‘Advanced Testing and Characterization of Bituminous Materials ~ Loizos, Partl, Scarpas & Al-adi (eds) © 2009 Taylor & Francis Group, London, ISBN 978-0-415-55854-9. First-principles investigation of the multiple phases in bituminous materials: The case of asphaltene stacking A.JM. Schmets, N. Kringos & A. Scarpas Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, The Netherlands CP. Duit Faculty of Applied Sciences, Delft University of Technology, Delft, The Netherlands G. Schitter Faculty of Mechanical, Maritime and Materials Engineering, Delft University of Technology, Delft, The Netherlands T, Pauli Western Research Institute, Wyoming, USA ABSTRACT: The existence of a microstructure in bituminous materials can explain this materials tendency to self heal, By means of atomic force microscopy and small angle neu- tron scattering the existence of the microstructure is demonstrated once again. Then itis sug- gested that the observed microstructure may very well be a consequence of the aggregation of asphaltenic molecules (providing internal volumes with a low H/C-ratio). This is a rather unadventurous suggestion, because flocculation of asphaltenes costs the global oil industry millions of dollars per day. A model asphaltene (a coronene-like molecule) has been selected for calculating semi-empirically the tendency to form stacks. It was found that there is some tendency to cluster, but the effect is not strong enough to be responsible for the observed microstructure. 1 INTRODUCTION 1.1 Multiple phases and a healing mechanism for bitumen Ithas been known for long that materials which can autonomously mitigate the initial stages of damage due to fatigue or impact loading exhibit a dramatic improvement of service life. Better service life characteristics automatically imply a more sustainable and more efficient utilization of the raw materials. In the composite material that is commonly used in road pave- ments it is the bituminous binder that will become a rare material eventually. Moreover, it is this component, which amounts for 5 to 8 weight percent of the pavement structure, where damage in the form of cracks will occur first, deteriorating the performance of the pavement as a whole, And finally, it has been known for long that it is this component of the asphalt system that has the ability to mitigate damage partially, ic. it exhibits the property of self healing (Little and Bhasin 2007). Thus we motivate that a better understanding of the healing behavior of pavement materials should start from the level of the binder. If the bituminous binder were a perfect homogeneous material, the existence of cracks and damage would occur because of yielding and traditional fracture mechanics applies. However, ithas been demonstrated (Pauli 2001), and we'll provide here some evidence from additional experiments, that bitumen at ambient temperatures are far away from being an homogeneous material. In contrary, it appears that bitumen exhibits a very rich microstructure, whose exact appearance depends on the type and composition of the bitumen and the thermal history of 143the material, Speculating that the various phases in the bitumen possess a different stiffness, means that under loading conditions cracks may be expected to form preferentially at the interfaces between the phases, i, at those locations in the material where there is a significant gradient in stiffness. In another contribution to this volume by Kringos et al. a model is presented that utilizes the existence of multiple phases in the bitumen to explain the observed self healing behavior ‘The authors formulate a theory that explains how the phases can reversibly appear and dis- appear by thermodynamically driven spinodal decomposition. The reversible formation and disappearance of the microstructure in the bitumen is then believed to be the major reason for the longer than expected service life time observed in these materials. In this paper we show additional evidence of the rich microstructure in bitumen, and w test the hypothesis that the phases are formed by stacked asphaltene molecules. IL 1.2. Experimental signature of multi-phase behaviour Extensive Atomic Force Microscopy (AFM) studies performed at the Western Research Insti- tute (Pauli 2001) have shown that bituminous materials appear to possess a rich microstrue- ture, Similar experiments have been performed at TU Delft on 70-100 pen-grade bitumen, a grade that is typically used in Dutch road constructions (Schitter et al 2008, Schmets & Schitter 2009). Also these experiments showed conclusively, as can be seen in Figure 1, that the material possesses a very tich and complex microstructure. The typical bee-structures, also observed by Pauli, are clearly visible. The bee-structures (coined ‘Marangoni bees’) have the shape of jagged ellipsoids (the envelope is an ellipsoid), with long axes that measure typi- cally about 10 um. Although the evidence provided by the AFM experiment does very decisively proof that bituminous materials possess multiple phases, it should be noted that the AFM technique does essentially probe the surface of the material under investigation. The observed phase behavior might then be only a surface effect, rather than a pattern that is present throughout the bulk of the material, For this reason other experimental techniques