optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent...

7
Optimization of Powder Injection Molding of Feedstock Based on Aluminum Oxide and Multicomponent Water-Soluble Polymer Binder Berenika Hausnerova, 1 Lucie Marcanikova, 1 Petr Filip, 2 Petr Saha 1 1 Polymer Centre, Faculty of Technology, Tomas Bata University in Zlı ´n, TGM 275, 762 72 Zlı ´n, Czech Republic 2 Department of Mechanics of Fluids and Disperse Systems, The Institute of Hydrodynamics, Academy of Sciences of the Czech Republic, Pod Patankou 5, 166 12 Prague, Czech Republic Analyses crucial to optimize powder injection molding of feedstock based on aluminum oxide powder and multicomponent polymeric binder are provided with the aim to obtain defect-free, high density parts. As the critical step of the process is the flow of highly filled (60 vol%) compound into a mold cavity, rheologi- cal properties supplemented by thermal and pressure- volume-temperature characteristics are measured and described. Upon shear deformation the feedstock undergoes structural changes, which are quantified in terms of yield stresses obtained using Herschel-Bulk- ley and Casson methods. Further, the rheological model is developed to describe the flow behavior of the feedstock in the whole shear rate range measured. Thermogravimetric analysis is performed to optimize debinding step of the process, and two possible ways of the binder removal are proposed: purely thermal and combined solvent/thermal. The quality of the final sintered parts is demonstrated on scanning electron microscopic images of their surfaces. POLYM. ENG. SCI., 51:1376–1382, 2011. ª 2011 Society of Plastics Engineers INTRODUCTION Powder injection molding (PIM) is an effective (high added value) and attractive alternative to the traditional processes (machining, investment casting) for the produc- tion of complex-shaped small parts. It combines a com- mon processing route for plastics—injection molding— with metallurgical sintering. PIM process might be generally divided into four con- sequential steps: (1) compounding metal or ceramic pow- der with a polymers mixture (called ‘‘binder’’) to obtain homogeneous highly filled feedstock, (2) injection mold- ing of prepared feedstock into a mold with a required design resulting in a ‘‘green’’ part, (3) thermal and/or sol- vent removal of a polymer binder creating a ‘‘brown’’ part, and (4) sintering the remaining powder structure to a high density component [1]. The process phase that still requires clarification and optimization is the flow of highly filled feedstock into a mold cavity during injection molding since the defects in the final parts (after sintering) are created during molding and cannot be reduced or eliminated during the following steps as debinding and sintering [2]. Thus, investigation of flow properties is of crucial im- portance in the PIM process optimization. While the rhe- ology of suspensions of noninteracting spheres seems to be well established, the understanding rheological behavior of multiphase systems as those intended for PIM represents a difficult task. For the cavity filling with mini- mized jetting a pseudoplastic flow that relieves process- ability is required [1]. Although, it is a common cause for unfilled polymers, PIM compounds show a complicated sensitivity to variations with shear rate, even if the binder behaves in a Newtonian fashion. Upon powder loading, the Newtonian plateau becomes reduced or disappears. It has been widely accepted that the change into non-Newto- nian behavior arises from the disruption of agglomerates formed by particles [3]. The two mechanisms affecting viscosity can be discerned: the agglomerates’ destruction during shearing causes a decrease in the amount of sus- pending fluid trapped among particles, and thus viscosity decreases due to the drop of an effective volume fraction of powder [4], and simultaneously the change in viscosity is related to the dissipation energy rising from the rotation and distortion of particle agglomerates (e.g., as shown in Ref. 5). A yield point often appears at low shear rates as an in- dication of temporary particle network structure within Correspondence to: Berenika Hausnerova; e-mail: [email protected] Contract grant sponsor: The Grant Agency of the Czech Republic; contract grant number: 103/08/1307; contract grant sponsor: The Ministry of Educa- tion, Youth and Sports of the Czech Republic; contract grant number: MSM 7088352101. DOI 10.1002/pen.21928 Published online in Wiley Online Library (wileyonlinelibrary.com). V V C 2011 Society of Plastics Engineers POLYMER ENGINEERING AND SCIENCE—-2011

