Journal of Elastomers and Plastics

4 downloads 0 Views 3MB Size Report
Jun 26, 2014 - van Gurp–Palmen plot. Rheological data can be presented by plotting the phase angle versus the absolute value of the complex modulus.
Journal of Elastomers and Plastics http://jep.sagepub.com/

Relationship between fractal, viscoelastic, and aging properties of linear and radial styrene −butadiene−styrene polymer-modified bitumen Zelimir Jelcic, Vesna Ocelic Bulatovic, Vesna Rek and Kristina Jurkas Markovic Journal of Elastomers and Plastics published online 26 June 2014 DOI: 10.1177/0095244314538437 The online version of this article can be found at: http://jep.sagepub.com/content/early/2014/06/26/0095244314538437

Published by: http://www.sagepublications.com

Additional services and information for Journal of Elastomers and Plastics can be found at: Email Alerts: http://jep.sagepub.com/cgi/alerts Subscriptions: http://jep.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://jep.sagepub.com/content/early/2014/06/26/0095244314538437.refs.html

>> OnlineFirst Version of Record - Jun 26, 2014 What is This?

Downloaded from jep.sagepub.com by guest on July 30, 2014

Article

Relationship between fractal, viscoelastic, and aging properties of linear and radial styrene–butadiene–styrene polymer-modified bitumen

Journal of Elastomers & Plastics 1–33 ª The Author(s) 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0095244314538437 jep.sagepub.com

ˇ elimir Jelcˇic´1, Vesna Ocelic´ Bulatovic´2, Z Vesna Rek2 and Kristina Jurkasˇ Markovic´3

Abstract The present study is aimed at evaluating the morphological and interfacial properties of the polymer and bitumen (BIT) in linear styrene–butadiene–styrene (SBS-L) block copolymer and radial SBS (SBS-R) block copolymer-modified BIT (PmB) by fractal analysis. Fluorescence microscopy technique coupled with image analysis was used to measure the particle size distribution and fractal dimension of PmB morphology formed in mixtures. Fractal analysis approach was proposed as quantitative description method to evaluate nonuniformity of phase-separated SBS polymer particles/fibrils in PmB. Fractal-like structures in fluorescence micrographs of PmB morphology with the polymer phase-separated particles/fibrils were quantified using interface D[BW], BIT-rich phase D[BþBW], and SBS-rich phase D[WþBW] box-counting fractal dimensions. The overall morphological structure became more compact or more fibril organized as the polymer content attained overcritical values. The relatively high values for the BIT-rich phase D[BþBW] fractal dimension suggests that the PmB mixtures have a structure with extremely high space-filling capacity. Polymer-BIT mixture properties by the conventional laboratory performance tests and by the rheological measurements before and after the Rolling Thin Film Oven Test have been correlated with the PmB mixtures fractal dimensions.

1 PLIVA Croatia Ltd, Research and Development Group, Teva Active Pharmaceutical Ingredients (TAPI), Zagreb, Croatia 2 Faculty of Chemical Engineering and Technology, University of Zagreb, Zagreb, Croatia 3 Institute of Civil Engineering Croatia, Zagreb, Croatia

Corresponding author: Zˇelimir Jelcˇic´, PLIVA Croatia Ltd, Research and Development Group, Teva Active Pharmaceutical Ingredients (TAPI), Prilaz baruna Filipovic´a 29, Zagreb HR10000, Croatia. Email: [email protected]

Downloaded from jep.sagepub.com by guest on July 30, 2014

2

Journal of Elastomers & Plastics

Keywords SBS, polymer-modified bitumen, rheology, fractal, aging

Introduction Polymer-modified bitumen (BIT; PmB) is generally considered to provide prolonged life or enhanced pavement performance of BIT.1–4 BIT modified with thermoplastic elastomers, styrenic block copolymers, due to their high viscoelasticity, higher creep and recovery under the traffic frequency, and enhanced BIT resistance to permanent deformation.5 The structure of styrene–butadiene–styrene (SBS) block copolymer consists of three block chains of SBS, with a two-phase morphology of spherical glassy polystyrene block domains within a matrix of rubber polybutadiene. The formation of PmB is based on the dissolution and/or fine dispersion of polymer in BIT and on the compatibility of the polymer/BIT blend. BIT and PmB properties are still defined on conventional tests such as penetration and ring and ball temperature. The rheological specifications are adopted by the Strategic Highway Research Program (SHRP).6–8 The factors affecting BIT and PmB aging include characteristics of BIT, composition of BIT and PmB, polymer content, and structure and phase interactions.4 In this article, the influence of content and structure of SBS copolymers (i.e. SBS with a linear structure (SBS-L)9 and SBS with a branched/radial structure (SBS-R)), as BIT polymer modifier on the rheological properties of PmB were researched in a range of temperatures under defined traffic frequencies with dynamic shear rheometry (DSR). The rheological properties of unaged PmB and of PmB after artificial thermo-oxidative aging in Rolling Thin Film Oven Test (RTFOT) were determined. SBS-L and SBS-R polymer-modified 70/100 BIT showed differences in morphology, storage stability, and in-phase separation diagram, while their rheological properties were found to be strongly morphology related. PmB displayed a similar morphology and that the polymer was seen as a discontinuous phase. The polymer-modified binder (PmB) compatibility properties are generally assessed by microscopy (morphology) and storage stability tests. PmB morphology and storage stability depend on the chemical nature of the base BIT, the characteristics and contents of the polymer, as well as the manufacturing process.10–13 BIT aging during production, application, and their service life is a very complex process causing the deterioration of asphalt pavements.14 Morphology and storage stability for a given PmB are influenced by the temperature and the thermal history of the BIT. Visual texture in a fluorescence image is often an important indication as to the nature or state of the PmB system being studied. The qualitative descriptors do not fully characterize the complex shapes observed. The problem is then to provide a quantitative measure of the observable texture. The use of fractal dimension has been investigated to provide a single number that is directly related to observed texture. The light objects in the fluorescence images of the PmB are assumed to exhibit the self-affine properties in a certain range of scales and they are treated as self-affine statistical fractals. The results of the fractal analysis of the selfaffine random light or SBS-rich phase objects in fluorescence images are used to classify these objects in various PmB. SBS-rich phase objects are not fractal from a mathematical

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

3

point of view because self-similarity cannot be observed over the entire range of sizes— only over some limited intervals. Image analysis of two-dimensional (2D) images of the phase-separated polymer-modified BIT mixtures yield limited data set on the irregularity features, such as area, perimeter, and length of phase-separated objects. In general, there are two basic methods of characterizing aggregates from 2D image data.15 The first method is to consider the scaling relationship between both the perimeter and the area. The second method is to quantify the fractal-like structures using interface D[BW], BITrich phase or ‘‘classical’’ box counting D[BþBW] and D[WþBW] fractal dimensions. This article evaluates the fractal descriptors (Df) of SBS/BIT mixtures morphology as a function of SBS content and aging. It is to be noted that results of different fractal analysis methods differ. This fact is caused by systematic error of different fractal analysis approaches. Different methods may quantify different aspects of the image, and thus, it is important to establish some criteria for choosing a particular method. Since not all methods give identical results for the same representation of an object, our approach has been to evaluate the usefulness of the different methods to compute Df in order to relate various computation results for Df to PmB characteristics as a function of SBS content and aging.

Experimental Sample preparation and conventional measurements have been considered as described in the work by Ocelic´ Bulatovic´ et al..16,17 The 70/100 penetration grade base BIT was modified by SBS-L, commercial grade Kraton D1101, with the content of polystyrene to be 31 wt%, and SBS-R, commercial grade Kraton D1184, with 30 wt% polystyrene content, both fabricated by the Shell Chemicals Company (Germany). The SBS-L- and SBS-R-modified BIT are produced with various contents of polymer. The samples are labeled as SBS-L-(wt%) or SBS-R-(wt%) for the SBS-Land SBS-R-modified BIT, respectively.

Optical microscopy The PmB samples with and without SBS compatibilizer were observed under an optical fluorescence microscope (BX-51, magnification 100; Olympus, Tokyo, Japan) to evaluate the changes qualitatively in the morphology as a consequence of addition and interaction between the components. All images were exported as tagged image file format files with 2040  1536 pixels and 1.14 pixel/mm. The limiting resolution was approximately 2 mm. Morphology is defined as the partition of polymer-rich and BIT-rich phases. The so-called bulk morphology of BIT-rich phase and that of polymer-rich phase are related to differences in rheological behavior.

Fractal analysis approaches Fractal dimension values were computed for the microscopic fine-sized texture in the fluorescence images of PmB by the monofractal approaches. The area–perimeterderived fractal dimensions DF,AP has been derived18,19 from the fluorescence images

