University of Southern California Dissertations and Theses

Mechanical behavior and microstructure optimization of ultrafine-grained aluminum alloys and nanocomposites

(USC Thesis Other)

Mechanical behavior and microstructure optimization of ultrafine-grained aluminum alloys and nanocomposites

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Mechanical Behavior and Microstructure
Optimization of Ultrafine-grained Aluminum
Alloys and Nanocomposites

By
YUZHENG ZHANG

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)

August 2015

Copyright 2015 Yuzheng Zhang
Acknowledgements

With my utmost gratitude, I would like to thank my advisor and mentor, Professor Steven Nutt,
for his constant support and trust as his student. I am very grateful that Prof. Nutt encouraged me
to pursue a PhD degree and welcomed me into his research group. I greatly admire his
knowledge and vision in a broad range of fields. Prof. Nutt’s office door is always open for me
when I seek guidance and suggestion in both work and life. My PhD would not be accomplished
without the research environment and people of the group he leads.

I would also like to express my gratitude to Byungmin Ahn (a former post doc fellow in the
group) for unselfishly sharing all his knowledge with me. This dissertation would not be possible
if not for his professional training and discussion. My congratulations go to Byungmin for
becoming a professor and starting a new life with his family in South Korea.

I gratefully appreciate the support from John Curulli and Matthew Mecklenburg of the center for
electron microscopy and microanalysis (CEMMA) at USC. With their trainings and supports, I
quickly developed the microscopy knowledge and skills which are beneficial to my research
study and career. I spent a great time working with them at CEMMA.

I would like to acknowledge the collaborators at the University of California at Davis: Dr. Troy
Topping, Hanry Yang, Dr. Tao Hu, Prof. Lavernia, and Prof. Schoenung for the valuable
suggestions and supports in my research study. I would also like to thank Prof. Langdon and Dr.
Shima Sabbaghian for helping me use the high pressure torsion instrument. My thanks also go to
Prof. Michael Kassner and Prof. Andrea Hodge for their time and effort to guide me through the
dissertation and defense. My acknowledgements extend to all the people in the research group
for their companies during my PhD life, especially the lab managers, Ezra Pryor, Rohan Panikar
and Yunpeng Zhang, who keep the lab running.

Finally, I would like to express my sincere gratitude and love to my whole family especially to
my parents and my wife, Renjie. I am very lucky and grateful to have your endless love and
support throughout my life.

Table of Contents

Abstract 1

Chapter 1
Introduction and review of nanocrystalline and ultrafine-grained materials

2
1.1 What are nanocrystalline and ultrafine-grained materials? 2
1.2 Synthesis of nanocrystalline and ultrafine-grained materials 7
1.2.1 An overview of the synthesis methods 7
1.2.2 Cryomilling 15
1.2.3 High pressure torsion 22
1.3 Enhancing strength 26
1.3.1 Dislocation 26
1.3.2 Grain boundary strengthening 27
1.3.3 Precipitate and dispersoid strengthening 32
1.3.4 Solid solution strengthening 33
1.3.5 Work hardening 33
1.3.6 Extrinsic reinforcing 34
1.4 Paradox of strength and ductility 35
References 49

Chapter 2
Experimental techniques
56
2.1 Characterization 56
2.1.1 Scanning electron microscope 56
2.1.2 Electron backscatter diffraction 59
2.1.3 Transmitted Kikuchi diffraction 62
2.1.4 Energy dispersive X-ray spectroscopy 65
2.1.5 Transmission electron microscope 68
2.1.6 Advanced sample preparation 71
2.2 Mechanical testing 76
2.2.1 Vickers microhardness measurement 76
2.2.2 Instrumented nanoindentation 79
2.2.3 In situ micro-tensile testing 81
2.2.4 Digital image correlation 85
References 89

Chapter 3
Dynamic micro-strain analysis of cryomilled Al alloys using digital image
correlation
91
3.1 Motivation 91
3.2 Experimental procedures 94
3.2.1 Specimen preparation 94
3.2.2 Alignment of EBSD scanned area to DIC strain field 96
3.2.3 E-beam lithography (EBL) patterning 98
3.2.4 In situ micro-tensile testing 99
3.2.5 In-plane micro-strain analysis 101
3.3 Results and discussions 101
3.3.1 EBL random speckle pattern for DIC 101
3.3.2 Intrinsic image distortion correction 103
3.3.3 Micro-strain measurement of ultrafine-grained Al-Mg alloy 105
3.4 Conclusions 111
References 112

Chapter 4
Microstructural evolution of trimodal Al alloys
114
4.1 Motivation 114
4.2 Experimental procedures 114
4.3 Results and discussions 115
References 117

Chapter 5
Micro-strain evolution and toughening mechanisms in trimodal AA 5083
118
5.1 Motivation 119
5.2 Experimental procedures 122
5.2.1 Material Synthesis 122
5.2.2 Characterization 123
5.2.3 Mechanical testing 124
5.3 Results and discussion 125
5.3.1 Microstructure 125
5.3.2 In situ tensile testing 130
5.3.3 Micro-strain measurement 132
5.3.4 Toughening mechanisms 134
5.3.5 Fracture analysis 137
5.4 Conclusions 138
References 140

Chapter 6
Two-step SPD processing of a trimodal Al-based nano-composite
144
6.1 Motivation 144
6.2 Experimental procedures 148
6.2.1 Material synthesis 148
6.2.2 Microstructure 149
6.2.3 Mechanical testing 150
6.3 Results and discussion 151
6.3.1. Microstructure 151
6.3.2 Microstructure homogeneity 158
6.3.3 Annealing 159
6.3.4 Tension 162
6.3.5 Fracture mechanism 164
6.4 Conclusions 166
References 168

Chapter 7
Historic aspects and future challenges
172
References 180

Chapter 8
Infusion quality and elution effect of functional light weight ceramic proppants
181
8.1 Motivation 182
8.2 Experimental procedures 185
8.2.1 Single pellet compressive test 185
8.2.2 Characterization 185
8.2.3 Elution 187
8.3 Results and discussions 188
8.3.1 Pellet morphology and load to crush a single pellet 188
8.3.2 Pore statistics 189
8.3.3 Infusion quality 191
8.3.4 Elution effect 193
8.4 Conclusions 196
References 197

List of Tables 198
List of Figures 199

Abstract

Nanocrystalline (NC) or ultrafine-grained (UFG) materials exhibit significantly improved
strength and hardness compared to the coarse grain (CG) counterpart according to the
well-known Hall-Petch effect. In this work, aluminum alloy (AA) 5083 was selected as
the target material due to its low density, good weldability and excellent corrosion
resistance. Refined grain size was achieved by severe plastic deformation techniques
including cryomilling and high pressure torsion (HPT) which both introduced large
amount of plastic strain into the materials and thus refined grain sizes down to
nano-scale.

Despite enhanced strength and hardness, the NC or UFG AA 5083 is very brittle due to
the lack of plasticity mechanisms. This brittleness can cause numerous risks and failures
during material service or even during production. The major challenge of this work is to
restore ductility of the NC or UFG AA 5083 while maintaining its high strength.
Toughening approaches described in this work follow the materials science paradigm
which says “microstructure connects fabrication with properties”. The mechanical
properties of the end products were tailored by altering the microstructure via different
processing routes. We seek an optimized microstructure which can balance the strength
with the ductility and maximize the fracture toughness of the NC/UFG AA 5083.

1

Chapter 1 will discuss the general background, modern synthesis methods, strengthening
mechanisms and toughening efforts of the NC/UFG materials. Chapter 2 will introduce
the methods and experimental procedures for microstructure characterization and
deformation mechanism investigation. Chapter 3 will describe an in situ micro-strain
measurement of an UFG Al-Mg alloy at sub-micron scale. Chapter 4 will focus on the
microstructural evolution of the cryomilled Al/B
4
C powder during processing. Chapter 5
will investigate the failure mechanisms and the toughening effect of the ductile regions in
a trimodal AA 5083 nanocomposite. Chapter 6 will demonstrate the attempt to enhance
the toughness by optimizing the morphology and spacing of the ductile regions in the
trimodal nanocomposite. At last, chapter 7 will discuss some historic aspects of this work
and the challenges that are lying ahead.

2

Chapter 1
Introduction and review of nanocrystalline and
ultrafine-grained materials

1.1 What are nanocrystalline and ultrafine-grained materials?
It is necessary to clarify some nomenclatures before we start this chapter. The definition
of nanocrystalline (NC) materials is any solid crystals that consist of grains less than 100
nm in size. Similarly, crystals with grain size from 100 nm to 1 um are called
ultrafine-grained (UFG) materials. To be more specific, the grains mentioned above refers
to grains separated by high angle grain boundaries (misorientation between neighboring
grains > 15
o
). NC and UFG materials have raised tremendous research interests due to
their unique microstructures as well as improved physical and mechanical properties
[1-5]. The outstanding properties of NC materials were summarized in an early review
article by Gleiter [1]. Since then, a large amount of studies and review articles have been
published to demonstrate the enhanced mechanical properties and new fabrication
methods of NC and UFG materials were discussed. Several high-impact review articles in
this field were listed in Table 1.1.

3

Table 1.1 List of review articles contributing to the community of nanocrystalline and
ultrafine-grained materials
Year Author Title Journal
1989 H. Gleiter Nanocrystalline materials Prog. Mater. Sci.
1989 R. Birringer Nanocrystalline materials Mater. Sci. Eng. A
1992 H. Gleiter Materials with ultrafine microstructure:
retrospective and prospective
Nanostruct. Mater.
1995 C.
Suryanarayana
Nanocrystalline materials : a critical
review
Intl. Mat. R.
1999 J. R. Weertman Structure and mechanical behavior of
bulk nanocrystalline materials
MRS Bull.
2000 C.
Suryanarayana
Nanocrystalline materials – current
research and future directions.
Hyperfine Interact.
2000 R. Z. Valiev Bulk nanostructured materials from
severe plastic deformation
Prog. Mater. Sci.
2000 H. Gleiter Nanostructured materials: basic concepts
and microstructure
Acta Mater.
2001 C.
Suryanarayana
Mechanical alloying and milling Prog. Mater. Sci.
2003 K. S. Kumar Mechanical behavior of nanocrystalline
metals and alloys
Acta Mater.
2005 D. Wolf Deformation of nanocrystalline materials
by molecular-dynamics simulation
Acta mater.
2006 M. A. Meyers Mechanical properties of nanocrystalline
materials
Prog. Mater. Sci.
2006 D. B. Witkin Synthesis and mechanical behavior of
nanostructured materials via cryomilling
Prog. Mater. Sci.
2008 A. P. Zhilyaev Using high-pressure torsion for metal
processing: Fundamentals and
applications
Prog. Mater. Sci.

4

As first described by Hall [6] and Petch [7], the small grain size results in material
strengthening owing to an increased volume fraction of grain boundary area inside the
material. This increasing strength with decreasing grain size is formulated by the famous
Hall-Petch equation (Equation 1).
𝜎𝜎 𝑦𝑦 = 𝜎𝜎 0
+ 𝑘𝑘 ∙ 𝑑𝑑 − 1/ 2
(Eq. 1)
Where 𝜎𝜎 𝑦𝑦 is the yield strength, 𝜎𝜎 0
is the strength needed to initiate plastic deformation
in the material with infinitely large grain size, 𝑘𝑘 is the Hall-Petch coefficient related to
material itself and 𝑑𝑑 is the average grain size of the material. From the Hall-Petch
equation, we can see that reduced grain size increases the yield strength of one solid
crystal without changing its chemical composition. (Strengthening mechanism will be
discussed in section 1.3.) Since then, many alloy metals and systems were studied to
verify the Hall-Petch relationship. Subsequent studies discovered that this relationship is
only valid for a certain range of grain size since intuitively one can not achieve
indefinitely large yield strength when grain size is indefinitely small. As researchers push
the grain size limit to ~ 10 nm, the “yield strength vs. grain size” relation deviates from
the Hall-Petch equation [3]. Interestingly, the materials get softer after the grain size is
reduced below a critical value (~ 10 nm). This phenomenon is called the inverse
Hall-Petch effect or the Hall-Petch breakdown [8]. Mathematically, we can treat this
phenomenon as the changes of the Hall-Petch coefficient 𝑘𝑘 at different grain size range.
A typical plot describing the Hall-Petch effect is shown in Figure 1.1. The Hall-Petch
coefficient 𝑘𝑘 (the slope) decreases as grain size is reduced below 100 nm. In the inverse
5

Hall-Petch region (grain size < 10 nm), we can think of a negative 𝑘𝑘 value associated
with the equation. The fact of this changing 𝑘𝑘 value implies changing deformation
mechanisms as we tune the grain size. The origin of the Hall-Petch effect and the
Hall-Petch breakdown will be discussed in depth in section 1.3.2.

Fig. 1.1 Schematic plot of the Hall-Petch relation and the Hall-Petch breakdown as a
function of grain size [3]

6

1.2 Synthesis of nanocrystalline and ultrafine-grained
materials
1.2.1 An overview of the synthesis methods
Due to excellent properties and emerging applications of NC and UFG materials,
tremendous efforts have been proposed to fabricate them in an effective and economical
way. Over the past three decades, a wide variety of synthesis techniques were developed
and studied to achieve NC and UFG materials in thin film and bulk form. All of these
synthesis methods generally belong to two categories – the “top-down” and the
“bottom-up” approach. As shown in Figure 1.2, the top-down approach starts with
macroscopic feature size and breaks it to nano regime. In contrast, the bottom-up
approach builds nanostructures from assembly of individual molecules or atoms. The
top-down method includes several severe plastic deformation (SPD) techniques such as
high pressure torsion (HPT), equal channel angular pressing (ECAP), mechanical
alloying (MA) and accumulative roll bonding (ARB). Examples of the bottom-up
methods are inert gas condensation (IGC), electrodeposition, physical vapor deposition
(PVD) and chemical vapor deposition (CVD). The bottom-up approaches are popular in
microelectronics and coating industry where crystal structure is essential, but not
thickness. In contrast, the top-down methods are good at producing NC materials in bulk
form which have major applications in mechanical engineering. These synthesis methods
are summarized in Table 1.2. In this section, several popular synthesis methods will be
briefly discussed to give you an overview of synthesis methods of NC and UFG
7

materials.

Fig. 1.2 Schematic illustration of top-down and bottom-up approaches

8

Table 1.2 List of popular synthesis methods for nanostructured materials using top-down
and bottom-up approaches
Top-down Bottom-up
Severe plastic deformation (SPD):
• High pressure torsion (HPT)
• Equal channel angular pressing
(ECAP)
• Mechanical alloying (MA)
• Accumulative roll bonding(ARB)
• Inert gas condensation (IGC)
• Electrodeposition
• Physical vapor deposition
(PVD)
• Chemical vapor deposition
(CVD)

Bottom-up approach
Inert gas condensation
As a bottom-up method, inert gas condensation (IGC) [4,9] can be used to make
nanocrystals with size less than 10 nm of metals or metal oxides with low melting point
such as Ag [10], Cu, Ni and SnO
2
[11]. The applications of these nanoparticles extend to
fields of electronics, optics, catalysis and magnetic data storage [12]. IGC is a physical
process during which the source metal is heated, vaporized in a vacuum chamber (~ 10
-4

mbar). The metal vapor is then cooled by ultra-pure inert gas (e.g. Ar and He) flowing
with a typical pressure of 10 – 50 mbar. As a result of collision between metal atoms and
inert gas molecules in chamber, the metals atoms condense and form small crystals in
nanosize. Eventually, the particles can be collected on a suitable substrate such as a metal
filter or a liquid nitrogen cold finger. The crystal structure and particle size of the end
product highly depend on collision frequency and nucleation rate of nanoparticles [11].

9

Electrodeposition
Electrodeposition is another widely used bottom-up approach which can make
nanocrystalline thin films [5, 13-15]. Electrodeposition is basically a reverse process of a
galvanic cell. During deposition, anode metal is oxidized and dissolved in the electrolyte.
This anode metal is then re-deposited on the cathode forming a layer of NC thin film.
Although this technique is able to produce super small grain size of 20 - 40 nm [15], the
final product is only in form of thin films which limit it to coating application. Another
generic issue associated with electrodeposition is the porosity remained in the product
which can be detrimental to its mechanical behavior [15].

Physical vapor deposition
Physical vapor deposition (PVD) is another general coating technique which is capable of
making nanostructured thin films. PVD has a wide range of applications in
microelectronics, hard coating industry and aerospace, etc. PVD is a general term for
many dry deposition techniques which involve pure physical processes. PVD always
involves atom “knock-out” from the source materials either by thermal energy, high
electric field, electron, plasma or laser, etc. and redeposition on the target substrate.
Variants of PVD include laser assisted PVD, electron beam assisted PVD and cathodic
arc PVD, etc. The resulting coating usually has higher hardness and better adhesion
strength than the one made using electrodeposition. One example of the PVD
applications is to synthesize highly nanotwinned Cu foil using magnetron sputtering [16].
10

During magnetron sputtering, inert gas (e.g. Argon) is ionized forming high density
plasma near the source metal. The high energy gas ions are accelerated toward the source
metal by high voltage and bombard the source atoms out of the surface. The sputtered
metal atoms are re-deposited on a substrate sitting on the opposite side of the source
metal forming a layer of NC thin film.

Chemical vapor deposition
Chemical vapor deposition (CVD), as revealed by its name, is a chemical coating process
governed by chemical reactions. Typically, the substrate or wafer is exposed to reactive
gas precursors which decompose on the substrate to form desired thin film via chemical
reaction. CVD is commonly used in the semiconductor industry to produce high purity
thin layers of metals. Popular CVD variants include metal organic CVD (MOCVD),
plasma enhanced CVD, atomic layer CVD and vapor phase epitaxy, etc. As an example
for metal organic CVD (MOCVD), high purity metal organic gas precursor flows in a
reactor and dissociated into volatile gas and NC solid deposits on a substrate. The surface
reaction of metal organic leads to condition for epitaxial growth of complex
semiconductor layered structures.

Top-down approach
The top-down method uses a completely different approach which starts from existing
bulk CG materials and breaks them into NC materials by introducing a significant
11

amount of plastic strain. The major top-down method is often called severe plastic
deformation (SPD) which includes high pressure torsion (HPT), equal channel angular
pressing (ECAP), accumulative roll bonding (ARB) and mechanical alloying (MA).
(Note: some literature does not consider MA as a SPD method. But in this study, I prefer
to include MA as one of the SPD methods because a large amount of deformation is
introduced during milling. )

Equal channel angular pressing
Equal channel angular pressing (ECAP) technique produces significant plastic strain
through an extrusion process and greatly increases dislocation density without an overall
change in sample dimensions. During ECAP, a metal billet in the form of rod or bar is
pushed through an angled channel (usually 90
o
~ 110
o
) using a hydraulic press at room
temperature or elevated temperature. A large amount of strain is introduced when the
billet is passing through the angled corner leading to increased dislocation density and
grain refinement. To achieve uniform and optimal grain refinement, this process may be
repeated for several passes leading to uniform UFG or NC microstructures [17-19]. The
final grain size (saturated grain size) is controlled by a number of factors including the
number of passes, the processing temperature, the strain rate and the dynamic recovery
rate of processed materials. Generally speaking, more passes, low temperature, high
strain rate can result in finer grain size on materials with limited dynamic recovery rate.
However, like most other SPD materials, the resulting UFG or NC structure achieved by
12

ECAP is not thermally stable which means significantly grain growth can be observed at
elevated temperature [20,21].

High pressure torsion
High pressure torsion (HPT) is another active research field in making NC or UFG Cu, Al
and Ti pure metals and their alloys [22-24]. In HPT, a disk specimen is placed between
two anvils. One anvil is rotated with respect to the other one to create large torsion force
under a very high compressive pressure (typically ~10
9
Pa). The middle region of the disk
experiences the least amount of deformation. Therefore, multiple turns are required to
achieve a uniform NC structure from the middle to the edge regions. HPT can produce
typical grain size of 100 – 200 nm depending on processing temperature, torsion speed,
materials and the number of turns the anvil rotates [22]. Similar to ECAP processed
materials, the thermal stability of HPT specimens is usually very low. Some study reports
a micro-hardness decrease at temperatures as low as 50
o
C [25]. Another obvious
disadvantage of HPT is that the final product size is limited by the size of the anvils
which is typical a disk shape with a diameter of several centimeters. In this work, HPT
was used as one of the processing steps to shape the microstructure of the aluminum alloy
5083 based metal matrix composite. You will see more details and the microstructure
reshaping by HPT in section 2.2.3 and chapter 6.

13

Accumulative roll bonding
Accumulative roll bonding (ARB) is the only rolling SPD method that can produce UFG
materials continuously [26.27]. Regular rolling introduces limited grain size reduction
due to large cross section reduction of the end product. However, ARB deforms materials
through rolling and bonding repeatedly. The rolled materials are cut into halves, stacked
back to original dimensions and rolled again. This process can repeat many cycles until
the desired UFG structure is achieved. ARB not only serves a rolling process but also a
bonding purpose. To get a nice bonding interface, the surface of rolled materials is often
greased and wire-brushed before re-stacking. And this roll-bonding process is often
carried out at elevated temperature right below the recrystallization temperature of the
processed material to facilitate deformation and bonding [27]. The resulting
microstructures are pancake or ultrafine lamellae elongated along the rolling direction.
Despite high strength of ARB processed materials, tensile elongation is often limited due
to early plastic instability [28].

Mechanical alloying
Mechanical alloying (MA) is often referred to high energy ball milling during which
particles or powders are fractured and cold-welded repeatedly [29]. The milled product is
usually a powder which requires additional consolidation steps to form a bulk material.
The beauty of this method is that it allows powders of different materials to be milled
together during milling which gives a lot of opportunities to engineer or design
14

nanocomposite materials. As a major synthesis method in this work, NC or UFG AA
5083 powders were fabricated using one variant of MA technique called cryomilling. As
illustrated by its name, cryomilling is a ball milling process at cryogenic temperature
(usually liquid nitrogen temperature 77 K). The liquid nitrogen instantly removed the
heat generated during milling and thus suppressed grain growth leading to NC powders in
a shorter time. The following section will focus on the cryomilling technique in more
details.

1.2.2 Cryomilling
Cryomilling is one variant of the MA process mentioned above. Regular room
temperature ball milling has a particular challenge in making NC or UFG aluminum and
its alloys. The final grain size of aluminum after regular milling can hardly reach
submicron or nano region due to high rate of dynamic recovery and high stacking fault
energy of aluminum. During room temperature milling, aluminum can quickly reach a
saturated grain size (no further grain size reduction) when the work hardening rate
balances with its high dynamic recovery rate (i.e. dislocation annihilation rate). Moreover,
the milled materials have poor thermal stability and are subject to recovery and grain
growth in subsequent high temperature post processing steps.

In contrast, cryomilling is a ball milling technique at cryogenic temperature. In the
present study, cryomilling was carried out in an attrition mill as shown in Figure 1.3. The
15

capacity of the attrition mill varies from 100 g to several hundred of kg. During
cryomilling, materials in powder form are sheared, fractured and cold-welded via
repeated collisions with steel milling balls in the liquid nitrogen slurry. Grain refinement
process during cryomilling can be divided into three major stages: (1) shear band
formation increases dislocation density, (2) low angle grain boundaries (LAGBs) evolves
from dislocation movement at certain level of strain via GB sliding and rotation, (3)
LAGBs transform to high angle grain boundaries (HAGBs) with excessive deformation
[30]. The cryogenic temperature quickly dissipates the heat generated from milling and
thus slows down the dynamic recovery. Therefore, cryomilling is capable of reaching
smaller grain size in a shorter time compared to room temperature ball milling.
Surprisingly, cryomilled powders have excellent thermal stability which results in limited
recovery and grain growth in the following high temperature processing steps such as
degassing and consolidation. In chapter 6, you will see no significant grain growth in
cryomilled AA 5083 after annealing at 400
o
C for one hour. The good thermal stability is
attributed to aluminum nitride (AlN) dispersoids segregated at grain boundaries [31,32],.
The grain boundaries movement is pinned down by these dispersoids which lead to very
good thermal stability. The nitrogen source is the liquid nitrogen used during cryomilling
which can not be found in unmilled AA 5083. Major processing elements and parameters
of cryomilling and post processing steps will be discussed in the following context.

16

Fig. 1.3 (a) Picture of an attrition mill used for cryomilling, (b) schematic picture
showing the ball mill cylinder used for cryomilling [29]

Process control agent
During milling, the powder particles can agglomerate due to cold-welding, especially for
ductile materials. This powder agglomeration can result in low milling efficiency and
therefore increase milling time and cost. An organic surface-active substance called the
process control agent (PCA) is often added to prevent excessive cold welding [29].
Stearic acid (C
18
H
36
O
2
) is a common organic acid used as PCA for ductile FCC metals
[29]. Many researchers have used stearic acid (C
18
H
36
O
2
) as PCA to prevent excessive
cold welding of Al powders during milling. The adsorbed surface-active PCA can lower
the surface energy of the particle surface. Because the energy increase ∆ 𝐸𝐸 for the
physical size reduction (more surface area) of a particle is expressed by
∆ 𝐸𝐸 = 𝛾𝛾 ∆ 𝑆𝑆
where 𝛾𝛾 is the surface energy (energy per unit area) and ∆ 𝑆𝑆 is the increase of the
17

surface area. Therefore, lowering 𝛾𝛾 by PCA will facilitate the particle fracture, reduce
the effect of cold-welding and increase the milling efficiency and yield [29]. However,
the addition of PCAs could alter the milling process. If too much PCA is used, it will
lower the quality of the final product due to excessive contamination. In the present study,
a minimum amount of PCA (0.2 wt% of stearic acid) was added to the attrition mill.

Hot vacuum degassing
The addition of PCA inevitably introduced contaminants such as hydrogen and oxygen
[33,34]. If not treated properly, the amount of hydrogen content will be large enough to
deteriorate ductility. Therefore, a hot vacuum degassing process is required to pump
hydrogen content out of the NC powders. During degassing, the powders are vacuumed
at elevated temperature (~ 400
o
C) for several hours which inevitably increases the grain
size [35]. After degassing, the powders are transferred to a glove box and later sealed in
an aluminum can for storage. A schematic illustration of the degassing process is shown
in Figure 1.4. Inevitably, slight grain growth resulted from degassing is a major concern.
The average grain size of as-cryomilled powder (~50 nm) increases to ~200 nm after
degassing at ~0.8 T
m
of Al. As expected, higher degassing temperature resulted to larger
grain growth [35].

18

Fig. 1.4 Schematic diagram of degassing process

Consolidation
To produce bulk materials, consolidation is performed on the cryomilled powders along
with the aluminum can which seals the powders. Because the consolidation technique
plays a critical role on the grain size and the porosity level of the final product, the
methods we used to consolidate NC powders are critical to the mechanical properties of
the final bulk products. The available consolidation techniques include hot isostatic
pressing (HIP), cold isostatic pressing (CIP), extrusion, rolling, quasi-static forging (QI
forging) and spark plasma sintering (SPS). HIP did a great job consolidating cryomilled
powders to a full density. However, the hot environment during pressing is a big concern
due to grain growth. To avoid significant grain growth, CIP replaced HIP as the primary
consolidation process.

19

However, for both HIPped and CIPped samples, the consolidation pressure is not high
enough to break prior particle boundaries (PPBs) which are boundaries of cryomilled
powders prior to consolidation [36]. Secondary consolidation method is often necessary
to introduce further strain and break prior particle boundaries (PPBs) as shown in Figure
1.5. PPBs are generally caused by native oxide layer formed on the powder particles
during transportation and handling. The presence of PPBs can lead to crack initiation and
thus poor ductility. The secondary consolidation includes extrusion, rolling and forging. A
newer forging method, often called Ceracon forging, dual mode dynamic (DMD) forging
or quasi-isostatic (QI) forging, has been investigated and used as both primary and
secondary consolidation methods. During QI forging, the uniaxial pressure was
transferred to a quasi-isostatic pressure via a granular pressure transmitting medium
(PTM), typically graphitic particulate. The high isostatic pressure applied in QI forging
helps to break PPBs and reduce porosity after consolidation. As a result, the QI forged Al
samples exhibited better fracture toughness than ones consolidated using other methods
such as HIP, CIP and conventional forging [37]. Due to this excellent result, QI forging
can be used as both primary and secondary consolidation methods.

Spark plasma sintering (SPS) is another new consolidation technique which applies
current and pressure simultaneously. When SPS is applied to NC powders, densification
can be achieved at significantly lower temperatures in a shorter time compared with
conventional sintering, thereby limiting grain growth and preserving the microstructure
20

[38]. Cryomilled brass consolidated by SPS was reported to have a compressive strength
of 950 MPa, which is significantly higher than that of conventional brass alloy (200~400
MPa) [38].

