Vous êtes sur la page 1sur 6

J. Mater. Sci. Technol., Vol.24 No.

2, 2008 227

A Comparison between Direct and Indirect Laser Sintering of Metals


Montasser M.Dewidar1,2)† , Jae-Kyoo LIM3) and K.W.Dalgarno4)
1) Department of Mechanical Design and Materials, High Institute of Energy, South Valley University, Aswan, Egypt
2) Department of Mechanical Engineering, Chonbuk National University, Chonju, Korea
3) Automobile Hi-Technology Research Institute, Faculty of Mechanical & Aerospace System Engineering, Chonbuk
National University, Duckjin 1-664-14, Jeonju, JB 561-756, South Korea
4) School of Mechanical and Systems Engineering, Stephenson Building, Claremont Road, Newcastle University,
Newcastle upon Tyne NE1 7RU, United Kingdom
[Manuscript received August 16, 2006, in revised form January 28, 2007]

Layer manufacture technologies are gaining increasing attention in the manufacturing for the production of
polymer mould tooling. Layer manufacture techniques can be used in this potential manufacturing area to
produce tooling either indirectly or directly, and powder metal based layer manufacture systems are considered
as an effective way of producing rapid tooling. Mechanical properties and accuracy are critical for tooling. This
paper reports the results of an experimental study examining the potential of layer manufacturing processes to
deliver production tooling for polymer manufacture. A comparison between indirectly selective laser sintering
and directly selective laser sintering to provide the tooling was reported. Three main areas were addressed
during the study: mechanical strength, accuracy, and build rate. Overviews of the results from the studies
were presented.
KEY WORDS: Selective laser sintering; Layer manufacture; Mould tooling; Rapid tooling

1. Introduction

Rapid prototyping or layer manufacturing tech-


nologies were first introduced in 1989 as a design vi-
sualization tool and have revolutionized the products
way of designing and manufacturing today. Proto-
typing equipment used to make plastic models for vi-
sualization 10 years ago is now used to make invest-
ment casting patterns, sand casting cores and molds,
patterns for secondary processes, functional proto-
types, and mould inserts for injection molding. Se-
lective laser sintering (SLS) is one of the layer man-
ufacture technology that can produce ceramic and
metallic tools. SLS is an advanced rapid prototyp- Fig.1 Schematic of SLS process
ing technology, which can shorten the manufacturing
time cycle, hence reducing the production cost and
increasing competitiveness[1–5] . In the SLS process, green part is subsequently post-processed by sinter-
a three-dimensional object is created layer-by-layer ing and infiltration to produce a high-density metal
from the bottom to the top from heat fusible pow- part. Infiltration is a process that wicks a molten
dered materials with heat generated from a laser. A phase into the open pores of porous structure by cy-
thin layer of heat fusible powdered material is deliv- cling to a temperature between the melting points of
ered onto the top of the build cylinder by a roller (see the infiltrant and the skeleton.
Fig.1). A heat generating laser beam traces across Direct SLS processes use metal powders directly
this layer, sintering specific areas according to the in- without any binder. Direct SLS eliminates the need
structions of the computer-aided design (CAD) file. for post-processing. Ongoing material and process
The platform lowers slightly and a new fresh thin layer developments have increased the productivity of the
of powder is delivered. Then, the laser scans selected systems and the quality of the resulting parts. SLS
areas of this layer, which bond to the previous layer. can be used as a stand-alone production technique,
The process continues, layer by layer, until the part or combined with conventional toolmaking processes.
is complete, enabling even highly complex parts and Direct SLS of composite metal powder blends can be
tools to be built up quickly and fully automatically. used to produce metal parts. In such systems, a low
The unsintered powder stays in the container during melting point component is melted and employed as
the process, serving as a natural support for the part. a matrix in which the higher melting point compo-
SLS of metals is divided into indirect and direct nents sit. Typical binary phase systems investigated
methods. Indirect SLS involves the use of a poly- include Ni-Cu, Fe-Cu and Cu-Pb/Sn[6,7] . The dis-
mer binder to produce a green part. The metal pow- advantage of the above processing routes is that the
der particles are coated with the polymer and the components produced exhibit the mechanical prop-
action of the laser melts the polymer, bonding the erties and characteristics of their weakest composite
metal particles together to produce a green part. The phase, thus lacking the full mechanical functionality
required for heavy-duty tasks.
Direct SLS of metals is a promising rapid proto-
† Ph.D., to whom correspondence should be addressed,
typing technique that allows manufacturers to pro-
E-mail: dewidar5@hotmail.com.
228 J. Mater. Sci. Technol., Vol.24 No.2, 2008

