2.12: Segregation of elements in 6060_1 and 6060_3 alloys a) Si, b)Mg, c)Fe, d)Mn
40

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
(c) (d)
Figure 2.13: Segregation of elements in 6060_2 and 6060_4 alloys a) Si, b)Mg, c)Fe, d)Mn
Segregation of elements for 6060_1 and 6060_3 is depicted in Figure 2.12, while for
the 6060_2 and 6060_4 in Figure 2.13. The elements depicted are Si in (a), Mg in (b), Fe in
(c) and Mn in (d). Elemental microsegregation is evident in all cases; however the elements
do not exhibit the same degree of microsegregation. A microsegragation index δi is defined
as
CM  Co
 
Co
(20)
where CM is the maximum composition in the segregation profile and Co is the composition
of the first solid to form (at zero weight fraction solid). The calculated values of the
segregation index appear in Table 2.3. The element possessing the highest
41

26/03/2018 03:27:43 EEST - 137.108.70.7
microsegregation is Mn with larger microsegregation index compared to the other
elements. Si and Mg have a higher diffusion coefficient in Al and therefore should
homogenize easier than Fe and Mn.
Table 2.3: Segregation index for the 6060 alloy variants

Alloy δFe δSi δMg δMn
6060_1 8.08 8.29 7.33 10,7
6060_2 8.08 8.29 5,45 15,5
6060_3 6.6 8.29 5,58 10.7
6060_4 6.6 8.19 4,88 9,84
2.4.4. Transformation of α to β-AlFeSi after solidification
The Scheil simulation described above defines the phase fractions present in the
microstructure at the end of solidification. According to the isopleth sections in Figure 2.4
transformation of α to β-AlFeSi takes place below the solidus temperature during
equilibrium cooling. However, taking into account the high cooling rates of the solidified
ingots, following the direct-chill casting operation, it should be evaluated whether this
transformation produces any appreciable amount of β-AlFesi phase. A kinetic simulation
using the DICTRA computational kinetics software was performed for the 6082 alloy. The
geometric model used in a previous work [25] was employed and is shown in Figure 2.14.
The calculation domain consists of two regions. The first region corresponds to the fcc
matrix phase while the α-AlFeSi is taken as a dispersed spheroidal phase. The second region
was attached to the right of the first region and corresponds to the β-AlFeSi phase. The
width of the fcc region was taken to be one half of the mean secondary dendrite arm
spacing (10 μm) and the width of the β-AlFeSi corresponded to the mole fraction of the β-
AlFeSi as calculated by the Scheil simulation.
42

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.14: The geometric model for the kinetic simulation using the DICTRA
computational kinetics software performed for the 6082 alloy
For the composition of the 6082 alloy, the mole fraction of β-AlFeSi corresponds to a
width of the β-AlFeSi region equal to 0.0111 μm. The temperature during the simulation
was considered to drop from 600 to 200 οC in 1 h corresponding to a cooling rate of 0.11Ks-
1. In order to take into account the impeding effect of the dispersed α-AlFeSi phase on the
diffusivities of the elements in the fcc matrix phase, a labyrinth factor [26], taken equal to
the volume fraction of the α-AlFeSi phase [27] was employed. The result of the simulation is
given in Figure 2.15 which shows the position of fcc/β-AlFeSi interface as a function of
simulation (cooling) time.
Figure 2.15: The position of fcc/β-AlFeSi interface as a function of simulation (cooling) time
The final position of the interface at 9.99977 μm, to the left of the initial position
(10μm), corresponds to an increase of the volume fraction of β-AlFeSi phase by 2.07 % of
the amount of β-AlFeSi formed during solidification. This value is considered negligible and
43

26/03/2018 03:27:43 EEST - 137.108.70.7
therefore the amount of the β-AlFeSi phase, used for the experimental validation and the
construction of the maps in Section 4.7, was taken directly from the Scheil simulations.
2.5. Experimental validation
The purpose of this section is to provide the necessary experimental data, regarding
phase identification and phase fractions, for the validation of the computational
methodology used in this work. As mentioned above, the validation was performed on alloy
6082, since it exhibited the largest fraction of intermetallics due to its higher alloy content.
The as-cast microstructure of 6082 is depicted in the SEM micrograph Figure 2.16a-c.
The EDS signal from areas 1 and 2 consisted of Mg and Si indicating that this phase is Mg2Si.
The signal from area 3 consisted of Al, Fe, Si and Mn indicating that it is the α-AlFeSi phase,
which contains Mn. The signal from area 4 consisted of Al, Fe and Si (without Mn) and was
identified as β-AlFeSi. In order to further verify the phase identification, the composition of
the α-AlFeSi was determined quantitatively.
(a) (b)
(c)
Figure 2.16: The as-cast microstructure of 6082 alloy as observed in the SEM
44

26/03/2018 03:27:43 EEST - 137.108.70.7
It should be mentioned that the interaction volume of the EDS measurement
performed in this work was about 1μm2 while the average size of α-AlFeSi particles on
which the EDS signal was taken was 20-30 μm2. The results are listed in Table 2.4 and
compared with measured literature data. The measured composition of α-AlFeSi lies within
the range reported in the literature and, therefore, the phase can be identified as α-AlFeSi.
Table 2.4: Chemical composition of α-AlFeSi phase (mass %)

Phase Si Fe Mn Reference
α-AlFeSi 8-12 11-13 14-20 [1]
10 -12 10-15 15-20 [28]
5.5-6.5 5.1-27.9 14-24 [29]
5-7 10-13 19-23 [30]
6.3-11.5 7.4-14.3 5.5-9.8 This work
Furthermore elemental area mapping in 6082 alloy was performed in order to

identify the presence of the elements in the observed intermetallic phases. The elemental
map images in Figure 2.17 depict the spatial distribution of elements, specifically Al, Mg, Si,
Fe, and Mn. In this way it is possible to identify phases in the microstructure that would be
difficult to identify by optical microscopy. Figure 2.17 depicts a SEM micrograph in the
centre with corresponding EDS maps for Al in (a), Si in (b), Mg in (c), Fe in (d) and Mn in (e).
The EDS analysis indicates that the black oval particles are Mg2Si strengthening phase
because of the presence of Mg and Si element (Figures 2.17b-c) and absence of Al signal
(Figure 2.17a). Besides the Mg2Si phase other Si containing phases are the intermetallics α
and β-AlFeSi. The low solid solubility of Fe in aluminum leads to the formation of the above-
mentioned intermetallics and this has been confirmed in Figure 2.17d with a strong iron
signal. The presence of Mn in Figure 2.17e indicates that this phase is the α-AlFeMnSi phase.
45

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
(c) (d) (e)

Figure 2.17: Elemental area mapping (SEM/EDS) in 6082 alloy a) Al, b)Si, c)Mg, d)Fe, e)Mn
Besides the identification of phases in the SEM, it is important to validate the matrix
composition, adjacent to the intermetallics, close to the boundary. The results are shown in
Figure 2.18, with a SEM microstructure in (a), the Scheil simulation for the
microsegregation of elements in (b) and an EDS spectrum from the matrix region adjacent
to the boundary in (c). The measured compositions are compared against the calculated
data from the Scheil simulation in Table 2.5. The agreement between measured and
calculated compositions is considered satisfactory.
46

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.18: Matrix composition adjacent to boundary intermetallics: (a) SEM micrograph,
(b) Scheil microsegregation profile and (c) EDS signal
47

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 2.5: Comparison between calculated and measured (EDS) matrix composition
adjacent to boundary intermetallics
Matrix composition adjacent to boundary
Measured Mass Calculated

Element
(%) (%)
Al 97.8
Si 1.2 1.32
Mg 0.7 0.59
Mn 0.3 0.5
Fe 0 0
Using the procedure outlined in the experimental procedures section, the SEM phase
identification was correlated with optical micrographs. Typical microstructures are
depicted in Figure 2.19a-c for 6082, 6063 and 6060 respectively. The microstructure
consists of Al-rich solid solution dendrites while the intermetallic phases are segregated at
the secondary dendrite arm boundaries. These phases are identified as Mg2Si (black), α-
AlFeSi (grey plate-like with rounded edges) and β-AlFeSi (grey plate-like). These
observations are also in very good qualitative agreement with available metallographic
observations in the literature [2, 31].
In Figure 2.20 6082 alloy is depicted and the morphology of the intermetallic
compound α-AlFeSi appears with a typical lamellar morphology known as “Chinese-script”
structure.
48

26/03/2018 03:27:43 EEST - 137.108.70.7
(a)
(b)
(c)
Figure 2.19: Typical microstructure of alloys (a) 6082, (b)6063 and
(c)6060
49

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.20: Typical morphology of intermetallic α-AlFeSi in 6082 known as “Chinese-script”
Further metallographic examination has been performed in the 6082 alloy. The Scheil
simulation on 6082 (Figure 2.6a) indicated that the last liquid solidifies by a quaternary
eutectic consisting of FCC + β-AlFeSi + Mg2Si + Si at 544οC. The amount of this quaternary
eutectic is small compared with the primary phases and forms when less than 10% liquid
remains, in agreement with ref. [31]. The quaternary eutectic constituent is depicted in
Figures 2.21a-b.
The presence of spherical Mg2Si and eutectic phases are also shown in Figure 2.21b.
(a) (b)
Figure 2.21: The (a) quaternary eutectic structure and spherical Mg2Si and eutectics of
6082 alloy in as-cast condition
50

26/03/2018 03:27:43 EEST - 137.108.70.7
Following phase identification, the measurement of phase fractions was performed by
quantitative metallography using the image analysis software Image –J. Low magnification
micrographs were selected in order to have the most representative microstructure. In
Figure 2.22, representative pairs of micrographs (left) and Image-J displays (right) are
shown.
Scheil calculations of phase fractions are compared with experimental measurements
in Table 2.6 for α-AlFeSi, β-AlFeSi, Mg2Si and quaternary eutectic (FCC + Mg2Si + β-AlFeSi +
Si). For the case of iron intermetallics the sum of α and β-AlFeSi is compared with
experimental data. Taking into account the experimental error involved when measuring
low volume fractions, the agreement between the Scheil simulation predictions and the
experimental measurements is considered satisfactory. Therefore computed fractions from
the Scheil simulations can be used for the phase fraction mapping of intermetallic phases in
the as-cast microstructure. This is discussed in the following sections.
51

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 2.22: Microstructures (left) and corresponding Image-J displays used for the
measurement of phase fractions
52

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 2.6: Comparison between calculated and measured phase fractions in the as-cast
6082 alloy
Measured (x10-3)
Calculated (x10-3)
Quantitative
Scheil simulations
metallography
β-AlFeSi 2.57
6.54 6.07
α-AlFeSi 3.97
Mg2Si 3.81 3.65
Quaternary eutectic 3.8 4.1
2.6. Effect of Mn on Mg2Si and β-AlFeSi phases
The results of the Scheil simulations have shown that increasing the Mn level from a
low 0.03% to a high 0.45mass% has two major effects, for the high-strength alloys with
more than 0.4mass% Si:
 reduction of β-AlFeSi mole fraction
 reduction of Mg2Si mole fraction
These reductions are expressed for β-AlFeSi as
f 0.2 Mn  f 0.03Mn
f  
f 0.03Mn
(21)
and for Mg2Si as
2 Si  f Mg 2 Si
0.2 Mn 0.03 Mn
f Mg
f Mg 2 Si  0.03 Mn
f Mg 2 Si
(22)
where f, denotes the mole fractions of these phases at the two levels of Mn content 0.03 and
0.2 mass%. These reductions are plotted in Figure 2.23 for the β-AlFeSi phase and Fig.24a, b
for the Mg2Si phase as a function of Si content (above 0.4mass%) for various levels of Mg
ranging from 0.5 to 1.2 mass%. In general the reduction in the β-AlFeSi is up to 15% while
the respective reduction in the Mg2Si can reach 18%. Regarding the β-AlFeSi phase, with
increasing the Si content (for fixed Mg content) the reduction first increases and then
decreases. The maximum reduction of β-AlFeSi is located between 0.5-0.6%Si depending on
53

26/03/2018 03:27:43 EEST - 137.108.70.7
the Mg level. In this range of Si, the reduction in β-AlFeSi increases with Mg, reaching a
maximum 15% for 0.8%Mg and then decreases for higher Mg contents. The maximum
reduction in β-AlFeSi is 15% for 0.5Si-0.8Mg (0.2Mn).
Figure 2.23: Mn-induced reduction of β-AlFeSi phase as a function of Si alloy content for
various levels of Mg in the alloy
Regarding the reduction in the Mg2Si phase, two behaviours are sheen. The first is
shown in Figure 2.24a for the 0.4-0.6 Mg level (mass%) where the reductions in Mg2Si are
higher and can reach values up to 27% depending on the Si level. In this range the lowest
reduction in Mg2Si (2%) appears for the 0.8Si-0.6Mg composition. The second is shown in
Figure 2.24b for the higher level of Mg between 0.7 and 1.2 mass% where the reductions in
Mg2Si are lower and limited to values up to 13%. Here the minimum reduction (0%) in
Mg2Si appears for the 0.5Si-1.2Mg composition. It is clear that the Mg-Si compositions for
the highest reduction in β-AlFeSi do not coincide with those for lowest reduction in Mg2Si.
54

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
Figure 2.24: Mn-induced reduction in Mg2Si phase as a function of Si alloy content for (a) 0.4-0.6 and
(b) 0.8 1.2 Mg (mass %) in the alloy
2.7. Mapping of intermetallic phases
Alloys 6060 and 6063 contain 0.03 mass% Mn while alloy 6082 contains 0.45 mass%
Mn. Therefore, Scheil calculations for phase fraction mapping were conducted for these two
Mn levels. The mole fraction of β-AlFeSi and Mg2Si, is depicted in the contour plots of fixed
mole fractions in Figures 2.25 (a) and (b) respectively as a function of Mg and Si content for
the two Mn levels of 0.03 and 0.45 mass %. The mole fraction of the β-AlFeSi phase
decreases towards the lower right corner of the contour plot, indicating that low β-AlFeSi
can be obtained at lower Si and higher Mg contents. The 0.45 % Mn iso-β lines lie above the
respective 0.03 % Mn iso-β lines, indicating that for fixed Mg content, higher Mn allows the
use of higher Si, without further increase of β-AlFeSi (Si stabilizes β-AlFeSi). The spacing
between the iso-β lines for the two Mn levels increases towards the upper left corner of the
contour plot, indicating that the effect of Mn (in reducing β-AlFeSi) becomes stronger at
higher Si and lower Mg contents.
55

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
Figure 2.25: Contour plots (maps) of fixed mole fractions of (a) β-AlFeSi and (b) Mg2Si
phases in the as-cast microstructure of Al-Mg-0.2Fe-Mn alloys (full lines 0.03
Mn, dotted lines 0.45 Mn, mass%)
Regarding the Mg2Si phase (Figure 2.25b) it is clear that the mole fraction increases
towards the upper right corner of the contour plot, indicating that an increase of both Mg
and Si results in an increase of Mg2Si, with Mg having a stronger effect. The 0.45 % Mn iso-
Mg2Si lines lie above the 0.03 % Mn lines, indicating that a higher Mg content is required to
obtain the same Mg2Si mole fraction in the microstructure.
Mapping of both the Mg2Si phase and the β-AlFeSi phase across the Mg-Si space can be very
useful for alloy design purposes. Such maps are shown in Figure 2.26 (a) and (b) for the two
Mn levels of 0.03 and 0.45 % Mn respectively. The mole fractions of the β-AlFeSi phase
range from 0.001 to 0.005 while those for the Mg2Si range from 0.002 to 0.009.
56

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
Figure 2.26: Contour plots (maps) of fixed mole fractions β-AlFeSi (solid lines) and Mg2Si
(dotted lines) in as-cast microstructure of Al-Mg-Si-0.2Fe-Mn alloys for (a)
0.03 Mn and (b) 0.45 Mn, mass%. Compositions for alloys 6060, 6063 and
6082 are indicated on the maps.
The position of alloys 6060, 6063 and 6082 is also indicated on the maps. All three
alloys possess a low fraction of β-AlFeSi indicating an excellent extrudability potential. In
addition, the 6082 alloy possesses a higher fraction of Mg2Si indicating a higher strength
potential compared to 6060 and 6063. The Mg2Si is maximized towards the upper right
corner while the β-AlFeSi phase is minimized towards the lower right corner of the maps.
This allows selection of optimum compositions for favourable as-cast microstructures in
terms of the two phases. The Mn effect can be summarized as the potential to move to
higher Mg and Si compositions in order to obtain higher amount of Mg2Si without at the
same time increasing the β-AlFeSi phase in the as-cast microstructure.
57

