Anda di halaman 1dari 6

J. Mater. Sci. Technol., Vol.23 No.

2, 2007 277

Element Analysis of Instrumented Sharp Indentations into Pressure-


sensitive Materials
Minh-Quy LE1)† and Seock-Sam KIM2)
1) Department of Mechanics of Materials & Structures, Faculty of Mechanical Engineering, Hanoi University
of Technology, Dai Co Viet Road, Hanoi, Vietnam
2) School of Mechanical Engineering, Kyungpook National University, Daegu 702-701, South Korea
[Manuscript received January 13, 2006, in revised form April 14, 2006]

Finite element analysis was carried out to investigate the conical indentation response of elastic-plastic solids
within the framework of the hydrostatic pressure dependence and the power law strain hardening. A large
number of 40 different combinations of elasto-plastic properties with n ranging from 0 to 0.5 and σy /E
ranging from 0.0014 to 0.03 were used in the computations. The loading curvature C and the average
contact pressure pave were considered within the concept of representative strains and the dimensional analysis.
Dimensionless functions associated with these two parameters were formulated for each studied value of the
pressure sensitivity. The results for pressure sensitive materials lie between those for Von Mises materials and
the elastic model.
KEY WORDS: Finite element analysis; Indentation; Mechanical Properties; Pressure-sensitive
materials

1. Introduction (see Fig.1). However, the strain-hardening of materi-


als has not explicitly involved under a functional re-
Indentations have been widely used to deter- lation in their work above. Ganneau et al.[23] for-
mine the hardness of materials[1] . Actually, instru- mulated a functional form relating the indentation
mented indentation techniques are developing exten- hardness to the cohesion and the frictional-angle of
sively to characterize various materials including met- materials.
als, metallic alloys, ceramics, glasses, polymers, and In the present work, the dimensional analysis and
coated materials, etc[2] . The dimensional analysis and the concept of representative strain are extended to
the concept of representative strain are widely used in study instrumented sharp indentations of pressure-
instrumented sharp indentations to formulate many sensitive materials. Attention is focused on the influ-
dimensionless functions, which relate indentation pa- ence of pressure sensitivity into the evolution of the
rameters to the indenter geometry and the indented loading curvature C and the average contact pressure
material0 s mechanical properties such as elastic mod- pave . Their relationship with the material0 s mechan-
ulus, yield stress and strain hardening exponent[3–6] . ical properties is formulated for each given value of
However, most of previous research works are based the pressure sensitivity.
on Von Mises criterion[7] , in which the influence of
hydrostatic pressure on material plastic deformation 2. Theoretical Backgrounds
is neglected.
In practice, hard metals, ceramics, bulk metallic 2.1 Yield criterion
glasses and polymers have been known to exhibit hy- In the classical plasticity theory, it is generally as-
drostatic pressure dependent plastic behavior. The sumed that hydrostatic pressure has no effect on ma-
pressure-sensitive yielding occurs from the basic flow terial plastic deformation, and plastic dilatation is ne-
mechanism in polymers[8–14] from phase transforma- glected. The Von Mises yield criterion is widely used
tion in ceramics[15] . Voids and other forms of defects to model this class of materials such as metals and
can also result in macroscopic pressure sensitivity[16] . metallic alloys[7] . However, for many materials in-
The Mohr-Coloumb and Drucker-Prager yield cluding hard metals, ceramics, glasses, and polymers,
functions have been suggested to represent the yield pressure-sensitive yielding and plastic volumetric de-
behavior of these materials[17–22] . Moreover, finite formation are exhibited. They are so-called pressure-
element analyses (FEA) of sharp indentation within sensitive materials and modeled according to Drucker-
the framework of the above yield criteria have been Prager yield criterion as follows[24] :
undertaken for elastic-plastic materials with linear
Φ(σij ) = σe + aσm − (1 − a/3)σy = 0 (1)
strain hardening[17] , and for elastic-perfectly plastic
materials[18,19,23] . Giannakopoulos and Larsson[17] and p
presented a general response of the pressure-sensitive σm = σkk /3 and σe = 3J2 (2)
materials under sharp indentation. We showed that
the pressure sensitivity and the strain hardening where σm and σe are the hydrostatic stress and Mises
increase the loading curvature C and the average equivalent effective stress, respectively; a is a mate-
contact pressure pave , and decrease the residual in- rial constant √
that measures the pressure sensitivity of
dentation depth-maximum indentation depth ratio yielding; (a/ 3 is the ratio of the plastic volumetric
strain to the effective plastic shear strain, a<3/2) and
† Ph.D., to whom correspondence should be addressed, σy denotes the initial uniaxial yield stress in compres-
E-mail: quylm@mail.hut.edu.vn. sion. The Von Mises yield criterion is recovered when
278 J. Mater. Sci. Technol., Vol.23 No.2, 2007

