Ind. Eng. Chem. Res. 1988,27, 143-149
sistant Secretary for Fossil Energy through the Morgantown Energy Technology Center. The authors are indebted to Joseph Ausikatus of the Union Carbide Corporation for providing the silicalite molecular sieve used in this study.
Nomenclature C = concentration of SO2 in stack gas, g of S02/cm3gas D = adsorber bed depth, cm k = rate constant, cm3/(gmin) N = adsorption capacity, g of S02/cm3bed R = gas constant, 1.987 cal/(moEK) t = time, min T = temperature, K V = linear flow rate, cm/min (v/v) = volumetric concentration Subscripts
0 = at the adsorber inlet or at t = 0 b = break point
c = critical Registry No. SOz, 7446-09-5.
Literature Cited Bohart, G. S.; Adams, E. Q.J . Am. Chem. SOC. 1920,42, 523. Brown, G. N.; Torrence, S. L.; Repik, A. J.; Stryker, J. L.; Ball, F. J. Chem. Eng. Prog. 1972, 68(8), 13. Chriswell, C. D.; Gjerde, D. T. Anal. Chem. 1982,54, 1911. Chriswell, C. D.; Shultz-Sibbel, G. M. W.; Gjerde, D. T.; Fritz, J. S.; Coleman, W. E. Talanta 1982, 29, 447. Collins, J. J.; Fornoff, L. L.; Manchanda, K. D.; Miller, W. C.; Lovell, D. C. Chem. Eng. Prog. 1974, 70(6), 84.
143
Du Pont TGA Data Analysis Program, Wilmington, DE, 1982, p 8. Engdahl, R. B.; Rosenburg, H. S.CHEMTECH 1978, Feb, 118. Flanigen, E. M. Pure Appl. Chem. 1980, 52, 2191. Flanigen, E. M.; Bennett, J. M.; Grose, R. W.; Cohen, J. P.; Patton, R. L.; Kirchne, R. M.; Smith, J. V. Nature (London) 1978,271(9), 512. Grindley, T.; Kim, S. S.; Gorski, E. E.; Steinfeld, G. Proceedings of the Fifth Annual Contractors Meeting on Contaminant Control in Coal-Derived Gas Streams, U S . Department of Energy, Morgantown Energy Technology Center, May 1985, p 33. Klein, S. M. M.S. Thesis, Iowa State University, Ames, 1982. Moore, W. E. Proceedings of the Fifth Annual Contractors Meeting on Contaminant Control in Coal-Derived Gas Streams, U. S. Department of Energy, Morgantown Energy Technology Center, May 1985, p 1. Ramalho, R. S. Introduction to Wastewater Treatment Processes; Academic: New York, 1983. Slack, A. V. Sulfur Dioxide Removal from Waste Gases;Noyes Data Corporation: Park Ridge, NJ, 1971. Smith, F. B. Ph.D. Dissertation, Iowa State University, Ames, 1970. “Sulfur Dioxide Processing”. Chem. Eng. Prog. 1975, 1; reprint published by American Institute of Chemical Engineers, New York, 1975. Treybal, R. E. Mass Transfer Operations;McGraw-Hill: New York, 1968. Vermeulen, T. “Adsorption and Ion Exchange”. Section 16 In Chemical Engineers Handbook;Perry, R. H., Chilton, C. H., Eds.; McGraw-Hill: New York, 1973. Wright, G. D. Ph. D. Dissertation, Virginia Polytechnic Institute and State University, Blacksburg, 1979. Yang, R. T.; Shen, M. AIChE J . 1979, 25(5), 811.
Received f o r review December 8, 1986 Revised manuscript received August 27, 1987 Accepted October 5, 1987
Solvent Effects during Reactions in Supercritical Water Susan H. Townsend,+Martin A. Abraham,l Gilbert L. Huppert,l Michael T. Klein,*+ and Stephen C. Paspekt Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716, and Standard Oil (Ohio),Independence, Ohio 44131
T h e reactions of fully hydrocarbon and also heteroatom-containing coal model compounds were examined in water over a range of reduced water densities, 0 < ,or < 2.1. Dibenzyl ether (DBE), phenethyl phenyl ether (PPE), guaiacol, benzyl phenyl ether, and benzylphenylamine all underwent parallel pyrolysis and hydrolysis. The hydrocarbons 1,3-diphenylpropane and 1,2-diphenylethane and the heterocycle dibenzothiophene underwent neat pyrolysis only, and phenyl ether, dibenzofuran, and carbazole were stable during reaction both neat and in water. The solvolysis was a t saturated carbon t o which was attached a heteroatom-containing leaving group. Detailed kinetics analysis of the reactions of DBE, PPE, and guaiacol allowed decoupling of the rates of pyrolysis and hydrolysis and permitted correlation of the hydrolysis rate constant with the solvent dielectric constant. Good correlation on this Kirkwood plot suggests the hydrolysis proceeds through a transition state that is more polar than the reactants. The thermodynamics of fluids in the vicinity of their critical temperatures and pressures has led to the recognition of supercritical fluid (SCF) solvent extraction as a promising separation technology (Jezko et al., 1982; Paulaitis et al., 1983). For example, the extreme sensitivity to pressure of the solubility of naphthalene in ethylene at 12 and 35 OC, shown in Figure 1 (Diepen and Scheffer, 1953; Tsekhanskaya et al., 1964), is qualitatively similar to the sensitivity of the density ( p ) of water to pressure at 374 “C, shown in Figure 2. Extraction a t a pressure in the upper asymptotic region of Figures 1 and 2 could be followed by precipitation via a relatively small reduction in pressure and therefore form the basis of a pressuredriven separation scheme. University of Delaware. Standard Oil (Ohio).