Downloaded from jep.sagepub.com by guest on July 30, 2014

4

Journal of Elastomers & Plastics

according to the relation: perimeter ¼ const  areaðDF;AP =2Þ , where DF,AP is the area– perimeter fractal dimension (embedded into 2D). DF,AP can be obtained from the slope   of the linear fit slope ¼ DF;AP 2 from the log–log plot. Modified box-counting method. When comparing the performance of various fractal dimension estimators, there is a need to compare the estimated fractal dimension D on the physical grounds as they represent the corresponding PmB phases. The fractal analysis method uses pixel intensity height profiles extracted from the fluorescence images and presented at threshold m-heights (scaling along the intensity surface axis) in an x-y image plane (scaling along the grayscale image plane). Counting black, white, and partially black squares separately can modify this method. The three fractal dimensions D[BW] (interface dimension), D[WþBW] (SBS-rich phase mass dimension), and classical box dimension D[BþBW] (BIT-rich phase or matrix mass dimension) are capable of describing morphological differences in PmB (HarFa20). The method is based on a simple principle where a square mesh of various sizes " is laid over the image object. A log– log plot of number of boxes N(") against mesh size " then yields a line of slope equal to D. Dimension D[BþBW] is then referred to as a ‘‘classical’’ box dimension. D[BW] characterizes properties of black and white or BIT and SBS interface. Therefore, five independent fractal dimensions can be computed; the most important the interface D[BW], the BIT-rich phase or classical box counting D[BþBW], and SBS-rich phase fractal index D[WþBW] (arises by adding number of white squares NW and remaining number of white and black squares NBW that contain just interface part of fractals). It should be noted that the interface D[BW], and the classical box counting D[BþBW] fractal index do not necessarily describe the same physical property of the objects and thus may have different values. In particular, they diverge at the morphological extremes of space-filling and tenuous objects. Fractal indices D[BW] characterize the complexity of the phase-separated surface, whereas the D[BþBW]) represents the complexity of the general phase-separated structure. The fractal dimension of image is established in the whole range of threshold conditions (fractal analysis–range). The ‘‘fractal spectrum’’ represents the fractal dimension as a function of threshold condition (e.g. fractal dimension as a function of masked intensity or shade of gray value, m).21–23 Higuchi fractal dimension. The Higuchi fractal dimension (Dh) is a measure of irregularity and is calculated for time series directly in the time domain.24 Every horizontal row and every vertical column of the image is extracted and taken as a separate one-dimensional (1D) signal. This approach leads to (n þ m)-many signals, with n being the number of image columns and m the number of image rows..25 Several lengths, that is, L(d), of the signal or curve are calculated, and a double logarithmic plot, ln[L(d)] versus ln(d), is used to estimate the actual dimension value. The assumption is that a fractal signal scales according to the following: Lðd Þ  d Dh . Fourier fractal dimension. The Fourier fractal dimension (Dfft) can be determined from the fast Fourier transform (FFT) of a fluorescence image. In the Fourier power spectrum method of fractal dimension estimation,26 the 2D Fourier transform of the digital image

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

5

is calculated first and the 2D-power spectrum is obtained. The latter is then reduced to a 1D-power spectrum by averaging over circles. The average 1D-power spectrum P(f ) of the surface is a function of the frequency f. The FFT power spectrum provides a direct measurement of the fractal dimension of surfaces. The slope and intercept of the amplitude versus frequency curve (on log–log axes) give the dimension and lacunarity as a function of direction.

Image texture However, it is recommended to combine fractal dimension measurements with other texture descriptors since fractal sets may share the same fractal dimension and yet have different textures. Image texture can be loosely defined as a descriptor of local brightness variation from pixel to pixel in a small neighborhood through an image. Alternatively, texture can be described as an attribute representing the spatial arrangement of the gray levels of the pixels in a region of a digital image.27 Visual textures are generated by the interaction of light with a rough surface28 and contain spatial variations in intensity, each visible to some degree, and on the whole, densely and evenly arrayed over the field of view. Another approach consists of performing a global characterization of phaseseparated phases using image texture description.29 Standard measures like features extracted from gray-level occurrence matrices (GLCM) calculate second-order statistics of the gray-level distributions or spatial gray-level difference statistics. GLCM have already proved their ability to classify using texture different images.30 GLCM are second-order statistics and use the relationship between groups of two (usually neighboring) pixels in the original image. It compares one pixel with another in the digital image. If one wants to compute the GLCM of an image matrix, three parameters are of importance: first, the length of translation; second, the angle of displacement; and finally, the number of gray levels in the image.29 A quantitative analysis of the GLCM through 14 textural descriptors is possible, although typically only a few of these are widely used.31–33 Many of the statistics suggested by Haralick et al. produce highly correlated texture features34–36 which is not desirable. Clausi37 studied the relationship between the statistical parameters and concluded that entropy, contrast, and correlation compose a preferred set of statistical parameters. Five of the most commonly used descriptors (angular second moment (ASM), contrast, correlation, inverse difference moment (IDM), and entropy) are used to extract textural features from the GLCM of the fluorescence grayscale images. These equations are presented in Table 1. The GLCM calculates how often a pixel with i gray-level (grayscale intensity or tone) value occurs either horizontally, vertically, or diagonally to adjacent pixels with the j value. ASM and uniformity are measures of textural uniformity of an image. ASM reaches its highest value when the gray-level distribution has either a constant or a periodic form, for a homogenous image. Contrast measures the amount of local variations in an image. IDM measures image homogeneity. This parameter achieves its largest value when most of the occurrences in GLCM are concentrated near the main diagonal. IDM is inversely proportional to GLCM contrast. Entropy measures the disorder of an image, and it achieves its

Downloaded from jep.sagepub.com by guest on July 30, 2014

6

Journal of Elastomers & Plastics

Table 1. Equations used to extract textural features by GLCM.a,b Contrast Contrast ¼

nP 1 nP 1

Mi; j ði  jÞ2

i¼0 j¼0

ASM or energy ASM ¼

nP 1 nP 1

Mi;j 2

i¼0 j¼0

Correlation Correlation ¼

nP 1 nP 1 i¼0 j¼0

Entropy Entropy ¼

n P n P

Mij ðimÞðjmÞ si sj

Mi;j log Mi;j

i¼1 j¼1

IDM or homogeneity IDM ¼

n1 P n1 P i¼0 j¼0

Mi;j 1þðijÞ2

GLCM: gray-level co-occurrence matrices; ASM: angular second moment; IDM: inverse different moment. a Mij is the element i, j of the normalized symmetrical GLCM, n is number of gray levels in the image, m is the GLCM mean or an estimate of the intensity of all pixels, and s is the variance of the intensities of all reference pixels in the relationships that contributed to the GLCM. b Based on the work by Haraclick.29

largest value for a heterogeneous image. The GLCM parameters were calculated38 by the Texture Analyzer v0.4.

Saliency An image detail appears salient when one or more of its low-level features (e.g. size, shape, luminance, color, and texture) exceeds the overall feature variation of the background. The computed visual saliency methods that compute visual saliency are based on the assumption that feature saliency is inversely related to feature occurrence (i.e. rare features are more informative and therefore more salient or surprising than features that occur more frequently). In this view, interesting image details correspond to the locations of maximal self-information.39 Saliency is the combination of size and contrast and indicates how important this object is relative to other objects. The saliency corresponds to the total amount of gray, or mass (that may be a better name for this), in the object in comparison with the surrounding area, the most complete characteristic of its importance. Thus, the saliency corresponds to the SBS-rich phase in the PmB.

2D sample correlational map The basic concept of 2D correlation analysis40 may be regarded as a form of manipulation and comparison of two vectors measured at two independent spectral variables. The pairwise comparison of two vectors generates a correlation intensity usually plotted

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

7

as a contour map. The intensity of a synchronous 2D correlation spectrum represents the simultaneous or coincidental changes of two separate spectral intensity variations measured at two data points for the externally defined variable, such as, in our case, the SBS content. The 2D sample correlation technique was intended here for the study of the sample content and aging perturbation. Correlation between bands is found through the so-called synchronous and asynchronous spectra that correspond to the real and imaginary parts of the cross-correlation of spectral intensity at two points. In a synchronous 2D map, the peaks located in the diagonal (autopeaks) correspond to changes in intensity induced by the external perturbation and are always positive. The cross-correlation peaks indicate an in-phase relationship between two bands involved. Asynchronous maps show not-in-phase cross-correlation between the bands, and this gives an idea of the course of the events produced by the perturbation. The 2D contour maps have a color range starting at red (highest correlation) and passing through yellow to blue (lowest correlation). Heterospectral correlation analysis is a variant of 2D correlation spectroscopy, where a dynamic spectrum measured by one technique (e.g. fractal analysis) may be compared with another dynamic spectrum measured with a completely different probe (e.g. rheological analysis). Correlation peaks appear at both diagonal and off-diagonal positions. Individual autopeaks and crosspeaks can be readily assigned to various thresholds in fractal analysis or temperature for the rheological analysis.

Results and discussion The critical SHRP temperature values for permanent deformation, when before aging ðG =sin d Þ  1kPa, and after aging ðG =sin Þ  2:2kPa, are higher for PmB than that of the base BIT.5 PmB with higher polymer contents have higher critical temperatures, which means better temperature resistance to permanent deformation. Better rheological properties and resistance to temperature can be also noted in conventional tests, such as penetration test, ring and ball temperature test, and elastic recovery test.5 Decrease in penetration and an increase in the softening point with the polymer addition has been observed.41,42 The conventional properties of the radial SBS-R PmB are superior to those of the linear SBS-L PmB even with less uniform polymer dispersion in the BIT.

van Gurp–Palmen plot Rheological data can be presented by plotting the phase angle versus the absolute value of the complex modulus. This plot is the so-called van Gurp–Palmen plot, which is temperature invariant and provides a technique to verify the time–temperature superposition (TTS) principle. However, this is not only application of this plot, it should be pointed out that this plot provides other application to analyze rheological behavior of polymer melts and solutions. The van Gurp–Palmen method43,44 consists of performing a frequency sweep in the linear regime and plotting phase angle  (strain–stress phase lag in linear rheological regime) against the magnitude of the complex modulus, |G*(!)|, normalized to the plateau modulus, GN0 . However, we have extended this approach in the temperature domain where now the complex modulus, |G*(T)|, normalized to GN0

