You are on page 1of 12

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/authorsrights
Author's personal copy
Aging of asphaltic binders investigated with atomic force microscopy
L.M. Rebelo
a
, J.S. de Sousa
a,
, A.S. Abreu
a
, M.P.M.A. Baroni
b
, A.E.V. Alencar
c
, S.A. Soares
c
,
J. Mendes Filho
a
, J.B. Soares
d
a
Departamento de Fsica, Universidade Federal do Cear, Caixa Postal 6030, Campus do Pici, 60455-760 Fortaleza, Cear, Brazil
b
Instituto Federal de Educao, Cincia e Tecnologia, 01109-010 So Paulo, Brazil
c
Departamento de Qumica, Universidade Federal do Cear, Brazil
d
Departamento de Engenharia de Transporte, Universidade Federal do Cear, Brazil
h i g h l i g h t s
The aging of bitumen is investigated with Atomic Force Microscopy.
The aging process cause an increase in size of the asphaltene micelles.
Short term aging induces the formation of fractal-like microstructures.
Stiffness increases half (one) order of magnitude for short (long) term aging.
Viscosity increases half-order of magnitude, mainly during short term aging.
a r t i c l e i n f o
Article history:
Received 20 June 2013
Received in revised form 19 August 2013
Accepted 6 September 2013
Available online 25 September 2013
Keywords:
Atomic force microscopy
Aging effects
Asphalt binders
a b s t r a c t
We investigated the short and long term aging of asphalt cement (AC) with different AFM techniques
(topography, phase and friction imaging and nano-indentation experiments). The aging process induces
a growth and nucleation of the asphaltene micelles with a concomitant reduction of the maltene phase,
whereas the short term aging induces the formation of fractal-like micellar structures. The friction inves-
tigation shows that the aging processes reduce the binder friction coefcient by 50%, and this reduction
occur predominantly during the short term aging, while the growth of the micelles occur predominantly
during the long term aging. The micro-indentation experiments revealed that the aging processes cause a
stiffening of the AC lm (half-order of magnitude for short term aging, and one order of magnitude for
long term aging). The aging process also increased the apparent viscosity of the AC lms by half-order
of magnitude.
2013 Elsevier Ltd. All rights reserved.
1. Introduction
Aging of asphalt cement (AC), or bitumen, is a key aspect that
can lead to premature deterioration of asphalt pavements [1].
ACs are affected by oxygen and ultraviolet radiation, which are
the main factors of the aging process. It occurs primarily during
mixing, but also during compaction and service. The material
undergoes chemical alterations that affect its mechanical proper-
ties, making it more viscous and brittle, thereby interfering in its
behavior under repetitive efforts. Several other factors affect the
aging process of the mixture: (i) asphalt characteristics, (ii) nature
of aggregates, (iii) particle size distribution, and (iv) air void con-
tent. Plant related parameters such as mixing temperatures and
time can also inuence mixture performance. Good adhesion be-
tween the asphalt binder and aggregates is also crucial for the con-
struction of durable pavements. One of the most common
problems that reduces the lifetime of pavements is the loss of
adhesion due to inltration of water between the aggregate and
the binder [2]. Therefore, the microscopy investigation of the adhe-
sion characteristics of asphalt binders is a very important topic of
research to future developments of pavement technology.
One of the greatest difculties to understand and predict as-
phalt pavement behavior is the high variability among different
AC sources with respect to their chemical composition and micro-
structure [36]. The worldwide increase of trafc volume demands
more resistant pavements, reduced maintenance interventions,
and increased life cycle. Temperature susceptibility is another
important variable. ACs must exhibit good rheological perfor-
mance in a wide range of temperatures, offering exibility in low
temperatures and rigidity in high temperatures to avoid thermal
and fatigue cracking, and permanent deformations. Although
empirical data and mechanistic approaches provide good indica-
tions of pavement performance, the fact that ACs are mixed with
0016-2361/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.fuel.2013.09.018

Corresponding author. Tel.: +55 8533669017.


E-mail address: jeanlex@sica.ufc.br (J.S. de Sousa).
Fuel 117 (2014) 1525
Contents lists available at ScienceDirect
Fuel
j our nal homepage: www. el sevi er . com/ l ocat e/ f uel
Author's personal copy
aggregates, along with the high variability of environmental and
loading conditions to which pavements are subjected, there is a
growing need to understand how the asphalt microstructures
inuence the overall behavior of ACs [7,8] and to dene strategies
to engineer the asphalt properties to improve its performance. One
common strategy is to modify ACs with polymers (e.g. SBS and
EVA) to increase its elasticity response and concomitantly reduce
its viscosity [9,10]. Anti-oxidant additives are also commonly used
to reduce the asphalt susceptibility to oxygen [1114].
Approaches such as X-ray diffraction [15], size-exclusion chro-
matography (SEC) [16] and various microscopy techniques, e.g.
scanning electron microscopy (SEM) [17], transmission electron
[18], (TEM) phase contrast [19], polarized light [20], laser-scanning
[21], uorescence [22], and atomic force microscopy (AFM) [7], have
been used to investigate asphalt microstructure. The combined use
of different AFM modes and various experimental conditions pro-
vides a basis for a comprehensive examination of the micromechan-
ical properties of asphalt [7]. In particular, AFM can be used to
provide insights into surface topography, phase separation and
mechanical properties such as stiffness, adhesion, viscosity and fric-
tion. The AFM technique has been already employed to investigate
the microstructure of bitumen. To mention a few, Pauli et al. used
AFMto characterize bituminous binders and their respective proper-
ties, correlating the surface morphology with the constituents in the
bitumen [23]. Masson et al. used phase-detection AFM to evaluate
bitumen morphology and proposed a system to classify bitumen
into three distinct groups, based on the different domains or visible
phases [5]. Aging effects have also been investigated by AFM. Zhang
et al. found that the overall surface stiffness increased and the bitu-
men surface became more solid-like, but the extent of these changes
was dependent on aging conditions [1].
