You are on page 1of 6

Met. Mater. Int., Vol. 23, No. 4 (2017), pp.

720~725
doi: 10.1007/s12540-017-6745-2

High-Temperature Oxidation of AZ91-0.3%Ca-0.1%Y Alloy in Air


1 1 2,*
Dong Bok Lee , Muhammad Ali Abro , and Bong Sun You

1
School of Advanced Materials Science & Engineering, Sungkyunkwan University, Suwon 16419, Republic of Korea
2
Materials Commercialization Center, Korea Institute of Materials Science, Changwon 51508, Republic of Korea
(received date: 20 October 2016 / accepted date: 1 December 2016)

AZ91 magnesium alloys containing 0.3% Ca and 0.1% Y were cast, and their oxidation behavior was investigated
between 425 and 600 °C in atmospheric air to examine roles of Ca and Y during oxidation. During casting, Ca
formed Al2Ca particles intergranularly, and reduced the amount of Al12Mg17 particles, while most of yttrium
existed as Al2Y particles inter- and intra-granularly in the alloy. The AZ91 alloy oxidized fast above 425 °C, lead-
ing to complete ignition. By contrast, AZ91-0.3Ca-0.1Y alloy oxidized very slowly up to 550 °C. Calcium, which
is more active than Mg, preferentially oxidized to CaO at the surface of the MgO-rich oxide to suppress the oxi-
dation, evaporation and diffusion of Mg during the initial oxidation stage. Such suppression was due to the quite
low vapor pressure and high stoichiometry of CaO. Calcium also suppressed the formation of less oxidation-resis-
tant Al12Mg17 through forming oxidation-resistant Al2Ca in the alloy from the initial oxidation stage.
Keywords: alloys, casting, oxidation, scanning electron microscopy (SEM), magnesium, calcium

1. INTRODUCTION (Mg-5Al-0.28Mn in wt%) [19] by forming an yttrium-rich oxide


on the alloy surface. However, the Y-addition increased the
Magnesium is used as light-weight structural components manufacturing cost, and the Ca-addition decreased rapidly
in automotive and aerospace industries because of its low den- the tensile properties because of precipitation of brittle, coarse
sity, high specific strength, and good castability, machinability Al2Ca along grain boundaries. By minimizing the amount of
and creep resistance [1,2]. However, its poor oxidation resis- Ca not to deteriorate the elongation and by adding a bit of Y to
tance is a big concern [3-5]. Magnesium oxidizes rapidly above improve the elongation, You et al. [20] have developed the
450 °C, and begins to ignite in several spots above 500 °C, cost-effective AZ91-0.3%Ca-0.1%Y (wt%) alloy having good
resulting in complete disintegration into pulverized oxides [6-8]. tensile properties and non-flammability. The combined addi-
During casting, Mg melts are shielded with the SF6 gas, which tion of Ca and Y had very attractive synergistic effects on the
is the most potent greenhouse gas with a life-time of 3200 ignition resistance and tensile properties when compared to
years and a global warming potential of 23900 times of CO2 [9]. the individual addition of Ca or Y. The improved ignition
To prevent ignition during casting, 10-100 ppm of Be, which resistance was attributed to the formation of the (CaO, Y2O3)-
is more active than Mg, is added in industries [6,10]. Not to mixed underlayer below the (MgO, CaO)-mixed layer, as depicted
use the SF6 gas and quite toxic beryllium through increasing in Fig. 1. This oxidation mechanism is valid for Mg(l), where
the oxidation resistance, small amounts of Ca and rare earths the liquid-gas reaction occurs. During application as structural
(REs) were alloyed in Mg alloys [7]. Calcium, which is more components, Mg(s) reacts with the oxidizing gas so that a different
active than Mg, effectively retarded the oxidation and igni- mechanism may govern. This study aims at characterizing the
tion through forming a thin CaO-rich outer layer during oxi- oxidation behavior of the AZ91-0.3Ca-0.1Y alloy below the
dation [11-13]. The formed inner layer was a (MgO, CaO)-
mixture in the case of Mg+(0.5-3) wt% Ca alloys [11], or a
(MgO, CaO, Al2O3)-mixture in the case of AZ91+(0.3-5) wt%
Ca alloys [12]. During casting and oxidation, the oxygen active
element, Y, protected pure Mg by forming Y2O3 [14,15], Mg-
Y-Ce alloys by forming Y2O3 and Ce0.202Y0.798O1.601 [16,17],
WE43 alloy (Mg-4Y-0.6Zr-3RE in wt%) [18] and AM50 alloy

