numerical and experimental analysis of anisotropic shrinkage during sintering
TRANSCRIPT
Numerical and Experimental Analysis of Anisotropic
Shrinkage during Sintering
Zhou Penga, Jianzhong Xiaob , Siyang Meic and Shijun Huangd
State Key Laboratory of Material Processing and Die & Mould Technology, Huazhong University of Science and Technology, Wuhan 430074, P. R. China
[email protected], [email protected], [email protected],[email protected]
Keywords: sintering; anisotropic shrinkage; master sintering curve; numerical simulation
Abstract. With a view of simulating the dimensional changes during sintering of rectangle green
bodies, the thermo-mechanical behavior of zirconium powder compacts at high temperature is
investigated. In order to better describe the behavior of anisotropic shrinkage, the revised Master
Sintering Curve is modified. Finite element calculations are then carried out on the green body
according to the modified equation with the different shrinkages coefficients at the different stages
of sintering. The possible causes of the anisotropic shrinkage are explained by macro-surface
energy model. Numerical shape predictions have been compared with experimental data, which are
considered to be in good agreement.
Introduction
A standard manufacturing process for ceramic parts involves several separate stages. Sintering is
the last step of the role of different processing steps, which has shown product shaping (powder
preparation, consolidation, and densification) [1]. In order to reduce the costs of manufacturing, it is
very important to control the dimensional change during sintering, which has been widely studied in
recent years.
Since the late 1940s, the sintering theory has received great attentions, a lot of successful
approaches have been made to understand the mechanisms of shrinkage which mainly based on
three states, namely the early sintering stages [2,3], the intermediate shrinkage stages [4,5] and the
late sintering stages [6]. In the ordinary way, researchers used the concept of the master sintering
curve (MSC) to predict and control sintering [7]. The MSC is an empirical curve that provides a
characteristic measure of the densification of a material over a given density range [8].
Unfortunately, all this previous work ignored the effect of anisotropic shrinkage which can cause a
common and serious problem in ceramics production [9-12].
Though Toussaint’s model [13] considered a lot of additional effects including anisotropic
shrinkage, he based his work on thermo-elastic equation which ignored the different sintering states.
Thus, the modle is inconvenience for application.
Improved accuracy in the predictions demands not only proper characterization of the
densification of material but also constitutive models of increased realism which are capable of
predicting dimensional changes during sintering. To explore the role of the anisotropic behavior, we
employ numerical simulations of a three-stage law. This paper concentrates on the application of
sintering models to explain some of the trends observed in the anisotropic shrinkage measurements
as sintering progresses.
Experimental procedures
Raw materials. Commercially available Y-TZP (NANO) with average size between 50 and 100 nm
is used in this experiment. The nano powder is first heat treated to 100 ºC for 24 h prior to mixing
for de-agglomeration. Feedstock with 59 wt.% solid content is shown in
Table 1. The tested specimens are cubic samples (3.0mm × 4.4mm × 54mm) and injection moulded
by SZL50g Moulding Machine.
Advanced Materials Research Vols. 233-235 (2011) pp 3068-3073Online available since 2011/May/12 at www.scientific.net© (2011) Trans Tech Publications, Switzerlanddoi:10.4028/www.scientific.net/AMR.233-235.3068
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP,www.ttp.net. (ID: 141.219.44.39, Michigan Technological University, Houghton, USA-09/09/14,10:10:24)
Table 1. Composition’s weight of feedstock
Feedstock Weight of powder (g) PW (g) EVA (g) SA (g)
59% 800 65.55 17.85 0.84
Debinding and sintering. After slow thermal debinding profile, the green body with very slow
heating rates (from 25 ºC to 100 ºC at 2 ºC /min; to 400 ºC at 5 ºC /min; to 25 ºC at 0.5 ºC /min) is
carried out in atmosphere environment. The slow heating rate can prevent defects such as
microcracks that are induced during the binder extraction process; while debinding to 400 ºC
provides strength to the debound sample for easy handling. Sintering with heating rate (from 25 ºC
to 400 ºC at 3 ºC /min; to 900 ºC at 5 ºC /min, and hold for 1 h, to 25 ºC at 5 ºC /min) is carried out
in atmosphere environment.
Modelling
Gillia et al.[14] used the model that considers the behavior of isotropic ceramic compacts during
sintering as a continuous process. And the total strain rate can be decomposed into two reversible
parts (shrinkage, thermal expansion) and two irreversible parts (free sintering, Newtonian
viscosity). While taking into account the mechanical aspects, some models are based on the three
different sintering stages despite the fact that the sintering mechanisms of these three stages may
not be the same. For instance, the shrinkage behavior can be quite different according to the
sintering time [15].And the situation becomes much more complicated in the case of an anisotropic
material [16]. There are different shrinkages in different sintering stages, so we can define different
coefficients of Thermal-Elasticity Model. The shrinkage in the first stage exhibited a linear
increase, and then exhibited a parabolic increase in the second stage followed by a linear increase in
the final stage.
