Вы находитесь на странице: 1из 14

J Mater Sci (2013) 48:41914204

DOI 10.1007/s10853-013-7232-x

Prediction models for the yield strength of particle-reinforced


unimodal pure magnesium (Mg) metal matrix nanocomposites
(MMNCs)
Chang-Soo Kim Il Sohn Marjan Nezafati J. B. Ferguson Benjamin F. Schultz
Zahra Bajestani-Gohari Pradeep K. Rohatgi Kyu Cho

Received: 13 December 2012 / Accepted: 11 February 2013 / Published online: 23 February 2013
 Springer Science+Business Media New York 2013

Abstract Particle-reinforced metal matrix nanocomposites (MMNCs) have been lauded for their potentially
superior mechanical properties such as modulus, yield
strength, and ultimate tensile strength. Though these
materials have been synthesized using several modern
solid- or liquid-phase processes, the relationships between
material types, contents, processing conditions, and the
resultant mechanical properties are not well understood. In
this paper, we examine the yield strength of particle-reinforced MMNCs by considering individual strengthening
mechanism candidates and yield strength prediction models. We first introduce several strengthening mechanisms
that can account for increase in the yield strength in
MMNC materials, and address the features of currently
available yield strength superposition methods. We then
apply these prediction models to the existing dataset of
magnesium MMNCs. Through a series of quantitative
analyses, it is demonstrated that grain refinement plays a
significant role in determining the overall yield strength of
most of the MMNCs developed to date. Also, it is found
that the incorporation of the coefficient of thermal expansion mismatch and modulus mismatch strengthening
C.-S. Kim (&)  M. Nezafati  J. B. Ferguson 
B. F. Schultz  Z. Bajestani-Gohari  P. K. Rohatgi
Materials Science and Engineering Department, University
of Wisconsin-Milwaukee, Milwaukee, WI 53211, USA
e-mail: kimcs@uwm.edu
I. Sohn
Materials Science and Engineering Department,
Yonsei University, Seoul 120-749, South Korea
K. Cho
U.S. Army Research Laboratory, Weapons and Materials
Research Directorate, Aberdeen Proving Ground,
MD 21005, USA

mechanisms will considerably overestimate the experimental yield strength. Finally, it is shown that work-hardening during post-processing of MMNCs employed by
many researchers is in part responsible for improvement to
the yield strength of these materials.

Introduction
Recent advances in nanotechnology have enabled the
incorporation of nano-sized reinforcements in metal matrix
composite materials in an effort to achieve mechanical
performances (i.e., strength and modulus) that were unattainable through the conventional micron-sized reinforcements. These MMNCs containing nanoparticles (NPs) have
been claimed to exhibit several potentially advantageous
properties that might enable their use in applications for
automotive, aerospace, construction, and military sectors.
Aluminum (Al) and magnesium (Mg) are most commonly
used as base matrix materials in light-weight MMNCs as
they offer high specific mechanical properties due to their
low densities (2.7 and 1.7 g/cm3 for Al and Mg, respectively) when compared to iron. The development and
application of high strength and light-weight MMNC
materials could have significant impact as a replacement
for heavier traditional metals or composites with resultant
savings in fuel economy. For these reasons, various
materials, synthesis, and processing methodologies and
conditions have been proposed to develop high strength,
particle-reinforced MMNCs [115].
To design and synthesize MMNCs with desirable properties, it is first necessary to develop a theoretical understanding of how the various potential strengthening
mechanisms combine to determine the strength of these
materials and under which conditions the desired

123

4192

J Mater Sci (2013) 48:41914204

microstructures can be produced. However, despite the


extensive efforts that have been made to synthesize MMNCs
with enhanced strength, a generally agreed upon theory to
describe the strengthening mechanism of these MMNCs has
not yet emerged. In the present work, we revisit the individual
strengthening mechanisms that may be present in particlereinforced MMNCs, and we test the previously proposed
theoretical models to predict the strength of MMNCs. Then,
we provide a thorough analysis based on the results from
prediction models and experimental observations. In
Strengthening mechanisms section, we review the candidate strengthening mechanisms that account for the various
aspects of strengthening of particle-reinforced MMNCs.
Next, various summation methods of contributions from
individual strengthening mechanisms to predict the strength
of synthesized MMNCs are introduced, and predicted
strengths using analytical models are compared with experimental measurements. In the current work, we have focused
on unimodal MMNCs in examining the yield strength
improvement. Multimodal MMNCs (e.g., bimodal or trimodal MMNCs) that are composed of fine and coarse matrix
crystals were not considered here because the strength of such
multimodal materials is routinely and relatively accurately
estimated by the rule of mixtures applied to the properties of
the fine and coarse phases. In our analysis, we primarily
employed the experimental data available for pure Mg
MMNCs. We analyzed the experimental data for pure Mg
MMNCs because, although there is a set of data for alloyed
Mg MMNCs [8, 10], we have only a few data points for each
of the alloy type. We, therefore, note that some of the material
parameters addressed in strengthening mechanisms and
strength prediction models are specifically relevant to predict
the properties of pure Mg-based MMNCs.

Strengthening mechanisms
Strength enhancement in MMNCs is commonly attributed
to the following three mechanisms: grain refinement,
Orowan strengthening, and coefficient of thermal expansion (CTE) mismatch (a.k.a., Forest) strengthening.
Grain refinement strengthening (HallPetch
strengthening)
There is evidence that the introduction of NPs into a metal
matrix either by liquid- or solid-state processing results in a
grain refinement effect [16, 17]. The HallPetch empirical
relation has traditionally been used to describe the effect of
average grain size (D) on the yield strength (ry) and/or
hardness (H) of metallic polycrystalline materials, e.g.,
k

ry r0 pyD ; where r0 and ky are experimentally

123

determined parameters. The parameter r0 represents the


yield strength of a single crystal in the absence of any
strengthening mechanisms except the solid solution effect,
and ky is the magnitude by which crystal boundaries in a
polycrystalline material resist slip. In cases where the
addition of particles reduces the size of the grains in the
MMNCs compared to the un-reinforced metal processed
under the same conditions, there will be an improvement to
yield strength due to grain refinement, DrGR:


1
1
DrGR ky p  p
1
DMMNC
D0
where DMMNC and D0 are the average grain diameters of
polycrystalline matrix materials in the MMNC composite
and unreinforced material, respectively. This equation
assumes that the HallPetch parameters ky and r0 remain
unchanged in the composites during processing. Unfortunately, the HallPetch parameters in a given alloy in a
given state cannot be predicted, and must determined
empirically for a given microstructure taking into consideration of such factors as precipitation condition and
increased dislocation density or texturing resulting from
mechanical post-processing. Thus, for the empirically
determined parameters to be valid for the purpose of calculating the grain refinement strength contribution, only the
grain size can vary and all other influences on the strength
of the material must, as much as possible, remain
unchanged [1820].
Table 1 shows the list of available HallPetch parameters for pure and alloyed Mg materials at room temperature
with available grain sizes that were synthesized by a
variety of processing routes, e.g., casting, rolling, extrusion, equal-channel angular processing (ECAP), powder
metallurgy (PM), etc. [19, 2130]. As demonstrated in
Table 1, it is generally accepted that r0 and ky values for
pure Mg materials fall in the ranges of 17.735 MPa and
p
250300 MPa lm; respectively, with few exceptions. In
the case of Mg alloys, the values of r0 and ky varied more
p
widely in the range of 3130 MPa and 150350 MPa lm;
respectively.
While it can be shown that the addition of NPs to a
metal matrix acts to refine the grain size [16, 17], it should
be noted that the NPs may only restrict the size of the
grains in the MMNC rather than nucleate new grains. For
such indirect impacts of NP addition, the following Zener
formula [31] is sometimes used to describe the grain
boundary pinning of added particles during processing and
post-processing (e.g., hot extrusion).
Dm

