The low-cost solvent can be converted selectively to useful polymer feedstocks.

Sheldon C. Sherman
Alexei V. Iretskii
Mark G. White
David A. Schiraldi

Despite the huge effort expended toward development of liquid crystalline polyesters and other high-performance rigid-rod polymers over the past 30 years, these materials account for sales of just 10 million lb/year in the United States (1). In contrast, the annual sales of poly(ethylene terephthalate) (PET) amount to more than 2000 times as many pounds per year, in spite of its more modest physical and thermal properties. The root cause of this dramatic difference in sales is, of course, selling price.

While PET sells typically in the $0.50/lb range, rigid-rod polyesters sell for $5–10/lb (2). Although manufacturing process differences between traditional commodity polymers and high-performance materials certainly contribute to the high cost of the latter substances, the costs of rigid-rod monomers, such as p-hydroxybenzoic acid, 6-hydroxy-2-naphthoic acid, trans-stilbene-4,4´-dicarboxylic acid, and 4,4´-bibenzoic acid (BBA), play a dominant role in determining the ultimate polymer costs to end users.

Our work had its genesis in a 1997 challenge from KoSa (then Hoechst Celanese) management to find a commercially viable route to the production of 4,4´-BBA at ~$0.50/lb. Previous methods, typically based on alkylation of biphenyl (3, 4) or recovery of oxidation unit byproducts (5), produced 4,4´-BBA salable in the $3/lb range. At such a price, bibenzoate polymers would be relegated to low-volume niche markets and could not justify new capital investments. The team assembled at KoSa and the Georgia Institute of Technology continued a long-term collaboration in strong-acid catalysis and aromatic functionalization that recently developed highly stereospecific carbonylation of aromatic feedstocks, such as toluene (6).

Bibenzoate-based polymers
Highly ordered, smectic liquid crystalline polymers have been produced from 4,4´-BBA and a series of linear diols ranging from 1,3-propanediol to 1,12-dodecanediol (7–11), as well as glycol ethers such as diethylene or triethylene glycol (12). With the impending commercialization of relatively low-cost 1,3-propanediol for poly(propylene terephthalate) fiber applications (13), poly(propylene bibenzoate) may represent a significant new market opportunity. More complex copolyesters based on 4,4´-BBA and other diacids, such as terephthalic acid (14), 2,6-naphthalenedicarboxylic acid (14), 3,4´-BBA (15), and p-hydroxybenzoic acid (16); polyesters based on 4,4´-BBA and two or more diols (17–19); and 4,4´-BBA copolyester amides (20) have all been shown to exhibit potentially salable thermal and mechanical properties. Copoly(ester–ether) elastomers, capable of withstanding harsher operating conditions than the transitional poly(butylene terephthalate)-based elastomers, have also been described (21, 22). These families of high-temperature, high-mechanical-strength, and rapid-processing-cycle materials, of only academic interest with $3/lb feedstocks, could become significant commodities when produced from $0.50/lb monomers.

The requirement for low-cost 4,4´-bibenzoate monomer demands a high-technology solution. The current processes for synthesizing a desired regioisomer of BBA are costly because stoichiometric amounts of expensive feedstocks are required. For example, the Suzuki coupling process requires benzene substituted with Cl and B(OH)₃. Our solution to this problem was to develop a catalytic process in which the feedstock is toluene. Any process based on a biphenyl-containing structure as starting material would fail to meet the low-cost requirements for this effort.

Toluene benefits from a consistently low commodity price and relatively high stability under most chemical processing conditions. The oxidative coupling of toluene to form dimethyl biphenyl (DMBP) was developed in the 1970s; however, this Pd-catalyzed process suffered from poor regioselectivityóall six isomers were made in significant yields (23, 24). We were successful in isomerizing the products of this process to make a simple mixture of only three isomers (3,3´-, 3,4´-, and 4,4´-DMBP). From this mixture, the desired isomer (either 3,4´- or 4,4´-DMBP) could be collected by fractional crystallization, and then the unwanted products could be recycled to extinction. This approach is similar to that used to make p-xylene. Once the desired DMPB isomer(s) are produced, they can be oxidized to the corresponding BBA monomers.

