Determination of Local Elastic Modulus of Soft Biomaterial ...
Transcript of Determination of Local Elastic Modulus of Soft Biomaterial ...
Determination of Local Elastic Modulus of Soft Biomaterial Samples Using
AFM Force Mapping
By
Faieza Saad Ejweada Bodowara
Bachelor of Science, Benghazi University
Master of Science, Benghazi University
A Dissertation
submitted to the Department of Chemistry of
Florida Institute of Technology
in partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
In
Chemistry
Melbourne, Florida
January 2018
ii
Determination of Local Elastic Modulus of Soft Biomaterial Samples Using AFM
Force Mapping
a dissertation by
Faieza Saad Ejweada Bodowara
Approved as to style and content
_______________________________________
Boris Akhremitchev, Ph.D. Committee Chairperson
Associate Professor, Department of Chemistry
________________________________________
Joel Olson, Ph.D.
Associate Professor, Department of Chemistry
________________________________________
Kurt Winkelmann, Ph.D.
Associate Professor, Department of Chemistry
________________________________________
Vipuil Kishore, Ph.D.
Associate Professor, Department of Chemical Engineering
Michael Freund, Ph.D. __________________________
Professor and Head, Department of Chemistry
iii
Abstract
Determination of Local Elastic Modulus of Soft Biomaterial Samples Using
AFM Force Mapping
by
Faieza Saad Ejweada Bodowara
Major Advisor: Boris Akhremitchev, Ph.D.
Conventional methods of mechanical testing cannot measure properties of soft
materials at the nanoscale. In fields such as tissue engineering, it is important to distinguish
the bulk elastic modulus from the surface elastic modulus and to characterize the spatial
distribution of material with non-uniform stiffness. One of the important new methods of
testing is the force mapping using the atomic force microscope. The existing force
mapping approaches often suffer from omitting important effects that might result in
artifacts. The most important effects include taking into account sample’s adhesion and
viscoelasticity and considering more realistic probe shapes. In this work, we have applied
a method that take into consideration probe shape and sample adhesion and developed
elastic modulus mapping. This inclusive approach can be applied to samples with adhesion
that exhibit indentation large enough that requires indenter shape models more
sophisticated than the traditional paraboloid shape model. Sample viscoelasticity can be
revealed by comparing parameters obtained in opposite scanning directions.
The application developed methodology is illustrated with the study of composite
biomaterial scaffolds that are used to support the differentiating cells. Samples are
composed of four groups: uncrosslinked electrochemically aligned collagen (ELAC),
iv
uncrosslinked Bioglass incorporated ELAC (BG-ELAC), genipin crosslinked-ELAC and
crosslinked electrochemically compacted collagen (ECC, unaligned). The force mapping
on BG-ELAC sample did not exhibit any area with high elastic modulus (measured
modulus was around 0.6 MPa), indicating that there is no Bioglass particle protrusion
observed at the BG-ELAC surface. Elastic modulus of ELAC molecules appears softer (~
0.1 MPa) for samples without Bioglass added. Adding genipin crosslinker to collagen
threads made the ELAC and ECC samples stiffer even more than uncrosslinked BG-ELAC
with characteristic values of elastic modulus approximately 0.97 MPa and 1.28 MPa for
crosslinked-ELAC and crosslinked ECC respectively. More extensive studies are
necessary to fully investigate effect of crosslinking ratio and adding Bioglass at different
concentrations on the elastic modulus values of collagen samples. Methodology reported
here is a suitable tool for such studies and can be applied to other soft and potentially
heterogeneous biomaterial samples.
v
Contents
Symbols vii
Abbreviations ix
List of Figures x
List of Tables xvi
Acknowledgement xix
Chapter1 Introduction
1.1 Why are Mechanical Properties Measured? 1
1.2 Mechanical Properties in Nanoscale 2
1.3 Parameters for Studying Mechanical / Nanomechanical Behavior of Materials
4
1.4- Mechanical Properties Measurements 5
1.5 Nanoindentation Techniques 15
1.5.1 Nanoindentation by Atomic Force Microscope 17
1.5.2 Forces during the Tip-Sample Contact 20
1.6. Measurements of Nanomechanical Properties by AFM 22
1.6.1 Tensile/ Bending Testing 24
1.6.2 Applications of Dynamic AFM to Characterize Mechanical
Properties 25
1.6.2.1 Force Modulation AFM 25
1.6.2.2 AFM Phase Imaging 29
1.7 Current Applications of AFM Nanoindentation on Soft Materials and Associated
Challenges 30
1.7.1 Effects of Viscoelasticity in Force-Indentation Measurements 32
vi
1.7.2 Cantilever and Tip Characteristics 34
1.7.3 Effects of Indentation depth 35
1.7.4 Effects of Model Selection 37
1.7.5 Problems with Contact Point Identification 39
1.8 The Aim of Approach 40
1.9 References 45
Chapter 2 Experimental Section 68
2.1 Samples and AFM Setup 68
2.2 Experimental Design 69
2.3 Force Mapping 73
2.4- Data Analysis 74
2.4.1 Calibration of Cantilever Sensitivity and Force Constant 74
2.4.2 Indentation Models 78
2.5 References 83
Chapter 3 Results 85
Chapter 4 Discussion 98
4.1 Elastic Modulus Measurements 98
4.2 Viscoelasticity 102
4.3 Heterogeneity of Surface 107
4.4 Depth Sensing and Tip Shape Artifacts 111
4.5 Conclusion 118
4.6- References 120
vii
Symbols
H Hardness of material
Pmax Maximum load
A Projected area
ν, Poisson ratio
E Elastic ( Young’s modulus)
E* Reduced Young’s modulus
S Stiffness of material
Smax The initial contact stiffness
P Applied load
h The deformation of sample
R Radius of curvature
a Radius of contact area
hf The final depth of the residual hardness impression
hs The drift of the surface at perimeter of contact
hc Vertical displacement during the contact
hmax Total displacement at any time during loading
Vt Photodiode voltage
∆Zp Vertical displacement of piezo actuator
Zp The position of the base of the cantilever
U1 The potential energy of the cantilever spring
U2 The potential energy pf tip-sample interaction
Ut Minimum potential at the equilibrium position of the tip
Us Potential energy due to the sample deformation
viii
Uc The energy resulting from the bending of the cantilever
Ucs Tip-sample potential caused by surface forces
d The cantilever deflection
kc The spring constant of the cantilever
ks Sample stiffness
kinteraction Spring constant of tip-sample interaction
D Tip-sample separation
δ The indentation depth
L The length of the cantilever
w Cantilever width
θ Cantilever thickness
C* A constant related to the selection of tip-sample contact point
α The semivertical angle of the tip
ix
Abbreviations
LH2 Liquid hydrogen
CNTs Carbone nanotubes
ECM Extracellular matrix
SEM Scanning electron microscopy
TEM transmission electron microscopy
AFM Atomic force microscope
ROC Radius of curvature
vdW van der Waals
PEO polyethyleneoxide
LRW London Resin White (plastic)
JKR Johnson, Kendall and Roberts Theory
DMT Derjaguin, Muller and Toporov Theory
ELAC Electrochemically aligned collagen
BG-ELAC Bioglass incorporated ELAC
ECC Electrochemically compacted collagen
Std Standard deviation of values
x
List of Figures
Figure 1.1 shows a stress- strain curve and illustrates different kinds deformation
5
Figure 1.2 shows Hertz model for studying elastic contact between two surfaces with
different radii and elastic constant 6
Figure 1.3 shows a load-displacement indentation curve for typical materials with an
elastic-half space characteristics 8
Figure 1.4 shows a scheme of the across section through the indentation of a typical
sample in the case of elastic-half space 12
Figure 1.5 illustrates the key parameters in Fig. 1.4 for an elastic half space material
13
Figure 1.6 shows a principle scheme of atomic force microscope 18
Figure 1.7 shows a deflection-displacement curve involving different types of tip-sample
interactions 19
Figure 1.8 represents the forces between tip and sample surface as a function of tip-sample
distance 22
Figure 1.9 displays force curves of three different materials: a) elastomer, b) purely plastic
(gold) and c) the most common elastic-half space material (graphite) 23
Figure 1.10 shows the scheme of bending/tensile AFM testing 25
xi
Figure 1.11 shows force-modulation imaging of carbon fiber and epoxy composite.
Panels in the figure show a) topographic image, b) stiffness in the force modulation image.
Here bright spots correspond to stiff areas. Image width is 32 µm 27
Figure 1.12 illustrates topography (A) and elasticity (B) of etched LRW embedded
sample. Panal B shows the difficulties of measuring elastic differences in the
components 29
Figure 1.13 shows the dependence of elastic modulus of polyethylene glycol on indenter
rate and temperature. The elastic modulus changes significantly at high indenter rate and
low temperature 34
Figure 1.14 represents a cross section showing osteons. A unit of cells, an osteocyte, has
a 5-20 µm diameter, is 7µm deep, and contains 40-60cells 42
Figure 1.15 displays a cartoon of an artificial scaffold with cells attaching between
collagen fibrils in microscale 43
Figure 2.1 shows photograph of 1 cm long and 1 mm wide filament of ELAC sample
deposited on a microscopic glass slide 70
Figure 2.2 shows SEM image for the tip of a long wide rectangular cantilever used in the
experiments 71
Figure 2.3 shows images for clean-dry glass surface before fixing the sample. The A
panel shows topographic image and the B panel shows the deflection image 72
xii
Figure 2.4 shows the typical deflection (A) and height (B) image of ELAC sample before
force mapping at scan size 20 μm 73
Figure 2.5 The dependence of voltage value on the cantilever motion. The repulsive force
bends the cantilever up and shifts the laser, and hence the voltage, to the positive value
75
Figure 2.6 shows a deflection vs. Z displacement curve on a hard surface 76
Figure 2.7 shows a typical force-distance curve 78
Figure 2.8 Shows a typical force-indentation curve 79
Figure 2.9 shows comparison of paraboloidal and hyperboloidal shapes that have the same
radius of curvature R. In the figure, α is the semi-vertical angle of hyperboloidal probe
80
Figure 3.1 A-D show deflection (left) and height (right) images at 15 µm scan size. Panels
A-D show images of typical ELAC, 30% BG-ELAC, crosslinked-ELAC, and crosslinked
unaligned collagen samples, respectively 86
Figure 3.2 displays fitting typical approach (A) and withdraw (B) force plots of ELAC
sample using paraboloidal probe shape. Data analyzed to maximum compressive
force of Fmax = 1 nN. 87
xiii
Figure 3.3 shows fitting typical approach (A) and withdraw (B) force plots of BG-ELAC
sample using paraboloidal probe shape. Data analyzed to maximum compressive force of
Fmax= 1 nN 87
Figure 3.4 displays fitting typical approach (A) and withdraw (B) force plots of
crosslinked-ELAC sample using paraboloidal probe shape. Data analyzed at Fmax = 1
nN. 88
Figure 3.5 shows fitting typical approach (A) and withdraw (B) force plots of crosslinked
unaligned collagen sample using paraboloidal probe shape. Data analyzed at
Fmax = 1 nN 88
Figure 3.6 shows the maps of elasticity of crosslinked-ELAC at the three different scan
sizes, and approach and withdraw scan directions using 1 nN maximum force and
paraboloidal probe shape model. Numbers on axes indicate pixel position. Blue pixels
show places were collected force plots could not be analyzed 90
Figure 4.1 displays the average values of Young’s modulus of ELAC, BG-ELAC,
crosslinked-ELAC and crosslinked-ECC collagen threads at A) Fmax = 1 nN, and B) Fmax
= 2 nN. Standard deviations are shown as error bars 99
Figure 4.2 shows the force maps of one spot of crosslinked ECC unaligned sample at the
three different scan sizes 100
xiv
Figure 4.3 displays the approach-withdraw offset for ELAC, BG-ELAC, crosslinked-
ELAC and crosslinked-ECC, unaligned samples at Fmax = 1 nN using paraboloidal tip
model. The numbers inside the chart are the positions of histograms ± one standard
deviation error in this position 104
Figure 4.4 shows the approach-withdraw offset for the different spots using data collected
with 15 μm scan size 105
Figure 4.5 displays the slow increase of viscoelasticity mean value of crosslinked-ELAC
with increasing the scan sizes as more higher modulus areas are tested 107
Figure 4.6 shows force maps and their corresponding histograms of ELAC sample at 15
µm scan size 108
Figure 4.7 shows elastic modulus maps and corresponding histograms measured on ELAC
sample at 640 nm scan size 109
Figure 4.8 shows elastic modulus maps and their corresponding histograms of BG-ELAC
at different scan sizes: a) 640 nm, b) 3 µm, and c) 15 µm 110
Figure 4.9 displays force maps of ELAC and BG-ELAC, crosslinked-ELAC and
crosslinked-ECC, unaligned samples at 640 nm scan size. The higher modulus samples
show sharper features 111
xv
Figure 4.10 Force maps and corresponding histograms of ELAC, BG-ELAC, crosslinked-
ELAC and crosslinked-ECC, unaligned samples using paraboloidal tip at different
indentation depths 114
Figure 4.11 The paraboloidal (gray) and hyperboloidal (blue) probes as expected at small
( Fmax = 1 nN) and large ( Fmax = 2 nN) indentation depths 116
Figure 4.12 in panels a and b show the difference in E values by using paraboloid and
hyperboloid probes over different scan sizes for crosslinked-ELAC sample at Fmax = 1 nN
and 2 nN, respectively 117
xvi
List of Tables
Table 3.1. Elastic modulus values and their corresponding standard deviations (Std) of
sample 1, 1st spot map (1024 force curves) of ELAC sample using maximum compressive
forces of 1 nN and 2 nN, extracted by fitting with JKR model 90
Table 3.2. Elastic modulus values and their corresponding standard deviations (Std) of
sample 1, 2nd spot map (1024 force curves) of ELAC sample using maximum compressive
forces of 1 nN and 2 nN, extracted by fitting with JKR model 91
Table 3.3. Elastic modulus values and their corresponding standard deviations (Std) of
sample 2, 1st spot map (1024 force curves) of ELAC sample using maximum compressive
forces of 1 nN and 2 nN, extracted by fitting with JKR model 91
Table 3.4. Elastic modulus values and their corresponding standard deviations (Std) of
sample 2, 2nd spot map (1024 force curves) of ELAC sample using maximum compressive
forces of 1 nN and 2 nN, extracted by fitting with JKR model 92
Table 3.5. Elastic modulus values and their corresponding standard deviations (Std) of all
spots on ELAC samples using maximum compressive forces of 1 nN and 2 nN, extracted
by fitting with JKR model 92
Table 3.6. Elastic modulus values and their corresponding standard deviations (Std) of
sample 1, 1st spot map (1024 force curves) of 30% BG-ELAC sample using maximum
compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model 93
xvii
Table 3.7. Elastic modulus values and their corresponding standard deviations (Std) of
sample 1, 2nd spot map (1024 force curves) of 30% BG-ELAC sample using maximum
compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model 93
Table 3.8. Elastic modulus values and their corresponding standard deviations (Std) of
sample 2, 1st spot map (1024 force curves) of 30% BG-ELAC sample using maximum
compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model 94
Table 3.9. Elastic modulus values and their corresponding standard deviations (Std) of
sample 2, 2nd spot map (1024 force curves) of 30% BG-ELAC sample using maximum
compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model 94
Table 3.10. Elastic modulus values and their corresponding standard deviations (Std) of
all spots on 30% BG-ELAC samples using maximum compressive forces of 1 nN and 2
nN, extracted by fitting with JKR model 95
Table 3.11. Elastic modulus values and their corresponding standard deviations (Std) of
one force map (1024 force curves) of crosslinked-ELAC sample using maximum
compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model 95
Table 3.12. Elastic modulus values and their corresponding standard deviations(Std) of
1st spot on crosslinked unaligned collagen sample using maximum compressive forces of
1 nN and 2 nN, extracted by fitting with JKR model 96
xviii
Table 3.13 Elastic modulus values and their corresponding standard deviations (Std) of
2nd spot on crosslinked unaligned collagen sample using maximum compressive forces of
1 nN and 2 nN 96
Table 3.14. Elastic modulus values and their corresponding standard deviations (Std) of
all spots on crosslinked unaligned collagen sample using maximum compressive forces of
1 nN and 2 nN, extracted by fitting with JKR model 97
Table 4.1. Elastic modulus values (MPa) and their standard deviations (Std) for one spot
on crosslinked-ELAC at (Fmax = 1 nN and 2 nN) using paraboloidal and hyperboloidal
models 115
xix
Acknowledgement
I would like to thank Almighty God for giving me the strength and ability to
undertake and accomplish this research satisfactorily. By his blessings, my dream to study
the Ph. D in new and interesting science became reality. The financial support of the
Libyan Ministry of Higher Education played an important role during the first five years
of this study. I would like to thank my country, Libya, for giving me an opportunity to
start my Ph. D program in the United States of America.
I would like to thank Dr. Boris Akhremitchev, my advisor and the source of
valuable scientific insights. Thank you for accepting me in your laboratory, engaging me
in my desirable field and enhancing my ability as a scientific researcher. I would like to
acknowledge the Department of Chemistry at Florida Institute of Technology for providing
me extensive courses and valuable seminars. I would also like to thanks my committee
members, Dr. Joel Olson, Dr. Kurt Winkelmann and Dr. Vipuil Kishore for their interest
in my research, and their wise extensive guidance. I wish to express my deep gratitude to
Dr. Vipuil Kishore, along with Madhura Nijsure and Kori Watkins, for enabling me to
reach my goal of studying a very interesting area by providing me samples, and for being
infinitely patient. I am also grateful to Anad Mohamed Afhaima, my friend and colleague,
for her very useful discussions, advice, and emotional support.
I would like to express my thanks to my Dean, Dr. Hamid Rassoul, who is a source
of much wisdom and endless encouragement. He was wonderful at solving problems,
removing obstacles, and offering good suggestions. I will never forget Dr. Mary Sohn for
xx
her help and support while was serving as the head of the Chemistry Department. Her
good attitudes were continued by Dr. Michael Freund, who took over that position and
continuously encouraged and supported my work.
My family has always been most supportive and encouraging. I would like to take
this opportunity to thank them all, my mother, brothers and sisters, whose love and
guidance were my motivation to pursue my goal, especially my mother, her prayer and
sacrifices for me was what made me brave enough to accept the challenges. Last, but
certainly not least, I am extremely grateful to my husband, who has stood by me throughout
this endeavor and who has never doubted my ability to succeed, your love has sustained
me.
1
Chapter 1 Introduction
1.1 Why are Mechanical Properties Measured?
The use of different kinds of materials depends on their mechanical properties, from
hard structural materials to soft biomaterials. Knowing the mechanical properties of
materials can facilitate their improvement [1]. For example, the cross linking process is
sometimes recommended in plastics for improving their resistance to mechanical and
chemical influences [2]. Here are several examples of processes where knowing the
mechanical properties of polymeric materials is essential for successful development.
Because of the rapid development of plastics and its derivatives, the plastic recycling
process has become one of the largest global concerns over the last several decades. To
prevent pollution and to recover both energy and material, mechanical properties of plastic
materials have been manipulated to produce easily recycled plastics. To achieve this goal,
investigation of the properties that are responsible for plastic degradation, e.g. the ability
to granulate, and hence to reuse the reground materials have definitely been successful [3].
In the space field, a tank made from thin polymer is used to store liquid hydrogen LH2,
which is used to protect the astronauts from the high level of cosmic radiation at cryogenic
temperatures. Investigation of mechanical properties such as permeability and Young's
modulus of the film has been performed at cryogenic temperatures to monitor the stiffness
of the LH2 container and hence to preserve the LH2 level [4]. In membrane manufacturing,
mechanical properties for membranes used for water filtration, methanol fuel cells, coating
and other important applications are undoubtedly worthwhile. Membranes should exhibit
a proper strength that keeps their integrity during the achievement of their role even in
2
critical environments such as high pressure or temperatures [5]. Lastly, it was recently
demonstrated that stiffness of substrates affects development and locomotion of cells [6,
7]. Therefore, knowing the local elastic modulus is important for the development of
approaches aiming at controlling cell behavior.
1.2 Mechanical Properties at Nanoscale
Development of the nanocomposite industry has provided new functional materials.
Mechanical performance can be controlled by adjusting the mechanical properties of
material in nanoscale [8], even better than those of micro-particle-filled composites [9].
For example, the interfacial strength between the fiber and matrix in nanocomposite
polymeric material is responsible for the general behavior of this material [8]. Carbon
nanotubes (CNTs) have the highest strength and stiffness and the lowest density among all
known materials [10]. Mixing polymeric material with CNTs creates smart (damage
resistant with high surface to volume ratio) polymer matrix nanocomposites and improves
the material mechanics [11]. CNTs are also widely used in polymer reinforcement and
coating surfaces that require a robust mechanical performance such as in gears, drive shafts,
body panels in airplanes [12,13], and instrumentation sensors such as probes in atomic
force microscopy [14].
