Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

January 2001

Materials Letters 47 Ž2001. 140–144


www.elsevier.comrlocatermatlet

Load dependence of low-load Knoop hardness in ceramics:


a modified PSR model
Jianghong Gong ) , Zhenduo Guan
Department of Materials Science and Engineering, Tsinghua UniÕersity, Beijing 100084, People’s Republic of China
Received 5 June 2000; accepted 19 June 2000

Abstract

Utilizing experimental data of the Knoop hardness versus applied load for several ceramics, the applicability of the
proportional specimen resistance ŽPSR. model proposed by Li and Bradt to describing the indentation size effect ŽISE.
observed in a relatively wide range of applied load was examined. The PSR model was proven to be insufficient for
representing the experimental data and was modified based on the consideration of the effect of machining-induced
residually stressed surface on the hardness measurements. q 2001 Elsevier Science B.V. All rights reserved.

Keywords: Indentation; Hardness; Knoop; Residual stress; Ceramics; Proportional specimen resistance

1. Introduction With the increase in the load, the measured hardness


decreases. Such a phenomenon is generally referred
The hardness characterization of advanced ceram- to as indentation size effect ŽISE.. Clearly, a com-
ics is extremely important because that many of the plete characterization of the hardness property for a
advantages of the advanced ceramics for use at near given material needs the knowledge of the quantita-
ambient temperatures are associated with their rela- tive description of the ISE.
tively higher hardness compared to metals. On the Recently, Li and Bradt w3x developed a propor-
other hand, however, it is usually difficult to make tional specimen resistance ŽPSR. model to describe
any quantitative comparisons between the published the ISE observed in the Knoop hardness testing. In
hardnesses for a given material which were mea- this model, the applied test load, P, and the resultant
sured by different workers in different load ranges, indentation dimension, d, are predicted to follow the
for the conventional hardness measurements are usu- relationship
ally load dependent w1,2x. When a very low load is
applied to a ceramic, the measured hardness is high. P s a1 d q a2 d 2 . Ž 1.
The observed ISE is considered to be a conse-
quence of the indentation-size proportional resistance
of the test specimen as described by a1-term and the
)
Corresponding author. Fax: q86-10-62771160. a 2-term can be related directly to the load-indepen-
E-mail address: gong@tsinghua.edu.cn ŽJ. Gong.. dent hardness.

00167-577Xr01r$ - see front matter q 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 5 7 7 X Ž 0 0 . 0 0 2 2 5 - 1
J. Gong, Z. Guanr Materials Letters 47 (2001) 140–144 141

From a viewpoint of practice applications, an ranging from 1.47 to 45.57 N using a commercial
equation used for describing the ISE should give an low-load hardness tester. From the start of the con-
accurate fit of the experimental data over a wide trolled release of the indenter onto the material sur-
range of applied test load. When examining the face to the development of a well-defined indenta-
works of Li and Bradt w3x more closely, however, tion impression on the same surface took about 30 s.
one can find that only a relatively narrow range of The length of the long diagonal, d, of each indenta-
applied load was considered. For example, in the tion was measured immediately after unloading. For
original paper w3x in which the PSR model was indentations made with an applied load of 45.57 N,
proposed, the basic conclusions were deduced only the length of the short diagonal, dX , was also mea-
based on analyzing four data points measured in the sured. All the indentation tests were carried under
load ranging from 50 to 300 g. Therefore, it seems to ambient laboratory conditions.
be necessary to make a further study on the applica-
bility of the existing PSR model to describing the
ISE observed in a wide range of applied load. This
forms the underlying basis of this communication, 3. Data analyses
which addressed the ISE in some typical ceramics
and reported measurements of the Knoop hardness.
The Knoop hardness, H, is defined as the ratio of
The observed ISE was further analyzed using the
the applied load, P, to the projected area of indenta-
PSR model. It was shown that the existing PSR
tion, A, i.e.,
model is not suitable for representing the observed
ISE and a modification of this model was proposed. P P
Hs s 14.229 . Ž 2.
A d2

Fig. 1 illustrates the measured Knoop hardness


2. Experimental
profiles as a function of the applied load for each
material tested. Each data point represents an aver-
Eight different materials, which are listed in Table age of measurements from ten tests. For the sake of
1, were selected for the present study. All the speci- conciseness, all the error bars corresponding to the
mens were received in slab form with machined flat, data points are omitted. The maximum scatter, typi-
parallel surfaces. Specimens were mounted in resin cally 5–6% of the hardness value, usually occurs at
and polished with successively finer diamond pastes the lowest load employed and reduces as the applied
until a mirror-like surface was achieved. Knoop load increases. Clearly, for each material, there is a
hardness measurements were conducted at test load significant ISE in the measured Knoop hardness.

