Free iodine can result from the combination of two iodine atoms or from the action of HzOzon hydrogen iodide:
HzOz
1. + I . + 2 HI
.-+
+
I2
+ 2Hz0
Iz
(12) (13)
Similarly molecular hydrogen can be formed as follows:
H*
+ He
+
Hz
(14)
Hydrogen peroxide may be produced by the combination of hydroxyl free radicals and water:
HOH
+ HO.
--+
HzO:!
+ H.
(15)
Additional formation of hydrogen peroxide during ultrasonic irradiation may take place by a mechanism shown by Weissler (1960):
HO*
+ HO.
+
HzOz
(16)
Decomposition of hydrogen peroxide may occur thermally or through interaction with free iodine or hydrogen iodide. The steady-state formation of hydrogen peroxide in aqueous methyl iodide solutions was constant a t 1.25 X 10-6 mole per (liter) (minute). Under identical ultrasonic conditions, pure water yielded hydrogen peroxide a t a constant rate of 3.4 X 10-5 mole per (liter) (minute). The difference was lost through reaction. Thus for a 40-minute ultrasound period, 0.86 X mole per liter of hydrogen peroxide was consumed. However, Figure 2 evidently shows that this consumption cannot account for all of the products formed. A free radical mechanism of the type postulated in this section is required to account for the observed product distribution.
Literature Cited
Anbar, M., Pecht, I., J . Phys. Chem. 68, No. 6, 1462 (1964). Currell, D. L., Wilheim, G . , Nagy, S., J . Am. Chem. SOC.85, 127 (1963). Ingold, C. K., Hughes, E. D., J . Chem. Soc. (London) 1935, p. 225; 1936,p. 225; 1937,p. 1183. Jennings, B. H., Townsend, S. N., J . Phys. Chem. 65, 1574 (1961). Lindstrom, O., J . Acoust. SOC.Am. 27, No. 4, 654 (1953). Lliboutry, L., J . Chzm. Phvs. 41, 13 (1944). Lur’e, Y . Y . , Kandzas, P. F., Mokina, A. A., Russ. J . Phys. Chem. English Transl. 36, No. 12, 1424 (1962). Lur’e, Y . Y . , Kandzas, P. F., Mokina, A. A , , Russ. J . Phys. Chem. English Transl. 37, NO. 1, 6 (1963). McTigue, P. T., Buchanan, A. S., Trans. Faraday SOC.55, 1153 (1959). Moelwyn-Hughes, E. A,, Proc. Roy. SOC.A164, 295 (1938). Petry, R. C., Schuler, R. H., J . Am. Chem. SOC.75, 3796 (1953). Prakash, S., Pandey, J. D., C h m . Anal. 46, No. 12, 607 (1964a). Prakash, S., Pandey, J. D., J . Indian Chem. SOC.41, No. 5, 316 (1964b). Prakash, S., Pandey, J. D., Tetrahedron 21, 903 (1965). Prakash, S., Pandey, J. D., Ichhaporia, F. M., Indian J . Chem. 2, 76 (1964a). Prakash, S., Pandey, J. D., Singh, R., J . Prakt. Chem. 26, No. 4, 140 (1964b), in English. Prakash, S., Prakash, S., J . Acoust. SOC.Am. 32, 138 (1960). Prakash, S., Prakash, S., J . Sci. Ind. Res. India 24, 629 (1965). Prudhomme, R. O., Grabar, P., J . Chzm. Phys. 46, 323 (1949). Roberts, J. D., Caserio, M. C., “Basic Principles of Organic Chemistry,” p. 301, W. A. Benjamin, Inc., 1964. Schuler, R. H., Chniiel, C. T., J . Am. Chem. SOC.75, 3792 (1953). Srivastava, S. C., Nature 182, No. 4627, 47 (1958). Srivastava, S. C., Berkowitz, N., Can. J . Chem. 41, 1787 (1963). Sturm, J., Schwarz, J. A , , Radzation Res. 17, 531 (1962). Tuszynski, T. M., M.A.Sc. thesis, University of Toronto, 1966. Weissler, A., J . Acoust. SOC.Am. 32, 283 (1960). Weissler, A., J . Am. Chem. SOC.81, 1077 (1958). Weissler, A., Cooper, H. W., Snyder, S., J . A m . Chem. SOC.72, 1769 (1950). Weissler, A , , Pecht, I., Anbar, M., Science 150, No. 3701, 1288 (1965).
Acknowledgment
The authors acknowledge the helpful suggestions of W. H. Burgess, A. J. Szonyi, and U. de Boni.
RECEIVED for review April 7, 1967 ACCEPTED February 5, 1968 Work supported by the Ford Foundation.
OXIDATION OF HYDROGEN CHLORIDE IN A MICROWAVE DISCHARGE W I L L I A M W . C O O P E R , l H A R O L D S . M I C K L E Y , A N D R A Y M O N D F. B A D D O U R
Department of Chemical Engineering, Massachusetts Institute of Technologv, Cambridge, Mass. 02739
+
+
The oxidation reaction 4HCI 02 -+ 2 H 2 0 2C12 was carried out with yields of up to 55% by use of a 2450-megacycle microwave discharge. Experimental results confirm the theoretical prediction that the ratio of the electric field strength to gas density is the important parameter. A postulated kinetic scheme is shown to be supported by the experimental measurements.
