OXYGEN GENERATION BY THE REACTION OF POTASSIUM SUPEROXIDE AND SEA WATER J .
1.
M A N G A N A R O '
Research and Development Center, General Electric Co. Schenectady, N . ~
k'.
The reaction of potassium superoxide with synthetic sea water yields oxygen which is evolved a t a relatively rapid rate and hydrogen peroxide which may subsequently decompose into water and additional oxygen. The kinetics of the rapid oxygen evolution were studied. A rate constant (in cubic centimeters per second per square inch) was defined and determined for several temperatures and KOH concentrations. Temperatures considered ranged between 10' and 24OC. Hydroxide concentrations ranged between ON and 3 N KOH. Addition of FeC13 to the reaction solution enhanced the decomposition rate of hydrogen peroxide sevenfold. Oxygen generated by this method was reasonably pure.
THEreaction of potassium superoxide (KOz) with water has received increasing attention recently (Mel'nikvov and Firsova, 1962; Petrocelli and Capotosto, 1964; Petrocelli and Krausa, 1963; Vannenberg, 1962; Vol'nov, 1964) owing to its applicability as an oxygen source aboard space vehicles. Russian chemistry has been particularly active in this field (Petrocelli, 1965). T h e literature has primarily treated the reaction of potassium superoxide with water vapor, while comparatively little has been reported of its reaction with liquid water. The ease with which potassium reacts with liquid water distinguishes it as a convenient source of oxygen for marine application such as sonobuoys and artificial atmospheres. This investigation was undertaken to obtain design data for the reaction of potassium superoxide with liquid water under a variety of conditions. The three main items considered were: reaction scheme, reaction kinetics, and purity of the oxygen generated. The effects of solution temperature and alkalinity on the kinetics of the oxygen evolution reaction were studied. Solution alkalinity is of interest in applications which involve a confined region such as a sonobuoy, where a local increase in alkalinity can be expected. The effect of distilled water on reaction kinetics compared to synthetic sea water was also considered. Experimental
The apparatus used to obtain kinetic data is diagramatically illustrated in Figure 1. T h e apparatus consisted of a 1000-cc. round-bottomed flask in which the oxygen was generated. A squat cylindrical pellet of potassium superoxide ( % inch in diameter by h inch
' Present address, Chemical Engineering Department, Manhattan College, Bronx, N . Y. 10471 Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 1, January 1970
thick) was dropped into 700 cc. of aqueous solution in the flask by means of an attached rotating elbow. The pellet was formed by pressing 1 gram of KO2 powder in a die a t 16,300 p s i . for 1 minute. The round-bottomed flask was connected by a 4-inch length of Tygon tubing to an inverted graduated cylinder where the oxygen was collected. Four electrical probes projected into the graduated cylinder a t three measured volume intervals of 36.7 cc. each. The graduated cylinder was provided with a valved side stem a t the top, used to draw the desired amount of aqueous saline solution into the cylinder. The saline solution functioned as an electrical conductor in the circuit shown in Figure 1. Evolved oxygen displaced the saline solution from the cylinder, causing the solution meniscus successively to pass and break contact with the four probes. An oscilloscope monitoring these electrical events produced a three-step pattern. The length of each step of the trace was proportional to the length of time required to evolve the respective 36.7 cc. of oxygen. A constant current power supply was the electrical source. Thus, the raw data for determining the reaction kinetics were the three individual time periods required to evolve oxygen for the three 36.7-cc. volume intervals. Temperature control was provided for the generating flask. Three temperature levels were studied: 24", 20", and 15"C. However, one experiment was made at 10"C. using 1N KOH. Three concentration levels of KOH in the synthetic sea water were considered: ON, lN, and 3N. When KOH or the superoxide was added to the sea water solution, several hydroxides were precipitated. An experiment was also made with 5N KOH in distilled water a t 24°C. The source of potassium superoxide was the Callery Chemical Co., Callery, Pa. A synthetic sea water with the brand name Sea Salt was used. 1
THE
++LJ COOLING SYSTEM
Figure 1 . Diagram &apparatus
The Reaction
A number of reaction schemes have been proposed. The first is the “direct formation” of oxygen with the absence of “delayed generation”:
2K02
+ H20 = 2KOH + 3/2Oz
(1)
Another scheme predicts direct production of oxygen along with the hydrogen peroxide anion which subsequently decomposes, releasing 0 2 :
2 K 0 2 + H 2 0 = 2K+ + 2HO;
+ %OZ
Table 1. Volume of Oxygen Immediately Evolved per Gram of Potassium Superoxide
Temp.,
c.
24 20 15 10
ON KOH,
cc .
136 135 134
...
1N KOH,
cc .
144 146 138 126
3N KOH,
cc.
137 137 137
...
5N KOH,
Distd.
