I
A. G. VEITH
B. F.
Goodrich Research Center, Brecksville, Ohio
Oxidation Rate Measurements of Hevea Rubber Vulcanizates The oxidation behavior of typical vulcanizates is related to simpler olefins. Exceptions to similarity are attributed to type of cure or filler
IN
,
1946, Bolland (3)reported on oxidation behavior of the olefin, ethyl linoleate, and later such work on other low molecular olefins was continued (7, 2, 3, 4 ) . As a result, the mechanism for oxidation of olefins a t relatively low temperatures is fairly well known. The reactions proceed via a chain free-radical mechanism for which formal kinetic schemes have been devised. It is logical that oxidation of unsaturated polymers follow a pattern similar to that for simpler olefins; therefore, an investigation was undertaken to examine oxidation behavior of typical vulcanizates and relate this to known oxidation characteristics of monomolecular olefins. It is widely accepted that oxygen is responsible for the major degradation of polymers or rubbers. Therefore, variations in resistance to oxidation are directly applicable to production of rubber articles for extended life service.
Experimental All vulcanizates were prepared by standard milling techniques. Pale crepe rubber was used throughout, and cures were made between 1-mil aluminum sheets. All vulcanizate sheets were 0.016 to 0.020 inch thick (Table I). All rate measurements were made in an apparatus designed to measure volume of oxygen consumed a t constant pressure and temperature. This apparatus is similar to that described by Tobolsky (6) and Shelton and others (5). Pressure for all oxidations was 760 f 15 mm. of mercury except when volume readings were taken which were a t 760 mm. Water and carbon dioxide, evolved during oxidations, were absorbed by Ascarite and Dri-rite. Temperature control to &O.Io C. was obtained with a Thermistor control unit. A large aluminum block was used to contain sample tubes. T o ensure that authentic and nondiffusion limited rate measurements were obtained, sample thickness was kept within the limits set u p by Shelton and others (5). Plots of oxygen consumed us. time
were generally typical of an autocatalytic reaction-Le., convex to the time axis. At several points along such a curve, instantaneous oxidation rate was determined by measuring tangents a t these points; instantaneous rate us. extent of oxidation was then plotted.
General Discussion Bolland found that for ethyl linoleate, the experimentally determined rate and extent of oxidation were linearly related. This corresponds to Y
=
yo
+ k (ROOH)
(11
where Y = oxidation rate, ye = oxidation rate a t zero oxygen uptake, (ROOH) = hydroperoxide concentration, and k = a constant. Here, Y is the instantaneous rate and all reacted oxygen resides in ROOH groups. This equation, typical of an autocatalytic reaction, is applicable over a large extent of oxidation. Parameter ye is the intercept on the ordinate and corresponds to oxidation rate a t zero time or zero extent of oxidation. This is defined as the initial rate. I t was originally thought that in oxidaation of rigorously purified olefins, ro was a measure of rate for direct reaction of oxygen with the olefin. Recent work by Bateman and others (7) has shown that Y, for direct reaction with pure hydrocarbon cannot be determined accurately by this approach. In catalyzed oxidation of olefins which constitute a different system, the significance of ro is altered. I t represents initial rate of catalyzed oxidation, and extrapolation is not subject to the above criticism. Here, ro is a measure of the
Table 1. Pale crepe Zinc oxide Stearic acid Sulfur MBT
DPG MBTS Aldehyde-ammonia
rate at which the added initiator or catalyst initiates oxidation chains. In olefins not rigorously purified, the initial rate represents initiation of oxidation chains by the parent hydroperoxide of the hydrocarbon. Generally, in oxidation rate data for natural rubber vulcanizates, an equation formally identical to Equation 1 was followed : = r, k,(Oz), (2)
+
Here, Y and yo have their former significance; k, is defined as an autocatalytic constant, and (O?),is the amount of oxygen reacted or consumed by the polymer. Equation 2 represents data u p to approximately 1 mmole of oxygen per gram of polymer, which is almost complete degradation of the physical properties of an unprotected vulcanizate. (ROOH) of Equation 1 and ( O Q )ap, pear identical and by implication, all oxygen consumed by the polymer or vulcanizate is united in ROOH groups. T h a t a proportionality does undoubtedly exist between ROOH and (O&, does not mean that all oxygen consumed is in ROOH groups. Oxidation of A 1:5 polyolefins (squalene) is more complicated than some of the simpler olefins (5) and no equivalence between ROOH and (O?),can be claimed. It is felt that a certain proportion of oxygen molecules that react do reside in hydroperoxide groups and subsequent decomposition of these groups will yield chain-initiating radicals. The autocatalytic nature of the oxidation supports this. Figure 1 shows some plots of instantaneous rate us. extent of oxidation for a series of natural rubber vulcanizates. A
Formulations Used C D
A 100.0 5.0
100.0
5.0
5.0
2.0 3.0
2.0 3.0
2.0 3.0 0.7
0.5
... ... ...