ORK done with d.c. arcs has shown that chemical reacwtions carried out in high temperature plasmas may have the required activation or free energy for reaction supplied through electron excitation. However, the fact that these processes operate a t very high gas temperatures of from 2000’ to 10,000’ K. (Bronfin, 1963) restricts their applicability; and they can be used only when the thermodynamics or kinetic
Present address, Mass. 02139. 400
Inc., 341 Vassa,. St,, Cambridge,
I&EC FUNDAMENTALS
sequence is such that the desired product may be removed in a quench step. T h e d.c. arc has the further drawback of requiring electrodes whose presence can present a serious contamination problem and, if metal, may be catalytic. Problems associated with arc processes may be avoided through the use of an electrodeless discharge contained within a dielectric vessel. Such a discharge is a nonequilibrium phenomenon, in that the gas temperature is not equal to but is less than the electron temperature. This difference allows the excitation of active chemical species through electron bom-
bardment withoht the simultaneous development of very high gas temperatures. For example, in the work reported here the gas temperatures were found to lie in the range of 500’ to 700’ C. while the chemical activation indicates that electron temperatures over 10,OOOo K. were present. This combination of relatively low gas temperature and high concentration of chemically active species presents a promising medium for carrying out reactions. T h e work reported here was carried out in a microwave discharge at 2450 megacycles per second. This corresponds to a free space wavelength of slightly over 12 cm. Although McCarthy (1954) indicates that a microwave discharge is a particularly rich source of chemically excited species, the physics show no reason why microwaves should be any better for excitation than any other high frequency. As long as this frequency is high enough so that all free electrons are not swept out of the discharge volume over the period of a half cycle, there should be no difference in discharges. If these electrons are lost over this time period and must be replaced in each half cycle, then the discharge is indistinguishable from a d.c. glow discharge. Because of the collisional excitation process, the molecule excited knows only that it was struck by a n electron of a certain energy, not how the electron gained this energy. This picture is in direct contrast to various quantum absorption processes such as photochemical reactions. There selectivity is obtained by radiating the reactants a t the particular wavelength corresponding in energy to the desired process. Although microwave power sources were developed for radar applications during World War I1 and the physical mechanism of gaseous breakdown was studied extensively during the next decade (Brown, 1951 ; Herlin and Brown, 1948a, 1948b), their use as a means of providing active species for chemical reaction was not reported until 1954 in a paper by McCarthy (1954). Certainly, part of the reason for the lack of interest in them as a chemical processing tool arose from the costs associated with this type of power. Nevertheless, since that time there have been a number of investigations (Frazer and Holzmann, 1958; McTaggart, 1965; Shaw, 1959; Streitweiser and Ward, 1963; Vastola and Wightman, 1964) into microwave plasma chemistry. I n spite of thr work that has been done, little attention has been given to controlling the species produced or to maximizing the production of species considered desirable for the reaction under study. T h e theoretical considerations presented below indicate that, by regulating the ratio of the electric field strength to gas density, the energy of the electrons may be controlled. T h e cross sections for the various excitation processes are also a function of the electron kinetic energy, and the product distribution from the discharge will be determined by control of this energy. Within the limitations imposed by the available equipment, work was done to test the hypothesis that the ratio of the electric field strength to gas density governs chemical synthesis in the microwave discharge. T h e oxidation of hydrogen chloride was chosen for study. T h e reaction, as shown below, is thermodynamically favorable and free from side reactions but will not readily proceed because of kinetic hindrances. 4HC1
+
0 2
= 2C12
+ 2H20
AFzssO = -18.196 kcal.