Fig.1 Schematic representation of a conical indentation: (a) axisymmetric model of the indenter and specimen,
(b) typical indentation load-depth curve

as shown schematically in Fig.2. A simple elastic-


plastic, true stress-true strain behavior is assumed to
be
σ = E · ε, (σ ≤ σy )
σ = K · εn , (σ ≥ σy ) (4)
0
where E is the Young s modulus, K a strength coeffi-
cient, n the strain hardening exponent, σy the initial
compressive uniaxial yield stress and σy the corre-
sponding yield strain, such that

σy = Eεy = Kεny (5)

Here the yield stress σy is defined at zero offset strain.


The total strain, ε, consists of elastic strain εe and
plastic strain εp :

ε = εe + εp (6)

The representative strain εr , defined by Dao et al.[3]


corresponds to the strain accumulated beyond the
Fig.2 Power law elastic-plastic stress-strain behavior yield point εy .
ε = εy + εr (7)

a=0. More information on the pressure-sensitive yield where εy is the strain reached at the yield stress, σy .
criterion can be found in literature [11,24]. With Eqs.(5) and (7), when σ>σy , Eq.(4) becomes
A direct measurement of the pressure-sensitivity ³ E ´n
index a relies on shear experiment under pressure. σ = σ y 1 + εr (8)
σy
An alternative method to determine a is to perform
compressive tests under pressure p; let σyp denote the To complete the material constitutive description,
compressive yield stress when superimposed by hy- Poisson’s ratio is designated as ν, and the incremen-
drostatic pressure p[25] : tal theory of plasticity with Von Mises criterion (when
a=0) or Drucker-Prager criterion is assumed.

a= (3)
3+α 3. Finite Element Model
σyp −σy
where α= p . In the case of transformation ce- Since the indentation problem of a rigid cone into
ramics, Chen[15] reported that a is 0.95 for Mg-PSZ half-space is axisymmetric (Fig.1(a)), only one-half of
and 1.33 for Ce-TZP, and a is about 1.19 for ZrO2 - the system is used in the modeling. Therefore, elastic-
containing ceramics[25] . Further, a is in the range plastic indentation was simulated using the axisym-
from 0.024 to 0.11 for steels[26,27] , and from 0.17 to metric capacities of the MARC finite element code.
0.43 for polymers[28] . Given that the projected contact area for a conical
indenter, a Berkovich indenter and a Vickers inden-
2.2 Power-law elastic-plastic behavior ter are A=πh2 tan2 θ; A=24.56h2 ; and A=24.50h2 , re-
Elastic-plastic behavior of many engineering solid spectively; the indenter was thus modeled as a rigid
materials can be modeled by a power law description, cone with a half-included angle of θ=70.3◦ . This angle
J. Mater. Sci. Technol., Vol.23 No.2, 2007 279