Upload: berenika-hausnerova

Post on 06-Jul-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

Optimization of Powder Injection Molding ofFeedstock Based on Aluminum Oxide andMulticomponent Water-Soluble Polymer Binder

Berenika Hausnerova,1 Lucie Marcanikova,1 Petr Filip,2 Petr Saha11 Polymer Centre, Faculty of Technology, Tomas Bata University in Zlın, TGM 275, 762 72 Zlın, Czech Republic

2 Department of Mechanics of Fluids and Disperse Systems, The Institute of Hydrodynamics, Academy ofSciences of the Czech Republic, Pod Patankou 5, 166 12 Prague, Czech Republic

Analyses crucial to optimize powder injection moldingof feedstock based on aluminum oxide powder andmulticomponent polymeric binder are provided withthe aim to obtain defect-free, high density parts. Asthe critical step of the process is the flow of highlyfilled (60 vol%) compound into a mold cavity, rheologi-cal properties supplemented by thermal and pressure-volume-temperature characteristics are measured anddescribed. Upon shear deformation the feedstockundergoes structural changes, which are quantified interms of yield stresses obtained using Herschel-Bulk-ley and Casson methods. Further, the rheologicalmodel is developed to describe the flow behavior ofthe feedstock in the whole shear rate range measured.Thermogravimetric analysis is performed to optimizedebinding step of the process, and two possible waysof the binder removal are proposed: purely thermaland combined solvent/thermal. The quality of the finalsintered parts is demonstrated on scanning electronmicroscopic images of their surfaces. POLYM. ENG.SCI., 51:1376–1382, 2011. ª 2011 Society of Plastics Engineers

INTRODUCTION

Powder injection molding (PIM) is an effective (high

added value) and attractive alternative to the traditional

processes (machining, investment casting) for the produc-

tion of complex-shaped small parts. It combines a com-

mon processing route for plastics—injection molding—

with metallurgical sintering.

PIM process might be generally divided into four con-

sequential steps: (1) compounding metal or ceramic pow-

der with a polymers mixture (called ‘‘binder’’) to obtain

homogeneous highly filled feedstock, (2) injection mold-

ing of prepared feedstock into a mold with a required

design resulting in a ‘‘green’’ part, (3) thermal and/or sol-

vent removal of a polymer binder creating a ‘‘brown’’

part, and (4) sintering the remaining powder structure to a

high density component [1].

The process phase that still requires clarification and

optimization is the flow of highly filled feedstock into a

mold cavity during injection molding since the defects in

the final parts (after sintering) are created during molding

and cannot be reduced or eliminated during the following

steps as debinding and sintering [2].

Thus, investigation of flow properties is of crucial im-

portance in the PIM process optimization. While the rhe-

ology of suspensions of noninteracting spheres seems to

be well established, the understanding rheological

behavior of multiphase systems as those intended for PIM

represents a difficult task. For the cavity filling with mini-

mized jetting a pseudoplastic flow that relieves process-

ability is required [1]. Although, it is a common cause for

unfilled polymers, PIM compounds show a complicated

sensitivity to variations with shear rate, even if the binder

behaves in a Newtonian fashion. Upon powder loading,

the Newtonian plateau becomes reduced or disappears. It

has been widely accepted that the change into non-Newto-

nian behavior arises from the disruption of agglomerates

formed by particles [3]. The two mechanisms affecting

viscosity can be discerned: the agglomerates’ destruction

during shearing causes a decrease in the amount of sus-

pending fluid trapped among particles, and thus viscosity

decreases due to the drop of an effective volume fraction

of powder [4], and simultaneously the change in viscosity

is related to the dissipation energy rising from the rotation

and distortion of particle agglomerates (e.g., as shown in

Ref. 5).

A yield point often appears at low shear rates as an in-

dication of temporary particle network structure within

Correspondence to: Berenika Hausnerova; e-mail: [email protected]

Contract grant sponsor: The Grant Agency of the Czech Republic; contract

grant number: 103/08/1307; contract grant sponsor: The Ministry of Educa-

tion, Youth and Sports of the Czech Republic; contract grant number:

MSM 7088352101.

