Kinetic Investigations of p-Xylene Oxidation to Terephthalic Acid with a

Article pubs.acs.org/IECR

Kinetic Investigations of p‑Xylene Oxidation to Terephthalic Acid with a Co/Mn/Br Catalyst in a Homogeneous Liquid Phase Meng Li,†,§ Fenghui Niu,§ Daryle H. Busch,‡,§ and Bala Subramaniam*,†,§ †

Department of Chemical and Petroleum Engineering, ‡Department of Chemistry, and §Center for Environmentally Beneficial Catalysis, University of Kansas, Lawrence, Kansas 66045, United States S Supporting Information *

ABSTRACT: Kinetic investigations of the liquid phase oxidation of p-xylene (pX) to terephthalic acid (TPA) with Co/Mn/Br catalyst were performed in a stirred 50 mL Parr reactor at 200 °C and 15 bar pressure under conditions wherein product precipitation is avoided. The oxidant (O2) was introduced by sparging into the liquid phase at constant gas-phase O2 partial pressure. Apparent kinetic rate constants, estimated by regressing experimental conversion data to a pseudo-first order lumped kinetic model, are at least an order of magnitude greater than those reported in the literature for similar catalytic reactions. We attribute this difference to the presence of gas−solid and liquid−solid mass transfer resistances in the previous studies wherein the TPA product precipitates as it forms, trapping intermediate products and slowing down their oxidation rates. Our results also indicate that it is not possible to completely eliminate the gas−liquid mass transfer limitations associated with the fast intermediate oxidation steps, even when operating without solids formation and at high stirrer speeds. Other types of reactor configurations are therefore needed to better overcome gas−liquid mass transfer limitations. Systematic studies of bromide concentration effects show that the observed reaction rates become zero order in bromide concentration at sufficiently high bromide levels where the elimination of intermediate 4-(bromomethyl)benzoic acid by oxidation is favored. Further, the rate constants do not show any statistically significant dependence on pX concentration as suggested in other reports involving the traditional three-phase gas−liquid−solid reaction system. This again confirms that the formation of a solid phase hinders the overall oxidation rate, resulting in much smaller apparent rate constants.

1. INTRODUCTION Terephthalic acid (TPA) is a commercially important aromatic compound used mainly as a major precursor of polyethylene terephthalate (PET) polymer, which is produced by polycondensation of ethylene glycol with TPA. More than 90% of the worldwide TPA production goes to make PET, consumed primarily for the manufacture of polyester fibers (dominant use in Asia), solid-state or bottle-grade resins (dominant use in North America and Europe) and polyester film. Worldwide demand for TPA is steadily growing at a rate of 7−8%, with current worldwide capacity exceeding 50 million metric tons per annum.1 The Asia region accounts for about 70% of the current world TPA capacity.2 Most of the commercial TPA processes are based on the core technology originally developed by Mid-Century Corporationa liquid phase bromine-promoted catalytic oxidation of p-xylene (pX) that uses either air or molecular oxygen.3 The well-known Amoco (acquired by BP in 1997) Mid-Century (MC) process is currently the leading TPA technology.4 It involves a stirred oxidation reactor in which air is dispersed into the liquid phase containing pX and Co/Mn/Br based catalyst dissolved in aqueous acetic acid.5 The pX oxidation is typically performed at 190−205 °C and 15−30 bar.6 The reactors are lined with titanium to resist corrosion by hydrobromic acid present in the reaction mixture. The crude TPA solid from the oxidation reactor contains 1000−4000 ppm of 4-carboxybenzaldehyde (4-CBA) due to incomplete oxidation.7 Most polyester applications require TPA that contains less than 25 ppm 4-CBA and less than 150 ppm p-toluic acid (p-TA),8 both © 2013 American Chemical Society