In this work, several AFM-related techniques such as topogra-
phy, phase and friction imaging, and force-volume (FV) analysis
are employed to investigate the nano-morphology and nano-rheol-
ogy of one type of AC before and after aging. The AC aging was in-
duced with the well known RTFOT [24] and PAV [25] processes,
which simulates short and long term aging, respectively.
2. Materials and methods
The present study makes use of one type of bitumen (50/70
penetration grade), processed by at Lubnor/Petrobrs and pro-
duced in Fazenda Campo Alegre (Esprito Santo, Brazil). Thin Layer
Cromatography (TLC-FID) was used to determine SARA fractions of
the bitumen, which resulted in 12% of saturates, 43% of aromatics;
18% of resins and 28% of asphaltenes. Differential Scanning Calo-
rimetry determined that the wax content in the bitumen is 0.25%.
2.1. Aging process
The rolling thin lm oven test (RTFOT) was used to simulate
short-termaging. It measures the effect of heat and air on a moving
lm of semi-solid asphaltic binder. A temperature of 163 C and
duration of 85 min is expected to produce aging effects comparable
to average asphalt plant conditions. The pressure aging vessel
(PAV) test simulates long-term aging equivalent to 510 years of
in-service pavements [26]. The PAV method was used to age RTFOT
residues.
2.2. AFM experiments
An AFM (Nanoscope IIIa, Bruker, Santa Barbara, CA, USA) was
used to measure topography, phase, friction and mechanical
properties of AC lms. Samples are measured in both contact and
intermittent (tapping) modes of operation. In tapping mode
(topography and phase), we used rectangular silicon cantilevers
(TESP7, Bruker) with a nominal spring constant of 42 N/m. In con-
tact mode (topography, deection, friction and FV analysis), we
used V shaped cantilevers (OTR8, Bruker) with a nominal spring
constant of 0.57 N/m. All images were acquired at room tempera-
ture and normal pressure, with a scan rate of 1 Hz and resolution
of 512 512 lines.
For most AFM measurements (topography, phase and friction
images), AC was heated and a small amount was deposited on
the center of a 13 mm diameter glass slide. The AC on the coverslip
was heated up to 150 C during 2 min until it became fully spread
on the glass surface thus forming an uniform lm. The samples
were cooled at room temperature in a closed chamber to avoid
exposure to contaminants during 24 h. For FV analyses, the sample
preparation was slightly different: the AC was deposited on the
glass slide such that only half of it is covered. This step is essential
for the force curves calibration, which was performed on the ex-
posed hard surface [27].
We have also measured AC properties using the AFM in lateral
force mode (LFM) [28,29]. The schematics of this technique is de-
picted in Fig. 1. In this operation mode, the AFM system detects
the torsion of the cantilever around its axis as the tip scans the
sample surface laterally. The torsion amplitude provides an indi-
rect measure of the friction coefcient between the cantilever tip
and the surface. Therefore, LFM maps offer a convenient method
to identify regions on which friction is higher or lower, which
can be readily correlated with asphalt microstructures.
For all types of measurements, three coverslips were prepared
for each sample. For topography, phase and friction images, we ob-
tained three images of different regions for each coverslip in order
to explore various regions of the lm at the frequency of 5 Hz. For
the FV analysis, we used ve different indentation rates: 0.5 Hz,
5.0 Hz, 10 Hz, 15 Hz and 30 Hz. Each map is composed of 32 32
force curves equally distributed in a region of 50 lm. The large
amount of data points allows a good statistical validation of our
measurements.
3. AFM data analysis
Fig. 2 shows the main features observed in conventional force-
distance AFM curves measured in viscoelastic and adhesive sam-
ples like ACs. Several of those curves were obtained with the FV
contact mode in different regions of the AC lms in order to deter-
mine the micro-rheological properties of those lms. In this sec-
tion, we describe the models used in the analysis of AFM force
curves to extract the following properties: elasticity (Youngs)
modulus [3032], apparent viscosity [33] and adhesiveness.
3.1. Elasticity modulus
AFM force curves exhibit the form d = f(z), where d is the canti-
lever deection and z is the corresponding translation of the piezo-
electric actuator. A schematics of a typical deection-displacement
curve measured in viscoelastic samples is shown Fig. 2. The hyster-
esis in the approach/retract cycle is a consequence of the viscoelas-
tic response of the sample. A maximum deection of 50 nm is
imposed to avoid excessive indentation. Beyond the contact point
(z
0
, d
0
), the actual cantilever deection is Dd = d d
0
, where d
0
is
the cantilever deection far away from the sample surface. The
corresponding piezo-actuator displacement is Dz = z d
0
, where
d
0
also represents the piezo displacement for which the cantilever
touches the sample surface. The sample indentation d is obtained
with d = Dz Dd. The force deecting the cantilever is obtained
by Hookes law F = k
c
Dd, where k
c
is the cantilever spring constant.
