Available online at www.sciencedirect.

com

CERAMICS
INTERNATIONAL
Ceramics International 40 (2014) 7661–7667
www.elsevier.com/locate/ceramint

Weibull distribution of the fracture strength of 99% alumina ceramic

reshaped by cold isostatic pressing
Yan Fu, Zhendong Tao, Xianqin Houn
Key Laboratory of Inorganic Functional Materials in Shandong Province, School of Material Science and Engineering, University of Jinan, Jinan 250022, China

Received 16 September 2013; received in revised form 5 December 2013; accepted 13 December 2013
Available online 21 December 2013

Abstract

The mechanical reliabilities of 99% alumina ceramics preformed by dry pressing using various pressures and reshaped by cold isostatic
pressing (CIP) at 150 MPa are investigated using the maximum likelihood method for the estimation of the Weibull parameters of the results. The
results indicate that among the various values of dry pressing and sintering temperatures applied, a maximum Weibull modulus of 13.9974 and
high reliability were attained for 75 MPa dry pressing followed by subsequent 150 MPa cold isostatic pressing (CIP) and a 1580 1C sintering
temperature. In addition, excessive preforming pressure caused dangerous flaws, called hardening defects, which were not repaired automatically
by the CIP and sintering process, thus acting as a source of fracture and reducing the reliability of the specimen strength.
& 2014 Published by Elsevier Ltd and Techna Group S.r.l.

Keywords: Weibull distribution; Cold isostatic pressing; 99% alumina ceramic; Fracture strength; Maximum likelihood estimation

1. Introduction a certain stress value of the material can be withstood before

failure occurs for a given probability of failure and confidence
Griffith strength theory accounts for the fracture of brittle level. Weibull was the first to use a statistical technique to
materials that is usually initiated from tiny cracks and flaws that describe the scatter of fracture, and his fundamental assumption
are distributed in the material in a disorderly and unsystematic was the weakest link hypothesis (the specimen fails if its
manner [1]. Fracture strength is one of the vital mechanical weakest volume element fails). In recent years, a significant
properties of structural ceramics [2]. The strength of materials number of researchers have used the Weibull distribution as a
depends on the size of the major flaw varying from one to statistical tool to evaluate the performance of ceramics. Because
another, which leads to a discrete distribution of the fracture the location parameter for brittle materials is usually observed to
strength. In this framework, an intimate correlation between the be zero [3], the two-parameter Weibull distribution is used
flaw size and the scatter of the data of the strength is realized. The for the present calculations [3]. Thus, the value of the shape
fracture of ceramics originates from the presence of defects in the parameter characterizes the spread of the fracture strength.
microstructure that are mostly caused by the production process Seidel et al. [4] studied the effect of processing on the reliability
of the material. Normally, the forming method, forming pressure of alumina ceramics in 1996, demonstrating that the change
and sintering temperature are the sources of defects (large grains, in the forming method used in his study was of no help in
pores and cracks). Therefore, it is important to develop a increasing the Weibull modulus, and the reason for the increase
production process to obtain ceramics with good mechanical in strength was the elimination of soft agglomerates. Nevarez-
properties and stable performances. Rascon et al. [5] used the Weibull distribution to evaluate the
The reliability of alumina ceramics is mainly determined by compressive strength, hardness and fracture toughness of
their fracture strength and Weibull modulus, which are greatly ZTA and ATZ nanocomposites reinforced by Al2O3 whiskers.
affected by the forming process. Reliability means ensuring that Danzer [6] studied the correlation between fracture and defect
statistics and discussed the Weibull distribution of very small
n
Corresponding author. Tel.: þ 86 13705403795; fax: þ 86 0531 82767385. specimens. Studies demonstrated that the defect density predicted
E-mail addresses: [email protected], [email protected] (X. Hou). by the Weibull distribution was higher and that some interaction

