Characterizing Micromechanical Properties of Asphalt

Atomic Force Microscopy

Introduction:

Due to limited understanding of the micromechanical behavior of asphalt, it is often difficult to predict the service life of HMA. The varying chemical and structural compositions of crude oil source combined with the complex rheological and thermoplastic behavior of asphalt adds more to it. This high level of variability between different asphalt sources result in different asphalt chemical composition and microstructure. To characterize micromechanical properties of asphalt, a good understanding of current methods used in characterizing the chemical composition and microstructure of asphalt is required.

Following methods study the asphalt molecular composition:

• Gas chromatography,
• Size-exclusion chromatography (SEC) or GPC,
• Mass spectroscopy,
• Fourier transform infrared spectroscopy.

To study the asphalt microstructure in details following microscopy techniques are used widely:

• Scanning electron microscopy,
• Atomic force microscopy (AFM).
• Transmission electron microscopy,
• Phase contrast microscopy etc. 

Theory and Components of AFM:

Conventional microscopes such as the optical and electron microscopes have magnification power ranging from 1000X to 100,000X but these can generate image in 2D only. The vertical dimension of the surface irregularities, height or depth such as holes, pits etc. cannot be measured with these microscopes. With a sharp tip that probes upon the surface, AFM can not only scan the surface topography with high magnification factor (up to 1,000,000X) but also measure the Z dimension, with higher resolution than X-Y plane. Since it is a mechanical imaging instrument, it can measure physical properties of the sample in addition to imaging. This is the reason why AFM is a widely used tool in the study of asphalt microstructure. AFM can be used into surface topography, phase separation and mechanical properties such as adhesion and stiffness at the micro-scale. Following characteristics make AFM an amazing instrument for imaging and measuring in the nanoscale [1]:

1. Atomic-scale Sensitivity: Unlike traditional instruments, AFM is of much smaller dimension to maintain the built-in sensitivity at atomic scale.
2. Fabrication Technology: The cantilever tip, the main component for imaging, can be fabricated as per the requirement of testing, i.e. surface structure of the samples tested. So the AFM can be customized to handle any project.
3. Motion Control: AFM has precise motion control technology to accurately scan and position the probe above the sample surface.

Figure 1 presents a typical commercial atomic force microscope. The basic components of a typical AFM are below [1]:

1. Stage: In AFM stage, the sample is kept for imaging and testing. It hosts the scanner containing the cantilever probe, the optical microscope and the motion control system in order to move the sample in X, Y and Z direction.
2. Control Electronics: This component generates the electronic signals for moving the motors located on the stage along with the scanner. It also converts the scanned images into a digital form that can be processed by a computer.
3. Computer: Computer is required for acquiring and post-processing the images.

Basic Mechanism

Normally, an AFM can operate in following three methods [2]:

1. Surface Sensing: A cantilever with a sharp tip is probed over the sample surface to scan for images in surface sensing method. Two types of forces work between the tip and sample surface. In close range, the attractive forces govern as the tip approaches the surface and the cantilever deflects towards the surface. With further decrease in tip-surface distance, repulsive forces become active and deflect the cantilever away from the surface.
2. Detection Method: The cantilever deflections due to the attractive and repulsive forces are detected with a laser. A highly dense laser beam can provide excellent spatial resolution in combination with a photodiode detector of high sensitivity. When the cantilever probes over the sample surface, any cantilever deflection will reflect off the incident beam in a slightly different direction. Thus, a position-sensitive photodiode (PSPD) can record gradual deflections of the cantilever when the tip passes over an elevated or higher surface.
3. Imaging: The surface topography of the region of interest on the sample surface can be imaged with this method. As the cantilever deflects away or attracts towards the sample surface based on the surface irregularities, the PSPD monitors and records the ongoing defection of the cantilever tip. A feedback loop controls how much the tip should be above the sample surface and maintains the laser position as constant. This helps the AFM generating a topographic image of the sample surface.