26/03/2018 03:27:43 EEST - 137.108.70.7
2.8. Conclusions
From the results presented in this chapter the following conclusions can be drawn:
Computational alloy thermodynamics, based on the CALPHAD approach, has been applied
for the calculation of equilibrium phase constitution as a function of alloy composition and
temperature. Isopleth phase diagrams in multicomponent alloys have been calculated
permitting the determination of the stability limits of the various phases in the system.
This approach was also applied to carry out Scheil-Gulliver solidification simulations
and calculate the resulting microsegregation of elements and phases in the as-cast
microstructure.
Segregation profiles for elements and phases across the grain were determined as a
function of alloy composition.
Standard metallographic techniques including optical microscopy and scanning
electron microscopy (SEM) were applied to reveal the microstructure of the as-cast and
homogenized materials. Phase fractions were measured by applying quantitative image
analysis. The results of the simulations were confirmed experimentally.
The variation of the mole fractions of the extrudability-limiting β-AlFeSi phase and the
strengthening Mg2Si phase with alloying elements can be mapped over the useful range (0-
1.2 mass%) in the Mg-Si composition space.
The constructed maps indicate that low mole fractions of β-AlFeSi are associated with
lower Si and higher Mg compositions. On the other hand, high mole fractions of Mg2Si are
associated with both higher Si and Mg compositions, with Mg possessing a stronger effect.
Construction of maps for different levels of Mn has shown that addition of Mn could allow
for higher alloying with Mg and Si, in order to obtain higher amount of Mg2Si, without at the
same time increasing the β-AlFeSi intermetallic phase in the as-cast microstructure
58

26/03/2018 03:27:43 EEST - 137.108.70.7
2.9. References
[1] G. Mrowka-Nowotnik, J. Sieniawski, M. Wierzbinska, Arch. Mater. Sci. Eng. 28(2007) 69-
76.
[2] Y.L. Liu, S.B. Kang, J. Mater. Sci. 32 (1997) 1443-1447.
[3] P.Mukhopadhyay Inter. Sch. Res. Net. ISRN Metallurgy (2012)
[6] W.D. Fei, S.B. Kang, J. Mater. Sci. Lett. 14 (1995) 1795-1797
[7] H. Tanihata, T. Sugawara, K. Matsuda, S. Ikeno, J. Mater. Sci. 34 (1999) 1205-1210.
[8] A. Verma, S. Kumar, P.S. Grant, K.A.Q. O’Reilly, J. Alloys Compd. 55 (2013) 274-282.
[9] A. Wimmer, J. Lee, P. Schumacher, Berg-und Hutten. Monatshefte 157 (2012) 301-305.
[10] Y.L. Liu, S.B. Kang, J. Mater. Sci. 32 (1997) 1443-1447
[11] Y. Du, Y.A. Chang, S.D. Liu, B.Y. Huang, F.Y. Xie, Y. Yang, S.L. Chen, Z. Metallkunde 96
(2005) 1351-1362
[12] J.O. Andersson, T. Helander, L. Höglund, P. Shi, B. Sundman, Calphad 26 (2002) 273
312.
[13] H.L. Lukas, S.G. Fries, B. Sundman, Computational thermodynamics, the Calphad
method, Cambridge university press, 2007
[14] J. Ansara, A.T. Dinsdale, M.H. Rand: COST 507: Thermochemical database for light
metal alloys, European Community publication, 2, EUR 18449 (1991).
[15] G. Inden, Physica 103B (1981) 82-100
[16] B. Sundman, Agren, J., Journal of Physics and Chemistry of Solids 42 (1981) 297-301.
[17] J. Lacaze, L. Eleno, B. Sundman, Metall. Mater. Trans. A 41A (2010) 2208-2215.
[18] E. Scheil, E. Metallkde, 34 (1942) 70.
[19] C.H. Gulliver, J. Inst. Metals, 9 (1913) 120.
[20] R. W. Balluffi, S. M. Allen, W. G. Carter, “Kinetics of Materials” Wiley- Interscience
(2005)
[21] N. Belov , D. Eskin, A. Aksenov Elsevier (2005) 17-33
[22] M. Wu, A. Ludwig, Metall. Mater. Trans. A 38A (2007) 1465-1475.
[23] M. Wu, J. Li, A. Ludwig, A. Kharicha, Comput. Mater. Sci. 79 (2013) 830-840
[24] ASTM Specification, E-522 (1998)
[25] Haidemenopoulos, G.N., Kamoutsi, H., Zervaki, A.D., Jour. Mat. Proc. Tech. 212 (2012),
2255-2260
[26] Rayleigh, J.W, Ph. Mag. 34, (1892) 481–502
[27] Zhang, W., Du, Y., Peng, Y., Wen, G., Wang, S., In. Jour. R. Met. H. Mat. 104, (2013) 721-
735
[28] Mondolfo L.F., 1976. Aluminum Alloys: Structure and properties. Butterworths,
London-Boston
[29] Warmuzek, M., Rabczak, K., Sieniawski, J., Jour. Mat. Proc. Tech. 175, (2005) 421-426.
[30] Liu, Y.L., Kang, S.B., Kim, H.W., Mat. Let. 41 (1999), 267-272.
[31] Mrowka-Nowotnik, G., Sieniawski, J., Wierzbinska, M., Arch. Mat. Sc.Eng. 28 (2007), 69-
76.
59

26/03/2018 03:27:43 EEST - 137.108.70.7
60

26/03/2018 03:27:43 EEST - 137.108.70.7
3. Simulation of Homogenization
3.1. Introduction
Homogenization is a significant link of the process chain of the extrudable 6xxx Al-
alloys. It follows casting and takes place before extrusion. The main purpose of
homogenization is, therefore, to remove the detrimental effects of microsegregation left
over by the casting process and to “condition” the material for the extrusion process. In that
respect, homogenization can be considered as a “key process”, since it controls in a large
degree the extrudability of the alloy. The homogenization has received considerable
attention, mostly in experimental terms, both in laboratory and industrial scale. There is a
large amount of published work, briefly reviewed in the next section, on the effects of the
homogenization parameters on the degree of homogenization. However, despite its
importance, the work on simulation of the homogenization process is limited. This is due to
the complexity of the process, since there is a series of phase transformations, which take
place concurrently, in a multiphase and multicomponent system, such as the Al-Mg-Si-Fe-
Mn system under consideration. The most important phenomena, which take place during
homogenization of the as-cast billets, are the following:
1. Removal of the microsegregation of the alloying elements
2. Removal of non-equilibrium low melting point eutectics, particles and segregation
gradients which provide areas with low melting point in order to avoid cracking or tearing
during subsequent hot working
3. Shape control (spheroidization) of remaining intermetallics and dispersoids with sharp
edges such as Fe-based intermetallics, which give poor workability
4. Formation of secondary particles (dispersoids) for grain size control during extrusion or
rolling.
5. The dissolution of the Mg2Si phase and the reprecipitation during cooling
6. The β → α-AlFeSi transformation.
The homogenization process parameters are the homogenization temperature,
homogenization time and the cooling rate from the homogenization temperature. The
correct selection of the homogenization temperature ensures that all necessary
61

26/03/2018 03:27:43 EEST - 137.108.70.7
transformations are activated and take place in an appreciable rate. The homogenization
time should also be selected carefully in order to ensure that the “conditioning” of the billet
for extrusion is accomplished. A longer, than necessary, homogenization time usually
results in increased costs and loss of productivity in the cast house. The cooling rate after
homogenization is also very important, since it controls the precipitation reactions in the
billet before extrusion. The cooling after homogenization is treated in Chapter 5. From the
above discussion, it is clear that the control of the homogenization parameters is an
important issue in the production of extrudable 6xxx Al-alloys.
The aim of the present chapter is to introduce a robust computational methodology
for treating the homogenization problem. This methodology constitutes a significant step
towards the design of the homogenization process. This is particularly important in the
development process of new alloys, since the design of homogenization can lead to
significant savings in cost and time for the alloy development.
3.2. Literature review
Current literature is reviewed in this section regarding the effect of homogenization

parameters on the removal of microsegregation and on the phase transformations taking
place during the homogenization process. As stated above, most of the work up to date is
experimental, either on a laboratory or on an industrial scale.
Homogenization comprises of three major steps, heating the as-cast billet with a
particular rate, holding at a constant temperature for a certain time, and cooling with a
proper cooling rate. The introduction of continuous homogenization lines has significantly
improved the temperature uniformity of homogenization allowing the holding times to be
minimized. Batch homogenization, however, provides flexibility in practices tailored for
different aluminum alloys. Soft 6060 and 6063 alloys are best homogenized at a higher soak
temperature than harder alloys such as 6061 and 6082 [1]. After direct chill (DC) casting,
the solid solution is supersaturated and the supersaturation increases from the center to
the edges of the dendritic branches because of microsegregation.
62

26/03/2018 03:27:43 EEST - 137.108.70.7
Major reactions during homogenization involve the dissolution of Mg2Si, coarsening of
dispersoids, spheroidization of undissolved particles and the transformation β-AlFeSi→α-
AlFeSi, which is important due to the low high-temperature ductility of the β-phase [2].
The homogenization practice has been investigated experimentally where both industrial
and academic researchers have focused on the influence of temperature and time of the
procedure on the microstructure and mechanical properties of the homogenized billets [3-
6]. Long-time homogenization treatments at 545 °C for 20 h, indicated that Mg2Si and Si
intermetallic compounds are dissolved into the matrix completely, and the remaining
phases were α(Al) matrix and Al15(FeMn)3Si2 or α-AlFeSi [7].
The controlling factor in the homogenization of an alloy is the diffusivity of the
respective alloying elements at the homogenization temperature. The higher the
homogenization temperature, the faster the elemental homogeneity can be obtained and,
thus, the more efficient this practice becomes in terms of industrial throughput. However
there exists an upper limit in the homogenization temperature. It has been found that
homogenization of 6082 alloy at temperature close to 540οC leads to partial local melting of
dendrite cell boundaries while at even higher temperatures full surface melting and alloy
degradation occurs [8]. The homogenization temperature, therefore, should not exceed the
solidus temperature in the particular alloy since this could result in localized melting. This
melting could cause structural damage of a type that cannot be repaired after
homogenization. This damage has been reported as excessive void formation, segregation,
blistering and cracking [9]. The exact solidus temperature of the material cannot be
predicted from the phase diagram because the microstructure is far from equilibrium. As a
result of microsegregation, regions exhibiting high concentration of alloying elements melt
at temperatures below the equilibrium solidus temperature. This is especially true for the
low-melting eutectics. The eutectic melting temperatures are presented in Table 1 for
several reactions.
The β-AlFeSi to α-AlFeSi transformation is probably the most important reaction
during the homogenization process. The α-AlFeSi has a cubic crystal structure and globular
morphology while the β-AlFeSi possesses a monoclinic structure and a plate-like
morphology, limiting the extrudability of the as-cast billet by inducing local cracking and
surface defects in the extruded material. The β to α-AlFeSi phase transformation
63

26/03/2018 03:27:43 EEST - 137.108.70.7
considerably improves the extrudability of the aluminum since the transformed α-particles
in the homogenized material improve the ductility and the surface quality of the extruded
material [10, 11].
Additional microstructural changes, such as the dissolution of Mg2Si or Si particles,
also occur during homogenization. However, since the Mg2Si or Si particles dissolve rather
fast, it is the β-to-α transformation kinetics, which determines the minimum
homogenization time that is needed to get the material in a suitable state for extrusion.
Table 3.1: Eutectic melting temperatures in Al-Mg-Si alloys [10]

Melting Temperature Reaction
~575oC Al + β – AlFeSi + Si → Liquid
o
~576 C Al + β – AlFeSi + Mg2Si → Liquid + α – AlFeSi
~577oC Al + Si → Liquid
o
~587 C Al + Mg2Si → Liquid
~612oC Al + β – AlFeSi → Liquid + αh – Al(FeMn)Si
o
~630 C Al + α – AlFeSi → Liquid + Al3Fe
Another transformation that takes place during holding at the homogenization

temperature is the spheroidization of the non-dissolvable α-AlFeSi phase. This issue will be
discussed in Chapter 4.
Cooling of the billet after homogenization is very important in order to obtain the
suitable microstructure to improve extrudability as well as the final mechanical properties
of the extrusion. A fully solutionized billet, as obtained at the end of homogenization
holding, is difficult to extrude due to solid solution strengthening. The motivation behind
controlled cooling is the establishment of a specific precipitation state [13]. During cooling
to room temperature precipitation of Mg2Si and other Fe, Mn and Cr-containing phases
takes place. The size and the density of these particles are crucial because they enhance the
recrystallization resistance of the material [13], which is beneficial for the final mechanical
properties. Respectively, the size and the density of Mg2Si particles must be such that they
can be dissolved easily before extrusion, leading to improved extrudability [14]. It has been
shown that smaller Mg2Si particles, obtained after cooling, dissolve faster during extrusion
[13-15]. The amount of Mg and Si in solid solution, and hence the amount of Mg2Si
precipitated, are highly affected by the cooling rate from the homogenization temperature.
64

26/03/2018 03:27:43 EEST - 137.108.70.7
Slow cooling tends to produce coarse Mg2Si particles while rapid cooling trap the Mg and Si
in solution with little or no Mg2Si precipitation [16,17]. A decrease in the cooling rate
increases the Mg2Si precipitation and decreases the amount of Mg in solid solution. Mn
additions up to 0.027% have no effect on Mg2Si precipitation [17]. However the effect of Mn
on the precipitation during cooling has not been investigated extensively. Zajac et al. [14]
and Nowotnik et al. [18] investigated the effect of the cooling rate on the final mechanical
properties of the 6005, 6082 and 6063 aluminum alloys. Usta et al. [26] and van de
Langkruis [15,19] studied the dissolution-coarsening kinetics of the Mg2Si particles during
reheating. Reiso et al. [16] correlated the cooling rate to the maximum extrusion velocity
for various chemical compositions of AlMgSi alloys. Birol et al. [20,21] studied the effect of
the homogenization temperature and time and the effect of cooling rate and the
microstructure of 6063 and 6082 alloys through metallographic techniques.
As mentioned above, most of the published work approaches the homogenization
process experimentally. The modelling or simulation work is rather limited. In a previous
doctoral thesis [22] at the Laboratory of Materials an attempt to simulate the dissolution of
the Mg2Si was presented. Kuijpers et al. [23] simulated the dependence of the β-AlFeSi to α-
Al(FeMn)Si transformation kinetics on the alloying elements and found a significant effect
of the Mn concentration as it increases the transformation rate. Haidemenopoulos et al. [24]
modeled the transformation of iron intermetallics during homogenization of 6xxx alloys
and also reported similar trends. However an integrated model, taking into account the
prior microsegregation and, most importantly, taking into account the concurrent Mg2Si
dissolution and β to α-AlFeSi transformation, in a multiphase, multicomponent system,
exhibiting grain size variations does not exist. It is the aim of this chapter to introduce such
a model. The computational methodology is presented in the following section. The
experimental validation of the model was performed on alloy 6082.
65

26/03/2018 03:27:43 EEST - 137.108.70.7
3.3. Methodology
3.3.1. Computational Method
The key elements of the computational approach to simulate the homogenization process
are the following:
 The fact that the grain size in the as-cast microstructure is not uniform is incorporated in
the model. A Dual-Grain Model (DGM) has been developed to serve as the computational
region. The region represents two neighboring grains, in contact, with significantly
different grain sizes. In this way the fact that the transformation rate is higher in the
smaller grains is taken into account. The DGM region serves as the representative volume
element (RVE) for the solution of the homogenization problem.
 The initial conditions in terms of composition and phases present in the DGM correspond
to the actual microsegregation of the as-cast microstructure. The composition as well as
the phase fractions profiles that were introduced as initial conditions was the Scheil
microsegregation profiles, calculated from the relevant Scheil simulations for the specific
alloy. The Scheil simulations were presented in Chapter 2.
 Two key phase transformations are treated concurrently in both grains of the DGM. These
are the Mg2Si dissolution and the β to α-AlFeSi transformation. Both are diffusional
transformations and their rate is controlled by the diffusion of alloying elements. Since
the diffusion of Mg and Si is faster than the diffusion of Fe and Mn, it is expected that the
dissolution of Mg2Si is faster than the β to α-AlFeSi transformation. In the DGM model,
only lattice diffusion through the matrix has been considered. Grain boundary diffusion or
dislocation assisted diffusion were not considered, since in the homogenization process
the diffusion takes place from the boundaries to the grain interiors in order to eliminate
the microsegregation gradients. In addition, the dislocation density of the material has
normal values, since the material has not undergone any cold deformation before
homogenization.
 The intermetallic phases were treated as non-diffusion phases. This means that they act
entirely as sinks or sources of alloying elements. Diffusion is restricted exclusively to the
matrix.
66

