Ind. Eng. Chem. Res. 1997, 36, 1391-1400
1391
Supercritical Water Oxidation Kinetics, Products, and Pathways for CH3- and CHO-Substituted Phenols Christopher J. Martino and Phillip E. Savage* Department of Chemical Engineering, The University of Michigan, Ann Arbor, Michigan 48109-2136
Phenols bearing -CH3 and -CHO substituents were oxidized in supercritical water at 460 °C and 250 atm. Experiments with each compound explored the effects of the reactor residence time and the concentrations of the phenolic compound and oxygen on the reaction rate. These experimental data were fit to global, power-law rate expressions. The resulting rate laws showed that the reactivity of the different isomers at 460 °C was in the order of ortho > para > meta for both compounds. Moreover, the CHO-substituted phenol was more reactive than the analogous CH3-substituted phenol, and all of these substituted phenols were more reactive than phenol itself under supercritical water oxidation conditions. Identifying and quantifying the products of incomplete oxidation allowed us to assemble a general reaction network for the oxidation of cresols in supercritical water. This network comprises three parallel primary paths. One path leads to phenol, a second path leads to a hydroxybenzaldehyde, and the third path leads to ring-opening products. The hydroxybenzaldehyde reacts through two parallel paths, which lead to phenol and to ring-opening products. Phenol also reacts via two parallel paths, but these lead to phenol dimers and ring-opening products. The dimers are eventually converted to ring-opening products, and the ring-opening products are ultimately converted to CO2. The relative rates of the different paths in the reaction network are strong functions of the location of the substituent on the phenolic ring. Introduction Supercritical water oxidation (SCWO) is a wastetreatment process that converts organic carbon to CO2 at reaction conditions that exceed the critical point of water (Tc ) 374 °C, Pc ) 218 atm). SCWO can treat a variety of wastes, including hazardous organic compounds (Modell, 1989) and sludges (Shanableh and Gloyna, 1991; Blaney et al., 1995). Water above its critical point is a good solvent for both oxygen and organic compounds, which allows these components to react quickly in a homogeneous fluid phase. Information about the kinetics and byproducts from oxidation of real aqueous pollutants is essential for the design, optimization, and control of reliable commercial SCWO reactors. A knowledge of the reaction kinetics allows one to calculate the residence times required for a desired destruction and removal efficiency in a commercial SCWO reactor. A knowledge of the identities and yields of reaction products allows one to identify processing conditions that minimize the production of undesired intermediate byproducts. Savage et al. (1995) reviewed previous SCWO research with model pollutants, and that review showed that phenolic compounds are the model pollutants studied most extensively at SCWO conditions. This attention is a reflection of phenol and substituted phenols having high water solubilities, which cause them to be prevalent in a wide variety of industrial waste waters (e.g., Yen et al., 1982; Jevtitch and Bhattacharyya, 1986). Phenol is the pollutant for which the SCWO kinetics, pathways, and mechanisms have been studied most thoroughly. Since the initial studies of phenol SCWO (Wightman, 1981), the kinetics of phenol oxidation in near-critical and supercritical water have been explored by several investigators (Thornton and Savage, 1992a; * Corresponding author. E-mail:
[email protected]. Phone: (313) 764-3386. Fax: (313) 763-0459. S0888-5885(96)00697-5 CCC: $14.00
Gopalan and Savage, 1995a,b; Rice et al., 1993; Kranjc and Levec, 1996; Tufano, 1993). The reaction network for phenol SCWO proceeds through the two parallel primary reaction pathways of ring opening and dimer formation (Thornton and Savage, 1992b; Gopalan and Savage, 1995b), and a mechanism for phenol SCWO has been proposed (Gopalan and Savage, 1994, 1995a). The kinetics for SCWO reactions of other phenolic compounds have received much less attention. Li et al. (1993) provide reaction kinetics and byproduct information for SCWO of o-chlorophenol, Yang and Eckert (1988) reported a kinetics study for p-chlorophenol SCWO, and Martino et al. (1995) provided reaction kinetics and byproduct information for o-cresol (omethylphenol). This last report showed that salicylaldehyde, or o-hydroxybenzaldehyde, was a major primary reaction product from o-cresol SCWO. This result placed value on a knowledge of the reactions of hydroxybenzaldehydes in supercritical water. The limited information available on the reactions of substituted phenols in supercritical water and the potential importance of hydroxybenzaldehydes as byproducts in the SCWO reaction network for phenolic compounds motivated this present study of the oxidation of CH3- and CHO-substituted phenols in supercritical water. A secondary motivation for the work reported here is a recognition of the value of structure-reactivity correlations. Studying a single class of compounds, such as substituted phenols, might allow one to develop a structure-reactivity relationship for the SCWO of that compound class. That is, one might be able to correlate the oxidation kinetics with some reactivity index that depends upon the identity of the substituent and its position on the aromatic ring. A structure-reactivity relationship can be expected for compounds with analogous reaction networks that proceed through the same rate-limiting step. The literature does not currently provide sufficient experimental data to undertake the task of exploring structure-reactivity issues for SCWO of real pollutants. © 1997 American Chemical Society
1392 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997
