Economically, the liquid ammonia process faces the requirements of a separate facility, a relatively remote location, cooling equipment, trained operators, extra safety protection, and a disposal problem. Extended Usage. The two processes-i.e., glycol and but)rl bromide-might be used in the recovery of spent fuels. I n these events, remo.te control facilities, collection of gas spray, conservation, and storage of wash liquids would be required Ruptured fuel rods are expected to be handled in 18-foot casks. preheated to 60' to 70' C., with suitable vents and drains. I a other than fuel piece cleaning, butyl bromide has been used for the removal O F residual sodium from deep, open retorts! whereas HB-40 [a mixture of partially hydrogenated terphenyls) (2) is preferred for enclosed systems.
Acknowledgment
T h e authors thank G. H. Hayes and J. P. Mills, Jr., for the construction and operation of the equipment. literature Cited
(1) Robb, R. L., North American .Aviation, Atomics International, Canoga Park, Calif., Spec. Rept. 8314 (1963). (2) Silverman, L., Sallach, R. A , I n d . E n g . Chem. 52, 231 (1960).
RECEIVED for review November 4: 1963 ACCEPTED May 6, 1964 Research supported by Empire State .Atomic DeLelopment Associates (ESXDX), a nonprofit corporation composed of S e w Yorkbased, investor-owned utility companies : Central Hudson Gas 8;
Electric Corp., Consolidated Edison Co. of New York, Inc., Siagara Mohawk Corp., Orange and Rockland Utilities, Inc., and Rochester Gas & Electric Corp.
NITRIC OXIDE DISTILLATION PLANT FOR
ISOTO PEI SEPARATION B. B. M c l N T E E R AND R O B E R T M . P O T T E R Los Alamos Scientijc Laboratory, Cnizerstty of Calzforma, L o s Alamos. 5.M .
A plant for enriching oxygen and nitrogen isotopes b y NO distillation, consisting of a purifier unit, a staged column 575 cm. long for continuous flow, and a uniform column for final batchwise enrichment, has been built and successfully operated. Over 600 (STP) liters of enriched NO have been produced, with an average and composition of 1.5% O", 21 % 01',and 25% N". Enrichments as large as 8.3% Oi7, 98,2% 01', 93.9y0 N15have been attained. Technological developments in low temperature distillation and purification of nitric oxide! are described. Exchange reactions among the isotopic species are negligibly slow in these columns.
and cokzorkers (7. 2: 4) in 1959 announced the measurement of the vapor pressure differences of the isotopic forms of nitric oxide (SO) for the t\vo nitrogen isotopes S I 4 and Sij and the three oxygen isotopes OI6>01',and Ol8. These differences \\-ere so large that distillation of N O became attractive as a possible method for separation of these isotopes. In 1961, Clusius, Schleich, and Vecchi ( 3 ) reported the results obtained \vith a packed 'column distilling S O . -At the time of the first announcement this laboratory had been interested in a lo\+;temperature distillation project using liquid oxygen for separation of the oxygen isotopes and changed its plans to include the study of S O distillation. T h e resulting plant has produced over 600 (STP) liters of enriched NO with a n average compositioii of 1.5% 0 ' 7 : 217, Ole, and 2570 XI5 a t a rate of 4 (STP) liters per day under steady operation. T h e description of this plant and its performance is the subject of this paper.
C
LUSIUS
The Flow Scheme
T h e plant consists of a purifier column to remove the chemical impurities found in commercial S O ; a composite column of three stages, column I, which is operated in a continuous flow fashion for the primary enrichment; and a second uniform column, column 11, \vhich operates batchwise, for further enrichment.. T h e enri'ihed product may be used as nitric oxide or. more often, is made to react chemically to produce water and nitrogen or ammonia. These isotopes are used by research groups in the L,os ,4lamos Scientific Laboratory.
