Vous êtes sur la page 1sur 18

This article was downloaded by: [Monash University Library]

On: 22 June 2013, At: 19:47


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

International Journal of Pavement Engineering


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gpav20

Molecular modelling and simulation of asphaltenes and


bituminous materials
Michael L. Greenfield

Department of Chemical Engineering, University of Rhode Island, Kingston, RI, 02881, USA
Published online: 13 Jun 2011.

To cite this article: Michael L. Greenfield (2011): Molecular modelling and simulation of asphaltenes and bituminous
materials, International Journal of Pavement Engineering, 12:4, 325-341
To link to this article: http://dx.doi.org/10.1080/10298436.2011.575141

PLEASE SCROLL DOWN FOR ARTICLE


Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.

International Journal of Pavement Engineering


Vol. 12, No. 4, August 2011, 325341

Molecular modelling and simulation of asphaltenes and bituminous materials


Michael L. Greenfield*
Department of Chemical Engineering, University of Rhode Island, Kingston, RI 02881, USA

Downloaded by [Monash University Library] at 19:47 22 June 2013

(Received 21 March 2011; final version received 22 March 2011)


Molecular level calculations targeting the behaviour of asphaltenes, resins and bitumens are reviewed with an objective of
relating molecular structure and interactions to macroscale physical, chemical and mechanical properties. The review
discusses molecular dynamics simulations of asphaltenes and bitumens and briefly summarises structure elucidation,
quantum mechanics, coarse graining and thermodynamic model approaches. The molecular architecture of asphaltenes in
simulations plays an important role, with continental and archipelago asphaltenes showing different packing tendencies in
isolated systems and solutions. Alkyl side chains bonded to fused aromatic rings interfere with multiple ring stacking layers.
Diversity of molecular structure increases asphaltene aggregate formation via alternatives that improve packing. Strongly
bent aromatic rings are vacuum simulation artefacts and are absent when enough molecules can solvate unbent fused
aromatic rings. Bitumen simulations provide methods to estimate the influence of asphaltenes on dynamic properties. Future
needs include methods to incorporate chemically specific parameters into modelling approaches that address larger length
and longer timescales.
Keywords: asphaltene; model asphalt; model bitumen; molecular dynamics; molecular modelling; statistical mechanics

1. Introduction
The bituminous materials used in paving asphalts exhibit
physical, chemical and mechanical properties that depend
on underlying chemical structure, processing conditions
and the modifiers employed. The intent of processing and
additive strategies is to alter the chemistry and phase
structure within a bitumen such that desired properties
(e.g. complex modulus jG * j, phase angle d) are obtained.
Similarly, the molecules classified as asphaltenes in
bitumen exhibit phase behaviour within bulk crude oil
such that aggregation and precipitation constrain the
temperatures and pressures over which crude oils can be
recovered and refined (Speight 1999). The goal of
understanding the physicochemical principles that control
the links among chemistry, structure and properties has
motivated work to understand interactions among saturate,
aromatic, resin and asphaltene (SARA) components within
bitumens and crude oils from a molecular perspective
using thermodynamic models and simulations. This paper
reviews such approaches in order to identify the state of the
art in describing asphaltenes and bituminous materials
using molecular scale models.
Models to be considered in this review fall in two
general classes. Molecular simulations (Allen and
Tildesley 1987, Leach 1996, Frenkel and Smit 2002) use
principles of statistical mechanics to determine properties
from the separations and orientations within packings of

*Email: greenfield@egr.uri.edu
ISSN 1029-8436 print/ISSN 1477-268X online
q 2011 Taylor & Francis
DOI: 10.1080/10298436.2011.575141
http://www.informaworld.com

molecules. Equation of state approaches will be


considered briefly. They combine free energy models for
the different types of interactions among portions of
molecules to determine the total free energy, from which it
is possible to predict phase behaviour. Molecular
simulations provide more chemical detail and potentially
accurate short-time dynamic behaviour, but require much
more significant computational resources. They are limited
to nanometre length scales (1 nm 1029 m), and bulk
effects are represented using periodic boundary conditions: the right surface of a simulation region connects
directly to the left surface, forming a loop, and analogous
connections are made for the front/back and top/bottom
surfaces. Molecular simulations calculate internal energy
and enthalpy in a straightforward manner through a force
field, which is a parameterised set of equations for the
thermodynamic internal energy as a function of the
distances between atoms. Common terms include stretching, bending and torsional twisting of bonds, charge
charge Coulomb interactions and dispersion interactions
between pairs of atoms not bonded to each other (Leach
1996). See Greenfield and Zhang (2009) for a description
of molecular simulations targeted towards a bitumen and
civil engineering perspective. Within molecular simulations, calculating the free energy difference between two
states (conditions of temperature, pressure and composition) remains a challenge in practice. Equation of state

Downloaded by [Monash University Library] at 19:47 22 June 2013

326

M.L. Greenfield

approaches do not provide dynamic mechanical relaxation


results, such as jG * j. A review of equation of state
approaches that addresses flocculation of asphaltenes has
been compiled by Pina et al. (2006).
This review includes studies on asphaltenes and
bitumens (termed asphalts in the USA). Bitumens include
asphaltenes as one component. Asphaltenes are defined as
the compounds that precipitate when bitumen or crude oil
is diluted with a straight chain alkane, which is typically
heptane or pentane (Roberts et al. 1996, Speight 1999).
Depending on the separation method, additional bitumen
fractions include resins, aromatics, and saturates or polar
aromatics, naphthene aromatics and saturates (Speight
1999). Much research has been conducted on asphaltene
chemistry, as described in books (Mullins and Sheu 1998,
Mullins et al. 2007) and a recent review (Mullins 2010).
Key questions involve the size of asphaltenes, as expressed
in their molecular weight M, and their molecular
architecture. The latter category includes the number and
size of fused rings and the extent to which asphaltenes
comprise a single set of fused rings (continental model)
compared with many small sets of fused rings connected
by branches (archipelago model). Both architectures
include additional side chains of assorted lengths. Bitumen
microstructure has been reviewed recently as well
(Lesueur 2009). Key questions include the extent to
which asphaltenes form nanophase domains (colloids in
the asphaltene literature, which are distinct from colloids
of soap molecules), which indicate correlated asphaltene
resin packing compared with more random packing.
A challenge for conducting molecular simulations of
bitumen is that the covalent bonding and atom identities
within a molecule need to be specified before the
simulation can begin. The diverse types of molecules
that are found in bitumens (Wiehe and Liang 1996) and
precipitate from solution under its defining conditions
complicate setting up the simulation. Traditional practice
has been to assume some molecular structure(s) based on
experimental data. The choices made in each study are
described here in some detail for this reason.
The review considers several simulation approaches.
Structure elucidation methods (Section 2) provide
molecular structures based on experimental data. Quantum
mechanics approaches (Section 3) provide accurate results
for specific interactions between asphaltenes and resins.
They can also account for electronic effects that are absent
in classical simulations. Molecular simulations of
asphaltenes (Section 4) provide direct results for
molecular level asphaltene interactions and packings.
Molecular simulations of bitumens (Section 5) extend the
ideas of Section 4 to models of a full bitumen, rather than
only asphaltene and/or resin. Coarse-grained methods
(Section 6) provide ways to access longer lengths and
times at the expense of molecular detail. Thermodynamic
models (Section 7) provide more approximate methods for

calculating free energies at the expense of dynamics and


transport properties. A summary concludes the review
(Section 8).
2. Structure elucidation of asphaltene molecules
A small number of studies have focused on so-called
structure elucidation: developing a computer algorithm
that generates molecular structures of asphaltenes that are
self-consistent with experimental characterisations. Those
data serve as input parameters to the algorithm. Outputs
include between one and many structures that can be used
in subsequent modelling studies or even can be evaluated
on their own for physical insights. Obtaining multiple
structures is consistent with a wide range of molecules
thought to exist within the asphaltene component. Diallo
et al. (2000) reviewed the history and uses of structure
elucidation programs prior to their applications to
asphaltenes.
Michael Klein and co-workers (Neurock et al. 1990,
1994, Klein 1994, Petti et al. 1994, Trauth et al. 1994,
Neurock and Klein 2000) employed multiple data sources
to create statistical approximations of resid: the petroleum
residue after crude oil distillation is complete. The resid
was represented by a collection of 103 to 104 noninteracting molecules, with properties calculated via
structure/property relationships (Trauth et al. 1994). The
range of different molecular size distributions was used as
a parameter in optimising the predictions from these
relationships (Trauth et al. 1994). Allowing for chemical
reactions among species did not improve accuracy; instead
the required computer time was better spent in simulating
a larger system without reactions (Petti et al. 1994).
Average molecular compositions, molecular weights and
atomic ratios calculated from the simulations were
reasonably consistent with experiment, as should be
expected since those were simulation inputs. The resulting
distributions suggested molecular weight ranges from 500
to 3000 g/mol for the Hondo resid considered. Asphaltenes
were required to contain multiple polycyclic regions, i.e.
an archipelago model (Trauth et al. 1994). Averaged over
the thousands in the collection, the molecules in the resid
contained 6.4 aromatic and 3.2 naphthenic rings in an
aromatic core, 9.3 naphthenic rings elsewhere in the
molecule and 1.1 side chains, 19.0 carbon atoms long.
Kowalewski et al. (1996) characterised the asphaltene
fraction precipitated from a Boscan (Venezuelan) crude oil
fraction. They then used a structure elucidation program
SIGNATURE previously written by Faulon to create a
single asphaltene molecule. This compound with
formula C618H722O10N12S20 statistically represented the
measured molecular properties of C:H ratio, C:O ratio and
relative sizes of aromatic vs. aliphatic components.
Finally, they simulated this molecule with molecular
dynamics and found little change in the energy or

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


conformation. Similar approaches have been applied in
coal research, as reviewed by Carlson and Faulon (1994).
Diallo et al. (2000, 2004) used SIGNATURE to devise
molecular structures that were consistent with characterisation of asphaltenes from Arab Berri and light Arabian
crude oils, resulting in 10 structures each. Groups in the
structures consisted of mixed naphthenic and aromatic
rings, thiophene, pyrrole and alkane chains. Though
 n , 2044 and 1280 g/mol, respectcomplete structures (M
ively) were visualised, their precise bonding patterns and
functional groups cannot easily be determined from the
figures. Required input information in the approach of
Diallo et al. was local bonding geometry. They found it
necessary to modify ring sizes manually in order to obtain
fused ring domains that were consistent with understandings about asphaltenes.
Sheremata et al. (2004) used an extension of the
approaches of Klein and co-workers. 1H and 13C NMR
data on Athabasca asphaltene provided guidance about the
relative concentrations of different chemical environments
(aromatic, aliphatic, naphthenic, etc). An archipelago
geometry was most consistent with their data, though the
relative compositions were also reproduced using
structures with only 25 or 50% of the experimental
molecular weight of 4190 g/mol. A set of asphaltene
structures was better than a single asphaltene molecule at
representing the distributions found in their characterisation data.
Boek et al. (2009) developed an asphaltene structure
generation approach that expanded on the work of
Sheremata et al. (2004). They ultimately generated
100,000 asphaltene structures and visualised eight of
them. Overall sizes were more in support of a continental
asphaltene model with sub-1000 g/mol molecular weight
than of an archipelago model with ca. 4000 g/mol.
Comparisons were based on an objective function that
tabulated differences between experimental measurements
and model estimates based on the devised molecular
structures.
Generated asphaltene structures provide some insights
into their own. Much more information about asphaltenes
and bitumens has been obtained through molecular models
and simulations based on proposed molecular structures.
3. Quantum mechanics applied to asphaltene
molecules
Specific interaction energies between asphaltene and resin
molecules have been calculated using methods from
quantum mechanics. These methods determine an
approximate numerical solution to the Schrodinger
equation and thus incorporate electronic effects of
chemical interactions that are neglected within a classical
force field. Slight differences in results arise among the
possible quantum mechanics approaches. Generally, all of