26/03/2018 03:27:43 EEST - 137.108.70.7
 The diffusion problem is one-dimensional and is treated in a multicomponent (Al-Mg-Si-
Fe-Mn) and multiphase system. The phases present in the system are the FCC aluminum
matrix and the dispersions of Mg2Si and the iron intermetallics β-AlFeSi and α-AlFeSi.
The DGM calculation domain, developed in order to simulate homogenization

considering the different grain sizes between neighboring grains is depicted in Figure 3.1.
Following metallographic examination, two sizes of 90μm and 20μm respectively
were selected for the construction of the DGM. The difference between the grain sizes is
significant and allows the study of the grain size effect on homogenization.
Figure 3.1: The Dual Grain Model considering the different grain sizes between
neighboring grains
The total region size is 110μm considering the two grains in contact. Due to the
symmetry, only half of the grains were considered.
In order to input the concentration profile of alloying elements as well as the phase
fractions profiles resulting from the Scheil simulations as initial conditions, the fraction
solid axis fs was converted to a distance axis via the relation
d
xs  f s (1)
2
67

26/03/2018 03:27:43 EEST - 137.108.70.7
where d is the grain size region of the DGM (90 or 20 μm). Thus the diffusion problem is
solved in one direction (x). The Mg2Si phase as well as the intermetallics α-AlFeSi and β-
AlFesi phases are taken as a dispersed phases in the FCC matrix.
Figure 3.2: Dispersed Phase model in DICTRA
This diffusion problem was solved with the computational kinetics package DICTRA
[25]. One basic model assumption is that the growth and dissolution rates of the precipitate
phases are very high compared with the diffusion rates in the matrix. Alternatively, the
matrix diffusion is the controlling mechanism of the overall kinetics. This assumption is
tolerable for the high homogenization temperatures, since the growth-dissolution rates are
very high and the particles reach the equilibrium state very fast. Consequently the local
conditions are rather concentration dependent than time-dependent. Locally, most of the
solute is trapped into the particles, reducing the concentration gradients and the diffusion
rate in the matrix. The “Dispersed Phase” module in DICTRA was employed as shown in
Figure 3.2. It treats problems involving diffusion through microstructures containing
dispersed precipitates of secondary phases. Diffusion is assumed to take place only in the
matrix phase. The dispersed phases are considered as “non-diffusion phases”. They act as
point sinks or sources of solute atoms in the simulation and their fraction is calculated from
the average composition in each node, assuming local equilibrium. The growth or
dissolution of phases leads to adjustments in the concentration profiles of the elements to
be used in the next time-step of the calculation. What changes during the calculations in
each time step is the volume fraction of phases and the local matrix compositions
(composition profiles) through the matrix diffusion.
The mathematical formulation of the multicomponent diffusion problem is based on
the Onsager principles and is described here briefly. A full description is presented in [26-
28].
68

26/03/2018 03:27:43 EEST - 137.108.70.7
Following the treatment in [26], the first Onsager principle is
 


T    J i Fi   J i (1)
 1  1 x
stating, that the rate of entropy production (free energy dissipation) is the sum of the
products between forces and fluxes. In the diffusion case the “forces” are the chemical
potential gradients, which drive diffusion, and “fluxes” are the diffusive fluxes. The second
Onsager principle states that the forces are linearly related to the fluxes by the expression
N
J i   Lik Fkj (2)
k 1
In the V-frame, where the volume is fixed, the fluxes of the N components are related
with the expression
N
  J
i 1
i 0 (3)
meaning that there are N-1 independent fluxes. The i are the atomic volumes of the
elements. This means that the Nth flux is

1
1
JN  

 J
i 1
i i (4)
Using this expression, equation (1) becomes

N 1

 
T     Ji ( i  i  N ) (5)
i 1
x N
and the “forces” Fi become

    N 
Fi    i  i  (6)
 x  N x 
so that the fluxes of equation (2) become
N 1 N 1
    N 
J i   Lij Fj   Lij  j  j  (7)
j 1 j 1  x  N x 
The fluxes in equation (7) are expressed in relation to the chemical potential gradients.
Now the fluxes will be written in the traditional Fick’s 1st law, so that they are related to
composition gradients.
The chemical potential of a component i depends on the concentration
69

26/03/2018 03:27:43 EEST - 137.108.70.7
i  i  c1, c2 ,..., cN 1  where i  1, 2,..., N 1 (8)
since there are N-1 independent concentrations. Additionally the Gibbs-Duhem relation
imposes an additional constraint that only N-1 chemical potentials can vary independently
so that
i N 1 i c j
 (9)
x j 1 c j x
The fluxes in equation (7) then become

N 1    N  ck
N 1 N 1

J i   Lij Fj   Lij   j  j  (10)
j 1 j 1 k 1  ck  N ck  x
and the interdiffusion coefficient can be identified as

N 1     N 
Dij   Lik  k  k
 c  c 
(11)
k 1  j  j 
The diffusion coefficient can be written as the product of two matrices

N 1
Dij   Lik  kj (12)
k 1
where Lik are the Onsager coefficients related to the atomic mobilities and  kj are the
relevant thermodynamic factors which express the dependence of the chemical potential on
composition. It is to be noted that in DICTRA, the mobilities are stored in the so-called
mobility database, while the thermodynamic factors are calculated via a link to Thermo-
Calc. In this way DICTRA takes into account the composition dependence of the diffusion
coefficient. The analysis of a N component system requires N-1 independent concentrations
and (N-1)2 independent interdiffusivities. In the Al-Mg-Si-Fe-Mn system considered in this
problem, the diffusion matrix contains 16 interdiffusivities.
Coming back to the DGM, the solution of the diffusion equation is performed under the
following boundary conditions. Considering a closed system, the boundary conditions are:
J i (0, t )  J i ( L, t )  0 (13)
Or in terms of concentration gradients

ci
 0 at x=0 and x=L (14)
x
70

26/03/2018 03:27:43 EEST - 137.108.70.7
The initial conditions for solving the diffusion problem are the results of the Scheil
calculations and converted over the diffusion distance for t=0 and can be expressed as
follows for the elements:
ci ( x,0)  cis ( x) , i=Mg, Si, Fe, Mn (15)
and for the initial fraction of the phases

f k ( x,0)  f kScheil , k=Mg2Si, α-AlFeSi, β-AlFeSi (17)
where cis ( x) and f kScheil are the compositions and phase fractions respectively resulting
from the Scheil simulation presented in Chapter 2.

In DICTRA the mobility parameter, M i for an element i in a given phase can be
described by a frequency factor M ko and activation energy Gi* and these are related in
equation:
M io G*
Mi  exp( i ) (12)
RT RT
where R is the gas constant and T is the absolute temperature. Both M i and Gi* are
composition dependent. In the spirit of the CALPHAD approach, in a multicomponent

system these parameters are expressed with a Redlich-Kister–Muggianu polynomial. It is
therefore important to state that the composition dependence of the diffusion coefficients
comes not only through the thermodynamic factor but from the mobilities as well.
The thermal cycle obtained for this study consisted of four stages: heating, holding at the
homogenization temperature for several hours, cooling to room temperature and pre-
heating (before extrusion). Figure 3.3 depicts the temperature – time profile of the
homogenization process mentioned above, which includes typical homogenization times
and temperatures.
71

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 3.3: The temperature – time profile of the homogenization process
Regarding the time step used in the simulations, the maximum time step allowed, or
the highest time step actually taken in the calculation, would affect the final result of the
simulation. This effect will be most evident if one or more of the diffusing species have low
solubility in the matrix phase. When this is the case, a supersaturation is created in the
matrix phase during a diffusion step and if too large a time step is allowed, then too much
supersaturation is created. For this reason a time step sensitivity analysis has been
performed to assure that there is no time step dependence of the solution.
3.3.2. Experimental procedures
In order to validate the results of the computations of the Dual Grain Model,
laboratory homogenization treatments were performed in selected alloys listed in Table 2.1
in order to study the progress of homogenization. The homogenization treatments were
performed in the temperature range of 540oC-580oC for 30min, 4hr, 6hr and 32 hr. In order
to alter the cooling rate, three cooling methods were used: water Quenching (Q), Air cooling
(AC) and Forced air cooling (FC). The effect of the cooling rate is discussed in Chapter 5.
After the homogenization heat treatment, the specimens were prepared for standard
metallographic examination in order to reveal the microstructure and apply quantitative
image analysis. The evolution of the elemental profiles (removal of microsegregation) was
performed by SEM-EDX measurements. Phase characterization using EDS elemental
72

26/03/2018 03:27:43 EEST - 137.108.70.7
mapping has also been performed in the alloys. The evolution of phase transformation with
homogenization time was performed with XRD analysis. Details of the experimental
methods are described below.
 Optical Metallography: Full metallographic analysis was performed on transverse
cross sections (perpendicular to the billet axis) along their diameter according to the
marking shown in Figure. 2.3 Specimen preparation included cutting with Struers
‘‘Accutom 2’’, mounting in resin, The samples were ground on 500 grit SiC water –
cooled paper, polish with 3μm and 1μm Diamond paste and finally polished with
0.05μm Colloidal Silica. The microstructure was revealed after etching with the
appropriate etchants for each alloy listed in Table 2.2. Examination of the
metallographic specimens was carried out on an Optical Metallographic Microscope,
Leitz ‘‘Aristomet’’ at magnifications 100 x –1000 x.
 Scanning Electron Microscopy/EDX analysis: Further metallographic investigation
was carried out on a JEOL SEM at several magnifications. Elemental area mapping
and line profiling were also performed for the major alloying elements.
 X-Ray Diffraction: X-Ray Diffraction analysis was carried out in order compare the
evolution of intermetallic phases after homogenization. The analysis was performed
by utilizing a Siemens D-5000 diffractometer operating with a Cu-Ka radiation. The
time for measurement was 0.008o/sec and t=4020sec.
73

26/03/2018 03:27:43 EEST - 137.108.70.7
3.4.1. Dual Grain Model
The results of the homogenization Dual Grain Model (DGM) are presented in this
section for the 6082 alloy. The evolution of the volume fraction of phases with time is
depicted in Figure 3.4a. Figure 3.4a depicts the isothermal holding at 540ο C up to 32h
(115200 sec). The Mg2Si dissolves quickly at the early stages of homogenization (below 500
sec). After an induction period of about 700 sec, the transformation β-AlFeSi→α-AlFeSi
starts. This is indicated by the reduction of the volume fraction of the β-AlFeSi phase and
the corresponding increase of the α-AlFeSi phase. The β-AlFeSi phase dissolves at a time
about 6000 sec (100 min), where the α-AlFeSi obtains its highest volume fraction and then
the volume fraction remains constant in the remaining holding time. In addition to the
holding period of 32 h, Figure 3.4b describes the cooling process with a cooling time of 2h
and the preheating prior to extrusion to 500οC which is obtained in 1h. Mg2Si re-
precipitation takes place during cooling from the homogenization temperature and the
volume fraction of Mg2Si is now much higher than the Mg2Si present in the as-cast
microstructure (at t=0). The respective values are 0.65% and 0.1% respectively. This is
attributed to the additional amount of Si coming from the dissolution of the Si-diamond
phase and the β-AlFeSi→α-AlFeSi transformation during homogenization, since the α-
AlFeSi phase is leaner in Si. Thus there is more Mg and Si available in the matrix to form
Mg2Si during cooling from the homogenization temperature. This Mg2Si dissolves again
during pre-heating prior to extrusion. Another interesting effect is depicted in Figure 3.4b.
The reverse transformation α-AlFeSi→β-AlFeSi seems to take place during cooling
and the direct transformation β-AlFeSi→α-AlFeSi takes place during billet preheating. This
is in line with phase stabilities since β-AlFeSi is stable at low temperatures. However in the
cooling rates encountered in industrial practice, the reverse transformation α-AlFeSi→β-
AlFeSi does not take place. Therefore after cooling from the homogenization temperature,
the iron intermetallic present in the microstructure is the α-AlFeSi phase, provided that the
direct transformation β-AlFeSi→α-AlFeSi during homogenization has been completed.
The spatial evolution of the Mg2Si dissolution and the β-AlFeSi→α-AlFeSi transformation
are presented below.
74

26/03/2018 03:27:43 EEST - 137.108.70.7
(a)
(b)
Figure 3.4 Temporal evolution of phase transformations during the homogenization
process a) Isothermal holding at 540οC for 32 h and b) holding at 540ο C for
32h cooling down (500οC/2h), preheating 500οC/1h.
75

26/03/2018 03:27:43 EEST - 137.108.70.7
3.4.2. Mg2Si dissolution
While the temporal evolution of the integrated volume fractions was depicted in
Fig.3.4, with the DGM model is possible to monitor the spatial evolution of phase fractions
at specific times during homogenization. The dissolution of the Mg2Si phase from the grain
boundaries and its re-precipitation in the grain interiors during cooling from the
homogenization temperature has been investigated in this way. The reduction of the mole
fraction of Mg2Si during homogenization at 540oC is shown in Figure.3.5. The t=0 curve is
the Mg2Si profile at the beginning of homogenization (as-cast condition, Scheil profile). As
seen in Figure.3.5, the Mg2Si phase dissolves rapidly and the complete dissolution of Mg2Si
takes only about 6 min at 580oC for 6082 alloy.
Figure 3.5: The spatial evolution of the Mg2Si phase from the grain boundaries and its re-
precipitation in the grain interiors
The dissolution of the Mg2Si is faster in the small grain (20μm) due to the shorter
diffusion distances. In addition the Mg2Si profile spreads away from the boundary in the
large grain. This is due to Mg and Si diffusion towards the grain interior causing
precipitation of Mg2Si in new locations. Then this Mg2Si also dissolves and in this way the
Mg and Si diffuse to the grain interior. Following the complete dissolution of Mg2Si during
isothermal holding at 580oC, the Mg2Si phase is re-precipitated during cooling from the
76

26/03/2018 03:27:43 EEST - 137.108.70.7
homogenization temperature in the grain interior as shown with the straight line in Figure
3.5. The Mg2Si fraction after cooling is uniform across both grains.
3.4.3. β-AlFeSi→α-AlFeSi transformation
The spatial evolution of the β-AlFeSi→α-AlFeSi transformation is presented in this

section. First in Figure.3.6a the initial profiles of β-AlFeSi and α-AlFeSi in the as-cast
structure are depicted. This is to compare with the final profiles after the completion of the
transformation in Figure 3.6b, where only the α-AlFeSi phase remains with a higher
fraction, due to the amount gained by the transformation.
(a)
(b)
Figure 3.6: The spatial evolution of the β-AlFeSi→α-AlFeSi transformation a) the

initial profiles of β-AlFeSi and α-AlFeSi in the as-cast structure, b) the
final profiles after the completion of the transformation
The intermediate stages of the transformation are depicted in Fig.3.7 (a-d) for times
15, 45min, 1h and 1 ½ hours. The following observations can be made:
77

26/03/2018 03:27:43 EEST - 137.108.70.7
 The profiles of α-AlFeSi and β-AlFeSi exhibit an exact spatial correspondence. Where
there is a drop in the fraction of β-AlFeSi, there is a rise in the fraction of α-AlFeSi at
the same location.
 The β-AlFeSi→α-AlFeSi starts at a certain distance from the boundary and the
transformation front moves towards the boundary. As a result the profile of α-AlFeSi
spreads towards the grain interior
 The phase fraction profiles are steeper and the transformation is faster in the
smaller grain.
The above results indicate that the β-AlFeSi→α-AlFeSi transformation is a transformation
which takes place at a region adjacent to the grain boundary by diffusion through the
matrix. The transformation takes place at a faster rate in the smaller grains.
(a)
(b)
78

