Surface microstructure of bitumen characterized by

Advances in Colloid and Interface Science 218 (2015) 17–33

Contents lists available at ScienceDirect

Advances in Colloid and Interface Science journal homepage: www.elsevier.com/locate/cis

Historical perspective

Surface microstructure of bitumen characterized by atomic force microscopy Xiaokong Yu a,⁎, Nancy A. Burnham b, Mingjiang Tao a,⁎ a b

Department of Civil and Environmental Engineering, Worcester Polytechnic Institute, 100 Institute Road, Worcester, MA 01609, USA Physics and Biomedical Engineering Departments, Worcester Polytechnic Institute, 100 Institute Road, Worcester, MA 01609, USA

a r t i c l e

i n f o

Available online 22 January 2015 Keywords: Bitumen Asphalt binder Atomic force microscopy Surface microstructures ‘Bee-structures’ Wax

a b s t r a c t Bitumen, also called asphalt binder, plays important roles in many industrial applications. It is used as the primary binding agent in asphalt concrete, as a key component in damping systems such as rubber, and as an indispensable additive in paint and ink. Consisting of a large number of hydrocarbons of different sizes and polarities, together with heteroatoms and traces of metals, bitumen displays rich surface microstructures that affect its rheological properties. This paper reviews the current understanding of bitumen's surface microstructures characterized by Atomic Force Microscopy (AFM). Microstructures of bitumen develop to different forms depending on crude oil source, thermal history, and sample preparation method. While some bitumens display surface microstructures with fine domains, flake-like domains, and dendrite structuring, ‘bee-structures’ with wavy patterns several micrometers in diameter and tens of nanometers in height are commonly seen in other binders. Controversy exists regarding the chemical origin of the ‘bee-structures’, which has been related to the asphaltene fraction, the metal content, or the crystallizing waxes in bitumen. The rich chemistry of bitumen can result in complicated intermolecular associations such as coprecipitation of wax and metalloporphyrins in asphaltenes. Therefore, it is the molecular interactions among the different chemical components in bitumen, rather than a single chemical fraction, that are responsible for the evolution of bitumen's diverse microstructures, including the ‘bee-structures’. Mechanisms such as curvature elasticity and surface wrinkling that explain the rippled structures observed in polymer crystals might be responsible for the formation of ‘bee-structures’ in bitumen. Despite the progress made on morphological characterization of bitumen using AFM, the fundamental question whether the microstructures observed on bitumen surfaces represent its bulk structure remains to be addressed. In addition, critical technical challenges associated with AFM characterization of bitumen surface structures are discussed, with possible solutions recommended. For future work, combining AFM with other chemical analysis tools that can generate comparable high resolution to AFM would provide an avenue to linking bitumen's chemistry to its microscopic morphological and mechanical properties and consequently benefit the efforts of developing structure-related models for bituminous materials across the different length scales. © 2015 Elsevier B.V. All rights reserved.

Contents 1. 2. 3. 4. 5. 6.

7. 8.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . AFM techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General microstructures observed on bitumen surfaces . . . . . . . . . . . . . The effect of sample preparation methods on bitumen microstructures . . . . . . The effect of thermal history on bitumen microstructures . . . . . . . . . . . . What is the chemical origin of the bee-structures? . . . . . . . . . . . . . . . 6.1. The asphaltenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2. The metal content . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3. The wax crystallization . . . . . . . . . . . . . . . . . . . . . . . . . Possible mechanisms responsible for the bee-structures . . . . . . . . . . . . . Are the microstructures observed on bitumen surfaces present in the bulk bitumen?

⁎ Corresponding authors. Tel.: +1 508 831 6487; fax: +1 508 831 5808. E-mail addresses: [email protected] (X. Yu), [email protected] (M. Tao).

http://dx.doi.org/10.1016/j.cis.2015.01.003 0001-8686/© 2015 Elsevier B.V. All rights reserved.

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

18 22 22 24 26 27 27 28 28 30 30

18

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

9. Technical challenges and recommendations 10. Conclusions . . . . . . . . . . . . . . Acknowledgment . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

1. Introduction Bitumen, also called asphalt binder, comes from the process of fractional distillation of petroleum (Considering the popularity of the words bitumen and asphalt binder in Europe and North America, respectively, these two words are used interchangeably throughout this paper.). Bitumen has been used in roof shingles for waterproofing, in cars and computers for vibration mitigation, and in paintings and inks as an additive for durability improvement. In addition, it has been widely used in road construction since ancient times (i.e., King of Babylon, 625– 604 BC) [1]. At the present time, 93% of more than 2.6 million miles of paved roads in United States are surfaced with bitumen. In the paving industry, bitumen acts as a glue binding the aggregates and filler particles together to form asphalt concrete. The efforts of producing more sustainable asphalt pavements are motivated by economic and environmental considerations (i.e., the increasing cost and the depletion of the non-renewable petroleum resources). Back in 1987, the American Association of State Highway and Transportation Officials initiated a five-year research program entitled ‘The Strategic Highway Research Program (SHRP)’, with one of its units focusing on developing performance-related specifications for asphalt binder and asphalt concrete. A profound advance from this project was the development of a set of performance-based testing methods to characterize the rheology, fatigue, and thermal cracking of asphalt binders. However, due to the complex chemistry of asphalt binder and the limited available techniques for microscopic investigation [2], the fundamental question of how the variations in chemical compositions of bitumen affect the overall performance of asphalt pavements remains unanswered.

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

31 32 32 32

Asphalt concrete is often considered as a composite consisting of constituents with lengths ranging from nano to macro scales, as shown in Fig. 1. Continuum mechanics, together with laboratory and field tests at the pavement length scale, are often used to predict the overall performance and in-service life of asphalt pavements. However, material properties at the micro and nano scales are often needed for macroscopic investigations. For example, the rheological property of bitumen, which contributes significantly to the final performance of asphalt pavements, is proven to be related to bitumen structures at the micro and nano scales [1]. Bitumen is a complex mixture of mostly hydrocarbons (greater than 90 wt.%) with different size and polarity, together with some heteroatoms such as sulfur, nitrogen, oxygen, as well as traces of metals (e.g., iron, nickel, and vanadium). Moreover, chemical compositions of bitumen vary with their crude sources. In order to better understand bitumen's chemistry and its rheological behavior, different methods have been developed to separate the molecules in bitumen into different chemical fractions, depending on their molecular size and polarity. Among these separation methods, SARA separation (saturates, aromatics, resins, and asphaltenes, with increasing molecular weight and polarity, as shown in Fig. 2) is frequently used [4]. Asphaltenes are commonly defined as the crude-oil fraction that is soluble in aromatic solvents (e.g., toluene or benzene) but insoluble in aliphatic solvents (e.g., n-heptane, or n-pentane). Fig. 3 shows the typical molecular structures of asphaltenes [5], which are typically large polycyclic–aromatic unit sheets containing heteroatoms, dipoles, and short aliphatic side groups [6]. Asphaltenes are the heaviest component in bitumen, and they have a strong tendency to aggregate due to the attraction between polyaromatic fused rings and precipitate out from an initially stable

microscale (mastic)

mesoscale 2 (asphalt) bitumen filler

asphaltenes maltenes

. . . .

stones (d > 2 mm) mortar voids

mastic nanoscale (bitumen)

sand (d ≤ 2 mm) mesoscale 1 (mortar)

macroscale (continuum)

Fig. 1. Multiscale model for asphalt mixtures, including the bitumen-scale, the mastic-scale (bitumen + filler), the mortar-scale (mastic + fine aggregate with d b 2 mm), the asphalt-scale (mortar + aggregates with d N 2 mm), and the macroscale [after 3].

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

Saturates

Aromacs

Resins

19

Asphaltenes

100%

Weight (%)

80% 60% 40% 20% 0% AAA-1 AAB-1 AAC-1 AAD-1 AAF-1 AAG-1 AAK-1 AAM-1 Fig. 2. SARA fractions for the core SHRP binders [after [7]].

mixture under less favorable conditions (i.e., low temperature). The other three fractions of bitumen, including saturates, aromatics, and resins, is collectively designated as maltenes. Definitions in SARA analysis are strictly operational ways to separate the essentially continuum polydisperse mixture into characteristic finite fractions according to their molar mass, aromatic content, and polarity. The rich chemistry of bitumen may result in a variety of intermolecular associations, which facilitate the formation of diverse microstructures. These structures at various length scales are related to the physical and mechanical properties of bitumen and asphalt concrete

[8]. Nellensteyn [9] and Pefiffer et al. [10] proposed the colloidal structure of bitumen, where the asphaltene micelles are stabilized by polar aromatics in a sea of naphthene aromatics and saturated compounds. The colloidal model helps explain the difference in rheological properties between so-called sol and gel bitumens, in which a colloidal index (CI) is defined as a ratio of the dispersed constituents (the aromatics and resins) over the flocculated constituents (saturates and asphaltenes) of bitumen [11]. A bitumen with a lower CI value (i.e., high asphaltene/ resin ratio) would lead to a network structure with high rigidity and elasticity, called a gel-type bitumen; whereas a sol-type bitumen has a

Fig. 3. Typical asphaltene molecular architecture. Molecules are roughly 750 Da, and there is a single fused aromatic ring system per molecule (island molecular architecture) [after [5]].