This extreme variation of density with pressure should be important in any application sensitive to the thermodynamic properties of a fluid. Classes of chemical reactions are well-known to be affected by solvent properties (Eckert, 1967; Moore and Pearson, 1981),such as the dielectric constant or solubility parameter. For example, the classic Kirkwood analysis (1934) separates the free energy of activation into electrostatic and nonelectrostatic components and shows the former to be a function of the solvent dielectric constant. This is usually demonstrated through analysis of a probe reaction in a set of molecularly different solvents that span a range of dielectric constants (Eckert, 1967; Moore and Pearson, 1981). Supercritical fluid solvents are interesting in that their dielectric constants may be varied continuously (Franck, 1978),and without change of molecular structure, through variations in density. Thus, just as solubility varied from gaslike to liquidlike
0888-5885/88/2627-0143$01.50/0 6 1988 American Chemical Society
144 Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 Table I. Experimental Conditions for Hydrolysis reactant
A. e
A -
goo'"'
r m O m
t, min - -__
T, "C
Pr,w
331.412
0.1.6
0.60
375-413
0-1.4
o x
420
0-1.6
0-60
383
0-2.1
0-45
284-384
0-2.1
0.120
386
0-2.0
0-60
380-500
0- 1.6
0-I80
405
0-1.5
0-60
405
0-1.5
0-60
500
0-1.5
0-60
450-550
0-1.5
0-h@
550
0-1 2
0. I b:)
Phenelhyl Phenyl Ether
0 A A
00001
Tsekhonskoyo et 0 1 . Tsekhanskoya e l 01.. Diepen on6 Schcffet, Diepan ond Schcllet.
12 C
1.3-Dlphenylpropane
35 C 12
35
C C
E. I
!,
I
8
I
I
1
0
IMO
2Mo
ym
IMO
scoc
Pressure (PSIA) ti
Figure 1. Solubilities of solid naphthalene in supercritical ethylene (Diepen and Scheffer, 1953; Tsekhanskaya et al., 1964).
C. Phonyl Elher
Dlphenylmelhane
'-. _. ~
.,,jrr
io30
3000
LOO0
,
i
5000
6000
P r e s s u r e (PSIA)
Figure 2. Gas- and liquidlike density of supercritical water a t 374 "C.
in Figure 1, it seems reasonable to expect reaction chemistry and solvent effects to vary from those expected in a gas to those expected in a liquid as water density spans the values of Figure 2. We have used the reactions of the slate of coal model compounds shown in Table I in supercritical water to probe these ideas for several reasons. First, the literature provides much detail about their thermal pathways, kinetics, and mechanism (Benjamin, 1978; Bruker and Kolling, 1965; Cronauer et al., 1979; Fixari et al., 1984; Gilbert and Gajewski, 1982; Hellyar, 1982; King and Stock, 1984; Klein and Virk, 1983; Miller and Stein, 1979; Poutsma and Dyer, 1982; Schlosberg et al., 1981a,b),which facilitates delineation of the effect of the SCF solvent. Second, their thermal reactions occur in the experimentally convenient range of 300-450 OC, which includes the critical temperature of water, T,= 374 OC. Finally, ample literature on the "extraction" of volatiles from coal and other macromolecules (Amestica and Wolf, 1984; Ceylan and Olcay, 1981; Modell, 1977; Paulaitis et al., 1983) with SCF solvents suggests interesting but presently obscure benefits from operation under supercritical conditions. The object of the present report, then, is to organize new and previous (Abraham and Klein, 1985; Lawson and Klein, 1985; Townsend and Klein, 1985) reactions in and with supercritical water into a general network of parallel pyrolysis and hydrolysis reactions, the latter of which appear to be influenced by chemical solvent effects.