26/03/2018 03:27:43 EEST - 137.108.70.7
(c)
(d)
Figure 3.7: Spatial evolution of β→α-AlFeSi transformation during homogenization for

times a) 15 b) 45min c) 1h d)1 ½ h
3.4.4. The effect of temperature and grain size
The effect of homogenization temperature and grain size on the dissolution of the
Mg2Si and the transformation of β→α-AlFeSi has been studied. As stated above, grain size
influences the transformation kinetics since it controls the distance over which diffusion of
alloying elements takes place. Homogenization temperature influences the diffusion rate.
For the 6082 alloy the weight fraction of Mg2Si phase is depicted in Figure 3.8a for 3
different homogenization temperatures, 540, 560 and 580oC for homogenization time
t=2sec. It is obvious that the Mg2Si dissolution is much faster at the high homogenization
temperature. Similarly the weight fraction of β-AlFeSi phase is depicted in Figure 3.8b for
the same temperatures and for homogenization time t=500sec. It is also evident that the
transformation rate increases with the homogenization temperature.
79

26/03/2018 03:27:43 EEST - 137.108.70.7
(a)
b)
Figure 3.8: The weight fraction of a)Mg2Si phase and b)β-AlFeSi phase for 3
different homogenization temperatures, 540, 560 and 580oC for
homogenization time t=2sec and 500sec respectively
Furthermore the dissolution of Mg2Si and β-AlFeSi has a different transformation rate
depending from the grain size. At 580οC and t=2sec the Mg2Si dissolution has been
completed only at the 20μm grain. The same stands the β-AlFeSi phase at t=500sec.
80

26/03/2018 03:27:43 EEST - 137.108.70.7
3.4.5. Removal of the microsegregation of the alloying elements
As stated in the introduction of this chapter, one of the aims of homogenization is the
removal of microsegregation, which is present in the as-cast microstructure.
The evolution of the concentration profiles of Mg, Si, Fe and Mn in the FCC phase during
homogenization at 540oC as predicted by the DGM model is given in Figure 3.9 a-d The
homogenization times considered are 0.5, 4, 8 and 32 hours. The t=0 curve corresponds to
the as-cast microsegregation profile (Scheil profile). The following remarks can be made:
 The concentration profiles for Mg, Si and Mn become more uniform with
homogenization time. The profile for Fe does not change appreciably due to the very
low diffusivity of Fe in Al.
 The average concentration of Mg and Si in the matrix interior (away from the
boundary) increase with homogenization time. This explains the larger amount of
Mg2Si precipitated during cooling discussed in the previous section.
 The profiles homogenize faster in the smaller grain
 The compositional fluctuations close to the boundary are due to the dissolution of
Mg2Si and the β-AlFeSi→α-AlFeSi transformation. These fluctuations decay with
homogenization time.
 The spatial evolution of the concentration profiles is consistent with the spatial
evolution of the phase fractions.
81

26/03/2018 03:27:43 EEST - 137.108.70.7
a) b)
c) d)
Figure 3.9: The evolution of the concentration profiles of Mg, Si, Fe and Mn in the FCC
phase during homogenization at 540oC
3.4.6. Homogenization Process Map of extrudable 6xxx alloys
The main parameters of the homogenization process are the homogenization

temperature and time. The selection of these parameters in industrial practice has been a
matter of thorough investigation in the past. In line with the phase fraction mapping,
presented in Chapter 2, a preliminary work is presented here as an initial effort to develop
homogenization process map, which will aid the selection of process parameters. The map
presented here regards the time for the β-AlFeSi→α-AlFeSi transformation. In order to
develop the maps, the DGM model was employed, for a series of simulations. The results are
presented as a function of the Mg:Si ratio in the alloy. The homogenization map based on
82

26/03/2018 03:27:43 EEST - 137.108.70.7
the β-AlFeSi→α-AlFeSi transformation is shown in Figure 3.10. Three homogenization
temperatures are considered, 540, 560 and 580oC. Except from the reduction of the β-
AlFeSi→α-AlFeSi transformation time with increasing the homogenization temperature,
another interesting effect is indicated. It appears that the β-AlFeSi→α-AlFeSi
transformation time diminishes with the increase of the Mg:Si ratio in the alloy. This is
consistent with the fact that Mg stabilizes the α-AlFeSi phase and Si stabilizes the β-AlFeSi
phase.
Figure 3.10: Homogenization process map of 6082 alloy based on the β-AlFeSi→α-AlFeSi
transformation
3.4.7. Experimental validation
 Homogenization 560oC AA6082 and AA6060
The purpose of this section is to provide the necessary experimental data, regarding
the validation of the phase transformations take place during homogenization e.g Mg2Si
dissolution, β→α-AlFeSi transformation and alloying element elimination.
As mention in previous chapter the validation was performed on alloys 6082 and on 2 6060
alloys. Homogenization heat treatments at 560οC for 30min, 4h, 8h and 32h have been
applied. The microstructure was revealed using the procedure outlined above i.e. light
83

26/03/2018 03:27:43 EEST - 137.108.70.7
optical microscopy, SEM and EDS analysis and elemental area mapping. X-Ray Diffraction
analysis was carried out in order to validate the evolution of intermetallic phases during
homogenization treatment as predicted by the DGM model.
Regarding the evolution of the phases present in the microstructure of the 6082
aluminum alloy the effect of the homogenization time is depicted in Figure 3.11 for 30min 2,
4 and 8h. The Mg2Si strengthening phase has dissolved completely and reprecipitated in the
grain interior during cooling from homogenization temperature.
84

26/03/2018 03:27:43 EEST - 137.108.70.7
6082 air cooled
30 min 4 hours 8 hours 32 hours
540
560
580
Figure3.11: The effect of the homogenization time of the 6082 aluminum alloy for 30min 2,4 and 8h
.
85

26/03/2018 03:27:43 EEST - 137.108.70.7
SEM micrographs of 6082 alloy are shown in Figure 3.12 a-e. The sharp intermetallic
phases of the as-cast microstructure become more rounded after homogenization. The
reprecipitation of Mg2Si in the grain interiors is evident in the micrographs.
(a)
(b) (c)
(d) (f)
Figure 3.12: SEM micrographs of 6082 alloy a) as- cast condition b) 0.5, c) 4h, d)8h, e)32h
86

26/03/2018 03:27:43 EEST - 137.108.70.7
The homogenized microstructure of the 6060 aluminum alloy is depicted in Figure
3.13. After homogenization the transformation of the β-AlFeSi→α-AlFeSi and dissolution of
the Mg2Si phase has been completed.
Figure 3.13: The homogenized microstructure of the 6060 aluminum
Furthermore more rounded α-AlFeSi particles are evident as well as a neckless

morphology depicted clearly in SEM micrograph in Figure 3.14a-b. The round-off or
sperioidization as well as the development of a neckless morphology of the α-AlFeSi phase
are very important additional stages during homogenization and will be discussed
extensively in Chapter 4.
Figure 3.14: a)Rounded α-AlFeSi particles and b) neckless morphology in SEM

micrographs
87

26/03/2018 03:27:43 EEST - 137.108.70.7
3.4.8. Transformation β-AlFeSi → α-AlFeSi
The β-AlFeSi→α-AlFeSi transformation was monitored experimentally with XRD. For

the 6082 alloy an XRD spectrum was taken after 0.5, 4, 8 and 32h homogenization at 560 oC
and cooling to room temperature. The as-cast condition was also analyzed with XRD. The
results appear in Figure 3.15. Diffraction peaks corresponding to the aluminum matrix,
Mg2Si, α-AlFeSi and β-AlFesi are evident. The β-AlFeSi peaks are replaced by α-AlFeSi peaks
in the 0.5h spectrum. This does not mean that the transformation is completed in 0.5 hours.
It is more probable that at 0.5 hours the fraction of the β-AlFeSi phase is small and beyond
the detection limit of the XRD.
Figure 3.15: XRD spectrum of the β-AlFeSi→α-AlFeSi transformation of 6082 alloy
The XRD spectrum for the 6060 alloy homogenized at 560oC for 0.5h is depicted in
Figure 3.16. The β-AlFeSi peaks have been replaced with α-AlFeSi peaks, indicating that the
transformation has been completed. The above results are in general agreement with the
DGM simulation results.
88

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 3.16: The XRD spectrum for the 6060 alloy homogenized at 560oC for 0.5h
3.4.9. Evolution of elemental profiles during homogenization
In order to experimentally validate the DGM predictions regarding the removal of

microsegregation (presented in 3.4.5 above), area analysis vs homogenization time and
elemental mapping were performed with EDS analysis. The area analysis was performed in
the matrix close to the boundary. The results are presented for each element in Figures
3.17-20.
 Profile and mapping for Fe
Due to the low diffusivity of Fe in Al, the Fe concentration does not change with
homogenization time. The simulation results are in agreement with the experimental data
as shown in Figure 3.17 a. The mapping depicted in Figure 3.17b-f indicates that Fe is
mostly concentrated in the α-AlFeSi phase.
89

26/03/2018 03:27:43 EEST - 137.108.70.7
(a)
(b)
(c) (d) (e) (f)

Figure 3.17: a) Simulation results and experimental for Fe element, b)profile mapping in
as – cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time
3.4.10. Profile and mapping for Mg
As indicated in Figure.3.18, the agreement between the simulation result and the
experimental data regarding the evolution of the Mg concentration with homogenization
time is good. The mapping for Mg indicates that Mg is absent from the boundary regions
90

26/03/2018 03:27:43 EEST - 137.108.70.7
where the iron intermetallics form. The spread of Mg in the matrix with homogenization
time is evident and is due to the diffusion of Mg to the grain interior.
(a)
(b)
(c) (d) (e) (f)

Figure 3.18: a) Simulation results and experimental for Mg element, b)profile mapping in
as – cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time
3.4.11. Profile and mapping for Si
As shown in Figure 3.19, the evolution of Si concentration with homogenization time

follows a similar trend with Mg. Although the DGM model shows a similar trend with the
91

26/03/2018 03:27:43 EEST - 137.108.70.7
experimental values, i.e. a reduction with homogenization time, the DGM underestimates
the Si at higher homogenization times. This discrepancy could be attributed to the Si
coming from the dissolution of the Si-diamond phase, which is present in the as-cast
microstructure.
(a)
(b)
(c) (d) (e) (f)

Figure 3.19: a) Simulation results and experimental for Si element, b)profile mapping in
as–cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time
3.4.12. Profile and mapping for Mn
As shown in Figure 3.20, the DGM model overestimates the Mn concentration in the
matrix especially at the short homogenization times, the difference decreases at longer
92

26/03/2018 03:27:43 EEST - 137.108.70.7
times. As indicated from the mapping, there is a strong Mn signal in the α-AlFeSi phase.
Therefore this discrepancy is due to the Mn retained in the α-AlFeSi phase and to the fact
that the thermodynamic description of the α-Al(Fe,Mn)Si phase is not yet optimized.
(a)
(b)
(c) (d) (e) (f)

Figure 3.20: a) Simulation results and experimental for Mn element, b)profile mapping in as –
cast condition, c)0.5, d) 4h, e) 8h, f) 32h homogenization time
93

26/03/2018 03:27:43 EEST - 137.108.70.7
3.5. Conclusions
From the work presented above the following conclusions can be drawn:
The DGM model has been developed to treat the homogenization process in
multicomponent and multiphase Al-alloys exhibiting a large variability of the as-cast grain
size. With the model it is possible to simulate the temporal and spatial evolution of phase
fractions and element concentration during homogenization.
Regarding the evolution of phase fractions during homogenization, the predictions of
the DGM model have been validated experimentally with XRD analysis. The evolution of the
β-AlFeSi→α-AlFeSi transformation is predicted by the DGM and is confirmed by XRD in the
same time scale.
The DGM predictions regarding the evolution of Fe and Mg concentrations with
homogenization time are in excellent agreement. There are only some discrepancies in the
profiles of Si and Mn. The Si discrepancy is due to the Si-diamond phase not taken into
account in the DGM. The Mn discrepancy is due to the Mn retention in the α-AlFeSi phase.
The DGM predicts the fast dissolution of Mg2Si during homogenization and its re-
precipitation during cooling. It also predicts its dissolution during preheating prior to
extrusion.
The DGM can describe the temporal and spatial evolution of the β- -AlFeSi
transformation. The spatial evolution exhibits an exact spatial correspondence. In addition
the starts at a certain distance from the boundary and the transformation front moves
towards the boundary.
Both the Mg2Si dissolution and the β-AlFeSi→α-AlFeSi transformation are faster in the
smaller grain as predicted by the DGM. In addition the concentration profiles of the
elements homogenize faster in the smaller grain.
The DGM predicts correctly the effect of homogenization temperature, with the rate of
both the Mg2Si dissolution and the β-AlFeSi→α-AlFeSi transformation to increase with the
homogenization temperature.
A preliminary attempt to develop homogenization process maps has been performed
using the DGM. These maps are, at present, based on Mg2Si dissolution and the β-AlFeSi→α-
AlFeSi transformation and can be used for the design of the homogenization heat treatment.
94

26/03/2018 03:27:43 EEST - 137.108.70.7
3.6. References
[1] B. Rinderer Mat Sc F. (2011) 264-275

[2] A. L. Dons, Journal of Light Metals 1 (2001) 133-149.
[3] G.Mrόwka-Nowotnik, Mech. and Mater. Eng. (2005)
[4] Li YJ, Arnberg L. Acta Mater (2003) 51-3415.
[5] Li YJ, Arnberg L. Mater Sci Eng A (2003), 347-130.
[6] Q. Du W.J. Poole, M.A.Wells, N.C. Parson, Acta Materialia 61 (2013) 4961-4973.
[7] Yan, L.-Z., Zhang, Y.-A., Li, X.-W., Li, Z.-H., Wang, F., Liu, H.-W., Xiong, B.-Q, Trans., Nonf.
Met. Soc. Of Ch. 24 (2014) 939-945.
[8] Cieslar, M., Bajer, J., Hájek, M., Očenášek, V. TMS Light Metals (2014), 237-241.
[9] P.K. Saha P.K., ASM International, (2000).
[10] N. C. W. Kuijpers, Ph.D.Thesis, TUDelft (2004)
[11] Sheppard, Kluwer Academic Publishers, (1999).
[12] S. Zajac,B. Bengtsson, C. Jonsson, Mat. Sc. Forum, Trans Tech, Switzerland, (2002)
[13] Kemal Deliji, Vanja Asanovi, Dragan Radonji,MTAEC9, 39 (4) (2005)101
[14] Polmear I.J., ‘Light Alloys, Metallurgy of the Light Metals’, Arnold, (1995).
[15] J. V. Langkruis, Ph.D. Thesis, TUDelft, (2000).
[16] O.Reiso, Ph.D. Thesis, Norwegian Institute of Technology, (1992).
[17] Halfdan Kristoffer Småbråten,Master thesis, Norwegian University of Science and
Technology, (2011)
[18] G. M.-Nowotnik, J. Sieniawski, Jour. Mat. Proc. Tech., (2005) 367-372,
[19] S.N Samaras,. G.N Haidemenopoulos, Jour., Mat., Proc., Techn. 194 (2007), 63–73
[20] Y. Birol, Jour., Mat., Proc. Tech. 148, (2004). 250-258
[21] Y. Birol, Met., Mat., Int. 20 (2014), 727-732.
[22] S.N. Samaras, PhD thesis, University of Thessaly Mech. Eng.(2006)
[23] N.C.W Kuijpers, Mat., Sc., Eng. A (2005), 394, 9–19.
[24] G.N Haidemenopoulos,. H Kamoutsi, A.D Zervaki, Jour., Mat., Proc., Techn., 212, (2012),
2255-2260.
[25] A. Borgenstam, L. Hoglund, J. Εgren, A. Engstrom, Jour. Ph. Equil. 21, (2000) 269-280
[26] Kinetics of materials, Book
[27] G. Inden, V. Ghetta (ed), Spr. (2008), 113-139.
[28] A. Engstrom, L. Hoglund, J. Agren, Met., Mat., Trans. 25A (1994) 1127-1134.
95

