Lvater vapor compared with an equilibrium composition of about 43%. This computation is not intended to imply that the gas composition close to the pellet surface is far removed from the equilibrium composition-in fact, at the low molar flow rates the n a t e r vapor concentration close to the pellet surface is undoubtedly considerably higher than the completely mixed concentration. In other words, this effect of molar flow rate on the reduction rate constant k can be explained in terms of an "effective" bulk stream gas composition. This so-called '.effective" bulk stream gas composition can be estimated in accordance with McKewan's experimental determination that the reduction rate constant k decreases essentially linearly u i t h increasing bulk stream water vapor concentration and becomes zero at the equilibrium concentration ( 3 ) T h e effective bulk stream gas composition may be estimated using the following equation relating k and the bulk stream composition
Lvhere c is the asymptotic value of k? P is the total pressure, PHI and P H g oare the partial pressures of hydrogen and of \cater vapor. respectively, and K is the equilibrium constant ( K = [PH20/PH2]eQ.). According to Equation 3 the effective bulk stream gas composition for a molar flow rate of 0.37 gram-mole/'sq. cm.-hr. at 816' C. is about ?,5% and at 1204' C. is aboat 29.27,. In a sense then, the difference between the effective bulk stream water vapor concentrations for 816' and 1204' C. at the lowest molar flow rate represents an indication of the additional work, in terms of increased gas flow rate. that is needed to achieve the asymptotic reduction rate at the higher temperature. Effect of Pellet Porosity. T h e maximum possible rate of reduction for any pellet occurs when sufficient H2 is transported to the pellet interior so that all available oxide sites are supplied \vith H2. As the porosity of the oxide layer increases, more sites are available to undergo reaction. If the porosity is high, H ? can readily diffuse into the oxide layer. Instead of having a sharp interface undergoing reaction as is observed \vith dense pellets, there will be a volume undergoing
reaction. Thus, because there are more reaction sites in a porous pellet. it should reduce faster than an equivalent dense pellet. This has been observed in the present and in previous studies. T h e experimental results of this study show that lower flow rates are required to obtain maximum or asymptotic reduction rates for dense pellets than for porous pellets (3)' T h e model used to analyze the data is derived by assuming that transport has no influence on the reduction of iron oxide pellets. If the above arguments are correct in including transport effects, why do the data fit the model? T h e answer to this question can only be speculated upon in a qualitative \Yay at the present time. It is recognized that reduction of a porous iron oxide pellet is not localized on a relatively sharp interface as it is for dense pellets, but occurs on an interface dispersed throughout the pellet. In the dense pellet, any finite area on the reaction interface loses oxygen at the same rate as any other finite area. I n the porous pellet this is not true. Material located near the surface of a partially reduced porous pellet is more highly reduced than material closer to the core. These oxygen concentration gradients in the solid during reduction are undoubtedly accompanied by concomitant water vapor concentration gradients in the gas phase dispersed throughout the pellet. T h e over-all reduction rate for porous pellets is therefore an .'integrated" value that accounts for the variety of conditions making up the complete reaction surface. T h e reduction proceeds topochemically at all points; hence. the proportionality between the rate constant, k: and the function [l (1 - R)113]for porous pellets may not be completely fortuitous. literature Cited (1) Kawasaki, E. P., "Kinetics of Iron Oxide Reduction with Hydrogen," Case Institute of Technology, Ph. D., Chemistry,
Inorganic. 1960. (2) McKewan, W. M., T r a n s . AZME 212, 791 (1958). (3) Ibid., 224, 2 (1962).
