Cold compaction of iron powders—relations between powder

morphology and mechanical properties

Part I: Powder preparation and compaction

D. Poquillon*, J. Lemaitre, V. Baco-Carles, Ph. Tailhades, J. Lacaze

CIRIMAT, UMR CNRS/UPS/INPT 5085, ENSIACET, 31077 Toulouse, France

Received 1 May 2001; received in revised form 1 January 2002; accepted 30 January 2002

Abstract

The effect of morphology of iron powders on their compaction behaviour has been studied in the case of compacts with a final relative

density below 0.8. Two powders, one with spherical grains and the other with spongy grains, have been synthetized in order to prepare green

compacts under pressures ranging from 100 to 350 MPa. The compaction behaviour of both powders has been experimentally described by

the relation between the applied compaction pressure and the relative density of the material. Different stages have been identified at

increasing compaction pressure: (i) fully elastic behaviour (Stage I), (ii) particle sliding (Stage II), (iii) particle irreversible deformation (Stage

III). Spongy powder-based compacts achieved greater density at the same compaction pressure than spherical powder-based compacts.

Models are proposed to describe the compaction behaviour of both powders. D 2002 Elsevier Science B.V. All rights reserved.

Keywords: Iron powder; Compaction; Particles morphology; Oxalate precursor; Compaction model

1. Introduction

Powder metallurgy (PM) is a highly developed method

of manufacturing reliable ferrous and non-ferrous parts. The

PM process is cost-effective because it minimises machin-

ing, produces good surface finish and maintains close

dimensional tolerances. Generally, the PM-forming process

includes cold compaction, sintering and finishing. The

successful compaction of the powder is the first step of

the process, which is also the last one in the case of green

compacts. The compacted part combines high porosity (20–

35%) and reasonable strength. This porosity can be a means

to lower structural weight in aeronautical applications (e.g.

shape magnets for sensors and motors) or to create oil-

retaining bearings, which are often referred to as self-

lubricant bearings.

Mechanical properties of materials prepared with such a

process are strongly influenced by the characteristics of the

powder (particle size and size distribution, particle shape,

structure and surface characteristics) and by the way com-

paction is carried out. Compaction induces very complex

states of stress in the powder. Lubrication, powder height in

the die, compaction rate and stress triaxiality strongly

influence the compaction process and, thus, affect the

properties of the products obtained [1]. Accordingly, opti-

misation of mechanical properties of green compacts

requires a better knowledge of the relation between powder

characteristics and mechanical behaviour of the material

during compaction (grain sliding and rearrangement, grain

deformation). Furthermore, the mechanisms for the devel-

opment of green strength during compaction are generated

by two phenomena, namely (i) particle sliding and inter-

locking and (ii) plastic deformation [2]. The initial stage of

compaction leads to rearrangement of the powder from a

loose array to close packing. As the pressure increases, the

contact area between the grains increases and particles

undergo extensive plastic deformation [3]. Green strength

has been found to be related to the contact area between

particles [3].

Generally, PM studies take into account particle size,

but neglect morphological parameters of powder grains as

they are more complex to define, measure and modify.

Grain size can be changed by simple grinding, but the

shape of the particles and their surface aspect results from

the chemical processes of powder preparation. In a pre-

vious study [4], it was shown that compacts of commercial

spherical or irregular iron powders have green strength up

0032-5910/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.

PII: S0032 -5910 (02 )00034 -7

* Corresponding author.

E-mail address: dpoquillon@ensct.fr (D. Poquillon).

www.elsevier.com/locate/powtec

Powder Technology 126 (2002) 65–74

to five times lower than those prepared from iron powder

with a spongy rough surface. To get a better understanding

of this phenomenon, we have chosen to study the influence

of such morphological parameters. Thus, two iron powders

were prepared to study the correlation between powder

morphology and mechanical properties (compaction and

three-point bending tests). The results obtained are pre-

sented in this paper and in a companion paper [5] devoted

to bending tests. In this first paper, the preparation of small

particles of metal with two different morphologies is

detailed. Compaction tests are then carried out on both

powders and the results are discussed in terms of micro-

structural effects. A model is proposed to simulate the

experimental curves relating the density of the compacts to

the compaction pressure. It is worth emphasising that most

of the previous investigations focused on the high relative

density range ( > 0.85). In this study, we were interested in

compacts with significant porosity, i.e. of relative density

under about 0.7.

