JO U RN A L O F M AT E RI AL S S CI EN C E 3 3 (1 99 8 ) 2 2 6 9 —228 1

Effect of inclusions on the tensile properties of

Al–7% Si–0.35% Mg (A356.2) aluminium casting
alloy
L. LI U , F. H. SAM UE L
Département des Sciences appliquées, Université du Québec à Chicoutimi,
555, boul. de l’Université, Chicoutimi(Québec), Canada G7H 4A4

The present study was performed on an A356.2 alloy. Two types of initial materials were
used, i.e. fresh and recycled. A total of 13 operations representing those normally applied in
aluminium foundries were simulated under dry atmospheric conditions (humidity
\15%—20%). The molten metal was cast into test bars which were T6 tempered prior to
tensile testing. The results show that holding the liquid metal for a long time, i.e. 72 h at
735 °C leads to sedimentation of most inclusions towards the bottom of the melting crucible.
However, a change in the surrounding humidity may cause absorption of hydrogen and,
hence, a large amount of porosity. Degassing using dry argon injected into the liquid metal
through a rotary impeller (speed \160 r.p.m) appears to be the best technique for inclusion
removal. The efficiency of this process is significantly improved when it is coupled with
filtration using ceramic foam filters (10 and 20 p.p.i). A linear relationship between alloy
ductility and logarithm of percentage inclusions has been established. Owing to decohesion
between the inclusions/oxide films and the surrounding matrix, cracks are easily initiated at
their interfaces, leading to unpredicted failure.  1998 Chapman & Hall

1. Introduction of inclusions/oxides, and can be expressed by the rela-

The detrimental effects of the presence of inclusions in tionship
aluminium are well documented [1—8]. Although
% elongation"1.06!0.66X (1)
much work has been reported on the study of inclu-
sions, the problem of measuring metal cleanliness where X represents the volume fraction of inclusions.
through a sensitive quantitative method still exists. Campbell [10] has reported that the oxide films
The difficulty is further compounded by the fact that that remain in the cast product will cause mechanical
the inclusions to be measured are usually very small, weakness and form leak paths that emerge through
about 10—20 lm in diameter, and present in trace the walls of the casting. In some cases, the large
amounts (about 10 p.p.m). The main inclusions that amount of porosity associated with the oxide films can
occur during the melting of aluminium alloys or hold- further enhance the weakness. Fractographic observa-
ing periods prior to casting are aluminium oxide tions reveal that crack initiation is related to the
(Al O ) as dispersed particles or oxide films, alumi- presence of these inclusions. Spinel inclusions are of-
2 3
nium carbide (Al C ), magnesium oxide (MgO), spinel ten seen to be surrounded by partial or complete
4 3
(MgAl O ), titanium diboride (TiB ), aluminium bor- voids, indicating debonding between the inclusions
2 4 2
ide (AlB), and titanium aluminide (TiAl ) [1—8]. and the matrix during fracture.
3
Many studies have been conducted on the role of Hedjazi et al. [11] examined the effect of non-metal-
non-metallic inclusions and oxide films in determining lic inclusions on the tensile properties of an Al—4.5%
the alloy mechanical properties. The presence of oxide Cu—1.5% Mg alloy. Their study showed that inclu-
films can lead to a significant scatter in the tensile sions which arise as a result of various physical and
properties of the cast alloy due to early initiation of chemical phenomena that occur during the melting
fracture. Removal of such inclusions/oxides is ex- and casting process were found to be non-uniformly
pected to result in more consistent and reproducible distributed within the cast product. Categorizing the
mechanical properties. inclusions/oxides as two types, namely, film and non-
Samuel et al. [9] studied the effect of melt cleanli- film, and further classifying the latter type according
ness on the mechanical properties of a 359 aluminium to their size as macro-inclusions ('50 lm) and
alloy reinforced with 20% SiC particles (average par- micro-inclusions ((50 lm), the authors concluded
ticle size \20 lm). Their work shows that the percent- that both film and non-film inclusions lower the duc-
age elongation varies linearly with the volume fraction tility, i.e. percentage elongation, to a greater extent
0022—2461 ( 1998 Chapman & Hall 2269
than the strength of the alloy. In their opinion, filtra- treated for 8 h at 540 °C, in a forced-air furnace, fol-
tion improves with bed depth as do the tensile proper- lowed by quenching in hot water (60 °C). The quen-
ties [12]. ched test bars were stabilized at room temperature for
Arai et al. [13] have observed that molten metal 24 h prior to hardening at 150 °C for 5 h.
filtration increases the strengths of 2014, 2017 and The solution treated test bars were pulled to frac-
7075 extrusions. The reason is a decrease in the non- ture at room temperature in an Instron Universal
metallic inclusions which disturb homogeneous pre- testing machine at a strain rate of 4]10~4 s~1.
cipitation. Ductilities of extrusions also increase by A strain gauge extensometer (50 mm range) was at-
molten metal filtration. The non-metallic inclusions tached to the test specimen for measuring the alloy
act as nucleation sites for dimple cleavages [13]. Rios ductility. The data were analysed using a special soft-
et al. [14] attribute the lower ductility and strength in ware designed for this work. Tensile properties, name-
the transverse direction (deformation axis normal to ly yield stress (YS) at 0.2% offset strain, ultimate
extrusion axis and tube radius) in hot-extruded tensile strength (UTS), and fracture elongation (EL%)
AA2014 aluminium tube specimens to the influence of were derived from the data aquisition system. The
coarse inclusions that fracture early during tensile volume fraction of non-metallic inclusions was meas-
deformation. The resultant cracks induce void forma- ured from macrophotographs of the specimen fracture
tion around the small inclusions in the matrix. surfaces, using the grid method.
The role of inclusions in initiating fatigue cracks
under dynamic loading is well documented. Chien
et al. [15] report that TiAl particles nucleate cracks 3. Results and discussion
3
in A201 alloys. It is expected that other types of 3.1. Fresh alloy
inclusions may act in the same way, particularly if The experimental conditions applied in the present
their size is large. Gerold [16] also found that in work are summarized in Table II, while tensile test
commercial alloys, intermetallic inclusions are addi- data are listed in Tables III—X.
tional sources for crack initiation. Thus, filtered metal It can be seen from Table III that a settling time of
should produce castings having an increased fatigue 72 h enhances the alloy properties, in particular, per-
resistance. centage elongation, which reaches almost 18%. It is
The present paper documents the tensile test results important to note the uniformity of the tensile proper-
obtained from a study of Al—7% Si—0.35% Mg ties of all castings regardless of the casting sequence.
(A356.2) alloy, where a total of 13 operations, typical Vigorous stirring appears to push the inclusions that
of those customarily used in the aluminium industry, settle close to the bottom of the crucible towards the
were applied to the alloy melts, to investigate the type upper surface of the molten metal. Depending upon
and amount of inclusions that could occur as a result the stirring speed and time, the thickness of the con-
of these operations, and their influence on the tensile taminated portion near the melt surface increases as
properties, in particular, the alloy ductility. shown in Table IV, where the first six castings exhibit
low ductility.
Degassing, on the other hand, is very effective in
2. Experimental procedure removing most of the inclusions and results in
The A356.2 alloy used in the present study was sup- a marked improvement in the tensile properties
plied by Guelph Foundry Works, Alcan International (12%—20% EL) for most of the castings, Table V.
Limited in the form of 12.5 kg ingots. The chemical However, if the inclusions in the upper portion of the
composition of the as-received alloy is shown in Table liquid metal following the degassing are not skimmed
I. The ingots were cut, cleaned, and melted in a silicon immediately, these inclusions (with heavier densities
carbide crucible of 28 kg capacity, using an electrical compared to that of liquid aluminium) tend to settle
resistance furnace. The melting temperature was held down towards the bottom of the crucible. Depending
at 735$5 °C. The maximum humidity was about upon the settling time, the molten metal will become
15%. No degassing was applied. The hydrogen level contaminated again, leading to unpredictable results.
was )0.1 ml/100 g Al as measured by an AlscanTM Table VI shows that in some cases, when the settling
apparatus. Prior to casting, melt-treatment agents time after degassing is large, e.g. 4 h, two test bars
such as strontium (400 p.p.m. Sr, added as Al—10% Sr produced simultaneously from the same casting can
master alloy) or TiB (in the form of Al—5% Ti—1% B, possess completely different ductilities, the latter being
2
measured in terms of a 0.02 wt % Ti addition) were controlled by the motion of inclusions during filling of
added to the melt. the mould, as shown schematically in Fig. 1.
Casting was done in a Stahl permanent mould (type Filtration using ceramic foam filters is one of the
ASTM B-108) heated at 425 °C. The test bars obtained most effective methods for inclusion removal. This
from the Stahl mould castings were solution heat technique is now being applied in most aluminium
foundries [17—22]. In the present work, round filter
discs (50 mm diameter, 15 mm thickness) were moun-
TA BLE I Chemical composition (wt %) of A356.2 alloy ted at the bottom of cylindrical stainless steel tubes.
Both the tube(s) and filter(s) were preheated at 450 °C,
Si Mg Fe Mn Cu Be Sr and placed firmly at the top of the sprue of the Stahl
mould prior to casting. The molten metal was poured
6.78 0.33 0.11 0.04 0.02 — —
through such tubes. Table VII shows the consistency
2270
TA BLE II Melt conditions for preparation of tensile test bars