Experimental Section All of the organic compounds used in this study were obtained from the Aldrich Chemical Company. The nominal purities were 99% for o-, m-, and p-cresol, 98% for o- and p-hydroxybenzaldehyde, and 97% for mhydroxybenzaldehyde. These nominal purities were taken into account when calculating concentrations of reactor feedstocks and calibrating analytical equipment. The oxidation experiments were performed in an isothermal, isobaric, plug-flow reactor constructed with Hastelloy C276 tubing. The reactor system contained separate 1/16 in. (1.59 mm) o.d. and 1.08 mm i.d. preheat sections for aqueous feed streams containing the organic compound and oxygen. The preheated feed streams mixed at the entrance to the reactor, which was either a 1- or a 4-m length of 1/8 in. (3.18 mm) o.d. and 1.40 mm i.d. tubing. Two different lengths were used to facilitate exploring a sufficiently wide range of residence times. The preheat lines and reactor tubing were immersed in a temperature-controlled fluidized bath of aluminum oxide particles operating at the desired reaction temperature. The reaction temperature was measured at the start of the reactor tube at the point where the two preheated streams are mixed. The reactor effluent was rapidly cooled and returned to ambient conditions prior to analysis. After the reactor reached steady-state conditions, the gaseous products were analyzed with an on-line gas chromatograph (GC) equipped with a thermal conductivity detector, and liquid effluent samples were collected for analysis. We used a high-performance liquid chromatography (HPLC) system described elswhere (Martino and Savage, 1997) to measure the concentrations of the phenolic compounds in both the reactor feed and in the aqueous reactor effluent. Other aqueous-phase reaction intermediates were identified from analyses employing a GC with a mass spectrometric detector and then quantified from analyses employing a GC with a flame ionization detector. The reactor system and analytical protocol are described in much greater detail elsewhere (Martino et al., 1995; Thornton and Savage, 1990). SCWO Global Kinetics This section provides the results of SCWO experiments with cresols and hydroxybenzaldehydes, and an analysis of those results, which led to global reaction rate laws for each compound. Our approach is to model the global kinetics using a power-law rate expression of the form
rate ) Ae-Ea/(RT)[organic]a[O2]b[H2O]c
We then fit the experimental data to this nonlinear algebraic equation to estimate numerical values for the rate-law parameters. Cresol Oxidation Kinetics. We previously performed 51 oxidation experiments using o-cresol as the organic reactant (Martino et al., 1995). The experiments were performed at various reactor conditions, with temperatures ranging from 350 to 500 °C and pressures ranging from 200 atm. to 300 atm. Thirtyone of the experiments were performed at 460 °C and 250 atm. In this section we revise our previously reported rate equation because we discovered that the o-cresol conversion was not measured accurately by our previous HPLC analysis. One of the major products from o-cresol SCWO, o-hydroxybenzaldehyde, eluted from the HPLC column at the same time as the o-cresol reactant. This previously undiscovered overlap of reactant and product peaks caused the measured concentration of o-cresol in the reactor effluent to be consistently high, which led to conversions that were systematically low. We are able to correct this problem and calculate accurate values for the o-cresol conversions because the molar yields of o-hydroxybenzaldehyde had been previously measured by a complementary GC analysis. These data, along with the HPLC response factors for o-cresol and o-hydroxybenzaldehyde, were used to calculate corrected values for the o-cresol conversion. Table 1 shows the revised values for the o-cresol conversion and the carbon tally, which we calculate as the percentage of carbon atoms flowing into the reactor that appear in the reaction products we quantify. This correction lowers the carbon tally values by an average of about 5%. A good indication of the effect of this correction on the carbon tally can be obtained by noting that the correction reduced the mean carbon tally from 102.6% to 97.4% for the 30 experiments at 460 °C and 250 atm where the carbon tally could be calculated. The standard deviation of the carbon tally decreased from 11.4% to 9.8%. The parameters in the global rate law for o-cresol disappearance were reevaluated using these revised data and the SimuSolv software package (Steiner et al., 1990), which employs the maximum likelihood method for parameter estimation. The resulting global rate equation is
roCR ) 107.1(2.4 exp
(-33 700RT( 9600) ×
[oCR]0.54(0.14[O2]0.35(0.22[H2O]1.46(0.69 (3)
(1)
Such simple power-law rate expressions possess utility in engineering applications such as design, economic evaluation, and computational fluid dynamics studies (Oh et al., 1996). We restrict our kinetics experiments to those wherein the oxygen concentration remains essentially constant throughout the reaction (O2 present in at least 200% stoichiometric excess), in which case the differential equation that describes the reactant conversion in an isothermal, isobaric, plug-flow reactor can be solved analytically to yield
X)1b c 1/1-R (1 + (a - 1)Ae-Ea/(RT)[organic]a-1 0 [O2]0[H2O]0τ) for a * 1 (2)
The rate has units of moles per liter per second, the concentrations are in moles per liter at the reaction conditions, and the activation energy is in calories per mole. The rate law parameters that changed the most by using the corrected conversions are the oxygen reaction order, which increased from 0.22 ( 0.22 to 0.35 ( 0.22, and the Arrhenius preexponential factor, which increased from 105.7(2.2 to 107.1(2.4. The uncertainties given here and elsewhere in this paper are the 95% confidence intervals. Table 2 presents the details of the nonlinear regression results, including the variances, σ2i and covari2 ances, σi,j , for each of the regression parameters. These variances and covariances are required to calculate the uncertainty in reaction rates determined from eq 3.
Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1393 Table 1. Corrected Values for the Conversion of o-Cresol for the SCWO Experiments Reported in Martino et al. (1995) concentration
corrections
temp. (°C)
press. (atm)
residence time (s)
o-cresol (µmol/L)
O2 (mmol/L)
reptd. conv. (%)
conv. (%)
carbon tally (%)
350.0 370.0 380.2 380.9 379.6 400.2 420.4 420.6 420.5 440.3 460.1 460.2 460.2 460.3 459.0 460.0 460.4 460.0 459.9 460.0 460.0 460.0 460.0 460.3 460.2 460.3 460.2 460.7 461.1 459.9 459.9 460.1 460.1 460.0 462.3 459.9 460.0 460.4 459.5 460.5 460.9 460.0 459.9 460.0 459.8 460.0 459.8 460.1 459.5 480.9 499.7
250 250 250 250 250 250 250 251 250 250 200 220 230 240 250 250 250 250 250 250 250 250 250 251 250 250 250 250 250 250 250 250 250 251 251 251 250 252 250 250 250 250 250 251 251 260 270 280 300 250 250
29.5 25.1 15.6 20.7 46.3 8.0 4.6 6.3 13.2 5.5 3.6 4.1 4.4 4.7 0.5 0.6 0.9 0.9 0.9 1.2 1.2 2.0 2.1 3.7 4.3 4.4 4.4 5.3 5.3 5.4 5.4 5.5 5.5 5.9 7.3 7.7 7.7 7.7 7.9 9.0 11.0 12.0 12.2 12.6 12.9 5.3 5.7 6.0 6.9 4.5 4.3
588 543 413 412 416 173 123 128 120 117 81.7 79.1 99.1 89.6 180 44.7 327 46.6 159 106 30.7 30.9 313 117 253 23.5 184 269 116 23.1 437 364 44.2 156 260 170 419 44.3 103 314 102 459 172 45.3 22.0 101 126 113 151 92.6 86.8
42.6 36.6 31.4 29.0 32.4 11.6 9.25 9.13 9.28 7.82 5.20 5.92 6.31 6.73 5.19 4.03 9.07 3.90 5.85 7.52 5.08 9.23 9.31 6.64 10.8 1.19 4.87 10.1 7.10 1.21 7.80 9.22 2.46 5.44 10.3 5.09 8.09 2.49 7.44 9.30 7.37 7.44 4.91 2.41 1.27 7.59 8.05 8.51 9.61 6.55 6.15
98.4 98.0 72.7 78.1 99.5 30.4 18.7 29.6 51.7 44.0 40.5 42.5 44.3 41.0 10.8 20.3 6.4 23.9 7.2 29.4 47.0 66.3 32.7 56.3 34.2 61.0 47.3 34.5 76.0 63.6 40.9 41.9 67.9 67.4 52.6 76.7 55.1 83.4 83.4 58.5 96.4 72.1 94.7 97.6 88.1 52.3 64.5 62.5 77.0 85.0 98.3
98.6 98.4 74.9 80.3 99.6 33.8 26.1 34.6 54.5 51.2 49.6 55.2 53.0 48.3 12.9 24.1 9.6 30.5 11.3 34.0 52.3 74.0 36.5 64.3 38.0 67.0 57.6 44.3 83.8 68.6 44.6 50.9 72.3 73.5 61.7 85.0 60.0 85.5 87.0 65.3 97.9 75.0 96.8 98.4 90.1 59.0 67.3 69.8 82.2 87.7 98.6
52.5 49.5 61.5 55.4 64.5 86.1 110.2 98.0 86.9 95.0 112.6 124.3 112.2 108.5 102.9 99.2 n.d.a 108.4 110.4 103.5 116.1 105.2 89.3 95.8 99.2 91.7 101.4 115.8 102.6 86.4 90.9 103.0 94.2 96.4 100.5 101.9 88.0 76.0 102.0 98.4 96.0 77.3 87.1 82.5 99.7 105.5 123.9 106.8 101.5 93.4 97.3
a
n.d ) not determined.
We investigated the oxidation of the m- and pmethylphenols in supercritical water at a single temperature and pressure of T ) 460 °C and P ) 250 atm. Tables 3 and 4 provide the experimental conditions and the results of the 13 SCWO experiments for m-cresol and the 12 SCWO experiments for p-cresol. Because all of the experiments were conducted at a single temperature and single water density, we could not determine values for the Arrhenius parameters or the water reaction order, c, in eq 1. Accordingly, the data for m- and p-cresol were modeled using a simpler global rate expression of the form a
rate ) k[organic] [O2]
b
(4)
The nonlinear algebraic expression analogous to eq 2 that relates the conversion to the process variables and
the parameters in the rate law of eq 4 is b 1/1-a X ) 1 - (1 + (a - 1)10k′[organic]a-1 0 [O2]0τ) for a * 1 (5)
We used SimuSolv to regress the values for the reaction orders a and b and the parameter k′. The nonlinear regression routine converged more readily when we used the parameter k′ instead of k, where k′ ) log10 k. The resulting parameters and their 95% confidence intervals are
rmCR ) 10-1.43(0.52[mCR]0.65(0.09[O2]0.61(0.25
(6)
for m-cresol and
rpCR ) 10-1.55(0.44[pCR]0.60(0.07[O2]0.54(0.19 for p-cresol.
