Structure-property relationships--linear and star-branched macrostructures.

Experimental

Synthesis--Model PBds

A one-gallon glass bowl reactor was used for polymerization, but the methods are similar to those of references using bottle polymerization (ref. 42) and vacuum line (ref. 83) techniques. The vessel was equipped with a solenoid-actuated steam and cold water temperature control system that utilized a stainless steel internal coil for heat transfer. The reaction temperature was monitored via thermocouples and controlled electronically. An air-driven impeller was used for agitation. Solvent and monomer were introduced through one port, while initiating and terminating agents were added via an injection port comprising a ball valve and leg that terminated below the liquid level of the reactor. Reactor contents were removed via a dip-leg. The reactor also was equipped with a nitrogen inlet and vent for pressure control, and was operated under positive pressure.

A stock n-butyllithium was prepared as a 1.03 M solution in hexanes. A silicon tetrachloride solution was prepared in hexanes to 0.5 M and stored over 3 A molecular sieves. Ethanol was purged and also held over 3 A molecular sieves. A premix solution was prepared by diluting butadiene monomer in hexanes to a 23.5% (w/w) solution and passing over a bed of silica gel until the level of impurities as titrated by n-butyllithium (using 2,2'-dipyridyl indicator, ethanol 99+%) was less than 2 ppm. A separate supply of dry hexanes was also prepared using similar techniques. Dry nitrogen was used for all pressure transfers both into and out of the reactor and as the atmosphere for both the bottled reagents and monomer storage tanks. The reaction vessel was prepared prior to each polymerization by scavenging with a solution of n-butyllithium in hexanes. After the cleaning solution was removed, the vessel was subsequently purged with dry nitrogen. 2,200 g of a butadiene in hexanes solution was added to the reactor for each polymerization to give a 13.6% (w/w) solution of monomer. Polymerizations were conducted at 65[degrees]C. n-butyllithium was added to the reaction mixture according to the desired molecular weights. A small excess of n-butyllithium was added with this charge to scavenge impurities present in the premix and vessel. Monomer conversion was monitored by residual analysis via gas chromatography (ref. 84). At full conversion, the polymerizations were either terminated with a slight molar excess of ethanol or a stoichiometric (Li:Cl); amount of Si[Cl.sub.4]. The terminated polymer solution was subsequently removed from the reactor and stabilized with BHT (1.0% (w/w) on polymer). The hexane was removed from the polymer solution by drying in a forced air oven at 70[degrees]C until no more weight loss was recorded.

Commercial SSBRs

The SSBRs used for compound testing were commercially available linear and star-branched products. S-25X35 (SSBR-A) and S-29X29 (SSBR-B) are prepared by a continuous solution anionic process and contain 37.5 parts per hundred rubber of aromatic oil. The rubber is stabilized with Polystay 100. SSBR-A displays a linear macrostructure, while SSBR-B is coupled with silicon tetrachloride.

Compounding and testing

Gum rubber testing

Raw gum rubber was characterized by several techniques. [sup.13]CNMR was used to determine the microstructure of the samples. Samples were dissolved in CD[Cl.sub.3] and run on a Unityplus 400 NMR spectrometer. A 10 mm probe was used at an operating temperature of 55[degrees]C. Differential scanning calorimetry was used to determine glass transition temperatures (Tg). Analyses were performed on a TA Instruments 2910 MDSC using a 2[degrees]C/min. linear heat rate with [+ or -] 1.5[degrees]C modulation amplitude, 60 second modulation period and a helium flow rate of 25 ml/min. Size exclusion chromatography (SEC) was used in conjunction with multi-angle laser light scattering (MALLS) and refractive index (RI) detection to perform the separation and molecular weight analysis. SEC was performed via Polymer Labs B and C mixed microgel columns using tetrahydrofuran as the carrier solvent and for sample preparation. MALLS measurements were carried out using a Wyatt Technologies miniDawn light scattering detector and a Hewlett Packard 1047A refractive index detector. The intrinsic viscosities of selected experimental samples were calculated using standard capillary viscometry techniques (ref. 85). A Ubbelohde-type viscometer was used in conjunction with an infrared sensor to determine the efflux time through the capillary. The viscometer apparatus was suspended in a constant temperature bath set at 30[+ or -]C. Toluene was used as the solvent blank and for sample dilution. Mooney viscometry was performed at 100[degrees]C using a large die with one minute of warm-up and four minutes of testing.