Table 1
Regression analysis results of the experimental data according to Eq. Ž6.
Material Sample P0 ŽN. a1 a2
denotation ŽNrmm. ŽNrmm2 .
Hot-pressed silicon nitride FD-02 y1.317 58.67 783.46
Hot-pressed silicon nitride FD-03 y1.740 70.65 759.92
SiC whisker reinforced Si 3 N4 SN-W y1.163 37.53 769.36
Sintered Al 2 O 3 Al 2 O 3 y0.989 32.57 891.34
Tetragonal zirconia polycrystal TZP y1.197 30.83 580.84
Normal sintered mullite Mullite y0.011 5.10 625.58
TiŽC,N.-based cermet TCN1 y0.645 40.60 730.11
TiŽC,N.-based cermet TCN2 y0.904 48.82 711.30
142 J. Gong, Z. Guanr Materials Letters 47 (2001) 140–144

Based on the analysis of a vast amount of re-


¨
ported experimental data, Buckle w4x has identified
three regions of hardness testing conditions: Amicro-
hardnessB, Alow-load hardnessB and Anormal hard-
¨
nessB. According to Buckle’s description, only the
Alow-loadB condition was definitely associated with
the ISE. In the AmicrohardnessB: region, experimen-
tal errors related to the smallness of the indentation
are usually significant and make it difficult to con-
duct repeatable measurements. In the Anormal hard-
nessB region, on the other hand, a nearly load-inde-
pendent hardness value is obtained. Therefore, a
complete description of the ISE needs only a set of
experimental data measured in the whole Alow-loadB
Fig. 2. Pr d plotted against d for samples FD-02 and TCN1.
range. As can be seen in Fig. 1, for each material, a
nearly constant hardness was observed on the
higher-load side, implying that the hardness-load
curves measured in the present study are sufficient that, for each material, all the data points fall into
for the complete description of the ISE. two separate sets. Using the experimental data mea-
Application of the PSR model proposed by Li and sured in the lower load region, a good linear relation-
Bradt to the experimental results shown in Fig. 1 can ship between Prd and d was obtained for each
be accomplished by plotting Prd versus d which material. Extrapolating the Prd y d straight line
yields a straight line with a slope equal to a 2-value fitted in the lower load region to the higher load
and an intercept equal to a1-calue. Fig. 2 presents the region, however, evident deviations appeared. Simi-
Prd y d plots for FD-02 and TCN1. It is evident lar phenomenon can also be observed by analyzing
the experimental data for other materials investi-
gated. These analyses indicate that the existing PSR
model does not provide a satisfactory description for
the observed ISE in ceramics and warrants a modifi-
cation.

4. Modification of the PSR model

Li and Bradt w3x analysis started with the


HaysrKendall approach. When examining the ISE
in the Knoop hardness testing of some metals, Hays
and Kendall w5x assumed that there exists a minimum
level of the applied load, W, below which permanent
deformation due to indentation does not initiate, but
only elastic deformation occurs. They introduced an
effective applied load, Peff Žs P y W ., and pro-
posed the following relationship between the applied
load and the resultant indentation size
P y W s a2 d 2 . Ž 3.
Fig. 1. Knoop hardness as functions of the applied test load for
materials tested. ŽB. FD-02; ŽI. FD-03; Žv . SN-W; Ž`. By analyzing the experimental data measured with
Al 2 O 3 ;Ž'. TZP; mullite; Ž%. TCN1; Ž\. TCN2. some single crystals, Li and Bradt w3x found that the
J. Gong, Z. Guanr Materials Letters 47 (2001) 140–144 143

