Motor Protein Inspired “Artificial Muscle” Actuator
ATHESISSUBMITTEDTO
THE SCIENCE AND ENGINEERING FACULTY
OF QUEENSLAND UNIVERSITYOF TECHNOLOGY
IN FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTEROF SCIENCEBY RESEARCH
Benjamin Alan Rudder
School of Chemical, Material and Structural Engineering Science and Engineering Faculty Queensland University of Technology
2017
Copyright in Relation to This Thesis
c Copyright 2017 by Benjamin Alan Rudder. All rights reserved.
Statement of Original Authorship
The work contained in this thesis has not been previously submitted to meet requirements for an award at this or any other higher education institution. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made.
QUT Verified Signature Signature:
Date: March 2017
i To my family
ii Abstract
The replacement, augmentation and imitation of human skeletal muscle is a narrow and specific area of study with broad implications. With the rise in demand for exoskeleton structures and aids, this technology becomes ever more relevant. Currently, however, existing systems are inadequate or have severe drawbacks in general and especially when applied to skin-contacting applications, such as high operating temperatures, voltages, and currents. There is also a lack of a more faithful adaptation or biomimicry of muscle to the bionics field.
The foundation is laid for a new form of artificial muscle, which more closely mimics the structure of skeletal muscle down to its most basic units myosin and actin motor proteins. An electro-mechanical drive method together with the geometrical properties of the sarcomere, aims to create an efficient, safe, and low temperature device which can be implanted or be in contact with skin. The feasibility of the principle is investigated, physical size limitations of each of the working elements (including micromagnets and electromagnets) ascertained, and two (larger scale) prototypes produced. A numerical simulation is also developed to enable optimization through refining the design with further conditions.
The biological analogies are explored and scalability discussed, in order to develop expan- sion from sarcomere level, to myofibril, muscle cell, and whole muscle level. These areas of investigation provided improvements and have presented a feasible new paradigm for the development of more biologically inspired and efficient artificial muscles, which relies on minimisation to improve efficiency and strength. Further research is required and development is in progress, but the results and idea are exciting and promising.
iii Keywords
Sliding filament theory, QUT, Artificial Muscle, Actuator
iv Acknowledgements
I would like to express my appreciation for the following people and organisations; each instru- mental in the completion of this work:
• My QUT supervisors: Professor Cheng Yan and Dr Wijitha Senadeera for their endless patience, stern encouragement, and professional manner.
• Dr. rer. nat. Oliver Schwarz, from Fraunhofer Institute for Manufacturing Engineering and Automation IPA and Universitat Stuttgart, with whom I began this journey and who put me on the path of this project with the initial area, and my inspiration towards biomimetics.
• QUT’s Science and Engineering Faculty and Higher Degree Research staff and manage- ment, who allowed me the flexibility and freedom to make my mistakes and learn new things.
In addition, I would like to show my appreciation for the support of my family and friends throughout this extended study, for both distracting me and focusing me when I needed it, and for being ever patient, understanding, and loving. There are many stories and many individuals, I cannot list you all here. Finally, this chapter is closing.
v Table of Contents
Abstract iii
Keywords iv
Acknowledgements v
List of Figures xi
List of Tables xii
1 Introduction 1
1.1 Motivation and significance ...... 1
1.2 Background ...... 1
1.3 Thesis Outline ...... 3
2 Literature Review 5
2.1 Muscles ...... 5
2.1.1 Definition ...... 5
2.1.2 Power density of muscle ...... 6
2.1.3 Myosin, Actin and the Sliding Filament Theory ...... 6
2.1.4 Action potential ...... 11
2.2 Existing Artificial Muscle Systems ...... 11
2.2.1 Fluidic Muscle ...... 11
2.2.2 Shape Memory Alloy ...... 12
2.2.3 Piezo Systems ...... 16
2.2.4 Electroactive Polymers ...... 22
vi 3 Fundamental Theory of Motion Realisation 27
3.1 Magnetics ...... 27
3.1.1 Basic Principles ...... 27
3.1.2 Magnetic Regions ...... 28
3.1.3 Landau-Lifshitz Energy Equation ...... 28
3.1.4 Single Domain Particles ...... 29
3.1.5 Magnetic Materials ...... 30
3.2 Electromagnets ...... 31
3.2.1 Basic Principles ...... 31
3.2.2 Equations ...... 31
3.2.3 Amperes law ...... 31
3.2.4 Biot-Savart law ...... 32
3.2.5 Dipole-dipole equations ...... 33
3.2.6 Micro-scale coils ...... 33
3.3 Nano- and Micro-system construction ...... 34
3.3.1 MEMS ...... 34
3.3.2 Manufacture Methods ...... 34
3.4 Macro-scale Materials ...... 38
4 Design Concepts and Process 41
4.1 Ring field ...... 41
4.1.1 Difficulties ...... 41
4.2 Stepping arms ...... 43
4.3 Parallelisation and Series ...... 46
4.4 Control ...... 47
4.5 Physical Calculations ...... 47
4.5.1 Qualitative Dimensional Analysis of Electromagnet ...... 47
4.6 Qualitative Design of Arm elements ...... 50
vii 5 Prototyping 54
5.1 Sarcomere Model ...... 54
5.1.1 Concept ...... 54
5.1.2 Manufacture ...... 61
5.1.3 Testing ...... 65
5.1.4 Results ...... 67
6 Conclusions 73
7 Future Work 73
References 81
viii List of Figures
1.1 Simplified depiction of the Sliding Filament action ...... 2
1.2 Cyclic Depiction of the Sliding Filament Theory ...... 3
2.1 Organisation of skeletal muscle[5] ...... 7
2.2 Sarcomere detail ...... 7
2.3 Sarcomere bands explained [9] ...... 8
2.4 Detail of myosin and actin ...... 9
2.5 Tension vs length of sarcomere[11] ...... 10
2.6 Summary of SMA types [17] ...... 13
2.7 Hysteresis curve in an SMA [16] ...... 14
2.8 Twinning explanation [16] ...... 14
2.9 Effect of the bending deformation strain on the TWSME ...... 15
2.10 Temperature of the TWSME formation ...... 15
2.11 Moving charge in quartz crystals [20] ...... 17
2.12 Construction of piezo bimorph actuator [23] ...... 18
2.13 Piezo bimorph connection methods [24] ...... 19
2.14 Piezo stroke amplification layouts [25] ...... 20
2.15 Piezo Walking Actuator steps (Author illustration) ...... 21
2.16 EAP basic types[27] ...... 23
2.17 Dielectric EAP function [28] ...... 23
2.18 Ionic EAP ion flow [30] ...... 24
3.1 Atomic origins of Magnetics[34] ...... 27
3.2 Illustration of Domain Walls [34] ...... 28
3.3 Crystal Structure of Neodymium Iron Borate [38] ...... 31
ix 3.4 Microelectric Fabrication Steps [49] ...... 37
4.1 Model of design 1 ...... 42
4.2 Visualisation of error states, error in orange ...... 42
4.3 Step illustration of vernier scale principle ...... 43
4.4 Active Arm Function ...... 44
4.5 Prototype design solid ...... 45
4.6 Single-acting through-passing model ...... 46
4.7 Double-acting Z-disk model ...... 47
4.8 Bending Arms ...... 50
4.9 Friction Force Diagram ...... 51
5.1 Solid design Axonometric view ...... 56
5.2 Solid design top view ...... 56
5.3 Solid design head view ...... 57
5.4 Solid design arm detail ...... 58
5.5 Solid design shell detail ...... 58
5.6 Solid design shaft detail ...... 59
5.7 Revised Shaft and Shell Elements ...... 62
5.8 Shaft head detail ...... 63
5.9 Revised Spool element, and section ...... 63
5.10 Assembly of prototype ...... 64
5.11 Assembly of prototype spool detail ...... 64
5.12 End view of assembly ...... 65
5.13 Rendering of prototype ...... 66
5.14 Image of produced prototype ...... 66
5.15 Magnet test setup ...... 67
5.16 Vertical pull test results ...... 68
x 5.17 Horizontal pull test results ...... 68
5.18 Theoretical force, 2A current ...... 71
7.1 Alternate arm design ...... 75
xi List of Tables
2.1 Calculations for estimating performance characteristics of a stripe actuator . . . 20
2.2 Electroactive Polymer type comparison ...... 25
3.1 Friction values [50] [51] [52] ...... 38
3.2 Polymer Material Data list ...... 40
xii Chapter 1
Introduction
1.1 Motivation and significance
Creating a compact and efficient artificial muscle actuator has been the topic of many institutes and researchers. The creation of a truly humanoid robot has been the dream of many. Replacing or supporting damaged muscles in humans with artificial ones is within the reach of a select few.