Compound mixing

The SSBRs were mixed in a 1,600 cc volume internal mixer equipped with a variable speed drive. Two-stage mixing was used. The carbon black non-productive stage was mixed at 60 rpm with a starting temperature of 65[degrees]C. The stage was mixed for six minutes or until an internal temperature of 160[degrees]C was reached. The silica non-productive stage was mixed using variable rpm control, starting at 60 rpm and 65[degrees]C. The silica stock was mixed at 160[degrees]C for two minutes, controlling rpm to maintain temperature. Productive stages were similar for both formulations, using initial conditions of 60 rpm and 65[degrees]C and mixing for two minutes or until the internal temperature reached 110[degrees]C.

Compound testing

The compounded stock was tested using standard techniques. An oscillating die rheometer was used for cure analysis (ASTM D2084). Both a parallel plate rheometer (model PBds) and a rubber process analyzer (PBds and SSBRs) were used for oscillatory shear testing. Testing was performed on a parallel plate rheometer using a TC501 temperature regulating device, a PP25 sensor and a 1.3 mm gap. A stress of 5,000 Pa was applied over a frequency range of 0.063 rad/s to 628.3 rad/s (0.01 Hz to 100.0 Hz) at various temperatures. Shear testing was also performed on a rubber process analyzer at 100[degrees]C under the varied frequencies and strains outlined in the text. Extrusion processing was evaluated using a Garvey die (ASTM D2230). Hardness was measured by the durometer A method (ASTM D2240 and DIN 53505). A solids analyzer was used to determine the dynamic properties of the cured stock. The test was performed in tension at 11 Hz and a heating rate of 3[degrees]C per minute at 0.1% strain. Finally, rebound testing was performed according to ASTM D1054 and DIN53512.

Results and discussion

In an attempt to link the solution and bulk rheological proper ties of linear and branched elastomers to the processing characteristics of filled rubber compounds, PBd samples were prepared using anionic polymerization. Four-arm stars were synthesized as models for branched polymers. The unmodified solution anionic polymerization of butadiene was chosen, since this precursor polymer results in the most efficient linking reaction possible, and silicon tetrachloride was used as the linking agent. Experimental samples were produced to form both linear and branched polymers at three different molecular weights. Both the microstructure and the macrostructure of these materials were fully characterized prior to the solution and bulk rheological testing and analysis.

In addition, commercially available linear and silicon-coupled solution SBR products were mixed in a model carbon-black filled compound. The stock was then laboratory tested to determine differences in processing and hysteresis as a function of the two macrostructures.

Model poly(butadienes)--microstructure analysis

Table 1 summarizes the microstructure characterization data. Vinyl levels below 10% (w/w) are typical of organolithium-initiated PBd polymerized in hexane solvent in the absence of vinyl directing modification (ref. 10). No significant differences in microstructure are apparent. All Tgs were sharp transitions characteristic of samples with intra- and inter-chain microstructural homogeneity.

Macrostructure analysis

By utilizing SEC techniques to provide the separation of polymeric molecules based on their hydrodynamic volume (ref. 86), MALLS can provide an absolute measurement of both molecular weight and root mean square radius of gyration for each polymer fraction (ref. 87). The dn/dc value used for the concentration measurements was 0.130 mL/g (ref. 88).

