You are on page 1of 8

Mat. Sci. & Eng. A288, 2000, 148-153.

RESIDUAL STRESS MEASUREMENT BY HERTZIAN INDENTATION


Y. Bisrat , S.G. Roberts Department of Materials, University of Oxford Parks road, Oxford, OX1 3PH, UK. Abstract We have used Hertzian indentation to measure surface stresses on ceramics. Surface stresses in brittle materials have the effect of shifting the minimum load to produce Hertzian fracture. Quantitative use of the method requires knowledge only of the materials Youngs modulus, Poissons ratio and its fracture toughness (or measurements from a stress- free specimen). We have used this method for surface stress measurements on Al2 O3 and soda lime float glass. Specimens of both types were stressed by bending. Our results are consistent with those obtained by other methods. The Hertzian technique thus provides a promising simple method for measuring surface residual stress in brittle materials. Keywords Hertzian indentation, surface residual stress, ring-on-ring bending, ceramics, glass, strength.

I. Introduction Indentation techniques have been widely used in the analysis and characterisation of fracture and deformation properties of brittle ceramics [1-3]. They provide a convenient means to study the effects of localised loading. Indentation fracture theory has given insight into a wide range of mechanical properties such as machining damage, erosion and wear. Indentation methods are gaining wide spread popularity as a simple and inexpensive technique for quantifying fracture toughness of brittle materials [1-5] as well as more conventional hardness measurements. Indentation testing is experimentally simple and widely used for quick material evaluation; the analysis may however, be complex. Indentation fracture theory concerns two basic type of contacts: sharp indenters - where the contact is essentially plastic up until fracture; and blunt indenters - where the contact is completely

Corresponding Author: Tel: +44 1865 273767, Fax: +44 1865 273764 E-mail: yordanos.bisrat@materials.ox.ac.uk

Mat. Sci. & Eng. A288, 2000, 148-153.

elastic up to the point of fracture. In this paper, we consider the use of blunt, spherical indenters Hertzian contact. The contact stress field for the Hertzian indentation is purely elastic and though complex, it is welldefined [6,7]. There have been several attempts to use Hertzian indentation to determine the fracture toughness of brittle materials [8-15]. The main principle depends on the interaction of the elastic stress field with a pre-existing surface flaw. The reasons why the Hertzian test has not been as popular as those using sharp indenters (e.g. Vickers indentation) may be due to: (i) the variability of the results obtained so far; (ii) the steep stress gradient which has made it difficult to obtain accurate estimates for stress- intensity factors for cracks driven by Hertzian loading; and (iii) the results of the analysis being very sensitive to the value of the Poissons ratio of the substrate [16]. There is, however, one significant advantage of Hertzian indentation over pointed indenters the substrates deformation is entirely elastic until fracture occurs. This avoids the complications arising from the somewhat ill defined indentation residual stress always associated with pointed indenters. The material properties that may be determined by this test include (a) fracture toughness of the nearsurface material, (b) the densities and sizes of surface cracks, and (c) residual stresses in the nearsurface material [15,16]. We give below a brief introduction to the studies of fracture toughness and residual stress using Hertzian indentation.

II. Hertzian contact When a hard sphere (radius R and elastic constants E1 , 1 ) is pressed with a normal load P on to a flat substrate (elastic constants E2 , 2 ), the contact radius is given by [16]
3RP a = * 4E
1 3

(1)

where

2 1 1 1 1 2 2 = + E* E1 E2

(2)

A complex but well defined stress field is set up around the contact site; the peak pressure under the sphere P0 is given by
P0 = 3P 2 a 2

(3)

The principal component of this stress field that is responsible for crack propagation is the radial stress. This component is tensile close to the contact surface and decays away from the contact centre; it also decreases rapidly with depth and finally becomes compressive [4,8,12]. Thus, calculating the stress intensity factors for surface flaws is not straightforward. Warren [16] has used the Nowell and Hills [17] approach of distributed dislocations to derive expressions for mode I stress intensity factor for surface flaws in a Hertzian stress field. This approach led to a new technique for the determination of fracture toughness (see section III) and is the basis for the residual stress measurement, described in section IV.

Mat. Sci. & Eng. A288, 2000, 148-153.

