Sunteți pe pagina 1din 12

Thermochimica Acta 651 (2017) 2233

Contents lists available at ScienceDirect

Thermochimica Acta
journal homepage: www.elsevier.com/locate/tca

An alternative method to identify critical temperatures for semisolid


materials process applications using differentiation
Gabriela Lujan Brollo, Ceclia Tereza Weishaupt Proni, Leandro Cssio de Paula,
Eugnio Jos Zoqui
Materials and Manufacturing Department, Faculty of Mechanical Engineering, University of Campinas UNICAMP, Campinas, SP 13083-860, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: A novel method is proposed for identifying critical temperatures in thixoformability analysis. The pro-
Received 8 November 2016 posed method applies the principles of differential calculus to semisolid transformation curves. The
Received in revised form 6 February 2017 method is compared with the conventional tangent method for determining the solidus and liquidus
Accepted 9 February 2017
temperatures using DSC curves and with the visual recognition method for identifying the knee, or
Available online 16 February 2017
unstable main eutectic transformation temperature. Numerical simulation is performed with Thermo-

Calc (under the Scheil condition) for A356 alloy, and the same alloy is used to generate experimental
Keywords:
Semisolid processing DSC data for cooling and heating cycles under different kinetic conditions (5, 10, 15, 20 and 25 C/min) in
Thixoformability order to compare the conventional methods for identifying critical temperatures and the novel method
Thermodynamic characterization described here. The ndings indicate that the proposed method, referred to here as the differentiation
method, is an efcient tool for identifying critical points such as the solidus, liquidus and knee tem-
peratures and that the results obtained using the method are in good agreement with results obtained
with traditional methods. The method also proved to be operator-independent, as it uses well-dened
mathematical/graphical criteria to identify critical points. Furthermore, the new method is easy to use,
as after differentiation the critical points become easily recognizable visual and numerical features such
as valleys and peaks (for the knee temperature) and zeros (for the solidus and liquidus) on the curve
being analyzed. The following formal denition for the eutectic knee, the rst of its kind in the literature,
is proposed: the peak/valley in the expected eutectic range of the differentiated DSC curve of the solidica-
tion/melting transformation. Application of the proposed denition yields results that agree well with
results obtained by the traditional visual recognition method, indicating its suitability for identication
of the eutectic knee, especially when a comparative study of different shaped curves is being performed.
A denition that allows suitable lower and upper limits for the SSM processing window to be identied
is also proposed.
2017 Elsevier B.V. All rights reserved.

1. Introduction is to ensure that the solid/liquid mixture can be easily controlled


and that the semisolid alloy microstructure has an spheroidal
The term semisolid materials (SSM) processing refers to the rather than a dendritic morphology. This can be achieved using
forming of metallic alloys in the semisolid temperature range [1,2]. several techniques, such as stirring during casting or reheating
SSM processing is divided into two main categories: rheocasting, in the alloy after it has been subjected to forming or grain-rened
which controlled cooling is applied to a liquid alloy so it can reach [49].
the semisolid state with a solid fraction of approximately 0.20.5 A spheroidal microstructure, which is extremely temperature-
[3], and thixoforming (or thixocasting), in which a solid billet is and time-dependent [10,11], ensures that the alloy exhibits
heated in a controlled manner up to a liquid fraction of approxi- semisolid thixotropic behavior when sheared [12,13]. The practical
mately 0.30.6 in the semisolid range [4]. The key to SSM processing result of this special rheology is smooth, laminar die lling, avoiding
defects usually found in conventional die casting, such as porosities,
and requiring the use of only a fraction of the force needed in con-
Corresponding author. ventional metal forming, with consequent energy savings. Other
E-mail addresses: gbrollo@fem.unicamp.br (G.L. Brollo), wyliah@gmail.com advantages include lower processing temperatures, reduced ther-
(C.T.W. Proni), leandrocp@fem.unicamp.br (L.C. de Paula), zoqui@fem.unicamp.br mal shock to the die, less solidication shrinkage, reduced material
(E.J. Zoqui). loss and lower process costs [1,2].

http://dx.doi.org/10.1016/j.tca.2017.02.010
0040-6031/ 2017 Elsevier B.V. All rights reserved.
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 23

Most commercial SSM processes use conventional A356 and a. Determining the solidus and liquidus temperatures. Current
A357 aluminum casting alloys [14]. Satisfying the demand for a simulation techniques can only precisely predict solidication
wider range of alloys for use in SSM processing represents a chal- phenomena [2224], as the theory governing melting has yet to be
lenge, as candidate alloys must have adequate thixoformability, i.e., fully elucidated, making selection of new alloys suitable for thixo-
their thermodynamic properties must allow them to be worked in forming processes difcult. When kinetic effects are signicant,
the semisolid state. This requirement is closely linked to control of this difculty is aggravated because of the changes in the shape of
the SSM process; as rst stated by Tzimas and Zavalianglos [15,16], the semisolid curve caused by the increase in the cooling/heating
the variation in liquid fraction (fl ) with temperature (T) of the candi- rate. This is an important issue in SSM processing research, par-
date alloy (a parameter that later came to be known as sensitivity ticularly in the case of thixoforming, as heating rates under real
(S) and is dened as dfl /dT) should be as low as possible so that operating conditions are typically 50100 C/min, posing signi-
temperature variations inherent to the process do not signicantly cant challenges for current simulation techniques.
affect the formation of solid or liquid, ensuring the reproducibility Differential scanning calorimetry (DSC) is normally performed
of the process. to determine experimentally how fl varies with T in semisolid
Alloys used in semisolid applications have a primary phase alloys. The technique can directly measure melting/solidication
followed by a (binary or ternary) eutectic transformation during heat ow rate during the solid-to-liquid/liquid-to-solid transition
solidication. Formation of the second phase usually causes a sharp under a broad range of kinetic conditions [25,26]. Flynns tech-
change in fl with T, i.e., a high dfl /dT, so the alloy should be processed nique, which involves integration of partial areas between the DSC
only in the primary-phase eld [17]. Another issue when processing curves and a chosen baseline, is used to obtain the fl or fs vs. T curve
in the eutectic region is the non-thixotropic behavior of the alloy normally used in thixoformability analysis [27].
because of the presence of sharp particles along with primary- Starting with the DSC curves, the liquidus and solidus are usu-
phase spheroids [18]. ally identied by what we here call the tangent method, in which
Since the study by Tzimas and Zavalianglos, several authors have two tangents to the DSC curve are plotted when its slope changes in
contributed to the development of thixoformability criteria: Liu relation to a dened baseline, indicating that a phase transforma-
et al. [19] propose the well-known and widely used three rules tion is starting or ending. The point at which these two lines meet
based on the fl vs. T semisolid transformation curve: (a) a maxi- is considered an approximation to the point at which the trans-
mum solidication interval of 130 C to prevent hot tearing; (b) an formation starts or ends [25]. This method is to a certain extent
fl range of 0.40.6 in the eutectic transition (known as the knee operator-dependent, as determination of a suitable tangent is based
in the fl vs. T curve) to prevent the billet collapsing as a result of on common sense and may be subject to measurement errors, as
processing the material with too much liquid and (c) a maximum seen in Fig. 1, where the choice of different pairs of tangents (a and
sensitivity of 0.03 C1 at fl = 0.5 to ensure the process is stable when b) leads to different temperatures (points A and B, respectively) to
liquid is being formed in the working window. Uggowitzer and mark the same transition in the curve.
Uhlenhaut [20] suggest that SSM processing should be performed b. Determining the knee. Despite its importance in thixoformabil-
at lower and higher values of fl and that forming parameters should ity analysis, there is no formal graphical or mathematical denition
be adjusted to take into account variations in the composition of the of the eutectic transformation in the literature. In practice it is
alloy. Finally, Zoqui et al. [21] propose that a much greater range of determined by visually identifying a knee-like shape in the fl vs.
alloys can be used in semisolid applications if the working window T curve (a technique we refer to here as the visual recognition
is not limited to fl values in the range 0.40.6 and the sensitivity method). This method is highly operator-dependent, increasing the
curve is analyzed in the entire semisolid range to evaluate all pos- chances of incorrectly identifying the knee temperature. The lack of
sible regions in which the alloy can be successfully thixoformed, a dened criterion for locating this temperature is especially critical
i.e., all temperatures at which dfl /dT < 0.03 C1 between the unsta- when a comparative analysis is required. This is illustrated in Fig. 2,
ble eutectic transformation (knee) and the temperature at which where the hypothetical semisolid curves have different shapes. The
melting is complete (liquidus). smoother the curve, the less accurately the knee temperature can
When analyzing the thixoformability of new alloys for use in be identied, making it difcult to compare this temperature for the
semisolid processing it is essential to identify (a) the solidus and different conditions being analyzed. This situation can occur when,
liquidus temperatures, which are required to plot an accurate for example, comparing curves generated under distinct kinetic
semisolid curve and (b) the knee point, without which the working conditions.
window cannot be dened. The following limitations of techniques In light of the issues discussed above, this paper proposes
used to identify these two points are discussed here: a method for accurately dening the critical transformation