Purifier Flow Systems. Commercial nitric oxide contains N20. and S O 2 as impurities. These impurities have boiling points of 77', 185", and 295' K.: respectively, as compared with 121' K. for S O . T h e X2 and SO2 are thoroughly removed by freezing the raw material to 77' K., pumping thereon. and subsequently boiling off the maior body of the mix at 121' K. However. the NCO is removed by this step to only a few tenths of 1%. If subsequently enriched in the plant, this impurity would deposit as an ice: clogging the column and associated lines. T o correct this, the material is passed through an auxiliary distillation column of a few plates in order to remove the SZO. T h e following scheme for gas purification is used in this plant. A s shown in Figure 1, commercial YO is condensed in a 2-liter stainless steel trap at 77' K. and the N2 impurity fraction is quantitatively removed by pumping to a few microns pressure. T h e trap is then warmed to no mofe than 153' K. and the gas transferred to a bank of gas cylinders (S-tanks) of 220-liter volume and stored at 5 to 15 atm. Residual gas in the trap is removed, carrying away the higher boiling impurities. T h e purifier column is continuously supplied with material from the S-tanks at the rate of 600 (STP) liters per day. T h e purified S O is removed from the top of the condenser. By adjusting the condenser temperature, as described subsequently, the NO pressure in the column is adjusted to 2.0 a t m . ; this gas pressure is reduced by a pressure-regulating valve to 1.5 atm. and bled at a controlled flow rate through a throttle needle valve to column I, which has a pressure of 0.75 atm. The S20 impurity builds u p as a solid on the surfaces of the boiler. After about 2 weeks' operation the NpO deposits plug the purifier. I t is then warmed to room temperature, its contents are pumped away, and it is rechilled to resume operaS1.
VOL. 4
NO. 1
JANUARY
1965
35
MS T- I -X-
Figure 1.
TO 'COL .I
VAG
VAC
PURIFIER
Purifier flow system
6 . Ballast volume, 4 4 liters C. Packed column Cryogenic pump Electrical heater, 10 watts LN. Liquid nitrogen LO. Liquid oxygen PRV. Pressure-regulating valve R. Raw NO S. 5-tanks (51, 44 liters each s v. Solenoid valve T-1. Throttle valve TC. Thermocouple, copper-constantan VAC. Vacuum CP.
H.
Figure
Symbols as in Figure 1 M-1, M-2. Mercury manometers MS. Sampling lead to mass spectrometer manifold
tion. T h e entering line to the boiler is as large as practical to prevent premature plugging. During the process of cleaning out the purifier, the effluent gases have been studied for information regarding the nature and amounts of the impurities. A rough volumetric analysis shows a few tenths of 1y0of S2O \vith trace amounts of NO2 to have been present in the feed gas to the purifier. These constituents are identified by their boiling points and their mass spectra. T o maintain a controlled holdup in the purifier system, the boiler is equipped with a thermocouple (copper-constantan) Lvhich serves as input to a simple on-off type controller which activates a solenoid valve in the supply line of the purifier at a setting a few degrees above the boiling point of NO. it'hen the purifier starts to suffer a depletion of its N O contents, the boiler begins to warm slightly, opening the supply solenoid valve. T h e supply flo\r rate is set to exceed the withdralval rate, so that liquid reserve in the purifier begins to build up: cooling the boiler and turning off the supply solenoid valve, thus starting the cycle again. If vithdrawal is stopped, the supply flow ceases. If the withdrawal rate exceeds the available supply rate (as limited b>- the throttle valve in series with the solenoid valve) so that the thermocouple reaches an emergency temperature level, withdrawal is stopped entirely by closing an exit solenoid valve. This emergency control circuit also protects the enriching columns in the event of failure of the purifier's liquid nitrogen supply, so that the accumulated N 2 0 is not accidentally fed to them. Column I Flow System, Purified NO flows continuously to the top of column I! which has the dimensions and characteristics given in Table I. This material, depleted in the heavier isotopic species. is withdrawn after its passage across the column top. This \vithdrawal rate is controlled by the thirdstage boiler temperature in a manner similar to that used for the purifier. If the boiler becomes too warm, the withdrawal flow is interrupted. If: holvever, the boiler approaches
liquid NO temperature, the exit flow is permitted at a rate in excess of the supply rate. These floiv rates are established by the follo\ving procedure (Figure 2) : .4fter observing the column operating pressure on mercury manometer M-1, we close the column-in and columnout valves and open the bypass valve. \+-e then adjust the two throttle valves, T-1 and T-2, so as to obtain the column operating pressure in the manifold and also the desired supply flow rate to the exhaust pump. The latter measurement is made by observing the mercury manometer: M - 2 , using the known pumping speed of the exhaust pump to compute the flow rate. This properly sets the supply throttle valve, T-1. \+'e then open to the column, close the bypass valve, and adjust the exhaust throttle valve, T-2, until the desired exhaust flow rate is obtained. Before startup, it is necessary to fill the columns with boiling liquid N O and slowly remove the isotopically depleted vapor from the condenser until the third-stage boiler starts to go dry. This wets the packing thoroughly, and presents a large surface for liquid-vapor contact. il'ithout this procedure: the two phases tend to channel, resulting in fewer theoretical plates. Although this preflooding operation requires a few days, it yields an enrichment of the heavy iqotopes so that the time involved is competitive with normal operation. The previously mentioned control system automatically controls the transition from preflood to normal operation. \Vhen point 3 of column I reaches a desired level of enrichment (1 to 2yc O"), as shown by mass analysis. a steady drawoff of a few grams per day is begun. The drawoff throttle valve allo\vs this gas to flo\v from the column to a cryogenic pump (a volume maintained at liquid nitrogen temperature), or simply to a n evacuated 44-liter gas cylinder whose contents may be periodically transferred elsewhere by a cryogenic pump. Column 11 Flow System. Column I1 has a flow net\rork similar to that of column I. The column length is divided into three equal segments by four points. 11-1, 11-2: 11-3, and
Table I. Construction Data for Columns Column I -~
Inside diameter, cm. Length, cm. Size of packing. mm.O Mass of packing, grams Boilup, watts a
36
Stage 1 5.22 190.5 2.3 X 4.5 X 4.5 6810 220
2. Column I flow system
Stage 2 2.36 193 0 1.3 X 2.5 X 2.3 960 60
Packing tyjes are stainless steel Heli-Pak, Podbielniak: Inc., 317 East Ohio St., Chicago, Ill.