of which are chain terminators in the polymerization process for PET production. A hydrogenation purification step is therefore required to upgrade the crude TPA to high purity polymer-grade TPA.9 The hydrogenation of 4-CBA is typically carried out in a fixed-bed reactor containing a carbon-supported palladium catalyst to quantitatively convert 4-CBA to p-TA at 275−290 °C and 70−90 bar.7,9 Polymer-grade TPA product is recrystallized from solution leaving the p-TA behind in solution. Currently, the main commercial TPA technology license holders and licensors are BP, DuPont, Dow Chemical, Mitsubishi Chemical, Eastman Chemical, Mitsui Chemicals, Interquisa, Hitachi and Grupo Petromex. BP is currently the world’s largest TPA producer with an annual capacity of 7.5 million metric tons in 2011.10 We recently reported a spray reactor concept as a greener alternative to the conventional MC process to produce highpurity TPA ( 97 wt %) (solvent), biphenyl (internal standard). Ultrahigh-purity (UHP) grade oxygen and industrial grade CO2 were purchased from Airgas and Linweld, respectively. 2.2. Apparatus and Procedure. The experimental apparatus employed in this work is shown in Figure 1. The 50 mL titanium stirred vessel (Parr Instrument Company, series 4560 mini Bench Top Reactor) is equipped with a magnetically driven impeller for stirring, a dual probe J-type thermocouple, and an OMEGA transducer for measuring the reactor temperature and pressure, respectively. The temperature and pressure are continuously recorded and controlled with the help of a LabVIEW data acquisition system. The reactor was modified to accommodate multiple inlet/outlet ports for (a) introducing O2 into the liquid phase via a dip tube/sparger, (b) sampling the liquid phase via a 1/16″ titanium dip tube, (c) withdrawing the vapor phase, and (d) providing a pressure relief valve. In a typical experimental run, the reactor was initially charged with a 35 mL solution of pX, Co/Mn/Br catalyst, and biphenyl 9019

dx.doi.org/10.1021/ie403446b | Ind. Eng. Chem. Res. 2014, 53, 9017−9026

Industrial & Engineering Chemistry Research

Article

→ TPA reaction). Thus, the O2 concentration in the liquid phase dominated by acetic acid is maintained constant. Accordingly, we adopted the pseudo-first-order kinetic model for the sequential reactions involving O2. The model equations based on these assumptions are given below:

standard (215 nm for pX and 254 nm for all other products). The HPLC samples were prepared from those collected during the reaction, and 15 mL of acetic acid was added to each sample. Figure 2 shows a typical chromatogram obtained from HPLC analysis of a sample. The carbon balance estimated from only the concentrations of the aromatic compounds in the liquid phase was found to be >95%.

dc1 = −k1c1 dt

3. LUMPED KINETIC SCHEME AND BATCH REACTOR MODEL None of the reported lumped kinetic models for pX oxidation to TPA account for 4-(bromomethyl)benzoic acid (BPTA), significant amounts of which are detected in our experiments. Considering its possible important role in the process, the formation and disappearance of BPTA are also included in our proposed lumped kinetic scheme as a parallel reaction, as shown in Figure 3. BPTA is formed by the substitution reaction

dc 2 = k1c1 − k2c 2 dt dc 3 = k 2c 2 − k 3c3 − k5c3 dt dc4 = k5c3 − k6c4 dt dc5 = k 3c3 + k6c4 − k4c5 dt

dc6 = k4c5 dt

with the following initial conditions: c1 = c10 ,

c 2 = c3 = c4 = c5 = c6 = 0

where the subscripts 1−6 denote pX, TALD, p-TA, BPTA, 4CBA, and TPA, respectively, as shown in Figure 3. Athena Visual Studio30 software was used to estimate the kinetic parameters. A nonlinear least-squares optimization method was implemented in the software and encoded using Fortran. The convergence criteria are as follows:

Figure 3. Proposed lumped kinetic scheme for pX oxidation to TPA.

Sik − Sik + 1

involving p-TA and HBr, and is oxidized in the presence of oxygen to 4-CBA. The proposed chemical reaction equations associated with the formation and oxidation of BPTA are shown in Figure 4.

Sik

≤ 10−3

n

Si ∑ (ci ,E − ci ,p)2 1

where Si is the sum of squares of component concentration residuals; cE and cP are experimental and predicted concentrations, respectively; n is the number of experimental data points (n = 10, in this case); and k is the iteration number. The following observations, based on published literature, provide valuable guidance for our experimental and modeling studies. It is generally believed that the p-TA → 4-CBA reaction is the slowest step in the overall oxidation sequence.21,22 If this is indeed the case, then the estimated first-order rate constant for this step (k3) should be several-fold less than the other rate constants in the sequential oxidation, and the overall oxidation rate would be the rate of the slowest step. In such a case, it is possible to assess the presence or absence of gas−liquid mass transfer limitations associated with the slowest step from the influence of stirrer speed on the estimated rate constant for that step. It must be recognized that the estimated rate constants associated with the nonrate determining steps could be merely mass transfer coefficients, since the O2 transport into the liquid phase is also a first-order rate process.