Fig.3 Finite element mesh of: (a) half-space, (b) zoom of the contact zone

than contact radius. These dimensions were found to


Table 1 Mechanical properties of material used be large enough to approximate a semi-infinite half-
in the computations space for indentations. This was evidenced by an in-
sensitivity of calculated results to further increase in
E/GPa σy /MPa σy /E
specimen size.
210 300 0.001428
10 30 0.003
Elements were finest in the central contact area
50 200 0.004 and became gradually coarser outwards. At the max-
90 500 0.00556 imum indentation depth, no less than 55 elements
130 1000 0.00769 came into contact. It enables an accurate determi-
10 100 0.01 nation of the real impression size. Frictionless roller
130 2000 0.01538 boundary conditions were applied along the center-
150 3200 0.02133 line and bottom. Outside surfaces were taken as free
10 270 0.027∗ surfaces. The interaction between the rigid indenter
10 300 0.03∗ and specimen was modeled by contact elements with-
Notes: ∗ correspond to unstable results for a=0.3 out friction. The residual stresses were not taken into
and a=0.58. Poisson0 s ratio, ν is fixed at 0.3. account in the analysis. Displacement-controlled pro-
Strain hardening exponent, n, is varied from 0, cedure was used in this work.
0.1, 0.3 to 0.5. a=0 (Von Mises materials), a=0.3 Since indentation tests have been used for a great
and a=0.58
variety of materials, ranging metals, metallic al-
loys, ceramics, polymers, glasses, etc., it is necessary
to model indentation using general though simpli-
fied descriptions for the mechanical properties of the
material, including power-law strain hardening and
pressure-sensitivity of yielding, which occurs even in
metals and metallic alloys[16,26,27] . Therefore, a large
number of 40 different combinations of elastic-plastic
properties with n ranging from 0 to 0.5 and σy /E
ranging from 0.0014 to 0.03 were used in the compu-
tations. This wide range of parameters covers mostly
engineering solid materials. Three different levels of
pressure-sensitivity were considered: a=0, a=0.3, and
a=0.58. The mechanical properties of material used
in the computations are given in Table 1.
Fig.4 Validation of the finite element model: indenta- The finite element model was well tested for con-
tion load-depth curve obtained by FEA and ex- vergence and accuracy, and then validated by com-
perimental data digitized in literature [3] paring the indentation load-depth curves obtained
by FEA with experimental data for 7075-T651 alu-
gives the same area-to-depth ratio as the Berkovich minum, which was investigated by Dao et al.[3] . The
or Vickers indenter, which are commonly used in mechanical properties of 7075-T651 aluminum are
instrumented-indentation experiments[2] . Since any taken as[3] : E=70.1 GPa, ν=0.33, σy =500 MPa, and
typical indentation experiments would involve blunt- n=0.12. The experimental indentation data were dig-
ing of the indenter tip, the cone tip was smoothed by itized. Figure 4 shows a good agreement between the
a sphere of radius much smaller than the indentation results obtained by FEA and those from experiments.
depth. This also eliminates any possible convergence Further, the results obtained by Dao et al.[3] for di-
errors due to sharp corners. mensionless functions related to the loading curvature
The specimen was modeled as a large cylinder C and the average contact pressure pave in the case
represented by around 2100 large strain four-node of Von Mises materials are reproduced in the present
axisymmetric elements (Fig.3). The radius and the work as indicated in the following section. This also
height of the sample are equal or seventy times larger proves the validity of the finite element model.
280 J. Mater. Sci. Technol., Vol.23 No.2, 2007

Fig.5 Dimensionless function II1 constructed for a=0.3 using three different values of the representative strain
εr : (a) εr =0.01, (b) εr =0.0164, (c) εr =0.033. A representative strain εr =0.0164 allows the construction of
II1 to be independent of strain hardening exponent n

