2050 Ind. E n g . C h e m . Res.

1988,27, 2050-2055
British Petroleum, Br. Patent 930 457, 1963a. Ioka, M.; Wakabayashi, M.; Oguchi, Y. Japan Kokai 7830996,1978.
British Petroleum, Belg. Patent 625 486, 1963b. Nagata, S.; Shimoda, S.; Sudo, T. Clays Clay Miner. 1974,22, 285.
British Petroleum, Br. Patent 625 269, 1963c. Occelli, M. L. Znd. Eng. Chem. Process Res. Deu. 1983, 22, 553.
British Petroleum, Belg. Patent 621 204, 1963d. Perez-Pariente, J. Ph.D. Thesis, Institute of Catalisis, Madrid, Spain,
British Petroleum, Fr. Patent 1376 095, 1964. 1984.
British Petroleum, Neth. Patent Appl. 506 019, 1965a. Pupko, S.Fr. Patent 1383 960, 1965.
British Petroleum, Belg. Patent 659 679, 196513. Rautureau, M. Ph.D. Thesis, Orleans, France, 1974.
British Petroleum, Ger. Offen 2 108276, 1971. Rautureau, M.; Mifsud, A. C. R. Hebd. Seances. Acad. Sci., Ser. C
Coenen, J. W. E.; Linsen, B. G. Physical and Chemical Aspects of 1975,281, 1071.
Adsorbents and Catalysts; Academic: London, 1970. Rautureau, M.; Mifsud, A. Clay Miner. 1977,22, 285.
Corma, A.; Miranda, M. A.; Perez-Pariente, J. React. Kinet. Catal. Reucroft, P. 0.; Pare!&, H.; Ganesan, P.; Russell; S. N.; Davis, B. H.
Lett. 1983,23, 153. Appl. Catal. 1982,3, 79.
Corma, A.; FornBs, V.; Mifsud, A.; PBrez-Pariente, J. Clay Miner. Serna, C.; Fernindez Alvarez, T. An. Quim. 1975, 71, 371.
1984, 19, 673. Serna, C . ; Van Scoyoc, G. E. Proc. Znt. Clay Conf. 1978, 197.
Corma, A.; Mifsud, A.; Perez-Pariente, J. Clay Miner. 1986,21, 69. Shil, L. N.; Bhatla, S. Znd. Eng. Chem. Prod. Res. Dev. 1986,25,530.
Dandy, A. J.; Nadiye-Tabbiruka, M. S.Clays Clay Miner. 1975,23, Singer, A.; Galan, E. Palygorskite-Sepiolite. Ocurrences, Genesis
428. and Uses; Elsevier: Amsterdam, 1984.
Feitknecht, W.; Berger, A. Helu. Chim. Acta 1942, 25, 1543. Spenadel, L.; Boudart, M. J. Phys. Chem. 1960, 64, 205.
Fernindez Alvarez, T. Proc. Znt. Clay Conf. 1972, 571. Suzuki, N.; Tsutsumi, K.; Takahashi, H. Zeolites 1982,2, 51, 185.
Fernindez Hemindez, M. N.; Fernhdez Alvarez, T. An. Quim. 1982,
79, 342. Received for review October 14, 1987
Fuentes, S.; Figueras, F. J. Catal. 1978,54, 397. Revised manuscript received June 27, 1988
Fuentes, S.; Figueras, F.; GBmez, R. J. Catal. 1981, 68, 419. Accepted July 7, 1988

Dehydrogenation of Butan-8-01on Zinc Oxide Catalyst: A Continuous

Stirred Tank Reactor Study
Sankaran Ravi and T. S. Raghunathan*
Department of Chemical Engineering, Indian Institute of Technology, Bombay 400 076, India

The kinetics of dehydrogenation of butan-2-01 on a zinc oxide catalyst was studied in a specially
fabricated continuous stirred tank reactor using annular ring-shaped disks of the catalyst. Ex-
perimental data were collected in the temperature range 400-490 "C and a t W/FA,,'sranging from
5X to 48 X lo4 g of catalyst/mol of butan-2-ol/s. An interesting feature concerning the rates
of reaction is the occurrence of a maximum in conversion at constant temperature. A dual-similar-site
mechanism with surface reaction as the rate-controlling step together with negligible K R and K s
is shown to account for all the observed results. Above 475 "C, however, a dual-dissimilar-site
mechanism with hydrogen desorption as the rate-controlling step also explains the data. An evaluation
of the internal diffusion effects on the basis of Weisz-Hicks criterion indicates that the effectiveness
factors were probably close to unity under most of the experimental conditions. Thus, the rate
equation proposed, viz., r = k p A / ( l + K A ~ A )could
~ , be said to describe the true kinetics of this
reaction.

