boiling points around 100" C., the entire evaporation apparatus is suspended in a tall form beaker over boiling water. Thp reflux action of the water warms the entire unit, and minimizes condensation of solvent in the upper chamber of the manifold. Other refluxing solvents may be used to give different temperature ranges, providing they are compatible with the grommet material. For volatile or thermolabile solutes, the apparatus van be used in a freeze drying mode, but a t the cost of longei evaporation times. After the evaporation is completed and the system brought back to atmospheric pressuie, the vial is removed and sealed with a ieptum. The vial may be weighed before and after the evaporation, to prrmit determining the net weight of the reciduo. The apparatus was tested by evaporating 0.3- to 3-mg. quantities of m-
phenetidine (either as the free base or as the hydrochloride salt) dissolved in 25 ml. of solvent. Methylene chloride was used as the solvent for the base and was evaporated in about 30 minutes. Aqueous 0.5N HC1 was used for the salt, and was evaporated in about 2 to 2.5 hours. Recoveries were determined by measurement of ultraviolet absorption. Low recoveries (less than 90%) were obtained when both types of samples were taken to complete dryness. The losses may be due to volatility of the base (b.p. 254" C.), but are attributed principally to entrainment of micro particles in the gas and vapor stream for the hydrochloride salt. Evaporation to incipient dryness-i.e., to the point where about 250 PI. of solution remains-consistently gave recoveries within the 95-1OOojO range for the salt. Recoveries for the free base under
these conditions, although significantly improved, were still somewhat low, Although higher recoveries could be obtained by using milder conditions for the evaporation (i.e., slower air flow), it was more expedient to evaporate the base onto a small quantity (40-60 mg.) of powdered sodium bisulfate added to the receiver. By thus forming a nonvolatile salt in a difficultly entrainable form, quantitative recoveries (well within +5% of target) could be obtained reproducibly and rapidly over the 0.3- to 3-my. range. ACKNOWLEDGMENT
The assistance of Peter Dockery in the glassblowing is gratefully acknowledged. LITERATURE CITED
(1) Adler, N., ANAL. CHEM. 36, 2291 (1964).
Automatic Vapor Sampling for Gas Chromatographic Analysis G. Szhkely, G. R&z, and G. Traply, Department of Physical Chemistry, Technical University of Budapest, Budapest, Hungary
sampling method for A vapor mixtures condensing a t high temperatures has been developed; it COYTISUOUB
eliminates mechanically moving parts of construction. The arrangement is especially suitable for drawing samples from reactors or other flow systems. In industrial and laboratory practice it frequently is necessary to analyze vapor mixtures condensing a t relatively high temperatures by a continuous sampling technique and to effect precise analyses using small samples. The main problem is to withdraw the sample, because the sample must be maintained above the condensing temperature. Sampling equipment for this purpose dcwibed in the literature is made of Teflon or metal, and construction involves moving parts. Handling this equipment is, therefore, not easy; moreover, methods do not lend themsel\es t o automation and can only be used to a temperature limit.
a pressure drop should be maintained a t the end of the system to keep Hz flowing through the column. At the same time, air necessary for the operation of the flame ionization detector is drawn through. A sample for analysis is obtained by closing the mercury valve for a few seconds by the timer. This causes the vapor to be sampled to pass capillary 10 toward the chromatographic column via the sampling pipe, 11. After a short time the mercury valve is opened again to let in the carrier gas which carries the sample through the column. Because in one
44.
cycle the timer switches only once, resetting is performed by the switching valve, 3, filled with an electrolytee.g., [NH4]&04 solution-and having electrodes in its vessels. When the liquid level is the same in both sides, only one contact closes. After closing the mercury valve the increasing pressure in the system before the valve drives the liquid levels apart in the manometer, thereby activating the reswitching circuit. To control the duration of the closing phase of the mercury trap, the height of the contact can be adjusted in the side of the manometer
-
-
~~
1 5.
air 7 vu6uum
6 9.
EXPERIMENTAL
Flow Design. We have solved the problem by the arrangement shown in Figure 1. I n the sampling part, 11, the pressure is slightly above atmospheric. There is excess pressure ahead of the chromatographic column, 6, when the mercury trap is open. This causes a small portion of the carrier gas to enter connection 11 via the capillary choke. As the chromatographic column has much higher flow resistance than the capillary tube,
6.
2 4 e Figure 1. 7. 2. 3. 4.
5. 6.