4adp
3Vp

where a is a proportionality constant, and Dm, dp, and Vp


are the smallest grain size of matrix, reinforcement particle

J Mater Sci (2013) 48:41914204

4193

Table 1 HallPetch parameters for pure and alloyed Mg materials


Alloy type

Processing route

Pure Mg

AZ31

Grain size (lm)

r0 (MPa)

p
ky (MPa lm)

Ref

Rolled

43172

35

291

[21]

Rolled

16

278

[22]

Extruded

1389

158

[22]

Cast

361500

17.7

250

[23]

Rolled

1.4140

10131

160348

[24]

Extruded

2.511

80130

182303

[24]

ECAP

232

3085.2

170205

[24, 25]

EBW

1140

62

202

[19, 24]

FSP

2.66.1

10

161

[25]

FSW

16.4

119.5

[26]

AZ91

Extruded

0.33.7

133

[27]

Mg2Zn

Cast
Cast

0.35
55340

3.1

166
470

[27]
[28]

Extruded

[29]

MgAlMnCa

Cast

23.3

192

[29]

Powder metallurgy

0.40.5

[29]

ZM21

725

154

80

[23]

Mg97Zn1Y2

Extruded

49

188

[30]

ZK30, ZE10, ZEK100

Hydrostatic procedure

320

200

210

[23]

AZ61

Extruded

8150

348

[22]

1100

247

[22]

ECAP equal-channel angular processing, EBW electron beam welding, FSP friction stir processing, FSW friction stir welding

diameter, and volume fraction of particles, respectively.


However, it is highly questionable if Eq. (2) can be applied
directly to the processing of MMNCs. Rather, it is
empirically proposed that Dm will follow Eq. (3) that
does not explicitly include the effect of particle diameter,
dp, but is instead based on the grain size, D0, that results
from processing conditions when particles are absent. This
has been shown to provide a reasonably good prediction of
resultant grain size for a specific cooling rate in the liquidphase processing [17].
Dm 

D0
1 pVp

13

where p is the proportionality constant that describes the


refining power of the reinforcement, which can be empirically determined.
Orowan strengthening
Orowan strengthening, sometimes called second-phase
particle strengthening, is another primary mechanism that
could result in considerable strength increases in particlereinforced MMNCs. The Orowan strengthening model
describes improvements in strength in the matrix material
based on the resistance of small hard particles against the

motion of dislocations. The dislocation movement proceeds with passing by these NP obstacles by first bowing,
then reconnecting, and finally forming a dislocation loop
around the particles. Formation of these loops in principle
leads to high work-hardening rates with improved strength
[3235]. In composites reinforced by spherical particles,
the maximum tensile and shear stresses occur at the surface
of the particles and they decrease with distance from the
surface of particles. Since the Orowan-type of strengthening is observed when the reinforcement particles are sufficiently small, it is proposed that the strength increases are
only effective for the particles with sizes under 1 lm [36].
The original Orowan formulation predicts an inverse relationship between the strength increase (DrOrowan) and the
interparticle mean free path (k) given by:
DrOrowan /

Gm b
k

where Gm and b are the shear modulus of matrix and the


magnitude of Burgers vector, respectively. The Ashby
Orowan formulation is a more advanced revision to Eq. (4)
that is widely used in predicting the strength improvements
[37] and takes the following form:
 
dp
AGm b
DrOrowan
ln
5
2pk
r

123

4194

J Mater Sci (2013) 48:41914204

where A is a constant depending on the materials and types


of dislocations (e.g., edge or screw), dp is the reinforcement
particle diameter, and r is the dislocation core radius (i.e.,
the size of the dilated or diminished lattice with a dislocation), respectively.
Several variations of the general form of the Ashby
Orowan equation (Eq. (5)) have been proposed to account
for the proportionality constant, particle distributions, and
lattice dilation effects in various materials [3234, 3843].
Table 2 summarizes the list of AshbyOrowan-type equations (Eqs. (6) through (10)) that have been previously
proposed for predicting strengthening effects of composites
based on pure or alloyed Mg materials. For most of these
relations, it was assumed that the reinforcement particles
are randomly distributed in the matrix, and one particular
slip system is uniquely dominant and more highly stressed
than others. In addition, to simplify the mathematical calculations, the particles were assumed to be spherical;
otherwise, the derived formula must be corrected incorporating the particle geometry factor [37, 42]. Among the
expressions listed in Table 2, Eq. (6) along with the corresponding interparticle spacing (Eq. (6a)) is the most
widely used formula to estimate the Orowan-type contribution to the yield strength improvement of particle-reinforced MMNCs [33, 38, 39]. The activation of the Orowan
strengthening mechanism is particularly important because
it does not necessarily involve the reduction of ductility of

MMNCs as other dislocation density-based strengthening


mechanisms, such as CTE and/or modulus mismatch
mechanisms do.
Coefficient of thermal expansion mismatch
strengthening
When a MMNC is quenched from the processing temperature
to room temperature, volumetric strain mismatch between the
monolithic matrix and reinforcement particles may occur due
to differences in CTE, which will subsequently produce
geometrically necessary dislocations (GND) around reinforcement particles to accommodate the CTE difference.
This mechanism is based on Taylor strengthening that
describes the increase in the flow stress in the matrix due to
the presence of the GND. When the length of a generated
dislocation loop is assumed as pdp, the CTE mismatch
strengthening (DrCTE) can be estimated by [44, 45]:
p
p
11
DrCTE 3bGm b qCTE ;
qCTE

12Vp DaDT
1  Vp bdp

11a

where b is the dislocation strengthening coefficient, qCTE is the


density of dislocations generated from the CTE difference, Da
is the CTE difference between the matrix and the reinforcement particles, and DS is the temperature change, respectively.

Table 2 Orowan equations and material parameters used for particle-reinforced Mg-based alloys or MMNCs in the previous papers [3234,
3842]
Formula
d

mb
ln 2bp (6)
DrOrowan 0:13G
k

In references [38, 39],


k dp 1=2Vp 1=3  1

6a

h
i

mb
ln rD0 B 7
DrOrowan 6AG
4pk

Composition

b (nm)

dp
(nm)

Pure Mg (4.9 wt% C-coated Ni)

0.35

20

99.9 % Mg

Pure Mg (99.8 %)

0.32

0.30

100

Mg, 5.25 wt% Y, 3.5 wt% RE,


0.5 wt% Zr

0.321

0.27

AZ91

0.321

Mg5 wt% Zn

0.069 10
11

0.29

Mg3.0 at.% Zn

0.32

ZK60

Vp
(vol%)

Gm
(GPa)

Ref

0.4

11.60

[38]

[39]

16.5

[33]

[32]

3.10

45

[42]

15

[40]

0.28

16.6

[41]

[34]

0.88

AZ31 Mg (8 wt% Al2O3)

(A = 1/(1 - m), B = 0.6 for screw, A = 1, B = 0.7


for edge dislocations)
1=2

DrOrowan
DrOrowan

1:33Gm bVp
dp

(8)


Gp
b
m
2pk 1m

dp
b

9


p  1 dp 9a
In Ref. [40], k 0:953
ln

Vp

lndb
0:4Gm b p

p
k
1m

10
DrOrowan M
q

q
p

1
10b
d 23dp 10a;k d
4Vp

b, m, dp, k, Vp, and Gm denote the Burgers vector, Poissons ratio, average particle diameter, interparticle spacing, particle volume fraction, and
shear modulus, respectively

