Computerized optimization of a white sidewall.

Compounds for tire white sidewalls are formulated to deliver a variety of properties including:

* A cure rate compatible with processing and cure of the green tire composite;

* hardness and stress/strain properties;

* adhesion to adjoining tire components;

* resistance to tear and cut/crack propagation;

* resistance to abrasion;

* resistance to degradation by oxygen and ozone;

* whiteness, both original and after UV aging; and

* lowest possible cost (ref. 1).

Typical formulations employ a blend of elastomers. Saturated elastomers such as halobutyl rubber (BIIR, CIIR) and terpolymers of ethylene-propylene with a diene rubber (EPDM) are used to impart oxygen- and ozone-aging resistance to the compound. They are used in combination with general purpose diene elastomers such as natural rubber (NR), cis-polyisoprene (IR), polybutadiene (BR) and its copolymers containing styrene (SBR). Titanium dioxide (Ti[O.sub.2]) is the whitening agent of choice for tire white sidewall compounds. Its chief limitation is the high cost per unit volume. Clay is normally used to reinforce the compound (ref. 2). Talc is used in many formulations as an extending filler and has been reported to increase the resistance of compounds to ozone degradation (ref 3). Compounds are normally cured with a blend of alkyl phenol disulfide (APSS) and sulfenamide or thiuram accelerators to balance flex fatigue and ozone-aging resistance with adhesion and cut resistance (ref. 4).

The simultaneous optimization of tire white sidewall properties is a complex task requiring the manipulation of large volumes of data and estimates of the reliability of predicted trends from the acquired data. This can reasonably be achieved only by the extensive use of statistical design and analysis using a computer.

Experimental

Rubber compounds were mixed using a two-stage mix in a laboratory internal mixer. All compounds were cured for a time equivalent to [T.sub.90] + 5 minutes in a platen press. Testing was carried out according to the procedures shown in table 1.

Table 1- test methods and equipment used
Rubber property   Test method        Equipment

Cure              ASTM D2084-87      MDR2000
Flex fatigue      ASTM D813-87       De Mattia
Rebound           ISO 4662-1986      Zwick 5109
Hardness          ISO 471            Zwick 5109
Stress/strain     ASTM D412-87       Instron 4204
Abrasion          ASTM D2228-88      Pico abrader
Tear strength     PPG CD-25-43       Instron 4204
Adhesion          ASTM D1876         Instron 4204
Whiteness         PPG BFC-101        Hunter D25-PC2
UV aging          ASTM D4329         Q-U-V
Ozone aging       ASTM D3395-86(a)   OREC

(a) - Dynamic conditions of 25% extension, modified to include
on/off cycling of ozone @ 50 pphm

The three rubber formulations shown in table 2 were used to establish a baseline performance for representative tire white sidewall systems. Formulation #1 is typical of a system used in bias passenger tires (ref. 5), formulation #2 is suggested for use in radial tires for original equipment applications (ref. 6) and formulation #3 is a low-cost compound that might be used on tires in an after-market application (ref 3).

Table 2 - tire white sidewall formulations
Formulation                            1       2       3
Ingredient (parts per hundred)
CIIR(a)                               20      55      50
EPDM(b)                               20      20      20
NR(c)                                 40      25      30
SBR(d)                                20       0       0
Titanium dioxide                      30      35      20
Talc(f)                                0      34       0
Clay(g)                               30      32      70
Stearic acid                           1       2       1
Petroleum wax(h)                       1     1.5     1.5
Paraffinic processing oil(i)           0       4       0
Ultramarine blue                     0.2     0.2     0.2
Zinc Oxide                            10       5       5
Rubbermakers sulfur                  0.5     0.8     0.5
APSS(j)                              1.3     1.3     1.3
Benzothiazyl disulfide (MBTS)(k)     0.8     1.0     0.8

Total                              174.8   216.8   200.3

(a) - Chlorobutyl 1066, Exxon Chemical Americas;(b) - Vistalon
2504, Exxon Chemical Americas;(c) - Natural rubber CV60;
(d) - SBR 1502, Ameripol Synpol Co.;(e) - Titanox 1 000, NL
Industries;(f) - Mistron Vapor, Cyprus Industrial Minerals Co.;
(g) - Nucap 290, J.M. Huber Co.;(h) - Sunolite 240, Witco-Organics;(i)
- Sunpar 2280, Sun Oil Co.;(j) - Vultac 5, Pennwalt
Co.;(k) - Thiofide, Monsanto Co.

