Ind. Eng. Chem. Res. 1992,31, 2456-2459
2456
at temperatures near the critical temperature. Dibenzop-dioxin appears later in the reaction network. 2. The phenoxyphenols and biphenol react on roughly the same time scale as phenol, but dibenzofuran has a higher persistence and is more resistant to oxidative degradation in SCW. 3. The primary reactions leading to the multiring products accounted for 43% of the phenol that initially reacted at 380 O C and 278 atm. Thus, the formation of undesired and potentially hazardous higher molecular weight compounds constitutes an appreciable portion of the initial reaction chemistry under these conditions. 4. The formation of potentially hazardous reaction products is a factor to consider in the design of SCW oxidation pmeasea for phenolic wastes. Further research is required to identify the process variables and process confiiations that minimize the formation of such products. Acknowledgment We thank Doug LaDue for experimental assistance. This project was supported by the National Science Foundation (CTS-8906860, CTS-8906859, and CTS9015738) and the Shell Faculty Career Initiation Fund. H, WStw NO. PhOH, 108-95-2; o - P ~ O C ~ H ~ O2417-10-9; p-PhOC6H40H, 831-82-3; o-OHC6H4C6H4-o-OH,1806-29-7; dibenzofuran, 132-64-9; dibenzo-p-dioxin, 262-12-4.
Born, J. G. P.; Louw, R.; Mulder, P. Formation of Dibenzodioxins and Dibenzofurans in Homogeneous Gas-Phase Reactions of Phenols. Chemosphere 1989,19,401-406. Haar, L.; Gallagher, S.; Kell, G. S.NBSINRC Steam Tables; Hemisphere Publishing: Washington, DC, 1984. Modell, M. Processing Methode for the Oxidation of Organics in Supercritical Water. U.S. Patent 4,338,199, July 6, 1982. Modell, M. Processing Methods for the Oxidation of Organics in Supercritical Water. US. Patent 4,543,190, Sept 24, 1985. Modell, M. Supercritical-Water Oxidation. In Standard Handbook of Hazardous Waste Treatment and Disposal; Freeman, H. M., Ed.; McGraw-Hik New York, 1989; Section 8.11. Nileson, C.; Andersaon, K.; Rappe, C.; Westermark, S. Chromatographic Evidence for the Formation of Chlorodioxina from Chloro-2-Phenoxyphenole. J. Chromutogr. 1974, W,137-147. Shaub, W. M.; Tsang, W. Dioxin Formation in Incinerators. Environ. Sci. Technol. 1983, 17, 721-730. Thornton, T. D. Phenol Oxidation in Supercritical Water: Reaction Kinetics, Products, and Pathways. Ph.D. Thesis,The University of Michigan, 1991. Thornton, T. D.; Savage, P. E. Phenol Oxidation in Supercritical Water. J. Supercrit. Fluids 1990, 3, 240. Thornton, T. D.; Savage, P. E. Kinetics of Phenol Oxidation in Supercritical Water. AZChE J. 1992, 38, 321-327. Thornton, T. D.; LaDue, D. E. IIJ; Savage, P. E. Phenol Oxidation in Supercritical Water: Formation of Dibenzofuran, Dibenzo-pdioxin,and Related Compounds. Environ. Sci. Technol. 1991,25, 1507.
Yang, H. H. Homogeneous Catalysis in the Oxidation of p-Chlorophenol in Supercritical Water. Ph.D. Thesis, The University of Illinois, 1988. Yang, H. H.; Ekkert, C. A. Homogeneoua Catalysis in the Oxidation of p-Chlorophenol in Supercritical Water. Znd. Eng. Chem. Res. 1988, 27, 2009.
Literature Cited Bhore, N.; Klein, M. T.; Bischoff, K. B. The Delplot Technique: A New Method for Reaction Pathway Analysis. Ind. Eng. Chem. Res. 1990, 29, 313.