Decreasing sizes of materials to nanoscopic scale has become desirable to achieve
durability and strength [15]. Over the last several years, understanding and probing
nanomechanical properties have become of significant interest. In the biological field, the
investigation of nanomechanical properties can lead to identifying biofunctions and
3
diagnosing many health problems [16]. For example, decreased rigidity is a marker for
cancer cells, in addition to other diseases such as Alzheimer's dementia, and vascular and
kidney diseases that occur due to losing the elasticity of cells [17], while stiffening of red
blood cells in malaria is correlated with the deaths due to this disease [18]. Therefore,
changes in mechanical properties of cells may affect disease progression.
One important recent application of nanoscience technology is in the field of tissue
engineering, where substrates called scaffolds are tailored to support growing cells. The
scaffolds are designed to mimic the in vivo chemical, topographical, and mechanical
properties of the organ [19]. To build a strong tissue, scaffolds first should be mechanically
strong to be easily handled. Second, the scaffold and tissue site in which the former is to
be implanted have to be mechanically consistent with each other. Third, scaffolds should
promote cell adhesion, which also strongly depends on consistency of the mechanical
properties between cells and scaffolds that should be similar to that between cells and the
extracellular matrix, ECM [20-30].
Development of materials with new and improved functions by manipulating
nanoscale characteristics requires quantification of the mechanical properties of materials
in that scale. This requires knowledge of important parameters, methods and theories that
have been used to determine nanomechanical properties of materials.
1.3 Parameters for Studying Mechanical / Nanomechanical Behavior of Materials
One of the first quantities that has been determined as a mechanical characteristic
of material is the hardness, H, which is a measure of material resistance to indentation, or
4
how much the material will deform under a load [31]. When a hard ball is pressed with a
fixed normal load into a smooth hard surface and equilibrium has been reached, the
indenter and hence the load are removed and the diameter of the permanent impression is
measured. Then, the hardness of the smooth surface is the ratio of the maximum load (at
the equilibrium Pmax) to the projected area, A, of the indentation [32]
H = 𝑃𝑚𝑎𝑥
𝐴 (1.1).
One more mechanical property of materials is the Poisson’s ratio, ν, which
expresses how much the material extends orthogonally to the direction of applied force. It
is the ratio of the transverse / orthogonal strain to the strain along the direction of elongation
[33]. The Poisson’s ratio for soft materials approximately equals 0.5 [9].
The elasticity (elastic modulus) of material, E, is the ratio of stress to strain, where
stress is the deformation force per contact area, and the strain is the amount of the
deformation relative to the unperturbed state [34]. The elastic modulus is the material's
resistance to being elongated or compressed elastically. It is a constant value which can be
extracted from the linear portion of a stress-strain diagram. Figure 1.1 illustrates the stress
/ strain relationship for different kinds of materials. The deformation is elastic as long as
the relation is linear.
5
Str
ess
Strain0
A
B
C
D
E
Elastic region
Plastic region
A – Elastic limit
B – Yield strength
D – Ultimate strength
E – Fracture pointYoung Modulus
Stress
Strain
Figure 1.1 shows the stress- strain curve and illustrates different kinds of deformation. The slope
of the linear region is the elastic (or Young’s) modulus [http://www.totalmateria.com/]
The stiffness, S, of material is an amount of force, P, that is required to deform the material
structure [35]
S = 𝑃
ℎ (1.2)
where h is the deformation of the sample caused by the force along the same degree of
freedom.
1.4 Mechanical Properties Measurements
One of the first analytical approaches to study mechanical properties of materials
was by Boussinesq in 1885 [36]. He first used potential theory to compute stresses and
displacements for a homogenous-elastic body loaded vertically by a rigid-axisymmetric
6
indenter [36]. Then, the same procedure was used to derive solutions for other common
indenter geometries such as conical and cylindrical [37, 38]. In 1896, Hertz recognized
that the elastic contact plays a key role in mechanical properties determination [31] since
the first response to interactions of two objects is elastic deformation [39]. He considered
two elastic bodies with different radii of curvature and different elastic constants in contact
at a point. Figure 1.2, [39] illustrates Hertz’s model for the elastic response when R1= ∞
for a flat surface and R2 = R for a spherical indenter [39].
h
a
Sample
RP
Indenter
Figure 1.2 shows Hertz model for studying elastic contact between two surfaces with different radii
and elastic constant [39].
By applying the load P, the bodies deform and create a contact area with radius of contact,
a, which increases with the penetration depth, h, according to the equation:
𝑎 = √𝑅ℎ (1.3)
7
For elastic sample, removal of the load decreases the contact area back to the initial point
contact. For an applied load, P, and the indenter radius R, the reduced elastic modulus E*
can be obtained from the contact area radius – load dependency:
𝐸∗ = 3𝑃𝑅
4𝑎3 (1.4)
where
1
𝐸∗=
(1 − 𝜈12)
𝐸1+
(1 − 𝜈22)
𝐸2 (1.5)
E1 and ν1 are the elastic modulus and the Poisson ratio of the flat surface, and E2 and ν2 are
the elastic modulus and the Poisson ratio for the indenter. Unlike Boussinesq's model,
Hertz's model can be used to analyze the effects of non-rigid indenters, since it can deal
with a wide range of indenter elastic constants [31]. Hertz's approach is a basis for many
experimental and theoretical works in contact mechanics [40]. However, Hertz's model
can be used only in the case of a very low surface energy (surface adhesion), and the
indenter load should be much higher than surface forces so the latter can be neglected.
Meanwhile, the load should be limited to prevent wear and plastic deformation [41, 42].
Most materials exhibit a mixture of elastic and plastic behavior, and the indentation
process might exhibit inelastic deformation. Moreover, application of the Hertzian model
is limited to few simple shapes of indenters. Consequently, this approach has been replaced
by elastic-half-space analysis, which was introduced by Sneddon [43]. He developed the
Hertizian theory further by using general indenters of axisymmetric shape that can also be
8
described as a solid of revolution of a smooth function (e,g, a flat ended cylindrical punch,
a parabolic of revolution, and a cone, and other shapes) to indent elastic half-space.
In the early 1970s, as an extension of Sneddon’s work, Bulychev, Alkhin, and co-
workers [44-48] were interested in measuring the elastic modulus from load-displacement
indentation. They used a microhardness testing machine to obtain load-displacement
curves, Figure 1.3 [35], and determined the elastic modulus by using the equation:
S = d𝑃
dℎ =
2
√𝜋 𝐸𝑟 √𝐴 (1.6)
hf
Displacement
Lo
ad
hmax
Pmax
S = dP / dhloading
unloading
Figure 1.3 shows the load- displacement indentation curve for typical materials with an elastic-
half-space characteristic. hf represents the final depth of the residual hardness impression [31].
where S is the stiffness of the upper linear portion (the elastic behavior) of the unloading
data, Er is the reduced modulus, and A is the projected area of elastic contact, which is the
optically measured area of hardness impression. Therefore, the modulus of material can
9
be determined by measuring the initial unloading stiffness from the slope of the unloading
curve and the contact area [31]. This equation was originally established for a conical
indenter in elastic contact theory, though Bulychev et al. obtained reasonable results with
this equation using spherical and cylindrical indenters, and expected the same applicability
for other indenter geometries with square and triangular cross sections. Parr, Oliver and
Brotzen also emphasized that the equation is suitable for any indenters with a body of
revolution of smooth function (spherical and cylindrical) [49]. In turn, King in 1987
confirmed by the finite element calculation that this equation can be applied also on
indenters of square and triangular cross sections, which are not bodies of revolution of a
smooth function [50].
Due to the fact that thin film industries and hence mechanical investigations on a
very small scale have become one of the foremost concerns during 1980's, instruments of
submicron indentations, which are called traditional indenters, have been developed [51-
54]. However, imaging a hardness impression for a very small indentation as a procedure
to obtain the contact area was very difficult and time consuming. Focusing on this problem,
Oliver, Hutching and Pethica proposed a simple method to determine the contact area in
terms of the indenter area function; it is a cross-section area of the indenter as a function
of the distance from its tip (contact depth) [55,56]. At the peak load, the material deforms
and takes the shape of the indenter at some depth, which can be obtained from the load-
displacement data in Figure. 1.3. There are two obvious depths that can easily be
extrapolated from the load-displacement curve, the maximum displacement at the
10
maximum load, hmax, and the residual depth after final unloading, hf. Then, the projected
area of contact can be estimated from the shape function [56].
In 1986, Doerner and Nix provided an approach, which was considered as the most
comprehensive method until 1992, interpreting the load-displacement data acquired from
deep-sensing indentation instruments to determine the elastic modulus of thin films.
According to this approach, the elastic indentation can be subtracted from the total
displacement to calculate the hardness of material, and the elastic modulus is extracted
from the initial unloading stages (unloading stiffness) [57]. By extrapolating the depth at
zero load, the area of contact can also be calculated using the Oliver, Hutching, and Pethica
approach [55, 56] to find the resolution. The authors observed in some materials that the
elastic behavior of the initial unloading stages is similar to that of a flat cylindrical punch.
In other words, during this stage the area of contact remains unchanged and hence its
unloading portion is linear. Therefore, the slope of the linear portion dP / dh (unloading
stiffness) can be determined. However, the flat-punch approximation is not completely
suitable for real material behavior. It was found that a large number of materials do not
have any linear part in their unloading curve. In 1992 [31], the indentation load-
displacement data for fused silica, soda-lime glass, a single crystal of aluminum, tungsten,
quartz, and sapphire were obtained using Berkovich (a three sided pyramid) indenters, and
the nonlinearity in unloading curves for each material was observed. Dynamic
measurements of contact stiffness proved that the reason for this curvature reflects changes
in the contact area continuously as the indenter is withdrawn [31].
11
To analyze these complex data, modeling Berkovich indenters, a method of taking
into account the curvature in unloading curves and determining the depth in the absence of
linear portions have been reported by Oliver and Pharr [31]. Based on the Sneddon theory,
which had derived analytical solutions for several indenter geometries, conical and
paraboloid of revolution shapes were selected to compare with elastic unloading of
Berkovich indenters. Like Berkovich indenter, the conical indenter has a cross section area
that is a function of h2. Additionally, the conical indenter tip is similar to that of the
Berkovich indenter; they have similar geometries at a small indentation depth. The
paraboloid of revolution was chosen as a realistic indenter (the real indenters are rounded
at the apex because it is impossible for indenter to be perfectly sharp). Previous methods
had attempted to manipulate the unloading curvature, in which a straight line of the upper
portion of the unloading curve was taken to measure the slope (the stiffness). However,
the stiffness value depends on how much of the data points are in this line. Further, the
stiffness obtained from the first and final unloading are very different because of the creep
in the first curve. Attempts have been made to minimize the creep by holding the indenter
for a period of time before the final unloading, yet a problem related to the data point-line
fitting still exists.
It was found that the stiffness obtained from the power law, Sneddon's expression,
for the first unloading data is only a little greater than that for the final curve. This indicates
that the creep is not observed and hence the power law is recommended to be used for
analyzing data. Parameters used in this analysis are shown in Figure 1.4 [31], which is a
scheme of a cross section of indentation.
12
a
P
hf
hc
hs
hmax
Surface profile after
load removalIndenter
Surface profile under load
Initial surface
Figure 1.4 shows a scheme of the across section through the indentation of a typical sample in the
case of elastic half space [31].
In this figure, hmax is the total displacement at any time during loading, hc is the vertical
displacement during the contact, hs is the drift of the surface at the perimeter of contact,
and hf is the final depth of the residual hardness impression after the indenter is fully
withdrawn and the surface has elastically recovered. Another scheme that illustrates the
key parameters is given in Figure 1.5 [31]. Based on Sneddon's model, the loading-
unloading data were analyzed in the elastic response region. Because of the nonlinearity,
the initial contact stiffness, Smax, was measured only at the peak load, Pmax,
Smax = 𝑃𝑚𝑎𝑥
ℎ𝑚𝑎𝑥 (1.7)
Then, the elastic modulus can be determined by
Er = 𝑆𝑚𝑎𝑥√𝜋
2 √𝐴 (1.8)
13
Load
hmax
Pmax
S = dP / dhloading
unloading S
hc forE = 1
hc forE = 0.72
Possible range
for hc
Figure 1.5 illustrates the key parameters in Fig. 1.4 for an elastic half space material [31].
To obtain the contact area, A, the indenter geometry and hence its area function, f (h),
should be known. Then, the projected area can be calculated at the peak load by
A = f (hc) (1.9)
where
hc = hmax – hs (1.10)
The hmax can be experimentally measured at the peak load. According to Sneddon's
expression for the shape of the surface outside the contact area [44], hs for the conical
indenter is given by:
hs = (𝜋 –2) (ℎ – ℎ𝑓)
𝜋 (1.11)
14
where (h – hf) represents only the elastic displacement. The Sneddon's expression of the
stiffness, S, over this distance for the conical indenter is:
S = 2 𝑃
(ℎ – ℎ𝑓) (1.12)
Therefore: for the projected area at the peak load Pmax
hs = ɛ 𝑃𝑚𝑎𝑥
𝑆𝑚𝑎𝑥 (1.13)
where ɛ equals to 2(π –2) /π ≈ 0.72, 0.75 or 1 for a conical indenter, a paraboloid of
revolution indenter or a flat punch indenter, respectively. Hence, hs for the flat punch
equals Pmax / Smax, then hc is given by:
hc = hmax – 𝑃𝑚𝑎𝑥
𝑆𝑚𝑎𝑥 (1.14)
which is the same as that in the Doerner and Nix expression for a flat punch. For a conical
indenter:
hc = hmax – 0.72 𝑃𝑚𝑎𝑥
𝑆𝑚𝑎𝑥 (1.15)
and for a paraboloid of revolution:
hc = hmax – 0.75 𝑃𝑚𝑎𝑥
𝑆𝑚𝑎𝑥 (1.16)
15
Using ɛ = 0.72 or 0.75 depends on which of the indenter geometries are the best for
describing the experimental data. Depending on the power law exponent, m, values that
fit in the unloading curve, the unloading was best described by a paraboloid of revolution
[31]. Although the difference with the conical ɛ is insignificant, there are other reasons for
selecting paraboloid geometry to represent the Berkovich indenter in this experiment. The
pressure distribution around the tip of the paraboloid is more realistic. Second, it is
relatively difficult to deal with a conical indenter because any plastic deformation on the
apex leads to complete loss of important characteristics, especially the elastic properties
[31].
1.5 Nanoindentation Techniques
Traditional indentations for measuring mechanical properties have additional
indirect steps; they are restricted by imaging the surface after the experiment in order to
determine the area of hardness impression (the plastically deformed area) [58]. Scanning
electron microscopy (SEM) or transmission electron microscopy (TEM) is required after
microindentation experiments. Further, imaging the shallow indenters by these techniques
is often unclear and leads to inaccurate measurements. For these reasons, traditional
indentation measurements have become inconvenient and time consuming. After
development of nanoscale science, traditional indentation measurements became more
tedious and difficult for acquiring reliable results since microindenters do not allow for the
investigation of nanomechanical properties. In the last 20 years, it was realized that there
is a deep relationship between nanoscale interactions and the macroscale behavior of
materials [9]. This implication has been used to improve and develop materials in the
16
engineering and biological fields [9]. For example, dispersing nanoparticles uniformly in
polymer matrices improves their mechanical properties such as modulus, hardness, and
even damage resistance.
Nanoindentation technique has been developed and widely used to determine
elastic modulus, hardness, and other mechanical properties at nanoscale, e. g. in studying
properties of thin-films [59]. Its applications have dramatically grown during the past 20
years in medical and microbiological fields to diagnose and monitor the mechanical
properties of cells that are linked to disease progression [60]. In the nanoindentation
approach, an indenter with a known tip geometry and nanoscale radius is loaded with
gradually increasing tiny forces to penetrate the sample by a small amount of depth, and
the mechanical properties can be estimated using the indentation force-displacement data
[60, 58]. The advantage of this technique over traditional indentations is the ability to
directly determine mechanical properties by analyses of indentation load-displacement
data alone, without the need for imaging the hardness impression [61]. It was found that
using nanoindenters with soft materials results in the “skin effect”, which is the
overestimation of modulus at small indentation depth [62]. This occurs because of the
existence of viscoelastic behavior or sample creep (especially when a sharp probe is used),
which leads to an error in the surface detection and hence to the use of an incorrect
projection contact area in the calculation of the modulus. In addition to the nonlinearity of
the stress-strain relationship, the skin effect can also originate from analyzing the data by
using a model that does not take surface adhesion into the account [62]. Indentations by
nanoindenters can only be analyzed by the Oliver and Pharr method, in which a small
17
indentation is made by an arbitrary sharp indenter (e.g. Berkovich indenter), and does not
take surface adhesion into account [63].
1.5.1 Nanoindentation by Atomic Force Microscope, AFM
A number of investigations of nanomechanical properties for soft materials and thin
layers, as in biomaterial, biological, and microbiological studies, by non-destructive and in
situ measurements have dramatically increased during the last decades [64-70]. In the
diagnostic field, measuring nanomechanical properties of cells is becoming a useful tool to
monitor disease progression [71].
At present, the direct local probing of mechanical properties of hard, soft, thin layer,
and complex materials non-destructively and in situ with nanoscale resolution has become
a reality after the development of the atomic force microscope, AFM [72]. It is considered
to be the most accurate technique for quantitative measurements of mechanical and
dynamic properties at nanoscale [17]. A small tip diameter (radius of curvature down to
<10 nm) makes the AFM instrument able to apply small loads (<1 nN) locally and for
relatively hard samples produces very small deformation (0.02 nm) [17]. Consequently, a
very high spatial image resolution can be obtained by AFM [17] and local topographic
information can be gained. Spatial resolution in topographic imaging is related to the
mechanical properties. AFM is a versatile tool that can be used for variety of materials
whether conducting or insulating, and under variety of conditions (vacuum, air or liquid
medium) [58, 60, 71, 72].
18
The general principle of AFM technique is fairly simple as illustrated in Figure 1.6.
The sample mounted on a piezo-ceramic electric scanner is contacted by a sharp tip fixed
at the end of a flexible cantilever that is held in the probe holder. During indentation
measurements, the tip first goes down to contact with the sample and then indents into the
surface until a previously specified value of compressive force is applied to the sample
[67]. This part of the motion cycle is called “approach”. Then the probe reverses its motion
and brings the probe to the initial position. This part of the motion is called “withdraw”.
Once the tip senses the surface forces, the flexible cantilever bends, and its deflection is
measured with high precision by the optical lever method. It is based on focusing a laser
beam on the top of the probe, so the cantilever deflection causes the laser reflection to a
new position on a segmented photodiode as shown in Figure 1.6.
Piezo scanner
Sample
AFM cantilever, kc
Laser
diode
Segmented
photodiode
Cantilever
deflection
measurement
Force
Deflection
DV
Figure 1.6 shows the principle scheme of the atomic force microscope.
19
Thus the deflection changes the photodiode voltage, Vt, which is monitored as a
function of the vertical displacement of the piezo actuator, ΔZp as shown as deflection vs.
displacement graph in Figure 1.7 [67]. During the probe approach the deflection voltage
remains constant (in the absence of long-range forces) until the tip jumps to contact with
the surface under influence of short-range (nanometer scale) surface attractive forces [67].
This results in a rapid decrease in deflection voltage (region B-C in Figure 1.7). A further
displacement of the piezo actuator increases the repulsive forces that deflect cantilever in
the opposite direction, causing an increase in the Vt signal (region C-D in Figure 1.7).
When the repulsion reaches the maximum, the unloading process starts, and the cantilever
deflection will gradually decrease until the tip separates from the surface. In many cases,
during the unloading process the tip adheres to the surface, leading to an increase in the
attractive forces (negative value) until the force reaches maximum tensile value (point E in
Figure 1.7). Then the tip separates from the sample (region E-F in Figure 1.7).
-1.50
-1.00
-0.50
0.00
0.50
1.00
1.50
Repulsive
region
3002001000 400 500Tip
def
lect
ion
vo
ltag
e, V
t
ΔZp, (nm)
D
C
BF
A
E
Attractive
region
Transition
region
Figure 1.7 shows deflection-displacement curve involving different types of tip-sample interactions
[67]
1.5.2 Forces during the Tip-Sample Contact
20
In AFM measurements surface forces play a major role in forming nanometer scale
contacts between the two surfaces [58]. Therefore, to achieve nondestructive
characterization with high spatial resolutions, the AFM operation should be within the
limited range of attractive and repulsive forces. Two major forces during tip-sample
interactions are the repulsive force and the attractive van der Waals (vdW) force [73, 74].
The repulsive force results from the overlap of electrons between the tip and sample atoms;
therefore, it is a short range (~ angstroms) and localized force. The long-range vdW force
(typical range 5-15 nm) originates from induced dipoles (polar molecules attract nonpolar
molecules by changing the arrangement of their electrons and thus inducing the dipole
moment) and/or dispersion force interactions and, unlike the repulsive force, it is not very
localized and can include hundreds of the nearest atoms [75]. This force is responsible for
the nanoscale jump to contact and for the same pair of materials varies depending on
surface roughness. Therefore, especially for biological and soft samples, to keep the
sample undamaged and to minimize the error in measurements of elastic parameters, the
vdW force should be controlled. This can be achieved for example by performing
experiments in a liquid medium. It was found that the AFM measurements have higher
signal to noise ratio in water than in a vacuum or air because of the significant reduction
of vdW force [76, 77]. This substantial decrease in vdW force is mainly attributed to
increasing the uniformity of the polarizable system created during the tip-sample contact
in a liquid medium [75].