(7)
1394 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Table 2. Regression Statistics for the Parameters in the o-Cresol Global Rate Expression log A Ea a b c 2 σlog A 2 σEa σ2a σ2b σ2c 2 σlog A,Ea 2 σlog A,a 2 σlog A,b 2 σlog A,c σE2 a,a σE2 a,b σE2 a,c σ2a,b 2 σa,c 2 σb,c
7.080 3.370 × 104 0.5381 0.3485 1.461 1.467 2.332 × 107 4.615 × 10-3 1.212 × 10-2 1.167 × 10-1 5714 -7.871 × 10-3 2.787 × 10-2 3.503 × 10-1 -52.08 62.46 1525 -5.327 × 10-3 -5.770 × 10-4 -5.681 × 10-3
Table 3. Experimental Conditions and Results from SCWO of m-Cresol at 460 °C and 250 atm initial concn. res. time (s)
m-cresol (µmol/L)
O2 (mmol/L)
conv. (%)
carbon tally (%)
3.5 5.0 5.0 5.0 5.1 5.1 7.4 7.4 7.5 7.5 7.5 7.5 10.2
356 38.4 46.0 180 106 343 59.5 270 69.6 172 303 498 311
9.03 5.03 9.38 9.35 5.16 8.95 3.80 7.04 7.14 3.83 9.73 6.89 9.29
11.3 20.3 29.4 23.7 10.7 12.6 23.8 24.1 34.3 20.8 18.1 14.0 34.9
92.4 88.0 84.7 85.2 94.6 n.d. 89.3 83.5 79.2 84.6 93.9 93.0 81.7
Table 4. Experimental Conditions and Results from SCWO of p-Cresol at 460 °C and 250 atm initial concn. res. time (s)
p-cresol (µmol/L)
O2 (mmol/L)
conv. (%)
carbon tally (%)
3.5 4.9 4.9 4.9 4.9 5.5 7.3 7.3 7.4 7.4 7.4 10.5
351 33.4 42.6 103 107 359 52.2 314 63.0 154 157 293
9.42 9.27 5.07 9.19 5.07 8.72 7.02 9.94 3.91 7.10 3.92 9.43
21.7 51.0 40.4 42.3 29.5 18.0 58.2 41.9 38.2 42.8 32.4 43.9
n.d. 87.7 90.2 78.8 n.d. 93.2 83.3 n.d. 84.5 n.d. 81.3 82.2
Table 5 contains the parameter estimates and the variances and covariances for the regression analysis of each of the cresols. Hydroxybenzaldehyde Oxidation Kinetics. The data from the SCWO experiments for o-, m-, and p-hydroxybenzaldehydes, which appear in Tables 6-8, were treated in the same manner as the data from the m- and p-cresol experiments. We performed 11 SCWO experiments for o-hydroxybenzaldehyde, 20 for m-hydroxybenzaldehyde, and 14 for p-hydroxybenzaldehyde. We focused on shorter residence times than those used
Table 5. Regression Statistics for the SCWO of Cresols k′ a b σ2k′ σ2a σ2b 2 σk′,a 2 σk′,b 2 σa,b
m-cresol
p-cresol
-1.431 0.6549 0.6140 5.485 × 10-2 1.711 × 10-3 1.283 × 10-2 2.801 × 10-3 2.027 × 10-2 -1.848 × 10-3
-1.553 0.6031 0.5369 3.720 × 10-2 9.907 × 10-4 7.057 × 10-3 2.579 × 10-3 1.232 × 10-2 -6.927 × 10-4
Table 6. Experimental Conditions and Results from SCWO of o-Hydroxybenzaldehyde at 460 °C and 250 atm initial concn. res. time (s)
o-HB (µmol/L)
O2 (mmol/L)
conv. (%)
carbon tally (%)
0.54 0.55 0.68 0.68 0.71 1.4 1.4 1.4 1.5 1.5 1.6
30.4 272 25.2 290 71.0 27.3 65.5 261 25.7 293 91.0
6.34 6.44 4.18 4.17 3.83 6.99 3.67 6.74 4.15 4.17 3.89
84.0 52.6 84.0 41.5 84.9 100.0 100.0 65.8 100.0 46.8 56.3
103.9 87.1 81.7 80.5 77.5 125.4 76.9 83.3 86.2 73..2 117.1
Table 7. Experimental Conditions and Results from SCWO of m-Hydroxybenzaldehyde at 460 °C and 250 atm initial concn. res. time (s)
m-HB (µmol/L)
O2 (mmol/L)
conv. (%)
carbon tally (%)
0.54 0.67 0.67 0.68 0.71 0.91 1.4 1.4 1.5 1.5 1.5 1.5 1.5 3.1 4.1 4.1 4.1 10.2 10.2 10.3
95.3 75.1 305 23.5 81.5 88.6 74.7 301 23.1 23.7 71.9 84.5 255 90.6 14.8 66.1 162 32.9 79.7 71.9
4.66 6.86 6.92 6.97 3.32 5.06 3.68 7.08 7.14 2.96 7.23 5.36 2.94 5.00 3.68 2.49 3.66 6.02 8.92 8.98
22.3 1.0 1.0 4.9 19.1 31.8 18.7 8.3 23.8 14.7 8.4 27.1 19.5 54.3 29.4 20.1 18.2 48.4 62.6 80.9
89.6 104.3 101.2 95.1 85.3 89.3 n.d. 99.0 94.7 n.d. 102.6 90.8 96.5 76.5 98.9 90.5 91.1 87.9 81.3 117.0
for the SCWO of the cresols, because the hydroxybenzaldehydes reacted more quickly. All but four experiments had residence times between 0.5 and 4.1 s. Using the data in Tables 6-8, we were able to regress the rate law parameters in eq 4. The resulting global rate expressions for the hydroxybenzaldehydes at T ) 460 °C and P ) 250 atm are
roHB ) 10-0.78(1.24[oHB]0.47(0.11[O2]0.57(0.53
(8)
rmHB ) 10-0.31(1.23[mHB]0.98(0.13[O2]-0.33(0.38 (9) rpHB ) 10-1.81(0.62[pHB]0.77(0.11[O2]-0.02(0.21
(10)
Note that the oxygen reaction orders for m- and p-hydroxybenzaldehyde are statistically indistinguish-
Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1395 Table 8. Experimental Conditions and Results from SCWO of p-Hydroxybenzaldehyde at 460 °C and 250 atm initial concn. res. time (s)
p-HB (µmol/L)
O2 (mmol/L)
conv. (%)
carbon tally (%)
0.72 0.92 0.92 0.92 0.93 0.94 1.5 3.1 3.1 3.1 3.1 3.2 4.0 10.2
74.8 19.3 21.0 203 68.4 63.4 67.5 17.2 19.6 186 57.1 68.7 71.5 35.1
5.64 2.96 7.21 7.18 5.43 3.02 5.56 3.34 7.72 7.74 3.41 5.44 2.34 6.00
16.5 40.2 18.2 11.8 14.1 15.4 23.8 64.2 41.1 32.7 38.5 46.4 24.5 76.2
95.5 90.7 107.6 98.1 98.5 93.7 94.3 89.8 108.4 94.0 92.4 84.7 98.2 105.7
Table 9. Regression Statistics for the SCWO of Hydroxybenzaldehydes k′ a b σ2k′ σ2a σ2b 2 σk′,a 2 σk′,b 2 σa,b
o-HB
m-HB
p-HB
-0.7801 0.4769 0.5723 3.013 × 10-1 2.453 × 10-3 5.482 × 10-2 6.115 × 10-3 1.187 × 10-1 -1.885 × 10-3
-0.3124 0.9831 0.3300 3.390 × 10-1 4.045 × 10-3 3.177 × 10-2 2.866 × 10-2 9.223 × 10-2 4.552 × 10-3
-1.812 0.7669 -0.02186 8.029 × 10-2 2.656 × 10-3 8.716 × 10-3 9.524 × 10-3 1.670 × 10-2 -8.319 × 10-4