2. Experimental data

2.1. Preparation and characteristics of the powders

Iron powders were prepared by reduction [6] of two

FeC2O4 � 2H2O oxalate particles having different morpho-

logical characteristics. These differences resulted from the

precipitation methods used. To get spherical shaped oxalate

particles (Fig. 1a), ferrous iron was precipitated by ammo-

nium oxalate in aqueous solution. The second type of

FeC2O4 � 2H2O oxalate was obtained with the reaction of

ferrous iron with oxalic acid in hydro-alcoholic medium

(Fig. 1b).

The preparation of the iron powders was then achieved

through direct decomposition of the oxalic precursors un-

der pure H2 (99.995%) in a specially designed furnace

(Type: ELTI 1731/S, volume close to 110 l). The rate of

gas flow was 580 l h� 1. The temperature was first held at

150 jC for 3 h to eliminate H2O from the ferrous oxalate

FeC2O4 � 2H2O. Afterwards, thermal decomposition of

ferrous oxalate with spherical morphology was performed

at about 500 jC. Fig. 2a shows the typical shape of the

powder thus obtained. This powder is made of more or less

spherical grains. This powder will be denoted powder S.

The average particle size is 15 Am. It can be seen in Fig.

2a that each grain is an agglomerate of small crystallites

with a typical size of about 0.5–1 Am. A processing

temperature higher than 500 jC has been tried, but resulted

in the growth of individual crystallites, and the final

roughness of the spherical powder particles decreased. The

thermal decomposition of the acicular ferrous oxalate was

performed also under pure H2, for a processing tempera-Fig. 1. Photograph of the oxalate precursors: (a) spherical ferrous oxalate

precursor; (b) Acicular ferrous oxalate precursor.

Fig. 2. Photograph of the iron powder: (a) S, spherical morphology; (b) E,

spongy morphology.

D. Poquillon et al. / Powder Technology 126 (2002) 65–7466

ture of about 600 jC. The resulting powder, denoted

powder E, appeared like agglomerates of tangled acicular

metallic crystallites. Fig. 2b shows a typical micrograph of

powder E. The grains are more irregular in shape than

those of powder S; they are rough and spongy and are

composed of crystallites of about 1–2 Am in size. The

average grain size is about 30 Am.

Various experiments were performed to characterise the

two types of powder after their preparation. The specific

surface area of the powders was measured following the

Brunauer, Emmet and Teller method [7] of adsorption of a

single layer of N2 at the surface of the powder grains. This

was made with a MICROMERITICS FLOWSORB II 2300

device. The specific surface area obtained for powder S was

2 m2 g� 1, whereas the specific surface area measured for

powder E was 0.6 m2 g� 1. Also, the apparent density of the

powders was evaluated by measuring the mass of powder

filling a container of known volume. This measurement was

performed without vibration of the powder. The mass of

powder divided by this volume defines the apparent density

of the powder. The values found were 1.173 g cm� 3 for

powder E and 1.169 g cm � 3 for powder S. The actual

density of the powders was evaluated through a He pycn-

ometer MICROMERITICS ACCUPYC 1330. The meas-

ured values were 7.73 g cm� 3 for powder E and 7.64 g

cm � 3 for powder S. These values are to be compared to the

density of pure body-centered cubic Fe which is 7.86 g

cm � 3. This difference is mainly due to surface oxidation of

the powder. This point is evidence by thermogravimetric

analysis detailed in the next paragraph.

The difference between the estimated apparent density of

the powders and that of iron was investigated by oxidising

the powders to evaluate their metallic Fe content. This was

performed by thermogravimetric analysis (TGA) on 20 mg

samples with an automatic thermobalance SETARAM

TGA92 under air during one heating up to 1000 jC at

3 jC/h. The iron content was calculated according to the

following oxidation reaction:

ð1� xÞFeþ xFe3O4 þ3

4� 1

2x

� �O2Z

1

2þ x

� �Fe2O3:

The content (in weight) of metallic Fe obtained with this

method was about 96.3% for powder E and 93.4% for

powder S. Assuming that the remaining mass is due to

Fe3O4, we obtained the following value of density 7.71 g

cm � 3 for powder E and 7.60 g cm� 3 for powder S. These

TGA results are found to be in agreement with the previous

estimates of the density (He pycnometer); thus, closed

porosity can be neglected for both powders.