Experiment Type of Weight of Melt Mould Additive Settling Stirring Stirring Degassing Filter Humidity
charge charge temperature temperature time speed time time size
(kg) (°C) (°C) (h) (r.p.m.) (min) (min) (p.p.i) (%)

3 Fresh 18 735$5 435$3 — 72 — — — — 18

4 Fresh 18 735$5 435$3 — 0 125—130 20 — — 15
5 Fresh 18 735$5 435$3 — 0 125—130 — 45 — 17
15 Fresh 18 735$5 435$3 — 4! 125—130 — 45 — 13
6A—6G Fresh 25 735$5 435$3 — 0 — — — 10 17
6H—6N Fresh 25 735$5 435$3 — 0 125—130 — 45 10 17
9A—9F Fresh 20 735$5 435$3 Al—Ti—B 1! 125—130 15 — — 17
9G—9L Fresh 20 735$5 435$3 Al—Ti—B 0 125—130 — 45 — 17
7A—7H Fresh 25 735$5 435$3 Al—Sr 0 125—130 15 — — 21
7I—7P Fresh 25 735$5 435$3 Al—Sr 0 125—130 — 45 — 21
8A—8E Fresh 25 735$5 435$3 — 72! 125—130 15 — — 16
8F—8J Fresh 25 735$5 435$3 — 0 — — — 10 16
8K—8O Fresh 25 735$5 435$3 — 0 125—130 — 45 — 20
11 Scrap 20 735$5 435$3 — 0 — — — — 19
10 Scrap 20 735$5 435$3 — 72 — — — — 15
12A—12F Scrap 20 735$5 435$3 — 0 125—130 20 — — 15
12G—12L Scrap 20 735$5 435$3 — 0 125—130 — 45 — 15
14A—14F Scrap 20 735$5 435$3 — 72! 125—130 15 — — 17
14G—14K Scrap 20 735$5 435$3 — 0 125—130 — 45 — 17
13A—13G Scrap 20 735$5 435$3 — 0 — — — 10 20
13H—13N Scrap 20 735$5 435$3 — 0 125—130 — 45 10 20

! Stirring/degassing before settling.