123

J Mater Sci (2013) 48:41914204

4195

The activation of CTE mismatch strengthening mechanism in MMNCs is, however, unclear because many
observations report the absence of CTE effects on the
strength improvement of MMNCs [4548]. For example,
Vogt et al. [45] tested the true stressstrain curves of
MMNC specimens with various heat treatments and
quenching conditions and they confirmed that the CTE
mismatch strengthening does not occur in MMNCs produced by PM techniques. Also, Redsten et al. [46] have
proposed that CTE mismatch strengthening can be ignored
below a critical reinforcement particle size limit, which, in
Mg/Y2O3 and Al/Al2O3 composites, corresponds to reinforcement particle sizes of approximately *70 and
*80 nm, respectively, when the processing temperature is
assumed as 300 C. Therefore, based on their work, if the
reinforcement particle size is small enough (i.e., below the
range of 7080 nm), the contribution from CTE mismatch
strengthening may be considered negligible or minor
compared with those from grain refinement or Orowan
strengthening mechanisms.
Other strengthening mechanisms
In addition to grain refinement, Orowan, and CTE mismatch mechanisms, the modulus mismatch and load-bearing strengthening mechanisms have been offered as
mechanisms capable of improving the strength of MMNCs.
The modulus mismatch strengthening mechanism also
describes the generation of GND when a MMNC is subjected to a compressive loading such as in hot extrusion.
Because of the presence of reinforcement NPs, many GND
must be created to accommodate the moduli difference
between the matrix and particles, and distortion deformation subsequently results during the post-processing.
Therefore, this modulus strengthening must be applied only
to the MMNCs where any suitable post-processing is
involved including compressive loading. If we again
assume that the length of a generated dislocation loop is
pdp, then the strength improvement by the modulus mismatch (DrModulus) is approximated by [44]:
DrModulus
qModulus

p
p
3aGm b qModulus ;

6Vp
e
bdp

12
12a

where a is the material-specific coefficient, qModulus is the


density of GND generated by the modulus mismatch
mechanism, and e is the bulk strain of composites,
respectively. Activation of the modulus mismatch
strengthening mechanism is, however, also unclear for
reasons similar to those described in the CTE mismatch
strengthening section.

The load-bearing strengthening mechanism explains the


direct strengthening contribution from the presence of
reinforcement particles (i.e., Vprp, where Vp and rp are the
volume fraction and the yield strength of particles,
respectively). If we assume a well-bonded spherical reinforcement particle to the matrix, then the yield strength of
particles, rp, can be represented by 1/2rm, where rm is the
yield strength of matrix. Therefore, the strength improvement by the load-bearing mechanism (DrLoad) is expressed
as [7, 38, 39, 49]:
1
DrLoad Vp rm
2

13

The influence of this load-bearing mechanism on the


MMNCs is, however, in general small enough to be ignored.
For example, when Vp = 0.010.05, the strength improvements
from the load-bearing mechanism are estimated as only 0.5
2.5 % of the original yield strength of matrix.

Strength prediction models for unimodal MMNCs


To predict the strength improvement of particle-reinforced
unimodal MMNCs, the following three superposition
approaches have been used: arithmetic summation, quadratic summation, and compounding methods. The physical basis for arithmetic and quadratic summation methods
have been put forward based on dislocation theory applied
to single crystals where quadratic summation should be
applied for obstacles to dislocation motion on the same
structural scale, and arithmetic summation should be used
for obstacles at significantly different scales [50]. Though
this scheme seemed to be confirmed experimentally by
Eberling and Ashby [51], the latest experimental evidence
from Lagerpusch et al. [52] does not support this scheme.
The compounding method is based on the assumption that
the stress applied to the material is transferred from the
relatively weak matrix material to the significantly stronger
reinforcement, enabling the material to withstand higher
stresses before the matrix yields. This is based at root on
the modified shear lag mechanism originally proposed by
Nardone and Prewo [48] for metal matrix microcomposites
(MMCs). It is characteristic of this proposed mechanism
that the yield strength of the composite depends on the
yield strength of the matrix, but not the strength of the
reinforcement. Thus the increase in strength due to load
transfer to the reinforcement can be treated mathematically
by multiplying the matrix yield stress by an improvement
factor. The compounding method, therefore, treats all
strengthening mechanisms as load-transferring mechanisms, which can be represented mathematically using a
series of improvement factors. Currently, there is no consensus as to which model or physical basis best represents

123

4196

J Mater Sci (2013) 48:41914204

reality, with the result that the prediction of strength relies


on intuition and empiricism and often strength predictions
take a form similar to that used to describe conventional
MMCs [5356]:

inappropriate material properties and expressions including


a much too large single-crystal yield strength (r0) of pure
Mg matrix in the HallPetch expression.

r rm Dr

Compounding method

14

where r, rm, and Dr are the yield strength of MMNCs,


grain size adjusted yield strength of matrix (i.e.,
p
rm r0 ky = DAlloy ), and the strength improvement by
incorporation of NPs, respectively.
Arithmetic summation method
The arithmetic summation method simply adds the contribution of individual strengthening mechanisms in a linear fashion. It assumes that different mechanisms do not
influence each other and they can independently contribute
to the final yield strength of composites.
Dr DrGR DrOrowan DrCTE DrModulus DrLoad
15
The individual contributions of strengthening mechanisms
are indicated by subscripts GR (grain refinement),
Orowan, CTE, Modulus, and Load. This method
has not been applied in many cases simply because such
linear arithmetic summation has been shown to predict higher
yield strengths compared to experimental data [39, 44].
Quadratic summation method
The quadratic summation method was originated by Clyne
and co-workers [57, 58] for the composites comproed of
micron-sized reinforcements. They assumed that the individual strengthening mechanisms interact with each other
and the sum of the squares of individual strengthening
contribution is proportional to the square of the total yield
strength improvement.
Dr

q
DrGR 2 DrOrowan 2 DrCTE 2 DrModulus 2 DrLoad 2

16
There are several prior efforts to estimate the yield
strength of particle-reinforced MMNCs based on this
quadratic summation approach [39, 44, 5759].
Depending on the processing routes and other conditions,
some of the strengthening factors are omitted in applying
Eq. (16). Several studies have reported that the quadratic
method shows better agreement between predictions and
experimental observations than other methods [39, 44, 60].
For example, Sanaty-Zadeh [60] has claimed that the
predicted strength using Eq. (16) gives the most accurate
values in reproducing the results of experimental testing.
However, the conclusion in [60] was based on several

123

The compounding approach differs from the previous two


summation methods in that the strengthening contributions
from the NPs are not added as Eq. (14) but are multiplied to
the original matrix yield strength. In other words, the
influences from each strengthening mechanism are
repeatedly counted as multiplication factors as follows:
r rm  Drf ;
Dr rm Drf  1;

17
17a

with




DrGR
DrOrowan
DrCTE
Drf 1
1
1
rm
rm
rm



DrModulus
DrLoad
 1
1
rm
rm

17b

As one of the representative models employing the


concept of compounding method given in Eqs. (17) and
(17b), Zhang and Chen [61] have demonstrated that their
approach produces the yield strength of composites very
close to the experimental findings of Mg/Y2O3 MMNCs
[62]; they proposed that the yield strength of MMNCs is
predicted by the following equation:



1
AB
r 1 Vp
rm A B
;
18
2
rm
s
12DaDTVp
A 1:25Gm b
;
18a
bdp 1  Vp
 
dp
0:13Gm b
B  1  ln
18b
2b
3
1
dp 2Vp 1
In this Zhang and Chens model [61], the effects of
Orowan, CTE mismatch, and load-bearing mechanisms are
considered, but grain refinement effects are ignored. In
other words, they used the monolithic yield strength of the
pure Mg material without NPs for rm for all materials, even
though the monolithic Mg material contained grains with
much larger size (49 lm) than the MMNCs with Al2O3
reinforcements (11 and 14 lm).
Experimental datasets and material properties for pure
Mg MMNCs
Table 3 summarizes the currently available microstructural
and mechanical property data for pure Mg MMNCs with

J Mater Sci (2013) 48:41914204

4197

Table 3 Reinforcement types, sizes, and volume fractions, processing routes, matrix grain sizes, and observed mechanical properties of pure Mg
MMNCs from [5, 6370]
Reinforcement type (size)

Al2O3 (50 nm)

Al2O3 (300 nm)

Al2O3 (A: 50 nm) ?


