The mechanical properties of RFL adhesives and influence on cord-in-rubber composites.

Many technical rubber goods, such as tires, hoses and conveyer belts, contain textile reinforcements for high strength and dimensional stability. These composites can only perform optimally if the adhesion between the reinforcement and the rubber is sound, so that forces acting on the rubber can be transferred to the cords.

The classical way to obtain good adhesion between rayon or nylon to rubber is to dip the cords in a resorcinol-formaldehyde-latex (RFL) suspension. RFL is an aqueous system containing a resorcinol-formaldehyde resin, which mainly determines the adhesion to the textile material, and a latex, responsible for adhesion to rubber. Because of the high surface polarities of rayon and nylon, strong physical and probably even chemical bonding takes place between the RF resin and the cord surface. For polyethylene terephthalate (PET) the surface has a low reactivity, while for aramid the active groups on the surface are sterically shielded by the aromatic nuclei and the accessibility to resin molecules is reduced by the high crystallinity of the aramid. Therefore, a predip is applied to PET and aramid, followed by an RFL dip, comparable to that used for nylon.

Most research on RFL adhesive systems has focused on the relationship between RFL compositions and the resulting adhesion of the dipped cords to rubber. However, because the RFL also affects the long term mechanical performance of the composite, the influence of its recipe on both the mechanical properties of the cured dip as well as the cord properties has to be known. To prepare a thin RFL layer representative for dipped cords (about 10 um), one can spray or cast the RFL onto mercury (refs. 1 and 2) or some non-adhering surface like PTFE (ref. 3) or etched glass (ref. 4). In practice, during the vulcanization of dipped cords in a rubber matrix, the curing agents present in the rubber diffuse into the RFL dip (refs. 4-7), resulting in crosslinking of the latex polymer. Besides promoting adhesion between the dip and matrix rubber, the vulcanization also influences the mechanical properties of the RFL dip. In addition to the difficulties encountered in handling such thin layers with their susceptibility to builtin stresses during the drying process, it is very difficult to homogeneously add a vulcanization system to the RFL using these techniques. Therefore in order to obtain appropriate information on the RFL adhesive properties, a straightforward procedure to obtain manageable RFL samples having the capability of adding a vulcanization system was deemed essential to this study.

Experimental

Materials

The RFL aqueous dispersions used in this study consisted of a resorcinol-formaldehyde resin (RF) and a latex (L) made up of a terpolymer of styrene, butadiene and vinylpyridine. The resin was prepared either by the condensation of formaldehyde and resorcinol, or by the addition of formaldehyde to the commercially available preformed resorcinol formaldehyde resin, Penacolite R2200 (Indspec Chemical). Resin condensation took place in the presence of either ammonia or sodium hydroxide, or a combination of both. For the latex, Pliocord VP 106 (Goodyear Chemicals), having a 15% vinylpyridine content was used.

For the vulcanization of the RFL layers, stearic acid, ZnO, Crystex Sulfur (Akzo Chemicals) and Perkacit MBTS (Akzo Chemicals) were used.

Procedure for obtaining RFL layers

The following procedure for obtaining RFL layers was developed:

* Drying an ample amount of RFL dip in shallow PTFE pans at room temperature and a low relative humidity (30% or less);

* Simulation curing of the RFL in an oven at 230[degrees]C for 90 seconds;

* Mixing the RFL with a curing system (2 wt% ZnO, 2 wt% stearic acid, 1 wt% sulfur and 1 wt% MBTS) on a rolling mill between 35-40[degrees]C and calendering into sheets.

The above procedure is a simple and reproducible method for obtaining RFL layers, offering the possibility of adding different curing agents to the adhesive.

Cure behavior of the RFL

Table 1 describes standard RFL formulations used to promote the adhesion of aramid, nylon and polyester to NR compounds. Throughout this article, the following compositions, based on Formulation Al, were evaluated; L/RF corresponds to weight ratios while F/R represents molar ratios:

a. At constant F/R = 1.7; L/RF = 3.5, 4.7, 5.7, 6.7, 8

b. At constant L/RF = 4.7; F/R = 1.0, 1.4, 1.7, 2.0, 2.5

In practice, sufficient adhesion is obtained for F/R ratios between 1.5 to 2 and with L/RF ratios falling between 5 and 7. The cure characteristics of the resulting RFL layers (e.g. scorch time, vulcanization time and possible network breakdown) were monitored using a Monsanto rotating disk rheometer at 150[degrees]C with a 3[degrees] arc.