Received for review March 23, 1992 Revised manuscript received August 3, 1992 Accepted August 17, 1992
Kinetics of the Catalyzed Supercritical Water-Quinoline Reaction Zhuangjie Li and Thomas J. Houser* Chemistry Department, Western Michigan University, Kalamazoo, Michigan 49008-3842
The kinetics of the catalyzed reaction between supercritical water and quinoline was studied over the temperature range of 400-500 "C.The reaction rate is first-order with respect to both quinoline and ZnCla (catalyst) and inversely proportional to water concentration. These observations are consistent with a Langmuir-Hinshelwood heterogeneous mechanism which involves competitive adsorption on the catalytic surface, with water much more strongly adsorbed. The Arrhenius parameters for the rate constant were an activation energy of 112 kJ/mol and preexponential fador of 1.7 X los mol/(L*g.s). This activation energy is well below the C-C and C-N bond energies in aromatic heterocycles. Introduction The possible use of supercritical fluid extraction (SFE) of coal to obtain cleaner, more versatile fluid products has been of significant interest. Some fluids have the opportunity to participate as reactants at process conditions, which may yield extracts of very different compositions than those obtained from other treatments and which will be dependent on the fluid used. Thermodynamic consideration of SFE leads to the prediction that the enhanced solubility (volatility) of the solute may be several orders of magnitude (Gangoli and Thodos, 1977; Williams, 1981; Whitehead and Williams, 1975). Thus,this method combines many of the advantages of distillation with those of extraction. Because of this interest in SFE and in the destruction of hazardous materiala by supercritical water (SW)oxidation, several studies have been reporting some of the basic chemistry that may be taking place during coal
extraction and oxidation at these conditions (Houser et al., 1986,1989;Abraham and Klein, 1985; Tounsend and Klein, 1985; Lawson and Klein, 1985; Helling and Tester, 1987,1988; Thornton and Savage, 1990; Jin et al., 1990). One of the major concern of the current program ie the removal of nitrogen from organic model compounds thought to be representative of structures found in fossil fuels and that may be present during the destruction of hazardous materials. Because of the difficulty of removing heterocyclic nitrogen, experiments were initiated by extensively examining the reactivities of quinoline (Q)and isoquinoline, as well as briefly examining the reactivities of other compounds (Houser et al., 1986). The selection of water as the fluid was based on its physical and chemical properties (Frank, 1968). Zinc chloride was chosen as a catalyst because of its reported catalytic activity for hydrocracking aromatic structures (Salim and Bell, 1984).
0888-5886f 92f 2631-2456$03.O0f 0 0 1992 American Chemical Society
Ind. Eng. Chem. Res., Vol. 31, No. 11, 1992 2457 Table I. Kinetic Data"
quinoline (mL) time (h) Ireaction NH3 yield ko x l@/s k X 1V (mol/(L.gs)) quinoline (mL) time (h) % reaction ko x 106/s k x 10' (mol/(L.g.s)) volatile product yields ((mol of product/ mol of quinoline reacted) X 100) toluene ethylbenzene xylene aniline phenol mixture lb mixture 2' quinaldine mixture 3d
400 "C, 260 bar 2 2 6 12 12 21 32 20 5.9 5.5 3.5 3.2
2 24 42 33 6.3 3.7
2 48 58 38 5.0 3.0
24 45 31 6.9 4.1
4 24 43 28 6.5 3.8
450 "C,349 bar ,2 2 3 6 20 38 2.07 2.21 12.2 13.0
2 12 57 1.95 11.5
2 24 81 1.92 11.3
1 6 30 1.65 9.7
4 6 30 1.65 9.7
1.0 1.3 7.0 10.4 6.0 4.0 15.7 9.1 32
0.9 0.9 0.9 4.6 14.4 6.3 3.0 10.1 6.5 23
0.7 0.6 0.5 2.7 15.3 9.3 1.6 6.4 5.9 47
0.7 1.4 1.6 2.9 4.0 4.3 4.0 7.4 1.4 25
0.6 1.2 1.0 4.2 2.0 8.0 6.0 11.4 10.0 39
0.31 0.60
0.10 0.39
0.12 0.20
0.42 0.60
0.26 0.56
1.4 1.0 2.0 2.5 0 5.0 2.5 8.6
1.1
2.8 21
3"
nonvolatile product yields (g/g quinoline reacted) char tar
1
500 "C, 435 bar quinoline (mL) time (h) % reaction NH3 yield
ko x 105/s k x 10' (mol/(L.gs))
1 3 65 42 9.7 57
2 3 62 56 9.0 53
4
3 69 43 10.8 64
0.5 6 79 60 7.2 42
1 6 79 65 7.2 42
2 6 75 72 6.4 38
4 6 79 68 7.2 42
"All experiments used 0.2 g of ZnC1, catalyst and (H,O) of 11.8 mol/L. k = ko (H,O)/g of ZnClz. bThe values for mixtures were estimated assuming an average response factor for the components. Mixture 1 consists of indane, cresols, toluidine, and methylindans. Mixture 2 consists of xylenols, xylidines, and a little naphthalene. Mixture 3 consists of methylquinoline isomers other than quinaldine (2-methylquinoline) and of dimethylquinoline isomers.