All data requests were transmitted to the rubber compounding lab using personal computers (PCs) linked to a central VAX computer system. Lotus 123 templates were used to provide automatic calculation of weights needed for mixing and green rubber inserts for curing based on recipe, specific gravity of ingredients and batch factor, as well as default values for requested tests. Lotus 123 macro functions extract the specific gravity and other information such as list price (ref. 7) from a file which is maintained on the VAX.

Data from the Monsanto MDR2000, Instron 4204, Zwick 5109 and other rubber testing lab equipment not used in this study are acquired by PCs as digitized files which are saved in Lotus 123 files on the VAX. All other data are saved on the VAX as Lotus 123 files with data entered manually. Macros in Lotus 123 were used to create summaries of the variables studied and properties of interest. The extract-to function of Lotus 123 was used to create ASCII files which were then analyzed using PC-SAS software.

Results and disucssion

Designed compound studies

In order to investigate the compounding ingredients which control rubber properties, a series of [2.sup.kp] factorial screening designs (ref. 8) were performed. In addition to the variables shown in the formulations above, precipitated silica was used to replace clay in order to improve the tear strength and flex fatigue resistance of the compound. In order to provide efficient cure in the silica-filled rubber systems, diphenyl guanidine (DPG) was used in partial replacement of the APSS accelerator. The levels of variables studied are summarized in table 3. For maximum efficiency, the initial designs were [2.sup.7-4] fractional factorials (ref. 9), including a replicated midpoint, and were connected with common edge values from the three formulations. Each design was analyzed separately for the effect of each of the variables studied using the General Linear Models functions in SAS software. When no reversals (common terms with statistically significant but opposite effects at the 95% confidence level) were found, the common points from the designs were used to create designs with larger orthogonal arrays of data. This concept is illustrated in two dimensions in figure 1. A series of rotatable central composite designs were then carried out for variables which were found to be statistically significant in order to determine coefficients for use in a series of quadratic equations to predict rubber properties.

Table 3 - compound variables studied
Ingredient         Range studied, phr

CIIR                     20 - 55
NR                       25 - 40
SBR                       0 - 20
Ti[O.sub.2]              15 - 35
Silica(a)                12 - 35
Talc                      0 - 45
Clay                      0 - 70

Ultramarine blue        0.2 - 0.4
APSS                    0.5 - 1.3
MBTS                    1.0 - 1.8
DPG(b)                    0   0.9
Sulfur                  0.5 - 1.0

(a) - Hi-Sil 243LD, PPG Industries;(b) -DPG, Mosanto Co.

The data in table 4 show the compounding ingredients which influence the original and aged Hunter whiteness index of the white sidewall compound with their coefficients per phr used in the formulation.

Table 4 - effects on compound whiteness
Material                          Coefficient   [r.sup.2]

Original Hunter whiteness index

Ultramarine blue                     52.5          0.999
Ti[O.Sub.2]                           1.5          0.998
Clay                                  0.9          0.997
Silica                               -0.4          0.999
APSS                                -10.7           0.97

UV-aged Hunter whiteness

Ultramarine blue                     28.8          0.993
Ti[O.sub.2]                           1.8          0.993
Clay                                 -0.7          0.994
DPG                                  -100           0.98
APSS                                 -180           0.98
IR                                   -1.1           0.96

Table 5 is a summary of the compounding ingredients which had the most significant effects of the cure and physical properties of the white sidewall compound.