Recently interest in this “Deacon reaction” has increased markedly as hydrogen chloride from chlorination processes has become a troublesome by-product. However, most efforts in this area have been directed toward the development of a
better catalyst to take advantage of the favorable low temperature thermodynamics. T h e use of a low temperature plasma presents another way of carrying out this reaction in a region where the thermodynamics are favorable. While in a n over-all chemical sense the microwave discharge supplies the necessary activation energy for reaction, from a kinetic viewpoint the plasma provides the active species which initiate the reaction. Because low temperature thermodynamics are favorable, there is no necessity for a rapid quench step to freeze out products a t some favorable high temperature equilibrium point. Theoretical
The conditions for the breakdown of a gas to form a plasma have been derived by a number of workers (Brown, 1951; Kontaratos and Demetriades, 1965) and are not discussed in detail here. Basically, the criterion is that the rate of electron production through ionization equals the rate of loss from diffusion, recombination, and attachment. Here the emphasis is placed on the chemical excitation occurring in the plasma. A solitary electron oscillating in a high frequency electromagnetic field and not undergoing collisions receives no net energy gain from the field. This situation changes markedly when the electron undergoes collisions. I n this case the equation of motion must be written to include the effects of momentum transfer on collision. For elastic collisions, the equation of motion is md2z/dt2
+ mvmu = -qEo
sin wt
(1)
where the electric field has been assumed to have components only in the z direction. I n terms of the electron’s velocity, Equation 1 may be rewritten as: du/dt
+ vmu = ( - q E , / m )
sin ut
(2)
Assuming that the initial velocity at time t = 0 is itself zero, integration of Equation 2 yields: dt/dt = (-qE,/m)(v,
sin ut
- w cos u t - .)/(a2 + vm2)
(3)
The work done on the electron in a time dt as it moves a distance dz under the influence of the field can be expressed as: dW = F
. dz
= -qEo
sin wtdt
(4)
Substituting Equation 3 into Equation 4 and integrating from t = 0 to t = 2 7 r / ~results in the following expression for the work done on the electron over the period of a cycle:
+ urn2)
W = q2E02nvm/mw(w2
(5)
Since the frequency of the field is related to the radian frequency by the simple relation f = w/2n, the power input to this electron is given by the product of the work per cycle and the frequency. T h e resulting equation for the power input per electron is:
Through a series of these elastic collisions, the electrons on the average progressively gain energy until finally they have enough energy to undergo a n inelastic collision. T h e result of such a collision is the excitation of the atom or molecule with which the electron collides. I n this way the active chemical species which cause reaction are produced in the plasma. Vasilev (1950) has performed a n energy balance o n these colliding electrons. From this calculation he has determined the fraction of the energy transferred on collision which goes VOL. 7
NO. 3 A U G U S T 1 9 6 8
401
into the excitation of a particular level of the struck particle. H e has shown that the distribution of this energy among the various possible levels for excitation is a function of the electron temperature only. Hence, electron temperature is the parameter that determines the excited species produced in a n electric discharge. Since the average energy of the electrons is related to their temperature and kinetic energy by the equation : e, = ( 3 / 2 ) k T e =
‘/2
mii2
(7)
the variable which controls the electrons’ energy will control their temperature. Equation 7 implies that the electrons’ distribution may be approximated by a Maxwellian. Massey and Cannon ( 1 9 6 5 ) have developed the equations necessary to determine this important physical parameter. Equation 6 may be rewritten in terms of an effective field strength defined as:
molecule. Such collisions produce the free radicals which are the actual chemical species responsible for reaction. Since activation may follow only from those collisions of proper energy, its differential rate mayibe written as: d R = vRdn,
where R is the activation rate, vR is the activation collision frequency, and dn, is the number density of electrons in the proper energy range to cause activation. T h e activation collision frequency may in turn be written in terms of the activation cross section as: VR
= NQRv
v, = o N Q
VR
q2Eeif2/2mvm=
~V,E~
Rearrangement of Equation 10 gives: e,
= q2Eefi2/2m6vm2
This equation shows that basically it is the ratio of the electric field strength to gas particle density which determines the average electron energy. T o convert this to the more commonly reported variable E/p, use is made of the perfect gas laws to express the number density of molecules as:
N = (6.02 X 1OZ3)(2.24 X 1 0 4 ) ( p / p , ) ( T , / T , )
(13)
where po and To are the standard pressure and temperature. T h e average electron energy in ergs then becomes:
402
l&EC FUNDAMENTALS
(18)
+
(19)
dn = ndf
If all collisions were elastic and there were no electric field present, the electron energy distribution function would be Maxwellian. The field and inelastic collisions will perturb the distribution; but this will be ignored, and the Maxwellian used as a n approximation. Within the limits of this approximation, the fraction df is given by Equation 2 0 : df = (3/e0)3’2(~/21r)1i2e-3~~2~~ de
(20)
Here e, is the same average electron energy appearing in Equations 7 to 14, but e is the actual energy of a particular electron. Substituting Equations 18 through 20 into Equation 15 results in the following expression for the differential activation rate: d R = ( n N ) (l/mn)1~z(3/~o)3’2Q~~e~3f~~fo de
T h e total activation rate may be obtained by integrating Equation 21 over all possible electron energies. This integral running from zero to infinity may be expressed as:
R Since under the conditions of operation the electron collision frequency is approximately 10“ per second and the microwave radian frequency is 10’0 per second, the effective field is very nearly the same as the applied field E,, and no correction via Equation 8 is necessary. Brown and MacDonald ( 1 9 4 9 ) have shown the importance of the ratio of electric field strength to pressure as a variable in correlating breakdown data. Work carried out by Lunt and Meek (1936) showed that hydrogen dissociation in a d.c. glow discharge was a function of this group, but no work to date has demonstrated its importance in synthesis reactions. T o understand the importance of the electron energy in governing chemical synthesis, consider that the primary initiation process is the inelastic collision of a n electron with a gas
= NQR(2e/m)112
T h e number density, dn,, of electrons in the proper energy range between E and e f de is given by the product of the total electron density, n, and the fraction of electrons in the energy de: range e to E
(11)
T h e use of Equations 7 and 9 transforms Equation 1 1 to:
(17)
Combining Equations 16 and 17 yields the following expression for the activation collision frequency:
(9 )
I n these equations D is the average electron speed, N is the gas density, and Q is the elastic collision cross section. Finally, a quantity 6 is defined as the fraction of an electron’s energy which is transferred upon collision. For a perfectly elastic collision, 6 = 2m/M. At steady state there is a balance between the rate at which energy is being imparted to the electrons from the field and the rate a t which the electrons are passing it on to the particles with which they collide. This balance may be expressed as:
(16)
I n Equation 16 N is the density of molecules with which the electrons collide, Q R is the collision cross section for activation, and u is the electron speed. This speed may be expressed as a function of the electron kinetic energy as: u = (2e/m)1’2
and the collision frequency may be written in terms of the collision cross section as:
(15 )
so-
= ( n N )( l / m ~ ) ~ / ~ ( 3 / e , ) ~ QRee-3e/2eode /~
(22)
T h e integral is a function only of the average electron energy E, which, through Equation 14, is a function of E/p. T o perform the integral in Equation 22, data on the initiating reaction cross section, Q R , as a function of E , the electron energy, are required. Then using Brown’s reported data (1959) on electron temperature as a function of E/p, Equation 7 may be used to plot the integral as a function of E/p. Except for the multiplicative factor ( n N ) , the reaction rate expression is a function of E/p alone. As a first approximation, this multiplicative factor will be taken as a constant. This simplification may be justified for the experimental conditions employed in this work. The discharge operating a t pressures of 5 to 50 mm. of mercury is attachment or recombination controlled. I n this case the electron density, n, varies roughly inversely
prossuro gauge
0
scavenger trap
sample trap
I
I
cooling water
Figure 1 .