able from zero. Thus, over the range of reaction conditions studied, the oxygen concentration does not affect the rate of reaction of these compounds. Table 9 provides a complete listing of all of the regression statistics. The rate expressions of eqs 8-10 describe the net rate of disappearance of the particular hydroxybenzaldehyde through all competing reactions. At SCWO conditions, these competing reactions can include pyrolytic and hydrolytic decomposition in addition to oxidative destruction. Martino and Savage (1997) reported that hydroxybenzaldehydes do indeed react in supercritical water even in the absence of oxygen. They found that of the three isomers, o-hydroxybenzaldehyde pyrolyzed the most rapidly, and m- and p-hydroxybenzaldehyde pyrolyzed more slowly. We desired to separate the contributions from pyrolysis and oxidation so that we could determine the kinetics of the oxidation reaction alone. To accomplish this separation, we write the total rate of disappearance of hydroxybenzaldehyde as the sum of the rate of pyrolysis plus the rate of oxidation.
rate ) ratepyrolysis + rateoxidation ) kp[organic]ap + ko[organic]ao[O2]bo (11) Martino and Savage (1997) provide reliable rate laws for the pyrolytic contribution, so the parameters kp and ap are known. Thus, we used these parameters and the experimental data in Tables 6-8 to estimate values of ko, ao, and bo in eq 11. We used a fourth-order RungeKutta method with 100 steps to integrate the governing differential equation numerically, and we performed the parameter estimation using the Solver routine in a Microsoft Excel spreadsheet. The objective function was the sum of the squares of the differences between
Figure 1. Parity plot for p-hydroxybenzaldeyde pyrolysis and oxidation rate expression ([ denotes pyrolysis experiments and b denotes oxidation experiments).
calculated and experimental hydroxybenzaldehyde conversions. The results of this analysis appear below.
roHB ) 10-1.17[oHB]0.86 + 10-0.98[oHB]0.39[O2]0.68 (12) rmHB ) 10-3.57[mHB]0.52 + 10-0.99[mHB]1.37[O2]0.28 (13) rpHB ) 10-2.83[pHB]0.80 + 10-2.01[pHB]0.72[O2]0.02 (14) Figure 1 is a parity plot, which shows the conversion of p-hydroxybenzaldehyde calculated from the rate law in eq 14 as a function of the conversion measured experimentally. The rate law is able to correlate the data from experiments both with and without added oxygen. Note that the horizontal bars on the data points in Figure 1 indicate the range of conversions observed for each steady-state experiment. Structure and Reactivity The section above provides global reaction rate expressions for the SCWO of six different substituted phenols. Five of these rate expressions apply only to reactions at a single temperature of 460 °C and a pressure of 250 atm; thus we will select these conditions as the basis for a comparison of the oxidation kinetics of different phenolic compounds. We used the rate laws reported in this paper for cresols and hydroxybenzaldehydes, along with those in the literature for phenol and o-chlorophenol to calculate pseudo-first-order rate constants for SCWO of each phenolic compound under identical conditions. For the hydroxybenzaldehydes, which disappear through both thermal and oxidation reactions, we calculated one pseudo-first-order rate constant (from eqs 8-10) that includes both effects and a second rate constant (from the second term in eqs 1214) that excludes the thermal component. All of these pseudo-first-order rate constants, k′′, were calculated at reactant concentrations of [organic] ) 10-4 M and [O2] ) 7 × 10-3 M, which fell within the range of conditions studied experimentally. Table 10 provides these rate constants along with the 95% confidence interval when these uncertainties could be calculated. The 95% confidence interval is related to the variance in the calculated rate, which can be obtained by using the variances (σ x2i) and covariances (σ x2i,xj) of the rate law parameters in Equation 15. This
1396 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Table 10. Pseudo-First-Order Rate Constants for SCWO of Phenolic Compounds at 460 °C and 250 atm, with [Organic] ) 100 µmol/L and [O2] ) 7 mmol/L compound
k′′ (1/s)
phenol (Gopalan and Savage, 1995b) o-chlorophenol (Li et al., 1993) o-cresol m-cresol p-cresol o-hydroxybenzaldehydea (eq 8) o-hydroxybenzaldehydeb (eq 12) m-hydroxybenzaldehydea (eq 9) m-hydroxybenzaldehydeb (eq 13) p-hydroxybenzaldehydea (eq 10) p-hydroxybenzaldehydeb (eq 14)
0.0308 0.0379 0.179 ( 0.023 0.0423 ( 0.0031 0.0754 ( 0.0047 1.20 ( 0.29 0.939 0.111 ( 0.029 0.0842 0.147 ( 0.020 0.117
a Rate constant for reactant disappearance through both thermal and oxidative reactions, calculated as the rate from eqs 8-10 divided by the reactant concentration. b Rate constant for reactant disappearance through only oxidative reactions, calculated as the rate from the second term in eqs 12-14 divided by the reactant concentration.
