Tutorial

Tutorial on the Phase Behavior of Asphaltenes and Heavy Oils

Introduction
From oil wells through refinery processing of petroleum, solid asphaltenes tend to precipitate to cause plugging, fouling, and coking . Since the economic viability of petroleum companies requires processing large volumes of petroleum, any interruption of the flow or of the heat transfer through metal surfaces by the precipitation of asphaltenes is extremely costly. Therefore, it is imperative to be able to predict when and why asphaltenes precipitate in order to mitigate plugging, fouling, and coking. However, petroleum is difficult to characterize because it is a mixture of 100,000 to 1,000,000 different molecules without a repeating molecular unit. In particular, the asphaltene fraction is difficult to characterize because of its lack of volatility and because of its tendency to self-associate. In addition, scattering data show that the asphaltenes are not in true solution in petroleum but are a colloidal dispersion with particles of the order of 40 A in diameter(1). Once more, there is far from agreement of what the interactions are in petroleum that drives the asphaltenes to precipitate. Nevertheless, through all this smoke screen of complexity and confusion, much progress in the phase behavior of petroleum is being made and rather simple models are producing surprisingly positive and practical results.

Chemical Characterization
Asphaltenes are defined as the toluene soluble fraction that precipitates from petroleum when an excess (25 to 40 times) of n-heptane (or n-pentane) is mixed with petroleum and waiting at least four hours before filtering. This definition reveals little about asphaltenes except that it is the least soluble fraction in petroleum. In fact, no one has positively identified the chemical structure of even one asphaltene molecule and if they did, they would have over 100,000 structures to identify to reveal the entire set of molecules in the asphaltene fraction of one crude oil. Therefore, it is constructive to investigate the properties of the entire asphaltene fraction and compare them with the properties of other fractions of petroleum macromolecules.

Figure 1. Petroleum Separation Protocol

Fractionation
Figure 1 shows one method for separating petroleum. The first step is a vacuum distillation. This is done to correspond to refinery separation, to remove the fraction that would evaporate with solvents, and to concentrate on the more difficult to characterize fraction, the vacuum resid. Except for a slight amount of inorganic contaminates (salt, rust, clay, etc.) petroleum crudes are toluene soluble. However, during conversion a black solid, toluene insoluble fraction can be formed that we call coke. After evaporating off the toluene, asphaltenes are precipitated with n-heptane and removed by filtration. The resin fraction is removed from the heptane solution by adsorption on Attapulgus clay and desorption with a mixture of toluene and acetone. Meanwhile, the heptane is evaporated off the remaining oil and the saturates are precipitated by methyl ethyl ketone at dry ice temperatures. The remaining oil after evaporation of the methyl ethyl ketone is the small ring aromatics (often called "aromatics"). An example set of elemental analyses, molecular weight, and 13C NMR aromaticity values for the fractions of Cold Lake vacuum resid are shown in Table I. The asphaltene fraction is typically the highest molecular weight, most aromatic, highest hetroatom (atoms other than carbon and hydrogen) fraction of crude oil. By comparing the composite total of the fractions with the starting vacuum resid, one determines good balance except for oxygen and hydrogen. This indicates that petroleum tends to oxidize during laboratory separation. Thus, more detailed studies of oxygen functionality should be done on fractions separated and stored under an inert atmosphere. The fact that the saturate fraction contains 15% aromatic carbon shows that the separation is far from exact but that this represents the most saturated of the fractions.

Molecular Weight Measurement
The measurement of even the average molecular weight of asphaltenes is a challenge because they are nonvolatile, self-associate to form colloids in solvents, and strongly adsorb to most surfaces. The lack of volatility causes mass spectrometry to measure values that are low; adsorption causes gel permeation chromatography to yield low values; and self-association produces high values with measurements in solvents, such as vapor pressure osmometry. The lesser of these evils appears to be vapor pressure osmometry in the best solvent, o-dichlorobenzene based upon two-dimensional solubility parameter studies(2), and at the highest temperature, 130º C, for the technique. Evidence indicates that these conditions break-up the asphaltene self-association(3).