l&EC PROCESS DESIGN A N D DEVELOPMENT
Stage 3 1.73 191.8 1.3 X 2.5 X 2.3 510 40
Column I1 1.09 580 0.8 X 1 . 8 X 1 . 7 690 2.0
11-4 reading from top I O bottom. At these locations are feed lines, sampling lines. and thermocouples. It has similar control circuitry, using the bottom thermocouple to control the withdrawal flow from the top. The general flow scheme is to circulate enriched gas past the top and to remove product from the bottom.
VAC
'8
ATM.
75'K
Equipment and Techniques
VAPOR
Insulating Techniques. I n a n isotope separation installation such as this, requiring hundreds of theoretical plates operating a t low temperatures, it is necessary to maintain meticulously adiabatic conditions for stable control of flow rates. Only vacuum insulation and careful prevention of stray thermal radiation flux will suflice. For example, column 11, whose boiler power is 2 watts, would receive 12 watts of thermal radiation if unshielded. The method used initially was to insert within the vacuum region a cooled radiation shield which surrounded the distil1ai:ion column. Radiant energy striking this shield was removed by liquid oxygen refluxing in a helical coil of 0.95-cm. diameter copper tubing, soft-soldered to a thinwalled (0.10 cm.) brass tube of IO-cm. diameter and 732-cm. length and also to the column condenser, thereby ensuring that the temperature of the boiling fluid was nearly that of the column. Conduction by the brass produced a temperature uniformity of 0.1 '. The recent advent of aluminized Mylar film has allowed a far simpler technique to be used. Low temperature components are loosely wrapped with 20 to 30 layers of wrinkled Mylar film, so that direct thermal contact between the plastic layers is confined to point contact. This multiplicity of shiny radiation shields reduces the radiated power by a factor of 100 or more. Although the finite vapor pressure of this plastic limits the ultimate vacuum in the jacket, the low temperature surfaces act as effective cold traps and vacua of mm. H g are typical. Condenser Refrigeration. T h e maintenance of a temperature of 121 ' K . by a primary refrigerant of liquid nitrogen a t 75' K. (Los .4lamos, N. M., has a barometric pressure of only 53 cm. of Hg) can be attained in a variety of ways. Certainly, no portion of the condenser region that is exposed to NO can be permitted to be colder than 1IO' K., the freezing point of NO. An attractive method developed for this plant is the use of an intermediate refrigerant boiling at the desired temperature. This secondary refrigerant fluid must possess a triple point temperature below the primary refrigerant temperature and a critical temperature above the desired column temperature. Such a material is oxygen. At 121' K. this substance has a vapor pressure of 10.78 atm. Figure 3 shows an idealized arrangement for effecting the desired condenser temperature. T h e pool of liquid oxygen, upon receiving the heat of condensation of N O from the condenser, boils under a synthetic atmosphere of helium a t 10.78-atm. pressure and is condensed by liquid nitrogen, just as water boiling in an open reflux still is condensed at just the rate which maintains atmospheric pressure. In practice, these simple considerations are modified to soine extent as follows. Under very low power the heat may be conducted by the metal walls of the oxygen region without involving the boiling process of the oxygen. This is noticeable during startup procedure before any j Y 0 has been added to the distillation column. when the oxygen may drop well below 121' K . and approach 75' K . Auxiliary power must be supplied a t such ' times before N O is added. T h e finite condensrr efficiency of the column condenser requires that some small temperature difference exist between the S O and the boiling oxygen. Typically, this is 3' and is easily obtained by using a somewhat lower helium pressure.
