Materials Science and Engineering A308 (2001) 38 44

www.elsevier.com/locate/msea

Effects of bismuth and antimony additions on the microstructure

and mechanical properties of AZ91 magnesium alloy
Yuan Guangyin a,*, Sun Yangshan b, Ding Wenjiang a
a

Institute of Foundry Engineering, College of Materials Science and Engineering, Shanghai Jiaotong Uni6ersity, Shanghai,
200030, Peoples Republic of China
b
Department of Materials Science and Engineering, Southeast Uni6ersity, Nanjing 210096, Peoples Republic of China
Received 22 June 2000; received in revised form 8 December 2000

Abstract
The effects of bismuth and antimony additions on the microstructure and mechanical properties of AZ91 alloy have been
studied. Results show that a small amount of bismuth or antimony additions to AZ91 increases the yield strength and creep
resistance significantly at elevated temperatures up to 200C. The highest creep resistance has been obtained from the alloy with
combined additions of bismuth and antimony. The activation energies of the steady-state creep for AZ91-based alloys studied
were close to that of pure magnesium self-diffusion, indicating that dislocation climb is responsible for the creep mechanism under
the present conditions. Microstructural observations reveal that the additions of bismuth or antimony have the effect of refining
the b (Mg17Al12) precipitates in as-cast alloys and suppressing discontinuous precipitation of the b phase effectively during the
aging process. Some rod-shaped Mg3Bi2 or Mg3Sb2 particles distributed mainly at grain boundaries have been observed in the
alloys with bismuth or antimony additions. Both Mg3Bi2 and Mg3Sb2 have a high thermal stability and play important roles in
improving creep resistance of the alloys at elevated temperatures. 2001 Elsevier Science B.V. All rights reserved.
Keywords: Antimony; Bismuth; Magnesium alloy; Mechanical properties; Microstructure

1. Introduction
Magnesium alloys are being increasingly used in the
automobile industry in order to save weight, thereby
reducing fuel consumption and pollution. Alloy AZ91
(Mg 9Al 0.8Zn 0.2Mn) is the most favored magnesium alloy, being used in approximately 90% of all
magnesium cast products [1]. It has excellent castability,
and sound room-temperature mechanical properties,
and in its high-purity form (AZ91E) shows a good
corrosion resistance. However, the range of applications for this alloy is limited by the low strength and
poor creep resistance at temperatures in excess of
120C, which makes it unsuitable for many of the
components in automobile engines [1,2]. Although
some magnesium alloys containing rare earth (RE)
elements developed in the past few years [2] have good
* Corresponding author. Tel.: +86-21-62932549; fax: + 86-2162932548.
E-mail address: gyyuan@mail1.sjtu.edu.cn (Y. Guangyin).

mechanical properties, even at moderate temperatures,

these alloys are more expensive than common Mg
alloys such as AZ91. In addition to developing new
alloy systems, optimization of existing alloys via traceelement additions is a powerful route to the improvement of the performance of magnesium as a structural
metal [3]. The work reported here is a part of the
program to develop cheap and creep-resistant magnesium alloys based on AZ91. In this paper, the effects of
a small amount of bismuth and antimony additions on
the microstructure and mechanical properties as well as
the creep resistance of the AZ91 alloy at elevated
temperatures up to 200C have been studied.

2. Experimental procedures
Eleven alloys (the compositions of which are listed in
Table 1) were prepared in a mild steel crucible under
the protection of a mixed gas atmosphere of SF6 (1
vol.%) and CO2 (bal.) using commercial stock. The base

0921-5093/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 5 0 9 3 ( 0 0 ) 0 2 0 4 3 - 8

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

39

3. Results

composition of the alloys studied was Mg 9Al

0.8Zn 0.2Mn (alloy1) conforming to that of AZ91.
Different amounts of bismuth and antimony were
added in the other alloys, so that the effects of bismuth and antimony addition and variation of their
concentrations on the microstructure and mechanical
properties could be studied. The melt was held at
720C for 20 min, and then poured into permanent
molds. Tensile specimens with a gage section of 15
mm 3.5 mm2 mm were cut by electric spark machining from the ingots. Before testing, all the specimens were heated at 420C for 12 h followed by
water quenching, then aged at 200C for 16 h (T6
condition). Microstructural observations of as-cast
and heat-treated specimens were conducted using optical and scanning electron microscopy (SEM). Microanalysis and determination of crystal structure of the
phases in the alloys studied were carried out by X-ray
energy-dispersive spectroscopy (XEDS) and X-ray diffraction (XRD), respectively.