Process chemistry
Toluene coupling over Pd(II). The oxidative coupling of toluene proceeds smoothly at 150 °C and 375 psig oxygen partial pressure over a Pd(II) acetate catalyst mixed with equimolar amounts of pentanedione. Turnover rates of 15–20 h^–1 are typical with yields of the six isomers totaling 25% at similar conversions. The regioselectivity for a typical run is shown in Table 1, top; for structures of the DMBP isomers, see Figure 1.

At higher conversions, the production of the unwanted trimers and tetramers becomes significant. Our first approach to solve the regioselectivity problem was to replace pentanedione with trifluoromethanesulfonic acid (triflic acid, CF₃SO₃H) (25). Although the selectivity of this process was encouraging, the turnover rates were too small, and the palladium catalyst had to be regenerated ex situ. Subsequent work focused on isomerizing the mixture of DMBP isomers produced from the oxidative coupling of toluene over Pd(OAc)₂. Early work by Lin and McCauley showed that superacids would preferentially form the meta isomer of xylene at mild conditions (26). A similar result was described by Olah in the superacid-catalyzed isomerization of diisopropylbenzene to form the meta isomer in nearly 100% yield (27). These results prompted our attempts to isomerize the products of the toluene coupling reaction using a superacid as the catalyst.

Isomerization of DMBP in triflic acid. In separate experiments, we isomerized a mixture of all six isomers at temperatures between 25 and 100 C with acid/substrate mol ratios of 5:1 to 50:1. These model reaction studies showed that excess acid and higher temperatures favored the reaction rate and the regioselectivity to 3,4´-DMBP. Consider as one example the reaction of DMBP at 100 °C for 10 min (Table 1, center).

In a subsequent experiment, we isomerized the same substrate diluted in toluene at room temperature. For these tests, we varied the acid/substrate mol ratio from 4:1 to 180:1 (Table 1, bottom). Increasing the acid/substrate ratio from 4:1 to 180:1 at a constant reaction time of 20 h increased the conversion from ~1% to equilibrium conversion and produced a product mixture that contained only 3,3´- and 3,4´-DMBP (70% total).

Molecular modeling of isomerization
We attempted to model the isomerization reaction by a reaction mechanism shown in Figure 2. The substrate is first protonated by triflic acid to form the carbocation complex. The complex then isomerizes to form the most thermodynamically stable carbocations, which, we conjecture, determine the final product distribution. In the last step, the carbocation is reacted with water to form neutral hydrocarbons.

We calculated the energies of the protonated carbocation complexes using a molecular modeling tool known as AM-1 (28). The results of these calculations are shown at the bottom of Table 1. Note that the predictions are similar to the observed products of the equilibrated mixture. Most remarkably, both the predicted and observed results show the complete absence of isomers containing ortho methyl groups. It is the selective removal of these three isomers that makes this result important to the commercialization of the process. Moreover, such equilibrated isomerization permits ìharvestingî of the desired isomers and recycling the unwanted ones to extinction.

The isomerization kinetics of a single DMBP isomer (4,4´-DMBP) were determined in an isothermal batch reactor operating at very low conversions (<10%). We measured the initial reaction rates at each of three temperatures (5, 20, and 35 °C) using five initial concentrations of 4,4´-DMBP in an inert solvent (octane) at each temperature. The initial reaction rates were plotted versus initial substrate concentration for each isotherm (Figure 3).

The data were correlated by the following kinetic rate equation that shows shifting-order kinetics in the substrate.