It is estimated that at least 1 in 190 Americans are living with a lost limb [1] with this number increasing over time, due to diabetes and other illnesses. In England, there are roughly 11,000 lower limb amputations alone, per year, [2] and in Australia this figure is around 8,000. [3] Whilst these amputations are the most prevalent, they are not the only ones, and this is an indicator of how many people are using prosthetics. Functional artificial muscles would aid these people in the use of their prosthetics, or in some cases eliminate the need for traditional prosthetics.
In addition to the replacement of lost or damaged muscles, an ever-increasing field of de- velopment is that of exoskeletons and other body-mounted robotic supports. Obviously, a more efficient and effective artificial muscle or actuator which is applicable in such environments would be of interest to this field in particular.
1.2 Background
Muscles themselves are difficult to replicate in a controllable manner, and many actuators are not efficient enough to be worthwhile. Festo’s and other fluidic muscles are elegant approx- imations to the broad action of muscles, but require too many systems. Shape memory alloy actuators have potential, but require too much current and generate too much heat to be used in contact with skin. Piezo crystal distortion requires the use of thousands of volts - to have
1 CHAPTER 1. INTRODUCTION 2 this near the body is also not desirable. There must be a better solution, which has not yet been considered. Biomimetics is an approach to engineering which accepts that nature, through evolution, is the best engineer - biomimetics takes evolutionary developments and applies them to engineering design processes. From biology, comes the inspiration for this project.
The essential elongation and contraction action of skeletal muscle has been used with suc- cess with fluidic muscles, however this project aims to go deeper, to a finer level than a muscle. A muscle is actually a bundle of fibers, which are in turn bundles of fibers (Cells), which contain a bundle of fibers (Myofibrils), which are made up of repeating segments (Sarcomeres) consisting of yet smaller fibers (Myosin and Actin). At this level, these fibers are actually indi- vidual molecules - motor proteins. These proteins individually ’crawl’ along each other, and the summation of these contractions forms the greater contraction that we see. This crawling motion is accomplished by bulbous ’heads’ on the Actin filament bonding with bonding locations on the Myosin filaments, then a power stroke, and then disbonding. This occurs symmetrically, so that each section gets shorter. This is known as the slidingfilament theory (see Section 2.1.3)
Figure 1.1: Simplified depiction of the Sliding Filament action
This research aims to apply this principle of action to attempt to solve the problem of the aforementioned inefficient, large, and hot actuators. By implementing the sliding filament theory with existing (or newly developed) technology, a possibly revolutionary actuator will be created. Permanent and electro magnets were used to approximate the bonding, powerstroke and disbonding of the filaments, reproducing the biological system as closely as possible. It should be efficient and applicable to a broader range of tasks than existing actuators, with a focus on bionic applications.
This type of development has not been investigated before, so it is novel in the field. As such, broad knowledge is collected, in order to gain a scope of the possibilities of the device. CHAPTER 1. INTRODUCTION 3
Figure 1.2: Cyclic Depiction of the Sliding Filament Theory
1.3 Thesis Outline
In order to achieve the end goal of developing a biologically inspired actuator, and understand- ing of the various broad areas of the biology and electromechanics is required. This thesis contains chapters exploring relevant literature, methods of inducing motion, design processes, and prototypes.
1. Chapter 2: Literature Review
Chapter 2 summarizes a review of the current literature in the fields pertaining to muscles. The first subsection pertains to muscles and the most fundamental acting unit of the muscle, the sarcomere. The molecular interaction of the contraction of muscle, the sliding filament theory, is explained in-depth, along with other pertinent details. The second subsection outlines and explains the shortcomings of existing artificial muscle systems currently available and in development, including fluidic muscles, shape memory alloys, piezoelectric systems, and electroactive polymers. The aim is that the result of the project overcomes these shortcomings. CHAPTER 1. INTRODUCTION 4
2. Chapter 3: Fundamental Theory of Motion Realisation
The third chapter outlines the technology required to realise the force and therefore motion of the device. The fundamental theoretical background of magnetics, magnetic regions, micromagnetics and electromagnets are explained. To compliment, Micro Elec- trical and Mechanical Systems and their production methods are elaborated, as well as some considerations made in the selection of materials for prototyping.
3. Chapter 4: Design Concepts and Process
Chapter 4 expands upon the design process of the actuator. Two models inspired by the sliding filament theory are discussed − one briefly, and the other with the aim of implementation, as it is closer to the biological inspiration and also less complicated and more efficient. Higher levels of the design are also discussed, expanding from the single acting unit to a group dynamic, modeled by activation energy. Physical calculations and models are also developed, qualitatively developing the feasibility of the device.
4. Chapter 5: Prototyping
Chapter 5 details the prototyping process, including the models created in a solid mod- eling software, Solidworks, for the purposes of 3D printing and testing. Photos of the actually produced model are also included, as well as an analysis of some results obtained in the production.
5. Chapter 6: Conclusions
The conclusions section summarises the results and determines that the production of the novel actuator is feasible. Further work is also discussed, to improve upon the results and design of the device. Chapter 2
Literature Review
2.1 Muscles
2.1.1 Definition
The basic function of muscle is to generate force. Anatomically and functionally, muscles can be divided into two types, smooth and striated. Striated or striped muscle can be further divided into skeletal muscle and cardiac (heart) muscle. Regardless of the type, all muscles share the following basic properties [4]:
• Conductivity: A muscle has the ability to conduct an action potential.
• Irritability: When stimulated, the muscle will react.
• Contractility: A muscle can shorten or produce tension between its ends.
• Relaxation: A muscle can return to resting properties after contraction.
• Distensibility: A muscle can be stretched by a force outside of the muscle itself. The muscle is not injured as long as it is not stretched past its physiological limits.
• Elasticity: The muscle will resist elongation and will return to its original position after passive or active elongation. Elasticity is the opposite of distensibility.
Smooth muscle and striated muscle can easily be differentiated from each other in a variety of ways, including appearance. For example, smooth muscle is uni-nucleated and contains sarcomeres (the functional units of muscle) that are arranged at oblique angles to each other; under a polarised-light microscope smooth muscle appears to be relatively featureless as a result of the orientation of its sarcomeres. On the other hand, striated muscle contains protein arrays called myofibrils that are parallel to each other and thus form striations or stripes. Cardiac muscle can be easily identified as distinct from skeletal muscle by appearance and differences in function, such as an intrinsic ability to contract [5]
5 CHAPTER 2. LITERATURE REVIEW 6
2.1.2 Power density of muscle
The power density of muscle is related to the type of muscle, the way in which the muscle is constructed by the body, as well as the animal in which it is. For example, in some birds, the power density of mitochondria in flight muscles running at a high temperature, is in the range that a little over 1 ml of mitochondria is required to sustain 1 Watt of mechanical power output. On this basis, a muscle with equal volumes of mitochondria and myofibrils should be able to deliver a specific power of about 430 W/kg, at an operating frequency around 40 Hz for non- fibrillar, or 230 Hz for fibrillar muscle. The limiting specific power should be twice this level in either case, i.e. about 860W/kg. [6] This is obviously quite efficient and a good aim to attain.
2.1.3 Myosin, Actin and the Sliding Filament Theory
Biological muscles are formed by repeating subsystems of differing scales. Muscles as the average person understands them are large bundles of fibres which contract together to create force. However, this is just the largest level. Muscles are actually a sequence of bundles of a series of microscopic fibres.
Skeletal muscles are those that are attached to bones by tendons, and that can be consciously controlled. Muscles are composed of bundles of muscle fibres, separated by a connective film. These fibres, muscle cells, are approximately 50 micrometres in diameter, and up to several centimetres in length [7] are in turn separated by a connective tissue known as endomysium. Inside this membrane is the sarcolemma - the cell membrane of a muscle cell. The muscle cell can be as long as the entire muscle, and contracts. However, the contraction is created by smaller contractions of sub-elements in the muscle cell, each adding a minute contraction, which sum to a large action. These elements are within the sarcoplasm; the cytoplasm of the muscle cell. This sarcoplasm consists of many small fibres called myofibrils, which are, in turn, consistent of yet smaller filaments; myofilaments. There are two types of filaments - myosin and actin. The combination of one of each filament forms the smallest unit of muscle contraction; a sarcomere. Myosin, the thicker filament, is about 15nm in diameter, and actin, the thinner, is about 7nm in diameter [7] Figure 2.1 shows this information in a more understandable manner, focusing on the continually reducing scale of muscle. CHAPTER 2. LITERATURE REVIEW 7
Figure 2.1: Organisation of skeletal muscle[5]
Figure 2.2: Sarcomere detail
Figure 2.2 better illustrates the final levels of magnification. As seen in the illustration, there are named zones in the structure of the myofibril. Again, the thick filament mentioned above is myosin, and the thin is actin.