I ATM.
10.78 ATM.
121' K
I
i Figure 3. Idealized arrangement for condenser temperature control
Since the oxygen pressure at 75' K . is finite although small (0.1 5 atm.), the helium may become slightly contaminated with oxygen. I n the event of failure of the liquid nitrogen supply, sizable quantities of oxygen transfer to the helium reservoir, both depleting the liquid oxygen boiler and admixing with the helium, thereby raising the total pressure. If the nitrogen supply is then replenished. the connecting line sustains a concentration gradient between nearly pure helium at the condenser region and highly impure helium in the reservoir, with diffusion being the only equalizing process. This results in a higher condenser temperature. These considerations apply only to abnormal conditions and are of no consequence during normal operation. Care should be exercised in the assumptions made regarding the volume occupied by a reservoir of vigorously boiling oxygen. Its effective density may be only half that of stagnant liquid and the increased liquid level may accentuate the metallic conduction mentioned above or even short-circuit the whole portion of the apparatus which sustains the desired temperature gradient. Thus, one should err on the side of scarcity of liquid oxygen. In practice, this simple and attractive system has performed well and is recommended despite the apparent complications of having an additional boiling fluid. T o our knowledge, this is a new technique for low temperature distillation technology. ,4s an attempt to obtain the desired temperature in another manner, we have also used a copper rod 2.54 cm. in diameter and 45.7 cm. long as a thermal resistor. With one end in liquid nitrogen and 40 watts conducted along such a rod, the desired 45' K. temperature difference is sustained. T h e warm end of this rod serves as the S O condenser, with an electrical heater to adjust its temperature. The most serious disadvantage of the system is that the massive copper block is so sluggish in its transient response that nonlinear, sustained oscillations have been observed under some circumstances. Gas Handling. Once the NO is well purified, it is only moderately corrosive. Copper tubing exposed to this gas will show only traces of copper nitrate. It is also possible to use pressure gages with phosphor bronze Bourdon tubes. Nevertheless, an effort has been made to use stainless steel for these applications. Valves have Monel bellows and Kel-F seats. .4 marked increase in corrosive action is observed at all points where exhausted NO mixes with air, forming NOg. For example, the exhaust chamber of the mechanical vacuum pump suffers serious corrosion. Manometers exhibit some slight corrosion of the mercury, which dirties the glass. Cryogenic pumps of 250- to 2000-cc. volume are of welded stainless steel and are immersed in liquid nitrogen to pump NO. which freezes on the Lvalls as a very low vapor pressure VOL. 4
NO. 1
JANUARY 1 9 6 5
37
ice. T h e inlet tube to such a volume must not be allowed to become too cold, because of the danger of plugging. Upon warming, a pressure exceeding the triple point pressure of 19 cm. of Hg is an indication of melting. Because of the health hazard of N O and itsoxidation products, the equipment is placed in well ventilated areas. Instruments for detection of NO2 are always available. SampIing Techniques. T o obtain a few (STP) cubic centimeters of NO gas for mass spectrometric analysis the columns are equipped at various points with 0.046-cm. i.d. capillary lines which conduct samples to a valve manifold located near the mass spectrometer. To obtain a sample, this line must be purged with fresh gas. Samples from the control panels are obtained by filling a previously evacuated line connected to the valve manifold.