R = kK[substrate]/(1 + K[substrate])

Values of the rate constant k and the equilibrium constant K were extracted for the three temperatures and plotted on a semilogarithmic plot to deduce the activation energy, 22.6 kcal/mol, and the heat of reaction, 3.2 kcal/mol. This rate equation may be placed into a continuous, stirred tank reactor (CSTR) design equation to show that 70% conversion of the substrate may be achieved in a space-time of 1.6 vol/vol/min when the reaction temperature is 100 °C. This space-time can be realized in commercial-size reactors.

Process flowchart
The overall process to convert toluene into purified DMBP products is shown in Figure 4. The process is separated into two major plants: a toluene coupling section and a DMBP isomerization section.

Toluene coupling. Toluene and air are fed to a CSTR, which is held at a nearly constant temperature of 150 °C. Two recycle streams join with this reactor to supply unreacted toluene and regenerated palladium catalyst to the reaction vessel. The reactor effluent passes to a heat exchanger and then to a pressure-reducing flash pot. Here, unused air is recovered along with some toluene and water vapor. The liquid passes to a vessel where the palladium is reduced in hydrogen, followed by a sintered metal filter to recover the Pd(II) as metallic palladium particles. This technology, developed by Ube, is used to precipitate the palladium from the hydrocarbon stream without reducing the aromatics to cycloalkanes (29).

Two filters may be used to allow continuous operation. The filters may be backflushed with chemicals to activate the palladium metal particles for subsequent oxidation by a catalytic amount of nitric acid and stoichiometric amounts of acetic acid. This palladium ion recovery is completed in a separate vessel, which is part of the palladium ion recycle loop. The hydrocarbon filtrate passes to a distillation column where toluene and water are separated from the DMBP isomers. Water is removed before the toluene returns to the coupling CSTR.

DMBP isomerization. This section may be operated to generate a purified isomer of either 3,4´-DMBP or 4,4´-DMBP. The mixture of DMBP isomers is cooled from the distillation temperature (~290 °C) to the reaction temperature (~100 °C) for isomerization in the stirred vessel. Dry triflic acid is recycled to this vessel along with the unwanted DMBP isomers that are separated from the desired products in the crystallizer. This acid–DMBP mixture is sent to a mixer–neutralizer where water and base are added. The mixture is then sent to a decanter where the aqueous and hydrocarbon phases are separated.

The hydrocarbon phase is sent to a crystallizer where the desired isomer or isomer mixture is separated from the unwanted isomers, which are returned to the isomerizer. The aqueous phase, containing the acid salt, is first sent to a distillation column to remove the water at a partial vacuum at moderate temperature (30–50 °C). Haldor–Topsoe patented the details of this process to recover dry triflic acid from its monohydrate (30). The triflic acid is now freed from its salt in a separate distillation unit that also regenerates the base used to reclaim the acid from the hydrocarbon.

The purified isomer of choice is determined by the operation of the separations and isomerization units. If these units are sized so as to make 4,4´-DMBP, then either 3,4´- or 4,4´-DMBP may be synthesized in the same units, by operating at different conditions. The main difference between the operations is the volume of recycle streams: The manufacture of 4,4´-DMBP requires higher recycle flow rates because it comprises only 10% of the coupled products leaving the toluene coupling section (see Table 1, center).

Process economics
Using the flowchart in Figure 4, we estimated the fixed capital costs and the product costs for a 50 million lb/year plant to make either of two products: >98% purity 3,4´-DMBP or >98% purity 4,4´-DMBP. These cost figures are shown in Table 2. The estimates show that 50 million lb/year of 3,4´-DMBP can be produced for an investment of $43.8 million, and the operating cost per pound of product is $0.37. The investment cost is $55.6 million for 4,4´-DMBP, and the cost per pound is $0.43. The major difference in the cost of manufacturing these compounds is the larger fixed capital investment required in the acid recovery section of the DMBP isomerizer for making 4,4´-DMBP. This larger investment is a direct consequence of having to recycle a greater amount of unwanted isomers.