*Corresponding author: bsyou@kims.re.kr Fig. 1. Schematic illustration of the oxide structures of Ca-added and
©KIM and Springer (Ca, Y)-added Mg melts [20].
High-Temperature Oxidation of AZ91-0.3%Ca-0.1%Y Alloy in Air 721

melting point of Mg (i.e., 650 °C). The AZ91 alloy (Mg-9Al-


0.9Zn-0.18Mn in wt%) was chosen because it is the most widely
used Al-rich, high-strength cast Mg alloy. The microstructure,
oxidation rate, oxide scales, and oxidation mechanism of the
AZ91-0.3Ca-0.1Y alloy were investigated.

2. EXPERIMENTAL PROCEDURE

The AZ91 alloy with and without 0.3Ca-0.1Y was melted


at 720 °C in a low-carbon steel crucible using an electrical
resistance furnace under a 90%CO2+10%SF6 protective gas.
The starting metals were magnesium (99.93% pure), alumi-
num (99.9% pure), zinc (99.9% pure), calcium (99.9% pure)
and yttrium (99.9% pure). The melt was poured into a perma-
nent mold preheated to 200 °C, and then air-cooled to ambient
temperature. The cast ingot was cut into coupons 5×10×10
3
mm in size, ground to a 1000 grit SiC finish, degreased, and
oxidized between 425 and 600 °C in flowing atmospheric air
using a thermogravimetric analyzer (TGA). The samples were
inspected using a field-emission scanning electron micro-
scope (FE-SEM) equipped with an energy dispersive spec-
trometer (EDS), an electron probe microanalyzer (EPMA),
an Auger electron spectrometer (AES), a high-power X-ray
diffractometer (XRD) with Cu-Kα radiation at 40 kV and 100
mA, and a transmission electron microscope (TEM operated
at 200 kV) equipped with an EDS (5 nmφ spot size). The TEM
sample was prepared by milling using a focused ion beam
system after carbon coating. The etchant used for the micro-
structural observation was acetic-picral (5 g picric acid+10
ml glacial acetic acid+10 ml H2O+70 ml ethanol).

3. RESULTS AND DISCUSSION

Figure 2 shows the XRD/EPMA/EDS results of the AZ91-


0.3Ca-0.1Y alloy. The XRD pattern shown in Fig. 2(a) detected
α-Mg as the major phase, and β-Al12Mg17 as the minor one.
During casting, Ca reacted with Al to precipitate Al2Ca along
grain boundaries, and unreacted Ca dissolved in the α-Mg
matrix [21,22]. During casting, Y in Mg-Y alloys formed
Mg24Y5 [14], and Y in Mg-Al-Y alloys formed Al2Y usually
at the grain center [19,23-26]. However, Al2Ca and Al2Y par-
ticles were absent in Fig. 2(a) owing to their small amounts.
They precipitated along boundaries of α-Mg grains (Fig. 2(b)).
Zinc dissolved in Al2Ca and β (Fig. 2(c)). Circles marked in
Figs. 2(b) and 2(c) indicate the location of Y, which was added to
the alloy by 0.1%. Al2Y existed inter- and intra-granularly, and
was incorporated with some Mg, Zn, Mn, and Ca. One of the
EDS spectra of impure Al2Y precipitate is shown in Fig. 2(d).
Figure 3 shows the oxidation curves of AZ91 with and Fig. 2. AZ91-0.3Ca-0.1Y alloy. (a) XRD pattern, (b) EPMA image
without 0.3Ca-0.1Y. The vertical axes are the sum of weight (etched), (c) EPMA maps of (b), (d) EDS spectrum of an encircled area.
gain due to the formation of oxides and weight loss owing to
the spallation of oxides. AZ91 kept oxidizing fast with time at AZ91 oxidized to MgO plus some MgAl2O4 and ZnO [8]. By
425 °C, and completely ignited at 450 °C after 12.2 h (Fig. 3(a)). contrast, AZ91-0.3Ca-0.1Y oxidized with negligible weight
722 Dong Bok Lee et al.