Downloaded from jep.sagepub.com by guest on July 30, 2014

8

Journal of Elastomers & Plastics

at low temperatures. Taking advantage of the low plateau value complex modulus G* measured at low temperatures is extrapolated to zero complex modulus GN0 . In this limit, limT! 0 G ¼ GN0 , complex modulus G* can be an effective quantity for ‘‘fingerprinting’’ phase-separated PmB. This fingerprints of the PmB can be easily presented by extending the van Gurp–Palmen plot to include the nonlinear rheological behavior in the thermal domain. This plot is very sensitive to morphology and can also be used to obtain extrapolated values of GN0 of the blends.45 Unlike the master curve, the van Gurp–Palmen plot does not require any shifting, thus temperature-independent curves can be constructed when the material obeys TTS principle.46 Typically, expected viscoelastic behavior describes a decreasing phase angle  and increasing complex modulus as the temperature decreases. Relaxation curves of the BIT show a single, broad profile approaching phase angle  ¼ 90 as complex modulus G* decreases, which is characteristic of linear chain structure. However, curves of PmB exhibit distinct inflections, indicating that additional constraints are present during the relaxation process. These inflections are characteristic of a long-chain branched structure or physical network entanglements. It can be seen that the addition of SBS polymer leads to a binder that performs more elastic under high temperatures/low frequencies compared with normal BIT, the PmB had a more elastic response under these conditions. It is also apparent that aging does not alter binder properties. The form of the van Gurp–Palmen plot is very similar for the nonaged (before RTFOT) and aged (after RTFOT) systems. Asphalt binders do not exhibit sudden changes in their behavior with respect to time or temperature. Therefore, any discontinuities revealed by the visual inspection of the test data in graphical format can be attributed to testing errors. These errors arise from the experimental setup due to changing the parallel plate testing geometry at 30 CC. Phase angle  is a very sensitive parameter to small change in rheology as the phase angle is approximately equal to the derivative of the logarithm of the stiffness with respect to frequency according to the following equation47:    d ðln G ð!; T ÞÞ ð!; T Þ ffi 2 d ln ! 48 The phase angle  is a direct measure  0 for low-temperature performance. When  phase angle  is plotted versus jG ðT Þj GN , there is a clear discrimination between PmB with different SBS-L content. PmB exhibit a lower phase angle than pure BIT for  0.01. PmB samples with 2–3 wt% SBS-L can exhibit a higher phase jG ðT Þj GN0 up to  angle for jG ðT Þj GN0 > 0:01. The different rheological behaviors of samples are observed in van Gurp–Palmen plot (Figures 1 and 2) that presents  versus G*. It is well established that compatibility and morphology of two phase systems can be studied by the related plot. It has been reported when the curve has a downward trend lower than its complex modulus, a droplet–matrix morphology is being formed.49,50 As shown in Figure 1, there is only one circular arc in the curve as homogeneous region. Above a certain temperature, a tail or a second circular arc appears on the left-hand side of the arc, corresponding to the sample of symmetric or off-symmetric composition, respectively, which indicates the appearance of a second relaxation mechanism and denoting the

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

9

Figure 1. van Gurp–Palmen plots of SBS-L-modified BIT samples (before RTFOT): BIT (♦), 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt% (●). SBS-L: styrene–butadiene–styrene linear; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

Figure 2. van Gurp–Palmen plots of SBS-L-modified BIT samples (after RTFOT): BIT (♦), 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt% (●). SBS-L: styrene–butadiene–styrene linear; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

formation of a second phase. Such plots in the phase-separated region clearly display two different relaxation mechanisms corresponding to the two different phase domains, which means that the droplet–matrix and co-continuous morphologies should be responsible for such features. From the characteristic shape of our samples, it can be concluded that dispersion is common in all morphologies of the samples. This observation was confirmed by fluorescence micrographs discussed in the morphology section (Figures 3 and 4).

Downloaded from jep.sagepub.com by guest on July 30, 2014

10

Journal of Elastomers & Plastics

Figure 3. van Gurp–Palmen plots of SBS-R-modified BIT samples (before RTFOT): BIT (♦), 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt% (●). SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

Figure 4. van Gurp-Palmen plots of SBS-R-modified BIT samples (after RTFOT): BIT (♦), 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt% (●).SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

The van Gurp–Palmen plots were used as a tool to investigate compatibilization of immiscible blends by Macaubas and Demarquette and Schulze et al.51,52 Terminal rheological response of phase-separated blend is similar with long-chain branching effect and should show characteristic shape in van Gurp–Palmen plot. As illustrated in Figures 1 and 2, above a certain temperature, that is, in phase-separated state, the curves do not merge into a single curve and show a temperature-dependent behavior, which means that TTS does not hold. It is should be pointed out that van Gurp–Palmen plots

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

11

seem to be more sensitive than mater curves to test the validity of TTS principle. In general, the curve of the homogeneous linear polymer melt finally reaches a plateau at 90 when going from high to low complex modulus |G*| values in general. In contrast to that, such plots of the two-phase samples exhibit characteristic shapes in low complex modulus |G*| regions. The phase-separated samples of symmetric composition pass a maximum and then descend with decreasing complex modulus |G*|, whereas the phase-separated ones of off-symmetric composition have a developed vale between the maximum and the 90 plateau. The low-frequency phase angle  of the BIT sample is close to 90 , which is indicative of a flow behavior presented by the viscoelastic fluid. As the polymer loadings increase to 5–7 wt%, the low-frequency  decreases remarkably to lower than 45 , indicating a rheological fluid–solid transition in that ternary system. The transitions for the PmB can be indicated around 35–38 C and about 63–73 C. These transitions are almost smeared after aging (after RTFOT) and are just slightly observable for the PmB with the 5 and 7 wt% of SBS-L at about 48–50 C. These characteristic behaviors should be corresponding to droplet–matrix or fibril morphologies, respectively. The presented results suggest that the effect of the SBS-L is more pronounced in the terminal zone (high tan ). This can be understood if one considers that the characteristic time of the measurement should be such that the effects of phase separation are probed.

2D correlation spectrum map In this scheme, a set of rheological spectra of complex modulus G*(T, wt%), where T is the temperature (in degree celcius) and wt% is the SBS content, as presented in the 2D cross-correlation mapping representation provides a surprisingly simple and direct method for detecting the transition temperatures. The location of the minima or maxima in a 2D gradient map enables us to identify thermal changes in SBS blend environment in PmB. The interphase transitions for the PmB can be indicated around 5–10 C and most probably arises from a maltene–asphaltene interfacial region of mixed composition likely rich in resins. Another transition (increasing with polymer content) at about 45 C is corresponding to the 70/100 BIT softening point. Transition at higher temperature, around 63–73 C, corresponds to the polystyrene-rich phase glass transition of mixed polystyrene–polybutadiene phase, with a maximum at 68 C.

Morphology analysis and particle size of SBS-modified BIT The compatibility between polymer and BIT is critical to the properties of PmB.53 The microstructure of PmB analyzed the distribution and fineness of polymer in the BIT matrix using the fluorescence microscopy for a better understanding of the properties of the PmB. The micrographs of 2–4 wt% SBS-L and 2–5 wt% SBS-R PmB show the particulate structure, confirming the formation of a weak gel at this concentration. However, large flat BIT areas are present. A flat-like structure was formed within the 2–3 wt% samples and fibrils were much more numerous for 4–7 wt% PmB. There was almost no interconnection between SBS particulates and flat BIT. However, a significantly stronger gel was formed when the concentration of SBS was further increased to 5–7 wt%. The fluorescence images showed a fibril-like structure with significant interactions forming a 3D network. The gel

Downloaded from jep.sagepub.com by guest on July 30, 2014

12

Journal of Elastomers & Plastics

Table 2. Fluorescence microscopy images of SBS-L-modified BIT with the composition of samples. SBS-L (wt%)

Before RTFOT

After RTFOT

3

4

7

SBS-L: styrene–butadiene–styrene linear; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test

network was getting denser with increasing the concentration of SBS in PmB. The coalescence of the polymer droplets can be observed by studying fluorescence images of the PmB as the blend is annealed by RTFOT. BIT and polymer mixture structure, demonstrated by the fluorescence images, consists of a dispersion of round BIT droplets in a polymer matrix,53 without shear stress (Tables 2 and 3). However, when SBS polymer content

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

13

Table 3. Fluorescence microscopy images of SBS-R-modified BIT with the composition of samples. SBS-R (wt%)

Before RTFOT

After RTFOT

2

4

7

SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

increases, the round forms change to a wavy lens form and then to the elongated, fibril forms. In order to confirm our speculation, we have investigated the resulting morphologies of the phase-separated samples with fluorescence microscopy. The variations of average size distribution with polymer content are summarized in Table 4. It is evident that the median size apparently shifts toward smaller size as the sample changes from a viscous fluid-like sample to a viscoelastic sample to an elastic solid-like sample (large

Downloaded from jep.sagepub.com by guest on July 30, 2014

14

Table 4. DF,

Journal of Elastomers & Plastics

AP

of SBS-L- and SBS-R-modified BIT before and after RTFOT. DF,

AP

SBS-L SBS (wt%) 2 3 4 5

SBS-R

Before RTFOT

After RTFOT

Before RTFOT

After RTFOT

– 1.413 1.668 –

1.197 1.491 1.493 –

1.581 1.567 1.384 1.389

1.397 1.380 1.381 1.465

DF, AP: area–perimeter-derived fractal dimensions; SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene– butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

amount of polymer). As shown, droplet–matrix and fibril domains are observed for the initial PmB blends with 2–4 wt% and for 7 wt% of SBS-L, respectively. In the images, the SBS-L-rich regions (light or white regions) are undoubtedly dispersed phase. These observations confirm that the co-continuous and droplet–matrix morphologies are responsible for the characteristic complex rheological behavior mentioned above. The compatibility between polymer and BIT is critical to the properties of PmB.54 The morphology of PmB is investigated using fluorescence microscopy by characterizing the distribution and the fineness of polymer in the BIT matrix. The morphology of SBS-L- and SBS-R-modified BIT samples is shown in Tables 2 and 3. The lightweight SBS fractions appear in the continuous BIT phase at low polymer content (2 wt% SBS), as small polymer droplets are swollen by BIT. By increasing the polymer concentration, although polymer continuity disperses in the BIT, a continuous polymer phase tends to appear in the studied systems. Thus, the final volume fraction of the polymer-rich phase is clearly higher than the initial one due to the swelling by maltenic components of the base BIT. No significant changes are found with the addition of SBS-R (low content) in the modified BIT sample. The general trend has been observed on aging in the SBS-modified BIT that the binder becomes more homogeneous due to both polymer chain scission and a better compatibility of the smaller polymers chains with the BIT molecules is achieved.55 However, the morphology of the polymer is quite different from the BIT modified with increasing the polymer content.56 Polymer/BIT compatibility is indeed a dynamic concept and that compatible systems are those with a slow creaming rate. In fluorescence microscopy technique, it is not difficult to measure the real sizes of polymer inclusions in BIT matrix. As the content of SBS is increased, up to 4–5 wt%, the modified BIT of relative uniformity is obtained. It can be seen that the white SBS particles disperse in the BIT matrix with the size of around 30 mm or less. At the polymer concentration of 3 wt%, the content of maltenes (more compatible with the polymer) is probably sufficient to swell the macromolecules without inducing instability in the micellar structure of the BIT. BIT is traditionally considered as a dynamic colloid system consisting of a suspension of highmolecular weight micelles dispersed in a lower molecular weight oily medium.57 The

