Download as pdf or txt
Download as pdf or txt
You are on page 1of 61

Prog. Polym. Sci.

27 (2002) 627±687
www.elsevier.com/locate/ppolysci

Rubber±®ller interactions and rheological properties


in ®lled compounds
Jean L. Leblanc*
Polymer Rheology and Processing, Universite P. & M. Curie (Paris 6), 60 rue Auber, F-94408 Vitry sur Seine Cedex, France
Received 25 July 2001; accepted 30 August 2001

Abstract
Filled rubber compounds are complex polymer systems that exhibit a number of singular ¯ow properties
markedly different from those of un®lled, molten polymers. In addition to usual hydrodynamics (or volume
fraction) effects, reinforcing ®llers such as carbon blacks or high-structure silica, impart modi®cations in ¯ow
properties whose origin is assigned to strong interactions arising between the elastomer and the ®ller particles. The
report discusses the nature of rubber±®ller interactions and their effects on rheological properties of uncured
materials. The concept of rubber±®ller mesophase is ®rst introduced in order to underline the fundamental scaling
problem that exists when attempting to relate phenomena occurring in the nanometer range to ¯ow singularities,
essentially observed in the macroscopic range. Then ¯ow singularities exhibited by ®lled rubber compounds are
brie¯y described, before interactions between ®llers and elastomers are reviewed with respect to ®ller character-
istics. Bound rubber is consequently considered, as a macroscopic result of rubber±®ller interactions, and its
importance is stressed as the obvious link towards ¯ow singularities. Eventually dimensional aspects in ®lled
rubber compounds are discussed in detail, since they offer the most likely key to understand the relationships
between bound rubber and ¯ow properties. q 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Filled rubber compounds; Complex polymer systems; Rheology; Flow properties; Rubber±®ller interactions;
Carbon black; Silica; Bound rubber; Dimensional aspects

Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628
2. From rubber±®ller mesophase to ¯ow behavior: a scaling problem . . . . . . . . . . . . . . . . . . . . . . . . . . 629
3. Singular ¯ow properties of ®lled rubber compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 630
3.1. Non-linear viscoelastic behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 631
3.2. Wall slippage effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634
3.3. Lower extrudate swell and smoother melt fracture defects with increasing ®ller content . . . . . . . 635

* Tel.: 133-1-4960-5782; fax: 133-1-4960-7066.


E-mail address: jleblanc@ccr.jussieu.fr (J.L. Leblanc).

0079-6700/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.
PII: S0 0 7 9 - 6 7 0 0 ( 0 1 ) 0 0 04 0 - 5
628 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

3.4. Anisotropic effects in ¯ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 636


4. Interactions of ®llers and elastomers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
4.1. Filler particle size and structure as primary parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 638
4.2. Filler surface characteristics as secondary particle parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 640
5. Bound rubber as macroscopic result of rubber±®ller interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
5.1. Bound rubber: a long known effect in rubber science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 645
5.2. Measurement of bound rubber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 647
5.3. Bound rubber formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651
5.4. Bound rubber and the morphology of uncured ®lled rubber compounds . . . . . . . . . . . . . . . . . . . 657
5.5. Bound rubber theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658
5.6. Storage maturation effects on bound rubber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
6. Dimensional aspects in ®lled rubber compounds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
6.1. Length scale of rubber±®ller interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664
6.2. Dimensional aspects in ®lled rubber compounds morphology . . . . . . . . . . . . . . . . . . . . . . . . . . 669
7. Bound rubber and ¯ow properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671
8. Conclusions and perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674
Appendix A. Wall slip in capillary ¯ow: revisiting the Mooney approach . . . . . . . . . . . . . . . . . . . . . . . . 675
Appendix B. Estimating macromolecular dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 680
Appendix C. Equations to calculate possible inter aggregate distances in model systems . . . . . . . . . . . . . 682
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683

1. Introduction

Most usage of elastomers would be impossible without the reinforcing character of certain ®llers, such
as carbon blacks and structured silica. With an implicit reference to tire technology, reinforcement is
usually de®ned as the ªimprovement in abrasion, tear, cutting and rupture resistance, in stiffness and
hardness of vulcanized compounds through the incorporation of ®nely divided (mineral) particlesº [1].
Quite a large variety of powdered minerals can be compounded with elastomers but not all have
reinforcing capabilities, and essentially two classes of powdered minerals have been found to offer
signi®cant reinforcing effects: carbon blacks and high-structure silica.
Reinforcement concerns ®nished rubber parts, that means vulcanized materials; but it is quite remark-
able that ¯ow properties of rubber compounds begin to signi®cantly differ from those of un®lled
materials when the ®ller has reinforcing capabilities. In addition to usual hydrodynamics (or volume
fraction) effects, reinforcing ®llers impart indeed other modi®cations in ¯ow properties whose origin is
assigned to strong interactions arising between the elastomer and the ®ller particles.
Discovered early in the 20th century (at Silverton, UK, in 1910), the reinforcing character of carbon
black has been extensively exploited in pragmatic rubber engineering and, as a matter of fact, called for
signi®cant developments in ®ller manufacturing, from the ages old, polluting, lamp black process to the
modern environmental friendly furnace technology. Carbon black reinforcement became a subject of
scienti®c interest only in the 1940s, thanks to the development of suitable investigation tools and the
growing use of synthetic rubbers in demanding applications, namely automotive and truck tires. Silica
reinforcement started to be considered in the 1980s when the bene®t of surface treatment with silanes
was recognized, with the need for lower rolling resistance tire being an important driving force in recent
research efforts.
Between carbon black and elastomers, interactions occur spontaneously but have to be (silane)
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 629

promoted in the case of silica. Today, the most sophisticated investigation techniques are used to
characterize reinforcing ®llers and to understand the very origin of rubber±®ller interactions. For
long the interest was focused on the reinforcement in vulcanized materials, and the dif®culties in
compounding and processing highly loaded systems were solved in an essentially practical manner
without paying much attention to their singular rheology. Since interactions between the various
compounding ingredients obviously take place in the early time of material preparation, i.e. during
mixing, it is quite logic to expect some links between the peculiar ¯ow properties induced by the ®ller in
the uncured state of the compound, and the reinforcement obtained after vulcanization. Rubber±®ller
interactions and reinforcing effects in vulcanisates have been regularly covered in a number of review
papers and textbooks [2,3] but the role played by such interactions in the ¯ow properties of (uncured)
compounds is a subject of relatively recent concern [4,5].
The present report will be concerned primarily with the nature of rubber±®ller interactions and their
effects on rheological properties of uncured materials. With respect to available information, more
importance will be given to rubber±carbon black systems than to rubber±silica ones. Despite certain
common ¯ow singularities, both systems appear to be very different, with little, if any, chemistry
involved in rubber±black interactions but quite signi®cant effects due to silanisation in rubber±silica
systems. Science however cannot progress if some kind of guideline is not sought to sort out what would
otherwise appears as a mere collection of singular observations. In order then to somewhat extend the
scope of the subject treated in this report, it is worth underlining that a number of systems are exhibiting
¯ow properties with common features, in such a manner that it is useful to de®ne them as `complex
polymer systems'. Whatever their chemical nature and composition, any complex polymer system has
the following characteristics:

an heterogeneous nature, i.e. there are several phases


at least one of the phase is a viscoelastic material, i.e. a polymer
there are strong interactions between phases

Such characteristics lead to the self-development of a particular morphology in the material. Like
other complex systems, uncured rubber compounds are subjected to various stimuli and constraints,
either arising from within or being imposed from without, and therefore, their morphology is likely to
evolve accordingly.
The leading idea in the present review is that the morphology in rubber compounds arises from
speci®c interactions between the elastomer and the ®ller particles, and that the resulting structure affects
the ¯ow properties. The objectives of all research work in the ®eld are to understand the origin and the
development of rubber±®ller interactions and the relationships between the resulting morphology and
the ¯ow properties. The challenge is that one is necessarily dealing with a multidisciplinary approach to
understand how microscopic phenomena, i.e. the rubber±®ller interactions, affect macroscopic (or bulk)
properties, i.e. the rheological behavior.

2. From rubber±®ller mesophase to ¯ow behavior: a scaling problem

Rheology is the science of deformation and ¯ow of matter, whose investigation tools essentially result
from continuum mechanics considerations, with such important concepts as (shear) stress, deformation
630 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

rate and viscosity, normal stress differences and elasticity, to name a few. Whatever the testing mode,
any rheometrical technique implies the simultaneous assessment of force (or pressure, or torque), rate (or
output, or rotational speed, or frequency) and temperature. Through the appropriate rheometrical equa-
tions, such basic measurements are converted into quantities of rheological interest, for instance shear or
extensional stress and rate in isothermal condition. The rheometrical equations are established by
considering the test geometry and the type of ¯ow involved, with respect to several hypotheses dealing
with the nature of the ¯uid and the boundary conditions; the ¯uid is generally assumed to be homogeneous
and incompressible, and ideal boundaries (in the continuum mechanics sense) are considered, for instance no
wall slip. With respect to the heterogeneous nature of complex polymer systems, one sees immediately that
there are some underlying problems in using blindly most standard rheometrical test methods with such
materials. So-called `anomalies' are indeed observed with respect to the behavior of pure, simple polymers,
implicitly considered as `normal', whilst the rheology of the latter is already far from meeting the ¯ow
properties of the ideal Newtonian ¯uid (i.e. a constant shear viscosity, independent of the deformation rate,
and an extensional viscosity equal to three times the shear viscosity, as stated in the Trouton law).
In what rubber compounds are concerned, it is now well established Ð as we shall see below Ð that there
are relationships between the magnitude of rubber±®ller interactions and the severity of ¯ow singularities. It
is clear that, when considering how an elastomer (macro) molecule interacts with a ®ller particle, one is
dealing with effects taking place at nanometer scale (i.e. 10 29 m), in regions of the material that can be called
the rubber±®ller mesophase. How such effects are able to in¯uence ¯ow measurements made in the cubic
centimeter range is the fundamental question that will be addressed throughout this report.
Rubber compounds are highly complex polymer systems in which various solid and liquid ingredients
are dispersed in an elastomer matrix that, by nature, exhibits a strong viscoelastic character. In a number
of technical formulations, the rubber matrix is also not simple, frequently a mixture of two or several
elastomers, which obviously further adds to complexity, since most high molecular weight polymers are
incompatible. For the sake of a sound rheological discussion, it is worth classifying the various
compounding ingredients in three categories: viscoelastic materials, powdery solids and viscous liquids.
All those ingredients have zero to limited mutual solubility or compatibility, even if they can be
thoroughly dispersed in each other. Certain major ingredients, namely the rubber(s) and the reinforcing
®ller(s), develop strong interactions that lead to the development of a structured material and the
associated peculiar ¯ow properties. In ®rst approximation, the effects of minor ingredients can be
considered as second order ones when compared with rubber±®ller interactions, whilst current experi-
ence suggests that certain chemicals in small concentration can have important rheological effects,
particularly on boundary conditions, just because they can migrate freely in rubber compounds from high
to low pressure (or strain) region. Consideration of the scaling problem cannot therefore be totally absent in
the discussion and there are two manners to approach it: either from top to bottom, when starting from the
observation of ¯ow singularities (the macroscopic view) and seeking down for explanation in terms of
rubber±particles interactions (the microscopic view), or from bottom to top, i.e. the reverse path. Whatever
the approach is, the task is very challenging and an overall understanding is still far from being achieved.

3. Singular ¯ow properties of ®lled rubber compounds

Filler rubber compounds exhibit very different rheological properties when compared with pure,
un®lled elastomers, for instance:
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 631

disappearance of any linear viscoelastic region,


wall slippage effects,
lower extrudate swell with increasing ®ller content,
smoother melt fracture defects with increasing ®ller content,
anisotropic effects in ¯ow, etc.

A detailed discussion of such properties of uncured ®lled compounds is beyond the scope of this report
and the reader may refer to several review papers [6±10]; only basic illustrations and comments will be
made hereafter.

3.1. Non-linear viscoelastic behavior

Most molten pure, high molecular weight polymers exhibit a typical shear viscosity function, at
constant temperature, with the occurrence of a so-called Newtonian plateau at low shear rate, where
the viscosity is a constant, i.e.:

h0 ˆ lim h…g_ † …1†


g_ !0

where g_ is the shear rate, h…g_ † the shear dependent viscosity and h 0 the so-called Newtonian limit
viscosity, characteristic of the material at the temperature considered and generally related to the weight
average molecular weight Mw through a scaling law, i.e.:

h0 / Mwb …2†

where the exponent is experimentally found close to 3.5. The Newtonian viscosity plateau corresponds
to the linear viscoelastic region of the polymer and its disappearance is a common feature in the ¯ow
properties of many ®lled systems, including suspensions of ®nely divided minerals in Newtonian ¯uids,
providing the ®ller level is high enough for interactions to occur between particles. With carbon black
®lled rubber compounds, this effect is particularly important since it appears at relatively low quantities
of ®ller (in the 10±15% weight range) and is further enhanced by the complexity of the black structure.
Data published by Montes et al. [11] are particularly illustrative in this respect, as shown in Fig. 1.
As can be seen in the left part of the ®gure, the Newtonian plateau is no longer observed with 20 phr of
reinforcing carbon black (N326) and because the viscosity is plotted versus the shear stress, the curve
obtained suggests the occurrence of a yield stress, below which no ¯ow would be observed (i.e. the
viscosity would be in®nite). A similar behavior is seen (right part of the ®gure), when substituting a non-
reinforcing grade (N990) by a reinforcing one (N326). Corresponding observations have been reported
with other ®lled rubber compounds, either SBR [12,13] or EPDM [14] based. In generating such data,
the authors had to use three types of instrument: a `sandwich' rheometer in the low rate (or stress) region,
a variable speed Mooney viscometer in the medium range and a capillary rheometer for the highest
range. There are mathematical models for the viscosity function that explicitly consider the occurrence
of a yield stress s c, for instance the Herschel±Bulkley equation:

s ˆ sc 1 K…lg_ †n …3a†
632 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 1. Effect of ®ller level and structure on the shear viscosity function of rubber compound; drawn with data by Montes et al.
[11].

that leads to write the following equation for the viscosity±shear stress function:
!
K ln21
h…s† ˆ s s . sc …3b†
s 2 sc

where K and n are power law parameters, l is a characteristic time needed for unit consistency and g_ the
shear rate. Would one ®t (by non-linear regression) the data in Fig. 1 with Eq. (3b), it is obvious that
de®nite values would be obtained for the yield stress s c. One notes that Eq. (3b) has a mathematical
singularity for s ˆ sc and has no physical meaning for s , sc :
Observations in the low shear rate range are however tedious and require special instruments, gener-
ally R&D prototypes, e.g. the `sandwich' rheometer, whose accuracy can be improved. A recently
developed sliding cylinder rheometer was used to investigate the very low shear rate region with ®lled
rubber compounds [15]. Basically the instrument submits to simple shear an annular sample contained in
the gap between two coaxial cylinders. The shearing force is applied with a dead weigh and an optical
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 633

linear transducer is used to record with a PC the displacement of the inner (shearing) cylinder with an
accuracy of 1 mm. The device can thus measure very low displacement rates, in the 1 mm per 10 23 s
range, which corresponds to a shear rate of 10 26 ±10 25 s 21 depending on the viscosity of the tested
material. As shown in Fig. 2, down to 10 25 s 21, no Newtonian plateau is observed (left part of the ®gure)
with this instrument when testing a standard carbon black ®lled polybutadiene compound, but one would
also hardly detect the occurrence of a yield stress on the h (s ) function (see right part of the ®gure). One
might thus argue about the yield stress as a true characteristic of ®lled materials since it is essentially an
extrapolated value and obviously related to instrument capabilities. Nevertheless, one could otherwise
consider that when the shear viscosity becomes very large, lets say higher that 100 kPa s, below a certain
value of the shear stress, then such a yield stress is an engineering reality.
Other rheological experiments reveal however that the disappearance of the linear viscoelastic region
with increasing ®ller content is a general characteristic of most practical rubber compounds.
Fig. 3 was drawn with data generated by Dick and Pawlowski [16] using the Rubber Process Analyzer
RPA 2000, a torsional dynamic rheometer, especially designed for testing very stiff materials. Strain
sweep experiments at ®xed frequency and temperature were performed on a series of compounds with
various carbon black or silica contents and quite a high amount of processing oil. Results are presented in
terms of elastic modulus G 0 versus percent strain. The plateau that corresponds to a linear viscoelastic
behavior is readily seen either with the gum rubber or with low ®lled materials. Above a certain ®ller
level, the linear behavior is lost and compounds exhibit a modulus drop similar to the so-called Payne
effect [17] in vulcanized rubber. Would lower strain magnitude be considered, unfortunately outside the
range of the instrument, it is nevertheless possible that a linear plateau exists for highly ®lled materials
but anyway inaccessible to experience and of no signi®cance in terms of process engineering.

Fig. 2. Viscosity function of a carbon black ®lled polybutadiene compound showing the absence of Newtonian plateau down to
10 25 s 21; data from BarreÁs and Leblanc [15].
634 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 3. Strain sweep experiments on carbon-black and silica ®lled SBR compounds with a torsional dynamic tester; data from
Dick and Pawslowsky [16].

3.2. Wall slippage effects

Wall slippage is a long-standing problem in polymer rheology and much research effort was related to
slip effects in association with the rheological characterization of rubber and its compounds, probably
with respect to the complex formulation of such materials. Above a critical ®ller content (10±15%
weight), rubber compounds exhibit slippage at the wall of laboratory or processing equipments,
obviously further enhanced when lubricating ingredients are present in the formulation.
Easily demonstrated in qualitative experiments [18±21], wall slippage is dif®cult to quantify and
when applied to ®ller rubber compounds, the well known Mooney method [22] gives sometimes striking
and obviously wrong results, for instance slip volume rate larger than the overall volume rate [23,24] or
negative wall slip rate [25]. This apparent non-sense results from the hypotheses drawn by Mooney in
developing his method, namely that wall slip rate is simply proportional to shear stress and that the
material is incompressible (see Appendix A for details). With rubber compounds, a number of disturbing
effects become signi®cant and, therefore, the Mooney method is no longer valid. Some of these effects,
for instance (pseudo) compressibility, are clearly related to the presence of the ®ller and, as
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 635

Fig. 4. Capillary rheometer experiments with a carbon-black ®lled EPDM compound exhibiting complex pressure-slippage
effects; data from Jepsen and Raebiger [27].

demonstrated through experiments with a pressurized rotational rheometer, slippage readily occurs with
®lled rubber compounds at low applied pressure [26]. Extrusion is obviously pressure ¯ow and conse-
quently wall slippage in capillary rheometer experiments may be expected to be strongly pressure
dependent, as indeed observed by several authors and explicitly considered in a recent purposely-simple
method to treat capillary rheometer data [19]. Wall slippage and compressibility appear to be effects of
opposite signs in (capillary) extrusion ¯ow.
Typical data by Jepsen and Raebiger [27] illustrate well the dif®culty in considering quantitatively
wall slip effects in the extrusion ¯ow of ®lled rubber compounds. As shown in Fig. 4, capillary rheometer
data with two dies of different diameters do not superimpose below g_ a ˆ 100 s21 : However, it is clear
that the Mooney treatment for wall slip would make sense only for the data between 1 and 100 s 21 since
in the very low shear rate range, a plot of the apparent shear rate versus 1/D at constant shear stress,
would yield negative wall slip rate. The fact that the ¯ow curves assessed with capillary dies of different
diameters do converge in the high rate region, where high extrusion pressure are applied, is nevertheless
a clear indication that wall slip effects are pressure dependent with ®lled rubber compounds.

3.3. Lower extrudate swell and smoother melt fracture defects with increasing ®ller content

Reinforcing ®llers in general, and particularly high structure carbon blacks improve the extrusion
characteristics of elastomers by decreasing extrudate swell. As illustrated by Fig. 5, the extrudate swell
decreases with increasing carbon black content, an effect reported by many authors [1]. With relatively
636 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 5. Effect of carbon black level on post-extrusion swelling.

low structure blacks, it has been observed [28] that the effect of ®ller content on post-extrusion swelling
can be represented by a single curve when the swelling ratio B (i.e. extrudate cross-section over die cross
section) is divided by the volume fraction of the elastomer, i.e. (1 2 f ) where f is the ®ller volume
fraction. With more reinforcing blacks, this single correction for elastomer volume fraction does not give
anymore a master curve and, in addition, reduces the B/(1 2 f ) value, at constant stress, more than
expected. Moreover, at constant ®ller level, a relationship has been found between dibutyl phtalate
absorption data for carbon blacks (a measure of particle structure Ð ASTM D2414) and post-
extrusion swelling of SBR compounds [29,30]. This indicates that high-structure blacks restrain
post-extrusion swelling by a more complex mechanism than a mere hydrodynamic effect. If,
according to well established considerations [31], post-extrusion swelling is considered as the
recovery of elastic strain stored during die ¯ow, it can be suggested that reinforcing ®llers have a
kind on dampening effect, likely related with the complex compound morphology due to rubber±
®ller interactions [5].
The bene®cial effect of reinforcing carbon blacks in reducing the severity of extrudate distortion when
it occurs is pragmatically exploited for long in the rubber industry but has received very little research
interest. The critical shear for the onset of extrudate defects (the so-called melt fracture) increases both
with black level and with the speci®c area of carbon black particles [32].

3.4. Anisotropic effects in ¯ow

Flow induced anisotropy effects are typical of high molecular weight polymer materials, readily
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 637

Fig. 6. Demonstrating ¯ow induced anisotropy effects in injection molding ®lled rubber compounds; drawn with data from
Nakashima et al. [33].

develop in converging ¯ow conditions and are further enhanced when reinforcing ®llers are present. In
rubber technology, a simple practical evidence is offered by the so-called `grain effect' in calendered
rubber sheets, i.e. a ®lled rubber compound exhibits higher tensile properties in the calendering direction
than in the perpendicular one. The anisotropic shrinkage of injection molded rubber parts offers another
demonstration, as illustrated by Fig. 6, drawn with data of Nakashima et al. [33]. Using various ®lled
compounds, plaques where molded with a central injection gate in such a manner that radial ¯ow
occurred; curing was made at the end of the molding process and tensile specimens were cut as indicated.
As can be seen, measured tensile modulus signi®cantly depends on the stretching direction with respect
to molding ¯ow direction, thus demonstrating the persistence of ¯ow induced anisotropy effect after
vulcanization.
In macromolecular systems, the internal spatial organization depends on the environment and on local
singularities, and therefore ¯ow-induced anisotropy is expected. All the more with chain molecules that
are known to develop entanglements and, obviously, the presence of reinforcing ®llers and likely
associated long-range interactions are enhancing this character.
638 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 7. Classi®cation of ®llers according to average particle size.

4. Interactions of ®llers and elastomers

4.1. Filler particle size and structure as primary parameters

Links between the capability of certain minerals to impart reinforcement to elastomers and the
development of interactions between ®ller particle surfaces and rubbers date back to the earliest obser-
vations in rubber technology, and as underlined above, related effects are observed in the rheological
properties of uncured rubber compounds. In what the ®ller is concerned, the single most important
parameter in such effects is the average particle size, as shown in Fig. 7. Particles larger than 10 3 nm do
not have reinforcing capabilities (at best) or have a detrimental action, and generally increase viscosity
by a mere hydrodynamic effect. Reinforcement is readily obtained with sizes smaller than 100 nm but
particle structure appears as a more decisive factor. Two classes of minerals have been found to offer
signi®cant reinforcing capabilities, carbon black and silica, providing their manufacturing technology is
such that quite complicated tri-dimensional objects involving several elementary particles of appropriate
size are obtained. The spatial complexity of such objects is referred as the `structure' of the ®ller.
Detailed discussions about the structural aspects of reinforcing ®llers are found in textbooks [2,3] and
only basic elements needed in the context of the report will be brie¯y outlined here. High-resolution
electron microscopy has long shown that carbon black is made up of complex arrangements of spherical
entities whose diameter ranges from 10 to 90 nm. The elementary black particles, the so-called colloidal
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 639

black, are spheres made up of broken quasi-graphitic layers whose stacking gives edges with a steps-like
structure, as suggested by recent observations with the atomic force microscope [34,35]. These particles
exist in various forms of aggregation, depending on the manufacturing process. The extent to which
aggregates form complex tri-dimensional objects is referred as `structure'. Four shape types can be
recognized for carbon black aggregates [36]: spheroidal, ellipsoidal, linear and branched; whilst some
rough correlations can be expected between such structural aspects and the reinforcing capabil-
ities of the ®ller, there is no single relationship between a particular type of aggregate shape and
the standard classi®cation of carbon black and, furthermore, several shapes may coexist in a
given grade. Several aggregates further give weak, giant assemblies called agglomerates. Carbon
black structure gives rise to reinforcing effects and therefore the aggregate is the smallest form of
a given carbon black grade well dispersed in an elastomer that will still keep all the reinforcing
capabilities of the ®ller. This follows from the fact that colloidal blacks have poor, if any,
reinforcing properties and that agglomerates persisting in a vulcanized compound are failure
initiation sites.
The reinforcing character of carbon black grades is accordingly related to both the size of the
elementary particle and the structure of aggregates essentially, whilst certain singular ¯ow properties
of ®lled rubber compounds can be shown to depend also on other parameters. The ASTM classi®cation
of carbon black re¯ects the importance of carbon black structure, as shown in Table 1. All carbon blacks
are graded with respect to a four-character code, i.e. Nxyz, where N stands for `normal curing', meaning
that the ®ller does not interfere much with vulcanization chemistry and xyz are three digits describing the
reinforcing character. The ®rst digit x was referring to the average typical size of the elementary particle
in ASTM D1765-86, and is now inversely proportional to the average speci®c area, according to the
most recent version of the standard, ASTM D1765-96. The two digits yz form a number that describes
the structure of the aggregate; whilst no rules are given in the standard how to assign the yz digits to a
given black, it is generally agreed that the higher yz, the more intricate the aggregate and hence its
reinforcing character, for instance N340 is a more reinforcing grade that N327, whilst their elementary
particles have essentially the same diameter, in the 26±30 nm range. A low structure black may have less
than 20 elementary particles per aggregate, though a high structure one may consist of aggregates with
up to 200 particles, as long observed on electron microphotographs [37].

Table 1
Carbon black classi®cation

Classi®cation Nxyz ASTM D1765-86 ASTM D1765-96


x Average (elementary) particle size (nm) N2 average speci®c area (m 2/g)

0 1±10 .150
1 11±19 121±150
2 20±25 100±120
3 26±30 70±99
4 31±39 50±69
5 40±48 40±49
6 49±60 33±39
7 61±100 21±32
8 101±200 11±20
9 201±500 0±10
640 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Table 2
Silica typical properties

Property Fumed silica Precipitated silica

N2 (BET) surf. area (m 2/g) 50±380 140±250


Elementary particle diameter (nm) 7±40 15±100
DBP adsorption (ml/100 g) 200±280 175±285
CTAB ads. surf. area (m 2/g) 160±220 110±200
pH (4% suspension in H2O) 3.6±4.5 6.0±9.0

Reinforcing silica, either precipitated or `fumed' grades, exhibit a similar complex geometry,
from elementary spherical particles to aggregates, agglomerates and clusters (due to hydrogen
bonding). Owing to their high manufacturing cost, silica obtained by pyrogenation (high tempera-
ture hydrolysis) have received less attention in rubber reinforcement than precipitated ones that,
nowadays, are readily used in the manufacture of so-called `green tires'. Silica elementary
particles are linked by chemical bonds to form `string of pearl' structures (or aggregates) with
dimensions in the 50±500 nm range. As for carbon black, silica aggregates cannot be broken
during mixing but are prone to form agglomerates which in turn gives relatively loose clusters
through hydrogen bonding, owing to the speci®c surface chemistry of silica (see below). There is
no standard classi®cation for silica which are essentially distinguished with respect to their
speci®c area, within quite a broad range, 50±380 m 2 g 21, as measured by nitrogen adsorption
(see Table 2). At similar reinforcing capabilities, silica speci®c area are larger than for their
carbon black equivalent, in line with their respective bulk speci®c gravity, i.e. 2.2 g cm 23 for
silica, 1.8 g cm 23 for carbon black.
In discussing rubber±®ller interactions, there is thus a problem of dimensions to consider, from
the colloidal particle (which no reinforcing capability) to the aggregate, the agglomerate and
eventually the pellet, by order of increasing size, in comparison with relevant dimensions for
macromolecules (see Appendix B). Fig. 8 highlights this comparison, using typical dimensions
that can be considered either for carbon black and silica or for a rubber macromolecule. Boxes
indicate the objects and dimensions that are likely to be relevant in terms of rubber±®ller
interactions.

4.2. Filler surface characteristics as secondary particle parameters

Apart from speci®c surface area as discussed above, it sounds logical to expect rubber±®ller inter-
actions to depend upon the surface activity of the particles. How surface activity is de®ned is quite
unclear as many phenomena might be involved, from Van der Waals proximity forces to speci®c
chemical interactions. There are no standard methods to measure surface activity.
Rubber grade carbon blacks contain some quantities of chemically combined hydrogen (0.2±
1.0%), oxygen (0.1±4.0%) and even sulfur (up to 1.0%) depending on the quality of the feed-
stock and the process. A large variety of oxygen containing functional groups, most in minute
quantities, has been detected in carbon blacks, for instance carboxyl and hydroxyl groups, phenol,
lactones, quinones, ketones, aldehydes, hydroperoxydes (Fig. 9). It seems however that the
presence of oxygen complexes is not essential for reinforcement in most rubbers, with the notable
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 641

Fig. 8. Relevant dimensions in rubber±®ller interactions.

exception of polar elastomers, e.g. butyl rubber. As convincingly demonstrated by Gessler et al.
[38] oxygen functionality on carbon black is a requisite to high-order interactions between ®ller
and butyl rubber. Indeed when the surface oxygen on channel black is removed by devolatilisa-
tion, signi®cant loss in reinforcement (damping properties) of butyl rubber is observed. Similarly,
furnace blacks are activated (for butyl rubber) when oxygen is added by heating them at 250±
3008C in a stream of oxygen. With non-polar elastomers, i.e. most general purpose rubbers,
including NR, BR, SBR, EPR and EPDM, the occurrence of chemical reactions with functional
groups on carbon black surface is far less convincing. The heating of carbon black below 8008C
does not result in graphitization of the ®ller, i.e. there are no signi®cant changes in the crystal-
linity of the inner particles, but it does remove most of the chemisorbed surface oxygen.
However, rubber±®ller interactions remain essentially unaffected, as shown by Dannenberg's
experiments with SBR compounds [39,40]. There is a consensus today to consider that the strong
rubber±carbon black interaction is not necessarily resulting solely from chemical reactions invol-
ving oxygen complexes at the surface of particles.
The situation is totally different with silica whose surface is occupied by sizable quantities of siloxane
and silanol groups, giving rise to hydrogen interaction with either `free' or `bound' water. Free water is
easily removed by drying at 105±2508C, while bound water is only released at 900±10008C and results
in fact from the condensation of vicinal silanol groups. As can be expected, the free moisture content of
silica strongly affects the rheological and curing properties of silica ®lled rubber compounds. Another
642 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 9. Surface chemistry of carbon-black and silica.

important consequence of the oxygen rich surface chemistry of silica are the strong inter-particles
interactions through hydrogen bonding that, on one hand give rise to poorer dispersibility than carbon
black, and on the other hand permit the use of suitable chemical promoters, e.g. bis(triethoxysilylpro-
pyl)tetrasulfane, TESPT, to form covalent bonding with (unsaturated) elastomer. Of course, certain
specialty elastomers such as polydimethylsiloxanes naturally interact with silica, thank to their similar
chemistry.
An attempt to quantify the role of ®ller surface chemistry is to consider the so-called `surface activity',
which may be assumed to be assessed through the surface energy g s, which consists of two components
[41], i.e.:
gs ˆ gsd 1 gsp …4†
where gds and g ps are the so-called dispersive and polar (or speci®c) components, respectively.
Such properties are measured by Inverse Gas Chromatography (IGC) [42], a technique in which the
®ller to be characterized is used as the stationary phase and the injected solute is the so-called probe. In
practice, the ®ller particles are carefully poured into a stainless steel column with an appropriate
diameter, typically a few millimeters. Suitable model chemicals, in dilute solution, are used as probes
in order to quantify their interactions with the ®ller. When the probe is operated at in®nite dilution,
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 643

information is obtained concerning the adsorption of a solute on a solid surface, by use of the Henry's
law that considers the standard free energy of transferring one mole of vapor from a gas phase (at the
standard pressure of 1 atm or 101 kPa) to a standard state on the surface [43]. By injecting a series of
homologous n-alkanes (e.g. pentane to decane) as probes, the dispersive component of the surface
energy gds can be obtained from the free energy of adsorption, by considering the slope of the measured
standard free energy for adsorption vs the number of carbon atoms for different n-alkanes. The speci®c
(or polar) component is obtained from the difference in the free energy of adsorption between a polar
probe and a real or hypothetical n-alkane with the same surface area (see details [41,42,44±47]). Table 3
gives typical data as reported in literature.
Carbon blacks exhibit a high dispersive component, actually proportional to their speci®c surface

Table 3
Surface energy components for carbon black and silica

Filler Speci®c Dispersive Polar component Source


sp
surface area component gds (at Ibenzene (g sp) (at
N2 (BET) 1508C) (mJ/m 2) 1508C) (mJ/m 2)
(m 2/g)

Carbon black
N110 140.0 270.4 120.0 [45]
N220 118.0 235.2 103.9 [45]
N326 83.2 186.5 90.2 [45]
N330 76.5 196.9 85.9 [45]
N330 80.0 150.4 80.2 [46]
N347 85.8 192.9 87.9 [45]
N550 39.7 134.4 75.0 [45]
N550 43.2 173.4 75.0 [47]
N660 39.4 124.7 71.1 [45]
N762 32.5 126.4 77.7 [45]
N762 24.0 132.8 74.0 [47]
N774 29.0 118.1 63.8 [46]
N880 12.3 113.1 63.9 [47]
N990 7.9 71.8 56.6 [45]
N990 10.3 78.7 58.8 [47]
Silica
Ultrasilw a VN2 134.0 21.3 b 73.1 c [41]
Ultrasilw VN2 134.0 22.9 64.0 [45]
Ultrasilw VN3 181.0 27.5 b 79.6 c [41]
Ultrasilw VN3 181.0 34.3 71.9 [45]
Aerosilw d 130 128.0 27.3 b 48.1 c [41]
Aerosilw 130 128.0 30.7 45.4 [45]
Aerosilw 200 189.0 40.9 c 57.4 c [41]
Aerosilw 200 189.0 44.3 55.1 [45]
Aerosilw 300 300.0 ,55.0 ,51.0 [44]
a
Ultrasil is Degussa trade mark for precipitated silica.
b
Calculated from data in Table III in source paper.
c
Read on Fig. 16 in source paper.
d
Aerosil is Degussa trade mark for fumed silica.
644 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

area, and a relatively low polar component, nearly constant, whatever the grade. On the reverse,
reinforcing silica grades have a low dispersive component, but a high gsp : On that basis, some authors
underline that with carbon blacks, the reinforcing effects is essentially achieved by means of strong
®ller-to-rubber interactions, and that the polar component is re¯ected by a weak carbon black network.
The opposite picture of gds and gsp with silica explains why ®ller±®ller interaction dominates in silica
®lled compounds, thus leading to the formation of silica networks and the associated dif®culties in
dispersing such a material in non-polar elastomers. Particles networks are detrimental to reinforcement
and call for the need to chemically promote the rubber±®ller interaction by using bi-functional chemi-
cals that, on one side, react with silanol groups that densely cover the surface of silica particles and, on
the other side, undergo covalent bonding with rubber chains.
Recently, new equilibrium gas adsorption techniques were used to analyze the surface energy distri-
bution of carbon blacks [48,49]. Four different energetic sites were found, whose nature is depending on
the model considered for the surface of the ®ller. For instance, if one considers with some authors
[50,51] that the surface of elementary carbon black particles consists of turbostatic graphitic crystallites
beside areas of amorphous carbon, then one could assign such energetic sites to different features:

Boundaries between two neighbor crystallites,


Boundaries between crystallite and amorphous areas,
Graphitic surface of a crystallite,
Surface of amorphous carbon.

The fraction of higher energetic sites is then found to decrease with the reinforcing capabilities of the
black and to disappear almost completely through graphitization. It can therefore be hypothesized that
these sites play an important role in ®ller±®ller and ®ller±elastomer interaction. However, because of
the relatively low energies of these sites, weak interaction forces (London, Van der Waals,¼) between
carbon black surface and unsaturated segments in the rubber backbone may prevail, with obviously a
signi®cant cumulative effect due to the large number of such sites.
With respect to the main subject of this report, one might consider that the exact nature of the surface
of carbon black particles is detailed knowledge in comparison with macroscopic phenomena involved in
¯ow properties. The key information would appear to be the demonstration of the surface roughness of
elementary carbon black particles, whether one consider a disordered array of graphitic crystallites or a
step like surface resulting from a spiraling growth process. It is quite remarkable that nearly all carbon
blacks have a unique surface roughness on atomic scales with a surface fractal dimension df < 2:6; in
line with the most recent theoretical models for physical concepts of surface growth that are nowadays
found to be valid in many different ®elds in nature. One such theoretical approach is the so-called Toom
model, i.e. a relatively simple probabilistic cellular automaton that appears to be `generic' for a variety
of physical patterns [52]. In the context of the present report, such fundamental theoretical works might
seems quite remote, but it is striking to see that through calculation with the appropriate equations, step-
like surface patterns are built that are very similar to what is observed on transmission electron micro-
scopy images of carbon black [53]. It seems thus that there is a convergence of experimental and
theoretical results that describe the rough surface of elementary carbon black particles as a pattern of
overlapping scales.
Another indirect demonstration of the importance of the very surface of ®ller particles in the devel-
opment of strong interactions with polymer materials is provided by the fact that, when chemically
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 645

grafting short hydrocarbon chains on carbon blacks, the reinforcing capabilities of the ®ller are either
lost or at least severely penalized. Vidal et al. [54] showed, for instance, that the esteri®cation (via
carboxyl groups at the surface) of carbon blacks, using either methanol or hexadecanol, is associated
with a strong decrease of the interactions the ®ller can develop with any molecule. The surface free
energy is found to decrease whilst the surface energy homogeneity is increasing. As expected, when
grafting ®ller particles, inter-particle interactions are lower, an effect that might be bene®cial for
dispersion mechanisms; but polymer±®ller interactions are also decreasing, which results in a severe
drop in reinforcement.

5. Bound rubber as macroscopic result of rubber±®ller interactions

5.1. Bound rubber: a long known effect in rubber science

When an elastomer and a reinforcing ®ller are mixed, strong interactions occur in such a manner that a
good solvent of the polymer can extract only a free rubber portion, leaving a highly swollen rubber±®ller
gel. Bound rubber (BdR) is, by de®nition, the rubber content of that gel.
This effect is known for more than 80 years in the case of carbon black ®lled compounds and because
of the striking parallelism between the amount of bound rubber and the reinforcing capabilities of the
black, it is considered one of the major factors in reinforcement and often a global measure of surface
activity of the ®ller. Nowadays however, one would consider BdR as a sometimes misleading measure-
ment of surface activity on one hand because only a portion of the non-extractable elastomer may be
effectively adsorbed and, on the other hand, because single chains are likely to develop multi-attachment
on the ®ller surface. For a given elastomer, the amount of bound rubber at ®xed ®ller content depends on
a number of factors, namely the surface area, structure and surface activity of the ®ller, the state of
mixing and the storage maturation of the compound. The chemical structure and the macromolecular
characteristics (molecular weight, polydispersity, branching, etc.) of the rubber are also important
parameters affecting bound rubber content. For long, BdR was considered as a striking phenomenolo-
gical aspect in carbon black reinforcement, without sound consideration for its very origin. With respect
to the previous section, it is obvious that bound rubber is an `all inclusive' macroscopic property surely
re¯ecting many diverse (nanoscopic) effects, some of which that became clear only recently, due to
developments in advanced analytical instrumentation and techniques (e.g. atomic force microscopy,
nuclear magnetic resonance,¼).
It is worth underlining however that a rubber compound is quite a complex system, subject to many
in¯uencing parameters, most of them still poorly understood. It is ®rstly a dense system, consisting of a
relatively stiff matrix in which various ingredients are dispersed, more or less well, depending on the
ef®cacy or the achievement of the mixing operations. In terms of viscoelasticity, the matrix, i.e. the
rubber, is at rest on its `rubbery plateau', a state of matter intermediate between a glassy solid and a
(highly) viscous ¯uid, in which macromolecules undergo segmental motions characterized by two main
groups of characteristic response times. Short characteristic times concern local ¯uctuation in chain
conformation, longer times correspond to large-scale segmental motion. The dispersed ingredients, i.e.
the ®ller, the chemicals, obviously interact or at least interfere with the rubber matrix but the time scale
for such interactions (or interferences) is obviously depending on the characteristic response time of the
polymer material. Filler surface information such as surface activity are obtained in conditions that are
646 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

far to correspond to those prevailing in a dense system; for instance, the chemical probes used in IGC are
small molecules, in a dilute environment. A bulk effect such as bound rubber is occurring in the actual
compound, i.e. a dense environment. One cannot therefore expect more than a loose correlation between
bound rubber and the surface activity of ®ller, as measured by techniques like IGC.
Bound rubber in carbon black ®lled compounds is a phenomenon whose description can be traced
back in literature as early as 1925 [55], when the poor swelling in petroleum naphta of raw rubber mixed
with carbon black and the dif®culties in separating the two mutually adsorbed substances were observed.
The term `bound rubber' seems to have been coined in 1937 by Fielding [56] whilst, prior to this date,
the capability of carbon black in insolubilizing rubber had already been investigated by several authors
who used the rather inappropriate wording `carbon gel' to describe the phenomenon. Since that period,
survey papers were quite often published, that dealt with the mechanism and factors affecting it, as well
as its in¯uence on rubber properties [1,2,57±60]. To consider that bound rubber results from the
adsorption of the polymer by the ®ller is an assumption of common sense, made in a number of
pioneering works. However, with respect to strict physico-chemical considerations, the adsorption of
the polymer on the particle surface would involve the existence of a physical equilibrium and conse-
quently the possibility to completely recover the elastomer by suitable physical manipulations. Experi-
mental proofs for the (at least partial) reversibility of bound rubber formation were obtained only
recently by Wolff et al. [61], who performed measurements at temperatures ranging from room tempera-
ture to 1008C. Below 708C, only a slight decrease in bound rubber with increasing temperature is
observed; above 808C, there is a severe drop in BdR content which however does not exclude the
possibility that some rubber chains remains attached to ®ller particles, thus resisting to extraction by
a boiling solvent. Kida et al. [62] reported similar observations on compounds prepared by mixing ISAF
(,N220) carbon black with benzene solution of synthetic polyisoprenes of various molecular weights
…M w ˆ 490±1280 kg mol21 †: Filled compounds were recovered from the mixed solution by a freeze-dry
technique and BdR was assessed through the usual solvent extraction method. When varying the
extraction temperature in the 25±1108C range, bound rubber content showed a decrease of around
45%. Bound rubber formation Ð and the associated rubber±carbon black interactions Ð appears
thus as mainly a physical process, while some chemisorption cannot be excluded with available experi-
mental evidences.
From fundamental works in the 1950s on adsorption from polymer solution, it is known that there is a
relationship between the polymer molecular weight and the amount of bound rubber. For instance,
Kolthoff and coworkers observed that low-molecular-weight polymer fractions are adsorbed rapidly
from solution by carbon blacks. Over a period of time, however, they are replaced by higher MW
polymer with little or no change in the total amount of adsorbed polymer [63,64]. As we will see
below, such a view is developed in most theoretical works on bound rubber formation. Jenkel and
Rumbach [65] obtained similar results for several plastic polymers and inferred that each macro-
molecule is adsorbed at one or a few points on the ®ller surface. Such experiments were performed
in conditions (polymer in solution) that are far from meeting those prevailing in an actual compound, i.e.
a dense system, and it has been shown that bound rubber formed during mixing involves a greater
amount of rubber than could be adsorbed from solution [66].
It is well established today that adsorption of polymers on solid surfaces from solution occurs by a
multi-contact mechanism that involves widely separated segment of the polymer chain. Consequently,
there are pronounced hysteretic effects since desorption would require the simultaneous release of all
contact points. Such an event has obviously a very low probability when the number of contacts is large.
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 647

Adsorption hysteresis however does not prove chemisorption. Qualitatively the same conclusions may
be drawn when the bound rubber is formed either by adsorption from a solution or during the mixing
process, namely that the high MW polymer is preferentially bound [67±69]. This means that statistically
any type of bonding of polymer to ®ller will preferentially involve the larger molecules. When a ®ller
and rubber are dry mixed, the resulting quantity of bound rubber obviously re¯ects all possible adsorp-
tion mechanisms as well as intermolecular entanglements. Again, because of the hysteretic nature of
physical adsorption, chemical interaction between the polymer and the polymer is not necessary for the
development of bound rubber, at least in the case of high structure carbon blacks. High structure silica
are also able to physically adsorb rubber segments but, with respect to their peculiar surface chemistry,
much more pronounced effects are obtained when chemical interactions are deliberately produced
through the appropriate chemicals, e.g. silanes. As we will see below, such a view is developed in
most theoretical work on bound rubber formation.

5.2. Measurement of bound rubber

In principle, measuring bound rubber is a simple but tedious technique, which consists essentially in
determining the weight percent of insoluble polymer remaining on the carbon black, following the
swelling of small pieces of the uncured compound in a good solvent of the elastomer. The technical
dif®culties are related to the excessive swelling, typically as much as 50 times the original volume of the
immersed sample, and the handling of a weak gel, hopefully coherent if the ®ller level is large enough,
that contains all the ®ller particles, a small portion of the original rubber content and typically more that
90% solvent. To the author's knowledge, the only attempt for a standard test is the method described in
the French norm AFNOR T45-114 [70]. Most published results on bound rubber do not refer to any
normalized method but the techniques used look rather similar to the French norm, with however
variations in solvent choice or extraction time.
Gessler et al. [38] used the following method: 0.5±1.0 g of sample were diced into a number of small
cubes that were introduced into a suf®cient quantity of solvent (generally 100 cm 3) contained in a tightly
capped vessel, that was gently inverted once each day to enhance the diffusion of the extractable
elastomer in the liquid. The rubber sample was maintained in the solvent at room temperature for several
days in order to presumably achieve full extraction of labile rubber. When possible, the wet, swollen
rubber pieces were separated from the solvent with a suitable screen, washed with fresh solvent and
blotted quickly with ®lter paper. When the gel was too tenuous to handle, the swollen pieces were
collected and washed in a weighed ®lter paper. Bound rubber content was obtained from difference in
sample weight, before and after the extraction. The solvent was evaporated by means of a controlled air
stream over a speci®ed time interval. As underlined by the authors, this method is valid only if the shape
of the test sample is kept during the whole immersion period, and disintegration or other disturbances are
avoided. Whilst less detailed, the method used by Dannenberg [39,40] seems similar, except that
vacuum drying at 508C was used to recover the extracted sample.
A fully detailed method is described by Leblanc and Hardy [71], as follows: around 0.5 g of sample is
cut in small pieces and introduced in a steel wire basket previously weighed (m1); the basket is closed
and weighed (m2) and suspended in 150 ml of solvent, in a vessel closed with a aluminum foil to
minimize evaporation. The solvent is agitated with a magnetic stirrer. After 72 h at room temperature,
the basket is slowly removed from the solvent, dried a few hours at room temperature, then for 24 h
under vacuum at 408C. Complete drying is checked by constant ®nal weight (m3). The amount of bound
648 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

rubber (in % of initial rubber content of the compound) is given by:


m0 2 …m2 2 m3 †
BdR…%† ˆ 100 …5†
m0
where m0 is the rubber content in the sample, given by m0 ˆ …m2 2 m1 †100=wcpd with m1 the mass of the
empty basket and wcpd the total formulation (in phr; part per hundred rubber); m2 is the mass of the basket
plus the unextracted sample; m3 is the basket plus the extracted, dried sample. Wolff et al. [61] used
essentially the same method, with however some differences in sample size (0.2 g), extraction duration
(7 days) and a smaller quantity of solvent (25 ml) that was renewed after three days.
Roychoudhury and De [72] conditioned rubber±carbon black mixes for 7 days at room temperature
before immersing in 300 ml solvent a 150 mesh stainless-steel cage containing a 0.5 g sample cut in
1 mm 3 pieces. The solvent was renewed every 24 h and, after 72 h, the cage (with the bound rubber and
the ®ller) was removed from solvent and vacuum dried at room temperature to constant weight. Wu et al.
[73] weighed precisely 0.5 g of ®lled silicone rubber, wrapped them in ®lter paper and immersed the
wrap in 100 ml solvent at room temperature for 9 days; solvent was renewed every 3 days. After 9 days,
the wrap was recovered, immersed in acetone in order to displace the remaining toluene, and dried.
Bound rubber was obtained by weighing. Hamed and Hat®eld [74] suspended strip samples weighing
about 100 g in 1800 ml solvent for various times up to 168 h, by replacing the solvent every 24 h. After
extraction, samples were dried for 48 h under vacuum at room temperature, and the carbon black content
of the dried specimens were determined from density measurements with a commercial gravitometer.
Whilst essentially performing bound rubber measurements, these authors expressed their results in phr of
black.
KaraÂsek and Meissner [75] cut small samples of measured mass, about 1 g, out of compression
molded sheets (20 min at 1008C) and placed them in weighed pouches made of polyamide mono®lament
fabric. Closed pouches were put into 200 cm 3 bottles and 100 ml of solvent were added. The solvent was
changed three times during 5 days and the extracts were collected for analysis. After extraction, the
pouch containing the swollen, coherent gel was dried and weighed in order to obtain the mass of
polymer±®ller material. Thereafter the pouch with the dried gel was burnt in a laboratory furnace
and the mass of the residue (essentially the ®ller) was measured. From these data, the bound rubber
was calculated by taking into account the ash content of the fabric and the ignition loss of the ®ller.
The above procedures can be called `standard' static extraction methods. If the swelling time varies
with authors, it must obviously be suf®cient for the optimum extraction of the free rubber to be achieved.
There is however no implicit veri®cation that this is indeed the case.
In order to overcome this testing de®ciency and to further increase the capabilities of the technique,
Leblanc and Stragliati [69] developed a method to investigate the kinetics of extraction of the unbound
fraction of rubber compounds. As illustrated in Fig. 10, a special glass device is used to perform the
swelling of test samples in a ®xed quantity of solvent for well-de®ned periods. The sample, cut in small
pieces, is weighed in a steel wire basket and disposed in the glass vessel with a polytetra¯uoroethylene
valve at the bottom. A know quantity of solvent (100 ml) is poured in the vessel and left in contact with
the sample; after various periods of time, the solvent (which contains some extracted species) is
collected through the valve and another portion of pure solvent is introduced in the vessel for a further
extraction period. The procedure is repeated until complete extraction is achieved. Aliquot quantities of
collected extracts are evaporated under vacuum at 508C and the extracted quantity assessed by weighing
the dry residue. Test data are thus the extracted rubber (g) for various extraction periods (h). Except for
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 649

Fig. 10. Extraction kinetics method for the assessment of bound rubber; principle of the method.

model binary formulation (i.e. rubber 1 black), a correction must be made for non-rubber soluble
ingredients, generally the oil, the stearic acid and certain chemicals. By considering that at any time
ti, the non-rubber soluble ingredients are extracted proportionally to the labile rubber, the corrected
amount of unbound elastomer is obtained from:
( P )
WS … Pj †
‰XC Ši ˆ ‰XA Ši 1 2 …6†
Pcpd ‰XA Šf

where WS is the sample mass (g); Pj are the soluble ingredients parts (phr); Pcpd the overall formulation
parts (phr); [XC] and [XA] the corrected and apparent extracted rubber (g); subscript i corresponds to time
ti; and subscript f corresponds to the ®nal time of the extraction experiment. After this correction is made,
the results are expressed in percent extracted rubber versus the initial gum rubber content in the
compound, and used to draw the extraction kinetics curve (Fig. 11). A model for the extraction kinetics
is then used to ®t such data by non-linear regression, according to:
‰%extractedŠt ˆ …100 2 ‰BdRŠ†{1 2 exp…2bt†} …7†

where [%extracted]t is the percent extracted rubber at extraction time t, [BdR] the bound rubber content
for an in®nite extraction time, b a kinetic extraction parameter and t the time. This technique does not
only provide the absolute bound rubber content (at the extraction temperature considered) with a de facto
compensation for experimental scatter, but also allows additional information to be obtained by analyz-
ing extracts, for instance by GPC. The residue can also be recovered, dried under vacuum, and the BdR
650 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 11. Extraction kinetics curve for a polybutadiene compound; data from Leblanc and Stragliati [69].

content crosschecked by thermogravimetry analysis (TGA). The % bound rubber assessed through the
extraction kinetic method is quite reproducible (^1%) when the test material is well mixed.
The above methods are applicable only if the swollen rubber±®ller gel remains coherent during the
extraction process, which is generally the case providing the ®ller level is high enough, i.e. higher than or
close to the so-called percolation level, ,10±15 vol%, and the dispersion quality satisfactory, i.e. no
more agglomerates. When the swollen sample disintegrates, the extracted solution is no longer clear and
in extreme cases, it becomes very dif®cult, if not impossible, to recover the rubber±®ller complex.
KaraÂsek and Meissner [76] have observed that, at low ®ller concentrations, i.e. below or slightly above
the percolation level, there are some ®ller±polymer particles whose size is smaller than the mesh size of
the polyamide pouch used to contain the ®lled rubber samples during the extraction process. Such
solvent dispersed ®ller±polymer particles have colloidal sizes, color in black the extraction solvent
and are arbitrarily considered by these authors as belonging to the extractable polymer because their size
does not differ suf®ciently from that of the ®ller particles (aggregates). When they no longer obtained a
coherent mass during their extraction experiments at high temperature, Wolff et al. [61] centrifuged the
(colored) extraction solvent at room temperature and assessed the polymer content of the black sediment
using thermogravimetry analysis (TGA). They found that some polymer still remains on the carbon
black, even in the no coherent mass.
Another question of interest in considering bound rubber measurement techniques is the choice of the
solvent. Prior to 1970, benzene was used by a number of authors before it was banned for health safety
reasons, and replaced by toluene or xylene, considered less harmful.
The solvent effect on bound rubber measurement was investigated by Gessler et al. [38] with butyl
rubber compounds. They performed BdR assessment with benzene, cyclo-hexane, ethyl-benzene, cyclo-
hexene, p-xylene, chloroform and carbon tetrachloride. The solvent effect was found very small. Indeed
with respect to a mean bound rubber content of 36.2% across the seven solvents, the average deviation
from this mean was just 5.8%. While this deviation is higher than the standard test precision with one
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 651

solvent, i.e. ^2.0%, these authors concluded that there is no signi®cant solvent effect. This conclusion is
largely supported by others works [57,68] and, in the author's laboratory, using a slow evaporating (and
low cost) solvent such as toluene became current practice as no difference was found when using either
toluene, cyclo-hexane or carbon tetrachloride. With silica ®lled compounds, performing bound rubber
measurements with mixtures of solvent that contains either organic acid [73] or ammoniac [61], allows
to selectively destroy silanol bonding and therefore to obtain additional information.

5.3. Bound rubber formation

Preparing a ®lled rubber compound is a complicated process that depends on a considerable number of
variables, as re¯ected by the complexity of industrial mixing lines. The interactions between rubber and
®ller particles are obviously initiated in the early stage of mixing but it is known for long that the process
is not instantaneous and has a kinematical character. Indeed, for rubber±®ller interactions to take place,
one must have ®rst a good wetting of particles by polymer chains, a physical process that is obviously
concomitant with the dispersion and distribution of ®ller aggregates. Whilst not interpreted in this
respect, earlier results by Cotten [77,78] show that bound rubber increases with mixing time until a
plateau value is reached, as illustrated in Fig. 12. Leblanc et al. [79] performed similar experiments with
an instrumented mixer and expressed their results versus speci®c mixing energy, a more universal
parameter than mixing time. As shown in Fig. 13, with a standard carbon black ®lled SBR compound,
a minimum of about 1800 MJ m 321 energy is needed before a fully developed rubber±®ller network is
obtained. One notes also that the extrapolation of bound rubber data towards zero mixing energy would

Fig. 12. Bound rubber formation during mixing; experiments of Cotten [77,78].
652 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 13. Bound rubber formation; experiments with an instrumented mixer by Leblanc et al. [79].

give negative values, thus indicating that there is an induction process, probably related to the break-
down of agglomerates and the wetting of fragments (i.e. the aggregates) by the rubber.
In discussing bound rubber formation, one cannot therefore dissociate the development of rubber±
®ller interaction from the complex events that take place during rubber mixing. A number of phenom-
ena, either purely physical or physico-chemical or even chemical reactions, have been identi®ed as
concomitant events occurring within a mixer when the various ingredients of a rubber formulation are
being intimately dispersed. For the sake of discussion, one can neglect the role of minor ingredients (on a
weight basis) and, in a ®rst approximation, pay attention only to the effects concerning the dispersing
matrix, i.e. the rubber, and the material to be dispersed, i.e. the reinforcing ®ller. It is reasonable to
consider that certain elementary events are sequential. For instance, it is obvious that particle wetting
somehow occurs before dispersion, and with respect to relevant ®ller dimensions (see Fig. 8), agglom-
erates must be reduced to component aggregates before the latter are randomly dispersed throughout the
rubber matrix, whilst completely uniform dispersion is never to be expected, of course. Rubber±®ller
interactions, as evidenced by bound rubber formation, can occur only when there is suf®cient particle
wetting and, as agglomerates are progressively downsized into component aggregates, more and more
particle surface is becoming accessible to rubber segments for rubber±®ller interaction to develop.
Whilst there is surely a certain degree of concomitance, it seems logical to consider that wetting and
dispersion are elementary mechanisms likely faster than the development of topological constraints
between rubber chain segments and the very surface of ®ller particles. Not much is known about the
wetting mechanisms of a ®ller particle surface by a rubber macromolecule in the bulky environment of
an actual compound but, with respect to the operational conditions of practical mixing, one might expect
wetting to be either faster or subject to a similar kinetics as dispersion.
Several authors, through relatively simple macroscopic techniques, have studied ®ller dispersion
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 653

Fig. 14. Typical torque curve as obtained with a Brabender plastograph.

during rubber mixing. Cotten [77] used a Brabender EP2 mixer equipped with a cam type mixing head to
perform a series of experiments in an oil-extended butadiene rubber and considered how bound rubber
develops with respect to the so-called power curve, i.e. the plot of the torque vs. mixing time. He
obtained the usual two peaks power curve and, when performing optical dispersion assessments accord-
ing to ASTM D 2663 [80], combined with density measurements, he noted that 97% of the carbon black
was wetted by the polymer when the second power peak was reached, but most of the ®ller was then
present as agglomerates as large as 100 mm in diameter. Reduction of agglomerates into aggregates
occurred after the second peak when the power curve exhibited an exponential decay to some limiting
value. This prompted Cotten to ®t this part of the power curve with the following equation (see Fig. 14):
 
P…t† 2 P1
ln ˆ 2kt …8†
P2 2 P1
where P…t†; P2 and P1 are torque values at time t, at the second peak and after `in®nite' mixing
respectively, and k a rate constant, that is proportional to the rate at which carbon black is being
dispersed. Experiments in Brabender type of mixers are somewhat arti®cial with respect to factory
mixing but it is the mathematical form of Eq. (8) that is worth consideration. Indeed this equation
describes a process obeying ®rst order kinetics and implies that the rate of disintegration of agglomerates
is proportional to the quantity of agglomerates at any given instant of time.
Coran and Donnet [81,82] performed a series of experiments using a Brabender plastograph and a
very ingenious experimental approach. Using a special grip, they took tiny samples of material from the
mixing chamber at various times during the process. By observing the samples with an optical
microscope, they rated the state of dispersion on a 0±10 scale with respect to a series of reference
654 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

photographs (dispersion rating DR ˆ 10 corresponds to optimum dispersion). Then, using the rating
values, they demonstrated dispersion kinetic for carbon blacks in Natural Rubber, according to the
following equation:
 
a 2 kt
DR…t† ˆ DR1 2 exp …9†
n
where DR(t) and DR1 are the dispersion ratings, respectively, for a mixing time t and for an `in®nite'
mixing time, a a function of a delay time t0 before ®ller dispersion starts, n an empirical ®t parameter to
relate the percent undispersed carbon black to the dispersion rating and k the rate constant. Again such
experiments are laboratory scale ones, but Eq. (9) corresponds also to dispersion kinetic of order one and
was found valid for carbon black dispersion in other elastomers (SBR, EPDM, IR, IIR, BR) [83]. The rate
constant was found to depend on the ®ller grade (for a given elastomer) and to re¯ect the fact that the more
complex the black structure, the faster the dispersion process, as previously observed by Cotten [78].
One would expect similar conclusions to be drawn from experiments performed with larger mixers,
when considering the rate of energy consumption. Because instrumented large mixers are needed, such
experiments are rare but, for instance, Leblanc and Lionnet [84] recorded the speci®c mixing energy
versus time when preparing a series of EPDM and NR compounds in a Banbury mixer. They used an
upside down mixing procedure in order to be able to subtract the curve recorded when mixing the
un®lled formulation from the curve obtained with the ®lled one. The net mixing energy associated with
®ller dispersion was then obtained. As illustrated in Fig. 15, a net change of slope is observed when the
®ller dispersion process is completed (lower part of the ®gure), as re¯ected by the ¯at trace of the net
energy absorption rate curve (upper part).
This kinematical character of the ®ller dispersion and the associated formation of bound rubber hide
obviously a highly complex combination of various concomitant phenomena, still essentially ignored,
but likely very important for the full control of the mixing operation. Their study involves considerable
experimental dif®culties but, with the on-going progress in advanced analytical techniques (e.g. solid
NMR), interesting information might be expected in the future, as suggested by earlier results of Cashell
and McBrierty [85] showing that the rubber spin concentration varies with the number of cold two rolls
milling passes until a plateau level is reached.
Such a kinetic might not be valid with other types of ®llers, e.g. silica, for which ®ller±®ller dry
interaction, for instance hydrogen bonding, are known to be signi®cant. Silica±rubber interaction is a
much more complicated subject than in the case of carbon black, essentially because the surface
chemistry of silica is playing the signi®cant role. It is for instance remarkable that silica±SBR interaction
looks different when solution and emulsion rubbers are considered. It is likely that surfactant residue in
emulsion SBR are able to somewhat hinder the interaction between the ®ller surface and the rubber. To the
author's knowledge, no such study of ®ller dispersion kinetics has ever been made with other polymer±®ller
systems, and would particles wetting by the polymer be dif®cult or would strong inter-particles interactions
exist, more complex mathematical models than the equations above are likely to be needed. The
important result in Cotten's and Coran's works is that dispersion has a kinetic character, as aspect of
rubber mixing substantiated by practical experience, for instance time controlled mixing procedures.
In factory mixing, carbon black wetting, agglomerate disintegration, aggregate dispersion and bound
rubber formation occur concomitantly even if, from a conceptual point of view, such events are likely to
be somewhat subsequent. With respect to experimental works described above and common observation
in mixing room, one would reasonably consider that wetting and agglomerate dispersion are faster
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 655

Fig. 15. Power and speci®c mixing energy curves as recorded with an instrumented mixer; data from Leblanc and Lionnet [84].

processes that bound rubber formation. Accordingly, a purposely-simple model for carbon black
dispersion and bound rubber formation was recently considered [86].
This model, depicted in Fig. 16 is based on the simple consideration that, before rubber±®ller inter-
actions occur that lead to bound rubber formation, carbon black particles must be wetted by the rubber
and suf®ciently dispersed (i.e. agglomerates must be suf®ciently split into their component aggregates).
Wetting and dispersion on one hand, and bound rubber formation on the other hand are obviously
concomitant and very complex but, with respect to their likely different rates, it is attractive to model
them as a sequence of two consecutive, irreversible, ®rst-order processes, in agreement with experiments
by Cotten and Coran, as discussed above. A standard reasoning made in analogy with chemical kinetics
leads to the following equation for the dependence of bound rubber level on mixing energy:
 
k2 k1
‰%BdRŠME ˆ ‰%BdRmax Š 1 2 exp…2k1 ME† 1 exp…2k2 ME† …10†
k2 2 k1 k2 2 k1
where [%BdRmax] is the maximum bound rubber level that can be achieved during mixing for the
656 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 16. A purposely simple model for carbon black wetting and dispersion, and bound rubber formation.

formulation considered, ME the mixing energy (MJ m 23), k1 and k2 the rate constants of respectively
wetting and dispersion of carbon black aggregates, and bound rubber formation.
Fig. 17 shows that this model accounts well for the variation of bound rubber with increasing mixing
energy (data [79]; ®t parameters are k1 ˆ 1:285; k2 ˆ 1:306; BdRmax ˆ 17:6%). As expected, rate
constants k1 and k2 strongly depend on compound formulation. When the rate constants have close
values, wetting and dispersion are relatively slow processes and consequently the formation of bound
rubber is somewhat delayed. According to this model, quite high levels of mixing energy are needed for
the full development of rubber±®ller interactions. At ®rst sight, such levels of energy would appear quite
unrealistic in regard to operation of industrial mixers, but factory mixing is a full sequence of events, not
at all restricted to internal mixers, but further involving a number of necessary steps after dump, during
which signi®cant amounts of extra mixing energy are involved in material treatment.

Fig. 17. Modeling carbon black wetting and dispersion, and formation of bound rubber.
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 657

As discussed in Section 3, ®ller addition drastically changes the rheological properties of rubber,
namely the shear viscosity. Because of the kinetic character of ®ller dispersion and bound rubber
formation, one may expect the viscosity of a rubber formulation to strongly depends on the achievement
of rubber±®ller interactions, as indeed observed [77]. Despite the concomitance of events that take place
in the mixer, the optimum adsorption of rubber segments on ®ller aggregate surface is likely to be the
ultimate controlling one. Direct relationships between bound rubber level and rheological properties are
thus logically expected consequences.

5.4. Bound rubber and the morphology of uncured ®lled rubber compounds

Providing the ®ller content is above or near the percolation level, the bound rubber measurement
technique yields a swollen, coherent sample at the end of the extraction process. This clearly indicates
that rubber±®ller interactions lead to a tridimensional morphology in uncured ®lled rubber compounds.
Microphotographic evidences of such a three-dimension structure were published in the early 1970s
[87,88] and several authors used advanced Nuclear Magnetic Resonance (NMR) techniques to obtain
detailed information about how the molecular ¯exibility is modi®ed by the presence of ®ller particles.
Kaufman et al. [89] obtained proton relaxation time T2 results on high cis-polybutadiene and EPDM
compounds that demonstrated the presence of three distinct regions in the ®lled rubber, with different
degrees of molecular mobility exhibited by the constituent rubber chains. NMR measurements by
McBrierty and coworkers [90,91] essentially con®rm that, in a ®lled compound, there is a region of
unbound rubber with mobility comparable to the pure gum, bound rubber in an outer shell around the
carbon black particles which is somewhat less mobile, and an inner shell of tightly bound rubber which
experiences very limited motion on the T2 time scale, i.e. ,10 24 s.
Recent publications, using various NMR techniques, essentially con®rm that, in a well-dispersed,
uncured compound, isolated aggregates induce restrictions to the mobility of rubber chains, in such a
manner that three rubber regions can be distinguished. NMR relaxation experiments re¯ect the hetero-
geneity of ®lled rubber materials, with a fast decay in the 20 ms range that is assigned to the immobilized
part of the bound rubber shell. At longer times (above 100 ms), there is mobile fraction that contributes to
the relaxation and in between (in the 50 ms range), there is an intermediate mobility component. NMR
relaxation curves are decomposed into components, using the appropriate number of distribution
functions. With respect to Gaussian functions, non-linear least square algorithms are used in quantita-
tively treating normalized relaxation curves M…t† with the following equation:
" !2 #
X t
M…t† ˆ fi exp 2 …11†
i T2;i

where fi is the fraction of rubber having relaxation time T2;i and t the time.
For instance, Yatsuyanagi et al. [92] performed pulse NMR measurements at 100 MHz on a series of
SBR 1502 compounds with various levels of HAF (,N330 and ,N339) and SAF (,N212) and
analyzed the results with respect to bound rubber content (72 h. extraction in toluene at room tempera-
ture). By considering solid echo sequence, the proton spin±spin relaxation time T2 was obtained and
reduced in several components by ®tting the signal with a set of Weibull functions. Above 2308C, ®lled
compounds exhibit three components in T2, a long T2L, an intermediate T2M and a short one T2S that are
assigned to three different types of bound rubber. The un®lled compound shows a simple signal that
658 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 18. Bound rubber components from NMR experiments; data from Yatsuyanagi et al. [92].

corresponds to T2L and, as illustrated in Fig. 18, there are clear relationships between T2M and T2S and
bound rubber content. In agreement with other publications, the T2 components can therefore be
considered as a clear indication of three types of rubber in a ®lled compound: the extractable (i.e.
unbound) fraction and two types of bound rubber, with different molecular mobility. Other authors
[93,94], also performing NMR experiments, have essentially con®rmed such results with various rubber
formulations.
All together, the various experimental results discussed above suggest that the morphology of a
carbon black ®lled compound can be depicted as in Fig. 19 [95]. Very close to ®ller particles, there
is a thin layer of tightly bound rubber, which is likely to behave in a ¯ow ®eld exactly as the aggregate.
Then there is a region of loosely bound rubber, i.e. chains attached to the particles Ð through the tightly
bound rubber region Ð but able to undergo very large deformation during ¯ow. This region eventually
forms connective ®laments between rubber±®ller aggregates. The third portion is the unbound rubber,
so-called because it can readily be extracted from the uncured compound by a good solvent of the
elastomer. Obviously this rubber region does not interact with the ®ller particles.
Extractable rubber accounts for 70±90% of the gum elastomer of the compound, depending on the
formulation. The tightly and loosely bound rubber fractions are up to 30% of the gum. The connecting
®laments, which are readily seen on published microphotographs [87,88], ensure the coherence of the
swollen rubber±black gel in a good solvent. Silica ®lled rubber compounds would also ®t the model
described in Fig. 19. However, in case of silanisation, the rubber±®ller mesophase is obviously more
complicated but no clear information is yet available.

5.5. Bound rubber theory

Bound rubber can be viewed as a gel of (carbon) particles with the bonding agent consisting of the
longer polymer molecules. Such a consideration is the root of many theories, intending to explain the very
origin of the phenomenon, with the goal to derive mathematical models that would allow the bound rubber
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 659

Fig. 19. Pictorial description of the morphology of carbon-black ®lled rubber compounds.

content of a given rubber formulation to be calculated from a set of fundamental material parameters. One of
the earliest bound rubber theories is due to Villars [96] who considered a gel of ®ller particles bonded by the
largest polymer molecules. He assumed that rubber segments adhere to `elemental sites' on ®ller particles,
the area of which being de®ned as the area occupied by one adsorbing rubber segment. The size of the
elemental site is thus an adjustable parameter. The mathematical relationships obtained by Villars are rather
dif®cult to handle, and he reported a mean elemental area of around 30 nm 2 with natural rubber, that
would correspond to an adhering rubber segment of up to 800 isoprene units, i.e. about one-®fth of the
weight-average molecular weight. This theory was found inadequate by Kraus and Gruver [68] with
respect to results obtained with narrow-distribution polybutadiene compounds. However, based on the
observed dependence of bound rubber on the square root of M  w ; they calculated the molecular weight
distribution (MWD) of the (extractable) unbound rubber and obtained good agreement with GPC data.
In 1974, Meissner [97] considered bound rubber as the effect of random adsorption of structural units
of polymer on `reactive sites' assumed to exist on the ®ller surface. A major difference with the Villars's
approach is that the size of the adsorbed polymer segment is ®xed and identi®ed with the structural unit
of the polymer. No hypothesis was drawn as to the chemical or physical nature of such reactive sites and
the following equations were derived for the bound rubber fractions:for monodisperse polymers:
!
M w cSp
BdR ˆ 1 2 exp …12a†
A0 NAv
for random disperse polymers:
4
BdR ˆ 1 2 ! …12b†
 w cSp 2
M
21
A0 NAv
w
where BdR is the bound rubber fraction (g of bound polymer/g of total polymer in the formulation), M
660 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

the weight average molecular weight of gum polymer (g/mol), c the ®ller concentration (g/g of gum
polymer), Sp the speci®c surface area of ®ller (m 2/g), NAv the Avogadro number (6.023 £ 10 23) and A0 the
area of one active site on the ®ller particle (nm 2). The latter parameter is in fact an adjustable one since
Meissner derives its value from experimental data (i.e. bound rubber measurements). For instance, with
natural rubber±HAF black systems (M  w ˆ 2:6 £ 105 g mol21 ; Sp ˆ 85 m2 =g), A0 is found equal to
2
52 nm , whilst for SBR1500-various blacks (M  w ˆ 2:5 £ 105 g mol21 ; Sp ˆ 35±140 m2 =g), A0 is
2
found equal to 83 nm . From Kraus and Gruver data [68] on SBR-ISAF black systems (M  w ranging
5 21 2 21 2
from 0.5 to 4.2 £ 10 g mol ; Sp ˆ 115 m g ), Meissner obtained A0 ˆ 120 nm : In another publica-
tion [98], he calculated A0 ˆ 112 nm2 using the same data on BR and SBR compounds from Kraus and
Gruver, A0 ˆ 12 nm2 from Cotten's data on SBR compound with 50 phr N347 carbon black …Sp ˆ
92 m2 g21 † [99] and A0 ˆ 180 nm2 from Shiga's data [100] on EPR and EPDM compounds with
80 phr FEF black …Sp ˆ 44 m2 g21 †:
When derived from experimental results, the area A0 of one active site on the ®ller particle varies thus
in quite a wide range, i.e. 12±180 nm 2. It can be shown [101] that using either Eqs. (12a) or (12b), for a
given carbon black (i.e. constant Sp) at constant level (i.e. constant c), signi®cant differences in bound
rubber content are obtained when A0 varies from 0 to 60±80 nm 2 while M  w is below 300±400,000 g/mol.
With respect to the molecular weight of most industrial gum elastomers, A0 is thus the key parameter
about which, unfortunately, no clear meaning was provided in the original Meissner's theory. KaraÂsek
and Meissner [75] suggested that, for a given polymer±®ller system, A0 be calculated from:
Sp M0
A0 ˆ …13†
kNAv
where M0 is the molar mass of the polymer structural unit and k a constant. The latter is derived from the
application of the Flory's theory for polyfunctional crosslinking [102] to polymer±®ller gel systems, i.e.
interactions between ®ller particles and polymer chains result in a coherent gel like structure embedded
in the extractable elastomer. Therefore, the fraction of crosslinked units of polymer qcr is proportional to
®ller fraction c, i.e.: qcr ˆ kc: The proportionality constant k has, however, to be derived by ®tting
experimental data. For instance, k was found equal to 1.72 £ 10 24 and 1.68 £ 10 24, respectively, for NR
and SBR compounds with fume silica (Sp ˆ 50 m2 g21 ; average particle size: 40 nm; aggregate
structure). Accordingly A0 equal to 32.0 and 32.5 nm 2 for the respective compounds. In another publica-
tion, KaraÂsek and Meissner [76] confronted theory and experiments with a series of SBR 1502 …M w ˆ
5 21  5 21  5
2:36 £ 10 g mol † and NR (M w ˆ 3:25 £ 10 g mol and M w ˆ 1:40 £ 10 g mol † compounds 21

with either carbon blacks (N220, N330, N550 and N990) or fume silica. They found that in the low
and medium ®ller concentrations, several aspects are correctly predicted by the theory, namely the total
bound rubber but that the method of ®ller incorporation into the polymer can somewhat in¯uence the
values of the adjustable parameters of the theory. In SBR compounds, they derived A0 values equal to
62.5 nm 2 (N220), 46.3±67.6 nm 2 (N330), 50.5 nm 2 (N550) and 65.1 nm 2 (N990). With NR compounds,
the ®ller surface area per one reactive site A0 was found equal to ,20 nm 2 with fumed silica (Aerosilw
OX50).
An interesting aspect in Meissner's theory is the equation obtained for the molecular weight distribu-
tion of the unbound rubber, i.e.:

1
wE …y†dy ˆ w…y†e2qy dy …14†
1 2 BdR
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 661

where y is the number of structural units in a polymer chain, wE(y)dy the weight fraction of polymer
having y between y and y 1 dy, q the fraction of adsorbed structural units …ˆ M0 cSp =A0 NAv †: This
equation predicts that, with respect to MWD of the initial gum rubber, the one of extractable rubber
is shifted towards the lower molecular weight, and therefore the bound rubber is made up of the largest
rubber chains, as indeed observed [68,69].
With respect to silica±polysiloxane mixtures, Cohen-Addad [103,104] submitted interesting theore-
tical developments for systems in which the mechanism of polymer adsorption on ®ller particles is well
identi®ed as the formation of hydrogen bonds between oxygen atoms on the polymer chain and silanol
groups located on silica surface. Permanent gels are obtained and, whilst extraction experiments with a
good solvent demonstrate that signi®cant amounts of polymer are strongly adsorbed on silica particles,
the mechanism is clearly chemical [105]. By considering that polysiloxane chains obey Gaussian
statistics, Cohen-Addad derived the following relationship for bound rubber, i.e.:
p
M 0 cSp q
BdR ˆ a
M …15†
A0 ea NAv
where M  0 is the average weight of one skeletal bond (equal to 37 g mol 21 for siloxane), A0 the average
area (on the ®ller particle) associated with one hydrogen bond (equal to 0.55 nm 2 for silanol group),
ea < 1 a numerical factor accounting for chain stiffness and surface coverage, and M n the number
average molecular weight of the polymer; c, Sp and NAv have the same meaning as in Meissner's theory.
Using this equation, one calculates, for instance, that a mixture of PDMS …M n ˆ 325 000† with 29 phr of
2 21
fume silica …Sp ˆ 150 m g †; has a bound rubber fraction of 0.46, a value reasonably close to the
experimental data (around 0.52). The above equation predicts that bound rubber increases with the
square root of the number average molecular weight of the initial gum polymer, as experimentally
observed by Kraus and Gruver with narrow-distribution polybutadienes [68]. The interesting aspect
in Cohen-Addad's work is the clear identi®cation of the anchor site (one silanol group) and of the
anchoring element (one siloxane group), which is permitted by the recognition of the chemical nature of
the adsorbing process. One notes however that the surface of the anchor site on silica particle is at least
100 times smaller than the reactive site in Meissner's theory (0.55 nm 2 versus 30±180 nm 2).
Cohen-Addad and Frebourg studied mixtures of low cis-1,4 polybutadiene (38% cis, 54% trans, 8%
vinyl) with various levels of N220 …Sp ˆ 115 m2 g21 † and developed an adsorption model to relate the
bound rubber fraction with the number average molecular weight of the gum [106]. In line with the
quantitative description already offered for PDMS±silica mixtures, they assumed that there are elemen-
tary interactions Ð of unspeci®ed nature Ð between monomer units of the polymer backbone and the
carbon black surface. Then they considered that any single polymer molecule can bridge only two
aggregates, whilst several elementary binding sites are involved in the interaction between one polymer
molecule and one aggregate. Eventually they established the following relationship for bound rubber:
 2  
nB  2n c2 1 2b nB  nc
BdR ˆ …1 2 2b† M M …16†
2C…n†NAv 2C…n†NAv
where b is the fraction of bound polymer chains involved in the bridging of aggregates (the other
fraction of bound chains are just dangling and partly desorbed), nB the number of elementary interaction
sites on the carbon black particle, C (n) the average number of monomer units involved in the binding of
one chain [note: n is obviously the average number of skeletal bonds in one chain]; M n ; c, and NAv have
662 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

the same meaning as above. Cohen-Addad and Frebourg used this equation to ®t their data with a master
curve, obtained b ˆ 0:35 and estimated that there are around 0.042 sites per nm 2 where polybutadiene
chains are strongly bound to carbon black, independently of the polymer molecular weight. One notes
however that such a value is notably low since, providing the actual law of adsorption obeys Gaussian p
statistics, the number of contact points of one chain with the ®ller surface should be proportional to M  n:
By considering that, in contrast with silica±PDMS systems where the interaction process is essentially
chemical, there is a broad distribution of the enthalpies of adsorption of monomer units in their poly-
butadiene±carbon black (N220) materials, the hypothesis was drawn that, would the enthalpy of
solvent±surface interaction be stronger than the enthalpies of adsorption, the washing process of ®lled
rubber mixtures goes along with a partial elimination of chains from the adsorbed layer on ®ller
particles. This was somewhat con®rmed by NMR analysis which showed that in the studied systems,
less than 0.001 monomer unit per nm 2 are involved in the polymer±®ller interaction process. In other
words, on average one monomer unit would be ®xed per 1000 nm 2 of carbon black surface, a result that
largely differs from the surface area derived from the application of Meissner's theory to bound
rubber data. Subsequently, Meissner and Karasek [107] critically reviewed several assumptions in
the Cohen-Addad and Frebourg theoretical treatment and noted that it fails to predict the
preferential adsorption of longer chains and the existence of a so-called `gel point', i.e. a critical
®ller content for a coherent swollen material to be obtained. They also demonstrated that the
experimental data obtained by these authors could be as well described by the Meissner's theory
and its developments.
In the author's opinion, most theories for bound rubber ignore, however, an important aspect: the
effect of the chemical nature of the elastomer: it is, indeed, common knowledge that, for a given type and
concentration of carbon black, bound rubber content is strongly dependent on rubber type [38,71,78].
Moreover, no consideration is given to the (known) mixing effects on bound rubber (see Section 5.3
above), and storage effects, as discussed below, are ignored.
If rubber±carbon black interactions have a pure physical nature, one could consider that they are
likely to be topological constraints exerted on chain segments by the appropriate (geometrical) elements
on the surface of ®ller particles. This simplistic view was considered by Leblanc [101] who assumed
that, from a strong interaction to occur, the surface topology of a given carbon black particle must locally
encounter the conformation of a chain segment equal to at least three structural units typical of the
elastomer considered. This would be possible, provided the polymer segment of structural units and the
®ller topological site have the corresponding reciprocal geometry, in the appropriate orientation, and at
the right time. During mixing, the probability of such favorable events is obviously quite high. Once this
topological interaction has taken place, it is quite obvious that, in order to release it, the free portions of
the chain must exert on the constrained units not only suf®cient stresses but also in the appropriate
direction, a process that would require quite high energy level to be statistically signi®cant for bound
rubber to vanish. The key point in this reasoning is that a logical estimate can now be offered for the size
of an `active' site on the ®ller particle, with respect to the dimensions of the monomer unit; A0 can for
instance be considered as 2±3 times the half lateral surface of a monomer unit, easily calculated with
C±C and C±H bond length. The Cohen-Addad approach for Silica±PDMS systems (Eq. (16) above) can
then be applied to carbon black±diene rubber systems since a calculated A0 value can be used, with quite
a good agreement, and an appropriate explanation for the low BdR values obtained with Ethylene±
Propylene Rubber compounds (i.e. longer segments, ,C24, must conform their morphology with the
particle surface, and event that has a lower probability).
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 663

5.6. Storage maturation effects on bound rubber

Considerations in Section 5.4 underlined an important aspect of factory rubber mixing: at dump, a
rubber compound is far from equilibrium in terms of rubber±®ller interactions. Not only because the
theoretical mixing energy level for full bound rubber formation cannot realistically be achieved in
practical operations, but also because at dump, the rubber matrix has been submitted to such high
stresses that a certain time is needed for relaxation to proceed towards an equilibrium state. If one
sees rubber±®ller interaction as being essentially a physical process, then bound rubber is at best the
adsorption±desorption balance of rubber segments on carbon black sites at a given time (and tempera-
ture). As long as an equilibrium is not reached, such an adsorption±desorption mechanism is likely to
further evolve while the compound is at rest. Therefore bound rubber is expected to vary upon storage
until an equilibrium value is reached, as indeed previously observed by several authors [38,60,71]. How
fast this equilibrium bound rubber is reached depends on the chemical nature of the rubber, on the
compound formulation, on the mixing and the dump compound storage conditions.
When viewing rubber±®ller interactions as topological constraints effects, it is quite obvious that
there will be a competition between short and long rubber chains during the adsorption±desorption
process. During mixing, owing to the strong ¯ow ®elds prevailing in the mixer, one considers that all
molecular fractions of the polymer have equal probability to have chain segments in close contact with
the appropriate sites on ®ller particles, in order to develop a topological interaction. After dump, when
the material is at rest, one could consider that segments from short chains are progressively displaced
from active sites by segments from long chains, leading to an increase in the average molecular weight of
bound rubber. Consequently, Leblanc [101] extended the Cohen-Addad's equation in order to include a
variation of bound rubber BdR versus storage time (at constant temperature), as follows:
 p q
M0 cSp  n …0† 1 ‰M n …1† 2 M  n …0†Š…1 2 exp…2bt†
BdR…t† ˆ M …17†
A0 NAv
where M0 is the weight of one skeletal bond, c the ®ller concentration (g/g of gum polymer), Sp the
speci®c surface area of carbon black (m 2/g), A0 the average area of one interactive site (nm 2), NAv the
Avogadro number, M  n …1† the number average molecular weight (g/mol) of chains involved
 n …0† and M
in bound rubber respectively immediately at dump and after an in®nite storage time, t the time and b a
kinetic parameter describing the storage maturation. Fig. 20 shows experimental data ®tted with Eq.
(17). As can be seen, the agreement is excellent and the effect of the chemical nature of the elastomer on
the storage maturation behavior is well taken into account.
The above modeling concerns storage effects at room temperature but one might expect adsorption±
desorption processes of rubber segments to be strongly affected by temperature and any other external
in¯uence, e.g. pressure, moisture, etc. Bound rubber re¯ects rubber±®ller interactions which, in addition
to ®ller content effects, strongly affect the ¯ow properties of rubber compounds. Consequently, for any
formulation, there are direct relationships between rheological properties and bound rubber content. For
instance, when bound rubber varies on storage, so does the viscosity, whatever are the manners the
variations are obtained. It is known for long that the bound rubber content of carbon±polymer master-
batches is increased by heating [38,99,108] as illustrated in Fig. 21. BdR variation on storage at room
temperature was modeled by Leblanc and Stragliati [69] with the following equation:
j p k
BdR…t† ˆ BdR0 1 …BdRf 2 BdR0 † 1 2 exp…2k t † …18†
664 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 20. Modeling storage maturation effects on bound rubber; data from Leblanc and Hardy [71], ®tted with Eq. (17).

where BdR(t), BdR0 and BdRf are, respectively, the bound rubber at time t, before heat treatment and
after an `in®nite' exposure time, and k a rate parameter. Some of the literature data used in drawing
Fig. 21 were ®tted with Eq. (18). As can be seen, the ®t is excellent, which suggests that the adsorption±
desorption mechanism discussed above remains valid whatever the temperature, with obviously
desorption favored at higher temperature.

6. Dimensional aspects in ®lled rubber compounds

6.1. Length scale of rubber±®ller interactions

Fig. 19 offered a representation of the likely morphology of ®lled rubber compounds, based on
information obtained through a number of investigation techniques as well as common sense observa-
tions. Whilst drawn with respect to carbon black ®lled materials, all the aspects of this pictorial
representation are applicable to other ®ller systems, providing that strong interactions develop (or are
developed) between the rubber matrix and the dispersed particles. As discussed in previous sections,
rubber±®ller interactions can be considered from a nanoscale point of view, which obviously allows to
somewhat understand the very origin of the peculiar rheological properties of rubber compounds. Flow
properties of materials, however, are worth consideration mainly from a macroscopic point of view, i.e. a
scale length that is several decades larger than the basic phenomena involved in bound rubber formation.
In order to really appreciate how local rubber±®ller interactions succeed in affecting ¯ow properties, it is
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 665

Fig. 21. Effect of heating on bound rubber content; data from Brennan and Lambert [108] (BR compounds), Gessler et al. [38]
(SBR compounds).

necessary to consider dimensional aspects or, in other words, to appreciate the length scale of the
phenomena involved.
The intermediate property between local rubber±®ller interactions and the rheological behavior is
evidently the soft tri-dimensional network associated with bound rubber formation, that is depending in a
signi®cant manner upon the surface activity and the speci®c surface area of the ®ller particles, plus
obviously the state of dispersion. As we shall see below, advanced NMR techniques allow certain key
dimensions in the bound rubber phenomenon to be assessed but, because material ¯ow involves
necessarily large strain deformation, one needs to capture the morphology of ®lled rubber compounds
from a broader point of view.
When considering the effect of dispersed (solid) particles on the ¯ow properties of ®lled polymer
materials, one would write the following general equality:
Pcpd ˆ Ppolym F…F; d; Ssp ; z; x_ ; T; ¼†

where Pcpd and Ppolym are the rheological property considered, respectively for the ®lled compound and
the dispersing (polymer) matrix, F the volume fraction of the ®ller, d the particle size, Ssp and z ,
respectively, the speci®c surface area and the structure of the ®ller, x_ the strain rate and T the tempera-
ture. In addition the type of ¯ow should be considered because there is no reason for a shear ¯ow or an
extensional ¯ow property to be affected in the same manner by the presence of a ®ller. Since the early
considerations by Einstein, who treated the problem of a dispersion of rigid spheres of equal diameter in
a Newtonian medium as involving only simple hydrodynamic interactions, there have been a number of
666 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

proposals for a suitable mathematical form of the function F, whilst none ever tried to consider all the
parameters.
In the context of the present review, it is interesting to note that, when the well known equation
derived by Guth et al. [109,110]for the (Newtonian) shear viscosity of suspensions, was found
unsatisfactory with ®lled rubbers, a modi®ed relationship was submitted by White and Crowder
[111], as an attempt to consider both the bound rubber effect and the anisotropy of ®ller particles, i.e.:
hcpd ˆ hpolym …1 1 a1 f FEff 1 a2 f 2 F2Eff 1 ¼† …19†

where h cpd and h polym are the shear viscosity (at constant rate of deformation), respectively, for the ®lled
compound and the rubber matrix, f the so-called anisometry factor (i.e. equal 1 for spheres), a1 and a2
adjustable parameters depending on the considered system and F Eff the effective volume fraction of
®ller, considered to be the sum of the ®ller fraction, plus the bound rubber fraction, i.e.: FEff ˆ F 1
FBdR : The bound rubber fraction is obtained from:
BdRrcpd
FBdR ˆ …20†
rgum Wcpd
where BdR is the bound rubber content (in percent weight of the gum content of the formulation), r cpd
and r gum the speci®c gravity of the compound and the gum rubber respectively, and Wcpd the overall phr
(parts per hundred rubber) of the compound. One recovers obviously the Guth et al. equation by using
a1 ˆ 2:5; a2 ˆ 14:1; f ˆ 1 and no bound rubber, i.e. FBdR ˆ 0: One notes that in Eq. (19) further terms
might be considered for a better agreement with experimental data, and that gum and compound are
implicitly considered as having the same shear sensitivity, an assumption somewhat in contradiction
with most observations (see Fig. 1). The important aspect in White and Crowder's proposal is however
that the bound rubber fraction contributes to the viscosity increase due to the ®ller, in other words the
adsorbed elastomer is regarded as a layer covering the surface of the ®ller particles, which results in a
larger `effective' volume of the latter and hence more important hydrodynamic effects.
Uncured ®lled rubber compounds exhibit peculiar ¯ow properties because, owing to strong rubber±
®ller interactions, they develop a tri-dimensional morphology. Such materials can consequently be
viewed as soft (®lled) networks, interpenetrated by the extractable, un®lled elastomer. Qualitative
details about bound rubber structure as obtained by NMR techniques (cf. Section 5.4) are surely
important in explaining the origin of the phenomenon but, in order to understand how this affects
rheological properties, a quantitative view is needed.
When analyzing elastomer±carbon black interaction and its effect on viscosity, Pliskin and Tokita
[112] developed the simple model of a shell of bound rubber around ®ller particles. They considered that
this shell has an average thickness DR0 and derived the following expression for bound rubber content:
   
FrSp
BdR ˆ DR0 fA 1 G 100 …21†
12F
where BdR is the percent bound rubber (versus the initial gum content in the compound), fA the fraction
of the total ®ller surface accessible to the gum rubber, F , r and Sp the volume fraction, the density
(1.8 g cm 321) and the speci®c surface area of the carbon black, respectively, and G the fraction of
insoluble rubber gel. For a series of compounds with the same rubber and different carbon blacks, a
plot of BdR vs FrSp =1 2 F is thus expected to be linear, and the slope of the line gives the bound rubber
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 667

Table 4
Average thickness of bound rubber layer for various elastomers

Elastomer Layer thickness DR0 (nm) Source

cis-1,4 polybutadiene 7.5 ^ 0.3 [112]


96% cis-1,4 polybutadiene [BR 1220] 6.6 ^ 1.3 [90]
40 cis/50 trans/10 vinyl BR 6.8 ^ 0.2 [112]
SBR (23% styrene) [SBR 1500] 5.4 ^ 0.2 [112]
SBR 1500 ‰ML100 ˆ 50Š a 6.0 ^ 2.0 [61]
SBR (15% styrene) 5.2 ^ 0.1 [112]
Emulsion BR b 5.5 ^ 0.3 [112]
Solution SBR 4.1 ^ 0.2 [112]
Butyl rubber 2.8 ^ 0.2 [112]
EPDM (ethylene±norbornene) 3.5 ^ 0.2 [112]
EPDM (1,4-hexadiene) c 3.5 ^ 0.8 [112]
EPDM (dicyclopentadiene) 2.3 ^ 0.4 [112]
EPR 2.7 ^ 0.1 [112]
a
Evaluated with 17 carbon blacks with different Sp and loadings.
b
17.9% gel content.
c
12.4% gel content.

shell thickness DR0, providing one makes a reasonable hypothesis about the value of fA (for instance, for
a perfect dispersion, one might consider fA ˆ 1). Table 4 gives the bound rubber average layer thickness
for various elastomers, as estimated by this method.
It is clear that the Pliskin and Tokita approach gives only a rough estimate of the bound rubber layer
®rstly because a uniform layer over the whole ®ller surface is assumed and secondly because the
possibility that rubber molecules are adsorbed on several (neighbor) particles is not considered. Conse-
quently, if the range of carbon black type and loading is large, a plot of BdR vs FrSp =1 2 F yields data
scattered between two lines of different slopes, as illustrated by Fig. 22. This suggests obviously
that the linear relationship expected by Tokita and Pliskin is depending on the level and
characteristics of the carbon black, and the particular value given to the parameter fA. It is
nevertheless remarkable that, whatever the rubber considered, the average bound rubber layer
ranges between 2.5 and 8.0 nm.
By combining NMR relaxation time measurements, i.e. T2,1 (inner region) and T2,2 (outer
region), with the Pliskin±Tokita approach, O'Brien et al. [90] estimated the average thickness
of the tightly and loosely bound rubber regions. For a linear 96% cis-1,4 polybutadiene (BR
1220) with SRF (,N774) carbon black at various loadings (20±100 phr), they obtained a mean
total bound rubber layer of 6.97 ^ 0.17 nm with a tightly bound rubber layer of around
1.33 ^ 0.46 nm. LuÈchow et al. [94] estimated the tightly bound rubber fraction from 1H NMR
relaxation experiments and used the following equation to calculate the thickness of the immo-
bilized rubber layer, i.e.:

f1 mrubb
DR1 ˆ …22†
mblack rrubb Sp
668 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. 22. Pliskin±Tokita plot for various carbon-blacks at different loadings, according to Wolff et al. [61].

where f1 is the immobilized rubber fraction, mrubb and mblack, respectively, the masses of rubber
and carbon black, r rubb the density of the rubber and Sp the speci®c surface area of the black.
They found DR1 in the 1.0±2.2 nm range for a series of E-SBR 1500 compounds with 50 phr
N220 type carbon black. Using data published by these authors and Eq. (22), one calculates that
for all bound rubber (i.e. tightly and loosely bound) the mean layer thickness is in the 1.9±
4.7 nm range. If one uses Yatsuyanagi et al. [92] data (i.e. tightly and loosely BdR fractions) and
Eq. (22), mean layers thickness values are obtained in the same range with, as expected, a clear
effect of the carbon black level and structure. The higher the ®ller content or the more complex
the aggregate structure, the thicker both the tightly and loosely BdR layers.
There is thus a convergence of experimental data to support the length scale features indicated in Fig.
19. The bound rubber shell increases the effective aggregate size by up to 7±8 nm and this shell consists
of polymer with two very different degree of mobility. Likely the same macromolecules are involved in
the tightly and loosely bound rubber, which means that segments of the same chain, close to ®ller
surface, are so constraint that they are in a near glassy state, whilst others have the same mobility as the
extractable rubber. The latter rubber fraction does not interact with the ®ller and accounts for a sizeable
part of the viscoelastic behavior of the material, but the origin of the strong non-linear viscoelastic
character of ®ller rubber compound must be sought in the tri-dimensional morphology whose existence
is asserted by the retained coherence of highly (solvent) swollen samples, even if NMR (and other)
techniques are not capable to distinguish the loosely BdR involved in the so-called `connective
®laments' from the soft shell surrounding ®ller particles.
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 669

6.2. Dimensional aspects in ®lled rubber compounds morphology

An uncured ®lled elastomer compound can thus be viewed as a system of interpenetrated networks.
Complex rubber±®ller units, connected by `®laments', self-organize in a soft, highly deformable
network that is interpenetrated by the extractable polymer whose rheological behavior is controlled
by entanglements. Simple considerations of the dimensional aspects of such a complex morphology
allow complementing this qualitative description.
As underlined in Section 4.1, the relevant dimensions in a well-dispersed compound are the sizes of
®ller aggregates and of the random coil. In addition, with respect to previous sections, the inter-aggre-
gate distance is an important element in describing the morphology of ®lled rubber compounds. Direct
observations of ®lled materials are now possible, thanks to new techniques such as the Atomic Force
Microscope, but essentially two-dimensional pictures are obtained that must be considered with respect
to rheological properties, i.e. volume ¯ow. Simple descriptions, based on common sense approxima-
tions, can be considered to capture the very dimensional aspects of the rubber±®ller morphology in ®lled
compounds.
Aggregates are objects with an intricate geometry, relatively well re¯ected by speci®c surface area
measurements, but their volume is a parameter more dif®cult to capture, all the more if the effective one,
involving the bound rubber, has to be considered. Measuring directly the volume of an aggregate is not
feasible and only estimations from two-dimensional aggregate projections (as obtained with Transmis-
sion Electron Microscopy) are possible, in agreement with the aggregate `¯oc simulation' by Medalia
[113] and its subsequent developments [114,115]. Most aggregates exhibit anisometry (typically
between 1 and 3) in such a manner that the best `envelope' would be an equivalent ellipsoid. However,
the measurement of typical dimensions for such ellipsoids would require stereoscopic electron
microscopy, not yet readily available, despite promising recent progress [36]. With respect to aggregate
two-dimensional projections, Medalia used the simpler equivalent spherical envelope and through the
appropriate semi-theoretical approach, he established the following relationship for the equivalent
sphere diameter:

DEqSph ˆ 1:432dp …1 1 2:139 £ 1022 DBPA†1:093 …23†

where dp is the mean diameter of elementary particles of the aggregate and DBPA is the di-butyl
phthalate absorption number (cm 3/100 g) [116]. The constants in this equation have been derived
from TEM observations on a large selection of carbon blacks, combined with several reasonable
hypotheses.
Using published data [36,117], the equivalent sphere diameters either measured or calculated through
Eq. (23) for a series of commercial carbon black are compared with so-called Stokes diameters, when
available, measured through disc centrifuge photo-sedimentometry by Patel and Lee [118,119] (Table
5). The Stokes diameter concerns the volume of gyration of a particle freely sedimenting in a ¯uid, and
can be considered as the (overestimated) maximum diameter of interaction of a single aggregate. As can
be seen, there is a certain coherence in the data with most reinforcing grades exhibiting aggregate sizes
in the 70±100 nm range.
A reasonable fraction, let say 0.5, of either the equivalent sphere diameter or the Stokes diameter can
consequently be used to estimate inter-aggregate distance in a ®lled compound, providing an appropriate
model is chosen for the distribution of particles in the volume of the material. The simplest approach is to
670 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Table 5
Equivalent sphere diameters of carbon black aggregates

Filler type Measured equiv. Calculated a equiv. Measured stokes Source


sphere diameter (nm) sphere diameter (nm) diameter (nm)

N110 70 99 85 [117±119]
N119 63 [117]
N166 79 [117]
N219 70 [117]
N220 79 117 108 ^ 7 [117±119]
N234 97 [118,119]
N242 83 [117]
N326 91 ^ 4 108 102 ^ 6 [36,118,119]
N327 75 [117]
N330 89 153 136 ^ 6 [117±119]
N339 111 ^ 7 150 113 ^ 17 [36,118,119]
N347 102 [117]
N351 121 185 126 [117±119]
N358 212 153 [118]
N440 90 [117]
N550 324 235 ^ 10 [118,119]
N642 186 ^ 7 [36]
N650 305 ^ 25 [36]
N660 260 ^ 16 313 267 ^ 23 [36,118,119]
N774 316 261 [118,119]
N990 782 436 [118,119]
a
Using Medalia's equation (Eq. (23)) and elementary particle diameters published in Refs. [2,3].

distribute aggregates, considered as spheres of equivalent volume, on a geometrically simple three-


dimensional lattice. Two basic situations are worth consideration, either a simple cubic or a face-
centered cubic model, the latter permitting obviously a more compact arrangement. As illustrated in
Fig. 23, arranging equivalent spheres on such lattices allows several inter-aggregate distances to be
calculated with respect to ®ller volume fraction (see Appendix C).
Let us consider for instance a standard SBR formulation 1 with various levels of carbon black below
and above the theoretical percolation level (i.e. F < 0:12; around 30 phr). By taking half the mean
Stokes diameter as an estimate of the equivalent sphere of in¯uence of a reinforcing grade carbon black,
all the possible inter-aggregate distances in either the cubic or the face-centered cubic lattices were
calculated, as shown in Table 6.
Such simplistic calculations correspond with the data of Patel and Byers [120] who found that, at
optimum loading for failure properties, the inter-aggregate distance is around 21 nm, among a number of
reinforcing grade blacks (i.e. N110 to N375), regardless of the elastomer system, SBR 1500 or SBR
1712. If one considers the standard SBR compound with 50 phr reinforcing carbon black (last column in
Table 6), one sees that inter-aggregate distances are in the 20±70 nm range. Mean radius of gyration of

1
SBR 1500: 100 (r ˆ 0.92 g/cm 3); carbon black: variable (r ˆ 1.84 g/cm 3); zinc oxide: s (r ˆ 5.57 g/cm 3); stearic acid: 3
(r ˆ 0.92 g/cm 3); processing oil: 5 (r ˆ 0.98 g/cm 3); trimethylquimolein: 1 (r ˆ 1.08 g/cm 3); N-idopropyl-N 0 -phenyl-p-
phenylene diamin: 1 (r ˆ 1.17 g/cm 3).
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 671

Fig. 23. Simple geometrical model for ®lled rubber systems; da is the diameter of a sphere equivalent to the ®ller aggregate.
Equations to calculate typical distances are given in Appendix 3.

molten polyole®ns (PE and PP) are in the 40±50 nm range [121] and one would estimate the same range
for an industrial diene elastomer (see Appendix B). Then, by considering a mean random coil diameter
of 50 nm, it follows that in most compounds, a sizeable portion of the elastomer is in contact with at least
two aggregates, thus forming bound rubber, whilst there is enough room for a few chains to be embedded
in the system, without contact with ®ller particles. The ¯ow behavior of such a system is thus not only
controlled by disentanglement of polymer chains, but also by the adsorption±desorption kinetics of
bound rubber. Conversely, data in Table 6 for ®ller volume fractions below or near the percolation level
clearly suggest that when ®ller particles are separated by distances similar or larger that the mean
random coil diameter, rubber±®ller interactions are less in¯uential on ¯ow properties, whilst never-
theless existing.

7. Bound rubber and ¯ow properties

Bound rubber is the macroscopic result of rubber±®ller interactions and the vehicle that bridges the

Table 6
Inter-aggregate distances in model systems

Equivalent sphere diameter: 50 nm


Filler volume fraction F ˆ 0:10 F ˆ 0:15 F ˆ 0:20
Cubic lattice
e (edge) (nm) 37 26 19
d (short diagonal) (nm) 73 57 47
g (large diagonal) (nm) 100 81 69
Face-centered cubic lattice
u (edge) (nm) 88 70 59
x (plan) (nm) 47 35 27
672 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

gap between effects taking place at nanometer scale and ¯ow properties. Direct relationships between
BdR and rheological quantities can therefore be expected and have been indeed observed. Leblanc and
Staelraeve [122] studied the variations upon storage at room temperature of Mooney viscometer data and
bound rubber for high cis-1,4 polybutadiene and natural rubber compounds with different levels of
reinforcing black. They observed linear relationships between either the so-called Mooney peak
or the ML(114) viscosity and BdR. Storage time affects all studied properties, and the nature of
the elastomer as well as the ®ller content have also signi®cant effects. The bound rubber was
clearly found more sensitive to storage time than is the Mooney viscosity, but it must be noted
that the rotor motion (at 2 RPM) induces some mixing within the sample; consequently the
rubber±®ller morphology is somewhat altered when the ML(114) reading is made; indeed at
this moment, the rotor has already made eight turns. The initial Mooney peak probes the
immediate response of the rubber±black network at the beginning of the test and quite logically
exhibits a stronger dependence on the storage time. Similar observations have subsequently been
made with a number of different ®lled rubber compounds. Fig. 24 presents a compilation of such

Fig. 24. Linear relationship between Mooney peak and bound rubber for carbon black ®lled rubber compounds; drawn with data
from Leblanc and BarreÁs [86] (BR, EDPM and NBR cpds), Leblanc and Staelraeve [122] (BR/N330 and NR/N330 cpds), and
Leblanc [123] (SBR cpds).
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 673

observations made in the author's laboratory [86,122,123]. For a given compound, the Mooney
peak and the BdR vary proportionally.
Recently, Schaal and Coran [124] studied the rheological behavior of a silica ®lled (60 phr
VN3 1 4.8 phr TESPT) solution SBR (25% styrene, 55% vinyl), using namely a variable speed Mooney
viscometer, capable to be operated in the 10 22 ±10 1 s 21 range. They considered the variation of
compound properties upon storage at 708C and, as shown in Fig. 25, they found also a direct relationship
between bound rubber and the Mooney peak measured at 0.052 s 21 (around two decades lower than the
standard Mooney viscometer). Despite the experimental scatter, it is quite clear that a linear equation
exists between both properties. Furthermore, these authors performed capillary rheometer experiments
in order to observe extrudate quality as a parameter sensitive to changes in mixing conditions, storage
time and temperature, and ®ller type. They found an excellent correlation …r 2 ˆ 0:92† between extrudate
roughness and the Mooney peak measured at 1208C and 0.052 s 21. If one considers that extrudate
fracture re¯ects an excessive elastic response of the material upon high shear ¯ow conditions, one
may expect extrudate roughness and bound rubber to be correlated since the latter is associated with
the soft three-dimensional morphology described in Fig. 19.
Such data highlight the very origin of the well-known processability variations of ®lled rubber
compounds, as daily experienced on the factory ¯oor. In rubber technology, mixing is the ®rst in a
series of operations whose purpose is to prepare quite a complex polymer system of various ingredients,
viscoelastic materials, powdery solids, low viscosity liquids and other chemicals. An adequate

Fig. 25. Relationship between Mooney peak and bound rubber for silica ®lled SBR compounds; drawn with data from Schaal
and Coran [124].
674 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

dispersion state of formulation ingredients must be achieved before the material is shaped and
vulcanized in order to fabricate parts with appropriate properties and performance. With respect to
the previous sections, it is quite clear that, when mixing the various ingredients of a ®lled rubber
formulation, in addition to dispersion, one needs also to promote and develop rubber±®ller interactions.
There are kinetics aspects in the development of such interactions, whatever they are, either essentially
physical in the case of carbon black or chemical with silane promoted silica. The complete development
of rubber±®ller interactions requires not only high levels of mixing energy, but also some resting time
under no or mild shearing/temperature conditions. This is achieved through the various steps of rubber
processing, and rubber compounds must be viewed as complex heterogeneous materials with the
capability to self organize in a well-de®ned morphology. This self-organization capability is thus the
macroscopic re¯ection of very strong rubber±®ller interactions taking place at nanoscopic scale.

8. Conclusions and perspectives

The ¯ow properties of ®lled rubber compounds arise from their heterogeneous nature and the strong
interactions that develop between their various ingredients. In focusing on rubber±®ller interactions, one
has obviously neglected to consider other effects due to minor (on a weight basis) ingredients. This was
deliberate of course, with respect to the scope of the present review, but it is well known that small
amounts of appropriate compounding ingredients, e.g. lubricants, can have tantamount effects on ¯ow
properties, for instance by promoting wall slippage.
In what rubber±®ller interactions are concerned, there is nowadays a body of evidences that support
the view that complex adsorption±desorption processes take place between the surface of ®ller particles
and rubber segments. Because rubber segments belong to macromolecules, such localized effects, whilst
occurring in the nanometer range, result in (relatively) long-range modi®cations in terms of ¯ow
mechanics. In this context, bound rubber appears as a key property of ®ller rubber compounds, obviously
an all-inclusive one, depending on the chemical nature of the elastomer, the size and structure of the
®ller, the mixing conditions and the storage time and temperature. At a given temperature and in rest
conditions (i.e. no shear), there is thus a rubber±®ller mesophase whose dynamics reaches at best an
equilibrium between adsorption and desorption of rubber segments on appropriate sites on ®ller
particles. This means obviously that, if the temperature is changed or the material is submitted to
shear, this equilibrium is likely to be displaced. This provides a qualitative explanation for many
singularities in the ¯ow properties of rubber compounds.
In quantitative terms, the effect of ®ller type and level on rheological properties is still far to be
conveniently modeled. The rubber±®ller morphology might indeed be considered as the very origin of
the strong non-linear viscoelasticity of ®lled materials, but the manner to capture such informations in
constitutive equations is still an unsolved (and essentially not addressed) question. Below the so-called
percolation level, one might expect advanced viscoelastic modeling, for instance the K-BKZ model as
modi®ed by Wagner, to allow adequate prediction of ¯ow properties, essentially because the rubber±
®ller interactions result in a larger effective ®ller volume fraction without long range effects imparted by
the connective ®laments that develop when the ®ller content level increases. Above the percolation
level, it is readily a soft three-dimensional network that is ¯owing when a suf®cient stress is applied. It is
obvious that ¯ow is possible only because there are substantial desorption±adsorption processes in the
rubber±®ller mesophase. One needs therefore to introduce some kinetic aspects for this mesophase
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 675

effect in the modeling of the ¯ow properties of ®lled rubber compounds, thus combining concepts and
ideas from continuum mechanics and molecular theories. A very challenging perspective, indeed.

Appendix A. Wall slip in capillary ¯ow: revisiting the Mooney approach

No subject is of longer standing in rheology than wall slippage and despite numerous contributions,
there is so far no easy method to assess in an undisputable manner, the wall slip rate, except for some
recent advanced techniques [125], unfortunately applicable only to transparent materials. The method
proposed by Mooney [22] some 70 years ago is still used nowadays by a number of laboratories; thanks
to its apparent simplicity, but experience shows that it does not yield meaningful results always,
particularly with ®lled rubber materials. A careful reading of Mooney's paper allows to avoid disap-
pointment or even misuses of this testing technique, because of the number of assumptions and hypoth-
eses, some explicit, others implicit, considered by the author when writing his paper.
With respect to Fig. A1, standard considerations of steady simple laminar ¯ow in a cylindrical duct
lead to the following equations:
Pressure gradient
DP P 2 P2
ˆ 1
2L L
Shear stress
rDP
srz ˆ
2L

Fig. A1. Simple laminar ¯ow in a cylindrical duct in case of wall slip according to Mooney.
676 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Wall shear stress


RDP
sw ˆ sRz ˆ
2L
Mooney (1931) considers that, in case of wall slip, the velocity has two components
v ˆ vg 1 vS …A1†
where vg is the slip velocity component and vs the shear velocity component.
The slip coef®cient b is de®ned as
v g ˆ sw b …A2†
Then
ZR 1
v ˆ sw b 1 srz dr
r h
where h is the shear viscosity: 2 h_ ˆ s=g and g the shear rate.
ZR r DP 1  ZR 1 
R DP DP
vˆ b1 dr ˆ Rb 1 r dr
2L r 2L h 2L r h

The output (cm 3/s) is obtained by integrating v over the duct cross-section, i.e.
ZR ZR  DP  ZR 1 
Qˆ 2prv dr ˆ 2p Rb 1 r dr r dr
0 0 2L r h

ZRDP ZR p DP ZR 1 
Qˆ p Rbr dr 1 r dr r dr
0 L 0 L r h
" 2 #R (" 2 #R )
DP r DP r ZR 1 ZR r 2 1
ˆp Rb 1p r dr 1 r dr ˆ 0
L 2 0 L 2 r h 0 0 2 h

DP DP ZR 1 3
Q ˆ pR3 b1p r dr
2L 2L 0 h
one has
R DP DP s s R
sw ˆ ) ˆ w ˆ and dr ˆ ds
2L 2L R r sw
then, by substitution and change of variable
R3 Zsw 1 3
Q ˆ pR2 sw b 1 p s ds …A3†
s3w 0 h
2
In his paper, M. Mooney used the concept of `¯uidity' w de®ned by dv=dr ˆ sw; where s is the shear stress, w is thus the
reverse of the (shear) viscosity: w ˆ 1=h:
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 677

Eq. (A3) is rearranged as

Q s b 1 Zsw 1 3
3
ˆ w 1 3 s ds …A3a†
pR R sw 0 h

When differentiating this equation with respect to 1=R; keeping s constant, one obtains:
 
Q
2 3
pR  ˆ vg …A4†
1
2
R

since sw b ˆ vg :
Eq. (A4) is identical to Eq. (8) obtained by Mooney in his 1931 paper 2 and immediately suggests an
experimental procedure to assess the wall slip rate vg. The volume output Q is measured from capillaries
of different radii while keeping the wall shear stress constant. According to Eq. (A4), any observed
changes in Q=pR3 are entirely due to slippage and are independent of the viscosity of the material,
Mooney concluded.
Eq. (A2) can be written in order to evidence a term 4Q=pR3 ; which is the wall shear rate in a tube for a
Newtonian ¯uid, i.e.

4Q 4sw b 4 Zsw 1 3 4vg 4 Zsw 1 3


ˆ 1 s ds ˆ 1 s ds
pR3 R s3w 0 h R s3w 0 h

The viscosity of a Newtonian ¯uid is a constant and consequently, one immediately obtains
" #s
4Q 4vg 4 s4 w 4vg s
3
ˆ 1 3 ˆ 1 w
pR R sw 4h 0 R h

The term sw =h is by de®nition the `true' wall shear rate i.e. corrected for wall slippage and by contrast,
4Q=pR3 ; is called the `apparent' wall shear rate. It follows that the above equation can be written in the
condensed form

4vg
g_ a ˆ 1 g_ w …A5†
R
Eq. (A5) is often used in literature to introduce the so-called Mooney method to assess wall slippage in
capillary dies but is not appearing explicitly in Mooney's paper. This is obviously of no consequence if
one keeps two important aspects in mind: (1) that the above development refers to a Newtonian ¯uid,
de®ned by h ˆ s=g_ ˆ cst ; and (2) that the wall slip is considered as being simply proportional to the wall
shear stress (cf. Eq. (A2)).
If the ¯uid is non-Newtonian, obviously Eqs. (A3) and (A3a), as well as the experimental technique
suggested by M. Mooney still apply, but Eq. (A5) is no longer valid because a correction for the non-
Newtonian character has to show-up in deriving the wall shear rate from the output and, in addition, a
possible non-linear dependence of vg on s w should be considered.
678 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Starting from Eq. (A1), one would indeed write that in case of wall slip, the overall (volume) output
has two components, one due to shear and another due to slip, i.e. Q ˆ Qslip 1 Qshear ; hence
ZR pR3 Zsw 2
Q ˆ pR2 v g 1 pr 2 g_ …r† dr ˆ pR2 vg 1 s f…s† ds …A6†
0 s3w 0
where the functional f …s† ˆ g_ expresses the shear stress±shear rate relationship for a non-Newtonian
¯uid, through the so-called (shear) viscosity function. With respect to the `apparent' wall shear rate,
de®ned as g_ a ˆ 4Q=pR3 ; and therefore needing a correction for both wall slip (if any) and the non-
Newtonian character of the ¯uid, Eq. (A5) is rewritten as follows:
4Q 4vg 4 Zsw 2
g_ a ˆ ˆ 1 s f …s† ds …A6a†
pR3 R s3w 0
Differentiating with respect to s w yields
Zsw
s3w 4s2w f …sw † 2 4 s2 f …s†3s2w ds
dg_ a 4 dvg 0
ˆ 1
dsw R dsw s6w
4 dvg 4f …sw † 3 4 Zsw 2
ˆ 1 2 s f …s† ds
R dsw sw sw s3w 0
or, with respect to Eq. (A5)
 
dg_ a 4 dvg 4f …sw † 3 4vg
ˆ 1 2 g_ 2
dsw R dsw sw sw a R
It follows that
4g_ w dg_ 4 dvg 3g_ a 12vg
ˆ 2 1 2
sw dsw R dsw sw sw R
where g_ w ˆ f …sw † or
1 sw 1 sw 3g_ a 3vg
g_ w ˆ dg_ a 2 dvg 1 2
4 dsw R dsw 4 R
By proceeding to the substitutions, dsw =sw ˆ dln sw ; dg_ a ˆ g_ a dln g_ a and dvg ˆ vg dln vg ; one
obtains
1 dln g_ a vg dln vg 3g_ a 3vg
g_ w ˆ g_ a 2 1 2
4 dln sw R dln sw 4 R
    …A7†
3 1 dln g_ a 3vg dln vg
g_ w g_ a 1 1 2 11
4 3 dln sw R dln sw
By noting dIn sw =dIn g_ a ˆ n (the so-called `¯ow index'), one has
   
3 3n 1 1 3vg 1 dln vg
g_ w ˆ g_ a 2 11
4 3n R 3 dln sw
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 679

or
   
3n 1 1 3vg 1 dln vg
g_ w ˆ g_ a 2 11 …A7a†
4n R 3 dln sw
where 3n 1 1=4n is the so-called Rabinowitch correction for non-Newtonian character. If the ¯uid is
Newtonian, n ˆ 1 and one recovers Eq. (A4) provided that dln vg =dln sw ˆ 1; in other terms, if and only
if, the wall slip rate vg remains simply proportional to s w, as explicitly considered by Mooney when he
de®ned his `slip coef®cient' b through Eq. (A2) above.
In fact the equality vg ˆ sw b is a very strong assumption that appears not supported by published
results, since a number of authors have either postulated or experimentally observed that a scaling
dependence of the form vg ˆ Asm w (with m . 1) is more appropriate (Fig. A2).
If this is the case, it obviously follows that dln vg =dln sw ˆ m and Eq. (A7a) is rewritten as:
 
3n 1 1 vg
g_ w ˆ g_ a 2 …m 1 3† …A8†
4n R
or
   
4n 4n…m 1 3† vg
g_ a ˆ g_ 1 …A8a†
3n 1 1 w 3n 1 1 R
with n ˆ 1 and m ˆ 3; one immediately recovers Eq. (A5).
If one considers with some authors [126] that m < 1=n is a reasonable assumption, then Eq. (A8a)
further simpli®es as:
 
4n 4vg
g_ a ˆ g_ w 1 …A8b†
3n 1 1 R

Fig. A2. Typical wall slip rate±shear stress function found experimentally.
680 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

which in fact is Eq. (A5) in which the so-called Rabinowitsch correction for the non-Newtonian
character has been applied to the (slip corrected) wall shear rate.
In summary, the non-Newtonian character of the ¯uid does not invalidate the Mooney method, as
expressed by Eq. (A4), which yields measurement of vg at constant wall shear rate. However, Eq.
(A5) should only be considered for Newtonian ¯uid when the wall slip rate is strictly propor-
tional to s w. In general, such conditions are obviously highly restrictive and Eqs. (A7) or (A8)
have to be preferred.
Filled rubber compounds exhibit a strong non-Newtonian ¯ow behavior and whilst not well docu-
mented, it is likely that their wall slip velocity is more than simply proportional to s w. Consequently, it is
not really surprising that the Mooney method for wall slip rate gives wrong results when used with such
materials.

Appendix B. Estimating macromolecular dimensions

The random coil is the most probable con®gurationp of a long chain molecule, whose
p typical dimen-
sions are the mean-square end-to-end distance r and the radius of gyration R 2g ; i.e. the root-
2

mean-square distance of the elements of the chain from its center of gravity. For linear polymers,
both quantities are simply related: r2 ˆ 6R 2g :
For a ¯exible chain of N segments of length b, e.g. polymethylene±(CH2)n ±, the end-to-end distance is
assessed using
p p
r2 ˆ b N …A9†
and the radius of gyration with
q p
r2
R 2g ˆ p …A10†
6
Diene elastomers are however chains with restricted ¯exibility due to conformations imposed by the
double bond and more complex equations must be used, as follows
" !#0:5
p 1 1 1 cos a 1 1 j cos2
a…1 2 j† 2
2
r2 ˆ bN 11 …A11†
2 1 2 cos a 1 2 j …1 1 j†2 1 2cos2 a…1 2 j†2
where a is the supplement to the C±C±C angle (i.e. 109828 0 ) and j the rigidity coef®cient, whose value
is taken as 0.5 by most authors. For large N, a good approximation is
p  1 0:5
2 2 1 1 cos a 1 1 j
r ˆ bN …A12†
2 1 2 cos a 1 2 j
or
p h i0:5
r2 ˆ 3b2 N

since cos a < 1=3 and j ˆ 0:5:


For polymethylene, a typical value for the segment length is 0.25 nm but the representative segment
for a diene elastomer is obviously longer. A fair estimation can be made with respect to the so-called
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 681

®ber period that can be either calculated from a plan model of the macromolecule or readily measured on
crystallized material. Let us consider the case of polybutadiene. The following data are found in
literature [127]:

cis-1,4 polybutadiene: ®ber period: theoretical: 0.915 nm, experimental: 0.89 nm


trans-1,4 polybutadiene: ®ber period: theoretical: 0.505 nm, experimental: 0.483 nm
1,2 polybutadiene: ®ber period (theor.): isotactic: 0.65 nm, syndiotactic: 0.51 nm

One could then consider the following mean values:

cis-1,4 polybutadiene: 0.903 nm


trans-1,4 polybutadiene: 0.494 nm
1,2 polybutadiene: 0.580 nm

Industrial butadiene rubbers exhibit the above conformations in various extents and a weighed
average of the above half ®ber period values is likely to provide a reasonable estimation of the
representative segment length bm ; using

fcis f f
bm ˆ x 1 y trans 1 z 12 …A13†
2 2 2

where fcis ; ftrans and f12 are the mean ®ber periods of the cis, trans and 1,2 con®gurations
respectively and x, y and z the fractions of such con®gurations. If for instance, one considers
a high cis-1,4 polybutadiene (i.e. 98% cis-1,4; 1% trans-1,4; 1% vinyl-1,2), the representative
segment length would be 0.448 nm.
The number of representative segments can be considered as equivalent to the mean degree of
polymerization, i.e.

PD ˆ Mn =M0 …A14†

where Mn is the number average molecular weight and M0, the weight of the monomer unit. A typical
commercial high cis-1,4 polybutadiene (e.g. NeoCis Br-40, Enichem Elastomeri) has the following
macromolecular characteristics [69]: Mn ˆ 142; 000 g=mol; Mw ˆ 493; 000 g=mol:
The number of representative segments is thus around 2630 (with pM0 ˆ 54 g=mol) and, using Eq.
(A12) withpbˆ 0:448 nm; the calculated end-to-end distance is r2 ˆ 39:8 nm; and the radius of
gyration is R g ˆ 16:2 nm:
2

The radius of gyration provides in fact an estimate of the spherical volume of space that a given
macromolecule needs to freely experience all its possible conformations. The above calculations show
that this spherical volume has a diameter of around 32 nm. However, this diameter must be considered as
a low estimate of the sphere of in¯uence of the macromolecule whose diameter is likely 1.5 times larger.
Fig. A3 shows how the random coil diameter (and 1.5 times this diameter) varies with molecular weight.
An average diameter of 50 nm is thus a reasonable representative dimension of a rubber macromolecule
in the context of rubber±®ller interactions.
682 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

Fig. A3. Effect of Mn on random coil diameter (Rg ˆ radius of gyration).

Appendix C. Equations to calculate possible inter aggregate distances in model systems

Cubic lattice
0 r  1
3 p
 
B 6 C
e ˆ da B p 
 2 1 C < da 0:806
p 
 2 1
@ 3 A
F 3
F
0 s
p 1
3 p 2
B C  
B C
d ˆ da B p
3

 2 1C < da 1:140
p 
 2 1
B 3 C
@ F A 3
F

0 s
p 1
3 p 3
B C  
B 2 C 1:396
B C
g ˆ da B p3  2 1C < da p3  2 1
@ F A F

Face-centered cubic lattice:


0 r 1
3 2p
 
B 3 C
u ˆ da B p 
 2 1 C < da 1:279
p 
 2 1
@ 3 A
F 3
F
0 r
p 1
3
p  
B 3 2 C 0:905
B C
x ˆ da @ p3  2 1A < da p3  2 1
F F
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 683

References

[1] Medalia AI, Kraus G. Reinforcement of elastomers by particulate ®llers. In: Mark JE, Erman B, Eirich FR, editors.
Science and technology of rubber, 2nd ed. San Diego: Academic Press, 1994. p. 387±418 (Chapter 8).
[2] Donnet JB, Voet A. Carbon black physics, chemistry and elastomer reinforcement. New York: Marcel Dekker, 1976.
[3] Donnet JB. Carbon black. New York: Marcel Dekker, 1993.
[4] Leblanc JL. Flowing from art to science. Eur Rubber J 1989;171(10):33±5.
[5] Leblanc JL. Insight into elastomer-®ller interactions and their role in the processing behaviour of rubber compounds.
Prog Rubber Plast Technol 1994;10(2):112±29.
[6] White JL, Tokita N. Elastomer processing and application of rheological fundamentals. J Appl Polym Sci 1967;11:321±
34.
[7] Dizon ES, Papazian LA. The processing of ®ller-reinforced rubber. Rubber Chem Technol 1977;50:765±79.
[8] White JL. Processability of rubber and rheological behavior. Rubber Chem Technol 1977;50:163±85.
[9] Nakajima N, Harrell ER. Analyzing steady-state ¯ow of elastomers. Encylopaedia of ¯uid mechanics, vol. 7. Houston:
Gulf Publ Co, 1988. p. 703±23 (Chapter 24).
[10] Leblanc JL. Modern approaches to rubber processing problems. Kautch Gummi Kunstst 1990;43(10):883±92.
[11] Montes S, White JL, Nakajima N. Rheological behaviour of rubber carbon black compounds in various shear histories.
J Non-Newtonian Fluid Mech 1988;28:183±212.
[12] Shin KC, White JL, Nakajima N. Extrudate character and post-extrusion shrinkage of rheologically characterized rubber
Ð carbon black compounds and their interpretation. J Non-Newtonian Fluid Mech 1990;37:95±108.
[13] Shin KC, White JL, Brzoskowski R, Nakajima N. Rheological behaviour and extrudate shrinkage of rubber±carbon
black compounds. Kautch Gummi Kunstst 1990;43:181±8.
[14] Song HJ, White JL, Min K, Nakajima N, Weissert FC. Rheological properties, extrudate swell, and die entry extrusion
¯ow marker experiments for rubber Ð carbon black compounds. Adv Polym Technol 1988;8:431±9.
[15] BarreÁs C, Leblanc JL. Recent developments in shear rheometry of uncured rubber compounds. Part 1: design, construc-
tion and validation of a sliding cylinder rheometer. Polym Testing 2000;19:177±91.
[16] Dick JS, Pawlowski H. Applications of the rubber process analyzer in characterizing the effects of silica on uncured and
cured compound properties. Rubb Div Mtg, ACS, Montreal, Canada 1996 (Paper no. 34).
[17] Payne AR, Whittaker RE. Low strain dynamic properties of ®lled rubber. Rubber Chem Technol 1971;44:440±78.
[18] Turner DM, Moore MD. The contribution of wall slip in the ¯ow of rubber. Plast Rubber Proc 1980;5:81±4.
[19] Turner DM, Bickley AC. The effect of shear history on the rheology of unvulcanised rubber. Plast Rubber Proc
1981;1:357±62.
[20] Ma CY, White JL, Weissert FC, Isayev A, Nakajima N, Min K. Flow patterns in elastomers and their carbon black
compounds during extrusion through dies. Rubber Chem Technol 1985;58:815±29.
[21] Montes S, White JL, Nakajima N, Weissert FC, Min K. An experimental study of ¯ow and stress ®elds in a pressurized
Mooney viscometer. Rubber Chem Technol 1988;61:698±716.
[22] Mooney M. Explicit formulas for slip and ¯uidity. J Rheol 1931;2:210±22.
[23] Geiger K. Rheologische charakteristierung von EPDM-Kautschukmischungen mittels Kapillarrheometer-Systemen.
Kautsch Gummi Kunstst 1989;42:273±83.
[23] Mourniac B, Agassant JF, Vergnes B. Determination of the wall slip velocity in the ¯ow of a SBR compound. Rheol Acta
1992;31:565±74.
[25] Leblanc JL, Navarat T. Assessing the wall slip rate of complex polymer systems with the Mooney method. 13th Int Cong
Rheol, Cambridge, UK, Aug 21±25, Proc 2000;3:52±4.
[26] White JL, Han MH, Nakajima N, Brsoskowski R. The in¯uence of materials of construction on biconical rotor and
capillary measurements of shear viscosity of rubber and its compounds and considerations of slippage. J Rheol
1991;35:167±89.
[27] Jepsen C, Raebiger N. Untersuchungen zum Wandgleitverhalten von Kautschukmischungen an einem Hochdruck-
Kapillar-Viskosimeter. Kautsch Gummi Kunstst 1988;41:342±52.
[28] Newman S, Trementozzi QA. Barus effect in ®lled polymer melts. J Appl Polym Sci 1965;9:3071±89.
[29] Cotten GR. In¯uence of carbon black on processability of rubber stocks II. Extrusion shrinkage. Rubber Chem Technol
1979;52:187±98.
684 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

[30] Cotten GR. In¯uence of carbon black on processability of rubber stocks IV. Kinetics of extrusion shrinkage. Rubber
Chem Technol 1979;52:199±206.
[31] Tanner RI. Recoverable elastic strain and swelling ratio. In: Collyer AA, Clegg DW, editors. Rheological measurement,
2nd ed. London: Chapman & Hall, 1988. p. 492±515 (Chapter 16).
[32] Cotten GR. Signi®cance of extensional ¯ow in processing rubbers. Plast Rubber Proc 1979;4:89±95.
[33] Nakashima K, Fukuta H, Mineki M. Anisotropic shrinkage of injection Ð molded rubber. J Appl Polym Sci
1973;17:769±78.
[34] Donnet JB, Custodero E. Ordered structures observed by scanning tunnelling macroscopy at atomic scale on carbon
balck surfaces. Carbon 1992;30:813±7.
[35] Donnet JB. Black and white ®llers and tire compound. Rubber Chem Technol 1998;71:323±48.
[36] Gruber TC, Herd CR. Anisometry measurements in carbon black aggregate populations. Rubber Chem Technol
2000;70:727±46.
[37] Medalia AI. Effect of carbon black on dynamic properties of rubber vulcanizates. Rubber Chem Technol 1978;51:437±
523.
[38] Gessler AM, Hess WM, Medalia AI. Reinforcement of elastomers with carbon black. Part IV. Interaction between
carbon black and polymer. Plast Rubber Proc 1978;3:141±56.
[39] Dannenberg EM. Primary structure and surface properties of carbon black Ð Part 1: a study of four ISAF type carbon
blacks of varying structure. Rubber Age 1966;98:82±90.
[40] Dannenberg EM. Primary structure and surface properties of carbon black Ð Part 2: properties of commercial carbon
blacks of varying structure and particle size. Rubber Age 1966;99:81±5.
[41] Wang MJ, Wolff S, Donnet JB. Filler-elastomer interactions. Part I: silica surface energies and interactions with model
compounds. Rubber Chem Technol 1991;64:559±76.
[42] Saint-Flour C, Papirer E. Gas-solid chromatography: a quick method of estimating surface free energy variation induced
by the treatment of short glass ®bers. J Colloid Interface Sci 1983;91:69±75.
[43] Conder JR, Young CL. Physico-chemical measurements by chromatography. New York: Wiley, 1979.
[44] Papirer E, Balard H, Vidal A. Inverse gas chromatography: a valuable method for the surface characterization of ®llers
for polymers (glass ®bres and silicas). Eur Polym J 1988;24(8):783±90.
[45] Wang MJ, Wolff S, Donnet JB. Filler-elastomer interactions. Part III: carbon black surface energies and interactions with
elastomer analogs. Rubber Chem Technol 1991;64:714±36.
[46] Darmstadt H, Roy C, Kalliaguine S, Cormier H. Surface energy of commercial and pyrolytic carbon blacks by inverse
gas chromatography. Rubber Chem Technol 1997;70:759±68.
[47] Darmstadt H, Cao NZ, Pantea DM, Roy C, SuÈmmchen L, Roland U, Donnet JB, Wang TK, Peng CH, Donnelly PJ.
Surface activity and chemistry of thermal carbon blacks. Rubber Chem Technol 2000;73:293±309.
[48] SchroÈder A, KluÈppel M, Schuster RH. Ober¯aÈchenaktivitaÈt von Furnaceruûen I. Bestimmung der Ober¯aÈchenrauheit
mittels statischer Gasadsorption, Monolagenbereich. Kautsch Gummi Kunstst 1999;52:814±22.
[49] SchroÈder A, KluÈppel M, Schuster RH. Characterisierung der Ober¯aÈchenaktivitaÈt II. Bestimmung der Ober¯aÈchenrau-
heit von Furnaceruûen mittels statischer Gasadsorption, Multischichtenbereich. Kautsch Gummi Kunstst 2000;53:257±
65.
[50] Zerda T, Xu W, Yang H, Gerspacher M. The effects of heating and cooling rate on the structure of carbon black particles.
Rubber Chem Technol 1998;71:26±37.
[51] Gerspacher M, Farrell CPO. Tire compound materials interactions. Kautsch Gummi Kunstst 2001;54:153±8.
[52] Barabasi AL, Araujo M, Stanley HE. Three-dimensional Toom model: connection to the anisotropic Kardar±Parisi±
Zhang equation. Phys Rev Lett 1992;68:3729±32.
[53] Heinrich G, KluÈppel M. A hypothetical mechanism of carbon black formation based on molecular ballistic deposition.
Kautsch Gummi Kunstst 1991;44:419±23.
[54] Vidal A, Hao SZ, Donnet JB. Modi®cation of carbon black surfaces, effects on elastomer reinforcement. Kautsch Gummi
Kunstst 2001;54:159±65.
[55] Twiss DF. The theory of vulcanisation. J Soc Chem Ind 1925;44:106±8.
[56] Fielding JH. Impact resilience in testing channel black. Ind Engng Chem 1937;29:880±5.
[57] Stickney PB, Falb RD. Carbon black±rubber interactions and bound rubber. Rubber Chem Technol 1964;37:1299±340.
[58] Kraus G. Reinforcement of elastomers by carbon black. Adv Polym Sci 1971;8:155±237.
[59] Blow CM. Polymer±particulate ®ller interaction. Bound rubber phenomena. Polymer 1973;14:309±23.
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 685

[60] Dannenberg EM. Bound rubber and carbon black reinforcement. Rubber Chem Technol 1986;59:512±24.
[61] Wolff S, Wang M-J, Tan E-H. Filler Ð elastomer interactions. Part VII. Study on bound rubber. Rubber Chem Technol
1993;66:163±77.
[62] Kida N, Ito M, Yatsuyanagi F, Kaido H. Studies on the structure and formation mechanism of carbon gel in the carbon
black ®lled polyisoprene rubber composite. J Appl Polym Sci 1996;61:1345±50.
[63] Kolthoff IM, Kahn A. Sorption of GR-S-type rubber by varbon black I. Sorption from benzene solution by Graphon.
J Phys Colloid Chem 1950;54:251±6.
[64] Kolthoff IM, Gutmacher RG. Sorption of GR-S type of polymer on carbon black III. Sorption by commercial blacks.
J Phys Chem 1952;56:740±5.
[65] Jenckel E, Rumbach B. Formation of bound rubber of GRS-S type polymer with carbon blacks. Z Electrochem
1950;55:612±8.
[66] Sweitzer CW, Goodrich WC, Burgess KA. The carbon black gel complex in cold rubber reinforcement. Rubber Age
1949;65:651±62.
[67] Duke J, Taft WK, Kolthoff IM. Formation of bound rubber of GR-S type polymers with carbon blacks. Rubber Chem
Technol 1952;25:500±16.
[68] Kraus G, Gruver JT. Molecular weight effects in adsorption of rubbers on carbon black. Rubber Chem Technol
1968;41:1256±70.
[69] Leblanc JL, Stragliati B. An extraction kinetics method to study the morphology of carbon black ®lled rubber
compounds. J Appl Polym Sci 1997;63:959±70.
[70] French norm AFNOR T45-114. DeÂtermination du caoutchouc lie au noir de carbone, June 1989.
[71] Leblanc JL, Hardy P. Evolution of bound rubber during the storage of uncured compounds. Kautsch Gummi Kunstst
1991;44:1119±24.
[72] Roychoudhury A, De PP. Elastomer-carbon black interaction: in¯uence of elastomer chemical structure and carbon
black surface chemistry on bound rubber formation. J Appl Polym Sci 1995;55:9±15.
[73] Wu J, Shen Z, Hu D, Huang J, Chen N. Study on bound rubber in silicone rubber ®lled with modi®ed ultra®ne mineral
powder. Rubber Chem Technol 2000;73:19±24.
[74] Hamed GR, Hat®eld S. On the role of bound rubber in carbon-black reinforcement. Rubber Chem Technol 1989;62:143±
56.
[75] KaraÂsek L, Meissner B. Experimental testing of the polymer-®ller gel formation theory. Part I. J Appl Polym Sci
1994;52:1925±31.
[76] KaraÂsek L, Meissner B. Experimental testing of the polymer-®ller gel formation theory. Part II. J Appl Polym Sci
1998;69:95±107.
[77] Cotten GR. Mixing of carbon black with rubber I. Measurement of dispersion rate by changes in mixing torque. Rubber
Chem Technol 1984;57:118±33.
[78] Cotten GR. Mixing of carbon black with rubber: IV. Effect of carbon black characteristics. Plast Rubber Proc Applic
1987;7(3):173±8.
[79] Leblanc JL, Evo C, Lionnet R. Composite design experiments to study the relationships between the mixing behaviour
and the rheological properties of SBR compounds. Kautsch Gummi Kunstst 1994;47:401±7.
[80] ASTM D 2663. Standard test methods for carbon black, dispersion in rubber.
[81] Coran AY, Donnet JB. The dispersion of carbon black in rubber. Part I: a rapid method for assessing the quality of
dispersion. Rubber Chem Technol 1992;65:973±97.
[82] Coran AY, Donnet JB. The dispersion of carbon black in rubber. Part II: the kinetics of dispersion in natural rubber.
Rubber Chem Technol 1992;65:998±1015.
[83] Coran AY, Ignatz-Hoover F, Smakula PC. The dispersion of carbon black in rubber. Part IV: the kinetics of carbon black
dispersion in various polymers. Rubber Chem Technol 1994;67:237±51.
[84] Leblanc JL, Lionnet R. Determining the components of mixing energy when preparing rubber compounds in instru-
mented internal mixer. Polym Engng Sci 1992;32:989±97.
[85] Cashell EM, McBrierty VJ. An electron spin resonance investigation of carbon black-®lled polybutadiene. J Mater Sci
1977;12:2011±20.
[86] Leblanc JL, BarreÁs C. Bound rubber: a key factor in understanding the rheological properties of carbon black ®lled
rubber compounds. Rub Div Mtg, Am Chem Soc Chicago, Illinois, Paper no. 70, April 13±16, 1999.
[87] Kruse J. Rubber microscopy. Rubber Chem Technol 1973;46:635±785, see Figs. 29 and 30.
686 J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687

[88] Ban LL, Hess WM, Papazian LA. New studies of carbon rubber gel. Rubber Chem Technol 1974;47:858±94, see Fig. 3.
[89] Kaufman S, Slichter WP, Davies DD. Nuclear magnetic resonance study of rubber±carbon black interaction. J Polym Sci
A-2 1971;9:829±39.
[90] O'Brien J, Cashell E, Wardell GE, Mc Brierty VJ. An NMR investigation of the interaction between carbon black and
cis-polybutadiene. Macromolecules 1976;9:653±9.
[91] Kenny JC, McBrierty VJ, Rigbi Z, Douglass DC. Carbon black ®lled natural rubber 1. Structural investigation. Macro-
molecules 1991;24:436±43.
[92] Yatsuyanagi F, Kaidou H, Ito M. Relationship between viscoelastic properties and characteristics of ®ller-gel in ®lled
rubber system. Rubber Chem Technol 1999;72:657±72.
[93] Litvinov VM, Steeman PAM. EPDM-Carbon black interactions and the reinforcement mechanisms, as studied by low-
resolution 1H NMR. Macromolecules 1999;32:8476±90.
[94] LuÈchow H, Breier E, Gronski W. Characterization of polymer adsorption on disordered ®ller surfaces by transversal 1H
NMR relaxation. Rubber Chem Technol 1997;70:747±58.
[95] Leblanc JL. Insight into elastomer-®ller interactions and their role in the processing behaviour of rubber compounds.
Plast Rubber Proc Technol 1994;10(2):110±29.
[96] Villars DS. Studies on carbon black. III. Theory of bound rubber. J Polym Sci 1956;21:257±71.
[97] Meissner B. Theory of bound rubber. J Appl Polym Sci 1974;18:2483±91.
[98] Meissner B. Bound rubber theory and experiment. J Appl Polym Sci 1993;50:285±92.
[99] Cotten GR. In¯uence of carbon black on processability of rubber stocks I. Bound rubber formation. Rubber Chem
Technol 1975;48:548±57.
[100] Shiga S. Relationships between molecular structure and mixing mechanism. Rubber Chem Technol 1987;60:14±24.
[101] Leblanc JL. A molecular explanation for the origin of bound rubber in carbon black ®lled rubber compounds. J Appl
Polym Sci 1997;66:2257±68.
[102] Flory PJ. Principles of polymer chemistry. Ithaca, New York: Cornell University Press, 1953.
[103] Cohen-Addad JP. Silica±siloxane mixtures. structure of the adsorbed layer: chain length dependence. Polymer
1989;30:1820±3.
[104] Cohen-Addad JP. Sol or gel-like behaviour of ideal silica±siloxane mixtures: percolation approach. Polymer
1992;33:2762±7.
[105] Cohen-Addad JP, Touzet S. Poly(dimethylsiloxane)-silica mixtures: intermediate states of adsorption and swelling.
Polymer 1993;34:3490±8.
[106] Cohen-Addad JP, Frebourg P. Gel like behaviour of polybutadiene/carbon black mixtures: NMR and swelling properties.
Polymer 1996;37:4235±42.
[107] Meissner B, KaraÂsek L. Gel-like behaviour of polybutadiene±carbon black compounds. Polymer 1998;39:3083±6.
[108] Brennan JJ, Lambert DH. Rubber-black interaction in¯uence on cure level of vulcanizates. Rubber Chem Technol
1972;45:94±105.
[109] Guth E, Simha R. The viscosity of suspensions and solutions. III. The viscosity of sphere suspensions. Kolloid-Zeits-
chrift 1936;74:266±75.
[110] Guth E, Gold O. Hydrodynamical theory of the viscosity of suspension. Phys Rev 1938;53:322.
[111] White JL, Crowder JW. The in¯uence of carbon black on the extrusion characteristics and rheological properties of
elastomers: butadiene rubber and styrene±butadiene rubber. J Appl Polym Sci 1974;18:1013±38.
[112] Pliskin I, Tokita N. Bound rubber in elastomers: analysis of elastomer, ®ller interaction and its effects on viscosity and
modulus of composite systems. J Appl Polym Sci 1972;16:473±92.
[113] Medalia AI. Morphology of aggregates. I. Calculation of shape and bulkiness factors; application to computer-simulated
random ¯ocs. J Colloid Interface Sci 1967;24:393±404.
[114] Medalia AI. Morphology of aggregates. IV. Effective volume of aggregates of carbon black from electron microscopy;
application to vehicle adsorption and to die swell of ®lled rubber. J Colloid Interface Sci 1970;32:115±31.
[115] Medalia AI, Richards LW. Tinting strength of carbon black. J Colloid Interface Sci 1972;40:233±52.
[116] ASTM D2414. Standard test methods for carbon black, n-dibutyl phthalate absorption number.
[117] Ban LL, Hess WM. Current progress in the study of carbon black microstructure and general morphology, Coll. CNRS,
Le Bischenberg-Obernay, France, 1973, Sept. 24±26, Proceedings p. 81±115, Ed. CNRS (Paris), ISBN 2-222-01749-1.
[118] Patel AC, Lee K. Characterizing carbon black aggregate via dynamic and performance properties. Elastomerics
1990;122(3):14±8.
J.L. Leblanc / Prog. Polym. Sci. 27 (2002) 627±687 687

[119] Patel AC, Jackson DC. Carbon black characterization, Part 1: Effects of dynamic parameters on the behaviour of carbon
black in rubber. J.M. Huber Corp., poster presented at IKT'91, 1991, June 25, Essen, Germany.
[120] Patel AC, Byers JT. The in¯uence of tread grade carbon blacks at optimum loadings on rubber compound properties.
Rubber Ind 1982;34(4):9±13.
[121] Brandrup J, Immergut EH, Grulke EA. Polymer Handbook. 4th ed. New-York: Wiley, 1999. p. V.10 (PE) and V.22 (PP).
[122] Leblanc JL, Staelraeve A. Studying the maturation process of freshly mixed rubber compounds and its effects on
processing properties. J Appl Polym Sci 1994;53:1025±35.
[123] Leblanc JL. Practical rheology of rubber compounds. Rubber World 2001;223(6):28±33.
[124] Schaal S, Coran AY. The rheology and processability of tire compounds. Rubber Chem Technol 2000;73:225±39.
[125] MuÈnstedt H, Schmidt M, Wassner E. Stick and slip phenomena during extrusion of polyethylene melts as investigated by
laser-Doppler velocimetry. J Rheol 2000;44:413±27.
[126] Person TJ, Denn MM. The effect of die materials and pressure dependent slip on the extrusion of linear low density
polyethylene. J Rheol 1997;41:249±65.
[127] Champetier G, et al. Chim Macromol 1972:483±7 Tome II, Hermann Ed., Paris.

You might also like