thesis glink

150
Chapter 1 Introduction Chapter 1. 1 Overview 1 Chapter 1 Introduction Chapter 1. 1 Overview This thesis describes improvements to the Shape Deposition Process (SDM) for the development of tool steels die inserts. SDM is a rapid prototyping process, which is used at Stanford University and Carnegie Mellon University to develop metallic parts by rapid fusion of powder metal by laser deposition or by plasma melting of metal wire. The approach given here was to develop an understand the microstructure produced by the laser deposition of tool steels. This understanding is essential to building die inserts. Die inserts have stringent material requirements because of the intense service conditions to cast aluminum parts. The presented research developed three significant outcomes: 1. The microstructual and morphological characterization of laser layered deposition of carbon steels. 2. A set of design rules which one must consider when depositing carbon steels with laser based layered manufacturing processes. 3. A methodology for using phase transformation and other microstructual knowledge to design improved parts, Designing for Microstructural Manipulation (DFM 2 ) was also developed. The benefits of characterizing carbon steel microstructure and cataloging the morphological evolution are the application of this knowledge to reduce part deflection and enhance part properties. Also, the carbon steel deposition methodology will increase building throughput, reduce part failure, and extend the range of the applications that can use laser layered manufacturing. The benefits to DFM 2 are the ability to leverage the transient heating patterns of laser layered manufacturing to manipulate phase percentage, grain size, and phase location to enhance wear

Upload: melculturbat

Post on 03-Oct-2014

53 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Thesis Glink

Chapter 1 Introduction

Chapter 1. 1 Overview

1

Chapter 1 Introduction

Chapter 1. 1 Overview

This thesis describes improvements to the Shape Deposition Process (SDM)

for the development of tool steels die inserts. SDM is a rapid prototyping process,

which is used at Stanford University and Carnegie Mellon University to develop

metallic parts by rapid fusion of powder metal by laser deposition or by plasma

melting of metal wire. The approach given here was to develop an understand the

microstructure produced by the laser deposition of tool steels. This understanding is

essential to building die inserts. Die inserts have stringent material requirements

because of the intense service conditions to cast aluminum parts. The presented

research developed three significant outcomes: 1. The microstructual and

morphological characterization of laser layered deposition of carbon steels. 2. A set of

design rules which one must consider when depositing carbon steels with laser based

layered manufacturing processes. 3. A methodology for using phase transformation

and other microstructual knowledge to design improved parts, Designing for

Microstructural Manipulation (DFM

2

) was also developed.

The benefits of characterizing carbon steel microstructure and cataloging the

morphological evolution are the application of this knowledge to reduce part

deflection and enhance part properties. Also, the carbon steel deposition methodology

will increase building throughput, reduce part failure, and extend the range of the

applications that can use laser layered manufacturing. The benefits to DFM

2

are the

ability to leverage the transient heating patterns of laser layered manufacturing to

manipulate phase percentage, grain size, and phase location to enhance wear

Page 2: Thesis Glink

Chapter 1 Introduction

Chapter 1. 2 Thesis Outline

2

resistance, deformation, and strength. Conventional Shape Deposition Manufacturing

techniques do not leverage knowledge of the deposited microstructure to create

improved parts or improved designs.

Chapter 1. 2 Thesis Outline

Chapter 2 provides a detailed background material on rapid tooling processes,

the SDM process, and other laser deposition processes. Chapter 3 describes the

experimental procedures used to analyze microstructure and material properties.

Chapter 4 describes the initial investigation into using tool steels with the SDM

process to make small die casting inserts (less than 25 mm in thickness) and low aspect

ratios. Chapter 5 details the characterization of SDM deposited 400 series Martensitic

Stainless steels and other carbon steels. Chapter 6 discusses utilizing the

characterization to select die cast insert material and to reduce part deflection. Chapter

7 describes design rules for laser depositing carbon steels in a new process which

focuses on depostion microstructure, Designing for Microstructural Manipulation.

Chapter 8 provides a short conclusion to this thesis. Additional information is

included in the appendix.

Chapter 1. 3 Individual / Group Work Statement

Part of the work presented here represents group work. The Stanford Rapid

Prototyping Lab is an environment in which teamwork and collaboration of ideas is

encouraged for the purpose of enriching and progressing the pace of research. All of

the work presented was developed under this environment. The idea, testing and use

of the martensitic expansion to reduce CTE shrinkage deflection, and the development

of Designing for Microstructural Manipulation process are completely my own work.

The parametric modelling of this phenomena in laser deposited martensitic steels is

also my own work. Most of the analysis by X-ray diffraction, transmission electron

microscope, electron microbeam analysis, and pixel-phase color analysis are also my

own work. The development of “Blue,” the pixel color intensity matching program,

was designed by Rudolph Leitgeb. Many samples were etched, polished, and analyzed

by Monikka Mann, Tony Nguyen and Tonya Huntley. Ms. Mann helped to perfect the

Page 3: Thesis Glink

Chapter 1 Introduction

Chapter 1. 3 Individual / Group Work Statement

3

electron backscattering probe technique for bulk carbon steel samples. Finite element

modelling was performed by Alexander Nickel. Many of the 316L SDM tooling

efforts were built by John Fessler, Alexander Nickel, and Xiochun Li. The laser SDM

process was perfected by John Fessler and Alexander Nickel. Any errors presented in

this work are my responsibility.

Page 4: Thesis Glink

Chapter 2 Background

Chapter 2. 1 Tooling Industry

4

Chapter 2 Background

Chapter 2. 1 Tooling Industry

The prototype tooling industry is about 300 Billion dollar a year industry.

Prototype parts are often used by designers to test alternative designs. Prototype parts

can range from conceptual to functional. Barkan and Iansiti developed a detail study

of the levels a prototyping stages which can occur during the design process (2.1).

Simple models or mockups are conceptual parts which can be made from the simplest

materials and processes. They can be made of plastics, paper, etc. They can be formed

by gluing, simple machining, or blade shaping. Mock up parts do not have to

necessarily fit tight tolerances. The primary purpose for conceptual parts are “look

and feel” attributes. They are typically used in the early part of the design cycle.

Subsystem and mechanical prototypes have a wide range of model classes

which very in level of integration and tolerances. Parts in these two subclass range

from moderate to highly tight tolerances. They can also be semi-functional to

functional. These parts are used for design validation. Often these parts are used to

determine part fit within packaging requirements. These parts made be made from

traditional prototyping techniques or rapid prototyping techniques.

Show parts are subsystem prototypes which high accuracy but are used

primarily for display or review purposes. These parts are not typically made of

conventional engineering materials, but are often painted or decaled to have show

quality finishes. These parts are typically made very quickly from foams, ren board,

and epoxy plastics. They often require manual finishing to insure tight tolerances and

part dimensions.

Page 5: Thesis Glink

Chapter 2 Background

Chapter 2. 1 Tooling Industry

5

Breadboards are subsystem prototypes which are very simple with low

tolerances but exhibit part function. These types of are typically used for testing

concept function or local part/subsystem function. The ability to make changes to

these systems rapidly is a prime feature of breadboards. Tooling to build such

prototypes is relatively simple, typically glue, solder, simple linkages, etc.

Mechanical prototypes are functional prototypes which typically support all

the functions of the final model but do not necessarily represent the final size or shape

of the actual part. These prototypes may include several breadboard prototypes,

Tooling for these types of prototypes, typically consists of off the shelf technology.

For examples, standard housing may be used for prototype printers. Non standard

metal and plastic parts may be machined from simple stock materials.

Engineering prototypes are classes of models which are functional prototypes

like the mechanical prototypes but are typically made out the same engineering

material as the final part. The size or footprint of the prototypes match the final

design. These prototypes are used as final design checks or limited field testing. Low

volume prototype casting processes are used for exotic parts as opposed to machining

them out of stock. Metal parts may be sand cast or gravity poured. Even though parts

may be made from the design intent engineering material, quality may still differ from

the production run parts.

Production prototypes are prototype parts which are made from the same

engineering material that the final part will be made and manufactured from similar

processes. The purpose of these parts is to test the manufacturing process and

production part quality. These prototypes will have the material characteristics of

production intent manufacturing. Therefore, early cycle time scenarios and

production volumes can be forecasted from these prototypes. Also, reliability or

failure studies on the actual part can be run with prototypes. Unlike many engineering

prototypes, the parts could be included within validation cycles. Tooling for these

prototypes are very expense and usually require long lead times because actual tooling

inserts are required. As the Figure 2.1 below shows, changes to prototype design at

this point is very expense.

Page 6: Thesis Glink

Chapter 2 Background

Chapter 2. 2 Conventional Tooling Process

6

Figure 2.1 Adapted from Barkan and Iansiti (1993)

Chapter 2. 2 Conventional Tooling Process

To build prototype parts many conventional tooling processes are used. There

are two approaches which are usually taken to make prototypes parts: part simulation

and process simulation. Part simulation typically involves using simple processes like

material removal to just get prototype parts made. Similar stock material may be

machined to get a model. Other part simulations includes the models, mock-ups, and

low level simulations. Prototypes built with process simulation are cast, drawn forged,

stamped, etc. to build parts which will be similar to the production intent pieces.

Process simulations include the mechanical, engineering, and production prototypes.

The two types of tooling processes are common to both part and process simulations

are material-removal processes and casting. Joining processes are also common but

will not be described explicitly.

Chapter 2. 2.1 Material-Removal Processes

Traditional tooling processes for building parts typically encompasses material

Models, Mock-Ups

ComputerSimulation

Subsystem

Mechanical

Engineering

Production

1K

10K

100K

Timing

Page 7: Thesis Glink

Chapter 2 Background

Chapter 2. 2 Conventional Tooling Process

7

removal process. These processes include cutting, abrading, burning, and eroding.

Cutting processing involve single or multiple point cutting tools such as milling or

drilling bits. Abrading processes involve grinding, polishing or sanding. Burning and

eroding processes involve utilizing electricity, chemicals, heat, or hydrodynamics to

shape or remove material.

Milling and drilling processes remove material by using a single point or multi-

point tool to shear material (chips) away from the workpiece, the material being

formed into the part. The workpiece is typically fixtured so that the cutting tool can

remove material by rotating the tool while feeding the workpiece toward the cutter’s

tool face. High tolerances and sharp corners can be achieved by these processes.

These processes are typically coupled with other conventional tooling processes.

Similar processes include lathing, turning, planing and reaming.

Grinding, polishing and sanding processes are very similar to milling and

drilling operations in that it is a chip removal process with the cutting tool being the

individual abrasive grain. More chip deformation occurs with highly abrasive

processes because of the highly negative rake angles of the grains. When grinding,

high temperatures can be reached at the surface. These raised temperatures can cause

tempering, burning or heat-checking at the surface of the workpiece. Heat checking is

cracking at the surface which leads to low toughness and low fatigue and corrosion

resistance. The temperature gradients within the workpiece. Similar processes

include lapping,

Electrical discharge machining is a material removal process which erodes

metals by spark discharges. A shaping tool called the electrode delivers DC power to

the workpiece. The workpiece is submerged in dielectric fluid. When the voltage

potential difference between the electrode and workpiece is sufficiently high, a

transient spark discharges through the fluid, removing a small amount of the

workpiece. Even though EDM has a localized effects on the workpiece, the first 500

µ

m of a tool steel workpiece may have undergone significant phase changes. The

average rate of removal is 10

-6

to 10

-4

mm

3

with discharges repeating between 50kHz

to 500kHz.

Electrochemical machining is another material removal process which unlike

electroplating deposits or build up material, erodes material. The workpiece is

submerged in an electrolytic fluid which is a current carrier. As the electrolyte moves

over the workpiece, metal ions are washed away. This constant flowing of electrolyte

keeps the ions from plating on to the tool. The tool serves as a cathode and the

Page 8: Thesis Glink

Chapter 2 Background

Chapter 2. 2 Conventional Tooling Process

8

workpiece serves as a anode. This process does have a tendency of eroding sharp

corners, developing uneven flat sections, loosing tight tolerances. Another similar

material removal process is electrochemical grinding.

Thermally assisted machining or hot machining uses a heat input to lower local

yield strength to allow for easier or more efficient machining requiring lower cutting

forces. The heat input can be a torch, electron beam, laser, or plasma arc. Because

high temperatures are involved and uniform workpiece temperatures are hard to

achieve, the microstructure of the full workpiece may be affected. If the high energy

beams sources and machining conditions are well regulated, only local microstructure

will be affected.

Hydrodynamic machining or abrasive water jet machining is a material

removal process which uses a jet of water to remove material. The water pressure can

be as high as 1600 MPa. Up to a depth of 7.5 m/min of material can be removed. This

process also has a localized temperature and deformation on the workpiece.

These processes are needed because they typically can provide higher

dimensional accuracy and smoothness of surface finish than casting, forming, or other

shaping processes. Also, they can produce features with sharp corners or flatness

which cannot be formed by other shaping processes.

Material removal process typically have a localized influence on the workpiece

or part. Phase transformation, plastic deformation, and surface residual stress resulting

from removal processes, typically occur very close to the cutting surface and not

within the bulk influence. The chips or material removed absorb most of the heat.

Page 9: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

9

Figure 2.2 Percentage of heat generated which is absorbed by workpiece, tool and chip as a function of removal speed. (Adapted from Manufacturing Engineering and Technol-ogy,2.3)

Chapter 2. 3 Casting

Casting is one of the oldest methods of manufacturing dating back to 4000

B.C. Casting is not limited to metals, but can also be used with plastics, glasses, and

ceramics. Casting is most often used because it can produce very complex shapes.

These shapes may also have cast features like internal cavities and hollow sections.

Large parts can be produced by casting. Many hard to work with materials can be

shaped much more easily with casting than other processes.

Metal casting processes are of most interest to the topic presented and will be

discussed in detail in this section. Casting processes for other materials will not be

discussed. Metal casting processes can be divided into two categories expendable and

permanent.

Chapter 2. 3.1 Sand Casting

One of the earliest forms of casting metal is sand casting. Sand casting

consists of using a pattern shaped like the final cast shape to make an imprint in sand.

This imprint or cavity will be filled with molten metal. The sand will also having

gating or flow systems for the metal to enter the cavity of the sand mold. Many large

parts are cast with this method like engine blocks and pump housings.

Energy

(%)

Workpiece

Tool

Chip

Removal Speed

Page 10: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

10

The sand used for these molds are typically have a silica base making them

have high resistance to temperature. When the casting cools, they shrink, and the sand

mold collapses around the part. If the sand did not collapse hot tears or cracks would

form in the casting. The sand is molded into the cope (the top of the mold and the drag

(the bottom of the mold. Cores which are used to represent interior surfaces like

hollows or cylinders can also be made out of sand. They are held in place in the sand

mold to cast these features.

All types of metals can be sand cast. Shape complexity of sand cast parts can

be quite complex. There is no limit on part size, but small size parts are very hard to

cast because of the difficult of maintaining gating and regulating metal flow into these

cavities. The accuracy of tolerances of sand cast parts is lowest when compared to all

other methods of casting. Typical surface finishes range from 5-25

µ

m. Lastly, one

sand mold is usually made for one part. As an expendable mold, the sand mold is

destroyed after casting. The sand material is typically reusable. The grain

development in sand casting has dendrite grains and must be heat treated.

Figure 2.3 Sand Casting Mold System

Chapter 2. 3.2 Shell Casting

Shell-mold casting was developed in the 1940’s. A fine sand is coated and

fired upon a mounted pattern made of ferrous or aluminum material. The fired fine

sand is now a highly accurate mold. It is removed from the metal pattern. The shell is

removed and often supported by sand, and gated. The shell is from 5-10 mm thick.

Cope

Drag

Mold Cavity

Vent Pour basin (cup)

Flask

Sand

Parting Line

GateChoke

Blind Riser

Runner

Open riserCore(sand)

Page 11: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

11

High precision gears and other precision small parts. With proper gating, multiple

parts can be cast at one time. Most metals can be cast by shell casting.

Although, the these molds have high tolerances, the fine grain sand does not

permit much venting. Trapped gasses can cause improper filling of the mold. These

molds are also susceptible to porosity and tears. Casting weight should not exceed a

few hundred kilograms. Shape complexity is limited. Achievable surface finish is in

the range of 1-3

µ

m.

Chapter 2. 3.3 Lost Foam Casting

An aluminum or metal die is formed to replicate a mold for a casting.

Polystyrene beads are placed in the mold and heated. The polystyrene expands to fill

the mold. The polystyrene casting is then placed in a sand filled container with gating.

The inflowing metal evaporates the foam. The dissolving foam causes the metal to

solidify faster than in sand casting leading to directional solidification of the metal.

Because of the sand, the vaporizing polystyrene vents easily.

Unlike the other processes no parting lines, cores, or riser systems are needed.

The process for the most part is inexpensive with the polystyrene, sand and containing

units being relatively inexpensive. Only the aluminum shaping die can be costly.

Therefore, low volume runs can be expensive.

Lost foam castings have no size limit. Casting from this process can have

surface finishes from 5- 20

µ

m. Dimensional accuracy from this process are better

than sand castings.

Figure 2.4 Lost foam casting of a (A.) water pump housing and a (B) bearing plate. (Cour-tesy of Diversa Cast Tech.)

Chapter 2. 3.4 Plaster Mold Castings

A plaster made of gypsum or calcium sulfate with talc and silica flour is poured

A.

B.

Page 12: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

12

over a pattern and set. The pattern is then removed and baked at 120C. Molten metal

is poured into the mold. The maximum metal temperature is 1200C so only aluminum,

magnesium and other non-ferrous metals can be cast. Patterns for this process have to

be of fine quality. Wood patterns cannot be used because the liquid plaster will cause

swelling to pattern ruining the mold. Also the plaster does not permit much venting of

gases. Thus, the part must be poured in vacuum or under pressure.

The high strength of these molds allow them to maintain good dimensional

accuracy much higher than sand or lost foam casting. Surface finishes from this

process are 1-2

µ

m.These molds have very low thermal conductivity. Therefore the

casting cools more slowly yielding a much more uniform grain structure and less

warpage. The maximum size limit is about 50 kg. This process is best suited for low

production runs because of the expense of producing patterns. Also, the time to make

these molds as well as the entire molding process is quite lengthy.

Figure 2.5 A. Plaster Molds drying in an oven. B. The drag portion of a plaster mold for an air compressor housing. C. Aluminum 356 air compressor housings from a plaster mold.

A.

C.

B.

Page 13: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

13

Chapter 2. 3.5 Ceramic-Mold Casting

Ceramic Mold Casting is very similar to Plaster Castings except that refractory

high-temperature materials like zircon or aluminum oxide instead of gypsum or

calcium sulfate. This is a cope and drag technique. A ceramic slurry covers a pattern

which has been put in a flask, holding container. Once the slurry is set, the pattern is

removed. The slurry is then dried and burnt to remove volatile matter. It is then baked.

Often, to improve strength of mold, fireclay is added to the backings of the mold. This

additional processing is called the Shaw process.

These high temperature molds can be used with all metal including ferrous

alloys. Parts cast in these molds can weigh as much as 700 kg. Castings can have

surface finishes of 1-2 mm. Dimensional accuracy is also very high. This process is

very expensive.

Typical parts made with this method are impellers, cutters, dies for metal working, or

molds for plastic parts.

Figure 2.6 A Ceramic Mold made with the Shaw Process

Chapter 2. 3.6 Investment Casting

Investment casting or the lost wax process is an old process dating back to

4000 B.C. First a metal die is made which is used to cast a pattern of the intended

part. Wax and plastic is injected into the die. Several wax pattern are created. Special

care in handling patterns must be done in order keep them from breaking or distorting.

The patterns are then attached to a pattern assembly or tree. The tree will help develop

Cope

Drag

Parting Line

Fireclay Backup

Ceramic Facing

Page 14: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

14

gating in the future investment molds. The tree is then invested with refactory

materials by dipping the tree into a ceramic slurry. This dipping is repeated over and

over to build up thickness of the coating. This mold is now dried in air and then

heated to about 100C in an inverted position to melt out the wax. This may take up to

12 hours. Four additional hours are spent firing the mold to 650 -1050 C to drive off

any remaining water.

Figure 2.7 Steps 1-3 of the investment casting process

Figure 2.8 Steps 4-7 of the investment casting process

The mold can now be filled with molten metal. Once the metal has solidified,

the mold can be broken up to remove the castings. Highly accurate and complex

castings can be made. Surface finishes will range from 1-3 mm. Parts cast in this

method should be under 100 kg. The development of patterns and the use of labor

throughout the process can be very expensive. Investment castings are most cost

effective at high production volumes.

Injecting Wax or Plastic

Pattern Tree

SlurryCoating

Pattern Melt out Pouring Shakeout

Detachingof Castings

Page 15: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

15

Chapter 2. 3.7 Hard Mold Casting

For hard mold casting, a metal mold made from cast iron, steel, bronze,

graphite, or refractory metal alloys. The mold is machined with gating. Internal

cavities are maintained by cores made from metal, plaster or sand and placed in the

mold prior to casting. Various parts of the mold which are sensitive to high wear can

have inserts.

To increase the life of the mold, the surfaces of the mold is coated with a

ceramic slurry like sodium silicate. These coatings serve as a parting agents or

thermal barriers to control the rate of cooling of the casting. Ejector pins may also be

placed in these molds. The mold halves are clamped together and then heated to 150-

200C to aid metal flow through the mold and reduce thermal damage to the mold. The

molten metal is then poured through the gating system to fill the mold. Once the

casting solidifies, the mold is opened and the casting is removed. The mold may be

water cooled or cooled by fins.

All metals can be cast in this method, but high metaling point metals like steels

need dies built or heat resistant materials. Surface finishes range from 2-3

µ

m.

Maximum part weight made in this method is about 300 kg. Achievable shape

complexity is from moderate to low. Typical parts made in this method are

kitchenware, connecting rods and gear blanks. Accuracy of castings is very high.

Minimum part thicknesses allowable is 2 mm. Machinery costs and labor can make

this process quite expensive. This process is most economical for high volume runs.

Chapter 2. 3.8 Low Pressure Casting

Low pressure casting or pressure pouring is a process similar to hard mold

casting. The process involve mold haves which are clamped together and filled by

molten metal forced upward by gas pressure. The mold may be made from graphite or

metal. The pressure is maintained until the metal has completely solidified in the

mold. The metal may also be driven upward to fill the mold by a vacuum. The

vacuum helps remove dissolved gases lowering porosity.

Very high quality casting are made with this process. Surface finishes range

from 1-3 mm. All metals can be cast in this process. Castings have very high

accuracy, but moderate to low complexity. The process is very expense because of

equipment. Typical parts made by this process are railroad wheels.

Page 16: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

16

Figure 2.9 Low Pressure Casting Systems

Chapter 2. 3.9 Die Casting

Die casting was developed in early 1900’s. Molten metal is forced into a die

cavity by pressure ranging from .7 -700 MPa. In order to achieve, to die cast a special

machine is required. Two basic types of die cast systems are available: hot-chamber

and cold chamber systems.

The hot chamber process involves using a piston which traps a specific volume

of molten metal and injects it into a die cavity. The shot chamber or piston path way is

heated. The die cavity is formed by two die halves called an ejector die and cover die.

The ejector die is movable. The die halves clamp together to receive the shot of

metal. The cavity is kept under pressure until the metal solidifies. The dies are cooled

by circulating water or oil through various cooling channels in the die blocks. This

cooling aids in improving die life and in rapid cooling of metals. Usually cycle times

can reach up to 900 shots per hour. Zinc, tin and other low metaling point alloys are

cast using the hot chamber process.

Air Tight Chamber

Ladle

Refractory Tube

Air Pressure

Casting

Mold

Molten Metal

Page 17: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

17

Figure 2.10 Hot Chamber Die Casting System

The cold chamber process involves molten metal which is placed in an

injection cylinder. The injection cylinder or shot chamber is not heated. The metal is

then forced into a two part die cavity. The cavity is very similar to the ones used in the

hot chamber process. The metal injection pressures average about 20-150 MPa.

Aluminum, magnesium and copper are commonly cast in this process. Other high

melting temperature metals can also be cast in this manner.

Figure 2.11 Cold Chamber Die Casting System

Dies are typically made from H13 or other hot working steels. Wear on dies

increases with the temperature of the molten metal. Surface cracking from repeated

heating and cooling of dies, heat checking, is a major result of die wear. Conventional

die cast dies can last for more than 500,000 shots before significant wear occurs.

Die cast parts have high accuracy and can maintain tight tolerances. Bearing

surfaces can be produced by die casting. Surface finishes are between 1-2 mm.

Maximum casting weigh from less than .05 kg to 50 kg. The minimum thickness of

Molten Metal

Furnace

HydraulicShot Cylinder

Nozzle

Cover DieEjector Die

Pot

Die Cavity

Ladle

Plunger Rod

Hydraulic Cylinder

Stationary PlatenEjectorPlaten(moves)

Shot Sleeve

Cavity

Ejector Die Half

EjectorBox

Page 18: Thesis Glink

Chapter 2 Background

Chapter 2. 3 Casting

18

sections can be a thin as .5 mm. Volume production for die casting is typically high.

The expense of making die inserts, getting needed equipment and the time to produce

die inserts makes only large volume runs economical.

Chapter 2. 3.10 Centrifugal Casting

Centrifugal casting process has been used since the 1800’s. The process uses

spins a mold about an axis of rotation. Molds are typically made of graphite, steel or

iron. All types of metals can be cast in this method. Often the inside of the mold

cavity is coated with a refactory lining to reduce mold wear. Typical parts cast in this

method are pipes, gun barrels and street posts.

Centrifugal cast parts have high degree of accuracy and very little porosity.

The surface finish will range from 2-10

µ

m. Maximum part weight is above 5000 kg.

The cost of the mold and equipment is quite expensive so only large volume runs are

economically feasible.

Figure 2.12 Centrifugal Casting

Chapter 2. 3.11 Crystal Growing Casting

Crystal Growing casting processes originate from the 1960. The has a

corkscrew gate into the cavity chamber. The corkscrew constriction is designed so that

only favorably oriented grains can grow. The die cavity is contained on a moving

platform. The die cavity is heated by heat baffles which radiate heat. As the

constriction only permits a single favorably oriented crystal to grow.

Rollers

Spout

DriveShaft

Free RollerDriveRoller

Mold

Molten MetalMold

Page 19: Thesis Glink

Chapter 2 Background

Chapter 2. 4 The Use of Conventional Tooling to Make Prototype Parts

19

As the platform lowers, the radiant heat of the baffles causes the single to grow

and fill the die cavity. All other grains are stopped at the walls of the constriction.