3.1. Microstructure
The microstructure of as-cast alloy 1 (AZ91) consisted of the matrix (a-Mg) and b(Mg17Al12) precipitates, which were mainly distributed at grain
boundaries, as shown in Fig. 1a. Fig. 1b is a SEM
micrograph taken from as-cast alloy 4 containing 2%
of bismuth, from which it can be seen that the b
precipitates appeared finer than that shown in Fig.
1a. In addition, some rod-shaped particles distributed
both in grains and at grain boundaries were observed.
Microanalysis carried out by XEDS indicated that
these precipitates have an approximate composition of
Mg 38Bi 4Al 1Zn (wt.%). The X-ray diffraction
(XRD) pattern taken from the as-cast alloy 4 is
shown in Fig. 2, in which peaks can be indexed as
arising from three phases, a-Mg (matrix), b(Mg17Al12)
and Mg3Bi2, which has a hexagonal structure of D52
[4]. On the basis of the microanalysis and XRD pat-

Table 1
Chemical compositions and mechanical properties of the alloys investigated
Alloy code

1
2
3
4
5
6
7
8
9
10

11

Composition
(wt.%)

Mg9Al0.8Zn
0.2Mn
Mg9Al0.8Zn
0.2Mn0.5Bi
Mg9Al0.8Zn
0.2Mn1.0Bi
Mg9Al0.8Zn
0.2Mn2.0Bi
Mg9Al0.8Zn
0.2Mn3.0Bi
Mg9Al0.8Zn
0.2Mn0.1Sb
Mg9Al0.8Zn
0.2Mn0.4Sb
Mg9Al0.8Zn
0.2Mn0.7Sb
Mg9Al0.8Zn
0.2Mn1.0Sb
Mg9Al0.8Zn
0.2Mn0.5Bi
0.4Sb
Mg9Al0.8Zn
0.2Mn1.0Bi
0.4Sb

RT tensile

150C tensile

Ulti. (MPa)

Yield (MPa)

Elong. (MPa)

Ulti. (MPa)

Yield (MPa)

Elong. (MPa)

222

106

5.3

175

99

232

132

4.7

180

132

24.3

250

166

4.6

180

134

21.7

265

184

4.4

184

138

19.6

224

162

3.2

172

119

12.1

238

134

5.0

183

121

34.5

264

177

4.5

185

138

34

257

172

4.4

180

135

26.8

244

165

2.7

173

133

18.9

262

172

3.6

183

138

16.3

269

178

3.3

187

140

14.8

35

40

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

Fig. 1. Microstructure of as-cast alloys (a) alloy 1, (b) alloy 4, (c) alloy 7, (d) alloy 11.

Fig. 2. X-ray diffraction patterns of as-cast alloy 4.

Fig. 3. X-ray diffraction patterns of as-cast alloy 7.

tern, these rod-shaped precipitates were identified as

Mg3Bi2.
Similar microstructure was observed in the alloys
with antimony addition. A small amount of antimony
addition also caused the refinement of the b precipitates
and the formation of rod-shaped particles, as shown in

Fig. 1c, a SEM microgragh taken from as-cast alloy 7

containing 0.4% of antimony. Fig. 3 shows the XRD
pattern taken from as-cast alloy 7, in which some peaks
were indexed as arising from the Mg3Sb2 phase, which
has the same crystal structure, D52 [5], as that of
Mg3Bi2. Microanalysis revealed that the average com-

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

position of these rod-shaped particles was approximately Mg 40.34Sb 1.59Al (wt.%). Thus, these rodshaped particles can be identified as Mg3Sb2. With the
combined addition of bismuth and antimony, both
Mg3Bi2 and Mg3Sb2 particles formed in the microstructure. Fig. 4 shows the XRD pattern of as-cast alloy 11
containing 1% bismuth and 0.4% antimony and reveals
that both phases, Mg3Bi2 and Mg3Sb2, existed in the
alloy. Fig. 1d shows the as-cast microstructure of alloy
11. It can be seen that the b precipitates in this alloy
became much finer and more uniform than that in the
base alloy (Fig. 1a). Since the morphologies of Mg3Bi2
and Mg3Sb2 were very similar, it is difficult to identify
them unless a microanalysis were performed.