Al2O3 (B: 300 nm)

Al2O3 (50 nm)

Process

DMD ? hot extrusion,


(20.25:1) 150 ton

ZrO2 (2968 nm)

Al2O3 (1.0 lm)

VpA (vol%)

D (lm)

ry (MPa)

VpB (vol%)

rUTS (MPa)

Ref.
ef (%)

49 8

97 2

173 1

7.4 0.2

11 3

146 5

207 11

8.0 2.3

press T = 250 C

0.66
1.11

14 4
14 2

170 4
175 3

229 2
246 3

12.4 2.1
14.0 2.4

DMD

0.00

49 8

97 2

173 1

7.4 0.2
12.5 1.8

DMD

PM ? hot extrusion,
(20.25:1) 150 ton

DMD ? hot extrusion,


(20.25:1) 150 ton
PM

DMD ? hot extrusion,


(20.25:1) 150 ton
press T = 250 C

Al2O3 (300 nm)

Mechanical properties

0.00

press T = 250 C
Y2O3 (29 nm)

Grain Size

0.22

press T = 250 C
Y2O3 (29 nm)

Reinforcement

PM ? hot extrusion,
(20.25:1) 150 ton

0.70

62

214 4

261 5

1.10

61

200 1

256 1

8.6 1.1

2.50

41

222 2

281 5

4.5 0.5

0.00

0.00

36 4

116 11

168 10

9.0 0.3

0.50

4.50

24 8

139 26

187 28

1.9 0.2

0.75

4.25

27 9

138 13

189 15

2.4 0.6

1.00

4.00

3.0 0.3

31 7

157 20

211 21

0.00

60 10

132 7

193 2

4.2 0.1

0.22

61 18

169 4

232 4

6.5 2.0

0.66

63 16

191 2

247 2

8.8 1.6

1.11

31 13

194 5

250 3

6.9 1.0

0.00

49 8

97 2

173 1

7.4 0.2

0.22
0.66

10 1
61

218 2
312 4

277 5
318 2

12.7 1.3
6.9 1.6

0.00

60 10

132 7

193 2

4.2 0.1

0.20

25 3

156 1

211 1

15.8 0.7

0.70

13 2

151 2

202 2

12.0 1.0

0.00

49 8

97 2

173 1

7.4 0.2

0.22

82

186 2

248 4

4.7 0.2

0.66

52

221 5

271 6

4.8 0.7

1.11

21

216 4

250 6

3.0 0.2

1.10

11 4

182 3

237 1

12.1 1.4

1.10

11 3

172 1

227 2

16.8 0.4

0.00
0.17

20 3
19 3

134 7
214 4

193 1
214 4

7.5 2.5
8.0 2.8

0.70

18 3

244 1

244 1

8.6 1.2

[5]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

press T = 250 C
Y2O3 (3050 nm)

PM ? microwave
sintering ? hot extrusion,
(25:1)
T = 250 C

[70]

In processing methods, PM and DMD indicate powder metallurgy and disintegrated melt deposition, respectively. VpA and VpB are the volume
fractions of A (primary) and B (secondary, if any) reinforcement particle types, dp is the particle diameter, ry is the yield strength, rUTS is the
ultimate tensile strength, and ef is the stain to failure, respectively

corresponding references [5, 6370]. It includes the types,


sizes, and volume fractions of reinforcement NPs, processing
routes, Mg matrix grain sizes (dp), yield strengths (ry), ultimate tensile strengths (rUTS), and the stain to failure (ef). For
our analysis, Mg-based MMNCs are selected over Al-based
MMNCs to investigate the strengthening, because there is
little data available that include the necessary microstructural
and mechanical properties for Al MMNCs though they also
have been widely studied in recent years. Using the data in

Table 3, a thorough analysis has been carried out to examine


the individual strengthening mechanisms and the strength
summation methods previously addressed in Strengthening
mechanisms and Strength prediction models for unimodal
MMNCs sections. Throughout the analysis, we have particularly taken into account the strengthening improvement
from grain refinement and Orowan mechanisms, because they
are considered as primary strengthening factors to enhance the
overall strength of particle-reinforced MMNCs. In testing the

123

4198

contributions of individual strengthening mechanisms and the


adequacy of the strength prediction methods, we must first
identify the proper material properties such as HallPetch and
Orowan parameters along with relevant strengthening mechanism expressions. For the grain refinement strengthening, the
HallPetch proportionality constant ky must be determined for
Eq. (1). In the current work, we have adopted ky =
p
291 MPa lm based on Ono et al.s measurement [21] considering that the general consensus on ky for pure Mg is 250
p
300 MPa lm. For Orowan, CTE, and modulus mismatch
strengthening mechanisms, we used the following data for
pure Mg based on the physical property values found in prior
literatures [33, 44, 61]: Gm = 16.5 GPa, Burgers vector
b = 0.32 nm, Poissons ratio m = 0.3, the dislocation
strengthening coefficient b = 1.25 in Eq. (11), and the CTE
values of pure Mg aMg = 28.4 9 10-6 K-1, Al2O3 aAl2 O3 =
8.1 9 10-6 K-1, Y2O3 aY2 O3 = 9.3 9 10-6 K-1, and ZrO2
aZrO2 = 10.3 9 10-6 K-1, respectively. We assumed that
CTEs are all constant over the entire processing and service
temperature ranges. This assumption will generate an error in
the range of at most *3.2 % depending on the processing
temperatures and materials types considering the CTE value
changes of *10 % with difference temperatures. Finally,
Eq. (6) along with (6a) is used to estimate the Orowan contribution, DrOrowan, because it is the most widely used formula
and is claimed to generate reasonable strength prediction
results consistent with experimental measurements [33, 38,
39, 61].