Fig. 1.5 Prior particle boundaries in consolidated NC Al samples [39]

In sum, the fabrication of cryomilled NC materials includes cryomilling, hot vacuum
degassing, primary consolidation and secondary consolidation. Possible fabrication routes
are illustrated in Figure 1.6. If we study the processing parameters of each step by trial
and error, we can easily build up more than 100 possible fabrication routes. One major
objective of this research project is to find out the fabrication recipe resulting to the
optimal NC Al alloys with the help of microstructure characterization and mechanical
testing.
21

Fig. 1.6 Fabrication routes of cryomilling

1.2.3 High pressure torsion
The pioneer work and general principles of high pressure torsion (HPT) can be traced
back to the high pressure experiments performed by Bridgman in early 1900’s [40].
However, the application of high pressure torsion as a metal processing method was
realized only about two decades ago [41]. The basic principle of HPT is that greater
torsional strain can be applied to a material without fracture if a longitudinal compressive
22

load is simultaneously applied along the torsion axis. In modern HPT apparatus, the
sample (often a thin disk) is placed in a depression between two anvils as shown in
Figure 1.7. The bottom anvil will be raised upwards until the disk is fully enclosed by
two anvils under high compressive pressure (~ several GPa). The bottom anvil then
rotates with respect to the top one during which a large amount of shear strain can be
introduced via the friction between specimens and anvils [42]. The true accumulative
strain applied by HPT can be expressed by Equation 2 [43]:
𝜀𝜀 = ln (
2 𝜋𝜋𝜋𝜋 𝜋𝜋 ℎ
) (Eq. 2)
Where N is the number of turns of anvil revolution, r is the radial distance from the disk
center, and h is the initial thickness of the disk. The equivalent von Mises strain is given
by a relationship of the form [44]:
𝜀𝜀 𝑒𝑒𝑒𝑒
=
2 𝜋𝜋𝜋𝜋 𝜋𝜋 √ 3 ℎ
(Eq. 3)

23

Fig. 1.7 (a) Schematic illustration of quasi-constrained HPT, (b) pictures of HPT setup
used in this work

One important concern of HPT sample is the microstructure inhomogeneity due to a
torsional strain gradient from the periphery to the center of the HPT disk. A uniform
microstructure and properties can be achieved by increasing number of turns of HPT [22].
The grain size will reach a limit value when the strain hardening rate balances the
dynamic recovery rate at high strain level [45]. This limit value will be first reached by
24

the peripheral region where largest torsional strain is imposed. The center region will
eventually reach the limit grain size when the accumulative strain is large enough as
increasing number of turns. As an alternative approach to obtain homogeneous structure,
specimens in the form of ring was utilized in HPT processing since the torsional strains
are the same for all areas in the ring sample [46]. Besides number of turns, another
processing parameter, the applied load, has little influence over the measured torque and
therefore the microstructure and the properties [47].

Another inherent issue of HPT is the size limitation. Typical HPT samples are small disks
with a diameter of 10 mm and 1 mm in thickness. Fabrication of large bulk HPT sample
was attempted using larger and deeper depression on the anvils [48]. However, HPT is
only capable of refining surface grains of the bulk sample because it is very difficult to
impose torsional strain deep inside the materials when sample is larger and thicker.
Nevertheless, it is still feasible to achieve surface hardening using bulk HPT processing.

Previously, HPT was used as a consolidation method for cryomilled powders [23]. In this
work, grain refinement and consolidation are not the main purposes of utilizing HPT.
Instead, we used HPT as a secondary SPD method following cryomilling to reshape the
microstructure and achieve optimal size and spacing of the coarse grain regions in a
trimodal Al-based composite. The effect of HPT on cryomilled samples will be evaluated
in Chapter 6.
25

1.3 Enhancing strength
The extraordinarily high strength exhibited by NC or UFG materials is attributed to not
only the Hall-Petch effect (grain boundary strengthening) mentioned in section 1.1 but
also to a combination of other strengthening mechanisms including precipitate and
dispersion strengthening, solution strengthening and work hardening (dislocation
strengthening). These strengthening mechanisms can be from different various sources
including material composition and synthesis. For instance, grain boundary strengthening
is a result of grain size refinement during SPD. Precipitate and dispersion strengthening is
often due to heat treatment or other high temperature processes. Solution strengthening is
an intrinsic mechanism in which solute atoms play a critical role. Work hardening is
achieved by additional strain introduced to materials during cold work such as extrusion,
rolling or forging. These strengthening mechanisms combined contribute to the final
strength of the end product. In this section, we will discuss the origin of each
strengthening mechanism and estimate their contributions to cryomilled aluminum alloys
based on previous studies. A fun-to-read book by Ashby [materials undergrad textbook] is
recommended to those who are devoted to metal strengthening.

1.3.1 Dislocation
Before we jump into strengthening mechanisms, it is worthwhile to take a look at how
crystalline materials deform. When an external load (compressive or tensile) is applied, a
crystalline material will respond by deformation. As the load increases from zero, each
26

atom in the crystalline lattice is slighted shifted away from its equilibrium position. The
lattice structure is distorted but no atomic bond is broken at this stress level. The small
distortion of each atom accumulates and results in a dimensional change of the material.
When the external load is removed, the material can recover to its original dimensions
just like a spring. This recoverable deformation is called elastic deformation.

When the applied load exceeds a critical value (yield point), the material will have
permanent deformation (called plastic deformation) even after the load is retracted. The
stress required to initiate this non-recoverable deformation is the strength of the material.
The plastic deformation is a result of the generation and movement of linear lattice
defects called dislocations. A deep explanation of dislocations can be found in a book by
Weertman [49]. The presence of dislocations allows materials to yield at a much lower
stress than the ideal strength calculated from inter-atomic force. Dislocations are
responsible for plastic deformation nearly in all crystalline metals. Almost all
strengthening mechanisms involve stopping mobile dislocations. When most dislocations
are trapped and reluctant to move inside materials, higher stress is required to force
dislocations become mobile again, or in other words, deform plastically.

1.3.2 Grain boundary strengthening
Grain boundary (GB) strengthening is the main hardening mechanism for achieving high
strength NC materials compared to conventional CG materials. The strengthening is
27

governed by the Hall-Petch relationship (Equation 1) which predicts an increase of
strength as deceasing grain size. To understand the origin of the Hall-Petch effect, it is
imperative to study how deformation mechanism changes when grain size approaches
nano-region.

Dislocation pile-up
Dislocation pile-up is a classic explanation for the Hall-Petch effect [50,51]. In CG
materials, applying stress creates dislocations in the center of grains (Frack-Read source
[49]). As discussed in the previous section, dislocation movement is the microscopic
reason for plastic deformation. These dislocations move inside a grain under stress and
eventually stop and pile up at GBs as illustrated in Figure 1.8. Coarse grains are big
enough to accommodate many dislocations inside the grain before there is no more room
for the pile-ups. Each dislocation in the pile-up applies a repulsive force which “pushes”
the dislocations near grain boundaries and “helps” them move across GBs to realize
plasticity. In nanocrystalline or ultrafine grains, dislocation density saturates quickly due
to limited grain size. Moreover, dislocations meet GBs more often due to increasing
volume fraction of GB areas. As a result, additional stress is needed to break dislocation
pile-ups and make dislocations mobile again.

28

Fig. 1.8 Illustration of dislocation pile-ups in a coarse grain

Grain boundary sliding / shearing
When dislocation motion is impeded by GBs, alternative GB-based plasticity
mechanisms can take over in NC materials. Various GB-based mechanisms have been
directly observed using in situ TEM nanomechanical testing [52]. Grain boundary sliding
is believed to be a major mechanism when grain size is less than 50 nm [53]. Grain
boundary sliding is reported to be a deformation mechanism responsible for
superplasticity at elevated temperature [54]. Figure 1.9 is a schematic illustration of the
grain boundary shearing process during which a group of grains slide against another
layer under applied stress.
29

Fig. 1.9 Schematic illustration of grain boundary sliding [53]

Grain boundary rotation
Grain boundary rotation is another alternative GB-based mechanism responsible for
plasticity in NC materials. It was proposed that the nano-sized grains can rotate during
plastic deformation and align several grains with different orientation into a single grain
[55]. This deformation mechanism is schematically shown in Figure 1.10 where the slip
plane is represented by a line in each grain. Grain coalescence can be achieved during GB
rotation. Under shear stress, multiple neighboring grains can rotate to the same
orientation, and thus eliminate GBs among these grains. As a result, a larger grain is
formed creating more easy paths for dislocation motion and hence extending plasticity.

30

Fig. 1.10 Schematic illustration of grain boundary rotation [53]

Inverse Hall-Petch effect
The Hall-Petch slope (coefficient k) becomes negative when the grain size is smaller than
a threshold value (typically ~10 nm for metals). This phenomenon is called the inverse
Hall-Petch effect or Hall-Petch breakdown [8,56,57]. Classic pile-up theory can no longer
explain the Hall-Petch breakdown since nano-sized grains are unable to support
dislocation pile-ups. The microscopic reason of the inverse Hall-Petch effect is still
undetermined. Several models were proposed to explain the inverse Hall-Petch effect as
this discussion continues in the science community.

One dislocation-based model predicted a decrease in dislocation energy as a result of
small grain size and showed how this reduced energy affected dislocation motion which
31

led to the inverse Hall-Petch effect [58]. In some models, the inverse Hall-Petch effect is
also believed to be a result of various diffusion processes [59]. In another model, grain
boundary and grain interior are treated as two distinctive phases and special mechanical
properties are assumed in grain boundary phase [60]. Grain boundary rotation is another
possible cause of the inverse Hall-Petch effect. As discussed above, grain boundary
rotation can cause grain coalescence which results in local softening of nanocrystalline
materials.

1.3.3 Precipitate and dispersoid strengthening
Precipitate/dispersoid strengthening (also called age hardening) is a hardening strategy
using heat treatment used to increase the yield strength by introducing second phase
precipitates in alloy systems. With the temperature change during heat treatment, the
solubility changes in an alloy system. This change leads to the formation of second phase
particles called precipitates or larger intermetallic particles named dispersoids distributed
preferably on GBs. These precipitate/intermetallic particles are able to pin down mobile
dislocations (Orowan mechanism [61]) and thus increase the yield strength of the material.
Specifically, Al
3
Mg
2
(β phase) precipitates and Al
6
(Mn,Fe) dispersoids can be formed
intergranularly and intragranularly in AA 5083 H131 during hot vacuum degassing, HIP
or Qi forging which are carried out at elevated temperatures. For the interest of this study,
precipitate/dispersoid strengthening is not prominent for the non-heat-treatable AA 5083.
In fact, β phase precipitates is the major reason for sensitization which makes AA 5083
32

subject to stress corrosion cracking (SCC) and Al
6
(Mn,Fe) dispersoids are potential sites
for early crack nucleation which will be discussed in Chapter 5.

1.3.4 Solid solution strengthening
Solid solution strengthening is due to the diffusion of an alloying element to a metal
matrix. For AA 5083, Mg is added to Al matrix to form an Al- 4.5 (wt. %) Mg alloy. The
Mg solute atoms interact with dislocations and impede dislocation motion. Trace amount
of Cr, Mn, Fe and Si can be found in AA 5083 in the form of intermetallic particles. The
overall effect is increasing the yield strength of the alloy system compared to pure metals.
The strength increase due to alloying can be expressed by Equation 3
∆ 𝜏𝜏 = 𝐶𝐶 ∙ 𝐺𝐺 ∙ ∆ 𝑐𝑐 ∙ 𝜀𝜀 4/ 3
(Eq. 3)
Where ∆ 𝜏𝜏 is the shear strength increase due to alloying, C is a constant, G is the shear
modulus, ∆ 𝑐𝑐 is the change in lattice parameter due to solute atoms, and 𝜀𝜀 is the strain
caused by misfit in the lattice.

1.3.5 Work hardening
The polycrystalline material can be hardened by applying additional strain to increase
dislocation density. Dislocations can be obstacles for themselves. As the dislocation
density increases, it is more likely to have dislocation-dislocation interaction and thus
impede dislocation motion. The increasing stress after yielding point observed in a
typical tensile stress and strain curve is a typical example of work hardening. The amount
33

of strength increase due to work hardening can be described by Equation 4.
∆ 𝜏𝜏 = 𝛼𝛼 ∙ 𝐺𝐺 ∙ 𝑏𝑏 ∙ 𝜌𝜌 1/ 2
(Eq. 4)
Where ∆ 𝜏𝜏 is the shear strength increase caused by work hardening, 𝛼𝛼 is a constant
ranging from 0.3 to 0.6, G is the shear modulus, b is the Burgers vector and 𝜌𝜌 is the
dislocation density which is ~ 10
16
/m
2
for annealed and cold-worked metals [62]. A lot of
cold working methods such as extrusion, drawing and rolling are effective strengthening
ways for non-heat-treatable alloys. In this work, work hardening of AA 5083 was
achieved during secondary consolidation such as extrusion and forging.

1.3.6 Extrinsic reinforcing
Metallic materials can be reinforced simply by dispersing extrinsic hard substances in the
metal matrix leading to metal-matrix composites (MMC). The common reinforcements
are ceramic, another metal or organic compound in the forms of particles, fibers or
whiskers. Common MMC systems such as Al/Al
2
O
3
, steel/BN, Ti/SiC are widely used in
automobile and aerospace industries. These reinforcement materials not only strengthen
the matrix but also improve other physical properties such as wear resistance and thermal
conductivity. In this work, AA 5083 MMC was reinforced by boron carbide particles to
gain extra strength in addition to grain boundary strengthening. Fabrication and
properties of this MMC will be discussed in section 1.4.

34

1.4 Paradox of strength and ductility
One problem that is haunting NC or UFG materials constantly is the lack of ductility at
room temperature. Despite enormous improvement in strength, the elongations to failure
(specially the uniform elongation) of most NC or UFG materials are limited to a few per
cent which inevitable leads to low fracture toughness [36,63]. This is a well-known
paradox which states that “materials may be strong or ductile, but rarely both at once
[64,65].” The brittle nature of NC or UFG materials is not surprising since small grains
have low dislocation storage capacity and many plastic deformation mechanisms
available in CG materials are “locked” or “disabled” in NC or UFG materials in order to
enhance strength. According to Considère’s criterion (Equation 5), the onset of necking
(deforming instability) occurs easily in NC or UFG materials due to low work hardening
capability of small grains.
�
𝜕𝜕𝜕𝜕
𝜕𝜕𝜕𝜕
�
𝜕𝜕 ̇ ≤ 𝜎𝜎 (Eq. 5)
Therefore, plasticity is rarely seen because strain localization and crack nucleation take
place long before uniform yielding could occur. The brittleness and low fracture
toughness will disqualify many NC or UFG materials from commercial applications and
eliminate the use of some fabrication processes that require material plasticity. In the
decade-long battle fighting poor ductility, researchers have proposed numerous
toughening strategies to increase both strength and ductility or to improve the overall
toughness of NC or UFG materials. Several toughening methods will be discussed below.

35

High strain SPD
Several papers report that repeating SPD passes / rounds could result in improvement of
both strength and ductility. This finding seems to be contradictory to the paradox of
strength and ductility since more SPD processing will generally produce smaller grain
size and worse ductility. In an early report by Valiev [66], pure Cu (99.996%) and pure Ti
(99.98%) were processed by ECAP (section 1.2.1) and HPT (section 1.2.3) at room
temperature, respectively. Both Cu and Ti samples exhibited simultaneous increase of
both strength and ductility when they experienced more passes of ECAP and turns of
HPT. This phenomenon indicates an interesting fact that a SPD process (ECAP or HPT)
increases the strength of a material accompanied with loss of ductility as more plastic
deformation is applied during SPD. However, when the applied deformation (strain)
reaches a critical level, both strength and ductility can be improved by further
deformation as shown in Figure 1.11. This toughening strategy is also found to be valid
for various metals and alloys [67-69]. The mechanism of this approach is still not fully
understood. Microstructural investigations suggested that the improvement of both
strength and ductility may be attributed to new deformation mechanisms (e.g. GB sliding)
that are enabled during high strain SPD. The advantages of this toughening approach are
wide applicability to many metals and alloys and straight-forward processing step
(increasing SPD passes/turns). The possible drawback is the low thermal stability of high
strained materials since both strength and ductility can be lost under elevated temperature.
Nevertheless, this unique approach is amazing to me since unlike other toughening
36

strategies it is able to increase ductility without any compromise of strength.

Fig. 1.11 Engineering stress and strain curves of (a) pure Cu processed by ECAP, and (b)
pure Ti processed by HPT [66]

37

Lowering GB misorientation
Misorientation between neighboring grains has direct influence on both strength and
ductility of NC or UFG materials [70,71]. High angle grain boundaries (HAGBs) with
large misorientation are more effective to stop mobile dislocations, while low angle grain
boundaries (LAGBs) are relatively weak barriers for dislocation motion. The idea of this
approach is straight-forward. By controlling the proportion of HAGBs and LAGBs, one
can tailor the ductility and strength to a desired level. In other words, ductility can be
improved by sacrificing some amount of strength. However, controlling GB
misorientation and accordingly their spatial distribution are not easy. Hu reported a
lamellar structure of Al 7xxx alloys fabricated using cryomilling followed by two
additional thermomechanical processing – rotary swaging and high strain rate extrusion
[72]. Each lamellar band was composed by ultrafine grains with LAGBs which leads to
an elongation improvement as shown in Figure 1.12. However, the ductility was
increased at the expense of strength using this approach.
38

Fig. 1.12 Engineering tensile stress and strain curves of conventional Al 709x alloy by
slow rate extrusion (CG-SE), cryomilled UFG alloy after slow rate extrusion (UFG-SE),
after rotary swaging (UFG-RS) and after a combination of rotary swaging and high rate
extrusion (UFG-RS-HRE) [72]

Nano twinning
A third method to increase both strength and ductility is to introduce coherent twin
boundaries in materials with inherently low stacking fault energy. For fcc metals, a
coherent twin boundary is defined as a (111) mirror plane at which the stacking sequence
is reversed. Twin boundaries are capable of strengthening materials like regular GBs via a
slip transfer mechanism [73]. In addition, twin boundaries are more stable than regular
GBs due to the fact that the energy of twin boundaries is much lower than that of regular
39

GBs. The strengthening of twin boundaries is directly determined by the spacing between
neighboring twins (defined as twin thickness). A systematic study of twin thickness in a
pure nano-twinned copper (nt-Cu) synthesized using electrodeposition [74] showed a
maximum strength can be achieved when twin thickness is ~ 15 nm. Another interesting
fact was that reducing twin thickness could result in excellent elongation (10% or more)
as shown in Figure 1.13. This approach is capable of making materials with both
extraordinarily high strength and ductility but limited to a family of materials with low
stacking fault energy. Furthermore, precise twin thickness control can be only achieved
by a few bottom-up synthesis methods which makes bulk fabrication difficult.

Fig. 1.13 True tensile stress and strain curves for nt-Cu with various twin thickness
(denoted by the number following nt in unit of nanometer). CG and UFG twin-free
counterparts are included for reference. [74]

40

Multi-scale toughening
Similar to ductile-phase toughening, domains with multi-scale grain size distribution can
be used to increase the elongation of brittle NC or UFG materials. In this study, a
multi-scale microstructure was investigated as a major toughening strategy for cryomilled
materials. The unique microstructure consisting of two (bimodal) or more (multi-modal)
areas with different grain size distribution (CG or UFG) can be achieved by a variety of
methods such as thermomechanical processes, stress-induced grain growth and mixture
of powders. The key idea behind this approach is to trade off some strength to gain
ductility. In this section, we will discuss the concept of bimodal and trimodal AA
accordingly with the synthesis technique and the resulting mechanical properties.

Bimodal structure
In a bimodal microstructure, ductile CG regions are uniformly dispersed in a brittle UFG
matrix as shown in Figure 1.14. The ductile CG regions in which plasticity is readily
available contribute to ductility. In contrast, the UFG matrix provides a strong back-bone
responsible for high strength of the material. Due to material’s powder state during
cryomilling, the bimodal structure can be easily achieved by blending NC (cryomilled)
and CG (unmilled) powders before degassing and consolidation. The volume fraction and
the size of the CG regions can be simply tailored by controlling the amount and the
particle size of the unmilled powders. This is a huge engineering advantage of
cryomilling over other synthesis methods.
41

Fig. 1.14 Schematic illustration of a bimodal structure with CG regions evenly dispersed
in a UFG matrix

The tensile curves of a bimodal Al - 7.5 Mg with changing CG fraction are plotted in
Figure 1.15 and corresponding mechanical properties are listed in Table 1.3 [75]. Caution
is advised to readers that the tensile data listed above was tested using a non-standard
miniature dog-bone specimen (dimensions will be provided in Chapter 3). Therefore,
direct comparison with standard tensile data is not recommended due to the size effect of
tensile samples [76]. All mechanical tests in this dissertation were performed at room
temperature unless specified. The cryomilled AA has an impressive five-fold increase in
strength. However, it is not surprising that the elongation of 100% cryomilled Al alloy
dramatically drops to only ~ 1.4% compared to that of conventional Al alloy (~16%). To
improve the ductility, unmilled CG powders were blended with cryomilled powders
before consolidation forming the bimodal structure as discussed earlier. As shown in
42

Figure 1.15, the ductility monotonically increased with CG addition at the cost of
moderate decrease in strength.

Fig. 1.15 Tensile curves of conventional, 100% cryomilled and bimodal Al-7.5Mg [75]

Table 1.3 Tensile properties of conventional, 100% cryomilled and bimodal Al-7.5Mg
[75]
Samples
0.2% Yield
strength
(MPa)
Ultimate tensile
strength
(MPa)
Elongation
(%)
100%
Cryomilled Al-7.5 Mg
641 847 1.4
Cryomilled Al-7.5 Mg +
15% unmilled
630 778 2.4
Cryomilled Al-7.5 Mg +
30% unmilled
554 734 5.4
Commercial Al-7.5 Mg 145 281 16
43

The deformation and failure mechanism of bimodal materials were investigated in depth
via microstructure imaging and in situ techniques [77,78]. When the bimodal grain
structure is under tension, micro-voids and cracks initiate mainly in the brittle UFG
matrix and at the CG/UFG interface. The ductile CG regions impede the crack growth
and propagation by crack tip blunting [77]. The schematic illustration is shown in Figure
1.16. According to the strain measurement during an in situ tensile test, plastic strain was
localized mainly in the CG regions (Figure 1.17) due to higher dislocation storage
capacity. In addition, the CG regions exhibited better strain hardening ability than the
UFG matrix and thus contributed to uniform elongation during tensile deformation [78].

44

Fig. 1.16 Schematic of deformation and fracture mechanism of a bimodal grain structure
under tension along extrusion direction [77]

Fig. 1.17 Visible light micrograph of the bimodal structure of AA 5083 (left) and local
strain level of the same area (right) [78]

45

Trimodal structure
Metal matrix composites are attractive materials in a wide range of structural applications.
In practice, the reinforcement of conventional aluminum and other alloys with
second-phase ceramic particulates or fibers leads to very significant increases in the
stiffness and strength of the material. The bimodal structure can be further strengthened
by adding ceramic reinforcement such as boron carbide (B
4
C) particles leading to a
trimodal structure which consists of CG regions, UFG matrix and reinforcing phase as
shown in Figure 1.18. The mechanical properties of the trimodal Al based composites
were listed in Table 1.4 with comparison to conventional Al alloy and 100% NC Al alloy.
The trimodal sample showed limited compressive strain-to-failure (0.8%). After
annealing at 723 K for 2 hours, the trimodal sample has a slightly higher strain-to-failure
(2.5%) with a moderate decrease in compression yield strength.

46

Fig. 1.18 Schematic illustration of a trimodal structure with CG regions dispersed in a
UFG matrix reinforced with hard particles

Table 1.4 Mechanical properties of the trimodal Al 5083 compared with a CG sample and
a 100% cryomilled NC sample. “O” stands for O type heat treatment. “10/50” indicates
10 wt% of B
4
C and 50 wt% of CG addition. “A” stands for annealing at 723 K (450
o
C)
for 2 hours [79].
Samples
0.2% Yield
strength
(MPa)
Tensile
elongation
(%)
Compressive
strain to failure
(%)
Al 5083-O [80] 123 (tensile) 16% -
NC Al 5083 [80] 334 (tensile) 8.4% -
Trimodal 10/50 [61] 1064
(compressive)
- 0.8%
Trimodal 10/50-A [61] 1058
(compressive)
- 2.5%

47

The fabrication of the trimodal AA is straight-forward. The B
4
C particles were blended
and cryomilled together with Al powders in liquid nitrogen slurry forming a good
interfacial bonding between the Al powders and the reinforcing phase. After cryomilling,
the composite powders mixed with CG powders (similar to bimodal materials) were
degassed and consolidated to bulk form. The grain refinement process was affected by
the presence of B
4
C particles during cryomilling. The influence of B
4
C particles on the
microstructural evolution of the Al nanocomposite will be discussed in Chapter 4. The
deformation and failure mechanism of the trimodal AA 5083 and the attempt to
microstructure optimization will be described in Chapter 5 and Chapter 6, respectively.

The toughening strategy by bimodal or trimodal structure is easy to understand and
feasible due to the powder nature of cryomilling. No other synthesis method can provide
such high level of control on microstructural design. Thanks to cryomilling, uniform
bimodal and trimodal structures can be achieved in bulk form which is favored by future
commercialization. However, cryomilling is not “cheap” and requires large amount of
capital investment on equipment because it involves many processing steps such as
degassing, forging and extrusion. In addition, it is difficult to achieve nano-sized grains
(< 100nm) in bulk sample due to grain growth during subsequent thermomechanical
processing.