327

the quantum chemistry methods require too much


computer time to be applied in simulations of how
asphaltene and bitumen systems evolve over time; thus,
classical force fields are used in dynamics simulations.
Further discussions are available in general references
(Parr and Yang 1989, Leach 1996).
Murgich et al. (2001) used density functional theory
(DFT) (Parr and Yang 1989) calculations with the VoskoWilk-Nusair (VWM) functional and DNP basis set to
estimate the adsorption energies between a hematite Fe2O3
surface and small molecules; the latter represent portions of
an asphaltene or resin molecule. They found that side chains
decreased adsorption energy and that little noticeable charge
transfer occurred during adsorption. They further presented
correlations between adsorption energy and aromaticity,
hydrogen/carbon ratio and molecular surface area. Disruption of intermolecular packing and energetics due to side
chains was consistent with the findings in classical
aggregation calculations that are described below.
Alvarez-Ramirez et al. used quantum mechanical DFT
calculations to estimate interaction energies and the
accuracy of the classical COMPASS force field. They used
an asphaltene structure proposed by Groenzin and Mullins
(2000) that had a pyrrole nitrogen heteroatom and seven
fused aromatic rings. An asphaltene adsorbing to an iron
oxide surface, representing pipeline material, exhibited
chemisorption based on the binding energy (AlvarezRamrez et al. 2004). Asphaltene asphaltene, asphaltene resin and resin resin interaction energies depended
on intermolecular orientation, e.g. face-to-face, T-shaped
and edge-to-edge, as a function of the closest distance
between molecules (Alvarez-Ramirez et al. 2006).
Energies calculated using the classical force field showed
differences in well depth and interaction distance
compared with the quantum mechanical results determined using the generalised gradient approximation
(GGA) functional.
Stoyanov et al. (2008b) performed DFT calculations
with the Harris approach to compare the energetics of a
combination of one archipelago and two continental
asphaltenes, representing a previously proposed Maya
system (Takanohashi et al. 2004) described below, in
aggregated and dissociated states. Partial charges were
calculated along six possible dissociation pathways and were
used in a coarse-grained solvation calculation based on the
3D-RISM method. Potential of mean force results suggested
more favourable aggregation in 1-methylnaphthalene and
more favourable dissociation in quinoline with increasing
temperature. Stoyanov et al. (2008a) also used DFT with the
PW91 functional to compare the relative adsorption energies
of substituted thiophenes onto ion-substituted zeolites, with a
motivation of improving the catalytic conversion of
asphaltenes to smaller hydrocarbons.
Schmets et al. (2009) used semi-empirical molecular
orbital calculations to investigate the stability of stacked

Downloaded by [Monash University Library] at 19:47 22 June 2013

328

M.L. Greenfield

asphaltenes. They used a coronene-like structure with


seven fused rings that was similar to but more symmetric
than those proposed by Ruiz-Morales and Mullins (2007).
Pendant alkyl side chains were omitted in the interest of
computational speed, which also reduced the possibility of
inter-asphaltene steric hindrance. For a system of two
isolated asphaltenes, they found only repulsive interactions between molecules. For sets of three to eight
asphaltenes, they found a weakly favourable parallel
between rings. This
geometry with separations of 5.55 A
implies that asphaltene aggregation occurs more to lessen
repulsions among molecule pairs rather than to enhance
strong interactions.
An exciting use of quantum mechanics of asphalts was
in determining the possible chemical structures of
asphaltene molecules. Ruiz-Morales (2002) used a variety
of quantum mechanical approaches to calculate the
difference between the highest occupied and the lowest
unoccupied molecular orbital energies. This difference
(the HOMO LUMO gap) relates to the experimentally
measurable fluorescence spectrum. Results for proposed
asphaltene structures were compared with those for the
well-defined polycyclic aromatic hydrocarbon molecules.
The results, in combination with experimental findings on
asphaltenes by others, established asphaltene characteristics that are consistent with several experimental results.
These specifics were outlined (Ruiz-Morales 2002): 5 to
10 rings make up the aromatic core, the core is not linear
nor zigzag, and the core is not a fully resonant structure.
Controversy over the possible structures of asphaltene
molecules and their molecular weight has continued, as
described in a recent review (Mullins 2010). Later work
(Ruiz-Morales and Mullins 2007) used absorption and
fluorescence spectra (i.e. colour) calculated by using the
semi-empirical ZINDO/s method to reinforce the idea that
asphaltenes contain 4 to 10 fused rings, with a majority
containing 6 to 8. Results were in support of an island
model (their term for a continental model) instead of an
archipelago model, since the latter would lead to
molecules higher in molecular weight than are found in
measurements conducted under the most dilute conditions,
which minimise aggregation. The black colour of
asphaltenes further supports the island model, given the
alkane-to-aromatic carbon ratio (Ruiz-Morales and Mullins 2007). Further molecular orbital calculations
suggested that the frequency dependence of absorbance
(i.e. the spectrum) limited the maximum size to
approximately 15 fused aromatic rings (Ruiz-Morales
et al. 2007). Ultraviolet absorption and fluorescence
spectra calculated for a distribution of fused ring sizes
were consistent with the experimental results for
asphaltenes, while the spectra calculated for ring sizes
applied in the structure elucidation method of Sheremata
et al. (2004) were found to be inconsistent (Ruiz-Morales
and Mullins 2009).

Two key conclusions may be drawn from the quantum


mechanics studies. Specific interactions are qualitatively
consistent with force field calculations, supporting the
appropriate use of force fields in classical molecular
simulations of asphaltenes and bitumens. Visualisations of
quantum chemistry results support a planar geometry for
fused aromatic rings. In addition, properties based on the
electronic structure, which require quantum mechanics to
model them, support the continental model of asphaltenes,
as shown by Ruiz Morales and co-workers and advocated
by Mullins (2010).

4. Molecular simulations of asphaltene molecules


The vast majority of molecular simulations related to
bitumen have focused on model asphaltene molecules and
their interactions with solvents, resins and other
asphaltenes. The traditional motivating question is the
nature of these interactions that lead to inopportune
asphaltene precipitation during crude oil processing. The
same fundamental intermolecular interactions occur
within a bitumen pavement. For example, force transmission across a bitumen layer between rock aggregates
ultimately relies at small enough length scales on the
forces that an asphaltene molecule can exert on its
neighbours. The question of whether those neighbours are
asphaltenes, resins, aromatics or saturates influences the
efficiency of force transmission compared to energy
dissipation, i.e. storage vs. loss modulus. While many
papers cited in this section do not discuss pavement
applications, their results on the interactions of model
asphaltene molecules can inform more direct bitumen
simulations, such as discussed in Section 5.
Brandt et al. (1995) constructed a statistical mechanical model for asphaltene stacking. They used molecular
dynamics simulations to estimate required energy
differences for several cases: a single asphaltene molecule
in liquid solution compared with one in bulk asphaltene
and a single asphaltene molecule within a stack compared
with one in bulk asphaltene. They noted difficulty stacking
asphaltene molecules on a molecular level, due to steric
hindrance (the organic chemistry term for spatial
interference) between pendant alkyl chains. They found
that predicted stack heights in their model were very
sensitive to solvent quality (interaction energy of solvent
compared with asphaltene phase) at low asphaltene
volume fractions but less so at high volume fractions.
They noted that . . . no attempt was made to calculate
. . . the average energy of an asphaltene unit sheet in a
liquidlike asphaltene phase, as this would require a
prohibitively expensive and complex simulation (Brandt
et al. 1995). Many of the papers described in this review
document the progress since 1995 towards conducting
such complex simulations in increasingly accurate ways.

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


Several early molecular simulations of asphaltenes
used classical force fields to study the associations
anticipated to occur among atoms of asphaltene and resin
molecules isolated in a vacuum. An important limitation to
recognise in these calculations is that the energy of an
asphaltene dimer (or trimer or tetramer, etc.) is compared
with the energy of an isolated asphaltene molecule.
The limitation is that a more desired comparison would be
between an asphaltene aggregate and an isolated
asphaltene molecule in solution, in a bitumen phase or
within an entire crude oil. The propensity of asphaltenes to
exist in solid or liquid phases, instead of a gas phase,
emphasises that asphaltenes prefer to be solvated by a
polar or non-polar hydrocarbon rather than surrounded by
an empty space such as a vacuum. Compared to two
isolated asphaltene monomers, an asphaltene dimer in
vacuum enables a portion of each molecule to be solvated
by the other asphaltene molecule; thus, it is not surprising
that the energy for the pair is lower. Calculations solely in
a vacuum do not have the capability to clarify the change
in potential energy that will occur when two fully solvated
asphaltenes are replaced by a fully solvated dimer. Even a
minimum energy calculation for solvated molecules is
unable to calculate the free energy difference, because an
entropy change occurs as a result of the change in possible
solvent arrangements around two asphaltene monomers
compared with one asphaltene dimer. Entropy corresponds
to the number of possible arrangements (Frenkel and Smit
2002).
Rogel (1995) was motivated by asphaltene precipitation
due to aggregation and conducted molecular dynamics
simulations of two different asphaltene molecules (A and B).
Both were continental models and contained 24 and 11 fused
aromatic rings, respectively. Both contained sulphur and
nitrogen heteroatoms, while A also contained oxygen.
The Hildebrand solubility parameter was estimated for
monomer, dimer, trimer and tetramer configurations.
The predicted solubility parameter was comparable to
experimental values (1721 MPa1/2) for a monomer of one
of the asphaltene structures, but the predictions were a bit
lower (1415 MPa1/2) for the other asphaltene monomer and
for aggregates of either structure. Note that 1 MPa1/2 1
(J/cm3)1/2 0.49 (cal/cm3)1/2. Rogel (1997) also modelled
solubility parameter distributions using group contribution
methods. Simulations (Rogel 1995) in toluene, heptane and a
system with both solvents revealed differences in intermolecular asphalteneasphaltene potential energy, though
the simulation results depended on the initial molecular
arrangements. This indicates poor sampling during the short
45 ps simulation. Images of optimised aggregate structures
indicated significant bending of aromatic rings to accommodate alkane side chains. This is potentially a computational side effect of the vacuum environment imposed on
the calculation.