Figure 4. Scheme for formation and oxidation of BPTA.

The reported power law kinetic models for pX oxidation to TPA mainly assume pseudo-first-order oxidation kinetics (in excess of O2). Cao et al.19 applied the first order kinetic model at low pX conversions wherein no solid phase was formed; that is, the system was considered homogeneous. Wang et al.25 also deduced from their fractional kinetic model that when the reactant (pX) concentration was low enough compared to O2, the first order kinetic model was valid. In our experiments, TPA precipitation is avoided and the O2 concentration was maintained in excess (at least 5-fold with respect to the pX

4. RESULTS AND DISCUSSION 4.1. Temperature and Pressure Profiles. Figure 5 shows the temperature and pressure profiles corresponding to run 3 (Table 1). Since low enough initial pX concentration (0.025 9020



Article

(beyond the range of uncertainty) and tends to level off around 1000 rpm. In contrast, the other rate constants in the oxidation sequence (k1, k2, and k4) do not vary in this stirrer speed range. These observations suggest that while it is possible to overcome gas−liquid mass transfer limitations associated with the p-TA → 4-CBA step in the stirred reactor, the kinetics associated with other rates are indeed much faster as reported in previous studies. In addition, the lowest rate constant (k6) is invariant in this stirrer rpm range suggesting that it is not influenced by gas−liquid mass transfer limitations. Note that there is an apparent decrease in rate constants with a further increase in the stirring rate to 1400 rpm. This is probably due to the formation of a deep meniscus at the higher agitation speed that decreases the liquid level below that of the sparger (either partly or fully) used for introducing O2 into the reactor. Hence, we conclude that the maximum rate of TPA formation in this stirred reactor occurs beyond 1000 rpm. Figure 6 shows the experimentally observed concentrations at various stirrer speeds. The fitted concentration profiles at 1000 rpm (solid line) match the experimental data quite well. Gas−liquid mass transfer coefficient for the pX oxidation system at conditions corresponding to Run No. 4 in Table 1 (at 1200 rpm) was estimated. Details of estimation are provided in the Supporting Information. The product of the gas−liquid mass transfer coefficient (kL) and the gas−liquid interfacial area per unit volume (av), kLav, was estimated to be 65 min−1. The fact that the mass transfer rate constant (kLav) is roughly an order of magnitude greater than the rate constant associated with the slowest oxidation step (k3) suggests that gas−liquid mass transfer limitations associated with the p-TA → 4-CBA step are eliminated at these operating conditions. However, the gas−liquid mass transfer rate constant is of the same order of magnitude as other rate constants in the oxidation sequence (k1, k2, and k4), suggesting that O2 saturation of the liquid phase might not have been achieved even at high stirring speeds. In such a scenario, any further increase in the overall TPA formation rate may be achieved only with a reactor configuration that provides enhanced gas−liquid mass transfer area such that O2 saturation of the liquid phase is not ratelimiting. This is achieved in a spray reactor in which the liquid phase is dispersed as fine droplets into a continuous gas phase.13 4.3. Comparison of Estimated Rate Constants with Literature Data. Table 3 compares the estimated reaction rate constants in this work with the reported data from literature. Even though different reaction conditions and kinetic models are employed, the relative rate constants of the various steps

Figure 5. Temporal temperature and pressure profiles during intrinsic kinetic studies of pX oxidation in a stirred reactor.

