Fuel 300 (2021) 121015

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Innovative application of waste polyethylene terephthalate (PET) derived

additive as an antistripping agent for asphalt mixture: Experimental
investigation and molecular dynamics simulation
Rui Li a, Zhen Leng a, *, Jun Yang c, Guoyang Lu a, Man Huang b, Jingting Lan a,
Hongliang Zhang b, Yawei Bai d, Zejiao Dong e
a
Department of Civil and Environmental Engineering, The Hong Kong Polytechnic University, Hong Kong Special Administrative Region
b
Key Laboratory for Special Area Highway Engineering of Ministry of Education, Chang’an University, Xi’an, China
c
School of Transportation, Southeast University, Nanjing, China
d
Henan Province Highway Management Center of Toll and Loan, Zhengzhou, China
e
School of Transportation Science and Engineering, Harbin Institute of Technology, Harbin, China

A R T I C L E I N F O A B S T R A C T

Keywords: Moisture-induced damage represents one of the major challenges to the durability of asphalt pavement, which
Polyethylene terephthalate (PET) may lead to various premature distresses, such as cracking, ravelling and potholes. To reduce the moisture-
Moisture damage induced damages, antistripping agents have been commonly used in asphalt mixtures. On the other hand, the
Adhesion
fast-growing waste polyethylene terephthalate (PET) from the end-of-life plastic products poses a serious
Antistripping
Molecular dynamics simulation
problem to the environment. How to recycle and reuse the waste PET is becoming a major concern. This study
aims to develop a value-added recycling approach of reusing waste PET as an antistripping agent for asphalt
mixture. To achieve this objective, waste PET was first collected and went through an aminolysis reaction to yield
the amine-based PET additive, which was used to modify bitumen. The infrared spectrum of the PET additive was
then characterized, followed by the evaluation of moisture susceptibility through the boiling test and indirect
tensile strength (ITS) test. In addition, molecular dynamics (MD) simulation was conducted to analyze the effect
of the PET additive on the density and the cohesive energy density (CED) of the binder, and the interface bonding
between binder and aggregate at the molecular level. The results from both experiments and MD simulation
consistently indicate that the waste PET derived additive can effectively increase the resistance to moisture-
induced damage of asphalt mixture.

1. Introduction the mineral aggregates are hydrophilic, their affinity to water normally
exceeds that to bitumen [7–9]. Thus, the binder film around aggregate is
Asphalt pavements are subjected to various damages caused by prone to be replaced by water as water diffuses to the asphalt-aggregate
vehicle loads and environmental effects, among which, the moisture- interface.
induced damage is recognized as one of the most important factors To reduce the moisture susceptibility, additives such as liquid fatty
affecting the durability of pavements [1]. Moisture-induced damage amines and hydrated lime have been applied as the antistripping agents
often leads to stripping of asphalt mixture [2], as a result of the loss of for hot-mix asphalt [10,11]. Due to the convenience of application,
interface adhesion between bitumen and aggregate or the reduction of liquid amines are more favored [12]. Various studies have been con­
cohesion within asphalt mastic under the effect of water [3]. It has been ducted to investigate the effect of amine-based antistripping agents on
reported that moisture-induced damage could also accelerate premature moisture-induced damages of asphalt mixtures, and the results proved
failures of pavements such as cracking, ravelling and potholes [4,5]. The that these antistripping agents can effectively reduce the moisture sus­
potential of moisture-induced damage is largely dependent on the ceptibility [13–15].
physico-chemical properties of bitumen and aggregate [6]. As most of Polyethylene terephthalate (PET) is a plastic polymer, which is

* Corresponding author.
E-mail address: zhen.leng@polyu.edu.hk (Z. Leng).

https://doi.org/10.1016/j.fuel.2021.121015
Received 24 February 2021; Received in revised form 29 April 2021; Accepted 6 May 2021
Available online 15 May 2021
0016-2361/© 2021 Elsevier Ltd. All rights reserved.
R. Li et al. Fuel 300 (2021) 121015

