Powder Technology 268 (2014) 269–278

Contents lists available at ScienceDirect

Powder Technology
journal homepage: www.elsevier.com/locate/powtec

Review

Experimental analysis and finite element simulation of the co-sintering

of bi-material components
H. Ou, M. Sahli ⁎, J.-C. Gelin, T. Barrière
FEMTO-ST Institute/Applied Mechanics Department, ENSMM, 24 chemin de l’épitaphe, 25000 Besançon, France

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the use of numerical simulations to describe solid state diffusion of the sintering stage
Received 12 June 2014 during a Powder Injection Moulding (PIM) process for micro-bi-material components based on a thermo-
Received in revised form 5 August 2014 elasto-viscoplastic model. The sintering behaviour was studied with dilatometer experiments, gravitational
Accepted 14 August 2014
beam-bending and free sintering tests. As a complement to this experimental study, a finite element simulation
Available online 27 August 2014
of the operation was performed. The simulations were based on constitutive equations identified from specific
Keywords:
experiments performed for each blend at different sintering heating rates and loadings. Finally, the simulation
Modelling results are compared to other experimental and simulation results to evaluate the reliability of the proposed
Simulation model. The numerical results addressing shrinkage and the relative density were found to be fully consistent
Sintering with the experimental observations.
Fine 316L stainless steel powders © 2014 Elsevier B.V. All rights reserved.
Copper
Bi-material components

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
2. Experimental procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
2.1. Powders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
2.2. Binder system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
2.3. Feedstock elaboration and characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3. Modelling of the sintering process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3.1. Constitutive model for the sintering process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3.2. Sintering parameter identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
4. Numerical identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
5. Numerical simulations of the sintering process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
5.1. Boundary and initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
5.2. Numerical results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
6. Experimental validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

1. Introduction
Sintering is a manufacturing process used for making various parts
from metal or ceramic powder mixtures. Sintering can be defined as
the thermal transformation of bulk materials into compact solids at
⁎ Corresponding author.
E-mail address: mohamed.sahli@ens2m.fr (M. Sahli).
temperatures below their melting point [1–4]. The main phenomenon
is solid state diffusion, which occurs along the different interfaces of
the crystals and through any vacancies. Changes in the microstructure
during sintering have been previously demonstrated (see Fig. 1).
Based on previous experience, the sintered material has to be homoge-
neous and characteristics such as porosity need to be controlled.
Sintering is a complex process influenced by many factors including
temperature, sintering time, pressure and atmospheric composition, all

http://dx.doi.org/10.1016/j.powtec.2014.08.023
0032-5910/© 2014 Elsevier B.V. All rights reserved.
270 H. Ou et al. / Powder Technology 268 (2014) 269–278
Fig. 1. Microstructure of raw, formed, and sintered metallic products.
Fig. 2. Schematic diagram of the metal injection moulding (MIM).
of which determine the final properties of the sintered product [5–8].
Modelling can be used to optimise and better understand the sintering
process to improve the quality of the sintered components. Modelling
of the sintering process is one of the most challenging problems in the
field of material modelling. There are different approaches to modelling
sintering processes, ranging from continuum phenomenological models
to micromechanical and atomistic ones. Different sintering models have
been reviewed in Refs. [9–12].
The first analyses of sintering were based on the famous two sphere
models that provided a simplified description of inter-particle neck
growth and densification through miscellaneous diffusion mechanisms
[13–15]. Schoenberg at al. [16] compared analytical calculations and fi-
nite element analysis simulations (FEA) to describe the sintering of a
barium titanate cylindrical component composed of a high-density
and low-density layer. Song et al. [17] investigated simulating the
sintering process of 316L stainless steel powder components with a
thermo-elasto-viscoplastic model. In this paper, we focus on numerical
simulations associated with the sintering stages for micro-bi-material
components in an assembly or separately at different powder volume
loadings. Frenkel [18] and Kuczynski [19] studied mechanisms for
neck growth and shrinkage during early sintering stages (particle bond-
ing) using a two-sphere model. Coble [20] developed a cylindrical pore
model. A spherical pore model for the later sintering stages was devel-
oped by McKenzie and Shuttleworth [21]. Barriere et al. [22,23] investi-
gated the optimal process parameters by proposing adapted finite
Table 1
Characteristics of the WC–Co powders.
Powder Particle shape d10
Fine 316L stainless steel Spherical 1.80 μm
Copper Irregular 3.73 μm
element modelling and simulation software for the metal injection
moulding stage (MIM) based on a bi-phasic model. More sophisticated
models that address the superposition of various sintering mechanisms
have been developed by Ashby [24], Arzt [25], and Exner and Arzt [26].
Clearly, modelling can help optimise the process parameters to bet-
ter obtain near-net shape and crack-free components. The purpose of
this paper is to simulate the sintering of bi-material components using
a finite element method. First, we introduce constitutive equations
that describe the behaviour of both materials during sintering. These
equations, which include both thermo-elastic and viscous terms, have
been adjusted with data extracted from specific sintering experiments
by the inverse method. Then, we plotted the results from the numerical
simulation in terms of dimensional changes and relative densities. Fig. 2
shows a schematic of the bi-metal injection moulding process.
2. Experimental procedures
2.1. Powders
Fine spherically shaped 316L stainless steel powder (D50 = 3.4 μm)
and irregularly shaped copper powder (D50 = 6.3 μm) were provided
by Sandvik Osprey Company®. This shape is generally more appropriate
for obtaining a feedstock that behaves with a low viscosity. The powders
had densities equal to 7.9 g cm−3 and 8.9 g cm−3, respectively.
d50 d90 Density Tap density
3.40 μm 6.02 μm 7.90 g/cm3 4.60 g/cm3
6.34 μm 10.81 μm 9.82 g/cm3 4.63 g/cm3
 H. Ou et al. / Powder Technology 268 (2014) 269–278 271

