Gas-Blending Apparatus Donald N. Hanson and Arturo Maimoni, University of California Radiation Laboratory, Berkeley, Calif.
A
capable of preparing synthetic gas mixtures of any composition with an accuracy to about 0.01% was required for the calibration of gas analytical techniques. Although a number of designs for such equipment have been reported, none seemed to have the required accuracy (1, 3-6). N APPARATUS
The apparatus devised was permanently mounted in a n aluminum frame with 3111 bedding compound (Minnesota Mining and Manufacturing Co.). The apparatus has two calibrated burets in which the pure components are measured. Each buret acts as one leg of a mercury manometer, the other leg being open to the atmosphere. After the pressures and volumes of the individual gases are dctermined, the burets are interconnected for blending. The diameters of the tubes in the different sections of the burets were chosen to minimize the percentage error in the PV product. If a standard reading error of 0.1 mm. is assumed, the diameter of the tubing is selected so that the percentage error in reading the volume is about the same as that in reading the pressure, taking into consideration the additional source of error introduced by the capillary correction. This point is discussed in greater detail by Cook ( 2 ) .
is closed and the system evacuated for admission of the second constituent into buret 1. The mercury level in 1 is adjusted to obtain the proper volume ratio between the two gases. After the pressure in the system is below 5 microns, as shoxn by a vacuum thermocouple gage in the vacuum-system manifold, the second gas may be admitted to buret 1. Stopcock B is nom closed and the system is evacuated. Once the pressure in the system is below 5 microns, stopcocks E and H are opened, letting the mercury rise in the intermediate tube until it is above stopcock C, which is noT7 closed. The mercury level is allowed to rise until it reaches point K . At this point the blending apparatus is charged and ready for determination of the pressure and volume of the gases in the burets This is done by connecting each buret to the manometer tube, by opening stopcocks G and D or F . For certain volume and pressure ratios it is possible to open the three stopcocks simultaneously and adjust the amount of mercury in the system so that the mercury meniscus in each
buret is in a calibrated section. For the apparatus described, mixtures containing about 6, 17, 41, and 52% of either component could be read in thls fashion. For other mixtures, the determination could not be carried out simultaneously. The blending apparatus is disconnected from the vacuum system and immersed in a well-stirred water bath, designed specifically for this purpose and provided with long plate-glass windows, to allow accurate reading of the mercury levels. The mercury levels and the position of the fiduciary mark are read with a cathetometer to the nearest 0.1 mm.; the apparatus was always carefully leveled before a set of readings mas taken. The atmospheric pressure and temperature of the water bath are recorded and the blending apparatus is ready to be connected to the vacuum system for transfer of the blend to the apparatus to be calibrated. Once the blending apparatus is connected, stopcocks G and either D or F are closed, A and B are opened, and
-
Gorl-le'
k4
L
CALIBRATION
The volume calibration of the burets was obtained by mighing thr mercury drained from the appropriate sections. To avoid gas bubbles, the apparatus was filled lvith mercury after the burets had been evacuated to less than 10 microns There is a small volume uncertainty, due to small amounts of stopcock grease from stopcocks A and B that may extrude and adhere to the top sections of the burets; for this reason these stopcocks n.ere lubricated with a minimal amount of grease. Successive volume calibrations were internally consistent. and the volume uncertainty was below what would produce a 0.01% error in composition. BLENDING
The apparatus is connected to the vacuum-system manifold and pumped out before admission of the gases t o be blended. The gas to be present in highest concentration in the final mixture is the one admitted to buret 2 . The mercury level in 2 is adjusted in the calibrated section above stopcock D , and the first gas is admitted. The pressure of the gas is adjusted to be very close to atmospheric. Stopcock A 158
ANALYTICAL CHEMISTRY
I
i
' I
1
Monometer
\
c
BdQET 2
the gas is transferred back and forth betwcm burcte 1 and 2 for mixing. The amount of time dcvoted to this operation is important. As much as 40 minutes was required for proper mixing and using shorter mixing times resulted in inconsistent results. This is probably the main drawback of this piece of equipment; however, it can be improved appreciably by including another bulb, of about 100-cc. capacity, helow the calibrated section of buret 1, to allow complete transfer of the gases from one buret to the other and to increase the amount of miyinp; in any given transfer. After the gases are properly blended, the mercury level in the central tube is lowered helow the mercury cutoff, and the blend is introduced into the vacuum-system manifold by opening stopcock C. From the values of pressure and volume of the gases the value of the PV product was calculated, by use of the ideal-gas law. The temperature of thc water bath was noted to ascertain the mercury-density correction
