Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Advances in Colloid and Interface Science 120 (2006) 57 – 67

www.elsevier.com/locate/cis

Nanomechanical measurements with AFM in the elastic limit


John R. Withers, D. Eric Aston ⁎
Department of Chemical Engineering, P.O. Box 441021, University of Idaho, Moscow, Idaho 83844-1021, United States
Available online 18 May 2006

Abstract

With increasing interest in nanoscience and nanotechnology, the fundamental underpinnings of what makes materials strong and durable are
under critical investigation. Recent findings suggest that when materials are reduced in extent to nanoscopic proportions, they exhibit enhanced
strength, specifically in the form of higher moduli than are measured on macroscopic objects of the same composition. Force-deformation
behavior of nanostructures subjected to concentrated loads, such as with atomic force microscopy (AFM), can yield detailed information and
insight about their local mechanical properties. We review and evaluate the effectiveness of deformation and indentation tests used in determining
the elastic modulus of nanobeams, nanosprings, thin films, biological samples, dendrimers, and fluid droplets. Obstacles yet remain in the
determination of absolute, quantitative modulus data at the nanoscale. In spite of basic limitations, recent developments in advanced
nanomechanical techniques will facilitate improvement in our understanding of material strength and aging from molecules and colloids to the
macroscale.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Atomic force microscopy; Nanomechanics; Elastic modulus; Nanoindentation; Three-point bend test

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
2. Atomic force microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3. Bend tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1. Nanowires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.2. Nanosprings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4. Nanoindentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.1. Living cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2. Dendrimers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5. Liquid interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
6. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

1. Introduction International trends in nanoscience funding indicate the growing


popularity of his vision and the push toward nanotechnology [4–
In his classic 1959 presentation “There's Plenty of Room at the 6]. We yet understand little of the behavior of matter at the
Bottom,” Richard Feynman speculated that we would observe a nanoscale, both in general and in specific cases, even of
multitude of new material properties in substances if we could mechanical properties that are well understood at the macroscale
arrange atoms individually the way we want them [1–3]. [7–9]. Something as simple as the bending of an elastic beam or
plate must be directly confirmed or recharacterized at the
⁎ Corresponding author. Tel.: +1 208 885 6793; fax: +1 208 885 7462. nanoscale due to the basic issue of larger surface-to-volume
E-mail address: aston@uidaho.edu (D.E. Aston). ratios of these objects, where the interfacial material of the system
0001-8686/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.cis.2006.03.002
58 J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67

under study can have a significant effect on properties and ly, a diode laser reflects off the back of the cantilever onto a
behaviors other than classically defined. quadrant photodetector, which senses cantilever bending and
Traditional optical microscopy lacks the resolution to in- twisting. If the cantilever spring constant is known, the can-
vestigate phenomena of colloidal dimensions adequately, and tilever deflections may be converted to quantitative force data.
electron and X-ray techniques are greatly limited either by Cantilever calibration is not a trivial procedure, however. Can-
environmental (e.g., liquids) or material property (e.g., con- tilevers are sold with typical force constants that may, in fact,
ductivity, cross-section for energy beam interactions) restric- vary by an order of magnitude from reported average values.
tions. Today, very few electron microscopes are capable of the Alternately, the values can be calculated from the cantilever's
true atomic resolution [10] required for fundamental studies on spring geometry [3,12,19,21,22]. For a uniform, rectangular
intermolecular and colloidal behavior of two- or three-body cross-section, the cantilever's spring constant is given by
interactions, for example. But scanning electron microscopes kc = Ewt3/4l3, where w is the cantilever's width, l is its length,
(SEMs) have played a crucial role in the study of mechanics t is its thickness, and E is the elastic modulus. This, however, is
with nanowires and nanosprings, and high-resolution optical seldom accurate enough for the required precision of quanti-
microscopy is useful for locating these so-called one-dimen- tative AFM studies. Most cantilever probes are rectangular or
sional objects on test substrates. In many cases, this is feasible triangular with a “two-beam” geometry connecting at the tip.
because of lengths exceeding a few microns that scatter enough Many AFM cantilevers are also coated with one or more layers
light for adequate contrast. The advent of atomic force micros- of metal for reflectivity and other surface modifications.
copy (AFM) [11–13] marked the beginning of significant ad- From the deflection of the cantilever, we calculate tip-sample
vancement toward more routine molecular-scale imaging in force data using Hooke's law:
three quantified dimensions with the simultaneous measure-
ment of an additional one or more physical properties. While the F ¼ kc z; ð1Þ
literature shows a nearly exponential increase in studies of
paired interactions and nanomaterials, our ability to construct where F is the magnitude of the force acting between the tip and
“from the bottom up” requires a still deeper understanding and sample, kc is the cantilever spring constant, and z is the can-
further advances in many disciplines for Feynman's visions to tilever deflection at its free end.
become reality. AFM may be known best for its ability to generate high-
Researchers employ the AFM in various ways to determine resolution topographical images. In most imaging modes, a
sample mechanical properties [14–17], especially the elastic, or feedback system senses instantaneous cantilever deflection and
Young's, modulus. These generally include nanodeformation adjusts scanner elements to maintain a constant interaction
tests, such as three-point bend tests and the more common between the tip and sample. The instrument records and plots
nanoindentation. We examine the application of these methods scanner adjustments as surface topography. In fact, the AFM
in the mechanical study of nanostructures, evaluating their may operate in any of various modes, depending on the inter-
utility and effectiveness in estimating elastic moduli and related action energies or forces of interest. For example, selective
mechanics. chemical functional groups may be attached to the probe,
generating force data reflecting sample composition. The image
2. Atomic force microscopy produced will thus be a convolution map of chemical makeup,
not merely surface topography [23,24]. Also, the AFM can
Atomic force microscopes are themselves nanomechanical record twisting movements of the cantilever, which represent
instruments. AFM employs a sharp, cantilever-mounted probe to frictional forces acting between the tip and sample [25].
raster scan surfaces. Image resolution can be very high–scientists Force spectroscopy generates a force–distance curve for a
have observed subatomic-scale features [18]–but depends on a single location on the sample. This is a plot of the magnitude of
variety of factors including tip sharpness, acoustic isolation of the the force acting between tip and sample versus the position of
instrument, sampling medium, AFM controller precision, etc. The the scanner in the direction normal to the substrate. Force–
tip and sample positions are manipulated relative to each other distance curves hold a wealth of information about the sample's
with piezoelectric or other (e.g., electromagnetic coils) actuators. mechanical properties. Points of discontinuity, the slopes of the
The AFM precisely controls the tip location on the sample by approach and retract curves, as well as any observed hysteresis
managing the voltage applied to the scanners [3,12]. These are all cede hints to surface behavior [12,26–29]. The difficulty
arranged either with three independent, orthogonal piezoelectric arises in interpretation and deconvolution of multiple phenom-
blocks or in a tube configuration. Piezoelectric scanner perfor- ena. Hysteresis, for example, is the result of adhesion, surface
mance can be limited due to nonlinearities in the scanner, material deformation, and/or nonlinear performance of the instrument,
creep, noise and drift in the high voltage supply, or thermal drift of such as piezoelectric or other transducers for scanning and the
the AFM apparatus itself [3,12,19]. probe detection sensor, e.g., photodiode.
A variety of attractive and repulsive forces act between the Another consideration is that the AFM does not directly
tip and sample [3,12,20], such as van der Waals, electrostatic, measure the actual tip-sample separation distance. Rather, it
and capillary forces. To some extent, such forces can be con- controls and/or measures the vertical scanner position, the can-
trolled by altering the sampling medium; for example, sampling tilever deflection, and the sample deformation [28]. We often
under water can eliminate the effect of capillary forces. Typical- collect an array of force–distance curves at discrete sites over an
J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67 59