commonly used to make plastic bottles. The fast-growing consumption Table 1

of plastic bottles has created a serious environmental problem in recent Basic properties of bitumen.
years. As the polymeric PET bottles are not bio-degradable, how to Penetration (25 ◦ C, 0.1 mm) Softening point (◦ C) Viscosity at 135 ◦ C (mPa⋅s)
recycle and reuse the waste PET bottles poses a major challenge to
64.5 48.5 477.5
environment. The application of waste PET in asphalt mixture provides a
possible solution to reuse this waste material. Due to the fact that the
melting point (~250 ◦ C) of PET is very high [16], it is not feasible to granite aggregate and mineral filler were obtained from a local supplier.
prepare a homogeneous PET modified bitumen with the wet process. The aggregate gradation for WC10 (a dense-graded wearing course
Instead, most previous studies used the dry process to prepare PET mixture with a nominal maximum aggregate size of 10 mm commonly
modified asphalt mixture, i.e., the PET flakes were added into asphalt used in Hong Kong) was used as shown in Fig. 1.
mixture directly during the mixing of asphalt binder and aggregate. It
was found that PET increased the fatigue life of asphalt mixtures to some 2.2. Aminolysis of waste PET
extent, while the resistance to moisture damage decreased [17–19],
which was mainly due to the poor adhesion between asphalt binder and The aminolysis of waste PET was carried out following the procedure
the PET flakes. On the other hand, chemical treatment of waste PET for as presented in Fig. 2. Prior to the aminolysis treatment, the PET bottles
the preparation of PET derived additive has been found to be a prom­ were first cleaned with normal detergent and then cut into small flakes,
ising alternative to the aforementioned dry process [20–22]. Recently, which were subsequently dried in an oven at 80 ◦ C for 2 h. A 500 ml
the aminolysis product of waste PET was employed to modify bitumen four-necked round flask equipped with a stirrer, reflux condenser and
by Leng, et al. [20,23], and the results highlighted that the PET derived nitrogen gas, was heated in an oil bath and charged with waste PET
additive significantly increased the adhesion between asphalt binder flakes and TETA. The weight ratio of PET to TETA was 1:2 based on a
and aggregate, therefore reducing the moisture-induced damage. previous study [20]. The aminolysis reaction of PET was conducted at
Nevertheless, the adhesion mechanism between the PET additive around 140 ◦ C for 2 h. After this treatment, the homogeneous product
modified asphalt binder and aggregate has not been fully understood was cooled down to room temperature and then filtrated with deionized
yet. Previous studies have indicated that molecular dynamics (MD) water for at least three times to remove the excess TETA and by-products
simulation can provide fundamental insights in understanding the (ethylene glycol). Finally, the substance remaining on the filter paper
interface adhesion between binder and aggregate from the molecular or was collected and dried in an oven at 60 ◦ C for 2–3 h to obtain the PET
atomic level [24,25]. It is therefore meaningful to explore the interface additive. Fig. 3 shows the chemical reaction between PET and TETA
performance between the PET additive modified binder and aggregate during aminolysis.
employing the MD simulation.
MD simulation has been widely used to study various properties of
asphalt binder and mixture, including the density, cohesion, and 2.3. Preparation of PET additive modified asphalt binder
particularly adhesion properties [25–28]. Xu and Wang [24] investi­
gated the interface adhesion between binder and aggregate using the To prepare the PET additive modified asphalt binder, approximately
bond energy parameters in both dry and wet conditions obtained from 500 g of bitumen was first heated to fluid condition at 150 ◦ C. The pre-
MD simulation. Their results indicated that both the aggregate and weighted PET additive was then added into the binder and stirred at a
binder types affect the interfacial adhesion. The study by Yao, et al. [29] speed of 500 rpm for 1 h. PET modified binders with two different
indicated that aggregate tends to affiliate more to water than bitumen, weight percentages of PET additive, 1 wt% and 3 wt% based on bitumen,
as the adhesion energy between aggregate and water is larger than that were prepared and denoted as PETB-1 and PETB-3, respectively. The
between aggregate and bitumen binder. Zhang, et al. [21] recently neat binder Pen 60/70 without PET additive was used for comparison.
verified that the glycolysis treated PET can improve the adhesion be­
tween binder and aggregate through MD simulation. The aminolysis 2.4. Experimental characterizations
product of waste PET showed strong antistripping ability in our previous
study [20], and the MD simulation can be applied to provide more in- Infrared spectra measurements were performed using a Bruker
depth understanding of the effect of PET additive on the interface Tensor II Fourier Transform Infrared (FT-IR) spectrometer. The IR
adhesion between binder and aggregate from the molecular perspective. spectrum of PET additive was measured from 600 cm− 1 to 4,000 cm− 1
The main objective of this research is to investigate the antistripping with a resolution of 4 cm− 1.
performance of the PET derived additive in asphalt mixture through The conventional bitumen binder tests of penetration at 25 ◦ C,
both experimental investigation and MD simulation. To achieve this
objective, the aminolysis treated waste PET additive was first used to 120
produce modified bitumen. The moisture susceptibilities of the PET
additive modified bitumen and asphalt mixture were then evaluated
100
experimentally through the boiling test and indirect tensile strength
Percentage passing (%)

(ITS) test. Finally, the MD simulation was implemented to evaluate the

80
density and cohesive energy density (CED) of the PET additive modified
binder, and the adhesion between the binder and aggregate in both dry
and wet conditions. 60

2. Materials and methods 40

2.1. Materials 20

In this study, the PET additive was obtained through aminolysis 0

treatment of waste PET. Waste PET bottles were collected after proper 0.01 0.1 1 10 100
identification. Triethylenetetramine (TETA) was purchased from Sigma- Size (mm)
Aldrich. Neat bitumen with a penetration grade of 60/70 (Pen 60/70)
was used, and its basic properties are presented in Table 1. Crushed Fig. 1. Gradation of aggregate.

2
R. Li et al. Fuel 300 (2021) 121015

Fig. 2. Aminolysis treatment of waste PET bottles to produce PET additive: (a) waste PET bottles, (b) PET flakes, (c) aminolysis reaction, and (d) the prepared PET
additive ready for use.

O O
NH
C C O CH2 CH2 O + H2N
n NH NH2

PET TETA

O O
NH NH C C NH NH

H2N NH NH NH2

PET additive
Fig. 3. Aminolysis reaction of PET. The PET flakes reacted with TETA at a temperature of 140 ◦ C for 2 h to prepare the PET derived additive.