3,4´-BBA´A potential rigid-rod monomer
Although the original motivation for the work was the development of low-cost 4,4´-BBA monomer, the process that we developed can clearly deliver a feedstock for that acid or for the 3,4´ isomer. Relatively little is known about the polyesters of 3,4´-BBA. Poly(ethylene 3,4´-bibenzoate) homopolymer is described as a high-temperature amorphous material, with a T_g of 104 °C (15). Our own comparison of poly(ethylene terephthalate-co-bibenzoate) isomers concluded that although the terpolymers of 4,4´-BBA were semicrystalline, rapidly crystallizing materials across the entire range of monomer concentrations, the analogous copolymers of 3,4´-BBA were amor phous at bibenzoate levels of 25 mol% (31). In that regard, 3,4´-BBA could be thought of as a higher-temperature alternative to isophthalic acid in PET-based copolymers.

Given its efficiency at disrupting crystallinity and producing high-T_g amorphous polymers, we believe that 3,4´-BBA may be an excellent feedstock for high-temperature, supertough amorphous polyamides and for high-impact polyurethanes.

Future directions
Having successfully developed highly stereoselective pro cesses for the carbonylation of toluene to p-tolualdehyde (6), and now for the oxidative coupling of toluene to dimethylbiphenyls, we believe that strong-acid catalysis can be used to upgrade relatively low-cost refinery product streams to specific molecules of interest to the polymer industry. With such a strategy, ultimate polymer costs can be reduced, bringing high-performance polymers within cost requirements for more general use. Examples recently demonstrated in our labs that could represent such opportunities include the upgrading of a low-cost refinery stream to dimethyldecalins and the reaction of ethylene with toluene to produce dimethylstilbenes. Both systems require considerable work on conversions and selectivities, but they could pave the way to low-cost routes to 2,6-naphthalenedicarboxylic acid and stilbenedicarboxylic acid.

Acknowledgments
We thank Robert Majeski for his support and encouragement of this project, as well as his expertise in developing the economic estimates. We also thank KoSa for its generous support.