When the mold is complete, a single crystal casting in the shape of the die cavity

The parts have a high degree accuracy, good surface finish. Equipment is and

the die cavity is very expensive. Part strength is higher than conventional castings

because it is a single crystal.

Figure 2.13 Single Crystal Casting

Chapter 2. 4 The Use of Conventional Tooling to Make Prototype Parts

To produce simple models or mockups which to progress the design process,

stock, material can be machined to produce prototypes. When material removal

processes are used to make prototypes, the grain structure of the metal prototypes

resembles that of the stock material. If the stock material is rolled, then the prototype

will have a rolled grain structure. Parts made in this manner typically will not have the

strength or yield characteristics of the production part. For example, an aluminum

valve cover machined from wrought stock will not have the structure of a die cast

valve cover.

Prototype parts for shell or investment casting can be cheaply done with sand

castings. The grain structures will be very similar. Validation or other lifecycle testing

could occur on these prototypes because of the similarity of the grain structure. It is

more difficult to develop prototyping techniques which will yield similar grain

structure and size that the conventional metal mold process produce. However, to

simulate die cast prototypes, special heat treatment must be used to gain

microstructure similar from any other type of casting or stock machined material.

Growing Single Crystal

Constriction to orient and grow 1 preferred grain

Radiant Heat

Chill Plate

Page 20: Thesis Glink

Chapter 2 Background

Chapter 2. 4 The Use of Conventional Tooling to Make Prototype Parts

20

Producing prototype die cast parts is a tradeoff between time, money, and material

properties. The heat treatment of stock materials will be costly and time consuming.

Using sand casting technology to produce one die cast prototype would be relatively

inexpensive. However, the additional processes needed to attain similar microstructures

and material properties will consume design lead time. Figure compares the average yield

strength of die cast parts to cast

Figure 2.14 The Cost to Produce 1 Part versus the as cast or produced grain size

Figure 2.15 Comparison of Yield Strength for various types of Castings .

Cost to Produce 1 Part -Without Heat Treatment

Incr

easi

ng G

rain

Siz

e

Sand

Cutting EDM Milling ECM Drilling Hydro

ThermalLost Foam

Ceramic

ShellInv Casting

S. Crystal

Permanent Mold

Die Casting

Centrifugal

0 5 0 100 150 200

Sand Castings

Plaster CastingsGrain Refined

Chilled-cast

Plaster CastingGrain-Refined

Die Castings

Yield Strength (MPa)

Page 21: Thesis Glink

Chapter 2 Background

Chapter 2. 5 The Need for Rapid Tooling.

21

Chapter 2. 5 The Need for Rapid Tooling.

As the preceding Figure 2.1 shows, the ability to make design changes is most

cost effective during the early stages of the design cycle. Today, concurrent or

simultaneous engineering that have multi-discipline design teams can now design

manufacturing processes along with new parts (2.2). The ability to test part features

and function in engineering materials and at the same time closely prototype the

manufacturing processes can occur under these new design paradigms with rapid

tooling.

Conventional tooling methods are too costly and require too much lead time to

practically use them to construct prototype parts. A typical die cast insert can take up

to 4 months to build and heat treat. Testing preliminary designs with conventional

tooling methods would eventually reduce the number of prototypes possible during

limited design times. An alternative method is required to produce functional

prototyping tooling, rapid prototyping.

Rapid tooling is the needed alternative process. Rapid tooling is the use of

solid froufrou processes to rapidly construct die insert or other forming tools to build

actual parts in simulated manufacturing rigs. Rapid tooling methods are typically

faster and less expensive than conventional tooling methods.

Chapter 2. 6 Solid Freeform Fabrication

Solid freeform fabrication is a process of building three dimensional objects in

a layered fashion. The three dimensional object is built in 2 dimensional layers

typically in an automated fashion. Therefore, complex parts can be built quite when

resolved in to two dimensional structures. Objects which cannot be built with

conventional manufacturing paradigms like high speed milling can not make

conformal cooling passages. Solid freeform fabrication techniques can build cooling

passages which follow intricate part surfaces because of the 2D layering approach.

For solid freeform fabrication a computerized model typically represented by

CAD (computer aided modelling) is then divided into layers by a hierarchical

algorithm. This algorithm divides the part by prioritizing layer order with shaping and

depositing precedences. The part is then built in the z direction, layer by layer with a

rapid prototyping process. The Figure 2.16 models the process.

Page 22: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

22

Figure 2.16 A CAD model is divided into layers and then programed and built by a rapid proto-typing process.

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tooling

Several commercial process have been used in rapid tooling.

Stereolithography, selective laser sintering, and laminate object manufacturing are all

primarily polymeric in nature and have been either used as a tool to create an insert or

have been used directly as the tooling insert.

Chapter 2. 7.1 Stereolithography

Stereolithography is a polymer based process which uses a laser or ultaviolet

light to cure an epoxy or plastic resin. This process was commercialized by 3D

Systems in and Beta tested by General Motors in . The part substrate or starting point

is a movable elevator platform which rests on the surface a large vat of curable resin.

The laser draws a pattern in the epoxy resin which solidifies and acts as bonded

supports on the platform. The supports are between 5 mm -10 MS thick which is about

20 to 100 layers. With each drawn layer the elevator submerges the part to rewet the

surface.

Once the support layers are drawn, the first layer of the part is then scanned.

The scanning process begins with the laser curing the outline of a 2D cross section of

motion control trajectories

data exchange

format

ComputerizedSolid Model

Physical Object

CAD System

Automated Process Planner

Automated Fabrication MachineRapid Prototyping Process

Page 23: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

23

the part. The inside of the cross sections are then rastered or weaved scanned. The

platform and scanned layers are then rewet by resin with the submerging of the part

beneath the liquid resin surface. After a 30 second (or shorter) wait time, the part is

raised to just below the surface the distance of one layer. Next the laser begins

scanning of the next layer. This process continues until the part is completed.

Once the part is completed, it is drained and then removed from the vat. The part is

then rinsed in a solvent to remove any uncured resin from the part surface. It is then

detached from the platform and placed in an ultraviolet curing oven for a post cure.

This process insures part strength and rigidity.

Figure 2.17 Stereolithography

Figure 2.18 SLA Part building on platform

UV LightSource

UV Curable Liquid

Liquid Surface

Platform

FormedPart

Vat

Page 24: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

24

Chapter 2. 7.2 Stereolithography Rapid Tooling

This process can be used for undercut and novel overhanging features which

cannot be easily produced with conventional machining. Thus, this process has been

used to advance rapid tooling in several ways: as EDM electrodes, the negative blank

to form the tool or as the positive insert. Researches have begun using

Stereolithography patterns to make investment cast die inserts.

Electrical discharge machining (EDM) is a tooling process which expends high

amounts to shape metal by burning or vaporization. Tool steel die inserts often used

EDM to build complex shapes and contoured surfaces. By plating Stereolithography

electrodes with copper, these electrodes have been used successfully to rough, semi-

rough and finish metal parts (2.4).

Prototype vacuum casting molds have been made with stereolithograhy (2.5).

These type of molds are ideal for stereolithography because pressure requirements and

casting temperatures are low. This process typically uses pressure to have the material

flow to all parts of the mold cavity. Rapid pressure changes typically do not occur.

Polyurethane plastic parts have been made with these stereolithography rapid tooled

molds.

In addition to vacuum casting stereolithography rapid prototyping technology

was used to create injection molding inserts, These insert unlike the vacuum casting

require much higher cycle pressures. Thermoplastic parts have been built with these

tooling inserts. Thermo-plastic specimens made using epoxy inserts and steel inserts

were compared in tensile strength, impact strength, and bifringence stress. The epoxy

tooled specimens had properties within 5 -10% of the parts molded with steel inserts.

These epoxy tooled parts had greater tensile strength, lower impact strength and lower

birefringence stress levels than there steel tooled parts (2.6).

Polyurethane and thermoplastic materials seem to perform well in

stereolithography tooling. Experiments have shown that as many as 500 parts have

been shot from a single epoxy die (2.7) . However, more demanding plastics like ABS,

polycarbonate and glass filled nylons have not fared as well because of the higher

melting temperatures have caused warping or galling of the parts. Thin walled features

like ribs and bosses are very vulnerable. One research investigation using vapor

deposited metal coating of nickel, copper, and zirconium nitride upon

stereolithography epoxy insert set. Regardless of the coating wear damage was

evident fairly quickly. The maximum number of parts, 14 parts were produced with

the copper coated tool before tool failure (2.8).

Page 25: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

25

Figure 2.19 A, Vacuum Casting Using a SLA Models B. Injection Molds made with SLA Technology

The ability to use stereolithograpy inserts seems to be a tradeoff analysis

between timing, tolerances, temperature(2.9), material, and volume.

Stereolithography tooling can produce prototype plastic parts quickly, giving the

designers the ability to get near-production parts quickly. However, the designer

ability to sufficient cool the part will dictate the materials and volume which can be

successfully produced via these inserts. Although more research is occurring,

currently, the best option to produce tightly tolerances parts in materials like ABS or

nylon still require metal inserts.

Chapter 2. 7.3 Selective Laser Sintering

Selective laser sintering is a SFF process which sinters powered material in a

layer by layer automated fashion to produce a part. Metal and plastic powders can be

used in this process. The powder is coated with a polymer binder and laser intensity is

used to melt the binder and fuse the powder together. The process pioneered by

University of Texas was commercialized by DTM Corporation.

The commercial process has a powder bed with a roller which spreads a thin

layer of material on to a platform. The laser then scans the outline and rasters the

B.

A.

Page 26: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

26

interior of a 2 D layer. Once the layer is completed the platform indexes downward.

The roller then once again spreads a thin and even layered of material across the top of

the last layer. The laser then draws the next layer. This process of lowering,

spreading, and scanning continues until the part is completed. The part must be

removed from the platform and surrounding powder bed and shaken or air blown to

remove excess powder.

Unlike the stereolithography process the build chamber is heated to just below

the glass-transition temperature or melting point of the material or binder. This reduces

the amount of laser power energy needed to consolidate the part. It also reduced part

stress because local part temperature is only raised slightly above the bulk part. The

chamber is also filled with nitrogen to make sure that chemical reactions do not occur.

Typically this process builds parts from polymer materials. ABS, nylon, glass-

filled nylon, and polycarbonate plastics have been used in this process. Metals, like

low carbon steel have also been used. However, these parts have to undergo additional

sintering and infiltration processes. Metal powder used in this process are simply not

heated enough to attain fusion. Only the polymer powder coating the powders is

melted. Additional heat treatment is needed to sinter the metal material and vaporize

out polymer binder. Once this done, the process must be infiltrated to fill pores. The

circular powder when lased simply cannot fuse or shrink together enough to eliminate

pores because of the spherical nature of the powders. Copper infiltration is used to fill

pores and produce a dense part. The final metal part is a hybrid or composite with

reasonable strength and thermal conductivity (2.10, 2.11).

Figure 2.20 SLS Parts Being shaken out of Powder Support Bed.

Page 27: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

27

Chapter 2. 7.4 SLS Rapid Tooling

Most rapid tooling activity have developed from the non-metal SLS process.

Part masters in wax have been used to make rapid investment casting molds for

General Motors. The SLS master is dipped into a ceramic slurry and coated. The

slurry coated part is then heated in a furnace, fusing and solidify the ceramic slurry

while burning or vaporizing the original SLS master. The surviving slurry is now a

perfect pattern to produce cast metal

Injection molding inserts have been built using the metal SLS process (2.12).

Cavity inserts and mold components have been successfully made with SLS for low

part volume injection molding. Fixturing methods have to be considered when using

these inserts. Good contact must be established so that these infiltrated molds will

dissipate heat well and reduce residual stress from thermal cycling. Surface finish in

infiltrated dies needs to be improved. Currently a lot of post processing is required

finish the cavities to acceptable levels. Over fifty polypropylene injection molded

parts have been in one of these infiltrated die sets (2.13).

Die casting molds have also been built for magnesium applications. Aluminum

die casting, because of metal reactivity can not be used in copper composite inserts.

Accuracy problem have limited the development of tooling insert using this approach.

Varying part geometries and wall thicknesses of die inserts lends themselves to non-

uniform shrinkage during the sintering process. Tooling for injection molding or die

casting need to have accurate tolerancing or simpler modelling techniques could be

used to build the prototypes. Sandcasting for metals and direct SLA parts for plastics

can built parts with higher accuracies than using SLS tooled inserts.

Page 28: Thesis Glink

Chapter 2 Background

Chapter 2. 7 Commercial Rapid Prototyping Processes used in Rapid Tool-ing

28

Figure 2.21 Rapid Steel Infiltrated SLS Injection Molding Inserts

Chapter 2. 7.5 Laminate Object Manufacturing

Laminate object manufacturing (LOM) is the process of building prototypes by

assembling 2 dimensional sections of adhesive material together. The material is

typically long sheets of adhesive backed paper. During preprocessing, the part is

divided into 2-D layers which are the thickness of the paper. The part is built upon a

wooden substrate which is screwed on to a steel plate. The plate is supported in the

LOM machine for stability. A few layers are adhered to the substrate by the machine

advancing the paper on to the board. A heated roller then presses the paper on to board

melting the adhesive to it. A laser then scans the outline of the paper and hatch a

pattern over the interior of the outline. The paper is then advanced forward. The

outline acts a separation so that only the unused paper, everything in the exterior of the

outline advances. This is repeated for a few layers to aid with part removal.

Now 2D part cross sections are scanned by the laser over the substrate. The

exterior of the cross section is hatched by additional laser scans. This continues until

the part is completed. When the part is completed, it surrounded by excess paper

composite which is similar to a type of sacrificial material. It was needed to support

the part during the building cycle. It is removed in a process called decubing. The

hatching process allows the sacrificial material to be removed as cubes. This process

can be very time consuming. When the part is finally decubed and removed from the

Page 29: Thesis Glink

Chapter 2 Background

Chapter 2. 8 Adapted Rapid Tooling Processes

29

substrate, it must be sealed to prevent swelling from humidity and moisture. Epoxy or

wood sealants are typically applied to the LOM parts.

Figure 2.22 Laminate Object Manufacturing

Chapter 2. 7.6 LOM Tooling

LOM paper material parts are most often used with investment or sand

casting. This type of tooling is often used to make cores or patterns. These paper parts

look like wood parts which is similar to traditional pattern parts. LOM is excellent for

tooling applications, because it can build very large tooling beds. Where the SLA and

SLS have size constraints, LOM has a build size of 50.8 cm x 50.8 cm x 50.8 cm.

Also, the paper parts are combustible and are quite suitable for investment casting

because they can be burned out during the firing of the investment shell. LOM inserts

have been used in blow molding, hydroforming, and injection molding. LOM inserts

have been used successfully to die cast magnesium inserts (2.15).

Chapter 2. 8 Adapted Rapid Tooling Processes

Many rapid tooling processes have arisen from the combination of traditional

prototyping methods and commercial rapid prototyping processes. The most advanced

of these are the Nickel Transfer Molding and Keltool Process.

Laser

Roll of Paperor Material

Part in CubePaper Support

Used Paper Roll

Page 30: Thesis Glink

Chapter 2 Background

Chapter 2. 8 Adapted Rapid Tooling Processes

30

Chapter 2. 8.1 Nickel Transfer MoldingNickel Transfer Molding is a process which has been commercialized by

CEMCOM corporations. From the CAD file of the part an SLA master is developed to

simulate the die insert set top and bottom faces needed to build the part. The SLA is

then placed in a nickel plating bath and plated. Several millimeters are plated on to the

model. The plated model is then suspended into a supporting frame box. The box is

then filled with a ceramic slurry which solidifies around the model. The SLA model

itself establishes the parting line. Once the slurry has hardened the two halves formed

by the SLA part’s parting line. The newly formed die set is removed from the SLA

model. The interiors of each halve are polished. The completed inserts can now be

used in standard injection molding frames.

Figure 2.23 Nickel Plating Transfer Process: (Courtesy of CEMCOM)

Although these molds wear faster than soft tooling P20 molds, die sets made

in this manner have made over 45,000 glass filled nylon parts.

Part is translated into anSLA part which representnegatives of mold cavities.

SLA part is supported in Nickel plating bath.

When plating has finished,The SLA part is fixtured ina supporting frame. The SLA nickel plated part forma parting line within support

The cavities are filled on both sides with a ceramicslurry.

Page 31: Thesis Glink

Chapter 2 Background

Chapter 2. 8 Adapted Rapid Tooling Processes 31

Figure 2.24 NPTP Demolding Process: (Courtesy of CEMCOM)

Figure 2.25 NPTP: Ejector Pins are Placed in Die Cavities

Chapter 2. 8.2 KeltoolKeltool is a process commercialized by 3D Systems, which also uses a SLA

model as a master. Keltool is a sintering technology which creates die inserts from

powdered material. The cavity and core of the tool to create the prototype part are

designed from the 3D CAD file. These core and cavity is then built by

stereolithography. These SLA parts are typically highly detailed. These SLA parts are

now called the master patterns.

Next these SLA masters are used with Room Temperature vulcanized silicone

rubber molding process. Molds of these cavities and cores are created by suspending

them in a frame and filling the frame silicone rubber. The SLA masters are removed

from the newly formed silicone molds. These molds are then filled with a mixture of

Once the slurry has hardened, the two halvesare demolded, leaving2 die inserts.

Ejector pins and other die cavity utilities are drilled orinserted. A complete dieset is now ready for injectionmolding.

Page 32: Thesis Glink

Chapter 2 Background

Chapter 2. 8 Adapted Rapid Tooling Processes 32

A6 tool steel powder, tungsten carbide powder, and epoxy binder. When the epoxy

binder mixture has cured within the mold, a green part is made. The green part is then

de-molded and sintered. The part which is the insert is sintered in a hydrogen-

reduction furnace. During the sintering process, the binder material is burned off

leaving a brown part which is a composite of A6 steel and tungsten carbide. The

composite insert now has voids from the burned out binder. The insert is then

infiltrated with copper to make the part fully dense. The final insert is 70% steel and

tungsten carbide and 30% copper. The insert is heat treatable and can achieve

hardness of 40-44 Rc.

Figure 2.26 The Keltool Process (Courtesy of 3D Systems)

Chapter 2. 8.3 Metal Spray ToolingEarly in the 1900’s, Dr. M.U Schoop found that by pouring molten metal into a

high pressure gas stream found that the metal would particulate into drops and deposit

in coatings. Schoop found similar results by passing metallic powder through flame.

Both of theses experiments led to the development of equipment to spray metal in wire

form. Common object sprayed in the 1930’s were dental light bulbs, refrigeration cold

plates, turbine wheels, and brake drums with soft metals like zinc, lead, or copper.

Most metal spraying had been accomplished by electric arc or flame from

Keltool Insert

SLA Cavity & CoreModels

Injection MoldedParts

Page 33: Thesis Glink

Chapter 2 Background

Chapter 2. 9 Other Rapid Tooling Processes 33

oxyacetylene torches. By the early 1960’s, a similar process is plasma spraying of

coatings began depositing hard metal or ceramic coatings. Thermal barriers and

oxidation or corrosion resistant coatings are the most common uses of metal spraying.

These coatings can also be designed to reduce wear and friction. Surfacing of die

inserts can be accomplished with spray processes.

In addition to now surfacing pumps with wear resistant coatings, heat and

corrosion resistant coating on electrical boards, massive depositions of spraying have

been attempted.

Chapter 2. 9 Other Rapid Tooling Processes

In addition to Stanford University, other academic or national lab researchers

have been developing processes to produce tooling inserts. The most advance of these

processes are the 3D Printing Process and LENS.

Chapter 2. 9.1 3D PrintingThe 3D Printing process was developed by Massachusetts Institute of

Technology in 19 . The process is very similar to inkjet printing technology. To build

a metal tooling insert, liquid binder is selectively secreted on powdered material. On a

layer by layer basis, the binder is applied in a process very similar to the way ink is

ejected on to the paper from an inkjet printer. The liquid hardens binding the powder

to the part. When the part is completed, it now a matrix of metal and binder. The part

is then sintered removing the binder. The part is then infiltrated with a metal or epoxy

to make a dense part (2.16).

Molding inserts have been built with this process. A ceramic mold made from

alumina powder and colloidal silica binder was built using the 3D Printing process.

The mold was made up of 100 powder binder layers. The mold was used to create a

bras casting in a gravity poured process. For the gravity poured process, the brass is

heated to melting temperature and poured in to the mold set under atmospheric

pressure (2.17). This tooling process has been licensed by Extrude-Hone company

Chapter 2. 9.2 LENS ProcessLaser Assisted Net Shaping process is a direct metal fabrication process which

can produce fully dense parts which can be used as die inserts, patterns, or metal

casting. A ND:YAG lase is used to melt metallic powder or powder mixtures. This is

an additive process building the part layer by layer.

Page 34: Thesis Glink

Chapter 2 Background

Chapter 2. 10 Shape Deposition Manufacturing 34

The LENS process uses a computer model to develop a laser deposition path

which represent the part in 2D cross-sections. The sections are deposited successively

in Z direction. The laser paths are changed with each layer. Alternative layers are

deposited at 90-degree angles to the previous layer. The part is built on a moving

platform which can move in X-Y translation. The powder injection nozzle moves

upward to compensate for the building of part height in the z direction.

The process has been used to make fully functional metal parts and metal

molds for injection molding. The LENS process has even been used to repair injection

molding molds. However surface finish and dimensional accuracy are a problem for

this technology. A LENS mold may require manual processing to produce good

surface finished and dimensional accuracy. Optomec is attempting to commercialize

the process.

Figure 2.27 LENS Process (Courtesy of Optomec)

Chapter 2. 10 Shape Deposition Manufacturing

Shape Deposition Manufacturing is an SFF process which was started at

Carnegie Mellon University and developed by Stanford University to produce laser

Powder DeliveryLaser Beam

X Translation

Y Translation

Page 35: Thesis Glink

Chapter 2 Background

Chapter 2. 10 Shape Deposition Manufacturing 35

deposited structures. Unlike other SFF processes like stereolithography and laser

sintering, Shape Deposition Manufacturing is a layered manufacturing process which

builds fully dense metal parts by incremental deposition and CNC shaping of material

layers (2.18, 2.19). First, a computer aided design model of a part is sliced into layers.

The layers are in the z-direction and derived by custom planning software. Next a layer

is deposited. The layer is deposited as near-net shape. This near-net shaped layer is

then milled to final dimensions by a 5-axis CNC mill. Support material is then

deposited around the layer to protect the features of this layer and provide a base for

overhanging features in following layers. The next layer of the part material is then

deposited, and the process continues.

Figure 2.28 The Shape Deposition Manufacturing

The combination of layered manufacturing and sacrificial support material

enables the production of complex features such as undercuts or conformal cooling

channels (2.20). Also this technique lends itself to the production of multi-material

structures. For instance, an insert can be produced which is primarily a hard ferrous

alloy with copper deposits for enhanced thermal conductivity. Using the multi-

material strategy, sensors can also be embedded in the die during the build sequence to

develop “smart dies.”

A laser / powder deposition system was used to deposit material for testing and

for construction of a test die (2.21). The system uses a 2.4 kW Neodymium YAG Laser

to fuse metallic powders into fully dense material. The laser is delivered by fiber

optics to an end effector mounted on a four-degrees-of-freedom robotic arm. The end

RemoveDeposit Part withSupport Material

RemoveSupportMaterial

ShapeDeposit MetalWith Laser

Depositand ShapeSupportMaterial

RemoveSupportMaterial

Page 36: Thesis Glink

Chapter 2 Background

Chapter 2. 11 SDM Tooling Efforts 36

effector focuses the light on the substrate, creating a melt pool. Metal powder is

added to the melt pool via a powder feed tube and a bead of deposited metal is created

as the robot transverses the substrate. This technique, which is similar to laser

cladding, has been very effective in forming fully dense metal layers. Nitrogen gas

shrouds the deposition to help prevent oxides from forming during the deposition

process.

Figure 2.29 Powder placement during the SDM Process

Chapter 2. 11 SDM Tooling Efforts

Several tooling inserts have been built using the SDM process. The GM

injection molding insert, the Alcoa tool injection molding insert were built for industry

partners of the Stanford Rapid Prototyping Lab.

substrate

deposited layer

laser

direction of travel

powder

Page 37: Thesis Glink

Chapter 2 Background

Chapter 2. 11 SDM Tooling Efforts 37

Figure 2.30 Equipment Setup for the SDM process

Chapter 2. 11.1 GM Injection Molding ToolThe SDM process was used to produce an injection mold for an electronics

compartment cover with snap fit tabs. The part was split into three main elements a

support level, cooling channel level, and a feature level. The support level is simply

the base of the injection molding insert. This level has no distinct features and merely

needed to allow the insert to fit into the molding base. The cooling channel level

contains the concentric cooling channel design needed to control temperature within

the insert and reduce warpage of the insert. The feature level is the top part of each

insert which will serve to mold and eject the part.

By segmenting the insert design in this manner, one can plan the deposition

and shaping of the part to enhance build time and reduce tool wear. The support level

of the insert 15 mm thick plate of 316L stainless steel. As described above a substrate

is deposited upon to build parts with the SDM method. Thus the support level and part

substrate are combined. Next, 316L stainless steel powder is then deposited and fused

to form the channel level of the part. After, a thickness of 10 mm is deposited, cooling

end effector

nitrogen shroud

powder feed tube

Page 38: Thesis Glink

Chapter 2 Background

Chapter 2. 11 SDM Tooling Efforts 38

channels are then machined in the deposit. The passages are then filled with microcast

copper to preserve the integrity of the channels as the next layers of the part are

deposited.

The feature level was laser deposited and near net-shaped by CNC on layer by

layer basis. Layer thickness averaged .25 mm. When the structure was completed, the

copper channels where etched, removing the copper so that channels could be used.

many features because of size or taper angle had to be EDM.

The inserts where placed in a 10 ton injection molding machine. The inserts

were prepped to fit into the mold base. About 20 nylon parts were run to see if the

inserts worked. As expected with such a short run, no visible die wear occurred.

Figure 2.31 GM Tool

Chapter 2. 11.2 Alcoa Injection Molding ToolA set of injection molding inserts were made for Alcoa using the SDM process.

The inserts are a composite stainless steel tool. Residual stresses caused warpage or

deformation to the GM tool during the deposition process. This warpage added

additional machining and heat treating hours to the deposition process. To combat this

during the deposition of the Alcoa inserts, the interior of the insert is deposited with

invar instead of 316L stainless. Because of the intricacy of the inserts, only two

planning levels are available: the channel level and the feature level. Microcast

copper is deposited into cooling channels.