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

15

PmB samples have a tendency to separate into two phases, one is (SBS) polymer-rich phase and the other BIT-rich phase, because the introduction of any polymer disturbs the dynamic equilibrium and reduces the homogeneity of the BIT system.58 The fluorescence image brightness increases with the increase in SBS content and decreases with RTFOT aging.59 DF,AP, derived from the fluorescence images (Table 4), is almost stable for the SBS-L-modified BIT. DF, AP decreases for SBS-R-modified BIT, before RTFOT, and are constant after RTFOT. The size but not the shape of the SBS-rich phases changes with RTFOT aging. The insensitivity of fractal dimensions DF, AP with SBS content confirms, within the scale range 1—30 mm, that aggregation and fibrillating phenomena are dominating the fractal nature of these PmB gels. However, a slight decrease in fractal dimensions DF, AP is manifested with the SBS content (before RTFOT; Table 4) thus indicating that the surface smoothing has some effect on this decrement. The PmB adopts a surface fractal dimension value that depends on the rugosity of the final material after the aging process (RTFOT). The surface fractal dimension is larger for samples with higher SBS content subjected to RTFOT than for materials treated at higher temperatures. The structure and distribution of the polymer varies with the polymer modifier content and that the structure of the SBS polymer remains almost intact when exposed to RTFOT. The compatibility of the components in the SBS-based PmB or the effect of the compatibilizer agents was qualitatively evaluated by comparing the morphology or the sample fluorescence micrographs. The extension or degree of interaction between the components can be related to a higher or lower homogeneity in the shape and distribution of the light areas (corresponding to the polymer) in a dark BIT matrix.60–62 There is more homogeneity in the micrograph of the SBS-based PmB modified with SBS-L (Table 2) compared with the sample modified with SBS-R (Table 3), in which the SBS-rich phase are larger and longer. Thus, it is possible to infer that for the mix conditions here employed, the SBS-L is better dispersed than SBS-R due to a more favored interaction between the SBS linear structure and BIT. The polymer dispersion within BIT is distinctly fine throughout the BIT, at lower polymer content. However, the biphase polymer particles surrounded by continuous BIT changes into a network formation of polymer and BIT fibrils. The SBS polymer elongates with the increase in modifier content to improve the BIT binder resistance to deformation. The morphology studies can assert the physical interactions that have occurred during processing. The polymer may dissolve and/or disperse into BIT, enhancing the mechanical properties of the mixture. The modified BIT has gained improved viscoelastic behavior with remarkable enhancement in the mechanical properties by the addition of SBS due to the physical and chemical interactions.

Monofractal analysis by modified box counting method Color or black-and-white pictures of PmB may be analyzed to provide the percentage of polymer areas or volume. As useful as the information contained in these percentages may be to predict the extent and the kinetics of preferential phase separation in PmB, one would undoubtedly want a more detailed description of the geometry of polymer patterns and some way to relate this geometry to known morphological features of the PmB.63,64

Downloaded from jep.sagepub.com by guest on July 30, 2014

16

Journal of Elastomers & Plastics

Figure 5. Spectra of the SBS-rich phase fractal indices D[WþBW] before RTFOT (a) and after RTFOT (b) by the modified box counting method of SBS-L-modified BIT as function of threshold (m) and the SBS-L content: 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt.% (●).SBS-L: styrene–butadiene–styrene linear; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

Fractal analysis is potentially suitable for an objective quantification of spatial heterogeneity because it is believed to be effective in helping to characterize complex systems that are hard to describe using conventional Euclidean geometry. In this respect, the very intricate details exhibited in phase-separated patterns can be characterized by fractal analysis. The calculation of fractal dimension D highly depends on the selected range of threshold values, regardless of the possible existence of fractal nature (Figures 5 and 6). The terms of the fractal analysis by the modified box counting method show a gradual decrease in BIT-rich phase fractal index D[BþBW], or increase of ‘‘interface’’ D[BW], and SBS-rich phase fractal index D[WþBW] with increasing SBS-L content, reflecting

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

17

Figure 6. Spectra of the SBS-rich phase fractal indices D[WþBW] before RTFOT (a) and after RTFOT (b) by the modified box counting method of SBS-R-modified BIT as function of threshold (m) and the SBS-R content: 2 wt% (■), 3 wt% (▲), 4 wt% (×), 5 wt% ( ▲ ), and 7 wt% (●).SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

the disappearance of small features and particles. This behavior is known and can be described in more general terms as fractal coarsening of the surface due to increased phase separation leading to the fibril structure. In practical applications of fractal geometry to real systems, images like those shown constitute the starting point of analyses. Available methods for the evaluation of fractal dimensions however require the number of grayscale levels to be reduced to just two, black and white. In other words, the images need to be thresholded. Various automatic algorithms could be used to this end. The first, or high threshold, special case of thresholding consists of considering any pixel with a grayscale value m < 255 should

Downloaded from jep.sagepub.com by guest on July 30, 2014

18

Journal of Elastomers & Plastics

Table 5. The interface D[BW], BIT-rich phase D[BþBW] and SBS-rich phase D[WþBW] fractal indices of SBS-L and SBS-R-modified BIT, before and after RTFOT, by the modified box counting method (at threshold m ¼ 129). Interface D[BW] fractal index SBS-L (wt%) 2 3 4 5 7 SBS-L (wt%) 2 3 4 5 7 SBS-R (wt%) 2 3 4 5 7 SBS-R (wt%) 2 3 4 5 7

Before RTFOT 0.813 + 0.160 1.662 + 0.028 1.736 + 0.057 1.555 + 0.211 1.669 + 0.115 After RTFOT 1.124 + 0.451 1.701 + 0.000 1.737 + 0.007 1.632 + 0.343 1.910 + 0.023 Before RTFOT 1.593 + 0.274 1.208 + 0.585 1.416 + 0.394 1.613 + 0.313 1.899 + 0.000 After RTFOT 1.684 + 0.000 1.649 + 0.129 1.631 + 0.295 1.707 + 0.283 1.817 + 0.000

BIT-rich phase D[BþBW] fractal index

SBS-rich phase D[WþBW] fractal index

2.000 + 1.998 + 1.997 + 1.998 + 1.953 +

0.000 0.000 0.001 0.002 0.017

0.870 + 1.639 + 1.710 + 1.535 + 1.799 +

0.171 0.026 0.053 0.209 0.062

1.999 + 1.998 + 1.993 + 1.998 + 1.968 +

0.001 0.000 0.003 0.002 0.023

1.157 + 1.669 + 1.738 + 1.600 + 1.931 +

0.418 0.000 0.006 0.338 0.022

1.999 + 2.000 + 1.997 + 1.995 + 1.994 +

0.001 0.000 0.004 0.003 0.000

1.561 + 1.201 + 1.409 + 1.619 + 1.873 +

0.272 0.520 0.390 0.288 0.000

1.998 + 1.998 + 1.999 + 1.996 + 1.998 +

0.000 0.001 0.000 0.005 0.000

1.660 + 1.626 + 1.607 + 1.681 + 1.772 +

0.000 0.094 0.262 0.291 0.000

SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

become white. Alternatively, adopting a low threshold, one could consider that any pixel with a grayscale level m > 0 should become black. Between these two extremes, one may take as a medium threshold the value that splits the grayscale into two equal parts, pixels with grayscale threshold value m 128 become white, and pixels with grayscale threshold value m > 128 become black. The original image has been transformed into grayscale images. The first approach is to define a good threshold for which we obtain the highest contrast, meaning the number of zeroes proportional to the number of ones (50%) or threshold around nearly m ¼ 128. SBS components of PmB showed fractal self-similar properties that were evaluated by fractal dimensions obtained by threshold at m ¼ 129 (Table 5). However, the fractal indices at threshold m ¼ 129 threshold did not revealed significantly different values that would enable the discrimination of SBS content effect. To threshold or ‘‘segment’’ digitized image, one