propagation of errors formula gives the variance in a quantity y, which is a function of several parameters, xi. The index l in eq 15 is the number of parameters in the rate law. l
σ2y )
∑ i)1
() ∂y
∂xi
2
l-1
l
∑ ∑ i)1 j)i+1
σ2xi + 2
( )( ) ∂y
∂y
∂xi ∂xj
σ x2i,xj
(15)
By comparing the values of k′′ in Table 10, we can in effect compare the reaction rates for these phenolic compounds at this particular set of conditions. Table 10 reveals that all of the substituted phenols that have been studied thus far in our laboratory oxidize more quickly than phenol itself, with o-chlorophenol being the most resistant of the substituted phenols studied. This observation suggests that phenol is a good “worst-case” compound to use in engineering and economic evaluations of SCWO, because most other phenolic compounds will be easier to treat via SCWO. Table 10 also reveals that the location of the substituent influences the reactivity of the substituted phenol in supercritical water. For both the cresols and the hydroxybenzaldehydes, the ortho-isomer has the highest rate of reaction and the meta-isomer has the lowest. Finally, the data in Table 10 show that the identity of the substituent influences the reaction rates. For example, the relative reactivity of oortho-substituted phenols is 1, 1.2, 5.8, 39 as the group resident at the ortho-position varies through the sequence -H, -Cl, -CH3, and -CHO, respectively. Reaction Products and Pathways In this section we report the identities and molar yields of the reaction products formed from the SCWO of cresols and hydroxybenzaldehydes. The molar yields of liquid-phase products were calculated as the molar flow rates of product in the reactor effluent divided by the molar flow rate of reactant in the reactor feed stream. The molar yields reported for CO and CO2, on the other hand, were normalized by dividing the molar yields calculated in the manner described above by the number of carbon atoms in the reactant. SCWO of Cresols. Our previous work with o-cresol oxidation identified 19 intermediate liquid-phase products and detected the presence of 10 additional compounds (Martino et al., 1995). Our present results from m- and p-cresol SCWO do not display such a rich
product spectrum. We detected relatively few products of incomplete oxidation from m- and p-cresol. Other than phenol and the corresponding hydroxybenzaldehydes, the only product we detected was indanone. Figures 2-4, which show the molar yields of the major products from SCWO of o-, m-, and p-cresol, respectively, as a function of the conversion of the cresol, provide considerable insight into the governing reaction network for SCWO of cresols and the relative rates of the different paths. These figures show that for all three cresols, the yield of CO2 always exceeded the yield of CO. This behavior has also been observed for the SCWO of phenol and o-chlorophenol (Li et al., 1992). The yield of CO2 consistently exceeding the yield of CO for all phenolic compounds studied to date, even at low conversions, indicates that CO is not an important intermediate in the reaction network for CO2 production from phenolic compounds. Rather, there must be some path for CO2 formation that bypasses CO. A strong possibility is decarboxylation of carboxylic acid intermediates formed via ring-opening reactions. A second trend that all three cresols display is that at cresol conversions of less than 50% the yield of the hydroxybenzaldehyde always exceeds the yield of phenol. This behavior demonstrates that oxidation of the -CH3 substituent in cresols is favored over elimination of the substituent. Comparing the yields of CO and CO2 with the yield of phenol reveals that o-cresol is unique in that these yields are comparable, even at the lower o-cresol conversions. For both m- and p-cresol, on the other hand, the yield of CO + CO2 is much greater than the yield of phenol. This observation indicates that elimination of the ortho-substituent (from o-cresol or perhaps the o-hydroxybenzaldehyde product) is more facile than elimination of the analogous meta- and para-substituent. The m- and p-cresols appear to favor ring-opening reactions and COx formation over elimination of the substituent. This difference in the relative reactivity for these competing pathways is consistent with the dissociation energy of the C-C bond between the aromatic ring and a carbon-containing substituent in ortho-substituted phenols being lower than that in meta- and para-substituted phenols. A final observation we make from Figures 2-4 is that hydroxybenzaldehydes are the major low-conversion products from SCWO of o- and p-cresol whereas CO and CO2 are the most abundant products from m-cresol SCWO. This difference indicates that the -CH3 substituent in o- and p-cresol is more readily attacked and oxidized than is the -CH3 substituent in m-cresol. This difference can be understood by recognizing that a carbon-centered radical at the ortho- or para-position enjoys a greater resonance stabilization than does an analogous radical at the meta-position (Gopalan and Savage, 1994). To summarize this section, we offer Figure 5 as a general reaction network for SCWO of cresols. This network shows three parallel paths for cresol SCWO. One leads to a hydroxybenzaldehyde via oxidation of the methyl substituent, another leads to ring-opening products and the formation of CO and CO2, and the last leads to phenol via demethylation. The relative importance of the parallel pathways depends on the specific cresol isomer being oxidized. Figure 5 shows that phenol and hydroxybenzaldehydes are key organic intermediates in the reaction network, so the network can be expanded by including
Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1397
Figure 2. Molar yields of products from SCWO of o-cresol at 460 °C and 250 atm.