Sinker EDM Feature

Wire EDM Feature

Zone 1

Zone 2

Zone 3

Page 39: Thesis Glink

Chapter 2 Background

Chapter 2. 12 Other Developing Laser Deposition Technologies 39

Once the bulk shape of the insert had been deposited and shaped, the copper

channels are etched out of the insert. Next specific tapers and part level features that

cannot be CNC machined were electro-discharge machined. The molding inserts

where completed and sent to Alcoa in December of 1998. Because of budget cuts the

tool was never tested.

Figure 2.32 Alcoa Injection Molding Inserts

Figure 2.33 The Interior of the Alcoa Tool

Chapter 2. 12 Other Developing Laser Deposition Technologies

There are other laser based technologies which could potentially be used to

StainlessSteel

Invar

Legend

Copper

Page 40: Thesis Glink

Chapter 2 Background

Chapter 2. 12 Other Developing Laser Deposition Technologies 40

build die casting prototype inserts and other forms of rapid tooling. Some of the

outcomes of this research will be able to benefit not only SDM laser process but other

laser technologies as well.

Chapter 2. 12.1 Laser-induced Vacuum Arc DepositionLaser-induced vacuum arc deposition is a process which combines the

controllability of pulsed laser deposition with vacuum arc technology (2.22). This

allows for very small droplets which produces a fine films. Typically amorphous

carbon films are made. These films are very hard and have excellent wear resistance

and low friction. This method has been used to deposit hard films on metallic

substrates. This technology could potentially be used to face die inserts to make them

more wear resistant yielding longer insert die life. Potentially could allow softer steels

like 316L to become more wear resistant so they could potentially be used to prototype

die casting or glass-filled nylon injection molding.

Chapter 2. 12.2 CO2 Laser DepositionThis process uses a CO2 laser to solidify metallic powder (2.23). The substrate

translate in the X,Y, and Z directions as powder is fed into the interaction zone on the

substrate. The laser power ranges between 300-400W and is focussed on to the

substrate in a donut shape with a 600 mm diameter. Helium gas is used as a shield gas.

Stainless steel 304L parts have been built with this method. Tool steel inserts may also

be able to be made with this process.

Chapter 2. 12.3 Pulsed Laser DepositionExcimer Nd:YAG or CO2 lasers are used to produce vapor of plasma states to

deposit or grow thin films (2.24). Ceramic thin films have been grown by pulsed lasers

on stainless steel, hard metal, Si, SrTiO3, and ZrO2. This technology can be used for

hardfacing tools. If multiple layer films can be built without of loss of adhesion or

delamination, this technology may be able to build feature level of die inserts.

A similar process called laser implant deposition uses a KF excimer laser to

deposit and incorporate silicon on the surface of stainless steel.

Page 41: Thesis Glink

Chapter 2 Background

Chapter 2. 12 Other Developing Laser Deposition Technologies 41

Chapter 2. 12.4 Laser Direct CastingLaser Direct Casting is a laser cladding process which uses a coaxial nozzle to

deposit and laze metal powders (2.25). Metal powder is injected into a laser generated

melt pool. The substrate translates in the X, Y, and Z directions in order to build 3D

parts. The laser power used in this process 400 W - 1400W and speeds of 500 - 1000

mm/min. Fully dense parts have been built with this process.

Chapter 2. 12.5 Laser CladdingLaser cladding is a process very similar to Laser direct casting (2.26). With

this process CAD.CAM systems are uses to develop the laser path also known as

cladding tracks. AISI 1045 steel plate and steel rollers have been deposited upon with

this process. In addition to metal depositions, metal matrix composites with ceramics

have been made. Cutting dies and stamping dies have been made with this process. A

similar process was developed at Los Alamos National Laboratory called Direct Laser

Fabrication (2.27). It is a near-net shape technology which uses CAD/CAM with a

high energy laser beam to produce fully dense parts.

Another laser cladding technique used powder blowing to place metallic

powder in the path of the laser for fusion to the substrate (2.33). A 5 KW CO2 laser is

used to solidify the powder. This process allows for very fine microstructure, no

porosity, uniform layer thickness and little dilution of material into substrate. The

HAZ produced is very small while the interface between cladding and substrate is very

sharp.

Chapter 2. 12.6 Laser induced Chemical Vapor DepositionAn argon ion laser beam is used to grow films of titanium nitride (2.28). These

films arise from direct laser pyrolysis of TiCl/4N/2/H/2 gas at atmospheric pressure.

These deposited films are hard, rough, and porous. Tool steel substrates have been

used in this process. The porosity of these coating may lend it unsuitable for tool

facing or build tool inserts. Other laser-induced chemical vapor deposition process are

at elevated pressures of 40 mbar and pulses at 100 to 600mW for nickel-iron films

(2.29). A similar process using an ArF excimer laser and low power CO2 laser to

produce pyrolytic laser chemical vapor deposition (2.32). This process has been used

to coat tool steels as well as small industrial tools.

Page 42: Thesis Glink

Chapter 2 Background

Chapter 2. 13 Requirements for Die Cast Inserts 42

Chapter 2. 12.7 Laser Fused Spray ToolingMolybdenum powder is predeposited on a steel surface by plasma spraying.

This coating is then fused by a continuous wave Nd:YAG laser (2.31). In addition to

densification of the predeposited molybdenum, alloying with the steel substrate

occurs. Surfaces treated by this method have excellent wear properties. The process is

monitor for sound emissions to determine crack intensity during alloying to evaluate

laser and coating parameters as well as process quality. As long as deformation of the

substrate or die can be controlled, this process could be used to face die inserts.

Another process similar to the molybdenum spray process, is laser-surface

melting (2.30). A 3KW CO2 laser is used melt plasma spray surfaces of ceramics or

metal. These surfaces are produced by transferred plasma jet technology. The laser is

used to improve the homogeneity of plasma sprays which inherently contains voids,

cracks or pores. Plasma oatings100-200 µm thick have been homgenitized with laser

surface melting.

Chapter 2. 12.8 Hot wire laser depositionA laser beam is used to melt a hot wire electrode (2.34). Two millimeter thick

coatings have been built by this system. The high temperature gradients and intensity

of the laser interaction with the wire and substrate causes limited dilution and limited

penetration into the substrate. Corrosion resistant coatings have been made using this

technology. This technology could be used to hard face metal injection molding or

die casting inserts.

Chapter 2. 13 Requirements for Die Cast Inserts

To rapid tool die cast inserts, a designer must understand the material and

functional requirements that the inserts must have in order to successfully produce

castings. Understanding the service requirements of the environments in which the

inserts will be used is essential to designing viable parts.

Die casting results in abrupt thermal and pressure changes on the insert during

the injection of the molten aluminum. An insert for aluminum die casting may

encounter temperature changes from 150oC to 670oC and pressure changes from

ambient to 142 MPa in a cycle time as short as 20 seconds. To cycle through these

changes in temperature and pressure, die casting inserts must be made of material

Page 43: Thesis Glink

Chapter 2 Background

Chapter 2. 13 Requirements for Die Cast Inserts 43

which possess the following characteristics:

•Low Coefficient of Expansion for high thermal fatigue resistance

•High hardness (44-48 Rc) for wear resistance (2.35)

•High modulus of elasticity or impact resistance to avoid deformation

from galling and heat checking

•Moderate thermal conductivity to produce castings of similar micro-

structures as H13 production inserts (on the order of 24 W/mK)

When designing die cast inserts, the material used must be able to survive these

conditions. Also, the material characteristics must have similar thermal properties to

insure proper microstructual development of castings produced by the inserts.

Additional requirements may be added because of processing requirements of laser

based deposition, particularly requirements of the SDM process. Initially the only

additional requirement is high corrosion resistance. SDM deposits sacrificial material

to support undercut features or preserve cooling channels shaped within the die.

Utilizing the 400 series martensitic stainless steel materials, will also add requirement

to insure the production of sound inserts with minimal deformation.

Page 44: Thesis Glink

Chapter 2 Background

Chapter 2. 13 Requirements for Die Cast Inserts 44

2.1 Barkan, P and Iasiti, M. (1993). “Prototyping: A tool for Rapid Learning in Prod-uct Development.” Concurrent Engineering: Research and Applications 1: 125-134.

2.2 Barkan, P. (1991). “Strategic and Tactical Benefits of Simultaneous Engineering.” Design Manufacturing Journal (Spring): 39-41.

2.3 Kalpakjian, Manufacturing Engineering & Technology, 19952.4 Leu, Ming C., “Feasibility study of EDM tooling using metallized stereolithogra-

phy models,” Technical Paper - Society of Manufacturing Engineers, Proceed-ings of the NAMRX XXVI Conference Atlanta, GA, USA

2.5 Kai, Chua Chee, “Integrating rapid prototyping and tooling with vacuum casting for connectors,” International Journal of Advanced Manufacturing Technol-ogy, v14 n 9 1998. pp. 617-623.

2.6 Polosky, Quentin F., Mechanical property performance comparison for plastic parts produced in a rapid epoxy tool and conventional steel tooling, Annual Technical Conference - ANTEC, Conference Proceedings. Special Areas Annual Technical Conference - ANTEC, Conference, Proceedings v 3 1998, p 2972-2976.

2.7 Rahmati, Sadegh and Dickens, Philip, Stereolithography for injection mould tool-ing Rapid Prototyping Journal. Rapid Prototyping Journal v 3 n 2,1997. p 53-60.

2.8 Burns, David T., Malloy, Robert A, McCarthy, Stephen P., Analysis of metal coat-ing effects on stereolithography tooling for injection molding, Annual Techni-cal Conference - ANTEC, Conference Proceedings, Proceedings v 1 1998, p 888-892.

2.9 Janczyk, M., Rapid stereolithography tooling for injection molding: The effect of cooling channel geometry, Journal of Injection Molding Technology. Journal of Injection Molding Technology v 1 n 1 Mar 1997. pp. 72-78.

2.10 Lakshminarayan, U., McAlea, K., Girouard, D., and Booth, R., Manufacture of iron-copper composite parts using selective laser sintering (SLS**T**M), Advances in Powder Metallurgy and Particulate Materials, v. 3, p 13/77-13/85.

2.11 Klocke, F.,Celiker, T., and Song, Y.-A., Rapid metal tooling, Rapid Prototyping Journal. Rapid Prototyping Journal v 1 n 3, 1995. pp. 32-42.

2.12 Hornig, C. and Lohner, A., Direct laser sintering of metal powder, Kunststoffe Plast Europe. Kunststoffe Plast Europe v 87 n 11, Nov 1997. pp. 72.

2.13 Killander, Lena Apelskog, Rapid Mould: Epoxy-infiltrated, laser-sintered inserts, Rapid Prototyping Journal. Rapid Prototyping Journal v 2 n 1,1996. pp. 34-40.

2.14 Pak, Sung S, Prototype tooling and manufacturing through Laminated Object Manufacturing (LOM), International SAMPE Symposium and Exhibition (Proceedings), v43 n 1 1998. SAMPE, Covina, CA, USA. pp. 685-692.

2.15 Warner, Merlin C., Rapid prototyping for die casting: today's applications and future developments, Die Casting Engineer. Die Casting Engineer v 40 n 2 Mar-Apr.,1996. 4pp.

Page 45: Thesis Glink

Chapter 2 Background

Chapter 2. 13 Requirements for Die Cast Inserts 45

2.16 Sachs, E, Williams, P., Brancazio, D., Cima, M., and Kremmin, K., Three-Dimen-sional Printing. Rapid tooling and prototypes directly from a cad model, Pro-ceedings of Manufacturing International '90. Part 4:, 1990, p 131-136.

2.17 Sachs, Emanuel, Cima, Michael, Brancazio, David, Curodeau, Alain, and Shalon, Tidhar, Three dimensional printing. Rapid fabrication of molds for casting, American Society of Mechanical Engineers, Production Engineering Division (Publication) PED. Advances in Integrated Product Design and Manufacturing American Society of Mechanical Engineers, Production Engineering Division (Publication) PED v 47. Publ by ASME, New York, NY, USA. pp. 95-10.

2.18 Rahmati, Sadegh, Rapid Prototyping Journal, Vol. 3, No .2, p. 53, (1997)2.19 Pintat, M. and Greulich, M., Proceedings Solid Freeform Fabrication Sympo-

sium, The University of Texas at Austin, pp. 74 (1995).2.20 Merz, R., Prinz, F. B., Ramaswami, K., Terk, M., and Weiss, L. E, Proceedings

Solid Freeform Fabrication Symposium, The University of Texas at Austin, p. 1, (1994).

2.21 Fessler, J. R., Merz, R., Nickel, A. H., and F. B. Prinz, Proceedings Solid Free-form Fabrication Symposium, The University of Texas at Austin, p.117, (1996).

2.22 Rebholz, C.,Scheibe, H.-J., Schultrich, B., and Matthews, A. Mechanical and tri-bological properties of hard aluminium-carbon multilayer films prepared by the laser-arc technique, Surface & Coatings Technology v107 n 2-3 Sep 10 1998. pp. 159-167,1998.

2.23 Kahlen, Franz-Josef,Kar, Aravinda, Watkins, Tom, and Burl, C., Stress analysis in rapid manufacturing, Laser Institute of America, Proceedings. Laser Institute of America, Proceedings v 83 n 2 1997. Laser Inst of America, Orlando, FL, USA. pp. E76-E83.

2.24 Kreutz, E.W., Pulsed laser deposition of ceramics - fundamentals and applica-tions, Applied Surface Science. Applied Surface Science v 127-129 May1 1998. pp. 606-613.

2.25 McLean, Mark A, Shannon, Geoff J., and Steen, William M., Laser Direct Cast-ing high nickel alloy components, Advances in Powder Metallurgy and Partic-ulate Materials v 3, 1997. Metal Powder Industries Federation, Princeton, NJ, USA. pp. 21-3-21-16.

2.26 Hu, Y.P., Chen, C.W., and Mukherjee, K., Analysis of powder feeding systems on the quality of laser cladding,Advances in Powder Metallurgy and Particulate Materials, Advances in Powder Metallurgy and Particulate Materials v 3, 1997. Metal Powder Industries Federation, Princeton, NJ, USA., p 21-17-21-31.

2.27 Lewis, Gary K., Lyons, Peter, Direct laser metal deposition process fabricates near-net-shape components rapidly, Materials Technology. Materials Technol-ogy v 10 n 3-4 Mar-Apr. 1995. pp. 51-54

2.28 Reisse, Guenter and Ebert, Robby, Titanium nitride thin film deposition by laser CVD, Applied Surface Science. Applied Surface Science v 106 Oct 2, 1996. pp. 268-274.

2.29 Maxwell, J.L., and Pegna, J., Deangelis, D.A., and Messia, D.V., Three-dimen-

Page 46: Thesis Glink

Chapter 2 Background

Chapter 2. 13 Requirements for Die Cast Inserts 46

sional laser chemical vapor deposition of nickel-iron alloys, Materials Research Society Symposium Proceedings. Advanced, Laser Processing of Materials - Fundamentals and Applications, Materials Research Society Sym-posium Proceedings v 397 1996.,Materials Research Society, Pittsburgh, PA, USA. pp. 601-606.

2.30 Pujar, M.G., Dayal, R.K., and Singh Raman, R.K., Microstructural and aqueous corrosion aspects of laser-surface-melted type 304 SS plasma-coated mild steel, Journal of Materials Engineering and Performance v 3 n 3 June 1994. pp. 412-418.

2.31 Haferkamp, H., Gerken, J., Toenshoff, H.K., and Marquering, M., Laser alloying of molybdenum on steel surfaces to increase wear resistance, , Proc 1995 9 Int Conf Surface Modif Technol 1996. Minerals, Metals & Materials Soc (TMS). pp. 547-564.

2.32 Zergioti, I.Zervaki, A., Hatziapostolou, A., Haidemenopoulos, G., and Hontz-opoulos, E., Deposition of refractory coatings by LCVD, Optical and Quantum Electronics. Optical and Quantum Electronics v 27 n 12 Dec 1995. pp. 1377-1383.

2.33 Yellup, J.M., Laser cladding using the powder blowing technique, Surface & Coatings Technology. Surface & Coatings Technology v. 71 n 2 Mar 1995. pp. 121-128.

2.34 Bouaifi, Belkacem and Bartzsch, Jorg, Surface protection by laser beam deposi-tion with hot wire addition, Welding Research Abroad. Welding Research Abroad v 40 n 8, Aug-Sept 1994. pp. 31-33,1994.

2.35 Skoff, J. V., Die Casting Engineer, Vol. 31, n. 1 , p. 58, (1987)

Page 47: Thesis Glink

Chapter 3 Testing Procedures

Chapter 2. 13 Requirements for Die Cast Inserts 47

Chapter 3 Testing Procedures

The investigative approach to designing successful carbon steel inserts or any

parts using SDM laser technology has been to understand, at a fundamental level,

microstructural development. Microstructure and processing conditions have been

shown to be highly correlated to material properties like strength and hardness.

Residual stress development has also been correlated to the microstructure.

Figure 3.1 Microstructure and grain growth.

To begin to understand the microstructure, a number of procedures had to be

Page 48: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 1 Grain Size 48

used to analyze SDM processed microstructure. Measurements of grain size, phase

volume, and material properties were studied to characterize SDM parts. Both manual

and automated processes were used.

Chapter 3. 1 Grain Size As Figure 3.1 illustrates, grain size influences material properties. With

increasing grain size, strength and hardness decrease. ASTM standard E112-96

describes procedures for measuring average grain size. These procedures characterize

two dimensional grain size which is exposed by sectioning test samples. Manual and

automated methods exist to measure grain size. Both methods were used to measure

grain sizes as well as verify technique accuracy.

Chapter 3. 1.1 Point Intercept MethodThree to five horizontal lines are placed across a metallographic image. The

length of the lines calibrated to the same scale as features within the image are drawn.

Next the number of times a grain boundary intersects one of the lines is summed. The

line intercept count (PL) is the number of intersections counted divided by the total

length of the lines.

Figure 3.2 Point Intercept Method

The PL is used to calculate the surface area in a unit volume. It is needed to

estimate the grain diameter. This method can also be accomplished with the use of a

circle. The circle must be larger than the largest grain. Total length is simply the

circumference of the circle. This method, also known as the Hillard method, and

reduces directional bias when counting intercepts.

Chapter 3. 1.2 Automated Point Intercept using Photoshop Plug InA program written by Reindeer Inc. uses pixel value measurements of grain

PL = counts / total length

PL = 5/ (3*100µm)

=.016 counts /µm

SV = 2* PL = .032 counts / µm

D = 8/(3SV) = 83µm

Page 49: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 49

boundary images to measure the equivalent diameter and shape of grains in a

metallograph. The metallograph is prepared by thresholding the RGB (Red-Green-

Blue) image so that black represents the grain and white represents grain boundaries.

Depending on how the image is etched and photographed, the image may be inverted

so that the grain will be thresheld as black and the grain boundaries are white. Other

aspects of this program can measure grain perimeter, grain center of gravity in x and y

coordinates.

Figure 3.3 Grains of SDM Tool Steel Analyzed by Photoshop Plug In

Chapter 3. 2 Volume Measurement and Phase Confirmation

Chapter 3. 2.1 Point Intercept MethodThe ASTM E562-1995 procedure uses a grid of 36 points area to measure

phase volume fraction. The grid is a 6 x 6 square area which is placed over a

metallograph. The phase or features of interest which intersect grid lines are counted.

The number of intercepted grid points divided by the total number of grid points (36)

is an estimate of the volume fraction. The volume fraction of a phase or constituent is

the fraction of the volume of the structure that it occupies.

Accuracy depends upon selecting several grids. For 10% relative accuracy,

625 fields would need to be analyzed for a 2% volumetric phase and 63 fields would be

needed for a 20% volumetric phase using a 32 point grid.

Chapter 3. 2.2 X-ray DiffractionX-ray diffraction was used in two ways: Phase Identification and Stress

analysis. For randomly oriented samples, quantitative measurements of volume

fraction of constituents like martensite, or austenite can be made from X-ray

diffraction patterns because the total integrated intensity of all the diffraction peaks for

10µm Inverted and Thresholded Image

Page 50: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 50

Figure 3.4 Point Intercept Method for Volume Analysis

each phase is proportional to the volume fraction of that phase. Moreover, if the

crystalline phases or grains of each phase are oriented randomly then the integrated

intensity of any single diffraction peak is proportional to the volume of the fraction of

that phase.

Copper A radiation was used for most of the quantitative measurements.

Because of background emissions or limited penetration depth of copper, the beam

count time was extended 30 sec. per .25o.

As figure 3.5 shows characteristic peaks of austenite and martensite existed

Figure 3.5 Analysis of Integrated X-ray Intensity for Phase Volume Analysis.

Point Count, Pp

Pp = 3/36 = 1/12 = 8%

of

C%austenite +C%martensite + C%carbides =1

v = volume of cell unitp=multiplicity factorF = structure factor, F=f(f)e-2M =temperature factor

Assumptions: polycrystalline specimens,

completely filling the incident beam at all angles.

form of flat plate of effectively infinite thickness, randomly oriented grains, making equal angles with incident and diffracted beams

0

50

100

150

200

250

X R

ay D

iffra

ctio

n C

ount

s 300

350

400

4 1 4 2 4 3 4 4 4 5

Two Theta Angle4 6 4 7 4 8

111 Austenite Plane

110 Martensite Plane

IausteniteImartensite

=Raustenite * C%austenite

Rmartensite * C%martensite

R = ( 1v2 )[F2 * p * ( 1+cos2 2Θ

sin 2 Θ*cosΘ)(e−2 M )

Where:

Page 51: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 51

with in an x-ray scan of the SDM tool steel samples. For peak separation, higher

angles were used to calculate integrated intensities to determine phase percentages.

X-ray stress measurements were also made. X-ray methods permit the

determination of the surface stress components which characterize the existing stress

system. X-rays measure biaxial stress within the surface because no stress exists at the

free surface.

Figure 3.6 Full X-ray diffraction Pattern

The relation of stress and strain are shown in Figure 3.7 . Using this formula

we can determine σx by finding the difference of strain in two angle in plane xz, and

σy can be found by measuring strain in yz plane. Strain can be measured by x-ray

diffraction of lattice parameters by the use of Bragg’s Law as shown in Figure 3.8.

Lastly, stress can be approximated by measuring strain at number of angles in a

particular plane and plotting the strain versus the square of the sines of each angle.

From the slope of the line the stress can be attained as shown in Figure 3.9. This

method is often called the Sin2Ψ method.

0

5 0

1 0 0

1 5 030 50 70 90

11

0

Co

un

ts 11

1 M

art

en

site

2theta

11

1A

ust

en

ite

00

2A

ust

en

ite

21

0 A

ust

en

ite 20

0 M

art

en

site

02

2 A

ust

en

ite

11

2 M

art

en

site

22

0 M

art

en

site

31

0 M

art

en

site

21

1 M

art

en

site

31

1 A

ust

en

ite

Counts

111

Aus

teni

te

111

Mar

tens

ite

00

2 A

uste

nite

200

Mar

tens

ite

020

Aus

teni

te

211

Mar

tens

ite11

2 M

arte

nsite

311

Aus

teni

te

220

Mar

tens

ite

310

Mar

tens

ite

30 50 70 90 11090% Martensite5.3% Austenite

Page 52: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 52

Figure 3.7 Relation of Stress and Strain

Figure 3.8 Braggs Law of X-ray Diffraction and Strain

Figure 3.9 Stress Approximation by Sines.

Another method of determining stress measurements is called Fastress. A

Fastress machine was used to determine stress and retained austenite. This system uses

Ψεψ

σxσy εψ = 1/E (σx[(1+v)sin2ψ -v]-vσy)

εψ =Strain in xz plainσx = stress in x directionσy = stress in y directionE = Youngs Modulusv = Poisson’s Ratio

εψ2-εψ1 = 1/E (σx[(1+v){sin2ψ2 -sin2ψ1} ]

σx = (E/(1+v))[1/{sin2ψ2 -sin2ψ1})(ε ψ2-εψ1)

Z

X

Y

Ψεψ

σxσy

εψ = (dψ-do)/doεψ2-εψ1 = (dψ2-dψ1)/do

d = 2 sin θ l λd = lattice spacing2θ = diffraction Angle of X-rayλ = X-ray wavelength employed

If Ψ1 = 0 then:

σx = (E /(1+v))[1/{sin2ψ2)]* (dψ2-dψ1)/do

=Bragg Law

sin2ψ

εψslope = σx (1+v)/E

Ψ1, Ψ2, Ψ3 ...

Page 53: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 53

two chromium targets and four counters. The two x-ray beams are positioned so that

two incident beams strike the fixed specimens at 0o and 45o for simultaneous

measurement. When the counters are centered on the lines of diffraction, a voltage

proportional to the difference of the angular positions is recorded. Since the voltage is

a linear function of (2ψ1-2ψ2), the stress can be measured directly. The Fastress acts as

a stress gauge which must be calibrated with known stresses. It has been designed to

measure stresses in ferritic and martensitic material.

Figure 3.10 Equations of Stress for Fastress Machine

Cross sections of laser deposited tool steels where analyzed by the Fastress

technique to gain a stress profile of the material in the as-deposited condition. The

Sin2Ψ method was used to measure the average stress of a 316L stainless steel deposit

and the average stress of the substrate for the 316L deposit. All measurements were

made with the substrates still attached to the laser deposit.

Chapter 3. 2.3 Scanning Electron MicroscopeScanning Electron Microscopes (SEM) were patterned after light microscopes

and yield similar information about topography, morphology, and composition. SEM

can relay topographical information about the surface features of an object, its texture

or other detectable features limited to a few manometers. SEM can relay

morphological information about the shape, size and arrangement of the particles

making up the object that are on the surface or have been exposed by grinding or

chemical etching. All detectable morphological features are limited to a few

nanometers. Lastly, SEM can relay compositional information about elements and

compounds making up the sample relative to the surface can be determined but limited

to 1 micrometer in diameter area.

Light microscopes can also be used for topographical and morphological

Ψεψ

σxσy

If Ψ1 = 0 then:

σx = (E /(1+v))[1/{sin2ψ2)]* (dψ2-dψ1)/do

With negligible error replace do with dψ2 or dψ1[]σx = (E /(1+v))[1/{sin2ψ2)]* (dψ2-dψ1)/dΨ1differentiate Bragg Law : ∆d/d= - cot θ ∆2θ/2

σx = (E cotθ (2(2ψ1-2ψ2))/{2(1+v))[1/sin2ψ2)]}

Page 54: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 54

information. However, the SEM has a much higher resolution. Resolution with a

light microscope is 0.0002 mm while with a scanning microscope it is 0.000000001

mm. SEM can attain higher magnifications. A light microscope can magnify and

object up to 1,000X, while the scanning microscope may go up to 400,000X.