The morphology of the powders is thought to be

responsible for the difference in Fe3O4 content because

the specific surface area of powder S was much higher than

that of powder E as indicated above. The larger average

size of both powder particles and constitutive crystallites of

powder E can explain its smaller specific surface area with

respect to powder S even if the grains of powder S

appeared more compact as illustrated in Fig. 2. These

differences in size are coherent with the difference in

specific surface area previously noted. Thus, powder S

with a larger specific surface area is more reactive than

powder E during storage in air. Consequently, powder S

presents a larger ability to develop Fe3O4 at its surface.

Finally, it was thought useful to have a quantitative

parameter to compare the ‘‘roughness’’ of the two mor-

phologies. It was found that the angle of repose for the

powders could conveniently provide such an information.

This quantity is frequently used in soil mechanics to

quantify the ability of sand or clay to flow. Angles of

repose below 45j indicate a good ability of the material to

flow and, as a consequence, the use of such material as an

embankment is not adequate. A triplicate test of the angle

of repose was performed under the standard AFNOR NF

A95-113 to compare the sliding and rearrangement facili-

ties of the powders. Powder S exhibited a 38F 0.5j angle

of repose, indicating a good ability of the grains to slide

and rearrange, whereas powder E exhibited a 52F 0.5jangle of repose, indicating that the irregular shape of the

grains should obviate easy interparticle sliding.

2.2. Compaction tests

A uniaxial SPECAC press was used to carry out the

compaction tests. As only the lower punch could move,

single-end pressing was carried out. The load imposed

was measured, but not the load transmitted to the upper

punch. This is unfortunate as this would have allowed us

to quantify the wall friction effects. Furthermore, the

elastic compliance of the apparatus was not taken into

account. Two different types of green compacts were

produced: cylindrical specimens and bars. Both types of

specimens were pressed at room temperature. The final

compaction pressure was varied between 100 and 350

MPa. Zinc stearate in acetone solution was used as die

lubricant. Compactions were performed in air, i.e. no

vacuum or controlled atmosphere was imposed during

compaction. The bars were designed to be used for the

three-point bending tests. Those tests are detailed and

described in a companion paper [5], but the correspond-

ing density data will be detailed in the present paper. Iron

powders were compacted in a hardened steel rectangular

die, which was 32.36 mm in length and 13.28 mm in

width. These dimensions are comparable to those of

normalised rupture tests (MPIF standard 15). Each speci-

men corresponded to about 6 g of iron powder. The

uncertainty about their final density due to the uncertain-

ties on dimensions and weight is 0.5%. The die used to

compact the cylindrical specimens was 10 mm in diam-

eter. The initial surface finish of the die walls and of the

punch was below 1 Am. Each cylindrical specimen

required less than 1 g of powder. The compaction of

the cylindrical specimens was followed with a MITU-

TOYO comparator.

D. Poquillon et al. / Powder Technology 126 (2002) 65–74 67

At the end of each compaction, the final pressure was

maintained for 2 min before the ejection of the green

compact. After compaction, the compacts were weighed

with an uncertainty of F 0.001 g and their dimensions

measured with a micrometer calliper to F 0.01 mm. The

relative density of the compacts was then estimated from

these values, with a total uncertainty of 0.5% for the bars

and 2% for the cylinders. From the final height of the

compact and from the displacement of the lower punch in

the case of the cylinders, it was, thus, possible to evaluate

the curve giving the applied pressure vs. density. It is worth

noting that, because of this method, the uncertainty in the

density is larger at the lower applied pressures. Every

experiment has been repeated two or three times under a

strict procedure. Although the lower punch displacement

was manual, it was found that the compaction curves were

highly reproducible.