in tensile properties for castings marked 6A to 6G, Modification of A356.2 alloy with strontium in the
which were obtained following such a procedure. The range of 250—300 p.p.m. is recommended in order to
combined effect of degassing and filtering resulted in change the eutectic silicon morphology from acicular
ductilities as high as 16%, especially for the last cast- to fibrous and enhance the alloy properties [27—31].
ings in the casting sequence, i.e. those taken from the This type of melt treatment is expected to shorten the
melt towards the bottom of the melting crucible. solution heat-treatment time required to achieve more
Grain refining of an Al—Si casting is normally done or less complete spheroidization of the eutectic silicon
through the addition of small amounts of TiB to the particles. However, the addition of strontium causes
2
melt in the form of Al—Ti—B master alloys [23—26]. a significant increase in the porosity volume fraction
Various types of master alloys can be used, such as which, in turn, affects the alloy strength. Table IX
Al—10% Ti—1% B, Al—5% Ti—1% B, and Al—7.5% depicts the tensile properties of strontium-modified
Ti—7.5% B. In the present work, Al—5% Ti—1% B was alloy obtained in the present study. Apparently, de-
chosen, being commonly used in aluminium foundries. gassing leads to contamination of the liquid metal due
The amount needed to be added was calculated on the to the reason mentioned above.
basis of a 0.02 wt % Ti addition. As was observed When the liquid metal is held at a sufficiently high
earlier in our study, such a small amount of Al—Ti—B temperature, e.g. 735 °C, for a lengthy period of time
master alloy is enough to introduce an extremely large (\72 h), most of the inclusions will settle to the bot-
number of TiB particles into the melt. These particles tom of the crucible. Thus, it is expected that the upper
2
act as nucleants for the crystallization of the liquid portion of the liquid metal (i.e. that in the upper
metal during solidification. Table VIII depicts the one-third of the crucible) is reasonably clean, Table X.
maximum tensile properties (castings A—C) and min- Using ceramic foam filters for the next one-third por-
imum tensile properties (castings D—F), with respect to tion of the melt proved to be effective in removing the
percentage elongation. It can be seen that although settled inclusions. As the amount of liquid metal re-
the alloy strength is high, the associated ductilities do maining thereafter was fairly little, degassing resulted
not necessarily follow the same trend. Degassing the in disturbing the settled inclusions at the bottom of
second half of the liquid metal (i.e. that in the lower the crucible, and, hence, random ductilities.
portion of the crucible) did not necessarily reduce the
difference in percentage elongation obtained from the
two test bars produced simultaneously from the same 3.2. Recycled (scrap) alloy
casting. This observation has been explained sche- The recycled material used in the present work consis-
matically in Fig. 2, where the amount of clean metal ted of a mixture of gates and runners collected from
(or rather, its depth with respect to the crucible dia- the fresh alloy castings (except those containing TiB
2
meter) after degassing is much less than what would be and strontium), as well as any remaining metal that
expected from the thickness of the dross and inclu- was not used for making test bars. The main type of
sion-rich layers. inclusions observed in the recycled (scrap) alloy was
2271
TA BLE II I Tensile test results of experiment 3 (fresh alloy) TA BLE V Tensile test results of experiment 5 (fresh alloy)

Experiment YS (MPa) UTS (MPa) % Elongation Experiment YS UTS % Elongation

(MPa) (MPa)
3A1 108.472 271.125 12.62
3A2 106.018 264.933 11.681 5A1 125.420 299.926 12
3B1 101.294 253.005 12.61 5A2 115.091 287.570 16.97
3B2 101.612 253.729 9.25 5B1 117.277 293.065 6.321
3C1 98.833 247.075 11.47 5B2 118.649 296.347 10.786
3C2 106.286 263.865 13.52 5C1 111.375 278.310 12.17
3D1 103.818 259.404 9.15 5D1 115.691 287.556 13.97
3D2 106.066 264.851 9.95 5D2 111.678 278.923 8.163
3E1 103.777 259.424 14.61 5E1 129.985 295.230 11.08
3E2 103.273 257.315 13.81 5E2 104.963 262.148 9.25
3F1 106.852 265.382 15.81 5F1 101.198 252.791 13.06
3F2 103.073 257.611 10.91 5F2 116.601 291.155 16.71
3H1 107.590 268.815 16.81 5G1 116.450 291.100 10.604
3H2 109.024 272.518 12.51 5G2 117.580 293.189 14.17
3I1 109.555 273.690 15.51 5H1 119.097 297.492 12.85
3I2 107.507 268.484 11.483 5H2 118.994 296.554 13
3J1 121.635 277.827 19.15 5I1 120.545 297.154 12.06
3J2 106.528 265.271 11.18 5J1 119.401 297.540 15.52
3K1 112.195 277.620 17.91 5J2 115.684 288.611 16.88
3K2 107.479 268.560 14.12 5K1 115.877 289.487 19.05
5K2 121.359 300.188 16.5

TA BLE IV Tensile test results of experiment 4 (fresh alloy)

TA BLE V I Tensile test results of experiment 15 (fresh alloy)
Experiment YS UTS % Elongation
(MPa) (MPa) Experiment YS UTS % Elongation
(MPa) (MPa)
4A1 102.156 254.929 5.67
4A2 109.093 272.725 7.95 15A1 113.961 284.522 9.691
4B1 113.395 281.785 7.409 15A2 112.588 281.192 10
15B1 115.071 287.384 16
4B2 112.347 280.599 6.758
4C1 111.692 279.179 7.375 15B2 115.071 287.625 10.2
4C2 102.005 254.688 4.933 15C1 114.554 286.370 14.1
15C2 114.236 284.867 10.41
4D1 116.808 291.852 11.12
4D2 114.009 284.777 8.197 15D1 113.795 284.302 16.5
4E1 110.872 277.172 10.05 15D2 103.246 257.618 5.423
15E1 114.429 285.501 14.2
4E2 111.334 278.124 8.267
4F1 108.672 271.732 14.36 15E2 115.933 289.652 10.25
4F2 112.326 280.606 10.81 15F1 112.726 281.219 16.9
15G1 115.912 289.507 15
4G1 112.389 280.957 14.63
4G2 109.320 273.242 12.65 15G2 103.990 259.390 4.238
4H1 112.451 281.006 12.45 15H1 107.900 268.091 13.2
15H2 106.411 265.706 5.29
4H2 115.933 289.418 13.77
4I1 108.376 270.801 14.81 15I1 116.167 289.983 14.7
4I2 117.773 294.065 13.71 15I2 104.549 261.031 7.75
15J1 103.349 257.852 16.7
4J1 117.056 291.783 13.56
4J2 114.609 282.074 10.51 15J2 110.389 275.635 13.5
4K1 107.135 267.781 15.11 15K1 111.168 277.407 11.7
15K2 114.905 287.087 12.8
4K2 115.919 289.507 12.65
4L2 117.146 292.472 16.91 15L1 115.774 288.928 12.4
15L2 113.561 283.867 10

