Beruflich Dokumente
Kultur Dokumente
Indentation of Piezoelectric
A. E. Giannakopoulos
Department of Materials Science
Materials
and Engineering, The present paper deals with theoretical and computational analysis of quasi-static, nor-
Massachusetts Institute of Technology, mal indentation of a transversely isotropic, linear elastic, piezoelectric half-space by a
Cambridge, MA 02139 rigid spherical indenter. The contact is axisymmetric, nonconforming, monotonically ad-
Mem. ASME vancing with load, frictionless and adhesionless. The indenter was modeled either as
perfect conductor or as perfect insulator. The mechanical and electrical fields below the
surface were examined. The issues of mechanical and dielectric strength due to indenta-
tion were examined using Weibull statistics of surface imperfections. The particular cases
of PZT-4, PZT-5A, BaTiO 3 , and Ba 0.917Ca 0.083 TiO 3 indented by rigid punches having
either zero electrical potential or zero electric charge were solved with finite element
analysis. S0021-89360002502-2
Journal of Applied Mechanics Copyright 2000 by ASME JUNE 2000, Vol. 67 409
21i 22 1 i , 21i 22 1 i , 8 0a M 6 M 1 M 2 M 5
. (23b)
c 11c 12 r M 1 M 4 M 2 M 3
21i 22 1 i . (17)
The constants M i are given in the Appendix. In order to have a
Inverting the Hankel transforms, the solution at the surface (z nontrivial solution, M 1 M 4 M 2 M 3 and M 5 M 8 M 6 M 7 , which
0) can be represented in a general form as are always satisfied for known piezoelectric ceramics. The above
analytical expressions agree with those derived by Chen and Ding
17 who used the potential theory method; the present results are
u z r,0 M 1 A 1 M 2 A 2 J 0 r d , (18)
0
more straightforward and simpler.
zz r,0
0
M 5 A 1 M 6 A 2 J 0 r d , (19)
4 Insulating Sphere
Equations 810 and 13 can be cast as a system of two dual
integral equations for the interior and the exterior problem.
The continuity contact condition gives the relation between the
r,0 M 3 A 1 M 4 A 2 J 0 r d , (20)
0
penetration depth h and the contact radius a, h2a 2 /D, which is
exactly as for the uncoupled mechanical indentation. The analyti-
D z r,0
0
M 7 A 1 M 8 A 2 J 0 r d , (21)
cal expressions for the contact pressure p(r), the charge distribu-
tion under the contact q(r), and the resultant force P, are given in
a previous work 18. Of interest are the relations for the surface
radial displacement and electric potential and were derived in this
where the constants M i are defined in the Appendix and J 0 , J 1 are
work.
Bessel functions of zeroth and first-order, respectively. The solu-
The radial displacements at the surface are
tion at infinity ( z 2 r 2 ) tends asymptotically to the point
force and point charge results. Equation 9 eliminates the A 3 ( )
function. The remaining unknown functions A 1 ( ), A 2 ( ) can be
found from the remaining surface electrical and mechanical
u r r,0
8a 3 M 9 M 8 M 10M 7
3 rD M 1 M 8 M 2 M 7
r2
1 1 2
a
3/2
;
macroscopic response is independent of c 12 .
2a 2 /D M 3 M 8 M 4 M 7 2 r2 a
r,0 2 2 arcsin
M 1 M 8 M 2 M 7 a r
3 Conducting Sphere
Equations 812 can be cast as a system of two dual integral
equations, for the interior problem (0ra) and for the exterior
r
a
1
a2
r2
; ra. (25b)
problem (ra). The solution follows from Sneddon 15 and is The constants M i are given in the Appendix. In order to have a
unique according to Walton 16. nontrivial solution, M 1 M 8 M 2 M 7 and M 5 M 8 M 6 M 7 , which
Using the continuity contact condition du z (a,0)/dr2a/D, are always satisfied for known piezoelectric ceramics.