This paper presents the kinetic data for the reaction of SW with Q.
and MS fragmentation pattern.
Experimental Section The experiments were carried out in a small (47 cm3) cylindrical, stainless steel, batch reactor, which was not equipped with a valve for the collection of gaseous products for analysis or for the addition of an inert gas for pressurization. The reactor was loaded with the appropriate amounts of Q, water, and ZnC12 catalyst, and then the reactor was purged with argon and bolted closed using a copper gasket. The reactor was placed in a preheated, fluidized sand bath furnace for the required reaction time; about 15 min was required to reach 375 O C . Following reaction, the vessel was air cooled and opened, the reaction mixture was removed, and the water and organic layers were separated. Portions of methylene chloride solvent were used to rinse the reactor and extract the water layer. These portions were combined with the organic layer and additional solvent was added to a standard volume for quantitative determinations made gas chromatographically (GC) using peak area calibrations from known solutions. The components for these solutions were identified mass spectrometrically (MS). In all experiments there were certain limitations on the GC-MS product determinations. Some components could not be separated completely, and these are reported as a total yield of mixture using an average calibration factor. In addition, some products are reported as an isomer of a probable structure as deduced from the molecular weight
Results Table I summarizes the results of the temporal and initialQ concentration variations on the extents of reaction at the three temperatures used. The apparent firsborder rate constants, ko,show no significant trend with time or initial concentration (Co), and their scatter indicates the reproducibility of the data for this complex reaction. Possible causes for these variations may be small differences in temperature due to reactor location in the sand bath and/or differenma in exposed surface area of catalysts due to agitation of the reactor. However, an examination of the results from other kinetic studies in SW indicates that variations in extents of reaction of 10% are not unusual (e.g., Abraham and Klein, 1985; Thornton and Savage, 1990), which would be more than enough to account for our reproducibility. Calculated 1/2-order and 3/2-orderrate constants showed trends with changes in C, that were independent of temperature. The 3/2-orderrate constants were approximately inversely proportional to Co'/2,whereas the 2-orderconstants were approximately proportional to Col 2. Thus, the fmtrorder dependence of the rate on Q is confirmed. The product distributions at 450 "C that are tabulated are typical, and similar results were obtained at 400 and 500 OC. It should be noted that phenol appears to be a secondary product, whereas aniline and alkylquinolines, quinaldine (2-methylquinoline),and mixture 3 are formed as intermediates, their yields first increasing and then
7
2468 Ind. Eng. Chem. Res., Vol. 31, No. 11, 1992 Table 11. Effect of Water Concentration on Rate" waterconcn calcd added % k, X lo5 k X lo' (mol/L) press. (bar) ZnCl, (a) reacted (s-') (mol/(L.a.s)) 2.36 11.2 0.2 64 9.44 314 1.95 11.5 0.2 57 11.8 349 1.51 10.8 0.2 48 14.2 379 0.1 42 1.70 9.0 268 7.11 0.1 32 0.89 8.4 314 9.44 0.83 9.7 0.1 30 11.8 349 0.64 9.0 0.1 24 14.2 379 aConditions for these experiments were time 12 h, temperature 450 "C, and 2.00 mL of quinoline.
