1468
INDUSTRIAL AND ENGINEERING CHEMISTRY
When activity and selectivity are simultaneously tested for significance, the same factors are found effective but effect of factor E is now least doubtful, while effects of factors F and G are on the boundary between significance and nonsignificance. The power of multiple balance in experimental design has been illustrated by an experiment of 32 runs on effects of eight preparation variables on performance of catalysts. Each conclusion m-aa as precise as if all 32 runs had been devoted t o testing that conclusion alone. ACKN0WLEDGB.IENT
The authors are indebted to A. W. McIGnnej- for technical assistance in the preparation of this paper, and to the referees for several valuable suggestions. NOMENCLATURE
A , B. C---H a, b, c---h
Di D2
= factors, independent variables = upper levels of factors
= effect of changing factor level on activity
effect of changing factor level on selectivity ( E ) = matrix of effects of factor E F = observed value of ratio of two variances FZm, z ( n - l ) = observed value of ratio of two variances Piith degrees of freedom indicated F E = observed value of ratio of two variances for factor E Pa = upper critical value a t significance level 01 =
iJJ l )
.m
n TIS
sp
Vol. 46, No. 7
= sum of matrices used in evaluation of error = determinant of ( J ) = number of degrees of freedom for effect
= number of runs = number of degrees of freedom for error = = = =
observed correlation coefficient Observed standard deviation for observed standard deviation for selectivity ratio of ~ J toI I J + REFERENCES
(1) Anderson, T. W., unpublished lecture notes. (2) Bartlett, hf. S., J . Roy. Statistical SOC.(Supplement), B9, 176
(1947).
(3) Brownlee, K. A,, “Industrial Experimentation,” 3rd American ed., Chap. 11, New York, Chemical Publishing Co., 1950. (4) Brownlee, K. A., Kelly, B. K., and Loraine, P. K., Biometnka, 35. 268 (1948). (5) Davies, O.’L., and Hay, W. A., Biometrics, 6 , No. 3 , 233 (1950). ~ c 291 s , (1945). (6) Finney, D. J., Ann. E u Q ~ ~ 12, (7) Finney, D. J., J . Agr. Sci., 36, 184 (1946). (8) Fisher, R. A., “Design of Experiments,” 4th ed., London,
Oliver and Boyd, 1947. (9) Hotelling, H., Ann. Math. Statistics. 2 , 360 (1931). (10) milks, S. S., Biometrika, 24, 471 (193%).
RECEIVED for review November 5 , 1953. ACCEPTED X a r c h 9, 1954. Presented before the Division of Industrial and Engineering Chemistry, Symposium on Statistics in the Design of Experiments, a t t h e 124th Meeting of the A h r E R I C A h . C H E M I C A L SOCIETY, Chicago, 111.
ixation of Nitro en in a Crossed Sy’ILLIA11 S . PARTRIDGE, RANSQ3I B. PARLIN, AND BRUNO J. ZKOLISSKI1 D e p a r t m e n t of C h e m i s t r y , University of Utah, Salt Lake C i t y , U t a h
1118 study arose in connection with an extensive investigation on the induction of chemical changes in a high frequency arc established betu-een a pair of metal electrodes. Through the specific advantages of various kinds of electrical discharges in bringing about a chemical change in systems characterized by high energies of activation and/or instability of the final products-e.g., endothermic processes-are well k n o m , little can be said as to whether ions, molecular or atomic fragments, or some combination of these are involved in the detailed mechanism. An investigation of the formation of nitric oxide when air is subjected to the action of such an arc in the frequency range of from 1 to 10 megacycles was initiated, and the present discussion forme a part of this general study. I n a recent series of patents, Cotton ( 2 ) disclosed the use of a new kind of electrical discharge for the fixation of nitrogen in air. Employing a discharge tube in a.hich four metal electrodes are symmetrically arranged in a plane, one pair of opposing electrodes was connected to a high voltage (ea. 2000 volts), 60-cycle power source; the other pair of electrodes was supplied with a radio-frequency energy source in the range of 1 to 10 me. This arrangement permitted him simultaneously to establish a high frequency and a low frequency discharge in the same spatial region. Essentially, Cotton’s claim has been that this superposition of a high and a low frequency discharge permits operation of a discharge in a relatively high pressure range (100 to 700 mm. of mercury), which a t the same time exhibits all the visible characteristics of a true glow discharge with none of the usual disadvantages of an arc-namely] high currents and high temperat ures. Under these favorable conditions of operation] conversion 1
Present address, Stanford Research Institute, Stanford, Calif.