In the middle section of the diagram, there are some lines shown. The Z-line or Z-disk (from German Zwischenscheibe, meaning between-disk) is a disk which is orientated perpen- dicularly to the fibre. The sarcomere is defined as the region between two Z-disks. The centre of CHAPTER 2. LITERATURE REVIEW 8 actin proteins are all connected to Z-disks. Between the Z-disks a long, elastic protein holds the myosin proteins in the centre, known as Titin. It also provides some of the tension that allows an over-extended muscle to return to shape and usefulness [7] Titin serves other purposes also, but these are not relevant to the study[8]. The following diagram illustrates more definitely the arrangement and constituents of the sarcomere.
Figure 2.3: Sarcomere bands explained [9]
In the relaxed state, there is only a small overlap between the actin and myosin. Each myosin filament is surrounded by six actin filaments in a hexagonal arrangement, as seen figure 2.3. The overlapping of the fibres forms striations; a variance in colour when observed in polarised light. In such an observation, the Z-line has a very dark colouration when observed. The actin filaments, which, in the relaxed state, are the main constituents of the I-band and Z-disk, have a dark colour due to their thinness, which polarises the light. This contrasts with the myosin filaments, being thicker, and polarise the light less and appear brighter in the polar-filtered scans. The diagram details the other regions and also shows the hexagonal arrangement of the CHAPTER 2. LITERATURE REVIEW 9
fibres, as well as detailing the M-line - the myosin-only area where they are linked together, the Myosin equivalent of the Z-line, at which the Actin filaments are joined (and the Titin also).
Figure 2.4: Detail of myosin and actin
Figure 2.4 shows a closer view of a single strand of each of Myosin and Actin. It shows the essential elements of the sliding filament theory actions. The bottom strand is the bundle of myosin chains (in six series radially, as previously discussed, here called the myosin rod). From this, the individual bulbous heads of the myosin chains are detailed. Along the actin protein at the top of the diagram (the series of larger balls at the top of the screen) are myosin bonding sites, which are normally blocked by the curled troponin strands. When the muscle is activated, calcium causes the tropo-myosin to move, revealing bonding sites for the myosin heads. [7] The myosin heads then form chemical cross-bridges with the revealed actin sites, holding onto the site. Immediately on bonding, the myosin releases some Adenosine DiPhosphate (ADP) and inorganic phosphate(Pi) and bends, in the picture above towards the left, pushing the actin to the left and the myosin rod to the right. This is known as the power stroke. In the action of this power stroke, after the release of the phosphates, a molecule of adenosine triphosphate (ATP) binds to the myosin, the crosslink is broken. [10] The ATP molecule is then split into ADP and CHAPTER 2. LITERATURE REVIEW 10
Pi, the energy from which re-extends the myosin head to its initial position. This action is then repeated for as long as calcium maintains the availability of cross-bridging sites on the actin filament.
Due to the way in which the muscle functions at this level, with overlapping strands and a varying amount of available cross-bridging sites and myosin heads, a piecewise-linear trend with regards to strength occurs. This is because the more contracted the strand, the more available sites there are to bond.
Figure 2.5: Tension vs length of sarcomere[11]
Figure 2.5 illustrates the trend mentioned in the previous paragraph. Starting with the point marked 1 (rightmost) in the diagram, the tension or force in the strand when the actin and myosin strands are completely overextended and not overlapping is zero, as there are no available binding sites nor heads to join to, logically; this case is extreme and only possible in diseased or torn muscles. Between that and the second point is a linear increase in strength as more and more heads are gradually exposed to more bonding sites. This reaches a maximum at point 2, until there are no longer any more heads, due to the reaching of the M line as mentioned above, where the myosin proteins are bundled together and the directional change occurs. This value is then constant until the actin proteins begin to overlap at 3, and therefore reduce the number of binding sites available in the correct direction. The decay is then linear as the strands are slowly overlapping. The trend is then steeper and again linear once the myosin strands begin to collide with the Z-disks and create irregular bonds and the entire process is chaotic and very CHAPTER 2. LITERATURE REVIEW 11 inefficient. This is also an extreme case and only possible in diseased muscles.
2.1.4 Action potential
The Action Potential (AP) is the electrical potential across muscle, generated when the muscle is being activated. When skeletal muscle is moved, the control is via electrical signals delivered by the brain. These signals reach the muscle sites and cause the release of calcium and other substances to enable the muscles to function, amplifying the signal. A process which details this activation energy is known as an electromyography, and generates an electromyograph. In order to create these electromyographs, the activation potential is measured. This voltage is measurable by a number of methods, generally in two categories; invasive and non-invasive. Invasive methods involve the insertion of a number of electrodes into the muscle itself, in order to read the voltages very accurately. Non-invasive methods, as the name suggests, avoid the painful insertion of the electrodes, at the cost of reduced precision.
2.2 Existing Artificial Muscle Systems
The creation of artificial muscles has been an interesting problem for engineers and scientists throughout history. There are currently four main areas of development in artificial muscle actuator developments, and their inherent positive points and negative points in terms of a bionic use will be outlined below:
2.2.1 Fluidic Muscle
Method of Actuation
Patented by Festo, they function on the Poisson effect in that when a pliable material is stretched axially, its perpendicular cross-section becomes thinner, and if its cross section were to increase, its length would decrease[13].
Festos fluidic muscle consists of an elastomeric tube attached to two end-pieces. Air is pumped into the tube, which causes it to inflate and therefore the two end caps are pulled towards the centre. The motion is rather efficient due to the lack of sliding friction, as well as the stick-slip energy wastage generally associated with piston based cylinders [12]. CHAPTER 2. LITERATURE REVIEW 12
It appears to mimic the shape of an entire human muscle as a unit closely, as well as exhibiting pliability and sensitivity, two of the defining elements of muscle.
Strengths and Limitations
Quoted by Festo (in their official flyer) [13] to be able to exert 10 times the amount of force of a comparably sized pneumatic cylinder actuator, they are relatively strong. They provide a stroke of 25% of their length. A 10mm diameter cylinder can exert a force of 630N, giving 8 MPa, with the larger cylinders of 20 and 40mm diameter having a lower pressure output of around 4.8 MPa.
They already have many uses in industry and novel robotic systems. They have been employed by Festo to create interesting bionic applications such as the bionic kangaroo and their elephant-trunk-inspired gripping mechanism. It is a pliable muscle, meaning it has the advantage of being able to operate in environments working closely with fragile objects and people.
However, they can be difficult to control accurately, due to their pliability and elasticity[14]. In addition, they have the major drawback of requiring a pneumatic system to function, which can be bulky and costly, and require relatively large cylinders to operate untethered, which then require charging. [12]
2.2.2 Shape Memory Alloy
Shape Memory Alloys, (SMA), are materials that change shape, stiffness, position, natural frequency, and other mechanical characteristics, in response to temperature or electromagnetic fields [15] There has been increasing development in the field continually moving forward, especially in the area of aircraft, but also in reversible fastening and coupling, deployment and retrieval components, and toys. [16]
Method of Actuation
SMAs can be ‘trained to take different shapes in their cooled and heated states. In their cooled state (for one-way SMAs) they are generally super elastic, owing to their martensitic structure. CHAPTER 2. LITERATURE REVIEW 13
They can be bent and deformed at will. When the alloy is heated above its memory transfer temperature, the alloy changes to its austenite phase, and returns to the form it ‘remembers before the deformation.
Figure 2.6: Summary of SMA types [17]
SMAs behave in this way due to changes in their physical structure and the shape of the crystal lattice that forms them. In a regular material (one not an SMA; regular steel, for example), when the crystal lattice is deformed due to physical pressure on the piece, the shape change is accommodated by slip-deformation of the lattice. This is generally a permanent process in martensite, as it changes the structure in an irreversible way. In an SMA material, these deformations are accommodated in a reversible way. This way is known as twinning and allows shape changes but not volume changes. Essentially it provides a way of movement without permanent changes in the crystal structure. CHAPTER 2. LITERATURE REVIEW 14
The transformation from austenite to martensite is exothermic and releases heat, but the transformation from austenite to martensite in SMAs are such that it is reversible and exhibits hysteresis. [16]
Once the sample is deformed, if it is then heated above its memory transfer temperature, it will return to its original shape, where its entire lattice is again in its Austenite structure. These changes are more simply illustrated in the following images.