Tensile properties of the RFL

In order to determine the effect of the curing system, tensile properties of the RFLs from table 1, with and without added curing agents, were determined in accordance with ASTM D412-87. Sheets of RFL were heated for 20 minutes in a hydraulic press at 150[degrees]C, producing layers with an average thickness of 3 mm. Dumbbell specimens (Die C) were stenciled and measurements were made on Instron, model 4501, equipped with an extensometer. The crosshead speed was 50 mm/min., and the initial separation of the extensometer clamps was 15 mm.

Table 1 - RFL formulations

                           A1       A2      B1        B2        C

Demineralized water       295.7    289.1   326.7     320.8     279.4
Ammonium hydroxide, 25%    12.9     -       -         -         -
Sodium hydroxide, 5%        -       19.5     12.6     12.9      14.8
Preformed resin, 70%       49.7     49.7      -        -        37.9
Resorcinol                  -       -        22.9      23.5      -
Formaldehyde, 37%           -       -        32.7     33.6      25.7
Demineralized water        110.1   110.1     59.3     75.7      97.4
Vinypyridine latex, 40%    508.7   508.7    519.5    533.5     513.8
Formaldehyde, 37%          22.9     22.9      -        -        -
Ammonium hydroxide, 25%     -       -        26.3     -         31.0

                         1000.0   1000.0   1000.0     1000.0  1000.0

L/RF weight ratio           4.7      4.7      6.0      6.0       5.7
F/R molar ratio             1.7      1.7      2.0      2.0       2.2

The same procedure was used to judge the influence of the L/RF and F/R ratio by measuring the properties of the RFLs described previously.

Dynamic mechanical properties of the RFL

Dynamic mechanical thermal analysis of the RFLs in table 1 were made in torsion with a Rheometrics RDA 700. A temperature sweep between -100 and +200[degrees]C was made with a 1% amplitude at a frequency of 1 Hz. From -100 to 0[degrees]C, measurements were taken at 3[degrees]C intervals (heating rate 3.7[degrees]C/min), and from 0[degrees]C to 200[degrees]C the interval was 5[degrees]C (heating rate 5.3[degrees]C/min). The sample, cut from the same sheets used in the tensile tests, were 50 x 10 mm with an average thickness of 3 mm.

Influence of RFL formulation on cord properties

For obtaining dipped cord properties, Diolen 1125T PET and Twaron 1001 aramid technical yarns were used. These yarns are supplied possessing an adhesive activator, and therefore require only an RFL dip in order to provide sufficient adhesion to rubber. From the PET yarns a two-ply cord of the 1100 dtex x 1Z435 x 2S435 turns/m construction was employed while the construction of the aramid cord was 1680 dtex x 1Z330 x 2S330 turns/m.

Both cords were dipped with RFLs A1, A2 and C, as described in table 1. The following dipping conditions were applied:

PET: RFL (25 wt%), 120"-150[degrees]C-4.5N / 30"-240[degrees]C-22N / 30"-240[degrees]C-4.5N

Aramid: RFL (25 wt%), 120"-150[degrees]C-9N / 90"-230[degrees]C-9N

The dipped cords were vulcanized in a standard NR compound and were tested at room temperature for static adhesion using a peel test as described in ASTM D4393-85.

The dipped cords were further evaluated in a disk fatigue (GBF) test in order to provide an impression of the (compression) fatigue behavior of the cords after cyclic loading. Although the GBF test is described in ASTM D885-62T, it is not fully standardized. Specimen construction and dimensions were described earlier (ref. 8). All tests were carried out on the same rubber as used in the adhesion test at 40 Hz under the following loading conditions:

PET: 0% elongation, 20% compression, 24 hours - 15% elongation, 20% compression, two, four, six and eight hours

Aramid: 2% elongation, 14% compression, 36 hours.

After testing, the cord was removed from the rubber matrix and its residual strength was assessed.

Results and discussion

Cure behavior of the RFL

Rheometric studies on the RFL layers provide an insight into the vulcanization behavior as is illustrated in figure 1. A fast curing reaction is noted for the RFL vulcanized without curing agents. The final product is brittle, suggesting that the latex rubber has been crosslinked by the RF resin. A similar process may be operative when a dipped cord is over-cured during the dipping process. In such cases adhesion between RFL and rubber matrix is reduced due to fewer accessible reactive sites in the RFL. With the addition of a curing system to the RFL, a sulfur vulcanization is evident as seen by the scorch and cure behavior (figure 1).

