Buckling-driven delamination of carbon nanotube forests
Parisa Pour Shahid Saeed Abadi, Shelby B. Hutchens, Julia R. Greer, Baratunde A. Cola, and Samuel Graham
Citation: Appl. Phys. Lett. 102, 223103 (2013); doi: 10.1063/1.4802080
View online: http://dx.doi.org/10.1063/1.4802080
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v102/i22
Published by the AIP Publishing LLC.
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 12 Jul 2013 to 131.215.71.79. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
Buckling-driven delamination of carbon nanotube forests
Parisa Pour Shahid Saeed Abadi,
1
Shelby B. Hutchens,
2
Julia R. Greer,
2
Baratunde A. Cola,
1,3,a)
and Samuel Graham
1,3,a)
1
George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, 771 Ferst Drive,
Atlanta, Georgia 30332, USA
2
California Institute of Technology, 1200 E. California Blvd. MC 309-81, Pasadena, California 91125, USA
3
School of Materials Science and Engineering, Georgia Institute of Technology, 771 Ferst Drive, Atlanta,
Georgia 30332, USA
(Received 13 February 2013; accepted 1 April 2013; published online 4 June 2013)
We report buckling-driven delamination of carbon nanotube (CNT) forests from their growth
substrates when subjected to compression. Macroscale compression experiments reveal local
delamination at the CNT forest-substrate interface. Results of microscale flat punch indentations
indicate that enhanced CNT interlocking at the top surface of the forest accomplished by
application of a metal coating causes delamination of the forest from the growth substrate, a
phenomenon not observed in indentation of as-grown CNT forests. We postulate that the
post-buckling tensile stresses that develop at the base of the CNT forests serve as the driving force
for delamination.
V
C
2013 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4802080]
Carbon nanotube (CNT) forests are composed of nomi-
nally vertically aligned, ultra-high aspect ratio nanotubes
with various degrees of alignment and entanglement along
their height.
1–6
Their unique thermal, electrical, photonic,
and mechanical properties, which in part arise from the com-
plex nano- and micro-structural detail of these materials,
render them potential candidates for many applications. Their
mechanical response, which potentially affects the functional-
ity of CNTs in these applications, has been investigated by
various methods including nano- and micro-scale indentation
of films of CNT forest and compression of pillars comprised
of CNT forest.
1,4–31
The deformation mechanism was found
to be dominated by periodic buckling in the case of CNT for-
est micro-pillars
5,11,12,14,31
and by the combination of buck-
ling and shear offsets in flat punch indentations of CNT forest
films.
1,4,10,13,17,19
While the visualization of nanoscale and
microscale compression of CNT forests adds useful insight
into the mechanics of these structures, it is not able to capture
all deformation. Alternate failure modes may occur during
the deformation of larger scale CNT forests that could be rel-
evant to component or system level applications.
We report on the delamination of CNT forest films as
revealed by macro-compression experiments. Post-
compression scanning electron microscope (SEM) images of
1cm 1 cm CNT forests reveal local delamination of the
films at nominal pressures as low as 50 kPa. Delamination
driven by in-plane buckling has been observed in monolithic
thin films where the biaxial compressive residual stress in
the film led to local buckling. This buckling subsequently
drove lateral propagation of the delamination at the inter-
face.
32,33
Buckling in the case of monolithic films is the “thin
plate” type of buckling, i.e., in the direction perpendicular to
the film surface. In contrast to this mechanism, the delamina-
tion of CNT forests reported here, is related to the column
buckling of high aspect ratio CNTs in the direction
perpendicular to the CNTs, i.e., parallel to the film surface.
We show that the occurrence of delamination is facilitated
by the bending moments acting on the CNT-substrate inter-
face as a result of local buckling. We also show that stress
concentrations associated with the non-uniformity of CNT
heights play a role in delamination during macro-
compression. Furthermore, we demonstrate a method for tai-
loring the degree of interaction among CNTs to cause
delamination in micro-scale indentation of CNT forests. This
method consists of coating CNT tips with a 1 lm-thick layer
of aluminum, which results in delamination during the on-
edge indentations. Delamination does not occur during the
on-edge indentations in the as-synthesized forests. This
emphasizes the significant effect that mechanical constraints
can have on the occurrence of delamination and informs the
design of structures to take advantage of this phenomenon.