Figure 2. Basic AFM set-up [3]

Figure 2 shows a basic set-up diagram of an AFM. It performs following basic concepts to operate in component level [1]:

1. Piezoelectric Transducers: The purpose of the piezoelectric transducer is to convert the electric potential to mechanical motion, in order to control the motion of the probe when it moves across the sample surface. Typically crystalline or amorphous, these materials can change geometry when subjected to potential difference at the two ends. The amount of change in shape depends on the type of the piezoelectric material and the magnitude of electric potential applied. If the expansion coefficient of a piezoelectric material is 0.1 nm, then the material will expand 0.1 nm if a potential of one volt is applied.
2. Force Transducers: The purpose of the force transducer is to measure the force between the sample surface and the probe. As long as the probe touches the sample surface, the force transducer gives an electrical signal which is amplified with an amplifier. After the probe touches the sample surface, the signal should increase monotonically with the increase in force between the sample surface and the probe. Typical capability of a force transducer is up to 10 pico Newton.
3. Feedback Control: Feedback controller continuously monitors the force between the sample surface and the probe tip (Figure 3) and wheels the piezoelectric ceramic responsible for the relative distance between the sample surface and the probe tip.

Figure 3. Basic Mechanism of an AFM [1]

Basically, these three components of an AFM help to put an image together. In a nutshell, the force transducer measures the force between the sample surface and the probe tip, the feedback controller retains a fixed force by regulating the height increase of the Z piezoelectric transducer and the X-Y piezoelectric ceramic moves the probe in the horizontal plane upon the sample surface. The electric potential from the Z piezoelectric ceramic is then converted into an image.

Figure 4. Atomic Interaction (force-separation curve)

When the probe tip and the sample surface are not in contact, following forces (Figure 4) tend to occur:

1. Repulsion: When the sample surface and the probe tip are very near (within a few Angstroms), an exchange interaction occurs among the atoms of the sample and the tip since their electronic orbitals overlap at such atomic distance. This short-range Coulomb interaction creates an intense repulsive force between the tip and sample surface.
2. Attraction (Van der Waals): Due to the intermolecular Van der Waals forces, resulting from the polarization interaction of nearby particles, an attractive force occurs between the tip and sample surface.

An AFM can be operated in the following modes: contact mode (CM), non-contact mode (NCM) and spectroscopy mode (SM). Each of the mode has their advantages and condition for appliance. In SM mode, the micromechanical properties of a sample can be obtained by moving the probe tip vertically, in respect to the test surface. NCM mode is useful for examining elastic or soft samples like asphalt binder since the force exerted is much less than CM mode.

Limitations of AFM:

A common problem when measuring with AFM is artifacts in images. The four primary sources (West & Starostina, 2009) [4] of artifacts are:

1. Probe tips,
2. Piezo-scanners,
3. Image processing or feedback,
4. Vibrations.

To minimize the likelihood of probe artifacts, the probe geometry should be much smaller than the features of the images being measured. Using a probe that is not the optimal size for the application can result in features on a surface appearing too large, features in an image appearing too small, strangely shaped objects or repeating strange patterns in an image. A general rule of thumb to detect image processing artifacts is: “if the image has no noise in it, then the data has likely been compromised (West & Starostina, 2009) [4].” 

Previous Works on AFM

When crude oil is distilled, we get asphalt binder. Several factors affect the mechanical properties of asphalt, such as: temperature, repetitive loading, time/rate of loading and aging. Various distresses in HMA is directly dependent upon the mechanical behavior of asphalt binder. Pavements designers and engineers are required to make sure that the asphalt binder applied in the mix should be able to resist fatigue or temperature cracking, rutting and other sort of distresses. To improve the quality of asphalt binder during the process of refining or by applying additives, proper understanding of the chemical components of asphalt is required.

Almost 82-88 % of asphalt is carbon. The rest 8-11 % is hydrogen atom. However, there are presence of heteroatoms of oxygen, sulfur and nitrogen which are responsible for the polarity in asphalt (Lesueur 2009 [5]; Petersen 1984 [6]). Based on the source of crude oil and also the refining process, the chemical components vary from binder to binder. Since the amount of each chemical component in asphalt binder is extremely diverse, it is sensible to categorize the components of asphalt binder based on their different characteristics rather than modeling the engineering properties of each of the component. Among the common characteristics such as polarity, ionic character and molecular size distribution, polarity has been found as more suitable in relating asphalt binder rheological properties with the chemical properties.