Solvent-Resid Phase Diagram(3)
By investigating the properties of the fractions of the resids of a number of crudes before and after conversion, no single property was found to distinguish fractions from each other. However, as shown in Figure 2, a plot of molecular weight versus hydrogen does achieve this objective as each fraction occupies a unique area. The hydrogen content is an inverse measure of aromaticity similar to H/C atomic ratio because the carbon content of all petroleum fractions, converted or not, is nearly constant (80-85%). This indicates that asphaltenes are insoluble in n-heptane because of a combination of high molecular weight and high aromaticity. In addition, the solvent-resid phase diagram provides a better definition of asphaltenes and the other fractions than the usual separation definition as well as being a more quantitative version of the heavy oil map than the conceptual molecular weight versus polarity plot of Long(4).

Physical Interactions
Primary Interaction
Based upon the solvent-resid phase diagram and investigating the solubility behavior of asphaltenes in a number liquids(2), the author has come to the conclusion that the primary interaction between asphaltene molecules is the van der Waals attraction between large flat areas of polynuclear aromatics. This conclusion is not generally accepted and others at this conference will offer polarity, hydrogen bonding, acid-base, charge transfer between aromatics (p-p bonding), and porphyrin interactions(5) as alternatives. While these other interactions might contribute, they are not required to model most phase behavior applications. One exception might be for petroleum-water emulsions where hydrogen bonding between asphaltenes and water at the oil-water interface is quite important(6). Nevertheless, the high solubility of asphaltenes in carbon disulfide(2), a high solubility parameter liquid that is incapable of specific interactions, indicates that asphaltene interactions are not unusual by type but by magnitude. As a result, we use the one-dimensional solubility parameter to measure the attractive interaction between petroleum macromolecules and relate it inversely with hydrogen content as shown in Figure 2.

Polynuclear Aromatic Size Distribution
To date, the aromatic ring size distribution of asphaltenes have not been directly measured. However, recent advances in multidimensional, solid state NMR indicate that this will be accomplished in the near future. Nevertheless, indirect measurements(7) indicate that the maximum aromatic ring size in asphaltenes is 5 or 6 but there is likely to be two or more polynuclear aromatics (PNA) per molecule. Still, this is much smaller than the polychickenwire aromatic structures that many investigators have proposed. By studying discotic liquid crystals that contain a one to four ring aromatic core with aliphatic side chains (Figure 3), we have learned that these tend to stack in columns, without the need for huge aromatics(1). This is what we propose as the interaction in asphaltenes, except the asphaltene stack is not high but there are two or more stack interactions per molecule. The multiple aromatic cores greatly promote association and insolubility. Since the resin fraction strongly interacts with asphaltenes but only weakly associates

Figure 2. The Solvent - Resid Phase Diagram

with other resin molecules, we picture resins to have one PNA per molecule like discotic liquid crystals, and also like discotic liquid crystals they are soluble in n-heptane.

Colloidal Dispersion-Solution Hybrid Model of Petroleum
Pfeiffer and Saal(8) devised the original colloidal model of petroleum to explain rheological data. A form of this model, shown in Figure 4, remains today. According to the present version of this model, asphaltenes associate to form microscopic solids that are dispersed by the amphiphilic resins. The resins are attracted to asphaltenes on one end of the resin molecule and to small ring aromatics on the other end. While small ring aromatics act as a solvent for the asphaltene-resin dispersion, saturates act as a nonsolvent. Thus, this model is neither a solution nor a colloidal model but a hybrid of both. Modern support for this model is that the addition of resins to asphaltenes in toluene reduces its radius of gyration, measured by small angle x-ray scattering(9). In addition, the addition of low concentrations of amphiphilic model compounds greatly increases asphaltene solubility(10). As most models, this is an over simplification that enables an approximation of reality within a certain degree of accuracy. Once the limitations are determined by comparing with data, directions for improving the model are usually suggested.

