Kinetics of Cuprous Iodide Oxidation with Air - Industrial & Engineering

J. Michael Davidson and Donald H. Glass. Industrial & Engineering Chemistry Research 2007 46 (14), 4772-4777. Abstract | Full Text HTML | PDF | PDF w/...
0 downloads 0 Views 338KB Size
KINETICS OF CUPROUS IODIDE OXIDATION WITH AIR H E I N Z

J .

NEUBURG'

Chemical Engineering Department, Universidad Tkcnica del Estado, Casilla 10233, Santiago, Chile The kinetics of oxidation of cuprous iodide w i t h air was studied in a batch system by the thermogravimetric technique between 300' and 400' C. Cuprous iodide precipitated as microparticles was used, and all experiments were carried out using a large excess of oxygen. The reaction was autocatalytic, with an activation energy of 13.5 kcal./g.-mole. Reaction rate data were correlated by the semi-empirical equation ( d x l d t ) = 3.53

X 103exp(-13, 5 0 0 / R T ) ~ "1~- (

x ) ~ ~

by using linear regression. With this equation and the residence time distribution function for series of equal-size backmix flow vessels, conversion was estimated v5. residence time and number of ideal backmix stages under the assumption of completely segregated flow. These results were compared with those for plug flow of solids.

THEoxidation

of cuprous iodide to cupric oxide is of interest in the copper industry, because it is part of some process for copper recovery from dilute solutions of cupric sulfate. As this reaction has not been studied from a kinetic point of view before, the purpose of this paper was to provide some preliminary kinetic information useful for design. The reaction is:

Cur12 (s) +

0 2

(g) = 2 c u o (s) + I? (g)

(1)

Heat of reaction and standard free energy change at 298.16"K. are: A H o = -26.52 and AGO = -25.85 kcal.1 g.-mole of Cu&. The reaction is.. exothermic a t higher temperatures, and the estimated values for the equilibrium constant a t 500" and 1000°K. ( K = 3.8 x lo4 and K = 1.7 x lo6, respectively) show that the reaction can be considered irreversible within this temperature range. Present address. 30 Charles St West, Toronto 189, Ontarlo, Canada

Equipmeni

The apparatus was a thermobalance assembled (Figure 1) by attaching a wire frame to one end of a H3 Mettler balance. A glass pail (about 2 cm. in diameter and 1 cm. high) was attached to the other end of the frame by means of a thin glass chain. The pail, containing the solid sample, hung freely inside the furnace. The furnace was made from a glass tube of 35-mm. i.d. and 43.5-cm. total length, and was used in a vertical position. I t was heated by an electrical resistance (DS Kanthal flat-wire 0.1 x 1 mm.) wrapped around all its length. The furnace was mounted inside an insulating tube which protected it mechanically and provided a thermostated layer of air. The pail hung in the part of the furnace where longitudinal temperature gradients were minimum. A thin scabbard of glass (about 1-mm. diameter) closed at one end was used as the guide and

I FURNACE 2 BALANCE 3 AUTOTRANSFORMER 4 GAS CONTROL VALVE

5 POTENTIOMETER 6 WASHING FLASK 7 FLOW METER

8 CONSTANT PRESSURE DEVICE

9 WATER PUMP

IO WATER CONTROL VALVE

Figure 1. Diagram of apparatus

Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 2, 1970 285

support of an 0.08-mm. iron-constantan thermocouple embedded directly in the solid sample. The air flow for the reaction was induced by a water pump (Figure 1).

10

0.8

Experimental Procedure

To prepare cuprous iodide, pure iodine was placed in a glass pail with water, and sulfur dioxide was bubbled through it until the iodine completely dissolved and a solution of H I and H S 0 4 was formed. This solution was introduced simultaneously with cupric sulfate solution of equal normality to an agitated vessel and SO, was bubbled through it again. Once precipitation was complete, the clear acid solution was decanted. The precipitate was washed with water until neutral, filtered, and dried a t 90" C. Finally the dry cake was reduced in size and screened. Particles under Tyler mesh 115 were used for this work. These particles, observed under the microscope, appeared as agglomerates of smaller subparticles whose diameter was not determined. Air flow was 0.3 liter per minute (STP), which provided a t least 17 times the oxygen consumed by the reaction. No attempt was made to establish the effect of oxygen partial pressure on reaction rate. The heating of the reactor was hand-controlled by an autotransformer. Once the reactor was a t steady state at the desired temperature level, a previously weighed sample of cuprous iodide (about 1 gram) was introduced, and the time was registered. Each 3 minutes the weight of the reacting solid was read (&O.OOl gram) until no more variation was observed. Runs were carried out a t 302.20, 316.2", 331.5", 350.5", 376.0°, and 400.0" C. Results