Dehydrogenation of butan-2-01 (sec-butyl alcohol) is an point is that the measurable exit conditions of concen-
industrially important reaction in that it yields methyl trations and temperature correspond to those inside the
ethyl ketone or butan-2-one, an industrial solvent, as the reactor. The rate of a reaction represented by the stoi-
chief product. This reaction is usually conducted in the chiometric equation
temperature range 325-425 "C on a catalyst such as zinc
oxide activated with Na2C03,Cu-Ni on Kieselguhr, Raney A=R+S (1)
copper etc. (Stephenson, 1966). A literature survey re- may readily be obtained from the equation
vealed that there are no published kinetic data on zinc
oxide catalyst. Miller and Wu (19721, however, report that
the activation energies of the reaction on a zinc oxide
catalyst with and without an oxygen pretreatment me 1.8 where X A is the exit fractional conversion of A in a CSTR
and 28.9 kcal/mol, respectively. No attempt was made by operating with a mass of catalyst equal to W and molar
Miller and Wu to present a rate equation for this reaction feed rate of A equal to F A @ When (1)occurs on a solid
on the zinc oxide catalyst. The present study was un- surface and the reaction mixture itself is in the gas (or
dertaken with the aim of obtaining a rate equation for the vapor) phase, it is necessary that any kinetic data collected
dehydrogenation of butan-2-01 on a zinc oxide catalyst. be such that it is possible to eliminate or take into account
It is well-known that the reactor system that yields the external physical transport resistances. In this respect,
direct rate information on any reaction is the well-mixed CSTR affords certain advantages over a fixed bed reactor
continuous stirred tank reactor or the CSTR. This type in that the agitation control can be achieved independently
of reactor is characterized by uniformity of concentration even at low feed rates of the reactants. An experimenter
of the species in the reaction together with constancy of is thus at a greater advantage with a CSTR for kinetic
temperature in the entire reactor space. An additional studies.
The experimental reactor system employed in the
* To whom all correspondence should be addressed. present study amounted (Shah, 1982) to a well-mixed re-
0888-5885/88/2627-2050$01.50/0 0 1988 American Chemical Society
 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988 2051

REACTOR

@ GUN METAL BUSH BEARING

SCALE -1 1
( All dimension are in m m )

Figure 2.