softening point and rotational viscosity from 120 ◦ C to 175 ◦ C were representative molecule types were commonly used [32,33]. Since the
carried out following ASTM D5, D36 and D4402, respectively. objective of this study is to evaluate the effect of PET additive on the
The adhesion between bitumen and aggregates was evaluated antistripping performance of the binder, the three-component bitumen
through the boiling water test in accordance with ASTM D3625. Around model [34] was adopted for its simplicity. The three components are
400 g of coarse aggregates with the size of 5–10 mm were mixed with asphaltene (C64H52S2), 1,7-Dimethylnaphthalene (C12H22), and n-
bitumen for the boiling test. To prepare a layer of binder film with a docosane (n-C22H46), which represent asphaltene, naphthene aromatic
thickness of 8–10 µm, 2 wt% of binder based on aggregates was applied and saturate, respectively. Fig. 4 illustrates the molecular structures of
[6,30]. The mixtures were subsequently boiled for 10 min in a beaker. the three-component bitumen model and that of the PET additive.
Finally, the boiled mixtures were scanned with a normal office scanner
to obtain the images. 2.5.2. Bulk binder model
The moisture susceptibility of PET modified asphalt mixture was The molecular compositions of the neat bitumen and PET modified
then evaluated through the indirect tensile strength (ITS) test at 20 ◦ C bitumen (PETB) are presented in Table 2, the number of molecules were
according to ASTM D4867. The compacted asphalt mixture with an air selected based on previous studies [34–36]. The mass fractions of
void of 6–8% was first prepared with the Marshall compactor. Two asphaltene, 1,7-dimethylnaphthalene, and n-docosane in the neat
groups of specimens were fabricated and subjected to dry and wet bitumen are 20.7 wt%, 19.7 wt%, and 59.6 wt%, respectively. According
conditioning, respectively. The dry group specimens were conditioned to the experiment results, 1 wt% of PET additive could already achieve
at room temperature (23 ± 0.5 ◦ C) for 24 h, and the wet group specimens good antistripping performance, thus only 1 PET additive molecule was
were immersed in the 60 ◦ C water for another 24 h following the dry added to construct the PETB model, which corresponds to 1.97 wt% of
conditioning procedure. The ITS tests for the two groups of specimens PET additive based on bitumen.
were then conducted, and the tensile strength ratio (TSR) between the In this study, the commercial software BIOVIA Materials Studio was
wet conditioned strength and the dry strength were then calculated. used to perform the molecular dynamics (MD) simulation. The COM­
PASS (Condensed-phase Optimized Molecular Potentials for Atomistic
Simulation Studies) force field, developed using ab initio and empirical
2.5. Molecular dynamics simulation parameterization techniques by Sun [37], has been commonly applied to
describe the interactions of bituminous materials [38–40]. The Nose-
2.5.1. Molecular models Hoover-Langevin (NHL) thermostat and the Andersen barostat were
Bitumen, a by-product originating from the crude oil distillation, is employed throughout the simulation to control the temperature and
composed of various hydrocarbons. The main components of bitumen pressure, respectively. A cut-off distance of 15.5 Å was used to speed up
can be divided into asphaltenes, resins, aromatics, and saturates, based the computation. To build the bulk binder model, an amorphous cell
on their polarity and aromaticity [31]. Bitumen constitutes a wide range with an initial density of 0.1 g/cm3 in 3D periodical condition was
of different molecules whose structures are still not totally clear, which constructed first. After a geometry optimization process using the
makes it challenging to study this material using MD simulation. steepest descent algorithm, the system was then annealed for five cycles
Considering the complexity of bitumen, it is almost impossible to from 300 K to 500 K. Following a pre-equilibrium run for 100 ps with a
construct a molecular model that represents its accurate composition at time step of 1 fs under canonical ensemble (NVT), the system went
present. Instead, simplified models that are composed of 3–12

3
R. Li et al. Fuel 300 (2021) 121015

Fig. 4. Molecular structures of (a) n-docosane (n-

C22H46), (b) 1,7-dimethylnaphthalene (C12H12), (c)
Asphaltene (C64H52S2), and (d) PET additive
(C20H38O2N8). Note that the grey atoms represent
C, the white atoms represent H, the yellow atoms
represent S, the blue atoms represent N, and the
red atoms represent O. (Please refer to the online
version for the colors of the atoms). (For interpre­
tation of the references to colour in this figure
(a) n-docosane (n-C22H46) (b) 1,7-dimethylnaphthalene (C12H12) legend, the reader is referred to the web version of
this article.)

(c) Asphaltene (C64H52S2) (d) PET additive (C20H38O2N8)

Table 2
Molecular compositions of binder.
Sample-ID Asphaltene (C64H52S2) 1,7-dimethylnaphthalene (C12H12) n-docosane (n-C22H46) PET additive PET additive wt%

Neat bitumen 5 27 41 / /
PETB 5 27 41 1 1.97

through a 200 ps relaxation and shrank to a stable state with the first cleaved in the (1 0 0) surface, which was then used to build a
isothermal-isobaric (NPT) ensemble. Finally, another simulation run of supercell with the size of 35 Å × 35 Å × 23 Å (Fig. 6). Geometry opti­
100 ps was performed with the canonical ensemble to further equilibrate mization of the supercell was performed before constructing the
the system (Fig. 5). Unless additionally stated, the temperature and aggregate-bitumen interface. It was reported that the surface silicone
pressure were set as 298 K and 0.0001 GPa (1 atm), respectively, during atoms in siliceous aggregate are attached with hydroxyl groups (–OH)
the simulation. [41]. According to previous studies, the density of the surface hydroxyl
groups was set as 4.56 OH/nm2 [42,43].
2.5.3. Aggregate-bitumen interface model The aggregate-bitumen interface was built by placing a confined
Silica (SiO2), commonly used in the aggregate model for asphalt bitumen layer on top of the aggregate layer. To avoid the interaction
mixture simulation, was adopted in this study to represent aggregate. A across the mirror image in the z-direction, a vacuum layer of 30 Å was
unit silica model with the lattice parameters of a = b = 4.91 Å, c = 5.402 added above the bitumen layer. After geometry optimization, the
Å, α = β = 90◦ , and γ = 120◦ normally applied to build the siliceous interface was equilibrated at the canonical ensemble for 200 ps. Fig. 7
aggregate model was selected in this study. The unit silica model was shows the initial configuration and the equilibrated configuration of the
interface model in dry condition.
To evaluate the influence of water on the properties of binder-
aggregate interface, 200 water molecules were placed between the
aggregate layer and bitumen layer, based on the experience of the pre­
vious studies [24,44]. The interface model with water was then

Fig. 5. Bulk bitumen model after equilibrium. A total of 73 molecules are

included, containing 4026 atoms. (Please refer to the online version for the Fig. 6. Supercell of SiO2. The size of the SiO2 supercell is 35 Å×35 Å×23 Å, and
colors of the atoms). the surface silicon atoms are attached with hydroxyl groups –OH.