(4) Smith, N. D., McKewan, W.M., Blast Furnace, Coke Oven, and Raw Materials Committee, Iron and Steel Division. AIM& 21st Proceedings, p. 3 , Interscience, \Viley, Kew York. 1962. RECEIVED for review August 27. 1964 ACCEPTED March 5 . 1965
MASS TRANSFER MODEL FOR KOLBESCHMITT CARBONATION OF 2-NAPHTHOL P. G. P H A D T A R E A N D L. K. D O R A I S W A M Y ,Vattonal Chemical Laboratory, Poona, Indta
EVERAL
studies have been reported on the chemical
S mechanism of the Kolbe-Schmitt reaction and these have been reviewed, but n o theories have been advanced which might provide a n engineering basis for the carbonation of phenols by the Kolbe-Schmitt (or modified) procedures. T h e object of the present Lvork \vas to study the carbonation of 2-naphthol to 2,3-hydroxynaphthoic acid as a typical KolbeSchmitt reaction (using one of the modified procedures), and to propose a rraction model, supported by experimental 274
I&EC
PROCESS D E S I G N A N D DEVELOPMENT
evidence, which might serve as a starting point for a more rational engineering approach. 2,3-Hydroxynaphthoic acid, commercially known as BOX acid, is prepared by the action of dry COZon dry alkali naphtholate. T h e conversion, x " : obtained is of the order of 357, and yields 70%. T h e method consists of preparation of anhydrous sodium 2-naphtholate from 2-naphthol and aqueous caustic soda by vacuum dehydration; carbonation of the anhydrous salt with COZunder pressure at 230' to 270' C.,
A model i s proposed for the Kolbe-Schmitt carbonation of 2-naphtho1, supported b y experimental evidence, which might be a starting point for a rational approach to this problem. Based on mass transfer considerations, a irate equation has been derived which represents the data with a standard deviation of 1.59. Several justifiable assumptions have been made in the derivation of this equation; one important assumption i s that the composition of the product crust i s uniform. In the absence of a direct method for measuring the composition of the crust, the proportion of tar in the crust has been taken as an index of uniformity. Up to a reacti,on time of about 32 minutes, about 20% i s tar, and the proposed mechanism holds satisfactorily in this region. Beyond this period tar formation rises rapidly, and the equation no longer represents the data. It appears that the proposed model can b e used for the carbonation of phenols in general.
\vith intermittent distillation of 2-naphthol; and purification of the final product. T h e scheme of the reactions may be represented as follows:
ONa COONa
+
HzS04
-
Several modifications to this so-called dry method have been suggested : T h e use of a liquid medium for both dehydration a n d carbonation-e.g., 2-naphthol, dioxane, pyridine, toluene, and xylene. T h e use of sodium hydride with anhydrous alkali naphtholate during carbonation. Elimination of the dehydration step by using anhydrous K 2 C 0 3directly with 2-naphthol for carrying out the carbonation a t high temperature ,and pressure (Marasse method). Preparation of 2-hydroxy-1 -naphthoic acid and its subsequent conversion to 2-hydroxy-3-naphthoic acid (BOS acid).
Of the methods used since 1924, the one employing a liquid medium (7-4. 6. 77-75) has advantages which include reduction in mechanical load. and this method (using kerosine oil as liquid medium) \vas taken for theoretical analysis and study. Earlier work in this laboratory had clearly established the feasibility of this process ( 9 , 70); a conversion of about 36YG was obtained, but the yield was somewhat lo\ver (about 65%) than in the dry method. T h e process is, however, capable of further improvement, particularly with respect to dehydration.
hydrolysis of this complex gives free 2-naphthol and sodium carbonate in the presence of COZ. T h e total reaction is represented by
T h e free 2-naphthol does not form sodium naphtholate again, since in its fused form it is inert to alkali. In another side reaction 1 mole of water is formed per mole of xanthone, which is perhaps the main constituent of byproduct tar, and this water is again free to react as described above. T h e proportion of products formed \\ill depend on the rates of the different reactions involved. T h e principal over-all reaction can, ho\vever. be easily recognized : disappearance of sodium 2-naphtholate by reaction with CO,. This presupposes the complete conversion of 2-naphthol to the sodium salt during dehydration, a condition which can be ensured by continuing the dehydration until the entire theoretical quantity of water formed by Reaction 1 is removed. LVhatever the nature of the side product formed, it is possible to visualize the diffusion of CO? through a mixture of products as one of the steps in the reaction. T h e following reaction mechanism is proposed, based on models generally used for gas-solid reaction (5, 7, 8).