Schematic of apparatus
first reactant
I
discharge ton@
A /
\
Figure 2. Side wall iniection, separated feed reactor .
with the pressure (Brown, 1959), while the gas density, N , varies directly with the pressure. These effects tend to compensate each other to make the product approximately constant. Experimental
Figure 1 shows a schematic diagram of the experimental apparatus. T h e microwave power was generated by a Raytheon PGM-100 unit capable of a n output of up to 1 kw. a t 2450 megacycles per second. T h e isolater section prevented reflected power from interacting with the magnetron. Since the power output could not be reduced below 500 watts when low power experiments were to be run, it was necessary to install a flap-type attenuator in the wave guide system. T h e power absorbed in the discharge itself was calculated from the
incident power reading obtained from the directional coupler and the voltage standing wave ratio obtained from the slotted section. T h e power itself was coupled into the plasma by means of a tapered terminating section (Fehsenfeld, Evenson, and Broida, 1965). A straight quartz tube was used for runs in which the reactant gases were premixed. The reactor shown in Figure 2 was used when only one gas was subjected to the discharge, followed by mixing downstream with the other. USP grade oxygen and 99% pure anhydrous hydrogen chloride were fed directly from the cylinders without further purification. The flows were monitored at 1 atm. with Matheson No. 600 series rotameters and were always in stoichiometric 4-to-1 ratio of HCl to 0 2 . T h e sample loop shown in Figure 1 provided a means of trapping out all of the products in a liquid nitrogencooled cold trap. These were warmed to room temperature, collected in a caustic solution, and analyzed via a potassium iodide titration of the free chlorine. Results and Discussion
I n the plasma, the electron energy determines the probability or cross section for the formation of the necessary active chemical species for reaction. This cross section is not, however, a single-valued function of electron energy but rather passes through a maximum. Physically, the reason for such a maximum is that the cross section represents the wave mechanical interaction between the impinging electron and the internal energy levels of the particle being activated. If the electron moves either too slowly or too quickly, there is little energy exchange. I n the discharge itself the activation process is thought to be a combination of the two reactions below:
+ 0HC1 + e = H + C10 2
+ e = 0
Cross-section data on the first of these have been reported by Schulz (1962) and on the,second by Fox (1957). I n both cases monoenergetic electron beam techniques were used to plot the curves, both of which showed the expected maximum. Using these cross-section data and Equation 22, the activation rate was plotted as a function of E/p. The procedure was to select a value of e, and then graphically integrate Equation 22. Repeating this process a t different values of e, yielded a curve of reaction rate as a function of eo. Equation 7 VOL. 7
NO. 3 A U G U S T 1 9 6 8
403
o' -
016
w
\
go I
A\A
012. 012-
0
W
A '
Or0 S+J A
\
-
0.08.
004
0.00
-
0
Figure 3.
