Вы находитесь на странице: 1из 9

Two-dimensional modeling to compute plastic zone in front of compact

tension sample of a multiphase material


Yunan Prawoto
*
, Roslinda Idris, Nazri Kamsah, Nasir Tamin
Dept. of Applied Mechanics, Faculty of Mechanical Engineering, University Technology Malaysia, 81310 UTM, Skudai, Johor, Malaysia
a r t i c l e i n f o
Article history:
Received 6 September 2009
Accepted 20 September 2009
Available online 13 October 2009
Keywords:
Fatigue crack propagation (FCP)
Multiphase
Plastic zone
a b s t r a c t
More and more components subjected to dynamic high stress loading are heat treated to get multiphase
microstructure as their nal products, instead of conventional tempered martensite. It is gaining popu-
larity due its manufacturing benets, including the elimination of the tempering process, and its better
fatigue performance. Multiphase of ferrite and martensite benets its fatigue performance by its high
toughness yet maintains the high tensile properties due to controlled microstructure. The plastic zone
size in front of the crack is the key to fatigue performance. In this research, a multiphase (polygonal fer-
rite and martensite) microstructure was developed. Manipulation method to use commercial software to
calculate the plastic zone size in front of a crack is presented.
Two-dimensional modeling of squares having variation of ferrite fraction was used as local models. The
shape of the ferrites was derived from actual microstructure of multiphase material. These local models
were put in front of the CT specimen in a global model. The analysis result shows the variation of the plas-
tic zone sizes as the ferrite fraction varies and saturates at about 65% ferrite fraction.
2009 Elsevier B.V. All rights reserved.
1. Introduction
It is generally understood that among other forms of micro-
structure available in industrial steel, martensite has the best ten-
sile properties [1]. However, at the same time it also has poor
toughness due to its brittleness. It has been known and practiced
extensively that tempering increases the toughness while sacric-
ing the materials tensile properties. Tempered martensite, also
formerly known as sorbite, is the microstructure resulting from
quenching followed by tempering.
From the fracture mechanics point of view, the as-quenched
microstructure has a sharp needle shape, or acicular carbide. This
shape somehow implies high brittleness. When quenched material
is tempered, the acicular carbide changes to less acicular which
also implies low brittleness. Typically for industrial steel, temper-
ing at around 550 C for approximately 6090 min is the most
commonly used method to achieve usable toughness. However,
in addition to gaining toughness, the corresponding material also
loses considerable tensile properties. This method produces homo-
geneous tempered martensite, which is good for preventing the
crack initiation, but not the best for preventing fatigue crack prop-
agation (FCP) due to its resulting low toughness properties, which
is inherent to martensite.
There is no doubt that controlling and determining the FCP rate
is the most essential part of the fracture mechanics design ap-
proach. The best way to acquire it is by knowing the metallurgical
mechanisms as well as the continuum mechanics of fatigue crack
propagation. Nonetheless, it is difcult to combine these ap-
proaches since in the area where the continuum mechanism ap-
proach can be used, the metallurgical inuence is small and vise
versa, see Fig. 1 [2]. Furthermore, research on microstructure often
lacks the continuum mechanics approach; the same way research
on continuum mechanics lacks microstructural observation.
In this research, the comprehension of fracture mechanics is
used for the development of microstructure. Ultimately, the goal
is to achieve a microstructure that resists fatiguing more than
the microstructure conventionally available.
1.1. Metallurgical mechanism of fatigue cracking
The early development of fracture modeling divides cracking
into two stages. The rst stage, as shown in Fig. 2, is characterized
by propagation of the crack on a plane oriented approximately 45
to the stress axis and by crystallographic fracture facets orientated
at grain boundaries [3].
Subsequently, the propagation enters Stage B where the plane
of crack propagation is approximately 90 to the stress axis and
the fracture surface is covered by striations running parallel to
the crack propagation front. Striations on fracture surfaces look like
tidemarks. Failure resulting from high strain fatigue takes place
0927-0256/$ - see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.commatsci.2009.09.014
* Corresponding author. Tel.: +60 167 279048; fax: +60 755 66159.
E-mail address: yunan.prawoto@gmail.com (Y. Prawoto).
Computational Materials Science 47 (2009) 482490
Contents lists available at ScienceDirect
Computational Materials Science
j our nal homepage: www. el sevi er . com/ l ocat e/ commat sci
predominantly by Stage B crack propagation; whereas at very low
strains, the majority of the fatigue life is spent in Stage A. Even at
low strains, however, most of the fracture surface of a failed ductile
specimen is covered with striations; this indicates that Stage B also