There is another type of attractive force when performing AFM measurements in
ambient air at considerable humidity that is called the capillary force. It results from
21
condensation of water at the place of tip-sample contact, thus creating a liquid meniscus
bridging the tip and sample surfaces [78]. It might lead to excessive attraction and hence,
inaccurate measurements. One convenient way to eliminate capillary force is to perform
the experiment in a liquid medium [79].
The tip-sample interaction forces acting between the tip and the surface during the
approach-withdraw cycle are illustrated in Figure 1.8. The cantilever, which is considered
as a Hookean spring, has a potential energy of U1 = 1
2 kc d
2, where kc is the spring constant
and d is the deflection of the cantilever. The tip-sample interaction has a potential energy
U2 that includes the sum of energies of the attractive and repulsive forces. The second
derivative of this potential with respect to the tip-sample separation D is the spring constant
of interaction k interaction= -dU22 / d2D.This spring constant is positive for repulsive forces
and negative for attractive forces. The equilibrium position of the tip corresponds to the
minimum of total potential, Ut=U1 + U2. If kc<-kinteraction, the cantilever will jump in contact
with the sample surface, and if kc > -kinteraction, the cantilever will jump out of contact with
sample surface [39]. For elastic deformation, the total potential of the system can be
described by:
Ut = Ucs(D) + Uc (Zc) + Us (δ) (1.17)
= Ucs (D) + 1
2 kc Zc
2+ 1
2 ks δ
2 (1.18)
22
where Ucs represents the tip-sample interaction potential caused by surface forces, Uc is the
energy resulting from the bending of the cantilever, Us is the potential energy due to the
sample deformation, ks is the sample stiffness, which is a resistance to the sample
deformation, δ is the indentation depth, and D is the actual tip-sample separation D = Zp–
(Zc + δ ) where Zp is the position of the base of the cantilever (piezo position). At the
contact point, D equals 0 and Zp = Zc + δ. At equilibrium, maximum indentation, we have
ks δ = kc Zc [71].
Figure 1.8 represents the forces between tip and sample surface as a function of tip-sample distance
[https://www.researchgate.net/].
1.6 Measurements of Nanomechanical Properties by AFM
The first investigation of mechanical properties using AFM was by Burnham and
Cotton in 1989 [80]. They used the force mode of AFM for three different materials. Pure
23
elastic (elastomer), pure plastic (gold), and a material in between (graphite) were indented
by a tungsten tip with radius R = 100-200 nm and attached to cantilever with kc = 50 ± 10
N/m. If such a cantilever is bent by 20 nm, then the applied load can be as large as 1000
nN. The characteristic force curves of the three materials are shown in Figure 1.9 [80].
Sneddon's solution for a flat-ended cylindrical indenter [43] was used to calculate the
elastic modulus values, which were comparable with what had been obtained by other
methods. This work started application of AFM as a method to directly determine the
mechanical properties of a wide range of materials.
Ideally elastic deformation
Penetration depth (um)
UnloadingNo
rmal
lo
ad (
N)
Elastic-plastic deformation
Penetration depth (um)
Loading
No
rmal
lo
ad (
N)
Plastic deformation
Penetration depth (um)
Loading
No
rmal
lo
ad (
N)
UnloadingLoading
Unloading
Figure 1.9 displays the force curves of three different materials: a) elastomer, b) purely plastic
(gold) and c) the most common elastic half-space material (graphite) [44].
Later, different AFM methods were developed to measure mechanical properties
of materials. These methods are bending/tensile test [81-85], dynamic AFM [86-90], as
well as nanoindentation methods [91-96].
24
1.6.1 Bending / Tensile Testing
In a bending / tensile test, the tensile load is applied on the fibrous specimen and the
resulting elongation is measured. The sample is mounted on a microscope stage equipped
with a camera to determine the deformation and geometrical dimensions of the sample
[85]. The load acting on the sample is determined by a piezo resistive AFM cantilever
whose tip is placed at the free end of the specimen. Then, the elastic modulus of the sample
can be determined using standard beam theory [84] assuming a uniform rectangular cross
section cantilever:
E = 4 𝐿3
𝐹
𝛿
𝑤 𝜃3 (1.19)
where L is the length of the cantilever, w is the cantilever width, 𝜃 is the cantilever
thickness, and F/δ is the slope of load-displacement curve. This method is suitable for in
situ applications; however, it has significant drawbacks [84]. First, while the force
measurements are carried out by the piezo resistive AFM cantilever, the displacement
cannot be measured by AFM because of the hysteresis in the piezoelectric AFM cantilever;
hence, an additional device such as SEM is needed to measure the displacement accurately
[84]. Even with additional accessories, accurate and reproducible results are limited to
measurements only in a vacuum state [84] and for only stretched materials such as
polyethyleneoxide (PEO) nanofiber as illustrated in Figure 1.10 [85].
25
Piezo-resistive
AFM cantilever
tip
Glass
fiber
Super
glue
UV glueMasking
tape
Microscope
stage
Coverslip
Stretched PEO
nanofiber
Microscope lens
Figure 1.10 shows the scheme of bending / tensile AFM testing [85].
1.6.2 Applications of Dynamic AFM to Characterize Mechanical Properties
Dynamic atomic force microscope techniques have been developed to obtain high
resolution topographic images and compositional information about sample surfaces in
gentle and non-destructive ways [86]. This method is especially suitable for soft samples
that can be easily damaged by high adhesion, friction or high electrostatic forces, or for
materials that cannot be tightly fixed on their substrates. Dynamic AFM methods include
the amplitude modulation AFM (tapping mode) and the non-contact mode.
1.6.2.1 Force Modulation-AFM, FM-AFM
The dynamic mode technique called force-modulation, AFM, allows the imaging
of areas on hard surfaces. In this technique the sample height is modulated when measuring
a topographic image and a map of surface elasticities is created (force modulation image)
by observing the modulation of cantilever amplitude at the frequency of sample modulation
26
[97]. After keeping the tip-sample force constant by the feedback loop, a small motion
(~1-10 nm amplitude) ΔZm, is introduced to the sample surface in Z direction. This leads
to an additional cantilever deflection, d, and hence a variation in the tip-sample force. The
upper and lower positions of the cantilever are measured to calculate the cantilever
deflection d. Then, the quantity d / ΔZm is used to build the force modulation image. The
amount of variation of the average force depends on the dynamic response represented in
the viscoelastic properties of the sample. This response is characterized by the sample
spring constant, ks, as follows:
ks = 𝛥𝐹
𝛥𝑍m–𝑑 (1.20)
where (ΔZm - d ) is the sample deformation by the acting force ΔF. In the absence of
adhesion forces (no energy dissipation) we have:
kc d = ks (ΔZm–d) (1.21)
ΔZm / d = 𝑘𝑐
𝑘𝑠 +1 (1.22)
If kc /ks << 1, the sample is much stiffer to be deformed and the contrast based on
surface elasticities cannot be obtained. Therefore, stiff cantilevers (depending on the
elastic modulus of the sample, kc > 1 N /m for polymeric samples and up to ~1000 N/m for
hard samples) are necessary for acquiring the force modulation images. During force
modulation imaging the soft areas deform more than hard areas on the sample surface (the
27
cantilever deflection d is less over the soft area). According to the Hertz model (with no
adhesion force), the force ΔF acting on a surface of modulus Es, and required to deform it
by pressing a spherical indenter of radius R to the depth δ is expressed as:
ΔF = ( Es2R δ3)1/2 (1.23)
The first application of force modulation model was by P. Maivald et al. in 1991
[97], in which a contrast resulting from variations in local surface elasticity and
topographic images were obtained for carbon fibers and epoxy composite, E ranges from
3.5 GPa to 2.3 × 102 GPa. Figure 1.11 [97] displays these images, for which a stiff
cantilever (kc = 3000 N /m) was fabricated with a diamond tip of radius (R = 10 μm) to
apply a repulsive force of 10–3 N in air.
Figure 1.11 shows force-modulation imaging of carbon fiber and epoxy composite. Panels in the
figure show a) topographic image, b) stiffness in the force modulation image. Here bright spots
correspond to stiff areas. Image width is 32 µm. The figure reproduced with permission from IOP
science [97].
28
As illustrated in the topographic image in Figure 1.11, carbon fibers show up higher
than the surrounding epoxy. The force modulation image exhibits the variation in local
surface elasticity and also reveals this structure across the cross section of fibers. Features
in the force modulation image generally do not match perfectly with features in the
topographic image confirming that each image has its own information source [97]. This
technique can be used to characterize nanoscale heterogeneity of mechanical properties
and surface defects. However, application of this technique is limited to relatively hard
samples [98]. On soft samples and samples with adhesion, force-modulation imaging
might produce excessive lateral forces that will damage samples. Moreover, on sticky
samples, the adhesive force adds to the average compressive force. Consequently this
might produce false contrast in the image. Force modulation measurements were
previously carried out on tissue embedded in London Resin White, LRW (a hydrophilic
acrylic resin used for embedding biological samples) [99], which have modulus values
almost equal 0.5 GPa. The image of elasticity was damaged during measurements as
shown in Figure 1.12 [99].
29
Figure 1.12 illustrates topography (A) and elasticity (B) of etched LRW embedded sample. Panal
B shows the difficulties of measuring elastic differences in the components [98].
1.6.2.2 AFM Phase Imaging
Tapping mode imaging is performed by exciting cantilever oscillation at or near its
resonance frequency. Then the tip at the end of the oscillating cantilever interacts with the
sample surface, and the local force gradient shifts the resonance frequency of the cantilever.
This in turn changes the amplitude of the cantilever oscillation. Imaging is performed by
scanning across the sample surface and using a feedback loop to maintain the amplitude of
the cantilever oscillation. It is noticed that the phase of the oscillating cantilever depends
on the mechanical and adhesive properties of the sample. More specifically, the phase shift
for the cantilever excited at its resonance frequency is proportional to the energy dissipated
during each contact of the tip with the sample. Phase shift can be measured along with the
topographic image. Contrast in the phase image indicates the variation in elastic and/or
30
adhesive properties of the sample surface [100]. Therefore, phase imaging cannot be used
to obtain quantitative information about the elastic modulus.
1.7 Current Applications of AFM Nanoindentation on Soft Materials and Associated
Challenges
The ability to continuously and accurately measure stiffness at various indentation
depths [101] nominates Nanoindentation methods to be used in determination of
nanomechanical properties of a wide variety of materials [102]. However, the relatively
large indentation forces (several μN) applied in this technique restricts it from being used
with soft materials, on which the nonlinear stress-strain relationship likely occurs. For
example, the strong indentation of biological cells can cause a disruption of cellular
components [103]. Furthermore, the skin effect, which is the overestimation of modulus
at small indentation depth and hence exceeding the elastic limit, occurs with soft samples
measured by nanoindenters, in which the data are analyzed by the Oliver-Pharr model
[104]. Another downside in nanoindenters is the need of additional optical device to
complete the measurements. This external procedure causes time consuming and decreases
the accuracy of measurements, which became depend on the resolution of this device [101].
The recent applications on soft materials, such as tissue engineering and biological
fields, concern with investigation of nanomechanical properties and submicro features on
the outermost surface. Further, imaging and concurrently determining nanomechanical
characterization under physiological conditions became very important for either
monitoring cell progression or improving biomaterials [101]. These multiple purposes, in
31
addition to the versatility of using different models to analyze data, demand the using of
AFM nanoindentation for applying forces of a few piconewton and achieve very small
deformation (~ 0.1 nm) [101], and to simultaneously image the indentation site under
desirable conditions [ 051 ]. Recently, AFM nanoindentation has been used to study the
skin effect of different polymers using probes with different radii (22, 810 and 1030 nm)
and analyzing the data by different models (Oliver-Pharr, JKR, Hertz and DMT) [ 041 ].
This report stated that the skin effect disappeared by using the large radius probes and
analyzing the data by either JKR or DMT, which take the adhesion into account. When
using the dull probes, the results started to be close to the bulk moduli at indentations 2-3
nm, which considered as the smallest depth for soft samples to reach the bulk modulus.
However, in the case of sharp tip the bulk moduli could not be reached until the indentation
was 90 nm. Consequently, the skin effect originates from either using a sharp tip that can
break the stress-strain linearity, or using improper model such as Oliver-Pharr to extract
modulus values [104]. In biological field, AFM nanoindentation was applied to distinguish
between healthy and disease cells and also monitor the disease progression [106]. Intensive
studies have been made by AFM nanoindentation on normal, benign and tumor cells [107.
108]. It was found that the flexibility of tumor cell diffusion can be attributed to its lower
stiffness with respect to normal cell [108, 106]. Because of the AFM ability to characterize
micro and nanoscale feature sizes, AFM nanoindentation has widely applied to study
microorganisms and viruses. Mechanical heterogeneity of bacterial surface and its biofilm
have been examined [109]. Determination of local stiffness on viral surfaces demonstrated
that their elasticity related to the virus inactivation reaction [110]. Additionally, this
approach has been used to determine elastic modulus of hydrogels and thin layers of soft
32
materials [68, 111, 104,112, 113], and some of the results were agree with macroscopically
measured values [68, 121]. More developed AFM nanoindentation method has been
performed using force-volume mode [114], which also called force mapping, in which a
matrix of force curves are collected across the sample surface. It has been widely used
with soft samples as a solution to avoid the lateral forces that result from contact-imaging
[115]. During force mapping, the tip is completely separated from the surface before
indenting the other point. Some of its recent applications are to produce maps of elasticity
and topographic images for human cells at different stages of disease progression.
Nonetheless, challenges for AFM nanoindentation to achieve accurate
measurements on soft samples still exist. These difficulties are related to either sample
nature such as viscoelasticity and adhesion phenomena, or experimental setup represented
in probe geometry and size, the cantilever stiffness, identification of probe-sample contact
point [104], and using improper model to extract the Young's modulus of materials [101].
1.7.1 Effects of Viscoelasticity in Force-Indentation Measurements
Viscoelastic materials have important applications as adhesives, coatings,
biosensors and lubricants [111]. Therefore, determining the viscoelastic behavior
accurately for these materials has recently a considerable concern. Attempts have been
made to access the viscoelastic property of soft materials; one of them is the study the
dependence of tip-sample adhesion on loading speed and also on temperature, but these
procedures have lack of quantitative information regards the viscoelastic characterization
[116]. To investigate the interfacial viscoelasticity of materials, microparticles were
attached to a tipless AFM to study the adhesion with mica surface [117, 118] and then
33
analyzing the data by different theories to explore the adhesion contributions [117,
120,121]. Recently, Cappella et al proposed an approach to determine the viscoelasticity
by studying the variation of elastic modulus with changing both time and temperature of
the experiment [122]. Yet, these attempts used indirect ways to investigate the viscoelastic
behavior. It used the frequency of voltage instead of piezo displacement rate, which directly
related to the complete travel of the piezo that can reflect the irreversible sample
deformation [122, 123]. Another suggestion to provide information about viscoelasticity
was the strain rate. Using a conical probe in AFM nanoindentation to explore the
viscoelasticity in term of the strain rate was arguable procedure. Although the self-
similarity of the conical probe as an axisymmetric shape, the stress field from this tip is
highly complicated because the surface points located in the tip domain are not
experiencing the same strain during the indentation process [111]. For this reason, indenter
rate was introduced instead of strain rate to characterize viscoelastic behavior and its
repeatability. In this work [111], the variations of Young's modulus with both temperature
and the indenter rate changes were identified, which reflected the viscoelastic behavior of
poly propylene glycol material. Figure 1.13 shows the dependence of evaluated Young's
modulus on both temperature and indenter rates.
34
Temperature, K
Ela
stic
Modulu
s, M
Pa
0
10
20
30
40
50
60
290 295 300 305 310 315
8400 nm/s
2790 nm/s
897 nm/s
Figure 1.13 shows the dependence of elastic modulus of polyethylene glycol on indenter rate and
temperature. The elastic modulus changes significantly at high indenter rate and low temperature
[111].
The plots illustrate the high dependence of modulus on the indenter rate at low
temperature. It was found that the viscoelastic property is reproducible as long as the basic
AFM steps have achieved [111].
1.7.2 Cantilever and Tip Characteristics
Even though the sharp tip with ROC~ 10 nm provides high resolution and can sense
the supported cellular membranes [124], the tip of large ROC is recommended for soft and/
or thin [74] samples to minimize plastic deformation and reduce damaging [125, 126]. The
large tip radius exerts a slight pressure and contact gently with soft samples. However, in
many cases the dull probes likely provided incorrect values of elastic modulus if the sample
roughness would not be taken into account [101]. Spherical tips have ROC similar or even
larger than the roughness of biological surfaces. This leads to the incorrect estimation of
35
the tip-sample contact point. Therefore, a relatively sharp probe with ROC smaller than
sample roughness should be considered [127]. The Geometrical changes in the tip shape
have significant effects on elastic modulus values. Using spherical probe with (ROC ~ 2)
μm causes a dramatic decrease in modulus compared to that using pyramidal probe (ROC
~ 50 nm) [128]. On the other hand, using probes with the same ROC but different spring
constants can affect Young's moduli. The spring constant is a marker to what extent the
cantilever can deflect, and by this deflection the mechanical properties of samples will be
detected [101]. To achieve an accurate detection for surface elasticity, the cantilever
stiffness should be close to that of sample. Indentation of soft materials by stiff cantilever
can exceed the elastic limit and damage the sample. Soft cantilever will be damaged if it
scans a much stiffer surface.
1.7.3 Effects of Indentation depth
There are two main effects related to the penetration depth and lead to obvious errors in
elastic modulus values. First, the indentation that similar or less than sample roughness
affects the true tip-sample contact area. In other words, very small indentation in biological
surfaces leads to miss contact with many points beneath the tip curvature and hence large
error in elastic modulus calculations [101]. To solve this problem, the penetration depth
must be much larger than sample roughness [129]. The rule of thumb can be applied in
this case, which stated that the roughness / indentation depth ratio should be ≤ 1
10 [101].
The second reason is represented in the substrate effect, which is the overestimation of
Young's modulus [101], and can be seen in the case of compliant material with small
36
thickness [130, 131]. This effect appears at the indentation site that either pile-up when
the tip indents the soft thin film deposited on a hard surface, or sink-in when the indentation
occurs at a hard material fixed on soft substrate [132]. To eliminate the substrate effect,
the indentation depth must be less than 10% of the entire sample thickness [31, 133]. Based
on Oliver and Pharr's, models have been developed to improve the accuracy of modulus.
Hay and Crawford [134] were enabled to eliminate the substrate effect and accurately
determine elastic modulus through an indentation depth up to 25% from the sample
thickness. However, it was noted that in AFM indentation measurements substrate effects
might be expected when the radius of contact area is similar to the sample thickness, even
though sample indentation might still be small in comparison to the sample thickness [135].
Therefore estimation of geometry of tip-sample contact is necessary in order to understand
whether the substrate effect can be avoided in particular experiment. Overall, the main
steps to deal with soft thin materials in AFM nanoindentation technique are to examine the
sample roughness and its total thickness, then perform an indentation ten times more than
roughness scale. For thin samples using sharp probe and low indentation force might be
necessary to avoid systematic error associated with hard substrate. For analyzing data of
thin films, using infinite thickness models may cause large errors in elastic modulus values
[135, 136, 137]. In the latter reference, Akhremitchev and Walker used Dhaliwal and Rau
model [138] to calculate elastic modulus for finite sample thickness and indicate that the
errors result from models of linear elasticity can be an order of magnitude.
1.7.4 Effects of Model Selection
37
Models that do not take into account the obvious phenomena of materials cannot
provide accurate values of modulus. Even though Hertzian and Oliver-Pharr models are
considered as conventional methods for many materials, challenges occur when applying
these methods on compliant and hydrated biomaterials [113]. For example, the Oliver
Pharr model can only deal with the elastic-plastic response of the sample without taking
the adhesion into account [139], resulting in large errors in modulus values of adhesive
biological samples. Other viscoelastic models have been proposed to calculate the storage
modulus of viscoelastic materials under a very low load to achieve only Hertzian
interactions [140], but the very small indentation can increase the error in modulus values.
Applying load much higher than adhesion forces or performing the experiments under
water can also minimize the adhesion and facilitate using the viscoelastic models.
However, using high load or water might damage the sample or change its physical
properties, respectively. Therefore, a comprehensive model that compromise between all
significant phenomena is required to diminish the error ratio. In fact, it is still complicated
to develop a model that can deal with both viscoelastic and adhesion properties of material
[141]. This can be attributed to the dependence of both adhesion and viscoelastic properties
on frequency.
The obstacle to use Hertz model with adhesive samples is that the adhesion property
of biological samples enlarges the contact area between tip and sample surface, which
conflicts with Hertzian concept that requires a very small contact area to ignore the
adhesion. The JKR model can be applied to model the indentation by a large spherical
probe that creates a large contact area, and then the adhesion can be taken into account.
