Content uploaded by Georgios Pipintakos
Author content
All content in this area was uploaded by Georgios Pipintakos on May 16, 2023
Content may be subject to copyright.
Supervisors Prof. Dr. Wim Van den bergh | Dr. Hilde Soenen
Thesis submitted for the degree of Doctor in Applied Engineering
Faculty of Applied Engineering | Antwerp, 2022
Towards an enhanced understanding of the
oxidative ageing mechanisms in bitumen
Georgios Pipintakos
Towards an enhanced understanding of the oxidative ageing mechanisms in bitumen Georgios Pipintakos
Faculty of Applied Engineering
Bouwkunde
Towards an enhanced understanding of the oxidative
ageing mechanisms in bitumen
Thesis submitted for the degree of doctor in Applied Engineering
at University of Antwerp to be defended by
Georgios PIPINTAKOS
Supervisors:
Prof. Dr. Wim Van den bergh
Dr. Hilde Soenen Antwerp, 2022
ii
Jury
Prof. D
r. Christophe Vande Velde
University of Antwerp, Facul
ty of Applied Engineering,
iPRACS
research group
Chair
Prof. D
r. Wim Van den bergh
University of Antwerp, Faculty of Applied Engineering,
EMIB research group
Promotor
D
r. Hilde Soenen
Nynas NV
Promotor
Prof. D
r. Evangelos Manthos
Aristotle University of Thessaloniki, Department of Civil Engineering,
Highway Engineering Laboratory
Secretary
Prof. D
r. Sabine Van Doorslaer
University of Antwerp, Faculty of
Sciences,
Bimef
research group
Member
Prof. D
r. Aikaterini Varveri
Delft
University of Technology, Civil Engineering and Geosciences,
Pavement Engineering
research group
Member
Prof. D
r. Bernhard Hofko
Technical University of Wien
, Institute of Transportation,
Christian Doppler Laboratory for Chemo
-Mechanical Analysis of Bituminous
Materials
Member
D
r. Xiaohu Lu
Nynas AB
Member
Contact:
Georgios Pipintakos
University of Antwerp, Faculty of Applied Engineering, EMIB research group
Groenenborgerlaan 171, 2020 Antwerp, Belgium
Email: gpipintakos@gmail.com
©2022 Georgios Pipintakos
All rights reserved
ISBN: 97890
57287671
Deposit number: D/2022/12.293/37
TABLE OF CONTENTS
ABSTRACT ......................................................................................................................... VI
SAMENVATTING .............................................................................................................. VIII
ACKNOWLEDGEMENTS ....................................................................................................... X
LIST OF ABBREVIATIONS ................................................................................................... XII
1 INTRODUCTION ........................................................................................................... 1
GENERAL BACKGROUND ..................................................................................................... 1
THEORETICAL BACKGROUND ................................................................................................ 3
Hypotheses of the ageing mechanisms in bitumen ............................................. 3
Hypotheses of the bitumen’s structure ................................................................ 5
Chemistry and performance of bitumen .............................................................. 7
PROBLEM STATEMENT ....................................................................................................... 9
RESEARCH OBJECTIVES AND SCOPE...................................................................................... 10
RESEARCH OUTLINE ......................................................................................................... 11
2 RESEARCH METHODOLOGY ........................................................................................ 15
SUMMARY ..................................................................................................................... 15
MATERIAL SELECTION ...................................................................................................... 16
SPECTROSCOPIC TECHNIQUES ............................................................................................ 16
Fourier Transform Infrared (FTIR) spectroscopy ................................................ 16
Electron Paramagnetic Resonance (EPR) spectroscopy ..................................... 18
Time-of-flight Secondary Ion Mass Spectrometry (TOF-SIMS) ........................... 20
Proton Nuclear Magnetic Resonance (1H-NMR) spectroscopy .......................... 21
GRAVIMETRIC TECHNIQUES ............................................................................................... 23
Dynamic Vapour Sorption (DVS) ........................................................................ 23
THERMOANALYTICAL TECHNIQUES ...................................................................................... 24
Differential Scanning Calorimetry (DSC) ............................................................ 24
Wide Angle X-ray Diffraction (WAXD) ................................................................ 25
MICROSCOPIC TECHNIQUES ............................................................................................... 26
Atomic Force Microscopy (AFM) ........................................................................ 26
Confocal Laser Scanning Microscopy (CLSM) ..................................................... 28
RHEOLOGICAL CHARACTERISATION ..................................................................................... 29
Dynamic Shear Rheometer (DSR) ....................................................................... 29
EXPERIMENTAL CHALLENGES ............................................................................................. 32
3 AGEING MECHANISMS IN BITUMEN VIA KINETICS ...................................................... 37
SUMMARY ..................................................................................................................... 37
OBJECTIVES ................................................................................................................... 38
iv
MATERIALS AND METHODS .............................................................................................. 38
OXIDATION PHASES......................................................................................................... 40
FTIR .................................................................................................................... 40
EPR .................................................................................................................... 42
OXIDATION PRODUCTS .................................................................................................... 44
TOF-SIMS ........................................................................................................... 44
REACTIVITY AND DIFFUSIVITY ............................................................................................ 50
FTIR .................................................................................................................... 50
DVS .................................................................................................................... 52
HIGHLIGHTS OF THE CHAPTER ........................................................................................... 54
4 AGEING MECHANISMS IN BITUMEN VIA STANDARDISED LAB SIMULATIONS .............. 55
SUMMARY .................................................................................................................... 55
OBJECTIVES ................................................................................................................... 56
MATERIALS AND METHODS .............................................................................................. 57
OXIDATION PRODUCTS IN BITUMEN .................................................................................... 58
FTIR .................................................................................................................... 58
EPR .................................................................................................................... 60
1H-NMR ............................................................................................................. 61
TOF-SIMS ........................................................................................................... 64
OXIDATION PRODUCTS IN SARA FRACTIONS ........................................................................ 66
FTIR .................................................................................................................... 66
1H-NMR ............................................................................................................. 68
DSC .................................................................................................................... 69
HIGHLIGHTS OF THE CHAPTER ........................................................................................... 70
5 MECHANISMS OF SURFACE MICROSTRUCTURE IN BITUMEN ...................................... 73
SUMMARY .................................................................................................................... 73
OBJECTIVES ................................................................................................................... 74
MATERIALS AND METHODS .............................................................................................. 75
MECHANISMS OF BITUMEN MICROSTRUCTURE ..................................................................... 76
DSC .................................................................................................................... 76
WAXD ................................................................................................................ 78
EVOLUTION OF BITUMEN MICROSTRUCTURE ........................................................................ 80
CLSM .................................................................................................................. 80
HIGHLIGHTS OF THE CHAPTER ........................................................................................... 83
6 EFFECT OF LAB AGEING ON BITUMEN MICROSTRUCTURE .......................................... 85
SUMMARY .................................................................................................................... 85
OBJECTIVES ................................................................................................................... 86
MATERIALS AND METHODS .............................................................................................. 87
MORPHOLOGY AND BEE COVERAGE.................................................................................... 89
WAVEFORM CHARACTERISTICS OF BEE STRUCTURES ............................................................... 95
BEE DETECTION USING DEEP LEARNING AND 2-D FFT ............................................................. 97
HIGHLIGHTS OF THE CHAPTER .......................................................................................... 100
7 CHEMISTRY AND RHEOLOGY OF BITUMEN WITH AGEING ......................................... 101
SUMMARY ................................................................................................................... 101
OBJECTIVES ................................................................................................................. 102
MATERIALS AND METHODS ............................................................................................ 103
Pearson’s r and Sample Correlation Matrix ..................................................... 104
Non-parametric Wilcoxon Rank Sum exact test ............................................... 104
Multivariate Factor Analysis ............................................................................ 105
Factor analysis regression ................................................................................ 106
THE EFFECT OF LAB AGEING PROTOCOL .............................................................................. 106
LINK BETWEEN CHEMISTRY AND RHEOLOGY BY AGEING STATE ................................................ 111
HIGHLIGHTS OF THE CHAPTER .......................................................................................... 117
8 THERMODYNAMICS MODELLING OF AGEING MECHANISMS IN BITUMEN ................. 119
SUMMARY ................................................................................................................... 119
OBJECTIVES ................................................................................................................. 119
MATERIALS AND METHODS ............................................................................................ 120
MECHANISM-BASED REACTION-DIFFUSION MODEL .............................................................. 121
SARA COMPOSITIONS AND FAST-RATE OXIDATION PHASE ..................................................... 123
HIGHLIGHTS OF THE CHAPTER .......................................................................................... 127
9 CONCLUSIONS AND RECOMMENDATIONS ................................................................ 129
CONCLUSIONS .............................................................................................................. 129
Molecular level ................................................................................................. 129
Microscale ........................................................................................................ 131
Phenomenological and mechanism-based modelling ...................................... 132
RECOMMENDATIONS ..................................................................................................... 133
REFERENCES .................................................................................................................... 137
RESEARCH DISSEMINATION ............................................................................................. 155
vi
ABSTRACT
The ageing phenomenon in asphalt and its binding medium, the bitumen, is well
documented in the scientific literature with regard to its rheomechanical effects. To
understand the ‘whys’ behind these alterations one should seek additionally on the
chemistry of bitumen.
This dissertation supports experimentally the hypothesis of an oxidation scheme
consisting of a fast and a slow rate-determining phase. This is achieved by utilising
various unmodified bituminous binders of different origin of crude source, refinery
process and performance both in oxidation kinetics and with standard lab ageing
simulations. The findings of Electron Paramagnetic Resonance (EPR) and Fourier
Transform Infrared (FTIR) Spectroscopy manage to distinguish the two phases, while
Proton Nuclear Magnetic Resonance (1H-NMR) Spectroscopy and Time-of-Flight
Secondary Ion Mass Spectrometry (TOF-SIMS) unravel the main oxygenated products
in bitumen. Chemical investigations of the SARA fractions show additionally the effect
of ageing on the different fractions. Moreover, each contribution of the coupled
reaction-diffusion is studied with FTIR and Dynamic Vapour Sorption (DVS).
Next, the fundamental mechanisms for the development of a peculiar microstructure,
the bee structures in bitumen, were explored and validated via Differential Scanning
Calorimetry (DSC) and Wide Angle X-ray Diffraction (WAXD). In this dissertation, the
hypothesis that the crystallisable compounds in bitumen are the main reason for such
structures is adopted, and thus various waxy binders were studied for the effect of lab
ageing with Atomic Force Microscopy (AFM) and Confocal Laser Scanning Microscopy
(CLSM). Image processing methods allowed to conclude that the bee coverage is
reduced upon ageing.
Additionally, the oxygenated products as revealed by the ageing mechanisms in
bitumen were linked via multivariate statistics to advanced rheological parameters
extracted via the Dynamic shear Rheometer (DSR). Convergence of the fast rate-
determining phase and the short-term lab ageing was found both for chemistry and
rheology. The dissertation ends with the description of a thermodynamics of
irreversible processes model for the fast rate-determining oxidation phase, with the
model accounting for reasonable changes of the SARA fractions with oxidation.
All in all, this dissertation provides a deeper scientific insight into the oxidative ageing
mechanisms in bitumen and clarifies the relationship between chemical and
rheological characteristics, which may contribute as a guideline to a more sustainable
road infrastructure in the future.
viii
SAMENVATTING
Het verouderingsverschijnsel van asfalt en zijn bindmiddel, het bitumen, is heden goed
gedocumenteerd in de wetenschappelijke literatuur, met betrekking tot de
reomechanische effecten. Om het 'waarom' achter deze veranderingen te begrijpen is
aanvullend onderzoek naar de chemie van bitumen noodzakelijk.
Dit proefschrift ondersteunt experimenteel de hypothese van een snel en een traag
oxidatieregime. Verschillende ongemodificeerde bitumineuze bindmiddelen zijn
onderzocht, verschillend in oorsprong (crude source), raffinageproces en prestaties.
Hierbij werden zowel de oxidatiekinetiek als de standaard labverouderingssimulaties
toegepast. De bevindingen van Electron Paramagnetic Resosonance (EPR) (EPR) en
Fourier Transform Infrared (FTIR) Spectroscopy slagen erin de twee fasen te
onderscheiden. De Proton Nuclear Magnetic Resonance (1H-NMR) Spectroscopy en
Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) laten toe om de
belangrijkste reactieproducten te ontrafelen. Chemisch onderzoek van de SARA-
fracties toont bovendien het effect van veroudering op de verschillende fracties.
Daarnaast werd de bijdrage van een gekoppelde reactie-diffusie ook succesvol
bestudeerd met FTIR en Dynamic Vapour Sorption (DVS).
De mechanismen voor de ontwikkeling van een bijzondere microstructuur in bitumen,
de bijenstructuur (bee-structure), werd onderzocht en gevalideerd met Differential
Scanning Calorimetry (DSC) en Wide Angle X-ray Diffraction (WAXD). In dit proefschrift
wordt de hypothese aangenomen dat kristalliseerbare verbindingen in bitumen de
belangrijkste reden zijn voor dergelijke structuren. Hiervoor werden verschillende
washoudende bindmiddelen na verschillende verouderingsprocedures onderzocht
met Atomic Force Microscopy (AFM) en Confocal Laser Scanning Microscopy (CLSM).
Door middel van beeldverwerkingsmethoden werd aangetoond dat tijdens het
verouderingsproces de bijendensititeit vermindert.
Zuurstofhoudende componenten, bepaald via de hierboven chemische testmethoden,
konden gekoppeld worden aan reologische parameters, door gebruik te maken van
multivariabele statistiek. Convergentie van de snelle reactiefase en de korte termijn
veroudering in het labo, werd gevonden voor zowel de chemische als reologische
paremeters. Het proefschrift eindigt met een thermodynamisch model, dat rekening
houdt met veranderingen van de SARA-fracties tijdens oxidatie, en waarin de
verouderingsmechanismen van bitumen voor de snelle reactiestap konden
geïmplementeerd worden.
Dit proefschrift geeft een dieper inzicht in de oxidatieve verouderingsmechanismen in
bitumen en verduidelijkt de samenhang tussen chemische en reologische kenmerken,
wat in de toekomst als een leidraad kan bijdragen tot een duurzamere
wegeninfrastructuur.
x
ACKNOWLEDGEMENTS
During this journey, I see myself like an unmanned ship at the port of Antwerp which
needed deck staff to balance, sail and explore the oceans. Luckily, I had two
experienced captains with me. Captain Soenen who had fought with big sea monsters
(we can call them in our story DSRs and AFMs) and helped the boat to survive by them
in each hardship and Admiral Van den bergh. The Admiral became more like a friend
as he was affectionate with his boat and took care of it when the weather was bad. It
was only then that the boat slowly sailed and passed from the port of Rotterdam where
Lieutenant Varveri with pleasure boarded on the ship. She knew this boat as she had
assisted in the design of it, in Delft some years ago. The idea of this boat actually was
shaped back in Greece thanks to Lieutenant Manthos.
There were people who joined the boat’s crew, shared their passion, wisdom and
experience and left the boat or watched it sailing from far away. Ηοnourable mention
to Commanders Mühlich, Blom, Ching, Lommaert, Tsakalidis and Sjövall. The ship was
eventually floating, however, what is a ship without night talks, with a flask of rum,
with good friends at the gunwale? Sailors they call them, but actually, these were the
most vivid and important to accomplish the mission of this journey. They were literally
living inside the boat and giving joyful life to it. Sailors Jacobs, Margaritis, Papini,
Hasheminejad, Ghalandari, Pieters, Borinelli, Barisoglu, Barros, Omranian, Hernando,
Vuye thank you for your services that reached more than your predefined duties. What
can be said also for all those sailors that were telegraphing their best vibes to the boat.
The boat is proud to feel the support of Sailors Tziampiris, Chatzistavrou, Kougioumtzi,
Gavras, Papathanassi, Karamoschos, Georgiadis, Pantsi, Emmanouilidis, Vasileiadis,
Tsikoulas, Papadakis, Teletinis and the Royal Dutch Navy: Tzortzinakis, Mastoras,
Presvyri, Papaioanou, Seleridis, Mouratidis. The boat is also grateful for the loyal
companion of Comrade Maria that accompanied it almost till the end of its journey.
International diplomacy of the boat resulted in some of the strongest alliances with
Austria and a network of allies known as the ‘Young Ageing Crew’. The boat is more
than honoured to have established together with the Shipowners Mirwald, Camargo
and Omairey this network.
And the boat from just floating started to sail full speed ahead. Of course, there were
tsunamis, hurricanes and storms. Tough winters and summer dryness. The boat would
have never managed to kick-off any journey if it was not the family of it that provided
all the necessary supplies to be designed. Τουλίτσα και Λεφτεράκο, σας ευχαριστώ και
είμαι περήφανος για σας. Circumferential family support to the boat was more than
crucial to complete the journey, as the boat was travelling around the world during
these 1720 days. Thank you Warrant Officers Fenia, Nina, Sofie, Christiana, Kostas,
Paschalina, Giannis, Kallinikos, Thomy, Vasilis, Efi, Zacharoula and Angelos. Fleet
Admirals Efthimia and Giorgos, sincerely thank you and I will always remember you.
And the boat was approaching the end target. It started wondering, as rumours say, is
it really the journey that matters or the final destination. Lieutenant Commanders Lu,
Van Doorslaer, Hofko and Commander Vande Velde assisted the boat to find the real
meaning of its journey.
And the boat arrived. People were waving as it was approaching and cheering with
bottoms up. In the end, the boat realised one thing for sure: You can never enjoy the
ups and downs of the heavy sea if you do not relax and let yourself being drifted by
the flow. Flow is People. People are Flow.
Georgios Pipintakos
Antwerp, December 2022
xii
LIST OF ABBREVIATIONS
1D FFT
One-dimensional Fast Fourier Transform
2D FFT
Two-dimensional Fast Fourier Transform
AFM
Atomic Force Microscopy
AlkP
α-alkyl Protons
ARI
Aromaticity Index
AroP
Aromatic Protons
ATR
Attenuated Total Reflectance
CBOR
Carbon-Based Organic Radicals
CLSM
Confocal Laser Scanning Microscopy
COI
Carbonyl Index
CW
Continuous Wave
DIC
Differential Interference Contrast
DSC
DSR
Differential Scanning Calorimetry
Dynamic Shear Rheometer
DVS
Dynamic Vapour Sorption
EPR
Electron Paramagnetic Resonance
FE
Finite Element
FTIR
Fourier-Transform Infrared spectroscopy
1H-NMR
Proton Nuclear Magnetic Resonance
MetP
Methyl Protons
M-TFO(T)
Modified Thin Film Oven (Test)
MtlP
Methylene Protons
OleP
Olefinic Protons
PAV
Pressure Ageing Vessel
RTFO(T)
Rolling Thin Film Oven (Test)
SANS
Small-Angle Neutron Scattering
SARA
SAXS
Saturates, Aromatic, Resins, Asphaltenes
Small-Angle X-Ray Scattering
SHRP
Strategic Highway Research Program
SOI
TIP
Sulfoxide Index
Thermodynamics of Irreversible Processes
TOF-SIMS
Time-of-Flight Secondary Ion Mass Spectrometry
TTSP
Time-Temperature Superposition Principle
WAXD
Wide Angle X-ray Diffraction
1
1 INTRODUCTION
General background
In Belgium, as well as in the major part of the world, private transportation on the
mainland is primarily conducted by individuals by making use of asphalt roads. The
word ‘asphalt’ origins from the ancient Greek word ‘άσφαλτος’ which literally
translates to something that cannot fail or collapse. Although asphalt has been
invented and intended to be used as an impeccable material, it has been shown that
the reality is far from this goal. To construct an asphalt mixture, whatsoever the type
is, one needs typically three main components, namely mineral aggregates, filler and
bitumen. For example in Belgium the preferred type for surface layers of lower load
severity is the asphalt concrete whereas another type, that of stone mastic asphalt, is
commonly used for heavily trafficked surface layers. While the asphalt mixture types,
differ between them i.e. in terms of air voids content and gradation curves, they all
have something in common: the existence of a binder known in European terminology
as bitumen. The latter is a material derived from crude oil and acts as the binding
medium in asphalt pavements due to its superior viscoelastic, adhesive and water-
proofing performance [1]. Recently released data report that 85% of the annual
bitumen world production (87 million tons) is used for road paving [2].
The condition of the road network is considered indirectly a metric of the level of
society and therefore authorities make significant efforts to keep it to an acceptable
degree. This is succeeded, among others, via various efforts to prolong the service life
of asphalt pavement and predominantly via a preventive maintenance plan. However,
it is often the case in Belgium, as in other developed countries, to face a number of
distress types in pavement. Experts have identified the reasons behind certain
distresses resulting in the lack of adequate asphalt performance. Some of the most
pronounced ones, i.e. block cracking, are related solely to one of the ingredients of
1. Introduction
2
asphalt, the bitumen. As such, bitumen has long been a matter of interest not only for
the scientific community but also for the asphalt industry in general since several
generated failure types are linked to it.
It all starts from the fact that bitumen is an organic material and that oxygen is
omnipresent in the atmosphere. Hence, bituminous binders are composed of a variety
of organic molecules consisting of about 85% carbon, 10% hydrogen, heteroatoms
such as nitrogen (0-2%), oxygen (0-2%), sulfur (0-9%) and traces of metals such as
vanadium, iron and nickel [3,4]. Among the myriad of chemical functional groups of
virgin bitumen are the hydroxyl groups of phenols, imino groups of pyrrolic
compounds, as well as carbonyl groups of ketones, carboxylic acids and 2-quinolones
[5,6]. The lesser amount of heteroatoms in bituminous organic molecules not only
modulate the polarity but also constitute chemical functional groups that can react
and change. In particular, a wide variety of sulfur-containing compounds occur
preferentially within the bituminous binders, such as sulfides, disulfides, sulfoxides,
ring compounds (thiophenes, benzothiophenes and dibenzothiophenes) and their
alkyl derivatives [7]. Due to its vast number of organic constituents bitumen becomes
prone to oxidation during exposure to mixing and environmental conditions [8].
Oxygen, on the other hand, exists in about 21% in the atmospheric air and although
essential for the continuity of fauna and flora of earth, in the case of bitumen it results
in some irreversible changes.
Albeit other irreversible (volatilisation and condensation), and reversible (physical
hardening) physicochemical processes take place in bitumen, the inevitable reaction-
diffusion of oxygen in it, widely known as ‘oxidative ageing’, is considered of utmost
importance since it generally deteriorates the pavement performance [9,10]. To date,
a lot of studies have already demonstrated the negative impact of oxidative ageing
resulting in increased brittleness of bitumen, responsible for cracking and ravelling in
asphalt scale [6,11–13]. However, it should be noted that moderated ageing may
indicate also positive effects on pavement if one is able to control it, i.e. by reducing
severe deformations and rutting.
From the combination of the two, the organic nature of bitumen and the atmospheric
oxygen, researchers were only able to intervene in the ‘lesser of the two evils’, the
bitumen. In parallel, recent literature has shown that other reactive oxygen species in
the atmosphere, like ozone and nitrogen oxides, have a strong oxidation potential to
bitumen [14,15]. As such, following a brute-force approach, a plethora of additives and
rejuvenating agents have been introduced with the potential to improve the bitumen’s
performance or to compensate for certain irreversible chemical alterations [16,17].
Hitherto, a number of studies stressed that chemistry and microstructure may be the
3
key element to unwrap the association of ageing with the bitumen’s performance
[6,18–22] and the necessity to bridge the different scales, from molecular to nano-,
micro- and further to meso- and macroscale [23–25]. However, this proved to be far
from an easy-going task since the bitumen’s exact composition depends on the
refinery process, the origin of crude oil and additionally alters with ageing, a fact that
makes the unboxing of this complex organic blend even more challenging [26,27].
Theoretical background
Hypotheses of the ageing mechanisms in bitumen
The unravelling of the underlying mechanisms of oxidative ageing has been a tedious
task already from the 1960s. The scientific community, historically, turned from studies
into isothermal oxidation kinetics to efforts to simulate the in-situ ageing in the lab
and, later, to an incorporation of various antioxidants which were believed to prohibit
the process of autoxidation as the binder reacts with oxygen, generating new
compounds that may continue to react with oxygen [18]. It has been concluded that a
thorough understanding of the oxidation products and the effect of oxygen on
fundamental bitumen properties is of paramount importance. As a result of this
acknowledgement, there is an increasing interest over the past decade back to the
investigation of the bitumen’s ageing mechanisms in order to tackle once and for all
the precedential uncertainties [28–30].
With regard to the hypotheses of the underlying mechanisms, independent studies
support that during the ageing process, the active functionalities of bitumen molecules
are decomposed through the oxidative dehydrogenation of polycyclic
perhydroaromatics generating intermediate hydroperoxides [31,32]. Through the
years, different mechanisms have been proposed to describe this phenomenon
ranging from an oxycyclic reaction mechanism [33] up to a dual sequential binder
oxidation scheme [6,34]. According to the latter, two major oxidation phases may
exist, namely the chemically distinct and rate-determining “fast-spurt” and
“slow/long-term” [18,19,34,35]. The existence of free radicals during this dual-
oxidation route will most likely provoke the chemical reactions resulting eventually in
the formation of polar sulfoxides from non-polar sulfides and polar ketones (as well as
anhydrides and carboxylic acids in smaller amounts) from benzylic carbon moieties [9].
This idea has gained considerable support since a direct link with the asphalt
production stages can be established [36].
1. Introduction
4
Previous kinetic studies have revealed also differences in the reaction rate and the
formation of the two major oxidation products, namely carbonyls and sulfoxides [35].
It has been reported that the higher temperature during short-term ageing, in the
production stage, has a stronger effect on the intensity of oxygenated products than
the lower average temperature during long-term ageing, in service life [37,38].
Moreover, the literature emphasises that an increase of 10 °C may double the reaction
rate and a relatively high temperature (above 120 °C) may cause thermal
decomposition of the sulfoxides [34,39]. In reality, the temperature varies between
the different stages of the service life; when open to traffic, extreme temperatures
greater than 80 °C are never reached, while the temperatures during the paving stages
depend on the asphalt mixture application (hot, warm, cold).
In order to mimic such in-situ changes of bitumen due to oxidation, a common practice
is to utilise routine tests simulating the short- and long-term ageing. More specifically,
the Rolling Thin-Film Oven Test (RTFOT) [40] followed by the Pressure Ageing Vessel
(PAV) [41] are most commonly used to mimic the elevated temperature during
production and paving and the weather conditions during use-life respectively.
According to literature conditioning in PAV of 20 hours at a pressure of 2.1 MPa
corresponds to 7-10 years of field ageing [42–44], where the exact equivalence
depends on the bitumen and type of the asphalt mixture. It is still a matter of debate
if these standardised protocols simulate in a similar manner the ageing in-situ, not only
in terms of performance but also in terms of chemistry, as other factors i.e. humidity,
ultraviolet light, reactive oxygen species may affect additionally the reaction routes
[14]. Of pragmatic importance is to examine whether the artificial ageing simulations,
widely used in the asphalt sector, account for a fair correspondence with the ageing
mechanisms and products, reported previously as hypotheses in kinetic studies
[19,35]. However, it appears that experimental validation of the underlying
mechanisms has been confined primarily to sulfoxide and carbonyl formation. As such,
considerable research has been devoted to the determination of these functional
groups via chemical analysis such as Fourier-Transform Infrared (FTIR) spectroscopy
[37,45,46].
When it comes to a classification of bitumen, a rather simplistic approach is to
characterise bitumen’s composition based on solubility classes, by utilising appropriate
solvents. This technique was introduced already in the 1970s [47] and is known
nowadays as the SARA fractionation due to the four main derived solubility-based
categories, namely Saturates, Aromatics, Resins and Asphaltenes. Different techniques
have been proposed for this classification, which can also result in slight differences
even for the same bituminous samples [48,49]. Many studies have highlighted that,
5
due to ageing, bitumen exhibits a shift of these fractions from aromatics to resins and
finally to asphaltenes, whereas saturates are considered in general unreactive when
employing their relative content [20,24,42,50,51]. Moreover, asphaltenes are
considered the largest and most polar constituents which account primarily for the
overall bitumen’s viscosity [52,53], while in the literature they have been also reported
to precipitate in n-heptane because of their strong dispersive π-π interactions [53]. In
the past, resins were believed to contribute as a stabilising medium in bitumen, by
means of asphaltenes micelles surrounded by them [52,54]. Challenges mainly arise
when specific products of the hypotheses proposed so far for the ageing mechanisms
are needed to be confirmed experimentally and additionally in the way that these
products are distributed within the SARA fractions.
Hypotheses of the bitumen’s structure
In order to understand better the molecular associations of bitumen, researchers have
extensively brainstormed over the years both with regard to the internal structure of
bitumen at nanoscale level, but also for the microstructure (surface and bulk) of it at
microscale level.
Already in the dawn of the 20th century, a colloidal structure model for the intrinsic
architecture of bitumen was proposed [55], with more detailed descriptions of this
assumption documented by Nellensteyn a decade later [56]. The latter study
supported that asphaltenes constitute a colloidal suspension together with the
maltenes, which consists of aromatics, resins and saturates, while resins play a
stabilising role in bitumen [57]. This idea was developed further based on the
association with the rheological response which was believed to vary between sol
binders (Newtonian behaviour) and non-linear gel ones (non-Newtonian behaviour),
while most of the bituminous binders behave in a situation between the two cases
[58,59]. The difference between the two extrema was attributed to the
interconnectivity of the asphaltenes micelles in the gel phase, whereas the full
dispersion and successively lack of interaction of the asphaltenes micelles was
responsible for a sol type of bitumen’s structure. The intermediate situation implied
the co-existence of the two phases [58]. Relationships between the different colloid
types in bitumen and the ratio of asphaltenes and saturates content over the sum of
resins and aromatics have been established, to characterise bitumen’s colloidal
stability [60].
The main opponents of the colloidal model argued already in the 1990s that bitumen
is a purely homogeneous fluid known as dispersed polar fluid, which gave its name to
1. Introduction
6
this theory concerning the intrinsic architecture of bitumen [61–63]. The main
argumentation in disregarding the colloidal theory was generated around the lack of
an elastic plateau for the gel binders [52] and the lack of a thermodynamic basis for
the separation of bitumen in different phases. More specifically, an assumed phase
separation would reduce the system’s entropy and would require an enthalpy
compensation which is in contradiction to the micelle theory where asphaltenes are
considered to be gathered. In the homogeneous fluid theory, the bitumen molecules
are considered to be in a mutual solution including a range of solubility parameters in
such a way that everything can be kept soluble. Additionally, the supporters of the
dispersed fluid theory believed that the monotonic time dependence and the
unimodal relaxation spectrum with regard to the viscoelastic response of bitumen
were the basis to support such a homogeneous theory.
The most recent colloid theories, known as the Yen-Mullins model, support that
asphaltenes play a dominant role as ‘island’ structures in the molecular architecture of
crude oils and therefore also of bitumen [64,65]. Asphaltenes are believed to appear
in ‘islands’ which in sufficient concentration can form near-spherical nanoaggregates,
consisting of six to ten molecules stacked together followed by clustering in higher
concentration [65,66]. Validation of such disk-shaped, core-shell nano-sized structures
for the asphaltene nanoaggregates has also been provided experimentally with
advanced techniques of Small-Angle Neutron Scattering (SANS) and Small-Angle X-Ray
Scattering (SAXS) [67].
On bitumen surfaces, on the other hand, micro-sized, wave-like, rippled patterns have
been observed utilising different microscopic techniques [68,69]. These were referred
by Loeber and his peers as bee structures [70], and have been a matter of debate in
the research community concerning their origin and the underlying mechanisms. It has
long been considered that the bee-like structures are related to the asphaltene
fraction of bitumen [71], whereas other studies attribute their formation to a
crystallisation of the waxy components in bitumen [72–75]. Little research, however,
has been devoted to microstructural patterns observed in the bulk of the bitumen,
mainly studied in fractured bituminous samples. In these fracture surfaces, bee
structures were able to be captured only upon reheating [76].
Concerning the wax theory for the mechanisms behind the bee structure formation,
bitumen, apart from the solubility-driven SARA fractions, can include another
crystallisable fraction. This crystallisable fraction may be present in the heavy crudes
used to process bitumen or it may be generated during certain refinery procedures i.e.
visbreaking or hydrocracking in the type of paraffinic or microcrystalline wax [77–79].
Typically, paraffinic waxes consist of linear n-alkanes, while microcrystalline wax is
7
made up of aliphatic hydrocarbons with a considerable amount of iso- and
cycloparaffins [4]. The paraffinic waxes crystallise upon cooling in large, thin, flat plates
(macrocrystalline waxes) while the microcrystalline waxes form small micrometre-
sized crystals. Differences between the two wax types are observed in the molecular
weight and the boiling temperatures; microcrystalline waxes have a higher molecular
weight and boiling range [80]. Past studies have also reported different geometries,
shapes and sizes (1-10 μm) for the formed crystals [81,82].
In practice, Differential Scanning Calorimetry (DSC) is used as an effective tool to
determine the presence and properties of crystalline waxes, including glass transitions,
exo- and endothermic phenomena [74,83]. Alternative tools able to complementarily
provide deeper insights into the wax crystallinity exist but are rather uncommon for
bitumen. For example, 13C Nuclear Magnetic Resonance and Gas Chromatography
combined with Mass Spectrometry have hardly been used to characterise the presence
of crystallinity in bitumen [77].
Chemistry and performance of bitumen
Bitumen has generated much interest in the scientific community in an effort to
understand better its performance from different physicochemical and mechanical
perspectives [53,84]. Previous research has also stressed the importance of bitumen’s
chemistry for its performance [10,18,85].
The studies presented so far provide evidence that the ageing-produced ketones and
carboxylic acids are of high polarity, generating strong associations, expressed through
their Van der Waals forces. Subsequently, the polar compounds of bitumen may
interact with each other [86]. Possible chemical changes could induce stronger
interactions and change the bitumen structure which may have implications for the
mechanical behaviour. Given this, there is a growing body of evidence that an increase
in apparent molecular weight due to increased molecular interactions can reduce the
mobility of molecules to flow which, in turn, will influence the bitumen rheology
[10,87–89].
There is also a widespread recognition that the severity of ageing can be tracked by
capturing the change in certain functional groups [10,36,45,46,88]. Nevertheless, it
remains an open question whether the fundamental chemistry and rheology in
bitumen are affected at a similar level by oxidative ageing. Thus, many studies have
focused on the association between simply derived FTIR ageing indices which
represent the oxygenated functional groups, i.e. carbonyls, and parameters essential
for workability, such as viscosity, taking into account the ageing effect [5,6]. Similar
1. Introduction
8
efforts have been undertaken to link other cumulative FTIR ageing indices, i.e.
sulfoxides and carbonyls, with more sophisticated rheological indices derived with a
Dynamic Shear Rheometer (DSR), accounting for a single frequency or the whole range
of frequencies in the area under a rheological master curve [90,91]. In parallel, other
works have utilised the same techniques for differentiating the effects of different
long-term protocols in corresponding ageing parameters [92].
In addition, empirical parameters i.e. penetration and softening point, have their
merits but reach their limitations, when dealing with modified or special binders
containing additives and/or extenders, and thus are no longer considered to be the
best performance indicators [93]. Instead, advanced rheological parameters have
been introduced in the last decades, which not only are associated with basic bitumen
concepts, such as ductility and relaxation but also with distress phenomena and
concepts of durability, rutting and thermal-cracking [16,84,93–96]. It appears that in
order to examine the correspondence of the intensity of the ageing products and the
simultaneous evolution of advanced rheological indices, extensive fundamental
information of chemistry is needed besides the oxygenated FTIR functional groups. Of
particular complexity is to incorporate fundamental products, i.e. the free radicals,
which have been assumed to play a crucial role in the hypotheses for the mechanisms
of a potential ageing scheme [18,97]. One able to implement in a scheme such factors
will have at least a fair overview of the bitumen’s chemistry so that controlled
conclusions can be drawn for the corresponding changes of rheology.
When it comes to the effect of the SARA fractions in performance, via a systematic
blending of them it has been demonstrated that asphaltenes are primarily responsible
for the high viscosities and the non-Newtonian rheological properties of bitumen
[18,98]. By remixing different weight ratios of the asphaltenes and the maltenes
obtained from different source bitumen, the binders’ characteristic microstructures
have been found to correlate well with their bulk thermal and mechanical properties
[99]. It has also been shown that, among different fractions, asphaltenes display the
lowest temperature susceptibility [98,100,101], and significantly contribute to
bitumen’s stiffness, rigidity, and elasticity [27,85,102]. On the other hand, for a better
fatigue resistance, limiting the content of asphaltenes in bitumen would be rather
beneficial [103]. In a study conducted during the Strategic Highway Research Program
(SHRP), the thickening effect of asphaltenes has been examined in a set of bitumen
produced from different crude oils using different techniques [104]. By defining a
relative viscosity of bitumen as the quotient of the viscosity of the whole bitumen
divided by the viscosity of the maltene fraction at a given temperature and rate of
shear, the compatibility of the bitumen could be assessed. Evidently, a deep
9
understanding of asphaltenes and maltenes is essential to additionally establish the
relationships between bitumen chemistry and rheology.
Finally, the crystallisable compounds are of utmost importance to be investigated not
only for scientific reasons. From a practical point of view, in crude oils, the presence of
wax may relate to undesirable effects such as precipitation and pipe plugging or
deposition problems, depending on the applied temperature and pressure [105]. In
bitumen, the presence of waxes affects, among other properties, the stiffness at low
service temperatures which can induce slow hardening, increasing the risk for low-
temperature cracking at the corresponding asphalt mixtures. On the other extreme, at
high service temperatures, waxes in bitumen may result in additional softening, which
increases the risk of permanent deformation [106–109]. These indicate the
importance of wax crystallinity and knowledge of the melting and crystallisation
temperatures in bitumen.
Problem statement
Following the fast-evolving technological and environmental trends, the alarming need
for designing more sustainable and durable structures, including roads, has led
towards an attitude to prolong the lifetime of materials and improve their composition
in order to resist the majority of the possible deteriorative mechanisms. Furthermore,
as bitumen is a non-renewable organic material, a disciplined usage is required against
its depletion which is stimulated additionally due to ageing. However, a thorough
understanding of the fundamental mechanisms governing bitumen’s chemistry and
performance is still missing and hypotheses are yet to be reviewed and validated.