the early oxidation stage. The oxide scale was about 0.3 μm-
thick. Ca, which is more active than Mg, oxidized preferen-
tially at the oxide surface. Such was similarly observed during
the oxidation of Mg-(0.5-3)Ca [11], Mg-5Ca-1Zn [5], AZ31-
0.3Ca [22], AZ91-(0.3-5)Ca [27], and AZ91-2Ca-0.06Be [28] in
wt%. It is seen that the outwardly diffusing tendency in the
oxide scale decreased in the order of Ca2+, Mg2+, and Al3+. Ca
and Mg oxidize to the highly stable CaO and MgO, respec-
tively, which have Schottky defects where the outward diffu-
sion of cations occurs faster than the inward diffusion of anions
[29]. According to the MgO-CaO binary phase diagram, CaO
and MgO have negligible mutual solid solubility [30]. Hence,
the (CaO, MgO)-mixed oxide scale existed as a supersatu-
rated state. The minor alloying element, Zn, diffused up to the
surface, because ZnO also grew by the outward diffusion of
interstitial Zn2+ ions [31]. A small amount of sulfur that came
from the shielding gas employed during casting was also incorpo-
rated in the oxide scale and the alloy. On the other hand, Mg
alloys have poor oxidation resistance for the following rea-
sons. (1) The Pilling–Bedworth ratio of MgO is 0.81 so that the
Fig. 3. Weight gain versus oxidation time curves. (a) AZ91 alloy, (b) tensile stress develops in the MgO scale, facilitating microc-
AZ91-0.3Ca-0.1Y alloy.
racking and spallation. Oxygen migrates easily through microc-
racks and the scale-spalled region to react directly with the
alloy [6-8]. (2) Oxidation of Mg to MgO gives off a large heat,
which heats the local area to considerably high temperatures.
The vigorous reaction with oxygen, ignition in several spots,
and ever-increasing linear oxidation rates result in the forma-
tion of the characteristic cauliflower-like MgO nodules. Mg
alloys eventually burn down to powder. (3) The outward dif-
fusion of Mg2+ ions to form MgO develops Kirkendall voids
at the scale/alloy interface, which deteriorate the scale adher-
ence by acting as stress concentration sites and also act as
channels for inward oxygen and outward cation transport [8]. (4)
Mg has a high vapor pressure, which keeps increasing with
increasing temperature [32]. The similarity of Mg and Ca is
that they readily oxidize exothermically to form highly stable
MgO and CaO, which have Schottky defects where cations
diffuse faster than anions. The dissimilarity is that Ca is more
active, and has the lower vapor pressure than Mg. For exam-
o
ple, at 525 °C, the equilibrium vapor pressure of Ca, PCa(s) , is
-6 o -4
1.4 × 10 atm, which is smaller than PMg(s) = 2.4 × 10 atm [33].
Once the stoichiometric CaO having a quite low vapor pres-
o -34
sure (e.g., PCaO (s ) = 1.37 × 10 atm at 525 °C [33]) covers the
Fig. 4. AES depth profiles of AZ91-0.3Ca-0.1Y alloy taken after oxi-
dation at 550 °C for 2 h in air. The sputtering rate was 55 nm/min for surface, the evaporation of Mg, the direct contact of Mg with
the reference SiO2. oxygen to form MgO, and the ionic transport in the oxide
scale would all be suppressed. It is noted that the Pilling–Bed-
gains up to 550 °C (Fig. 3(b)). A small bump in the oxidation worth ratio of CaO is 0.64 [31], which is even small than 0.81
curve at 575 °C was due to the local ignition. AZ91-0.3Ca-0.1Y of MgO. Hence, CaO as well as MgO develops the tensile
oxidized rapidly from the beginning at 600 °C. Nonetheless, stress. Ca gives off even more heat than Mg during oxidation.
the small addition of Ca and Y beneficially increased the oxidation For example, the standard heat of formation, ΔGfo (kJ/mol), is
resistance of AZ91. -549 for CaO(s), and -512 for MgO(s) at 550 °C [33]. Since
Figure 4 shows AES depth profiles of AZ91-0.3Ca-0.1Y Ca has positive and negative effects on oxidation, we propose
taken after oxidation at 550 °C for 2 h, which was taken to study that the role of Ca is not prevention but suppression of oxida-
High-Temperature Oxidation of AZ91-0.3%Ca-0.1%Y Alloy in Air 723