26/03/2018 03:27:43 EEST - 137.108.70.7
96

26/03/2018 03:27:43 EEST - 137.108.70.7
4. Quantification of Homogenization
state
4.1. Introduction
As discussed in Chapter 3 the aim of homogenization is to reduce the

microsegregation of elements and phases that are concentrated at grain boundaries and
interdendritic areas. In 6xxx alloys the undesirable β-AlFesi intermetallic phase, which
limits extrudability, transforms to a more rounded α-AlFeMnSi intermetallic phase during
homogenization. After the transformation β→α-AlFeSi is complete, the α phase undergoes
coarsening and spheroidization, adopting a “necklace” morphology, which enhances the
extrudability of the billet. This explains the fact that the actual homogenization times in
industrial practice are longer than the times required for Mg2Si dissolution and the
completion of the β→α-AlFeSi transformation. The morphological changes of the α-AlFeSi
phase have only be described qualitativelly in the published literature and a
“quantification” of the homogenization state does not exist. In this chapter the
homogenization state has been quantified by means of quantitative metallography in order
to describe the morphology of the intermetallic phases. An index-based methodology has
been developed. The aspect ratio and circularity are two indices, among others, that can be
used to characterize the homogenization state. These parameters can be determined by
quantitative metallography, involving image analysis. A fully homogenized billet, with the
potential for high extrudability should have all β-AlFeSi transformed to α with necklace
morphology and appropriate values of aspect ratio and circularity.
4.2. Literature review
As discussed in previous chapters the presence of intermetallic phases with sharp

edges can impair the deformability of 6xxx extrudable alloys especially when located in the
grain boundary regions [1-4]. Extrudability is qualitatively defined by the maximum
production speed attainable for a given press capacity while still obtaining the desired
mechanical properties, surface quality and geometric tolerances of the extrudate. As
97

26/03/2018 03:27:43 EEST - 137.108.70.7
described earlier, the extrudability increases due to the β to α transformation. The sharp
tips of the β particles initiate micro-cracks during the deformation and, therefore, lead to
surface cracking of the extrudate. The transformed α-particles are more rounded and
therefore cause less local cracking at the surface.
Although there have been a number of investigations on the homogenization
treatment of 6xxx series aluminum alloys in recent years, most of the research has been
focused on the evolution of microstructure as discussed extensively in Chapter 3 (see
literature review Ch3). Moreover there has been no comprehensive quantitative study on
the definition of homogenization state where the dependence of intermetallic phases and
their morphology during the homogenization is investigated. The correlation of
extrudability and intermetallic phase morphology has been investigated only qualitatively.
Studies have been made on the microstructural evolution during the homogenization of
AA7020 aluminum alloy concerning the dissolution of detrimental particles, which degrade
the hot workability of the AA7020 alloy e.g phases and grain boundary (GB) particles
during homogenization treatment [5-7]. It was found that the dominant process is the
spheroidization of the GB particles; together with the decrease in the width of the GB
particles, spheroidization was considered a key mechanism in the microstructural evolution
during homogenization [1, 8, 9]. In the present thesis the state of homogenization has been
quantified by determining specific indices by means of quantitative metallography. The
study was performed for the 6060_3 variant of a 6060 alloy. The methodology and results
are described below.
4.3. Methodology
The chemical composition of the 6060_3 alloy investigated is shown in Table 2.2
(Chapter 2). Three homogenization heat treatments, in laboratory scale, were investigated,
following the simulation conditions from Chapter 3, holding at 540ο, 560ο, and 580oC for 4h
followed by air cooling (Table 4.1). These conditions ensured the complete transformation
of β→α-AlFeSi.
98

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 4.1 Homogenization heat treatments
Homogenization Homogenization
Alloy Condition Temperature Time
( C)
ο (h)
6060_3 As-Cast - -
6060_3A Homogenized 540 4
6060_3B Homogenized 560 4
6060_3C Homogenized 580 4
4.3.1. Quantitative metallography
After the homogenization heat treatment, the specimens were prepared for standard
metallographic examination involving optical microscopy, SEM and image analysis as
discussed in previous Chapters, in order to reveal the microstructure. The as-cast as well as
the homogenized microstructures were characterized for intermetallic phases and the
particles were categorized in three morphological types as rounded particles, pinched
particles and particles exhibiting a neckless formation. It should be noted that pinched
particles are those that are in the initial stage of separation in smaller rounded particles
towards the formation of a neckless group. The number of images processed and the
number of particles measured for each alloy appears in Table 4.2.
Table 4.2: Number of images analyzed and number of particles measured per billet for the
quantification of the homogenization state
Number Number
Alloy
of Images of particles
6060_3 58 106
6060_3A 57 161
6060_3B 56 150
6060_3C 58 133
The quantification of the homogenization state was based on the measurement of

indices that describe the size and shape of the intermetallics and indicate the state of
homogenization. The indices employed were the following and are shown in Table 4.3:
99

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 4.3: The indices employed for the quantification of the homogenization state
A ratio of the major to the
minor diameter of a particle, 𝑑𝑚𝑎𝑥
Aspect ratio which can be used when the 𝐴𝑅 =
𝑑𝑚𝑖𝑛
major axis does not cross a
particle outline.
Ferret The longest caliper length
Circularity is a measure of
how closely a particle
resembles a circle. It varies 𝑝2
Circularity
from zero to one in magnitude 𝐶=
4𝜋𝐴
with a perfect circle having a
value of one
Same as circularity but more

𝑝2
Circularity of the sensitive on the edges when 𝐶=
edges the length of a particle does 4𝜋𝐴
not change
4.3.2. Image analysis
After the standard metallographic observation, a higher magnification and resolution

of the particles, through the use of SEM analysis, have been applied and shown in Figure
4.1a for a group of particles. It is important to use the appropriate magnification with a
suitable numerical aperture [10] Fig4.1b. The Image J software display is depicted in
Fig4.1c while Table 4.4 shows the respective measurements for each particle in the group.
100

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
2
1
3
5
4
6
(c)
Figure 4.1: SEM image used for the measurement of indices: (a) Low magnification image,
(b) high magnification isolation of the group of particles, (c) image J display
used for the measurement of the indices
101

26/03/2018 03:27:43 EEST - 137.108.70.7
Table 4.4: Respective measurements for each particle referring to Figure 4.1
Area
No Diam(max) Diam(min) Circularity Perimeter Feret(max)
(polygon)
1 3.289 0.574 5.200 7.427 0.844 3.308
2 3.655 0.376 5.190 8.405 1.083 3.656
3 1.792 0.389 2.103 4.395 0.735 1.793
4 1.123 0.632 1.439 3.080 0.524 1.160
5 1.069 0.534 1.433 2.820 0.441 1.068
6 0.976 0.489 1.384 2.712 0.423 1.000
7 2.369 0.526 2.764 6.027 1.045 2.482
4.3.3. Statistical Analysis
In order to have credible results that could allow comparison between

homogenization treatments, a large number of particles were measured. The area of
measurement (scanned area) was kept constant for all homogenization treatments. In order
to get credible results the data were processed statistically. In order to decide if the samples
were comparable, the Kruskal-Wallis test was used. The Kruskal-Wallis test [11] is a non-
parametric method for testing whether samples originate from the same distribution and it
is used to compare two or more independent samples of equal or different sample size.
With a confidence level of 99% it was proved that the samples derive from different
distributions. As a result the samples are not comparable without further processing. In
order to compare between the dissimilar samples, the “Bootstrapped Mean” [12]method
was used. Bootstrapping is a non-parametric statistical technique that allows accurate
estimations about the characteristics of a population to be made, when the examined
sample size is limited. As it is non-parametric, the method can be used to compare between
samples derived from different distributions, such as Normal and LogNormal distributions.
It works by recursively calculating the preferred parameter, like the mean or the median,
for a part of the sample and then combining the results to make robust estimates of
standard errors and confidence intervals of the population parameter. In this case a 95%
confidence interval was used, while the standard error was kept to a minimum by using a
102

26/03/2018 03:27:43 EEST - 137.108.70.7
large number of iterations. This process leads to comparable statistical parameters for each
measurement.
The results of the quantification of homogenization heat treatment are presented in

this section. Regarding the homogenization state of the alloys studied the effect of
temperature is depicted in Fig.4.2a-d in the as cast condition, homogenized at 540ο, 560ο
and 580οC respectively.
4.4.1. Metallographic Investigation
The as-cast microstructure depicted in Fig. 4.2a and consists of characteristic “Chinese
– script” morphology of intermetallic α-AlFeSi phase. After homogenization treatment at
540οC and 560οC this morphology still remains in the microstructure. In the contrary, in the
580οC temperature the α-phase seems to have reached a state of satisfactory
spheroidization.
103

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
(c) (d)
Figure 4.2: Metallographic images, (a) as-cast, (b) homogenized at 540oC, (c) 560oC and (d)
580oC
4.4.2. Characterization of the particles
Three stages of morphological changes during homogenization

 1st stage/ Rounded particles
As discussed in the previous section following the β to α-AlFeSi transformation the

intermetallic α-phase undergoes spheroidization. In the first stage of this process the plate–
like particles exhibit a slight decrease in their width. Although they do not exhibit complete
104

26/03/2018 03:27:43 EEST - 137.108.70.7
spheroidization the particles become more rounded at the edges Figure 4.3. It was found
that this stage is more frequent for temperatures 540ο and 560ο C.
Figure 4.3: Images indicating the rounding of the edges of the particles during
homogenization
 2nd stage/ Pinched Particles
At the second stage an initial perturbation can be observed in the particles. Particles
take an ellipse shape with more rounded edges. This stage characterized by the trend of the
particles to be separated into smaller rounded particles. These discontinuities are depicted
in Figure.4.4. The driving force for this behavior is reduction of strain energy associated
with large particles.
Figure 4.4: Imaged revealing the pinching process during homogenization
105

26/03/2018 03:27:43 EEST - 137.108.70.7
3rd Stage/ Neckless Particles
The reduction of surface energy of the α-phase is the driving force for spheroidization.
With this process, the total interface area between the matrix and the α-AlFeSi phase is
reduced. The particles finally adopt a spherical shape at the third stage as depicted in
Figure. 4.5
Figure 4.5: Images revealing the spherical shape at the third stage
It was found that for the two homogenization temperatures 540ο and560ο C, although
the morphological changes towards spheroidization were activated; only the first stage was
completed. On the other hand for the homogenization at 580οC the α-particles reached the
third stage of spheroidization involving the neckless formation.
4.5. Results and discussion
The mean values of microstructural indices, Aspect ratio, Feret, Circularity and
Circularity of Edges have been measured for all conditions investigated. The mean value
reported possesses a 95% confidence while the upper and lower points are the 97.5% and
2.5% cut-offs to exclude extreme values. From these data, which are shown in Figure 4.6 a-d
some general remarks can be made.
Regarding the Aspect Ratio (Figure 4.6a) the heat treatments of 6060_3B and 6060_3C
alloys (560ο and 580ο C respectively) have a lower value compared to 6060_3A alloy
(540οC). However 3C alloy, has a lower Feret value (Figure 4.6b) and almost equal with 3A
106

26/03/2018 03:27:43 EEST - 137.108.70.7
alloy. This means that although the size of the particles is small for 3A and 3C alloys, the 3C
alloy had more spheroidized particles.
The effect of homogenization temperature on the circularity of the particles is
depicted in Fig 4.6c. Again 3C alloy has the lowest value. This is in a good agreement with
the fact that 3C alloy reached the 3rd stage of spheroidization while alloy 3B reached the
2nd stage. Alloy 3A reached only the 1st stage, which explains the low value of circularity of
edges (Figure 4.6d) as in that stage, particles exhibited an initial spheroidization at their
edges. From the above analysis it is concluded that the treatment possessing
homogenization indices for better extrudability is 580οC for 6060_3C.
(a) (b)
(c) (d)
Figure 4.6: The values of indices for the as-cast and homogenized alloys: (a) aspect ratio, (b)
feret, (c) circularity and (d) circularity of edges
107

26/03/2018 03:27:43 EEST - 137.108.70.7
4.6. Conclusions
An index-based method to quantify the homogenization state has been developed.

Indices such as the aspect ratio, feret, circularity and circularity of edges have been
determined in order to characterize the stage of spheroidization of the α-AlFeSi phase,
following the β to α-AlFeSi transformation. The effect of the homogenization temperature
was studied in a 6060 alloy. The major conclusions are the following:
 The α-AlFeSi particles, after the completion of the β to α-AlFeSi transformation undergo
morphological changes leading to spheroidization. This process can be divided in three
stages: (1) rounding of edges, (2) pinching and (3) neckless formation.
 The evolution of the morphological changes can be described quantitatively by the use of
indices, which are sensitive to homogenization process parameters, such as the
homogenization temperature or time.
 Through the use of these indices, it was found that the 580oC homogenization temperature
leads to a complete spheroidization of the α-AlFeSi phase, involving a neckless
formation. The lower homogenization treatments resulted in intermediate stages involving
rounding of edges or particle pinching.
 The method requires the measurement of a large number of particles and the
implementation of a statistical analysis in order to be credible.
4.7. References
[1] Sheppard, Kluwer Academic Publishers, (1999).

[2] F.Y. Xie, T. Kraft, Y. Zuo, C.H. Moon, Y.A. Chang, Acta Mater., 47 (1999) 489-500.
[3] J.S. Robinson, Mater. Sci. Technol., 19 (2003) 1697-1701.
[4] L.L. Rokhlin, T.V. Dobatkina, N.R. Bochvar, E.V. Lysova, J. Al., Compd., 367 (2004) 10-16.
[5] A.R. Eivani, H. Ahmed, J. Zhou, J. Duszczyk, Met., Mat., Trans., A, 40 (2009) 717-728.
[6] A.R. Eivani, H. Ahmed, J. Zhou, J. Duszczyk, Met., Mat., Trans., A, 40 (2009) 2435-2446.
[7] A.R. Eivani, H. Ahmed, J. Zhou, J. Duszczyk, Cr and Mn”, Mat., Sc., Tech., (2010)
[8] J.D. Robson, P.B. Prangnell, Acta Mater., 49 (2001) 599-613.
[9] X. Fan, D. Jiang, Q. Meng, L. Zhong, Mater. Lett., 60 (2006) 1475-1479.
[10]ASTM Specification, F1877 (1998)
[11]D. Wayne App. Nonpar. Stat. (2nd edition) (1990), 226-234
[12] B. Efron Boca Raton, FL: Chapman & Hall/CRC (1993)
108

26/03/2018 03:27:43 EEST - 137.108.70.7
5. Precipitation during Cooling from the
Homogenization Temperature
5.1. Introduction
An important step in the process chain of extrudable Al-alloys is the cooling from the
homogenization temperature. As discussed in the previous chapters of this thesis,
homogenization involves holding the material at a high temperature for a certain time
period (holding time). During this time several transformations of the as-cast
microstructure take place, including the dissolution of Mg2Si, the transformation of β-AlFeSi
to α-AlFeSi, the spherodization of the α-AlFeSi and the reduction of elemental
microsegregation. During cooling from the homogenization temperature, new
transformations are activated, the most important being the re-precipitation of Mg2Si. The
assumption is made here that although the β-AlFeSi is the thermodynamically stable phase
at low temperatures, the transformation of α-AlFeSi to β-AlFesi does not take place at any
appreciable rate. This is true, especially for the cooling rates encountered in the current
industrial homogenization practices and is due to the very low diffusivity of Fe in the Al
matrix. A DICTRA calculation in Chapter 3 indicated that there is actually no transformation
of α to β-AlFeSi during cooling of the cast billet. Therefore, the only transformation to be
studied is the precipitation of the strengthening Mg2Si phase. This chapter deals with three
issues:
1. Precipitation of Mg2Si and hardening of the homogenized billet

2. Hardening potential of homogenized billets
3. Effect of excess-Si on the Continuous Cooling Precipitation (CCP) diagrams
Regarding the first issue, a simulation is developed to describe precipitation of Mg2Si

during continuous cooling from the homogenization temperature. The KWN model,
previously employed for isothermal aging of aluminum alloys [1], will be used in the
present case, under non-isothermal conditions (continuous cooling). The results of the
KWN model are the fraction and size of precipitates. A suitable strength model is then
109