I
I
'
4
~
~
I
I
16
8
l2
1 €4
-
I
.20 24 voWcmlmrn 1 Hg
1
I
I
28
32
1
1
Integrated hydrogen chloride dissociative attachment cross section as a function of
ElP
-
1
I
1
I
I
1 -
I
I
I
I
I
-
A
1
1
100
200
Figure 4. Mixed feed of 0.579
I
400
500 600 watts of power absorbod in tho discharge
300
Conversion vs. power at 10 mm. of
I&EC FUNDAMENTALS
I
800
900
pressure
X IO-' mole of 02 per second and 2.32 X 1 0-4 mole of HCl per second
was then used together with Brown's data to transform the abscissa into the variable E / p rather than eo. Figure 3 shows the results of this calculation using Fox's cross-section data on the dissociative attachment of hydrogen chloride. A very similar curve is obtained from Schulz's data on the oxygen dissociative attachment, except that the maximum occurs for a larger value of the electron energy. Since the processes of activation occurring in the discharge are fast compared to the residence time of thr gases in the discharge zone (Kondrat'ev, 1964), the apparent rate of chlorine formation will be governed by the gas flow rate. I n contrast, the extent of reaction or conversion will be independent of gas flow rates and determined by the relative concentrations of electrons of the appropriate energy to promote the forward as compared to the reverse reaction. T h e concentration of these electrons will be determined by their relative rates of formation balanced against their rates of loss by all processes. The rate of formation of electrons of a certain energy level is a function 404
Hg
I
700
of E / p , as shown previously. Consequently, if this hypothesis be true, the net conversion of hydrogen chloride to chlorine will be a function of E / p independent of gas flow rates. However, this last statement is no longer true when the power input to the discharge is not in sufficient excess to ensure that production of active species is not the limiting step. I n one set of experiments conversion data were taken using a premixed feed of HCl and 0 2 in stoichiometric (4 to 1) ratio, and a reactor diameter of 11 mm. I n Figure 4 conversion is shown as a function of absorbed power a t a fixed pressure of 10 torr and flow rate of 2.9 X 10-4 mole per second. I n Figure 5 conversion is shown as a function of absorbed power a t a fixed pressure of 20 torr and flow rate of 2.9 X 10-4 mole per second. I n Figure 6 conversion is shown as a function of pressure a t a flow rate of 1.7 X 10-4 mole per second and a t absorbed power levels of 208 and 342 watts. T h e data of Figures 4, 5, and 6 may also be used to test the hypothesis that E / p controls the discharge reaction. Although
1
70
-
60
-
.Es .-E
1
I
1
I
I
I
1
-
I
-
50-
0 c
-
8 406
c
2: a
-
al
30-
I
Bo
100
I
-
I
'5
i+
B
4c-
C
sc
30-
I
400
X 1 0-4
I
500
I .
600
I
700
I
800
900
mole of 0 2 per second and 2.32 X 1 0-4 mole of HCI per second
I
I
I
/O-Q-S\
50-
%
300
Conversion vs. power at 2 0 mm. of Hg pressure
Figure 5. Mixed feed of 0 . 5 7 9
60
1
200
I
I
I
1
-I
-
-
1
A 4
P/
-
-
8
-
Eii x 10
-
-
0 208 watts of 0 342 power absorbed
I
I
I
I
I
I
I
I
I
5
10
15
20
25
30
35
40
45
it was not possible to measure the electrical field directly, under the experimental conditions employed the power absorbed in the discharge is very nearly proportional to the square of the electric field strength (Brown, 1959). Hence, on Figure 7 the d a t a of Figures 4, 5, and 6 are plotted with conversion as the ordinate and the square root of the absorbed power divided by the pressure as the abscissa. This method of plotting correlates data covering the range: absorbed power 100 to 700 watts, pressure 5 to 40 torr, and flow rate 1.7 to 2.9 X mole per second. Comparison of the curve of Figure 7 with that of Figure 3 shows the qualitative similarity between the experimental results and the approximate theoretical calculations. T h e 11-mm. diameter reactor shown in Figure 2 was used when only one of the reactant gases was passed through the discharge, the other being added through the side a r m downstream. Data taken with this reactor and shown in Figures 8 and 9 demonstrate that the active species required for reaction
have a finite lifetime outside the discharge zone and that reaction may take place in this plasma-free region. To explain the data which indicate that reaction may occur when only one gas is subjected to the discharge, a kinetic scheme must be developed depending on a n active energy-carrying species having a mean lifetime on the order of 10 msec. Consider first the case when only oxygen is passed through the discharge, followed by mixing with hydrogen chloride, T h e postulated kinetic scheme is:
+ HCl = OH + C1 OH + HCl = H2O + C1 c1 + c1 = Cl2
0
AF29s0
= 0.586 kcal.
AF2980 =
- 15.043 kcal.
AF2980 = -50.204 kcal.
This scheme is supported by the following facts. I t is well established (Elias et al., 1959) that oxygen atoms are the major product of a microwave discharge in oxygen, although smaller amounts of the lA0 megastable state are also formed VOL. 7
NO. 3
AUGUST 1 9 6 8
405
60
I
I
I
1
1
I
I
I
1
c
Y
P
10
A 10 mm Hg pressure
-
0 20 mm Hg pressure 0 208 watts absorbed
I
I
I
I
60 I-
O 'h
I
V 342 watts absorbed
A new initiating step is needed to explain the data which show reaction occurring when the oxygen is injected downstream from a hydrogen chloride discharge. Del Greco and Kaufman (1962) and Foner and Hudson (1962) report the reaction of hydrogen atoms with oxygen molecules via a third body collision. Their reaction sequence is shown below with the free energy changes for each step.
-I
F
I
0 lOmm Hg pressure
A20mm Hg pressure
20
+ + M = HOz f M H + HOz = 2 0 H H + HOs = HzO + 0 H
t
a2
0;.l
I
I
1
1 L
1
1 1
2 3 5 cm. sbparation from HCL input to O2 discharge 1
Figure 8.
