INDUSTRIAL AND ENGINEERING CHEMISTRY
1040 Log P =
E*
- AH logp f c
LITERATURE CITED
Amerongen, G. J. Van, J . Appl. Phys., 17, 972 (1946). Amerongen, G. J. Van, J . Polymer Sci., 2, 381 (1947). Zbid., 5, 307 (1950). Barrer, R. M., “Diffusion in and through Solids,” Cambridge University Press, London, 1951. Barrer. R. M.. Kolloid 2..120. 177 (1951). Barrer; R. M.; Trans. Faiaday S O C . , 628, ’ ~ ~644 , (1939). Zbid., 36, 644 (1940). Barrer, R. M., and Skirrow, G., J . Polymer Sei., 3, 549, 564 (1948). Boer, J. H. de, and Fast, J. D., Rec. trav. chim. pays. bus., 57,317 (1938). Bradley, R. S., J . Chem. Soc. (London), 1934, p. 1910. Braune, H., 2.p h y s i k . Chem.;llO, 147 (1924). Brubaker, D. W., and Kammermeyer, K., Anal. Chem., 25, 424 (1953). Brubaker, D. W., and Kammermeyer, K., IND.ENQ. CHEM., 44, 1465 (1952). Zbid., 45, 1148 (1953). Zbid., 46, 733 (1954). Doty, P. M., Aiken, W. H., and Mark, H., I b i d . , 38, 788 (1946). Doty, P. M., Aiken, W. H., and Mark, H., IND.ENQ.CREM., ANAL.ED.,16, 686 (1944). Friedman, I. R., thesis, Dept. Chem. Eng., Polytechnic Institute of Brooklyn, 1952. Glasstone, S., Laidler, K. J., and Eyring, H., “Theory of Rate Processes,” McGraw-Hill Book Co., New York, 1941. Jost, W., “Diffusion in Solids, Liquids, and Gases,” Academic Press, New York, 1952. Osburn, J. O., and Kammermeyer, K., Ibid., 46, 739 (1954). Othmer, D. F., IND. ENQ.CHEM.,32, 841 (1940). Othmer, D. F., and Thakar, M. S., Zbid., 44, 1654 (1952). I b i d . , 45, 589 (1953). Othmer, D. F., and White, R. E., I b i d . , 34, 952 (1942) Rabinowitch, E., Trans. Faraday Soc., 33, 1225 (1937). Weller, S., and Steiner, W. A,, J . A p p l . Phys., 21, 279 (1950). West, C. J., Kunz, W. B., and Sears, G. B., “Permeability of Organic Materials to Gases,” Series 169, Parts I & 11,Institute of Paper Chemistry, Appleton, Wis., 1948.
Equation 10 indicates that if the permeability constant is plotted on logarithmic paper against the vapor pressure of a reference liquid, always taken at the same temperature, a straight line results, the slope of which is given by the value
E*
-
AH
L
This has been demonstrated graphically in Figures 1 and 2 and by numerous other plots which are not given here. I n each case the slopes of the lines are represented by this value, which is shown to be constant throughout the entire temperature range, since the lines are straight and hence the slopes are constant. It is thus apparent, since the values of L, the molar latent heat of the reference substance, are known as a function of temperature, that if either E* or A H is known, the other may be calculated from the slopes of these lines at the particular temperature. NOMENCLATURE
Gas volumes are measured at standard pressure and temperature.
C
D Do
= constant of integration =
= E = E* = e = AH =
K
=
P
=
Po
=
p
=
R T
=
KO = L =
=
Vol. 41, No. 5
diffusion constant, sq. cm. per second diffusion constant independent of temperature, same as D activation energy of permeation, calories per mole activation energy of diffusion, calories per mole base of natural logarithm heat of solution, calories per mole solubility constant cc./sq. cm./cm. Hg solubility constant independent of temperature, same as K heat of vaporization of reference substance, calories per mole permeability constant, cc. cm./second X sq. cm. X cm. Hg permeability constant independent of temperature, same as P vapor pressure of reference substance, mm. of Hg or atmospheres gas constant, cal./moleoo K. absolute temperature, K.
RECEIVED for review August 5, 1954. A C C ~ P T I DNovember 13, 1954. Previous articles of this series have appeared in INDUSTRIAL AND ENQINEIRINQ CHEMISTRY during 1940,1942,1943, 1944,1945,1946,1948,1949,1960, 1951,and 1953; Chem. & Met. Eng., 1940; Chimie et Industris (Paris),1948; Euclides (Madrid), 1948; Sugar, 1948; Petroleum Refiner, 1951, 1952, and 1953; Proc. 111 World Pet. Conoreas (Leiden), 1951; Prop. X I International Congress Pure and Applied Chemistry (London), 1948.