yields tvere obtained for the formation of nitric oxide greatly in excess of those normally encountered in the Birkeland-Eyde process, Of especial interest is the appearance of critical frcquencies for the high frequency power which increased the yields by factors of from 2 to 10. Some of these frequencies appeared to be characteristic of the reaction mixture only, while others \yere characteristics of the electrode material. The former were designated as “critical reaction frequencies” and the latter as “critical electrode frequencies.” I n addition, a best or critical pressure of operation was found a t a value of 335 mm. of mercury. APPARATUS AND PROCEDURE
For the preparation and handling of the air csed in the study, a conventional flow system was employed.
L2
Figure 1. Apparatus
July 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY I
12
I
I
I
I
-
I
I
I
-
FLOW RATE: 7 CC/ SEC FREQUENCY. 2 6 M C
__
ELECTRODE GAP* 6 MM
\
\
\
'\
_.
4 0 0
IO
20
30
40
% LOW Figure 2.
50
60
70
80
90
The normal procedure followed was to establish the high f r e quency discharge a t a suitable pressure and flow rate and then apply the low frequency power by adjusting transformer TI. This resulted in a luminescent discharge which pervaded the entire volume circumscribed by the four electrode tips, with the a p pearance of a glowing flat disk. The discharge exhibited more of the characteristics of an arc than a low pressure discharge: A copper-constantan thermocouple immersed in a mercury well built into the side of the discharge tube indicated temperatures in the range of 100' to 350" C., varying directly with the pressure and power used. It has not been found possible to operate the present equipment in such a manner as to obtain a discharge of the type described by Cotton.
00
FREQUENCY POWER
Effect of Varying Ratio of Power Frequency
1469
'*
1
I
I
I
I
I
Jo
6dO
I
FLOW ELECTRODE RATE'7GAP: C C / S6 E CM U
,/*
FREQUENCY: 3.4 M C
A spherical borosilicate glass discharge tube ( 125-ml. capacity) was equip ed with four conical electrodes arranged symmetrically in a &an,. The electrodes, approximately 22 mm. in length, were fashioned from electrol tic copper and tapered to a sharp point (solid angle, 18"). Tge large end of the electrodes was drilled and threaded to fit 9-32 steel bolts, which in turn were silver soldered to tungsten wire 0.040 inch in diameter; the latter was sealed into a male standard-taper ground-glass joint. This arrangement permitted variation of the electrode gap distance. As a source of radio-frequency power, a 150-watt BC-375-E U. S. Army transmitter was used, capable of supplying frequencies in the range of 1.5 to 12.5 mc., voltages up to 1000 volts, and a maximum current of 0.5 ampere. One electrode of the discharge tube was directly connected to the antenna lead of the transmitter and the opposing electrode was grounded. No further coupling arrangements were attempted. To measure the high-frequency power, two radio-frequency probes, PI and PZ in Figure 1, were designed and constructed which rectified the high frequency current and the resulting direct curient measured with a vacuum tube voltmeter, VTVM. One probe, Pz,was so placed as to measure the voltage drop across a 100-ohm, noninductive resistance, R1, connected in series with the grounded side of the discharge tube. The details are shown in Figure 1. An oscilloscopewas used to measure the power factor (cos e). Under