Strengthening prediction for pure Mg MMNCs


Figure 1 shows stacked bar graphs that quantify the predicted individual contribution of grain refinement (DrHP),
Orowan (DrOrowan), and CTE mismatch (DrCTE)
strengthening mechanisms colored by pink, yellow, and
orange contrasts, respectively, to the overall yield strength
improvement of the pure Mg MMNCs listed in Table 3.
These bars were generated using the representative grain
refinement (Eq. (1)), Orowan (Eqs. (6), (6a)), and CTE
mismatch (Eqs. (11), (11a)) theories, and the material/
processing parameters given in Experimental datasets and
material properties for pure Mg MMNCs section. One
experiment with a fixed NP volume fraction is represented
by single bar graph, and a set of experiments is clustered
together as indicated by the reinforcement types and references along the x-axis. The potential strength improvements from modulus mismatch and load-bearing
mechanisms are purposefully not included in this figure.
Contribution from the modulus mismatch strengthening is
relatively difficult to estimate because the bulk strain in
Eq. (12a) cannot be easily identified, and the load-bearing
effect is likely to provide only a minor contribution. From

123

J Mater Sci (2013) 48:41914204

Fig. 1 Contribution of different strengthening mechanisms, grain


refinement (DrGR), Orowan (DrOrowan), and CTE mismatch (DrCTE),
based on the experimental data in Table 3 (Color figure online)

Fig. 1, it is clear that the grain refinement strengthening


(pink bars) and the CTE mismatch strengthening (orange
bars) are both predicted to be large in magnitude, while the
effects of Orowan strengthening (yellow bars) are relatively minor. Here, it should be mentioned that these plots
are based on the presumption of the full activation of the
three strengthening mechanisms.
Figure 2 shows a comparison chart between the predicted
yield strength increase (Drtheory) and the measured yield
strength increase (Drexperiments). Drexperiments was simply
extracted from the measurement data in Table 3, and
Drtheory were calculated by the superposition of the three
(i.e., grain refinement, Orowan, and CTE mismatch)
strengthening components, and the modulus mismatch effect
as shown in Fig. 1. A bulk strain (e) value of 0.001 was used
for the modulus mismatch mechanism assuming only a
minimal contribution of this mechanism. As seen in the
figure, we tested the predicted yield strengths from the three
strength summation methods (i.e., arithmetic, quadratic, and
compounding), and additionally, we also examined the
Zhang and Chen (ZC) summation method [61]. Their
approach was particularly examined since, in some previous
reports, this model has been demonstrated to reasonably
reproduce the experimental yield strengths of MMNCs [7,
61]. In Fig. 2, the dotted diagonal line denotes the one-toone correspondence between the theory and experimental
measurements. From the vertical positions of symbols in
Fig. 2, it is apparent that the majority of theoretical predictions, even though the minimal contribution from the
modulus mismatch strengthening was approximated, significantly overestimate the yield strengths of MMNCs (i.e.,
data points are above the dotted line), except some estimates
by quadratic summation and the ZC models. In general, the
compounding and the arithmetic summation models predict

J Mater Sci (2013) 48:41914204

Fig. 2 Comparison between experimental (Drexperiments) and theoretical (Drtheory) yield strength improvements using arithmetic
summation, quadratic summation, compounding, and ZC [61] methods with grain refinement, Orowan, CTE, and modulus mismatch
strengthening mechanisms (Color figure online)

the highest and the next higher Drtheory, and the quadratic
summation and ZC methods estimate relatively lower
Drtheory; Dr (compounding) [ Dr (arithmetic) [ Dr (quadratic) and Dr (ZC). This trend can be easily anticipated
from the nature of mathematical expression for these summation methods and from the deficiency of the grain
refinement component in ZC approach. The significance of
the missing grain refinement effect in the ZC model is
clearly seen in the chart as it consistently produces negative
deviation from the dotted line when Drexperiments increases
(the orange triangle symbols in the Drexperiments range of
75125 MPa). Higher Drexperiments generally related to the
higher grain refinement effect as will be subsequently
demonstrated; thus, it can be inferred that ZC approach
tends to show underestimation in the yield strengths as the
grain refinement effect increases. From Fig. 2, it is plausible
that the full activation of CTE and modulus mismatch
strengthening mechanisms is unlikely to occur due to the
overestimation.
Since most predicted yield strengths based on grain
refinement, Orowan, CTE, and modulus mismatch mechanisms exhibit considerable positive deviation from the
experimental observations, and it is not clear that CTE and
modulus mismatch mechanisms are truly active, we tested
the Drtheory based only on HallPetch and Orowan mechanisms excluding the effects of CTE and modulus mismatch strengthening. In Fig. 3, the predicted Drtheory
values are plotted with reference to the experiments using
arithmetic, quadratic, and compounding superposition
methods. The dotted diagonal line again represents exact
agreement between the theory and experiment. As apparently seen from Fig. 3, the theory now predicts the correct

4199

trend, but underestimates the yield strength of Mg MMNCs


for the majority of experiments listed in Table 3. This trend
showing negative deviations likely results from underestimating the contribution from grain refinement and/or the
Orowan mechanism. In determining the grain refinement
p
contribution, we used ky = 291 MPa lm treating
MMNCs as pure Mg materials; however, it may be that the
presence of NPs results in a higher ky in Mg MMNCs
causing a greater contribution of grain refinement that
would raise the predicted theoretical data points in Figs. 2
and 3. Since all materials analyzed have been subjected to
mechanical post-processing treatments, increased strength
due to work-hardening is another factor that may explain
the discrepancy in the predicted strength when only grain
refinement and Orowan strengthening are considered. Postprocessing using extrusion or rolling involves large
amounts of deformation and generates dislocations that
likely lead to improved yield strength in MMNCs. Such
work-hardening effect has apparently never been considered when strengthening mechanisms for MMNCs are
discussed, likely because it is the result of post-processing
and not directly attributable to the incorporation of NPs.
In Fig. 3, there are two distinct datasets as enclosed by
ellipses A and B to show the opposite extreme deviations in
the yield strength predictions. It is thought that Drtheory is
overestimated for the dataset A (ZrO2, Vp = 1.11 vol%,
DMMNC = 2 lm [68]) because of the significantly smaller
matrix grain size; as the grain size decreases to near (or
below) 1 lm range, the effect of grain refinement is
anticipated to abruptly increase by the HallPetch expression (Eq. (1)), but it seems that Eq. (1) may overestimate
the contribution of grain refinement for the dataset A as

Fig. 3 Comparison between experimental (Drexperiments) and theoretical


(Drtheory) yield strength improvements using arithmetic summation,
quadratic summation, and compounding methods with grain refinement
and Orowan strengthening mechanisms (Color figure online)

123

4200

J Mater Sci (2013) 48:41914204

shown in the far right bar in Fig. 1. Although there is a


report to address that the HallPetch relation is valid down
to *75 nm in some metallic composites [71], it is currently unclear if the HallPetch relation in MMNCs is still
active in the experimental dataset A. For the dataset
B (Y2O3, Vp = 0.66 vol%, DMMNC = 6 lm [66]) in Fig. 3,
Drtheory is much lower than the experimental measurement.
This can be presumably explained by the extensive generation of dislocations from the work-hardening. It is currently unclear why the work-hardening effect would be so
prevalent in only this sample, but the significantly reduced
ductility (i.e., strain to failure) from 12.7 to 6.9 % for this
system, as shown in Table 3, supports the generation of
massive dislocations by work-hardening.
To examine the effects of only grain refinement, we
have plotted the Drexperiments in Fig. 4 using the grain size
and yield strength data in Table 3 as a function of the grain


1
p1 , where DMMNC and D0
size reduction, i.e., p

D
D
MMNC

denote the matrix grain size of MMNCs and pure Mg


materials without NPs, respectively. In the figure, the solid
line is drawn to represent only the contribution from grain
refinement strengthening to Drtheory, meaning that the
p
slope of the solid line is 291 MPa lm. Therefore, if the
Drexperiments data fall on the solid line, it can be understood
that grain refinement is the only active mechanism. If the
experimental data lie above the solid line, it is then considered that mechanisms other than grain refinement
strengthening may influence the overall yield strength
improvement for the corresponding MMNCs. From Fig. 4,
it is seen that, in general, Drexperiments increases with the


