Influence of Polyphosphoric Acid on the Consistency and Composition of Formulated Bitumen: Standard Characterization and NMR Insights

Over the recent years, bitumen modification with polymers, acids, or mineral fillers has gained relevance to adjust its performance properties. This work reports the use of polyphosphoric acid (PPA) for the modification of formulated bitumen. With this objective, an in-depth literature review on PPA modification was firstly performed. Subsequently, five individual refinery components were selected for the preparation of bitumen blends, namely, asphaltic residue, vacuum residue, and three lube oils extracts. Seven binary/ternary bitumen blends were prepared and then treated with PPA. Afterwards, the five components and the unmodified and PPA-modified bitumen were characterized by standard methods (penetration, softening point, and penetration index), SARA analysis, elemental analysis, and 31P and 1H nuclear magnetic resonance (NMR) spectroscopy. The results evidenced higher asphaltenes and lower saturates/resins contents in PPA-modified bitumen. The NMR data suggest that the paraffinic chains became longer, the content of condensed aromatics increased, more substituted aromatic structures appeared, and α-hydrogen in aromatic structures diminished. These findings disclosed the improved consistency and oxidation stability of PPA-modified bitumen blends.


Introduction
The European Committee for Standardization defines bitumen as a virtually nonvolatile, adhesive, and waterproofing dark brown viscous organic material derived from crude oil or present in natural asphalt [1]. Due to its binding and hydrophobic properties, bitumen is commonly used for the construction of pavements and waterproof systems [2,3]. Bitumen properties are related to its composition, which in turn depends strongly on the type and origin of the crude oil. However, the precise identification of the molecular composition and structure of bitumen is almost impossible; therefore, they are often characterized by fractionation techniques [3,4]. For instance, the SARA method is based on 2 Journal of Analytical Methods in Chemistry The structure of bitumen is controversial but the colloidal model, proposed by Rosinger [8] though generally attributed to Nellensteyn [9], still prevails, especially for explaining the interactions between modifiers and bitumen fractions [6,10]. According to this model, bitumen is considered as a suspension of asphaltene micelles peptized by resins, dispersed in an oily medium constituted by saturates and aromatics [3,[10][11][12]. Depending on the relative contents of these fractions, bitumen exhibits different rheological properties used for its classification into three distinct types, specifically: "sol" (viscous) bitumen which has a colloidal structure constituted by noninteracting micelles and behaves as Newtonian fluid; "gel" (elastic) bitumen which is characterized by a three-dimensional colloidal structure and high resilience; and "sol-gel" (viscoelastic) bitumen which exhibits an intermediate behavior with elastic effects in the initial stages of deformation and a colloidal structure characterized by the presence of supermicelles [13,14].
Over the recent years, bitumen has become a highly technical material with modifiers (e.g., polymers, acids, and mineral fillers) gaining relevance in the control of the performance properties of modified bitumen [3,15]. Polyphosphoric acid (PPA) is an example of a reactive modifier that can be used by itself or in conjugation with a suitable polymer, in this case, with considerable economic benefits [16]. In general, the concentrations of PPA in bitumen are in the range from 0.2 to 1.2 wt.% [7] with an optimum at 1.0 wt.% [17]. Typically, commercial PPA is a mixture of phosphoric acid (H 3 PO 4 ), pyrophosphoric acid (H 4 P 2 O 7 ), triphosphoric acid (H 5 P 3 O 10 ), and higher oligomers [18].
Bitumen refined from a specific crude oil has characteristic properties related to its consistency, namely, softening point and penetration value at 25 ∘ C (expressed in dmm) which is the base for the European grading system [19]. These properties can be shifted up or down in the refining process or by posterior blending with different grade bitumen or an additive (e.g., a polymer, PPA, or a combination of both) [20]. Bitumen modification with a polymer or with PPA stiffens the product at higher temperatures, improving the resistance to permanent deformation, without negative effects at low temperatures (i.e., between −30 and 5 ∘ C) [18]. It is noteworthy that PPA has been reported to improve the behavior of modified bitumen also at low temperature (≤5 ∘ C) [21] and to impart antioxidant properties [18].
In this work, a detailed review of issued patents and scientific literature on the performance of straight-run bitumen (i.e., obtained directly from crude oil distillation) modified with PPA has been accomplished and is presented in the Appendix. On the other hand, since the influence of PPA on formulated bitumen is not reported in the scientific literature, this is the main focus of this paper. It has been studied using phosphorous ( 31 P) and proton ( 1 H) nuclear magnetic resonance spectroscopy to evaluate structural modifications, SARA and elemental analyses to evaluate chemical transformations, and standard grading tests (softening point, penetration value, and penetration index) for performance evaluation.

