You are on page 1of 4

J O U R N A L O F M A T E R I A L S S C I E N C E L E T T E R S 1 8 (1 9 9 9 ) 1007 – 1010

Correlation between Young’s modulus and porosity


in porous materials
J . K O V ÁČI K
Institute of Materials and Machine Mechanics, Slovak Academy of Sciences, Račianska 75, P.O. Box 95, 830 08
Bratislava, Slovak Republic
E-mail: ummsjk@savba.savba.sk

During the last few years, progress in predicting the called a percolation threshold, at which a solid phase
elastic properties of porous materials over an entire forms a continual network spanning the whole system.
porosity range has been closely related to the power-law At and above the percolation threshold, the geometri-
empirical relationship of Phani and Niyogi [1] cal, physical and mechanical properties of the system
µ ¶ behave as
p f
E = E0 · 1 − (1) E ∝ (n − n c ) f for n ≥ n c (2)
pc
where E is the effective Young’s modulus of porous where E is an investigated property, n is the volume
material with porosity p, E 0 is Young’s modulus of fraction of the solid material and f is the critical expo-
solid material, pc is the porosity at which the effective nent theoretically predicted for the investigated prop-
Young’s modulus becomes zero and f is the parameter erty.
dependent on the grain morphology and pore geome- The percolation theory expects that the values of the
try of porous material [1]. As was noted by Wagh et al. critical exponents are universal, i.e., they do not de-
[2], fittings of the experimental data to Equation 1 often pend on the structure and geometrical properties of the
give pc = 1 [1, 3] and do not explain the data accurately. system but only depend on the dimension of the prob-
In recent experimental works, either pc ≡ 1 is prefer- lem [9]. On the other hand, the value of the percolation
ably used [4–6] or a linearized model ( f ≡ 1) by Lam threshold depends significantly on the structure. In a
et al. [7] is used, where pc is considered to be an initial three-dimensional (3D) structure, the experimental val-
powder porosity. ues of the percolation threshold as low as 0.06 vol %
In this letter, it will be shown that the empirical rela- and as high as 60 vol % were found [10, 11].
tionship shown in Equation 1 is identical with the per- When in Equation 2 the porosity instead of the vol-
colation theory equation for the behavior of Young’s ume fraction of powder is used, p = 1 − n, one obtains
and shear modulus with porosity. Further, the applica-
bility of the percolation model for Young’s modulus of E ∝ ( pc − p) f for p ≤ pc (3)
porous materials will be demonstrated and the results
will be discussed. Hence, Equation 3 ought to satisfy the boundary con-
The porous materials are preferably prepared from dition E = E 0 ; at p = 0 it can be expressed as follows
powders, the particle size and shape of which can vary µ ¶f
significantly. During the powder consolidation, various pc − p
E = E0 for p ≤ pc (4)
porosities can be achieved by varying the technologi- pc
cal parameters such as temperature, external pressure or
time. Compacting starts from just touching powder par- which is identical to Equation 1.
ticles and goes to the lower porosity by the creation and The percolation theory predicts f = 2.1 for Young’s
growth of the necks between particles. The subsequent modulus in 3D [12]. This value is determined for an in-
closure of the pore channels leads to the elimination finite cluster or when all dimensions of the system tend
of the pores. Three various porosity ranges can be usu- to infinity. In the continuum, however, this universal be-
ally identified, e.g., Danninger et al. [8] observed for havior is often affected by the finite size of the system
sintered iron the following porosity ranges: [13], thus giving a characteristic exponent instead of a
1. porosity ≤3%: fully isolated pores of nearly spher- critical exponent. It implies that for all porous materi-
ical or elliptical shape als the characteristic exponent f ought to be nearly the
2. porosity ≥20%: fully interconnected pores of same. On the other hand, the value of the percolation
complex shape threshold is a function of the powder size, shape, their
3. porosity between 3% and 20%: both isolated and distributions and the preparation method. This is often
interconnected pores are present in various amounts. missed in the literature [1, 3, 6], where the experimen-
This indicates that the powder consolidation is in tal data from various sources are joined in an attempt
general a connectivity problem, which is studied by to obtain the wider porosity range for the investigated
the percolation theory [9]. According to the percola- material, thus leading to the unrealistic values of the
tion theory, there exists a critical volume fraction nc , fitting parameters pc and f .
0261–8028 °
C 1999 Kluwer Academic Publishers 1007
T A B L E I Fitting results for powder size influence on Young’s mod- T A B L E I I I Fitting results for porosity range influence on Young’s
ulus porosity dependence for porous Th2 O [14] (χ 2 is a minimization modulus porosity dependence for sintered iron
function)
Sintered iron Porosity
Th2 O powder Porosity reference range E 0 (GPa) pc f χ2
size (µm) range E 0 (GPa) pc f χ2
[16] 0–0.22 212.1 ± 3.5 0.41 ± 0.09 1.13 ± 0.36 33.2
0–2 0–0.33 263.9 ± 3.4 0.37 ± 0.02 1.22 ± 0.10 18.7 [8] + [16] 0–0.28 213.2 ± 3.1 0.42 ± 0.05 1.19 ± 0.28 31.5
2–4 0–0.39 264.2 ± 3.9 0.52 ± 0.04 1.36 ± 0.17 18.9
4–44 0–0.27 262.9 ± 3.3 0.46 ± 0.06 1.13 ± 0.23 11.3