Model. T h e reaction mixture consists of sodium 2-naphtholate, dispersant (in this case, kerosine of a particular cut), and products of reaction. I t is assumed that: T h e naphtholate particles are in a state of suspension in the reaction mass. T h e reaction takes place by diffusion of COz across an equivalent film onto the particle surface. T h e composition of the product layer is uniform throughout the reaction. T h e particle is spherical in shape. T h e chemical reaction is extremely fast. .c. D .C
Reaction Model
Main Reactions. T h e principal reactions that occur during the carbonation of sodium 2-naphtholate are the formation of BOS acid with a n equivalent quantity of free naphthol, and the subsequent conversion of BOS acid and naphthol to a variety of compounds which together may be termed as tar. T h e presence of the slightest traces of moisture during carbonation results in side reactions which decrease the yield of BON acid. \Vater forms strong chelation with sodium 2-naphtholate, preventing addition of C O , to give BON acid. T h e
c,
i
0
Figure 1. Reaction model 1. 2. 3.
Equivalent kerosine film Outer product shell U n r e a c t e d sodium naphtholate
VOL. 4
NO. 3
JULY
1965
275
Figure 1 shows a sphere of naphtholate of radius R, on whose surface the reaction takes place by the diffusion of COS across a n equivalent kerosine film of thickness Z L . As the reaction progresses, the reaction plane retreats inside the sphere, leaving a n outer crust of products and a n unreacted sphere of the naphtholate. Thus, if at any time ’6 the reaction plane is situated at a distance, 7, from the center, the thickness of the product shell is given by ( R - r ) . T h e shell can be visualized as a mixture of products and kerosine oil. T h e rate of diffusion of C o t through this shell will depend on its effective diffusivity. With this model of the reaction, a n equation can be derived for the rate of carbonation of sodium 2-naphtholate from considerations of mass transfer (also heat transfer). Development of Equation. At any time 8, the rate of diffusion of COr across an equivalent film of kerosine outside the product shell may be ivritten as nA
= kL(CO -
c)
(5)
where kL
-VA =
KL4 T R2Coa ZLr ZLr R(R - r)
+
or a R2Cor’ LVA -- r fkL4 _ k v_ (l -_ r‘)
+
~
where r’ =
R
(fraction unreacted)
and
k.11
R -
1
ZLff
For every mole of COn 2 moles of naphtholate react ing this in mind, the rate of diffusion of CO;!can be equated to the instantaneous rate of disappearance of the naphtholate
D
= ZL
T h e number of moles diffusing per unit time is then given by =
.VA
kL 4 a R2(C0 - C )
(6)
where the right-hand side represents the rate of the chemical reaction. Using parameter r’ as defined earlier: Equation 14 may be rewritten as:
T h e rate of diffusion of CO, across the product layer is given
= - 8 pa
by nA =
kL’(C -
c,)
.4ssuming the reaction to be very fast-i.e.,
(7)
dr
R2rf2R ~
de
From Equations 13 and 15, we may write :
C, = 0 (16)
7’)Jdr’
where
If x represents the total conversion of the naphtholate to various products
T h e number of moles diffusing per unit time is therefore given
Integrating Equation 15 between the limits
by
e
=
0,r ‘
=
e
=
e, r t
=
1
and where 4 T rR represents the geometric mean of the two surfaces, 4 T R2 and 4 a r2. T h e effective diffusivity, D e , appearing in this equation may be written as
D,=aD
(9)
where D is the diffusion coefficient and cy is a n over-all constant \\hich accounts for the porosity of the product crust as \tell as the lack of linear passage through it. Equation 4 may now be rei\ritten as
we obtain, on simplification
0 = A.,f[0.33(1 - y3)
+ k.,f(0.17 - 0.5 y2 + 0.33
j3)]
where
Since y = (1 - x ) , Equation 18 can be used to predict reaction time as a function of over-all conversion.