plays an important role in low strain fatigue because this process
actually fractures most of the cross-section of the specimen. Since
crack propagation rates in Stage B growth can reach values of mi-
crons per cycle, the phenomena associated with the growth mech-
anism is fairly large and relatively easy to observe.
The initiation of fatigue crack propagation is known to occur
along slip bands, in grain boundaries, in second-phase particles,
and in inclusion or second-phase interfaces with the matrix
phases. The sequence of events for the initiation of fatigue cracks
is as follows; during cyclic plastic deformation, dislocations either
emerge at the surface of the metal or pile up against obstacles. If
dislocations continuously emerge at the surface rather than pile-
up against obstacles, then slip bands that eventually become cracks
appear in the central portions of the grains where the ow stress is
lower. This was shown by Kung and Fine [4].
It is known that many types of obstacles can cause dislocation
pile-ups during cycling, including grain boundaries, inclusions,
and oxide lms. Dislocation pile-ups result in an increase in elastic
energy. When the strain energy density exceeds the surface energy,
a condition which favors the initiation of micro cracks begins. This
can lead to a slip-band crack in the matrix, de-cohesion along a
grain boundary, or cracking of a second-phase. These mechanisms
govern the initiation of cracks, which will continue to coalesce, and
later develop into a crack.
A fatigue crack, once initiated, can also grow by a mechanism of
reversed slip [5]. Several stages of fatigue crack propagation are
shown in Fig. 3. A sharp crack in a tension eld causes a large stress
concentration at its tip where slip can occur easily. The material
above the crack (Steps 1 and 2 in the gure) may slip along favor-
able slip planes in the direction of maximum shear stress. Due to
this slip, the crack opens, but it also extends in length. Slip can
now occur on another plane, Step 3. Work hardening and increas-
ing stress will nally activate other parallel slip plane, which leads
to a blunt crack tip, Step 4. During the rising load stage of the cycle,
the crack is propagated by the amount of Da. Plastic deformation
has occurred in a small region embedded in elastic surroundings.
During load release, the elastic surroundings will contract and
the plastically deformed region, which has become large, does
not t any more in its surroundings. In order to make it t, the elas-
tic material will exert compressive stresses on the plastic region
during the decreasing load stage of the cycle. These compressive
stresses will close and re-sharpen the crack tip, Step 5. The cyclic
opening and closing of the crack (Steps 17) will develop the stri-
ation patterns.
The question of whether one cycle corresponds to one striation
spacing, however, remains unclear. Some researchers claim to have
Fig. 1. Schematic plot of the area grouping and their mechanisms [2].
Fig. 2. Schematic representation of the two stages of crack propagation [3].
Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490 483
proven that each fracture surface striation is associated with one
stress cycle [6]. Others say, that based on the damage accumula-
tion mechanism, the striation spacing does not necessarily corre-
spond to the crack growth after one cycle [7]. Tanaka has shown
experimentally that for microstructurally short cracks, the resis-
tance to crack growth is greatly inuenced by the material micro-
structure, such as the grain orientation and grain boundaries, see
Fig. 4 [8].
1.2. The effect of the plastic zone
The effect of microstructure on FCP has been actively re-
searched recently (e.g., [9]). Quantitatively it is easier to manage
with the plastic zone concept. It has been reported that the PZ
can be visualized experimentally with various experiments (e.g.,
[10]). A detailed explanation of the plastic zone is discussed in
the next section. Here, it is only accentuated that for the same con-
dition, larger plastic zones have tendency to slow FCP more. Pavlou
[11] is one among several researchers that have proven experimen-
tally that larger plastic zones result in more efcient retardation of
the FCP. Fig. 5 illustrates that larger plastic zone due to some over-
load retards the FCP. This phenomenon is widely accepted among
practitioner in aircraft industries.
2. Theory
2.1. Stress intensity factor
The concepts of fracture mechanics are concerned with the ba-
sic ideas for developing methods of predicting the load-carrying
capabilities of structures containing cracks. The fracture mechanics
approach is based on a mathematical description of the character-
istic stress eld surrounding any crack in a loaded body.
To explore the characteristics of the stress eld surrounding a
crack in a loaded body, one can start from the Westergaard func-
tion [12]. The Westergaard function is a complex solution to the
Airy stress functions. To do this, consider a coordinate system X,
Y, Z in a stressed solid. At each point (x, y, z) one can dene the
stresses r
x
; r
y
; r
z
; s
xy
; s
xz
; s
yz
. Neglecting the body forces, for two-
dimensional problems, equilibrium equations are:
@r
x
@x