0272-8842/$ - see front matter & 2014 Published by Elsevier Ltd and Techna Group S.r.l.
http://dx.doi.org/10.1016/j.ceramint.2013.12.053
7662 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667

among the defects should occur for a very small amount of where s is the fracture strength, F is the applied force at the
specimens. Fan et al. [7] discussed the porosity dependence of the fracture point, s is the span between supports, b is the width of the
Weibull modulus for hydroxyapatite and other brittle materials. In specimen, and h is the thickness of the specimen.
their study, a total of 540 hydroxyapatite specimens were sintered The Weibull distribution [9–16] has been widely used to
and fractured to extensively study the Weibull modulus. Warren analyze the scatter of fracture strength. Weibull parameters are
[8] discussed the effect of the test method and threshold stress on linked with the reliability of specimens. The Weibull distribu-
the Weibull modulus of fracture of brittle materials. He demon- tion in this study is determined for four groups.
strated that the Weibull modulus determined using different testing In terms of the Weibull distribution, the Weibull modulus,
techniques for fracture strength would be the same if there was no m, is of great importance, where m can be defined in the two-
threshold stress for fracture; in contrast, if this threshold stress did parameter Weibull distribution:
exist, the test methods would affect the value of the Weibull
FðxÞ ¼ P ¼ 1  exp½  ðs=sc Þm : ð2Þ
modulus even though specimens were formed using the same
method. where P is the cumulative probability of failure, m is the
In this paper, the cumulative distribution function, a two- Weibull modulus or the Weibull shape parameter, s is the
parameter Weibull distribution, is used to assess the mechan- fracture strength, and sc is the characteristic strength or scale
ical reliabilities of 99% alumina ceramics preformed by dry parameter.
pressing with various pressures and reshaped by cold isostatic
pressing (CIP) at 150 MPa. With the analysis of the Weibull
parameters, the forming parameter of a 99% alumina ceramic 3. Results
with high reliability is investigated.
As observed in Fig. 1, the XRD patterns of the raw material (the
lattice constants are a¼ 5.028 Å, b¼ 5.028 Å, and c¼ 13.730 Å)
2. Experimental procedures
and the specimens sintered at 1500 1C and 1580 1C (the lattice
constants are a¼ 4.758 Å, b¼ 4.758 Å, and c¼ 12.991 Å) display
2.1. Specimen preparation
distinct diffraction peaks. The primary phase can be identified as
α-Al2O3, which has a hexagonal structure. The peaks at 2θ¼
Through spray granulation, pure (99%) alumina powder (Zhe-
43.351, 35.121, 57.491, and 25.571 [17] correspond to the
jiang Ultrafine Powder & Chemicals Co., Ltd.) with an average
characteristic diffraction peaks of α-Al2O3. No peaks of another
size of 100 μm was preformed into rectangular bars with dimen-
phase can be detected. When the sintering temperature reaches
sions of 45  8  6 mm3 via dry pressing. The pressures used for
1500 1C, the distinct diffraction peaks are the same as those of the
the dry pressing were 50 MPa (group A), 75 MPa (group B) and
specimen sintered at 1580 1C. However, both sets of peaks are
100 MPa (group C). The bar-shaped compacts were then reshaped
shifted to the right compared with the X-ray diffraction pattern of
by CIP at 150 MPa. To compare these samples with the specimens
the raw material, which might be caused by the diminution of the
reshaped by CIP, control experiments were established. Specimens
lattice constants. In addition, the orientation of grain growth might
were formed by dry pressing only at 100 MPa (group D). The
lead to the shift of the diffraction peaks. Moreover, a small quantity
number of specimens in each group was 100. Specimens of each
of fluxing agent in the raw material might form a solid solution,
group were divided into five equal parts and sintered in an electric
which could also cause the diffraction peaks to shift. Furthermore,
furnace at 1500 1C, 1520 1C, 1540 1C, 1560 1C and 1580 1C. The
specimens were then polished and numbered for various tests to
determine their firing shrinkage, bulk density and fracture strength.