20

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

Fig. 4. The modified colloidal structure of bitumen. The asphaltene micelles are pictured as spherical to illustrate the concepts of solvation layer (the resin shell) and effective volume. The oily dispersion medium refers to the maltenes [1] with permission by Elsevier.

higher CI value (i.e., high resin/asphaltene ratio), forming domains of asphaltene micelles which exhibit Newtonian behavior. Between these two extremes, a majority of bitumens has intermediate rheological properties due to a mixed “sol–gel” structure. The asphaltene micelles were further confirmed from SAXS/SANS experiments [12–14]. Using the same techniques, Storm et al. further developed the colloidal model of bitumen by means of a solvation parameter that quantifies the increase of the volume fraction of solid phase due to the absorbed resins [15, 16]. Now, all the bitumens can be considered to have the same structure with variant amount of solvated asphaltenes (Fig. 4). This modified colloidal model allows better understanding of the basic features of bitumen rheology and also more complex phenomena such as bitumen aging, the diffusion of rejuvenating agents into an aged bitumen, and the effects of different modifiers (polymers, acids or mineral fillers) [1]. Although the colloidal model is generally useful for the explanation of rheological properties of most of the paving grade bitumens, it does not consider the waxes and the structural changes they generate, which are thought to be related to the reduced performance of asphalt concrete at low temperatures. Therefore, a better understanding of bitumen structures at the micro and nano scales and their relationships to bitumen chemical compositions is needed before a complete mechanical description can be reached. Recently, atomic force microscopy (AFM) has been explored to characterize bitumen's diverse

Quadrant Photodiode

A B

microstructures and their changes at different temperatures due to its high resolution (down to microns and nanometers) and easy sample preparation [2,3,17–28]. Note that the word microstructure can cause some confusion as to nomenclature. In material science, the term microstructure often refers to the arrangement of phases and defects within the bulk of a material. AFM mainly studies the surface microstructures of a material, which may or may not represent the material's bulk structures. To avoid confusion, we will use the word microstructures throughout this paper to describe the surface morphology of bitumen as observed in AFM images. The first attempt at investigating bitumen's microstructures using AFM was the work of Loeber et al. [21]. They reported rippled microstructures on surface of a thin-film bitumen and coined the term ‘beestructures’ to describe the wavy patterns that resemble the yellow and black stripes of a bumble bee (Fig. 6(a)). Since then, more and more AFM studies on microscopic characterization of bitumen followed [2,3, 17–20,22–28]. These studies indicated that AFM could be a powerful tool for better understanding of bitumens' microscopic morphological and mechanical properties, which will benefit the practice of performance-related design for asphalt pavements and other industrial processing techniques involving asphalt binder and crude oil in general. In this case, questions that draw the researchers' attention include: what are the factors that would affect the development of bitumens' microstructure? What chemical components are responsible for the ‘beestructures’? Are the microstructures observed on the sample surfaces representative of bitumens' bulk structures? What is the correlation between bitumens' microstructures and their physical and mechanical properties? On the other hand, using AFM for binder characterization is nontrivial. Technical challenges, such as tip contamination and the influence of operational conditions on the measured properties are often encountered, and data interpretation of AFM images of the complex binders is not as straightforward as it seemed. A literature review summarizing the reported studies regarding the above significant questions would provide timely guidance for making further progress in this important research field and benefit the effort of applying AFM for studies of other challenging materials (i.e., other chemical fractions derived from crude oil, polymer blends, biomaterials, and nanocomposites) as well. This review paper starts with a brief overview of the AFM technique including the basic principle and the different operating modes, as prior knowledge for understanding its application in microscopic characterization of bituminous materials. In the next section, a general description

Mirror Laser

C D

Sample z

Piezoelectric Scanner

y x

Fig. 5. Schematic illustration of key components in an AFM set-up. A flexible cantilever with a tip at the end is connected with a piezoelectric element. A laser beam is reflected off the back of the cantilever and collected in a photodiode. As the tip is scanning the sample surface, the microscopic morphology and mechanical properties of the sample can be obtained from the laser signal [after [33]].

Table 1 Proposed chemical origins of the bee-structures in asphalt binder. Binder source

Sample preparation method

AFM mode

Imaging temperature

What are the bees?

Loeber [21]

A gel-type binder with unknown crude source Five different binders with different PEN (7–151), crude sources unknown SK-70 binder from Korea Eight SHRP binders from different crude sources, asphaltene-doped binder Twelve SHRP binders from different crude sources and a PEN 85/100 binder with unknown source Eight SHRP binders (same as Pauli ([21])), neutral fractions and maltene fractions A PEN 30/45 binder Eight SHRP binders (same as Pauli (2001)), neutral fractions and maltene fractions, wax-doped binders Four binders from different crude sources with same PEN (70/100) Same as Soenen [35] Three binders from different crude sources

Heat-cast

Contact mode, tapping mode

Room temperature

Asphaltene fractions

Heat-cast

Non-contact mode, pulsed-force mode

Room temperature

Asphaltene fractions

Heat-cast Solution-cast

Tapping mode Contact mode, LFM, tapping mode

Room temperature Room temperature

Asphaltene fractions Asphaltene fractions

Heat-cast

Tapping mode

Room temperature

Heat-cast

Tapping mode

Room temperature

Metallic cations and the size and shape of the molecular planes Wax content

Heat-cast Heat-cast & solution-cast

Tapping mode Tapping mode

Room temperature, thermal cycles (25 °C–160 °C) Room temperature, thermal cycles (25 °C–43 °C)

Wax content The interaction between waxes and the remaining binder components

Heat-cast

Tapping mode

Room temperature, thermal cycles (25 °C–80 °C)

Wax content

Heat-cast Heat-cast

Peak Force QNM mode Tapping mode

Thermal cycles (30 °C–60 °C) Thermal cycles (25 °C–75 °C)

Wax content Asphaltene contents steer the size of the bee-phase; metal content affects the area fraction of the bee-phase; wax is related to a hysteresis effect during the heating-cooling cycle. Wax provides the nucleation centers for the bee-phase; asphaltene fractions work as an affinity to other species such as waxes Wax provides the nucleation centers for the bee-phase; asphaltene fractions work as an affinity to other species such as waxes

Jäger [3] Zhang [28] Pauli [42] Masson [23]

Schmets [44]

de Moraes [39] Pauli [2]

Soenen [35] Das [17] Nahar [25]

Sourty [26]

Two binders from Middle East and European crudes, respectively

Solution-cast

Tapping mode

Room temperature, thermal cycles (25 °C–120 °C)

Fischer [20]

Six binders with different PEN (10–220)

Heat-cast

Tapping mode

Thermal cycles (30 °C–65 °C)

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

Authors

‘PEN’ represents the penetration grade of asphalt binder.

21

22

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

of the diverse microstructures of bitumens from different crude sources is provided, followed by the investigations of the effects of sample preparation methods and thermal history on the development of bitumen microstructures. Further, a comprehensive evaluation of the different opinions regarding the chemical origin of the typical ‘bee-structures’ is presented. Subsequently, the possible mechanisms responsible for the formation of ‘bee-structures’ in bitumen are briefly discussed. The succeeding section addresses whether the microstructures observed on thin films of bitumen samples are also present in the bulk bitumen. Concerns and issues that are related to data acquisition and image analysis for applying AFM into microscopic characterization of bitumen are highlighted throughout the paper and also summarized at the end. We finish with conclusions and remarks for future work.

Bruker, in which the cantilever with resonant frequency of hundreds of kHz oscillates at low frequency (~2 kHz). This mode works similarly to pulsed-force mode AFM [34]. Peak Force Tapping mode allows simultaneous capture of microscopic morphology and mechanical properties of the sample surface. In lateral force microscopy mode, compliant cantilevers are often used to avoid causing damage on the sample surface, and the torsion of the cantilever around its axis is detected as the tip scans the sample surface laterally. The above AFM techniques have been applied into investigating microstructures of thin-film bitumen samples by different research groups (Table 1) [2,3,17–28], and their advantages and disadvantages are summarized in Section 9.

2. AFM techniques

3. General microstructures observed on bitumen surfaces

With the inspiration of invention of scanning tunneling microscope, AFM was developed by Benning and coworkers in 1986 [29]. The basic working principle of AFM is that a sharp tip attached at the end of a cantilever probes the specimen surface, with a laser beam focused at the end of the cantilever reflecting into a photodetector to track the surface topography (Fig. 5). Further, the AFM technique has been advanced with diverse imaging modes for microscopic characterization of phase separation, mechanical properties and other phenomenon of various materials [30–32]. Brief explanations of the different AFM imaging modes are as follows. In contact mode AFM, the tip is in contact with the sample at constant height or constant load during scanning, which is suitable for topographic imaging and mechanical property measurements of stiff, less adhesive samples. In dynamic mode AFM, including non-contact mode and intermittent-contact mode (also called tapping mode™), the cantilever is oscillating at or near its resonance frequency. Because the force between the tip and the sample is usually low, dynamic mode AFM is good for imaging soft, delicate samples. Non-contact mode differs from tapping mode in that the tip does not contact the sample surface. In tapping mode, the oscillating cantilever touches the sample surface at the end of its swing and both topographic and phase contrast images are obtained. Phase contrast image is generated by recording the phase shift between the driving force and the tip response, and therefore contains information about variations of sample mechanical properties. Peak Force Tapping™ mode was recently developed by

Loeber et al. [21] first used AFM to study the microstructures of a geltype bitumen that was supposed to have a higher asphaltene content. Samples for AFM measurements were prepared by the heat-cast method (Section 4) in which a bead of hot binder at 140 °C was dropped on a steel disk support and cooled down to room temperature. AFM cantilevers with spring constants of 0.5 N/m and 50 N/m were used for contact mode and tapping mode, respectively. Bee-like microstructures, of several micrometers in diameter and tens of nanometers in height were observed from the pure binder for the first time from the AFM topographic image (Fig. 6(a)). Interestingly, it was found that these ‘bee-structures’ disappeared after repetitive scanning on the same location of the sample, and a network structure appeared. The network structure consisted of interconnected particles with a diameter of 100 nm, which was also observed in the SEM image of the deoiled sample prepared by leaking out the oil phase using a filter paper as a support for the asphaltene structures (Fig. 6(b)). Similar percolating network structures were also observed by Sourty et al. [26] on AFM tapping mode images of solution-cast thin films (Section 4) of the asphaltene fractions of two binders from different crude sources. Loeber et al. [21] suggested that the disappearance of the bee-structures after several repetitive scans was probably because the AFM tip had swept away the oil phase and consequently the network structures below the oil phase appeared. The network structures were related to the asphaltene and resin particles, but the chemical origin of the bee-structures remained unknown.

Fig. 6. (a) AFM tapping mode image of a gel-type bitumen shows typical bee-structures [21] with permission by John Wiley and Sons. (b) AFM phase contrast image of C7 asphaltene-B15%5000 shows the network structures of asphaltenes [26] with permission by John Wiley and Sons.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

23

Fig. 7. (a) AFM phase contrast image of bitumen ABF with a color contrast of phase angle of ~80°. (b) AFM topographic image of bitumen AAM-1under a vaccum with a color contrast of height of ~10 nm. (c) AFM phase contrast image of an asphalt binder from Petro-Canada at 22 °C under vacuum with a color contrast of phase angle of ~170°. The four different domains were named catanaphase (bee-shaped), periphase (around catanaphase), paraphase (solvent regions) and salphase (high phase contrast spots) as shown. The scan area for all the images are 15 × 15 μm2 [23] with permission by John Wiley and Sons.