The SARA analysis divides the asphaltenes and maltenes of asphalt into four polar fractions commonly known as the Saturates, Aromatics, Resins and Asphaltenes where the Asphaltenes are the most polar and the Saturates are the least polar. Several researchers strained to relate the amount of polar fractions present in asphalt with their engineering properties. Corbett (1969) [7] confirmed that the presence of more aromatics and saturates soften the asphalt, but asphaltenes and resins increase the stiffness. Dealy (1979) [8] showed that the asphalt becomes more viscous if the asphaltene quantity is increased and vice versa for maltenes. After analyzing binder chemistry at molecular level, Robertson (1991) [9] found that polar molecules affect the elastic part of asphalt’s viscoelastic behavior whereas the non-polar part affect the viscous part.  However, after investigating physical and chemical properties of asphalt binder from two different sources, Michalica et al. (2008) [10] found that the binder that had lower asphaltene content exhibited more stiffness than that with higher asphaltene content. This finding confirms that stiffness of the binder cannot be directly related to the SARA fractions since the mechanical and physical properties of these components vary from source to source. In addition, it is not wise to conclude that an asphalt binder could have higher stiffness just because the asphaltene content is high. These finding led to the study of asphalt microstructure to better explain and relate the chemical and physical properties of asphalt with microscopic tools like AFM.

Researchers frequently observed the “bee” like structures in the dispersed phase of aged binder with AFM and proposed their presence as a correlation between increased amount of asphaltene and consecutive higher stiffness of the binder (Loeber et al. (1996) [11], Pauli et al. (2001) [12]).  In later work, Pauli et al. (2011) hypothesized the presence of bee structures in asphalt as a result of aliphatic chains.

Jäger et al. (2004) used AFM in NCM (non-contact mode) and SM (spectroscopy mode) to provide insight into surface topography and mechanical properties of two asphalts (B50/70 and B160/220) at room temperature. They also observed the “bee-shaped” structures in the topography images in both types of asphalt which were randomly distributed. In the surface topography, the alternating lower and higher elevations form the bees. They did not find any association with the change in topography with alternating asphalts. However, they identified four subdomains from the analysis of surface topography, each having different mechanical properties.

Figure 5. NCM result for (a) B50/70 and (b) B160/220; (c) characteristic dimensions, d [nm] and ΔH [nm] of one “bee” [Jäger et al. (2004)]

Jäger et al measured the characteristic dimensions, i.e., the maximum topographical change ΔH and the distance d between the higher parts within one “bee” (Figure 5).

Figure 6. PFM plots: (a) V_s [V] and (b) V_ad [V] of B70/100 (T = 10C); V_s [V] of B50/70 using (c) lower penetration and (d) higher penetration (T = 15C) [Jäger et al. (2004)]

Figures 6(a) and (b) show plots of the output quantities V_s (stiffness) and V_ad (adhesion) obtained from PFM measurements for B70/100 at T = 10C, and Figure 2(c) shows V_s for B50/70 at T = 15C. According to these results, the four subdomains identified from the surface topography (Figure 5) exhibit different mechanical properties. Hereby, the parts of the surface with larger stiffness [see light areas in Figure 6(a)] show a lower adhesive behavior, and vice versa.

Figure 7. Mean values of relative stiffness [%] of the four phases for (a) B50/70 and (b) B70/100; (c) average relative stiffness [%] for considered types of bitumen. [Jäger et al. (2004)]

In order to gain insight into the stiffness properties of the four subdomains observed for both bitumens shown in Figures 6(a) and (c), the relative stiffness, associated with the mean value of V_s over selected parts of the PFM plots, are given in Figures 7(a) and (b). The values corresponding to the soft part of the matrix and to the soft part of the “bees” are almost the same. However, they were obtained from different areas of the PFM plots, separated by the hard-matrix phase surrounding the “bee-shaped” structures. Figure 7(c) shows the mean values of the relative stiffness [%] obtained from averaging over the whole scanning domain (50 X 50μm).

