Strength of Materials, Vol. 46, No.
1, January, 2014
INVESTIGATION ON THE MICROCRACKING OF CERAMIC MATERIALS
USING THE ACOUSTIC EMISSION METHOD
A. V. Drozdov UDC 539.3/4:666.3
This paper presents the results of investigations on the microcracking of composite ceramic
materials, such as electrical engineering porcelain and silicon nitride ceramics, using the acoustic
emission technique. The analytical expression describing the kinetics of microcrack accumulation is
refined. The influence of the loading rate on the strength characteristics of ceramics and the special
features of acoustic emission during their cracking are studied. The dependence of the total acoustic
emission count on the stressed surface area of specimens is found. The notion of the microcracking
modulus of ceramic composite materials is introduced and its mismatching with the Weibull
modulus in strength tests of ceramics is shown.
Keywords: ceramic materials, microcracking, acoustic emission, total count, Weibull distribution, microcracking
modulus.
Introduction. The acoustic emission (AE) technique is useful for studying the processes of fracture in
ceramic materials both at the macro and micro levels [1, 2]. This is due to the fact that the above processes exhibiting
the breaking of the bonds in the structure of brittle materials are accompanied by the release of elastic energy which
can be recorded by highly sensitive equipment. For ceramic materials under loading, the microcracking and
subcritical crack growth (SCG) are the main fracture processes and, accordingly, the AE sources.
The stage of SCG preceding catastrophic fracture of many ceramic materials can be detected by AE signals.
It was found that for polycrystalline ceramic materials the total AE count, N , is directly proportional to the growing
crack length increment during the time of observance [2], while the count rate N& , in its turn, is proportional to the
subcritical crack growth rate [1, 3]. It is usual practice to describe the dependence of the crack growth rate V on the
stress intensity factor K1 by an exponential equation [1] of the form:
n
da æ K ö
V= = A * çç 1 ÷÷ = A (K1 ) n , (1)
dt è K1c ø
where a is the crack length, t is the time, A * , A, and n are the material-, temperature-, and environment-
dependent parameters, and K1c is the critical stress intensity factor.
The stress intensity factor K1 is related to the stress acting in the material s and the crack length a by the
following relationship:
K1 = Ys a , (2)
where Y is the geometric factor.
As a result of numerous experimental investigations [1–6], it was found that for the majority of ceramic
materials at room temperature, the parameter n (the crack growth rate index) is within 30 and 300, whereas its lower
Pisarenko Institute of Problems of Strength, National Academy of Sciences of Ukraine, Kiev, Ukraine
([email protected]). Translated from Problemy Prochnosti, No. 1, pp. 89 – 100, January – February, 2014.
Original article submitted May 22, 2013.
0039–2316/14/4601–0071 © 2014 Springer Science+Business Media New York 71
values ranging from 12 to 16 are typical of glass. The analysis of expressions (1) and (2) testifies that the subcritical
crack growth at a constant or increasing load is an avalanche-like process observed at a relatively short final portion
of loading and terminated by catastrophic fracture.
The other typical source of AE under mechanical loading of ceramics is the formation of non-interacting
cracks dispersed in the bulk or on the surface of the material. This kind of microcracking takes place in
heterogeneous ceramic materials and occurs due to the anisotropy in the elastic properties of structural components
of the material and the difference in their thermal expansion coefficients, which leads to the occurrence of additional
residual stresses [7]. The microcracking can also be initiated by surface cracks that are formed during the mechanical
treatment of ceramic materials [1]. A positive role of the microcracking lies in the fact that this is one of the possible
mechanisms to increase the fracture toughness of ceramic materials [7, 8]. Thus, in the development of ceramic
materials, the goal can be to produce controlled cracks which would increase the fracture toughness values while
maintaining the high strength [8]. Therefore, the knowledge of the regularities underlying the microcracking process
in ceramic materials is important both in the study of special features of their mechanical behavior and fracture at the
micro- and macrolevels, and in the development of new materials with an enhanced fracture toughness.