Fig. 1. Hypothetical endothermic DSC curve (left). Zoom-in on the region at the end of the transformation (right): two different pairs of chosen tangents (pair a continuous
gray lines; pair b dashed gray lines) result in two different points (A and B, respectively) to mark the same transition.
24 G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233

software by applying the Flynn method [27] of integration of partial


areas under the DSC curves, using the tangent and differentiation
methods to determine the solidus and liquidus. The differentiation
method also allowed several critical temperatures to be identi-
ed, such as the knee and the upper and lower limits of the SSM

working window. Origin software was used for further integration
and differentiation operations. Analysis of the heating and cooling
cycles allowed both melting and solidication transformations to
be analyzed. The use of several heating and cooling rates allowed a
comparative study of the kinetic behavior of the alloy to be carried
out.

Thermodynamic simulation was performed with Thermo-Calc
to generate the fl vs. T curve for the A356 alloy under the non-
equilibrium Scheil condition, which is based on the following
assumptions: (a) equilibrium exists at the solidliquid interface;
Fig. 2. Hypothetical fl vs. T curves with different shapes. Identication of the knee (b) no back diffusion occurs in solid phase that has already formed;
becomes less accurate as the curve becomes smoother. The sharpest curve, a, has (c) the composition throughout the entire liquid phase is homo-
a clearly identiable knee (point A), while the knees of the smoother curves, b and geneous and (d) the solid and liquid phases have equal densities
c, cannot be clearly identied. The segments BB and CC , respectively, correspond [23]. The results of this simulation were used as a reference in the
to possible positions for these knees.
analysis of the experimental curves. The percentages of Al, Cu, Mg,
Fe and Ti were taken into account in the simulation.
temperatures in the semisolid state. The proposed method com-
bines principles of differential calculus and features of the DSC 3. Results and discussion
curves as an alternative to the tangent method for identifying
liquidus and solidus temperatures and to the visual recognition 3.1. The proposed method
method for determining the knee on the fl vs. T curves of candi-
date alloys for semisolid applications. Criteria for an SSM working This section considers some practical aspects of the proposed
window are proposed, and the new method is used to determine method, which we call the differentiation method (DM). Specif-
further semisolid critical points and specify a suitable window. ically, it describes steps that should be followed to accurately
identify several critical temperatures in the semisolid range, such
2. Experimental procedure as the solidus, liquidus and eutectic knee, and determine a suit-
able working window for SSM processing. The theory underlying
To evaluate the applicability of the proposed method, thermo- the proposed method is presented in the Appendixes. Appendix
dynamic simulation and thermal analysis tests were performed A contains a complete graphical and mathematical characteriza-
with grain-rened A356 commercial alloy. The chemical compo- tion of a single phase transformation in the semisolid state and
sition, presented at Table 1, was determined with a BILL OES discusses features of the transformation that allow the knee tem-
optical spectrophotometer. Grain renement was achieved by perature, which should be avoided in SSM processing, to be formally
adding an Al5.0 wt%Ti1.0 wt%B master alloy. A356 alloy was cho- dened and determined. Appendix B describes a methodology for
sen because of its broad application in semisolid processing and determining the solidus and liquidus temperatures by extending
because extensive information about its phase transformation tem- the reasoning presented in Appendix A to a complete solidica-
peratures is available in the literature, making validation of the tion/melting reaction so that the semisolid range can be accurately
proposed method easier. identied. Finally, Appendix C describes how other critical points
Thermal analysis to determine the experimental fl vs. T curve used to determine the temperature range for rheocasting and thixo-
for the A356 alloy was performed with a NETZSCH STA 409C DSC forming operations can be identied.
system. The DSC sample was heated to 700 C at 5, 10, 15, 20 and The method proposed here consists of the following steps:
25 C/min and then cooled to room temperature at the same rates.
Heat ow rate (here called HFR) and temperature were monitored (a) Starting with the HFR (usually in mW/mg) vs. T curves gen-
by thermocouples so that heating and cooling HFR vs. T curves could erated from the results of the DSC experiments (the black
be generated. All DSC tests were carried out using the same speci- curves in Fig. 3a, cooling, and b, heating), the corresponding
men to avoid that differences in the chemical composition along the rst derivative with respect to temperature, i.e., the dHFR/dT
billet inuenced the results. Prior to data acquisition, the DSC sam- (in mW/mg C1 ), can be plotted for the complete temperature
ple was heated to 700 C and then cooled to room temperature at range tested, regardless of whether the experiment was per-
the lowest rate (5 C/min) to also guarantee chemical homogeneity formed to map the solidication (cooling cycle rheocasting)
along the specimen. To apply the differentiation method, HFR data or melting (heating cycle thixoforming) phenomenon. The
was derived with respect to temperature, originating the variation rst derivatives correspond to the blue curves in Fig. 3a and b,
of the heat ow rate in relation to temperature along the cooling respectively.
and heating process (dHFR/dT vs. T curves). The fl vs. T relation- (b) By analyzing the behavior of this derivative curve (see

ship was determined with NETZSCH Protheus thermal analysis Appendixes A, B and C), the initial and nal temperatures of the