Table 2 shear viscosity vs. shear rate curve is shown in Fig. 3. The viscosities of
Characteristics of the different binder components. both feedstocks are considered suitable for injection moulding.
Binders Density [g/cm3] Melting temperature [°C]

Stearic acid (SA) 0.89 70

3. Modelling of the sintering process
Paraffin wax (PW) 0.91 60
Polypropylene (PP) 0.90 160 3.1. Constitutive model for the sintering process

Compared with conventional trial and error methods, numerical

1000
316L-60% 316L-62% simulations of sintering using finite element methods can be more ef-
316L-64% 316L-66% fective by minimising the effects of tooling and processing parameters
[27]. The model chosen for this investigation is a phenomenological
Shear viscosity [Pa.s]

Copper-60%
100 model based on continuum mechanics and uses thermo-elasto-
viscoplastic formulations [17]. The strain rate consists of three parts:
the elastic strain rate ε̇e , the thermal strain rate ε̇th and the viscoplastic
strain rate ε̇vp . These are related by the following equation [28]:
10

devðσ Þ σ m −σ s
ε̇ ¼ C eσ̇þ αΔṪI þ þ I; ð1Þ
2G 3K
1
where Ce is the elastic compliance matrix, α is the thermal expansion co-
100 1000 10000
Shear rate [s-1]
efficient, Δ Ṫ is the incremental temperature rate, I is the second order
identify tensor, α is experimentally determined using a dilatometer,
σm = tr(σ)/3 is the trace of the stress tensor, I, G is the shear viscosity
Fig. 3. Rheological characterisation of the 316L stainless steel and copper feedstocks
obtained at 160 °C using the same capillary (Ø = 1 mm, L = 16 mm). modulus, K is the bulk viscosity modulus, and σs is the sintering stress.
The variables G, K and σs are material parameters that still need to be
2.2. Binder system determined.
The elastic–viscous analogy is used to define the shear and bulk vis-
The binder components were polypropylene (PP), paraffin wax cosity moduli for sintering materials [29]:
(PW) and stearic acid (SA). The PP primary binder is used to retain
ηp η
the component shape after injection moulding and debinding. The Gp ¼
 ; Kp ¼
p  ð2Þ
main effect of the PW secondary binder is to decrease the feedstock 2 1 þ νp 3 1−2ν p
viscosity and increase the replication ability of the feedstock. The SA
surfactant is used to facilitate powder wetting. The composition of where ηp and νp are the uniaxial viscosity and the viscous Poisson's ratio
binder systems used with the stainless steel and copper feedstocks is of a porous material, respectively. Song et al. derived the following
40 vol% for PP, 55 vol% for PW and 5 vol% for SA. The characteristics of relationship to define the uniaxial viscosity ηep through bending tests
the binder systems and powders are shown in Tables 1 and 2. in a dilatometer [30]:
!
4 3
e 1 5ρa gLs PLs
2.3. Feedstock elaboration and characterization ηp ¼ þ ð3Þ
δ̇ 32h2 4bh3
By employing the proposed binder system, 316L stainless steel feed-
stock with a 60–66% solid loading fraction and copper feedstock with a where δ̇ is the deflection rate at the centre of the specimen, ρa is the ap-
solid loading fraction of 60% were used for the experiments. The critical parent density, g is gravity, P is the external load, and Ls, b and h are the
solid loading amounts were determined from a subsequent study [4,34]. distance between the two supporting rods and the width and thickness
The mixing of the powders and binders was carried out using a twin- of the specimen, respectively [30].
screw mixer at 160 °C and 30 rpm for 30 min. A Rosand RH2000® Bordia [30] and Scherer [31] relate a phenomenological expression
capillary rheometer was used to characterise the rheological character- to calculate Poisson's ratio of the sintered material as follows:
istics of the feedstocks. The viscosity was measured at 160 °C over a rffiffiffiffiffiffiffiffiffiffiffiffiffi
predetermined shear rate range (102–104 s− 1), which is usually the 1 ρ ρ0
νp ≈ ;ρ ¼ ; ð4Þ
shear rate range occurring during the injection moulding stage. The 2 3−2ρ ð1 þ λÞ3

