You are on page 1of 87

Hot Extrusion of Carbon Nanotube - Magnesium Matrix Composite Wire

A thesis presented to

the faculty of

the Russ College of Engineering and Technology of Ohio University

In partial fulfillment

of the requirements for the degree

Master of Science

Vishal S. Dongare

December 2014

© 2014 Vishal S. Dongare. All Rights Reserved.


2

This thesis titled

Hot Extrusion of Carbon Nanotube - Magnesium Matrix Composite Wire

by

VISHAL S. DONGARE

has been approved for

the Department of Mechanical Engineering

and the Russ College of Engineering and Technology by

Frank F. Kraft

Associate Professor of Mechanical Engineering

Dennis Irwin

Dean, Russ College of Engineering and Technology


3

ABSTRACT

DONGARE, VISHAL S., M.S., December 2014, Mechanical Engineering

Hot Extrusion of Carbon Nanotube - Magnesium Matrix Composite Wire

Director of Thesis: Frank F. Kraft

The main goal of this research was to develop a lab-based process to produce a

carbon nanotube (CNT)-Mg matrix composite wire via hot extrusion. An apparatus was

developed for use with a 250 kN servo hydraulic MTS® machine to perform hot

extrusion in this research. The CNT-Mg billets with 3 wt.% CNTs were provided by

external researchers. The scope of the project involved modeling the process analytically

and performing a series of extrusion trials to determine flow properties of the composite

material. Material parameters for the classical Zener-Hollomon model was determined

using compression testing to characterize the plastic behavior of pure Mg as a function of

temperature and strain rate. To show consistency with the Zener-Hollomon model, three

extrusion trials with pure Mg were conducted. Extrusion trials were also performed on

the CNT-Mg composite material. To produce a refined composite structure of CNTs in

the Mg matrix, re-extrusion trials were performed with re-pressed CNT-Mg composite

billets. These new billets were produced from the extruded wire obtained in the previous

extrusion. The effect of CNT addition on Mg and the effect of re-extrusions on the

composites’ flow properties were assessed. The addition of CNTs increased the apparent

flow stress of Mg by 10 – 30%. The apparent flow stress of the composite was further

increased by 50 – 70% as a result of repressing and re-extrusion, compared to pure Mg.


4

ACKNOWLEDGMENTS

The author would like to thank thesis advisor Dr. Frank Kraft for his for all his

support and guidance. The author would also like to thank Thesis committee members

Dr. Cotton, Dr. Ali and Dr. Savin for their valuable inputs to this research. Finally, the

author is thankful to International Copper Association (ICA) for providing financial

support for this work.


5

TABLE OF CONTENTS

Page

Abstract ............................................................................................................................... 3
Acknowledgments............................................................................................................... 4
List of Tables ...................................................................................................................... 7
List of Figures ..................................................................................................................... 8
Chapter 1 - Introduction .................................................................................................... 11
1.1 Background ............................................................................................................. 11
1.2 CNT- metal matrix composites ............................................................................... 13
1.3 Direct Hot Extrusion ............................................................................................... 14
1.4 Mathematical modeling of extrusion ...................................................................... 15
1.5 An estimation of extrusion pressure ....................................................................... 15
1.6 Objectives ............................................................................................................... 16
Chapter 2 - Literature Review........................................................................................... 17
2.1 Different processes to fabricate CNT-metal matrix composites ............................. 17
2.1.1 Powder metallurgy ........................................................................................... 18
2.1.2 Electrochemical Deposition ............................................................................. 19
2.1.3 Melting and casting .......................................................................................... 19
2.1.4 Plasma spraying ............................................................................................... 20
2.1.5 Some innovative ways of producing MMCs ................................................... 20
2.2 Strengthening mechanism of a metal matrix composite ......................................... 20
2.3 CNT-Cu and CNT-Al matrix composites ............................................................... 21
2.4 CNT-Mg matrix composites ................................................................................... 22
Chapter 3 - Equipment and Procedure .............................................................................. 25
3.1 Overall extrusion assembly and setup .................................................................... 25
3.2 Ram stem design ..................................................................................................... 29
Chapter 4 – Flow Stress Analysis ..................................................................................... 31
4.1 Development of an extrusion model ....................................................................... 31
4.2 Parameters for the flow stress estimation ............................................................... 34
6

4.3 Compression testing ................................................................................................ 34


4.4 Friction ring testing ................................................................................................. 35
Chapter 5 - Plan of Work .................................................................................................. 39
5.1 Compression testing on pure magnesium ............................................................... 40
5.2 Extrusion with pure magnesium billets .................................................................. 42
5.3 Extrusion with the CNT-Mg composite billets ....................................................... 42
Chapter 6 - Results and Discussion .................................................................................. 45
6.1 Compression test results ......................................................................................... 45
6.1.1 Strain rate dependence of the flow stress of pure Mg ...................................... 51
6.1.2 Strain rate and temperature dependence of the flow stress of pure Mg ........... 53
6.1.3 Validation of the Zener-Hollomon model........................................................ 57
6.2 Extrusion results ..................................................................................................... 62
6.2.1 First extrusion of CNT-Mg composite billets .................................................. 65
6.2.2 Re-extrusion of CNT-Mg composite billets .................................................... 68
6.2.3 Third extrusion of CNT-Mg composite billet no. 2 ......................................... 71
6.3 Extrusion results analysis and discussion ............................................................... 72
Chapter 7 – Summary and Conclusions ............................................................................ 78
References ......................................................................................................................... 81
7

LIST OF TABLES

Page

Table 1 : Essential parameters for flow stress calculation ................................................ 34


Table 2 : Coefficient of friction µ and friction factor (𝑚𝑓) for pure Mg at 300C and
450C ................................................................................................................................. 38
Table 3 : Compression testing plan for pure Mg .............................................................. 40
Table 4 : Test plan for extrusion on pure Mg billets......................................................... 42
Table 5 : Hot extrusion test plan with CNT-Mg composite billets ................................... 43
Table 6 : Flow Stress Values of pure Mg at various conditions ....................................... 46
Table 7 : Average constant flow stress values for pure Mg at 300C and 0. 1 s-1 strain rate
........................................................................................................................................... 49
Table 8 : Material constants for pure magnesium at 300C and 450C............................. 53
Table 9 : Experimental flow stress values by Sheng and Shivpuri in tensile testing and
flow stress values predicted by Zener-Hollomon model for pure Mg .............................. 62
Table 10 : Maximum pressure required to extrude the CNT-Mg billets with 1 inch lengths
........................................................................................................................................... 73
Table 11 : Densities of the CNT-Mg composite billets before first, second and third
extrusion trials as received ................................................................................................ 74
8

LIST OF FIGURES

Page

Figure 1 : A schematic of single walled carbon nanotube capped on both sides with C60
fullerene hemispheres (adapted from American Scientist, magazine of Sigma Xi, The
Scientific Research Society. Illustration by Aaron Cox.) ................................................. 11
Figure 2 : a) Annotated sketch of the extrusion apparatus, b) Photo of the apparatus
assembled on the MTS machine before extrusion trial ..................................................... 26
Figure 3 : Extrusion assembly insulated with white silica insulation during the trial.
Composite wire (encircled) is protruding out of the gas tube. .......................................... 28
Figure 4 : a) a CAD model of extrusion die, b) Cross section of extrusion die. Die entry
and exit are also indicated. ................................................................................................ 29
Figure 5 : Photo of a composite billet, as received, similar to the ones that were used for
extrusion............................................................................................................................ 29
Figure 6 : Calibration curves to determine a) Coefficient of friction (𝜇), and b) friction
factor (𝑚𝑓) from ring compression testing [54] .............................................................. 36
Figure 7 : a) The specimen prepared for testing is on the left with the thermocouple
attached to it. The sample on the right has been compressed at an elevated temperature. 37
Figure 8 : Photo of experimental setup for compression testing on pure Mg ................... 41
Figure 9 : Flow stress data as a function of true strain for pure Mg at strain rates 0.01 s-1,
0.1 s-1 and 1 s-1 at 300C.................................................................................................... 45
Figure 10 : Flow stress data as a function of true strain for pure Mg at strain rates 0.01 s-1,
0.1 s-1 and 1 s-1 at 450C.................................................................................................... 46
Figure 11 : A typical stress strain curve for pure Mg at constant strain rate and
temperature ....................................................................................................................... 47
Figure 12 : Compression data for tests 1 – 3 at stain rate 0.1 s-1 and at 300C to evaluate
experimental accuracy ...................................................................................................... 49
Figure 13 : Compression test data for trial with pure Mg at a strain rate of 0.1 s-1 at 300C
........................................................................................................................................... 50
9

Figure 14 : Compression test data for trial with pure Mg at a strain rate of 1 s-1 at 450C
........................................................................................................................................... 51
Figure 15 : A plot of ln𝜎 versus ln𝜀 estimated from the compression data for pure Mg . 52
Figure 16 : A plot of 𝜎 as a function of ln𝜀 ...................................................................... 54
Figure 17 : A graph showing ln𝜀 plotted against reciprocal of absolute temperature (1/𝑇)
........................................................................................................................................... 55
Figure 18 : A plot of ln𝜎 as a function of ln𝑍 .................................................................. 56
Figure 19 : A graph showing comparison between compression data for tests 1 to 11 and
a Zener-Hollomon model .................................................................................................. 57
Figure 20 : Extrusion data for pure Mg at a constant ram speed of 0.076 mm/s at 530C 58
Figure 21 : Extrusion data for pure Mg at a constant ram force of 17,780 N at 530C .... 59
Figure 22 : Photo of a distorted portion of wire due to die load-up.................................. 60
Figure 23 : Extrusion data for pure Mg at 470C and at 0.127 mm/s ram speed .............. 61
Figure 24 : A typical extrusion pressure curve showing various components of total
extrusion work .................................................................................................................. 63
Figure 25 : Experimental data for first extrusion on the CNT-Mg billet 1 at 500C and at
constant ram speed of 0.076 mm/s.................................................................................... 64
Figure 26 : a) A schematic of die and container showing semi-die angle 𝜶 and extrusion
pressure P, b) The formation of dead metal zone due to friction during extrusion. The
effective die angle 𝛼𝑒 is indicated. c) Metal flow in the middle of extrusion, d) An
instance where ram starts pushing dead metal into the deformation zone at the end of
extrusion............................................................................................................................ 65
Figure 27 : Pressure data for first extrusion on CNT-Mg billets 1 to 3 at a constant ram
speed of 0.076 mm/s at 500C superimposed with the prediction of extrusion pressure for
pure Mg using Equation 8 and Zener-Hollomon model at the same conditions .............. 66
Figure 28 : Temperature data for first extrusion trials on CNT-Mg billets 1 to 3 ............ 67
Figure 29 : Experimental data for first and second extrusion on the CNT-Mg billet no. 1
at a constant ram speed of 0.076 mm/s at 500C .............................................................. 69
10

Figure 30 : Experimental data for first and second extrusion on CNT-Mg billet no. 2 at a
constant ram speed of 0.076 mm/s at 500C ..................................................................... 70
Figure 31 : Temperature data for second extrusion trials on CNT-Mg billets 1 and 2 ..... 70
Figure 32 : Experimental data for the third extrusion on CNT-Mg billet no. 2 ................ 71
Figure 33: Experimental data for the extrusion trials with pure magnesium billet and re-
pressed magnesium billet at a constant ram speed of 0.127 mm/s at 470C ..................... 75
Figure 34: Temperature data for the extrusion trials with pure magnesium and re-pressed
magnesium billet ............................................................................................................... 76
11

CHAPTER 1 - INTRODUCTION

1.1 Background

Carbon nanotubes (CNTs) are being used in a variety of applications recently in

engineering and in the medical field. With the discovery of CNTs in 1991 [4], carbon

nanotubes have shown a tremendous potential in various structural and electrical

applications. CNTs are reported to have excellent mechanical and electrical properties,

such as a tensile strength of 50 – 200 GPa, Young’s modulus of 800 GPa and a capacity

of carrying electric current up to a density of 4 × 109 A/cm2 (300% of aluminum’s) [2 –

4]. Figure 1 shows a schematic of single walled carbon nanotube, a cylinder of graphite

sheet with a very high length to diameter ratio [5]. Multi walled nanotubes consist of

several such concentric graphite cylinders [5].