4
R. Li et al. Fuel 300 (2021) 121015

Fig. 7. Initial configuration (a) and equilibrated configuration (b) of the interface model in dry condition. A vacuum layer of Å was placed on top of the bitumen
layer to avoid interaction across the mirror image.

subjected to the same equilibrium procedure as described for the

0.35
interface model without water. Fig. 8 shows the initial configuration and
the equilibrated configuration of the interface model in wet condition.
0.30

3. Experimental results 1,636 1,547

0.25

Absorbsance (a.u.)
3.1. FT-IR test 0.20

Fig. 9 illustrates the FT-IR spectrum of the PET additive. It can be 0.15
observed that the distinct peak at 1,720 cm− 1 corresponding to the 3,320
carbonyl (–C– – O) stretching vibration of the ester group in PET mole­ 0.10
cules disappeared, while the two peaks at 1,636 cm− 1 and 1,547 cm− 1
corresponding to the characteristic carbonyl (–C– – O) stretching vibra­ 0.05

tion and N–H bending vibration of the amide group in PET additive,
respectively, emerged. In addition, the peak at 3,320 cm− 1 attributed to
0.00

the stretching vibration of the N–H in the amine group also showed up.
This result indicated the successful aminolysis of the waste PET by 4000 3000 2000 1000
TETA. -1
Wavenumber (cm )

3.2. Basic properties of binders Fig. 9. FT-IR spectrum of PET additive. The peaks at 1,636 cm− 1 and 1,547
cm− 1 were attributed to the carbonyl (–C–
–O) stretching vibration and N–H
Table 3 compares the values of penetration and softening point of the bending vibration of the amide group in PET additive, respectively, and the
base bitumen and the PET additive modified bitumens. It can be seen peak at 3,320 cm− 1 corresponds to the stretching vibration of the N–H in
that penetration decreased while the softening point increased slightly amine group.
with the incorporation of the PET additive. In addition, the binders with
1 wt% and 3 wt% of PET additive resulted similar values of penetration
Table 3
and softening point.
Penetration and softening point of binders.
The viscosity of PET additive modified binders as a function of
temperature from 120 to 175 ◦ C is shown in Fig. 10. It is observed that Sample-ID Penetration (25 ◦ C, 0.1 mm) Softening point (◦ C)

the viscosity increased a little bit with addition of PET additive at lower Pen 60/70 64.5 48.5
temperature (120 ◦ C). With the increase of temperature, the viscosity PETB-1 53.6 50.0
PETB-3 53.2 49.6
become similar for all the binders. In fact, the viscosity did not show
much difference with temperatures above 150 ◦ C. The underlying reason
is that the PET additive may form a crystal at ambient temperature [23], 3.3. Boiling test
due to the linearity of its molecule structure and the presence of amino
groups that leads to hydrogen bonds between the molecules. When The images of the binder coated aggregate after the boiling tests were
temperature increases, the additive melts and turns into liquid. Overall, analyzed to compare the moisture susceptibility of PET additive modi­
it is concluded from these tests that the PET additive does not have much fied asphalt binders. Fig. 11 presents the original scanned images after
influence on the basic properties of bitumen binder. boiling for 10 min and the corresponding binary images of the loose
asphalt mixtures. The binary images were obtained through image
analysis using MATLAB®, and the white area ratios representing the

Fig. 8. Initial configuration (a) and equilibrated configuration (b) of the interface model in wet condition. 200 water molecules were inserted between the aggregate
layer and bitumen layer.

5
R. Li et al. Fuel 300 (2021) 121015

aforementioned, the PET additive may form crystal structures. When the
concentration of PET additive increases, part of the additive can attract
with each other and the amount of PET additive at the bitumen-
aggregate interface may not change significantly. Therefore, the mois­
ture susceptibility of the asphalt mixture can be effectively reduced with
the incorporation of 1 wt% of PET additive.

3.4. ITS test

To further evaluate the effect of PET additive on the resistance to

moisture-induced damage of asphalt mixture, the ITS tests were con­
ducted on the compacted asphalt mixtures containing different dosages
of PET additive. Fig. 13 shows that the ITS values for the dry group were
very close, around 1,000 kPa. However, after conditioning in water at
60 ◦ C for 24 h, the ITS of the control asphalt mixture decreased sharply
to 665.9 kPa, while there is no significant change in the wet conditioned
ITS values for the PET modified asphalt mixtures. The TSR for the
Fig. 10. Viscosity measurement as a function of temperature. control asphalt mixture was around 60%, which increased dramatically
to over 90% with the incorporation of only 1 wt% of PET additive. With
percentages of the stripped aggregates were computed [45] as shown in further increase of the PET additive dosage, the TSR did not change
Fig. 12. What stands out in the graph is that the boiling loss area for the much. The fracture interface images of the samples in the wet-
control mixture was around 18.0%, which dropped considerably to conditioned group after the ITS test are presented in Fig. 14. It is clear
around 10.0% with only 1 wt% of PET additive. With further increasing that most of the Pen 60/70 bitumen was stripped off from the aggregate
of the PET additive dosage, the boiling loss area did not change much. As surface (Fig. 14(a)), representing a distinct adhesive failure. However,

Fig. 11. Original (left) and binary (right) images after boiling tests: (a) Pen 60/70, (b) PETB-1, and (c) PETB-3.