(10) Experimental
From a consideration of mass balance it is clear that the moles of COS diffusing across the kerosine film should be equal to the moles diffusing across the product layer. Thus, from Equations 6 and 10, Equation 11 may be derived by appropriate rearrangement : C = cy
RCo(R - r ) ZLr R(R - r)
+
Substituting Equation 11 in 6 and simplifying: 276
I&EC PROCESS DESIGN A N D DEVELOPMENT
Equipment. To test the validity of the proposed reaction model, an experimental assembly was set up, the principal features of which are shown diagrammatically in Figure 2. I t was provided with a n anchor-type stirrer, rotating through a water-cooled stuffing box. In the flanged lid of the reactor were openings for charging, vapor outlet (used during distillation of the solvent), and COz inlet. A thermowell of appropriate length !vas also provided. T h e reactor was heated by electrical resistance Lvire \vound over its outer surface! and heat input was controlled through a Variac and ammeter. T h e reactor temperature was continuously recorded on a Cam-
1
2 3
4 5 6 7
8 9 10 11 12 13
REACTOR. 2 0 0 B x 2 0 0 H G T PRESSURE GAUGE CARBON DIOXIDE INLET KEROSENE OUTLET THERMOWELL CHARGING HOLE CONDENSER DISCHARGE VALVE STIRRER SHAFT WITH P U U E Y INTERMEDIATE PULLEY MOTOR W I T H PULLEY ELECT. RESISTANCE WIRE STIRRER
Figure. 2 Experimental assembly
bridge circular recorder. T h e entire reaction assembly was tested for a pressure of 300 p.s.i.g. a t 350' C. and was found to be leakproof. Procedure. Kerosine oil (boiling range, 160' to 265' C.) and 2-naphthol (97 to 987, pure) were charged in the reactor in a predetermined ratio. Stirring was then started and N a O H (exact stoichiometric quantity) was added as 507, liquor. Heating was commenced and the temperature brought to about 50' C. A condenser was connected to the reactor as shown in Figure 2, and the temperature was raised to 105' C., when a mixture of kerosine oil and wa.ter started to distill over. T o ensure a constant ratio of kerosine oil to 2-naphthol in the reactor? carefully dried kerosine was introduced into the reactor through a graduated reservoir a t the same rate a t which it distilled over from the reactor. By this distillation almost all the water could be removed in the first 2 to 3 hours. T h e last traces of moisture came o u t in the form of droplets in the distilling kerosine oil. T h e quantitative removal of water was ascertained in each run by measuring the water removed. Dehydration was considered to be complete only Ivhen the \vater removed was equal to the water added plus that formed during the reaction. To mak.e sure that the last traces of moisture had been completely removed, distillation a n d addition of kerosine oil were continued until there were three complete change-overs of kerosine oil from the reactor. T h e ratio of 2naphthol to dispersant in the reactor was then adjusted to the value required during carbonation. At the end of this adjustment, the temperature in the reactor reached a stable value of 230' to 235' C. in the majority of runs. C 0 2was then introduced through a silica gel t r a p (after the condenser was disconnected and all the openings were plugged) a t a slow, regulated rate a t the desired temperature. An appreciable absorption of COSwas noticed in the first half hour a n d feeble absorption u p to about 3 hours, after which there was practically no absorption. T h e time required for bringing the reactor to the desired conditions after introducing C O , and that taken u p in removing the COn after the reaction were kept a t the minimum. 2 to 3 minutes on a n average. This time was not included in the recorded carbonation time. T h e reaction \\,as stopped after the desired time interval, and the products of reaction were worked u p . Details of postreaction treatment have been described by Phadtare ( 9 ) :and only l.he salient features are given below.