The SEM images are obtained by using a very small electron probe (or electron

spot) scanning over the surface of the specimens and by mapping the detected electron

signals from each specimen pixel onto the corresponding pixel of the Cathode Ray

Tube (CRT) or Charge Coupled Device (CCD), i.e. the screen. Typically a device

called an electron gun at the top of the apparatus produces a stream of monochromatic

electrons. The stream is condensed by the first condenser lens which is usually

controlled by the “coarse probe current knob”. This lens is used to form the beam and

limit the amount of current in the beam and works with aperture of the condenser to

eliminate the high-angle electrons from the beam. A second condenser lens focuses the

electrons into a thin, tight, coherent beam and is controlled by the “fine probe current

knob” The aperture of the second condenser again eliminates high-angle electrons

from the beam. A set of coils then scans the beam The dwell time is typically

microseconds. The final lens which is the Objective, focuses the scanning beam onto

specimen. When the beam strikes the sample, electrons scatter or excited within the

specimen. These electrons are detected with various instruments and counted. A pixel

value corresponding to the number of counts is displayed on the CRT /CCD. The pixel

intensity corresponds to the counts: (the higher the count the brighter the pixel). This

process is repeated until the grid scan is finished and then repeated, the entire pattern

can be scanned 30 times per second.

For this research, SEM was used to examine fracture surfaces such as

solidification and stress cracks. The ease of specimen preparation, high resolution,

and extensive field of depth make the SEM an invaluable analysis tool. The emissive

mode of SEM which utilizes low-energy secondary electrons emitted from the

specimen surface produces high resolution images contrasting surface roughness or

height of topographic features making the examination of cracks or fracture to within

4 nm.

Samples are mounted in bakelite or conductive SEM plastic mount. If a non-

conductive mount is used, carbon leads are painted from sample to the backside of the

mount to allow for charging of sample for imaging. Sample are polished up to .05 µm

silicate slurry finish.

Page 55: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 55

Chapter 3. 2.4 Transmission Electron MicroscopyTransmission Electron Microscopy (TEM) was used to confirm crystalline

structures of martensite and austenite. In the standard mode, bright field and dark

field images were taken. A bright field image is produced when only the direct beam is

used for image formation. A dark field image is formed when the diffracted beam is

used for image formation.

Lattice images were also taken. These images were used to index diffraction

patterns to confirm the existence of phases. Double diffraction patterns were

differentiated based upon intensities. Angles are measured from the intensity of the

patterns to determine poles and stereographic projects of planes. Bases on determined

planes fcc, bct, and bcc phases were found.

Images in transmission electron microscope form when incident electrons are

scattered by the specimen and focused by one or more electromagnetic lenses.

Electrons scatter elastically, without energy loss (velocity and wavelength remain

unchanged) if they hit the nuclei of specimen atoms. Electrons scatter inelastically

with loss of energy (velocity decreases and wavelength increases), when the orbital

electrons of the specimen atoms are hit. Inelastic events generally involve deflections

through small angles (<10-4 radians) and cause specimen damage . The amount of

scattering is proportional to the thickness of the specimen or atomic number. Thick

specimens, or those with large number atomic numbers, scatter more electrons than

thin specimens or ones with low average atomic number. TEM specimens are typically

thinner than 50 nm. This reduces the number of collisions because most electrons pass

through the specimens without scattering. However, the electrons that do scatter are

sufficient to produce images but without causing specimen damage.

TEM differs from light microscopy techniques in different ways. Optical

lenses are generally made of glass with fixed focal lengths. TEM uses magnetic lenses

which are constructed with ferromagnetic materials and windings of copper wire

producing a focal length which can be changed by varying the current through the coil.

Magnification in the light microscopes is generally changed by switching between

different power objective lenses mounted on a rotating turret above the specimen. or

by changing to different power oculars (eyepieces). The magnification of TEM arises

changing focal lengths but the objective remains fixed. While light microscopes have

small depth of fields, the TEM have large depth of fields allowing for the full sample to

be in focus simultaneously.

A bright field image is produced when only the direct beam is used for image

Page 56: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 56

formation. In other words, unscattered electrons of the incident beam combine with

scattered electrons which have been modified or refocus by passage through objective

aperture. Dark areas in the bright field image arise from specimen regions which

scatter electrons widely and into the objective aperture.

A dark field image is formed when the diffracted beam is used for image

formation. If only scattered electrons are used (unscattered electrons are removed), t a

dark field image is produced. The viewing screen is dark unless there is a specimen

present to scatter electrons. Dark field images typically have higher contrast than

bright field images. However, since the intensity is greatly reduced, longer

photographic exposures required.

The objective aperture can be displaced sideways to intercept the main

unscattered electrons. To study specific crystallographic orientations the apertures may

be placed off-axis or the beam may be tilted. This dark field image may be of poor

quality because the aperture accepts off-axis electrons subject which are subject to

larger aberrations (spherical and chromatic) than those on the optic axis. However, if

the incident electron beam is tilted at such an angle that it is intercepted by the aperture

and the diffracted beam of interest travels down the objective lens axis, only minimum

aberrations exist. These aberrations are similar to those suffered by a bright field

image. However, both of these methods allow only certain diffraction spots/rings to be

transmitted so only specific crystallographic orientations will be highlighted in the

image.

These crystallographic or lattice images expose crystal lattices. Crystals are

composed of groups of atoms repeated at regular intervals in three dimensions with the

same orientation. This group of atoms or the collection of points form is the space

lattice or lattice of the crystal. A crystal lattice can be indexed so that material phases

can be identified.

Chapter 3. 2.5 Electron Microbeam ProbeElectron microprobe analysis (EMPA) is a non-destructive method for

determining the chemical composition. EPMA allows one to quantitatively determine

the chemical composition of nearly all solids on the micron scale. This is achieved by

collecting x-rays that are emitted from atoms which have become “excited” by a

primary beam of electrons. Localized specimen chemistry can be achieved which can

provide insight into the degree of chemical homogeneity which will affect bulk

mechanical, electrical and thermodynamic qualities.

Page 57: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 57

EMPA uses a high-energy focused beam of electrons to generate X-rays

characteristic of the elements within a sample from a volumes as small as 3

micrometers diameter. The resulting X-rays are diffracted by analyzing crystals and

counted by detectors. Chemical composition is determining by comparing the

intensity of X-rays from standards of known composition with those from the

unknown materials. The measurements are corrected for the effects of absorption and

fluorescence within the sample.

EMPA uses an electron beam current from 10 to 200 nanoamps to excite X-

rays. This beam current is about 1000 times greater than the beam used in SEM

analysis. These higher beam currents produce more X-rays from the sample and

improve both the detection limits and accuracy of the resulting analysis. Analysis

locations upon the specimen can be selected by using a transmitted-light optical

microscope mode or SEM mode, which allows positioning accurate to about 1

micrometer. The resulting data can yield quantitative chemical information in a

textural context. Variations in chemical composition within a material, such as a

mineral grain or metal, can be determined.

In this research, EMPA was used to measure chemical composition of material

interfaces and areas with graded material transitions. EMPA was very important in

determining the sharpness of interfaces within the micrometer range.

Chapter 3. 2.6 EtchingTEM and X-ray diffraction are very timely/expensive operations to perform

phase identifications. TEM samples may take weeks to prepare. With one martensitic

stainless steel sample, to dimple the sample for TEM took 2 weeks in the ion mill.

Because of very low penetration in ferrous based samples, when using X-ray

diffraction, a sample must run for about 1 week for a full scan from 40-120o to gather

enough counts to differentiate peaks. The analysis which has gone into this research

has spanned over 300 samples.

Samples were mechanically polished to .05 micron with an AlO3 slurry. The

following etchants were used to analyze the samples to determine phase compositions.

Light microscopy techniques were used to take picture of the samples.

Page 58: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 58

Table 3.1 Etchants Used to Analyze Laser Material

Chapter 3. 2.7 Blue: Automated Volume Analysis A program called BLUE was developed to quantify material phases in steel

samples. These steel samples were etched in different reagents which colored

different crystalline structures with specific colors. Optical light microscope

photographs were taken at 125X magnification to 200X magnification. These

photographs were all taken at similar cyan, magenta, and yellow levels and similar

contrast, brightness and sharpness levels as well. By standardizing the parameters used

during the light microscopy photographs, automated analysis is able to be completed.

This technique allows one to compare the amounts of the different phases among all

the samples.

This automation can occur by using a pixel value analysis technique. An

earlier form of Blue assigns different ranges of grayscale pixel value to martensite and

other two austenite. Calibrated samples which have all ready been categorized for

phases were analyzed. This earlier form of blue is able to predict the amount of

martensite and austenite to within 2% of values reported values.

However, to resolve or separate colored phases of brown (light brown and

dark), blue and red, full RGB color pixel values must be used. Thus, Blue was created.

It is designed to allow the user to standardize an input file to represent all pixel values

of a specific phase. For example, a filter called blue209 contains all pixel values which

correspond to sigma ferrite, which have been revealed by potassium metabisulfide

Etchants Application Technique Phase Exposure

Picral Reagent Swabbed Exposes Grain Bound-aries

Sodium Meta-Bisulfite Immersed Darkens as Quenched martensite

Potassium Meta-Bisulfite Swabbed Darkens All Martensite

Muramaki’s Reagent Boiling Etched Blue-Sigma Ferrite, Brown-Delta Ferrite

Klemm’s Reagent Swabbed Blue - Ferrite, Brown-Martensite

Page 59: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 59

reagent. The optical photomicrograph and the filter are saved as xpm files, which is a

text format of the graphic values. The xpm format allows for programming logic to

analyze the filter and the optical photo . The accuracy of blue is once again within 2%

of X-ray diffraction measurements.

Table 3.2 Comparison of X-ray Diffraction and Blue results for SDMT Steel Sample

In addition to just matching pixels, the beauty of BLUE is that it is able

identify additional area of the phase of interest by identifying a pixels proximity to that

of a given phase. This searching parameter is controlled by user inputs, so the

program can be told how far to deviate from defined pixel values.

Technique Martensite Austenite

X-ray Diffraction 90% 5.3

Blue Analysis 89% 7%

Page 60: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 2 Volume Measurement and Phase Confirmation 60

Figure 3.11 Color Separtion of Blue Technique

Chapter 3. 2.8 Electron Backscattering ProbeIn the final stages of this research, a new technique was discovered which can

used by others in the future who want to understand the material phases within bulk

deposits. Electron Backscattering Probe (EBSP) is a technique similar to SEM. This

technique enables crystal orientation to be determined on thicker surfaces than TEM.

To produce measurements, a stationary beam interacts with the surface of the crystal.

The electrons backscatter in a direction opposite the incoming beam and are captured

by on a phospor screen with a low intensity video camera. Orientations can be

determined by Bragg’s law. Unlike TEM which takes large amounts of time to prepare

carbon steel samples, preparation of EBSP samples are quite simple. Bulk samples are

polished by mechanical means to a .05 µm silica finish. They are etched and then

Delta Ferrite Filter

Sigma Ferrite FilterColor Image

Page 61: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 3 Material Property Test 61

polished again by only the .05 µm solution. Samples can also be quite large (over

1cm2 in area) giving maximum localized orientation information. These

measurements can be automated to get a complete surface mapping of crystalline

information. However, special filters have to be written for BCT formation. Further

testing of this technique for use with carbon samples is undergoing.

Figure 3.12 Comparison of TEM and EBSP

Chapter 3. 3 Material Property Test

To determine the material properties of tools steels deposited by laser

deposition, testing had to occur. Three types of tests which were preformed: tensile

testing, Charpy Impact testing and hardness testing. Tensile testing are able to

measure yield and ultimate tensile strength. Impact resistance a property which is very

necessary for dies inserts can be measured by Charpy Impact testing. Hardness testing

can give information on a more localized area. Microhardness testing was used to

yield layer by layer information or even interfacial informations.

Chapter 3. 3.1 Tensile TestingASTM E8-1998 describes methods for producing samples for tensile testing.

Tensile specimens, when tested, are able to yield material property information like,

yield strength, yield point elongation, tensile strength, elongation and reduction of

area.

BCC/BCT EBSP Pattern

37.757.6

[211] BCTTEM Index Pattern

Beam = [011]

Page 62: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 4 Part Functionality Testing 62

Chapter 3. 3.2 Charpy Impact TestingASTM E23-1998 describes the method of using notched-bar impact testing of

metallic materials by using a Charpy apparatus. It is used to test impact resistance.

Chapter 3. 3.3 ASTM E18-1998 describes Rockwell Hardness testing. ASTM E10-1998

describes Brinell hardness testing. Also, ASTM E92-1982 describes Vickers hardness

testing.

Chapter 3. 4 Part Functionality Testing

To determine if a part can actually be used in service or if the part possess

functionality. For this research three characteristic of functionality were tested:

deflection, wear resistance and thermal resistance. Deflection is an indication of

whether a part is able to meet geometrical requirements. Wear and thermal resistance

are indications of whether a part was able to meet service or environmental

requirements.

Chapter 3. 4.1 Deflection TestingOne measure of part integrity is surface warp or deflection. When layered

manufacturing parts exhibit deflection, it is a symbol of residual stress or shrinkage

imbedded to the part during the deposition or building sequence. A part with low

warpage tends to have low residual stress conditions or has built in a manner to

compensate for shrinkage. Currently, the shape manufacturing process has deflection

measured from the curvature of the substrate on the order of 1-2 mm.

Warpage or deflection is also a concern for the silicon processing industry.

Warp can significantly affect the yield of semiconductor device processing. Producers

and consumers of silicon products use the measurement of warp to determine if

dimensional characteristics of a silicon wafer will satisfy geometrical requirements.

Likewise, warp in layered manufacturing parts, will mean that additional processing is

required to insure that outer part dimensions are within required tolerances. Although

additional processing steps like machining may bring outer dimension back in to

specifications, internal geometry cannot be repaired very easily.

Silicon warpage is measured by ASTM F657-1992. A granite plate is used to

define a surface indexing 3 points on the bottom of the silicon wafer. The maximum

Page 63: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 4 Part Functionality Testing 63

radius of curvature is measured using this plane as a base. Deflection of beam deposits

is measured in a similar way to silicon warpage. For manual measurements, the

substrate and deposit are positioned on fixture which supports the substrate level to

granite plate. The deposit is now pointing towards the granite plate. The now exposed

surface of the substrate is used to take measurements. The edge of the exposed surface

of the substrate is grounded at zero. A height gauge is then moved across the substrate

in a grid with measurements taken every 6 mm. The largest absolute deviation in

height from zero is defined as the maximum deflection.

A coordinate measuring machine at the Stanford Linear Accelerator Center

was used to measure deflection. The beam deposit was again fixtured with the deposit

point down toward a granite plate. A 3 x 24 point grid was used to measure deflection.

This method was very fast and very accurate.

Figure 3.13 Measurement of Deflection

Chapter 3. 4.2 Dunk Test Specimen The Jack Wallace Dunk test is a die casting simulation. This test has been

calibrated so that one cycle is equivalent to 100 die casting shots. A cycle consists of

lowering a prefabricated fixture into molten aluminum (667 oC) while cold water (20oC) is injected into the interior cavity of the fixture. This test is also a measurement of

the thermal fatigue of the material.

D=DeflectionD

Page 64: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 5 Modelling of Results 64

Figure 3.14 Thermal Dunk Test Specimen

Chapter 3. 4.3 Die Casting

Die Casting was described in Chapter 2. 3.9. Die casting was used to test an

insert built by Shape Deposition Manufacturing. This is the ultimate test of insert

viability.

Chapter 3. 5 Modelling of Results A problem as complex as modeling phase transformation in laser deposition

can be attacked in many ways each with limitations. Laser deposition is a 3D

dimensional problem which involves many variables: laser scanning speed, laser focal

length, laser spot size, laser path, the number of laser scans, nitrogen shroud gas flow

rate, cooling rate, material properties, material combinations, substrate material and

thickness, bolting conditions, layer thickness, machining parameters, subsequent

processing steps, process step order, etc. Design of experiments like the Taguchi

approach can be used if judgements can be made to simplify the system components to

key parameters which can be stringently manipulated to plan experiments [3.8, 3.9].

Parameters like scanning speed and laser power can be manipulated for a planned

experiment. It is more difficult and at times impossible to regulate levels of certain

variables like the chromium content of a stock alloy for a planned experiment.

For the laser deposition system, it is at times, difficult to separate endogenous

(input) variables or exogenous (output) variables. Sometimes effects can be masked

because it may take several endogenous variable to produce one exogenous response.

Often DOE’s require a basic understanding of the system relationships to produce

accurate results. Because of these problems: the inability to control certain variable

50.8 mm

177.

8 m

m

177.

8 m

m

50.8 mm 50.8 mm

38.1 mm

Page 65: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 5 Modelling of Results 65

inputs and the daunting inability to separate variables, typical DOE methods may be

inadequate. When a judgment about the magnitude of influence and interelationship

of variables is needed other non-traditional methods may need to assessed. In cases

like the laser deposition system, it may be impossible to make such judgement calls.

Laser deposition is a complex problem which can be looked at as an ill-defined

system because there is insufficient knowledge to throughly define the systems and

key interactions. At present no theory exists which can account for all of the primary

relationships or interaction of variables to define system responses such as phase

formations or grain sizes. Therefore, modeling the system based on a theory driven

approach is difficult and can often lead to errors because of inappropriate assumptions.

Often, when building the model, one has to know things about the system that are

generally impossible to know without extensive testing.

These unknowns may not only cause the inappropriate selection of key

controls or variables, but the model structure can be compromised by the insuffiencient

knowledge about interference factors or influencing factors. This uncertainty can

confuse the modeling in several ways:

1. The selection of variables as endogenous variables or exogenous variables. 2. The

functional form of the relationships between variables and system dynamics. 3.

Proper understanding of the origin or description of error.

A data driven approach can overcome some or all of the problems associated

with ill-defined systems. Often data is analyzed by statistical means for model

formation. However, this type of defining process typically needs to have priori

knowledge about the structure of the system to produce the mathematical model

[3.10]. Data mining techniques can be implemented to gain system knowledge and

modelling information. Data mining techniques include data visualization, tree-based

methods and methods of mathematical statistics like multivariate regressions as well

as those for knowledge extraction from data using self-organizing modelling [3.13,

3.13].

Data mining is an interactive and iterative process of numerous subtasks and

decisions-making steps such as data selection and pre-processing, choice and

application of data mining algorithms and analysis of the extracted knowledge. Many

automated data mining programs try to limit the involvement of users in the overall

data mining process and the inclusion of existing a priori knowledge. Thus, the

process becomes more automated and more objective.

To tackle the phase transformation element of the laser deposition problem

Page 66: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 5 Modelling of Results 66

which is complex and expense (in time, operations, etc.) to solve and to ascertain

primary influences of parameters the Group Method of Data Handling (GMDH)

procedure was used [3.11]. GMDH looks for simple relationships in multiple level

systems.

The goal of GMDH and other similar tools are to predict behavior by means of

parametric or nonparametric models. Parametric models are adaptively created from

data by the Group Method of Data Handling (GMDH) in the form of networks of

optimized transfer functions (Active Neurons). Nonparametric models are selected

from a set of variables analyzing one or more patterns of a trajectory of past behaviors

which are analogous to a chosen reference pattern. Both approaches of self-organizing

modeling include not only core data mining algorithms but also an iterative process of

generation of alternative models with growing complexity, their evaluation, validation

and selection.

At present, GMDH algorithms present a method to identify and forecast

relationships in cases of noisy and short input sampling. In contrast to neural

networks, the results are explicit mathematical models, obtained in a relatively short

time. KnowledgeMiner is a software tool which uses GMHD. It is an easy-to-use

modelling tool which realizes twice-multilayered neuronets and enables the creation

of time series, multi input/single output and multi input/multi output systems (system

of equations). Successful applications are shown in the field of analysis and prediction

of characteristics of stock markets in financial risk control modelling [3.10].

Page 67: Thesis Glink

Chapter 3 Testing Procedures

Chapter 3. 5 Modelling of Results 67

3.1 Kalpakjian, Manufacturing Engineering & Technology, 19953.2 Norton, John T., “Review of Methods of X-ray Stress Measurement,3.3 Kocks, U.F., Tome,C. N. and Wenk H.-R., “Texture and Anisotropy,” New York:

Cambridge University Press, 1998, pp. 167 -177.3.4 E. Beraha and B. Shpigler, Color Metallography, (1977).3.5 B. L. Averbach, L. S. Castleman, and M. Cohen, “Measurement of Retained Auste-

nite in Carbon Steels,” Transactions of the ASM, (1949) Vol. 42, pp112-120.3.6 N. Williams and C. Carter, Transmission Electron Microscopy - Diffraction, Vol 2,

(1996), pp.267-288.3.7 Burgman, Patrick, “Design of Experiments The Taguchi Way,” Manufacturing

Engineering, May 1985, pp. 44-46.3.8 Ross, Philip J., “Taguchi Techniques for Quality Engineering,” McGraw Hill, New

York: 1996.3.9 Ranjit, R. A Primer on the Taguchi Method, Van Nostrand Reinhold, New York, pp.

145-155, (1990).3.10 Lemke, F.; Mueller, J.A. (1997): Self-Organizing Data Mining for a Portfolio

Trading System. Journal for Computational Intelligence in Finance, 5(1997) pp. 3

3.11 Farlow, S.J. (ed.) (1984): Self-organizing Methods in Modeling. GMDH Type Algorithms. Marcel Dekker. New York, Basel

3.12 Madala, H.R.; Ivakhnenko, A.G. (1994): Inductive Learning Algorithms for Com-plex Systems Modelling,CRC Press Inc., Boca Raton, Ann Arbor, London, Tokyo

3.13 Sarle, W.S. (1995): Neural Networks and Statistical Models. in: Proceedings of 19th Annual SAS User Group International Conference. Dallas. pp. 1538-1549.

Page 68: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 1 The Need for A New Material 68

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 1 The Need for A New MaterialAs described in Chapter 2. 10, the SDM process is an additive and subtractive

process which involves the deposition and shaping of material. When used with

metals, part material is deposited by laser deposition and shaped by CNC machining.

A sacrificial material is often used to support over-hanging features or to maintain the

integrity of hollow internal geometries within the part. For example support features

are often used to fill cooling channel passages as well as to support undercut surfaces.

To remove sacrificial material, an etching process is used. This removal process

requires that part materials have high corrosion resistance and sacrificial material to

have low corrosion resistance.

Common materials used to produce parts in Shape Deposition Manufacturing

(SDM) have been 316L stainless steel, copper, copper bronze, and invar. Stainless

steel-316L is a soft machinable material with high corrosion resistance. It is often

used as part material. However, it has a very high coefficient of thermal expansion

(CTE). When 316L is deposited, the melted powder upon solidification shrinks. This

shrinkage cause 316L parts to deform.

Copper is used as a support material. It is often plasma deposited and shaped

to support 316L parts. The low corrosion resistance of copper allows, exposed

surfaces of copper to be etched away and removed. The high thermal conductivity

makes it an attractive option for depositing within multimaterial parts to improve heat

transfer.

Copper bronze can also be used as a sacrificial material, and it can be deposited

Page 69: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 1 The Need for A New Material 69

by laser. Its low corrosion resistance make it ideal for this etching process. It also

soluble with 316L. Thus, for depositing gradient structures, copper bronze is a good

candidate. However, the thermal conductivity is very similar to 316L so no thermal

advantage is gained by using copper bronze within the deposited part.

Invar is used as part material and for graded structures. Functionally graded

parts have been researched by the Stanford Rapid Prototype Lab. Two materials are

blended in the power feeding unit by allowing material mixing in situ by changing

powder feed flow rates. For example, Invar has a low CTE from 400C to room

temperature. It is during this temperature range when most residual stress develops

which causes part deformation. Therefore, invar has been used in graded structures

called functionally gradient parts. The powder feeding system used for SDM consists

of 3 automated powder feeders. Each can be filled with different material and

controlled individually. Figure 4.1 shows the results of depositing both invar and

stainless in over a 48 mm span. The measurements for this graph were taken by

electron microprobe analysis. As one material decreases the other increased. A

smooth compositionally graded structure is created. This technique has been used to

build multi-material parts. The invar is deposited on the interior of the part while

stainless steel is deposited on exterior surfaces or as barriers near low corrosion

resistant materials.

Figure 4.1 Graded Structure of Invar and 316L - Measured by Microprobe

0

0.2

0.4

0.6

0.8

1

0 8 1 6 2 4 3 2 4 0 4 8

Percent Compositon of Stainless-Invar Bar

%316L %Invar

Per

cent

Com

posi

tion

mm 100% Stainless Steel 100% InvarVariation within 95% confidence Level is + .6%

Page 70: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 1 The Need for A New Material 70

As graded material with 316L, Invar has been used to build many parts.

However, Invar has low corrosion resistance and low machineability. Because of these

two properties, production of parts completely from invar is limited.

While each of these materials have been successfully used to produce parts,

each is inadequate to produce die cast inserts. Stainless 316L, Invar, copper, and

copper bronze are not wear resistant enough to survive molten injection of aluminum.

Nickel in 316L and Invar are soluble in molten aluminum. Alloying and precipitation

of die material with casting material produces an unusable material combination.

Copper and copper bronze in addition to solubility, have melting temperatures in the

range of molten aluminum. Mold deformation or dissolution would occur if used in

aluminum die casting. Also none of these materials attain the hardness needed for

wear resistance specified for die casting.

Table 4.1 Current SDM Materials: Balance Iron

Also, in addition to the requirements needed for die casting listed in Chapter 2.

13, the material needs to meet requirements of laser deposition in SDM. Also, any

new material should yield more applicability than just aluminum die cast inserts. The

process to find new materials should attempt incorporating materials which have

properties that would lower part deformation, or improve deposition characteristics,

etc. For example, if it is possible, the new material should have a low CTE or have

improved wetting characteristics so that additional materials can be used with SDM.

C% Mn% Si% Cr% Ni% Mo% Cu% Al% Sn%

Copper 0 0 0 0 0 0 99.9 0 0

Copper Bronze

0 0 0 0 .5 0 86.5 6 .2

Invar .03 .3 .2 0 40 0 0 0 0

316L .02 1.74 .73 17.3 13.1 2.66 0 0 0

Page 71: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 1 The Need for A New Material 71

Table 4.2 Typical Materials used with SDM

Therefore the requirements of materials incorporated in the SDM process to

build aluminum die casting inserts can be divided into to camps, Tier I and II. Tier one

requirements cannot be compromised. Tier II. requirements are additional

requirements which are to be sought but are not decision limiting criteria:

Tier I. Requirements

• Low Coefficient of Expansion for high thermal fatigue resistance

•High hardness (44-48 Rc) for wear resistance (2.35)

•High modulus of elasticity for high resistance to deformation to avoid

galling and heat checking

•Moderate thermal conductivity to produce castings of similar micro-

structures as H13 production inserts (on the order of 24 W/mK)

•Minimal to trace amounts of Nickel to reduce the potential alloying or

dissolution.