area of the sample surface to determine effects of sample of varying thickness were bound to copper grids, resulting in
heterogeneity. This enables the AFM to produce spatially re- adhesive, edge-clamped, free-standing films over each grid
solved maps of both topography and other sample material square. The films subjected to point loading with AFM were
properties near the surface that are gleaned from force or energy modeled as beams but are closer to rectangular plates, even though
profiles. Commercially available tools, such as pulsed-force significant stretching of the polymeric (membrane) material may
mode (PFM) [30–32] and force-volume imaging [33], accom- be expected to skew the results significantly from classic behavior.
plish this mapping with improved automation. In these cases, Local deformations in AFM bend tests cannot be neglected as in
force, adhesion, and stiffness data are readily collected and some classic experiments, but a satisfactory solution for nano-
made available for offline interpretation and analysis. Other mechanics is yet to be developed. We expect to observe discrep-
physical interactions between the probe and sample may also be ancies between actual deformations and predicted values based
mapped with more difficulty, such as energy loss or long-range only on global bending model predictions at this stage in devel-
forces. It is primarily the stiffness data that are of interest to opment for both suspended films and nanowires [17].
nanomechanical studies.
3.1. Nanowires
3. Bend tests
Researchers employ various deformation tests to study the
Nanowires and nanosprings have been highly studied for moduli of elasticity of nanowires. In general, the expressions used
their sometimes remarkable electrical and mechanical properties to relate elastic modulus to observed deflection and applied force
[15,34–39]. Some nanowires exhibit surprisingly high tensile come from classic texts on strength of macroscopic materials. This
strength, possibly due to the presence of fewer mechanical defects approach is valid when the beams follow linear elastic theory of
per unit length than in their macroscopic analog [39]. Being able isotropic materials and have high length-to-thickness ratios [8,34].
to synthesize these structures reliably and understand their A beam can be supported in various ways resulting in mathe-
behavior will be vital to their ultimate application as future matical description via boundary conditions of the type of sup-
building blocks in everything from circuits to microelectrome- ported ends. The two extremes in behavior are set by having both
chanical or nanoelectromechanical systems (MEMS or NEMS), ends either free or built-in (that is, clamped or fixed) to resist both
environmental sensors to biomimetic implants, hydrogen pro- torque and slip. The first is the simply supported beam (Fig. 1a)
duction and storage to smart composite materials, just to name the and the latter is the clamped end configuration (Fig. 1b). Other
more popular examples [4–6]. beam bending behavior may be described by mixed boundary
Deformation tests to determine the mechanical properties of conditions, or by modifications thereof, for example, a limited-slip
edge-supported films are similar to bending nanowires and are or limited-torque end that might be related to adhesion and
plagued with similar difficulties of experimental setup and data tribology.
interpretation at the nanoscale. While a significant literature The general bending equation expresses beam deflection, z,
base exists on bulge tests [40], the “two-dimensional” analog of in the plane of applied stress as a function of the beam's
beam bending, there is, surprisingly, little in the way of true moment, M [7]:
nanomechanical investigations with AFM [41], nanoindentation d2 z
[42], or other techniques, that is, nanoscopic in both morpho- EI ¼ ¼ −M : ð2Þ
dx2
logy and stress, e.g., nanoNewton point loads. Edge-supported
film deflections may be interpreted from expressions derived for The first moment of inertia, I, depends on the beam's cross-
the classic centrally loaded plate deformations [9] or from sectional shape and physical dimensions. (If the beam shape
extended models for membranes and shape memory materials changes during measurement as may be expected for nano-
[42]. Yang [41] also described research in which polymer films mechanical studies, the in situ determination of I may be

Fig. 1. Schematic representation of beam support models: (a) simple and (b) clamped.
60 J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67

intractable for nanowires–or even unsolvable under AFM test segment of constant compliance. Young's modulus can then be
designs–and will at worst need to be considered a second computed from Eq. (7) with kw and the geometry of the nano-
variable parameter for interpretive analysis [17]; this will be a wire setup (from AFM and/or SEM measurements).
future topic for expansion.) The moment M depends on the We find that for beams with identical material properties and
magnitude of the concentrated point load, F, the load location physical dimensions, where the only difference is the support
measured from the origin down the long axis (i.e., the x- scheme, the midpoint deflection for the simply supported beam
coordinate), a, and the length of the beam, L. M is negative will be larger by a factor of four, as shown in Fig. 2 [7]. This
when the z-axis is defined as positive in the downward idealized scenario is of course complicated for systems where
direction, convenient for AFM bending studies. In the case of the supporting substrate contributes to the overall compliance
built-in or clamped ends, M is expressed as being measured by the AFM experiments. That is, there will be
multiple deformations convoluted into the process if the sup-
Fb Fab2 Fa2 b ports for the nanowire ends are not themselves rigid or if the
M¼ x− 3 ð L−xÞ− 3 x; x V a; ð3Þ
L L L supported portions of the nanowire itself distort during the test.
and We are currently investigating models for “springy” and sliding
supported ends that allow these additional deformations to be
Fb Fab2 Fa2 b
M¼ x−F ð x−aÞ− 3 ð L−xÞ− 3 x; xza: ð4Þ fitted; however, under these scenarios with multiple unknown
L L L parameters, the goal of quantifying nanowire elasticity with
Substitution of Eq. (3) or Eq. (4) into Eq. (2) followed by certainty from a single type of AFM experiment would again be
double integration with respect to x provides the standard forms indeterminate, as with the above case of changing cross section.
for the bent beam profiles as a function of applied load, re- We have used AFM to study polymer nanowires suspended
presentative of AFM force-distance profiling methods: across a ridged substrate [17]. We applied a force at discrete
points down the length of the suspended nanowire sections and
Fb2 x3 Fab2 x2 used Eqs. (5)–(7) to calculate effective elastic modulus. Local
EIz ¼ − ð2a þ L Þ þ ; x V a; ð5Þ
6L3 2L2 indentation was considered to be additive to the bending mode
and was merely subtracted as an offset as a first-order approxi-
Fb2 x3 Fab2 x2 mation; computational work is on-going to determine the ac-
EIz ¼− ð 2a þ L Þ þ curacy of this assumption. The deflection profile generated
6L3 2L2 ð6Þ
F 3 during the measurements seem to validate the assumption of
þ x−aÞ ; xza: fixed ends; that is, the shape of the bending nanowire is better
6
fitted with clamped ends than with simply supported ends. Wu
Relationships of these and similar forms can be used to et al. [16] have extended this method into the study of yield
calculate Young's modulus directly, assuming elastic behavior strength and nanowire plasticity of gold nanowires. Our own
for classic mechanical comparisons. For nanomechanics, values investigations on silver nanowires have shown only elastic
calculated for E are currently interpreted as “apparent” elasticity behavior.
since we have yet to investigate nanoscale bending in full depth. Wong et al. [39] conducted experiments with only one fixed
Boundary conditions are determined by the effect of the support end, while the balance of the wire rested on the substrate, free to
scheme on the shape of the profile constructed from deflection slide in a sweeping lateral motion (Fig. 3). Scanning the wire
data versus location on the nanowire [17]. with AFM in contact mode, they recorded lateral forces at
Several papers [16,34,37] include the following equation,
which is valid at the midpoint for a mid-beam point load.
Letting a = b = x = L/2 yields E = FL3/192zI. Alternately, we can
substitute a known spring constant of a beam, e.g., nanowire,
and calculate its elasticity directly:

kw L3
E¼ : ð7Þ
192I

The slope of an AFM force–distance curve taken against a


surface is the observable spring constant of the mechanical
system, ksys, composed of the cantilever probe being used and
the material under test, which is ideally characterized by the
cantilever and the (nanowire) sample being springs in series:
1 1 1
¼ þ :
ksys kc kw

We obtain kc from a calibration cycle [21,22,43,44], and we Fig. 2. Comparison of deflection profiles for simply supported beam (solid) and
obtain ksys from the slope of the force–distance curve over a beam with clamped or fixed ends (dashed). Units are normalized.
J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67 61

(Δ1sinθ1 + Δ2sinθ2), where F is the applied load, k is the


nanospring constant, Δ1 and Δ2 are the elongations of the spring
segments to either side of the applied load beyond their relaxed
states, and θ is the angular displacement of the spring segments,
as shown in Fig. 4.
If we let the tip act at the spring's midpoint, employ appro-
priate trigonometric identities, and assume uniform geometries
and material properties for the length of the spring, the equation
may be written explicitly as a function of vertical displacement
only:
0 1
B L=2 C
Fig. 3. Lateral AFM scanning, overhead view of single fixed-end nanowire scan. F ¼ 2k z@1− qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiA; ð8Þ
2
ðL=2Þ þ z 2

various positions along the length of the beam. Their data


indicated that the wire returned to its original position after its where L is the linear distance separating the spring's two
release by the AFM tip, suggesting that the deflection was anchored ends. The bending spring constant, k, can be ex-
elastic and the effects of wire-surface friction were not ob- pressed in terms of the physical geometry and properties of the
served. Lateral forces increased linearly once the tip made spring material:
contact with the nanowire, following Eq. (1) to yield a can-
Gd 4
tilever-like spring constant. Further, the slopes of the lateral k¼ ; ð9Þ
force–distance curves, k(x), decreased for scans recorded at 8D3 N
increasing distances, x, from the fixed end. The k(x) versus where d is the wire diameter, D is the nanospring coil diameter,
x curve qualitatively agreed with the expression k(x) = 3EI/x3 N is the number of involved nanospring coils, and G is the shear
[39]. An effective elastic modulus can be determined from any modulus of the material ideally related to elastic modulus [47].
force curve, but Wong et al. considered them in toto for a more Substituting Eq. (9) into Eq. (8) yields the following result:
general calculation. 0 1
Gd 4 B L=2 C
3.2. Nanosprings F¼ @1− qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiAz: ð10Þ
4D3 N 2
ðL=2Þ þ z2
In 2001, researchers at the University of Idaho synthesized the
first coiled nanosprings [45] with both coil and wire diameters of All values may be observed with AFM and/or SEM with the
the order of tens of nanometers. Currently, AFM mechanical exception of the shear modulus, which may be estimated from
investigations of silicon oxide nanosprings are underway; while the ideal relationship E = 2(1 + ν)G, where ν is Poisson's ratio
the measurements have become almost routine, their interpreta- for the spring material. For transverse loading on the spring, the
tion with respect to boundary condition uncertainty remains a relationship between F and z is not expected to be linear. Under
difficult issue. Various other one-dimensional objects similar to load within the elastic limit, the nanospring wire is twisted rather
these nanosprings have come under study. Silicon springs of than stretched axially in the more usual testing geometry for a
much larger pitch and coil diameter were reported to exhibit an coiled spring. It is not subjected to the same effective loading in
electromechanical response to conducting mode AFM [15]; these the way its nanowire analog would be in the three-point bend test.
long-axis measurements are prone to simultaneous compression Another complication is the determination of Poisson's ratio.
and bending, which can lead to indeterminate conclusions about Values are tabulated for common bulk materials, but may in fact
material properties. Chen et al. [37] describe a method in which a
nanocoil is clamped between two cantilevered AFM tips–one
with a very compliant cantilever, another of an order of magnitude
higher–by electron beam-induced deposited (EBID) residual
hydrocarbons in a SEM environment. In this case, the force
applied to the spring is determined from the observed deflection
of the more compliant cantilever, and the nanocoil elongation is
determined, both using SEM. A spring constant for the coil is then
found by dividing the applied load by the total elongation, as in
Eq. (1). They assumed that the hydrocarbon “glue” does not
deform, viz., fixed ends.
Bend tests may be conducted for nanosprings, as with the
nanowires, to find Young's modulus. As long as the AFM tip is
not located at either end of the spring, the force applied can be
expressed with simple geometric relationships [46]: F = k Fig. 4. Lateral view of nanospring AFM bend test.
62 J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67