Figure 2.7: Hysteresis curve in an SMA [16]
Figure 2.8: Twinning explanation [16]
The one-way SMA effect is not so useful to the applications in this project, but is useful to CHAPTER 2. LITERATURE REVIEW 15
know. However, SMAs can also have a two-way effect; these are known as Two Way Shape Memory Alloys (TWSMAs)
Two-way shape memory effect
The transition and training of TWSMAs are complex and complicated processes which, for brevity, fall outside the scope of this study. However, it is essentially performed through training the material to remember the shape it deforms to in each state through over-deformation in the martensite state, repetitive cycling in each state, pseudoelastic cycling, a combination of the two previous, and constrained temperature cycling. These are briefly illustrated in Figures 2.9 and 2.10 respectively:
Figure 2.9: Effect of the bending deformation strain on the TWSME
Figure 2.10: Temperature of the TWSME formation
TWSMAs are the more interesting SMAs in the implications of this study, as they could be used to drive changes in systems and as actuators. CHAPTER 2. LITERATURE REVIEW 16
Strengths and Limitations
Some variations of SMAs have placed their Carnot efficiency at around 25%, and their thermal efficiency around 5% [18] Some examples of wax-based SMAs have given an output of up to 300N for 25mm strokes of a dia. 20mm x 60mm length actuator. The major limitation of SMA- based actuators especially in the application of prosthetics and other body-contacting devices is the cooling and heating required. The technology is still young but the heating and cooling of the individual strands of the SMA actuators have rather long delay times, and often require high temperatures. They generate or are operated under a considerable amount of heat, and must vent this. As development progresses, this may become less of a problem, but at the time of writing it is still inhibitive.
2.2.3 Piezo Systems
Piezoelectric effects are those in which crystalline materials, usually ceramics, display a rela- tionship between axial dimensional rate of change and electrical output.[19] When the crystal of piezo material is suddenly deformed, it outputs a short burst of electrical energy. This effect also happens in reverse; electrical potential can create a dimensional change in the crystal structure. When the piezoelectric material is stressed, a shift of the positive and negative charge centres in the material occurs, which creates an electric field. [20]
Piezo systems are often used for sensors and fine actuators, especially in the areas of sound and optics. [21]
The stroke of a piezo device is given by the applied voltage by the d33, or piezo electric coefficient, which relates the efficiency of the material in transferring electrical energy to me- chanical energy. This movement does not depend on the dimensions of the piezo element, but has a different value in the different directions in the crystal lattice. This stroke is, in most elements, a small fraction of a percent. There are continually new and surprising piezo- materials being discovered and developed, and even cellulose has been discovered to exhibit a piezoelectric effect. [22] CHAPTER 2. LITERATURE REVIEW 17
Figure 2.11: Moving charge in quartz crystals [20]
Method of Actuation
There are a number of actuators utilizing the Piezo effect to produce movement. They fall into three major categories: stack, stripe, and walking motors.
Stack actuators
Piezo Stack actuators are generally linear actuators which act by utilising thin layers of piezo materials stacked perpendicular to the line of action. This allows the small strain of the piezo crystals expansion to sum with lower voltage requirements than a single solid crystal. The voltage applied across the element is limited by the thickness of each element. A stack of two elements will have twice the movement at the same applied voltage as a single element, and three times for three elements, etc. Within the stack actuator category there is a further set of sub-categories, separated by construction geometry and voltage allowances.
Low voltage actuators are co-fired multilayer actuators. They are created by casting a ceramic and organic slurry to form a thin tape, which is then dried, electroded with a thin metal electrode (often silver palladium), stacked, laminated and fired to create a dense package. The package is cut to size, exposing the electrodes, then electrically connected. Operating voltages are usually up to 200 Volts. [21] CHAPTER 2. LITERATURE REVIEW 18
High voltage stack actuators are constructed using thicker sintered poled ceramic layers, with thin metal electrodes between the ceramics. These layers are then bonded together using quality adhesive. They are often enclosed in metal casing with mechanical pre-stress, and designed to manage the heat generated during operation. These high voltage actuators function on up to 1000 volts.
Strengths and Limitations
American Piezo company claims that their actuator can support loads of 7kN/cm2. Standard purchasable piezo stack actuators are quoted to have a high load capability of up to 10 000N, with a strain of 0.15-0.20% of their height. Piezo actuators are extremely fast-acting, and capable of precision movements due to their short ranges and thousand-volt input range. The chief drawback of Piezo actuators of this type is their short stroke.
Piezo Stripe Actuators
Stripe actuators consist of a conductive central electrode which is bonded to two piezoelectric plates. These plates are usually poled in the same direction. This means that, when the voltage is applied, one side of the strip expands due to the positive voltage and the other contracts due to the negative one, thus working together. This arrangement is known as a bimorph.[23]
Figure 2.12: Construction of piezo bimorph actuator [23] CHAPTER 2. LITERATURE REVIEW 19
Piezo bimorph actuators are used for vibration sensing and other fine sensor applications, but also for actuation and fine position control, vibration generation, and light load operations such as textile manufacture.
Figure 2.13: Piezo bimorph connection methods [24]
Strengths and limitations
This arrangement of piezo crystals allows for greater displacement at lower voltage. It does, however, reduce the amount of force applicable. The actuators can be very small and have been used in examples of microsystem actuation. The deflection and strength of the actuators are given by American Piezos catalog as:
The development of actuators which amplify the force and motion of the bimorph piezo actuator are continuing currently, which have the layout in figure 2.14 in summary:
Piezoelectric ‘walking’ actuators
In this design of Piezo actuator, the Piezo action is utilised to vibrate a shaft either linearly or rotationally by small amounts. This is accomplished by using modules of unimorph actuators linked together in an h shape such that they clamp the actor shaft. An orthogonally arranged and adhered unimorph then contracts or extends to move the first unimorph and the shaft to create the movement, and then the initial unimorph releases. The second unimorph returns to CHAPTER 2. LITERATURE REVIEW 20
Characteristic Calculation Result −6 2 2 Total Deflection 2.2×10 ×([lf ] ÷[h] )([V ]) Mm −5 3 Compliance 26.4 × 10 × ([lf ] ÷ – ([w])([h]3) Blocking Force [deflection] ÷ [compliance] N 5 2 Resonance 3.2 × 10 × ([h] ÷ [lf ] ) Hz Frequency l f = free length of actuator h = thickness of actuator w = width of actuator V = voltage (150 V maximum)
Table 2.1: Calculations for estimating performance characteristics of a stripe actuator
Figure 2.14: Piezo stroke amplification layouts [25] CHAPTER 2. LITERATURE REVIEW 21
its original size to relocate the first one. This action is repeated at high frequency, enabling the movement of the slide or shaft. Figure 2.15 details the motion:
Figure 2.15: Piezo Walking Actuator steps (Author illustration)
In the above figure, the resultant motion of the bottom blue shaft is to the left. There are a few variations on the arrangement of these parts, but the operating principle remains the same. CHAPTER 2. LITERATURE REVIEW 22
Benefits and limitations
This arrangement of actuator allows for very precise positioning of microscope slides as the current state of the art. The drive method means a rotational drive can rotate continually further than 360 degrees, and a linear slide can act the entire length of the slide. The arrangement is not self-sensing and as such requires encoding to locate. Driving is dependent on friction and hence can change depending on the surface and any defects, and is very sensitive to these. The arrangement does not require lubricant and has few moving parts and is, therefore, very simple. The lack of lubricant means functioning in extremely cold environments or in vacuum is possible without modification, allowing the application to be used in space. A major limitation for the application of this category of actuator is the slow rate of motion, and the complexity.
2.2.4 Electroactive Polymers
Electroactive Polymers (EAPs) are a more recent development finding use in a variety of ap- plications [26] , also those outside of traditional engineering and into spaces such as art and architecture.
Method of Actuation
Electroactive polymers are numerous in configuration, but generally fall into two classes: Ionic and Dielectric.
Electronic/Dielectric EAP
These EAPs function by utilising electrostatic attraction forces. They do this by having (gener- ally noble) conductive electrodes parallel to each other, separated by a non-conductive polymer. In this they function as a variable capacitor. As the electrodes are given a charge, they are pulled together, squeezing the pliable polymer between them and causing a change in the dimensions orthogonal to the plates. These actuators typically require high voltages to act, up to thousands of volts, but use minimal energy, as the moving elements are simply electrons. They can operate in many environments with little issue.
