You are on page 1of 12

Construction and Building Materials 255 (2020) 119332

Contents lists available at ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Molecular dynamics simulation of distribution and adhesion of asphalt


components on steel slag
Jinzhou Liu, Bin Yu, Qianzhe Hong
School of Transportation, Southeast University, Sipailou #2, Nanjing 210096, China

h i g h l i g h t s

 Steel slag increases the concentration of each component of asphalt on mineral surface.
 Coulomb interaction dominates adhesion for strongly alkaline minerals.
 Steel slag has the higher adhesive capacity with asphalt, following by limestone and granite.

a r t i c l e i n f o a b s t r a c t

Article history: Steel slag, as an industrial waste, is a potential alternative to replace aggregate in asphalt mixture, but the
Received 10 January 2020 interaction mechanism between asphalt and steel slag is still not understood. In this study, asphalt-steel
Received in revised form 6 April 2020 slag interfacial model was established to carry out molecular dynamics simulation to measure the inter-
Accepted 22 April 2020
action of the two materials. The radial distribution function and relative concentration were defined to
Available online 27 April 2020
illustrate the aggregations of asphalt components on steel slag under both dry and humid conditions.
And the weakly alkaline and acidic minerals were also used as controls. The results show resin in asphalt
Keywords:
is more likely to gather around weakly alkaline (limestone) and strongly alkaline (steel slag) minerals
Molecular dynamics
Concentration distribution
while the opposite is true for acid minerals (granite). For acidic and weakly alkaline minerals, Van der
Adhesion Waals effect accounts for most of the adhesion while for strongly alkaline minerals, Coulomb interaction
Steel slag dominates. Steel slag has the highest adhesive capacity, following by limestone and granite under both
Asphalt components dry and humid conditions. The interfacial strength of asphalt and minerals is tightly relevant to the chem-
ical compositions of both aggregate and asphalt. The interaction mechanisms of steel slag and asphalt at
the molecular level help evaluating steel slag and selecting suitable asphalt in field application.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction slag. Ahmedzade P and Sengoz B [2] evaluated the mechanical


characteristics of four different asphalt mixtures. They believed
Asphalt pavements are widely used due to the high strength, the rough texture of steel slag is beneficial for the performance
good stability and wear resistance, however the constructions con- of asphalt concrete. And the electrical conductivity of steel slag is
sume large amount of high-quality aggregate. With the shortage of superior to that of limestone mixture.
aggregate resources and increase in transportation costs, it is nec- Regarding the interfacial bonding, numerous studies have pro-
essary to find alternative materials. As an industrial by-product, ven asphalt and steel slag have good adhesion properties. Xie
steel slag has the characteristics of excellent wear resistance and and Chen [3] introduced a new type of interface tensile test,
anti-slip performance. The embedded frame formed by particles demonstrating slag and asphalt display better bonding strength
is exceedingly stable in road applications and the available studies than conventional road building materials. Shen [4] proposed the
have verified the feasibility of replacing ordinary aggregates with surface pores of slag and the chemical reaction of steel slag and
steel slag during road use. Wu and Xue [1] applied steel slag asphalt enhance the adhesion of steel slag and rubberized asphalt.
instead of aggregates in stone mastic asphalt (SMA) mixtures then For this purpose, they carried out tensile and net adsorption tests
evaluated their performances. The results showed both the low and observed the microstructure of the interfacial transition zone
and high temperature performances are improved by adding steel via scanning electron microscope (SEM). The chemical reaction
mechanism of steel slag and rubberized asphalt was also clarified
E-mail address: yb@seu.edu.cn (B. Yu) by Fourier Transform Infrared Spectrometer. Liu [5] proved that

https://doi.org/10.1016/j.conbuildmat.2020.119332
0950-0618/Ó 2020 Elsevier Ltd. All rights reserved.
2 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

steel slag has good adhesion to asphalt due to the new organic sil- line minerals such as CaCO3 and acidic minerals such as SiO2
icon compound formed by the chemical reaction between asphalt were employed.
and ferric oxide. Meanwhile the mineral compositions of steel slag
react chemically with asphalt. Xiao [6] introduced an image analy-
2. Simulation details
sis method to quantitatively estimate adhesion of asphalt mortars
with aggregate using the area of asphalt coating identified by
2.1. Asphalt model
lmage J. The conclusions also prove steel slag has better moisture
damage resistance.
Materials Studio (MS), an Accelrys’ commercial software, was
Water damage is a common disease happening in asphalt
employed to simulate molecular dynamics of steel slag and differ-
pavements. With the generation of hydrodynamic pressure,
ent components of asphalt. The steel slag-asphalt interface system
water gradually reduces the bonding of interface. As a result,
model consists of steel slag and asphalt layer. To successfully sim-
asphalt detaches from the surface of mineral aggregate. Several
ulate an interface system, the atomic models of asphalt and steel
theories characterizing adhesion mechanism have been proposed
slag need to be constructed in the MS.
to describe the bonding between asphalt and minerals [7–13].
Asphalt is a compound containing thousands of complex com-
The widely accepted acid-base theory indicates the combination
ponents, mainly including hydrocarbons and functional groups
of acidic components in asphalt and alkaline components in
such as sulfur, nitrogen and oxygen, which makes it difficult to pre-
aggregate contribute to the adhesion, which is also consistent
cisely depict the chemical structure and predict the properties.
with the observed phenomenon [7–9]. Thermal analysis reveals
According to the American Society for Testing and Materials
that the combination of asphalt and minerals tends to reach an
(ASTM) D4124-09, asphalt is composed of four components:
energy stable state. The surface energy is actually energy changes
asphaltenes, resins, aromatics, and saturates. These four compo-
of different substances during the bonding process [10–13].
nents can be obtained by the selective adsorption-analysis method
Moreover, experimental and microscopic tests have been pro-
proposed by Corbett [40], which has been proved to be more in line
posed to probe the mechanism of interfacial adhesion, including
with the true asphalt properties. The asphalt model selected in this
SEM [14], X-ray computed tomography (CT) [15], atomic force
article mainly relies on the 12-component asphalt model proposed
microscopy (AFM) [16–19], dynamic mechanics analyzer (DMA)
by Li and Greenfield [41]. The model representations of the 12
[20], optical microscope (OM) image, and dynamic shear rheome-
components are shown in Figs. 1–4. Table 1 provides some con-
ter (DSR) [21]. These experimental methods have achieved some
stituent parameters of asphalt. The molecular structures of the
results in the measurement of asphalt-aggregate interface failure
above components are derived by petroleum analysis in sedimen-
behavior. However, adhesion involves the adsorption of asphalt
tary rocks [42–45].
by minerals, which is difficult to explain the physical adhesion
To build asphalt model, the component molecules were set with
that occurs at molecular scale by traditional macro tests and
an initial density of 0.1 g/cm3 in an empty box to eliminate the
micro tests.
accumulation effect between asphalt components. The smart
Molecular Dynamics (MD) is an emerging technique to describe
deceleration method was used to optimize the initial structure
interactions between substances. It applies the principles of New-
by 5000 iterations, afterward 200 ps MD simulation was performed
tonian mechanics, and simulation data can be adopted to avoid
under constant pressure and temperature (NPT) ensemble to fur-
tedious experimental testing methods. MD simulation can reveal
ther obtain a balanced molecular structure with a time step of
the energy dissipation process of the molecular compounds chang-
1 fs, where the pressure was 1.0 atm and the temperature was
ing with space and time. It can interpret interface features or fail-
set to 298 K. The optimized asphalt structure is shown in Fig. 5.
ure behaviors on a molecular scale. Recently, researchers have used
MD simulations to unveil the connection between the chemical
compositions and physical as well as mechanical behavior of 2.2. Steel slag surface model
asphalt mixtures [22–26], including thermodynamic properties
such as density and solubility parameters, oxidative aging X-ray diffraction experiments of steel slag were conducted.
[27,28], interfacial interaction and adhesion properties [29–34]. It Through X-ray diffraction analysis in Fig. 6 and literature review
is indicated that MD simulation is feasible in studying the adhesion [3,5,39], it was found steel slag is mainly composed of tricalcium
of asphalt and mineral aggregates. silicate (C3S). The cell of C3S is shown in Fig. 7. To simulate steel
A review of literatures reveal the surface interaction investiga- slag, the C3S unit cell was cleaved to obtain a (0 0 1) surface. The
tions between steel slag and asphalt are mainly focused on selection of Miller plane (0 0 1) was established by evaluating
macroscopic performance testing [35–37] and material surface the influence of different Miller planes on the simulation results
properties [3–6,38,39]. The distribution and adhesion mechanism of interface adhesion. The simulation values from the three Miller
of asphalt components on steel slag during the combination pro- surface ((0 0 1), (0 1 0), (1 0 0)) display slight differences [27]. In
cess is rarely concerned. Therefore, this study applied MD simula- addition, the Miller plane (0 0 1) is the most stable surface [32].
tion to establish models for characterizing asphalt-steel slag After geometric optimization, the crystal surface was enlarged to
interfacial behaviors. The absorption differences between asphalt form a super unit cell by constructing a repeating unit in x and y
components and steel slag were quantitively determined. The directions. A vacuum slab was included on top of the extended
article is mainly divided into the following sections: firstly, MD molecules surface to generate a three-dimensional periodic bound-
simulation details were elaborated and an atomic interface model ary model.
of asphalt-steel slag was established; secondly, by comparing the
density, and other thermodynamic parameters of asphalt model, 2.3. Steel slag-asphalt system model
the accuracy of the selected force field and parameters were ver-
ified; thirdly, MD simulation was carried out to observe the dis- The Amorphous cell module was used for system assembly. A
tribution and adhesion of various components of asphalt on vacuum layer of 50 Å was added above the asphalt layer to prevent
steel slag. The moisture sensitivity of the interface was also con- the influence of periodic conditions. After assembly, the energy-
sidered; and lastly, the adhesion mechanism of asphalt and steel minimization function of Forcite module was used to obtain the
slag was summarized at molecular scale. To fully understand the most stable phase of the system based on smart descent method.
difference between steel slag and conventional aggregates, alka- The most stable structure of the system is shown in Fig. 7(a).
J. Liu et al. / Construction and Building Materials 255 (2020) 119332 3