had to be called in for exploring the existence of the multiple phases throughout the bulk of the material. The tech- nique of choice would be a small angle neutron scattering technique, because of the ability of 0 50pm 0 15 um Figure 1. Microstructure of 70-100 pen-grade bitumen observed with a Vyoo AFM. Clearly visible are the ‘Marangoni bees’. The right image is a detail of the left image, and reveals more details of the complex beauty of this microstructure, 1441.104 1.05 4 1.004 0.95 4 oso 4 7 PIP, oss 4 - oso 754 4 070-4 06s 4 “a 0 2 4 6 8 10 12~=—O18 18 Spin echo length [um] Figure 2. SESANS response to the 70-100 pen-grade bitumen. The dotted red line would be the response if there were no microstructure present in the sample (with features between 20 nm and 20 micrometer), neutrons to ‘look inside materials’ and the accessibility of the observed length scales of the microstructure by sinall angle techniques. The 70-100 pen grade bitumen that was studied with atomic force microscopy has been subjected to small angle neutron scattering experiments. The bitumen has been dispersed on a1 mm thick aluminum plate of 3 cm by 5 cm, and was mounted in the Spin Echo Small Angle Neutrons Scattering instrument (SESANS) in the TU Delft research reactor. SESANS. (Rekveldt 1996) is a small angle instrument that produces real-space data, in contrast to tra- ditional small-angle scattering instruments. First results as depicted in Figure 2, suggest that there are patterns present in the material with a typical spatial dimension of about 5to 10 jum, very similar as was observed with the AFM (Schmets & Bouwman 2009). The observation of a microstructure in the material with SESANS is only possible if there is some sort of a ‘contrast’ present in the sample. In the case of SESANS this would be a contrast in scattering length density. Given the composition of bitumen that consists mainly out of carbon and hydrogen atoms the contrast in scattering length density is most likely caused by a spatial variation of the hydrogen to car- bon ratio (H/C-ratio) in the bitumen. Thus we may conclude from Figure 2 that the bitumen under scrutiny is a non-homogeneous material with respect of H/C-ratio. 2 COLUMNAR PHASES OF STACKED ASPHALTENES Althongh the existence of the microstructure in bitumen has been confirmed, the exact com- position of the observed phases is still heavily debated, One scenario would be that one of the phases consists of stacked asphaltene molecules (Bergman et al. 2003, Groenzin et al. 2000), because these molecules (essentially sheets of graphite dressed with aliphatic ligands and hetero atoms) have a very different number of hydrogen atoms per carbon compared 145nematic phase columnar phase Figure 3. Nematic and columnar ordering of the disk-like (‘asphaltenic’) molecules, to more aliphatic components in the bitumen. This is due to the pericondensed aromatic rings that constitute the backbone of the group of molecular species that have been called asphaltenes (formally asphaltenes are defined as the fraction that is separated out of bitumen by n-heptane). Typically asphaltene molecules have an H/C-ratio of 1.1, where the bitumen as a whole will have an H/C-ratio close to 1.5. Coagulation or stacking of the asphaltenes would produce volumes of realy distinct H/C- ratios, and could hence be the observed feature in the SESANS experiment. It is also well known that flat and rigid structures of aromatic macrocycles possess extensive molecular -orbitals in which the electrons are delocalized on both sides of the core plane, The n-orbitals of different (asphaltene) molecules tend to interact with each other, giving rise to various ordered, liquid crystalline, structures as shown in Figure 3. This type of structures (discotics) are widely studied (eg. (Piris 2004)), and are known to be very common for porphyrin-like molecules (another component of every bitumen, though occurring in very small concentra- tions therein), The stacks of asphaltene molecules might very well represent one of the phases that are observed. And even if the stacks of asphaltene molecules do not constitute the main ingredi- ent of one of the phases observed, their stacking behavior can be at the onset of the forma- tion of the microstructure, ie, associations of asphaltene molecules catalyze the formation of microstructure, It should also be noted that the mutual attraction of asphaltene molecules is a well known problem in the oil industry. Pipelines get blocked by hard and glassy material It is a known fact that this material are just precipitated asphaltenes; so, at least in the pres- ence of many light fractions, asphaltenes like to stick together and form a insoluble phase, causing blockage of oil pipelines. The remainder of this text will deal with the formation of stacks of model asphaltene molecules. 