the relation between the penetration depth h and the contact radius
a is h2a 2 /D. Other continuity conditions, e.g., zz (a,0)0 or 5 Mechanical and Dielectric Strength
D z (a,0)0, are possible but could lead to jumps in the slope of
When investigating the mechanical and dielectric strength of
the surface deformation at the contact perimeter and will not be
brittle materials with spherical indenters, it is often the case that
addressed further in this work.
the most critical region is at the surface, close to the contact
The surface vertical displacement outside the contact area (r
perimeter ra, z0. Therefore, it seems important to examine
a) is
the stresses and the electric flux in that region. From the boundary
u z r,0 h/ 2 r 2 /a 2 arcsin a/r r/a 1 a 2 /r 2 . conditions and the constitutive relations, it can be shown that at
(22) the surface (z0)
zz E z 0 for r0 , rr 0 for ra , (26)
The radial displacement at the surface inside the contact area
(0ra) is and therefore for z0, ra, Eq. 6 gives
rr r,0
r2
8 c 11c 12 M 9 M 4 M 10M 3 a 3
M 1 M 4 M 2 M 3 3D
with zero potential ( 0 0), Eqs. 2829 and 3435 can be
expressed in a more compact form as
P C c P/r 2 ra ,
(36)
M 6 M 1 M 2 M 5 0 a
, (28) where C c is a constant which depends on the material properties
M 1 M 4 M 2 M 3 c 11c 12 and the electrical conditions of the indenter. Equation 33 simpli-
whereas using Eq. 25b for the insulating, the surface radial stress fies to 19
outside the contact radius (ra) is found to be
rr r,0
8a c 11c 12 M 9 M 8 M 10M 7
3
. (29)
ln c P a 2 0
C c P/a 2
g c CcP
a 2
1 d. (37)
which is in the classic form proposed by Weibull 22, with n 1
2a 2 /D M 3 M 8 M 4 M 7 2 2r a r 2 a 2 being Weibull modulus.
E r 2 arcsin 2 ;
M 1 M 8 M 2 M 7 a r ar The previous analysis assumes zero residual surface stresses.
However, Pohanka et al. 23,24 showed that tensile internal
ra. (31) stresses due to polarization degrade the strength of BaTiO3 and
Therefore, for piezoelectric materials indented by a perfect insu- PZT-4 piezoelectric ceramics. Pohanka et al. found that machin-
lator, the maximum magnitude of electric flux is at the contact ing defects at the surface act as sources of failure. Their results are
perimeter in accord with the permittivity analysis of Buessem et al. 25
who found tensile internal stresses in constrained piezoelectric
4a/D M 3 M 8 M 4 M 7 grains that undergo cubic-to-tetragonal transformation. Pohanka
max E z2 E r2 E r a,0 . (32) et al. assumed a relation between the microcrack size c the ap-
M 1 M 8 M 2 M 7
plied stress and the internal stress R according to linear elastic
If the piezoelectric constants are zero, then M 3 0 and M 7 fracture mechanics
0 and the problem decouples to a mechanical indentation and to
a rigid dielectric problem of a transversely isotropic half space. R C A / c, (40)
5.1 Statistical Aspects of Mechanical Strength. The me- where C A is a constant that depends on the fracture energy G c ,
chanical strength of piezoelectric materials can be related to the the elastic modulus, c 33 and, weakly, on the crack configuration
density distribution of the surface microcracks. Spherical indenta- C A G c c 33. Pohanka et al. found that the internal tensile stress
tion has been used successfully to assess the distribution of mi- depends on the microcrack size c and the grain size d. Their
crocracks on glass surfaces e.g. Argon 19. Such approach can experimental results may be approximated in a simple way as
be justified in the case of piezoelectrics from fracture mechanics R R0 d/ cC R / c, (41)
analysis and experiments e.g., see 20 that indicate the me-
chanical strain energy release rate as the most suitable parameter where R0 is the internal stress for cd. Combining 40 and 41
for fracture criterion for non-conducting cracks this may not be
true for conducting cracks. Various surface stress combinations C A C R c, (42)
can be used in order to examine the statistical aspects of strength where C A C R , i.e., G c c 33 R0 d,
