CHE1IIC.kL hCTIO?; I S T H E GLOW DISCHARGE IT. T H E S T S T H E S I S OF OZOSE* B'i A . KEITH RItEJYER AXD J. V-.ITESTHAVER
The innumerable investigations on the synthesis of ozone in various types of electrical discharges. by a ray bombardment, and by ultra-violet radiation indicate that ozone is readily formed around positive oxygen ions, and possibly around oxygen atoms and excited oxygen molecules. A direct coniparison of the relative reactivities of these various forms is completely lacking. Considerable evidence has been accumulated which indicates that inert gases such as nitrogen are positive catalysts in ozone formation. The function of the catalyst or of any of the mechanisms involved, however, is a matter of conjecture, The difficulty in the study of ozone formation is an inherent one, arising out of its extremely unstable nature. It is almost impossible to obtain the direct formation under conditions where decomposition is absent or negligible. For this reason most investigators have confined their study to conditions a t or near equilibrium, and so can say little regarding the mechanism. Recently Lind' in synthesizing ozone by a ray bombardment, has obtained a ratio of ozone molecules to oxygen ions of h l 5 = 1.5. Busse and Daniels?, using cathode rays, report a yield of one molecule of ozone per I j electron volts; this is equivalent to a minimum ratio of 11,'s = I . Hunt3, using the glow discharge, obtained as a lower limit 11)K = 0.03. These ratios are all fairly high and show that decomposition has been reduced to a low value. I n the earlier papers of this series4a technique has been developed for the synthesis of ammonia and of nitrogen dioxide in the glow discharge wherein only the for-xard reaction is in evidence, the back reaction having been eliminat,ed by freezing out the products with liquid air as fast as formed. This technique has the further advantage of enabling a distinction to be made b e h e e n the relative reactivity of positive ions, neutral atoms, and excited molecules. It was thought advisable, therefore, to apply the method to the synthesis of ozone. d comparison of the relative efficiencies of 0 + 2 ions and atoms in forming ozone is very pertinent at this time in view of the interest centered on the presence of ozone in the upper atmosphere. Experimental The apparatus and method of procedure were identical t o those described previously, except that a double 11eLeod gauge was used which enabled Fertilizer and Fixed Kitrogen Investigations, Bureau of Chemistry and Soils, U. S. DPpartment of Agriculture. J. rlm. Chem. Soc.. 51, zijI (19291. (J. .4m. Chem. Soc, 5 0 . 3271 f1928!. J. Am. Chem. Soc., 51, 30 (19291. J. Phys. Chem., 33, 883 (19~9:; 34, 1j9,0000 11930).
CHEMICAL A C T I O S IK THE GLOW DISCHARGE
1281
pressures t o be read accurately up to z cm. of mercury. I t was necessary to modify the discharge tube so that the positive column could be studied; the tube used is illustrated in Fig. I . The procedure was as follows: After the system has been well evacuated the desired amounts of gases were admitted and trapped by a mercury cut off. Liquid air was raised on the tube and, with the discharge maintained a t constant current, readings of thr hIcLeod gauge and electrostatic voltmeter were made a t intervals of I,'Z minute until the pressure showed no further decrease. The time-pressure and timeroltage curres were plotted immediately after each run, and, except for occasional errors in readings, the points were found to lie accurately on a smooth curve. Synthesis in the Negative Glow In all the experiments performed to date it has been impossible to freeze out ozone in the negative glow of a discharge. This is true for discharge tubes in which the positive column is absent, i.e., with electrodes less than 3 cm. apart, as well as for those tubes in which the positive column is uncooled, only the negative glow being immersed in liquid 2ir. The inability t o synthesize ozone in the negative glow is probably due to the fact that in this region ozone is decomposed as fast as formed. Indeed this was shown to FIG.I be a possible explanation by first freezing Discharge Tube for Ozone Formation out ozone on the anode and surrounding walls and then reversing the polarity. Kot only did the pressure rise showing a n ozone decomposition, but a visual observation showed all ozone to be removed from the walls to a distance of about 3 centimeters from the cathode. Gas phase decomposition of appreciable amounts of ozone was found to occur almost instantaneously upon the starting of the discharge. Synthesis in the Positive Column The Dtscharge Tube. The synthesis of ozone takes place readily in the positive column of B discharge tube immersed in liquid air. The ozone is deposited on the walls of the tube as a blue liquid film. Since the vapor pressure of ozone a t liquid air temperature is in the neighborhood of 0.2 mm., synthesis can not be carried below this pressure. Ozone is rapidly decomposed in the discharge a t room temperature. By allow4ng the liquid ozone to evaporate into the discharge, therefore, the same oxygen can be used for an extended series of runs. Apparently the con-