Fig. 4. Heating cycles for beam-bending and free sintering testing in the vertical dilatometer for (a) 316L stainless steel and (b) copper powders.
272 H. Ou et al. / Powder Technology 268 (2014) 269–278

Fig. 5. Beam-bending test in a vertical SETSYS® evolution dilatometer: (a) setup for the TMA measurement and the probe, (b) geometry of the sample and the sample support and
(c) setup for free-sintering test.

where ρ is the relative density ρ0 and λ is the uniaxial shrinkage, which [33]. Quinard et al. [34,35] performed experiments using a horizontal di-
is defined as: latometer equipped with a beam-bending test support for the 316L
stainless steel feedstock, at powder volume loadings up to 60%. The ver-
L−L0 tical dilatometer was intensively used in this study. Material parameters
λ¼ ð5Þ
L0 were determined for the 316L stainless steel feedstock at high solid
loadings from 62% to 66% and for the copper feedstock at solid loadings
where L0 and L are the length of the specimens before and after of 60%.
sintering. To properly sinter the material, high sintering temperatures and a
The following equation is used to determine the sintering stress [32]: fast heating rate were used. The densification of the 316L stainless
steel and copper powders occurs at temperatures greater than 900 °C
σ s ¼ Bρ
C
ð6Þ and 700 °C, respectively. The sintering temperatures of the 316L stain-
less steel and copper powders are 1360 °C and 1000 °C, respectively.
where B and C are material parameters identified from dilatometry In addition, heating rates of 5 °C/min, 10 °C/min and 15 °C/min were
experiments. Using these proposed constitutive equations, the related used for identification tests in the vertical dilatometer. Fig. 4 illustrates
material parameters can be determined. the heating cycles used with 316L stainless steel and copper powders
for beam-bending and free sintering tests in the vertical dilatometer.
3.2. Sintering parameter identification The setup used for the beam-bending tests in a SETSYS® vertical di-
latometer is shown in Fig. 5. The associated probe is made up of a base
A beam-bending test process was proposed to identify the sintering
parameters for the parts during the high temperature sintering stage

Fig. 6. Uniaxial viscosity vs. temperature from the beam-bending tests conducted in a Fig. 7. Uniaxial shrinkage vs. temperature from free-sintering tests in a vertical dilatome-
vertical dilatometer for 316L stainless steel feedstock at powder volume loadings of ter for the 316L stainless steel feedstock at powder volume loadings of (a) 62% and
(a) 62% and (b) 64%. (b) 64%.
 H. Ou et al. / Powder Technology 268 (2014) 269–278 273

Table 3 4. Numerical identification

Material parameters B and C used in the sintering process for both the 316L stainless steel
and copper feedstocks.
The sintering stress is the driving force for densification. It depends
Stainless steel ρ b 0.8 ρ ≥ 0.8 on the surface tension, relative density, size and shape of the particles,
feedstock as well as the grain and pore sizes that vary with the microstructural
B (Pa) C B (Pa) C
7 −1 evolution during the sintering process. The identification algorithm is
60% 5 °C/min 3.91 × 10 0.30 1.47 × 10 −93.13
10 °C/min 7.26 × 108 9.79 5.73 × 102 −58.77 designed for the proper identification of material parameters B and C
15 °C/min 1.13 × 108 3.74 3.04 × 104 −39.42 used in the sintering stress model to optimise the numerical simula-
tions. The following equation was proposed to calculate the stress dur-
62% 5 °C/min 6.50 × 103 5.25 2.48 × 103 −5.43
10 °C/min 1.11 × 103 0.05 1.53 × 103 −6.33 ing the sintering stage [32]:
15 °C/min 6.76 × 103 4.65 1.70 × 103 −8.79
1 dL 1
64% 5 °C/min 7.15 × 103 5.23 8.52 × 102 −10.41 ¼ αΔṪ− σ : ð7Þ
10 °C/min 3.98 × 103 3.35 3.09 × 103 1.22 L dt 3K p s
15 °C/min 6.20 × 103 4.64 2.08 × 102 −23.27