Bitumen Modification with PPA and Its Characterization
The interaction of PPA with bitumen model compounds was studied by several authors, as summarized in the Appendix. From these studies, it became clear that PPA modification follows different mechanisms depending on the chemical composition of bitumen which is strongly related to the crude oil geographical source. Overall, the reactivity of PPA increases with the polarity of the asphaltene fraction because it enhances PPA dissociation (into PPA − and H + ) [3] disrupting the hydrogen bond network formed within the agglomerates of asphaltene micelles. Consequently, the molecular weight of the asphaltene fraction is lowered and the distribution of asphaltenes in the remaining fractions (i.e., saturates, aromatics, and resins, usually called maltene phase) is improved, shifting the bitumen towards a more elastic gel-type structure, as represented schematically in Figure 1 [13,17].
There is a direct relationship between the chemical composition and colloidal structure of bitumen and its performance [13]. Therefore, bitumen is often characterized by related parameters, such as the index of colloidal instability (defined as the ratio of the sum of asphaltenes and saturates to the sum of aromatics and resins), the colloidal indices Ip (i.e., resins-to-asphaltenes ratio) and Is (i.e., ratio of asphaltenes to maltenes), the penetration index (PI), and other rheological measurements.
Several rheological studies of PPA-modified bitumen focused on its performance and stability. For instance, Giavarini et al. [13] reported that PPA-modified bitumen exhibits higher penetration index (PI) and thus improved thermal susceptibility in comparison to unmodified bitumen. Giavarini et al. [16] and Bonemazzi and Giavarini [14] performed rheological studies with polymer-modified bitumen and proved that PPA changes the bitumen structure towards gel type, thus improving the stabilization interactions between the polymer and the bitumen components. Edwards et al. [22] performed rheological tests at low temperature (−25 ∘ C) and established that bitumen modification is influenced by the amount and type (grade) of PPA used and also by the composition of bitumen. Baldino et al. [21] validated the capacity of PPA to decrease the glass transition temperature (Tg) and to increase the stiffness of modified bitumen.
Another approach is to use spectroscopic techniques to detect specific functional groups in highly complex hydrocarbon mixtures such as bitumen [23][24][25][26]. In particular, solution-state 1 H and 13 C nuclear magnetic resonance (NMR) spectroscopy can be used to identify and quantify the aromatic and aliphatic moieties. This is of utmost importance to understand the molecular interactions between the SARA fractions of bitumen and to explain the influence of modifiers on bitumen performance [27]. For instance, Michon et al. [28,29] used quantitative 13 C NMR data coupled with molecular weight distribution data to estimate structural parameters of bitumen such as, for example, aromaticity and the average number of naphthenic and aromatic rings per molecule. Their final purpose was to define a bitumen "fingerprint" [28], particularly for the aromatic part, in order to propose Journal of Analytical Methods in Chemistry  a mechanism for bitumen oxidation [29]. The chemical transformations associated with bitumen ageing were investigated by Siddiqui and Ali [30] and Siddiqui [31]

Preparation of Bitumen Blends
Modified with PPA. The bitumen blends (penetration grade 70/100) used in this work contained asphaltic residue (AR), vacuum residue (VR), and aromatic extracts SN1, SN2, or SN3, in the proportions presented in Table 1. These bitumen blends were provided by a Portuguese refinery. Bitumen modification with PPA was performed in a glass reactor, equipped with a mechanical stirrer and a temperature controller. Bitumen blends (1000 ± 0.01 g) were heated to approximately 135 ∘ C and then PPA grade 105% (supplied by Innophos, Cranbury, USA) (8.00 ± 0.01 g) was added, which corresponds to 0.8 wt.%. The mixture was allowed to react for 30 minutes at 135 ∘ C with constant stirring (ca. 245 rpm).