The experimental architecture field seems to use these actuators with enthusiasm, due to CHAPTER 2. LITERATURE REVIEW 23
Figure 2.16: EAP basic types[27]
Figure 2.17: Dielectric EAP function [28] CHAPTER 2. LITERATURE REVIEW 24 their interesting form effect and also their ease of manufacture for non-precision tasks. In that field, the high voltages required are not of concern and the strength is also of little matter in cosmetic or air/light- directing applications. For such applications, they can be made by simply coating some form of acrylic polymer sheeting with a conductive powder (such as graphite) on both sides and applying a high voltage across the two, in the example given, 4000V [29]
Ionic EAPs
Ionic EAPs also exist in a number of forms. In these actuators, the density of ionic elements at the boundary of the polymer is affected by an applied electric field. Initially, the ions of the polymer are randomly oriented and located. With the application of an electric field, these orient themselves and flow towards the electrode of opposite polarity. The ionic elements flowing to the side of the polymer creates an increase in the volume of the side of the polymer, creating a bending action. The most prevalent under this subtype are IPMCs, or Ionic Polymer Metal Composites. These are thin ionomeric membranes plated with noble metal surfaces which have large practical ranges of motion. The idea is also promising for biomimetic uses, as collagen fibres exhibit natural ionomeric properties.
Figure 2.18: Ionic EAP ion flow [30] CHAPTER 2. LITERATURE REVIEW 25
EAP Type Electronic EAP Ionic EAP Advantages Rapid Respose(ms) Naturally Bi-Direction DC hold Low Voltage Large forces Can have Bistable High energy density Disadvantages Single direction Requires Electrolyte High Voltage (100MV/m or Requires encapsulation if not 20MV Ferro EAP) immersed Low Electromechanic cou- pling No DC hold (except certain) Slow (fraction of second) Weak Electrolysis breaks down 1.23V
Table 2.2: Electroactive Polymer type comparison
There is a drawback to the ionic functioning of the EAP, however. The Ionic EAP requires immersion in an electrolytic substance to enable the flow and resupply of ions.
Progress is still being made on this problem, and the state-of-the-art presents numerous novel solutions. Generally, these involve gels in absorbent materials, often carbon nanotubes and other highly complex materials. [31]
A promising area of advancement in the field has been the experimentation with bucky-gel actuators, in which nano-carbon is dispersed in he ionic-liquid gel contained within the actuator, or other variations in which carbon nanofibres are orientated in order to maximise the developed motion. [32]
Benefits and limitations
A summary of the benefits and limitations of the respective elements here is given in the following graphic:
Currently, the EAP actuators available are not sufficient for use in the desired applications. Every year further progress is made, and there exists a competition in which EAP muscles compete against humans in an arm-wrestling match. So far, humans have always won. The CHAPTER 2. LITERATURE REVIEW 26
field is promising, however, and deserves attention. There are various studies progressing, and even NASA has taken interest. [33] Chapter 3
Fundamental Theory of Motion Realisation
3.1 Magnetics
3.1.1 Basic Principles
All magnetism is an effect of flowing electric charges. This is explained by Maxwells equations, as one of the fundamental laws of nature. At a rather small level, these moving charges are the electrons spinning around atoms, as shown in Figure 3.1
Figure 3.1: Atomic origins of Magnetics[34]
As different atoms have different electron arrangement and electron affinities, as well as spins, they have varying innate magnetism, as the charges of the subatomic particles affect directional fields. Molecules contain multiple atoms, and can, due to said affinities, have a more specific orientation than others. When these molecules, now forming magnetic dipoles, align, as they do in materials exhibiting ferromagnetism, they form permanent magnets.
27 CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 28
3.1.2 Magnetic Regions
A magnet is not uniformly magnetic, but has many small regions of uniform magnetic align- ment. These magnetic regions are smaller than the grains of the metal, in such that one grain of a metal can have a number of magnetic regions. Where the regions of alignment finish and change direction, a domain wall is formed, in which the alignment is somewhere between those of the neighbouring domains. [34]
Figure 3.2: Illustration of Domain Walls [34]
These magnetic regions do not form completely at random. The formation and size of the regions are not easily controlled; however, they do seek to maintain minimum possible energy within the system. This energy is given by the Landau-Lifshitz energy equation, detailed in the next subsection.
3.1.3 Landau-Lifshitz Energy Equation
Static micromagnetics wishes to solve the spatial distribution of the magnetisation of a material M at equilibrium. The energy of the system is stable at a minimum of energy, E, this is given by the following equation:
E = Eex + ED + Eλ + Ek + EH (3.1)
where:
• Eex is the exchange energy, the energy due to the exchange interaction between dipole molecules in ferromagnetic (and other) materials. As it is lowest when the dipoles are all CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 29
oriented in the same direction, it is responsible for magnetisation of magnetic materials.
• Ed is magnetostatic energy, a self-energy, from the interaction of different parts of the material. It is increased by the volume of the magnetic field outside of the domain. It encourages the magnetisation to be parallel to the surfaces of the sample so that field lines will remain within the material. Reducing this value is the main reason for the creation of magnetic domains.
• Eλ is magnetoelastic anisotropy energy. This energy is due to magnetostriction in the material, and is minimised when the axis of magnetisation of the domains are parallel.
• Ek is magnetocrystalline anisotropy energy. As crystals exhibit magnetic anisotropy, they are easier to magnetise in one direction than others. This term is minimised when the magnetisation is along the easy axis. As the crystal lattice in the grains of the material is arranged randomly, this causes randomosity in the dominant orientation.
• EH is Zeeman energy. Essentially, this is the energy of an applied field to the magnet. This energy is minimised when the domains of the material are aligned with the applied field.
Applying a magnetic field to a ferromagnetic material generally causes the domain walls to move to decrease this energy. This is how ferromagnetic materials are magnetised, and how certain materials remain magnetised.
3.1.4 Single Domain Particles
The field energy saved is proportional to the cube of the domain size, while the domain wall energy is proportional to the square of the domain size.
This means there is a balance point where no more will form at approximately the size of 10−4 -10−6 m.
This size is inherently the optimal grain size for a magnetic material, as it can then be most simply unidirectionally aligned. CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 30
3.1.5 Magnetic Materials
Neodymium is a rare earth element used to make up 18% of the material used in flint for lighters, and also the strongest available permanent magnet. It is the second most abundant of the rare-earth elements, and is almost as abundant as copper. [35]
It is found in many minerals including all lanthanide minerals such as monazite and bastna- site. It is mined mainly in Brazil, China and Australia as well as others to a much lesser extent. Approximately seven thousand tonnes are mined a year.
Magnets made using Neodymium also contain Iron and Boron. These minerals are mined and measured and put together into a vacuum induction furnace to create the alloy. In some situations, other alloys are added to give different properties such as corrosion resistance. The vacuum furnace is used so that oxygen does not corrupt the mixture of metals and prevent them from forming the correct alloy.
The resultant ingots are broken down with a process known as Hydrogen Decrepitation, or similar, and thereafter jet milled down in an inert environment to a micron sized powder, of maximum size of 3 microns.
This powder is processed in one of a number of ways to form a neodymium magnet. In one way, this powder is fed into a press and a magnetic field is applied, to align the inherent magnetic regions. It is pressed and sintered to mould the pieces together, and coated in protective Zinc. Finally, it is magnetised in the preferred direction under a field of approximately 5 Teslas, in order to induce the permanent field and lock it into position. NdFeB alloy magnets are the strongest magnetisable material as yet known. Its molecular formula is Nd2Fe14B, and this allows the molecules to become extremely magnetisable. [36]
This is due to the crystal structure of the Nd2Fe14B. It has a tetragonal structure, with the neodymium and iron having a mirror symmetry in two orthogonal planes, meaning that each symmetrical plane cancels the other’s effective magnetisation direction, leaving the crystal with a single magnetising plane, thereby enabling a very strong magnetising capability. At room temperature, this allows for a magnetisation of approximately 1.6T at room temperature. [37] CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 31
Figure 3.3: Crystal Structure of Neodymium Iron Borate [38]
3.2 Electromagnets
3.2.1 Basic Principles
Electromagnetism is one of the fundamental forces in nature, alongside gravitational and prox- imity forces. In this section, however, electromagnetism is considered on a micro scale rather than at an atomic or sub-atomic level.
As a charge moves, it creates a magnetic field which opposes its movement. In this way, the more current changes, the more magnetic field is created. This is easily seen in the first of the Maxwell equations, translated into the second of Gausss laws.
3.2.2 Equations
3.2.3 Amperes law
Amperes law is perhaps the most relevant and most indicative equation in relation to electro- magnets and solenoids. It is one of the Maxwell equations, forming a part of the basis of classical electromagnetism. It is given by the following equation: CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 32
I ZZ B · dl = µ0 J · dS = µ0Ienc (3.2) C S
in which C is a closed curve, B is the magnetic B-field in Tesla, l is length, J is total current
−2 density (in Am ), S is the surface through which it passes, µ0 is the permeability of free space, and Ienc is the total current passing through the surface.