III. Fracture toughness KIC determination by Hertzain Indentation Warren [16] has shown that for every load, there is one size and position of flaw that gives maximum stress intensity factor Kmax of all possible flaws in the surface. Kmax increases monotonically with applied indentation load. Ring cracks can form only if this Kmax equals the fracture toughness KIC of the material. Hence, there is a minimum load Pmin below which no fracture occurs. The principle of the method for determining KIC is thus, to find this Pmin . This is done by performing a series of Hertzian tests, noting the fracture loads. It is experimentally found that there is a definite minimum load to fracture Pmin (see section VI). The fracture toughness KIC can then be calculated as [16]
E *Pmin = CR
1 2

K IC

(4)

where C is a dimensionless constant that depends on the Poissons ratio of both the indenter ball and test piece. The values of C for various values of can be found on Table 1 [16] for the case of elastically similar test piece and test ball. Reliable results for KIC have been reported for glass and alumina [16]. The advantages of this technique are that it requires minimum specimen preparation, and there is no need to measure crack size; it relies mainly on performing enough tests, typically 2030 in total, to ensure the tests find one flaw close to critical flaw size and position, giving reliable value for Pmin . IV. Residual stress measurement Brittle materials, such as glass and ceramics, have low fracture toughness and tend to fail from defects that exist on the surface due to manufacturing process or handling. One of the popular practices of reinforcing them is to introduce compressive stresses at the surface. This can be done using a variety of techniques: ion-exchange or thermal tempering process in soda- lime glass [18-22]; phase transformation in partially stabilised ZrO 2 [23] and by the formatio n of low thermal expansion solid solution surface layers on Al2 O3 and glass ceramics [23]. Also, surface grinding has been shown to introduce a compressive layer on some ceramic structural components [24-29] which in some cases can more than compensate for the damage introduced, and give high strengths. Several techniques have been employed to quantitatively measure surface stress and in some instances to study the stress profile with depth in brittle materials: X-ray diffraction [30-31] in polycrystalline ceramics; deflection methods [31-33]; birefringence methods for transparent materials [34-36]; strain gauge techniques [37]; and acoustic waves to qualitatively measure the temper of heattreated glass [38]. Vickers indentation has been used as a simple microprobe method to measure the stress in tempered glass surfaces [39]. Although this technique can determine absolute surface stress to considerably better than a factor of 2, its advantage of experimental simplicity is offset by the uncertainties associated with the determination of some test parameters [31,39] particularly the residual stress field around the indentation. Recently, Roberts et al. [40] have described a method based on Hertzian indentation. A short summary of the technique, an extension of the use of Hertzian indentation to determine KIC, is given below.

Mat. Sci. & Eng. A288, 2000, 148-153.

If a specimen is tested using Hertzian indentation after a compressive stress is introduced on the surface, the minimum load to produce Hertzian fracture (see section III) shifts to a higher value. This leads to an apparent increase in fracture toughness. If the residual stress is assumed to be uniform over the depths of the surface cracks, a rough estimation of the stress can be given as [40]
0

K IC S K IC = 1.12 (c)

(5)

where c is the flaw size giving the minimum load for a reference, stress- free specimen with fracture toughness 0 KIC, and SKIC is the apparent fracture toughness for a stressed specimen, calculated from minimum fracture load via Eqn. 4.

V. Experimental technique To experimentally validate the technique described in section IV, we used a ring-on-ring bending jig to induce a known compressive stress on the surface of the test materials. During loading the inner portion of the specimen is bent into a spherical curvature whereby the top surface is under compression and the bottom surface is under tension. This bending stress, which is uniform and biaxial, is measured using strain gauges mounted on the specimen surfaces. For each test, four strain gauges were mountedtwo on each surfaceon a float glass disc with diameter 65 mm and thickness 2 mm. The specimen was then mounted on top of a ring of steel ball bearings on the lower part of the jig. Another set of ball bearings on the upper surface acted as the other loading ring. Figure 1 shows the schematic diagram of the experimental set up. Hertzian indentations were performed on the centre of the compressive surface to evaluate this method for calculating the stresses. A CK10 testing machine (Engineering Systems, Nottingham, UK) was used to carry out the indentation (see Figure 2). This machine uses an acoustic emission sensor to detect the formation of a ring crack, which allows the technique to be applicable to opaque materials. In the present study, alumina discs, supplied by Morgan Matroc Ltd., UK., and glass discs supplied by Pilkingtons Ltd., UK., were used. 5 mm diameter alumina balls were used as indenters for the alumina discs whereas 2 mm diameter steel balls were used for glass discs. It was not possible to use glass spheres for indentation as they broke before forming a ring crack on the disc. Every sample was tested under two conditions unstressed and stressed. Alumina discs were tested as received, as it is assumed the surface finish would provide ample flaws. The glass discs were abraded with 1200 SiC grit. Not more than 25 indentations were possible on any specimen as the test area of the stressed specimens, which rests on a middle support, is only 1 cm in diameter and because indentations should not be closer than four or five crack diameters, typically 200 m, to prevent one indentation affecting another.

Mat. Sci. & Eng. A288, 2000, 148-153.

P
Inner loading ring Acoustic sensor
Acoustic sensor

Strain gauges Outer loading ring Support tower


Indenter shaft Bending jig

Fig. 1. Schematic diagram of the experimental set up.

Fig. 2. Experiment in progress.