Experimental Section Model compounds were reacted neat and in solvent a t the conditions outlined in Table I, which is divided into three sections: section A lists diaryl reactants with
Carbazole
mm Dlbenzolhlophrne
three-atom linkages; section B lists the aryl alkyl ether, guaiacol, and diaryl reactants with two-atom linkages; and section C lists both diaryl reactants with one-atom linkages and also heterocyclic compounds. Reduced water loadings ( p , = pW/pw,J ranged from 0.0 to 2.1. Except for phenethyl phenyl ether, which was synthesized by using Williamson condensation (Mamedov and Khydyrov, 1962), all reactants, solvents, and GC standards were commercially available and used as received. A typical experimental procedure was as follows: measured amounts of the reactant, solvent, and the demonstrably inert (Townsend and Klein, 1985) internal standard biphenyl were loaded into "tubing-bomb" reactors comprising one lI4-in. stainless steel Swagelok port connector and two end caps. Sealed reactors were immersed into a fluidized sand bath held a t the desired reaction temperature, *2 "C, for the reaction time. The reaction temperature was attained by the reactors in about 2 min; this heat-up period was small compared to ultimate reaction times of up to 180 min and was, in any case, similar for all runs. Reactions were quenched by immersion in a cold water bath. Thus at low conversion the reaction system was essentially a ternary (reactant-H20-biphenyl) mixture, the phase behavior of which depending on the amount of water added to the reactor, i.e., pw = g of waterlreactor volume. Phase equilibrium estimates were accomplished by using a flash calculation scheme (King, 1980) modified for a fixed volume input (instead of pressure). The generalized Peng-Robinson (1976) equation of state, which has been shown (Thies, 1985) to provide an accurate correlation of the phase behavior of coal model compounds and supercritical methanol, was used to determine species' fugacity coefficients.
Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 145 Table 11. Experimental Results reactant
pyrolysis
products hydrolysis '.O
z
p
c
1
+
{gym 0.41 0.1
xxxx
! ;, ,
,&
,a; ,
8
0.0
-
E: Qfa
0-a
QfT
c
r
m
xxxx C.
Ux2
xxxx
m-0
xxxx
QCJa
xxxx
qm
xxxx
Phenyl Ether
Dlphmylmclhanc
Dibcnzoluran
a
xxxx
Dlbenlothiophcne
XXXX
No Reaction
The flash calculation indicated that the neat pyrolysis = 0) was 75% liquid and 25% vapor phase and that a two-phase mixture existed for 0 < p, < 0.3. For p , 1 0.3, where the vast majority of the experiments were executed, the initial reaction mixture was a single phase. This is the region of focus in the present report, although a report detailing with the phase equilibrium calculations and their interesting implications for p , < 0.3 is to follow. Reaction products were collected as a single phase in either acetone or tetrahydrofuran. Subsequent product identification was by GC/MS, and routine quantitation was by GC on an H P 5880 instrument equipped with an SE-54 or DB-5 fused silica capillary column and flame ionization detector. Response factors were estimated by analysis of standard mixtures. (p,
Results The experimental results suggest that reaction in supercritical water consists of parallel pyrolysis and hydrolysis reaction pathways. These results are summarized in Table 11, which organizes, as before, the three classes of model compounds studied and their reaction products. The discussion of the experimental results is through sections pertaining to the reactions of an individual model compound. Neat pyrolysis results are presented first and serve as a background against which the reactions in solvent may be discerned and quantitated. Three-Atom Linkages. Dibenzyl Ether. Neat pyrolysis of dibenzyl ether (DBE) at 374 "C led to toluene and benzaldehyde as major and primary products. Its reaction in water a t 374 "C led to benzyl alcohol, toluene, benzaldehyde, and oligomers. DBE decomposition in
0.0
0.1
, B
,
8
,
.y 0.4
0.6
0,t
1.0
Reduced Water Loading
,
-1
A
Y
1.1
1.4
1.6
(Pr)
Figure 3. Reaction of dibenzyl ether in water a t 375 " C and 20 min: dependence of product selectivities on pr.