Table 1
Chemical composition of the A356 alloy (in wt%) at pressure p = 0.1 MPa.a

Si Mg Fe Cu Zn Ti Otherb

6.78 0.14 0.42 0.08 0.35 0.07 0.18 0.04 0.14 0.04 0.23 0.05 < 0.23

a
Maximum of 0.23 wt% other residual elements and Aluminum = Balance.
b
Other residual elements such Cr, Ni, Mn etc.
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 25

Fig. 3. Illustration of the proposed differentiation method. Determination of the solidus, liquidus, eutectic knee and SSM working window using the rst derivative of the
HFR curve for solidication (a) and melting (b) phenomena; partial integration of the HFR curve within the semisolid range for solidication (c) and melting (d) phenomena;
the liquid fraction and its rst derivative, liquid fraction sensitivity, as a function of temperature showing the SSM processing temperature range for both solidication and
melting phenomena (e).

(solidication or melting) reaction, i.e., the liquidus and solidus TSSMI and TSSMF ) suggested earlier, the manageable tempera-
temperatures, can be accurately determined. These are high- ture range in which semisolid processing can be performed is
lighted in red in Fig. 3a and b and correspond to the closest nally dened. This is shown schematically in Fig. 3e as SSM
points to the semisolid transition at which the derivative is zero. process.
The derivative curve also allows the knee, which characterizes
the unstable eutectic transformation and should be avoided in 3.2. Application of the differentiation method to A356 alloy
semisolid processing, to be precisely identied. This point cor-
responds to the rst peak (for cooling) or valley (for heating) 3.2.1. Thermodynamic characterization using CALPHAD
on the derivative curve and is indicated in purple in Fig. 3a and simulation

b. The derivative curve also allows a working window for SSM Thermodynamic simulation using Thermo-Calc under the
processing to be dened. This is the region between the tem- Scheil condition for A356 alloy yielded the curve shown in Fig. 4.
peratures TSSMI (initial) and TSSMF (nal), the two points in green The arrows indicate the formation of the following phases in liq-
in Fig. 3a and b where the derivative curve crosses the baseline uid during solidication: (1) an FCC aluminum primary phase
above the eutectic knee. (L + Al-fcc), which starts to solidify at Tliquidus = 620.8 C; (2) a Si-
(c) Once the solidus and liquidus temperatures have been deter- rich main eutectic (L + Al-fcc + Si), marking the knee at 574.0 C
mined, the curve corresponding to the partial integration of the and fl = 0.48; (3) an AlFeSi Fe-rich -phase (L + Al-fcc + Si + ); (4)
HFR within the semisolid range (the red curve in Fig. 3c and d), an AlFeSi Si-rich phase (L + Al-fcc + Si + + ) and, nally, (5)
i.e., the fl vs. T curve, can be drawn. Then, in possession of the tertiary AlFeMgSi-rich particles, such as Al8 FeMg3 Si6 and Mg2 Si
knee and the suggested temperatures (TSSMI and TSSMF ) for the compounds (L + Al-fcc + Si + + + AlFeMgSi particles). Complete
working window, the corresponding liquid fraction for these solidication occurs at Tsolidus = 538.8 C.
critical points can be determined. The eutectic forms within a temperature range rather than at
(d) Finally, once the solid-to-liquid transition curve has been a well-dened point characteristic of an invariant transformation.
drawn, the derivative of this curve (the green curve in Fig. 3e) This is because of the presence of residual elements, such as Fe, Cu,
can also be drawn. This is the liquid fraction sensitivity curve Ti and Mg, which increase the complexity of the eutectic transition,
(dfl /dT) and shows the region where semisolid processing can extending the temperature range in which it occurs.
be adequately controlled, i.e., where dfl /dT < 0.03 C1 . By com- The formation of tertiary AlFeMgSi compounds in the semisolid
bining this information with the working window (between temperature range does not affect semisolid processing, as this is
26 G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233

higher temperatures). Because of kinetic effects, higher heating


rates result in DSC curves with higher exothermic/endothermic
HFR levels, in agreement with the literature [25]. This is because
transformations that have less time in which to occur (i.e., faster
reactions, corresponding to higher rates) require a larger input of
thermal energy per unit temperature than slower ones, leading to
larger HFR valleys/peaks.
Further tertiary phase transformations previously observed in
the Scheil curve were not detected by the DSC equipment, probably
because of the low HFR generated during the associated pro-
cesses. This in turn is a result of the small mass fraction associated
with these reactions compared with the two main transformations
(Si-eutectic and Al-fcc). As can be seen in Fig. 4, these smaller
reactions represent less than 20% of the total mass transformed
in the semisolid temperature range and are masked by the larger
Fig. 4. fl vs. T curve based on the results of the numerical simulation for A356 alloy
Si-eutectic peak/valley.

(Thermo-Calc under the Scheil condition). The arrows indicate phase transforma- The tangent method was used on the outer edges of the Si-
tions during solidication. eutectic and Al-fcc valleys/peaks in the DSC curves in Fig. 5a and b to
determine the solidus and liquidus temperatures. These are shown
in Table 2 in the columns headed Tan for the heating and cooling
performed above the temperature at which these particles pre- cycles and the ve cooling/heating rates. As an example, the results
cipitate. The amount, distribution, size and morphology of these obtained using this method for the 20 C/min cooling/heating rate
particles are only relevant because of their impact on the mechan- are shown in green for the cooling cycle (Fig. 5a) and light blue for
ical properties of the nal thixoformed or rheocast component, as the heating cycle (Fig. 5b).
they are present in the solidied microstructure. For this reason, The DSC cooling and heating curves differentiated with respect
their effects will not be discussed here. to temperature (dHFR/dT vs. T) for A356 alloy are shown in Fig. 5c
and d, respectively. These were used to identify ve temperatures
3.2.2. Identication of critical points by differentiation of the DSC of relevance in semisolid processing: the solidus, liquidus, eutec-
curve tic knee and upper and lower limits of the working window. The
DSC tests at several cooling and heating rates for A356 alloy solidus and liquidus are shown in blue and correspond to the rst
resulted in the curves shown in Fig. 5a and b, respectively. There and last points at which the curves are zero; the eutectic knee
are two main transformations in the solidication/melting reac- is the rst peak or valley; and the upper and lower limits of the
tion: one related to the Si-rich eutectic, represented by the rst working window, TSSMI and TSSMF , are the rst and second point,
valley/peak (at lower temperatures), and another related to the respectively, after the eutectic knee where the curve crosses the
Al-rich primary phase, represented by the second valley/peak (at horizontal axis.