329

Murgich et al. (1996) conducted molecular mechanics


(energy minimisations) using different isolated asphaltene asphaltene, resin resin and asphaltene resin packings to study the stability energies of different size
aggregates. The asphaltene structure was based on prior
spectroscopy results and contained 27 rings; it was similar
to asphaltene A of Rogel (1995). Resins contained five to
eight rings. Both had long alkyl side chains. The cff91
force field was used. Stable packings displayed parallel
separations that
ring ring interactions centred at 3.8 A
were in good agreement with experimental scattering data
(Storm et al. 1994b). Aggregates beyond three asphaltenes
were unstable due to steric repulsion induced by the alkyl
side chains. Noticeable bending of aromatic rings was
evident; this was attributed to favourable non-bonded
interactions that compensated for the significant energies
required to alter the geometry of covalent bonds.
Working together, Murgich, Rogel and co-workers
(Murgich et al. 1998) studied the associations of
asphaltene and resin onto kaolinite, a clay found in crude
oil reservoirs. They attributed the majority of the potential
energy to van der Waals interactions (70%), rather than
Coulomb interactions (20 30%) or hydrogen bonding (5
10%). Murgich (2002) later discussed this distinction at
length. However, the definition of binding energy, i.e. the
distinction between total energy and change in energy
between separated and adsorbed molecules, is not fully
clear in their work. We note that the changes in energy
contributions between the separated and adsorbed states
are what indicate important components of the adsorption
mechanism. In addition, those adsorbed systems were
simulated in the presence of a vacuum, meaning that
adsorption can potentially play a solvation role. Again,
associating with any other compound via van der Waals
interactions may be more favourable than the absence of
interactions in a vacuum.
Murgich et al. (1999) used molecular mechanics to
study adsorption of resin molecules to a particular
asphaltene molecule that had been suggested (Strausz
et al. 1992) as representative of Athabasca oil sands
asphaltene. This large archipelago asphaltene (M , 6200)
contained 10 islands of fused rings attached via alkyl
chains. The nine resins ranged from zero to three fused
rings and contained either thiophene, pyrrole, pyridine or
one of assorted oxidised functional groups. Interactions
between asphaltene and resin based on the cff91 force field
were stronger than between asphaltene and solvent, as
emphasized in later work (Murgich and Strausz 2001). The
interactions were also dependent on chemical specificity
of the resin and the asphaltene adsorption site. Later work
(Murgich et al. 2002) found that the presence of a water
molecule that bridges two aggregated asphaltene molecules can lead to significantly more favourable interaction energy. Murgich (2003) reviewed these findings and

Downloaded by [Monash University Library] at 19:47 22 June 2013

330

M.L. Greenfield

visualised a minimised configuration of an asphaltene with


an archipelago geometry.
Additional simulations by Rogel (2000) compared
interaction energies of isolated asphaltene asphaltene and
asphaltene resin dimers. The average structures used
were based on speciations of crude oils that were either
stable or unstable to asphaltene precipitation. Stable and
unstable asphaltenes contained 12 and 16 22 fused rings,
respectively; they also contained sulphur and nitrogen
heteroatoms. Resins contained four to nine rings and no
heteroatoms. The cvff force field was used in a
combination of molecular dynamics simulations and
energy minimisations. The asphaltene dimer stabilisation
energy originated from van der Waals interactions and was
stronger for the structure based on the less stable oil.
Exchanging an asphaltene in the dimer for a resin
molecule was unfavourable energetically. The distribution
of solvent molecules within a thin layer on a resin surface
was also considered. Compared to heptane, naphthalene
and toluene were better able to swell resin molecules
towards the solution phase.
Rogel and Leon (2001) simulated adsorption at low
and high concentration of ethoxylated dodecylphenol
surfactants on surfaces of asphaltenes; they explained that
these surfactants are used to lessen asphaltene precipitation. They based the asphaltene structures on work
described above (Rogel 1995) and calculated energetics
using the cvff force field within molecular dynamics
simulations. They found more favourable adsorption at
high surface concentrations (20 compared with 1 adsorbed
molecule) and noted that the thickness of the polar head
group layer was independent of the size of the head
group. The adsorbed molecules apparently packed such
that the entire head group could remain close to the
surface.
Follow-up simulations (Rogel and Carbognani 2003)
used molecular dynamics with the cvff force field to
compare how changes in asphaltene chemistry affected
bulk asphaltene density. The 12 proposed asphaltene
molecular structures ranged from 8 to 19 fused rings and
C4 to C7 side groups; these led to different C:H ratios.
Some structures (but not all) contained heteroatoms.
Periodic boundary conditions were used to represent bulk
asphaltene within the simulation. Though the equilibration
times in the simulations were longer than in their past work
(5 ns of simulated annealing, then 50 ps of equilibration at
298 K, compared with 15 30 ps of equilibration earlier),
the predicted densities (1.0 1.1 g/cm3) were well below
the experimental values (1.15 1.3 g/cm3). A correlation
from the literature led to better agreement with
experimental density for the proposed asphaltene
structures.
Sheu (1998) simulated a toluene solution using a
diverse set of 64 asphaltene molecules, representative of
Ratawi vacuum resid. A Gaussian distribution with a range

of 3 to 11 fused rings was represented simultaneously.


The MOPAC6 approach for electrostatics and the Dreiding
force field for all other energy terms were used; toluene
solvent was represented using a dielectric continuum
background. Simulations for 250 ps at 0.015 wt% showed
no aggregation beyond dimers. Simulations at 5.0 wt%
showed formation of various size aggregates, which
Sheu described as loose and irregular. One set showed
seven stacked aromatic rings, comparable to a model of
Yen (1972).
Diallo et al. (2000, 2004) used short molecular
dynamics simulations across a wide range of temperatures
to predict density, heat capacity, isothermal compressibility and solubility parameter for the asphaltene structures
they had generated (see Section 2). Simulations of
individual asphaltene molecules in bulk, using threedimensional periodic boundary conditions, led to density
estimates of 1.08 1.18 g/cm 3 for the Arab Berri
asphaltenes and 1.05 1.36 g/cm3 for the light Arabian.
Solubility parameter estimates were 17.6 21.6 MPa1/2 and
12.9 20.2 MPa1/2, respectively. For the light Arabian
system, results for the set of 10 asphaltenes simulated
within a single system led to more representative values of
1.2 g/cm3 and 18.6 MPa1/2. A step change in the light
Arabian isothermal compressibility suggested a glass
transition near 5758C for the asphaltenes, though results
for molar volume showed too much scatter to infer a
change in slope. The scatter was potentially due to
incomplete structural relaxation, as simulations proceeded
at a given temperature for only 75 ps, which is likely to be
shorter than the single molecule relaxation time.
Takanohashi et al. (2003a, 2003b) simulated a system
of three asphaltene molecules chosen to represent the
experimental characteristics of Khafji asphaltenes.
The three molecules two continental with a central core
of seven or nine fused aromatic rings and one archipelago
with two sets of seven fused rings joined by an alkyl link, all
with side chains were aggregated in a vacuum
environment, and the stability was probed at 300, 523,
573 and 673 K. Aggregation was probed by monitoring the
distance between parallel rings and the hydrogen bonding
distance between hydroxyl groups. Removing side chains
and replacing heteroatoms with carbon led to larger
dissociations of the aggregate at high temperature over the
100 ps time probed. Periodic boundary conditions were
employed, though individual molecules spanned the full
box size and thus potentially could induce artificial
correlations. Adding 1-methylnaphthalene solvent did not
disrupt this packing, while quinoline changed packing
dramatically within less than 100 ps. Similar findings were
made in systems of multiple asphaltene molecules that
were designed to represent Khafji, Maya and Iranian Light
asphaltenes (Takanohashi et al. 2004). The presence of
quinoline modified the parallel stacking found in each
system. Heating from 573 to 673 K disrupted packing in

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


two of the three systems, while the presence of 1methylnaphthalene did not change packing. These
simulations were limited to relaxations over their 100 ps
duration.
Pacheco-Sanchez et al. (2003) simulated a set of 96
asphaltene molecules that were initially separated from
one another and in vacuum. The continental asphaltene
structure was initially proposed by Groenzin and Mullins
(2000) and had eight fused rings and one thiophene
sulphur heteroatom. Over 40 ps the asphaltenes aggregated
into some dimers, trimers and larger aggregates, which
then changed little over 100 ps. The intermolecular
packings among aromatic rings showed face-to-face,
T-shaped and offset p-stacked geometries (this terminology is used by Leach (1996)). Within aggregates, there
were multimodal distributions of sulphur sulphur interaction distances in the g(r) pair distribution function.
These simulations showed some questionable attributes:
fused aromatic rings folded up to 458 in some cases, and
bin width cited for computing the pair
the 1.9 A
distribution function was more coarse than some of the
g(r) details shown.
Follow-up work (Pacheco-Sanchez et al. 2004a)
focused on the geometries and intermolecular separations
found at energy minima for four systems of different
continental asphaltenes that each contained 32 molecules.
The molecular architectures used had been designed by
Speight and Moschopedis (1979) for a Venezuelan crude
oil (16 fused rings; sulphur, nitrogen and oxygen
heteroatoms), by Groenzin and Mullins (2000) for a
California asphaltene (described above), by Murgich et al.
(1996) for a Venezuelan crude oil (described above) and by
Zajac et al. (1994) for a Mayan crude oil (12 fused rings;
sulphur and nitrogen heteroatoms). Computed structure
factors S(k) were compared with experimental results from
the literature attributed to Yen et al. (1961). Results for the
different asphaltene simulations agreed best with S(k) of
different asphaltene sources, meaning that asphaltene
chemistry differences were noticeable among the simulations and in experimental characterisation data. Related
systems in which aliphatic side chains were removed and
energy was subsequently re-minimised showed different
packings and similar chemistry-dependent trends. Significant and unphysical bending of fused aromatic rings still
occurred for larger asphaltenes, as discussed above for
simulations of isolated asphaltene structures. Short
simulations (4 ps equilibration, 12 ps sampling) addressed
increased pressures and found corresponding increases in
mass density (Pacheco-Sanchez et al. 2004b). Significant
asphaltene dissociation occurred at the lowest pressure in a
uniform dielectric background that was representative of
toluene. Asphaltene aggregation persisted in a dielectric
background representative of pentane.
Carauta et al. (2005b) used molecular dynamics to
simulate an asphaltene dimer surrounded by solvent to learn