the dimensions of each body, that each body can be approximated


as a semi-infinite half-space (i.e., no bulk deflections), and that
deformations occur within the elastic limit. In 1971, Johnson et al.
[65] proposed corrections to this model to account for adhesive
forces between the bodies in what is now called the JKR theory.
Still, most nanoindentation methods require corrections to these
idealized models. Further corrections on the initial contact regime
are also available, having more pertinance in adhesion studies
rather than indentation [20,66,67]. The Oliver and Pharr technique
is commonly used after initial indenter contact [57,58,68–71],
extending from prior research conducted by Sneddon [64] and
others [72–74].
Nanoindentation collects data from a complete loading and
unloading cycle to generate a particular force–distance curve
(Fig. 5), very similar to an AFM force profile. Force versus
penetration depth (or displacement) is desired instead of the force
versus separation of typical AFM force spectroscopy or nanowire
bend tests. In the case of a purely elastic deformation, there will
be no hysteresis between the loading and unloading curves. Con-
sideration of plasticity or viscoelasticity is a much more complex
problem; much of our understanding in this area is the result of
Fig. 5. Sample nanoindentation force–displacement curve for elastoplastic
experimentation and finite element simulation [63].
loading and unloading.
Oliver and Pharr consider the cantilever-surface system as a
two-spring system, as above. The total measured displacement
vary over a sample surface and are completely unknown when is a combination of the load frame (AFM cantilever) dis-
working with new materials, adding another dimension of un- placement, hl, and that of the sample, hs: htot = hl + hs. We con-
certainty to the mechanics of nanosprings. vert this expression to compliances by dividing the force, F,
giving effectively the same relationship as that for the spring
4. Nanoindentation constants in series described above for bending: Ctot = Cl + Cs,
where compliance is simply the inverse of the stiffness, S (C = 1/S).
Nanomaterials find increasing use in modern technologies This is analogous to bending but characteristically different in the
such as thermal or protective coatings, dielectrics, nonlinear actual process of indenting. Cs is given by
optics [48], data storage technologies [49], etc. Mechanical pffiffiffi
properties–such as high elasticity–are indicators of successful p 1
Cs ¼ pffiffiffi : ð11Þ
long-term material performance, making it important to be able 2Er A
to determine these characteristics quickly and reliably for many
A is the contact area, and Er is the effective, or reduced, Young's
different length scales. One of the most common and direct ways
modulus of the system given by
of probing local mechanical properties of films or other extended
surfaces is by nanoindentation [48,49], which is most common for 1 1−m2s 1−m2i
objects that are not suspended. This method employs a sharp, rigid ¼ þ : ð12Þ
Er Es Ei
tip to probe a sample, recording applied force and penetration
depth. AFM can be used as a nanoindenter, but specialized depth- Eq. (11) holds for any indenter that can be described as a
sensing instruments are also commercially available; some re- body of revolution of a smooth function, and it is not limited to a
search groups construct their own [49]. At present, the most specific geometry. Oliver and Pharr suggest an iterative scheme
common techniques for measuring elasticity with high spatial to determine Cl and A, where an initial estimate for a perfect
resolution are indentation [50–52] and various force modulating
methods with AFM [17,53–56].
A host of nanoindentation methods are available, employing
various tip shapes and relying on different model assumptions
[57–62]. While most, if not all, models were developed in some
fashion from the Hertz contact problem in the late 19th century
[3,8,63], currently applied methods and advanced developments
for indentation modeling are more directly connected to Sneddon's
“major contribution” in this area [63,64]. Hertz analyzed the
problem of the elastic contact between two spherical surfaces with
different radii and elastic moduli. His model assumes elliptical
contact area, frictionless contact, small contact area compared to Fig. 6. Nanoindentation schematic of indenter tip and surface.
J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67 63

Berkovich tip is A ≅ 24.5hc2: hc is the contact depth (Fig. 6), sured values of Er requires that Poisson's ratio be known. As we
given by hc = hmax − hs; hmax is measured at the peak load, and hs will discuss later, when the ratios are not known, we can map
can be determined from the force–penetration data, hs = εFmax/S; only relative material property differences for the sample. Also,
ε varies with indenter geometry, as in Table 1. The factor of 24.5 the validity of this approach depends on there being an ideal
is not accurate for other geometries, and especially for AFM indenter geometry or, in the worse case, precise modeling of the
indentation, the individual tip in use must be independently real tip shape, which is computationally intensive.
characterized with electron micrographs or one of the standard tip Indentation testing is very common in the determination of
imaging deconvolution techniques [19,75–80]. Commercial elastic modulus and hardness of supported thin films
AFM probes now come with a variety of comparably sharper [49,51,68,71,81,82], which introduces the added complication
tips with high aspect ratios such that the contact mechanics may of substrate influences [81] due to finite sample thickness
be substantially different from classic indentation experiments. In [54,56] and/or the presence of multiple layers [83].
any event, using higher-order fits for the area–indentation rela-
tionship (A ∼ hc2) is theoretically dubious and yet may provide 4.1. Living cells
some practical improvement in data correlation.
Initial estimates of Cl and Er are obtained by plotting Ctot Extensive progress has been made over the years in under-
versus A1/2 for several indentations made in a standard reference standing the surface properties of microorganisms using a variety
sample, such as aluminum. Revised calculations of A will alter of methods [84], such as micropipette aspiration, dynamic
estimates of Cl and Er, so the process is repeated until con- reflection interference contrast spectroscopy, scanning acoustic
vergence. An alternate method of finding A is to image the microscopy, infrared laser traps, and various magnetometric anal-
plastic impression with AFM [70]. For a highly plastic reference yses. Most of these methods average properties over compara-
material, this should give about the same projected area as the tively large areas with only modest spatial resolution.
contact area at maximum load. Again, plotting Ctot against A1/2 It is relatively straightforward to create topographic images of
yields Cl and Er from the slope and y-intercept of a linear fit. a cell sample using AFM [85,86]. Determination of its mechanical
Once this quasi-calibration method is complete, we observe properties can be accomplished using a nanoindentation process
the force-penetration curves for our samples of interest. The like that discussed above, even under actual physiological con-
unloading portion of the curves is fitted to a power law ex- ditions [85,87,88]. Force curves are recorded and converted into
pression: F = α(h − hf)m, where the parameters α, hf, and m are force-indentation plots using appropriate models. These profiles
all determined by a least squares fitting procedure of the curve; can then be analyzed to give quantitative information about the
hf is the final unloading depth; m is related to the geometry of surface, such as E. Nanoindentation has been accomplished on
the indenter; α accounts for geometric constants, the sample animal cells including glial cells, platelets, cardiomyocytes, ma-
elastic modulus, the sample's Poisson ratio, the indenter elastic crophages, endothelial cells, epithelial cells, fibroblasts, osteo-
modulus and the indenter's Poisson ratio [57,60]. The stiffness blasts, and more [33,87,89–95]. AFM force measurements have
is found by analytically differentiating this expression and also been applied to microbial specimens [96], resulting in
evaluating the derivative at the peak load and displacement, spatially well-resolved maps of mechanical properties [33,97].
Fmax and hmax: Understanding biomechanical mechanisms may one day be key in
strengthening bone, growing cartilage, improving cardiac con-