Figure 1 : A schematic of single walled carbon nanotube capped on both sides with C60

fullerene hemispheres (adapted from American Scientist, magazine of Sigma Xi, The

Scientific Research Society. Illustration by Aaron Cox.)

In various studies, it has been argued that mechanical and electrical properties of a

base metal can be improved significantly by introducing CNTs into a metal matrix.
12

Hence, CNT- metal matrix composite (MMC) is gaining attention because of its

potentially superior overall properties compared to a pure base metal [6].

Metals, such as copper, silver, and aluminum, were once believed to have the

lowest electrical resistance at room temperature; hence, they have been extensively used

to transport electricity. Copper is the most commonly used metal to produce current

carrying wires. It has a better tensile strength, ductility, corrosion resistance, and thermal

conductivity compared to other conducting materials [7] and it is also available in

abundance. Silver is a metal with higher electrical conductivity than copper, but it is

considerably more costly. Gold is sometimes used in plating electrical contacts [8]

because of its high resistance to oxidation. Aluminum wire is also widely used in

electrical power transmission because it is lightweight and more economical than copper.

Magnesium has slightly lower electrical conductivity than aluminum. However, it is

rarely used as a current carrying conductor due to its lower electrical conductivity, poor

ductility and a 30% higher cost than aluminum. Some non-metals, such as graphite and

silicon, have also been used to carry electricity. They typically have inferior electrical

properties compared to copper at room temperature [8].

There is an upper limit to the electrical conductivity of pure metals. A fraction of

generated electrical power is lost in transportation because of the use of conventional

metals [7]. According to the Energy Information and Administration (EIA), 7% of the

total power (approximately 290 billion kWh) is lost during transmission in the USA [9].

Power is lost because of the heat generation which is directly proportional to the

resistance of a conductor and the square of a current flowing through it. The energy lost
13

in transmission can be reduced by increasing the voltage or by reducing the resistance of

a conductor. Hence, a reduction in transportation loss may be achieved by improving

electrical properties of a conductor. A CNT-metal matrix composite with improved

electrical and mechanical properties can be considered a viable alternative to traditional

electrical conductors.

1.2 CNT- metal matrix composites

Effective MMC fabrication is needed to achieve desired properties in the

composite. In one study, Hjortstam et al. [10] concluded that the metal matrix composites

(MMCs) with 30 – 40% CNTs by volume may have up to double the electrical

conductivity than normal metals. They further suggested that an appropriate

manufacturing process needs to be developed to fabricate the composites so that CNTs

can be used to their full potential. They also mentioned that nanotubes should preferably

be aligned with the direction of electrical current, they should be well distributed within

the matrix, and they should make sufficient contact with the metal matrix material. High

electrical conductivity and sufficient mechanical strength and ductility are some of the

essential properties of a good electrical conductor. Embedded CNTs in metal matrix

composites may increase electrical conductivity, and they may also increase tensile

strength, ductility, and improve other useful mechanical and thermal properties of the

base metal [10].

Recently, magnesium has attracted a lot of attention as an engineering metal due

to its low density (1.74 g/cm3). Magnesium alloys are considered as a feasible option for

lightweight applications in automotive and aerospace industries. Few studies have been
14

conducted to investigate the possible improvement of overall properties of magnesium by

adding CNTs. The main focus of this research is to develop an approach and technical

data by which CNT-Mg matrix composite wire can be produced by hot extrusion.

External researchers (ICA funded) have developed proprietary approaches to synthesize

CNT-Mg mixture by which extrusion billets for this research were produced. The

development of a manufacturing process to convert these billets into wire was a primary

objective of this research. The billets were hot extruded through a specially designed

conical die to produce the composite wire.

1.3 Direct Hot Extrusion

In this research, CNT-Mg matrix composite billets were processed via direct hot

extrusion to obtain a composite wire. Oberg et al. [11] explained that extrusion is a

plastic deformation process, where the material is pushed through the die to produce a

desired shape. They further classified extrusion into three types depending on the

working temperature: cold, warm, and hot extrusion. They also mentioned that in hot

extrusion, the material is pushed through the die above the recrystallization temperature

of the work piece. The effect of strain hardening is avoided at high temperature and

hence, the press loads are reduced [11]. Moreover, large strain, under high temperature,

refines the grain structure and produces uniform physical properties [12]. Other

advantages of hot extrusion of non-ferrous metals involve achieving low tolerances,

material savings, and design flexibility [12].


15

1.4 Mathematical modeling of extrusion

Process modeling involves determining the extrusion work per unit volume (𝑤𝑎 ),

which is mathematically equal to the extrusion ram pressure. The components of work

required for extrusion are listed below.

1. Ideal deformation work in the die (𝑤𝑖 )

2. Friction work in the die (𝑤𝑓𝑑 )

3. Friction work in the container (𝑤𝑓𝑐 )

4. Redundant (non-uniform or non-ideal) work in the die (𝑤𝑟 )

Ideal work is the energy required for uniform deformation of the material,

neglecting the effect of friction and non-uniform deformation. The force required to

overcome friction in the conical die is essentially constant throughout extrusion, since

material always remains in the die, once it is filled. On the other hand, friction force in

the container decreases with the billet length. Redundant work is the energy required to

overcome non-uniform deformation of the material passing through a deformation zone

[13].

1.5 An estimation of extrusion pressure

Various components of the extrusion work, such as ideal work, friction work, and

redundant work were determined to estimate the actual extrusion work per unit volume in

this research. Actual extrusion work per unit volume, 𝑤𝑎 , is the summation of all these

components and it is also a function of the billet’s flow stress 𝜎 [14]. Hence,

determination of flow stress of the material, at a given temperature and a strain rate,

becomes a necessity. An estimation of the flow stress from the extrusion data is discussed
16

in Chapter 3. The actual work per unit volume (𝑤𝑎 ) is mathematically equal to the ram

pressure exerted on the billets [13]. Thus, the cross sectional area of the container bore

multiplied by 𝑤𝑎 is the required extrusion force. Force estimations determined in this

manner were used as a guide to develop a research plan and subsequent material and

process model.

1.6 Objectives

The main goal of this research was to develop a lab-based process to produce a

metal matrix composite wire with CNTs. The lab process used a 250 kN servo-hydraulic

MTS® machine to provide the extrusion force. The CNT-Mg matrix composite billets

were provided by ICA funded researchers. The objectives of this study were:

o To develop and evaluate a lab based process to produce a wire from CNT-Mg

matrix composite billets

o To design and fabricate an apparatus to perform hot extrusion in the lab

o To model the process mathematically and to determine flow properties of CNT-

Mg matrix composite material from process data

o To determine the effect of CNTs on the flow properties of magnesium and

compare process parameters to pure wrought magnesium.


17

CHAPTER 2 - LITERATURE REVIEW

Various techniques of producing CNT-metal matrix composites are discussed in

this section. Also, a possible strengthening mechanism of MMC is briefly reviewed

followed by a discussion on magnesium, copper and aluminum based CNT- metal matrix

composites.

2.1 Different processes to fabricate CNT-metal matrix composites

CNT- metal matrix composites are produced using a variety of processes. Powder

metallurgy is a commonly adapted technique for the synthesis of MMCs [6]. Other

techniques, such as electrochemical deposition, melting and casting, thermal plasma

spraying, nano-scale mixing and molecular mixing, have also been developed in the

effort of obtaining composites with well dispersed CNTs in the metal matrix [6]. Bakshi

et al. [6] stated that the primary requirement for MMC fabrication is that CNTs should be

well distributed in the metal matrix to obtain uniform properties. They further suggested

that the surface interactions and the chemical stability between the reinforcing material

and the base metal play an important role in achieving desired mechanical properties.

CNTs have a tendency to form clusters because of their large surface area which may

lead to an overall reduction of the mechanical strength of a composite [6]. Hence, proper

distribution and interfacing of CNTs into the metal matrix while blending become

important aspects of the research in this field. Moreover, retention of undamaged CNTs is

expected from a process used to prepare a metal matrix composite [6]. Thus analyzing

different production processes for the fabrication of CNT-MMC is necessary.


18

2.1.1 Powder metallurgy

Most of the research with the powder metallurgy technique has been undertaken

to develop aluminum and copper based MMCs [15-17]. A few attempts were also made

to fabricate MMCs with other base metals, such as Mg, Ag, Ti, and Ni [6]. The basic

steps involved in this technique are mixing CNTs with the metal powder followed by

compaction and sintering. Sintering is a heat treatment process that bonds, densifies, and

strengthens the compacted material [18].

Martin et al. [19] demonstrated that powder particles experience both position

rearrangement and plastic deformation together during compaction. Compacted material

is still porous in nature. Hence, to model the densification of powder metal during

compaction, constitutive equations for porous materials can be used. Some of the well-

known models are as follows: Green’s model [20], Gurson’s model [21], Kunh and

Downey’s model [22], and Shima and Oyane’s model [23]. However, in the case of near

fully dense materials, consideration of porosity may complicate the flow stress analysis.

In the current research, the billets are expected to be near fully dense. Also the process

will be performed at an extrusion ratio of 60 and at temperatures greater than 500C. The

extrusion ratio is defined as the ratio of the cross section area of a billet to the cross

section area of a wire. Ryabicheva and Usatyuk [24] concluded that copper rod produced

by extrusion of porous billets (with 95% density), at an extrusion ratio of 16.8 and a

temperature of 920C, was nearly fully consolidated. In another study, Goh et al. [25]

claimed that a theoretical density of 99.5% was achieved after hot extrusion of CNT-Mg

billets with 2 wt.% CNTs at 350C and at an extrusion ratio of around 20. Nevertheless,
19

the effect of porosity on the flow stress is assumed to be negligible to avoid mathematical

complexity in the present study.

2.1.2 Electrochemical Deposition

Electrochemical deposition has been sometimes used to produce MMCs. A one

dimensional metal composite with CNT coating can be obtained using this method [6].

These composites are used for various applications, such as nano-sensors, electrodes, and

magnetic recorder head and interconnects [6]. Even though the electrochemical

deposition technique has been explored significantly, the application of this process is

limited since this technique is only used to create a coating of CNTs on the surface of the

base material. Homogeneous dispersion of CNTs within the matrix could only be

achieved with further post deposition processing.

2.1.3 Melting and casting

The technique of melting the matrix metal and solidification with the addition of

CNTs has also been used to produce MMCs. However, this process may be limited to the

matrix metals with a relatively low melting point. Moreover, at elevated temperatures,

CNTs tend to form clusters and a homogeneous dispersion of CNTs within the matrix is

difficult to achieve [6]. A few attempts have been made using a casting approach where

alloy powders mixed with CNTs were compacted, melted, and cast to produce MMCs

[25, 26]. In one study, Zhou et al. [27] showed that hardness and wear resistance of the

CNT-Mg composite were increased with the increase in volume fraction of CNTs.

However, very few studies have been performed to analyze the strengthening of MMCs

fabricated solely by the melting and casting operations.