331

the extent to which the aggregated rings increase their


separation. The asphaltenes featured 8 to 13 fused aromatic
rings, 5 naphthene (saturated) rings and C5 to C7 pendant
alkyl chains; they were designed to represent asphaltenes
from Brazilian crude oils that tend to aggregate. Over the
100 ps of the simulation, separations between fused rings
in toluene, from 3.8 to 4.7 A
in
increased from 3.8 to 5.3 A

butane and isobutane, and remained unchanged at 3.8 A in


heptane. In related work, Carauta et al. (2005a) used
molecular dynamics and quantum mechanics calculations on
isolated systems to compare different enantiomers for the
naphthenic portions of two smaller asphaltene molecules:
each had only six fused aromatic rings, three naphthenic
rings, two side chains and a hydroxyl group. Slight
differences in energy existed among the possible bonding
arrangements for the naphthene rings in each molecule.
Asphaltene dimers showed favourable energies compared
with isolated asphaltene monomers in vacuum. Visualised
conformations in both papers illustrated how pendant side
chains can impose significant steric hindrance on p stacking
of aromatic rings.
Kuznicki et al. (2008) simulated continental and
archipelago asphaltenes dissolved together in water,
toluene and heptane. The continental asphaltene had
eight fused rings; the archipelago connected sets of five
and six fused rings. Both had pyrrole (aromatic N H),
sulphide, ether and ketone heteroatom groups. The mixed
asphaltenes aggregated with each other in all three liquid
solutions, with stacking of fused rings potentially assisted
by the lack of pendant alkyl chains on one set of fused
rings in the archipelago-type molecule. Aggregation was
fastest in water, and the aggregate in toluene was smaller
than in heptane. In mixed water/toluene systems, the
asphaltene aggregate tended to be located in the toluene
phase if enough mobility was available to escape an initial
placement in water, such as if phase separation occurred in
the local region surrounding the aggregate. In a case of
solvents initially separated in the simulation box,
asphaltene aggregates remained in water throughout the
10 ns simulation. When a charged carboxylic acid group
was added at the end of an alkyl chain on the continental
asphaltene, those molecules localised at the toluene/water
interface and initiated aggregation there, with all groups of
the asphaltenes other than the charged acid group located
in the toluene phase. A key result was that the presence of
the charged acid group was a prerequisite for the
asphaltene to show surface-active behaviour; otherwise,
asphaltenes localised within a bulk phase.
Later work (Kuznicki et al. 2009) studied VO-79
(a commercially available molecule that resembles an
asphaltene), a continental asphaltene and two variations of
the continental asphaltene that had charged carboxylic acid
groups at the end of pendant alkyl chains. All four showed
aggregation after 7 ns in a mixed water/toluene system. The
charges in two asphaltenes tethered to the interface, while all

Downloaded by [Monash University Library] at 19:47 22 June 2013

332

M.L. Greenfield

other groups were in the toluene phase. The uncharged


asphaltenes were in the toluene phase as far as possible from
the surface. Individual asphaltene molecules were oriented
close to perpendicular to the surface. Self-diffusion mostly
occurred parallel to the interface as little distance
perpendicular to the interface could be explored before
reaching the other toluene/water phase boundary. Diffusion
coefficients were of order 1026 cm2/s, which is typical for
molecules in a low-viscosity fluid. Fused aromatic rings
exhibited p stacking. Visualisation of the aggregates
suggested smaller sizes than in systems with fewer pendant
alkyl chains (Kuznicki et al. 2008), though this was not noted
or quantified in that paper.
Headen, Boek and co-workers conducted molecular
dynamics simulations of several asphaltene molecules with
structures determined using the method of Boek et al. (2009).
Continental asphaltenes, archipelago asphaltenes and resins
were surrounded by toluene or heptane solvent at 7 wt% and
were simulated for 20 ns at 300 K. In the results presented by
Boek et al. (2009) for six asphaltene molecules of one
structure in either heptane or toluene, pair distribution
function g(r) suggested closer association in heptane
solution. The simulations presented by Headen et al.
(2009) showed that nanoaggregates in toluene form, break
and rearrange over nanosecond timescales. Continental
asphaltenes aggregated in heptane for longer times, but the
average g(r) pair distribution results were the same in both
solvents. Resin molecule aggregates formed and broke over
faster timescales. Aggregates of archipelago asphaltenes
showed less precise geometries and larger rearrangement
times, due to their larger size. Average orientations were near
parallel at close spacings and reduced to angles u < 60+ at
longer spacings; orientation always remained closer to
parallel than to random packing (cos u < 1=3). Two
alternate methods for calculating the free energy between
asphaltenes (i.e. potential of mean force) led to results that
were quantitatively inconsistent but did suggest little
difference in aggregation free energy between toluene and
heptane. Follow-up work (Headen and Boek 2011a)
simulated asphaltene aggregation in supercritical CO2 in
the presence and absence of limonene. Aggregation was
stronger than in heptane or toluene. The presence of 50 wt%
limonene led to larger asphalteneasphaltene spacing and
spacing fluctuations, thus disrupting aggregation. Additional
work (Headen and Boek 2011b) used constrained molecular
dynamics simulations to evaluate the potential of mean force
between an asphaltene molecule and a calcite surface.
The calculated free energy of adsorption was considered
reasonable, given the absence of solvent, the lack of
interactions among competing adsorbents and approximations in the force field.
Tarefder and Arisa (2010) simulated a physical
mixture of asphaltene and oxygen molecules. The
asphaltene was taken from Groenzin and Mullins (2000).
Glass transition temperatures were reported based on

temperature-dependent density, but the corresponding


figures showed cases in which the thermal expansion
coefficient decreased above Tg rather than increased. We
note that this is counter to the expected physical behaviour
that amorphous systems above Tg expand more than
glasses with temperature. They presented the simulations
as indicative of oxidation, but no chemical reactions
were allowed or enabled to occur in the calculations.
Changes they attributed to oxidation are instead more
likely due to the response of asphaltene molecules to the
presence of gas phase oxygen within the same simulation
box.
The overall set of asphaltene simulation results
indicates several significant results. The molecular
architecture of asphaltenes plays an important role:
continental and archipelago asphaltenes show different
packing tendencies. Alkyl side chains bonded to fused
aromatic rings interfere with multiple layers of p p
stacking. Diversity of molecular structure improves the
ability of asphaltenes to form aggregates; the aggregation
propensity for mixed asphaltene systems compared to a
single structure is one example. The visualisations by
Kuznicki et al. (2008, 2009) enable an important
observation about the trade-offs between solvation and
bending of fused aromatic rings. Compared to isolated sets
of asphaltene molecules or asphaltene resin clusters,
solutions retained parallel stacking of aromatic rings while
bending of fused aromatic rings did not occur. This
suggests that bent rings in isolated asphaltene dimers are
likely an artefact: though solvation is energetically
preferable to a vacuum interface by more than the cost
of bending fused aromatic rings, achieving solvation using
smaller molecules (which can include resins, aromatics,
saturates and/or solvent) is even more preferable
energetically. The many possible additional arrangements
of such combinations of molecules, compared to
arrangements of separated molecules, can potentially be
preferential in terms of entropy as well.

5. Molecular simulations of multicomponent


bitumens
The simulations described above focused on the molecular
interactions and thermodynamic properties of asphaltene
molecules. The motivation was increasing an understanding of how interactions among asphaltenes influence
aggregation and precipitation behaviour within an entire
crude oil and in solutions in solvents such as toluene or
heptane. As computer speeds increased, the asphaltenes in
target systems of interest evolved from molecules or
clusters isolated in a vacuum to being contained in a liquid
phase.
The role of asphaltene molecules is somewhat different
when considering bitumens. Rather than separating from

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


the majority phase through aggregation and precipitation,
asphaltenes play a significant role within the bitumen and
contribute strongly to the overall mechanical response.
The objective of bitumen molecular simulation is to
understand how the different molecules within a bitumen
interact in order to provide the dynamic mechanical
properties such as complex modulus G * which make
bitumens useful to combine with sand, rocks, additives and
sometimes polymers to make an asphalt pavement. Limits
to nanometre length scales and tens of nanosecond
timescales preclude conducting molecular simulations of
full-fledged pavements. Indeed, these small scales even
limit the choice and diversity of molecules that can be
studied within a single model bitumen system. Choosing
the molecules that constitute the model, bitumen is an
important part of a molecular simulation.
Initial molecule-based calculations on model bitumens
focused on systems that contained a single average
molecule whose chemistry was reflective of the bitumen as
a whole. Pauli et al. (2005) used correlations built into
commercial software to estimate density, index of
refraction and surface tension for the average molecular
structures proposed (Jennings et al. 1993) for the core
asphalts of the Strategic Highway Research Program
(SHRP). Results were compared with experiment for the
same asphalts and were also used to estimate Hildebrand
solubility parameters. Density and index of refraction
showed agreement in approximate magnitude between
experiment and correlation, though relative orderings
among different asphalts were not correct. Although
surface tension was underpredicted by ca. 15%, relative
trends among the average structures of SHRP core asphalts
seemed more consistent with experiment. Applying the
correlations to well-defined hydrocarbon molecules
suggested that alicyclic ring compounds (i.e. saturated
interconnected rings) would have properties most similar
to the SHRP core asphalts in terms of surface energy as a
function of carbon number. An alicyclic compound with
13 rings was predicted to show the best agreement. Its
estimated density was intermediate between the densities
of alkane and aromatic compounds, falling near the
estimated and measured densities of the SHRP core
asphalts.
Zhang and Greenfield (2007a) were the first to attack
the problem of simulating a bitumen using multiple
compounds in the same system. They began with
continental asphaltenes proposed in two separate literature
sources (Artok et al. 1999, Groenzin and Mullins 2000).
They chose dimethylnaphthalene as a resin and n-C22 as a
saturate; the latter was of a size that was most common in
the asphaltene structure elucidation study of Kowalewski
et al. (1996). The numbers of each molecule were chosen
based on composition analyses of vacuum resid
(Storm et al. 1994a). Molecular dynamics simulations
were based on the optimized potentials for liquid

333

simulations (OPLS) force field and were conducted across


temperatures of 2 35 1708C, with equilibration for 1.5 ns
followed by sampling for 3 ns or longer. Estimates of
thermal expansion coefficient and isothermal compressibility were reasonable for a bitumen, though the density
was a bit lower than that of SHRP core asphalts due to
heteroatom concentrations that were too low and saturate
concentrations that were too high. The simulations also
yielded results for the packing and orientation among
molecules in the simulation (Zhang and Greenfield 2007b).
The continental asphaltene2 with longer branches, eight
fused rings and a thiophene sulphur heteroatom, originally
proposed by Groenzin and Mullins (2000), took on
arrangements that were oriented more towards parallel
within close pair separations as temperature increased.
In contrast, arrangements were increasingly towards
perpendicular for the continental asphaltene1 with shorter
branches, a larger aromatic core of 10 fused rings and two
thiophene sulphur atoms; this molecule was originally
proposed by Artok et al. (1999). Both cases were subject to
sampling errors due to the long times required for
asphaltenes to rearrange. Intermolecular orientations were
random at distances much longer than the first neighbour
separation. Paired correlation function g(r) results suggested
that the model resin dimethylnaphthalene molecules did not
preferentially pack in the immediate vicinity of asphaltenes;
instead they packed further away, with asphaltene molecules
occupying the nearest neighbour positions. Those dimethylnaphthalene molecules closest to the asphaltene were
oriented towards being parallel to the fused aromatic rings
(angles of 5308).
A criticism of that initial bitumen molecular
simulation work is that the model system presents a
polarity gap: while the asphaltenes contain fixed aromatic
rings and heteroatoms (via thiophene rings), the
dimethylnaphthalene resin is much less polar and can
be considered a naphthene aromatic instead of a polar
aromatic. To address this concern, Zhang and Greenfield
(2008) proposed a new system in which the asphaltene
with longer branches (Groenzin and Mullins 2000) was
combined with a wide range of compounds. Polar
aromatic resins included benzoquinoline, ethylbenzothiophene and pentylthiophene. The first arose as an
example in a review of cracks in asphalt (Masson and
Lacasse 2000); the others were considered as model
resins by Murgich et al. (1999). Ethyltetralin was chosen
as a naphthene aromatic as it combined aromatic and
saturate rings within a single compound. n-C22 was
retained as a saturate. The relative population of each
SARA component and the distribution of nitrogen- and
sulphur-containing aromatics were chosen to be selfconsistent with the detailed NMR results of Jennings et al.
(1993) for the SHRP AAA-1 core asphalt. Results for
density, thermal expansion coefficient and isothermal
compressibility (Zhang and Greenfield 2008) showed