Macro-compression testing was performed on CNT for-
ests grown on 1 cm 1 cm Si substrates. Nominal pressures
of 50 and 100 kPa were applied to the CNT forests by plac-
ing 0.5 kg and 1 kg weights on the samples. SEM images
taken from deformed CNT forests are shown in Fig. 1.
Periodic buckles form close to the substrate (Fig. 1(a)), as
was the case in the indentation of CNT forests grown with
the same recipe.
4
The location and commencement of such
folds has been shown to be a function of the local density,
tortuosity, and entanglement of the CNTs, which commonly
vary along the height of CNT forests.
1,4,5
Buckle formation
is not uniform in any of the compressed macroscale CNT
forests. The number of buckles along any given edge region
range from zero to 20. This is likely due to the inherent
nonuniformity in the height of the CNT forests—the height
variance in each sample was approximately 10%–50%,
which led to a nonuniform distribution of the compressive
stresses within the sample. Height variation is a common
consequence of growing the macroscale CNT forests by
chemical vapor deposition (CVD)
19,34
likely caused by flow
and temperature variations in the growth chamber. We
observe a local delamination of the CNT forest films from
a)
Authors to whom correspondence should be addressed. Electronic addresses:
sgraham@gatech.edu or cola@gatech.edu.
0003-6951/2013/102(22)/223103/5/$30.00
V
C
2013 AIP Publishing LLC102, 223103-1
APPLIED PHYSICS LETTERS 102, 223103 (2013)
Downloaded 12 Jul 2013 to 131.215.71.79. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
the growth substrates (Figs. 1(b) and 1(c)), which occurs in
all tested CNT forests within a span of 1 mm of the edge of
the sample under 50 kPa of nominal pressure. The rest of the
edge—a span of 9 mm—remains bonded. The lateral extent
of the delaminated region increases to greater than 2 mm
when the nominal pressure is doubled. No such delamination
is observed in micro-scale flat punch indentation testing of
CNT forests grown with the same recipe and loaded to in-
dentation stresses higher than the nominal stress applied in
macroscale compression.
4
Similar delamination occurs in micro-indentation of
CNT forests whose top surface is coated with a 1 lm-thick
Al layer, deposited through e-beam evaporation. Such
delamination is not observed during identical tests on the
uncoated samples. Indentations to the depth of 20 lm were
performed with an 80 lm wide flat punch on the edge of the
140 lm tall CNT forest coated with a 1 lm-thick Al.
Indentations were performed in situ using a dedicated nano-
mechanical instrument, SEMentor,
4,5
a combination of SEM
and a nanomechanical module similar to a nanoindenter.
SEMentor allowed for direct observation of the deformation
while concurrently gathering load and displacement data.
Fig. 2 illustrates the deformation mechanism of such a repre-
sentative CNT forest coated with a 1 lm-thick Al film.
Images are frames from the recorded video in SEMentor (see
supplementary material
35
). Fig. 2(a) shows the edge of the
coated CNT forest upon its initial contact with the indenter
before the application of pressure. At very small strains, a
buckle initiates at 35 lm above the substrate, which repre-
sents 25% of the total height (Fig. 2(b)). Lateral propagation
of this buckle is the dominant deformation mode up to a
strain of 5.5%, when the CNT film begins to detach from the
substrate directly underneath the indenter and vertical cracks
appear in the film along the projected indenter edges
(Fig. 2(c)). Fig. 2(d) illustrates the maximum displacement
of the indenter. At this point, the detached region expands
laterally by 230 lm, and the vertical distance between the
delaminated CNT forest base and the substrate reaches a
maximum of 17 lm. Vertical cracks also propagate along the
CNT heights and span over 70% of the film thickness but do
not appear to break through the top Al coating. The CNT tips
remain adherent to the coating throughout the test.