The transversely-isotropic thermo-elastic law contains thermal expansion coefficients in the
different stages during sintering and the elastic modulus, poisson ratios and shear modulus of the
green body of ceramics. We can consider thermal expansion coefficients of green body as negative
values, since it will generate shrinkage deformation. It can be measured by thermal analysis
experiments or the shrinkages after sintering. The shrinkage rates of the X, Y and Z axial direction
are xε , yε and zε . Measured at the end of free sintering tests, they are supposed to be proportional
to the variation in temperature 0 fT T T∆ = − ( 0T is the initial temperature and fT the maximal
sintering temperature) as
x xTε α=∆ (1)
y yTε α=∆ (2)
z zTε α=∆ (3)
Where xα , yα and zα are the thermal expansion coefficients of X, Y and Z axial direction.
These parameters are negative since the materials shrinks when the temperature increases.
During the sintering, the sample is subjected to a stress yσ in the Y axial direction; the induced
deformations in the three directions are calculated from the following relations:
1
y
y y TE
σ
ε α′ = + ∆ (4)
1
x y x
vT
Eε σ α′ =− + ∆ (5)
Advanced Materials Research Vols. 233-235 3069
1
z y z
vT
Eε σ α′ =− + ∆ (6)
It can be deduced that
1
y
y y
Eσ
ε ε
=′ −
(7)
x x
y y
vε ε
ε ε
′ −=−
′ − (8)
The last parameter is the shear modulus G, which cannot be identified from previous
experimental data. We estimated it as
( )1
2 1
EG
v=
+ (9)
To the cubic sample, we can deduce that
( )( )( )1 1 1 1v x y zε ε ε ε= − − − − (10)
vv
T
ε
α =∆
(11)
The shrinkage of the body is different in different sintering stages:
1 1 2 2 3 3v T T Tε α α α= ∆ + ∆ + ∆ (12)
In Eq. 12, The thermal expansion coefficients of initial stage, intermediate stage and final stage
are α1, α2 and α3, and the corresponding temperature variations are ∆T1, ∆T2 and ∆T3, respectively.
It can be measured by experiment.
Table 2 summarises the main values of the thermo-elastic parameters that have been obtained
from experimental data. The thermal expansion coefficients are found with a good precision, a few
percents, whereas the accuracy of the elastic moduli is much poorer, between 10 and 25%. Table 2. Main values used for numerical simulations
α(ºC -1) -3.64×10
-4
E(GPa) 5.87
Glt(GPa) 4.11
νt 0.30
Numerical simulation
For the present study, several finite element software, either commercial or academic, have been
devoted to the macroscopic simulation of sintering in order to predict shape changes of green body
[17]. We used 3D module of ANSYS code with the transversely-isotropic constitutive equations
previously presented (Eq. 4 , Eq. 5 and Eq. 6 ) and the material parameters listed in Table 2.
Calculations are carried out with the plane stress assumption. The contact between the cuboid
green body and the load bearing board is supposed to be frictionless. Sintering is simulated by
applying to the green body a uniform temperature of 1450 ºC. For describing horizontal sintering
the green body is assumed to be also submitted to its own weight. In these cases, the calculated
thermo-elastic deformation of the green body resulting from these conditions is assumed to be its
final deformation after the sintering cycle.
Fig. 1 compares the results before and after sintering. The size change of the green body
indicated different extents of the shrinkage in X, Y and Z axial direction. The blue part represents
3070 Fundamental of Chemical Engineering
the body after sintering, and the larger part denotes the body before sintering. In Table 3, the data
show the experimental and numerical shrinkage, and they have a maximum relative error of 5%.
The final density calculated from equations was in good agreement with the values determined
using the three-stage orthotropic thermo-elastic law.