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

19

could in principle proceed by trial and error until one achieves thresholding that appears reasonable, that is, coincides with some a priori idea one may have about the two categories of pixels one attempts to separate. Unfortunately, this procedure is very subjective and may lead to biases when one is trying to compare images, or in the analysis of time sequences of images of a given object, for example, under evolving lighting conditions. To palliate these difficulties, we have adapted the 2D correlation approach to extract the threshold that is needed to discern the SBS content effect. The basic idea is to transform an image of PmB to a number that can be used to classify the total image. The interface D[BW] and SBS-rich phase D[WþBW] fractal indices (at threshold m ¼ 129) by the modified box counting method of modified BIT as function of the SBS-L content before and after RTFOT level for SBS-L content above 3 wt% indicating phase transition. The BIT-rich phase fractal indices D[BþBW] before and after RTFOT just drop above 5 wt% of SBS-L. In the 2D correlation spectra approach, set of fractal index spectra A(m, wt%), where m is the threshold and wt% is the SBS content, is presented in the 2D cross-correlation mapping representation that plots the values of the fractal indices over the space of SBS content versus threshold provides a surprisingly simple and direct method for detecting the phase transition. The data in a 2D spectrum are usually presented in the form of a contour plot, which resembles a topographical 3D map. The ordinate and abscissa dimensions are threshold m values. The cross-correlations of fractal indices are shown in the third dimension, above and below the plane of the plot, (both positive and negative). The construction of such a plot helps to directly visualize the entirety of complex spectral events occurring during a phase transition phenomenon. Autopeaks that are along the diagonal do not really give any information. The cross-peaks that are positioned off the diagonal are helpful. The location of the minima or maxima in a 2D gradient map enables us to identify changes in SBS blend environment at specific threshold which are undergoing the phase transition process. The 2D sample correlation spectra of the interface D[BW] and BIT-rich phase D[BþBW] fractal indices for SBS-L-modified BIT as functions of polymer content (before RTFOT) reveal in the asynchronous and argument maps cross-peaks at threshold values nearly m ¼ 30 and m ¼ 55 and autopeak at threshold value m ¼ 30. The 2D sample correlation spectra of the fractal indices interface D[BW] and BIT-rich phase D[BþBW] for SBS-L-modified BIT as functions of polymer content after RTFOT do not produce peaks due to very similar morphology. However, the fractal indices D[BW], D[BþBW], and D[WþBW] (at threshold m ¼ 55) for SBS-L-modified BIT as functions of polymer content indicate the phase transition at 4 wt% of SBS-L before RTFOT. The morphology after RTFOT does not change much by polymer content, but, still, the fractal indices are larger than before RTFOT. The differences between the interface D[BW], BIT-rich phase D[BþBW], and SBS-L-rich phase D[WþBW]fractal indices before and after RTFOT are minimal around 4 wt% of SBS-L and is increasing afterward with the polymer content above 4 wt% of SBS-L. The differences between the interface D[BW] and SBS-R-rich phase D[WþBW] before and after RTFOT are decreasing with the polymer content above 2 wt% of SBS-R. Differences in BIT-rich phase D[BþBW] fractal indices before and after RTFOT are minor. The differences between the interface D[BW], BIT-rich phase D[BþBW], and SBS-L-rich

Downloaded from jep.sagepub.com by guest on July 30, 2014

20

Journal of Elastomers & Plastics

Figure 7. 2D sample correlation asynchronous spectrum of the interface fractal indices D[BW] (a) and the SBS-rich phase fractal indices D[WþBW] (b) for SBS-R-modified BIT as functions of polymer content, before RTFOT. 2D: two-dimensional; SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

phase D[WþBW] fractal indices before and after RTFOT are decreasing with the larger value of the corresponding fractal indices before RTFOT. The 2D sample correlation spectra of the fractal indices D[BW] and D[WþBW] for SBS-R-modified BIT as functions of polymer content (before RTFOT) (Figure 7) reveal that in the asynchronous and argument maps, cross-peaks occur at threshold values nearly m ¼ 129. Other thresholds are not so sensitive. The 2D sample correlation spectra of the fractal indices D[BW] and D[BþBW] for SBS-R-modified BIT as functions of polymer content after RTFOT (Figure 8) do not produce cross-peaks, just weak autopeaks, due to very similar morphology. However, the fractal indices D[BW], D[BþBW], and D[WþBW] (at threshold m ¼ 129) for SBS-R-modified BIT as functions of polymer content indicate the phase transition at 3–4 wt% of SBS-R before RTFOT. The morphology after RTFOT does not change much by polymer content, but, still, the fractal indices are larger than before RTFOT. Spatial heterogeneity is an important property displayed by PmB that has applicability consequences. In PmB, the fibril network pattern of the polymer phase may not be maintained. Areas not well homogenized may exhibit a greater degree of phase separation. The spatial distribution may also vary functionally over the course of (aging) time. The extent of spatial heterogeneity is not well described by just imaging analysis methods, but it may be well be by quantitative fractal analysis methods. Larger fractal index value indicates that the texture of the image fluctuates markedly. Fluctuation implies the appearance of increased roughness in the overlapping region of BIT-rich and SBS-rich phases. Therefore, the texture roughness increases as fractal indices increases. Because the morphology of phases can influence their properties, our results implied that the fractal analysis has importance in PmB characterization.

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

21

Figure 8. 2D sample correlation synchronous spectrum of the interface fractal indices D[BW] (a) and the SBS-rich phase fractal indices D[WþBW] (b) for SBS-R modified BIT as functions of polymer content, after RTFOT. 2D: two-dimensional; SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test. Table 6. Dh of SBS-L- and SBS-R-modified BIT, before and after RTFOT. Dh SBS-L

SBS-R

SBS (wt%)

Before RTFOT

After RTFOT

Before RTFOT

2 3 4 5 7

1.448 + 1.754 + 1.788 + 1.583 + 1.427 +

1.602 + 1.858 + 1.761 + 1.725 + 1.665 +

1.675 1.666 1.660 1.689 1.645

0.149 0.003 0.036 0.194 0.088

0.202 0.000 0.014 0.183 0.034

+ 0.166 + 0.304 + 0.091 + 0.109 + 0.351

After RTFOT 1.527 + 1.635 + 1.760 + 1.822 + 1.685 +

0.355 0.000 0.029 0.020 0.000

Dh: Higuchi fractal dimension; SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

Higuchi’s fractal algorithm For calculation of fractal dimension of fluorescence images, one may use quick and easy data series algorithm proposed by Higuchi24 (Table 6). The Higuchi dimension of the 2D images is calculated using Higuchi’s algorithm, and it is shown that both regions of interests and directional dependencies can be evaluated independently of the whole picture.25 The calculations in a particular x- or y direction reveal that the Higuchi fractal dimensions get through a maximum around 3–4 wt% of SBS-L/BIT mixtures before RTFOT, and after RTFOT are almost constant. Also, the Higuchi Dh fractal

Downloaded from jep.sagepub.com by guest on July 30, 2014

22

Journal of Elastomers & Plastics

Table 7. Dfft of SBS-L and SBS-R modified bitumen, before and after RTFOT. Dfft SBS-L

SBS-R

SBS (wt%)

Before RTFOT

After RTFOT

Before RTFOT

After RTFOT

2 3 4 5 7

0.781 + 0.469 + 0.509 + 1.037 + 0.348 +

0.524 + 0.747 + 0.420 + 1.282 + 0.826 +

0.701 + 0.101 0.757 + 0.018 0.831 + 0.421 0.833 + 0.524 0.875 + 0.207

0.665 1.082 1.090 0.909 1.681

0.187 0.117 0.014 0.346 0.149

0.049 0.000 0.029 0.499 0.956

+ 0.156 + 0.000 + 0.713 + 0.020 + 0.000

FFT: fast Fourier transform; Dfft: FFT fractal dimension; SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

dimensions are almost constant for SBS-R/BIT mixtures (before RTFOT) and get through a maximum around 5 wt% of SBS-R after RTFOT. The results show that the extracted fractal features are distinct for spherical and fibril phase-separated regions of SBS/BIT mixtures.

FFT fractal dimension FFT fractal dimension Dfft (Table 7) is considerably higher after RTFOT than before RTFOT, for high (5 wt%) SBS-L content. At lower SBS-L content, the FFT fractal dimensions Dfft before and after RTFOT are not discernibly different. However, for the SBS-R-modified BIT, the FFT fractal dimension Dfft (Table 7) is considerably higher after RTFOT than before RTFOT.

GLCM image texture The GLCM entropy of modified BIT, before and after RTFOT, increases with the SBS-L content indicating increase in disorder and heterogeneity. The entropy after RTFOT is larger than before RTFOT, indicating apparent heterogeneous morphology in the SBS-L-modified BIT after RTFOT. The GLCM entropy of SBS-R-modified BIT is almost same before and after RTFOT. However, a step transition is observed at about 4–5 wt% of SBS-R, indicating more heterogeneous morphology for higher SBS-R loadings. Still the GLCM entropy of modified BIT before and after RTFOT is linearly correlated: SBS-L (R2 ¼ 0.873) and SBS-R (R2 ¼ 0.985). The heterogeneity is induced stronger for the SBS-L-modified BIT than for the SBS-R-modified BIT. When the image is not texturally uniform, many GLCM elements have very small values, which imply that entropy is very large. Therefore, entropy is inversely proportional to GLCM ASM term (Table 8). The critical SHRP temperatures for permanent deformation are increasing with the increase of the GLCM entropy term for the SBS-L-modified BIT; however, this relation is weak for the SBS-R-modified BIT. Elastic recovery (at 25 C) is related to the GLCM correlation, IDM, and entropy terms for the SBS-L-modified BIT (weakly

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

23

Table 8. GLCM terms of modified BIT as function of SBS-L polymer content (before and after RTFOT), ASM, Con., Corr, IDM, and Ent. GLCM terms ASM103 Con. Corr 103 IDM Ent. ASM103

Con.

Corr 103 IDM Ent.

SBS-L SBS (wt%) 2 3 4 5 7

Before RTFOT 27 4 3 0.75 3

23.4 321.6 379.6 174.1 93.7

3 1 0.59 2 0.43

After RTFOT 0.94 0.43 0.29 0.21 0.34

4.05 6.67 7.46 7.77 7.74

23 2 4 0.38 4

182.5 512.0 242.6 1056 570.7

3 0.53 0.62 0.42 0.24

0.73 0.24 0.34 0.10 0.20

4.47 7.83 7.12 8.99 8.72

0.47 0.65 0.61 0.17 0.12

6.35 5.54 5.17 8.48 8.14

SBS-R Before RTFOT 2 3 4 5 7

2 6 13 0.20 0.24

192.3 304.5 125.0 446.5 515.7

1 1 1 0.82 0.48

After RTFOT 0.34 0.45 0.56 0.09 0.11

7.15 6.12 5.67 8.95 9.08

5 100 17 0.39 0.45

361.0 172.5 89.75 324.7 288.0

0.75 0.44 2 0.75 2

GLCM: gray-level co-occurrence matrices; ASM: angular second moment; Con.: contrast; Corr: correlation; IDM: inverse different moment; Ent.: entropy; SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene– butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test; BIT: bitumen.

before and strongly after RTFOT). However, the SBS-R-modified BIT elastic recovery (at 25 C) is not at all related to any of the GLCM terms. Ring and ball temperature is strongly increasing with the increase of the GLCM entropy term for the SBS-L- and SBS-R-modified BIT (except for the SBS-R-modified BIT after RTFOT).