Illustrative experimental data are shown in Table I. Direct readings were time, mass, and temperature. Conversion at any time, x,, is obtained directly from mass readings according to Equation 2. For each time

0.6

0

5

8

0.4

u 0.2

0 0 0

20

40

80

60

100

120

Figure

M , = 0.999 gram MF= 0.420 gram Q = 0.33 literlmin tz,

Ma >

Min. G u m

T,

c.

0

0.999

0.000

3

0.990 394.0

0.016

6

0.965

401.4

0.059

9

0.894

403.6

0.181

12

0.799

401.1

0.345

15

0.691 400.2

0.532

18

0.589

400.6

0.708

21

0.502

397.6

0.858

24

0.449

400.0

0.950

27

0.420

402.2

1.000

286

r,

X,

160

2. Conversion vs. time

interval the average reactions rate and average conversion were obtained by Equations 3 and 4.

x L = ( M ,- MI) / ( M ,- M,)

(2)

(4) Constancy of temperature during each run was within 12OC. except for the first 3 minutes, when thermal steady state was to be achieved. Figures 2 and 3 are plots of the experimental conversion us. time and reaction rate us. conversion data. An Arrhenius plot of the experimental data was made on the following basis: I t was assumed that the Arrhenius law accounts for the effect of temperature on the reaction rate. Hence, we can write

(dxldt)= k, exp(-E/RT).f(x) Table I. lime vs. Mass and Conversion vs. Reaction Rate

140

TIME, min

(5)

Integration of Equation 5 at constant temperature between two fixed levels of conversion x1 and xz leads to

Mole CUI Converted Initial Moles CUI x Min.

0.0078

0.00518

0.0371

0.01439

0.1200

0.04087

0.2634

0.05469

0.4387

0.06218

0.6201

0.05872

0.7833

0.05009

0.9042

0.03051

0.9750

0.01670

Ind. Eng. Chem. Process Des. Develop., Vol. 9, No. 2, 1970

Hence, a straight line results if log 1 / 0 2 - tl) is plotted against 1 / T , if the rate data follow the Arrhenius law (Figure 4). Conversion levels x1 and x 2 were chosen as 0.2 and 0.4, respectively, because of greater confidence in the experimental data in that range. Analysis of Figure 3 leads to the following conclusions: The shape of the curves of reaction rate us. conversion is the same for all temperature levels. These curves show a maximum, which is characteristic of autocatalytic reactions. This shape would not be observed if diffusional resistance (gas-film or pore diffusion resistance) were the rate-controlling step. Hence, chemical reaction was assumed to be the controlling step in the temperature range explored. This assumption is consistent with the absence of any significant change of apparent activation

72

I

I

1

I

I

I

I

I

I

40O.O0C 64

56

48 c

I

C

'1 -

40

0

c

c'

2

32

z

P ci

2

\ \i

24

16

8

0

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

1.0

0.9

CONVERSION

Figure 3. Reaction rate vs. conversion

energy in the Arrhenius plot (Figure 4 ) , which would be observed if change in the controlling step occurs when the temperature level is increased. The same behavior was observed for other reactions by Bond and Clark (1960) and Pavlyuchenko and Rubinchik (1951). So, considering that the reaction is practically irreversible and autocatalytic, a rate model of the type

10-2 0.8

0.6

r IC '-

0.4

-. E.

e

r = k , exp(-E/RT) xb (1 - x)'

(7)

was selected for correlation of the rate data; parameters k,, E , b, and c were obtained by linear regression in an I B M 360 digital computer. With these values Equation 7 becomes:

r = (3.53 x

4

r

0.9

io3min .--I) 1.4

This equation correlates the experimental data with an average arithmetical deviation of 12.9%, which seems reasonable, taking into account that the values of r were obtained by directly differentiating by finite increments without previous smoothing of the data. However, Equation 8 does not apply in the neighborhood of x = 0 and x = 1.0; it predicts r = 0 for x = 0, which is not in accordance with the experimental fact that the reaction starts by itself with pure CUI. The different values of b and c take into account the asymmetry observed in the curves in Figure 3.

