a r t i c l e i n f o a b s t r a c t
Article history: A commercial AZ91 magnesium alloy (nominal composition Mg9%Al; 1%Zn; 0.3%Mn, balance Mg in
Received 20 April 2009 weight percent) reinforced with SiC particles and modied by the addition of Si has been used in this
Received in revised form 14 June 2009 study. Formation of an in situ composite (MgMg2Si) results in strong bonding between Mg2Si and
Accepted 19 June 2009
the matrix interface. Samples were deformed in compression in the temperature interval from room tem-
Available online 26 June 2009
perature up to 300 C. Stress relaxation tests were performed with the aim to reveal the thermally acti-
vated processes. Reinforcing effect of SiC and Mg2Si particles decreases with increasing temperature. The
Keywords:
estimated values of the activation volume as well as the activation enthalpy indicate that the main ther-
A. Metal matrix composites
B. Plastic deformation
mally activated process is connected with a rapid decrease of the internal stress. Fracture properties were
C. Stress relaxation studied in impact tests at various temperatures. A ductility enhancement was found at 200 C and tem-
D. Scanning electron microscopy peratures above 200 C.
E. Casting 2009 Elsevier Ltd. All rights reserved.
Magnesium alloy
Composite
Mechanical properties
Thermally activated process
Fractography
0266-3538/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compscitech.2009.06.016
Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264 2257
where G is the shear modulus, a1 is a constant describing interac- A commercially available AZ91 magnesium alloy (the nominal
tion between dislocations, b is the Burgers vector of dislocations composition Mg9%Al; 1%Zn; 0.3%Mn; balance Mg in weight per-
and qt is the total dislocation density. The effective shear stress r* cent) was reinforced with 10 vol.% of SiC particles (SiCp) and an
acts on dislocations during their thermally activated motion when addition of 3 wt.% of Si. The size of the SiCp exhibited about
they overcome short range obstacles. The mean velocity of disloca- 20 lm. Composite was prepared by the squeeze cast technology.
tions v is connected with the plastic strain rate by the Orowan The preform was inltrated by the liquid alloy using two-stage
equation: application of the pressure. The samples were subjected to a stan-
e_ 1=wqm bv ; 3 dard T6 thermal treatment (annealing for 18 h at 413 C in the pro-
tective argon atmosphere, then ageing for 8 h at 168 C [11])
where qm is the density of mobile dislocations and w is Taylor ori- Compression tests were carried out at an initial strain rate of
entation factor. The most common equation used in describing the 3.3 105 s1 over a wide temperature range from room temper-
average dislocation velocity as a function of the effective stress is an ature (RT) to 300 C. Dimensions of the samples used in compres-
Arrhenius type. The plastic strain rate e_ for a single thermally acti- sion tests were 5 5 10 mm3.
vated process can be expressed as: Sequential stress relaxation tests were performed at increasing
e_ e_ 0 expDGr =kT; 4 stress along a stressstrain curve. Duration of the SR was 300 s.
Ductility of the alloy at room temperature is very low and there-
where e_ 0 is a pre-exponential factor containing the mobile disloca- fore the SR tests were performed at elevated temperatures. On
tion density, the average area covered by the dislocations in every the other hand, recovery during the SR test was observed at
activation act, the Burgers vector, the vibration frequency of the dis- 300 C and hence, the results obtained at this temperature were
location line, and the geometric factor. T is the absolute tempera- not taken into account because the above given equations
ture and k is the Boltzmann constant. DG(r*) is the change in the describing the SR were derived under an assumption that the
Gibbs free enthalpy depending on the effective stress r* = r ri internal stress ri is constant during the SR. Only in the rst SR
and the simplest form is test at 300 C, recovery may be considered to be negligible, i.e.
DGr DG0 V r DG0 Vr ri : 5
ri is constant.
