Standardization and Stability of 0.1 N Sodium Thiosulfate Solutions in Hot Weather SIGURD 0. RUE, Chemical Research Laboratory, Ethyl Corporation, Detroit, Mich.
S
EVERAL investigators have shown that, by properly
bicarbonate, as in the Willard and Furman method ( I g ) , is therefore believed to be unnecessary. Incidental experiments mere carried out to determine whether a shift in the starch-iodine end point occurred as the temperature was increased to 40" C. The effect proved to be negligible, certainly not more than equivalent to 0.02 ml. of thiosulfate solution. The proposed procedure for thiosulfate standardization followed directly from the above experiments.
choosing experimental conditions, various standards for 0.1 N sodium thiosulfate solutions may give results agreeing within 0.1 per cent ( I S ) . Of these standards, potassiuni dichromate is one of the best because of its ease of purification and the stability of its solutions. The usual standardization methods even for this reagent, however, become unreliable a t the high summer temperatures prevalent in many localities (11,18). The problem is further complicated by the possible instability of the thiosulfate solutions themselves (7). Various methods of standardizing and stabilizing 0.1 N sodium thiosulfate solutions have been systematically studied under hot weather conditions. I n the course of the investigation a dichromate procedure was developed which has given highly accurate results over a temperature range of 23" to 40" C.
PROPOSED DICHRONATE METHOD. Accurately weigh 0.2 to 0.22 gram of purified potassium dichromate into a 500-ml. Erlen-
TABLEI. EFFECTOF TEMPERATURE ON THE VOSBURGHMETHOD Reaction Temperature rJ
c.
Titration Temperature
c.
SatS1034
s
23 0.10165 23 25 0.10169 40 0.10182 40 25 0.10192 40 40 0 Each value represents average of at least 3 determinations: normalities are corrected to 23' C.
meyer flask. Carefullv wash down any dichromate adhering to the funnel or sides of the flask. Dilute to 125 ml. with distilled water, and add 5 grams of potassium iodide, low in iodates. Rotate the flask until the iodide has dissolved. Add 5 ml. of 6 N hydrochloric acid with constant swirling. Carefully wash down the sides of the flask, so that a layer of water is formed on top of the solution. Stopper the flask and set the sample in the dark for about 6 minutes. Add 165 ml. of distilled water. Titrate with 0.1 N sodium thiosulfate solution until the solution becomes greenish-yellow, rotating the flask so that no excess of sodium thiosulfate is in contact with the acidic solution at any time. Add 0.6 per cent starch solution, using from 2 ml. at 20" to 3 ml. a t 40" C. Wash down the sides of the flask, and continue the titration t o the end point. For a corresponding volumetric procedure: add about 45 ml. of 0.1 N potassium dichromate solution, accurately measured, to the Erlenmeyer flask, and proceed as described above. To determine the accuracy of the dichromate procedures, it appeared desirable to compare them with an exact method involving the use of pure iodine. The iodine was weighed in a mixture of potassium iodide and water, as suggested by Treadwell and Hall (16). It was found that weighing bottles containing mixtures of 2.5 grams of the iodide with 0.5 to 0.8 mi. of water lost weight a t the same rate when the stopper was removed, whether iodine had been added to the solution or not. Hence, losses resulting from the vaporization of iodine are believed to be negligible. I n order to determine the quantity of water vapor lost during the transfer of the crystalline iodine, weighed samples of cadmium iodide, clay plate, sodium nitrate, and iodine itself were poured into the potassium iodide-water mixtures contained in 50 x 15 mm. weighing bottles with male ground stoppers. The results showed that a positive correction of 0.20 * 0.05 mg. must be applied. The uncertainty in the correction for a 0.5-gram sample of iodine is about 0.01 per cent of the total weight. These experiments were made in a room thermostatically controlled to 23" to 25" C. and having a low relative humidity, and no claim is made for the reliability of the method under other atmospheric conditions. The procedure given below was adapted from that of Treadwell and Hall in the light of the above experiments and of the results obtained by Popoff and Xhitman ( I S ) . Under these conditions very closely reproducible results were obtained. The method was of great value in this research, but it cannot be recommended for routine work. Baker's resublimed iodine was further purified by sublimation, first from a potassium iodide mixture and then alone. A third sublimation in a current of dry nitrogen (2) did not perceptibly change the result's.
Standardization Methods The effect of temperature on the Vosburgh dichromate method (1'7) was first studied.