tion of Mg by partially replacing MgO with CaO in the oxide


scale. Yttrium is as active as Ca, and can improve the oxidation
resistance, as outlined in Fig. 1. However, no Y was detected
in the oxide scale, because it was mostly bound to Al2Y particles.
As such, it could not diffuse into the oxide scale. The enhanced
oxidation resistance observed in Y-added Mg alloys was mainly
attributed to Y dissolved in the alloy.
Figure 5 shows XRD/SEM/TEM-EDS results of AZ91-
0.3Ca-0.1Y after oxidation at 550 °C for 20 h, which was obtained
to study the intermediate oxidation stage. In Fig. 5(a), the α-
Mg matrix and β-Al12Mg17 precipitates were detected with-
out MgO, because the oxide scale was still too thin. At 550 °C,
precipitates such as Al2Ca (m.p. = 1079 °C) and Al2Y (m.p. =
1485 °C) do not melt, but Al12Mg17 (m.p. = 458 °C) melts. This
led to the grain growth of the alloy (compare Figs. 2(b) and
5(b)). In Fig. 5(c), the scale was partially detached off from the
matrix, which happened during the TEM sample mounting
owing to the accumulated residual stress despite thinness of
the scale. It was about 2.1 μm-thick, nanocrystalline, single-
layered, and microscopically inhomogeneous in composition
(Figs. 5(c-d)). Although concentrations displayed in Fig. 5(e)
are inaccurate because the oxygen signal is susceptible to
absorption owing to its low characteristic energy, the followings
can be seen. The scale consisted of Mg, Al, Ca, and Zn, in a
descending order of concentration (Fig. 5(e)). A blocky Al2Ca
particle existed at the left side of Fig. 5(c). Its position sug-
gested that the oxidation was controlled by the inward diffu-
sion of oxygen. During the early oxidation stage, not only the
inward migration of oxygen but also the outward transport of
Ca, Mg, and Zn occurred to a certain extent (Fig. 4). During
casting, the initially added Ca formed Al2Ca particles along
grain boundaries, and also partially dissolved in α-Mg grains
[22]. Hence, Ca was detected in the oxide scale and Al2Ca
particle (Fig. 5(e)). There was a lesser amount of oxygen in
Al2Ca, implying that Al2Ca was somewhat oxidation-resistant
because of its high Al content (Fig. 5(e)). On the other hand,
Al12Mg17 is the preferential oxidation site at 550 °C, because it
melts at 458 °C [7]. This harmful Al12Mg17 was partially replaced
with the more oxidation-resistant Al2Ca when Ca was added.
This is another beneficial role of Ca, besides the incorpora-
tion of CaO in the MgO scale. In Figs. 5(c) and 5(e), the spot
1-3, 4-7, 8-9, and 10 corresponded to the (Mg, Al, Ca, Zn)-
containing oxide, the (Mg, O)-dissolved Al2Ca particle, the
(Mg, Al, Ca, Zn)-containing oxide, and the oxygen-affected
matrix, respectively. The primary element at the oxide surface
was changed from Ca (Fig. 4) to Mg (Fig. 5(e)) because the
supply of Ca from the alloy became limited, as the oxidation
progressed.
Figure 6 shows SEM/EPMA results of AZ91-0.3Ca-0.1Y
after oxidation at 575 °C for 20 h. In the low magnification Fig. 5. AZ91-0.3Ca-0.1Y alloy after oxidation at 550 °C for 20 h. (a)
photo shown in Fig. 6(a), a thin oxide layer formed on the XRD pattern, (b) SEM cross-sectional image, (c) TEM cross-sec-
coarse matrix grains. The small rectangle denoted in Fig. 6(a) tional image, (d) elemental maps of (c), (e) concentration profiles
was analyzed using EPMA, as shown in Fig. 6(b). This indicates along the spot 1-10 marked in (c).
724 Dong Bok Lee et al.