Approximately, 30% of the total deformation is recovered
upon unloading (see Figs. 2(e) and 2(f)). The recovery for
the crack opening along the delaminated CNT-substrate
interface on the edge is equivalent to the difference between
the maximum and the final vertical dimension of the open-
ing, divided by the maximum vertical dimension. This recov-
ery is calculated to be approximately 50% (Figs. 2(d) and
2(e)).
The indentation stress-strain data are shown in Fig. 2(f).
The axial indentation strain is calculated by dividing the
measured vertical displacement by the initial height of the
CNT forest, averaged over 1 cm width. To calculate the axial
indentation stress, the nano-indenter force is divided by the
homogenized contact area underneath the indenter. This area
is equivalent to the nanoindenter footprint and remains con-
stant during the test. Data points corresponding to the images
in Figs. 2(a)–2(e) are marked on the indentation stress-strain
data in Fig. 2(f) and reveal the presence of two distinct defor-
mation regimes: (1) before buckling, characterized by a lin-
ear loading slope of 75 MPa, and (2) after buckling, where
the stress vs. strain relationship remains linear but the slope
decreases to 4–6 MPa. Transition between the two regimes
occurs gradually, over a strain range of 1%–2%. Indentation
stress dropped from 885 to 870 kPa at point (c) (see inset in
Figure 2(f)). This drop is likely due to the initiation of
FIG. 1. SEM images of the CNT forests base, illustrating permanent defor-
mation after macro-compression. (a) SEM image of an edge region with
multiple buckles but no delamination. (b) SEM image of a location illustrat-
ing multiple buckles and delamination of CNTs. (c) A magnified view of a
region where CNTs buckled and interface delamination occurred.
FIG. 2. SEM micrographs (a–e) and stress-strain data (f) illustrating the de-
formation of a CNT forest coated with a 1 lm Al film that was indented at
its edge. Images correspond to points a–e on the indentation stress-strain
curve in (f).
223103-2 Pour Shahid Saeed Abadi et al. Appl. Phys. Lett. 102, 223103 (2013)
Downloaded 12 Jul 2013 to 131.215.71.79. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
delamination and formation of vertical cracks, which occur
faster than the prescribed displacement rate. At this point,
the load drops to maintain the constant prescribed displace-
ment rate and the loading slope decreases from 6 to 4.7 MPa
(point (c)). In situ SEM video of the indentation and the cor-
responding video of the stress-strain curve are provided in
the supplementary material.
35
Such a deformation signature
and concurrent stress-strain data are consistent over multiple
locations (Figure S2 in the supplementary material
35
).
To gain insight into the stress distribution within the
CNT forest that causes its delamination from the substrate,
we employed the Euler-Bernoulli beam bending theory. This
theory provides a means for stress calculation under a small
deflection in an elastic, isotropic, and homogenous beam.
Although the classic beam bending theory does not fully cap-
ture the complexities within the CNT forest structure such as
inelasticity, inhomogeneity, and anisotropy, it provides useful
information about the likely stress distribution under load.
We show that it provides a reasonable estimation of the ten-
sile stresses when combined with the deformations observed
in situ and the ex situ measurement of the local effective elas-
tic modulus. The classic beam bending theory dictates that
the bending moment, M , and curvature, j, are related by
M ¼ EIj ¼
EI
R
; (1)
where E, I, and R represent the elastic modulus, second
moment of inertia, and radius of curvature, respectively. The
strong van der Waals driven interactions among the CNTs
render Eq. (1) ineffective in describing the bending of each
individual nanotube. Instead, it is reasonable to apply Eq. (1)
to describe the deformation of coalesced groups of CNTs
that form because of van der Waals interactions and there-
fore buckle in unison. Although the eventual buckling of
such bundles of CNTs is inelastic, their initial bending
before buckling is observed to be mainly elastic. This is evi-
denced by the greater than 50% recovery in the crack open-
ing displacement along the delaminated CNT-substrate
interface on the edge upon release of the load (Figs. 2(d) and
2(e)). To account for the inhomogeneity of the CNT forests
along the height, the effective moduli of the base of CNT
forests were measured by ex situ nanoindentation into the
base of a macroscale CNT forest. The CNT forest was sepa-
rated from the growth substrate by tensile loading and the
nanoindentation was performed using a conical tip (more
details are provided in the supplementary material
35
).