In [9–11], the approach was developed to describe the microcracking process during loading of specimens of
polycrystalline alumina and electrical engineering porcelain (which is a composite ceramic material containing
quartz particles in the glass matrix). The foundation of this approach is the Weibull statistical distribution that is
based on the “weakest-link” concept. According to this distribution, the probability of fracture P due to the stress
value s is defined by expression [12]:
é æ s - s öm ù
P = 1 - exp ê- çç n
÷÷ ú , (3)
ê è s0 ø ú
ë û
where s n is a stress below which the probability of fracture is equal to zero, s 0 is the characteristic stress, or the
normalizing parameter, and m is the Weibull modulus, or the homogeneity coefficient.
Expression (3) is known as the three-parameter Weibull distribution, which is transformed into the
two-parameter distribution for s n = 0:
m
é æ s öù
P = 1 - exp ê- çç ÷÷ú . (4)
ë ès0 øû
Since the empirical function of the fracture probability is equally matched with the distributions of (3) and
(4) for a limited number of tested specimens, then, in order to assess the strength of ceramic materials, a more simple
two-parameter Weibull distribution (4) is most often used.
The basis of the analysis of the process of microcracking in ceramic materials [9–11] is the notion that the
material consists of elementary volumes, and the occurrence of a microcrack in each of the volumes is governed by
the two-parameter Weibull distribution. For a low density of non-interacting microcracks formed, i.e., at low values
of the fracture probability P, it is possible to make use of the following approximate expression [7]:
m
h æ s ö
= çç ÷÷ , (5)
c +1 è s 0 ø
where h is the number of microcracks being formed and c is the total number of potentially possible microcracks.
Since the microcracking of composite ceramic materials is accompanied by a discrete AE, the events thus
recorded, or the value of the total count N , are proportional to the number of microcracks being formed and
therefore the stress s versus total AE count N curves can also be described by a power dependence similar to (5).
In [9, 11], special features of AE are analyzed for the electrical engineering porcelain under loading. As a result, it
was found that the power exponent m calculated from the AE parameters (m = 4) agrees with the results of
determining the Weibull modulus during the strength tests of specimens. Similarly, during testing of aluminum oxide
72
Fig. 1. Typical s s u versus total AE count N curves under loading of ceramic materials.
(Lucalox) it was found that m =10, and this, as is stated in [10], corresponds to the Weibull modulus for
polycrystalline ceramic materials. Unfortunately, the statistical strength characteristics of the studied materials are
not directly presented in [10] and, therefore, the assertions about the correspondence between the power exponent m
during their microcracking and the Weibull moduli probably need more justification. It is noteworthy that in [9–11]
special features of microcracking were analyzed from the results of recording the AE count rate, which is the
derivative of the total AE count, N , having a more pronounced, as compared to the latter, discrete character, and is
more dependent on the parameters of the recording equipment, which makes difficult the analysis and processing of
experimental data.
The goal of the present study is to refine the regularities underlying the microcracking of heterogeneous
polycrystalline ceramic materials during their mechanical loading by using the AE method.
Selection of Materials for Investigations. To select specific ceramic materials exhibiting microcracking, the
strength tests of different ceramic materials in bending were preliminarily performed with AE recording. For this
purpose, special equipment was developed that ensures three- and four-point bending of ceramic specimens and the
recording of AE parameters. The total count which is an integral parameter reflecting the kinetics of the processes in
both micro- and macrofracture of ceramic materials under their loading was taken as an informative AE parameter.
For comparison purposes, the relative stress s s u versus total AE count N curves (s u is the ultimate
strength in bending) were determined for each of the materials under study. As a result of the performed tests, three
typical curves were obtained under loading of ceramic materials (Fig. 1).
For curve 1 (Fig. 1), the low AE activity or its practical absence is observed up to the very instant of
fracture, separate jumps in the total count take place, which can be due to the material fracture at the points of load
application to the specimen or the friction between the loading elements and the specimen surface. The use of thin
polyvinylchloride or teflon pads between the loading rollers and the specimen surface makes it possible to eliminate
or considerably reduce this phenomenon. The described type of the AE curve is typical of high-strength ceramic
materials (the ultimate strength is 200 MPa or higher), single crystals or amorphous materials such as glass.
The low AE activity at the initial portion of loading and noticeable increase in the total count N is observed
shortly before the specimen fracture at a constant load or a slight stress relaxation, which are terminated by its
catastrophic fracture (a sharp drop in the load) – curve 2 in Fig. 1. This testifies to the absence of a noticeable
microcracking at the initial stage of loading and the presence of the subcritical crack growth stage in these materials.