DOI 10.1002/pen.21928

Published online in Wiley Online Library (wileyonlinelibrary.com).

VVC 2011 Society of Plastics Engineers

POLYMER ENGINEERING AND SCIENCE—-2011

Page 2: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

melt (e.g., [6, 7]), although some authors claim that yield

stress might be a result of extrapolation rather than a real

phenomenon [8, 9] as pointed out by Kurzbeck et al. [3].

However, from the engineering viewpoint, the Casson

method [5] based on an energy dissipation mechanism or

the empirical Herschel-Bulkley model [10] are widely

accepted ways of yield stresses evaluation.

Upon further increase of shear rates the particle struc-

ture is destroyed, particles and polymer orientate and

order in the flow direction to allow interparticle motion,

and the viscosity is dominated by hydrodynamic interac-

tions [11] resulting in shear thinning.

Highly concentrated compounds (about 50 vol% solids

and higher) may exhibit a radical change on their flow

curves accompanied by distortions of the extrudate sur-

face expressing themselves similarly to the spurt flow of

e.g., linear polyethylene [12, 13]. The mechanism of these

flow instabilities is however different as investigated and

reported in [14–16].

Because PIM is a high-pressure molding process, flow

behavior, and compressibility of the powder and polymer-

based binder in a pressurized melt-stage are key indicators

for the assessment of processing conditions. The pressure-

volume-temperature (PVT) characteristic provides infor-

mation about the specific volume of feedstock at the

molding temperature and the pressure necessary for the

production of defect-free, injection molded parts.

Although the shrinkage of a green body is needed to be

incorporated into the mold design, PVT studies on PIM

materials are reported scarcely [17–20]. Further, Greene

and Heaney [19] proved that the holding pressure could

be effectively used to control the dimensional stability of

the final sintered parts. An excessive pressure may cause

relaxation problems and higher shrinkage as reported also

by Laddha et al., [17] for aluminum oxide (56 vol%)

feedstock. This results in a suggestion of choosing the

holding pressure level after appropriately referring to

PVT-diagram in such a way that the residual pressure in a

cavity before mold opening is near to the atmospheric

one.

In this article, the PIM process of aluminum oxide

(alumina) feedstocks is optimized. Although metals such

as stainless steels (316L, 17-4PH) are nowadays prevail-

ing PIM material, aluminum oxide parts (thread guides)

represent the earliest PIM application dating back to the

1930s [21]. The advantage of alumina powder as the most

widely used PIM ceramic [22] over the metals consists in

a combination of good mechanical properties with a low

specific weight. The PIM alumina products find their

application in areas where are exposed to extreme condi-

tions such as high temperatures, corrosive atmosphere, ab-

rasive conditions, or high loads at extreme temperatures

[23]. Due to the above properties, alumina powder is uti-

lized in chemical and electronic industries as chemical

processing and heat treatment equipments, melting cruci-

bles and inject printheads, and an increasing potential of

ceramics is registered also in a medical sector as cardiac

rhythm management, cardiovascular and endoscopic devi-

ces, neurological sensors and stimulators, cochlear

implants, dental implants and abutments, fluid handling,

and small joints.

The aim is to produce nondefect parts via careful con-

trol of the flow properties of the feedstock. As already

pointed out, the problems created during flow into a mold

appear during/after debinding and/or sintering, and there-

fore, their solution consists in rheology. In addition, ther-

mogravimetric analysis (TGA) to set the conditions for

debinding is provided. In the recent years, solvent debind-

ing combined with a thermal method becomes common

as decreases a cycle time. Organic solvents predominantly

used (n-heptane, acetone, trichlorethane) are flammable,

carcinogenic and environmental pollutants [24]. Thus,

the water-soluble binder components (polyethylene

glycol, polyvinyl alcohol, agar, etc.) appear as favorable

alternative.

EXPERIMENTAL

Materials

In this study, highly compressive superground alumi-

num oxide (alumina) powder MARTOXID1 MR70

(Albemarle Corporation) with a specific surface area

(BET) 6–10 m2/g, bulk density � 0.90 g/cm3, green den-

sity 2.20-2.40 g/cm3, and fired density (1,6008C, 2 h)

3.80-3.92 g/cm3 was used, see Fig. 1. The particle size

distribution was measured using ANALYSETTE 22

MicroTec plus (FRITSCH GmbH, Germany) equipment,

and Table 1 compares the obtained data with the data

sheet determined with CILAS 1064 (CILAS, France).