€3
... ... ... ...
100.0
... ... ...
100.0 5.0 2.0 3.0
... 1.0 ... ...
VOL. 49, NO. 10
E
F
100.0
100.0 5.0 2.0
... ... 4.0 ... ... ... 1.0
OCTOBER 1957
3.0 ... ... 0.6 ...
1775
0.4
0.2
EXTENT
I .o
0.8
0.6
of OXIDATION (moles 0 2 / g . ) x IO
3
Figure 1. Rate vs. extent of oxidation for MBTS vulcanizates cured for 90 minutes at 140" C. Oxidation at 90" C. A. Gum vulcanizate 6. 20% by volume of HAF black C. 10% by volume of calcium carbonate
linear relation holds for both the filled and unfilled vulcanizates accelerated with mercaptobenzothiazyl disulfide (MBTS). Because vulcanizates cannot be considered pure hydrocarbons, 7 0 must indicate initial oxidation rate catalyzed by some active chemical species in the vulcanizate, the nature of which can vary with type of vulcanizate and loading filler. Therefore, history of a given vulcanizate will influence 70 which can be assumed to reflect previous treatment and nonrubber content, including fillers and other compounding ingredients, of the vulcanizate at the beginning of oxidation. Significance of k , is exactly the same as that for oxidation of simple olefins. I t defines the velocity at which the initial rate is accelerated by production of chemically active groups Tvhich can subsequently generate radicals to initiate oxidation.
Gum Vulcanizates The effect of state of curve was investigated by measuring the oxidation Table II.
rate of a MBT accelerated stock (Mix A) over a cure range of 10 to 160 minutes. A linear dependence was found for both 70 and k , us. combined sulfur (Figures 2 and 3). Correlation coefficients for these plots are 0.77 and 0.97 for ro and k , respectively. Experiments with added zinc sulfide gave evidence that organic combined sulfur was responsible for the behavior of 70 and k,. Oxidation behavior, however, of a vulcanizate cannot be predicted if the amount of combined sulfur is known. Although Table 11: listing effects of various accelerators on r0 and k,, shows a general parallelism between combined sulfur and values of ro and k,. there are excepIions. In general, the more active accelerators produce vulcanizates of lowered resistance to oxidation. Although combined sulfur is high for N-cyclohexyl-2-benzthiazyl sulfenamide (CBS), k , is not large in proportion. Thus, effects other than combined sulfur, influence oxidation behavior; combined sulfur, however, can be used to approximate oxidation susceptibility.
Effect of Various Accelerators on ro and k, at 100' Accelerator
Mix E, P.H.R.
Mercaptobenzothiazole (MBT) Mercaotobenzothiazvl disulfide (MBTS) (MBT) Diphenylguanidine (DPG) N-Cyclohexyl-2-benzylthiazyl sulfenamide (CBS) Tetramethylthiuram disulfide (TMTD) Tetramethylthiuram monosulfide (TMTM) Tellurium diethyl dithiocarbamate (TDEDC) Zinc dimethyl dithiocarbamate (ZDC) a
b
Mole 0 2 g . - l min.-1 Min.-l 40 min. at 135' C .
1776
INDUSTRIAL AND ENGINEERING CHEMISTRY
C.