SLS on a DTM Sinterstation 2000. The fragile green


part produced consists of steel powder held in a poly-
mer matrix. The second step in the process was a
sintering furnace cycle. In this step the green parts
were placed into a furnace and heated to 1120◦ C with
a heating rate of 2◦ C/min. During the heating phase
the polymer was burned away leaving the steel skele-
ton, which was then sintered traditionally to a porous
steel structure at 1120◦ C for 3 h. The sintered parts
were then cooled to room temperature at a cooling
rate of 3◦ C/min. The atmosphere in the furnace was
inert gas with a mix of 30% hydrogen and 70% ni-
trogen recommended[16] . Finally this structure was
infiltrated with molten bronze, which was absorbed
from the base of the parts by a wicking action, which
drew the bronze through the porous matrix resulting
in a near fully dense composite. The furnace was held
at 1050◦ C for 3 h. The heating rate, cooling rate, and
the atmosphere inside the furnace were the same as
for the sintering cycle.
To measure the strength of the RapidSteel 2.0
material four cubic test blocks with dimensions of
80 mm×80 mm×80 mm were built from RapidSteel
2.0, using Sinterstation 2000. The processing pa-
rameters to manufacture the green parts were the
DTM default setting. Each block was cut into six-
teen 20 mm×20 mm×80 mm slices cut either hori-
zontally, or vertically and were machined to dumb-
bells (see Fig.2).
Tensile testing was conducted using a Dartec Uni-
Fig.2 (a) 80 mm cubic RapidSteel 2.0 block cut into hor-
izontal sections, (b) 80 mm cubic RapidSteel 2.0
versal Testing Machine at a fixed crosshead speed
block cut into vertical sections, (c) dimensions of of 0.05 mm/s. Some samples from fractured speci-
RapidSteel tensile specimens mens were further evaluated by scanning electron mi-
croscopy (SEM, JEOL JSM 35CF).
In order to study the bulk accuracy, and manu-
duce both prototype and production tools. Recently, facturability of small features of the RapidSteel 2.0
new generations of different material systems have material, two blocks with small features, as shown
been introduced to improve strength, surface quality, in Fig.3 were built to characterise the ability of the
and tool life[8–15] . The aim of this work is to compare process to manufacture small negative and positive
between direct and indirect SLS of metals in terms of (protruded and depressed) features. A family of fea-
mechanical properties, accuracy, and rate of building. tures, fabricated from a number of cylinders and bars
was defined and applied to 80 mm cubic block. The
2. Experimental same features were made for the negative and posi-
tive blocks. Coordinate Measuring Machine (CMM)
RapidSteel 2.0 process was used to create parts was used to measure the dimensions of all features
using indirect SLS. The first step in the process was for green parts, brown parts, and infiltrated parts to
to build a green part from coated 316 stainless steel determine the effects of SLS process and the furnace
powders. These were produced through conventional cycles on the accuracy of the small features of parts.