2.2. Specimen characterization and Weibull analysis

X-ray diffraction (XRD, Bruker AXS, D-76181 Karlsruhe,

Germany) was used to analyze the crystalline phase of the raw
materials and fired specimens; a scanning electron microscopy
system (SEM, FEI, QUANTA FEG 250, America) was used to
examine the microstructural features of the fractured speci-
mens. The volume density of the specimens was measured
using the Archimedes method. The fracture strength of the
specimens was tested using the three-point bending test with a
support distance of 30 mm on a universal testing machine
(MTS, CMT5504, China).
All the specimens were bent to fracture with the loading center
perpendicular to the widest surface. The fracture strength was
calculated using the following equation: Fig. 1. X-ray diffraction patterns of material powder (a¼ 5.028 Å, b¼5.028 Å,
c¼13.730 Å) and specimens sintered at 1500 1C and 1580 1C (a¼4.758 Å,
s ¼ 3Fs=ð2bh2 Þ: ð1Þ b¼ 4.758 Å, c¼12.991 Å).
 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667 7663

Fig. 2. Fracture strength for (a) group A, (b) group B, (c) group C and (d) group D.

undeveloped grain growth in the specimens sintered at 1500 1C

causes the intensities of the peaks to decrease. xi=[σ1 σ2 …σ20];
Fig. 2 shows the fracture strength for the four groups of
specimens sintered at five different temperatures. As observed
in Fig. 2, the fracture strength appears to be affected by the Plati=wblfit(xi)
preforming pressure. The data of group B show the most stable
state when the sintering temperature is 1580 1C.
In each group, a two-parameter Weibull distribution was applied Plati=A(σ0) B(m)
to analyze the fracture strength data for each validly fractured
specimen. A total of 400 specimens yielded a valid fracture. For
simplicity and accuracy, the maximum likelihood estimation was [M,V]=weibstat(A,B)
used to estimate the Weibull modulus and scale parameter [18].
MATLAB was used to write a program to estimate the Weibull
parameters. The detailed commands are shown in Fig. 3. M=expectation,V=variance
The Weibull modulus, scale parameter, expectation and
variance are determined after entering the values of the fracture Fig. 3. The detailed commands for the maximum likelihood estimation in
strength of each group. The detailed data are presented in MATLAB.
Table 1.
As observed in Table 1, the Weibull modulus appears to be when the preforming pressure is relatively lower. In addition, the
affected by the pressure of dry pressing. Ignoring some of the Weibull moduli of specimens sintered before 1580 1C differ
individual data points, the Weibull moduli of the specimens irregularly, which might be caused by the incomplete densification
reshaped by CIP in this experiment are relatively high, except of specimens, grain growth and other reasons. In terms of (1) and
for that of group C, indicating that one type of defect dominates the (2), it is important to note that the values of the Weibull moduli
fracture process in almost all of the specimens preformed by CIP are relatively higher except when the dry pressing pressure was
7664 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667

Table 1
Weibull modulus, s0, expectation M, and variance V for (a) group A, (b) group
B, (c) group C and (d) group D.

T/1C s0/MPa m M V

(a)
1580 225.7056 8.8553 0.5131 0.0048
1560 227.5228 12.3756 0.6188 0.0037
1540 220.4052 12.4386 0.6219 0.0037
1520 217.7454 10.7151 0.5773 0.0042
1500 194.7364 11.6710 0.6094 0.0040

(b)
1580 224.6861 13.9974 0.6544 0.0033
1560 209.7745 11.7054 0.6064 0.0039
1540 213.0256 9.0371 0.5233 0.0048
1520 216.9435 12.6666 0.6279 0.0036
1500 185.7625 7.8912 0.4854 0.0053