M) was charged in the reactor, the reactor temperature rise (caused by the heat of reaction) was within 1 °C, which allowed us to assume that the reactor was operated under isothermal conditions. By replenishing O2 as needed from an external reservoir, the reactor pressure was maintained constant (15 bar) except for brief, insignificant pressure fluctuations during sampling. 4.2. Effect of Stirring Rate. The intrinsic kinetic rates of the sequential steps during pX oxidation under MC process conditions (200 °C, 15 bar) are very fast and tend to be limited by gas−liquid mass transfer resistance. The experiments (Runs 1−5 of Table 1) were carried out in the stirred reactor to evaluate the effect of stirring speed on the reaction rates. Figure 6 shows the temporal product distributions at various stirring speeds. Effective rate constants, estimated from the reactant and product concentration profiles as explained previously, are shown in Table 2. Clearly, the slowest step in the series reaction sequence is the formation of 4-CBA from p-TA, and that in the parallel reaction is the oxidation of BPTA to 4-CBA. At all stirrer speeds ranging from 600 to 1400 rpm, the rate constants associated with these steps (k3 and k6) are several-fold lower than those associated with the other steps. The uncertainties associated with the larger rate constants are greater, as might be anticipated with the measurement of faster reaction rates. For instance, because the first two reactions in the oxidation sequence are so fast, the sampling errors associated with TALD and p-TA will be substantially more. In contrast, the uncertainties associated with the rate constants of the slowest steps (k3, k5, and k6) are much lower. For the sequential oxidation steps, the rate constant associated with 4-CBA formation from p-TA (k3) increases from 600 to 800 rpm

Figure 6. Experimental and simulated product distributions at different stirring speeds [reaction conditions: T = 200 °C, P = 15 bar; initial pX = 25 mM, Co = 12.5 mM, Mn = 12.5 mM, Br = 32.5 mM; O2/CO2 (mol/mol) = 1:1]. 9021



Article

Table 2. Estimated Reaction Rate Constants at Different Stirring Speeds (Initial [pX] = 0.025 M; All Other Experimental Conditions Are as Shown in Table 1) k1 (min−1)

stirrer speed (rpm) 600 800 1000 1200 1400 a

11.7 11.8 11.7 13.6 13.9

(1.72) (1.96) (1.85) (2.31) (2.75)

a

k2 (min−1) 33.2 37.1 39.0 42.1 36.8

k3 (min−1)

(18.9) (26.0) (27.1) (32.9) (28.9)

3.90 4.45 5.89 5.56 4.49

(0.548) (0.764) (1.10) (1.03) (0.793)

k4 (min−1) 17.5 18.6 22.5 23.2 20.5

(4.78) (5.24) (6.83) (6.77) (5.91)

k5 (min−1) 4.12 4.71 6.00 5.81 4.88

(0.568) (0.799) (1.10) (1.05) (0.844)

k6 (min−1) 0.421 0.423 0.414 0.435 0.401

(0.0405) (0.0450) (0.0406) (0.0364) (0.0369)

95% confidence level shown in parentheses.

Table 3. Comparison of Estimated Rate Constants with Literature Data model type

k1 (min−1)

k2 (min−1)

k3 (min−1)

k4 (min−1)

ref

fractional kinetic model

0.176 0.732 1.9(10−3) 0.38

0.725 1.422 7.6(10−2) 0.99

0.0361 0.139 6.6(10−4) 0.123

0.338 0.873 1.86(10−2) 0.44

22 26 19 23

11.74

39.03

5.889

22.50

power law assumption

a

this worka

remarks parameter re-estimated for Wang et al.’s model22 first order in reactant concentrations 0.65 order in pX concentration, first order in other reactant concentrations run no. 3 in Table 1

Stirrer speed = 1000 rpm; initial [pX] = 0.025 M; all other operating conditions are as listed in Table 1.

Figure 7. Color changes observed in the reaction solution due to the formation of cobalt bromides31 at various bromide concentrations in acetic acid: [Co(II)] = 12.5 mM.

follow the same trend: k3 < k1 < k4 < k2. Our investigations concur with previous reports that the slowest step is the oxidation of p-TA to 4-CBA. However, the estimated reaction rate constants in this work are at least an order of magnitude greater than those reported in the referenced studies. We attribute this to the fact that the reaction was performed under homogeneous conditions in this work, avoiding product precipitation in the liquid phase and thereby eliminating gas− solid and liquid−solid mass transfer limitations. It is worth pointing out that Cincotti et al.21 found that their first order kinetic model fits the experimental data well even at initial pX concentrations ([pX] = 4.33−9.75 M) that are higher than industrial conditions. The reaction was performed at low temperatures (80−130 °C) to slow down the kinetics, and the first order kinetic model matches well with experimental data prior to the formation of the solid products. On the basis of this evidence, it seems plausible that the apparent change in order at high pX concentrations reported in some reported studies is