This force is transmitted to the sample causing an indentation. In
Hertz contact theory, the loaddisplacement relationship for coni-
cal indenters is given by [3032]:
16 L.M. Rebelo et al. / Fuel 117 (2014) 1525
Author's personal copy
F
cone

2
p
E
1 m
2
tanhd
2
; 1
where E and m represent the elasticity modulus and Poissons ratio
of the indented material, respectively. The samples are considered
to be virtually incompressible (m = 0.5). h represents the half-open-
ing angle of the conical tip. The above expression is used to t the
experimental (z, d) curves to extract the elasticity modulus E. For
this, one must choose an interval to which the tting must be ap-
plied, as shown in Fig. 2. The detection of the contact point (z
0
, d
0
)
is performed as follows: d
0
can be determined by averaging an arbi-
trarily chosen region of the non-contact portion of the curve. As for
z
0
, the best solution for the analysis of few curves is to visually
determine the contact point. For a large number of curves, the best
strategy is to treat z
0
as a tting parameter as well.
3.2. Apparent viscosity
For viscous samples, the response to the cantilever force is com-
posed of an elastic and a viscous component (F
T
= F
el
+ F
v
), such that
the work done by the cantilever is partially lost by internal friction,
generating an hysteresis in the approach/retraction cycle. The work
difference between approach and retraction (DW
T
= DW
el
+ -
DW
v
= W
(app)
W
(ret)
) can be calculated directly from the (z,d)
curves as:
DW
T
k
c
Z
z
2
z
1
d
app
d
ret
dz: 2
Since DW
el
= 0, DW
T
= DW
v
and DW
v
is numerically equal to the en-
ergy lost by friction in the cell, and can be calculated by:
DW
v

Z
z
2
z
1
F
app
v
F
ret
v

dd: 3
Since the elastic force in Hertzian contact theory takes into account
the contact area between indenter and sample, we assume that the
viscous component should also consider the contact area and thus
can be modeled as F
v
= gdA/dt, where A = pd
2
tan
2
h is the contact
area between the conical indenter and sample, and g is the apparent
viscosity. The area A changes over time due to (i) sample relaxation
and (ii) the movement of the cantilever with respect to the sample.
Assuming that the main contribution is due to the latter, one can
make the following simplication dA/dt = (dA/dz)(dz/dt) = v
z
dA/dz,
where v
z
is the velocity of the piezo movement. This approximation
leads to an analytical form of DW
v
that can be used to calculate the
apparent viscosity directly from (z, d) curves as:
g
k
c
ptan
2
hv
z

R
z
2
z
1
d
app
d
ret
dz
d
2
2
d
2
1

app
d
2
2
d
2
1

ret
: 4
3.3. Adhesiveness
A typical force curve measured in a material exhibiting adhe-
siveness to the AFM tip is shown in Fig. 2. During cantilever retrac-
tion, the cantilever undergoes a negative deection, until its out-
pulling force exceeds the adhesion forces, losing contact with the
sample. The point at which the adhesion force between the probe
and the sample is maximum corresponds to the minimum deec-
tion point (Fig. 2). Using these general characteristics, the adhe-
siveness of bitumen can be deduced qualitatively in two different
ways: (i) the maximum force of adhesion between the probe and
the sample, and (ii) calculating the work of adhesion forces. The
work is obtained by calculating the area of the deection-displace-
ment curve whose deection is negative, i.e., the portion corre-
sponding to the adhesion forces.
4. Results
4.1. Topography and phase images analysis
Fig. 3 shows representative images (topography and phase in
tapping mode) of each sample studied in this work. Both unaged
Fig. 1. Schematics of the lateral force microscopy (LFM). The cantilever scans the
surface perpendicularly to its axis. The friction between the cantilever tip and
surface induces a torsion around the cantilever axis. The friction force is measured
by tracking the lateral movement of the reected laser beam in the photodetector.
The larger the torsion, the larger is the lateral movement of the laser beam which
leads to larger voltage reading of the photodetector.
Fig. 2. Schematics of a typical AFM deection-displacement curve obtained from
asphalt cements, exhibiting hysteresis between approach and retract curves, and
adhesiveness. The tting of these curves between a given interval of deection
[d
1
, d
2
] with Hertz model provides an estimate of the sample elasticity moduli E
[3032]. From the hysteresis, we employed the method of Ref. [33] to determine the
apparent viscosity. The strength of the adhesive forces is qualitatively determined
by means of the work done by the cantilever in the region of the negative cantilever
deection.
L.M. Rebelo et al. / Fuel 117 (2014) 1525 17
Author's personal copy
and aged samples show that the AC surface is composed of do-
mains of asphaltenes (micelles) in a sea of hydrocarbons (maltene).
We did not observe the presence of bee structures in our samples
because the amount of wax in the bitumen used in this work is
negligible [5].
The striking differences between unaged and aged samples are
the following. (i) The topography images of aged samples display
nearly the same features of the phase images, while in unaged
samples the topography image shows a nearly at surface, exhib-
iting only small dark regions in the center of the asphaltene mi-
celles. These dark spots are similar to the so called sal phase
present in some types of bitumen investigated by Masson et al.
[5]. In fact, those micelles can only be observed in phase images
for unaged ACs. (ii) The phase contrast between asphaltene do-
mains and the hydrocarbon sea is inverted between unaged and
aged samples. The brighter the image, the larger is the phase lag
difference between the input sinusoidal cantilever signal and its
response. The phase images in Fig. 3 clearly shows that the phase
lag in the asphaltene domains is smaller compared to the hydro-
carbon sea in unaged samples, while this trend is inverted in aged
samples. Disregarding complicated adhesive effects, it is well ac-
cepted that regions displaying small phase angles in tapping mode
images exhibit enhanced storage moduli, while brighter regions
exhibit enhanced loss moduli. In that sense, considering that most
of the volume of the AC lm is composed of hydrocarbon sea in
both unaged and aged samples, unaged samples should exhibit
Fig. 3. Topography (left panels) and phase images (right panels) of a unaged (ab), RTFOT aged (cd) and PAV aged (ef) AC lms. The lateral dimensions of the images are
10 lm 10 lm. The scale of topography images is 20 nm and for phase image is 10.