Table 5 - effects on compound physical
properties
Ingredient     Property              Coefficient   [r.sup.2]
Silica         T[S.sub.2]                   0.13       0.999
               [T.sub.50]                   0.23       0.999
               [T.sub.90]                   0.53       0.999
               Break strength               0.19       0.999
               Pico abrasion                0.60        0.95
               Tear strength                0.57       0.997
               Flex fatigue                 0.12       0.992

Total filler   MH                           0.03        0.98
               Hardness                     0.11       0.999
               Rebound                     -0.06       0.991
               Elong. to break             -3.01        0.95

APSS           MH                           4.10       0.999
               T[S.sub.2]                  -4.27       0.997
               [T.sub.50]                  -2.99        0.96
               [T.sub.90]                  -1.83        0.95
               Break strength               1.21        0.96
               Elong. to break           -227.54       0.999
               Ozone-aging res.             11.1       0.991

DPG            MH                           2.50       0.998
               T[S.sub.2]                  -4.79       0.999
               [T.sub.50]                  -7.00       0.999
               [T.sub.90]                  -9.79       0.999
               Break strength               5.12       0.999
               Elongation to break        -13.22        0.97
               Ozone-aging res.            -12.0        0.97

An example of the relationship between compound ingredients and rubber physical properties is shown in figure 2, which depicts the linear dependence of tear strength of the rubber compound on the silica level for formulation #3. The points indicated by solid squares depict the measured tear strength obtained by compounds formulated with direct replacement of clay by silica in the base formulation. The points indicated by open circles are all data containing silica, with varying levels of the other ingredients. The regression line (solid line) is calculated from the interactions of all variables for the direct replacement of clay with silica in the formulation.

Computer optimization

PC-SAS programming language was used to predict optimum formulations using the equations with interactions generated for the significant variables. Each of the variables studied was changed in increments which approximate the minimum observable difference for the ingredient over the entire range studied for the variables as shown in table 2. For example, curatives were varied in increments of 0.025 phr, fillers and polymers in increments of 1.0 phr and ultramarine blue in increments of 0.01 phr. From these variations over 300 million compound formulations were processed by the computer.

The predictions were first compared to the cure values measured by the MDR2000 for each of the three initial formulations. An example of a predicted rheometer curve which matches the specified values for formulation #3 is shown in figure 3. It should be noted that univariate statistical procedures were used. For matching cure properties to a rheometer cure curve necessary to fit a current factory process, multivatiate statistical procedures would allow the user to define a desired curve shape as well as the limits of the predicted cure curve.

Predicted formulations which matched the maximum torque (MH), time to two point rise in torque (T[S.sub.2]), time to 50% cure ([T.sub.50]) and time to 90% cure ([T.sub.90]) values within [+ or -]20% for each individual property, with at least three of the properties within 15% of the initial values, were transferred to the next algorithm. The second algorithm compared the stress/strain, hardness and rebound properties predicted by the equations with the initial compounds. Formulations which matched the target values within [+ or -]10% for breaking strength, elongation at break, modulus, hardness and rebound at 3 [degrees] and 100 [degrees] C were then transferred to the final algorithm.

The final algorithm selected formulations which had predicted improvements in at least one of the following properties with no detriments in any of these properties:

* Pico abrasion index;

* Hunter whiteness, original and UV-aged;

* tear strength;

* flex fatigue resistance;

* ozone-aging resistance; and

* cost.

Compound optimization

From these calculations several general strategies for producing a tire white sidewall compound with improved properties at minimum cost were evident. In order to reduce cost, the level of Ti[O.sub.2] should be the minimum level needed to obtain original whiteness and maintain aged whiteness, and the level of talc should be increased to as high a level as possible without reducing the break strength of the compound. In order to increase the resistance to tear, flex fatigue and abrasion, reinforcing precipitated silica was used as a replacement for clay. Data in table 5 and figure 2 show the effectiveness of this replacement strategy. Diphenyl guanidine was found to impart a number of advantages to the compound. It was the most efficient accelerator, particularly in the silica-filled compounds, having little effect on the scorch safety and resulting in large decreases in the [T.sub.90] values (table 5). DPG also had less pronounced deleterious effects on whiteness and elongation than the APSS accelerator (tables 4 and 5). An additional advantage of the diphenyl guanidine accelerator over several of the sulfenamide accelerators studied is that it does not generate volatile nitrosamine compounds during the cure cycle (ref. 10).