Given the increasing publications over the last two decades [2,3,17, 20,22–24,26,27] inspired by Loeber's first exploratory effort of applying AFM for microscopic characterization of bitumens, an important question that is worthy of investigating is the suitability of the AFM technique for studying microscopic morphology of bituminous materials. A round robin study [35] conducted by the RILEM (Réunion Internationale des Laboratoires et Experts des Matériaux, systèmes de construction et ouvrages) technical committee demonstrated that microstructures of four different types of bitumens were repeatable across the different laboratories when the consistency of sample preparation and conditioning procedure were guaranteed. The reproducibility and stability of using AFM for characterizing bitumen microstructures were also confirmed by two other studies [24,36], in which inter-laboratory comparison was employed on the bitumen samples from different crude sources and the samples were either virgin, aged, or rejuvenated. Therefore, AFM is a suitable tool for fast screening and microscopic characterization of bituminous materials. Depending on the residue crude sources from which the bitumens are derived, their thermal history, and also their sample preparation methods, various microstructures are observed on surfaces of bitumen films [2,3,17–28]. Although bee-structures with wavy patterns are relatively more commonly seen in some bitumens, a few other bitumens

with low wax content (i.e., AAA-1, AAG-1) show relatively homogenous morphology without apparent structural features. In addition, microstructures with fine domains, flake-like domains, dendrite structuring, and flower-like domains also appear on surfaces of other bitumens [2, 26,37]. An appropriate demonstration of the diverse microstructures of different bitumens can be found in the work conducted by Masson et al. [23], in which thirteen SHRP binders were imaged using tapping mode AFM. Thin-film binder samples were prepared by the heat-cast method in an effort to maintain the solid-state structure of the binders. Various microstructures were observed on surfaces of thin-film bitumens from different crude sources, but no correlation between the AFM morphology and the SARA fractions was established. Microstructures of the thirteen binders were further classified into three distinct groups. One group showed fine domains down to 0.1 μm (Fig. 7(a)), another displayed flake-like domains of about 1 μm (Fig. 7(b)), and a third group exhibited up to four different domains of different sizes and shapes, similar to Loeber's observation of the bee-structures. The four different domains (Fig. 7(c)) were named as catanaphase (bee-shaped), periphase (around catanaphase), paraphase (solvent regions) and salphase (high phase contrast spots) based on the meaning of these Greek words. On the other hand, the paraphrase can be absent on surfaces of some

Fig. 8. (a) AFM topographic image of 1% Sasobit/YNP binder blend showing dendrite structuring [37] with permission by Elsevier. (b) AFM phase contrast image of a binder derived from Middle East crude exhibiting flower-like domains [26] with permission by John Wiley and Sons.

24

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

bitumens where the bee-shaped microstructures are dispersed in the continuous matrix phase [36,38]. Note that there is no consensus among the asphalt research community on how to name the elliptical bee-structures in bitumen morphology obtained by AFM, and both ‘catanaphase’ and ‘bee-structures’ are popularly used in the publications. In this review paper, we will use the word ‘bee-structures’ to describe the dispersed wavy patterns in bitumen microstructures in general. Dendritic microstructures (Fig. 8(a)) were observed on surfaces of solution-cast thin films of Sasobit wax modified asphalt binders, and these microstructures appeared only dependent on the Sasobit concentration, regardless of the binder sources investigated [37]. In a study conducted by Pauli et al. [2], upon doping 4 wt.% hexacontane into AAG-1 bitumen using solution-cast method, the bee-structures appeared much more clustered together forming dendrite branching and exhibiting a fractal pattern. In addition, flower-like domains with a characteristic lateral size of about 10 μm (Fig. 8(b)) were observed on solution-cast thin film of a binder derived from Middle East crude, as confirmed by bright field optical microscopy and transmission electron microscopy [26]. The diverse microstructures observed on surfaces of bitumen samples using AFM are certainly related to their crude sources. In addition, due to the viscoelastic nature of bitumen and the complicated molecular associations among the chemical compositions, one cannot ignore the effects of sample preparation methods and thermal treatment on the development of bitumen microstructures. For instance, the flower-like structures [26] are probably related to the toluene residue that might be trapped in the thin-film specimen due to incomplete evaporation from solution-cast method. Therefore, the effects

of sample preparation methods and thermal history on formation and evolution of bitumen microstructures are evaluated in the following two sections. 4. The effect of sample preparation methods on bitumen microstructures Heat-cast and solution-cast methods are frequently used for sample preparation of AFM characterization of bituminous materials [2,26,27, 36], and they differ from each other in terms of the temperature and the amount of bitumen required to obtain a smooth surface and whether a solvent is involved or not. At the end, the two methods produce thin films with various thicknesses, which would subsequently affect the development of bitumen microstructures. To make heat-cast thin films [2,3,19,25,36,39], a bead of binder (amount varies depending on the effective area of the substrate) is dropped onto a substrate (microscope slide or steel plate), which is heated for 5 min on a hot plate at about 115 °C, a temperature high enough to melt the binder, but not so high to cause rapid oxidation. Once the binder becomes liquid, it is spread out with a blade or spatula to form a thin film. This hot film is then left on the hot plate undisturbed for an additional 10 min to allow the surface to flow to a smooth and glossy finish. Another practice to obtain heat-cast thin films involves rotating the substrate on which the hot binder is deposited at 5000–8000 r.p.m. on a spin coating device [39,40]. The thin films are then cooled to room temperature, stored in a nitrogen gas chamber to prevent dust deposition and aging and annealed at room temperature for a minimum of 24 h before AFM imaging. The film thickness is usually in the order of

Fig. 9. AFM topographical images (all 40 × 40 μm2) of binder AAK-1 as a function of film thickness: (a) 1 μm film (z-scale = 350 nm), (b) 0.1 μm film (z-scale = 90 nm) and (c) 0.01 μm film (z-scale = 300 nm) [2] with permission by Taylor & Francis.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

25

Fig. 10. AFM phase contrast image of the gel-type bitumen ordered by increasing film thickness, ranging from 170 nm to 10 μm (a–h). The area fraction of the flower-like asphaltene aggregate domain initially increased with the film thickness until it covered up the entire film surface, and then the visible fraction of this domain became smaller as the film thickness further increased [26] with permission by John Wiley and Sons.

micrometers or thicker, depending on the amount of bitumen used and the surface area that the bitumen occupies. For the solution-cast method [2,26,41], (1.000 ± 0.005) g of bitumen is dissolved in (10.00 ± 0.01) ml of toluene (concentration of the solution varies depending on the preferred sample thickness). The sample solution is allowed to stand for approximately 24 h to assure complete dissolution of the bitumen. A thin film of bitumen is obtained by depositing a certain amount of the solution (e.g., 20 μl) on a spinning glass microscope slide in a centrifuge. The solution-cast samples are stored in the nitrogen gas chamber to protect them from dust and aging and help remove toluene in the thin film. Complete evaporation of the solvent residue in the film can be verified by Fourier Transform Infrared Microscopy [42]. Solution casting usually produces bitumen samples thinner than the heat-cast ones. The film thickness varies from nanometers to micrometers, depending on the rotation speed and the concentration of the bitumen solution. Masson et al. [23,43] preferred the heat-cast method for preparing bitumen samples for AFM morphological characterization and argued that the original solid-state structure of the binder can be maintained by this approach. They compared the microscopic features of five bitumens prepared by heat-cast method from their group and those of the same bitumens prepared by solution-cast approach from Pauli's group [42] and claimed that solution-cast films provided fewer structural features compared to heat-cast samples. Pauli et al. [2] studied the effect of film thickness on bitumen's morphology by making solution-cast films with thicknesses of 10, 100, and 1000 nm. As shown in Fig. 9, the 1000 nm and 100 nm images of binder AAK-1 exhibited bee-structures, but the number and size of the structures changed dramatically relative to the film thickness. The 10 nm film had small lenticular structures rather than bee-structures. For the other binder AAM-1, the outlier studied by both Schmets [44] and Pauli [2], the thickness effect on morphology was also obvious but only the thinnest 10 nm sample displayed the bee-structures. The authors also reported that the structuring as seen in AFM images appeared to become essentially constant for solution-cast films thicker than 1000 nm, as would be typically found in asphalt concrete used in pavement construction. Further, the effect of film thickness on microstructural change was attributed to the amount of wax available in the film

and the migration of the wax from the bulk of the film to the surface driven by solvent and thermal gradients. In other words, for a thicker film, more wax accumulated at the sample surface helped the growth of ‘3D’ microstructures; whereas ‘2D’ microstructures were developed on thinner films due to less wax and the restriction of the underlying substrate. The dependence of bitumen microstructures on film thickness was also investigated by Sourty et al. [26]. Thin-films of a gel-bitumen and a sol-bitumen were prepared by solution casting at different rotation speeds from solutions with different concentrations, and the film thickness had a wider range (i.e., 40 nm to 10 μm) than that in Pauli's study (i.e., 10 nm to 1000 nm) [2]. Flower-like and flake-like domains were observed on thinner films (less than hundreds of nanometers) of the gel- and sol-bitumens, respectively. However, the growth behavior of microstructures of the two binders was different as the film thickness increased. For the gel-bitumen, the area fraction of the flower-like domain initially increased with the film thickness until it covered up the entire film surface, and then the visible fraction of this domain became smaller as the film thickness further increased (Fig. 10). Sourty et al. [26] claimed that the growth of the flower-like pattern was due to the diffusion-limited aggregation type-growth of asphaltenes under a strong concentration gradient, in which the process of toluene evaporation would contribute to the development of the flower-like asphaltene aggregations. Whereas the decrease of the area fraction of the flowerlike domain after the film thickness further increased was related to high concentrations (high viscosities), at which the aggregation was more dominated by shear-induced ballistic coagulation, and hence decreased with decreasing rotation speed. For the sol-bitumen, the flakelike domain grew in amount and size with film thickness, but there was no sign of the diffusion-limited aggregation type-growth mechanism as occurred with the gel-bitumen. This was attributed to a low sticking probability, as would be expected for a smaller difference in aromaticity and polarity between the asphaltene and maltene fractions in this sol-bitumen. Debate continues about the effect of sample preparation on bitumens' microstructures. On the one hand, as part of the round robin study [35] conducted by the RILEM technical committee, Pauli's group reported that no evidence was found that the solution-cast films with