water at p , = 1.6 was about 3.5 times faster than neat pyrolysis; benzyl alcohol was the major and essentially the only primary product at this density. The yield of benzyl alcohol reached a maximum and then decreased at longer thermolysis times as it reacted to oligomers. The influence of p, on the selectivity (s, = y J x ) to benzyl alcohol, toluene, and benzaldehyde is presented in Figure 3 for the reaction of DBE in water at 375 "C at a constant reaction time of 20 min. Selectivity to benzyl alcohol increased from less than 0.1 to about 1.5 as pr increased from 0.0 to 1.6. Selectivities to toluene and benzaldehyde, the products of DBE pyrolysis, both decreased from about 0.6 to less than 0.1 as pr increased from 0.0 to 1.6. The foregoing suggests that DBE reaction in water comprised two simultaneous reaction pathways, the first being identical with the neat pyrolysis reported in the literature (Schlosberg et al., 1981a) and also like thermolysis in hydrogen donor (Bruker and Kolling, 1965; Cronauer et al., 1979; Simmons and Klein, 1985). The second path was hydrolysis of 1 mol of DBE to 2 mol of benzyl alcohol. Kinetics analysis in terms of this network provided estimates of the rate constants for reaction at 374 "C (Townsend and Klein, 1985). For reaction at p , = 1.6, the value for the pseudo-first-order pyrolysis rate constant, s-l, was much smaller than the value for h, = 4.0 X the pseudo-first-order hydrolysis rate constant, h i = lo-* SI. Decoupling the concentration of water from k2' afforded the "true" second-order rate constant k 2 (L/(mol s)) as will be discussed below. Phenethyl Phenyl Ether. The major primary products from the neat pyrolysis of phenethyl phenyl ether (PPE) at 400 "C were phenol and styrene; phenol was thermally stable, and styrene underwent secondary reaction to ethyl benzene, toluene, benzene, and other minor products. Additional minor products included benzaldehyde, 1,2-diphenylethane, diphenylmethane, 1,3-diphenylpropane, and materials of molecular weight higher than that of PPE. P P E reaction in water at 400 "C and p , = 1.6 also led to phenol, styrene, and the other products of neat pyrolysis, but in addition afforded phenethyl alcohol. Reactions of P P E in H2180showed incorporation of the label into the phenethyl alcohol. The influence of p , on the selectivity to products at 400 O C probed the operative pathways further, as illustrated in Figure 4. For a constant reaction time of 30 min, selectivity to phenol was essentially constant at its average of 0.70 f 0.06 at all reduced densities. Selectivity to styrene increased from near zero at pr = 0.0 to about 0.30 and then decreased slightly for pr L 0.73. This occurred because the secondary decomposition of styrene was suppressed by the increased water density, which was further shown by the decrease in selectivity to ethyl benzene, toluene, and benzene with increasing water density. The
146 Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988
.-
c
0.7
0.4
"% J 0
@I\ &
, , , ,
O.Oi.---,
00
0.5
1 .o
,
Reduced Water L o a d i n g
, , 1.5
, -,
.
I
20
00
(pr)
Figure 4. Reaction of phenethyl phenyl ether in water a t 400 "C and 30 min: dependence of product selectivities on pr.
selectivity to phenethyl alcohol increased from zero at pr = 0.0 to 0.41 at pr = 0.99 and then decreased slightly for pr > 0.99. This decrease is consistent with a greater degree of secondary decomposition of the alcohol at higher values of PI. These results suggest that the overall reaction of PPE in water is by two paths, the first being pyrolysis to phenol and styrene and the second being hydrolysis to phenol and phenethyl alcohol. The neat pyrolysis pathway is identical with that described in the literature (Klein and Virk, 1983). The hydrolysis of P P E was equivalent to the addition of 1 mol of water to 1 mol of P P E to product 1 mol each of phenol and phenethyl alcohol. 1,3-Diphenylpropane. Neat pyrolysis of 1,3-diphenylpropane (DPP) at 420 "C led to toluene and styrene, with the latter product undergoing rapid secondary conversion to other products, including ethylbenzene. Minor products included 1,2-diphenylethane, benzene, and npropylbenzene, and all were present in molar yields of less than 0.05. Reaction of DPP in water a t 420 "C and pr = 1.6 led to the same products in approximately the same molar yields as did pyrolysis. Therefore, neat reaction of DPP, which proceeded along a single pathway to toluene and styrene, was not accompanied by hydrolysis during its reaction in water. Two-Atom Linkages. Guaiacol. Catechol was the major and primary product from the neat pyrolysis of guaiacol a t 383 "C; secondary decomposition of catechol was observed at long reaction times. Minor products included o-cresol, phenol, and trace amounts of methanol and an intractable high molecular weight material. The pseudo-firsborder disappearance rate constant determined through linear least-squares analysis was 5.48 X s-'. Reaction of guaiacol in water at 383 "C and pr = 1.9 yielded catechol and methanol as the major and primary products, and the increase in their selectivities was at the expense of the yields of char and the minor pyrolysis products o-cresol and phenol. The dependence of product selectivities on pr is illustrated in Figure 5 for reaction of guaiacol at 383 "C at the constant reaction of 30 min. Selectivity to catechol, which resulted from both pyrolysis and hydrolysis,increased from about 0.14 during neat pyrolysis to greater than 0.72 a t pr = 2.1. Selectivity to methanol increased from nearly 0.0 to 0.58 as pr increased from 0.0 to 2.1. The high molecular weight material (char) was not observed from reaction of guaiacol at pr > 0.5. These results are consistent with reaction of guaiacol in water proceeding along parallel pyrolysis and hydrolysis paths (Lawson and Klein, 1985). The pyrolysis pathway led primarily to catechol and char, whereas the hydrolysis pathway involved the addition of 1 mol of water to 1 mol of guaiacol, producing 1mol each of catechol and methanol
04
08
1 2
1 6
Reduced Water Loading
20
2 4
(pr)
Figure 5. Reaction of guaiacol in water a t 383 dependence of product selectivities on pr.