CC.
HzO, CC.
137
124
... ... ...
... ...
...
(2a)
followed by
2H02 = 20H-
+0
(2b)
2
The first of these reactions (Equation 2a) yields oxygen a t a relatively rapid rate; the second (Equation 2b), a t an appreciably slower rate. Oxygen which is rapidly generated is referred to as “immediately available”; oxygen produced by an inherently slower process, such as Equation 2b, is called “delayed oxygen.” In actual practice, little difficulty was experienced in distinguishing between immediate and delayed processes. Thus, it was clear when the immediate process concluded and the delayed continued. Still another reaction scheme, considered by George (1955), was the formation of H 2 0 2and immediate oxygen evolution according to
2 K 0 2 + 2Hz0 = 2KOK
+ H ~ O+Z 0
2
(3a)
followed by the slower reaction,
Hz02 = H20
+
ZO2
(3b)
From measurement of the immediate and delayed oxygen evolved per gram of KO, for given conditions, the predominant reaction may be deduced. If Reaction 1 is the chief reaction, a simple calculation gives the amount of O2 which would be immediately evolved (all volumes a t STP unless otherwise stated): Immediate oxygen = 236 cc. per gram of KO, Delayed oxygen = 0 cc. per gram of KO, For Reactions 2a and 2b: Immediate oxygen = 78.7 cc. per gram of KO, Delayed oxygen = 157.3 cc. per gram of KOs and for Reactions 3a and 3b: 2
Immediate oxygen = 157.3 cc. per gram of KO2 Delayed oxygen = 78.7 cc. per gram of KO2 Table I summarizes the data obtained for several temperatures and initial KOH concentrations in synthetic sea water. All columns except the first indicate volumes of oxygen generated (corrected to STP) a t the specified temperature and initial KOH concentration. These experiments were performed with an initial solution of 700 cc. The data show an immediate oxygen volume ranging between 124 and 146 cc. with a meap of about 137 cc. This suggests that Reactions 3a and 3b may be predominant, with some contribution from Reactions 2a and 2b. A further experiment was performed in which potassium superoxide was placed in 700 cc. of 1N KOH and synthetic sea water at 24°C. and 3.5 grams of FeC13 added to catalyze the peroxide decomposition. The following results were obtained: Immediate oxygen = 134 cc. per gram of KO, Delayed oxygen = 74 cc. per gram of KO2 The delayed oxygen in this case was evolved over a period of 3 !4 hours. Comparing these figures with theoretical calculations based on Reactions 3a and 3b: %;o of theoretical immediate oxygen = 85% % of theoretical delayed oxygen = 95%
The over-all oxygen actually claimed was 88.2%. The less than 100% recovery might be accounted for on the basis of inadequate reaction time for the delayed reaction and impurities in the sample, such as KOH, KZCO?, and Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 1, January 1970
K L O L . Since the delayed oxygen constitutes the smaller portion of the oxygen evolved and its recovery is fairly good, impurities in the superoxide sample are believed to be responsible for the low over-all recovery.
20
L
Kinetics
Attention is now turned to the kinetics of immediate oxygen evolution. The kinetics of oxygen evolution by the delayed reaction is not considered in this paper. A knowledge of the surface area upon which the reaction occurs is necessary. T o this end, 1 gram of KOz powder was pressed a t 16,300 p s i . for 1 minute into a cylindrical pellet j, inch in diameter and '/& inch thick, giving a known initial area. The effect of surface area on reaction rate (uncorrected for reacting area) is well illustrated by comparing the reaction rates for the above described pellet with that for 1 gram of K O L powder. I n the former case, 120 cc. of oxygen are evolved in approximately 12 seconds; in the latter, 120 cc. are evolved in approximately 1 second. The density of the KO? pellet may be varied within limits by varying the amount of superoxide charged to the die and the pressure of compression. Experiments on two pellets of different densities (31.42 and 27.2 grams per cubic inch) indicate little effect of superoxide packing density on the rate of reaction, within the narrow range studied. However, greater variations in the packing density might be expected to have significant effect. A rate expression for the rapid oxygen evolution of the following form is assumed:
(4) The rate constant contains both the temperature and concentration dependence of the reaction. The total surface area of the cylindrical pellet is readily expressible as:
A = 2nr ( r + h )
(5)
where both r and h are functions of time. The radius and height may be related to the volume of O2 evolved after time t by the following
V ( t ) = bpn (dho - r'h)
(6)
The constant b was taken to have the value of 137 cc. per gram of KOJ. If the reductions in radius and height are assumed linear with time, there result