4. Results and Discussion influence of the choice of the representative strain on


the evolution of the function II1 , for a=0.3. For each
4.1 Loading curvature C value of a, the representative strain has been defined
Figure 1(b) shows the typical indentation load- as a strain level where II1 can be constructed indepen-
depth response of an elastic-plastic material to sharp dently of the strain hardening exponent n: εr =0.033[3]
indentation. From the dimensional analysis and ge- for a=0 (Von Mise materials), εr =0.0164 for a=0.3,
ometrical similarity of a conical/pyramid indenter, and εr =0.0076 for a=0.58.
Cheng and Cheng[6] have demonstrated that the in- The ratio C/σr is higher for larger a. As E ∗ /σr
dentation load P during loading is proportional to the increases and reaches high values, C/σr becomes con-
square of the indentation depth h: stant. For each value of a, the curve of C/σr vs E ∗ /σr
is equivalent for elastic-work hardening materials[29] .
P = Ch2 (9) It includes three successive parts: a linear part (elastic
behavior: C/E ∗ =(2/π)tanθ), a curved part (elastic-
where C is the loading curvature, which is a measure plastic behavior) and a horizontal line (full plastic-
of the resistance of the material to indentation. C can ity) as indicated in Fig.6. As a increases, the sizes
be obtained by a least square fitting procedure from of the linear part and the curve part increase and
the loading part of an indentation P -h curve. It was the size of the fully plastic part decreases. When
verified, for all computational simulations with mod- a=0.58, the fully plastic part seems to disappear and
eled materials, that the P -h curve is well reproduced the elastic-plastic regime is dominant. Figure 7 shows
by Eq.(9). the evolution of C/E ∗ vs E ∗ /σr . With increasing a,
For Von Mises materials (a=0), which was consid- C/E ∗ increases and reaches a limit in values as 1.39,
ered by Dao et al.[3] , the authors defined the dimen- 1.56 and 1.75 for a=0 (Von Mises materials), a=0.3
sionless function II1 as below: and a=0.58, respectively. It can be concluded that
³ E∗ ´ if C/E ∗ is higher than 1.4, the pressure-sensitivity is
C probably involved.
II1 ,n = (10)
σr σr Figures 6 and 7 show that the results for pressure-
sensitive materials are limited by those for Von Mises
where σr is the stress corresponding to a representa-
materials as lower bound and the elastic model as up-
tive strain εr , and E ∗ is the reduced elastic modulus
per bound. With increasing a, the results for pressure-
for a rigid indenter:
sensitive materials can approach the elastic model
E (C/E ∗ =1.78). Therefore, the material has more elas-
E∗ = (11) tical behavior with increasing a. Thus, the representa-
1 − ν2
tive strain associated with the dimensionless function
In the case of a Berkovich indenter, a representa- II1 =C/σr decreases as a increases.
tive strain of 3.3% allowed Dao et al.[3] to construct a
function II1 independent of the strain hardening ex- 4.2 Average contact pressure
ponent. The function II1 was constructed with the Tabor[1] used the slip-line theory to derive the rela-
results of a parametric study of 76 cases with various tion between hardness and yield stress for rigid plastic
elastic-plastic parameters representing the behavior of solids as follows:
typical engineering metals (Eq.(A1)). Recently, Bu- H ≈ pave = C0 Y (12)
caille et al.[4] constructed the function II1 by a sim-
ilar numerical study on 24 cases. Further, the au- where C0 is a constraint factor (C0 =3). For elastic-
thors showed that three times fewer simulations are perfectly-plastic, Tabor[1] showed that C0 is around
sufficient to reproduce the results presented by Dao 3. Based on experimental observations for Von Mises
et al.[3] and constructed the function II1 for indenter materials, earlier studies by Tabor[1] and Johnson[29]
angles 60◦ , 50◦ , and 42.3◦ . showed that the hardness can be also estimated by
The function II1 for a equal to 0.3 and 0.58 are the following formula:
constructed in the present study (see the appendix H ≈ pave = C0 σr (13)
for the list of functions). Figure 5 clearly shows the
J. Mater. Sci. Technol., Vol.23 No.2, 2007 281

Fig.9 Variation of pave /E ∗ vs σr /E ∗ with three different


Fig.6 Variation of the dimensionless function II1 vs levels of the pressure sensitivity
E ∗ /σr with three different levels of the pressure
sensitivity
To study the average contact pressure, the di-
mensionless function II2 =pave /σr suggested by Dao
et al.[3] was used as follows:
³ E∗ ´ p
ave
II2 ,n = (14)
σr σr