Column Construction. Columns are made of standard size stainless steel tubing filled with the appropriate size of packing given in Table I. Cone-shaped packing supports of metal screen hold the packing in place at each end of a stage. T h e first and second stage heater interior surfaces are machined with helical channels to enhance liquid contact. These heater blocks are of copper, in order to ensure good temperature uniformity. All inlet tubes are hard-soldered into drilled holes in the columns. Thermocouples of copper-constantan are soft-soldered to the outer column surface. Although care is exerted to keep impurities from the columns, a further precaution, to avoid plugging of the drawoff and sampling lines at the bottoms of the columns, consists of inserting such lines 15 cm. or more above the bottom. thereby establishing a small settling reservoir for frozen impurities. T o avoid buckling due to thermal expansion, the columns are suspended from a major flange near the base of the liquid nitrogen Dewar. A l l associated tubing must be provided with expansion relief by coils or zigzag bends. These tubes are in the vacuum region outside the Mylar radiation shielding. Vacuum jacket sections are made of l j - c m . diameter brass tubing of 150-cm. length flanged at each end, which may be removed successively from the bottom during disassembly. All metal surfaces and packing are degreased and dried before assembly. Heaters. Boiler heat at the bottom of each stage is applied external to the copper heater blocks, so that the heater must operate in the vacuum region. This has posed some difficulties in the use of electric heaters because of the tendency of heater wires and leads to overheat in such a n environment, resulting in their burnout. Disk heaters soldered to the heater blocks have proved to be the most successful electric heaters for such use. An alternative device is the "water heater," which is a copper rod, one end of which is soldered to the 121' K. heater block and the other end maintained a t 283' K. by circulating chilled water. ,4 40-watt heater is made of such a rod 0.95 cm. in diameter and 10.2 cm. long. T h e 20-watt heater is of the same length but reduced to 0.67-cm. diameter. Their freedom from failure makes these heaters attractive. However, they require a reliable water supply. This plant is protected against a circulating water failure by a n auxiliary water supply. It is basic to this concept that once the heater is installed there remains limited flexibility in heater power adjustment. O n the other hand, there is no danger of high temperaturr damage. At present, the power input to the third stage boiler is 40 watts; at the second stage, 20 \vatts; at the first stage, 160 watts by the use of four 40-\vatt heaters. T h e boilup in such a plant is the power input to a stage plus all heaters below it. so that the third stage boilup is 40 watts; the second, 60 watts; and the first, 220 watts. Analytical Methods
Although it is possible to analyze SO directly by mass spectrometry, the six isotopic forms show at only the four primary mass numbers, 30 through 33. Instead, a gas sample at a pressure of 2 cm. of Hg is sparked for 90 seconds with a radio-frequency high voltage source. This results in a 907, dissociation into molecular nitrogen and oxygen. T h e remain38
I & E C PROCESS D E S I G N A N D D E V E L O P M E N T
ing NO is frozen out with liquid nitrogen and the mixture introduced to the spectrometer. Nitrogen isotopic molecules appear a t masses 28, 29, and 30; oxygen molecules a t 32, 33: 34, 35, and 36. Mass 34 has contributions from 0 2 ' ; as well as O'6018. T h e 0 2 ' ' contribution is calculated from 32- and 33-peak heights. T h e results of an analysis are expressed as the atomic per cent 0 l 6 : O", and 0 ' 8 in the oxygen, and the atomic per cent K1* and W5 in the nitrogen. The computations are standard for isotopic forms of a diatomic gas. This does not: of course, describe the molecular abundances of the six NO species. In the analysis of an isotopic mixture of N O there are five independent variables, which are constrained by the determination of the two independent variables of the oxygen analysis and the single independent variable of the nitrogen analysis: leaving two degrees of freedom in the molecular composition. However, when a mixture of NO molecules is in exchange equilibrium with random pairing of the S's and O's, the relation
(1) holds, where i refers to the nitrogen isotope and j to the oxygen isotope and \vhere x t j is the molecular fraction of an N O species, y t the nitrogen analysis, and c j the oxygen analysis. By measuring the mass spectrum of an unsparked NO sample and that of the dissociated nitrogen-oxygen mixture, one can tell whether the sample is in exchange equilibrium. S o measurement a t tnis laboratory during this project has ever revealed a significant deviation from exchange equilibrium in these gas samples. T h e attainment of equilibrium in the measured spectrum can result from rapid reaction in room temperature gas and does not imply that the material in the columns is in local exchange equilibrium. O n the contrary, in the body of the columns the exchange rate is so slow that the six NO species are essentially frozen as stable nonreactive molecules (see column performance). Although the mass spectrometer analyses gave good agreement between measured atomic abundances in the supply NO and reported values (0.0377, 0": 0.204% 0 ' 8 , 0.365% W) in the earliest phase of our operations, the subsequent high concentration samples have produced a significant memory problem in the mass spectrometer. Accordingly, normal S O is run regularly and the discrepancies Lvith the reported abundances are applied as corrections to the sample analyses. xi1 = YtCj
Theory for Nitric Oxide Separation
T h e single-stage equilibrium between a liquid with multicomponent mole fractions y t and vapor mole fractions X I for an ideal solution and a perfect gas vapor is given by Raoult's law as Yt
YL - PfYi - at--
xt
P*Xi
(2)
Xi
where P is the vapor pressure of the pure substance and component 1 is taken as the most volatile species-namely N'4016. Representing et = 1 e t , where e t 1962. Electric heaters had failed repeatedly, and "Lvater heaters" were then installed throughout the plant. A successful production period from May 18 to June 24. 1962, sufEced to fill all current demands. .4nother production run of column I during the first quarter of 1963 \vas made t s obtain feed material for column 11, Lvhen a serious power failure caused extensive damage to all plant equipment. During a recent run. from June 3. 1963, to October 1. 1963. column I performed \vel1 and smoothly. For performance analysis of column I. the data obtained June 19 to 22, 1962, are used. During this time a reasonably steady set of compositions was obtained a t the four sampling points, T-1. 1, 2. and 3, under a continuous dra\voff of 5.9 (STP) liters per day. These data are shown in 'rable 111. For theoretical comparison Lvith this experiment. the fourcomponent system N"0'6, N ' 4 0 1 ~X14018, ~ and S 1 j 0 1 6 is assumed. A Runge-Kutta method is used to integrate Equation 16 numerically. Data for the boilup rates, Q : are taken from Table I : Values for ~ i are j from 'Table 11. Starting Lvith the experimental point 3 data, a n integration is performed requiring 1 I O theoretical plates, where point 2 experimental concentrations are bracketed. T h e experimmtal point 2 concentrations are also used as a starting point and an integration is VOL. 4
NO. 1
JANUARY
1965
39
80 70 IO
60
5
40
50
30 20
% %
I
10
5
0.5
2 I
0 5
01
02 01
0 05
0 05 0.01
0
n ni
50
100
I75
275
"."I
0
50
I90
Figure 4.