Fig. 1. Comparison of simulation of the cubic sample before and after sintering. The blue part is
the green body after sintering, and the white part is the green body before sintering
Table 3 shows the differences between the experimental group and numerical group are always
less than 5%. The shrinkages in Y and Z axial direction are particularly better predicted (less than
2%) than in X axial direction (less than 5%). Considering the complexity of Thermal-Elasticity
model, the result could be acceptable. Table 3. Comparison between experimental measurements and numerical prediction of the sample
Axial direction Experimental shrinkage (%) Numerical shrinkage (%) Error (%)
X 21.7 22.7 4.61
Y 18.67 18.3 1.98
Z 19.09 19.2 0.58
Discussion
The large defects and cracks caused by the effect of anisotropic shrinkage may be developed into a
complex shape of compact during sintering. The formation of the compact requires a
time-consuming machining involving diamond tools for reshaping which markedly increases the
production cost of ceramics. Anisotropic sintering shrinkage is observed in most of the forming
processes—e.g. injection molding, extrusion, tape casting. The possible causes of shape changes
during sintering are: ①heterogeneous green density field owing to powder-die friction or
non-uniform compression ratio during die pressing, ② structural anisotropy induced by
unidirectional die compaction, ③creep under the stresses created by the weight of the part, ④
heterogeneous temperature field in the part [18]. And Zavaliangos et al. [19] believed a number of
possible sources (including prior compaction, elongated particles, gas pore pressure, interface
porosity and crystallographic texture) of sintering anisotropy based on simulations of a
two-dimensional array of particles.
The pervious models described anisotropy shrinkage were too complex, for example, Wakai and
Akatsu used the linear sintering stress tensor and the viscosity tensor’s model to understand
anisotropic viscosities and shrinkage [20], however, uncertainty increases because that the model
contained too much variables which are inexact measured in practice, thus it is very important to
build a model which is suitable for application. During sintering, any mechanism that does not
cause the particle centers move closer will not lead to the densification. Instead, coarsening will
occur as matter is transferred from the surface to the neck region of the particles. The nondensifying
mechanisms involve: surface diffusion, lattice diffusion from the surface, and vapor transport. Even
Advanced Materials Research Vols. 233-235 3071
though these mechanisms do not lead to the densification, they do have an effect on accelerating or
alleviating the densification progress, because they cause a decrease in the neck curvature, which in
turn slows the rate of the densifying mechanisms. Accompanying the neck growth is an increase in
the compact strength. The mechanisms that lead to densification are grain boundary diffusion,
lattice diffusion from the grain boundary and plastic flow. All these possible sintering mechanisms
produce neck growth and surface energy reduce. For this reason, the rate at which the width of the
neck increases can be used to measure the rate of sintering. Furthermore, it can be used to compare
the relative rates of those mechanisms. Additionally, the rate of shrinkage of the powder compact
can be used to measure the rate of sintering for the densifying mechanisms. From the macroscopic
model, the rate of shrinkage of the cubic sample also used to measure the rate of sintering in
different stages.
For the phenomenon of shrinkage in sintering experiments, the free sintering process of ceramic
is the process of lowering its surface energy as well as free energy from the theory of diffusion. The
less specific surface area of ceramics, the less surface energy goes. The length, width, and height of
specimen after sintering are l, b and h, respectively, so the specific surface area S is as follows:
(13)
If l = b = h, the value of takes minimum value as well as the surface energy, when
the value of (l + b + h) is definite in the cuboid model. The variation of l, b and h tends to be
equivalent from the viewpoint of the trend of drop in free energy. However, as Fig. 2, the value of l
is larger than its b and the value of b is larger than its h in specimen, thus the relationship of
shrinkage rate is as follows: l b hε ε ε> > . It is identical to our experiments and numerical
simulations.
Fig. 2. The cubic model of macro-surface energy. The length, width, and height of specimen after
sintering are l, b and h, respectively, the value of l is larger than its b and the value of b is larger
than its h in specimen, thus the relationship of shrinkage rate is as follows: l b hε ε ε> > .
Conclusions
This study proposed a valuable approach for predicting the final deformation of anisotropic Y-TZP
(NANO) ceramics cubic sample during sintering. And the new predicting approach takes the
advantage of the classical method based on the three-stage Model. Numerical shape predictions and
experimental data are considered to be in good agreement. The results showed the differences
between the experimental group and numerical group are always less than 5%. The shrinkages in Y
and Z axial direction are particularly better predicted (less than 2%) than in X axial direction (less
than 5%). The possible causes of the anisotropy in shrinkage during sintering are discussed by
macro-surface energy model. The less specific surface area of ceramics, the less surface energy
goes. And the accuracy in predictions is improved by considering the additional effect of anisotropy
shrinkage.
3072 Fundamental of Chemical Engineering
Acknowledgements
We are grateful to Mrs. Wei Wei and Mr. Dai Wei in State Key Laboratory of Material Processing
and Die & Mould Technology of Huazhong University of Science and Technology for their kind
support.