Downloaded by [Monash University Library] at 19:47 22 June 2013

334

M.L. Greenfield

noticeable improvements over the earlier results for a


three-component model bitumen. Matching the heteroatom and aromatic carbon concentrations enabled the
density to reach a value typical of a bitumen after 2 ns of
equilibration and 8 ns or more of sampling at 1 atm
pressure and temperatures of 25 1708C. Pair distribution
function results showed that the benzoquinoline and
ethylbenzothiophene preferentially packed near themselves and each other, but not near the asphaltene: g(r) for
asphaltene polar aromatic pairs was less than 1
throughout the first neighbour shell, rising to 1 (i.e.
random packing) at larger distances. Within this system
and its more gradual range of polarity, resins still did not
pack in the immediate vicinity of asphaltenes.
Zhang and Greenfield (2007c) also studied the dynamics
within model bitumens. They extended their earlier work by
continuing the simulations under NVT conditions for up to
28 ns. They used the instantaneous stress fluctuations to
calculate the viscosity (Allen and Tildesley 1987); this was
successful at 1708C, which corresponds to a hot-mix
temperature, but failed to converge at lower temperatures.
To estimate viscosity h at lower temperatures, they followed
ideas previously applied to liquids (Mondello and Grest
1997, Gordon 2003a) on applying the DebyeStokes
Einstein relationship,

tr K

vp h
;
kB T

which balances the rotation rate of a molecule in a uniform


environment against the viscous forces that impose drag on
that rotation. K is a numerical prefactor that depends on
shape and on the drag boundary conditions; vp is the average
volume of a single rotating molecule; kB is the Boltzmann
constant; T is the absolute temperature and tr is the average
rotational relaxation time of the molecule. The underlying
idea is that Equation (1) suggests that viscosity and rotational
relaxation time both scale with temperature as h=T , tr , so
assessing the temperature dependence of rotational relaxation time leads to an estimate of the temperature
dependence of viscosity. The rotational relaxation time itself
was estimated by the rate that an asphaltene changed its
direction, as indicated by a vector normal to the fused
aromatic rings. The time correlation function of this vector
decays to zero at long enough times, and the decay time can
be estimated through extrapolation if relaxation is
incomplete over the simulation duration.
Results from this approach (Zhang and Greenfield
2007c) were very good. The temperature-dependent
viscosity of naphthalene, calculated as a test case, agreed
quantitatively with measured values reported in the
literature. The temperature dependence of asphaltene
rotational relaxation time showed a nonlinear behaviour
on an Arrhenius plot and was more consistent with a
Vogel Fulcher dependence. The asphaltene relaxation

time calculated for high-temperature conditions was


within the range of average relaxation times measured
experimentally (Groenzin and Mullins 2000) for asphaltenes in toluene solutions of similar viscosity. Calculations
for the different asphaltene types showed different
temperature dependences and thus suggested different
bitumen viscosities. Results were less good in the model
AAA-1 system (Zhang and Greenfield 2008). The larger
relative concentration of smaller molecules, such as the
pentylthiophene polar aromatic, allowed for increased
molecular motion at each temperature and thus for a lower
viscosity. At 1708C the viscosity was about half that of the
earlier systems, whereas at 258C (Zhang and Greenfield
2010) it was lower by a factor of about 103. The viscosity
of this model system is much too low for it to be
considered an accurate mechanical model of a bitumen.
Larger polar and naphthene aromatic molecules are
required in order to describe speciation and dynamics
simultaneously.
Follow-up work (Zhang and Greenfield 2010) has
compared the dynamics of different molecule types and
the coupling among relaxation mechanisms. The earlier
works from Zhang and Greenfield (2007c, 2008) reported
that self-diffusion coefficients Di for each compound i in
model bitumen systems followed an Arrhenius dependence. As viscosity should also scale inversely with T/D
(the Stokes Einstein relationship), the product of diffusion coefficient and rotational relaxation time should be
constant when diffusion and rotation are coupled
(Lombardo et al. 2006). This was found to be the case
for small- and moderate-size compounds but not for the
largest compounds. The asphaltene molecules showed a
tendency for rotation to slow more than diffusion as
temperature decreased. A more general theoretical
analysis (Ngai 1999) suggests that the viscosity remains
coupled to rotation in this scenario, which emphasises how
asphaltene chemistry can play a dominant role in bitumen
mechanical properties.
Molecular simulations have also been used to study
other aspects of model bitumen systems. Zhang and
Greenfield (2008, 2010) studied how the presence of a
polystyrene chain in the bitumen phase modifies
intermolecular packing and dynamics. Increases in
viscosity were found in each of the model bitumen
systems. The rotational relaxation time increased and the
self-diffusion coefficient decreased for each compound,
while the temperature dependences were essentially
unchanged. Diffusion rotation decoupling for the polystyrene was very similar to that of the asphaltene,
indicating that both types of molecules control viscosity.
Intermolecular packing among asphaltenes and resins
showed few qualitative changes compared to the cases
without polymer. The length- and timescale limitations of
molecular dynamics simulations limited the extent to
which polymer effects could be modelled explicitly on the

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


molecular level. For example, the phase domain
microstructure that is characteristic of a bitumen modified
with styrene butadiene styrene (SBS) block copolymer
exhibits textures over scales larger than those accessible in
a single simulation (Masson et al. 2003). Alternate
methods are needed to address the properties of
polystyrene-rich phases, bitumen-rich phases and the
mechanical properties of the resulting composite.
Bhasin, Little and co-workers (Little et al. 2009,
Bhasin et al. 2011) used molecular dynamics simulations
of model bitumens to address questions of healing rate in
response to crack formation. The system initially included
a nanocrack, which was filled during the simulation via
molecular diffusion into it. Starting with average bitumen
structures (Jennings et al. 1993) and a model bitumen of
Zhang and Greenfield (2007a), they varied the balance
among asphaltene, resin and saturate concentrations and
the saturate chain length in order to relate differences in
diffusion coefficient (and indirectly in healing rate) to
differences in bitumen speciation. Measures derived from
infrared spectroscopy that were proposed earlier (Kim et al.
1990) for quantifying the chemistry dependence were
calculated and utilised to interpret the direct simulation
results. Diffusion coefficients calculated during 50 ps of
simulation depended on proximity to the interface and
showed smooth dependences on the predefined chemistry
parameters. Diffusion coefficients and thus healing rate
increased with increasing CH2/CH3 ratio, which corresponds to a longer chain length or to less branching.
Decreased diffusion rate also correlated with increases in
methylene methyl hydrogen to carbon ratio, which
corresponds to increased alkane branching. The simulation
results were consistent with earlier experimental findings
from FTIR spectroscopy.
Wang and Lu (2009) drew upon prior literature (Zhang
and Greenfield 2007a), while presenting an overview of
molecular simulations of bitumen at an interface with
quartz. More detailed results (Lu and Wang 2010a) drew
further from the literature (Artok et al. 1999, Baljon and
Robbins 1999, Zhang and Greenfield 2007a) while
outlining how they placed a three-component system
proposed by Zhang and Greenfield (2007a) near a quartz
surface and imposed shear. Shear stress was monitored for
104 time steps (approximately 10 ps) and showed sizeable
fluctuations. Interaction strengths between bitumen and
quartz were considered qualitatively acceptable, though
the shear stress on the quartz in the system (of 7 nm height)
was 4 105 GPa after conversion from the simulationscale units reported. Much longer averaging times would
be required to determine accurate estimates from the
simulation results, in part as the quartz surface would have
at the 1 m/s sliding velocity over the
moved by only 0.1 A
4
10 time steps shown in the plotted results, assuming a
typical 1 fs time step. The box height makes this
correspond to a shear rate of 1.4 108 s21. In subsequent

335

work, Lu and Wang (2010b) conducted brief molecular


dynamics simulations of two model asphalt systems. One
was an average asphalt molecule that they attribute to
Groenzin and Mullins (2001), but that appears to have
been proposed by Rogel (1995). The other system was the
multicomponent system of Zhang and Greenfield (2007a)
that included an asphaltene molecule proposed by Artok
et al. (1999). Lu and Wang (2010b) neglected to note prior
simulations of this system. Cohesive energy densities were
reported after 50 100 ps of simulation. The corresponding
solubility parameters of d , 15.0 (J/cm3)1/2 for the
average molecule and d , 16.0 (J/cm3)1/2 for the threecomponent system are close to those calculated for
asphaltene aggregates by Rogel (1995) and lower than
those of Diallo et al. (2000, 2004), as described earlier in
this review. In comparison, solubility parameters of 15.1,
16.0 and 18.2 (J/cm3)1/2 are reported (Elliott and Lira
1999) for heptane, nonane and toluene, respectively. The
similarity in solubility parameter between asphaltene and
heptane suggests that further refinements in energetics
may be necessary if these simulation predictions have
converged, as similar solubility parameters imply mutual
solubility. Shear and bulk viscosities were reported, but the
simulations are certainly shorter than the duration of
approximately 70 200 rotational relaxation times that are
recommended (Mondello and Grest 1997, Cui et al. 1998,
Gordon 2003b) for convergence of Green Kubo viscosity
calculations. Results from Zhang and Greenfield (2007c)
suggest that simulation durations of order seconds, not
nanoseconds, would be required for direct viscosity
calculations at these temperatures due to slow rotation of
asphaltene molecules. In this work, Lu and Wang also used
finite difference estimates of strain derivatives in a
minimum energy crystal configuration to calculate elastic
constants of granite and calcite.
Ren et al. (2010) were interested in quantifying the
molecular size (e.g. radius of gyration) as a function of
solubility class and chemical environment. They proposed
structures for what they called the polycyclic aromatic and
heavy resin fractions of Tahe and Suizhong residues. Both
fractions were distinct from asphaltenes but had similar
molecular weight ranges. MD simulations in vacuum
showed significant bending of aromatic rings in order to
increase intramolecular contacts. Simulations in solution
showed more extended side chains and less bent aromatic
rings, presumably since the molecule can now be solvated
by toluene, a thermodynamically good solvent. The larger
radii of gyration in solution compared to in vacuum were
consistent with good solvent behaviour, as is found in
polymer solutions.
In total, bitumen molecular simulations have made an
exciting start. Questions about composition effects have
begun to be asked. Needs exist for more advanced
compositions that simultaneously account for proper
SARA balance, heteroatom chemical speciation, and

336

M.L. Greenfield

asphaltene diversity while achieving dynamics that show


proper rates and temperature dependence. Comparisons
with direct experimental characterisation data, such as the
approach pursued by Bhasin and co-workers, will assist
these efforts and will provide further quantitative support
to simulation results.