The powder was compounded with a commercial

multicomponent binder Licomont EK 583-G (Clariant,

Switzerland), which is partially water-soluble with a den-

sity 1.05-1.15 g/cm3 and the softening point at 105-

1158C. During mixing in a blade kneader at 1608C for

2 h a surfactant (1 wt% oleic acid) was added. Subse-

quently, 60 vol% feedstock in a form of pellets was

acquired from a single-screw extruder.

Methods

Rheological properties of the alumina feedstock were

measured using a capillary rheometer Rheograph 2001

(Gottfert, Germany) at shear rates from 101 to 104 sec21

at temperatures 150, 160, and 1708C. The length-to-diam-

eter (L/D) ratio of capillary was 30. The apparent viscos-

ity values are presented since the data measured with an

orifice capillary (L/D ¼ 0.12/1) were rather scattered.

Rheological behavior of the binder was determined with

the help of a rotational rheometer Physica MCR501

(Anton Paar, Austria). Shear viscosities were measured at

shear rates ranging from 1021 to 6 3�102 sec21 and at

temperatures from 150 to 1708C in steps of 58C.

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2011 1377

Page 3: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

PVT characteristics were obtained with a PVT-100

(SWO, Germany) apparatus. The specific volumes were

examined at pressures and temperatures in the range 30–

200 MPa and 50-2508C, respectively, in a measurement

mode of isobaric heating with a heating rate of 58C/min.

Thermal properties were determined via a differential

scanning calorimeter (DSC) Pyris 1 DSC (Perkin-Elmer)

in a sealed aluminum pans under nitrogen atmosphere

with heating/cooling rate 108C/min at temperatures rang-

ing from 45 to 1808C.The TGA was performed with the samples of dimen-

sions (3 3 3 3 2) mm with a SETSYS Evolution

1200TG (Setaram, France) thermogravimeter from ambi-

ent temperature to 1,2008C.The green and brown parts’ surface morphologies as

well as the surface of the sintered parts were observed

with a scanning electron microscopy (SEM) using Vega II

(Tescan, Czech Republic) microscope operated at 10 kV;

all samples were coated with a thin layer of gold using a

polaron sputtering apparatus.

RESULTS AND DISCUSSION

From Fourier Transform Infrared Spectroscopy analysis

(not included) it is supposed that the binder contains poly-

olefines, paraffin waxes, and polyethyleneglycols. Its

rheological characteristic was acquired at five different

temperatures in the range 150-1708C in steps of 58Cusing the rotational rheometer. The effect of temperature,

especially at shear rates higher than 10 sec21, is not so

pronounced with increasing temperature above 1608C(Fig. 2). Overall level of binder viscosity lies in the range

proposed for PIM technology, that is, \0.1 Pa.s at the

processing shear rate to provide PIM mixtures with vis-

cosity below 1,000 Pa.s [25].

Filling the binder with 60 vol% alumina powder

resulted in the feedstock with flow properties far from

required above. Fine alumina powder is relatively

hydroscopic and binder is rather sensitive to destabili-

zation in water resulting in enhanced viscosity. This

problem can be solved via improving the ‘‘flowability’’

of the system with dispersants and lubricating agents.

Reduction of viscosity about one order of magnitude

has been reported by Lin and German [26] for

56 vol% alumina powder in a paraffin wax added with

4 wt% of stearic acid (SA). In case of Chan and

Lin [27] SA molecules adsorption on alumina powder

surface changed the flow course from dilatant to

pseudoplastic.

The rheological data obtained for 60 vol% alumina

feedstock modified with 1 wt% oleic acid are depicted in

Fig. 3. Oleic acid has been chosen with regard to investi-

gation by Tseng [28] comparing the effect of stearic acid,

TABLE 1. Particle size distribution of alumina powder.