18 L.M. Rebelo et al. / Fuel 117 (2014) 1525
Author's personal copy
an enhanced uid-like behavior compared to aged samples. (iii)
Comparing only the phase images of aged ACs, we also notice that
the phase contrast between darkest and brightest areas is different
in PAV aged AC compared to RTFOT aged AC. As the phase scale is
the same for all images, we observe that micelle regions in PAV
aged ACs are not as bright as in RTFOT aged samples. Besides that,
the micelles in PAV aged samples display fractured-like features,
indicating that both aging processes play an important and distin-
guished role in the AC micelle structure. (iv) As a general trend, all
phase images conform well to the AC colloidal model of Yen et al.
where the asphalt binder can be structured in either sol or gel struc-
tures [34,8]. In the sol conguration the asphaltenes form micelles
surrounded by a sea of hydrocarbons, while in the gel conguration
the asphaltene micelles tend to form large and disordered clusters.
In this sense, unaged ACs display sol behavior in most of the cases
observed in our study, while the aged samples still display sol
behavior, but exhibiting growth and clustering of the micelles. (v)
We have also observed in aged ACs that the number of micelles re-
duced followed by an increase in their average sizes as compared to
unaged ACs. This is agreement with recent the work of Leguern et al.
that demonstrated that the amount of asphaltenes in ACs increase
during aging processes [6], and with the work of Wu et al. that
showed that aging processes favor the clustering of asphaltene mi-
celles forming larger ones [35]. Based on those experimental nd-
ings, we attibute the observed clustering of micelles to an increase
in the content of asphaltenes induced by the aging processes.
Figs. 4 and 5 show phase images of a rich set of interesting fea-
tures observed in RTFOT aged samples. For example, Fig. 4 display
images taken in other regions of the same coverslip used in Fig. 3.
Fig. 4(a) and (b) exhibit sol-like arrangement, with the micelles
exhibiting coreshell like structure. Interestingly, Fig. 4(c) and (d)
show the coreshell micelles disposed in a gel-like arrangement.
Fig. 5(a) and (b) show branch-like structures of resins around the
asphaltenes which probably indicates the onset formation of the
shell structure around the micelles, while Fig. 5(c) and (d) show
an interesting pattern of the transition between isolated (top-left
region) and a fractal-like clustering of coreshell micelles (bot-
tom-right region). Motivated by these ndings, we calculated the
fractal dimension of these images in the next section. The presence
of both sol and gel structures, and the presence of a region display-
ing a transition between both in Fig. 5(d) possibly indicates that
the RTFOT process is an intermediate aging step which is in half-
way between solgel transition.
4.2. Fractal analysis
The fractal dimension D
f
indicates the amount of space occupied
by an object considering its degree of roughness and irregularity at
different scales, and should exhibit fractional values. The method
used to calculate the fractal dimension D
f
of the AFM phase images
is the box-counting dimension. In this method, a binary image is
covered with squares of a certain size using the minimum number
of squares to completely cover the object [36]. This process contin-
ues repeatedly with smaller squares. The box-counting dimension
is then given by:
D
f
lim
L!0
log NL
log1=L
5
Fig. 4. Different morphological features observed in RTFOT aged lms. In one region, the phase images (a and b) shown micelles exhibit a coreshell like structure. The lateral
size of these images is respectively 10 lm
2
and 50 lm
2
. In another region, the topography (c) and phase (d) images displayed morphologies compatible with the gel model of
Yen [34].
L.M. Rebelo et al. / Fuel 117 (2014) 1525 19
Author's personal copy
where N(L) is the number of squares of size L needed to cover the
image [38,37,36]. The box-counting dimension is quite similar to
Hausdorff dimension, which considers the number N(r) of balls of
radius r nedeed to cover the object. When r is small, N(r) is large,
i.e., as r approaches zero, 1/r
d
is larger, where d is the Hausdorff
dimension [38,37,39]. Both methods result in the same value for
many shapes, but there are some exceptions [38,37]. Fig. 6 shows
reconstructed versions of Fig. 5(d) with both box-counting and
Hausdorff methods. In this particular case, we obtained a fractal
dimension of 1.9007 and 1.8738, respectively. Several other images
were reconstructed which leaded to D
f
values ranging between
[1.8271 1.9127] with both methods.
Several studies report that asphaltenes suffer precipitation and
form fractal-like structures during phase separation process
[4042]. Using X-ray scattering at low angles (SAXS), several
authors suggest that clustered asphaltene structures in solvents
Fig. 5. Phase images of RTFOT aged AC. Images (A), (B), (C) and (D) have the respective lateral dimensions: 5, 10, 25 and 50 lm. Image (D) exhibits an interface between a
region composed of micelles and a region composed of fractal structures.
Fig. 6. Reconstruction of phase image of Fig. 5(D) with both Hausdorff and box-counting methods. The fractal dimension of the micelle structure is 1.8738 and 1.9007,
respectively.
20 L.M. Rebelo et al. / Fuel 117 (2014) 1525
Author's personal copy
(e.g. toluene and bitumen) are consistent with structures
exhibiting with fractal dimension D
f
2 [4345]. Besides that,
Raghunathan et al. determined the Hausdorff fractal dimension
of asphaltene polymers as ranging between 1.6 and 2.0 [46].