Column I performance, June 19 to 22,
1962 Theoretical predictions o f composition from next lower ex. perimental sampling point of column
performed with the second stage boilup, until a set of theoretical (bracketing experimental point 1) concentrations is reached, requiring 140 plates. Similarly, 50 plates from experimental point 1 concentrations yields the theoretical 7 -1 concentrations. Figure 4 shoivs, then, the computed distribution of the four isotopic molecules (also listed in Table 111), \vith good agreement throughout column I . Column 11. Column I1 was flooded with normal NO, deflooded, and started with 20 watts of power on December 8> 1962. Product from column I was added during December, and finally a flow of this material was established across the top of column 11. Approximately 150 (STP) liters of product was circulated, dropping from 1.5% to 0.7%., OIi>producing a column holdup of 30 (STP) liters of 4% 0". During the first part of January the serious departure from exchange equilibrium was noted, and on January 16, 1963, an exchanger was started a t the bottom of the column. Periodic withdrawals from the bottom of the column prevented 0 1 8 species from pushing the X''01' out the top. This scheme slo\rly built LIP the 0 1 7 concentration to 8.287, a t point 11-2. However, a Point 11-4 exhibited 98.2% 0 1 s and 93.97, " 5 . power failure caused loss of performance of the column on March 31> 1963. Apparently the exchanger was performing adequately, but many more exchangers would be needed at various points along the column to achieve reasonable exchange throughout. Also during the last half of January the power of this column was reduced by raising the control temperature a t 11-4 until a final value of 2 watts was attained. This yielded substantial increases in the average 01; concentration due to the lowered holdup and the resulting increase in the number of plates from 300 to 600. Following the power failure: the entire column material was removed and measured as 34.5 (STP) liters with 4.2% 0": 27.47, 0 1 g 3 and 54.3% X15. 40
NUMBER OF PLATES
300
NUMBER OF PLATES
l&EC PROCESS DESIGN A N D DEVELOPMENT
Figure 5. Column II atomic concentration distribution, January 14, 1963 Assumption: fast exchange Theoretical curves fitted at 11-4. perimental data
Poor agreement with ex-
The theoretical treatment of column I1 consists entirely of analysis of equilibrium conditions xrithin the column by application of Equation 9. Typical are the data of January 14: 1963, shown in Table I\-. The primary object of this discussion is the role of exchange equilibrium \rithin the column. The data are treated under the t\\-o extreme assumptions of very rapid and ver)- slow isotopic exchange reaction rates. .&SUMPTION. FASTEXCH-ANGE.This assumption alloivs the atom per cents given in Table I\- to be used directly in Equation 9. Starting \rith 11-4 concentrations. the distributions a t various plates above are showm in Figure 5. Although the vertical lines displab-ing the experimental anal!-5es are free to be shifted in position (assuming different number of plates for a section). it is evident that there is no consistent location under this assumption. Furthermore! the number of plates, lvhich should be several hundreds for such a column, is considerabl>too lo\v and apparently uneven. In brief. the experimental results are inconsistent lvith this assumption ASSUMPTION^ EXCHANGE . ~ B s E ? ; T . Under this assumption \ve are squarely faced irith the analytical dilemma previously discussed. \Ve can measure atomic abundances in the column but not the abundance of the six molecular forms. I n terms of independent variables (since EzcI = 1. eic.), \ r e measure three numbers (sa)- cg, cg. and y?)whereas five (of the six h l j ) actually enter in the processes in the column. Table IV.