A. KEITH BREWER AND J. W. WESTHAVER
1282
centration of ozone in the uncooled discharge is very low since no observable contamination of the mercury in the 1lcLeod gauge or mercury cut-off could be detected when this procedure was followed. A large variety of discharge tubes was tried in studying the factors influencing the mechanism. The results in all cases were concordant, indicating that the rate of synthesis was proportional t o the length of column immersed and increased somewhat with the bore of the tube. The results presented in this paper were obtained with the tube illustrated in Fig. I . /40
/20
too
40
20
FIG. 2
The Effect of Pressure. In Fig. z are given the data obtained from a typical run in oxygen. The oxygen was purified by a series of fractional distillations of liquefied tank oxygen and also by calcining potassium permanganate; the results were identical in either case. The voltages recorded are for the fall of potential across the immersed positive column. It will be noted that for constant current the linear time-pressure relationship previously reported in connection with the synthesis of ammonia and nitrogen dioxide does not hold rigorously for the synthesis of ozone. These slight variations in rate with pressure were the cause of a large amount of experimentation. The decrease in rate above I O mm. results largely, if not entirely, from a temperature effect due t o the slow warming up of the gas and
CHEMICAL ACTION I N THE GLOW DISCHARGE
1283
the electrodes. The slowing down in the rate commencing a t 2 . 5 mm. and culminating in a sharp break a t 1.5 mm. has been traced to a slight decomposition of the ozone film covering the walls. The break in the curve a t 1.5 mm. resulting in a sudden increase in rate is concomitant with a rise in potential drop across the column. Like the slowing down between 2.j and 1.5 mm., it is also connected with the ozone film on the walls, since both are entirely absent when there is no ozone film covering the walls. This was shown by starting the synthesis a t an initial
Weft Input
tu Poarfive Column
FIG.3
pressure of z mm. It was at first thought that the break might be related t o the Kirkby effect5 but the voltage characteristics of this pressure range at room temperature contained no such break. It was later found that the potential drop across the column could be increased by several fold if a very small amount of ozone was allowed t o evaporate into the discharge. For a constant current the increased potential drop occasioned by the sudden evaporation of a trace of ozone into the discharge causes an increase in the power input t o the positive column, a greater part of which is expended in producing additional ionization. The action of evaporated ozone, therefore, is t o produce a regenerative or auto-catalytic effect on the rate of synthesis.
Rate of Synthesis and Power Input. The relationship between rate and power input is illustrated in Fig. 3. The points on these lines were computed from a series of curves similar t o 6PM. Mag., 2, 913 (1926).
1284
A . KEITH BREWER AND J. W. WESTHAVER
Fig z 1 but for different currents The difficulty in folloming the reaction for the higher power inputs accounts for the observed scattering of the points about their respective line. The results are distinctly different from those reported for ammonia and nitrogen dioxide in that the rate i s more nearly proportional to the power input than to the current Curve I represents the reaction rates measured a t a pressure of I mrn. with currents ranging from z t o I j m a. The initial oxygen pressure was 2 m m Eflecf of Nifrogen on Ozone Forrnafion
FIG.4
Higher currents could not be used in these particular experiments without danger of melting the electrodes. It will be seen from curve I that the reaction rate a t low pressures and currents is proportional to the power input in the positive column. The yield of 150 gms. per kw.hr. or one molecule per I I .9 electron volts was the maximum obtained in these experiments. This is equivalent to a minimum 31,S ratio of I . 19 ozone molecules per 0 + 2 ion. Curve z is for reaction rates measured a t 3 . j mm., and currents from 5 t o 50 m.a.; it will be noted that the yield falls off slightly for the higher currents. Curve 3 gives the yields a t I Z mm. pressure with currents from I O to 40 m.a. At this pressure the discharge could not be maintained below I O m.a., and it was also necessary to start the discharge with an induction coil. The yield is much less than for the lower pressures. It is evident from these data that, while the yield decreases materially for the higher pressures and current densities, it tends to approach a constant value a t all pressures provided the current density is sufficiently small. Since