VI. Results and Discussion For every series of 25 indentations, the fracture loads were sorted in ascending order so that probability to failure could be assigned to each fracture load. The probability to failure versus fracture load of the glass specimen is shown in Fig. 3. There is a clear minimum fracture load, Pmin and a progressive increment in Pmin with increasing compressive stress. The steep gradient of the probability curve represents the initiation of ring cracks over a very narrow range of loads close to the minimum load, Pmin . We believe this is because the mild abrasion provided a large flaw density near the ideal size for fracture at Pmin [41]. The surface stresses as read from the strain gauges, and as calculated from the shift in Pmin are shown in Table 1. Probability to failure

1.2 1 0.8 0.6 0.4 0.2 0 0 50 100 150 200


43 MPa 0 MPa 21 MPa

Load (N)
Fig. 3. Probability to failure versus load plot : Pmin is 39N, 56N and 75N at stress levels equal to 0 MPa, 21 MPa and 43 MPa respectively .

The value of the stress calculated from Hertzian test varies from that obtained from strain gauges. This was initially thought to be due to elastic mismatch between indenter and test specimen (as glass balls that are strong enough were not available, a steel indenter was used for the glass specimen). The analysis described in sections III and IV, however, assumes that both the test piece and indenter are elastically similar and consequently the dimensionless constant C in Eqn. 4 is dependent only on Poissons ratio. Johnson et al. [42] noted that elastic mismatch between contacting bodies gives rise to the presence of shearing traction and hence modification of the contact stress fields. This leads to C being affected not only by but also the degree of elastic mismatch. Warren et al. [43] have given a quantitative assessment of the effect of elastic mismatch on both the contact field and the stress

Mat. Sci. & Eng. A288, 2000, 148-153.

intensity factor experienced by a surface crack, giving a method for calculation of the modified dimensionless constant C. Table 1 shows the fracture toughness and stress values calculated using a modified value of C, taking the lowest possible value of coefficient of friction suggested by Johnson et al. [42]. The stress values are in better agreement but the fracture toughness (0.6 MPam1/2 ) is lower than what is quoted in most of the literature (0.8 MPam1/2 ). Figures 4 and 5 show the probability to failure versus fracture load plots for alumina discs tested with 5mm diameter alumina balls. The probability curves show some scatter, which is probably due to the rough as-received 20 m finish. The shift in the minimum load to fracture with increased compressive stress is, however, obvious. The calculated values for the residual stress are shown in Table 2. The stress values obtained are in fairly good agreement with what is indicated by the strain gauges. However, the values are 50% lower than the strain gauge reading. Elastic mismatch can not be the cause of the variation as both indenter and specimen are alumina. It is possible that mechanical play in the central support under the indentation site might also influence the results; there are fluctuations in the strain gauge readings during testing which indicate this might be the case. This is under current investigation.

Probability of fracture

0.8 0.6 0.4 0.2 0 150 200 250 300 350

Probability of fracture

0.8 0.6 0.4 0 MPa 0.2 0 150 200 250 300 350 132 MPa

0 MPa

83 MPa

Load (N)

Load (N)

Fig. 4. Probability to failure vs load of Alumina A. The minimum load to fracture Pmin is 196 N when unstressed and 235 N when stressed to 83 MPa.

Fig. 5. Probability to failure vs load of Alumina B. The minimum load to fracture P min is 207 N when stressed and 265 N when stressed to 132 MPa.

Mat. Sci. & Eng. A288, 2000, 148-153.

Table 1. Residual stress results for mildly abraded glass specimen. Stress from Strain Gauge (MPa) 0 25+1 43+1
*

Pmin

KIC

(N) 39+2 56+8 75+6

(MPa m1/2 ) 0.91+0.01 1.09+0.03 1.27+0.01

* Stress from KIC with Hertzian corrected C test (MPa) (MPa m1/2 ) 0.66+0.01

Stress from Hertzian With corrected C for Elastically dissimilar (MPa) 35+1 66+1

45+1.4 90+1.2

0.80+0.03 0.92+0.01

KIC is equal to real 0 KIC for stress free case or apparent SKIC for stressed specimen.

Table 2. Residual stress results for as-received alumina specimens.

Unstressed Specimen Pmin (N) KIC (MPa m1/2 )


0

Stressed Pmin (N) KIC (MPa m1/2 )


S

Residual stress (MPa) From Hertzian Tests 47+1.8 68+1.2 From Strain Gauge 83+8

A B

196+11 207+6

2.36+0.07 2.43+0.04

235+11 265+2

2.59+0.06 2.76+0.01

132+6

VII. Conclusion We have used the Hertzian indentation technique to measure residual stress on ceramics and glass specimens. Although it needs further refining, the Hertzian method has a potential for an easy, straightforward measurement of near surface residual stress in brittle materials to a good first approximation [27,40]. This would give a useful technique for both quality and process control of engineering ceramic components.