dF
dh
jhmax ;Fmax
¼ amðhmax −hf Þm−1 :
tractility, and constructing tissues for artificial organs [98].
Hertzian-based models are still commonly used in the
evaluation of elastic modulus [84,88,94,97,99]. Several rela-
Doerner and Nix [74] took their initial unloading slope to be tionships have been published for applied force vs. tip
linear, assuming the contact area remained constant. Oliver and penetration depth, varying again with indenter shapes. For ex-
Pharr [63] demonstrated that the unloading slope is never linear ample [94,97], the conical and paraboloidal tips would be of the
in practice; the contact area changes immediately and contin- following forms:
uously during unloading. The Oliver and Pharr method predicts 2
actual behavior reasonably well, within 4% of values reported in Fcone ¼ tanaEr d2 ;
p
the literature for materials with isotropic elastic properties [70].
There are some difficult points to consider in the use of this 4
model. Error is more pronounced in materials that show a large Fparaboloid ¼ Er R1=2 d3=2 :
3
elastic recovery [58]. To compute the modulus from the mea-
In the case of soft samples and sufficiently rigid probes, Eq.
(12) reduces to
Table 1
Typical values for ε and m
Es
Indenter geometry ε m Er ¼ :
1−ʋ2s
Flat-end cylinder 1 1
Parabaloid 0.75 1.5 These models have been demonstrated to be adequate in
Cone 0.73 2
many applications even though cells do not meet the model
64 J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67

assumptions. In particular, living cells are anisotropic and will Hyperbranched polymers are often more rapidly prepared and
not necessarily behave elastically without a viscous component, more economical than ideal dendrimers for certain applications.
or loss modulus [95]. Hertzian models are not appropriate where The high symmetry of perfectly branched dendrimers, their
adhesive forces are present. Another difficulty is that Poisson's cascade synthesis method, and the lack of a molar mass dis-
ratio is not measured in the indentation experiment. In practice, persity (assuming ideal growth) show a closer similarity to
it is often arbitrarily assigned the constant value of 0.5 (the case biomolecules like DNA and polypeptides rather than classical
for incompressible bodies), and we would expect sensitive polymers [101].
stiffness measurements across the surface of a cell to resolve Study of hyperbranched molecules has increased in recent
real property variations in local deformation because of inter- years, with about 2000 publications in 2004 [102]. The field is
cellular heterogeneities. too new to draw conclusions about the ultimate usefulness of
Other sources of error arise from characteristics of the AFM, dendritic polymers and far too broad to review current efforts
which are issues in any quantitative experiment. Choice of can- completely here. It is widely believed that they offer the chance
tilever spring constant and accurate calibration thereof (see for the development of new products, but their complexity and
Section 2) are primary examples. If a sample is soft relative to broad variety of structures present a monumental challenge.
the sensitivity of the AFM, the sample will deform in response to It is thought that certain functional groups can be placed at
the tip before the cantilever deflects measurably and the true the sterically congested core of dendritic polymers, giving them
contact point will not be detected. Large uncertainty in the contact highly selective reactivity [103,104], although lately, the inter-
point leads to significant errors in the estimation of the indentation est seems to have turned more to studies of surface functional
depth, elastic moduli, or any other parameter defining the contact group behaviors [105–108]. Dendritic polymers may also find
behavior [33,84,93]. The soft cellular matter requires an equally application in medicine, catalysis [104], or nanoelectromecha-
soft cantilever. The wide range of AFM tip shapes and the small tip- nical systems (NEMS) [109]. AFM has been used to create
sample contact area make the contact geometry difficult to define. topographical images of dendritic polymers [110], and AFM
Dimitriadis et al. [99] write that the error introduced into the chemical force modes have allowed the study of reactivity at the
calculated elastic modulus by the aggregate of tenuous assump- peripheral functional groups including branch termini [24,111].
tions can be an order of magnitude. They derive corrections to the Researchers at Iowa State University [112] pursued mechan-
Hertz model that cover the full range of contact force, tip radius, ical studies of dendrimers similar to those used in the study of
and sample thickness. In many instances–even in materials other living cells, using Hertzian contact models. Force–distance data
than cells–variations of stiffness are more interesting than quan- showed that compression behavior varied under low and high
tifying elasticity, as indicated by Heinz and Hoh [33,87]. Studies normal loads with large initial deformation followed by a stiffer
of spatial and temporal changes in mechanical properties on a response at higher applied forces. This is consistent with
relative scale can lead to many insights without a calibrated behavior of a compliant sample on a rigid substrate. It appeared
standard for Young's modulus. In that case, these significant that higher-generation dendritic molecules possessed a much
errors in absolute property measurements are no longer important. higher elastic modulus, due to higher internal molecular stiff-
ness and a crowded periphery. As for cells, indentation studies
4.2. Dendrimers of dendrimers and the like are more complicated than nano-
indentation on flat films and other extended surfaces.
The first hyperbranched molecule was patented in 1987
[100]. Its architecture immediately interested synthetic che- 5. Liquid interfaces
mists, but it took some time to attract the mainstream, possibly
because of difficulty in characterization due to irrational solu- Liquid interfaces of gas bubbles and droplets demonstrate
bility and irreproducible molecular weight. Most of the research some resistance to deformation under an applied load, exper-
effort has come from industry. imentally similar to elasticity measurements. As such, we
Hyperbranched molecules manifest unique properties as a observe effective spring behavior in the fluid interfaces, which
result of their intrinsic globular structure and large number of may be deconvoluted to quantify various interfacial phenomena
terminal functional groups. Unlike dendrimers, which are “per- such as colloidal interactions or local surface tension. Much
fectly” branched, hyperbranched polymers have elements of effort has been expended in the attempt to characterize this
conventional polymers, namely molecular weight polydispersity, behavior, with some discussion as to the suitability of Hookean
isomerism, and geometrical shapes. A normalized branching spring models. Studies have examined the interaction of
factor was introduced to describe the degree of perfection of the bubbles/droplets with colloid particles [29,112–119] as well
branching [100,101]: fbr = (T + B)/No, where T is the mole fraction as the interaction between two liquid particles [120]. Unlike
of monomers at the terminal position, L is the number of un- solid surfaces, interaction models for fluid interfaces do not
branched monomer units, B is the number of fully branched usually require a Born repulsion term in the net interparticle
monomer units, and No = T + B + L. Thus, the branching factor for force relationship since physical contact with a fluid is a wetting
a fully branched polymer is one, such as a dendrimer, and fbr of a condition [20,114].
linear polymer is zero. To a greater degree even than in studies of cells or polymers,
In practice, dendrimers and hyperbranched polymers are a significant stumbling block to force measurements on highly
usually discussed together using the term dendritic polymers. deformable systems is that there is no clear transition from
J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67 65