20

2.1.4 Plasma spraying

Plasma spraying is an innovative way of MMC fabrication. Laha et al. [28] have

suggested that molten or semi-molten particles can be sprayed onto the base material to

obtain a desired composite by impact and solidification. Bakshi et al. [29] claimed that

the homogeneous dispersion of CNTs in MMCs was obtained by plasma spraying.

Additionally, they found that the degree of porosity was decreased by using this

technique. They further claimed that retention of undamaged CNTs was achieved.

However, mechanical behavior of the composite under plastic loading was not

investigated.

2.1.5 Some innovative ways of producing MMCs

A few innovative methods, such as molecular mixing [30, 31], vapor deposition

[15, 32, 33], friction stir processing [34], and nano-scale dispersion [35, 36] have been

developed to obtain the metal composites with CNTs. But these techniques require

further investigation in order to use their full potential. Additionally, some of these

techniques may have to be combined with other methods, such as, pressing, sintering and

hot extrusion, to obtain promising results.

2.2 Strengthening mechanism of a metal matrix composite

Mechanical behavior of fiber reinforced materials has been analyzed extensively

using the shear lag model [37-39]. This model is based on the assumption that stress

transferred to the fiber is a shear stress at the surface and that there is a perfect interface

bonding between metal particles and a reinforcement material.


21

The average normal stress in the loading direction using this model can be given

by,
𝜎𝑚
𝜎𝑓 = 𝜎𝑚 + 𝑠 (1)
2
where 𝜎𝑓 is the applied normal stress, 𝑠 is the aspect ratio (the ratio of length to diameter

of a reinforcing fiber), and 𝜎𝑚 is the flow stress of the matrix. The flow stress 𝜎 of the

composite can be predicted using following equation.


𝑉𝑓 𝜎𝑚 𝑠
𝜎= + 𝜎𝑚 (2)
2
where 𝑉𝑓 is the volume fraction of the fiber elements.

The shear lag model may be adapted to analyze the MMCs with CNTs. However,

this model overestimates the flow stress in the direction of loading since it considers a

perfect alignment of the reinforcement material. CNTs may not be perfectly aligned to

the loading direction. Ryu et al. [40] proposed a generalized shear lag model. They

suggested that the aspect ratio s can be expressed as a function of misalignment angles.

However, further investigations need to be performed in order to obtain promising

results.

2.3 CNT-Cu and CNT-Al matrix composites

Copper and aluminum are widely used in the transportation of electricity.

Therefore, an improvement in electrical and mechanical properties of Cu and Al has been

the primary focus recently. Overall properties are related to the volume fraction of CNTs

added to the base material [6]. Powder metallurgy was the first technique attempted to

fabricate the CNT-Cu matrix composite in order to improve conductivity of copper [6].

Very little or no improvement was found in the conductivity of the CNT-Cu matrix
22

composite by Yang et al. [41]. However, they found an increase in hardness, yield

strength, and elastic modulus. In another study [30], about 200% increase in yield

strength, and 70% increase in elastic modulus of the CNT-Cu matrix composite

compared to pure copper was found using molecular mixing technique.

Analysis of thermal properties may be useful to estimate electrical properties of

MMCs because of the correlation between the two. Ngo et al. [42] showed that the CNT-

Cu matrix composite obtained by electro deposition with 40 vol.% CNTs possessed lower

thermal resistivity than unreinforced copper.

According to Tu et al. [16], the CNT-Al matrix composite material prepared by

spark plasma sintering possessed a higher yield strength and elastic modulus than

unreinforced aluminum. Xu et al. [43] found a 66% increase in the electrical resistivity of

the CNT-Al composite compared to unreinforced aluminum. They concluded that

insufficient surface interactions and possible clustering of CNTs may have caused

resistance to the electricity transfer.

2.4 CNT-Mg matrix composites

As previously indicated, magnesium has gained a lot of attention due to its

suitability as a structural metal in lightweight applications. Pure Mg has inferior

mechanical properties and it is thus alloyed or otherwise reinforced to improve its

mechanical strength. Particulate reinforced Mg composites have been more of a primary

focus recently than fiber reinforced Mg composites, due to increased production rate,

reduced costs and simpler manufacturing [25]. The CNT-Mg matrix composites are
23

becoming popular because of their superior overall properties compared to magnesium

[25].

Powder metallurgy (PM) [44-46] and disintegrated melt deposition (DMD) [25,

47] are the commonly used techniques to fabricate CNT-Mg matrix composites. Goh et

al. [25, 44, 48] studied the effect of weight fractions of CNTs on magnesium. They found

an improvement in the yield strength (11%) and ductility (69%) when 1.3 wt.% CNTs

were incorporated into Mg matrix. However, a significant deterioration of the properties

was observed when CNTs of more than a threshold of 1.3 wt.% were added. Correno et.

al. [46] claimed that a homogeneous dispersion of CNTs was achieved and Young’s

modulus of the composite was increased by 9% when 2 wt.% CNTs were mixed in

magnesium matrix. However, no investigation was done on the flow properties of the

CNT-Mg matrix composites.

A few attempts were also made to analyze other properties of the CNT-Mg matrix

composite material. These include corrosion resistance, fatigue strength and thermal

stability. Although the composite was found to be more thermally stable than monolithic

Mg [44], corrosion resistance [47] and fatigue strength [48] were reduced significantly

due to the presence of CNTs. A few studies were directed toward the effect of CNTs

incorporated in magnesium alloys such as, AZ91D [45] and AZ31 [34]. An improvement

in mechanical strength [45] and grain structure [34] was observed in CNT-Mg matrix

composite material compared to the monolithic Mg. However, no analysis was performed

on the flow stress of the CNT-Mg matrix composite material. Also, no study investigated

the effect of CNTs on the flow properties of magnesium. These studies also lack a
24

comparative analysis between the CNT-Mg matrix composite material and pure wrought

Mg with respect to plastic loading.

In this research, a mechanical behavior of the CNT-Mg matrix composite material

under plastic loading at elevated temperature was evaluated. The work was in the context

of the hot extrusion of small CNT-Mg billets into wire. The maximum extrusion pressure

was modeled mathematically with respect to strain rate and temperature. The material

was assumed nonporous with homogenous properties throughout the analysis. The flow

properties of magnesium were analyzed using compression testing. A Zener-Hollomon

model was proposed to establish flow properties of pure wrought Mg. Hot extrusion was

performed in the lab to obtain the CNT-Mg composite wire. Process parameters were

compared between the CNT-Mg composite and pure wrought Mg. The effect of CNTs on

the flow properties of magnesium in extrusion was also investigated in this research.
25

CHAPTER 3 - EQUIPMENT AND PROCEDURE

Forming of wire from the CNT-Mg matrix composite billets was achieved via

direct hot extrusion at temperatures of around 500 – 550C in this research. The extrusion

ratio was approximately 60. Small billets were provided by ICA funded researchers and

they were extruded into wire of near 12 AWG (American wire gauge) dimensions (~ 2

mm diameter). As a part of this effort, an extrusion apparatus was developed for use with

a servo hydraulic MTS testing machine. All related components of the apparatus were

designed with the 3D modeling software Solid Edge™. A nickel based super alloy ATI

720™ was chosen for the key components of the apparatus due to its superior strength at

elevated temperatures [49].

3.1 Overall extrusion assembly and setup

Figure 2 shows a simple annotated sketch and photo of the experimental extrusion

apparatus. The apparatus consists of components including the die and die holder, die

backer, container (mantle and liner), ram stem and ram stem holder. A somewhat similar

apparatus was developed previously for the extrusion of copper multi-channel tubing [49,

50]. In the present study, extrusion was performed with billets and special conical die to

form wire. The apparatus uses a 250 kN servo hydraulic MTS machine to provide

pressure for extrusion. The ram control is PC-operated using object oriented

programming. The ram can be controlled through force and speed control. The apparatus

was mounted to the MTS machine using upper and lower mounts developed previously

[49]. All key components in surface contact with each other were sprayed with boron
26

nitride, a high temperature lubricant and a releasing agent. It prevents sticking of the

components with each other at high temperature [50].

a) b)

Figure 2 : a) Annotated sketch of the extrusion apparatus, b) Photo of the apparatus

assembled on the MTS machine before extrusion trial

Temperatures of the container and the die were maintained using separate

controllers. Eight cartridge resistance heaters are inserted in the container and, two

heaters are inserted in to the die holder. These heaters have special grooves to

accommodate thermocouples. Feedback to the controllers is by two K-type

thermocouples placed directly in contact with the container heater and the die heater.

Using this feedback, the controllers activate solid state relays to maintain the
27

temperatures by the heaters. Additionally, two K- type thermocouples were used to

monitor temperatures at two key locations, die entry and container inner wall.

Figure 3 shows the extrusion apparatus insulated with white silica insulation during the

trial. The insulation prevents excessive heat loss to the surroundings and ensures a near

isothermal process.

The extruded wire is shielded with argon gas within the tube indicated in Figure

3. During extrusion, the ram stays stationary and the container is pushed upwards by

hydraulic pressure of the machine to achieve direct extrusion. The billets in this manner

are effectively pressed down by the ram/dummy block through conical die to form wire.

A graphite block directs wire from the die backer exit to the argon gas-tube. The

experiments were performed at relatively low speeds so that sufficient cooling of the wire

takes place in the argon atmosphere to prevent oxidation of wire. The argon gas is heavier

than air. Therefore, to avoid gas from escaping via the tube exit without entering in the

apparatus, an upward bend is provided in the tube.


28

Figure 3 : Extrusion assembly insulated with white silica insulation during the trial.

Composite wire (encircled) is protruding out of the gas tube.

The CAD model of the extrusion die and its cross section are presented in Figure

4. Die entry and exit are also indicated. Figure 5 shows a photo of the CNT-Mg matrix

composite billet as received from ICA funded researchers. A semi-die angle of 45o and a

bearing length of 0.02 inch were selected as design parameters for the die. Some reasons

behind selecting a high semi-die angle are that it decreases friction work in the die, it

minimizes un-extruded material and it may promote homogeneous dispersion and

redistribution of CNTs in the metal matrix. One possible drawback is that a high semi-die

angle increases non-uniform deformation work.


29

Semi-die angle Entry

Exit

a) b)

Figure 4 : a) a CAD model of extrusion die, b) Cross section of extrusion die. Die entry

and exit are also indicated.

Figure 5 : Photo of a composite billet, as received, similar to the ones that were used for

extrusion

3.2 Ram stem design

The extrusion pressure is applied to the billets by the ram stem via the dummy

block. An important design constraint of the ram stem is potential buckling and yielding

at an elevated temperature. The yield strength (𝜎) and elastic modulus (𝐸) of its material
30

(ATI 720™) at 600 C are 170 kpsi and 25,000 kpsi respectively [51]. The maximum

forces available for extrusion to avoid buckling and yielding (in compression) of the ram

stem were determined using expressions given in Equations 3 and 4 respectively. The

formula in Equation 3 is known as Euler’s formula.


𝜋 2 𝐸𝐼
𝐹𝑏 = (3)
(𝐾𝐿)2
𝐹𝑐 = 𝜎𝐴 (4)

where 𝐹𝑏 is the critical force in buckling, 𝐼 is the area moment of inertia, 𝐾= 2 is the

length factor (for length fixed on one end and free on the other end), 𝐿 is the unsupported

length of ram stem, 𝐹𝑐 is the critical force in yielding, and 𝐴 is the cross section area of

ram stem. The maximum forces available at 600C for extrusion using the buckling and

yielding criteria (Equations 3 and 4) are 43,220 lbs and 50,650 lbs respectively.
31

CHAPTER 4 – FLOW STRESS ANALYSIS

4.1 Development of an extrusion model

The development of a functional process model is important for future prototype

production to scale-up the process. The analytical model to determine ram pressure with

respect to ram speed and extrusion temperature is presented in this section. The ram

pressure is mathematically equal to the total work per unit volume (including components

of ideal and redundant plastic deformation and friction with the tooling). The total work

per unit volume or the extrusion work (𝑤𝑎 ) is a summation of the work required in the die

(𝑤𝑑 ) and the work required in the container (𝑤𝑐 ) [14].