Masson et al. (2005) [2] investigated 13 bulk specimens of asphalt binder. They prepared each sample to test in AFM by heat-casting method (described later). This method ensures the solid-state structure of asphalt since in solid-state of asphalt, the rheological properties are dominant. Masson et al used PDM, an intermittent contact AFM mode, to obtain phase-lag and topographic images at room temperature (Figure 8a). The AFM was a JEOL JSPM-5200 machine. MikroMasch probes with beam-shaped cantilevers were used. The tip radius, spring constant and free resonance frequency of the probe was 10 nm, 40 Nm^-1 and 160 kHz, respectively. Images were taken at several locations of the surface of each sample. Similar “bee structures” were observed. However, previously unseen morphologies of the binders were classified into three groups. The first group exhibited a fine dispersion (0.1–0.7 μm) in a homogenous matrix. The second group displayed domains of about 1 μm. The third group revealed up to four different phases termed as ‘catana’, ‘peri’, ‘para’ and ‘sal’ phases, which were found of vastly different sizes. Masson et al. also performed the topographic profile analysis on the catana phase, i.e., bee structures found in the PDM images.

 

(a)      (b)

Figure 8. (a) PDM image of binder PC (image size: 15 × 15 μm), (b) Typical line profile from topographic image. Maxima and minima portraits the pale and dark strips in the topographic image. A and B are the height and spacing [13].

They also performed the elemental analysis in each bitumen type and correlated the metal content (ppm) with the % area of catana phase. The area was calculated with an image analysis software (Clemex SPM). The saturate (S), polar aromatic (PA), asphaltene (As) and naphthene aromatic (NA) compositions as presented in Table 1 did not show any correlation with the PDM (phase detection microscopy) morphology. However, this research showed a good correlation between the area of the “bees” from PDM images and the elements in asphalt such as vanadium and nickel (Figure 9). This indicates that binder morphology and molecular arrangement seem to be partly governed by the polarity of the metallic cations.

Table 1. Features of asphalt binder studied (Masson et al., 2005)

Bitumen	Source/type	Composition*					Elemental analysis
		S	NA	PA	As		N (%)	S (%)	V (p.p.m.)	Ni (p.p.m.)	Fe (p.p.m.)
PC	Unknown	9	27	43	20		0.7	4.4	–	–	–
AAA	Lloydminster	11	32	37	16		0.5	5.5	174	86	<2
AAB	Wyoming sour	9	33	38	17		0.5	4.7	222	57	16
AAF	West Texas sour	12	34	39	13		0.5	3.4	91	36	100

*S, saturates; NA, naphthene aromatics; PA, polar aromatics; As, asphaltenes.

Grover et al. (2012) [14] studied the micro-rheology of these phases using AFM before and after oxidative aging upon six SHRP designated binders and presented differences between the properties amongst the various microstructures within an asphalt binder, in addition to the influence of oxidative aging on these properties. Nano-indentation creep measurements of each phase-separated regions showed heterogeneous domains in asphalt with different mechanical properties. Oxidative aging caused considerable microstructural changes within these domains, including variations in phase structure, properties and distribution. Clustering, phase dispersion and materialization (bee structure) are the three typical microstructural changes observed in this study.

Prior to aging, asphalts AAB, AAD and ABD each consisted of two distinctive phases: dispersed and continuous. Aging increases the percentage of higher stiffness components in asphalt binder which is the prime reason for increase in stiffness after aging. In addition, the dispersed and continuous phase also showed an increase in stiffness.

The Agilent 5400 AFM system was operated in NCM for phase imaging and SM for nano-scale indentation in order to estimate the micro-mechanical properties of asphalt. All imaging was performed with a Silicon Nitride tip mounted on the free end of a conical shaped PPP-NCL cantilevers. The average resonant frequency of the cantilever is 175 kHz and a force constant ranging from 21-98 N/m.