As magnetic and electric fields are intrinsically linked by nature, the constitutive equation is given by B = µ0 H which relates the B and H fields with the permeability of free space.
Amperes law can be used to show the effects of current in a wire to create the electro magnet.
3.2.4 Biot-Savart law
Closely related to Amperes law/s, the Biot-Svart is another law which can be used to describe the magnetic field that is generated an electric current. It combines the effects of Amperes law and Gausss law for magnetism. It is dependent on the magnetostatic condition, in that the fields created are established and unchanging.
In the interest of this research, the most useful form of the Biot-Savart Law would be the equation relating B field at a location r, as follows:
Z µ0 Idl × r‘ B(r) = 3 (3.3) 4π C |r‘|
Here, I is the current, C is the path over which the current flows, dl is the differential of the direction of current flow (not charge, which is in the opposite direction), r = r-l, l is the displacement vector from the wire element to the point of calculation, and µ0 is the magnetic constant.
This law, in order to calculate the integral, relies on the superposition principle of magnetic fields, in that each contributing element of the coil superimposes in vector form for each other.
The law will be useful in order to modularise the calculation of the B-field created by each loop of the electromagnet, in order to sum them and find the absolute field for the calculation of the moment, thus leading to the force generated by the electromagnet-permanent magnet interaction. CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 33
3.2.5 Dipole-dipole equations
3µ0 5(m1r˙)(m2r˙) F (r, m1, m2) = [(m1r˙)m2 + (m2r˙)m1 + (m1m˙ 2)r − r] (3.4) 4πr5 r2
The dipole-dipole equation given above relates the magnetic moments (m1 and m2) in space, separated by the vector r (with magnitude | r | and unit vector rˆ) . This equation is a highly- idealised one, used to theoretically generate the forces of atoms with spins on each other, and as such the structure of molecules. However, as it is based on known physical constants, the formula may be of some use in calculations and shall be investigated.
1 The dipole-dipole equation states that the fall-off of a magnetic field falls off with r2 . This though is a far-field. Near field (inside the source), the equation becomes much more complicated, and unnecessary for the purposes of this work.
3.2.6 Micro-scale coils
There are a number of layouts for currently existing Micro-scale coils in the areas of electronics and micro-actuators. (See section 3.3 for more information about production methods).
Microcoils used for actuation are generally in the format of loops printed onto circuit boards. The field of use of these coils and their purposes generally the construction of inductors to receive signals; however, there are increasing developments in the area of microactuators.
Traditionally, such coils are limited to planar structures printed as a layer on the silicon wafer of the circuit, using the traditional manufacturing techniques mentioned in the following chapter.
In addition to flat coils, there has recently been impressive developments in the area of small profile 3D coils. In particular, the work of Wen [39] has developed extremely compact, self- rolling coils, which are a 3D structure of SiN strips with pre-patterned metal conducting layers. This enables the structure to take up much less space, for the same quality factor and inductance.
This process is rather complicated, and involves carefully tuning the tensions in the layers of the silicon substrate and the overlaying metal layer. However, the technology is reliable enough to produce very high-quality coils which can be tuned, and take up to 100 times less area on a surface. This novel concept creates 3D structures from a 2D process, and could be a promising advancement and useful construct paradigm for later iterations of this project. [39] CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 34
With current technology, a feature size of 10µm can be attained relatively simply [40] (see Section 3.3 on production of Micro systems).
3.3 Nano- and Micro-system construction
3.3.1 MEMS
Micro Electro-Mechanical Systems (MEMS) is a technology that can be generally defined as miniaturised mechanical and electro-mechanical elements created by various techniques of microfabrication. The critical dimensions of MEMS devices can vary from below microns (hence Micro-) up to several millimetres. Being systems, the composition of the devices can be from simple structures with no moving parts, to complicated systems with a number of moving elements, control systems, and electronics. A main criterion under the definition of MEMS is that some element has some sort of mechanical function, whether or not the element can move. [41]
The complexity of these devices has been increasing steadily as better manufacturing meth- ods are developed and perfected. It is occurring currently that MEMS versions of various sensors now outperform the macro versions in practical use.
A number of the methods of production used in the integrated circuit industry can be translated into producing MEMS devices, meaning that batch processes can be used to reduce per-unit costs of MEMS devices, and as such these devices are becoming continually more accessible.
3.3.2 Manufacture Methods
There are three main manufacture steps for MEMS. They are the key processes in the man- ufacturing of most exisiting MEM systems, and also for the production of integrated circuit boards (ICBs), which was their original purpose before MEMS processes became in focus. [42] These methods form a chain and are often iterated numerous times to enable multilayer or 3D structures to be created.
• Deposition Processes CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 35
The first process is the deposition of material, the title deposition process. It is obviously a very important technology group in the production of MEMS. There are two main deposition categories for use in the creation of MEMS; Physical deposition and chemical deposition.
Physical vapour deposition (PVD) is a collection of methods which consist of a process of transference of material from a target to a surface. This can be performed by ion beams, in which an ion beam takes an atom from the target and allows it to relocate to the substrate, or it can be evaporated from the target material via heat (thermal evaporation) or an electron beam (e-beam evaporation) in a vacuum system, or a variety of other methods PVD systems allow usual coating thicknesses of up to 40 microns. [43]
Chemical deposition relies on the chemical reaction of the substrate and an additive to grow the required thickness of the desired material. This can be achieved by streaming a source gas onto the substrate, which then reacts and forms the material desired, in a process known as Chemical Vapour Deposition. Within this category are also a number of differing variations of the process with regards to temperature and pressure ranges, as well as aids such as plasma, customizable for each desired substrate and reagent. Each different technique has a specific set of application regions and benefits, from speed of deposition, thickness, purity, oxygenation, hydrogenation, and various others such as doping and metal additives.[44]
These methods are numerous and specific for each task, and as such wont be discussed in this work until further steps and developments can supply the specific requirements to investigate within boundaries, which will fall outside the scope of the project. However, it could be foreseen that, for this project, it would be necessary to have a copper or other conductive metal be involved without corrupting the material.
• Patterning
Patterning is the preparation of the shapes of the material to be used for the MEMS device. The method generally used is lithography. Lithography is the transferral of an image by selectively masking or exposing a (flat) surface to a treatment so as to resist adherence or etching by another process. This was originally specifically for ink printing, but has in modern times become common as a process for semiconductors and by association, MEMS. [45] CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 36
In the lithography specific to the MEMS processing context, a photosensitive material is used to take up a pattern when exposed to a certain type of radiation. The radiation changes the properties of the material in question, allowing it to be selectively removed or treated to stay, and the pattern to be transferred as a mask to the underlying substrate.
As technology tends to minimise, making smaller patterns is always the aim. However, there is always a limit to certain technologies. In this case, the use of radiation as the transferral mechanism reaches its limit when the mask is so thin that the radiation used to expose it is diffracted around the edge of the mask too much to be effective. This is known as the diffraction limit. The diffraction limit means that only higher-nanometer (i.e. 50nm or more) resolutions can be obtained. [46]
Using another type of lithography, either electron beam or ion beam, allows this limit to be overcome, due to not requiring a mask and the more direct nature of the control of electron beams. This allows nanometre-scale elements, but very slow tracking. [46]
• Etching
The final process of the manufacture of MEMS is etching, in which selected material is removed from the bulk of the patterned layer.
In wet etching, this material is removed by dissolution through its immersion in a bath of chemical. The selected material is removed much more quickly than the desired region, and if the selection is careful, the rates are well matched. A multitude of etching chemicals or acids are available for various combinations, but the exact details of these fall outside the scope of the current work.
In certain materials, the etching process is isotropic, in which it occurs at the same rate in all directions in three dimensions. This means that a narrow rectangular notch to be etched would, after etching, become a v-shaped groove in the substrate.
However, in some materials which are single crystals, such as some forms of silicon, anisotropic rates of etching occur, which means that the rate is different in different direc- tions inside the crystal. When silicon is etched in KOH (potassium hydroxide), the silicon h111i planes etch around 1% of the speed of the other planes. This changes the shape of the etched notch from the shape masked. These are just a few of the complications and things which need to be taken into account in the wet etching process.[47] CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 37
Dry etching is a more expensive operation than wet etching technologies. It uses gases which are ionised and either react or not with the substrate; the former being named Reactive Ion Etching, and the latter Sputter etching. They have the capability to create many geometries and side-wall shapes, and are thus an enabling technology in MEMS creation. [48]
Figure 3.4: Microelectric Fabrication Steps [49] CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 38
Nylon PET Aluminium Steel Nylon 0.3 0.2 0.4 0.35 PET 0.19/0.25 0.4 0.38 Aluminium 1.15/1.4 [0.3] 0.61/0.47 Steel 0.31/0.23
Table 3.1: Friction values [50] [51] [52]
3.4 Macro-scale Materials
For the construction of the prototypes, different materials would need to be used to facilitate testing of the concept. These would obviously be different to those for the micro-scale pro- totypes, and would also depend heavily on the manufacturing processes available and decided upon.