Fig. 5. HFR vs. T curves (a and b) and dHFR/dT vs. T curves (c and d) for A356 alloy at cooling/heating rates of 5, 10, 15, 20 and 25 C/min. Five temperatures of interest for
semisolid processing were determined by the differentiation method: the solidus, liquidus, knee and upper and lower limits of the working window. Examples of these
temperatures for a 20 C/min cooling/heating rate are highlighted in the gure.
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 27

Table 2
Critical temperatures on the DSC curve and its derivative determined using the tangent method (tan) and differentiation method (DM) for the A356 alloy. The results of the
numerical simulation (under the Scheil condition) are shown as a reference at pressure p = 0.1 MPa.a

Cycle Rate ( C/min) Tsolidus ( C) TKnee ( C) TSSMI ( C) TSSMF ( C) Tliquidus ( C)

Tan DM DM DM DM Tan DM

Thermo-Calc Scheil 538.8 574.0 574.0 617.0 620.8

5 543.0 536.9 561.9 569.4 601.9 609.4 610.4


10 535.4 527.4 560.9 568.5 597.0 605.9 607.9
Cooling 15 520.0 517.8 560.0 567.3 597.5 607.5 610.4
20 519.8 517.9 557.3 567.3 599.8 617.3 620.4
25 512.9 512.9 557.9 562.9 595.4 607.9 610.4

5 563.0 557.5 582.5 590.0 605.0 625.7 627.5


10 562.0 557.5 587.5 600.0 615.0 630.8 635.0
Heating 15 558.5 555.0 590.0 605.0 615.0 633.5 642.5
20 559.6 555.1 587.5 602.5 617.5 635.8 647.5
25 559.5 555.1 597.5 609.5 614.0 647.5 650.0
a
Standard uncertainties u are u(T) = 0.05 C.

The locations corresponding to these temperatures on the respectively, slightly lower and higher than the corresponding
curves in Fig. 5c and d are indicated by arrows, and the correspond- points calculated using the tangent method. Hence, the alternative
ing temperatures for the heating and cooling cycles and the ve method proposed here provides a slightly larger semisolid interval
cooling/heating rates are shown in Table 2 in the column headed than the traditional method. This result is easily explained by the
DM. Again, as an example, the results for a cooling/heating rate inherent differences in the way the extreme semisolid tempera-
of 20 C/min are shown on the differentiated curves (Fig. 5c and tures are measured in each method. In the differentiation method,
d) in dark blue for the cooling cycle and red for the heating cycle. the temperature is determined as soon as the DSC curve changes
The solidus and liquidus are also shown on the original DSC curve its slope, and the start or end of a phase transformation is indicated
(Fig. 5a and b) so that a comparison with the results obtained using by a zero on the differentiation curve (dHFR/dT), while in the tan-
the tangent method can be made. gent method, the temperatures identied by the points at which
the tangents cross over tend to correspond to a delayed start to, or
3.2.3. Identication of the liquidus and solidus earlier end of, the transformation, slightly shortening the semisolid
To compare the tangent and differentiation methods in terms of interval. These differences can be clearly observed in Fig. 5a and b.
the liquidus and solidus identied for A356 alloy using each method Fig. 7 shows the fl vs. T curves for solidication/melting of A365
and shown in Table 2, these critical temperatures are plotted in the alloy after partial area integration of the DSC curves. The solidus and
form of a semisolid range in Fig. 6. Generally, regardless of whether liquidus were identied using the tangent (light blue/green curves)
the tangent or differentiation method is used, the Scheil values (the and differentiation (red/dark blue curves) methods and the values
dashed lines in the graph) lie between the cooling temperatures in Table 2. The black curve shows the variation in fl with T under
(dark blue/green bars) and heating temperatures (red/light blue the Scheil condition, as in Fig. 4. The cooling and heating cycles are
bars) for both the tangent and differentiation methods. Heating clearly divided into two separate groups in the graph, and the curve
temperatures are higher than cooling ones, and a slight increase corresponding to the Scheil condition lies between them.
in the semisolid interval is observed as the heating/cooling rate The relative positions of the curves are explained by the fact that
increases. These trends will be explained later in this section, when the higher the cooling/heating rate (i.e., the farther from equilib-
the graph in Fig. 7 is analyzed. rium conditions the process is), the more the phase transformations
Both methods showed good agreement between their respec-
tive values for each condition, with mean variations (standard
error) of Tsolidus = 4.06 0.72 C and Tliquidus = 4.07 1.09 C. The
solidus and liquidus temperatures determined with the DM are,

Fig. 7. fl vs. T curves produced by integration of partial areas under cooling/heating


DSC curves with respect to temperature for A356 alloy at 5, 10, 15, 20 and 25 C/min
Fig. 6. Solidus and liquidus temperatures for A356 alloy measured using the tangent cooling/heating rates using the tangent and differentiation methods to identify the
and differentiation methods at ve heating/cooling rates. The values predicted by solidus and liquidus points. The curve predicted by numerical simulation (Scheil
numerical simulation (Scheil condition) are shown as a reference. condition) is shown as a reference.
28 G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233

are delayed and extended in the temperature range, i.e., at higher Table 3
Knee parameters obtained by the differentiation (DM) and visual recognition (Vis)
cooling/heating rates the imposed heat input is not followed by
methods for A356 alloy at several cooling/heating rates at pressure p = 0.1 MPa. Val-
the diffusion processes responsible for phase transformations, so ues that could not be determined are shown as NI. Values predicted by numerical
these are forced to occur over a wider temperature range than simulation (Scheil condition) are shown as a reference.a
that predicted for equilibrium conditions. This effect is amplied as
Cycle Rate ( C/min) TKnee ( C) fl Knee
the cooling/heating rate increases. The result is that cooling curves
move toward lower temperatures (left) and heating curves toward Vis DM Vis DM
higher temperatures (right) as the cooling/heating rate increases. Thermo-Calc