66% 5 °C/min 3.09 × 103 3.73 2.44 × 103 −6.17 The proper strategy consists of identifying parameters B and C in
10 °C/min 5.70 × 103 4.97 1.81 × 103 −1.70 Matlab® that determine the numerical shrinkage curve according to
15 °C/min – – – –
Eq. (8). This curve is matched to that obtained from the free sintering
Copper feedstock ρ b 0.87 ρ ≥ 0.87 tests. Therefore, the minimisation algorithm is used to fit the simula-
B (Pa) C B (Pa) C tions as best as possible to the experimental curves by adjusting the
2 −1 physical parameters [32]:
60% 5 °C/min 8.38 × 10 0.29 3.89 × 10 −76.88
10 °C/min 2.28 × 103 3.34 5.16 × 10−2 −49.56 8
15 °C/min 3.41 × 102 −3.51 4.20 × 10−3 −119.70 >
> min F ðxÞ
>
< n 

2
X  m e 
F ðxÞ ¼ λ T i; x −λ T i; x  ; ð8Þ
>
>
>
: i¼1
x ¼ ½B; C 
with two knives and a rod with a knife-shaped cross section. A load
equal to 5 cN was applied at the centre of the specimen through the
rod. The specimens have a rectangular shape of 14 mm in length, where λe is the experimental uniaxial shrinkage obtained from the dila-
5.5 mm in width and 1 mm in thickness. tometry tests, λm is the numerical uniaxial shrinkage, F(x) is the mean
The uniaxial viscosity curve vs. the different sintering temperatures residual squares of the tolerance, where i = 1,…, n indicates the differ-
for the different loaded 316L stainless steel feedstocks were observed ent sintering temperatures and x is the set of material parameters that
and shown in Fig. 6. For identical heating rates, the higher the feedstock need to be identified.
solid loading, the higher the uniaxial viscosities are at the same temper- The sintering cycle for the micro-bi-material components discussed
ature. This is related to the fact that the more powder used, the higher in our investigations is detailed in Fig. 4. By coupling the shrinkage ob-
the related viscosity becomes. tained from the identification tests in the dilatometer, the sintering pro-
Complementary tests were performed in a compression configura- cess is divided into two stages depending on the critical value of the
tion using a vertical dilatometer, as indicated in Fig. 5c. The length of relative density, which is 0.8 for the 316L stainless steel and 0.87 for
the cylindrical specimens is 10 mm, and the diameter is 5 mm. The the copper. The algorithm for the identification of the material parame-
rod and base both have flat surfaces. The uniaxial shrinkage curve vs. ters B and C during the sintering stage for the 316L stainless steel feed-
the different sintering temperatures for the 316L stainless steel feed- stock and copper feedstock was implemented in Matlab©. The identified
stocks were observed and shown in Fig. 7. Shrinkage begins to occur values are presented in Table 3.
at approximately 1000 °C and rapidly increases at a temperature
above 1050 °C until approximately 1200 °C. At the same temperature, 5. Numerical simulations of the sintering process
significant shrinkage is obtained for the specimens fabricated at lower
powder loading. The reason is that when the powder loading is high, 5.1. Boundary and initial conditions
more pores are produced and the components shrink more obviously
after sintering. The same procedure was also applied to cylindrical The cylinder and the ring components were separately injected for
pre-sintering samples using copper feedstocks at a solid loading equal the 316L stainless steel and copper feedstocks, respectively. The cylin-
to 60%. drical component was first sintered at 1360 °C, followed by a second

Fig. 8. FE meshes of the cylindrical micro-part, ring micro-part and the plate support prior to the simultaneous sintering stages.
274 H. Ou et al. / Powder Technology 268 (2014) 269–278

Fig. 9. FE meshes of the micro-bi-material component and the plate support prior to the simultaneous sintering stages.

