Fuel 140 (2015) 34–43

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Relation between bitumen chemistry and performance

Per Redelius a,⇑, Hilde Soenen b
a
Nynas AB, SE-149 82 Nynäshamn, Sweden
b
Nynas NV, Noorderlaan 183, BE2030 Antwerp, Belgium

h i g h l i g h t s

 Interactions between hydrocarbon molecules is very important for the physical properties of the product.
 The strength of some of the interactions may be estimated by solubility parameters.
 Large aromatic systems contribute with important interactions not determined in the solubility parameters.
 The viscosity of a heavy petroleum product may be estimated from molecular weight and aromaticity.
 The behaviour of the fractions ‘‘asphaltenes’’ may be explained in terms of molecular interactions.

a r t i c l e i n f o a b s t r a c t

Article history: The relation between the chemical composition of bitumen and the physical properties, such as viscosity
Received 23 April 2014 or viscoelasticity is very important for the use of bitumen as an advanced construction material. In view
Received in revised form 30 July 2014 of the limited understanding of the exact molecular structure of the millions of different molecules
Accepted 11 September 2014
constituting bitumen, many different proposals for the structure of bitumen have been published over
Available online 1 October 2014
the last 100 years. In this paper we propose that interactions between the molecules are the main deter-
minants for the physical properties of bitumen. We discuss different types of interactions, typical for
Keywords:
hydrocarbons, such as dispersive London interactions, polar interactions and pi–pi interactions, and
Bitumen
Interactions
how the strength of the interactions may be estimated by different methods. We also discuss one class
Chemistry of molecules in bitumen, defined by non-solubility, the asphaltenes, and why they contribute strongly
Asphaltenes to the thickening effect of bitumen, but we also conclude that asphaltenes are not the only molecules
Viscosity contributing to viscosity. Finally we show how the use of molecular parameters like molecular weight
and aromaticity can be used to give a good estimate of the viscosity of heavy hydrocarbons of different
composition.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Ever since the production of bitumen as a distillation residue

Bitumen is a black adhesive material produced from crude oil.
The most common use is as the binder in asphalt. But it is used
for many other purposes, including roofing applications and other
types of water tightening as well as many other applications. Long
before crude oil started to be produced (1859) bitumen was found
in nature as ‘‘natural asphalt’’ in many places around the world.
The very extensive use of asphalt developed in parallel to the
enhanced demand of good roads to meet the growing use of cars
in the society.
⇑ Corresponding author. Tel.: +46 8 520 650 00, Mobile: +46 70 514 29 72; fax:
+46 8 520 176 38.
E-mail addresses: per.redelius@nynas.com (P. Redelius), hilde.soenen@nynas.
com (H. Soenen).
URL: http://www.nynas.com (P. Redelius).
from petroleum, there have been attempts to clarify the true chem-
ical structure of bitumen. It was very soon stated that bitumen
consists mainly of hydrocarbons with small amounts of heteroat-
oms including sulphur, nitrogen and oxygen and traces of metals
such as vanadium, nickel and iron. In the early days (1924)
Nellensteyn speculated that bitumen was a colloidal dispersion
of carbon particles dispersed in oil [1]. The chemistry of bitumen
has sometimes been considered to be a mystery and ‘‘very compli-
cated’’. When consulting literature, among relevant and high class
investigations, also unsupported speculations and misinterpreta-
tions have been published. It is not always easy for the newcomer
to navigate between contradictory claims and statements.
It has sometimes been assumed that once we know the chem-
istry of bitumen we will be able to predict its performance as a
construction material, as well as specify the properties of ‘‘good’’

http://dx.doi.org/10.1016/j.fuel.2014.09.044
0016-2361/Ó 2014 Elsevier Ltd. All rights reserved.
 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43 35

bitumen. However, the knowledge about the molecules is only of

limited help for understanding bitumen, and advanced modern
analytical techniques always provide average results which are
not easily translated to physical properties or performance proper-
ties. Instead separation techniques have been used to divide bitu-
men into fractions. Many techniques, using different properties
of the molecules for separations have been used. These fractions
have then been correlated to physical properties sometimes rather
successful but more commonly without finding any general corre-
lation. The fractions are generally very heterogeneous and are only
defined by the method of separation. There is no ‘‘gap’’ in chemical
properties between the fractions. A common misunderstanding is
to consider the fractions as ‘‘the components in bitumen’’ and
claim that bitumen consists of a mixture of three, four or five types
of compounds instead of a continuum of molecules. The continuum
consists of relatively large hydrocarbons with different size, polar-
ity and aromaticity. The smallest size of the hydrocarbons is
defined by the processing conditions of the crude oils (cut point
in the vacuum distillation tower) and the largest size is defined
by the crude oil.
Fig. 1. ‘‘Typical’’ bitumen molecule.
In this paper we will give an explanation how the chemistry of
bitumen determines the physical properties and performance
The smallest size of the bitumen molecules varies depending on
properties of the bitumen as a construction material in asphalt.
type of crude oil and grade of bitumen. It is determined by the
cut point at the distillation of the crude oil. It is possible to
estimate that the smallest size of the molecules is around 20 car-
2. Bitumen at a molecular level
bons and goes up to the largest size in the residue. This is still
under discussion among researchers and there are a lot of different
Today the chemistry of the molecules constituting bitumen can
opinions. Based on the latest findings [2–5] it seems unlikely that
be considered to be fairly well known although more than 95% of
there are any significant quantities of molecules larger than about
the molecules have never been isolated or identified and can thus
1500 g/mol (about 110 carbons).
be said to be unknown to mankind. How can we then claim that
The most typical chemical property of bitumen is its heteroge-
the chemistry is well known? One pragmatic way of estimating
neity, particularly the unique combination of the large number of
the chemistry of bitumen is to study the composition of the lighter
similar but still different molecules, which makes it meaningless
fractions from crude oil, the distillates. The light distillates with the
to describe bitumen with one or a few ‘‘typical’’ average molecules,
lowest boiling points are well known in detail, while the composi-
but the whole range of molecules must be taken into consideration
tion of the heavier distillates is only known in more general terms.
to make realistic chemical description of bitumen.
The fact that bitumen is a residue does not mean that the principle
structure of the molecules changes during the transition from a
distillate to a residue. It is just that the molecules become larger 3. Separations
and thus have a higher boiling point. Some contributions to the
new chemistry are collected in Ref. [2] and further detailed Traditional analytical techniques like ultraviolet spectroscopy
information about the chemistry of the molecules can be found (UV), infrared spectroscopy IR, nuclear magnetic resonance NMR,
in a series of references by McKenna et al. [3–5]. All these results or mass spectroscopy MS have given substantial information about
confirm the theory which was already presented in the 1980s by the average chemical composition of bitumen. Most bitumens give
Boduszynski et al. [6] but which attracted limited attention. more or less identical spectra and no general correlations between
All molecules in bitumen are hydrocarbons with small amounts physical properties and any particular functional group as identi-
of sulphur, nitrogen and oxygen and traces of metals like vanadium fied by the techniques above are known. The only technique which
and nickel. The hydrocarbons may be described by a number of has been frequently used is IR, which permits identification of
core structures consisting of polyaromatic structures containing carbonyls and sulphoxides formed as a result of ageing. Instead dif-
different number of fused rings, saturated polycyclic structures ferent techniques to separate bitumen into fractions have been
also with different number of rings and combinations of these. developed. Over the years many different techniques based on dif-
All these core structures contain saturated hydrocarbon side chains ferent separation principles have been tried. Some examples are:
of different chain length and different substitution patterns. The
number of possible isomers is almost unlimited. That is the reason  Reactivity with sulphuric Rostler Sternberg [7]
why bitumen consists of millions of different molecules, almost acid
none of them in sufficiently large quantities that it is possible to  Solubility Hoiberg [8]
isolate and characterize them. So even if the structure is known  Size (SHRP SEC separation) SHRP report [9]
in principle, the exact structure is unknown. Based on average ana-  Charge (SHRP IEC SHRP report [10]
lytical data it is possible to suggest an average ‘‘bitumen molecule’’ separation)
as illustrated in Fig. 1. The molecule has never been proven in  Polarity (SARA) Several different procedures
crude oil or bitumen, but if it exists it will fall in the bitumen frac- [11]
tion. The size is large enough to give a boiling point above the cut
point for the most heavy distillate, it contains sulphur in thiophe-
nic structure, which is the most common structure for sulphur Today the most popular technique is the separation into satu-
based heterocyclic molecules, and it contain aromatics and satu- rates, aromatics, resins and asphaltenes (SARA). There are several
rates in approximately the average amount known for bitumen. different techniques claiming to do the same separations, for
36 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43
example ASTM D2007, ASTM D4127 and a technique using thin
layer chromatography and a flame ionisation detector (TLC–FID)
also referred to as IatroscanÒ [12]. It is clear that the different
methods give fractions with different chemical composition
although the fractions are sometimes given the same names. One
example is the fraction in a TLC–FID experiment which does not
move with any of the solvents used for development (i.e. stays at
the application spot) which usually is reported as ‘‘asphaltenes’’.
In IP 143 the asphaltenes are defined as the fraction of bitumen
which is insoluble in n-heptane. It is very obvious that these
fractions are not identical [12].
The SARA separations have caused some confusion, since it has
sometimes been perceived as if bitumen consists of four types of
components, saturates, aromatics, resins and asphaltenes. Many
textbooks give descriptions of each type of fraction and the
particular properties of each fraction. It is however important to
realise that the fractions are empirical slices of a continuum of
molecules. Each fraction is very heterogeneous and consists of a
range of different molecules which only satisfy the condition for
the separation. This also explains the difficulty to have a good
reproducibility. When the condition is changed only a little, a part
of the molecules fall into the neighbouring category. A better way
of interpreting the SARA fractions is to see them as a distribution of
polarity of the molecules in bitumen Fig. 2, although strictly speak-
ing, asphaltenes do not fit in a polarity scale since they are defined
by solubility and not by polarity.
4. Molecular interactions
The most important factor for understanding the relation
between chemistry of bitumen and physical properties is the
molecular interactions. There are many different types of molecu-
lar interactions but for the purpose of bitumen which mainly con-
sists of non-polar hydrocarbons the dominating interactions are
the London dispersive interactions [24]. Also contributing are polar
interactions and hydrogen bonding interactions which are due to
the content of the more electronegative elements nitrogen and
oxygen. There are also pi–pi interactions which for non-polar mol-
ecules have been considered as a special case of dispersive
interactions.
5. Dispersive interactions
The dispersive interactions are also called London forces. The
London forces are due to temporary dipoles caused by fluctuations
in the electrons surrounding a molecule. The fluctuations will act
as an attraction in all directions between the molecules, which will
influence the physical properties by for example increased boiling
point, increased viscosity, increased heat of vaporisation, etc.
Although dispersive interactions are considered to be a very weak
interaction between molecules, it is certainly the most important
Fig. 2. SARA fractions as a distribution of polarity in bitumen.
for non-polar hydrocarbons and is also the most important interac-
tion for the physical properties of bitumen.
Dispersive interactions can be derived from electronic polariz-
ability and refractivity using the following equation [18]:
   2 
3 M n 1
a¼ ð1Þ
4pNA d n2 þ 2
where NA is Avogadro’s number, M is molecular weight, d is density
and n is the refractive index. The equation is primarily for estimat-
ing the polarizability of pure compounds but it could also be used to
estimate dispersive interactions in hydrocarbon mixes such as
crude oils or bitumen [17]. The same data can be used for estimat-
ing transport properties including viscosity, diffusivity and thermal
conductivity [18]. The polarizability of a number of model
hydrocarbons, aromatic and non-aromatic was calculated using
the algorithm in ChemSketchÒ (Table 1).
A straight correlation with number of carbons and thus
molecular weight was found (Fig. 3). The same can be seen when
calculating molar refractivity [17]. Eq. (1) does however not take
different shapes or pi–pi interactions of the molecules into consid-
eration. This might influence the true dispersive interaction since
the strength of the interactions is due to the surface area of the
molecules. There is also no particular contribution from polyaro-
matic structures other than by refractive index and density in the
calculation, although large polyaromatic systems are expected to
give more contribution to dispersive interactions than molecular
weight only. Whether this extra contribution is due to the more
elongated or flattened shape of polyaromatic molecules or a true
contribution from the p-electrons will be discussed below.
6. Polar and hydrogen bonding
The polar and hydrogen bonding interactions are due to an
imbalance of the electron distribution in a molecule where atoms
with different electronegativity are connected by covalent bonds.
Such molecules are also called permanent dipoles. The polar and
hydrogen interactions are considered to be stronger than the
dispersive interactions which are true for small molecules. For
large hydrocarbons, with few polar groups, like the molecules in
bitumen the dispersive interactions are dominating. Elemental
analysis of bitumen has revealed that the amount of electronega-
tive elements nitrogen (0.3–0.8%) and oxygen (0.5–1.1%) are rela-
tively small in bitumen. Since it is known that the nitrogen and
oxygen are relatively evenly distributed among the molecules in
bitumen, we may conclude that no strong polar molecules exist.
It is likely that the dispersive interaction dominates in bitumen,
but polar interactions are also present and have been shown to cor-
relate with the elastic part of the complex modulus of bitumen
[20]. Bitumen is known to contain much more sulphur than nitro-
gen and oxygen. Sulphur has electronegativity very close to carbon
and hydrogen and does not contribute substantially to polarity.
7. Pi–pi interactions
Another type of interaction which could be expected between
bitumen molecules are p interactions which may appear between
delocalised pi electrons in molecules [21]. One kind of p–p interac-
tion is the charge transfer complexes, which require that at least
one of two interacting molecules have strong electron donating
or electron attracting substituents on the aromatic core to give a
strong difference between the two overlapping electron clouds.
This is not typical for the polyaromatic molecules in bitumen,
although some porphyrin molecules or metal organic molecules
present in bitumen may be able to form such complexes. The mol-
ecules with one or two fused rings seem not to provide any ‘‘extra’’
 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43 37

Table 1
List of model substances used to estimate polarizability. Toluene and heptane may not appear in bitumen but the other molecules might.

Molecular formula Mol weight (g/mol) Polarizability (1024 cm3) Structure

C7H8 92 12.3

C7H16 100 13.7

C23H24 300 39.5

C31H64 437 57.7

C32H36OS 469 61.0

C32H44OS 477 60.0

C40H50O 547 72.9

C49H58S 679 91.3

 The interactions listed above do not in isolation, but there are

always combinations of two or more types of interactions
between bitumen molecules.
 There is a strong effect from the solvents (or the other mole-
cules) on the strength of the interactions.

The many different types of interactions and the uncertainty of

Fig. 3. Polarizability for different size of molecules.
contribution from pi–pi interactions over the expected dispersive
interactions. For larger aromatic systems with four or more fused
rings the contribution is significant, and increases with the number
of rings [25]. This is also seen in a higher boiling point for polyar-
omatic hydrocarbons compared to saturated hydrocarbons of the
same size. If an aromatic ring is substituted with a more electro-
negative element (nitrogen or oxygen) the pi–pi interaction is
enhanced. A large aromatic system can also interact with highly
polar external molecules [21] which might have given rise to the
suggestions that polar interactions are of major importance for
bitumen, for example in the SHRP microstructural model.
8. Quantification of chemical interactions
To be able to predict physical properties in bitumen we need to
estimate the molecular interactions. There are a number of factors
which make it difficult:
 The exact structure of all the molecules in bitumen is not
known.
the exact structure of the molecules in bitumen make calculations
of molecular interactions very uncertain. A practical empirical
approach to estimate the molecular interactions is to use Hansen
solubility parameters (HSP) [13,14]. The HSP gives a value on three
of the most important interactions for hydrocarbons, the dispersive
interactions (dD), polar interactions (dP) and hydrogen bonding
(dH). It was shown that the strongest interactions in bitumen are
provided by the dispersive interactions and less, but still impor-
tant, by the polar interactions and the hydrogen bonding.
It was noted that the pi–pi interaction is not part of the HSP. It is
not obvious that it should be, because in small aromatic systems
the pi–pi interactions are considered as part of dispersive interac-
tions. It could also be that pi–pi interactions appear as hydrogen
bonding interactions in the HSP, which is indicated in Fig. 4.
It has also been noticed that hydrogen bonding interactions of
bitumen seem to increase as a result of ageing. Typical solubility
parameters for bitumen, some of its fractions and some solvents
are listed in Table 2. The numbers in the table are averages of
the molecules in bitumen and its fractions. The variation in solubil-
ity parameter in the bitumen fractions is better illustrated by the
solubility spheres, which are made from the HSP of all the mole-
cules in bitumen (Fig. 5). From Table 2 it is obvious that the differ-
ences between bitumen, asphaltenes (precipitated by n-heptane)
and maltenes are relatively small. From Ref. [26] we may estimate
the solubility parameter distance (H) between two materials,
expressed as the ratio of cohesion energy densities:
38 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43

Fig. 4. Some examples of delta on saturated and aromatic rings.

determined by a combination of molecular weights and molecular

Table 2 interactions. The true molecular weight distribution of bitumen is
Comparison of HSP of asphaltenes and maltenes compared to the original Venezuelan very difficult to determine. Most common methods for molecular
bitumen. Also included is the HSP of n-heptane, the precipitant. weight determination make use of solutions, and the solvents are
Sample dD MPa0.5 dP MPa0.5 dH MPa0.5 Radius generally not good enough for the big range of different molecules
Bitumen 18.4 3.9 3.6 5.8 found in bitumen resulting in associations giving unrealistic high
Asphaltenes 19.6 3.4 4.4 5.3 Mw. This is true for Vapour Phase Osmometry (VPO) and Gel Per-
Maltenes 17.7 5.8 2.5 6.7 meation Chromatography (GPC). The most correct molecular
n-Heptane 15.3 0 0 weights are likely found using Field Ionisation Mass Spectroscopy
Toluene 18.0 1.4 2.0
(FIMS) [27] or other Mass Spectroscopic techniques [3]. In this
work we have used GPC with a Refractive Index (RI) detector pre-
suming that the relative molecular weights are correct. A diode
array UV/Vis detector was used to estimate the relative amount
of polyaromatic molecules as the molecules with absorption at
the highest wavelength.
Asphaltenes
10. Boiling point and viscosity
Maltenes
Another indirect estimate of molecular weight is to compare
boiling points which are roughly related to molecular size, as long
as we compare saturated hydrocarbons. But from the estimations
of dispersive interactions from Eq. (1) and Fig. 3 it is evident that
the dispersive interactions increase linearly with molecular sizes,
which also influence the boiling points. An attempt to make a cor-
relation for crude oils is demonstrated in Fig. 6 [19], two crude oils
and their residues are compared by High Temperature Gas Chro-
matography (HTGC). The Boscan residue with cut point 400 °C is
bitumen with a viscosity of 574 Pa s at 60 °C, while Russian B res-
idue with cut point 564 °C is a soft residue with a viscosity of
68.6 Pa s at 60 °C. The cut point is the temperature at which all vol-
Fig. 5. Solubility spheres of the maltene fraction and the asphaltene fraction. atile hydrocarbons have been distilled off. The residue contains
hydrocarbons with boiling points from the cut point and up. It is
H ¼ ðRAÞ2 =R2M
" # evident from Fig. 6 that the more viscous Boscan residue has a
ðdP2  dP1 Þ2 ðdH2  dH1 Þ2 . 2
2 much wider range of molecules with different sizes, from rather
¼ ðdD2  dD1 Þ þ þ RM ð2Þ small with a boiling point of about 400 °C up to a substantial frac-
4 4
tion which did not pass through the GC column at 740 °C and thus
where RA = is the HSP difference between a material and solvent have a higher boiling point and corresponding larger size. It
and RM = is the radius of the solubility sphere. The ratios of cohesion appears as the residue from Russian B consist of molecules with
energy densities for some different combinations are listed in larger size and stronger interactions and thus are expected to give
Table 3. higher viscosity compared to Boscan residue with relatively lower
When comparing the numbers in Table 3, it should be noticed boiling molecules of smaller size or weaker interactions. But we
that the asphaltenes and maltenes are not compounds, but very find the opposite. The reason is to be found by the very heavy mol-
heterogeneous fractions, while Eq. (2) was developed for pure ecules or strongly interacting molecules with a boiling point higher
compounds. But it is evident that the asphaltene fraction defined than 740 °C which are not detected by the HTGC. These molecules
as insoluble in n-heptane cannot at the same time be insoluble are strongly contributing to the viscosity, likely by very strong
in the maltenes due to the small difference in cohesion energy dif- interactions. From Fig. 6 it is evident that Russian B residues almost
ference indicating good solubility. lack molecules with higher boiling point than 740 °C while Boscan
contains a relatively large quantity although it is not possible to
quantify the amounts from Fig. 6.
9. Molecular weight

So far we have discussed the interactions between the 11. Molecular interactions and asphaltenes
molecules in bitumen. Another important property of the mole-
cules is the molecular weight or more exactly the molecular weight One obvious manifestation of molecular interactions is the pre-
distribution. We may speculate that the viscosity of bitumen is cipitation of a fraction of bitumen or crude oil named asphaltenes,
 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43 39

Russian B: crude oil

Russian B: residue

Mass 20% (BP < 560ºC)

C17 C25 C37

Boscan crude oil Boscan residue

Mass 20% (BP < 400ºC)

C17 C25 C27

Fig. 6. High temperature simulated distillation of two crude oils (left) one light (Russian B) and one heavy (Boscan) and residues after distillation of the two crude oils (right).

Table 3 The main reason for the precipitation of asphaltenes in n-hep-

Cohesion energy difference for combination of materials.
Material 1 Material 2
Asphaltenes n-Heptane
Maltenes n-Heptane
Asphaltenes Maltenes
which are defined as the fraction precipitating upon dilution with
n-alkane, usually n-heptane (IP 143/04). The asphaltene fraction is
a very heterogenic fraction consisting of molecules with a molecu-
lar weight ranging from about 400 amu up to about 1500 amu and
the most characteristic property compared with the maltenes is
the higher aromaticity and higher average molecular weight. The
fraction consists of molecules with polycyclic aromatic cores with
three to eight fused rings substituted with saturated hydrocarbon
chains or rings. It is known that some asphaltene fractions contain
saturated long chain n-alkanes, which sometimes have been con-
sidered ‘‘impurities’’. However n-alkanes with over about 40 car-
bons are insoluble in n-heptane and are expected to precipitate
as asphaltenes if they are present in the sample [22].
tane is the weak molecular interactions with n-heptane, which
H makes the fraction in bitumen (or any oil) having the strongest
3.7 interactions to precipitate due to its poor solubility. The concentra-
1.4 tion of n-alkane needs to be enough to neutralise the solubilising
0.5 effect of the other molecules (maltenes) and create a ‘‘chemical
gap’’ between the precipitate and the solvent. This critical concen-
tration of n-heptane varies between different bitumen and crude
oils. There has been much of speculation about the nature of the
interactions causing the asphaltenes to precipitate. Some claim
that the precipitation is due to polar interactions, while others
have concluded that the polarity of the asphaltene molecules is
too low to account for the strong associations upon precipitation.
This may be estimated based on the content of nitrogen and oxy-
gen in the asphaltene fraction. The amount is about 3% in total
which is too low to make the large hydrocarbon molecules suffi-
ciently polar to result in a precipitation. Solubility parameters of
asphaltenes indicate that dispersive interactions are more impor-
tant than the polar interactions. As we noticed in Fig. 3 the disper-
sive interactions become stronger for larger molecules, which also
result in lower solubility in n-heptane. The fact that asphaltenes
are always more aromatic than the maltenes suggests that pi-pi
40 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43

interactions may have a significant contribution to interactions

Asphaltene content %
25
and thus to insolubility in n-alkanes. We may conclude that many 20
types of interactions are contributing to the insolubility of the
15
asphaltene fraction in n-heptane. The determination of the
solubility parameters shows that the dispersive interactions are 10
the strongest, while the polar and hydrogen bonding interactions 5
are less important. The contribution from pi-pi interactions is not 0
easy quantified. It is however known that larger polyaromatic sys- 0 200000 400000 600000
tems, as in asphaltenes, have a significant contribution from pi–pi Viscosity Pas at 25ºC
interactions [25] so we estimate that aromaticity of the asphalt-
enes are important for the strength of the molecular interactions. Fig. 7. Asphaltene content versus viscosity on SHRP bitumens. Data from Ref. [15].

We may speculate that the same interactions which make the

asphaltenes insoluble in n-alkanes are also the most important
contributors to the high viscosity of bitumen. To test the hypothe-
sis we may use data from the extensive analysis of bitumens 3
collected for the Strategic Highway Research Program (SHRP)
2,5
sponsored by Federal Highway Administration (USA) and con-

Log visc rao

ducted between the years 1988 and 1992. It should be noticed that 2
the binders were collected as ‘‘commercial’’ bitumens used for 1,5
highway construction and are thus limited in variation of proper- 1
ties to fulfil the US specification at that time. The set of samples
0,5
collected by the authors of this paper [30] and which are used
for construction of Fig. 9 contains a wider variation of samples 0
0 10 20 30
including some not suitable for road construction.
n-heptane asphaltenes %
In Ref. [15], Pauli and Branthaver, examined the thickening
effect of asphaltenes in a set of bitumens from the SHRP program. Fig. 8. Log viscosity ratio bitumen/maltenes versus content of n-heptane
The bitumens are produced from different crude oils using asphaltenes.
different techniques and are sometimes mixtures. If we compare
a selection of data from Refs. [15,16] listed in Table 4 we may
notice some relations.
By calculating the ratio between the viscosity of the bitumens
to the viscosity of the maltenes it is possible to estimate the thick-
ening effect of the asphaltenes, which is considerable and varies
with different bitumens. There is however no relation between
the asphaltene content and the viscosity of the bitumen, Fig. 7. This
indicates that the variation of the thickening effect is not only a
concentration effect but due to different composition of asphalt-
enes from different bitumens. If, however, we consider the thicken-
ing effect of the asphaltenes expressed as the ratio between
viscosity of the bitumen and the maltenes, Fig. 8, we notice a fairly
good correlation of R2 = 0,72. This figure demonstrates that it is not
only the asphaltenes which are responsible for the thickening
effect but also the maltenes. This becomes obvious if we consider
the variation in viscosity of the maltenes which is a factor of 300
while the bitumens vary with a factor of 10. The function of the Fig. 9. Relation between molecular parameters, size and aromaticity, with viscosity
maltenes is twofold; first the solubility power of the maltenes at 90 °C.

Table 4
Viscosity of bitumens and maltenes.

Bitumen Grade Pen 25 °C Bitumen Maltenes Ratio n-Heptane asphaltenes

Pas 25 °C 0.1 rad/s Pas 25 °C 0.1 rad/s Visc Bit/Visc Mal 25 °C % w/w
AAA-1 150/200 160 37,020 804 46 15.8
AAB-1 AC-10 98 118,600 3020 39 17.3
AAC-1 AC-8 133 90,100 10,520 9 9.9
AAD-1 AR-4000 135 57,280 329 174 20.2
AAE-1 60/70 73 184,500 530 348 22.9
AAF-1 AC-20 55 381,700 25,000 15 13.4
AAG-1 AR-4000 53 248,900 82,460 3 5
AAH-1 AC-10 95 159,500 1804 88 15.9
AAJ-1 AC-20 67 201,100 15,251 13 10.6
AAK-1 AC-30 70 251,700 1620 155 20.1
AAM-1 AC-20 64 364,400 101,800 4 3.7
AAN-1 85/100 90 122,700 3720 33 15.7
AAS-1 AC-20 64 164,100 734 224 18.4
AAV-1 AC-5 121 46,700 5334 9 9.7
AAW-1 AC-20 64 288,700 8962 32 17.9
AAX-1 AC-20 51 275,900 24,057 12 12
 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43
influences the strength of the interactions of the asphaltenes and
second, the maltenes may also contain molecules with strong
interactions which contribute to the thickening of the bitumen
but not so strong that they are insoluble in n-heptane. Since we
have a continuum of molecules the borderline of solubility is
heavily dependent on type of precipitant. This is the reason for
the difference in asphaltenes precipitated by different n-alkanes.
At this point it is important to remember that asphaltenes is a
heterogenic, empirical fraction, determined by its insolubility in
n-heptane, and not expected to give a correct picture of all interac-
tions contributing to the viscosity. It is true that removing asphalt-
enes always decreases the viscosity of bitumen but the effect varies
substantially with different crude oils.
12. Ageing of bitumen
When bitumen is exposed to air it hardens. There bigamy be
significant differences in the extent of hardening between different
types of bitumen. The increased stiffness is likely due to increased
interactions between the molecules. It is not yet fully understood
which type of interactions are the most important. There are basi-
cally three different hypotheses:
 During reaction with oxygen there is a condensation of bitumen
molecules resulting in an increase in size of the molecules.
 The reaction with oxygen increases the polarity of the bitumen
molecules causing stronger polar interactions.
 The ageing increases the aromaticity which increases pi–pi
interactions and dispersive interactions.
There is experimental evidence for all three hypotheses above.
GPC experiments have proven that the molecular weight increases
after ageing. However the GPC experiments are always performed
with the bitumen sample dissolved in a solvent, commonly tetra-
hydrofuran (THF), to lower the viscosity enough to let the sample
flow easily through the GPC columns. THF might not be a perfect
solvent for all molecules in bitumen, and we might expect some
molecules to make associations strong enough to give an apparent
higher molecular weight in the GPC column.
The formation of carbonyls and sulphoxides on oxidative ageing
increases the polarity of some bitumen molecules. The monitoring
of carbonyls and sulphoxides from IR spectrograms has become
one of the most common techniques to monitor ageing of bitumen.
It has been speculated that the increase of the polarity of the bitu-
men is the reason for the hardening of the bitumen. Although the
increase in carbonyls is very evident, it is rather small, and it is
unlikely that the increased polarity can fully explain the very
strong hardening.
The increase in aromaticity is rather well known but has not
attracted so much attention as the formation of carbonyls. One of
the reasons may be that it is a little more difficult to monitor
increases in aromaticity and it is not so evident how increased aro-
maticity contributes to the hardening.
13. Aromaticity of bitumen
We have now discussed the importance of dispersive
interactions for the viscosity of bitumen. From Eq. (1) we learn that
molecular size, refractive index and density are part of the estima-
tion of polarizability which is a measure of strength of dispersive
interactions. The refractive index and density is also heavily depen-
dent on aromaticity. To further understand the relation between
chemistry of bitumen and its physical properties it is important
to be able to measure aromaticity with a good precision. In this
work we have used three techniques:
41
 IR using the peak around 1600 cm2 indicating C@C double
bonding, which in case of bitumen is always aromatic.
 GPC-UV/VIS where the absorption above a certain wavelength is
determined. Only larger aromatic systems give absorptions at
longer wavelength.
 NMR using combination of C13NMR and H1 NMR.
The first technique (IR) is expected to give an estimate of the
total aromatic content, while the GPC UV/VIS gives a relative esti-
mate of the content of larger aromatic systems. For NMR we still
have a limited number of analyses. A remarkable observation is
that the NMR diagrams from different bitumen and particularly
unaged and aged bitumen are very similar. It is only after integra-
tion of some of the peaks that we may calculate differences.
14. Prediction of viscosity
The discussion above indicates that the most important interac-
tions for the viscosity and the physical properties of bitumen are
London Dispersive interactions, which may be estimated by
molecular weight, and aromatic p–p interactions which may be
estimated from amount of large polycyclic aromatic systems. For
testing this hypothesis we selected a set of 27 different binders
covering many different types of binders; distilled from single
crude as well as blends, blown bitumens and products produced
by special refining processes such as cracking and precipitation.
More data on the binders are presented in Ref. [30]. The samples
were tested with GPC to get a relative value of the molecular
weight, and with UV/VIS absorbance to get an estimate of the
amount of larger aromatic systems. Details of the analyses are pre-
sented in Ref. [30]. The viscosity was taken at 90 °C to ensure that
we had a fairly Newtonian liquid. The best fit was achieved with
log(Mw) + log(UV abs 380 nm)/0.35 (Fig. 9).
It is to be noticed that none of the binders, even when processed
in such a way that the composition of the binder are very different
from ‘‘normal’’ distilled binders, do show up as outliers. This sug-
gests that we have a reasonable correlation between chemistry
and physical properties. The point with a very high viscosity may
contribute too much to the correlation. If the correlation is calcu-
lated without this particular point R2 = 0.78. In the set of 27 binders
there are 5 binders from the same crude source. Three of them are
distilled and two are oxidised. If a correlation with only these 5
binders are plotted a correlation of R2 = 0.92 is calculated, Fig. 10.
The correlation R2 = 0.84 for the full set is not very high, but consid-
ering the very diverse selection of samples and the simple test
methods it should be considered very good. The main error is prob-
ably due to the GPC method which does not give correct molecular
Fig. 10. Relation between size, aromaticity and viscosity for 5 selected binders.
42 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43
weights, but more relative values. The correlation is also based on
averages, while a complete description needs to take the full range
of different molecules into consideration.
If the correlation is tested at lower temperatures, it becomes
less good. Obviously other factors which influence viscosity
appear. This could be an increased influence from polar interac-
tions, or it could be the effect of wax which will start to crystallise
once the temperature goes below approximately 60 °C.
15. Phase separation in bitumen
In the discussions above we have identified interactions
between bitumen molecules as one of the most important reasons
for the high viscosity and stiffness of bitumen binders. From
Differential Scanning Chromatography (DSC) we know that most
bitumen show a kind of phase transition on cooling from high tem-
perature (100 °C). The first transition is usually visible at around
40–60 °C and has been considered as crystallisation of paraffin
wax. The presence of wax crystals in bitumen has been confirmed
by several microscopy techniques [28]. Also on the surface of bitu-
men, wax in different shapes has been detected with Atomic Force
Microscopy (AFM). The shape and size of the crystals varies with
type of bitumen as well as temperature history. The wax is consid-
ered to consist mainly of long chain n-alkanes which form so
strong molecular interactions at low temperature that they crystal-
lize inside the bitumen. As with all types of molecules in bitumen,
the wax consists of a range of different molecules [29]. A more
careful investigation has shown that after the first initial crystalli-
sation peak, there is a range of molecules which either crystallize
or go through a phase separation. At low temperature the mobility
of the molecules becomes very low and it can take very long time,
several hours or days, to reach equilibrium. Thus not all transfor-
mations can be detected in a normal DSC scan with a cooling rate
of 10 °C per minute. These effects have been detected by using DSC,
but stopping the scan at certain temperature to wait for equilib-
rium. The melting or solubilisation of the new phase is then seen
in the subsequent heating scan.
It should be noticed that the model above does not give any
support for the idea that the asphaltene fraction as such is present
as particles or as aggregates in a colloidal solution. It is very unli-
kely that a highly heterogeneous phase such as the asphaltenes
should behave as a pure substance. The big difference in solubility
parameter between n-heptane, defining asphaltenes, and maltenes
makes it certain that the same fraction cannot be insoluble in the
maltene phase. We can however not rule out the possibility that
some of the molecules in the asphaltene fraction may be insoluble
in the maltenes and actually be dispersed rather than dissolved.
One group of substances that could be expected to form associ-
ations are the porphyrins containing metals like vanadium or
nickel. They might have the ability to form so called charge transfer
complexes. The concentration of these complexes is however
expected to be too low to influence the physical properties to a
higher degree.
16. Conclusion
The physical characteristics and the performance of bitumen
are a result of the molecular interactions taking place between
the components present in bitumen. Different types of interactions
are present, for example dispersive, polar, hydrogen bonding or pi–
pi interactions.
The interactions cannot be measured directly in the complex
mixture constituting bitumen.
One empirical method to determine dispersive, polar and
hydrogen bonding interactions is the determination of Hansen
Solubility parameters which may be visualised in a 3D diagram
as a solubility spheroid.
It is obvious that larger molecules, as well as large polyaromatic
systems give higher interactions and thus a higher viscosity to
bitumen.
It is possible to estimate the viscosity of bitumen based on
molecular weight and aromaticity.
Asphaltenes are a very heterogenic fraction defined by their
solubility. It contains the molecules in bitumen with strongest
dispersive interactions as well as the strongest pi–pi interactions,
but not giving the full range of properties to estimate the physical
properties of the bitumen. It is important for the properties and the
interpretation of test results to consider bitumen as a continuous
range of molecules with respect to size, polarity and aromaticity.
In most bitumen a two phase system appears at lower temper-
atures below 50–60 °C. This phase is generally considered to be
crystallizing waxes, but may consist of a variety of different types
of molecules. This crystallization will also influence the physical
properties of bitumen.
References
[1] Nellensteyn FJ. The constitution of asphalt. Inst Petrol Technol
1924;10:311–25.
[2] Mullins OC, Sheu EY, Hammami A, Marshall AG. Asphaltenes, heavy oils, and
petroleomics. Springer; 2007.
[3] McKenna AM, Blakney GT, Xian F, Glaser PB, Rodgers RP, Marshall AG. Heavy
petroleum composition. 2. Progression of the Boduszynski model to the limit of
distillation by ultrahigh resolution FT-ICR mass spectrometry. Energy Fuels
2010;24:2939–46.
[4] McKenna AM, Purcell JM, Rodgers RP, Marshall AG. Heavy petroleum
composition. 1. Exhaustive compositional analysis of Athabasca bitumen
HVGO distillates by Fourier transform ion cyclotron resonance mass
spectrometry: a definitive test of the Boduszynski model. Energy Fuels
2010;24(5):2929–38.
[5] McKenna AM, Marshall AG, Rodgers RP. Heavy petroleum composition. 4.
Asphaltene compositional space. Energy Fuels 2013;27(3):1257–67.
[6] Boduszynski MM, Altgelt KH. Composition and analysis of heavy petroleum
fractions. New York: CRC Press; 1994.
[7] Rostler FS, White RM. Composition and changes in composition of highway
asphalts 85–100 penetration grade. Proc Assoc Asphalt Paving Technol
1962;31:35–89.
[8] Hoiberg AJ, Garris Jr WE. Analytical fractionation of asphalts. Anal Chem May
1944;16:294–302.
[9] Petersen JC, Robertson RE, Branthaver PM, Duvall JJ, Kim SS, Anderson DA, et al.
Binder characterisation and evaluation. Test methods 1994;4:79–85 [SHRP-A-
370].
[10] Petersen JC, Robertson RE, Branthaver PM, Duvall JJ, Kim SS, Anderson DA, et al.
Binder characterisation and evaluation. Test methods 1994;4:65–76 [SHRP-A-
370].
[11] Corbett LW. Composition of asphalt based on generic fractionation, using
solvent deasphaltening, elution–adsorption chromatography, and densimetric
characterization. Anal Chem 1969;41(4):576–9.
[12] Ray JE, Oliver KM, Wainwright JC. The application of the Iatroscan TLC
technique to the analysis of fossil fuels. Petroanalysis 81. Institute of
Petroleum (GB); 1982. p. 361–88.
[13] Redelius PG. Solubility parameters and bitumen. Fuel 2000;79:27–35.
[14] Redelius P. Hansen solubility parameters of asphalt, bitumen, and crude oils.
In: Hansen CM, editor. Hansen solubility parameters; a user’s handbook. CRC
Press; 2007.
[15] Pauli AT, Branthaver JF. Relationships between asphaltenes, Heithhaus
compatibility parameters, and asphalt viscosity. Pet Sci Technol
1998;16(9&10):1125–47.
[16] Jones IV DR. SHRP material reference library: asphalt cements: a concise data
compilation. Washington, DC: National Research Council; 1993. SHRP-A-645.
[17] Vargas FM, Chapman WG. Application of the one-third rule in hydrocarbon
and crude systems. Fluid Phase Equilib 2010;290:103–8.
[18] Riazi MR, Roomi YA. Use of refractive index in the estimation of
thermophysical properties of hydrocarbons and petroleum mixtures. Ind Eng
Chem Res 2001;40:1975–84.
[19] Lu X, Kalman B, Redelius P. A new test method for determination of wax
content in crude oils, residues and bitumen. Fuel 2008;87:1543–51.
[20] Redelius P, Soenen H. Correlation between bitumen polarity and Rheology.
Road Mater Pavement Des 2005;6(3):385–405.
[21] Hunter CA, Sanders JKM. The nature of p–p interactions. J Am Chem Soc
1990;112:5525–34.
[22] Redelius P. Asphaltenes in bitumen, what they are and what they are not. Road
Mater Pavement Des 2009;10(1):25–43.
[24] Israelachvili JN. Intermolecular and surface forces. 3rd ed. Elsevier; 2011.
 P. Redelius, H. Soenen / Fuel 140 (2015) 34–43 43

[25] Grimme S. Do special noncovalent p–p stacking interactions really exist?
Angew Chem Int Ed 2008;47:3430–4.
[26] Hansen CM, editor. Hansen solubility parameters; a user’s handbook. CRC
Press; 2007.
[27] Boduszynski MM, McKay JF, Latham DR. Asphaltenes where are you? Proc
Assoc Asphalt Paving Technol 1980;4:123–43.
[28] Lu X, Langton M, Olofsson P, Redelius P. Wax morphology in bitumen. J Mater
Sci 2005;40:1893–900.
[29] Redelius P, Lu X, Isacsson U. Non-classical wax in bitumen. Road Mater
Pavement Des 2002;3(1):7–21.
[30] Soenen H, Redelius P. The effect of aromatic interactions on the elasticity of
bituminous binders. Rheol Acta 2014;53(9):741–54.