Table 1 gives the main characteristics of the green

compact bars, whereas Table 2 lists those of the cylindrical

specimens. In these tables, the relative density is defined as

the apparent density divided by the actual density measured

with the He pycnometrer. Two main tendencies were noted:

(i) the cylindrical samples achieved generally lower relative

density than bars; (ii) powder E had better compressibility

than powder S. It is well known that die wall friction

depends on the compact height and on the die geometry

even when a lubricant is used [1,8,9]. The difference in

compressibility observed during compaction between bars

and cylindrical specimens could have been investigated by

an analysis of the transmitted load which could not be made

with the rig employed during this study.

Fig. 3 compares the green densities achieved with the

final compaction pressure for the bars and cylinders.

Although powder S had a slightly higher density than

powder E before compaction and also had a lower angle of

repose, it is observed that powder S had a slightly lower

compressibility in the range of pressures studied. This could

mean that the morphological differences between the two

powders affect compaction only at low pressures, i.e. when

grain sliding controls the changes in density. Fig. 3 shows

also that the density is a linear function of the compaction

pressure for relative density lower than about 80%. This

result is in agreement with previous data obtained for copper

[10], iron [11] and aluminium [12] powders, which have

been also plotted in Fig. 3.

The compaction response of the powders is shown in

Fig. 4, where it is observed that all the curves show the

same features. The response of the powders during single-

end pressing can be divided into three different stages,

whatever the powder morphology (E or S) is. During Stage

I (0 to 15 MPa), densification increases very slowly; then in

Stage II, densification increases abruptly until the start of

the Stage III where the densification rate decreases. The

limit between Stages I and II is the end of linearity of the

first part of the curve. This limit seems to depend on the

initial relative density of the powder. The denser the initial

Table 1

Green compact (bar specimen) characteristics

Compact

label

Compaction

pressure (MPa)

Relative

density (%)

Weight

(g)

Height

(mm)

S100.1 100 48.86 6.010 3.75

S100.2 100 48.65 5.919 3.71

S100.3 100 48.65 5.915 3.70

S150 150 54.21 6.113 3.43

S200.1 200 58.66 6.108 3.17

S200.2 200 58.86 6.321 3.27

S250 250 62.10 6.227 3.05

S290.1 290 65.72 6.232 2.89

S290.2 290 65.57 6.307 2.93

E100.1 100 54.96 5.959 3.27

E100.2 100 52.02 5.991 3.47

E150 150 59.95 6.192 3.11

E200.1 200 59.86 6.554 3.29

E200.2 200 59.75 6.507 3.28

E250 250 64.44 6.265 2.93

E290.1 290 71.29 13.017 5.50

E290.2 290 71.44 13.056 5.50

Table 2

Green compact (cylindrical specimens) characteristics

Compact

label

Compaction

pressure (MPa)

Relative

density (%)

Weight

(g)

Height

(mm)

PS200.1 200 55.7 0.694 2.07

PS200.2 200 55.6 0.694 2.08

PS350.1 350 64.1 0.815 2.12

PS350.2 350 62.4 0.795 2.10

PE200.1 200 60.3 0.595 1.62

PE200.2 200 59.8 0.595 1.64

PE350 350 68.4 0.802 1.93

Fig. 3. Relative density of compacts vs. the final pressure applied.

Comparison of the results obtained during this study with results obtained

for copper [10], iron [11] and aluminium [12] powders.

D. Poquillon et al. / Powder Technology 126 (2002) 65–7468

powder is, the lower the limit between Stages I and II.

However, as the initial density of powder E is on average

lower than that of powder S, it is difficult to separate the

effects of morphology from those of the initial density.

The end of Stage II and, thus, the beginning of Stage III, is

less easy to determine on the curves. We fixed it at

P1 = 35 MPa, where densification rate has clearly started

to decrease.

Compaction of iron powders during Stages I and II is

comparable to soil compaction, e.g. clays [13–16]. This

type of behaviour is due to elastic deformation of the

particle heap during Stage I and to particle rearrangement

by interparticle sliding during Stage II. Stage I is totally

reversible. While intense irreversible particle sliding and

rearrangement occur during Stage II, it is assumed to

correspond to negligible plastic deformation and energy is

mainly dissipated by interparticle friction. The transition

from Stage I to Stage II is related to an abrupt increase of

densification. Stage III starts when particle interlocking

generates plastic deformation, which is first localised at

the contact areas between particles [2,8].