sintering at 1000 °C, with the ring placed around the cylinder on a rigid
plate.
The whole set of equations based on the thermo-elasto-viscoplastic
model was implemented in the ABAQUS/Standard Finite Element
Software with a user-supplied material routine (UMAT). The assembly
is composed of the micro-bi-material components, and the support is
related through Figs. 8 and 9. The plate support is assumed to be a
rigid body during the simulation, and the coefficient of friction between
the specimens and the support is set at 0.5. The plate support and
the micro-bi-material component elements considered for the simula-
tion are R3D4 and C3D8R elements, respectively. In addition, the
«Pressure/Over-closure» contact was selected for the components.
5.2. Numerical results and discussion
Fig. 10 shows the predicted numerical shrinkage distribution for the
micro-bi-material specimen after sintering. The calculation was based
on the assumption that there was no contact between the two compo-
nents, i.e. that the two parts are not linked together (i.e. two totally sep-
arate parts). A surface–surface contact with the coefficient of friction
between the specimens is set at 0 was reported. Solid loadings of 62%
and 64% were selected for the 316L stainless steel feedstock, whereas
a 60% solid loading was used for the copper feedstock. The maximum
shrinkage occurs at the cylinder surfaces and illustrates the distortion
often induced in the final sintered component.

Fig. 10. Numerical final shrinkage (z-z) of the sintered micro-bi-material component (316L stainless steel feedstock at a solid loading of 62% and 64% and copper feedstock at a solid loading
of 60%) after the sintering stage (at a heating rate: 5 °C/min).

Fig. 11. Numerical final shrinkage of the sintered micro-bi-material component (316L stainless steel feedstock at solid loadings of 60%, 62%, 64%, 66% and copper feedstock at a solid loading
of 60%) after sintering at 1000 °C (heating rate: 15 °C/min, hold time: 120 min).
 H. Ou et al. / Powder Technology 268 (2014) 269–278 275

Fig. 12. Final numerical relative density of the sintered micro-bi-material component after sintering at 1000 °C and heating rate equal to: (a) 10 °C/min and (b) 15 °C/min (hold time:
120 min).

Fig. 13. Final numerical relative density of the sintered micro-bi-material component (316L stainless steel feedstock at a solid loading of 60% and copper feedstock at a solid loading of 60%)
after sintering at 1000 °C (heating rate: 5 °C/min, hold time: 120 min).

The final shrinkage in the micro-bi-material components after the
numerical simulations for a powder volume loading from 60% to 66%
for the 316L stainless steel feedstock and a powder volume loading of
60% for the copper feedstock is illustrated in Fig. 11. A distortion exists
on the contact surface between the 316L stainless steel and copper.
This is likely the result from unbalanced sintering shrinkage kinetics be-
tween the two materials. Careful control over the powder volume load-
ing is important to control the shrinkage between the materials. Indeed,
the outer part exhibits more shrinkage to ensure constant contact be-
tween the two materials.
Besides of shrinkages, the relative densities have been simulated as
well for these micro-bi-material components injected with different
solid loading. The relative density distributions of the sintered micro-
parts have been illustrated in Figs. 12 and 13. As it is shown in the fig-
ures, the relative densities are generally homogeneous for most of the
simulations, in which the variations have been well controlled within
1%. It can be also seen that the relative density of micro-parts increases
steadily with increasing of heating rates. This indicated that the increase
of the heating rates is in favour of the densification of micro-gears. The
relative density of the specimens can be also increased with the incre-
ment of the sintering temperature and solid loading.
6. Experimental validation
During our investigations, bi-material micro-injection moulding
equipment was used to inject both 316L stainless steel and copper feed-
stock into die mould cavities. The mould was composed of three-plates
to separate the die cavities for each feedstock so that the interface be-
tween both feedstocks can be well controlled. The resulting component
from this bi-material injection stage is shown in Fig. 14a. The central cyl-
inder (Ø3 mm × 5.6 mm) is injected with the 316L stainless steel feed-
stock, whereas the ring (Ø6 mm × 2.4 mm) around the central cylinder
corresponds to the copper feedstock. In addition, the central cylinder
and the ring can be injected separately, as shown in Fig. 14b.
The main injection parameters are the same for both the simultaneous
and separate injection operations, including an injection temperature of

Fig. 14. (a) Micro-bi-material component injected with the 316L stainless steel feedstock (solid loading of 60%) and copper feedstock (solid loading of 60%), (b) ring component injected
with the copper feedstock and (c) cylinder component injected with the 316L stainless steel feedstock.
276 H. Ou et al. / Powder Technology 268 (2014) 269–278

Fig. 15. Photographs of the micro-specimens obtained after injection, debinding and sintering: (a) cylindrical-shaped parts fabricated from the stainless steel feedstock loaded at 64% and
(b) ring-shaped parts fabricated from the copper feedstock loaded at 60%.