Scheil 574.0 0.48
Hence, the outer and inner edges of each group are progressively
5 563.9 561.9 0.50 0.45
farther from and closer to equilibrium, respectively. Between the 10 560.5 560.9 0.43 0.43
curves corresponding to the lowest cooling/heating rates (5 C/min) Cooling 15 560.4 560.0 0.47 0.47
lies the Scheil curve, which is nearer equilibrium than any of the 20 558.4 557.3 0.44 0.43
25 556.0 557.9 0.48 0.50
experimental curves. In addition to causing a delay in the semisolid
transformations, higher rates result in a larger driving force (super- 5 582.0 582.5 0.55 0.56
cooling/superheating) for solidication/melting, making them start 10 588.4 587.5 0.54 0.51
Heating 15 NI 590.0 NI 0.51
sooner [23]. Consequently, faster cycles have a larger semisolid
20 NI 587.5 NI 0.43
interval. Faster cooling curves have higher liquidus points than 25 NI 597.5 NI 0.54
slower ones, and a similar trend is observed for heating curves, a
Standard uncertainties u are u(T) = 0.05 C and u(fl ) = 0.01.
with faster heating curves having slightly lower solidus tempera-
tures. This explains the trends shown earlier in Fig. 6. For heating
curves, the decrease in solidus temperature is difcult to see in operator-dependent. For the 15, 20 and 25 C/min heating curves,
Fig. 7 due to the reduced extend of the difference (4 C) compared the knee could not be identied and the corresponding values are
to the magnication of the graph, that has a x-axis scale ranging therefore missing in their respective elds in Table 2 (indicated as
from 520 to 650 C (T = 130 C) in the limited page space. Further- NI, not identiable) and in Fig. 8. This problem does not arise
more, the separation of the lower limit for these specic curves with the differentiation method, as it uses mathematical/graphical
occurs at extremely low liquid fractions, in which the curves are procedures to identify the knee that do not change even when it is
practically horizontal and join together. However, this trend can be imperceptible to the human eye.
clearly seen in Table 1. In Fig. 8, as in Fig. 7, the knee is at a higher temperature in heating
Another feature of the curves in Fig. 7, especially in heating cycles (red points) than in cooling ones (dark blue points), and the
cycles, is that the faster the rate, the softer the knee shape of the values corresponding to the Scheil condition lie in between (dashed
Si-eutectic transformation. This smoothing is a result of the effect lines). The reasons for this have already been discussed. Analysis of
mentioned above whereby the farther the process is from equi- the results where a knee could be identied by the visual recogni-
librium conditions, the greater the temperature range over which tion method, i.e., excluding the 15, 20 and 25 C/min heating curves,
phase transformations occur. revealed that there was good agreement between both methods
Finally, it can be observed that the curves generated using each and that the mean difference (and standard error) in the knee tem-
method for a particular condition are very similar, indicating that perature and knee liquid fraction between the two methods was
the choice of method (differentiation vs. tangent) does not signif- only Tknee = 0.22 0.4 C and fl (knee) = 0.0083 0.008.
icantly affect the shape of the curve. Hence, the results for both Therefore, the differentiation method proved to be (a) suitable
methods are coherent. for accurately identifying the knee point and in good agreement
In summary, the differentiation method proved to be (a) able with the visual recognition method; (b) operator-independent, as
to accurately identify the solidus and liquidus temperatures, as it it uses the same mathematical/graphical procedure to identify the
yielded results that were in good agreement with those obtained by knee and therefore does not require common sense to identify
the tangent method; (b) operator-free, as it uses the same math- a knee-like shape in the curve, as a result of which measure-
ematical/graphical procedure to identify the solidus and liquidus ment inaccuracies are avoided; (c) technically easier to use than
points, eliminating the need for common sense when choosing the the visual recognition method, as the knee is represented by a
most appropriate tangents so that measurement inaccuracies are
avoided and (c) technically easier to use than the tangent method,
as the solidus and liquidus points are represented by points where
the curve being analyzed is equal to zero instead of changes in the
slope of the curve (features that may be barely identiable in some
situations).

3.2.4. Identication of the knee


To analyze the effectiveness of the differentiation method in
identifying the unstable Si-eutectic transformation temperature for
A356 alloy, the standard visual recognition method for identifying
the knee was used with the red and dark blue fl vs. T curves in
Fig. 7. The values of temperature and liquid fraction determined in
this way are shown in the Vis (Visual) columns in Table 3. The
values of knee temperature obtained earlier from the derivatives
of the DSC curves (Tknee column in Table 2) were used with the
integrated fl vs. T curves to obtain the corresponding knee liquid
fraction shown in the last DM column of Table 2. The temperature
and liquid fraction data in Table 3 is shown graphically in Fig. 8.
Fig. 8. Knee temperature and liquid fraction determined by the visual recognition
The knee for the 10 C/min heating curve could only be identied and differentiation methods for A356 alloy at several cooling/heating rates. Values
very imprecisely by the visual recognition method and was highly predicted by numerical simulation (Scheil condition) are shown as a reference.
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 29

Fig. 9. Comparative behavior of dHFR/dT vs. T (dark blue), fl vs. T (bright blue) and dfl /dT vs. T (light blue) curves derived from DSC data for semisolid transformation of A356
alloy at cooling rates of 5 (b), 10 (c), 15 (d), 20 (e) and 25 (f) C/min. The values predicted by numerical simulation (Scheil condition) are shown at the top-left as a reference
(a).

valley/peak in the curve being analyzed instead of an unstable knee- an overview of the graphical behavior as the semisolid transfor-
like shape and (d) much better able to identify the knee than the mation progresses. The Scheil condition is shown as a reference in
visual recognition method, as seen for the 15, 20 and 25 C/min Figs. 9a and 10a (black curve). The ve critical temperatures identi-
heating curves. ed earlier by differentiating the DSC curves (shown in the solidus,
For some authors, identication of the knee becomes less knee, TSSMI , TSSMF and liquidus columns in Table 1) are indicated on
important for thixoformability analysis as the cooling/heating rate the curves for all the conditions analyzed.
increases due to the smoothening of the knee-like shape caused To establish an adequate working window for SSM processing of
by kinetic effects. As the transformation at these higher rates does A356 alloy by the differentiation method, as described in Appendix
not occur at a specic point but over a progressively broader tem- C of this work, the maximum sensitivity (dfl /dTmax ) in the proposed
perature range, identication can become meaningless in practical interval (between TSSMI and TSSMF ) was read on the graph to assess
terms. Nevertheless, the method proposed here can still be use- the thermodynamic stability. The liquid fractions at the boundaries
ful for intermediate cooling/heating rates, for which the knee is of the proposed interval (fl SSMI and fl SSMF ) were also identied so
still present although only roughly identiable, as seen for the that the range of liquid fractions during the process was known.
10 C/min heating condition in this study. The method can also be These data are shown in Table 4 for all the conditions analyzed
used for further comparative analysis, e.g., to gain a better under- and are highlighted for illustrative purposes on the curves for the
standing of how kinetic effects inuence knee temperature under 20 C/min cooling and heating cycles in Figs. 9 and 10.
different heating/cooling conditions. Apart from the 10 C/min cooling condition, all the sensitiv-
ity values in Table 4 are below the 0.03 C1 limit, so the original
3.2.5. Determining the semisolid processing range interval suggested by the differentiation method was maintained.
The dHFR/dT vs. T curves and the resulting fl vs. T and dfl /dT vs. In the case of the 10 C/min cooling cycle, the upper limit of the
T (sensitivity) curves for A356 alloy for all the conditions analyzed working window (the region of maximum sensitivity) was low-
here are shown in Fig. 9 (cooling) and Fig. 10 (heating) and provide ered to the next temperature in the interval at which the sensitivity
30 G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233