Fig. 7. Friction maps of (a) pure, (b) RTFOT aged AC, and (c) PAV aged AC. The vertical scale (in Volts) measures the cantilever torsion. All maps were measured with a lateral
scanning frequency of 8 Hz.
0.3 0.2 0.1 0 0.1 0.2 0.3
0
0.5
1
1.5
2
x 10
5
C
O
U
N
T
PURE
0.3 0.2 0.1 0 0.1 0.2 0.3
0
0.5
1
1.5
2
x 10
5
C
O
U
N
T
RTFOT
0.3 0.2 0.1 0 0.1 0.2 0.3
0
0.5
1
1.5
2
2.5
x 10
5
C
O
U
N
T
PAV
VOLTAGE (V)
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
PURE RTFOT PAV
negative peak
positive peak
VOLTAGE (V)
VOLTAGE (V)
(a) (b)
(c) (d)
Fig. 8. Histograms constructed from the friction images of (a) pure, (b) RTFOT aged AC, and (c) PAV aged AC. Those histograms were tted with a double gaussian curves
(black lines). The r
2
values of each dataset are r
2
pure
0:999; r
2
RTFOT
0:987, and r
2
PAV
0:996. (d) Average torsion of the cantilever (in Volts) in the maltene phase (blue) and in
the micelles (red). (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)
L.M. Rebelo et al. / Fuel 117 (2014) 1525 21
Author's personal copy
Therefore we can conclude that our fractal analysis of aged ACs is
in agreement with other results in the literature.
4.3. Friction analysis
Fig. 7 shows representative friction images of pure, RTFOT and
PAV aged samples obtained with the lateral force mode of the
AFM. The images represent the raw AFM friction data without
any enhancement. The horizontal lines in the top region of the
maps are due to the fast lateral scanning which is necessary to
avoid adhesion of the tip. The friction maps show some similarities
with the phase images discussed previously: (i) the micelle struc-
tures are readily visible in this AFM mode, (ii) the size of the mi-
celles increase with aging, and (iii) the image contrast of the
pure AC is larger compared to the ones of aged samples. Since
the contrast is the difference between the lowest and highest can-
tilever torsion (measured in Volts), we conclude that the friction
coefcient between the cantilever tip and sample is larger for un-
aged samples. The similar image contrast between aged samples
reveal that they have similar friction coefcients. The voltage scale
of the images span from negative to positive values, corresponding
to the torsion experienced by the cantilever which is recorded by
the photodetector on its left and right quadrants, respectively.
The more distant the value is from zero (untorsioned cantilever),
the highest the friction between tip and sample surface. Based on
the images of Fig. 7, we clearly see that positive voltages are asso-
ciated to the surface of the micelles, while negative voltages repre-
sent the surface of maltene phase.
In our friction analyses we produced 27 maps (nine maps for
each sample type). From those images we constructed friction his-
tograms to provide insights about the overall effect of aging on the
friction coefcient of ACs. These data are shown in Fig. 8. The his-
tograms are composed of two broad peaks: a peak at negative volt-
ages corresponding to the overall friction response of the maltene
phase, and a peak at positive voltages describing the friction distri-
bution of the micelles. The histogram of pure AC is composed of
0 0.2 0.4 0.6 0.8 1 1.2
0
50
100
150
200
C
O
U
N
T
SLOPE
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 0.2 0.4 0.6 0.8 1 1.2
0
20
40
60
80
100
120
140
C
O
U
N
T
SLOPE
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 0.2 0.4 0.6 0.8 1 1.2
0
50
100
150
200
250
C
O
U
N
T
SLOPE
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
PURE AC
10Hz
RTFOT AC
10Hz
PAV AC
10Hz
(a) (b)
(c) (d)
(e) (f)
Fig. 9. Slope maps of pure (a), RTFOT (c) and PAV AC lms (e). Their respective histograms are shown in (b), (d) and (f).
22 L.M. Rebelo et al. / Fuel 117 (2014) 1525
Author's personal copy
two symmetric peaks and the cantilever torsion spans from0.3 V
to + 0.3 V. The histogram of RTFOT also exhibit two symmetric
peaks, but the cantilever torsion spans from 0.2 V to +0.2 V. As
for the PAV histogram, the peaks are no longer symmetric, with
the peak at positive voltages being nearly twice as large as com-
pared to the negative peak. Besides that, the cantilever torsion
spans from0.2 V to +0.2 V. From these observations, it is possible
to conclude that the aging processes affect the (i) friction coef-
cient of the samples, and the (ii) relative composition of the mi-
celles and maltene phase. The rst observation can be better
visualised by tting all histograms with a double gaussian curve
and tracking the center of each peak. This is shown in Fig. 8(d).
The average cantilever torsion in the micelles reduces from
+0.1 V in pure AC to +0.05 V in RTFOT aged AC, which represents
a 50% of reduction of the friction coefcient. Besides that, the
counts of data points remains nearly the same, meaning that the
size of the micelles grows only slightly in the short term aging.
From RTFOT to PAV, the cantilever torsion remains unchanged
(+0.05 V), but the count of data points increases strongly. This
means that the long term aging does not noticeably change the
friction coefcient, but strongly increases the percentage of the mi-
celles at the expense of a reduction in the maltene phase. As for the
evolution of the maltene phase with aging, the cantilever torsion
reduces from 0.1 V to 0.05 V in the short term aging, while
the long term aging induces a nearly zero torsion in the cantilever.
We remark that the presence of image artefacts due to fast scan-
ning do not change the above analysis because their only effect
is an small increase in the mean width of the peaks in the histo-
grams of Fig. 8.