Characterization of Bitumen Components and Blends.
The SARA fractions of the individual components and the seven bitumen blends (Bit j , j = 1-7) were analyzed according to the International Standard method IP-469 using Iatroscan MK-6 TLC-FID instrument (Iatron Laboratories Inc.) [33]. Data acquisition was performed with the DataApex Clarity Lite software, version 3. Each sample was analyzed in triplicate and the results (peak areas) were averaged and considered as the weight percentage of the corresponding SARA fractions.
Elemental analysis of carbon, hydrogen, and nitrogen in bitumen samples was performed in a Leco TruSpec5 instrument (628 Series) following the standard procedure ASTM D5291 [34]. Sulfur content was determined by energydispersive X-ray fluorescence (EDXRF) in Oxford Instruments Lab-X3500, following the International Standard method IP 336 [35]. The analyses were performed using three replicated samples.
For NMR analysis, the bitumen samples (0.20 ± 0.01 g) were dissolved in deuteriochloroform (CDCl 3 , Sigma-Aldrich, 99.8 atom% D) (1 mL) and analyzed in triplicate. Solution-state 1 H NMR measurements were obtained in a Bruker Avance 300 spectrometer (operating at 300.13 MHz) under the following conditions: spectral window of 6887 Hz, with a 30 ∘ pulse width (3.63 s), acquisition time 2.4 s, relaxation delay 1 s, 32768 data points, and 128 scans. Solution-state 31 P NMR spectra were acquired at room temperature in a Bruker Avance 300 spectrometer at a nominal frequency of 121.5 MHz, over a period of 7 minutes, using a pulse repetition rate of 1 s. Chemical shifts were referenced to orthophosphoric acid at 0 ppm.
Bitumen penetration (Pen, expressed in dmm) at 25 ∘ C and the ring and ball softening point (SP, in ∘ C) were measured by standard methods [19]. The penetration index (PI) was calculated by [36] where ( ∘ C) is the temperature used to measure Pen.