11-2 11-3 11-4
Column II Experimental Data, Jan. 14, 1963
3.33 3.58 1.97
5.12 31.3 74.7
26.5 55 1
62.0
~
Table V.
Predicted Distribution of Molecular Species for Column II, January 14, 1963
0 17,
AY'40 '6,
.~14017,
>Vl4Ols,
,~150 16,
Point
71
Xll, C /cT
212. 50
X13,
XZl,
70
X?Z,
11-1 11-2 11-3 11-4
304 1 194 1 104 1
97 37 66 1 4 12 28 0 322
0 60 3 15 3 12
29 51
0 566
37 14
7 25 52 23
89 06 84 01
0 0 0 1
0
~~
s'o
0 138 5 59
A-018,
"c
X?3,
70
000
0 000
041 465 404
0 030 1 794 37 56
If: however, we consider tic0 points in a column which is a t equilibrium, one distinguished from the other by the superscript 0, there are six independently measured atomic concentrations and six related unknoums: the five previous x i t s and the number of plates, n, between the t\vo points in the column. T h e equations relating these are
+
cI. = x ~ J ~ 2 1 ; ~
yf
=
xil
+ + ,xi2
j '
+ +
= ~150
~21';
= xtlO
xi2'
ri3;
j
=
+
1, 2, 3
xi?;
(1 7 )
i = 1, 2 (18)
and from Equation 8
Despite the formidable multiplicity of these relations, it has bern possible to solve for the s i j ' s , xilO's, and n (see Appendix). LVith this procedure we use the 11-3 and 11-4 data and calculate the results in Table V, shown in Figure 6. for the six molecular forms. These computed results are expressed as predicted atomic fractions and are shown in Figure 7 and Table \'I. As before, the assignment of n for the 11-2 and 11-1 points is somewhat arbitrary. T h e analytical difficulties for 0 ' 8 determination a t the 11-1 level make the discrepancy there credible, so that the over-all picture appears consistent and believeable. Another index of the degree of isotopic exchange present in the column is contained in the formation of the apparent equilibrium constants of the two exchange reactions ~ 1 4 0 1 6
Assumption: exchange absent Theoretical curves fitted a t 11-4 and 11-3 data
I
514017
I THEO.
EXP
-0 ' 8
x
017 NI5
A
-
0
+ +
~ 5 0 1 6= xi5016
=
~
0
1
8
wjoii
+ +
~ 1 4 0 1 6 ;
K, =
x73,x11,/x13x21
~ 4 0 1 6 ;
K? =
x?2xll/x12x21
For random pairing, each K assumes the value 1. In the complete absence of exchange however, the heavy-heavy combinations X1jO1*and Xl5O1' would not appear (being virtually absent from the feed gas): so that x23 = X B ? = 0 and Kl = K B = 0. Thus? accepting the values of Table V as '.observed" values of molecular abundances at various points in the column, we may form the indicated ratios and evaluate a n "observed" K , and K2 a t each point. Values near 1 indicate exchange equilibrium; values near 0. its absence. I t turns out that K I a n d Kf are independent of the sampling location in the column: and that a t all points shown in Table V the values K1 = 0.014. K2 = 0.034 obtain. This indicates that, while a small amount of exchange has occurred in the material of the colllmn, the material is far removed from exchange equilibrium, whatever be the source or location of the reaction Lvhich has occurred. ~~
Table VI. 0
IO0
200
300
NUMBER OF PLATES
Figure 7. Column I1 atomic concentration distribution, January 14, 1963 Assumption: exchange ab,ient Theoretical curves fitted a t 11-4 and 11-3 data. G o o d agreement with experimental data and approximate uniformity of plates per section
Distribution of Atomic Fractions for Column II, January 14, 1963
Point
11-2 Calcd Obsd 11-1 Calcd Obsd
017,
%rc
0'8.
%;o
3 19 3 33
5 62 5 12
0 60 0 59
0 14 0 09
VOL. 4
NO. 1
JANUARY
A75,
yc
25 1 26 5 1 89 2 00
1965
41
Klein, Spindel. and Stern (5) have recently shown the sensitive dependence of this exchange reaction upon thr presence of N203 as a n impurity. l h e foregoing observations are consistent with their conclusions if it is hypothesized that no impurity is present in column I1 material. 'This hypothesis is probable because the material which serves as feed to column I1 has undergone not only the initial purification by the purifier column, but a n additional purification from the small settling reservoir a t the bottom of column I (see section on column construction). Acknowledgment
.4ckno\dedgment is made of the mutual interest and the harmonious relationship \vith the concurrent project of oxygen isotope separation by steam distillation directed by John S. Drury a t Oak Ridge Sational Laboratory. Members of this lahoratory \vho contributed to the successful conclrision of this program a r e : R. D. Foxvler and E. S. Robinson, supervisors; D. E . .4rmstrong and .4.C . Briesmeister, engineering design and management; F. A . Guevara and \\'. F,. \'r'agemaii. professional services; J. G . Montoya, R . C. I'andervoort. and R . E. Pruner, operations and plant assembly; P. J. Pallone and K . \Y. Syquist, machining; and L. C . Lapatnick and h l . S.Donaldson, LASL summer students.