[CHART OMITTED]

Figures 2 and 3 show the influence of the L/RF and F/R ratios on the cure behavior of RFL dips based on variations in RFL A1 described earlier. Cure behavior was determined by the scorch time ([ts.sub.2]), time to 90% cure ([tc.sub.90]), and the cure rate index (CRI) which is defined as:

CRI = ([M.sub.h] - [M.sub.1])/([tc.sub.90] - [ts.sub.2]) with [M.sub.h] = maximal torque level

[M.sub.1] = minimal torque level

[CHART OMITTED]

Figure 2 shows that at higher L/RF ratios, the CRI decreases sharply. Since [ts.sub.2] and [tc.sub.90] are shown to be constant, this effect is attributed to a decreasing difference between maximal and minimal torque levels. The cure behavior of pure VP rubber, having the same vulcanization system (solid figures), corresponds with that of an RFL having a very high L/RF ratio. From figure 3 it can be seen that with an increasing F/R ratio, [ts.sub.2] remains constant at about 1.5 minutes, while [tc.sub.90] increases from 13 to 19 minutes. The CRI on the other hand displays a maximum at an F/R ratio of 1.5, which corresponds to the stoichiometric ratio for crosslinking between resorcinol and formaldehyde in the resin. With additional formaldehyde, the CRI drops considerably. This can be attributed to not only an increase in [tc.sub.90], but also to an increase in [M.sub.1]. Therefore high RFL cure reactivity is expected with low L/R ratios in combination with an F/R of about 1.5. Overall, the cure behavior of RFL dips containing vulcanizing agents seems to be completely rubber like. The applied curing system results in cure times normal for tire compounds, while the resulting scorch times are rather short.

Tensile properties of the RFL

Figure 4 shows the stress-strain behavior for three RFL dips described in table 1 (formulations A1, B1 and B2), with and without added curing agents. In the absence of curing agents, the resulting RFL layers are weak, due to little to no interaction between resin and latex. However, analogous to rubber, the addition of a vulcanization packet to the RFL significantly increases modulus and breaking strength upon curing. For formulations A1 and B1, the elongation at break now exceeds 200%. These differences clearly demonstrate the importance of curing agent diffusion from the rubber into the RFL during vulcanization. Limitation of this diffusion negatively influences the mechanical properties of the RFL layer which may result in cohesive failure within this layer during composite loading. Diffusion limitation can occur with rubbers having short scorch times, high "flow viscosities" ([M.sub.1]), or when a peroxide curing system is used. In this last situation, the short lived radicals or radical oligomers may be terminated in the rubber before entering the RFL layer.

[CHART OMITTED]

Figure 5 plots the tensile secant modulus at 2% strain against both the L/RF and F/R ratios, taken from samples based on variations in RFL A1, as described in the section on cure behavior. All RFLs were cured using a standard curing system. At a constant F/R ratio (1.7), the dip modulus decreases with increasing amounts of latex. For reference pure latex has a modulus of 0.9 MPa. Increasing the resin's formaldehyde content at a constant L/RF ratio, results in a higher dip modulus. This can be explained in that increases in formaldehyde increase both resin molecular weight and crosslink density, which results in a greater reinforcing effect. Within the upper range of the F/R ratios tested, the dip modulus still increased. Nevertheless, it is expected that at very high F/R ratios, the modulus will eventually decrease due to relatively lower percentages of resorcinol.

[CHART OMITTED]

In addition to F/R and L/RF ratios affecting adhesion, initial base selection may also influence dip curing and ultimately adhesive behavior. An alkaline environment is necessary for RF resin condensation, and as can be seen from figure 6, the influence of either ammonia or sodium hydroxide on cured dip properties can be quite different. Within each set, A or B, both the resin and latex levels are identical with the difference being the base employed (ammonia for A1 and B1; sodium hydroxide for A2 and B2). Dip mechanical properties corresponding to the formulations employing sodium hydroxide results in a greater modulus, lower breaking toughness (area under curve) and lower elongation at break as compared to their respective analogs based on ammonia. The lower modulus of ammonia containing RFL can be expected if ammonia is incorporated into the resin network, producing dimethylene amine crosslinks, instead of methylene crosslinks, as was shown by [13.sup.C] NMR techniques (ref. 9):

with [NH.sub.3]:>-[CH.sub.2]-NH-[CH.sub.2]-< with NaOH:>-[CH.sub.2]-<

By incorporating two formaldehydes per crosslink, more flexible bridges are produced. Since twice as many formaldehydes are consumed in chain extension between resorcinol molecules, less are available for crosslinking. Both factors result in a less extended, and more flexible network which is manifested by a lower RFL modulus.