the large number of oxide films that floated to the that ranged between 12% and 18%, Table XII,
surface during melting. Tables XI—XV summarize the compared to 5%—14% shown in the previous case,
tensile properties that were obtained following several Table XI.
foundry operations. One of the serious problems that could occur dur-
Table XI exemplifies the effect of oxide films on the ing melting is if the melt is vigorously stirred: such an
alloy strength immediately after melting. As can be operation would disturb a large portion of the oxide
seen, the ductility improved gradually along the length films in suspension near the upper surface of the liquid
of the melting crucible. Holding the liquid metal at metal, and incorporate them into the bulk of the
735 °C for a period as long as 72 h led to flotation melt. Their concentration within the melt will deter-
of most of the oxide films to the upper surface of mine the alloy properties, Table XIII. These properties
the molten metal which could be removed easily, are found to be more or less comparable with those
and sedimentation of heavy inclusions to the bottom shown in Table XI. Thus, degassing may be the better
of the crucible. This process resulted in elongations way to remove both inclusions and oxide films. The
2272
Figure 1 Schematic diagram showing the motion of inclusions before and after degassing. (a) After melting (experiment 6.11), (b) after
degassing (experiment 5), (c) after degassing#settling 4 h (experiment 15), (d) after degassing/stirring#settling 72 h (experiment 3.1).

TA BLE V I I I Tensile test results of experiment 9 (fresh alloy)

Experiment YS UTS % Elongation

TA BLE VI I Tensile test results of experiment 6 (fresh alloy) (MPa) (MPa)

Experiment YS UTS % Elongation 9A1 117.312 292.948 9.9

(MPa) (MPa) 9A2 114.278 285.522 7.843
9B1 108.314 270.553 6.116
6A1 110.975 277.303 11.1 9B2 114.519 286.122 6.664
6A2 117.463 293.451 11.05 9C2 116.077 290.100 9.2
6B1 117.029 288.776 8.63 9D1 116.070 289.893 7.987
6B2 119.325 291.700 12.13 9D2 112.071 279.296 8.136
6C1 112.078 279.847 12.77 9E1 115.436 288.432 9.27
6D2 125.730 282.985 9.27 9E2 116.836 291.941 12.3
6E1 116.994 292.045 7.95 9F2 118.249 294.010 11.5
6E2 111.961 278.613 10.07 9G1 116.001 289.700 9.6
6F1 111.313 277.448 7.503 9G2 113.492 283.460 9.3
6F2 119.415 298.347 13.86 9H1 112.906 281.144 6.967
6G1 116.477 290.217 8.92 9H2 107.403 268.498 5.466
6G2 114.705 285.356 11.13 9I1 119.256 295.809 11.5
6H1 108.927 271.580 6.9 9I2 111.258 278.082 7.23
6H2 118.173 289.128 13.78 9J1 116.181 290.273 9.35
6I1 121.945 288.818 12.74 9J2 154.924 386.803 11.25
6I2 114.209 284.791 14.55 9K1 113.402 283.185 14
6J1 117.822 293.155 11.46 9K2 116.367 290.852 15.8
6J2 115.533 287.522 11.112 9L1 117.236 292.996 9.9
6K1 120.890 301.725 12.68 9L2 116.112 289.838 13.6
6K2 115.319 287.632 9.85
6L1 116.257 288.894 17.18
6L2 123.345 304.483 14.68
6M1 116.201 286.632 9.36 effectiveness of the degassing process will depend dir-
6M2 115.602 288.852 7.834 ectly on the amount of molten metal.
6N1 116.498 289.576 14.92
The combined effect of sedimentation (72 h at
6N2 113.899 284.060 17.15
735 °C) and degassing on the alloy tensile properties is
2273
Figure 2 Schematic diagram showing the effectiveness of degassing with respect to amount of liquid metal. (a) Full crucible (experiment 5), (b)
partially full crucible (experiments 6, 7, 9).

TA BLE IX Tensile test results of experiment 7 (fresh alloy) TA BLE X Tensile test results of experiment 8 (fresh alloy)

Experiment YS UTS % Elongation Experiment YS UTS % Elongation

(MPa) (MPa) (MPa) (MPa)