was effected with the aid of a chromel-alumel thermo-

couple entering through the exit at the bottom of the re-
actor and positioned at a distance of about 1 mm below
the tip of the stirrer shaft. Traces of temperature-time
recordings during the reactor operation showed that the
'L - - - U I
---l temperatures remained within f1.5 "C.
A- Feed system G - Motor Experimental Procedure
B - Proheater H-Thermocouple position in Reactor Preliminary blank experiments showed that there was
-
C Furnace S-Soap Filmmeter no catalytic activity of the metal parts of the reactor with
D-Reactor Th -Thermocouple which the vapors of butan-2-01 came into contact. The
E-Stutting boa Ms-Motor starter general procedure of any run was to send a constant liquid
F- Condensers
alcohol flow (f3%) to the evaporator-preheater, and the
resulting hot vapors were sent into the catalytic reactor.
Figure 1. Schematic diagram of the setup. The exit vapors were cooled and condensed to obtain al-
actor where annular disks of zinc oxide were rotated at cohol- and ketone-free exit gases. The liquid collected was
such speeds that external mass-transfer resistances were analyzed for the constituents on a gas-liquid chromato-
essentially absent. The details of the experimental system graph on a column filled with 5% Carbowax 20M on
and procedure are discussed below. Chromsorb W with the aid of a hydrogen flame ionization
detector. The peaks obtained were very sharp and the
Experimental Section peak heights were taken to characterize the composition.
A schematic diagram of the experimental setup com- The liquid reaction mixture when analyzed indicated the
prising the feed system, the reactor, and the condenser is presence of methyl ethyl ketone and butan-2-01 only. The
shown in Figure 1. A brief description of the reactor is ratio of the ketone to alcohol in any experimental run was
as follows. compared with a calibration curve of this ratio plotted
Reactor Details. The reactor body was made of against the molar composition of the ketone in syntheti-
stainless steel by drilling out of a 50- X 70-mm cylindrical cally prepared samples. The fractional conversion of the
rod a cylindrical space 33 mm in diameter and 50 mm in alcohol, XA,was thus obtained. Traces of dehydration
depth, the last 5 mm rounded off to avoid stagnant regions. product (but-2-ene) were observed on virgin catalyst, and
A 100-mm-diameter, 12-mm-thick flange having six 9-mm this ceased within 30 min of operation.
holes at a PCD of 91 mm was provided on the top part of Catalyst Preparation. Several batches of zinc oxide
the reactor body to be mated with an identical closure were prepared starting from either basic zinc carbonate
flange carrying the stirrer assembly and the stuffing box. or zinc nitrate hexahydrate (supplied by Robertson
The stirrer assembly itself consisted of a 6-mm cylindrical Johnson Co.). The various catalysts prepared were pelleted
rod fitted with propeller blades and catalyst arms on which in the form of annular disks (13 X 4 mm) for use in the
annular disks of the catalyst could be mounted. A section reactor. Some of the catalyst preparations did not yield
of the reactor, stirrer, and the stuffing box is shown in mechanically strong disks, and hence only such prepara-
Figure 2. tions that resulted in mechanically strong disks were tested
Catalyst disks of 13-mm 0.d. and 4-mm i.d. of differing in the kinetic study. Of the various catalysts prepared,
thicknesses in the range 0.9-2.5 mm were prepared in a the zinc oxide obtained by the decomposition of the basic
hand-operated press at a load of about 2 tons. zinc carbonate at 400 "C for 2 h followed by kneading with
Freshly prepared catalyst powder was used to form these distilled water and drying at 110 "C for 2 h gave the highest
disks without any addition of binders, die lubricants, etc. conversion coupled with good mechanical stability of the
Plate 1 shows a view of the catalyst disk and the stirrer disks. All the kinetic results reported in this work were
assembly. The entire reactor containing the catalyst disks obtained on this preparation. The details of the perform-
was housed in a detachable two-piece furnace provided ance of the other preparations tested are given elsewhere
with a proportional temperature controller to maintain the (Ravi, 1980).
temperature in the reactor at any level in the range in- In any given run, upto four disks of the catalyst could
vestigated. The temperature measurement in the reactor be charged in the reactor, and this indeed was one of the
2052 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988

Table I. Exuerimental Data

at 425 "C at 450 "C at 475 " C a t 490 " C
WIF X A r X A r X A r X A r
2.79 0.06 0.022 0.09 0.032 0.16 0.056 0.20 0.072
3.64 0.12 0.033 0.18 0.049 0.26 0.071 0.31 0.085
9.14 0.26 0.028 0.37 0.040 0.47 0.051 0.52 0.057
12.51 0.31 0.025 0.41 0.033 0.52 0.042 0.59 0.047
14.42 0.82 0.057 0.895 0.062 0.93 0.064 0.93 0.064
18.21 0.40 0.022 0.49 0.027 0.60 0.033 0.64 0.035

o
major limitations employed in the present study. The only
way to increase the mass of the catalyst was to increase
the thickness of the pellets which might bring in significant
pore diffusion effects. 009