Figure 3. Example Discotic Liquid Crystals

Figure 4. Colloidal Dispersion - Solution Hybrid Model of Petroleum

Phase Behavior of Petroleum
With the hybrid model of petroleum one can see that asphaltenes are held in a delicate balance that can be easily upset by the addition of saturates or the removal of resins or small ring aromatics. Petroleum in reservoirs often contains large amounts of dissolved gases, especially methane (called live oils). During production as the pressure drops, the solubility parameter of the dissolved methane decreases and above the bubble point some crudes, especially those containing high fractions of saturates, precipitate asphaltenes(11). In addition, during enhanced oil recovery when compressed gases, such as ethane or carbon dioxide, are injected into the reservoir, it can lower the solubility parameter of the mixture and precipitate asphaltenes(12). Once crudes are brought to the surface, they are commonly mixed in the field, in tankers and/or pipelines, and in tanks at the refinery. It has not generally been known but the blending of crudes can upset the delicate balance and precipitate asphaltenes(13). When crudes or resids are thermally processed, the side chains are cracked off asphaltenes and resins that causes resins to combine and form more asphaltenes and the asphaltenes to become less soluble(14). On cooling, the less soluble asphaltenes with fewer resins as natural dispersants can precipitate and form hot filtration sediment(15). In addition, if the thermal conversion is carried far enough, the asphaltenes can precipitate as a liquid crystalline phase (or carbonaceous mesophases) and quickly combine to form coke(14). In addition to these undesirable asphaltene precipitation events, in the deasphalting process gaseous alkanes (propane, butane, or pentane) are designed to precipitate asphalt and extract the more desirable deasphalted oil for forming heavy lube base stocks or for forming a better feed for the catalytic processing of resids.

When an alkane is mixed with petroleum, a certain amount can be blended before asphaltenes begin to precipitate (the flocculation point). As more alkane is added the amount of phase that is precipitated increases but approaches an asymptotic value at high ratios of alkane to oil(16). As the number of carbons in the normal alkane is increased, the asymptotic quantity of heavy precipitated phase decreases(16). However, as the number of carbons in the normal alkane is increased, the volume of alkane at the flocculation point first increases but goes through a maximum (about C9) and decreases. Recently, Cimino et al(17) have shown that this latter effect is due to the entropy of mixing different size molecules that can be described by the Flory-Huggins equation. Of course, if the mixture includes components that are near or above their critical temperature, such as in a live oil, then volume expansion effects are important. In these cases increasing the temperature can lower asphaltene solubility. However, in dead oils (without components that are near or above their critical temperature) increasing temperature always increases asphaltene solubility

Thermodynamic Models
How does one account for colloidal dispersions, asphaltene association, and the large number of molecules in petroleum to develop a thermodynamic model of petroleum? Does one include polarity, hydrogen bonding, acid-base, charge transfer between aromatics (p-p bonding), and porphyrin interactions? How does one best describe volume expansion effects and the entropy of mixing of molecules of different sizes? Many of the papers in the three sessions on the Thermodynamics of Asphaltenes and Heavy Oils will try to answer some of these questions. However, it is doubtful that all the papers together will answer all of the questions.

Some completely ignore the complexity and use a simple model. One approach is based upon the hypotheses that the flocculation point occurs at a unique solubility parameter of the mixture for an oil and that this mixture solubility parameter is the volume average solubility parameter of the components(13, 18,19,20). This is all that is needed for a conservative prediction if the blending of oils will precipitate asphaltenes(14, 21). However, improved predictions need a colloidal component to the model. By adding an equation of state, supercritical fluids can be included(19, 20). Others combine the regular solution model with the Flory-Huggins entropy of mixing and an equation of state(17, 21,22). Another approach is to use more modern, and more complex, solution models. The SAFT equation of state was originally developed to model petroleum(23,24), including the association of asphaltenes, but has had its greatest success when applied to polymers. Several groups have attempted to account for colloidal effects and the dispersant activity of resins(25, 26). Pan and Firoozabadi(27) have even modeled the ability of amphiphilic compounds to increase asphaltene solubility. All of these are valid approaches. However, what is needed is the simplest combination of solution and colloidal models that enables solving practical petroleum phase behavior problems to the required accuracy with the input of easily obtained experimental data.