3. CALCULATION OF ASPHALTEN] -ASPHALTENE INTERACTION 3.1 A model asphaltene molecule for molecular simulation Over the years the understanding of the exact nature of asphaltene molecules has changed significantly, Initially it was believed that asphaltene are very large (M, > 10000 amu’s) mol- ecules, Over time it was found that this high molecular weight picture of asphaltenes could be an artifact of the m1 interaction of the rings, i. asphaltene molecules were always observed as stacked clusters instead as an isolated molecule. It was through the work of Mullins and co-workers that the exact nature of the asphaltene component of crudes and bitumen was finally elucidated (Bergman 2003, Groenzin & Mullins 2000) that (petroleum derived) asphaltenes consist on average out of 4-10 fused rings, and 146Mullins-asphaltene Ruiz-Morales asphaltene M,= 708 amu M,=300 amu fue = 1.22 ic =0.5 Figure 4, (a) Average structure of a petroleum derived asphaltene (Groenzin & Mullins 2000) and (b) the ‘coronene'-like model system used in our simulation of the stacking behaviour of asphaltenic molecules. that their molecular weight averages between $00 and 800 atomic mass units. They suggest the structure depicted in Figure da as a kind of average asphaltene molecule For the context of this research, the asphaltene model molecule as presented by Mullins is far too complex to start with. For reasons of computational efficiency, one would like to have a molecule that shares the polycyclic nature of a true asphaltene, though a high degree of symmetry would speedup calculations, and would also lead to non-tilted, ie. parallel, association of the asphaltenes. Therefore we remove the ligands of the asphaltene and find a suitable ring structure. Consequently we have come to a coronene-like molecule, a choice that was also suggested by the work of Ruiz-Morales (Ruiz-Morales 2007). The molecular structures discussed in the text are displayed in Figure 4 3.2. Semiemperical methods The calculations of the interaction between our asphaltene representing molecules were per- formed using the computer code VAMP (Clark 2002). This programme is a semi-empirical molecular orbital package that has been optimized to be numerically stable and fast. It includes routines implementing the natural atomic orbital/point charge model for molecu- lar electrostatic properties, which guarantees accurate calculation of dipoles and quadru- poles, and high quality molecular electrostatic potentials. Moreover, the VAMP code outputs 2 number of molecular properties, for example ionization potentials, multipole moments, molecular polarizabilities and potential-derived charges. The programme is especially useful for geometry optimization and as such it is a very useful tool to prepare large scale complex molecular dynamics simulations involving many different molecular types (a task that the authors are currently undertaking) Semi-empirical techniques use the same Linear Combination of Atomic Orbitals-Self Consistent Field theory as true ab initio (i.e. only the type of atoms and their initial posi- tions are required) codes. The main difference is that many of the more complex integrals are removed or replaced by simple approximations. Empirical parameters and functions are used to repair for the errors that occur in the result by removing those integrals. These empirical parameters are fitted to reproduce experimental data 147The molecular wave function, ¥, is built up by a Slater-determinant consisting of molecular spin orbitals, Y, These ate described by the usual LCAO method, in which the molecular orbit- als are obtained as a linear combination of the atomic orbitals, = Dcai®e w @ Molecular orbitals can then be built up by an iterative procedure that optimizes the coef- ficients Cy, . Once the molecular wave function has been found, VAMP uses it to evaluate the energy of the molecule, the atomic forces, and many other electronic properties. The energy and the atomic forces are used to optimize the geometry of the molecule to a stationary point at which the atomic forces are ideally all zero. Semi-empirical programs are used for molecules that are too large for geometry optimi- zation by (rue ab initio methods like the Hartree-Fock method. At present (March 2008) semi-empirical codes like the one we utilized in the current context are used for molecules comprising 50 to 400 atoms. The asphaltene representing molecule, C,,H,,, which we con- sider in this contribution, consists of 36 atoms. So for calculating their mutual interactions semi-empirical methods are currently indispensible. 3.3 Structural optimization of a single molecule Before the association behavior of our model-asphaltene molecules can be studied, the geometry of an individual molecule should first be optimized, using the same semi-empirical method as will be used for the remainder of the calculations. For the geometry optimization we utilized the ‘neglect of diatomic differential overlap’ (NDDO) approximation with the PM3 hamiltonian and the electron spin was treated in a restricted Hartree-Fock manner, The structure was considered ‘optimized’ when the energy was converged to between two subsequent self consistent steps to Sx 10" eV atonr". The resulting optimized structure, together with the calculated bond lengths, is given in Figure 5 below. The calculated values for the various bond lengths in the molecule are in good agreement with spectroscopic data of similar molecular systems. This gives confidence for the accurate- ness of the computational method that was selected in this research effort. 3.4 Do the asphaltene molecules stack? Because of symmetry considerations it is clear that the parallel arrangement of two of the ‘asphaltenes’ must be the energetically most favorable. Therefore only single point energy Figure 5. Optimized structure of our ‘model’ asphaltene. 