Fig. 10. Comparative behavior of dHFR/dT vs. T (dark red), fl vs. T (bright red) and dfl /dT vs. T (light red) curves plotted using DSC data for semisolid transformation of A356
alloy at heating rates of 5 (b), 10 (c), 15 (d), 20 (e) and 25 (f) C/min. The values predicted by numerical simulation (Scheil condition) are shown at the top-left as a reference
(a).

satised the 0.03 C1 criterion. The resulting interval was con- alloy for all the cooling and heating cycles. The results are shown in
sidered for SSM processing and had the following corrected Fig. 11a and b in terms of temperature and liquid fraction, respec-
parameters: TSSMF = 596 C, Tcooling = 27.5 C, fl SSMF cooling = 0.83 tively.
and fl SSM cooling = 0.32 and dfl /dTmax = 0.029 C1 . The data in Three features related to the temperature range for SSM
Table 3 were then used to plot the working windows for A356 processing can be observed in Fig. 11a:

Table 4
Parameters for the working window for SSM processing of A356 alloy determined by the differentiation method for ve cooling/heating rates at pressure p = 0.1 MPa. The
values predicted by numerical simulation (Scheil condition) are shown as a reference.a

Cycle Rate ( C/min) TSSMI ( C) TSSMF ( C) TSSM ( C) fl SSMI fl SSMF fl SSM dfl /dTmax ( C1 )

Thermo-Calc Scheil 574.0 617.0 43.0 0.48 0.92 0.44 0.019

5 569.4 601.9 32.5 0.55 0.92 0.37 0.026


10 568.5 597.0 28.5 0.51 0.86 0.35 0.032
Cooling (rheocasting) 15 567.3 597.5 30.2 0.54 0.87 0.33 0.017
20 567.3 599.8 32.5 0.52 0.88 0.36 0.016
25 562.9 595.4 32.5 0.55 0.91 0.36 0.016

5 590.0 605.0 15.0 0.65 0.84 0.19 0.014


10 600.0 615.0 15.0 0.69 0.88 0.19 0.014
Heating (thixoforming) 15 605.0 615.0 10.0 0.74 0.87 0.13 0.014
20 602.5 617.5 15.0 0.67 0.88 0.21 0.015
25 609.5 614.0 4.5 0.75 0.82 0.07 0.016
a
Standard uncertainties u are u(T) = 0.05 C, u(fl ) = 0.01 and u(dfl /dT) = 0.005 C1 .
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 31

Fig. 11. Working window for SSM processing obtained by the differentiation method for A356 alloy at different cooling (rheocasting) and heating (thixoforming) rates:
temperature range (a) and liquid fraction range (b). The values predicted by numerical simulation (Scheil condition) are shown as a reference.

(a) In cooling cycles the working window temperatures for SSM the conditions analyzed: TSSM cooling = 31.0 0.9 C and TSSM

processing are lower than the temperatures for heating cycles. This heating = 11.9 2.0 C (mean standard error). This result shows
is a result of the kinetic effects already discussed in Section 3.2.3. another thermokinetic advantage of rheocasting over thixoform-
Cooling and heating curves move toward lower and higher temper- ing: the former is more controllable, as the semisolid material is
atures, respectively, as the cooling/heating rate increases because stable over a wider temperature range. It also suggests that the
of the delayed semisolid phase transformations and greater tem- working window for thixoforming processes should be dened to
perature range over which the transformations occur. In practical extend over a wider temperature range when kinetic effects are
terms, this result shows that rheocasting operations should be per- signicant.
formed at lower temperatures than thixoforming operations to Two of the three results discussed here related to the SSM
achieve more controllable processing conditions. In fact the authors processing temperature range are also reected in the SSM liq-
believe that this particular nding is not a unexpected result, since uid fraction range shown in Fig. 11b. Because the SSM temperature
the solidication phenomena is time dependent due to diffusion range for cooling cycles is wider than for heating cycles (Fig. 11a),
as well as segregation of Silicon and the residual elements into the corresponding liquid fraction ranges are also wider, as shown in
the liquid, i.e., the solidication of the primary Alfcc and solid solu- Fig. 11b: fl SSM cooling = 0.35 0.01 and fl SSM heating = 0.16 0.026
tion depends upon segregation. The melting phenomena is less (mean standard error). Similarly, because the SSM temperature
time dependent, since all the silicon particles as well as most of the range remains almost unchanged between cooling cycles (Fig. 11a),
residual elements are already conned in the secondary (eutectic) the liquid fraction range for cooling also remains essentially
volume of the material, therefore the TSSMI that marks the nal unchanged (Fig. 11b), varying by only 0.05 among all the cycles
eutectic melting is expected to be higher when the primary Alfcc (from 0.32 to 0.37). In contrast, the liquid fraction for heating cycles
and solid solution starts to melt. varies by 0.12 (from 0.07 to 0.19). As stated for the temperature
(b) Another feature related to kinetic effects is that the work- range analysis, these results also show that rheocasting operations
ing window for SSM processing remains almost unaltered as the are superior to thixoforming operations in terms of process control
cooling rate increases, varying by only 5 C (27.532.5 C) between parameters.
the smallest and largest window; however, the window generally However, although the SSM working window temperatures are
becomes smaller as the heating rate increases, varying by 10.5 C lower for cooling cycles than for heating cycles, the SSM liquid
(4.515.0 C) between the highest and lowest heating rate. This fraction ranges overlap. For rheocasting the liquid fraction varies
behavior shows that the melting phenomenon is more affected by between 0.55 and 0.92, and for thixoforming it varies within a nar-
kinetic effects than the solidication one. A practical consequence rower range (0.650.88) contained within this interval. This result
of this is that rheocasting processes are more stable than thixoform- indicates that because of the thermokinetic behavior of the melt-
ing ones for processing conditions farther from thermodynamic ing and solidication phenomena, their critical points are in similar
equilibrium. regions of reacted mass (liquid or solid fraction) even though the
The signicant reduction in the working window for heat- transformations occur in different temperature ranges.
ing cycles (especially for a heating rate of 25 C/min, for which For the purposes of comparison, Table 5 shows the working
TSSM = 4.5 C) indicates that the delayed semisolid transforma- windows for A356 alloy for all the cooling and heating condi-
tions and greater temperature range over which these occur tions analyzed using three different approaches: the differentiation
because of kinetic effects is more marked for eutectic melting than method and the criteria proposed by Liu et al. [19] and Zoqui et al.
for -phase melting, i.e., TSSMI is more delayed than TSSMF , making [21], which are described in Section 1 of this work.
the interval between them narrower as the heating rate increases. The traditional denition of the working window proposed by
Thus, the working window for higher heating rates should perhaps Liu et al. does not consider the thermodynamic behavior of individ-
be dened so as to extend over a wider range by raising TSSMF , ual alloys in the semisolid state (which is strongly affected by the
since the progressive smoothing of the transformation curve as composition of the alloy and kinetic factors) and limits the process
the heating rate increases reduces the importance of identifying to a dened fl of 0.40.6. The result is that for all the conditions
critical points. This issue was discussed earlier for the knee temper- analyzed the unstable eutectic knee lies within the chosen work-
ature (Section 3.2.4) and arises because the transformations occur ing window, which is not recommended for various reasons already
in a progressively wider temperature range as the kinetic effects discussed in this work (working windows containing the knee are
intensify. highlighted in bold in Table 4). Furthermore, some conditions are
(c) A further feature of Fig. 11a is that the SSM processing discarded merely because the sensitivity is greater than 0.03 C1
temperature range for cooling is wider than for heating for all at fl = 0.5. Examples of this are the 5 C/min cooling and heating
32 G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233