1
p1 , which

degree of grain size refinement, p
D
D
MMNC

confirms the positive contribution from the grain refinement strengthening mechanism to the overall yield
strength. This further indicates that as the grain size of
MMNCs decreases, the relative contribution from grain
refinement to the total Dr monotonically increases. In
Fig. 4, about two thirds of Drexperiments data are reasonably
explained by the sum of grain refinement effect and a
constant offset caused by other strengthening mechanism(s). In other words, when the solid HallPetch line is
shifted up by *35 MPa as indicated by the dotted line, it
reasonably predicts two thirds of Drexperiments data, which
signifies that the yield strength improvement of current
MMNC synthesis is highly dependent on the grain size of
the synthesized matrix. Further, this indicates that the
specimens showing much improved yield strength generally contain smaller matrix grain sizes. Hence, much of the
improvement is not directly attributable to the NPs themselves, but rather results from the reduced grain size that
comes from the incorporation of the NPs. One desirable
outlier from the dotted trend line is given by the symbol

123

Fig. 4 Contribution from the grain refinement strengthening mechanism to the overall yield strength improvement (Color figure online)

C (the purple square near the top), which corresponds to the


datasets of ellipse B in Fig. 3. In this experiment, as
described in Fig. 3, Drexperiments is much higher than the
dotted HallPetch trend line, meaning that a significant
portion of Drexperiments seems to be accounted for by
mechanisms other than grain refinement. However, the
significantly reduced strain to failure indicates that the
improvement seems not to be coming from the desirable
Orowan mechanism but rather from the massive generation
of dislocations resulting from mechanical post-processing.
If the grain refinement mechanism predominantly determines the yield strength of MMNCs especially for the finegrain materials, it is very difficult to clarify or determine
which summation method will most truly predict the
experimental measurement values because all of the three
superposition models are identical in the case of only a
single strengthening mechanism. From the fact that the
yield strength improvements can be approximated by the
simple sum of grain refinement effect and a constant (i.e.,
*35 MPa) irrespective of the NP volume fractions or
particle size, it is doubtful that the Orowan strengthening
mechanism is in fact active in these materials. Given that
the *35 MPa seems to be roughly independent of NP
composition, size, and volume fraction, it is similarly
doubtful that CTE or modulus mismatch mechanisms are
active and it would seem that increased dislocation density
from post-process work-hardening is the best and only
remaining candidate to account for this relatively modest
strength increase.
In Fig. 5, the predicted Drtheory are plotted as a function
of NP volume fraction, Vp, with two different particle
diameters, dp, of (a) 30 and (b) 100 nm, respectively.
Drtheory were predicted for the Orowan mechanism alone,
as well as the three basic superposition methods using both

J Mater Sci (2013) 48:41914204

Fig. 5 Expected yield strength improvement (Drtheory) using arithmetic summation, quadratic summation, and compounding methods
as a function of particle volume fractions (Vp) with particle diameter
(dp) of a 30 nm and b 100 nm (Color figure online)

Orowan and CTE mismatch mechanisms. Contribution


from the CTE mismatch strengthening was considered to
test the activation of the GND generation-based strengthening mechanism. The effect of grain refinement strengthening is not included for now because it is not directly
related to the NP addition as mentioned before. To determine the contribution from CTE mismatch strengthening,
we used the temperature difference DT of 280 K, and the
CTE difference Da of 19 9 10-6 K-1 for Eq. (11a). From
Fig. 5a, b, it is clear that the Drtheory is strongly influenced
by the particle diameter, dp, as well as the volume fraction,
Vp. As shown by the black curves (i.e., Orowan), Orowan
strengthening mechanism predicts the yield strength
improvement of 50 and 20 MPa with 2 vol% of NP additions when the dps are 30 and 100 nm, respectively. It is,

4201

therefore, critical to add fine NPs into the matrix in utilizing the Orowan-type mechanism. By the activation of
CTE mismatch mechanism, the improvement is expected
to be greatly increased; with the same condition (i.e.,
Vp = 2 vol%, and dp = 30 and 100 nm), Drtheory predicts
130 and 70 MPa using the quadratic summation. These
predictions are already close to the observed yield strength
improvements for the majority of samples in Table 3,
which disproves the activation of GND generation-type
strengthening mechanisms.
Despite the grain refinement strength enhancement of
MMNCs being caused indirectly by the addition of NPs, it is
worth theoretically examining a possible effect of NP
additions on ky. As stated earlier, a possible change in ky due
directly to NPs may account for the disagreement between
predicted and experimental yield stress improvements. Figure 6 shows the predicted yield strength increase by the
grain refinement mechanism, DrGR, with fixed values of the
p
HallPetch slope, ky, at 150, 291, and 350 MPa lm. The
p
p
minimum 150 MPa lm and maximum 350 MPa lm
values for the ky were selected from Table 1 treating them as
lower and upper bounds for the DrGR prediction in Mg
p
MMNCs. The red curve is based on ky 291 MPa lm;
which was used for the analyses in Figs. 24 based on Ono
et al.s report [21]. DrGR in Fig. 6 were computed assuming
that the elemental metal matrix without NPs (i.e., D0 in Eq.
(1)) has the original grain size of 50 lm. When the matrix
grain size of MMNCs (DMMNC) is decreased from 50 to
5 lm, DrGR is estimated to be 45115 MPa, which is
already greater than the strength improvement factors other
than the grain refinement effect based on the properties of
pure Mg observed in Fig. 4 (i.e., the difference between the
solid and dotted lines, about 35 MPa). If the grain size is
reduced below 5 lm, DrGR is estimated to exhibit an abrupt
increase due to the fine size effect, but it is not certain that
the HallPetch expression with constant ky in Eq. (1) can
still be applied to the MMNCs with such fine grain sizes
(i.e., under *1 lm) because it may overestimate the effect
of grain refinement as illustrated in the data points surrounded in the ellipse A in Fig. 3. Here, we want to mention
that the grain refinement effect in Mg MMNCs would be
much greater than that in Al-based MMNCs, because the

p
HallPetch slope, ky, for Al materials  68 MPa lm is
in general much smaller compared to the slope for Mg
because of the multiple slip systems for Al [72]. Therefore,
the contribution from the grain refinement in Al MMNCs is
expected to be smaller than that in Mg MMNCs, but it still
would occupy a considerable portion in the yield strength
increase.
Finally, 3D prediction maps for the yield strength
increase in particle-reinforced Mg MMNCs using (a) grain
refinement and Orowan strengthening, and (b) grain

123

4202

J Mater Sci (2013) 48:41914204

Fig. 6 Expected grain refinement contribution (DrGR) as a function


of matrix grain diameter of MMNC, DMMNC. Initial diameter, D0, was
assumed as 50 lm (Color figure online)

refinement strengthening mechanisms are presented in


Fig. 7a, b, respectively. The map in Fig. 7a was constructed using the quadratic summation method, and the
prediction in Fig. 7b is based on the sum of grain refinement effect and 35 MPa, i.e., the contribution from other
sources as indicated in Fig. 4. The color legend represents
the predicted Dr in the unit of MPa by the NP volume
fraction, Vp, and the matrix grain size, D. We used same
processing/material conditions (DS = 280 K, Da = 19.0
9 10-6 K-1 in Eqs. (11) and (11a)) and assumed the
particle diameter, dp, as 30 nm for Fig. 7a. The map based
only on grain refinement and Orowan (Fig. 7a) will act as
the prediction for desired strength improvement that would
be attained without loss of ductility, and the other map
including the grain refinement effect and a constant sum
(Fig. 7b) can be used to reasonably approximate the
observed yield strength improvement of Mg MMNCs
resulting from the grain refinement effect and probably
from work-hardening. From Fig. 7a, in theory, the yield
strength improvement will reach *280 MPa with submicron grain size (*1 lm) and 7.5 vol% NP additions.
Given that the baseline of yield strength of monolithic
polycrystalline Mg materials is around 100 MPa, and the
maximum strength improvement is found as *130 MPa in
Figs. 2 and 3 with the exception of one outlier, Dr of
*280 MPa would be regarded as a significant strength
improvement in the MMNC community. It must be noted
that the predictions presented in Fig. 7a are based on the
assumption that the NPs are not agglomerated and evenly
distributed during the synthesis procedure, i.e., the Orowan
strengthening mechanism is fully active. It is, however, still
very challenging to obtain such ideal microstructures even
using the best currently available synthesis technologies for
solid- and liquid-phase processing routes. Therefore, it is