resultant W-value is too large to have a physical ings of the parameters a1 and a 2 in Eq. Ž6. are the
meaning. Thus, they suggested that W is not a same as those in Eq. Ž1..
constant but increases with the indentation size and By the way, it should be pointed out that both
is directly proportional to it, i.e. Eqs. Ž6. and Ž1. are of the same form that has been
¨
applied by Buckle w8x when utilizing a polynomial
W s a1 d. Ž 4. series representation of the applied load to the ISE.
Eq. Ž6. differs in form from Eq. Ž2. only in the
To a first approximation, Li and Bradt considered P0-term.
Eq. Ž4. to be similar to the elastic resistance of a Fig. 3 illustrates P versus d for all the eight
spring with the opposite sign to the applied test load. materials tested in the present study. The solid lines
Substituting Eq. Ž4. into Eq. Ž3. yields Eq. Ž1., the in these plots are obtained by a conventional polyno-
relationship between the applied load and the resul- mial regression according to Eq. Ž6.. It is evident
tant indentation size, which was predicted in the PSR that Eq. Ž6. is proven sufficiently suitable for the
model. representation of the experimental data in each case.
It should be pointed out that, when using the PSR The best-fit values of the parameters included in Eq.
model to explain the ISE in brittle ceramics, an
important factor should be taken into account. To
obtain a flat, parallel and mirror-like surface suitable
for the hardness measurements, machining and pol-
ishing are usually unavoidable processes for the
specimen preparation. As mentioned above, all the
specimens used in the present study were also re-
ceived with machined and polished surfaces. The
surface machining and polishing processes, which
remove material mechanically, may introduce resid-
ual stresses on the surface of brittle materials. Stud-
ies on ceramic grinding have shown w6,7x that the
residually stressed layer is compression near the
surface and tensile underneath. Although a quantita-
tive analysis relating the effect of the existence of
such a residually stressed surface on the hardness
measurements is still lacking, there is reason to
believe that, if it is true that the resistance, W, of a
specimen with a residually stressed surface can be
simulated as a spring, such a Aspring’ must have
been in a state of compression or tensile, rather than
stress-free, before being subjected to indentation. If
this were the case, Eq. Ž4. would be corrected as

W s P0 q a1 d, Ž 5.
where P0 is related to the residual surface stresses in
the test specimen.Substituting Eq. Ž5.into Eq. Ž3.
yields

P s P0 q a1 d q a2 d 2 . Ž 6.
Clearly, Eq. Ž6. may be regarded as a modified Fig. 3. Indentation size versus the applied test load for test
form of the existing PSR model. The physical mean- materials.
144 J. Gong, Z. Guanr Materials Letters 47 (2001) 140–144

ulus ratio. For each material tested, the dXrd ratio


was measured for Knoop indentations made with an
applied load of 45.57 N and the results are shown in
Fig. 4 as a function of the a2ra1-value. It is evident
that there exists a strong correlation between dXrd
and a1ra2 . This seems to be a sound support to the
validity of the modified PSR model established
above.

Acknowledgements

The authors would like to thank Mr. Jianjun Wu


X
and Mr. Zhe Zhao for their assistance in carrying on
Fig. 4. Variation of d r d with the a1 r a2 ratio. the experimental work.

Ž6. for each material are summarized in Table 1. The


relatively small, negative values of P0 seem to be References
reasonable estimations of the magnitudes of the w1x D.J. Clinton, R. Morrell, Mater. Chem. Phys. 17 Ž1987. 461.
residual surface stresses in the test specimens, which w2x J.B. Quinn, G.D. Quinn, J. Mater. Sci. 32 Ž1997. 4331.
had been subjected to a careful polishing after ma- w3x H. Li, R.C. Bradt, J. Mater. Sci. 28 Ž1993. 917.
chining. w4x ¨
H. Buckle, in: J.H. Westbrook, H. Conrad ŽEds.., The Science
Following the analysis of Li and Bradt w3x, the of Hardness Testing and Its Research Application, Metal Park,
American society for Metals, OH, 1973, p. 453.
parameter a2 in Eq. Ž6. is a measure of the true w5x C. Hays, E.G. Kendall, Matall. 6 Ž1973. 275.
hardness of the test materials while, if the friction w6x D. Johnson-Walls, A.G. Evans, D.B. Marshall, M.R. James, J.
effect between the indenter facets and the test speci- Am. Ceram. Soc. 69 Ž1986. 44.
men may be ignored, the parameter a1 can be related w7x R. Samuel, S. Chandrasekar, T.N. Farris, R.H. Licht, J. Am.
to the elastic modulus. On the other hand, the ratio Ceram. Soc. 72 Ž1989. 1960.
w8 x ¨
H. Buckle, ¨ ¨
Mikroharteprufung, Berliner Union Verlag,
of the short diagonal length, dX , to the long diagonal Stuttgart, 1965.
length, d, of a Knoop indentation has been suggested w9x D.B. Marshall, T. Noma, A.G. Evans, J. Am. Ceram. Soc. 65
w9x to be a measure of the hardness-to-elastic-mod- Ž1982. C-175.

You might also like