4
Journal of Analytical Methods in Chemistry

SARA and Elemental Analysis of Bitumen Components and Blends.
The SARA fractions of the individual bitumen components and of the seven bitumen blends before and after PPA modification are presented in Tables 2 and 3, respectively. Table 3 also presents the elemental analysis of bitumen before PPA modification. The binary blends (Bit1, Bit2, and Bit3) prepared with asphaltic residue (AR) and one of the extracts (SN1, SN2, or SN3) present similar contents of asphaltenes and resins (within experimental error). In contrast, the concentration of saturates and aromatics differs for the blends prepared with extracts SN1 and SN2 (i.e., Bit1 and Bit2, resp.) and with SN3 extract (i.e., Bit3). For the former, saturates were not detected and the aromatic content is naturally higher (ca. 71 wt.% versus 65.5 wt.% for Bit3) due to the nature of the extracts (see Table 2). However, the differences in elemental composition of the three bitumen blends are negligible (see Table 3).
It is noteworthy that the SARA fractions for binary blend Bit3 (78.39 wt.% AR + 21.61 wt.% SN3 extract) are similar to those for Bit7, a binary blend prepared with asphaltic and vacuum residues (30 and 70 wt.%, resp.; see Table 1). The most significant difference is the high content of saturates (3.0 wt.%, Table 3) in Bit7 which can be related to its higher content of vacuum residue (VR). In fact, VR contains 5.5 wt.% of saturates whereas AR, the main component of the other blends, contains no saturates (see Table 2). Again, there are no significant differences for the elemental chemical composition of Bit7 in comparison to the other bitumen samples.
The results for the ternary blends (Bit4-Bit6) prepared with AR, VR, and one of the extracts are quite similar (within experimental error) in what concerns the SARA fractions and elemental composition (Table 3). Overall, these results are consistent with the origin of the blends as they were all prepared with components derived from Arabian light crude oil.
It is worth noting that the elemental analysis of formulated bitumen after PPA addition is not reported in Table 3, as the results were statistically equivalent. In fact, from the literature review, it became clear that PPA modification follows different mechanisms, like redox reactions and disruption of hydrogen bond networks, and none of these mechanisms are expected to change the overall elemental composition of PPAmodified bitumen. A final comment concerning the absence of saturates in Bit1 and Bit2 is in order. In fact, these blends were prepared with extracts SN1 and SN2 which contain ca. 4 wt.% of saturates (Table 2) and thus the absence of saturates is puzzling. A possible explanation is that the content of saturates is below the TLC-FID apparatus detection limit (detection limit 0.8%).
The influence of PPA on bitumen composition (SARA fractions) can be visualized in Table 3. In general, PPA modification decreases the content of saturates and resins and increases the fraction of asphaltenes in agreement with the literature [13,37]. The higher content of asphaltenes in PPAmodified bitumen explains the experimental results (Table 4), namely, higher softening points (ca. 1-3 ∘ C increment), lower penetration values (ca. 10 dmm), and, consequently, an improvement of the penetration index (PI) in accordance with the results reported by Oyekunle [38]. The higher values of PI indicate a change in the colloidal structure of PPAmodified bitumen blends which is expected to improve their thermal susceptibility [13].
The effect of PPA on bitumen modification has been associated with redox reactions involving polycondensed aromatic structures that act as reducing agents for phosphoric acid oligomers (HPO 3 ) , which are therefore oxidized [13]. As a result, one would expect the conversion of part of the resins fraction into asphaltenes and the conversion of the saturates and aromatics fractions into resins [13]. However, the increase of the aromatics fraction observed in this work, after PPA addition, suggests that part of the resins were converted to asphaltenes and part were converted to aromatics. 31 P NMR Spectroscopy. The 31 P NMR spectrum of the PPA used in this work is presented in Figure 2. According to the literature [37,39], the resonance peak at 0 ppm is attributed to phosphorus in H 3 PO 4 , a smaller peak at −13 ppm is assigned to phosphorus in end groups of PPA chains, and a much   smaller peak around −26 ppm is attributed to phosphorus in the middle groups of PPA chains. The PPA-modified bitumen blends were also analyzed by 31 P NMR spectroscopy, at room temperature. The spectra for the seven blends (see Figure 3) are very similar revealing a broad signal approximately at 1 ppm. This resonance peak, located in the vicinity of the phosphorous resonance peak in H 3 PO 4 , was assigned to hydrophobic interactions between phosphorous and polycondensed aromatic structures typically present in asphaltenes [23]. The absence of resonance peaks assigned to phosphorus in phosphate chains (end and middle groups) suggests that phosphoric esters were not formed during bitumen modification with PPA, in agreement with the results of Miknis and Thomas [39] and in opposition to the earlier hypothesis of Orange et al. [40]. Furthermore, these results indicate that PPA is hydrolyzed completely back to H 3 PO 4 , probably due to the presence of residual water in bitumen [39].

Analysis of PPA and Modified Bitumen by
The broader spectral peaks for PPA-modified bitumen blends (Figure 3), in comparison with PPA (Figure 2), are probably due to higher viscosity of the PPA bitumen blends or to the formation of charge transfer complexes between phosphorous and the asphaltene polycondensed aromatic structures [39,41]. Another hypothesis is that small amounts of paramagnetic impurities typically present in bitumen (e.g., vanadium and nickel) may decrease the relaxation time and, consequently, increase peak broadening due to the Heisenberg uncertainty principle [42]. 1 H NMR Spectroscopy. The 1 H NMR results for unmodified and PPA-modified bitumen blends are presented in Table 5. The spectra were divided into 12 unequal segments corresponding to well-defined chemical shifts associated with specific groups (i.e., hydrogen types), though minor overlapping may occur [23]. Peak integration in the selected segments generated, after normalization, the relative content of the various proton types ( Table 5). The mean values were obtained by averaging the results from the spectra of three aliquots of the same bitumen formulation.