that most particles of Al12Mg17 and Al2Ca were present along


grain boundaries in which Zn was dissolved (Fig. 6(b)). Here,
Y spots matched with Al spots, because Y existed in the impure
Al2Y particles that formed inter- and intra-granularly. Yttrium
was hardly recognizable in the oxide scale. By contrast, Ca was
found in the MgO-rich oxide scale as well as grain boundaries
of the α-Mg grains. Mn was weakly, and randomly distributed
in the alloy. Since the thin Ca-containing, Mg-rich oxide scale
was not an impervious layer, the internal oxidation of Mg
occurred locally (see the Mg and O map).
AZ91-0.3Ca-0.1Y oxidized rapidly at 600 °C from the begin-
ning, as depicted in Fig. 3(b). At 600 °C, Mg oxidized fast, a
large heat evolved, vapor pressure of Mg became excessive,
and Al12Mg17(ℓ) around the surface oxidized fast. Hence, the
characteristic cauliflower-like oxide nodules developed locally,
as shown in Fig. 7(a). MgO-rich nodules provided a channel
for fast oxygen and Mg transport. Nodules that consisted of
pulverized oxides were gone during epoxy mounting, as shown
in Fig. 7(b). Here, the dotted line indicates the original alloy
surface. The etched image shows that the grain size of the alloy
was coarse, and non-uniform.

Fig. 6. AZ91-0.3Ca-0.1Y alloy after oxidation at 575 °C for 20 h. (a) 4. CONCLUSION


SEM cross-sectional image (etched), (b) EPMA maps of rectangular
box denoted in (a).
The initially added Ca formed Al2Ca particles along grain
boundaries. Al2Ca improved the oxidation resistance by resisting
the oxidation to a certain extent. The formation of Al2Ca auto-
matically reduced the amount of the less-oxidation resistant
Al12Mg17 by consuming Al in the alloy. This fact further improved
the oxidation resistance. The dissolved Ca in the α-Mg matrix
diffused outwardly to form the CaO-dissolved MgO layer.
The superficial CaO suppressed the vaporization and oxida-
tion of Mg, and ionic diffusion in the oxide scale, additionally
improving the oxidation resistance. Addition of Y was effec-
tive in the liquid phase oxidation [20], but the beneficial effect
of Y was not shown in the solid phase oxidation because the
initially added Y was bound to Al2Y particles inter- and intra-
granularly. However, the dissolved Y in the α-Mg matrix could
improve the oxidation resistance by diffusing into the oxide
scale.

ACKNOWLEDGEMENT

This work was supported by the project “Development of


the High-Efficiency Low-Emission Future Energy Produc-
tion Technology (EO15580)” of National Research Council
of Science & Technology (NST) grant by the Korea govern-
ment (MSIP) (No. CRC-15-07-KIER).

REFERENCES

Fig. 7. AZ91-0.3Ca-0.1Y alloy after oxidation at 600 °C for 30 min. 1. D. H. Cho, J. H. Nam, B. W. Lee, J. Y. Park, H. J. Shin, and I.
(a) SEM top view, (b) SEM cross-sectional image (etched). M. Park, Korean J. Met. Mater. 53, 220 (2015).
High-Temperature Oxidation of AZ91-0.3%Ca-0.1%Y Alloy in Air 725