References

1. East, A. J.; Golder, M. Polyesters, Thermoplastic. In Kirk-Othmer Encyclopedia of Chemical Technology, 4th ed.; Kroschwitz, J. I., Ed.; Wiley-Interscience: New York, 1996; Vol. 19, pp 641–647.
2. Plastics News, March 6, 2000, p 23.
3. Notte, P.; Poncelet, G.; Remy, M.; Pardonois, P.; VanHoecke, M. U.S. Patent 5,233,111, 1993.
4. Periana, R. A.; Schaefer, G. F. U.S. Patent 5,068,407, 1990.
5. Ward, B. C. Hoechst Celanese Corp., unpublished results.
6. Schiraldi, D. A.; Kenvin, J. C. U.S. Patent 5,910,613, 1998.
7. Lee, D.-K.; Tsai, H.-B. J. Appl. Polym. Sci. 1997, 65, 893–900.
8. Jackson, W. J.; Morris, J. C. Liquid Crystalline Polymers; ACS Symposium Series 435; American Chemical Society: Washington, DC, 1990; pp 16–32.
9. Jackson, W. J., Jr.; Morris, J. C. U.S. Patent 5,011,878, 1990.
10. Jackson, W. J., Jr.; Morris, J. C. U.S. Patent 4,959,450, 1988.
11. Jackson, W. J., Jr.; Morris, J. C. U.S. Patent 5,037,946, 1990.
12. Perez, E.; Benevente, R.; Bello, A.; Perena, J. M.; VanderHart, D. L. Macromolecules 1995, 28, 6211–6219.
13. Adams, J. Chemical Week, Dec 8, 1999, p 19.
14. Krigbaum, W. R.; Asrar, J.; Torkiumi, J.; Ciferri, A.; Preston, J. J. Polym. Sci. Polym. Lett. Ed. 1992, 20, 109–115.
15. Mang, M. N.; Brewbaker, J. L. U.S. Patent 5,138,022, 1991.
16. Alms, G. R.; Samuels, M. R.; Waggoner, M. G. U.S. Patent 5,250,654, 1992.
17. Morris, J. C.; Jackson, W. J., Jr. U.S. Patent 5,057,595, 1990.
18. Morris, J. C.; Jackson, W. J., Jr. U.S. Patent 5,037,947, 1990.
19. Morris, J. C.; Jackson, W. J., Jr. U.S. Patent 5,037,946, 1990.
20. Gupta, B.; Halke, W.; Shepherd, J. P.; Sawyer, L. C. World Patent 95/17446, 1995.
21. Lee, H.-B.; Liu, J.-L.; Tsai, Y.-S. J. Appl. Polym. Sci. 1996, 59, 1027–1031.
22. Lee, H.-B.; Lee, D.-K.; Liu, J.-L.; You, J.-W. J. Polym. Res. 1995, 2, 109–114.
23. Itatani, H.; Yoshimoto, H. Chem. Ind. 1971, 674.
24. Iataaki, H.; Yoshimoto, H. J. Org. Chem. 1973, 38, 76.
25. Xu, B.-Q.; Sood, D.; Iretski, A. V.; White, M. G. J. Catal. 1999, 187, 358–366.
26. McCaulay, D. A.; Lien, A. P. J. Am. Chem. Soc. 1952, 74, 6246–6250.
27. Olah, G. A. U.S. Patent 4,547,606, 1984.
28. Sherman, S. C.; Iretskii, A. V.; White, M. G.; Gumienny, C.; Tolbert, L. M.; Schiraldi, D. A. J. Am. Chem. Soc., submitted for publication, 2000.
29. Itatani, H.; Yoshimoto, H.; Shiotani, A.; Yokota, A.; Yoshikiyo, M. U.S. Patent 4,294,976, 1979.
30. Hommeltoft, S. I. U.S. Patent 5,759,357, 1998.
31. Brown, D. M.; Clary, R. S.; Lee, C. D.; Monroe, W. G., IV; Vaughn, D. A.; Ragheb, R. T.; Erter, J. W., III; Schiraldi, D. A. Polym. Preprints 2000, 41, 123–124.

Sheldon C. Sherman is an undergraduate student in the School of Chemistry and Biochemistry at the Georgia Institute of Technology, and he expects to earn a B.S. degree in chemistry in December 2001. He has been employed by the School of Chemical Engineering for 5 years as a research associate.

Alexei V. Iretskii is a postdoctoral researcher in the School of Chemical Engineering, Georgia Institute of Technology. He received his undergraduate degree in chemistry and his Ph.D. in coordination chemistry from the Institute of Technology, St. Petersburg, Russia. His research interests include organometallic and coordination chemistry and their applications to catalysis.

Mark G. White is a professor of chemical engineering at the Georgia Institute of Technology (School of Chemical Engineering, Atlanta, GA 30332-0100; 404-894-2822; mark.white@che.gatech.edu), where he has taught for 23 years. He received a B.S. degree from the University of Texas, Austin, an M.S. degree from Purdue University, and a Ph.D. from Rice University, all in chemical engineering. His research interests include homogeneous and heterogeneous catalysis.

David A. Schiraldi is the lead chemist at KoSa (1551 Sha Lane, Spartanburg, SC 29307; 864-579-6549; dschiral@earthlink.net) in long-range polymer chemistry R&D, and he is responsible for KoSa’s leveraged research programs with universities. He is an associate editor for the Journal of Applied Polymer Science, a Chemical Innovation Heart Cut contributor, and a member of the International Advisory Board of the Journal of Industrial and Engineering Chemistry. He received a B.S. degree in chemistry from the University of California, San Diego, and a Ph.D. in organometallic chemistry from the University of Oregon.

Return to Top