An improved, in-depth knowledge of bitumen oxidation is considered crucial for
reviewing the protocols for artificial lab ageing in order to adapt them in a controlled
manner. In addition, acquiring this knowledge may be the key answer to intervene into
bitumen’s composition more systematically in the near future i.e. via targeted
modifications which will prohibit certain oxidation paths of ageing and by
implementing more appropriate modifiers or rejuvenators when recycling in order to
reverse certain performance characteristics. In parallel, the underlying mechanisms
controlling the surface microstructure in bitumen have been confined to single
ambient temperature observations by limited techniques, while a real-time evolution
of microstructure is still missing. In addition, the effect of oxidative ageing on the
microstructure is often contradictory in the literature and it is limited in human-based
interpretations.
1. Introduction
10
Currently, a comprehensive and mechanism-based approach for modelling the
multiphysics ageing process in bitumen is not available. Existing efforts to model
oxidative ageing focus either on particular aspects at certain scales or employ a purely
phenomenological approach aiming to capture the entire process over all scales,
without accounting for compositional changes. Meanwhile, the scientific community
has not yet understood and implemented completely the basics of bitumen in such
phenomenological models in order to make the correct decisions that are of
paramount importance for the final output of interest, the performance of it.
Research objectives and scope
Hence, the scope of the current research project is to design a useful handbook for
every stakeholder dealing with bitumen and is interested in exploring the ‘why’ behind
certain changes related to ageing and the microstructure of bitumen which have a
direct effect on its behaviour.
Contrary to the brute-force approaches used up to now in the literature the
methodology adopted in this doctoral dissertation initiates from the molecular level
and scales up to the microscale of bitumen (Figure 1-1). More specifically, this research
project investigated and attempted to give answers to a series of unexplored,
conflicting and open questions that exist in the literature up to this moment.
The main objectives of this dissertation are summarised herein.
1. It attempts to employ a number of state-of-the-art experimental techniques
as the ground truth of the hypotheses proposed so far for the ageing
Meso
Macro
Micro
Molecular
Figure 1-1 The hierarchical scales of asphalt ageing
11
phenomenon in oxidation kinetics and also by utilising standardised lab ageing
simulations.
2. Via ageing kinetics, it estimates fundamental properties of the coupled
contributions of reaction and diffusion of oxygen in bitumen.
3. It aims to investigate how the polarity-based fractions of bitumen are
chemically affected upon ageing.
4. The dissertation discusses the main oxygenated products in bituminous
binders accounting for variability between the selected materials in terms of
type, refinery process and origin.
5. It provides support for the mechanisms of the bitumen’s microstructure and
offers a sophisticated way to comprehensively evaluate the effect of ageing on
certain morphological patterns.
6. This thesis seeks to elucidate the influence of different laboratory ageing
protocols on chemistry and rheology and to provide a framework to link them.
7. After reviewing the ageing mechanisms, efforts to implement this
compositional information in a mechanism-based model accounting for
diffusivity and reaction completion time, are undertaken.
Research outline
The dissertation is outlined in nine Chapters, while the content of the thesis
conceptually is distinguished into five main parts. All of them are illustrated
schematically in Figure 1-2. Part I introduces the reader to the topic and the problem
statement of this work (Chapter 1) and it provides the research methodology used in
this study versus the possible obtained results (Chapter 2).
In part II, the main laboratory oxidation products for a variety of conventional
bituminous binders differing considerably in terms of origin, distillation process and
composition are explored in Chapters 3 and 4 by FTIR, Electron Paramagnetic
Resonance (EPR), Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) and
Proton Nuclear Magnetic Resonance (1H-NMR) spectroscopy. In parallel, back-
calculation of diffusivity and solubility of oxygen as well as the activation energy via
kinetics is performed by utilising a Dynamic Vapour Sorption (DVS) analyser and thin
film oxidation kinetics, respectively. In Chapter 4, the same binders are investigated in
the lab with RTFOT and PAV. In the same Chapter 4, for two of the aforementioned
1. Introduction
12
binders extensive chemical analysis, performed additionally in its bitumen fractions
before and after PAV, is also discussed.
Part III of this book dives into the mechanisms accompanying the changes in bitumen’s
surface microstructure for two bituminous binders differing in compositional
information in terms of natural wax. In Chapter 5, Differential Scanning Calorimetry
(DSC) and Wide Angle X-ray Diffraction (WAXD) are combined with optical
observations via a CLSM in an effort to link the thermal transition phenomena, the
crystallinity of bitumen and the observed microstructural patterns of it, known as the
bee structures. In the same part III of the dissertation in Chapter 6, a wider range of
the two types of binders are examined for their microstructural properties related to
the bee area percentage, waveform characteristics and probabilistic values of the bee
structures. Commercial software packages and deep learning techniques are
employed for the bee structure identification. Coupling of Confocal Laser Scanning
Microscopy (CLSM) and Atomic Force Microscopy (AFM) demonstrates a clear effect
of ageing on specific traits of these patterns.
Part IV of this book in Chapters 7 and 8 respectively, directs to the phenomenological
and mechanism-based modelling approaches performed. In Chapter 7 the
fundamental information obtained for the three binders in part II is first compared
with regard to the convergence of different ageing protocols for their chemistry and
rheology. Various statistical methods (Bivariate analysis, Wilcoxon exact Rank-sum
test, Factor analysis) assisted in identifying relationships among the chemo-rheological
parameters. The work is completed in Chapter 8 with the first efforts to model in a
Thermodynamics of Irreversible Processes (TIP) model the ageing mechanisms
supported in part II for bitumen, accounting simultaneously for its fractions.
Finally, in Chapter 9 (part V), the main conclusions extracted from this dissertation are
reviewed and the way to assist in future research are discussed. Recommendations for
future work are proposed as well as suggestions to link the current observations with
field performance.
13
Figure 1-2 Schematic structure of the research
1. Introduction
14
*This chapter is redrafted from: G. Pipintakos et al., Application of Atomic Force (AFM), Environmental
Scanning Electron (ESEM) and Confocal Laser Scanning Microscopy (CLSM) in bitumen: A review of the
ageing effect, Micron (2021). https://doi.org/10.1016/j.micron.2021.103083
2
2 RESEARCH METHODOLOGY
Summary*
In this Chapter 2, the research methodology adopted during this project is scrutinised
with respect to its working principles and capabilities for bitumen. The theoretical
background of the techniques and analyses used is briefly explained so that the reader
becomes familiar with them. An overview of the usage of the most appropriate
techniques in order to evaluate the ageing process in bitumen both from a chemical
and rheological perspective, is given in Table 2-1. Finally, the challenges to harmonise
the information obtained between the various techniques are acknowledged and
discussed.
Table 2-1 Overview of experimental techniques in this dissertation and target goals
Experimental technique
Purpose of use
FTIR
Oxygenated products, oxidation phases and activation energies
EPR
Identification of free radicals and oxidation phases
1H-NMR
Chemical classification of proton regions
TOF-SIMS
Oxygenated products
DVS
Diffusivity and reactivity
DSC
Thermal transitions
WAXD
Crystallinity
AFM
Identification of bee structures and their characteristics
CLSM
Identification of bee structures and their characteristics
DSR
Rheological characterisation
2. Research methodology
16
Material selection
For a holistic viewpoint of the ageing process, it is necessary to vary the type of the
examined binders so that the obtained results can be generalised as much as possible.
With this in mind, the binders selected for all the subsequent experimental techniques
were varying in four main aspects: 1. their empirical performance, 2. their origin of
crude source, 3. their wax presence and 4. their manufacturing process. To better
accommodate for the reader the association between the results of the performed
experiments and the type of binder, the exact details of the binders are given in the
corresponding section of ‘Materials and Methods’ of each Chapter.
Spectroscopic techniques
Fourier Transform Infrared (FTIR) spectroscopy
2.3.1.1 Working Principles
The general operation of FTIR is based on the absorption and interaction of infrared
radiation with the chemical bonds of the material under investigation. An infrared
beam is emitted to the specimen and activates certain chemical bonds in its molecules,
which start to vibrate at specific frequencies. Thus, the change of the bond or structure
at the molecular level is depicted by the infrared absorption during vibration, which is
related to a specific frequency for each bond. The vibrations of these bonds are crucial
for identifying the type of functional groups in a material, which works as a kind of
‘chemical fingerprinting’. These vibrations can be either stretching of the atoms in a
molecule or bending by means of a change in angle between two bonds.
The two general modes used for performing FTIR analyses are in transmittance and in
reflection. In the first mode, the radiation is sent through the sample and specific
wavelengths are absorbed. For the second mode, typically, the addition of an
Attenuated Total Reflectance (ATR) fixture allows total reflection, as radiation enters
the FTIR crystal and reflects off the internal surface in contact with the sample. The
reflection can be a single or multiple ones depending on the utilised crystal. In this
mode, in the case of multiple reflections, the infrared beam is reflected by the sample
multiple times inside the crystal. It should be noted that the infrared wave is capable
to penetrate only a few μm in the sample, and the prism is manufactured in such a way
to ensure the total reflectance of the beam. Commonly used crystals of diamond and
germanium have refractive indices which are larger than the testing materials to
ensure internal reflectance. Finally, the sample absorbs specific frequencies and the
17
beam loses energy at this particular frequency for each functional group. The radiation
after reflection is collected by a detector connected to a software which can
demonstrate in a spectrum the wavenumber against absorbance or transmittance.
2.3.1.2 Bitumen application
Undeniably, FTIR is a versatile, easy-going spectroscopic tool not only for chemists and
bitumen producers but also for road engineers in order to evaluate, among others, the
ageing severity of bitumen and to trace certain additives. For the ageing severity, the
functional groups of sulfoxides and carbonyls are more often used as indices whereas
other indices are used to track i.e. the aromatisation process in bitumen.
In the current research project, the FTIR analysis was performed in reflection mode
with a Thermo Scientific Nicolet iS10 and iS50 FTIR spectrometer equipped with an ATR
fixture and a Smart Orbit Sampling Accessory. At least three replicates were measured
per sample in different ageing states used in this study. The collected spectra were
acquired with 32 repetitive scans and ranged from wavenumbers between 400 cm-1
and 4000 cm-1 with a resolution of 4 cm-1.
After the spectra acquisition, the protocol as described in [45] for the determination
of the normalised intensity of certain functional groups, was followed. More
specifically, the areas around certain peaks of an infrared spectrum are calculated. To
do this, a baseline is introduced based on the limits of each band given in Table 2-2
and the area that is enclosed is computed based on the trapezoidal rule that
approximates a definite integral. For the quantification of the corresponding
normalised index, Equation 2-1 to Equation 2-4, are finally applied.
Table 2-2 Band limits of the utilised FTIR functional groups [45,110,111]
Functional group Bond vibration
Band limits for
baseline (cm
-1
)
Area around
peak n, (An)
Long chain alkanes
(CH2)n rock (n ≥ 4) bending
734-710
724
Aromatic structures C=CH adjacent out of plane
bending
783-734 743
Aromatic structures C=CH adjacent out of plane
bending
838-783 814
Aromatic structures C=CH singlet out of plane
bending
912-838 864
Sulfoxides
S=O stretching
1047-995
1030
Branched aliphatic
structures
CH3 symmetric bending 1390-1350 1376
Aliphatic structures
CH3 asymmetric bending
1525-1395
1460
Aromatic structures
C=C stretching
1670-1535
1600
Carbonyls
C=O stretching
1753-1660
1700
2. Research methodology
18
Aliphatic structures
C-H symmetric stretching 2880-2820 2862
Aliphatic structures
C-H asymmetric stretching
2990-2880
2953
1030
norm. sulfoxide index (SOI)
n
n
A
A
=∑
Equation 2-1
1700
norm. carbonyl index (COI) n
n
A
A
=∑
Equation 2-2
1376
1460 1376
norm. branched aliphatic index (BAI) A
AA
=+
Equation 2-3
1600
norm. aromaticity index (ARI)
n
n
A
A
=∑
Equation 2-4
Electron Paramagnetic Resonance (EPR) spectroscopy
2.3.2.1 Working Principles
EPR is a spectroscopic technique able to identify paramagnetic centres and molecules
in a material, i.e. the components containing unpaired electrons. It is particularly
useful to characterise organic radicals and transition-metal ions. The basic idea of this
technique is based on the fact that atoms, ions, molecules or molecular fragments
which have an odd number of electrons exhibit characteristic magnetic properties. In
parallel, it is known that each electron is a charged particle which spins around its axis
causing a magnetic moment (intrinsic property) and orients when placed inside a
strong magnetic field of an EPR, where electromagnetic radiation is applied
monochromatically. The alignment of the unpaired electrons creates two spin states
which have also different energies. Hence, an electron can move between the two
energy levels by absorption or emission of a photon of energy when resonance
conditions are fulfilled. Although different frequencies and magnetic field values can
in principle be used the majority of EPR experiments are performed in the X-band (9-
10 GHz) by keeping the photon frequency fixed and varying the external magnetic field
incident on the sample under investigation until it matches the energy of the
microwaves in the X-band. In this state the unpaired electrons can move freely
between the two spin states and the absorption that is recorded is converted into a
spectrum.
19
Concerning the different existing EPR types, these can be mainly categorised in high-
field high-frequency, pulsed and Continuous-Wave (CW) measurements depending on
the source of radiation and the type of it. Many EPR applications make use of CW EPR
as it has the advantage over pulse EPR to detect with high sensitivity at room
temperature and is less expensive than pulsed EPR.
2.3.2.2 Bitumen application
CW EPR spectra of bitumen samples were recorded in this research project with a
Bruker Elexsys E680 spectrometer mounted with an ER 4102ST TE102 mode resonator
working at ⁓9.75 GHz (X-band). Polypropylene Eppendorf tubes were used as sample
holders ensuring that the total bitumen quantity did not overfill the cavity. Preliminary
measurements in the studied binders showed a two-component EPR spectrum
consisting of contributions of a vanadyl centre (VO2+, S = 1/2) and an organic carbon-
centred radical. The former is characterised by an axial powder pattern (Figure 2-1
[left]) with typical hyperfine splitting due to the interaction of the electron with the 51V
nucleus (I = 5/2), and the latter gives rise to an unresolved single line close to the free
electron value (ge = 2.0023).
Figure 2-1 Room temperature CW X-band EPR spectra of indicative bitumen sample with its simulations
[left] and its signal intensity contrasted with the square root of the power at different g-values, with
insets showing a linear fit of selected powers [right]
Power saturation measurements showing different relaxation rates (Figure 2-1 [right])
confirmed the presence of the two species, namely organic carbon-centred radicals
and VO2+. For all the subsequent measurements, 0.5 mW was chosen, because this was
close to the highest microwave power level before either species became saturated
(i.e. signal intensity vs √microwave power was linear). At least three replicates were
measured per sample state with the centre field at 341 mT, sweep width of 20 mT,
2. Research methodology
20
resolution of 2048 points, modulation amplitude of 0.1 mT and modulation frequency
of 100 kHz over 2 scans. Simulation of the experimental spectra with Matlab2018b
using the EasySpin-6.0 module [112], gave the EPR parameters of the two species as
well as the relative amounts of spins between the two (‘weights’). Next, the number
of spins in each sample was estimated by comparing the double integral of each
spectrum with those of TEMPO (2,2,6,6-tetramethyl-1-piperidinyloxyl, a chemical
compound used commonly as a structural probe for radicals’ characterisation) in
toluene solution standards. More specifically, the number of VO2+ or organic radical
spins in each sample was estimated by comparing the weights used in the spectral
simulations. As last step, division by the mass of the sample gave the number of spins
per gram of sample.
Time-of-flight Secondary Ion Mass Spectrometry (TOF-SIMS)
2.3.3.1 Working Principles
The analysis principle of typical TOF-SIMS includes the bombardment of the outermost
surface layers of a sample with high-energy primary ions and analysis of the emitted
(secondary) ions with respect to mass-to-charge ratio (m/z). In general, the emitted
particles by the collision process of the primary ions vary considerably and can include
atoms, molecular fragments, intact molecules and molecular complexes which when
energetically charged can be analysed by a TOF analyser providing eventually a mass
spectrum.
The obtained mass spectra provide molecular information about the sample surface
whereas imaging of specific ions is accomplished by scanning the primary ion beam
over a selected analysis area and acquiring separate mass spectra in each pixel [113].
In principle, the selection of the number of pixels in the TOF-SIMS analyses, i.e. the
number of data points within the analysis area, is based on the fact that the pixel size
should be matched with the diameter of the primary ion beam. The acquired data can
be displayed in a variety of ways, i.e. total area mass spectra and ion images that show
the spatial signal intensity distribution of selected secondary ions on the sample
surface by means of chemical mapping. Another option is to display mass spectra from
selected regions of interest, providing detailed chemical information about specific
structures within the analysis area.
2.3.3.2 Bitumen application
In this research project, the TOF-SIMS analyses were conducted in a TOF-SIMS IV
instrument (IONTOF GmbH, Germany) using 25 keV Bi3+ primary ions and low-energy
electron flooding for charge compensation. During analysis, the sample temperature
21
was kept between -50 °C and -80 °C to prevent diffusion/segregation of the sample in
the vacuum environment of the TOF-SIMS instrument. For the sample preparation,
bituminous samples of 1 mm thickness were deposited on silicon wafer substrates 10
mm x 10 mm and subsequently allowed to cool down according to the protocol
reported in [4]. The surface information obtained by Bi3+ primary ions penetration has
been estimated to be just a few nanometres [113].
Positive and negative ion spectra were acquired over analysis areas of 200 µm x 200
µm (128 x 128 pixels) or 500 µm x 500 µm (256 x 256 pixels) at three locations of each
sample, with the instrument optimised for high mass resolution (m/Δm ≈ 3000-6000).
The spectra for all the samples were analysed based on the intensity of the areas of
selective mass-to-charge-ratios, assigned to specific ions, normalised by the total
recorded ions’ intensity (counts). To obtain more in-depth insights into the chemical
mapping in selected specimens additional positive and negative ion data were
acquired (i.e. high-resolution images) with the instrument optimised for high image
resolution (lateral resolution ≈ 0.5 µm). In all the cases, the instrument was controlled
and the data was acquired by the IONVAC software (IonTOF), while they were
processed using Surfacelab 7 (IonTOF).
Proton Nuclear Magnetic Resonance (1H-NMR) spectroscopy
2.3.4.1 Working Principles
NMR spectroscopy was initially developed to study the properties of the atomic nuclei
whereas later was realised that can be used to study the structures of organic
compounds. The principle is based on the fact that nuclei, such as hydrogen, which
have an odd number of protons and/or an odd number of neutrons possess a property,
called spin, which allows them to be studied by NMR quite similar to the principles of
EPR. In other words, a nucleus with a spin has a magnetic moment and generates a
magnetic field which when placed between the poles of a strong applied magnetic field
makes the nuclei align with or against this applied field, while the difference between
the two spin states depends on the strength of the applied magnetic field and
successively on the frequency of the spectrometer. Flipping the spin states of nuclei
requires an amount of energy to be absorbed which is the radiation in the radio
frequency region of the electromagnetic spectrum. This radiation absorption
generates signals whose frequency depends on the difference between the spin states.
Hence, the chemical shift of a particular nucleus results from the modification of the
local magnetic field at the location of the nucleus by the shielding of the external
magnetic field by the electrons around the nucleus, which is dependent on the binding
2. Research methodology
22
state of the atom in question. Finally, an NMR spectrometer is able to detect these
signals and depict them in a graph of frequency versus intensity, known as the NMR
spectrum. In modern pulsed NMR, the magnetic field is held constant while the radio
frequency pulse covers a range of frequencies so that all the nuclei can come
eventually into resonance. Most frequently, the NMR spectra are recorded in solutions
with deuterated solvents. It is the deuterium signal that is later used to monitor the
difference with the onset of the chemical shifts. It is finally worth mentioning that the
origin of the terminology of nuclear magnetic resonance derives from the fact that the
nuclei are in resonance (flipping their spin state) with a corresponding radio frequency
radiation.
2.3.4.2 Bitumen application
1H-NMR is a superior spectroscopic technique able to characterise the molecular
structure of a material, even for a complex one, such as bitumen. It can assist with the
characterisation of the relative amount of different types of aliphatic, olefinic and
aromatic protons in bitumen [114–116] and therefore can also be applied to capture
the proton distribution upon ageing in bitumen. In this research work, data collection
of all the bituminous binders at room temperature was conducted with a high-
resolution liquid-state Bruker Avance III HD 400 MHz smart probe spectrometer (32
scans). Preliminary measurements with CDCl3 (deuterated chloroform) as solvent
showed interference with the aromatic region signal of bitumen, while different
concentrations of dissolution with 5 and 40 mg of bitumen showed a poor and an
overloaded 1H-NMR signal respectively. Therefore, ± 20 mg of bitumen or its fractions
were dissolved in 650 μl of deuterated tetrachloroethane (C2D2Cl4) inside borosilicate
NMR tubes (ASTM Type 1 Class B glass) of diameter 5 mm and a wall thickness of 0.4
mm. To ensure an adequate dissolution of the solvent with the specimen all vials
containing the dissolved samples were additionally placed inside an ultrasonic water
bath for 10 minutes before the 1H-NMR analysis. The repeatability was provided by
means of standard deviations.
After the analysis, steps were first taken to accurately calibrate the starting chemical
shift based on the difference of the residual solvent peak (6.00 ppm) with respect to
TMS (tetramethylsilane) (0.00 ppm) [117]. The analysis of the results was performed
in the MestreNova spectral analysing software following the integration of the typical
proton chemical shift regions reported in [118], neglecting each time the protons
associated to heteroatoms and the residual solvent at 6.00 ppm. The latter study
classifies a typical 1H-NMR spectrum of bitumen in five main groups, given in Table 2-3.
Normalisation of all the integrated areas with the exact sample mass allows for a fair
23
comparison of the relative percentage distribution of the different types of protons
identified in bituminous samples.
Table 2-3 Typical groups of protons in bitumen
Gravimetric techniques
Dynamic Vapour Sorption (DVS)
2.4.1.1 Working Principles
DVS is an automated gravimetric analyser able to determine with high precision mass
balance changes of resolution of ±0.1 mg due to a gas flow in the sample under
investigation. Modern DVS devices allow the calibration of the humidity level, partial
pressure, the mix of different gasses and the temperature control under which a
specified program is executed in a predefined time. It is a particularly useful device to
examine vapour sorption/desorption, adsorption and absorption phenomena of a
material.
2.4.1.2 Bitumen application
In this work access to DVS, Surface Measurement Systems was provided by the
Pavement Engineering research group of Delft University of Technology. The DVS was
utilised to study the uptake of oxygen vapour into bituminous mastics. To that end,
the relative humidity was kept to a minimum in the samples prior to testing by placing
them inside a vacuum environment with silica gel for one week before the experiment.
The humidity was recorded during this drying stage to ensure that there was no effect
of induced humidity in the subsequent gravimetrical changes. This was only possible
with bituminous mastics and not with bitumen. The samples and the reference pan of
the DVS were successively exposed to dry air (20% O2 + 80% N2) for a total period of 7
days under isothermal conditions at 75 °C. Preliminary studies with the added filler of
the mastics under the exact same conditions showed no change of mass due to dry air,
whereas an increase of the initial mass of mastics was observed for the bituminous
Designation
Chemical
shift range
Type of protons
Major proton peak
in this region
Hmethyl 0.5-1.0
Aliphatic hydrogen on C
γ
and the CH
3
beyond the
Cγ to aromatic rings
Methyl
Hmethylene 1.0-2.0
Aliphatic hydrogen on C
β
and the CH
2
beyond the
Cβ to aromatic rings
Methylene
Hα-alkyl
2.0-4.0
Aliphatic hydrogen on Cα to aromatic rings
-
Holefinic
4.0-6.0
Olefinic hydrogen
-
Haromatic
6.0-9.0
Aromatic hydrogen
-
2. Research methodology
24
mastics. In the current research project, this was attributed completely to oxygen
reaction-diffusion in bitumen.
Thermoanalytical techniques
Differential Scanning Calorimetry (DSC)
2.5.1.1 Working Principles
Nowadays the three most common types of DSC are heat-flux, power compensation
and fast-scan. Whatever the type is, the working principles are somewhat similar for
the material characterisation and vary with respect to the sensor and supplied power
to the specimen. In general, with this thermodynamical tool, a specified temperature
program is applied by means of heat flow both to the investigated sample and a known
reference material and the amount of heat difference to match the exact same
temperature between the two is measured as a function of temperature. DSC enables
the investigation of endothermic or exothermic phase transition phenomena (melt,
crystallisation, demixing, remixing and others), second order phenomena (glass
transitions) and chemical reactions (thermal curing, specific heat capacity and others).
At the end of the temperature program, the recorded data of a DSC result in a curve
of heat flux versus temperature known as thermogram which can be also used to
calculate the enthalpy of a specific transition.
2.5.1.2 Bitumen application
A TA Instrument, model 2920 DSC was used in this research project. For each analysis,
samples of approximately 15 mg were hermetically sealed into a sample pan, then
heated to 110 °C and kept at this temperature for 15 minutes. The data were first
recorded during cooling to -110 °C and heating to 110 °C, both at a rate of 10 °C/min.
To examine the effects of annealing for DSC, thermograms were also acquired for
selected samples in the same cooling and heating cycles, after 24 hours at 25 °C and
the data of the heating ramp were recorded. The measured thermal characteristics of
bitumen before and after the annealing period of 24 hours include onset, end and peak
temperatures as well as enthalpies for the wax crystallisation and melting. For the
quantification of the enthalpies, integration of the area enclosed between the
experimental DSC curve and a linear extrapolation from the melt data was performed.
Additionally, the glass transition temperatures were reported from the same
thermograms.
25
Wide Angle X-ray Diffraction (WAXD)
2.5.2.1 Working Principles
This technique is commonly used to obtain information for crystalline materials by
making use of the fact that these ordered patterns cause a beam of incident X-rays to
diffract in specific directions. The analysis of the scattered angles and intensities of the
diffracted rays by a detector can be used to obtain information about the cause of
them which is commonly ascribed to nanosized structures. The principle of the
technique is based on the scanning of the sample with a wide-angle goniometer where
the scattering intensity is typically plotted as a function of the 2θ angle. With WAXD
the degree of crystallinity can be estimated at a specific temperature or in a range of
temperatures during cooling/heating scans. Additionally, the patterns of the internal
structure of the crystal can be determined.
2.5.2.2 Bitumen application
To collect the WAXD patterns, a X33 double-focusing camera of the EMBL in HASYLAB
was used on the storage ring DORIS of the Deutsches Elektronen Synchrotron (DESY),
Hamburg, Germany, at a wavelength of 1.5 Å [119]. The bituminous samples in their
liquid state at high temperature were inserted in small sample holders and sealed with
aluminium foil [120]. After cooling to room temperature, the samples were stored for
3 days at room temperature prior to collecting a WAXD pattern at 25 °C. In addition,
WAXD patterns were collected during a cooling/heating scan at 10 °C/min from 100 °C
down to -40 °C and again up to 250 °C after 2 min at -40 °C. For this time resolved
measurement, patterns were collected in consecutive frames of 6s, which
corresponded to a temperature resolution of one pattern per °C. The temperature was
controlled by a Mettler FP-82HT hot stage, flushed with cooled nitrogen to achieve the
required cooling rate. After a background correction, the WAXD patterns were
normalised to their integral to ensure a constant scattering mass. A crystallinity index
was extracted from the normalised patterns by integrating the intensity of the 110 and
200 orthorhombic reflections, separated from the liquid scattering using linear sectors.
For the calibration of the scattering angles, the 110 and 200 reflections of a quenched
linear polyethylene were used. The obtained data were treated as a function of the
scattering angle but converted to the scattering angle expected in the case CuKα
radiation would have been used (λ = 1.542 Å) instead of λ = 1.5 Å. Such a
representation facilitates comparison with literature data, which most often make use
of CuKα radiation.
2. Research methodology
26
Microscopic techniques
Atomic Force Microscopy (AFM)
2.6.1.1 Working Principles
Imaging with the AFM microscope has evolved over the past years but the general
working principles remain the same, rooted back in its invention in 1986 [121]. More
specifically, the imaging of the target uses a cantilever, with a sharp tip that is used to
scan the sample surface by interacting with it. When the cantilever experiences a force
between the tip and the sample, measured by a laser beam that is placed at the end
of the cantilever, it deflects and shifts up according to the Hooke’s law. This results in
a deviation of the laser beam from its original position, which is measured as a voltage.
This voltage is translated into a variety of forces (e.g. mechanical contact force,
capillary forces etc.) or relative height and an image can be made based on all the
measured differences [121,122].
There are different operational modes of an AFM, but among others (near-contact,
pulsed force, lateral force mode) the most common ones are the dynamic (tapping)
and the static (contact) mode [21]. With contact mode, one refers to the mode where
the tip stays in contact with the sample at constant load, while the surface is moving
in the plane directions. This mode is able to generate both frictional and topographic
data. Tapping mode, on the other hand, occurs when the cantilever, oscillating up and
down at its resonance frequency, is gently tapped on the surface to reach contact and
then shifts in the plane directions when the tip is lifted away from contact. This mode
provides phase and topographic data and generally reduces the risk of destroying the
surface compared to the static mode where the developing forces are higher [122].
2.6.1.2 Bitumen application
The surface microstructure of bitumen is still a subject that is not fully understood and
completed. Over the years, there were a lot of attempts to examine the microstructure
of bitumen, with the most commonly used technique being the optical microscope.
AFM proved to be a promising alternative and was adopted by bitumen researchers
soon after its invention [70]. The results give information about the topography and
phase contrast of the sample and are therefore especially useful for multi-phase
materials such as bitumen. Using AFM, experts aimed to characterise the topography,
stiffness, tackiness and molecular interaction at the micro-level of materials [123]. In
parallel, mechanical properties of bitumen such as adhesion, rigidity, hardness and
modulus of elasticity can be estimated by employing AFM with the nanoindentation
27
technique [124]. This technique penetrates the bitumen surface with a tip of defined
geometry [125].
The fact that AFM is time-efficient and has a relatively simple sample preparation,
makes its use favourable. Attention should be though given to the consistency of the
preparation procedures as it may introduce differences if the handling procedure and
storage conditions vary significantly between the samples. In general, sample
preparation can be done by two different techniques for bituminous binders, either
the spin or the heat casting method.
Spin casting includes a spinning plate, in which the sample is cast on. The centrifugal
forces of the plate provoke the sample to spread evenly so that it eventually results in
a thin film for the imaging. Beforehand, the bitumen is dissolved in a solvent which
evaporates after the spinning. Allen and his coworkers suggested the elimination of
the residual solvent to preserve the sample in an airtight heated vacuum desiccator
followed by a purification step with dry nitrogen [126]. For the heat casting method,
the bitumen is heated and stirred at a temperature of 110-130 °C, according to the
bitumen type to be workable, and profoundly mixed so that a specimen has no more
oxidants or dust compared to a replica of the same bitumen tank [74]. Next, a bitumen
drop of 15-30 mg is placed on a conductive sample holder and held horizontally on a
heating plate (set at the same temperature as the heating temperature of bitumen) so
it would spread evenly and become a flat surface. The samples are normally tested
directly in the AFM and then held in a dust-free environment before testing again at
different time intervals so that the microstructures can settle and their evolution can
appear on the bitumen’s surface.
In the current research project, AFM imaging was recorded using an Asylum Research
MFP-ED AFM in tapping mode. A resonance frequency 300 kHz and spring constant 26
Nm-1 of an AC 160 TS cantilever tip were utilised for this study. With regard to the
instrumental settings, the vertical Z-axis displacement of the measuring head is
constrained to 15 μm, while the sensor noise is less than 25 nm with an average
deviation in a 0.1-1 kHz bandwidth for this device. In order to scan the bitumen surface,
X and Y activators are used with a travel distance limited to 90 μm, with sensor noise
in these directions less than 0.5 nm as an average deviation in a 0.1-10 kHz bandwidth.
All the binders were recorded in a scan size between 20x20 to 40x40 μm so that for
each sample the number of observed microstructures has a statistically meaningful
calculation.
To minimise the effects of thermal history and other effects like the nature of the
sample substrate [73,127], the preparation method in [69] was adopted. In this
procedure, a small quantity of bitumen is heated to 160 °C for 15 minutes and after
2. Research methodology
28
proper stirring with a metal spoon, a hot droplet is placed on a 76x26 mm pre-cleaned
glass slide (Thermo Scientific) with ground edges and a frosted end. To achieve a flat
surface the glass slide with the droplet is reheated on a hot stage for one minute. The
samples are then let to cool down and stored in a dark and dust-free place before the
measurements. All the scans are performed in air at 25 °C after a storage time of 24
hours, to allow the microstructures to develop and stabilise [73]. The analysis and
processing of the AFM images were performed using the open-source, image editing
software package Gwyddion (version 2.58) [128].
Confocal Laser Scanning Microscopy (CLSM)
2.6.2.1 Working Principles
The CLSM has become a popular method in biology, biomedical, and material sciences
in the past few decades [129,130]. The basic concept of confocal microscopy was first
developed and patented by Marvin Minsky in 1957 to image the neural networks in
unstrained preparations of brain tissues [131]. Unlike conventional optical
microscopes, confocal microscopes apply a spatial pinhole in front of the detector to
focus the laser beam on just one predefined depth. It was during the 1970s and 1980s
that the advances in computer and laser technology led to the development of the
CLSM [132]. In CSLM, the light source, which is a laser beam, based on depth selectivity
allows for optical sectioning. The information gained from this focal point is projected
on a pinhole in front of the detector, which ensures that only the light from the small
area of the sample, which is irradiated, is detected [133]. Apart from the reflectance
or transmission mode, CLSM can also operate in fluorescence mode. This is done by
using a different laser light source with a lower wavelength. In fluorescence mode, the
molecules absorb the high energy (short wavelength) light and after a short lag period
(fluorescence lifetime) emit a lower energy (longer wavelength) light.
Comparison of images taken by conventional optical microscopy and CLSM in reflected
light mode has indicated the higher resolution and thus the better quality of the
images obtained with CLSM [134]. Finally, in a CLSM, the image is created by scanning
the surface point by point. If this is done in the x-y plane for different depths in the z-
direction, the 2D images can be reconstructed rather fast into a 3D representation and
without direct contact with the surface, which is the biggest advantage of CLSM over
other microscopic techniques [134]. By successive scanning, at different focal points,
a topographic image is finally obtained [132].
29
2.6.2.2 Bitumen application
CLSM is a relatively new technique to investigate the microstructure of bitumen. This
technique has been used both in reflectance and fluorescence mode on bitumen. Next
to its capability of creating 3D reconstructions, using CLSM to investigate bitumen
requires little pre-treatment.
CLSM in reflectance mode has been used by researchers to determine the size,
distribution, and shape of the asphaltene particles [134] or the classification of wax
morphology [81]. Furthermore, in the last few years, researchers started to use this
technique to observe fluorescence centres in bitumen. It was demonstrated that
bitumen exhibits fluorescence when irradiated with 488 nm wavelength light [135].
Therefore, to investigate bitumen using CLSM in fluorescence mode, the samples are
typically irradiated with 488 nm wavelength laser light, while observation of emitted
signals is mostly done in the 500-550 nm wavelength range [136–138].
Other than using CLSM to investigate virgin bitumen, this technique was also used to
study the morphology of polymer-modified bitumen [139–141] and recycled asphalt
shingle blended with asphalt binder [137]. Moreover, CLSM has been widely used to
explore the morphology of epoxy asphalt binder [142], phase-separated
microstructure and dispersion of the asphalt rubber in epoxy asphalt [143,144], as well
as the morphology and phase separation of polymer-modified epoxy asphalt binders
[145–147].
In the current study, a Keyence VK-X1000 CLSM was used, mounted with a VK-D1
motorised XY-stage. Recording of topographical images was conducted with a laser
and white light source. Among a variety of available lenses, the Nikon Lens Plan Apo
EPI with a magnification of 150 times, 5 nm lateral resolution and 10 nm axial
resolution was utilised. The lens magnification was chosen so that the obtained images
were of a similar scan size and contain a statistically meaningful number of bee
structures. All the images were captured in ambient conditions using the same sample
preparation and storage time as for the AFM measurements. The images were further
processed with the commercially available software package VK MultiFileAnalyzer
(version 2.2.0.93). During image processing, a surface shape correction was applied as
well as a smoothing of the surface to eliminate any noise.
Rheological characterisation
Dynamic Shear Rheometer (DSR)
2. Research methodology
30
2.7.1.1 Working Principles
This type of testing is based on the oscillation of the upper of two parallel plates in a
DSR where a sandwiched sample is placed in between, while the base plate remains
clamped. The sample is subjected to sinusoidal torque or sinusoidal angular
displacement of constant frequency or strain under a specified temperature,
frequency/strain program. As such, rheological properties can be extracted as the
result of the applied torque and displacement, converted into shear stress or strain
respectively. The ratio of these amplitudes is then used to calculate the norm of the
complex modulus (G*) and the phase angle (δ) of the investigated material in a range
of temperatures and frequencies (under a strain-controlled test). Apart from the norm
of the complex modulus, other basic dynamic parameters of paramount importance
are the storage or elastic modulus G’ (in-phase component of G*) and the loss modulus
G’’ (out-of-phase component of G*). From a practical point of view, the appropriate
selection of the controlled stress or strain is crucial to fulfil the linearity requirements
of the viscoelastic range within which the majority of rheological tests are performed.
Thus, DSR gives an indication of the resistance of the material to deformation due to
stress, understood by the complex modulus. In parallel, one of the primary functions
of a DSR is to address the nature of the material varying from viscous to elastic
depending on the temperature and frequency. For example, in low phase angle values,
the material behaves as an elastic solid, whereas a high phase angle indicates a
behaviour of a viscous fluid. These parameters are often not the last step but rather
an intermediate one by making further use of the time-temperature superposition
principle (TTSP), when applicable. More specifically, the data extracted from
isothermal plots can be shifted, based on the TTSP, and further used to model the
complex modulus and the phase angle in a wider range of frequencies/temperatures
with a continuous curve, known as a master curve.