noncontact to contact [33,119]. Surface deformation occurs similar to hard wall contact with a solid surface. But keff is
prior to contact due to the extended range of surface forces and different depending on where it is measured; greater deformation
fluid dynamics. Thus, the zero of separation cannot be deter- yields a larger linear spring constant. Both experiment and theory
mined with AFM in the same way as for rigid bodies. point to a nonlinearly deflecting fluid–fluid interface, though
Gillies et al. [119] summarize a variety of ways in which there are certain experimental circumstances under which a linear
researchers have defined zero separation. In some cases, the approximation is very accurate [118].
scanner position is recorded, and all separations are considered Furthermore, this provides a possible explanation for the
relative to an arbitrary zero [121,122]. In others, deformation is apparent transition observed by Hartley et al. from the expo-
modeled as a Hookean spring and contact is considered to be the nential force behavior of overlapping double layers, when the
point at which the linear compliance curve reaches zero force, drop acts like an ideal spring switching to the “simple linear
either upon approach or during the retracting of the probe, relationship” of force with apparent separation. This is not the
which is only valid if the surfaces come into intimate contact same as linearity of the interfacial deformation with load. To
[123–125]. Others take the zero to be where deformed surfaces maintain the analogy, if the oil–water interface is like a spring
in contact exhibit zero force. This occurs due to a balance of van but with an increasing stiffness, then keff may be expressed as a
der Waals attraction and steric (or other) repulsions, and the weak function of drop deflection. It might be useful to fit
amount of deformation can be substantial [126–128]. The zero experimental data with a power function, as is common for
of separation is sometimes taken to be that corresponding to the indentation, to convert F(z) curves into F/R vs. separation: R is
maximum force. This approach also ignores the flattening of the the system radius of curvature, used to normalize data [20];
deformable bodies and can be in serious error [29,114,129,130]. F = kdn, where d is the drop deflection, k is constant, and n is
Contact is sometimes taken to occur at the point of the first slightly increasing as d increases (values of n taken from fitting
measurable force. This is appropriate for contact forces of the results). If the stiffness is indeed equivalent to the interfacial
essentially “zero range” but not for long-range forces such as tension, then the dn term is merely the z-component of a surface
those of the electrostatic double layer. Burnham et al. [131] integral of the interfacial curvature, which has no analytical
defined contact as the position where repulsion is first detect- solution. It remains to be shown that physical conditions, if any,
able, indicated by a change in curvature of the force-distance would lead to this behavior. Empirically, there seems to be a
profile, i.e., a break point. A related method fixes the zero of connection between the onset of the drop dimpling inward and
relative separation at an arbitrarily small force [132], which can the deviation from linear interfacial deflection, that is, the drop
serve to eliminate the uncertainty of defining the onset of apex curvature inversion, from convex to concave. It is likely a
deviation from the baseline of no interaction, defined at large geometric consequence of measuring only the normal force
probe-sample separations. Mulvaney et al. [115] fitted the component of a curved film with increasing interfacial area
nonlinear Poisson-Boltzmann equation for rigid bodies by taking part in the force measurement [29,114,133].
adjusting both the zero of the experimental separation and the
surface potential. Each of these approaches has its limitations; in 6. Concluding remarks
order to have meaning, zero separation or contact must be
interpolated from experimental data in some self-consistent The advent of AFM was of great import to nanomechanics.
manner. In all events, the definitions of the distance or sepa- Phase contrast, pulsed-force and other intermittent contact modes
ration origin are made for experimental and/or modeling con- provide high spatial resolution of surfaces, highlighting inhomo-
venience and are of limited theoretical use. There is yet no way geneities and relative surface property differences. Instrumenta-
to control a fluid–fluid interface directly for all separations of tion and modeling have matured–and are constantly evolving–to
interest. where they offer a useful look into nanoscale mechanical per-
Colloid probe force measurements have demonstrated formance. Now, nanomechanical investigations of molecular
Hookean spring behavior in air bubbles and oil droplets in the changes that precede a material yielding, creeping, crazing,
constant compliance region. Theoretical work [113] has con- cracking, or otherwise failing at a macroscopic level are feasible.
firmed this experimental finding, showing that the experimen- One continuing problem with the determination of absolute
tally applied loads are well within the linear region, and that the force data lies in knowing Poisson's ratio. In general, this is
effective spring constant of a fluid interface, keff , is practically obtained from tabulated data for known, bulk materials, or it is
equal to the interfacial tension [132]. However, this is highly given an assumed value, as in the study of cells. Another problem
dependent on geometry and scaling. It is suspected that this is that the tip-sample contact area is difficult to quantify, which
would only be the case within some limiting regime, like that of can lead to large errors. Still, relative, local surface properties are
small interfacial deflections. easily mapped, and absolute, quantitative data are unnecessary
If the interface were to behave as a Hookean spring, then it for many experiments and applications in any case.
should be possible to convert the force versus total distance The true challenge will be to determine the local effects of
displaced into the true force-separation profile simply by adding making nanomechanical measurements. The authors are cur-
the quantity of force divided by keff: separation = distance + F/keff. rently investigating energy loss in polymer nanowires with
The quantity F/keff is the interfacial displacement normally bending and nanoindentation experiments to determine how
inferred from AFM experiments on air bubbles or oil droplets. AFM alters molecular conformations. Preliminary studies show
The value of keff is the slope of F(z) at high load, considered to be accelerated aging across repeated deformation experiments,
66 J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67