Table 5
SSM working windows obtained by the differentiation method and the criteria proposed by Liu et al. [19] and Zoqui et al. [21] for A356 alloy at several cooling/heating rates
at pressure p = 0.1 MPa. Bold values represent intervals containing the unstable eutectic knee. Empty boxes represent conditions for which dfl /dT(0.5) > 0.03 C1 .a

Cycle Rate ( C/min) SSM working window

TLiu ( C) TZoqui ( C) TDIF ( C) fl Liu fl Zoqui fl DIF

5 565610 569602 0.51.0 0.50.9


10 559581 561595 568597 0.40.8 0.50.9
Cooling (rheocasting) 15 556575 560610 567597 0.51.0 0.50.9
20 556578 557620 567600 0.41.0 0.50.9
25 552570 558610 563595 0.51.0 0.50.9
0.40.6
5 582627 590605 0.51.0 0.60.8
10 584592 587635 600615 0.51.0 0.70.9
Heating (thixoforming) 15 586594 590642 605615 0.51.0 0.70.9
20 586597 587647 602617 0.41.0 0.70.9
25 592601 597650 609614 0.51.0 0.70.8
a
Standard uncertainties u are u(T) = 0.05 C, u(fl ) = 0.01 and u(dfl /dT) = 0.005 C1 .

cycles, for which the processing intervals are not shown in Table 4. inherent aspects of SSM processing. The scope of this work is
By using this criterion, the entire stable region above the knee for limited to a thermodynamic characterization and for this reason
these two cooling and heating cycles is simply ignored, showing other (but not less important) points concerning the applicability
that this method may not evaluate the full potential of a candidate of new alloys to SSM processing are not discussed and are strongly
alloy or kinetic condition for SSM processing. suggested as future research.
The criteria proposed by Zoqui et al. signicantly increase the
number of alloys that could potentially be considered for SSM 4. Conclusions
processing by using only the position of the eutectic knee as a
constraint (lower limit) on the working window and by analyzing An alternative method for identifying critical points in thixo-
the sensitivity throughout the entire interval (dfl /dT < 0.03 C1 ). formability analysis and a formal denition for the knee point
The processing window with this approach is the widest of the have been proposed. The method can be used with solidi-
three windows determined with the approaches considered here cation and melting transformations. A new SSM processing
and generally extends from the knee to the liquidus temperature. window has been suggested for both rheocasting and thixoforming
However, it allows SSM processing to take place near the region of operations.
the curve where eutectic particles are still present in the semisolid The ndings indicate that the proposed method is an effective
slurry and the liquid fraction is extremely high, potentially having tool for identifying critical points such as the solidus, liquidus and
an adverse effect on the slurry rheology during processing as well knee temperatures. The results agree well with those obtained
as on the microstructure of the nal product. using traditional methods. The differences between the solidus
The differentiation method seeks to improve on the approach and liquidus temperatures measured with the proposed method
proposed by Zoqui et al. by narrowing the working window. It does and with the tangent method were Tsolidus = 4.06 0.72 C
this by moving the lower boundary of the window farther from the and Tliquidus = 4.07 1.09 C, and the differences between the
unstable eutectic knee and the upper boundary farther from the liq- knee temperatures and liquid fractions measured with the pro-
uidus, decreasing the likelihood of eutectic particles being present posed method and with the visual recognition method were
in the slurry and avoiding an excessive amount of liquid during Tknee = 0.22 0.4 C and fl knee = 0.0083 000.8. The proposed
SSM processing. Despite these differences, both the differentiation method also proved to be operator-independent, as it uses well-
method and the approach proposed by Zoqui et al. consider the dened mathematical/graphical criteria to identify critical points,
characteristic thermodynamic behavior of a particular alloy under thereby avoiding measurement inaccuracies. Furthermore, it is eas-
specic kinetic conditions and not only are more compatible with ier to use than other methods, as critical points are turned into
real SSM operations but also allow more commercial alloys to be easily recognizable visual and numerical features of the curve being
used in SSM processing. Finally, although it predicts a small working analyzed after second-order differentiation, such as valleys, peaks
window for thixoforming operations, the differentiation method is and zeros, instead of slight changes in slope or knee-like shapes,
suitable for use with both rheocasting and thixoforming operations which may be barely identiable in some situations, especially
as in practice in a commercial environment it is easier to heat billets when kinetic effects are more signicant.
for semisolid processing than to cool material in order to rheocast The suggested formal denition for the knee point (the
it. peak/valley in the expected eutectic range of the differentiated DSC
A remark should be made concerning the use of the suggested curve of the solidication/melting transformation) yields results that
working window (presented in Appendix C) for real forming oper- agree well with those obtained using traditional approaches, indi-
ations in the semisolid state. It is known that in addition to cating the suitability of this denition for identication of the knee,
thixoformability, several factors play a determining role in SSM particularly when the aim is to compare different shaped curves.
processing, such as microstructure and rheology of the semisolid The proposed working window allows suitable lower and upper
slurry as well as further processing limitations such as hetero- limits for SSM processing to be identied.
geneous temperature and chemical composition of the billet.
Furthermore, specic characteristics and issues associated to rheo- Acknowledgments
casting and thixoforming operations raise signicant practical
differences between these two operations. The thermodynamic The authors would like to thank the Brazilian research fund-
analysis presented here is only the initial step for the character- ing agencies FAPESP (So Paulo Research Foundation Projects
ization of candidate alloys for SSM process and an adaptation of 2013/09961-3 and 2015/22143-3), CNPq (National Council for Sci-
the suggested working window should be performed in order to entic and Technological Development Project 470572/2011-6)
adequate the observed thermodynamic features to these other and CAPES (Federal Agency for the Support and Improvement of
G.L. Brollo et al. / Thermochimica Acta 651 (2017) 2233 33