A side view of the CNT deformation in the image shown
in Fig. 3(a) is depicted schematically in Fig. 3(b), which
illustrates the out-of-plane bending of the structure in the
y-direction. The curvature observed in the deformed CNT
forest provides evidence of the presence of a bending
moment in the direction schematically depicted in Fig. 3(c).
This bending moment produces the normal tensile and com-
pressive stresses. A differential element of the curved section
(marked in Fig. 3(b) with a circle) is used in Fig. 3(c) to
demonstrate the bending moment and the stress distribution
projected in two dimensions. The bending moment in the
depicted direction induces tensile stresses on the outer sur-
face of the element. The normal stress due to bending, r,is
related to the bending moment, M,by
r ¼
My
I
; (2)
where y, depicted in Fig. 3(c), represents the distance from
the neutral axes. By combining Eqs. (1) and (2), the tensile
stress at the CNT-substrate interface at the beginning of
delamination is estimated by
r ¼
Ey
max
R
; (3)
where y
max
and R are estimated using the deformations
observed in the in situ SEM videos and E is estimated from
the ex situ nanoindentation testing of the base of CNT forests
released from the substrate (more details are provided in the
supplementary material
35
). Substituting the estimated values
into Eq. (3) results in r in the range of 0.4–0.8 MPa at incipi-
ent delamination.
To compare this analytical estimate with experimentally
measured interfacial strength, we perform macroscale tensile
testing of the substrate-supported CNT forests. The tensile
experiments, which are performed by applying tensile loads
on grips bonded to both the growth substrate and CNT tips,
resulted in full delamination of CNT forests from the sub-
strates and reveal the adhesion strength between CNT fo rest
and substrate to be between 0.2 and 0.4 MPa. This ran ge of
adhesive strength falls within the same order of magnitude
as the theoretically estimated tensile stresses due to bending
proposed above, which lead to delamination at the interface.
This agreement suggests that the tensile stresses developed
in the entangled bundles of CNTs after buckling are suffi-
ciently high to drive the delamination between the CNT for-
est and the growth substrate. This strengthens the evidence
FIG. 3. (a) Micrograph from Fig. 2(c)
with the addition of a vertical line at the
middle of the indenter to guide the eye.
The inward arrow ( inside a circle) is
the direction of the buckling, (b) illustra-
tion of the side view of the out-of-plane
deformation of the CNT forest along the
vertical line in (a). The left arrow shows
the direction of the buckling. (c)
Schematic of the bending moment and 2-
dimensional stress distribution in an ele-
ment highlighted in (b). (d) Curvature of
the section of CNT forest under the
buckle location.
223103-3 Pour Shahid Saeed Abadi et al. Appl. Phys. Lett. 102, 223103 (2013)
Downloaded 12 Jul 2013 to 131.215.71.79. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
that tensile loading due to out-of-plane bending of the CNT
forest is the main reason for delamination.
No buckling-followed-by-delamination behavior, as
observed in the Al-coated CNT samples, is observed in the
uncoated CNT forests when comparing samples indented to
the same depth, 30 lm (Ref. 4) (Fig. 4(a)). This suggests that
even if the tensile stresses are also developed within the
uncoated CNT forests, they are lower than the interfacial ad-
hesion. The as-grown samp les had more relaxed mechanical
boundary constraints, which enabled additional mechanisms
to carry the strain, for example, by forming vertical shear
offsets along the indenter edges that propagated to the sur-
face (Fig. 4(a) ). This causes the CNTs directly under the in-
denter to collectively separate from the rest of the CNT
forest as a block, presumably because of their low resistance
to shear as has been previously reported.