38
However, the JKR model has some limitations since it cannot be valid for a wide range of
materials such as visoelastic samples, and works only with a perfectly spherical tip
[113].To develop this model and make it more versatile, a model which based on the
adhesive interaction between AFM tips and samples, called two-point method, has been
proposed in previous work [142]. Many problems related to soft thin films and large
adhesive materials have been overcome by modeling the induced adhesive tip-sample
interactions. In this method, only a passive indentation is induced by tip-sample adhesion,
minimizing the interference from the substrate compared to what is found using the
indentation method [142]. Furthermore, it does not require locating the tip-sample contact
point, which is a challenge in adhesive soft samples, since the adhesive interactions are
represented by fitting the whole retraction force curves. In this report, the two point method
was tested on a series of poly (dimethyl siloxane) polymer samples with different degrees
of crosslinking, and the results were consistent with macroscopic data [143]. Because of
the high accuracy of this model [144, 140], it was later applied in several researches. One
of them was for calculating the elastic modulus of poly (vinyl acetate), PVAc, films in both
glassy and rubbery states. First, elastic modulus values were calculated from measured
force curves by using Hertz equation. Then, the modulus values were used to get the JKR
theoretical curves, which exhibited a good agreement with the experimental data in the
case of glassy samples, indicating the validation of two-point method for calculating the
elastic modulus of glassy polymers. On the other hand, a little deviation between the
experimental data of rubbery samples and the theoretical curves can be observed. This
inconsistency might be attributed to the viscoelastic interaction between the probe and
sample, indicating that the two-point method works only with elastic manner [140, 142]
39
1.7.5 Problems with Contact Point Identification
One of the most common problems during nanoindentation of soft materials is
misidentification of contact point, Guo and Akhremitchev [145] stated that almost 10% of
systemic errors are related to incorrect selection of contact point. Missing the contact point
can result from either the high level of noise in the instrument or the high adhesion on the
sample surface, which considerably exists in biological materials [144]. Attempts have
been made to correctly identify the tip-sample contact point on soft and adhesive samples.
Jaasma et al [146] extracted the contact point from the linear relationship of derivative of
cantilever deflection to its base position. Extrapolation data to zero value yields the point
at contact or separation position. However, this method requires data to be in the elastic
limit (very small indentation). On the other hand, at this position the interactive forces are
significant and cause high level of noise that can affect the accuracy of derivation. To
manipulate this situation, measurements might be performed at sufficiently large
indentation to minimize the level of noise [146]. Yet, this leads definitely to damage the
soft biological surfaces. Another suggestion has been introduced by Guo and
Akhremitchev [145], their idea start from the fact that accurately determination of Young’s
modulus from Hertizian expression (equation 1.24) is limited with a correct value of the
indentation depth, δ.
F = 4 𝐸 𝑅
12
3(1−𝜈2) 𝛿
3
2 (1.24)
Where F is the load, R is the probe’s radius of curvature, δ is the indentation, ν is the
Poisson ratio of the elastic solid, E is the Young’s modulus. Yet, it is difficult to obtain
40
the accurate value of δ because of the hardly determine the tip-sample contact point.
Therefore, Guo and Akhremitchev wrote the Hertzian equation in a way by which the
Young’s modulus can be calculated from the dependence of applied force on the tip-sample
distance, D, regardless their contact point. They states that the relation between D and δ
can be written as:
δ = C − D (1.25)
Therefore, the Hertzian equation can be expressed in terms of equation (1.25) as follows:
𝐹2
3 = C* − [4𝐸𝑅
12
3(1−𝜎2)]
2
3
(1.26)
C* is another constant that reflects the selection of tip-sample contact point.
From this linear equation, the post contact region points can be plotted and then the
Young’s modulus can be extracted from the slope of the line. Noticeable, a systematic error
of only ~ 10% might occur, which results from either the probe radius of curvature or the
estimated Poisson’s ratio. This uncertainty is found to be less than that of cantilever spring
constant [145]. Yet, this approach can be used only in elastic limit where the indentation
is non-interactive [144].
1.8 The Aim of Research
Methods of measuring elastic properties of materials described above suffer from
different limitations that prevent comprehensive characterization of mechanical properties
of soft samples. Specifically, techniques capable of mapping mechanical and adhesive
properties often provide only a relative map of elastic properties (e.g. force modulation and
41
phase imaging described above). Techniques capable of extracting the absolute value of
the local modulus (force-indentation measurements) often report only statistical values of
the elastic modulus without providing information regarding the spatial distribution of
moduli values [147-150]. More advanced studies that generate maps of elastic moduli of
soft samples do not consider the effects of adhesion and viscoelasticity [151,152]. It is the
goal of the current study to develop and test methodology that is capable of measuring
mechanical and adhesive properties of soft materials and to simultaneously report
heterogeneity and spatial distribution (mapping) of the measured parameters, including the
effects of depth sensing, adhesion, and viscoelasticity.
The developed approach is used to characterize the elastic properties of collagen-
Bioglass samples that are important substrates in tissue engineering efforts [153]. Selection
of these samples for testing measurement methodology is rationalized below.
The rational design of artificial scaffolds for tissue engineering requires inclusive
studies of the extracellular matrixes (ECMs) to successfully mimic their characteristics.
ECMs are natural scaffolds that provide structural support to cells and surrounding tissues
[152]. Figure 1.14 shows an image of the natural ECM of bone tissue.
42
Figure 1.14 represents a cross section showing osteons. A unit of cells, an osteocyte, has a 5-20
µm diameter, is 7 µm deep, and contains 40-60 cells [https://www.78stepshealth.us/].
ECMs mainly contain bioactive molecules such as proteins, glycosaminoglycans
and glycoproteins. Quantities and local distribution of the ECM components determine
the mechanical properties, shape, and function of the whole organ. For example, collagen
is the most common component in bone and skin tissues [154]. Therefore, collagen can be
the main component for designing an artificial scaffold in bone engineering. However,
collagen concentration must be controlled since it can affect both cell-cell adhesion and
scaffold porosity, which has basic roles in cell-scaffold adhesion. Moreover, an increase
in collagen concentration results in high crosslinking of ECM, which leads to a limitation
in the supply of gases and nutrients to the cells and release of a hypoxia-related signaling
process [155]. Scaffold stiffness should also be controlled, since it is responsible for
mechanical strength during and after implantation [156]. Meanwhile, ECM stiffness values
are different from one tissue to another. The stiffness values of osteogenic, myogenic, and
neuronal tissues are 34 kPa, 11 kPa, and 0.1 kPa, respectively [156]. Therefore, the
artificial scaffold should have a stiffness value compatible with that kind of tissue.
43
Moreover, cells interacting with scaffolds interact with the superficial layer of the
biomaterial, while the mechanical testing of scaffold materials typically measures the total
modulus. Thus, it is important to compare the elastic properties at the surface with the bulk
values. In addition, the spatial heterogeneity of mechanical properties on a biologically-
relevant length scale should be characterized to provide additional information for scaffold
design.
Experimental testing will be performed on collagen samples loaded with different
amounts of Bioglass, to mimic the ECM of bone tissue. Figure 1.14 shows a cartoon of an
artificial scaffold as expected at microscale.
Figure1.15. displays a cartoon of an artificial scaffold with cells attaching between collagen fibrils
in microscale
Elastic properties of collagen samples will be mapped using the AFM force
mapping technique. The variation of the elastic modulus with depth will be sensed by
analyzing portions of the force curves corresponding to different penetration depths
44
(different levels of maximum applied force). The spatial heterogeneity of the modulus will
be characterized by using different scan sizes, corresponding to the pixel-to-pixel distances
of 20, 95, and 470 nm. This will allow the separation of systematic variations in the elastic
modulus with characteristic dimensions from random point-to-point variations.
In order to perform depth sensing, relatively sharp AFM probes will be used.
Therefore the tip of the probe might indent the sample to a depth noticeably exceeding the
radius of curvature of the probe. However, standard methods of extracting elastic
properties in indentation measurements utilize very simple probe shape models (e.g.
conical or paraboloidal tip models). The tip of the AFM probe is usually somewhat
rounded at the apex and then transitions to a conical shape. Consequently, if a simple probe
shape model is used in data analysis, then the effects of the probe shape might be perceived
as the depth dependence of the elastic modulus. To test and correct for this possible artifact,
the hyperboloidal tip shape model will be used because it has both of these features:
rounded tip shape that transitions into cone [137]. Data analysis will explicitly take into
account sample’s adhesion to avoid artifacts related to testing mechanical properties for
sticky samples [142].
45
1.9 References
[1] Su, C.; Chen, Y.; Yu, P.; Song, M.; Chen, W.; Guo, S.F., Linking the thermal
characteristics and mechanical properties of Fe-based bulk metallic glasses, Journal of
Alloys and Compounds 2016, 663, 867-871.
[2] Greskovič, F; Varga, J.; Dulebová, L., The Utilize of gamma radiation on the
examination of mechanical properties of polymeric materials, Metabolism 2012, 51, 245-
248.
[3] Dulebová, L; Badida, M.; Duleba, B., Influence of environment on bending stress of
composite with regranulate, Fascicule 2012, 2, 1584-2665.
[4] Johnsona ,W.L.; Youngquista , R. C.;Gibsonb , T. L.; Jolleyb, S. T.; Williamsa, M. K.,
Measurement of the mechanical properties of thin film polymers at cryogenic temperatures,
Advances in Cryogenic Engineering 2014, 1574, 162-169.
[5] Esteban, A.; Franceschini; Horacio R. Corti, Elastic properties of nafion,
polybenzimidazole and poly [2,5-benzimidazole] membranes determined by AFM tip
nano-indentation, Journal of Power Sources 2009, 188, 379-386.
[6] Pelham Jr., R. J.; Wang, Y., Cell locomotion and focal adhesions are regulated by
substrate flexibility, Proceedings of the National Academy of Sciences of. U.S.A 1997, 94,
13661-13665.
46
[7] Discher, D.E.; Janmey, P.; Wang, Y.L.; Tissue cells feel and respond to the stiffness of
their substrate. Science 2005, 310, 1139-1143.
[8] Vanlandingham, M. R.; McKnight, S. H.; Palmese, G. R.; Elings, J. R.; Huang,
X.;Bogettl, T. A.;Eduljee, R. F.; Gillespie, J. W. Jr., Nanoscale indentation of polymer
systems using the Atomic Force Microscope, The Journal of Adhesion 1997, 64, 31-59.
[9] Devaprakasam, D.; Hatton, P.V.; Möbus, G; Inkson, B.J., Nanoscale tribology, energy
dissipation and failure mechanisms of nano- and micro-silica particle-filled polymer
composites, Tribology Letters 2009, 34, 1-11.
[10] Ali, S. H. R.; Bedewy, M. K.; Etman, M. A.; Khalil, H. A. and et al., Morphology and
properties of polymer matrix nanocomposites, International Journal of Metrology and
Quality Systems 2010, 1, 33-39.
[11] Ali, S. H. R., Mechanical, tribological properties and surface characteristics of
developed polymeric materials reinforced by CNTs, SAE International Journal of Fuels
and Lubricants 2015,8, 35-40.
[12] Boskovic; Bojan, Carbon nanomaterials for transport, Cambridge nanomaterials
technology Ltd, Nanomagazine, UK. 2014.
[13] Shojaeefard, M. H.; Khalkhali, A.; Khakshournia, Sh., Multi objective optimization
of a CNT/polymer nanocomposite automotive drive shaft, The 3rd Int. Conference on
Design Engineering and Science, Pilsen, Czech Republic, 2014.
47
[14] Chen; Wei; Yao; Yuan; Kim, Y.; Young, A Review on carbon nanotube probes for
microscopy applications, experiments in micro/nanoscience and engineering term,
Department of Mechanical Engineering, 2014.
[15] John R.; Withers, D.; Aston, E., Nanomechanical measurements with AFM in the
elastic limit, Advances in Colloid and Interface Science 2006, 120, 57-67.
[16] Jason I.; Kilpatrick ; Revenko, I.; Brian J.; Rodriguez, Nanomechanics of Cells and
biomaterials studied by atomic force microscopy, Advanced. Healthcare Materials 2015,
4, 2456-2474.
[17] Tang, B.; Ngan, A. H. W.; Pethica, J. B., A method to quantitatively measure the
elastic modulus of materials in nanometerscale using atomic force microscopy,
Nanotechnology 2008, 19, 1-7.
[18] Hosseini, S. M.; Feng, J. J., How malaria parasites reduce the deformability of infected
red blood cells, Biophysical Journal 2012, 103, 1-10
[19] Rehfeldt, F.; Engler, A. J.; Eckhardt, A.; Ahmed, F.; Discher, D.E., Cell responses to
the mechanochemical microenvironment-implications for regenerative medicine and drug
delivery, Advanced Drug Delivery Reviews. 2007, 59, 1329-39.
[20] Janson, I. A.; Putnam, A. J., Extracellular matrix elasticity and topography: material-
based cues that affect cell function via conserved mechanisms, Journal of Biomedical
Materials Research Part A 2014, 103, 1246-1258.
48
[21] Engler, A.; Bacakova, L.; Newman, C.; Hategan, A.; Griffin, M.; Discher, D.,
Substrate compliance versus ligand density in cell on gel responses, Biophysical Journal
2004, 86, 617-628 .
[22] Engler, A. J.; Sen, S.; Sweeney, H. L.; Discher, D. E., Matrix elasticity directs stem
cell lineage specification, Cell 2006, 126, 677-689 .
[23] Discher, D. E.; Mooney, D. J.; Zandstra, P. W., Growth factors, matrices, and forces
combine and control stem cells, Science 2009, 324, 1673-1677.
[24] Discher, D. E.; Janmey, P.; Wang, Y. L., Tissue cells feel and respond to the stiffness
of their Substrate, Science 2005, 310, 1139-1143 .
[25] Buxboim, A.; Ivanovska, I. L.; Discher, D. E., Matrix elasticity, cytoskeletal forces
and physics of the nucleus: How deeply do cells 'feel' outside and in? Journal of Cell
Science 2010, 123, 297-308 .
[26] Kim, S. N.; Jeibmann, A.; Halama, K.; Witte, H. T.; Wälte, M.; Matzat, T.; Schillers,
H.; Faber, C.; Senner, V.; Paulus, W.; Klämbt, C., ECM stiffness regulates glial migration
in drosophila and mammalian glioma models, Development 2014, 141, 3233-3242 .
[27] Dingal, P. D. P.; Discher, D. E., Material control of stem cell differentiation challenges
in nano-characterization, Current Opinion in Biotechnology 2014, 28, 46-50 .
49
[28] Halliday, N.; Tomasek, J., Mechanical properties of the extracellular matrix influence
fibronectin fibrel assembly in vitro, Experimental Cell Research 1995, 217, 109-117 .
[29] Zajac, A. L.; Discher, D. E., Cell differentiation through tissue elasticity-coupled,
myosin-driven remodeling, Current Opinion in Cell Biology 2008, 20, 609-615.
[30] Engler, A. J.; Griffin, M. A.; Sen, S.; Bönnemann, C. G.; Sweeney, H. L; Discher, D.
E., Myotubes differentiate optimally on substrates with tissue-like stiffness: pathological
implications for soft or stiff microenvironments, Journal of Cell Biology 2004, 166, 877-
888.
[31] Oliver, W. C; Parr, G. M., An improved technique for determining hardness and elastic
modulus using load and displacement sensing indentation experiments, Journal of
Materials Research 1992, 7, 1564-1583.
[32] Tabor, D., A simple theory of static and dynamic hardness, Proceedings of the Royal
Society 1947, A 192, 247-274.
[33] Weast, R.C., Handbook of chemistry and physics, Chemical Rubber Publishing
Company 1976.
[34] Vinckiera, A.; Semenza, G., Measuring elasticity of biological materials by atomic
force microscopy, FEBS Letters 1998, 430, 12-16.
50
[35] Baumgart, F., Stiffness - an unknown world of mechanical science? Injury 2000, 31,
14-23.
[36] Boussinesq, J., Application des potentiels à l'étude de l'équilibreet du mouvement des
solid es élastiques, Paris 1885.
[37] Love, A. E. H., The stress produced in a semi-infinite solid by pressure on part of the
boundary, Philosophical Transaction of the Royal Society. 1929, A 228, 377-420.
[38] Love, A. E. H., Boussinesq's problem for a rigid cone, The Quarterly Journal of
Mathematics 1939, 10, 161-175.
[39] Pethica, J. B., Tip surface interactions in STM and AFM, Physica Scripta 1987, T19,
61-66.
[40] Johnson, K. L., Contact mechanics, Cambridge University Press, Cambridge, UK
1985.
[41] Schwarz, U. D.; Zwőrner, O.; Kőster, P.; Wiesendanger, R., Quantitative analysis of
the frictional properties of solid materials at low loads. I. Carbon compounds, Physical
Review 1997, B 56, 6987-6996.
[42] Müller, M. K.; Geiss, R. H.; Hurley,D. C., Contact mechanics and tip shape in AFM-
based nanomechanical measurements, Ultramicroscopy 2006, 106, 466-474.
51
[43] Sneddon, I. J., The Relation Between Load and Penetrationin The Axisymmetric
Boussinesq Problem for a Punch of Arbitrary Profile, International Journal of Engineering
Science 1965, 3 , 47-57.
[44] Ternovskii, A. P.;Alikhin, V. P.; Shorshorv, M. Kh.; Khrushchev, M. M.; Skvortsov,
N. V., Mechanical testing of materials by micro pressing technique, Zavodskaya
Laboratoriya 1973, 39, 1242.
[45] Bulychev, S. I.; Alekhin, V. P.; Shorshorv, M. Kh.; Ternovskii, A. P.; Shnyrev, G.
D., Determining Young’s modulus from the indentor penetration diagram; Zavodskaya
Laboratoriya 1975, 41, 1137-1141.
[46] Bulychev, S. I; Alekhin, V. P.; Shorshorv, M. Kh.; Ternovskii, A. P., mechanical
properties of materials studied from kinetic diagram of load versus depth of impression
during microimpression, Strength Mater 1976, 8, 1084-1089.
[47] Shorshorv, M. Kh.; Bulychev, S. I.; Alekhin, V. P., Work of plastic and elastic
deformation during indentation, Soviet Physics Doklady 1982, 26, 769-771.
[48] Bulychev, S. I; Alekhin, V. P, Method of kinetic hardness and microhardness in testing
impression by an indentor, Zavodskaya Laboratoriya 1987, 53, 76.
[49] Pharr. G. M.; Oliver, W. C.; Brotzed, F. R., on the generality of relationship among
stiffness, contact area, and elastic modulus during indentation, Journal of Materials
Research 1992, 7, 613-617.
52
[50] King. R. B., Elastic analysis of some punch problems for a layered medium,
International Journal of Solids Structures 1987, 3, 1657-1664.
[51] Pethica, J. B., Ion implantation into metals, Pergamon Press. 1982, 147-156.
[52] Loubet, J. L.; Georges, J. M.; Marchesini, O.; Meille, G., Vickers indentation curves
of magnesium-oxide (MgO), Journal of Tribology 1984, 106, 43-48.
[53] Newey, D.; Wilkens, M. A.; Pollock, H. M., An ultra low load penetration hardness
tester, Journal of Physics E: Scientific Instruments 1982, 15, 119-122.
[54] Stone, D.; Lafontaine, W. R.; Alexopoulos, P.; Wu, T. W.; Yu Li Che, An
investigation of hardness and adhesion of sputter deposited AI on Si by a continuous
indentation test, Journal of Materials Research 1988, 3, 141-147.
[55] Pethica, J. B.; Hutchings, R.; Oliver, W. C., Hardness measurements at penetration
depth as small as 20 nm, Philosophical. Magazine 1983, 48, 593-606.
[56] Oliver, W. C.; Hutchings, R.; Pethica, J. B., Microindentation techniques in materials
science and engineering, American Society for Testing and Materials 1986, 90-108.
[57] Doerner, M. F.; Nix, W. D., A method for interpreting the data from depth-sensing
indentation instruments, Journal of Materials Research 1986, 1, 601-609.
[58] Burnham, N. A.; Colton, R. J.; Pollock, H. M., Interpretation of force curves in force
microscopy, Nanotechnology1993, 4, 64-80.
53
[59] Yang, S.; Zang, Y-W, Analysis of nanoindentation creep for polymeric materials,
Journal of Applied Physics 2004, 95, 3655-3666.
[60] Cinar, E.; Sahin, F.; Yablon, D., Development of a novel nanoindentation technique
by utilizinga dual-probe AFM system, Beilstein Journal of Nanotechnology 2015, 6, 2015-
2027.
[61] Oliver, W. C.; Pharr, G. M., Measurement of hardness and elastic modulus by
instrumented indentation: Advances in understanding and refinements to methodology,
Journal of Materials Research 2015, 6, 2015-2027.
[62] Dokukin, M. E.; Sokolov, I., On the measurements of rigidity modulus of soft
materials in nanoindentation experiments at small depth, Macromolecules 2012, 45, 4277-
4288.
[63] Pharr, G. M. Measurement of mechanical properties by ultralow load indentation,
Materials Science & Engineering A: Structural Materials: Properties, Microstructure and
Processing 1998, 253, 151-159.