4.4. Microrheology analysis
The slope of the force curve is a dimensionless quantity that
provides quantitative information about the mechanical nature of
the sample. In our case, it indicates if the sample is more or less
deformable. A map representative of each sample and their respec-
tive slope histograms of approach curves is shown in Fig. 9. These
histograms show mainly two peaks in their distributions. The left
peak corresponds to the slope on the AC surface. The right peak
centered around slope 1 (indicating an innitely rigid surface)
corresponds to the region of the substrate used for calibration.
The maps in Fig. 9 reveals that the aging process increases the
stiffness of AC lms. For example, the slope on top of the unaged
AC lm ranges between 0.2 0.6, and its average slope is around
0.3. For the RTFOT aged AC, the main slope contribution ranges be-
tween 0.5 0.7, and its average slope is around 0.6. For the PAV
aged AC, the main slope contribution ranges between 0.5 0.8,
and its average slope is approximately 0.7. The maps also show
that lms of unaged and PAV aged ACs exhibit a pretty uniform
slope distribution, while the RTFOT aged lm exhibits slope inho-
mogeneities in the surface. This qualitatively agrees with the fact
that different AC components may contribute differently for the
macroscopic rheological properties of the material. We remark that
the micellar structures shown in Figs. 37 could not be spatially re-
solved in the slope maps because of the reduced resolution
(32 32) and low maximum force employed in the FV
measurements.
We have also mapped the elasticity modulus E and apparent
viscosity g over the AC lms (three different locations in three dif-
10
9
8
7
6
5
l
o
g
1
0
[
E
(
P
a
)
]
4 6 8
1
2 4 6 8
10
2 4
FREQUENCY (Hz)
PURE AC
PAV AC
RTFOT AC
8
7
6
5
4

l
o
g
1
0
[
(
P
a
.
s
)
]
4 6 8
1
2 4 6 8
10
2 4
FREQUENCY (Hz)
PURE AC
PAV AC
RTFOT AC
5
4
3
2
1
0
A
D
H
E
S
I
O
N

W
O
R
K

(
f
J
)
4 6 8
1
2 4 6 8
10
2 4
FREQUENCY (Hz)
PURE AC
PAV AC
RTFOT AC
60
50
40
30
20
10
A
D
H
E
S
I
O
N

F
M
I
N

(
n
N
)
4 6 8
1
2 4 6 8
10
2 4
FREQUENCY (Hz)
PURE AC
PAV AC
RTFOT AC
(a) (b)
(c) (d)
Fig. 10. Average elasticity modulus E (a) and apparent viscosity g (b) as a function of the vertical scan frequency (which is proportional to the indentation speed) of unaged
and aged AC lms. The dispersion bars were determining by averaging nine maps (with resolution of 32 32 data points) for each sample. Average adhesiveness of AC lms
estimated with two different methods: (c) work done by the adhesive forces, and (d) minimum adhesive force.
L.M. Rebelo et al. / Fuel 117 (2014) 1525 23
Author's personal copy
ferent lms for each sample) similarly to what have been done
with the slope in Fig. 9. The average values of E and g are both
shown in Fig. 10. We have changed the vertical scan frequencies
to investigate the response of the lms with varying indentation
speeds. For all frequencies, the unaged AC is always less stiff than
the aged AC. Between aged ACs, the PAV one is approximately half-
order of magnitude stiffer than the RTFOT one, and almost one or-
der of magnitude stiffer than the unaged AC. The dispersion bars
have amplitude of one order of magnitude, indicating that the AC
lms are indeed composed of inhomogeneous components that
contribute differently to the rheological properties of the bitumen.
The average apparent viscosity of the lms is shown in
Fig. 10(b). As an overall trend, the apparent viscosity decreases
more than one order of magnitude with a modest increase of fre-
quency. The unaged AC is half-order of magnitude less viscous than
the aged samples. The aged samples exhibit similar apparent vis-
cosities except for the indentation frequency of 0.5 Hz, for which
the RTFOT aged AC and the unaged AC exhibit similar apparent vis-
cosities. As in the case of the elasticity modulus, the amplitude of
the dispersion bars is very large, reinforcing the idea that the di-
verse structures of the AC lm contribute differently to the macro-
scopic properties of bitumen.
The reason why E increases while g decreases with the indenta-
tion speed can be explained in terms of the Deborahs number De
dened as De = t
c
/t
p
where t
c
refers to the stress relaxation time
and t
p
refers to the time scale of observation [47]. t
c
is material
dependent and in the case of AC, it is of the order of a minute
[48]. t
p
is related to the inverse vertical scan frequency. Therefore,
Deborahs number increases with an increase in frequency. At high
De, the deformation properties are dominated by elasticity exhibit-
ing solid-like behavior and reduced viscosity. At low De, materials
properties are predominantly viscous exhibiting uid-like behav-
ior. In that sense, it does not sound unreasonable that the apparent
viscosity of RTFOT aged AC is only slightly larger than for unaged
AC at 0.5 Hz, since RTFOT causes a short term aging of AC, and both
should exhibit similar relaxation times t
c
.
Finally, Fig. 10(c) and (d) shows the effect of the aging processes
in the adhesion properties of AC calculated with the two methods
described previously. We noticed that the minimal adhesion force
increases with aging. Taking into account both the absolute and
work value of the adhesion force, the PAV aged AC exhibits en-
hanced adhesiveness properties as compared to RTFOT and pure
ACs. However, the adhesiveness response presented a complicated
dependence with the indentation speed which requires further
investigation.