[CHART OMITTED]

Dynamic mechanical properties of the RFL

Dynamic mechanical analysis (DMA) can provide information on both the morphology as well as the mechanical properties of viscoelastic materials. Although DMA is widely used on polymeric materials, so far little has been published with regard to the use of this technique on RFL layers (refs. 1-3). Figures 7 and 8 show typical plots of the complex modulus G(*) and the loss tangent, tan &, as a function of temperature for the RFL layers studied. Following Rahrig (ref. 2), we have labeled the low temperature relaxation between -30 and -40[degrees]C as the [alpha]-relaxation. The higher temperature relaxation, which is very broad and sometimes barely visible, is labeled the [alpha]'-relaxation. The [alpha]-relaxation arises from molecular motion in the VP terpolymer, which as a pure polymer has a [T.sub.g] of about -47[degrees]C (ref. 10). The [alpha]'-relaxation occurs over the temperature range where one might expect to see molecular motion arising from the RF resin phase.

[CHART OMITTED]

The samples shown in figures 7 and 8 represent RFL formulations A1 and A2, with resin condensation in the presence of ammonia and sodium hydroxide, respectively. Comparison of the two figures shows that the [alpha]-relaxation decreases in magnitude and that a more distinct [alpha]'-relaxation is shifted to higher temperature with the employment of sodium hydroxide (figure 8) in place of ammonia (figure 7). These results can be interpreted suing the RFL model shown in figure 9. Here the RF resin forms a continuous network in which the VP polymer is dispersed as distinct rubbery inclusions. In effect this model represents a semi-interpenetrating network. When the VP particles are in contact with one another the curing agents are able to diffuse throughout the entire system. During RFL vulcanization, not only do the rubber particles undergo crosslinking, but also an interaction between the RF resin and the latex particles is being brought about. At this point the molecular motion of the terpolymer molecules is restricted by the resin, and as explained earlier from tensile results, using sodium hydroxide during resin condensation results in a much stiffer resin network, and the stiffer the resin, the more restriction occurs. Also in accordance with the tensile results, the moduli at temperatures above 0[degrees]C are about seven times higher for sodium hydroxide condensed RFL. The loss tangent of the latter is slightly lower up to 60[degrees]C, but increases at higher temperatures.

[CHART OMITTED]

To show the influence of the absence of a vulcanization system, the dynamic mechanical properties of RFL A1, cured without curing agents, are shown in figure 10. Compared to figure 7, the [alpha]-relaxation lies at a somewhat lower temperature (-37 against -34[degrees]C) and has a greater magnitude. The [alpha]'-relaxation is also more distinct, and the modulus is much lower. These findings concur with those taken from tensile measurements, and suggest less interaction between the resin and the VP polymer.

[CHART OMITTED]

The influence of RFL formulation on cord properties

The L/RF ratio is known to significantly influence the adhesion properties of the dipped cord. At too high L/RF ratios, composite failure occurs either at the cord/RFL interface as a consequence of too few reactive resin groups, or via a cohesive failure within the RFL itself due to a low mechanical strength of the RFL layer. On the other hand, when too much resin is used (lower L/RF ratios), RFL/cord adhesion may be sufficient, but failure could occur at the RFL/rubber interface. Optimal adhesion is normally achieved at L/RF weight ratios between five and seven.

With varying F/R molar ratios, maximum adhesion is normally found between 1.5 and 2.0. With less formaldehyde, a lower resin crosslinking density results which affects the mechanical properties of the RFL; too much formaldehyde increases the resin's molecular weight, thus hindering resin penetration into the cord surface.

Table 2 shows the properties of PET and aramid cords dipped with different RFL adhesives. The RFL compositions do not appear to affect the cord properties in any significant way. No differences in cord stiffness are noted for either PET or aramid dipped with RFL C and Al; this is in accordance with the similar moduli characteristics of these RFLs. The higher modulus noted for RFL A2 increases the cord stiffness substantially, especially in combination with the aramid cord. Any correlation between adhesion and RFL composition and modulus could not be substantiated in these experiments, given that all three dips had about the same L/RF and F/R ratios.

Table 2 - influence of RFL on dipped cord properties

RFL:                            C         A1        A2
RFL modulus-2% (MPa)          6.8         10        70
PET:
Dip pick-up (w%)              6.7        6.8       9.2
Breaking strength (N)         139        140       138
Elongation at break (%)      13.7       13.4      12.9
Taber cord stiffness (-)     0.42       0.40      0.53
Adhesion (N/2 cm)(a)     300 (60)   280 (30)   300 (60)
GBF (%)(b), 0/-20%, 24h        90         91        87
Aramid:
Dip pick-up (w%)              7.2        8.3       6.2
Breaking strength (N)         513        504       505
Elongation at break (%)       5.2        5.1       5.1
Taber cord stiffness (-)     0.35       0.36      0.52
Adhesion (N/2 cm)(a)     270 (20)   300 (60)   290 (40)
GBF (%)b, +2/-14%, 36 h        47         42       22