26

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

sufficient thickness had significant effect upon the microstructures observed by AFM, and neither solvent nor aging effects were observed. On the other hand, Masson et al. [23,43] reported differences in the observed microstructures of the same bitumens prepared by both heatcast [23] and solution-cast [42] methods. Solution-casting seems to have had more significant effect on the formation of bitumen morphology because the molecular interactions in bitumen are probably changed by the solvent and the evaporation process; whereas the heat-cast approach helps preserve the solid-state morphology of bitumen. Another advantage of the heat-cast method is that the finished films often have a thickness of a few microns, which is more of less close to the film thickness of asphalt binders as used in the pavement construction; whereas solution-cast samples are usually thinner than a micron. Therefore, heat casting is more popular than solution casting, based on a comparison of the numbers of the publications that adopted these two methods (Table 1). When the solvent-cast method is used, one may want to try annealing the thin films at a temperature equivalent to that occurring in the heat-cast approach or applying freezedrying after depositing the bitumen solution onto a substrate, which allows complete evaporation of the solvent and minimizes alteration to molecular interactions. For both sample preparation methods, surface roughness of the thin films should be small enough to eliminate possible artifacts due to uneven tip-sample contact during AFM imaging. To better understand the effects of sample preparation and film thickness on the development of bitumen morphology, AFM characterization of the same bitumen prepared by both heat-cast and solution-cast methods with a range of film thicknesses is recommended. 5. The effect of thermal history on bitumen microstructures With bitumen's viscoelastic nature, its thermal properties such as glass transition temperature (Tg) and enthalpy changes associated with the crystallized fraction have been well characterized by

Differential Scanning Calorimetry (DSC) [45]. Further, the microstructural changes of bitumen as a function of temperature were visualized by using optical thermomicroscopy techniques such as polarized light microscopy and phase contrast microcopy. At room temperature, bitumens with higher crystallized fractions (6.1 wt.%–33.9 wt.%) appeared to have a larger quantity of well-organized domains as compared to the featureless morphology of two bitumens with lower crystallized fractions (0.2 wt.%, 3.2 wt.%). Under thermal cycles in a temperature range of 25 °C–100 °C, the organized domains disappeared upon heating and then reformed upon cooling, and the phenomenon of crystallization was generally reversible. The morphological changes from optical microscopy were correlated with enthalpy changes from the DSC measurements. Therefore, the effect of thermal history on bitumen microstructures is often related to the crystallized fractions in bitumen, which can be paraffin waxes, microcrystalline waxes, or certain aromatics and molecules with polar functional groups, as discussed later (Section 6.3). As compared to optical microscopy, an AFM equipped with a temperature controller offers better resolution of the microstructural changes of bitumen subjected to thermal cycling [2,17,35,39,43]. The morphological evolution of a bitumen thin-film at different temperatures is related to the bitumen crude source and sample preparation method. For instance, dos Santos et al. [38] reported that beestructures evident on surface of a heat-cast bitumen film were absent from its non-annealed film, which was prepared by merely buttering the bitumen at room temperature with no heat treatment. Nevertheless, for most bitumen-related applications, heat treatment is inevitably involved at different stages. Similarly, thermal cycling is often applied to heat-cast and solution-cast bitumen samples, which is the main focus for the following discussions. For waxy bitumens subjected to heating and cooling cycles in the temperature range from room temperature (~25 °C) to 80 °C, the beestructures changed considerably [2], and their evolution was consistent

Fig. 11. AFM topographical images of Bit-C as a function of temperature and corresponding DSC heating curve. The change in the phases with increase in temperature was correlated with the DSC curve, which indicates that the behavior and appearance of the microstructure is very much related to the wax behavior [35] with permission by Springer.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

with Claudy's observations of the reversibility of wax crystallization as a function of temperature. For instance, de Moraes et al. [39] verified that the overall sample morphology was dependent on the thermal history and the analysis temperature. Bee-structures that appeared at room temperatures were completely dissolved into the matrix phase at 70 °C and they started to reappear and grow upon cooling. These microstructural changes on the bitumen surfaces were closely correlated with the DSC test results, consistent with what Soenen, Besamusca et al. [35] reported (Fig. 11). In addition to the above efforts of qualitative comparison of microstructural changes of bitumen at different temperatures, morphologies of bitumens subject to thermal cycles were also evaluated quantitatively [25]. For heat-cast asphalt binders under thermal cycles (25 °C–75 °C), a hysteresis effect between the cooling and heating paths in the temperature range of 40 °C–60 °C was observed. The signature of the hysteresis was the morphological differences in terms of the long axis of the beestructures, the aspect ratio of the elliptical ‘bees’, and the phase fraction. Although the origin of the hysteresis effect remains unclear, the thermal drifting effect from the AFM instrument itself might cause ‘shifting’ of the microstructures and therefore contribute partially to the morphological changes for sequentially acquired images. Thermal drifting is more significant at elevated temperature [24,29,46] and should be calibrated for quantitative comparison of bitumen microstructures as a function of temperature. It was also found that the microstructural features tended to align themselves with increasing temperature; whereas at lower temperatures (25 °C to 45 °C) the domains were almost randomly oriented, a phenomenon that was also noticed by other research groups [47,48]. This effect was most pronounced for binders that possess a lower phase fraction of bee-structures, and it was explained by either the lack of the steric hindrance by neighboring domains or by the viscosity decrease with the increasing temperature. In addition, microstructures of the same bitumen quenched from two different temperatures (90 °C and 180 °C) happened to display different morphologies in terms of the area fraction and the size of the bee-structures. The authors speculated that bitumen may be characterized by a ‘reset temperature’, above which the ‘memory’ of its previous microstructural state would be erased. This phenomenon may provide guidance for the mixing process of bitumen, aggregates, filler and other additives in an asphalt plant. The annealing effects on the microstructural changes of solutioncast bitumen films observed by Sourty et al. [26] are rather complicated and intriguing compared to the heat-cast thin films in Nahar's work [25]. Thin films of both gel- and sol-bitumens were prepared at different solution concentrations and subjected to the same annealing procedure. The thin films were heated at a rate of 20 °C/min up to 60 °C for stage I and 120 °C for stage II, and kept at these respective temperatures for 2 min, followed by cooling down to room temperature for 15–30 min prior to AFM imaging. Microstructural changes of these two binders subjected to the above temperature–time trajectories were quite different. For the gel-bitumen, after a heat treatment to 60 °C, the lowconcentration (5 wt.% of binder in toluene solution) sample showed further aggregation of asphaltenes, and this was explained by the temporary lowering of the viscosity, which facilitates precipitation of asphaltene components. After a second treatment to 120 °C, the aggregated structures broke down into dispersed particles with elongated shape. For the high-concentration (50 wt.%) sample, which already had a dense asphaltene aggregation, the first heat treatment showed no further aggregation, rather some breakdown, followed by a further significant breakdown into elongated particles after heating to 120 °C. For the sol-bitumen from a different crude source, the original flakelike patterns in the low-concentration sample evolved into a coreshell feature after heat treatment to 60 °C; however after further heating to 120 °C and cooling down again, these core-shell patterns disappeared, followed by the appearance of flower-like asphaltene aggregation domains. The major effect of the heat treatments on the high-concentration sol-bitumen is the appearance of the bee-

27

structures with size increasing with the temperature. In this study it remains unknown how the toluene evaporation process during solution-cast sample preparation contributed to the development of the binders' microstructure. As a complementary effort to the above studies that are dedicated to examining bitumen microstructures at temperatures above room temperature, Masson et al. [43] used cryogenic AFM to investigate the bitumen microstructures at low temperatures. Typical imaging temperatures were determined to be 22 °C, − 10 °C, − 27 °C, − 55 °C and −72 °C based on the glass transition temperatures (Tg) of bitumen. At room temperature, the three SHRP bitumens showed catana, peri, and para phases. Given that Tg is governed by chemical composition, the catana, peri, and para phases were assigned to domains rich in asphaltenes, naphthene and polar aromatics, and saturates, respectively. Upon cooling to low temperatures (− 10 to − 30 °C), viscous domains (called the salphase, Fig. 7(c)) appeared in the saturates-rich paraphase and alkyl-aromatics-rich periphase due to phase segregation. These domains retained liquid-like phases at temperatures well below the material's glass transition temperature, which may have important implications for the use of bitumen at low temperatures. Bitumen's microstructures change significantly as a function of temperature, regardless of the crude source and its sample preparation methods. For waxy bitumens, there is an evident correlation between the morphological evolution in AFM images and the enthalpy changes in DSC tests, and the crystallization–dissolution of bee-structures is reversible. In addition, the size and ordering of microstructural features vary as temperature changes, and a hysteresis effect exists around the melting temperature of the waxes. These microstructural variations under different thermal treatments are supposed to be related to changes of bitumen's physical and mechanical properties at large scales [2]. 6. What is the chemical origin of the bee-structures? AFM is a promising tool for investigating three-dimensional, high resolution bitumen morphology at the meso and nano scales, a considerable advance as compared to conventional optical microscopy. The development of microstructures on surfaces of thin-film bitumens is related to its crude source, sample preparation method, and thermal history, as discussed in the previous sections. Among the variety of microstructures, bee-structures are observed for many bitumens. In order to understand the chemical origin of the bee-structures, different research groups have tried various methods, such as adding the potential components or precipitating out the irrelevant fractions. Due to the complex chemistry of bitumen, different conclusions were drawn regarding the chemical origin of the bee-structures, as summarized in Table 1. Firstly, some authors [3,28,49] suggested that the beestructures consist of the most polar fraction in bitumen, namely the asphaltenes; then Masson et al. [23] reported a poor correlation between the area fraction of bee-structures and asphaltene fractions, and instead they observed a good correlation between the area of the beestructures and the metal contents according to the measurements conducted for thirteen SHRP binders; more recently, several research groups [2,17,19,35,39,44] started to attribute the bee-structures mainly to clustering of the non-polar waxes. The detailed efforts on investigating the chemical compositions responsible for the bee-structures are discussed below. 6.1. The asphaltenes Pauli et al. [42] studied eight SHRP binders that contain various amounts of asphaltenes and waxes using different AFM techniques such as lateral force microscopy (LFM), contact mode and tapping mode. Bee-structures as reported by Loeber [21] were observed in the two binders (AAD-1 and AAK-1) having the largest amounts of the asphaltene fraction (~20 wt.%). For both binders, bee-structures were identifiable in both LFM and contact mode images. For binder AAD-1,

28

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

Fig. 12. AFM topographic image of (a) a virgin SK-70 bitumen supplied by SK Corp., Korea and (b) the same bitumen without asphaltenes (both 15 × 15 μm2 area). The color contrast covers a height variation of ∼120 nm [28] with permission by John Wiley and Sons.