•Corrosion Resistance or the ability to alloy with corrosion resistant

materials

Tier II. Requirements

•Good Machineability

•Low CTE

•Commercially available from reliable source

Melting Temperature (C)

CTE(µm/m•K)

300oC -20oC

Copper 1083 16.9

Copper Bronze

1053 16.2

Invar 1427 1.6

316L 1385 17.5

Page 72: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 72

Chapter 4. 2 Initial Selection of Tool Steels

The hardness and strength requirements were the guiding criteria to select

materials. The typical tool building material for SDM was 316L stainless steel. This

is an austenitic steel which has the crystal lattice structure of face-centered cubic

(FCC). A FCC alloy has many good characteristics, low-temperature toughness and

excellent weldability, and typically good corrosion resistance. However, they are often

susceptible to stress-corrosion cracking. However, the main problem is that the

relatively low yield strength allows that these FCC alloys can only be hardened by

coldworking, precipitation, or solid solution strengthening. Precipitation and solution

hardening involve special heat treatment processes to either change the bulk to one

phase or cause precipitation of a second phase in the part matrix. All three treatments,

would be difficult to perform on a SDM 316L part especially if it were a multi-material

part (with Invar and copper) because of the multiple melting point /thermal phase

characteristics or deformation to final part dimensions.

The low yield strength was actually one of the guiding principles for choosing

316L. When one looks at beam deformation, maximizing R or the radius of curvature

would minimize the amount of warp or deflection which the deposited structure would

have. Stainless-316L has a high E/σ ratio.

Figure 4.2 Maximizing R to Minimize Deflection

y

R

Maximize R by Maximizing

Page 73: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 73

Table 4.3 Comparison of E/σ with Common SDM Metals

This criteria though does not seem to help when trying to find hardenable

material for build die inserts. Two types of materials which did meet the hardness

criteria were the 400 series martensitic stainless steel and tool steels. The 400 series

martensitic stainless steels stainless. Theses material have high modulus, can attain

high hardnesses. They have moderate to high corrosion resistance. The high

hardnesses does pose a challenge for machining, but it has similar machining

characteristics to 316L. Therefore, the expertise of depositing and machine stainless

could be exploited for these new materials.

Tool steels will also be able to attain the hardnesses required for die casting

inserts. However, the corrosion resistance of these materials was questionable.

Chromium which forms carbides that reduces corrosion susceptibility is low in many

tool steels. Also many allowing agents are also included in the composition of tool

steels.

Eight materials were originally tested. The alloys consisted of two high speed

tool steels T15 and M2, H13 a hot working steel and five industrial alloys similar in

carbon and chromium compositions to the 400 series stainless: Polar, Cryotherm,

Pyro 1, Pyro 2, Spraco 2.

E x104 MPaσ MPa

(Annealed)E/σ

Copper 110.4 69 15977

Copper Bronze

96.5 207 4660

Invar 14.1 276 511

Low Carbon Steel

20.0 558 358

316L 19.3 234 825

Page 74: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 74

Table 4.4 Initial Tool Steel Selection for SDM Processing

The materials were tested for hardness, deposition quality and corrosion

resistance. The material was SDM deposited in samples measuring 30 mm x 35 mm x

4 mm were deposited from each of the metal powders. Each sample was produced

using standard deposition parameters of 100% laser power, 25 mm/sec laser scanning

rate, and 30 g/min. powder feed rate. Each sample was evaluated on the basis of

deposition quality, hardness, surface wetting and acid etch resistance. Deposition

quality was evaluated through visual observation of cracking and surface oxidation.

Hardness testing was performed with a Rockwell Hardness Indenter (ASTM E18-

1995) and 10 individual tests per sample. Etch resistance testing was determined in

accordance to ASTM A262-93a-1996 by placing the sample in 65o C nitric acid for 30

minutes. After an etch test, the microstructure of each sample was analyzed for

corrosion susceptibility based on etch structure classification (ASTM A262-93a-

1996).

C% Mn% Si% Cr% Ni% Mo% Co% W% P% V%

T15 1.6 0 0 4 0 1 5 13 0 5

M2 .8 .3 .3 4 0 0 0 6 0 2

Polar .01 0 0 10.5 0 0 0 0 0 0

Cryo1 .01 0 0 15 0 0 0 0 0 0

Pyro1 .01 0 0 12 0 0 0 0 0 0

Pyro2 .01 0 0 14 0 0 0 0 0 0

H13 .4 0 1.05 5 0 0 0 0 0 1.1

316L .02 1.74 .73 17.3 13.1 2.66 0 0 0 0

Page 75: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 75

Table 4.5 Testing Results of Tool Steel in SDM Processing

Spraco 2 was the only alloy to successfully meet all the requirements of

hardness, etch resistance and deposition quality. This alloy was selected as the SDM

Tool Steel. Analysis of photomicrographs of the Spraco 2 deposit after the nitric bath

showed no ditch structures or end grain pits which would indicate susceptibility to

intergranualar corrosion. M2 and H13 steels can only be considered if they can be

used with corrosion resistant functionally graded structures such as a structure with

H13 on the outside and 316L deposited upon faces of etchable surfaces.

Alloy Composition Hardness Deposition Etching

T15

4% Cr, 5% Co, 13%W

1.6% C, 5% V, 1% Mo 63 Rc Pitted Fast

M2

4% Cr, .3% Si, 6% W

.8% C, 2% V, .3% Mn 58 Rc Good Fast

Polaris

Industrial Alloy

10.5% Cr 42 Rb Good Slower

Etching

Cryo 1

Industrial Alloy

15% Cr 40 Rb

Visible

Cracking Slower

Etching

Pyro 1 12% Cr, 0% Co <1%

C

30 Rc Cracking Slow

Pyro 2 14% Cr, 0% Co <1%

C

44 Rc

Good

(Cracked

during cooling)

Slight,

Very Slow

Spraco 2 16% Cr, 0% Co <1%

C

40 Rc Good None

H13 5% Cr. 0% Co .4%C 50 Rc Good Etching

Page 76: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 76

Spraco tool was selected as the new SDM tool steel. Metallographic analysis

showed that using the deposition parameters, as for 316L were not sufficient because

some deposits had many voids. To reduce the voids and create an efficient deposition

routine, the deposition parameters had to be optimized.

Chapter 4. 2.1 Optimization of Deposition Parameters

A Taguchi based design of experiment was used to find optimal deposition

parameters for SDM Tool steel. The optimal parameters should provide the maximum

material deposition rate which produces fully dense deposits and material within the

acceptable hardness range. Remelting of the substrate should also be kept to a

minimum to preserve the geometry of previous layers. Three factors were

investigated: laser power, laser scan speed, and powder feed rate. A Latin square array

of 8 experiments varying the three factors at two levels was completed.

Table 4.6 Experimental Parameters

The full factorial design was chosen to directly characterize the two factor and

three factor interactions. Tool steel samples of 30 mm x 10 mm x 5 mm were

deposited at each of the 8 experimental settings. Hardness was measured as described

in Chapter 3. Metallographs were taken at optical magnification from 50-1000x.

From these photomicrographs, the laser penetration depth, porosity and pore density

were calculated. D. Gentry, et al. (4.1). showed that optical measurements of porosity

were within 93% of the accuracy of traditional porosimetry methods.

A Taguchi function for maximizing the outcome, F was chosen to correlate the

observed measurements with the optimal condition (4.2). The optimal setting is found

when F is maximized with the following equation :

F = .4 H* + .4(1- P) +.2 D*

Parameter Level 1 Level 2

Laser Power 100% 90%

Laser Scanning Rate 25 mm/sec 35 mm/sec

Material Feed Rate 30 g/min 15 g/min

Page 77: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 77

where H* is the Hardness normalized by 48 Rc, Pi is the porosity, D* is the depth of

penetration normalized by 100 µm. Once the optimized results were obtained, tensile

specimens were deposited. ASTM Standard E8-1995 was used for the tensile testing

of specimens.

The factors, (A = Laser Power, B = Laser Scanning Rate, and C = Powder Feed

Rate) contribute to the optimal deposit conditions as described in . Using the

optimizing function, (equation 1), the following deposition parameters were selected:

100% laser power, 35 mm/sec scan rate, and a powder feed rate of 15 g/min.

According to the data, the laser scan rate is the single most important factor in the

optimizing function, as well as for controlling hardness, penetration depth, and

porosity individually. Specifically, the faster the scanning rate within the limits

examined, the harder the material, the lower the porosity and the lower the penetration

depth. Laser power independent of the other factors made little contribution to the

outcome.

Table 4.7 Analysis of Variance Table for Optimal Response: A = laser power, B = scan rate, C = powder feed rate

The signal to noise ratio of the data is 37.9 and mean squared deviation is

.0002. These value indicate a minimal scatter of data and a highly reliable optimum

(4.1).

Degree

of

Freedom

Sum of

Squares

Mean

Squares

(Variance)

Variance

Ratio

Pure Sum of

Squares S'

Percent

Contribution

A 1 0.0006 0.0006 1.23 0.0001 Pooled

B 1 0.07 0.07 143.27 0.07 47.39

C 1 0.0001 0.00007 0.13 -0.0004 Pooled

AxBxC 1 0.05 0.05 96.72 0.05 31.88

AxB 1 0.004 0.004 7.23 0.003 2.08

AxC 1 0.02 0.02 36.19 0.02 11.72

BxC 1 0.004 0.004 7.45 0.003 2.15

Total 15 0.2 0.0000

ERROR 0 0.004 0.0005 1.00 0.008 4.78

Page 78: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 2 Initial Selection of Tool Steels 78

Chapter 4. 2.2 Material PropertiesAt this optimized level, the material structure and properties were analyzed.

Metallographs were made of cross sections of the SDM tool steel deposits made at the

optimum conditions.

The metallograph shown below shows a cross section of SDM Tool steel etched with

prical acid. The material structure is primarily martensite (dark, fine regions) with

retained austenite (light bands) . The material exhibited a hardness of 45 Rc. and was

found to be 99.9 % dense. The pore density is .002 pores/ mm2 with an average pore

diameter of 29 µm (4.3). The mechanical properties from tensile testing are shown in

Table 4.8 compared with standard H13 tool steel.

Table 4.8 SDM Tool Steel (as deposited) Vs. H13 (tempered)

As compared to H13 which has been heat treated , SDM tool steel has a lower

yield and tensile strengths and less elongation (4.4). However, these properties for the

SDM Tool steel are in the as deposited condition. For prototype inserts, these

property levels may be sufficient. The high modulus of the SDM tool steel is a result

of the martensitic matrix. The retained austenite should add increased toughness and

reduce the possibility of cracking or galling (4.5).

Mechanical Property SDM Tool Steel H13

Ultimate Tensile Strength 879 MPa 1482 MPa

.2% Offset Yield Strength 510 MPa 1344 MPa

Reduction of Area 2% 38%

Modulus of Elasticity 254 MPa 210 MPa

Elongation (in 25 cm gage

length)

2% 14 % (50 cm gage

length)

Page 79: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 3 Building A Small Scale SDM Die Cast Insert 79

Figure 4.3 Metallograph of Austenite (light) and Martensite (dark) in SDM Tool Steel

Chapter 4. 3 Building A Small Scale SDM Die Cast InsertTo test the SDM process and the new SDM tool steel, a die cast insert needed

to be built. To minimize the possibility of failure within a die casting machine, a small

scale insert was built. To design the insert a set of guidelines was imposed. The both

the part thickness and feature depth would be minimized while maximizing the ratio

between the two specifications. These constraints accomplish two things, the

deposition time would be minimized and the complexity of the die insert would also be

simplified. A ratio of 4 to 1, part thickness to feature depth was maintained.

An insert was built using the optimal parameters for SDM Tool steel and

cycled in an overflow of a production die set in a 600 ton die casting machine. The

insert, the Stanford paper clip (Figure 4.4) had outer dimensions of 6.4 cm x 6.4 cm x

1.9 cm. By using the SDM additive and subtractive processes, the insert was

deposited and fabricated in layers. A pocket of aluminum bronze, 1 cm x 1 cm x.32

cm, was deposited in layers .32 cm below the S feature of the paper clip. The

aluminum bronze was added to illustrate the multi-material capability of SDM.

100 µm

Page 80: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 3 Building A Small Scale SDM Die Cast Insert 80

Figure 4.4 SDM Tool Steel Insert

A number of tests were performed on the insert prior to die casting. Melted

wax was poured in the die to test part removal. Thermal fatigue testing was performed.

The insert was thermally cycled by pouring molten aluminum 380 alloy at 670oC into

the insert, allowing the casting and insert to air cool to about 100oC and removing the

casting. The insert was then air cooled to room temperature. This would be repeated

seven times with different amounts of insert preheat.

The original 5o taper was found to be insufficient for the removal of the wax

from the insert. Ejector pins, gating, and venting were added to facilitate material

removal (Figure 4.7). The insert was tested for thermal fatigue resistance with gravity

poured aluminum.

Figure 4.5 shows the thermal cycling history of the insert. Aluminum 380

castings without warpage or dimensional shifts, proper surface finish and low porosity,

were produced with an insert preheat of 660oC and a 5 minute dwell time in the closed

stand. Figure 5 compares the microstructures of the gravity poured insert and a die

cast part from a traditional H13 insert. The gravity poured parts showed more

columnus structure due to dwell times on the order of 25 times longer. However, the

hardness of the castings from the SDM insert and H13 insert are similar (55 Rb and 57

Rb, respectively).

Page 81: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 3 Building A Small Scale SDM Die Cast Insert 81

Figure 4.5 Gravity Casting produced Thermal Cycling

The die casting machine parameters used to produce castings are listed in Table

4.9. Castings were evaluated microstructural homogeneity under the optical

microscope and compared to castings produced by an H13 insert. The average grain

diameter was calculated by the intercept method in accordance with ASTM Standard

E112-1995.

Table 4.9 Die Casting Parameters

Parameter Setting

Shot Pressure (No Intensifier) 103 MPa

Intensifier Pressure

27.6 MPa

Die Preheat (2 hours) 121oC

Die Temperature while running 82oC

30

130

230

330

430

530

630

730

0 50 100 150 200 250T

empe

ratu

re (

C)

Time (Min.)

Page 82: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 3 Building A Small Scale SDM Die Cast Insert 82

Figure 4.6 A. Die Casting of Aluminum B. Gravity Cast Structure

For actual die casting, the features were remachined with a 15o taper to

facilitate easy removal of the part without ejector pins. The insert was positioned in an

overflow of a valve body die set where ejector pins could not be used. The increased

taper angle resulted in sharp thin edges in the “S” feature as shown in Figure 4.7.

a b

Figure 4.7 Remilled Insert: 15o Taper added to all walls (a) and close up of fine f

Figures 4.8 and 4.9 show, the position of the insert in an overflow section of a

production H13 die casting tool. The cycle time for producing the casting shown in

Figure 4.10, from injection of the aluminum to ejection of the casting was 70 seconds.

The SDM insert was run in the die cavity for over 150 shots with no visible evidence

of wear, fatigue, or cracking.

Both the castings from the SDM insert and H13 overflow sections had a

hardness of 57 Rb. Metallographs of a castings from the two sections are shown in

Figure 4.10.

A. B.60 µm 60 µm

10 mm 1 mm

Page 83: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 3 Building A Small Scale SDM Die Cast Insert 83

Figure 4.8 SDM Insert in Overflow

Figure 4.9 SDM Insert placed in overflow of Saturn Valve Body

have a similar structure with agglomerated cell patterns. The ASTM E112-1995

intercept method was used to calculate average grain diameter. The average grain

diameters for the castings from the SDM insert and the H13 insert were 8.5 µm and

8.6 µm, respectively.

Chapter 4. 3.1 SDM Tool Steel use for small inserts The SDM tool steel used for the die casting insert exhibited higher hardness

than expected for this material. This material deposited through flame spray or

conventional welding operations will have a hardness in the range of 20-28 Rc. The

material experienced rapid quenching during air cooling which initiated a martensitic

transformation without going through the other diffusion limited phases. The

martensitic phase is responsible for the high hardness of this low carbon alloy. Alloys

with higher carbon content should produce material with even higher hardnesses, but

10 mm

Page 84: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 4 Deflection Vs. SDM Tool Steel 84

such materials would be difficult to mill. The low carbon content in SDM tool steel

will allow the SDM process to produce structures within the range of 35-45 Rc.

The corrosion resistance of the SDM tool steel assures that the rapid

transformation is not causing sites for intergranular corrosion (4.6). Certain heating

conditions can allow chromium to be depleted from the alloy matrix and form

chromium carbides on the grain boundaries. This depletion weakens corrosion

resistance. This condition is corrected by tempering and other heat treatments. The

rapid cooling does not allow this precipitation to occur. This is one reason that SDM

tool steel artifacts could be used without heat treatments. In addition to giving good

corrosion resistance, the high chromium content will also give the insert high heat

resistance up to 1000oC (4.7).

These experiments show that SDM is a viable method for rapidly producing

die casting inserts. SDM die cast inserts can produce time savings over conventional

tool and die methods. Multi-material dies with better heating properties can be used

to shorten dwell times and still attain proper cell structures. Moreover, the

possibilities of producing small prototype die inserts “as deposited”, without heat

treatment, can reduce die production lead times further.

Figure 4.10 Die Cast Part from the SDM Mold.

Chapter 4. 4 Deflection Vs. SDM Tool Steel

An interesting observation was made with beams deposited with SDM tool

steel. One beam was built at 20 mm/s and another at 30 mm/s. The beams were 127

mm x 4 mm x 12.5 mm on a 152 mm x 52,4 x 6.25 mm substrate. The beam built at 20

mm/s had an average hardness of 46 Rc, and the beam built at 30 mm/s had an average

hardness of 38 mm/s. The beam seemed to possess less deflection with increasing

Cast from H13InsertCast from SDM Insert

30 µm

Page 85: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 4 Deflection Vs. SDM Tool Steel 85

hardness. This is exactly the opposite effect seen in 316L beams. Increasing hardness

in 316L beams seamed to increase deflection.

Figure 4.11 Increasing Hardness reduced deflection in SDM Tool Steel

The test suggests that increasing hardness may have influence upon deflection.

Further testing would have to help calibrate whether the change in deflection was truly

the result of hardness or whether it can be attributed to smaller grain size, phase

percentage or other factors which can contribute to increased hardness.

Figure 4.12 Increased Hardness increased deflection in 316L beams

-0.2

0

0.2

0.4

0.6

0.8

20 40 60 80 100 120 140

SDM Tool Steel (46 Rc )

SDM Tool Steel (38 Rc)

1.5

1.0

0.5

0.0

Distance Along Beam (mm)

Def

lect

ion

(mm

)

-0.2

0

0.2

0.4

0.6

0.8

20 40 60 80 100 120 140

Stainless Steel (316L) (90 Rb)

Stainless Steel (316L) (86 Rb)

defle

ctio

n (m

m)

distance (mm)

Increasing Hardness

1.5

1.0

0.5

0.0

Distance Along Beam (mm)

Def

lect

ion

(mm

)

Page 86: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 5 Attempting to Build Larger Scale Inserts 86

Chapter 4. 5 Attempting to Build Larger Scale Inserts

The unheat treated SDM die insert made from the alloy known as SDM tool

steel work well with the small scale die insert, producing over 150 parts without any

signs of wear or cracking. However, when the a much larger insert was attempted, the

Dunk test specimen, catastrophic failure occurred. This insert has a part thickness to

feature level thickness of 177 mm to 171 mm. In an attempt to minimize the build or

deposition time, the insert was deposited on its side, changing the deposition part

thickness to feature ratio to 55 mm to 40 mm. A dunk test specimen was being built in

1mm thick layers at the optimized setting for SDM tool steel. Every 10 layers, the part

was allowed to cool before more deposition occurred. With only 5 mm left to be built

the 50 mm part fractured during deposition.

Figure 4.13 Cracked Deposit: 50 layers of SDM Tool Steel Deposited

Metallographic examinations were performed upon crack surfaces. Etchants

were used to determine the amounts of martensite, austenite, and ferrite were located

in the area of the crack. The first interesting finding was the stridation of the regions of

tempered to as-quenched martensite. The grain size of the tempered martensite was on

average 3 times larger than grains within the as-quenched region. The tempered

region has cementite precipitation on the former austenitic boundaries of the tempered

martensite. The larger tempered martensite packets are an indication of extended

reheat temperatures or extended heating times. Very close to the crack sigma ferrite

stringers were found. Also along the crack face sigma ferrite seemed to be propagated.

The metallographic examination helped us to determine that the heating

characteristics of layers is retained as the deposit is built. Tempering of layers occurs

but is not complete. Grain size differences is quite apparent. The initiating crack

surface started in area of brittle martensite. Moreover, sigma ferrite was found along

the crack interface. It is difficult to determine if sigma ferrite existed before crack

Page 87: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 5 Attempting to Build Larger Scale Inserts 87

failure. However, the appearance of sigma ferrite does indicate under certain

deformation or heating conditions sigma ferrite can develop.

It became quite obvious that by building parts with tools steels, new designing

or planning rules are required. Unlike our experience with austenitic material, these

carbon steels can develop multiple phases. Our understanding of these phases and

phase development would also need to be challenged.

Figure 4.14 Microstructure of the cracked layers of SDM tool steel deposit

Figure 4.15 Sigma Ferrite has developed along crack edge

1 mm

Dark regions representas-quenched martensite.

Light regions representtempered martensite and other phases.

100µm

Delta Ferrite formed oncrack interface.

Delta Ferrite Stringerwithin deposit.

Page 88: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 5 Attempting to Build Larger Scale Inserts 88

Figure 4.16 Close up of tempered and as-quenched layers of SDM deposit.

When examining the crack surface in the interior of the part, voids were also

discovered. This void growth had not been encountered before. Smaller samples were

then built at the previously optimized levels. Void growth was now at 30%. When

samples of the original powder was tested at the optimal settings, void percentage was

under 1%. The powder characteristics or composition had changed.

100 µmAs-Quenched

Tempered Martensite and

Martensite

Other phases

Page 89: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 5 Attempting to Build Larger Scale Inserts 89

Figure 4.17 500 and 1000 X Views of Cracked SDM Deposit.

Chemical analysis showed that the silicon content of the powdered alloy had

changed from initial specifications. Also, the particle size had become coarser. When

refining the powder mesh to between 90 µm-120 µm, the void percentage reduced or

was eliminated. However, instability of the powder source and cleanliness of powder

was now questionable. A new material for carbon steel deposition would have to be

developed. However, the profoundly different microstructure of carbon steels,

required that a more fundamental understanding of the deposition process needed to

developed.

20 µm40 µm

40 µm20 µm

Page 90: Thesis Glink

Chapter 4 Material Selection and Design of Small Scale Die Inserts

Chapter 4. 5 Attempting to Build Larger Scale Inserts 90

.

4.1 Gentry, D., Humbert, L., and Burlot, Rene, C. R. Academy of Science, Paris, Vol 309:II, p. 1481, (1989)

4.2 Ranjit, R. A Primer on the Taguchi Method, Van Nostrand Reinhold, New York, pp. 145-155, (1990).

4.3 Homand-Etienne, F., and Houper, R., International Journal of Rock Mechanics and Mining Sciences and Geomechanics, p. 125, (1989)

4.4 ASM Engineering Property of Steel, ASM, pp. 458-461, (1982).4.5 Wayman, Michael, The Metals Black Book: Ferrous Metals, Vol. 1., 2nd Edition,

Cast 1 Publishing Inc., Edmonton, Alberta, pp. 14-37, (1995).4.6 Metals Handbook, ASM, Vol 2, pp. 253-254, (1974).4.7 Metals Handbook, Lyman, Taylor, Vol. 2, 8th Edition, American Society of Metals,

Metals Park, Ohio, pp. 221-283, (1964).

Page 91: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 4. 5 Attempting to Build Larger Scale Inserts 91

Chapter 5 Characterization Tool Steels

In this chapter, the characterization of carbon steels on a layer by layer basis

was preformed to understand the cell and layer morphology of depositing tools steels.

A generalization of carbon steel (>.05%C) morphology in layer manufacturing is

made which can be aid in designing and building carbon steel parts. The term carbon

steel is used to designate the set of steels characterized. The austenitic steel 316L is

also characterized to serve as a comparison. New tools for determining microstructural

phases are adapted for layered manufacturing with laser deposition: Shaeffler Welding

Diagram and the Solidification Modes Diagram.

When originally attempting to deposit carbon steels, the same optimization

strategy designed for laser deposition of 316L stainless steel and aluminum bronze

was used. This strategy emphasizes trying to maximize layer thickness and minimize

the build completion time. Using this laser deposition strategy which emphasized

maximizing throughput was simply inadequate for depositing carbon steels because of

phase transformations. Phase transformations of materials had not been encountered

before. In order to improve tool steel deposition a more fundamental understanding of

the deposition of tools steels from a microstructual level needed to occur. Also, since

SDM Tool (SDMT) steel was no longer the most reliable powder choice, more

materials had to be tested. The opportunity to investigate higher carbon materials

would be beneficial to the quest to understand the affect of parameters on material

phases and layered deposition microstructure.

Page 92: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 1 The Addition of 400 Series Stainless Steels and Other Carbon Steels92

Chapter 5. 1 The Addition of 400 Series Stainless Steels and Other Carbon Steels

With the quality of SDMT alloy now in question, the opportunity to change or

add materials to the characterization search was now presented. Martensitic stainless

steel alloys can attain the hardnesses need for die cast tooling, and were readily

available. Multiple sources of powder could be found. The 400 series stainless can

possess several different quantities of carbon, so the possibility of developing a

understanding of the relationship of material constituents to factors like hardness or

strength now existed. For conventional processes, like quenching and tempering,

correlations currentlt exist to quantify the effects of material composition upon

hardness levels. Thus, by using the 400 series, one could develop a more meaningful

understanding of carbon steel deposition than just developing a carbon steel for die

casting. The composition of tested martensitic steels are compared against the other

carbons steels investigated and 316L in the table below:

Table 5.1 Carbon Steels: Martensitic, Hot Working, Tool Steels, and Commercial Alloys

When one looks at the E/σ ratio for these steels, we can see that only one has a

ratio higher than 316L stainless. However, most are much worse:

C%Mn%

Si% Cr% Ni%Mo%

Nb%

Al% P% W V S%

410 .06 .17 .53 12.5 .07 0 0 0 .017 .007

420 .45 .49 .54 13.6 0 0 0 0 .017 .007

431 .18 .69 .57 15.6 1.78 0 0 0 .016 .003

SDM Tool Steel I

.01 1 1.23 21.5 2.8 0 .84 0 0 0

H13 .4 0 1.05 5 0 0 0 0 0 0 1.1 0

M2 .8 .3 .3 4 0 0 0 0 0 6 2 0

Rock .15 16 1

316L .02 1.74 .73 17.3 13.1 2.66 0 0 0 0

Page 93: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 1 The Addition of 400 Series Stainless Steels and Other Carbon Steels93

Table 5.2 Comparison of E/σ of Tool Steels

The coefficient of thermal expansion of these materials are lower than 316L but

not lower than Invar in the 20oC - 400oC range. It is during this range the residual

stress begin to develop. At temperatures above this range, the yield strength of the

material has been exceeded so the plastic deformation of the material relieves any

additional stress accumulation. SDM tool steel is similar in composition to 440C.

