This Proceedings of the Sixth Solid Freeform Fabrication (SFF) Symposium, held at The
University of Texas in Austin on August 7-9, 1995, was attended by over 150 national and
international researchers. Papers addressed SFF issues in computer software, machine design,
materials synthesis and processing, and integrated manufacturing. The continued growth in the
research, application and development of SFF approaches was readily apparent from the increased
participation over previous years and the diverse domestic and foreign attendees from industrial
users, SFF machine manufacturers, universities, and government. The excitement generated at the
Symposium reflects the participants' total involvement in SFF and the future technical health of this
growing technology. The Symposium organizers look forward to its being a continuing forum for
technical exchange among the expanding body of researchers involved in SFF.
The Symposium was again organized in a manner to allow the multi-disciplinary nature of
the SFF research to be presented coherently, with various sessions emphasizing computer issues,
machine topics, and the variety of materials aspects of SFF. To avoid parallel sessions, a poster
session was organized. We believe that documenting the constantly changing state of SFF art as
represented by these Proceedings will serve both the people presently involved in this fruitful area
as well as the large flux of new researchers and users entering the field.
Several important issues surfaced during a plenary discussion at the end of the meeting.
Considerable interest was expressed in the availability of related topics on the worldwide web. In
response, The University of Texas at Austin Laboratory for Freeform Fabrication homepage
(http://shimano.me.utexas.edu/sffl) now includes links to all sites currently published by our home
page, including all locations submitted at the meeting. This current list of web locations is also
included at the end of this proceedings volume. We will be pleased to update the list by notification
of one of the Symposium Proceedings editors.
Another issue which would benefit a majority of SFF researchers is formation of a research
infrastructure manufacturing network. Interest was expressed in the formation of a library of
"public domain" .STL files. Clemson University has created this, and Elaine Persall is the contact
person. contact information is in the participant index.
The editors would like to extend a warm "Thank You" to Sue Ferentinos for her detailed
handling of the logistics of the meeting and the Proceedings, as well as her excellent performance
as registrar and problem solver during the meeting. We also acknowledge the support efforts of
Vicki Lehmeier and Cindy Pflughoft throughout. We would like to thank the organizing
committee, the session chairmen, the attendees for their enthusiastic contributions, and the speakers
both for their significant contribution to the meeting and for the relatively prompt delivery of the
manuscripts comprising this volume. We look forward to the continued close cooperation of the
SFF community in organizing the Symposium. We also want to thank ONR through Grant No.
N00014-95-1-0424, ARPA, and The Minerals, Metals and Materials Society for co-sponsoring the
Symposium with the Mechanical Engineering Department and the Center for Materials Science and
Engineering at the University of Texas at Austin.
Organizing Committee: Dick Aubin, United Technologies Research Center;
Joel W. Barlow, The University of Texas at Austin;
Joseph J. Beaman, The University of Texas at Austin;
David L. Bourell, The University of Texas at Austin;
Robert L. Brown, The Gillette Company;
Michael Cima, Massachusetts Institute of Technology;
William Coblenz, ARPA;
Richard Crawford, The University of Texas at Austin;
Samuel Drake, University of Utah;
Steven Fishman, Office of Naval Research;
Harris L. Marcus, University of Connecticut at Storrs;
Sean O'Reilly, Ford Motor Company;
Fritz Prinz, Stanford University;
Emanuel Sachs, Massachusetts Institute of Technology;
Greg Sanders, Lost Foam International;
Susan Smyth, General Motors Corporation;
Ralph Wachter, Office of Naval Research;
Michael Wozny, Nat'l Institute of Standards and TechnoIogy
Several technical challenges exist in adapting Three Dimensional Printing (3DP) to processing of dense ceramic structures. The sintering rate of particulate bodies depends on the sintering mechanism, average powder size, and initial packing density. Fine powders are necessary to ensure appreciable densification rates from powders which sinter by solid state transport. A critical packing density exists for such powders below which densification does not occur. Special build strategies are, therefore, required for 3DP of ceramic structures. We have successfully demonstrated five approaches to produce dense ceramic components by 3DP. First, spray-dried granules of fine ceramic powders are spread in the existing 3DP equipment and bound using a latex binder through an ink-jet print head. The resulting components are then isostatically pressed to raise the green density to a point that the parts will fully densify when fired. A second approach uses glass powders that sinter by a viscous sintering mechanism. Such bodies sinter to full density at all initial green densities. Spray-dried granules of fine glass powders are spread and bound with latex followed by directly sintering to full density. Both of these approaches produce rather large linear shrinkage because of the low overall packing density. Large glass particles have a much higher packing density and produce bodies that sinter to full density because of the rapid viscous sintering. This third technique produces fully dense parts with linear shrinkage of about 15%. The fourth approach involves glass infiltration of porous ceramic bodies. Our results indicate that this technique can produce dense parts with less than 1% linear shrinkage. Finally, the 3DP process has been modified to permit deposition of fine powders as slurries, rather than dry powders. The resulting process considerably increases the bed density and the resulting fine ceramic parts can be sintered to full density without intermediate isopressing.