suggesting that a kind of strain hardening or crystallization is [32] Sun C, Aston DE, Berg JC. J Colloid Interface Sci 2002;248:96.
taking place. The fundamental reasons for increased apparent [33] Heinz WF, Hoh JH. Trends Biotechnol 1999;17:143.
[34] Sundararajan S, Bhushan B, Namazu T, Isono Y. Ultramicroscopy
elastic modulus in nanostructures remain unproved. Even 2002;91:111.
though significant obstacles remain, the collection of recent [35] Salvetat J-P, Kulik AJ, Bonard J-M, Briggs GAD, Stockli T, Metenier K,
advancements in nanomechanical measurements of material et al. Adv Mater 1999;11:161.
strength are laying a strong foundation to improve our under- [36] Salvetat-Delmotte J-P, Rubio A. Carbon 2002;40:1729.
[37] Chen X, Zhang S, Dikin DA, Ding W, Ruoff RS, Pan L, et al. Nano Lett
standing of basic material behavior such as beam bending and
2003;3:1299.
plastic deformation. [38] Xie S, Li W, Pan Z, Chang B, Sun L. J Phys Chem Solids 2001;61:1153.
[39] Wong EW, Sheehan PE, Lieber CM. Science 1997;277:1971.
Acknowledgements [40] Jansen KMB, Gonda V, Ernst LJ, Bressers HJL, Zhang GQ. J Electron
Packag 2005;127:530.
The authors gratefully acknowledge partial support for this [41] Yang AC-M. Mater Chem Phys 1995;41:295.
[42] Poilane C, Delobelle P, Lexcellent C, Hayashi S, Tobushi H. Thin Solid
work from the National Science Foundation (EPS0132626, Films 2000;379:156.
EPS0447689) and the W.M. Keck Foundation. [43] Torii A, Sasaki M, Hane K, Okuma S. Meas Sci Technol 1996;7:179.
[44] Jericho SK, Jericho MH. Rev Sci Instrum 2002;73:2483.
References [45] McIlroy DN, Zhang D, Kranov Y, Norton MG. Appl Phys Lett
2001;79:1540.
[1] Feynman R. There's Plenty of Room at the Bottom. California Institute of [46] Zhang H-F, Wang C-M, Buck EC, Wang L-S. Nano Lett 2003;3:577.
Technology; 1959. [47] Higdon A, Ohlsen EH, Stiles WB, Weese JA, Riley WF. Mechanics of
[2] Anderson PW. Science 1972;177:393. Materials. 4th ed. New York: John Wiley and Sons; 1985.
[3] Meyer E, Hug HJ, Bennewitz R. Scanning Probe Microscopy. New York: [48] Du B, Tsui OKC, Zhang Q, He T. Langmuir 2001;17:3286.
Springer; 2004. [49] Nix WD. Mater Sci Eng A 1997;234–236:37.
[4] Fritz S. Understanding Nanotechnology. New York: Warner Books; 2002. [50] Heuberger M, Dietler G, Schlapbach L. Nanotechnology 1994;5:12.
[5] Ratner M, Ratner D. Nanotechnology: A Gentle Introduction to the Next [51] Domke J, Radmacher M. Langmuir 1998;14:3320.
Big Idea. New Jersey: Prentice Hall; 2003. [52] Li X, Gao H, Murphy CJ, Caswell KK. Nano Lett 2003;3:1495.
[6] Atkinson WI. Nanocosm. New York: Amacon; 2003. [53] Mazeran P-E, Loubet J-L. Tribol Lett 1997;3:125.
[7] Timoshenko S. Strength of Materials, Part I. 3rd ed. Malabar, FL: Krieger [54] Jourdan JS, Cruchon-Dupeyrat SJ, Huan Y, Kuo PK, Liu GY. Langmuir
Publishing Company; 1958. 1999;15:6495.
[8] Timoshenko SP, Goodier JN. Theory of Elasticity. 3rd ed. San Francisco: [55] Price WJ, Leigh SA, Hsu SM, Patten TE, Liu G-Y. J Phys Chem A
McGraw-Hill Book Company; 1970. 2006;110:1382.
[9] Timoshenko S. Theory of Plates and Shells. 3rd ed. New York: McGraw [56] Sun Y, Akhremitchev B, Walker GC. Langmuir 2004;20:5837.
Hill; 1987. [57] Gerberich WW, Yu W, Kramer D, Strojny A, Bahr D, Lilleodden E, et al.
[10] Hetherington C. Mater Today 2004;7:50. J Mater Res 1997;13:421.
[11] Binnig G, Quate CF, Gerber C. Phys Rev Lett 1986;56:930. [58] Fischer-Cripps AC. Vacuum 2000;58:569.
[12] Wiesendanger R. Scanning Probe Microscopy and Spectroscopy: [59] Chizhik SA, Huang Z, Gorbunov VV, Myshkin NK, Tsukruk VV.
Methods and Applications. New York: Cambridge University Press; Langmuir 1998;14:2606.
1994. [60] Baker SP. Thin Solid Films 1997;308–309:289.
[13] Bottomley LA, Coury JE, First PN. Anal Chem 1996;68:R185. [61] Li X, Bhushan B. Mater Charact 2002;48:11.
[14] Kim J-H, Yeon S-C, Jeon Y-K, Kim J-G, Kim Y-H. Sens Actuators A [62] Pietrement O, Troyon M. J Colloid Interface Sci 2000;226:166.
2003;108:20. [63] Oliver WC, Pharr GM. J Mater Res 1992;7:1564.
[15] Singh JP, Liu D-L, Ye D-X, Lu T-M, Wang G-C. Appl Phys Lett [64] Sneddon IN. Int J Eng Sci 1965;3:47.
2004;84:3657. [65] Johnson KL, Kendall K, Roberts AD. Proc R Soc Lond A 1971;324:301.
[16] Wu B, Heidelberg A, Boland JJ. Nat Mater 2005;4:525. [66] Greenwood JA. Proc R Soc Lond A 1997;453:1277.
[17] Shanmugham S, Jeong J, Alkhateeb A, Aston DE. Langmuir [67] Barthel E, Roux S. Langmuir 2000;16:8134.
2005;21:10214. [68] Chen S, Liu L, Wang T. Surf Coat Technol 2005;191:25.
[18] Giessible FJ, Hembacher S, Bielefeldt H, Mannhart J. Science 2000;289:422. [69] Du B, Liu J, Zhang Q, He T. Polymer 2001;42:5901.
[19] Gibson CT, Watson GS, Myhra S. Scanning 1997;19:564. [70] VanLandingham MR, Villarrubia JS, Guthrie WF, Meyers GF. Macromol
[20] Israelachvili JN. Intermolecular and Surface Forces. 2nd ed. San Diego: Symp 2001;167:15.
Academic Press; 1992. [71] Pharr GM. Mater Sci Eng A 1998;253:151.
[21] Sader JE, Larson I, Mulvaney P, White LR. Rev Sci Instrum 1995;66:3789. [72] Loubet JL, Georges JM, Meille G. In: Blau PJ, Lawn BR, editors.
[22] Cleveland JP, Manne S, Bocek D, Hansma PK. Rev Sci Instrum 1993;64:403. Microindentation Techniques in Materials Science and Engineering.
[23] Frisbie CD, Rozsnyai LF, Noy A, Wrighton MS, Lieber CM. Science Philadelphia: ASTM; 1986. p. 72.
1994;265:2071. [73] Loubet JL, Georges JM, Marchesini O, Meille G. J Tribol 1984;106:43.
[24] Noy A, Vezenov DV, Lieber CM. Annu Rev Mater Sci 1997;27:381. [74] Doerner MF, Nix WD. J Mater Res 1986;1:601.
[25] Schwarz UD, Koster P, Wiesendanger R. Rev Sci Instrum 1996;67:2560. [75] Bykov V, Gologanov A, Shevyakov V. Appl Phys A 1998;66:499.
[26] Nalwa HS. Advances in Surface Science, vol. 38. San Francisco: [76] Mazeran P-E, Odoni L, Loubet J-L. Surf Sci 2005;585:25.
Academic Press; 2001. [77] Nie H-Y, McIntyre NS. Langmuir 2001;17:432.
[27] Vickerman JC. Surface Analysis—The Principal Techniques. New York: [78] Tabet MF, Urban III FK. J Vac Sci Technol B 1997;15:800.
John Wiley and Sons; 1997. [79] DeRose JA, Revel JP. Microsc Microanal 1997;3:203.
[28] Cappella B, Dietler G. Surf Sci Rep 1999;34:1. [80] Markiewicz P, Goh MC. Langmuir 1994;10:5.
[29] Aston DE, Berg JC. Ind Eng Chem Res 2002;41:389. [81] Shulha H, Kovalev A, Myshkin N, Tsukruk VV. Eur Polym J
[30] Rosa-Zeiser A, Weilandt E, Hild S, Marti O. Meas Sci Technol 2004;40:949.
1997;8:1333. [82] Nie H-Y, Motomatsu M, Mizutani W, Tokumoto H. Thin Solid Films
[31] Krotil H-U, Stifter T, Waschipky H, Weishaupt K, Hild S, Marti O. Surf 1996;273:143.
Interface Anal 1999;27:336. [83] Cekada M, Panhan P. Vacuum 2001;61:235.
J.R. Withers, D.E. Aston / Advances in Colloid and Interface Science 120 (2006) 57–67 67