𝑤𝑎 = 𝑤𝑑 + 𝑤𝑐 (5)

The work expended in the die includes ideal work, non-uniform or redundant

work and friction work. Ideal work is required for uniform deformation of material

through the die. Material at the center of billet experiences pure elongation in extrusion

and it is associated with the change of cross section of billet [52]. The material near

container wall, on the other hand, undergoes extensive shear deformation due to container

and billet friction [52]. Redundant work is the energy expended to overcome this shear

deformation.

The work required in the die can be expressed by Equation 6, as a function of

ideal work and an efficiency term ( ), which accounts for redundant and friction work in

the die [13],


𝜎 ln 𝑅
𝑤𝑑 = (6)
𝜂
32

where 𝜎 the flow stress of the material, 𝑅 is the extrusion ratio, the ratio of the cross

section areas of the billet/container to the wire. The efficiency 𝜂 can be estimated from

extrusion data (or even from analytical methods).

The work in the container is due to friction between material and container wall.

It is expressed in Equation 7 assuming constant interfacial shear stress [53].

4𝜎 𝑚𝑓 𝑙𝑏
𝑤𝑐 =
√3𝑑𝑏 (7)

where 𝑙𝑏 is the length of billet, 𝑑𝑏 is the diameter of billet/container bore, and the friction

factor 𝑚𝑓 is equal to 1 assuming sticking friction between billet and container. The 𝜎⁄√3

term in Equation 7 represents the von Mises shear flow stress of the material. The initial

billet diameter is slightly less than the container bore to allow easy insertion, but the billet

expands or “upsets” to the container bore diameter at the start of extrusion.

The extrusion pressure (𝑃𝑒 ) is equal to the total work done per unit volume (𝑤𝑎 )

on the material such that, 𝑃𝑒 = 𝑤𝑎 . Hence, the extrusion pressure can be expressed as a

function of flow stress (𝜎) by combining Equations 5-7.

ln 𝑅 4 𝑚𝑓 𝑙𝑏
𝑃𝑒 = 𝜎 [ + ]
𝜂 √3𝑑𝑏 (8)

The process model presented in Equation 8 is used to predict the maximum pressure

required to perform extrusion at given temperature and strain rate.

For metals, flow stress (𝜎) at a constant hot working temperature can be modeled

using Equation 9 which relates 𝜎 to strain rate (𝜀̇) and strain rate sensitivity (𝑚) [13].

𝜎 = 𝐶𝜀̇ 𝑚 (9)
33

The material constants 𝐶 and 𝑚 can be estimated from extrusion data of trials

performed at different strain rates. Strain rate 𝜀̇ in the above material model is the total

strain divided by the time required for the deformation to occur. This is the time average

of strain rate in the deformation zone of the die.

ln 𝑅
𝜀̇ =
𝑡𝑠 (10)

where 𝑅 is the extrusion ratio, and 𝑡𝑠 is the straining time.

Time for straining the material 𝑡𝑠 can be calculated using Equation 11.

𝑉𝑑𝑒𝑓
𝑡𝑠 =
𝑣𝑟 𝐴𝑐 (11)

where 𝑉𝑑𝑒𝑓 is the volume of deformation zone, 𝐴𝑐 is the cross section area of the

container, and 𝑣𝑟 is the ram speed. 𝑉𝑑𝑒𝑓 and 𝐴𝑐 are geometrical parameters of the conical

die and the container respectively.

The data obtained from the extrusion trials also includes ram displacement 𝑟𝑑 .

These data can be numerically differentiated to obtain ram speed (𝑣𝑟 ) using Equation 12,

which is also known as the midpoint difference formula.

𝑟𝑑(𝑖+1) − 𝑟𝑑(𝑖−1)
𝑣𝑟𝑖 =
𝑡𝑖+1 − 𝑡𝑖−1 (12)

Using Equations 10 – 12, strain rate (𝜀̇) can be calculated for use in Equation 9.
34

4.2 Parameters for the flow stress estimation

The important parameters for the calculation of flow stress are listed in Table 1.

Table 1 : Essential parameters for flow stress calculation

Billet diameter 𝑑𝑏 16.25 mm (0.6350 in)


Wire diameter 𝑑𝑤 2.05 mm (0.0808 in)
Semi die angle 𝛼 45o
Volume of deformation 𝑉𝑑𝑒𝑓 552.24 mm3 (0.0337 in3)
Total ideal strain 𝜀 4.12
Billet cross section area 𝐴𝑐 204.31 mm2 (0.3167 in2)

4.3 Compression testing

For more elaborate analysis of the flow properties of pure Mg, high temperature

compression tests were performed in this research. For most metals, strain rate

dependence of the flow stress varies with temperature [13]. Zener and Hollomon

developed a model that incorporates both strain rate and temperature dependence of the

flow stress [13]. This model is presented in Equation 1, where  is the flow stress of the

material and 𝑍 is the temperature compensated strain rate or Zener-Hollomon parameter.

Equation 13 is similar to Equation 9, where  is replaced by 𝑍, and 𝐶𝑧 and 𝑚𝑧 are the

material constants.

𝜎 = 𝐶𝑧 𝑍 𝑚𝑧 (13)
35

The parameter 𝑍 takes both strain rate and temperature into account. It can be

calculated using Equation 14, where 𝑄 is the activation energy, 𝑅 is the gas constant, and

𝑇 is the absolute temperature.

𝑄
𝑍 = 𝜀̇ exp ( ) (14)
𝑅𝑇

Zener and Hollomon used the Arrhenius rate equation to develop their flow stress

model [17]. They suggested that the plastic straining at elevated temperatures can be

quantified as a rate process. The Arrhenius rate equation is presented in Equation 15.

−𝑄
𝐾 = 𝐶𝑝 exp ( ) (15)
𝑅𝑇

where 𝐾 is the rate constant, 𝑅 is the universal gas constant, 𝑇 is the absolute

temperature, 𝑄 is the activation energy, and 𝐶𝑝 is the pre-exponential factor.

4.4 Friction ring testing

In a metal forming process, friction at the die-work piece interface plays an

important role in determining forming load, metal flow properties, surface quality and

final structure of the product [54]. Therefore, exact quantitative evaluation of friction is

important to analyze a metal forming process. Ring compression testing is a widely used

technique for this purpose. The method utilizes dimensional changes that occur due to

compression loading on the test material. The evaluation of friction is independent of

compression force and yield strength of the material. Figure 6 shows calibration curves

to determine coefficient of friction 𝜇 (Coulomb’s law), and friction factor 𝑚𝑓 (interface

frictional shear factor) associated with die-work piece interface.


36

a) b)

Figure 6 : Calibration curves to determine a) Coefficient of friction (𝜇), and b) friction

factor (𝑚𝑓 ) from ring compression testing [54]

Four ring compression tests were performed with pure Mg specimens. A photo of

the specimen before and after the test is shown in Figure 7. The specimens were 16 mm

(0.63″) in diameter (OD), 5.3 mm (0.21″) in height (H) with an internal diameter (ID) of

8 mm (0.32″). These specimens were cut to the recommended dimensional ratio of 6:3:2

(OD: ID: H) [54].


37

Figure 7 : a) The specimen prepared for testing is on the left with the thermocouple

attached to it. The sample on the right has been compressed at an elevated temperature.

The test setup included the MTS machine and Ameritherm 5 KW induction heater

with a temperature controller. A thermocouple was welded directly to the specimen to

maintain the process temperature via controller. The tests were performed with and

without the lubricant (boron nitride). In one study [50], Vaitkus found that boron nitride

was the most effective lubricant in high temperature ring testing of copper. The results

from ring tests performed on pure Mg at 300C and 450C are presented in Table 2. The µ

and 𝑚𝑓 were reduced considerably when a lubricant was used. These results were used in

compression test data analysis with Mg.


38

Table 2 : Coefficient of friction (µ) and friction factor (𝑚𝑓 ) for pure Mg at 300C and

450C

Change in internal Change in


Lubricant Temperature (C) µ 𝑚𝑓
diameter (%) height (%)
No lubricant 300 26.5 35.5 0.4 1
Sticking
No lubricant 450 47.3 44 0.57
friction*
Boron nitride 300 22.7 40.3 0.2 0.7
Boron nitride 450 37.7 44 0.4 1

*The value of 𝑚𝑓 varies from 0 (no friction) to 1 (sticking friction). For data points above

𝑚𝑓 = 1 curve, sticking friction is assumed.


39

CHAPTER 5 - PLAN OF WORK

To achieve the main goal of this research, namely to produce 12 AWG CNT-Mg

matrix composite wire, the following steps were set forth.

a) To design the process related tooling (3D modeling software was used and

components were manufactured by an outside vendor).

b) To analyze the process mathematically to project extrusion loads at each

processing step, with respect to temperature and ram speed.

c) To estimate the maximum extrusion force required to perform hot extrusion.

d) To determine the flow properties of the composite material from the process data

and to determine the effect of CNTs at each processing step.

e) To compare the CNT-Mg matrix composite flow properties to that of pure

wrought magnesium.

In order to compare properties of the CNT-Mg composite with magnesium at the

desired conditions, it was necessary to establish a material model for magnesium. The

Zener-Hollomon model is proposed to predict magnesium properties using compression

testing and data analysis in this research. Friction coefficients for this analysis were

obtained from ring compression testing. To verify the proposed Zener-Hollomon model

and to support its applicability, three extrusion tests were performed with magnesium

billets. Once the material model for pure magnesium was established, its flow properties

were compared to that of the CNT-Mg composite in hot extrusion.


40

5.1 Compression testing on pure magnesium

To define material properties of magnesium using the Zener-Hollomon model,

several compression tests were performed on pure Mg rods. The test plan is presented in

Table 3.

Table 3 : Compression testing plan for pure Mg

Test # Temperature (C) Strain Rate (s-1)


1 300 0.01
2 300 0.1
3 300 0.1
4 300 0.1
5 300 1
6 300 1
7 450 0.01
8 450 0.01
9 450 0.1
10 450 1
11 450 1

The MTS machine and Ameritherm 5 KW induction heater with a temperature

controller were used for the test setup (Figure 8). Tests 1-3 were used to establish the

uncertainty for compression testing using this setup. As a specimen is compressed, its

length deceases. Therefore, in order to keep a constant strain rate, the MTS program

Teststar-IIs was divided into discrete steps. For each 0.635 mm (0.025″) ram

displacement, ram speed was reduced accordingly to ensure a constant strain rate. The

speed for each step was calculated taking into account the remaining length of the

specimen.
41

Figure 8 : Photo of experimental setup for compression testing on pure Mg

The tests were performed by placing K-type thermocouple under the specimen.

This thermocouple was used to give a feedback signal to the controller that maintained a

constant process temperature. A second thermocouple was welded directly to the

specimen to take accurate temperature readings during compression.

The compression anvils and the specimen were sprayed with boron nitride for

lubrication. Once the steady temperature was reached (within 2C), the tests were started.

The flow stress of the material in compression was calculated using Equation 16. This

equation is obtained from slab analysis which is based on evaluating force balance on a

differential slab (a thin slice) of the material under deformation [13].