Typical predicted properties of an optimized compound based on comparisons to formulation #1 are shown in table 6. Also included are the actual physical property values measured for the compound which was mixed using levels specified in the optimized formulation. The increases in tear strength (145%) and pico abrasion (13%), coupled with both a decrease in cost of 9% and obtaining Hunter whiteness values which are equal-to or better-than formulation #1 are notable compound improvements. The excellent agreement obtained between predicted and measured rubber properties is evident in the values shown in table 6, where the only significant deviation from the predicted properties is the [TS.sub.2] scorch safety. The scorch value measured is slightly less than that predicted, but still significantly improved over that measured in formulation #1. For all formulations studied, the measured physical properties of the rubber compounds were within 15% of the computer-generated predictions. In most cases the values were within observed experimental reproducibility for the measurement of the property.

Table 6- computer optimization of form. #1

                         #1       Optimization
Ingredient
Clay                      30            0
[TiO.sub.2]               30           24
APSS                      1.3         0.9
DPG                       0           0.3
Silica                    0            12
Talc                      0            30
Ultramarine blue          0.2         0.4
Physical property                Predicted    Actual
MH, dNM                   9.51       9.96       9.78
[T.sub.90], minutes      13.80      12.89      12.96
[T.sub.50], minutes       6.32       7.32       6.95
[TS.sub.2], minutes       3.92       5.33       4.66
Break strength, kPa      14.66      15.12      14.93
Elongation to break, %   723         782        745
Pico abrasion index      38          44          43
Hunter whiteness index   60.0       60.98      61.73
Tear strength, N/mm       5.84      15.13      14.26
Cost/lb, $                 .93               0.85

Physical properties obtained from the compound mixed based on the predicted optimum of formulation #2 are shown in table 7. As observed previously, significant improvements were obtained in tear strength (38%) and whiteness (7% original and UV-aged) at an equivalent compound cost. An increase in flex fatigue resistance of 22% was also obtained using the optimized formulation.

Table 7- computer optimization of form. #2

                                    #2       Optimization
Ingredient
Clay                                32            0
[TiO.sub.2]                         35            27.5
APSS                                1.3            0.8
DPG                                 0              0.6
Silica                              0             17.5
Talc                                34            42.5
Ultramarine blue                    0.2           0.3
Property
[T.sub.90], minutes                 14.6          13.3
[TS.sub.2], minutes                  3.4           3.4
Hardness                              58            56
Pico abrasion index                   37            42
Hunter whiteness index, original      59.6         63.9
- UV aged                             20.1         21.6
Tear strength, N/mm                   12.8         17.6
Cut growth (100 kc), mm               10.1          7.9
Cost/Ib, $                            1.01         1.00

The properties obtained from the compound mixed based on a computer-optimized recipe derived from formulation #3 are shown in table 8. In this formulation the replacement of 70 phr of clay with 25 phr of reinforcing precipitated silica and 40 phr of talc yields a compound with a 325% increase in tear strength, a 39% increase in abrasion resistance and a 31% improvement in UV-aged Hunter whiteness index, coupled with a reduction in compound cost of 7%.