48

References
[1] H. Gleiter, Prog. Mater. Sci. 33, 223, 1989.
[2] R. Birringer, Mater. Sci. Eng. A, 117, 33, 1989.
[3] C. Suryanarayana, Intl. Mat. R., 40, 41, 1995.
[4] K. Lu, Mater. Sci. Eng. R., 16, 161, 1996.
[5] H. Gleiter, Acta Mater., 48, 1, 2000.
[6] E. O. Hall, Proc. Phys. Soc., B64, 747, 1951.
[7] N. J. Petch, J. Iron Steel Inst., 174, 25, 1953.
[8] C. E. Carlton and P. J. Ferreira, Acta Mater., vol. 55, no. 11, pp. 3749–3756, Jun.
2007.
[9] R. Birringer, H. Gleiter, H. Klein, and P. Marquardt, Phys. Lett., vol. 102A, no. 8, pp.
365–369, 1984.
[10] M. Turker, Mater. Sci. Eng. A, vol. 367, no. 1–2, pp. 74–81, Feb. 2004.
[11] K. M. Lee, D. J. Lee, and H. Ahn, Mater. Lett., vol. 58, no. 25, pp. 3122–3125, Oct.
2004.
[12] A. Djekoun, N. Boudinar, A. Chebli, A. Otmani, M. Benabdeslem, B. Bouzabata,
and J. M. Greneche, Phys. B Condens. Matter, vol. 404, no. 20, pp. 3824–3829, Nov.
2009.
[13] L. Lu, S.X. Li and K. Lu, Scripta Mater. 45, 1163, 2001.
[14] C. Schuh, T.G. Nieh and T. Yamasaki, Scripta Mater. 46, 735. 2002.
[15] C. Schuh, T.G. Nieh and H. Iwasaki, Acta Mater., 51, 431, 2003.
49

[16] Y. Zhao, T. A. Furnish, M. E. Kassner, and A. M. Hodge, J. Mater. Res., 27, 3049,
2012.
[17] M. Furukawa, Z. Horita, M. Nemoto and T.G. Langdon, J. Mater. Sci., 36, 2835,
2001.
[18] R.Z. Valiev and T.G. Langdon, Prog. Mater. Sci., 51, 881, 2006.
[19] R.Z. Valiev, Y. Estrin, Z. Horita, T.G. Langdon, M.J. Zehetbauer and Y.T. Zhu, JOM
58, 33, 2006.
[20] J. Wang, Y. Iwahashi, Z. Horita, M. Furukawa, M. Nemoto, R.Z. Valiev and T.G.
Langdon, Acta Mater., 44, 2973, 1996.
[21] F.J. Humphreys, P.B. Prangnell, J.R. Bowen, A. Gholinia and C. Harris, Phil. Trans.
R Soc. Lond. A 357 1663, 1999.
[22] A. P. Zhilyaev and T. G. Langdon, Prog. in Mater. Sci., 53, 893, 2008.
[23] Z. Lee, F. Zhou, R. Z. Valiev, E. J. Lavernia and S. R. Nutt, Scr. Mater., vol. 51, no. 3,
pp. 209–214, Aug. 2004.
[24] A. Zhilyaev, G. Nurislamova, B. K. Kim, M. Baró, J. Szpunar, and T. G. Langdon,
Acta Mater., vol. 51, no. 3, pp. 753–765, Feb. 2003.
[25] H. Jiang, Y. Theodore Zhu, D. P. Butt, I. V. Alexandrov and T. C. Lowe, Mater. Sci.
and Eng. A, 290 128, 2000.
[26] N. Tsuji, Y . Saito, S. H. Lee, and Y . Minamino, Adv. Eng. Mater., vol. 5, pp. 338–344,
2003.
[27] Y. Saito, H. Utsunomiya, N. Tsuji, and T. Sakai, Acta Mater., vol. 47, no. 2, pp. 579–
50

583, 1999.
[28] N. Tsuji, Y . Ito, Y . Saito and Y . Minamino, Scipta Mater. 47. 893, 2002.
[29] C. Suryanarayana, Prog. Mater. Sci., vol. 46, no. 2001, pp. 1–184, 2001.
[30] H.J. Fecht, Nanostruct. Mater., 6, 33, 1995.
[31] H. Yang, T. D. Topping, K. Wehage, L. Jiang, E. J. Lavernia, and J. M. Schoenung,
Mater. Sci. Eng. A, Jul. 2014.
[32] I. Roy, M. Chauhan, and E. J. Lavernia, Metall. Mater. Trans. A, vol. 37, no. March,
pp. 721–730, 2006.
[33] L. Shaw, M. Zawrah, J. Villegas, H. Luo, and D. Miracle, Metall. Mater. Trans. A,
34A, 159, 2003.
[34] A.M. El-Sharik, U. Erb, G. Palumbo and K.T. Aust, Scripta Metall. Mater., 27, 1185,
1992.
[35] B. Ahn, A.P. Newbery, E.J. Lavernia and S.R. Nutt, Mater. Sci. Eng. A, 463, 61,
2007.
[36] A. P. Newbery, B. Ahn, T. D. Topping, P. S. Pao, S. R. Nutt, and E. J. Lavernia, J.
Mater. Process. Technol., vol. 203, pp. 37–45, Jul. 2008.
[37] A. P. Newbery, B. Ahn, P. Pao, S.R. Nutt and E.J. Lavernia, Adv. Mater. Res., 29−30,
21, 2007.
[38] H. Wen, T. D. Topping, D. Isheim, D. N. Seidman, and E. J. Lavernia, Acta Mater.,
vol. 61, no. 8, pp. 2769–2782, May 2013.
[39] F. Tang, C. P. Liao, B. Ahn, S. R. Nutt, J. M. Schoenung, Powder Metallurgy, 50,
51

307-312, 2007.
[40] P. W. Bridgman. J. Appl. Phys., 14:273, 1943.
[41] R. Z. Valiev, A. V. Korznikov and R. R. Mulyukov, Mater. Sci. Eng. A, vol. 168, pp.
141–148, 1993.
[42] M. Kawasaki and T. G. Langdon, Mater. Sci. Eng. A, vol. 498, pp. 341–348, 2008.
[43] R. I. Kuznetsov, V . I. Bykov, V . P. Chernyshev, V . P. Pilyugin, N. A. Yefremov and A.
V. Pasheyev, “Plastic deformation of solid bodies under pressure: Equipment and
methods”. Preprint 4/85, IFM UNTs AN SSSR, Sverdlovsk, USSR; 1985 (In
Russian).
[44] R. Z. Valiev, Y . V . Ivanisenko, E. F. Rauch, and B. Baudelet, Acta Mater., vol. 44, no.
12, pp. 4705–4712, 1996.
[45] C. Xu, Z. Horita and T. G. Langdon. Acta Mater., 55:203, 2007.
[46] Y . Harai, Y . Ito and Z. Horita, Scripta Mater., 58:469, 2008.
[47] F. Wetscher, A. V orhauer and R. Pippan, Mater. Sci. Eng. A, 213, 410–411, 2005.
[48] G. Sakai, K. Nakamura, Z. Horita and T. G. Langdon, Mater. Sci. Eng. A, 406, 268,
2005.
[49] J. Weertman and J. Weertman, “Elementary Dislocation Theory”, Macmillan, New
York, 1966.
[50] C.S. Pande, R.A. Masumura and R.W. Armstrong, Nanostruct. Mater., 2, 323, 1993.
[51] T. R. Smith, R.W. Armstrong, P. M. Hazzledine, R. A. Masumura and C. S. Pande,
Pittsburgh, PA, Mater. Res. Soc. (MRS) Proc. 362, p. 31, 1995.
52

[52] Q. Yu, M. Legros, and A. M. Minor, MRS bulletin, vol. 40, January, 62-68, 2015.
[53] M. A. Meyers, A. Mishra and D. J. Benson, Prog. Mater. Sci., 51, 427, 2006.
[54] T. G. Langdon, Mater. Sci. Eng. A, vol. 174, pp. 225–230, 1994.
[55] D. Jia, K.T. Ramesh and E. Ma, Scripta Mater. 42, 73, 2000.
[56] T. G. Nieh and J. Wadsworth, Scr. Metall. Mater., 25, 955, 1991.
[57] J. Eckert, J. C. Holzer, C. E. Krill and W. L. Johnson, J. Mater. Res., 7, 1751, 1992.
[58] R. O. Scattergood and C. C. Koch, Scr. Metall. Mater., 27, 1195, 1992.
[59] G. A. Malygin, Phys. Solid State, 37, 1248, 1995.
[60] G. J. Fan, H. Choo, P. K. Liaw and E. J. Lavernia, Mater. Sci. Eng. A, 209, 248,
2005.
[61] J. Ye, B. Q. Han, Z. Lee, B. Ahn, S. R. Nutt, and J. M. Schoenung, Scr. Mater., vol.
53, no. 5, pp. 481–486, Sep. 2005.
[62] A. Marc and A. Meyers, “Mechanical Behavior of Materials”. 1999: Prentice-Hall,
Inc.
[63] Y. Zhang, T. D. Topping, H. Yang, E. J. Lavernia, J. M. Schoenung, and S. R. Nutt,
Metall. Mater. Trans. A, vol. 46, pp. 1196–1204, 2015.
[64] R. Valiev, Nature, 419, 887, 2002.
[65] R. Valiev, Nat. Mater., 3, 511, 2004.
[66] R. Z. V aliev and I. V Alexandrov, J. Mater. Res., vol. 17, no. 1, pp. 5–8, 2002.
[67] F. D. Torre, R. Lapovok, J. Sandlin, P.F. Thomson, C.H.J. Davies and E.V . Pereloma,
Acta Mater., 52, 4819, 2004.
53

[68] P. Kumar, C. Xu, and T.G. Langdon, J. Mater. Sci., 44, 3913, 2009.
[69] T. Mungole, P. Kumar, M. Kawasaki, and T. G. Langdon, J. Mater. Res., vol. 29, no.
21, pp. 2534–2546, Oct. 2014.
[70] T. Niendrof, J. Dadda, D. Canadinc, H.J. Maier and I. Karaman, Mater. Sci. Eng. A,
517, 225, 2009.
[71] D. Canadinc, E. Biyiki, T. Niendorf and H.J. Maier, Adv. Eng. Mater., 13, 281, 2011.
[72] T. Hu, K. Ma, T. D. Topping, B. Saller, A. Yousefiani, J. M. Schoenung and E. J.
Lavernia, Scr. Mater., vol. 78–79, pp. 25–28, May 2014.
[73] X. Zhang, Acta Mater. 52, 995, 2004.
[74] L. Lu, X. Chen, X. Huang, and K. Lu, Science, vol. 323, pp. 607–610, 2009.
[75] Z. Lee, D. B. Witkin, V. Radmilovic, E. J. Lavernia, and S. R. Nutt, Mater. Sci. Eng.
A, vol. 410–411, pp. 462–467, Nov. 2005.
[76] Y. H. Zhao, Y. Z. Guo, Q. Wei, T. D. Topping, A. M. Dangelewicz, Y. T. Zhu, T. G.
Langdon, and E. J. Lavernia, Mater. Sci. Eng. A, vol. 525, no. 1–2, pp. 68–77, Nov.
2009.
[77] Z. Lee, V. Radmilovic, B. Ahn, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans.
A, vol. 41, no. 4, pp. 795–801, Oct. 2009.
[78] B. Ahn, E. J. Lavernia, and S. R. Nutt, J. Mater. Sci., vol. 43, no. 23–24, pp. 7403–
7408, Sep. 2008.
[79] B. Ahn, PhD dissertation, University of Southern California, 2008
[80] J.R. Davis, “ASM Specialty Handbook: Aluminum and Aluminum Alloys”,
54

Materials Park, OH: ASM International (1993) p. 675.

55

Chapter 2
Experimental techniques

In this chapter, I will discuss the principles, applications and limitations of the
experimental techniques used in this dissertation. The experiments can be divided into
two general categories: microstructure characterization and mechanical testing which
include scanning electron microscope (SEM), electron backscatter diffraction (EBSD),
transmitted Kikuchi diffraction (TKD), energy dispersive X-ray spectroscopy (EDS),
transmission electron microscope (TEM), hardness measurement, and in situ
micro-tensile testing.

2.1 Characterization
2.1.1 Scanning electron microscope
Scanning electron microscope (SEM) is a versatile imaging tool with nanometer spatial
resolution. SEM scans the region of interest in a raster fashion using a focused electron
beam. As the electron beam raster scans from left to right and top to bottom, signals
excited at each point can be collected simultaneously using corresponding detectors. The
grey scale value of each step on the digital micrograph represents the signal intensity of
that spot on the surface. When the focused electron beam hits a surface, the sample can
generate a variety of signals in response, such as secondary electrons, backscattered
electrons, X-rays, Auger electrons and cathodoluminescence as shown in Figure 2.1.
56

Fig. 2.1 Signals generated by the primary beam of SEM

Among them, low energy secondary electrons (SEs) are the most commonly used
imaging signals for surface topography due to the edge effect. SEs are named “secondary”
because they are electrons knocked out of the sample by the primary electron beam.
Enhanced emission of secondary electrons can be seen from the edges or peaks (more
free surface area for SE to escape the sample) on the sample surface. Therefore, this edge
effect can generate a strong topographical contrast. Backscattered electrons (BSEs) are
primary electrons that are bounced back by the sample. The “bouncing” probability is
directly related to the atomic number of the target sample (how “heavy” they are). BSEs
are very sensitive to atomic number and thus can generate contrast from chemical
composition. BSEs can also generate diffraction patterns when certain diffraction
condition is met (see section 2.1.2). The energy of X-rays is measured to identify
57

elemental information on the sample surface as a spectroscopy technique called energy
dispersive X-ray spectroscopy (EDS) which will be briefly discussed in section 2.1.3.

SEM samples should be conductive otherwise the surface will be charged quickly by
incident primary electron beam (charging effect) which makes imaging impossible.
Non-conductive specimens can be either coated with a conductive layer (gold, carbon or
platinum) or imaged under low vacuum mode. In low vacuum mode, a small amount of
gas (e.g. nitrogen) is released in the vacuum chamber. These gas molecules are ionized
near the imaging surface due to the impact with high energy electron beam. The ions can
neutralize the surface to help reduce charging effect. However, a loss of spatial resolution
is expected using low vacuum mode since the electron beam is inevitably broadened by
the gas molecules.

SEM sample preparation is sometimes required to minimize the topography contrast and
thus highlight the defects, voids or phase contrast. However, little contrast can be
achieved from crystal orientation (BSEs can generate crystalline contrast to some extent
but not enough to extract orientation information). Therefore, we must seek help from
other characterization methods such as EBSD and electron diffraction in TEM.

58

2.1.2 Electron backscatter diffraction
Electron backscatter diffraction (EBSD) is a common SEM-based tool which can be used
to index and measure crystalline orientation, grain size, GB misorientation, texture,
coincident site lattice (CSL), phase and semi-quantitative strain analysis. EBSD can
detect a wide range of crystalline materials including but not limited to metals, alloys,
ceramics, semiconductors and minerals. Therefore, EBSD has many applications in
metallurgy, aerospace, automotive, microelectronics and earth science.

The schematic setup of an EBSD system is illustrated in Figure 2.2. During an EBSD
scan, a focused electron beam strikes a tilted sample (usually ~70
o
) and the electrons
scattered by lattice is a cone-shape diffraction which intersects with a phosphor screen to
form a pattern called Kikuchi band. This Kikuchi pattern can be used to identify the
crystalline orientation of the surface. The fluoresced pattern is then sent to a CCD camera
connected to a computer. The Kikuchi pattern generated from the surface area is directly
determined by the crystalline orientation of that spot. The software analyzes and indexes
the pattern via Hough transformation which transforms lines into points. As the electron
beam scans through the area of interest, a complete crystalline orientation map or inverse
pole figure is achieved from which a bunch of information (mentioned above) can be
extracted. The grain orientation is usually indexed by colors as shown in Figure 2.3.
Black areas in Figure 2.3 are unindexed points due to either impurities or local strain.

59

In this work, EBSD was carried out in a JSM-7001 field emission SEM (JEOL Inc.)
equipped with Hikari detector (EDAX Inc.). When scanning UFG AA, low acceleration
voltage (<10 kV) and low probe current were often selected to reduce the energy of
incident electron beam because higher acceleration voltage allows electrons to penetrate
deeper into the sample and generate multiple overlapping Kikuchi patterns which makes
pattern indexing impossible. Low pattern intensity due to low voltage can be
compensated by increasing the exposure (dwell time) or the average frame number to
minimize noise level at the expense of long scanning time (~hours). However, long
scanning time is usually not desired due to possible sample drift and cost consideration.
One has to play with different combinations of acceleration voltage, probe current and
scanning time to find out the best EBSD parameters for a specific application.

60

Fig. 2.2 Schematic of an EBSD system inside a SEM vacuum chamber

Fig. 2.3 EBSD crystal orientation map (inverse pole figure) of a coarse grain aluminum
alloy

61

Sample preparation is very critical to Kikuchi pattern quality. An ideal EBSD sample
should have a smooth and clean surface at sub-micron level. Both abrasive polishing and
ion polishing are commonly used for EBSD sample preparation. The abrasive polishing
often involves pre-grinding and coarse polishing using fine SiC papers followed by
vibratory polishing using colloidal silica nanoparticle for 1 hour. The abrasive polishing
is capable of polishing large surface area (several centimeters by centimeters) but subject
to introducing surface contamination and strain due to its wet and abrasive nature. An
alternative method is to use the ion beam to physically mill the target surface. The ion
polishing offers the best polishing quality which is damage-free to sample’s original
structure. But the polished area is usually restricted to several millimeters by several
millimeters. The ion polishing can be fully automated once the sample is correctly
aligned with the ion beam which will be discussed in section 2.1.6. Both abrasive and ion
polishing were used in the following chapters to prepare large in situ tensile dog-bone
samples and small EBSD specimens.

2.1.3 Transmitted Kikuchi diffraction
Transmitted Kikuchi diffraction (TKD) is a modified EBSD technique for small grain
size measurement. Conventional EBSD has a spatial resolution of ~ 50 nm which is
largely limited by the interaction volume of the electron beam. Although the spot size of a
modern field emission SEM is very small (~ several nanometers), the Kikuchi pattern can
be generated from a much larger volume since the electron beam can scatter deep into the
62

bulk SEM sample and have a interaction volume that is much larger than the beam spot
size. Each signal has its characteristic interaction volume depending the type and energy
of the signal. As shown in Figure 2.4, the interaction volume of BSEs can extend to
several hundred nanometers deep in the bulk specimen. Therefore, it is possible to collect
multiple Kikuchi patterns from very small grains within the interaction volume. The
overlapping patterns can deteriorate the indexing confidence and thus lead to unreliable
EBSD data that can be just random noise. To overcome this physical limitation on EBSD
resolution, Keller and Geiss proposed using thin foil specimen (~ 100 nm) instead of bulk
sample to reduce the electron-specimen interaction [1]. Keller and Geiss named this
EBSD variant the “transmission EBSD” (t-EBSD) due to the fact that the Kikuchi
patterns are formed by the forward-scattered transmitted electrons as in a transmission
electron microscope. The name “t-EBSD” was later changed to transmitted Kikuchi
diffraction (TKD) since the term “EBSD” already indicates a backscatter process [2,3].
To avoid confusion among readers, we will use transmitted Kikuchi diffraction (TKD) in
the following context. In TKD, the thin foil specimen does not give electron beam
opportunity to broaden within the material and thus can improve the spatial resolution as
illustrated in Figure 2.5. The experimental setup of TKD in Figure 2.6 and more detail
can be found in Chapter 5. One drawback of TKD is that thin foil specimen generates low
pattern intensity which often leads to long scanning time. In addition, making thin foil
specimen is sometimes not trivial. In this work, the thin foil specimens were prepared
using focused ion beam which will be discussed in section 2.1.6.
63

Fig. 2.4 Illustration of the interaction volume of the SEM signals [4]

Fig. 2.5 Schematic of the conventional EBSD process (left) and the thin foil TKD (right)
[1]

64

Fig. 2.6 Schematic of the TKD experimental setup for a thin foil specimen

2.1.4 Energy dispersive X-ray spectroscopy
Energy dispersive X-ray spectroscopy (EDS or EDX) is another SEM-based technique
that can identify elemental composition in seconds. When an electron beam interacts with
the specimen, electron-beam-excited X-ray emits in all directions. A silicon drift detector
(SSD) measures the energy of a captured X-ray photon by the number of ionization
events triggered by this photon. The generated charged particles “drift” to a collection
electrode by a transversal field. Then the current is amplified and converted to digital
signals. The modern EDS detector has a very high count rate of ~ 100k counts per
seconds and an energy resolution of ~ 0.2 KeV. An EDS spectrum can be gathered in a
few seconds which is very fast compared to other X-ray techniques such as X-ray
diffraction (XRD) and wavelength dispersive X-ray spectroscopy (WDS). A typical EDS
65

spectrum is shown in Figure 2.7. The spectrum is composed of two types of X-rays: the
continuous X-ray and the characteristic X-ray. The continuous X-ray which forms the
background in Figure 2.7 is caused by deceleration of electrons which are deflected by
nucleus of the specimen. Therefore, the continuous X-ray is also called Bremsstrahlung
(German word for “braking”) X-ray radiation. The elemental information is hidden in the
characteristic X-rays which are represented by peaks in Figure 2.7. When a valence
electron is knocked out by the incident electron beam, another electron from the outer
shell will quickly fill the hole left behind by the valence electron and release energy in
the form of X-ray. The energy of the X-ray photon is characteristic of the electron
structure of the target element. Therefore, it can be used to determine the elements on the
surface. Based on the relative intensity of the characteristic peaks, relative atomic
percentage of each selected element can be calculated. Because EDS uses focused
electron beam as excitation radiation, it has excellent spatial resolution compared to other
spectrum techniques. One can easily pin point the elemental information of a small area
of 1 x 1 µm. The spatial resolution of SEM-EDS is again limited by the
electron-specimen interaction volume. Theoretically, the lowest element EDS can detect
is lithium because hydrogen and helium only have one atomic orbital. In practice, the
lowest element that a windowless EDS detector can identify is boron.

EDS is a fast and convenient technique for elemental detection. However, EDS is not
efficient at detecting low Z elements due to low X-ray emission. In addition, EDS is not
66

considered as a quantitative tool due to the error caused by X-ray absorption by
neighboring atoms. Although quantitative EDS is possible using a reference sample with
known composition, most EDS data is semi-quantitative based on correction model that
is saved in the built-in analysis software. In addition, one should take extra caution when
calculating carbon or oxygen content using EDS. It can often lead to huge error due to the
presence of hydrocarbon contamination either in the SEM chamber or on the sample.
Another limitation that EDS has is relatively low energy resolution which will cause peak
overlap for elements with close atomic number. Other quantitative techniques such as
wavelength dispersive X-ray spectroscopy (WDS), X-ray photoelectron spectroscopy
(XPS) and X-ray diffraction (XRD) should be used as complementary tools to EDS.

67

Fig. 2.7 A typical EDS spectrum of a mineral showing characteristic X-ray and
continuous X-ray

2.1.5 Transmission electron microscope
A transmission electron microscope (TEM) is not only a high resolution electron
microscope but also a versatile analytical tool widely used in physical science and life
science. TEM allows microstructure examination, crystal structure investigation, and
chemical composition analysis via high resolution imaging, X-ray spectroscopy and
electron energy loss analysis. Unlike the SEM (section 2.1.1), a TEM image is generally
formed using electrons that transmit though a thin foil specimen. Therefore, a good TEM
specimen should have a thickness less than 100 nm to achieve electron transparency.
Thanks to the high energy electron beam and thin specimens, TEM is able to deliver a
68

stunning spatial resolution of ~ several angstroms (~ 10
-10
m) [5]. Columns of atoms can
be readily resolved when the defocus, sample tilt and scope alignment are optimized.

Figure 2.8 shows the picture and schematic of the TEM (JEOL 2100F) used in this study.
The grain structure is imaged using diffraction contrast. When a parallel electron beam
hits the thin foil crystals, the lattice structure is able to diffract the beam according to
Bragg’s diffraction law [5]. The transmitted electrons (unscattered) is used for bright field
imaging. Objective aperture can be inserted to increase the contrast. Dark field imaging
can be achieved if we only use diffraction beam to form an image when the aperture is
moved on one of the diffraction spots and blocking the primary transmitted beam. In this
work, both TEM bright field and dark field were used to investigate the microstructure
and measure the grain size of highly deformed aluminum samples processed by SPD
(section 1.2) since the nano-sized grains and internal strains in these SPD samples make
SEM and EBSD analysis almost impossible. In addition to diffraction contrast, it is not
uncommon to use thickness-mass contrast, phase contrast and Z (atomic number) contrast
[6]. Thickness-mass contrast rises due to thickness variation across the thin foil specimen.
Thick and dense area will cause more absorption and diffraction of the electron beam
which leads to dark area in the image. Phase contrast stems from the interference between
transmitted and diffracted beams through a crystalline sample. Lattice fringe can be
obtained when transmitted and diffracted beams are coherent. The phase contrast is very
useful in interface study and defects analysis (e.g. dislocation, impurities and stacking
69

fault). Coherent beam source (field emission gun) and thin specimens are usually
required to get phase contrast. Z contrast comes in handy when studying the chemical
composition. Z contrast can be obtained using high angle annular dark field (HAADF)
imaging mode. HAADF is accompanied with the scanning TEM (STEM) mode during
which the beam is focused to a fine spot (similar to SEM) instead of using a parallel
beam. As the focused beam scans through the region of interest, the transmitted beam will
be used to generate bright field STEM image and diffracted beam will be used for annular
dark field (ADF) or high annular dark field imaging (HAADF) which is very sensitive to
the atomic number of the target material. The transmitted beam can be also used to
analyze the energy loss of the electrons which is characteristic of the elements and
chemical states in the specimen. The elastically and inelastically scattered electrons can
be separated when the beam goes through a magnetic field (energy filter). The electrons
which pass the energy filter forms an energy loss spectrum which can be used to identify
chemical states of each element existing in the specimen. This technique is called
electron energy loss spectroscopy (EELS) which was used in this work to identify
precipitates such as oxide, nitride and β phase (Al
3
Mg
2
) at the grain boundaries.

70

Fig. 2.8 Picture and schematic of a JEOL field emission TEM (2100F) equipped with
Gatan Orius and image filter camera

2.1.6 Advanced sample preparation
Cross section ion polishing
Most SEM characterization requires a clean and smooth sample surface to reveal the
intrinsic microstructure other than the surface roughness. In addition, a defect-free and
flat surface is a strict requirement in order to get reliable EDS and EBSD data. Traditional
abrasive/mechanical polishing requires extensive labor and is likely to introduce
polishing defects. Delicate structure is subject to destruction using abrasive polishing. An
alternative method is to use ion beam to physically sputter away the surface material. In
71

this work, most clean sample surfaces for SEM imaging, EDS and EBSD collection were
prepared using the ion polishing. Figure 2.9 shows the picture the cross-section ion
polisher (IB-09010CP, JEOL Inc.) used in this work. The ion polisher is equipped with an
argon ion beam source and a visible light microscope to help align the sample with the
ion beam. The ion beam is not perpendicular to the surface of interest. Instead, the beam
is parallel to the surface and “slicing” the target cross section which gives its name
“cross-section” polisher. The schematic of the polishing process is illustrated in Figure
2.10. The Ni alloy shield plate covering the top of the sample is a protection against ion
beam to ensure a sharp “cut” on the cross section. The sample stage will swing during
polishing to obtain a uniform surface milling. Figure 2.11 shows the cross section of a
ceramic particle prepared by abrasive polishing and ion polishing. Fine void structure can
not be seen via abrasive polishing due to the surface roughness. In contrast, ion polishing
produced a clean and defect-free cross section which clearly exhibits the fine structure of
interconnected pores. In this work, aluminum surface is ion polished for EBSD (section
2.1.2), EDS (section 2.1.4), and SEM microstructure imaging.
72

Fig. 2.9 Cross-section ion polisher (IB-09010CP, JEOL Inc.), image courtesy by JEOL
USA

Fig. 2.10 Schematic of the cross-section ion polishing

73

Fig. 2.11 Surfaces of a ceramic particle prepared by (a) abrasive polishing and (b) ion
polishing

Focused ion beam lift-out
Focused ion beam (FIB) is a popular tool widely used in material research and
semiconductor industry. Most FIBs are usually equipped with a dual beam system
utilizing both electron and ion beams. The electron beam and ion beam intersect at an
angle of 52
o
as shown in Fig 2.12. The additional ion beam is capable of milling, etching
and depositing at micron scale, making FIB a “nano-machining” tool.

The most common application of FIB is TEM thin foil specimen preparation. This
process is often called in situ lift-out because the TEM specimen is lifted out from the
base material and the entire process is performed in situ inside the FIB chamber. During
lift-out, a thin slice of a cross section piece is separated from the base material by making
trenches around the area of interest. The isolated slice is then “glued” to a probe or a
needle which can be precisely manipulated at a speed of 100 nm/s. Finally, the probe
transfers the thin slice onto a specialized TEM grid for further thinning of the specimen.
74

The lift-out specimen will be TEM-ready when out of the FIB chamber. A typical lift-out
specimen has an electron-transparent area of ~ 10 x 10 µm and less than 100 nm in
thickness. One big advantage of the FIB lift-out process is that it allows a site-specific
TEM specimen preparation which is ideal for materials with heterogeneous structure.
However, FIB can easily cause beam damage and create amorphous layer on the
specimen. Therefore, low energy ion beam is usually used for a final surface cleaning to
reduce the beam damage.

Fig. 2.12 Schematic of the FIB dual beam system

In this work, lift-out was performed to make thin foil specimens for TKD and TEM using
a dual beam FIB system (JIB-4500, JEOL Inc.). More details will be discussed in Chapter
5 and 6.

75

2.2 Mechanical testing
2.2.1 Vickers microhardness measurement
Vickers hardness test is an easy and reliable technique for measuring the hardness of a
material [7]. Figure 2.13 shows the Vickers tester (FM-1e Vickers indenter) used in this
work.

Fig. 2.13 Picture of the Vickers microhardness tester

In a hardness test, a diamond indent tip is stamped on a polished surface under a constant
load and leaves am indent of pyramid shape. The Vickers hardness value H
v
is easily
defined by Equation 6.
76

𝐻𝐻 𝑣𝑣 =
𝐹𝐹 𝐴𝐴 (Eq. 6)
Where F is the applied load (in the unit of kilogram-force) during indentation and A is the
area (in unit of square millimeters) of the indent caused by the diamond tip when the load
is retracted. In practice, load F is a parameter the user can freely select. Therefore,
measuring H
v
becomes a problem of measuring the indent area. Due to the fixed geometry
of the diamond tip, the area A can be calculated via the length of two diagonals using an
empirical equation (Equation 7).
𝐴𝐴 ≈
𝑑𝑑 2
1. 8 5 4 4
(Eq. 7)
Where d is the average of two diagonals ( 𝑑𝑑 =
𝑑𝑑 1
+ 𝑑𝑑 2
2
) of the indent in the unit of
millimeters as shown in Figure 2.14.

Fig. 2.14 Diagonal measurement of a Vickers indent on an Al-Mg alloy

In Chapter 6, Vickers hardness maps of the HPT-processed Al disks were measured to
77

track the microstructure evolution as a function of distance to the disk center. A matrix of
indents with a spacing of 300 µm was made in one quarter of the HPT disk as shown in
Figure 2.15. The spacing is selected so that each indent will not be influenced by the
plastic zone of the previous one [7]. The size of each indent is ~ 50 µm which can be
larger than the CG regions in some heavily deformed samples. Therefore, readers should
keep in mind that the hardness map is an average of local hardness values which does not
represents the real structure of the trimodal samples. More details and results can be found
in Chapter 6.

Fig. 2.15 Visible light micrograph of the matrix of Vickers hardness indents with a
spacing of 300 μm

78

2.2.2 Instrumented nanoindentation
Conventional indentation techniques have difficulties in measuring mechanical properties
on thin film samples or samples with fine heterogeneous structures because of the large
indentation tip and difficulty of accurate indent size measurement. Nanoindentation
technique developed by Oliver and Pharr [8] provides a solution for measuring
mechanical properties locally at nano-scale. Unlike Vickers hardness test (section 2.2.1),
the indent made by a nanoindenter is only in the order of sub-micrometer which is very
difficult to measure directly using optical microscopes. Instead, the area of indents is
automatically calculated using an empirical function of the tip displacement. Modern
instrumented nanoindentation technique is capable of acquiring multiple mechanical
properties such as elastic modulus, loss modulus, hardness and yield strength through a
single indentation.