123

Fig. 7 3D prediction maps for the yield strength increase in particlereinforced Mg MMNCs using a grain refinement and Orowan
strengthening, and b grain refinement strengthening mechanisms
(Color figure online)

required that more advanced synthesis technologies must be


developed to prevent agglomeration of NPs with high volume contents and to effectively reduce grain size of matrix
without generation of massive deformation dislocations.
Summary
In this work, we have reviewed the theoretical mechanisms and
summation methods to predict the yield strength of particlereinforced MMNCs, and we have performed analyses using Mg
MMNCs to study these mechanisms and summation methods.
The following are the main points addressed in the current
document.

It was found that all of the arithmetic, quadratic, and


compounding methods using grain refinement, Orowan,

J Mater Sci (2013) 48:41914204

CTE, and modulus mismatch strengthening mechanisms in general overestimate the yield strength, and
ZC method underestimates the strength especially when
the effect of matrix grain size becomes dominant. On
the other hand, it is demonstrated that the three
conventional summation methods using only grain
refinement and Orowan mechanisms produce yield
strength values lower than the experimental ones.
Since the activation of CTE and modulus mismatch
mechanisms are unlikely to occur, and the yield strength
prediction based only on grain refinement and Orowan
mechanism generally showed underestimation, it is
inferred that work-hardening from post-processing of
MMNCs may contribute to the overall strength improvement. Such work-hardening effect is supported by the
reduction in failure strains of MMNCs.
Our analysis shows that, in most samples, the measured
yield strength increases linearly with the inverse of the
square root of matrix grain size, which confirms that the
grain refinement effect predominantly determines the
strength of MMNCs, in which case all summation
methods become identical to the HallPetch relation
with about a 35 MPa adjustment to r0. Therefore, given
the current lack of a sufficient strengthening contribution by other mechanisms it is currently not feasible to
test which summation method produces the most
consistent results with experimental observations.
We present the prediction map for the yield strength
improvement of Mg MMNCs. In theory, the yield
strength of Mg MMNCs would reach 380 MPa with a
submicron matrix grain size and 7.5 vol% particle
addition. In attaining MMNC products with such theoretical strength, it is, however, still considered as a huge
roadblock for MMNC synthesis technologies to minimize
the particle agglomeration while maintaining uniform
distributions, and also to prevent excessive generation of
dislocations in effectively reducing matrix grain size.

Acknowledgements This work is primarily supported by the


Research Growth Initiative (RGI) Award from University of Wisconsin-Milwaukee (UWM). Partial support from the U.S. Army
Research Laboratory (US ARL) under Cooperative Agreement No.
W911NF-08-2-0014 is also acknowledged. The views, opinions, and
conclusions made in this document are those of the authors and
should not be interpreted as representing the official policies, either
expressed or implied, of Army Research Laboratory or the U.S.
Government. The U.S. Government is authorized to reproduce and
distribute reprints for Government purposes notwithstanding any
copyright notation herein.

References
1. Ye J, Han BQ, Lee Z, Ahn B, Nutt SR, Schoenung JM (2005) Scr
Mater 53:481. doi:10.1016/j.scriptamat.2005.05.004

4203
2. Tang F, Hagiwara M, Schoenung JM (2005) Scr Mater 53:619.
doi:10.1016/j.scriptamat.2005.05.034
3. Li Y, Zhao YH, Ortalan V, Liu W, Zhang ZH, Vogt RG,
Browning ND, Lavernia EJ, Schoenung JM (2009) Mater Sci Eng
A 527:305. doi:10.1016/j.msea.2009.07.067
4. Li Y, Lin YJ, Xiong YH, Schoenung JM, Lavernia EJ (2011) Scr
Mater 64:133. doi:10.1016/j.scriptamat.2010.09.027
5. Hassan SF, Gupta M (2004) Mater Sci Technol 20:1383. doi:
10.1179/026708304X3980
6. Tun KS, Gupta M (2008) J Mater Sci 43:4503. doi:
10.1007/s10853-008-2649-3
7. Hassan SF, Tan MJ, Gupta M (2008) Mater Sci Eng A 486:56.
doi:10.1016/j.msea.2007.08.045
8. Paramsothy M, Hassan SF, Srikanth N, Gupta M (2009) Mater
Sci Eng A 527:162. doi:10.1016/j.msea.2009.07.054
9. Yang Y, Lan J, Li X (2004) Mater Sci Eng A 380(2004):378. doi:
10.1016/j.msea.2004.03.073
10. Cao G, Kobliska J, Konishi H, Li X (2008) Metall Mater Trans A
39A:880. doi:10.1007/s11661-007-9453-6
11. Dutkiewicz J, Litynska L, Maziarz W, Haberko K, Pyda W,
Kanciruk A (2009) Cryst Res Technol 44:1163. doi:10.1002/crat.
200900455
12. Ahn JH, Kim YJ, Chung H (2008) Rev Adv Mater Sci 18:329
13. Mohammad Sharifi E, Karimzadeh F, Enayati MH (2011) Mater
Des 32:3263. doi:10.1016/j.matdes.2011.02.033
14. Mazahery A, Abdizadeh H, Baharvandi HR (2009) Mater Sci Eng
A 518:61. doi:10.1016/j.msea.2009.04.014
15. Yao B, Hofmeister C, Patterson T, Sohn YH, Van den Bergh M,
Delahanty T, Cho K (2010) Compos A 41:933. doi:
10.1016/j.compositesa.2010.02.013
16. Schultz BF, Ferguson JB, Rohatgi PK (2011) Mater Sci Eng A
530:87. doi:10.1016/j.msea.2011.09.042
17. Ferguson JB, Sheykh-Jaberi F, Kim CS, Rohatgi PK, Cho K
(2012) Mater Sci Eng 558:193. doi:10.1016/j.msea.2012.07.111
18. Mallick A, Vedantam S, Lu L (2009) Mater Sci Eng A 515:14.
doi:10.1016/j.msea.2009.03.002
19. Wang YN, Huang JC (2007) Mater Trans 48:184. doi:
10.2320/matertrans.48.184
20. Mann G, Griffiths JR, Caceres CH (2004) J Alloys Compd
378:188. doi:10.1016/j.jallcom.2003.12.052
21. Ono N, Nowak R, Miura S (2003) Mater Lett 58:39. doi:
10.1016/S0167-577X(03)00410-5
22. Wang HY, Xue ES, Xiao W, Liu Z, Li JB, Jiang QC (2011) Mater
Sci Eng A 528:8790. doi:10.1016/j.msea.2011.07.052
23. Andersson P, Caceres CH, Koike J (2003) Mater Sci Forum
419422:123. doi:10.4028/www.scientific.net/MSF.419-422.123
24. Yuan W, Panigrahi SK, Su JQ, Mishra RS (2011) Scr Mater
65:994. doi:10.1016/j.scriptamat.2011.08.028
25. Kim HK (2009) Mater Sci Eng 515:66. doi:10.1016/j.
msea.2009.02.039
26. Afrin N, Chen DL, Cao X, Jahazi M (2008) Mater Sci Eng A
472:179. doi:10.1016/j.msea.2007.03.018
27. Han BQ, Dunand DC (2000) Mater Sci Eng A 227:297. doi:
10.1016/S0921-5093(99)00074-X
28. Bohlen J, Dobron P, Meza Garcia E, Chmelik F, Lukac P, Letzig
D, Kainer KU (2005) Adv Eng Mater 8:422. doi:10.1016/
j.msea.2006.02.469
29. Elsayed A, Kondoh K, Imai H, Umeda J (2010) Mater Des
31:2444. doi:10.1016/j.matdes.2009.11.054
30. Hagihara K, Kinoshita A, Sugino Y, Yamasaki M, Kawamura Y,
Yasuda HY, Umakoshi Y (2010) Acta Mater 58:6282. doi:
10.1016/j.actamat.2010.07.050
31. Zener C, quoted by Smith CS (1948) Trans AIME 175:15
32. Szaraz Z, Trojanova Z, Cabbibo M, Evangelista E (2007) Mater
Sci Eng A 462:225. doi:10.1016/j.msea.2006.01.182