41

When the as-cast specimens of the alloys were solution-treated at 420C for 12 h (T4 condition), almost all
of the b phase (Mg17Al12) dissolved into the matrix, and
the microstructure of the alloy1 consisted of only the
a-Mg matrix. However, the solution treatment at 420C
did not result in dissolution of Mg3Bi2 or Mg3Sb2
particles in the alloys with bismuth or antimony additions, as shown in Fig. 5a and b (SEM micrographs
taken from specimens of alloy 4 and alloy 8 after
solution treatment). This indicated that both Mg3Bi2
and Mg3Sb2 have a high thermal stability at elevated
temperatures.
Another important effect of bismuth or antimony
addition on the microstructure was suppressing discon-

Fig. 4. X-ray diffraction patterns of as-cast alloy 11.

Fig. 5. Microstructure of solution-treated specimens (a) alloy 4 (T4 solution treatment) showing the rod-shaped Mg3Bi2 phase; (b) alloy 8 (T4
solution treatment) showing the rod-shaped Mg3Sb2.

Fig. 6. Microstructure of aged specimens: (a) alloy 1 (T6 treatment) showing a large amount of lamellar discontinuous precipitation (D.P.) at grain
boundaries; (b) alloy 4 (T6 treatment) showing plate-shaped continuous precipitates (C.P.) and rod-shaped Mg3Bi2 particle at the grain boundary.

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

42

Table 2
Creep data for magnesium alloys AZ91+Bi/Sb
Alloy code

Temperature (C)

Stress (MPa)

Creep rupture life, tf (h)

Elongation, l (%)

Creep rate, m; min (S1)

Alloy 1

150
150
175
200
200
200
150
150
175
200
200
200
150
150
175
200
200
200
150
150
175
200
200
200

50
70
50
30
40
50
50
70
50
30
40
50
50
70
50
30
40
50
50
70
50
30
40
50

\100a
267
35
\100a
\100a
14
\100a
402
\100a
\100a
\100a
75
\100a
764
\100a
\100a
\100a
108
\100a
956
\100a
\100a
\100a
138

2.4
27.0
5.0
2.5
6.0
18.0
1.5
14.0
1.7
1.7
1.8
15.5
1.4
16.4
1.5
1.3
2.0
13.0
0.8
9.4
1.1
0.9
1.2
10.0

5.0108
1.4107
3.4107
1.9107
8.9107
2.5106
9.6109
3.1108
6.8108
3.7108
1.9107
5.4107
4.7109
1.6108
3.5108
1.9108
9.9108
2.9107
3.1109
1.3108
2.3108
2.1108
7.5108
1.9107

Alloy 4

Alloy 7

Alloy 11

(100 h)

(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)
(100 h)

The test was stopped after creep at the temperature for 100 h.

tinuous precipitation of Mg17Al12. For the alloy AZ91,

aging at 200C caused the formation of Mg17Al12
precipitates, and the precipitation of Mg17Al12 occurred
in two forms, discontinuous precipitation at grain
boundaries and continuous precipitation within grains
[6]. In aged specimens of alloy1, which is free of
bismuth and antimony, the lamellar precipitation
(discontinuous precipitation) was dominant in the
vicinity of grain boundaries, as shown in Fig. 6a. With
bismuth or antimony addition, discontinuous
precipitation was greatly suppressed and fine
plate-shaped precipitates (continuous precipitation)
were observed at the grain boundaries (see Fig. 6b).

3.2. Mechanical properties

3.2.1. Tensile properties
Table 1 lists the tensile data for all specimens tested.
It can be seen that bismuth and antimony addition to
the AZ91 alloy had a significant influence on tensile
properties. At both ambient temperature and an elevated temperature of 150C, the ductility decreased
slightly, while the yield strength significantly increased
with increasing bismuth or antimony addition. The
maximums of yield strength were obtained from alloy 4
(containing 2% Bi), alloy 7 (containing 0.4% Sb) and
alloy 11(containing 1% Bi and 0.4% Sb), respectively. A
further increase of bismuth or antimony addition resulted in a slight reduction in yield strength. Therefore,

the amount of bismuth or antimony addition to the

AZ91-based alloys must be limited within a certain
range.