Fig. 16. Photographs of the bi-material micro-specimens obtained after injection, debinding and sintering using copper feedstock loaded at 60% and stainless steel feedstock loaded at 60%.

220 °C, an injection pressure of 40 bars, and injected volumes of 45 mm3
for the 316L stainless steel feedstock and 60 mm3 for the copper feed-
stock. Figs. 15 and 16 show the correctly debinded and sintered compo-
nents. There are no obvious defects or cracks at the interface of the
different feedstocks, even along the interface zone.
The dimensional analyses were carried out on the three sintered
specimens injected with different feedstock loaded from 60% to 64%.
The same debinding kinetics rate was used for both feedstocks. The in-
jection, debinding and sintering stages of ring micro parts were per-
formed after sintering the cylindrical micro-parts. The mean shrinkage
in the main directions is illustrated in Figs. 17–19. The results show
that the feedstock loaded at 60%, with a sintering cycle utilizing a heating
rate of 5 °C/min results in the maximum shrinkage. Very high shrinkage
in thickness direction was also observed for the ring component injected
with copper feedstock loaded at 60%. This same phenomenon has been
encountered by other researchers, such as N.H. Loh et al. [36]. The
main parameters affecting the final part size have already been studied
and include the metal powder morphology, binder ingredients and
proportions, mixing conditions, mould design, moulding parameters,
20%
5°C/min 10°C/min
Numerical
Experimental
15%
10%
rheological behaviour of the moulding materials, debinding, sintering,
equipment and even environmental conditions. Among these often inte-
grally related factors, the most sensitive are the powder volume loading
and the mould design [36].
The numerical shrinkage curves are shown in Figs. 17–19. The nu-
merically determined curves compare well with the experimental
ones for the three different heating rates. Additionally, these results
were confirmed by the choice of the sintering stress model and the
identification method in our work.
In addition, the relative densities of the sintered micro-components
(cylindrical, ring and bi-material components) were measured by the
water displacement method (Archimede method) for the feedstocks
loaded from 60% to 64%, as reported in Table 4. The relative density
was calculated as the ratio of the measured density to the theoretical
density. All specimens subjected to density measurements were ultra-
sonically cleaned in an alcohol bath for 10 min and rinsed in distilled
water for an additional 10 min to remove any contaminants. The
specimens were then dried with adsorbent paper towels. The average
random and systematic errors in the density measurements were
quantified as 1%. The sintered micro-parts using copper or stainless
steel feedstocks reached relative maximum densities of 92% and 97%,
15°C/min
respectively.
As an example, the relative density distributions (for a solid loading
of 60% and a heating rate of 5 °C/min) of the sintered micro-specimens
are illustrated in Fig. 13. The relative densities were in perfect agree-
ment with the experimental densities, especially for the sintering cycles
with relatively low heating rates.

7. Conclusions
5%

In powder metallurgy, a comprehensive design of powder forming

0%
Outside diameter (Ø=6±0.01) Inside diameter (Ø=3±0.01) Thickness (e=2.4±0.02)
Fig. 17. Comparison between the experimental and numerical shrinkage for the ring com-
ponent after the sintering stage using different heating rates (copper feedstock, sintered
temperature: 1000 °C, solid loading: 60%, units: mm).
and sintering processes is desired to improve the quality of the prod-
ucts. Numeric simulations provide useful information for the process
design. In this work, a method for simulating the microscopic behaviour
of micro-bi-material components during the sintering stage was pro-
posed using a 3D finite element method. A phenomenological model
describing the thermal/mechanical behaviour during the sintering
 H. Ou et al. / Powder Technology 268 (2014) 269–278 277

a b
20% 15%
5°C/min 10°C/min 15°C/min 60% 62% 64%
15% Numerical Numerical
Experimental 10% Experimental

10%

5%
5%

0% 0%
Outside diameter Length (L=5.6±0.02) Outside diameter Length (L=5.6±0.02)
(Ø=3±0.01) (Ø=3±0.01)

Fig. 18. Comparison between the experimental and numerical shrinkage for the cylindrical component after the sintering stage using (a) different heating rates and (b) different solid
loadings (sintered temperature: 1360 °C, stainless steel feedstock, units: mm).