Fraction (%)

Particle Size (lm)

CILAS 1064 ANALYSETTE 22

\10 0.200–0.400 0.240

\50 0.500–0.800 0.643

\90 1.500–3.000 2.573

FIG. 2. Temperature-dependent viscosity versus shear rate of multi-

component binder.

FIG. 3. Temperature-dependent viscosity versus shear rate of alumina

feedstock.

FIG. 1. Scanning electron micrograph of alumina powder.

1378 POLYMER ENGINEERING AND SCIENCE—-2011 DOI 10.1002/pen

Page 4: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

oleic acid, and 12-hydroxystearic acid (2 wt%) on the

flow behavior of 60 vol% alumina feedstock. Their meas-

urements at high shear rates region (1,000–15,000 sec21)

showed similar effect of SA and oleic acid on mixture

viscosity at 1508C. In addition Persson et al. [18] demon-

strated that 1% of SA added to the iron-based feedstock

has the same effect as 2% of SA, decreasing viscosity

four times.

The viscosity values (Fig. 3) are about three to two

orders of magnitude higher in comparison to binder viscos-

ity at the corresponding shear rates. At lower shear rates

(up to 500 sec21) the viscosity of feedstock decreases with

increasing shear rate suggesting particle or binder molecule

orientation and ordering with flow. When the shear rate

reaches 500 sec21 particles cannot form layers and slide

over each other as firstly reported by Hoffman [29], and

shear thinning turns into a dilatant flow. Similar phenom-

enon was observed for all investigated temperatures. There

is still considerable uncertainty about the source of such

behaviour. Jansma and Qutubuddin [30], who studied this

phenomenon using different viscometers, showed that it

could not be an experimental artefact due to the wall slip.

The mechanism proposed by Barnes [31] is that with

increasing shear stress (rate) the layers formed in the pseu-

doplastic flow region becomes disrupted, and at a certain

shear stress (rate) are fully eliminated (flow turns into dilat-

ant). It implies that every highly concentrated suspension

exhibits dilatant flow if proper flow conditions (depending

on filler concentration, particle size distribution as well as

viscosity of a polymer component) are selected.

For the alumina feedstock investigated, this structure

restructuralization appears repeatedly what reflects in

repeated changes between pseudoplastic and dilatant flow

behavior. Herschel-Bulkley [10] and Casson [5] models

applied to the rheological data resulted in similar values

of yield stress as can be seen from Table 2, corresponding

well to Kurzbeck et al. [3] studying inorganic pigment/

paraffin wax compounds, supporting Casson’s idea of

energy dissipation mechanism responsible for viscosity

variation with shear rate.

Herschel-Bulkley and Casson models cannot describe the

flow properties of alumina feedstock in the whole range of

the measured data because classical empirical models are

primarily determined for characterization of monotonous de-

pendence of viscosity on shear rate or stress. As the course

of data in Fig. 3 is consecutively shear thinning, shear thick-

ening and again shear thinning, a more complicated model

has to be applied. Among other things it implies a model

with the increased number of empirical parameters. As even

for more complicated shear-thinning (i.e., still monotonous)

materials lacking their ‘‘symmetry’’ (in other words -the

drop-off in viscosity from first Newtonian plateau is mir-

rored by its rapid leveling-off towards second Newtonian

plateau) Roberts et al. [32] used so-called generalized Ellis

model with eight parameters, it is natural that for this non-

monotonous behavior, we apply a model with the same

number of parameters. Further, the model should relate vis-

cosity with shear stress and not shear rate because for two

phase materials the stress is continuous across the interface

whereas the deformation rate is not, see Lomellini and Ferri

[33]. This is exactly the problem with interpretation of rheo-

logical data in Hoffman [29] where he claims that with

increasing volume fraction of suspension the respective

curves viscosity vs. shear rate exhibit discontinuous behav-

ior. If the rheograms are transformed to the relation viscosity

vs. shear stress then there is possible to eliminate the prob-

lem with a seeming discontinuity. For rheological modeling

of the data introduced in Fig. 3 the eight-parameter model

published in Filip et al. [34] is applied

Z ¼ Z1 expð�f1Þb1 þ expðf1Þ þ expð�f1Þ þ

Z2 expð�f2Þb2 þ expðf2Þ þ expð�f2Þ

(1)

where

f1 � f ðg; c1; p1Þ ¼ logðc1tÞp1 ; f2 � f ðg; c2; p2Þ ¼ logðc2tÞp2(2)

TABLE 2. Yield stresses values calculated from Herschel-Bulkley and

Casson models.