1s

1.6

1.8

Figure 4. Arrhenius plot

to its individual residence time by Equation 9. For the case of a reactor with ideal plug-flow solids, the individual residence time is the same for all particles and equal For this case the average to the mean residence time conversion 7 is related to the mean residence time through Equation 10.

Reactor Performance

The effect of backmixing on the reactor performance may be estimated on the basis of the kinetic data if segregated flow, as well as a large excess of air and absence of diffusional effects, is assumed. Conversion of any particle at the reactor exit is related

I .7

( 1 0 ~ 1 ,~ )OK-'

dx,

y

L=L -

t = J o

X d x

r

Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 2, 1970 287

For series of equal-sized backmix reactors, the average exit conversion is given by Equation 11 (Levenspiel, 1962) -

x = 1-

m

(1 - X I ) E@,)dt,

(11)

where the residence time distribution for j equal-sized reactors [E(t,)]in terms of mean residence time is (Levenspiel, 1962)

which combined with Equation 11 gives

The main trouble with the computation of X arises from Equation 9, because of the nonvalidity of the rate equation (Equation 8) in the proximity of x = 0. This trouble was bypassed in computations by neglecting the time required to go from x = 0 to x = 0.004. With this approach, Equations 9 and 8 can be combined to give

Equation 14 was proved to reproduce the experimental data with reasonable accuracy. X was computed as follows: First, values of ( L i t ) were computed for x , = 0.004, x, = 0.006, x , = 0.008, . . . , x, = 0.998 by integration of Equation 14 by Simpson’s rule with increments ( x , ) = 0.001; and Equation 13 was integrated by the trapezoidal rule with a variable increment of ( L i t ) by using the sets of (LE/ 7 ) us. x, previously obtained. All the calculations were performed on an IBM 360 digital computer, and the results of exit conversion us. the dimensionless parameter ( h t ) are presented in Figure 5 for different values of ideal backmix stages. This figure shows the strong negative effect of backmixing on reactor performance (when the number of backmix stages, j , decreases) a t high conversion levels, as required to avoid large iodine losses in the process of copper recovery. So, for example, for a conversion level of 99S, the values of parameter ( h t ) are about 210, 30, and 3.6 for 1, 2, and m backmix stages. At 350°C. ( h = 0.0653 min.-’) those values correspond to residence times of 3220, 460, and 55 minutes, respectively. The convenience of increasing the number of ideal backmix stages or approaching plug flow of solids is evident and should be taken into account for proper reactor design. Conclusions

Under the conditions of reaction diffusional resistances appear t o be unimportant. Computations made on the basis of the kinetic study results lead to the conclusions that backmix of solids must be reduced to a minimum in designing the industrial scale reactor for this reaction.

Figure 5. Relation of exit conversion to

(kr)

Nomenclature

b = reaction order of cuprous iodide c = reaction order of cupric oxide E = activation energy, cal./g.-mole E(tJ = residence time distribution function j = number of backmix stages K = equilibrium constant k = reaction rate constant, min.-’ k , = frequency factor, min.-’ X2

k# = constant in Arrhenius plot, k o / L d x / f ( x ) 1

M f = final mass of sample, g. M , = mass of sample a t reading i , g. M + l = mass of sample at reading i + 1, g. M , = initial mass of sample, g.

& = air flow, litersimin.

R = 1.987 cal./g.-mole K. O

r = reaction rate, moles CUI converted/ initial moles

CuI/min.

T = absolute temperature, K. - time, min.

t, = individual residence time of a particle for conversion, x,,min. t = mean residence time, min.

x = average conversion for each time interval, moles

CUI converted/initial moles CUI x, = conversion of a particle a t any time, moles CUI convertediinitial moles CUI x = average exit conversion of backmix reactor,

moles CUI converted/initial moles CUI Literature Cited

Bond, W. D., Clark, W. E., “Reduction of Cupric Oxide with Hydrogen,” U. S. Atomic Energy Commission, Oak Ridge National Laboratory, Rept. ORNL-2815 (1960). Levenspiel, O., “Chemical Reaction Engineering,” pp. 282, 364, Wiley, New York, London, 1962. Pavlyuchenko, M. M., Rubinchik, Y. S., J . A p p l . Chem. USSR 24, 751 (1951).

Acknowledgment

The author is grateful to Jaime A. Maymb for aid in the elaboration of this work. 288

Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 2, 1970

RECEIVED for review May 19, 1969 ACCEPTED September 11, 1969