(c)
1580 284.2463 4.0112 0.2217 0.0038
Fig. 4. SEM image of the raw material.
1560 309.0777 6.7623 0.3999 0.0048
1540 245.0394 6.8940 0.4208 0.0051
1520 212.7108 5.3475 0.3383 0.0053
1500 193.1959 8.8155 0.5207 0.0050 porosity of the specimens, well-grown grains and a uniform grain
(d)
size. All of these factors lead to a higher reliability of specimens
1580 287.4833 6.6126 0.3962 0.0049 preformed by dry pressing (75 MPa) and reshaped by CIP
1560 239.9311 5.6315 0.3493 0.0051 (150 MPa) with a high Weibull modulus compared with
1540 213.5262 9.6625 0.5453 0.0046 Fig. 4a (incomplete growth of the grain and grain boundary)
1520 198.0753 6.5439 0.4154 0.0055 and Fig. 5c (oversintering of the grain and the hardening defect).
1500 194.6201 7.1417 0.4476 0.0055
Comparing Fig. 5c with Fig. 5d, oversintering of the grain and
the hardening defect of Fig. 5c lead to a lower Weibull modulus
in the specimens preformed by dry pressing (100 MPa) and
75 MPa combined with CIP at a sintering temperature of 1500 1C. reshaped by CIP than in the specimens formed by dry pressing
It is obvious that the value of the Weibull modulus is relatively (100 MPa) only, even though it has a higher bulk density and
high, up to 13.9974 (the preforming pressure was 75 MPa, the CIP firing shrinkage (Fig. 6, Fig. 7).
pressure was 150 MPa, and the sintering temperature was Fig. 6 shows the firing shrinkage [19] for the 4 groups of
1580 1C), indicating that the fracture strength values are centra- specimens. The specimens preformed by dry pressing and
lized, which is apparent in Fig. 2. The Weibull modulus of (4) is reshaped by CIP exhibit a higher value of firing shrinkage than
relatively lower than that of (1) and (2). The Weibull modulus those formed by dry pressing only except when the preforming
provides paramount information about the scatter of the fracture pressure is 100 MPa. In this case, the shrinkage is lower than
strength, but it cannot dominate the reliability of ceramics alone. in the specimens formed by dry pressing only at 1500 1C.
To determine the reliability of the specimens or the synergistic In addition, the firing shrinkage for specimens formed at
action of the shape parameter and scale parameter, the expectation different pressures increases with increasing sintering tempera-
and the variance should be investigated. ture. A high sintering temperature (below the melting point)
As observed in Table 1, when the preforming pressure is contributes to molecular motion and sintering.
75 MPa, the subsequent CIP pressure is 150 MPa and the Fig. 7 shows the bulk density for the 4 groups of specimens.
sintering temperature is 1580 1C, the reliability is the highest As observed in Fig. 7, the specimens preformed by dry pressing
(the expectation is the highest, the variance is the lowest and and reshaped by CIP exhibit a higher bulk density than those
the shape parameter is the highest). formed by dry pressing only; however, when the preforming
Fig. 4 shows the microstructure of the raw material with an pressure is lower and the sintering temperature is 1580 1C, the
average size of 100 μm. Fig. 5 shows the fracture surfaces for value of the bulk density is higher. In addition, the bulk density of
specimens with the same value of fracture strength formed using the specimens sintered below 1580 1C does not correspond to the
different methods and sintered at 1580 1C. In Fig. 5d, crack-like shrinkage. This result may occur because the bulk density was
voids entrapped in grains are present, and intergranular fracture is measured using the Archimedes method. The density is thus
dominant. In addition, the size of the grains is not uniform, which measured with apparent porosity, ignoring the closed pores that
indicates that specimens formed by dry pressing only with the might be caused by hardening defects and crack-like voids.
pressure of 100 MPa have a non-homogeneous microstructure, In addition, the contact angle between the grain (undeveloped
leading to a decrease in the Weibull modulus and an enlargement growth) and the pore is too small to overcome the energy of grain
in the range of the fracture strength. In Fig. 5b, an appropriate boundary migration (removed with grain boundary movement).
preforming pressure (75 MPa) yields better densification, less The expansion of pores at the high temperature will lead to the
 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667 7665

Fig. 5. SEM images under the same magnification (5 μm) of fracture surface for specimens sintered at 1580 1C. (a) formed by dry pressing only at 100 MPa,
(b) preformed by dry pressing at 50 MPa and preshaped by CIP at 150 MPa, (c) preformed by dry pressing at 75 MPa and preshaped by CIP at 150 MPa and
(d) preformed by dry pressing at 100 MPa and preshaped by CIP at 150 MPa.

Fig. 6. Firing shrinkage for 4 groups of specimens sintered at five different Fig. 7. Bulk density for 4 groups of specimens sintered at five different
temperatures. temperatures.