Phenol

Py r r ol e Thiophene
Fig. 1. Molecular representation of asphaltene components. (We use gray to indicate carbon, white to indicate hydrogen, red to indicate oxygen, blue to indicate nitrogen, and
yellow to indicate sulphur). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

PHPN DOCHN
Fig. 2. Molecular representation of aromatic components.

Xh 2 3 4
i
The interface modeling method was similar when considering Etotal ¼ k2 ðb  b0 Þ þ k3 ðb  b0 Þ þ k4 ðb  b0 Þ
moisture effect. The moisture sensitivity of the interface was con- b
sidered by adding 200 water molecules between steel slag surface Xh i
þ k2 ðh  h0 Þ2 þ k3 ðh  h0 Þ3 þ k4 ðh  h0 Þ4
and asphalt layer [27,28,33,34], which corresponds to 10% mass
h
fraction of 12-component asphalt. The optimized steel slag- X
water-asphalt interface model is shown in Fig. 7(d). Besides, silica þ ½k1 ð1  cos/Þ þ k2 ð1  cos2/Þ þ k3 ð1  cos3/Þ
/
and limestone were also considered as two ordinary types of X X ; ;
X
aggregates. þ k2 v2 þ kðb  b0 Þðb  b0 Þ þ kðb  b0 Þðh  h0 Þ
v ;
The key of molecular simulation is to determine the type of b;b b;h
X
force field. The force field, also called potential energy, is a function þ ðb  b0 Þ½k1 cos/ þ k2 cos2/ þ k3 cos3/
that is related to the coordinates of each molecule in the Cartesian b;/
coordinate system. The structure of a molecule will change with X
þ ðh  h0 Þ½k1 cos/ þ k2 cos2/ þ k3 cos3/
time and space during its movement. In other words, the force field h;/
is a function connecting the potential energy and molecular config- X X
urations. The Condensed-Phase Optimized Molecular Potentials for þ kðh;  h;0 Þðh  h0 Þ þ kðh  h0 Þðh;  h;0 Þcos/
b;h h;h;/
Atomistic Simulation Studies (COMPASS) force field is widely 2 !9 !6 3
exploited for covalent compounds and has been tried to analyze X qi qj X r0ij r0ij
the interface of asphalt mixtures [30,33]. The potential functions þ þ eij 42 3 5 ð1Þ
i;j
r ij i;j
rij rij
contained in COMPASS force field are [47]:
4 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

Quinolinohopane Thio-isorenieratane

Benzobisbenzothiophene Pyridinohopane

Trimethylbenzene-oxane
Fig. 3. Molecular representation of resin components.

that the adsorption contribution of asphaltene and resin generated


by hydrogen bonding on quartz surface was negligible (less
than1%).
The combination of organics and inorganics provides possibility
to investigate the interfacial interaction in hybrid systems with
COMPASS force field. Moreover, COMPASS has been successfully
utilized for asphalt and mineral such as quartz [27,29–33,50–52],
Squalane calcite [27,32,33], and C3S [53,54]. Therefore, the COMPASS force
field was employed for calculation.
A MD run with constant volume and temperature (NVT) ensem-
ble for 500 ps was conducted at 298 K with a time step of 1 fs. The
initial velocity was randomly determined by system. During simu-
lation, the Nose-Hoover-Langevin (NHL) thermostat was employed
in the interface model to provide a stable temperature in set con-
ditions. The Van der Waals force was calculated by the atom-based
Hopane summation method and the electrostatic force was determined
through the Ewald summation method. To reach the balance of
Fig. 4. Molecular representation of saturate components.
obtaining the dynamic trajectories as accurate as possible and
improving the computational efficiency, we set the calculation
The COMPASS field includes bonding interactions (bonding range of the potential energy function as 15.5 Å meanwhile the
stretch (b), angular bonding (h), torsion angle(/), out-of-plane dynamic trajectory data was outputted every 500 steps. According
angle (v), and cross-coupling items) and non-bonded interactions to the literatures, the setting conditions of the simulation parame-
(electrostatic effect represented by Coulombic equation and vdW ters have been verified to be adequate to obtain the most stable
forces represented by Lennard–Jones function) [47,48]. The COM- energy system [27].
PASS field is the first molecular force field that unifies organic
molecular systems and inorganic molecular systems while they
were treated separately in previous work. However, the classical 3. Results and discussion
force fields have limitations. One of the limitations is they assume
that breakage/formation of bonds does not occur at interface. Sim- 3.1. Verifications of force field parameters
ilarly, the total charge of system remains constant during simula-
tion [26]. In addition, the energy of hydrogen bond is not In the light of the significance of density in distinguishing
accurately defined in COMPASS. Whereas Murgich [49] illustrated asphalt characteristics, dynamic simulation of asphalt molecules
J. Liu et al. / Construction and Building Materials 255 (2020) 119332 5

Table 1
Characteristics of 12-component AAA-1 asphalt model [41,46].