This type of curve was observed in tests of ceramic materials with a relatively large grain size.
Curve 3 in Fig. 1 exhibits monotonic increase in the total AE count (and the count rate), which is observed
during loading of specimens up to the instant of their catastrophic fracture. The stress versus total AE count N
curves are intrinsic for the electrical engineering porcelain, silicon nitride material such as SNCCM – silicon nitride
composite ceramic material – [14] and some other composite ceramic materials, which is caused by the process of
their gradual microcracking [9].
73
In view of the above, the materials that exhibited microcracking accompanied by AE were chosen as objects
for further investigations (curve 3 in Fig. 1), namely: the electrical engineering porcelain consisting of a glassy phase
with particles of quartz or mullite uniformly distributed therein, and the three-phase silicon nitride ceramics
containing grains of silicon nitride and magnesium oxide. It is noteworthy that both materials exhibited linear
stress–strain curves.
Special Features of Experimental Data Processing. To investigate the special features of microcracking,
the strength tests of specimens of 3.5 ´ 5.0 ´ 50 mm in size were carried out under conditions of four-point bending
with a load span of 40/20 mm. The tests were performed using the equipment that include a precision loading device
and the system for recording the AE parameters [13]. To eliminate extraneous acoustic noise between the loading
rollers and the specimen, the pads of thin polyvinyl chloride tape were placed.
Since AE signals were practically absent during the initial portion of loading, and the character of the
recorded curves was smooth, with no noticeable knees and kinks up to the very instant of fracture, an attempt was
undertaken to approximate the dependence of the total count N on the acting stresses s with a more general
expression resulting from the three-parameter Weibull distribution:
ì0 at s £ s A ,
N =í m (6)
î B (s - s A ) at s > s A ,
where  and m are the parameters of microcracking, and s A is the stress corresponding to the appearance of AE
signals.
It was assumed a priori that the index of microcracking m (referred to as the microcracking modulus) differs
from the Weibull modulus m determined from the dispersion of the values of the ultimate strength of the specimens.
The modulus m was determined from the experimental data for each specimen with the use of the least square
method by transforming expression (6) to the form:
ln( N ) = m ln(s - s A ) + ln(B ). (7)
Since fixation of the value of stress s A is problematic, and this procedure has a quite biased character, the
microcracking modulus m was determined by the successive search of various possible values of stress s Ai .
At that the sum of squares of the deviations S i of the calculated data N cj from the experimental ones N ej
served as the measure of agreement of the proposed power function (6) for the fixed value of s Ai :
k
Si = å ( N cj - N ej ) 2 , (8)
j =1
where k is number of points for approximation.
For convenience of representation and comparison of the calculation results obtained for each value of s Ai ,
the reduced value of the sum of squares of the deviations of the calculated and experimental data S i was
determined by the formula
Si
Si = , (9)
S max
where S max is the maximum value of the sum of squares of the deviations for the considered range of variation in
the stresses s A .
The results of calculating the dependence of S i on s A for specimens of electrical engineering porcelain
and silicon nitride ceramics tested in bending testify to its clearly defined minimum (Fig. 2). With a variation of s A
from 0 to 20 MPa for the specimen of electrical engineering porcelain, the reduced value of the sum of squares of the
74
Fig. 2. Dependence of S i on s A for the specimen of electrical engineering porcelain
(the loading rate is 0.75 MPa/s).
deviations was decreased by a factor of about 1000. The microcracking modulus m and the stress s A were
determined for the minimum value of the above dependence.
This behavior can be due to the fact that the value of the total count N up to the instant of fracture is of the
order of 10 5 , while the number of AE events reflecting the number of the recorded microcracks is from 1. 5 × 10 4 to
2. 0 × 10 4 . This is a sufficiently large number and therefore the nonuniformity of the microcracking process under the
material loading is not noticeable. Similar S i versus s A dependences were also obtained for specimens of silicon
nitride ceramics. It is noteworthy that 90–95% of the recorded experimental stress s versus total AE count N
curves obtained in testing specimens of both of the materials had a clearly defined minimum of the calculated S i
versus s A dependences. The test data of separate specimens, for which small jumps in the total count N were
observed during the initial portion of loading, did not undergo any processing, which is probably caused by poor
mounting of the specimen on the bearing supports or a possible presence of foreign particles in the contact areas
between the loading rollers and the specimen.