in the absence of spontane-
21. Since the radial stress, rr (r,0), is also the maximum ous cracking.
tensile principal stress at the surface, the simplest stress criterion The failure probability analysis may be equivalently reformu-
would involve alone. If g c () is the critical stress distribution lated in terms of a critical distribution of microcrack sizes at the
for semi-elliptical cracks per unit surface area, then the probability surface, using 42
of no fracture below load P in the entire surface outside the con-
tact circle a is g c c V 1 n 1 1 n 1 C A C R n 1 1 c 1n 1 /2 . . . . (43)
c P exp 2
a P
r dr
0
r, P
g c d . (33)
Assuming that the essential statistic parameters are the same for
the poled and unpoled specimens and that the first term in 43
dominates the problem, then the condition for similar probability
The relation between a and P is given by Giannakopoulos and of failure for both poled and unpoled cases gives the following
Suresh 18: relation between the critical indentation loads:
e P exp 2
0
r dr
E P
0
g e E dE . (45)
gularity of the problem. The contact radius was resolved with 24
elements. The outer boundary was at least 20a away from the
contact regime. With reference to Fig. 2, the outer boundary con-
The electric field intensity factor at the tips of the surface flaws ditions were vertically constrained along the sides CD and AB. In
is analogous to the stress intensity factor see, for example, 20. addition, sides CD and AB were given zero electrical potential. In
Thus, pre-existing surface flaws are expected to have a similar all cases, the uncoupled problem (e i j 0) was also solved. The
role in both mechanical failure and dielectric breakdown. Indeed, contact stresses were found to be in agreement within five percent
Yamashita et al. 28 and Kishimoto et al. 29 found similarity error when compared to analytic results.
between the shapes of mechanical and dielectric strength distribu- Four piezoelectric materials were analyzed, PZT-4, PZT-5A,
tions for piezoelectric ceramics of small grain size, in both ferro- BaTiO3, and 95 percent BaTiO3 5 percent CaTiO3. Their fully
electric and paraelectric state. These experimental results suggest poled, room temperature mechanical, dielectric, and piezoelectric
that properties are shown in Table 1. The data were collected from
g e E W 1 n 1 1 n 1 E n 1 1 . . . , (46) Jaffe et al. 4 and Bechmann 31. The indenting sphere was
taken to be rigid and approximated by an axisymmetric paraboloid
where the Weibull modulus n 1 is the same with that of g c (), Eq. of diameter D, which was kept the same in all cases. For the
38, and V 1 ( n 1 1 ) W 1 E ( n 1 1 ) . coupled cases, the sphere was taken either as a perfect conductor
with zero electric potential ( 0 0), or as a perfect insulator with
5.3 The Case of an Indenter With Nonzero Electric Poten- no surface electric charge distribution (D z 0). A constant aver-
tial. The case of the indenter being a perfect conductor currying age pressure P/( a 2 )33.84 GPa was used in all calculations
nonzero potential 0 is interesting for the reason that the normal the results for different average pressures scale according to
contact stress and the surface electric charge at the contact area Table 2.
(0ra) have square-root radial singularity, as also noted by The normalized applied load P/(D 1/2h 3/2), maximum principal
Chen and Ding 17 tensile stress max 1 , and the average electric charge distribution
8 M 4 M 5 M 3 M 6 0.1875Q(D/a 3 ) are tabulated in Table 2. These are results that are
zz r,0 a 2 r 2 useful for macroscopic observations of the mechanical strength
D M 1 M 4 M 2 M 3 due to indentation.
The contours of principal tensile stress, 1 , are shown in Fig.
