402
ANALYTICAL CHEMISTRY
level the method described seems to be precise. The results obtained on two samples of commercial titanium are also shown in this table. These happened t o be low in boron. As a check on the precision and accuracy of the method, several recovery experiments were carried out by adding known amounts of boric acid to samples of low-boron titanium (Brand 11). Total boron was then determined by the procedure as described (Table 11). At the 0.05 and 0.10% levels the recoveries are good. At the 0.005% level the relative error is rather large, but even here the results should be adequate for most practical purposes. The probable accuracy of the method is *0.002% for samples containing 0.5 to 0.1% boron.
indicated quantity of the elements as in the color development step of the procedure and reading against a reagent blank as the reference solution. The vanadium-carminic acid complex does not exhibit such a sharp absorption peak when read against water or sulfuric acid, but the shift in the absorption curve as measured against the blank produces the spectrum shown. This system does not follow Beer's law, and in practice the interference of vanadium becomes negligible when less than 0.3% is present in the sample. Larger quantities may be tolerated if the vanadium content is not more than ten times the boron content. Vanadium is not retained a t all by the ion exchange resin under the conditions specified, so some other method of removing it would be necessary if too much were present.
INTERFERING ELEMENTS
Most of the common elements in reasonable quantities are known t o cause no interference in the carminic acid method for boron (2). I n amounts that are likely to occur in titanium alloys, the follon-ing do not interfere, as shown by tests in this laboratory: chromium, aluminum, zirconium, nickel, iron, tungsten, columbium, tantalum, and silicon. Vanadium represents a possible serious interference, which apparently has not been reported previously. The absorption spectra of Figure 1 illustrate the magnitude of the interference a t various wave lengths, when 10 times as much vanadium as boron is present. These curves were obtained by treating the
LITERATURE CITED
(1) Babko, A . K., Volkova, A. I., Zhitr. Obshchel Khim. 21, 1949-56 (1951). 22, 567 (1950). (2) Hatcher, J. T., Wilcox, L. V.,ANAL.CHEM. (3) Hoffman, J. I., Lundell, G. E. F., J . Research Natl. Bur. Standards 22, 465 (1939). (4) Latimer, W. M., Hildebrand, J. H., "Reference Book of Inorganic Chemistry," 317, LIacmillan, New York, 1940.
(5) Newstead, E. G., ANAL.CHEY.(in press). (6) Norwitz, G., Codell, M., Anal. Chim. Acta 1 1 , 233 (1954). (7) Smith, W. C., Goudie, A. J., Sivertson, J. N., ANAL. CHEM.27, 295 (1955). RBCEIVED for review July 28. 1955. Accepted December 4, 1955.
Moisture Permeability of Glove Materials for Control led-Atmosphe re Boxes J. H. ROWAN
Y-72Plant, Union Carbide Nuclear Co., Oak Ridge, Tenn. The moisture permeability of glove materials has been found to be the limiting factor in the maintenance of dry atmosphere boxes. A comparison of available materials for this use has shown the superiority of butyl rubber over all other materials tested.
A
TTEMPTS to maintain extremely dry atmospheres in glove boxes for handling moisture-sensitive materials have met with repeated failure in spite of rigorous sealing techniques and free use of desiccants. Esperience has indicated that a major path by which moisture enters such boxes is through the gloves, which form a significant part of the inside area of a glove box (on the order of 3000 sq. cm. per glove). I n t h e work reported here, the moisture permeability of samples from commercially available gloves, and of other materials which showed promise or were of interest, was measured. 4 search of available literature revealed that the moisture permeability of natural rubber films is dependent upon many factors, such as the age of purified latex before vulcanization (3), the kind and amount of filler used (4),plasticization and calendering (S), and vulcanization conditions ( 1 ) . From the work of Corwin and Karr ( d ) , who demonstrated the superiority of butyl rubber tubing in equipment for organic carbon-hydrogen analyses, it was concluded that this type of material might also be a superior material for gloves. The work of Thomas and coworkers ( 5 ) also suggested the superiority of butyl rubber. A s no butyl rubber gloves were commercially available, sample sheets with thicknesses equivalent to other glove materials were obtained for testing. Other natural and synthetic materials were also tested for comparison.