%
X
ka X
Combined Sulfurc
0.5 0.5
2.2 2.2
6.3
4.7
1.58 1.23
0.51 0.I(
3.2
9.3
2.71
0.5 0.2 0.2 0.15 0.15
5.5 13.4 14.0 18.8 19.3
6.7 31.2 24.0 3.5 18.2
2.40 2.72 2.68 2.60 2.65
yo
Although not strictly a compounding variable, temperature of oxidation was investigated. 'Three gum vulcanizates were selected-Mix C , a typical mercaptobenzothiazole (MBT) mix, Mix D using diphenylguanidine (DPG) as the accelerator, and Mix E without metal oxide and cured with aldehyde-ammonia (AA) (a condensation product of acetaldehyde and ammonia), and sulfur only, Table 111. For diphenylguanidine, one of the exceptions referred to earlier, oxidation is not auto-catalytic but represented by a linear oxygen uptake-time curve; k , is zero and the initial rate prevails for at least the first millimole of oxygen per gram consumed. Diphenylguanidine has the greatest initial rate of all three. For aldehyde-ammonia, oxidation is strongly autocatalytic as shown by the high values for k,; its initial rates are slightly greater than for mercaptobenzothiazole which is the most resistant to oxidation with respect to both ro and k,. The effect of temperature on the two oxidation parameters is indicated in Tables I11 and IV. For mercaptobenzothiazole and aldehyde-ammonia the activation energies for yo are larger than those for k , by some 10 kcal. The mean value for AE,* for mercaptobenzothiazole and aldehydeammonia is 21 kcal. The value of AE,* for the benzoyl peroxide catalyzed oxidation of squalene ( a hexaisoprene) is 22.8 kcal. ( 6 ) . This amount of agreement between oxidation of isoprenic olefins and vulcanizates indicates that the sequence of steps in both systems could be similar and affected by temperature changes to the same extent. Effect of loading Fillers
For most technical applications, finely divided fillers are incorporated into the rubber, but relatively few general investigations into the effect of these common
i 0
08
T.
0 6
i -
00
1.0 PERCENT
e .o COMBINED
3.0 SULFUR
Figure 2. Initial rate r, vs. percentage of combined sulfur, cured for 10 to 160 minutes a t 135" C. ( a t 90" C.)
O X I D A T I O N RATE MEASUREMENTS Table 111.
Effect of Temperature on Oxidation Rates (Cured 100 minutes at 135' C.)
ro X 107 ka X
lo4
ro X 107 ka x 104
ro X 107 ko x 104
Table IV. Plot ro vs. l / T kavs. l/T
MBT DPG 800 c . 0.45 2.58 5.9 900 c. 1.50 6.3 10.5 1000 c . 4.90 16.2 23.8
AA
0.62 8.6
...
29.0
...
6.6 45.9
(Kilocalories) MBT DPG 32 25 19 .t
AA 32 22
loading fillers on oxidation rates have been reported. Van Amerongen (7) recently reported on the effect of some carbon blacks in natural rubber. Earlier Winn, Shelton, and Turnbull (8) reported on the effect of carbon blacks in GR-S. For this study several common inorganic or mineral fillers were selected and also several carbon blacks, both channel and furnace types. Surface areas were determined by nitrogen absorption measurements. I t was assumed a t the beginning that differences in oxidation behavior of loaded vulcanizates could arise principally from two sources: first, from chemical or physical nature of the filler
le
/
/
a
PERCENT
COMBINED
SULFUR
Figure 3. Autocatalytic constant k, percentage of combined sulfur, cured at 10 to 160 minutes at 135" C. (at 90" C.) v5.
Filler
Oxidation Rate Measurements at
Filler-Rubber Int.erfacia1 Area of 10 Sq. M. per Gram ro X ko x 107" 104b
Volume Loading, 20% Interfacial ro X ko X 107, 1046 Areac
CaCOa
1.45
14.1
24.4
CaSiO4.zHnO ZnO, fine ZnO, coarse Clay HPC EPC
1.80 1.1 0.70 0.95 2.75 3.30
1.7
47.8
1.60
0.0
8.0 5.9 5.7 5.8 3.8
15.9
1.25
6.05
67.7 54.8
5.6 5.4
4.2 5.0
SAF HAF
3.45 3.30
5.0 4.7
69.3 40.1
6.40 4.40
3.7 6.9
VFF FF MAF HMF SRF
3.75 3.35 3.20 3.20 3.40
4.0 5.2 6.3 5.5 4.9
39.1 25.0
4.70 4.15
5.3 5.0
18.5 12.3
2.90 3.35
6.2 6.1
Base vulcanizate, no filler
1.60
7.1, 6.8
...