Components of the applied stress (ri, r*) were estimated using
Here DG0 is the Gibbs free enthalpy necessary for overcoming a Lis method [14,15]. The SR curves were tted to the power law
short range obstacle without the stress and V = bdLc is the activation function in the form:
volume where d is the obstacle width and Lc is the mean length of
1 1
dislocation segments between obstacles. r ri am 11m t t0 1m ; 10
The stress relaxation technique is very useful method to study
the thermally activated processes and to reveal the dominant pro- where a, t0 and m are tting parameters.
cess occurring during plastic deformation. In a stress relaxation The impact tests were performed by the Charpy pendulum in
(SR) test, the specimen is deformed to a certain stress, r0, then the temperature range from RT to 300 C. The samples with dimen-
the machine is stopped and the stress is allowed to relax. The stress sions of 10 10 55 mm3 were machined from the billet using
decreases with the time t. The specimen can be again reloaded to a standard tools. The U-type notch has been created by the electro
higher stress (load) and the SR test may be repeated. The time erosive cutting to 2 mm depth. High temperature tests were con-
derivative r_ = dr/dt is the stress relaxation rate and r = r(t) is ducted with the preheated samples in an electric furnace. The frac-
the ow stress at time t during the SR. The stress relaxation tests ture surfaces were investigated immediately after the impact tests
are very often analysed under the assumption that the SR rate is using a scanning electron microscope (SEM) in order to prevent
proportional to the strain rate e_ , according to [12], as: possible oxidation.
Samples for the microstructure analysis were prepared by the
e_ r_ =M; 6 conventional mechanical polishing and etching using glycol solu-
tion (1 ml HNO3, 24 ml water, 75 ml ethylene glycol). In order to
where M is the combined modulus of the specimenmachine set.
obtain detailed three-dimensional morphology the surfaces of
The stress decrease with the time during the SR can be described
samples were deeply etched using a mixture of 36 ml phosphoric
by the well known Feltham equation [13]:
acid (H3PO4) and 64 ml ethyl alcohol. The samples were etched
Drt r0 rt alnbt 1; 7 for 3060 s by an electrolytic method at the current
I = 0.05 mA cm2 and the voltage of 0.7 V. For the phase identica-
where r(0) r0 is the stress at the beginning of the stress relaxa- tion, a SEM analysis was performed on the samples that undergone
tion at time t = 0, both etching and deep etching procedure.
2258 Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264
Fig. 2. (a) SEM micrograph of initial microstructure etched using glycol solution; (b) Chinese script type particle; (c) c-phase (electron compound Al12Mg17) and lamellae
discontinuous precipitate (c + d); (d) primary Mg2Si coarse dendritic crystal entrapped by d-Mg.
Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264 2259
Fig. 3. SEM micrographs of the same places as shown in Fig. 2 after using deep etching.
Fig. 4. SEM micrographs represent SiC particle with serrated interface and primary Mg2Si coarse dendritic crystal after using glycol etching (a) and deep etching (b).
Fig. 5. The sample under observation using SEM energy dispersive analysis: (a) SEM image and (b) map of Al distribution. Note higher concentration of Al in the zones
adjacent to SiCp and no occurrence in the vicinity of Mg2Si, the concentration of Al in the c-phase (Al12Mg17) is appropriate.
2260 Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264
(MPa)
300
The true stresstrue strain curves obtained in compression at
various temperatures are shown in Fig. 6. Samples were deformed 200
either to fracture at temperatures up to 200 C or to predetermined
strain at higher temperatures. A at shape of the stressstrain 100
curves obtained at higher temperatures than 150 C is obvious
from Fig. 6, indicating the equilibrium between hardening and
softening. The temperature dependence of the yield stress as well 0
0 50 100 150 200 250 300 350
as the maximum stress is introduced in Fig. 7. A weak temperature
dependence of the yield stress up to 200 C is followed by the more temperature (C)
rapid decrease at higher temperatures (Fig. 7). It is interesting to
Fig. 7. Temperature dependence of maximum stress and yield stress.