4,8,9
In contrast, the
rigid constraint at the top of the coated CNT forests drives
their deformation via out-of-plane bending of the buckled
area, similar to Mode I crack opening. Larger crack-opening
displacements correspond to lower radii of curvature, which
in turn generate a greater bending moment and higher tensile
stress acting on the interface between the CNT forest and the
substrate. On-edge indentation of similar CNT forests coated
with 10 thinner Al films of 100 nm results in fracture of the
coating and in deformation mechanisms similar to those
in the uncoated CNT forests, i.e., with no delamination
(Fig. 4(b)). SEM images of the three characteristic deforma-
tion morphologies, corresponding to no coating, 100 nm-
thick Al coating, and 1 lm-thick Al coating are illustrated in
Fig. 4.
Different from the delamination of the coated CNT sam-
ples as a result of the top strain constraint in micro-
indentation, the detachment of the localized interfacial
regions (Fig. 1) is likely caused by the non-uniform heights
of the CNTs in the macro-compressed samples. The variation
in height was measured to be on the order of tens of microns
over the lateral extent of hundreds of microns (Fig. S3 in the
supplementary material
35
). Compressing such a structure
results in the originally taller CNTs bending and buckling
first, causing lateral displacements within the forest before
full contact with all CNTs is established. This likely leads to
the formation of local stress concentrations within the sam-
ples, i.e., generating the local tensile stress. This tensile
stress drives the detachment of certain interfacial regions.
The non-constant height distribution likely has a minimal
effect during the microscale tests of CNT forests because the
wavelength of their height variation is on the order of the
nanoindenter dimensions or larger.
In summary, we describe the mechanisms of delamina-
tion of 130–190 lm-thick CNT forest films, grown via CVD,
from their Si substrates. Both macro-compression experi-
ments of as-grown CNT forests and micro-compression
experiments of Al coated CNT forests display loss of interfa-
cial cohesion at sufficient deformation. In macroscale com-
pression of as-grown CNT forests, localized delamination
initiates under nominal pressure as low as 50 kPa. The mech-
anism for delamination under macro-compression may be
explained by the development of local stress concentrations
within the film, which imposes a local tensile load that is suf-
ficiently high to overcome the adhesion strength between
CNTs and the substrate. These local stress inhomogeneities
are likely a result of the height variation at the CNT forest
surface and the van der Waals interactions between CNTs. In
the case of micro-scale indentation of the CNT forest coated
with a 1 lm-thick Al film, the rigid constraint formed as a
result of the applied metal coating is responsible for the
delamination. Simple analytical beam bending theory esti-
mates the magnitude of the tensile stress to be on the same
order of magnitude as the measured CNT-substrate adhesion
strength, 0.4–0.8 MPa. These findings demonstrate that tuning
the mechanical boundary constraints provides an important
design parameter for attaining greater reliability in devices
and materials that utilize CNT forests. For example, continu-
ous or percolative thermal and electrical conduction paths
through the CNTs can be severed by delamination, which
would cause decay in the functionality of the CNT forest film
as an interface material.
36
Such a tunable delamination can
also be used in the design of micro-electromechanical sys-
tems. For instance, decreased electrical conductance in CNT
forests coated with metal films could potentially be utilized
to design nano-based mechanical pressure switches and
impact sensors that work based on interface delamination.
The details of the experimental procedures are as fol-
lows. CNT forests were grown on 1 cm 1 cm Si substrates
coated with Ti (30 nm)/Al (10 nm)/Fe (3 nm) by electron
beam evaporation (e-beam). Low pressure CVD (LPCVD)
was used for synthesis of CNT forests in a commercial CVD
system (Black Magic Pro 4
00
, Aixtron SE). The growth
FIG. 4. Post-indentation deformation morphologies on the edge of (a) an
uncoated CNT forest and CNT forests coated with (b) 100 nm and (c) 1 lm
Al films. Scale bar in each image corresponds to 50 lm.
223103-4 Pour Shahid Saeed Abadi et al. Appl. Phys. Lett. 102, 223103 (2013)
Downloaded 12 Jul 2013 to 131.215.71.79. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://apl.aip.org/about/rights_and_permissions