the conditions maintained in this investigation, the power factor was found to be close to unity: In no case was its value less than 0.96. The energy yields as reported in grams of nitric acid per kilowatt-hour are for all practical purposes the same as yields per kilovolt-ampere-hour. For the low frequency discharge a 60-cycle power source was employed as shown in Figure 1. Two radio-frequency chokes, LI and Lz,blocked the high frequency current out of the low frequency circuit. Standard alternating current meters were used to measure the power (current times voltage) dissipated in the 60-cycle circuit. TZrepresents a center-tapped power transformer capable of delivering 500 to 2400 volts. Fine adjustment of the voltage was obtained by the use of a variable autotransformer, TI, in the primary of
Tz, Although it is theoretically possible t o have several different oxides of nitrogen as well as ozone formed in air (free of water and carbon dioxide) subjected to an electrical discharge, detectable quantities of only nitric oxide and nitrogen dioxide were found. These results are in accord with those of Willey (3) and Briner ( 1 ) . The total acidic component of the exit gases (nitric oxide and nitrogen dioxide) was determined by trapping a known volume (250 ml.) in an excess of standard base and back-titrating with standard acid. Earlier studies showed that the oxides of nitrogen readily attack the ordinary Apiezon stopcock grease and this was replaced with silicone grease on all ground-glass connections. The flow rate was measured a t 25' C. and 640 mm. of mercury and not corrected to standard temperature and pressure,
4 1
/ P
14
d
4
0 ' Od
LOO
lA0
LO
PRESSURE,
4bo MU
'
Effect of Pressure on Yield
Figure 3.
The efficiency of the process may be represented in various ways, two of which are considered here. Taking the standard heat of formation for nitric oxide as 21.6 kcal. per mole, the percentage of the theoretical yield, based on energy consumption, is given as XI
= 3.97
x
10-2 y
%
where y is the observed yield in grams of nitric acid per kilowatthour. On the basis of the utilization of the oxygen present in the entering air, the percentage of the theoretical yield becomes
Xz
= 15.5 p y / F P
%
where p = apparent power, watts (summed for both high and low frequency discharges) F = flow rate, cubic centimeters per second P = pressure, mm. of mercury Three typical runs with the present apparatus are given, together with the corresponding percentage yields, in Table I. RESULTS AND DISCUSSION
Over the range of conditions maintained in this study, the crossed discharge enhances the energy yield of nitric oxide over that obtained from the use of high frequency power alone. The effect of varying the ratio of high frequency to low frequency power is shown in Figure 2, where the energy yield reported as grams of nitric acid per kilowatt-hour is plotted against per cent of low frequency power a t two different pressures. RIaxima in
INDUSTRIAL AND ENGINEERING CHEMISTRY
1470
TABLE I. DATAFROX TYPICAL RUNS Run
Pressure, hIm.
470
100
448
300
456
236
Voltage 465 420 480 540 460
Amp.
0.100 0.012 0.140 0.050 0.010 0.086
540
F l o d r a t e , 7 oo./sec.
e
Current,
Yield, G./ Frequency Kw.-Hr. XI, % 2.6mc. 2.1 0.083 60 cycles 3.4mc. 9,4 0.373 60 cycles 2.6mc. 8.4 0.334 60 cycles
Xz, % 2.40
6.65
4.04
Gap distance, 6 mm.
PRESSURE; 235 ‘LOW
RATE;
MU
~CC/SEC
ELECTRODE GAP; 6
YM
*t0
Figure 4.