Literature Cited 1. Wiehe, I.A. and Liang, K.S., Fluid Phase Equilibria, 117, 201 (1996).
2. Wiehe, I.A., Fuel Sci. & Tech. Int'l., 14, 289 (1996)
3. Wiehe, I.A., Ind. Eng. Chem. Res., 31, 530 (1992).
4. Long, R. B., Advances in Chemistry Series 195, J.W. Bunger and N.C. Li, ed., ACS, Washington, DC, 17 (1981).
5. Kaminski, T.J., Fogler, H.S. and Mairal, A., Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling , Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils I (1999).
6. Spiecker, P.M. and Kilpatrick, P, Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third International Symposium on the Thermodynamics of Asphaltenes and Heavy Oils I (1999).
7. Speight, J.G. and Pancirov, R.T., Liq. Fuels Technol., 2, 287 (1984).
8. Pfeiffer, I. And Saal, R.N.J., J. Phys. Chem., 44, 139 (1940).
9. Espinat, D., du Petrole, F., and Ravey, J.C., SPE 25187 (1993).
10. Chang, C.-L. and Fogler, H.S., Langmuir, 10, 1749 (1994).
11. Cimino, R., Correra, S., Del Bianco, A., and Lockhart, T.P., Asphaltenes: Fundamentals and Applications, E.Y. Sheu and O.C. Mullins, ed., Plenum Press, NY, 126 (1995).
12. Dandekar, A.Y., Andersen, S.I., and Stenby, E.H., Prepints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils II (1999).
13. Wiehe, I.A., Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third International Symposium on the Thermodynamics of Asphaltenes and Heavy Oils III (1999).
14. Wiehe, I.A., Ind. Eng. Chem. Res., 32, 2447 (1993).
15. Wiehe, I. A., ACS Div. Pet. Chem. Meeting, San Francisco (1997).
16. Speight, J.G., The Chemistry and Technology of Petroleum, 2nd ed., Marcel Dekker, NY (1991).
17. Cimino, R., Correra, S., Del Bianco, A., and Lockhart, T.P., Asphaltenes: Fundamentals and Applications, E.Y. Sheu and O.C. Mullins, ed., Plenum Press, NY, 121 (1995).
18. Wiehe, I. A. and Kennedy, R. J., US patent 5,871,634 (1999) World patent WO 98/26026.
19. Andersen, S. A., 8th Int. Conf. On Fluid Properties & Phase Equilibria for Chemical Process Design (1998).
20. Andersen, S. I. and Pedersen, Carsten, Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils III (1999).
21. Hirschberg, A., deJong, L.N.J., Schipper, B.A., and Schipper, J.G., SPEJ, June, 283 (1984).
22. Fahim, M.A., Al-Sahhaf, T.A., and Al-Sayrafi, A., Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils III (1999).
23. Huang, S. H. and Radosz, M., Fluid Phase Equilibria, 70, 33 (1991).
24. Zhuang, M. S. and Thies, M.C., Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils III (1999).
25. Leontaritis, K.J. and Mansoori, G.A., AIChE Spring National Meeting, Houston (1993).
26. Pan, H. and Firoozabadi, A., SPE 38857, SPE Annual Technical Conference and Exhibition, San Antonio (1997).
27. Pan, H. and Firoozabadi, A., Preprints of AIChE International Conference on Petroleum Phase Behavior and Fouling, Third Int'l Symposium on the Thermodynamics of Asphaltenes and Heavy Oils III (1999).