note that the softening and the transition from brittle to ductile
behaviour above 200 C corresponds with the onset of plasticity
of Mg2Si as published by Takeuchi et al. [20] (see also [21]). How-
ever, the dislocation glide in non-basal slip planes in the Mg ma- Bf DaDT 1
qT ; 12
trix, which is usually observed at higher temperatures (higher b1 f t0
than about 200 C) [2224], may involve not only the increased where t0 is the minimum size of the reinforcing phase particles, b is
compression plasticity but also the observed transition from brittle the magnitude of the Burgers vector of the newly created disloca-
to ductile behaviour as well. The Mg matrix with the hexagonal tions, B is a geometrical constant, Da is the difference between
structure provides only two independent basal slip systems at the two thermal expansion coefcients and DT is the temperature
ambient temperature. Non-basal slip systems activity strongly de- variation. A higher dislocation density in the composite material
pends on structural parameters like grain size, texture, alloying yields in a higher level of the internal stress. For the strengthening
elements and strain gradients which may arise in the vicinity of due to the presence of dislocations generated by the differential
inclusions. A comparison of the temperature dependence of the CTE, the following equation may be used:
yield stress of the alloy and composite is introduced in Fig. 8. As
it follows from Fig. 8, the addition of the reinforcing phase in- DrT a1 wGbqT 1=2 : 13
creases the yield stress of the monolithic alloy. The difference be-
Strengthening mechanisms in MgSi alloys prepared by rapid
tween the yield stresses of the composite and alloy decreases
solidication procedure were studied by Mabuchi and Higashi
with increasing temperature. The composite strengthening de-
[28]. They calculated various contributions to strengthening and
pends on the geometrical parameters of composites (size and
found that the main strengthening effect has the small grain size
shape of bres or particles, volume fraction of the reinforcement,
and the small Mg2Si particles. Rapid decrease of the strength at
interparticle distance), on the physical properties of the all compo-
temperatures higher than 200 C has been attributed to the coars-
nents (thermal expansion coefcient, bonding at the matrixrein-
ening of particles. Larger particles (>110 lm) or very small parti-
forcement interface), microstructure and mechanical properties
cles (<50 nm) are required to attain high strength at elevated
of all components [25].
temperatures above 200 C.
Metal matrix composites are characterized by a large difference
The reinforcing phase contributes to strengthening mainly
in the thermal expansion coefcient (CTE) of the matrix and cera-
through an increased dislocation density (arising from the ther-
mic reinforcement (CTE of Mg exhibits 26 106 K1 [20],
mal and geometrical mismatch). The load transfer from matrix
CTE(SiC) = 6.6 106 K1 [26] and CTE(Mg2Si) = 7.6 106 K1
to reinforcing particulates plays in this composite only a mar-
[19]). Even small temperature changes generate thermal stresses
ginal role.
in the matrix. Thermal stresses can be partially released by disloca-
tion generation in the vicinity of the interface. The density qT of
thermal dislocations in the matrix can be calculated [27].
200
400 RT 50C AZ91
100C AZ91+SiC+Si
160
300
150C
02 (MPa)
(MPa)
120
200
200C
250C 80
100
300C
0 40
0.00 0.04 0.08 0.12 0 50 100 150 200 250 300 350
temperature (C)
Fig. 6. Compression true stresstrue strain curves at initial strain rate Fig. 8. Temperature dependence of the yield stress of composite in comparison
e_ 8:3 105 s1 . with the unreinforced alloy.
Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264 2261
300 40
150C 35
250
ap 150C
30
200 250C
(MPa)
25
V/b3
150
i 20
100
15
50
10
0 5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Fig. 9. A part of the true stresstrue strain curve obtained at 150 C. The points of Fig. 11. Strain dependence of activation volume obtained for two temperatures.
rap on the curve indicate the stresses at which the SR tests were performed, ri and
r* are the internal and effective stresses.
atures are plotted against the effective stress, r*, the values appear
to lie at one line master curve (Fig. 12). Kocks et al. [30] sug-
3.3. Stress relaxation tests
gested an empirical equation between the Gibbs enthalpy DG
and the effective stress r* in the following form:
A part of the true stresstrue strain curve of the alloy measured
p q
at 150 C with points indicating the stresses at which the SR tests r
DG DG 0 1 ; 14
were performed is shown in Fig. 9. Components of the applied
r0
stress the internal stress, ri,, and the effective stress, r*, are given
in Fig. 9. It is obvious that the internal stress ri is a substantial con- where DG0 and r0 are Gibbs enthalpy and the effective stress at 0 K.