123

4204
33. Habibnejad-Korayem M, Mahmudi R, Poole WJ (2009) Mater
Sci Eng A 519:198. doi:10.1016/j.msea.2009.05.001
34. Zhang Z, Yu H, Wang S, Wang H, Min G (2010) J Mater Sci
Technol 26:151. doi:10.1016/S1005-0302(10)60025-4
35. Nguyen QB, Gupta M (2008) Compos Sci Technol 68:2185. doi:
10.1016/j.compscitech.2008.04.020
36. Miller WS, Humphreys FJ (1991) Scr Metall 25:33. doi:
10.1016/0956-716X(91)90349-6
37. Ashby MF (1968) The theory of the critical shear stress and work
hardening of dispersion-hardened crystals. In: Proceeding of
second Bolton landing conference on oxide dispersion strengthening. Gordon and Breach, Science Publishers, Inc., New York,
p 143
38. Sun Y, Choi H, Konishi H, Pikhovich V, Hathaway R, Chen L, Li
X (2012) Mater Sci Eng A 546:284. doi:10.1016/j.msea.2012.
03.070
39. Goh CS, Wei J, Lee LC, Gupta M (2007) Acta Mater 55:5115.
doi:10.1016/j.actamat.2007.05.032032
40. Robson JD, Stanford N, Barnett MR (2010) Scr Mater 63:23. doi:
10.1016/j.scriptamat.2010.06.026
41. Rosalie JM, Somekawa H, Singh A, Mukai T (2012) Mater Sci
Eng A 539:230. doi:10.1016/j.msea.2012.01.087
42. Zeng X, Zou H, Zhai C, Ding W (2006) Mater Sci Eng A 424:40.
doi:10.1016/j.msea.2006.02.021
43. Ferguson JB, Lopez H, Kongshaug D, Schultz B, Rohatgi P
(2012) Metall Mater Trans A 43:2110. doi:10.1007/s11661011-1029-9
44. Dai LH, Ling Z, Bai YL (2001) Compos Sci Technol 61:1057.
doi:10.1016/S0266-3538(00)00235-9
45. Vogt R, Zhang Z, Li Y, Bonds M, Browning ND, Lavernia EJ,
Schoenung JM (2009) Scr Mater 61:1052. doi:10.1016/j.
scriptamat.2009.08.025
46. Redsten AM, Klier EM, Brown AM, Dunand DC (1995) Mater
Sci Eng A 201:88. doi:10.1016/0921-5093(94)09741-0
47. Nardone VC (1987) Scr Metall 21:1313. doi:10.1016/0036-9748
(87)90105-0
48. Nardone VC, Prewo KM (1986) Scr Metall 20:43. doi:
10.1016/0036-9748(86)90210-3
49. Ramakrishnan N (1996) Acta Metall 44:69. doi:10.1016/
1359-6454(95)00150-9
50. Kocks UF, Argon AS, Ashby MF (1975) Prog Mater Sci 19:224
51. Ebeling R, Ashby MF (1966) Phil Mag 13:805
52. Lagerpusch U, Mohles V, Baither D, Anczykowski B, Nembach
E (2000) Acta Mater 48:3647. doi:10.1016/S1359-6454(00)
00172-5

123

J Mater Sci (2013) 48:41914204


53. Chawla N, Andres C, Jones JW, Allison JE (1998) Metall Mater
Trans A29:2843. doi:10.1007/s11661-998-0325-5
54. Chawla N, Habel U, Shen YL et al (2000) Metall Mater Trans
A31:531. doi:10.1007/s11661-000-0288-7
55. Chawla N, Shen YL (2001) Adv Eng Mater 3:357. doi:
10.1002/1527-2648(200106)3:6\357:AIDADEM357[3.3.CO;2-9
56. Arsenault RJ (1984) Mater Sci Eng 64:171. doi:10.1016/
0025-5416(84)90101-0
57. Clyne TW, Withers PJ (1995) An Introduction to Metal Matrix
Composites. Cambridge University Press, Cambridge
58. Hull D, Clyne TW (1996) An introduction to composite materials. Cambridge University Press, Cambridge
59. Lilholt H (1985) Deformation of multi-phase and particle containing materials. In: Proceedings of the 4th rise international symposium on metallurgy and materials science, Roskilde, Denmark
60. Sanaty-Zadeh A (2012) Mater Sci Eng A 531:112. doi:10.1016/
j.msea.2011.10.043
61. Zhang Z, Chen DL (2006) Scr Mater 54:1321. doi:10.1016/j.
scriptamat.2005.12.017
62. Hassan SF, Gupta M (2006) Compos Struct 72:19. doi:10.1016/
j.compstruct.2004.10.008
63. Hassan SF, Gupta M (2008) J Alloy Compd 457:244. doi:10.1016/
j.jallcom.2007.03.058
64. Wong WLE, Karthik S, Gupta M (2005) J Mater Sci 40:3395.
doi:10.1007/s10853-005-0419-z
65. Hassan SF, Gupta M (2005) Mater Sci Eng A 392:163. doi:10.1007/
s11661-005-0344-4
66. Hassan SF, Gupta M (2007) J Alloy Compd 429:176. doi:
10.1016/j.jallcom.2006.04.033
67. Hassan SF (2011) Mater Sci Eng A 528:5484. doi:10.1016/
j.msea.2011.03.063
68. Hassan SF (2006) Creation of new magnesium-based material
using different types of reinforcements. Dissertation, National
University of Singapore
69. Hassan SF, Gupta M (2006) Mater Sci Eng A 425:22. doi:10.1016/
j.msea.2006.03.029
70. Tun KS, Gupta M (2007) Compos Sci Technol 67:2657. doi:
10.1016/j.compscitech.2007.03.006
71. Misra A, Hirth JP, Hoagland RG (2005) Acta Mater 53:4817. doi:
10.1016/j.actamat.2005.06.025
72. Wong WLE, Gupta M (2007) Compos Sci Technol 67:1541. doi:
10.1016/j.compscitech.2006.07.015

Вам также может понравиться