2. J. Luo, Y. Yan, J. Zhang, and L. Zhuang, Met. Mater. Int. 22, 20. B. S. You, Y. M. Kim, C. D. Yim, and H. S. Kim, Magnesium
637 (2016). Technology 2014 (eds. M. Alderman, M. V. Manuel, N. T.
3. S. Y. Park , S. K. Kim, and D. B. Lee, Korean J. Met. Mater. Hort, and N. R. Neelameggham), p. 325, TMS, USA (2014).
54, 390 (2016). 21. K. Ozturk, Z. K. Liu, and A. A. Luo, Magnesium Technol-
4. Y. H. Kim and W. J. Kim, Met. Mater. Int. 21, 374 (2015). ogy 2003 (ed. H. Kaplan), p. 195, TMS, USA (2003).
5. K. Y. Ji and D. B. Lee, Korean J. Met. Mater. 54, 645 (2016). 22. D. B. Lee, Corros. Sci. 70, 243 (2013).
6. F. Czerwinski, Corros. Sci. 86, 1 (2014). 23. H. W. Chang, D. Qiu, J. A. Taylor, M. A. Easton, and M. X.
7. F. Czerwinski, J. Metals 64, 1477 (2012). Zhang, J. Magnes. Alloys 1, 115 (2013).
8. F. Czerwinski, Acta Mater. 50, 2639 (2002). 24. Z. Zhao, Q. Chen, Y. Wang, and D. Shu, Mat. Sci. Eng. A 515,
9. R. de Richter and S. Caillol, J. Photoch. Photobio. C 12, 1 152 (2009).
(2011). 25. D. Qiu and M. X. Zhang, J. Alloy. Compd. 586, 39 (2014).
10. X. Q. Zeng, Q. D. Wang, Y. Z. Lu, W. J. Ding, Y. P. Zhu, et al. 26. W. J. Park and N. J. Kim, Scripta Metall. Mater. 32, 1747
Mat. Sci. Eng. A 301, 154 (2001). (1995).
11. B. S. You, W. W. Park, and I. S. Chung, Scripta Mater. 42, 27. B. H. Choi, B. S. You, W. W. Park, Y. B. Huang, and I. M. Park,
1089 (2000). Met. Mater. Int. 9, 395 (2003).
12. B. H. Choi, B. S. You, and I. M. Park, Met. Mater. Int. 12, 63 28. B. H. Choi, B. S. You, and I. M. Park, Met. Mater. Int. 12, 63
(2006). (2006).
13. S. H. Ha, J. K. Lee, H. H. Jo, S. B. Jung, and S. K. Kim, Rare 29. P. Kofstad, Nonstoichiometry, Diffusion and Electrical Con-
Metals 25, 150 (2006). ductivity in Binary Metal Oxides, p. 121, Wiley-Interscience,
14. X. M. Wang, X. Q. Zeng, Y. Zhou, G. S. Wu, S. S. Yao, and Y. USA (1972).
J. Lai, J. Alloy. Compd. 460, 368 (2008). 30. W. J. M. van der Kemp, J. G. Blok, P. R. van der Linde, H. A.
15. A. Prasad, Z. Shi, and A. Atrens, Corros. Sci. 55, 153 (2012). J. Oonk, A. Schuijff, and M. L. Verdonk, Calphad 18, 255
16. J. F. Fan, C. L. Yang, G. Han, S. Fang, W. D. Yang, and B. S. (1994).
Xu, J. Alloy. Compd. 509, 2137 (2011). 31. D. A. Jones, Principles and Prevention of Corrosion, 2nd ed.
17. J. F. Fan, G. C. Yang, S. L. Chen, H. Xie, M. Wang, and Y. H. p. 412, Prentice Hall, USA (1996).
Zhou, J. Mater. Sci. 39, 6375 (2004). 32. V. Fournier, P. Marcus, and I. Olefjord, Surf. Interface Anal.
18. N. V. Ravi Kumar, J. J. Blandin, M. Suéry, and E. Grosjean, 34, 494 (2002).
Scripta Mater. 49, 225 (2003). 33. I. Barin, Thermochemical Data of Pure Substances, VCH,
19. P. Y. Lin, H. Zhou, W. P. Li, W. Li, S. Z. Zhao, and J. G. Su, Germany (1989).
Corros. Sci. 51, 1128 (2010).

You might also like