[64] Fang, T. H.; Chang, W. J.; Tsai, S. L., Nanomechanical characterization of polymer
using atomic force microscopy and nanoindentation, Microelectron Journal 2005, 36, 55-
59.
[65] Sokolov, I.; Firtel, M.; Henderson, G. S., In situ high-resolution AFM imaging of
biological surfaces, Journal of Vacuum Science and Technology B 1996, 14, 674-678.
54
[66] Domke, J.; Radmacher, M., Measuring the elastic properties of thin polymer films
with the atomic force microscope, Langmuir 1998, 14, 3320-3325.
[67] Butt, H. J.; Cappella, B.; Kappl, M., Force measurements with the atomic force
microscope: technique, interpretation and applications, Surface Science Reports 2005, 59,
1-152.
[68] Dimitriadis, E. K.; Horkay, F.; Kachar, B.; Maresca, J.; Chadwick, R. S.,
Determination of the elastic modulus of thin gels using the atomic force microscope. In
biophysics of the cochlea: From molecules to models, Proceedings of the International
Symposium, Titisee, Germany, 2002; Gummer, A. W., Ed.; World Scientific: River Edge,
NJ, 2003, 177-178.
[69] Lin, D. C.; Dimitriadis, E. K.; Horkay, F., Elastic modulus maps of soft, rubber-like
gels and tissues from AFM nanoindentation, Biophysical Journal 2007, 514a−514a
[70] Tranchida, D.; Kiflie, Z.; Acierno, S.; Piccarolo, S., Nanoscale mechanical
characterization of polymers by atomic force microscopy (AFM) nanoindentations:
viscoelastic characterization of a model material, Measurement Science and Technology
2009, 20, 095702.
[71] Christoph, R.; Schaffer, T. E.; Kienberger, F.; Stracke, W.; Oberleithner, H., Vacuolar
structures can be identified by AFM elasticity mapping, Ultramicroscopy 2007, 107, 895-
901.
55
[72] Chizhik, S. A.; Huang, Z; Gorbunov, V.V.; Myshkin, N. K.; Tsukruk, V. V.,
Micromechanical properties of elastic polymeric materials as probed by Scanning Force
Microscopy, Langmuir 1998, 14,2606-2609.
[73] Bruch, L. W., Theory of physisorption interactions, Surface Science 1983, 125, 194-
217.
[74] Israelachvili, N., Intermolecular and surface forces, Academic, New York, 1985.
[75] Frank, O.; Goodman; Garcia, N., Roles of the attractive and repulsive forces in atomic-
force microscopy, Physical Review B 1991, 43, 4728-4731.
[76] Drake, B.; Prater, C. B.;Weisenhorn, A. L.; Gould, S. A. C.; Albrecht, T. R.; Quate,
C. F.; Cannell, D. S.; Hansma, H. G.; Hansma, P. K., Imaging crystals, polymers, and
processes in water with the atomic force microscope, Science 1989, 243, 1586-1589.
[77] Weisenhorn, A. L.; Hansma, P. K.; Albrecht, T. R.; Quate, C. F., Forces in Atomic
force Mocroscopy in air and water, Applied Physics Letters 1989, 54, 2651-2653.
[78] Fabié, L.; Durou, H.; Ondarçuhu, T., Capillary forces during liquid nanodispensing,
Langmuir 2010, 26, 1870-1878.
[79] Noy, A.;Vezenov, D. V.; Lieber, C. M., Chemical force microscopy nanoscale probing
of fundamental chemical interactions, Handbook of Molecular Force Spectroscopy 2008,
Springer, Boston, MA, 97-122.
56
[80] Burnham, N. A. and Colton, R. J., Measuring the nanomechanical properties and
surface forces of materials using an atomic force microscope, Journal of Vacuum Science
and Technology1989, A7, 2906-2913.
[81] Xu, Ni H, Li; Gao, H. S., Elastic modulus of amorphous SiO2nanowires, Applied
Physics Letters 2006, 88, 043-108.
[82] Jenga, Y-R.; Tsai, P-C., Molecular-dynamics studies of bending mechanical
properties of empty and C60-filled carbon nanotubes under nanoindentation, The Journal
of Chemical Physics 2005,122, 2247138.
[83] Xu, Li, X. D.; Sutton, W. J.; Mello, M. A., In situ nanoscale in plane deformation
studies of ultrathin polymeric films during tensile deformation using atomic force
microscopy and digital image correlation techniques, IEEE Transactions on
Nanotechnology 2007, 6, 4-12.
[84] Orso, S; Wegst, U. G. K.; Arzt, E., The elastic modulus of spruce wood cell wall
material measured by an in situ bending technique, Journal of Materials Science 2006, 41,
5122-5126.
[85] Tan, E. P. S.; Goh, C. N.; Sow, C. H.; Lim, C. T., Tensile test of a single nanofibre
using an atomic force microscope tip, Applied Physics Letters 2005, 86, 073115.
[86] García, R.; Magerle, R.; Perez, R., Nanoscale compositional mapping with gentle
forces, Nature Materials 2007, 6, 405-411.
57
[87] Tamayo, J.; Garcia, R., Effects of elastic and inelastic interactions on phase contrast
images in tapping-mode scanning force microscopy, Applied Physics Letters 1997, 71,
2394-3296.
[88] Lee, M.H.; Jhe, W., General theory of amplitude-modulation atomic force
microscopy. Physical Review Letters 2006, 97, 036104.
[89] Hurley, D. C.; Müller, M. K.; Kos, A. B.;Geiss, R. H., Nanoscale elastic-property
measurements and mapping using atomic force acoustic microscopy methods,
Measurement Science and Technology 2005, 16, 2167-2172.
[90] Xu, W.; Paula, M.; Adams, W.; Robertson, C. G., Measuring local viscoelastic
properties of complex materials with tapping mode atomic force microscopy, Polymer
2006, 47, 4798-4810.
[91] Ptak, A.; Kappl, M.; Butt, H-J., Modified atomic force microscope for high-rate
dynamic force spectroscopy, Applied Physics Letters 2006, 88, 263111.
[92] Oral, A.; Grimble, R. A.; Ȍzer,H. Ȍ.; Pethica, J. B., High-sensitivity noncontact
atomic force microscope / scanning tunneling microscope (AFM / STM) operating at
subangstrom oscillation amplitudes for atomic resolution imaging and force spectroscopy,
Review of Scientific Instruments 2003, 74, 3656-3663.
[93] Shulha, H.; Kovalev, A.; Myshkin, N.; Tsukruk, V. V., Some aspects of AFM
nanomechanical probing of surface polymer films, European Polymer Journal 2004, 40,
949-956.
58
[94] Heuberger, M. G.; Schlapbach, L., Mapping the local Young's modulus by analysis of
the elastic deformations occurring in atomic force microscopy, Nanotechnology 1994, 5,
12-23.
[95] Motoki, T.; Gao, W.; Kiyono, S.; Ono, T., A nanoindentation instrument for
mechanical property measurement of 3D micro/nano-structured surfaces, Measurement
Science and Technology 2006,17, 495-499.
[96] John, R.; Withers, D.; Aston, E., Nanomechanical measurements with AFM in the
elastic limit, Advances in Colloid and Interface Science 2006, 120, 57-67.
[97] Maivald, P.; Butt, H. J.; Gould, S. A. C.; Prater, C. B.; Drakei, B.; Gurieyt, J. A.;
Eiings, V. B.; Mansma, P. K., Using force modulation to image surface elasticities with
the atomic force microscope, Nanotechnology 1991, 2, 103-106.
[98] Yamanaka, K.; Nakano, S., Ultrasonic atomic force microscope with overtone
excitation of cantilever, Japanese Journal of Applied Physics 1996, 35, 3787-3792.
[99] Akhremitchev, B. B.; Brown, H, G.; Graner, S. R.; Walker, G. C., Force modulation
elasticity mapping of plastic-embedded, thin-sectioned skeletal muscle, Microscopy
Microanalysis 2001, 7, 32-38.
[100] Geer, R. E.; Kolosovb, O. V.; Briggs, G. A. D.; Shekhawat, G. S., Nanometer-scale
mechanical imaging of aluminum damascene interconnect structures in a low-dielectric-
constant polymer, Journal of Applied Physics 2002, 91, 4549-4555.
59
[101] Change, A. C. et al., Practical assessment of nanoscale indentation techniques for the
biomechanical properties of biological materials, Mechanics of Materials 2016, 98, 11-21.
[102] Calabri, L.; Pugno, N.; Valeri, S., Applied scanning probe methods XI (AFM
nanoindentation method: Geometrical effects of the indenter tip) 2009, 139-164.
[103] Charras, G. T.; Horton, M. A., Single Cell mechanotransduction and its modulation
analyzed by atomic force microscope indentation, Biophysical Journal 2002, 82, 2970-
2981.
[104] Dokukin, M. E.; Sokolov, I., On the measurements of rigidity modulus of soft
materials in nanoindentation experiments at small depth, Macromolecules 2012, 45, 4277-
4288.
[105] Guo, Q.; Xia, Y.; Sandig, M.; Yang, J., Characterization of cell elasticity correlated
with cell morphology by atomic force microscope, Journal of Biomechanics 2012, 304-
309.
[106] Nikkhah, Strobl, J. S.; De Vita, R.; Agah, M., The cytoskeletal organization of breast
carcinoma and fibroblast cells inside three dimensional (3-D) isotropic silicon
microstructures, Biomaterials 2010, 31 4552-4561.
[107] Zhou, Z.; Zheng, C.; Zhou, X.; Liu, Z.; He, Q.; Zhang, N.; Ngan, A.; Tang, B.;
Wang, A., AFM nanoindentation detection of the elastic modulus of tongue squamous
carcinoma cells with different metastatic potentials, Nanomedicine 2013, 9, 864-874.
60
[108] Plodinec, M. at al., The nanomechanical signature of breast cancer, Nature
Nanotechnology 2012, 7, 757-765.
[109] Baniasad, M. et al., Nanoindentation of Pseudomonas aeruginosa bacterial
biofilm using atomic force microscopy, Materials Research Express 2014, 1, 045411.
[110] Castellanos, M.; Carrillo, P. J.; Mateu, M. G., Quantitatively probing propensity for
structural transitions in engineered virus nanoparticles by single-molecule mechanical
analysis, Nanoscale 2015, 7 5654-5664
[111] Tranchida, D.; Piccarolo, S.; Soliman, M., Nanoscale mechanical characterization of
polymers by atomic force microscopy (AFM) nanoindentations: viscoelastic
characterization of a model material, Measurement of Science and Technology 2009, 20,
095702.
[112] Fang, T. H.; Chang, W. J.; Tsai, S. L. Nanomechanical characterization of polymer
using atomic force microscopy and nanoindentation. Microelectronics Journal 2005, 36,
55-59.
[113] Kohn, J. C.; Ebenstein, D. M., Eliminating adhesion errors in nanoindentation of
compliant polymers and hydrogels, Journal of the Mechanical Behavior of Biomedical
Materials 2013, 20, 316-326.
61
[114] Radmacher, M.; Cleveland, J. P.; Fritz, M.; Hansma, H. G.; Hansma, P. K., Mapping
Interaction Forces with the Atomic Force Microscope, Biophysical Journal 1994, 66, 59-
2165.
[115] Chopinet, L.; Formosa, C.; Rols, M. P.; Duval, R. E.; Dague, E., Imaging living cells
surface and quantifying its properties at high resolution using AFM in QI™ mode, Micron
2013, 48, 26-33.
[116] Garcıa, R.; Perez, R., Dynamic atomic force microscopy methods, Surface Science
Reports 2002, 47, 197-301.
[117] Vakarelski, I U.; Toritani, A.; Nakayama, M.; Higashitani, K., Deformation and
adhesion of elastomer microparticles evaluated by AFM, Langmuir 2001, 17, 4739-4745.
[118] Buzio,R.; Bosca, A.; Krol, S.; Marchetto, D.; Valeri, S.; Valbusa, U., Deformation
and adhesion of elastomer poly(dimethylsiloxane) colloidal AFM probes, Langmuir 2007,
23, 9293-9302.
[119] Sun, Y.; Walker, G. C., Viscoelastic response of poly(dimethylsiloxane) in the
adhesive interaction with AFM tips, Langmuir 2005, 21, 8694-8702.
[120] Maugis, D.; Barquins, M., Fracture mechanics and the adherence of viscoelastic
bodies, Journal of Physics D: Applied Physics 1978, 11, 1989-2023.
62
[121] Greenwood, J. A.; Johnson, K. L. Oscillatory loading of a viscoelastic adhesive
contact, Journal of Colloid and Interface Science 2006, 296, 284-291.
[122] Cappella, B.; Kaliappan, S. K.; Sturm, H., Using AFM force–distance curves to study
the glass-to-rubber transition of amorphous polymers and their elastic−plastic properties
as a function of temperature, Macromolecules 2005, 38, 1874-1881.
[123] Tsukruk, V. V.; Gorbunov, V. V.; Huang, Z.; Chizhik, S., Dynamic microprobing of
viscoelastic polymer properties, Polymer International 2000, 49, 441-444.
[124] Liu, B. H.; Li, L. K.; Huang, W. K.; Liao, J. D., Nanomechanical probing of the
septum and surrounding substances on Streptococcus mutans cells and biofilms, Colloids
and Surfaces B: Biointerfaces 2013, 110, 356-362.
[125] Ebenstein, D. M.; Pruitt, L. A., Nanoindentation of biological materials, Nanotoday
2006, 3, 26-33.
[126] Kaufman, J. D.; Klapperich, C. M., Surface detection errors cause overestimation of
the modulus in nanoindentation on soft materials, Journal of the Mechanical Behavior of
Biomedical Materials 2009, 2, 312-317.
[127] Yuan, Y; Verma, R., Measuring microelastic properties of stratum corneum, Colloids
and Surfaces B: Biointerfaces 2006, 48, 6-12.
63
[128] Han, L; Frank, E. H.; Greene, J. J.; Lee, H. Y.; Hung, H. H.; Grodzinsky, A. J.; Ortiz,
C., Time-Dependent Nanomechanics of Cartilage, Biophysical Journal 2011, 100, 1846-
1854.
[129] Baker, S. P., Between nanoindentation and scanning force microscopy: measuring
mechanical properties in the nanometer regime, Thin Solid Films 1997, 309, 289-296.
[130] Yang, Y. T.; Lin, C. C.; Liao, J. D.; Chang, C. W.; Ju, M. S., Continuous depth-
sensing nano-mechanical characterization of living, fixed and dehydrated cells attached
on a glass substrate, Nanotechnology 2010, 21, 1-7.
[131] Geerligs, M.; Van Breeman, L.; Peters, G.; Ackermans, P.; Baaijens, F.; Oomens, C.,
In vitro indentation to determine the mechanical properties of epidermis, Journal of
Biomechanics 2011, 44, 1176-1181.
[132] Chen, Xi; Vlassak, J. J., Numerical study on the measurement of thin film mechanical
properties by means of nanoindentation, Journal of Materials Research 2001, 16, 2974-
2982.
[133] Tsui, T. Y., Pharr, G. M., Substrate effects on nanoindentation mechanical property
measurement of soft films on hard substrates, Journal of Materials Research 1999, 14,
292-301.
[134] Hay, J.; Crawford, B., Measuring substrate-independent modulus of thin films,
Journal of Materials Research 2011, 26, 727-738.
64
[135] Akhremitchev, B. B.; Walker, G. C., Finite sample thickness effects on elasticity
determination using atomic force microscopy, Langmuir 1999, 15, 5630-5634.
[136] Kim, M. T., Influence of substrate on the elastic reaction of films for the
microindentation tests, Thin Solid Films 1996, 283, 12-16.
[137] Domke, J.; Radmacher, M., Measuring the elastic properties of thin polymer films
with the atomic force microscope, Langmuir 1998, 14, 3320-3325.
[138] Dhaliwal, R. S.; Rau, I. S., The axisymmetric Boussinesq problem for a thick elastic
layer under a punch of arbitrary profile, International Journal of Engineering Science
1970, 8, 843-856.
[139] Denisin, A. K.; Pruitt, B. L., Tuning the range of polyacrylamide gel stiffness for
mechanobiology applications, ACS Applied Materials & Interfaces 2016, 8, 21893-21902.
[140] Nguyen, H. K.; Fujinami, S.; Nakajima, K., Size-dependent elastic modulus of
ultrathin polymer films in glassy and rubbery states, Polymer 2016, 105, 64-71.
[141] Johnson, K. L., Contact mechanics and adhesion of viscoelastic spheres, ACS
Symposium Series 2000, 741, 24-41.
[142] Sun, Y.; Akhremitchev, B. B.; and Walker, G. C., Using the adhesive interaction
between Atomic Force Microscopy tips and polymer surfaces to measure the elastic
modulus of compliant samples, Langmuir 2004, 20, 5837-5845.
65
[143] Vorvolakos, K.; Chaudhury, M. K., The effects of molecular weight and temperature
on the kinetic friction of silicone rubbers, Langmuir 2003, 19, 6778-6787.
[144] David, C.; Lin; Hokay, F., Nanomechanics of polymer gels and biological tissues: A
critical review of analytical approaches in the Hertzian regime and beyond, Soft Matter
2008, 4, 669-682.
[145] Guo, S.; Akhremitchev, B. B., Packing density and structural heterogeneity of insulin
amyloid fibrils measured by AFM nanoindentation, Biomacromolecules 2006, 5, 1630-
1636.
[146] Jaasma, M. J.; Jackson, W. M.; Keaveny, T. M., Measurement and characterization
of whole-cell mechanical behavior, Annal of Biomedical Engineering 2006, 34, 748-758.
[147] Di Mundo, R.; Recchia, G.; Parekh, M.; Ruzza, A.; Ferrari, S.; Carbone, G., Sensing
inhomogeneous mechanical properties of human corneal Descemet's membrane with AFM
nano-indentation, Journal of the Mechanical Behavior of Biomedical Materials 2017, 74,
21-27.
[148] Fortier, H.; Variola, F.; Wang, C.; Zou, S., AFM force indentation analysis on
leukemia cells, Analytical Methods 2016, 8, 4421-4431.
[149] Kilpatrick, J. I.; Revenko, I.; Rodriguez, B. J., Nanomechanics of cells and
biomaterials studied by atomic force microscopy, Advanced Healthcare Materials 2015, 4,
2456-2474.
66
[150] Demichelis, A.; Divieto, C.; Mortati, L.; Pavarelli, S.; Sassi, G.; Sassi, M. P., Toward
the realization of reproducible atomic force microscopy measurements of elastic modulus
in biological samples, Journal of Biomechanics. 2015, 48, 1099-1104.
[151] Han, B.; Nia, H. T.; Wang, C.; Chandrasekaran, P.; Li, Q.; Chery, D. R.; Li, H.;
Grodzinsky, A. J.; Han, L., AFM-nanomechanical test: An interdisciplinary tool that links
the understanding of cartilage and meniscus biomechanics, osteoarthritis degeneration, and
tissue engineering, ACS Biomaterials Science and Engineering 2017, 3, 2033-2049.
[152] Kurland, N. E.; Drira, Z.; Yadavalli, V. K., Measurement of nanomechanical
properties of biomolecules using atomic force microscopy, Micron 2012, 43, 116-128.
[153] Madhura P. Nijsure, Meet Pastakia, Joseph Spano, Michael B. Fenn, Vipuil Kishore
Bioglass incorporation improves mechanical properties and enhances cell-mediated
mineralization on electrochemically aligned collagen threads. Journal of Biomedical
Materials Research Part A, 2017, 105A, 2429–2440.
[154] Emsley, C. G.; Knight, R. W.; Farndale, M. J. Barnes, J., Structure of the integrin
alpha 2 beta1-binding collagen peptide, Molecular Biology 2004 , 335, 1019-1028.
[155] Zaman, M. H.; Trapani, L. M.; Sieminski, A. L.; Mackellar, D.; Gong, H.; Kamm,
R. D.; Wells A.; Lauffenburger, D. A.; Matsudaira, P., Migration of tumor cells in 3D
matrices is governed by matrix stiffness along with cell-matrix adhesion and proteolysis,
Progressing of the National Academy of Sciences 2006, 103, 10894.
67
[156] Engler A.J.; Sen, S.; Sweeney H. L.; Discher, D.E., Matrix elasticity directs stem cell
lineage specification, Cell 2006 , 126, 677-689.