SEM observations of the surface of the green compacts

were performed in order to compare the morphology of the

grains after compaction at pressures between 100 and 290

MPa. Fig. 5a and b provides a comparison of the micro-

structure of powder S after pressing at 100 and 290 MPa,

respectively. Interparticle porosity seemed to be reduced and

the contact area between grains to be increased with

increased pressure. In Fig. 5c, it is seen, however, that the

shape of the grains is still identifiable. Also, it seemed that

the decrease of the porosity with increasing compaction

pressure proceeded through localised plastic deformation at

the contact zone between grains. Fig. 5d shows the micro-

structure of powder E after compaction at 290 MPa. In this

case, it appears impossible to identify the initial powder

grains because of large interlocking. The contact zones

between grains were less localised and involved more

crystallites than in the case of powder S. Observations made

on compacts pressed at various pressures showed that

particle interlocking was dominant and increased with

compaction pressure in the case of powder E. These differ-

ences between powder E and powder S can be correlated

with the higher density obtained for the former one than for

the latter at a given compaction pressure. Moreover, the

effects of wall friction during compaction were evidenced

for the specimens compacted at 290 MPa as experimental

observations indicated noticeable shear stress near the wall.

Intense and localised plastic deformations were observed in

the first 150 Am, under the surface in areas in contact with

either the lateral wall or the punches.

3. Numerical simulation of the compaction tests

The three-compaction stages depicted in the preceding

part have been described by means of two different mod-

Fig. 4. Relative density of the compacts vs. the applied compaction pressure.

D. Poquillon et al. / Powder Technology 126 (2002) 65–74 69

els: the CAM-CLAY model for Stages I and II, and a

power law model for Stage III. The CAM-CLAY model

was initially devoted to soils science and developed to

simulate the behaviour of clays, accounting for voids and

water content [13–16]. This model is intended to consider

elastic deformation as well as sliding of the grains between

each other. The CAM-CLAY approach has an excellent

ability to be transposed from soil mechanics to cold

compaction of spherical commercial iron powders [17–19].

The power law used in the model selected to describe Stage

III is supposed to take into account the plastic deformation of

the contact area between grains.

The parameters entering the constitutive equations of

both models were identified by means of a phenomenolog-

ical approach, i.e. by looking for the simplest expressions

allowing for fitting the experimental curves. The main

features of these models as well as the results for the three

successive stages of the compaction are described below.

As no quantitative experimental data was available to

access die wall friction, calculations were performed assum-

ing uniform compression in the die. Accordingly, the stress

state in the sample was assumed to be homogeneous with

axial symmetry.

The tensorial equations of the CAM-CLAY model have

been described previously [16]. As only homogeneous

uniaxial compression is considered, the mechanical equa-

tions can be simplified. The only possible deformation

occurs along the compression axis (let it be denoted z) so

that the strain tensor ¯e is constant throughout the sample

thickness. It has the following expression in axi-symmetric

coordinates: ¯e ¼ ezzez � ez , where ezz is the deformation

along the z-axis, ez is the unit vector in the axial direction

and � stands for the tensorial product. If q0 is the initial

relative density and q the current relative density of the

compact, the strain evolution during compaction is simply

given by: ezz=(q0/q)� 1. It is common practice in soil

mechanics to replace the relative density by the void ratio,

e, for materials with high porosity. This parameter is

defined as e=(1� q)/q and was adopted for the present

analysis. By substitution, ezz= q0(e� e0), where e0 stands

for the initial void ratio. In the following analysis, the

pressures are to be expressed in MPa.

The response of the material according to the CAM-

CLAY model depends on five parameters. Two parameters

are related to the initial state of the material, namely the

initial voids ratio, e0, and the yield pressure, Pi0. The

three other parameters are material intrinsic characteristics.

They are denoted M, j (the loading–unloading slope) and

k (the loading slope on the yield surface). The yield

surface is a function of the mean stress, p, and of the

deviatoric stress, q. Its shape is elliptical in a p–q graph.