Computed tomography produces sets of tomograms for medical interpretation. Typical interpretation consists of imaging and simple observation on a 2D display screen, so that feature extraction and tissue differentiation is based primarily on human expertise. Solid freeform fabrication offers the promise of fabrication of prostheses based on actual patient anatomy. Use of CT data for this purpose requires automated interpretation. This paper presents a system architecture based on neural networks for the segmentation and classification of tissues of interest in tomograms. This approach produces a quantitative recovery of the available information by applying a feed-forward neural net trained with the back-propagation algorithm. The neural network architecture selected was tested on fabricated CT image matrices of the lower extremity.
Three Dimensional Printing is a desktop manufacturing process in which
powdered materials are deposited in layers and selectively joined with binder from an ink-jet style
printhead. Unbound powder is removed upon process completion, leaving a three dimensional
part. Stainless steel injection molding inserts have been created from metal powder with the 3DP
The freedom to create internal geometry by the use of the 3D-Printing process allows for the
fabrication of molds with complex internal cooling passages. Tooling was developed with cooling
channels designed to be conformal to the molding cavity. A finite difference simulation was
constructed to study conformal channel design.
A direct comparison of the mold surface temperature during the injection cycle of a 3D Printed
mold with conformal channels and a mold machined with conventional straight channels was
completed. The conformal passages produced with the 3DP process provide the ability to
accurately control the temperature of the molding cavity throughout the process cycle. Surface
temperature measurements demonstrated that the inserts with conformal cooling channels exhibited
a more uniform surface temperature than the inserts machined with straight channels.
Issues such as powder removal and post processing of green parts with small cooling
channels were investigated.
One of the advantages of the Selective Laser Sintering (SLS) process is that a variety of materials
can be processed. However, the goal of being able to produce fully dense metal parts with no post
processing has been elusive. Using Selective Laser Sintering to produce metal parts with full
density without post processing poses a challenge since both the processing conditions and the
metal system must be controlled. This article describes two metallurgical mechanisms by which
loose metal powder beds could be sintered to nearly full density using a scanning laser beam. The
mechanisms are particle rearrangement during liquid phase sintering (LPS) and in-situ infiltration.
Some of the particles, when heated by the laser radiation, melt and form a liquid. Ifthis liquid has
certain physical properties (e.g., low viscosity and high surface tension) and wets the other solid
particles, then the SLS process can in theory produce dense layers by either mechanism. The
purpose of this study is to determine the process and material selection parameters required to
achieve fully dense parts during direct Selective Laser Sintering of metal.
The deposition rates using pure tetramethylsilane (TMS) as precursor are calculated numerically
for a ~od .grown by th~ .Selected Area Laser Dep?sition J?rocess. In particular, the dependence of
the kinettcs of deposItion on pressure of TMS IS examIned. The conditions for which volcano
d~pos~tion pr?files occur are also investigated. The results show that deposition rate increases
wIth IncreasIng pressure and then becomes saturated. In addition, adsorption-desorption
phenomena, rather than effects ofreactants depletion, are responsible for the volcano deposition
profile observed experimentally
(1995) Melvin III, Lawrence S.; Beaman Jr., Joseph J.