r = ro - ct h = h, - et
(74 (7b)
where c = a constant, inchisecond. We may write from Equations 6, 7a, and 7b
V ( t ) = bpr[r;ho- ( r o- ct)'(ho- c t ) ]
(8)
which is a cubic equation to be solved for c. Integrating Expression 4 between t = 0 and t = t results in a value for k-Le.,
I
3.36
3.48 (\IT)
1
3.54
3.60
x io3
Figure 2. Reaction rate constant
2n - 3c (ro-
I
1
I
3.42
k vs. l/J
crot2 chote ro +-3c + ruhot- __2 - 2 ~
+
-1c2t3 3
which, if the assumptions are reasonably valid, should be a constant dependent only on temperature and concentration. Calculational Procedure. The raw data obtained were the time periods required to evolve 36.7, 73.4, and 110.1 cc. of 0 2 . By using the above data in Equation 8, c was determined. With the value of c and the raw data, k was computed from Equation 10. For a run a t a given temperature and concentration of KOH, the values of V and t introduced into Equation 10 should not appreciably affect k . This was found to be the case, indicating that the assumptions possess some degree of validity. Results
A plot of log k us. l / T a t various KOH concentration levels in synthetic sea water is given in Figure 2. The graphs are approximately linear, indicating an exponential dependence of l / T . The rate constant for the reaction in distilled water (at 24°C.) was also determined and found to be 21.3 cc. per second per sq. inch. A higher KOH concentration results in a lower rate constant, probably due to the decrease in mole fraction of water. Unsuccessful attempts were made to free k of its concentration dependence. This is desirable, as it will allow analytical expression of the oxygen evolution rate. I t is believed that the complexity of the water mole fraction dependence is due to the effects of diffusion and reaction molecularity. Comment on Delayed Reaction. For ON KOH in sea water a t 24" C. the following rates were found for pelletized KO?:
(9)
Immediate reaction = 3600 cc. per hour Delayed reaction = 3 cc. per hour (for a 3.5-hour period).
Combining Equations 5 and 7 and performing the indicated integration of 9 gives an expression for k
Therefore, the immediate is about 1200 times faster than the delayed reaction. George (1955) has reported the delayed as being a t least 1/600 that of the immediate. Although the speed of the immediate reaction is dependent
k=--
V(t) A(t)dt
Ind. Eng. Chem. Process Des. Develop., Vol. 9, No. 1, Jonuory 1970
3
200 220
180 -
n w
160-
3 p
140-
0"
120-
w
/
/
/
o*p-E-s
k
i ,
SEA WATER I N KOH and F e C I S
0 0 -1
a
c
E
10080-
60 40 -
2o
o
t 0
l
2
'
"
4
l
6
l
8
l
IO
l
'
"
'
20
40
12 sec(cl)
min.(m) 0
'
min.(A) 0
TIME Figure
-
60 20
80 40
100 120 140 160 180 200 220 240 260 60 80 I00 I20 140 160 180 200 220
3. Total oxygen evolved v s . time
on the surface area, this comparison gives an idea of the orders of magnitude involved. When a catalyst such as FeCL (3.5 grams) was added to 1N KOH in synthetic sea water (700 cc.) the rate of delayed evolution (considered over a period of 3.5 hours) was increased to 21.2 cc. per hour (a sevenfold increase). Figure 3 is a plot of total oxygen evolved as a function of time with and without FeC13 (3.5 grams per 700 cc. of solution). Both experiments were performed in 1 N KOH and synthetic sea water a t 24°C. The circled data points refer to the time scale in seconds, while the square and triangular points refer to the time scale in minutes starting a t the indicated position. The curve which is initially lower and then becomes higher was obtained with the addition of FeCII. The other curve was obtained for 1 N KOH and synthetic sea water alone. Oxygen Purity. The oxygen generated by this method might be used by a fuel cell to power the instrumentation on a sonobuoy. For such an application, it is important to know the purity of the oxygen generated. Oxygen was generated from KO2 by the Kipp principle and run into a clean metal bomb for ' 2 hour a t a rate of approximately 2 cc. per second. Two samples were taken: one from the reaction of superoxide with distilled water and the other with synthetic sea water. The results of the gas analysis are given in Table 11. No appreciable difference was found between the results for distilled and synthetic sea water. It has been suggested that the results for actual sea water may differ from those for synthetic sea water, since the former contains organic substances which might undergo oxidation during the reaction and assume a gaseous form. Conclusions
The main reactions are given by Equations 3a and 3h: fast evolution of Or followed by the slower decomposition of hydrogen peroxide (delayed reaction). 4
"
Table II. Results of Gas Analysis
Constituents N, A
co, co
'N D
=
Concentration, P P M Distilled S3nthetzc water sea uater 1400 40
ND" 100 ED" 15
1000 30 KD' 100 SD" 15
none detected, less than