@s
xy
@y
0; and
@r
y
@y

@s
xy
@x
0: 1
If the displacements in x and y are u and v, respectively, the expres-
sions for the strains are:
e
x

@u
@x
; e
y

@v
@y
; c
xy

@u
@y

@v
@x
; 2
and the stressstrain relations:
Ee
x
r
x
vr
y
; Ee
x
r
y
vr
x
;
Ec
x
21 v
s
xy
; 3
where E is Youngs Modulus and v is Poissons Ratio. The equilib-
rium equation in Eq. (1) is automatically satised if
r
x

@
2
w
@y
2
; r
y

@
2
w
@x
2
; and s
yx

@
2
w
@x@y
: 4
The function w is called the Airy stress function. Substitution of Eqs.
(2) and (4) into Eq. (3), and differentiating twice leads to the com-
patibility equation:
@
4
w
@x
4
2
@
4
w
@x
2
@y
2

@
4
w
@y
4
0; 5
or:
r
2
r
2
w 0: 6
One can dene a complex function by
Zz ReZ i ImZ with z x iy: 7
For Z to be an analytic function, the derivative dZ/dz must be de-
ned unambiguously. This leads to the CauchyRiemann
conditions:
@ReZ
@x

@ImZ
@y
Re
dZ
dz
and
@ImZ
@x

@ReZ
@y
Im
dZ
dz
: 8
In the case of Mode I cracks it is convenient to use a function pro-
posed by Westergaard. The Westergaard function is:
Fig. 3. Possible model for fatigue crack propagation [5].
484 Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490
w ReZ yImZ; 9
where
dZ
dz
Z;
dZ
dz
Z and
dZ
dz
Z
0
. With the CauchyRiemann equa-
tions it follows that
r
2
ReZ r
2
ImZ 0; 10
which means that Eq. (9) automatically satises the compatibility
Eq. (6).
By using Eq. (4), the stresses can be determined as:
r
x
ReZ yImZ
0
; r
y
ReZ yImZ
0
;
and s
xy
yReZ
0
: 11
Any analytic function Z(z) will result in stresses dened by Eq. (11).
For a crack problem of Fig. 6, representing an innite plate un-
der biaxial stress, the stress function is:
Fig. 4. Propagation of fatigue crack in an Al alloy [6].
Fig. 5. Crack growth under cyclic loading: with some overload and without overload.
Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490 485
Z
r z

z
2
a
2
p ; where z x iy: 12
The function is analytic except for a 6 x 6 a; y 0. The boundary
stresses follow Eq. (11). At innity, where jzj !1, the result is
r
x
r
y
r and s
xy
0, and on the crack surface r
y
s
xy
0,
which means that the boundary conditions are satised.
It is more convenient to convert to a coordinate system with the
origin at the crack tip, hence z should be replaced by z a. Turn-
ing to the general problem, which is shown in Fig. 7, that has the
form:
Z
f z

z
p : 13
According to Eq. (11), both r
y
and s
xy
are zero at the crack surface.
The required real and constant value of f(z) at the crack tip is given
by the notation K
I
, hence
Z
jzj!0

K
I

2pz
p : 14
Taking polar coordinates from the origin with z re
ih
, the stress at
the crack tip can be calculated from Eqs. (11) and (14) to be:
r
x