It presents a maximum at the point ( pm, qm). M is the

Fig. 5. Observation of the lateral surface of compacts: (a) Powder S, final pressure 100 MPa; (b) Powder S, final pressure 290 MPa; (c) Powder S, final pressure

290 MPa; (d) Powder E, final pressure 290 MPa.

D. Poquillon et al. / Powder Technology 126 (2002) 65–7470

slope of the critical state line and is defined by M= qm/pm.In the CAM-CLAY model, this value is constant and

independent of the yield pressure, Pi0, reached. As only

uniaxial tests had been performed, triaxiality aspects of the

behaviour of the powders were not investigated. As a

consequence, the M parameter was not evaluated.

Fig. 6 shows schematically the response of the material

according to the CAM-CLAY model. It is seen that the

void ratio first decreases slowly from e0 as the compaction

pressure increases. During this first stage, only elastic

deformation of the grains is involved. Thus, during unload-

ing, the material should follow the same curve. The sur-

face roughness of the grains prevents interparticle sliding

and reversible strain comes from skeletal elastic deforma-

tion of the particle arrangement. As the yield pressure, Pi0,

is reached, the rate of change in void ratio increases

abruptly. This is intended to reproduce the sliding of the

grains between each other during Stage II. In the case of

unloading, the compacts have an elastic response with the

same slope as that during Stage I. However, the overall

deformation due to sliding is not reversible as illustrated in

Fig. 6.

To describe Stage I, it was found convenient to relate e

and log10(PC) according to the linear relation: e = e0�jlog10(PC), where PC is the compaction pressure. More-

over, the value j = 0.2 was found to allow fitting of the

Fig. 6. Schematic representation of the evolution of the void ratio with the

compaction pressure according to the CAM-CLAY model. Definition of the

slopes j and k.

Fig. 7. Predicted evolution of the void ratio with the applied compaction pressure during stages I and II (curves with small symbols). Comparison to

experimental measurements (large symbols). The interrupted line represents the limit between stages I and II.

D. Poquillon et al. / Powder Technology 126 (2002) 65–74 71

curves whatever the initial density and the shape of the

powder grains. A comparison between predicted and meas-

ured curves is shown in Fig. 7. The end of Stage I

corresponds to a compaction pressure equal to the yield

pressure, Pi0. By means of an iteration procedure, it was

found convenient to set Pi0 = 12 + e0 for all cases. This limit

between Stages I and II is also drawn on Fig. 7. It is clearly

seen that during initial compaction, the start of grain sliding

depends either on the initial void ratio or on the shape of the

grains, or else on both of them. Unfortunately, the present

results were not sufficient to determine the actual effect of

each of these quantities.

To describe Stage II, a linear relation between e and

log10(PC) was considered as for Stage I. Moreover, it was

thought convenient to introduce the initial void ratio e0 as

Stage I did not lead to significant compaction. It was finally

found that most of the features of the curves could be

satisfactorily reproduced by considering the following

expression for the irreversible compaction:

e ¼ eðPi0Þ � klog10Pc

Pi0

� �,

with:

k ¼ e0 � 3:5

1:48� log10ðPi0Þ:

It is worth noting that the slope k depends on e0 in such a

way that particle sliding increases as the initial void ratio is

increased. The critical value e0 = 3.5 could correspond to

compacts initially so dense that compaction should start

directly in Stage III. Moreover, it should be emphasised that

the selected expressions of e and k could be chosen as

identical for both powders. This may come from the fact that

powder S had initially a relative density higher than powder

E. As it was mentioned previously, there is not enough

experimental information to separate the effects of particle

size and those of particle shape. However, Fig. 7 shows that

a quite good agreement between simulations and experi-

mental data could be obtained for Stage II also.

The beginning of Stage III was fixed at P1 = 35 MPa,

where densification rate has clearly started to decrease.

Modelling of Stage III was achieved with the following

power law q = q1 + a(P�P1)n, where q1 and P1 are the

relative density and the compaction pressure, respectively,

Fig. 8. Evolution of the relative density of the compacts during stage III. Comparison of the predicted evolution (dotted lines) with measurements

(symbols).