Vol. 46, No. 7
function of pressure is shown in Figure 3. For both kinds of discharges, the yields are almoet directly proportional to the pressures, with no indications that a critical pressure for maximum yield exists for either kind of discharge. €Ion-ever, for the limited pressure range studied, in which a stable crossed discharge could be maintained, the latter is far more effective. The higher yields may possibly be due to a larger effective volume for the discharge plasma. I n employing the crossed discharge in the oxidation of nitrogen, Cotton observed almost resonancelike peaks in the yield-frequency curves. Six of these peaks were independent of the electrode material and characteristic of gases only, and are called critical reaction frequencies. Eight other frequencies were found to be characteristic of the electrode material only, in the present case copper, and are called critical electrode frequencies. Thesc characteristic frequencies were observed in the frequency range of from 0.48 to 16 6 me., though insufficient data are given t o report half-widths. The band spread of these critical peaks calculated a t the base line has an average deviation of about 15%. For this study, a frequency range of from 2.1 to 4.6 n as chosen, which embraced four of the critical electrode frequencies for copper a t values of 2.5, 2.8, 3.1, and 1.15 me., as u-ell as a critical reaction frequency for the oxidation of nitrogen at 4.25 me. The trials were carried out a t a constant pressure of 235 mm.. a flow rate of 7.0 cc. per second (25” C., 640 mm. of mercury), and an electrode gap distance of 6 mm. Yields were obtained for thc. crossed discharge as well as for the pure high frequency discharge. The results observed are shown in Figures 4 and 5. The energy yields in grams of nitric acid per kiloxatt-hour are independent of the frequency Tithin the experimental error, with no evidence of the existence of critical frequencies for either kind of a discharge
i 26
28 FREOJENCY,
30
32
MC
Effect of Frequency on Yield
the yields were obtained a t 50 and 2074, respectively, for the two curves a t pressures of 100 and 235 mm. of mercury. I n this series of runs, the flow was kept a t a constant value of 7.0 cc. per second (25’ C., 640 nun. of mercury) and the high frequency at 2.6 me. The reproducibility of the data is shown in the curve for 100 mm. of mercury. It is usually of the order of 5%> although for a series of successive runs under identical conditions the average deviation was found t o be 0.2%. I t was not possible to extend the measurements beyond a 50% contribution of the low frequency power. Excessive overheating and rapid oxidation of the copper electrodes occurred, interfering with the maintenance of a stable discharge. Similar difficulties were encountered when the measurements were extended to higher air pressures. Above pressures of 300 mm. of mercury, the 60-cycle discharge had a tendency to arc across to the low frequency electrodes and also to the grounded high frequency electrode. Assuming a monotonic variation of the yield with the percentage of low frequency power (beyond the maximum), extrapolation of the curve for 235 mm. of mercury gives a value of approximately 3 grams of nitric acid per kilowatt-hour for the pure 60cycle discharge. This is to be compared with a yield of 6.1 grams of nitric acid per kilowatt-hour for a pure high frequency discharge, In contrast, Cotton obtained maximum yields for a unit ratio of high frequency input a t a pressure of 335 mm. A11 yields obtained here are a factor of from 5 to 15 smaller in magnitude than those reported by Cotton. The effectiveness of a crossed discharge and a pure high frequency discharge in the conversion of air into nitric oxide as a
42 44 FREQUENCY, Y C
Figure 5.
48
Effect of Critical Electrode Frequency on Yield
Cotton also reports an enhancement in his yields whenever a critical electrode frequency is choeen that overlaps a critical reaction frequency, thus giving rise to a “compound” peak in the curve of yield us. frequency, The results are shown in Figure 5 , where a critical electrode frequency should exist a t a value of 4.15 mc. and a critical reaction frequency a t a value of 4.25 me. Considering an average deviation of 15% for the band spread-Le., Av 0.6 mc.-of each critical yield peak, strong overlapping
July 1954
INDUSTRIAL AND ENGINEERING CHEMISTRY
should take place. However, for both the pure high frequency and the crossed discharge, the yields were independent of the frequency within the experimental error. LITERATURE CITED
(1) Briner, E., Helv. Chim. Acta, 19, 287, 308, 320 (1936). (2) C o t t o n , W. J., Trans. Electrochem. SOC.,P r e p r i n t s (1946); U. S.
P a t e n t s , 2,468,173-4-5-6-7, 2,485,476-7-84, a n d 2,485,480-1 (1946) : Chem. Eng., 54, 252 (September 1947).