Figure 3. Molar yields of products from SCWO of m-cresol at 460 °C and 250 atm.
Figure 4. Molar yields of products from SCWO of p-cresol at 460 °C and 250 atm.
Figure 5. Cresol SCWO reaction network.
the reaction paths for these compounds. The SCWO of phenol has been thoroughly investigated, and a proven reaction network is available in the literature (Gopalan and Savage, 1995b). This network proceeds through two parallel primary paths that produce either phenol dimers or ring-opening products. Both the dimers and
the ring-opening products are ultimately oxidized to CO2. The SCWO of hydroxybenzaldehydes, on the other hand, has been unexplored until now. The next section, which discusses the reaction network for SCWO of hydroxybenzaldehydes, will thus allow us to complete the reaction network for SCWO of cresols. SCWO of Hydroxybenzaldehydes. The major products from SCWO of hydroxybenzaldehydes were CO, CO2, and phenol, and the corresponding carbon tally, reported in Tables 5-7, was always close to 100%. In fact, only 6 of the 45 experiments with hydroxybenzaldehydes had carbon tallies that were not within the range of 100% ( 15%. In addition to the three major products, indanone and trace amounts of phenol SCWO byproducts such as xanthone and fluorenone were also detected in some experiments. Figures 6-8 display the molar yields of the major products from SCWO of o-, m-, and p-hydroxybenzaldehyde, respectively, as a function of the conversion of the hydroxybenzaldehyde. These data reveal that for all three hydroxybenzaldehydes, the molar yield of CO2 always exceeded that of CO. This observation is fully consistent with the SCWO behavior of all other phenols studied to date. The data also show that although all three hydroxybenzaldehydes led to the same products, the relative abundance of the different products varied
1398 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997
Figure 6. Molar yields of products from SCWO of o-hydroxybenzaldehyde at 460 °C and 250 atm.
Figure 7. Molar yields of products from SCWO of m-hydroxybenzaldehyde at 460 °C and 250 atm.
Figure 8. Molar yields of products from SCWO of p-hydroxybenzaldehyde at 460 °C and 250 atm.
for the three isomers. For o-hydroxybenzaldehyde the yield of phenol was greater than the yield of COx, for m-hydroxybenzaldehyde the yield of COx was greater than the yield of phenol, and for p-hydroxybenzaldehyde the yields of COx and phenol were comparable. Phenol formation from hydroxybenzaldehydes involves loss of the CHO group, and a higher phenol yield from ohydroxybenzaldehyde is consistent with the ortho-C-C bond being weaker than the meta- and para-C-C bonds. The initial slopes of the molar yield vs hydroxybenzaldehyde conversion curves in Figure 6-8 are nonzero for both CO2 and phenol, which is indicative of these being primary products. That is, their formation from the hydroxybenzaldehyde involves only one “slow” step. For the case of CO2 formation, it is difficult to envision a direct reaction of hydroxybenzaldehyde to high yields of CO2, so we interpret the non-zero initial slope for CO2 as an indication that the steps that ultimately produce CO2 are much faster than the primary ring-opening reactions. This interpretation is consistent with a previous quantitative kinetics study of the phenol SCWO reaction network (Gopalan and Savage, 1995b). Figure 9 summarizes this discussion of hydroxybenzaldehyde products and pathways by showing a reaction network for hydroxybenzaldehyde SCWO. There are two parallel reactions. One leads to phenol, and the other leads to ring-opening reactions and ultimately
Figure 9. Hydroxybenzaldehyde SCWO reaction network.
CO2. We can combine the reaction network in Figure 9 for hydroxybenzaldehydes, the reaction network for phenol in the literature (Gopalan and Savage, 1995b), and the reaction network in Figure 5 for cresols to assemble a more complete reaction network that shows the conversion of cresols and its important intermediate byproducts into CO2. Figure 10 provides this overall reaction network for the SCWO of cresols. Summary and Conclusions The oxidation of CH3- and CHO-substituted phenols in supercritical water proceeds more rapidly than the oxidation of phenol itself. For a given substituent,
Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 1399 p ) parar ) rate R ) gas constant T ) absolute temperature t ) time X ) reactant conversion Greek Letters σ2 ) variance or covariance τ ) residence time Subscripts 0 ) at reactor entrance o ) oxidation p ) pyrolysis c ) critical value
Literature Cited Figure 10. Combined reaction network for the SCWO of CH3and CHO-substituted phenols.