2.7.1.2 Frequency sweeps
Two DSR devices, an Anton Paar MCR 101 and an MCR 500 were used for the
rheological assessment of the bituminous binders. The former was used for the low-
temperature range, -30 °C to +10 °C, with a plate-plate geometry of 4 mm diameter,
whereas the latter was used for higher temperatures ranging between 0 °C to +40 °C
and +30 °C to +70 °C with 8 and 25 mm diameter of plates respectively. Strain levels of
0.02 %, 0.05 % and 1 % with increasing diameter were found adequate to perform
strain-controlled frequency sweeps within the linear viscoelastic region (LVER).
Therefore, frequency sweeps from 0.1 to 10 Hz were conducted according to EN14770
in sandwiched samples of 1.75, 2.00 and 1.00 mm thickness by increasing diameter.
31
Processing of the DSR raw data was performed with the rheological software RHEA
(version 2.0.2) in order to construct master curves based on the Christensen-Andersen
model [148]. Various advanced rheological parameters related i.e. to non-load related
cracking of bitumen can be extracted by this master curve. Recent studies have shown
the evolution of these meaningful rheological parameters with ageing as well as the
relationships between them [93,149–151]. Hence, a selection among these rheological
indicators is considered of paramount importance to evaluate the performance of
bitumen. An overview of the rheological parameters used in this work alongside their
definition and the effect of ageing on them is provided in Table 2-4.
Table 2-4 Advanced rheological parameters
Rheological value
Formula
Definition
Plate
geometry
Evolution
with ageing
Indication
Crossover modulus
(Gc)
Respective values
derived by a master
curve where a phase
angle is δ=45°
The modulus,
frequency and
temperature at the
point where the loss
modulus G΄΄
coincides with the
storage modulus G΄
8 and 25 mm
↓
Elastic/viscous
transition,
brittleness,
hardness [95,152]
Crossover
frequency (ωc)
Crossover
temperature (Τc)
↑
Rutting and
cracking [153]
Shape parameter
(R)
R = logG
g
− logG
c
where Gg is the glassy
modulus and Gc is the
crossover modulus
Difference between
the moduli of glassy
behaviour and
crossover point
4, 8 and 25
mm
↑
Width of
relaxation
spectrum, fatigue
cracking [95,148]
Glover-Rowe (G-R)
G−R =
G∗∙(cosδ)2
sinδ
where G∗ and δ the
complex shear modulus
and phase angle at
15 °C, 0.005 rad/s
A parameter based
on the complex
modulus and phase
angle under a
specific frequency in
correspondence
with ductility at the
same temperature
8 mm
↑
Durability, non-
load related
cracking [94,154]
2. Research methodology
32
and specific
elongation rate
ΔΤ
c
∆T
c
= T
c,G
− T
c,m
where Tc,G and Tc,m the
temperatures where the
DSR values of G(60) =
143 MPa and
m(60) = 0.275
respectively [155]
Difference of two
lower continuous
grading
temperatures
related to creep
stiffness and
relaxation
4 and 8 mm
↓*
Ageing-induced
cracking [154]
*As real number
2.7.1.3 Cup and bob test
An MCR 102 from Anton Paar was selected to be used from the available DSR devices
due to the relatively high torque (200 mNm) of this device compared to the previously
mentioned DSRs. The fixture of cup and bob was used for the viscosity measurements
in a range of shear rates 1-100 s-1 according to EN13302. All the bituminous binders of
this research project behave as Newtonian fluids in all the ageing states and in the
whole range of tested temperatures 110-180 °C. The determination of viscosity is an
important factor in the asphalt process as it has a direct effect on mixing with the rest
asphalt components and eventually on the mixture’s workability and a satisfactory
asphalt compaction. Especially the threshold viscosity value of 3 Pa·s at 135 °C has
been used to ensure the pumpability of the examined bitumen as well as a guideline
to determine mixing and compaction temperatures in the field [156]. The dynamic
viscosity value at 135 °C (η135oC), which has been reported to increase with ageing, was
used further in this work.
Experimental challenges
Although the techniques used in this research project are robust enough, limitations
may still exist, especially when investigating a multi-component mixture like bitumen.
This needs to be acknowledged when reviewing and comparing the findings of each of
the techniques as the results may be contradictory.
First of all, FTIR and TOF-SIMS are in fact both surface analysis methods with different
penetration depths, of a maximum of 2-3 micrometres for FTIR-ATR [157], and only a
few nanometres for TOF-SIMS [113]. In TOF-SIMS an air-cooled surface is investigated,
33
while in FTIR, the interface as formed against the ATR diamond crystal is investigated.
In literature, it has been demonstrated that bitumen may exhibit a different
composition depending on the substrate and the environment [4,68,76,158].
Especially, the reproducibility of TOF-SIMS results is an issue that needs special care as
the thin film flatness may affect the obtained spectra. In the current study the
preparation procedure, the thermal history of the binders and the control of film
thickness were kept constant.
Secondly, sample preparation and sample thermal history are different for both the
FTIR and TOF-SIMS tests: in FTIR a small hot drop or thin film is placed directly onto
the crystal and tested. In TOF-SIMS, hot bitumen drops are first placed on a substrate,
which is then shortly reheated and air-cooled, to achieve a flat surface. The analysis
temperature is much lower in TOF-SIMS compared to the other spectroscopic
techniques, which could for example influence the extent of crystallisation of potential
waxy components. For the 1H-NMR tests, the bitumen is dissolved in deuterated
tetrachloroethane, and although several steps were taken to obtain dissolved samples
(i.e. a solvent with appropriate solubility, ultrasonic bath and visual inspection), there
is still a risk of some undissolved species. Preliminary trials to dissolve asphaltenes
failed although the temperature and dissolution time were considerably increased to
obtain a homogeneous solution. Asphaltene particles were still visible and as such no
1H-NMR measurements were carried out on pure asphaltenes due to a risk of high
variability between replicates or misinterpretation of the obtained spectra.
Moreover, the detection sensitivity for various chemical compounds is dependent on
the respective technique, i.e. FTIR is very sensitive to capture basic oxygenated
products (carbonyls and sulfoxides) while it is rather insensitive to changes such as a
further aromatisation process [159]. Additionally, in TOF-SIMS only the charged
species formed (and not the neutral ones) after a chemical bombardment on the
surface are detected. The 1H-NMR by definition will only capture hydrogenated
compounds, as such the formation of PAH due to the aromatisation process is not
expected to be captured since they do not contain a hydrogen in the middle. In
addition, CW EPR shows both bulk and surface paramagnetic centres and its spectral
intensity will depend on the time after generation of the radicals and, therefore,
centres that are too short living will not be detected in the EPR experiment.
Another point of consideration is the standard ageing protocols performed when
investigating bitumen oxidation, which are not conducted in a closed system.
Therefore, there is a possibility that volatiles, present or formed during ageing may
leave the sample. This effect would be the same for all the analyses performed on the
sample after the ageing tests. Bituminous samples could also not be studied in DVS as
2. Research methodology
34
the change of recorded mass was within the drift of the microbalance and the samples
were not porous enough to obtain more drastic changes.
When it comes to the thermoanalytical techniques, in DSC the acquired signals are
typically very broad, they can be combined with exothermic transitions upon reheating
and the thermal history of the samples before testing may significantly affect the DSC
signals. It is also important to highlight the differences that exist between the
thermoanalytical and microscopic techniques which can be related to the
specifications and working principles of each technique; the DSC and WAXD tests are
related to bulk properties, while the bee structures via AFM or CLSM are a surface
phenomenon.
With regard to the surface investigation of bitumen, there is a wide range of
microscopic techniques available to evaluate the microstructure which vary in terms
of working medium and resolution. A comparison of AFM and CLSM in Table 2-5
reveals that the tapping mode in which the AFM operates allows to record more
detailed height patterns and thus depict phase changes as a result of the shift of tip
oscillation. This residual effect of tip oscillation may affect slightly the reported
topography, although it is considered a near-contact method. In contrast to AFM, the
multi-focus imaging capability of the CLSM makes the recording of non-contact images
of non-smooth surfaces with large height differences possible. The difference in the
resolution of the two microscopies can explain also to some extent differences in the
exact length of certain microstructures of the same sample.
Table 2-5 Comparison between AFM and CLSM
Microscope
Radiation
source
Working
medium
Specimen
mounting Best resolution
Cost of
equipment
AFM
Microcantilever
probe air
Aluminium stubs
or glass slides 0.5 nm ++
CLSM
Laser light
air
Glass slide
1 nm
++
Currently, AFM is the most common instrument used to investigate the microstructure
in bituminous binders. AFM requires minimum sample preparation and can operate in
ambient conditions. Other than the microstructure, AFM is also able to provide
information on the mechanical properties of the sample on the nanoscale. However,
this method is highly sensitive to vibrations during the measurements and can only
conduct measurements on a limited scanning area. Other limitations of AFM bitumen
application are the fact that the surface of the prepared sample should be relatively
smooth and of sufficient thickness to exclude surface-driven effects [160,161].
35
Moreover, tacky and liquid samples cannot be measured with an AFM in the tapping
mode.
It has also been proven that the handling temperature has a significant impact on the
final obtained images, and as such for a fair comparison between microscopic findings
by different research groups, the thermal history, among other factors, during the
sample preparation must remain the same. It is known that the microstructural
properties of bitumen are strongly dependent on the isothermal annealing, and
cooling rate. This becomes alarming since bitumen can age or undergo steric hardening
during or after the sample preparation stage or bitumen samples might collect dust
particles before the images are acquired.
For the rheological characterisation of bitumen not only the volatilisation during the
ageing simulations but also the type of bitumen, temperature equilibrium time and
plate geometry compliance may challenge the validity of the TTSP and thus certain
extracted rheological parameters should be handled with extra caution.
Finally, to additionally provide the potential reader with an idea of the repeatability of
the conducted measurements, since average values are used in the following, the
coefficient of variation was ranging between 0-30% for the spectroscopic,
thermoanalytical and microscopic techniques whereas the obtained rheological
parameters were within 0-10 %.
2. Research methodology
36
*This chapter is redrafted from: G. Pipintakos et al., Experimental investigation of the oxidative ageing
mechanisms in bitumen, Construction and Building Materials (2020).
https://doi.org/10.1016/j.conbuildmat.2020.119702
3
3 AGEING MECHANISMS IN
BITUMEN VIA KINETICS
Summary*
Chapter 3 aims to offer insights into the validity of commonly held hypotheses
regarding the oxidation phases of ageing in bitumen, the fast- and the slow-rate phase,
and explore the main oxidation products formed upon ageing. In order to evaluate
possible differences between bitumen types, the penetration grade as well as the
bitumen production process was varied. Thus, the ageing of three different binders
was first studied by FTIR and EPR spectroscopy in oxidation kinetics. The formation of
oxygen-containing molecular structures on the bitumen surface during ageing was
studied with TOF-SIMS. The results of FTIR reveal a gradual increase of sulfoxides upon
ageing, while the EPR results show an increase of organic carbon-centred radicals. In
parallel, TOF-SIMS results provide evidence for an increase of oxygenated compounds,
such as SOx-, HOx- and NOx-containing compounds. It appears also that paramagnetic
metal species, such as vanadyl-porphyrins, are insusceptible during ageing. The
cumulative FTIR index of carbonyls and sulfoxides showed also a different gradual
increase with varying conditioning temperatures and allowed to estimate the reaction
rates and activation energies for two of the studied binders. DVS tests were also
efficiently utilised for bitumen-filler (mastics) systems and were coupled with FE
simulations to allow for a rather accurate prediction of the diffusivity parameters.
Overall, the findings of this Chapter are in agreement with a mechanism comprising
two rate-determining phases and support the formation of different oxygenated
products. It is believed that the experimental approach used in this Chapter may
contribute further to an improved understanding of the ageing mechanisms in
bitumen.
3. Ageing mechanisms in bitumen via kinetics
38
Objectives
Chapter 3 addresses a number of questions regarding the ageing mechanisms of
bitumen via oxidation kinetics. An important issue is the validation of the previously
proposed oxidation schemes [6]. This was achieved by utilising a number of
spectroscopic and gravimetric techniques. In particular, support was provided by FTIR,
EPR and TOF-SIMS spectroscopy as well as DVS. Links between the results of the
spectroscopic techniques under predefined oxidation time and temperature review
certain hypotheses for the ageing mechanisms of bitumen. Additionally, this Chapter
attempts to address the effect of increased temperature during short-term ageing on
subsequent isothermal reaction kinetics in bitumen as well as to discuss the
differences in reactivity and diffusivity of bitumen by means of their activation energy
and diffusion coefficients/solubility respectively. A graphical summary including the
main objectives for each technique is presented in Figure 3-1.
Figure 3-1 Flowchart of the experimental part of Chapter 3 and objectives
Materials and Methods
Three bituminous binders were used as specified in Table 3-1, namely, a hard binder A
and two soft binders B and C. Binders A and B originate from an acidic, wax-free crude
oil, and differ only in the degree of distillation processing. Binder C is a visbroken
39
residue, containing natural wax (crystallisable compounds) and coming from a
different crude oil. The bituminous binders were selected so as to vary as much as
possible with regard to their empirical properties in order to observe possible
differences arising from the bitumen type.
Table 3-1 Basic properties of the bituminous binders
*N/D=Not Determined
For the mastic sample preparation for DVS, binders A and B were blended at 130 °C
with a filler-to-bitumen ratio (mg/mg) of 1, so that the contribution of both can be
equally balanced. A commercially available inert filler (Quartz) consisting of 100 %
quartz after mineralogical analysis [162], was used to eliminate possible chemical
reactions between oxygen and filler and simultaneously to accelerate the diffusion
part due to the increased accessibility of oxygen into the more porous mastics
compared to virgin bitumen.
To simulate the oxidative ageing of the binders, a Modified Thin Film Oven test (M-
TFOT) was used: a binder film of approximately 1 mm thick was aged in a ventilated
oven. For the investigation of oxidation kinetics, three different temperatures were
employed i.e. 25, 50 and 75 °C for different time intervals based on each temperature.
Table 3-2 provides the exact time intervals measured per temperature considering as
a rule of thumb that the overall reaction rate doubles per 10 °C increase. For
exploratory purposes, the effect of bitumen film thickness was also investigated at the
highest temperature of 75 °C using a thin film apparatus (Figure 3-2) able to produce
films as thin as 100 μm (0.1 mm). This was done by heating both the apparatus and the
binder and then the apparatus passed on top of the binder pool to create the required
thickness in a pastry-like fashion. Additionally, the effect of short-term ageing by
means of laboratory RTFOT on the reaction kinetics was investigated for both film
thicknesses at 75 °C, indicated as RTFOT+75 in Table 3-2.
After performing the ageing conditioning, the top surface of the bituminous samples
A and B was measured in all the time intervals, temperatures and film thicknesses in
FTIR, while for bitumen C was studied only for the time intervals at 50 °C. In addition,
all the binders were tested in TOF-SIMS at the virgin state and after 8 days M-TFOT at
50 °C. In all cases, the film thickness was kept minimum in order to exclude, as much
Material
Property
Binder
Test Method
Bitumen
A
B
C
Penetration 25 °C (0.1 mm)
16
189
190
EN1426
Softening point (°C)
61.1
37.5
39.2
EN1427
Penetration index, Ip
-1.06
-1.46
-0.63
EN12591
Viscosity 135 °C (mm2/s)
1285
203
N/D*
EN12595
Performance Grade (PG)
76-16
52-28
52-22
AASHTO MP1
3. Ageing mechanisms in bitumen via kinetics
40
as possible, the diffusion effect from the coupled reaction-diffusion phenomenon of
ageing. By minimising the film it is assumed that the diffusion effect will be eliminated
and primarily oxidation will be the dominant process [163–165]. However,
instrumental constraints for the EPR analyses resulted in thicker films (3-5 mm) which
were aged directly in polypropylene tubes and may have experienced different
diffusion. Since in EPR analyses the entire bitumen sample was measured, the number
of spins derived from the EPR spectra was divided by the exact mass of the sample in
order to extract a fair comparison value independently of the small fluctuation of
thickness compared to FTIR and TOF-SIMS films.
Table 3-2 Studied time intervals per oxidation temperature
Figure 3-2 Thin Film applicator for bitumen
For the ease of the reader, a comparative overview by means of ageing ranking of the
three binders per technique upon completion of the ageing kinetics at 50 °C, is given
in Table 3-3.
Table 3-3 Ageing ranking per technique for the three binders after kinetics completion
Experimental technique
Ageing severity (increasing order)
FTIR
A
B
C
EPR
C
A
B
TOF-SIMS
C
A
B
DVS
A
B
-
Oxidation phases
FTIR
To track the oxidation phases assumed previously in the literature, focus was initially
given exclusively on sulfoxides, which are more prone to be produced during oxidation
Temperature (°C)
Time intervals (hours)
Film thickness (mm)
25
0/672/1344/2016/2688/3360
1.0
50
0/168/336/504/672/840/1008/1176/1344
1.0
75
0/8/24/96/168/336/504/672
1.0 & 0.1
RTFOT+75
0/8/24/96/168/336/504/672
1.0 & 0.1
41
at lower temperatures, like the one studied (50 °C). A typical evolution of this index in
kinetics at 50 °C is depicted in Figure 3-3 for bitumen A.
Figure 3-3 Spectra of the evolution of sulfoxides at 50 °C M-TFOT for bitumen A
The results of FTIR analyses show a steep increase of the sulfoxide index (normalised
intensity) followed by a steady milder increase for all three binders (Figure 3-4).
Binders A and C were found to have almost completed the initial rapid increase at
about 5 days, whereas binder B reached this transition point at about 2 days. It
becomes apparent that at 8 days of controlled ageing kinetics the slow-rate oxidation
reaction has been initiated for all the binders.
Figure 3-4 Evolution of the FTIR sulfoxide index over ageing time at 50 °C M-TFOT and the fitting used,
based on a dual-sequential model [6,19]
Given that sulfoxides are considered to be one of the end products of both a fast and
a slow reaction [18], oxidation kinetics were approximated in the following way for the
product evolution. Assuming pseudo first-order kinetics and that the rate of the slow
reaction is rather small, the evolution of an end product P, of a dual-sequential
oxidation in a specific temperature can be described by Equation 3-1 [19].
3. Ageing mechanisms in bitumen via kinetics
42
'
( ) (1 ) '
f
kt
f ss
Pt P e P k t C
−
∞∞
=−+ +
Equation 3-1
where () is the amount of product as a function of time;
∞ and
∞ are amounts
of product from the fast and slow reactions, respectively at the reaction endpoint;
′
and
′ are pseudo first-order rate constants of the fast and slow reactions,
respectively; and is a constant [19]. The FTIR results for all three binders can be
reasonably fitted to Equation 3-1 with their parameters given in Table 3-4.
Table 3-4 Parameters used for fitting Equation 3-1 to the FTIR sulfoxide index at 50 °C M-TFOT and their
R2
Binder
∞ [×10-3 a.u.]
′ [×10-6 s-1]
∞
′ [×10-6 s-1]
[×10-3 a.u.]
R2
A
8.5 ± 0.2
7.3 ± 2.2
1.33 ± 0.03
4.69 ± 0.08
0.99
B
6.3 ± 0.1
13.8 ± 1.5
0.67 ± 0.04
4.14 ± 0.01
0.99
C
7.9 ± 0.7
5.9 ± 1.2
1.00 ± 0.07
8.12 ± 0.65
0.98
From this analysis,
′ was determined to be fastest for binder B, followed by A then C.
Moreover, the percentile increase of the normalised sulfoxide intensity of the virgin
binder up to the completion of the ageing treatment used here, was evaluated. This
percentile increase demonstrates that binder A suffered from a harsher oxidation
effect (317.9 %) followed by binder B (224.5 %) and C (148.4 %), under the same ageing
conditions.
Semi-quantitative methods to analyse the FTIR spectrum can be useful for identifying
and characterising the evolution of specific oxidation products. Consistent with
previous studies, this work demonstrated an initial rapid increase of sulfoxides
followed by a slow-rate formation [6,18]. Hence, the kinetics of the sulfoxide index can
establish a simple way to distinguish different oxidation rates. At least, they can give a
rough estimation for the completion time of the fast-rate phase and the initiation of
the slow one under the given oxidation kinetics.
EPR
The time-dependent evolution of the EPR spectra of the three binders was investigated
in the same timeframe as with the FTIR, up to 56 days at 50 °C. The graphs in Figure
3-5 [right] and Figure 3-5 [left] respectively, show the evolution of the organic carbon-
centred radicals and VO2+ species over ageing time. Interestingly, in all cases, the
amount of VO2+ spins remained relatively constant as the samples were aged (a zeroth-
order line can be fitted with the error bars), while an increase was observed for the
organic carbon-centred radicals (Figure 3-5 [left]). A comparison of the binders used in
this study indicates also that the relative increase of organic carbon-centred radicals
43
in binder C was about 1.5-2.0 times higher in comparison with binders A and B, while
binder A keeps slightly increasing after 56 days.
Figure 3-5 Evolution of the carbon-centred radical at 50 °C M-TFOT and the used fitting, based on a dual-
sequential or a fast reaction model [left] and VO2+ centres over ageing time at 50 °C M-TFOT [right]
Although the results for the evolution of the carbon-centred radical EPR signal could
also be fitted using Equation 3-1 (parameters given in Table 3-5), for each binder the
contribution from the slow reaction term was almost negligible and for binder C it was
even negative, which is contrary to a dual fast and slow production of end products
model. These carbon-centred radicals are believed to be only produced and also
subsequently consumed during the slow-rate phase. Since they are observable, their
rate of production must be faster than their rate of consumption. In all cases, the
evolution of the carbon-centred radical EPR signal is dominated by a mono-exponential
component for the total radical evolution which strongly suggests that their rate of
production is kinetically controlled by the fast-rate phase. This means that the
previously hypothesised oxygen-centred free radicals, assumed by Petersen and his
colleagues to be generated in the fast-rate phase [6], are able to immediately abstract
protons attached to benzyl rings to yield carbon-centred radicals. In contrast, since the
subsequent consumption of these carbon-centred radicals appears to be much slower,
it must be governed by the overall kinetics of the slow-rate phase. Once the system
enters the slow-rate phase, the rate of oxygen-centred free radical production and
thus carbon-centred radical production will become comparable to their rate of
consumption, and therefore, as can be seen in Figure 3-5 [left], no net increase in the
amount of carbon-centred radicals is observed. Therefore, a two-phase oxidation
scheme can still be supported by the obtained EPR results.
In line with these hypotheses, the
′ values obtained for the radical production were
within a similar range to those for sulfoxide production. Similarly, the
′ for binder B
was again the fastest, which suggests that both organic carbon-centred radicals and
3. Ageing mechanisms in bitumen via kinetics
44
sulfoxide production in the fast reaction may be controlled by the same rate-limiting
process.
Table 3-5 Parameters used for fitting the EPR spin counting of the radical signal at 50 °C M-TFOT and
their R2
Binder
Reaction(s)
∞
[×1017 spins g-1]
′
[×10-6 s-1]
∞
′
[×1010 spins
g
-1
s
-1
]
[×1017 spins
g
-1
]
R2
A
Fast & slow
4.1 ± 0.5
1.9 ± 0.4
5.2 ± 1.4
16.0 ± 0.2
0.98
Fast only
6.1 ± 0.3
1.0 ± 0.1
N/A
16.0 ± 0.2
0.97
B
Fast & slow
3.4 ± 0.4
3.7 ± 0.9
2.2 ± 1.4
11.0 ± 0.2
0.97
Fast only
4.0 ± 0.2
2.6 ± 0.4
N/A
11.0 ± 0.2
0.96
C
Fast & slow
8.7 ± 0.9
1.7 ± 0.3
-0.73 ± 0.02
21.0 ± 0.4
0.97
Fast only
8.5 ± 0.4
1.8 ± 0.2
N/A
21.0 ± 0.4
0.97
It is interesting to mention that, in contrast to FTIR results, the harsher, relative to the
unaged state, oxidation effect was observed for binder C. This difference can be
explained by the fact that binder C was the result of a visbreaking process, which is a
mild cracking process, resulting in radical formation. This can explain the higher initial
organic carbon-centred radical concentration observed in bitumen C (Figure 3-5 [left]).
In addition, the VO2+ signal appears to stay constant over the ageing time in Figure 3-5
[right], for the three investigated binders. It can be exploited as an indicator of the
vanadium content in petroleum [166]. Of particular interest is that this assignment
seems to remain unaffected by ageing and could be potentially used as a marker for
the origin of the bitumen.
Together the two spectroscopic techniques (FTIR and EPR) give evidence for the two
rates of a dual-oxidation route. Given this, FTIR supports that between 2 and 5 days
for all the examined binders the fast-rate reaction has finished, whereas EPR proves
that carbon-centred radicals, which are believed to be produced and subsequently
consumed during the slow-rate phase, actually evolve with the fast-rate phase kinetics.
The fact that free organic carbon-centred radicals still exist after 56 days of M-TFOT
ageing, suggests towards the continuation of the slow-rate oxidation phase until
complete reaction/termination. The ageing time interval of 8 days was used
afterwards for TOF-SIMS surface analyses to unravel the main products after the
combined effects of the fast- and slow-rate phase.
Oxidation products
TOF-SIMS
45
TOF-SIMS was used to analyse the molecular changes and oxidation products on the
bitumen surfaces upon ageing. Positive and negative ion spectra were acquired for the
three binders (A, B and C) in the unaged state and after 8 days of ageing, at which point
the fast-rate phase is mainly completed and the slow-rate phase has been initiated.
Representative spectra of the unaged and aged samples of bitumen A and C are
presented in Figure 3-6 and ion assignments of peaks relevant to ageing are listed in
Table 3-6 and Table 3-7. No clear changes due to ageing can be observed in the major
peaks of the spectra, indicating that the effect of ageing on the molecular surface
structure is relatively small. In contrast, clear differences can be observed between the
spectra of bitumen A and C, mainly reflecting the wax content of bitumen C. As has
been reported previously in the literature wax segregates effectively in bitumen to
form a thin layer of wax, largely covering the surface, which can be observed in the
spectra of bitumen C [167]. The higher intensities of peaks in bitumen C represent
aliphatic species, whereas the spectrum of the wax-free bitumen A displays higher
intensities of peaks representing aromatic species, as well as N- and S-containing
organics.
Figure 3-6 Negative [left] and positive [right] ion spectra of bitumen A and C before and after M-TFOT at
50 °C for 8 days
3. Ageing mechanisms in bitumen via kinetics
46
Table 3-6 Utilised oxygenated peaks for positive
ions of TOF-SIMS
Ion Observed
mass (m/z)
Assignment of
molecular
structure
CH3O+
31.020
Ox-containing
C3H5+
41.047
Aliphatic
C2H3O+
43.020
Ox-containing
C3H7+
43.066
Aliphatic
C4H7+
55.068
Aliphatic
C4H9+
57.087
Aliphatic
C3H7O+
59.049
Ox-containing
C5H7+
67.066
Aliphatic
C5H9+
69.087
Aliphatic
C5H11+
71.117
Aliphatic
C6H9+
81.087
Aliphatic
C6H11+
83.112
Aliphatic
C6H13+
85.135
Aliphatic
C7H11+
95.106
Aliphatic
C7H13+
97.129
Aliphatic
C8H13+
109.128
Aliphatic
C9H7+
115.059
Aromatic
C9H11+
119.105
Aliphatic
C10H8+
128.068
Aromatic
C13H9+
165.079
Aromatic
Table 3-7 Utilised oxygenated peaks for negative
ions of TOF-SIMS
Ion Observed
mass (m/z)
Assignment of
molecular
structure
O-
15.994
Ox-containing
OH-
17.004
HOx-containing
CN-
26.003
N-containing
C3H2-
38.016
Aliphatic
C3H3-
39.024
Aliphatic
C2HO-
41.006
HOx-containing
CNO-
42.002
NOx-containing
CHO2-
45.001
HOx-containing
C4H3-
51.023
Aliphatic
CSO-
59.967
SOx-containing
SO2-
63.963
SOx-containing
C4HO-
65.003
HOx-containing
C5H5-
65.038
Aliphatic
C2H3SO-
74.989
SOx-containing
C6H5-
77.037
Aliphatic
SO3-
79.961
SOx-containing
SO4H-
95.959
SOx-containing
C3NO-
96.999
NOx-containing
Although the major peaks in the TOF-SIMS spectra are essentially unchanged, the
effect of ageing can be observed by consideration of oxygen-containing fragment ions,
including SOx-, NOx- and Ox-containing organic ions, which are present at lower signal
intensities and mainly in the negative spectra (see Table 3-6 and Table 3-7).
The oxygen-containing fragment ions of the three binders were analysed based on
normalised signal intensities and categorised into SOx-containing (Figure 3-7 [left]) and
HOx-containing fragments (Figure 3-7 [right]). A strong increase in the intensity of the
fragments with the generic formula RSOx and RHOx can be observed, indicating the
formation of sulfoxide- and oxygen-related compounds as a result of ageing.
Additionally, for all binders the amount of cyanate fragments, with generic formula
RNO-, also increased during oxidation. It is important to note here that signal
intensities of the same ion can be compared to indicate concentration differences
between samples of the specific species that it represents.
47
Figure 3-7 Intensity of SOx-containing [left] and HOx-containing compounds via TOF-SIMS for all three
binders [right] before and after 8 days M-TFOT at 50 °C
Comparing the different binders, the relative effect of ageing is generally higher for
bitumen A compared to bitumen B, but the strongest effect is observed for the wax-
containing bitumen C. Furthermore, bitumen A and B show no effect of ageing on the
intensities of fragment ions without oxygen content, including CN- and
aliphatic/aromatic hydrocarbon fragments. For bitumen C, however, ageing results in
increased intensities of CN- (Figure 3-8 [left]), as well as in hydrocarbon fragments
representing aromatic species, and decreased intensities of aliphatic hydrocarbon
fragment ions, see Figure 3-8 [right].
Next to sulfoxide formation during the slow reaction, alcohol groups may also be
produced [6,168]. The sulfoxide intensity observed with FTIR has been assigned, up to
now, completely to this functional group. However, an overlap with other HOx-
containing groups may exist in the corresponding infrared absorption band (around
1100 cm-1), indicating that alcohols and sulfoxides may coincide. Whereas the origin
of the increase of this infrared absorption band upon ageing is not clear, the observed
increase of HOx-containing fragments with TOF-SIMS is consistent with the formation
of alcohols and/or carboxylic acids. However, a clear differentiation is not possible
without a more elaborated statistical analysis and/or complementary chemical
information. It can thus be speculated, in general, that ageing results in the formation
of more polar species, e.g. alcohols or carboxylic acids, which then may affect the
molecular interactions and eventually the rheology of bitumen.
3. Ageing mechanisms in bitumen via kinetics
48
Figure 3-8 Intensity of nitrogen-containing compounds in all binders [left] and intensity of aliphatics in
binder C [right] before and after 8 days M-TFOT at 50 °C
Previously [169,170], it was found that nitrogen-containing compounds can be
identified in oxidised binders without significantly changing upon ageing [18]. In this
study, the increase of RNO- fragments indicates that additional oxidation products,
those of nitrogen-containing compounds are present upon ageing, which should be
possibly considered in a future oxidation scheme. It can also be expected that the
concentration of the different heteroatoms may have an impact on the intensity of the
different fragments.
Figure 3-9 High mass range of negative [left] and positive ion spectra of bitumen C before and after 8 days
M-TFOT at 50 °C [right]
In Figure 3-9, TOF-SIMS spectra of bitumen C are presented at a higher mass range,
m/z = 350-1200, where the peaks to a larger extent correspond to intact or nearly
intact molecular species. The envelope of peaks at m/z 500-1000 in both negative and
positive ion spectra can be assigned to intact wax molecules, as previously articulated
by Lu [167], for which the intensity reduction upon ageing is consistent with the
reduced intensities of the aliphatic fragment ions upon ageing (Figure 3-8 [right]).
Interestingly, ageing of bitumen C produces a new envelope of peaks at m/z 600-800
in the negative ion spectrum. Although unambiguous identification of these peaks is
not possible due to the limited mass resolution, mass separation of m/z = 14 between
equivalent peaks and the exact mass of the peaks is consistent with molecules
comprised of nearly saturated aliphatic hydrocarbon chains with an added SOx
49
functional group. For example, a peak observed at m/z 689.63 is consistent with
C45H85SO2, containing both an aliphatic and a small aromatic structure in which the S=O
may be included. The latter may be taken as a consequence of oxidation and its
subsequent sulfoxide formation.
For the positive ions, the effect of ageing was mainly observable in binder C. A clear
increase in the intensities of O-containing organic ions was observed, reflecting the
formation of oxidation products on the bitumen surface (Figure 3-10). Furthermore,
positive ions of aliphatic fragments show a decreasing trend as can be seen in Figure
3-10, in contrast to PAH which increase upon ageing (Figure 3-11), consistent with the
observations for the negative ions (Figure 3-8 [right]).
Figure 3-10 Intensity of O-containing and aliphatic fragments in selected positive ions before and after 8
days M-TFOT at 50 °C for bitumen C
Figure 3-11 Intensity of PAH fragments in selected positive ions of bitumen C before and after 8 days M-
TFOT at 50 °C
Finally, in order to investigate the spatial distribution of the wax fraction and the
ageing-related molecular species on the surface of bitumen C, high-resolution TOF-
SIMS images were generated (Figure 3-12). A clear phase separation with a governing
3. Ageing mechanisms in bitumen via kinetics
50
aliphatic phase is observed, with particles about 5-10 μm in size which covered most
of the surface. Apparent wax-related particles (represented by aliphatics) cover most
part of the surface and the spaces between these particles display increased signal
intensities of ageing-related ions, such as O-, CN- and (H)SOx- (Figure 3-12), as well as
aromatics.
Figure 3-12 High-resolution TOF-SIMS images of selected negative fragment ions on the surface of bitumen
C before [top] and after 8 days M-TFOT at 50 °C [bottom]. The signal intensities in the images are given as
the maximum number of ion counts per pixel (MC) and total ion counts in the entire image (TC)
Reactivity and diffusivity
FTIR
Exploration of the reaction kinetics of three different isotherms and two film
thicknesses in FTIR was conducted for binders A and B by employing the sum of the
ageing indices of carbonyls and sulfoxides. The results of the oxidation kinetics in
Figure 3-13 were fitted assuming a first-order reaction rate expression according to
Equation 3-2, where k is the reaction rate coefficient, t is the ageing time interval and
P is the sum of the oxidation products of carbonyl and sulfoxide, assumed herein
independent of oxygen concentration.
(1 )
P
t
kP
∂
∂
=
−
Equation 3-2
51
Figure 3-13 Experimental data points of FTIR ageing indices at the three isotherms for M-TFOT at 50 °C of
1 mm thickness
Assuming the effect of oxidation temperature T on the oxidation reactions, an
Arrhenius type equation can explain the reaction rate based on Equation 3-3, where
Ea is the activation energy, R is the universal gas constant and A is the reaction factor.
a
E
RT
k Ae= ⋅
Equation 3-3
Fitting the oxidation kinetics data of Figure 3-13, based on Equation 3-2 and extracting
the reaction rates (Table 3-8) allows the estimation of the activation energy by linear
fitting of the natural logarithm of the rate constant versus 1/RT at three temperatures
(Figure 3-14).
Table 3-8 Reaction rates of a first-order reaction
Temperature
(°C)
Film thickness
(mm)
k (bitumen A)
(1/h)
k (bitumen B)
(1/h)
25
1.0
3.78 E-06
3.34 E-06
50
1.0
2.05 E-05
1.06 E-05
75
1.0
6.11 E-05
5.08 E-05
75
0.1
7.65 E-05
5.89 E-05
RTFOT+75
1.0
2.81 E-05
2.62 E-05
RTFOT+75
0.1
3.55 E-05
3.21 E-05
The results of Table 3-8 reveal that there is an increasing trend of the reaction
constants for both binders with increasing reaction temperature, which confirms the
validity of the Arrhenius temperature dependency of the k constants. Interestingly, the
reaction constants at 75 °C show that kinetics of the samples previously subjected to
RTFOT, experience a slower reaction rate which is almost halved. Moreover, the
findings of the two investigated film thicknesses at 75 °C support that for thinner films
of 100 μm the reaction is the governing process of the coupled reaction-diffusion
phenomenon, which is in line with the slightly higher rate constant compared to the 1
mm films for both the unaged and RTFOT aged binders.
3. Ageing mechanisms in bitumen via kinetics
52
Comparing the two binders, bitumen B shows a lower rate constant at all
temperatures, while Figure 3-14 depicts the goodness of fitted activation energies as
expressed by the high coefficients of determination. Based on this fitting, bitumen B
shows a lower Ea=46.7 KJ/mol compared to the slightly higher one of bitumen A
(Ea=48.2 KJ/mol), which indicated that bitumen B has slightly higher temperature
sensitivity.
Figure 3-14 Linear fit for the extraction of activation energies
DVS
The experimental results of the DVS tests for the corresponding bituminous mastics
are given over time in Figure 3-15. It can be seen that both mastics of bitumen A and
B are not able to reach the saturation level (plateau) within the specified testing time.
However, the acquisition of data points for the change of mass allows for the back-
calculation of parameters governing the reaction procedure and, thus, the prediction
of the terminal time with completed oxidative reactions. To succeed this, traditional
Fickian laws have been revised by scholars with the potential to account for both
effects and to discuss more accurately the effect i.e. on oxygen diffusivity [19]. Past
studies have reported values for the diffusion coefficients of oxygen in bitumen
ranging between 10-10 m2/s and 10-15 m2/s [171].
Simulations and optimisations in a multiphysics Finite Element (FE)-based software
package (COMSOL version 6.0) were performed assuming a reaction-diffusion model
for the DVS data of the following Equation 3-4 [110], where cO2 is the concentration of
O2, D is the oxygen diffusion coefficient, k is the oxygen reaction constant and cB is the
concentration of reactive bitumen components.
22
2
2
B
oo
co
t
D
c kc c
∂
∂
= ∇
−
Equation 3-4
53
Figure 3-15 DVS experimental data and optimised FE simulations for the mastics of binders A and B
Figure 3-15 depicts the convergence between the experimental results and model
simulations that can accurately predict the overall diffusion and reaction evolution.