5. Conclusions
In this work we AFM methods to investigate the micro-mor-
phology and micro-rheology of one type of bitumen before and
after aging, induced with RTFOT and PAV methods. Morphologi-
cally, we have found that AC lms are mainly composed of asphal-
tene micelles dispersed in a hydrocarbon matrix. In unaged
samples, the micelles structures cannot be resolved in the topo-
graphic images, only in phase imaging. For aged samples the micel-
lar structure is readily detected in both topographic and phase
imaging. The aging process induces a growth of the average sizes
of the micelles which are in sol form in the unaged samples. The
growth of those micelles seems to reach a regime in which they be-
come structured in a network of connected micelles with open
voids (occupied by the hydrocarbon matrix) for long-term aging
(PAV). The short-term aging (RTFOT) lies in between, where we
found fractal-like micellar structures with fractal dimension D
f
ranging between [1.8271 1.9127]. Those values are comparable
with other works in the literature describing fractal structures
made of asphaltenes [4346]. The phase images also show that
the hydrocarbon matrix exhibits higher loss modulus compared
to the asphaltene micelles in unaged samples. While in aged ones,
we observed the opposite trend. The friction imaging revealed that
the short term AC aging reduces the average friction coefcient by
50% in both micelle and maltene phases, while the long term aging
does not cause any further reduction in the friction coefcient. The
friction images conrmed that the aging causes an increase in the
of micelle phase with a concomitant reduction in the fraction of
maleness. Besides that, we were able to demonstrate that this in-
crease occur mainly in the long term aging. The nano-indentation
experiments performed in FV mode shown that the aging pro-
cesses cause a stiffening of the AC lm (half-order of magnitude
for RTFOT process, and one order of magnitude for PAV process).
The aging process also increased the apparent viscosity of the AC
lms by half-order of magnitude.
Acknowledgment
LMR, JSS, JMF and JBS acknowledge the nancial support from
Brazilian National Research Council (CNPq).
References
[1] Zhang HL, Yu JY, Feng ZG, Xue LH, Wu SP. Effect of aging on the morphology of
bitumen by atomic force microscopy. J Microscopy 2011;246:11.
[2] Ribeiro EA, Ferreira WLG, Branco VTFC, Soares SA, Soares JB, Mateos A.
Avaliao da resistncia ao dano por umidade em misturas asflticas por meio
de mtodos adotados no Brasil e na Espanha. Rev Paviment 2011;20:19.
[3] Masson JF, Price T, Collins P. Dynamics of bitumen fractions by thin-layer
chromatography/ame ionization detection. Energy Fuel 2001;15:955.
[4] Masson JF, Perc SB. Solventeless ngerprinting of bituminous materials: a
high-resolution thermogravimetric method. Thermochim Acta 2005;436:35.
[5] Masson JF, Leblond V, Margeson J. Bitumen morphologies by phase-detection
atomic force microscopy. J Microscopy 2006;221:17.
[6] Le Guern M, Chailleux E, Farcas F, Dreessen S, Mabille I. Physico-chemical
analysis of ve hard bitumens: identication of chemical species and
molecular organization before and after articial aging. Fuel 2010;89:3330.
[7] Allen R, Little D, Bhasin A. Structural characterization of micromechanical
properties in asphalt using atomic force microscopy. J Mater Civ Eng
2012;24:1317.
[8] Lesueur D. The colloidal structure of bitumen: consequences on the rheology
and on the mechanisms of bitumen modication. Adv Colloid Interface Sci
2009;145:42.
[9] Soares SA, Lucena MCC, Soares JB. Characterization and thermal behavior of
polymer-modied asphalt. Mater Res 2004;7:529.
[10] Pamplona TF, Amoni BC, Alencar AEV, Lima APD, Ricardo NMPS, Soares JB, et al.
Asphalt binders modied by SBS and SBS/nanoclays: effect on rheological
properties. J Braz Chem Soc 2012;23:639.
[11] Iwan sk M, Mazurek G. Hydrated lime as the anti-aging bitumen agent.
Procedia Eng 2013;57:424.
[12] Apeagyei AK. Laboratory evaluation of antioxidants for asphalt binders. Constr
Build Mater 2011;25:47.
[13] Wu S, Han J, Pang L, Yu M, Wang T. Rheological properties for aged bitumen
containing ultraviolet light resistant materials. Constr Build Mater
2012;33:133.
[14] Ouyang C, Wang S, Zhang Y, Zhang Y. Improving the aging resistance of the
styrene-butadiene-styrene copolymer modied asphalt by addition of
antioxidants. Polym Degrad Stab 2006;91:795.
[15] Siddiqui MN, Ali MF, Shirokoff J. Use of X-ray diffraction in assessing the aging
pattern of asphalt fractions. Fuel 2002;81:51.
[16] Davison RR, Glover CJ, Burr BL, Bullin JA. Size exclusion chromatography of
asphalts. In: Wu Chi-san, editor. Handbook of size exclusion chromatography
and related techniques. Revised and expanded. CRC Press; 2003 [Print ISBN:
978-0-8247-4710-7].
[17] Loeber L, Sutton O, Morel J, Valleton J-M, Muller G. New direct observations of
asphalts and asphalt binders by scanning electron microscopy and atomic
force microscopy. J Microscopy 1996;182:32.
[18] Zare-Shahabadi A, Shokuhfar A, Ebrahimi-Nejad S. Microstructure and
properties of nanoclay reinforced asphalt binders. Defect Diffus Forum
2010;579:297.
[19] Hesp SAM, Iliuta S, Shirokoff JW. Reversible aging in asphalt binders. Energy
Fuel 2007;21:1112.
[20] Claudy P, Ltoff JM, Rondelez F, Germanaud L, King GN, Planche JP, et al. A
new interpretation of time dependent physical hardening in asphalt based on
DSC and optical thermoanalysis. In: ACS symposium on chemistry and
characterization of asphalts. Washington, DC; 1992.