factor, but was not used to reduce the PV product to a standard temperature, because both gases are a t the same temperature and the per cent correction would be the same for both gases. The number of moles of each gas was computed from the PV product and the value of the actual molar volume of the gas. For the gases used in this workhydrogen and nitrogen-the above calculation procedure is justified, because the deviations from idcal gas behavior would introduce errors smaller than 0.01% in the final molar composition of the blend. The pressure readings were corrected for the capillarity effects of the tubes of the burets. The inside diameter of the manometer tube, 18 mm., was chosen to eliminate the need for any correction. The hollow bore of stopcocks A and B was partially filled with sealing wax, to eliminate the dead volume and the possibility of having an unmixed pocket of gas. For the same
reason the mercury level in the central tube is raised to point K . This dcsign of blending apparatus, in which the gases are measured in two independent burets, performed very satisfactorily after the effect of mixing time was ascertained and allowed for. It is believed that the average error in the composition of a blend was of thc order of 0.01'36, as evidenced by the data obtained in the calibration of an optical interferometer. LITERATURE CITED
(1) Bus~y, R. H., Barthauer,
G. L.,
Metier, A. V., IND.Esa. CHEM.,ANAL.
ED. 18,407 (1946).
(2) Cook, M. W., "Solubili;y of Hydrogen m Ron-Polar Solvents, Univ. Cali-
fornia Radiation Laboratory, Rept. UCRL-2459 (January 1054). (3) Langor, A,, Rev. Sn'. Inslr. 18, 101
Constant-Volume Fraction Collector for Column Chromatography
W. J. Wechter,' J. E. McCarty, and 5. E. Fisher, Department of Chemistry, University of California, 10s Angeles 24, Calif. XPERIENCE
with automatic fraction
E collectors has shown that a collector
which measures the eluate directly is preferable. Volumes measured by time flom (5, 6) will decrease appreciably as the resistance to flow increases with gradual packing of the column. The flow rate is also sensitive to temperature changes, and volumes measured hy time flow may vary as much as 50% owing to the decrease in density and viscosity attended by a corresponding temperature increase. Volumes measured by drop count (2, 8) are subject t o three major difficulties: (1) Changes in concentration of the developing buffer will result in variations as great a6 30% ( 4 ) ; (2) temperature variations will affcct drop size and lead to inequities in the volumes collected; and (3) the column must be run slowly, so that individual drops may he counted by the photoelectric system. Fractions collccted by weight (7) will vary invcrscly as the density of the eluate. The density can change considerably during the course of a chromatogram. Although siphon-type collectors (4, 6) overcome all of the ahove difficulties, they are subject t o others. As these collecton retain eluent in their siphon tubes, there is partial mixing of successive fractions. Such receivers 1
are subject t o considerable losses by evaporation, except for the special siphon apparatus designed by Mader (6). Dutton and Castle ( I ) overcame these difficulties, in the design of their simple collector, which wm actuated by a photoelectrir volume device controlling a solenoid lifted iron ball valve. This equipment was designed for use with the Technicon fraction collector for volumes of from 5 to 20 ml. This automatic f r y t i o n collector collrets accurate volumes from 7 to 500 ml. (if the column is flowing at such a rate that thc eluate delivered during the time the dump valve is open is not a significant fraction of the volume of cluntc collected). As the receiver flask
Figure 1. Representative receiver flasks 135. loo-. and 500-ml.)
. .
is glass (Figure 1) and the valve impermeable to low surface tension solv e n t s i . e . , n-pentane-virtually any solvent employed in chromatography can be employed. The machine provides automatic changing of sample flasks by volume actuation, and an automatic shutoff after the last sample. This apparatus combines the virtues of loiv price (about 5180 for parts and materials exclusive of fabrication time), a wide range of adaptability with respect to fraction size, constint volume fractions, and good separation of the eluate from consecutive fractions. OPERATION
When the machine (see F i y r e 2)
~
I
Present address, The Upjohn Co.,
Kalamaaoo, Mich.
VOL. 31, NO. 1, JANUARY 1959
159