1471
(3) Willey, E. J. B., Proc. Roy. SOC.(London),A127, 511 (1930). RECEIVED for review May 19, 1953. ACCEPTED March 13,19:4. Presented before Section 14, Physical and Inorganic Chemistry, at the X I I t h International Congress of Pure and Applied Chemistry, New Tork, N. Y.,September 1951. Abstract from a dissertation submitted by W. 8. Partridge in June 1951 to the Graduate School of the University of Utah in partial fulfillment of the requirements for the degree of doctor of philosophy. Work supported by a grant from the University of Utah Research Fund and the Atomic Energy Commission.
Viscosity of Nitric Oxide-Nitrogen Dioxide System in Liquid Phase H. H. REAMER, G. N. RICHTER, AND B. H. SAGE Calvornia Znstitute of Technology, Pasadena, Calif.
N
0 EXPERIMEKTSL data for the viscosity of the nitric
oxide-nitrogen dioxide system were found. Pure nitrogen dioxide in the liquid phase was investigated by Thorpe and Rodger (60). Scheuer ( 1 7 ) studied the viscosity of this pure compound, but his results differ markedly from the measurements of Thorpe and Rodger. Measurements of the viscosity of pure nitrogen dioxide in the liquid phase were made a t pressures up to 5000 pounds per square inch in the temperature interval between 4O0and280' F. (10). The volumetric and phask behavior of mixtures of nitric oxide and nitrogen dioxide has been described (19). These data extend to pressures in excess of 5000 pounds per square inch a t temperatures from 40' t o 340' F. and serve as the basis for the volumetric corrections required to determine the absolute viscosity of this binary system. They are in reasonable agreement with the measurements of Purcell and Cheesman ( 7 ) for temperatures a t which the two investigations may be compared. Baume and Robert (1) also studied the phase behavior of the nitric oxidenitrogen dioxide system a t temperatures below 68" F., and Wittorf ( 6 2 ) determined the limits of solubility of nitric oxide in nitrogen dioxide. The effect of pressure and temperature upon the specific volume of nitrogen dioxide was investigated (9, 18), and a review of the available data for this compound was presented.
square inch or 0.2%, whichever wa8 the larger measure of uncertainty. Hydrodynamic characteristics of rolling ball viscometers were and Hubbard investigated by Watson ( $ I ) , Hersey and Shore (j), and Brown (6) and have been considered in the application of this instrument (2). The apparatus was calibrated with n-pentane in the liquid phase using the critically chosen values of Rossini (11) for the viscosity of this compound a t atmospheric pressure. .4n equation of the following form (6)was used to estabIish the viscosity from the measured roll time: 7 =
ef
Ae(aB - a/) - e
The coefficients A and B were determined from the measured roll times with n-pentane as a function of temperature and were
I
I I
I-----l
METHODS AND APPARATUS
The present measurements were made with a rolling ball viecometer of a type proposed by Flowers ( 3 ) and developed by Hersey (4, 5 ) . The instrument employed was described in connection with measurements of the viscosity of ammonia ( 2 ) . This equipment involved a stainless steel tube inclined at an angle of approximately 15" down which a closely fitting steel ball was permitted to roll. The time of traverse of the ball between two sets of three coils located near the ends of the tube was determined electronically. A centrifugal pump was employed to return the ball to the upper end of the tube and to bring the BYEtem to a, uniform composition and temperature. The exit from the roll tube n-as closed during measurements of the time of traverse of the ball down the tube. Roll times were measured with a probable error of 0.2% and the temperature of the roll tube was known, with respect to the international platinum scale, within 0.1" F. Pressures were determined by use of a balance (1.4)calibrated against the vapor pressure of carbon dioxide a t the ice point. The pressure of the fluid within the instrument was established through a balanced aneroid-type diaphragm (18) and was known within 2 pounds per
I 1000 2000 30CO 4000 PRESSURE, POUNDS PER SOUARE INCH
Figure 1. Viscosity of the Liquid Phase of a Mixture Containing 0.920 Weight Fraction Nitrogen Dioxide
'