Saliency Saliency (the combination of size and contrast) indicates how important this object is relative to other objects. Saliency of modified BIT with the SBS-L polymer increases with the polymer content (Table 9). Saliencies of modified BIT with the SBS-R polymer are almost constant and slightly fall down with the polymer content (Table 9). Saliencies of the PmB with the SBS-L content larger than 5 wt% are stable to aging due to the formation of continuous elastic network formed by the SBS-L fibrils. The saliencies of the PmB with the SBS-R are stable to the aging without any apparent change. The overall change in saliency is very small and not significant. Still, saliency of the SBS-Lmodified BIT, before and after RTFOT, is relatively well correlated to the length of the phase-separated particles. Saliency of the SBS-L- and SBS-R-modified BIT, before and after RTFOT, is decreasing with the increase in the particle/fibril (90 percentile) length.

Downloaded from jep.sagepub.com by guest on July 30, 2014

24

Journal of Elastomers & Plastics

Table 9. Effect of polymer content on the saliency of modified BIT as function of SBS-L and SBS-R modified BIT (before and after RTFOT). Saliency SBS-L

SBS-R

SBS (wt%)

Before RTFOT

After RTFOT

2 3 4 5 7

6917 + 8017 + 7821 + 8632 + 7472 +

7543 9041 6963 8523 7589

18 2073 1716 4040 1028

+ 1295 + 2672 + 760 + 1983 + 1413

Before RTFOT 8558 8280 8054 8431 8067

+ 1947 + 1939 + 1943 + 2114 + 1555

After RTFOT 7656 8343 7780 8014 9696

+ 1433 + 2872 + 1912 + 1593 + 4223

SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test; BIT: bitumen.

Rheological relations to morphology As already described, the rheological properties of PmB are affected by the phase structure or polymer morphology in the BIT.65 A fine dispersion of polymers and a continuous polymer phase in the modified binders are normally desired to achieve an optimal rheological improvement.66,67 The modified BIT þ 5 wt% SBS-L þ 5 wt% SBS-R have displayed homogeneous polymer network morphology, thus showing a rubbery-like behavior at phase angle of 30 –60 and stiffness level around 1 MPa. However, for the other two modified binders, the polymer is observed as a separated phase (the light objects in fluorescence images); consequently, the rubber-like behavior is much weaker. The 2D-sample correlation spectrum maps of the SBS-L-modified BIT, before RTFOT, reveal that to the interface fractal index D[BW] at threshold levels around threshold m ¼ 30 (positive correlation) and m ¼ 210 (negative correlation). The high-temperature complex modulus is related mainly to the interface fractal index D[BW] at threshold levels around m ¼ 100–150 (positive correlation). The BIT-rich phase fractal index D[BþBW] is related to the low-temperature complex modulus at threshold levels around m ¼ 30 (positive correlation) and related to the high-temperature complex modulus at threshold levels around m ¼ 65 (negative correlation). The SBSrich phase fractal index D[WþBW] is related to the low-temperature complex modulus at threshold levels around m ¼ 210 (negative correlation) and related to the hightemperature complex modulus at threshold levels around m ¼ 100–150 (positive correlation; Figures 9 and 10).

Rheological fractal dimension The storage modulus (G0 ) and loss modulus (G00 ) change with temperature, between 5 C and 80 C, for (0–7 wt%) SBS PmB. G0 and G00 decreased gradually, up to gel point, about 10–40 C. At the gel point, it was frequently found, for chemical irreversible gels and also for physical gels of biopolymers,68–71 that G0 < G00 , while for physically

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

25

Figure 9. 2D sample correlation spectrum map of the logarithm of the complex modulus (ln G*) (in temperature range from 5 C to 80 C; abscissa) and the SBS-rich phase fractal index D[WþBW] (image threshold m ¼ 0–254; ordinate) of the SBS-L-modified BIT, before RTFOT as functions of polymer content: (a) synchronous, (b) asynchronous. 2D: two-dimensional; SBSL: styrene–butadiene–styrene linear; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

Figure 10. 2D-sample correlation spectrum map of the logarithm of the complex modulus (ln G*) (abscissa: temperature range from 5 C to 80 C;) and the SBS-rich phase fractal index D[WþBW] (ordinate: threshold m ¼ 0–254;) of the SBS-R-modified BIT, before RTFOT as functions of polymer content: (a) synchronous and (b) asynchronous. 2D: two-dimensional; SBS-R: styrene–butadiene–styrene radial; BIT: bitumen; RTFOT: Rolling Thin Film Oven Test.

Downloaded from jep.sagepub.com by guest on July 30, 2014

26

Journal of Elastomers & Plastics

Table 10. Linear regression terms of the ‘‘gel point’’ temperature versus SBS weight fraction (%) for SBS-L- and SBS-R-modified BIT. The confidence criteria are acceptable at Fsignif < 0.05.

SBS-L Before RTFOT After RTFOT SBS-R Before RTFOT After RTFOT

Linear slope

Linear intercept

R2

Fsignif(1,4)

2.67 + 0.87 2.97 + 0.62

7.40 + 3.60 18.5 + 2.6

0.702 0.848

0.037 0.009

1.90 + 0.33 2.87 + 0.43

10.91 + 1.79 18.98 + 2.60

0.891 0.917

0.004 0.002

SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test; BIT: bitumen.

reversible gels,72–74 it has been observed that G0 > G00 . In the present work, the PmB showed G0 smaller than G00 , in the range above ‘‘gel point’’ temperatures. The volume fractions ’ occupied by the SBS particles/fibrils were calculated from the mass fraction, w, that is, ’ ¼ w/, where  is the density. From plots of elastic modulus (G0 ) of the PmB with temperature, the gel point values, G0 were determined.75 G0 (¼G00 ) values at the gel point temperature of PmB dispersions were plotted against SBS content ’. The double logarithmic plot76 resulted in reasonable straight lines (R2  0.82) for both SBS types, fulfilling the power law relationship inferred by weak link approach: G0 / j1=3Df . The fractal dimensions of these two types of SBS-modified BIT, before RTFOT, were calculated from the slope of the lines as Df ¼ 1.7400 (+0.0008) for SBS-L and Df ¼ 1.9125 (+0.0004) for SBS-R. The fractal dimensions of the SBS-modified BIT, after RTFOT, strongly increases, Df ¼ 2.5141 (+0.0019), for SBS-L and is almost the same for SBS-R, Df ¼ 1.9516 (+0.0003). This was interpreted as PmB, with SBS-L after RTFOT, having morphology with highly convoluted surfaces. The higher G0 values of SBS-L after RTFOT PmB were attributed to the higher rigidity of their structure. The structural and dynamic properties of PmB are based on the idealized model of (SBS-) elastomeric and BIT matrices. For the dynamic properties, the time–temperature–concentration superposition principle applies well for the relaxation of PmB. This scaling behavior has the underlying implication that the rheological properties (i.e. steadystate shear viscosity) of PmB possess thermorheological simplicity. In addition, interestingly, the gel point temperature (or temperature where G0 ¼ G00 ¼ tan  ¼ 1) as a function of SBS concentration exhibits Arrhenius temperature dependence, if the concentration variation is regarded as an inverse of gel point temperature variation. The introduced SBS exerts the similar effect to the thermodynamic variables (e.g. pressure and temperature) on the polymer dynamics. The gel point temperature linearly depends on the SBS weight fraction (%) for SBS-L- and SBS-R-modified BIT, before and after RTFOT (Table 10). The gel point temperature for BIT is then, as extrapolated, 9.2 C (+2.5 C) and 18.7 C (+0.3 C), before and after RTFOT respectively. Finally, the logarithm of the SBS weight fraction concentration scales linearly as a function of the inverse of the gel point temperature. The results indicate an Arrheniuslike relationship, which is in good agreement with equation:

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

27

Table 11. DEa (kJ mol1; from the logarithm of the SBS weight fraction versus the inverse ‘‘gel point’’ temperature) for SBS-L- and SBS-R-modified BIT. The confidence criteria are acceptable at Fsignif < 0.05.

SBS-L Before RTFOT After RTFOT SBS-R Before RTFOT After RTFOT

DEa (kJ/mol)

R2

Fsignif(1,3)

37.2 + 9.9 47.3 + 15.2

0.826 0.763

0.03 0.05

67.4 + 13.3 56.8 + 15.5

0.895 0.817

0.01 0.03

DEa: apparent activation energy; SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test; BIT: bitumen.