One common diamond tip used in nanoindentation is the Berkovich tip which is a
four-face pyramid. The shape of the indent left by Berkovich tip is a triangle with equal
sides as shown in Figure 2. 16. In a nanoindentation test, upon contacting the sample
surface, the load applied on the tip increases at a constant rate until it reaches a preset
maximum value. Then the maximum load is held for several seconds before unloading.
Figure 2.17 shows a typical load-displacement curve of a nanodentation test where P
max
is
the maximum load, S is the calculated stiffness, h
max
is the maximum tip displacement
into the sample and h
f
is the depth of the indent when load is retracted. The stiffness is
79

calculated from the slope of the unloading line since the unloading curve only involves
elastic recovery. Detail analysis of hardness and stiffness measurement can be found in
reference [8] and [9].

Fig. 2.16 (a) SEM image of an indent by a Berkovich tip, (b) Geometry of a Berkovich
tip [10]

Fig. 2.17 The load-displacement curve of a nanoindentation test [9]

80

In this work, the hardness values of the CG region and UFG matrix were measured
independently in a trimodal sample by indenting each small region using nanoindentation.
The sample surface has a smooth finish to eliminate errors caused by surface roughness.
Moreover, sample should be mounted carefully to make sure the loading direction is
parallel to the surface normal. More discussion about nanoindentation can be found in
Chapter 4.

2.2.3 In situ micro-tensile testing
Understanding the deformation and failure mechanisms is critical to improving the
macroscopic mechanical performance of a material. The capability of “seeing”
deformation and failure of a material at small scale has been attractive to the materials
science community for decades. Many in situ mechanical testing techniques have been
developed to study material deformation and fracture at various scales. For example, an in
situ TEM straining stage was introduced in the late 1950s [11] to study the behavior of
dislocation at nano-scale and its interaction with other defects such as twins, GBs and
impurities. In addition, SEM-based in situ mechanical testing is more suitable for studying
the material behavior under straining at micron-scale [12]. In this work, the deformation
and failure behavior of the cryomilled Al was investigated under the SEM using a
small-form-factor tensile stage. Local strain measurement becomes possible at submicron
resolution when the in situ tensile testing is combined with the digital image correlation
81

(DIC) technique. This section will introduce an in situ micro-tensile setup inside a SEM
vacuum chamber and the local strain measurement using DIC.

A micro-tensile tester (see Figure 3.3) equipped with a custom SEM stage was used for an
in situ tensile test carried out in a SEM chamber. The micro-tensile stage is ideal for
non-standard samples with limited geometries. The compact size of the stage allows it to
be placed in a SEM vacuum chamber. However, the small size does not compromise the
performance. The stage can provide a maximum load up to 5 kN and has a slow strain rate
of ~ 10
-4
s
-1
which is ideal for quasi-static condition. The dimensions of the miniature
dog-bone tensile specimens are shown in Figure 2.18.

Fig. 2.18 Dimensions of the miniature dog-bone specimens used for in situ tensile testing
(left) and small HPT disk samples (Images not scaled)

82

During the tensile test, the dog-bone specimen is grabbed by two clamps tightened by the
screws shown in Figure 2.19 (a). The friction force on the contact area between the
specimen and the grip provides tensile force up to several hundred N. However, grip
slipping often occurs due to insufficient friction force especially for small HPT specimens
with small contact area. The slipping leads to inaccurate stress-strain curves with incorrect
elastic modulus and overestimated strain. Custom grips were designed to prevent slipping
and minimize errors. Figure 2.19 (b) shows a custom grip with extra flanges which can
stop the slipping and support tensile force in addition to friction.

Fig. 2.19 (a) Illustration of specimen gripping using two types of fixtures, (b) the custom
grip with extra flanges to prevent slipping

83

Readers should be aware that this micro-tensile setup uses non-standard testing condition
and custom sample sizes. Therefore, comparing the tensile results using the micro-tensile
test with other standard tests is meaningless and misleading. In this work, the comparison
between samples was made only when the testing condition and the sample geometries are
identical.

84

2.2.4 Digital image correlation
Digital image correlation (DIC) is a non-contacting and highly adaptable technique for
in-plane strain field measurement. DIC can be used at different scales ranging from large
infrastructures such as houses and bridges [13], to micro-structures of alloys and
composites [14,15].

The principle of the correlation method involves automated tracking of pattern change
between two images which can be taken using regular digital camera, light microscope or
SEM depending on the feature size of interest [16]. The pattern tracking procedure is
schematically described in Figure 2.20 where S and S
1
represent an un-deformed state and
a deformed state, respectively.

85

Fig. 2.20 Illustration of the strain measurement using DIC from the initial S to the
deformed state S
1
[16]

If M and N (un-deformed state) are described by (x
m
,y
m
) and (x
n
,y
n
), then the coordinates
of M
1
and N
1
can be expressed by a first order Taylor expansion:

𝑥𝑥 𝑛𝑛 1
= 𝑥𝑥 𝑚𝑚 + 𝑢𝑢 𝑚𝑚 + �1+
𝜕𝜕𝜕𝜕
𝜕𝜕𝜕𝜕
�
𝑀𝑀 ∆ 𝑥𝑥 + �
𝜕𝜕𝜕𝜕
𝜕𝜕 𝑦𝑦 �
𝑀𝑀 ∆ 𝑦𝑦 (Eq. 8)
𝑦𝑦 𝑛𝑛 1
= 𝑦𝑦 𝑚𝑚 + 𝑣𝑣 𝑚𝑚 + �1+
𝜕𝜕 𝑣𝑣 𝜕𝜕 𝑦𝑦 �
𝑀𝑀 ∆ 𝑦𝑦 + �
𝜕𝜕 𝑣𝑣 𝜕𝜕𝜕𝜕
�
𝑀𝑀 ∆ 𝑥𝑥 (Eq. 9)

The definition of the correlation coefficient C is
𝐶𝐶 =
∑ [ 𝑓𝑓 ( 𝜕𝜕 𝑛𝑛 , 𝑦𝑦 𝑛𝑛 ) − 𝑓𝑓 𝑑𝑑 ( 𝜕𝜕 𝑛𝑛 1
, 𝑦𝑦 𝑛𝑛 1
)]
2
𝑁𝑁 ∈ 𝑆𝑆 ∑ 𝑓𝑓 ( 𝜕𝜕 𝑛𝑛 , 𝑦𝑦 𝑛𝑛 )
2
𝑁𝑁 ∈ 𝑆𝑆 (Eq. 10)
Where 𝑓𝑓 ( 𝑥𝑥 𝑛𝑛 , 𝑦𝑦 𝑛𝑛 ) and 𝑓𝑓 𝑑𝑑 ( 𝑥𝑥 𝑛𝑛 1
, 𝑦𝑦 𝑛𝑛 1
) are grey scale distribution of the un-deformed pattern
86

and the deformed pattern. Coefficient C is a number evaluating the correlation level of all
spot patterns in the area S (called the subset). In other words, the DIC algorithm correlates
and tracks the surface pattern by mathematically seeking a minimal coefficient C within
the subset. When the pattern is successfully correlated before and after the deformation,
the strain can be calculated by tracking the relative position of spot patterns before and
after deformation.

The accuracy of DIC significantly relies on the pattern quality. The ideal spot size should
be 2 or 3 pixels for a given image [17]. For example, if we are measuring the deformation
of a house, the speckle we paint on the wall should be several centimeters in diameter. As
we move to a smaller field of view, For instance, to measure the micro-strain among
ultra-fine grains, the field of view we are looking at is just 20 by 20 um using SEM. Table
2.1 shows the optimal feature size for the corresponding magnification in the
field-emission SEM used in this study. At such small scale, nano-sized speckle patterns
are required for accurate strain measurement. Therefore making a fine pattern is the major
challenge. In this work, the surface will be patterned using e-beam lithography or silica
nanoparticle or simply using the intrinsic patterns of the trimodal samples. The
experimental detail of submicron DIC will be discussed in Chapter 3.

87

Table 2.1 Optimal pattern size and step size required for DIC using JEOL 7001F SEM
Magnification Optimal pattern size (nm) Area step size (nm)
3,000x 97 19
5,000x 57 11
6,000x 48 10
7,000x 41 8
8,000x 35 7
10,000x 28 6

88

References
[1] R. R. Keller and R. H. Geiss, J. Microsc., vol. 245, no. July 2011, pp. 245–251,
2012.
[2] P. W. Trimby, Ultramicroscopy, vol. 120, pp. 16–24, Sep. 2012.
[3] P. W. Trimby, Y. Cao, Z. Chen, S. Han, K. J. Hemker, J. Lian, X. Liao, P. Rottmann,
S. Samudrala, J. Sun, J. T. Wang, J. Wheeler, and J. M. Cairney, Acta Mater., vol. 62,
pp. 69–80, Jan. 2014.
[4] Leng Y. “Materials characterization: Introduction to microscopic and spectroscopic
methods”, Singapore: Wiley, 2008.
[5] D. B. Williams and C. Barry Carter, “Transmission electron microscopy”, New York,
Plenum, 1996.
[6] R. S. Rai and S. Subramanian, Prog. Cryst. Growth Charact. Mater., vol. 55, no. 3–4,
pp. 63–97, 2009.
[7] A. E. Giannakopoulos, P. L. Larsson, and R. Vestergaard, Int. J. Solids Struct., vol.
31, no. 19, pp. 2679–2708, 1994.
[8] W.C. Oliver and G.M. Pharr, J. Mater. Res. 7, 1564, 1992.
[9] G.M. Pharr, W.C. Oliver and F.R. Brotzen, J. Mater. Res. 7, 613, 1992.
[10] A.C. Ficher-Cripps, “Nanoindentation”, NY: Springer- Verlag, 2002.
[11] H.G.F. Wilsdorf, Rev. Sci. Instrum. 29 , 323, 1958.
[12] T. J. Turner, P. A. Shade, J. C. Schuren, and M. A. Groeber, Model. Simul. Mater. Sci.
Eng., vol. 21, no. 1, p. 015002, Jan. 2013.
89

[13] S. W. Kim and N. S. Kim, Procedia Eng., vol. 14, pp. 195–203, 2011.
[14] Y. Zhang, T. D. Topping, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans. A, vol.
45, no. 1, pp. 47–54, May 2013.
[15] Y. Zhang, T. D. Topping, H. Yang, E. J. Lavernia, J. M. Schoenung, and S. R. Nutt,
Metall. Mater. Trans. A, vol. 46, pp. 1196–1204, 2015.
[16] Y .H. Huang, C. Quan, C.J. Tay and L.J. Chen, Opt. Eng. 44(8) 087011-1, 2005.
[17] P. Zhou and K.E. Goodson, Opt. Eng., vol. 40, pp. 1613–20, 2001.

90

Chapter 3
Dynamic micro-strain analysis of cryomilled Al
alloys using digital image correlation

3.1 Motivation
Understanding deformation and fracture mechanisms is essential to improving
mechanical properties of new materials. Microscopic investigations of new materials
commonly rely on transmission electron microscopy (TEM), scanning electron
microscopy (SEM) and focused ion beam (FIB). In recent years, new characterization
techniques and methods have evolved to exploit the capabilities of these instruments. For
example, in situ SEM observation of micro-strain evolution has become an attractive
approach for studying the deformation mechanisms of advanced structural materials.
Likewise, the use of digital image correlation (DIC) has been extended to small-scale
deformation measurements. In this work, we focus on the direct observation of how
micro-strain evolves among the grains of an UFG Al-Mg alloy. To our best knowledge,
no study has been done on the in-situ micro-strain measurement of an UFG Al-Mg alloy
using digital image correlation at sub-micron level.

DIC [1,2] is a non-contact, adaptable metrology technique for in-plane or out-of-plane
strain field measurement that can be utilized on a variety of length scales ranging from
civil engineering structures [3] to micro-structures of metallic specimens [4,5]. The DIC
91

algorithm tracks a greyscale pattern on the deforming surface step-by-step in a small area
called a subset. To track the full-field surface deformation, an isotropic random speckle
pattern is required on the specimen surface. This speckle pattern can be either intrinsic
(from existing surface features) or extrinsic, as in a deposited pattern. The optimal feature
size of a speckle pattern is reportedly 2-3 pixels for a recorded image [6]. Therefore,
different patterning methods are needed to meet this requirement at different scales.
While there have been multiple investigations reporting the use of DIC in making
micro-scale strain measurements, one of the major challenges in such endeavors is
producing a nano-scale, random and isotropic speckle pattern required for DIC. In this
study, the area of interest is only about 20 by 20 µm which requires a nano-scale speckle
patterns for DIC analysis.

Various approaches have been employed to generate extrinsic DIC patterns on different
substrates at reduced scales. The most common patterning method for small-scale DIC
involves using a high quality airbrush to spray micro-scale paint patterns on a substrate
[7,8]. Another convenient approach is generating grid patterns using a grid mask [9,10].
Intrinsic patterns are also available when the surface of specimens exhibited small-scale
features [11]. Nanoparticles are often utilized for DIC patterns because of the small
sizes and low cost [12,13]. In addition, focused ion beam systems have been used to
deposit platinum dots in a slow and controllable process [13]. However, all of these
92

studies highlight certain common drawbacks of the patterning methods employed,
including lack of pattern density control, coarse feature size, non-random patterns and
substrate dependence.

In this work, e-beam lithography (EBL) was used to make nano-scale gold speckle
patterns on the aluminum surface. EBL was introduced by Sutton [19] as a DIC
patterning method. However, patterns from the early works were not small enough for a
DIC analysis among the ultrafine grains. In this study, EBL parameters were optimized to
achieve a 45nm gold dot pattern in an area of 20 by 20 µm. EBL is commonly used for IC
(integrated circuit) fabrication in the semiconductor industry and to produce nano-scale
functional devices. EBL utilizes a focused electron beam to generate patterns across a
resist-coated substrate. A polymer resist film is required as an electron beam resist layer.
One conventional positive resist material for EBL is polymethyl-methacrylate (PMMA).
As the highly focused electron beam scans across the resist following the pattern design,
the areas exposed by the beam can be subsequently dissolved in a developer solution,
while the remaining unexposed PMMA remains intact in the developing process.
Subsequently, the pattern on the resist layer can be transferred to the substrate via
post-developing processes such as direct etching and lift-off according to different
applications. Finally, the residual polymer resist material can be simply dissolved away
with acetone. The EBL patterning method is substrate-independent, repeatable and
designable. The feature size limit of EBL patterns is determined primarily by the
93

proximity effect attributed to electron forward- and back-scattering in the resist layer
[14].

After the specimen was patterned, tensile tests were performed in situ in an SEM using a
micro-tensile testing module. To reveal the effect of microstructure on deformation
evolution, DIC strain analysis was performed on deformed samples and later correlated
with a crystalline orientation map obtained using electron backscattered diffraction
(EBSD) technique.

3.2 Experimental procedures
3.2.1 Specimen preparation
The bulk specimens used in this investigation were consolidated from cryomilled AA
5083 (Al-4.4Mg-0.7Mn-0.15Cr wt %) powders produced by Valimet, Inc. (Stockton, CA).
Cryomilling was performed in liquid nitrogen for 8 hours using a modified Szegvari
attritor. Stainless steel milling balls were used with a 32:1 ball-to-powder weight ratio.
During the milling, 0.2 wt% ‘stearic acid’ (stearic acid: palmitic acid = 1:1) was added as
a process control agent (PCA). Because of the large scale of the project, cryomilling
was performed by a commercial entity (DWA Aluminum Composites, operating under the
supervision of UC Davis and Pratt & Whitney Rocketdyne, Inc., Canoga Park, CA).
After milling, the powder was transferred to a glove box in a liquid nitrogen medium,
94

ensuring that atmospheric contamination of the cryomilled powders was minimized.
Further details of cryomilling are available elsewhere. [15-17]

After cryomilling, the powders were canned and hot vacuum degassed to remove the
PCA and physi-adsorbed moisture prior to consolidation by quasi-isostatic (QI) forging
(Advanced Materials and Manufacturing Technologies, LLC, AM2T, Granite Bay, CA).
QI forging was conducted in two steps at 350 ºC, followed by hot rolling at 450 ºC in
multiple steps (Niagra Specialty Metals, Akron, NY). After rolling, the UFG plate
dimensions were ~19 mm thick and ~610 mm in diameter. Details pertaining to the
processing and subsequent properties of this plate have been reported elsewhere [18].

Miniature dog-bone tensile specimens were produced by electrical discharge machining
(EDM) and featured gauge (aligned with the rolling direction) dimensions of 4 mm
(length) x 3 mm (width) x 1.5 mm (thickness) as shown in Figure 3.1. A notch 200 µm
wide and 500 µm long was machined in the center of the gauge section to introduce stress
concentration. As load was applied, plastic deformation was initially confined to the
near-notch area, constraining the strain-localized area for DIC analysis.

As a reference point to conventional tensile data, a cylindrical ASTM E8M subsize
specimen [5] was tested in addition to the miniature dog-bone specimen used for DIC
analysis. The subsize specimen had nominal dimensions of ~15 mm gauge length × ~3
95

mm diameter.

Fig. 3.1 Geometry of a dog-bone tensile specimen and the notch area

3.2.2 Alignment of EBSD scanned area to DIC strain field
Grain structures were revealed using EBSD. During an EBSD scan, the diffraction
pattern generated was indexed by matching to a library of Kikuchi patterns. An EBSD
scan was conducted prior to EBL speckle patterning because the residual polymer layer
would block the scattered signal from the sample beneath it. Before scanning, the tensile
specimen was ground and polished through 240, 400, 600, 1200, 2400 and 4000 grit
abrasive papers and diamond suspensions (6 µm, 3 µm and 1 µm). Finally, the specimen
was fine polished using colloidal silica suspension on a vibratory polisher (Buehler) for 1
hour until the surface showed a mirror-like finish. To later match the region of interest
96

(ROI) with the EBSD scan region, reference marks were first deposited using EBL
patterning, as shown in Figure 3.2 (b). Direct observation of the ROI was not possible in
EBL machine because electrons exposure from imaging would cause unfavored patterns
on the ROI. A Cartesian coordinate on the sample surface was therefore necessary to
locate the ROI in EBL. Three additional reference marks were required for this
coordinate on the sample surface. These three marks were situated far from the ROI to
allow locating the area without exposing it to electrons.

An EBSD scan of the ROI was carried out at 15kV acceleration voltage, with a probe
current of 13 and a working distance of 15mm using JEOL JSM-7001F SEM with EDAX
TSL orientation image mapping system. An area of 20 × 20 µm was mapped using a step
size of 50 nm. The raw EBSD data was first processed using a grain confidence index (CI)
standardization procedure. A secondary process called single iteration grain dilation was
carried out to reduce un-indexed points and noise level. Following the EBSD scanning,
the speckle pattern was transferred onto the scanned area using EBL. Using these
reference marks, we were able to ensure that the EBSD scanned area matched the EBL
patterned area for DIC analysis.

97

3.2.3 E-beam lithography (EBL) patterning
The polished surface was first spin-coated with PMMA A4 at 5k RPM for 30 seconds.
Post-baking of the coated sample was carried out on a hot plate at 453K (180
o
C) for 2
minutes. To achieve a fine EBL pattern, the smallest aperture (7.5 µm) and a low
acceleration voltage (10 kV) were selected. Dosage test was performed to determine the
optimal dose factor. The pattern designed by the CAD software consists of a 30 by 30 µm
speckle pattern and an orientation mark as shown in Figure 3.2 (a). After this pattern was
transferred onto the PMMA layer, the sample was immediately developed by immersing
in IPA:MIBK (3:1) solution for 45 seconds. The exposed PMMA regions were dissolved
away during the developing process. After developing, the sample was rinsed in IPA for 1
minute and dried using compressed air. To harden the residual polymer layer, the sample
was again baked at 373K (100
o
C) for 2 minutes. A layer of gold 20 nm thick was then
deposited using a sputter coater. Finally, the residual PMMA was dissolved using
acetone.

Fig. 3.2 (a) A random speckle pattern with a central target mark; (b) the same pattern
after deformation; the arrows indicate the positions of reference marks
98

3.2.4 In situ micro-tensile testing
Micro-tensile tests were carried out in situ in an SEM vacuum chamber at room
temperature using the micro-tensile module shown in Figure 3.3. The dimensions of the
micro-tensile stage (146 mm x 108 mm x 39 mm) are small enough to be placed inside an
SEM chamber. Mounted inside an SEM vacuum chamber, the micro-tensile stage can be
controlled by a computer and a control unit via a feed-through. The micro-tensile module
was capable of achieving a maximum load of 5kN. The SEM used in this study (JEOL
JSM-6610) featured a tungsten filament, and image distortion was evaluated prior to the
actual tensile test. A series of SEM images were recorded prior to deformation of the
sample. DIC analysis was then conducted on these non-deformed images to determine the
pseudo-strain introduced by image distortion. This distortion effect was minimized to
0.01% by reducing specimen charging effect and by image averaging. Details of
distortion correction will be discussed later.

99

Fig. 3.3 Micro-tensile stage placed in SEM chamber

A strain rate of 4.18 x 10
-4
s
-1
was selected for a quasi-static tensile testing. During
micro-tensile tests, SEM images were recorded after the test was interrupted every five
seconds. After each stop, the sample was held for at least one minute to allow the load to
stabilize. To ensure each image was captured in an identical position, a target mark was
designed in the center of the speckle pattern as shown in Figure 3.2 (a). Each deformed
image was aligned by centering on the target at a higher magnification. Secondary
electron images were acquired, and the acceleration voltage and the working distance
were set to 15 kV and 12 mm respectively to minimize the charging effect.

100

3.2.5 In-plane micro-strain analysis
A series of deforming images were imported into the 2D-DIC software (VIC-2D,
Correlated Solutions Inc.), and an un-deformed image was used as a reference image.
Deformed images were then compared to the first reference image. All the SEM images
were recorded at the same magnification with a spatial resolution of 13.7 nm/pixel.
During deformation, the greyscale speckle pattern was tracked in a small window, termed
the subset. A subset size of 70 and step size of 1 were selected. The correlation process
initiated from the central target area. Finally, Lagrangian strains were calculated based on
the relative displacement of speckle patterns.

3.3 Results and discussions
3.3.1 EBL random speckle pattern for DIC
A gold speckle pattern was deposited on the sample surface using EBL, and dosage tests
were performed to determine the optimal dose level. The base dose level was 800
pAs/cm
2
. The total dose level was the base level multiplied by a dose factor. Figure 3.4
shows patterns produced with dose factors of 1, 2, 5 and 7, respectively. Dose factors of 1
and 2 both generated patterns with horizontal gold rods because the dose level was not
sufficient to make through holes to the aluminum substrate. On the other hand, dose
factors of 5 and 7 resulted in vertical gold columns - ideal for the DIC speckle pattern. To
minimize the exposure time, a dose factor of 5 was selected. These gold columns had an
average height of 20 nm, ensuring that good contrast could be achieved via secondary
101

electron imaging (SEI). Because SEI does not rely on atomic number contrast, this DIC
pattering method is generally substrate-independent.

Fig. 3.4 EBL patterns with exposure (a) dose factor 1, (b) dose factor 2, (c) dose factor 5,
and (d) dose factor 7

The diameter of the gold columns is ~45 nm, which is close to the optimal feature size
(40 nm) for a 20 by 20 µm field of view [6]. To evaluate the pattern quality, a pattern
intensity histogram was constructed, as shown in Figure 3.5. This greyscale (from 0 to
255) histogram is plotted against the numbers of pixels. As shown in Figure 3.5, a
bell-shaped distribution of pattern intensity was observed. Patterns with a bell-shaped
intensity distribution are ideal for accurate correlation results [8]. Because of the small
feature size and high contrast, the EBL pattern is well-suited to sub-micron DIC analysis
for the UFG Al-Mg alloy.

102

Fig. 3.5 Pixel greyscale histogram of the EBL speckle pattern

3.3.2 Intrinsic image distortion correction
The accuracy of the in situ DIC analysis depends on the quality of the speckle pattern and
the SEM images. SEM imaging quality is limited by the SEM resolution, specimen
contrast and the intrinsic image distortion. Although SEM provides sufficient resolution
and contrast for the current DIC analysis, image distortion can degrade the correlation
accuracy, especially at higher magnification. Pseudo-strain can be introduced due to
image distortion (as opposed to real surface deformation).

To evaluate the influence of image distortion, a series of images of the same region
before loading were recorded at different times. Pseudo-strain was calculated by
correlating to the reference image taken at t = 0 sec. A random distribution of
pseudo-strain was observed at five different time delays. This pseudo-strain was not
103

spatial but temporal distortion in nature. The average value of the pseudo-strain
introduced by image distortion was plotted in Figure 3.6. The average value of the
pseudo-strain was independent of the time lapse. The effect of image distortion can be
minimized using an image integration process that averages image data from multiple
scans [19]. In this study, the scan speed was set to 20 seconds per scan and the averaging
number of 2 was selected. Therefore, two images were integrated together with a total
scan time of 40 seconds. Compared to a single 40-second scan, the integrated image has a
higher signal-to-noise ratio and a more homogeneous pseudo-strain distortion instead of
step changes in x and y directions [19]. As shown in Figure 3.6, the average and the
maximum value of this pseudo-strain for the SEM used (JEOL JSM-6610) is on the order
of 0.01% and 0.5%, respectively.

Fig. 3.6 Pseudo-strain introduced by SEM image distortion

104

3.3.3 Micro-strain measurement of ultrafine-grained Al-Mg alloy
Micro-tensile testing results
Micro-tensile tests were performed at room temperature using a constant strain rate of
4.18x10
-4
s
-1
. The engineering stress-strain curve obtained from a micro-tensile stage is
shown in Figure 3.7. The engineering stress was calculated using the applied load divided
by the cross section of the gauge length (1.5 ×3 mm). The engineering strain was defined
by the change of the gauge length divided by the original gauge length (4 mm). The
micro-serrations occurred when the test was interrupted for imaging. The sudden drop of
stress followed by gradual increase at each serration was caused by dislocation relaxation
processes [20]. The mechanical properties of the UFG Al-Mg alloy, including Young’s
modulus, 0.2% offset yield strength, ultimate tensile strength (UTS) and elongation, are
tabulated in Table 3.1. As a comparison, tensile data extracted from a standard tensile test
(ASTM E8M) are also listed in Table 3.1.

Fig. 3.7 Engineering stress and strain curve for the UFG Al-Mg alloy
105

Table 3.1 Mechanical properties of the UFG Al-Mg alloy
Young’s
modulus
(GPa)
Yield strength
σ
y
(MPa)
UTS
(MPa)
Elongation to
failure
UFG Al-Mg
alloy
(microtensile)
63 397 503 14.6%
UFG Al-Mg
alloy
(ASTM E8M)
70 370 450 16.9%

In-plane micro-strain evolution among grains
The patterned area was located along one of the shear bands near the notch where the
shear strain was initially localized due to the stress concentration. Figure 3.2 (b) shows
the patterned area after plastic deformation. Because the patterned area was located along
the shear band, the shear strain e
xy
was the major strain component during deformation.

The EBSD inverse pole figure map of the UFG Al-Mg alloy is shown in Figure 3.8. An
average confidence index (CI) of 0.65 was achieved, indicating that above 90% of the
data points are indexed correctly. The UFG Al-Mg alloy has an average grain size of 560
nm with a fraction of 92.4% high-angle grain boundaries (15
o
-180
o
misorientation) and
7.6% low-angle grain boundaries (5
o
-15° misorientation). The black regions in Figure 3.8
are regions in which the confidence index was less than 0.1. Low values of confidence
index can be caused by uneven surfaces, voids, highly deformed grains, or
non-crystalline phase. The black areas at the four corners are caused by the reference
106

marks which are covered with a layer of polymer. The large low-CI area in the top part of
the map is due to the presence of voids. The rest of the low-CI regions are attributed to
uneven areas or residual polishing debris deposited during the polishing process.

Fig. 3.8 EBSD inverse pole figure map for the UFG Al-Mg alloy; Black and white
boundaries indicate high angle grain boundaries and low angle grain boundaries,
respectively

Finally, the EBSD inverse pole figure was overlaid on the correlated strain map by
matching the four corner reference marks and the orientation mark. The micro-strain
evolution (Figure 3.9) was recorded from 180 - 240 seconds of the micro-tensile test,
during which the macro-strain increased from 9.3% to 12.1%.

As evident from Figure 3.9, the micro-strain did not initiate uniformly over the entire area.
The strain was localized within larger grains and extended across grain boundaries into
107

adjacent smaller grains. This observation is consistent with the bimodal deformation
model proposed by Lee et al [21] who claimed that larger grains were likely to exhibit
larger plastic deformation than fine grains, and also reported that micro-voids and cracks
nucleated at interfaces of coarse and fine grains due to localized deformation. This
phenomenon was first observed using a light microscope at a much lower magnification
and resolution [22]. The EBL nano-scale patterning method in the present work affords
opportunity to observe the micro-strain evolution within individual grains of the UFG
Al-Mg alloy. However, micro-voids and cracks were not observed because of the polymer
layer covering the patterned area.