APPARATUS AND PROCEDURE
Three conditione of relative humidity and temperature were selected for the permeability tests: 24" C. and 58% relative humidity (corresponding to normal conditions in a glove when the box is not in use), 24' C. and 100% humidity (corresponding to the atmosphere on the exposed side of the glove just after use), and 37" C. and 100% humidity (corresponding to the conditions within a glove when it is actually in use), All materials were subjected to the first two sets of conditions, and materials of special interest were further subjected to the last set of conditions. The apparatus was, in part, modeled after similar equipment used by Florin (4).The atmosphere box and sample holder are shown diagrammatically in Figure 1 as they were arranged for elevated temperature and humidity conditions. For measurements a t 24" C. and elevated humidity, the heat lamp was disconnected, but the water tray (with cellulose sponge maze to increase evaporation surface area) was left in place and the atmosphere box was kept closed. At 24" C. and .58% relative humidity (normal laboratory conditions) the water tray was removed, and one side of the atmosphere box was opened to the laboratory, so that room air could be circulated over the sample by an external fan. The temperature and humidity within the atmosphere boy were observed with a wet-dry bulb thermometer. Normal laboratory conditions were also monitored with the aid of a recording Hythergraph. Throughout each test the temperature in the box was observed to be constant within fl o C., while the relative humidity a t 58% was constant within &2%. It was not possible to maintain exactly 100% relative humidity, because the atmosphere box was neither sealed nor insulated. The test sample consisted of a disk 25 cm. in diameter, of which a test area of 346 sq. cm. was exposed to the atmosphere in the box. Each sample Bras smoothed over the face of the sample holder, and was held in place by the Teflon O-ring and the retaining ring, shown in Figure 1. Precautions were taken to prevent leaks caused by channeling under the edge of the sample. The sample was under no tension, as indicated by a tendency to balloon slightly when argon was swept through the system.
403
V O L U M E 28, NO. 3, M A R C H 1 9 5 6 Table I. Permeability of Materials Testeda Total HzO Perme% ability, Relative Total colTemp., HumidTime, lected, y/Hr./ Sq. Cm. Hr. hfg. O C. ity 24 58 112.7 13.4 0.34 24 100 112.5 19.1 0.49 37 100 46.7 24.9 1.54 2 0.74 1.06 24 58 43.7 7.4 0.49 24 100 39.8 8.6 0.62 37 100 39.9 23.6 1.71 3 1.09 1.23 24 58 47.7 49.8 4.8 24 100 49.7 103.0 6.3 37 100 23.7 177.0 21.6 4 0.64 1.21 24 58 23.6 50.4 6.2 24 100 23.9 99.7 12.1 37 100 24.8 506.9 59.1 5 0.74 1.25 24 8.8 58 39.7 121.6 162.7 16.6 24 100 28.3 8.1 6 0.43 1.32 24 58 8 7 24.2 190.1 17.1 24 100 32.2 7 0.64 1.14 24 10.2 58 63.7 224.6 24 100 87.6 638 9 20.8 a Order of analysis was 5 4 6 , 3 7, 2 1. All room temperature studies were made first, followed by' eievateh tedperature studies. b 1. Butyl rubber sheeting N o . 6948 45-50 Duro. 2. Butyl rubber sheeting N o . 6935' 3&35 Duro. 3. Synthetic rubber A, sample cui from elbow-length gauntlet glove designed for handling corrosive materials. 4. Synthetic rubber A, sample cut from shoulder-length gauntlet glove designed for use in glove boxes. 5. Natural cloth-backed rubber white sample cut from shoulder-length gauntlet glove designed for handling borrosive materials. 6 . Synthetic rubber B, shoulder-length gauntlet glove with universal hand designed for glove box. 7. Synthetic, rubber C. sample ,cut from elbow-length gauntlet glove designed for handling corrosive materials.