ortho-substituted phenols are the most reactive, and meta-substituted phenols are least reactive. For a given point of substitution, CHO-substituted phenols are more reactive than CH3-substituted phenols. Quantitative reaction rate laws for each phenolic compound appear in this paper. The oxidation of cresols in supercritical water proceeds through three parallel paths, demethylation (to form phenol), oxidation of the methyl group (to form hydroxybenzaldehyde), and ring opening. The ringopening path most likely proceeds through carboxylic acid intermediates, the decarboxylation of which account for the high yields of CO2. The relative rates of the three parallel paths are isomer specific and consistent with the bond energies and resonance stabilization expected in substituted phenols. The oxidation of hydroxybenzaldehydes in supercritical water proceeds through two parallel paths. One path leads to phenol, and the other involves ring opening and CO2 formation. Acknowledgment This work was supported by the Department of Energy University Coal Research program under Grant DE-FG22-92PC92536 and by an equipment grant from the National Science Foundation (CTS-9311300). Nomenclature A ) Arrhenius preexponential factor a ) reaction order for the organic compound b ) reaction order for oxygen c ) reaction order for water CI ) confidence interval CR ) cresol (methylphenol) Ea ) activation energy HB ) hydroxybenzaldehyde k ) reaction rate constant k′ ) log(k) k′′ ) pseudo-first-order rate constant l ) number of parameters M ) moles per liter m ) metan ) number of experiments P ) pressure o ) ortho-
Blaney, C. A.; Li, L.; Gloyna, E. F.; Hossain, S. U. Supercritical Water Oxidation of Pulp and Paper Mill Sludge as an Alternative to Incineration; Hutchenson, K. W., Foster, N. R., Eds.; ACS Symposium Series 608; American Chemical Society: Washington, DC, 1995; pp 444-455. Gopalan, S.; Savage, P. E. Reaction Mechanism for Phenol Oxidation in Supercritical Water. J. Phys. Chem. 1994, 98, 2646. Gopalan, S.; Savage, P. E. Phenol Oxidation in Supercritical Water: From Global Kinetics and Product Identities to an Elementary Reaction Model; Hutchenson, K. W., Foster, N. R., Eds.; ACS Symposium Series 608; American Chemical Society: Washington, DC, 1995a; pp 217-231 Gopalan, S.; Savage, P. E. Reaction Network for Phenol Oxidation in Supercritical Water: A Comprehensive Quantitative Model. AIChE J. 1995b, 41, 1864-1873. Jevtitch, M. M; Bhattacharyya, D. Biotreated Coal Liquefaction Wastewater: Identification of Organics by Composite RO Membrane Concentration, HPLC, and GC/MS. Env. Prog. 1986, 5, 130-134. Krajnc, M.; Levec, J. On the Kinetics of Phenol Oxidation in Supercritical Water. AIChE J. 1996, 42, 1977-1984. Li, R., Thornton, T. D.; Savage, P. E. Kinetics of CO2 Formation from the Oxidation of Phenols in Supercritical Water. Env. Sci. Technol. 1992, 26, 2388. Li, R.; Savage, P. E.; Szmukler, D. 2-Chlorophenol Oxidation in Supercritical Water: Global Kinetics and Reaction Products. AIChE J. 1993, 39, 178. Martino, C. J.; Savage, P. E. Thermal Decomposition of Substituted Phenols in Supercritical Water. Ind. Eng. Chem. Res. 1997, 36, 1385-1390. Martino, C. J.; Savage, P. E.; Kasiborski, J. Kinetics and Products from o-Cresol Oxidation in Supercritical Water. Ind. Eng. Chem. Res. 1995, 34, 1941-1951. Modell, M. In Standard Handbook of Hazardous Waste Treatment and Disposal; Freeman, H. M., Ed.; McGraw Hill: New York, 1989; Section 8.11. Oh, C. H.; Kochan, R. J.; Charlton, T. R.; Bourhis, A. L. ThermalHydraulic Modeling of Supercritical Water Oxidation of Ethanol. Energy Fuels 1996, 10, 326-332. Rice, S. F.; Steeper, R. R.; LaJeunesse, C. A. Destruction of Representative Navy Wastes Using Supercritical Water Oxidation. Sandia National Lab Report SAND94-8203; Sandia National Laboratories: Albuquerque, NM, 1993. Savage, P. E.; Gopalan, S.; Mizan, T. I.; Martino, C. J.; Brock, E. E. Reactions at Supercritical Conditions: Fundamentals and Applications. AIChE J. 1995, 41, 1723-1778. Shanableh, A.; Gloyna, E. F. Supercritical Water Oxidations Wastewater and Sludges. Water Sci. Tech. 1991, 23, 389. Steiner, E. C.; Rey, T. D.; McCroskey, P. S. SimuSolv Modeling and Simulation Software Reference Guide; The Dow Chemical Company, Midland, MI, 1990. Thornton, T. D.; Savage, P. E. Phenol Oxidation in Supercritical Water. J. Supercrit. Fluids 1990, 3, 240-248. Thornton, T. D.; Savage, P. E. Kinetics of Phenol Oxidation in Supercritical Water. AIChE J. 1992a, 38, 321.
1400 Ind. Eng. Chem. Res., Vol. 36, No. 5, 1997 Thornton, T. D.; Savage, P. E. Phenol Oxidation Pathways in Supercritical Water. Ind. Eng. Chem. Res. 1992b, 31, 2451. Tufano, V. A Multi-Step Kinetic Model for Phenol Oxidation in High Pressure Water. Chem. Eng. Technol. 1993, 16, 186-190. Wightman, T. J. M.S. Thesis, University of California at Berkeley, 1981. Yang, H. H.; Eckert, C. A. Homogeneous Catalysis in the Oxidation of p-Chlorophenol in Supercritical Water. Ind. Eng. Chem. Res. 1988, 27, 2009. Yen, T. F.; Tang, J. I. S.; Washburne, M.; Cohanim, S. Analytical Methods for Hazardous Organics in Liquid Wastes from Coal Gasification and Liquefaction Processes, 1982; EPA-600/4-82-
038; U.S. Environmental Protection Agency, U.S. Government Printing Office: Washington, DC, 1982.
Received for review November 4, 1996 Revised manuscript received January 31, 1997 Accepted February 3, 1997X IE960697Q
X Abstract published in Advance ACS Abstracts, April 1, 1997.