3.2.2. Creep beha6ior

The most important result from bismuth and antimony additions was in creep resistance. Creep tests
were performed on selected specimens of alloys 1, 4, 7,
and 11 in the T6 condition, over the temperature range
of 150 200C and at an applied stress of 3070 MPa.
The results are summarized in Table 2. It can be seen
that bismuth and antimony additions to the base alloy
resulted in a significant increase in creep resistance. At
a temperature of 200C and an applied stress of 50
MPa, the creep rupture life of the base alloy (alloy 1)
was only 14 h. With 2% bismuth (alloy 4) or 0.4%
antimony (alloy 7) addition, the creep rupture life increased to 75 and 108 h, respectively. A further improvement in creep resistance was obtained in alloys
with a combined addition of bismuth and antimony.
The creep rupture life of alloy 11 (containing 1% Bi and
0.4% Sb) was almost one order of magnitude higher
than that of its base alloy, and its steady creep rate
decreased to 1.910 7 s 1, one order of magnitude
lower than that of the base alloy. Typical creep curves
are shown in Fig. 7, from which the decrease in steadystate creep rate with bismuth and antimony addition
can be clearly seen.

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

43

Fig. 7. Typical creep curves of alloys investigated: (a) at 200C, 30 MPa; (b) at 200C, 50 MPa.

The results of creep tests also showed a linear relationship between log m; min vs. log | (Fig. 8) for the alloys
in T6 condition over the temperature and stress range
investigated. The variations in creep rate with the reciprocal temperature (1/K), are shown in Fig. 9 for the
four alloys tested at 50 MPa.

4. Discussion
At elevated temperatures, the steady-state creep rate
m; min as a function of stress and temperature is well
described by the relationship [7]:
m; min = A| n exp( Q/RT)

(1)

where A is a dimensionless constant, n is the constant

called the stress exponent, Q is the activation energy for
creep, R is the gas constant, and m; min is the steady-state
creep rate.
Examination of the variation of log m; min with log |
and temperature, on the basis of Eq. (1) and Fig. 8,
produces stress exponents of 5.2, 4.9, 4.6 and 4.4 for
alloys 1,4,7 and 11, respectively. These values are close
to the stress exponents of pure polycrystalline magnesium (56) obtained by Vagarali and coworkers [8].
The activation energies are found to be 129, 134, 137
and 140 kJ mol 1 for alloys 1, 4, 7 and 11, respectively
(see Fig. 9), which are all close to the activation energy
of self-diffusion in pure magnesium, 135 kJ mol 1 [9].
This indicates that the dislocation climb is the rate-controlling step in the creep of the alloys at 200C. The
activation energy of high-temperature creep can be
altered due to the solute atoms [9]. Microanalysis
showed that the solid solubility of Bi in the AZ91 alloy
was about 0.1 wt.%. Therefore, the higher activation
energies of alloys containing Bi comparative to AZ91
may be related to the existence of some Bi as solid
solution solutes, which may influence the mobility of
matrix atoms. Moreover, very recent aging studies carried out by the authors on AZ91-based alloys containing Bi showed that the Bi addition slowed down the
aging process, which also supports the above hypothe-

sis. For alloy 7 containing Sb, another mechanism may

exist because the solid solubility of Sb in AZ91 was
very low. However, we observed the presence of Sb in
the b-Mg17Al12 phase, which may contribute to the
stability of b phase and provide some resistance to
coarsening during elevated temperature creep. Further
investigations are under way.
For common Mg Al alloys, the tensile strength and
creep resistance at elevated temperatures are poor [2]. A
previous investigation [10] proposed that in Mg Albased alloys, the sliding of the grain boundaries is an
important part of the deformation mechanism at elevated temperatures. The discontinuous precipitation
has been considered to be responsible for excessive
deformation at elevated temperatures because the pres-

Fig. 8. Variation of creep rate with stress for alloys tested at 200C.

Fig. 9. Variation of creep rate with reciprocal temperature for alloys

tested at 50 MPa.