D. Poquillon et al. / Powder Technology 126 (2002) 65–7472

at the end of Stage II. a characterises the material’s ability

to be deformed, while n is equivalent to a hardening

coefficient. Both parameters were considered as constants

to be fitted to experimental data. The best fit between

calculated and experimental curves showed that the value

of n could be the same for powders S and E. This would

mean that the mechanism for dissipation of the energy

related to plastic deformation should be the same for both

powders. It was, thus, decided to use one single value of n

for this model parameter, which was set equal to 0.55 for

all simulations.

Given the value of n, the best value for a was found to be

1.69� 10� 2 MPa � 0.55 for powder S and 1.94� 10� 2

MPa � 0.55 for powder E. This difference relates to the

higher densification achieved with powder E than with

powder S. Fig. 8 compares the calculated and experimental

curves. It is seen that a satisfactory agreement is obtained

for the whole range of pressure, which means that this

simple model is able to reproduce most of compaction

process when simulated as a uniaxial compression.

4. Conclusion

The effect of morphology of iron powder on their

compaction behaviour has been studied in the case of

compacts with final porosity higher than 0.3. Two pow-

ders, one with spherical grains and the other with spongy

grains, have been synthetized in order to prepare green

compacts under pressures ranging from 100 to 350 MPa. It

was found that the density of the compacts varied linearly

with the compaction pressure in the range investigated,

with a higher ability to be compacted for the spongy

powder than for the spherical one. As the sliding of the

grains is easier for the spherical powder, such a difference

relates to a larger plastic deformation of the spongy

powder particle. This was clearly evidenced by SEM

observations of the compact surface. The compaction

behaviour of both powders has been experimentally

described by the relation between the applied compaction

pressure and the relative density of the material. The

curves show that compaction proceeds along three succes-

sive stages involving elastic deformation of the grains

(Stage I), interparticle sliding with elastic deformation

(Stage II) and, finally, plastic deformation of the grains

(Stage III). It was observed that particle sliding and

rearrangement occurs at higher pressure in the case of

spongy powder-based compacts. These green compacts

achieved a slightly higher density at a given compaction

pressure than spherical powder-based compacts. These

observations are in agreement with the higher angle of

repose and higher capability for plastic deformation of the

spongy powder than the spherical one.

The whole compaction process has been simulated with

two models, the CAM-CLAY model for Stages I and II and

a power law model for Stage III. It is worth emphasising

that calculations were made with the same values of the

model parameters for both powders, apart from a, whichcharacterises the deformation capability of the powder.

Quantitative differences were, thus, accounted for only

through the initial void ratio which was higher for the

spongy powder than for the spherical one. A better

description of Stage III would rely on a characterisation

of the evolution of the porosity during compaction and,

particularly, in separating interparticle and intraparticle

voids. Ozkan and Briscoe [9] have performed such a study

by means of laser profilometry on powders of a alumina.

This could be applied to the powders investigated in this

study. It could also be of great interest to confirm the

validity of the three successive stages assumed for model-

ling. For example, the beginning of plastic deformation

during compaction could be checked by acoustic emission

[20].

The behaviour of powders during uniaxial compaction

did not allow us to identify the yield surface of the CAM-

CLAY model (coefficient M ), which is effected by inter-

locking and grain sliding. It may be guessed that powders E

and S would have a more different behaviour during shear

test (or isostatic compaction) than during uniaxial compac-

tion. This could be investigated by means of the exper-

imental rig developed by Pavier and Doremus [21], who

showed shear stress to be more favourable to compaction

than hydrostatic stress. Knowledge of the yield surface of

the powders would allow further description of be the

transition from grain sliding to particle interlocking and

the beginning of plastic deformation. The heterogeneities in

the compacts could then be simulated.

References

[1] B.J. Briscoe, S.L. Rough, The effects of wall friction in powder com-

paction, Colloids and Surfaces, A: Physicochemical and Engineering

Aspects 137 (1998) 103–116.

[2] J.A. Lund, Origins of green strength in iron P/M compacts, Interna-

tional Journal of Powder Metallurgy & Powder Technology 18 (2)

(1982) 117–127.