2 M 6 M 1 M 2 M 5 0
, (47) 3a for the uncoupled case of PZT-4, in Fig. 3b for the coupled
M 1 M 4 M 2 M 3 a 2 r 2 case of PZT-4 with conducting sphere and in Fig. 3c for the
coupled case of PZT-4 with insulating sphere. The contours of
8 M 7 M 4 M 8 M 3 principal tensile stress are shown in Fig. 4a for the uncoupled
D z r,0
D M 1 M 4 M 2 M 3
2 M 1 M 8 M 2 M 7 0
a 2 r 2 . (48)
M 1 M 4 M 2 M 3 a 2 r 2
A compressive stress singularity is expected to create mi-
crodamage at the contact perimeter; the electric charge singularity
is expected to depole the material at the contact perimeter. These
effects attenuate with decreasing indenters diameter D. It is then
expected that due to the local damage, energy will be dissipated
and the probability of macroscopic indentation-induced cracking
be lowered. High positive electric potential would increase the
tensile radial stress at the contact perimeter, Eq. 28, and there-
fore increase the probability of macroscopic cracking. On the
other hand, high negative electric potential could result in lower
tensile stresses at the contact perimeter, Eq. 29, and reduce the
probability of macroscopic cracking.
The electric potential, however, cannot be arbitrarily negative.
The stability condition d P/dh0 for a0 and d P/dh0 if a
0 is satisfied if
0 D M 6 M 1 M 2 M 5 4a 2 M 4 M 5 M 3 M 6 0, (49)
Fig. 2 Overall view of the finite element mesh used in the
which also guarantees that the contact stresses are compressive present calculations; details of the mesh close to and away
everywhere, zz (r,0)0. High negative electric potential may from the contact area are included
Elastic
Stiffness
Coefficients
GPa PZT-4 PZT-5A BaTiO3 (Ba0.917Ca0.083TiO3
C 11 139.00 121.00 166.00 158.00
C 33 115.00 111.00 162.00 150.00
C 44 25.60 21.10 42.90 45.00
C 12 77.80 75.40 76.60 69.00
C 13 74.30 75.20 77.50 67.50
Piezoelectric PZT-4 PZT-5A BaTiO3 (Ba0.917Ca0.083TiO3
Coefficients
C/m2
e 31 5.200 5.400 4.400 3.100
e 33 15.10 15.80 18.60 13.50
e 15 12.70 12.30 11.60 10.90
Dielectric PZT-4 PZT-5A BaTiO3 (Ba0.917Ca0.083TiO3
Constants
(109 F/m)
11 6.461 8.107 11.151 8.850
33 5.620 7.346 12.567 8.054
denter. This indicates that either the contact perimeter or the con-
tact center are the most critical regions where contact induced
piezoelectric depoling may first occur. The maximum magnitude
of electric flux is higher for the conducting type of indenter.
It was found earlier that if the critical stress distribution of
microcracks per unit surface area, g c , remains the same for both
the unpoled and the poled material, then the probability of no
fracture under spherical indentation depends on the electric con-
tact conditions. The present analysis predicts that for PZT-4 the
probability of no failure at indentation load, P, is higher for the
unpoled material and lower for the poled material indented by the
same sphere. Therefore, indentation induced cracking is expected
for the poled PZT-4 material at lower load than the unpoled ma-
Fig. 4 Maximum tensile principal stress distribution for terial. In addition, the probability of no failure is slightly higher
spherical indentation of 95 percent BaTiO3 5 percent CaTiO3 ;
for the indentation of the poled material with an insulating in-
a uncoupled case P a 2 33.84 GPa, b coupled case,
with indenter being a perfect conductor of zero electric poten- denter than with a conducting indenter. Therefore, indentation in-
tial P a 2 33.84 GPa c coupled case, with indenter being duced cracking is expected for the poled PZT-4 material indented
a perfect insulator of zero surface electric charge P a 2 by a conducting indenter at lower load than indented by an insu-
33.84 GPa lating indenter of the same diameter.