Ignition Limits of Hydrogen Peroxide Vapor CHARLES N. SATTERFIELD, PETER J. CECCOTTI, H. R. FELDBRUGGE
AND ALONZO
Massachusetts Institute of Technology, Cambridge, Mass.
DATA ON IGNITION LIMITS
0
0
0
of mixtures of hydrogen peroxide vapor and inert diluent gases are of practical importance wherever appreciable concentrations of hydrogen peroxide are encountered
A
N EARLIER paper ( 4 ) described studies of the explosive
characteristics of hydrogen peroxide vapor, including measurements of the ignition limits at atmospheric and subatmospheric pressures for systems in which the hydrogen peroxide vapor was diluted with water vapor or mixtures of oxygen and water vapor. The principal results were as follows: At atmospheric pressure the ignition limit occurred at 26.0 mole % hydrogen peroxide, when either a hot wire or a spark gap initiator was used, and this limiting composition was unchanged when the ratio of water vapor t o oxygen in the diluent gas was varied over a considerable range, and the total pressure was kept equal t o 1 atmosphere. The ignition limit was found t o increase to 33 mole % ’ hydrogen peroxide at 200 mm. of mercury total pressure, and t o 55 mole % hydrogen peroxide at 40 mm. It waB also found that catalytic surfaces such as copper, silver, and platinum, when introduced a t room temperature into vapors of
May 1955
INDUSTRIAL AND ENGINEERING CHEMISTRY
explosive composition, will usually initiate an explosion, while clean, relatively noncatalytic materials, such as aluminum, will initiate explosions a t somewhat elevated temperatures. In the investigation reported here, the effects of various diluent gases on the ignition limit were studied (at 200 mm. of mercury total pressure), as well as the effect on the ignition limit a t subatmospheric pressures of packing the explosion bulb with Raschig rings. As before, the term “ignition limit” is defined as the minimum concentration of hydrogen peroxide in a vapor mixture under specified conditions, below which powerful external ignition cannot propagate an explosion.
1041
explosive and nonexplosive composition region. If the hydrogen peroxide concentration is slightly above the ignition limit, an audible pop is heard. As the concentration is gradually reduced, the phenomenon becomes attenuated and explosion is said to occur if the vapor cloud hovering in the condenser is expelled suddenly on closing the spark in the explosion bulb, even if there is no audible report. This is the same criterion of explosion limit as used previously and has been found to be highly reproducible.
EXPERIMENTAL
The apparatus was that previously used for ignition limit studies a t subatmospheric pressure ( 4 ) . It consisted essentially of a hydrogen peroxide boiler, from which vapors passed continuously to the explosion bulb comprising a round 500-ml. flask fitted with a spark gap ignition source. Vapors leaving the explosion bulb passed through a condenser, and then noncondensable gases were removed through a vacuum system. The apparatus was designed so that li uid condensate samples could be removed without ‘disturbing %e steady-state operation. The apparatus included a vacuum pump, manostat, surge tank, and leveling device to control the rate of flow of hydrogen peroxide solution to the boiler. The pressure in the explosion bulb could be varied from atmospheric down to about 30 mm. of mercury. As before, diluent gas was introduced into the hydrogen peroxide boiler through a fritted-glass plate inserted in the wall of the boiler a t the base of the heated arm. I n studies of the effect of diluent gases, aluminum electrodes were used for the 5-mm. continuous spark gap in the explosion bulb, with voltage supplied from a Ford spark coil; the corresponding power dissipation was about 6 watts. When the explosion bulb was packed with Rashig rings, i t was necessary to reduce the spark gap to 3 mm. to insert it into the packing. About 9 watts of power were dissipated under these conditions. The operation of the leveling device and boiler has been described. A t steady state the boiler delivered continuously to the explosion bulb a homogeneous vapor of constant composition and a t a constant rate. By careful cleaning of all surfaces, the amount of hydrogen peroxide decomposition occurring on boiling could be reduced to a negligible amount. I n the studies on diluent gases other than oxygen, the oxygen concentration in the gas and vapor mixture was never more than 0.4% and usually 0.2% or less. Thirty minutes to 1 hour were allowed for the attainment of steady state, which was indicated by a constant boiling rate and constant com2osition of samples of the condensate. The composition of the vapor in the explosion bulb was calculated from the known rate a t which the diluent gas was metered into the hydrogen peroxide boiler and the measured composition and rate of condensation of the liquid from the condenser. A slight correction was applied to the vapor composition to allow for the amounts of water vapor and oxygen produced by decomposition occurring in the boiler, the amount of which was calculated from the difference between the hydrogen peroxide