process was proposed, and the involved material parameters were iden- References
tified from the results of specific experiments. The model makes no as-
sumptions about the shapes of the particles and their relative sizes. [1] W. Limberg, T. Ebel, F. Pyczak, M. Oehring, F.P. Schimansky, Influence of the sintering
However, the model correctly reproduces the free-sintering mecha- atmosphere on the tensile properties of MIM-processed Ti 45Al 5Nb 0.2B 0.2C,
Mater. Sci. Eng. A 552 (2012) 323–329.
nism. In a second step, bi-material components were simulated with [2] I. Ul Mohsin, D. Lager, W. Hohenauer, C. Gierl, H. Danninger, Finite element sintering
the Abaqus© finite element code. We analysed the results of these calcu- analysis of metal injection molded copper brown body using thermo-physical data
lations in terms of shape changes and relative densities at different solid and kinetics, Comput. Mater. Sci. 53 (2012) 6–11.
[3] N.H. Mohamad Nor, N. Muhamad, A.K.A. Mohd Ihsan, K.R. Jamaludin, Sintering pa-
loadings. The simulation results are in a good agreement with the ex- rameter optimization of Ti-6Al-4V metal injection molding for highest strength
perimental data, as determined by vertical dilatometry. using palm stearin binder, Proc. Eng. 68 (2013) 359–364.
The numeric simulation described here will be used in the future to [4] M. Sahli, C. Millot, J.-C. Gelin, T. Barrière, The manufacturing and replication of
microfluidic mould inserts by the hot embossing process, J. Mater. Process. Technol.
support optimising the thermal cycle and component geometry so that
213 (2013) 913–925.
near-net shape and crackle components are obtained. Further exten- [5] M. Chmielewski, J. Dutkiewicz, D. Kaliński, L. Litynska-Dobrzynska, K.
sions of the viscoelastic model, modelling of the individual powder par- Pietrzak, A. Strojny-Nedza, Microstructure and properties of hot-pressed
molybdenum-alumina composites, Arch. Metall. Mater. 57 (2012)
ticles and pores and the formulation of the microstructural changes are
687–693.
presently under development. [6] M. Chmielewski, D. Kaliński, K. Pietrzak, W. Włosiński, Relationship between mixing
conditions and properties of sintered 20AlN/80Cu composite materials, Arch. Metall.
Mater. 55 (2010) 579–585.
[7] D. Kaliński, M. Chmielewski, K. Pietrzak, An influence of mechanical mixing and hot-
pressing on properties of NiAl/Al2O3 composite, Arch. Metall. Mater. 57 (2012)
694–702.
[8] W. Weglewski, M. Basista, M. Chmielewski, K. Pietrzak, Modelling of thermally in-
duced damage in the processing of Cr-Al2O3 composites, Compos. Eng. 43 (2012)
255–264.
[9] V. Tikare, M. Braginsky, D. Bouvard, A. Vagnon, Numerical simulation of microstruc-
tural evolution during sintering at the mesoscale in a 3D powder compact, Comput.
Mater. Sci. 48 (2010) 317–325.
[10] S. Martin, M. Guessasma, J. Léchelle, J. Fortin, K. Saleh, F. Adenot, Simulation of
sintering using a non smooth discrete element method. application to the study
of rearrangement, Compu Mater. Sci. 84 (2014) 31–39.
[11] G. Largiller, L. Dong, D. Bouvard, C.P. Carry, A. Gabriel, Deformation and crack-
ing during sintering of bimaterial components processed from ceramic and
metal powder mixes. Part II: numerical simulation, Mech. Mater. 53 (2012)
132–141.
[12] J. Pan, Modelling sintering at different length scales, Int. Mater. Rev. 2 (17) (2003)
69–85.
[13] C. Herring, Surface tension as a motivation for sintering, in: W.E. Kingston (Ed.),
The Physics of Powder Metallurgy, McGraw-Hill, New York, 1951, pp. 143–179.
Fig. 19. Comparison of the experimental and numerical shrinkage for the bi-material [14] G.C. Kuczynski, The mechanics of densification during sintering of metallic particles,
component after sintering at different heating rates (copper and stainless steel feedstocks, Acta Metall. 4 (1956) 58–61.