7A1 113.133 282.654 9.2 8A1 112.892 282.061 7.95

7A2 112.726 281.475 10.65 8A2 107.445 268.471 6.441
7B1 116.643 291.017 11.7 8B1 115.650 288.880 14.5
7B2 111.478 278.655 11.532 8B2 114.416 285.991 9.75
7C1 106.673 266.347 7.519 8C1 103.404 258.238 4.18
7C2 115.450 288.294 9.4 8C2 110.492 275.807 9.176
7D1 107.155 267.685 7.755 8D1 112.154 280.309 9.7
7D2 112.754 281.702 11.25 8D2 111.409 278.344 6.151
7E1 110.237 275.504 7.216 8E1 111.485 278.537 9.3
7E2 113.974 284.708 10.5 8E2 110.734 276.800 10.9
7F1 113.692 284.040 8.982 8F2 113.836 284.122 8.319
7F2 115.153 287.570 13.45 8G1 110.630 276.372 10.5
7G1 106.479 266.064 6.49 8G2 113.299 282.895 9.023
7G2 113.678 283.791 8.759 8H1 114.443 283.626 12
7H1 110.823 276.490 8.499 8H2 109.300 273.056 6.9
7H2 116.581 290.969 13.4 8I1 115.084 287.618 8.413
7I1 109.368 273.063 13.35 8I2 110.113 275.145 6.749
7J1 106.652 266.526 7.414 8J1 115.960 289.280 10.8
7J2 113.202 282.854 10.4 8J2 111.451 278.441 6.982
7K1 98.199 245.221 5.485 8K1 96.254 240.146 2.659
7K2 110.148 274.883 9.15 8K2 113.685 283.984 10.95
7L1 99.460 248.620 5.923 8L1 91.641 228.080 2.214
7L2 104.673 261.617 7.8 8L2 111.382 278.386 6.98
7M1 99.033 247.482 3.705 8M1 89.366 223.405 1.937
7M2 114.960 287.342 12.5 8M2 112.223 280.392 7.459
7N2 117.580 293.465 12 8N1 108.913 272.228 7.431
7O1 96.847 242.097 4.441 8N2 112.582 281.433 10.7
7O2 111.085 277.386 10.65 8O1 95.744 238.767 2.371
7P1 99.488 248.365 4.232 8O2 113.519 283.467 6.546
7P2 104.039 259.707 11.3

shown in Table XIV. It is interesting to note the role of appropriate. It is also observed that the alloy elonga-
settling time in improving the tensile properties of the tion improves gradually with increasing casting se-
castings corresponding to the upper portion of the quence. Applying degassing reduces the tensile
crucible, whereas degassing tends to enhance the duc- properties of the first casting (i.e. 13H), as explained
tilities of castings made from metal close to the cru- schematically in Fig. 8 (see later). Beyond that, the
cible bottom. Casting 14G is a good example of alloy exhibits consistent improvement (13L).
inclusion concentration caused by inclusion floata- Table XVI and XVII summarize the average tensile
tion. properties of the A356.2 alloy studied for the different
Filtration of liquid aluminium with 10 p.p.i. ceramic experimental conditions listed in Table II.
foam filters is an important process of inclusion/oxide For castings made from fresh ingots, degassing of
removal. Table XV shows that, due to floatation of a large quantity of liquid metal (namely, a full crucible
oxide films to the upper surface of the liquid metal, in the present case) seems to be the best method for
filtration with 10 p.p.i. filters might not be very effec- inclusion removal and, hence, for achieving the high-
tive, and a finer filter size (i.e. 20 p.p.i.) would be more est percentage elongation. However, when applied to
2274
TA BLE XI Tensile test results of experiment 11 (scrap alloy) TA BLE X I I I Tensile test results of experiment 12 (scrap alloy)

Experiment YS UTS % Elongation Experiment YS UTS % Elongation

(MPa) (MPa) (MPa) (MPa)

11A1 98.102 245.152 4.585 12A1 110.417 275.800 7.57

11A2 103.377 258.176 5.088 12A2 110.334 275.676 6.405
11B1 105.562 263.610 7.116 12B1 108.582 271.229 7.155
11B2 100.219 250.447 4.663 12B2 106.769 266.892 7.506
11C1 106.452 169.955 7.372 12C1 106.390 265.782 10.8
11C2 103.680 258.866 5.424 12C2 111.616 278.999 8.039
11D1 109.844 273.159 6.908 12D1 112.216 280.323 9.4
11D2 110.934 277.069 9.12 12D2 111.630 278.965 9.042
11E1 113.650 283.764 10.4 12E1 113.912 284.481 11.6
11E2 113.747 284.357 10 12E2 114.243 181.821 10.25
11F1 111.223 277.600 12.05 12F1 114.188 285.122 13.1
11F2 111.961 279.634 10.223 12F2 112.782 281.668 12.1
11G1 110.541 276.303 8.845 12G1 115.043 287.473 12.3
11G2 104.901 261.948 5.58 12H1 114.540 286.170 11.6
11H1 111.651 278.827 8.34 12H2 112.382 280.840 8.9
11H2 96.930 242.256 10.2 12I1 114.416 285.894 9.8
11I1 113.981 176.409 11.02 12I2 120.497 298.443 12.7
11I2 110.423 275.759 10.3 12J1 116.015 289.480 17.5
11J1 110.954 277.055 13.4 12J2 115.484 288.632 15.5
11J2 113.623 284.019 14.45 12K1 111.982 279.785 11.1
11K1 111.010 277.158 14.2 12K2 114.809 286.949 12.35
11K2 112.320 280.316 13.1 12L1 110.610 276.276 11.4
11L1 109.341 273.152 10.5 12L2 113.381 277.255 14.2
11L2 113.974 284.481 10.8

TA BLE XI I Tensile test results of experiment 10 (scrap alloy) 3.3. Ductility—percentage inclusions/
Experiment YS UTS % Elongation
oxide films relationship
(MPa) (MPa) Fig. 3a—g show the size and distribution of non-metal-
lic inclusions/oxide films found on the fracture surfa-
10A1 110.803 276.517 16.5 ces of test bars obtained from fresh ingots. The
10A2 113.119 282.688 10 corresponding elongations range from 19%—1.5%.
10B1 112.844 281.730 14.5
Similarly, Fig. 4 shows the oxides observed on the
10B2 107.672 269.084 7.285
10C1 114.540 286.280 9.449 fracture surfaces of test bars obtained from scrap. The
10C2 114.871 287.073 11.9 area fraction of inclusions/oxides was measured from
10D1 113.657 283.481 13.4 photographs similar to those shown in Figs 3 and 4,
10D2 115.622 288.632 14.8 using the grid method. The relationship between per-
10E1 114.085 284.681 14.5
centage elongation and log area percentage inclu-
10E2 113.864 284.570 15.9
10F1 111.465 278.165 10.5 sions/oxide films is presented in Fig. 5. A linear
10F2 113.726 282.006 12.4 relationship expressed as
10G1 111.340 278.227 9.4
10G2 109.224 272.118 6.285 percentage elongation"9.5!2.98 log area percentage
10H1 114.629 285.536 13.6
10H2 112.264 280.392 11.2 inclusions/oxide films (1)
10I1 109.996 274.400 13
10I2 115.140 287.459 10.8 is obtained, with a good fitness coefficient (R2\0.9) .
10J1 112.540 281.102 10.6 Table XVIII lists the total inclusions, harmful inclu-
10J2 111.878 279.606 8.232 sions and oxide films as measured by the PoDFA
10K1 97.054 242.449 16 technique [32, 33], and the corresponding tensile
10K2 111.230 277.586 11.6
properties measured for similar melt experimental
10L1 113.064 282.399 15.8
10L2 112.526 281.233 15.8 conditions. It is evident from this table that the oxide
10M1 107.714 269.070 12.9 films have a far more deleterious effect on the mechan-
10M2 112.609 281.330 14.8 ical properties. This observation is in accordance with
the oxides shown in Figs 3 and 4.