References
[1] J.C. Agarwal: Adv. Ceram. Mater. Vol. 1 (1986), p. 32
[2] J. Frenkel: J. Phys (USSR). Vol. 9 (1945), p. 385
[3] G.C. Kuczynski: Met. Trans. Vol. 185 (1949), p. 896
[4] R.L. Coble: J. Appl. Phys. Vol. 32 (1961), p. 787
[5] W.D. Kingery and M. Berg: J. Appl. Phys. Vol. 26 (1955), p. 1205
[6] J.K. Mackenzie and R. Shuttleworth: Proc. Phys. Soc. Sec. B. Vol. 62 (1949), p. 833
[7] H. Su and D.L. Johnson: J. Am. Ceram. Soc. Vol. 79 (1996), p. 3211
[8] W.Q. Shao, S.O. Chen, D. Li, H.S. Cao, Y.C. Zhang and S.S. Zhang: Sci. Sinter. Vol. 40
(2008), p. 251
[9] J.S. Sung, K. D. Kood and J.H. Park: J. Am. Ceram. Soc. Vol. 82 (1999), p. 537
[10] P.M. Raj and W.R. Cannon: J. Am. Ceram. Soc. Vol. 82 (1999), p. 2619
[11] D.S. Park and C.W. Kim: J. Mater. Sci. Vol. 34 (1999), p. 5827
[12] A. Shui and N. Uchida: Powder. Technol. Vol. 127 (2002), p. 9
[13] F. Toussaint and D. Bouvard: J. Mater. Process. Tech. Vol. 147 (2004), p. 72
[14] O. Gillia and D. Bouvard: Adv. Powder. Metall. Part. Mater. Vol. 2 (1996), p. 739
[15] S. Kiani, J. Pan and J.A. Yeomans: J. Am. Ceram. Soc. Vol. 89 (2006), p. 3393
[16] A. Jagota, P.R. Dawson and J.T. Jenkins: Mech. Mater. Vol. 7 (1988), p. 255
[17] H.G. Kim, O. Gillia, P. Doremus and D. Bouvard: Int. J. Mech. Sci. Vol. 44 (2002), p. 2523
[18] O. Lame, D. Bouvard and H. Wiedemann: Powder. Metall. Vol. 45 (2002), p. 181
[19] A. Zavaliangos, J.M. Missiaen and D. Bouvard: Sci. Sinter. Vol. 38 (2006), p. 13
[20] F. Wakai and T. Akatsu: Acta. Mater. Vol. 58 (2010), p. 1921
Advanced Materials Research Vols. 233-235 3073
Fundamental of Chemical Engineering 10.4028/www.scientific.net/AMR.233-235 Numerical and Experimental Analysis of Anisotropic Shrinkage during Sintering 10.4028/www.scientific.net/AMR.233-235.3068
DOI References
[4] R.L. Coble: J. Appl. Phys. Vol. 32 (1961), p.787.
http://dx.doi.org/10.1063/1.1736107 [5] W.D. Kingery and M. Berg: J. Appl. Phys. Vol. 26 (1955), p.1205.
http://dx.doi.org/10.1063/1.1721874 [6] J.K. Mackenzie and R. Shuttleworth: Proc. Phys. Soc. Sec. B. Vol. 62 (1949), p.833.
http://dx.doi.org/10.1088/0370-1301/62/12/310 [7] H. Su and D.L. Johnson: J. Am. Ceram. Soc. Vol. 79 (1996), p.3211.
http://dx.doi.org/10.1111/j.1151-2916.1996.tb08097.x [8] W.Q. Shao, S.O. Chen, D. Li, H.S. Cao, Y.C. Zhang and S.S. Zhang: Sci. Sinter. Vol. 40 (2008), p.251.
http://dx.doi.org/10.2298/SOS0803251S [9] J.S. Sung, K. D. Kood and J.H. Park: J. Am. Ceram. Soc. Vol. 82 (1999), p.537.
http://dx.doi.org/10.1111/j.1151-2916.1999.tb01798.x [10] P.M. Raj and W.R. Cannon: J. Am. Ceram. Soc. Vol. 82 (1999), p.2619.
http://dx.doi.org/10.1111/j.1151-2916.1999.tb02132.x [11] D.S. Park and C.W. Kim: J. Mater. Sci. Vol. 34 (1999), p.5827.
http://dx.doi.org/10.1023/A:1004770520830 [12] A. Shui and N. Uchida: Powder. Technol. Vol. 127 (2002), p.9.
http://dx.doi.org/10.1016/S0032-5910(02)00004-9 [13] F. Toussaint and D. Bouvard: J. Mater. Process. Tech. Vol. 147 (2004), p.72.
http://dx.doi.org/10.1016/j.jmatprotec.2003.11.019 [15] S. Kiani, J. Pan and J.A. Yeomans: J. Am. Ceram. Soc. Vol. 89 (2006), p.3393.
http://dx.doi.org/10.1111/j.1551-2916.2006.01252.x [17] H.G. Kim, O. Gillia, P. Doremus and D. Bouvard: Int. J. Mech. Sci. Vol. 44 (2002), p.2523.
http://dx.doi.org/10.1016/S0020-7403(02)00189-3