6
R. Li et al. Fuel 300 (2021) 121015

20 groups in PET additive can form the strong hydrogen bonds with the
hydroxyl groups at the aggregate surface, and 2) the alkaline amine
groups can also increase the interaction between the granite aggregate
and bitumen binder.
15
Boiling loss area (%)

4. MD simulation

10 4.1. Density

During the isothermal-isobaric (NPT) ensemble equilibration, the

volume of the bitumen system decreased gradually, and the density
5 increased accordingly. The density evolutions of the neat bitumen and
PET additive modified bitumen (PETB) are shown in Fig. 15. It can be
seen that the densities of both neat bitumen and PETB increased sharply
0 within the first 30 ps and then stabilized after 50 ps. Considering the low
content of the PET additive, no significant difference was observed be­
Pen 60/70 PETB-1 PETB-3
tween the densities of the two binders. The equilibrated densities
Fig. 12. Comparison of boiling loss area. (averaged over the last 50 ps) of the neat bitumen and PETB are 0.906 g/
cm3 and 0.907 g/cm3, respectively. The densities are consistent with
no significant stripping was observed in the asphalt mixtures containing those in the previous studies using the three-component bitumen model
PET additive, indicating that the failure occurred within the bitumen [24,34].
binder or mastic phase, which is typically cohesive failure. Such phe­
nomenon indicated that the PET additive effectively increased the 4.2. Cohesive energy density (CED) of binder
resistance to moisture-induced damage of asphalt mixture. The granite
aggregate is acid, and the base bitumen binder is also acidic because of The cohesive energy density (CED) is used to quantify the magnitude
the presence of carboxylic and/or sulfonic groups [46]. Thus, the of attraction between different molecules inside a substance. It is ob­
bonding between the granite aggregate and base bitumen binder is poor. tained from the difference of the total nonbond energy and the intra­
The PET additive contains amine groups, which is an alkaline additive. molecular nonbond energy, and the square root of CED is defined as the
The PET additive is expected to increase the bonding between the solubility parameter δ (Eq. (1)).
granite and bitumen binder for the following two reasons: 1) the amine √̅̅̅̅̅̅̅̅̅̅
δ = CED (1)

1200 120

Dry strength
1000 Wet strength
100
TSR

800 80
ITS (kPa)

TSR (%)

600 60

400 40

200 20

0 0
Pen 60/70 PETB-1 PETB-3
Fig. 13. ITS of asphalt mixtures with different concentrations of PET additive at 20 ◦ C (TSR represents the ratio between the wet tensile strength to the dry ten­
sile strength).

Fig. 14. Fracture interfaces for the asphalt mixtures that had been subjected to 24 h water conditioning at 60 ◦ C: (a) Pen 60/70, (b) PETB-1, and (c) PETB-3.

7
R. Li et al. Fuel 300 (2021) 121015

Neat bitumen PETB potential energies of the binder and the aggregate separated at the
equilibrium state, respectively, and A is the interface contact area.
1
Because of the hydrophilic nature of mineral aggregate, water is
0.9
more likely to be absorbed onto the aggregate surface. As a result,
0.8
asphalt binder is prone to detach from the aggregate in the presence of
0.7
Density (g/cm3)

water. The adhesion energy ΔEadhesion water and the work of adhesion
0.6
ΔWadhesion water in the presence of water are calculated using Eq. (4) and
0.5
Eq. (5), respectively. A lower value of ΔEadhesion water represents a lower
0.4
energy potential for water to displace the binder from the aggregate
0.3
surface [47], or the displacement is more difficult to happen and the
0.2
resistance to moisture-induced damage is stronger.
0.1
0 ΔEadhesion water = ΔEbit water + ΔEagg water − ΔEbit agg (4)
0 50 100 150 200 250
Time (ps) ΔWadhesion = ΔEadhesion (5)
water water /A