K
I

2pr
p cos
h
2
1 sin
h
3
sin
3h
2

;
r
y

K
I

2pr
p cos
h
2
1 sin
h
2
sin
3h
2

; 15
s
xy

K
I

2pr
p sin
h
2
cos
h
2
cos
3h
2
:
The parameter K
I
in these equations is known as the stress intensity
factor.
2.2. Plastic zone in fatigue crack propagation
It was discussed in detail previously that the stresses in the
vicinity of a crack tip are of the form of those written in Eq. (15),
or can be simplied to
r
ij

K
I

2pr
p f
ij
h; 16
where r and h are the polar coordinates of a point with respect to
the crack tip, and K
I
is the stress intensity factor. This equation indi-
cates that all the stresses tend to innity as r ! 0. However, the
theoretical very high elastic stresses near the tip exceeds the yield
strength of real materials and therefore, induces a small plastically
deformed volume called the crack tip plastic zone (PZ), as illus-
trated in Fig. 8.
The size of the PZ can be predicted as follows. Eq. (16) can be
rewritten in terms of stresses in y direction in front of the crack as:
r
y

K
I

2pr
p : 17
To obtain the corresponding PZ in Fig. 7, we can substitute the
stress r
y
in Eq. (17) with the yield strength r
YS
and r with PZ. The
result is
PZ
I

1
2p
K
I
r
YS

2
: 18
Eq. (18) is called the rst prediction, which is a rough estimation of
the plastic zone. It was obtained by assuming that the stress distri-
bution is elastic. The subscript I in this equation is used to describe
the rst estimation of PZ, as opposed to the second estimation.
However, this prediction is not strictly correct because it is based
on an elastic crack tip solution. By assuming an elasticperfectly
plastic solution, it is found that the corrected PZ is as much as twice
the size of the rst prediction of PZ. Although we know that this
assumption is also not strictly correct and will give an overestima-
tion of the PZ, it is worth mentioning here [13]. Fig. 9 illustrates how
Irwin [14] models the correction. Referring to Fig. 9, in order to
maintain the force balance, the shaded area in (a) should be the
same as the shaded area in (b), or the following expression must
be true:
Z
PZ
I
0
K

2pr
p dr PZ
II
r
YS
: 19
Solving Eq. (19),
PZ
II

2
p
q
K

PZ
I
p
r
YS
: 20
Combining Eq. (20) with Eq. (19) will easily lead to
Fig. 6. Mode I crack under bi-axial stress.
Fig. 7. General Mode I problem. Fig. 8. The basic concept of how plastic zone is formed near the tip.
486 Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490
PZ
II
2 PZ
I

1
p
K
I
r
YS

2
: 21
Knowing that Eq. (18) was derived based on a linear elastic assump-
tion, which gives an underestimation, and Eq. (20) was derived
based on an elasticperfectly plastic assumption, which gives an
overestimation, we now have a general idea that the plastic zone
size should be somewhere in between them.
2.2.1. The shape of plastic zone
Theoretically the plastic zone shape varies depending on the
distance along the crack front to a free surface. The two extreme
cases are on the surface and in the middle of a thick sample. On
the surface, which is governed by plane stress, the plastic zone
shape differs from the inside of the sample, which is governed by
plane strain.
Using the von Mises yield criterion, the plane stress plastic zone
boundary [15] is given by:
r C
K
max
r
YS

2
1
3
2
sin
2
h cos
h
2

: 22
Similar to Eq. (22), the plane strain plastic zone boundary is given
by:
r C
K
max
r
YS