−1
ℎ𝑖 2 2𝜇𝑅𝑖 2𝜇𝑅𝑖
𝜎 = 2𝑃 {( ) [𝑒𝑥𝑝 ( )− ( ) − 1]}
𝜇𝑅𝑖 ℎ𝑖 ℎ𝑖 (16)
42

where 𝜎 is the flow stress, 𝑃 is the axial pressure, ℎ𝑖 and 𝑅𝑖 are the instantaneous height

and the instantaneous radius of the specimen respectively and µ is the coefficient of

friction. The µ values obtained by ring compression testing were used in the analysis.

5.2 Extrusion with pure magnesium billets

The extrusion trials performed with pure Mg billets (to validate the proposed

Zener-Hollomon model) are presented in Table 4.

Table 4 : Test plan for extrusion on pure Mg billets

Test # Temperature (C) Ram speed (mm/s) Ram force (N)


1 530 0.076 -
2 530 - 17,780
3 470 0.127 -

The deformation efficiency and friction factor for the analysis were selected such

that Equation 8 and the Zener-Hollomon model provided the best fit to experimental data.

These trials were used to verify the accuracy of the Zener-Hollomon model proposed in

this research for magnesium in hot extrusion. Experimental pressures for these trials were

compared with the pressure predicted by Equation 8 where flow stress 𝜎 for magnesium

was calculated using Zener-Hollomon model, to determine the accuracy.

5.3 Extrusion with the CNT-Mg composite billets

A series of extrusion trials were performed with the CNT-Mg composite billets to

produce wire with a refined microstructure. The data were used to analyze flow stresses
43

of the material. The list in Table 5 shows the plan for the hot extrusion trials with the

process parameters.

Table 5 : Hot extrusion test plan with CNT-Mg composite billets

Trial # Billet # Extrusion Material Temperature (C) Ram speed (mm/s)


1 1 First CNT-Mg 500 0.076
2 2 First CNT-Mg 500 0.076
3 3 First CNT-Mg 500 0.076
4 4 First CNT-Mg 500 0.076
5 1 Second CNT-Mg 500 0.076
6 2 Second CNT-Mg 500 0.076
7 3 Second CNT-Mg 500 0.076
8 2 Third CNT-Mg 550 0.076

Extrusion parameters (speed and temperature) for CNT-Mg billets were dictated

by the physical attributes that ICA funded researchers desired to achieve in the composite

material. As such, there were only a limited amount of trials and extrusion conditions that

could be considered with the composite material. The desired extrusion temperature was

500 – 550C. The CNT-Mg composite billets consisted of approximately 3 wt.% CNTs.

These 0.625" diameter billets were synthesized and processed by the external researchers

(ICA funded) using magnesium mixed with CNTs in a proprietary process. The extruded

wire was returned to the researchers for re-processing into new billets. The reason behind

re-processing was to develop optimum distribution and alignment of CNTs into the metal

matrix. Re-processing involved producing new billets from the extruded composite wire.
44

After extrusion of the initial composite billets, they were re-processed twice by external

researchers and re-extruded subsequently to promote the desired CNT distribution.

Ram force and displacement data were recorded for each extrusion trial at the

sampling rate of 10 Hz using the data acquisition system of MTS machine. Temperature

data were taken at 1 Hz using a separate data acquisition module.

The flow properties of the CNT-Mg composite were analyzed at each processing

step in this research. The effect of CNT additions to a magnesium matrix was analyzed

by comparing first extrusion data to pure Mg. The effect of change in CNT distribution

(due to reprocessing) on flow properties of the CNT-Mg composite was then analyzed

using re-extrusion data.


45

CHAPTER 6 - RESULTS AND DISCUSSION

6.1 Compression test results

The experimental data for compression testing of pure Mg at strain rates 0.01 s-1,

0.1 s-1 and 1 s-1 are shown in Figures 9 and 10 at temperatures 300 C and 450 C


respectively. True strain 𝜀 was calculated using 𝜀 = 𝑙𝑛 ℎ0 where ℎ0 is the initial height
𝑖

and ℎ𝑖 is the instantaneous height of the sample. Flow stress 𝜎 was estimated using

Equation 16.

Figure 9 : Flow stress data as a function of true strain for pure Mg at strain rates 0.01 s-1,

0.1 s-1 and 1 s-1 at 300C


46

Figure 10 : Flow stress data as a function of true strain for pure Mg at strain rates 0.01 s-1,

0.1 s-1 and 1 s-1 at 450C

The constant flow stress values for pure Mg representing varying strain rates and

temperatures are given in Table 6. An average flow stress was determined at points where

the slope of the curve is close to zero for each test.

Table 6 : Flow Stress Values of pure Mg at various conditions

Flow stress (MPa)


Strain Rate (s-1) 300C 450C
0.01 22.4 5.6
0.1 30.2(Avg.) 7.7
1 38.7 12.0
47

At temperatures above 200C, a dynamic recrystallization occurs in magnesium

[55]. This solid-state metallurgical reaction is a softening process that promotes a

constant flow stress with respect to strain, at a constant temperature and strain rate. Also,

at high temperatures, additional slip systems are activated in Mg crystal structure which

improves workability [55].

In a typical stress strain curve for pure Mg (Figure 11), a well-defined peak stress

is obtained. A strain hardening is followed by strain softening at the initial stage [55].

After this phase, equilibrium between the strain hardening rate and the softening rate is

achieved. A stress at this steady-state deformation is generally considered to characterize

flow properties of magnesium at given conditions.

Figure 11 : A typical stress strain curve for pure Mg at constant strain rate and

temperature
48

In the data presented in Figures 9 and 10, the ratio of strain at the peak stress

(𝜀𝑝𝑒𝑎𝑘 ) and the steady-state strain (𝜀𝑠𝑡𝑒𝑎𝑑𝑦 ) increased with increasing temperature but it

𝜀
decreased with increasing strain rate. The ratio 𝜀 𝑝𝑒𝑎𝑘 = 1 would mean that the steady
𝑠𝑡𝑒𝑎𝑑𝑦

state was achieved instantly. The ratio of the peak stress (𝜎𝑝𝑒𝑎𝑘 ) and steady-state

𝜎𝑝𝑒𝑎𝑘
(𝜎𝑠𝑡𝑒𝑎𝑑𝑦 ) stress decreased as temperature increases. However, the ratio increased
𝜎𝑠𝑡𝑒𝑎𝑑𝑦

with increasing strain rates. The 𝜀𝑠𝑡𝑒𝑎𝑑𝑦 and 𝜎𝑠𝑡𝑒𝑎𝑑𝑦 were considered at the point where

the slope of the curve is close to zero. From above analysis, it can be concluded that

relatively more time and energy are required to achieve steady-state flow stress in

compression at lower temperatures and higher strain rates.

The data for compression tests 1 – 3, at temperature 300C and strain rate 0.1 s-1,

are presented in Figure 12. These tests were performed to determine consistency and

experimental accuracy of compression testing. The average flow stress values for pure

Mg are listed in Table 7.


49

Figure 12 : Compression data for tests 1 – 3 at stain rate 0.1 s-1 and at 300C to evaluate

experimental accuracy

The flow stress of pure Mg was estimated with an uncertainty of ± 0.6 MPa, with

compression testing under same conditions in Figure 12. Using the statistical analysis, it

can be shown that this uncertainty corresponds to a confidence interval of 95%.

Table 7 : Average constant flow stress values for pure Mg at 300C and 0. 1 s-1 strain rate

Trial # Flow Stress (MPa) Flow Stress (psi)


1 30.3 4398.0
2 29.5 4280.2
3 30.7 4449.6
50

Figure 13 and 14 present compression data with pure Mg for two trials to show

temperature and ram speed variation during the entire course of the test. The trial

presented in Figure 13 was performed at a strain rate of 0.1 s-1 at 300C with pure Mg

billets. The strain rate and process temperature for the trial presented in Figure 14 was 1

s-1 and 450C respectively.

Figure 13 : Compression test data for trial with pure Mg at a strain rate of 0.1 s-1 at 300C
51

Figure 14 : Compression test data for trial with pure Mg at a strain rate of 1 s-1 at 450C

The graphs show that the tests were performed at fairly constant temperatures and

strain rates. The variation in temperature was within 10C throughout the tests. The strain

rate for the trial in Figure 13 was nearly constant. This was essentially the same for the

trial in Figure 14 with strain rate only slightly increasing toward the end of the test (to

about 1.1 s-1). This indicates the ability to maintain constant strain rate at high speeds.

Strain rate was within 8% of the nominal value and this would only correspond to about

2% variation in flow stress.

6.1.1 Strain rate dependence of the flow stress of pure Mg

The flow stress (𝜎) is expressed as a function of strain rate (𝜀̇) at constant

temperature in Equation 17 (an established equation), where 𝑚 is the strain rate

sensitivity and 𝐶 is the strength coefficient. Equation 18 is derived from Equation 17 that

can be used to determine 𝐶 and 𝑚 from experimental data.


52

(17)
𝜎 = 𝐶𝜀̇𝑚

ln 𝜎 = 𝑚 ln 𝜀̇ + ln 𝐶 (18)

Figure 15 : A plot of ln 𝜎 versus ln 𝜀̇ estimated from the compression data for pure Mg

In Figure 15, ln 𝜎 is plotted as a function of ln 𝜀̇. The y-intercept (= ln 𝐶) and

slope (= 𝑚) were determined by fitting a linear curve through the data points. The

material constants 𝐶 and 𝑚 for pure magnesium at temperatures 300C and 450C are

listed in Table 8. Results show that as temperature increases strain rate sensitivity

increases and the strength constant decreases for magnesium. This is a typical behavior

for most metals.


53

Table 8 : Material constants for pure magnesium at 300C and 450C

Temperature (C) Strength constant 𝐶 (MPa) / (psi) Strain rate sensitivity 𝑚


300 3.66 / 530.84 0.118
450 2.46 / 356.79 0.165

6.1.2 Strain rate and temperature dependence of the flow stress of pure Mg

A Zener-Hollomon model analysis to study strain rate and temperature

dependence of the flow stress (with one equation) is presented in this section. Equation

13 can be written as Equation 19. Using Equation 19, the strength coefficient 𝐶𝑧 and the

strain rate sensitivity 𝑚𝑧 for Zener-Holloman model can be obtained from an ln 𝜎 versus

ln 𝑍 plot.

ln 𝜎 = 𝑚𝑧 ln 𝑍 + ln 𝐶𝑧 (19)

However, in order to determine the Zener-Hollomon parameter (𝑍), the activation

energy (𝑄) needs to be determined first. Strain rate can be expressed as a function of

temperature at a constant flow stress as in Equation 20 that is derived from Equation 14.

A plot of ln 𝜀̇ versus 1/𝑇 can be used to determine activation energy from Equation 20.

𝑄 1
ln 𝜀̇ = − + ln 𝑍
𝑅 𝑇 (20)

The graph in Figure 16 illustrates the relation between 𝜎 and ln 𝜀̇ at temperatures

300C and 450C. From the graph, 5 MPa, 10 MPa, and 15MPa lines were identified as

constant flow stress lines and corresponding ln 𝜀̇ values were obtained at 300C and

450C using extrapolation.


54

Figure 16 : A plot of 𝜎 as a function of ln 𝜀̇

The ln 𝜀̇ data are plotted against the reciprocal of corresponding absolute

temperatures in Figure 17. The slopes are equal to −𝑄/𝑅 (from Equation 20), where 𝑅 =

8.314 𝐽/𝑚𝑜𝑙 𝐾 is the universal gas constant. An average of the slopes was used to

estimate 𝑄. The average activation energy (𝑄) for pure Mg was determined to be

159.5 𝐾𝐽/𝑚𝑜𝑙.
55

Figure 17 : A graph showing ln 𝜀̇ plotted against reciprocal of absolute temperature

(1/𝑇)

Equation 20 can be applied to calculate the values of ln 𝑍 corresponding to ln 𝜀̇.