tribution to the applied stress rap at the onset of the stressstrain For the effective stress it follows:
curve. The internal stress, ri, increases with increasing strain up to " 1=q #1=p
kT e_ 0
approximately 5% of the plastic strain, than decreases with increas- r r 0 1 ln ; 15
ing strain. On the other hand, the effective stress component DG0 e_
increases with increasing strain and at e = 0.25 ri = r*. Similar
where p and q are phenomenological parameters reecting the
curves estimated at 250 C are shown in Fig. 10. The effective stress
shape of a resistance obstacle prole.
component is higher than the internal stress and it increases with
The possible ranges of values p and q are limited by the condi-
increasing strain while the internal stress decreases with increas-
tions 0 < p 6 1 and 1 6 q 6 2. Ono [31] suggested that Eq. (14) with
ing strain for e > 0.007. The internal stress of several magnesium
p = 1/2, q = 3/2 describes a barrier shape prole that ts many
alloys has been studied by Milicka et al. [29] in creep experiments.
predicted barrier shapes. Thermodynamics generally denes the
They found that the internal stress ri reects the creep resistance
activation volume as
of the materials. Experimental internal stresses determined in
creep well correspond to those determined in SR tests under com- dDG d ln e_ d lnr_
V kT kT : 16
parable testing conditions. dr T
dr dr
T
The activation volumes V were estimated using Eqs. (7) and (8).
Eq. (16) can be rewritten as
As usual, the values of V divided by b3 for samples deformed at 150
and 250 C are plotted against strain in Fig. 11. It can be seen that p q1 p1
DG0 pq r r
the activation volumes decrease with increasing strain for both V 1 : 17
r0 r0 r0
temperatures measured. If the activation volumes for both temper-
50
250C
120
40
ap 150C
(MPa)
250C
80 30
* fit
V/b3
20
40 i
10
0
0.00 0.04 0.08 0.12 0.16 0
20 40 60 80 100 120
* (MPa)
Fig. 10. A part of the true stresstrue strain curve obtained at 250 C. The points of
rap on the curve indicate the stresses at which the SR tests were performed, ri and Fig. 12. The plot of the activation volume against the thermal stress r* estimated
r* are the internal and effective stresses. for two deformation temperatures and t of all experimental points using Eq. (17).
2262 Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264
Fig. 13. Fractographic observation (SEM): (a) surface perpendicular to the fracture surface of specimen fractured at RT shows the Mg2Si dendrite with many microcracks (see
arrows); (b) cleavage fracture of primary Mg2Si dendritic crystal entrapped by d-phase, fractured at 300 C; (c) and (d) the opposite sites of fractured specimen at 200 C with
pyramid-like shape of primary Mg2Si dendritic crystal (c) and rest of Mg2Si in opposite site of fracture surfaces (d).
The values of the activation volume should lie at the curve given 3.4. Impact tests
by Eq. (17). Using binominal expansion in Eq. (17), the activation
volume should depend on the effective stress Vapp / (r*)n. All data The observation of fracture surfaces may help to identify operating
estimated for the activation volume at both temperatures studied fracture mechanism (s) of constituents at different temperatures and
are introduced in Fig. 12 together with the t according to Eq. the bonding abilities of their interfaces. The investigation was concen-
(17). The activation enthalpy calculated from Eq. (17) has been trated on the U-notch side of the fracture surface, where bending stres-
estimated to be DG0 = 0.95 0.05 eV. Similar activation enthalpy ses are assumed to be generated during the impact testing. Specimens
was found for AX41 magnesium alloy reinforced by 13 vol.% of were fractured predominantly in a brittle manner even at 300 C. The
short Safl bres [32]. The values of the activation volume and measured fracture energies were almost identical for all temperatures
the activation enthalpy may help to identify thermally activated tested (about 1 J). This implies that on one hand the composite parti-
processes. The dislocationdislocation interaction mechanism has cles contribute to the stiffness, and on the other hand they serve as
an activation volume ranging between 102104b3, with the activa- the crack initiators. The main features of fracture surfaces in the inves-
tion volume and enthalpy varying with strain. A rapid decrease in tigated temperature range were similar: brittle fracture of SiCp and
the internal stress with temperature (see Fig. 10) indicates that Mg2Si phases (see Figs. 13 and 14) and slightly ductile fracture of the
softening is connected with dynamic recovery. d-phase (see Figs. 15 and 16).