As a result of the performed calculations, it has been found that the microcracking modulus m for both of
the specimens of electrical engineering porcelain and those of silicon nitride ceramics is within the range from 2.4 to
3.7 with an average value of 3.0. It is important that the values of m and s A determined from the minimum of the
S i versus s A dependence are practically independent of the number of experimental points (5–50) used for the
calculation. The value of the microcracking parameter B in expression (6) depends on the AE equipment sensitivity
and the quality of the acoustic contact between the specific specimen and the receiving AE transducer. For the
electrical engineering porcelain, the scatter in the values of B is a factor of two, whereas for the silicon nitride
ceramics, it is a factor of four times.
Influence of the Loading Rate on the Strength Characteristics and Microcracking Parameters. The
special features of AE accompanying the microcracking of electrical engineering porcelain and silicon nitride
ceramics were investigated in four-point bending strength tests of specimens of 3.5 ´ 5.0 ´ 50 mm in size at different
loading rates. To obtain the representative data, the number of the tested specimens for each of the loading rates was
no less than ten. The testing results and microcracking parameters of the above materials are summarized in Table 1.
The ultimate strength distribution curves for the chosen materials at different rates of loading are given in Fig. 3. The
Weibull modulus was determined by the least square method from the experimental data presented in the
ln(ln(1 (1 - P (s )))) - ln(s ) coordinates. As is seen, the variation of the loading rate has a strong influence on the
mean strength value of electrical engineering porcelain, while retaining a small scatter in the strength of separate
specimens and high values of the Weibull modulus m = 31 to 36 (Fig. 3a). For the silicon nitride ceramics, an
insignificant influence of the loading rate on its ultimate strength is observed with a relatively large scatter in the
obtained data and lower values of m (15–21) – see Fig. 3b.
75
TABLE 1. Results of Strength Tests for Specimens of Electrical Engineering Porcelain and
Silicon Nitride Ceramics under Four-Point Bending
Material Loading rate Number Ultimate strength m sA , m
s& , MPa/s of specimens su , MPa MPa
Electrical 0.04 10 73.1 32 14.0 3.1
engineering 0.75 22 79.3 31 17.5 3.0
porcelain 12.00 12 85.6 36 21.0 3.0
Silicon nitride 0.15 33 172.0 21 27.5 2.9
ceramics 1.80 32 176.0 16 37.0 3.0
50.00 33 181.0 15 47.0 3.0
a b
Fig. 3. Ultimate strength distributions of electrical engineering porcelain (a) and silicon nitride ceramics (b)
at different loading rates: (1) s& = 0.04 MPa/s, m = 32; (2) s& = 0.75 MPa/s, m = 31; (3) s& = 12 MPa/s,
m = 36; (4) s& = 0.15 MPa/s; (5) s& = 1.8 MPa/s; (6) s& = 50 MPa/s.
a b
Fig. 4. Averaged stress s versus total count N curves at different loading rates for the electrical
engineering porcelain (a) and the silicon nitride ceramics (b): (¢) s& = 12 MPa/s; (p) s& = 0.75 MPa/s;
() s& = 0.04 MPa/s; (r) s& = 0.15 MPa/s; () s& = 1.8 MPa/s; (£) s& = 12 MPa/s.
76
a b
Fig. 5. Averaged stress s versus total count N curves for specimens of silicon nitride ceramics
under four- (a) and three-point (b) bending.
The stress s versus total AE count N curves for both the specimens of electrical engineering porcelain and
those of silicon nitride ceramics are similar but, as was noted above, they exhibit significant scatter in the absolute
values of the total count N at the given stress level. Using these curves, the microcracking parameters m and s A
were determined for each of the specimens from the minimum condition for the reduced value of the sum of
deviation squares S i . The mean values of these parameters at different loading rates for each of the specimen groups
are given in Table 1.
By analyzing the calculated data, it is possible to conclude that the mean values of the microcracking
modulus m for the electrical engineering porcelain and the silicon nitride ceramics practically coincide and are, in
fact, independent of the loading rate (m = 2.9 to 3.1). At the same time, an increase in the loading rate results in the
increase in the value of s A from 14 to 21 MPa for the electrical engineering porcelain and from 27.5 to 41.0 MPa
for the silicon nitride ceramics. The averaged stress s versus total AE count N curves for each of the specimen
groups are presented in Fig. 4. It is significant that for both of the materials, the curves are satisfactorily described,
even at the lowest loading rates, by an exponential dependence (6) and contain no portion of the increase in N at a
constant stress prior to the fracture. This can be indicative of the absence of any noticeable subcritical macrocrack
growth for these materials.