VIII. References [1] B.R. Lawn, J.Am.Ceram. Soc. 81 (1998) 1977. [2] B.R. Lawn, T.R. Wilshaw, J.Mater.Sci. 10 (1975) 1049. [3] B.R. Lawn, Fracture of Brittle Solids, Cambridge University Press, 1993, Ch. 8. [4] P. Ostojic, R.McPherson, Int. J. Fract. 33 (1987) 297.

Mat. Sci. & Eng. A288, 2000, 148-153.

[5] R.F. Cook, G.M. Pharr, J. Am. Ceram. Soc.73 (1990) 787. [6] M.T. Huber, Ann. Phys. 14 (1904) 153. [7] W. B. Morton, L. J. Close, Phil. Mag. 43 (1922) 320. [8] F. C. Frank, B. R. Lawn, Proc. R. Soc. Lond. A 229 (1967) 291. [9] B.D. Powell, D. Tabor, J. Phys. D : Appl. Phys. 3 (1970) 783. [10] Hj. Matzke, V. Meyritz, J. L. Routbourt, J. Am. Ceram. Soc. 66 (1983) 183. [11] M. Laugier, J. Mater. Sci. 19 (1984) 254. [12] Idem, J. Mater. Sci. Lett. 4 (1985) 1542. [13] K. Zeng, K. Breder, D. J. Rawcliffe, Acta Metall. 40 (1992) 2601. [14] R. Warren, Acta Metall. 26 (1978) 1759. [15] T. R. Wilshaw, J. Phys. D: Appl. Phys. 4 (1971) 1567. [16] P. D. Warren, J. Eur. Ceram. Soc. 15 (1995) 201. [17] D. Nowell, D. A. Hills, J. Strain Anal. 22 (1987) 177. [18] S. S. Kistler, J. Am. Ceram. Soc. 45 (1962) 59. [19] J. S. Olcott, Sci. 140 (1963) 1189. [20] O. S. Narayanswanny, R. Gardon, J. Am. Ceram. Soc. 52 (1969) 554. [21] R. Gardon, J. Non-Cryst. Solids 73 (1985) 233. [22] M. J. C. Hill, I. W. Donald. Glass Tech. 30 (1989) 123. [23] H. P. Krichner, R. M. Gruver, R. E. Walker, J. Am. Ceram. Soc. 51 (1968) 251. [24] R. F. Cook, B. R. Lawn, T. P. Dabbs, P. Chantikul, ibid. 64 (1981) c121. [25] D. B. Marshall, A. G. Evans, B. T. Khuri Yakub, J. W. Tien, G. S. Kino, Proc. R. Soc. Lond. A385 (1983) 461. [26] I. A. Chou, H. M. Chan, M. P. Harmer, J. AM. Cerm. Soc. 79 (1996) 2403. [27] H. Z. Wu, C. W. Lawrence, S. G. Roberts, B. Derby, Acta Mater. 46 (1998) 3839. [28] F. F. Lange, M. R. James, D. J. Green, J. Am. Ceram. Soc. 66 (1983) C16. [29] B. R. Lawn, A. G. Evans, D. B. Marshall, J. Am. Ceram. Soc. 63 (1980) 574. [30] M. R. James, J. B. Cohen, in: H. Hermon (Ed)., Treatise on Materials Science and Technology, Academic Press, New York, 1980, pp. 1-62. [31] S. Chandrasekar, M. M. Chandhri, Phil. Mag. A 67 (1993) 1187. [32] D. J. Green, F. F. Lange, M. R. James, J. Am. Ceram. Soc. 66 (1983) 623. [33] S. Chandrasekar, M. C. Shaw, T. N. Farris, B. B. Bhushan, Adv. Inform. Storage Sys. 1 (1991) 353. [34] A. J. Monack, E. E. Beeton, I:Glass Ind. 20 (1939) 127. [35] Idem., II: Ibid. 5 (1939) 185. [36] Idem., III: Ibid. 6 (1939) 223. [37] I. L. Resnick, R. E. Mould, J. Soc. Glass Tech. 35 (1951) 487T. [38] K. M. Jassby, A. Aharoni, Am. Soc. For Non-destructive Testing (1981) 502. [39] D. B. Marshall, B. R. Lawn, J. Am. Ceram. Soc. 60 (1977) 86. [40] S. G. Roberts, C. W. Lawrence, Y. Bisrat, P. D. Warren, D. A. Hills, J. Am Cearm. Soc. 82 (1999) 1809. [41] S. G. Roberts, Proc. Brit. Ceram. Soc. 59 (1999) 45. [42] K. L. Johnson, J. J. OConnor, A. C. Woodward, Proc. Roy. Soc. A334 (1973) 95. [43] P. D. Warren, D. A. Hills, J. Mater. Sci. 29 (1994) 2860.

You might also like