Fig. 14. (a) Microcracks (arrows) between matrix and SiC particles near the fracture surface at RT; (b) brittle fracture of SiCp (300 C) with serrated interface and (c) teeth in
the surface after delamination of SiCp at RT (SEM).
Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264 2263
Fig. 16. Comparison of d-phase breakage at RT (a) and 300 C (b), mechanism for both is transcrystalline ductile fracture with plastically transformed walls (SEM).
2264 Z. Trojanov et al. / Composites Science and Technology 69 (2009) 22562264
The strength improvement is very probably due to the increase [8] Raynor GV. The physical metallurgy of magnesium and its alloys. Pergamon
Press; 1959. p. 33.
of dislocation density (thermal and geometrically necessary
[9] Okamoto H. In: Massalski TB, editor. Binary alloy phase diagrams. ASM
dislocations). International; 1990.
[10] Nayeb-Hashemi AA, Clark JB. MgSi (MagnesiumSilicon). In: Massalski TB,
Stress relaxation tests: editor. Binary alloy phase diagrams. ASM International; 1990.
[11] Polmear IJ. Magnesium alloys and applications. Mater Sci Technol
1994;10:116.
The internal stress estimated from the stress relaxation tests [12] Dotsentko VI. Stress relaxation in crystals. Phys Status Solidi (b)
decreases with increasing deformation temperature. 1979;93:1143.
[13] Feltham P. Stress relaxation in magnesium at low temperatures. Phys Status
The values of the apparent activation volumes are in the order of Solidi 1963;3:13406.
101102b3. [14] Li JCM. Dislocation dynamics in deformation and recovery. Can J Appl Phys
1967;45:493509.
[15] De Batist R, Callens A. On the analysis of stress relaxation experiments. Phys
Fracture properties: Status Solidi (a) 1974;21:5915.
[16] Jiang QC, Wang HY, Wang Y, Ma BX, Wang JG. Modication of Mg2Si in MgSi
The composite studied is very brittle at all temperatures as alloys with yttrium. Mater Sci Eng A 2005;392:1305.
[17] Crawley AF, Milliken KS. Precipitate morphology and orientation relationships
proved by impact tests. in an aged Mg9% Al1% Zn0.3% Mn alloy. Acta Metall 1974;22:55762.
The Mg matrix exhibits a slightly plastically deformed fracture [18] Candan S. Effect of SiC particle size on corrosion behavior of pressure
surface at and above 200 C. inltrated Al matrix composites in a NaCl solution. Mater Lett
2004;58:36015.
The main contribution to fracture comes from the primary Mg2Si
[19] Liu Y, Liu X, Xiufang B. Grain renement of MgAl alloys with Al4C3SiC/Al
dendritic crystals and from the matrix/SiCp interface. master alloy. Mater Lett 2004;58:12827.
The effect of SiCp seems to be harmful rather than benecial. [20] Takeuchi S, Hashimoto T, Suzuki K. Plastic deformation of Mg2Si with the C1
structure. Intermetallics 1996;4:S14750.
[21] Lee JC, Byun JY, Park SB, Lee HI. Prediction of Si contents to suppress the
Acknowledgements formation of Al4C3 in the SiCp/Al composite. Acta Mater 1998;46:177180.
[22] Yoshinaga H, Horiuchi R. On the ow stress of a solid solution Mg Li alloy
single crystals. Trans JIM 1963;4:13441.