Fig.3 (a) Test blocks to assess small feature accuracy, (b) dimensions of small features
J. Mater. Sci. Technol., Vol.24 No.2, 2008 229
laser power ranging from 50 to 80 W, scan speed
1 mm/s, and scan spacing 0.682 mm, which recom-
mended from our previous work[8] .
To measure the strength of the parts produced
using direct SLS, many samples were made. Some
samples were further conventionally sintered to in-
crease density. Sintering was performed in a vacuum
(10−3 Pa). Samples were heated at a rate of 10◦ C/min
to sintering temperature (1240◦ C), held for 1 h, then
furnace cooled to room temperature. Other samples
were infiltrated to full density using the same process
as for indirect SLS process. Tensile testing for direct
SLS samples was conducted under the same condi-
tions as RapidSteel 2.0 samples. A surface profilome-
Fig.4 Variation in tensile strength of RapidSteel 2.0 ter Form Talysurf 120L from Taylor Hobson Ltd., was
sample cut horizontally used to measure surface roughness.

3. Results and Discussion

The resulting variations in the strength for all six-


teen samples in the block, which were cut horizontally
as a function of distance from the place of bronze infil-
tration are plotted in Figure 4. A pattern of degrada-
tion appears with strength with decreasing distance
from the place of bronze infiltration in both the ver-
tical and horizontal directions. Figure 5 shows maxi-
mum strength of brown samples (after sintering cycle
only). It is clear from this figure that the strength
is approximately constant at 33±3 MPa. It is also
clear from the above results shown in Figs.4 and 5
Fig.5 Variation in tensile strength of RapidSteel 2.0 sin- that the infiltration cycle plays an important role in
tered samples (brown part) the indirect SLS process, where the strength of infil-
trated part has been increased to fifteen times higher
The direct SLS machine includes a 250 W contin- than that of the brown part. The variation in hard-
uous wave CO2 laser with a minimum beam size of ness with infiltrated samples and brown samples is
1.1 mm in diameter at the focal length. Galvanome- shown in Fig.6. It can be clearly seen from this fig-
ter controlled mirrors direct the laser beam within a ure that the hardness of infiltrated samples is lower
70.0 mm diameter build area which is housed within a than that of brown samples due to the presence of
300 cm3 (L=460 mm, H=260 mm, and D=250 mm, the bronze. The 0.2% yield strength of RapidSteel
where L is the length, H the height and D the depth) 2.0 material varied from 250 to 350 MPa, and the
process chamber capable of sustaining a variety of at- ultimate strength ranged between 375 to 693 MPa,
mospheric conditions. In this study, the atmosphere with the elastic modulus remaining relatively con-
inside the chamber was high purity argon (bottled stant at 256 GPa. Figure 7(a) shows that the den-
argon at 99.9% purity) to minimise oxidation effects. sities of brown samples are relatively constant about
The material was gas atomised high-speed steel (HSS) 4.34 g/cm3 , and Fig.7(b) shows that the densities of
powder. The grade of HSS processed was chosen for infiltrated samples are about 8 g/cm3 . Also it is no-
its high solidus-liquidus range, which was considered ticed from Fig.7(b) that the samples farthest away
to offer the greatest scope for appropriate process con- from the bronze position are only about 1% less dense
trol. The processing conditions for this material were than those close to the bronze position, however, this

Fig.6 (a) Variation in hardness with cut horizontally infiltrated RapidSteel 2.0 samples, (b) variation in hardness
with brown block RapidSteel 2.0 samples
230 J. Mater. Sci. Technol., Vol.24 No.2, 2008

Fig.7 (a) Density measurements for RapidSteel 2.0 samples without infiltration (brown block), (b) density mea-
surements of RapidSteel 2.0 infiltrated material