Figure 10. Phase images of binder AAB and location of force measurements for the two trial experiment methods (each image shows a 25 × 25 μm surface area; color scale shows relative differences in phases based on tip frictional resistance during scanning [14].

Grover et al. identified the distribution of different phases in each asphalt types (Figure 10), compared the deflection in each types (Figure 11b) and observed the line profile (Figure 11c) to see variation in the surface of each phase. They conducted creep measurements to differentiate the time-dependent creep-like deformation of each identified phases between unaged and aged binder (Table 2).

Figure 11. (a) Phase image; (b) creep measurements; (c) profile extraction for asphalt AAB [14] 

Table 2. Comparison of Phase Distribution and Deflection Data [14]

Grover also performed the statistical analysis (Table 3) of the surface texture, roughness and topography fluctuations to assess the microstructural phase differences of each asphalt types. A notable observation from the comparison of surface texture statistics is that S_p, S_v and S_z increased for each asphalt due to aging, which clearly shows that age-induced structural phase change is also accompanied by increased roughness.

Table 3. Comparison of Surface Texture Statistical Analysis [14].

Grover et al. (2014) [15] evaluated these phases for asphalt binders with varying concentrations of the different polar fractions. They observed that oxidative aging increased the area of ‘catana’ phase. This finding was consistent with the results from previous studies (Lin et al. 1995 [16]; Petersen 1984 [6]). Grover et al. (2014) [15] also reported that the concentration of different polar fractions and chemical parameters present in the binder had a significant and consistent influence on the size, distribution and micro-rheology of the different phases. In addition to the imaging mode, in this study, the AFM was also operated in spectroscopy (SM) mode. The SM mode helps to measure the adhesion (pull-off force) of soft materials such as asphalt binders by moving the tip vertically with respect to the sample surface. In order to obtain qualitative pull-off force measurements in the units (nN) reported in this paper, the detected tip-deflection data, reported in least significant bits (LSB) units, were converted to units of volts (V). The data values reported in volts are then converted to nanometers, D_max (nm), where D_max is the maximum vertical raw deflection of the cantilever tip given in the respective units. Finally, the pull-off force can be computed from the following equation:

Fpull-offnN=DmaxnmXk (nNnm)

The pull-off (adhesion) forces were obtained via force–distance curves by initiating an approach of the AFM tip into the asphalt followed by retraction of the tip from the asphalt. During the retraction phase of the test, the force required to remove the tip from the material surface transitions from a maximum negative force to a near-zero force, thus revealing the pull-off force of the tip-surface interaction as noted in Figure 14. By allowing the tips to interact with each phase of the two different asphalts, i.e. continuous, dispersed, ‘bee’ structure and ‘bee’ casing phases, the microstructural adhesion measurements were obtained. Six measurements were extracted from each phase, and the statistical analysis of the adhesion measurements is given in Table 4.

Table 4. Pull-off (adhesion) forces of discrete asphalt microstructural phases

A study conducted by Jahangir et al (2015) [17] evaluates the microstructural change in asphalt binder resulting from RTFO and PAV aging.  In this study, creep indentation technique was applied on asphalt binder using AFM. The creep data helped to obtain viscoelastic properties of the various phases detected in asphalt samples from two different sources. The effect of these changes on the evolution of damage in asphalt binder subject to tensile deformation was also investigated. Numerical simulations were performed with these data to compare the effect of tensile strains on the internal stress distribution of the binder, i.e. the induced damage with the experimental results. A microloading apparatus (Figure 14), fabricated specifically to induce tensile strains in the binder, was used and the resulting damage was observed with AFM.

Figure 13. AFM phase image showing interstitial phase of asphalt binders (a) A and (b) B (50 μm² X 50 μm²).

In order to quantitatively assess the changes in the microstructure due to the applied tensile strain, image analysis techniques were employed using the open source image analysis tool ‘ImageJ’. Various geometric attributes were obtained and a comparison of these attributes was done before and after strain was applied to the test specimens. Due to phase separation and changes that occurred in the bee casing and interstitial phases (Figure 13), comparisons here are only made on the basis of the discrete bee phase. Following Table 5 compares the distribution of the bee structures before and after load is applied to binders A and B.