Table 2 shows the SEC characterization for macrostructure determination. Target molar masses are given; the coupled samples show both targets for the base (uncoupled parent) and the coupled molecular weight. Both the weight average molecular weight ([M.sub.w]) and the polydispersity index (PD) are shown for each sample. The [M.sub.w] and PD of the individual parent and coupled peaks of the star samples are also shown. For the coupled samples, the [M.sub.w] and PD values given under the "overall" heading are for the complete bimodal distribution. The PD values of the linear samples and the base peaks of the coupled samples are quite low as expected for batch-produced anionic PBd. The peaks representing the star fraction of the coupled samples are also quite low, the first indication of a near-quantitative linking reaction. Theoretically, the linking reaction should produce PD values narrower than that of the parent chains (ref. 89), but the parent chains used in this study have such narrow molecular weight distributions that the effect is not seen. A bimodal distribution for the coupled samples is inferred by the increase in the overall PD of the sample.

Several methods for calculating the coupling efficiency of linking reactions have been reported. By taking the number average molecular weight ratio of coupled to base polymer, a value for the average number of chains attached about the linking atom (or arm functionality, f) can be calculated. The functionality for each coupled sample is given, and these values are very close to the tetrafunctional target (f = 4). The weight percent of coupled product can be calculated from this ratio if one assumes that only tetrafunctional and unreacted base polymer are present (ref. 90). However, current computer-based MALLS software allows for the calculation of weight percent coupled product by simply dividing the area under the coupled peak by the total area of both coupled and base peaks. This technique assumes that there is no loss of mass of the sample either during filtering or by retention on the SEC column, an assumption checked and validated by a mass balance calculation. The coupling efficiencies for the star samples are given in table 2. These samples averaged approximately 91% coupling efficiency. By the above analysis, it has been shown that the samples prepared are linear and star macrostructures with near identical molecular weights (linear peak versus coupled peak), low polydispersities and high coupling efficiencies.

As previously noted, MALLS measurements provide not only molecular weights but also radius of gyration. Comparing the size of a random coil in solution is one method for determining the branching character of the sample. Branched samples should have a smaller hydrodynamic volume than their linear counterparts (refs. 91 and 92). Star-shaped macromolecules, when compared to linear species at equal molar mass, have a density in solution that is much higher due to the conformation restriction about the multi-functional linking node. Using SEC, star and linear samples of the same molar mass will have different elution times (linear first), and will require correction if using standard linear poly(styrene) calibration plots for molecular weight. However, MALLS detection determines the weight average molecular weight of the sample regardless of elution time. A plot of the overall z-average root mean square radius of gyration ([Rg.sub.z.sup.2]) as a function of weight average molecular weight for the samples is shown in figure 2. A conformation plot of the root mean square radius of gyration ([Rg.sup.2]) as a function of molecular weight can provide a measure of the relative density of a given sample at the same molecular weight. Figure 3 shows a conformation plot of the linear and branched sample pair C and I. In each case, the linear sample had a higher [Rg.sup.2] than the star sample at a given molar mass.

[FIGURES 2-3 OMITTED]

Dilute solution viscometry has also been utilized to characterize the molecular weight of polymeric samples. Solution viscosity can be calculated as a function of molecular weight through the Mark-Houwink-Sakurada equation, which empirically relates intrinsic viscosity ([eta]) to the viscosity average molecular weight. This equation is shown in equation 3.

(3) [eta] = K[M.sup.a]

K and a are constants that are defined by the polymer microstructure. The measurement of intrinsic viscosity is sensitive to changes in macrostructure. The value of exponent (a) in the equation is a function of the shape of the sample in solution, and is dependent on the chain microstructure, macrostructure and solvent. Many experimentalists have tried to derive expressions for the shape of a polymer sample in solution based on the interaction with the solvent, and have produced equations to calculate conformations that correlate with (a) (refs. 57, 93 and 94).

The intrinsic viscosity can be calculated by extrapolating viscosity data as a function of concentration to infinite dilution. The Huggins equation (ref. 95) (equation 4) uses the reduced viscosity in this measurement, while a variation on this equation uses the natural logarithm of the relative viscosity (ref. 96) (equation 5).