SARA Molecules Number in model Molecular representation Molar mass (g/mol) Solubility fractions (%)
Model Experiments
Asphaltene Phenol 3 C42H54O 575.0 16.50 16.20
Pyrrole 2 C66H81N 888.5
Thiophene 3 C51H62S 707.2
Aromatic PHPN 11 C35H44 464.8 30.60 31.80
DOCHN 13 C30H46 406.8
Resin Quinolinohopane 4 C40H59N 554.0 38.10 37.30
Thio-isorenieratane 4 C40H60S 573.1
Benzobisbenzothiophene 15 C18H10S2 290.4
Pyridinohopane 4 C36H57N 530.9
Trimethylbenzene-oxane 5 C29H50O 414.8
Saturate Squalane 4 C30H62 422.9 10.70 10.60
Hopane 4 C35H62 483.0

values with the previous literature. The simulated density values


are consistent with previous experimental ones.
Cohesive energy density (CED) is a measurement of bonding
between molecules. It is the total energy required to move a mole-
cule outside the range of intermolecular attraction. CED also gives
an expression for intermolecular interactions. Since the inter-
molecular interactions in asphalt are achieved by electrostatic
forces and van der Waals effect, CED and the solubility parameters
can be separated into two components, as shown in equations (2)
and (3).

CED ¼ d2 ð2Þ

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
d¼ d2v dw þ d2ele ð3Þ
Fig. 5. Asphalt model for the interface system (asphaltene: blue; aromatic: green;
resin: yellow; saturate: pink). (For interpretation of the references to colour in this
where CED is cohesive energy density, d is solubility parameter;
figure legend, the reader is referred to the web version of this article.) dvdW and dele represent van der Waals effect and electrostatic inter-
actions between all atoms in asphalt binder model.
From Table 2, the simulation values fall within the range given
with a step of 1 fs at a temperature of 298 K and a pressure of in the literatures. For the asphalt model selected, the van der Waals
1.0 atm under NPT ensemble was performed to verify whether force component is much larger than electrostatic interaction ones,
the asphalt and the force field selected above can yield precise sim- this means that the adhesion of asphalt molecules is mainly pro-
ulation values. The Berendsen barostat acts as a pressure con- vided by van der Waals forces.
troller. The density of asphalt model was calculated after 200 ps As a typical viscoelastic material, asphalt exhibits different rhe-
simulation and Table 2 lists the comparison of the simulation ological behaviors at different temperatures: glassy state, vis-

1500
3 1 - CaO
5 2 - Ca(OH)2
2 3 - CaCO3
2
1200
5 4 - Ca2SiO4
6 5 - Ca3SiO5
6 - Ca2Fe2O5
Intensity (CPS)

900
4 7 - Fe2O3
5
66 1
3 4
600 5 7
4 5 4 4
51 2 3 6 6 5 53
6 54 76 7 1 46 5
3
300 33 6

0
10 20 30 40 50 60
2θ (degree)

(a) XRD patterns (from experiments) (b)Cell model


Fig. 6. XRD patterns and cell model of steel slag.
6 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

Fig. 7. Interfacial model. (a) asphalt-steel slag. (b) asphalt-SiO2. (c) asphalt-CaCO3. (d) asphalt-water-steel slag. (e) asphalt-water-SiO2. (f) asphalt-water-CaCO3.

Table 2 significance. The NPT ensemble simulation at temperatures of


Density, CED and d of asphalt model. 263 K, 273 K, 278 K, 283 k, 288 K, 293 K, 298 K, 308 K, 318 K,
Properties Simulation Previous experiment or and 328 K were carried out while the other simulation conditions
values simulation values [55] remained unchanged. The change of specific volume with temper-
Density (298.15 kg/cm3) 1.004 1.01–1.04 ature is plotted in Fig. 8.
CED (108J/m3) 3.29 3.19–3.32 The specific volume exhibits an increasing trend as temperature
d ((J/cm3)1/2) 18.139 13.30–22.50 rises in Fig. 8. At low temperatures, the specific volume increase
dvdW ((J/cm3)1/2) 17.854 N/A
rate is significantly lower than that at high temperatures, which
dele ((J/cm3)1/2) 1.257 N/A
corresponds to the two different viscoelastic states of asphalt
molecule. The two different regions were fitted separately to
obtain the intersection. The fitting results indicate the glass transi-
coelastic state and viscous flow. The glass transition temperature tion temperature of the selected asphalt is about 295.3 K, which
Tg is characterized by a critical temperature during the change of corresponds to the range given in the literature [56]. Therefore,
viscoelastic state of asphalt with temperature. Asphalt exhibits thanks to the consistency of the significant parameters of simula-
high modulus brittle state below glass transition temperature tion results with the previous literatures, the developed 12-
while viscoelastic state above. When the temperature is lower than component asphalt model was transferred to the COMPASS force
Tg, asphalt will display larger stiffness and less prone to deform, field for further analysis.
which corresponds to the high modulus of asphalt. When the tem-
perature is higher than Tg, the stress and strain of asphalt under 3.2. Distribution of asphalt compositions on steel slag
load behave evident time-temperature dependence. Tg is consid-
ered closely relevant to low temperature cracking. Therefore, the Fig. 9 shows the temperature and energy changes during the
glass transition temperature has a considerable physicochemical simulation. The energy fluctuation is less than 5% after 500 ps sim-
J. Liu et al. / Construction and Building Materials 255 (2020) 119332 7

around weakly alkaline minerals. This is due to the charge effect


Specific volume
1.015 of weakly alkaline minerals and the polarity of resin. The distribu-
Fitting line of specific volume
tions of the remaining components are more uniform.
In Fig. 10 (b), saturate was firstly adsorbed and accumulated at
Specific volume (cm*3/g)