Table 5. Microstructural analysis of asphalt binders A and B.

FE simulation and experimental results show that applying strain resulted in damage/phase separation concentrated in the interstitial zone between neighboring bee structures, defined as load-induced phase separation. This study proposes that at each of the domains within asphalt microstructure, the regions with high stress intensity act as a center for damage formation. However, aging reduces the difference in the rheological properties such as shear modulus, stiffness, hardening etc. of each microdomain. This makes the internal stress distribution more homogenous which leads to fewer nucleation sites for damage formation. This study suggests that evaluating the asphalt microstructure and micro-rheology is vital in perceiving how the damage formulates in asphalt binder and it can be better engineered to increase the overall durability.

Figure 14. Microloading frame [17]

In a recent study, Pauli et al (2014) [18] investigated the contact adherence energy of binder thin films with AFM force-displacement mechanism. This technique utilizes ultra-thin bitumen films (500–800 nm) as test sample and a glass micro-bead cantilever tip as the test probe. The mechanism involves applying direct tension between the bitumen and glass tip when they are in contact, and consecutively fracturing and creating the adherence. The work required to fracture the contact is measured in respect to temperature and rate of separation between the tip and binder. These studies were performed to measure surface energy of asphalt binder as it relates to the fatigue properties and self-healing behavior of asphalt binder. Adhesive properties were found to be a function of separation rate and test temperature by exhibiting the transition behavior from viscous to viscoelastic nature. Sample results of such adherence properties are presented in Table 6. In this paper, adherence energy is quantified and interpreted in terms of the DMT (Derjaguin, Muller, Toporov) model. Several test results also showed that at higher temperature (i.e, above the ambient temperature), the adherence energy diminishes and the surface energy prevails.

Table 6. Regression Coefficient (W₀, n), Contact Area (A), Equilibrium Work of Adhesion (G₀), and Adherence Energy (G) Prepared at Three Different Film Thicknesses, Reported for Three Asphalts [18]

While previous study strongly support the notion that the amount and characteristics of different chemical components significantly affect the rheological and micromechanical properties of asphalt, much research is further required on the wide variety of asphalt binders depending on the individual sources, in order to identify which specific components the asphalt rheology in what quantity. More profound study is required to characterize the aging properties and the shift in micro-mechanical behavior of asphalt due to oxidative aging which affects the HMA surface in the long-term.

AFM Data Analysis Parameters

Based on the aforementioned discussion, following parameters are selected in this study to analyze AFM data:

• Phase identification from topographic image
• Adhesion
• Stiffness
• Maximum force (F_max)
• Area  calculation
• Line profile
• Histogram and statistical data

Sample Preparation Technique for AFM

Method 1 (Masson et al, 2006 [13]):

• 12 mm steel disk (an AFM substrate) or microscope slide was heated for 1-2 min on a hot plate at 115 °C to pour small amount of asphalt and prepare thin asphalt film.
• Liquid asphalt was spread out with a blade to form a round film (diameter: 5 mm).
• The hot film was gradually cooled for 1 min on the hot plate to form a smooth and glossy surface, and then subsequently cooled to room temperature.
• Then the asphalt film was annealed for a minimum of 24 hours before imaging to reduce the possibility of steric hardening.
• A glass cover slide was used to press the hot asphalt into a thin film for optical microscopy.
• The images were captured at room temperature and normal pressure with MikroMasch (MikroMasch, Portland, OR) and beam-shaped cantilever probes were used for the measurements.
• The nominal tip radius of cantilever probe was 10 nm. An average spring constant of 40 Nm^-1 was with free resonance frequency of 160 kHz. A commercial image analysis software, “Clemex SPM”, was used to calculate the area of the bee-like structures.