bee-structures from the topographic and phase contrast images using tapping mode had a higher resolution compared to those obtained from LFM and contact mode. On further inspection of a 5 wt.% asphaltene-doped binder AAD-1, smaller and increased numbers of bee-structures were observed in the phase contrast image compared to that of the original binder. These results, considered along with the findings from contact and friction modes, seemed to further support the hypothesis that the bee-structures might be associated with the binders' asphaltene content. However, the observation about the asphaltene-doped sample remains controversial because the authors ignored how the scan size for the doped sample was twice of the original binder, which may cause the visual increase of the number of the beestructures and decrease of the size of the bee-structures. Keeping the scan size the same or using standard image analysis tools for images with different scan sizes might be helpful to make a meaningful comparison. In addition, if the assumption about the correlation between the asphaltenes and the bee-structures holds, the other two binders (AAA-1 and AAB-1) that contain comparable total amount of asphaltenes and waxes as binders AAD-1 and AAK-1 are expected to show bee-structures as well. However, no distinguishing features appeared on surfaces of these two binders, which was attributed to the low viscosity of binders AAA-1 and AAB-1 compared to the other two binders. Jäger [3] and Zhang [28] adopted a different approach to assess the chemical composition of the bee-structures observed in AFM images. Instead of adding asphaltenes into the neat binder, both groups chose the alternative of separating the binders into their asphaltene and maltene fractions using n-heptane. In Jäger's study, the appearance of the beestructures in the original binder was confirmed by both non-contact mode AFM and reflected light microscopy. However, the beestructures disappeared on reflected-light-microscope image of the maltene fractions. Therefore, the authors concluded that asphaltenes might be related to the formation of the bee-structures. For a SK-70 bitumen with 10.58 wt.% of asphaltenes and 1.86 wt.% of wax content examined in Zhang's work, typical bee-structures were observed on the original neat bitumen; however, the topographic image of the maltene fractions that contain no asphaltenes did not show beestructures (Fig. 12). Based on this comparison, the appearance of the bee-structures was believed to be caused by the existence of asphaltenes in the bitumen. However, the authors further proposed that the bee-structures can be related to the wax contents as well, considering how the waxes can coprecipitate with asphaltenes during the deasphaltening process [50]. In this case, it is difficult to differentiate the contribution of the asphaltene and the waxes to the formation of the bee-structures.

6.2. The metal content At this point, it is tempting to attribute the bee-structures to associations of asphaltene components in bitumen, but Masson's work did not support this assessment [23]. For the thirteen SHRP binders imaged using tapping mode AFM, diverse microstructures appeared on the heat-cast thin films depending on their crude sources (Section 3). Interestingly, for bitumens that showed bee-structures, no correlation was found between the area of the bee-structures and the asphaltene content. Instead, a strong correlation existed between the area of the beestructures and the mass percent of vanadium and nickel metals in the binders. Therefore, they concluded that the bee-structures on bitumen surfaces were governed by the polarity defined by the metallic cations. However, the above observation seems to be inconsistent with how most of the metal contents (nickel and vanadium) in bitumen would be expected to be associated with the highly polar asphaltene fraction [51–53]. For instance, Pauli et al. [2] suggested that the presence of bee-structures in the asphaltene-free bitumen fractions (saturates and aromatics) indicated the irrelevance of asphaltenes and metal contents on development of the bee-structures because the removal of asphaltenes during separation should constitute an indirect removal of most of the metals. Therefore, the effect of metal content on bitumen microstructures remains controversial.

6.3. The wax crystallization Waxes in bitumen can come from different sources and with different chemical origins. Some waxes are originally present in crude oils and retained after the distillation, while others are added into bitumen as a modifier to reduce its viscosity for mixing and also for construction compaction purposes. The wax types usually present in asphalt and heavy oil residuals include paraffin wax, microcrystalline wax, and certain aromatics and molecules with polar functional groups which may crystallize upon cooling [54]. Paraffin wax refers to the group of n-alkanes with few or no branches that tends to crystallize in large and flat plates or needles; microcrystalline wax mainly consists of naphthenes and iso-alkanes and tends to form tiny microscopic needles [55]. DSC is often used to determine the wax content and wax crystallization and dissolution in bitumen [17,35,48]. However, wax content calculated from the endothermic peaks of DSC heating and cooling curves can be different, depending on the annealing procedure and heating/cooling rate, which has caused problems for many years [35,56]. Therefore, wax contents of different binders should be determined with a specific set of DSC operational parameters for meaningful comparison.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

The effect of waxes on bitumen quality and asphalt performance at the macro scale has been carried out by several groups [56–59]. Waxes in bitumen influence bitumen properties to a major or minor extent, determined by many factors including chemical compositions and rheological properties of bitumen, amount of wax in the bitumen, as well as chemical compositions and crystalline structure of the wax. Although wax can give rise to positive effects on bitumen, such as improved resistance to rutting at higher temperatures, it has normally been considered as a negative indication of the quality of the bitumen. For instance, waxy materials have long been associated with lowtemperature properties of asphalt pavements such as increased sensitivity to cracking [56]. Consequently, many bitumen specifications include requirements limiting the maximum wax content. Lesuer [1] suggested a complete account of bitumen rheology would need to incorporate a detailed description of the waxes, their corresponding structure changes and their interaction with other bitumen fractions, especially asphaltenes and resins. That the waxes play a direct role in the formation of bitumen microstructures was confirmed when the correlation between morphological changes obtained from optical microscopy (i.e., polarized light microscopy and phase contrast microcopy) and the thermal effects observed in DSC experiments was established by Claudy et al. [45]. Later Lu et al. [47] employed polarized light microscopy, confocal laserscanning microscopy, and the freeze fracture technique in combination with transmission electron microscopy to study the consequences of wax crystallization on bitumen structures. They found that wax crystals displayed different geometries (i.e., elongated, crescent-like, and flakes) with typical sizes of order of 1–10 μm depending on the crude source and crystallization conditions. In studies where AFM was applied for morphological characterization of waxy bitumen [2,17,35,39], the correlation between the microstructural changes in waxy bitumen and the enthalpy peaks in DSC curves was verified, in favor of the hypothesis that the wax contents in bitumen are associated with the formation of bee-structures. Another piece of evidence to support the association of the beestructures with the waxes is that bitumens with various amounts of crystalline materials showed different surface morphologies [2]. Two (AAA-1 and AAG-1) of the eight SHRP binders that have very low total crystalline material content (b0.2 wt.%) exhibited relatively smooth surfaces; whereas other binders with at least 1 wt.% of crystalline material showed bee-structures. However, it remains unclear why a relative difference in crystalline material content among different bitumens would cause significant difference in bitumens' microstructures. In addition, it is challenging to explain the relationship between the low mass weight of the wax content and the large area fraction of beestructures that are evident on bitumen surfaces. This question will be discussed in detail in Section 8. Effects of wax on bitumen morphology depend on factors such as chemical composition of the bitumen, the nature of the wax, and the asphaltene/wax interactions. A series of experiments was conducted to image the maltene components (including the maltenes, saturates, and aromatics) derived from dissolution/filtration techniques and the neutral fractions generated by ion exchange chromatography [2,44]. It is assumed that both the maltene fractions and the neutral components obtained from the above separation methods consist of a majority of the wax although not asphaltenes. According to Schmets' investigation [44], the bee-structures observed in the original bitumens (AAK-1 and AAC-1) appeared in their maltene fractions as well. Pauli et al. [2] looked into the morphologies of the sub-fractions of maltene phase (saturates and aromatics) before and after waxes were separated. Both saturates and aromatics exhibited bee-structures prior to wax separation whereas the waxfree saturates and aromatics did not show this type of structuring. Therefore, both groups believed that the waxes rather than the aphaltenes, which were precipitated out beforehand, are involved in the formation of bee-structures.

29

Different from the above studies focusing on the neat waxy bitumens (wax content greater than 1 wt.%), Pauli et al. [2] chose two base bitumens (AAA-1 and AAG-1) that exhibited little surface structuring due to their low wax content (b0.2 wt.%) and doped them with different model waxes at various concentrations. The model waxes used for doping included a microcrystalline wax (IGI 5788A) and four normal paraffin waxes (octacosane (C28H58), tetratetracontane (C44H90), pentacontane (C50H102), hexacontane (C60H122)). Two sets of wax-doping experiments were carried out. In the first one, the base bitumens were doped with different waxes at the same concentration (2 wt.%); and the other set involved doping the two bitumens with the same paraffin wax (hexacontane) at different concentrations (1 wt.%, 2 wt.%, 4 wt.%, and 6 wt.%). In all cases, the addition of paraffin or microcrystalline waxes resulted in the formation of bee-structures that are not typically observed in the base bitumens. In particular, binders doped with different types of wax at the same concentration showed different effects on the surface structures; different surface structuring was also observed when the same wax was doped into two different binders; the size of the bee-structures in wax-doped samples increased as the wax concentration increased and the beestructures appeared much more clustered together at higher concentrations. On the other hand, the two microcrystalline wax-doped samples of AAA-1 and AAG-1 exhibited similar morphology (i.e., the shape and size of the bee-structures), independent of their crude source. This might indicate that crystalline wax and paraffin wax modify bitumen's properties through different mechanisms because of the intrinsic differences in their chemistry and crystallization behavior. Therefore, AFM characterization demonstrated that microstructures of wax-doped bitumens depend on the wax concentration, wax type and the crude source of the base bitumens, consistent with the observations from Lu et al. [47], who used different techniques such as polarized light microscopy and confocal laser scanning microscopy. Based on the morphological characterization of wax-doped bitumens, the authors suggested that the interaction between crystalizing waxes and the remaining asphalt fractions is responsible for much of the structuring, including the wellknown bee-structures. In summary, as to the chemical origin of the bee-structures on surfaces of thin-film bitumens, very discrepant views are held due to bitumens' rich chemistry. The association between the trace metals and the asphaltene fraction due to their similar polarity makes it difficult for one to differentiate the effects from the asphaltenes and from the metal content on bitumen microstructures. The interactions between the waxes and the asphaltenes are even more complicated. This can be seen from the contradictory observations of morphologies of the bitumen maltene phase obtained from the deasphalteneing process. With the appearance of the bee-structures on surfaces of the original bitumens from different crude sources, the similar rippled patterns were reported to be absent [3,28] or present [2,44] in the maltene phase of bitumen. Understanding of where the waxes are prone to stay in the bitumen fractions during the separation process would help resolve the above contradictory opinions. Waxes can dissolve in the bitumen maltene phase at some temperatures, but they can also coprecipitate with asphaltenes during the deasphaltening process, generating waxy asphaltenes and black waxes [1,50]. A direct observation of a cross-contamination of the wax and the apshaltenes is from AFM tapping mode images of the separated asphaltene fraction where the stepwise features belonging to the wax content nucleated at the asphaltene aggregate domains [26]. Therefore, the absence or presence of the bee-structures in the bitumen maltene phase probably depends on the association of waxes with other bitumen fractions. When flocculated asphaltenes provide wax crystallization sites, little or no wax is left in the maltenes, and therefore bee-structures barely appear on maltene morphology. In contrast, the maltene phase exhibiting bee-structures might contain more waxes that do not coprecipitate with the asphaltenes. In addition, the solvent used for the deasphalteneing process would also affect the asphaltene–wax interatctions [50], causing variations in the