108

Fig. 3.9 Engineering shear strain e
xy
overlapped with grain structures as a function of
deformation time: (a) 180 seconds, (b) 190 seconds, (c) 200 seconds, (d) 215 seconds, (e)
225 seconds, and (f) 240 seconds

109

Fig. 3.10 The mean engineering shear strain evolution

The average micro-strain is plotted as a function of time in Figure 3.10. The average
value of the micro-strain saturated at about 220 seconds because a major crack developed
and propagated along the shear band on the opposite side of the notch root.

Note that the combined EBSD and DIC metrology described here is valid only when the
micro-strain is relatively low (less than 10%). Larger deformations can limit the use of
EBSD for the reasons that (1) distorted crystalline structure generates unrecognizable
diffraction patterns and (2) large distortions of grain geometry hampers accurate
matching between the DIC micro-strain field and the EBSD inverse pole figure map.
Moreover, in situ EBSD is not always feasible due to space constraints within the SEM
vacuum chamber and the long processing time for a single EBSD scan. Alternative
methods will be required to reveal a deforming microstructure and perform in situ tensile
110

testing simultaneously.

3.4 Conclusions
In this work, the micro-strain evolution of an UFG Al-Mg alloy was measured for the
first time at nano-scaled grain structures. The EBL patterning technique described here is
essential to generate the requisite fine speckle patterns with a feature size of 45 nm. The
EBL patterning method is repeatable, designable, scalable to various sizes, and
substrate-independent. Using the combined EBSD and DIC techniques, effects of
micro-structural features can be investigated by overlaying the crystalline orientation map
with the DIC strain contour map. Using these techniques, micro-strain localization was
observed in large grains in an UFG Al-Mg alloy and extended across grain boundaries
into adjacent smaller grains, indicating that larger grains were likely to exhibit larger
plastic deformation than fine grains. Micro-voids and cracks formed at interfaces of
coarse and fine grains to maintain the discontinuity in strain field. The patterning method
in this work is suitable for the direct observation of deformation mechanisms in a wide
range of nano-scaled homogeneous and heterogeneous materials.

111

References
[1] W. H. Peters and W. F. Ranson, Opt. Eng., 21: 427-432, 1982.
[2] M. A. Sutton, W. J. Wolters, W. H. Peters, W. F. Ranson, and S. R. McNeill, Image
Vision Comput., 1(3): 1333-1339, 1983.
[3] S. Yoneyama, A. Kitagawa, S. Iwata, K. Tani, and H. Kikuta, Exp. Tech., 31: 34-40,
2007.
[4] M. A. Tschopp, B. B. Bartha, W. J. Porter, P. T. Murray, S. B. Fairchild, Metall Mater
Trans A, 40: 2363-2368, 2009.
[5] J. Kang, Y. Ososkov, J. D. Embury, D. S. Wilkinson, Scr. Mater., 56: 999-1002, 2007.
[6] P. Zhou and K. E. Goodson, Opt. Eng., 40: 1613-1620, 2001.
[7] N. K. Kar, Y. Hu, B. Ahn, and S. R. Nutt, Compos. Sci. Technol., vol 72 (11)
1283-1290, 2012.
[8] T. A. Berfield, J. K. Patel, R. G. Shimmin, P. V. Braun, J. Lambros, and N. R. Sottos,
Exp. Mech., 47: 51-62, 2007.
[9] H. Jin, W. Y . Lu, and J. Korellis, J. Strain Analysis, 43: 719-727, 2008.
[10] Y. Tanaka, K. Naito, S. Kishimoto, and Y. Kagawa, Nanotechnology, 22: 115704,
2011.
[11] B. Ahn and S. R. Nutt, Exp. Mech., 50: 117-123. 2010.
[12] H. Wang, H. Xie, Y. Li, and J. Zhu, Meas. Sci. Technol., 23: 035402, 2012.
[13] A. D. Kammers and S. Daly, Meas. Sci. Technol., 22: 125501, 2011.
[14] C. Vieu, F. Carcenac1, A. Pépin, Y. Chen, M. Mejias, A. Lebib, L. Manin-Ferlazzo, L.
112

Couraud, and Launois H, Appl. Surf. Sci., vol 164 (1–4): 111–117, 2000.
[15] A. P. Newbery, B. Ahn, T. D. Topping, P. S. Pao, S. R. Nutt, and E. J. Lavernia, J.
Mater. Process. Technol., 203(1-3): 37-45, 2008.
[16] T. D. Topping, B. Ahn, Y. Li, S. R. Nutt, and E. J. Lavernia, Metall. and Mater. Trans.
A, 43(2): 505-519, 2012.
[17] D. B. Witkin and E. J. Lavernia, Prog. Mater. Sci., 51(1): 1-60, 2006.
[18] T. D. Topping, Y. Li, E. J. Lavernia, K. Manigandan, and T. S. Srivatsan, Adv. Mater.
Res., 410, 175-186, 2011.
[19] M. A. Sutton, N. Li, D. C. Joy, A. P. Reynolds, and X. Li, Exp. Mech., 47: 775-787,
2007.
[20] A. Seeger, Mater. Sci. Eng. A, 370, 50-66, 2004.
[21] Z. Lee, V. Radmilovic, B. Ahn, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans. A,
vol 41A, 795-801, 2010.
[22] B. Ahn, E. J. Lavernia and S. R. Nutt, J. Mater. Sci., 43, 7403-7408, 2008.

113

Chapter 4
Microstructural evolution of trimodal Al alloys

4.1 Motivation
To achieve desired microstructure of the Al-based trimodal nanocomposite, it is important
to understand the effect of reinforcing ceramic particles on the microstructural evolution
as a function of milling time. The purpose of this study was investigating the mechanism
of forming NC Al / B
4
C composites during cryomilling.

4.2 Experimental procedures
Al 5083 powder (90 wt%) and B
4
C particles (10 wt%) were blended during cryomilling
under liquid nitrogen temperature. Cryomilling was carried out at a temperature of –180
ºC using a Szegvari attritor with a ball-to-powder ratio of 32 : 1 and a rotation speed of
180 rpm. As always, 0.2 wt% stearic acid was added into the milling chamber as a
process control agent (PCA) to prevent the severe adhesion of the aluminum onto the
chamber and milling balls. Powders milled with different periods of time were retrieved
in order to study the microstructural evolution as a function of milling time.

The average grain size of each batch of powder was measured by XRD technique based
on the X-ray peak broadening. [1] The XRD scans were performed using Cu Kα radiation
in a Rigaku X-ray diffractometer. Microstructural characterization was conducted using a
114

field emission scanning electron microscope (JEOL JEM 7001F). The powder samples
were glued to a bulk piece using Gatan G1 resin. The SEM specimens were prepared by
JEOL cross sectional polisher. All the SEM images were captured using backscattered
electron signal to enhance the contrast form crystalline orientation.

4.3 Results and discussions
The microstructure of the trimodal powders cryomilled for 1, 2, 4, 8 and 12 hours is
shown in Figure 4.1. At the beginning of cryomilling, the grains in each powder particle
were sheared, flattened, and developed a elongated shape as shown in Figure 4.1 (a) and
(b). The average grain size of the powder milled for only one hour is about 1 um. After 2
hours of milling time, nanocrystalline regions were found among CG regions. After 4
hours, most of the regions have a grain size less than 1 um. Up to 12 hours, all areas
become nanocrystalline regimes with an average grain size of ~20 nm as shown in Figure
4.1 (e).

Figure 4.2 shows the size of B
4
C particles as a function of milling time. As we expected,
the B4C particles did not show major size reduction during cryomilling, because these
ceramic particles have good impact resistance and high hardness. Moreover, these B
4
C
particles were embedded in softer Al matrix which acted as a cushion damping the stress
applied on hard B
4
C particles. Figure 4.1 (f) shows the average grain size measured by
115

XRD. The average grain size drops rapidly at 4 hours of cryomilling which is consistent
with the SEM images shown in Figure 4.1.

Fig. 4.1 Microstructure evolution of a trimodal AA 5083 with 10 wt% of B
4
C as a
function of milling time

116

References
[1] J. He, J. Ye, E. J. Lavernia, D. Matejczyk, C. Bampton, and J. M. Schoenung, J.
Mater. Sci., 39, 6957, 2004.

117

Chapter 5
Micro-strain evolution and toughening
mechanisms in trimodal AA 5083

A trimodal metal matrix composite (MMC) based on AA (Al alloy) 5083
(Al-4.4Mg-0.7Mn-0.15Cr wt.%) was synthesized by cryomilling powders followed by
compaction of blended powders and ceramic particles using two successive dual mode
dynamic (DMD) forgings. The microstructure consisted of 66.5 vol. % ultrafine grain
(UFG) region, 30 vol. % coarse grain (CG) region and 3.5 vol. % reinforcing boron
carbide particles. The microstructure imparted high tensile yield strength (581 MPa)
compared to a conventional AA 5083 (242 MPa) and enhanced ductility compared to 100 %
UFG Al MMC. The deformation behavior of the heterogeneous structure and the effects
of CG regions on crack propagation were investigated using in situ scanning electron
microscopy (SEM) micro-tensile tests. The micro-strain evolution measured using digital
image correlation (DIC) showed early plastic strain localization in CG regions.
Micro-voids due to the strain mismatch at CG/UFG interfaces were responsible for crack
initiation. CG region toughening was realized by plasticity-induced crack closure and
zone shielding of disconnected micro-cracks. However, these toughening mechanisms did
not effectively suppress its brittle behavior. Further optimization of the CG distribution
(spacing and morphology) is required to achieve toughness levels required for structural
applications.
118

5.1 Motivation
Over the past few decades, nanocrystalline (NC) and ultrafine-grained (UFG) materials
have drawn attention due to the improved mechanical properties and the unusual grain
structures [1-5]. A wide variety of synthesis techniques have been reported to fabricate
bulk NC or UFG materials [6,7]. Among them, cryomilling is a promising synthesis
method for making NC or UFG materials in commercial quantities (30 ~ 40 kg) [8,10,32].
During cryomilling, gas-atomized metallic powders undergo severe plastic deformation
via high energy ball milling in cryogenic liquid slurry. During cryomilling these powders
are sheared, fractured and cold-welded back together refining the grain size to the NC
regime. The grain refinement process during milling can be divided into three major
stages [9]: (1) localization of high dislocation densities into shear bands, (2) low angle
grain boundaries (LAGBs) and subgrains evolving from dislocation rearrangement at
particular strain levels via recovery, and (3) LAGBs transforming to high-angle grain
boundaries (HAGBs) with excessive deformation by GB sliding and rotation. Cryogenic
temperature effectively dissipates the heat generated from milling and thus limits
recovery and grain growth [10]. In this study, Al alloy (AA) 5083 was selected as base
material because of its potential application in automobile, aerospace and marine
structures. Cryomilled Al alloy exhibits enhanced thermal stability due to the creation of
nanodispersed aluminum nitrides that pin grain boundaries and allow for the NC or UFG
microstructure to be preserved during thermomechanical processing and consolidation
[10,33-37] . Since cryomilled AA 5083 retain its refined grain structures, the strength of
119

resultant bulk products are significantly enhanced according to the Hall-Petch
strengthening mechanism [10-12]. However, NC or UFG materials usually suffer from
limited ductility and low toughness owing to minimal work hardening associated with
NC or UFG regions [13,14]. This limitation precludes the use of NC and UFG materials
from most engineering applications.

In an effort to enhance plasticity and mitigate the brittle behavior of cryomilled AA 5083
(Al-4.4Mg-0.7Mn-0.15Cr wt.%) coarse grain (CG) regions were introduced into a UFG
matrix [8,15]. One appeal of cryomilling is the ability to design multi-scale grain
structure by simply blending unmilled powders with milled powders [16,17,38]. Similarly,
one can produce a metal matrix composite (MMC) by adding reinforcing ceramic
particles during milling [18]. In this work, a trimodal microstructure was achieved by
introducing coarse grain (CG) regions into a UFG matrix reinforced by boron carbide
(B
4
C) particles. The introduction of CG regions reportedly improves the ductility of
cryomilled Al with only a moderate strength penalty [18,38], and B
4
C was selected
because it is the third hardest material known to man, surpassed only by diamond and
cubic boron nitride. Strong covalent bonds in B
4
C impart an extraordinarily high
hardness (25~40 GPa) but low density (~2.5g cm
-3
) [19].

Understanding the deformation and failure mechanisms associated with this
heterogeneous microstructure is essential to evaluating this complex microstructural
120

design and further improvement of the mechanical properties. Z. Lee [17] proposed a
toughening mechanism for a bimodal UFG Al-Mg alloy and schematically illustrated the
crack propagation through CG and UFG regions. Z. Zhang [20] discussed the fracture
mechanism of a UFG Al composite based on fracture analysis showing shear band
formation in UFG region followed by void initiation and micro-crack propagation. G. Fan
[16] reported a combined fracture mode of shear localization, cavitation and necking for a
bimodal UFG Al-Mg alloy under compressive tests. However, these post-failure analyses
did not provide clear answers to several key questions, such as: (1) how micro-strain
evolves in this multi-scale grain structure, (2) how cracks interact with CG regions along
the propagation path, and (3) what toughening mechanism(s) arise from the ductile CG
constituent. An in situ observation technique is one approach to answering these
questions. To our knowledge, no in situ straining experiments have been conducted to
directly observe the deformation and fracture behavior of a cryomilled trimodal Al MMC.
In this work, we analyzed micro-strain evolution, crack propagation and CG toughening
mechanisms at the micron scale using in situ tensile straining in a scanning electron
microscopy (SEM). Digital image correlation (DIC) was used to quantify the micro-strain
development within the heterogeneous microstructure of the trimodal MMC during a
tensile test. According to the findings from the in situ tensile test, the toughening
effectiveness of the CG inclusions was limited by sub-optimal CG distribution, resulting
in a tensile failure strain of 3.9%.
121

5.2 Experimental procedures
5.2.1 Material Synthesis
Nanocrystalline AA 5083 – B
4
C MMC powders were synthesized using a modified 1S
Svegvari attritor by ball milling in a liquid nitrogen slurry at cryogenic temperature
(cryomilling). Gas atomized AA 5083 powders obtained from Valimet, Inc. (Stockton,
CA) were blended with submicron B
4
C particulates obtained from H.C. Starck (Newton,
MA) with an average size of ~ 500 nm. This powder blend was then cryomilled for 12
hours at 180 rpm with a 32:1 ball-to-powder ratio and 0.2 wt. % stearic acid
(CH
3
(CH
2
)
16
CO
2
H). Stearic acid was used as process control agent (PCA) to prevent
excessive cold welding during cryomilling. Unmilled (as atomized) AA 5083 powder was
blended with cryomilled powder to reach a target composition of 66.5 vol. % UFG Al, 30
vol. % CG Al, and 3.5 vol. % B
4
C reinforcing particles. To remove any residual moisture
or PCA the powder blend was containerized in an Al 6061 can and hot vacuum degassing
at 500
o
C for 20 hours with a final vacuum level less than 1.0x10
-6
torr [39]. The
degassed powders were consolidated at Advanced Materials and Manufacturing
Technologies (Riverbank, CA) using dual mode dynamic (DMD) forging – a
quasi-isostatic forging process. During DMD forging, uniaxial compression was
converted to a quasi-isostatic pressure via a granular pressure transmitting medium
(PTM), surrounding the target material. DMD forging was performed at 400
o
C twice to
consolidate the powders and deform the bulk sample (after the containment can was
removed) to its final shape.
122

5.2.2 Characterization
The microstructures of this trimodal AA 5083 MMC and a conventional, armor grade AA
5083 H131 sample were characterized using a field-emission SEM (JSM-7001F, JEOL
Inc.). The conventional material, used as a baseline for comparison, is a strain hardened
plate of AA 5083 produced via ingot metallurgy, where the H 131 is a temper designation
that refers to the degree of strain hardening in the cold-rolled plate [40].The grain
structures of the conventional AA 5083 and CG regions of the trimodal sample were
analyzed using electron backscattered diffraction (EBSD). Due to the difficulty of
indexing UFG regions of trimodal nano-composite using conventional EBSD, the grain
structure of UFG regions was determined using transmitted Kikuchi diffraction (TKD)
[21,22]. The experimental setup for TKD collection is illustrated in Figure 5.1 (a). A thin
foil specimen was mounted on a TEM specimen blade and tilted from horizontal
orientation by 20
o
as shown in Figure 5.1 (a). The Kikuchi pattern was collected using an
accelerating voltage of 20kV at a working distance of 12 mm. The thin specimen used in
the TKD technique was prepared machined using a focused ion beam (FIB: JIB-4500,
JEOL Inc.) and mounted on a copper grid as shown in Figure 5.1 (b).

123

Fig. 5.1 (a) SEM chamber setup for TKD, (b) thin film sample for TKD prepared using
FIB.

5.2.3 Mechanical testing
Tensile tests were performed in situ in a SEM (JSM-6610, JEOL Inc.) using the
micro-tensile stage shown in Figure 5.2 (a). The samples were electrical discharge
machined (EDM’d) to a miniature dog-bone specimen as shown schematically in Figure
5.2 (b). In order to predict where failure will occur, a notch was deliberately made at the
gauge center to introduce a stress concentration on the target area. The tensile test was
conducted using a strain rate of 4.18 x 10
-4
s
-1
at room temperature. The tensile test was
paused every 10 seconds to record SEM images of regions of interest. Each SEM image
was centered on the same surface feature to ensure every image was captured at the
identical location. These SEM images of deforming regions were subsequently exported
to DIC software for micro-strain calculation. Details in DIC settings can be found
elsewhere [23]. Hardness of H131 and trimodal samples was measured using
instrumented indentation testing (NanoXP, MTS). Failure analysis was conducted on the
124

fractured surface after tensile tests.

Fig. 5.2 (a) Picture of the micro-tensile stage, (b) dimensions of a miniature dog-bone
tensile specimen

5.3 Results and discussion
5.3.1 Microstructure
The microstructure of the trimodal Al-based MMC was imaged using backscattered
electron signals in SEM. The bright CG regions were uniformly distributed in a dark
UFG matrix, as shown in Figure 5.3 (a). The CG regions had an average size of 165.7μm
and comprised 28.1% of the cross sectional area, which is consistent with the volume
fraction of unmilled powders blended during synthesis. The average spacing between
adjacent CG regions is ~150 𝜇𝜇𝜇𝜇 . The CG powders showed an aspect ratio of ~ 2.3 due to
flattening during consolidation. The dark tone in the UFG matrix was caused by B
4
C
particulates introduced during cryomilling. As shown in Figure 5.3 (b), no B
4
C was
observed in CG regions (CG powders were added after cryomilling). In Figure 5.3 (c),
125

B
4
C particles showed faceted shapes with an average size of 315 nm and an
approximately uniform dispersion in UFG matrix.

In addition to the dark B
4
C particles, bright particles of similar size were present in both
UFG and CG regions (see red arrow in Figure 5.3 (c)). A portion of these bright particles
were clustered at UFG - CG interface regions. Energy dispersive X-ray spectroscopy
(EDX) revealed that these particles presented Mn and Fe K lines (Figure 5.3 (d)). Thus,
these bright particles were identified as second-phase intermetallic particles (SPIPs), such
as Al
6
(Mn,Fe) [41]. These are normally present as grain refining dispersoids in AA 5083,
and were refined during cryomilling and thermomechanical processing (TMP) [42,43].
These closely spaced, submicron-sized particles potentially can provide Orowan
strengthening in both CG and UFG regions [18,24]. However, these SPIPs may also
reduce fracture toughness, as they are generally brittle and may provide sites for early
crack nucleation. Their influence on fracture toughness requires future investigation.
Close examination of SEM images showed no voids or defects in either CG or UFG
regions or at the Al/B
4
C interface before tensile loading. This observation indicated that
DMD forging was effective in consolidating Al powders to full density and eliminating
pre-existing micro-voids or micro-cracks.
126

Fig. 5.3 Backscattered SEM images of the microstructure of the trimodal sample at: (a)
low, (b) medium and (c) high magnifications. (d) Energy dispersive X-ray spectrum of
Al
6
(Mn,Fe) second-phase intermetallic particles (SPIPs)

Crystalline orientation maps of the trimodal and H131 samples were acquired to measure
the average grain size using EBSD. To achieve a high index rate of UFG regions in the
trimodal sample, TKD [21,22] was used to collect Kikuchi patterns as shown in Figure
5.4 (a). The spatial resolution using TKD (<10 nm) was significantly better than
conventional EBSD because the electron probe did not produce a broad interaction
volume in the thin TEM specimens. Each orientation map was cleaned using confidence
index (CI) standardization followed by single iteration grain dilation. Pixels with low CI
(<0.1) caused by B
4
C or SPIPs were excluded from the map as indicated by black areas
in Figure 5.4 (a). The average grain sizes in UFG, CG regions and H131 were 197 nm,
1.3 μm and 25.9 μm, respectively. The grain shape aspect ratio in UFG, CG regions and
H131 were 0.42, 0.54 and 0.44, respectively. The CG regions in the trimodal sample
127

exhibited certain textures. For example, the CG region in Figure 5.4 (b) showed (101)
texture. CG regions are likely to have textures because most CG powders were
single-crystalline particles before consolidation. During consolidation process, subgrain
boundaries were introduced in CG regions which inherit textures from their original
crystalline orientations. On the other hand, the UFG region in the trimodal sample and the
conventional H131 sample did not have preferred crystalline orientations (see Figure 5.4
(a) and (b)) because grain refinement during cryomilling leads to high angle grain
boundaries and random crystalline orientations.

To evaluate hardness, at least 30 indents were measured for each region using an
instrumented nanoindenter with a Berkovich tip. Hardness was plotted against (grain
size)
-1/2
(see Figure 5.5) indicating that the grain size strengthening effect followed the
Hall-Petch relationship [44,45] in a range from ~30 μm down to ~200 nm in grain size.
However, determination of the Hall-Petch coefficient was not possible due to the large
variation in hardness values. This variation is especially evident for the UFG matrix in
the trimodal sample because the true hardness value of AA 5083 grains in UFG regions
was inevitably biased by the presence of nearby B
4
C and Al
6
(Mn,Fe) dispersoids. The
hardness variation for the CG regions is attributed to different textures and Al
6
(Mn,Fe)
dispersoids in the CG regions. The Hall-Petch coefficient of AA 5083 was well
documented in the literature [10]. Table 5.1 presents a summary of the morphology,
average grain size and hardness of each region in trimodal and H131 samples.
128

Fig. 5.4 (a) Crystalline orientation map of UFG regions in the trimodal sample collected
by TKD. (b) Crystalline orientation map of CG regions in the trimodal sample and (c)
H131 using conventional EBSD

Table 5.1 Summary of the microstructure and hardness values for CG region, UFG
region and B
4
C of the trimodal sample and for the H131 sample
Avg. grain
size
Avg. grain
aspect ratio
Avg. region/particle
size
Hardness
(GPa)
CG 1.3 μm 0.54 165.7 μm 2.69 ± 0.98
UFG 197 nm 0.42 - 4.87 ± 1.61
H131 25.9 μm 0.44 - 1.40 ± 0.06
B
4
C - - 315 nm 24.8 ± 5.56

Fig. 5.5 Plot of hardness vs. grain size
-1/2
indicating that Hall-Petch relationship applies
in a range from ~30 μm down to ~200 nm in grain size
129

5.3.2 In situ tensile testing
Engineering stress - strain curves from the micro-tensile tests are plotted in Figure 5.6.
The serrated yielding (Figure 5.6) in the H131 curve was due to dynamic strain aging
(DSA) or Portevin–Le Chatelier effect, which is common in ductile solid-solution
materials [25]. DSA was not observed in the trimodal curve, indicating low dislocation
mobility and limited work hardening [46,47].

The 0.2% offset yield strength, ultimate tensile stress (UTS) and elongation to failure 𝜀𝜀 𝑓𝑓
for trimodal and H131 samples are tabulated in Table 5.2. The yield strength of the
trimodal Al-based MMC (581 MPa) was twice that of the conventional H131 sample
(242 MPa). The enhanced strength is attributed primarily to the refined grain structure
and presence of reinforcing B
4
C particles. A more complete analysis of the strengthening
mechanisms present in trimodal composites can be found in reference 48. The trimodal
sample showed 3.9% elongation to failure 𝜀𝜀 𝑓𝑓 with no necking in the gauge area prior to
failure. When compared to H131 the trimodal sample appears to fracture at a relatively
low strain with little evidence of work hardening. However, many carbide reinforced
MMCs, such as Al-SiC MMCs [26] are naturally brittle in tension and therefore the
observed tensile ductility of the trimodal sample (with both carbide reinforcement and
UFG regions) is significant. Therefore, the remarkable tensile ductility in the Al-B
4
C
MMC is attributed to the CG content in the trimodal microstructure.

130

The strain-to-failure values presented in this work are higher than those reported in
standard tests due to the size effect, instrument compliance and low strain rate. The
tensile specimens used in this work have reduced sizes and customized geometry to fit in
the micro-tensile module. Therefore, larger strain-to-failure values can be partially caused
by the size effect [49]. Instrument compliance is not negligible when testing high strength
trimodal specimens. The deformation of the micro-tensile module contributed to the
derived strain as well [49]. The low strain rate (~10
-4
s
-1
) used in this work is another
reason for higher strain-to-failure values. All the tensile tests in this work were conducted
using the same sample geometry and the same testing condition to allow direct
comparison between non-standard tests.

Fig. 5.6 Engineering stress and strain curves for the trimodal and H131 samples

131

Table 5.2 Mechanical properties of the trimodal Al-based MMC and H131 sample
Yield
strength
(MPa)
UTS
(MPa)
Elongation
to failure, 𝜺𝜺 𝒇𝒇 (%)
Trimodal 581 613 3.9
H131 242 394 29.6

5.3.3 Micro-strain measurement
The evolution of micro-strain in the trimodal Al-based MMC was analyzed using DIC, a
technique that can be used to calculate and visualize the micro-strain field on a deforming
surface during micro-tensile testing. For DIC, an isotropic random speckle pattern is
required on the specimen surface to track the full-field surface deformation. After
vibratory polishing, the dispersed B
4
C particles and SPIPs serendipitously provided
intrinsic speckle patterns for a field of view of ~ 50 by 50 μm.

Using DIC, the micro-strain field near the notch area at the yielding stage was mapped, as
shown in Figure 5.7. During the early stage of deformation, elastic displacements
occurred evenly in UFG and CG regions. As tensile loading increased, CG regions began
to yield, owing to the lower yield strength compared to the UFG matrix, while UFG
regions continued to deform elastically. As a result, the evolving plastic strain was
localized primarily in CG regions, as shown in Figure 5.7 (b). Similar finding was
132

observed in UFG Al alloys with a bimodal grain structure [50].

However, as strain increased, voids and cracks formed between the CG regions and the
UFG matrix. Figure 5.8 shows micro-cracks at CG/UFG interfaces prior to fracture.
Voids and cracks were absent prior to loading, so these micro-cracks formed during
deformation due to stress concentrations, and subsequently evolved to form macro-cracks.
From Figure 5.8, note that a shear band is present in the CG region but not in the UFG
matrix. The observation supports the hypothesis that CG regions bear more plastic
strain than the UFG matrix during deformation. As indicated by arrows in Figure 5.8,
micro-cracks also formed at large inclusions of SPIP.

Fig. 5.7 (a) SEM micrograph of the region of interest in DIC micro-strain analysis, CG
regions highlighted with dash line boundaries, (b) DIC strain field map of the same area
showing development of localized micro-strain in CG regions

133

Fig. 5.8 SEM micrograph of micro-voids formation due to a strain mismatch between CG
and UFG regions; Crack nucleation found in large inclusions of SPIP, indicated by the
arrow

5.3.4 Toughening mechanisms
Toughening mechanisms were revealed through dynamic observations of crack growth
during in situ straining experiments. These experiments were undertaken to understand
how cracks interacted with the CG regions, which were intended to enhance ductility and
toughness at a moderate cost in strength [27,28]. Two toughening mechanisms were
identified from the in situ straining experiments. First, cracks were blunted in the CG
region due to plasticity. When a crack reached a CG region, it was arrested, blunted, and
deflected, as shown in Figure 5.9 (a). Shear banding extended across this CG region.
Second, cracks were arrested in CG regions through zone shielding of disconnected
micro-cracks and micro-voids along the path of the shear band, as indicated in Figure 5.9
134

(b) (arrows). These micro-cracks and micro-voids effectively relaxed the stress triaxiality
at the crack-tip. As a result, the propagating crack tip deflected and advanced through the
relatively brittle UFG matrix. Complete decohesion was not observed at the CG/UFG
interface, indicating a strong metallurgical bond between the CG and UFG regions.

Fig. 5.9 (a) SEM backscattered electron micrograph showing a blunted crack tip trapped
in a CG region and a deflected crack propagating in UFG region. (b) SEM secondary
electron micrograph showing disconnected micro-cracks or micro-voids along the path of
the shear band

The observations described above revealed the role of CG regions in toughening the
trimodal composite. However, these toughening mechanisms were limited by the
availability of pathways for brittle crack propagation in the UFG matrix. As cracks
lengthened, the stress intensity factor increased at the crack tip, and the ductile CG
regions were eventually unable to restrain the crack. Consequently, despite localized
ductility, the trimodal sample fractured in a brittle manner. The ductile behavior stemmed
from deformation of CG regions before the crack propagation prevailed in the UFG
135

matrix.