a very limited amount of liquid metal, degassing

causes liquid contamination, instead, leading to the 3.4. Fractography
lowest values of elongation observed. When the cast- As shown in our previous work [32, 33], the main
ings are made from recycled materials, the obtained oxide inclusions in A356.2 alloy melts held for a long
ductilities are more or less the same, regardless of the period, i.e. 72 h, at 735 °C are aluminium oxide
type of melt treatment. This observation may be ex- (Al O ) and spinel (MgAl O ). Owing to plastic defor-
2 3 2 4
plicable in terms of the large amounts of oxide films mation during tensile loading, these oxides tend to
present in the recycled alloy melts. break into small fragments that are segregated in
2275
TA BLE XI V Tensile test results of experiment 14 (scrap alloy) TA BLE X VI Average tensile properties (fresh alloy)

Experiment YS UTS % Elongation Experiment YS UTS % Elongation

(MPa) (MPa) (MPa) (MPa)

14A1 95.338 256.853 10.9 3 106.54$4.8 264.53$8.27 13.20$2.79

14A2 95.472 257.280 11.3 4 111.86$4.34 279.186$10.69 11.03$3.4
14B1 89.204 240.008 5.761 5 116.65$6.28 288.72$12.35 17.59$19.65
14B2 92.296 248.951 7.869 15 111.18$5.11 277.56$12.83 11.42$4.02
14C1 93.389 251.702 11.9 6A—6G 116.12$4.29 286.34$7.08 10.37$1.98
14C2 93.146 250.585 12.5 6H—6N 116.76$3.64 289.06$7.72 12.44$3.2
14D1 95.977 258.487 13.3 9A—9F 114.92$2.9 286.88$7.16 8.89$1.97
14D2 93.466 252.061 9.598 9G—9L 117.88$12.08 294.22$30.11 10.33$3.07
14E1 95.779 258.259 12.2 7A—7H 112.29$3.29 280.44$8.15 9.77$2.12
14E2 93.811 252.915 8.803 7I—7P 106.05$6.82 264.90$16.99 8.45$3.32
14F1 97.536 262.893 11.3 8F—8J 112.68$2.37 281.17$5.68 8.85$1.89
14F2 95.127 256.418 10.4 8K—8O 104.53$9.98 261.03$25.12 5.92$3.45
14G1 91.268 245.965 5.208
14G2 93.696 252.233 10.4
14H1 95.133 256.584 11.9
14H2 92.859 250.392 9.2
14I1 93.964 253.302 12.7 TA BLE X VI I Average tensile properties (scrap alloy)
14I2 84.265 227.204 5.975
14J1 93.593 252.288 8.374 Experiment YS UTS % Elongation
14J2 95.344 256.984 13.2 (MPa) (MPa)
14K1 92.462 249.365 15.4
14K2 91.549 246.710 13.6 11 108.68$5.1 262.89$30.27 9.32$2.99
10 111.83$3.67 279.15$9.12 12.35$2.86
12A—12F 111.09$2.7 268.90$28.12 9.41$2.16
12G—12L 114.47$2.59 285.20$6.37 12.49$2.47
TA BLE XV Tensile test results of experiment 13 (scrap alloy) 14A—14F 94.21$2.17 253.87$5.92 10.49$2.16
14G—14K 92.41$3.16 249.10$8.51 10.60$3.37
Experiment YS UTS % Elongation 13A—13G 109.94$11.92 274.69$29.77 9.56$2.51
(MPa) (MPa) 13H—13N 111.99$5.69 279.75$14.26 10.44$2.31

13A1 107.583 268.905 5.382

13A2 107.955 269.663 5.604
13B1 107.203 267.988 9
13B2 70.805 176.919 8.741
13C1 115.153 287.859 10.9 mature failure. Another example of decohesion
13C2 110.251 275.441 5.938 between oxide particles and the matrix is shown in
13D1 118.373 295.775 9.3 Fig. 6d, where a lump of wrinkled layers of oxide films
13D2 115.422 288.259 10.3 is seen. The arrow in Fig. 6d indicates the passage
13E1 117.684 293.927 10.9
of a crack through the oxide film.
13E2 109.079 272.449 11.35
13F1 115.298 287.590 11.3 Besides the spinel particles, the other main type of
13F2 113.050 282.619 9.95 defect observed in the fracture surface is the presence
13G1 117.305 293.189 14.3 of large volume fractions of Al O oxide films, in the
13G2 114.043 285.060 10.9
2 3
form of crumpled layers. These films occur randomly
13H1 111.340 278.034 9.2 in the melt and, hence, test bars. Fig. 8a and b depict
13H2 102.763 256.797 7.996
13I1 116.684 291.493 7.934
examples of these films near the edge and the centre of
13I2 115.229 287.997 13.7 the test bar, respectively. The associated X-ray spec-
13J1 103.515 258.356 12.3 trum, Fig. 9, shows strong reflections due to alumi-
13K2 115.071 287.473 9.6 nium and oxygen, and a very weak magnesium
13L1 115.181 287.770 9.5 reflection, indicating the presence of small amounts of
13L2 116.153 290.080 13.25
MgO and spinel (MgAl O ).
2 4
The effect of oxide particles or films in weakening
the alloy through the initiation of cracks at their
interfaces with the matrix is best illustrated by Fig. 10.
certain locations, as shown in the optical macrograph It is evident from these fractographs that due to the
of Fig. 3g. The corresponding scanning electron difference in the rigidity (measured by Young’s
micrograph, Fig. 6a, reveals the presence of small modulus) of the matrix and inclusions, the crack initi-
particles (arrowed) in the fracture surface. A high- ates easily at particle/matrix interfaces, leading to
magnification fractograph, Fig. 6b, shows how the rapid crack propagation, large crack paths, and,
flow of liquid aluminium (shiny phase) is trapped hence, early failure.
within some of these oxide particles.
An interesting feature is exhibited in Fig. 6c, where
a large particle of spinel (as confirmed by the asso- 4. Conclusions
ciated EDX spectrum, Fig. 7) is seen debonded from 4.1. Fresh alloy
the surrounding matrix (arrowed). Such a type of 1. Under very dry conditions (humidity \10%—
decoherence would weaken the alloy, leading to pre- 13%), melt cleanliness by sedimentation, i.e. holding
2276
Figure 3 Optical macrographs showing the size and distribution of
non-metallic inclusions/oxide films in tensile-tested bars obtained
from fresh alloy, and corresponding to elongations of: (a) 19%, (b)
16%, (c) 10%, (d) 8%, (e) 5%, (f ) 3.5%, (g) 1.5%.