68
Chapter 2. Experimental Section
2.1 Samples and AFM setup
In this study, four different collagen-based biomaterial samples were prepared: 1)
uncrosslinked electrochemically aligned collagen (ELAC), 2) uncrosslinked Bioglass
incorporated ELAC (BG-ELAC), 3) crosslinked-ELAC and 4) crosslinked
electrochemically compacted collagen (ECC; unaligned). ELAC and BG-ELAC were
synthesized based on previously published protocol [1]. Briefly, acid soluble monomeric
collagen solution (Purecol, 3.1 mg/mL, Advanced Biomatrix, CA) was first dialyzed
against water for 24 hrs. Bioglass particles (GLO160P/-20; ~ 20 µm particle size; MO-
SCI Corporation, MO) were suspended at 200 mg / ml in water. Then, the Bioglass
particles were mixed with dialyzed collagen at the desired ratio, 30% Bioglass : 70%
collagen (w / w). This mixture was then applied between two stainless steel wire electrodes
( electrode spacing: 1.8 mm), and electric field was applied (3 V, 30 min). Electric field
triggers the formation of a pH gradient that in turn imparts a positive charge to the collagen
molecules close to the anode and negative charge to the collagen molecules close to the
cathode. Because of their repulsion with the same charge on the electrodes, the collagen
molecules self-assemble along the isoelectric point. During this process, the Bioglass
particles become entrapped within the collagen framework to form a highly aligned and
densely packed Bioglass incorporated ELAC thread (BG-ELAC) [1]. Collagen only
threads (ELAC) were made by the same way but without adding Bioglass particles. For
the third group of samples, ELAC threads (without Bioglass) were crosslinked via
incubation in 0.625% genipin solution (in 90% ethanol) for four hours at 37 ⁰C [2]. The
69
last group of samples were the crosslinked unaligned ECC collagen matrices which were
prepared in a similar manner but by replacing the stainless steel wire electrodes with planar
graphite electrodes [3].
2.2 Experimental Design
For AFM measurements, the samples were fixed on clean clipped corner plain
microscope slides [Thermo Fisher Scientific ESCO]. The thread diameter of ELAC
samples ranged from 250 µm to 300 µm, but appeared wider when deposited on the
substrate. The ECC unaligned samples are flat with a width of almost 5 mm. The cover
glasses were fixed with epoxy glue on plain glass microscope slides. Figure 2.1 shows a
photograph for a filament of the ELAC sample after fixing on the slide.
All AFM measurements were performed on fresh samples using a Molecular Force
Probe 3D AFM (MFP-3D, Asylum Research, Santa Barbara, CA). Samples were placed
on the AFM scanner, and two separate positions of the xy scanner were identified such that
AFM probe can be moved to these positions for force indentation measurements. Before
the AFM measurements, the samples were re-hydrated in a phosphate buffer saline (PBS)
pH ~ 7, to mimic the physiological environment. Then, the AFM head was placed properly
on the scanner and the measurements started from one of the identified positions. The
scanning was performed at two different spots for each sample (two samples for the
uncrosslinked and two for crosslinked collagen) and three different scan sizes to investigate
the large scale and short scale heterogeneity, respectively.
70
Figure 2.1 shows photograph of 1 cm long filament of ELAC sample deposited on a microscopic
glass slide. After deposition on glass apparent width is 1 mm.
All AFM measurements were performed using silicon nitride probes model NP-
S20 (Veeco, Santa Barbara, CA). To quantify force acting on the probe the cantilever
spring constant was measured by the thermal noise method [4], and the typical value was
approximately 100 pN / nm. Because of the heterogeneity of the Si3N4 cantilever material,
it might be suspected that the spring constant of cantilevers might vary significantly [5].
Therefore the spring constant was measured for each cantilever used in the measurements.
The manufacturer specifies radius of curvature of the AFM tip as approximately 20
nm. If the tip radius is very sharp it might damage the soft surface; if it is very dull, no
high spatial resolution can be obtained. The tip of AFM probe used in the elastic modulus
measurements was imaged using scanning electron microscope (SEM, Philips XL30).
Figure 2.2 shows the SEM image of the typical tip used in this research. The resolution of
this image is not sufficient to resolve features on a scale of 10 nm, but the image allows to
see that the probe has a sharp tip with a radius of curvature < 50 nm. This dimension
indicates that the probe was not significantly damaged during the AFM measurements.
71
Figure 2.2 shows SEM image for the tip of a long wide rectangular cantilever used in the
experiments.
The AFM measurements followed standard measurement procedure. After fixing
the cantilever in the probe holder and then in the AFM head, the laser beam was focused
on the back of cantilever close to the tip position to give the maximum reflection signal,
and the deflection was adjusted to zero. Then, the deflection sensitivity was calibrated on
the glass surface, and afterwards the spring constant was also calibrated using the built-in
thermal motion analysis procedure.
The measurements were carried out at three different spots that were previously
selected. One spot was selected on the glass area that was not covered with biomaterial.
This spot was used for measuring the optical lever sensitivity in PBS (typical values of
inverse sensitivity range from 27 nm / V to 31 nm / V) and to image the bare glass surface
for comparison. Figure 2.3 shows the glass surface image after cleaning by deionized water
and drying with a flow of N2.
72
Deflection
range = 10 nm
-200
200
nm
A B
Figure 2.3 shows images for clean-dry glass surface before fixing the sample. The A panel shows
deflection image and the B panel shows the topographic image. The topographic image is flat
because this scale is typical for collagen samples.
The other two spots were on the collagen sample with a distance of ~ 2 mm in
between. Two spots are selected to investigate the large scale heterogeneity of elastic
properties. At each spot, first, a contact mode image of a square area of 20 × 20 μm2 was
obtained to ensure the tip is scanning over the sample (not substrate) surface. Figure 2.4
shows typical image on ELAC sample in contact mode at a relatively low set point of 0.5
V. Low setpoint is selected to maintain the integrity of the sample surface.
Optical lever sensitivity was calibrated after each scan to perform accurate
indentation measurements [5].
73
Figure 2.4 shows the typical deflection (A) and height (B) image of ELAC sample with scan size
of 20 μm × 20 μm obtained before force mapping.
2.3 Force Mapping
Force curves were collected by moving the AFM probe towards and away from the
sample surface. Motion towards the sample surface was reversed once the deflection of
the cantilever reached the pre-determined trigger value. After each measurement, the
position of the sample was changed to a new location. Measurements continued until a
map of 32 × 32 force plots were collected [6]. The resulting force volume is a two-
dimensional grid across a specific area of sample surface, which can be used to extract a
map of the mechanical, adhesive, and viscoelastic properties over this area. During the
force mapping, the relative trigger was selected at 25 nm, meaning that the maximum
compressive force will be limited to 25·kc.≈ 2.5 nN for all force curves. The force curve
before the trigger is reached is called the approach force curve and after the trigger the
74
retracting curve starts. Collected data are stored on the computer and then analyzed using
custom software written for Matlab.
At each spot, the force mapping is performed over three different scan sizes, 640
nm × 640 nm, 3 μm × 3 μm, and 15 μm × 15 μm, corresponding to 20 nm, 95 nm, and 470
nm distances between two nearby force plots. The rationale for mapping small scan sizes
is to investigate the small scale heterogeneity of elastic modulus and to maintain the optical
lever sensitivity, since it might vary for large scan sizes.
2.4 Data Analysis
The AFM instrument measures cantilever deflection voltage and sample
displacement. To extract information about the local mechanical properties of materials
these data should be converted to force and indentation values. The experimentally
measured force vs. indentation dependence is compared with predictions of theoretical
models. In this section instrument calibration and several theoretical models are
considered.
2.4.1 Calibration of cantilever sensitivity and force constant
The AFM instrument measures cantilever deflection voltage and sample
displacement. To extract information about the local mechanical properties of materials
these data should be converted to force and indentation values. To obtain force and
indentation values, the cantilever deflection sensitivity should be calibrated.
When the laser beam reflects off the cantilever, it goes to the split photodiode and
produced the deflection voltage. When the end of the cantilever moves the deflection
75
voltage changes as illustrated in Figure 2.5. The sensitivity tells us how much the voltage
changes with a cantilever motion (the slope of ΔV vs. ΔZ), where ΔZ is the change in piezo
position, the deflection [7].
Calibration of optical lever sensitivity was carried out by changing the z piezo
position until the tip reached a hard surface (the deflection occurs) and change in the
deflection voltage was recorded. Figure 2.6 shows the relationship between changes in
deflection voltage and displacement, and the calibration of the optical lever sensitivity.
Figure 2.5 shows the dependence of voltage value on the cantilever motion. The repulsive force
bends the cantilever up and shifts the laser, and hence the voltage, to the positive value (Figure
adapted from http://www.nanophys.kth.se/)
Optical lever sensitivity is measured on a hard surface:
OLS (optical lever sensitivity) = ∆𝑉
𝑑 (2.1)
76
where ∆V is the change in the deflection voltage and d is the cantilever deflection.
Figure 2.6 shows a deflection vs. z displacement curve on a hard surface.
The measured value is used to convert the deflection voltage to the deflection distance at
other positions on the sample:
d = ∆𝑉
OLS= ∆𝑉 × Inv. OLS (2.2)
where Inv.OLS is the inverse optical lever sensitivity. Displacement of the base of
cantilever is the sum of the cantilever deflection d and the indentation δ.
Z = d + δ (2.3)
Then
∆Z = ∆V × Inv.OLS + δ (2.4)
77
Therefore, the cantilever sensitivity should be calibrated on surface that is much harder
than the samples (such as mica or a glass slide) to ensure that δ is close to zero in equation
(2.4) so that for calibration
Inv. OLS =∆𝑍
∆𝑉 (2.5)
The value of Inv.OLS should be relatively low (less than 100 nm / V); higher values
lead to a weak sensitivity, meaning that a large cantilever deflection is needed to achieve a
little change in voltage. Consequently, low sensitivity results in considerable contribution
of electronics noise to the measured signal, degrading performance of the instrument. Also,
low sensitivity might indicate that the AFM probe got damaged by handling or during
experiments. When the cantilever deflection is calibrated, the sample indentation can be
calculated as.
𝛿 = 𝑍 − ∆𝑉 × Inv. OLS (2.6)
The applied force can be calculated as follows:
F = kc × d (2.7)
where d is the cantilever deflection. Selecting cantilever with particular spring constant
requires taking into account several considerations. The most sensitive measurements can
be obtained when the cantilever spring constant is comparable to that for the sample.
However, the deflection voltage of a very soft cantilever in presence of adhesive forces
might exceed the voltage range of the photodiode. On the other hand, deflection of too
stiff cantilever will have significant contribution from the system electronic noise [4]. In
78
measurements spring constant of cantilevers was calibrated using the built-in thermal noise
method [4].
During the experiments, the Inv.OLS values in PBS were between 24 nm / V and
31 nm / V. The spring constant determined by the thermal noise method for a long wide
cantilever ranges from 98 pN / nm to 125 pN / nm, so the maximum applied forces
calculated by Eq. (2.6) range from 2.4 nN to 3.8 nN. After calibration, the force-
displacement curve can be obtained as shown in Figure 2.7.
Figure 2.7 shows a typical force-displacement curve
2.4.2 Indentation Models
Indentation values were obtained as described in eq. 2.6. However, it should be
noted that the displacement Z does not have zero corresponding to the sample surface, but
is determined up to some constant. Therefore the extracted indentation will also have some
offset. This issue in detail is discussed in section 1.7.5. This problem can be particularly
79
significant for soft samples [8]. Therefore, initially indentation is determined with an offset
(as shown in Figure 2.8) and the position of the contact point is used as an adjustable
parameter during data reduction.
-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0
0
0.5
1
1.5
2
2.5
3
Fo
rce,
nN
Indentation, nm
Surface
Figure 2.8 shows a typical force-indentation curve
For example, the force-indentation dependence without adhesion for paraboloidal
probe is fitted using the model [8]:
F = 4
3
𝐸
1–𝜎2 √𝑅 𝛿3 (2.7)
where δ is the indentation determined with offset as in Figure 2.8 with an added offset (𝛿 =
𝛿𝑒𝑥𝑝 + 𝛿𝑜𝑓𝑓), F is the applied force, R is the probe’s radius of curvature, σ is the Poisson
ratio of the sample and E is the elastic (Young’s) modulus of the sample. Then off is used
as a fitting parameter. The Poisson ratio cannot be obtained separately from the elastic
modulus in indentation measurements. Therefore another fitting parameter in this model
is the modified elastic modulus E*
80
𝐸∗ =𝐸
1–𝜎2 (2.8)
As can be seen from SEM image of the AFM probe, the initial curved apex then
transitions into somewhat conical shape. Therefore, if indentation is large then
paraboloidal shape model given by eq. 2.8 might be inaccurate. Therefore probe shape that
contains both the rounded apex and conical body can be considered. One of such shapes
is the hyperboloidal shape [8] that is shown in Figure 2.9.
Paraboloidal
Hyperboloidal
R
a
Figure 2.9 shows comparison of paraboloidal and hyperboloidal shapes that have the same radius
of curvature R. In the figure a is the semi-vertical angle of hyperboloidal probe.
For a hyperboloid probe profile force-indentation dependence cannot be derived as
a closed-form expression. However, this dependence can be calculated using the radius of
contact area as a parameter [8]:
𝐹 =𝐸𝑎3
(1−𝜎2)𝑅[𝜁2 +
𝜉
2 (1 − 𝜉2) {
𝜋
2 + 𝑎𝑟𝑐𝑡𝑎𝑛 (
1
2𝜉 −
𝜉
2)}] (2.9)
81
𝛿 = 𝑎2
2𝑅 𝜁 [
𝜋
2 + 𝑎𝑟𝑐𝑡𝑎𝑛 (
1
2𝜁−
𝜁
2)] (2.10)
where
𝜁 =𝑅 𝑐𝑜𝑡 (𝛼)
𝑎 (2.11)
Here a is the radius of contact area, α is the semi-vertical angle of the tip and other
parameters as described for the paraboloidal probe model eq. 3.7.
The modified elastic modulus values obtained from data analysis with the above
models that do not include adhesion are used as an initial guess to fit the data with models
that consider adhesion [6]. In fitting models with adhesion one more fitting parameter is
used: the energy of adhesion, . Here for both probe shape model the equations use the
radius of contact area, a, that connects force and indentation. For the paraboloidal probe
the JKR theory (after Johnson, Kendall and Roberts [9]) gives force and indentation as:
3 *3 *4
83
a EF a E
R (2.12)
2
*
2a a
R E
(2.13)
The JKR model takes into account elastic deformation of the sample surface, but it
ignores the long range attractive forces. Samples used in this study are soft and elastic
deformation is significant. Moreover measurements do not detect long range attractive
forces. Therefore it is reasonable to apply this model to analyze indentation data on
82
samples studied here. Similar model that takes adhesion into account can be derived for
the hyperboloidal probe model [6]:
* 2 2 2 2* 3
2 22arcsin 8
4
E A a A a AF a A E a
R a A
(2.14)
2 2
2 2 *
22arcsin
4
a A a A a
R a A E
(2.15)
Here a is radius of contact area, cot( )A R a , and other parameters as described above.
83
2.5 References
[1] Nijsure, M. P.; Pastakia, M.; Spano, J.; Fenn, M. B.; Kishore, V., Bioglass incorporation
improves mechanical properties and enhances cell-mediated mineralization on
electrochemically aligned collagen threads, Journal of Biomedical Materials Research
Part A, 2017, 105A, 2429-2440.
[2] Uquillas, J. A.; Kishore, V.; Akkus, O., Optimized crosslinking elevates the strength of
electrochemically aligned collagen to the level of tendons, Journal of Mechanical Behavior
for Biomedical Materials 2012, 15C, 176-189.
[3] Kishore, V.; Iyer, R.; Frandsen, A.; Nguyen, T-U., In vitro characterization of
electrochemically compacted collagen matrices for corneal applications, Biomedical
Materials 2016, 11, 055008.
[4] Proksch, R.; Schȁffer, T. E.; Cleveland, J. P.; Callahan, r. C.; Viani, M. B., Finite optical
spot size and position corrections in thermal spring constant calibration, Nanotechnology
2004,15, 1344-1350.
[5] Ducker, W. A.; Senden, T. J.; Pashley, R. M., Direct measurement of colloidal forces
using an atomic force microscope, Nature 1991, 353, 239-241.
[6] Sun, Y.; Akhremitchev, B. B.; and Walker, G. C., Using the adhesive interaction
between Atomic Force Microscopy tips and polymer surfaces to measure the elastic
modulus of compliant samples, Langmuir 2004, 20, 5837-5845.
84
[7] Attard, P., Measurement and interpretation of elastic and viscoelastic properties with
the atomic force microscope, Journal of Physics: Condensed Matter 2007, 19, 1-30.
[8] Akhremitchev, B. B.; Walker, C. G., Finite sample thickness effect on elasticity
determination using Atomic Force Microscopy, Langmuir 1999, 15, 5630-5634.
[9] Johnson, K. L.; Kendall, K.; Roberts, A. D., Surface energy and contact of elastic solids,
Proceedings of the Royal Society of London Series a-Mathematical and Physical Sciences
1971, 324, 301-313.
85
Chapter 3. Results
AFM images for all samples, ELAC, 30% BG-ELAC, crosslinked ELAC and
crosslinked unaligned collagen have been collected before and after force mapping to
confirm the integrity of measured surface after force mapping. Figure 3.1 shows the height
(right panel) and deflection (left panel) images for the typical sample of each type collected
after performing the force mapping measurements; scan size in all four images was 15 µm
× 15 µm. If there was any damage during the mapping, squares of 640 nm and 3 µm would
be visible in the middle of images below. Absence of such squares indicates no damage to
the sample surface using the selected range of forces during indentation measurements.
A
B
86
Figure 3.1A-D show deflection (left) and height (right) images at 15 µm scan size. Panels A-D
show images of typical ELAC, 30% BG-ELAC, crosslinked-ELAC, and crosslinked unaligned
collagen samples, respectively.
Approach and withdraw force plots have been acquired for each point in the map
(32 × 32 = 1024 force plot). The force-indentation data were analyzed using custom
software written for Matlab. Figures 3.3-3.6 show typical approach and withdraw force
curves for each sample. Force curves were analyzed using the least squares fit of the JKR
model eq. 2.12 and 2.13 to the data. For each force curve analysis provides elastic modulus
(E*, see eq. 2.8), position of the surface and for curves exhibiting adhesion, energy of
adhesion parameter as well.
C
D
87
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
500
0
1
2
3
For
ce,n
NMaximum
forceSurfac
e
A
-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0
0
1
2
3
Fo
rce,
nN
Relative Indentation, nm
B
Surfac
e
Maximum
force
Figure 3.2 displays fitting typical approach (A) and withdraw (B) force plots of ELAC sample using
paraboloidal probe shape. Data analyzed to maximum compressive force of Fmax = 1 nN.
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
500
0
1
2
3
For
ce,n
N
A
-
500-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
500
0
1
2
3
For
ce,n
N
Relative Indentation, nm
B
Maximum force
Maximum force
Surface
Surface
Figure 3.3 shows fitting typical approach (A) and withdraw (B) force plots of BG-ELAC sample
using paraboloidal probe shape. Data analyzed to maximum compressive force of Fmax= 1 nN.
88
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
50 0
0
1
2
3
For
ce,n
N
A
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
50 0
0
1
2
3
For
ce,n
N
Relative Indentation, nm
B
Maximum force
Surface
Surface
Maximum force
Figure 3.4 displays fitting typical approach (A) and withdraw (B) force plots of crosslinked-ELAC
sample using paraboloidal probe shape. Data analyzed at Fmax = 1 nN.
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
500
0
1
2
3
Forc
e,nN
A
-
500
-
450
-
400
-
350
-
300
-
250
-
200
-
150
-
100
-
500
0
1
2
3
Forc
e,nN
Relative Indentation, nm
B
Maximum
force
Surfac
e
Surfac
e
Maximum
force
Figure 3.5 shows fitting typical approach (A) and withdraw (B) force plots of crosslinked unaligned
collagen sample using paraboloidal probe shape. Data analyzed at Fmax = 1 nN.
89
Analysis of all force plots (1024 analyzed for each map) provides maps of elastic
modulus (E*) for each sample at two spots and three different scan sizes to characterize
local elastic modulus and investigate the large range and short range heterogeneity. As an
example, Figure 3.7 illustrates the heterogeneity in maps of elasticity over three scan sizes:
640 nm, 3 µm and 15 µm for crosslinked-ELAC sample. Separate maps are shown for
approach and withdraw scan direction. In analysis range of compressive forces was limited
to 1 nN maximum force; the JKR paraboloidal probe shape model was used.
5 10 15 20 25 30
5
10
15
20
25
30
Approach
Elastic modulus [MPa]; Mean 0.609 Std 0.133 Elastic modulus [MPa]; Mean 0.580 Std 0.119
5 10 15 20 25 30
5
10
15
20
25
30
Withdraw
640 nm
scan size
0.25 MPa
1.35 MPa
Elastic modulus [MPa]; Mean 1.340 Std 1.060
5 10 15 20 25 30
5
10
15
20
25
30
3 um
scan size
Elastic modulus [MPa]; Mean 0.881 Std 0.520
5 10 15 20 25 30
5
10
15
20
25
30
0.25 MPa
4.6 MPa
90
Elastic modulus [MPa]; Mean 1.480 Std 0.962
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.924 Std 0.470
5 10 15 20 25 30
5
10
15
20
25
30
15 um scan size
0.1 MPa
5.6 MPa
Figure 3.6 shows the maps of elasticity of crosslinked-ELAC at the three different scan sizes, and
approach and withdraw scan directions using 1 nN maximum force and paraboloidal probe shape
model. Numbers on axes indicate pixel position. Blue pixels show places were collected force
plots could not be analyzed.
The average of elastic modulus values for each map with the corresponding
standard deviation were calculated using Matlab program. Tables 3.1 to 3.14 display these
values for the four different samples.