00 =
a3
7iz
= 713
+
722
62'
f
723
c3'
= =
'4 =
211
+
c2 = X I 2
c3 = ,713
+
f
B ,721
610
=
x22
c20
x23
c3'
= x120 = x13'
XI10
+ + f
C .YZlO
x220
xz2 =
bvhere D
=
x230et2an,'D
2xilOecL~"
la.
eflinxl10
2a.
e*l?nvl20
3a.
eelantl30
+ ef2inxa10 = al + a2
lb.
xllo
2b.
x120
a3
3b.
.xl30
ec22nxg20
+
=
ef:1n.t2~0 =
'7ij
+ x210 = clo + = + = ,t2.'0
c20
,t230
c30
Any corresponding pair of these equations-e.g.: l a and 1bcomprises a set of t\vo simultaneous linear algebraic equations in the variables A l t o and . A Z $ , the solutions of which are x I f 0 = (1 fi)(ctoef'i" - a t ) xpiO = (1 f t ) ( a f - cloeflJ)
where =
f i
ec?,n
-
The application of the relations of set C yields the qil and completes the solution of the limited problem, assuming D and n to be kno\vn. To determine D and n \ve use the two new relations ?I
+
= xll
=
f
211'
x12 x12'
f
XI3
f
X13'
where y1 and 7 1 0 are the measured h'l4 atomic abundances a t the t\vo points. The solutiohs involve straightforlvard although complicated algebra and are simplified by the general rela tion €21 = f l i
+
€1
Lvhere e l is a separation factor for a n N1j species from a comparable "4 species. T h e results are that n must be found to
[I
involve the known quantities c i , c f o .cij: and the unknown quantities x i j 2 x i l o . T h e number of plates, n. is to be treated tentatively as kr.own, even though its evaluation is deferred to a later step in this development. T h e primary difficulty in solving these equations is that the presence of the xilo's in D causes set C to be nonlinear. Ho\vever, this set C may be transformed to a linear set by defining a set of ne\v unknowns
=
B
A
xll = xilOe"ln:D x12 = x120e(12n,'D
x23'
xl20ef12n
Substituting set C into set A yields
?lo
c1
qll =
~ 2 2 0 x23'
723
Appendix
I n the determination of mole fractions of NO species in a column, the relations
+ +
x1Zo
XU0
-
(1 - e-"'))y2][1
- (1 -
e-fl")ylO]
=
(1 -
x
satisfy the relation by successive approximation methods. In this relation y2 = 1 - yl. When a suitable n is found, D is computed from 1
r
l
1
T h e a i dre then formed, the xilo determined, and the computed. completing the solution of the problem.
= DXil
so that the e q u a t i o x of set C become
XijOew
yl,
Literature Cited
Each relation in set A is also to be multiplied by D , and the quantities ciD are to be regarded as knotvn constants, ai
(1) Clusius. K.. Schleich. K.. HelL. Chim.d c t a 41. 1342 (19581 (2) Clusius; K.: Schleich. K:, \'ecchi, M.. 16id..' 42, 2624 (f959). ( 3 ) Ibid., 44, 343 (1961).
(4) Clusius, K., Vecchi. M., Fischer. A , , Piesbergen, U . . Ibid., 42, 1975 ( l 9 5 9 ) , (5) Klein, F. S., Spindel. FV., Stern. M. J., J . Chim. Phys. 60, 148
c ~ D
,,r,,Q\
T h e determination of both n and D has thus been deferred until later. !$.irh these transformarions the problem is reduced to the set A a,
42
= qI1
l&EC
+ qll
B clo
= xllo
+
C q l l = xl,cef!ln
PROCESS D E S I G N A N D DEVELOPMEN1
\l-/UJ).
(6) Narten, A , J . C h f m . Phys. 34, 2198 (1961)
RECEIVED for review December 6, 1963 A C C E P T E D April 2'. 1964 b'ork performed under the auspices of the U.S. Atomic Energy Commission.