\.\.:hen the temperature came down to 90' C., the residual gas was released, water was added, and the mass was heated to about 125' C. to ensure complete dissolution of the product. T h e solution was again cooled to about 90" C., discharged, and diluted to 8.6 to 10.2 Tw. T h e diluted solution was heated to 70' to 80' C. by steam for 15 minutes, after which steam \vas stopped but stirring was continued for 1 hour, and then allo\ved to settle for hour. This procedure helped to separate the tar efficiently. T h e tar was removed by filtration. T h e filtrate consisted of two layers, the aqueous layer (containing BON acid and 2naphthol) and the organic layer. After separation, the aqueous layer was heated to 65' to 70" C. and the pH brought do\vn to 6.8 by 50yc sulfuric acid when 2-naphthol was precipitated out. This was filtered, the filtrate was heated to 70' to 80' C., and its pH was brought down to 2 with 50% sulfuric acid, when BON acid precipitated out. T h e slurry was then cooled to 50' C. and filtered, and the BOK acid was \vashed free of acid a n d dried under vacuum a t 40' to 50' C. 2-Kaphthol and BOX acid were estimated by using a standard diazo solution of p-chloroaniline, and the results were verified by iodometric titration. Results Effect of Agitation. According to the film theory, the eddy transport in a fluid bulk can be expressed as the product of molecular diffusion and concentration gradient in a n equivalent film. With increased agitation, the eddy diffusivity increases, reaching a near-constant value a t a certain agitation. This will correspond to a certain equivalent film thickness. In order to apply Equation 18. which has been derived on the basis of the film theory, the film thickness contained in the constant k , should obviously be independent of agitation. T h e first step in the study was therefore to examine the effect of stirrer r.p.m. on the reaction. T h e response chosen in this case was conversion of 2-naphtho1 to BOY acid (see Table I ) and the results a t three stirrer speeds are sholvn in Figure 3. Beyond 150 r.p.m., the effect of agitation does not appear to be significant. I n all the kinetic runs, a n r.p.m. of 160 was maintained to eliminate the VOL. 4
NO. 3
JULY
1965
277
7
THEORETICALLY x : ~ AT 8 - 1 7
0 9a
w
I
1
ACTUALLY THIS W E S W T m X D SINCE CALCULATED VALUES OF AM AND k u
RUN 8 j
0830 -
0 7-
-s
0 6-
'x
r/li /
x
v
" z 2 0.5-
0 z
B M-
;
a z >
W
z
i
0
I
TEMPERATURE CARBONATIMY
II
~-NAPTHOL-KEROSENE RATIO
I
I
-6
s o b -
HR.
- CALCULATED
- 5OPSlG
PRESSURE
I 10-
--260'C* TIME,
-1
e
11
EXPERIMENTAL
TEMPERATURE PRESSURE ~-HAPHTHOL/KEROSEHE} DURING CARBONATION
/
I I
-
26O.C. 7 0 PSlG
- ,:,
1 I I
a
L
I 20
I
I
40
64
I
so
I
I
I
XXI
120
14
I
1w
0
100
V
I
I
I
I
1P
20
30
40
CARBONATION TLME
STIRRER RPM
Figure
3. Effect of agitation on carbonation
effect of external agitation. T h e conversion plotted in Figure 3, x", is the definition normally used in industry, and should be distinguished from the other conversions showm in Table I. Rate Data. A series of experiments was then carried out under conditions where tar formation would be minimum, so that the product layer may be assumed to have a uniform composition. O n the basis of several earlier experiments (not recorded in this paper) (9, I O ) , the conditions established for this purpose were : 2-Naphthol to kerosine ratio Carbonation temperature Carbonation pressure Carbonation time
1:l-1:1.25 250-270" C. 50-70 p.s.i.g.