Conversion vs. position of HCI input
Side wall injection reactor at 327 watts absorbed and flows of 0.579 X 1 0-4 mole of 02 per second and 2.32 X 1 0-4 mole of HCI per second
(Foner and Hudson, 1956). Harteck and Kopsch (1931) have reported the first of these reactions to be the primary process by which oxygen atoms formed in a silent discharge react with hydrogen chloride. Spectra of the reaction zone taken by the present experimenters disclose the presence of all the postulated atoms and of hydrogen radicals. 406
I&EC FUNDAMENTALS
-
0 2
AFZg8O= -40.538 kcal.
AF2980= -40.248 kcal. AF2gao =
-47.877 kcal.
With the formation of oxygen atoms and hydroxyl radicals, this scheme merges into that proposed for thr oxygen discharge; and the atoms and radicals can react with the remaining hydrogen chloride to form chlorine. These reaction sequences must be consistent with the experimental results obtained with mixed feeds. As a basis for comparison, in Figures 8 and 9 the yields obtained with the separated feed reactor are extrapolated to zero input position of the second gas. When hydrogen chloride is the activated gas, Figure 9 indicates this extrapolated yield to be 50y0 a t both 10 and 20 torr; and when oxygen is activated, Figure 8 shows it to be 17% a t 10-mm. and 35% a t 20-mm. pressure. These figures compare with the 46 to 48% yields obtained using a mixed feed at the same 320-watt absorbed power level. Since experiment shows that premixing the gases does not sum the yields achieved unmixed, hydrogen chloride and oxygen cannot both be activated to the same extent when mixed as when passed through the discharge alone. T h e explanation of these observations lies in the relative cross sections for the activation processes. Because Figure 9 shows a higher yield a t zero separation than Figure 8, the data indicate that the hydrogen chloride is more effectively activated than the oxygen. This finding is consistent with data on their relative electron collision cross sections reported by Brown (1959) and with the fact (Fox, 1957; Schulz, 1962) that the maximum in the dissociation attachment cross-section curve for hydrogen chloride occurs at 0.9 electron volt while that for oxygen is a t 6.7 e.v.
OIOmm Hg prossure D2Omrn Hg pressure
-c
c
9.
Figure
Conversion vs. position of
0 2
input
Side wall injection reactor at 31 8 watts absorbed and flows at 0.579 of 0 2 per second and 2.32 X lo-' mole of 02 per second
0 1Omm Hg prassuro D 2Omm Hg pressure
X 1 0-4 mole
Let n be the atomic concentration in such a wafer-like element a t any time, T h e rate a t which these particles collide with the wall is given by the kinetic theory (Chapman and Cowling, 1961) as nu/4 collisions per square centimeter per second. Here u is the mean peculiar speed of the particles. T h e area available for collision is 2 m d z and the element has a volume of d d z . Let y denote the efficiency of recombination-that is, the number of recombinations per collision. Then the rate of change of density in this volume element as it moves along will be given by Equation 23 :
dn/dt = -y(nu/4)(2nrdz/nr2dz) =
-ynu/2r
(23)
This equation integrates to: In(n/n,) =
h A
timo Figure 10. flight
of
I
I
I
10 flight
20
30
from
I
-
40 50 O2 discharge to HCI input milliseconds
Oxygen atom concentration vs. time of
Side wall injection reactor at 327 watts absorbed and flows of 0.579 X 1 0 - 4 mole of 0 2 per second and 2.32 X 1 0-4 mole of HCI per second
I t is possible to model the reactor being used as a semiinfinite cylinder and consider the decay in active species concentration as a slug of material moves down the tube away from the discharge. T h e differential equations governing the bulk and diffusive flow have been set up and solved (Cooper, 1966) and show that the same result is obtained from a mass balance applied to a volume element as it moves down the tube after the discharge.
- (yu/2r,)t
(24)
Plotted on a semilogarithmic scale, this equation has a slope of - (yu/2r,). T h e data in Figures 8 and 9 are presented in terms of percentages and distances. Assuming that the mechanistic sequences postulated above are correct and knowing the temperatures in the region downstream from the discharge, it is possible to back-calculate the oxygen or hydrogen atom concentration a t the point of injection of the second gas and also the time of flight from the discharge zone to this point. These data have been replotted in terms of those variables in Figures 10 and 11. I n these latter figures the data a t the two pressures plot as parallel lines, as would be expected from Equation 24 which shows no pressure dependence. From the slope of the lines in Figures 10 and 11, it is possible to calculate the recombination coefficient. T h e only other information necessary is the radius of the tube and the mean peculiar speed of the oxygen or hydrogen atoms. The former was 0.55 cm. T h e latter can be calculated from a knowledge of the temperature in the region downstream from the discharge. From kinetic theory this mean peculiar speed is given by : u = (8kT/i~M)'/~
(25)
For the purposes of calculation on the oxygen data of Figure 10, a n average value of 625' K. was used. This yields a mean peculiar speed of 9.05 X I O 4 cm. per second. With this value VOL. 7
NO. 3
AUGUST 1 9 6 8
407
OlOmm Hg prossuro
A20m Hg prPsswo
-
--
a -
-
L
a
302
-eK c
-
aob
I
1
2
4
I '
1
8
6
t
1
1.2
- mi\\accondo
t h o of fUght from HCL discharge to O2 ':put
Figure 1 1 .