  Ea j ¼ j0 exp RTgel where ’ is the SBS weight fraction, R is the gas constant, and Tgel is gel point absolute temperature in kelvin. The smaller the value of the gel point apparent activation energy Ea and the higher the temperature, the smaller is the SBS weight fraction for gelling and the faster is the gelling (Table 11). The value of Ea is apparently larger for the SBS-R PmB (before RTFOT), which is attributed to the extra and stronger physical bonds formed between the BIT chain segments and the SBS-R polymer inclusion surface, causing the increase of the viscosity of SBS-R PmB. However, according to the figures’ standard deviation, the calculated gel point apparent activation energy Ea is almost the same for the SBS-L and SBS-R systems, before and after RTFOT. Physically, this suggests that the reduction of the viscosity, driven by an imposed steady-state shear rate, exhibits the equivalent result for the viscosity at a higher temperature, or for the case at a much lower SBS concentration. Adopting this scaling principle will enable us to accurately obtain the viscosity at low shear rate, which is difficult to accurately measure at low shear rate in a single experiment of PmB. Moreover, for high SBS loadings, it is also difficult to accurately measure the viscosity. Accordingly, we can equivalently measure the viscosity at a lower SBS concentration at a lower temperature and use the above scaling principle to obtain the one at high SBS loadings. These implications will find great contributions in the application of processing for PmB. Presumably, SBS in SBS-modified BIT increases in apparent volume due to swelling with maltenes incorporated from BIT and, at the same time, is dispersed into BIT, changing from coarse to fine morphologies.77,78 The microstructure of SBS-modified BIT can be broadly divided into the type in which BIT forms a continuous phase (hereinafter referred to as the ‘‘as-network type’’) and the type in which SBS forms a continuous phase (hereinafter referred to as the ‘‘SBS network type’’). The SBS concentration at which the change from the as-network to SBS network types occurs is generally given79 to be 6–8%. Rheological testing was a straightforward and precise way to determine the gel point temperature of the PmB system. This work shows that the development of the ‘‘networktype’’ morphology increases the gel point, indicating that fractal analysis can quantify this

Downloaded from jep.sagepub.com by guest on July 30, 2014

28

Journal of Elastomers & Plastics

Figure 11. The ‘‘gel point’’ temperature versus the ‘‘interface’’ D[BW] fractal index (threshold m ¼ 128); SBS-L: (^) before RTFOT, (&) after RTFOT; SBS-R: (~) before RTFOT, ( ) after RTFOT. SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

development. The gel point temperature increases with the interface D[BW] (threshold m ¼ 128; Figure 11), SBS-rich phase fractal index, and BIT-rich phase interface fractal index. Also, the gel point temperature increases with the SBS-rich phase D[WþBW] fractal index (threshold m ¼ 55) (Figure 12). Physical gelation and morphological structural ordering are related processes. Morphological ordering into apparent structures requires molecular mobility. However, morphological structure reduces mobility and may cause gelation. This gelation phenomenon will interrupt the morphological structuring process. Microphase-separating SBS block copolymers-modified BIT also show this interplay of gelation and morphological structuring, as determined by the SBS-rich phase D[WþBW] fractal index. However, it is obvious, and already shown, that the aging in RTFOT conditions influences BIT and SBS–polymer chemistry relating to the loss of volatiles and oxidation of the material (our results are not shown here). For evaluation of aging properties of PmB, chemical, rheological, and morphological methods are thus required. The Higuchi Dh nor the Fourier Dfft fractal dimensions cannot efficiently describe the gel point temperature, although some trend can be indicated that fractal features distinctively affect the gel point temperature before and after RTFOT. The failure of Higuchi or Fourier fractal dimension can be ascribed to the orientation source of these fractal dimensions determinations, inherent in these fractal dimensions calculations.80

Conclusions Our work provides some initial quantification of changes in morphology at the set SBS content. Polymer distribution of polymer/BIT mixture had obvious fractal characteristics. The morphology distribution of phase-separated PmB mixture was deduced based on the

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

29

Figure 12. The ‘‘gel point’’ temperature versus the SBS-rich phase D[WþBW] fractal index (threshold m ¼ 55); SBS-L: (^) before RTFOT, (&) after RTFOT; SBS-R: (~) before RTFOT, ( ) after RTFOT. SBS-L: styrene–butadiene–styrene linear; SBS-R: styrene–butadiene–styrene radial; RTFOT: Rolling Thin Film Oven Test.

fractal model; in general, the larger value of fractal dimension means more complex geometrical shape. Relationship between morphology and PmB properties may be better understood through exploring the fractal dimensions of the microstructure. This article used linear regression analysis to correlate fractal dimension of PmB mixture and the rheology of the mixture. Test results indicated that the fractal dimension of PmB mixture was controlled by the polymer content and dispersion. On the other hand, rheology is related to the fractal dimension. The fractal dimension can be used as a quantification index of PmB properties evaluation. The improved viscoelastic properties of the PmB have been demonstrated using the rheological parameters of high-temperature viscosity and (G*/sin ) values. The van Gurp–Palmen plots showed some sensitivity of rheological behavior toward different obtained morphologies. Image analysis confirms that the morphologies of the phase-separated samples are responsible for such fingerprints of these rheological functions, which can be qualitatively corroborated by utilizing fractal model. The comparison of the different correlations between rheological functions has shown that the van Gurp–Palmen plots are most sensitive to phase separation and morphology. They are the most appropriate for characterizing phase separation and morphology. The rheology is a sensitive tool for analyzing the structure of materials. Moreover, the success in the qualitative analysis of morphology by rheology made it possible to quantitatively evaluate the rheological properties of phase transition in two-phase polymer blends. The rheological properties of road BIT are improved by means of SBS-L and SBS-R polymer modification as proven by both conventional and rheological parameters complex modulus G*, complex viscosity *, and phase angle  obtained by DSR as shown in our previous research.5 Fractal dimension is a parameter of major importance in the study of PmB phase separation since it characterizes the way polymer physically penetrates into the BIT phase. Aging

Downloaded from jep.sagepub.com by guest on July 30, 2014

30

Journal of Elastomers & Plastics

process (RTFOT) can then infer that irregular SBS-phase rich aggregates with high borderbased fractal dimensions (Df close to 2) become spherical SBS-phase rich particles (Df close to 1) as a result of condensation and coagulation processes (due to RTFOT aging). Hence, both fractal and GLCM analyses confirm that the morphological reorganization of PmB, attributable to a gain of structural complexity, occurs at low SBS content and after RTFOT aging. Information obtained with the quantification of SBS-phase rich and BIT-rich phase morphology may improve the PmB climate and aging models. Funding The present work was financially supported by the Ministry of Science, Education and Sports of the Republic of Croatia (grant no. 125-125-2971-2578).

References 1. Lu X and Isacsson U. Modification of road bitumens with thermoplastic polymers. Polym Test 2000; 20: 77–86. 2. Lu X and Isacsson U. Chemical and rheological evaluation of ageing properties of SBS polymer modified bitumens. Fuel 1998; 77: 961–972. 3. Lu X and Isacsson U. Rheological characterization of styrene-butadiene-styrene copolymer modified bitumens. Const Build Mater 1997; 11: 23–32. 4. Rek V and Barjaktarovic´ ZM. Dynamic mechanical behavior of polymer modified bitumen. Mater Res Innov 2002; 6: 39–43. 5. Ocelic´ Bulatovic´ V, Rek V and Jurkasˇ Markovic´ K. Effect of polymer modifiers on the properties of bitumen. J Elast Plast. Epub ahead of print 22 January 2013. DOI: 10.1177/ 0095244312469964. 6. Lesueur D. The colloidal structure of bitumen: consequences on the rheology and on the mechanisms of bitumen modification. Adv Colloid Interface Sci 2009; 145: 42–82. 7. Yildirim Y. Polymer modified asphalt binders. Const Build Mater 2007; 21: 66–72. 8. Kennedy TW, Hubert GA, Harrigan ET, et al. Superior performing asphalt pavements (Superpave): the product of the SHRP asphalt research program publication SHRP-A-410. Washington, DC: Strategic Highway Research Program, National Research Council, 1994. 9. Zhang Q, Fan W, Wanga T, et al. Influence of emulsification on the properties of styrene-butadiene-styrene chemically modified bitumens. Constr Build Mater 2012; 29: 97–101. 10. Masson J-F, Collins P, Robertson G, et al. Thermodynamics, phase diagrams, and stability of bitumen-polymer blends. Energy Fuels 2003; 17: 714–724. 11. Pe´rez-Lepe A, Martinez-Boza FJ and Gallegos C. High temperature stability of different polymer-modified bitumens: a rheological evaluation. J Appl Polym Sci 2007; 103: 1166–1171. 12. Wilson A, Fuchs G, Scramoncin C, et al. Localization of the polymer phase in bitumen/ polymer blends by field emission cryo-scanning electron microscopy. Energy Fuels 2000; 14: 575–584. 13. Vasiljevic-Shikaleska A, Popovska-Pavlovska F, Cimmino S, et al. Viscoelastic properties and morphological characteristics of polymer-modified bitumen blends. J Appl Polym Sci 2010; 118: 1320–1330. 14. Soenen H, Lu X and Redelius P. The morphology of SBS modified bitumen in binders and in asphalt mix. In: Loizos A, Partl MN, Scarpas T and Al-Qadi IL (eds) Advanced testing and characterization of bituminous materials. Leiden, The Netherlands: CRC Press, 2009, pp. 151–162.