4. Discussion
reduction of shrinkage. All of these factors cause the Weibull
modulus of specimens sintered below 1580 1C to differ irregu- The results reveal that both crack-like voids and inherent
larly. In addition, the excessive pressure of preforming increases hardening (Fig. 8) defects exist in the specimens reshaped by
the rate of densification, leading to the bulk density of the CIP with an excessive preforming pressure (100 MPa), or dry
specimens preformed by dry pressing with pressures of 50 MPa pressing, such that subsequent CIP with the reshaped pressure up
and 75 MPa to be lower at 1500 1C. With increasing temperature, to 150 MPa could not restore the specimens. The defects cause
the shrinkage increases until full sintering. the Weibull modulus and reliability of group C to be lower than
7666 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667

Hardening defect
Excessive pressure
of preforming
Raw materials
(Spherical-like
particles)
Sintering process

Fig. 8. Forming process of hardening defect.

those formed by dry pressing only. To some extent, CIP cannot In addition, an appropriate preforming pressure (75 MPa) (dry
restore the inherent hardening defect when the pressure of dry pressing) promoted an increase of the Weibull modulus of
pressing is oversized. CIP makes the defect die-hard such that specimens up to 13.9974 and improved the reliability of this
grain growth during the sintering process with densification of the modulus; the expectation is 0.6544, and the variance is 0.0033.
specimens is of no benefit. Furthermore, excessive preforming An excessive preforming pressure (100 MPa) causes hardening
pressure causes a point contact to irreversibly transform into line defects, and subsequent CIP cannot restore the dangerous
contacts and face contacts, leading to a reduction in the mobility defect that results in a low Weibull modulus and low reliability
of the powder. The energy of the sintering process might of the specimens.
decrease, and the hardening defect [20] cannot be restored
automatically. In addition, most of the soft defects produced by
the preforming process can be restored automatically during the References
CIP process, and sintering contributes to increasing the Weibull
[1] Q. Yang, L.G. Tham, G. Swoboda, Relationship between refined Griffith
modulus when the preforming pressure is lower. criterion and power laws for cracking, Mechanics Research Communica-
For the firing shrinkage, at the range of low temperature, tions 31 (2004) 429–434.
densification caused by pressure has a stronger effect on the [2] N. Shinohara, M. Okumiya, T. Hotta, et al., Variation of the micro-
shrinkage of specimens than the temperature. When the structure and fracture strength of cold isostatically pressed alumina
ceramics with alteration of dewaxing procedures, J. Eur. Ceram. Soc.
sintering temperature is 1580 1C, the effect of densification
20 (2000) 843–849.
caused by pressure on shrinkage and sintering is crucial. [3] Z.Q. Meng, Z.Y. Guo, J.H. Guo, Weibull distribution of flexural strength
In addition, the hardening defect produced by dry pressing with of silica ceramics, Rare Metal Materials and Enqineering 38 (2009)
excessive pressure is difficult to eradicate in the low temperature 1178–1180.
phase. A lower preforming pressure increases the number of [4] J. Seidel, N. Claussen, J. Rödel, Reliability of Alumina Ceramics. 2:
Effect of Processing, J. Eur. Ceram. Soc. 17 (1997) 727–733.
pores between grains or large agglomerates that can be crushed
[5] A. Nevarez-Rascon, A. Aguilar-Elguezabal, E. Orrantia, et al., Compres-
by CIP. CIP promotes the densification of compacts. A high sive strength, hardness and fracture toughness of Al2O3 whiskers
preforming pressure combined with CIP causes poor mobility of reinforced ZTA and ATZ nanocomposites: Weibull analysis, Int. Journal
the powder. A close connection between powders is helpful for of Refractory and Metals and Hard Materials 29 (2011) 333–340.
the densification of a green body and is beneficial to increase the [6] R. Danzer, Some notes on the correlation between fracture and defect
statistics: Are Weibull statistics valid for very small specimens?, J Eur.
degree of firing shrinkage. A hardening defect produced by an
Ceram. Soc. 26 (2006) 3043–3049.
excessive preforming pressure is difficult to eliminate; thus, [7] X. Fan, E.D. Case, F. Ren, Y. Shu, M.J. Baumann, Part І: Porosity
with increasing preforming pressure, the degree of shrinkage dependence of the Weibull modulus for hydroxyapatite and other brittle
decreases, which is why specimens preformed by dry pressing materials, J. of the mechanical behavior of biomedical material 8 (2012)
(100 MPa) and reshaped by CIP (150 MPa) have lower Weibull 21–36.
[8] P.D. Warren, Fracture of brittle materials: effects of test method and
moduli and reliabilities.
threshold stress on the Weibull modulus, J. Eur. Ceram. Soc. 21 (2001)
Thus, an appropriate dry pressing pressure and sintering 335–342.
temperature are crucial to the shrinkage, bulk density, Weibull [9] A.D. Papargyris, Estimator type and population size for estimating the
modulus [21] and reliability when the specimens undergo a Weibull modulus in ceramics, J. Eur. Ceram. Soc. 18 (1998) 451–455.
subsequent CIP process [22,23]. [10] K. Franken, H.R. Maier, Weibull analysis of soldered MLC under
bending load stress, J. Eur. Ceram. Soc. 19 (1999) 1307–1310.
[11] S.L. Fok, B.C. Mitchell, J. Smart, B.J. Marsden, A numerical study on the
5. Conclusions application of the Weibull theory to brittle materials, Engineering
Fracture Mechanics. 68 (2001) 1171–1179.
The dry pressing pressure has considerable effects on the [12] D. Hudak, M. Tiryakioglu, On comparing the shape parameters of two
microstructures and reliabilities of 99% alumina ceramics Weibull distributions, Materials Science and Engineering A 528 (2011)
8028–8030.
reshaped by CIP. An appropriate dry pressing pressure [13] D. Stoyan, C. Funke, Stefan Rasche, Back to and beyond Weibull-The
(75 MPa) and an appropriate sintering temperature (1580 1C) hazard function approach, Computational Materials Science 68 (2013)
promote densification and grain growth of the specimens. 181–188.
 Y. Fu et al. / Ceramics International 40 (2014) 7661–7667 7667