Downloaded by [Monash University Library] at 19:47 22 June 2013

6. Coarse-grained simulations of asphaltene


molecules
Quantum mechanics can provide energies for a small
number of atoms or molecules but cannot be used for
asphaltene or bitumen dynamics. Classical molecular
dynamics simulations allow for more atoms but are limited
to nanometre lengths and tens of nanosecond durations. An
alternative to quantum mechanics and to fully detailed
molecular mechanics and dynamics calculations is to
determine a simpler functional form for molecule
molecule interactions, rather than atom atom interactions,
and then to conduct simulations using that so-called
potential of mean force (Leach 1996). Each molecule is
represented as a single site or as a few sites. This allows
simulations to span larger lengths and longer times, at the
expense of precise molecular detail.
Ortega-Rodriguez (2001) used a classical force field (socalled molecular mechanics) to calculate the interaction
energies between asphalteneasphaltene, asphalteneresin
and resinresin pairs in an orientation of parallel aromatic
rings. The asphaltene proposed by Zajac et al. (1994) and
described above was used. Long pendant side chains were
frozen in a single-molecule minimum-energy conformation,
with non-planar bending of the fused aromatic rings. Within
these limits, asphaltene asphaltene interactions were
approximately two and four times more favourable than
asphalteneresin and resinresin interactions, respectively.
The interaction energies were fit to an equation that
combined screened Coulomb and dispersion terms. Next
these potential functions were used in a coarse-grained
simulation (Ortega-Rodrguez et al. 2003). Each asphaltene
and resin molecule was represented by a sphere at its centre
of mass, and the solvent (heptane, toluene or pyridine) was
represented indirectly via its dielectric constant. Significant
aggregation of asphaltenes and resins was found for effective
heptane and toluene media, while packing remained
disordered in pyridine. The pyridine results were also
described accurately using the OrnsteinZernike integral
equation (Hansen and McDonald 2006). Follow-up work
(Ortega-Rodriguez et al. 2004) illustrated the utility of the
integral equation approach for mapping phase boundaries
between dissolution and aggregation that are self-consistent
with the averaged, spherically symmetric interaction
energies of the coarse-grained model. An advantage of this
approach is its speed over fully atomistic molecular
simulations. Disadvantages are the loss of detail, such as
the importance of orientation in asphaltene aggregation, and

the propagation of errors in chemical inputs when


determining the coarse-grained potential, such as the
possible inaccuracies that could result from chemical
structures that possess bent aromatic rings and immovable
side chains.
Muller and his students were joined by Murgich in
extending the work of Ortega-Rodriguez et al. They
accounted for orientation by representing each asphaltene
using a planar set of seven spheres, while each resin remained
a single sphere (Aguilera-Mercado et al. 2006). LennardJones parameters were obtained through regression to the
molecular mechanics results of Ortega-Rodriguez et al.
(2001). Solvent effects were incorporated via the dielectric
constant (for repulsions) and the Hamaker constant (for
attractions). Using Monte Carlo simulations, they found a
continuous transition in the extent of asphaltene aggregation
as a function of solvent quality. Aggregation trends as
functions of temperature and asphaltene concentration
followed experimental trends well. Resin molecules were
not found to solubilise asphaltenes. Instead, they filled gaps
between asphaltene molecules.
Boek et al. (2010) used stochastic solvent dynamics and
coarse-grained molecular dynamics to study asphaltene
deposition during capillary flow. The interactions among the
same asphaltene molecules studied by Headen et al. (2009)
were approximated at intermediate distances using a 1/r 2
dependence. The approximate well depth and length scale
were then used in a scaling argument to relate flow rate and
interaction strengths used in the stochastic simulations to
interpretations of experimental measurements.
Zhang et al. (2010) applied dissipative particle dynamics
(DPD) to petroleum systems. Rotational terms were added so
the fused aromatic rings of asphaltenes could be approximated as rigid planar units of fewer DPD particles. They
simulated a model of heavy crude oil and of oil components
in solvents. For model asphaltenes whose structure was
dominated by fused rings and which contained few short side
chains, they found significant long-range ordering. Increasing the number of side chains decreased aggregation
significantly. p-stacked, offset p-stacked and
T-shaped geometries were all observed within the simulation
of heavy crude oil. Emulsion microstructures were predicted
in simulations of heavy crude oil with water. The speed of
DPD simulations enabled long timescales to be reached.
The connections between the energy parameters and
molecular scale interactions are not yet well understood,
however, meaning that DPD simulations incorporate
chemical effects only approximately.
The few coarse-grained studies conducted to date have
begun to reveal the possibilities for using multiple scales
to address asphaltenes and bitumens. Maintaining
orientation effects, as in the works of Aguilera-Mercado
et al. (2006) and of Zhang et al. (2010), will be important
in future studies so the underlying chemistry and physics
are not lost.

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


7. Equations of state for asphaltene precipitation
More molecularly coarse molecular thermodynamics
models have been developed to address the conditions
under which asphaltene precipitation occurs. As with
asphaltene simulations, the motivation is to reduce
asphaltene-induced problems in crude oil processing.
This section of the review is limited to a sample of the
many thermodynamic models available in the literature.
For example, Victorov and co-workers (Victorov and
Firoozabadi 1996, Victorov and Smirnova 1999) proposed
an asphaltene precipitation model based on self-assembly
of asphaltenes and resins. The resin-stabilised asphaltene
micelle geometry was related to shape parameters. They
obtained parameters that led to good agreement with
precipitation data.
Some studies have applied the statistical associating
fluid theory (SAFT) equation of state (Chapman et al.
1988, 1990) to the question of asphaltene precipitation.
This equation of state determines the free energy via
individual contributions based on an ideal gas reference
state plus hard sphere repulsions, van der Waals
interactions, chain bonding and specific associations; see
Muller and Gubbins (2001) for a review. The primary
motivation in existing studies has been reproducing phase
behaviour that explains the precipitation of asphaltenes
during crude oil production and refinement.
Wu et al. (1998) developed a molecular thermodynamics framework for asphaltene oil phase equilibria,
which included a colloidal model and specific associations
between resins and asphaltenes. They applied it (Wu et al.
2000) to asphaltene precipitation. With this model, they
predicted regions of single or multiple phases among
asphaltene, resin and crude oil components. The
asphaltenes and resins were each represented as a single
component; the oils were treated as a continuous medium
(Wu et al. 1998). The oils in the model distinguish the
system from bitumen.
Chapman and co-workers (Ting et al. 2003, Gonzalez
et al. 2005, 2007) used a lumping scheme based on
chemical families (saturates, aromatics/resins and asphaltenes) to determine SAFT parameters. Their results predict
pressure-dependent density, toluene solubility and reservoir phase behaviour. Vargas et al. (2009b) recently
reviewed how perturbed chain (PC) SAFT can describe
pressure, temperature, polydispersity and processing (gas
injection and oil-based mud) effects on asphaltene
precipitation. Many of the precipitation phenomena are
described there in terms of changes in solvent solubility
parameter. For example, increasing temperature increases
the relative size of solvent molecules compared to
asphaltenes due to molecular packing effects. This
decreases the solvent solubility parameter and hence the
asphaltene solubility. Reduced variables based on cohesive
energy density (i.e. solubility parameters) can be used to

337

create phase diagrams of asphaltene phase behaviour that


unify precipitation results for different systems (Vargas
et al. 2009a).
Quinones-Cisneros et al. (2004, 2005) developed
approaches for predicting and correlating the density and
viscosity of light and heavy oils as functions of
temperature, pressure and composition. The composition
ranges specifically included terms for the sizes of
molecules found in unmodified bitumens. Their methods
combine a cubic equation of state with viscosity
correlations. The results are accurate in single and
multiple phase regions (i.e. the natural gas vapours that
evaporate from crude oil), particularly when data at some
conditions are used to tune the parameters that arise in the
so-called mixing rules for the parameters that depend on
composition and molecular weight.
This short section of the review includes only a few of
the thermodynamic models applied to questions of
asphaltene precipitation. The equation of state approaches
show promise for being extended to bitumen systems.
Defining appropriate thermodynamics questions, such as
conditions for co-existing bitumen, wax, and polymer
phases and partitioning of SARA components among
them, will be important for advancing the uses of equation
of state approaches in bitumen research.
8.

Summary and future prospects

The simulation and modelling studies of asphaltene and


bitumen systems conducted to date have focused on a few
main areas. The majority of papers have calculated
asphaltene asphaltene and asphaltene resin interactions
and have compared the effects of different asphaltene
molecular architectures. Asphaltene asphaltene associations were calculated in vacuum, and simulations in
condensed phases have found that associations persist and
are consistent with experimental knowledge of dissolution
in particularly good solvents (e.g. pyridine). Some
aggregations persist in toluene and in model bitumen.
Side chains impede extended packing, and the extent of
aggregation depends on the number of fused aromatic
rings and on whether bonding geometry shows a
continental or archipelago architecture. Diversity in
structure and composition appears important for achieving
a well-packed structure without unphysical concepts such
as bent aromatic rings; the multicomponent nature of
bitumen is important for achieving low-energy phases.
Quantum mechanics calculations provide important
support in favour of the continental asphaltene structure;
reasons include UV absorption, fluorescence and colour.
Asphaltene and polymer molecules control bitumen
dynamical properties through their rotations.
Future work in the area must ultimately address larger
scales. Greenfield (2009) has described the role of
molecular simulations within a multi-scale approach,

338

M.L. Greenfield

including ideas about how to combine calculations across


different scales within the context of bitumens. Upscaling
methods such as coarse graining of molecular details for
introducing parameters with chemical specificity into
methods that address larger length and longer timescales
will be important for achieving a path for implementing
molecular methods into more common bitumen design
methods.

Acknowledgements

Downloaded by [Monash University Library] at 19:47 22 June 2013

This work was supported by funds provided by the Rhode Island


Department of Transportation (Research and Technology
Division), the University of Rhode Island Transportation Center
and the Asphalt Research Consortium of the Federal Highway
Administration (contract DTFH61-07-H-00009), via a subcontract from the Western Research Institute.

References
Aguilera-Mercado, B., et al., 2006. Mesoscopic simulation of
aggregation of asphaltene and resin molecules in crude oils.
Energy Fuels, 20, 327 338.
Allen, M.P and Tildesley, D.J., 1987. Computer simulation of
liquids. New York: Oxford University Press.
Alvarez-Ramrez, F., et al., 2004. Docking of an asphaltene
molecular model on a Fe2O3 surface, an ab initio simulated
annealing. Petroleum Science and Technology, 22, 915926.
Alvarez-Ramirez, F., Ramirez-Jaramillo, E. and Ruiz-Morales,
Y., 2006. Calculation of the interaction potential curve
between asphaltene asphaltene, asphaltene resin, and
resin resin systems using density functional theory.
Energy Fuels, 20, 195 204.
Artok, L., et al., 1999. Structure and reactivity of petroleumderived asphaltene. Energy Fuels, 13, 287 296.
Baljon, A.R.C. and Robbins, M.O., 1999. A molecular view of
bond rupture. Computational and Theoretical Polymer
Science, 9, 35 40.
Bhasin, A., et al., 2011. Use of molecular dynamics to investigate
self-healing mechanisms in asphalt binders. Journal of
Materials in Civil Engineering, 23, 485 492.
Boek, E.S., Yakovlev, D.S. and Headen, T.F., 2009. Quantitative
molecular representation of asphaltenes and molecular
dynamics simulation of their aggregation. Energy Fuels,
23, 1209 1219.
Boek, E.S., et al., 2010. Multi-scale simulation and experimental
studies of asphaltene aggregation and deposition in capillary
flow. Energy Fuels, 24, 2361 2368.
Brandt, H.C.A., et al., 1995. Thermodynamic modeling of
asphaltene stacking. Journal of Physical Chemistry, 99,
10430 10432.
Carauta, A.N.M., et al., 2005a. Conformational search and
dimerization study of average structures of asphaltenes.
Journal of Molecular Structure (Theochem), 755, 1 8.
Carauta, A.N.M., et al., 2005b. Modeling solvent effects on
asphaltene dimers. Energy Fuels, 19, 1245 1251.
Carlson, G.A. and Faulon, J., 1994. Applications of molecular
modeling in coal research. Preprints American Chemical
Society Division of Fuel Chemistry, 39 (1), 18 21.
Chapman, W.G., Jackson, G. and Gubbins, K.E., 1988. Phase
equilibria of associating fluids: chain molecules with
multiple bonding sites. Molecular Physics, 65, 1057 1079.