O C
and 30 min:
c
u
3
U 0
L
a Reduced Water Loading
Figure 6. Reaction of benzyl phenyl ether in water at 377 O C and 5.6 min: dependence of product selectivities on p r .
as the major and primary products. Benzyl Phenyl Ether. Neat pyrolysis of benzyl phenyl ether (BPE) at 332 "C led to phenol and toluene as major and primary products; both were thermally stable. Minor products included, in order of decreasing yield, ohydroxydiphenylmethane (OHD), p-hydroxydiphenylmethane (PHD), diphenylmethane, benzaldehyde, benzene, and 1,2-diphenylethane. The apparent first-order disappearance rate constant, estimated by using leastsquares analysis, was 9.45 X loe4s-l for BPE pyrolysis at 332 "C. Reaction of BPE in water a t 332 "C at pr = 1.6 was almost 4 times faster than neat pyrolysis at the same temperature. Benzyl alcohol, produced iii only trace quantities during neat pyrolysis, was a major hydrolysis product and underwent secondary reaction. The selectivity to phenol increased by a factor of about 2 and the selectivity to toluene decreased by a factor of about 4 relative to the selectivities observed in neat pyrolysis. The minor products formed during neat thermolysis of BPE were also observed from its reaction in water. Figure 6 illustrates the dependence of product selectivity on pr for reaction of BPE at 377 "C and a constant reaction time of 5.6 min. Selectivity to benzyl alcohol increased from near zero under neat pyrolysis conditions to a maximum of about 0.5 a t pr of about 1.2. The decrease in selectivity at higher pr may be attributed to secondary decomposition of the alcohol, as described previously for reaction of DBE. The selectivity to phenol, both a pyrolysis and hydrolysis product, increased from about 0.6 to 0.85 as pr increased from 0.0 to 1.2; selectivity to phenol decreased slightly at higher pr. Selectivity to toluene decreased from 0.22 during neat pyrolysis to 0.03 for pr 2 1.2. Selectivities to OHD and PHD were 0.08 and 0.03 for pr = 0.0; these values increased about 3-fold to 0.21 and 0.12, respectively, as pr increased to 2.1. These observations imply that selectivities to these two rearrangement prod-
Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988 147
Reduced Water Loading (pr)
Figure 7. Reaction of benzylphenylamine in water a t 386 "C and 20 min: dependence of product selectivities on pr.
ucts were enhanced by the formation of a "solvent cage" as pr increased. These results suggest that reaction of BPE in water is a combination of a thermal pathway leading to phenol and toluene and a hydrolysis pathway yielding phenol and benzyl alcohol. The thermal pathway is like that reported in the literature for BPE thermolysis in tetrahydroquinoline (Bruker and Kolling, 1965; Schlosberg et al., 1981b) and thermolysis in tetralin (Sato and Yamakawa, 1985; Kamiya et al., 1979). The hydrolysis reaction is the addition of 1 mol of water to 1 mol of BPE, producing 1 mol each of phenol and benzyl alcohol. Benzylphenylamine. Neat pyrolysis of benzylphenylamine (BPA) at 386 "C yielded toluene, aniline, and benzalaniline as major and primary products; 1,2-diphenylethane and diphenylmethane were minor, primary products. The yields of toluene and aniline were roughly equal. BPA conversion was approximately 0.8 after 40 min, a t which time the yields of toluene, aniline, and benzalaniline were 0.4, 0.4, and 0.3, respectively. BPA reaction in supercritical ( p , = 1.2) water yielded benzyl alcohol as a major and primary product and benzaldehyde as a minor product in addition to those products from neat pyrolysis. BPA conversion was approximately 0.7 after 40 min, at which time the yields of toluene and aniline were approximately 0.5 and 0.2, respectively. Benzyl alcohol attained a maximum yield of 0.15 at 15 min, after which time its yield decreased to near 0.0. This is consistent with the secondary decomposition of benzyl alcohol observed during reaction of both DBE and BPE in water. The effect of ,or on product selectivities at 386 " C and a t a constant reaction time of 20 min is shown in Figure 7. The selectivity to aniline of 0.6 was relatively unaffected by the solvent density. Selectivity to toluene decreased from 0.4 for neat pyrolysis to nearly 0.0 at p, = 2.0. Conversely, the selectivity to benzyl alcohol was 0.0 for neat pyrolysis and increased to approximately 0.5 as pr increased to 2.0. Thus toluene and benzyl alcohol were produced through competitive pyrolysis and hydrolysis pathways. The rate of pyrolysis, and thus the selectivity to toluene, decreased, while the rate of hydrolysis, and therefore the selectivity to benzyl alcohol, increased as the reduced water density increased. These results are consistent with reaction of BPA in water following parallel pyrolysis and hydrolysis pathways, the former to toluene, aniline, and benzalaniline and the latter to benzyl alcohol and aniline. 13-Diphenylethane. Neat pyrolysis of diphenylethane (DPE) a t 500 "C afforded toluene as the major primary product; trans-stilbene, benzene, ethylbenzene, phenanthrene, and diphenylmethane were all minor primary products. Trace amounts of styrene and triphenylethylene were also present. Reaction in water a t 500 "C and pr =
1.4 also led to toluene as the major primary product, and the minor products were identical with those observed from neat pyrolysis. Thus, no additional pathways were identified for the reaction of DPE in water. Diarylalkanes, Diary1 Ethers, and Heterocycles. Neat pyrolysis and reaction in water of phenyl ether, diphenylmethane, dibenzofuran, and carbazole at 405,405, 500, and 550 "C, respectively, led to no significant observable products. Neat pyrolysis of the heterocycle dibenzothiophene (DBT) a t 550 O C led to a conversion of only 0.02 after 240 min; biphenyl was the primary product, and benzene was also found in lesser amounts. Reaction of DBT in water a t 550 OC (p, = 1.2) led to a conversion of 0.02 after 240 min; biphenyl and benzene were its products. We were therefore able to deduce no differences between neat pyrolysis of these compounds and their reaction in water.