[84] A-Hassan E, Heinz WF, Antonik MD, D'Costa NP, Nageswaran S, [111] Schneider M, Zho M, Papastavrou G, Akari S, Mohwald H. Langmuir
Schoenenberger C-A, et al. Biophys J 1998;74:1564. 2002;18:602.
[85] Fernandez JM. Proc Natl Acad Sci U S A 1997;94:9. [112] Shulha H, Zhai X, Tsukruk VV. Macromolecules 2003;36:2825.
[86] Le Grimellec C, Lesniewska E, Giocondi M-C, Finot E, Vie V, [113] Attard P, Miklavcic SJ. J Colloid Interface Sci 2002;247:255.
Goudonnet J-P. Biophys J 1998;75:695. [114] Aston DE, Berg JC. J Colloid Interface Sci 2001;235:162.
[87] Hoh JH, Schoenenberger C-A. J Cell Sci 1994;107:1105. [115] Mulvaney P, Perera JM, Biggs S, Grieser F, Stevens GW. J Colloid
[88] Bowen WR, Lovitt RW, Wright CJ. Biotechnol Lett 2000;22:893. Interface Sci 1996;183:614.
[89] Kinney JH, Balooch M, Marshall GW, Marshall SJ. Arch Oral Biol [116] Nespolo SA, Chan DYC, Grieser F, Hartley P, Stevens GW. Langmuir
1999;44:813. 2003;19:2124.
[90] Collinsworth AM, Zhang S, Kraus WE, Truskey GA. Am J Physiol Cell [117] Nguyen AV, Nalaskowski J, Miller JD. Miner Eng 2003;16:1173.
Physiol 2002;283:1219. [118] Chan DYC, Dagastine RR, White R. J Colloid Interface Sci
[91] Goldmann WH, Ezzell RM. Exp Cell Res 1996;226:234. 2001;236:141.
[92] Shoelson B, Dimitriadis EK, Cai H, Kachar B, Chadwick RS. Biophys J [119] Gillies G, Prestidge CA, Attard P. Langmuir 2001;17:7955.
2004;87:2768. [120] Dagastine RR, Stevens GW, Chan DYC, Grieser F. J Colloid Interface Sci
[93] Stolz M, Raiteri R, Daniels AU, VanLangingham MR, Baschong W, Aebi 2004;273:339.
U. Biophys J 2004;86:3269. [121] Burnham N, Dominguez D, Movery R, Colton R. Phys Rev Lett
[94] Touhami A, Nysten B, Dufrene YF. Langmuir 2003;19:4539. 1990;64:1931.
[95] Rico F, Roca-Cusachs P, Gavara N, Farre R, Rotger M, Navajas D. Phys [122] Weisenhorn A, Khorsandi M, Kasas S, Gotzos V, Butt H-J. Nanotech-
Rev E 2005;72:021914. nology 1993;4:106.
[96] Pesen D, Hoh JH. Biophys J 2005;88:670. [123] Preuss M, Butt H-J. Langmuir 1998;14:3164.
[97] Vinckier A, Semenza G. FEBS Lett 1998;430:12. [124] Fielden M, Hayes R, Ralston J. Langmuir 1996;12:3721.
[98] Huang H, Kamm RD, Lee RT. Am J Physiol Cell Physiol 2004;287:1. [125] Ducker W, Xu Z, Israelachvili J. Langmuir 1994;10:3279.
[99] Dimitriadis EK, Horkay F, Maresca J, Kachar B, Chadwick RS. Biophys J [126] Burnham N, Colton R. J Vac Sci Technol 1987;7:2906.
2002;82:2798. [127] Weisenhorn AL, Maivald P, Butt H-J, Hansma PK. Phys Rev B Condens
[100] Kim YH. J Polym Sci Part A Polym Chem 1998;36:1685. Matter Mater Phys 1992;45:11226.
[101] Voit B. J Polym Sci Part A Polym Chem 2000;38:2505. [128] Snyder BA, Aston DE, Berg JC. Langmuir 1997;13:590.
[102] Voit B. J Polym Sci Part A Polym Chem 2005;43:2679. [129] Considine R, Hayes R, Horn RG. Langmuir 1999;15:1657.
[103] Jansen JFGA, de Brabander-van den Berg EMM, Meijer EW. Science [130] Schmitt F-J, Ederth T, Weidenhammer P, Claesson P, Jacobasch J-J.
1994;266:1226. J Adhes Sci Technol 1999;13:79.
[104] Bosman AW, Janssen HM, Jeijer EW. Chem Rev 1999;99:1665. [131] Burnham NA, Colton RJ, Pollock HM. Nanotechnology 1993;4:64.
[105] Abdelhady HG, Allen S, Davies MC, Roberts CJ, Tendler SJB, Williams [132] Hartley PG, Grieser F, Mulvaney P, Stevens GW. Langmuir 1999;15:
PM. Surf Sci 2004;558:99. 7282.
[106] Frechet JMJ. Science 1994;263:1710. [133] Aston DE, Berg JC. In: Mendez-Vilas A, Mendez-Vilas As, editors.
[107] Tully DC, Frechet JMJ. Chem Commun 2001:1229. Science, Technology and Education of Microscopy: An Overview, vol. 2.
[108] Mezzenga R, Boogh L, Manson J-AE. Macromolecules 2000;33:4373. Spain: Formatex; 2003. p. 630.
[109] Craighead HG. Science 2000;290:1532.
[110] Li J, Piehler LT, Qin D, Baker JR, Tomalia DA. Langmuir 2000;16:5613.

You might also like