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

31

15. Rahmani NHG, Dabros T and Masliyah JH. Fractal structure of asphaltene aggregates. J Colloid Interface Sci 2005; 285: 599–608. 16. Ocelic´ Bulatovic´ V, Rek V and Jurkasˇ Markovic´ K. Influence of polymer types on bitumen engineering properties. Mater Res Innov 2013; 17: 189–194. 17. Ocelic´ Bulatovic´ V, Rek V and Jurkasˇ Markovic´ K. Polymer modified bitumen. Mater Res Innov 2012; 6: 1–6. 18. Almqvist N. Fractal analysis of scanning probe microscopy images. Surf Sci 1996; 355: 221–228. 19. Lung CW and Mu ZQ. Fractal dimension measured with perimeter-area relation and toughness of materials. Phys Rev 1988; B38: 11781–11784. 20. Available at: http//www.fch.vutbr.cz/lectures/imagesci/ (accessed 27 June 2013). 21. Jelcˇic´ Zˇ, Vranjesˇ N and Rek V. Long-range processing correlation and morphological fractality of compatibilized blends of PS/ HDPE/ SEBS block copolymer. Macromol Symp 2010; 290: 1–14. 22. Jelcic Z, Hauschild K, Ogiermann M, et al. Evaluation of tablet formation of different lactoses by 3D modeling and fractal analysis. Drug Dev Ind Pharm 2007; 33: 353–372. 23. Jelcˇic´ Zˇ, Mastelic´ Samardzˇic´ Z and Zrncˇevic´ S. Fractal analysis of catalysts surface morphologies on hydrogenation in process of 2-((1-benzylpiperidin-4-yl)methyl)-5,6-dimethoxy-2,3-dihydroinden-1-one hydrochloride synthesis. Appl Cat A 2013; 456: 30–43. 24. Higuchi T. Approach to an irregular time-series on the basis of the fractal theory. Phys D 1988; 31: 277–283. 25. Ahammer H. Higuchi dimension of digital images. PLoS One 2011; 6(9): e24796. 26. Vandenberg S and Osborne CF. Digital image processing techniques, fractal dimensionality and scale-space applied to surface roughness. Wear 1992; 159: 17–30. 27. IEEE Standard 610.4-1990. IEEE Standard Glossary for Image Processing and Pattern Recognition Terminology (withdrawn), 1990. 28. McGunnigle G and Chantler M. Rotation invariant classification or rough surfaces. IEEE Proc Vision Image Signal Proc 1999; 146: 345–352. 29. Haraclick RM. Statistical and structural approaches to texture. Proc IEEE 1979; 67: 786–804. 30. Bharati MH, Liu JJ and MacGregor JF. Image texture analysis: methods and comparisons. Chemom Intell Lab Syst 2004; 72: 57–71. 31. Haralick RM, Shanmugam K and Dinstein I. Textural features for image classification. IEEE Trans Syst Man Cybern 1973; 3: 610–621. 32. Weszka JS, Dyer CR and Rosenfeld A. Comparative study of texture measures for terrain classification. IEEE Trans Syst Man Cybern 1976; 5: 269–285. 33. Al-Janobi A. Performance evaluation of cross-diagonal texture matrix method of texture analysis. Pattern Recogn 2001; 34: 171–180. 34. Barber DG and LeDrew EF. SAR sea ice discrimination using texture statistics: a multivariate approach. Photogramm Eng Remote Sensing 1991; 57: 385–395. 35. Shokr ME. Evaluation of second-order texture parameters for sea ice classification from radar images. J Geophys Res 1991; 96: 625–640. 36. Baraldi A and Parmiggiani F. An investigation of the textural characteristics associated with gray level cooccurrence matrix statistical parameters. IEEE Trans Geosci Remote Sens 1995; 33: 293–303. 37. Clausi DA. An analysis of co-occurrence texture statistics as a function of grey level quantization. Can J Remote Sens 2002; 28: 45–62. 38. Available at: http//rsb.info.nih.gov/ij/plugins/download/GLCM_Texture.class (accessed 27 June 2013).

Downloaded from jep.sagepub.com by guest on July 30, 2014

32

Journal of Elastomers & Plastics

39. Zhang L, Tong MH, Marks TK, et al. SUN: a Bayesian framework for saliency using natural statistics. J Vis 2008; 8: 1–20. 40. Noda I and Ozaki Y. Two-dimensional correlation spectroscopy, applications in vibrational and optical spectroscopy. Chichester: Wiley, 2004. 41. Rek V, Vranjesˇ N and Barjaktarovic´ ZM. Evaluation of ageing properties of polymer modified bitumen. Mater Res Innov 2005; 9: 670–691. 42. Rek V and Barjaktarovic´ ZM. Dynamic mechanical behavior of polymer modified bitumen. Mat Res Innov 2002; 6: 39–43. 43. Trinkle S, Walter P and Friedrich C. Van Gurp-Palmen plot: a way to characterize polydispersity of linear polymers Rheol Acta 2001; 40: 322–328. 44. Trinkle S, Walter P and Friedrich C. Van Gurp-Palmen-plot II—classification of long chain branched polymers by their topology. Rheol Acta 2002; 41: 103–113. 45. Eklind H and Maurer FHJ. On the morphology determination of heterogeneous blends by melt-state dynamic mechanical spectroscopy. Polym Networks Blends 1995; 5: 35–45. 46. Van Gurp M and Palmen J. Time-temperature superposition for polymeric blends. Rheol Bull 1998; 67: 5–8. 47. Booij HC and Thoone GP. Generalization of Kramers-Kronig transforms and some approximations of relations between viscoelastic quantities. Rheol Acta 1982; 21: 15–24. 48. van de Ven M and Jenkins K. Rheological characterisation of some (polymer modified) bitumen and bitumen-filler system at compaction and in-service temperatures. In: Proceedings of 6th RILEM symposium performance testing and evaluation of bituminous materials (PTEBM’03) (ed MN Partl), Zurich, Germany, 2003, pp. 88–94. France: RILEM Publications SARL. 49. Li R, Yu W and Zhou C. Rheological characteristic of droplet-matrix versus co-continous morphology. J Macromol Sci B Phys 2006; 45: 889–898. 50. Li R, Yu W and Zhou C. Phase behaviour and its viscoelastic responses of poly (methyl methacrylate) and poly (styrene-co-maleic anhydride) blend systems. Polym Bull 2006; 56: 455–466. 51. Macaubas PHP and Demarquette NR. Time-temperature superposition principle applicability for blends formed of immiscible polymers. Polym Eng Sci 2002; 42: 1509–1519. 52. Schulze D, Roths T and Friedrich C. Classification of model topologies using the  versus G* plot. Rheol Acta 2005; 44: 485–494. 53. Loeber L, Sutton O, Morel J, et al. New direct observations of asphalts and asphalt binders by scanning electron microscopy and atomic force microscopy. J Microsc 1996; 182: 32–39. 54. Lewandowsky MH. Polymer modification of paving asphalt binders. Rubber Chem Technol 1994; 67: 447–481. 55. Mouillet V, Lamontagne J, Durrieu F, et al. Infrared microscopy investigation of oxidation and phase evolution in bitumen modified with polymers. Fuel 2008; 87: 1270. 56. Lesueur D. The colloidal structure of bitumen: consequences on the rheology and on the mechanisms of bitumen modification. Adv Colloid Interface Sci 2009; 145: 42–82. 57. Loeber L, Muller G, Morel J, et al. Bitumen in colloid science: a chemical, structural and rheological approach. Fuel 1998; 77: 1443–1450. 58. Gao G, Zhang Y and Zhang YX. Improved storage stability of LDPE/SBS blends modified asphalts. Polym Polym Compos 2002; 10: 229–236. 59. Fu H, Xie L, Dou D, et al. Storage stability and compatibility of asphalt binder modified by SBS graft copolymer. Constr Build Mater 2007; 21: 1528–1533. 60. Fawcett AH and McNally T. Blends of bitumen with polymers having styrene component. Polym Eng Sci 2001; 41: 1251–1264.

Downloaded from jep.sagepub.com by guest on July 30, 2014

Jelcˇic´ et al.

33

61. Brule´ B, Brion Y and Tanguy A. Paving asphalt polymer blends: relationships between composition, structure and properties. J Assoc Asphalt Pav 1988; 57: 41–64. 62. Sengoz B and Isikyakar G. Analysis of styrene-butadiene-styrene polymer modified using fluorescent microscopy and conventional methods. J Hazard Mater 2008; 150: 424–432. 63. Raghunathan P. Evidence for fractal dimension in asphaltene polymers from electron-spin-relaxation measurements. Chem Phys Lett 1991; 182: 331–335. 64. Janardhan AS and Ali Mansoori G. Fractal nature of asphaltene aggregation. J Petrol Sci Eng 1993; 9: 17–27. 65. Lu X, Soenen H and Redelius P. Rheological characterization of polymer modified bitumens. Ann Trans Nordic Rheol Soc 2011; 19. 66. Soenen H, Lu X and Redelius P. The morphology of bitumen-SBS blends by UV microscopy: an evaluation of preparation methods. Road Mater Pavem 2008; 9: 97–110. 67. Lu X, Soenen H and Redelius P. SBS modified bitumens, does their morphology and storage stability influence asphalt mix performance. In: The 11th ISAP international conference on asphalt pavements, Nagoya, Japan, 1–6 August, 2010. 68. Adolf D, Martin JE and Wilcoxon JP. Evolution of structure and viscoelasticity in an epoxy near the sol-gel transition. Macromolecules 1990; 23: 527–531. 69. Hodgson DF and Amis EJ. Dynamic viscoelasticity during sol-gel reactions. J Non-Cryst Solids 1991; 131-133: 913–920. 70. Laire D, Adam M, Emery JR, et al. Rheological behavior of an epoxy/amine system near the gel point. Macromolecules 1992; 25: 286–289. 71. Michon C, Cuvelier G and Launay B. Concentration dependence of the critical viscoelastic properties of gelatin at the gel point. Rheol Acta 1993; 32: 94–103. 72. Clark AH and Ross-Murphy SB. Structural and mechanical properties of biopolymer gels. Adv Polym Sci 1987; 83: 57–192. 73. Hossain KS, Miyanaga K, Maeda H, et al.. Sol-gel transition behavior of pure i-carrageenan in both salt-free and added salt states. Biomacromolecules 2001; 2: 442–449. 74. Nystro¨m B, Walderhaug H and Hansen FK. Rheological behavior during thermoreversible gelation of aqueous mixtures of ethyl (hydroxyethyl) cellulose and surfactants. Langmuir 1995; 11: 750–757. 75. Shih WH, Shih WY, Kim SI, et al. Scaling behavior of the elastic properties of colloidal. Phys Rev A 1990; 42: 4772–4779. 76. Marangoni AG and Rousseau D. Is plastic fat rheology governed by the fractal nature of the fat crystal network? J Am Oil Chem Soc 1996; 73: 991–994. 77. Hanyu A, Ueno S, Kasahara A, et al. Effect of the morphology of SBS modified asphalt on mechanical properties of binder and mixture. J East Asia Soc Transp Studies 2005; 6: 1153–1167. 78. Nakajima S, Deguchi T and Saito A. Investigations on morphology of thermoplastic elastomer (SB type TPE) in asphalt. Modified Asphalts 1999; 12: 13–21. 79. Lu X and Isacsson U. Chemical and rheological characteristics of styrene-butadiene-styrene polymer-modified bitumens. Transp Res Rec J Transp Res Board 1999; 1661: 83–92. 80. Asvestas P, Matsopoloulos G and Nikita K. A power differentiation method of fractal dimension estimation for 2-D signals. J Visual Commun Image R 1998; 9: 392–400.

Downloaded from jep.sagepub.com by guest on July 30, 2014