Thus, 440C is included to approximate SDM tool steel’s coefficient of thermal

expansion.

The corrosion resistance of most of the 400 series stainless is a bit lower than

316L. From the initial tooling search, a criteria of at least 12% chromium had been

designated to insure corrosion resistance. However, since SDM has the capability of

building gradient structures, the ability to deposit a high corrosion resistant material

next to a low corrosion resistant material, this concern was removed.

E x104 MPaσ MPa

(annealed)E/σ

410 22 241 913

420 20 345 580

431 20 665 301

M2

H13 21 1250 168

SDM Tool Steel

25.4 510 498

316L 19.3 234 825

Page 94: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 1 The Addition of 400 Series Stainless Steels and Other Carbon Steels94

Figure 5.1 Comparison of the Coefficient of Thermal Expansion of Tool Steels and SDM Mate-rials

For example, suppose cooling channels need to be deposit in a tooling insert.

Typically, the part would be planned with copper (the sacrificial material) being

deposited as the cooling channel and the hard carbon steel deposited around the

copper, forming the walls of the cooling channel. However, during the etch removal of

the copper, some of the low corrosion resistant carbon steel would also be etched.

However, if 316L stainless is deposited as a barrier between the copper and carbon

steel interface, the carbon steel interior of the die insert would be protected. Figure 5.2

shows the cooling channel arrangement. Figure 5.3 shows that it is possible to achieve

smooth gradients by changing the percentage of each metal during deposition. The

figure shows a graded deposition with SDM tool steel and 316L by laser deposition.

The dotted line represents the composition as detected by electron microprobe

measurements and calibrated to represent the increasing carbon content. The solid line

represents the hardness of carbon steel alloys with similar compositions. The smooth

transition enables the use of martensitic stainless or even lower corrosion resistant

carbon steels, like H13.

0

5

10

15

20

0 200 400 600 800 1000 1200 1400

Coefficient of Thermal Expansion

CT

Eµm

/(m

K

)

Temperature (K)

316

410

420440C

414430

Invar (Fe -36Ni)

Page 95: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 2 Differences in Depositing Carbon Steels 95

Figure 5.2 Graded Structure with Cooling Channel

Figure 5.3 Graded Deposition of 316L and a .15% Carbon Steel

Chapter 5. 2 Differences in Depositing Carbon Steels Our understanding of deformation, grain size, cell morphology, and other

aspects of laser layered manufacturing had developed over many years of laser

depositing invar, aluminum bronze, and 316L. However, none of these materials

undergo solid state phase transformation during laser deposition. These materials

remain face centered cubic or body centered cubic.

Chapter 5. 2.1 Stress States of Carbon SteelsTwo noticeable aspects of deposition changed: stress state and hardness. In

316L, tensile stresses, measured by X-ray diffraction, were located within the deposit

and compressive stresses were measured within the substrate. This is predominantly

from solidification shrinkage. The melted 316L powder will shrink upon solidification

and cooling. The coefficient of thermal expansion insures that the shrinkage will be

A A

Section A

Copper

316L

CarbonSteel

Die Insert

Cooling Channel

0

10

20

30

40

50

0 0.05 0.1 0.15

Hardened StainlessSS-Tool Steel Deposition

Roc

kwel

l C

% C

100% 316L100%

Carbon Steel

Page 96: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 2 Differences in Depositing Carbon Steels 96

upon the order of the thermal gradient. However, since the deposit is constrained by

the substrate it is placed in tension. The substrate is simultaneously placed in

compression.

Figure 5.4 Material Shrinkage

However, in carbon steels, some solid-state phase transformations can induce

either tension or compression within the laser layered deposits. Stress measurments

of cross sections of laser deposited 316L stainless steel and carbon steels were taken

by use of the Sin2Ψ and Fastress methods, respectively [Section 3. 2.2]. Figure 5.5

shows the results for the stress testing. Steels with percentage weight carbon between

.2-.8% can impose compressive stresses within the deposit and tensile stresses within

the

Figure 5.5 Stress development in SDM Deposits

substrate. When considering multiple phase depositions, one can consider it having a

As the materialshrinks the molecules

of the deposit are placed in tension

These tension forces cause

the moleculesof the substrate

to be in compression.

because it is constraint by substrate.

- 4 0 0

- 2 0 0

0

2 0 0

4 0 0

- 4- 3- 2- 101234

.2% C -Stainless Steel

.8% C - M21.7% C - Grey Cast Iron.1% C 316L

Str

ess

(M

PA

)

Depth (mm) - Note: 0 is substrate-deposit interface

σzz

Cross Sectionof Layers σ z

z

Page 97: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 2 Differences in Depositing Carbon Steels 97

matrix phase and inclusion phases. The balance between hydrostatic pressure arising

from the inclusion phase upon the matrix and stress from thermal contraction of the

matrix will dictate the magnitude and sign of the stress [6.17].

Figure 5.5 is a graph of residual stress measured at different points within the

bulk of a deposit which is still constrained by the substrate. The austenitic 316L

stainless steel deposit exhibits tension, while the substrate is in compression which is

common to single phase depositions. The austenite (20%) and martensite deposition

of a .2%C martensitic stainless steel has the deposit in compression and the substrate

displaying both compression and tension at different points. The two phase (matrix

and inclusion) deposit has changed the stress profile exhibited by the deposit.

The high carbon steel (.8% C) also displays mixed modes. At the top of the

deposit (the last deposited layer) the material is primarily austenite and martensite and

the material exhibits compression. The middle and bottom layers consist of 3 phases,

austenite, martensite and ferrite. The middle and lower regions of the deposit exhibit

tension.

The possibility of different stress states causes our current understanding of

laser deposition to be challenged. However, these different stress states could be

beneficial. The possibility of compression in the deposited layers could beneficially

affect layered deposition by improving crack resistance but it could also be detrimental

by reducing fatigue strength.

Chapter 5. 2.2 Hardness of Laser Deposited Carbon SteelsHardness for 316L and Invar were 90-96 Rb (200-230 HK) and 80-90Rb (175-

200 HK). This level of hardness indicates typically soft and reasonably ductile

material. Hardnesses for the carbon steels deposited by SDM’s 2400 W Nd:YAG laser

was now in the range of 35-65 Rc (350-700 HK). The higher the carbon content, the

100

200

300

400

500

600

-8-6-4-202468

410-Left Side 410-Middle

Mic

roha

rdne

ss (

HK

)

Distance (mm) Note: O is Deposit Interface

Deposit Substrate

Page 98: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 2 Differences in Depositing Carbon Steels 98

Figure 5.6 Microhardness of a 410 (SP) deposit on a low carbon steel cold roll substrate

higher the hardness ranked. Figures 5.6-5.9 are microhardness graphs of 410 and 420

martensitic stainless steels. Each structure is from 5-6 layers thick. The measurements

are taken in two places: 2 mm on the left edge of a deposit and in the middle of the

deposit. Each deposit is about 20 mm wide. Different build styles were used to try to

increase hardness by creating more martensite within the bulk structure. A single pass

style (SP) occurs when each layer is built with powder first being placed on the

substrate or previously deposited layer and then scanned by the laser. This sequence

continues over and over until the part is completed. A double pass style (DP) is very

Figure 5.7 Microhardness of a 410 (DP) deposit on a low carbon steel cold roll substrate

Figure 5.8 Microhardness of a 420 (SP) deposit on a low carbon steel cold roll substrate

isimilar with one important difference. Powder is preplaced upon the substrate or

previously deposited structure and then scanned by the laser. Immediately, the laser

scans the part again, a double pass. No additional powder is placed on the newly

deposited layer before the double pass occurs. This procedure of a powder-laser scan

combination followed by a powderless-laser scan is continued to produce each layer of

100

200

300

400

500

600

-8-6-4-202468

410R Left410R-Middle

Mic

roha

rdne

ss (

HK

)

Distance (mm) Note: O is Deposit Interface

Deposit Substrate

100

200

300

400

500

600

-8-6-4-202468

420-Left 420-Middle

Mic

roha

rdne

ss (

HK

)

Distance (mm) Note: O is Deposit Interface

Deposit Substrate

Page 99: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 3 Changes in Solidification Modes 99

the part. It was hoped that the second scan would reduce the retention of secondary

retained phases or extend the heat affected zone of the laser pass to harden the

material.

The left side of deposits seem to have much more variability in hardness than

the middle of the deposit. The bottom layers of the SP deposits in the middle of the

deposits are typically harder than the top layers of the same area. In 410, the DP

sequence hardened the deposit significantly, while the 420 DP deposit seems to have

soften a bit.

Figure 5.9 Microhardness of a 420 (DP) deposit on a low carbon steel cold roll substrate

Chapter 5. 3 Changes in Solidification ModesWhen building parts or die inserts by using a layer laser manufacturing

technique, there are two steps which should dictate the resultant phases: Solidification

and Layering. Solidification or the solidify of molten metal particles into a solid

deposit is a significant affect. With our laser deposition process, we are able to attain

solidification rates between 104 and 105 K/s. This is rapid solidification. By

undergoing rapid solidification, metastable phases can now exist at room temperature.

Austenite, martensite, and even sigma ferrite may exist. Under slower cooling

processes, martensite will not form, austenite should not be retained, and sigma ferrite

may not exist in iron based steels. If we can understand the solidification modes

produces by rapid solidification, then we may be able to predict in advance phases

produced in the deposition process and begin to prescribe processing changes or even

heat treatments for minimal layer deposits (1-3 layers).

Tools have been developed by Shaeffler (5.1) and Delong (5.2) to aid in

predicting solidification modes. These tools are dependent upon knowing the

100

200

300

400

500

600

-8-6-4-202468

420R-Left 420R-Middle

Mic

roha

rdne

ss (

HK

)

Distance (mm) Note: O is Deposit Interface

Deposit Substrate

Page 100: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 3 Changes in Solidification Modes 100

composition of the metals. The solidification diagram is based upon the fact that as

most iron based metals solidify from the liquid state, the will either solidify as

austenite or ferrite. Typically , the determining factor resides within the composition.

The amount of nickel or chromium within the metal is typically the determinant.

Nickel is an austenite stabilizing element and chromium is a ferrite-stabilizing

element. The ratio of Chromium to nickel is a good indication of which phase the melt

will solidify first. If the Cr/Ni ratio is higher than 1.5 the melt will solidify as ferrite.

If the Cr/Ni ratio is lower than 1.5 it will solidify as austenite (5.3).

Other elements may also be classified as austenite or ferrite-stabilizing.

Carbon and manganese are strong austenite stabilizing elements. Molybdenum,

silicon, and niobium are strong ferrite stabilizing elements. Austenite forming element

may extend the austenite range on the Fe-C phase diagram to lower temperatures.

Ferrite stabilizing element may extend the ferrite zone or shrink the austenitic phase

field (5.1).

The solidification diagram displays four modes of solidification which are

quantified by the calculation of nickel (Ni-Eq) and (Cr-Eq) chromium equivalence.

The nickel equivalence (Ni-Eq = Ni + 30 C + .5 Mn) and the chromium equivalence

(Cr-Eq = Cr + Mo + 1.5 Si + .5 Nb) attempt to incorporate the effects of the strongest

ferrite or austenite-stabilizing elements.

Figure 5.10 Solidification Diagram

The Solidification Diagram was designed to be used to determine the

solidification modes of weld metal. It uses the nickel and chromium equivalents to

determine the mode preference. Four solidification modes are represented on the

= Cr, Nb, Mo

Ni-E

quiv

alen

t = C

, NI

0

510152025

3035

40

5 10 15 20 25 30Cr-Equivalent

A

AFFA

F

Page 101: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 3 Changes in Solidification Modes 101

diagram: Austenitic (A), Austenitic-ferritic (AF), Ferritic (F) and Ferritic-austenitic

(FA). Austenitic solidification is a mode in which austenite solidifies as the primary

and only mode. No ferrite is present in the structure. Segregation occurs only to grain

boundaries. Austenitic-ferritic solidification has austenite solidifying as the primary

phase with ferrite forming at grain boundaries. The ferritic second phase may

partially transform to austenite at subsolidus temperatures. Ferritic solidification

occurs when ferrite solidifies at the primary and only phase. Ferritic-austenitic

solidification occurs when ferrite is the primary phase but a second phase, austenite

grows at grain boundaries.

Seven carbon steels were examined. M2, H13, Rock, SDMT, 410, 420, 431

were compared against 316L. Table 5.3 shows the calculated nickel equivalents.

Table 5.3 Nickel and Chromium Equivalents

Figure 5.11 Solifidification Modes as Determined by Nickel and Chromium Equivalents

Metal Cr-Eq Ni-Eq

410-Anval 12.500 2.1700

420-Anval 13.600 15.750

431-Anval 15.600 8.0800

SDMT 21.520 3.1450

M2 4.0000 28.000

H13 6.3500 14.000

Rock 21.000 1.5500

316L 19.960 13.905

0

5

1 0

1 5

2 0

2 5

3 0

3 5

4 0

5 1 0 1 5 2 0 2 5 3 0

Nic

kel E

qu

iva

len

t

Cr-Equivalent

410

420

431

316LH13

M 2

RockSDMT

A

AFFA

F

Page 102: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 4 The Influence of Layering upon Laser Deposited Carbon Steels102

However, rapid solidification of the laser melt pool allows for most of the

modes to solidify as austenitic and austenitic-ferritic (Figure 5.12). The rapid

solidification seems to suppress the ferritic transformation. When comparing these

results against other rapid solidification techniques, the suppression of the ferritic

transformation is supported for similarly composed metals especially stainless steel

(5.5,5.6).

Figure 5.12 Redrawn Solidification Diagram

Chapter 5. 4 The Influence of Layering upon Laser Deposited Carbon Steels

A single layer typically is not enough material deposition to build a part.

Several layers have to be deposited in order to build parts. Figure 5.13 shows a

deposition which has been etched with sodium meta-bisulfite to reveal the individual

weld passes.

Figure 5.13 SDM deposited 410 Stainless steel Vs. a diagram showing layered manufactur-ing

With each additional layer, prior layers begin to be reheated. The layering

process allows for multiple temperature gradient to exist within the deposition. Thus,

Ni-E

quiv

alen

t

05

10

1520

2530

35

40

5 10 15 20 25 30Cr-Equivalent

431

A

AF

FA

420

410

316LH13

M2

RockSDMT

2 mm

Build PlatformSubstrate

NN+1

N+2 ∆T1

∆T2

∆T3

Page 103: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 4 The Influence of Layering upon Laser Deposited Carbon Steels103

the phases of the solidified structure will change. Solid state phase transformations

may occur, grain growth, or tempering may occur. To account for just solidification

phases is insufficient.

The SDM deposition process is very similar to multipass welding. Tools like

the Shaeffler welding diagram could be potentially used to benefit SDM designers by

aiding them to understand phase development. By knowing phase developement,the

SDM designer can now make a more informed decision on material selection and

potentially gain insight upon necessary heat treatment regimes to remove unwanted

phases. However, the Shaeffler welding diagram was designed for lower solidification

rates. Conventional welding processes like acetylene torches or GMAW solidify

between 100 to 102 K/s. Solidification for this application has been calculated to be

between 104 to 105 K/s. Other laser based process can be as high as 107.

Figure 5.14 is a plot of the test metals on the traditional Shaeffler Welding

Diagram. The phases that the diagram predicts are only correct in one incident. Using

the nickel and chromium equivalents calculated above, one can see the predicted

phases are inadequate. For the Shaeffler Diagram to be useful for layered

manufacturing, it would have to be changed.

Figure 5.14 Plotting the test metals based on the Nickel and Chromium Equivalents

Other researches investigating rapid solidification have suggested altering the

Shaeffler diagram. Vitek et. (5.7) al and Katayama et. al. (5.8) have actually proposed

modifications of the Shaeffler Welding diagram to reflect rapid solidification. Instead

of modify the areas of the diagram, Self et. al (5.9) proposes modifying the definition

of the nickel equivalent. This approach seems to be stronger, in that trends to

modifying the definitions of nickel equivalents can be accomplished more easily than

redefining the boundaries of the entire diagram. In this study, the researched focused

Ni-E

quiv

alen

t

431 = A+M+F420 = A+M+F316L = A410 = A+MSDMT = A+MH13 = A+MM2 = A+M+F

The Prediction

Actual

H13

M2

0

5

10

15

20

25

30

0 5 10 15 20 25 30 35 40Cr-Equivalent

420

A+M

MartensiteA+M+F

δ-Ferrite

431

410M+F(δ)

Austenite

316L

SDMT

431=A+M+F

420 = A+M+F316L = A

410 =A+MSDMT =A+M

H13 =A+MM2 =A+M+F

Actual

A=Austenite, F=Ferrite, M=Martensite

Page 104: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 4 The Influence of Layering upon Laser Deposited Carbon Steels104

upon the martensitic regions of the Shaeffler Diagram. Other researcher’s data who

have looked at other rapid solidification processes or laser multipass welding

phenomena have predominantly studied metals falling in the austenitic and ferritic

regions. Combining their analysis with this research could provide a more

comprehensive revision to the Shaeffler welding diagram and a more robust solution.

Utilizing the group handling data method (GMDH), the Cr-Eq and Ni-Eq were

modified to place the rapid solidification metals in to the correct areas. The GMDH

was not only given the compositional elements of the metals to choose from but

melting temperatures, thermal conductivities, etc. to find the relevant modifications to

the Shaeffler Diagram. Also, other researchers data was also included in the pattern

recognition search (5.6,5.10,5.11).

Figure 5.15 Modified Shaeffler for Layered Laser Manufacturing

Figure 5.15 shows the modified Shaeffler Welding Diagram. Notice how the

the over all trend is to shift down and to the right for the more martensitic and tri-

phase steels and up and left for the more austenitic steels. (Figure 5.17 shows more of

the modelled data and includes other researchers data: Katayama (5.8), Vitek (5.7),

David(5.6) , Anjos (5.9), Wei (5.10). )

Figure 5.16 shows the proposed New Ni-Eq and Cr-Eq. Both the New Ni-Eq

and the New Cr-Eq show a dependence upon the Cr-Eq. This is similar to results

found by Self et al (5.9). Self found that the amount of chromium was relevant in the

retention of austenite and the formation of martensite . At high chromium the phase

change of interest in Self’s work is austenite-ferrite, but at low chromium, it becomes

austenite-martensite.

ST = SDMT

Ni-E

quiv

alen

t

Cr-Equivalent

0

5

1 0

1 5

2 0

2 5

3 0

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

A+M

MartensiteA+M+F

δ-FerriteM+F(δ)

Austenite

LS H13

4 1 04 2 0

4 3 1M2

ST309A

316L

Page 105: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 4 The Influence of Layering upon Laser Deposited Carbon Steels105

Researchers have shown that the level of chromium can become an agent to

determine how certain phase-stabilizers behave. In addition to the strong austenite-

stabilizers listed earlier, there are others: Co, Cu, N, Al. There are also additional

ferrite-stabilizers: Ta, Ti, W, V, Z. However, during cryogenic solidification or high

service temperatures, some of these element along with Mn, Al, Mo, and Si may

change and stabilize the opposite modes (5.12, 5.13). At high solidification rates,

ferrite-stabilizers may act as austenite-stabilizers and vice-versa. One indicator of this

mode change, may be the level of chromium. In low chromium systems, ferrite-

stabilizers act like ferrite-stabilizers and austenite-stabilizers act like austenite-

stabilizers. In high chromium systems the “switching: properties has been shown to

occur.

Figure 5.16 The proposed NI-Eq and Cr-Eq

Figure 5.17 shows the limits of the proposed model. The distinctive separation

of high ferrite- low austenite content welds from fully ferritic welds is not very clear.

312B was fully ferritic in rapid solidification experiments. However, with the

traditional Shaeffler diagram. However, Katayama’s 100% Ferrite line may be more

appropriate for the true boundary line (5.8).

Figure 5.17 The Extents of the Laser Layered Manufacturing Shaeffler Model

(New Ni-Eq) = 11.977810 -0.368111 [W] + 1.598563 [Mo] + 0.579666 [Ni] - 0.262729 [Cr-Eq]

(New Cr-Eq) =0.666030 + 0.578787 [V] +5.386530 [Si] + 2.048034 [W] + 0.811128 [Cr-Eq]

Instability in the Model: Modelin accurately predicts the austeniteferrite content in this region.

100% Ferrite Line proposed by Katayama,et. al.

0

5

1 0

1 5

2 0

2 5

3 0

0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0

A+M

MartensiteA+M+F

δ-FerriteM+F(δ)

Austenite

LS H13MS

R

410420

431 M2ST

446

312B309A

316L316B

304A

316A

Ni-E

quiv

alen

t

Cr-Equivalent Solidification Rate: 104-106

Page 106: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 5 Phase Location 106

Chapter 5. 5 Phase LocationThe improved Shaeffler Welding Diagram for Layer Manufacturing does not

account for the exact percentage of austenite, martensite or ferrite within the deposited

material except for 100% single consituency regions. Regions of multiple phases now

only account for the existence of a phase as oppose to the quanitity. This exists for two

reasons: 1. The data is being fit to the Shaeffler Welding Diagram thereby eliminating

the phase percentage and diagram location relationship. 2. The data is an average of

the amount of phases which exist within the bulk deposit.

The original Shaeffler Welding Diagram was designed to aid designers to not

only predict the existence of phases within the welded materials but also to predict the

phase percentages. Because data is being fit into diagram, it was very difficult to

maintain the phase percentages as described by the diagram and successfully redefine

the Ni & Cr equivalences to accurately describe a wide set of material compositions

and encompass the full effects of rapid solidification. To produce a diagram which can

acurately describe phases percentage would require redrawing the entire Shaeffler

Diagram.

The model uses the average phase values to position the material within the

correct Shaeffler zone. A laser deposit which is more than 3 layers thick can be

divided into 3 disctinct regions: bottom, middle and top.

Figure 5.18 Top, Middle, and Bottom sections of carbon steels

Figure 5.18 shows two sets of pictures of microstructural zones of laser

A1. Top 420

B1. Top 420

A2. Middle 420

A3. Bottom 420

B2. Middle 420

B3. Bottom 420

50 µm

50 µm

50 µm

50 µm50 µm

50 µm

SS-Single Pass

SS-Double Pass

SS-Single Pass

SS-Double Pass

SS-Single Pass

SS-Double Pass

Page 107: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 5 Phase Location 107

deposited stainless steel. The micrographs labeled “A” were built with single scans.

Single pass scans are a deposition sequence of placing the powder upon the substrate

or previously deposited layer and then scanning the powder with the laser heat. The

micrographs labeled “B” were built with double scans. Double pass scans are a

deposition sequence of placing the powder upon the substrate or previously deposited

layer, scanning it with the laser heat, and then immediately scanning the surface again

with laser heat without the addition of new powder. These two samples are included to

show how different the microstructure can look if the laser solidification-reheating

sequence is changed even with the same lasing conditions (laser power, scanning

speed, powder feed rate, etc.).

The bottoms of both carbon steel deposits have a significant amounts as-

quenched martensite, characterized by dark brown to black because of reactive

etching. The lighter brown shown in all 6 photos is tempered martensite. Blue

indicates ferrite. The shiny or unreacted boundaries are mixtures of retained austenite

and ferrite. Table 5.4 and 5.5 show the average phase percentage for the 420 deposits

Table 5.4 Phase Averages for each location in a 420 Single Pass Laser Deposit

Table 5.5 Phase Averages for each location in a 420 Double Pass Laser Deposit

Martensite Ferrite Austenite

Top 86% 7% 5%

Middle 82% 6% 9%

Bottom 76% 4% 12%

Average 81% 6% 9%

Martensite Ferrite Austenite

Top 77% 4% 14%

Middle 94% 0% 2%

Bottom 79% 2% 15%

Average 83% 2% 10%

Page 108: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 6 Grain Size of Laser Deposited Layered Carbon Steels 108

of both conditions (See section 3. 1.1 -3. 2.2 for methodology used to measure

phases). The differences in microstructure based upon location can be dramatic and is

determined by proximity to substrate ( which acts as a heat sink), the number of layers

deposited, and laser conditions. A single tool which could meaningfully characterize

all of these effects and still provide quantitative phase percentages is extremely

difficult to build. Multiple tools for each region, each lasing condition, and each

material would have to be constructed. Also, the regions may not scale as easily to

larger builds. When the total thickness of deposited layers extends beyond 25 mm,

significant changes may occur to the phases present. The continual reheating of lower

layers begins to form more equilibrium phases like cementite and banite. Therefore,

another set of Shaffler Diagram’s would have to be built.

As recounted in Chapter 2, there are several competitive laser deposition

processes which can also benefit from this knowledge. However, each of the processes

uses different deposition sequences, laser conditions, etc. A specific regional tool for

SDM could not be leveraged very well for their applications. A universal tool that can

inform the designers of potential phases that can arise during laser deposition, is much

more useful.

Chapter 5. 6 Grain Size of Laser Deposited Layered Carbon Steels

Grain size is indicative of the material properties of the laser deposit. The

grain structure of austenitic steels is very different from the laser deposited carbon

steels. Austenitic steels formed lenticular shaped cells while the carbon formed from

lathe to plate -like structures. No direct correlation could be found between powder

size to grain size. However, by using GMDH a relation ship was developed which can

predict equivalent diameters of the grains. Equivalent diameters are based on the area

of the grain.