I
I
14
16
Hydrogen atom concentration vs. time of night
Side wall injection reactor at 31 8 watts absorbed and flows of 0.579 and 2.32 X 1 0-4 mole of HCl per second
and the measured slope of 107 seconds-1 for the lines in Figure 10, the oxygen wall recombination coefficient is calculated to be 13.0 X 10-4. For hydrogen chloride the corresponding temperature is 685' K., which results in a coefficient of 8.62 X 10-4. T h e number obtained with the oxygen data compares with a value of 7.1 X lO-4reported by Greaves and Linnett (1958) for recombination on quartz, No value was found in the litera-
Table I.
X 10-4 mole of
0 2
per second
ture with which to compare the recombination coefficient for hydrogen chloride. T h e agreement between the oxygen recombination coefficient calculated from the experimental data and that reported in the literature provides support for the proposed mechanism which postulates the oxygen atom as the active initiating species. The steeper slopes of the lines in Figure 11 indicate a faster decay rate of the active species in the hydrogen chloride
Diameter and Flow Effects (Mixed Feed)
Minimum Power to Maintain Discharge, Watts 127 151 111 149 132
Reactor Diameter, Mm. 2 11
Flow Rate, Range of M l . / M i n . at 7 Atm. Pressure, Absorbed Range of HCl 0 2 M m . Hg Power, Watts Conversions, yo 20 340 85 127-382 34.8-50.2 340 15 1-709 20 85 42.8-51.3 85 10 340 29.9-52.7 ._ 111-687 340 36.3-37.2a 22 85 20 149-275 10 340 85 132-472 48.8-51.9 680 10 11 170 165 47.6 ... 680 185-420 38.548.2 20 170 185 170-384 10 1360 340 170 21.3-37.P 20 1360 179-420 340 23.1-5 1 . 1 179 a Low yields result from incompletefilling of crosssection of reactor discharge at this higher pressure and large diameter. Reactant gases may pass through discharge region without passing through discharge. Lowyields refrect inabilib of cold trap to collect allproducts at this highflow rate and lower pressure. Table II. Energy Requirements
Run No.
435 436 437 438 439 440 441 442 443 444 445 447 448 451 452
408
Flow Rate, M l . / M i n . at 1 Atm. HCl 02 680 170 680 170 680 170 680 170 680 170 1360 340 1360 340 1360 340 1360 340 1360 340 1360 340 1360 340 340 1360 1360 340 1360 340
l&EC FUNDAMENTALS
Pressure, Mm. Hg 20 20 20 20 10 20 20 20
10 10
%
Conversion 41.4 45 .O 38.5 48.2 47.6 42.8 45.6 51.1 33.3 47.4 21.3 23.1 35.6 36.8 37.9
Power Absorbed, Watts 185 266 340 420 165 265 338 420 252 408 170 179 338 384 31 1
Kw.-Hr./Lb. Clz 3.45 4.57 6.80 6.72 2.67 2.38 2.86 3.17 2.91 3.33 3.08 5.65 3.68 4.04 3.17
than for the oxygen. If hydrogen atoms are indeed the initiator there, this faster decay would be expected, because a t the same temperature hydrogen atoms have a mean peculiar speed four times that of the oxygen atoms and hence more opportunity to collide with the wall and be lost. Energy Requirements
T h e amount of power being fed into the gas greatly exceeded that required to dissociate all of the reactant material. For example, with a flow rate of 0.579 X 10-4 mole per second of oxygen, only 28.4 watts would be needed to dissociate it completely to atoms. This compares with 111 watts, the minimum power to maintain the discharge. Several attempts were made to make more efficient use of this power. An increase in reactor diameter which would reduce diffusion losses of electrons did not reduce the power required to break down the gas. T h e lack of any effect of diameter, or more specifically surface area (Table I), on the power required for breakdown indicates that the discharge is recombination or attachment rather than diffusion controlled. At the relatively high pressures of this work, this loss mechanism is not unexpected; and Brown (1959) mentions its importance. I t is also consistent with the dissociative attachment activation scheme presented above. Temperatures of the gas were measured in the region just downstream from the discharge with a quartz-sheathed thermocouple. T h e values measured were in the range of 500’ to 700’ C. Heat transfer calculations (Cooper, 1966) showed errors due to recombination on the probe to be less than 10’ C. An energy balance around the system showed that to within =tlO’% the power input to the discharge as measured by the microwave power meter was the same as rate of increase in sensible heat in the reactor’s cooling oil. Simply placing the oil-filled reactor jacket in a microwave field did not cause any significant increase in oil temperature. These facts indicate that the excess energy going into the discharge produced activated species which ultimately gave up their energy a t the wall. Since more than enough energy was being put into the gas to break the molecular bonds, the flow rates were increased to try to make more efficient use of this energy. This increase in flow rates led to a reduction in energy consumption per pound of chlorine formed. Table I1 shows the results when the flows were first doubled and then quadrupled over those of the data in Figures 5 and 6. T h e lowest amount of energy required was 2.4 kw.-hr. absorbed in the discharge per pound of chlorine. T h e capacity of the cold trap sampling system to remove the products limited any further increases in flow rates. I n addition to increasing flow rates, further reduction in energy requirement might br obtained by using a generator which has a pulsed rather than a . continuous output. I n this way a sharp peak of power would be used to break down the gas a t intervals corresponding to the residence time of the reactant gases in the reactor. Of course, the minimum energy requirement is set by that needed to produce enough of the active species required to initiate the reaction sequence. Acknowledgment
T h e authors gratefully acknowledge the assistance on both the practical and theoretical aspects of the work that was given by Peter H . Dundas and Alexis T. Bell. One of the authors (W.W.C.) was supported during part of this work by a National Science Foundation Cooperative Graduate Fellowship.