Table 3.1. Elastic modulus values and their corresponding standard deviations (Std) of sample 1, 1st spot map (1024 force curves) of ELAC sample using maximum compressive forces of 1 nN and 2 nN, extracted by fitting with JKR model.
Scan
sizes
Scan
direction
E*, MPa
Fmax 1 nN Fmax 2 nN
640 nm
Approach
Std
0.112
0.058
0.317
0.053
Withdraw
Std
0.319
0.099
0.569
0.151
3 um
Approach
Std
0.077
0.040
0.155
0.069
Withdraw
Std
0.156
0.076
0.219
0.139
15 um
Approach
Std
0.165
0.079
0.227
0.105
Withdraw
Std
0.176
0.082
0.253
0.110
91
Table 3.2. Elastic modulus values and their corresponding standard deviations (Std) of sample 1,
2nd spot map (1024 force curves) of ELAC sample using maximum compressive forces of 1 nN and
2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax 1 nN Fmax 2 nN
640 nm
Approach
Std
0.104
0.063
0.151
0.091
Withdraw
Std
0.135
0.102
0.221
0.148
3 um
Approach
Std
0.098
0.028
0.124
0.036
Withdraw
Std
0.111
0.041
0.149
0.054
15 um
Approach
Std
0.091
0.032
0.117
0.048
Withdraw
Std
0.116
0.054
0.145
0.082
Table 3.3. Elastic modulus values and their corresponding standard deviations (Std) of sample 2,
1st spot map (1024 force curves) of ELAC sample using maximum compressive forces of 1 nN and
2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax 1 nN Fmax 2 nN
640 nm
Approach
Std
0.139
0.007
0.146
0.005
Withdraw
Std
0.129
0.010
0.140
0.006
3 um
Approach
Std
0.141
0.052
0.132
0.034
Withdraw
Std
0.122
0.044
0.128
0.037
15 um
Approach
Std
0.114
0.031
0.116
0.027
Withdraw
Std
0.111
0.035
0.114
0.032
92
Table 3.4. Elastic modulus values and their corresponding standard deviations (Std) of sample 2,
2nd spot map (1024 force curves) of ELAC sample using maximum compressive forces of 1 nN and
2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax 1 nN Fmax 2 nN
640 nm
Approach
Std
0.090
0.006
0.075
0.003
Withdraw
Std
0.094
0.006
0.081
0.005
3 um
Approach
Std
0.103
0.008
0.087
0.007
Withdraw
Std
0.103
0.008
0.095
0.009
15 um
Approach
Std
0.118
0.011
0.109
0.019
Withdraw
Std
0.113
0.013
0.132
0.037
Table 3.5. The average of elastic modulus values and their corresponding standard deviations (Std)
of all spots on ELAC samples using maximum compressive forces of 1 nN and 2 nN, extracted by
fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax 1 nN Fmax 2 nN
640 nm
Approach
Std
0.111
0.034
0.172
0.038
Withdraw
Std
0.166
0.054
0.253
0.077
3 um
Approach
Std
0.105
0.045
0.125
0.036
Withdraw
Std
0.122
0.038
0.148
0.061
15 um
Approach
Std
0.122
0.038
0.142
0.049
Withdraw
Std
0.114
0.076
0.161
0.065
93
Table 3.6. Elastic modulus values and their corresponding standard deviations (Std) of sample 1,
1st spot map (1024 force curves) of 30% BG-ELAC sample using maximum compressive forces of
1 nN and 2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax= 1 nN Fmax = 2 nN
640 nm
Approach
Std
0.510
0.112
0.533
0.093
Withdraw
Std
0.488
0.111
0.563
0.101
3 um
Approach
Std
0.572
0.199
0.578
0.181
Withdraw
Std
0.534
0.197
0.601
0.212
15 um
Approach
Std
0.527
0.131
0.538
0.111
Withdraw
Std
0.499
0.130
0.564
0.118
Table 3.7. Elastic modulus values and their corresponding standard deviations (Std) of sample 1,
2nd spot map (1024 force curves) of 30% BG-ELAC sample using maximum compressive forces
of 1 nN and 2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax= 1 nN Fmax = 2 nN
640 nm
Approach
Std
0.529
0.101
0.587
0.094
Withdraw
Std
0.463
0.107
0.612
0.100
3 um
Approach
Std
0.570
0.149
0.625
0.129
Withdraw
Std
0.525
0.132
0.643
0.132
15 um
Approach
Std
0.605
0.199
0.676
0.173
Withdraw
Std
0.590
0.169
0.736
0.280
94
Table 3.8. Elastic modulus values and their corresponding standard deviations (Std) of sample 2,
1st spot map (1024 force curves) of 30% BG-ELAC sample using maximum compressive forces of
1 nN and 2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax= 1 nN Fmax = 2 nN
640 nm
Approach
Std
0.578
0.058
0.709
0.068
Withdraw
Std
0.719
0.106
0.852
0.106
3 um
Approach
Std
0.542
0.154
0.761
0.210
Withdraw
Std
0.804
0.296
1.020
0.340
15 um
Approach
Std
0.392
0.198
0.568
0.299
Withdraw
Std
0.593
0.373
0.796
0.371
Table 3.9. Elastic modulus values and their corresponding standard deviations (Std) of sample 2,
2nd spot map (1024 force curves) of 30% BG-ELAC sample using maximum compressive forces
of 1 nN and 2 nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax= 1 nN Fmax = 2 nN
640 nm
Approach
Std
0.777
0.164
0.756
0.137
Withdraw
Std
0.704
0.136
0.771
0.123
3 um
Approach
Std
0.826
0.225
0.792
0.178
Withdraw
Std
0.731
0.187
0.786
0.171
15 um
Approach
Std
0.778
0.205
0.780
0.173
Withdraw
Std
0.770
0.155
0.782
0.162
95
Table 3.10. The average of elastic modulus values and their corresponding standard deviations
(Std) of all spots on 30% BG-ELAC samples using maximum compressive forces of 1 nN and 2
nN, extracted by fitting with JKR model.
Scan
sizes
Std, Scan
direction
E*, MPa
Fmax= 1 nN Fmax = 2 nN
640 nm
Approach
Std
0.600
0.109
0.646
0.098
Withdraw
Std
0.594
0.115
0.700
0.107
3 um
Approach
Std
0.628
0.182
0.689
0.175
Withdraw
Std
0.649
0.203
0.762
0.213
15 um
Approach
Std
0.575
0.183
0.640
0.189
Withdraw
Std
0.613
0.206
0.715
0.231
Table 3.11. Elastic modulus values and their corresponding standard deviations (Std) of map (1024
force curves) of crosslinked-ELAC sample using maximum compressive forces of 1 nN and 2 nN,
extracted by fitting with JKR model.
Scan
sizes
Std, scan
direction
E*, MPa
Fmax= 1 nN Fmax= 2 nN
640 nm
Approach
Std
0.609
0.133
0.715
0.126
Withdraw
Std
0.580
0.119
0.769
0..145
3 um
Approach
Std
0.881
0.520
1.490
1.210
Withdraw
Std
1.340
1.060
2.240
1.940
15 um
Approach
Std
0.924
0.470
1.670
1.190
Withdraw
Std
1.480
0.962
2.450
1.860
96
Table 3.12. Elastic modulus values and their corresponding standard deviations (Std) of 1st spot on
crosslinked unaligned collagen sample using maximum compressive forces of 1 nN and 2 nN,
extracted by fitting with JKR model.
Scan
sizes
Std, scan
direction
E*, MPa
Fmax= 1 nN Fmax= 2 nN
640 nm
Approach
Std
0.922
0.226
1.390
0.251
Withdraw
Std
2.470
0.628
3.100
0.521
3 um
Approach
Std
0.732
0.187
1.220
0.219
Withdraw
Std
1.790
0.587
2.640
0.473
15 um
Approach
Std
0.570
0.316
0.887
0.449
Withdraw
Std
1.280
0.730
1.890
0.886
Table 3.13 Elastic modulus values and their corresponding standard deviations (Std) of 2nd spot
on crosslinked unaligned collagen sample using maximum compressive forces of 1 nN and 2 nN,
extracted by fitting with JKR model.
Scan
sizes
Std, scan
direction
E*, MPa
Fmax= 1 nN Fmax= 2 nN
640 nm
Approach
Std
1.040
0.196
1.760
0.329
Withdraw
Std
2.120
0.650
2.990
0.710
3 um
Approach
Std
0.842
0.241
1.360
0.361
Withdraw
Std
1.530
0.538
2.230
0.560
15 um
Approach
Std
0.772
0.253
1.140
0.343
Withdraw
Std
1.320
0.466
1.980
0.510
97
Table 3.14 Elastic modulus values and their corresponding standard deviations (Std) of all spots
on crosslinked unaligned collagen sample using maximum compressive forces of 1 nN and 2 nN,
extracted by fitting with JKR model.
Scan
sizes
Std, scan
direction
E*, MPa
Fmax= 1 nN Fmax= 2 nN
640 nm
Approach
Std
0.981
0.211
1.575
0.290
Withdraw
Std
2.295
0.639
3.040
0.616
3 um
Approach
Std
0.787
0.214
1.290
0.290
Withdraw
Std
1.660
0.562
2.435
0.517
15 um
Approach
Std
0.671
0.285
1.014
0.396
Withdraw
Std
1.300
0.600
1.930
0.698
Results shown in these tables indicate that the samples are relatively soft with
modulus E* in are range from approximately 0.1 MPa to approximately 3 MPa. It should
be noted that to obtain values of corresponding elastic modulus, results for modulus E*
should be multiplied by factor (1 - 2). The Poisson ratio ranges from 0.5 to 0, so the
factor ranges from 0.75 to 1. This represents the inherent uncertainty in extracting elastic
modulus using this method. This uncertainty can be reduced if reasonable guess about
Poisson ratio can be made.
98
Chapter 4. Discussion
4.1 Elastic modulus measurements
Previous section shows parameters obtained by fitting the JKR model to data
collected on four different types of samples. Results show that for ELAC, BG-ELAC,
crosslinked-ELAC and crosslinked-ECC samples the average extracted moduli obtained
from approach curves using 1 nN maximum compressive force are 0.12 MPa, 0.61 MPa,
0.97 MPa and 1.28 MPa with 0.05 MPa, 0.17 MPa, 0.54 MPa and 0.41 MPa standard
deviation, respectively. The same order was found at Fmax = 2 nN. Figure 4.1 shows these
average data with their standard deviations shown as error bars. In previous work, it was
found that Bioglass incorporated within ELAC threads improves the elastic modulus
significantly [1]. The higher standard deviation value of BG-ELAC than for ELAC sample
indicates less uniformity of the former. Crosslinked samples have higher Young’s modulus
than uncrosslinked collagen threads. The same observation has been reported in previous
work [2], which stated that the modulus value of soft materials decrease with decreasing
the crosslink ratio [2]. Kishore and coworkers determined the ultimate tensile stress (UTS)
for ECC material with and without treatment with genipin crosslinker [3]. They observed
a significant increase in UTS by adding genipin to the ECC matrices [3]. The crosslinked-
ECC unaligned exhibits higher modulus than crosslinked aligned, crosslinked-ELAC
sample. The higher std value of elastic modulus for cosslinked-ELAC than that of
crosslinked ECC unaligned sample might be attributed to the presence of regions with
significantly different values of the modulus as shown in Figure 3.6. No such regions are
99
observed for unaligned sample. Figure 4.2 shows the force maps of one spot of crosslinked
ECC unaligned sample at the three different scan sizes.
E, MPa
ELAC BG-ELAC Crosslinked-
ELAC
Crosslinked-
ECC
0
0.4
0.8
1.2
1.6 A
ELAC BG-ELAC Crosslinked-ELAC Crosslinked-
ECC
0
1
2
3
E, MPa
B
Figure 4.1 displays the average values of Young’s modulus of ELAC, BG-ELAC, crosslinked-
ELAC and crosslinked-ECC collagen threads at A) Fmax = 1 nN, and B) Fmax = 2 nN. Standard
deviations are shown as error bars.
100
Elastic modulus [MPa]; Mean 1.040 Std 0.196
Approach
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 2.120 Std 0.650
5 10 15 20 25 30
5
10
15
20
25
30
Withdraw
Elastic modulus [MPa]; Mean 1.530 Std 0.538
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.842 Std 0.241
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 1.320 Std 0.467
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.772 Std 0.253
5 10 15 20 25 30
5
10
15
20
25
30
640 nm
3 um
15 um
100 nm 100 nm
1 um 1 um
5 um 5 um
0.65 MPa
3.35 MPa
0.20 MPa
2.75 MPa
0.20 MPa
2.50 MPa
Figure 4.2 shows the force maps of one spot of crosslinked ECC unaligned sample at the three
different scan sizes.
Elastic modulus values measured here can be compared with elastic modulus values
of collagen gels, in which collagen molecules are organized in hydrogels with a lot of
empty space between individual molecules, and with the elastic modulus is in the sub-kPa
101
range [4]. Therefore, the densification-alignment and unalignment processes of the
investigated samples very significantly increases modulus in comparison with hydrogels.
The measured values are close to the elastic modulus of stiffer biological materials like
cartilage and meniscus (100 kPa – 1 MPa range) [5, 6]. However, this value is significantly
(about two orders of magnitude) lower than the elastic modulus of protein materials in
crystalline state [7].
The elastic modulus of similarly prepared BG-ELAC samples was reported earlier
in paper by Kishore and co-workers [1]. Results for the bulk modulus obtained using the
dynamic mechanical analyzer were approximately 10 times higher than values reported
here. It might be suggested that in part the observed difference results from uncertainty in
the radius of curvature of the AFM probe. However, the nominal value of the tip radius
was specified with 5 nm uncertainty. This means that if the extreme values of 15 nm and
25 nm were used, they would provide modulus values different by ~13% from the value
obtained using the expected radius. The measured elastic modulus is proportional to R–½
(see eq. 2.7). Therefore the radius should be approximately 100 times lower to reconcile
the observed difference, making the probe unrealistically sharp with 0.2 nm radius of
curvature. Therefore other explanation for observed difference is needed. This difference
in modulus values can be attributed to the testing direction. The modulus value of
anisotropic material depends on the direction of the applied force. In the previous work
[1], the applied force of the tensile test used to determine the elastic modulus is parallel to
collagen alignments, whereas the force applied by AFM is perpendicular to the alignments.
Furthermore, the observed difference can be explained by noting that AFM tests elastic
102
modulus of the near-surface layer. Therefore it can be suggested that the surface of samples
is considerably softer than the middle part of material. From the typical force-indentation
curves shown in Figure 3.2 – 3.6, it can be seen that the soft surface layer extends to at
least several tens of nanometers into the sample surface because after some relatively short
distance (about 50 nm – 150 nm) the fit curves start to deviate noticeably from the data. It
appears that samples are stiffer beyond the surface layer. For ELAC sample the elastic
half-space model fits the indentation data reasonably well up to the depth of approximately
150 nm (Figure 3.2) and for BG-ELAC sample the range of good fit is approximately 50
nm (Figure 3.4). For crosslinked-ELAC, the reasonable fitting is at depth up to
approximately 35 nm while that for crosslinked-ECC (unaligned) at ~ 25 nm. The
consistence of these depths coincides with that order in Figure 4.1. By applying a specific
force the tip travels largest distance into the softest, and the shortest into the stiffest
material. The upturn of the force-indentation data from the fit lines generated using the
JKR model can be observed for all samples.
4.2 Viscoelasticity
As with most biological and soft materials [8, 9, 10], our samples exhibit viscoelastic
behavior that can be seen as hysteresis between the approach and the withdraw force curves
when the probe applies compressive force to the sample (hysteresis at tensile forces is
caused by adhesion) [9, 11, 12]. Generally such hysteresis in the region of compressive
forces can be attributed to either plastic or viscoelastic behavior [13]. After the sample is
deformed plastically, the deformation persists after removal of the force. However, the
viscoelastic behavior is a time dependent deformation that occurs after the force is
103
applied/removed to/from the sample [14]. Therefore, if force curves collected on the same
spot continue to exhibit hysteresis then observed hysteresis is of viscoelastic nature. For
all samples tested here the continuously reproduced hysteresis indicates the viscoelastic
behavior. Figures 3.3 – 3.6 show this hysteresis as a difference in slope and offset between
approach and withdraw curves for ELAC, BG-ELAC, crosslinked-ELAC and crosslinked-
ECC, unaligned, samples in the region of compressive (positive) forces.
The model that is used here to analyze indentation data does not explicitly include
viscoelastic effects. The observed viscoelasticity can be quantified as differences in the
elastic modulus obtained for the approach and withdraw force curves. Such difference
would mainly manifest itself as difference in the slope of force-indentation curves near the
point of reversal of probe’s motion. Therefore higher-force regions of force curves should
be considered to compare viscoelastic behavior across the samples. This type of hysteresis
is typical for viscoelastic polymeric materials [15] and biomaterials [16].It can be noticed
that the approach and withdraw force curves shown in Figures 3.3 – 3.6 have similar slopes
for low compressive forces (below ~1 nN). Consequently effects of viscoelasticity can be
quantified as horizontal offset between approach and withdraw curves. Such offset is zero
for purely elastic samples. Figures 4.3 and 4.4 show histograms of approach-withdraw
offsets obtained for different samples at different scan sizes, as indicated in corresponding
panels.
Figure 4.3 compares force curve offsets at different scan sizes and Figure 4.4
compares values obtained at different spots on the sample surface. Offsets seem to be
similar at different scan sizes with BG-ELAC samples exhibiting lower offset than the
104
other two samples. Large offset observed on ELAC sample might be expected because
this sample is not crosslinked. Somewhat surprising is the low offsets measured for BG-
ELAC sample observed at different scan sizes and at different spots.
-50 0 50 100 150 200 250 3000
20
40
60
80
100
120
-50 0 50 100 150 200 250 3000
20
40
60
80
100
120
-50 0 50 100 150 200 250 3000
50
100
150
200
250
-50 0 50 100 150 200 250 3000
20
40
60
80
100
120
140
160
300
Oc
cu
ran
ce
Occ
ura
nce
Approach-withdraw offset, nm Approach-withdraw offset, nm
Approach-withdraw offset, nm Approach-withdraw offset, nm
Approach-withdraw offset, nm Approach-withdraw offset, nm
58.7 ± 8.6ELAC (3 um scan size)
6.4 ± 1.7ELAC (15 um scan size)
8.0 ± 0.3BG-ELAC (3 um scan size)
9.5 ± 0.3BG-ELAC (15 um scan size)
Oc
cu
ran
ce
Oc
cu
ran
ce
-50 0 50 100 150 200 2500
50
100
150
200
250
Crosslinked-ECC(3 um scan size)
Occ
ura
nce
-50 0 50 100 150 200 250 3000
20
40
60
80
100
120
140
160
180
17.8 ± 0.6Crosslinked-ECC
( 15 um scan size)
Occ
ura
nce
15.2 ± 0.3
Figure 4.3 displays the approach-withdraw offset for ELAC, BG-ELAC and crosslinked-ECC,
unaligned samples at Fmax = 1 nN using paraboloidal tip model. The numbers inside the chart are
the positions of histograms ± one standard deviation error in this position.
105
1 st spot 2 nd spot
-50 0 50 100 1500
20
40
60
80
100
120
-50 0 50 100 1500
20
40
60
80
100
120
140
160
-50 0 50 100 1500
20
40
60
80
100
120
140
160
-50 0 50 100 1500
50
100
150
200
250
-50 0 50 100 1500
50
100
150
200
250
25.3 ± 0.2
Crosslinked-ECC
-50 0 50 100 1500
20
40
60
80
100
120
140
160
180
6.4 ± 1.7
ELAC (15 um scan size)
10.6 ± 0.5
ELAC
9.5 ± 0.3
BG-ELAC
3.3 ± 0.3
BG-ELAC
17.8 ± 0.6
Crosslinked-ECC
Occ
ura
nce
Occ
ura
nce
Occ
ura
nce
Approach-withdraw offset, nm Approach-withdraw offset, nm
Figure 4.4 shows the approach-withdraw offset for the different spots using data collected with 15
μm scan size.
This reveals that the Bioglass particles affect the viscoelastic behavior of collagen
threads: It appears that samples with added Bioglass exhibit more elastic-like behavior.
For ELAC sample, it can be seen that the offset distribution dramatically changed
106
throughout the short scale of the scan size (from 58.7 − 8.6 to 6.4 + 1.7 from 3 μm to 15
μm, respectively).
For crosslinked-ECC, the viscoelasticity is larger than for BG-ELAC sample and
this is consistent at different scan sizes and spots. In the case of crosslinked-ELAC a
continuous change in viscoelastic offset was observed throughout changing the scan scales.
Figure 4.5 shows the slow increase of the mean value of approach-withdraw offset for
crosslinked-ELAC sample over the three different scan sizes, 640 nm, 3 μm and 15 μm.
This regularity might be attributed to the fact that this sample consists of regions with
relatively low and high elastic modulus, as seen in Figure 3.8. Only “low” modulus region
was tested at scan size of 640 nm, and this region corresponds to lower viscoelastic offset
that is shown on the left panel of Figure 4.5. As higher modulus areas appear in the image,
so the more pronounced is the right hand side shoulder in the histograms. These
observations are consistent with earlier notion that crosslinking increases elastic modulus
but also increases viscoelastic offset between approach-withdraw curves.