concentration fed to the boiler and that collected from the condenser. N o correction was made for water vapor carried off uncondensed in the diluent gas from the condenser exit; i t was calculated that neglecting this factor did not affect the reported ignition limit by more than about 0.201, a t the most. Hydrogen peroxide concentrations were determined by volumetric analysis with potassium permanganate, using the recommended procedure of Huckaba and Keyes ( 2 ) . The diluent gas flow rate was measured by an orifice calibrated for each gas against a gas wet-test meter. The rate of boiling varied between 3.5 and 10 grams of hydrogen peroxide solution per minute. After steady state was reached, attempts a t initiating an explosion were made by closing the continuous spark gap. Five attempts were made, spaced a t intervals of 4 minutes each to enable the system to recover from disturbances caused by sparking or explosion. There is no sharp boundary between the
N 0
2 5 % H202 35 50% INERT G A S
55
45 MOLE
65
PERCENT H1 0
2 5 % He0
25%H20,
0% INERT GPS 15% H z O
I N E R T GAS
RESULT
ON SPARKING
Op Np He 0 0
6 p d p
A p
F l V E EXPLOSIONS ON FIVE A T T E M P T S NO EXPLOSIONS
ON F I V E A T T E M P T S
ONE TO FOUR EXPLOSIONS
ON F I V E
The results are shown in Figures 1, 2, and 3. Each group of five attempts a t initiating explosion a t a given vapor composition is represented by a single point. The various vapor compositions studied may be divided into three categorics, represented by three kinds of points:
1. No explosion obtained in five attempts. 2. One or more, but less than five, explosions obtained in five attempts . 3. Five explosions obtained in five attempts. I n some cases the explosion was so violent on the first, second, or third attempt that all the liquid was blown out of the boiler. Under these conditions no further attempts a t explosion were made, and the composition was located in the explosive region. I n the experiments with the packed explosion bulb, advantage was taken of the fact that the water vapor-oxygen ratio in the diluent gas had been found to have no effect on the explosion limits a t 760 and 200 mm. of mercury and it was assumed that there was likewise no effect of moderate oxygen proportions a t pressures down to 50 mm. of mercury. Therefore in these studies small amounts of oxygen gas were fed continuously to the boiler as a convenient method of varying the proportion of hydrogen peroxide in the total vapor mixture within the explosion bulb. This enabled the operator by suitable choice of composition of the starting mixture, rates of boiling, and rates of oxygen input, to obtain points below, on, and above the ignition limit during the course of a single run. The explosion bulb was completely packed with a packing consisting of borosilicate glass Raschig rings, 14 mm. in outside
INDUSTRIAL AND ENGINEERING CHEMISTRY
1042
diameter and 14 mm. in length and fire-polished on both ends. The spark gap was placed in the approximate center of the bulb, closely surrounded by pieces of the packing. The effect of noncondensable diluents on the ignition limit was studied a t a total pressure of 200 mm. of mercury. At this pressure the ignition limit exists a t a fairly low mole fraction of hydrogen peroxide, thus allowing study of substantial noncondensable gas fractions. Moreover, the ignition limit is easier t o determine and more reproducible than a t substantially lower pressures. On the other hand, the pressure is sufficiently low that
0 % Go, 30% H20 NLARGED SECTION GO* 10
40
eo
80 U*O
Vol. 41, No. 5
concentrations of a few mole per cent of the total vapor composition. At higher concentrations, however, carbon dioxide dampens the propagation of the explosion and the effect becomes progressively greater as the carbon dioxide concentration is increased. Effect of Packing Explosion Bulb. The ignition limit was determined a t four pressures: 200, 150, 100, and 50 mm. of mercury. At each pressure from 15 t o 20 runs were made, each consisting usually of a series of five attempts a t one vapor composition, as above. The ignition limits were determined t o be, respectively, 40.5, 42.9, 48.6, and 59.8 mole % hydrogen peroxide. I n these studies, from 10 to 20 mole yo of oxygen was added t o the gas mixture as a control technique, as described above. The remainder was water vapor and hydrogen peroxide. During the couise of the work a t 200 mm. of mercury the explosion bulb was repacked to eliminate the effects of a particular orientation; this had no effect on the results. The results are shown on Figure 3 together with, for purposes of comparison, the ignition Iimit curve previously determined as a function of pressure in an unpacked bulb. As expected, packing the bulb shifts the ignition,limit to higher hydrogen peroxide concentrations, but the shift is not great. DISCUSSION
The data in general are characterized by a high degree of internal consistency. There is somewhat more spread in the data a t 50 and 100 mm. of mercury than a t the higher pressures, because the ignition limit is more difficult to determine a t these low pressures. 5 0 % GO, 30% HzO
Figure 2.