2277
 Figure 5 Percentage elongation—log area percentage inclusions/
oxide films relationship.

large amount of liquid. The first casting, however,

should be rejected, because it is taken from the melt
surface layer that is rich in inclusions/oxides.
3. Filtration using 10 or 20 p.p.i. ceramic foam
filters is highly recommended for inclusion removal in
a dry atmosphere.
4. Mechanical stirring normally used to ensure dis-
solution of grain refiners or modifiers may be replaced
by degassing using a rotary impeller.
5. Oxide films have a much more deleterious effect
on the mechanical properties compared to that ex-
pected from other inclusions.

4.2. Recycled (scrap) alloy

Holding the liquid metal for a lengthy period of time
improves the melt cleanliness similar to that men-
tioned above. However, due to the presence of excess-
ive amounts of inclusions and oxide films, degassing
and/or filtration may be a more appropriate proced-
ure to follow.
In both cases, i.e. fresh and recycled alloys, cracks
are initiated at the inclusion (or oxide)—matrix inter-
face, followed by their relatively rapid propagation,
leading to the formation of large crack paths through
the matrix and, hence, premature failure.

Figure 4 Optical macrographs showing the size and distribution of

non-metallic inclusions/oxide films in tensile-tested bars obtained
from recycled (scrap) alloy, and corresponding to elongations of: (a) Acknowledgements
7.5%, (b) 5%, (c) 3.5%. This study was supported by several industrial part-
ners (research and development organizations and
foundries) in Canada, the United States and Mexico.
These include Bomem, Inc. (Québec, Canada), Cercast
the liquid metal at 735 °C for \72 h, results in the Group (Québec, Canada), CQRDA (Québec, Canada),
settlement of inclusions to the bottom of the melting FUQAC (Québec, Canada), SËCAL (Québec, Cana-
crucible, leading to good mechanical properties. In- da), Grenville Castings Limited (Ontario, Canada),
crease in the humidity level, however, will introduce General Motors Powertrain Group (Saginaw, USA),
hydrogen to the melt, and result in a large porosity Nemak Corporation (Monterrey, Mexico), KB Alloys,
volume fraction deleterious to the mechanical proper- Inc. (Robards, USA), HI-TECH Ceramics (New York,
ties. USA). Their financial and in-kind support is gratefully
2. Degassing using a rotary impeller appears to be acknowledged. The authors would also like to
the most effective method for removing both hydrogen thank Dr A.M. Samuel for help with the fractography
and inclusions, provided it is applied to a reasonably work.
2278
TA BLE XVI II Inclusion/oxide—tensile properties relationships for A356.2 alloy

Serial Total inclusion Total harmful inclusion Oxide film YS UTS % Elongation
no. (mm2 kg~1) (mm2 kg~1) classification (MPa) (MPa)

Thin Thick

1 0.06 0.02 None None 123.40 310.10 19.20

2 0.02 0.02 Moderate Moderate 118.54 294.43 14.82
3 0.24 0.02 Moderate Moderate 116.60 290.61 13.88
4 0.68 0.24 Moderate Moderate 112.52 281.16 13.38
5 0.17 0.14 Light Moderate 108.61 271.13 13.16
6 0.54 0.19 Moderate Moderate 114.45 284.68 13.15
7 0.18 0.15 Light Light 118.03 294.58 12.66
8 0.03 0.01 Moderate —! 112.25 253.31 12.29
9 0.37 0.22 Light Light 112.70 280.06 12.26
10 4.24 3.78 None None 104.35 260.70 11.54
11 0.24 0.19 Heavy Moderate 111.81 279.28 11.31
12 0.85 0.21 Light Light 114.00 284.83 10.93
13 0.36 0.29 Moderate Light 115.47 288.44 10.78
14 0.01 0.00 Heavy —! 112.65 281.34 10.67
15 1.41 1.12 Heavy Moderate 111.36 278.27 10.60
16 0.64 0.35 Moderate Moderate 115.59 287.79 10.03
17 1.25 0.36 Moderate Moderate 114.92 286.61 9.85
18 0.05 0.03 Moderate Heavy 110.46 276.02 9.32
19 0.03 0.03 Heavy Heavy 114.80 286.83 9.11
20 0.04 0.01 Moderate —! 106.01 264.83 8.24
21 1.68 0.08 Light —! 113.61 283.79 7.63
22 0.03 0.01 Moderate —! 107.73 244.76 7.21
23 0.10 0.07 Heavy Heavy 98.39 245.87 7.18
24 0.93 0.48 Moderate Moderate 109.25 272.51 6.95
25 0.04 0.02 Moderate Heavy 101.82 254.35 5.36
26 1.40 1.12 Heavy Heavy 89.40 223.40 1.94

! Classification indeterminable.