44

Y. Guangyin et al. / Materials Science and Engineering A308 (2001) 3844

ence of many lamellae (discontinuous precipitation) in

the vicinity of grain boundaries will provide more surfaces on which sliding can take place easily. In the
present investigation, the discontinuous precipitation of
Mg17Al12 in the AZ91 alloy was suppressed effectively
by the bismuth and antimony additions so that the
sliding of grain boundaries during the deformation of
tensile and creep tests at elevated temperatures was
reduced.
On the other hand, the b (Mg17Al12) precipitates play
an important role in strengthening the alloys at the
room temperature in Mg Al-based alloys. However,
the melting point (437C) of the b phase is low, and it
can readily soften and coarsen with increasing temperature due to accelerated diffusion [11]. Since the discontinuous precipitates b phase is mainly distributed at
gain boundaries, their softening and coarsening weaken
the grain boundaries at elevated temperatures and become the key factors accounting for the low creep
resistance of these alloys. Additions of bismuth and
antimony to the AZ91 alloy result in the formation of
rod-shaped particles (Mg3Sb2 and Mg3Bi2) that straddle
the grain boundaries, with some of them reaching into
both grains (see Fig. 5). The high melting points of
Mg3Sb2 (1228C) and Mg3Bi2 (823C) imply the high
stability of these particles in the interesting temperature
range, which has been confirmed as no apparent
changes in size or shape can be observed after solution
treatment at 420C for 12 h (see Fig. 5). Generally, the
most effective particles/precipitates are those that
coarsen slowly, thus providing a fine dispersion to resist
dislocation movement and grain boundary sliding at
elevated temperatures. In the present investigation, for
this reason, both Mg3Sb2 and Mg3Bi2 particles are
particularly effective in improving the creep resistance
of the alloys studied. Since the melting point of Mg3Sb2
is higher than that of Mg3Bi2, it seems more effective in
strengthening the alloys at elevated temperatures. TEM
investigations on the dislocation structure in the creep
specimens of the alloys studied are under way.

5. Conclusions
Small amounts of bismuth or antimony additions to
the Mg 9Al 0.8Zn(AZ91)-based alloys result in significant increases in yield strength and creep resistance

but slight decreases in ductility at elevated temperatures

up to 200C. The highest creep resistance has been
obtained from the alloy with combined additions of
small amounts of bismuth and antimony.
The activation energies of creep for the alloys studied
are similar to the energy of self-diffusion of pure magnesium. Therefore, dislocation climb is considered to be
the rate-controlling mechanism of steady-state creep of
AZ91-based alloys at 200C. The steady-state creep rate
exponents of alloys 1, 4, 7 and 11 are 5.2, 4.9, 4.6 and
4.4, respectively, which also support this conclusion.
Bismuth or antimony additions refine the b precipitates in the microstructure of as-cast alloys and suppress discontinuous precipitation of the b phase
effectively during aging process.
Bismuth or antimony additions cause the formation
of some rod-shaped Mg3Bi2 or Mg3Sb2 particles, both
of which have a high thermal stability at elevated
temperatures. The melting point of Mg3Sb2 is higher
than that of Mg3Bi2, so the addition of antimony seems
to be more effective than that of bismuth on improving
the creep resistance at elevated temperatures.

Acknowledgements
This research was supported by the Ministry of Education PRC. The authors wish to thank Professor
Wang Jiannong for his helpful discussions about the
creep mechanism.

References
[1] D.R. Lesuer, G.T. Kipoures, JOM 7 (1995) 17.
[2] A. Luo, M.O8 . Pekguleryuz, J. Mater. Sci. 29 (1994) 5259.
[3] M.O8 . Pekguleryuz, M.M. Avedesian, Proc. Magnesium Alloys
and their Applications, DGM, Germany, 1992, p. 213.
[4] ASTM X-Ray Powder Diffraction Date File 4-0464.
[5] ASTM X-Ray Powder Diffraction Date File 3-0375.
[6] E.F. Emley, Principles of Magnesium Technology, Pergamon
Press, London, 1966, p. 233.
[7] J. Cadek, Creep in Metallic Materials, Elsevier, Amsterdam,
1988.
[8] S.S. Vagarali, T.G. Langdon, Acta Metall. 29 (1981) 19691982.
[9] J.E. Dorn, Creep and Recover, American Society for Metals,
Metals Park, OH, 1957, pp. 255 283.
[10] P. Humble, Mater. Forum 21 (1997) 45 56.
[11] G.V. Raynor, The Physical Metallurgy of Magnesium and its
Alloys, Pergamon Press, London, 1959, p. 103.