decreasing with time. In general, the NH, yields are higher at 500 "C, indicating that nitrogen removal is more effective at higher temperatures. The chars were methylene chloride insoluble solids, and the tars were nonvolatile materials dissolved in the solvent. The results of varying the water concentration at a fixed temperature and reaction time are presented in Table 11. The product of ko and water concentration is reasonably constant at constant catalyst loading. Thus, the rate appears to be inversely proportional to water concentration. It should be noted that previous data show that SW increases the rate over that of pyrolysis (Houser et al., 1989). The effect of the catalyst amount on the rate is shown in Table 111. It is assumed that the catalysis is a heterogeneous process with ZnCl, as a liquid (mp 283 "C); thus the amounts are reported in grams rather than concentration units. This assumption appears reasonable since the solubilities of similar salts in SW at 450 "C are well below the smallest loading used, 0.83 w t %. For example, the solubilities at 450 "C and 250 bar were about 0.03 wt % for NaCl (Bischoff and Pitzer, 1989) and about 0.0005 wt % for CaCl, (Martynova, 1976). The data show that the rates of reaction are proportional to catalyst loading; the apparent first-order rate constants are proportional to the mass of ZnCl,, other conditions being held constant. In addition, larger amounts of ZnC1, are somewhat more effective in removing the nitrogen as ammonia. Taking into consideration the concentration dependencies of the rate, the rate equation for the consumption of Q is of the form rate = -d(Q)/dt = k(Q)(ZnC12)/(H20) (1) The average rate constants are summarized in Table IV and have the temperature dependence described by eq 2: k = 1.7 X lo5 exp(-13530 f 1830/T) (mol/(L.g.s)) (2) Discussion A possible reaction scheme to account for the products was previously reported (Houser et al., 1986) and thus has not been repeated here. The observation of methylquinolines and of dimethylanilines (xylidines) and dimethylphenols (xylenols) demonstrate that CHI ( x = 1-3) groups are present for methylations, probably as radicals. Apparently after an initial rupture of the CN bond in the 1,2 position of Q, the hydrocarbon side chain is shortened, giving riee to theae CHI species. The effectiveness of ZnCl, to catalyze the disappearance of Q and facilitate the removal of amino groups from anilines to form primarily Table 111. Catalyst Effect" water pressure (bar) ZnCh ( 9 ) time (h) % reaction NH3 yield ( % ) ko X 1O6/8 k x lo4 (mol/(L.ge))
349 0.1 12 30 26 0.83 9.7
349 0.2 12 57 23 1.95 11.5
349 0.4 12 79 47 3.61 10.7
349 0.8 12 94 43 6.51 9.6
"All experiments were at 450 O C and with 2 mL of quinoline.