Higher Education) for providing nancial support for this study. The [12] O. Lashkari, R. Ghomashchi, The implication of rheology in semisolid metal
authors are also indebted to the Faculty of Mechanical Engineering processes: an overview, J. Mater. Process. Technol. 182 (2007) 229240,
http://dx.doi.org/10.1016/j.jmatprotec.2006.08.003.
at the University of Campinas for the practical support very kindly [13] M. Perez, J.-C. Barb, Z. Neda, Y. Brchet, L. Salvo, Computer simulation of the
provided. microstructure and rheology of semisolid alloys under shear, Acta Mater. 48
(2000) 37733782, http://dx.doi.org/10.1016/s1359-6454(00)00161-0.
[14] M. Paes, E.J. Zoqui, Semisolid behavior of new AlSiMg alloys for
Appendix A. Supplementary data thixoforming, Mater. Sci. Eng. A 406 (2005) 6373, http://dx.doi.org/10.1016/
j.msea.2005.07.01.
[15] E. Tzimas, A. Zavaliangos, Materials selection for semisolid processing, Mater.
Supplementary data associated with this article can be found, in Manuf. Proc. 14 (2) (1999) 217230, http://dx.doi.org/10.1080/
the online version, at http://dx.doi.org/10.1016/j.tca.2017.02.010. 10426919908914819.
[16] E. Tzimas, A. Zavaliangos, Evaluation of volume fraction of solid in alloys
formed by semisolid processing, J. Mater. Sci. 35 (21) (2000) 53195330,
References http://dx.doi.org/10.1023/A:1004890711322.
[17] Q. Han, S. Viswanathan, The use of thermodynamic simulation for the
[1] D.H. Kirkwood, M. Sury, P. Kapranos, H.V. Atkinson, K.P. Young, Semisolid selection of hypoeutectic aluminumsilicon alloys for semisolid metal
Processing of Alloys, vol. 124, Springer, 2010, pp. 172, http://dx.doi.org/10. processing, Mater. Sci. Eng. A364 (2004) 4854, http://dx.doi.org/10.1016/j.
1007/978-3-642-00706-4. msea.2003.06.002.
[2] M.C. Flemings, Behavior of metal alloys in the semisolid state, Metall. Trans. A [18] S.G. Irizalp, N. Saklakoglu, Effect of Fe-rich intermetallics on the
22A (1991) 957981, http://dx.doi.org/10.1007/bf02651227. microstructure and mechanical properties of thixoformed A380 aluminum
[3] E.J. Zoqui, Alloys for semisolid processing, in: J. McGeough (Ed.), alloy, Eng. Sci. Technol., Int. J. 17 (2014) 58e62, http://dx.doi.org/10.1016/j.
Comprehensive Materials Processing, vol. 5, Elsevier Ltd., 2014, pp. 163190, jestch.2014.03.006.
http://dx.doi.org/10.1016/B978-0-08-096532-1.00520-3. [19] D. Liu, H.V. Atkinson, H. Jones, Thermodynamic prediction of thixoformability
[4] H.V. Atkinson, Modelling the semisolid processing of metallic alloys, Prog. in alloys based on the AlSiCu and AlSiCuMg systems, Acta Mater. 53
Mater. Sci. 50 (2005) 341412, http://dx.doi.org/10.1002/chin.200543242. (2005) 38073819, http://dx.doi.org/10.1016/j.actamat.2005.04.028.
[5] E.J. Zoqui, J.I. Gracciolli, L.A. Lourencato, Thixo-formability of the AA6063 [20] P.P.J. Uggowitzer, D.I. Uhlenhaut, Metallurgical aspects of SSM processing, in:
alloy: conventional production processes versus electromagnetic stirring, J. G. Hirt, R. Kopp (Eds.), Thixoforming-Semi-Solid Metal Process, Wiley-VCH,
Mater. Process. Technol. 198 (2008) 155161, http://dx.doi.org/10.1016/j. 2009, pp. 2942, http://dx.doi.org/10.1002/9783527623969.
jmatprotec.2007.06.089. [21] E.J. Zoqui, D.M. Benati, C.T.W. Proni, L.V. Torres, Thermodynamic evaluation of
[6] K. Sukumaran, B.C. Pai, M. Chakraborty, The effect of isothermal mechanical the thixoformability of AlSi alloys, CALPHAD: Comput. Coupling Phase Diagr.
stirring on an AlSi alloy in the semisolid condition, Mater. Sci. Eng. A369 Thermochem. 52 (2016) 98109, http://dx.doi.org/10.1016/j.calphad.2015.12.
(2004) 275283, http://dx.doi.org/10.1016/j.msea.2003.11.036. 006.
[7] L. Sang-Yong, L. Jung-Hwan, L. Young-Seon, Characterization of Al 7075 alloys [22] K. Chang, S. Liu, D. Zhao, Y. Du, L. Zhou, L. Chen, Thermodynamic description
after cold working and heating in the semisolid temperature range, J. Mater. of the AlCuMgMnSi quinary system and its application to solidication
Process. Technol. 111 (2001) 4247, http://dx.doi.org/10.1016/s0924- simulation, Thermochim. Acta 512 (2011) 258267, http://dx.doi.org/10.
0136(01)00494-0. 1016/j.tca.2010.11.009.
[8] P.K. Seo, C.G. Kang, The effect of raw material fabrication process on [23] S. Nasi, D. Emadi, R. Ghomashchi, Semi solid metal processing: the fraction
microstructural characteristics in reheating process for semisolid forming, J. solid dilemma, Mater. Sci. Eng. A 507 (2009) 8792, http://dx.doi.org/10.
Mater. Process. Technol. 162163 (2005) 402409, http://dx.doi.org/10.1016/ 1016/j.msea.2008.11.052.
j.jmatprotec.2005.02.012. [24] X. Yan, F. Xie, M. Chu, Y.A. Chang, Microsegregation in Al4.5Cu wt.% alloy:
[9] E. Tzimas, A. Zavaliangos, Evolution of near-equiaxed microstructure in the experimental investigation and numerical modeling, Mater. Sci. Eng. A302
semisolid state, Mater. Sci. Eng. A289 (2000) 228240, http://dx.doi.org/10. (2001) 268274, http://dx.doi.org/10.1016/s0921-5093(00)01825-6.
1016/s0921-5093(00)00908-4. [25] G.W.H. Hhne, W.F. Hemminger, H.-J. Flammershein, Differential Scanning
[10] Y. Chen, J. Wei, Y. Zhao, J. Zheng, Microstructure evolution and grain growth Calorimetry, Springer Ltd., 2003, pp. 298, http://dx.doi.org/10.1007/978-3-
behavior of Ti14 alloy during semisolid isothermal process, Trans. Nonferrous 662-06710-9.
Met. Soc. China 21 (2011) 10181022, http://dx.doi.org/10.1016/s1003- [26] Y. Birol, Solid fraction analysis with DSC in semisolid metal processing, J.
6326(11)60815-7. Alloys Compd. 486 (2009) 173177, http://dx.doi.org/10.1016/j.jallcom.2009.
[11] M. Reisi, B. Niroumand, Growth of primary particles during secondary cooling 06.165.
of a rheocast alloy, J. Alloys Compd. 475 (2009) 643647, http://dx.doi.org/10. [27] J.H. Flynn, Analysis of DSC results by integration, Thermochim. Acta 217
1016/j.jallcom.2008.07.090. (1993) 29149, http://dx.doi.org/10.1016/0040-6031(93)85104-H.

S-ar putea să vă placă și