Temperature (8C)

Yield Stress (kPa)

Herschel-Bulkley Casson

150 50 47

160 45.5 44

170 40 37

FIG. 4. Temperature-dependent shear stress versus shear rate of alu-

mina feedstock; solid lines represent data fitting by the rheological

model.

TABLE 3. Parameters of rheological model applied to viscosity data

of alumina feedstock.

Parameter (2) [f(T) ¼ 1þ(170 2 T)/80]

a1 108 � f(T) a2 0.02 � f(T)b1 21.9 b2 21.99986

c1 0.034 c2 0.00072

p1 0.635 p2 0.028

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2011 1379

Page 5: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

The results of the fitting the experimental data with

this model are shown in Fig. 4 using the relation

s ¼ g � _c, the parameters are summarized in Table 3. The

parameters g1 and g2 represent the values of first and sec-

ond Newtonian plateau, respectively, and they dominantly

influence the maxima of the corresponding peaks. The pa-

rameters bi, ci, and pi significantly participate in the ‘‘geo-

metrical forming’’ of the left (i ¼ 1) and right (i ¼ 2)

peaks. The three parameters consecutively determine the

rate of pseudoplastic/dilatant change, horizontal shift

(along the abscissa), and steepness of the respective peaks

(for more detail see David and Filip [35]). It is necessary

to point out that six out of eight parameters are fixed and

the same for all temperatures, the remaining two depend

on the identical empirically derived temperature-depend-

ent function. It is supposed that the empirical parameters

can be further linked to the materials characteristics when

the corresponding database will be created.

The PVT data of the alumina feedstock under iso-

baric heating regime is shown in Fig. 5. As it can be

seen the temperature transitions are imperceptible. In

contrast, Persson et al. [18] reported (using the same

device) 420 stainless steel feedstock transition zones

corresponding to the particular components of their

commercial binder. Similarly, Wei et al. [20] when

using PVT-100 device for 85 wt% alumina with paraf-

fin wax based binder system obtained the transition

zones indicating clearly the binder components. The

result obtained for alumina feedstock might be

explained as the consequence of the multicomponent

character of the binder whose particular components

have overlapping melting zones as can be seen from the

DSC data (Table 4). Subsequently, DSC data reveals a

slight drop of melting temperature of binder after 60

vol% addition of alumina powder. Specific heat

capacity of the feedstock (cp), determined at three tem-

peratures in the range 120-1608C in 208C steps, varies

with temperature as follows: (1.072 6 0.055), (1.080 60.066), and (1.083 6 0.080) J/(g 8C), respectively.

During debinding process, the binder must be com-

pletely removed before starting the sintering cycle to hold

the shape of the ceramic part. From the debinding meth-

ods, the thermal removal of polymer based binders is

widely used to remove organic components before sinter-

ing. TGA is mostly employed to design a debinding

cycle.

The optimal conditions for debinding and following

sintering of the investigated alumina feedstock are deter-

mined as follows: the heating rate 508C/h from ambient

temperature to 808C, then the heating rate is slowed down

to 108C/h for temperatures in the range 80–2808C, and

finally sintering temperature 1,6008C is reached at the rate

1008C/h. During thermal debinding the binder can leave

the part as liquid or vapor. Liu and Tseng [36] propose

evaporation as a dominant mechanism for the low-molec-

ular-weight binder systems.

Thomas-Vielma et al. [37] investigated for alumina

with HDPE based binder that solvent/thermal combination

decreases the time of a debinding cycle by cca 4 h in

comparison to purely thermal debinding. Figure 6 shows

the weight loss for both thermal and combined solvent/

thermal debinding process. The first stage of major

weight-loss occurs in the temperature range �320–4008C,

FIG. 5. PVT characteristic of alumina feedstock under isobaric regime.