MOLE PERCENT H20
0 9. GO2 SO%H20
Ignition limit i n hydrogen peroxide-water vapor-carbon dioxide system
F I V E EXPLOSIONS ON F I V E b T T E M P T S
0
0 NO EXPLOSIONS O N F l V E A T T E M P T S
0 Q A
Five explosions on five attempts No explosion on five attempts One to four explosions on five attempts Total pressure, 200 mm. of mercury
d ONE
a 0 3
e 500 -
I D A T A P O I N T S REfER
I
the violence of the explosion is decreased to a point where it does not represent a major hazard to equipment and to operators. I n previous studies at atmospheric pressure a number of explosions were sufficiently violent to shatter a part of the apparatus. Pressures of 200 mm. and below were studied in the packed explosion bulb experiments for the reasons enumerated above, but also because of the interest in conjunction with distilling processes for concentrating hydrogen peroxide, which operate under vacuum.
T O STUblE5
IN
PACKED V E S S E L )
E'
I G N I T I O N LIMIT. UNPACKED VESSEL
.: 4 0 0 a
0 3
300
--
a
IGNITION LIMIT, PACKED V E S S E L
d
5 200 t-
100
-
RESULTS
The results of the studies on the effect of diluent gases are shown in Figures 1 and 2. Each of the large triangular plots represents only a portion of the total possible range of composition, The maximum diluent gas compositions studied were limited by the difficulties of working with the very highly concentrated hydrogen peroxide solutions required. Effect of Oxygen, Nitrogen, and Helium. It was found previously (4) that, a t atmospheric pressure, the ignition limit of 26.0 mole yo hydrogen peroxide was not changed by varying the water vapor-oxygen ratio in the diluent gas over a considerable range. The present studies a t 200 mm. of mercury are shown in Figure 1. The ignition limit a t this total pressure is 32.5 mole % hydrogen peroxide and there is again no measurable effect of oxygen on the limit when the oxygen concentration is varied from 0.0 to 39.0%. The same results were obtained with helium and nitrogen. As shown in Figure 2, the effect of carbon dioxide is unique among the gases studied. At 200 mm. of mercury it has no effect upon the ignition limit of hydrogen peroxide when it is present in
T O F O U R E X P L O S I O N S ON FIVE
ATTEMPTS
I 25 H20,
I
I
I
I
I
I
I
50 55 60 65 70 C O N C E N T R A T I O N , M O L E P E R C E N T ( R E M A I N D E R I S H Z O AND 02)
30
35
'40
45
Figure 3. Effect.of pressure on ignition limit in hydrogen peroxide-water vapor-oxygen system Curve for unpacked vessel from ( 4 )
Interpretation of the results of the effect of the various diluents on the ignition limit can only be speculative a t present. It was indicated previously that the most probable mechanism for the decomposition of hydrogen peroxide involved straight chains initiated by hydroxyl radicals
HzOz
OH HOZ
+ HzOz
+ HzOz
+
(1)
20H HOz
+ +
OH
+ HzO
+ H20
$-
0 1
with appropriate chain-stopping steps, This hypothesis was based upon the energy changes associated with various possible
*
.
INDUSTRIAL AND ENGINEERING CHEMISTRY
May 1955
Table I. Calculated Adiabatic Reaction Temperatures of Limiting Compositions
HzOz,
mole % ’ 32.5
37.0 42.5 26.0
Species HzO, mole % 67.5 27.5 45.5 35.0 74.0
Other, mole Yo n 4 0 . 6 0% 4 0 . 0 Nz 4 0 . 0 He 1 7 . 5 COz 2 2 , 5 Cor 0
Total Pressure Mm. ~ g i 200 200
200 200 200 200 760
Reaction Temp., O
c.