1.010
the position of 10 Å in the asphalt-SiO2 interface system. According
1.005 Glass transition area to previous research [57], the interaction between acidic minerals
and asphalt is so weak that it is difficult to affect the aggregation
state of asphalt near the minerals. The results in the diffusion rate
1.000
of the asphalt are approximately consistent with the molecular
weights of the components. The order of the first peak in Fig. 10
0.995 (b) is: saturate, asphaltene, resin and aromatic, which further con-
firms the reliability of this conclusion.
Tg point 295.3K
0.990 In Fig. 10 (c), due to the strong alkaline properties of C3S, the
peak of resin was also found firstly near 6.8 Å from the mineral sur-
260 270 280 290 300 310 320 330 face in the asphalt-C3S interface system, followed by asphaltene
Temperature (K) with a continuous wave peak near 11 Å. It is worth noting the val-
ues of the two peaks in the asphalt-C3S model are significantly lar-
Fig. 8. Changes of specific volume with temperature.
ger than the corresponding ones in the CaCO3 and SiO2 systems.
Meanwhile the distances between asphalt and mineral are smaller,
ulation, which means the system reached equilibrium. To charac- indicating the charge interaction between steel slag and asphalt
may be greater so that more asphalt molecules are attracted.
terize the molecular aggregation state of asphalt component near
steel slag, the radial distribution function (RDF) of 4 components The impacts of water on the adhesions of asphalt to mineral
were also simulated, with the results plotted in Fig. 11. The aggre-
were exported from the trajectories. RDF is a measure of the distri-
bution probability of other particles around the reference particle, gations of the four components are more uniform in moist condi-
tions and moisture delays the appearance distance of the g(r) peak.
reflecting the aggregation of the selected molecule with the speci-
fied reference molecule. Assuming that the number of molecules in In the asphalt-water-CaCO3 interface system, the peak of resin
with a value of 1.06 does not appear until 15.6 Å, much smaller
the range r~r + dr around the reference molecule in the system is
dN, the radial distribution function is defined as: than the value in the moisture-free state. The aggregation state
of the four components is consistent with CaCO3 for the asphalt-
water-SiO2 interface system. The g (r) peak of saturates is delayed
qgðrÞ4pr2 ¼ dN ð4Þ
from the initial 10 Å to 28 Å. It can be deduced water greatly
expand the distance between saturate and SiO2, which may be
where q is the interface density, given the total molecules amount
due to the lighter molecular mass of saturate. And the g(r) peaks
in the system is N. Then:
of the other components also decrease in different amplitude. It
Z 1 Z N
is encouraging the aggregation of asphalt components is not signif-
qgðrÞ4pr2 dr ¼ dN ¼ N ð5Þ icantly affected by moisture in the asphalt-water-C3S interface sys-
0 0 tem. The peak g (r) of resin decreases from 2.25 at 6.8 Å to 1.87 at
The RDFs of asphalt components were analyzed from the last 11.5 Å; the peak g (r) of asphaltene postpones from 2.24 at 11 Å to
200 ps based on the NVT simulation. The uppermost atom on min- 2.06 at 14.6 Å. Benefiting from the strong charge attraction, the
eral surface was selected to be reference particles. In Fig. 10 (a), the peak g (r) of asphalt component is still much higher than CaCO3
distribution of resin is remarkably different from other compo- and SiO2, which demonstrates the resistance of steel slag to water
nents in the asphalt-CaCO3 interface system. A ‘‘chubby” peak damage is stronger than conventional weak alkaline and acidic
was observed at 10 Å, which means resin is more likely to gather minerals.

450 25000
Potential energy
Kinetic energy
20000 Non-bond energy
Potential energy (kcal/mol)

400
Total energy
Temperature (K)

15000
350

10000

300
5000

250
0

200 -5000
0 100 200 300 400 500 0 100 200 300 400 500

Time (ps) Time (ps)

(a)Temperature (b) Energies


Fig. 9. Temperature and energy fluctuations in MD simulations.
8 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

3.0 3.0 3.0


asphaltene asphaltene
asphaltene
resin resin
2.5 2.5 2.5
resin
saturate saturate
aromatic saturate
aromatic
aromatic
2.0 2.0 2.0
g(r)

g(r)
g(r)
1.5 1.5 1.5

1.0 1.0 1.0

0.5 0.5 0.5

0.0 0.0 0.0


0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
r (Å) r (Å) r (Å)

(a) Asphalt-CaCO3 (b) Asphalt-SiO2 (c) Asphalt-C3S


Fig. 10. RDFs of 4 asphalt components in different models.

3.0 3.0 3.0


asphaltene asphaltene asphaltene
resin resin resin
2.5 2.5 2.5 saturate
saturate saturate
aromatic aromatic aromatic
2.0 2.0 2.0
g(r)

g(r)

g(r)
1.5 1.5 1.5

1.0 1.0 1.0

0.5 0.5 0.5

0.0 0.0 0.0


0 10 20 30 40 50 0 10 20 30 40 50 0 10 20 30 40 50
r (Å) r (Å) r (Å)

(a) Asphalt-water-CaCO3 (b) Asphalt-water-SiO2 (c) Asphalt-water-C3S


Fig. 11. RDFs between 4 asphalt components and aggregate surface in different models.

The particles tend to be directional during movement. In previ- the saturate component with the lowest mass fraction has its first
ous studies, mean squared displacement (MSD) was widely peak at 45.5 Å and the relative concentration is about 4.81%. Resin
adopted for analyzing the diffusion of asphalt components is more easily adsorbed by minerals, providing interfacial
[28,31,52]. However, MSD merely reflects the speed of diffusion adhesion.
so that it is difficult to characterize the distribution of asphalt com- In Fig. 12 (b), the concentration distribution of saturate in the
ponents in different directions. SiO2-asphalt interface is consistent with its diffusion law. A peak
The relative concentration of the asphalt components can be with a concentration of 6% first appears at 25.1 Å. Then following
obtained by calculating the atom density distribution within a unit by resin, a peak of 5.4% relative concentration appears at a similar
thickness slab parallel to the yz, zx and xy planes. The relative con- position of 23.7 Å. It is emphasizing the peak concentration of resin
centration of a set of atoms in a slab is calculated as follows: is lower than that in the CaCO3-asphalt interface. The distribution
of asphaltene and aromatic are relatively uniform, which indicate
RC i ¼ ci =cbulk ð6Þ
the bonding capacity of acid minerals and asphalt components is
weak. It is difficult to affect the distribution of asphalt components
ci ¼ N i =V i ð7Þ
through the adsorption of minerals on asphalt. In addition, resin
where ci and cbulk are slab and system concentrations, respectively; has a high activity in the diffusion process and is easier to adhere
V i is the slab volume; N i is the atom amount in slab. to either acidic or alkaline minerals.
Compared with RDF, the concentration distribution further pro- In Fig. 12 (c), the concentration of the four components on the
vides concentration values in different directions (such as x, y, and surface of steel slag is higher. As a strongly alkaline mineral, resin
z directions) near mineral surface, thus the difference in the disper- is also first adsorbed by steel slag. An 11% concentration peak of
sion of asphalt components in different directions can be revealed. resin appears at 24.0 Å; a 12% concentration peak of asphaltene
Relative concentration in Z directions was calculated based on the appears at 28 Å; the concentrations of aromatic and saturate also
dynamic trajectory. In Fig. 12 (a), the distribution of resin is con- show an increasing trend. The quantitative analysis of the compo-
centrated and uneven in the CaCO3-asphalt interface. A ‘‘narrow nents concentration explains the better adhesion of asphalt and
and high” peak with a 6% relative concentration appears at 25 Å steel slag.
away from the aggregate, indicating resin in the asphalt is more The concentrations of asphalt on mineral surfaces with water
likely to adhere to the surface of CaCO3 during diffusion process. were also analyzed. As revealed in Fig. 14, water makes the distri-
In contrast, the distributions of asphaltene, saturate, and aromatic bution of each component of asphalt more uniform.
are more uniform. The first peak of aromatic appears at 37 Å with a In Fig. 13 (a), water significantly reduces the peak concentration
relative concentration of 4.16%; the asphaltene shows a ‘‘broad and and appearance distance of resin in the CaCO3-water-asphalt inter-
fat” continuous peak at 36.8 Å with a relative concentration of 5%; face system. A concentration peak of 3.87% does not appear until
J. Liu et al. / Construction and Building Materials 255 (2020) 119332 9