sintering cycle, is 208 MPa. The benchmark value of


the modulus of elasticity of the infiltrated samples is
130±5 GPa. The low value of modulus of elasticity of
these samples compared with RapidSteel 2.0 material
occurs because the orientation of the agglomerates is
parallel to the direction of scanning. The direction of
the tensile tests is carried out parallel to the sintered
layers in both direct and indirect SLS process.
The density of samples without infiltration was
found to be 5.13±0.2 g/cm3 ; the density of sam-
ples after a furnace sintering cycle was found to
be 5.5±0.1 g/cm3 , while the density of samples af-
ter infiltration was found to be 7.76±0.2 g/cm3 .
The hardness of infiltrated samples was found to be
36.2±0.2 Rc.
Fig.8 Typical variation in strength of sintered and infil-
trated direct SLS samples
Analysis of the RapidSteel 2.0 material under a
microscope reveals spherical metal particles, located
within a bronze matrix. Further detailed etching with
small variation in density appears to weaken the inter- ferric chloride revealed large bronze grains, which de-
nal structure. The maximum strength of A1, which is crease in size with distance from the bronze infiltra-
close to the bronze position is 671.3 MPa, where the tion source. Other analysis results by scanning elec-
maximum strengths of B3, and C3, which are farthest tron microscopy (SEM), shown in Fig.9(a), confirm
away, are 376.25 and 437.16 MPa, respectively. these findings. Figure 9(b) shows a brown sample,
Figure 8 shows the typical stress-strain curve for which reveals spherical metal particles ranging from
two samples. The first sample was made using the di- 34 to 55 µm in diameter. The dark colour shows in-
rect SLS process followed by a sintering furnace cycle, ternal porosity within the material.
and the second was made using the direct SLS process The studies related to accuracy have considered
followed by an infiltration cycle. The maximum ten- bulk accuracy and surface finish, and the accuracy
sile strength of the infiltrated samples is 426.3 MPa, and manufacturability of features. The accuracy of
while the maximum strength of the samples after a components manufactured using the indirect and di-

Fig.9 (a) SEM of RapidSteel 2.0 sample (dark colour represents the steel particles, while the bright colour
represents the bronze), (b) SEM of brown RapidSteel 2.0 sample (dark colour represents the pores)
J. Mater. Sci. Technol., Vol.24 No.2, 2008 231