Results and Discussion

Table I lists the experimental data with respect to the
rates of reaction at various conversion at 425 "C. The
complete set of data used in this study is available in Ravi
(1980). Attempts were made to fit the data into equations
derived on the basis of assumed mechanisms, and the
various equations tested are shown in Table 11. At the
outset, it may be remarked that none of these expressions
fit the data in the entire range of conditions, meaning the
data analysis revealed one or more negative constants of
the 14 equations shown in Table 11. However at 475 and
490 "C,it was found that the dual-dissimilar-site mecha-
nism with hydrogen desorption as the rate-controlling step
described the data well. The rate equation under these
conditions is given by
~~(PA/PR)
r= (3)
1 + K$(s@A/PR)
The equilibrium constants in the reaction, K,, are reported
by Kolb and Burwell (1945). The partial pressures of the i-
three components in the reaction are related to XA; thus, L l l l l l l l l l i l l l
0 0.1 0.2 0.3 0.4 05 06 0.7
1 - XA XA Convorsion X A
P A = - PR=PS=- (4) Figure 3. Rate of reaction versus conversion with temperature as
a parameter.
at a total pressure of 1 atm, corresponding to a pure alcohol
feed to the reactor. The estimates obtained for k and Ks A second differentiation of (6) may be carried out to obtain
in ( 3 ) are shown in Table 111. the second derivative of r with respect to X A which can
Dual-Similar-Site Mechanism with Negligible Ad- then be evaluated at XA,, given by (7); thus,
sorption Equilibrium Constants of Ketone and Hy-
drogen. In an attempt to find a single rate equation that 2 k [ ( l - KA)' - ( 1 + KA)']
would describe the kinetics of the reaction in the entire
range of experimental conditions, the rates of reaction were
plotted against X , at various temperatures (Figure 3). It
is seen that these curves exhibit a maximum, and this led
to the necessity of examining such rate equations which
display this characteristic. One such rate equation is
[ (1

Thus, the sign of (d2r/dXA2)Tis the same as that of the

numerator of the right-hand side of (€3)) viz., 2k(-4K~),
+ KA)' - ( 1 - KA)'

Le.,
l + K A
I
negative (<O) when both k and KA are positive. In other
kPA words, the rate of reaction ( r ) passes through a maximum
r = (5)
(1 + KAPA)' according to ( 5 ) at XAmgiven by ( 7 ) . The corresponding
maximum rate of reaction is readily obtained by substi-
By differentiation of ( 5 ) with respect to XA,one observes tuting (7) in ( 5 ) together with (3). Thus,
that
r, = k / 4 K A (9)

(&IT + + ( 1 - KA)XA)' +
= ( - 2 k [ x ~ ( ( K 1~)

(1 - X A ) ' ] [ (-~KA') + ( 1 - K A ) ' X A ] ) / ( [ ( K+A1 ) +

For purposes of examining the validity of ( 5 ) with respect
to the data of the present study, this equation was rear-
ranged in the form

( 1 - K A ) X A ] (6) ~)

Equating the right-hand side of (6)to zero, one finds that and a linear regression was performed. The results of the

(
&)T = 0 at X A , , = --
1-KA
KA+1
=
KA-1
KA+1
~ (7)
estimates of k and KA are shown in Table IV.
A few additional comments are in order concerning the
rate equation ( 5 ) . A comparison of (5) with (H) of Table
 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988 2053
Table 11. Possible Mechanisms in the Reaction A = R + S on a Solid Catalyst and the Corresponding Rate Equations
mechanism rate-limiting step rate eq
single site, A and R adsorption of A
as the adsorbed species

surface reaction

desorption of R
k (;: 2)
single site, A and S adsorption of A
as the adsorbed species

surface reaction

desorption of S
k (;: E)
dual similar site adsorption of A

surface reaction between

adsorbed alcohol and an
active site (adjacent)
desorption of R
k (E 2)
1 + KAPA+ KRKe-
PA
Ps
+K~ps
desorption of S (J)

dual dissimilar site, A and R adsorption of A

adsorbed on one type of k ( P A - Y )
active site and S adsorbed
on a second type of active center

surface reaction

desorption of R

desorption of S

I1 clearly shows that the former is a special case of the may prevail with respect to XA,m. The estimates of KA of
latter in which K Rand K,s are negligible. -A further point the present study comply with this condition. It may also
emerges from (7) in that K A must. be greater than unity be remarked that at least two earlier workers reported the
(KA > 1) in order that physically meaningful situations validity of the rate equation of the form shown in (5).
2054 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988