The function II2 for different levels of pressure-


sensitivity are constructed within the concept of rep-
resentative strain in the present study as shown in
Fig.8. Similar to the procedure taken to obtain the
representative strain for II1 , it is found that for a=0
(Von Mises materials), the results reported by Dao
et al.[3] is reproduced with a representative strain
of 8.2%, and εr =0.068 for a=0.3, and εr =0.048 for
a=0.58 (see the appendix for the list of functions). It
is found that the representative strain decreases as a
Fig.7 Variation of C/E ∗ vs E ∗ /σr with three different increases.
levels of the pressure sensitivity Equation (14) was used to consider the evolution
of the dimensionless function pave /E ∗ vs σr /E ∗ for
various values of n ranging from 0 to 0.5 (Fig.9).
The pressure- sensitivity a increases pave /σr as well as
pave /E ∗ . With increasing a, the dimensionless func-
tion II2 =pave /σr increases. pave /σr varies from 2.66
to 4.20 and from 3.79 to 6.65 for a=0.3 and a=0.58,
respectively. The ratio pave /E ∗ increases with increas-
ing σr /E ∗ . However, pave /σr decreases with increas-
ing σr /E ∗ especially for high value of a.
Similar to the observation on the loading cur-
vature C, results on the average contact pressure
for pressure-sensitive materials are limited by those
for Von Mises materials as lower bound and the
elastic model as upper bound as shown in Fig.8
(pave /E ∗ =1/2tanθ∼0.178[29] ). With increasing a, the
results for pressure-sensitive materials tend to the
elastic model. However, the upper bound is very far
Fig.8 Variation of the dimensionless function II2 vs from the results for pressure-sensitive materials in this
σr /E ∗ with three different levels of the pressure case.
sensitivity
5. Conclusion
where σr is calculated at a representative strain
εr =0.08. The loading curvature C and the average contact
Experiments showed that the factor C0 in Eq.(12) pressure pave are studied within the context of repre-
is about 4.5 for silica and glasses[30] , about 8.2 for sentative strain and dimensional analysis. The results
ZrO2 -containing ceramics[25] and in order of 20-30 for on these two parameters for pressure-sensitive mate-
cementitious materials[31–33] . rials lie between those for Von Mises materials and
282 J. Mater. Sci. Technol., Vol.23 No.2, 2007