Investigation of the Influence of Specimen Size and Loading Mode on the Special Features of
Microcracking. It is of interest what distribution (either volumetric or surface) of dispersed microcracks is observed
under loading of ceramic composite materials. When it is considered that the surface of materials can interact with
the environment, which results in the activation of the breaking of some bonds, the microcracking process can be
predominantly a surface one. A possible formation of surface cracks resulting from the mechanical treatment of
specimens can also serve as a source of the occurrence of microcracks under their loading.
In order to identify the volumetric or surface character of microcracking, specimens of silicon nitride
ceramics of 3.5 ´ 5.0 ´ 50 and 5.0 ´ 5.0 ´ 100 mm in size were tested under four-point bending at comparable loading
rates. For specimens of 3.5 ´ 5.0 ´ 50 mm in size, the distance between the inner and outer loading rollers was 20 and
40 mm, respectively, the loading rate was 1.8 MPa/s, whereas for specimens of 5.0 ´ 5.0 ´ 100 mm in size, it was 40
and 80 mm, respectively, and loading rate was 1.5 MPa/s. The ratios of the effective stressed volumes and squares of
the surfaces for the specimens of the above dimensions are, respectively, 0.35 and 0.5. The averaged stress s versus
total AE count N curves are shown in Fig. 5a. The values of the total count for two dimension types of specimens at
the same stress are within 0.45 and 0.55, which corresponds to the ratio of the stressed surface areas of the
specimens.
Additionally, two groups of specimens of 3.5 ´ 5.0 ´ 50 mm in size were tested under three-point bending.
The specimens were placed differently on supports (Fig. 5b). The ratios of the effective stressed volumes to the
77
TABLE 2. Strength Test Results for the Silicon Nitride Ceramics under Different Loading Schemes
Loading Specimen Number Ultimate m s A *, m*
scheme dimension, mm of specimens strength su , MPa MPa
Four-point 5.0 ´ 3.5 ´ 50 32 176 16 37.0 3.0
bending 5.0 ´ 5.0 ´ 90 12 158 17 47.0 3.1
Three-point 3.5 ´ 5.0 ´ 50 23 213 15 – –
bending 5.0 ´ 3.5 ´ 50 32 212 26 – –
Note. The mean values are indicated by an asterisk.
stressed surface areas were 1.0 and 0.7, respectively. As is seen from Fig. 5, the total AE count correlates with the
stressed surface area of the specimens.
It is noteworthy that the absolute value of the total AE count under three-point bending was less than that
under four-point bending, and in the majority of cases, the dependence of the total count N on the stress s for each
of the specimens could not be satisfactorily described by expression (6). The minimum values of the relative
deviations S i were more than 4% (for the tested specimens it was frequently impossible to determine these minima
at all), the stress corresponding to the appearance of AE signals s A varied from 0 to 100 MPa, whereas the
microcracking index m had wide scatter (from 1.45 to 6.3). This effect can be due to a nonuniform stress state of the
specimen, a lower number of microcracks being formed, an inadequate sensitivity of the equipment and also a
comparatively high contribution of noisy signals, which could not be eliminated completely. Therefore, for the case
of three-point bending, the microcracking parameters were not calculated.
The microcracking modulus m that characterizes the process of microcrack accumulation and is obtained for
specimens tested in a three- and four-point bending configuration is substantially less than the Weibull modulus
(Table 2), which does not correspond to the assumption of matching of these quantities [9, 11].
Thus, the comparison of the test results for specimens of different sizes under different loading schemes
shows that this material exhibits the microcracking dispersed primarily on its surface, i.e., microcracks are formed on
the material surface.
For further investigations, it seems reasonable to perform the tests of materials at a uniform stress-strain state
of the specimen, namely, in tension. This would make it possible to verify and refine the regularities underlying the
AE that accompanies the microcracking and also to develop a model of microcracking of ceramic composite
materials. However, the tensile tests of ceramics are not an easily solved problem. This is due to the complexity to
ensure the uniform tension of the specimen with elimination of possible bending moments, and also the necessity to
have available a sufficiently large amount of material.