8
8 asphaltene 14
asphaltene resin asphaltene
7
7 resin saturate 12 resin

Relative concentration (%)

Relative concentration (%)


Relative concentration (%)

saturate aromatic saturate


6
6 aromatic
aromatic 10
5
5
8
4 4
6
3 3

2 4
2

1 1 2

0 0 0
20 30 40 50 60 70 20 30 40 50 60 70 20 25 30 35 40 45 50 55 60
Z direction (Å) Z direction (Å) Z direction (Å)

(a) Asphalt-CaCO3 (b) Asphalt-SiO2 (c) Asphalt-C3S


Fig. 12. Relative concentration in Z direction.

12
8 8 asphaltene
asphaltene asphaltene
resin
7 resin 7 resin 10 saturate

Relative concentration (%)


saturate
Relative concentration (%)

Relative concentration (%)

saturate aromatic
6 aromatic 6 aromatic
8
5 5
6
4 4

3 3 4

2 2
2
1 1
0
0 0 20 25 30 35 40 45 50 55 60
20 30 40 50 60 70 80 20 30 40 50 60 70 80
Z direction (Å)
Z direction (Å) Z direction (Å)

(a) Asphalt-water-CaCO3 (b) Asphalt-water-SiO2 (c) Asphalt-water-C3S


Fig. 13. Relative concentration in Z direction with water.

Fig. 14. Asphalt-aggregate interface model at equilibrium. (a) asphalt-CaCO3; (b) asphalt-water-CaCO3; (c) asphalt-SiO2; (d) asphalt-water-SiO2; (e) asphalt-C3S; (f) asphalt-
water-C3S.
10 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

36.3 Å. The asphaltene peak is postponed to 40.4 Å with its peak W adhesion ¼ DEasphaltaggregate =A ð8Þ
concentration basically unchanged. The saturate shows a peak of
6.96% at 49 Å which can be interpreted water promotes the diffu- DEaspahtaggregate ¼ Etotal  ðEasphalt þ Eaggregate Þ ð9Þ
sion of light mass components. Nevertheless, the change in aro-
matic is not significant. For alkaline minerals, water reduces the where W adhesion is the adhesion work of asphalt and steel slag;
adhesion between asphalt and minerals via inhibiting the concen- DEasphaltaggregate is the bonding energy of interface; A is the area
tration of resin near the mineral surface. enclosed by the interface;Etotal is the whole potential energy of
In Fig. 13 (b), the first peak of saturate is delayed from 25.1 Å to interface; Easphalt and Eaggregate are the potential energies that makes
43.9 Å in the SiO2-water-asphalt interface, but the peak concentra- up asphalt molecules and steel slag molecules.
tion remains almost unchanged. The highest peak of resin with a The adhesion works are summarized in Fig. 15. The interfacial
concentration of 5.2% appears at 49.8 Å, indicating for acidic min- energy is composed of non-bond energy under either dry or humid
erals, water reduces the adhesion between asphalt and minerals by conditions. Interestingly, there is no bonding effect in the energy
increasing the distance. In other words, water is more easily composition of interface adhesion, that is, the reason for the inter-
adsorbed on acidic minerals surface, leading to the detaching of face adhesion is the interaction of non-bond terms [33]. Mean-
asphalt from the aggregate surface [58]. while the mineral types have an effect on W adhesion . Van der
In Fig. 13 (c), although the peak concentrations of resin and Waals effect constitutes the majority of the total energy composi-
asphaltene are reduced in the C3S-water-asphalt interface system, tion in the asphalt-SiO2 and asphalt-CaCO3 interface system. Less
compared with dry conditions, they still maintain a high level, with than 1% is made up by electrostatic interactions. With water exists,
a resin relative concentration of 6.4% at 24.6 Å, an asphaltene rel- the adhesion of asphalt to CaCO3 and SiO2 are weakened and van
ative concentration of 10.2% at 29.2 Å, an aromatic relative concen- der Waals energy is significantly reduced during the process. It is
tration of 8.3% at 34.5 Å, and a saturate relative concentration of reduced from 124.6 mJ/m2 to 28.1 mJ/m2 in the asphalt-CaCO3
10% at 37.1 Å. They are much higher than those at the same loca- interface and from 82.2 mJ/m2 to 16.2 mJ/m2 in the asphalt-SiO2
tion of CaCO3 and SiO2. The phenomenon of relative concentration interface.
is consistent with the conclusion obtained by RDF. As asphalt mix- Water increases the distance between asphalt and mineral com-
ture is subject to water erosion, the concentration of asphalt near pared to dry condition. Van der Waals force has a characteristic
the mineral surface gradually decreases, leading to a decline in that it decreases as the distance between molecules increases.
the bonding strength. Despite this, the adhesion of steel slag with The moisture thus weakens the interface adhesion. The adhesion
asphalt is still better than ordinary minerals such as granite and work under dry conditions is in a rank of steel slag (C3S) > limestone
limestone. (CaCO3) > basalt (SiO2), which is consistent with the experimental
This can provide researchers some suggestions when selecting findings. Steel slag as a strongly alkaline mineral, has the adhesion
different types of aggregates for mixing and paving asphalt pave- work of 2220.4 mJ/m2, which is much larger than weak alkaline
ments. Asphalt species can be optimized according to the type of (406.3 mJ/m2) and acid minerals (91.4 mJ/m2). Reflected in
aggregate, and the component proportions of asphalt can be appro- Fig. 15 (c), the W adhesion of steel slag interface is mainly composed
priately regulated within the allowable range, so as to provide mix- of Coulomb energy with a magnitude of 2159.9 mJ/m2. The remain-
tures with more excellent performance. ing 60.5 mJ/m2 comes from van der Waals energy, which only
accounts for a minor part of the total energy.
3.3. Adhesion of steel slag with asphalt The large amount of Ca2+ contained in steel slag has a strong
electrostatic interaction with asphalt. In Fig. 14 (e) and (f), asphalt
The durability and water damage resistance of asphalt mixture and steel slag are tightly bonded due to strong electrostatic inter-
depend on the adhesion provided by aggregate and asphalt. The action. The adhesion works under wet conditions are as follows:
asphalt-mineral interface structure after MD simulation is shown steel slag (C3S) > limestone (CaCO3) > quartz (SiO2), which are con-
in Fig. 14. The distance between steel slag and asphalt is the small- sistent with the conclusion under dry conditions. The W adhesion of
est, whereas this value is the largest for SiO2. The distance can C3S is 1998.6 mJ/m2, only being 9.99% lower than that in dry con-
directly reflect the degree of interaction, indicating steel slag has ditions. However, the adhesion works of CaCO3 and SiO2 are greatly
the strongest adhesion to asphalt while quartz has the weakest reduced by 76.70% and by 80.15%, respectively. It can be concluded
adhesion. To quantify the ability of steel slag combined with acidic minerals have the weakest adhesion to asphalt, followed by
asphalt, the adhesion work was calculated, as is defined as the weakly alkaline mineral, and strongly alkaline minerals have the
energy required by dividing the phases in the interface into two best adhesion to asphalt. The reason is wealthy positive charges
independent ones. The adhesion work is calculated by the follow- on the surface of strong alkaline minerals are more attractive to
ing equations. asphalt and meanwhile are less sensitive to humidity.