Based on the optimisation of the reaction-diffusion model by unitising experimental
data from DVS, useful parameters can be extracted, presented in Table 3-9. The
diffusion coefficient can therefore be predicted for bitumen mastics and considering
that the filler remained unchanged during the DVS tests, this coefficient has also an
application for the corresponding binders A and B. In addition, the range of these
coefficients lies well within the values reported previously in the literature [110]. With
regard to the reaction rates, they are rather underestimated as compared to a pure
reaction model of Equation 3-2, while they are in agreement with the trend found
previously in Table 3-4. This can be explained by the fact that a coupled reaction-
diffusion model accounts for more details i.e. the oxygen supply and gradient in the
film, as oxidation kinetics evolve at 75 °C.
Table 3-9 Simulated FE reaction-diffusion parameters based on modified Fickian law
Property
Mastic of bitumen A
Mastic of bitumen B
O2 solubility [mol/m3]
0.940
1.079
D [m2/s]
8.26 E-11
7.90 E-11
k [m3/(s·mol)]
7.83 E-06
1.03 E-05
cB [mol/m3]
59.381
37.497
Finally, the values of oxygen concentration at the ideal saturation level can be obtained
via the FE simulated model and by definition are equal to the oxygen solubility in
mastics and respectively in bitumen. The predicted values are also within the range of
previously reported values in literature [19,110]. All in all, comparing the two binders
A and B they present similar values for the diffusion coefficient while bitumen B
exhibits higher oxygen solubility. It is finally worth noting that the reactive component
concentration of bitumen A is higher than that of bitumen B. SARA analysis of bitumen
3. Ageing mechanisms in bitumen via kinetics
54
A (see Chapter 8), corroborates this observation as it showed higher asphaltenes and
resins percentage.
Highlights of the chapter
FTIR supports the hypotheses of two rate-determining oxidation phases, a
fast- and a slow-rate and a rapid sulfoxide formation during the fast-rate
phase.
The completion point of the fast-rate phase has been reached for all binders
around 8 days under oxidation kinetics at 50 °C via FTIR and EPR.
EPR measurements demonstrate that the overall amount of organic carbon-
centred radicals evolves predominantly with the fast-rate phase.
EPR also shows that in this study vanadyl-porphyrin species remain unchanged
during lab ageing kinetics.
FTIR and TOF-SIMS spectrometry indicate that sulfoxide-, nitrogen- and
oxygen-containing compounds, e.g. alcohols and/or carboxylic acids, are
formed after the occurring fast- and slow-rate phases.
An increase of aromatics and the accompanying decrease of aliphatics is only
observed for the visbroken, waxy binder.
RTFOT short-term ageing influences the reaction rate constants of a universal
first-order reaction kinetics model.
DVS in combination with FE model simulations can rather adequately estimate
the diffusion coefficient of mastics and bitumen in the case that an inert filler
is used.
The oxygen solubility, diffusion coefficient and reactive components
concentration of the binders of this work agree well with reported values in
literature.
Overall, the insights gained from this Chapter may be particularly interesting for laying
the groundwork for the underlying oxidation mechanisms in bitumen with
standardised lab ageing simulations, examined in Chapter 4.
*This chapter is redrafted from: G. Pipintakos et al., Exploring the oxidative mechanisms of bitumen after
laboratory short- and long-term ageing, Construction and Building Materials (2021).
https://doi.org/10.1016/j.conbuildmat.2021.123182
X. Lu et al., Analysis of asphaltenes and maltenes before and after long-term aging of bitumen, Fuel (2021).
https://doi.org/10.1016/j.fuel.2021.121426
4
4 AGEING MECHANISMS IN
BITUMEN VIA STANDARDISED
LAB SIMULATIONS
Summary*
Chapter 4 focuses on the identification of the intermediate and final oxygenated
products after short- and long-term laboratory ageing simulated with RTFOT and PAV
respectively. Three binders were investigated in this study, two originated from the
same wax-free crude source, while the third was obtained from a different source,
containing natural wax, and followed a different manufacturing process. FTIR
spectroscopy demonstrated a clear increase of the sulfoxide and carbonyl functional
groups upon ageing for all the binders independently of origin, manufacturing or
performance. EPR spectroscopy showed an increase of the organic carbon-centred
radicals after short-term ageing (RTFOT), whereas after PAV ageing these radicals
remained constant in the two wax-free binders originating from the same crude
source, and even decreased for the third, waxy binder. 1H-NMR spectroscopy reported
differences in the relative distribution of protons between the binders in the unaged
state, and similar minor changes after both ageing steps regardless of the binder’s
crude source and distillation. The results of TOF-SIMS revealed that SOx- and (OH)x-
containing compounds are produced after the sequentially occurring short- and long-
term ageing in both wax-free binders, whereas an almost constant behaviour of
aliphatics after PAV ageing can be seen for the same binders. In addition, the SARA
4. Ageing mechanisms in bitumen via standardised lab simulations
56
fractions of the waxy and the softer straight-run binder were also investigated by FTIR,
1H-NMR and DSC before and after RTFOT+PAV and the findings support that a large
part of carbonyls and sulfoxides upon ageing shifts to the asphaltenes, while waxes
can mainly be found in the maltenes fraction.
Objectives
Artificial lab ageing simulations are widely used in the asphalt sector, as it is believed
to capture more realistically the view of bitumen in-situ than kinetics and accelerate
considerably the lab simulation. It is of importance to examine if the hypotheses about
the ageing mechanisms via kinetics account for a fair correspondence with
standardised lab simulations of bitumen. Thus, Chapter 4 aims primarily to identify the
basic ageing compounds (radicals, ions, protons and chemical products) formed after
artificial ageing. Specifically, the proposed approach employs the most appropriate
advanced spectroscopy (FTIR, EPR, 1H-NMR and TOF-SIMS), described in Chapter 2, in
an attempt to capture the structural and chemical changes upon ageing with routine
laboratory tests. It manages also to account for the effect of both the crude source,
the binder’s penetration grade and the distillation process on the mechanisms upon
standardised ageing. The distribution of the oxygenated products in the different
bitumen fractions (asphaltenes and maltenes) is also investigated via spectroscopy and
DSC. The methodology adopted in this Chapter alongside the main objectives is shown
schematically in Figure 4-1.
Figure 4-1 Flowchart of the experimental part of Chapter 4 and objectives
57
Materials and Methods
The same bituminous binders A, B and C (Table 3-1) explored via kinetics were used
also in this Chapter to reveal differences related to the crude source, refinery process
and wax presence.
The oxidative ageing in the lab was simulated for the short term-ageing by RTFOT
according to EN 12607-1 [40] and for the long-term ageing with the PAV according to
EN 14769 [41]. Typically for RTFOT, 35 grams of unaged virgin bitumen are placed in
each bottle inside a rotating carousel for 75 minutes, where temperature and fresh air
flow are set at 163 °C and 4 L/min. This test simulates the high temperatures occurring
during the production and is linked with the initial fast-rate oxidation phase of a dual-
sequential oxidation scheme. The PAV protocol, on the other hand, was followed for
all the RTFOT aged samples, in an attempt to reproduce in a short period of time the
effect of many years in the field. The PAV procedure requires 50 grams of bitumen to
be placed in pans inside a pressure chamber, which results in a thin film of bitumen
able to be sufficiently aged. In the current investigation, the temperature was set at
100 °C with an air pressure of 2.1 MPa for a total duration of 20 hours. In order to
simplify the nomenclature for each ageing state of the three binders, the following
code ‘X-ageing state’ indicates the type of binder X (A, B or C) and the ageing state
(Unaged, RTFOT or PAV).
An overview of the ageing ranking of the three binders per technique as well as their
overall rheological evaluation after PAV, is given in Table 4-1.
Table 4-1 Ageing ranking per technique for the three binders after PAV
Experimental technique
Ageing severity (increasing order)
FTIR
A
B
C
EPR
C
A
B
TOF-SIMS
C
A
B
1H-NMR
C
B
A
DSR
C
B
A
To investigate additionally the effect of lab ageing on the fractions, SARA separation in
the straight-run bitumen B and the waxy bitumen C was additionally performed
according to DIN 51595. Based on this protocol, about 60 g of hot bitumen are poured
into an Erlenmeyer flask, where n-heptane solvent is added in a ratio of 100 mL of
solvent per 1 g of sample. After slightly heated on a hot plate, the Erlenmeyer flask is
shaken until all the bitumen is partially dissolved and the other part precipitates. The
solution with the precipitates (asphaltenes) is cooled to room temperature for 30
minutes. Then a first filtration is performed on a Whatman 41 filter paper. The
asphaltenes on the filter paper are washed repeatedly with the hot solvent until the
4. Ageing mechanisms in bitumen via standardised lab simulations
58
filtrate becomes colourless. After drying the filter paper, the precipitates can be
collected in a beaker. The dried asphaltenes are washed one more time with n-heptane
and are filtered on a vacuum filter. Maltenes fraction can also be collected from all n-
heptane solutions after removing the solvent by a rotary evaporator. Therefore,
asphaltenes in a powder-like state and maltenes in the virgin and after the sequentially
occurring RTFOT+PAV state were further used in the analytical methods.
Oxidation products in bitumen
FTIR
The indices presented in Equation 2-1 to 2-4 were determined for the three bituminous
binders under investigation before and after laboratory short (RTFOT) and long-term
ageing (PAV). The results are consistent with the view that an increase in ageing
severity is observed after standardised ageing for the main oxidative indices, namely
the sulfoxide and the carbonyl index [46,172]. Figure 4-2 depicts the evolution of the
two indices and the effect of each ageing state is discussed herein in percentage
increase of the virgin state.
Figure 4-2 Evolution of carbonyl [left] sulfoxide index [right] with lab ageing
A relative comparison of the formation rates can be performed in terms of percentage
differences. The results of bitumen A for the carbonyl index show an increase of
150.7% and 600.1% after RTFOT and PAV respectively, which indicates a more rapid
and steep increase compared to bitumen B (79.2% and 157.5%) which differed only in
empirical performance (Figure 4-2 [left]). The presence of carbonyl groups for these
two binders in the unaged state can be explained by the acidic nature of the crude oil
from which they originated. On the contrary, for bitumen C-Unaged, a negligible initial
carbonyl index was obtained based on the same analysis procedure. As such
percentage differences are meaningless for bitumen C after RTFOT (with an initial
59
unaged index equal to zero), but an increase in carbonyls after RTFOT ageing is clear.
The significant increase of carbonyls is apparent in bitumen C after PAV which
increased (484.7%) compared in this case with the index in RTFOT. Therefore, it can be
speculated that the carbonyl formation rate was even faster for bitumen C than A,
although a numerical straight-forward comparison of the percentage increase cannot
be performed.
A detailed inspection of the initial sulfoxide index for C-Unaged implies that this is
significantly higher than for bitumen A and B. Concerning the sulfoxide index the three
bituminous binders show a similar percentage increase after RTFOT (A-95.3%, B-91.4%
and C-91.9%). The effect of the bitumen type seems to be very small on the sulfoxide
formation rate after short-term ageing but after PAV ageing the sulfoxide indices
increase more for binders A and B (A-218.9%, B-246.0%) compared to the increase of
the sulfoxide index of bitumen C (123.1%).
Overall, the bitumen type has little effect on the sulfoxide formation rate but seems to
affect the carbonyl formation differently. On the other hand, straight-run bitumen
from the same crude shows a similar increase of sulfoxide upon ageing in the lab, with
the hard bitumen A exhibiting a more rapid carbonyl increase compared to bitumen B.
The representative indices (Equation 2-3 and 2-4) for the aromatisation of bitumen
were also investigated in this Chapter. In the past, it has been assumed that
aromatisation increases primarily the planarity of perhydroaromatic rings in the fast-
rate reaction but it can also cause the aromatisation of alkyl-substituted naphthenic
rings during the slow-rate phase [18,173]. Moreover, an interplay between the
formation of aromatic structures and a reduction of aliphatics (occurring in a lesser
amount) can be hypothesised, which would result in a mutual change of the
aromaticity and the branched aliphatic index. Surprisingly in this investigation, the FTIR
indices related to these compounds were constant for both the aromaticity and
branched aliphatic index in all the ageing states (Figure 4-3). It was postulated that
FTIR was not sensitive enough to capture changes related to aromatisation with the
suggested protocol followed herein for the index calculation. Moreover, alcohol
regions around 3200-3500 cm-1, seem not to be affected by ageing for all the binders,
whereas an overlap in the C-O stretch of possible alcohol formation could exist with
the sulfoxide increase in the FTIR spectra.
4. Ageing mechanisms in bitumen via standardised lab simulations
60
Figure 4-3 Branched aliphatic [left] and aromaticity index [right] with lab ageing
EPR
The evolution of the organic carbon-centred radicals and the VO2+ species after routine
laboratory ageing simulations is given in Figure 4-4. Overall, the carbon-centred
organic radicals showed an increase after short term-ageing with RTFOT for all the
three bituminous binders of this work with a significant drop in the number of spins
for bitumen C observed upon PAV. Bitumen A and B kept a constant number of spins
of carbon-centred organic radicals between short- and long-term ageing. The previous
kinetics’ study of Chapter 3 suggested that the oxygen-centred radicals, such as •OH,
may abstract protons attached to benzyl rings and could yield to carbon-centred
radicals [30]. The point at which the plateau of this type of radicals appears is
considered the onset of the slow-rate phase. Hence, the differences between bitumen
A, B and C suggest that the intermediate oxidation products, such as the organic
carbon-based radicals, are affected differently by ageing. Furthermore, the different
initial higher number of organic radical spins in bitumen C is possibly related to the
visbreaking process of this bitumen. In general, the free organic carbon-centred
radicals appear already in the unaged samples and this may be a critical parameter for
the subsequent ageing severity of each bitumen.
When it comes to the VO2+ centres, identified with the EPR analyses, the present work
confirms for these metal ions earlier observations of Chapter 3, namely that these
centres remain in general unaffected by the oxidation process, as it is evidenced by
the stabilisation of the number of spins in all the ageing states (Figure 4-4 [right]). This
again raises a significant point about a possible correlation between the vanadyl
species and the vanadium content, present in bitumen’s composition since it can be
exploited as an indicator and marker of the origin following [166].
61
Figure 4-4 Evolution of carbon-centred radicals [left] VO2+ centres [right] with lab ageing
Up to this point, the differences between bitumen C and binders A and B may affect
the initial number of spins of organic radicals in the unaged state. The different
penetration grade between bitumen A and B implies also that the distillation grade of
the same crude oil may have an effect on the VO2+ centres with the softer bitumen B
presenting the higher number of VO2+ centres spins.
1H-NMR
Spectroscopic 1H-NMR analyses were evaluated to gain insights into possible changes
of proton distribution upon ageing and thus the chemical composition of bitumen. An
inspection of the spectra in all the ageing states for bitumen A (Figure 4-5) revealed a
large peak at 0.90 ppm assigned to alkyl (methyl) protons and a peak at 1.28 ppm
related to alkyl (methylene) protons, something which was observed also for binders
B and C. After long-term ageing with PAV, bitumen A showed a peak around 1.67 ppm
ascribed to alkyl (methine) protons. The same peak was observed already for bitumen
B-RTFOT and C-RTFOT with the intensity of the peak increasing after PAV. The broader
peak between 2.00-2.30 ppm can be assigned to protons on Cα next to carbonyl and
the peak at 2.30 ppm can be assigned to a benzylic proton. No significant peaks were
observed in the olefinic proton region (4.00-6.00 ppm) in any of the ageing states. The
broader signal from 6.50 to 8.50 ppm is related to the aromatic protons.
4. Ageing mechanisms in bitumen via standardised lab simulations
62
Figure 4-5 Example of 1H-NMR spectra in all the ageing states of bitumen A with their main proton peaks
and regions
Although spectral peaks appear somewhat similar apart from the prominent peak at
1.67 ppm upon ageing, integration of specific regions given previously in Table 2-3 (for
which relative percentage distribution was determined) allows for the discussion of
the chemical alterations upon standardised ageing. The relative occurrence of protons
in different chemical environments/ageing states is given in Figure 4-6. All the analysed
samples present a negligible presence of olefinic protons independent of the
bitumen’s ageing state. For clarity, the standard deviations of this specific region are
not presented, whereas standard deviations of all the other regions are included in
Figure 4-6. In binders A and B, the governing region seems to be the methylene,
followed by methyl, α-alkyl and aromatic proton regions, while for bitumen C the
methyl and α-alkyl region appeared in different decreasing percentage order. The main
differences between bitumen A and B are the lower methyl region and the higher
methylene region for bitumen A. Since methylene is the region that contributes
predominantly to the total spectral region of the aliphatic protons
(=Hmethylene+Hmethyl+Hα-alkyl) its lower percentage for bitumen B is reasonable based on
the different distillation grade of the two bituminous binders. In addition, bitumen A
63
and B differ from bitumen C mainly by the lower fraction in the aromatic and the higher
fraction in the methyl regions.
The effect of the ageing state with regard to the relative percentage of protons was
also assessed and presented in Figure 4-6. More specifically, in the hard bitumen A,
the methylene proton region appears to increase slightly from 59.0% in A-Unaged to
59.5% in A-RTFOT and remains constant in A-PAV (59.6%). The aromatic protons follow
the opposite trend (6.0% for A-Unaged to 5.5% for A-PAV). The region linked with the
methyl protons fluctuated for bitumen A (19.3% - A-Unaged, 19.8% - A-RTFOT, 19.1%
- A-PAV).
Figure 4-6 Relative percentage distribution of protons in different ageing states of bitumen A, B and C
Bitumen B originating from the same crude oil presents a similar relative percentage
proton distribution after laboratory short- and long-term ageing. It is interesting that
the α-alkyl proton zone from 16.0% in B-Unaged decreased in B-RTFOT (13.2%) and
PAV (14.2%) compared to the unaged state. A fluctuation for the methyl protons was
observed for bitumen B (increased percentage after RTFOT and decreased percentage
values upon PAV), a trend which was observed also for bitumen A. Changes in the
methylene proton region showed also an upward relative percentage trend with
ageing.
Bitumen C, which gave the most apparent difference in the FTIR and the EPR results,
showed a different initial proton relative percentage distribution compared to A and
B-Unaged (higher aromatic and α-alkyl proton percentages, lower methyl proton
percentages) but demonstrated, in general, the same trends with ageing. More
specifically, bitumen C showed a decreasing trend after short-term ageing with RTFOT
for the α-alkyl proton region shifting respectively from 16.5% in the C-Unaged to 15.7%
in C-RTFOT and in 15.8% in the C-PAV, something which was observed also for bitumen
B. Similar to binders A and B, increasing trends in the methylene and methyl proton
regions of bitumen C were also observed, with the methyl proton region increasing
with RTFOT ageing. The aromatic region appeared to decrease with ageing for bitumen
C.
4. Ageing mechanisms in bitumen via standardised lab simulations
64
Obstacles of exploratory studies to identify chemical differences upon ageing in 1H-
NMR were overcome by grouping the main proton categories, acknowledging always
the limitations that may exist for the application of this technique to bitumen.
Therefore, bitumen A and B exhibited in general similar relative percentage
distribution of protons as expected for the same crude oil in the unaged state but also
upon laboratory ageing where the percentage distribution of protons was quite
similar.
TOF-SIMS
Molecular investigation of the oxygenated products after the combined effect of short-
and long-term ageing was conducted with TOF-SIMS on the surface of the three
bituminous samples. It has been noted that ageing may affect the compatibility of the
wax present in bitumen C to appear more pronounced upon ageing on the surface of
the films. Although clear differences cannot be seen between the different ageing
states i.e. in the spectra of bitumen A (Figure 4-7) a thorough inspection of certain
fragment ions assisted to form a more clear view of the products. M/z values in the
negative ion spectra of all the bituminous tested samples were classified into groups
of ion fragments with generic formulas RSOx and RHOx given in Table 4-2. Selected m/z
values in the positive ion spectra were assigned to aromatic and aliphatic ion
fragments (Table 4-2).
The findings of the (OH)x-containing fragments of bitumen A and B and partially for
bitumen C point out that other products such as alcohols/ethers or carboxylic acids
can be formed. This is evident by the higher intensity of most of the (OH)x-containing
fragments (attributed possibly to alcohols) after PAV compared to the unaged state
which was not observed for specific ion fragments (C2OH-, C6OH-) of bitumen C (Figure
4-8 [left]), most probably as a result of the prominent presence of wax in its surface.
65
Figure 4-7 Example of negative [top] and positive
[bottom] TOF-SIMS ion spectra of bitumen A in the
unaged state and after PAV.
Table 4-2 Utilised negative and positive
oxygenated ion fragments before and after
PAV
Ion
Observed
mass (m/z)
Assignment of
molecular structure
OH-
17.003
(OH)x-containing
C
2
OH-
41.005
(OH)
x
-containing
CHSO-
60.974
SO
x
-containing
C
2
H
3
SO
74.989
SO
x
-containing
C
6
HO-
89.001
(OH)
x
-containing
C3H7+
43.055
Aliphatic
C
4
H
7
+
55.054
Aliphatic
C
4
H
9
+
57.070
Aliphatic
C
5
H
7
+
67.049
Aliphatic
C
5
H
9
+
69.068
Aliphatic
C
5
H
11
+
71.088
Aliphatic
C
6
H
9
+
81.065
Aliphatic
C
6
H
11
+
83.085
Aliphatic
C
6
H
13
+
85.104
Aliphatic
C
7
H
11
+
95.079
Aliphatic
C
7
H
13
+
97.102
Aliphatic
C
9
H
7
+
115.033
Aromatic
C
10
H
8
+
128.036
Aromatic
C
13
H
9
+
165.035
Aromatic
Figure 4-8 Intensity of (OH)x-containing compounds [left] and SOx-containing compounds [right] with lab
ageing
Similar trends follow for the increase of the ion fragments with generic formula RSOx
given in Figure 4-8 [right]. The results of bitumen A and B support the formation of
considerable amounts of sulfoxide-containing compounds after the sequential short
and long-term ageing as the intensities of A-PAV and B-PAV remain higher than the
unaged state. Caution should always be taken concerning the specific sulfoxide-
containing fragments observed with TOF-SIMS which give a partial view of the total
sulfoxides present in bitumen which FTIR might capture in a greater penetration depth;
this means that limitations should be again acknowledged when comparing the
4. Ageing mechanisms in bitumen via standardised lab simulations
66
different techniques. Potentially, the results of the two surface techniques together
could be used to exploit the surface effects of oxidation in bituminous films. Bitumen
C seems to reduce the sulfoxide-containing ion fragments with ageing something
which does not agree with the FTIR results. This may be explained by the segregation
of wax, present in this bitumen, on the surface of the bituminous film, resulting in
aliphatic compounds which cover most of the surface [30,167].
Figure 4-9 Intensity of PAH fragments [left] and aliphatic fragments [right] with lab ageing
The analysis of selected ions in the positive ion spectra (Figure 4-9 [left]) assigned to
PAH is also discussed. The intensity of PAH of B-Unaged and A-Unaged increased
slightly after PAV, whereas a significant drop of PAH with ageing for bitumen C was
captured, probably due to the increase of the waxy particles on the surface upon
ageing. Finally, binders A and B showed an almost constant behaviour (or a slight
increase) of aliphatic ion fragments upon PAV contrary to bitumen C which showed a
decrease for specific aliphatic ion fragments (Figure 4-9 [right]). The intensity of the
aliphatics for bitumen C appeared the highest, a fact that can be explained by the wax
present in this binder and the association of wax-related particles with aliphatics on
the surface of this bitumen.
Oxidation products in SARA fractions
FTIR
The chemical fingerprinting of asphaltenes and maltenes of binders B and C that differ
in terms of wax presence, as well as of their parent bitumen samples were investigated
in FTIR. The effect of ageing on the bitumen was somewhat as expected. Increases
were mainly seen in carbonyl and sulfoxide signals. As shown already in Figure 4-2 only
bitumen B shows a carbonyl signal in the unaged state, related to the fact that this
binder is produced from an acid crude oil or a crude with high acidity.
67
For asphaltenes, the IR spectra were different from the bitumen samples and maltenes
(which displayed very similar spectra). The signals associated with the stretching or
bending vibrations of saturated hydrocarbons at about 2920 cm-1, 2850 cm-1, 1460 cm-
1 and 1376 cm-1 are much smaller compared to those for the binders.
Differences between the bitumen fractions and the effect of long-term lab ageing are
further shown by normalised peak areas in Figure 4-10. Due to the considerable
differences in the symmetric and asymmetric bending vibrations of the aliphatic
groups at 1460 cm-1 and 1376 cm-1 for the different fractions, the normalised areas
were preferred to be used for the fractions instead of the indices introduced in Chapter
2. For the area normalisation, the procedure described in [17] was adopted. For this
procedure, first, all the spectra are shifted to zero absorbance at a fixed wavenumber
of the lowest spectrum absorbance. Then an absorbance correction factor is applied
to scale the spectra at the asymmetric stretching vibration of the aliphatic structures
at 2923 cm-1. The entire spectrum is then multiplied by this ratio factor.
For bitumen C, it is clear that in the unaged state, the asphaltenes do not contain
carbonyls, which all fall into the maltene fraction (see Figure 4-10). After long-term lab
ageing, the asphaltenes contain a large part of the carbonyl and sulfoxide signals,
suggesting that a lot of molecules become insoluble in warm n-heptane after they have
reacted with oxygen to form carbonyl and sulfoxide groups (Figure 4-10). This is the
case for both binders B and C and can explain the increase of the more polar fractions
in asphaltenes due to ageing. The observation is in agreement with that made by
Mirwald and his group [92]. For the maltene fraction, on the other hand, the effect of
ageing is rather small for bitumen B in comparison to bitumen C.
Besides those observations already made on carbonyls and sulfoxides, an obvious
difference was found in the normalised aromaticity area, being ranked as asphaltenes
> bitumen > maltenes, while this area was slightly decreased upon ageing in the
maltenes.
4. Ageing mechanisms in bitumen via standardised lab simulations
68
Figure 4-10 Normalised absorbance peak areas of carbonyls [left] and sulfoxides [right] before and after
PAV for binders B and C
1H-NMR
In Figure 4-11, the relative hydrogen percentage distributions of different types of
protons (methyl, methylene, α-alkyl, olefinic, aromatic) are plotted as the average of
the two replicates with their standard deviation. The percentage distributions are
shown for the maltenes, whereas for the corresponding bitumen before and after
long-term lab ageing, the results can be found previously in Figure 4-6. No results are
shown for asphaltenes since this fraction was not completely dissolved in deuterated
tetrachloroethane. Figure 4-11 indicates that the hydrogen percentage distributions of
the two binders are quite different in the maltene fractions. The maltenes from
bitumen B show a higher methyl region, lower α-alkyl region and lower aromatic
region compared to the maltenes from bitumen C. The differences between the
maltenes fractions can already be seen in the two bitumen samples (Figure 4-6),
possibly indicating that the binders were produced from different crude sources
and/or by different refining processes. The two graphs together also show that ageing
affects in a similar manner the two maltene fractions and their parent binders, i.e. an
increase is observed for methyl and methylene regions and a decrease seems to hold
for aromatic and α-alkyl regions. In all the cases, the olefinic region is negligible. The
changes upon ageing are more obvious for the maltene fraction of bitumen C as
compared to that of bitumen B. This is also the case when the two bitumen samples
were compared.
69
Figure 4-11 Relative percentage distribution of protons in maltenes of bitumen B and C before and after
PAV
DSC
The DSC thermograms of the parent bitumen C and its maltenes in the heating scan
revealed an endothermic reaction, which is interpreted as melting of the wax in this
bitumen. Thus, the melting enthalpy can be used as a measurement of the wax present
in this sample. No results of the thermal properties are shown for asphaltenes since
asphaltenes were not melted under the testing conditions. Bitumen C and its maltenes
fraction displayed a similar melting starting temperature of slightly above -15 °C, but
they differ in melting out temperatures, which was about 90 °C for the parent bitumen
C and 80 °C for the maltenes. The wide range of temperatures during which the melting
takes place points towards the presence of molecules with different melting points.
The melting of a given molecule also depends on the surrounding chemical
compositions, and this is the case when comparing the bitumen and the maltene
fraction. On the other hand, the wax melting starting temperature and melting out
temperature were not affected by ageing. For bitumen B, as well as its maltene
fraction, no wax crystallisation and melting were detected by DSC but only a glass
transition temperature. This supports that bitumen B is a wax-free bitumen.
Regarding the melting enthalpy in Figure 4-12 [left] the maltene fraction of bitumen C
shows a higher value than the corresponding bitumen. Figure 4-12 [left] also shows
that, after long-term ageing, the melting enthalpy of the bitumen becomes lower,
while for the maltenes the effect of ageing is likely the opposite, i.e. higher melting
enthalpy. These results suggest that for the investigated bitumen C, more wax is
present in the maltenes when the bitumen is fractionated, especially after long-term
ageing. Finally, the effect of glass transition temperature is clear between the maltenes
and the parent binders B and C, namely the maltenes experience consistently a lower
glass transition temperature both before and after long-term lab ageing for both
4. Ageing mechanisms in bitumen via standardised lab simulations
70
binders. Ageing seems to decrease the glass transition temperature for the maltenes
of the two binders whereas the opposite trend is found for the parent binders (Figure
4-12 [right]).
Figure 4-12 Melting enthalpies of parent bitumen C and its maltenes [left] and glass transition
temperatures [right] of binders B and C and their maltenes with lab ageing
Highlights of the chapter
FTIR supports the increase of the oxygenated indices of sulfoxide and carbonyl
with lab ageing, for all binders, independently of bitumen source, distillation
process and performance.
Differences in the extent of carbonyl and sulfoxide formation are observed
between the various binders, especially for those of different empirical
performance, with long-term lab ageing.
EPR analyses explore the role of the intermediate oxidation products, the
organic carbon-centred radicals, which were increased for all binder types with
short-term lab ageing.
The EPR VO2+ centres remain relatively constant and are almost insusceptible
to lab ageing.
1H NMR demonstrates slight changes in the relative proton distribution upon
lab ageing for all the studied binders.
Sulfoxide-containing compounds via TOF-SIMS are increased significantly after
the sequentially short-and long-term lab ageing for the straight-run binders.
(OH)x-containing fragments suggest that apart from carbonyls, other products,
such as alcohols and carboxylic acids, can be produced with lab ageing
regardless of the bitumen type.
71
After lab ageing of bitumen, a large part of the carbonyl and sulfoxide signals
can be found in the asphaltene fraction.
1H-NMR relative proton distribution changes with lab ageing in a similar
manner both for maltenes and the parent bitumen.
FTIR indicates that asphaltenes are the most aromatic fraction followed by the
parent bitumen and maltenes.
After bitumen fractionation, more wax is found in the maltenes fraction
compared to the parent bitumen, and this is even more evident when the
bitumen is lab aged.
The ageing mechanisms of bitumen with standardised lab simulations were
explored in this Chapter. A comparison between the oxidation products of routine
lab ageing simulations and the ones found via kinetics is of certain value to be
performed in order to understand more clearly the intensity of the chemical
changes that take place in bitumen. This will be partially covered in Chapter 7 of
the current dissertation.
4. Ageing mechanisms in bitumen via standardised lab simulations
72
*This chapter is redrafted from: G. Pipintakos et al., Crystallinity of bitumen via WAXD and DSC and its
effect on the surface microstructure, Crystals (2022). https://doi.org/10.3390/cryst12060755
5
5 MECHANISMS OF SURFACE
MICROSTRUCTURE IN
BITUMEN
Summary*
It is well documented that most bituminous binders contain crystallisable material and
all of them consist of asphaltenes. The crystallisable fraction, often referred to as
paraffinic or natural wax, is associated with the bitumen’s origin and has an influence
on its rheological performance and microstructure. Before diving into a systematic
investigation of the ageing effect on bitumen’s microstructure, an improved
understanding of the underlying mechanisms is of paramount importance. In general,
there are two main theories with regard to the appearance of bitumen’s surface
microstructure, namely due to asphaltenes and because of wax in bitumen. Chapter 5
adopts the second theory and explores the potential crystallisation and melting
process of a waxy and a wax-free bitumen via three different approaches: DSC, WAXD
and CLSM. The findings of this Chapter reveal that the DSC transitions of the waxy
bitumen are in good agreement with the corresponding occurrence of WAXD signals
and to some extent with the formation and disappearance of the surface
microstructures which were followed in real-time at two cooling and heating rates.
WAXD results additionally demonstrate that the crystalline material in bitumen is
organised in an orthorhombic unit cell, typical for straight-chain aliphatic structures.
On the other hand, DSC and WAXD support the lack of crystallinity for the wax-free
bitumen which is a hypothesis that could explain its featureless CLSM surface. Overall,
the originality of this Chapter resides in the disclosure of connections between
5. Mechanisms of surface microstructure in bitumen
74
crystallographic properties and thermal transitions and examines a certain hypothesis
of the possible mechanisms behind the surface micromorphology of bitumen.
Objectives
The aim of Chapter 5 is first to compare the thermal transitions observed in DSC,
associated with the melting and crystallisation behaviour of natural waxes in bitumen,
to the crystallographic findings via WAXD, during cooling and heating in the same
temperature range. This was based on the hypothesis that wax crystallinity is one of
the possible reasons in the literature for the formation of various morphological
structures on a bitumen surface, with the other one being the link of asphaltenes with
the microstructure [102]. Following the first hypothesis, two bituminous binders, a
waxy binder A and a wax-free binder B, were used in this Chapter. The effect of
annealing at room temperature was also investigated in DSC and CLSM for the waxy
bitumen. Overall, this Chapter manages to provide an explanation for a certain
hypothesis about the mechanisms governing the formation of certain microstructural
patterns in bitumen, known as bee structures. As mentioned in Chapter 1, if a
relationship between these microstructures with the wax presence holds, will be of
paramount importance to be better understood for the prevention of a number of
implications in the performance of bitumen (i.e. increased low-temperature cracking)
and its production (i.e. deposition problems). A schematic overview of the research
methodology adopted in this Chapter 5 together with the main goals of it is provided
for the ease of the reader in Figure 5-1.
Figure 5-1 Flowchart of the experimental part of Chapter 5 and objectives
75
Materials and methods
Based on the preliminary DSC wax determination proposed in Chapter 4, in this
Chapter a waxy bitumen A and a wax-free bitumen B were selected with the potential
to observe differences in crystallinity and surface morphology. Their empirical
properties are summarised in
Table 5-1. In this work, the selected softer bitumen A was the residue of a visbreaking
conversion process during which additional wax may have been formed, while bitumen
B is a harder straight-run bitumen, originating from a different crude, containing no
wax.
Table 5-1 Basic properties of the waxy and wax-free bitumen
The two binders, A and B, were further tested in DSC cooling/heating cycles and after
annealing as well as at ambient temperature and cooling/heating cycles in WAXD. To
follow the microstructure evolution on CLSM, an in-house thermal fixture was
assembled by the Road Engineering Research Section (Figure 5-2), consisting of
electrical variable resistors that were fixed accordingly to obtain the required
temperature on top of a metallic plate. The temperature of this small heating element
was adjusted based on calibrated thermocouples inserted in the bituminous sample
on the glass plate. Investigations on the waxy bitumen’s microstructure were
conducted in cooling and heating cycles, in a range of temperatures between 10 to 80
°C at a cooling/heating cycle of 1.25 °C/min and a lower cooling/heating cycle using 0.5
°C/min rate. Preliminary investigations showed featureless surfaces for temperatures
above 65 °C in both cycles as well as for the wax-free bitumen B which was not
investigated further. Additionally, the effect of annealing at room temperature was
also investigated for the waxy bitumen after 24 hours in a heating scan.
Material
Property
Binder
Test Method
Bitumen
A
B
Penetration 25 °C (0.1 mm)
80
16
EN1426
Softening point (°C)
45.8
61.1
EN1427
Penetration index, Ip
-1.20
-1.06
EN12591
5. Mechanisms of surface microstructure in bitumen
76
Figure 5-2 Bituminous sample placed on the in-house heating element under CLSM
Differential Interference Contrast (DIC) CLSM images with a laser and white light
source on the surface of the bituminous samples were recorded during real-time
microstructure evolution. A DIC CLSM image has the advantage of introducing
moderated contrast to optical images, producing a pseudo-3D effect. To additionally
assess the evolution of the bee structures a meaningful, scale-free metric was used,
namely the bee area percentage (bee coverage). The image post-processing is based
on a 90-95% height or depth threshold, using the height differences between the
valleys and peaks of the bee structures and the rest of the bituminous surface [174].
More specifically, masks of the bee structures are created based on these height
thresholds both for valleys and peaks and division with the total image area allows the
determination of the bee area coverage.
Mechanisms of bitumen microstructure
DSC
DSC scans obtained during cooling/heating for both binders and heating after 24 hours
annealing temperature at 25 °C for bitumen A are shown in Figure 5-3. The cooling scan
of bitumen A shows an exothermic transition with a maximum at around +30 °C, which
is attributed to the wax crystallisation. This transition is indicated by the area A in
Figure 5-3 [top]. It continues down to the glass transition which appears for bitumen
A at around -30 °C. The cooling scan of the wax-free bitumen B confirms the absence
of crystallisable components as no exothermic transition can be observed whereas the
glass transition temperature for this bitumen can be found around -15 °C.
77
Figure 5-3 DSC cooling scan [top], DSC heating scan for bitumen A and B and heating after annealing at
25 °C for 24 hours for bitumen A [bottom]
In the subsequent heating ramp of the waxy bitumen A, denoted as ‘heating scan after
cooling’, an endothermic transition can be clearly observed (area C), which is related
to wax melting. The melting end temperature is much higher than the corresponding
crystallisation onset temperature for this bitumen indicating the need of nucleation at
a certain degree of undercooling. The heating scan of bitumen A also reveals a small
exothermic signal close to the glass transition, which is associated with a cold
crystallisation in the literature [175,176], denoted as area B in Figure 5-3 [bottom].
Both heating and cooling scans of bitumen A show wide temperature ranges for
crystallisation and melting, potentially indicating large differences between the
crystals’ sizes and perfection. Changing wax concentrations in the liquid while the
crystallisation is progressing, may also have contributed to the width of the transition.