CHEMICAL A C T I O S IS T H E GLOR' DISCHARGE
FIG.5
fffecfqf Hehum an Ozone Formahon
Time (minutes)
FIG.6
1285
1286
A. KEITH BREWER AND J. W. WESTHAVER
ozone is decomposed with extreme rapidity in the discharge the decrease in yield at high current densities is evidently due to gas phase decomposition. Such a decomposition would be particularly evident a t high pressures where the rate of diffusion of the newly formed ozone molecules to the walls is slow. Mixtures of Oxygen with Other Gases Sitrogen. A typical run to determine the effect of nitrogen on the rate of synthesis is shown in Fig. 4. The partial pressure of nitrogen for this run 'was 2 . 5 mm. Of particular interest is the sudden break in the time-pressure curve occurring a t 8 mm. d break similar to this was always observed, the pressure a t which it took place increasing with the partial pressure of nitrogen. Coincident with this break was a sharp rise in voltage, although neither pure nitrogen nor pure oxygen in the absence of an ozone layer showed any such voltage rise, nor did such a rise in potential gradient occur for any mixture of nitrogen and oxygen in the uncooled discharge. It seems probable that this break is an accentuation of the break in nitrogen-free oxygen which occurs a t about I . j mm., and is due t o the same basic cause, namely, the evaporation of ozone from the film covering the walls.
Argon. Fig. 5 shows the effect of a partial pressure of 2 mm. of argon on the rate of synthesis. No break in the time-pressure curves was noted a t any pressure or composition. The rate showed only a gradual decrease with increasing percentage of argon. Helaum. The characteristic effect of helium on the rate of synthesis is shown in Fig. 6. While helium shows no appreciable break in the time-pressure curve, this effect apparently is not completely absent as is evidenced by the break in the voltage curve occurring a t about 4.5 mm. pressure in the particular run illustrated. It is possible that this break is due t o traces of nitrogen in the helium, but this is improbable since the helium was allowed t o stand for days over active charcoal immersed in liquid air. The Effect of Percent added Gas. The data obtained in a complete series of runs similar t o those illustrated in Figs. 4, 5 , and 6, in which the composition of the mixture was changed by small intervals from pure oxygen to pure added gas, are summed up in Fig. 7. The nitrogen curve is especially interesting in that nitrogen is often referred t o as a positive catalyst in ozone formation. The results clearly show that under the condition of these experiments no such property can be attributed t o nitrogen. The results are peculiar in that the potential gradient through the positive column rises t o a maximum for a mixture containing about 5 5 per cent nitrogen. This result must be due t o the presence of the
CHEMICAL ACTION I N THE GLOW DISCHARGE
1287
ozone layer, for in the uncooled discharge no such maximum in the voltage curve could be observed, the values in all cases lying between that of pure nitrogen and pure oxygen. The voltage curve for added argon in the cooled discharge is similar to that obtained when uncooled, since all points fall between the values for pure oxygen and pure argon. The effect of argonon the yield is not distinctive, the yield decreasing nearly proportionally to the p-rcent of argon present. Yield of Ozone in the Positive Columr,
FIG.7
Helium in its voltage characteristics lies intermediate between nitrogen and argon. The yield curve is peculiar in that helium up to 40% has no retarding effect on the amount of ozone synthesized; in small amounts it even enhances the yield. In referring to the t.ime-pressure curve in Fig. 2 , it will be seen that the effect of helium on t,he reaction rate is small although the percent' helium increases with decreasing pressure. The enhancement in the yield found for small amounts of added helium a t higher pressures was found to be absent for similar compositions a t low pressures. The effect is probably an indirect one in that helium will permit a more ready diffusion of ozone t o the walls and thus decrease the gas phase decomposition which is present at, high pressures. Since the ionization potential of helium is about I O volts higher than that of oxygen, the rate of formation of 0 2 ions is not materially effected by the presence of small amuunts of helium.
Computation of Ozone Yield The total measured volume of the static system was 4.30 liters; the average temperature was estimated to be zo°C. A change in pressure of I mm. represents
1288
A . KEITH B R E W E R A S D J.