A sieve feed system has been designed for use with the Selective
Laser Sintering process. The sieve feed system uses
electrostatic charge to help apply polymer powder to a green
powder bed. The sieve feed system was found to help the
application of polymer powder as measured by a 10 to 15%
increase in final part density. The sieve feed system has
many potential applications, including material property design,
and material mixing during the sintering process.
(1995) Melvin III, Lawrence S.; Beaman Jr., Joseph J.
Metal powder was applied for the Selective Laser Sintering
process using the sieve feed system and a magnetic field.
The magnetic field had a negative effect on final part quality
as measured by a reduction in final part density. This
negative effect is theorized to be due to the shape and orientation
ofthe magnetic field. It appears possible to change
the field to a benevolent orientation.
Current methods ofjoining of ceramic components may compromise the strength, chemical
resistance, or high temperature properties of the resulting ceramic parts. A new method of
joining, Ceramic Joining by Selective Beam Deposition, creates an all-ceramic joint
between two or more ceramic components through selective decomposition of a gas
precursor. An all-ceramic joint not only preserves the valuable properties of the ceramic
materials joined, but may be tailored to match the coefficient of thermal expansion ofthe
original material(s). The added material may be the same as one or both of the joined
Inaterials, or may be a composite material. This preliminary work explores the effect of
scanning speed and precursor pressure on Selective Beam Deposition ofsilicon carbide
A simple model for estimating the emissivity ofthe surface of a powder bed from
knowledge only ofthe bed porosity and solid emissivity is presented. Estimates by this
model are compared with experimental measurements for powder beds of alumina, silicon
carbide, and iron. Agreement within the uncertainty ofthe measurements, ± 10%, is
A revised view factor for the prediction of the thermal conductivity of powder beds at high
temperatures that includes a radiation contribution to the conductivity is presented.
Comparison of predictions by this equation with 424 measured values shows the
predictions to be accurate to within a ±30% relative error.
SelectiveLaser ReactionSinteringis a variation ofselective laser sintering (SLS) that incorporates
anjn~situreaction underthe·scannedbeamtofabricate shapes from materials not directly accessible by
traditional SLS. Thispaperclescribesaninvestigation into the production of silicon nitride (Si3N4) shapes
by lasersinteringsiliconpowderinanammonia (NH3) atmosphere. The effect of gas pressure and the
importance of gas/laserinteractionsarediscussed. Single and multiple layer shapes are fabricated. The
material is analyzed by x~ray diffraction spectroscopy (XRDS) for phase content and scanning electron
microscopy (SEM) for macrostructure. Data is presented that demonstrates conversion rates from silicon to
silicon nitride on the order of 85%.
Selective Area Laser Deposition (SALD) has demonstrated the ability to deposit
controlled shapes of silicon carbide and silicon nitride by using a laser beam to decompose a
precursor gas. The goal of the work here is to include titanium among the list of SALD
materials, although this goal has not yet been reached. This paper describes the selection of
precursors and the results of some SALD experiments using the first precursor explored,
titanium tetrachloride. The results of precursor gas mixture, pressure, and laser power on
deposition composition and rates are discussed.
(1995) Rao, T. Srinvasa; Bourell, D.L.; Marcus, H.L.
Production of structurally sound parts by any rapid prototype technique is essential, because
fully functional features are necessary where application testing is required. In the present work,
a powder blend of A1203/AI (3:1 by weight) was mixed with ammonium dihydrogen phosphate
and subjected to selective laser sintering (SLS) using a C02 laser. An attempt has been made
to increase the powder bed density by introducing vibration to the part cylinder. These SLS
processed preforms were then subj ected to a secondary heat treatment in a hydrogen
atmosphere and to hot isostatic pressing. Densification behavior of these Al20 3/Al composite
preforms is discussed.