due to mass transfer limitations stemming from intermediate products being trapped in the solids. 4.4. Effect of Bromide Concentration. Different bromide concentrations (runs 3 and 6−10 of Table 1) were used in the kinetic study experiments for pX oxidation to TPA to investigate the effect of bromide concentration on the reaction rate. Figure 7 shows the color change of the reaction solution corresponding to the various bromide concentrations in run 3 and runs 6−10 in Table 1, respectively. The solution changes color from blue to pink when bromide concentration is changed from 34.8 mM to 0.91 mM. The reason of the color change is probably associated with the formation of cobalt bromides.31 Figure 8 shows the component concentrations with different bromide concentrations. Surprisingly, high concentrations of BPTA were detected during the pX oxidation. Although it is well recognized that the p-TA to 4-CBA conversion is the rate determining step, the slowest step is BPTA oxidation according to the estimated rate constants (Table 4). The decrease of the 9022



Article

Figure 8. Experimental and simulated product distributions with various bromide concentrations (initial [pX] = 0.025 M; stirring rate =1000 rpm; all other experimental conditions are as shown in Table 1).

Table 4. Estimated Reaction Rate Constants at Different Bromide Concentrations (Initial [pX] = 0.025 M; Stirring Rate =1000 rpm; All Other Experimental Conditions Are as Shown in Table 1)a CBr (mM) 34.8 15.8 9.44 4.93 2.40 0.91 a

k1 (min−1) 11.7 11.8 11.6 8.25 6.45 7.80

(1.91) (1.60) (2.28) (2.36) (1.70) (2.66)

a

k2 (min−1) 39.0 39.6 45.6 56.3 17.6 14.5

(27.1) (25.1) (43.2) (48.2) (11.0) (8.92)

k3 (min−1) 5.89 (1.10) 7.75 (1.45) 5.64 (1.27) 2.43 (0.648) 0.927 (0.183) 0.419 (0.0719)

k4 (min−1) 22.5 29.3 25.4 11.8 7.84 3.87

(6.83) (10.3) (12.8) (6.51) (4.08) (1.93)

k5 (min−1)

k6 (min−1)

6.00 (1.10) 7.69 (1.42) 4.17 (0.955) 1.17 (0.413) 0.223 (0.119) 0.0585 (0.0477)

0.414 (0.0405) 0.307 (0.0227) 0.261 (0.0359) 0.205 (0.0913) 0.152 (0.125) 0.0658 (0.0599)

95% confidence level shown in parentheses.

attains a maximum, increased when lowering the cobalt concentration. It is reported that as much as 99% of the initial inorganic bromide is converted to benzylic bromide and its concentration remains approximately the same until the oxidation of the hydrocarbon is nearly complete.34 Partenheimer,35,36 as well as Saha and Espenson,37 also found that a decrease in Co/Mn/Br catalytic activity is attributed to the benzylic bromide formation and that benzylic bromide, unlike the inorganic bromide, has virtually no activity or promotional effect in Co/Mn/Br autoxidations. The mechanism of BPTA oxidation has not been well studied. Metelski et al.34 developed a proposed

bromide concentration will slow down the BPTA disappearance rate. Partenheimer32 illustrated the role of bromide in the Co/ Mn/Br catalyst system during the oxidation of pX to TPA. When bromide is added, there is a large increase in activity and selectivity due to the rapid electron transfer from cobalt to bromide. Kamiya et al.33 investigated the effect of bromine/ metals ratio on the reaction rate and found that at a cobalt concentration of 0.05 M, the pX → TPA oxidation rate rapidly increased to a maximum value as the molar Br/Co ratio was increased to 1 and remained constant up to a ratio of 9. The optimum Br/Co ratio, at which the pX → TPA oxidation rate 9023



Article

invariant when the confidence intervals are taken into account. This invariance is to be expected in the case of intrinsic rate constants. However, as shown in Figure 10, the rate constants

mechanism for benzyl bromide oxidation (Figure 9). According to this mechanism, benzyl bromide is oxidized in the presence

Figure 10. Effect of initial pX concentration on the estimated rate constants23

are reported to decrease with the increase of initial pX concentration in a three-phase reaction system.23 This again confirms that the formation of a solid phase introduces mass transfer resistances, thereby lowering the rates and associated rate constants.