Indeed, as the crystallinity progresses upon cooling, the remaining melt becomes wax
depleted by which lower temperatures are needed to induce further crystallisation
[177]. This effect, which is common for crystallisation from mixed melts is mirrored for
melting upon heating. On the other hand, bitumen B appears to have no apparent
transition phenomena which occur during the heating scan.
Further, the melting scan recorded after the annealing for bitumen A presents a
different shape as illustrated by the green curve in Figure 5-3 [bottom]. In this case,
the melting signal consists of two peak temperatures, with the annealing temperature
being located between these two peak temperatures of melting, indicating that during
5. Mechanisms of surface microstructure in bitumen
78
the annealing period, recrystallisation takes place forming crystals with a higher
melting temperature.
In Table 5-2, the onset and peak crystallisation temperatures (Tc), the peak and end
melting temperatures (Tm) and the crystallisation and melting enthalpies (ΔΗ) are
presented. The main difference between the crystallisation and melting enthalpy is
mainly due to the additional cold crystallisation during the heating scan. The
undercooling effects are again obvious since Tm-end is much higher than the Tc-onset.
Furthermore, the melting temperature range of bitumen A before and after annealing
slightly differs in accordance with the results of a Round Robin study where the same
bitumen was used [74]. This past study also presented the repeatability and
reproducibility between the different laboratories before and after annealing and the
melting, crystallisation onset and peak of the current study lie adequately well within
the previously reported range. In the current study of Chapter 5 changes in the shape
of the melting after the second heating were additionally observed, while the
annealing procedure does not influence significantly the melting enthalpy. This
witnesses the small effects of annealing on the overall transition phenomena.
Table 5-2 Summary of DSC thermal transition data of binders A and B
N/D=Not determined, N/A=Not applicable
WAXD
Initially, a WAXD investigation for both bituminous binders was performed at room
temperature. As can be seen in Figure 5-4 in contrast to bitumen B only bitumen A
contains two sharp crystalline reflections in the WAXD patterns which are due to
orthorhombically organised paraffins as described by Bunn and of which the labelled
110 and 200 reflections are most prominent [178]. The data are shifted along the
intensity axis to avoid overlap and reflections are labelled with their Miller indices. This
observation is in line with the absence of crystallinity expected from the wax-free
crude and initially confirmed with DSC.
Material
Cooling cycle
Heating cycle
Bitumen
Tc-onset
[°C]
Tc-peak
[°C]
ΔH
[J/g]
Tm-end
[°C]
Tm-peak
[°C]
ΔH
[J/g]
A
42.0
29.4
4.5
84.0
54.0
7.2
A after annealing
N/D
N/D
N/D
91.0
52.6
6.7
B
N/A
N/A
N/A
N/A
N/A
N/A
79
Figure 5-4 Normalised synchrotron WAXD patterns of bitumen A and B at room temperature (25 °C)
In a second experiment, WAXD patterns for bitumen A, which showed crystalline
reflections at room temperature, were recorded during cooling and heating. A
representation of the temperature-dependent WAXD patterns during the cooling scan
is shown in Figure 5-5, in which the scattered intensities are presented by means of a
grey scale. Darker regions correspond to lower scattering intensities while the 110 and
200 reflections are labelled accordingly. It appears that the orthorhombic 110 and 200
reflections are present between roughly +35 and -40 °C and increase upon cooling.
Figure 5-5 Time-temperature resolved synchrotron WAXD patterns during cooling at 10 °C/min from 100
to -40 °C for bitumen A
From the integral of the 110 and 200 reflections, a crystallinity index can be obtained
as described in Chapter 2. The index was arbitrarily normalised to its value at -40 °C
during the heating run. In Figure 5-6 this crystallinity index is plotted during the cooling
(black data) and heating (red data) scans. For better visualisation of the crystallinity
evolution within bitumen A, a smooth sigmoidal fitting curve was added. At low
5. Mechanisms of surface microstructure in bitumen
80
temperatures, both the cooling and heating scans show a plateau of the crystallinity
up to 10 and 30 °C respectively, followed by a steep decrease in intermediate
temperatures. At high temperatures, the wax within bitumen A is fully molten and the
crystallinity equals zero. The end melting temperature is around 70 °C for the heating
scan which is in agreement with the DSC reported temperature. Similar to DSC,
undercooling effects are observed by WAXD and the onset of crystallinity at around 50
°C corresponds very well to the DSC temperatures. The starting value of the
crystallinity in the heating run also seems to be slightly higher than the end value of
the cooling run, indicating that additional crystallinity was generated during the short
isothermal stay at -40 °C.
Figure 5-6 Evolution of synchrotron WAXD-based crystallinity index of bitumen Α during cooling and
heating, where dots are experimental data points and full lines are sigmoidal guides
Evolution of bitumen microstructure
CLSM
With CLSM the percentage of the bee structures (% of their total area) for bitumen A
is given in Figure 5-7, within the temperature range between +10 to +80 °C, during
cooling and heating at two different rates. Only the microstructure evolution of
bitumen A was investigated, as the non-waxy bitumen B showed no microstructural
characteristics in accordance with other studies [72,174].
The analysis of the CLSM images shows that during the faster cooling/heating rate the
bees start to form at +50 °C and their percentage increases quickly when cooling
further between +50 and +30 °C. In the subsequent heating cycle, the bee percentage
decreases in a slightly larger temperature range, between +30 and +60 °C. For the
81
cooling and subsequent heating cycle recorded at a slower rate, the effects are very
similar but the temperatures are shifted to higher values. This indicates that during a
slow cooling, there is more time to form larger and more perfect crystals which in the
corresponding heating cycle have a higher melting temperature. In addition, in the
slower heating and cooling scan the bee percentage seems to develop in two steps, a
steep change between 0 and 2% and then a less steep change up to 2.7% bee area
Some possible reasons for this observation could be related to the parallel viscosity
increase/decrease in the binder upon cooling/heating which at a certain point may
start to hinder the crystallinity formation or the phase separation of the melt in
paraffin rich and paraffin poor phases before crystallisation.
In general, the observations support that the appearance/disappearance of WAXD
patterns, coincide well with the observed changes in the bee structures and with the
thermal observations in DSC. Especially, if one considers that the experimental
conditions are not exactly the same, as cooling rates are much slower in CLSM
compared to DSC and WAXD and that microscopy is following a surface feature while
DSC and WAXD are measuring bulk properties.
Figure 5-7 Evolution of CLSM bee structures during heating and cooling scans, where markers are
experimental data points whereas full and dash lines are polynomial guides
In Figure 5-8 and Figure 5-9, selected temperatures of the cooling respectively heating
scan for bitumen A (rate 1.25 °C/min) are represented. It can be seen in Figure 5-8 that
progressively the bee structures are formed during cooling. The bee structures appear
at temperatures below 50 °C and they can be clearly seen at 45 °C. Similarly, as shown
in Figure 5-9, the bee structures disappear from the surface at temperatures greater
than 60 °C during heating, in line with the end melting temperature of DSC. Apart from
the bee structure formation upon cooling, it was noticed that the surface becomes
5. Mechanisms of surface microstructure in bitumen
82
coarser upon further cooling, whereas any potential vertical-like artefacts are due to
small sample movements during cooling/heating.
Figure 5-8 CLSM-DIC images of bitumen A at selected temperatures during the fast cooling scan (1.25
°C/min)
Figure 5-9 CLSM-DIC images of bitumen A at selected temperatures during the fast heating scan (1.25
°C/min)
For the sample that was annealed 24 hours at room temperature (Figure 5-10), this
was prepared by placing a drop of binder on a hot plate (120 °C) and cooling it almost
immediately to room temperature. When heating this binder at 1.25 °C/min, it was
noted that in this case, the bees disappeared already at 55 °C, so even faster as
observed in the case of Figure 5-8 and Figure 5-9, where a sample was first cooled and
subsequently heated at 1.25 °C/min rate. Hence, it seems that the cooling rate before
recording the heating scan, rather than an isothermal annealing, determines at what
temperature the bees disappear in the subsequent heating scan.
83
Figure 5-10 CLSM-DIC images of bitumen A at selected temperatures during the heating scan (1.25
°C/min) after 24 hours annealing at 25 °C
Highlights of the chapter
DSC and WAXD relate thermal transitions to changes in the X-ray diffraction
pattern.
The correspondence between the DSC, WAXD and CLSM and the temperature
ranges, where patterns are formed and disappear, confirm that wax
crystallinity is the reason for the endo- and exothermic transitions and
support, based on a wax hypothesis, the bee structure formation in bitumen.
The wax-free bitumen displayed a glass transition and no other thermal effects
or surface structure.
Annealing affects the shape of the broad DSC endotherm of the waxy bitumen
with a negligible effect on the melting enthalpy.
The WAXD results at room temperature showed crystalline diffractions for the
waxy binder.
The diffraction pattern by WAXD showed that the crystalline material in the
waxy bitumen has an orthorhombic unit cell.
Heating and cooling scans revealed that the WAXD pattern for the waxy
bitumen disappeared at around 70 °C (in heating) and was formed at around
50 °C (in cooling).
The CLSM indicates that on the surface of the waxy binder bee structures are
formed, upon cooling, which disappear upon heating.
The cooling/heating rate has an influence on the temperatures where bee
structures are formed or disappear.
5. Mechanisms of surface microstructure in bitumen
84
*This chapter is redrafted from: G. Pipintakos et al., Coupling AFM and CLSM to investigate the effect of
ageing on the bee structures of bitumen, Micron (2021). https://doi.org/10.1016/j.micron.2021.103149
6
6 EFFECT OF LAB AGEING ON
BITUMEN MICROSTRUCTURE
Summary*
Some theories postulate the effect of crystalline wax on the so-called bee structures
of bitumen’s microstructure while others support that asphaltenes is the governing
reason for such structures in bitumen. Although it is widely accepted that ageing has
an effect on this unique microstructure yet conflicting literature exists on clear trends.
Chapter 6 explores the effect of lab ageing on the bee structures of bitumen,
employing two advanced microscopic techniques: an AFM and a CLSM. Having
supported to some extent a certain hypothesis about the mechanisms that govern the
bitumen microstructure in a previous Chapter, here four waxy and two wax-free
bituminous binders are investigated before and after sequential laboratory short- and
long-term ageing. Chapter 6 demonstrates that the number of bees per μm2 and the
bee area coverage decrease with ageing, whereas their size is increasing, contributing
to the hypothesis that ageing has a clear effect on the microstructure. A systematic
analysis of the waveform characteristics is also provided for the peaks and valleys as
well as the shape probabilistic values of the bee structures. In conclusion, the results
of the two techniques are in good agreement, reporting similar trends upon ageing for
the bee coverage. Differences are mainly identified in the waveform calculations.
Additionally, the efforts undertaken to capture bitumen’s microstructure via deep
learning and advanced image processing techniques are conceptually explained. The
systematic investigation in this work (commercial image processing or sophisticated
machine learning) assists in enhancing the understanding of the effect of ageing on the
surface microstructure.
6. Effect of lab ageing on bitumen microstructure
86
Objectives
Although considerable research has been devoted to the identification of the bee-like
structures and their accompanying phases there is no consensus on how these are
affected by ageing. Typically, 3 to 4 phases can be identified in bitumen’s surface with
AFM imaging: the catanaphase (bee structure), the periphase (around catanaphase),
and the perpetua phase which can be distinguished in the paraphase (solvent regions)
and the salphase (high phase contrast spots) [74,179]. A simplified example of this
phase detection is given in the example Figure 6-1 before and after ageing.
Figure 6-1 Typical phases of surface bitumen microstructure before [left] and after ageing [right] in AFM
phase contrast and topography respectively (reprinted by permission from [74])
Yet controversy exists whether prolonged and more severe ageing increases [124,180–
184], decreases [185,186] or fluctuates [90,126,187] i.e. the length and area of the
bee structures. On the other hand, a number of studies [124,126,183–185,188–191]
found consistent changes with ageing on the surface stiffness, adhesive and cohesive
strength and force, with most of them claiming an increase of these properties.
Consequently, no clear view exists about the wavelength, height and roughness of the
sequence of peaks and valleys of the catanaphase of the bee structures as well as the
bee area percentage and length in bitumen upon ageing.
Acknowledging this gap in the literature, Chapter 6 aims to systematically examine the
change of the most apparent bitumen surface microstructure, widely known as bee
structures, upon ageing. Based on the hypothesis that wax crystallinity may be the
reason for such structures, it examines four bituminous binders containing natural wax
with respect to their bee characteristics of the catanaphase and two control wax-free
binders before and after long-term laboratory ageing by utilising AFM and CLSM. This
analysis adds a clearer view to the existing literature and attempts to establish links
between the observations of the two microscopic techniques and their convergence.
The experimental flowchart together with the objectives of Chapter 6 is given in Figure
6-2.
87
Figure 6-2 Flowchart of the experimental part of Chapter 6 and objectives
Materials and Methods
The bituminous binders of this Chapter were selected to contain surface bee
structures, based on the partial support provided for the wax hypothesis about the
mechanisms of bitumen’s microstructure in Chapter 5. Preliminary microscopic
investigations on several wax-free binders confirmed that such binders display
featureless surface topography before as well as after ageing, independent of the
bitumen type.
Thus, four waxy binders designated with letters A to D, having similar empirical
properties were chosen as well as two wax-free control binders E and F, of different
penetration classes. For consistency, the images of only one of the two wax-free
binders are presented hereafter.
The waxy binders differ in the refinery process, more specifically, bitumen B, C and D
are derived after a mild form of thermal cracking, namely visbreaking. Bitumen C was
further treated in an air-blowing unit at elevated temperatures, a process that results
in an increase in stiffness and softening point making the product more suitable for
road applications. For bitumen A, the exact production details were not available. All
four waxy binders contained natural wax (crystallisable compounds) which was
confirmed by the melting enthalpies derived from DSC heating scans. Wax-free
binders, like binders E and F, do not show endo- or exothermic transitions. Basic
empirical properties and DSC melting enthalpies of all the binders are given in Table
6-1.
Routine laboratory ageing simulations were performed for all the binders according to
the European standards for short- and long-term ageing [40,41] simulated with RTFOT
followed by PAV for 20 hours. For all the subsequent microscopic observations the
bitumen in the unaged state was designated as ‘X unaged’ and as ‘X aged’ after the
Quantification
of the bee
structure
characteristics
AFM
CLSM
Unaged and
RTFOT+PAV
samples
4 waxy and 2
wax-free
binders
6. Effect of lab ageing on bitumen microstructure
88
sequence of RTFOT and PAV, where X refers to the letter representing one of the
binders.
Table 6-1 Empirical and DSC properties of the bituminous binders in the unaged state
Property
Bitumen
Test Method
A
B
C
D
E
F
Penetration 25 °C (0.1 mm)
42
52
52
67
187
64
EN1426
Softening point (°C)
51.3
49.0
49.8
46.8
38.5
47.7
EN1427
Penetration index, Ip
-1.27
-1.37
-1.17
-1.36
-1.80
-1.23
EN12591
Melting enthalpy (J/g)
2.8
6.8
6.2
7.6
0.0
0.0
[107]
Next, images were captured in the two ageing states for all the examined binders with
AFM and CLSM. After sample preparation, as described in Chapter 2, scans are
performed in air at 25 °C after a storage time of 24 hours, to allow the microstructures
to develop and stabilise [73].
For the AFM image processing, a flat baseline plane was first obtained for the images.
Then the contour scale was fixed according to the height of each topographical image
to stress the height differences. For the identification of the bee structure, different
masks were applied for the peaks and valleys of the topographical image, by imposing
specific boundaries for the height and slope. Consequently, the total bee
microstructure area was assessed. Bee profiles, wave properties and statistics were
additionally evaluated before and after ageing. An example of the bee detection for
bitumen A original is shown in Figure 6-3. The two colours represent the peaks and
valleys while the overlapping is excluded by implying the height threshold above or
below the reference plane where no bees occur. For the CLSM image processing, first,
the same procedure described in Chapter 5 was followed for the bee characteristics of
the topographical CLSM images, whereas additionally a deep learning-based object-
detection model that can detect the bee patterns in the DIC CLSM images was
proposed.
Figure 6-3 Example of the masks applied in AFM topography for peaks [left] and valleys [right] in
bitumen A unaged
89
Morphology and bee coverage
AFM images of topography, phase contrast and 3D views are presented before and
after ageing in Figure 6-4. The corresponding topography and 3D views are given for
CSLM in Figure 6-5. As expected, the reference wax-free binder E is featureless before
and after long-term lab ageing as witnessed both by AFM and CLSM topography images
and can be seen in Figure 6-6. The same featureless bitumen surface displayed also the
wax-free binder F. Furthermore, a visual inspection of the AFM phase images for
bitumen A and D shows that they present two distinguished phases namely the so-
called catanaphase, which is the bee structures, and the periphase surrounding the
bee structures. These phases are becoming more pronounced upon ageing. For
bitumen A the periphase forms an elliptical shape before and after ageing, while for
bitumen D a star shape can be seen before ageing transforming into a spherulite phase
after ageing. The elongated bee structures of bitumen C are also striking for the
catanaphase of bitumen C before ageing, turning into a spherulite phase after ageing.
A-unaged
A-aged
B-unaged
6. Effect of lab ageing on bitumen microstructure
90
B-aged
C-unaged
C-aged
D-unaged
D-aged
Figure 6-4 AFM topography [left], phase contrast [middle] and 3D view [right] images of waxy binders
before and after long-term lab ageing
91
On the other hand, CLSM is not able to detect the surrounding environment of the bee
structures, although the apparent catanaphase is observable. The height of the valleys
and peaks of the catanaphase seems to be higher for bitumen B and C as can be
revealed by the height extrema of the 3D views in AFM and CLSM. A detailed analysis
of the wave characteristics of the bee structures is provided in the following section of
this Chapter.
A-unaged
A-aged
B-unaged
B-aged
6. Effect of lab ageing on bitumen microstructure
92
C-unaged
C-aged
D-unaged
D-aged
Figure 6-5 Topography [left] and 3D view [right] CLSM images of waxy binders before and after long-term
lab ageing
93
E-unaged
E-aged
Figure 6-6 Topography of wax-free binder E before [up] and after long-term lab ageing [down] for AFM
[left] and CLSM images [right]
When it comes to the effect of ageing, it can be easily observed from Figure 6-4 and
Figure 6-5 that for bitumen A, B and D long-term lab ageing results in a decrease in the
number of independent bees per μm2 and an increase in their size in line with limited
observations in literature [184]. To clarify these visual observations the processing of
the masks, applied with commercial image processing software packages, for the
catanaphase is discussed further. In Table 6-2, the relative bee density is expressed via
the total bee area percentage (bee coverage) and the number of identified bees per
μm2. Table 6-2 indicates that the number of bees per μm2, as well as their bee
coverage, are reduced for all the waxy binders after long-term lab ageing. The data
additionally support that ageing has the same decreasing trends on certain
characteristics of the bee structures, independently of the bitumen type.
Based on the results of the previous Chapter 5, the change of the bee properties can
be interpreted as additional crystallisation of the present wax due to ageing and
additional investigation into the changes of melting enthalpy upon ageing needs to be
performed to confirm it. For the unaged state, it seems that the wax content as
expressed indirectly via the melting enthalpy in Table 6-1 has a weak link with the total
bee area. Moreover, the compatibility of the aged bitumen matrix and the wax should
be investigated i.e. by doped wax content into aged wax-free binders to obtain a clear
view of the reason for the changes in the bitumen microstructure.
6. Effect of lab ageing on bitumen microstructure
94
Table 6-2 Bee coverage and number of bee structures per μm2 before and after lab ageing
Bitumen
Bee coverage (%)
Number of bees/μm2
Unaged
Aged
Unaged
Aged
AFM
CLSM
AFM
CLSM
AFM
CLSM
AFM
CLSM
A
3.85
3.43
1.97
2.83
0.084
0.147
0.052
0.067
B
4.08
4.01
3.46
3.03
0.023
0.045
0.014
0.028
C
3.17
3.29
2.71
2.75
0.006
0.005
0.004
0.003
D
3.12
2.75
0.96
1.13
0.004
0.023
0.003
0.005
The average size of bee structures was calculated from the bee coverage and the
number of bees per μm2, which is shown in Figure 6-7 [top]. The results of Figure 6-7
[top] demonstrate that for all the waxy binders the average size of the bee structures
is increasing upon lab ageing as expected from an initial visual inspection, with the
exception of bitumen D via CLSM. Overall, the average bee size varies between 0.5 and
8 μm2 and a larger size seems to be associated with a higher melting enthalpy and wax
content. For the latter care should be taken for the existing differences in thermal
history between DSC and image acquisition. It is also apparent from the linear
correlations in Figure 6-7 [bottom] that the two microscopic techniques, AFM and
CLSM, are showing similar information.
Figure 6-7 Average bee size derived from AFM and CLSM before and after lab ageing [top] and their
relationships [bottom]
95
Waveform characteristics of bee structures
The typical wavy shape of the bee structures was further assessed. An example of such
a wave profile along with its calculated waveform characteristics is shown in Figure
6-8. In the statistical analysis, for each valley and peak of all the bee structures the
peak height, the valley depth, the wavelength and the projected length were
measured. Average data with their standard deviation are presented in Figure 6-9.
Figure 6-8 Schematic picture of a bee profile and its waveform characteristics in bitumen B
The analysis of these characteristics for the two microscopic techniques examines
further their convergence. In Figure 6-9 [left] AFM shows that bitumen C presents the
most elongated bee microstructure indicated by the highest average projected length
for bitumen C followed by bitumen D, B and A. The projected length from the AFM
data seems to decrease with lab ageing. CLSM on the other hand presents an average
projected length that is similar for bitumen B, C and D, whereas bitumen A shows the
smallest bee length from both microscopies. The effect of ageing in CLSM projected
length is not consistent as bitumen A and C increase in length whereas B and D
decrease upon ageing. The resolution of the two microscopies may have affected the
comparison of the length and wavelength between them. Given this, the bee structure
length size ranges between 2000-20000 nm from AFM analysis whereas the CLSM
images range between 2000-7000 nm.
Of significant importance is to elucidate the effect of ageing on the wavelength of the
waveform that this microstructure presents. AFM demonstrates a wavelength
between 500-1100 nm whereas the CLSM findings show a wavelength between 550
and 3500 nm. Bitumen A and B seem unaffected by ageing concerning the distance of
the valleys and peaks since their wavelength remains the same, whereas bitumen D
reduces its wavelength with ageing. These observations are confirmed by both
techniques.
6. Effect of lab ageing on bitumen microstructure
96
Figure 6-9 Wavelength and projected length [left] and wave height and depth [right] and for all waxy
binders before and after lab ageing
The average peak height in Figure 6-9 [right] exhibits values as captured from AFM
between 15-30 nm whereas CSLM shows relatively higher values 15-50 nm for the
examined binders. Similarly, the valley depth ranges between 10-30 nm for AFM while
CLSM reports a depth between 15-100 nm. The direction of ageing is not clear for these
two wave characteristics, however, each technique shows a consistent decrease or
increase of both values upon ageing.
It is also critical to appreciate the true impact of ageing on the waveform shape for
each binder. Therefore, typical shape probability distribution measures were utilised,
namely kurtosis, skewness and roughness. Kurtosis originates from the Greek word
‘kyrtos’ which translates as curved, and a value around 3 shows a normal distribution
shape of a wave with a few outliers, whereas negative values indicate a so-called
platykurtic distribution with shorter tails. The latter is the case of the wavy formations
obtained via AFM images as can be seen in Figure 6-10 [left]. On the other hand, CLSM
(Figure 6-10 [right]) shows a relatively normal shape distribution of the waveforms as
the values are around 3. The effect of ageing is not consistent between the waxy
binders for this characteristic. Skewness provides also some insights into the
probability distribution of the wave asymmetry with a value equal to 0 showing a
balanced and symmetric waveform. The results of Figure 6-10 show that the skewness,
as extracted by AFM, presents a slightly positive value, which assumes the tail is on the
left of wave distribution when CLSM fluctuates with regard to this measure. Finally,
the surface roughness was consistently reduced upon ageing for bitumen B and D both
from AFM and CSLM data. It should be noted that although kurtosis and skewness are
unitless the roughness estimations are given in nm and are merged in the same Figure
6-10.
97
Figure 6-10 Shape probabilistic values with lab ageing for all the waxy binders via AFM [left] and CLSM
[right]
Bee detection using deep learning and 2-D FFT
The two microscopic tools proved robust enough to analyse the characteristics of the
bee structures before and after lab ageing using commercial software packages for
image processing. However, their convergence is worth to be discussed. First of all,
AFM can capture clearly in phase contrast the different phases, whereas CLSM cannot
identify phases other than the catanaphase. It was found that both techniques agree
with regard to the bee coverage and show a sufficient relationship for the bee size.
Through waveform characteristics, it becomes possible to highlight the differences
between AFM and CLSM for the average peak height, wavelength and length with
CLSM reporting in general higher values. From the shape probabilistic measures, clear
differences can be seen for kurtosis and surface roughness. The observed differences
can originate either from the microscopic technique itself or they can depend on the
image processing.
Therefore, Chapter 6 proposes additionally a more sophisticated deep learning
algorithm to detect the location of the bee structures in order to investigate their
surface roughness. It also suggests a new method based on two-dimensional Fast
Fourier Transform (2-D FFT) [192] to identify in the most precise way the wavy
characteristics of the bee structures. This additional analysis of Chapter 6 offers to the
existing literature a versatile image detection tool and serves towards overcoming a
longstanding challenge with respect to the true impact of ageing on the bee structures.
Deep learning is a popular computational method that enables computers to learn
from experience and is a subset of machine learning that uses a hierarchical artificial
neural network to carry out the desired tasks. This powerful technique has
dramatically improved the state-of-the-art in many domains such as visual object
recognition and object detection [193]. In this Chapter, deep learning was used for DIC
CLSM images of a reference waxy bitumen to detect the bee structures on the surface
of the bitumen microstructure before and after long-term lab ageing. A YOLOv3-based
6. Effect of lab ageing on bitumen microstructure
98
algorithm developed in a Matlab environment is trained and used [194]. Contrary to
many other object detection methods, YOLOv3 applies a single neural network to the
full image and by dividing the image into multiple regions it predicts bounding boxes
and their probabilities for each region. These bounding boxes are weighted by the
predicted probabilities.
As presented in Figure 6-11, the proposed deep learning network structure in this
Chapter consists of the feature extraction network in SqueezeNet, followed by two
detection heads at the end, with the second detection head able to detect smaller bee
structures. The network input size is selected to be (277, 277, 3), and all the DIC CLSM
images and the bounding boxes are resized to match the input size. Next, with the help
of the estimateAnchorBoxes function of Matlab, the number of anchor boxes is
selected equal to six with the three larger ones used in the first detection head and the
smaller ones assigned to the second detection head. Anchor boxes are a set of
predefined bounding boxes of a certain height and width that can help improve the
speed and efficiency of the detection portion of the network. Finally, the system is
trained on GPU of a laptop with Intel Core(TM) i7-9850H with a clock speed of
2.60 GHz, 32.0 GB RAM, and a GPU of NVIDIA GeForce MX150.
Figure 6-11 Flowchart of the deep learning network used for the bee structures detection
Next, the bee structures on 21 DIC CLSM images are labelled using the Image Labeler
application in Matlab. Afterwards, using a data augmentation procedure including
random rotation, up to 10% scaling of the images and colour jitter (with no saturation),
the number of labelled structures is increased. This allows to increase the variety of
the training data, which can improve the accuracy of the trained network. The trained
algorithm is finally run for all the images captured from the bitumen surfaces,
detecting eventually the present bee structures. After bee detection, the number and
the area of the bounding boxes in each image can be calculated and compared in
different ageing states. For the reference waxy bitumen used for this detection, the
trends with ageing were the same as with commercial image processing.
Output
Input
Feature extraction network
(SqueezeNet)
Detection heads
99
Certain bee characteristics, such as roughness, can be calculated by the proposed
deep-learning bee detection. However, another bee characteristic, prone to human
error with commercial image processing, is the wavelength. To overcome this obstacle
and automate this process it is necessary to develop a method that can estimate the
orientation of each bee (angle of the axis of the bee with respect to the x-axis) in an
image. This is done by developing an algorithm based on 2-D FFT, which is an efficient
and popular method to remove periodic noise from digital images [195]. More
specifically, 2-D FFT in digital image processing is a tool allowing to decompose an
image into its sine and cosine components. The output of this transformation
represents the image in the frequency domain. After applying a 2-D FFT on an image
and plotting the magnitude of the output in the logarithmic scale, two spots with
higher values along the direction of the bee structure were visible. All the main
frequency contributors to this image in all directions are found by calculating the
radially averaged power spectrum. However, it is not possible to obtain the main
frequency contributing to the bee pattern in the desired direction. Therefore, it was
required to find the direction of the bee in a DIC CLSM image. To find this direction,
the image is rotated from 0 to 180 degrees and the averages of the spectrum of the
middle 22 pixels (selected based on the range of the bee dimensions) on the x-axis are
calculated for each rotation angle. An example of it is depicted in Figure 6-12. Hence,
the maximum value shows the direction of the bee. After the bee is rotated, the values
along the axis of each bee are selected. Then, the spectrum of this height data is
calculated using a one-dimensional Fast Fourier Transform (1-D FFT). The maximum
value in the expected region provides the dominant frequency and, consequently, the
wavelength of the bee.
Using the developed methodology, the wavelength of all detected bee structures can
be more accurately estimated showing now a more consistent increasing trend with
ageing (from 0.50 μm in the unaged state to 0.55 μm after PAV) for the reference waxy
0 50 100 150 200
Angle [deg]
0.8
1
1.2
1.4
1.6
Figure 6-12 CLSM DIC in frequency domain (magnitude of the 2D FFT) [left] and average of the middle 22
pixel values on the x
-axis of the spectrum image rotated between 0 to 180 degrees [left]
6. Effect of lab ageing on bitumen microstructure
100
bitumen, whereas the values with commercial image processing were rather
overestimated (see Figure 6-10).
Highlights of the chapter
The effect of ageing on the bee structures of waxy binders is systematically
investigated using commercial image processing techniques for AFM and
CLSM.
Lab ageing results consistently in a decrease in the number of bees per μm2
and the bee coverage percentage.
AFM and CLSM indicate an increased size of the bees with lab ageing.
Assumptions for the change in the bee properties can be attributed to
additional wax crystallisation due to ageing or/and the compatibility of the
aged bitumen matrix with the wax.
A systematic waveform evaluation for the peak height, the valley depth, the
projected length of the bee and the wavelength is provided, while there is no
clear tendency that can be found with lab ageing.
Basic probabilistic measures such as kurtosis, skewness and surface roughness
of the bee waveform are provided, whereas the effect of ageing on them is
not clear.
A comparison of the AFM and CLSM highlights some differences in certain bee
characteristics i.e. wavelength and kurtosis.
Advanced image processing techniques, based on a deep learning network and
2-D FFT, are developed in this Chapter to more accurately detect and analyse
the bee structures in bitumen.
*This chapter is redrafted from: G. Pipintakos et al., Do chemistry and rheology follow the same laboratory
ageing trends in bitumen?, Materials and structures (2022). https://doi.org/10.1617/s11527-022-01986-w
7
7 CHEMISTRY AND RHEOLOGY
OF BITUMEN WITH AGEING
Summary*
The ageing of bitumen has received great attention both from a chemical and
rheological perspective due to its direct impact on asphalt performance. However,
open questions with respect to the convergence of the synchronous ageing changes in
rheology and chemistry of bitumen still exist. Chapter 7 addresses these alterations of
chemistry and rheology and attempts to propose a phenomenological link fed by
fundamental chemical information. To that end, three binders of different type were
used in four different laboratory ageing states. A number of spectroscopic techniques
and rheological testing were employed to derive corresponding chemical and
rheological parameters. In parallel, various statistical methods (Bivariate analysis,
Wilcoxon test, Factor analysis) assisted in identifying the relationships among the
chemo-rheological parameters and simplifying the number of variables. The results of
this Chapter demonstrate clearly that chemistry and rheology are following similar
changes when considering laboratory ageing based on the fast-rate phase of a dual
oxidation scheme and short-term lab ageing. Finally, this work manages to establish
an exclusive linking framework, by means of a training dataset, for a number of newly-
introduced rheological parameters. All in all, the results of this Chapter might be
particularly interesting for future interventions in the chemical composition of
bitumen, considering its effect on performance.
7. Chemistry and rheology of bitumen with ageing
102
Objectives
Alternative lab ageing simulations in thin bitumen films under constant conditions of
temperature and pressure at different time intervals offer the advantage of oxidation
kinetics and have been proven appropriate to track the time evolution of ageing
chemistry and the transition point of the rate-determining phases [30,196]. For this
transition point (completion of the fast- and initiation of the slow-rate phase) in
kinetics, it is still uncertain whether also chemistry would converge if similar
rheological performance with standardised lab protocols can be obtained. Hence, the
aim of the current Chapter is twofold.
On one hand, Chapter 7 addresses the open question of whether chemistry and
rheology follow the same laboratory ageing trends independent of the conditions of
the laboratory ageing protocol which is used and, on the other hand, aims to provide
a statistical linking framework for advanced rheological parameters using fundamental
chemistry of bitumen as input.
The chemo-rheological analyses and statistical methods used for the binders of
Chapters 3 and 4 alongside the objectives of Chapter 7 are schematically given in Figure
7-1.
Figure 7-1 Flowchart of the experimental part, statistical analysis and objectives of Chapter 7
103
Materials and Methods
The same three bituminous binders A, B and C of Chapters 3 and 4, differing in crude
source, distillation and penetration grade were examined to compare their changes in
rheology and chemistry during ageing. The reader is referred to Table 3-1 for the basic
properties of these binders. Four different laboratory ageing states (unaged state
(fresh), RTFO, RTFO+PAV and one time interval of kinetics) were utilised with the
primary goal to reveal possible simulation-related differences in bitumen’s chemistry
and rheology. More specifically, in the initial investigation of Chapter 3, the binders
were aged with a M-TFOT up to 56 days at different time intervals and oxidation
kinetics were studied and revealed that the completion of the fast-rate oxidation
phase and the initiation of the slow-rate oxidation phase for all the three binders was
found to be around 8 days. This time interval was additionally used to evaluate the
chemical and rheological level of this transition point between the two rate-
determining phases with standard lab protocols. In all the subsequent analyses the
binder designation is followed by the laboratory ageing simulation used whereas all
the parameter values are the average of two replicates for the rheological and three
replicates for the chemical values.
Fundamental information derived in Chapters 3 and 4 from advanced spectroscopy is
used as chemical input in a statistical framework. The COI, SOI, and ARI indices were
eventually utilised for FTIR, the carbon-cased organic radicals CBOR and vanadyl
centres (VO2+) for EPR, the chemical distribution of methyl (MetP), methylene (MtlP),
α-alkyl (AlkP), olefinic (OleP) and aromatic (AroP) protons for 1H-NMR, as well as
representative positive and negative ions for each category of (OH)x, SOx-containing
and PAH TOF-SIMS ion fragments, (OH-, C2OH-, C4OH-), (CHSO-, C2H3SO-), and (C9H7+,
C10H8+, C13H9+) respectively.
The Chapter is additionally designed with the optimum goal to account for the DSR
extracted rheological parameters of R, ΔΤc, Glover-Rowe (G-R), ωc, Gc, Tc and viscosity.
After deriving all the necessary chemical and rheological parameters, the Wilcoxon
exact test and the Pearson’s r of a bivariate analysis are used as metrics for the
equivalence of the properties between standardised ageing and the reaction-rate
transition point in kinetics. Employing multivariate statistics by means of sample
correlation matrix and factor analysis, the number of original chemical and rheological
variables is reduced and further implemented into fitting equations of rheology as a
training dataset, using a so-called factor analysis regression. This work finally stresses
the goodness of the existing fit between the latent variables of chemistry and the
advanced parameters of rheology. In order to accommodate for the potential reader
7. Chemistry and rheology of bitumen with ageing
104
the use of the utilised statistical methods, their theoretical background and application
of them are briefly revisited in the following subsections.
Pearson’s r and Sample Correlation Matrix
In multivariate data analysis, the association or relationship between the different
variables is key. There are different measures to express the linear relationship
between two variables. Unlike the sample covariance, the Pearson’s product-moment
correlation coefficient (r) is scale-invariant. That is, its value does not depend on the
units or dimensions used. Therefore, the relationships between the different chemical
and different rheological parameters between the ageing states were first inspected
by the Pearson’s r, using the statistical software JMP Pro 16. From a mathematical
point of view, Pearson’s rij between two variables Xi and Xj is defined according to
Equation 7-1.
ij
ij ji
ii jj
s
rr ss
= =
Equation 7-1
where sij is the sample covariance between Xi and Xj. If 0 << rij < 1 then a positive linear
relationship holds between the two variables, while rij ~ 0 indicates no linear
relationship. On the other extrema, for values -1 < rij << 0 the variables correlate
linearly with a negative relationship. To judge additionally if the assumed correlation
is strong or weak the p-value is employed in sample statistics. It shows, in simple
words, the probability that a difference has been generated by a random chance. A
significance level is usually set (< 0.05) indicating the percentage of risk to reject a
certain hypothesis, known as the null hypothesis. For p-values greater than the
significance level, they report that the null hypothesis is most likely to occur. For
Pearson’s r and the accompanying correlation between the variables Xi and Xj the p <
0.05 witnesses that the null hypothesis (r = 0) in this case is not true, and therefore
gives evidence for a certain relationship. In the second phase of the analysis, the
chemical and rheological variables were used to construct sample correlation matrices
containing as elements the Pearson’s r between the different variables.
Non-parametric Wilcoxon Rank Sum exact test
Non-parametric statistical hypothesis tests are commonly used for small samples,
typically with a number of observations n < 15, and especially when assumptions about
the underlying data distribution cannot be done. The Wilcoxon exact test is used to
test the null hypothesis for each pair of the two levels of a variable X. It is the non-
105
parametric version of the equivalent parametric two-sample t-test. Also in this case
the p-value is the key factor for rejecting or not the null hypothesis. Using a significance
level of 0.05 for a two-tailed test the null hypothesis is rejected when p < 0.05 which
can be interpreted as the fact that the medians of the two populations differ
significantly, otherwise for p > 0.05 the null hypothesis cannot be rejected and
therefore provide evidence that the medians tend to be equal. The Wilcoxon exact test
was used in this work to examine together with the Pearson’s r the two pairs (RTFO
and M-TFOT) of each rheological and chemical index.