T A B L E I I Fitting results for preparation method influence on Young’s


modulus porosity dependence for porous Si3 N4 [15]

Si3 N4 doped Porosity


with range E 0 (GPa) pc f χ2

5 wt % MgO 0–0.44 315.7 ± 6.7 0.53 ± 0.03 1.66 ± 0.18 14.1


5 wt % CeO2 0–0.38 284.1 ± 4.3 0.44 ± 0.02 1.13 ± 0.12 28.0
12 wt % Y2 O3 + 0–0.12 264.9 ± 3.6 0.31 ± 0.19 0.94 ± 0.75 29.2
8 wt % SiO2

Figure 2 Influence of preparation method on Young’s modulus porosity


dependence for porous Si3 N4 [15] doped with: (—䊊—) 12 wt % Y2 O3 +
8 wt % SiO2 ; (- -䊉- -) 5 wt % MgO; (· · · 䉫 · · ·) 5 wt % CeO2 .

Figure 1 Influence of powder size on Young’s modulus porosity depen-


dence for porous Th2 O [14]: (- - 䊊 - -) 0–2 µm; (· · · 䊉 · · ·) 2–4 µm;
(—䉫—) 4–44 µm.

The influence of the powder size can be illustrated


using the data for sintered Th2 O [14]. The percolation
threshold increases with increasing size of used Th2 O Figure 3 Young’s modulus porosity dependence for sintered iron: (䊊)
powder (see Table I). It is due to the fact that the finer data by Panakkal et al. [16], (䊉) extended by Danninger et al. [8]. Solid
line is Equation 4 for all data.
powder (0–2 µm) possesses higher packaging capabil-
ity than the larger one (2–4 µm). The fitting results for
the powder size of 4–44 µm deviates from this (see
Fig. 1), but the porosity range for this powder is signif- The same data by Panakkal et al. [16] in the porosity
icantly smaller than for finer ones. range 0–0.22 were extended by Danninger et al. [8]
The influence of the preparation method can be to the porosity range of 0–0.28. Even such a small in-
shown using porous Si3 N4 [15]. The characteristic ex- crease of the porosity range significantly increases the
ponent for porous Si3 N4 doped with 5 wt % MgO is values of the obtained fitting parameters (see Table III
much higher than characteristic exponents for porous and Fig. 3).
Si3 N4 doped either with 5 wt % CeO2 or 12 wt % Y2 O3 The influence of the powder shape can be illustrated
+ 8 wt % SiO2 (see Table II and Fig. 2). Also in this case, using sintered α-Al2 O3 trigonal corundum [17]. There,
the small porosity range of Si3 N4 doped with 12 wt % an egg-shaped powder possesses smaller packaging ca-
Y2 O3 + 8 wt % SiO2 gives a very low estimate of the pabilities and, therefore, the higher percolation thresh-
fitting parameters. old than the spherical powder (Figs. 4 and 5). Hence,
It indicates that the increase of the porosity range up the sonic measurements of Young’s modulus were per-
to the percolation threshold can lead to the increase of formed for porosity from 0.2 up to pc , and it was nec-
the values of fitting parameters. This can be demon- essary to include the Young’s modulus of solid alumina
strated using the experimental data of sintered iron: E 0 = 401 GPa [17] into the fitting data. Table IV shows
1008
T A B L E I V Fitting results for powder shape influence on Young’s modulus porosity dependence for porous alumina [17]