Figure 18 shows the compression data where ln 𝜎 is plotted against ln 𝑍. A linear fit is

applied to the data points to obtain the strength coefficient (𝐶𝑧 ) and the strain rate

sensitivity (𝑚𝑧 ) for the Zener-Hollomon model. The constants 𝐶𝑧 and 𝑚𝑧 for pure Mg

were determined to be 0.11 MPa and 0.178, respectively. A coefficient of determination

for the linear fit is R2 = 0.98 .

The average (or equilibrium) flow stress values for tests 1to11 are presented in

Figure 19 along with the Zener-Hollomon model developed for pure Mg. The graph
56

shows that the prediction of the behavior of pure Mg in hot compression by the model

was fairly close with the maximum discrepancy of about 2.5 MPa (10%).

Figure 18 : A plot of ln 𝜎 as a function of ln 𝑍


57

Figure 19 : A graph showing comparison between compression data for tests 1 to 11 and

a Zener-Hollomon model

6.1.3 Validation of the Zener-Hollomon model

Extrusion test data to support the Zener-Hollomon model for pure Mg is presented

in Figures 20 and 21. Superimposed onto these plots are the curves for either ram

pressure or ram speed predicted by Equations 8 and 21 respectively. Equation 21 can be

derived by solving and rearranging Equation 8 for strain rate in terms of ram speed 𝑣𝑟 ,

where 𝜎 is replaced by the Zener-Hollomon model flow stress (Equations 13 and 14).

1

𝑉𝑑𝑒𝑓 −𝑄 𝐶 ln 𝑅 4𝑚𝑓 𝑙𝑏 𝑚
𝑣𝑟 = exp ( ) [ ( + )]
𝐴𝑐 ln 𝑅 𝑅𝑇 𝑃 𝜂 √3𝑑𝑏 (21)
58

Figure 20 shows data for the constant speed trial on pure Mg at 530C. The ram

speed was maintained at 0.076 mm/s. Friction factor (𝑚𝑓 ) was determined to be 0.5 from

the slope of Figure 1. The slope is equal to 4𝜎𝑚𝑓 /√3𝑑𝑏 , where 𝑑𝑏 is the container bore

diameter. Figure 26 shows data for the trial with a constant ram force of 17,780 N / 4000

lbs. The deformation efficiency was determined from constant force data. This method of

using constant force data to evaluate efficiency is expedient as it is calculated over a

range of ram speeds (and corresponding strain rates) [49]. Efficiency of the extrusion die

(𝜂) was determined to be 0.35 (or 35%) experimentally. The losses are attributed to

friction and redundant work. Equation 8 predicts extrusion pressure within 4% in Figure

20 (neglecting the data at the end where material flow changes).

Figure 20 : Extrusion data for pure Mg at a constant ram speed of 0.076 mm/s at 530C
59

Figure 21 : Extrusion data for pure Mg at a constant ram force of 17,780 N at 530C

Figure 21 shows erratic dips in ram speeds after 150 seconds. For the period of

150 – 200 seconds, ram speed was inconsistent. This most likely occurred because of the

die load-up due to resistance in the path of the extruded wire through the shielding-gas

tube. The distortion of wire due to the die load-up is shown in Figure 22.
60

Figure 22 : Photo of a distorted portion of wire due to die load-up

At temperatures greater than 400C, the strength of pure Mg is very low. Any

small resistance to the wire as it exits the die can cause a compressive stress great enough

to cause buckling and subsequent filling (or loading-up) of the die at the exit area. The

resistance may result in erratic dips in ram speeds during force controlled extrusion.

Equation 21 predicted ram speeds with a maximum discrepancy of about 15% (neglecting

data points in the die load-up region). This discrepancy can be attributed to the

uncertainties in evaluating friction factor and the data inconsistency due to resistance in

the path of extruded wire.

To check the applicability of the Zener-Hollomon model, a third trial was

performed on pure Mg with a different temperature and ram speed than first two trials.

The extrusion data are shown in Figure 23. The trial was performed with constant ram

speed of 0.127 mm/s at 470C. Equation 8 underestimated the ram pressure slightly but

within 6.5%. The data points near the end of extrusion do not follow the analytical model.

The ram pressure values showed some inconsistency initially from a displacement of

about 4 to 8 mm. This can also be attributed to loading of the die. The resistance leading
61

to the “die loading” was most likely due to the wire travel in the shielding-gas tube,

which forces the wire to make severe bend in order to keep the heated wire shielded with

argon gas (which is heavier than air).

Figure 23 : Extrusion data for pure Mg at 470C and at 0.127 mm/s ram speed

Nevertheless, this analysis supports the Zener-Hollomon model within range of

test temperatures and ram speeds/strain rates. To further support the proposed model, its

results were compared to experimental data from Sheng and Shivpuri [56] in Table 9.

These experimental data were obtained by performing tensile testing on magnesium at a

strain rate of 1.7 x 10−4 s −1 . From Table 9, it is evident that Zener-Hollomon model

predicted the flow stress of magnesium within 4 MPa. Therefore, this model may be
62

considered as a reasonably effective model to characterize the flow stress of Mg at even

lower temperatures and strain rates than that considered in this research.

Table 9 : Experimental flow stress values by Sheng and Shivpuri in tensile testing and

flow stress values predicted by Zener-Hollomon model for pure Mg

Flow Stress (MPa) by Flow stress (MPa) by


Temperature C
Sheng and Shivpuri ZH model
150 65 69
200 35 32
250 21 17

6.2 Extrusion results

A typical extrusion curve of ram pressure versus ram displacement plot is

presented in Figure 24. The region under the curve is divided into two parts. The bottom

part represents ideal and redundant work and the top part represents friction work in the

container.
63

Figure 24 : A typical extrusion pressure curve showing various components of total

extrusion work

From the graph, it is evident that the ideal and the redundant component of total

extrusion work are independent of ram displacement. On the other hand, friction work

decreases with displacement because it depends on the instantaneous billet length in

contact with the container. Hence, the curve declines gradually with negative slope. The

pressure curve presented in Figure 25 shows a similar trend. The trial presented in Figure

25 was performed on the CNT-Mg billet no. 1 at a constant ram speed of 0.076 mm/s at

500C. However, an increase in pressure was observed toward the end of extrusion,

indicating a change in material flow pattern as the dummy block/ram approached the die

entrance.
64

Figure 25 : Experimental data for first extrusion on the CNT-Mg billet 1 at 500C and at

constant ram speed of 0.076 mm/s

During extrusion, metal flow may not fully follow the 45° semi-die angle of the

die. The effective die angle and flow field that develops apparently extends into the

container. This is disrupted when the dummy block/ram is near the die entrance. This

instance is shown in Figure 26d, where ram starts pushing material from dead zone into

the die that results in the increase in ram pressure. An increase in the extrusion force

“noise” during the second half of the plot is possibly due to some back extrusion.
65

Figure 26 : a) A schematic of die and container showing semi-die angle 𝜶 and extrusion

pressure P, b) The formation of dead metal zone due to friction during extrusion. The

effective die angle 𝛼𝑒 is indicated. c) Metal flow in the middle of extrusion, d) An

instance where ram starts pushing dead metal into the deformation zone at the end of

extrusion

6.2.1 First extrusion of CNT-Mg composite billets

Figures 27 and 28 present data for first extrusion performed with 3 wt.% CNT-Mg

composite billets (no.’s 1 to 3). These trials were performed at constant ram speed of

0.076 mm/s (0.003 in/s) at 500C. The speed 0.076 mm/s translates to a strain rate equal

to 0.116 s-1 (Equations 10 and 11). Ram pressure is plotted against ram displacement in
66

Figure 27. The extrusion pressure for pure Mg predicted using the Zener-Hollomon

model and Equation 8 is also graphed for comparison. The comparison was used to

analyze the effect of CNTs in magnesium matrix.

Figure 27 : Pressure data for first extrusion on CNT-Mg billets 1 to 3 at a constant ram

speed of 0.076 mm/s at 500C superimposed with the prediction of extrusion pressure for

pure Mg using Equation 8 and Zener-Hollomon model at the same conditions


67

Figure 28 : Temperature data for first extrusion trials on CNT-Mg billets 1 to 3

As mentioned earlier, the pressure curve typically declines with the negative slope

during extrusion. The curves in Figure 27 show similar pattern declining gradually. From

the data presented in Figure 27, the extrusion pressure was within 5% (neglecting the

region at the end) for trials on billet 1 and 2. However, for billet 3, pressure was about

20% higher than for billet 1. This indicates that flow properties may vary significantly

from one billet to another. Equation 8 was used to estimate extrusion pressure for pure

Mg in Figure 27. The differences in extrusion pressures for pure Mg and the CNT-Mg

billets 1 and 2 are approximately 10% and 15% for respectively. In the case of billet 3,

the pressure is about 30% higher than pure Mg.


68

The temperature data at different locations in the apparatus for trials on billets 1 to

3 are shown in Figure 28. The temperature at the die entry may be considered as the most

representative temperature of the billet during extrusion. The temperature of the

deforming billet is not easy to measure. Hence, the die entry temperature was used for the

analysis. The temperature data indicate that these first extrusion trials were fairly

isothermal. The data show consistency in die entry temperatures for these trials. The die

temperatures were generally within 3C. At low extrusion speeds, a large portion of the

deformation/friction heat dissipates, thereby promoting near isothermal extrusion [14].

The die and the container control temperatures remained constant because thermocouples

were located directly adjacent to the heaters which were maintained accurately by solid

state relay system. During extrusion, the container wall temperature decreased as much as

8C for first extrusion trials. This can be attributed to the evacuation of the billet from the

respective region of the container as extrusion progressed.

6.2.2 Re-extrusion of CNT-Mg composite billets

CNT-Mg composite wire obtained from first set of extrusion trials was returned to

the external researchers to produce new billets. The second extrusion / re-extrusion was

performed on the new billets. These trials were performed at 0.076 mm/s ram speed and

at 500C (same conditions as the first extrusion). The ram pressure data are presented in

Figures 29 and 30. The extrusion pressure for new billets 1 and 2 increased by about 25%

and 28% respectively during re-extrusion. Temperature data for second extrusion trials is

shown in Figure 31. Similar to the first extrusion, the second extrusion trials were also
69

nearly isothermal. Die temperatures were almost constant throughout the process

averaging around 500C ± 2C.

Figure 29 : Experimental data for first and second extrusion on the CNT-Mg billet no. 1

at a constant ram speed of 0.076 mm/s at 500C


70

Figure 30 : Experimental data for first and second extrusion on CNT-Mg billet no. 2 at a

constant ram speed of 0.076 mm/s at 500C

Figure 31 : Temperature data for second extrusion trials on CNT-Mg billets 1 and 2
71

6.2.3 Third extrusion of CNT-Mg composite billet no. 2

The extruded wire from the second extrusion was returned to the external

researchers to produce a new billet. A third extrusion was performed on this new billet.

The Figure 32 and 33 show extrusion data for the third trial on the processed billet no. 2.

The ram speed was maintained constant at 0.076 mm/s. The third extrusion was

performed at 550C, a slightly higher temperature than previous trials. The external

researchers suggested performing third extrusion at the temperature closer to the melting

point of magnesium (650C) to facilitate more homogenous CNT distribution.