Fig. 15. Adhesion work of interface models at dry and humid conditions.
J. Liu et al. / Construction and Building Materials 255 (2020) 119332 11

Table 3
Interface interaction and debonding work.

Models DEasphalt-water (kcal/mol) DEaggregate-water (kcal/mol) DEasphalt-aggregate (kcal/mol) Wdebonding (mJ/m2)


Asphalt-CaCO3 8913.23 389620.03 377133.17 1660.63
Asphalt-SiO2 8597.08 59705.58 51023.79 39.37
Asphalt-C3S 9929.87 1022265.25 1001791.00 4899.85

Due to the hydrophobicity of asphalt and the hydrophilicity of while Coulomb interaction dominates for strongly alkaline
aggregate, water is prone to separate the two-phase interface. To minerals.
clarify the influence of aggregate chemical composition on interfa- (4) For both the dry and wet conditions, steel slag has the higher
cial moisture resistance, this study calculated the debonding work adhesive capacity with asphalt, following by limestone and
proposed in previous literatures [28,33,34]. The debonding work is granite. Moisture reduces the adhesion work of steel slag,
stipulated as the energy needed for water to detach the asphalt out limestone, and granite by 9.99%, 76.70%, and 80.15%,
of aggregate, as shown in Eqn. 10. respectively.
(5) The interface strength between asphalt and minerals against
wdebonding ¼ ðDEasphaltwater þ DEaggregatewater
water erosion is related to the chemical composition of
 DEasphaltaggregate Þ=A ð10Þ aggregate as well as the asphalt compositions.
where wdebonding is debonding work;DEasphaltwater is the interfacial
bonding energy of asphalt with water; DEaggregatewater is the interfa- CRediT authorship contribution statement
cial bonding energy of aggregate with water; DEasphaltaggregate is the
interfacial bonding energy of asphalt with aggregate; A is the area Jinzhou Liu: Software, Investigation, Data curation, Formal
enclosed by the interface. analysis, Validation, Writing - original draft, Writing - review &
From Table 3, The debonding works of the three interface sys- editing. Bin Yu: Conceptualization, Methodology, Writing - review
tems are all negative. There is a direct relationship between the & editing, Supervision, Funding acquisition. Qianzhe Hong:
debonding work with the type of aggregate. A negative debonding Resources, Software.
work indicates moisture does not need to absorb energy when the
asphalt is detached from aggregate, meaning the separation behav- Declaration of Competing Interest
ior of interface occurs naturally. The absolute value of debonding
work reflects the difficulty with which the interface system can The authors declare that they have no known competing finan-
be separated. The larger the absolute value, the stronger the ability cial interests or personal relationships that could have appeared
of interface to resist water damage. The Wdebonding of asphalt-SiO2, to influence the work reported in this paper.
asphalt-CaCO3 and asphalt-C3S are 39.37 mJ/m2, 1660.63 mJ/
m2, and 4899.85 mJ/m2, respectively. This indicates limestone Acknowledgements
is more resistant to water attack than acidic minerals. Steel slag
has the best anti-debonding ability due to its strong electrostatic This work was supported by National Natural Science Founda-
effect. In general, the resistance of interface water damage is not tion of China: [Grant Number 51878163]; Natural Science Founda-
only affected by the characteristics of asphalt but also by the chem- tion of Jiangsu Province: [Grant Number BK20171359].
ical composition of minerals.

References
4. Conclusions
[1] S. Wu, Y. Xue, Q. Ye, Y. Chen, Utilization of steel slag as aggregates for stone
In this research, the interaction of asphalt and steel slag was mastic asphalt (SMA) mixtures, Build. Environ. 42 (2007) 2580–2585, https://
doi.org/10.1016/J.BUILDENV.2006.06.008.
explored by molecular dynamics simulation. Limestone and gran-
[2] P. Ahmedzade, B. Sengoz, Evaluation of steel slag coarse aggregate in hot mix
ite were employed as controls. Different asphalt-mineral interface asphalt concrete, J. Hazard. Mater. 165 (2009) 300–305, https://doi.org/
models were established and verified. RDF as well as relative con- 10.1016/J.JHAZMAT.2008.09.105.
centration were calculated to evaluate the differences in adsorp- [3] J. Xie, Z. Chen, L. Pang, S. Wu, Implementation of modified pull-off test by UTM
to investigate bonding characteristics of bitumen and basic oxygen furnace
tion of asphalt components. The adhesion work was defined to slag (BOF), Constr. Build. Mater. 57 (2014) 61–68, https://doi.org/10.1016/J.
characterize the adhesive performance of minerals and asphalt CONBUILDMAT.2014.01.083.
under both dry and humid conditions. Throughout the research [4] A. Shen, C. Zhai, Y. Guo, X. Yang, Mechanism of adhesion property between
steel slag aggregate and rubber asphalt, J. Adhes. Sci. Technol. 32 (2018) 2727–
work, the following conclusions are obtained: 2740, https://doi.org/10.1080/01694243.2018.1507505.
[5] W. Liu, H. Li, H. Zhu, P. Xu, Properties of a steel slag-permeable asphalt mixture
(1) The resin in asphalt is more likely to gather around weakly and the reaction of the steel slag-asphalt interface, Materials 12 (2019),
https://doi.org/10.3390/ma12213603.
alkaline (limestone) and strongly alkaline (steel slag) miner- [6] Z. Xiao, M. Chen, S. Wu, J. Xie, D. Kong, Z. Qiao, C. Niu, Moisture susceptibility
als due to the charge action to provide good adhesion prop- evaluation of asphalt mixtures containing steel slag powder as filler, Materials
erties. The interaction between acid minerals and asphalt is 12 (2019), https://doi.org/10.3390/ma12193211.
[7] J.A.N. Scott, Adhesion and disbonding mechanisms of asphalt used in highway
weak so that the distribution of asphalt is relevant to the construction and maintenance, 47, 1978, pp. 19–48.
molecular weight of the components; [8] C.J. Van Oss, R.J. Good, M.K. Chaudhury, Additive and nonadditive surface
(2) Steel slag increases the concentration of each component of tension components and the interpretation of contact angles, Langmuir 4
(1988) 884–891, https://doi.org/10.1021/la00082a018.
asphalt on mineral surface. Even when subject to water ero-
[9] H. Visser, Interfacial forces in aqueous media, Powder Technol. 82 (1995) 209–
sion, the peak concentrations of resin and asphaltene are still 210, https://doi.org/10.1016/0032-5910(95)90005-5.
higher than those of granite and limestone; [10] E. Masad, C. Zollinger, R. Bulut, D. Little, R. Lytton, H. Khalid, R. Davis, T.
(3) The adsorption of minerals on asphalt is composed of non- Scarpas, E. Fini, A. Guarin, Characterization of HMA moisture damage using
surface energy and fracture properties, Asphalt Paving Technology:
bond interactions. For acidic and weakly alkaline minerals, Association of Asphalt Paving Technologists-Proceedings of the Technical
van der Waals effect constitutes the majority of the adhesion Sessions. 75 (2006) 713–754.
12 J. Liu et al. / Construction and Building Materials 255 (2020) 119332