Analysis of Unmodified and PPA-Modified Bitumen by
Some conclusions may be drawn from the 1 H NMR results (Table 5) more explicitly presented in Figure 4. In general, the content of aliphatic hydrogen (segments 1 and 2, Table 5) in PPA-modified bitumen is different from that in the parent unmodified bitumen. The signals in the segments associated with the branching (0.5-1.0 ppm) and the length (1.0-1.7 ppm) of paraffinic chains are systematically higher, though rather similar. The signals in segment 3 (1.7-1.9 ppm), that is, in the region of chemical shifts associated with sulfur adjacent hydrogen atoms (1.8-3.0 ppm) [27], decrease after PPA modification, thus suggesting more substituted aromatic structures.
The slightly lower signals in segment 6 (2.4-3.5 ppm), Figure 4, which are associated with the chemical shifts of -hydrogen ( -CH, -CH 2 ) indicate the occurrence of oxidation reactions, thus explaining the improved stability of PPA-modified bitumen [27]. The signals associated with bridging hydrogen atoms (segment 7, 3.5-4.5 ppm) and hydrogen atoms in monoaromatic structures (segment 9, 6.0-7.2 ppm) have both a tendency to decrease   Table 5. after PPA modification (see Figure 4). This might be related to a slight increase in the content of hydrogen atoms associated with polyaromatic structures (segments 10, 11, and 12, in Table 5), suggesting an increase in content of condensed structures as reported in the literature [13].
A final note about the residual presence of olefins, detected in the region 4.5-6.0 ppm (segment 8, Table 5). In fact, since bitumen is allegedly free of olefins, due to the process conditions in the refinery, its presence can be attributed to cracking reactions eventually occurring during NMR sample preparation (recall that bitumen reheating was necessary to facilitate sample handling) [54].

Conclusions
Binary and ternary bitumen blends formulated with asphaltic residue, vacuum residue, and aromatic-extracts derived from refined Arabian crude oil were analyzed before and after modification with polyphosphoric acid (PPA). SARA analysis revealed that, in general, PPA modification increased the fraction of asphaltenes in bitumen and decreased the saturates and resins fractions abundance. These changes improved the performance properties, namely, the penetration index (PI), which indicates higher thermal stability of PPA-modified bitumen blends. With respect to the mechanism of PPA actuation, the 31 P NMR proved that PPA is hydrolyzed back to phosphoric acid and reacts with bitumen, probably in the form of charge transfer complexes with polycondensed aromatic structures. This fact was disclosed from the presence of the single resonance peak assigned to phosphoric acid and the absence of peaks assigned to phosphorus in the end and middle groups of phosphate chains. The PPA disrupts the agglomerates of asphaltene micelles, which promotes the distribution of asphaltenes in the maltene phase, thus increasing the content of the aromatic fraction. Furthermore, part of the resins is converted into the aromatic fraction.
Concerning the 1 H NMR studies, results suggest that PPA improves the oxidation stability of the bitumen blends and increases the concentration of condensed structures. The former was evidenced by the content drop of -hydrogen (range 2.4-3.5 ppm) after PPA modification, as these hydrogen atoms are associated with oxidation susceptibility.

Appendix
This appendix presents a literature review concerning bitumen modification with polyphosphoric acid (PPA). Table 6 lists the patents issued since 1973 when Alexander [55] patented a process using PPA to improve bitumen viscosity. Table 7 summarizes the scientific studies performed to evaluate the effect of PPA on bitumen, or bitumen model compounds, and the main characteristics of the ensuing PPAmodified bitumen.     (ii) Addition of PPA alone provides better improvements than modification with LDPE (with and without PPA) [53] a EVA: ethylene vinyl acetate, HDPE: high density polyethylene, LDPE: low density polyethylene, PE: polyethylene, PPMA: polypropylene-maleic anhydride, SBR: styrene-butadiene rubber, and SBS: styrenebutadiene-styrene block copolymers.
c SHRP: Strategic Highway Research Program.