Fig.10 Dimensional accuracy of the 80×80×80 block part

rect SLS processes have been assessed through the duced by indirect SLS need sintering and infiltrating,
manufacture and measurement of a range of com- which takes about 2 d. Sintered parts produced by
ponents and tools. Figure 10 shows the dimen- direct SLS only need infiltrating (about 1 d) to im-
sional accuracy of the green, brown and infiltrated prove the density and the mechanical properties of
blocks. The nominal dimensions of the blocks are the part.
80 mm×80 mm×80 mm. It is noticed from Fig.10
that the green parts shrink by about 0.1 mm in all di- 4. Conclusions
rections during the furnace sintering, and swelled by
0.05 mm in the X direction, 0.1 mm in the Y direc- (1) Both direct and indirect SLS process can gen-
tion and more shrinkage occurred in the Z direction erate components with sufficient mechanical proper-
through infiltration. Generally, the accuracy of the ties to act as mould tools.
block is within ±0.2 mm in all directions. The sur- (2) Neither direct nor indirect SLS is currently ca-
face roughness, Ra of RapidSteel 2.0 components after pable of making a net shape component at the accu-
infiltration is typically in the range 6–10 µm. racy and surface roughness levels mould tooling ap-
The absolute error in both X and Y directions plications demand.
for the samples produced by direct SLS was found to (3) For the two material systems studied here in-
be ±0.4 mm. The error of direct SLS parts is bigger direct SLS is capable of building components at a sig-
than that of indirect SLS parts, which may be due nificantly higher rate than direct SLS. Other work[10]
to the big size of beam diameter (1.1 mm). There is on direct metal SLS with powder mixes has shown
no significant variation in dimensions after the infil- that some direct metal SLS systems can operate at
tration cycle. An average roughness Ra in the range more competitive build rates (of the order of to days
40–50 µm was obtained after sintering, which reduced manufacture tool), but further assessment of this is
to 40–45 µm after infiltration. Many mould tools in- required.
clude small positive and negative features, and the (4) The variation in accuracy caused by furnace
degree to which these could be formed with the indi- cycles is very limited for both processes, so good sin-
rect and direct SLS processes was clearly of interest. terstation process control is essential to ensure accu-
Figure 3 shows the test blocks, which were used to racy.
characterise the ability of the indirect SLS process to (5) With regard to using the indirect SLS pro-
manufacture small positive and negative features. A cess to generate full production tooling, it is possible
family of features, made up of a number of bars and to generate near net shape tools which require post-
cylinders, shown in Fig.3(a), was defined and applied processing (finishing operation) to meet production
to 80 mm cubic block. The feature set was the same specification. There are geometric limitations on this
for the positive and negative feature blocks, and an process with regard to small features (2 mm or less).
example of each is shown in Fig.3(b). The manufac-
tured blocks were then measured to determine what
features the process was capable of manufacturing to REFERENCES
within the ±0.2 mm range identified above as the gen-
eral accuracy figure. For the indirect SLS process the [1 ] C.S.Edson, S.Masanari, O.Kozo and L.Tahar: Int. J.
conclusion was that any feature with a dimension of Mach. Tool. Manuf., 2006, 46, 1459.
less than 2 mm would not generally be accurate to [2 ] J.H.Liu, Y.S.Shi, Z.L.Lu, Y.Xu, K.H.Chen and
S.H.Huang: Mater. Sci. Eng. A, 2007, 444(1), 146.
within ±0.2 mm. Because of the large beam diameter
[3 ] H.Asgharzadeh and A.Simchi: Mater. Sci. Eng. A,
no attempt has been made to manufacture features
2005, 403(1), 290.
on this scale using direct SLS. For use in production
[4 ] L.Hongjun, F.Zitian, H.Naiyu and D.Xuanpu: J.
tooling applications both indirect and direct SLS com- Mater. Process. Technol., 2003, 142(3), 710.
ponents would require substantial finishing. [5 ] E.Berry, J.M.Brown, M.Connell, C.M.Craven,
The build time of the indirect SLS pro- N.D.Efford, A.Radjenovic and M.A.Smith: Med. Eng.
cess of RapidSteel 2.0 for a solid block Phys., 1997, 19(1), 90.
(80 mm×80 mm×80 mm) is about 21 h, while the [6 ] J.P.Kruth, L.Froyen, L.Van Vaerenbergh, P.Mercelis,
build time for the same block using the direct SLS M.Rombouts and B.Lauwers: J. Mater. Process.
process would be about 17 d. The green parts pro- Technol., 2004, 149, 616.
232 J. Mater. Sci. Technol., Vol.24 No.2, 2008

[7 ] L.Lu, J.Y.H.Fuh and Y.S.Wong: Laser-Induced Ma- [12] D.King and T.Tansey: J. Mater. Process. Technol.,
terials and Processes for Rapid Prototyping, Kluwer 2003, 132, 42.
Academic Publishers, Norwell, USA, 2001. [13] D.Gu and Y.Shen: Mater. Lett., 2006, 60(29), 3664.
[8 ] M.Dewidar, K.W.Dalgarno and C.S.Wright: Advances
[14] D.Gu and Y.Shen: J. Mater. Process. Technol., 2007,
in Manufacturing Technology XV, John Wiley and 182(1), 564.
Sons, 2001, 181.
[9 ] A.Simchi: Mater. Sci. Eng. A, 2006, 428, 148. [15] J.W.Xie, P.Fox, W.O0 Neill and C.J.Sutcliffe: J. Mater.
[10] S.Kumar and J.P.Kruth: Mater. Design, 2007, 28(2), Process. Technol., 2005, 170(3), 516.
400. [16] DTM: Guide to materials: rapidSteelTM 2.0 used to
[11] A.Simchi, F.Petzoldt and H.Pohl: J. Mater. Process. produce rapid tool LR mold inserts0 . DTM Corpora-
Technol., 2003, 141, 319. tion, DCN: 8002-10001, August, 1998.

Vous aimerez peut-être aussi