Chapman, W.G., et al., 1990. New reference equation of state for


associating liquids. Industrial and Engineering Chemistry
Research, 29, 1709 1721.
Cui, S.T., Cummings, P.T. and Cochran, H.D., 1998. The
calculation of viscosity of liquid n-decane and n-hexadecane
by the Green Kubo method. Molecular Physics, 93,
117 121.
Diallo, M.S., et al., 2000. Thermodynamic properties of
asphaltenes: a predictive approach based on computer
assisted structure elucidation and atomistic simulations. In:
T.F. Yen and G.V. Chilingarian eds. Asphalts and asphaltenes
(Developments in Petroleum Science Series). Vol. 40B.
New York: Elsevier, 103 127.
Diallo, M.S., et al., 2004. Thermodynamic properties of
asphaltenes through computer assisted structure elucidation
and atomistic simulations. 1. Bulk Arabian light asphaltenes.
Petroleum Science and Technology, 22, 877 899.
Elliott, J.R. and Lira, C.T., 1999. Introductory chemical
engineering thermodynamics. Upper Saddle River, NJ:
Prentice Hall.
Frenkel, D. and Smit, B., 2002. Understanding molecular
simulation. 2nd ed. San Diego, CA: Academic Press.
Gonzalez, D.L., et al., 2005. Prediction of asphaltene instability
under gas injection with the PC-SAFT equation of state.
Energy Fuels, 19, 1230 1234.
Gonzalez, D.L., et al., 2007. Modeling of asphaltene precipitation due to changes in composition using the perturbed
chain statistical associating fluid theory equation of state.
Energy Fuels, 21, 1231 1242.
Gordon, P.A., 2003a. Characterizing isoparaffin transport
properties with Stokes Einstein relationships. Industrial
and Engineering Chemistry Research, 42, 7025 7036.
Gordon, P.A., 2003b. Influence of simulation details on
thermodynamic and transport properties in molecular
dynamics of fully flexible molecular models. Molecular
Simulation, 29, 479 487.
Greenfield, M.L., 2009. Bitumen at the molecular level:
molecular simulations and chemo-mechanics. In: N. Kringos,
ed. Chemomechanics of bituminous materials (Proceedings
of the International Workshop on the Chemomechanics of
Bituminous Materials), Delft, the Netherlands: Group of
Mechanics of Infrastructure Materials, 9 15.
Greenfield, M.L. and Zhang, L., 2009. Developing model asphalt
systems using molecular simulation, Technical report Project
No. 000216, University of Rhode Island Transportation
Center available on-line via http://www.uritc.uri.edu.
Groenzin, H. and Mullins, O.C., 2000. Molecular size and
structure of asphaltenes from various sources. Energy Fuels,
14, 677 684.
Groenzin, H. and Mullins, O.C., 2001. Molecular size and
structure of asphaltenes. Petroleum Science and Technology,
19, 219 230.
Hansen, J.P. and McDonald, I.R., 2006. Theory of simple liquids.
3rd ed. Amsterdam: Academic Press.
Headen, T.F., Boek, E.S. and Skipper, N.T., 2009. Evidence
for asphaltene nanoaggregation in toluene and heptane
from molecular dynamics simulations. Energy Fuels,
23, 1220 1229.
Headen, T.F. and Boek, E.S., 2011a. Molecular dynamics
simulations of asphaltene aggregation in supercritical carbon
dioxide with and without limonene. Energy Fuels, 25, 503508.
Headen, T.F. and Boek, E.S., 2011b. Potential of mean force
calculation from molecular dynamics simulation of asphaltene
molecules on a calcite surface. Energy Fuels, 25, 449502.

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


Jennings, P.W., et al., 1993. Binder characterization and
evaluation by nuclear magnetic resonance spectroscopy,
Technical report SHRP-A-335, Strategic Highway Research
Program Available on-line from http://gulliver.trb.org/
publications/shrp/SHRP-A-335.pdf.
Kim, Y.R., Little, D.N. and Benson, F.C., 1990. Chemical and
mechanical evaluation on healing mechanism of asphalt
concrete. Journal of the Association Asphalt Paving
Technologists, 59, 240 275.
Klein, M.T., 1994. Monte Carlo simulation of asphaltene
structure and reactivity. Preprints American Chemical
Society Division of Petroleum Chemistry, 39, 208 209.
Kowalewski, I., et al., 1996. Preliminary results on molecular
modeling of asphaltenes using structure elucidation programs in conjunction with molecular simulation programs.
Energy Fuels, 10, 97 107.
Kuznicki, T., Masliyah, J.H. and Bhattacharjee, S., 2008.
Molecular dynamics study of model molecules resembling
asphaltene-like structures in aqueous organic solvent
systems. Energy Fuels, 22, 2379 2389.
Kuznicki, T., Masliyah, J.H. and Bhattacharjee, S., 2009.
Aggregation and partitioning of model asphaltenes at
toluene water interfaces: molecular dynamics simulations.
Energy Fuels, 23, 5027 5035.
Leach, A.R., 1996. Molecular modelling: principles and
applications. Essex, UK: Addison Wesley Longman.
Lesueur, D., 2009. The colloidal structure of bitumen:
consequences on the rheology and on the mechanisms of
bitumen modification. Advances in Colloid and Interface
Science, 145, 42 82.
Little, D.N., Bhasin, A. and Greenfield, M., 2009. Intrinsic
healing in asphalt binders measurement and impact of
molecular morphology. In: N. Kringos, ed. Chemomechanics
of bituminous materials (Proceedings of the International
Workshop on the Chemomechanics of Bituminous Materials),
Delft, the Netherlands: Group of Mechanics of Infrastructure
Materials, 55 59.
Lombardo, T.G., Debenedetti, P.G. and Stillinger, F.H., 2006.
Computational probes of molecular motion in the Lewis
Wahnstrom model for ortho-terphenyl. Journal of Chemical
Physics, 125, 174507.
Lu, Y. and Wang, L., 2010a. Molecular dynamics simulation to
characterize asphalt aggregate interfaces. In: J.D. Frost, ed.
Characterization and behavior of interfaces. Amsterdam:
IOS Press, 125 130.
Lu, Y. and Wang, L., 2010b. Nanoscale modeling of the
mechanical properties of asphalt and aggregate. In: R.A.
Tarefder, Y.R. Kim, Z. You and L. Wang eds. Pavement and
materials: testing and modeling in multiple length scales.
Reston, VA: American Society of Civil Engineers (ASCE),
43 53.
Masson, J.F. and Lacasse, M.A., 2000. A review of adhesion
mechanisms at the crack sealant/asphalt concrete interface.
In: A.T. Wolf, ed. Proceedings of the third International
RILEM Symposium on Durability of Building and Construction Sealants. Paris: RILEM, 259 274.
Masson, J.F., et al., 2003. Thermodynamics, phase diagrams,
and stability of bitumen-polymer blends. Energy Fuels,
17, 714 724.
Mondello, M. and Grest, G.S., 1997. Viscosity calculations of nalkanes by equilibrium molecular dynamics. Journal of
Chemical Physics, 106, 9327 9336.
Muller, E.A. and Gubbins, K.E., 2001. Molecular-based
equations of state for associating fluids: a review of SAFT

339

and related approaches. Industrial and Engineering Chemistry Research, 40, 2193 2211.
Mullins, O.C., 2010. The modified Yen model. Energy Fuels,
24, 2179 2207.
Mullins, O.C. and Sheu, E.Y., 1998. Structures and dynamics of
asphaltenes. New York: Plenum Press.
Mullins, O.C., et al., 2007. Asphaltenes, heavy oils, and
petroleomics. New York: Springer.
Murgich, J., 2002. Intermolecular forces in aggregates of
asphaltenes and resins. Petroleum Science and Technology,
20, 983 997.
Murgich, J., 2003. Molecular simulation and the aggregation of
the heavy fractions in crude oils. Molecular Simulation,
29, 451 461.
Murgich, J. and Strausz, O.P., 2001. Molecular mechanics of
aggregates of asphaltenes and resins of the Athabasca oil.
Petroleum Science and Technology, 19, 231 243.
Murgich, J., Rodrguez M, J. and Aray, Y., 1996. Molecular
recognition and molecular mechanics of micelles of some
model asphaltenes and resins. Energy Fuels, 10, 68 76.
Murgich, J., et al., 1998. Interatomic interactions in the
adsorption of asphaltenes and resins on kaolinite calculated
by molecular dynamics. Energy Fuels, 12, 339 343.
Murgich, J., Abanero, J.A. and Strausz, O.P., 1999. Molecular
recognition in aggregates formed by asphaltene and resin
molecules from the Athabasca oil sand. Energy Fuels, 13,
278 286.
Murgich, J., et al., 2001. A molecular mechanics-density
functional study of the adsorption of fragments of
asphaltenes and resins on the (001) surface of Fe2O3.
Petroleum Science and Technology, 19, 437 455.
Murgich, J., et al., 2002. Molecular mechanics and microcalorimetric investigations of the effects of molecular water on
the aggregation of asphaltenes in solutions. Langmuir,
18, 9080 9086.
Neurock, M. and Klein, M.T., 2000. Monte Carlo simulation of
asphaltene structure, reactivity and reaction pathways. In:
T.F. Yen and G.V. Chilingarian, eds. Asphalts and
asphaltenes. (Developments in Petroleum Science Series)
Vol. 40B. New York: Elsevier, 59 101.
Neurock, M., et al., 1990. Monte Carlo simulation of complex
reaction systems: molecular structure and reactivity in
modeling heavy oils. Chemical Engineering Science,
45, 2083 2088.
Neurock, M., et al., 1994. Molecular representation of complex
hydrocarbon feedstocks through efficient characterization
and stochastic algorithms. Chemical Engineering Science,
49, 4153 4177.
Ngai, K.L., 1999. Alternative explanation of the difference
between translational diffusion and rotational diffusion in
supercooled liquids. Journal of Physical Chemistry B, 103,
10684 10694.
Ortega-Rodriguez, A., et al., 2001. Interaction energy in Mayaoil asphaltenes: a molecular mechanics study. Petroleum
Science and Technology, 19, 245 256.
Ortega-Rodrguez, A., et al., 2003. Molecular view of the
asphaltene aggregation behavior in asphaltene resin mixtures. Energy Fuels, 17, 1100 1108.
Ortega-Rodriguez, A., et al., 2004. Stability and aggregation of
asphaltenes in asphaltene resin solvent mixtures. Energy
Fuels, 18, 674 681.
Pacheco-Sanchez, J.H., Zaragoza, I.P. and Martnez-Magadan,
J.M., 2003. Asphaltene aggregation under vacuum at
different temperatures by molecular dynamics. Energy
Fuels, 17, 1346 1355.