5. CONCLUSIONS The kinetics of the liquid phase oxidation of p-xylene (pX) to terephthalic acid (TPA) was investigated in a stirred reactor under homogeneous conditions (i.e., low pX concentrations employed to avoid TPA precipitation). A first order kinetic model was developed based on a lumped reaction scheme that includes a parallel step describing the formation of 4(bromomethyl)benzoic acid (BPTA). Kinetic rate constants regressed from the experimental data are shown to be at least an order of magnitude greater than those reported in the literature. This is attributed to the fact that the conversion data were obtained under homogeneous conditions in this study (unlike previous literature data that were obtained in gas− liquid−solid systems) avoiding solid formation in the liquid phase, and thereby eliminating gas−solid and liquid−solid mass transfer limitations. Our investigations concur with previous reports that the slowest step is the oxidation of p-toluic acid to 4-carboxybenzaldehyde. Systematic investigations of the effects of stirrer speed reveal that it may not be possible to completely eliminate the gas−liquid mass transfer limitations in even a laboratory scale stirred reactor. This further confirms that other reactor configurations (such as a spray reactor in which the

Figure 9. Proposed scheme for the autoxidation of benzyl bromide (Adapted from ref 34. Copyright 2000 American Chemical Society).

of Co(III), leading to the recovery of bromide ions. Simultaneously, Co(III) is reduced to Co(II) by hydrogen bromide (faster) and also by benzyl bromide, albeit at a much slower rate. Partenheimer35 reported that the competing mechanisms of oxidation and solvolysis for the disappearance of benzylic bromide occur at approximately the same rates under the studied experimental conditions. 4.5. Effect of Substrate Concentration. Experiments with different pX concentrations (runs 6, 11, and 12 of Table 1) were carried out to study the effect of substrate concentration on the regressed rate constants. Solubility studies show that approximately 1.6 g of TPA is completely dissolved in 100 g of acetic acid at 200 °C and 15 bar,12 suggesting a maximum pX concentration of 0.1 M to maintain a homogeneous liquid phase. Table 5 shows the estimated rate constants. It was found that the rate constants at the different concentrations are

Table 5. Estimated Reaction Rate Constants at Different pX Concentrations (P = 15 bar; T = 200 (°C; Stirring Rate =1000 rpm; All Other Experimental Conditions Are As Shown in Table 1)a run no. 12 6 11 a

CpX,0 (mM) 14 25 35

k1 (min−1) 11.8 (2.56) 11.8 (1.60) 11.0 (2.10)

a

k2 (min−1)

k3 (min−1)

k4 (min−1)

k5 (min−1)

k6 (min−1)

40.1 (42.0) 39.6 (25.1) 38.9 (29.9)

5.20 (1.27) 7.75 (1.45) 8.81 (2.36)

18.73 (6.60) 29.3 (10.3) 30.3 (16.9)

5.64 (1.37) 7.69 (1.42) 5.87 (1.61)

0.225 (0.0255) 0.307 (0.0227) 0.302 (0.0445)