VOSBURGHMETHOD.Accurately weigh 0.2 to 0.22 gram of purified potassium dichromate into a 750-ml. Erlenmeyer flask. Add 100 ml. of distilled water and when the dichromate is dissolved add 5 ml. of 6 N hydrochloric acid. Add 2.5 grams of analytical grade potassium iodide crystals, stopper the flask with a clean stopper, and then rotate the flask to mix the solutions thoroughly until the iodide has dissolved. Set the flask in the dark for 5 minutes. Then add 300 ml. of distilled water, and titrate with 0.1 N sodium thiosulfate solution of known temperature until within 3 t o 5 ml. of the end point, bein careful not to rotate the flask too vigorously. Add starch an3 complete the titration. Reaction and titration temperatures were varied separately by adding distilled water of the proper temperature at convenient points of the procedure. If a high reaction temperature was desired, the flask was placed in a n oven set at that temperature during the dichromate-iodide reaction. The results summarized in Table I show that the reaction temperature had little effect on the final result, although an increase in temperature doubtless accelerated the reaction. On the other hand, a n increase in temperature of the solutions during the titration always resulted in higher normalities, indicating a loss of iodine by volatilization. It was found that this loss could be largely avoided, even a t 40" C., by increasing the potassium iodide concentration a t the beginning of the titration from 0.5 to 1.4 per cent. Likewise, reversing the order of addition of acid and the iodide should minimize a possible initial loss of iodine which otherwise might occur before the reagents are completely mixed. The possibility of the iodide ion undergoing atmospheric oxidation was considered. Results obtained by carrying out the reaction at 40" C. in a n atmosphere of nitrogen did not differ significantly from those obtained in air. The use of
IODINEMETHOD.Add 0.7 ml. of distilled water directly to 2.5 grams of potassium iodide in a 50 X 15 mm. weighing bottle
(male ground stopper) by means of a 1-ml. pipet. Let the bottle remain open for a few minutes to permit the evaporation of any 802
ANALYTICAL EDITION
October 15, 1942
803
calculated for 23" C., taking account of the expansion of both water and glass. Baker's analyWillard tical grade potassium dichromate was further and purified for these determinations by thrice reVosburgh Iodine Furman Iodine Proposed Iodine Method Method ?rIethod" Method Method Method crystallizing and fusing. 0.10080 0.10092 0.10091 0.10117 0,10086 0.10078 It is evident from the data in Table I1 that 0.10093 0.10087 0.10096 0.10113 0.10082 0.10079 0.10086 0,10086 0.10101 0.10113 0 10080 0.1008T satisfactory results may be obtained a t moderate 0,10088 0,10093 0.10097 0.10113 0.10078 0,1007T temperatures (20" to 25" C.) by any of the 0,10118 0.10117 0.10068 0.10082 0.10088 0.10074 0.10084 0.10085 0,10111 0.10117 0.10076 0.10086 dichromate procedures under consideration. 0,10085 ..... 0.10120 . . . . . 0.10074 . . . 0 . ioogn ..... . . . . . . . . . . 0.10068 Results obtained a t 40" C. by the proposed 0.10084 0.10080 ..... .......... method (Table 111) do not differ significantly 0.10081 ..... 0.10098 o:ioiO5 n:ioii5 0,10078 0:ioos:! Average 0,10087 0.10086 from the true value, but those obtained by the 0.00018 0.00019 0.00029 0.00004 0.00017 o . oooin Range 4.2 11.5 2.2 6.1 Standard deviation X 10' 5 . 6 6.8 Vosburgh and the Willard and Furman methods -0. 00004 Difference from iodine value - 0.00010 + 0 . 00001 are high by 0.45 per cent and 0.20 per cent, rea Each value represents average of two determinations made with same dichromate soluspectively, and are not very consistent. Cooltion. Standard deviation of t h e two values in each pair, a measure of t h e titration errors, was found t o be 4.8 X 10-6. ing the solution before titrating (Table IV) is unimportant in the proposed dichromate method e v e n a t 30" t o 40" C.,^provided that the starch TABLE111. STANDARDIZATION OF THIOSULFATE SOLUTION AT solution is reasonably sensitive. The proposed method is 40" C. simpler than the Killard and Furman method and is adapted (Normalities corrected t o 23' C.) Willard to the use of individually weighed samples as well as the and Iodine more rapid volumetric procedure. The method is therefore Vosburgh Furman Proposed Methodb, Method Method5 Method 23' C recommended for the standardization of 0.1 N thiosulfate 0.10164 0.10145 0.10120 solutions, particularly a t temperatures of 30" C. and higher, 0.10160 0,10129 0.10117 ..... 0.101i2 ..... 0.10126 0.10122 0.10138 ..... 0.10138 0.10120 OF THIOSULFATE SOLUTION AT 23" C. TABLE11. STLUDARDIZATIOS
I
. . I
. . t . .
0.10174 0.10155 0.10156 0.10160 Average 0.00036 Range Standard deviation X 10' 12.1 Difference from iodine value + O . 00045
0.10135
..... .....
0.10135
0.10124 0,10122 0.10126 0.10122 0.00009 2.9 0.00007
..... . . .
0:ioiiS
0,00004 0.00019 2.2 7.7 00020 ..... a Each value represents average of two determinations made with same dichromate solution. Standard deviation within pairs, due t o titration errors, was found t o he 6 0 X 10-6. b These d a t a , from Table 11, are included for comparison with dichromate method values obtained in hot room.