Downloaded by [Monash University Library] at 19:47 22 June 2013

340

M.L. Greenfield

lvarez-Ramrez, F. and MartnezPacheco-Sanchez, J.H., A


Magadan, J.M., 2004a. Morphology of aggregated asphaltene structural models. Energy Fuels, 18, 1676 1686.
Pacheco-Sanchez, J.H., Zaragoza, I.P. and Martnez-Magadan,
J.M., 2004b. Preliminary study of the effect of pressure on
asphaltene disassociation by molecular dynamics. Petroleum
Science and Technology, 22, 927 942.
Parr, R.G. and Yang, W., 1989. Density-functional theory of
atoms and molecules. New York: Oxford University Press.
Pauli, A.T., et al., 2005. Assessment of physical property
prediction based on asphalt average molecular structures.
Preprints American Chemical Society Division of Petroleum
Chemistry, 50, 255 259.
Petti, T.F., et al., 1994. CPU issues in the representation of the
molecular structure of petroleum resid through characterization, reaction, and Monte Carlo modeling. Energy Fuels,
8, 570 575.
Pina, A., Mougin, P. and Behar, E., 2006. Characterisation of
asphaltenes and modelling of flocculation: state of the art. Oil
and Gas Science and Technology Revue IFP, 61, 319 343.
Quinones-Cisneros, S.E., et al., 2004. Viscosity modeling and
prediction of reservoir fluids: from natural gas to heavy oils.
International Journal of Thermophysics, 25, 1353 1366.
Quinones-Cisneros, S.E., Andersen, S.I. and Creek, J., 2005.
Density and viscosity modeling and characterization of
heavy oils. Energy Fuels, 19, 1314 1318.
Ren, W., et al., 2010. Molecular size characterization of heavy oil
fractions in vacuum and solution by molecular dynamic
simulation. Frontiers of Chemical Engineering in China,
4, 250 256.
Roberts, F.L., et al., 1996. Hot mix asphalt materials, mixture
design, and construction. 2nd ed. Lanham, MD: National
Asphalt Pavement Association.
Rogel, E., 1995. Studies on asphaltene aggregation via
computational chemistry. Colloids Surface A, 104, 85 93.
Rogel, E., 1997. Theoretical estimation of the solubility
parameter distributions of asphaltenes, resins, and oils from
crude oils and related materials. Energy Fuels, 11, 920925.
Rogel, E., 2000. Simulation of interactions in asphaltene
aggregates. Energy Fuels, 14, 566 574.
Rogel, E. and Carbognani, L., 2003. Density estimation of
asphaltenes using molecular dynamics simulations.
Energy Fuels, 17, 378 386.
Rogel, E. and Leon, O., 2001. Study of the adsorption of alkylbenzene-derived amphiphiles on an asphaltene surface using
molecular dynamics simulations. Energy Fuels, 15,
1077 1086.
Ruiz-Morales, Y., 2002. HOMO-LUMO gap as an index of
molecular size and structure for polycyclic aromatic
hydrocarbons (PAHs) and asphaltenes: a theoretical study.
I. Journal of Physical Chemistry A, 106, 11283 11308.
Ruiz-Morales, Y. and Mullins, O.C., 2007. Polycyclic aromatic
hydrocarbons of asphaltenes analyzed by molecular orbital
calculations with optical spectroscopy. Energy Fuels, 21,
256 265.
Ruiz-Morales, Y. and Mullins, O.C., 2009. Measured and
simulated electronic absorption and emission spectra of
asphaltenes. Energy Fuels, 23, 11691177.
Ruiz-Morales, Y., Wu, X. and Mullins, O.C., 2007. Electronic
absorption edge of crude oils and asphaltenes analyzed by
molecular orbital calculations with optical spectroscopy.
Energy Fuels, 21, 944 952.
Schmets, A.J.M., et al., 2009. First-principles investigation of the
multiple phases in bituminous materials: the case of
asphaltene stacking. In: A. Loizos, M.N. Partl, T. Scarpas

and I. Al-Qadi eds. Advanced testing and characterisation of


bituminous materials. Vol. 1. Leiden, NL: CRC Press,
143 150.
Sheremata, J.M., et al., 2004. Quantitative molecular representation and sequential optimization of Athabasca asphaltenes.
Energy Fuels, 18, 1377 1384.
Sheu, E.Y., 1998. Self association of asphaltenes: structure and
molecular packing. In: O.C. Mullins and E.Y. Sheu, eds.
Structures and dynamics of asphaltenes. New York:
Plenum Press, 115 144.
Speight, J.G., 1999. The chemistry and technology of petroleum.
New York: Marcel Dekker.
Speight, J.G. and Moschopedis, S.E., 1979. Some observation on
the molecular nature of petroleum asphaltenes. Preprints
American Chemical Society Division of Petroleum Chemistry, 24, 910 923.
Storm, D.A., et al., 1994a. Molecular representations of Ratawi
and Alaska North slope asphaltenes based on liquid- and
solid-state NMR. Energy Fuels, 8, 561 566.
Storm, D.A., et al., 1994b. A comparison of the macrostructure of
Ratawi asphaltenes in toluene and vacuum residue.
Energy Fuels, 8, 567 569.
Stoyanov, S.R., Gusarov, S. and Kovalenko, A., 2008a.
Modelling of bitumen fragment adsorption on Cu and
Ag exchanged zeolite nanoparticles. Molecular Simulation,
34, 943 951.
Stoyanov, S.R., Gusarov, S. and Kovalenko, A., 2008b.
Multiscale modelling of asphaltene disaggregation.
Molecular Simulation, 34, 953 960.
Strausz, O.P., Mojelsky, T.W. and Lown, E.M., 1992.
The molecular structure of asphaltene: an unfolding story.
Fuel, 71, 1355 1363.
Takanohashi, T., et al., 2003a. Molecular dynamics simulation of
the heat-induced relaxation of asphaltene aggregates.
Energy Fuels, 17, 135 139.
Takanohashi, T., Sato, S. and Tanaka, R., 2003b. Molecular
dynamics simulation of structural relaxation of asphaltene
aggregates. Petroleum Science and Technology, 21, 491 505.
Takanohashi, T., Sato, S. and Tanaka, R., 2004. Structural
relaxation behaviors of three different asphaltenes using MD
calculations. Petroleum Science and Technology, 22,
901 914.
Tarefder, R.A. and Arisa, I.R., 2010. Molecular dynamic
simulation of oxidative aging in asphaltene. In: R.A.
Tarefder, Y.R. Kim, Z. You and L. Wang, eds. Pavement
and materials: testing and modeling in multiple length
scales. Reston, VA: American Society of Civil Engineers
(ASCE), 16 30.
Ting, P.D., Hirasaki, G.J. and Chapman, W.G., 2003. Modeling
of asphaltene phase behavior with the SAFT equation of
state. Petroleum Science and Technology, 21, 641 661.
Trauth, D.M., et al., 1994. Representation of the molecular
structure of petroleum resid through characterization and
Monte Carlo modeling. Energy Fuels, 8, 576 580.
Vargas, F.M., et al., 2009a. Development of a general method for
modeling asphaltene stability. Energy Fuels, 23, 1147 1154.
Vargas, F.M., et al., 2009b. Modeling asphaltene phase behavior
in crude oil systems using the perturbed chain form of the
statistical associating fluid theory (PC-SAFT) equation of
state. Energy Fuels, 23, 11401146.
Victorov, A.I. and Firoozabadi, A., 1996. Thermodynamic
micellization model of asphaltene precipitation from
petroleum fluids. AIChE Journal, 42, 1753 1764.
Victorov, A.I. and Smirnova, N.A., 1999. Description of
asphaltene polydispersity and precipitation by means of

Downloaded by [Monash University Library] at 19:47 22 June 2013

International Journal of Pavement Engineering


thermodynamic model of self-assembly. Fluid Phase
Equilibria, 158 160, 471480.
Wang, L. and Lu, Y., 2009. Atomistic investigation into the
chemo-mechanics properties of bitumen rock interface. In:
N. Kringos, ed. Chemomechanics of bituminous materials
(Proceedings of the International Workshop on the
Chemomechanics of Bituminous Materials). Delft, the
Netherlands: Group of Mechanics of Infrastructure
Materials, 65 68.
Wiehe, I.A. and Liang, K.S., 1996. Asphaltenes, resins, and other
petroleum macromolecules. Fluid Phase Equilibria, 117,
201 210.
Wu, J., Prausnitz, J.M. and Firoozabadi, A., 1998. Molecularthermodynamic framework for asphaltene-oil equilibria.
AIChE Journal, 44, 11881199.
Wu, J., Prausnitz, J.M. and Firoozabadi, A., 2000. Molecular
thermodynamics of asphaltene precipitation in reservoir
fluids. AIChE Journal, 46, 197 209.
Yen, T.F., 1972. Present status of structure of petroleum heavy
ends and its significance to various technical applications.
Preprints American Chemical Society Division of Petroleum
Chemistry, 17, F102 F114.
Yen, T.F., Erdman, J.G. and Pollack, S.S., 1961. Investigation of
the structure of petroleum asphaltenes by X-ray diffraction.
Analytical Chemistry, 33, 1587 1594.

341

Zajac, G.W., Sethi, N.K. and Joseph, J.T., 1994. Molecular


imaging of petroleum asphaltenes by scanning-tunnelling
microscopy: verification of structure from 13C and proton
nuclear magnetic resonance data. Scanning Microscopy, 8,
463 470.
Zhang, L. and Greenfield, M.L., 2007a. Analyzing properties of
model asphalts using molecular simulation. Energy Fuels,
21, 1712 1716.
Zhang, L. and Greenfield, M.L., 2007b. Molecular orientation in
model asphalts using molecular simulation. Energy Fuels,
21, 1102 1111.
Zhang, L. and Greenfield, M.L., 2007c. Relaxation time,
diffusion, and viscosity analysis of model asphalt systems
using molecular simulation. Journal of Chemical Physics,
127, 194502.
Zhang, L. and Greenfield, M.L., 2008. Effects of polymer
modification on properties and microstructure of model
asphalt systems. Energy Fuels, 22, 3363 3375.
Zhang, L.Q. and Greenfield, M.L., 2010. Rotational relaxation
times of individual compounds within simulations of
molecular asphalt models. Journal of Chemical Physics,
132, 184502.
Zhang, S.F., et al., 2010. Aggregate structure in heavy crude oil:
using a dissipative particle dynamics based mesoscale
platform. Energy Fuels, 24, 4312 4326.

Vous aimerez peut-être aussi