107
50
13.2 ± 0.3
15 um scan size
50
Approach-withdraw offset, nm
-20 -10 0 10 20 30 400
20
40
60
80
100
120
140
160
180
200
8.4 ± 0.1
Occ
ura
nce
Approach-withdraw offset, nm
100
120
140
160
180
-20 -10 0 10 20 30 400
20
40
60
80
10.7 ± 0.8Crosslinked-
Approach-withdraw offset, nm
-20 -10 0 10 20 30 40 500
20
40
60
80
100
120
140
160
Crosslinked-ELACELAC
3 um scan sizeCrosslinked-ELAC
640 nm scan size
Figure 4.5 displays the slow increase of viscoelasticity mean value of crosslinked-ELAC with
increasing the scan sizes as higher modulus areas are included in testing.
In the comparison of mean values of offset for the four types of sample at 15 μm scan size,
the offset distribution is overlapped within the two spot ranges of ELAC and BG-ELAC
samples, revealing the similar manner in viscoelasticity at the given scan size. On the other
hand, crosslinked materials show clearly increase in the offset most probable values from
each other and the first two samples.
4.3 Heterogeneity of Surface
The topographic map of ELAC sample in Figure 2.4 shows variations that could be
associated with local organization of collagen threads. The corresponding map and
histogram of elastic modulus are shown in Figure 4.6. It can be observed that the
distribution of elastic modulus is not wide with clear peak around the most probable value.
As discussed in the previous sub-section, the elastic modulus extracted from the withdraw
force curves is shifted towards higher values. Distance between force plots collected in
this map is approximately 0.5 µm, and considerable point-to-point variation in the elastic
108
modus observed in the modulus maps here indicate that the characteristic scale for variation
in the elastic modulus for this sample is less than 0.5 µm. The force maps collected at 640
nm scan size shown in Figure 4.7 exhibit considerably smoother variation in the modulus
with characteristic scale of modulus variation of 100 nm. This scale is similar in size with
the size of the tip-sample contact area estimated as 2(R·δ) ½ which reaches value of ~110
nm at maximum indentation. Figures 4.6, and 4.7 show the maps and histograms for both
scan directions, approach and withdraw. The withdraw maps show slightly higher
variation in distribution of modulus and thus higher standard deviations. Higher width of
distribution of the withdraw curves can be attributed to the higher uncertainty in the
modulus extracted by simultaneously taking into account the adhesion of samples.
5 10 15 20 25 30
5
10
15
20
25
30
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.109 Std
0.019
Elastic modulus [MPa]; Mean 0.132 Std
0.037
Approach Withdraw
5 um 5 um
0 0.1 0.2 0.30
50
100
150
200
250
300 Elastic modulus [MPa]
0 0.1 0.2 0.30
50
100
150
200
250
300 Elastic modulus [MPa]
0.08 MPa
0.34 MPa
Figure 4.6 shows force maps and their corresponding histograms of ELAC sample at 15 µm scan
size
109
Unlike the ELAC sample, which does not have a significant increase of
heterogeneity with increasing scan size, the standard deviation of elastic modulus for BG-
ELAC is considerably higher and it increases with increasing the scan size, as shown in
Figure 4.8. This is expected for composite materials since the larger scan size can include
more and more features with various mechanical properties. This consistence has been
shown in previous report [18], in which the samples appears to be strongly heterogeneous
with increasing the scan size.
Elastic modulus [MPa]
Mean 0. 075 Std 0.004
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]
Mean 0.081 Std 0.005
5 10 15 20 25 300
5
10
15
20
25
30
Approach Withdraw
100 nm 100 nm
0.07 0.08 0.09 0.100
20
40
60
80
100Elastic modulus [MPa]
0.07 0.08 0.09 0.100
20
40
60
80
100
120 Elastic modulus [MPa]
0.035 MPa
0.079 MPa
Figure 4.7 shows elastic modulus maps and corresponding histograms measured on ELAC sample
at 640 nm scan size.
110
5 10 15 20 25 30
5
10
15
20
25
30
5 10 15 20 25 30
5
10
15
20
25
30
5 10 15 20 25 30
5
10
15
20
25
30
a)
b)
c)
100 nm
1 um
5 um
.
Elastic modulus Mean 0.570 MPa, Std 0.149
Elastic modulus Mean 0.605 MPa, Std 0.199
0 0.5 1 1.5 20
20
40
60
80
100 Elastic modulus [MPa]
0 0.5 1 1.5 20
20
40
60
80
100
120 Elastic modulus [MPa]
0 0.5 1 1.50
50
100
150 Elastic modulus [MPa]
2
Std = 0.10 MPa
Scan size= 640 nm
Std = 0.15 MPa
Scan size = 3 um
Std = 0.2 MPa
Scan size = 15 um
0.23 MPa
0.91 MPa
0.20 MPa
1.10 MPa
Elastic modulus Mean 0.530 MPa, Std 0.101
0.2 MPa
1.70 MPa
Figure 4.8 shows elastic modulus maps and their corresponding histograms of BG-ELAC at
different scan sizes: a) 640 nm, b) 3 µm, and c) 15 µm.
Elastic map with scan size 640 nm for BG-ELAC shows noticeably sharper variation of
features than for ELAC sample. This can be attributed to the considerably lower size of
the contact area; here characteristic size 2(R·δ) ½ is approximately 60 nm. Crosslinked-
ECC sample, which has a resolution of approximately 45 nm shows features with more
sharp dimensions than BG-ELACs’. Figure 4.9 characterizes some features of the four
types of sample at the smallest scan size, 640 nm, in which the point to point distance
equals 20 nm. The small features reflect the high modulus of surface, on which the tip-
111
sample contact area is smaller than that on the lower modulus materials. Even though
crosslinked-ELAC has a characteristic size even less than BG-ELACs’ (approximately 53
nm), it exhibits sudden variations in force the map, indicating that the feature dimensions
are smaller than 53 nm.
5 10 15 20 25 30
5
10
15
20
25
30
ELAC threads at 640 nm scan size
5 10 15 20 25 300
5
10
15
20
25
30
BG-ELAC at 640 nm scan size
5 10 15 20 25 30
5
10
15
20
25
30
5 10 15 20 25 30
5
10
15
20
25
30
Crosslinked-ELAC at 640 nm scan size Crosslinked-ECC, at 640 nm scan size
100 nm100 nm 100 nm 100 nm
0.34 MPa
0.16 MPa 0.20 MPa
0.87 MPa
0.25 MPa
1.35 MPa
0.65 MPa
3.35 MPa
Figure 4.9 displays force maps of ELAC and BG-ELAC, crosslinked-ELAC and crosslinked-
ECC, unaligned samples at 640 nm scan size. The higher modulus samples show sharper features.
4.4 Depth Sensing and Tip Shape Artifacts
Tables 3.5, 3.10, 3.11 and 3.14 display the elastic modulus values of samples,
ELAC, BG-ELAC, crosslinked-ELAC and crosslinked-ECC, unaligned samples at
maximum forces of 1 nN and 2 nN, using a JKR model for paraboloidal tip shape [18]. For
all samples there are differences in elastic modulus values determined at different level of
compressive forces. By using the t-test these differences are significant although their
elastic maps exhibit unobvious differences. In a previous study [19], there was a slight
difference in modulus values between the core and shell of the collagen fibril, and both
showed similar morphological structures [20, 15]. However, findings for individual
collagen fibrils cannot be directly applied here because of difference in preparation
112
conditions and modification procedures of these materials. Figure 4.10 displays 15 µm ×
15 µm elastic modulus maps for ELAC, BG-ELAC, crosslinked-ELAC and crosslinked-
ECC, unaligned samples obtained using a paraboloidal tip model at different levels of
maximum (depth) applied force. First, The BG-ELAC maps show essentially no different
topographic features indicating absence of objects with different elastic moduli at depth up
to ~250 nm.
Elastic modulus [MPa]; Mean 0.227 Std 0.105
5 10 15 20 25 30
5
10
15
20
25
30
0 0.200 0.400 0.6000
20
40
60
80
100
120 Elastic modulus [MPa]
0 0.100 0.200 0.300 0.400 0.500 0.6000
20
40
60
80
100
120 Elastic modulus [MPa]
Elastic modulus [MPa]; Mean 0.165 Std 0.079
5 10 15 20 25 30
5
10
15
20
25
30
ELACF
max = 1 nN F
max = 2 nN
113
Elastic modulus [MPa]; Mean 0.392 Std 0.198
5 10 15 20 25 30
5
10
15
20
25
30
5 10 15 20 25 30
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.568 Std 0.299
Fmax
= 1 nN Fmax
= 2 nN
BG-ELAC
0 0.5 1 1.5 20
20
40
60
80
100
120
Elastic modulus
[MPa]
0 0.5 1 1.5 20
20
40
60
80
100
120
140
Elastic modulus [MPa]
5
10
15
20
25
30
Elastic modulus [MPa]; Mean 0.924 Std 0.470
5 10 15 20 25 30
Crosslinked-ELAC
Elastic modulus [MPa]; Mean 1.670 Std 1.190
5 10 15 20 25 30
5
10
15
20
25
30
Fmax
= 2 nNFmax
= 1 nN
0 1 2 3 4 5 60
20
40
60
80
100
120
140
160
Elastic modulus [MPa]
60 1 2 3 4 50
20
40
60
80
100
120
140
Elastic modulus [MPa]
114
Crosslinked-ECC, unaligned
Elastic modulus [MPa]; Mean 1.140 Std 0.343
5 10 15 20 25 30
5
10
15
20
25
30
100
Fmax
= 1 nN Fmax
= 2 nN
Elastic modulus [MPa]; Mean 0.772 Std 0.253
5 10 15 20 25 30
5
10
15
20
25
30
0 0.500 1.00 1.50 2.00 2.500
20
40
60
80
100
120 Elastic modulus [MPa]
0 0.50 1.00 1.50 2.00 2.500
20
40
60
80
Elastic modulus [MPa]
Figure 4.10 Force maps and corresponding histograms of ELAC, BG-ELAC, crosslinked-ELAC
and crosslinked-ECC, unaligned samples using paraboloidal tip at different indentation depths.
Data in the previously mentioned tables and histograms indicate that the samples
appear stiffer and with higher variation (higher std) at higher indentation force. The higher
standard deviation with increasing indentation depths might reflect higher heterogeneity in
direction perpendicular to sample’s surface within the sample. Increase in the elastic
modulus with increase of the applied force has observed when measuring elasticity of
living cells [21]. However, for soft sample such as corneal membrane, the indentation
profiles indicated considerable uniformity of material with depth [20]. The increase of
modulus values with the indentation depth in part can be attributed to the tip shape artifact
and not to actual increase of elastic modulus with increase of depth sensing. The AFM tip
115
does not have ideal shape of paraboloid of revolution, but rather shape with rounded apex
that becomes a cone, similar to shape of hyperboloid [22]. To avoid this tip shape artifact
in the depth sensing, studies the hyperboloid probe with adhesion model [18] was applied.
As an example, the results of crosslinked-ELAC samples are shown in Table 4.1.
Table 4.1. Elastic modulus values (MPa) and their standard deviations (Std) for one spot on
crosslinked-ELAC at (Fmax = 1 nN and 2 nN) using paraboloidal and hyperboloidal models.
Scan sizes
And
directions
E, MPa,
Standard
deviation
Fmax, JKR Fmax, Hyperboloid
1 nN 2 nN 1 nN 2 nN
Approach
640 nm
withdraw
Mean
Std 0.609
0.133
0.715
0.126
0.604
0.133
0.701
0.135 Mean
Std 0.580
0.119
0.769
0..145
0.593
0.118
0.720
0.146
Approach
3 μm
withdraw
Mean
Std 0.881
0.520
1.490
1.210
0.836
0.433
0.989
0.587 Mean
Std 1.340
1.060
2.240
1.940
0.864
0.362
0.959
0.495
Approach
15 μm
withdraw
Mean
Std 0.924
0.470
1.670
1.190
0.879
0.478
1.160
0.776 Mean
Std 1.480
0.962
2.450
1.860
0.868
0.560
1.200
0.739
To achieve the same indentation by both paraboloid and hyperboloid probe shapes
with the same radius of curvature, the higher force should be applied with the hyperboloidal
probe because the former shape displaces more material as illustrated in Figure 4.11.
116
Figure 4.11 compares shapes of the paraboloidal (gray) and hyperboloidal (blue) probes and
indicates expected indentation depths at two different force levels, as indicated in the figure.
However, the results in Table 4.1 show that using hyperboloid probe model
decreases the modulus values obtained at different indentation forces, indicating that the
paraboloidal tip underestimated the modulus values. This artifact has increased with
increasing the applied force. By using t-test, there are insignificant differences between
paraboloid and hyperboloid modulus values in the case of Fmax = 1 nN, whereas the
differences became significant at Fmax = 2 nN. Figures 4.12 a and b show the increase of
modulus difference between paraboloidal and hyperboloidal model at Fmax = 2 nN over the
given scan sizes on crosslinked-ELAC sample.
117
Figure 4.12 in panels a and b show the difference in E values by using paraboloid and hyperboloid
probes over different scan sizes for crosslinked-ELAC sample at Fmax = 1 nN and 2 nN,
respectively.
The same effect was also found in ELAC material. The reason for increasing the
tip artifact with increasing the indentation depth can be attributed to the considerable
118
change in the tip shape at larger indentation corresponding to Fmax = 2 nN as illustrated in
Figure 4.11.
4.5 Conclusion
A comprehensive approach was developed to determine and characterize surface
distribution of elastic modulus for soft biomaterials at nanoscale under physiological
conditions. The developed method takes into account sample adhesion and can be used for
probes with geometry that deviates from simple paraboloid of revolution geometry. The
hyperboloidal probe shape is added into the analysis in order to test for probe shape
artifacts. Elastic modulus values obtained for modified collagen threads are in the range
of hundreds of kilopascals. The characteristic dimensions of surface features visible in
elastic maps are consistent with estimated size of the tip-sample contact area. Even though
the visual differences between modulus maps at different maximum forces are
unobservable, difference of mechanical properties measured at the surface layer and at the
deeper probe penetration are significant as confirmed by the t-test.
No Bioglass protrusions at the surface were detected for BG-ELAC sample,
however elastic modulus of surface collagen molecules appear stiffer than for ELAC-
sample. It appears that Bioglass affects packing order of the surface collagen molecules
even though it is not present at the top sample layer. Crosslinked materials showed higher
elastic modulus than BG-ELAC threads, however crosslinking also increases viscoelastic
features observable in indentation experiments. Further more extensive studies are
necessary to determine the concentration dependence of the Bioglass, or crosslinker,
affecting the elastic modulus of the surface of BG-ELAC threads. This can be extended
119
with comprehensive multi-disciplinary study of effects of crosslinking ratio or adding
Bioglass on cell response [1, 3]. The rationale for using this method is its ability to separate
the elastic modulus in microscale dimensions and of superficial layer that is sensed by cells
from the elastic modulus of the bulk of substrate.
The developed technique allows the detection of viscoelasticity of the surface layer
of the material. This opens an opportunity to determine to what degree viscoelastic
properties (in contrast to the purely elastic) of the surface layer affect cell differentiation
and growth. However, to fully explore this idea the data analysis should be extended to
take viscoelastic properties into account explicitly. To start, it can be performed by
adopting viscoelastic models that do not take adhesion into account [21-23], and then
utilize much more complex models that includes both adhesion and viscoelasticity [24-26].
120
4.6 References
[1] Madhura P. Nijsure, Meet Pastakia, Joseph Spano, Michael B. Fenn, Vipuil Kishore
Bioglass incorporation improves mechanical properties and enhances cell-mediated
mineralization on electrochemically aligned collagen threads, Journal of Biomedical
Materials Research Part A, 2017, 105A, 2429-2440.
[2] Kohn JC1, Ebenstein DM., Eliminating adhesion errors in nanoindentation of compliant
polymers and hydrogels, Journal of the Mechanical Behavior of Biomedical Materials
2013, 20, 316-326.
[3] Kishore, V.; Iyer, R.; Frandsen, A.; Nguyen, T-U., In vitro characterization of
electrochemically compacted collagen matrices for corneal applications, Biomedical
Materials 2016, 11, 055008.
[4] Nama, S.; Hub, K. H.; Buttec, M. J.; Chaudhuria, O. Strain-enhanced stress relaxation
impacts nonlinear elasticity in collagen gels, Proceedings of the National Academy of
Sciences 2016, 113, 5492-5497.
[5] Kilpatrick, J. I.; Revenko, I.; Rodriguez, B. J., Nanomechanics of Cells and
Biomaterials Studied by Atomic Force Microscopy, Advanced Healthcare Materials 2015,
4, 2456-2474.
[6] Han, B.; Nia, H. T.; Wang, C.; Chandrasekaran, P.; Li, Q.; Chery, D. R.; Li, H.;
Grodzinsky, A. J.; Han, L., AFM-Nanomechanical Test: An Interdisciplinary Tool That
Links the Understanding of Cartilage and Meniscus Biomechanics, Osteoarthritis
121
Degeneration, and Tissue Engineering, ACS Biomaterials Science & Engineering 2017, 3,
2033-2049.
[7] Guo, S.; Akhremitchev, B. B., Investigation of Mechanical Properties of Insulin
Crystals by Atomic Force Microscopy, Langmuir 2008, 24, 880-887.
[8] Vanlandingham, M. R., et al., Nanoindentation of polymers: An overview,
Macromolecule 2001, 167, 15-43.
[9] Dagastine, R. R.; et al., Dynamic forces between two deformable oil droplets in water,
Science 2006, 313, 210-213.
[10] Ahn, D.; Shull, K. R., JKR studies of acrylic elastomer adhesion to glassy polymer
substrates, Macromolecules 1996, 29, 4381-4390.
[11] Efremov, Y. M.; et al., Measuring nanoscale viscoelastic parameters of cells directly
from AFM force-displacement curves, Scientific Reports 2017, 7, 1541.
[12] Attard, P., Measurement and interpretation of elastic and viscoelastic properties with
the atomic force microscope, Journal of Physics: Condensed Matter 2007, 19, 473201.
[13] Recco, A. A. C.; et al., Energy dissipation in depth-sensing indentation as a
characteristic of the nanoscratch behavior of coatings, Wear 2009, 267, 1146-1152.
122
[14] Wenger, P. E.; et al., Mechanical properties of collagen fibrils, Biophysical Journal
2007, 93, 1255-1263.
[15] Moeller, G., AFM nanoindentation of viscoelastic materials with large end-radius
probes, Journal of Polymer Science Part B: Polymer Physics 2009, 47, 1573-1587.
[16] Chandrasekaran, P.; Doyran, B.; Li, Q.; Han, B. A.; Bechtold, T. E.; Koyama, E.; Lu,
X. L.; Han, L., Biomechanical properties of murine TMJ articular disc and condyle
cartilage via AFM-nanoindentation, Journal of Biomechanics 2017, 60, 134-141.
[17] Benguigui, L.; Boue, F., Homogeneous and inhomogenous polyacrylamide gels as
observed by small Angle neutron scattering: A connection with elastic properties,
European Physical Journal B 1999, 11, 439-444.
[18] Sun,Y.; Akhremitchev, B. B.; Walker, G. C., Using the Adhesive interaction between
atomic force microscopy tips and polymer surfaces to measure the elastic modulus of
compliant samples, Langmuir 2004, 20, 5837-5845.
[19] Strasser, S.; et al., Structural investigations on native collagen type I fibrils using
AFM, Biochemical and Biophysical Research Communications 2007, 354, 27-32.
[20] Di Mundo, R.; Recchia, G.; Parekh, M.; Ruzza, A.; Ferrari, S.; Carbone, G., Sensing
inhomogeneous mechanical properties of human corneal Descemet's membrane with AFM
nano-indentation, Journal of the Mechanical Behavior of Biomedical Materials 2017, 74,
21-27.
123
[21] Wei, F. N.; Yang, H. T.; Liu, L. Q.; Li, G. Y., A novel approach for extracting
viscoelastic parameters of living cells through combination of inverse finite element
simulation and atomic force microscopy, Computer Methods in Biomechanics and
Biomedical Engineering 2017, 20, 373-384.
[22] Akhremitchev, B. B.; Walker, G. C., Finite sample thickness effects on elasticity
determination using atomic force microscopy, Langmuir 1999, 15, 5630-5634
[23] Woodhead, A. L.; de Souza, M. L.; Church, J. S., An investigation into the surface
heterogeneity of nitric acid oxidized carbon fiber, Applied Surface Science 2017, 401, 79-
88.
[24] Andrews, J. W.; Bowen, J.; Cheneler, D., Optimised determination of viscoelastic
properties using compliant measurement systems, Soft Matter 2013, 9, 5581-5593.
[25] Chyasnavichyus, M.; Young, S. L.; Tsukruk, V. V., Probing of polymer surfaces in
the viscoelastic regime, Langmuir 2014, 30, 10566-10582.
[26] Haiat, G.; Phan Huy, M. C.; Barthel, E., The adhesive contact of viscoelastic spheres,
Journal of the Mechanics and Physics of Solids 2003, 51, 69-99.