sintered temperature: 1000 °C, solid loading: 60%, units: mm). [15] R.L. Coble, Initial sintering of alumina and hematite, J. Am. Ceram. Soc. 41 (1958)
55–62.
[16] S.E. Schoenberg, D.J. Green, A.E. Segall, G.L. Messing, A.S. Grader, P.M. Halleck, Stress-
es and distorsion due to green density gradients during densification, J. Am. Ceram.
Table 4 Soc. 89 (2006) 3027–3033.
Density of the sintered micro-parts fabricated from the copper and stainless steel [17] J. Song, J.C. Gelin, T. Barrière, Experiments and numerical modelling of solid state
feedstocks. sintering for 316L stainless steel components, J. Mater. Process. Technol. 177
(2006) 352–355.
Density [g/cm3] Relative density [18] J. Frenkel, Viscous flow of crystalline bodies under the action of surface tension, J.
Phys. USSR 9 (1945) 385–391.
Heating rate Solid 60% 62% 64% 60% 62% 64%
[19] G.C. Kuczynski, Self diffusion in sintering of metallic particles, Metal Transac. 185
[°C/min] loading
(1949) 169–178.
Cylindrical 5 7.20 7.21 7.26 92.30 92.43 93.07 [20] R.L. Coble, Sintering of crystalline solids. I. Intermediate and final state diffusion
components 10 7.29 7.33 7.37 93.46 93.97 94.48 models, J. Appl. Phys. 32 (1961) 787–792.
15 7.46 7.51 7.58 95.64 96.28 97.18 [21] J.K. McKenzie, R. Shuttleworth, A phenomenological theory of sintering, Proc. Phys.
Ring 5 8.15 – – 90.95 – – Soc.: B 62 (1949) 833–852.
[22] T. Barriere, J.C. Gelin, B. Liu, Improving mould design and injection parameters in
components 10 8.23 – – 91.85 – –
metal injection moulding by accurate 3D finite element simulation, J. Mater. Process.
15 8.31 – – 92.74 – –
Technol. 125–126 (2002) 518–524.
278 H. Ou et al. / Powder Technology 268 (2014) 269–278
[23] T. Barriere, B. Liu, J.C. Gelin, Determination of the optimal process parameters in
metal injection molding from experiments and numerical modeling, J. Mater. Pro-
cess. Technol. 143–144 (2003) 636–644.
[24] M.F. Ashby, A first report on sintering diagrams, Acta Metall. 22 (1974) 275–289.
[25] E. Arzt, The influence of an increasing particle coordination on the densification of
spherical powders, Acta Metall. 30 (1982) 1883–1890.
[26] H.E. Exner, E. Arzt, Sintering processes, in: R.W. Cahn, P. Haasen (Eds.), Phys. Metall.,
30, 1996, pp. 2628–2662.
[27] P.C. Yu, Q.F. Li, J.Y.H. Fuh, T. Li, L. Lu, Two-stage sintering of nano-sized yttria stabi-
lized zirconia process by powder injection moulding, J. Mater. Process. Technol.
192–193 (2007) 312–318.
[28] M. Gasik, B. Zhang, A constitutive model and FE simulation for the sintering process
of powder compacts, Comput. Mater. Sci. 18 (2000) 93–101.
[29] J. Song, T. Barriere, B. Liu, J.C. Gelin, G. Michel, Experimental and numerical analysis
on sintering behaviours of injection moulded components in 316L stainless steel
powders, Powder Metall. 53 (2010) 295–304.
[30] R.K. Bordia, G.W. Scherer, On constrained sintering. I. Constitutive model for a
sintering body, Acta Mater. 36 (1988) 2393–2397.
[31] G.W. Scherer, Sintering inhomogeneous glasses: application to optical waveguides,
J. Non-Cryst. Solids 34 (1979) 239–256.
[32] A. Peterson, J. Agren, Constitutive benhavior of WC–Co materials with different grain
size sintered under load, Acta Mater. 52 (2004) 1847–1858.
[33] D.C. Blaine, R. Bollina, S.J. Park, R.M. German, Critical use of video-imaging to ratio-
nalize computer sintering simulation, Comput. Ind. 56 (2005) 867–875.
[34] C. Quinard, T. Barriere, J.C. Gelin, Development and property identification of 316L
stainless steel feedstock for PIM and μPIM, Powder Technol. 190 (2009) 123–128.
[35] C. Quinard, J. Song, T. Barriere, J.C. Gelin, Elaboration of PIM feedstocks with 316L
fine stainless steel powders for the processing of micro-components, Powder
Technol. 208 (2011) 383–389.
[36] N.H. Loh, R.M. German, Statistical analysis of shrinkage variation for powder injec-
tion molding, J. Mater. Process. Technol. 59 (1996) 278–284.