Discussion The overall reaction in supercritical water of the compounds listed in Table I can thus comprise both pyrolytic and hydrolytic pathways. Neat reaction leads to a pyrolysis product slate, and the addition of water may alter the overall selectivity toward a hydrolysis product slate. The presence of the hydrolysis path and the increase in reaction rate along this path with increases in p, may be considered as primary solvent effects. However, interesting secondary chemical solvent effects also impact activity and selectivity, as follows. The experimental results are summarized in Table TI in terms of the structures of the reactants, pyrolysis products, and hydrolysis products. In section A, the reactions of the three-atom bridged diaryl ethers include both pyrolysis and hydrolysis. The diarylalkane (DPP), on the other hand, afforded only pyrolysis products, even in the presence of high concentrations of water. Section B shows the same to be true of a family of two-atom bridged diaryl ethers and alkanes. In the presence of water, guaiacol, benzyl phenyl ether, and benzylphenylamine underwent both pyrolysis and hydrolysis, but 1,2-diphenylethane underwent only pyrolysis, even at large pr. Thus, sections A and B of Table I1 collectively suggest that the hydrolysis reaction requires the presence of a heteroatom in the reactant. Additionally, however, section C shows several heterocyclic reactants to be inert to hydrolysis; the mechanism of hydrolysis is therefore likely constrained further to require a reactant with a saturated carbon to which is attached a heteroatom-containing leaving group. The preceding observations are concisely summarized in terms of the mechanistic hypothesis of Figure 8. This single step summarizes each hydrolysis reaction of Table I1 and depicts the nucleophilic attack of the solvent on a saturated carbon atom to which is attached a heteroatom-containing leaving group. This mechanism thus would not allow for either hydrolysis of a diarylalkane or a heterocycle devoid of a saturated carbon. The reaction presented in Figure 8 is between neutral molecules likely proceeds through a polar transition state. It should thus be susceptible to the classic Kirkwood analysis (1934), wherein the free energy of activation is apportioned into electrostatic and nonelectrostatic parts, the former being dependent upon the solvent dielectric constant. The associated Kirkwood plot correlates the rate constant for solvolysis linearly with (e - l ) / c , where t is the solvent dielectric constant (Moore and Pearson, 1981). The Kirkwood plot shown in Figure 9 is for the hydrolysis kinetics of DBE, PPE, and guaiacol. The hydrolysis rate constants, k,, were decoupled from the pyrolysis rate constants by monitoring both solvolysis
148 Ind. Eng. Chem. Res., Vol. 27, No. 1, 1988
drolysis path increases with water loading for several reasons, the most important being the increase in rate of a reaction that is positive order in the concentration of one of its reactants. Increasing the water density also raises the water dielectric constant, which stabilizes the polar hydrolysis transition state over the less-polar reactants.
R
Compounds
L
DEE
Ph
OCH Ph
PPE
CH,Ph
OPh
Guaiacol
H
Ho O x
BPE
Ph
OPh
BPA
Ph
NHPh
Acknowledgment
Figure 8. Mechanism for the reaction observed with supercritical water.
Guoiacol at 656 K
I
. . . ..
*.
0
* *
I 3000010~
.____. 05
06
08
07
09
13
(&-I)/.