...
.*.
*.
14.4
...
...
Activation Energies, AE"
10
Table V.
Moleg.-lmin.-l
Min.-l
90' C.
Effect of Varying Interfacial Area Interfacial krz X To x 104~ 107~ AreaC
5.1 10.8 17.2 24.4
... ... ...
11.6 24.4 38.6 54.8 93.8
...
8.45 17.8 28.2 40.1 68.6
... ...
... ... ...
1.80 1.70 1.40 1.40
13.0 13.2 12'.6 13.4, 14.4
... ... ...
... ...
2.50 2.80 3.30 5.60 10.1
...
7.0 7.5 7.8 7.0
...
... ...
...
... ...
2.25 2.75 4.25 4.70 7.80
... ... ... ...
8.6 8.5 5.6 4.7 6.7
... e . .
...
Sq. m.g.-'
surface, and second, from surface area alone. If surface area available to form an interface with the rubber is held constant, something about the nature of the surface can be learned. A series of measurements was made with all fillers a t an equivalent interfacial area per gram of rubber (assuming uniform dispersion). This value was selected as 10 sq. meters per gram of rubber. Mix F was used as a masterbatch for all filled vulcanizates. The time, 90 minutes at 140" C., was chosen to give a full cure and completely exhaust the postcuring capabilities of the vulcanizate. Table V lists values for r, and k, for all fillers studied. For mineral fillers except calcium silicate, initial rate is somewhat less than for the gum base vulcanizate (control) and their addition is beneficial in regard to ro. Also, k , is slightly depressed with an important exception-the value for calcium carbonate is greater than the control by a factor of 2. This filler, although not affecting the initial rate, enhances the autocatalytic character of the oxidation, and seems to assist in decomposing hydroperoxides or other active intermediates formed in the oxidation propagation step. In contrast, calcium silicate has a higher initial rate; but it is a more desirable loading
filler from an oxidation viewpoint because k , is smaller and, therefore, the initial rate does not grow large quickly. Initial rate for all carbon blacks is reasonably constant with a mean value of 3.30 x mole gram-' min.-'. Values for k , are also reasonably constant and their general level lower than that of the control vulcanizate. Trends of data in Table V can best be shown by Figures 4 and 5. Figure 4 is a plot of k , us. the interfacial surface area for all vulcanizates, and Figure 5 is a plot of r, us. interfacial surface area for carbon black vulcanizates only. Calcium carbonate has been omitted from Figure 4 because its k , is unique. Linear regression curves have been calculated and are fitted to experimental points in both figures. Both of these correlations are significant (Table
VI). No correlation exists between the volume of carbon black and r,. Table V I clearly indicates this. In this table 1-P is the Confidence Level for the calculated coefficient. Clearly both ro and k , are dependent upon interfacial area of filler. Interpretation of these findings can only be speculative a t this time; however, the general decrease in k , with VOL. 49,
NO. 10
OCTOBER 1957
1777
IO
0
1
0
- a
k ._
X
X
E
t
'6 x D
Y
4
I-
2
l
0-Furnace Black @ - Z i n c Oxide
0 -Clay
-
Summary
0 20
0
40 INTERFACIAL
Figure 4.
ao
60
SURFACE
AREA
mz ' : 9
Autocatalytic constant k, vs. interfacial surface area for oxidation a t
90' C. increased interfacial area may be due to the termination of free radical chains on the filler surface. This explanation suffices for all fillers except calcium carbonate which behaves differently. Increase in r. with interfacial surface area of carbon black may result from the condition of vulcanizates before oxidation. Van Amerongen (7) reported that ability to absorb and retain oxygen in both cured and uncured mixes varies directly with the surface area of the black.
I t is plausible to suggest that dissolved oxygen reacts with the rubber during the cure to produce peroxides or other chain-initiating species whose decomposition then causes r, to be greater. However, experiments failed to support this hypothesis. Mixed stocks were pumped at