If the prototype is to be created in a subtractive material method such as CAM-milling and turning operations, care needs to be taken as to the selection of each material in order to maximise the performance of each piece, and minimise friction.
Common materials for such manufacture of prototyping would be Nylon (P olyamide PA6), PET (polyethelynetetrephthalate), Aluminium and Steel, due to their ease of working and availability, as well as relatively low cost.
For ease of reference, the following table has been constructed comparing the coefficient of friction (static/sliding [greased]), if available, between the above mentioned materials:
The values in the table above are from many sources, due to the complexity of friction coefficients in general. The exact composition and situation of the parts are incredibly important in the design. With reference to the table, were the part to be designed and manufactured by traditional subtractive manufacturing techniques, the combination of Polyamide and Alu- minium parts would be used, due to their relative densities and workability, as well as their low combined friction coefficient (even lower when a chemical lubricant is used such as silicone- based lubricant), and also due to the fact that generally polymer friction is actually lower with differing materials than identical, so for example Polyamide - Polyamide would perform worse than Polyamide -Aluminium or Polyamide Steel. [53]
However, it was decided that the optimal way for creating a prototype would be to use rapid CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 39 prototyping and in particular 3D printing, due to the authors proficiency in CAD-CAM and the ability to create more intricate and minimised profiles for the parts, to approach a reasonable scale for the device.
The additive manufacturing machines (3D-printers) available for the purposes of the exper- iment were:
• Dimension SST 768 Fused deposition modelling, build area 203 x 203 x 305mm, accu- racy +/- 0.3mm, ABS material
• Alaris30 Polyjet Photopolymer jetting, build area 300 x 300 x 150mm, accuracy +/- 0.1-0.2 mm, Photopolymer material.
• ProJet 3510 SD Multijet printing technology, build area 298 x 185 x 203mm, accuracy +/- 0.025 - .05mm, many proprietary materials with varying properties.
As accuracy in form for a small part was key in the production and due to the fact that the Alaris was out of order for the time of production, the ProJet was the machine to be used. This only supported the choice of manufacture, however, as ABS materials are inherently brittle due to their layered production and to the fragility of the bonding between layers. This is less of an issue in ProJet materials due to their different manufacturing methods.
The following table is taken from official documentation supplied by QUT Rapid Prototyp- ing services to help choose the correct material for rapid prototyping, sourced from 3D Systems. [54]
As can be seen in the table, the preferred material would be Visijet M3 Black, due to its ability to be deformed before breaking, as the flexibility and return to the original shape of the part the most critical are the creation of the prototype. CHAPTER 3. FUNDAMENTAL THEORY OF MOTION REALISATION 40
Brand ProJet ProJet ProJet ProJet ProJet ProJet 3510SD 3510SD 3510SD 3510SD 3510SD 3510SD Material Visijet M3 Visijet M3 Visijet M3 Visijet M3 Visijet M3 Visijet M3 X Black Crystal Proplast Navy Techplast Tensile 49 35.2 42.4 26.2 20.5 22.1 Strength Tensile Modu- 2168 1594 1463 1108 735 866 lus Tensile Elon- 8.3 19.7 6.83 8.97 8 6.1 gation Flexural 65 44.5 49 26.6 28.1 28.1 Strength
Table 3.2: Polymer Material Data list Chapter 4
Design Concepts and Process
4.1 Ring field
Operating on the principles of such devices as polywells [55] this design intends to create a moving magnetic field in a tube, that would in turn move a micromagnet or a small piece of ferromagnetic material through the tube in a controllable manner with fine control.
It would consist of conductive loops contained within non-magnetic medium (i.e. SiO, SiN), which would be switched by rotating tube which allows the control of the magnetic field inside it, as shown in Figure 4.1
4.1.1 Difficulties
A primary limitation in this design is the material strength to resist the tension from the repulsion created by neighbouring loops of wire within the structure. Another perhaps greater limitation is the creation of the switching tube, as it would be incredibly small and must rotate. Yet another difficulty would be electrical contact with each loop and the tube, bushes would be necessary and difficult to create the contacts, and would also increase friction. Error states as the tube rotates makes contact between adjacent loop entries would occur. See figure 4.2 for a visualisation of the states.
All of these problems with the rotating tube could be removed by controlling each loop by the use of transistor switches, meaning that every loop or group of loops would have to be individually controlled, which would be feasible but with a high degree of complexity. This idea was abandoned in favour of the second (elaborated in the following section), due to the higher complexity of this model, although a review could be made in the future if further inspiration is found.
41 CHAPTER 4. DESIGN CONCEPTS AND PROCESS 42
Figure 4.1: Model of design 1
Figure 4.2: Visualisation of error states, error in orange CHAPTER 4. DESIGN CONCEPTS AND PROCESS 43
4.2 Stepping arms
This design is inspired by the Vernier sliding scale and various other electromagnetic propulsion methods, but applied with the myosin-actin principle of motion.
The best way to visualise the initial principle of action is by using the vernier indicating and sliding scale. Each “attachment zone” on the Actin strand (Consult section 2.1.3) is located at a point on the scale, with each Myosin-head being on the indicating scale (red arrow).
Figure 4.3: Step illustration of vernier scale principle
When the head and the attachment zone are aligned (in the first image, the two arrows), the next indicating point (number 1) is the nearest non-aligned point to the other scale, and if this point is then made to be aligned to its coinciding point (image 2), then the next, number 2, is the closest in advance.
N graduations on the indicating scale (bottom) correspond to N-1 graduations on the fixed scale. So Indicating scale spacing is (N-1)/N by the unit length. Each neighbouring position is therefore 1/N units away from its nearest point.
As an analogy to the sliding filament theory, the heads and bonding sites would be approxi- mated by permanent- and electro- magnets.
In this way, the individual units can be spaced a relatively large distance apart to minimise interference from their individual magnetic influences.
The initial design consisted of a very close analogy to the myosin-actin action, of which there is a rough sketch in Figure 4.4.
As seen in the graphic, there are a number of phases to the motion. Following the magnet on the arm from the top, the phases are as follows. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 44
In the first phase, the electromagnet is turned on to attract the magnet, functioning as the movement action. In the analogy above of the vernier scale, this would be the second set of lines, misaligned by the minimum amount (D/N).
In phase two, the electromagnet and permanent magnet are touching, and the maximum hold strength of the element is reached. The unit has moved now by (D/N).
In phase three, the succeeding set of magnets were attracted and moved forward, thus moving the sliding core forward by the step. This magnet was then compressed into the wall, bending the arm holding the magnet. Depending on the angle of the contact, and the pivot method of the beam, the arm would bend and the magnet would have some force both normal and parallel to the interface.
In the fourth phase, the unit has moved another step forward, and the arm has bent. The force has become sufficient that the electromagnet has begun to slide from the ledge. This phase would be transient and immediately after, the electromagnet will have passed into the next void, getting ready for the next alignment.
In these four phases, the functioning of the device is summarised, and the basic behaviour established.
Taking further influence from the physical reality of the Figure 4.4: Active Arm myosin and actin filaments, a three dimensional array of the Function head and bonding sites would be a convenient and efficient arrangement.
In the biological arrangement of myosin and actin filaments as mentioned above, each thick strand is surrounded by 6 thin ones (see section 2.1.3). This is a convenient arrangement for design also, and allows for a regular tessellating structure based on triangles. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 45
Also essential is a description of the behaviour of each element in the Myosin-Actin analogy.
Figure 4.5: Prototype design solid
• Myosin head element: As in the biological system, the myosin head elements would be active elements; forming linkages and providing force and movement. The strongest form of electromagnet would be the ideal element for this. They would be switched and controlled by the system.
• Actin head element: In the original design, the Actin bonding sites were to be immobile passive elements, formed by permanent magnets embedded in a holding matrix. As previously informed, Neodymium would be preferred. This was later revised, however, allowing the passive elements to be also heads on moving arms, whilst still passive. This is discussed further in the Prototyping section.
• Shaft element: The shaft element would be in the electromechanical solution an analogy to the Myosin filament bundle in the biological form. It has the role of both locating the active heads where they are supposed to be, transferring the forces generated by the action, and also carrying the circuitry involved in switching and controlling.