95% confidence level shown in parentheses. 9024



Article

(3) Landau, R.; Heights, R.; Saffer, A. Process for the preparation of terephthalic acid. U.S. Patent 2833818, 1958. (4) Tomás, R. A. F.; Bordado, J. C. M.; Gomes, J. F. P. p-Xylene oxidation to terephthalic acid: A literature review oriented toward process optimization and development. Chem. Rev. 2013, DOI: 10.1021/cr300298j. (5) Raghavendrachar, P.; Ramachandran, S. Liquid-phase catalytic oxidation of p-xylene. Ind. Eng. Chem. Res. 1992, 31, 453−462. (6) Weissermel, K.; Arpe, H.-J. Industrial Organic Chemistry; VCH Publishers, Inc.: New York, 1997. (7) Ure, A. M.; Parker, D. Methods, processes, and system for treating and purifying crude terephthalic acid and associated process streams. WO 2010/122304 A1, 2010. (8) Hashmi, S. M. A.; Al-Luhaidan, S. Process for preparing purified terephthalic acid. EP 1671938 A1, 2006. (9) Meyer, D. H. Fiber-grade terephthalic acid by catalytic hydrogen treatment of dissolved impure terephthalic acid. U.S. Patent 3584039, 1971. (10) BP. Annual Report and Form 20-F 2011. bp.com/annualreport (accessed Jan. 2013). (11) Subramaniam, B.; Busch, D. H.; Niu, F. Spray process for selective oxidation. WO 2010/111288 A2, 2010. (12) Li, M. A Spray reactor concept for catalytic oxidation of p-xylene to produce high-purity terephthalic acid. Ph.D. Dissertation, University of Kansas, Lawrence, KS, 2013. (13) Li, M.; Niu, F.; Zuo, X.; Metelski, P. D.; Busch, D. H.; Subramaniam, B. A spray reactor concept for catalytic oxidation of pxylene to produce high-purity terephthalic acid. Chem. Eng. Sci. 2013, 104, 93−102. (14) D’oro, P. C.; Danoczy, E.; Roffia, P. Low temperature oxidation of p-xylene. Oxid. Commun. 1980, 1 (2), 153−162. (15) Morbidelli, M.; Paludetto, R.; Carra, S. Gas−liquid autoxidation reactors. Chem. Eng. Sci. 1986, 41, 2299−2307. (16) Jacobi, R.; Baerns, M. The effect of oxygen transfer limitation at the gas-liquid interphase. Kinetics and product distribution of the pxylene oxidation. Erdoel Kohle Erdgas P. 1983, 36 (7), 322−326. (17) Sun, W.; Pan, Y.; Zhao, L.; Zhou, X. Simplified free-radical reaction kinetics for p-xylene oxidation to terephthalic acid. Chem. Eng. Technol. 2008, 31 (10), 1402−1409. (18) Cao, G.; Pisu, M.; Morbidelli, M. A lumped kinetic model for liquid-phase catalytic oxidation of p-xylene to terephthalic acid. Chem. Eng. Sci. 1994, 49 (24B), 5775−5783. (19) Cao, G.; Servida, A.; Pisu, M. Kinetics of p-xylene liquid-phase catalytic oxidation. AIChE J. 1994, 40 (7), 1156−1166. (20) Cincotti, A.; Orrù, A.; Cao, G. Effect of catalyst concentration and simulation of precipitation processes on liquid-phase catalytic oxidation of p-xylene to terephthalic acid. Chem. Eng. Sci. 1997, 52 (21/22), 4205−4213. (21) Cincotti, A.; Orrù, A.; Cao, G. Kinetics and related engineering aspects of catalytic liquid-phase oxidation of p-xylene to terephthalic acid. Catal. Today 1999, 52, 331−347. (22) Wang, Q.; Li, X.; Wang, L.; Cheng, Y.; Xie, G. Kinetics of pxylene liquid-phase catalytic oxidation to terephthalic acid. Ind. Eng. Chem. Res. 2005, 44, 261−266. (23) Yan, X. Data mining macrokinetic approach based on ANN and its application to model industrial oxidation of p-xylene to terephthalic acid. Chem. Eng. Sci. 2007, 62, 2641−2651. (24) Partenheimer, W. Methodology and scope of metal/bromide autoxidation of hydrocarbons. Catal. Today 1995, 23, 69−158. (25) Wang, Q.; Cheng, Y.; Wang, L.; Li, X. Semicontinuous studies on the reaction mechanism and kinetics for the liquid-phase oxidation of p-xylene to terephthalic acid. Ind. Eng. Chem. Res. 2007, 46, 8980− 8992. (26) Dong, Y.; Yan, X. Hybrid model of industrial p-xylene oxidation incorporated fractional kinetic model with intelligent models. Ind. Eng. Chem. Res. 2013, 52, 2537−2547. (27) Qian, F.; Tao, L.; Sun, W.; Du, W. Development of a free radical kinetic model for industrial oxidation of p-xylene based on artificial

liquid is the dispersed phase) are needed to completely overcome gas−liquid mass transfer limitations. The effects of bromide and initial pX concentrations on the reaction rate were also investigated. The reaction rate shows mixed order dependence on bromide concentration; positive order at relatively low values and zero order at higher bromide concentrations where the intermediate BPTA elimination by oxidation is favored. Decreases in the reaction rate constants with increasing pX concentration, reported in traditional threephase reaction systems, were not observed under the homogeneous liquid phase conditions of our study. This further confirms that the presence of the solid TPA phase hinders the reaction rate and the kinetic rate constants reported in the literature are influenced by mass transfer limitations.