(a)Figures between brackets represent rubber coverages of the peeled surface

(b)Relative to cords from untested specimens

The retained cord strengths, measured after block fatigue testing, for either PET or aramid cords treated with RFL C and Al are equivalent (table 2). Aramid tested in combination with RFL A2 results in a considerably lower cord strength than aramid treated with RFL C or Al. For PET, the difference is not statistically significant under these testing conditions. Table 3 summarizes the GBF test results for dipped PET cords under more severe testing conditions. These results indicate that RFL A2 has an adverse affect on the cord fatigue properties of PET as well. Although absolute differences in retained strength are small, the increased number of broken cords (which were not used to calculate the average) observed in combination with RFL A2 indicate that these differences are indeed significant.

Table 3 - retained strength of dipped PET cords after GBF testing

RFL:                        A1      A2
Dip pick-up (w%)           12.3     9.7
Breaking strength (N)      137      137
Elongation at break (%)    14.0     13.8
Taber cord stiffness (-)   0.24     0.31
GBF (%), +15/-20%(a):
2 hours                   86 (1)   80 (4)
4 hours                   89 (0)   76 (1)
6 hours                   88 (0)   77 (1)
8 hours                   84 (0)   80 (3)

(a)Figures between brackets represent the number of broken cords after testing, on a total of 6 tested cords

Although on dipped cords the layer of high modulus RFL A2 has a thickness of only about 10 [micro]m, it still can lead to an adverse effect on the fatigue properties of both PET and aramid cords. A possible explanation for this phenomenon is an increased restriction of the outer filaments, causing local increases in filament bending fatigue during compression. When RFL Al or C is applied to the cord, the lower RFL modulus allows the filaments to move more easily with respect to each other during loading. As a result, the applied tension is more evenly spread across the total cord bundle, thus inducing less cord damage and a higher retained strength. Another possibility is that the cyclic loading causes small cracks in the more brittle layer of RFL A2 to form, after which the cord filaments are damaged during compression loading.

Conclusions

The developed test procedure is a strong tool to gain insight into the relationship between formulation and properties of RFL adhesives, and it offers the possibility to improve the long term mechanical properties of cord/rubber composites.

Rheometric studies suggest that a sulfure vulcanization mechanism is operative in the RFL adhesive layers when curing agents are present. In practice, the curing agents diffuse from the rubber into the RFL. Vulcanization strongly improves the RFL's mechanical properties, which in turn are mainly determined by the latex to resin ratio (L/RF), the formaldehyde to resorcinol ratio (F/R), and by the base used for resin condensation, i.e. ammonia or sodium hydroxide.

The strong improvement in mechanical properties of the RFL by vulcanization emphasizes the importance of allowing the rubber curing agents to diffuse into the RFL during the rubber scorch time.

Dynamic mechanical thermal analysis showed that RFL adhesive systems are two-phase materials, with considerable interaction between the RF and the latex phases. Higher resin stiffness results in increased restriction of the latex particles, with the loss tangent at working conditions remaining about the same.

Neither the dipped cord tensile properties nor simple static adhesion are influenced by the RFL modulus, if the L/RF and F/R ratios remain constant. Fatigue tests show that retained strength drops when a sodium hydroxide condensed, high-modulus RFL is used.

References

(1.)G. Raumann, Text. Res. J. 38, 627 (1968).

(2.)D.B. Rahrig, J. Adhesion 16, 179 (1984).

(3.)B. Das, M. Girgis, J. Appl. Polym. Sci. 40, 1367 (1990).

(4.)M.S. Dostyan, D.M. Sandomirskii, R.V. Uzina, Sov. Rubber Technol. 9, 20 (1960).

(5.)A.G. Causa, in "Tire reinforcement and tire performance," R.A. Fleming and D.I. Livingston Eds., ASTM Spec. Tech. Publ. 694, 200 (1979).

(6.)B.C. Begnoche, R.L. Keefe, A.G. Causa, Rubber Chem. Technol. 60, 689 (1987).

(7.)G. Gillberg, L.C. Sawyer, A.L. Promislow, J. Appl. Polym. Sci. 28, 3723 (1983).

(8.)E.M. Winkler and M.C. de Jong, Rubber Chem. Technol. 63, 223 (1990).

(9.)A.J.J. de Wit, W. Dankelman, W.G.B. Huysmans, J. de Wit, Angew. Makromolek. Chem. 62, 7 (1977).

(10.)Product information, Goodyear Chemicals.