Apparatus for automatic vapor sampling
Container for carrier gas Differential manometer Switching manometer Timer Mercury valve Chromatographic column
Hl
7. 8.
9. 70. 7 7.
Flame ionization dector Differential manometer [for air stream) Hydrostatic pressure control Capillary as pneumatic choke Sampling connection to reactor
VOL. 38, NO. 8, JULY 1966
1097
open to atmosphere. It is necessary to avoid condensation in system parts 6, 7, 10, and 11 by keeping them a t an appropriate temperature. Apparatus. The chromatographic column was a tube 156 cm. long and 5.5 mm. in diameter. It was filled with fireclay ground to a grain size of 0.2 t o 0.3 mm. and coated with 5% polyglycol 600. Column temperahre was 64' C. and the hydrogen flow rate was 39.0 cc. per minute. The ionization detector was homemade and the signal was amplified on a pH meter (OP 401-Radelkisz). A selfbalancing potentiometer of 10-mv. full scale deflection (Type EPP 09) was used as a recorder. RESULTS
Reproducible samples are obtained for establishing quantitatively tmherelative amounts of the component (Figure 2). The figure relates to a series of measurements during which the sampling connection, 11, was flushed by mixtures of nitrogen streams which had previously been saturated by passing them through vessels containing benzene and cyclohexane a t controlled
Figure 2.
Chromatography of vapor mixture
temperature. The automatic sampling system introduced samples of these streams into the column. The actual sample size was not known because it is a function of the viscosity of the mixture passing connection 11. In most cases, however, it is not necessary to know the size of the samples because
only the relative composition is needed. By comparing with samples introduced with a syringe it is possible to estimate the automatic samples. These are approximately 0.1 to 0.5 ml. of vapor saturated a t room temperature. The method can be used without a temperature limit.
In-Place Fabrication of Fritted Glass Discs Derwood B. Bird, Biochemical Research Program, Veterans Administration Hospital, San Francisco, Calif.
has been a recurring need for T large numbers of various sized glass columns for use in chromatography and HERE
ion exchange in this laboratory. The desirability of fritted glass discs as bases for column beds and the frequently long waiting period for delivery from suppliers have indicated the value of quickly prepared, reasonably rugged, do-it-yourself columns which would satisfy an immediate need. Fritted glass discs from 3.0 mm. to 40.0 mm. have been sintered within the glass tubing. A single operation forms the disc and fuses it into position for use. PREPARATION
The materials required are borosilicate glass tubing, quartz sand, and asbestos fiber, filtering grade. Pieces of glass tubing are broken, then ground to an appropriate size in a porcelain mortar, cleaned by any efficient glassware cleaning method, and dried. Uniform sizing is best done by screening. Particles smaller than 60 mesh and larger than 120 mesh are easily fritted and produce a disc of uniform porosity. Larger or smaller particles may give fragile or fused products. 1098
ANALYTICAL CHEMISTRY
The quartz sand should approximate the size of the glass particles for more uniform faces on the finished discs. The asbestos fiber is chosen for looseness. If may be rubbed up or even partially fired to produce a soft, easily handled fiber. A firmly placed, but not extremely compacted, plug of asbestos is formed slightly below the desired level of the glass disc. This plug is nicely formed between the ends of two glass rods inserted below and above a wad of the asbestos fiber. Partially bored cork stoppers may be slipped on the rods for large discs. While the glass tubing is held in a vertical position and the lower glass rod is held against the asbestos plug, enough clean quartz sand is added to
Asbestos Fiber
Figure 1.
Diagram of packing sequence
form a layer about 5 inm. thick. The sand is leveled and packed lightly using the upper glass rod for tamping. Clean ground glass is added to the desired thickness for the disc and is leveled and packed as for the sand layer. A second quartz sand layer and finally another asbestos plug are added in the same manner as the first. After removal of the packing rods, the multi-layered section, as it appears in Figure 1, will remain in place and the disc is ready to be heated. DISCUSSION
If the packing is either too loose or too firm, an imperfect frit results. Too loose packing allows glass and sand particles to move during rotation. On the other hand, too firm packing traps air which expands the heated tubing as it softens, giving much the same result. Fritting is accomplished by heating the ground glass and sand area a t the tip of an open flame M hile slowly rotating the tubing. Small burners of the Bunsen micro type do an excellent job of fritting; a glass-working burner which produces a variable needle flame does a better job. Natural gas and air or oxygen are preferred.