The rate constant, k , for oxygen evolution decreases with increasing KOH concentration. The rate of the immediate reaction ranges from 17.1 cc./sec.-sq. inch in ON KOH (24°C.) to 8 cc./sec.-sq. inch in 3 N KOH a t 15"C., for the KOr pellets used. Reaction rate of the delayed O2 evolution without catalyst is on the order of l j l 2 0 0 that of the immediate reaction. The addition of a small amount of FeCli to synthetic sea water enhanced the rate of delayed oxygen evolution by sevenfold (at 24" C. and 1 N K O H ) . Samples of oxygen evolved by potassium superoxide reacting with distilled water and with synthetic sea water were reasonably pure. Nomenclature
A = pellet surface area a t time t , sq. inch b = volume of oxygen immediately evolved per gram of potassium superoxide, cc. (STP) e = speed a t which pellet radius and height decrease, inchisecond h = pellet height, inch k = rate constant for immediate oxygen evolution, cc./secondisq. inch r = pellet radius, inch t = time, seconds Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 1, January 1970
V(t) = total volume of oxygen evolved a t time t , cc. (STPI p
= packing density of potassium superoxide, gram/
cc. SUBSCRIPT
0 = initial value literature Cited
Petrocelli, A. W., Capotosto, A., Aerospace Med. 35, 440 (1964). Petrocelli, A. W., Krausa, D. L., J . Chem. Educ. 40, 146 (1963). Vannenberg, N.-G., “Progress in Inorgaqic Chemistry,” F. Cotton, ed., Interscience, New York, 1962. Vol’nov, I. I., “Peroxides, Superoxides and Ozonides of Alkali and Alkaline Earth Metals,” Nauka, MOSCOW, 1964 (in Russian).
George, P., J . Chem. SOC.70, 2367 (1955). Mel’nikvov, A. Kh., Firsova, T. P., Russ. J. Inorg. Chem. 7, 633 (1962). Petrocelli, A. W., Aerospace Med. 36, 1187 (1965).
RECEIVED for review March 28, 1966 RESUBMITTED August 7, 1969 ACCEPTED August 19, 1969
PERFORMANCE OF POROUS CELLULOSE ACETATE“ MEMBRANES
FOR THE REVERSE OSMOSIS SEPARATION OF MIXTURES OF ORGANIC LIQUIDS J.
K O P E C E K ’ A N D
s.
S O U R I R A J A N
Division of Chemistry, National Research Council of Canada, Ottawa, Canada The reverse osmosis technique is applicable for the separation of binary mixtures of alcohols and/or hydrocarbons, including azeotropic and isomeric mixtures. The pqrous structures of the cellulose acetate membranes used were affected by the composition of the feed solution in contact with them; hydrocarbon liquids tended to collapse their porous structures on continued contact. Hydrogen bonding and solubility parameter may offer valid criteria of preferential sorption for nonelectrolyte binary feed mixtures containing components whose solubilities are governed primarily by either polar or dispersion forces.
of work (Sourirajan, 1964) on the reverse osmosis separation of binary mixtures of some organic liquids using the Loeb-Sourirajan type porous cellulose acetate membranes illustrates the applicability of the technique for the separation of azeotropic mixtures, substances belonging to the same homologous series, isomeric substances, and mixtures of other organic liquids. EXTENSION
Experimental Details
Reagent grade organic compounds and porous cellulose acetate membranes (designated here as CA-NRC-18 type films) made in the laboratory, were used. The films were cast a t -10°C in accordance with the general method described earlier (Loeb and Sourirajan, 1963,1964; Sourirajar1 and Govindan, 1965) using the following composition (weight per cent) for the film casting solution: acetone 68.0, cellulose acetate (acetyl content = 39.8%) 17.0, water 13.5, and magnesium perchlorate 1.5. Membranes shrunk at different temperatures were used to give different levels
’ Present address, Institute of Macromolecular Chemistry, Prague, Czechoslovakia Ind. Eng. Chem. Process Des. Develop., Vol. 9,No. 1, January 1970
of membrane porosity and performance a t preset operating conditions. The experiments were carried out a t the laboratory temperature in the pressure range 250 to 1000 psig, using the reverse osmosis cell shown in Figures 1 and 2. The cell was a stainless steel pressure chamber consisting of two detachable parts. The film was mounted on a stainless steel porous plate embedded in the lower part of the cell through which the membrane-permeated liquid was withdrawn at atmospheric pressure. The upper part of the cell contained the feed solution under pressure in contact with the membrane. The two parts of the cell were clamped and sealed tight using rubber O-rings. Compressed nitrogen gas was used to pressurize the system. About 250 cc of feed solution were used each time. The feed solution was kept well stirred by means of a magnetic stirrer fitted in the cell about % inch above the membrane surface. The quantity of liquid removed by membrane permeation was small compared to the amount of feed solution in the pressure chamber. The compositions of the feed and the membrane-permeated 5