α-Al2 O3 Powder shape Porosity range E 0 (GPa) pc f χ2

Spherical–bending 0–0.39 399.6 ± 7.7 0.43 ± 0.02 1.25 ± 0.13 96.2


Spherical–sonic 0–0.39 398.7 ± 6.5 0.43 ± 0.01 1.15 ± 0.06 43.3
Egg-shaped–bending 0–0.42 425.2 ±10.9 0.47 ± 0.03 1.39 ± 0.18 234.1
Egg-shaped–sonic 0–0.43 401.6 ± 7.0 0.46 ± 0.01 1.20 ± 0.06 51.4

Figure 4 Young’s modulus porosity dependence for α-Al2 O3 trigo- Figure 6 Log–log plot of Young’s modulus versus 1 − p/ pc ; pc ≡
nal corundum [17], sonic measurement: (—䊊—) spherical powder; 0.37 is tap porosity: (· · · 䊊 · · ·) sintered alpha–two titanium alumide [5];
(· · · 䉬 · · ·) egg-shaped powder. (—䉬—) sintered glass [18], bending measurement.

Figure 5 Young’s modulus porosity dependence for α-Al2 O3 trigo- Figure 7 Log–log plot of Young’s modulus versus 1 − p/ pc for sintered
nal corundum [17], bending measurement: (— 䊊—) spherical powder; glass [18], bending measurement. pc ≡ 0.42 is initial powder porosity.
(· · · 䉬 · · ·) egg-shaped powder.

also that there exists significant difference between the percolation threshold corresponds to the initial pow-
sonic and bending fitting parameters for both powder der porosity. Unfortunately, the initial powder porosity
types. was mentioned only for sintered glass [18]. Setting this
The available experimental data of Young’s modulus value into Equation 1 also gives a good fit for high-
of porous materials are usually from a narrow low- porosity data (Fig. 7).
porosity range far from the percolation threshold. This In summary, the percolation model was found to de-
problem can in some cases be overcome by using the tap scribe fairly well the Young’s modulus–porosity depen-
porosity as the percolation threshold. The characteris- dence of porous materials. It was further shown that it
tic exponent can then be better estimated. Fig. 6 il- is necessary to have a porosity range as wide as possi-
lustrates this approach for sintered alpha–two titanium ble for the investigated material prepared by the same
alumide [5] and sintered glass [18]. From Table V, it can preparation method from the same type of the powder,
be seen that the obtained estimates of the characteris- to incorporate the property of the solid material into
tic exponent are more reasonable. The fitting results fitting process (when known) and to estimate the per-
for both materials, however, do not fit well with the colation threshold (preferably the initial powder poros-
high-porosity data, possibly because the value of the ity). Only when these requirements are fulfilled can one
1009
T A B L E V Influence of estimation of percolation threshold on the characteristic exponent for Young’s modulus of sintered alpha–two titanium
alumide [5] and sintered glass [18]

Material Porosity range E 0 (GPa) pc f χ2

Alpha–two titanium alumide 0–0.30 97.2 ± 3.3 0.45 ± 0.12 1.64 ± 0.62 12.1
Alpha–two titanium alumide (tap porosity) 0–0.30 95.9 ± 3.2 ≡0.37 1.18 ± 0.10 14.1
Glass–static 0–0.35 55.6 ± 4.3 0.53 ± 0.15 2.03 ± 0.98 1.0
Glass–static (tap porosity) 0–0.35 52.6 ± 2.7 ≡0.37 1.02 ± 0.13 32.6
Glass–static (initial porosity) 0–0.35 54.2 ± 2.6 ≡0.42 1.32 ± 0.15 29.3
Glass–dynamic 0–0.30 73.2 ± 2.6 0.95 ± 0.87 4.35 ± 4.57 14.4
Glass–dynamic (tap porosity) 0–0.30 69.4 ± 1.8 ≡0.37 1.28 ± 0.07 17.2
Glass–dynamic (initial porosity) 0–0.30 70.3 ± 1.8 ≡0.42 1.56 ± 0.08 15.7