Figure 6 Fracture surface of a tensile-tested bar obtained from an alloy melt held for 72 h at 735 °C, showing: (a) fragmentation of oxide
particles (corresponding to Fig. 3g), (b) the presence of liquid aluminium entrapped within oxide film particles, (c) debonding of a large spinel
particle from the surrounding matrix, (d) a crumpled mass of oxide films (the arrow denotes the crack path).

2279
Figure 7 The energy dispersive X-ray (EDX) spectrum correspond- Figure 9 Typical energy dispersive X-ray (EDX) spectrum corres-
ponding to the Al O oxide films shown in Fig. 8a. Note the very
ing to Fig. 6c, showing strong reflections due to oxygen, magnesium 2 3
and aluminium. weak magnesium reflection.

Figure 10 Fracture surfaces showing crack paths located at ox-

Figure 8 Fracture surface of a tensile tested bar obtained from ide/matrix interfaces in the samples corresponding to those shown
recycled alloy (Fig. 4c), showing a crumpled mass of oxide films: (a) in (a) Fig. 4a, and (b) Fig. 4c.
near the edge, and (b) close to the centre of the test bar.

References 3. D. A PEL IA N, in ‘‘1988 Electric Furnace Conference Proceed-

1. W. S I M M O N S , Indian Foundry J. August (1986) 21. ings’’, (Iron and Steel Society, Warrendale, PA, 1989) p. 325.
2. C. J. S IM EN S E N and G . BE R G , Aluminium 56 (1980) 4. D. A P ELIA N, R. M U T H AR A SA N and S . A LI, J. Mater. Sci.
335. 20 (1985) 3501.

2280
 5. H. C. CU M M IN G S , F. B. S TU L EN and W . C . S C H U LTE , 21. L. J . GA UC KL ER , M. M . W EBE R, C . C O N TI and M.
¹rans. Amer. Soc. Metals 49 (1957) 487. JA CO B- D UL IERE , ¸ight Metals (1985) 1261.
6. C. E . E C K ER T, R . E. MI LL ER , D . AP EL IAN and R . 22. T. CA LLA I S, M . RI CH A RD and M. S TU C K Y, Fonderie-
MU TH AR ASA N , ¸ight Metals (1984) 1281. Fondeur d’aujourd’hui 97 (1990) 22.
7. C. J . S I ME N SE N , Metall. ¹rans. 12B (1981) 733. 23. L. F . M O N D O LFO , ‘‘Grain Refinement in Castings and
8. J. GOB RE CH T, Geisserei 62 (1975) 263. Welds’’ (Metallurgical Society of AIME, 1983) pp. 3—50.
9. F. H . S A M U EL , H . LI U and A . M . S A M U EL , Metall. ¹rans. 24. M. M . G O ZO WS KI , D . A. S E N TER and G . K. S IG -
24A (1993) 1631. WO RT H, Metall. ¹rans. 18A (1987) 603.
10. J. E . C AMP BE LL, ‘‘Castings’’ (Butterworth-Heinemann, 25. G. K . S I G WO RT H and M . M . G O Z O W S K I, Amer. Foundry-
Oxford, 1991). men’s Soc. ¹rans. 93 (1985) 907.
11. D. H E DJ A ZI , G . H. J . BEN N E TT and V . K O ND I C , Metall. 26. S. K E N N ER KN E C H T, in ‘‘Advanced Casting Technology’’,
¹echnol. 3 (1976) 537. AGARD Conference Proceedings No. 325, Advisory Group
12. Idem, Br. Foundryman 68 (1975) 305. for Aerospace Research and Development, North Atlantic
13. K. A RA I , T . O N I SHI , M . G O TO and R. O TSU K A , J. Jpn Treaty Organization [AGARD/NATO], Brussels, Belgium,
Inst. ¸ight Metals 27 (1977) 319. 4—9 April 1982, pp. 1—52.
14. P. R. RI OS , J . C. BR U N O and A . S . M . FO N S EC A, J. Mater. 27. S. Z . LU and A . HE L LA W EL L, Metall. ¹rans. 18A (1987)
Sci. ¸ett. 10 (1991) 1346. 1721.
15. K. H . C H I EN , T. Z. K A TT A MI S and F. R. M O LL AR D , 28. D. AP E LI AN , G . K . S I G W O RT H and K . R. W H AL ER ,
Metall. ¹rans. 4A (1973) 1069. Amer. Foundrymen’s Soc. ¹rans. 92 (1988) 297.
16. V. G E R OL D , in ‘‘Fourth International Conference on Age- 29. R. D A S GU P TA , C . G . B R OW N and S . M A REK , ibid. 96
Hardenable Aluminium Alloys’’, Balatonfured, Hungary, May (1988) 297.
1985, Materials Science Forum 13—14 (1987) pp. 175—94. 30. P. D . H ES S and E . V . BL AC K M U N , ibid. 84 (1975) 87.
17. C. E . E C K ER T, R . E. MI LL ER , D . AP EL IAN and R . 31. M. G A RA T and R. S C AL L IE T, ibid. 84 (1978) 549.
MU TH AR ASA N , ¸ight Metals (1984) 1281. 32. L. L IU and F. H . SA MU EL , J. Mater. Sci. 33 (1997) 5901.
18. D. AP E LIAN and R. M U T H AR AS A N , J. Metals 32 (1980) 14. 33. Idem, ibid. 33 (1997) 5927.
19. D. AP ELI AN and R. MU T HA R ASA N, ‘‘Modelling of Inclu-
sion Removal of Melt Systems’’, presented at the 72th Annual
AIChE Meeting, San Francisco, CA, November 1979.
20. C. J . S IM EN S EN and U . H A RT VE D T, Z. Metallkde 76 Received 5 August
(1985) 409. and accepted 17 December 1997

2281