Method 2 (Grover et al., 2010 [19]):

• Approximately 1.5g of asphalt was mixed with 11g High Performance Liquid Chromatography (HPLC) grade toluene and kept for 12 hours to dissolve.
• A spin processor (Brand: Laurell WS-650S) equipped with vacuum chuck is required to cast the asphalts into thin films.
• Each microscope slide was pre-cleaned with methanol and dried with nitrogen before casting.
• The glass slide was placed on top of the vacuum chuck in such a way that it completely cover the O-ring prior to applying vacuum.
• Typical spinning rate = 1000 rpm for 15 seconds.
• Before spinning, adequate amount of asphalt solution should be applied upon the glass slide with a dropper pipet to completely cover the top of it.
• This casting method yields to an asphalt film thickness of approximately 1000±200 nm which is usually adequate for testing.
• To eliminate the residual solvent, i.e, toluene, used in the solution, the thin-film specimens was stored in an air-tight heated vacuum desiccator and ultra-high purity dry nitrogen passing through a super clean purifier was again passed though that vacuum desiccator.
• This nitrogen purge was applied for at least 2 hours.
• After that, each asphalt film as annealed for at least one hour at 50C to eliminate any micro capillaries that might be formed in the film while performing the nitrogen purge to vaporize toluene.
• The type of the tip mounted at the free end of cantilever probe used for AFM imaging and SM force indentation could be fabricated from silicon or silicon nitride and the shape is conical.

Method 3 (Zhang et al, 2011) [20]:

•  A single drop of asphalt heated at 140◦C was placed on a steel disk (dimension: 10mm × 10mm × 1mm) and then cooled to ambient temperature (about 5◦C). Dust was prevented by covering the steel with a glass cap. The asphalt steel disk was annealed for a minimum of 24 hours before imaging (Pauli et al., 2001).
• Both phase and topographic images were taken. An etched silicon probe was used.
• Cantilever length was 125 μm with a curvature radius ranging between 5-10 nm. The drive frequency and the scan rate was 260 kHz and 0.8 Hz respectively.
• Test was performed in tapping mode. The piezoelectric driver stimulates the cantilever into resonance oscillation in tapping mode.
• The dimension of images are within 15 μm × 15 μm region. The mean square roughness (Rq) of the binders was determined by the digital Instruments Nanoscope Software (version V5.30r3.sr3).

Method 4 (Jahangir et al., 2015 [17]):

• Jahangir et al. used an alternate method to reduce the possible influence of the solvent cast spin coating process on the resulting microstructure of the binder.
• Binder samples were prepared for tensile testing with a custom-designed Teflon mold.
• The purpose of this mold was to shape and support the specimens between two aluminum plates that were designed to be used in conjunction with the loading frame.
• The mold also ensures that the sample is of a consistent thickness (approximately 3.5 mm). The specimen length was allowed to vary from 10 to 20 mm. The specimen width, however, was fixed at 14 mm.
• To prepare and attach the samples to the loading frame, a heat casting protocol was used.
• A small sample of the asphalt binder was heated to 150°C, which allowed the sample to start flowing freely.
• The binder was then poured in the mold and cooled to room temperature at a slow rate (during a period of 30 min) and was allowed to consolidate between two metal plates at the two ends of the Teflon mold.
• Heat casting simply requires the asphalt to heat it up and letting it cool down slowly. The objective is to make the surface fairly smooth. A tiny bit of asphalt on can be put on glass and heat it up to the desired temperature. Then the asphalt should be allowed to soften slowly and become fairly smooth and flat.

In this research, we have followed the procedure described by Jahangir et al (2015) [17], for sample preparation techniques.

References:

[1] P. E. West, Introduction to Atomic Force Microscopy: Theory, Practice, Applications. Pacific Nanotechnology, 2006.

[2] P. Systems, “AFM Principle.” [Online]. Available: http://www.parkafm.com/index.php/medias/nano-academy/how-afm-works.

[3] A. Kronenberger, “Atomic Force Microscopy – Basics and Applications,” School of Engineering and Science, Duke University, 2006. [Online]. Available: https://users.cs.duke.edu/~reif/courses/molcomplectures/MolecularImaging/AFM/Atomic Force Microscopy(AFM) (Kronenberger).pdf.

[4] P. West and N. Starostina, “AFM Image Artifacts.,” Pacific Nanotechnology, Inc., Lot-Oriel Gruppe Europa, 2009.