Ductile phase toughening depends strongly on morphology and distribution of the ductile
phase in the brittle matrix. For example, a Zr-Ti-based bulk metallic glass (BMG)
toughened with ductile crystalline dendrites led to extraordinary toughness and tensile
ductility comparable to crystalline Ti alloys [29]. The success of this approach rested
largely on achieving a particular distribution (spacing and size) of the ductile dendritic
arms. A similar approach should be effective with the UFG-CG grain structure considered
here. For example, a CG/UFG laminate structure or a UFG matrix with a CG network
would effectively deflect and trap cracks at the early stage of crack formation. Such
microstructures can be achieved by multiple approaches. For example, one can alter the
size (and amount) of the unmilled powders blended with the cryomilled powders to
control spacing. Alternatively, one can employ an additional SPD step such as high
pressure torsion (HPT) or equal channel angular pressing (ECAP) to alter the shape and
spacing of the CG regions to laminate structures without significant grain refinement.

The microstructural length scale (L) and spacing (S) of CG inclusions can be optimized
when L and S ≈ 𝑟𝑟 𝑝𝑝 , where 𝑟𝑟 𝑝𝑝 is the characteristic size of the plastic zone associated
with the crack tip [29]. Assuming a mode I opening crack and plane stress condition
(surface cracks), 𝑟𝑟 𝑝𝑝
is given by the equation below [30].
136

𝑟𝑟 𝑝𝑝 =
𝐾𝐾 1𝐶𝐶 2
𝜋𝜋 𝜕𝜕 𝑦𝑦 2
(Eq. 11)
where 𝐾𝐾 1 𝐶𝐶 and 𝜎𝜎 𝑦𝑦 are fracture toughness and yield strength, respectively. For plane
strain condition (internal cracks), the plastic zone size will be ~1/6 of the above value.
We can choose the fracture toughness of a bimodal AA 5083 (~10 MPa·m
1/2
) as a
conservative estimate [31]. The yield strength on the other hand, can be taken from the in
situ tensile tests reported here. Although the in situ test is not a standard test, this strength
value can be used as a reasonable estimate, because sample geometry has negligible
effect on yield strength [49]. The 𝑟𝑟 𝑝𝑝 for the UFG matrix can be estimated as 𝑟𝑟 𝑝𝑝 ≈ 10 −
50 𝜇𝜇𝜇𝜇 , indicating that the optimal toughening effect can be achieved when the size and
spacing of CG inclusions are on the order of 10 − 50 𝜇𝜇𝜇𝜇 . For the current trimodal
sample, the average size and spacing of CG inclusions are on the order of ~150 𝜇𝜇𝜇𝜇
which are sub-optimal in terms of toughening effectiveness.

5.3.5 Fracture analysis
The fracture surface of the trimodal Al-based MMC was examined after micro-tensile
testing. Figure 5.10 shows a portion of the fracture surface at a CG/UFG interface. The
region illustrates the dual failure modes of brittle intergranular fracture in the UFG matrix
and a dimpled ductile fracture in the CG region. The relatively smooth layer between the
UFG and CG region arose from strain localization in CG regions prior to fracture.
Micro-voids were present in this layer, but full-length decohesion between the CG and
137

UFG regions was not observed. Negligible plastic flow was evident in the UFG matrix.
Instead, the plastic flow in this trimodal Al-based MMC was largely confined to CG
regions, which effectively blunted propagating cracks.

Fig. 5.10 SEM micrograph showing a fracture surface at the CG/UFG interface

5.4 Conclusions
In this work, a trimodal MMC was synthesized by cryomilling, blending with unmilled
powders and hot consolidation via DMD forging. Micro-strain development and failure
mechanisms were observed directly via in situ micro-tensile tests in a SEM. The in situ
observations demonstrated that the ductile CG regions were only partly effective in
blunting cracks that initiated in the UFG regions. Modifying the distribution (spacing and
morphology) of the ductile CG regions holds promise as a means to more effectively
toughen this brittle MMC. The cryomilling synthesis route – blending milled and
138

unmilled powders, affords direct control of the distribution more readily than other
processing methods. Furthermore, the morphology of the CG regions can be
modified/optimized by secondary SPD processes (e.g., HPT or ECAP) to arrest or delay
crack growth more effectively at the early stage of deformation (Attempts were made in
Chapter 6). Systematic effort to control the distribution and morphology of ductile CG
inclusions promises to provide clearer understanding of crack growth mechanisms in
bimodal alloys and provide guidelines for design of UFG microstructures with enhanced
strength and toughness.

139

References
[1] H. Gleiter, Progess Mater. Sci., vol. 33, pp. 223–315, 1990.
[2] R. Birringer, Mater. Sci. Eng. A, vol. 117, pp. 33–43, Sep. 1989.
[3] C. Suryanarayana, Int. Mater. Rev., vol. 40, no. 2, pp. 41–64, Jan. 1995.
[4] K. Lu, Mater. Sci. Eng. R, no. 16, pp. 161–221, 1996.
[5] H. Gleiter, Acta Mater., vol. 48, pp. 1–29, 2000.
[6] M. A. Meyers, A. Mishra, and D. J. Benson, Prog. Mater. Sci., vol. 51, no. 4, pp.
427–556, May 2006.
[7] R. Valiev, R. . Islamgaliev, and I. . Alexandrov, Prog. Mater. Sci., vol. 45, no. 2,
pp. 103–189, Mar. 2000.
[8] V. L. Tellkamp, A. Melmed, and E. J. Lavernia, Metall. Mater. Trans. A, vol. 32A,
no. September, pp. 2335–2343, 2001.
[9] G. Hardenbergstr, Nanostructured Mater., vol. 6, no. 95, pp. 33–42, 1995.
[10] D. B. Witkin and E. J. Lavernia, Prog. Mater. Sci., vol. 51, no. 1, pp. 1–60, Jan.
2006.
[11] K. M. Youssef, R. O. Scattergood, K. Linga Murty, and C. C. Koch, Appl. Phys.
Lett., vol. 85, no. 6, p. 929, 2004.
[12] K. . Kumar, H. Van Swygenhoven, and S. Suresh, Acta Mater., vol. 51, no. 19, pp.
5743–5774, Nov. 2003.
[13] E. Ma, Scr. Mater., vol. 49, no. 7, pp. 663–668, Oct. 2003.
[14] P. G. Sanders, J. a. Eastman, and J. R. Weertman, Acta Mater., vol. 45, no. 10, pp.
4019–4025, Oct. 1997.
[15] Y. Wang, M. Chen, F. Zhou, and E. Ma, Nature, vol. 419, no. 6910, pp. 912–5,
Oct. 2002.
[16] G. J. Fan, H. Choo, P. K. Liaw, and E. J. Lavernia, Acta Mater., vol. 54, no. 7, pp.
1759–1766, Apr. 2006.
140

[17] Z. Lee, V. Radmilovic, B. Ahn, E. J. Lavernia, and S. R. Nutt, Metall. Mater.
Trans. A, vol. 41, no. 4, pp. 795–801, Oct. 2009.
[18] Y. Li, Y. H. Zhao, V. Ortalan, W. Liu, Z. H. Zhang, R. G. Vogt, N. D. Browning,
E. J. Lavernia, and J. M. Schoenung, Mater. Sci. Eng. A, vol. 527, no. 1–2, pp.
305–316, Dec. 2009.
[19] K. M. Reddy, P. Liu, a Hirata, T. Fujita, and M. W. Chen, Nat. Commun., vol. 4, p.
2483, Jan. 2013.
[20] Z. Zhang, T. Topping, Y. Li, R. Vogt, Y. Zhou, C. Haines, J. Paras, D. Kapoor, J.
M. Schoenung, and E. J. Lavernia, Scr. Mater., vol. 65, no. 8, pp. 652–655, Oct.
2011.
[21] P. W. Trimby, Ultramicroscopy, vol. 120, pp. 16–24, Sep. 2012.
[22] P. W. Trimby, Y. Cao, Z. Chen, S. Han, K. J. Hemker, J. Lian, X. Liao, P.
Rottmann, S. Samudrala, J. Sun, J. T. Wang, J. Wheeler, and J. M. Cairney, Acta
Mater., vol. 62, pp. 69–80, Jan. 2014.
[23] Y. Zhang, T. D. Topping, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans. A,
vol. 45, no. 1, pp. 47–54, May 2013.
[24] J. Ye, B. Q. Han, Z. Lee, B. Ahn, S. R. Nutt, and J. M. Schoenung, Scr. Mater., vol.
53, no. 5, pp. 481–486, Sep. 2005.
[25] K. Peng, W. Chen, H. Zhang, and K.-W. Qian, Mater. Sci. Eng. A, vol. 234–236,
pp. 138–141, Aug. 1997.
[26] S. R. Nutt and J. M. Duva, Scr. Metall., vol. 20, no. 7, p. 1055, 1986.
[27] Z. Lee, D. B. Witkin, V. Radmilovic, E. J. Lavernia, and S. R. Nutt, Mater. Sci.
Eng. A, vol. 410–411, pp. 462–467, Nov. 2005.
[28] A. P. Newbery, S. R. Nutt, and E. J. Lavernia, J. Miner. Met. Mater. Soc., no.
April, pp. 56–61, 2006.
[29] D. C. Hofmann, J.-Y. Suh, A. Wiest, G. Duan, M.-L. Lind, M. D. Demetriou, and
W. L. Johnson, Nature, vol. 451, no. 7182, pp. 1085–9, Feb. 2008.
[30] R. W. Hertzberg, “Deformation and fracture mechanics of engineering materials,”
WILEY, Chapter 8, 1996.
141

[31] P. S. Pao, H.N. Jones and C. R. Feng, Mater. Res. Soc. Symp. Proceeding, vol. 791,
Q1.8.1, 2004.
[32] A. P. Newbery, B. Ahn, T. D. Topping, P. S. Pao, S. R. Nutt and E. J. Lavernia, J.
Mater. Proc. Tech., vol. 203 (1-3), pp. 37-45, 2008.
[33] R. W. Hayes, P. B. Berbon, and R. S. Mishra, Metall. Mater. Trans. A, vol. 35, pp.
3855–3861, 2004.
[34] T. J. Van Daam, and C. C. Bampton, US Patent. 2008, The Boeing Company
(Chicago, IL)
[35] O. Susegg, E. Hellum, A. Olsen and M. Luton, Micron Microsc. Acta, vol. 23, no.
1/2, pp. 223–224, 1992.
[36] Y. Li, W. Liu, V. Ortalan, W. F. Li, Z. Zhang, R. Vogt, N. D. Browning, E. J.
Lavernia, and J. M. Schoenung, Acta Mater., vol. 58, no. 5, pp. 1732–1740, Mar.
2010.
[37] F. Tang, C. P. Liao, B. Ahn, S. R. Nutt and J. M. Schoenung., Powder Metallurgy,
2007. 50(4): p. 307-312.

[38] L. Jiang, K. Ma, H. Yang, M. Li, E. J. Lavernia, and J. M. Schoenung, JOM, vol.
66, no. 6, pp. 898-908, 2014.

[39] Z. Zhang, S. Dallek, R. Vogt, Y. Li, T. D. Topping, Y. Zhou, J. M. Schoenung, and
E. J. Lavernia, Metall. Mater. Trans. A, vol. 41, no. 2, pp. 532–541, Nov. 2009.

[40] Metals Handbook. 10 ed. “Properties and Selection: Nonferrous Alloys and Special
Purpose Materials”, V ol. 2. 1990: ASM International.

[41] Y. J. Li, W. Z. Zhang, and K. Marthinsen, Acta Mater., vol. 60, no. 17, pp. 5963–
5974, Oct. 2012.

[42] G. Lucadamo, N. Y. C. Yang, C. S. Marchi, and E. J. Lavernia, Mater. Sci. Eng. A,
vol. 430, no. 1–2, pp. 230–241, Aug. 2006.
[43] T. D. Topping, B. Ahn, Y. Li, S. R. Nutt, and E. J. Lavernia, Metall. Mater. Trans.
A, vol. 43, no. 2, pp. 505–519, Aug. 2011.
[44] E. O. Hall, Proc. Phys. Soc. London, vol. 64, no. 381, pp. 747–753, 1951
142

[45] Petch, N.J., Journal of the Iron and Steel Institute, 1953. 174(1): p. 25-28.
[46] F. Tang and J. M. Schoenung, Mater. Sci. Eng. A, vol. 493, no. 1–2, pp. 101–103,
Oct. 2008.
[47] T. D. Topping, and E. J. Lavernia, 13th International Conference on Aluminum
Alloys, John Wiley & Sons, Inc., Hoboken, NJ, USA, 2012
[48] H. Yang, T. D. Topping, K. Wehage, L. Jiang, E. J. Lavernia, and J. M. Schoenung,
Mater. Sci. Eng. A, Jul. 2014.
[49] Y. H. Zhao, Y. Z. Guo, Q. Wei, T. D. Topping, a. M. Dangelewicz, Y. T. Zhu, T.
G. Langdon, and E. J. Lavernia, Mater. Sci. Eng. A, vol. 525, no. 1–2, pp. 68–77,
Nov. 2009.
[50] B. Ahn, E.J. Lavernia, and S.R. Nutt, J. Mater. Sci., vol. 43, pp. 7403, 2008.

143

Chapter 6
Two-step SPD processing of a trimodal Al-based
nano-composite

6.1 Motivation
Ultrafine-grained (UFG) materials, characterized by average grain sizes of < 1 µm, have
been studied extensively because of their unique properties [1-7]. Because of the small
grain size and large volume fraction of the grain boundary regions, UFG materials exhibit
greater strength than their coarse grain (CG) counterparts. The Hall-Petch effect predicts
an increase in yield strength with a decrease in grain size [8,9]. However, UFG materials
are inherently brittle at room temperature due to the suppression of plasticity mechanisms.
Additions of ceramic particles impart additional strength but further reduce ductility and
toughness. In this work, we present an approach to increasing the toughness of a UFG
Al-based metal matrix nano-composite using two-step severe plastic deformation (SPD)
processing.

SPD processing describes a set of top-down fabrication techniques used in making UFG
materials [10,11]. In SPD methods, a high level of plastic strain is applied to coarse grain
(CG) materials to refine the grain structure to the UFG regime [12]. Unlike bottom-up
techniques, such as electrodeposition or vapor deposition, SPD techniques can be used to
produce UFG materials with full density in bulk forms other than thin films. In this work,
144

the aluminum alloy (AA) 5083 was selected as the base material due to the low density,
weldability, corrosion resistance, and widespread use in automobile, aerospace, marine
and defense applications [13-15]. However, the relatively low strength of AA 5083
(relative to steel and titanium alloys) limits applications in high load bearing structures
and ballistic protection. Therefore, increasing the strength of AA 5083 using SPD
techniques represents a practical approach to expand the utility of this alloy to a broader
range of engineering applications. In this study, we describe the synthesis of a bulk UFG
AA 5083 nano-composites using a combination of cryomilling and high pressure torsion
(HPT).

Cryomilling [16], a variant of ball milling, was performed to refine the grain structure
beyond the refinement typically possible by other SPD techniques. During cryomilling,
the powders were repeatedly fractured and cold-welded while immersed in a slurry of
stainless steel balls and liquid nitrogen. During this process, the grain size of AA 5083
powders was refined to the nanocrystalline regime via severe plastic deformation [16].
The liquid nitrogen dissipated heat generated during milling, thereby suppressing
recrystallization and grain growth while simultaneously inhibiting oxidation at fractured
surfaces and promoting the formation of nitride phases [17-19]. Hot vacuum degassing
was performed to remove residual gas contamination and a two-step consolidation
process was carried out to consolidate the powders into bulk forms. Previous studies have
shown that the as-milled powders exhibited an average grain size of less than 100 nm
145

[16]. The subsequent degassing and consolidation processes introduced moderate grain
growth, leading to a final average grain size of ~500 nm in the bulk form [16].

Cryomilled AA 5083 is inherently brittle due to the lack of effective plasticity
mechanisms. In early efforts to improve toughness, unmilled CG powders were blended
with milled powders before consolidating the powders to bulk form [20,21]. The
microstructure after consolidation consisted of CG regions evenly dispersed in a UFG
matrix and thus a bimodal grain size distribution could be fabricated. The introduction of
CG regions in a UFG matrix reportedly imparts unusual deformation and fracture
mechanisms not unlike ductile-phase toughening. For example, it was observed that the
bimodal structure of an Al-Mg alloy exhibited increased ductility with a moderate
sacrifice in strength [21,22]. Similarly, it was reported that the CG regions in the bimodal
microstructure not only improved the global ductility, but also delayed plastic instability
[20]. Another study of bimodal aluminum showed that the plastic strain was localized in
the CG regions during deformation [23].

To afford ballistic protection to military vehicles, researchers have reinforced UFG
aluminum alloys with boron carbide (B
4
C) particles, adding the particles during
cryomilling [24,25] to produce a trimodal structure (UFG matrix, CG region and B
4
C
particles) as described later in this work. The trimodal AA 5083 exhibited a two-fold
increase in tensile strength compared to a conventional alloy (see section 3.4). This
146

remarkable increase in strength was attributed to a combination of B
4
C strengthening via
load transfer, grain boundary strengthening, and Orowan strengthening from dispersoids
[26]. During deformation, micro-voids nucleated in the brittle UFG matrix and evolved
into cracks. When a propagating crack reached a ductile CG region, the crack tip was
blunted and arrested due to the enhanced plasticity of the CG regions. However, in a
previous study, we observed that this toughening mechanism was ineffective because the
spacing and the morphology of the CG regions were far from optimal [27]. For effective
ductile-phase toughening, the size and spacing of the ductile regions should be
comparable to the plastic zone size of the brittle matrix [28]. Based on a previous
estimate [27], the optimal size and spacing of the CG regions for effective toughening are
~10 - 50 µm. In this work, we explore an approach to achieve near-optimal spacing and
shapes of the CG regions by applying post-consolidation shear strain using HPT.

HPT is another SPD technique for making ultrafine-grained materials in disk or ring form
[29]. During HPT, a large amount of torsional strain is applied on a disk sample under
high compressive pressure (typically ~6.0 GPa) [29]. In HPT, a disk sample is typically
placed in the depressions between two anvils and the lower anvil rotates to produce
torsional straining with a radial gradient (largest strain on periphery and lowest in center).
The compressive constraint ensures that the sample can sustain large amounts of strain
without fracture [29]. Processing by HPT was conducted on the trimodal AA 5083 as a
secondary SPD process to reduce the spacing of the CG regions and impart more lamellar
147

shapes. In the following sections, we describe a trimodal microstructure reshaped by HPT
and investigate the effects of this structure on the mechanical behavior and failure
mechanism of trimodal AA 5083.

6.2 Experimental procedures
6.2.1 Material synthesis
Gas atomized AA 5083 powder (Al-4.4Mg-0.7Mn-0.15Cr wt.%) together with submicron
B
4
C particles were ball milled (32:1 ball-to-powder ratio) in a liquid nitrogen slurry for
12 hours using a modified 1S Svegvari attritor at 180 rpm. Afterwards, unmilled (as
atomized) AA 5083 powder was blended with the milled powder to reach a trimodal
composition of 66.5 vol. % UFG Al, 30 vol. % CG Al, and 3.5 vol. % B
4
C reinforcing
particles. The mixed powder was sealed in a container for hot vacuum degassing at 500
o
C (773 K) for 20 hours. The degassed powders were then consolidated using dual mode
dynamic (DMD) forging, formerly known as quasi-isostatic or Ceracon™ forging.

After DMD forging, trimodal samples were machined into 9.8 mm disks as shown in
Figure 6.1. The plane normal of the disk was parallel to the forging direction. The
trimodal disks were abrasive-polished to a final thickness of ~0.82 mm. These disks were
processed by HPT at room temperature under a compressive pressure of 6.0 GPa. The
HPT processing in this work was conducted in a quasi-constrained condition at a rotating
speed of 1 rpm through total numbers of 1/4, 1, and 5 turns (revolutions). The processed
148

disk had a final thickness of ~0.69 mm with minor material outflow at the periphery.
Subsequent to HPT, samples were annealed for one hour at 100
o
C (373 K), 200
o
C (473
K), 300
o
C (573 K) and 400
o
C (673 K), individually.

Fig. 6.1 Dimensions of the HPT disk and two miniature dog-bone specimens machined
from the disk.

6.2.2 Microstructure
The microstructure of the sample processed by HPT was observed using a digital light
microscope (VHX-600, Keyence, USA) after preparing polished sections using SiC
papers and a colloidal silica suspension. The grain size distribution in the CG regions was
measured using electron backscatter diffraction (EBSD) in a field-emission scanning
electron microscope (SEM, JSM-7001F, JEOL Inc.). The specimen surface was
ion-polished for EBSD using a cross-section polisher (IB-09010CP, JEOL Inc.). Kikuchi
149

patterns were collected at a working distance of 15 mm with an acceleration voltage of 15
kV using a TSL orientation imaging system. Due to the difficulty in indexing material
with high dislocation density via conventional EBSD, the grain size of the UFG matrix
was characterized using the transmitted Kikuchi diffraction (TKD) method described
earlier [27]. A field-emission transmission electron microscope (TEM, JEM-2100F, JEOL
Inc.) was used to measure the grain size and identify small precipitates and dispersoids. A
focused ion beam (FIB) system (JIB-4500, JEOL Inc.) was used to prepare thin TKD
specimens and TEM thin foil specimens.

6.2.3 Mechanical testing
Microhardness measurements were carried out using an FM-1e Vickers indenter with a
maximum load of 200 gf and a dwell time of 10 s. The indents were made on a quarter
surface of the disk with a spacing of 300 µm because we assume the microstructure is
symmetrical with respect to the disk central axis. After the microhardness measurements,
disks were cut using an electrical discharge machine into two miniature tensile specimens
with gauge lengths of 1 mm and thicknesses of 0.5 mm, as shown in Figure 6.1. Tensile
tests were performed at room temperature with a constant strain rate (~4.8 x 10
-4
s
-1
)
using a micro-tensile stage [23]. The yield strength, ultimate tensile stress (UTS) and
elongation to fracture were calculated directly from the stress-strain curves. Fractured
surfaces were investigated after tensile tests using SEM. However, due to limited
materials, only one tensile specimen was tested for each sample.
150

6.3 Results and discussion
6.3.1. Microstructure
Microstructure after cryomilling
Prior to HPT processing, the microstructure of the cryomilled trimodal AA 5083
composite exhibited CG regions with an average size of ~150 µm (area equivalent
diameter) randomly dispersed in a UFG matrix, as shown in Figure 6.2 (a). The spacing
and the average aspect ratio of the CG regions were ~120 µm and 2.9 µm, respectively.
The average grain sizes in the CG regions and the UFG matrix are listed in Table 1. The
dark contrast of the UFG matrix derives from the B
4
C reinforcing particles (~400 nm)
which generate a weak backscatter signal due to the low atomic number, as shown in
Figure 6.2 (b). Al
6
(Mn,Fe) dispersoids of similar size to B
4
C particles were present in
both CG regions and the UFG matrix, appearing as bright particles in Figure 6.2 (b).
These second-phase intermetallic particles (SPIPs) can alter the strength of the material
via conventional Orowan strengthening. However, these second-phase particles may also
reduce the fracture toughness because these brittle intermetallic particles may provide
early crack nucleation sites under tension as shown in Figure 6.3. No preexisting
micro-cracks or voids were observed in the as-consolidated trimodal specimens,
indicating that the Al powders can be effectively consolidated into a full density bulk
sample using DMD forging.

151

Fig. 6.2 (a) Backscatter SEM micrograph of the trimodal structure, (b) Enlarged view of
the microstructure showing UFG matrix with B
4
C (dark particles) and CG regions with
Al
6
(Mn,Fe) dispersoids (bright particles).

Fig. 6.3 Secondary electron SEM image of a crack nucleation site on an intermetallic
particle in the trimodal AA 5083.

Microstructure reshaping by HPT
To achieve more a desirable spacing and morphology of the CG regions, HPT was
conducted on the cryomilled trimodal sample as a secondary SPD processing step. In the
following context, the “cryomilling + HPT” processed trimodal samples and the
“cryomilled-only” trimodal samples will be designated as “nT”, where n is the number of
152

turns performed during HPT. Thus, a cryomilled sample processed with 1 turn of HPT is
named “1T” and the “cryomilled-only” sample is labeled “0T” (without HPT
processing).

The morphology of the CG regions was markedly altered by the shear strain during HPT.
Figure 6.4 shows the microstructure of the HPT processed samples. In these images, the
bright and dark regions are the CG regions and the UFG matrix, respectively. The CG
regions started deforming at the disk periphery because the shear strain is a maximum at
the edge during torsion [30,31]. As shown in Figure 6.4 (a), (b) and (c), the morphology
of the CG regions evolved from oval shapes to a fine marbleized pattern as the number of
HPT turns increased. Figure 6.4 (d) shows the microstructure of sample “5T”. The width
and spacing of the CG strips are ~25 µm which lies within the optimal size and spacing
range (~10 to 50 µm). The aspect ratio and spacing of the CG regions in each sample are
summarized in Table 1 where all the measurements were taken near the edge of the disk
samples.

153

Fig. 6.4 Visible light microscope images showing microstructure evolution after HPT of
(a) 1/4 T, (b) 1T and (c) 5T. (d) Close-up view of the 5T specimen.

During HPT, both the UFG matrix and the CG regions experienced grain refinement
together with increased dislocation density. As shown in Figure 6.5 (a), sometimes a CG
region exhibited a single crystalline orientation accompanied by low-angle grain
boundaries, an indication that some of the un-milled CG powders were single crystals
prior to consolidation. The large shear strains applied during HPT introduced dislocations
and reduced the grain size in these CG regions. The EBSD inverse pole figure images
show a marked grain size reduction from an average size of 9.2 µm to 971 nm after 1/4
turn of HPT (see Figure 6.5 (b)). After five turns of HPT, the average grain size of the CG
regions dropped to ~260 nm which is comparable to that of the UFG matrix. The black
regions in Figure 6.5 (b) are un-indexed due to poor quality Kikuchi patterns during
154

EBSD scans. This low pattern quality indicates a higher dislocation density introduced
during HPT. The un-indexed regions comprised larger area fractions with increasing
numbers of HPT turns. In the UFG matrix, the grain size was further reduced from ~450
nm to ~200 nm after HPT.

As shown in Fig 6.6, the grain size finally reached a constant value of ~250 nm for both
the CG regions (Because CG regions no longer have coarse grains after HPT, we will
rename them as “ex-CG regions” to avoid confusion) and the UFG matrix. It was not
feasible to reduce the grain size further via more HPT turns because the strain hardening
rate was compensated by the dynamic recovery rate at such high strain levels. Grain size
measurements (even using TKD) became difficult (un-indexed Kikuchi patterns) due to
the high dislocation density and lattice distortion caused by shear strain during HPT.
Therefore, the average grain size of “1T” and “5T” samples are only approximate due to
the limited number of indexed grains. The size of B
4
C reinforcing particles remained
unchanged after HPT, as shown in Fig 6.6, because these hard brittle particles were
effectively shielded by the softer UFG matrix during HPT. No cracks or voids were
observed in the HPT processed samples because of the high compression loading during
HPT and strong Al/B
4
C interfaces [32].

155

Fig. 6.5 (a) EBSD inverse pole figure map of the CG regions of sample “0T”, showing an
average grain size of 9.2 µm in CG regions. (b) Grain size reduction (avg. grain size ~
971nm ) in CG regions after HPT (1/4 T).

156

Fig. 6.6 Average grain size and B
4
C size plot as a function of HPT turns.

Table 6.1. Summary of the microstructural change as a function of HPT turns.
Sample ID Avg. grain size
UFG* (nm)
Avg. grain size
CG/ex-CG*
(nm)
CG/ex-CG
aspect ratio*
CG/ex-CG
spacing*(nm)
B
4
C
Size (nm)
0T 441±158 9191±4502 2.9±1.4 125.8±45.0 409±158
1/4 T 283±79 971±487 5.6±1.8 72.5±37.6 384±173
1 T 177±94 237±134 7.6±3.7 49.8±31.9 418±214
5 T 204±60 260±52 24.2±11.5 24.7±8.2 439±208
* All measurements were made near the edge of each HPT disk sample.