Table 111. Constants k and K, in Equation 3 Table V. Asymptotic Effectiveness Factors at Various
temp, “C w 5 k , mol/((a cat) s) KR Temperatures: Experiments with W/FA, = 14.42 g of
catalyst/mol of A/h
475 8.52 0.027
490 6.57 0.008 temp, XAfrom the
“C ?asymptotic X~srptlltaaymptotic soh of (11)
Table IV. k,KA, and the Rate Constant of the Surface 425 1.762 0.465 0.482
Reaction in Equation 5 450 1.364 0.651 0.620
475 1.222 0.757 0.704
temp, W k , mol KA, 490 1.200 0.771 0.733
“C A/((g cat) s atm) atm-* 106k, = k / K A
425 1.72 4.41 3.90 these conditions but would involve the solution of the
450 1.70 3.52 4.83 nonlinear differential equation relating the concentration
475 1.66 2.64 6.29 of A at any position inside the catalyst disk and the size
490 1.61 2.11 7.63
of the disk together with the other parameters such as p ~ ,
Thus, Nondek and Kraus (1975) report that the initial KA, etc. It must be remembered that this equation is
rates of dehydrogenation of several C3-C7 secondary al- formulated to include the effects of (1)the peculiarities
cohols on various chromia catalysts are expressible by an of the kinetics, i.e., of (5), and (2) decreasing density of the
equation identical in form with that of (5). The same reaction mixture, owing to the stoichiometry of the reac-
reaction on a brass catalyst was studied by Thaller and tion. No detailed numerical solution was attempted at this
Thodos (1960) and was shown to have an identical rate stage, but an asymptotic approach to the effectiveness
equation. factors was taken, and this leads to the following expres-
It is useful to point out certain implications of (7) and sion:
(9) with respect to the variation of KA with increasing VaaVmDtotic
--
temperature. Thus, if KA decreases as the temperature
is increased (as is the case in the present study), (7) dictates
that XA,, should decrease with a temperature increase.
According to (9), the value of rm should increase at the
same time considering that k varies only slightly (Table
IV), the overall variation being about 6% decrease. Thus,
an anomalous situation is observed in the present study
in that the results conform with the latter implication The results of the evaluation of 7 at the four tem-
concerning rm but not the former. In other words, both peratures indicated are shown in ?%f%.
Table V lists
r , and XA,m increase with an increase of temperature. the calculated values of XA,ax,,fl/?)wymputic,
which could be
Internal Diffusion Effects. Attempts were made to looked upon as being proportional to the “pore-diffusion
assess the possible involvement of significant pore diffusion effects-corrected” rate of reaction at W / F = 14.42 g of
effects in the kinetic experiments of the study. At the catalyst/mol of A/h. The last column of Table V provides
outset, it may be pointed out that the effectiveness factors the solution of the cubic equation obtained from (2) to-
in situations represented by kinetic expressions such as gether with (4) and (5), i.e.,
(5) are decided by at least two parameters: (a) the modulus
combining the rate constant, effective diffusivity, shape
x~~+ bXA2 CXA d = 0 + + (12)
and size of pellet, etc.; (b) the adsorption coefficients such where
as KA together with the partial pressure levels on the 2(1 - KA2) + k ( W / F )
catalyst surface. In fact, in (l), there is bound to be a third b=
factor in the form of an “expansion factor” accounting for (1- K A ) ~
the density decrease of the reaction mixture due to the
stoichiometry of the reaction. A rigorous approach would (1 + K.J2 k(W/F) (13)
C= d=
involve the solution of a nonlinear differential equation, (1- K A ) ~ (1- K A ) ~
and it is a two-point boundary-value problem that needs
to be solved numerically. A simplified approach, however, Equation 1 2 relates the conversion (XA)to the operating
was resorted to: in the modulus of Weisz and Hicks ac- conditions in the CSTR, viz., temperature (which decides
cording to Satterfield (1971), it was assumed that, under k and KA) and WIF. The above equation was solved with
conditions when this modulus is quite small, the pore- the parameters k and KA as given in Table IV together
diffusion effects are unimportant and hence the effec- with W / F = 14.42 at the four temperatures shown. The
tiveness factor wouldbe essentially unity. In order to ob- good agreement clearly indicates the validity of the pro-
tain estimates of this modulus, it is necessary to have an posed rate equation in explaining the experimental data
estimate of De, the effective diffusivity of A in the reaction of the present study.
mixture. Based on the effective diffusivity data reported Thus, it may be concluded that eq 5 describes the ki-
for the ethanol-hydrogen system (Satterfield, 1971), netics of the dehydrogenation of butan-2-01 on the zinc
measurements of Defor H2-N2 in zinc oxide pellets (Ra- oxide studied. The constants k and KA which are tabu-
ghunathan, 1977) assumed that D e= 0.1 cm2/s for pur- lated in Table IV show the following temperature depen-
poses of the calculation of the modulus. dencies:
Calculations of the modulus, (Vp/S,,)2[r,,,/(oeC,)], at
the various operating conditions showed that this modulus
was much less than one except in the case of the runs with
W / F = 14.42 g of catalyst/mol of A/h. It may therefore
be inferred that the pore diffusion effects are probably
significant with respect to the experimental data obtained
in these experiments. A correct approach would be to It may be added here that (14)implies an activation energy
evaluate the effectiveness factors of the catalyst under of about 10.8 kcal/mol for the catalyst studied.
 Ind. Eng. Chem. Res., Vol. 27, No. 11, 1988 2055