[11] A. Bhasin, E. Masad, D. Little, R. Lytton, Limits on adhesive bond energy for [35] L. Ma, D. Xu, S. Wang, X. Gu, Expansion inhibition of steel slag in asphalt
improved resistance of hot-mix asphalt to moisture damage, Transport. Res. mixture by a surface water isolation structure, Road Mater. Pavem. Design
Record: J. Transport. Res. Board 2006 (1970) 2–13, https://doi.org/10.1177/ (2019), https://doi.org/10.1080/14680629.2019.1601588.
0361198106197000101. [36] Y. Zhang, X. Luo, I. Onifade, X. Huang, R.L. Lytton, B. Birgisson, Mechanical
[12] C. Miller, D. Little, A. Bhasin, N. Gardner, B. Herbert, Surface energy evaluation of aggregate gradation to characterize load carrying capacity and
characteristics and impact of natural minerals on aggregate-bitumen bond rutting resistance of asphalt mixtures, Constr. Build. Mater. 205 (2019) 499–
strengths and asphalt mixture durability, Transport. Res. Record: J. Transport. 510, https://doi.org/10.1016/j.conbuildmat.2019.01.218.
Res. Board 2267 (2012) 45–55, https://doi.org/10.3141/2267-05. [37] L. Liu, A. Zhou, Y. Deng, Y. Cui, Z. Yu, C. Yu, Strength performance of cement/
[13] A. Habal, D. Singh, Effects of warm mix asphalt additives on bonding potential slag-based stabilized soft clays, Constr. Build. Mater. 211 (2019) 909–918,
and failure pattern of asphalt-aggregate systems using strength and energy https://doi.org/10.1016/j.conbuildmat.2019.03.256.
parameters, Int. J. Pavement Eng. (2019) 1–13, https://doi.org/10.1080/ [38] M.J. Qazizadeh, H. Farhad, A. Kavussi, A. Sadeghi, Evaluating the fatigue
10298436.2019.1623399. behavior of asphalt mixtures containing electric arc furnace and basic oxygen
[14] L.D. Poulikakos, M.N. Partl, Investigation of porous asphalt microstructure furnace slags using surface free energy estimation, J. Cleaner Prod. 188 (2018)
using optical and electron microscopy, J. Microsc. 240 (2010) 145–154, https:// 355–361, https://doi.org/10.1016/j.jclepro.2018.04.035.
doi.org/10.1111/j.1365-2818.2010.03388.x. [39] Z. Chen, J. Xie, Y. Xiao, J. Chen, S. Wu, Characteristics of bonding behavior
[15] A. Bhasin, A. Izadi, S. Bedgaker, Three dimensional distribution of the mastic in between basic oxygen furnace slag and asphalt binder, Constr. Build. Mater. 64
asphalt composites, Constr. Build. Mater. 25 (2011) 4079–4087, https://doi. (2014) 60–66, https://doi.org/10.1016/J.CONBUILDMAT.2014.04.074.
org/10.1016/J.CONBUILDMAT.2011.04.046. [40] L.W. Corbett, Composition of asphalt based on generic fractionation using
[16] M. Guo, Y. Tan, J. Yu, Y. Hou, L. Wang, A direct characterization of interfacial solvent deasphaltening elution-adsorption chromatography and densimetric
interaction between asphalt binder and mineral fillers by atomic force characterization, Anal. Chem. 41 (1969) 576, https://doi.org/10.1021/
microscopy, Mater. Struct. 50 (2017), https://doi.org/10.1617/s11527-017- ac60273a004.
1015-9. [41] D.D. Li, M.L. Greenfield, Chemical compositions of improved model asphalt
[17] X. Yu, N.A. Burnham, R.B. Mallick, M. Tao, A systematic AFM-based method to systems for molecular simulations, Fuel 115 (2014) 347–356, https://doi.org/
measure adhesion differences between micron-sized domains in asphalt 10.1016/j.fuel.2013.07.012.
binders, Fuel 113 (2013) 443–447, https://doi.org/10.1016/j.fuel.2013.05.084. [42] T.B. Oldenburg, H. Huang, P. Donohoe, H. Willsch, S. Larter, High molecular
[18] H. Sui, G. Ma, Y. Yuan, Q. Li, L. He, Y. Wang, X. Li, Bitumen-silica interactions in weight aromatic nitrogen and other novel hopanoid-related compounds in
the presence of hydrophilic ionic liquids, Fuel 233 (2018) 860–866, https://doi. crude oils, Org. Geochem. 35 (2004) 665–678, https://doi.org/10.1016/J.
org/10.1016/j.fuel.2018.06.114. ORGGEOCHEM.2004.02.005.
[19] A.L. Lyne, V. Wallqvist, B. Birgisson, Adhesive surface characteristics of [43] M.P. Koopmans, J.W. De Leeuw, M.D. Lewan, J.S. Sinninghe Damste, Impact of
bitumen binders investigated by Atomic Force Microscopy, Fuel 113 (2013) dia- and catagenesis on sulphur and oxygen sequestration of biomarkers as
248–256, https://doi.org/10.1016/j.fuel.2013.05.042. revealed by artificial maturation of an immature sedimentary rock, Organic
[20] L. Mo, M. Huurman, S. Wu, A.A.A. Molenaar, Ravelling investigation of porous Geochem. 25 (1996) 391–426, https://doi.org/10.1016/S0146-6380(96)00144-
asphalt concrete based on fatigue characteristics of bitumen–stone adhesion 1.
and mortar, Mater. Des. 30 (2009) 170–179, https://doi.org/10.1016/J. [44] M.P. Koopmans, J.W. De Leeuw, J.S.S. Damsté, Novel cyclised and aromatised
MATDES.2008.04.031. diagenetic products of b-carotene in the Green River Shale, Org. Geochem. 26
[21] R. Moraes, R. Velasquez, H.U. Bahia, Measuring the effect of moisture on (1997) 451–466, https://doi.org/10.1016/S0146-6380(97)00025-9.
asphalt-aggregate bond with the bitumen bond strength test, Transp. Res. Rec. [45] L. Marynowski, M.J. Rospondek, R. Meyer zu Reckendorf, B.R. Simoneit,
2209 (2011) 70–81, https://doi.org/10.3141/2209-09. Phenyldibenzofurans and phenyldibenzothiophenes in marine sedimentary
[22] M. Guo, Y. Huang, L. Wang, J. Yu, Y. Hou, Using atomic force microscopy and rocks and hydrothermal petroleum, Organic Geochem. 33 (2002) 701–714,
molecular dynamics simulation to investigate the asphalt micro properties, https://doi.org/10.1016/S0146-6380(02)00040-2.
Int. J. Pavement Res. Technol. 11 (2018) 321–326, https://doi.org/10.1016/j. [46] Z. Du, X. Zhu, Molecular dynamics simulation to investigate the adhesion and
ijprt.2017.09.017. diffusion of asphalt binder on aggregate surfaces, Transport Res. Record: J.