30

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

separated fractions, and consequently the contentious morphological observations. As a result of the complicated interactions among the different chemical components, it is still a challenging task to tie the beestructures with specific chemical components in bitumen. A study conducted by Nahar et al. [25] reported that the concentration of asphaltenes seemed to affect the size of the bee-structures; the difference of the area fraction of the ‘bees’ might result from the difference of the total metal content (being the sum of the vanadium, nickel and iron content) among the binders; and a hysteresis effect observed between the heating and cooling paths in the temperature range of 40 °C to 60 °C was attributed to the wax melting and crystallization. However, only three binders were investigated and they might not be representative for various types of binders, therefore no rigorous statistical statement can be clearly drawn from this study. Among the different opinions as discussed above, it seems there is more evidence in favor of the hypothesis that the waxes present in bitumen contribute significantly to the formation of the bee-structures on bitumen surfaces. The most relevant proof of this hypothesis is the good correspondence between microstructure evolutions under thermal cycles with the peaks changes in DSC curves. On the other hand, one cannot exclude the effect of asphaltene fraction or metal content on the development of the bee-structures completely. The asphaltenes can provide crystallization sites for wax, and trace metals are often associated with asphaltene fraction, which modify the interaction between crystallizing waxes and the other bitumen fractions. Therefore, it is more reasonable to say that the complicated molecular interactions among the different chemical components in bitumen rather than a single chemical fraction are responsible for the diverse microstructures of the bitumen, including the bee-structures. On the other hand, further separation of the metal contents and the waxes from the asphaltene fraction and more accurate assessment of each of these components and their interactions would provide more insight for better understanding of the chemical origin of bitumen microstructures.

rectangular single crystals grow in a radical pattern extending from a central point. The narrow periodic undulations along the long axis through the center of each crystal ‘arm’ resemble the appearance of bee-structures. Phase field modeling revealed that these sectorized ripples across the long lamellar growth front were formed due to curvature elasticity coupled with strain recovery deformation. In addition, sectorized 3D chair-conformation morphologies were observed in the polyethylene single crystals, and the authors believed that these crystals were formed as a result of the development of screw dislocations during crystal growth [64]. Inspired by the theory of surface wrinkling observed for some polymers [65,66], Nahar et al. [67] pointed out that wrinkles of the beestructures could originate from a mismatch in coefficient of thermal expansion of the different phases in the microstructures. Upon cooling, stresses lead to buckling of the microstructures—the wavy pattern of the ‘bees’, and details of this pattern may eventually be translated into mechanical properties. Similarly, Lyne et al. [22] applied a laminate wrinkling model to explain the formation of the bee-structures, in which the bees and their surrounding areas were considered as one single bee laminate phase resting on a flat matrix layer. Under compressive loading, as during cooling of the bitumen from a melt, the stiffer bee laminate contracts less than the less stiff matrix phase and surface wrinkling is likely to occur (Fig. 13). This model is essentially similar to the lamellar buckling effect as predicted by ‘curvature elasticity’ theory [61]. The above mechanisms originating from the polymer field might explain the formation of bee-structures observed on bitumen surfaces; however, neither of them can fully reconcile the seemingly contradictory fact: the wax at low mass weight has resulted in the large area fraction of bee-structures that are evident on the surfaces of thin film bitumen. To resolve this concern, investigation into whether the beestructures observed on surfaces of thin-film bitumen are a consequence of migration of the waxes from the bulk bitumen under a thermal gradient or if they exist in the bulk is therefore of great relevance.

7. Possible mechanisms responsible for the bee-structures

8. Are the microstructures observed on bitumen surfaces present in the bulk bitumen?

Experiments conducted to date strongly suggest that the interactions between crystallizing waxes and other bitumen fractions lead to the microstructures observed on the surfaces of thin-film bitumen samples. However, the exact mechanism responsible for the formation of the bee-structures on bitumen surfaces is not fully understood yet. In view of the similarity between the rippled structures observed in polymer crystals and the bee-structures displayed on thin-film bitumen surfaces, mechanisms such as ‘curvature elasticity’, ‘screw dislocations’, and ‘surface wrinkling’ involved in the development of the rippled morphology of polymer systems might shed some light for understanding the growth of bitumen microstructures. The rippling patterns of the bee-structures might be related to theoretical elucidation that explains the single-crystal growth in polymer liquid crystalline systems [60–63]. Surface morphology of single crystals of polyethylene and syndiotactic polypropylene was investigated by Transmission Electron Microscopy or AFM imaging. Elongated

In applications related to bituminous materials, damage and failure of the structures under cyclic loading such as traffic and temperature on a macroscopic scale is believed to be linked to the bitumen's microstructures [1,68]. The valleys of the bee-structures as well as the interfaces between the dispersed and the continuous phases in bitumen's microstructures tend to be the weak points where micro cracks initiate, and they are inclined to grow under loading [38,68]. Therefore, a fundamental understanding of the presence and distribution of bitumen's microstructures should lead to better prediction of bitumen's mechanical properties in various applications. Lyne et al. [22] proposed that the chemical components that are responsible for the bee-structures might be separated from the bulk and transported to the surface of the binder thin film, similar to the bloom formation in chocolate due to the cocoa butter separating and migrating to the surface as a white haze [69]. In a system with multiple

λ 2A L

h

L- ∆L

Fig. 13. Schematic of surface wrinkling. The formation of bee-structures was explained by a surface wrinkling phenomenon in which the stiff thin bee laminate layer rests on a less stiff matrix layer. Under compressive loading, as during cooling of the bitumen from a melt, the stiffer material will contract less than the less stiff material when the material cools down, and surface wrinkling is likely to occur [22] with permission by Springer.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

phases, the interfaces among different phases are often associated with an excess interfacial energy, and the system prefers to lower its free energy by phase separation, which causes the phase with the lowest surface energy and density enriched at the surface of the composite material. Pauli et al. [2] had a similar explanation for the apparent differences in surface structuring as the thickness of the thin film samples was decreased (Section 4). The hypothesis was the wax contents in the bulk of the film diffused to the film surface under thermal convection during sample preparation, and they finally preferentially accumulated at the surface due to the interfacial effects. On the other hand, some other research groups argued that the principle microstructural features were not just entirely a surficial phenomenon. Schmets et al. [44] applied SANS to detect the wax distribution in the bulk bitumen, and demonstrated that the wax was present throughout the bitumen bulk phase. Qin's study [37] on Sasobit modified asphalt binders also suggested that the network structures appeared after the introduction of the Sasobit wax was developed not only on the surface of the sample as depicted by AFM images but also inside the bulk as implied by rheological tests. In Fischer's work [20], a thinfilm of asphalt binder was prepared by turning upside down the bead of binder extracted from a large binder reservoir; in this way, the actual surface to be analyzed can be considered as a cross section of the bulk. Microstructures from the sample representing the bulk of the binder were similar to the normally prepared samples, from which the author believed that the microstructures may develop throughout the bulky material. However, it was possible that the surface microstructures of the bulk sample had developed during the short period of heating treatment in order to obtain a relatively flat surface when the sample was prepared. In a follow-up study by the same group [18], a sandwiched sample was prepared by applying another steel substrate disk on top of the heat-cast thin film, followed by annealing at room temperature for at least 24 h. A freeze-fracture surface was generated by removing one of the substrate disks after the sample was stored in a freezer at −20 °C for 2 h, and imaged immediately under AFM before a substantial thermal energy from the ambient environment can be accumulated to the sample surface. In such a way, the fracture surfaces had not been in contact with air before imaging, which might help reduce the effects of surface mobility and surface tension as occurred in the conventional heat-cast approach. AFM phase contrast images of the thin-film bitumen exhibited two distinct phases, which were also observed on the planar view (depth of ~ 500 μm) from scanning acoustic microscopy. The authors therefore believed that bitumen is not homogenous in the bulk, and the observed surface microstructures do reveal some of the bulk features as well. It remains inconclusive whether the microstructures associated with the wax crystallites are only present on surfaces of bitumen thin-films or if they exist throughout the bulk phase of bitumen. As the heterogeneity of bitumen microstructures can cause stress concentration and non-homogeneous straining along microstructures' interfaces [68], answers to the above question are of great significance. 9. Technical challenges and recommendations Bitumen rheology is a determining factor for good performance of asphalt pavements and other applications. The macroscopic rheological behavior of bitumen is largely related to its diverse microstructures at the micro and nano scales. Studies conducted to date have shown that AFM is a suitable tool for characterization of microscopic morphology of bitumens from different crude sources and their microstructural changes as a function of temperature with reasonable repeatability when sample preparation and the instrumental operating conditions are consistent. Observations from AFM measurements can provide supplementary information for developing more accurate structure-related models mathematically describing bitumen rheology as a function of its composition. Although AFM is superior to other techniques such as optical microscopy, confocal laser-scanning microscopy, and scanning electron