Thickh m p l e ness, N0.b Mm. 1 0.81
Density, G./Cc. 1.21
ATMOSPHERE BOX WITH THERMOMETER AND FAN
7
i
SAMPLE HOLDER
4 TO WEIGHING ' I TUBES AND WET
TEST FLOW M E T E R
quantitative measure of the water content of gases whose moisture content approaches or is less than that of the desiccant. Willard and Smith (6) found the residual water a t equilibrium over anhydrous magnesium perchlorate a t 23" C. to be 5 X 10-4 mg. per liter, and over magnesium perchlorate trihydrate to be 2 X 10-3 mg. per liter. As the tank argon, as available in thie laboratory, was found to be so consistently dry that the dew point was below the estimated linit of sensitivity of an Alnor Dewpointer (130" A., corresponding LO a moisture content of less than 1 X 10-6 mg. per liter), the gas was "preconditioned" by passing through two 25-cm. towers packed with magnesium perchlorate followed by a cold trap of dry ice in acetone. By this procedure the moisture content of the argon sweep gas n-as increased to a level which gave a positive blank in the weighing tubes. Continuous use of the preconditioning train resulted in a gradual depletion of the available moisture in the preconditioning train. At the start of the experiment the blank was + l . 6 mg. in 65 hours with freshly filled towers; after the towers had been used for 73 days, the blank was -2.4 mg. in 23 hours. These blank values are insignificant for the purpose of relative comparison. However, as the body temperature studies (37" C.) were made after all room temperature studies (24' C.) had been completed, it is probable that the 3 i " C. permeability tests were affected by negative blanks. For these reasons the tabulated results have not been corrected for the presence or absence of residual moisture in the sweep gas. Consequently, the results are not absolute, although they provide an adequate basis for relative comparison. Experimental results along with descriptions of the materials sampled and analyzed are given in Table I. Experimentally determined thickness and density values are provided to give a more complete description. Tabulated thicknesses are average values for measurements made with micrometer calipers at various places on the sample tested. Density values were calculated from the mean thicknesses and weights of test areas. Time of analysis and weight of water collected are accumulated totals of all determinations for each sample. Individual determinations varied from 4 to 65 hours. The accumulated totals were used to calculate permeabilities because they were thought to yield a more valid measure. CONCLU SlON S
DL
PRE-CONDITIONED A R G O N SUPPLY
H 2 0 TRAY WITH SPONGE M A Z E
H E A T L A M P - CONTROLLED BY THERMOREGULATOR
Figure 1. Permeability apparatus Desired temperature and humidity conditions were established in the atmosphere box before the sample holder was mounted and connected into the system. "Preconditioned" argon was then alloR-ed to flow through the system for several hours prior to determination. The permeability of a sample was measured by determining the gain in weight of magnesium perchloratecharged U-tubes for a timed period and metered argon flow. The flow rate during flushing and measurement was 4 to 6 liters per hour. Moisture recovery did not appear to be influenced greatly by flow rate, unless the rate was extremely low. Duplicate determinations were made for most samples; additional determinations for samples of special interest were made t o assure attainment of apparent equilibrium. Whenever possible, a determination was continued until a significantly Feighable amount of water had been collected. Blank determinations were made by bypassing the sample holder. DISCUSSION AND RESULTS
The determination of weight changes in tubes filled with "anhydrous" magnesium perchlorate cannot be considered a
The results obtained for the samples tested indicate that butyl rubber has a moisture permeability about one tenth that of the next best material tested. Xone of the parameters mentioned in the introduction was known or controlled for this experiment; consequently, the results presented should not be considered as applicable to other materials which appear similar. No correction has been considered for variations in sample thickness, although permeability would be expected to follow an inverse relationship to thickness. ACKNOWLEDGMENT
The author gratefully acknowledges the editorial assistance of R. C. McIlhenny and J. 0. Hibbits. The Anchor Packing Co. provided the sample sheets of butyl rubber, LITERATURE CIT
Bowler, W. W., Ind. Eng. Chem. 44, 787 (1952). Corwin, A . H., Karr, C., Jr.. ~ ~ N A LC . m x 20, 1116 (1948). Dalfsen, J. W. van, Arch. Rubbercultuur 25, 483 (1941). Florin, A. E., Los Alamos Scientific Laboratory, Loa Alamos, iY.M., private communication. (5) Thomas, R. M., Lightbrown, I. E., Sparks, W. J., Frolich, P. K., Murphee, E. V., I n d . Eng. Chem. 32, 1283 (1940). (6) Willard, H. H., Smith, G. W., J . A m . Chem. SOC.44, 2255 (1922). RECEIVED for review Sugust 22, 1955. Accepted November 14, 1955. Based on work performed f o r the Atomic Energy Commission by Union Carbide and Carbon Corp., Oak Ridge, Tenn.