(4) [[eta].sub.sp]/c = [eta] + k'[eta]2c

(5) 1n[[eta].sub.r]/c = [eta] + k"[eta]2c

Here, k' and k" are constants for polymers of equal micro- and macrostructures but different molecular weights in the same solvent and obey the relationship k' - k" = 0.5. The values of these constants, however, have been shown not to correlate to the degree of branching (ref. 97). This study plots the results of both equations, but uses only the results of equation 5 for calculations. To calculate intrinsic viscosities, four concentrations of solute were used, and the efflux time was taken as an average of three runs at each concentration.

Typical plots of these measurements are shown in figures 4 and 5 for a linear structure (sample D) and a star structure (sample J), respectively, both at a nominal 250,000 g/tool [M.sub.w]. Table 3 contains a summary of the solution viscometry data for the selected experimental samples.

[FIGURES 4-5 OMITTED]

Interesting relationships can be seen when the data from light scattering and viscometry are plotted together. A typical Mark-Houwink-Sakurada plot of 1n[eta] as a function of 1n([M.sub.w]) is shown in figure 6. This figure includes both the experimental data and a trend line depicting the theoretical values. The theoretical trend line uses published constants of 3.90E-04 ml/g (K) and 0.713 (a) (ref. 98), along with the calculated branching ratio viscosity correction for the star macrostructure. There was good agreement between the experimental and the theoretical data. The calculated (a) value for the linear samples was 0.67 and that for the star samples was 0.70. As expected, the solution viscosity for the star macrostructure was less than that for the linear case. A review of published data for both linear and star-shaped poly(styrene) samples qualitatively gave similar results (ref. 15).

[FIGURE 6 OMITTED]

The analysis of reported solution viscosity versus molecular weight data in the literature also suggests that intrinsic viscosities decrease as the functionality of the star polymer increases (at constant molar mass). As the above theory asserts, the intrinsic viscosity should be a function of the hydrodynamic volume of the polymer chains in solution, regardless of macrostructure, with [Rg.sub.z.sup.2] an adequate approximation for the volume of the sphere defined by this construct (ref. 5). Figure 7 plots the intrinsic viscosity as a function of [Rg.sub.z.sup.2]. A good fit is seen for all samples regardless of the macrostructure.

[FIGURE 7 OMITTED]

The characterization data above supports the fact that the PBd samples had been prepared to the desired macrostructures. They exhibited the dilute solution behavior expected for the conformations of linear and four-arm star macromolecules. The structure of these samples had been set by the chemistry of their preparation, and verified by both light scattering and solution viscometry.

Table 1--microstructural analysis of the sample PBds

                      %       13C NMR         %
Sample    Type      vinyl    % cis-1,4    trans-1,4

A         Linear     8.5       42.4         49.1
B         Linear     8.5       42.9         48.6
C         Linear     7.7       42.5         49.8
D         Linear     8.5       43.8         47.7
E         Linear     7.7       45.4         46.9
F         Linear     8.4       43.9         47.7
G         Star       8.5       40.4         51.1
H         Star       8.5       40.1         51.4
I         Star       8.0       39.9         52.1
J         Star       8.5       40.3         51.2
K         Star       8.7       41.0         50.3
L         Star       8.4       40.3         51.3

                  Tg
Sample    (mdpt., [degrees]C)

A               -94.9
B               -94.7
C               -94.8
D               -95.2
E               -95.9
F               -95.8
G               -94.1
H               -94.3
I               -94.7
J               -94.2
K               -93.9
L               -94.6

Table 2--SEC analysis of the sample PBds

                    [M.sub.W]    [M.sub.W]
                    target(s)    precursor            Coupled
Sample     Type     (kg/mol)     (kg/mol)      PD     (kg/mol)