Since the test results for only two materials were presented and discussed in this paper, it is of interest to
extend the above specific features of the microcracking process to the other existing ceramic composite materials.
CONCLUSIONS
1. The use of the AE method allows one to record the microcracking process in composite ceramic materials
under their mechanical loading. Based on the data obtained for the electrical engineering porcelain and the silicon
nitride ceramics, a more accurate expression has been proposed, which will make it possible to describe the
microcrack accumulation and to obtain the stress s versus total count N curves under loading of the above
materials taking into account the stress corresponding to the appearance of AE signals.
2. The assessment of the statistical strength characteristics was performed for the materials under study, and
the values of the Weibull modulus characterizing the scatter in the ultimate strengths of the tested specimens at
different loading rates, specimen dimensions and loading schemes were determined. The notion of the micro-
cracking modulus that describes the kinetics of microcracking with a monotonic increase in the stresses has been
introduced and its values have been shown to differ essentially from those of the Weibull modulus for the given
materials.
78
3. For the silicon nitride ceramics, the dispersed microcracking is primarily a surface one, which follows
from the comparison of the values of the total AE count at the fixed level of stresses for different specimen sizes and
schemes of loading.
REFERENCES
1. A. G. Evans and T. G. Langdon, Structural Ceramics, Pergamon Press (1976).
2. A. V. Drozdov, V. O. Galenko, G. A. Gogotsi, and M. V. Swain, “Acoustic emission during micro- and
macrocrack growth in Mg-PSZ,” J. Amer. Ceram. Soc., 8, 1922–1927 (1991).
3. G. S. Pisarenko, G. A. Gogotsi, A. V. Drozdov, and A. N. Negovskii, “Acoustic emission studies of the
strength of ceramics under mechanical and thermal loads,” Strength Mater., 14, No. 4, 419–425 (1982).
4. G. S. Pisarenko, G. A. Gogotsi, and V. P. Zavada, “Crack resistance of a constructional ceramic,” Strength
Mater., 17, No. 4, 445–450 (1985).
5. S. Sclosa, D. F. Dailly, and G. W. Hastings, “Fracture toughness of hot isostatically pressed alumina,” Trans.
J. Br. Ceram. Soc., 81, 148–151 (1982).
6. A. H. de Aza, J. Chevalier, G. Fantozzi, et al., “Crack growth resistance of alumina, zirconia and zirconia
toughened alumina ceramics for joint prostheses,” Biomaterials, 23, 937–945 (2002).
7. S. M. Barinov, “Fracture toughness of structural engineering ceramics,” in: Results in Science and
Engineering. Technology of Silica-Based and Refractory Non-Metallic Materials [in Russian], Vol. 1,
Moscow (1988), pp. 72–132.
8. É. S. Gevorkyan, V. A. Chishkala, and O. M. Mel’nik, “Some special features of formation of the structure
of partially stabilized zirconia nanopowder-based materials using hot pressing with direct passing the
current,” Visn. Khark. Nat. Univ., Ser. Khimiya, 3, 50–54 (2011).
9. A. G. Evans and M. Linzer, “Failure prediction in structural ceramics using acoustic emission,” J. Amer.
Ceram. Soc., 56, 575–581 (1973).
10. A. G. Evans, M. Linzer, and L. R. Russel, “Acoustic emission and propagation in polycrystalline alumina,”
Mater. Sci. Eng., 15, 253–261 (1974).
11. A. G. Evans, “Residual stress measurements using acoustic emission,” J. Amer. Ceram Soc., 58, 239–243
(1975).
12. W. A. Weibull, “Statistical distribution function of wide applicability,” J. Appl. Mech., 18, 293–297 (1951).
13. A. V. Drozdov, V. V. Kutnyak, and A. N. Negovskii, “Experimental equipment for evaluation of the strength
characteristics of ceramics materials,” Strength Mater., 31, No. 5, 516–522 (1999).
14. G. A. Gogotsi, V. P. Zavada, and O. D. Shcherbina, “Strength and crack resistance of ceramics. Report 2.
Silicon nitride ceramic,” Strength Mater., 16, No. 12, 1656–1659 (1984).
79