2

3
2
sin
2
h 1 2v
2
1 cos h

: 23
Although all the approaches discussed here are based on the von
Mises yielding criterion, similar results can be obtained when calcu-
lated using the Tresca yielding criterion. Fig. 10 illustrates the shape
of the PZ derived from Eqs. (22) and (23). The shape of this PZ has
been veried by several researchers by various methodologies.
The shape of the PZ has been used to verify the variation in path
of the FCP in steel materials [13].
3. Computational method
A two-dimensional plane strain model was created to estimate
the size of the plastic zone in front of a crack. This model compares
the condition of material with two different microstructures, con-
ventional tempered martensite and multiphase polygonal ferrite-
martensite. Here, the conventional tempered martensite was trea-
ted as homogeneous as opposed to the multiphase material that
was treated as non-homogeneous material (see Fig. 11). Commer-
cially available FEA software, ABAQUS [16], was used to study the
plastic zone of two different microstructures. As a test specimen,
ASTM E647 [17] standard was chosen. Because of its symmetrical
nature, only a half of the specimen needs to be modeled. The model
was created and meshed in ABAQUS/CAE (ABAQUS pre-processor).
The following steps were taken for the FEA study: step 1 FEA model
verication followed by step 2, which is plastic zone area estima-
tion, and step 3 FEA of homogeneous/multiphase material with
variation of the ferrite phase.
3.1. FEA model verication by stress intensity factor comparison
Before comparing the plastic zone between two microstruc-
tures, the FEA model needs to be veried. The stress intensity fac-
tor, K, from numerical calculations and that of FEA were compared.
Fig. 10. Plane stress and plane strain condition plastic zone boundary as a function of h.
Fig. 9. (a) Elastic stress distribution and rst estimation of PZ (under-estimation of PZ), and (b) second estimation of PZ based on elasticperfectly plastic concept (over
estimation of PZ).
Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490 487
For a compact tension specimen, the relation between the applied
load, P, and the stress intensity factor, K, is as follows [18]:
K
P
B