26/03/2018 03:27:43 EEST - 137.108.70.7
employed to determine the strength (hardness) of the material at the end of the cooling
period. A new Matlab code has been developed [2] to handle the calculations under both
isothermal and non-isothermal conditions encountered in welding and the modification to
the non-isothermal conditions prevailing during cooling from the homogenization
temperature has been performed [3] at LoM. The results of the combined KWN and
strength models are experimentally validated.
Regarding the second issue, the hardening potential of the homogenized billets is
investigated experimentally for three alloys 6063, 6005 and 6082 as a function of
homogenization temperature and cooling rate. The aim here is to identify possible effects of
homogenization parameters on the ability of the material to age harden (hardening
potential) during subsequent aging treatments.
The third issue is related to the extrudability of the homogenized billet. It is of major
concern to the development of high extrudability. After homogenization the billets are
transferred to the extrusion press. Before extrusion, a preheating of the billets is performed.
The selection of the preheating temperature is crucial, since during extrusion the
temperature of the material rises due to the high deformation rate. Furthermore the exit
temperature from the extrusion press should not exceed a certain limit (e.g the solidus) in
order to avoid hot tearing resulting from local melting of low-melting temperature
eutectics. For high extrudability (high extrusion speeds) the preheating temperature should
be as low as possible. The equilibrium β-Mg2Si phase has a high solvus temperature and will
not dissolve during preheating. Therefore the cooling from the homogenization
temperature should aim at the precipitation of the metastable β΄-Mg2Si, which exhibited a
lower solvus. The precipitation of this metastable phase is influenced by the presence of the
so-called excess Si in the alloy. The KWN model was employed in order to quantify the
effect of excess Si on the precipitation of the metastable β΄-Mg2Si, in an effort to quantify the
effect of new alloy compositions on the CCP diagram. The approach was applied for the
6060 alloy.
Taking into account the three issues stated above as well as the results presented in
the previous chapters of this thesis, design rules are formulated for the achievement of high
extrudability.
110

26/03/2018 03:27:43 EEST - 137.108.70.7
5.2. Precipitation and hardening during cooling from the
homogenization state
5.2.1. Simulations with the KWN and Strength models
The majority of phase transformations, which take place in alloys, are diffusional and
proceed by thermally activated movements of atoms across a concentration gradient. Of
significant importance in relation to thermal industrial processing, are phase
transformations involving precipitation reactions resulting to a second phase in the form of
a particle population. During these transformations three basic physical mechanisms are
taking place, the nucleation of new particles the growth of the nucleated particles and the
coarsening of the particles resulting to a particle population which can be described by a
particle size distribution (PSD). The material final mechanical properties depend on the
particles volume fraction and radius and consequently on the shape of the PSD. Since the
original works of Lifshitz & Slyozov [4] and Wagner [5] only a few of the published
precipitation models are comprehensive to deal with non-isothermal transformations
treating nucleation, growth and coarsening taking place concurrently. The two main works
in this area were performed by Langer & Schwartz [6] who studied nucleation and growth
of droplets in near-critical fluids and Kampmann et al. [7,8] who proposed a numerical
procedure (KWN model) for the calculation of the particle size distribution in solid-state
transformations treating the concurrent nucleation, growth and coarsening of precipitates.
As mentioned above the KWN model was modified to handle precipitation of Mg2Si during
cooling from the homogenization temperature. The results of the model were introduced to
the strength model for the calculation of the final strength of the cooled billet, after
homogenization. The strengthening mechanisms considered we lattice resistance, solid
solution strengthening and precipitation strengthening. The details of the models are
discussed in [1-3]. Results will be presented for alloy 6063, where the experimental
validation was performed.
Three homogenization temperatures were considered 540, 560 and 580oC.
Homogenization time is not a parameter in these simulations, since it is considered that the
time is enough for complete dissolution of the Mg2Si phase during holding at the
homogenization temperature. Therefore all Mg and Si are in solid solution at the end of the
homogenization holding period. A critical parameter is the cooling rate. Two cooling
111

26/03/2018 03:27:43 EEST - 137.108.70.7
profiles were considered, since it was possible to accomplish only natural cooling and
forced cooling (with forced air circulation) in a laboratory scale. The cooling profiles are
shown in Figure 5.1 a, b. With natural cooling, cooling at 100oC is accomplished in 700 sec,
while with forced cooling the same temperature is reached in only 220 sec. In addition the
difference in homogenization temperature is evident only at the initial stages of cooling,
and then the three curves coincide.
Figure 5.1: Measured cooling profiles used in simulations and in the experimental
validation, (a) natural cooling, (b) forced cooling
The simulation results from the KWN model regarding the evolution of the volume
fraction and particle size of the Mg2Si phase as a function of time during cooling from the
homogenization temperature are shown in Figure 5.2. Regarding the volume fraction, it
appears that the homogenization temperature influences the initial stages of the
precipitation. Towards the end of cooling the curves corresponding to the three
homogenization temperatures coincide. The amount of Mg2Si formed under forced cooling
is slightly lower than the respective amount formed under natural cooling conditions, 0.79
and 0.82% respectively. However the final particle size at the end of cooling is lower in the
forced cooling condition, about 6.9nm, compared to 7.8nm for the natural cooling condition.
112

26/03/2018 03:27:43 EEST - 137.108.70.7
(a) (b)
Figure 5.2: Simulation results for the 6063 alloy: (a) evolution of Mg2Si volume fraction
during cooling for three homogenization temperatures 540, 560 and 580oC
and (b) evolution of particle size during cooling for natural and forced cooling
from 580oC homogenization temperature
Regarding the shape of the curves describing the size evolution during cooling, is
evident from Figure 5.2b that there is a competition between nucleation and growth
processes, which as mentioned in the Introduction of this chapter, take place
simultaneously. Three stages can be described. Following initial nucleation the mean size
drops due to the increase of the nucleation driving force with the decrease in temperature.
The rise of the mean size is followed by growth of the already nucleated particles. However
the further decrease of temperature during cooling induces a substantially higher driving
force for nucleation and the mean size is again reduced to its final value at the end of
cooling.
The respective particle size distributions (PSD) are given in Figure 5.3 for the 580oC
homogenization temperature and the two cooling profiles. The KWN simulation predicts
similar peak sizes for the two cooling profiles, but it also predicts a second peak and larger
particles for the natural cooling profile. This is due to the longer time available for the
precipitation process in the case of natural cooling.
113

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.3: Particle size distribution (PSD) predicted by the KWN model for the 6063 alloy
for the two cooling profiles, natural and forced cooling for the 580oC
homogenization temperature
The volume fraction of Mg2Si and the size evolution are used as input to the strength
model. The resulting hardness evolution is shown in Figure 5.4 for the two cooling profiles
and the three homogenization temperatures. The results indicate that forced cooling results
in slightly higher hardness 69 relative to 64.9 HV0.3 for natural cooling. This hardness
difference is attributed to the slightly finer dispersion of Mg2Si formed after forced cooling.
Figure 5.4: Evolution of hardness during cooling of the 6063 alloy for natural and forced
cooling and for three homogenization temperatures 540, 560 and 580oC.
114

26/03/2018 03:27:43 EEST - 137.108.70.7
5.2.2. Experimental validation
In order to validate the simulation predictions presented above, homogenization

experiments were performed in as-cast alloy 6063. Homogenization was performed in a
laboratory furnace at temperatures 540, 560 and 580oC for 8 hours. According to the
results presented in Chapter 3, this time is more than sufficient in order to dissolve all
Mg2Si present in the as cast microstructure. Two cooling methods were followed, natural
and forced air cooling, following the profiles depicted in Figure 5.1. The specimens were
then polished to a flat surface and microhardness measurements were taken with a Vickers
indenter and a load of 0.3 Kg. The hardness measurements appear in Table 5.1.
Table 5.1: Microhardness (HV0.3) as a function of cooling profile

Homogenization
Forced Cooling Natural Cooling
temperature (oC)
540 60.1, 59.2, 63.6, 64.1, 63.9 61.7, 61.8, 61.0, 61.2, 61.6
560 63.0, 63.8, 67.4, 64.3, 63.7 62.3, 60.0, 58.6, 61, 61.8
580 66.3, 66.4, 70.0, 65, 63.2. 64, 62.8, 61.0, 61.6, 61.2
Mean hardness 64.1 61.4
The hardness resulting from forced cooling is slightly higher than the one resulting
from natural cooling as predicted by the simulation. Regarding absolute values, the
simulation overestimates the hardness by 3-4 Vickers units in both cases, as shown in
Figure 5.4. Taking into account the approximations used in the KWN precipitation model as
well as the strength model, the agreement is considered satisfactory.
5.3. Hardening potential of homogenized billets

5.3.1. Introduction
An experimental study has been undertaken to identify the effect of homogenization

temperature and time on the hardening potential of the alloys during subsequent aging. It is
realized that a significant amount of Mg2Si is precipitated during cooling from the
115

26/03/2018 03:27:43 EEST - 137.108.70.7
homogenization temperature and, depending on the cooling rate, an amount of Mg and Si
still remains into solution. Under conditions of natural aging, the hardness of the material
could be further increased due to precipitation. Such conditions are encountered in
industrial practice in the time between billet homogenization and billet extrusion. These
two operations are usually performed in different industrial sites and, therefore, several
days could elapse between these two operations, during which natural aging is at play.
Referring to the hardening potential, it is defined as the hardness increase between the
homogenized state and the state after natural aging. The effect of homogenization
temperature, homogenization time, cooling rate and alloy composition on the hardening
potential was investigated experimentally [9]. The experimental procedures and the results
of this study are presented in the next sections.
5.3.2. Experimental procedures
Three alloys were investigated, 6082, 6005 and 6063, the chemical composition of
which is given in Table 2.2. The billets were sectioned and specimens were prepared and
homogenized according to the procedures described in Chapter 3. Three homogenization
temperatures 540, 560 and 580oC as well as four homogenization times 0.5, 4, 8 and 32
hours were selected. Similar homogenization parameters have been studied by other
researchers [10-13]. Three conditions regarding the cooling rate were studied: Water
Quenching (Q), air cooling (AC) and forced cooling (FC). A K-type thermocouple was
attached at the specimens in order to monitor the cooling profile. The cooling profiles are
depicted in Figure 5.5. The AC and FC specimens reach 150oC in 6 and 2.5 min respectively.
116

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.5: Homogenization cooling profiles
Brinnel hardness testing was performed in all specimens with a 5mm ball indenter
diameter and a load of 250Kg for 15 sec. The indentation diameter was measured by
microscopy. In order to determine the hardening potential, the hardness measurements
were performed immediately after cooling and also after 5 days of natural aging at room
temperature. The 5 day period has been also suggested in [12,14].
5.3.3. Results
Brinnel hardness is depicted as a function of homogenization temperature and cooling

rate in Fig. 6, 7 and 8 for alloys 6082, 6005 and 6063 respectively. Regarding the data for
the 6082 alloy in Figure 5.6, it appears that for the short homogenization times (0.5 and 4
hours) the quenched specimens possess higher hardness followed by the forced cooled and
air cooled specimens. This is due to the higher amount of Mg and Si into solution (solution
strengthening).
117

26/03/2018 03:27:43 EEST - 137.108.70.7
This hardness difference increases with alloying (606360056082). At longer

homogenization times the difference between the three cooling conditions diminished.
Similar situations can be described for the other two alloys as depicted in Figures 5.7 and
5.8 for 6005 and 6063 alloys respectively.
118

26/03/2018 03:27:43 EEST - 137.108.70.7
119

26/03/2018 03:27:43 EEST - 137.108.70.7
In all three alloys, the 8h homogenization treatment appears softer in almost all
homogenization temperatures and cooling rates. Moreover with increasing the Mg and Si
alloying elements the hardness increases for each specific cooling rate. This effect can be
seen more easily in Figures 5.9, 5.10 and 5.11, which depict the comparison between the
three alloys for a 4h homogenization time and for the 540, 560 and 580oC homogenization
temperature respectively.
120

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.9 Brinnel hardness immediately after cooling from 540oC homogenization
temperature
temperature
temperature
121

26/03/2018 03:27:43 EEST - 137.108.70.7
In all cases the homogenized alloys exhibited higher hardness differences between
them in the homogenized state in relation to the as-cast state. The quenched specimens
exhibit higher hardness relative to the air-cooled state only for the 540oC homogenization
temperature. These differences diminish at higher homogenization temperatures.
As mentioned above, the hardening potential of the homogenized billets was
investigated. The hardness of the homogenized billets after 5 days of natural aging is
depicted in Figures 5.12, 5.13 and 5.14 for the 6082, 6005 and 6063 alloys respectively.
122

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.12: Brinnel hardness after homogenization and natural aging for the 6082 alloy
123

26/03/2018 03:27:43 EEST - 137.108.70.7
It is evident that the quenching state (Q) exhibits higher hardening relative to the air
cooling (AC) and forced cooling (FC) states. In addition, the 8 h homogenization time, after
natural aging, does not correspond any more to the softer state. These effects can be better
examined directly by plotting the hardening potential of the alloys. The hardening potential
Hp can be expressed as
H NA  H AH
Hp  100
H AH
where, HAA is the hardness after 5 days of natural aging and HAH is the hardness immediately
after homogenization. The results are shown in Figures 5.15, 5.16 and 5.17 for the 6082,
6005 and 6063 alloys.
homogenization temperatures
124

26/03/2018 03:27:43 EEST - 137.108.70.7
homogenization temperatures
125

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.17: Hardening potential Hp for the 6063 alloy at (a) 540, (b) 560 and (c) 580 oC
homogenization temperatures.
For the 6082 alloy, the Q state exhibits the highest hardening potential followed by
the FC and AC states, this is due to the higher amount of Mg and Si retained in solid solution
after quenching from the homogenization temperature. The same effect is observed for the
6005 and 6063 alloys. Regarding the effect of homogenization time, the hardening potential
for the Q, FC and AC states increases with homogenization time, reaches a maximum at 8h
end then decreases. This is in line with the results presented above, indicating that the
softer state following homogenization is the one corresponding to the 8 h homogenization
time. It appears from these results that transformations other than the Mg2Si dissolution
during homogenization influence the Mg and Si in the matrix. The most probable influence
is from the β-AlFeSi→α-AlFeSi transformation, studied in Chapter 3, which liberates Si in
126

26/03/2018 03:27:43 EEST - 137.108.70.7
the Al matrix. This Si is combined with the Mg coming from the dissolution of eutectic
constituents and forms additional Mg2Si. For the 6005 alloy, the maximum hardening
potential also appears for the 8h homogenization time. However the difference between the
8h treatment and the other homogenization times diminishes with the homogenization
temperature. The same arguments hold for the 6063 alloy. Additionally in this case the
hardening potential at 0.5 and 4h homogenization times, increase with the homogenization
temperature. At 580oC homogenization temperature, the hardening potential of the 4 and
8h are practically the same. Finally for the lower alloyed 6063 and 6005 billets, the
hardening potential at shorter homogenization times (0.5 and 4h) increase with the
homogenization temperature.
The above results illustrate the complex interactions between alloying elements,
homogenization temperature, time and cooling rate on the strengthening mechanisms at
play and especially on solid solution and precipitation strengthening. More work is
required in this area, possibly involving a more rigorous approach, e.g an extended KWN
model in multicomponent systems, to quantify the observed behavior.
5.4. Effect of excess-Si on the CCP diagram

5.4.1. Introduction
Extrudability depends on the state of iron intermetallics and removal of

microsegregation during homogenization. However, extrudability also depends on the
strength of the billet prior to extrusion, i.e. the state of Mg2Si precipitation. Mg2Si
precipitation takes place during cooling from the homogenization temperature and also
during billet preheating. These Mg2Si precipitates should dissolve completely during
extrusion in order to develop a good response to the aging treatment. The rate of cooling is
important. It appears that a cooling cycle that produces a fine dispersion of β-Mg2Si is the
optimum for extrudability [15,16]. The reasons are: (a) finer particle dispersion, which
dissolves faster, (b) a lower dissolution temperatures because it possesses a lower solvus
temperature. Therefore during preheating, the β΄ will dissolve at a lower temperature, be
substituted by a fine dispersion of β-Mg2Si, which will dissolve readily during extrusion.
This enables the use of a lower preheating temperature, providing a larger window for
127