Table 8 -computer optimization of form. #3

                                   #3      Optimization
Ingredient
Clay                               70          0
[TiO.sub.2]                        20          15
APSS                               1.3         0.8
DPG                                 0          0.6
Silica                              0          25
Talc                                0          40
Ultramarine blue                   0.2         0.4
Physical Property
MH, dNm                            13.15      13.12
[T.sub.90], minutes                16.68      12.56
[T.sub.50], minutes                 7.24       6.39
[TS.sub.2], minutes                 3.54       4.40
Break strength, kPa                11.06      10.64
Elongation to break, %              507        684
Pico abrasion index                  18         25
Hunter whiteness index, original   63.64      68.24
 - UV aged                         29.43      38.50
Tear strength, N/mm                 5.53      24.83
Cost/Ib, $                          0.82       0.76

Predicted formulations can be targeted to have a broad range of specified properties. Targeted properties such as cure rates, stress/strain properties, tear strength, abrasion resistance, flex fatigue resistance, whiteness indexes and cost can be matched by the computer within seconds. Specific targeted properties can be emphasized by increasing the relative weight of the property in the overall regression strategy. Properties may be more precisely matched to the cure times of a tire production process or improvements in specific properties may be emphasized by the use of multivariate statistical procedures or using optimization strategies such as the desirability function developed by Derringer and Suich (ref. 11).

Summary

Using the computer and statistically designed compounding experiments, the factors which determine tire white sidewall performance have been investigated in detail. Regression equations to predict specific physical properties of white sidewall compounds were derived which accurately predicted significant improvements in compound properties.

The replacement of clay with reinforcing precipitated silica and talc provided increased compound tear strength, abrasion resistance and flex fatigue resistance, while maintaining cost. DPG provided efficient cure of the silica-filled compound without sacrificing whiteness properties and had the additional benefit of being a non-nitrosamine generating accelerator. The [TiO.sub.2] level was lowered to reduce the final cost of the compound without affecting whiteness when other variables were optimized.

The computer was used to optimize a wide range of predicted values of over 300 million formulations for rubber cure, physical properties and appearance. Compounds which have increased tear strength of 40 to 350%, improved abrasion resistance, improved flex fatigue resistance and increased original and UV-aged whiteness of 10 to 30%, with cost reductions of up to 10% were verified compared to current published white sidewall compounds for a variety of tire applications. Additional optimized compounds for a variety of applications are available within seconds from the PC-SAS programs.

References

[1.] R.S. Bhakuni, S.K Mowdood, W.H. Waddell, LS. Rai and D.L. Knight, "Tires" in Encyclopedia of Polymer Science a Engineering, Second Edition, Vol 16, J.L Kroschwitz, Editor, John Wiley & Sons, New York, 1989, p. 844. [2.] C Blackshaw and L Kristensen, "Blends of bromobutyl, EPDM and natural rubber in light colored compounds," ACS Rubber Division Meeting, May 1974. [3.] "Exxon chlorobutyl rubber, compounding and applications, Exxon Chemical, Houston, Texas (1989). [4.] A.E. Crepeau, "Natural rubber-halobutyl-ethylene propylene diene compounds for radial tire sidewall service," Journal of Elastomers and Plastics 9, 349 (1977). [5.] J. C Ambelang, "Pneumatic tire compounding," in the Vanderbilt Rubber Handbook, R.O. Babbit, Editor, R.T Vanderbilt Company, Inc., Norwalk, CT, 1978, p. 651. [6.] W.H. Waddell, R.S. Bhakuni, WW. Barbin ad P.H. Sandstrom, "Pneumatic tire compounding," in the Vanderbilt Rubber Handbook, R.F. Ohm, Editor, R.T Vanderbilt Company, Inc., Norwalk, CT, 1990, p. 605. [7.] "Compounding ingredients price list," Rubber World 205 (6), 37 (1991). [8.] R. A. McLean and V. A. Anderson, "Applied factorial and fractional designs," Marcel Dekker, Inc., New York, 1984. [9.] G. C. Derringer, "Statistical method," Rubber Chemistry and Technology 61, 381 (1988). [10.] K M. Davies D. G. Lloyd, and A. Orband, "The impact of n-nitrosamine regulations on sulfenamide selection," Kautschuk + Gummi Kuntststoffe, 120, February, 1989. [11.] G. C. Derringer and R. Suich, Journal of Quality Technology 12, 214 (1980).