Multivariate Factor Analysis
Exploratory factor analysis is a common method for re-expressing multivariate data.
The main assumption in this analysis is that the observed variance in the multivariate
data is attributable to a relatively small number of common factors, the so-called latent
factors. Prior to this analysis some conditions should be met and scrutinised, such as
the existence of one or more groups of variables that are highly correlated as
witnessed by a correlation matrix. In general, the mathematical framework of this
analysis is based on the idea that for p variables X = (Χ1, Χ2, ..., Χp) certain correlations
exist and depend on m<p common factors F = (F1, F2, … Fm) as well as on the
corresponding p error terms U. Each of the p variables is linearly dependent to the m
factors in the common factor analysis which in matrix notation can be written in
Equation 7-2.
X LF U= +
Equation 7-2
where L is the loading matrix containing the factor loadings lij which vary between -1
and +1 and can be interpreted as correlations between the original variables and the
factors. Of particular interest is that the squared factor loading lij2 is the percentage of
variance in variable Xi explained by the factor Fj, while the sum of the different squared
factor loadings for a variable Xi is known as communality. In other words, communality
shows the proportion of the variation in Xi accounted by all the common factors.
Additionally, in factor analysis, the given m number of factors is, initially, based on the
researcher’s choice and later is verified by the cumulative percentage of the total
variance explained by the selected number of factors. As a rule of thumb, the number
of factors is chosen in such a way that can explain more than 70% of the total variation
and has also a meaningful interpretation [93]. Simplifying the interpretation of the
factors is also necessary as obtained by the factoring method i.e. principal axis,
because the obtained factors are not unique. One possible simplification method is by
7. Chemistry and rheology of bitumen with ageing
106
rotating the factors i.e. by the Varimax rotation method, which is used in the statistical
software JMP Pro 16.
Factor analysis regression
Frequently the factor analysis is not an end step but rather an intermediate step to
further data analysis. One useful application of the factors is to use them as
independent variables in multiple linear regression, which is called factor analysis
regression. The goal is to model the linear relationship between the independent
factors derived from chemistry and a response variable, in this case, a rheological
value. Hence, the location of each of the original chemical variables in the reduced
factor space is needed, which is alternatively called the factor scores for the latent
factors. JMP Pro 16 allows exporting the factor scores which are used as independent
variables in a factor analysis regression described in Equation 7-3.
11 2 2i o i i m im
Y b bF bF b F
ε
= +⋅ + ⋅ + ⋅ +
Equation 7-3
where Yi is the dependent rheological variable i, Fij is the independent chemical factor
score j for the prediction of the dependent variable i, b0 is the y-intercept (constant
term), bj are the slope coefficients for each independent chemical factor score j and ε
is the residual (error term). Some assumptions such as the linearity, the independence
of observations, the normality of the residuals and the homoscedasticity should also
hold here for a factor analysis regression. Finally, a statistical metric to explain the
degree of variation of the outcome based on the variation of the independent variables
is the coefficient of determination (R2) with values close to 1 indicating a small
prediction error.
The effect of lab ageing protocol
The rheological behaviour of the binders is depicted by means of master curves in
Figure 7-2. A master curve is a useful illustration providing an overview of the effect of
the different ageing states on a broader frequency/temperature domain. Consistent
with previous works this study affirms that ageing increases the complex modulus and
decreases the phase angle of bitumen. This can be seen mainly in the low-
frequency/high-temperature domain of the master curves and the differences
because of ageing become more apparent for bitumen C in a wider frequency range.
Besides this, binder A as expected based on its bitumen grade presents higher values
of complex modulus in a wider low-temperature/high-frequency domain compared to
107
binders B and C. In general, it is believed that ageing at low temperatures, like the one
used in M-TFOT, affects to a greater level the low-temperature properties compared
to ageing performed at higher temperatures, like in RTFO. Moreover, high-
temperature ageing changes mainly the shape parameters of a master curve.
Counterbalancing these changes due to different ageing temperatures can be possibly
achieved with an appropriate selection of the ageing duration.
Of particular importance is the convergence of the M-TFOT and RTFO master curves
for all the binders. It has been shown that the transition point between the two rate-
determining oxidation phases under the specific conditions of M-TFOT occurs around
8 days. The findings here support an association between this ageing state (8 days M-
TFOT at 50 °C) and the standardised laboratory simulation for short-term ageing (RTFO
at 163 °C), while the binders exhibited their most extreme values after PAV ageing. The
comparability of the two ageing states is further investigated using statistical analysis.
7. Chemistry and rheology of bitumen with ageing
108
Figure 7-2 CA master curves shifted to a reference temperature of 15 °C for all the binders and ageing
states
To examine the initial observation of the rheological convergence of M-TFOT and RTFO
as well as the corresponding evolution of the chemical parameters, bivariate analysis
was first used. This analysis will allow concluding whether binders that differ in terms
of type follow the same evolution separately for their rheological and their chemical
changes, irrespective of the ageing simulation. The two ageing states are treated as a
dependent and independent variable with two sets of values (domains), namely
chemical and rheological, including all three binders. The chemical parameters include
all the parameters investigated and explained in the previous section, whereas the
rheological parameters extracted from the master curves are used as another domain.
Differentiation between the binders from a chemical perspective can be mainly seen
in their intensity as depicted in Figure 7-3 [top], something which is particularly
pronounced for the CBOR and ARI of bitumen C. Moreover, since the values differ in
terms of magnitude two bivariate plots are performed for each domain to exclude the
effect of magnitude errors. Figure 7-3 shows these bivariate plots per domain as well
as the bivariate fits that are performed. This analysis can be helpful in testing simple
convergence between the two states. As can be seen, the parameters of the two states
seem to fit well in a linear relationship between them. It can also be seen that with
regard to the radical formation of binder C the convergence of the two ageing states
is lower than the other two binders. The reason for this observation may be found in
the considerably different type of this binder considering its origin and refinery
process.
109
Figure 7-3 Bivariate plots of all chemical [top] and rheological parameters [bottom] between M-TFOT
and RTFO
To further clarify the relationship between M-TFOT and RTFO the statistics of the
bivariate analysis are provided in Table 7-1. The number of observations is based on
the average values for each chemical or rheological parameter, thus, it is much higher
in reality. The parameter estimation for coefficient b of the bivariate fits presents
values close to the unit which means that they correlate in a fashion y=x with the
intercept a interpreted here as the error between the values for each state. This holds
both for the rheological and the chemical parameters, as well as for low and high
magnitude values. In addition, the Pearson’s r shows a strong positive correlation of
this relationship with values close to 1. The significance of this positive correlation
between the two states for both sets of parameters is further justified by the p-values,
which are much lower than the threshold level of 0.05. The coefficient of
determination (R2 ~ 1) for the suggested fits is adequate to support the fitting choice
and establish a reliable relationship between the two states. That is, not only the initial
observation of the convergence of the master curves is valid but also the bivariate
analysis for the rheological values confirms this claim. In parallel, the chemical
parameters for the same ageing states show a significant level of convergence which
implies that chemistry can follow the rheological ageing trends in bitumen for different
ageing simulations and vice versa.
7. Chemistry and rheology of bitumen with ageing
110
Table 7-1 Summary of bivariate statistics between M-TFOT and RTFO
Parameters Fit
Magnitude of
parameters
Parameter
estimates
Pearson’s
r
Significance
(p-value)
R2
Observations
N
a
b
Chemical
RTFO= a +
b·M-TFOT Low 3.43E-6 1.009 0.933 <0.0001 0.979 33
High -5.38E+17 1.079 0.932 0.0068 0.868 6
Rheological
RTFO= a +
b·M-TFOT Low -0.227 0.912 0.997 <0.0001 0.994 13
High
-87610
1.037
0.999
<0.0001
0.997
5
To further ensure that each parameter in the previous domains for all the binders is at
least comparable in terms of distribution of values between the two ageing states, the
Wilcoxon Rank Sum exact test was performed. The selection of this test is explained
by the small number of observations for each numerical parameter. Table 7-2 shows
that the p-values for all the parameters suggest in favour of the null hypothesis (p >
0.05), therefore pointing to similar centres of location between the two levels of the
grouping variable of ageing state, in this case, M-TFOT and RTFO. The lowest p-values
which can be interpreted here as having the highest probability to show different
distribution between the two states were found for the (OH)x and SOx-containing ion
fragments as well as the SOI. This test strengthens the existence of convergence
between the two ageing states not only for their domains but also for the individual
parameters.
Together the two statistical approaches show that similar levels of rheological and
chemical parameters can be succeeded between the transition point (8 days) of the
oxidation kinetics with M-TFOT and routine lab simulation of short-term ageing. It is
fair enough to consider that although the ageing protocol differs i.e. RTFO incorporates
airflow and rotation contrary to M-TFOT, once the same ageing level has been reached
in terms of rheological performance, the chemistry of bitumen can follow the same
trend and vice-versa, by means of a balance of ageing temperature and time.
Another point, worth to be highlighted here, is the fact that although the temperature
difference in the two protocols differs significantly when examining more meaningful
derived rheological parameters, it seems that the change in duration can well
compensate for the temperature difference.
In the following statistical analysis, to avoid overloading the analysis with similar values
of RTFO and M-TFOT which converge to similar ageing severity both for rheology and
chemistry, only the chemorheological values of RTFO are utilised.
111
Table 7-2 Output of Wilcoxon tests for each parameter of M-TFOT and RTFO
Statistical method
Grouping variable
with 2 levels
Parameter
Analysis of numerical
parameter
p-value
Wilcoxon Rank Sum exact test
RTFO & M-TFOT
Chemical
SOI
0.4
COI
1.0
ARI
1.0
CBOR
0.7
VO2+
1.0
OH-
0.4
C2OH-
0.1
C4OH-
0.1
CHSO-
0.4
C2H3SO-
0.4
C9H7+
0.7
C10H8+
0.7
C13H9+
1.0
Rheological
G
c
1.0
Tc
0.7
ωc
1.0
R
1.0
ΔΤc
0.7
G-R
1.0
Link between chemistry and rheology by ageing state
It remains to be seen if the fundamental information can be used to estimate certain
rheological values and their change with ageing. Three ageing states (unaged, RTFO
and PAV) for all different binder types were chosen within the framework of
multivariate statistics in this study. Preliminary investigation of relationships is a
requirement so that one is able to apply more sophisticated exploratory analysis i.e.
factor analysis.
Hence, the parameters are first grouped into chemical and rheological. Correlation
matrices are formed and shown in Table 7-3 and Table 7-4. Typically a correlation
matrix is a symmetric matrix including as elements the exact correlations, as realised
by the Pearson’s r, between two parameters. This is particularly useful to decide on a
specific threshold under which correlations are important. For consistency, only one
part of the symmetric correlation matrices is given as well as only the strong
correlations (|r| > 0.7) are presented with increasing intensity of redish colour for
values of |r| between 0.7-0.8, 0.8-0.9 and 0.9-1.0 respectively. It is demonstrated
numerically that several correlations exist between the different chemical parameters,
a fact that suggests that multivariate exploratory factor analysis can be performed for
this category.
7. Chemistry and rheology of bitumen with ageing
112
Possible explanations for the different chemical correlations can be found based on
the chemical family that the compounds belong to or on their role in the ageing
process. To name some of them, the SOI and CBOR have been used to distinguish the
two rate-determining changes and can explain well the transition point of a fast-rate
followed by a slow-rate oxidation phase. PAH of C13H9+, C10H8+ and C9H7+ seem to
correlate well with ARI of the same chemical family. The SOx-containing fragments
show also a strong correlation between them. Possible explanations behind the strong
positive or negative correlations of specific protons with OHx-containing fragments
may be found in the change of molecular associations which needs to be revealed in
further studies. In general, it is believed that the individual contribution of each
chemical change could have a significant effect on rheological performance although
the relative increase may not be the same between the ageing states for rheology and
chemistry.
Table 7-3 Correlation matrix of chemical parameters
SOI
COI
ARI
CBOR
VO2+
OH-
C2OH-
C4HO-
CHSO
-
C2H3SO-
C9H7+
C10H8+
C13H9+
AlkP
MetP
OleP
MtlP
AroP
SOI
1.00
COI
1.00
ARI
1.00
CBOR
0.75
1.00
VO2+
1.00
OH-
0.87
1.00
C2OH-
-0.73
0.70
1.00
C4HO-
-0.77
0.81
0.96
1.00
CHSO
-
1.00
113
C2H3SO-
0.99
1.00
C9H7+
-0.77
-0.79
0.79
-0.72
0.85
0.94
1.00
C10H8+
-0.82
-0.76
0.81
0.86
0.95
0.99
1.00
C13H9+
-0.89
0.78
0.88
0.94
0.95
0.98
1.00
AlkP
0.72
-0.75
1.00
MetP
-0.93
0.91
0.93
0.86
0.90
0.96
-0.80
1.00
OleP
-0.74
0.79
-0.76
1.00
MtlP
0.82
0.83
1.00
AroP
0.78
-0.83
-0.72
-0.81
-0.79
-0.86
-0.91
-0.88
1.00
Next, to minimise the number of independent variables, i.e. chemical parameters, in
order to implement them in a fitting scheme of dependent ones i.e. rheological,
possible relationships of the dependent variables should also be identified. Previous
scholars have already reported a variety of relationships between these rheological
parameters. Taking into account these well-documented relationships, the correlation
matrix provided in Table 7-4 and the meaningful interpretation of the parameters
explained previously in Chapter 2, the governing and most important rheological
parameters were selected. Therefore, the crossover value Gc together with the R, ΔΤc
and η135 °C are chosen that can capture the viscoelastic response, the relaxation,
ageing-induced cracking and workability issues that occur due to ageing.
Table 7-4 Correlation matrix of rheological parameters
G
c
T
c
ω
c
R
ΔΤ
c
G-R
η
135οC
G
c
1.00
T
c
-0.74
1.00
ω
c
0.78
1.00
R
-0.87
0.84
1.00
7. Chemistry and rheology of bitumen with ageing
114
ΔΤ
c
0.88
-0.78
-0.90
1.00
G-R
1.00
η
135οC
0.95
1.00
Given that the conditions to perform a factor analysis are supported by the correlation
matrix of the chemical parameters, these parameters are further incorporated in the
statistical software JMP Pro 16. After investigating the number of factors that can
explain the total variance of the parameters, three factors proved satisfactorily to
explain the total variance of the chemical parameters. The cumulative percentage of
the variance explained by the three factors is 87.26% indicating that the three factors
can adequately describe the total variance according to a threshold level of 70%. The
rotated factor loading matrix and the communality estimates are also considered
crucial to better understand and interpret the formulation of the factors which are
able to reduce the number of the original chemical parameters. Table 7-5 clearly
indicates the borders of the three factors that are highlighted for the parameters with
a high factor load. More specifically, the rotated factor loadings (Varimax rotation
method) in Table 7-5 support that Factor 1 loads heavily on C13H9+, COI, MetP, C10H8+,
C4HO-, C9H7+, VO2+, C2OH-, ARI, AroP, OleP and AlkP and can account for 43.84% of the
total variance. This factor explains primarily the formation and consumption of the
ageing products. C13H9+, COI, MetP, C10H8+, C4HO-, C9H7+, VO2+, C2OH have a positive
effect on this Factor 1 while the proton regions have a negative effect. On the other
hand Factor 2 loads on SOI, MtlP, CBOR, OH- which are factors that have been shown
to have an association with the transition between the fast and slow-rate phase of an
oxidation scheme [30,97]. This Factor accounts for 24.79% of the total variance while
Factor 3 explains 18.64% of the total variance. Factor 3 loads heavily on the sulfoxide-
related products of CHSO- and C2H3SO-. Furthermore, the total proportion of the
variation per chemical variable by all the factors can be understood by the
communality of each variable. The total communality estimates (Table 7-5) are
acceptable ( > 0.7) for all the parameters accounted for by the three factors. The lowest
communality can be found for the vanadyl species VO2+, where a possible explanation
for this observation can be the differences that have been observed for binders of
different crude sources for this value; a fact that the multivariate statistics are not able
to capture adequately.
Table 7-5 Rotated factor loading and final communality of factor analysis for the chemical parameters
Chemical
Parameter
Rotated factor loading
Final
communality
estimates
Factor 1
Factor 2
Factor 3
C
13
H
9
+
0.884
-0.407
0.190
0.982
115
COI
0.854
0.162
0.023
0.756
MetP
0.846
-0.318
0.384
0.965
C
10
H8+
0.843
-0.512
0.096
0.981
C4HO-
0.782
-0.478
0.349
0.961
C
9
H
7
+
0.772
-0.594
0.106
0.959
VO2+
0.740
-0.320
-0.061
0.653
C
2
OH-
0.709
-0.367
0.546
0.936
SOI
0.041
0.901
-0.252
0.877
MtlP
-0.120
0.883
-0.259
0.861
CBOR
-0.405
0.872
0.004
0.924
OH-
-0.399
0.723
0.159
0.707
ARI
-0.761
0.331
-0.302
0.779
AroP
-0.932
0.149
-0.033
0.891
OleP
-0.750
-0.019
-0.511
0.684
AlkP
-0.788
-0.186
-0.647
0.926
CHSO-
0.114
-0.146
0.956
0.948
C
2
H
3
SO-
0.055
-0.167
0.941
0.917
Up to this point, it was possible to minimise the number of chemical parameters for all
the binders and ageing states in three factors by means of factor analysis. The factor
scores of this analysis were extracted and successively implemented in a factor analysis
regression, where the independent factors are the factors scores for each ageing state
of each bitumen and the dependent factors are the rheological parameters that have
been chosen from the correlation matrix (Table 7-4). The coefficients of the factor
analysis regression are shown in Table 7-6 together with the corresponding coefficient
of determination.
The advantage of these linking equations is among others that the number of
independent variables is relatively small and the goodness of fit is evidenced to be high
enough by the R2 values for the Gc, R and ΔΤc rheological parameters. The viscosity fit
is characterised as weak. The linking framework seems to differentiate the rheological
properties derived in different temperature ranges, where the higher testing
temperature results in a weaker fitting equation. It is also possible that the
incorporation of a variety of factors in the binder’s resistance to flow may overload the
fitting equation since this rheological parameter is rather simple compared to the
more advanced parameters derived from the master curves and can be fitted by fewer
chemical parameters i.e. carbonyls [6,90] . The estimation of the parameters of this
training dataset also shows the strength of the chemical parameters per binder and
their effect on rheology.
7. Chemistry and rheology of bitumen with ageing
116
Table 7-6 Factor regression analysis between chemical and rheological parameters
Rheological
parameter
Factor analysis
regression
Parameter estimates
R2
a
b
c
d
Gc
y = a + b·FS1 +
c·FS2 + d·FS3
15919191
2671103.7
-8653879
2144094.8
0.741
R
1.981
-0.055
0.335
-0.137
0.825
ΔΤ
c
-1.763
1.738
-3.021
1.391
0.914
η
135
ο
C
1467.022
738.742
235.789
-447.663
0.288
Finally, the actual and predicted (via the chemical analysis) rheological parameters are
plotted in Figure 7-4. The graphs highlight the convergence of the actual and predicted
values for three of the four parameters. Moreover, since correlations exist between
the crossover values one is able to estimate the rest of the rheological values. The fact
that the ageing effect can be taken into account in these values shows that
fundamental chemistry plays a significant role in bitumen’s performance. Looking back
to the fundamentals of bitumen chemistry and its evolution with ageing, as well as the
effect on the latent factors and the factor analysis regression will allow for the
understanding of the chemical indices that deserve special attention. Moreover, the
necessity to further support these observations, so that a clear quantification of the
crucial chemical components can be done, should be tackled by additional validation
datasets.
Figure 7-4 Statistically predicted and actual rheological values in bivariate plots of the training dataset
117
Highlights of the Chapter
A linking framework of a training dataset between the rheological
performance and bitumen’s fundamental chemistry is established.
A variety of statistical methods are utilised such as bivariate analysis,
correlation matrices, the Wilcoxon exact test, Factor analysis and regression
for the different variables.
This Chapter explores the convergence of bitumen chemistry among different
ageing simulations. In particular, investigation of two short-term ageing
simulations indicates for the examined dataset that besides rheology also
bitumen chemistry can follow similar ageing trends.
The phenomenological framework of this Chapter supports that advanced
rheological values, such as ΔΤc, can be fairly accurately estimated (R2 = 0,91)
only if a number of fundamental chemical parameters is taken into account.
For more basic rheological properties, i.e. viscosity, the number of chemical
parameters needed for an acceptable estimation may be less.
Multivariate statistics show their great potential to understand the association
of the laboratory ageing changes in bitumen’s chemistry and rheology.
7. Chemistry and rheology of bitumen with ageing
118
*This chapter is redrafted from: U. Mühlich et al., Mechanism based diffusion-reaction modelling for
predicting the influence of SARA composition and ageing stage on spurt completion time and diffusivity in
bitumen, Construction and Building Materials (2020).
https://doi.org/10.1016/j.conbuildmat.2020.120592
8
8 THERMODYNAMICS
MODELLING OF AGEING
MECHANISMS IN BITUMEN
Summary*
Reaction-diffusion models derived within the frame of continuum thermodynamics of
irreversible processes (TIP) for reacting mixtures enjoy a number of advantages over
purely phenomenological models. Chapter 8 explores the potential of this approach by
means of a particular model for the fast rate-determining phase of oxidation in
bitumen. Film ageing experiments are simulated considering different bitumen
compositions in terms of SARA fractions. The predicted fast-rate phase completion
times are contrasted with parameters characterising the initial composition and the
results demonstrate the consistency of the model accounting for the fast-rate
oxidative mechanisms of bitumen.
Objectives
In phenomenological modelling efforts, like the ones described in Chapter 7,
interrelationships between the different scales are captured almost exclusively by
purely empirical relations and phenomenological parameters. A concept for modelling
the oxidative ageing of bitumen which rests on TIP is proposed in this Chapter,
adopting a continuum perspective [24,197,198]. This model aims to establish direct
8. Thermodynamics modelling of ageing mechanisms in bitumen
120
links with the underlying molecular structure and the oxidative mechanisms of
bitumen, reviewed experimentally in Chapters 3 and 4, by means of physically
meaningful parameters such as solubility and molar volumes. The model presented in
this Chapter attempts additionally to distinguish between different SARA
compositions. Finally, to judge the capabilities of this model, the predicted oxidation
times are contrasted with experimental results and trends reported in the literature.
Materials and Methods
Two bituminous binders A and B of Chapters 3, 4 and 7 were chosen to track the
progress of oxidation for the sulfoxide evolution, depicted previously in Figure 3-4. As
mentioned in Chapter 3, a M-TFOT for bituminous films of 1 mm thickness, at 50 °C
was used for the ageing kinetics and the chemical fingerprinting of them was recorded
by FTIR at different time intervals.
To design reasonable SARA compositions as input for TIP modelling, the possible
ranges for the individual fractions in the literature were taken into account
[26,52,199]. As such the interval between zero and hundred percent is partitioned into
fragments of five percent. Three groups are defined with saturates content of 5, 10
and 15 %, respectively. Compositions are generated using standard combinatorics,
more specifically by generating combinations with repetitions excluding cases with less
than four individual fractions, i.e. compositions with only saturates and resins or only
saturates, resins and aromatics are excluded. In total, there are 1076 possible
compositions. However, the number reduces to 49 when applying the constraints
given in Table 8-1.
Table 8-1 Data in terms of SARA fractions with known constraints from literature
SARA fraction
Range [wt. %]
Constraints based on literature
Saturates
5-15
Saturates < Aromatics
Aromatics
40-65
Saturates ≤ Resins
Resins
20-45
Aromatics > Asphaltenes
Asphaltenes
5-25
Resins ≥ Asphaltenes
c
Asphaltenes Saturates
IResins Aromatics
+
=+
Equation 8-1
A further reduction is achieved by using the Gaestel index (Ic) in Equation 8-1 [200].
This index has been used in several studies as an indicator of the colloidal instability of
bitumen. It is believed that the increase of asphaltenes (included in the numerator of
this index) can lead to an unstable colloidal system, as this fraction and its
representative molecules are of the highest polarity in bitumen. Previous studies
utilising binders from different crude oils suggested limits for this index, within which
121
the bitumen is considered to perform adequately well [50]. Excluding all SARA
compositions with an Ic, not within 0.1 and 0.6, yields eventually 14 virtual
compositions, presented in Figure 8-1.
Figure 8-1 SARA compositions used in this study, where columns 6 and 12 refer to the compositions of
the real bitumens A and B, respectively, whereas all other compositions are generated artificially
In addition, the SARA composition of the two real binders A and B was determined by
the Thin Layer Chromatography-Flame Ionisation Detector (TLC-FID) method using an
IATROSCAN MK-6s. The fractionation was performed according to IP 469/01 [48]. For
this method, the bitumen sample is diluted in dichloromethane solutions, after which
the solution is applied on a silica-coated quartz rod. Then, different solvents are used
for the development of the bitumen sample. Three rods were prepared for each
bitumen and the rods were passed, afterwards, through the flame ionisation detector
inside the analyser. The chromatograph produced indicated four main peaks
attributed to each of the calculated SARA fractions used for this Chapter.
Mechanism-based reaction-diffusion model
The model proposed in this work considers a simplified version of the fast-rate
oxidation phase and its mechanisms. The corresponding reaction can be simplified as
follows. An oxygen molecule comes close to a saturated ring, situated between two
aromatic rings. Two hydrogen atoms are taken off from the saturated ring, which
converts the latter into an aromatic ring. If sulfur atoms are available, then the oxygen
and the two hydrogen atoms form a water molecule and sulfoxide.
To incorporate this mechanism into a reaction-diffusion model, the following five
species are defined, numbered in the following by α=1,…,5: oxygen (1), water (2),
aromatised compounds (ARA*) with one S=O (3), aromatisable compounds with traces
of sulfur (ARA) (4) and saturates (5). Each ARA member can only participate once in
the following simplified chemical reaction
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
123456* 78910 11 12* 13 14 15 16
Saturates Aromatics Resins Asphaltenes
8. Thermodynamics modelling of ageing mechanisms in bitumen
122
*
22
11 1O ARA ARA H O+→ +
Equation 8-2
ARA denotes an average molecule accounting for the effects of asphaltenes, aromatics
and resins. Saturates are considered separately because they do not participate in the
reaction. For the chemical potentials of the individual species Equation 8-3 is used with
the chemical potential of the pure species being μα0, molar fractions χα, and activity
coefficient γα.
0
ln ln
αα α α
µµ χ γ
=++
Equation 8-3
Regarding the activity coefficient, the most simple available model which accounts for
differences in size between the representative species is chosen [201] in Equation 8-4.
*
2
ln [ ]
v
RT
α
αα
γ δδ
= −
Equation 8-4
The presence of physically meaningful parameters such as molar volumes of pure
species in liquid state vα*, and solubilities δα, allows for establishing direct links
between molecular level and continuum scale. The mean solubility of the mixture,
δ
is given by Equation 8-5
*
1
*
1
N
N
v
v
αα α
α
ββ
β
δχ
δ
χ
=
=
=
∑
∑
Equation 8-5
Due to Equation 8-2, all molar fractions can be expressed in Equation 8-6 based on the
molar fractions of oxygen and water (χ1, χ2) and the initial molar fractions of saturates
χ50.
32
0
5 5 12
0
4 5 12 2
[1 ]
[1 ][1 ]
χχ
χ χ χχ
χ χ χχ χ
=
= −−
=− −− −
Equation 8-6
Assuming that only oxygen diffuses implies that all other species remain on average in
their spatial locations. Therefore, the model reduces to two essential mass balances in
Equation 8-7, which read in terms of molar concentrations
1
1
2
1divcJ
M
c
τ
τ
∂ + = −Λ
∂=Λ
Equation 8-7
123
with molar mass of oxygen, M1, diffusion flux vector
J
and reaction rate density Λ. The
partial derivative with respect to time is indicated by
τ
∂
while div is used for the spatial
divergence operator. The components of
J
with respect to a Cartesian coordinate
system defined by coordinates xj, j = 1, 2, 3 read in terms of molar fractions as follows.
1 1 1, 2 2,
[]
j jj
J Lj j
χχ
=−+
Equation 8-8
where χα,k indicates the gradient of a quantity χα with respect to the spatial coordinate
χk. The terms j1, j2 in Equation 8-8 are given by
15 15
11
1 12 1
12 1 12 2
1,1,
11
jj
χχ
χχ
χχ χ χχ χ
∂Α ∂Α
=++ =++
−− ∂ −− ∂
Equation 8-9
with
* 2* 2
11 5 5
15
[][ ]vv
ART
δδ δδ
−− −
=
Equation 8-10
Concerning the reaction rate density, the methodology proposed in [202] is used,
which relies on representing the reaction rates by polynomials in activities. Hence,
similar to the diffusion part, molecular level and activity models can be linked by means
of activity models with physically meaningful parameters and information about the
SARA fractions. As such, the final result for the reaction rate density is expressed in
* 2* 2
011 44
14 1 5 1 2 2
[][]
[[1 ][1 ] ] ex p vv
ΛΒ
RT RT
δδ δδ
χ χ χχ χ
−−
= − −− − +
Equation 8-11
where B14 is the reaction parameter, R is the ideal gas constant and T is the absolute
temperature. To simplify the discussion, only problems which are captured sufficiently
well by a one-dimensional version of the model are considered in the following. In this
case, there is only one active component of the diffusion flux vector in Equation 8-8
which is simply called J in the following and it reads
11 2 2
[]J Lj j
χχ
′′
=−+
Equation 8-12
where the prime indicates the derivative with respect to the spatial coordinate
denoted by z.
SARA compositions and fast-rate oxidation phase
To estimate the properties of the ARA-species, a pure ARA-mixture with mass fractions
4k
β
is considered, where k =1, 2, 3 refer to asphaltenes, resins and aromatics,
8. Thermodynamics modelling of ageing mechanisms in bitumen
124
respectively. The average molecular weight assigned to an ARA-molecule is computed
by means of the weighted arithmetic mean
0
3
4
444 4 0
15
, where 1
k
kk k
k
w
Mw
ββ
=
=Μ=
−
∑
Equation 8-13
and the molar volume is computed analogously. The ARA-solubility, δ4 , is estimated
by adjusting the concept of mean solubility in Equation 8-5. Hence,
3
*0
44 4
1
43
*0
44
1
kk k
k
mm
m
v
v
δχ
δ
χ
=
=
=∑
∑
Equation 8-14
where the molar fractions χ4k are obtained from initial mass fractions w4k0. The values
of δ4 given in Table 8-2 differ only slightly from the arithmetic mean δ4*=63246
2
/g cms
. In addition, assumptions that the properties of ARA and ARA* are the same
in terms of molecular mass, volume and solubility are made. Literature values
[203,204] used for all the subsequent computations are given in Table 8-3.
Table 8-2 Initial compositions and average parameters of an ARA-molecule used in simulations
Initial composition in mass fractions
Parameters used in simulations
SARA
number
Saturates
w50
Asphaltenes
w410
Resins
w420
Aromatics
w430
χ
5
0
Μ
4
[g/mol]
v
4
*
[cm3/mol]
δ
4
[
2
/g cms
]
1
0.150
0.200
0.200
0.45
0.421
1525
1335
63155
2
0.150
0.150
0.250
0.45
0.386
1318
1170
62807
3
0.150
0.200
0.250
0.40
0.426
1557
1364
62967
4
0.150
0.100
0.350
0.40
0.353
1144
1034
62277
5
0.100
0.150
0.200
0.55
0.271
1239
1101
63008
6
0.082
0.187
0.217
0.52
0.247
1356
1197
63036
7
0.100
0.200
0.250
0.45
0.310
1495
1313
62983
8
0.100
0.250
0.250
0.40
0.341
1721
1497
63137
9
0.100
0.150
0.350
0.40
0.285
1331
1185
62479
10
0.100
0.100
0.400
0.40
0.254
1136
1029
62156
11
0.050
0.200
0.200
0.55
0.167
1411
1240
63165
12
0.052
0.220
0.238
0.49
0.184
1520
1332
63095
13
0.050
0.200
0.300
0.45
0.173
1468
1293
62830
14
0.050
0.250
0.300
0.40
0.193
1682
1468
62974
15
0.050
0.200
0.350
0.40
0.176
1497
1320
62662
16
0.050
0.150
0.400
0.40
0.157
1313
1172
62353
Table 8-3 Parameter values of each species used in simulations
Species
Density
g/cm3
Molecular mass
g/mol
Molar volume
cm3/mol
Solubility
√MPa
2
/g cms
125
Oxygen
1.114
32
28.0
14.0
44272.2
Water
0.988
18
18.2
48.0
151790.4
Asphaltene
1.200
4500
3750.0
20.0
63246.0
Resins
1.054
990
940.0
19.0
60083.7
Aromatics
1.006
440
437.0
21.0
66408.3
Saturates
0.887
370
417.0
16.5
52177.9
Next, in order to estimate the required time for the completion of the fast-rate
oxidation phase (spurt completion time tSCT), the compositions of the two real binders
A and B and their ageing kinetics by means of sulfoxide evolution are contrasted. In
parallel, virtual film ageing experiments are performed by simulating the reaction-
diffusion process using the model presented in the previous section. The oxygen
uptake (CO2) can be computed by integrating the spatial distribution of oxygen
concentration in a bituminous film of thickness h at a given time t based on Equation
8-15.
2
1
0
() ( ,)
h
O
C t c z t dz=
∫
Equation 8-15
Furthermore, the simulated spurt completion time tSCTsim is defined as the time at which
oxygen uptake stops. The simulations require proper estimates regarding the
parameters L and Β14 of Equation 8-8 and Equation 8-11 that control the diffusion
speed and reaction rate respectively, which can be derived from the FTIR kinetics of
Chapter 3 for binders A and B. As such, thin films of h =1 mm are used in M-TFOT and
the oxygen uptake is estimated for similar films via the developed reaction-diffusion
model, whereas for the free oxygen surface the molar concentration value c1s = 8.1E-
07 mol/cm3 from literature, is used [19]. The parameters L and Β14 are eventually
estimated by applying a least square method for the two binders A and B, comparing
their experimental tSCTexp (where the change of the initial rapid increase is followed by
an almost constant slope during the slow-rate phase) and the tSCTsim. This demands that
22
,, ,,
[ ] [ ] min
sim sim sim sim
SCT A SC T A SCT B SCT B
tt tt− +− →
Equation 8-16
The convergence of the experimental and simulated tSCT in order to back-calculate the
required parameters is presented schematically in Figure 8-2.
8. Thermodynamics modelling of ageing mechanisms in bitumen
126
Figure 8-2 Schematic illustration for the determination of the spurt completion time via experiments and
simulations
After providing as input to the model reliable parameters by coupling it with
experimental results, the effect of SARA fractions on the tSCTsim was investigated for a
sensitivity analysis of the compositions of Table 8-2. Figure 8-3 [left] shows the
predicted spurt completion time tSCTsim for all virtual compositions as a function of the
molecular mass of an ARA molecule M4 and the initial weight percentage of saturates
w50. It should be noted that all simulations are performed with the same L and B14.
Hence, pressure, density and volume are kept constant. Therefore, an increase in M4
implies less ARA molecules per reference volume, in other words, less molecules to be
oxidised. As Figure 8-3 [left] illustrates, the model predicts consistently a decrease in
spurt completion time for given w50 and increasing M4. On the other hand, increasing
w50 while keeping M4 constant causes certainly a decrease in the number of ARA
molecules, but simultaneously the number of saturates molecules increases. Due to
the difference between the molecular masses of saturates and ARA, one ARA molecule
corresponds to about four saturates. Since individual mass fractions have to sum up to
one, increasing w50 by adding four saturates, for example, requires to remove only one
ARA molecule from the system. Therefore, diffusion becomes more time-consuming
because an oxygen molecule may need to pass more saturates to come close to an
ARA molecule. Thus, the model predicts an increase in spurt completion time with
increasing w50 which is consistent. Furthermore, Figure 8-3 [left] depicts that the spurt
completion time depends in a non-linear manner on M4 and w50. This is to be expected
because diffusion flux (Equation 8-8) and reaction rate density (Equation 8-11) are
highly nonlinear.
127
Figure 8-3 Spurt completion time as a function of the molecular mass of the ARA molecule for different
initial weight percentages of saturates [right] and asphaltene content [left].
The spurt completion time in Figure 8-3 [right] was found to decrease with increasing
asphaltene content. Assuming the same bitumen in two different ageing states,
according to experimental observations, the asphaltene content is evidenced to be
increased in the more severe ageing state. Thus, it can be suspected that in this ageing
state a sufficient amount of O2 has at least reacted with the hypothesised average ARA
molecule, indicating that the remaining time upon ideal spurt completion time will be
reduced compared to the initial less aged state. This is also adequately captured by the
model in Figure 8-3 [right].
Highlights of the chapter
Contrary to purely phenomenological reaction-diffusion models, models
developed within the TIP framework can account for bitumen’s composition
in terms of SARA fractions.
Hypotheses about oxidation mechanisms at the molecular scale are
implemented in a continuum thermodynamics model in this Chapter by means
of activity models.
The developed mechanism-based model captures expected trends in terms of
the completion time of the fast-rate oxidation phase in bitumen.
Coupling of experiments and modelling allows for the estimation of reliable
parameters that control the reaction rate and diffusion speed.
8. Thermodynamics modelling of ageing mechanisms in bitumen
128
9
9 CONCLUSIONS AND
RECOMMENDATIONS
Conclusions
The conclusions that can be drawn from the current doctoral thesis considering the
initial research objectives are summarised below per thematic section as given in
Figure 1-2.