157

6.3.2 Microstructure homogeneity
One potential drawback to HPT processing is the microstructural inhomogeneity that
invariably arises from the radial strain gradient in the HPT disks in the early stages of
HPT processing. Thus, a Vickers microhardness map was constructed to evaluate the
hardness change and monitor the microstructural homogeneity among different areas of
the disk. Only a quarter of the disk surface area was examined because it is reasonable to
anticipate that the microstructure is symmetric with respect to the central axis of the disk.
In Figure 6.7 (a), the hardness map of sample “0T” shows an average hardness of ~120
H
v
and the hardness variations stem from the heterogeneous trimodal structure. In Figure
6.7 (b), sample “1/4T” exhibited a hardness increase (~250 H
v
) starting from the edge,
while the center of the disk remained unchanged because the 1/4 turn of HPT was
insufficient to alter the microstructure near the disk center. As the number of turns
increases, the hardness increases at both the disk edge and center, as shown in Figure 6.7
(c) and (d), until there is saturation at ~300 H
v
. This hardness saturation was consistent
with the minimum grain size observed in Figure 6.6. Further torsional straining will not
reduce the grain size and therefore will not increase the hardness. As shown in Figure 6.7
(d), processing by HPT through 5 turns produced uniform hardness, grain size
distribution and ex-CG morphology throughout the disk except at the center of the disk.
The miniature dog-bone tensile specimens were machined to avoid the center to ensure
that the entire gauge area had uniform microstructure.

158

Fig. 6.7 Vickers microhardness map of a quarter surface of the HPT disks.

6.3.3 Annealing
As discussed in section 3.1.2, the average grain size of the ex-CG regions approached
that of the UFG matrix after several turns of HPT. As a result, the ductility of the ex-CG
regions decreased, thereby limiting the toughening effect. To restore some of the ductility
and toughening capacity, the HPT processed trimodal AA 5083 was annealed to achieve
recrystallization and grain growth in the ex-CG regions. Figure 6.8 shows the
microstructures in the ex-CG regions after 1 hour annealing at 373K (100
o
C), 473K
159

(200
o
C), 573K (300
o
C) and 673K (400
o
C). Recovery and recrystallization started at 373K
(100
o
C), as shown in the inverse pole figure map (Figure 6.8 (b)), which shows fewer
un-indexed areas (dark region) compared to the non-annealed “5T” sample (Figure 6.8
(a)). Grain growth was observed in the ex-CG regions after annealing at 373K (100
o
C)
and at higher temperatures (Figure 6.8 (c), (d) and (e)).

By contrast, the grain structure of the UFG matrix was preserved after annealing. Figure
6.9 (a) shows an EBSD image quality map of sample “5T” after 1 hour annealing at
673K (400
o
C). The dark contrast in the UFG matrix indicates a low average intensity of
the Hough peaks due to a high dislocation density and small grain size. Figure 6.9 (b) is
an STEM annular dark field image showing no significant grain growth in the UFG
matrix of the annealed “5T” sample. This result indicates the thermal stability of the
UFG matrix during annealing. In contrast, the CG regions (except for the grain
boundaries) exhibit good quality metric (bright areas in Figure 6.9(a)), a result of reduced
dislocation density and grain growth during annealing.

The thermal stability of the UFG matrix is attributed to the presence of oxides and
nitrides introduced during powder processing and cryomilling, respectively. These fine
precipitates and solute segregated to grain boundaries and effectively pinned boundary
movement and restricted dislocation glide [33]. The CG regions lacked thermal stability
because the CG powders were not cryomilled and thus contained no oxide and nitride
160

precipitates and solute to impede grain growth. Consequently, annealing restored the
ductility of the CG regions without softening the thermally stable UFG matrix. Note that
annealing in AA 5083 can potentially cause sensitization, a process in which magnesium
segregates to grain boundaries to form Al
3
Mg
2
(β phase) intermetallic compounds.
Sensitized AA 5083 is susceptible to both intergranular corrosion (IGC) and stress
corrosion cracking (SCC), particularly in corrosive environments at high temperature (>
50
o
C) [34-36].

Fig. 6.8 EBSD inverse pole figure maps of the CG regions in (a) non-annealed “5T”, (b)
“5T” annealed at 373K (100
o
C) for 1 hour, (c) “5T” annealed at 473K (200
o
C) for 1 hour,
(d) “5T” annealed at 573K (300
o
C) for 1 hour, and (e) “5T” annealed at 673K (400
o
C) for
1 hour.

161

Fig. 6.9 (a) EBSD image quality map of sample “5T” annealed at 673K (400
o
C) for 1
hour, (b) STEM annular dark field image of the UFG matrix in the annealed “5T” sample.

6.3.4 Tension
Room temperature tensile tests were performed at a strain rate of ~4.8 x 10
-4
s
-1
using a
micro-tensile stage. The tensile stress-strain curves for the “0T”, “1/4T” (annealed at
400
o
C), “1T” (annealed at 400
o
C), “5T” (annealed at 400
o
C), and a conventional
work-hardened AA 5083 are plotted in Figure 6.10. The yield strength, ultimate tensile
strength and fracture elongation are listed in Table II. To eliminate the size effect of the
tensile specimens, the properties listed in Table II were measured using identical
specimen geometry and testing conditions. As shown in Figure 6.10 (a), all trimodal AA
5083 samples exhibited high strength (> 600 MPa), at least twice that of conventional AA
5083 (~271 MPa). In particular, the annealed “5T” sample showed the highest yield
strength (~746 MPa), 24% greater than for sample “0T”. In addition, the “5T” (annealed
at 400
o
C) sample showed the greatest elongation (~1.9%) among all trimodal samples
(Figure 6.10 (b)). Note that this increase in both strength and elongation of the annealed
162

“5T” sample is attributed to the change in the spacing and shape of the CG regions after
HPT.

Fig. 6.10 (a) Tensile stress-strain curves of the conventional AA 5083, 0T, 1/4T (annealed
at 400
o
C), 1T (annealed at 400
o
C), and 5T (annealed at 400
o
C), (b) Partial enlargement of
the stress-strain curves near fracture.

Table 6.2 List of the tensile properties measured from the stress-strain curves
Sample ID
Yield strength
(MPa)
Ultimate tensile
stress (MPa)
Fracture
elongation (%)
Conventional AA 5083 271 324 16.9
0T 603 607 1.0
¼T (annealed at 400
o
C) - 611 1.0
1T (annealed at 400
o
C) 677 678 1.2
5T (annealed at 400
o
C) 746 781 1.9

163

6.3.5 Fracture mechanism
Fracture surfaces were examined after tensile tests to determine failure mechanisms. Two
distinctive failure modes were evident in the ductile CG regions and the brittle UFG
matrix as shown in Figure 6.11, where CG regions are artificially tinted to enhance
visibility. Figure 6.11 (a) and (b) are fracture surfaces of samples “0T” and “5T”
(annealed at 400
o
C), respectively. The CG regions of sample “0T” were elongated
perpendicular to the forging direction. In contrast, the CG regions of sample “5T”
(annealed at 400
o
C) had smaller spacings and no preferred direction of elongation. Figure
6.11 (c) shows brittle intergranular fracture observed in the UFG matrix. The CG region
in Figure 6.11 (c) shows a dimple and cup fracture surface, an indication of ductile
failure.

Fig. 6.11 Fracture surfaces of (a) “0T” and (b) “5T” (annealed at 400
o
C) with artificially
tinted CG regions. (c) SEM Micrograph showing intergranular fracture in the UFG
matrix and dimple-like ductile fracture in the CG region of sample “5T” (annealed at
400
o
C).

164

Fig. 6.12 Schematic illustration of the failure mechanisms in sample “5T” (annealed).

Figure 6.12 schematically shows the toughening process of the CG regions in sample “5T”
(annealed at 400
o
C). First, the majority of micro-cracks initiated at the brittle
intermetallic dispersoid sites under tensile loading. Because the spacing of the CG
regions was intentionally reduced to the plastic zone size of the UFG matrix [27], the
crack growth was effectively retarded due to a toughening mechanism in adjacent CG
regions, as shown in Figure 6.13, where a surface micro-crack was arrested at a CG
region. This delay in crack propagation led to the extended uniform elongation and
therefore higher strength of sample “5T” (annealed at 400
o
C), as shown in Figure 6.10
(b). As the load increased, micro-cracks grew and interconnected, quickly leading to
165

catastrophic fracture. The limited yielding and strain softening recorded in the
stress-strain curve (Figure 6.10) was attributed to the local necking of the CG regions, as
shown in Figure 6.12 (b). Sample “5T” (annealed at 400
o
C) exhibits a slightly higher
ductility (Figure 6.10). More CG regions were involved in toughening and thus
contributed to a higher elongation compared to sample “0T” in which a crack readily
avoided the CG regions and found “easy” propagation pathways in the UFG matrix.

Fig.6.13 Backscatter image of a surface micro-crack stopped at a CG region.

6.4 Conclusions
Most toughening strategies for ultrafine-grained materials enhance ductility while
sacrificing some amount of strength. In this work, we demonstrated that the strength and
ductility of trimodal AA 5083 composites (although still limited compared to
conventional AA 5083) can be improved simultaneously by judicious control of the
spacing and morphology of the CG regions. Increasing HPT turns (e.g. 20 turns) may
lead to even higher strength and ductility by activating new deformation mechanisms
166

(e.g., grain boundary sliding) in the UFG matrix [37-39]. The thermally stable nature of
the UFG matrix leads to selective grain growth in the ex-CG regions and introduces a
potential for high temperature applications for this material. However, this trimodal AA
5083 composite may not be suitable for use in corrosive environments due to possible
susceptibility to stress corrosion cracking (Al
3
Mg
2
precipitates introduced during
thermomechanical processing and subsequent annealing). In addition, the two-step SPD
synthesis involves multiple steps and processing variables that will inevitably entail
high-cost and low-yield production. A major practical limitation of HPT arises from the
thin plate geometry, a factor that will limit engineering applications. Alternative
fabrication techniques will be required to reduce the production cost and remove the
geometry limitation. Nevertheless, this work validates the toughening strategy in which
strong and hard phases, when combined with ductile regions of optimal size and spacing,
effectively increase strength while retaining acceptable toughness. The trimodal
microstructure described here can only be achieved through powder processing routes
that allow tailoring of multi-modal grain size distributions. These routes expand the
design space for materials with mixed microstructures and balanced properties not
achievable by conventional routes. It is anticipated that refinements of this approach will
lead to bulk materials having the toughness levels required for load-bearing structures
(~20 𝑀𝑀𝑀𝑀𝑀𝑀 ∙ 𝜇𝜇 1/ 2
) while simultaneously boosting the strength to ~700 MPa, thereby
overcoming the strength-ductility paradox of UFG materials.

167

References
[1] R. Z. Valiev, R. K. Islamgaliev, and I. V. Alexandrov, Prog. Mater. Sci., vol. 45, no.
2, pp. 103–189, Mar. 2000.
[2] R. Kapoor and J. K. Chakravartty, Acta Mater., vol. 55, pp. 5408–5418, 2007.
[3] T. G. Langdon, Acta Mater., vol. 61, no. 19, pp. 7035–7059, 2013.
[4] T. G. Langdon, Mech. Mater., vol. 67, pp. 2–8, 2013.
[5] Y. Huang and T. G. Langdon, Mater. Today, vol. 16, no. 3, pp. 85–93, 2013.
[6] L. S. Toth and C. Gu, Mater. Charact., vol. 92, pp. 1–14, 2014.
[7] R. Z. Valiev, Y. Estrin, Z. Horita, T. G. Langdon, M. J. Zehetbauer, Y. T. Zhu, JOM,
vol. 58, no. 4, pp. 33-39, 2006.
[8] E.O. Hall, Proc. Phys. Soc., B64, p. 747, 1951.
[9] N.J. Petch, J. Iron Steel Inst., 174, p. 25, 1953.
[10] Y.T. Zhu and T.G. Langdon, JOM, vol. 56, pp. 58-63, 2004.
[11] Y.T. Zhu, T.C. Lowe, T.G. Langdon, Scripta Mater., vol. 51, pp. 825-830, 2004.
[12] R.Z. Valiev, A.P. Zhilyaev, T.G. Langdon TG, Bulk Nanostructured Materials:
Fundamentals and Applications. Wiley, Hoboken, NJ, USA, 2014
[13] A. P. Newbery, S. R. Nutt, and E. J. Lavernia, JOM, vol. 58 (4), pp. 56–61, 2006.
[14] A. P. Newbery, B. Ahn, T. D. Topping, P. S. Pao, S. R. Nutt, and E. J. Lavernia, J.
Mater. Process. Technol., vol. 203 (1-3), pp. 37–45, Jul. 2008.
[15] D. Witkin, B.Q. Han, and E.J. Lavernia, Metall. Mater. Trans. A, vol. 37A, pp. 185–
194, 2006.
168

[16] D. B. Witkin and E. J. Lavernia, Prog. Mater. Sci., vol. 51, no. 1, pp. 1–60, Jan.
2006.
[17] R. W. Hayes, P. B. Berbon, and R. S. Mishra, Metall. Mater. Trans. A, vol. 35A (12),
pp. 3855–3861, 2004.
[18] O. Susegg, Micron and Microscopica Acta, vol. 23(1/2), pp. 223-224, 1992.
[19] Y. Li, W. Liu, V. Ortalan, W. F. Li, Z. Zhang, R. Vogt, N. D. Browning, E. J.
Lavernia, and J. M. Schoenung, Acta Mater., vol. 58 (5), pp. 1732–1740, 2010.
[20] G. J. Fan, H. Choo, P. K. Liaw, and E. J. Lavernia, Acta Mater., vol. 54, no. 7, pp.
1759–1766, Apr. 2006.
[21] Z. Lee, V. Radmilovic, B. Ahn, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans.
A, vol. 41, no. 4, pp. 795–801, Oct. 2009.
[22] Z. Lee, D. B. Witkin, V. Radmilovic, E. J. Lavernia, and S. R. Nutt, Mater. Sci. Eng.
A, vol. 410–411, pp. 462–467, Nov. 2005.
[23] Y. Zhang, T. D. Topping, E. J. Lavernia, and S. R. Nutt, Metall. Mater. Trans. A, vol.
45, no. 1, pp. 47–54, May 2013.
[24] L. Jiang, K. Ma, H. Yang, M. Li, E. J. Lavernia, and J. M. Schoenung, JOM, vol. 66,
no. 6, pp. 898–908, Apr. 2014.
[25] Z. Zhang, T. D. Topping, Y. Li, R. Vogt, Y. Zhou, C. Haines, J. Paras, D. Kapoor, J.
M. Schoenung, and E. J. Lavernia, Scr. Mater., vol. 65, no. 8, pp. 652–655, Oct.
2011.
169

[26] J. Ye, B. Q. Han, Z. Lee, B. Ahn, S. R. Nutt, and J. M. Schoenung, Scr. Mater., vol.
53, no. 5, pp. 481–486, Sep. 2005.
[27] Y. Zhang, T. D. Topping, H. Yang, E. J. Lavernia, J. M. Schoenung, and S. R. Nutt,
Metall. Mater. Trans. A, vol. 46, pp. 1196–1204, 2015.
[28] D. C. Hofmann, J.Y. Suh, A. Wiest, G. Duan, M.L. Lind, M. D. Demetriou, and W.
L. Johnson, Nature, vol. 451, no. 7182, pp. 1085–9, Feb. 2008.
[29] R.Z. Valiev, T.G. Langdon, Prog. Mater. Sci., vol. 51,pp. 881-981, 2006.
[30] R. Z. Valiev, Y. V. Ivanisenko, E. F. Rauch, B. Baudelet, Acta Mater., vol. 44, pp.
4705-4712, 1996
[31] F. Wetscher, A. Vorhauer, R. Stock, R. Pippan, Mater. Sci. Eng. A, vol. 387-389, pp.
809-816, 2004.
[32] Y. Li, Z. Zhang, R. Vogt, J. M. Schoenung, and E. J. Lavernia, Acta Mater., vol. 59,
no. 19, pp. 7206–7218, Nov. 2011.
[33] L. Hashemi-Sadraei, S. E. Mousavi, R. Vogt, Y. Li, Z. Zhang, E. J. Lavernia, and J.
M. Schoenung, Metall. Mater. Trans. A, vol. 43 (2), pp. 747–756, 2012.
[34] Y. Zhu, D. A. Cullen, S. Kar, M. L. Free, and L. F. Allard, Metall. Mater. Trans. A,
vol. 43, no. 13, pp. 4933–4939, Aug. 2012.
[35] S. Jain, M. L. C. Lim, J. L. Hudson, and J. R. Scully, Corros. Sci., vol. 59, no. 2012,
pp. 136–147, Jun. 2012.
[36] I. N. A. Oguocha, O. J. Adigun, and S. Yannacopoulos, J. Mater. Sci., vol. 43, no. 12,
pp. 4208–4214, Apr. 2008.
170

[37] R. Z. Valiev and I. V. Alexandrov, Y. T. Zhu, T. C. Lowe, J. Mater. Res., vol. 17,
no. 1, pp. 5–8, 2002.
[38] T. Mungole, P. Kumar, M. Kawasaki, and T. G. Langdon, J. Mater. Res., vol. 29, no.
21, pp. 2534–2546, Oct. 2014.
[39] T. Mungole, P. Kumar, M. Kawasaki, and T.G. Langdon, J. Mater. Sci., in.press,
DOI 10.1007/s10853-015-8915-2

171

Chapter 7
Historic aspects and future challenges

The cryomilled AA 5083 project was initiated by the need to reduce the overall weight of
the Assault Amphibious Vehicle (AAV7A1) of BAE Systems shown in Figure 7.1. The
AAV family of armored vehicles serves as transport and fire covering of infantry.
AAV7A1 is frequently deployed globally by the U.S. Marine Corps, including Iraq. The
weight of an AAV7A1 is 21 tons when empty and 27 tons when fully loaded for battle [1].
The vehicle’s hull is made from the work-hardened AA 5083 which contributes about 14
tons of the total vehicle weight. The overall weight reduction of the AAV7A1 is a primary
objective in order to lower the fuel consumption and expand the range of the vehicle
which is crucial in hostile battlefields. The unique cryomilling’s capability of tailoring the
strength and ductility and the B4C reinforced trimodal structure allows the use of AA
5083 in various parts of the vehicles, including ballistic protection armor. Lead by the
University of California at Davis, the cryomilled AA 5083 program is developed in
collaboration with the University of Southern California, Pittsburgh Materials Technology,
Inc. BAE systems, the Naval Research Laboratory, the Army Research Laboratory and
the University of California at Irvine [1].

172

Fig. 7.1 An Assault Amphibious Vehicle deployed in the city of Fallujah, Iraq (image
courtesy of the United States Armed Forces)

The effect of cryomilling process parameters, such as milling time, ball to powder ratio,
milling speed, and process control agent, on the final product quality has been studied
extensively and optimized [2-4]. The effects of post-processing steps such as degassing
[5], consolidation [4], extrusion [6], and forging [7] were systematically investigated to
establish the structure-processing-property relationship. Various shapes of the cryomilled
AA 5083 are available, including plate and sheet via the secondary consolidation methods
such as extrusion, rolling and forging. The post-processing parameter optimization is
currently underway to find out various routes to achieve desired microstructures and
properties.

Weldability of the cryomilled AA 5083 is also essential to the vehicle manufacturing.
Three welding techniques tungsten (inert gas welding, inertia welding and friction-stir
173

welding) were tested using the cryomilled AA 5083. Only a slight drop of hardness was
measured at the welded joints which indicate acceptable weldability of the cryomilled AA
5083 [1]. Corrosion is another concern of the cryomilled AA 5083 since the amphibious
AAV7A1 is susceptible to corrosive marine environment. Corrosion resistance of the
cryomilled AA 5083 could be tempered by sensitization that occurs during
thermomechanical processes such as degassing and consolidation at elevated temperature.
The corrosion behavior of the cryomilled AA 5083 was studied and compared with a
conventional AA 5083 [8]. The corrosion tests found that there is no significant loss of
the corrosion resistance of the cryomilled AA 5083 compared to the conventional
counterpart. However, we lack corrosion data from the bimodal or trimodal AA 5083 with
the addition of the CG regions. Sensitization may cause stress corrosion cracking in the
ubiquitous CG regions which leads to brittle failure of the vehicle structure or armor.

The biggest technical challenge of the cryomilled AA 5083 is improving the ductility
while keeping the strength uncompromised. The addition of the unmilled coarse grain
regions were explored hoping to gain ductility with a trade-off of strength. Despite these
efforts, lack of ductility is still an issue preventing the use of the cryomilled AA 5083 in
many engineering applications. Several problems need to be resolved or improved in
future studies. For instance, will the strength be preserved after forming? Is the
cryomilled material ductile enough to allow complex forming and production?

174

Despite the technical challenge, the economic cost of the cryomilled product can not be
ignored when we are considering its commercial application. Will the cost increase of the
cryomilled material be worth the overall property improvement compared with the
conventional alloy? If the cost of cryomilling is exceedingly high, the use of the
cryomilled material may be restricted in aerospace and military industries. According to
the cost model by Ye [9], a large portion of the total cost of the laboratory-scale
cryomilling attribute to raw materials, liquid nitrogen and labor cost as shown in Figure
7.2. The estimated unit price of the laboratory scale cryomilled AA 5083 is $320/kg
which is undoubtedly high for any practical application of the material. Let alone this
price does not include the cost of the consolidation methods such as HIP and extrusion.
Scaling-up the production is an effective of way to reduce the unit cost. Figure 7.3 shows
contributions of each cost element of the industry-scale cryomilling. Again, the cost
estimation only covers the cryomilling process and does not include any post-processing
of the material. The percentage of the labor cost drops significantly when the capacity of
the attritor mill increases as shown in Figure 7.3. As a result, the unit price of the
industrial cryomilled AA 5083 is reduced to $22/kg which is still expensive compared to
cast Al-MMCs produced by Alcan Inc. ($6/kg) and MC-21 Inc. ($2.2/kg) [9]. In addition,
this cost model is over optimistic because it assumes very low prices of the raw materials
and liquid nitrogen which are only feasible for large volume purchases. Therefore, the
cost of cryomilled AA 5083 greatly depends on the global demand for this material and
can be very expensive if the production scale is limited.
175

Fig. 7.2 Cost contributions by cost element of the laboratory-scale cryomilling [9]

Fig. 7.3 Cost contributions by cost element of the industry-scale cryomilling [9]

Reducing the cost of the cryomilled AA 5083 is a primary objective of its use in
commercial applications. In addition to scaling-up, cost reduction can be achieved by
176

yield improvement, liquid nitrogen waste reduction, optimization of processing
parameters and new design of the mill. Loss of powder during cryomilling is attributed to
powder welding on the tank wall and venting of the liquid nitrogen gas. If the yield of the
cryomilled powder can be increased to above 90%, the unit cost can be reduced to about
$10/kg [9]. Better tank insulation may be an effective way to reduce liquid nitrogen
consumption rate. New design of the continuous or circulation attritors may help to
increase the yield and reduce liquid nitrogen waste [10]. Process parameters also have
large contributions to the cost. Sometimes, cost can be reduced by changing parameters in
trade-off of the product quality. For example, increasing the milling speed will reduce the
milling time and thus decrease the liquid nitrogen and labor cost. However, higher
milling speed could result in high liquid nitrogen consumption rate which means that
more liquid nitrogen will be consumed to produce same amount of cryomilled powders.
Furthermore, higher milling speed will result in large centrifugal force that pushes the
milling balls against the tank wall and therefore lead to insufficient milling and low yield
during production. Figure 7.4 shows the total cost of the laboratory-scale cryomilling as a
function of the milling speed [9]. The cost is minimal and not speed-sensitive at the speed
range of 300 – 600 rpm. Product quality can then be optimized within this speed range.

177

Fig. 7.4 Total cost of the laboratory-scale cryomilled AA 5083 as a function of the milling
speed [9]

In sum, the future application and commercialization of the cryomilling largely depend
on the technical and economic challenges mentioned above. Ductility improvement and
cost reduction are key factors to the success of the cryomilled materials. Of course, it is
not easy to obtain both good properties and low cost simultaneously. Fortunately,
cryomilling has the capability of tailoring the microstructure and balancing properties
with cost for specific applications. New fabrication design is desired to optimize the CG
distribution effectively. Corrosion study is needed to justify the use of the bimodal and
trimodal AA 5083 in corrosive environment. Deeper scientific understanding of the
failure mechanisms will help enhance the toughness of the cryomilled materials and
178

benefit other research fields because the concept of bimodal and trimodal structure and
microstructure optimization will not be limited to cryomilling.

179

References
[1] P. Newbery, S. R. Nutt, and E. J. Lavernia, J. Miner. Met. Mater. Soc., no. April, pp. 56–61, 2006.
[2] D. B. Witkin and E. J. Lavernia, Prog. Mater. Sci., vol. 51, no. 1, pp. 1–60, Jan. 2006.
[3] J. Ye, B. Q. Han, Z. Lee, B. Ahn, S. R. Nutt, and J. M. Schoenung, Scr. Mater., vol. 53, no. 5,
pp. 481–486, Sep. 2005.
[4] T. D. Topping, B. Ahn, Y . Li, S. R. Nutt, and E. J. Lavernia, Metall. Mater. Trans. A, vol. 43,
no. 2, pp. 505–519, Aug. 2011.
[5] Z. Zhang, S. Dallek, R. Vogt, Y . Li, T. D. Topping, Y . Zhou, J. M. Schoenung, and E. J.
Lavernia, Metall. Mater. Trans. A, vol. 41, no. 2, pp. 532–541, Nov. 2009.
[6] Y . Li, Y . H. Zhao, V. Ortalan, W. Liu, Z. H. Zhang, R. G. Vogt, N. D. Browning, E. J.
Lavernia, and J. M. Schoenung, Mater. Sci. Eng. A, vol. 527, no. 1–2, pp. 305–316, Dec.
2009.
[7] A. P. Newbery, B. Ahn, T. D. Topping, P. S. Pao, S. R. Nutt, and E. J. Lavernia, J. Mater.
Process. Technol., vol. 203, pp. 37–45, Jul. 2008.
[8] E. Kuş, Corrosion, 62, 2, 152-154, 2006.
[9] J. Ye and J. M. Schoenung, Adv. Eng. Mater., vol. 6, pp. 656–664, 2004.
[10] A. Russel, Industrial Minerals, April, 57, 1989.

180

Chapter 8
Infusion quality and elution effect of functional
light weight ceramic proppants

In this chapter, I will present a side project during my last year of PhD. Although this
project is not directly related to the aluminum work, it is challenging and incorporating
many techniques and skills I learned from the aluminum project. This chapter focuses on
a practical engineering problem of the actual well behavior of porous ceramic proppants.

Abstract
A multi-functional kaolinite-based ceramic proppant is described that is intended to
maintain crack opening displacements while simultaneously carrying and releasing
chemicals for well treatment during hydraulic fracture (fracking). Chemical additive was
infused into porous proppants using three infusion methods. Subsequent to infusion, a
polymer coating was applied on the proppant pellets to control the chemical release rate
during well operation. The porous structure of the proppant was characterized using
low-vacuum backscatter SEM imaging and 3D x-ray microscopy, while infusion quality
was evaluated using energy dispersive X-ray spectroscopy (EDS). Load to crush a single
pellet was measured to investigate proppant mechanical behavior after infusion and
coating. The elution profile was systematically studied as a function of elution volume to
assess the effectiveness of the infused additive. The results indicate that ceramic
181

proppants can be imbued with additional functions to enhance efficiency of resource
extraction.

8.1 Motivation
The oil and natural gas stored in tight shale formations can be extracted due primarily to
two technological advances - horizontal drilling and hydraulic fracturing (also known as
fracking). Fracking involves the use of fluid and material to create small fractures in a
shale-rock formation in order to stimulate production rates from new or existing oil and
natural gas wells. In fracking, a wellbore is first drilled vertically down for 1,500 to 3,000
meters followed by horizontal drilling along the shale rock formation to maximize the
contact area between the wellbore and the reservoir. A wire installed with explosive
charges is then dropped in the well and set to fracture the wellbore casing and
surrounding rocks. These cracks are further opened and extended by flushing down
millions of gallons of high-speed fracking liquid under high pressure of 40 to 70 MPa.
When the fracking pressure is reduced, proppants (propping agents) are employed to prop
open the fracture. Multiple materials have been used for proppants including sand, glass
beads, walnut hulls, and metal shot.
Recent research proposed the use of ceramics sintered from bauxite or kaolinite powders
for light-weight proppants [1]. Because of the porous structure, this low-density proppant
requires less fracking liquid pressure to maintain proppant transport within the fracture,
thus reducing production costs. Unlike other porous low-density materials, the ceramic
182

proppant is inherently strong, an important characteristic to withstand high closure
stresses under well conditions. Unlike silica sand, the sintered ceramic proppant exhibits
a roughly spherical shape, creating more open space among proppant packs and thus
increasing the conductivity of the well. In addition, the spherical shape helps to decrease
the possibility of stress concentration at sharp edges and therefore also reduces the risk of
fragmenting into finer particles [2]. Another attractive feature of the ceramic proppant is
that its porous structure can be infused with chemicals used for well treatment.