[3] I.-M. Moon, J.-S. Choi, Dependence of green strength on contact area

between powder particles for spherical copper powder compacts,

Powder Metallurgy 28 (1) (1985) 21–28.

[4] V. Baco-Carles, J. Lemaıtre, D. Poquillon, Ph. Tailhades, J. Lacaze, A.

Rousset, Correlations entre la morphologie de poudres de fer metal-

lique, leur aptitude a la compaction et la resistance mecanique en

flexion de compacts crus, Colloque ‘‘Science et Technologie des Pou-

dres’’, Nancy, France, 3–5 Avril, 2001, For publication in Recents

Progres en Genie des Procedes.

[5] D. Poquillon, V. Baco-Carles, Ph. Tailhades, E. Andrieu, Cold com-

paction of iron powders—relations between powder morphology and

mechanical properties: Part II : Bending tests, results and finite ele-

ment analysis, Powder Technology, companion paper.

[6] V. Carles, P. Alphonse, Ph. Tailhades, A. Rousset, Study of thermal

decomposition of FeC2O4 � 2H2O under hydrogen, Thermochimica

Acta 334 (1999) 107–113.

[7] S. Brunauer, P.H. Emmet, E. Teller, The adsorption of gases in

multimolecular layers, Journal of the American Chemical Society

60 (1938) 39.

D. Poquillon et al. / Powder Technology 126 (2002) 65–74 73

[8] H. Lippmann, V. Mannl, R. Iankov, O. Beer, Numerical simulation of

density distribution during compaction iron powders, Archive of Ap-

plied Mechanic 67 (1991) 191–199.

[9] N. Ozkan, B.J. Briscoe, The surface topography of compacted ag-

glomerates; a means to optimize compaction conditions, Powder

Technology 86 (1996) 201–207.

[10] I.-H. Moon, K.-H. Kim, Relationship between compacting pressure,

green density, and green strength of copper powder compacts, Powder

Metallurgy 27 (2) (1984) 80–84.

[11] G.W. Halldin, K.T. Yung, T.H. Tsai, A new analysis of compressibility

and green strength of metal powder, Progress in Powder Metallurgy 37

(1982) 383–393.

[12] C.G. Tu, I.T. Chen, The pressure–density– strength relationships of

aluminium alloy powder compacts, Journal of the Chinese Institute of

Engineers 13 (3) (1990) 259–271.

[13] K.E. Roscoe, A.N. Schofield, C.P. Wroth, On the yielding of soils,

Geotechnique 9 (1968) 71–79.

[14] A.N. Schofield, C.P. Wroth, Critical State Soil Mechanics, McGraw-

Hill, New York, 1968.

[15] I.B. Burland, K.E. Roscoe, On the generalized stress–strain behaviour

of wet clay, Engineering Plasticity, Heyman-Leckie, Cambridge,

1968.

[16] F. Darve, Manuel de rheologie des geomateriaux, Ed. by Presses de

l’ENPC Paris, 1987.

[17] E. Pavier, P. Doremus, Mechanical behaviour of lubrificated iron

powder, Metal Powder Report 52 (10) (1997) 40–47.

[18] PM-Modnet-Computer-Modelling-Group, Comparison of computer

models representing powder compaction process. Powder Metallurgy,

Powder Metallurgy 42 (4) (1999) 301–311.

[19] X.K. Sun, S.J. Chen, J.Z. Xu, L.D. Zhen, K.T. Kim, Analysis of cold

compaction densification behaviour of metal powder, Materials Sci-

ences & Engineering, A: Structural Materials: Properties, Microstruc-

ture and Processing 267 (1999) 43–49.

[20] N. Parida, B.B. Pani, B.R. Kumar, Acoustic emission assisted com-

paction studies on iron, iron–aluminium and iron-cast iron powders,

Scripta Materialia 37 (11) (1997) 1659–1663.

[21] E. Pavier, P. Doremus, Triaxial characterisation of iron powder

behaviour, Powder Metallurgy 42 (4) (1999) 345–352.

D. Poquillon et al. / Powder Technology 126 (2002) 65–7474