[23] G. Xu, H. Wang, Molecular dynamics study of interfacial mechanical behavior Transport. Res. Board. (2019), https://doi.org/10.1177/0361198119837223.
between asphalt binder and mineral aggregate, Constr. Build. Mater. 121 [47] H. Sun, COMPASS: an ab initio force-field optimized for condensed-phase
(2016) 246–254, https://doi.org/10.1016/j.conbuildmat.2016.05.167. applications overview with details on alkane and benzene compounds, J. Phys.
[24] Y. Lu, L. Wang, Nanomechanics modeling of interface interactions in asphalt Chem. B 102 (1998) 7338–7364, https://doi.org/10.1021/jp980939v.
concrete: traction and shearing failure study, J. Multiscale Model. 10 (2019), [48] H. Sun, D. Rigby, Polysiloxanes: ab initio force field and structural,
https://doi.org/10.1142/s1756973718410044. conformational and thermophysical properties, Spectrochim. Acta Part A
[25] Y. Lu, L. Wang, Nanoscale modelling of mechanical properties of asphalt- Mol. Biomol. Spectrosc. 53 (1997) 1301–1323, https://doi.org/10.1016/S1386-
aggregate interface under tensile loading, Int. J. Pavement Eng. 11 (2010) 393– 1425(97)00013-9.
401, https://doi.org/10.1080/10298436.2010.488733. [49] J. Murgich, J.M. Rodríguez, A. Izquierdo, L. Carbognani, E. Rogel, Interatomic
[26] M. Shishehbor, M.R. Pouranian, R. Imaninasab, Evaluating the adhesion interactions in the adsorption of asphaltenes and resins on kaolinite calculated
properties of crude oil fractions on mineral aggregates at different by molecular dynamics, Energy & Fuels. 12 (1998) 339–343, https://doi.org/
temperatures through reactive molecular dynamics, Pet. Sci. Technol. 36 10.1021/ef9701302.
(2018) 2084–2090, https://doi.org/10.1080/10916466.2018.1531032. [50] Z. Xuefen, L. Guiwu, W. Xiaoming, Y. Hong, Molecular dynamics investigation
[27] Y. Gao, Y. Zhang, Y. Yang, J. Zhang, F. Gu, Molecular dynamics investigation of into the adsorption of oil-water-surfactant mixture on quartz, Appl. Surf. Sci.
interfacial adhesion between oxidised bitumen and mineral surfaces, Appl. 255 (2009) 6493–6498, https://doi.org/10.1016/j.apsusc.2009.02.021.
Surf. Sci. 479 (2019) 449–462, https://doi.org/10.1016/j.apsusc.2019.02.121. [51] G. Wu, L. He, D. Chen, Sorption and distribution of asphaltene, resin, aromatic
[28] G. Xu, H. Wang, Molecular dynamics study of oxidative aging effect on asphalt and saturate fractions of heavy crude oil on quartz surface: molecular dynamic
binder properties, Fuel 188 (2017) 1–10, https://doi.org/10.1016/ simulation, Chemosphere 92 (2013) 1465–1471, https://doi.org/10.1016/j.
j.fuel.2016.10.021. chemosphere.2013.03.057.
[29] G. Meng, T. Yiqiu, W. Jianming, Using molecular dynamics simulation to study [52] M. Huang, H. Zhang, Y. Gao, L. Wang, Study of diffusion characteristics of
concentration distribution of asphalt binder on aggregate surface, J. Mater. Civ. asphalt–aggregate interface with molecular dynamics simulation, Int. J.
Eng. 30 (2018) 4018075, https://doi.org/10.1061/(ASCE)MT.1943- Pavement Eng. (2019) 1–12, https://doi.org/10.1080/10298436.2019.1608991.
5533.0002258. [53] D. Tavakoli, A. Tarighat, Molecular dynamics study on the mechanical
[30] Z. Dong, Z. Liu, P. Wang, X. Gong, Nanostructure characterization of asphalt- properties of Portland cement clinker phases, Comput. Mater. Sci. 119
aggregate interface through molecular dynamics simulation and atomic force (2016) 65–73, https://doi.org/10.1016/j.commatsci.2016.03.043.
microscopy, Fuel 189 (2017) 155–163, https://doi.org/10.1016/ [54] W. Wu, A. Al-Ostaz, A.H.D. Cheng, C.R. Song, Computation of elastic properties
j.fuel.2016.10.077. of Portland cement using molecular dynamics, J. Nanomech. Micromech. 1
[31] M. Guo, Y. Tan, L. Wang, Y. Hou, Diffusion of asphaltene, resin, aromatic and (2011) 84–90, https://doi.org/10.1061/(ASCE)NM.2153-5477.0000026.
saturate components of asphalt on mineral aggregates surface: molecular [55] P. Wang, Z. Dong, Y. Tan, Z. Liu, Investigating the interactions of the saturate,
dynamics simulation, Road Mater. Pavement Des. 18 (2017) 149–158, https:// aromatic, resin, and asphaltene four fractions in asphalt binders by molecular
doi.org/10.1080/14680629.2017.1329870. simulations, Energy Fuels 29 (2015) 112–121, https://doi.org/10.1021/
[32] L. Chu, L. Luo, T.F. Fwa, Effects of aggregate mineral surface anisotropy on ef502172n.
asphalt-aggregate interfacial bonding using molecular dynamics (MD) [56] H.A. Tabatabaee, R. Velasquez, H.U. Bahia, Predicting low temperature physical
simulation, Constr. Build. Mater. 225 (2019) 1–12, https://doi.org/10.1016/ hardening in asphalt binders, Construct. Build. Mater. 34 (2012) 162–169,
j.conbuildmat.2019.07.178. https://doi.org/10.1016/j.conbuildmat.2012.02.039.
[33] Y. Gao, Y. Zhang, F. Gu, T. Xu, H. Wang, Impact of minerals and water on [57] D.N. Little, A. Bhasin, Using surface energy measurements to select materials
bitumen-mineral adhesion and debonding behaviours using molecular for asphalt pavement: Final Report for NCHRP Project, 2006.
dynamics simulations, Constr. Build. Mater. 171 (2018) 214–222, https://doi. [58] P. Mirzababaei, Effect of zycotherm on moisture susceptibility of Warm Mix
org/10.1016/j.conbuildmat.2018.03.136. Asphalt mixtures prepared with different aggregate types and gradations,
[34] G. Xu, H. Wang, Study of cohesion and adhesion properties of asphalt concrete Constr. Build. Mater. 116 (2016) 403–412, https://doi.org/10.1016/
with molecular dynamics simulation, Comput. Mater. Sci. 112 (2016) 161– j.conbuildmat.2016.04.143.
169, https://doi.org/10.1016/j.commatsci.2015.10.024.

You might also like