To revisit briefly these objectives, this thesis explored via a number of spectroscopic,
gravimetric, thermoanalytical, microscopic techniques and rheological
characterisation, the oxidative ageing mechanisms in bitumen. This was investigated
both for ageing kinetics and lab ageing simulations taking into account the origin,
manufacturing and performance differences between different binders and their
fractions. Moreover, this work examined the reason behind the surface microstructure
in bitumen and the effect of oxidative ageing on it. Finally, efforts to link the
fundamental chemistry with the rheological response were presented and a novel
thermodynamics reaction-diffusion model was shown in this dissertation.
Molecular level
In this part, at the molecular level, three binders varying in distillation, origin, wax
content and empirical performance as well as their fractions were used as the
reference material. Specific answers, as extracted by the experimental program (FTIR,
EPR, TOF-SIMS, 1H-NMR, DSC and DVS) of Chapters 3 and 4, are given with regard to
the objectives set initially for this research project.
9. Conclusions and recommendations
130
I. It became possible to support via EPR and FTIR a dual-oxidation scheme that
consisted of two rate-determining phases. The formation of organic carbon-
centred radicals and sulfoxide formation was in line with hypotheses about the
ageing mechanisms of bitumen supporting the existence of a fast and a slow
rate-determining phase.
II. Both ageing kinetics and routine lab ageing simulations of RTFOT and PAV
showed that in bitumen the relative amount of vanadyl-porphyrin metal
species remained constant, a fact that can be used to trace back the origin of
the crude source.
III. Oxygenated products after the completion of the fast-rate phase included
sulfoxide, nitrogen, aliphatic, PAHs and oxygen-containing compounds as
revealed by TOF-SIMS analyses, regardless of the bitumen type. For the latter,
these can be alcohols or carboxylic acids.
IV. The EPR analyses demonstrated that after RTFOT and PAV ageing the organic
carbon-centred radicals show a similar effect as with ageing kinetics, for the
underlying ageing mechanisms. The fast-rate phase governs the formation of
carbon-centred radicals which provokes the continuation of the slow-rate
phase in bitumen.
V. A chemical classification of bitumen via 1H-NMR was shown. Slight changes
were found with lab ageing for the relative proton distribution on bitumen,
irrespective of the bitumen type.
VI. SARA fractionation showed that asphaltenes after PAV are more oxidised and
aromatised than the corresponding aged maltenes as supported by FTIR.
VII. Chemical classification of the polarity-based fractions showed that bitumen
and maltenes change similarly with ageing when it comes to their relative
proton distribution.
VIII. The effect of crystallisable compounds in the bitumen composition was
apparent on the bitumen surface via high-resolution images of TOF-SIMS,
realised as aliphatics. Besides this the effect of bitumen’s source, refinery
process and type were minor when it comes to the ageing mechanisms. The
crystallisable compounds after SARA fractionation are gathered mainly in the
maltenes as investigated with DSC, with this effect being more obvious upon
PAV ageing.
IX. Exploration of different film thicknesses and temperatures allowed the
estimation of activation energies and reaction rates in bitumen. More
131
specifically, the thinner the bitumen film, the more pronounced the reaction
contribution of the coupled reaction-diffusion and thus the higher the reaction
rate.
X. A novel technique, namely DVS, was used in combination with an inert filler to
estimate the coupled reaction-diffusion parameters. Diffusion coefficients for
bitumen and oxygen solubility were able to be back-calculated using a
modified Fickian law and FE simulations.
Microscale
In this second part, consisting of the outcome of Chapters 5 and 6, a waxy and a wax-
free binder were used for supporting one of the hypotheses in the literature, based on
the presence of wax in bitumen, about the mechanisms of the bitumen microstructure
via DSC, WAXD and CLSM. Later, this batch was expanded into four waxy and two wax-
free binders to comprehensively evaluate microscopically (AFM and CLSM) the effect
of ageing on them and give answers to the initial objectives of this dissertation.
I. Considering the wax theory about bitumen’s surface, the crystallisable
compounds if present in bitumen may be the reason for the so-called bee-
structures. This was experimentally investigated following real-time
microstructure evolution with CLSM and utilising thermoanalytical techniques
of DSC and WAXD. Interrelationships of the endothermic and exothermic DSC
transitions, the X-ray diffraction patterns and the bee structures were
identified within the same temperature ranges, a fact that shows, based on a
certain hypothesis (wax theory), that crystallinity may be the reason behind
the bee structures' formation.
II. Only the waxy bitumen presented thermal transitions and crystalline
diffractions, while with WAXD it was shown that the crystalline material is
organised in an orthorhombic unit cell consisting most likely of n-alkanes.
III. The real-time microstructure evolution of the relative bee structure coverage
was investigated with an in-house heating/cooling apparatus and was in good
correspondence with the WAXD patterns and the relative crystallinity. The
cooling/heating rate affected the appearance/disappearance of the bee
structures.
IV. A systematic analysis of the long-term lab ageing with PAV was performed in
this project for four waxy binders with CLSM and AFM. Two image processing
methods of commercial software packages and a deep learning algorithm for
bee detection were proposed. It appeared that the normalised to the unaged
9. Conclusions and recommendations
132
state relative bee area coverage is a suitable metric for the effect of ageing on
the crystallisable compounds as it is scale-free.
V. Consistently, all the considered binders presented an increased size of the bee
structures after PAV, a decreased relative bee coverage and normalised
number of bee structures. Possible reasons for these changes may be found in
the additional crystallisation of the waxes with ageing and their compatibility
with the bitumen matrix.
VI. AFM and CLSM were in good agreement regarding the morphological
characteristics of the bee structures, whereas differences were observed for
some of their waveform characteristics (wavelength) and probabilistic
measures (kurtosis). This was attributed to the working principle of each
technique, with AFM being a near-contact method compared to CLSM.
Phenomenological and mechanism-based modelling
In the last part of this dissertation, in Chapter 7 the results of molecular level were
combined statistically for the three types of binders and were linked with their
rheological performance to shed light on the relationships of different ageing
simulations and the role of chemistry on performance. A thermodynamics model was
also formed in Chapter 8 accounting for the SARA composition via a sensitivity analysis
for sixteen different binders. Both Chapters 7 and 8 managed to meet the objectives
imposed at the beginning of this doctoral project.
I. The severity of the oxygenated products (free radicals, ions, protons and other
chemical compounds) after the completion of the fast rate-determining phase
of oxidation kinetics and RTFOT were compared with bivariate analysis,
correlation matrices and the Wilcoxon exact test. This analysis showed that
similar levels can be achieved between these ageing states.
II. Similarly, advanced rheological properties such as the crossover values and
ΔΤc also converged when chemistry reached similar levels independent of the
ageing simulation used. This enables a direct comparison of the oxidation
phases with lab ageing simulations, in other words, RTFOT can adequately
account for the short-term ageing phase, although different time duration, air
flow and temperatures are used compared to the M-TFOT.
III. A phenomenological framework to link the changes in fundamental chemistry
taking into account the oxidation mechanisms and the rheological response by
means of advanced DSR parameters was selected for a training dataset. It
employed useful multivariate statistical tools such as the Factor analysis and
eliminated the number of variables to only three.
133
IV. The results of this phenomenological model supported that, first, the ageing
effect can be sufficiently incorporated, while secondly, for basic rheological
properties, like viscosity, the model should be simplified in terms of chemistry
input parameters.
V. In order to account for a scheme which dives into the ageing mechanisms and
its SARA properties, a thermodynamics of irreversible processes model was
constructed in this project. The fast rate-determining phase and its
mechanisms were considered in this model and linked the chemical potentials,
the activity coefficients and the solubilities of the SARA fractions.
VI. A sensitivity analysis of the SARA fractions was performed and M-TFOT lab
simulations were combined with the mechanism-based model to back-
calculate realistic parameters to feed the model. This allowed discussing the
bitumen’s composition, accounting simultaneously for the oxidation
mechanisms.
VII. The designed model is able to efficiently predict trends for the completion
time of the fast rate-determining phase. For example, it demonstrated that a
more oxidised bitumen will contain more asphaltenes and thus the time until
the completion of the fast-rate phase will be reduced.
Recommendations
Based on the extensive work conducted around the fundamental chemistry of bitumen
and its microstructure, some potential improvements are identified at the end of this
project. Specific recommendations for ongoing and future studies are given in the
following.
I. The fundamental understanding of the ageing mechanisms in bitumen was
based on a limited number of unmodified and waxy binders of different crude
sources and refinery processes. A wider batch of binders including not only
the oxidative ageing effect on the bitumen matrix but also the polymer
degradation is recommended to be studied by future scholars to unravel the
balance between the two; the ageing and the polymer degradation. The
current dissertation can work as a guideline for newly proposed experimental
techniques used in order to further understand the ageing phenomenon from
a fundamental basis in binders modified with plastics, renewable, rejuvenated
and other sustainable bio-binders.
II. Out of this thesis, it is recommended to use the currently available lab ageing
protocols only for comparative studies between binders and the effect of
9. Conclusions and recommendations
134
certain additives. The effects of reactive oxygen species, humidity, and
ultraviolet radiation are not considered in the current standardised lab
protocols. However, researchers worldwide work independently on these
specific factors and it is recommended to bridge the independent findings for
the development of a more realistic model, closer to field ageing. This model
could contain a more precise conditioning and duration of the slow-rate/long-
term oxidation phase from a fundamental point of view, in order to predict
more accurately the in-situ ageing.
III. A global dataset will be of great importance to be made not only for
researchers but also for authorities and producers when designing new
asphalt mixtures and their components. This will assist in improving the
bitumen composition and prolong eventually the target of each stakeholder,
the lifetime of pavement. Initiating by this, in-situ binders can be extracted
and used as a reference with standard lab simulations. At least the binder
type, the date of paving, the polymer percentage and other information
regarding the bitumen’s composition and rheology should be recorded in a
global database by the contractors so that the role of each can be better
understood and quantified. Given that this information will be made available
the phenomenological model constructed in this dissertation can be validated
and enhanced by such datasets.
IV. The modelling results of this dissertation can be used in ongoing monitoring
programs within and outside the Road Engineering Research Section (like the
Fibre Bragg Gratings project) in order to back-calculate the
chemical/rheological behaviour of bitumen. In addition, to scale up in upper
mortar and mixture levels by including the rest components (filler, sand and
aggregates), it should be first made clear if the examined oxidised binder is
part of the coating of aggregates or of the ‘free’ binder in the mix. This can be
possibly achieved by making use of FTIR microscopy, or following techniques
used in the current dissertation such as the TOF-SIMS. Additionally, a more
sophisticated processing of NMR spectra can be used together with in-situ
EPR monitoring of asphalt.
V. In this project relationships between fundamental chemistry and rheology
were proposed via appropriate modelling. In the literature, empirical limits
for advanced rheological parameters such as the cracking initiation exist i.e.
G-R and ΔΤc. Following the same rationale and the phenomenological
framework proposed herein, limits for the bitumen’s chemistry can be
imposed. This means that future intervention via additives, antioxidants, and
135
rejuvenators can be manufactured with the optimum goal to reach this target.
Therefore, this thesis can be used as a reference and guideline in order to
compare the impact of different additives to each other and the reference
binder. The potential chemical limits will be strengthened when more
validation datasets will become available in the scientific community.
Validation datasets can also point towards the oxygenated products that are
affected primarily by oxidative ageing as expressed by the predictive model
coefficients.
VI. Conflicting recommendations exist until nowadays for the sample preparation
of bitumen for microscopic and spectroscopic investigations. The current
dissertation, after extensive CLSM and AFM microscopic imaging, suggests a
consistent sample preparation of such specimens between different labs in
order to keep the obtained images free from additional ageing and artefacts.
Similar sample preparation can be followed for the different spectroscopic
techniques, without necessarily the need for a heating plate and a flat
specimen. Additionally, at least three samples should be investigated in
different spots of the specimen to avoid bias in the interpretations. The
current dissertation recommends the following step-by-step sample
preparation:
a. The microscopic sample preparation should start with a high-temperature
treatment to create a workable and homogenous material. The
temperature and heating time are chosen based on bitumen type and
ageing state and is often between 110 and 160 °C. Even though higher
temperatures have also been used in the literature, this can cause
additional ageing in the material and is not recommended.
b. A sample holder (microscopic slide) is placed horizontally on a heating
plate set at a temperature 140 - 160 °C. A bitumen drop is placed on the
sample holder and left until a smooth surface (1-2 minutes) with sufficient
thickness is obtained.
c. The sample is left inside a dust-free environment for two hours to cool
down to the ambient temperature. If applicable, the cooling rate is kept
between 2 °C/min and 3.4 °C/min to obtain similar cooling conditions of
bitumen during asphalt mixture compaction.
d. Finally, images can be taken at different time intervals, using a non-contact
microscopic technique, to keep the surface of the samples free of any
artefacts.
9. Conclusions and recommendations
136
VII. The interpretation of the microstructure should be based in the near future
not only on human interpretations but also on image processing of the
obtained images which provides useful information with regard to meaningful
metrics affected by ageing. Following the different image processing methods
of this work, similar methods can be applied to modified binders as well as
rejuvenated ones.
VIII. In this research project, it was not possible to study the effect of nanoscale of
the multiscale ageing phenomenon. This should be taken into account by
future peers in order to link existing theories of bitumen’s structure i.e.
nanoaggregation model with bulk information. WAXD and SAXS are
recommended to be used before and upon ageing for this purpose and can
be modelled by molecular dynamic simulations.
IX. Finally, the thermodynamics model developed in this dissertation in Chapter
8 can be expanded to incorporate both the fast and the slow rate-determining
phase. Although the slow-rate phase can follow various chemical paths
compared to the fast-rate, with reasonable simplifications the fundamental
mechanisms can be finally used as the basis to scale up to the bitumen’s
performance
REFERENCES
[1] Eurobitume, The Bitumen Industry: A Global Perspective, 2011.
[2] Asphalt Institute Inc. and European Bitumen Association-Eurobitume, The
bitumen industry: A global perspective, 2015.
[3] J.S. Moulthrop, M. Massoud, The SHRP Materials Reference Library,
Washington DC, 1993.
[4] X. Lu, P. Sjövall, H. Soenen, Structural and chemical analysis of bitumen using
time-of-flight secondary ion mass spectrometry (TOF-SIMS), Fuel. 199 (2017)
206–218. https://doi.org/10.1016/j.fuel.2017.02.090.
[5] J.C. Petersen, R.E. Robertson, J.F. Branthaver, P.M. Harnsberger, J.J. Duvall, S.S.
Kim, H.U. Bahia, R. Dongre, C.E. Antle, M.G. Sharma, Binder Characterization
and Evaluation Volume 4 : Test Methods, Washington DC, 1994.
[6] J.C. Petersen, R. Glaser, Asphalt Oxidation Mechanisms and the Role of
Oxidation Products on Age Hardening Revisited, Road Mater. Pavement Des. 12
(2011) 795–819. https://doi.org/10.1080/14680629.2011.9713895.
[7] L.M. De Carvalho, C.S. Grassmann, C. Paulo, D. Bohrer, A.C. Fröhlich, C.K.
Hoinacki, F.R. Adolfo, L.E. Claussen, M. Cravo, L.F.M. Leite, Distribution of sulfur
compounds in Brazilian asphalt cements and its relationship to short-term and
long-term aging processes, Constr. Build. Mater. 117 (2016) 72–79.
https://doi.org/10.1016/j.conbuildmat.2016.04.111.
[8] P.E.Y. Wang, Y. Wen, K. Zhao, D. Chong, A.S.T. Wong, Evolution and locational
variation of asphalt binder aging in long-life hot-mix asphalt pavements, Constr.
Build. Mater. 68 (2014) 172–182.
https://doi.org/10.1016/j.conbuildmat.2014.05.091.
[9] F. Wang, Y. Xiao, P. Cui, J. Lin, M. Li, Z. Chen, Correlation of asphalt performance
indicators and aging Degrees : A review, Constr. Build. Mater. 250 (2020)
118824. https://doi.org/10.1016/j.conbuildmat.2020.118824.
[10] X. Lu, U. Isacsson, Effect of ageing on bitumen chemistry and rheology, Constr.
Build. Mater. 16 (2002) 15–22. https://doi.org/10.1016/s0950-0618(01)00033-
2.
[11] M.C. Cavalli, M. Zaumanis, E. Mazza, M.N. Partl, L.D. Poulikakos, Aging effect on
rheology and cracking behaviour of reclaimed binder with bio-based
rejuvenators, J. Clean. Prod. 189 (2018) 88–97.
https://doi.org/10.1016/j.jclepro.2018.03.305.
[12] X. Hou, F. Xiao, J. Wang, S. Amirkhanian, Identification of asphalt aging
characterization by spectrophotometry technique, Fuel. 226 (2018) 230–239.
9. References
138
https://doi.org/10.1016/j.fuel.2018.04.030.
[13] A.A.A. Molenaar, E.T. Hagos, M.F.C. van de Ven, Effects of Aging on the
Mechanical Characteristics of Bituminous Binders in PAC, J. Mater. Civ. Eng. 22
(2010). https://doi.org/10.1061/(ASCE)MT.1943-5533.0000021.
[14] J. Mirwald, D. Maschauer, B. Hofko, H. Grothe, Impact of reactive oxygen
species on bitumen aging – The Viennese binder aging method, Constr. Build.
Mater. 257 (2020) 119495.
https://doi.org/10.1016/j.conbuildmat.2020.119495.
[15] A. Sreeram, A. Masad, Z. Sootodeh Nia, D. Maschauer, J. Mirwald, B. Hofko, A.
Bhasin, Accelerated aging of loose asphalt mixtures using ozone and other
reactive oxygen species, Constr. Build. Mater. 307 (2021) 124975.
https://doi.org/https://doi.org/10.1016/j.conbuildmat.2021.124975.
[16] A. Margaritis, G. Pipintakos, G. Jacobs, D. Hernando, M. Bruynen, J. Bruurs, W.
Van den bergh, Evaluating the role of recycling rate and rejuvenator on the
chemo-rheological properties of reclaimed polymer-modified binders, Road
Mater. Pavement Des. (2021).
https://doi.org/10.1080/14680629.2021.1905700.
[17] B. Hofko, M.Z. Alavi, H. Grothe, D. Jones, J. Harvey, Repeatability and sensitivity
of FTIR ATR spectral analysis methods for bituminous binders, Mater. Struct. 50
(2017) 187. https://doi.org/10.1617/s11527-017-1059-x.
[18] J.C. Petersen, A Review of the Fundamentals of Asphalt Oxidation (E-C140),
Transp. Res. Rec. J. Transp. Res. Board. (2009).
https://doi.org/10.17226/23002.
[19] P.R. Herrington, Diffusion and reaction of oxygen in bitumen films, Fuel. 94
(2012) 86–92. https://doi.org/10.1016/j.fuel.2011.12.021.
[20] J. Mirwald, S. Werkovits, I. Camargo, D. Maschauer, B. Hofko, H. Grothe,
Understanding bitumen ageing by investigation of its polarity fractions, Constr.
Build. Mater. 250 (2020) 118809.
https://doi.org/10.1016/j.conbuildmat.2020.118809.
[21] M. Zhang, P. Hao, S. Dong, Y. Li, G. Yuan, Asphalt binder micro-characterization
and testing approaches : A review, Measurement. 151 (2020) 107255.
https://doi.org/10.1016/j.measurement.2019.107255.
[22] L. Eberhardsteiner, J. Füssl, B. Hofko, F. Handle, M. Hospodka, R. Blab, H.
Grothe, Influence of asphaltene content on mechanical bitumen behavior:
experimental investigation and micromechanical modeling, Mater. Struct. 48
(2015) 3099–3112. https://doi.org/10.1617/s11527-014-0383-7.
[23] H.R. Fischer, A. Cernescu, Relation of chemical composition to asphalt
microstructure – Details and properties of micro-structures in bitumen as seen
by thermal and friction force microscopy and by scanning near-filed optical
139
microscopy, Fuel. 153 (2015) 628–633.
https://doi.org/https://doi.org/10.1016/j.fuel.2015.03.043.
[24] U. Mühlich, G. Pipintakos, C. Tsakalidis, Mechanism based diffusion-reaction
modelling for predicting the influence of SARA composition and ageing stage
on spurt completion time and diffusivity in bitumen, Constr. Build. Mater. 267
(2021). https://doi.org/10.1016/j.conbuildmat.2020.120592.
[25] X. Yu, N.A. Burnham, M. Tao, Surface microstructure of bitumen characterized
by atomic force microscopy, Adv. Colloid Interface Sci. 218 (2015) 17–33.
https://doi.org/10.1016/j.cis.2015.01.003.
[26] The Shell Bitumen Handbook, 6th edition, ICE Publishing, 2015.
https://doi.org/10.1680/tsbh.58378.
[27] L. Loeber, G. Muller, J. Morel, O. Sutton, Bitumen in colloid science: A chemical,
structural and rheological approach, Fuel. 77 (1998) 1443–1450.
https://doi.org/10.1016/S0016-2361(98)00054-4.
[28] S. Ren, X. Liu, P. Lin, R. Jing, S. Erkens, Toward the long-term aging influence
and novel reaction kinetics models of bitumen, Int. J. Pavement Eng. (2022) 1–
16. https://doi.org/10.1080/10298436.2021.2024188.
[29] X. Jin, R. Han, Y. Cui, C.J. Glover, Fast-rate-constant-rate oxidation kinetics
model for asphalt binders, Ind. Eng. Chem. Res. 50 (2011) 13373–13379.
https://doi.org/10.1021/ie201275q.
[30] G. Pipintakos, H.Y. Vincent Ching, H. Soenen, P. Sjövall, U. Mühlich, S. Van
Doorslaer, A. Varveri, W. Van den bergh, X. Lu, Experimental investigation of
the oxidative ageing mechanisms in bitumen, Constr. Build. Mater. 260 (2020)
119702. https://doi.org/10.1016/j.conbuildmat.2020.119702.
[31] J. Knotnerus, Oxygen uptake by bitumen solutions as a potential measure of
bitumen durability, Am. Chem. Soc. Div. Pet. Chem. 16 (1971) D37–D59.
[32] T. Mill, The role of hydroaromatics in oxidative aging in asphalt, ACS Div. Fuel
Chem. Prepr. 41 (1996) 1245–1248.
[33] G. King, Oxycyclics : Understanding catalyzed oxidation mechanisms in bitumen
and other petroleum products, Fuel Sci. Technol. Int. 11 (1993) 201–238.
https://doi.org/10.1080/08843759308916063.
[34] P. Herrington, B. James, T.F.P. Henning, Validation of a Bitumen Oxidation Rate
Model, Transp. Res. Rec. J. Transp. Res. Board. 2632 (2017) 110–118.
https://doi.org/10.3141/2632-12.
[35] J. Petersen, P. Harnsberger, Asphalt Aging: Dual Oxidation Mechanism and Its
Interrelationships with Asphalt Composition and Oxidative Age Hardening,
Transp. Res. Rec. J. Transp. Res. Board. 1638 (1998) 47–55.
https://doi.org/10.3141/1638-06.
9. References
140
[36] P.K. Das, R. Balieu, N. Kringos, B. Birgisson, On the oxidative ageing mechanism
and its effect on asphalt mixtures morphology, Mater. Struct. 48 (2014) 3113–
3127. https://doi.org/10.1617/s11527-014-0385-5.
[37] B. Hofko, L. Porot, A.F. Cannone, L. Poulikakos, L. Huber, X. Lu, H. Grothe, K.
Mollenhauer, FTIR spectral analysis of bituminous binders : reproducibility and
impact of ageing temperature, Mater. Struct. 51 (2018).
https://doi.org/10.1617/s11527-018-1170-7.
[38] L.D. Poulikakos, C.F. A, D. Wang, L. Porot, B. Hofko, Impact of asphalt aging
temperature on chemomechanics, RSC Adv. 9 (2019) 11602–11613.
https://doi.org/10.1039/C9RA00645A.
[39] P.R. Herrington, Thermal decomposition of asphalt sulfoxides, Fuel. 74 (1995)
1232–1235. https://doi.org/10.1016/0016-2361(95)00039-8.
[40] EN12607-1, Bitumen and bituminous binders. Determination of the resistance
to hardening under the influence of heat and air–Part 1: RTFOT method, 2007.
[41] EN14769, Bitumen and Bituminous Binders. Accelerated Long-term Ageing
Conditioning by a Pressure Ageing Vessel (PAV), 2012.
[42] K. Zhao, Y. Wang, F. Li, Influence of ageing conditions on the chemical property
changes of asphalt binders, Road Mater. Pavement Des. (2019) 1–29.
https://doi.org/10.1080/14680629.2019.1637771.
[43] F. Liu, Z. Zhou, X. Zhang, Y. Wang, On the linking of the rheological properties
of asphalt binders exposed to oven aging and PAV aging, Int. J. Pavement Eng.
0 (2019) 1–10. https://doi.org/10.1080/10298436.2019.1608992.
[44] N.A. Al-Azri, S.H. Jung, K.M. Lunsford, A. Ferry, J.A. Bullin, R.R. Davison, C.J.
Glover, Binder oxidative aging in Texas pavements: Hardening rates, hardening
susceptibilities, and impact of pavement depth, in: Transp. Res. Rec., 2006: pp.
12–20. https://doi.org/10.3141/1962-02.
[45] J. Lamontagne, P. Dumas, V. Mouillet, J. Kister, Comparison by Fourier
transform infrared ( FTIR ) spectroscopy of different ageing techniques :
application to road bitumens, Fuel. 80 (2001) 483–488.
https://doi.org/10.1016/S0016-2361(00)00121-6.
[46] G. Tarsi, A. Varveri, C. Lantieri, A. Scarpas, C. Sangiorgi, Effects of Different Aging
Methods on Chemical and Rheological Properties of Bitumen, J. Mater. Civ. Eng.
30 (2018). https://doi.org/10.1061/(asce)mt.1943-5533.0002206.
[47] L.W. Corbett, R.E. Merz, Asphalt Binder Hardening in the Michigan Test Road
After 18 Years of Service, Transp. Res. Rec. (1975) 27–34.
[48] E. Institute, IP 469/01: Determination of saturated, aromatic and polar Flame,
compounds in petroleum products by thin layer chromatography and ionization
detection, 2006.
141
[49] N. Sakib, A. Bhasin, N. Sakib, Measuring polarity-based distributions ( SARA ) of
bitumen using simplified chromatographic techniques, Int. J. Pavement Eng.
8436 (2019) 1–28. https://doi.org/10.1080/10298436.2018.1428972.
[50] M. Paliukaitė, A. Vaitkus, A. Zofka, Evaluation of bitumen fractional composition
depending on the crude oil type and production technology, in: 9th Int. Conf.
Environ. Eng. ICEE 2014, 2014. https://doi.org/10.3846/enviro.2014.162.
[51] P. Mikhailenko, H. Baaj, Comparison of Chemical and Microstructural
Properties of Virgin and Reclaimed Asphalt Pavement Binders and Their
Saturate , Aromatic , Resin , and Asphaltene Fractions, Energy & Fuels. 33 (2019)
2633–2640. https://doi.org/10.1021/acs.energyfuels.8b03414.
[52] D. Lesueur, The colloidal structure of bitumen: Consequences on the rheology
and on the mechanisms of bitumen modification, Adv. Colloid Interface Sci. 145
(2009) 42–82. https://doi.org/10.1016/j.cis.2008.08.011.
[53] P. Redelius, H. Soenen, Relation between bitumen chemistry and performance,
Fuel. 140 (2015) 34–43. https://doi.org/10.1016/j.fuel.2014.09.044.
[54] D. Lesueur, Evidence of the Colloidal Structure of Bitumen, in: ISAP Int. Work.
Chemo-Mechanics Bitum. Mater., 2009.
https://doi.org/10.1016/j.cis.2008.08.011.
[55] A. Rosinger, Beiträge zur Kolloidchemie des Asphalts, Kolloid-Zeitschrift. 15
(1914) 177–179. https://doi.org/10.1007/BF01427821.
[56] F.J. Nellensteyn, The constitution of asphalt, J. Inst. Pet. Technol. 10 (1924)
311–323.
[57] J.M. Swanson, A Contribution to the Physical Chemistry of the Asphalts., J. Phys.
Chem. 46 (1942) 141–150.
[58] J.P. Pfeiffer, R.N.J. Saal, Asphaltic bitumen as colloid system., J. Phys. Chem. 44
(1940) 139–149.
[59] J.P. Pfeiffer, The properties of asphaltic bitumen, (1950).
[60] C. Gaestel, R. Smadja, K.. Lamminan, Contribution à la connaissance des
propriétés des bitumes routiers, Rev. Générale Des Routes Aérodromes. 466
(1971) 85–97.
[61] J.C. Petersen, R.E. Robertson, J.F. Branthaver, P.M. Harnsberger, J.J. Duvall, S.S.
Kim, D.A. Anderson, D.W. Christiansen, H.U. Bahia, Binder characterization and
evaluation: Volume 1, Rep. No. SHRP-A-367, Strateg. Highw. Res. Program, Natl.
Res. Counc. Washington, DC. (1994).
[62] D. Christensen, DW Anderson, Rheological evidence concerning the molecular
architecture of asphalt cements., in: Proc. Chem. Bitumen, Vol 2, Rome, Pp.,
1991: pp. 568–95.
[63] P.G. Redelius, The structure of asphaltenes in bitumen, Road Mater. Pavement
9. References
142
Des. 7 (2006) 143–162. https://doi.org/10.1080/14680629.2006.9690062.
[64] B. Schuler, Y. Zhang, F. Liu, A.E. Pomerantz, A. Ballard, L. Gross, V. Pauchard, S.
Banerjee, O.C. Mullins, Overview of Asphaltene Nanostructures and
Thermodynamic Applications, (2020).
https://doi.org/10.1021/acs.energyfuels.0c00874.
[65] O.C. Mullins, The Modified Yen Model, Energy & Fuels. (2010) 2179–2207.
https://doi.org/10.1021/ef900975e.
[66] O.C. Mullins, H. Sabbah, J. Eyssautier, A.E. Pomerantz, L. Barré, A.B. Andrews,
Y. Ruiz-Morales, F. Mostowfi, R. McFarlane, L. Goual, R. Lepkowicz, T. Cooper,
J. Orbulescu, R.M. Leblanc, J. Edwards, R.N. Zare, Advances in asphaltene
science and the Yen-Mullins model, in: Energy and Fuels, 2012: pp. 3986–4003.
https://doi.org/10.1021/ef300185p.
[67] J. Eyssautier, P. Levitz, D. Espinat, J. Jestin, I. Grillo, L. Barr, Insight into
Asphaltene Nanoaggregate Structure Inferred by Small Angle Neutron and X-
ray Scattering, J. Phys. Chem. B. (2011) 6827–6837.
https://doi.org/https://doi.org/10.1021/jp111468d.
[68] A. Ramm, M.C. Downer, N. Sakib, A. Bhasin, Morphology and kinetics of asphalt
binder microstructure at gas, liquid and solid interfaces., J. Microsc. 276 (2019)
109–117. https://doi.org/10.1111/jmi.12842.
[69] G. Pipintakos, N. Hasheminejad, C. Lommaert, A. Bocharova, J. Blom,
Application of Atomic Force (AFM), Environmental Scanning Electron (ESEM)
and Confocal Laser Scanning Microscopy (CLSM) in bitumen: A review of the
ageing effect, Micron. 147 (2021) 103083.
https://doi.org/10.1016/j.micron.2021.103083.
[70] L. Loeber, O. Sutton, J. Morel, J.M. Valleton, G. Muller, New direct observations
of asphalts and asphalt binders by scanning electron microscopy and atomic
force microscopy, J. Microsc. 182 (1996) 32–39.
https://doi.org/10.1046/j.1365-2818.1996.134416.x.
[71] A. Jäger, R. Lackner, C. Eisenmenger-Sittner, R. Blab, Identification of four
material phases in bitumen by atomic force microscopy, Road Mater. Pavement
Des. 5 (2004) 9–24. https://doi.org/10.1080/14680629.2004.9689985.
[72] G. Pipintakos, J. Blom, H. Soenen, W. Van den bergh, Coupling AFM and CLSM
to investigate the effect of ageing on the bee structures of bitumen, Micron.
151 (2021) 103149. https://doi.org/10.1016/j.micron.2021.103149.
[73] H. Fischer, L.D. Poulikakos, J.-P. Planche, P. Das, J. Grenfell, Challenges While
Performing AFM on Bitumen, in: N. Kringos, B. Birgisson, D. Frost, L. Wang
(Eds.), Multi-Scale Model. Charact. Infrastruct. Mater., Springer Netherlands,
Dordrecht, 2013: pp. 89–98. https://doi.org/10.1007/978-94-007-6878-9_7.
[74] H. Soenen, J. Besamusca, H.R. Fischer, L.D. Poulikakos, J. Planche, P.K. Das,
143
Laboratory investigation of bitumen based on round robin DSC and AFM tests,
Mater. Struct. (2014). https://doi.org/10.1617/s11527-013-0123-4.
[75] J. Blom, H. Soenen, N. Van den Brande, W. Van den bergh, New evidence on
the origin of ‘bee structures’ on bitumen and oils, by atomic force microscopy
(AFM) and confocal laser scanning microscopy (CLSM), Fuel. 303 (2021) 121265.
https://doi.org/10.1016/j.fuel.2021.121265.
[76] J. Blom, H. Soenen, A. Katsiki, N. Van Den Brande, H. Rahier, W. Van den bergh,
Investigation of the bulk and surface microstructure of bitumen by atomic force
microscopy, Constr. Build. Mater. 177 (2018) 158–169.
https://doi.org/10.1016/j.conbuildmat.2018.05.062.
[77] X. Lu, P. Redelius, Compositional and Structural Characterization of Waxes
Isolated from Bitumens, Energy & Fuels. 20 (2006) 653–660.
https://doi.org/10.1021/ef0503414.
[78] L. Carbognani, L. DeLima, M. Orea, U. Ehrmann, Studies on large crude oil
alkanes. II. Isolation and chracterization of aromatic waxes and waxy
asphlatnes., Pet. Sci. Technol. 18 (2000) 607–634.
https://doi.org/10.1080/10916460008949863.
[79] M. Djabourov, J. Volle, M. Kane, J. Lechaire, G. Frebourg, Morphology of
paraffin crystals in waxy crude oils cooled in quiescent conditions and under
flow, Fuel. 82 (2003) 127–135. https://doi.org/10.1016/S0016-2361(02)00222-
3.
[80] P. Claudy, J.M. Letoffe, G.N. King, J.P. Plancke, Characterization of asphalts
cements by thermomicroscopy and differential scanning calorimetry:
correlation to classic physical properties, Fuel Sci. Technol. Int. 10 (1992) 735–
765. https://doi.org/10.1080/08843759208916019.
[81] X. Lu, M. Langton, P. Olofsson, P. Redelius, Wax morphology in bitumen, J.
Mater. Sci. 40 (2005) 1893–1900. https://doi.org/10.1007/s10853-005-1208-4.
[82] P. Claudy, G.N. King, Caractérisation des bitumes routiers par analyse
calorimétrique différentielle (ACD). Analyse thermo-optique (ATO). Corrélation
entre propriétés physiques et résultats ACD, Bull. Liaison Des Lab. Des Ponts
Chaussees. (1992).
[83] P.M. Claudy, J.M. Létoffé, D. Martin, J.P. Planche, Thermal behavior of asphalt
cements, Thermochim. Acta. 324 (1998) 203–213.
[84] R. Hajj, S. Young, An analysis of theoretical and empirical relationships between
two asphalt binder cracking parameters, Road Mater. Pavement Des. 22 (2021)
S180–S196. https://doi.org/10.1080/14680629.2021.1906734.
[85] S. Weigel, D. Stephan, Relationships between the chemistry and the physical
properties of bitumen, Road Mater. Pavement Des. 19 (2018) 1636–1650.
https://doi.org/10.1080/14680629.2017.1338189.
9. References
144
[86] S.-C. Huang, H. Di Benedetto, Advances in Asphalt Materials, 2015.
https://doi.org/978-0-08-100269-8.
[87] R. Tauste, F. Moreno-Navarro, M. Sol-Sánchez, M.C. Rubio-Gámez,
Understanding the bitumen ageing phenomenon: A review, Constr. Build.
Mater. 192 (2018) 593–609.
https://doi.org/10.1016/j.conbuildmat.2018.10.169.
[88] A. Dony, L. Ziyani, I. Drouadaine, S. Pouget, S. Faucon-Dumont, D. Simard, V.
Mouillet, J.E. Poirier, T. Gabet, L. Boulange, A. Nicolai, C. Gueit, 1, MURE
National Project : FTIR spectroscopy study to assess ageing of asphalt mixtures,
in: 6th Eurasphalt Eurobitume Congr., 2016.
https://doi.org/dx.doi.org/10.14311/EE.2016.154.
[89] F.J. Ortega, F.J. Navarro, M. Jasso, L. Zanzotto, Physicochemical softening of a
bituminous binder by a reactive surfactant ( dodecenyl succinic anhydride , DSA
), Constr. Build. Mater. 222 (2019) 766–775.
https://doi.org/10.1016/j.conbuildmat.2019.06.117.
[90] L.D. Poulikakos, S. Santos, M. Bueno, S. Kuentzel, M. Hugener, M.N. Partl,
Influence of short and long term aging on chemical , microstructural and macro-
mechanical properties of recycled asphalt mixtures, Constr. Build. Mater. 51
(2014) 414–423. https://doi.org/10.1016/j.conbuildmat.2013.11.004.
[91] Q. Qin, J.F. Schabron, R.B. Boysen, M.J. Farrar, Field aging effect on chemistry
and rheology of asphalt binders and rheological predictions for field aging, Fuel.
121 (2014) 86–94. https://doi.org/10.1016/j.fuel.2013.12.040.
[92] J. Mirwald, S. Werkovits, I. Camargo, D. Maschauer, B. Hofko, H. Grothe,
Investigating bitumen long-term-ageing in the laboratory by spectroscopic
analysis of the SARA fractions, Constr. Build. Mater. 258 (2020) 119577.
https://doi.org/10.1016/j.conbuildmat.2020.119577.
[93] A. Margaritis, H. Soenen, E. Fransen, G. Pipintakos, G. Jacobs, J. Blom, W. Van
den bergh, Identification of ageing state clusters of reclaimed asphalt binders
using principal component analysis ( PCA ) and hierarchical cluster analysis (
HCA ) based on chemo-rheological parameters, Constr. Build. Mater. 244 (2020)
118276. https://doi.org/10.1016/j.conbuildmat.2020.118276.