the elastic model. C/E ∗ and pave /E ∗ increase with [12] J.A.Sauer, K.D.Pae and S.K.Bhateja: J. Macromol.
increasing pressure sensitive level. While the pressure Sci. Phys., 1973, 88, 631.
sensitivity increases, C/E ∗ for pressure-sensitive ma- [13] W.A.Spitzig and O.Richmond: Poly. Eng. Sci., 1979,
terials can approach the elastic model (C/E ∗ =1.78). 19, 1129.
The representative strains associated with the two [14] L.M.Carapellucci and A.F.Yee: Poly. Eng. Sci., 1986,
26, 920.
studied dimensionless functions decrease with increas-
[15] L.W.Chen: J. Am. Ceram. Soc., 1991, 74, 2564.
ing the pressure sensitivity. Dimensionless functions [16] H.Y.Jeong and J.Pan: Int. J. Solids Struct., 1995, 32,
associated with the two studied parameters are estab- 3669.
lished for each studied value of the pressure sensitiv- [17] A.E.Giannakopoulos and P.L.Larsson: Mech. Mater.,
ity. 1997, 25(1), 1.
[18] R.Vaidyanathan, M.Dao, G.Ravichandran and
S.Suresh: Acta Mater., 2001, 49(18), 3781.
Acknowledgement [19] M.N.M.Patnaik, R.Narasimhan and U.Ramamurty:
This work was partially supported by the Research Acta Mater., 2004, 52(11), 3335.
Program 2005 of Hanoi University of Technology, Viet- [20] P.E.Donovan: Acta Metall., 1989, 37(2), 445.
nam. [21] P.B.Bowden and J.A.Jukes: J. Mater. Sci., 1972, 7,
52.
[22] R.Quinson, J.Perez, M.Rink and A.Pavan: J. Mater.
REFERENCES Sci., 1997, 32(5), 1371.
[23] F.P.Ganneau, G.Constantinides and F.J.Ulm: Int. J.
[1 ] D.Tabor: The Hardness of Metal, Clarendon Press, Solids Struct., 2006, 43(6), 1727.
Oxford, 1951. [24] W.F.Chen and D.J.Han: Plasticity for Structural En-
[2 ] A.C.Fischer-Cripps: Nanoindentation, Springer- gineers, Springer-Verlag, New York, 1988.
Verlag, New York, 2002. [25] I.W.Chen and P.E.Reyes-Morel: J. Am. Ceram. Soc.,
[3 ] M.Dao, N.Chollacoop, K.J.Van Vliet, T.A.Venkatesh 1986, 69, 181.
and S.Suresh: Acta Mater., 2001, 49, 3899. [26] W.A.Spitzig, R.J.Sober and O.Richmond: Acta Met-
[4 ] J.L.Bucaille, S.Stauss, E.Felder and J.Michler: Acta all., 1975, 23, 885.
Mater., 2003, 51(6), 1663. [27] W.A.Spitzig, R.J.Sober and O.Richmond: Metall.
[5 ] N.Chollacoop, M.Dao and S.Suresh: Acta Mater., Trans., 1976, 7A, 1703.
2003, 51(13), 3713. [28] A.J.Kinloch and R.J.Young: Fracture Behaviour of
[6 ] Y.T.Cheng and C.M.Cheng: Mater. Sci. Eng.: R, Polymers, Elsevier Applied Science, London, 1983.
2004, 44(4-5), 91. [29] K.L.Johnson: Contact Mechanics, Cambridge Univer-
[7 ] R.Hill: The Mathematical Theory of Plasticity, sity Press, Cambridge, 1985.
Clarendon Press, Oxford, 1950. [30] I.J.McColm: Ceramics Hardness, Plenum Press, New
[8 ] W.Whitney and R.D.Andrews: J. Polym. Sci. Pol. York, 1990.
Sym., 1967, 16C, 2981. [31] M.Kholmyansky, E.Kogan and K.Kovler: Mater.
[9 ] S.S.Stenstein and L.Ongchin: Am. Chem. Soc. Poly., Struct., 1994, 27, 584.
1969, 10, 1117. [32] S.Iragashi, A.Bentur and S.Mindess: Adv. Cement
[10] S.Rabinowtz, I.M.Ward and J.S.C.Party: J. Mater. Res., 1996, 8(30), 877.
Sci., 1970, 5, 29. [33] G.Constantinides, F.J.Ulm and K.Van Vliet: Mater.
[11] D.C.Drucker: Metall. Trans., 1973, 4, 667. Struct., 2003, 36, 1991.

Appendix

Dimensionless function II1 for a=0 (Von Mises Materials, εr =0.033):


C1 h ³ E ∗ ´i3 ³ E ∗ ´i2 h ³ E ∗ ´i
II1 = = −1.131 ln + 13.625 − 30.594 ln + 29.267 (A1)
σ0.033 σ0.033 σ0.033 σ0.033
for a=0.3 (εr =0.0164):
C1 h ³ E ∗ ´i3 ³ E ∗ ´i2 h ³ E ∗ ´i
II1 = = −2.9584 ln + 41.6068 − 136.3520 ln + 147.2762 (A2)
σ0.0164 σ0.0164 σ0.0164 σ0.0164
for a=0.58 (εr =0.0076):
C1 h ³ E ∗ ´i3 ³ E ∗ ´i2 h ³ E ∗ ´i
II1 = = −11.1418 ln + 185.6190 − 879.5642 ln + 1353.2351 (A3)
σ0.0076 σ0.0076 σ0.0076 σ0.0076
Dimensionless function II2 for a=0 (Von Mises Materials, εr =0.082):
pave ³σ ´2 ³σ ´
0.082 0.082
II2 = = −15.4944 ∗
− 15.1699 ∗
+ 2.7497 (A4)
σ0.082 E E
for a=0.3 (εr =0.068):
pave ³σ ´2 ³σ ´
0.068 0.068
II2 = = −210.9303 ∗
− 31.3284 ∗
+ 4.1581 (A5)
σ0.068 E E
for a=0.58 (εr =0.048):
pave ³σ ´2 ³σ ´
0.048 0.0648
II2 = = −373.5926 ∗
− 96.6850 ∗
+ 6.6378 (A6)
σ0.048 E E

Anda mungkin juga menyukai