Fig. 15. Densities of neat bitumen and PETB during equilibrium. The
where ΔEbit water is the interaction energy between binder and water,
isothermal-isobaric (NPT) ensemble was implemented.
ΔEagg water is the interaction energy between aggregate and water, and
ΔEbit agg is the interaction energy between binder and aggregate. All the
Table 4 presents the computed CED of the neat bitumen and PETB.
energies were obtained at the equilibrium state.
With the incorporation of PET additive, the CED of PETB became larger
The energy ratio (ER), i.e., the ratio between the work of adhesion in
than that of the neat bitumen, indicating that PET additive increased the
dry condition and the work of adhesion in wet condition, is usually used
molecular interaction. The result could be explained by that the PET
to represent the moisture susceptibility of asphalt mixture [43,47]. A
additive is more polar than bitumen binder, thus both the van der Waals
larger value of ER indicates lower moisture sensitivity. The ER is given
force and electrostatic force between different molecules increased,
as,
leading to stronger cohesion of PETB.
ER = Wadhesion /Wadhesion water = ΔEadhesion /ΔEadhesion water (6)
4.3. Radial distribution
Table 5 shows the adhesion energy and work of adhesion between
binder and aggregate in the dry condition.ΔEvdw and ΔEelec represent the
Fig. 16 depicts the radial distribution of each component in the
energy difference associated with the van der Waals part and the electric
binder with respect to the aggregate surface. It can be seen that the
part, respectively. Since both bitumen and aggregate possess high sur­
concentration of 1,7-dimethylnaphthalene is the highest at the interface
face energy in isolated condition, when they attach to each other to form
for the neat bitumen, followed by n-docosane and asphaltene. The first
an interface, the total surface area becomes smaller, thus decreasing the
peak at around 2 Å in Fig. 16(b) belongs to the PET additive in PETB,
potential energy of the system. This is the reason why the interaction
indicating that the PET additive was the closest to the aggregate surface.
energies in the table are negative. It is clear to see that the absolute value
As demonstrated in Fig. 17, the amine groups in PET additive were very
of the adhesion energy ΔEadhesion and work of adhesion Wadhesion for PETB
close to the aggregate surface. Two reasons may contribute to such
and aggregate are larger than that of neat bitumen and aggregate,
phenomenon: (1) the higher polarity of PET additive makes it more
indicating that PET additive increased the interfacial adhesion between
affiliate to the polar aggregate surface, and (2) the amine groups in the
the binder and aggregate. Furthermore, it is observed that both the van
PET additive and the hydroxyl groups in the aggregate surface can form
der Waals and electric energy increased with the addition of PET addi­
hydrogen bond, which is much stronger compared with other non-
tive. The increase of interface adhesion can be mainly attributed to the
bonded interactions.
polarity of the PET additive and the formation of hydrogen bonding
between the PET additive and aggregate surface hydroxyl groups.
4.4. Work of adhesion between binder and aggregate Table 6 presents the adhesion energy in wet condition as well as the
calculated energy ratio. It can be noticed that the absolute value of
The work of adhesion Wadhesion is defined as the work required to ΔEagg water is larger than that of ΔEbit agg , which demonstrates that water
separate the binder from aggregate at the interface, which determines
is more likely to be absorbed to the aggregate surface than bitumen
the resistance to the interface detachment. A negative value of
binder due to the hydrophilic nature of the mineral aggregate. As shown
Wadhesion indicates attraction of the two parts, whereas a positive value
in Table 6, the ΔEadhesion water of the aggregate-water-PETB system is
indicates repulsion.
about 21% lower than that of the aggregate-water-NB system, indicating
The work of adhesion Wadhesion is given by Eq. (2), and the adhesion
that the coating of PETB on the aggregate is harder to be replaced by
energy ΔEadhesion is obtained from Eq. (3).
water compared with neat bitumen. On the contrary, the ER of the PETB
Wadhesion = ΔEadhesion /A (2) system reached 1.24, which is larger than that of the neat bitumen
system (0.88). The simulation results revealed that the moisture sensi­
ΔEadhesion = Etotal − (Ebinder + Eaggregate ) (3) tivity of the asphalt mixture was decreased by the incorporation of PET
additive, which is consistent with the findings obtained from the pre­
where Etotal is the total potential energy of the binder and aggregate ceding experiments.
interface system after equilibration, Ebinder and Eaggregate represent the

Table 4
Cohesive energy density (CED) of binder.
Sample-ID Van der Waals (106 J/m3) Electrostatic (106 J/m3) CED (106 J/m3) Solubility Parameter ((J/cm3)0.5)

Neat bitumen 326.4 − 1.221 332.4 18.232

PETB 334.5 0.610 342.6 18.509

Note: PETB represents the PET additive modified bitumen, in which 1 PET additive molecule was added into the bitumen model, corresponding to 1.97 wt% of the PET
additive based on bitumen.

8
R. Li et al. Fuel 300 (2021) 121015

Fig. 16. Radial distribution of each component respect to the aggregate surface: (a) neat bitumen; (b) PETB.

Fig. 17. Interfacial interaction of PETB and aggregate.

• The MD simulation showed that PET additive improved the cohesive

Table 5
energy density of the binder.
Adhesion energy of binder and aggregate in dry condition.
• The adhesion between the binder and aggregate was strengthened by
Sample- ΔEvdw (kcal/ ΔEelec (kcal/ ΔEadhesion (kcal/ Wadhesion (mJ/ PET additive in both dry and wet condition. PET additive tends to
ID mol) mol) mol) m2)
approach the binder-aggregate interface, and form strong hydrogen
Agg_NB − 201.201 − 1.897 − 208.03 − 118.03 bonding, thus increasing the interfacial adhesion. The higher polar­
Agg_PETB − 213.066 − 8.658 − 226.81 − 128.68 ity of PET additive in comparison with bitumen further increased the
Note: Agg_NB represents the interface between aggregate and neat bitumen in adhesion with the polar aggregate.
dry condition, and Agg_PETB represents the interface between aggregate and
PET additive modified bitumen in dry condition. Both the experimental results and MD simulation indicated that the
PET additive can improve the adhesion between asphalt binder and
aggregate, thus increasing the resistance to moisture-induced damage of
Table 6 asphalt mixture. It is therefore concluded that the PET additive obtained
Adhesion energy of binder and aggregate in wet condition and the calculated ER. from the aminolysis treated waste PET bottles can be successfully used
Sample-ID ΔEbit water ΔEagg water ΔEbit agg ΔEadhesion water ER as a sustainable antistripping agent in asphalt mixture. This technology
(kcal/ (kcal/ (kcal/ (kcal/mol) transferred the waste PET into a value-added additive that is applicable
mol) mol) mol) in the pavement industry. In future studies, the effects of the PET derived
Agg_water_NB − 105.1 − 200.0 − 73.9 − 231.3 0.88 antistripping additive on asphalt mixtures with different aggregates will
Agg_water_PETB − 145.1 − 158.3 − 120.5 − 182.9 1.24 be investigated, and the economic and environmental performances of
Note: Agg_water_NB represents the interface between aggregate and neat this technology will be evaluated through life cycle cost analysis and life
bitumen in wet condition, and Agg_water_PETB represents the interface between cycle assessment, respectively.
aggregate and PET additive modified bitumen in wet condition.
Funding Sources
5. Conclusions
The authors would like to acknowledge the funding support from the
In this study, the antistripping ability of the waste PET derived ad­ Hong Kong Environment and Conservation Fund through ECF Project
ditive on asphalt mixture was investigated through both laboratory tests 84/2017 and Science and Technology Project of Henan Provincial
and MD simulation. The following points summarize the main conclu­ Department of Transportation (Project number: 2020J6).
sions of this study:
CRediT authorship contribution statement
• The experimental results indicated that the boiling stripping area
decreased significantly with the incorporation of PET additive. In Rui Li: Conceptualization, Methodology, Experiments, Test result
addition, the TSR increased dramatically to over 90% with only 1 wt analysis, Manuscript drafting, Manuscript revision and editing. Zhen
% of PET additive, and the failure mode transferred from adhesive Leng: Conceptualization, Methodology, Test result analysis, Manuscript
failure to cohesive failure. revision and editing. Jun Yang: Test result analysis, Manuscript revision