phenols (Houser et al., 1986) indicates a probable ionic mechanism is also operating, at least with those reactions involving the nitrogen. The first kinetics concern is that of pressure effects on the rate. Since the change in pressure is caused by a change in reactant concentration, H20, it is not possible to identify the effects due to pressure alone (the addition of an inert pressurizing gas was not possible with the current system). Our observation that the pseudo-firstorder rate constant decreases with increasing water pressure is in sharp contrast to most other examples of rate constants determined in supercritical solvents reported in the literature (Chateauneuf et al., 19911, indicating negative volumes of activation (AV') for those reactions. Our results are consistent with the effect of water pressure on the V206-cataly& oxidation of 1,4-dichlorobenzenein SW (Jin et al., 1990). It has been shown (Eckert et al., 1986) that although solutes in solvents (CO, and C2H4) near their critical point exhibit very large negative partial molal volumes which may produce large negative AV* for some reactions, at relative pressures and temperatures about the same as those in our study the partial molal volumes approach zero. However, since these kinetics and partial molal volume studies dealt with homogeneous systems and primarily nonpolar solvents, their applicability to the current study is of questionable validity. Thus, we believe the effect of water concentration is most likely due to adsorption properties rather than to the AV'. It should be noted that the assumption that a negative AV* is consistent with rate data can be misleading. An examination of the oxidation of phenol data (Thornton and Savage, 1990), as reported in their Figure 6, shows that the initial rates appear independent of pressure, within experimental uncertainties. The water pressure effects appear to be on the equilibrium position. A second concern is that of the heterogeneity of the reaction. If there was a significant homogeneous contribution to the rate (catalyzed or uncatalyzed), then a linear least squares treatment of the pseudo-first-order rate constants vs ZnC1, loading listed in Table I11 should yield a nonzero, positive intercept. The intercepts obtained ( N O 5 ) were 0.09 f 0.63 and 0.10 f 0.10 s-l for pressures of 349 and 379 bar, respectively, indicating a negligible homogeneous contribution to the rate. The observations that water increases the rate of disappearance of Q over that of pyrolysis (Houser et al., 1989) but that the rate is inversely proportional to water concentration are consistent with a Langmuir-Hinshelwood surface-catalyzed mechanism, i.e., competitive adsorption of Q and water (Laidler, 1965). The rate equation is of the form rate
k&$w(Q)(H20)/[1
+
+ &(H20)l2
(3)
where k, is the surface rate constant and ICs are adsorption equilibrium constants for the reactants. If Kw(H,O) >> [ l + Kq(Q)],the equation reduces to the form of eq 1with the catalyst surface area included in k,. Finally, the activation energy of about 112 kJ/mol is about 4 times lower than the eathated C 4 and C-N bond 379 0.1 12 24 32 0.64 9.0
379 0.2 12 48 27 1.51 10.8
379 0.4 12 70 46 2.79 9.9
379 0.8 12 90
50 5.33 9.5
349 0.2 6 38 32 2.21 13.0
349 0.4 6 63 47 4.60 13.5
349 0.8 6
86 45 9.10 13.4
Ind. Eng. Chem. Res., Vol. 31, No. 11,1992 2459 Table IV. Average Rate Constants T (K) k X 10" (mol/(L.a.s)) 673 3.55 0.42 723 10.6 0.8 773 48 9
* *
Nb
6 18 7
"The f values are the 95% confidence limits of the mean. *Number of experiments.
energies of 490 and 445 kJ/mol, respectively, in aromatic heterocyclic rings (Houser et al., 1980). Thus, the homogeneous rupture of the C-N bond is not rate controlling and a concerted elimination of a small molecule is very unlikely in view of the products formed. Even a homogeneous radical chain reaction cannot have an E, less than half the bond energy of the initial bond rupture. It can be concluded that this low activation energy is consistent with a catalytic process, which, in view of chloride salt solubilities, is also heterogeneous.
Summary and Conclusions The major conclusions that are derived from this work are as follows: 1. The reaction of Q with SW is heterogeneously catalyzed by ZnClp with a negligible contribution from homogeneous reactions. 2. The rate is best described by an equation first order in both Q and ZnCl:, and inverse first order in water. This is consistent with a Langmuir-Hinshelwood surface-catalyzed, bimolecular mechanism with competitive adsorption between Q and water. The o b ~ e activation d energy, 112 kJ/mol, is well below the CN bond energy in Q, which is also consistent with a heterogeneous mechanism. 3. The product distribution's dependence on reaction time indicate that the initial step in the reaction is the rupture of the Q CN bond in the 1,2 position followed by fragmentation of the hydrocarbon side chain to provide species for ring alkylation. ZnClzis effective in promoting the removal of nitrogen, some of which was replaced by OH. Both higher temperatures and higher catalyst loadings gave more complete nitrogen removal in the form of ammonia. Acknowledgment Supported by, or in part by, the US. Army Research Office. Literature Cited Abraham, M. A,; Klein, M. T. Pyrolysis of Benzylphenylamiie Neat and with Tetralin, Methanol, and Water Solvents. Znd. Eng.