obtain the realistic value of the characteristic exponent References


for Young’s modulus of porous material. When they are 1. K . K . P H A N I and S . K . N I Y O G I , J. Mater. Sci. 22 (1987) 257.
not fulfilled, the obtained values are merely the best fit- 2. A . S . W A G H , R . B . P O E P P E L and J . P . S I N G H , ibid. 26
(1991) 3862.
ting parameters valid only for the investigated porosity 3. K . K . P H A N I and S . K . N I Y O G I , J. Mater. Sci. Lett. 6 (1987)
range. 511.
Almost all investigated materials possess the value 4. A . R . B O C C A C I N I , G . O N D R A C E K and E . M O M B E L L O ,
of the characteristic exponent f for Young’s modulus ibid. 14 (1995) 534.
in the range of 1.10–1.70. As was mentioned previ- 5. T . E . M A T I K A S , P . K A R P U R and S . S H A M A S U N D A R ,
J. Mater. Sci. 32 (1997) 1099.
ously, in 3D f = 2.1 [12]. Low value of f can probably 6. A . K . M A I T R A and K . K . P H A N I , ibid. 29 (1994) 4415.
be explained by the mechanical stability of the porous 7. D . C . L A M , F . F . L A N G E and A . G . E V A N S , J. Amer.
material at early stages of the consolidation process: Ceram. Soc. 77 (1994) 2113.
The stability of the system is provided by small vol- 8. H . D A N N I N G E R , G . J A N G G , B . W E I S S and R .
umes in and around the necks between the particles. S T I C K L E R , pmi 25 (1993) 170 and 219.
9. D . S T A U F F E R and A . A H A R O N Y , “Introduction to Percola-
There, the created bonds are significantly weaker than tion Theory,” 2nd ed. (Taylor & Francis, London, 1992).
inside the particles due to the high concentration of the 10. R . S C H U E L E R , J . P E T E R M A N N , K . S C H U L T E and H . P .
surface inhomogeneities, cracks and impurities. The re- W E N T Z E L , J. Appl. Pol. Sci. 63 (1997) 1741.
gions of unbound particles encapsulated inside the cre- 11. I . B A L B E R G and N . B I N E N B A U M , Phys. Rev. B 35 (1987)
ated skeleton can be also found. Uncertainty about the 8749.
12. M . S A H I M I , “Applications of Percolation Theory” (Taylor &
precise determination of the percolation threshold, Francis, London, 1994) p. 185.
however, can be another reason for low values of f. 13. I . B A L B E R G and N . B I N E N B A U M , Phys. Rev. B 33 (1986)
Only the porous Si3 N4 doped with 5 wt % MgO can 2017.
be clearly considered as a 3D system, f = 1.66 ± 0.18. 14. S . S P I N N E R , F . P . K N U D S E N and L . S T O N E , J. Res. Natl.
It is probably due to the fact that MgO improves Bur. Std. 67C (1963) 39.
15. G . de P O R T U and P . V I N C E N Z I N I , “XIII Siliconf.,” Vol. 1
the binding between Si3 N4 powder particles, thus in- (Budapest, Hungary, 1981) p. 265.
creasing the load-carrying volume near the percola- 16. J . P . P A N A K K A L , H . W I L L E M S and W . A R N O L D ,
tion threshold. The obtained result is similar to f = J. Mater. Sci. 25 (1990) 1397.
1.66 ± 0.07 for the Young’s modulus of porous alu- 17. J . C . W A N G , ibid. 19 (1984) 801 and 809.
minum foam (porosity range 0–0.85, pc ≡ 1) [19]. In 18. L . C O R O N E L , J . P . J E R N O T and F . O S T E R S T O C K ,
ibid. 25 (1990) 4866.
this case, the lower value of the characteristic exponent 19. J . K O V ÁČI K and F . S I M A N ČÍK , Scripta Mater. 39 (1998) 239.
is due to the finite size of the system.
It must be noted that all presented ideas need further
investigation in the field of Young’s and shear modulus Received 6 October 1998
of porous materials. and accepted 11 February 1999

1010

You might also like