26/03/2018 03:27:43 EEST - 137.108.70.7
temperature rise up to the exit temperature. This means that a higher extrusion speed can
be applied. Precipitation of β takes place below 400oC. A CCP diagram for 6060 alloys can be
used to design the relevant cooling cycle. As far as alloying is concerned, it appears that the
amount of the so-called excess Si (Siex) is important as it affects the precipitation kinetics in
Al-Mg-Si alloys. Siex leads to finer and denser precipitates by enhancing the nucleation
kinetics of GP zones. It has been reported [17] that the excess Si in the alloy enhances the
nucleation of the hexagonal β- Mg2Si on Si atom grids, which are composed of hexagonal
sub cells with lattice parameters a=b=4.05A and c=4.05A (Angstroms). For a given Mg and
Fe content, the amount of Siex depends on which intermetallic (α-AlFeSi or β-AlFeSi) is
present. During homogenization, the transformation β→α leads to an increase of Siex since
the β-AlFeSi contains more Si than the α-AlFeSi phase. The aim of this section is to
determine the shift of the Continuous Cooling Precipitation diagram due to the amount of
the excess Si in the alloy. The methodology was applied for the 6060_3 alloy. The chemical
composition of the alloy was given in Table 2.2 in Chapter 2.
5.4.2. Methodology
Construction of CCP diagrams can be accomplished experimentally by applying

Differential Scanning Calorimetry (DSC). A sample is cooled continuously from the
homogenization temperature and the endothermic or exothermic peaks arising from
particle dissolution or precipitation are recorded. Such a DSC spectrum is given in Figure
5.18 taken from [18], for a 6005 alloy for several cooling rates.
128

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.18: DSC spectra for 6005 alloy (0.57Mg-0.68Si) for cooling rates from 0.1 to 375
K/min [18]
The first peak during cooling is attributed to β-Mg2Si, while the second peak at lower
temperatures is attributed to β΄-Mg2Si precipitation. It is clear that the application of a
higher cooling rate suppresses the precipitation of the β-Mg2Si, leaving only the
precipitation of the β΄-Mg2Si. Literature studies regarding the effect of cooling rate on the
precipitation of Mg2Si are limited. Only one publication regarding the development of
Continuous Cooling Precipitation (CCP) diagram was found [18] and one publication
regarding the isothermal TTP diagram of 6082 [19]. The work presented in this section is
based on the work presented in [18] where the CCP diagrams for five alloys 6060, 6063,
6005, 6082(L) and 6082(H) were determined. The composition of the 6060 alloy studied in
the above work is 0.43 Mg-0.40Si. The aim is to quantify the shift of the CCP with the new
alloy composition of 6060_3. It is evident from the above work that the position of CCP is
influenced by mainly the Mg content and secondly by the excess Si in the alloy. Mg
influences directly the amount of Mg2Si while the excess Si influences the heterogeneous
nucleation sites of the β΄-Mg2Si. The methodology followed is the following:
129

26/03/2018 03:27:43 EEST - 137.108.70.7
 Non linear fitting of the β-nose and the β΄-nose of the CCP with the Mg content using the
data of the CCPs from the work [18].
 Use of the equations derived in step (a) to calculate the nose position for a model alloy.
 Displacement of the β and β΄ noses due to the excess Si in the alloy. Final position of the
CCP curve of the 6060_3 alloy.
5.4.3. Results
The CCP diagrams for the alloys investigated in [18] are depicted in Figure 5.19.
Mg=0.43 Mg=0.47
Mg=0.57
Mg=0.62 Mg=1.05
Figure 5.19: CCP diagram for 6060, 6063, 6005 and 6082 alloys [18]
130

26/03/2018 03:27:43 EEST - 137.108.70.7
The times for the β and β΄ noses of the CCP diagrams are given in Table 5.2 as a function of
Mg content
Table 5.2 Position of β and β noses from the CCP diagrams of Fig.2
Time (sec) Time (sec) Logt Logt
Mg(wt%)
(β-nose) (β-nose) (β-nose) (β-nose)
0.43 1200 450 3.07 2.65
0.47 180 100 2.25 2.00
0.57 170 25 2.23 1.39
0.61 40 10 1.60 1.00
1.05 15 1.8 1.17 0.255
Nonlinear curve fitting has applied to the data of Table 5.3 in order to get an expression for
the position of the noses as a function of Mg content. An Arrhenius expression was selected
for the fitting. The fitting results are displayed in Figure 5.20
(a) (b)
Figure 5.20: Arrhenius fitting of the data of Table 2 a) Fitting the β-nose, b) fitting β’ - nose
The expression for the β-nose is
logt = 1.13665 + 22.4678e-5.89242 Mg (1)

The expression for the β΄-nose is
131

26/03/2018 03:27:43 EEST - 137.108.70.7
logt = 0.186353+ 27.3091e-5.64495 Mg (2)
For Mg=0.38 for 6060_3 alloy the position of the noses are
β-nose at t=4750 sec
β΄-nose at t=3750 sec
Thus the β and β΄ noses are displaced by 1000 sec based only on the Mg content of the alloy.
The next step is to find the displacement of the β΄-nose due to the excess Si in the alloy. The
excess Si is the Si remaining in the alloy after subtracting the Si necessary for the formation
of Mg2Si and the Si present in the α-AlFeMnSi phase after the β-AlFeSi→α-AlFeMnSi
transformation during homogenization.
The relevant equations are [20]:
[Si]xs = [Si]alloy – [Si]α-AlFeMnSi – [Si]Mg2Si (3)
[Si]Mg2Si = [Mg]alloy {Asi/2AMg} = 0.577 [Mg]alloy (4)
[Si]α-AlFeMnSi = [Fe+Mn]alloy {ASi/3(AFe+AMn)} = 0.0664 [Fe+Mn]alloy (5)

where A is the atomic weight of the alloying element.
The results are displayed in Table 5.3
Table 5.3: Calculation of excess Si of the alloys (compositions in wt%)

Alloy Mg Si Fe Mn [Si]xs
6060 0.43 0.39 0.22 0.02 0.126
6063 0.47 0.50 0.19 0.03 0.214
6005 0.57 0.68 0.20 0.11 0.330
6082(L) 0.61 0.69 0.196 0.45 0.295
6082(H) 1.05 1.23 0.20 0.65 0.567
6060_3 0.38 0.47 0.2 0.03 0.288
132

26/03/2018 03:27:43 EEST - 137.108.70.7
A continuous cooling precipitation model is applied in order to calculate the start time
for the precipitation of β΄-Mg2Si phase. In this work the KWN model was employed.
In order to calculate the shift of the β΄-nose the following parameters were considered:
The β΄-solvus curve in the phase diagram was displaced by 70oC to lower temperatures
relative to the equilibrium β solvus curve.
Because the β΄-phase is semicoherent with the Al matrix, the interfacial energy was reduced
to γ=0.3 J/m2 relative to γ=0.5 J/m2 for the equilibrium β-phase, which is incoherent.
The effect of excess Si was taken into account with two adjustable parameters in the
KWN model. The first parameter, the total number of nucleation sites No, was increased in
proportionality with the excess Si in the alloy. For the 6060_3 model alloy, a value 2No was
selected, where No is the value considered for the 6060 alloy in Table 5.3. The second
parameter, the activation energy, was reduced exponentially with the amount of excess Si.
In this way the activation energy for the 6060_3 model alloy was reduced to the value ΔGo/7
relative to the value of ΔGo for 6060 in Table 5.3.
The results are shown in Figure 5.21. The β ΄nose is displaced by 1530 sec and is located at
about 2220 sec.
Figure 5.21: Effect of excess Si on the precipitation of β’-Μg2Si
The shift in the CCP diagram of the 6060_3 alloy is shown in Figure 5.22. The dotted β΄
line is shifted by taking into account only the Mg effect as per equation (2). The full line
corresponds to the β΄-nose, taking into account the effect of excess Si in the alloy.
133

26/03/2018 03:27:43 EEST - 137.108.70.7
Figure 5.22: CCP diagram of the 6060_3 alloy (red lines). The dotted β΄ line is shifted by
taking into account only the Mg effect. The β΄ nose (full line) is shifted by
taking into account the effect of excess Si in the 6060_3 alloy
The resulting CCP diagram can be used for the design of a homogenization cooling program,
in order to achieve precipitation of the metastable phase β΄-Mg2Si.
5.5. Design rules for high extrudability
The aim of the present section is to formulate design rules regarding the
homogenization practice in order to obtain high extrudability. From the literature and from
the results of this thesis, regarding the simulation of solidification and homogenization of
6xxx alloys, the following preliminary remarks can be made:
Extrudability is a complex property, not measurable directly. Extrudability is usually
determined at the extrusion shop with several ways, like the maximum extrusion speed
accomplished without surface defects or the maximum pressure, or even the contact time in
the press. However, since extrudability is a property, it depends on the microstructure,
which in turn depends on the composition and the process chain from melt treatment and
134

26/03/2018 03:27:43 EEST - 137.108.70.7
alloying down to the extrusion press. All elements of the process chain have an effect on
extrudability:
1. melt conditioning, cleaning and alloying,
2. DC casting – development of the as cast microstructure,
3. billet homogenization, involving heating and holding
4. cooling from the homogenization temperature
5. billet preheating at the extrusion shop.
The rest of the process chain is also important because it affects extrudability indirectly. So
billet preheating is followed by
6. extrusion,
7. cooling from the extrusion temperature (also solution temperature) and
8. aging for precipitation strengthening.
It appears that from the above list, the most important design rules regarding high
extrudability are:
Melt purity: It is an important and usually overlooked variable, which influences
extrudability and yield in the press shop. It is influenced by the melt conditioning, cleaning
and alloying practice for the control of inclusions, hydrogen, alkali metal salts and the
desired alloying level.
As-cast microstructure: The microstructure developed during DC casting is affected by
the solidification conditions. Grain size and DAS are the most important factors. They can be
controlled by the casting speed (solidification rate) and seeding for grain refining. A smaller
grain size and DAS improves the strength and also shortens the diffusion distances, leading
to a more efficient homogenization for a given homogenization cycle. In terms of phase
fractions, the results presented in Chapter 2 allowed the mapping of the Mg-Si space for
given Fe and Mn contents. Compositions can be selected to minimize the undesirable β-
AlFeSi intermetallic in the as-cast microstructure, in order to achieve a good starting point
for the homogenization to follow. It was shown that Mn reduces the amount of β-AlFeSi for
a given Mg-Si-Fe combination.
Homogenization process: The aim of homogenization is to reduce the microsegregation
of elements and phases that are concentrated at grain boundaries and interdendritic areas.
In 6xxx alloys the undesirable β-AlFesi intermetallic phase, which limits extrudability,
135

26/03/2018 03:27:43 EEST - 137.108.70.7
transforms to a more rounded α-AlFeMnSi intermetallic phase. With the simulations
presented in Chapter 3, it is possible to determine the required time for Mg2Si dissolution
and the transformation β→α-AlFeSi as a function of homogenization temperature and alloy
composition. It was shown that Mn has a beneficial effect on the kinetics of the β to α
transformation. After the transformation β→α is complete, the homogenization process
continuous since the α phase undergoes coarsening and spheroidization, adopting a
necklace morphology. Microstructural indices, such as the aspect ratio and the circularity,
were determined in Chapter 4 and can be used to characterize the homogenization stage
quantitatively. These parameters can be determined by quantitative metallography (image
analysis). A fully homogenized billet, with the potential for high extrudability should have
all β-AlFeSi transformed to α-AlFeSi, with necklace morphology and as aspect ratio and
circularity approaching unity. In addition all Mg2Si should be dissolved and Mg and Si
should be distributed as uniformly as possible in the grain interiors.
Cooling from the homogenization temperature: The rate of cooling is important. It
appears that a cooling cycle that produces a fine dispersion it β΄-Mg2Si is the optimum,
because it possesses a lower solvus temperature. Therefore during preheating, the β΄ will
dissolve at a lower temperature, be substituted by a fine dispersion of β-Mg2Si, which will
dissolve readily during extrusion. This enables the use of a lower preheating temperature,
providing a larger window for temperature rise up to the exit temperature. This means that
a higher extrusion speed can be obtained. The cooling program should be designed so that
precipitation of β-Mg2Si is avoided and only precipitation of β΄-Mg2Si takes place. A CCP
diagram like the one developed for the 6060_alloy in this Chapter can be used to design the
cooling cycle.
Preheating of billet prior to extrusion: It is important to match the preheating
conditions to the homogenization cooling conditions in order to get the most benefit of
homogenization. β-Mg2Si precipitates require higher temperatures to dissolve (higher
solvus) and have a detrimental effect on hot ductility. On the other hand the β΄-Mg2Si ppts,
are finer, semicoherent with the matrix and have a lower solvus temperature. Thus a lower
preheating temperature can be employed in combination with a high extrusion speed. This
will raise the exit temperature to a value sufficient for complete dissolution of Mg2Si.
136

26/03/2018 03:27:43 EEST - 137.108.70.7
5.6. Conclusions
From the results presented in this chapter the following conclusions can be drawn:
 Precipitation during cooling from the homogenization temperature was simulated with the
KWN precipitation model. The relevant particle size distribution (PSD), evolution of
volume fraction and particle size during cooling were calculated. The results were
introduced in a suitable strength model to calculate the resulting hardness. The simulation
results are in satisfactory agreement with experimental data. Forced air cooling results in
higher hardness in relation to natural cooling for the alloy 6063 investigated.
 The hardening of the homogenized billets was investigated experimentally. Regarding the
effect of the homogenization cooling rate, water quenching results in higher hardness
relative to air cooling and forced air cooling. The hardness difference increases with
alloying in the order 6063→6005→6082. The effect is attributed to the higher amounts of
Mg and Si retained in solid solution during cooling.
 Regarding the effect of alloy composition, homogenization temperature, homogenization
time and homogenization cooling rate on the hardening potential of the homogenized
billets, the alloys water that were quenched from the homogenization temperature
exhibited the higher hardening potential relative to the air and forced cooling. In addition,
the maximum hardening potential is exhibited for the 8h homogenization time,
irrespective of the homogenization temperature.
 For the lower alloyed 6063 and 6005 billets, the hardening potential at shorter
homogenization time increase with the homogenization temperature.
 The effect of excess Si on the precipitation of β΄-Mg2Si was determined using the KWN
precipitation model. The resulting shift in the continuous cooling precipitation (CCP)
diagram was calculated. This allows the design of a suitable cooling program in order to
avoid the precipitation of β-Mg2Si and promote the precipitation of β΄-Mg2Si. This allows
the use of a lower preheating temperature prior to extrusion and leads to the use of higher
press loads in order to obtain higher extrusion speeds.
 Specific design rules were formulated, regarding the as-cast microstructure, the
homogenization temperature and time as well as the cooling from the homogenization
temperature in order to obtain high extrudability.
137

26/03/2018 03:27:43 EEST - 137.108.70.7
5.7. References
[1].S. Samaras Ph.D Thesis, University of Thessaly, Mech. Eng (2006)

[2].A. Serafeim, Diploma Thesis, University of Thessaly, Mech. Eng. (2013)
[3].J.Fanikos, Diploma Thesis, University of Thessaly, Mech. Eng. (in progress)
[4].I. M. Lifshitz, V.V. Slyozov, J.Phys.Chem.Solids, (1961)19:35.
[5].C. Wagner, Z. Elektrochem., 65 (1961) 581.
[6].J. S. Langer, A. Schwartz, Physical Review A 21 (1980) 948.
[7].R. Wagner, R. Kampmann, In Mater. Sci. Techn. Weinheim 5 (1991) 21.
[8].R. Kampmann, H. Eckerlebe, R. Wagner, Mater. Res. Soc. Symp. Proc., 57 (1987) 52
[9].A.Tousias, Diploma Thesis, University of Thessaly, Mech. Eng. (2014)
[10]. O. Reiso, Marter. Forum, 28 (2004) 32-46
[11].A. Jackson, T. Sheppard, Alum., Assoc., 1 (1998) 1-20.
[12].Y. Birol, Trans., Nonferrous Met. Soc. China, 23 (2013) 1875-1881
[13].G. Mrowks-Nowotnik, J. Sieniawski, J. Mater. Proc. Techn. 162-163 (2005) 367-372
[14].M.A. Abdel-Rahman, Def., Dif., For., 303-304 (2010) 107-112
[15].S. Zajac, B. Bengtsson, C. Jonsson, Mater. Sci. Forum 396-402 (2002) 399-404
[16]. H. Zhu, M. Couper, A. Dahle, Jour., Met., 63, No.11 (2011) 66-71.
[17]. Marioara C.D., Andersen S.J., Stene T.N., Hasting H., Walmsley J., Vanhelvoort A.J.,
Holmestad R., Philos. Mag. 87 (2007) 3385-341
[18]. Milkereit B., Wanderka N., Schick C., Kessler O., Mat., Sc., Eng. A 550 (2012) 87-96.
[19]. Shang B.C., Yin Z.M., Wang G., Liu B., Huang Z.Q., Mat. Des. 32 (2011) 3818-3822
[20]. Asensio-Lozano J., Suarez-Pena B., Vander Voort G.F., Mat. 7 (2014) 4224-4242
138