Nomenclature
E = electric field strength, volts/cm. E , = amplitude of electric field, volts/cm.
f =
frequency, reciprocal seconds
F = force, dynes k = Boltzmann’s constant, 1.38 X 10-’6
erg per degree Kelvin m = mass of electron, grams M = molecular or atomic mass, grams n = active species density, moles/cc. no = active species density a t base of discharge, moles/cc. N = number density of gas molecules, number per cc. P = power, watts or ergs per second P = pressure, atmospheres or mm. H g P o = standard pressure a t 1 atm. 4 = electronic charge, coulombs Q = electron elastic collision cross section, sq. cm. R = activation rate, molecules per cc.-sec. r = radial position in reactor, cm. ro = radius of reactor, cm. t = time, seconds u = mean peculiar speed, cm./second T , = electron temperature, degrees Kelvin To= gas temperature, degrees Kelvin T o = standard temperature, degrees Kelvin u = velocity, cm./second o = average electron speed, cm./second W = work, ergs Y = recombination coefficient, dimensionless 6 = fraction of a n electron’s energy transferred on collision Eo = electron kinetic energy, ergs y m = collision frequency for momentum transfer, reciprocal seconds w = angular frequency, radians per second literature Cited
Bronfin, B. R., Sc.D. thesis in chemical engineering, Massachusetts Institute of Technology, Cambridge, Mass., 1963. Brown, S. C., “Basic Data of Plasma Physics,” M.I.T. Press, Cambridge, Mass., 1959. Brown, S. C., Proc. I R E 39, 1493 (1951). Brown, S. C., MacDonald, A. D., Phys. Rev. 76,1629 (1949). Chapman, S., Cowling, T. G., “The Mathematical Theory of Non-Uniform Gases,” 2nd ed., University Press, Cambridge, 1961. Cooper, W. W., Sc.D. thesis in chemical engineering, Massachusetts Institute of Technology, Cambridge, Mass., 1966. Del Greco, F. P., Kaufman, Frederick, Discussions Faraday SOC.33, 128 (1962). Elias, L., Ogryzlo, E. A., Schiff, H. J., Can. J . Chem. 37, 1680 (1959).
Fe‘hsenfgld, F. C., Evenson, K. M., Broida, H. P., Rev. Sci. Inst. 36,294 (1965). Foner, S . N., Hudson, R. L., J . Chem. Phys. 25, 601 (1956). Foner, S. N., Hudson, R. L., J . Chem. Phvs. 36.’ 2681 (1962). Fox, R. E., j.Chem. Phys. 26,’1281 (1957): Frazer, J. W.,Holzmann, R. T., J . Am. Chem. SOG.80, 2907 (1958). Greaves, J. C., Linnett, J. W., Trans. Faraday SOC. 54, 1323 (1958). Harteck, P., Kopsch, U., Ann. Physik. Chem. B12, 327 (1931). Herlin, M. A., Brown, S. C., Phys. Rev. 74, 291 (1948a). Herlin, M. A., Brown, S. C., Phys. Rev. 74, 910 (1948b). Kondrat’ev, N. M., “Chemical Kinetics of Gas Reactions,” Addison-Wesley, Reading, Mass., 1964. Kontaratos, N., Demetriades, S. T., Phys. Rev. 137(6A), 1685 (1965). Lunt, R. Winstanley, Meek, C. A., Proc. Roy. SOC.A157, 146 (1936). McCarthy, R. L., J. Chem. Phys. 22, 1360 (1954). McTaggart, F. K., Australian J . Chem. 18,937 (1965). Massey, J. T., Cannon, S. M., J . Appl. Phys. 36, 361 (1965). Schulz, G. J., Phys. Rev. 128, 178 (1962). Shaw, R. M., J . Chem. Phys. 30,1366 (1959). Streitweiser, Andrew, Jr., Ward, H. R., J . Am. Chem. SOG. 85, 539 (1963’1. VasiIev,’S. S., Zh. Fiz. Khim. 24, 1107 (1950). Vastola, F. J., Wightman, J. P., J . AppI. Chem. 14, 69 (1964). .
I
RECEIVED for review April 18, 1967 ACCEPTED February 14, 1968
VOL. 7
NO. 3
AUGUST 1968
409