A         Linear       200          --         --        --
B         Linear       200          --         --        --
C         Linear       250          --         --        --
D         Linear       250          --         --        --
E         Linear       300          --         --        --
F         Linear       300          --         --        --
G          Star      50->200        53        1.01      199
H          Star      50->200        56        1.00      211
1          Star      63->250        66        1.00      251
J          Star      63->250        62        1.01      243
K          Star      75->300        74        1.01      296
L          Star      75->300        71        1.01      292

                  [M.sub.W]
                   overall                  %
Sample     PD     (kg/mol)       PD      coupled     f

A          --        208        1.01       --        --
B          --        212        1.01       --        --
C          --        249        1.01       --        --
D          --        246        1.00       --        --
E          --        315        1.01       --        --
F          --        296        1.00       --        --
G         1.02       188        1.15       89       3.75
H         1.02       211        1.06       94       3.77
1         1.02       245        1.14       91       3.80
J         1.02       223        1.18       90       3.92
K         1.01       277        1.17       91       4.00
L         1.00       250        1.21       90       4.11

Table 3--summary of the intrinsic viscosity data

                    [M.sub.W]    [[eta]]
Sample     Type     (kg/mol)     (dL/g)

A         Linear       208        2.29
D         Linear       246        2.60
F         Linear       296        2.90
G          Star        199        1.52
J          Star        243        1.73
K          Star        292        1.98

References

(5.) P.J. Flory, Principals of Polymer Chemistry, Cornell University Press, Ithaca, 1953.

(10.) H.L. Hsieh and R.P. Quirk, Anionic Polymerization: Principals and Practical Applications, Marcel Dekker, Inc., New York, 1996.

(15.) B.J. Bauer and L.J. Fetters, Rubber Chem. Tech., 51, 406 (1978).

(42.) R.P. Zelinski and C.F. Wofford, J. Polym. Sci., 3, 93 (1965).

(57.) B.H. Zimm, J. Chem. Phys., 24, 269 (1956).

(83.) L.J. Fetters and M. Morton, Macromolecules, 7, 552 (1974).

(84.) Gas chromotography was performed on a Hewlett Packard 5890 Series II machine using a DB-5 60 m x 0.25 mm column with helium as the carrier gas. A Hewlett Packard 3396 Series II integrator was used for analysis.

(85.) F.W. Billmeyer, Jr., Textbook of Polymer Science, 3rd Ed., John Wiley and Sons, New York, 1984.

(86.) J.C. Moore, J. Polym. Sci., 2, 832 (1964).

(87.) P.J. Wyatt, Analytica Chimica Acta, 272, 1 (1993).

(88.) Determined in THF at 23[degrees]C for high molecular weight poly(butadiene). American Polymer Standards Corporation, 2003.

(89.) B.L. Johnson, H.E. Adams, E C. Weissert and K. Farat, The Proceedings of the International Rubber Conference, Maclaren and Sons, London, 1967.

(90.) J.C. Moore, J. Polym. Sci., 2, 832 (1964).

(91.) W.H. Stockmayer and M. Fixman, Annals. N. Y. Acad. Sci., 68, 323 (1953).

(92.) W.W. Graessley, "Characterization of macromolecular structure," in NAS Pub. 1573, D. McIntyre, ed., National Academy of Sciences, Washington, D.C., 1968.

(93.) P.J. Flory and T.G. Fox, J. Polym. Sci., 5, 745 (1950).

(94.) P. Debye and A.M. Bueche, J. Chem. Phys., 16, 573 (1948).

(95.) M.L. Huggins, J. Am. Chem. Soc., 64, 2,716 (1942).

(96.) E.O. Kraemer, Ind.Eng. Chem., 30, 1,200 (1938).

(97.) M. Morton, ZE. Helminiak, S.D. Gadkary and F. Bueche, J. Polym. Sci., 57, 471 (1962).

(98.) R. Endo, J. Rubber Ind. Japan, 35, 658 (1962).