Figure 9. Kirkwood plot for hydrolysis of DBE and guaiacol in supercritical water.
products’ yields at low overall substrate consumption, each at a different initial pr. Both k , and also the pyrolysis rate constant were dependent upon the initial water density, which was essentially constant for the low conversion data used for parameter estimation. Over the range 0.24 < pr < 2.1, the correlation of log k , vs ( t - 1)/t is reasonably linear for DBE, PPE, and guaiacol. This both lends support to the postulated polar transition states and also provides the basis for extracting information about the changing distribution of charge along the reaction coordinate (Moore and Pearson, 1981). Reaction in a SCF solvent is especially well suited for this analysis since its dielectric constant can be manipulated and controlled through pressure (density) alone. Often the construction of a Kirkwood plot is based on study of a series of solvent molecules (e.g., HzOmethanol-ethanol) (Eckert, 1967; Moore and Pearson, 1981), which can induce additional experimental and theoretical uncertainty. The Kirkwood correlation therefore suggests that the control of the solvolysis rate constant by variation of t can be effected through changes in density. For a SCF solvent, the pressure dependence of density, and thus e, near the critical pressure is strong (Franck, 1978) and represents a powerful lever to manipulate and control reactivity and selectivity with relatively modest changes in pressure.
Conclusions The overall reaction in supercritical water of a collection of coal model compounds with the common structural moiety of a saturated carbon to which is attached a heteroatom-containing leaving group involves both pyrolysis and hydrolysis components. The selectivity to the hy-
The authors acknowledge the experimental assistance of Dr. John R. Obst (USDA Forest Products Laboratory, Madison, WI) and James R. Lawson (University of Delaware) and also useful discussions with Dr. Obst and Dr. Mike Lemanski, Standard Oil, Cleveland, OH. We are grateful for the support of this work by the Petroleum Research Fund, administered by the American Chemical Society, the Department of Energy (DE-FG22-85PC80509), the Standard Oil Company, and the USDA Forest Products Laboratory.
Nomenclature n, = mole number of i P = pressure, atm R = reactant s, = selectivity = y , / x T = temperature, “C x = conversion = n,/nR,O yi = yield = nr/nR,O Greek Symbols t = dielectric constant pw = absolute water loading, g of water/volume of reactor pr = reduced water loading = p w / p w , c Subscripts
c = critical point i = species counter r = reduced property w = water 0 = initial Registry No. DBE, 103-50-4; P P E , 40515-89-7; DPP, 108175-0; BPE, 946-80-5; BPA, 103-32-2; DPE, 103-29-7; DBT, 13265-0; PhOPh, 101-84-8; PhCH2Ph, 101-81-5; guaiacol, 90-05-1; dibenzofuran, 132-64-9; carbazole, 86-74-8.
Literature Cited Abraham, M. A.; Klein, M. T. Ind. Eng. Chem. Prod. Res. Deu. 1985, 24, 300. Amestica, L. A.; Wolf, E. E. Fuel 1984, 63, 227. Benjamin, B. M. Fuel 1978,57, 378. Brdker, R.; Kolling, G. Brenstaff-Chem. 1965, 1965, 41. Ceylan, R.; Olcay, A. Fuel 1981, 60, 197. Cronauer, D. C.; Jewell, D. M.; Shah, Y. T.; Modi, R. J. Ind. Eng. Chem. Fundam. 1979,18, 153. Diepen, G. A. M.; Scheffer, F. E. C. J. Phys. Chem. 1953,57, 575. Eckert, C. A. Ind. Eng. Chem. 1967, 59, 20. J. Chzm. 1984,8(3), Fixari, B.; Abi-Khers, V.; Le Perchec, P. NOUU. 177. Franck, E. U. “Dielectric behaviour of polar fluids at high pressures”, In Organic Liquids: Structure, dynamics and chemical properties; Buckingham, A. D., Lippert, E., Bratos, s.,Ed.; Wiley: New York, 1978. Gilbert, K. E.; Gajewski, J. J. J. Org. Chem. 1982, 47(25), 4899. Hellvar. M. J. Ph.D. Thesis. Massachusetts Institute of Technology, -. Cimbridge, 1982. Jezko, J.; Gray, D.; Kershaw, J. R. Fuel Process Technol. 1982, 5, 229. Kamiya, Y.; Yao, T.; Oikawa, S. Prepn.-Am. Chem. SOC., Diu.Fuel Chem. 1979, 24(2), 116. King, C. J. Separation Processes; McGraw-Hill: New York; 1980. King, H.-H.; Stock, L. M. Fuel 1984, 63, 810. Kirkwood, J. G. J . Chem. Phys. 1934, 2, 351. Klein, M. T.; Virk, P. S. Ind. Eng. Chem. Fundam. 1983, 22, 35. Lawson, J. R.; Klein, M. T. Ind. Eng. Chem. Fundam. 1985,24,203. Mamedov, S.; Khydyrov, D. N. Zh. Obsh. Khim. 1962, 32, 1427.
