I
CHARLES N. SATTERFIELD and THEODORE W. STEIN' Department'of Chemical Engineering, Massachusetts Institute of Technology, Cambridge, Mass.
Decomposition of Hydrogen Peroxide Vapor on Relatively Inert Surfaces Studies with glass, polymer, and metal surfaces provide comparisons of their relative activities plus useful data on hydrogen peroxide vapor decomposition
THE
heterogeneous decomposition of hydrogen peroxide vapor has been the subject of several investigations, with data reported for partial pressures of hydrogen peroxide up to 1 or 2 cm. of mercury, at temperatures from room temperature up to 400' C., and on such surfaces as soft glass, borosilicate glass, quartz, and metallized glass. The work of previous investigators has been recently summarized (28), However, there have been inconsistencies in the data available, and the fundamental nature of the heterogeneous decomposition of hydrogen peroxide vapor has remained obscure. The purposes of this work were, first, to obtain a clearer picture of the important factors which affect the surface decomposition rate and, secondly, to obtain engineering information on materials that will cause the minimum decomposition on contact with hydrogen peroxide vapor. In the present work hydrogen peroxide vapor concentrations substantially greater than those used in previous work were studied, as well as a much greater variety of surfaces. Previous studies indicate that any
surface has an accelerating effect on hydrogen peroxide vapor decomposition. Most investigators have, reported that the reaction rate can be expressed by a first-order equation, although secondorder reactions, both retarded and unretarded by water, and zero-order reactions have also been suggested (72, 76, 78). The reaction is reportedly unretarded by additions of moderate quantities of air or carbon dioxide. The rate of reaction is much higher on metals or soft glass than on borosilicate glass or quartz, although it can be varied by one or more orders of magnitude on a borosilicate glass surface alone by changing its chemical and physical properties. The reported activation energies have ranged from 5 to 20 kcal. per gram-mole but most frequently have been in the vicinity of 10 kcal. per grammole. Homogeneous decomposition was insignificant under all conditions studied here; it does not become appreciable at these concentrations until temperatures of about 450' C. are reached.
Present address, Whiting Research Laboratories, Standard Oil Co. (Indiana), Whiting, hid.
The experimental apparatus consisted of a vaporization section and a decompOSitiOh section.
Experimental
I n the vaporization section a hydrogen peroxide-water solution was fed continuously to an electrically heated borosilicate glass boiler through a special leveling device described by Holmes (75) (insert, in diagram), which maintained a constant liquid level in the boiler. The' boiler pressure was regulated by an external helium system; the helium, being lighter than the hydrogen peroxide-water vapor, was prevented from entering the system by maintaining a visible helium-vapor interface in the reflux condenser atop the boiler. This system was capable of producing a steady flow of hydrogen peroxide vapor at a constant concentration. The hydrogen peroxide-water vapor mixture from the boiler first passed through a n entrainment separator and then through an electrically heated borosilicate glass preheater. A portion of the vapor at this point was withdrawn for analysis by condensing it at 5' C. and determining hydrogen peroxide content by potassium permanganate titration. The remainder of the vapor flowed through the reaction tube and was similarly condensed and analyzed. The vapor flow was divided between the upstream and downstream condensers by control of the pressure, VOL. 49, NQ 7
JULY 1957
1173
A
/
FEED
I
\
ORA I N
REFLUX CONDENSER INTERFACE
LI UM PP LY
LEVELLING PYREX BO1 L E R
DRAIN Hydrogen peroxide vaporization system
UPSTREAM THERMOCOUPLE
$,
STIRRER .,
TEMPERATURE
VAPOR FROM BOILER ORA1 N
WET
=c
PJ TO H E L I U M j S Y STEM f
M A N 0 METER
f
UPSTREAM SAMPLE
Decomposition apparatus shown, without insulation
1 174
INDUSTRIAL AND ENGINEERING CHEMISTRY 5
MANOMETER
TO ATMOSPHERE
DOWNSTREAM SAMPLE
HYDROGEN PEROXIDE DECOMPOSITION set by bubbling the oxygen leaving each condenser through a column of water before passing it through a wet-test meter. The collectors in which the liquid samples accumulated were so constructed that the samples could be removed without affecting the pressures in the condensers, or the flow division. The reaction tubes used in the study of glass surfaces were usually 1 inch in diameter and approximately 2 feet long, with ball and socket joints on the ends with which the tube was clamped into the system. When high polymer surfaces were to be studied, the sample was placed inside a glass tube. I n the study of metallic surfaces, a metal rod about 3 / ~ 2 inch in diameter and 6.5 inches long rested coaxially on small aluminum supports inside a borosilicate glass tube about 1 foot long. Each end of this tube had a diameter of 