Chem. Prod. Res. Dev. 1985,24, 300. Bischoff, J. L.; Pitzer, K. S. Liquid-vapor Relations of the System NaC1-H20: Summary of the P-T-X Surface From 300 to 500. Am. J. Sci. 1989,289, 217. Chateauneuf, J. E.; Roberts, C. B.; Brennecke, J. F. Laser Flash Photolysis Studies of Benzophenone in Supercritical COz. Presented at the American Institute of Chemical Engineera 1991Annual Meeting, Los Angeles, CA; American Institute of Chemical Engineers: New York, 1991; paper 91a. Eckert, C. A.; Ziger, D. H.; Johnston, K. P.; Kim, S. Solute Partial Molal Volumes in Supercritical Fluids. J. Phys. Chem. 1986,90, 2738. Frank, E. U. Supercritical Water. Endeavour 1968,27, 55. Gangoli, N.; Thodos, G. Liquid Fuels and Chemical Feedstocks from Coal by Sumrcritical Gas Extraction. Znd. EM. - Chem. Prod. Res. Dev. i977,'16, 208. Helling, R. K.; Tester, J. W. Oxidation Kinetics of Carbon Monoxide in SuDercritical Water. Enerav Fuels 1987. I. 417. Helling,-R. K.; Tester, J. W. Oxidation of Simple Compounds and Mixtures in Supercritical Water: Carbon Monoxide, Ammonia, and Ethanol. Environ. Sci. Technol. 1988,22, 1319. Houser, T. J.; McCarville, M. E.; Biftu, T. Kinetics of the Thermal Decomposition of Pyridine in a Flow System. Znt. J. Chem. Kinet. 1980, 12, 555. Houser, T. J.; Tiffany, D. M.; Li, Z.; McCarville, M. E.; Houghton, M. E. Reactivity of Some Organic Compounds with Supercritical Water. Fuel 1986,65,827. Houser, T. J.; Tsao, C.-C.; Dyla, J. E.; VanAtten, M. K.; McCarville, M. E. The Reactivity of Tetrahydroquinoline, Benzylamine and Bibenzyl with Supercritical Water. Fuel 1989, 68, 323. Jin, L.; Shah, Y. T.; Abraham, M. A. The Effect of Supercritical Water on the Catalytic Oxidation of 1,4-Dichlorobenzene. J. Supercrit. Fluids 1990,3, 233. Laidler, K. J. Reactions on Surfaces and in the Solid State. In Chemical Kinetics; McGraw-Hik New York, 1965; p 274. Lawson, J. R.; Klein, M. T. Influence of Water on Guaiacol Pyrolysis. Znd. Eng. Chem. Fundam. 1985,24,203. Martynova, 0. I. Solubility of Inorganic Compounds in Subcritical and Supercritical Water. In High Temperature, High Pressure Electrochemistry in Aqueous Solutions; de G. Jones, D., Staehle, R. W., Eds.; National Association of Corrosion Engineers: Houston, TX, 1976; pp 131-138. Salim, S. S.; Bell, A. T. Effects of Lewis Acid Catalysts on the Hydrogenation and Cracking of Three-Ring Aromatic and Hydroaromatic Structures Related to Coal. Fuel 1984, 63, 469. Thornton, T. D.; Savage, P. E. Phenol Oxidation in Supercritical Water. J. Supercrit. Fluids 1990, 3, 240. Tounsend, S. H.; Klein, M. T. Dibenzyl Ether as a Probe into the SupercriticalFluid Solvent Extraction of Volatile8 from Coal with Water. Fuel 1985,64,635. Whitehead, J. C.; Williams, D. F. Solvent Extraction of Coal by Supercritical Gases. J. Znst. Fuel 1975, 182. Williams, D. F. Extraction with SupercriticalGases. Chem. Eng. Sci. 1981,36, 1769. Receiued for review March 18, 1992 Revised manuscript received July 27, 1992 Accepted August 21, 1992