This work received a support from the Ministry of Education, [23] Courret A, Caillard D. An in situ study of prismatic glide in magnesium I. The
Youth and Sports of the Czech Republic by the project MSM rate controlling mechanism. Acta Metall 1985;33:144754.
0021620834. The authors are grateful also for the support offered [24] Courret A, Caillard D. An in situ study of prismatic glide in magnesium II.
Microscopic activation parameter.. Acta Metall 1985;33:145562.
by the Czech and Slovak authorities under the Exchange Pro- [25] Trojanov Z, Grtnerov, Drozd Z, Szraz Z, Lukc P. Short bres and particles
gramme MEB 0808116. P.P. and M.C. would like to acknowledge impact on compressive behaviour of an AZ91 alloy. Inter Foundry Res
the nancial support of the Slovak Grant Agency for Science under 2008;60:4452.
[26] Lide DR. editor. Handbook of chemistry and physics. 73rd ed.; 19921993.
the contract VEGA 1/0249/09.
[27] Arsenault RJ, Shi N. Dislocation generation due to differences between the
coefcients of thermal expansion. Mater Sci Eng 1986;81:17587.
References [28] Mabuchi M, Higashi K. Strengthening mechanisms of MgSi alloys. Acta Mater
1996;44:46118.
[1] Trojanov Z, Grtnerov V, Lukc P, Drozd Z. Mechanical properties of Mg [29] Milicka K, Trojanov Z, Lukc P. Internal stresses during creep of magnesium
alloys composites reinforced with short Safl bres. J Alloys Compd alloys at 523 K. Mater Sci Eng A 2007;462:2159.
2004;378:1926. [30] Kocks UF, Argon AS, Ashby MF. Thermodynamics and kinetics of slip. Prog
[2] Trojanov Z, Szraz Z, Lbr J, Lukc P. Deformation behaviour of an AS21 alloy Mater Sci 1975;19:1288.
reinforced by short Safl bres and SiC particles. J Mater Sci Technol [31] Ono K. Temperature dependence of dispersed barrier hardening. J Appl Phys
2005;162163:1318. 1968;39:18036.
[3] Hassan SF, Gupta M. Development of high performance magnesium nano- [32] Trojanov Z, Lukc P, Szraz Z. Strengthening and recovery in AX41
composites using nano-Al2O3 as reinforcement. Mater Sci Eng A magnesium alloy reinforced with short Safl bres. Kovov Mater
2005;392:1638. 2007;45:28391.
[4] Zheng MY, Wu K, Liang M, Kamado S, Kojima Y. The effect of thermal exposure [33] Grtnerov V, Chalupov M, Jger A, Trojanov Z. Inuence of temperature on
on the interface and mechanical properties of Al18B4O33w/AZ91 magnesium fracture mechanisms of magnesium composites. Mat Sci Forum
matrix composite. Mater Sci Eng A 2004;372:6674. 2005;482:3758.
[5] Ho KF, Gupta M, Srivatsan TS. The mechanical behavior of magnesium alloy [34] Balk J, Lukc P, Bohlen J, Kainer KU. Effect of temperature on tensile properties
AZ91 reinforced with ne copper particulates. Mater Sci Eng A of hot rolled AZ31 Mg alloy sheets. Kovove Mater 2007;45:13540.
2004;369:3028. [35] Yoo MH, Agnew SR, Morris JR, Ho KM. Non-basal slip systems in HCP metals
[6] Trojanov Z, Drozd Z, Kdela S, Szraz Z, Lukc P. Strengthening in MgLi and alloys: source mechanisms. Mater Sci Eng A 2001;319321:8792.
matrix composites. Compos Sci Technol 2007;67:196573. [36] Palcek P, Chalupov M, Hadzima B. Cracks propagation in AZ91 and AZ31
[7] Trojanov Z, Lukc P, Szraz Z. Strengthening and recovery in AX41 magnesium magnesium alloys. In: 21th Symposium DanubiaAdria on experimental
alloy reinforced with short Safl bres. Kovove Mater 2007;45:28391. methods in solid mechanics, Brijuni/Pula, Croatia, September 2004.