9
R. Li et al. Fuel 300 (2021) 121015

and editing. Guoyang Lu: Test result analysis, Manuscript revision and [19] Ahmadinia E, Zargar M, Karim MR, Abdelaziz M, Ahmadinia E. Performance
evaluation of utilization of waste polyethylene terephthalate (PET) in stone mastic
editing. Man Huang: Test result analysis, Manuscript revision and
asphalt. Constr Build Mater 2012;36:984–9.
editing. Jingting Lan: Experiments, Test result analysis. Hongliang [20] Leng Z, Sreeram A, Padhan RK, Tan Z. Value-added application of waste PET based
Zhang: Test result analysis. Yawei Bai: Test result analysis, Manuscript additives in bituminous mixtures containing high percentage of reclaimed asphalt
revision. Zejiao Dong: Test result analysis, Manuscript revision. pavement (RAP). J Cleaner Prod 2018;196:615–25.
[21] Zhang H, Huang M, Hong J, Lai F, Gao Y. Molecular dynamics study on
improvement effect of bis(2-hydroxyethyl) terephthalate on adhesive properties of
asphalt-aggregate interface. Fuel 2021;285:119175.
Declaration of Competing Interest [22] Merkel DR, et al. Waste PET chemical processing to terephthalic amides and their
effect on asphalt performance. ACS Sustain Chem Eng 2020;8(14):5615–25.
The authors declare that they have no known competing financial [23] Leng Z, Padhan RK, Sreeram A. Production of a sustainable paving material
through chemical recycling of waste PET into crumb rubber modified asphalt.
interests or personal relationships that could have appeared to influence J Clean Prod 2018;180:682–8.
the work reported in this paper. [24] Xu G, Wang H. Study of cohesion and adhesion properties of asphalt concrete with
molecular dynamics simulation. Comput Mater Sci 2016;112:161–9.
[25] Xu M, Yi J, Feng D, Huang Y, Wang D. Analysis of adhesive characteristics of
Acknowledgment asphalt based on atomic force microscopy and molecular dynamics simulation. ACS
Appl Mater Interfaces 2016;8(19):12393–403.
The authors would like to acknowledge the insightful discussions [26] Zhang L, Greenfield ML. Relaxation time, diffusion, and viscosity analysis of model
asphalt systems using molecular simulation. J Chem Phys 2007;127(19):194502.
about molecular dynamics simulation with Dr. Yangming Gao from [27] Lu Y, Wang L. Nano-mechanics modelling of deformation and failure behaviours at
Aston University in UK and Dr. Chi Zhang from ETH in Switzerland. asphalt–aggregate interfaces. Int J Pavement Eng 2011;12(4):311–23.
[28] Bhasin A, Bommavaram R, Greenfield ML, Little DN. Use of molecular dynamics to
investigate self-healing mechanisms in asphalt binders. J Mater Civ Eng 2011;23
References (4):485–92.
[29] Yao H, Dai Q, You Z. Chemo-physical analysis and molecular dynamics (MD)
[1] Collop A, Choi Y, Airey G, Elliott R. Development of the saturation ageing tensile simulation of moisture susceptibility of nano hydrated lime modified asphalt
stiffness (SATS) test. In Proceedings of the Institution of Civil Engineers-Transport, mixtures. Constr Build Mater 2015;101:536–47.
2004, vol. 157, no. 3, pp. 163-171: Thomas Telford Ltd. [30] Ling C, Moraes R, Swiertz D, Bahia H. Measuring the influence of aggregate coating
[2] Emery J, Seddik H. Moisture damage of asphalt pavements and antistripping on the workability and moisture susceptibility of cold-mix asphalt. Transp Res Rec
additives: causes, identification, testing and mitigation; 1997. 2013;2372(1):46–52.
[3] Airey GD, Choi Y-K. State of the art report on moisture sensitivity test methods for [31] Corbett LW. Composition of asphalt based on generic fractionation, using solvent
bituminous pavement materials. Road Mater Pavement Design 2002;3(4):355–72. deasphaltening, elution-adsorption chromatography, and densimetric
[4] Gubler R, Partl MN, Canestrari F, Grilli A. Influence of water and temperature on characterization. Anal Chem 1969;41(4):576–9.
mechanical properties of selected asphalt pavements. Mater Struct 2005;38(5): [32] Greenfield ML, Zhang L. Developing model asphalt systems using molecular
523–32. simulation: final model. University of Rhode Island. Transportation Center; 2009.
[5] Grenfell J, Ahmad N, Liu Y, Apeagyei A, Large D, Airey G. Assessing asphalt [33] Li DD, Greenfield ML. Chemical compositions of improved model asphalt systems
mixture moisture susceptibility through intrinsic adhesion, bitumen stripping and for molecular simulations. Fuel 2014;115:347–56.
mechanical damage. Road Mater Pavement Design 2014;15(1):131–52. [34] Zhang L, Greenfield ML. Analyzing properties of model asphalts using molecular