31

microscopy because of its high resolution and easy sample preparation, technical challenges and concerns related to instrumentation and interpretation of results for bitumen samples should be borne in mind. Firstly, for sample preparation, the conditioning environment during sample preparation (i.e., thermal annealing procedure, sample storage time) and the imaging parameters (imaging mode, tip and cantilever properties, drive frequency, set point, and temperature, etc.) dramatically influence bitumen microstructures [24,27,70] and should always be reported together with the AFM datasets. In addition, attention should be paid to obtain a smooth enough surface to eliminate surface roughness effects [27,71,72]. During imaging, tip contamination is a common problem for morphological characterization of bitumen at room temperature and above [3,21,27,42,73]. This issue is mainly caused by the adhesive nature of bitumen at intermediate and high temperatures. For instance, tapping mode phase contrast images clearly showed streaks on surface of binder AAD-1 in Pauli's study [49], and the authors believed that the tip had collected some materials from the viscous phase of the surface. Once the tip is contaminated with binder residue, the resolution for the subsequent scanning is probably reduced. To identify the tip contamination, Pauli et al. [49] observed an increase in cantilever mass after checking the free harmonic oscillation of the cantilever before and in-between scans. However, this method is only sensitive to a large amount of residue picked up by the tip. The study by Yu et al. [27] suggested that an adhesion standard provided a more sensitive means of inferring tip contamination. The adhesion standard was made of a glass slide coated with a thin layer of inert, hydrophobic chemical. By observing force curves on the adhesion standard before and after the adhesion measurement of the sample of interest, a significant change of measured adhesion force indicates tip contamination or change in tip size, shape or functionality. Imaging the thin-film bitumen samples in clean deionized water might help minimize tip contamination due to reduced tip-sample interaction force as long as the bitumen microstructures remain stable over a reasonable time period with no moisture damage. In addition, data acquisition using AFM should be performed in a scientific manner that generally applies to each type of samples. The scan size should be large enough to obtain representative surface features [24]. More importantly, statistical measurements have to be considered for AFM tests (repetitions and locations for data acquisition) and the measurement uncertainty (the standard deviation and standard deviation of the mean) should be reported; otherwise, the following analysis regarding microstructural parameters (i.e., number and size of the beestructures, distribution and orientation of the different micro-phases) among different samples will be less representative or meaningful [70, 72]. Post-processing of the AFM images is nontrivial for data interpretation. Raw images acquired from AFM are given with auto-scaled scale bar by default, and images of different samples often have different scale bars because of the dimensional difference of the sample features. Direct comparison of features from images with different scale bars can be misleading [2,20,35,44]. Therefore, resetting the scale bars of the different images to be the same is necessary for easier comparison of the microstructures. In addition, the rich microstructures of bitumen render visual comparison of the microstructural parameters difficult and unreliable [28,41,42]. In this case, professional image processing tool and mathematical models are recommended for quantitative analysis [19, 24,67]. To make an overall assessment of the various AFM techniques that have been applied in the available publications, they all have advantages and disadvantages for microscopic characterization of bituminous material. Contact mode and lateral force microscopy mode can be used for both morphological characterization and mechanical property measurements. However, these two modes are more likely subject to tip contamination; therefore they are not commonly used currently. Yu et al. [27] reported that contact mode AFM could not generate consistent images of solution-cast samples under ambient conditions because

32

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33

of the inherent stickiness of bitumens as compared to other materials. PeakForce Tapping mode was developed for simultaneous measurement of both morphology and mechanical properties, but more work needs to be done for applying the appropriate contact mechanic models to infer modulus of bituminous materials because of their viscoelastic nature [74]. Tapping mode is being more frequently used for microscopic characterization of bitumen because it allows simultaneous recording of surface topography and phase contrast of the material. However, the contrast in the phase image is associated with the combined effects of adhesion, stiffness, damping, and topographic features of the sample, which makes the phase image difficult to interpret [75]. For example, the phase contrasts of some bitumens turned out to be inverted [14, 25]. Although Masson et al. [23] attributed this phase contrast inversion to steric hardening or the slow oxidation of the bitumen surface in contact with air, it also can be an artifact from relative compressions of the micro phases in bitumen's microstructures, similar to the effect in Behrend's work [75]. Therefore, one should be very careful about the operational parameters in data acquisition and further interpretation of the phase contrast image. In addition, comparison of the microstructural changes among different binders under various treatments (temperature, pressure, and oxygen) would be better performed on the topographic image rather than the phase contrast image due to the combined effects of topographic features and mechanical property differences [20,24,67]. Moreover, even though AFM is suitable for morphological characterization of bitumen, it is limited in terms of investigating the chemical composition over the different micron-sized domains in bitumen microstructures. Combining AFM with other chemical analysis tools such as Fourier Transform Infrared Spectroscopy, X-ray diffraction, and Raman microscopy would provide complementary data. In addition, morphological investigation of bitumen thin-films only provides limited information, and more research should be conducted on characterization of bitumen's mechanical properties at the micro and nano scales, in order to establish the chemical–mechanical links for optimal design of bituminous materials. Efforts in using AFM to measure bitumen's mechanical properties such as adhesion, modulus, viscoelasticity, and friction are ongoing [3,19,27,70,76].

of great importance because the heterogeneity of bitumen microstructure governs the cracking mechanism in bitumen-related applications. On the other hand, one should keep in mind that AFM mainly studies surface microstructures of bitumen, which do not necessarily represent the three dimensional arrangement of the different phases within the bulk bitumen. In addition, more attention should be paid to resolving technical challenges associated with AFM characterization of bituminous materials, such as sample preparation, tip contamination, data acquisition, and image interpretation (Section 9). Many questions remain in the knowledge gap of the chemical–mechanical link of bitumen. For one, due to the complex chemistry of bitumen and the limitation of AFM in chemical composition characterization, it is still premature to assign the different morphological domains to certain bitumen chemical components (i.e., the SARA fractions). Better understanding of the associations of the asphaltenes with the metal traces and the waxes and further separation of these fractions would help identify the chemical origin of bitumen microstructures. Combining AFM with other chemical analysis tools that can generate comparable high resolution to AFM would also present an avenue to linking the chemistry to microstructures of bitumen. Secondly, even though bee-structures were observed in many bitumens and are related to the waxes, whether this bee-phase is relevant to binders' mechanical properties remains elusive. Publications examining the effect of waxes on macroscopic mechanical properties of bitumen may provide some insights for microscopic investigation of the effect of the waxes on bitumen rheology. In addition, the potential of applying AFM for characterization of bitumen microscopic mechanical properties should be fully explored, and the results should be compared with other nanoscale measurements such as nanoindentation or macroscale bulk measurements such as dynamic mechanical analysis. Consequently, the relationships among microstructures, and the mechanical and chemical properties of bitumen could be established, and a structure-related model of the mechanical properties of asphalt binder could therefore be developed, which would help engineer bitumen for a given application in a cost-effective and more sustainable manner. Acknowledgment

10 . . Conclusions Bitumen, found in natural deposits or a refined petroleum product, is primarily used as a glue binding the mineral aggregates together to form asphalt concrete. Hundreds of thousands of molecular species, ranging from non-polar hydrocarbon waxes (n-alkanes) to aromatic, heteroatom-containing polar molecules, exist within any particular bitumen. The rich chemistry of asphalt binder results in complex intermolecular interactions and interesting microstructures at the micro scale, which are closely related to bitumen's rheology. AFM has been proven to be a powerful tool for microscopic characterization of bitumen's morphology. Diverse microstructures (e.g., bee-structures, fine domains, flakelike domains, and dendrite structures, and flower-like domains) are observed on bitumen surfaces. Microstructures of bitumens develop to different degrees depending on their crude sources, sample preparation method, and thermal history. The identification of the chemical origin of the typical bee-structures has been evolving and more evidence points to the interaction between crystallizing waxes and the remaining non-wax chemical components in bitumen. However, the effects of asphaltenes and metal traces on bitumen microstructures cannot be excluded because of the complicated intermolecular associations in bitumen (i.e., coprecipitation of wax and/or metalloporphyrins in asphaltenes). ‘Surface wrinkling’ theory responsible for the rippled structures observed in polymer crystals might explain the formation of bee-structures on surfaces of thin-film bitumen. The question whether the microstructures observed on bitumen surfaces exist in bitumen's bulk remains to be addressed even though it is

The authors would like to thank Manfred Partl, Lily Poulikakos, and Salomé dos Santos in the Laboratory for Road Engineering/Sealing Components in EMPA Swiss Federal Laboratories for Materials Science and Technology for their great discussions regarding this review paper. References [1] Lesueur D. The colloidal structure of bitumen: consequences on the rheology and on the mechanisms of bitumen modification. Adv Colloid Interface Sci 2009;145(1–2): 42–82. [2] Pauli AT, Grimes RW, Beemer AG, Turner TF, Branthaver JF. Morphology of asphalts, asphalt fractions and model wax-doped asphalts studied by atomic force microscopy. Int J Pavement Eng 2011;12(4):291–309. [3] Jäger A, Lackner R, Eisenmenger-Sittner C, Blab R. Identification of four material phases in bitumen by atomic force microscopy. Road Mater Pavement Design 2004;5(Suppl. 1):9–24. [4] Corbett LW. Composition of asphalt based on generic fractionation, using solvent deasphaltening, elution–adsorption chromatography, and densimetric characterization. Anal Chem 1969;41(4):576–9. [5] Mullins OC. The asphaltenes. Annu Rev Anal Chem 2011;4:393–418. [6] Yen TF, Chilingarian GV. Asphaltenes and asphalts, 2, 2. Elsevier; 2000. [7] Mortazavi M, Moulthrop JS. The SHRP materials reference library. Strategic Highway Research Program, National Research Council; 1993. [8] Robertson RE. Chemical properties of asphalts and their relationship to pavement performance. Strategic Highway Research Program, National Research Council; 1991. [9] Nellensteyn F. The constitution of asphalt. J Inst Pet Technol 1924;10:311–23. [10] Pfeiffer JP, Saal R. Asphaltic bitumen as colloid system. J Phys Chem 1940;44(2): 139–49. [11] Loeber L, Muller G, Morel J, Sutton O. Bitumen in colloid science: a chemical, structural and rheological approach. Fuel 1998;77(13):1443–50. [12] Bardon C, Barre L, Espinat D, Guille V, Li MH, Lambard J, et al. The colloidal structure of crude oils and suspensions of asphaltenes and resins. Fuel Sci Technol Int 1996; 14(1–2):203–42.