26/03/2018 03:27:43 EEST - 137.108.70.7
6. Conclusions
The results presented in this thesis lead to the following conclusions:
Regarding the as-cast microstructures

 Computational alloy thermodynamics, based on the CALPHAD approach, has been
applied to perform Scheil-Gulliver solidification simulations. The resulting
microsegregation of elements and phases in the as-cast microstructure were
calculated as a function of alloy composition.
 The results of the simulations were confirmed experimentally by quantitative image
analysis for the measurement of phase fractions.
 The variation of the mole fractions of the extrudability-limiting β-AlFeSi phase and
the strengthening Mg2Si phase with alloying elements has been mapped over the
useful range (0-1.2 mass%) in the Mg-Si composition space.
 The constructed maps indicate that low mole fractions of β-AlFeSi are associated
with lower Si and higher Mg compositions. On the other hand, high mole fractions of
Mg2Si are associated with both higher Si and Mg compositions, with Mg possessing a
stronger effect.
 Construction of maps for different levels of Mn has shown that addition of Mn could
allow for higher alloying with Mg and Si, in order to obtain higher amount of Mg2Si,
without at the same time increasing the β-AlFeSi intermetallic phase in the as-cast
microstructure.
Regarding the homogenization process
 The Dual grain model (DGM) has been developed to treat the homogenization
process in multicomponent and multiphase Al-alloys exhibiting a large variability of
the as-cast grain size. With the model it is possible to simulate the temporal and
spatial evolution of phase fractions and element concentration during
homogenization.
139

26/03/2018 03:27:43 EEST - 137.108.70.7
 Regarding the evolution of phase fractions during homogenization, the predictions
of the DGM model have been validated experimentally with XRD analysis. The
evolution of the β-AlFeSi→α-AlFeSi transformation is predicted by the DGM and is
confirmed by XRD in the same time scale.
 The DGM predictions regarding the evolution of Fe and Mg concentrations with
homogenization time are in excellent agreement. There are only some discrepancies
in the profiles of Si and Mn. The Si discrepancy is due to the Si-diamond phase not
taken into account in the DGM. The Mn discrepancy is due to the Mn retention in the
α-AlFeSi phase.
 The DGM predicts the fast dissolution of Mg2Si during homogenization and its re-
precipitation during cooling. It also predicts its dissolution during preheating prior
to extrusion.
 The DGM can describe the temporal and spatial evolution of the β-AlFeSi→α-AlFeSi
transformation. The spatial evolution exhibits an exact spatial correspondence. In
addition the transformation starts at a certain distance from the boundary and the
transformation front moves towards the boundary.
 Both the Mg2Si dissolution and the β-AlFeSi→α-AlFeSi transformation are faster in
the smaller grain as predicted by the DGM. In addition the concentration profiles of
the elements homogenize faster in the smaller grain, indicating that a fine grain size
in the as-cast microstructure accelerates the homogenization process.
 The DGM predicts correctly the effect of homogenization temperature. The rate of
both the Mg2Si dissolution and the β-AlFeSi→α-AlFeSi transformation increase with
the homogenization temperature.
 A preliminary attempt to develop homogenization process maps has been performed
using the DGM. These maps are, at present, based on Mg2Si dissolution and the β-
AlFeSi→α-AlFeSi transformation and can be used for the design of the
homogenization heat treatment.
 The quantification of the homogenization state was achieved by determine specific
microstructural indices, such as the aspect ratio and the circularity of the
intermetallic phases. In this way it was possible to quantify the shape evolution of
the intermetallic compounds during homogenization.
140

26/03/2018 03:27:43 EEST - 137.108.70.7
Regarding the homogenization cooling
 Precipitation during cooling from the homogenization temperature was simulated

with the Kampman-Wagner Numerical (KWN) precipitation model. The relevant
particle size distribution (PSD), the evolution of volume fraction and the particle size
during cooling were calculated. The results were introduced in a suitable strength
model to calculate the resulting hardness. The simulation results are in satisfactory
agreement with experimental data. Forced air-cooling resulted in higher hardness in
relation to natural cooling for the alloy 6063 investigated.
 The hardness of the homogenized billets was investigated experimentally. Regarding
the effect of homogenization cooling rate, water quenching results in higher
hardness relative to air cooling and forced air cooling. The hardness difference
increases with alloying in the order 6063→6005→6082. The effect is attributed to
the higher amounts of Mg and Si retained in solid solution during cooling.
 Regarding the effect of alloy composition, homogenization temperature,
homogenization time and homogenization cooling rate on the hardening potential of
the homogenized billets, the alloys that were water-quenched from the
homogenization temperature exhibited the higher hardening potential relative to
the air and forced cooling. In addition, the maximum hardening potential was
exhibited for the 8h homogenization time, irrespective of the homogenization
temperature.
 The effect of excess Si on the precipitation of β΄-Mg2Si was determined using the
KWN precipitation model. The resulting shift in the continuous cooling precipitation
(CCP) diagram was calculated. This allows the design of a suitable cooling program
in order to avoid the precipitation of β-Mg2Si and allow the precipitation of β΄-Mg2Si.
This allows the use of a lower preheating prior to extrusion and leads to the use of
higher press loads in order to obtain higher extrusion speeds.
141

26/03/2018 03:27:43 EEST - 137.108.70.7
Regarding the design rules
 Specific design rules were formulated, regarding the as-cast microstructure, the
homogenization temperature and time as well as the cooling from the
homogenization temperature in order to obtain high extrudability.
 The microstructure developed during DC casting is affected by the solidification
conditions. Grain size and dendrite arm spacing (DAS) are the most important
factors. They can be controlled by the casting speed (solidification rate) and seeding
for grain refining. A smaller grain size and DAS improves the strength and also
shortens the diffusion distances, leading to a more efficient homogenization for a
given homogenization cycle.
 Phase fraction mapping, developed in Chapter 2, can be used for the selection of
alloy compositions to minimize the undesirable β-AlFeSi intermetallic in the as-cast
microstructure, in order to achieve a good starting point for the homogenization to
follow. In addition, it has been shown that Mn reduces the amount of β-AlFeSi for a
given Mg-Si-Fe combination.
 With the DGM model presented in Chapter 3, it is possible to determine the required
time for Mg2Si dissolution and the transformation β→α-AlFeSi as a function of
homogenization temperature and alloy composition. It has also been shown that Mn
has a beneficial effect on the kinetics of the β to α transformation. The DGM model
also indicated that a small as-cast grain size accelerates the kinetics of
homogenization, a point that was indicated above.
 Microstructural indices, such as the aspect ratio and the circularity, were
determined in Chapter 4 and can be used to characterize the homogenization stage
quantitatively. A fully homogenized billet, with the potential for high extrudability
should have all β-AlFeSi transformed to α-AlFeSi, with necklace morphology and
with aspect ratio and circularity approaching unity. In addition all Mg2Si should be
dissolved and Mg and Si should be distributed as uniformly as possible in the grain
interiors.
142

26/03/2018 03:27:43 EEST - 137.108.70.7
 The cooling from the homogenization temperature should be designed so that
precipitation of β-Mg2Si is avoided and only precipitation of β΄-Mg2Si takes place. A
CCP diagram like the one developed for a 6060 alloy in this Chapter 5 can be used to
design the cooling cycle.
 It is important to match the preheating conditions to the homogenization cooling
conditions in order to get the most benefit of homogenization. β-Mg2Si precipitates
require higher temperatures to dissolve (higher solvus) and have a detrimental
effect on hot ductility. On the other hand the β΄-Mg2Si ppts, are finer, semi – coherent
with the matrix and have a lower solvus temperature. Thus a lower preheating
temperature can be employed in combination with a high extrusion speed. This will
raise the exit temperature to a value sufficient for complete dissolution of Mg2Si.
143

26/03/2018 03:27:43 EEST - 137.108.70.7
144

26/03/2018 03:27:43 EEST - 137.108.70.7
7. Proposal for further research
The present thesis provided answers to specific questions and methodologies towards
the computationally-based design of extrudable Al-alloys. However, as it was expected, new
questions and new research topics were generated. Some of these topics are outlined
below:
Model integration
While the simulation of microsegregation has been linked to the DGM, the
precipitation during homogenization cooling is an independent simulation based on the
KWN model. It is proposed to link this model to the DGM in the sense that the results of the
DGM can be used as a direct input to the KWN model. A Matlab code could be written to
perform this task. In this way it would be possible to link also the preheating prior to
extrusion; so that all parts of the process chain prior to extrusion can be integrated. This
will increase the level of model integration and provide an easier-to-handle computational
tool for design purposes.
Homogenization cooling
In this thesis the KWN model was used to simulate the precipitation during
homogenization cooling. The current version of the model handles only one set of
precipitates in binary systems (matrix-precipitate) as in the Al-Mg2Si system. Furthermore
only equilibrium phases are considered. It is proposed to investigate the application of the
new software TC-Prisma, which is based on the KWN model, but additionally relies on the
thermodynamic and kinetic databases of Thermo-Calc and Dictra. The new versions of
Thermo-Calc provide values for the interfacial energies and contain data for metastable
phases, making it possible in the near future to simulate precipitation of metastable or
transition phases, such as the β΄-Mg2Si.
145

26/03/2018 03:27:43 EEST - 137.108.70.7
DGM model
The DGM for homogenization can be further refined through parametric simulation
studies such as (a) looking at different geometries (spherical or cylindrical), (b) simulating
different grain sizes, (c) taking into account increased diffusivities due to high dislocation
density. In addition the effect of composition such as the effect of Mn or other alloying
elements on the α-AlFeSi→β-AlFeSi transformation could be investigated.
Homogenization Process Maps
A preliminary attempt to generate a homogenization process map was presented in this

thesis based on the time for the completion of the α-AlFeSi→β-AlFeSi transformation. These
maps should be extended for other alloy compositions (with different Fe and Mn contents),
in order to generate a master process map for the homogenization of 6xxx alloys.
Shape change during homogenization
It appears that in addition to the Mg2Si dissolution and the α-AlFeSi→β-AlFeSi

transformation, a third transformation regarding the shape change of the α-AlFeSi phase
takes place during homogenization. This transformation has three stages (a) edge rounding,
(b) particle pinching and (c) neckless formation and spheroidization.
A model describing the shape evolution of the α-AlFeSi should be developed and linked to
the DGM in order to generate a complete homogenization model for 6xxx alloys.
Extrusion and aging
A next step to process integration is the link of the homogenization model with
available or to-be-developed models for hot extrusion based on the finite element method.
Extrusion is a deformation process carried out under high deformation rates at high
temperatures. Several metallurgical phenomena, such as recovery, recrystallization, phase
146

26/03/2018 03:27:43 EEST - 137.108.70.7
dissolution and reprecipitation take place concurrently during the extrusion process. The
material after extrusion is subjected to the final aging treatment for precipitation
strengthening. The simulation of extrusion and aging combining mechanics and metallurgy
seems to be a good research topic, which will allow the complete process integration and,
therefore, the full scale computational design of extrudable Al-alloys.
147

26/03/2018 03:27:43 EEST - 137.108.70.7
148

26/03/2018 03:27:43 EEST - 137.108.70.7
Publications by the author
1. P.I. Sarafoglou and G.N. Haidemenopoulos, Phase Fraction Mapping in the As-Cast
Microstructure of Extrudable 6xxx Aluminum Alloys, Int. Journal of Materials Research
105, No.12, pp. 1202-1208 (2014).
2. G.N. Haidemenopoulos and P.I. Sarafoglou, Extrudable Al-alloys: microsegregation
and homogenization, to appear in the Encyclopedia of Aluminum Alloys, Taylor and
Francis, NY, (invited paper, under preparation)
3. P.I. Sarafoglou, I. Aristeidakis, M. Tzini and G,N. Haidemenopoulos, Index-based
quantification of the homogenization state in extrudable Al-alloys, To be submitted in
Metals, 2016 (under preparation)
4. P.I. Sarafoglou and G.N. Haidemenopoulos, Simulation of the microsegregation and
homogenization, including hardening during homogenization cooling, of the 6082 Al-
alloy, to be submitted in Materials Science and Engineering A, 2016 (under
preparation)
5. P.I Sarafoglou and G.N. Haidemenopoulos, On the computational-based design of 6xxx
series extrudable Al-alloys, MSE Conference, Darmstadt, Germany, 23-25 September
2014.
6. P.I. Sarafoglou and G.N. Haidemenopoulos, Design of 6xxx alloys: Control of
intermetallic phases in the as-cast microstructure, 5th Hellenic Conference on
Metallic Materials, Volos, 2013
7. Maria-Ioanna Tzini, Panagiota Sarafoglou, Andreas Stieben, Gregory
Haidemenopoulos, Wolfgang Bleck, Austenite evolution and solute
partitioning during thermal cycling in the intercritical range of a medium-Mn steel,
Junior Euromat, 10-16 July, 2016, Lausane, Switzerland
In addition to the publications based on the thesis the following publications were
concluded:
149

26/03/2018 03:27:43 EEST - 137.108.70.7
1. L.A. Spyrou, P.I. Sarafoglou, N. Aravas and G.N. Haidemenopoulos, Evaluation of Creep
Damage of INCOLOY 800HT Pigtails in a Refinery Steam Reformer Unit, Engineering
Failure Analysis 45, pp. 456-469 (2014).
2. P.I. Sarafoglou, M.I.T. Tzini and G.N. Haidemenopoulos, Simulation of cyclic
transformations in the intercritical range of a 5Mn steel, Int. Journal of Materials and
Metallurgical Engineering, 1 (104) 2015 DOI:18:15344/IJMME/2015/104
3. G.N. Haidemenopoulos, P. I. Sarafoglou, P. Christopoulos and A.D. Zervaki,
Metallurgical Study of Initiation and Propagation of RCF cracks in Rails, ICEAF IV,
Skiathos 24-26 June, 2015
4. P.I. Sarafoglou, M.I.T. Tzini and G.N. Haidemenopoulos, Simulation of cyclic
transformations on a 0.2C-5Mn steel, ICEAF IV, Skiathos 24-26 June, 2015
5. G.N. Haidemenopoulos, P.I. Sarafoglou, P. Christopoulos, A. Zervaki, Rolling Contact
fatigue cracking in rails subjected to in-service loading, Fatigue and Fracture of
Engineering Materials and Structures, 2016
150

26/03/2018 03:27:43 EEST - 137.108.70.7

Dictra

Uploaded by

Copyright:

Available Formats

Dictra

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dictra

Uploaded by

Copyright:

Available Formats

University of Thessaly

Simulation and design

Development of design rules for high extrudability

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

National Technical University of Athens

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Figure 2.1: The process chain of extrudable Al-alloys

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Standard metallographic techniques including optical microscopy and scanning

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Mg-Si map of phase fractions in

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

Institutional Repository - Library & Information Centre - University of Thessaly

2.2.1. Computational Alloy Thermodynamics

Institutional Repository - Library & Information Centre - University of Thessaly

G is the Gibbs free energy of phase i

Gi  G o  Gid  G xs  G mag (2)

G o is the contribution of the pure components

G xs is the excess term corresponding to the deviation from ideality and

Institutional Repository - Library & Information Centre - University of Thessaly

G oi are the free energies of the elements

The free energy contribution of the pure elements is

Go, LIQ  X Al GAlo, LIQ  X Mg GMg

Where, for example, GMg

and for 923<Τ<6000Κ the free energy term is

Institutional Repository - Library & Information Centre - University of Thessaly

XS ,LIQ (0) (1) (2)

(0) (1) (2) (3) (0)

(0) (1) (2) (3) (4)

(0) (1) (2) (3)

Institutional Repository - Library & Information Centre - University of Thessaly

The β-AlFeSi phase is treated as a stoichiometric Al14Fe3Si3 intermetallic compound.