24 L.M. Rebelo et al. / Fuel 117 (2014) 1525
Author's personal copy
[21] Forbes A, Haverkamp RG, Robertson T, Bryant J, Bearsley S. Studies of the
microstructure of polymer-modied bitumen emulsions using confocal laser
scanning microscopy. J Microscopy 2001;204:252.
[22] Sengoz B, Isikyakar G. Analysis of styrene-butadiene-styrene polymer
modied bitumen using uorescent microscopy and conventional test
methods. J Hazard Mater 2008;150:424.
[23] Pauli AT, Branthaver JF, Robertson RE, Grimes W. Atomic Force Microscopy
investigation of SHRP asphalts. In: Proc Symp on heavy oils and residue
compatibility and stability, 221st national meeting. Division of Petroleum
Chemistry, San Diego, California, USA: American Chemical Society; 2001.
[24] Lee S-J, Amirkhanian Serji N, Shatanawi K, Kim KW. Short-term aging
characterization of asphalt binders using gel permeation chromatography
and selected Superpave binder tests. Constr Build Mater 2008;22:2220.
[25] Bahia HU, Anderson DA. The pressure aging vessel (PAV): a test to simulate
rheological changes due to eld aging. Special technical publication 1241,
Philadelphia: ASTM; 1994.
[26] Binard C, Anderson D, Lapalu L., Planche JP. Zero shea viscosity of modied and
unmodied binders. In: Proc. 3rd Eurasphalt and Eurobitume congress. 2004.
p. 1721.
[27] Cumpson PJ, Clifford CA, Hedley J. Quantitative analytical atomic force
microscopy: a cantilever reference device for easy and accurate AFM spring-
constant calibration. Meas Sci Technol 2004;15:1337.
[28] Mate CM, McClelland GM, Erlandsson R, Chiang S. Atomic-scale friction of a
tungsten tip on a graphite surface. Phys Rev Lett 1987;59:1942.
[29] Magonov SN. Surface analysis with STM and AFM. Experimental and
theoretical aspects of image analysis. VCH; 1996.
[30] Radmacher M. Studying the mechanics of cellular processes by atomic force
microscopy. In: Methods in cell biology. Cell mechanics, vol. 83. Elsevier; 2007.
[31] Radmacher M. Measuring the elastic properties of biological samples with the
AFM. IEEE Eng Med Biol Mag 1997;16:47.
[32] Santos JAC, Rebelo LM, Araujo AC, Barros EB, de Sousa JS. Thickness-corrected
model for nanoindentation of thin lms with conical indenters. Soft Matter
2012;8:4441.
[33] Rebelo LM, de Sousa JS, Mendes Filho J, Radmacher M. Comparison of the
viscoelastic properties of cells from different kidney cancer phenotypes
measured with atomic force microscopy. Nanotechnology 2013;24:055102.
[34] Yen TF. Asphaltene/resin plus oil interconversion: an investigation into
colloidal model of asphaltenes. In: Proc. workshop the chemical components
and structure of asphaltic materials. Rome, Italy; 1991. p. 10.
[35] Wu S, Pang L, Mo L, Chen Y, Zhu G. Inuence of aging on the evolution of
structure, morphology and rheology of base and SBS modied bitumen. Constr
Build Mater 2009;23:1009.
[36] Bushell GC, Yan YD, Woodeld D, Raper J, Amal R. On techniques for the
measurement of the mass fractal dimension of aggregates. Adv Colloid
Interface Sci 2002;95:1.
[37] Alligood KT, Sauer TD, Yorke JA. Chaos: an introduction to dynamical
systems. New York: Springer-Verlag; 1996.
[38] Ott E. Chaos in dynamical systems. Cambridge University Press;
1994.
[39] Hausdorff F. Dimension und aueres ma. Math Ann 1918;79:157.
[40] Sirota EB. Physical structure of asphaltenes. Energy Fuel 2005;19:1290.
[41] Rahmani NHG, Dabros T, Masliyah JH. Fractal struture of asphaltene
aggregates. J Colloid Interface Sci 2005;285:599.
[42] Henriques CB, Winter A, Koroish ET, Filho RM, Bueno MIMS. Estudo da
inuncia dos particulados no fenmeno de agregao dos asfaltenos por
espectroscopia de varredura ptica. Quim Nova 2011;34:424.
[43] Roux JN, Broseta D, Demsans B. Study os asphaltene aggregation:
concentration and solvent quality effects. Langmuir 2001;17:5085.
[44] Fenisteins D, Barr L, Broseta D, Espinat D, Livet A, Roux JN. Viscosimetric and
neutron scattering study of asphaltene aggregates in mixed toluene/heptane
solventes. Langmuir 1998;14:1013.
[45] Espinat D, Rosenberg E, Scarsella M, Barre L, Fenisteuin D, Broseta D. Colloidal
structure evolution from stable to occulated state of aspkatene solutions and
heavy crudes. In: Mullins O, Sheu EY, editors. Structures and dynamics of
asphaltenes. New York: Plenum Press; 1998. p. 145201.
[46] Raghunathan P. Evidence for fractal dimension in asphaltene polymers
from electron-spin-relaxation measurements. Chem Phys Lett 1991;182:
331.
[47] Reiner M. The deborah number. Phys Today 1964;17:62.
[48] Dourado ER, Simo RA, Leite LFM. Mechanical properties of asphalt binders
evaluated by atomic force microscopy. J Microscopy 2010;245:119.
L.M. Rebelo et al. / Fuel 117 (2014) 1525 25