[14] B. Defez, G. Peris-Fajarnés, et al., Influence of the load application rate [19] K. Mori, M. Ohashi, K. Osakada, Simulation of microscopic shrinkage
and the statistical model for brittle failure on the bending strength of behavior in sintering of powder compact, Int. J. Mech. Sci. 40 (1998)
extruded ceramics tiles, Ceramics Internatinal 39 (2013) 3329–3335. 989–999.
[15] X. Fan, E.D. Case, I. Gheorghita, M.J. Baumann, Weibull modulus and [20] I. Yu, Arch effect in high isostatic pressure compacts, J. Eur. Ceram. Soc.
fracture strength of highly porous hydroxyapatite, Journal of the 19 (1999) 2619–2623.
Mechanical Behavior of Biomedical Materials (2013). [21] M.A. Madjoubi, C. Bousbaa, M. Hamidouche, N. Bouaouadja, Weibull
[16] A. Saghafi, A.R. Mirhabibi, G.H. Yari, Improved linear regression statistical analysis of the mechanical strength of a glass eroded by sand
method for estimating Weibull parameters, Theoretical and Applied blasting, J. Eur. Ceram. Soc. 19 (1999) 2957–2962.
Fracture Mechanics. 52 (2009) 180–182. [22] I. Yu. Prokhorov, G. Ya, Akimov. Cold isostatic pressing as a method of
[17] L.D. Shang, Preparation and indentification, Chemical World (1994) pre-forming green ceramic ware, J. Eur. Ceram. Soc. 17 (1997) 129–131.
346–350. [23] W. He, Q.S. Wei, K. Liu, Y.S. Shi, J. Liu, Numerical simulation of cold
[18] X.Y. Li. Research on estimation for the three-parameter Weibull isostatic pressed alumina parts produced by selective laser sintering and
Distribution. Master Degree of Beijing Jiaotong University. part shape optimization, Ceramics International. (2013).