• Shell element: The shell element would be the electromechanical analogy to the surround- ing arrangement of Actin threads around the individual Myosin strand. Its role would be to maintain the location and orientation of the bonding sites and also function to transfer the force of the mechanism in tandem with the shaft element. It would encase the rest of the device, and also provide the connection between each parallel element. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 46
4.3 Parallelisation and Series
The biological model of Myosin and Actin consists of both parallel and series elements of each individual sarcomere, as shown above in section 2.1.3. This can be realised by the location and orientation of successive elements in the model. The single sarcomere unit length of the individual unit already would contain a number of single power units, but the unit can easily be stacked in series. This would give it the same effect as the myosin and actin filaments in the biological case, with a similar length-strength curve as in figure 2.5. Or, perhaps a more efficient but less accurate modeling would be made, by creating the units so that the passive unit can pass through the base of the active one, allowing a longer travel for fewer active elements. This is explained diagrammatically in figure 4.6.
Figure 4.6: Single-acting through-passing model
However, as the idea of this project is to approximate the biological muscle effectively and efficiently, the double acting tension motion will be kept.
In keeping this method, the Z-disks and A-zones of the sarcomeres would be kept and used as a feature, where each direction of the contractile units are joined, and where the control circuitry would be routed.
Using this model, for a given unit strength and travel, the total maximum pull strength of the actuator can be established by how many cells deep and wide the module is. Strength would sum in parallel and stroke length sum in the series direction. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 47
Figure 4.7: Double-acting Z-disk model
4.4 Control
The motion of the actuators would be very similar to a stepper motor. As a stepper motor functions using the reluctance of magnets to rotate, this has a similar effect as the proximity of the magnets used in this project.
The control algorithms would essentially be staggering the movement of each phase in series in order to maximise strength and minimise jolting, as is usual in a stepper motor configuration, and would thus have control over the performance of the actuator.
4.5 Physical Calculations
As the goal of this project is to create a minimal sized device to maximise power density, a feasibility study needs to be undertaken to establish the limitations and limiting factors of the device.
4.5.1 Qualitative Dimensional Analysis of Electromagnet
The most critical element to analyse in a small scale is the electromagnet. There already exist very small magnetic coils currently being developed as mentioned previously, yet it would be useful to know the limiting factors in further development. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 48
The most useful equations in this situation are the following:
Point Poles force
µq q F = m1 m2 (4.1) 4πr2
The point poles force equation is a simplified force equation typically used for calculating the force between two ideal dipole moments. For the purposes of this investigation, however, only the critical information is needed; that the force is related proportionally with the magnetic fields, and inversely proportionally with the square of the distance between the faces. This is a good incentive to move the magnets close to each other.
The forces between and the strength of permanent magnets are, as previously discussed, a difficult and complicated problem to calculate accurately quantitatively and as such would need careful simulation with the use of a variety of rules to obtain an accurate impression of the behaviour to be expected. It has the added complexity that the equations used for such calculations usually depend on small magnets at large distances. However, the equation given by [34] for the strength of the force between two identical bar magnets provides some insight, even with those assumptions: B LA B = o πx(x2L2) (4.2) 4
B is magnetic field, Bo is the field at the point of the end of the solenoid, x is distance, L is length of the magnet. The main important information in that equation is that the shape maximises the area whilst minimising the length, meaning a disk shape. Also important to notice is the inverse cubic relationship with distance.
Next, an equation for the strength of magnetic field generated by an electromagnet will provide another useful set of information.
µ Ni B = 0 (4.3) o 2L
This equation shows the relationship between B0 number of turns N, current in the wire i, and length of the solenoid L. In addition to the already established shape implications, it shows the importance of the number of coils and current. The number of coils would be limited to the size restrictions of the manufacturing process only, but the current in the coil has other restrictions. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 49
For example, the current in a coil generates heat and can melt its own wire or the encasing materials. The equation for the heating of a wire due to current, known as ohmic or Joule heating, is given by: P = IV = I2R (4.4)
Where P is the heat energy converted in the coil. This shows the main limiting factor of the design; the emittance of heat of the coil is dependent on the square of the current passing through it, multiplied by the resistance of the coil, R. The value for R is given by the following:
ρL R = (4.5) A
In which ρ is the resistivity of the material, L is the length and A is the cross-sectional area of the conducting material. This shows that with a constant material and Resistance, for each factor of increase of Length, the same factor of area needs to be decreased. This, however, is an inverse squared relationship with dimension of cross-section of area. This means that as a coil gets progressively smaller, its length decreases at a rate proportionally faster than the area of the wire forming it, meaning that eventually the shape would require a complete disk shape rather than a ring; consequently, an electromagnet would no longer occur.
This shows that there is a limit to how small the device can be, to handle specifications.
However, it is also important to consider the functioning of the device, as the duty cycle of the individual magnet pairs will also play a role in the limits of current and wire sizes that can be used.
Experimental and practical considerations may end up creating hard limits in the creation of the device, as, until hard lines are set, the electromagnet functioning system seems capable on a qualitative scale of performing to most physical scales, and of improving with decreasing scale.
Analysis [56] suggests that the limit to which electromagnetic forces dominate over electro- static forces in the micro domain is around 2 µm, assuming iron and nickel based alloys, which, therefore, provides a more solid guideline to the minimum size of the device. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 50
Force Calculations of Electromagnet Permanent magnet pair
In order to calculate the principle forces affecting the design of the magnetic system to be used in the device, a number of assumptions and overlapping equations are required. This is because the varying equations, including those given above, are valid within certain categorizations of magnet dimensions and distances of interaction.
4.6 Qualitative Design of Arm elements
The first step was to look at the slide-off force due to friction. As the two arms would feasibly bend equally, the actual angular deflection of the arms would be halved, as mentioned above. This gives rise to the diagrams below:
Figure 4.8: Bending Arms
The equation for the value of the friction force, which resists movement in any direction, is:
Ffr = µN (4.6)
In which Ff r is the force of friction, µ is here (differently to in other equations used in this project) the friction constant, most importantly the value of static friction, and N is the force normal to the plane of interaction. CHAPTER 4. DESIGN CONCEPTS AND PROCESS 51
Figure 4.9: Friction Force Diagram
From a simple analysis of the state of the device at the time, the following are attainable:
N = F sin θ (4.7)
P = F cos θ (4.8)
Which then means that the force when P = Ffr is given by (substituting the two above equations into the previous): F cos θ = µF sin θ (4.9) F cos θ = µ (4.10) F sin θ tan µ = θ (4.11)
Once the material is known, this will give the slip-off angle.
By analysing the general physical state at the time of slip-off, as above, and assuming that the arm behaves as a cantilever beam with a uniform bending profile subject to axial loading, the following equation can be used to find the angle of the end of the beam due to the moment and the force applied.
According to Shigley’s Mechanical Engineering Design [57] which uses this case, the total strain energy function can be found by:
Z h P 2y2 Z L (P h + Qx)2 Q2h P 2h U = dy + dx + + (4.12) 0 2EI 0 2EI 2EA 2EA
Which can then be used to find the angle of deflection caused by the applied moment created CHAPTER 4. DESIGN CONCEPTS AND PROCESS 52
by the force applied at distance h from the axis, as the following:
∂U Z h M 2 θA = = dMdy (4.13) ∂M 0 2EI Z h M Z h P y P y2 h = dy = dy = (4.14) 0 EI 0 EI 2EI 0
Which then leads to the useful design construct that:
P h2 θ = (4.15) A 2EI
And all of these factors are design factors that can be established by the geometry and properties of the constructing material.
An alternate way of calculating this angle would also be to begin at the beam deflection equation: d2y P P e + y + = 0 (4.16) dx2 EI EI In which y is deflection, x is distance along beam, P , E, and I are as in the previous equations, and e is the eccentricity, being the distance from the centroid axis that the force acts.[57] Applying the boundary conditions that y(x) = 0 and y(x) = 0 due to the fixed nature of the beam, the equation for deflection can be established to be:
r P y(x) = e(cos x − 1) (4.17) EI
The derivative of this function provides the slope of the line formed by the beam then:
q P −P e sin( EI x) y(x) = (4.18) q P EI EI
And the arctan of which can be used to find the actual angle of the end of the beam.
q P −P e sin( EI x) θ(x) = tan−1 (4.19) q P EI EI
These could also be combined with the previously attained function for the slide-off angle CHAPTER 4. DESIGN CONCEPTS AND PROCESS 53 of the basic design, to establish the maximum force before slide-off of the arm system with the basic design:
With the first equation: P h2 tan µ = θ = (4.20) 2EI 2EI P = tan µ (4.21) h2
Or for the second: Letting x be length (L)