148Distance a a) (by i t “el ney Dance 10%} Figure 6. (a) Calculated energy of the two molecule system. (b) Trajectory taken for single point energy calculations. ae Figure 7. (a) Energy per molecule as function of the number of molecules that participate in a stack and (b) a typical result for a stack of 4 model asphaltenes. calculations have been performed as a function of the asphaltene-asphaltene distance, see Figure 6b, Remarkably, we do not find a clear energy minimum (Figure 6a), only a clear repul- sive part up to 2A and a very shallow minimum at about 6 A . This suggests that there is no bonding between the two m-systems, at least not due to polarization effects of the fused ring systems. A subsequent geometry optimization of the system, where the forces on both molecules were minimized, gave a rather large equilibrium distance between both molecules of 641A. Then an equivalent procedure has been repeated for stacks of 3-8 molecules. Surprisingly enough, there is a much more distinguished minimum in the total energy of the 3 molecule system, with an equilibrium inter-molecular distance of 5.55A.. Adding more molecules to the stack does not change anything, what leads to the conclusion that these type of systems will most likely ocour in nature as stacks of 3 molecules. All this is clearly demonstrated in Figure 7a, where the energy per ‘asphaltene’ molecule is given as function of number of atoms that participate in the stack, ie. the calculated total energy of the system is divided by N (umber of molecules in the stack) times the energy of a free single molecule. When this number deviates from unity there is an energy advantage to form a stack. It is clear that the stack of 3 atoms is energetically favorable over a single molecule or a stack of 2 molecules Figure 7b shows how the minimum energy structure of a stack of molecules would look like. Obviously, these are not the structures we are after, because they are significantly smaller than the detection limit of the experimental techniques that we have used for studying the microstructure in the bituminous material. 1494 CONCLUSIONS It has been demonstrated by various means that bituminous materials possess a very rich microstructure, depending on the molecular make-up of the bitumen as well as its thermo- dynamic history. The micro-structural features can be distinguished by their C/H-ratio as SESANS results suggest. ‘A candidate with a distinct enough H/C-ratio, namely stacks of asphaltenes have been con- sidered. By means of semi-empirical methods it was found that a selected model-asphaltene prefers to stack in clusters of 3 molecules. This can not give rise to the observed micro- structural pattern. We dare to conclude that it is quite unlikely that the observed microstruc- tural features consist of aggregates of asphaltenes; even not if more realistic models are to be selected to represented the fraction of asphaltenes. ACKNOWLEDGEME) s The research described in this paper has been performed in the context of the ‘Delft Healing Consortium’. In this context, the authors gratefully acknowledge the support of Shell Global Solutions, Nynas AB, the Dutch Ministry of Public Works, Ooms Nederland Holding by, Delft Centre for Materials, the Western Research Institute and the US Federal Highway Administration. The authors thank the Reactor Institute Delft and Dr Wim Bouwman for allowing the use of the unique SESANS setup and the Delft Centre for Systems and Controls for allowing the use of their state-of-the-art AFM equipment. Finally, Robijn Niemeijer is thanked for other kinds of support REFERENCES Bergmann, U., Groenzin, H., Mullins, 0.C., Glatzel, P, Fetzer, J, & Cramer, SP, 2003. Chem Phys. Lett, 369, 184 Clark, T,, Alex, A., Beck, B, Burkhardt, F, Chandrasekhar, J, Gedeck, P, Horn, AJHLC, Hutter, ‘M.2002. Martin, B., Raut, G., Sauer, W, Schindler, T., & Steinke, T., Vamp 8.1, Erlangen, Groenzin H., Mullins 0.C, 2000, “Asphaltene molecular size and structure”, Journal of Physical Chem- istry, A 103, pp. 11237-11245 Little, D.N,, & Bashin, A. 2007. ‘Exploring Mechanisms of Healing in Asphalt Mixtures and Quantify- ing Its Impact’, in Self Healing Materials: an alternative approach to 20 centuries of materials sci- cence, Springer Series in Materials Science 100. Pauli, AT, Branthaver, LF, Robertson, RE., Grimes, W., & Eggleston, CM. 2001, “Atomic Force Microscopy Investigation of SHRP Asphalts”, American Chemical Society, Division of Petroleum Chemistry, 46(2), 104-110, Piris, J, 2004. Optoelectronic Properties of Discotic Materials for Device Applications, Delft University Press Rekveldt M-Th. 1996. Nucl. Instr. & Methods in Phys. Res. B 114 366 Ruiz-Morales, ¥., Molecular orbital calculations and optical transitions of PAHs and asphaltenes, in “Asphaltenes, Heavy Oils, and Petroleomics’, Mullins, 0.C., et al. (editor), Springer, 2007, Schitter, G., & Rost M.J. 2008. Scanning probe microscopy at video-rate, Materials Today, Volume 11, Supplement 1, Pages 40-48, Schmets, A.JM., & Bouwman, W.G. et al, 2009, ‘Small angle neutron scattering on bituminous materi- als with various wax contents’, to be published. Schmets, AJM, & Schitter, G. ot al. 2009. “Temperature and cooling rate dependence of the micro- structural features present in waxy crndes’, to be published, 150