Various chemicals are added to the fracking liquid to ensure proper fracking activity. The
chemical additives used in a typical fracture treatment depend on the conditions of a
specific well. A typical fracture treatment involves dilute concentrations of between 3 to
12 additives, including friction reducers, acids, surfactant, gelling agent, iron control
corrosion inhibitor, scale inhibitor, and biocide. Each additive serves a specific
engineering purpose [3]. For example, friction reducer is used to lubricate the fracking
water so that it can be pumped down to the long narrow bore and maintain a high
pressure at the end of pipe [4]. Various acids are commonly used to dissolve mineral
depositions in the pipe [5], and surfactant and gelling agent are used to increase the
viscosity of the fracking water [6]. Iron control prevents precipitation of metal oxide [5]
and corrosion inhibitor prevents the corrosion of the pipe [7]. Biocide kills bacteria which
produce corrosive matters [8], while scale inhibitor is used to prevent scale deposits on
183

proppant packs, production conduit or processing facilities that can reduce conductivity
and impair equipment during hydraulic fracturing [3,7,9,11].
Conventionally, these chemical additives are squeezed into perforations or added to
fracking liquid during fracking operation [11]. As an alternate approach, these additives
can be infused into the interconnected pores in the ceramic proppant to create a
multi-functional propping agent in the well treatment. Infused ceramic proppants can in
principle deliver additives throughout the entire production system in a more
cost-efficient and long-lasting way. In addition, a resin coating can be applied not only to
contain fines [2, 12] but also to control the chemical release rate. Understanding and
controlling the elution rate of the infused chemical will be critical to defining the market
value of the product and to increasing well productivity.

In this work, we present a systematic study of kaolinite-based multi-functional ceramic
proppants infused with a phosphonate type scale inhibitor [10]. Three infusion methods
were investigated and infusion quality was evaluated using energy dispersive X-ray
spectroscopy (EDS). To evaluate the chemical elution rate, infused proppant pellets were
“washed” by passing a certain volume of fracking fluid through the proppant pack. The
amount and distribution of additives remaining after elution were characterized using
EDS elemental mapping.

184

8.2 Experimental procedures
8.2.1 Single pellet compressive test
To measure the load required to crush a single proppant pellet, single pellet compressive
test was conducted using a micro-tensile stage (Deben, UK) with a maximum load of
5kN, as shown in Figure 8.1. To obtain statistical results, ten proppant pellets of each
batch were tested at a constant strain rate (~4.8 x 10
-4
s
-1
).

Fig. 8.1 Picture of the micro-tensile stage used for single pellet compressive test

8.2.2 Characterization
Kaolinite-based ceramic proppant pellets were chemically infused using three different
infusion methods, referred to as method A, B and C. The morphology of proppant pellets
was studied using digital light microscopy (Keyence VHX-Z100R). The pellet was
mounted in resin and dry-ground using abrasive papers to achieve a cross section through
its equator. The cross section was then ion-polished using a cross section polisher
185

(IB-09010CP, JEOL). The final cross section exhibited a smooth and damage-free surface
suitable for SEM imaging and EDS analysis. The microstructure of the porous structure
was imaged at 20 kV using backscattered electrons in a low vacuum environment to
minimize charging effects. To ensure high spatial resolution for resolving sub-micron
pores, the cross section image was formed by stitching over 60 high magnification
images. Using this composite image, the area fraction of pores was calculated as a
function of radial distance to reveal the pore distribution within a pellet. The cross section
was divided into concentric ring areas with a width of 25 μm, from which the area
fraction of pores was measured, as shown in Figure 8.2. The total number of pixels and
the number of black pixels in the i
th
circular area are N
i
and b
i
, respectively. Thus, the
pore fraction of the ring P can be calculated by the pixel number difference between two
neighboring circles according to Equation 1. Pore size distribution was quantified using
open source image analysis software (Image J). The equivalent pore size is defined using
Equation 2.
𝑀𝑀 =
𝑏𝑏 𝑖𝑖 + 1
− 𝑏𝑏 𝑖𝑖 𝜋𝜋 𝑖𝑖 + 1
− 𝜋𝜋 𝑖𝑖 (Eq. 1)
𝑑𝑑 = 2 �
𝑆𝑆 𝜋𝜋 (Eq. 2)
where d is defined as the equivalent pore size, and S is measured pore area from SEM
images.

186

Fig. 8.2 The cross section of a proppant pellet was divided into concentric circles. The
total number of pixels and the number of black pixels in the i
th
circular area are N
i
and b
i
,
respectively. The pore fraction of the ring can be calculated by the pixel difference
between two neighboring circles

To semi-quantify the chemical distribution inside proppant pellets, EDS elemental maps
were collected on the ion polished cross section at a working distance of 15 mm using 20
kV acceleration voltages. To ensure mapping quality, a 200 ms dwell time was selected
and 128 frames were scanned to form a spectrum map.

8.2.3 Elution
To measure the inhibitor releasing rate, the proppant pellets were eluted by fracking water
with 10%, 40% and 100% of the total elution volume. The chemical distribution after
elution was semi-quantified using EDS elemental mapping.

187

8.3 Results and discussions
8.3.1 Pellet morphology and load to crush a single pellet
The morphologies of bare, resin coated and inhibitor infused proppant pellets are shown
in Figure 8.3. The average thickness of resin coating is ~ 10.7 µm, as shown in the inset
of Figure 8.3 (b). The average diameter and the load to crush a single pellet of the bare,
resin coated and inhibitor infused pellets are presented in Figure 8.4. The coating and
infusion process did not significantly alter the average pellet size. The small variation for
each batch ensures a good inter-particle spacing, which enhances conductivity of the
proppant packs. In addition, the spherical shape of proppant pellets can reduce the risk of
crushing due to stress concentration on irregular surface features. The load to crush a
single pellet of the resin coated and infused pellets slightly increased (Figure 8.4)
compared to the bare pellets possibly because the resin coating contained fine production
[2] within the pellets and the infused chemicals filled pores so that the overall failure of a
pellet was delayed.

Fig. 8.3 Light microscope images of (a) bare proppant pellets, (b) resin coated pellets (the
inset shows a SEM micrograph of the resin coating layer), and (c) inhibitor infused
pellets
188

Fig. 8.4 Average pellet diameters of bare, resin coated, and inhibitor infused pellets

8.3.2 Pore statistics
The pore structure was characterized on a damage-free cross section, as shown in Figure
8.5 (a). The composite image benefits from both high spatial resolution and large field of
view, which is well-suited to pore analysis. The pore area fraction of a bare pellet
(without resin coating and chemical infusion) was quantified as a function of radial
distance from core to shell (Figure 8.5(b)). The calculated global porosity is ~ 21% which
is consistent with the density measurements at the fabrication site.

The non-uniform pore distribution shows a denser area in the shell region, which
potentially could act as a low permeability barrier for chemical infusion. Larger voids
were observed in the core region, which is attributed to the boundary of the seed particle.
189

However, the majority of the voids are less than 5 um in size, as shown in Figure 8.6.
These fine voids are considered to provide the major infusion channels for inhibitor.

Fig. 8.5 (a) Ring areas from which the pore fraction was calculated, (b) pore area fraction
as a function of radial distance

Fig. 8.6 Pore size distribution showing the majority of void size less than 5 um

190

Close examination of the cross section revealed that some large voids were iron-rich
impurities which exhibited a hollow inner structure surrounded by a dense shell. However,
these hollow particles did not contribute to permeability during chemical infusion
because of the surrounding closed shell. Zirconium silicate impurities from kaolinite
mineral were also observed as high density particles.

8.3.3 Infusion quality
The phosphorus-based scale inhibitor was infused into the proppant pellets using three
different methods, and the infusion quality for each method was evaluated using EDS
elemental mapping. Since the scale inhibitor is a phosphonate solution, phosphorus
contents were used as an indication of the inhibitor presence. The phosphorus distribution
for each pellet using different infusion methods is shown in Figure 8.7. A high
phosphorus concentration on the shell was observed for all three samples. However,
infusion method B failed to carry the inhibitor into the pellet core region while infusion
methods A and C resulted in a uniform distribution of phosphorus throughout the cross
section. Note that the bright particles labeled in Figure 8.7 (a) are zirconium silicate
impurities. These zirconium silicate particles appear in the phosphorus map because the
Zr peak overlaps with the P peak in the X-ray spectrum. The Zr peak was isolated using
wavelength dispersive X-ray spectroscopy (WDS), which has a higher energy resolution
than EDS.

191

Fig. 8.7 Phosphorus map of proppants infused by (a) method A, (b) method B, and (c)
method C.

Figures 8.8 and 8.9 show the sodium and chlorine maps because a salt-based solution was
used during infusion. A good correlation was found between sodium and chlorine, which
indicates that sodium and chlorine stay in a salt form during the infusion process. In
method C, a salt layer was deposited on the pellet surface after drying. This layer of salt
deposition may lead to poor adhesion for the encapsulation process.

Fig. 8.8 Sodium map of proppants infused using (a) method A, (b) method B, and (c)
method C

192

Fig. 8.9 Chlorine map of proppants infused using (a) method A, (b) method B, and (c)
method C

8.3.4 Elution effect
To determine the effects of elution, the resin-coated infused pellets (method C) were
eluted by 10%, 40% and 100% of a preset elution volume by flowing fracking water
through a fixed column of pellets at room temperature. After elution, the cross section of
the proppant pellets was analyzed using EDS. Three pellets from each batch were
analyzed to obtain statistical results and investigate pellet-to-pellet variation.

The chemical distribution maps of the proppant cross section are shown in Figure 8.10. In
Figure 8.10 (a), the inhibitor concentrated at the periphery below the resin coating after
10% elution volume. The resin layer acted as a polymer barrier which slowed down the
diffusion process and the chemical release rate. After 40% of elution volume, the
concentrated ring of the inhibitor disappeared, as shown in Figure 8.10 (b), indicating that
most inhibitor had diffused through the polymer barrier and diffused out of the pellet.
Consequently, there was a significant drop in the inhibitor concentration. In Figure 8.10
193

(b), a small amount of inhibitor and salt deposition was detected outside the resin coating,
which resulted from redeposition after drying.

After 100% elution volume, the majority of the inhibitor diffused out of the pellet, much
like the 40% eluted sample. Inhibitor remaining below the resin coating was observed
only in pellet #1, a result of pellet variation. However, a large amount of sodium chloride
was detected inside the pellet because the salt-based elution water reduced the driving
force for diffusion of sodium chloride. Based on the elution analysis, we observed a large
drop in the inhibitor concentration (from 10% to 40% elution volume), but no significant
change after 40% elution. The observations indicate inhibitor depletion after 40% of the
total elution volume, leading to a decrease of the chemical release rate.

194

Fig. 8.10 Chemical distribution of the proppant cross sections after (a) 10%, (b) 40% and
(c) 100% elution volume

195

8.4 Conclusions
We have described basic performance and structural characteristics of a multi-functional
ceramic proppant designed to deliver chemicals at controlled release rates for well
treatments, while simultaneously propping open fractures. Because of the inherently
porous structure of the proppant, the ceramic is readily infused with selected chemicals.
The combination of backscatter imaging and EDS allows characterization of
proppant-carried chemicals for the first time. The infused additives and the chemical
release rate can be selectively engineered to suit well conditions. In subsequent study,
high resolution micro-CT can be a good candidate providing tomographical information
on porosity and fraction of interconnected pores. Additional efforts should be focused on
controlling the chemical release rate to maintain the minimal chemical concentration in
the wellbore over a longer period. This multi-functional fracking agent can reduce
production costs by saving water and energy required to pump chemicals into the
wellbore, helping drilling activities profitable regardless of unpredictable change in oil
price. Cost models and field tests are needed to justify whether the improvements in well
performance and savings in water and energy is worth the proppant cost increase due to
additional infusion process.

196

References
[1] R. Rickards, H. D. Brannon, W. D. Wood, and C. J. Stephenson, SPE Production &
Operations, 212, 2006.
[2] T. Palisch, R. Duenckel, M. Chapman, S. Woolfolk, C. Ceramics, and M. C. Vincent,
SPE, vol. 119242, 2006.
[3] U.S. Department of Energy, Modern Shale Gas Development in the United States: A
Primer, 2009.
[4] J. K. Fink, Chapter 4, Hydraulic Fracturing Chemicals and Fluids Technology, pp
59, 2013.
[5] J. K. Fink, Chapter 17, Oil Field Chemicals, pp 233, 2003.
[6] J. J. Sheng, Enhanced Oil Recovery Field Case Studies, pp 117, 2013.
[7] M. M. Amro, J. Pet. Sci. Eng. 46 (4) 30, pp 243, 2005.
[8] J. K. Fink, Chapter 17, Hydraulic Fracturing Chemicals and Fluids Technology, pp
193, 2013.
[9] J. K. Fink, Chapter 11, Hydraulic Fracturing Chemicals and Fluids Technology, pp
129, 2013.
[10] D. E. Abd-El-Khalek, and B. A. Abd-El-Nabey, Desalination, vol 311, pp 227, 2013.
[11] Ghosh and X. Li, J. Pet. Sci. Eng. vol. 108, pp 250, 2013.
[12] M. Zoveidavianpoor and A. Gharibi, J. Nat. Gas Sci. Eng., vol. 24, no. 2015, pp 197,
2015.

197

List of Tables
Table 1.1 List of review articles contributing to the community of nanocrystalline and
ultrafine-grained materials
Table 1.2 List of synthesis methods for nanostructured materials using top-down and
bottom-up approaches
Table 1.3 Tensile properties of conventional, 100% cryomilled and bimodal Al-7.5Mg
Table 1.4 Mechanical properties of the trimodal Al 5083 compared with a CG sample and
a 100% cryomilled NC sample. “Al 5083-O” stands for the Al alloy with O type heat
treatment. “10/50” composite means 10 wt% of B4C and 50 wt% of CG. “A” stands for
annealing at 723 K for 2 hours
Table 2.1 Optimal pattern size and step size required for DIC using JEOL 7001F SEM
Table 3.1 Mechanical properties of the UFG Al-Mg alloy
Table 5.1 Summary of the microstructure and hardness values for CG region, UFG
region and B
4
C of the trimodal sample and for the H131 sample
Table 5.2 Mechanical properties of the trimodal Al-based MMC and H131 sample
Table 6.1 Summary of the microstructural change as a function of HPT turns
Table 6.2 List of the tensile properties measured from the stress-strain curves

198

List of Figures
Fig. 1.1 Schematic plot of the Hall-Petch relation and the Hall-Petch breakdown as a
function of grain size
Fig. 1.2 Schematic illustration of top-down and bottom-up approaches
Fig. 1.3 (a) Picture of an attrition mill used for cryomilling, (b) schematic picture
showing the ball mill cylinder used for cryomilling
Fig. 1.4 Schematic diagram of degassing process
Fig. 1.5 Prior particle boundaries in consolidated NC Al samples
Fig. 1.6 Fabrication routes of cryomilling
Fig. 1.7 (a) Schematic illustration of quasi-constrained HPT, (b) pictures of HPT setup
used in this work
Fig. 1.8 Illustration of dislocation pile-ups in a coarse grain
Fig. 1.9 Schematic illustration of grain boundary sliding
Fig. 1.10 Schematic illustration of grain boundary rotation
Fig. 1.11 Engineering stress and strain curves of (a) pure Cu processed by ECAP, and (b)
pure Ti processed by HPT
Fig. 1.12 Engineering tensile stress and strain curves of conventional Al 709x alloy by
slow rate extrusion (CG-SE), cryomilled UFG alloy after slow rate extrusion (UFG-SE),
after rotary swaging (UFG-RS) and after a combination of rotary swaging and high rate
extrusion (UFG-RS-HRE)
Fig. 1.13 True tensile stress and strain curves for nt-Cu with various twin thickness
199

(denoted by the number following nt in unit of nanometer). CG and UFG twin-free
counterparts are included for reference
Fig. 1.14 Schematic illustration of a bimodal structure with CG regions evenly dispersed
in a UFG matrix
Fig. 1.15 Tensile curves of conventional, 100% cryomilled and bimodal Al-7.5Mg
Fig. 1.16 Schematic of deformation and fracture mechanism of a bimodal grain structure
under tension along extrusion direction
Fig. 1.17 Visible light micrograph of the bimodal structure of AA 5083 (left) and local
strain level of the same area (right)
Fig. 1.17 Schematic illustration of a trimodal structure with CG regions dispersed in a
UFG matrix reinforced with hard particles
Fig. 2.1 Signals generated by the primary beam of SEM
Fig. 2.2 Schematic of an EBSD system inside a SEM vacuum chamber
Fig. 2.3 EBSD crystal orientation map (inverse pole Figure) of a coarse grain aluminum
alloy
Fig. 2.4 Illustration of the interaction volume of the SEM signals
Fig. 2.5 Schematic of the conventional EBSD process (left) and the thin foil TKD (right)
Fig. 2.6 Schematic of the TKD experimental setup for a thin foil specimen
Fig. 2.7 A typical EDS spectrum of a mineral showing characteristic X-ray and
continuous X-ray
Fig. 2.8 Picture and schematic of a JEOL field emission TEM equipped with Gatan Orius
200

and image filter camera
Fig. 2.9 Cross-section ion polisher (IB-09010CP, JEOL Inc.), image courtesy by JEOL
USA
Fig. 2.10 Schematic of the cross-section ion polishing
Fig. 2.11 Surfaces of a ceramic particle prepared by (a) abrasive polishing and (b) ion
polishing
Fig. 2.12 Schematic of the FIB dual beam system
Fig. 2.13 Picture of the Vickers microhardness tester
Fig. 2.14 Diagonal measurement of a Vickers indent on an Al-Mg alloy
Fig. 2.15 Visible light micrograph of the matrix of Vickers hardness indents with a
spacing of 300 μm
Fig. 2.16 (a) SEM image of an indent by a Berkovich tip, (b) Geometry of a Berkovich
tip
Fig. 2.17 The load-displacement curve of a nanoindentation test
Fig. 2.18 Dimensions of the miniature dog-bone specimens used for in situ tensile testing
(left) and small HPT disk samples
Fig. 2.19 (a) Illustration of specimen gripping using two types of fixtures, (b) the custom
grip with extra flanges to prevent slipping
Fig. 2.20 Illustration of the strain measurement using DIC from the initial S to the
deformed state S
1

Fig. 3.1 Geometry of a dog-bone tensile specimen and the notch area
201

Fig. 3.2 (a) A random speckle pattern with a central target mark; (b) the same pattern
after deformation; the arrows indicate the positions of reference marks
Fig. 3.3 Micro-tensile stage placed in SEM chamber
Fig. 3.4 EBL patterns with exposure (a) dose factor 1, (b) dose factor 2, (c) dose factor 5,
and (d) dose factor 7
Fig. 3.5 Pixel greyscale histogram of the EBL speckle pattern
Fig. 3.6 Pseudo-strain introduced by SEM image distortion
Fig. 3.7 Engineering stress and strain curve for the UFG Al-Mg alloy
Fig. 3.8 EBSD inverse pole figure map for the UFG Al-Mg alloy; Black and white
boundaries indicate high angle grain boundaries and low angle grain boundaries,
respectively
Fig. 3.9 Engineering shear strain e
xy
overlapped with grain structures as a function of
deformation time: (a) 180 seconds, (b) 190 seconds, (c) 200 seconds, (d) 215 seconds, (e)
225 seconds, and (f) 240 seconds
Fig. 3.10 The mean engineering shear strain evolution
Fig. 4.1 Microstructure evolution of a trimodal AA 5083 with 10 wt% of B
4
C as a
function of milling time
Fig. 5.1 (a) SEM chamber setup for TKD, (b) thin film sample for TKD prepared
Fig. 5.2 (a) Picture of the micro-tensile stage, (b) dimensions of a miniature dog-bone
tensile specimen
Fig. 5.3 Backscattered SEM images of the microstructure of the trimodal sample at: (a)
202

low, (b) medium and (c) high magnifications. (d) Energy dispersive X-ray spectrum of
Al
6
(Mn,Fe) second-phase intermetallic particles (SPIPs)
Fig. 5.4 (a) Crystalline orientation map of UFG regions in the trimodal sample collected
by TKD. (b) Crystalline orientation map of CG regions in the trimodal sample and (c)
H131 using conventional EBSD
Fig. 5.5 Plot of hardness vs. grain size
-1/2
indicating that Hall-Petch relationship applies
in a range from ~30 μm down to ~200 nm in grain size
Fig. 5.6 Engineering stress and strain curves for the trimodal and H131 samples
Fig. 5.7 (a) SEM micrograph of the region of interest in DIC micro-strain analysis, CG
regions highlighted with dash line boundaries, (b) DIC strain field map of the same area
showing development of localized micro-strain in CG regions
Fig. 5.8 SEM micrograph of micro-voids formation due to a strain mismatch between CG
and UFG regions; Crack nucleation found in large inclusions of SPIP, indicated by the
arrow
Fig. 5.9 (a) SEM backscattered electron micrograph showing a blunted crack tip trapped
in a CG region and a deflected crack propagating in UFG region. (b) SEM secondary
electron micrograph showing disconnected micro-cracks or micro-voids along the path of
the shear band
Fig. 5.10 SEM micrograph showing a fracture surface at the CG/UFG interface
Fig. 6.1 Dimensions of the HPT disk and two miniature dog-bone specimens machined
from the disk
203

Fig. 6.2 (a) Backscatter SEM micrograph of the trimodal structure, (b) Enlarged view of
the microstructure showing UFG matrix with B
4
C (dark particles) and CG regions with
Al
6
(Mn,Fe) dispersoids (bright particles)
Fig. 6.3 Secondary electron SEM image of a crack nucleation site on an intermetallic
particle in the trimodal AA 5083
Fig. 6.4 Visible light microscope images showing microstructure evolution after HPT of
(a) 1/4 T, (b) 1T and (c) 5T. (d) Close-up view of the 5T specimen
Fig. 6.5 (a) EBSD inverse pole figure map of the CG regions of sample “0T”, showing an
average grain size of 9.2 µm in CG regions. (b) Grain size reduction (avg. grain size ~
971nm ) in CG regions after HPT (1/4 T)
Fig. 6.6 Average grain size and B
4
C size plot as a function of HPT turns
Fig. 6.7 Vickers microhardness map of a quarter surface of the HPT disks
Fig. 6.8 EBSD inverse pole figure maps of the CG regions in (a) non-annealed “5T”, (b)
“5T” annealed at 373K (100
o
C) for 1 hour, (c) “5T” annealed at 473K (200
o
C) for 1 hour,
(d) “5T” annealed at 573K (300
o
C) for 1 hour, and (e) “5T” annealed at 673K (400
o
C) for
1 hour
Fig. 6.9 (a) EBSD image quality map of sample “5T” annealed at 673K (400
o
C) for 1
hour, (b) STEM annular dark field image of the UFG matrix in the annealed “5T” sample
Fig. 6.10 (a) Tensile stress-strain curves of the conventional AA 5083, 0T, 1/4T
(annealed), 1T (annealed), and 5T (annealed), (b) Partial enlargement of the stress-strain
curves near fracture
204

Fig. 6.11 Fracture surfaces of (a) “0T” and (b) “5T” (annealed) with artificially tinted CG
regions. (c) SEM Micrograph showing intergranular fracture in the UFG matrix and
dimple-like ductile fracture in the CG region of sample “5T” (annealed)
Fig. 6.12 Schematic illustration of the failure mechanisms in sample “5T” (annealed).
Fig. 6.13 Backscatter image of a surface micro-crack stopped at a CG region
Fig. 7.1 An Assault Amphibious Vehicle deployed in the city of Fallujah, Iraq (image
courtesy of the United States Armed Forces)
Fig. 7.2 Cost contributions by cost element of the laboratory-scale cryomilling
Fig. 7.3 Cost contributions by cost element of the industry-scale cryomilling
Fig. 7.4 Total cost of the laboratory-scale cryomilled AA 5083 as a function of the milling
speed
Fig. 8.1 Picture of the micro-tensile stage used for single pellet compressive test
Fig. 8.2 The cross section of a proppant pellet was divided into concentric circles. The
total number of pixels and the number of black pixels in the ith circular area are Ni and bi,
respectively. The pore fraction of the ring can be calculated by the pixel difference
between two neighboring circles
Fig. 8.3 Light microscope images of (a) bare proppant pellets, (b) resin coated pellets (the
inset shows a SEM micrograph of the resin coating layer), and (c) inhibitor infused
pellets
Fig. 8.4 Average pellet diameters of bare, resin coated, and inhibitor infused pellets
Fig. 8.5 (a) Ring areas from which the pore fraction was calculated, (b) pore area fraction
205

as a function of radial distance
Fig. 8.6 Pore size distribution showing the majority of void size less than 5 um
Fig. 8.7 Phosphorus map of proppants infused by (a) method A, (b) method B, and (c)
method C
Fig. 8.8 Sodium map of proppants infused using (a) method A, (b) method B, and (c)
method C
Fig. 8.9 Chlorine map of proppants infused using (a) method A, (b) method B, and (c)
method C
Fig. 8.10 Chemical distribution of the proppant cross sections after (a) 10%, (b) 40% and
(c) 100% elution volume

206

Abstract (if available)

Abstract Nanocrystalline (NC) or ultrafine-grained (UFG) materials exhibit significantly improved strength and hardness compared to the coarse grain (CG) counterpart according to the well-known Hall-Petch effect. In this work, aluminum alloy (AA) 5083 was selected as the target material due to its low density, good weldability and excellent corrosion resistance. Refined grain size was achieved by severe plastic deformation techniques including cryomilling and high pressure torsion (HPT) which both introduced large amount of plastic strain into the materials and thus refined grain sizes down to nano-scale. ❧ Despite enhanced strength and hardness, the NC or UFG AA 5083 is very brittle due to the lack of plasticity mechanisms. This brittleness can cause numerous risks and failures during material service or even during production. The major challenge of this work is to restore ductility of the NC or UFG AA 5083 while maintaining its high strength. Toughening approaches described in this work follow the materials science paradigm which says “microstructure connects fabrication with properties”. The mechanical properties of the end products were tailored by altering the microstructure via different processing routes. We seek an optimized microstructure which can balance the strength with the ductility and maximize the fracture toughness of the NC/UFG AA 5083. ❧ Chapter 1 will discuss the general background, modern synthesis methods, strengthening mechanisms and toughening efforts of the NC/UFG materials. Chapter 2 will introduce the methods and experimental procedures for microstructure characterization and deformation mechanism investigation. Chapter 3 will describe an in situ micro-strain measurement of an UFG Al-Mg alloy at sub-micron scale. Chapter 4 will focus on the microstructural evolution of the cryomilled Al/B4C powder during processing. Chapter 5 will investigate the failure mechanisms and the toughening effect of the ductile regions in a trimodal AA 5083 nanocomposite. Chapter 6 will demonstrate the attempt to enhance the toughness by optimizing the morphology and spacing of the ductile regions in the trimodal nanocomposite. At last, chapter 7 will discuss some historic aspects of this work and the challenges that are lying ahead.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Processing, mechanical behavior and biocompatibility of ultrafine grained zirconium fabricated by accumulative roll bonding

PDF

Mechanical behavior and deformation of nanotwinned metallic alloys

PDF

Deformation behavior and microstructural evolution of nanocrystalline aluminum alloys and composites

PDF

Microstructural evolution and mechanical properties of a copper-zirconium alloy processed by severe plastic deformation

PDF

Synthesis and mechanical behavior of highly nanotwinned metals

PDF

Properties of a ZK60 magnesium alloy processed by severe plastic deformation

PDF

Microstructural evolution and mechanical properties of an aluminum alloy processed by equal-channel angular pressing and high-pressure torsion

PDF

Shock wave response of in situ iron-based metallic glass matrix composites

PDF

The role of nanotwins and grain boundary plane in the thermal, corrosion, and sensitization behavior of nanometals

PDF

Tailoring compositional and microstructural complexity in nanostructured alloys

PDF

Atomistic modeling of the mechanical properties of metallic glasses, nanoglasses, and their composites

PDF

Development and characterization of hierarchical cellular structures

PDF

Modeling of bicontinuous nanoporous materials and investigation of their mechanical and hydraulic properties

PDF

Structure and behavior of nano metallic multilayers under thermal and mechanical loading

PDF

Understanding the formation and evolution of boundaries and interfaces in nanostructured metallic alloys

PDF

Large-scale molecular dynamics simulations of nano-structured materials

PDF

Long range internal stresses in deformed single crystal copper

PDF

Metallic syntactic foams synthesis, characterization and mechnical properties

PDF

Ageing and mechanical failure of fiber reinforced polymers

$Fatigue and fracture of pultruded composite rods developed for overhead conductor cables$

PDF

Fatigue and fracture of pultruded composite rods developed for overhead conductor cables

Asset Metadata

Creator Zhang, Yuzheng (author)

Core Title Mechanical behavior and microstructure optimization of ultrafine-grained aluminum alloys and nanocomposites

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Materials Science

Publication Date 07/29/2015

Defense Date 03/05/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag cryomilling,failure mechanism,metal matrix composite,OAI-PMH Harvest,ultrafine-grained materials

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Nutt, Steven R. (committee chair), Hodge, Andrea M. (committee member), Kassner, Michael E. (committee member)

Creator Email yuzhengz@usc.edu,yuzhengz1985@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-613286

Unique identifier UC11302809

Identifier etd-ZhangYuzhe-3738.pdf (filename),usctheses-c3-613286 (legacy record id)

Legacy Identifier etd-ZhangYuzhe-3738-0.pdf

Dmrecord 613286

Document Type Dissertation

Format application/pdf (imt)

Rights Zhang, Yuzheng

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA