Migration and Blooming of Waxes to the Surface of

Rubber Vulcanizates
S. H. NAH and A. G. THOMAS, Malaysian Rubber Producers' Research
Association, Brickendonbury, Hertford SG13 8NL, United Kingdom

Synopsis
When certain substances, notably waxes, are incorporated into rubber during vulcanization, the
surface of the vulcanized rubber may subsequently become covered by a film of the substance dif-
fusing out. This phenomenon, known as blooming, depends on the substance being soluble a t the
vulcanization temperature but only partially soluble at the temperature of blooming. A study has
been made using pure waxes in natural rubber vulcanizates with a range of crosslink densities. The
mass of bloomed material has been determined as a function of time, and the expected dependence
on the square root of the time has been found to hold over the anticipated range. It has been shown
that the kinetics of the process cannot be explained simply in terms of the degree of supersaturation
of the wax in the rubber, the observed rates being much too low. This appears to be related to the
precipitation of the wax in the body of the material. A theory has been developed based on a cal-
culation of the stresses set up around such a precipitated particle and the effect of the consequent
free-energy gradient on the rate of diffusion. Comparison with experiment shows satisfactory
agreement with the theory.

INTRODUCTION
It has been known for a long time that when certain substances, such as waxes,
are incorporated into rubber during the mixing process prior to vulcanization,
the vulcanized rubber sheet may, after some time, become covered with a film
of the substance. This phenomenon is known as blooming. The process of
blooming is of considerable practical importance as for example in the utilization
of wax bloom to protect the rubber against ozone attack. Although the basic
cause of wax blooming is known, its detailed mechanism is not well understood.
In this paper such a mechanism is proposed which explains the phenomenon
quite well quantitatively.

BLOOMING OF WAXES
The basic cause of wax blooming is derived from the fact that a t temperatures
above its melting point, typically 50 to 60°C, the wax is very soluble in the rubber,
but a t ordinary ambient temperatures the solubility is very much reduced. Thus
the wax dissolves readily in the rubber during mixing and remains in solution
during vulcanization; on cooling after vulcanization to below its melting point,
its solubility decreases abruptly and the wax becomes supersaturated in the
rubber. Some crystallization of the wax now occurs a t the surface of the rubber,
so that here the concentration of the dissolved wax is reduced to its equilibrium
value. This causes a concentration gradient to be set up in the rubber and wax
diffuses to the surface forming the characteristic bloom. Hence the basic
mechanism is diffusion produced by a concentration gradient which itself is
derived from the relative insolubility of the wax below its melting point.

Journal of Polymer Science: Polymer Physics Edition, Vol. 18, 511-521 (1980)
0 1980 John Wiley & Sons;Inc. 0098-1273/80/0018-0511$01.00
512 NAH AND THOMAS

This simple picture, however, is not adequate to explaiq the phenomenon

quantitatively as will be shown below. On the basis of this simple model, during
the initial stages of blooming the system may be treated as that of wax diffusing
out from a semi-infinite medium with the concentration of wax a t the surface
being given by the equilibrium solubility SO, and the rest of the rubber retaining
the initial concentration of wax wo in supersaturated solution. This leads to the
relation1
Mt/t1l2 = ~ ( W -
O SO)(D/T)'/~ (1)
where M t is the amount of wax bloomed per unit area after a time t , and D is the
diffusion coefficient of the wax in the rubber.
By using eq. (l),the rate of blooming M t l t 1/2 can be calculated and compared
with experimentally determined value. Such a comparison is shown in Table
I. The diffusion coefficient D, the solubility SO, and the amount of bloom were
determined as described in the Experimental section.
Table I thus shows that the simple theory leading to eq. (1)is incorrect, pre-
dicting blooming rates many times higher than observed. The rubbers used are
transparent, and visual observation of the samples indicates a reason for this
failure. On cooling it can be seen that wax precipitates out in the body of the
rubber so that the degree of supersaturation assumed in the simple theory cer-
tainly does not occur. This raises the question as to why, if precipitation occurs
in this way, any blooming to the surface occurs a t all. A mechanism for wax
blooming which will account for these observations is now suggested.

Model for Wax Blooming

I t is assumed that wax in excess of its equilibrium solubility precipitates out
as spheres around flaws or impurities in the rubber. As the wax crystals grow,
they have t o deform the rubber, producing a larger spherical hole; and hence a
pressure p will be set up in the wax. This stressed wax will be in local equilibrium
with the rubber in its immediate vicinity. In contrast, the rubber on the surface
will be in equilibrium with unstressed wax. Hence there will be a driving force
favoring the redissolving of the stressed wax and its reappearance on the surface
as unstressed wax. The following approach attempts to put this model on a
quantitative basis.
For the present purpose it is more convenient to consider the diffusion to be
produced by a free-energy gradient rather than a concentration gradient. This
can be done using the concept of the thermodynamic diffusion coefficient D T . ~
The flux F of diffusing substance is then given by

TABLE I
Rate of Bloomina of the Wax Docosane from Natural Rubber (NR) Vulcanizates
Docosane blooming rate
M t / t 1/2 x 106 ( g cm-2 sec-'l2) Calculated from
Vulcanizatesa Experimental eq. (1)
A 3.5 20.0
B 3.9 20.1
C 4.1 18.8
a For details about the rubber vulcanizates A,B,C, see the Experimental section.
 MIGRATION AND BLOOMING OF WAXES 513

where c is the concentration of diffusing liquid in mass per unit volume and d p l d x
is the rate of change of the chemical potential per mole with distance x; DT can
be identified with the ordinary diffusion coefficient D when the concentration
is small. The free energy difference associated with the transference of a
quantity of wax from an unstressed state into a stressed cavity against an elastic
stress p is given by
P=PV (3)
where V is the molar volume of the solid wax.
As diffusion progresses, the size of the particles near the surface will gradually
diminish and so the stresses on them will be relieved. If it is considered that there
are many crystals present, then the average free energy gradient will be given
by

and so from eq. (2), we have

F = -DT--cV d p (5)
RT d x
The pressure developed inside a spherical cavity in an infinite rubber block can
be shown to be given by3

where X is the extension ratio around the cavity and the rubber is assumed to
obey the statistical theory of rubberlike elasticity in having a stored energy
function of the form
W = (G/2)(X: + XZ, + XZ, - 3) (7)
where the X’s are the principal extension ratios and G is the elastic shear modulus.
If there are n seeding particles per unit volume each of initial radius ro then the
amount of wax present per unit volume of rubber is given by

where pw is the density of the solid wax, f l p w is the volume fraction of flaws
present, and S O is the solubility of the wax in the rubber. An apparent diffusion
coefficient D , can be defined such that

F=-D dw
-
a dx

and thus from eq. (5), we have

D, = D T - c V -dP
RT d w
Making use of eqs. (6) and (8),we obtain
so 7
D, = DT - v
2G(1/X2 + l / X 5 )
RT 3fX
In eq. (ll),c has been replaced by S O since the diffusing process takes place in
514 NAH AND THOMAS

the rubber matrix and it is assumed that the concentration of wax dissolved in
the rubber will be close to its equilibrium solubility value. Equation (11)shows
that the apparent diffusion coefficient depends on A, and hence it will be con-
centration dependent, the dependence being given implicitly by way of eq.
(8).
The diffusion equation that has to be solved with suitable boundary conditions
is
bw b
Dll -

This partial differential equation can be reduced to an ordinary differential

equation by introducing a new variable y such that1
y = x/2(At)lj2 (13)
where
A = 2D@oVG/3fRT
Using this substitution gives
d2X - 1
---[
dy2 X4+ X
(2x3 + 5) dX
-
(dy)
2

-2yh”(E)]
The above substitution can only be applied effectively for particular problems
of which the present one, a semi-infinite block with constant initial concentration
and a constant boundary condition, is the main example. At the free surface
( y = 0) the precipitated wax particles will have redissolved, the concentration
being solely due to the equilibrium solubility SO. The boundary condition here
is therefore X = 1. A long way from the surface the wax concentration will be
its initial value wo,which will be related to the value of A, A,, by eq. (8),thus

Since eq. (15) has to be solved numerically, beginning a t y = 0, an initial value

of ( d X / d y ) ,(dXldy)ohas to be assumed and the appropriate final asymtotic value
of X, X, found a t the end of the computation. Examples of the results of the
numerical integration of eq. (15) are shown in Figure 1which shows the relation
between X and y for different values of ( d Xldy)o. The corresponding values of
A, are also indicated in Figure 1. From such plots of X vs. y , the relation between
( d X/dy)o and A, can be found, and is shown in Figure 2. Thus, with the use of
eq. (16) and the curve in Figure 2, the value of ( d X / d y ) ocorresponding to a given
value of the initial concentration wo of wax may be obtained.
The rate of blooming is given by ( D , dwldx),=o. Substituting for D,, w, and
x using eqs. ( l l ) (8),
, and ( 1 3 ) ,respectively, and integrating with respect to t ,
gives the mass of bloom per unit area M t ,

Using eq. (171, it is thus possible to calculate the blooming rate if a n estimate
off can be made since all the other constants in it can be determined indepen-
dently. Fortunately the predicted blooming rate is remarkably insensitive to
the precise value o f f chosen. Thus, changing the value of f / p w from low2to
 MIGRATION AND BLOOMING O F WAXES 515

Y
Fig. 1. Theoretical relation between X and y for different values of (dX/dy)o.

increases the blooming rate by less than 20% for the rubbers considered here when
they contain about 13 parts of docosane. This uncertainty is less than that due
to probable uncertainties in the other parameters. The insensitivity to changes
in f arises from the fact that (dX/dy)o implicitly depends on f by way of eq. (16)
and the curve of Figure 2; and on substitution in eq. (17), some degree of can-
cellation occurs with the function off in the latter. A value off in the above range
is perhaps not unreasonable as the precipitation of the wax is assumed to occur
around flaws or impurities in the rubber and to are about the values
accepted for impurity concentrations.
The significance of the interfacial energy between the crystals and the rubber
is not immediately apparent. The wax bloom is not strictly in the form of a flat
sheet, but consists of crystals of some tens of microns in size. The size of the
crystals in the body of the rubber is not easy to estimate accurately but micro-
scopic observations suggest that it may be similar. Thus it is not certain in which
direction the interfacial energy will tend to make the wax migrate. An order of

A,
Fig. 2. Theoretical relation between ( d X / d y ) o and X, obtained by numerical integration of eq.
(15).
516 NAH AND THOMAS

2.5

~ 2.0

‘6
1.5
z

1.0

0.5

0 20 40 60 80 100 120
fI12/min”z

Fig. 3. Blooming of docosane in NR, vulcanizate A (01,B (A),C (0).

Initial concentration of
wax is 0.12 g cm-3.

magnitude estimate of the possible relative effects of interfacial and elastic

stresses can be made by comparing the elastic stress p [as given by eq. (6)] and
the term (2 X interfacial energy)/(radiusof particle). If the interfacial energy
is assumed to be of the same order as in simple liquids and is taken as 25 erg cm+,
then for a particle of radius 5 X lo-* cm, the interfacial energy term is about lo5
dyn cm-2. The elastic stresses are typically about 10 times or more greater than
this, and are therefore likely to dominate the process. A similar conclusion was
reached by Gent and Tompkins4 for the problem of cavitation in rubbers. Thus
in view of the uncertainty as to the relevance of the interfacial energy term in
the present problem and the indication that it may be relatively small in any case,
it has been neglected in the above analysis.

0 20 40 60 80 100 120
t”zlmin”2
Fig. 4. Blooming of eicosane in NR, vulcanizate A (01,B (A). Initial concentration of wax is 0.23
g cm-3.
 MIGRATION AND BLOOMING O F WAXES 517

t"'/min"'
Fig. 5. Blooming of docosane in NR vulcanizate B. Initial concentrations of wax are 0.03 g ~ r n - ~
( o ) 0.12
, g cm-3 (A),0.19 g crn-3 (0).
EXPERIMENTAL
In the present work pure materials were used rather than commerically
available waxes, and the paraffin waxes eicosane (C20H42) and docosane (C22H46)
were chosen. Their melting points were, respectively, 36.5 and 43.5"C. The
vulcanizates used were of natural rubber (SMR 5L) vulcanized with 1.0 ( A ) ,1.6
( B ) ,2.0 (C), and 3.0 ( D )parts of dicumyl peroxide (DCP) per 100 parts of rubber,
the vulcanization time being 10 min at 100°C followed by 60 min a t 160°C.
Before use, they were extracted in acetone for about 24 hr and then dried. The
sheets were 0.25 to 0.5 mm thick, and samples about 4 cm2 were generally
used.
The wax was incorporated by allowing the sample to absorb the desired amount
of wax by immersion in the molten material a t 60°C. T h e samples were then
removed from the wax but kept in the oven for about 72 hr a t 6OoC between
aluminum plates to allow the wax to distribute itself uniformly throughout the
rubber. The samples were quenched in cold water before the experiments began,
the plates effectively preventing blooming until they were removed from the
rubber.
The blooming experiments were carried out a t (22 f 1)OC. The amount of
bloom was found by periodically removing the wax which had bloomed to the
surface by means of scraping with a razor blade or using adhesive tape (Sellotape)
and measuring the weight loss. In this way it was possible to record the total
amount of wax that had come out as a function of time.
Eventually no more wax bloomed out, and the concentration of wax remaining
in the rubber then gave the solubility.
It was also necessary to determine the diffusion coefficients of the wax in the
518 NAH AND THOMAS

various vulcanizates. This was done by transfer measurements. I t has been

found5 that when a piece of rubber containing a nonvolatile liquid is placed in
initimate contact with a similar piece of rubber which has a different concen-
tration of the liquid, there is a transfer of liquid from the higher concentration
to the lower. If intimate contact has been properly maintained throughout the
rubber interface, measurement of the amount of liquid that is transferred as a
function of time enables the diffusion coefficient to be evaluated. I t is relatively
easy to achieve intimate contact between two rubber samples by clamping them
together with their major surfaces in contact. It is desirable to use a high pressure
initially to squeeze out any air bubbles but the pressure can then be reduced so
that the rubber is not significantly deformed. The diffusion coefficients de-
termined in this way have been found to be consistent with those found by other
technique^.^
For the present study, a sample which had finished blooming and therefore
had only dissolved wax in it was placed in intimate contact with a similar sample
which contained no wax. The mass M of wax per unit area of rubber surface that
diffused from the first sample to the second after time t was found by periodic
weighing. In accordance with diffusion theory, M was found to be initially
proportional to t1I2, as during the early stages the samples can be treated as
semi-infinite. From the slope of the M vs. t1/2 plots, the diffusion coefficient
can be obtained from the relation

where c , is the final concentration of wax in the rubber when transfer has ceased
and both samples have the same concentration.

RESULTS AND DISCUSSION

The results are plotted in the form of the mass of wax bloomed per unit area
of rubber Mt against the square root of the time t after removal of the aluminum
plates to initiate blooming. Some typical blooming curves are shown in Figures
3-5. In the initial stages M t is essentially proportional to t ll2 as would be ex-
pected for a diffusion-controlled process. I t should be noted, however, that this
proportionality would hold even if the diffusion coefficients were dependent on
concentration,' so that few conclusions can be drawn from this observation on
its own. A t sufficiently long times blooming ceases and the amount of wax re-
maining indicates the solubility in the rubber. It can be seen that in general the
more heavily crosslinked (harder) rubbers bloom more rapidly and the blooming

TABLE I1
Diffusion Coefficients and Solubilities of Docosane in NR a t 22°C
Diffusion coefficient Solubility
D X lo8 so x 102
Vulcanizate (cm2 sec-') (g cm-3)

A 3.4 2.4
B 3.3 2.3
c 2.9 2.3
D 2.3 2.0
 MIGRATION AND BLOOMING OF WAXES 519

TABLE 111
Diffusion Coefficients and Solubilities of Eicosane in NR a t 22°C
Diffusion coefficient Solubility
D X lo8 so x 102
Vulcanizate (cm2sec-') (E cm-3)
A 4.6 5.9
R 4.7 5.6
D 3.9 6.3

rate also increases with the concentration of wax. The results will be discussed
in more detail later.
The diffusion coefficients of the dissolved wax a t 22°C and their solubilities
in the rubbers used have also been obtained as described in the experimental
section and the values are presented in Tables I1 and 111. These tables indicate
that both the diffusion coefficients and solubilities vary only very slightly with
crosslink density for the rubbers studied. The moduli G of the vulcanizates A ,
B, C, D are, respectively, 2.26 X lo6, 2.78 X lo6, 3.74 X lo6, and 5.32 X lo6 dyn
cm-2.
From the initial slopes of the Mt vs. t lI2 blooming curves, the rate of blooming
M t / t can be found, and this experimentally determined rate can be compared
with that calculated from theory using eq. (17). In using eq. (17), the value of
the thermodynamic diffusion coefficient was taken as the ordinary diffusion
coefficient D since the concentration of the soluble diffusing wax was small. The
value of f / p w used to calculate the blooming rates was lop3.
A comparison between observed and calculated blooming rates is made in
Tables IV and V. The agreement, while not perfect, is probably little worse than
would be expected on the basis of the accumulated errors in the various param-
eters. Compared with the results shown in Table I the agreement has been
improved by an order of magnitude. The calculated results also show the correct
dependence of the blooming rates on the crosslink density or modulus and the
initial concentration of wax. The general agreement suggests that the basic
mechanism has been established over the concentration range studied. In
particular, the neglect of the interfacial energy as an important factor would

TABLE IV
Bloomina Rates of Docosane in NR Vulcanizates a t 22°C
Concentration Rate of blooming
of wax M t / t lI2 X lo6
wo x 102 (g cm-2 sec-1/2)
(g ~ m - ~ ) Vulcanizate Experimental Calculated
3 A 0.7 1.0
B 0.8 1.0
D 0.9 1.1

12 A 3.5 3.1
R 3.9 3.4
c 4.7 3.9

19 A 3.9 4.3
B 4.7 4.5
520 NAH AND THOMAS

TABLE V
Blooming Rates of Eicosane in NR Vulcanizates a t 22°C
Concentration Rate of blooming
of wax M t / t 1 / 2X lo6
wo x 102 (g cm-2 sec-'/2)
(a Vulcanizate Experimental Calculated
7 2.0 1.8
2.1 2.5
12 5.3 4.3
6.9 4.9
23 7.3 8.1
13.0 9.0

appear to be justified, as i t would lead for example to a lack of dependence on

modulus, contrary to observation. Some further experiments suggest that the
above theory will need elaboration to account for the behavior of very highly
crosslinked rubbers and high loadings of wax, in part because of the limitations
of the stored energy function, eq. (7),used here. It is hoped to treat these points
in a later publication.
The concentration distribution according to the theory is shown in Figure 6
for the cases illustrated in Figure 1,using eq. (8) and taking SO = 0.02 g cme3 and
flpw = 10-3. The strong concentration dependence of the apparent diffusion
coefficient predicted by the theory gives a steep region to the profile, approxi-
mating to a retreating boundary between the original and the surface concen-
trations. This is consistent with the visual observation that near the surface
the rubber becomes much more transparent, indicating a low wax concentration
with quite a sharp demarcation with the opaque crystalline region.

I I I I I

0 0.1 0.2 0.3 0.4 0.5

Y
Fig. 6. Theoretical concentration profiles for the wax in the rubber. The solubility of the wax
in the rubber is taken to be 0.020 g ~ m - y~ =, x/2(At)'l2.
 MIGRATION AND BLOOMING OF WAXES 521

CONCLUSIONS
T he model proposed in this paper appears to give an adequate explanation
for the magnitude of the blooming rates observed. The driving force is due to
elastic stresses set up around the precipitated wax particles and thus the
blooming rate is influenced quite strongly by the modulus of the rubber. On
the basis of this mechanism it would be expected that unvulcanized rubber would
show little blooming, and this is in accord with general observations on such
materials. The solubility of the wax is an important parameter and, because
of its sensitivity to temperature, the ambient temperature and the melting point
of the wax will be important factors influencing the blooming in practical ap-
plications.
The precipitation of the wax has been assumed to occur a t imperfections in
the material. The source of these is not yet clear, although impurities could
clearly be important. The values that have to be assumed for the volume fraction
of these seeding sites, about is not unreasonable in general magnitude for
impurity concentrations.
This work forms part of a program of research undertaken by the Board of the Malaysian Rubber
Producers’ Research Association.

References
1. J. Crank, The Mathematics of Diffusion, Oxford University Press, London, 1956.
2. G. S. Park, Trans. Furaday SOC.,46,684 (1950).
3. A. E. Green and W. Zerna, Theoretical Elasticity, Oxford University Press, London, 1954,
p. 104.
4. A. N. Gent and D. A. Tompkins, J . Polym. Sci., A-2, 7,1483 (1969).
5. E. Southern, Ph.D. thesis, University of London, 1969.

Received May 30,1979

Accepted July 26,1979