880
910 925 1020 930 1030
7 80
reaction steps, plus the study of Urey, Dawsey, and Rice ( 5 ) on the emission spectrum of hydrogen peroxide decomposed in a discharge tube. None of the diluent gases would be expected to be a direct participant in the decomposition of hydrogen peroxide except as a third body and this expectation is borne out by the nature of their effects. The role of the inert gas may be through its heat capacity, which affects the adiabatic reaction temperature; through its thermal conductivity, which affects the rate of heat dissipation from the reaction zone; or in its effect on the rate a t which free radicals formed in the reaction can dissipate by molecular diffusion. I n fuel-oxidant limiting mixtures, the heat capacity effect of the diluent appears to be generally the most important, although the other effects apparently enter in varying degree in different cases ( I , 3 ) . The adiabatic reaction temperatures attained by various limiting compositions here are listed in Table I. I n each case the initial temperature was taken t o be that corresponding to equilibrium with the liquid and gas phases in the boiler. Perhaps the most interesting results are that the limiting mixtures containing the highest fractions of helium or carbon dioxide I
1043
have reaction temperatures substantially higher than the others. As helium greatly increases both the rate of radical diffusion and thermal conductivity of the gas mixture, relative t o that expected with the other diluents, it is plausible that a higher reaction temperature is required t o propagate the explosion. The case of carbon dioxide is not so readily interpreted, although the results obtained here are similar to those obtained with some fueloxidant systems. Carbon dioxide usually has a greater dampening effect than nitrogen or helium, and this effect is sometimes greater than that attributable t o its heat capacity alone, as here. The reaction temperatures of the limiting composition mixtures are much lower here than those encountered in most fueloxidant systems. Indeed, of a large group of lower-limit mixtures listed by Egerton ( I ) , only those of hydrogen and carbon disulfide had flame temperatures below the 780” C. calculated for the limiting composition of hydrogen peroxide at 1 atmosphere pressure LITERATURE CITED
(1) Egerton, A. C., “Fourth (International) Symposium on Combustion,” p. 4, Williams Br. Wilkins, Baltimore, 1953. ( 2 ) Huckaba, C. E., and Keyes, F. G., J . Am. Chem. Soc., 70, 2578 (1948). (3) Mellish, C . E., and Linnett J. W., “Fourth (International) Symposium on Combustion,” p. 407, Williams & Wilkins, Baltimore, 1953. (4) Satterfield, C. N., Xavanagh, G. M., and Resnick, H., IND. ENG. CHEM., 43,2507 (1951). ( 5 ) Urey, H. C., Dawsey, L. H., and Rice, F. O., J . Am. Chem. SOC., 51, 1371 (1929). RECEIVED for review October 26, 1954. ACCEPTED December 23, 1954. Based in part on the M.S. thesis by A. H. R . Feldbrugge submitted to the Department of Chemical Engineering, Massachusetts Institute of Technology, 1951. A few of the data presented were obtained b y J. A. Pitoock. Financial support was received from the Office of Naval Research under Contract N5ori-07819, NR-092-008
General Density Equation for Glyceridic Oil-Solvent Mixtures CALCULATION OF DENSITY -COMPOSITION-TEMPERATURE DATA FROM OIL AND SOLVENT DENSITIES EVALD L. SIUU, FRANK C. MAGNE, ROBERT R. MOD,
AND
RENE L. DURRl
Southern Regional Research Laboratory, New Orleans, La.
DEAS
ITY-composition-temperature data for binary mixtures of glyceridic oils with various solvents are highly important in development and control of commercial processes such as solvent extraction, solvent refining, and solvent crystallization. Such data are required by the engineer for accurate calculations of heat transfer and fluid flow in connection with plant design. They also afford a basis for rapid determination of miscella concentrations a t the various stages of oilseed and oil processing. I n reporting the results of such density measurements in previous publications (2-4) no suitable method was found for accurate presentation of the data in a readily usable form. Adequate graphical or tabular representation in the form of density-composition isotherms is not feasible because of the size of the charts or the length of the tables required. The method finally resorted to involved fitting the data to an empirical four-constant cubic equation (3, 6) and tabulating the constants for the iso1
Present address, 3033 Forty-First St., Kew Orleans, La.
therms for each oil-solvent system a t temperature intervals of 10” C. Tedious calculations and graphical interpolations are then required t o construct graphs or tables of sufficiently complete data in a form for convenient use. Because the empirical equation is of the third order, the calculation of the density a t a given composition is involved and the reverse calculation of the composition of a mixture from its density is extremely intricate. All these methods of presentation have two additional disadvantages. I n the first place, they all give data which apply accurately only to the particular specimens of oil and solvent. Actually the density of a given oil may vary considerably, depending upon its source and its stage of refinement and the same is true for a given solvent, especially a mixed solvent such as commercial hexane. An approximate correction can be applied for such variations ( 2 , S ) , but this involves further calculations. Another disadvantage of the above methods is that interconversion between c.g.s. and English units and between Centigrade and Fahrenheit degrees requires considerable calculation and replotting.