TABLE 4. Melting peak temperatures of polymer binder and alumina

feedstock.

Material

Peak Melting Temperature (8C)

#1 #2 #3 #4

Binder 60.5 6 0.3 77.8 6 0.1 89.2 6 0.1 110.5 6 0.1

Feedstock 58.2 6 0.3 77.0 6 0.3 87.2 6 0.1 109.6 6 0.5

FIG. 6. TGA of binder removal.

FIG. 7. Scanning electron micrographs of final sintered alumina part in

a bulk (a) and detailed (b) resolutions.

1380 POLYMER ENGINEERING AND SCIENCE—-2011 DOI 10.1002/pen

Page 6: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

and the second takes place from � 420 to 4708C. The

first stage of binder removal should be closely equivalent

to the removing of binder wax and low-molecular sub-

stances. At this stage the evaporation is assumed to be a

dominant mechanism for the wax removal. Consequently,

the second stage of weight loss is attributed decisively to

the disposal of the low-molecular-weight polyolefines [36]

and polyethyleneglycols. Trunec and Cihlar [38] studied

the influence of various atmospheres for 59.8 vol% alu-

mina filled in EVA based binder. In their case, the atmos-

phere containing oxygen caused defected parts due to the

formation of nonvolatile layer from oxidative degradation

slowing down the evaporation of low-molecular-weight

components. On the opposite, binder removal in nitrogen

and carbon dioxide exhibited defect-free specimen. The

thermal debinding of the alumina feedstock, investigated

in this work, has been successfully carried out under air

atmosphere as can be demonstrated with the SEM pictures

of the final sintered aluminum oxide parts (Fig. 7). The

final porosity of the sintered part is only 1.1305%, which

in terms of the range acceptable for PIM parts (up to 2–

5%) [1].

TGA curves shows (Fig. 6) that during the solvent

debinding (608C, 3 h) there is extracted about 4 wt% of

water-soluble components prior to the thermal process.

During solving, water diffuses into the binder to react

with water-soluble substance, and its molecules diffuse

out of the sample through a network of pores formed by

remaining polymer backbone and alumina particles [39].

Finally, the scanning electron microscopy verifies the

TGA results. Alumina powder well covered with multi-

component binder shown in Fig. 8a is compared with the

state after removal of water-soluble components from the

feedstock (Fig. 8b). At the surface depicted in the Fig. 8c,

the binder is completely removed and the part has an

open presintered porosity.

CONCLUSION

Alumina powder grade for PIM technology was com-

bined with a commercially available multicomponent

binder. An oleic acid was used as modifier to attain suita-

ble viscosity level of 60 vol% feedstock. Rheological,

thermal, PVT, and morphological analyses together with a

proper tailoring of debinding conditions resulted in the

optimization of the production of nonporous homogenous

ceramic parts.

REFERENCES

1. R.M. German and A. Bose, Injection Molding of Metals andCeramics, Metal Powder Industries Federation, Princeton (1997).

2. P. Schwenzel and J.F. Petzold, EuroPM2009 Proc., 2, 129(2009).

3. S. Kurzbeck, J. Kaschta, and H. Munstedt, Rheol. Acta., 35,446 (1996).

4. C.R. Wildemuth and M.C. Williams, Rheol. Acta., 23, 627(1984).

5. N. Casson, ‘‘A Flow Equation for Pigment-oil Suspension of

the Printing Ink-type,’’ in Rheology of Disperse Systems,C.C. Mill, Ed., Pergamon, London, 84 (1959).

6. A.B. Metzner, J. Rheol., 29, 739 (1985).

7. B. Hausnerova, T. Honek, P. Saha, and T. Kitano, J. Polym.Mater., 21, 1 (2004).

8. H.A. Barnes and K. Walters, Rheol. Acta., 24, 323 (1985).

9. H.A. Barnes, J. Non-Newtonian Fluid Mech., 81, 133 (1999).

10. W.H. Herschel and R. Bulkley, Proc. ASTM, 26, 621

(1926).