W. WESTHAVER
2 7 3 X 4.30 X 6.06 X 1 0 ) ~ = 1.42 X 760 x 2 9 3 x 2 2 . 4
IO?''
molecules of oxygen
or 2 7 3 x 4 . 3 0 x 32 = . 0 0 7 5 3 grams of oxygen. 760 X 293 X 22.4
Denoting the rate of decrease of pressure in niillinieters per minute by R, the current in amperes by I, and the voltage drop through the positive column by V, the yield of ozone in grams ppr kw. hr. is
Y = ,00753 X 6 0 X
103
X It '11- = 451 K ' I V grams per kw. hr.
The number of ozone molecules formed per electron of current is
11
=
2
3 X 1.42 X
= ,252
IO~O
X
I
60 X r.jo X
10-l~
X R 'I
R I niolecules of ozone per electron.
The current as measured by a niilliameter is assumed to be the electronic current within the positive colunm, since the velocity of the positive ions as compared to electrons is negligible. The rate of reaction can be found from the slope of the time-pressure curve for constant current; 11 may be calculated by substituting in the above expression. The energy necessary to form one ozone molecule in the discharge is
11- x I IT v = - = ____ = 3.97 - electron volts per molecule. hI ,252 X H H Little is known about the actual number of ions formed by an electron in traversing the positive colunin of a discharge, but an upper limit for this value niay be found by dividing the fall of potential through the positive column by the ionization potential of the gas. For oxygen, the maximum number of ions that can be formed per electron is
Y I,,, = -ions per electron. 14.2
From this the mininiuni 11 ' S ratio is
It is apparent that the foregoing computations involve no assumptions as to the mechanism of ionization or of the resultant chemical action Table I illustrates the varioub value* of Y, 11,K,,,, (AI S'),,,, and v obtained a t different pressures from the data in Fig. 2
TABLE I Pressure (mm.)
R
I
,30 1.4j
,020
.58 1.60
,020
I
0.5
0.48
,020
13 9
I
j
I
,020
,020
v
Y
440 420
66.8
3j0 31j 215
;21
Nm..
101.0
16.4 18.3 19.9
31 .O 29.6 24.6
114.0
20.2
22.2
j8.0
j0.0
6.0
1j.2
v
(M/N)mio
26.8
,528
22.9
,618
18.5
,809
1j.6 35.8
,910 ,395
1289
CHEMICAL ACTIOS I N THE G L O W DISCHARGE
The table shows that as the pressure decreases the yield and (31 X)min increase to a maximum at I mm. At this pressure the energy expended in producing one ozone molecule is I 5.6 electron volts, an amount slightly greater than the ionization potential of oxygen. Line I of Fig. 3, however, represents a higher yield than could be obtained in the type of experiment recorded in the table. In this case a t low initial pressures and for low current densities an efficiency of 11.9 electron volts per ozone molecule was obtained, which is equivalent to (31!S)nlin= 1.19. I
Discussion of Results Khile conditions in the positive column in general may be said to resemble those in the negative glow, a very marked difference arises in virtue of the fact that the potential gradient through which an electron falls before entering the negative glow from the Crookes dark space is several hundred times that to be found along the positive column. For this reason the method employed in the prevous papers of this series for computing the number of positive ions formed in the region wherein synthesis takes place cannot be applied here. The effect of this marked difference in potential gradients causes the present results to differ in that the rate is not quite independent of the pressure, in that the yield in terms of electron volts is higher, and, especially, in that the rate is proportional to the power input instead of the current. Some insight into the mechanism involved in the reaction process may be had from a study of the physics of the discharge. As in the negative glow, so within the positive column, are to be found excited molecules, atoms, positive ions, and, in this case, an appreciable concentration of negative ions as well as electrons. Indeed the energy input, or electron energy, of the positive column may be considered as expended in the following ways. I. In elastic collision 2. In molecular excitation 3. In dissociation into atoms 4. In ionization Unfortunately none of these expenditures can be evaluated accurately; nevertheless it is possible to make some reasonable assumptions as to their values. It is definite that in the end all the energy consumed appears as heat or as chemical energy in the ozone formation. A lower limit for the energy lost in elastic impacts may be computed by assuming the fractional loss per collision to be nm/M, where m and 31 are the masses of the electron and molecule respectively. Such a computation involves the assumption that there is no affinity between an electron and a neutral molecule, which is not justifiable for oxygen. The values of 10% a t 8.0 mm., and about 1 7at~1.0mm., thus computed, are low by several fold. The energy expended in excitation and dissociation may be approximated by the intensities of the characteristic spectra. Johnson6 and Locknow? have 6
Proc. Roy. SOC., 105.4, 683 (19241. Astrophys. J., 63,Zoj (1926).