X. Yu et al. / Advances in Colloid and Interface Science 218 (2015) 17–33 [13] Dwiggins Jr C. A small angle X-ray scattering study of the colloidal nature of petroleum. J Phys Chem 1965;69(10):3500–6. [14] Ravey J, Ducouret G, Espinat D. Asphaltene macrostructure by small angle neutron scattering. Fuel 1988;67(11):1560–7. [15] Storm DA, Baressi RJ, Sheu EY. Rheological study of Ratawi vacuum residue in the 298–673 K temperature range. Energy Fuel 1995;9(1):168–76. [16] Storm DA, Sheu EY, DeTar MM. Macrostructure of asphaltenes in vacuum residue by small-angle X-ray scattering. Fuel 1993;72(7):977–81. [17] Das PK, Kringos N, Wallqvist V, Birgisson B. Micromechanical investigation of phase separation in bitumen by combining atomic force microscopy with differential scanning calorimetry results. Road Mater Pavement Design 2013;14(Suppl. 1):25–37. [18] Fischer H, Dillingh E. On the investigation of the bulk microstructure of bitumen—introducing two new techniques. Fuel 2014;118:365–8. [19] Fischer H, Stadler H, Erina N. Quantitative temperature-depending mapping of mechanical properties of bitumen at the nanoscale using the AFM operated with PeakForce Tapping mode. J Microsc 2013;250(3):210–7. [20] Fischer HR, Dillingh E, Hermse C. On the microstructure of bituminous binders. Road Mater Pavement Design 2013:1–15 [(ahead-of-print)]. [21] Loeber L, Sutton O, Morel J, Valleton JM, Muller G. New direct observations of asphalts and asphalt binders by scanning electron microscopy and atomic force microscopy. J Microsc 1996;182(1):32–9. [22] Lyne ÅL, Wallqvist V, Rutland MW, Claesson P, Birgisson B. Surface wrinkling: the phenomenon causing bees in bitumen. J Mater Sci 2013;48(20):6970–6. [23] Masson JF, Leblond V, Margeson J. Bitumen morphologies by phase‐detection atomic force microscopy. J Microsc 2006;221(1):17–29. [24] Nahar S, Dillingh B, Erkens S, Schmets AJ, Fischer H, Scarpas A, et al. Is atomic force microscopy suited as tool for fast screening of bituminous materials? Interlaboratory comparison study. Transportation Research Board 92nd Annual Meeting; 2013. [25] Nahar SN, Schmets AJM, Scarpas A, Schitter G. Temperature and thermal history dependence of the microstructure in bituminous materials. Eur Polym J 2013;49(8): 1964–74. [26] Sourty E, Tamminga A, Michels M, VELLINGA WP, Meijer H. The microstructure of petroleum vacuum residue films for bituminous concrete: a microscopy approach. J Microsc 2011;241(2):132–46. [27] Yu X, Burnham NA, Mallick RB, Tao M. A systematic AFM-based method to measure adhesion differences between micron-sized domains in asphalt binders. Fuel 2013; 113:443–7. [28] Zhang HL, Wang HC, Yu JY. Effect of aging on morphology of organomontmorillonite modified bitumen by atomic force microscopy. J Microsc 2011; 242(1):37–45. [29] Binnig G, Quate CF, Gerber C. Atomic force microscope. Phys Rev Lett 1986;56(9):930. [30] Burnham NA, Colton RJ. Measuring the nanomechanical properties and surface forces of materials using an atomic force microscope. J Vac Sci Technol A 1989; 7(4):2906–13. [31] Burnham NA, Dominguez DD, Mowery RL, Colton RJ. Probing the surface forces of monolayer films with an atomic-force microscope. Phys Rev Lett 1990;64(16):1931. [32] Butt H-J, Cappella B, Kappl M. Force measurements with the atomic force microscope: technique, interpretation and applications. Surf Sci Rep 2005;59(1):1–152. [33] Hanley SJ, Gray DG. Atomic force microscopy. Boca Raton, Fl: CRC Press, Inc.; 1995. [34] Rosa-Zeiser A, Weilandt E, Hild S, Marti O. The simultaneous measurement of elastic, electrostatic and adhesive properties by scanning force microscopy: pulsed-force mode operation. Meas Sci Technol 1997;8(11):1333. [35] Soenen H, Besamusca J, Fischer HR, Poulikakos LD, Planche J-P, Das PK, et al. Laboratory investigation of bitumen based on round robin DSC and AFM tests. Mater Struct 2013:1–16. [36] Yu X, Zaumanis M, dos Santos S, Poulikakos LD. Rheological, microscopic, and chemical characterization of the rejuvenating effect on asphalt binders. Fuel 2014;135: 162–71. [37] Qin Q, Farrar MJ, Pauli AT, Adams JJ. Morphology, thermal analysis and rheology of Sasobit modified warm mix asphalt binders. Fuel 2014;115:416–25. [38] dos Santos S, Partl MN, Poulikakos LD. Newly observed effects of water on the microstructures of bitumen surface. Construct Build Mater 2014;71:618–27. [39] De Moraes M, Pereira R, Simão R, Leite L. High temperature AFM study of CAP 30/45 pen grade bitumen. J Microsc 2010;239(1):46–53. [40] Nazzal MD, Kaya S, Gunay T, Ahmedzade P. Fundamental characterization of asphalt clay nano-composites. J Nanomechanics Micromech 2012;3(1):1–8. [41] Allen RG, Little DN, Bhasin A. Structural characterization of micromechanical properties in asphalt using atomic force microscopy. J Mater Civ Eng 2012;24(10):1317–27. [42] Pauli A, Branthaver J, Robertson R, Grimes W, Eggleston C. Atomic force microscopy investigation of SHRP asphalts: heavy oil and resid compatibility and stability. Prepr Am Chem Soc Div Petrol Chem 2001;46(2):104–10. [43] Masson JF, Leblond V, Margeson J, BUNDALO‐PERC S. Low‐temperature bitumen stiffness and viscous paraffinic nano‐and micro‐domains by cryogenic AFM and PDM. J Microsc 2007;227(3):191–202. [44] Schmets A, Kringos N, Pauli T, Redelius P, Scarpas T. On the existence of wax-induced phase separation in bitumen. Int J Pavement Eng 2010;11(6):555–63. [45] Claudy P, Letoffe J, King G, Plancke J. Characterization of asphalts cements by thermomicroscopy and differential scanning calorimetry: Correlation to classic physical properties. Fuel Sci Technol Int 1992;10(4–6):735–65.

33

[46] McCarron B. The investigation of 'bee-structures' in asphalt binders. Major Qualifying Project, Worcester Polytechnic Institute; 2012. [47] Lu X, Langton M, Olofsson P, Redelius P. Wax morphology in bitumen. J Mater Sci 2005;40(8):1893–900. [48] Masson J, Polomark G. Bitumen microstructure by modulated differential scanning calorimetry. Thermochim Acta 2001;374(2):105–14. [49] Pauli AT, Grimes W. Surface morphological stability modeling of SHRP asphalt. Abstr Pap Am Chem Soc 2003;225:U422-U422. [50] Thanh NX, Hsieh M, Philp R. Waxes and asphaltenes in crude oils. Org Geochem 1999;30(2):119–32. [51] Buenrostro-Gonzalez E, Andersen SI, Garcia-Martinez J, Lira-Galeana C. Solubility/ molecular structure relationships of asphaltenes in polar and nonpolar media. Energy Fuel 2002;16(3):732–41. [52] Qian K, Edwards KE, Mennito AS, Walters CC, Kushnerick JD. Enrichment, resolution, and identification of nickel porphyrins in petroleum asphaltene by cyclograph separation and atmospheric pressure photoionization Fourier transform ion cyclotron resonance mass spectrometry. Anal Chem 2009;82(1):413–9. [53] Speight JG. The chemistry and technology of petroleum. CRC press; 2014. [54] Srivastava S, Handoo J, Agrawal K, Joshi G. Phase-transition studies in biN nb/iN-alkanes and petroleum-related waxes—a review. J Phys Chem Solid 1993;54(6):639–70. [55] Kane M, Djabourov M, Volle J-L, Lechaire J-P, Frebourg G. Morphology of paraffin crystals in waxy crude oils cooled in quiescent conditions and under flow. Fuel 2003;82(2):127–35. [56] Edwards Y, Isacsson U. Wax in bitumen: Part II—characterization and effects. Road Mater Pavement Design 2005;6(4):439–68. [57] Edwards Y, Isacsson U. Wax in bitumen: Part I—classifications and general aspects. Road Mater Pavement Design 2005;6(3):281–309. [58] Redelius P. Asphaltenes in bitumen, what they are and what they are not. Road Mater Pavement Design 2009;10(Suppl. 1):25–43. [59] Tinsley JF, Jahnke JP, Dettman HD, Prud'home RK. Waxy gels with asphaltenes 1: characterization of precipitation, gelation, yield stress, and morphology. Energy Fuel 2009;23(4):2056–64. [60] Lovinger AJ, Lotz B, Davis DD, Schumacher M. Morphology and thermal properties of fully syndiotactic polypropylene. Macromolecules 1994;27(22):6603–11. [61] Mehta R, Keawwattana W, Guenthner AL, Kyu T. Role of curvature elasticity in sectorization and ripple formation during melt crystallization of polymer single crystals. Physical Review E 2004;69(6):061802. [62] Tsukruk VV, Reneker DH. Surface morphology of syndiotactic polypropylene single crystals observed by atomic force microscopy. Macromolecules 1995;28(5):1370–6. [63] Zhou W, Cheng SZ, Putthanarat S, Eby RK, Reneker DH, Lotz B, et al. Crystallization, melting and morphology of syndiotactic polypropylene fractions. 4. In situ lamellar single crystal growth and melting in different sectors. Macromolecules 2000; 33(18):6861–8. [64] Toda A, Okamura M, Hikosaka M, Nakagawa Y. Three-dimensional shape of polyethylene single crystals grown from dilute solutions and from the melt. Polymer 2005; 46(20):8708–16. [65] Brochard-Wyart F, de Gennes PG. How soft skin wrinkles. Science 2003;300(5618): 441-441. [66] Cerda E, Mahadevan L. Geometry and physics of wrinkling. Phys Rev Lett 2003; 90(7):074302. [67] Nahar S, Mohajeri M, Schmets AJ, Scarpas A, van de Ven MF, Schitter G. First observation of blending-zone morphology at interface of reclaimed asphalt binder and virgin bitumen. Transportation Research Board 92nd Annual Meeting; 2013. [68] Fischer HR, Dillingh B. Response of the microstructure of bitumen upon stress–damage initiation and recovery. Road Mater Pavement Design 2014:1–15 [(ahead-ofprint)]. [69] Lonchampt P, Hartel RW. Fat bloom in chocolate and compound coatings. Eur J Lipid Sci Technol 2004;106(4):241–74. [70] Rebelo L, de Sousa J, Abreu A, Baroni M, Alencar A, Soares S, et al. Aging of asphaltic binders investigated with atomic force microscopy. Fuel 2014;117:15–25. [71] Fischer H, Poulikakos LD, Planche J-P, Das P, Grenfell J. Challenges while performing AFM on bitumen. Multi-Scale Modeling and Characterization of Infrastructure Materials. Springer; 2013. p. 89–98. [72] Tarefder RA, Zaman AM. Nanoscale evaluation of moisture damage in polymer modified asphalts. J Mater Civ Eng 2009;22(7):714–25. [73] Dourado ER, Simao RA, Leite LF. Mechanical properties of asphalt binders evaluated by atomic force microscopy. J Microsc 2012;245(2):119–28. [74] Dokukin ME, Sokolov I. Quantitative mapping of the elastic modulus of soft materials with HarmoniX and PeakForce QNM AFM modes. Langmuir 2012;28(46): 16060–71. [75] Behrend O, Odoni L, Loubet J, Burnham N. Phase imaging: deep or superficial? Appl Phys Lett 1999;75(17):2551–3. [76] Allen RG, Little DN, Bhasin A, Lytton RL. Identification of the composite relaxation modulus of asphalt binder using AFM nanoindentation. J Mater Civ Eng 2012; 25(4):530–9.