[5] D. Lesueur, “The colloidal structure of bitumen: Consequences on the rheology and on the mechanisms of bitumen modification,” Advances in Colloid and Interface Science, vol. 145, no. 1–2. pp. 42–82, 2009.

[6] J. C. Petersen, “CHEMICAL COMPOSITION OF ASPHALT AS RELATED TO ASPHALT DURABILITY: STATE OF THE ART,” Transp. Res. Rec., no. 999, pp. 13–30, 1984.

[7] L. W. Corbett, “Composition of asphalt based on generic fractionation, using solvent deasphaltening, elution-adsorption chromatography, and densimetric characterization,” Anal. Chem., vol. 41, no. 4, pp. 576–579, 1969.

[8] J. M. Dealy, “Rheological properties of oil sand bitumens,” Can. J. Chem. Eng., vol. 57, no. 6, pp. 677–683, 1979.

[9] R. E. Robertson et al., “Chemical properties of asphalts and their relationships to pavement performance,” Asph. Paving Technol. Assoc. Asph. Paving Technol. Tech. Sess., vol. 60, pp. 413–436, 1991.

[10] P. Michalica, I. B. Kazatchkov, J. Stastna, and L. Zanzotto, “Relationship between chemical and rheological properties of two asphalts of different origins,” Fuel, vol. 87, no. 15–16, pp. 3247–3253, 2008.

[11] L. LOEBER, O. SUTTON, J. MOREL, J.-M. VALLETON, and G. MULLER, “New direct observations of asphalts and asphalt binders by scanning electron microscopy and atomic force microscopy,” J. Microsc., vol. 182, no. 1, pp. 32–39, 1996.

[12] A. T. Pauli, J. F. Branthaver, R. E. Robertson, W. Grimes, and C. M. Eggleston, “Atomic force microscopy investigation of SHRP asphalts,” Prepr. – Am. Chem. Soc., Div. Pet. Chem., vol. 46, pp. 104–110, 2001.

[13] J. F. Masson, V. Leblond, and J. Margeson, “Bitumen morphologies by phase-detection atomic force microscopy,” J. Microsc., vol. 221, no. 1, pp. 17–29, 2006.

[14] R. G. Allen, D. N. Little, and A. Bhasin, “Structural Characterization of Micromechanical Properties in Asphalt Using Atomic Force Microscopy,” J. Mater. Civ. Eng., vol. 24, no. 10, pp. 1317–1327, 2012.

[15] R. G. Allen, D. N. Little, A. Bhasin, and C. J. Glover, “The effects of chemical composition on asphalt microstructure and their association to pavement performance,” Int. J. Pavement Eng., vol. 15, no. 1, pp. 9–22, 2014.

[16] F. Lin and D. J. Meier, “A Study of Latex Film Formation by Atomic Force Microscopy. 1. A Comparison of Wet and Dry Conditions,” Langmuir, vol. 11, no. 7, pp. 2726–2733, 1995.

[17] R. Jahangir, D. Little, and A. Bhasin, “Evolution of asphalt binder microstructure due to tensile loading determined using AFM and image analysis techniques,” Int. J. Pavement Eng., vol. 16, no. 4, pp. 337–349, 2015.

[18] P. Troy, G. Will, C. Alec, and H. Shin-Che, “Adherence Energy of Asphalt Thin Films Measured by Force-Displacement Atomic Force Microscopy,” J. Mater. Civ. Eng., vol. 26, no. 12, Dec. 2014.

[19] R. G. ALLEN, “STRUCTURAL CHARACTERIZATION OF MICROMECHANICAL PROPERTIES IN ASPHALT USING ATOMIC FORCE MICROSCOPY,” Texas A&M University, College Station, 2010.

[20] H. L. Zhang, H. C. Wang, and J. Y. Yu, “Effect of aging on morphology of organo-montmorillonite modified bitumen by atomic force microscopy,” J. Microsc., vol. 242, no. 1, pp. 37–45, 2011.

[21] Nanounity, “Nano-Profiler Atomic Force Microscope.” [Online]. Available: http://www.nanounity.com/atomic-force-microscopy.php.