Dual-Dissimilar-Site Mechanism. It was remarked pA,pR,p s = partial pressures of A, R, and S, atm

earlier that the data obtained at 475 and 495 "Ccould be r = rate of reaction, mol/(s (g of catalyst))
described by (3) as well. That such a form of the rate robs= rate of reaction per unit volume of catalyst, mol/(s (cm3
equation describes the kinetics of this reaction on various of catalyst))
zinc oxides has been established by Spolnicki (1964)) Sex= external surface area of catalyst pellet, cm2
Nirdosh (1974),and Raghumthan (1977).An explanation T = temperature, "C or K
for the validity of both mechanisms a t the two higher V , = volume of catalyst pellet, cm3
temperaturea in the preaent study may roughly be sketched XA= fractional conversion of A
as follows. Under these reaction conditions, the rate of Greek Symbols
hydrogen evolution due to reaction was higher, and this = effectiveness factor
TJ
could possibly m u l t in a surface reduction of the zinc oxide pp = pellet density, g of catalyst/cm3
which in turn could be the precursor to the formation of R, S = methyl ethyl ketone and hydrogen, respectively
a second type of active site (Zn atom). Further, it is Registry No. Butan-2-01,78-92-2;butan-2-one, 78-93-3;zinc
postulated that hydrogen molecules tend to get adsorbed oxide, 1314-13-2.
strongly on these active sites. It is then probable that a
switch in mechanism from a dual similar site (surface re- Literature Cited
action controlling) to a dual dissimilar site (hydrogen de- Kesavulu, S.; Taylor, H. S. J . Phys. Chem. 1960,64, 1124.
sorption controlling) occurs under these conditions. That Kolb, H. J.; Burwell, R. L. J. Am. Chem. SOC.1945, 67, 1084.
two different types of active centers are indeed produced Miller, K. J.; Wu, J. L. J. Catal. 1972, 27, 60.
on zinc oxide during hydrogen treatment is reported by Nirdosh, I. Ph.D. Thesis, The University of Birmingham, 1974.
Kesavulu and Taylor (1960). Nondek, L.; Kraus, M. J. Catal. 1975,40,40.
Raghunathan, T. S. Ph.D. Thesis, The University of Birmingham,
Nomenclature 1977.
&vi, S. M.Tech. Dieaertation, Indian Institute of Technology, Bom-
A = butan-2-01 bay, 1980.
Cs = concentration of A at the catalyst surface corresponding Satterfield, C. N. Mass Transfer in Heterogeneous Catalysis, M.I.T.
to p M , = p M / R T , mol/cm3 Press: Cambridge, 1971.
De = effective diffusivity __
Shah, J. M. M.Tech. Dissertation, Indian Institute of Technology,
E = activation energy Bombay, 1982.
k = rate constant in (51, mol of alcohol/(s (g of catalyst) atm) SDohicki. J. Ph.D. Thesis. The Universitv of Birmineham. 1964.
Stephenson, R. M. Introduction to the Chemical ProcesHZnd&ries;
k l = rate constant in (3), mol of alcohol/(s (g of catalyst)) Van Nostrand: Reinhold, NY, 1966; p 269.
k, = k/KA Thaller, L. H.; Thodos, G . AZChE J. 1960, 6, 369.
KA, KR, KS = adsorption equilibrium constants of A, R, and
S, atm-' Received for review April 24, 1987
K , = equilibrium constant of (3) Accepted March 31, 1988