This paper describes the results to date ofresearch done to compare and contrast the
tribological properties ofnanostructured tungsten carbide-cobalt composites consolidated by
selective laser sintering (SLS) and conventional grain size composites ofthe same chemical
composition consolidated by conventional commercial methods. The powder preprocessing and
selective laser sintering methods will briefly be described. The tribological testing methods will be
discussed, and the tribological properties ofthe selective laser sintered and commercially
consolidated materials will be compared. It will be seen that the nanosized WC-Co composites
have far superior harness and wear resistance compared to their microsized counterparts.
The orientation of SFF-manufactured parts can have a significant effect on the
quality of the parts, in both surface effects and strength. Currently, orientation is
either ignored or set on the basis of experience. This paper takes some simple experiments
and creates quantitative measures relating different aspects of part quality to
orientation. This leads to several computational tools for optimizing the orientation
of a part for manufacture with SLS or SALD on the basis of part strength, surface
"aliasing" error, volumetric supports, and build time as an alternative to human
(1995) Turner, Irem Y.; Wood, Kristin L.; Busch-Vishniac, Ilene J.
In this work, the machine dynamic response in Selective Laser Sintering is investigated with
the purpose of determining the causes of scanning errors. Machine subcomponents are first
investigated to determine their potential effects on the laser beam positional accuracy. The
dynamics of the laser beam delivery system are identified as the major contributor to deviations
in the laser beam position. The moving-iron galvanometer scanner used in SLS machines is then
modeled, with the ultimate goal of understanding how its various components and parameters
affect part scanning accuracy. This work should provide a better understanding of the dynamics
of the laser beam delivery system and give insight on machine parameters that result in better
Alumina powders with mean particle size of 15 J.lm, 2 J.lm and 0.5 J.lm were processed by
SLS using polymer binders. The 2 J.lffi and 0.5 J.lm powders were given a thermal agglomeration
treatment before SLS. The green shapes after SLS were infiltrated with a colloid of aluminum
oxide. After infiltration the samples were given suitable heat-treatment to remove the polymer
binder by thermal decomposition followed by a sintering treatment at 16OOC. Green densities were
in the range of 45% of theoretical density for the agglomerates of 2 Jlm and 0.5 J.lffi powders while
it was about 36% of theoretical density for the 15 J.lffi powders. Sintered densities were about 55%
of theoretical density for the samples from agglomerates of 2 J.lffi and 0.5 Jlm powders while it was
42% of theoretical density for the samples from 15 J.lm powders. The strength of samples were
measured in the green state and after sintering to determine the effect of particle size.
A two-stage method has been devised for free form fabrication of nickel,
iron and copper based alloy parts with shape and property control equal or
superior to investment castings in the same base alloys. A major advantage
of the approach is the ability to utilise commercially available selective laser
sintering systems with virtually no modification from their standard
configurations for plastic model generation. We have demonstrated the
essential feasibility of shape, dimension and property control for complex,
low production volume rocket engine components and for tools and dies in
higher volume commercial production situations. This presentation is limited
in scope to a brief overview of our recent progress.
To cope with the increasing market competition, the concurrent engineering (CE) concept is
being adopted by many companies to reduce the cost and the cycle time for manufacturing quality
parts. To build a successful CE system where designers and manufacturing experts work
simultaneously, the appropriate management of the product information flow among the users is
essential. The product information include high-level data such as design intent, part functionality
and manufacturing processes, which traditional CAD systems cannot support. To support such
high-level information beyond geometric data in the CE system, feature-based CAD systems have
been introduced to associate engineering meaning to the shapes of the CAD model components. In
these systems, users can manipulate the CAD models in terms offeatures, and software algorithms
can simulate the human behavior by manipulating the high-level feature entities, as oppose to the
low-level geometric reasoning processes with blind searching algorithms.
One of the primary application of the current SFF processes is to fabricate design prototypes for
fast design verification: The process is identified to be a valuable tool in the CE environment
because it can reduce the significant amount of design cycle time. Therefore" it is desirable that the
SFF process software is fully integrated into the environment by taking a feature-based approach.
As the process requires extensive geometric reasoning procedures that are time consuming and
require complex algorithms, the feature-based approach is appropriate, and more intelligent
processing is possible. Also, an algorithm can be easily customized