11. D.M. Husband and N. Aksel, J. Rheol., 37, 215 (1993).

12. B. Hausnerova, P. Saha, J. Kubat, T. Kitano, and J. Becker,

J. Polym. Eng., 20, 237 (2000).

13. B. Hausnerova, P. Saha, and J. Kubat, Int. Polym. Proc., 14,254 (1999).

14. T. Honek, B. Hausnerova, and P. Saha, Appl. Rheol., 12, 72(2002).

15. U. Yilmazer, C.G. Gogos, and D.M. Kalyon, Polym. Comp.,10, 242 (1989).

16. P. Yaras, D.M. Kalyon, and U. Yilmazer, Rheol. Acta., 33, 48(1994).

17. S. Laddha, C. Wu, S. Vallury, G. Lingam, S. Lee, K. Sim-

mons, P. Thomas, B. Levenfeld, A. Varez, S.J. Park, S.

Ahn, R.M. German, and S.V. Atre, PIM Int., 3, 64 (2009).

18. H. Persson, B. Hausnerova, L. Nyborg, and M. Rigdahl, Int.Polym. Proc., 24, 206 (2009).

FIG. 8. Scanning electron micrographs of alumina feedstock after molding (a), water debinding (b), and

solvent/thermal debinding (c).

DOI 10.1002/pen POLYMER ENGINEERING AND SCIENCE—-2011 1381

Page 7: Optimization of powder injection molding of feedstock based on aluminum oxide and multicomponent water-soluble polymer binder

19. C.D. Greene and D.F. Heaney, Mater. Des., 28, 95 (2007).

20. W.C.J. Wei, R.Y. Wu, and S.J. Ho, J. Eur. Ceram. Soc., 20,1301 (2000).

21. L.J. Prakash, PIM Int., 1, 17 (2007).

22. R.M. German, PIM Int., 2, 45 (2008).

23. T. Moritz and R. Lenk, PIM Int., 3, 23 (2009).

24. W.W. Yang, K.Y. Yang, M.C. Wang, and M.H. Hon,

Ceram. Int., 29, 745 (2003).

25. R.M. German, Powder Injection Molding, Metal Powder

Industries Federation, Princeton (1990).

26. S.T. Lin and R.M. German, J. Mater. Sci., 29, 5207 (1994).

27. T.Y. Chan and S.T. Lin, J. Am. Ceram. Soc., 78, 2746 (1995).

28. W.J. Tseng, Mater. Sci. Eng. A., 289, 116 (2000).

29. R.L. Hoffman, T. Soc. Rheol., 16, 155 (1972).

30. J.B. Jansma and S. Qutubuddin, J. Rheol., 39, 161 (1995).

31. H.A Barnes, J. Non-Newtonian Fluid Mech., 56, 221 (1995).

32. G.P. Roberts, H.A. Barnes, and P. Carew, Chem. Eng. Sci.,56, 5617 (2001).

33. P. Lomellini and D. Ferri, ‘Polymer Melt Rheology Versus

Shear Rate or Versus Shear Stress?’ in Proc. 13th Int. Con-gress on Rheology, D.M. Binding, N.E. Hudson, J. Mewis,

J.-M. Piau, C.J.S. Petrie, P. Townsend, M.H. Wagner and K.

Walters, Eds., Cambridge (UK), 20–25 Aug 2000, Vol. 1,

pp. 118–120, Published by the British Society of Rheology,

Glasgow 2000.

34. P. Filip, J. David, and R. Pivokonsky, Acta Technica CSAV,51, 349 (2006).

35. J. David and P. Filip, Appl. Rheol., 14, 82 (2004).

36. D.M. Liu and W.J. Tseng, Ceram. Int., 25, 529 (1999).

37. P. Thomas-Vielma, A. Cervera, B. Levenfeld, and A. Varez,

J. Eur. Ceram. Soc., 28, 763 (2008).

38. M. Trunec and J. Cihlar, J. Eur. Ceram. Soc., 17, 203

(1997).

39. V.A. Krauss, A.A.M. Oliviera, A.N. Klein, H.A. Al-Qure-

shi, and M.C.J. Fredel, Mater. Process. Tech., 182, 268

(2007).

1382 POLYMER ENGINEERING AND SCIENCE—-2011 DOI 10.1002/pen