Using GMDH, a heuristic was developed to predict grain size. The factors

included at the beginning of modeling were material composition, powder size,

solidification rate, thermal conductivity, and melting point. Figure 5.20 displays the

relevant terms. The heuristic will enable designers to get an estimate of the grain size

of the structure if they are using a layering process with solidification rates from 104-

105 K/s.

Page 109: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 6 Grain Size of Laser Deposited Layered Carbon Steels 109

Table 5.6 Powder Diameter Vs. Grain Diameter

Figure 5.19 Examples of Grain Sizes

The heuristic’s components seem to be quite applicable. The carbon content

will influence phase formation as well as aid carbide forming elements. Carbide

forming element, particularly Mo and V, because they can act as grain size controlling

elements. The principal effect of the grain size controlling elements is to extend the

recrystallizing time of austenite. The longer time the it takes for austenite to

recrystallize, the more nucleation sites for ferrite to form. The more ferrite nucleation

site the smaller the ferrite grains will be (5.14). The chromium and Ni-Eq

(30*C+Ni+Mn), all aid in determining the phases which exist in the solidify melt. The

Metal Phase(s) Powder

SizeEquivalent Diameter

316L 100%-A 74 +34 µm 9.64 µm

410 SS 30%A+M 55 +32 µm 10.29 µm

431 SS 9%A+M+2%F

122 +49 µm 5.63 µm

420 SS 14%A+M+6%F

100 +60 µm 5.04 µm

H13 3%A+M 45 +20 µm 4.76 µm

SDMT 20%A +M 90 +20 µm 4.11 µm

Laser Deposited 420 SS

100 µm100 µm

Laser Deposited 316L Stainless

100 µm

100 µm

Laser Deposited 431 SS

100 µm

Laser Deposited 410 SS

100 µm

Laser Deposited 316L

Page 110: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 110

melting kinetics related to powder size, thermal conductivity and melting temperature

will all influence grain size.

Figure 5.20 Equation for Grain Size Predictor

Figure 5.21 Grain Size Comparison Actual Vs. Heuristic

Chapter 5. 7 Layer ThicknessLayer thickness is another characteristic of deposition which needs to be

understood. The ability to predict layer thickness will enable designers to accurately

forecast the number of layers needed to build a part and predict completion times.

Also, understanding the nature of layer morphology will enable more insightful

planning.

Since SDM is deposited at room temperature and without preheat, much of the

energy of the first layer of deposition is used to bring the substrate and build platform

to a steady state temperature with the fresh deposit. The low carbon steel substrate and

aluminum build platform act as a heat sink. As more layers are deposited, and the

distance from the top layer to the substrate increases, the upper layers of the deposit

begin to get warmer. The weld pool can be larger and more powder is able to solidify.

Thus layers size increases with z-height. However, the layer thickness does begin to

achieve a mean thickness. The last layer of the deposit is about 40-50% larger than

Average Grain Diameter (µm)= 22.8(µm) + {0.3 [V] + 3.7 [C] -0.0008/K [Melting Temperature] + 0.6[Cr] -0.1/µm [Powder Size] -0.8 [Mo]-0.3 [NI-EQ] -0.5ms/K [κ]}µm

0 2 4 6 8 1 0 1 2

4 1 0

4 2 0

4 3 1

SDMT

H13

3 1 6 L

Invar

Experimentally MeasuredCalculated Values

Equivalent Diameter (µm)

Page 111: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 111

mean size thickness. The last layer is not remelted like the prior layer so it maintains

the thickness of individual weld passes.

Table 5.7 displays layer values for 4 steels used in the research. Five layer

structures were built of each material. Two examples of 420 are shown. One was built

with single pass layers while 420-DP was built in a double pass condition:

Immediately following a powder melting laser pass, the laser scans the part again, this

time with no powder. All of the other beams were built with a single pass. The double

pass builds seem to slightly thicken the latter layers of the build.

Table 5.7 Layer Thickness for SDM Carbon Steels

Every time a deposit is allowed to cool to room temperature, the layer size

starts almost back over, as if it were a brand new deposit instead of a continued build.

Figure 5.22 shows a 60 layer (55 mm) structure. Notice how the layer thickness

deviates as the build gains in z-height. Figure 5.23 shows the same graph but this time

it is identified with every time with the break that was taken during the deposition.

The ability to control layer thickness seems to stem from scheduling breaks. If a

certain layer size is critical to part strength or critical internal location, like the

deposition strategy for the location of internal geometry, a cooling/deposition

sequence during the build process need to occur. Lastly, if more than 12 layers are

deposited at any point in time, the grain size of bottom layers may grow as well as

incur phase transformations.

Layer316L(µm)

SDMT(µm)

431(µm)

420(µm)

420-DP (µm)

5th Layer 1200 1286 1624 732 755

4th Layer 937 856 703 468 550

3rd Layer 899 904 797 413 503

2nd Layer 604 647 557 442 470

1st Layer 550 292 470 174 208

Page 112: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 112

Figure 5.22 Sixty Layer Deposition

GHDM was used to determine what the most significant factors effecting the

mean layer thickness. In addition to the material composition, material properties like

thermal conductivity, melting point and solidification rate, two addition factors were

added: fill rate and thermal diffusivity length.

Figure 5.23 The Effect of Cooling on Deposition

0

0.2

0.4

0.6

0.8

1

1.2

1.4

0 5 10 15 20 25 30

Ave

rage

Lay

er T

hick

ness

(m

m)

Z-Height (mm)

20 25 30 35 40 45 50

10.80.6

0.4

0.2

02

4

6

8

2

4

0 5 10 15 20 25 30

A

B

B

B A

A

Z-Height (mm)

Laye

r T

hick

ness

(m

m)

A = Overnight Cooling

(12-24hr breaks)B = Short Cooling Periods

(1-2hr breaks)

20 25 30 35 40 45 50

Page 113: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 113

Figure 5.24 shows the results of the model. Metals with various deposition fill

rates and scanning speeds were put in the model. However, Figure 5.25 just shows one

setting. Figure 5.25 shows a comparison between the model and the actual values of

mean layer thickness.

Figure 5.24 Heuristic for Predicting Mean Layer Size

Figure 5.25 Comparison of Mean Layer thickness: Predict Vs. Actual

Even thought the solidification rate does include the effects of the scanning

speed, it is the powder fill rate which make more of an impact upon mean layer

thickness. As Figure 5.26 shows, from 15 mm/s to 40 mm/s there is an insensitivity

to laser scanning speeds. The amount of energy that the SDM puts into weld is so

large that regardless of the speed adequate layer solidification is achieved. At speeds

higher than 40 mm/s, the 407 W/mm2 is no longer sufficient to get good layer

solidification with no porosity or retention or secondary phases.

The fill rate is extremely sensitive. Depositions below 23 g/min tend produce

too thin of layers and more secondary phases from solid state phase transformations

tend to occur. Depositions above 23 g/min produce too thick of layers, secondary

phases are retained from the melt.

Layer (µm) = 1176.7µm + {80.8 [V] -0.0008s/K [Solidification Rate] +11.1 [Ni] - 40.3 [Ni-EQ + 27.8 min/g [Fill Rate] - 5.5/µm [Powder Size]} µm

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0

4 1 0

4 2 0

4 3 1

SDMT

H13

3 1 6 L

Invar

Experimentally MeasuredCalculated Value

Thickness (µm)

30 mm/s at 23g/min

Page 114: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 114

Figure 5.26 Optimizing Deposition Parameters

Fill Rate (g/min)

Laser Scan Speed (mm/s)

%R

etai

ned

Aus

teni

te 0 5 10 15 20 25 30 35 40

0

5

10

1

5

20

2

5 3

0

35

4

0

15 20 25 30 35 40 45 50 55

Speed

Fill Rate

Page 115: Thesis Glink

Chapter 5 Characterization Tool Steels

Chapter 5. 7 Layer Thickness 115

5.1 Shaeffler, A. L., “Constitutional Diagram for Stainless Steel Weld Metal,” Metal Progress 56(5): 680-680B.

5.2 Delong, W. T. “Ferrite in Austenitic Stainless Steel Weld Metal,” Welding Journal 53(7):273-s -286-s.

5.3 Suutala, N. Takalo,T.,Moisio, “The relationship Between Solidification and Micro-structure in Austenitic and Austenitic-Ferrite Stainless Steel Welds,” Metallur-gical Transactions 10 (A), 512-514 (1979).

5.4 Robert, George, Krauss, George, and Kennedy, Richard, “Tool Steels,” ASM Inter-national, 1998.

5.5 Elmer, Walter, “The Influence of Cooling Rate on the Microstructure of Stainless Steel Alloys,” Lawrence Livermore Laboratory, September 1988.

5.6 David, S. A., Vitek, J. M., and Hebble, T. L., “Effect of Rapid Solidification on Stainless Steel Weld Metal Microstructures and Its Implications on the Shaef-fler Diagram,” Welding Research Supplement, October 1987, pp. 289-s- 300-s.

5.7 Vitek, J. M. , Dasgupta, A. , and David, S. A., “Microstructual Modification of Austenitic Stainless Steel by Rapid Solidification,” Metallurgical Transactions 14 (A), 1833-1841 (1983).

5.8 Katayama, S. and Matsunawa, “Solidification Microstructure of Laser Welded Stainless Steels”, Proc. ICALEO, p. 60, 1984.

5.9 Self, J. A., Matlock, D. K., and Olson, D. L., “An Evaluation of Austenitic Fe-Mn-Ni Weld Metal for Dissimilar Metal Welding,” WRC Bulletin, September 1984, p 282-s - 288-s.

5.10 Wei, M. Y. and Chen, C., “Fatigue Crack Growth Characteristics of Laser-Hard-ened 4130 Steel,” Scripta Metallurgica et Materialia, Vol 31., No. 10, 1994, pp. 1393-1398.

5.11 Rieker, C., D. G. Morris, and Morris, M. A., “Microcrystalline Surface Layers Created by Laser Alloying” Journal of Less-Common Metals, 145 (1988) 595-600.

5.12 Wallace, W., Trenouth, J. M., Daw, J.D., “Microstructual Instabilities in An Industrial Engine Vane,” Metallurgical Transactions A (Physical Metallurgy and Materials Science), Vol 7A, No 7, Jul 1976 p. 991-997.

5.13 Sakamoto, T., Nakagama, Y., and Yamauchi, I, “Effect of Mn on the Cryogenic Properties of High Austenitic Stainless Steels,” Advances in Cryogenic Engi-neering: Materials, Volume 32, 1986, p 65-71.

5.14 Reed-Hill, Robert E. and Abbaschian, Reza, Physical Metallurgy Principles, PWS-Kent Publishing: Boston, pp. 661-685.

Page 116: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 1 High Impact Resistant Materials 116

Chapter 6 Applications of Characterization

Now that the formation and transformation of carbon steels in layered laser

deposition is better understood, one can use this new understanding to solve

deposition problems. Knowledge of grain size and phase composition are used to

select high impact resistant materials and to reduce part deflection. These are just two

examples of how understanding the marriage of material microstructure and

deposition parameters can benefit layered manufacturing

Chapter 6. 1 High Impact Resistant MaterialsDie cast inserts need high impact resistant materials in order to survive the die

casting service environment. Figure 6.1 shows a pressure trace of a 6 ton die casting

machine . Dies in this machine will exhibit pressure changes from ambient to 20,000

psi and temperature changes of 150_C to 667_C in 12 seconds or less

Figure 6.1 Pressure Trace from a 6 ton Die Casting Machine

Shot Pressure (PSI)

14000

15000

16000

17000

18000

19000

20000

Time (Sec)

150m C

667m C

1.0 1.4 1.8 2.2 2.6

One second aftermetal injection.

Page 117: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 1 High Impact Resistant Materials 117

If molten aluminum is injected into a die with low impact resistance, the die

will gall or initiate heatchecking. Thin features will degrade, and the die inserts will

be rendered useless quickly.

Impact resistance is typically measured by charpy impact specimens (see 3.

3.2) at room temperature. Die casting die inserts are typically made out of H13 and

are austenitized and tempered. After these heat treatments, charpy specimen’s can

typically endure from 10-20 ft-lb of impact resistance.

Chapter 6. 1.1 Engineering Material SelectionsOur typical engineering understanding would have us select the material by

looking at the ratio of the modulous of elasticity to yield strength (Figure 6.2 and Table

6.1). According to this SDMT should have the best results. Although better than most,

Figure 6.2 Maximizing the E / σ relationship to increase Impact Resistance

Table 6.1 Comparison of E /o Tempered Relationship

E x104 MPaσ MPa -

(Tempered)E/σ

410 22 1005 219

420 20 690 290

431 20 738 271

SDM Tool Steel

25.4 510 498

H13 21 1344 156

316L 19.3 415 465

Minimize the Deformation By Toughening the Materialε *L =δδ h σ/E Minimize δ by maximizing E /σ

δ

Ιmpact TestingCharpy Impact Specimen

Page 118: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 1 High Impact Resistant Materials 118

431 was clearly the best. On the half size specimens, it showed almost twice the

impact resistance. If E/σ is not the appropriate measure to use, perhaps the

microstructual information can help.

Figure 6.3 Results of Charpy Testing

Chapter 6. 1.2 Microstructual Material SelectionImpact resistance is a function of grain size just as yield strength and many

other properties are. Also, the phases within the deposit will determine, impact

resistance. Martensite has a bulk modulus of 17.3 x 104 MPa while austenite has a

bulk modulus of 16.4 x 104 MPa and ferrite has a bulk modulus of 17.1 x 104 MPa

(6.1). The added toughness of martensite may be swaying the results. However, when

compare grain size and phase retention, we see that it is a combined result which gets

the best results. A small grain size and the reduction of secondary phases gives the

best impact resistance (Figures 6.4 and 6.5).

Figure 6.4 Grain Size (Equivalent Diameters) vs Impact Resistance

The additional heat treatment of H13 did not improve the results. This is a

0 2 4 6 8 1 0

4 1 0

4 3 1

4 2 0

SDMT

H13

Impact Resistance (ft-lb)

Note: H13 was austenitized.All other specimens are inthe as deposited condition.

Half Size Charpy Specimens(5mm x 10 mm x 55 mm T x W x L)

0 2 4 6 8 10

10.29

5.63

5.04

4.11

410

431

420

SDMTGra

in S

ize

(µm

)

Impact Resistance (ft-lb)

Page 119: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 1 High Impact Resistant Materials 119

trend seen in other metals. A series of experiments were run on the 420 martensitic

stainless steels (Figure 6.6). During conventional tempering temperatures between

225- 500 Co the formation of sigma ferrite is quite high. Sigma ferrite is a brittle

phase of ferrite . Because it is brittle, it often acts as a crack initiation point in

materials or reduce material properties. Typically, sigma transforms from metastable

delta ferrite at temperatures of 600-700 Co(6.2). Because conventional heat treatment

cycles do not seem to work, each metal will have to be tested to find appropriate heat

treatments.

Figure 6.5 Secondary Phase Constituents Vs. Impact Resistance

Figure 6.6 Sigma Ferrite Formation during Tempering Cycles of 420 Stainless Steel.

The metal 431 had the highest impact resistance. Even though its E/σ ratio is

not very high, the small grain structure and limited amount of secondary phases cause

its material properties to be superior. When comparing the result of 431 with full size

specimens, we see that the 431 results are comparable to conventionally rolled or

forged H13 which has been tempered (Figure 6.7). The results do suggest that there is

0 2 4 6 8 10

3 0

2 0

1 7

1 1

410

431

420

SDMT

% 2

nd P

hase

C

ompo

sitio

n

Impact Resistance (ft-lb)

Dan

gero

us H

eat

Tre

atm

ent

Zon

e

0

1 0

2 0

3 0

4 0

5 0

0 2 0 0 4 0 0 6 0 0 8 0 0 1000

Delta Ferrite Sigma Ferrite

Pha

se P

erce

ntag

e

Tempering Temperatures

Page 120: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 120

a possibility to produce die inserts which can work without heat treatment. This would

be an additional time savings for the designer and manufacturer.

Figure 6.7 Comparison of Layered Laser Deposited Carbon Steels and Conventionally formed and tempered H13.

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transformation

Laser deposition produces rapid solidification or quenching because of fast

heat conduction from the laser melt pool into the base metal. This solidification rate

can also be influenced by changing deposition parameters such as scanning speed and

laser power. When depositing carbon steels, rapid solidification can cause metastable

phases to occur, primarily martensite. Martensite formation is a diffusionless process

which occurs when the deposit cools from high temperatures (like the steel melting

point) to below the martensitic start temperature in less than 10 sec. The arrows shown

in Figure 6.8 illustrate the isothermal transformation of H13 and 410 stainless steel to

martensite by rapidly cooling to room temperature. Cooling slower than this would

not produce martensite but result in the production of ferrite and cementite.

When the laser melts the powder, the first solid metal to nucleate within the

liquid is austenite. Austenite has a face center cubic structure. The rapid quenching

causes the carbon in the austenite phase to transform to body center-tetragonal (BCT)

martensite trapping carbon in the BCT lattice which has not had a chance to diffuse.

0

5

10

15

20

25

25 30 35 40 45 50

Cha

rpy

V-n

otch

ed

Impa

ct e

nerg

y (

ft•lb

f)

Hardness (R c)

420/410 H13 (Tem, L)

H13 (2 Tem, L)

H13 (Tem, L)

H13 (Tem, T)H13-VAR (2 Tem, T)

H13-ESR (2 Tem, T)

431

410 420SDMT

Page 121: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 121

Figure 6.8 Isothermal Transformation Diagram of H13 and 410 steels. Compiled from [6.9].

This transformation results in an expansion in volume. This expansion can be

as big as 4%. This is the same order of magnitude as the thermal shrinkage which

arises from cooling of the metal. If martensite transformation can be induced and

controlled by layered laser deposition, then the potential to balance the solidification

shrinkage with phase volumetric expansion may exist.

The use of lasers to change part or surface microstructure has been extensively

researched for laser cladding and surface hardening. Several researchers have shown a

relationship between laser parameters and microstructual evolution for laser cladding

and surfacing. Wang et al. [6.5] showed that metastable phases can be produced on

material surfaces by laser quenching, a process in which laser melting and quenching

of a material occurs by using a very short laser pulses. Yang et. al. [6.6] used a CO2

laser to show that case depth or the depth of phase transformation and morphology of

materials can be influenced by laser power and laser scan rate. Fouquet et. al [6.7]

used a continuous wave CO2 laser to transform the surface of grey cast iron from

austenite to a mixture of austenite, cementite and martensite by using overlapping

multiple laser scan paths. Rieker, et. al [6.8] uses a remelting second pass of the laser

to homogenize the chemistry and microstructure of a laser hardened surface on ferritic

stainless steel.

200

400

600

800

1000

1200

1400

1600

1800

1 10 100 1000

T

empe

ratu

re (

C)

Time (s)

410

H13

A+9%FA+>9%F

A+F+C F+C

AA+F A+F+C

F+C

Ms

A=AusteniteF=FerriteC=CementiteM=Martensite

Page 122: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 122

Once the deposition of the beams was completed, the beams were tested for

deflection and phase composition. Table 7 lists the average results for each type of

material tested.

Table 7: Results of Deposition Experiments

Table 6.2 Phase Analysis and Deflection ResultsSelective etchants revealed martensitic and austenitic phases in all the metals

with the exception of 316L which is only austenitic. Martensitic stainless steel, 420,

had three phases: austenite, martensite, and ferrite (Figure 6.9 and Figure 6.10).

Figure 6.9 Pictures of 431 stainless: 1. Optical microscope with light brown reflecting tem-pered martensite, black - as-quenched martensite, white - retained austenite. 2. Bright-field image of TEM with white regions - martensite, black regions - austenite. @30,700 X.

Microhardness for the specimens ranged between 400-540 HK. For each of

Lase

r P

ass

Defl

ectio

n (m

m)

% A

s-qu

ench

edM

arte

nsite

(MA

Q)

% T

empe

red

Mar

tens

ite(M

T)

% A

uste

nite

% F

errit

e

MA

Q/M

T

410 1 .97 7 55 34 0 .125

420 1 1.26 16 65 9 6 .249

431 1 .432 29 62 4 2.3 .471

SDM -Tool 1 .89 9 68 20 0 .129

316L 1 .97 0 0 98 0 .0

410 2 1.14 4 88 6 0 .047

420 2 .33 28 55 10 2 .506

431 2 .33 30 58 6 3 .51

1. 2

M

A

A100 µm

Page 123: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 123

the metals, the hardness ranges were typically below the value of typical as-quenched

or hardened steel ranges. Figure 6.11 compares the quench/hardness ranges for the

martensitic stainless steels used in the experiments.

For certain materials increasing the martensite percentage reduced deflection

while it increased it in others. Figure 6.12 shows that by increasing the ratio of as-

quenched martensite to tempered martensite, deflection is reduced.

More as-quenched martensite is found in base layers than in N+1 layers

(middle and top layers). This suggests that the quenching rate is much faster near the

substrate than for subsequent layers. The double laser pass condition increases the

quenching rate for middle sections of the beams than the single pass condition. The

inset in figure 6.12 compares middle and bottom layers of 420 in single pass and

double pass condition.

Figure 6.10 Optical Microscope pictures of :1. 410 - SP, 2. 410 - DP, 3. 420 - SP, 4. 420 - DP.

Chapter 6. 2.1 As-Quenched Martensite MaterialsIncreasing the amount of as-quenched martensite retained in the beams

reduced the deflection of SDM deposited beams. As-quenched martensite allowed for

maximum volumetric expansion. Tempered martensite allowed the carbon to diffuse,

shrinking the lattice and reducing the amount of volumetric expansion achieved. SDM

laser manufacturing induced as-quenched martensite. However, reheating of layers

during the deposition of subsequent layers tempered much of the as-quenched

martensite. The addition of a second laser pass increased the quenching rate and

helped maintain more as-quenched martensite for most of the metals tested. Attaining

a ratio of at least .45 as-quenched to tempered martensite produced a 30-75%

reduction in deformation in the SDM deposited beams.

1. 2. 3. 4.20 µm

20 µm

20 µm

MT A MAQ A MAQ MT A F MAQ MT A MT MAQ

20 µm

Page 124: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 124

Figure 6.11 Comparison of the hardness values for martensitic stainless steel beams: 1 - Sin-gle Pass, 2-Double Pass. Quenching range compiled from [6.15].

These tests upon the SDM layered process indicate that continuous deposition

of material layers influences the microstructure of N, N-1, N+1 layers. The layer that

is being deposited, N, is obviously affected because of the melting and solidifying of

the material. The N-1 microstructure is affected because of the reheating of the layer

due to the newly deposited layer. The next layer, N+1 is affected because its grain

growth direction is seeded from the prior layer N. As more layers are deposited, more

heat is retained in the bulk. Grain growth in both austenitic and martensitic steels

increases with each additional layer. Grain size in layer N+1 is larger than in N.

However, in martensitic steels, this increased heat retention produces more tempered

martensite in the bulk and upper layers than near the interface between substrate and

first layer. The difference in the heat characteristics is an important element to

modelling and understanding the phase development within layered metal part.

Quench/Hardening Range for 410

300

350

400

450

500

550

600

650

700

0 5 10 15 20 25 30 35

Har

dnes

s (H

K)

Percentage of As-Quenched Martensite

410 (1)

410 (2)

420 (2)431 (2)420 (1)

431 (1)

Quench/Hardening Range for 431

Quench/Hardening Range for 420

Page 125: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 125

Figure 6.12 The effect of increasing MAQ/ MT Ratio on deflection.

Multiple phases may be both detrimental and beneficial to material properties.

A fully martensitic beam that is in the as-quenched phase may be too brittle to use. A

beam that has ferritic phases could develop sigma ferrite which is a brittle phase often

related to crack initiation points in metal. However, the retention of austenite or delta

ferrite has been shown to increase the toughness of martensitic parts [6.16].

Charpy impact specimens show a 20% drop impact resistance at the maximum

martensite condition. However, after heat treating a beam with negligible changes in

deflection at 200Co for 12 hours, the difference was ony 10%. Therefore, if the as

deposited condition is not sufficient, this special heat treatment may be enough to give

a prototype die insert the impact resistance that it needs.

Chapter 6. 2.2 Double Laser Pass Affects the Amount of As-Quenched Martensite

The double pass in most cases increased the amount of As-Quench martensite

retained in the deposition. As Figure 6.13 shows the double pass seems to improve the

ratio of as-quenched to tempered martensite significantly. The single pass has room

temperature powder with air gaps to heat and solidify, limiting the depth of the heat

affected zone (HAZ) where most of the as quench martensite populates. The double

pass has 250 Co fully solid material (98-99.5% dense with minimal pores) to remelt,

0.2

0.4

0.6

0.8

1

1.2

1.4

0 0.1 0.2 0.3 0.4 0.5 0.6

De

flect

ion

(m

m)

MAQ

/MT Ratio

420 (1)

420 (2)431 (1)

431 (2)

410 (2)

410 (1)

N-L

ayer

s

Primary Sitesfor As-QuenchedMartensite in 420

(1) SP (2) DP

MT

MA

Q-M

T

MA

Q-M

T

MA

Q-M

T

Substrate

20µm

Page 126: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 126

extending and densify the depth of the HAZ. Figure 6.14 shows the difference in a

single pass and double pass on 410 stainless. The dark brown is as-quenched

martensite, lighter brown is tempered martensite and white is retained austenite.

Notice the reduction in the amount of retained austenite in the HAZ of the double pass

deposition.

Table 6.3 Heat Treatment of 431 Laser Deposited Steels

Figure 6.13 Double Pass Vs Single Pass Laser Depositions

Specimen ConditionImpact

Resistance (ft-lb)

431-SP As Deposited 9.25

431-DP As Deposited 7.5

431-SP Heat Treated

12h @ 200Co 8.5

431-DP Heat Treated

12h @ 200Co8.5

0

0.1

0.2

0.3

0.4

0.5

0.6

0 5 10 15

Double Laser ScanSingle Laser Scan

Rat

io o

f As-

Que

nche

d M

arte

nsite

/T

empe

red

Mar

tens

ite

Ratio of Crequ

/ Niequ

Page 127: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 127

Figure 6.14 410 Stainless with Single Pass and Double Pass Configurations

200 µm

2 mm

200 µm

2 mm

Single Pass Double Pass

Page 128: Thesis Glink

Chapter 6 Applications of Characterization

Chapter 6. 2 Characterizing Material Composition to Induce Martensite Transforma-tion 128

6.1 Pan, H. H. and Weng, G. J., “Thermal Stress and Volume Change During A Cool-ing Process Involving Phase Transformation,” Journal of Thermal Stresses, 15:1-23, 1992, pp. 1-23.

6.2 Gill, T. P. S., Shankar, V., Pujar, M. G., and Rodriguez, P., “Effect of Composition on the Formation of d-Ferrite to s In Type 316 Stainless Steel Weld Metals,” Scripta Metallurgica Materialia, Vol. 32, No. 10, pp 1595-1600.