1290
A . KEITH BREWER A S D J. W. WESTHAVER
shown that excitation in the glow discharge operated on D.C. or 60 cycle A.C., for pure oxygen gives a series of bands in the positive column know: as the first negative system, extending from 4 8 j o b in the visible to z313A in the ultra-violet. Steubing8 has further shown the presence of a serjes of bands in the extreme ultra-violet covering the region from 1831&1919A, which are found in both the positive column and negative glow and also in absorption. This is generally referred to as the ozone-forming region. I n addition to these bands the “sezond negative band system’’ consisting of bands ranging from 68 j 3 A to 4 9 5 5 8 is excited in the high frequencydischarge and occurs also with the atomic excitation in the negative glow, especially a t pressures below z mm. The intensity of the emission decreases with increasing gas pressure due t o the increased loss of energy in elastic impacts, as was pointed out in the preceding paragraph. The intensities of the various bands have been shown by Johnson to be fairly unjformly distributed over the entire quartz violet region. The high intensity of these bands indicates that the energy consumed in excitation constitutes an appreciable fraction of the total energy input. Crew and Hulburtg have shown that dissociation in oxygen is inappreciable above about 0.5 mm., and further that the dissociation is nearly independent of the power input. An attempt was made t,o confirm these results by measuring the change in pressure other than that due to temperature. Since the discharge filled but a portion of the tube, only the order of magnitude of the dissociation could be measured. The percent dissociation a t pressures above I mm. for D.C. and A.C. currents was low; the dissociation was considerably higher for the high frequency discharge and appeared t o involve a large fraction of the molecules in the path of the discharge. The amount of energy expended in dissociation a t pressures above I mm. with the D.C. discharge used in t,he ozone experiments, however, cannot amount to more than a few percent. of the entire energy input. Since the voltage drop and ionization potential of oxygen are both known, an upper limit can be given for the number of positive ions formed per electron of current, this number being obtained by dividing the potential drop through the column by the ionization potential of oxygen. An exact determination, however, cannot be made, since the energy lost in non-ionizing impacts is indefinite. The actual number of positive ions formed per electron, in view of the above considerations, may fall below this upper limit by as much as 50 per cent. The distribution of ions in the positive column is determined by the potential gradients. Since the gradient is uniform throughout the length of the tube, except for local variations in the case of striations, the production of ions per unit length is constant. This accounts for the observed fact that the rate of chemical action is proportional to the length of column immersed. 8
Ann. Physik, (4), 33, 553 (1910). Phys. Rev., (2), 30, 124 (1927)
CHEMICAL ACTION I N THE GLOW DISCHARGE
1291
The concentration of positive and negative ions per unit volume is also constant except for a very thin layer adjacent t o the walls. The negative ions because of their greater mobility will diffuse to the walls more readily than will the positive, building up a negative charge on the walls sufficient to counteract the difference in mobility. The neutralization of ions on the walls of the tube increases as the bore of the tube is diminished. The increased yield observed for tubes of large bore indicates that ions thus neutralized may not contribute t o ozone formation. The rate of ion formation is also very nearly independent of the pressure over a comparatively wide range. This is a result of the fact that while the potential gradient is a direct function of the pressure, the electronic mobility is such a decreasing function of the pressure that the rate of production of positive ions remains quite constant, thus accounting for the linearity in the time-pressure relationship. I t is obvious from the foregoing discussion that the synthesis of ozone, like that of ammonia and nitrogen dioxide, is due largely if not entirely to positive ions. Since atoms and radiation in the extreme ultra-violet are present in the discharge the possibility exists that both atomic and photochemical mechanisms are also in evidence. Fortunately it is possible to give an idea of the relative efficiencies of these various processes. If neutral atoms contributed to the observed yield the effect must be small a t pressures above I mm., due to the low atomic concentration. The fact that the rate of synthesis fell rapidly to zero in the region where the atomic concentration was increasing indicates that neutral atoms are comparatively inactive in the reaction process. The design of the apparatus was such that a direct test could be made for t h e amount of