6. TRIBUTE TO PROFESSOR MASSIMO MORBIDELLI On the occasion of Professor Morbidelli’s 60th birthday, we wish him good health and cheer. We hope that this contribution is a fitting tribute to Professor Morbidelli’s outstanding contributions to the field of chemical engineering and to his profession. The diversity of topics in which Professor Morbidelli has made impactful contributions, ranging from parametric sensitivity in chemical reactors to adsorption-based separations by simulated moving beds to polymer reaction engineering, is truly impressive. It should come as no surprise that he has also published in the area of p-xylene oxidation to TPA, the subject of this dedication. It is especially a pleasure for one of the authors (B.S.) to have enjoyed a special friendship with Professor Morbidelli, sharing memorable times during their graduate student days at the University of Notre Dame, a sabbatical stay at ETH, Zürich, and a visit to Politecnico di Milano. It is indeed a distinct privilege to be invited to contribute to this special issue honoring him.

■

ASSOCIATED CONTENT

S Supporting Information *

Details of the estimation of the gas−liquid mass transfer coefficient at the reactor operating conditions; transient concentration data used in Figures 6 and 8. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: 785-864-2903. Fax: 785864-6051. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was supported in part by the National Science Foundation (EEC-0310689) and in part by the U.S. Department of Agriculture (2011-10006-30362). We gratefully acknowledge valuable discussions with Dr. Peter Metelski (BP Products North America Inc.) regarding the effects of bromide concentration on the reaction kinetics.

■

REFERENCES

(1) ICIS. ICIS pX, PTA, MEG Market Outlook Webinar: Asia and Middle East Markets. Personal contact (communicated May 2013). (2) ICIS. http://www.icis.com/Articles/2007/11/06/9076461/ purified-terephthalic-acid-pta-uses-and-market-data.html (accessed Feb. 2013). 9025



Article

neural network and adaptive immune genetic algorithm. Ind. Eng. Chem. Res. 2012, 51, 3229−3237. (28) Abrams, K. J. Process for preparing aromatic carboxylic acids with efficient energy recovery by the oxidation of aromatic hydrocarbon feedstocks. WO 96/11899 A1, 1996. (29) Viola, A.; Cao, G. A rapid direct analysis of p-xylene oxidation products by reverse phase HPLC. J. Chromatogr. Sci. 1996, 34, 27−33. (30) Athena Visual Studio, v 14.2; Athena Visual Software, Inc.: Naperville, IL, 2012. (31) Sawada, K.; Tanaka, M. Formation of bromo complexes of cobalt(II) in acetic acid. J. Inorg. Nucl. Chem. 1977, 39 (2), 339−344. (32) Partenheimer, W. A Chemical Model for the Amoco “MC” Oxygenation Process to Produce Terephthalic Acid. Catalysis of Organic Reactions; Marcel Dekker: New York, 1990; pp 321−346. (33) Kamiya, Y.; Nakajima, T.; Sakoda, K. Autoxidation of p-xylene catalyzed with cobalt monobromide in acetic acid. Bull. Chem. Soc. Jpn. 1966, 39 (10), 2211−2215. (34) Metelski, P. D.; Adamian, V. A.; Espenson, J. H. Mechanistic role of benzylic bromides in the catalytic autoxidation of methylarenes. Inorg. Chem. 2000, 39, 2434−2439. (35) Partenheimer, W. The complex synergy of water in the metal/ bromide autoxidation of hydrocarbons caused by benzylic bromide formation. Adv. Synth. Catal. 2004, 346, 297−306. (36) Patenheimer, W. The complex synergy of water in the metal/ bromide autoxidations. Part II. Effect of water and catalyst on the aerobic oxidation of benaldehydes and the effect of water on the elementary catalytic pathways. Adv. Synth. Catal. 2005, 347, 580−590. (37) Saha, B.; Espenson, J. H. Low-bromide containing MC catalyst for the autoxidation of para-xylene. J. Mol. Catal. A 2007, 271, 1−5.

■

NOTE ADDED AFTER ASAP PUBLICATION This paper was published ASAP on December 26, 2013. Minor text corrections were made to the Tribute section, and the corrected version was reposted on January 2, 2014.

9026


Kinetic Investigations of p-Xylene Oxidation to Terephthalic Acid with a

Recommend Documents