1 inch, but the central portion in which the metallic sample was placed had a diameter of 3/16 inch. Each reaction tube was immersed in a constant temperature bath to which it was sealed by stuffing boxes, with asbestos tape as the gasket material. At temperatures below 180' C. Nujol, a mineral oil, was used in the bath; above 180' C., a molten salt mixture containing 7 weight % sodium nitrate, 40 weight yo sodium nitrite, and 53 weight yo potassium nitrate, was used. The bath was agitated by a stirrer, and its temperature was regulated to within 1 2 ' C. by a thermoswitch which controlled a heater lying along the length of the bath. The temperatures were measured with iron-constantan thermocouples at three locations-in a well at the exit of the preheater and in the two ends of the constant temperature bath. The range of experimental variables studied was: partial pressure of hydrogen peroxide, 0.02 to 0.23 atm.; temperature, 110' to 250' C.; total pressure, 1 atm. The data obtained in each set of experimental conditions were the amount and composition of the hvdrogen peroxide-water liquid sample collected from each condenser, volume of oxygen leaving each condenser, upstream and downstream pressures, temperature of vapor leaving the preheater, and temperature of the bath. At each combination of experimental conditions, three sets of data were collected to avoid fortuitous errors. The inlet and outlet concentrations were not greatly different, so that the reaction tube may in general be regarded as a "differential reactor." The data reported were therefore calculated as the difference between the rate at which hydrogen peroxide enters and leaves the tube, divided by the total surface area. The hvdrogen
peroxide concentrations reported are the geometric mean of the inlet and outlet partial pressures. Two types of surfaces. borosilicate glass coated with boric oxide and 2s aluminum, were studied over a wide range of temperatures and compositions. Other surfaces were studied less extensively and frequently at only one temperature. The rates of heterogeneous reaction were sufficiently low to permit neglect of the rate of diffusion of hydrogen peroxide from the bulk vapor to the surface as a limiting factor. For example, the difference between the interfacial and bulk partial pressure was estimated, by the correlation for laminar flow in a wetted wall column (29), to be about 15yo in the studies with the most reactive glass tube surface (a borosilicate glass tube on which silver was ad,sorbed) when the average bulk partial pressure of hydrogen peroxide was 0.02 atm. O n a boric oxide-coated tube a t 180' C. the difference increased from 2% at 0.03 atm. of hydrogen peroxide partial pressure to 11% a t a partial pressure of 0.2 atm. Similarly on an aluminum surface, it increased from 3y0 at a bulk hydrogen peroxide partial pressure of 0.045 atm. to 13Oj, at 0.14 atm. Neglecting this effect does not affect the form of the equation derived for expressing the data, and the value of the correction factor called for is less than the reproducibility of the data. Where the reaction tube itself was the surface studied, the reported rate was based on the entire area of the tube, including the slight portion which protruded outside the bath. The connectors to upstream and downstream condensers were acid-treated borosilicate glass and had an area of about 5% of the reaction tube. Except when the most inrrt surfaces were studied, decomposition on connector surfaces could be neglected; in a few cases a small correction factor was applied based on the known decomposition rate of acidtreated borosilicate glass. Three groups of materials were studied: glasses, organic polvmers, and metals. T o compare the activities of different materials and different methods of surface treatment, rate studies were made at one common temperature chosen to give a readily measurable rate of decomposition: 215' C. for glass surfaces, 215' and 120' C. for polymers, and 150' C. for metals. For most surfaces data were obtained at several hyd'rogen peroxide concentrations, but to condense and summarize the results the decomposition rates reported here are values interpolated to a common partial pressure of, in most cases, 0.04 atm. for glasses and organic polymers and 0.03 atm. for metals.
For extrapolation of vapor phase decomposition data obtained on relatively inert surfaces, experience suggests use of an activation energy of about 6 kcal. and the assumption that the rate is proportional to the hydrogen peroxide partial pressure raised to about the 1.4 power. These are merely approximate figures for engineering use where more precise data are unavailable and are based primarily on studies with various glass surfaces and, to a lesser extent, on 2s aluminum. A more precise expression for the effect of concentration is given in the last section of this paper.
Glass Unaltered Surfaces. T h e rates of decomposition of hydrogen peroxide vapor were measured on the following glass surfaces, in the form of glass tubes: boric oxide fused onto borosilicate glass, fused silica,