W
p
2 a
1 a
3=2
0:886 4:64a 13:32a
2
14:72a
3
5:6a
4

24
where B and W are the specimens thickness and width, respec-
tively. a is the relative crack length (a/W), and P is the applied load.
For this analysis, an elastic material was used because the plastic
zone is not studied at this step. The following values were used
for both the numerical verication and the modeling, a 0:45,
W = 38.1 [mm], and B = 10 [mm].
For this global model, the element types used were CPE3 (3-
node linear element) and CPE4R (4-node bilinear element, with re-
duced integration) with the applied load P = 100 [N]. The Youngs
Modulus, E [GPa], was 185.5 and the Poissons Ratio, m was
0.3125. From this estimation result, it was found that the plastic
Fig. 11. Concept of the modeling used to calculate and compare the morphology and the size of the plastic zone in front of the crack tip. By embedding two different
microstructures, direct comparison of the size of the PZ can be calculated.
Fig. 12. Finite element modeling of the ASTM 647 sample. The multiphase microstructure (local model) was embedded to the sample (global model).
488 Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490
zone was within 1 mm in height, see Fig. 12. Therefore it was
determined that only 1 mm square area needs to be studied.
3.2. Plastic zone area estimation
Since the target of this FEA study is to show that the plastic zone
varies by changing microstructure, only the plastic zone area
around the crack tip needs to be analyzed with two different
microstructures. Although it is possible to mesh a whole model,
such attempt is both cumbersome and numerically erroneous.
The smarter way to do this was utilizing the sub-modeling or mul-
ti-level modeling. One millimeter square was made for the local
model.
3.3. FEA of homogeneous and multiphase material
The sub-modeling technique is used to study a local part of a
model with rened mesh based on the FEA result of a global model
with coarse mesh. There are two steps for sub-modeling. (1) Per-
form analysis of a global (whole) model to nd the boundary con-
dition of a local area. (2) Perform analysis of a local model based on
interpolation of the solution from the global model.
The rst step was done in the previous analysis. In this study, the
global model has approximately 10,000 elements. A new local area
model was meshed with approximately 20,000 elements using
PPM2OOF. Exactly the same meshed model was used for both the
homogeneous and multiphase microstructure. Material properties
Fig. 13. Appearance of the element mesh of local models. Each local model consists of approximately 20,000 elements.
Fig. 14. Representative of the calculation results, showing 10%, 40%, and 90% ferrite fractions.
Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490 489
were assigned properly to differentiate between these two models.
The following values were used for the local model: Ferrite, Youngs
Modulus [GPa] 124.7, Yield Stress [MPa] 359, and Poissons Ratio
was 0.3125. For Martensite, Youngs Modulus [GPa] 200.1, Yield
Stress [MPa] 1449, and Poissons Ratio was 0.3125.
Displacement of the local model boundary was interpolated
from the result of the global model automatically. Results of the
von Mises stress and plastic zone are shown in Fig. 12. Fig. 13
shows the local models having ferrite percentage from 10% to
90% with the interval of 5%. On these local models, the same
boundary conditions obtained from the global model were applied.
4. Analysis results
Typical analysis results are shown in Fig. 14. To represent other
data, 10%, 40%, and 90% a-Fe microstructures were presented here.
The size of the plastic zones of all models was than calculated.
Fig. 15 shows the results of the plastic zone normalized to the larg-
est calculation result. The size of the plastic zone increases from
10% almost linearly up until 40%. The slope gradually decreases un-
til the ferrite fraction is 65% and nally saturated. This implies that
from the fracture mechanics point of view, the optimum ferrite
fraction is about 65%. Beyond this point, the ferrite fraction addi-
tion becomes detrimental in term of its static properties, such as
yield strength and ultimate tensile strength.
This analysis results are in the near future will be veried
experimentally, and numerically using three-dimensional model.
5. Discussion and conclusion
A theoretical fracture mechanics approach was used to develop
a microstructure that is more resistant to the fatigue than the con-
ventional one. A nite element model, utilizing sub-modeling con-
cept was used to prove that the multiphase material has a larger
plastic zone size compared to the conventional homogeneous
material for identical load levels. Furthermore, the model was also
used to nd approximate optimum ferrite fraction to balance the
plastic zone size.
6. Notes for future plan
The shape of the ferrite in this research was based on micro-
structure, which was two-dimensional. Plane strain condition al-
lows analysis closed enough for estimation. However, three-
dimensional analysis would be giving different information and
is undergoing now. Furthermore, experiment is also planned for
very near future. The introduction of ferrite constituents into the
martensitic microstructure may also be potentially advantageous
to the corrosion resistance. To explore the effect on corrosion, a re-
search on it is also being considered.
References
[1] G. Krauss, Steels: Heat Treatment and Processing Principles, vol. 1, ASM
International, pp. 6472.
[2] M.E. Fine, Y.W. Chung, Fatigue failure in metals, ASM Handbook: Fatigue and
Fracture, vol. 19, ASM International, 1996, pp. 6472.
[3] C. Laird, The inuence of metallurgical structure in the mechanisms of fatigue
crack, Fatigue Crack Propagation, ASTM STP 415 (1967) 131168.
[4] C.Y. Kung, M.E. Fine, Metallurgical Transactions A 10A (1979) 603.
[5] C.Q. Bowles, D. Broek, International Journal of Fracture Mechanics 8 (1972) 75
85.
[6] P.J.E. Forsyth, D.A. Ryder, Metallurgica (1961) 117124.
[7] J. Lankford, D.L. Davidson, Acta Metallurgica 31 (1983) 12731284.
[8] K. Tanaka, Mechanics and micromechanics of fatigue crack propagation,
Fracture Mechanics: Perspectives and Directions (Twentieth Symposium),
ASTM STP 1020, 1989, pp. 151183.
[9] J. Oh, N.J. Kim, S. Lee, E.W. Lee, Materials Science and Engineering A340 (2003)
232242.
[10] Y. Prawoto, R.A. Winholtz, The growth rate and plastic zone size variations of a
fatigue crack grown in a tensile residual stress eld, in: Proceedings of the
Eighth International Conference on Nuclear Engineering, American Society of
Mechanical Engineers, paper 8785, 2000.
[11] D.G. Pavlou, Engineering Structures 22 (2000) 17071713.
[12] H.M. Westergaard, Journal of Applied Mechanics 61 (1939) A49A53.
[13] D. Broek, Elementary Engineering Fracture Mechanics, Kluwer Academic
Publishers, 1996.
[14] H.P. Rossmanith, Fracture Research in Retrospect An Anniversary Volume in
Honour of G.R. Irwins 90th Birthday, A.A. Balkema Publishers, Rotterdam,
Netherland, 1997.
[15] T.L. Anderson, Fracture Mechanics: Fundamental and Applications, CRC Press,
1994.
[16] ABAQUS Users Manual, in: Analysis, vol. II, ABAQUS Inc., 2006.
[17] ASTM International, Standard test methods for tension testing of metallic
materials, ASTM Designation E 647-04.
[18] ASTM International, Standard test methods for tension testing of metallic
materials, ASTM Designation E 8-04.
Fig. 15. Relation between ferrite fraction and plastic zone sizes.
490 Y. Prawoto et al. / Computational Materials Science 47 (2009) 482490

Вам также может понравиться