[94] C.J. Glover, R.R. Davison, C.H. Domke, Y. Ruan, P. Juristyarini, D.B. Knorr, S.H.
Jung, Development of a New Method for Assessing Asphalt Binder Durability
with Field Validation, Texas Dep. Transp. 1872 (2005) 1–334.
https://static.tti.tamu.edu/tti.tamu.edu/documents/0-1872-2.pdf.
[95] D.J. Mensching, N.H. Gibson, A. Andriescu, G.M. Rowe, J. Sias, Assessing the
Applicability of Rheological Parameters to Evaluate Modified Binders, in: Int.
Soc. Asph. Pavements Symp. 53rd Petersen Asph. Res. Conf., 2016: pp. 1–12.
https://www.researchgate.net/publication/299804374.
145
[96] D. Lesueur, M.D. Elwardany, J.P. Planche, D. Christensen, G.N. King, Impact of
the asphalt binder rheological behavior on the value of the ΔTc parameter,
Constr. Build. Mater. 293 (2021) 123464.
https://doi.org/10.1016/j.conbuildmat.2021.123464.
[97] G. Pipintakos, H. Soenen, H.Y.V. Ching, C. Vande Velde, S. Van Doorslaer, F.
Lemière, A. Varveri, W. Van den bergh, Exploring the oxidative mechanisms of
bitumen after laboratory short- and long-term ageing, Constr. Build. Mater. 289
(2021) 123182. https://doi.org/10.1016/j.conbuildmat.2021.123182.
[98] L.W. Corbett, Relationship between composition and physical properties of
asphalt and discussion, in: Assoc. Asph. Paving Technol. Proc, 1970.
[99] X. Yu, N.A. Burnham, S. Granados-Focil, M. Tao, Bitumen’s microstructures are
correlated with its bulk thermal and rheological properties, Fuel. 254 (2019)
115509.
[100] M.A. Poirier, H. Sawatzky, Changes in chemical component type composition
and effect on rheological properties of asphalts, Prepr. Symp. 35 (1990) 301–
307.
[101] G. Thenoux, C.A. Bell, J.E. Wilson, Evaluation of physical and fractional
properties of asphalt and their interrelationship, Transp. Res. Rec. (1988) 82–
97.
[102] B. Hofko, L. Eberhardsteiner, J. Füssl, H. Grothe, F. Handle, M. Hospodka, D.
Grossegger, S.N. Nahar, A.J.M. Schmets, A. Scarpas, Impact of maltene and
asphaltene fraction on mechanical behavior and microstructure of bitumen,
Mater. Struct. Constr. 49 (2016) 829–841. https://doi.org/10.1617/s11527-
015-0541-6.
[103] C. Wang, W. Xie, B.S. Underwood, Fatigue and healing performance assessment
of asphalt binder from rheological and chemical characteristics, Mater. Struct.
Constr. 51 (2018) 1–12. https://doi.org/10.1617/s11527-018-1300-2.
[104] A.T. Pauli, J.F. Branthaver, Relationships between asphaltenes, heithaus
compatibility parameters, and asphalt viscosity, Pet. Sci. Technol. 16 (1998)
1125–1147. https://doi.org/10.1080/10916469808949827.
[105] Z. Huang, S. Zheng, H.S. Fogler, Wax deposition: experimental
characterizations, theoretical modeling, and field practices, CRC Press, 2015.
[106] H. Ding, S.A.M. Hesp, Quantification of crystalline wax in asphalt binders using
variable-temperature Fourier-transform infrared spectroscopy, Fuel. 277
(2020) 118220. https://doi.org/10.1016/j.fuel.2020.118220.
[107] X. Lu, P. Redelius, Effect of bitumen wax on asphalt mixture performance,
Constr. Build. Mater. 21 (2007) 1961–1970.
https://doi.org/https://doi.org/10.1016/j.conbuildmat.2006.05.048.
9. References
146
[108] H. Soenen, P. Redelius, The effect of aromatic interactions on the elasticity of
bituminous binders, Rheol. Acta. 53 (2014) 741–754.
https://doi.org/10.1007/s00397-014-0792-0.
[109] H. Ding, Y. Qiu, A. Rahman, Low-Temperature Reversible Aging Properties of
Selected Asphalt Binders Based on Thermal Analysis, J. Mater. Civ. Eng. 31
(2019) 4018402. https://doi.org/10.1061/(ASCE)MT.1943-5533.0002625.
[110] R. Jing, Ageing of bituminous materials: Experimental and numerical
characterization, (2019).
[111] P. Bruice, Organic Chemistry 4th Edition, (2019).
[112] A. Stoll , S.; Schweiger, EasySpin, a comprehensive software package for
spectral simulation and analysis in EPR, J. Magn. Reson. 178 (2006) 42–55.
https://doi.org/https://doi.org/10.1016/j.jmr.2005.08.013.
[113] V. Thiel, P. Sjövall, Time-of-flight secondary ion mass spectrometry (TOF-SIMS):
Principles and practice in the biogeosciences, RSC Detect. Sci. 2015-Janua
(2015) 122–170. https://doi.org/10.1039/9781782625025-00122.
[114] N. Kim, Chemical Characterization of Gilsonite Bitumen, J. Pet. Environ.
Biotechnol. 05 (2014). https://doi.org/10.4172/2157-7463.1000193.
[115] J.C. Poveda, D.R. Molina, Journal of Petroleum Science and Engineering Average
molecular parameters of heavy crude oils and their fractions using NMR
spectroscopy, J. Pet. Sci. Eng. 84–85 (2012) 1–7.
https://doi.org/10.1016/j.petrol.2012.01.005.
[116] S. Werkovits, M. Bacher, J. Theiner, T. Rosenau, H. Grothe, Multi-spectroscopic
characterization of bitumen and its polarity-based fractions, Constr. Build.
Mater. 352 (2022) 128992.
https://doi.org/https://doi.org/10.1016/j.conbuildmat.2022.128992.
[117] L. J. O’Neil, P.E. Heckelman, C.B. Koch, K.J. Roman, The Merck Index, an
Encyclopedia of Chemicals, Drugs, and Biologicals, Merck Research
Laboratories; 14th Edition, 2006.
[118] C.O. Rossi, P. Caputo, G. De Luca, L. Maiuolo, S. Eskandarsefat, C. Sangiorgi, 1H-
NMR Spectroscopy : A Possible Approach to Advanced Bitumen
Characterization for Industrial and Paving Applications, Appl. Sci. 8 (2018).
https://doi.org/10.3390/app8020229.
[119] M.H.J. Koch, J. Bordas, X-ray diffraction and scattering on disordered systems
using synchrotron radiation, Nucl. Instruments Methods Phys. Res. 208 (1983)
461–469. https://doi.org/10.1016/0167-5087(83)91169-9.
[120] B. Goderis, H. Reynaers, R. Scherrenberg, V.B.F. Mathot, M.H.J. Koch,
Temperature Reversible Transitions in Linear Polyethylene Studied by TMDSC
and Time-Resolved, Temperature-Modulated WAXD/SAXS, Macromolecules.
147
34 (2001) 1779–1787. https://doi.org/10.1021/ma001759y.
[121] G. Binnig, C.F. Quate, C. Gerbe, Atomic Force Microscope, Phys. Rev. Lett. 56
(1986).
[122] P.K. Das, H. Baaj, S. Tighe, N. Kringos, Atomic force microscopy to investigate
asphalt binders: a state-of-the-art review, Road Mater. Pavement Des. 17
(2016) 693–718. https://doi.org/10.1080/14680629.2015.1114012.
[123] V. Bellitto, Atomic Force Microscopy - Imaging, Measuring and Manipulating
Surfaces at the Atomic Scale, BoD–Books on Demand, 2012.
https://doi.org/10.5772/2673.
[124] A. García, J.P. Aguiar-Moya, J. Salazar-Delgado, A. Baldi-Sevilla, L.G. Loría-
Salazar, Methodology for estimating the modulus of elasticity of bitumen under
different aging conditions by AFM, Road Mater. Pavement Des. 20 (2019) S332–
S346. https://doi.org/10.1080/14680629.2019.1588152.
[125] H. Ban, P. Karki, Y.-R. Kim, Nanoindentation test integrated with numerical
simulation to characterize mechanical properties of rock materials, J. Test. Eval.
42 (2014) 787–796. https://doi.org/10.1520/JTE20130035.
[126] R.G. Allen, D.N. Little, A. Bhasin, Structural Characterization of Micromechanical
Properties in Asphalt Using Atomic Force Microscopy, J. Mater. Civ. Eng. 24
(2012) 1317–1327. https://doi.org/10.1061/(asce)mt.1943-5533.0000510.
[127] S.N. Nahar, A.J.M. Schmets, A. Scarpas, G. Schitter, Temperature and thermal
history dependence of the microstructure in bituminous materials, Eur. Polym.
J. 49 (2013) 1964–1974. https://doi.org/10.1016/j.eurpolymj.2013.03.027.
[128] D. Nečas, P. Klapetek, Gwyddion: an open-source software for SPM data
analysis, Open Phys. 10 (2012) 181–188.
[129] B.V.R. Tata, B. Raj, Confocal laser scanning microscopy: Applications in material
science and technology, Bull. Mater. Sci. 21 (1998) 263–278.
https://doi.org/10.1007/BF02744951.
[130] W.B. Amos, J.G. White, How the confocal laser scanning microscope entered
biological research, Biol. Cell. 95 (2003) 335–342.
https://doi.org/10.1016/s0248-4900(03)00078-9.
[131] M. Minsky, Memoir on inventing the confocal scanning microscope, Scanning.
10 (1988) 128–138. https://doi.org/10.1002/sca.4950100403.
[132] N.S. Claxton, T.J. Fellers, M.W. Davidson, Laser Scanning Confocal Microscopy
(LSCM), SpringerReference. (2011).
https://doi.org/10.1007/springerreference_67032.
[133] S.W. Paddock, Principles and practices of laser scanning confocal microscopy,
Mol. Biotechnol. 16 (2000) 127–149. https://doi.org/10.1385/MB:16:2:127.
[134] H. Li, W.K. Wan, Investigation of the asphaltene precipitation process from cold
9. References
148
lake bitumen by confocal scanning laser microscopy, in: SPE Int. Heavy Oil
Symp., Society of Petroleum Engineers, 1995: pp. 709–740.
https://doi.org/10.2523/30321-ms.
[135] R.J. Mikula, V.A. Munoz, Characterization of emulsions and suspensions in the
petroleum industry using cryo-SEM and CLSM, Colloids Surfaces A
Physicochem. Eng. Asp. 174 (2000) 23–36. https://doi.org/10.1016/S0927-
7757(00)00518-5.
[136] S. Bearsley, A. Forbes, R. G. Haverkamp, Direct observation of the asphaltene
structure in paving-grade bitumen using confocal laser-scanning microscopy, J.
Microsc. 215 (2004) 149–155. https://doi.org/10.1111/j.0022-
2720.2004.01373.x.
[137] E.M. A., S. Saman, M.L. N., H. Marwa, D.W. H., D. Samer, New Approach to
Recycling Asphalt Shingles in Hot-Mix Asphalt, J. Mater. Civ. Eng. 24 (2012)
1403–1411. https://doi.org/10.1061/(ASCE)MT.1943-5533.0000520.
[138] D. Großegger, Microstructural aging of bitumen, in: 6th Eurasphalt Eurobitume
Congr., 2017. https://doi.org/10.14311/ee.2016.135.
[139] S.J. Rozeveld, E. Eugene Shin, A. Bhurke, L. France, L.T. Drzal, Network
morphology of straight and polymer modified asphalt cements, Microsc. Res.
Tech. 38 (1997) 529–543. https://doi.org/10.1002/(SICI)1097-
0029(19970901)38:5<529::AID-JEMT11>3.0.CO;2-O.
[140] Y.-J. Lee, L.M. France, M.C. Hawley, The Effect of Network Formation on the
Rheological Properties of SBR Modified Asphalt Binders, Rubber Chem.
Technol. 70 (1997) 256–263. https://doi.org/10.5254/1.3538430.
[141] D. Kaya, A. Topal, T. McNally, Correlation of processing parameters and ageing
with the phase morphology of styrene-butadiene-styrene block co-polymer
modified bitumen, Mater. Res. Express. 6 (2019) 105309.
https://doi.org/10.1088/2053-1591/ab349c.
[142] H. Yin, Y. Zhang, Y. Sun, W. Xu, D. Yu, H. Xie, R. Cheng, Performance of hot mix
epoxy asphalt binder and its concrete, Mater. Struct. 48 (2015) 3825–3835.
https://doi.org/10.1617/s11527-014-0442-0.
[143] J. Gong, Y. Liu, Q. Wang, Z. Xi, J. Cai, G. Ding, H. Xie, Performance evaluation of
warm mix asphalt additive modified epoxy asphalt rubbers, Constr. Build.
Mater. 204 (2019) 288–295.
https://doi.org/10.1016/j.conbuildmat.2019.01.197.
[144] Y. Liu, Z. Xi, J. Cai, H. Xie, Laboratory investigation of the properties of epoxy
asphalt rubber (EAR), Mater. Struct. 50 (2017) 219.
https://doi.org/10.1617/s11527-017-1089-4.
[145] Y. Liu, J. Zhang, R. Chen, J. Cai, Z. Xi, H. Xie, Ethylene vinyl acetate copolymer
modified epoxy asphalt binders: phase separation evolution and mechanical
149
properties, Constr. Build. Mater. 137 (2017) 55–65.
https://doi.org/10.1016/j.conbuildmat.2017.01.081.
[146] Y. Jiang, Y. Liu, J. Gong, C. Li, Z. Xi, J. Cai, H. Xie, Microstructures, thermal and
mechanical properties of epoxy asphalt binder modified by SBS containing
various styrene-butadiene structures, Mater. Struct. 51 (2018) 86.
https://doi.org/10.1617/s11527-018-1217-9.
[147] Y. Liu, J. Zhang, Y. Jiang, C. Li, Z. Xi, J. Cai, H. Xie, Investigation of secondary
phase separation and mechanical properties of epoxy SBS-modified asphalts,
Constr. Build. Mater. 165 (2018) 163–172.
https://doi.org/10.1016/j.conbuildmat.2018.01.032.
[148] D.W. Christensen, D.A. Anderson, Interpretation of dynamic mechanical test
data for paving grade asphalt cements (with discussion), J. Assoc. Asph. Paving
Technol. 61 (1992).
[149] L. Garcia Cucalon, F. Kaseer, E. Arámbula-Mercado, A. Epps Martin, N. Morian,
S. Pournoman, E. Hajj, The crossover temperature: significance and application
towards engineering balanced recycled binder blends, Road Mater. Pavement
Des. 20 (2019) 1391–1412. https://doi.org/10.1080/14680629.2018.1447504.
[150] R. Zhang, J.E. Sias, E. V Dave, Correlating Laboratory Conditioning with Field
Aging for Asphalt using Rheological Parameters, Transp. Res. Rec. 2674 (2020)
393–404. https://doi.org/10.1177/0361198120915894.
[151] D.W. Christensen, D.A. Anderson, G.M. Rowe, Relaxation spectra of asphalt
binders and the Christensen–Anderson rheological model, Road Mater.
Pavement Des. 18 (2017) 382–403.
https://doi.org/10.1080/14680629.2016.1267448.
[152] W.S. Mogawer, T. Bennert, A. Austerman, C. Ericson, AAPT symposium:
Investigating the aging mitigation capabilities of rejuvenators in high RAP
mixtures using black space diagrams, binder rheology and mixture tests, in:
Asph. Paving Technol. Assoc. Asph. Paving Technol. Tech. Sess., Association of
Asphalt Paving Technologist, 2015: pp. 705–738.
[153] D. Wang, A.C. Falchetto, C. Riccardi, L. Poulikakos, B. Hofko, L. Porot, M.P.
Wistuba, H. Baaj, P. Mikhailenko, K.H. Moon, Investigation on the combined
effect of aging temperatures and cooling medium on rheological properties of
asphalt binder based on DSR and BBR, Road Mater. Pavement Des. 20 (2019)
S409–S433.
[154] R.M. Anderson, G.N. King, D.I. Hanson, P.B. Blankenship, Evaluation of the
relationship between asphalt binder properties and non-load related cracking,
Asph. Paving Technol. Assoc. Asph. Paving Technol. Tech. Sess. 80 (2011) 615–
662.
[155] D. Christensen, D. Mensching, G. Rowe, R.M. Anderson, A. Hanz, G. Reinke, D.
9. References
150
Anderson, Past, present, and future of asphalt binder rheological parameters:
Synopsis of 2017 Technical Session 307 at the 96th Annual Meeting of the
Transportation Research Board, Transp. Res. Circ. (2019).
[156] T.W. Kennedy, G.A. Huber, E.T. Harrigan, R.J. Cominsky, C.S. Hughes, H. Von
Quintus, J.S. Moulthrop, Superior performing asphalt pavements (Superpave):
The product of the SHRP asphalt research program, (1994).
[157] A.M. Hung, A. Goodwin, E.H. Fini, Effects of water exposure on bitumen surface
microstructure, Constr. Build. Mater. 135 (2017) 682–688.
https://doi.org/10.1016/j.conbuildmat.2017.01.002.
[158] J. Mirwald, D. Nura, B. Hofko, Recommendations for handling bitumen prior to
FTIR spectroscopy, Mater. Struct. 55 (2022) 26.
https://doi.org/10.1617/s11527-022-01884-1.
[159] H. Soenen, X. Lu, O.-V. Laukkanen, Oxidation of bitumen: molecular
characterization and influence on rheological properties, Rheol. Acta. 55 (2016)
315–326. https://doi.org/10.1007/s00397-016-0919-6.
[160] H.R. Fischer, E.C. Dillingh, C.G.M. Hermse, Applied Surface Science On the
interfacial interaction between bituminous binders and mineral surfaces as
present in asphalt mixtures, Appl. Surf. Sci. 265 (2013) 495–499.
https://doi.org/10.1016/j.apsusc.2012.11.034.
[161] S.N. Nahar, A.J.M. Schmets, C. Kasbergen, G. Schitter, A. Scarpas, Self-healing
of bituminous materials by damage reversal at the microstructural scale, in:
Proc. 94th Annu. Meet. Transp. Res. Board, 2015: pp. 11–15.
[162] F. Mastoras, A. Varveri, M. van Tooren, S. Erkens, Effect of mineral fillers on
ageing of bituminous mastics, Constr. Build. Mater. 276 (2021) 122215.
https://doi.org/https://doi.org/10.1016/j.conbuildmat.2020.122215.
[163] Y. Cui, C.J. Glover, J. Braziunas, H. Sivilevicius, Further exploration of the
pavement oxidation model Diffusion-reaction balance in asphalt, Constr. Build.
Mater. 161 (2018) 132–140.
https://doi.org/10.1016/j.conbuildmat.2017.11.095.
[164] R. Karlsson, U. Isacsson, Laboratory studies of diffusion in bitumen using
markers, J. Mater. Sci. 38 (2003) 2835–2844.
https://doi.org/10.1023/A:1024476217060.
[165] R. Han, X. Jin, C.J. Glover, Petroleum Science and Technology Oxygen Diffusivity
in Asphalts and Mastics, Pet. Sci. Technol. (2013) 37–41.
https://doi.org/10.1080/10916466.2011.559506.
[166] J.H. Tannous, A. De Klerk, Quantification of the Free Radical Content of Oilsands
Bitumen Fractions, Energy & Fuels. 33 (2019) 7083–7093.
https://doi.org/10.1021/acs.energyfuels.9b01115.
151
[167] X. Lu, P. Sjövall, H. Soenen, M. Andersson, Microstructures of bitumen observed
by environmental scanning electron microscopy ( ESEM ) and chemical analysis
using time-of- flight secondary ion mass spectrometry ( TOF-SIMS ), Fuel. 229
(2018) 198–208. https://doi.org/10.1016/j.fuel.2018.05.036.
[168] M.F. Libert, I. Walczak, Effect of radio-oxidative ageing and pH on the release
of soluble organic matter from bitumen, in: Int. Conf. Sci. Res. Back-End Fuel
Cycle 21. Century, France, 2000: pp. 1–4.
[169] G. Boczkaj, A. Przyjazny, M. Kamin, Characteristics of volatile organic
compounds emission profiles from hot road bitumens, Chemosphere. 107
(2014) 23–30. https://doi.org/10.1016/j.chemosphere.2014.02.070.
[170] I.-H. Wang, J.L. Boucher, R.A. Romine, R.D. Rowlett, G.-D. Lei, Oxidation
chemistry in asphalt, Fuel Sci. Technol. Int. 11 (1993) 1–28.
https://doi.org/http://dx.doi.org/10.1080/08843759308916056.
[171] L. Ma, A. Varveri, R. Jing, S. Erkens, Comprehensive review on the transport and
reaction of oxygen and moisture towards coupled oxidative ageing and
moisture damage of bitumen, Constr. Build. Mater. 283 (2021) 122632.
https://doi.org/10.1016/j.conbuildmat.2021.122632.
[172] I. Gabrielle do Nascimento Camargo, B. Hofko, J. Mirwald, G. Hinrich, Effect of
Thermal and Oxidative Aging on Asphalt Binders Rheology and Chemical
Composition, Materials (Basel). 13 (2020) 4438.
https://doi.org/doi:10.3390/ma13194438.
[173] J.F. Branhaver, J.C. Petersen, R.E. Robertson, J.J. Duvall, S.S. Kim, P.M.
Harnsberger, T. Mill, E.K. Ensley, F.A. Barbour, J.F. Schabron, SHRP-A-368 Binder
Characterization and Evaluation Volume 2 : Chemistry, Washington, DC, 1993.
[174] J. Mirwald, B. Hofko, G. Pipintakos, J. Blom, H. Soenen, Comparison of
microscopic techniques to study the diversity of the bitumen microstructure,
Micron. 159 (2022) 103294. https://doi.org/10.1016/j.micron.2022.103294.
[175] L.C. Michon, D.A. Netzel, T.F. Turner, D. Martin, J.P. Planche, A 13C NMR and
DSC study of the amorphous and crystalline phases in asphalts, Energy and
Fuels. 13 (1999) 602–610. https://doi.org/10.1021/ef980184r.
[176] I.N. Frolov, E.S. Okhotnikova, M.A. Ziganshin, A.A. Firsin, Cold crystallization
event on DSC heating curves of bitumen, J. Therm. Anal. Calorim. 147 (2022)
5269–5278. https://doi.org/10.1007/s10973-021-10908-x.
[177] V.B.F. Mathot, Calorimetry and thermal analysis of polymers, Hanser
Publishers, 1994.
[178] C.W. Bunn, D.R. Holmes, Chain configurations in crystals of simple linear
polymers, Discuss. Faraday Soc. 25 (1958) 95–103.
https://doi.org/10.1039/DF9582500095.
9. References
152
[179] J. Masson, V. Leblond, J. Margeson, Bitumen morphologies by phase-detection
atomic force microscopy, J. Microsc. 221 (2006) 17–29.
https://doi.org/10.1111/j.1365-2818.2006.01540.x.
[180] S. Liu, A. Peng, S. Zhou, J. Wu, W. Xuan, W. Liu, Evaluation of the ageing
behaviour of waste engine oil-modified asphalt binders, Constr. Build. Mater.
223 (2019) 394–408. https://doi.org/10.1016/j.conbuildmat.2019.07.020.
[181] S. Wu, L. Pang, L. Mo, Y. Chen, G. Zhu, Influence of aging on the evolution of
structure , morphology and rheology of base and SBS modified bitumen, Constr.
Build. Mater. 23 (2009) 1005–1010.
https://doi.org/10.1016/j.conbuildmat.2008.05.004.
[182] W. Zhang, L. Zou, Z. Jia, F. Wang, Y. Li, P. Shi, Effect of thermo-oxidative ageing
on nano-morphology of bitumen, Appl. Sci. 9 (2019).
https://doi.org/10.3390/app9153027.
[183] L.M. Rebelo, J.S. De Sousa, A.S. Abreu, M.P.M.A. Baroni, A.E.V. Alencar, S.A.
Soares, J. Mendes Filho, J.B. Soares, Aging of asphaltic binders investigated with
atomic force microscopy, Fuel. 117 (2014) 15–25.
https://doi.org/10.1016/j.fuel.2013.09.018.
[184] A. Koyun, J. Büchner, M.P. Wistuba, H. Grothe, Rheological, spectroscopic and
microscopic assessment of asphalt binder ageing, Road Mater. Pavement Des.
23 (2022) 80–97. https://doi.org/10.1080/14680629.2020.1820891.
[185] M. Xu, J. Yi, Z. Pei, D. Feng, Y. Huang, Y. Yang, Generation and evolution
mechanisms of pavement asphalt aging based on variations in surface structure
and micromechanical characteristics with AFM, Mater. Today Commun. 12
(2017) 106–118. https://doi.org/10.1016/j.mtcomm.2017.07.006.
[186] H.L. Zhang, H.C. Wang, J.Y. Yu, Effect of aging on morphology of organo-
montmorillonite modified bitumen by atomic force microscopy, J. Microsc. 242
(2011) 37–45. https://doi.org/10.1111/j.1365-2818.2010.03435.x.
[187] A. Chen, G. Liu, Y. Zhao, J. Li, Y. Pan, J. Zhou, Research on the aging and
rejuvenation mechanisms of asphalt using atomic force microscopy, Constr.
Build. Mater. 167 (2018) 177–184.
https://doi.org/10.1016/j.conbuildmat.2018.02.008.
[188] P.E.Y. Wang, K. Zhao, C. Glover, L. Chen, Y. Wen, D. Chong, Effects of aging on
the properties of asphalt at the nanoscale, Constr. Build. Mater. 80 (2015) 244–
254. https://doi.org/10.1016/j.conbuildmat.2015.01.059.
[189] H.L. Zhang, J.Y. Yu, Z.G. Feng, L.H. Xue, S.P. Wu, Effect of aging on the
morphology of bitumen by atomic force, J. Microsc. 246 (2012) 11–19.
https://doi.org/10.1111/j.1365-2818.2011.03578.x.
[190] J.P. Aguiar-Moya, J. Salazar-Delgado, A. García, A. Baldi-Sevilla, V. Bonilla-Mora,
L.G. Loría-Salazar, Effect of ageing on micromechanical properties of bitumen
153
by means of atomic force microscopy, Road Mater. Pavement Des. 18 (2017)
203–215. https://doi.org/10.1080/14680629.2017.1304249.
[191] Y. Yuan, X. Zhu, L. Chen, Relationship among cohesion , adhesion , and bond
strength : From multi-scale investigation of asphalt-based composites
subjected to laboratory-simulated aging, Mater. Des. 185 (2020) 108272.
https://doi.org/10.1016/j.matdes.2019.108272.
[192] J. Zhu, R. Balieu, X. Lu, N. Kringos, Microstructure evaluation of polymer-
modified bitumen by image analysis using two-dimensional fast Fourier
transform, Mater. Des. 137 (2018) 164–175.
[193] A.R. Pathak, M. Pandey, S. Rautaray, Application of Deep Learning for Object
Detection, Procedia Comput. Sci. 132 (2018) 1706–1717.
https://doi.org/https://doi.org/10.1016/j.procs.2018.05.144.
[194] J. Redmon, A. Farhadi, Yolov3: An incremental improvement, ArXiv Prepr.
ArXiv1804.02767. (2018).
[195] A.A.-K.H. Abu-Ein, A novel methodology for digital removal of periodic noise
using 2D fast Fourier transforms, Contemp. Eng. Sci. 3 (2014) 103–116.
[196] G. Pipintakos, H.Y.V. Ching, U. Mühlich, H. Soenen, S. Van Doorslaer, P. Sjövall,
A. Varveri, C. Vande Velde, X. Lu, Experimental Validation of the Dual-Oxidation
Routes in Bituminous Binders, in: H. Di Benedetto, H. Baaj, E. Chailleux, G.
Tebaldi, C. Sauzéat, S. Mangiafico (Eds.), RILEM Bookseries, Springer
International Publishing, Cham, 2022: pp. 903–909.
https://doi.org/10.1007/978-3-030-46455-4_115.
[197] U. Mühlich, G. Pipintakos, C. Tsakalidis, Modelling of Oxidative Ageing in
Bitumen Using Thermodynamics of Irreversible Processes (TIP): Potential and
Challenges BT - Proceedings of the RILEM International Symposium on
Bituminous Materials, in: H. Di Benedetto, H. Baaj, E. Chailleux, G. Tebaldi, C.
Sauzéat, S. Mangiafico (Eds.), Springer International Publishing, Cham, 2022:
pp. 1359–1365.
[198] U. Mühlich, Towards a continuum thermodynamics framework for mechanism-
based modelling of oxidative ageing in bitumen, Int. J. Pavement Eng. 21 (2020)
1419–1428. https://doi.org/10.1080/10298436.2018.1546859.
[199] J. Read, D. Whiteoak, The Shell Bitumen Handbook, 5th edition, Thomas
Telford, 2003. https://doi.org/10.1680/sbh.32200.
[200] J.W.H. Oliver, Changes in the Chemical Composition of Australian Bitumens,
Road Mater. Pavement Des. 10 (2009) 569–586.
https://doi.org/10.1080/14680629.2009.9690214.
[201] J. Vidal, Thermodynamics, Editions OPHRYS, 1971.
[202] M. Pekař, I. Samohýl, The thermodynamics of linear fluids and fluid mixtures,
9. References
154
Springer, 2014.
[203] Diana Paola powers, Characterization and Asphaltene Precipitation Modeling
of Native and Reacted Crude Oils, Department of chemical and petroleum
engineering, Calgary, Albert, 2014.
[204] K. Akbarzadeh, H. Alboudwarej, W.Y. Svrcek, H.W. Yarranton, A generalized
regular solution model for asphaltene precipitation from n-alkane diluted
heavy oils and bitumens, Fluid Phase Equilib. (2005).
https://doi.org/10.1016/j.fluid.2005.03.029.
RESEARCH DISSEMINATION
Publications related to the dissertation
1. Georgios Pipintakos, H.Y. Vincent Ching, Hilde Soenen, Peter Sjövall, Uwe
Mühlich, Sabine Van Doorslaer, Aikaterini Varveri, Wim Van den bergh and Xiaohu
Lu. Experimental investigation of the oxidative ageing mechanisms in bitumen.
Construction and Building materials (2020).
2. Georgios Pipintakos, Hilde Soenen, Hong Yue Vincent Ching, Christophe Vande
Velde, Sabine Van Doorslaer, Filip Lemière, Aikaterini Varveri and Wim Van den
bergh. Exploring the oxidative mechanisms of bitumen after laboratory short- and
long-term ageing. Construction and building materials (2021).
3. Georgios Pipintakos, Navid Hasheminejad, Caitlin Lommaert, Anastassiya
Bocharova and Johan Blom. Application of Atomic Force (AFM), Environmental
Scanning Electron (ESEM) and Confocal Laser Scanning Microscopy (CLSM) in
bitumen : a review of the ageing effect. Micron (2021).
4. Georgios Pipintakos, Johan Blom, Hilde Soenen and Wim Van den bergh.
Coupling AFM and CLSM to investigate the effect of ageing on the bee structures
of bitumen. Micron (2021).
5. Georgios Pipintakos, Hilde Soenen, Bart Goderis, Johan Blom and Xiaohu Lu.
Crystallinity of bitumen via WAXD and DSC and its effect on the surface
microstructure. Crystals (2022).
6. Georgios Pipintakos, Caitlin Lommaert, Aikaterini Varveri and Wim Van den
bergh. Do chemistry and rheology follow the same laboratory ageing trends in
bitumen? Materials and Structures (2022).
7. Georgios Pipintakos, H.Y. Vincent Ching, Uwe Mühlich, Hilde Soenen, Sabine
Van Doorslaer, Peter Sjövall, Aikaterini Varveri, Christophe Vande Velde and
Xiaohu Lu. Experimental validation of the dual-oxidation routes in bituminous
binders. ISBM 2020: Proceedings of the RILEM International Symposium on
Bituminous Materials (2021).
8. Georgios Pipintakos, Lili Ma, Aikaterini Varveri, Ruxin Jing, Hilde Soenen and
Wim Van den bergh. Diffusivity and reactivity of oxygen in bitumen and mastics.
Advances in Materials and Pavement Performance Prediction: Proceedings of the
International AM3P Conference (2023)
156
9. Uwe Muehlich, Georgios Pipintakos and Christos Tsakalidis. Mechanism-based
diffusion-reaction modelling for predicting the influence of SARA composition and
ageing stage on spurt completion time and diffusivity in bitumen. Construction and
building materials (2021).
10. Uwe Muehlich, Georgios Pipintakos and Christos Tsakalidis. Modelling of
Oxidative Ageing in Bitumen Using Thermodynamics of Irreversible Processes
(TIP): Potential and Challenges. ISBM 2020: Proceedings of the RILEM International
Symposium on Bituminous Materials (2021).
11. Xiaohu Lu, Hilde Soenen, Peter Sjövall and Georgios Pipintakos. Analysis of
asphaltenes and maltenes before and after long-term aging of bitumen. Fuel
(2021).
12. Navid Hasheminejad, Georgios Pipintakos, Cedric Vuye, Thomas De Kerf, Taher
Ghalandari, Johan Blom and Wim Van den bergh. Utilizing deep learning and
advanced image processing techniques to investigate the microstructure of a waxy
bitumen. Construction and building materials (2021).
Co-authored publications during the doctoral trajectory
13. Alexandros Margaritis, Hilde Soenen, Erik Fransen, Georgios Pipintakos, Geert
Jacobs, Johan Blom and Wim Van den bergh. Identification of ageing state clusters
of reclaimed asphalt binders using principal component analysis (PCA) and
hierarchical cluster analysis (HCA) based on chemo-rheological parameters.
Construction and Building materials (2020).
14. Alexandros Margaritis, Georgios Pipintakos, Geert Jacobs, David Hernando,
Mats Bruynen, Jeroen Bruurs and Wim Van den bergh. Evaluating the role of
recycling rate and rejuvenator on the chemo-rheological properties of reclaimed
polymer-modified binders. Road Materials and Pavement Design (2021).
15. Alexandros Margaritis, Georgios Pipintakos, Aikaterini Varveri, Geert Jacobs,
Navid Hasheminejad, Johan Blom and Wim Van den bergh. Towards an enhanced
fatigue evaluation of bituminous mortars. Construction and Building materials
(2021).
16. Alexandros Margaritis, Navid Hasheminejad, Georgios Pipintakos, Geert
Jacobs, Johan Blom and Wim Van den bergh. The impact of reclaimed asphalt rate
on the healing potential of bituminous mortars and mixtures. International Journal
of Pavement Engineering (2021).
157
17. Alexandros Margaritis, Geert Jacobs, Georgios Pipintakos, Johan Blom, and
Wim Van den bergh. Fatigue Resistance of Bituminous Mixtures and Mortars
Containing High Reclaimed Asphalt Content. Materials (2020).
18. I. Rocha Segundo, S Jr. Landi, A. Margaritis, Georgios Pipintakos, E. Freitas, C.
Vuye, J. Blom, S. Tytgat, S. Denys and J. Carneiro. Physicochemical and Rheological
Properties of a Transparent Asphalt Binder Modified with Nano-TiO2.
Nanomaterials (2021).
19. S. R. Omranian, Hamzah, M.O., Georgios Pipintakos, W. Van den bergh, C. Vuye
and M.R.M. Hasan. Effects of Short-Term Aging on the Compactibility and
Volumetric Properties of Asphalt Mixtures Using the Response Surface Method.
Sustainability (2020).
20. Thomas De Kerf, Georgios Pipintakos, Zohreh Zahiri, Steve Vanlanduit and Paul
Scheunders. Identification of Corrosion Minerals Using Shortwave Infrared
Hyperspectral Imaging. Sensors (2022).
21. Johannes Mirwald, Bernhard Hofko, Georgios Pipintakos, Johan Blom and
Hilde Soenen. Comparison of Microscopic Techniques to study the Diversity of the
Bitumen Microstructure. Micron (2022).
22. Alexandros Margaritis, Georgios Pipintakos, Li Ming Zhang, Geert Jacobs,
Cedric Vuye, Johan Blom and Wim Van den Bergh. Introducing an Improved
Testing Method to Evaluate the Fatigue Resistance of Bituminous Mortars. ISBM
2020: Proceedings of the RILEM International Symposium on Bituminous Materials
(2021).
23. Seyed Reza Omranian, Michiel Geluykens, Myrthe Van Hal, Navid
Hasheminejad, Iran Rocha Segundo, Georgios Pipintakos, Siegfried Denys, Tom
Tytgat, Elisabete Fraga Freitas, Joaquim Carneiro, Sammy Verbruggen and Cedric
Vuye. Assessing the Potential of Titanium Dioxide Application on Asphalt
Pavements to Degrade Soot Deposition. Construction and Building materials
(2022).
24. Dheeraj Adwani, Anand Sreeram, Georgios Pipintakos, Johannes Mirwald, Yudi
Wang, Ramez Hajj, Ruxin Jing and Amit Bhasin. Interpreting the Effectiveness of
Antioxidants to Increase the Resilience of Asphalt Binders: First of its Kind Global
Interlaboratory Study. Construction and Building materials (under review).
25. Geert Jacobs, Georgios Pipintakos, Xander Van den Buijs and Wim Van den
bergh. Chemo-rheological equivalence of bitumen between different lab ageing
158
procedures: from binder to mixture. Road Materials and Pavement Design, Special
Issue: EATA 2023 (under review).
26. Michael Elwardany, Nusnin Akter, Panos Apostolidis, Terry Arnold, Mike
Aurilio, Lorena Garcia-Cucalon, Hamzeh Haghshenas, Kamal Hossain, David
Mensching, Johannes Mirwald, Virginie Mouillet, Georgios Pipintakos, Laurent
Porot, Nibert Saltibus, Pezhouhan Tavassoti-Kheiry, Sandra Weigel, Gayle King,
Jean-Pascal Planche, Gerald Reinke and Jack Youtcheff. A Review of FT-IR
Spectroscopy for Asphalt Bitumen Engineers and Scientists. Transportation
Research Board E-Circular (under review).