Figure 32 : Experimental data for the third extrusion on CNT-Mg billet no. 2
72

6.3 Extrusion results analysis and discussion

Extrusion pressure is a function of flow stress of the material. Hence, an increase

in pressure can directly correspond to an increase in flow stress of the work piece (at a

given ram speed and process temperature). Figure 27 shows that the processed material

with 3 wt.% CNTs effectively corresponded to an increase of 10 – 30% in the flow stress.

Extrusion of re-processed composite material also had a significant effect on the flow

properties. Flow stress increased by 25% and 28% overall due to second extrusion for

CNT-Mg billets 1 and 2 compared to first extrusion, respectively. These data are

presented in Figures 29 and 30. The third extrusion was performed on the CNT-Mg billet

no. 2 at a temperature 50 C higher than first and second extrusions. The flow stress was

reduced by as much as 15% during the third extrusion. This is most likely a combined

effect from a reduction in flow stress due to higher temperature, and further CNT re-

distribution. The formation of MgO due to reprocessing may also have affected the

change in flow stress.

Ram pressure at a given instant in extrusion represents that value required to

perform extrusion on the billet length at that instant. This value is also the total work

required per unit volume to extrude the material at that instant. Ram pressure values

required to extrude 1 inch (25.4 mm) billet lengths are presented in Table 10. The ram

pressure values of processed CNT-Mg composite are compared with that of pure

magnesium, predicted by Equation 8.


73

Table 10 : Maximum pressure required to extrude the CNT-Mg billets with 1 inch lengths

Maximum extrusion pressure (MPa)


CNT-Mg Ram speed First Second
Pure Mg % diff. % diff.
Billet # (mm/s) extrusion extrusion
1 0.076 82 92.5 12.8 123.5 51.6
2 0.076 82 96 17 126 53.6
3 0.076 82 102 24.4 138 68.3
4 0.076 82 100 22 - -

CNTs have excellent mechanical properties. Pure magnesium was strengthened

by adding CNTs and re-processing, consistent to what others have reported in the open

literature [25, 44, 46, 48]. CNT reinforcement to the Mg matrix structure increases flow

stress. This increase depends on factors such as distribution of CNTs within the matrix

and interface/contact between the CNTs and the metal matrix. The composite fabrication

process, CNT content, and the quality of CNTs (single walled, multi-walled etc.)

determine these factors in the bulk material.

It is evident that by processing magnesium with 3 wt.% multi-walled CNTs in an

Mg matrix using a proprietary technique, a 12 – 25% increase in apparent flow stress was

observed (see Table 10). MgO (which is 3 times denser than magnesium) can be

introduced in significant quantities during composite fabrication, particularly if

magnesium powder is used. This can also increase flow stress of the material and such

was the case in some initial trials of this research. The presence of MgO in the

magnesium matrix is detrimental to electrical conductivity because of its high resistance.

Proper fabrication methodology minimizes or even eliminates the formation of MgO. The

MgO content in the composite billets was not directly measured, however the densities of
74

CNT-Mg billets were equal to or lower than magnesium (Table 11). The density of pure

magnesium at 20C is 1.74 g/cm3.

Table 11 : Densities of the CNT-Mg composite billets before first, second and third

extrusion trials as received

Density g/cm3 before


Billet no. First extrusion Second extrusion Third extrusion
1a 1.699
1.710 -
1b 1.702
2a 1.713
1.709 1.72
2b 1.715
3a 1.708
1.710 -
3b 1.709
4a 1.708
- -
4b 1.710

CNTs have a natural tendency to form clusters causing poor distribution within a

metal matrix. Clustering of CNTs is not expected to increase the strength of the matrix

material to the same degree that well distributed CNTs would. The flow stress of CNT-

Mg composite increased by around 50 – 70% over pure Mg during the second extrusion

which is 30 – 35% higher than as received CNT-Mg composite. This appears to indicate

that CNT clusters were most likely “de-clustered’ and redistributed during extrusion and

re-extrusion and that CNTs were dispersed more homogeneously within the matrix. This

effectively increased the overall strength of the composite.

To assess the strength effect of MgO formation during Mg wire re-processing,

two additional extrusion trials were performed with pure magnesium. These trials
75

involved extrusion of a pure wrought magnesium rod (billet), and a magnesium billet that

was formed by re-pressing previously extruded (and oxidized wire). The wire was cut

into small pieces and pressed in a cylindrical die to form new re-pressed billets. The

average density of re-pressed billets was about 1.71 g/cm3. These trials were performed at

an extrusion ram speed of 0.127 mm/s and at 470C. Figures 33 and 34 show extrusion

data for the two trials.

Figure 33: Experimental data for the extrusion trials with pure magnesium billet and re-

pressed magnesium billet at a constant ram speed of 0.127 mm/s at 470C


76

Figure 34: Temperature data for the extrusion trials with pure magnesium and re-pressed

magnesium billet

The extrusion pressure for the pure Mg billet is compared to the pressure for the

re-pressed Mg wire in Figure 33. The data show that the pressure increased by only 2.5%

(average) due to re-processing (neglecting data at the end). This indicates that the MgO

formation during re-processing produced a minimal effect on Mg’s apparent flow stress.

Therefore, it can be concluded that much of the extrusion pressure increase of the re-

pressed CNT-Mg billets was due to refined distribution of CNTs in the magnesium

matrix. The MgO formation most likely resulted in a slight increase in the composite’s

apparent flow stress. The temperature data in Figure 34 show that the extrusion trials

were near isothermal. The die temperatures were almost constant throughout and within

3C.
77

The third and final extrusion trial was performed with re-pressed CNT-Mg billet

no. 2 (Figure 32). The apparent flow stress was decreased by as much as 15% during

third extrusion. This is attributed to the higher process temperature was higher (550C)

compared to earlier extrusions which were performed at 500C. Flow stress typically

decreases as temperature increases. In the case of pure magnesium, a 50C temperature

increase from 500C to 550C would result in 25% reduction in flow stress (at a strain

rate of 0.116 s-1). Hence, it is evident that a 15% decrease in flow stress was a combined

effect of a decrease due to higher temperature and an increase due to additional re-

processing of the composite material. However, the individual effect of these parameters

is very difficult to quantify exactly based on the data obtained in this research. By

performing more extrusion trials at different temperatures and strain rates, necessary data

could be obtained to analyze the effect of temperature on the flow properties of the CNT-

Mg composite.
78

CHAPTER 7 – SUMMARY AND CONCLUSIONS

In this research, a lab based process was developed to produce wire from the

CNT-Mg matrix composite billets via hot extrusion. An extrusion apparatus for use in a

250 kN MTS machine was designed with solid modeling software. Material parameters

for the classical Zener-Hollomon flow stress model were also developed for pure

magnesium using hot compression testing. The compression tests were performed on

pure magnesium at strain rates of 0.01 s-1, 0.1 s-1 and 1 s-1 and at 300 C and 450 C. From

the compression test data, the strength constant and strain rate sensitivity for the Zener-

Hollomon model for pure magnesium were determined to be 0.11 MPa and 0.178,

respectively. The proposed model correlated well to the compression test data with

coefficient of determination R2 = 0.98. For the analysis, the friction conditions between

the compression anvils and the magnesium specimen at 300C and 450C were evaluated

using ring compression testing. The analytical model was then used in analysis of hot

extrusion trials with pure magnesium. The extrusion model (which incorporated the

Zener-Hollomon model) predicted extrusion pressure within 6.5% and 4% at 470C and

530C, respectively. The maximum discrepancy in ram speed prediction from constant

force/variable velocity trial was 15%. The constant ram force trial was performed to

determine die efficiency in extrusion (𝜂 = 0.35). The friction factor (𝑚𝑓 = 0.5) in the

container was determined using constant ram speed extrusion.

Four hot extrusion trials were performed with the CNT-Mg composite billets to

produce wire. Extrusion was at 0.076 mm/s ram speed and at 500C. The composite

billets were fabricated with 3 wt.% multi-walled CNTs using a proprietary technique (by
79

ICA funded researchers). In order to promote a refined microstructure and homogenous

dispersion of CNTs, three re-extrusion trials were performed on the CNT-Mg composite

billets (produced from the wire obtained in the first extrusion). The extrusion data with

the CNT-Mg composite were then compared with the flow properties of pure Mg

predicted by the Zener-Hollomon model under the same conditions. The effect of adding

CNTs and (re-)processing on the flow properties of Mg was analyzed by comparing the

flow stress of the CNT-Mg composite in first set of extrusions to pure Mg. The flow

stress increased by 10 – 30% by processing 3 wt.% multi-walled CNTs in the Mg matrix.

CNTs acted as reinforcing fibers to strengthen the Mg matrix. The strength was further

increased by 25 – 28% overall during re-extrusion compared to as received CNT-Mg

composite. This indicates that CNTs may have dispersed more homogenously during re-

extrusion. A small amount of MgO formation inherent in re-processing may also have

contributed to this result.

By processing CNTs into the Mg matrix and by performing re-extrusion, the

mechanical strength of the wire was increased significantly over structurally weak pure

magnesium. Furthermore, CNTs clusters (potentially detrimental to the electrical

properties) within the matrix were probably fragmented and redistributed due to hot

working. The CNT-Mg composite wire prepared in this research could be considered as

one of the possible candidates for “future” current carrying conductors. However, further

research is necessary to analyze the properties of the CNT-Mg composite, such as

electrical and thermal properties, corrosion resistance, and ductility etc. and to explore its

potential as a possible candidate for “future” conductor.


80

For future work, several recommendations are provided. First, a material model

should be developed for the CNT-Mg composite to predict its plastic behavior similar to

the Zener-Hollomon model developed in this research. Second, more trials should be

performed with the CNT-Mg composite at different temperatures and strain rates to

analyze the dependence of flow stress (or flow stress sensitivity) on these process

parameters. The limited material available prevented this effort in this research. Lastly,

metallography and/or appropriate microscopy should be conducted on the CNT-Mg

composite samples to study the microstructure. Metallography, scanning electron

microscopy and EDS (Energy Dispersive Spectroscopy) analysis could be used to

analyze the CNT distribution, metallurgical changes, and grain structure of the resultant

composite due to CNT additions.