Ind. Eng. Chem. Res. 1988,27, 149-153 Miller, R. E.; Stein, S. E. Prepr.-Am. Chem. SOC.,Diu. Fuel Chem. 1979, 24,271. Modell, M. “Reforming of Glucose and Wood at the Critical Conditions of Water”. Technical Report, 1977; ASME Intersociety Conference on Environmental Systems, New York. Moore, J. W.; Pearson, R. G. Kinetics and Mechanism, 3rd ed.; Wiley: New York, 1981. Paulaitis, M. E., Penninger, J. M. L., Gray, R. D., Jr., Davidson, P., Eds. Chemical Engineering at Supercritical Fluid Conditions; Ann Arbor Science Ann Arbor, MI, 1983. Peng, D.-Y.;Robinson, D. B. Ind. Eng. Chem. Fundam. 1976,15,59. Poutsma, M. L.; Dyer, C. W. J . Org. Chem. 1982, 47,4903. Sato, Y.; Yamakawa, T. Znd. Eng. Chem. Fundam. 1985,24(1), 12.
149
Schlosberg, R. H.; Ashe, T. R.; Pancirov, R. J.; Donaldson, M. Fuel 1981a, 60, 155. Schlosberg, R. H.; Davis, W. H., Jr.; Ashe, T. R. Fuel 1981b, 60,201. Simmons, M. B.; Klein, M. T. Ind. Eng. Chem. Fundam. 1985,24(1), 55. Thies, M. C. Ph.D. Thesis, University of Delaware, Newark, 1985. Townsend, S. H.; Klein, M. T. Fuel 1985, 64, 635. Tsekhanskaya, Y. V.; Iomtev, M. B.; Mushinka, E. V. Russ. J. Phys. Chem. 1964, 38, 1173.
Received for review April 20, 1987 Revised manuscript received September 23, 1987 Accepted September 29, 1987
GENERAL RESEARCH Some Studies on Heat Transfer in Diverging-Converging Geometries C. M. Narayanan*+and B. C. Bhattacharyyat Department of Chemical Engineering, Regional Engineering College, Durgapur 713209, India, and Department of Chemical Engineering, Indian Institute of Technology, Kharagpur 721302,India
Transport phenomena in diverging-converging geometries have created a lot of interest recently. T h e present paper highlights the specific advantages of diverging-converging geometries in the augmentation of heat transfer a t the expense of a negligible increase in the pressure drop penalty. The work is confined to Newtonian, incompressible, and steady-state flow through the annular space between a diverging-converging tube and a n outer straight column and also t o heat transfer a t constant wall temperature. The work also highlights the dependence of heat-transfer augmentation on the design parameters such as the angle of constriction and segment length. The analysis of transport phenomena (momentum, heat and mass transport) connected with irregular geometries has always generated a lot of interest. Different studies have been made in the past to account for different geometric parameters. However, until very recently, generalized mathematical analyses of velocity, temperature, and concentration distributions in irregular geometries have not been conducted. One of the earliest studies in this connection may be attributed to that of Millsaps and Pohlhausen (19531, who plotted a family of velocity profiles for flow through convergent as well as divergent channels a t different Reynolds numbers on the basis of numerical calculations performed by them on the exact solutions proposed by Jeffery (1915) and Hamel (1916). Among the recent studies, the work of Payatakes et al. (1973) is most noteworthy. They determined the velocity distribution for Newtonian flow through periodically constricted tubes by performing a numerical solution to the flow equations. Flow dynamics for non-Newtonian flow through porous media has been analyzed by Sheffield and Metzner (1976). They have concluded that in such cases the “divergingconverging” character of the flow channels is a very influencing parameter that shall have to be accounted for. Some studies on heat-transfer enhancement in tortuous flow have been reported by Fujita and Hasegawa (1940) and Klepper (1973). Narayanan (1983) and Narayanan and Bhattacharyya (1978) have made elaborate matheRegional Engineering College. t Indian Institute of Technology.
matical analysis of momentum and heat transport characteristics in irregular geometries with special reference to diverging-converging tubes. Navier-Stokes equation for steady-state, two-dimensional flow (in the annular space between a diverging-converging tube and an outer straight column) in terms of dimensionless stream function (Bird et al., 1960) has been solved numerically by using the line successive overrelaxation method. The values of velocity components so obtained have been substituted in the energy equation so as to derive the temperature profile (also numerically). The concerned equations are
where
E*4 = E*2(E*2) E*2’---a 2
1 a +a2
&*2 r* ar* &*2 ReM = Lvop//.l $* = $ / V J 2 ff* = [Re~Prl-’ V,* = V,/Vo VI* = V I / V o r* = r / L z* = z / L T* = T/To Assuming “no slip” a t the solid walls and constant wall temperature, the boundary conditions have been devel-
0888-5885/88/2627-Ol49$01.50/0 0 1988 American Chemical Society