[6] Liu Y, Apeagyei A, Ahmad N, Grenfell J, Airey G. Examination of moisture simulation. Energy Fuels 2007;21(3):1712–6.
sensitivity of aggregate–bitumen bonding strength using loose asphalt mixture and [35] Ding Y, Huang B, Shu X, Zhang Y, Woods ME. Use of molecular dynamics to
physico-chemical surface energy property tests. Int J Pavement Eng 2014;15(7): investigate diffusion between virgin and aged asphalt binders. Fuel 2016;174:
657–70. 267–73.
[7] Cheng D, Little DN, Lytton RL, Holste JC. Surface energy measurement of asphalt [36] Yao H, Dai Q, You Z. Molecular dynamics simulation of physicochemical properties
and its application to predicting fatigue and healing in asphalt mixtures. Transp of the asphalt model. Fuel 2016;164:83–93.
Res Rec 2002;1810(1):44–53. [37] Sun H. COMPASS: an ab initio force-field optimized for condensed-phase
[8] Kandhal PS. Field and laboratory evaluation of stripping in asphalt pavements. applications overview with details on alkane and benzene compounds. J Phys
Transp Res Rec 1994;1454. Chem B 1998;102(38):7338–64.
[9] Fromm HJ. The mechanisms of asphalt stripping from aggregate surfaces. In [38] Gao Yangming, Zhang Yuqing, Gu Fan, Xu Tao, Wang Hao. Impact of minerals and
Proceedings of the association of asphalt paving technologies, vol. 43; 1974. water on bitumen-mineral adhesion and debonding behaviours using molecular
[10] Nazirizad M, Kavussi A, Abdi A. Evaluation of the effects of anti-stripping agents on dynamics simulations. Constr Build Mater 2018;171:214–22.
the performance of asphalt mixtures. Constr Build Mater 2015;84:348–53. [39] Dong Zejiao, Liu Zhiyang, Wang Peng, Gong Xiangbing. Nanostructure
[11] Sebaaly PE, Little D, Hajj EY, Bhasin A. Impact of lime and liquid antistrip agents characterization of asphalt-aggregate interface through molecular dynamics
on properties of idaho hot-mix asphalt mixture. Transp Res Rec 2007;1998(1): simulation and atomic force microscopy. Fuel 2017;189:155–63.
65–74. [40] Ding W, Jin H, Zhao Q, Chen B, Cao W. Dissolution of polycyclic aromatic
[12] Hesami E, Mehdizadeh G. Study of the amine-based liquid anti-stripping agents by hydrocarbons in supercritical water in hydrogen production process: A molecular
simulating hot mix asphalt plant production process. Constr Build Mater 2017;157: dynamics simulation study. Int J Hydrogen Energy 2020;45(52):28062–9.
1011–7. [41] Yoon HH, Tarrer AR. Effect of aggregate properties on stripping (no. 1171). Transp
[13] Park D-W, Seo W-J, Kim J, Vo HV. Evaluation of moisture susceptibility of asphalt Res Rec 1988.
mixture using liquid anti-stripping agents. Constr Build Mater 2017;144:399–405. [42] Zhuravlev LT. Concentration of hydroxyl groups on the surface of amorphous
[14] Iskender E, Aksoy A, Ozen H. Indirect performance comparison for silicas. Langmuir 1987;3(3):316–8.
styrene–butadiene–styrene polymer and fatty amine anti-strip modified asphalt [43] Xu Guangji, Wang Hao. Molecular dynamics study of interfacial mechanical
mixtures. Constr Build Mater 2012;30:117–24. behavior between asphalt binder and mineral aggregate. Constr Build Mater 2016;
[15] Aksoy A, Şamlioglu K, Tayfur S, Özen H. Effects of various additives on the 121:246–54.
moisture damage sensitivity of asphalt mixtures. Constr Build Mater 2005;19(1): [44] Gao Yangming, Zhang Yuqing, Yang Yang, Zhang Junhui, Gu Fan. Molecular
11–8. dynamics investigation of interfacial adhesion between oxidised bitumen and
[16] Modarres A, Hamedi H. Developing laboratory fatigue and resilient modulus mineral surfaces. Appl Surf Sci 2019;479:449–62.
models for modified asphalt mixes with waste plastic bottles (PET). Constr Build [45] Ling Cheng, Hanz Andrew, Bahia Hussain. Measuring moisture susceptibility of
Mater 2014;68:259–67. Cold Mix Asphalt with a modified boiling test based on digital imaging. Constr
[17] Modarres A, Hamedi H. Effect of waste plastic bottles on the stiffness and fatigue Build Mater 2016;105:391–9.
properties of modified asphalt mixes. Mater Des 2014;61:8–15. [46] Jada A, Salou M. Effects of the asphaltene and resin contents of the bitumens on the
[18] Baghaee Moghaddam T, Karim MR, Syammaun T. Dynamic properties of stone water–bitumen interface properties. J Petrol Sci Eng 2002;33(1-3):185–93.
mastic asphalt mixtures containing waste plastic bottles. Constr Build Mater 2012; [47] Bhasin A, Howson J, Masad E, Little DN, Lytton RL. Effect of modification processes
34:236–42. on bond energy of asphalt binders. Transp Res Rec 2007;1998(1):29–37.

10