6.3 L. Samuels, Metallographic Polishing By Mechanical Methods, 3rd Ed.1982.6.4 Fourney, “Surface Engineering of Tool and Die Steel,” ASM Specialty Handbook:

Tool Materials, (1995), pp.391-392.6.5 W. K. Wang, C.J. Lobb, and F. Spaepen, “Formation of Metastable Nb-Si Phases

by Picosecond and Nanosecond Pulsed Laser Quenching,” Materials Science and Engineering, 98, (1988) 325-328.

6.6 L.J. Yang, S. Jana, S. C. Tam, and L. E. N. Lim, “The Effects of Process Variables on the Case Depth of Laser Transformation Hardened AISI 01 Tool Steel Spec-imens

6.7 F. Fouquet and E. Szmatula, “Laser Surface Melting of a Pearlitic Grey Cast Iron,” Material Science and Engineering, 98 (1988) 305-308.

6.8 C. Rieker, D. G. Morris and M. A. Morris, “Microcrystalline surface layers created by laser alloying,” Journal of Less-Common Metals, 145 (1988) 595-600.

6.9 Crucible Steel (1949), Rickett, Waltin, Butler (1952), Wang, Lobb, & Spaepen, (1988)

6.10 R. Merz, F.B. Prinz, K. Ramaswani, M. Terk, and L.E. Weiss, “Shape Deposition Manufacturing,” Proc. of Solid Freeform Fabrication Symposium, The Univer-sity of Texas at Austin, (1994) pp.1-8.

6.11 J. R. Fessler, Merz, A.H, Nickel and F. B. Prinz, Proceedings of Solid Freeform Fabrication Symposium, The University of Texas at Austin, (1996) pp.117.

6.12 E. Beraha and B. Shpigler, Color Metallography, (1977).6.13 B. L. Averbach, L. S. Castleman, and M. Cohen, “Measurement of Retained Aus-

tenite in Carbon Steels, ”Transactions of the ASM, (1949) Vol. 42, pp112-120.6.14 N. Williams and C. Carter, Transmission Electron Microscopy - Diffraction, Vol

2, (1996), pp.267-288.6.15 D. Olson, T. Siewert, S. Liu and G. Edwards, “Selection of Wrought Martensitic

Stainless Steels,” ASM Handbook: Welds, Brazing, and Soldering, Vol. 6, (1993), pp.432-442.

6.16 G. R. Link, J. Fessler, A. Nickel and F. Prinz, “Rapid Tooling Die Case Inserts Using Shape Deposition Manufacturing,” Material and Manufacturing, Vol 13, No, 2, (1998), pp. 263-274.

6.17 H.H. Pan and G. J. Weng, “Thermal Stress and Volume Change During A Cooling Process Involving Phase Transformation,” Journal of Thermal Stress, Vol. 15, (1992), pp. 1-23.

Page 129: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 1 Aggregate and Localized Microstructual Influence 129

Chapter 7 Designing For Microstructual Manipulation

This chapter will focus upon designing with microstructure in mind. This

approach builds upon the ability of layered manufacturing to influence microstructure

on a localized as well as aggregate basis. Several design rules are introduced to aid

designers working with carbon steel deposition to produce viable parts .

Chapter 7. 1 Aggregate and Localized Microstructual Influence

Influencing or altering the microstructure of a formed object to achieve certain

materials properties is a common aspect of producing viable parts. Heat treatment is

often used to homogenize the microstructure of newly formed sand castings or

machined metal prototypes to attain uniform hardness and reduce residual stress. The

surface microstructure of metal parts like gears may be deformed by peening to

impose compressive stress to increase wear resistance. Die surfaces may be

impregnated with alloying agents to improve die life. Laser hardening has been used

on camshafts to improve wear resistance.

Layered manufacturing techniques of solid freeform fabrication have the

ability to influence or control the microstructure of the prototype. Like heat treatment

and casting processes, laser deposition can affect the microstructure of the part.

However, unlike these bulk processes, laser deposition techniques like surface

cladding can have a more localized effect (Figure 7.1). Yang et al. [6.6] showed that a

CO2 laser hardening process can produce a microstructually altered region near a

part’s surface 350 µm thick.

Page 130: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 1 Aggregate and Localized Microstructual Influence 130

Figure 7.1 Surface effects of processing techniques.

The ability of laser processing to affect both aggregate and localized

microstructure as well as the ability of layered manufacturing to affect both aggregate

and localized geometry lends itself naturally to begin to think about designing with

microstructural constraints or enhancments in mind. Often the designer is limited in

her ability to improve her design because they are only focusing upon the feature level

of design. They may focus upon the need for a part’s surface to be a bearing surface or

that two mating parts will rub, and not the underlying microstructure.

If one could begin to think about the design on a microstructural level one may

find opportunities to improve the design. For example , a designer is building tensile

bars. They need to be stiff but still have good elongation properties. With

conventional processes, his choices are limited. With layered laser technique like

SDM, one could improve elongation by placing austenitic grains within strategic

locations and martensitic in areas requiring stiffness (Figure 7.2). Considering

microstrure, or Designing for Microstructual Manipulation (DFM2) adds an additional

element of freedom to design.

Four thermal fatigue specimens were built without possessing a complete

0 500 1000 1500 2000 2500 3000

Fine Al2O3 Polish

Abrasive Paper

Surface Grinding

Turning Lathe

EDM

Laser Cladding

Laser Welding

Laser Surface Alloying

Cutoff Wheel

Hacksaw

Heat Treatment

Casting

Laser Deposition

Steel Brass

Micrometers of Surface Influence from Deformation or Alloying Thickness

(.1 µm)

Page 131: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 2 Applying DFM2 to Internal Geometry of Die Inserts 131

knowledge of the microstructure (Figure 7.3). Each one failed because the deposition

strategy was not superior to the effects of the microstructure. However knowledge

gained by these experiments, have increased the knowledge of carbon steel deposition

100 fold.

Figure 7.2 Freeing Designer with Microstructure

Figure 7.3 Four Thermal Fatigue Specimens

Chapter 7. 2 Applying DFM 2 to Internal Geometry of Die Inserts

Copper cooling bars can be deposited in die inserts to improve the thermal

conductivity of the dies. With improved conductivity, faster injection cycles can be

attained without compromising casting integrity or microstructure. Placing sharp

corners in the interior of a part builds in stress concentrations which can induce

solidification cracking (Figure 7.4). If the corners could be avoided entirely the

Constrained Designers

Part Needs Elongation needs elongation

•Choose Austenitic Material

Unconstrained Designer

Interior Section needs ElongationExterior needs StiffnessChoose Graded Material Deposition:•Martensitic Exterior•Austenitic Interior

more than stiffness.

SDMT Steel“Cracky”

SDMT Steel“Splity”

410 Steel“Brittle”

H13/410 Composite“Holey”

Page 132: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 2 Applying DFM2 to Internal Geometry of Die Inserts 132

Figure 7.4 Sharp Internal Corner of 316L and Copper

situation would improve, but the difference in thermal conductivity can still produce

very sharp thermal gradient differences which can also produce stress concentrations

(Figure 7.4). However, inducing delta ferrite near the sharp or thermally divergent

interface may be a better way to avoid solidification cracking. Figure 7.5 shows a

corner which was removed by the laser only to form to additional corners.

Figure 7.5 Using laser to eliminate sharp corners is not always successful.

A better way to eliminate solidification cracking is to introduce delta ferrite

near the stress concentration. From the carbon steel characterization, we know that

420 forms delta ferrite more easily than any of the other metals tested. The 420

material was applied after an initial powderless pass to warm the copper and insure

good bonding. Figure 7.6 shows the etched surface of the interface. Delta ferrite

forms near the interface. No cracks were observed.

500 µm

316L

316L

Copper

Solidification cracknear sharp CopperCorner.

100 µm

Original Corner

Additional Corner Created

Page 133: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 3 Applying DFM2 to Part Building Strategy 133

Figure 7.6 (A.) Corner Solidification crack in SS-CU. (B.) Corner Prep with 420 Stainless

Chapter 7. 3 Applying DFM 2 to Part Building Strategy

Chapter 7. 3.1 Part Substrate InterfaceGood bonding between the parts substrate and first layers of the part are

essential to prevent delamination. As the earlier characterization shows, the first layer

is the thinnest. This thin layer is subject to both residual stress in the substrate as well

as stress within the deposit. Plasma spray coatings are applied with pressure and heat

to coat a surface. However most of the bonding is purely mechanical than

metallurgical. Mechanical bondings are easier to rupture than metallurgical bonding.

Figure 7.7 shows the similarities between a plasma coating and a laser layered deposit

of H13 on low carbon cold roll steel.

Figure 7.7 A. Plasma Spray Coating, B. Single Pass interface of H13 and low carbon steel

50 µm50 µm

316L

Cu Cu

420

316L

δ

δ

Α. Β.100 µm

100 µmCu

Stainless Steel

100 µm

Good Bonding Poor Bonding

A. B.

H13

Cold Roll Steel

Page 134: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 3 Applying DFM2 to Part Building Strategy 134

Figure 7.8 A prepass before depositing improves substrate and deposit bonding.

One way to get a better bond is by doing a prepass of the laser before

deposition. A prepass is a powderless laser scan of the substrate immediately before

depositing. The prepass warms the substrate and roughens the surface to further aid in

bonding the first layer of deposition. The bonding region doubles in size making a

much stronger bond.

This is not the same as doing a double laser pass (a powder scan followed by a

second powderless laser scan). The interfacial bonding region of a single pass vs a

double pass is about the same, on the order of 100 µm.

Chapter 7. 3.2 Part Layering StrategyBecause carbon steels will continue to transform as multiple layers of laser

deposition are attempted no more than 25 mm of layers of continuous deposition

should be attempted. If one is trying to design a die insert which is 50 mm in

thickness, try to incorporate the substrate within the design. Many die insert designs

can be split into a feature level and a bulk part. The feature level includes the casting

surface and features beneath the part surface like conformal cooling channels or other

internal features. The bulk part is simply the rest of the tool which forms the base.

The bulk part should be machined from a similar metal and used as a big substrate for

the rest of the deposited features. Special care should be used to insure proper

bonding occurs.

100 µm

Good Bonding

H13

Cold Roll Steel

Page 135: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 3 Applying DFM2 to Part Building Strategy 135

Figure 7.9 Conceptual Splitting of a Die Insert

If this two part build system cannot be done, anneal the part and substrate every

25 mm of deposition. The only way to insure that unwanted phases do not occur

within lower parts of the deposition, or that unwanted residual stress arising phase

transformations or mismatches in coefficients of thermal expansion, is to anneal the

entire deposit. However, annealing is another form of heat treatment. Special care

must be taken to characterize the appropriate annealing conditions and procedures.

Certain temperatures may cause chromium carbides to form or redistribute upon grain

boundaries which can lead to integrannular corrosion during etching procedures or

part service. Cooling procedures from the annealing temperatures must also be well

characterized. If one does not understand the affects of certain heat treatments upon

the particular carbon steel, avoid depositing over 25 mm. In this authors opinion, this

caution can be applied not only for SDM but, for any high energy - rapid solidification

procedure attempting layered manufacturing of carbon steels.

Lastly, the number of layers which are deposited at any given time should be

limited to no more than 7-12 layers. The divergence of thicknesses within the

deposition can weaken the strength of the part. Uniform grain sizes and layer

thicknesses allow parts to have more homogenous properties and attain higher

strengths. Continuous deposition beyond these limits cause severe inhomogenity in

grain size and layer thicknesses. For example, charpy impact specimens were taken

from different positions within a large deposit of SDM Tool Steels (Figure 7.10 and

Table 7.1 ). Charpy 1 samples a have a slightly higher value but a very high deviation.

Tooling Insert Bulk Part Feature Level

= +

Page 136: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 3 Applying DFM2 to Part Building Strategy 136

Figure 7.10 Charpy impact specimens taken from large deposit

Table 7.1 Charpy Impact Resistance for variable layer sizes (Sample 1 Area) and consistent layer sizes (Sample 2 Area).

Charpy samples taken from region 2 have much more uniform properties. The slightly

lower measurements are probably because they were taken from a lower position

within the laser deposition and have been tempered inappropriately by subsequent

deposition layers. (The big deposit violates the 25 mm rule.)

Chapter 7. 3.3 Deposition Parameters

For the 2400 Watt ND:YAG laser which delivers an incident power of about

400 Watt, the speed as was shown earlier is not a significant factor. With almost all

Impact Resistance (ft-lb)

Average Charpy(ft-lb)

Standard Deviation

(ft-lb)

Charpy 1-a 9.5 7.83 1.53

Charpy 1-b 6.5

Charpy 1-c 7.5

Charpy 2-a 7.5 7.5 0.00

Charpy 2-b 7.5

10.80.6

0.4

0.2

02

4

6

8

2

4

0 5 10 15 20 25 30

A

B

B

B A

A

Z-Height (mm)

Laye

r T

hick

ness

(m

m)

A = Overnight Cooling

(12-24hr breaks)B = Short Cooling Periods

(1-2hr breaks)

Charpy 2Charpy 1

20 25 30 35 40 45 50

Page 137: Thesis Glink

Chapter 7 Designing For Microstructual Manipulation

Chapter 7. 3 Applying DFM2 to Part Building Strategy 137

speeds below 40 mm/sec, SDM is still delivering enough power to insure solidification

of the deposit. However, the fill rate is critical. Figure 7.11 shows a graph which

compiles at speeds from 15 to 50 mm/sec the effects of changing fill rate. Only one fill

rate reduces the amount of 2nd phase constituents at all speeds.

Figure 7.11 For Laser Deposition, Fill Rate is a Critical Parameter

Therefore, fill rate can be used as a way of reducing or increasing the amount

or retained or transformed secondary phases. Depending upon the nature of parts

design and ultimate use, fill rate can be an important way of achieving different

material characteristics. However, unless one is sure of the affects of secondary

phases, one should find a setting which minimizes the effects. For this research that

setting was 23 g/ min.

% R

etai

ned

Aus

teni

te

0

5

10

15

20

25

30

35

10 15 20 25 30 35Fill Rate (g/min)

Layers Too Thick-Maximum Retention of 2nd Phase

Layers Too Thin-Maximum Reheating-Multiple Phase Transformations

Page 138: Thesis Glink

Chapter 8 Conclusions

Chapter 8. 1 Characterization of Carbon Steels 138

Chapter 8 Conclusions

Deposition of carbon steels is feasible and delivers a whole range of potential

applications and capabilities to Shape Deposition Manufacturing which cannot be

achieved by traditional deposition metals. However, the addition of carbon steels

requires a more detail understanding of the microstructure than was necessary for

other deposition materials. The added knowledge benefits SDM and creates new

unique opportunities to improve deposition and more importantly designed objects.

Chapter 8. 1 Characterization of Carbon SteelsKnowledge of how phases forms and how grains and layers evolve are essential

to develope effective strategies to deposit tools steels. Because SDM involves a high

energy deposition source which induces rapid solidification and uses layered

manufacturing techniques which induce multiple temperature gradients through out

the deposition, traditional designer’s tools are ineffective. The Shaeffler Welding and

Solidification Diagrams have been used by low energy welding sources to predict

phase evolution. However, for SDM they were highly inaccurate. Proposed in this

research are two new modified versions of the Shaeffler and Solidification Diagrams

which can be used by designers to more accurately select metals for layered high

energy deposition.

A heuristic for predicting grain size and layer thickness have been proposed for

SDM. The heuristic identifies the specific variables in composition, and other more

process oriented variables like, the fill rate, solidification rate and powder size as

determinants for grain and layer size. Although the heuristics are directly applicable to

SDM, the parameters may lend incite in to other high energy processes. Also, the

Page 139: Thesis Glink

Chapter 8 Conclusions

Chapter 8. 2 Application of Characterization 139

understanding of how layer evolve within SDM may also aid others in understanding

there own processes.

Chapter 8. 2 Application of CharacterizationTwo applications of using characterization to improve deposition have been

highlighted by this research. Choosing high impact resistant materials for building die

casting inserts is dependent upon understanding the microstructure in addition to bulk

material properties. Because high energy deposition processes produce material

which are very different than traditional materials, they will display unusual

properties. The small grain sizes of the deposited materials lend to these unique

properties. However, knowledge of phase evolution is important. Certain phases in

addition to the grain size can improve properties while other may reduce properties.

Delta ferrite can help prevent solidification cracking in some metals while serving as a

crack initiation point in others. Martensite can add toughness and reduce deflection,

but can also reduce elongation and yield strength. Therefore, traditional material

property knowledge must be supplemented by knowledge of the microstructure when

working with nontraditional part forming methods.

Chapter 8. 3 Designing with Microstructure

Now that an understanding of how the microstructure of laser deposited carbon

steels develops within layered manufacturing techniques has been extracted, and how

this knowledge can be used to improve deposition, this knowledge can be further

extended to aid designers. Designers can use this knowledge and improve designs

intended for layered manufacturing. Improved part strength, wear resistance, and

resistance to deformation can be accomplished if one allows the microstructure to

guide design decisions. Sooner or later, the microstructure will win, and lesser designs

which do take microstructual concerns into account will fail. Design rules which

accomplish this have been proposed in this research.

Chapter 8. 4 Leverage of This KnowledgeAs indicated by Chapter 2, there are several other high energy deposition

processes, which are also attempting build structures in a layered fashion. They two

can benefit from understanding the nature of the microstructure. Metals prototyping

has lagged behind its plastic counterparts even though they share many similar

problems linked to thermal gradients and shrinkage. Yet, plastics rapid prototyping

Page 140: Thesis Glink

Chapter 8 Conclusions

Chapter 8. 4 Leverage of This Knowledge 140

has been quite successful in changing the design process for many designers world-

wide. Leveraging the nature of metal microstructure may be the key to advancing this

technology to every designer’s door.

Page 141: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 141

Appendix

Page 142: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 142

Chapter A. 1 Selecting a Laser Scanning Speed

When trying to determine a laser scanning speed knowing the thermal

diffusion length is a potential starting point. Thermal diffusion length measures the

penetration or diffusion of heat of a laser while it is incident upon a particular material.

To determine the interaction time, or the time a moving laser is incident upon a spot,

an approximation is provided by dividing the beam with by the velocity of the beam.

An approximation for the interaction time, tp, is spot size divided by scanning speed.

The thermal diffusivity, αth, should be known for each metal. For 2000 W absorbed

into substrate or metal at a 2.8 mm spot size, the calculated thermal diffusion length

values are listed in Table :

Table A.1 Thermal Diffusion Length for Several Carbon Steels

Select an initial speed which is approximately thickness of layer which you are

attempting to deposit. This may not be the most optimized speed but it is good starting

point for optimization and characterization.

20 mm/sec 30 mm/sec

316L 1.01 mm .824 mm

SDMT 1.31 1.07

410 1.33 1.08

420 1.28 1.05

431 1.19 .97

H13 1.31 1.07

LT 2αth

tp

( )=

Page 143: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 143

Chapter A. 2 Martensitic Start Temperature

The martensitic start temperature can be calculated with knowledge of the

constituents of the materials.

Ms=521-350 C-13.6Cr -16.6 Ni -25.1 Mn -30.1Si -20.4 Mo-1.07CR * Ni +21.9(CR+.73Mo)C This formula was developed by Self and Carpenter, “Phase Transformations and Alloy Stability,” (1986).

Table A.2 Martensitic Start Temperature for Materials Used in this Research

Material Ms in Co

410 324

420 284

431 214

H13 306

SDMT 208

M2 240

316L -298

Page 144: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 144

Chapter A. 3 Theoretical Volumetric Expansion

Theoretical Percentage of Volume Increase based upon 100% martensitic

transformation.

Lyman, Metals Handbook, 8th Ed.

C%

Aus

teni

te to

Mar

tens

iteVo

lum

e P

erce

ntag

e

Dim

ensi

onal

mm

per

mm

Aus

teni

te to

Fer

rite

and

cem

entit

e

Dim

ensi

onal

mm

per

mm

Sph

erod

ite -

Tem

pere

d m

arte

nsite

to e

utec

toid

- F

eC3

glob

ules

in

410 0.06 4.6082 0.0154 4.507 0.0144 -4.5074

420 0.45 4.4015 0.0147 3.646 0.0142 -3.6455

431 0.18 4.5446 0.0152 4.242 0.0155 -4.2422

SDMT 0.15 4.5605 0.0152 4.309 0.0155 -4.3085

H13 0.4 4.428 0.0148 3.756 0.0153 -3.756

1010-low carbon

0.1 4.587 0.0153 4.419 0.0155 -4.419

Page 145: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 145

Chapter A. 4 Grain Size Heuristic

For the chosen data basis the following best model was generated:

X26= + 9.84e-2z61 + 2.30e+0z62 + 5.82e+0

z61= + 1.00e+0z51

z51= + 1.00e+0z41

z41= + 1.00e+0z31

z31= + 1.00e+0z21

z21= + 1.00e+0z11

z11= + 2.89e+0X14 - 3.54e-1

z62= - 2.40e-1z51 + 1.18e+0z52

z51= + 1.00e+0z41

z41= + 1.00e+0z31

z31= + 1.00e+0z21

z21= + 1.00e+0z11

z11= - 5.14e+0X1 + 1.49e-3X19 - 6.23e-1

z52= - 2.78e-1z41 + 1.26e+0z42

z41= + 1.00e+0z31

z31= + 1.00e+0z21

z21= - 2.24e-1z11 + 8.88e-1z12

z11= + 5.42e+0X1 - 1.30e+0

z12= + 2.91e-1X7 - 4.79e-2X24 + 4.63e-1

z42= - 1.79e-1z31 + 1.09e+0z32

z31= + 1.00e+0z21

z21= + 1.00e+0z11

z11= + 1.94e-1X7 - 2.21e+0

z32= - 1.84e-1z21 + 1.12e+0z22

z21= + 1.00e+0z11

z11= + 1.12e+0X9 - 5.01e-1

z22= + 7.91e-1z11 + 4.42e-1z12

z11= + 2.91e-1X7 - 4.79e-2X24 + 4.63e-1

z12= - 1.52e-1X16 - 2.64e-1X20 + 7.66e+0

Page 146: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 146

Mean Absolute Percentage Error (MAPE): 3.07 %

Approximation Error Variance: 0.01669

OUTPUT VARIABLE:

X26 - avg cell

RELEVANT INPUT VARIABLES:

X14 - V

X1 - C

X19 - Melting

X7 - Cr

X24 - powder

X9 - Mo

X16 - NI-EQ

X20 - K

CHOSEN HEURISTICS:

Data Length: 9

Number of Input Variables: 20

Max. Lagged Time: 0

Model Type: input-output-model / exclusively linear / static

The embraced linear model is:

X26 = 22.838715 + 0.284510X14 + 3.738204X1 -0.000820X19 + 0.642496X7 -

0.125312X24 -0.769774X9 -0.278010X16 -0.483130X20

Page 147: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 147

Chapter A. 5 Layer Thickness

For the chosen data basis the following best model was generated:

X25= + 2.50e+1z41 + 2.10e+2z42 + 7.45e+2

z41= + 1.00e+0z31

z31= + 1.00e+0z21

z21= + 1.00e+0z11

z11= + 9.19e-1X14 - 1.12e-1

z42= - 1.59e-1z31 + 9.96e-1z32

z31= + 1.00e+0z21

z21= + 1.00e+0z11

z11= + 1.84e+0X2 - 1.25e+0

z32= + 2.14e-1z21 + 1.02e+0z22

z21= + 1.00e+0z11

z11= + 1.99e-1X7 - 2.18e+0

z22= - 4.51e-1z11 + 1.37e+0z12

z11= + 1.46e-5X27 - 3.68e+0

z12= - 1.01e-1X16 - 2.68e-2X24 + 3.29e+0

Mean Absolute Percentage Error: 9.40 %

Approximation Error Variance: 0.1993

OUTPUT VARIABLE:

X25 - avg layer

RELEVANT INPUT VARIABLES:

X14 - V

X2 - Mn

X7 - Cr

X27 - solid rate

X16 - NI-EQ

X24 - powder

Page 148: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 148

CHOSEN HEURISTICS:

Data Length: 9

Number of Input Variables: 22

Max. Lagged Time: 0

Model Type: input-output-model / exclusively linear / static

The embraced linear model is:

X25 = 2003.830933 + 22.995251X14 -61.247478X2 + 8.895004X7 -0.001405X27

-29.596924X16 -7.855000X24

Page 149: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 149

Chapter A. 6 New Chromium Equivalence Definition

For the chosen data basis the following best model was generated:

X17= + 3.11e-1z21 + 6.19e+0z22 + 1.72e+1 z21= + 1.00e+0z11 z11= + 1.86e+0X14 - 3.61e-1 z22= + 2.76e-1z11 + 9.99e-1z12 z11= + 3.15e+0X4 - 1.36e+0 z12= + 3.32e-1X5 + 1.31e-1X15 - 2.29e+0

Mean Absolute Percentage Error: 6.57 %Approximation Error Variance: 0.0512

OUTPUT VARIABLE: X17 - new Cr

RELEVANT INPUT VARIABLES: X14 - VX4 - SiX5 - WX15 - Cr-Eq

CHOSEN HEURISTICS: Data Length: 16Number of Input Variables: 15Max. Lagged Time: 0 Model Type: input-output-model / exclusively linear / static

The embraced linear model is:

X17 = 0.666030 + 0.578787X14 + 5.386530X4 + 2.048034X5 + 0.811128X15

Page 150: Thesis Glink

Appendix

Chapter 8. 4 Leverage of This Knowledge 150

Chapter A. 7 New Nickel Equivalence Definition

For the chosen data basis the following best model was generated:

X18= - 5.35e-1z21 + 4.52e+0z22 + 1.14e+1 z21= + 1.00e+0z11 z11= + 6.89e-1X5 - 2.58e-1 z22= + 3.37e-1z11 + 7.53e-1z12 z11= + 1.05e+0X9 - 5.97e-1 z12= + 1.70e-1X8 - 7.72e-2X15 + 3.93e-1

Mean Absolute Percentage Error: 19.96 %Approximation Error Variance: 0.3663

OUTPUT VARIABLE: X18 - new NI

RELEVANT INPUT VARIABLES: X5 - WX9 - MoX8 - NiX15 - Cr-Eq

CHOSEN HEURISTICS: Data Length: 16Number of Input Variables: 17Max. Lagged Time: 0 Model Type: input-output-model / exclusively linear / static

The embraced linear model is:

X18 = 11.977810 -0.368111X5 + 1.598563X9 + 0.579666X8 -0.262729X15