81

REFERENCES

[1] Iijima, S., 1991, "Helical Microtubules of Graphitic Carbon," Nature, 354(6348), pp.
56-58.
[2] Uddin, S. M., Mahmud, T., Wolf, C., Glanz, C., Kolaric, I., Volkmer, C., Höller, H.,
Wienecke, U., Roth, S., and Fecht, H., 2010, "Effect of Size and Shape of Metal Particles
to Improve Hardness and Electrical Properties of Carbon Nanotube Reinforced Copper
and Copper Alloy Composites," Composites Science and Technology, 70(16), pp. 2253-
2257.
[3] Hong, S., and Myung, S., 2007, "Nanotube Electronics: A Flexible Approach to
Mobility," Nature Nanotechnology, 2(4), pp. 207-208.
[4] Forró, L., Salvetat, J., and Bonard, J., 1999, "Electronic and Mechanical Properties of
Carbon Nanotubes," Proceedings of Nanotube, p. 99.
[5] Callister, W.D., and Rethwisch, D.G., 2013, "Fundamentals of Materials Science and
Engineering," John Wiley & Sons Inc., New York.
[6] Bakshi, S. R., Lahiri, D., and Agarwal, A., 2010, "Carbon Nanotube Reinforced Metal
Matrix Composites - a Review," International Materials Reviews, 55(1), pp. 41-64.
[7] Joseph, G., and Kundig, K.J., 1999, "Copper: Its Trade, Manufacture, Use, and
Environmental Status," ASM International, Ohio.
[8] Hammond, C. R., 2004, “Handbook of Chemistry and Physics, 81st edition”, CRC
press, Boca Raton, FL.
[9] Energy information and administration (EIA), 2012, from
http://www.eia.gov/tools/faqs/faq.cfm?id=105&t=3
[10] Hjortstam, O., Isberg, P., and Söderholm, S., 2004, "Can We Achieve Ultra-Low
Resistivity in Carbon Nanotube-Based Metal Composites?" Applied Physics A: Materials
Science & Processing, 78(8), pp. 1175-1179.
[11] Oberg, E. J., Franklin D. H., Holbrook L. R. and Henry H. H., 2000, "Machinery's
Handbook (26th ed.)," Industrial Press, New York.
82

[12] Society of Manufacturing Engineers, Dallas, D. B., and American Society of Tool
and Manufacturing Engineers, 1976, "Tool and Manufacturing Engineers Handbook: A
Reference Work for Manufacturing Engineers," McGraw-Hill, New York.
[13] Hosford, W.F., and Caddell, R.M., 2011, "Metal forming: mechanics and
metallurgy", Cambridge University Press, New York.
[14] Kraft, F. F., and Gunasekera, J. S., 2005, "Conventional Hot Extrusion," ASM
Handbook, Metalworking: Bulk Forming, 14, pp. 421-439.
[15] Chen, W.X., Tu, J.P., Wang, L.Y., Gan, H. Y., Xu, Z. D., and Zhang, X. B., 2003,
"Tribological Application of Carbon Nanotubes in a Metal-Based Composite Coating and
Composites," Carbon, 41(2), pp. 215-222.
[16] Tu, J. P., Yang, Y. Z., Wang, L. Y., Ma, X. C., and Zhang, X. B., 2001,
"Tribological Properties of Carbon-Nanotube-Reinforced Copper Composites,"
Tribology Letters, 10(4), pp. 225-228.
[17] He, C. N., Zhao, N., Shi, C., Du, X., Li, J., Li, H., and Cui, Q., 2007, "An Approach
to Obtaining Homogeneously Dispersed Carbon Nanotubes in Al Powders for Preparing
Reinforced Al-Matrix Composites," Advanced Materials, 19(8), pp. 1128-1132.
[18] Hausner, H. H., 1973, "Handbook of Powder Metallurgy," Chemical Publishing Co.
Inc., New York.
[19] Martin, C., Bouvard, D., and Shima, S., 2003, "Study of Particle Rearrangement
during Powder Compaction by the Discrete Element Method," Journal of the Mechanics
and Physics of Solids, 51(4), pp. 667-693.
[20] Green, R., 1972, "A Plasticity Theory for Porous Solids," International Journal of
Mechanical Sciences, 14(4), pp. 215-224.
[21] Gurson, A. L., 1975, "Continuum Theory of Ductile Rupture by Void Nucleation
and Growth. Part I. Yield Criteria and Flow Rules for Porous Ductile Media," Journal of
Engineering Materials and Technology, 99, pp. 2-15.
[22] Kuhn, H., and Downey, C., 1971, "Deformation Characteristics and Plasticity
Theory of Sintered Powder Materials," International Journal of Powder Metallurgy, 7, pp.
15-25.
83

[23] Shima, S., and Oyane, M., 1976, "Plasticity Theory for Porous Metals," International
Journal of Mechanical Sciences, 18(6), pp. 285-291.
[24] Ryabicheva, L., and Usatyuk, D., 2008, "Modeling of the Production Technology of
High-Dense Copper from a Porous Fibrous Billet," Transactions- Powder Metallurgy
Association of India Hyderabad, 34, pp. 34-37.
[25] Goh, C.S., Wei, J., Lee, L.C., and Gupta, M., 2006, "Simultaneous Enhancement in
Strength and Ductility by Reinforcing Magnesium with Carbon Nanotubes," Materials
Science & Engineering: A, 423(1), pp. 153-156.
[26] Bian, Z., Pan, M. X., Zhang, Y., and Wang, W. H., 2002, "Carbon-Nanotube-
Reinforced Zr52.5Cu17.9Ni14.6Al10Ti5 Bulk Metallic Glass Composites," Applied Physics
Letters, 81(25), pp. 4739-4741.
[27] Zhou, S. M., Zhang, X. B., Ding, Z. P., Min, C. Y., Xu, G. L., and Zhu, W. M.,
2007, "Fabrication and Tribological Properties of Carbon Nanotubes Reinforced Al
Composites Prepared by Pressureless Infiltration Technique," Composites Part A, 38(2),
pp. 301-306.
[28] Laha, T., Agarwal, A., McKechnie, T., and Seal, S., 2004, "Synthesis and
Characterization of Plasma Spray Formed Carbon Nanotube Reinforced Aluminum
Composite," Materials Science & Engineering: A, 381(1), pp. 249-258.
[29] Bakshi, S. R., Singh, V., Seal, S., and Agarwal, A., 2009, "Aluminum Composite
Reinforced with Multiwalled Carbon Nanotubes from Plasma Spraying of Spray Dried
Powders," Surface & Coatings Technology, 203(10), pp. 1544-1554.
[30] Cha, S. I., Kim, K. T., Arshad, S. N., Mo, C. B., and Hong, S. H., 2005,
"Extraordinary Strengthening Effect of Carbon Nanotubes in Metal-Matrix
Nanocomposites Processed by Molecular-Level Mixing," Advanced Materials, 17(11),
pp. 1377-1381.
[31] Kyung, T. K., Cha, S. I., and Soon, H. H., 2007, "Hardness and Wear Resistance of
Carbon Nanotube Reinforced Cu Matrix Nanocomposites," Materials Science &
Engineering: A, 449, pp. 46-50.
84

[32] Zhang, Y., Zhang, Q., Li, Y., Wang, N., and Zhu, J., 2000, "Coating of Carbon
Nanotubes with Tungsten by Physical Vapor Deposition," Solid State Communications,
115(1), pp. 51-55.
[33] Shu, J., Li, H., and Yang, R., 2006, "Cage-Like Carbon nanotubes/Si Composite as
Anode Material for Lithium Ion Batteries," Electrochemistry Communications, 8(1), pp.
51-54.
[34] Morisada, Y., Fujii, H., Nagaoka, T., and Fukusumi, M., 2006, "MWCNTs/AZ31
Surface Composites Fabricated by Friction Stir Processing," Materials Science and
Engineering: A, 419(1), pp. 344-348.
[35] Kwon, H., Estili, M., Takagi, K., Miyazaki, T., and Kawasaki, A., 2009,
"Combination of Hot Extrusion and Spark Plasma Sintering for Producing Carbon
Nanotube Reinforced Aluminum Matrix Composites," Carbon, 47(3), pp. 570-577.
[36] Noguchi, T., Magario, A., Fukazawa, S., Shimizu, S., Beppu, J., and Seki, M., 2004,
"Carbon Nanotube/Aluminum Composites with Uniform Dispersion," Materials
Transactions, 45(2), pp. 602-604.
[37] Cox, H. L., 1952, "The Elasticity and Strength of Paper and Other Fibrous
Materials," British Journal of Applied Physics, 3(3), pp. 72-79.
[38] Fukuda, H., and Chou, T. W., 1981, "Advanced Shear-Lag Model Applicable to
Discontinuous Fiber Composites," Journal of Composite Materials, 15(1), pp. 79-91.
[39] Clyne, T. W., 1990, "A Simple Development of the Shear Lag Theory Appropriate
for Composites with a Relatively Small Modulus Mismatch," Materials Science and
Engineering: A, 122(2), pp. 183-192.
[40] Ryu, H. J., Cha, S. I., and Hong, S. H., 2004, "Generalized Shear-Lag Model for
Load Transfer in SiC/Al Metal-Matrix Composites," Journal of Materials Research,
18(12), pp. 2851-2858.
[41] Yang, Y. L., Wang ,Y. D., Ren, Y., He, C. S., Deng, J. N., Nan, J., Chen, J. G., and
Zuo, L., 2008, "Single-Walled Carbon Nanotube-Reinforced Copper Composite Coatings
Prepared by Electrodeposition Under Ultrasonic Field," Materials Letters, 62(1), pp. 47-
50.
85

[42] Ngo, Q., Cruden, B. A., Cassell, A. M., Walker, M. D., Ye, Q., Koehne, J. E.,
Meyyappan, M., Li J, and Yang, C. Y., 2004, "Thermal Conductivity of Carbon
Nanotube Composite Films," Materials Research Society Symposium Proceedings,
812(1), pp. 179-184.
[43] Xu, C. L., Wei, B. Q., Ma, R. Z., Liang, J., Ma, X. K., and Wu, D. H., 1999,
"Fabrication of aluminum–carbon Nanotube Composites and their Electrical Properties,"
Carbon, 37(5), pp. 855-858.
[44] Goh, C. S., Wei, J., Lee, L., and Gupta, M., 2006, "Development of Novel Carbon
Nanotube Reinforced Magnesium Nanocomposites Using the Powder Metallurgy
technique," Nanotechnology, 17(1), pp. 7-12.
[45] Shimizu, Y., Miki, S., Soga, T., Itoh, I., Todoroki, H., Hosono, T., Sakaki, K.,
Hayashi, T., Kim, Y. A., Endo, M., Morimotob, S. and Koide, A., 2008, "Multi-walled
Carbon Nanotube-Reinforced Magnesium Alloy Composites," Scripta Materialia, 58(4),
pp. 267-270.
[46] Carreño-Morelli, E., Yang, J., Couteau, E., Hernadi, K., Seo, J. W., Bonjour, C.,
Forro, L., and Schaller, R., 2004, "Carbon Nanotube/Magnesium Composites, " Physica
Status Solidi(a), 201(8), pp. 53-55.
[47] Aung, N. N., Zhou, W., Goh, C. S., Nai, S. M. L., and Wei, J., 2010, "Effect of
Carbon Nanotubes on Corrosion of Mg–CNT Composites," Corrosion Science, 52(5), pp.
1551-1553.
[48] Goh, C. S., Wei, J., Lee, L. C., and Gupta, M., 2008, "Ductility Improvement and
Fatigue Studies in Mg-CNT Nanocomposites," Composites Science and
Technology, 68(6), pp. 1432-1439.
[49] Barkley, B., 2013, "Analysis of hot Isothermal Copper Extrusion for Multi-Channel
Profiles," M.S. thesis, Dept. of Mech. Eng., Ohio University, Athens, OH.
[50] Vaitkus, V., 2008, "A Process for the Direct Hot Extrusion of Hollow Copper
Profiles," M.S. thesis, Dept. of Mech. Eng., Ohio University, Athens, OH.
[51] ATI Allvac. "ATI 720 alloy ". (n.d.). from
http://www.alleghenytechnologies.com/aerospace/docs/aerospacedatasheets/ATI720
Nickel.pdf
86

[52] Pradip, K.S., 2000, "Aluminum Extrusion Technology," ASM International, Ohio.
[53] Kraft, F., and Kochis, J., 2013, "Hot Extrusion of Thin-walled Multi-channel Copper
Profiles," Journal of Mechanical Science and Technology (Transactions of ASME), 135,
pp. 1-9.
[54] Male, A. T., and Cockcroft M. G., 1964. "A method for the determination of the
coefficient of friction of metals under conditions of bulk plastic deformation," Journal of
the Institute of Metals, 93, pp. 38-46.
[55] Sitdikov, O., and Kaibyshev, R., 2001, "Dynamic Recrystallization in Pure
Magnesium," The Japan Institute of metals, 42(9), pp. 1928-1937.
[56] Sheng, Z. Q., and Shivpuri, R. 2006, "Modeling Flow Stress of Magnesium Alloys at
Elevated Temperature," Materials Science and Engineering: A, 419(1), pp. 202-208.
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!

!
!
Thesis and Dissertation Services

You might also like