ozone produced by radiation. This was accomplished by placing the liquid air in the usual position and passing the discharge between the electrodes in the upper portion of the tube in such a manner that the radiation passed directly into the cooled region. Various combinations of electrodes were tried in which the effect of radiation from the positive column, from the negative glow, and from the entire discharge were tested. D.C., 60 cycle, and high frequency discharges were used. I n no instance could a detectable amount of ozone be frozen out when the discharge itself did not take place in the portion of the tube immersed in liquid air. In these tests the intensity of radiation entering the cooled region must have been as great as that produced under normal operating conditions. It may be concluded, therefore, that if ozone is formed by the radiation produced in the discharge, the efficiency of this process is \'cry low, not of the same order of magnitude as that for positive ions. A determination of the ratio of ozone molecules formed to the number of 0 $ ions reacting involves a probable maximum uncertainty of about two-fold. The possible variation in the number of positive ions formed per electron of current has been discussed. There are two other uncertainties, namely, the loss of ions t o processes other than the formation of ozone, and the decomposition of ozone whether in the gas phase or on the walls. The latter of
I292
A. KEITH BREWER A S D J. W. WESTHAVER
these uncertainties has been shown to be negligible at low pressures and current densities. The former, however, must amount to several percent, for it includes t'hose ions which contribute to the positive ion current as well as the ions which diffuse to the walls or become neutralized without inducing ozone formation. In referring to the table it will be seen that the lower limit for the ratio M / N is close to I , whereas somewhat larger values than I are obtained under more favorable conditions. The value of N used assumes that the entire energy input goes into the production of ions, which, as has been pointed out, is necessarily incorrect, yielding too large a value possibly by as much as 50 per cent. Making reasonable assumptions as to the energy dissipated in processes other than ozone formation a value is obtained of the order of hf/K = 2 . Since it is impossible to obtain ratios between I and z by any simple mechanism, this seems the most probable value. The ratio M/S = 1.j obtained by Lind is in line with the contention that the true value is nearer z than I . These considerations are in reasonable agreement with the results of J. B. Johnsonlo who has shown that the energy lost in the ionization process is appreciable; the number of ions produced by a I O O volt electron, for instance, is two and one-half observed, and four calculated. This correction when applied t o the results of Busse and Daniels? would yield an M i x ratio of about, 2 , while the ratio obtained by Hunt3 would be raised to nearly I . In view of the results presented in this paper the most probable mechanism for ozone formation should not involve atoms nor excited moleculm and should have a yield given by XX, ?; = 2 . Also the high rate demands simplicity. These demands can best be met by a cluster mechanism. A possible mechanism, therefore, may be written:
ot For gas phase neutralization (203)+
+
2 0 2
=
(203)+
+ 0-?
= 203
For wall neutralization (z03)t
+e =
+ 02
203
-In alternative mechanisrii is
+
(O,)+ + - 0 - 2 = 2 0 3 In advancing a cluster mechanism it is necessary to assume that in cases where chemical action is possible the stability of a cluster is greater than would be calculated by the ,J. J. Thomsonl' method, wherein ions and molecules are treated as charged and uncharged conducting spheres. Since some time is required for the necessary clusters to build up, the probability of premature wall neutralization increases for decreasing tube bore, and for low total or partial pressures of oxygen.
0:
In 1'
0 2
=
Phys. Rev., ( 2 ) 10,609 (191j). J. J. Thomson: Phil. Mag., 47, 337 (1924).
CHEMICAL ACTION Ih’ THE GLOW DISCHARGE
Summary The rate of synthesis of ozone in the positive column of the glow discharge a t constant current is nearly independent of pressure from 1 5 t o 0.3 mm., but is proportional t o the power input for low pressures and current densities. The amount of ozone synthesized maximum efficiency, is proportional to the number of positive ions formed, the discharge yielding, a t one ozone molecule per I 1.9 electron volts or 150 grams per kw. hr. The minimum M/Y ratio is thus greater than one molecule per ion. No detectable synthesis can be attributed to neutral oxygen atoms or t o radiation produced in the discharge. The effect of foreign gases is t o decrease the yield. Xitrogen is most pronounced in this respect and shows no positive catalytic activity whatsoever. Argon decreases the yield nearly in proportion to the amount present. Helium in small percentages slightly enhances the jield.