+o.
+
Lvater from the inner wall of the bottle; then stopper and weigh accurately. Add 0.4 t o 0.5 gram of purified iodine directly to the moist salt with a small porcelain s oon. Stopper the bottle, tap it gently for a few seconds until al? the iodine dissolves, and wei h. The difference in weights plus 0.20 mg. equals the weight of t i e iodine added. Loosen the stopper, and drop the weighing bottle followed by the stopper into a 500-ml. Erlenmeyer flask containing a solution of 2 rams of potassium iodide in 150 ml. of distilled water. Rotate t i e flask only enou h to submerge the weighing bottle. Titrate at once with 0.1 #sodium thiosulfate solution with no swirling at all to within 3 ml. of the end point (calculated approximately). Then slowly rotate the flask so that most of the iodine is reduced before reaching the surface. Continue the titration to a pale yellow, slowly rotating the flask. -4dd 2 ml. of 0.6 per cent starch solution, and complete the titration.
FINAL TESTS. The different dichromate procedures were tested in rooms thermostatically controlled to 23" to 25" C. and 38" to 40" C:. The determinations in Table I1 for a given dichromate method, together with corresponding iodine tests, were carried out on the same day. rill the determinations given in Table I11 were completed during a single day, as were those of Table IV, the dichromate methods being alternated to minimize the relative effect of any change in temperature, lighting, or alertness of the operator. Separately weighed samples were used for the determinations by the Vosburgh and proposed methods in Tables I1 and 111. The Xillard and Furman method is not well adapted to the use of individually weighed samples; consequently, dichromate solutions were made up from accurately weighed portions of the salt, and aliquot portions were taken for the trials by this method. These solutions were also used for all the determinations in Table IV. The normality of the thiosulfate solutions as determined by the iodine method two days before tJhe hot-room tests was taken as the true normality of the solutions during the tests. All normalities mere
TABLEIv. STdNDARDIZATION O F THIOSULFATE SOLUTION 38" to 40" C. WITH COOLING BEFORE TITRATIONO
AT
(Xormalities corrected t o 23' C.) Willard and lodiur Vosburgh Furman Proposed Nethodb, Method RIethodo Method 2 3 O C. 0.10120 0.10113 0.10116 ..... 0.10117 0.10129 0.10115 ..... 0.10134 0.10127 0.10111 ..... 0.10107 0.10123 0.10113 ... 0.10126 0.10128 0.10119 Average 0.10121 0.10124 0.10115 o:ioii5 0.00027 Range 0.00016 0.00008 0.00004 Standard deviation X 106 10.1 6.6 3.0 2.2 Difference from iodine value + O . 00006 $0.00009 ..... 0.00000 Determinations were carried out in hot room, b u t cold distilled water was a d i e d t o solutions before titration. Average titration temperatures were 24 30°, an,d 28' C. for Vosburgh, Willard a n d Furman, a n d proposed methods,' respectively. b These data, f r o m Table 11, are included for comaarison with dichromat? method values obtained in hot room. Q
Stability Tests The decomposition of thiosulfate is generally attributed to the action of certain bacteria, light, oxygen of the air catalyzed by traces of copper ions, and possibly carbon dioxide (7). A large number of stabilizers have been advocated (4-7, 10, 16, SO), but apparently little is known as to their effectiveness in hot weather. The present investigation involves the periodic testing of variously treated thiosulfate solutions kept in a hot-air bath a t 40 * 1" C., and of control solutions kept a t 20" to 25" C. The solutions were made with freshly distilled water and stored in clean bottles, sterilized by steam. The bottles were of amber glass except two of colorless glass which were used for solutions to be exposed to the light. An individual Normax pipet was used for each solution, chiefly to avoid the possible spread of bacterial infection. The pipets to be used for the 40" solutions mere kept in the hot-air bath, STORAGE TESTS. The normalities of the various thiosulfate solutions were determined by titrating 25-ml. portions against 0.1 N iodine solutions, which were made up as suggested by Hillebrand and Lundell ( 3 ) , and were standardized by the arsenious acid method described by Popoff (12). Alkalitreated solutions were first neutralized with 0.01 h7 hydrochloric acid. Suitable volume corrections were applied to compensate for differences in temperature of the solutions.
INDUSTRIAL AND ENGINEERING CHEMISTRY
804
TABLE V.
STORAGE TESTS ON THIOSELFATE SOLUTIONS" (Normalitiea corrected t o 20° C . ) Nov.
7,s A.
N o special treatment N o preservative
5
0.10086
2.
K O preservative
XH
3.
No preservative, light
06:?0073 6.7 0.10062 6.5
lb.
B. Alkali treatment
4b. Sodium hydroxide, 0.15 g./L 5.
Sodium hydroxide, 0.15 g./L
6 s . Sodium carbonate, 0.2 g./L 7.
Sodium carbonate, 0.2 g./l.
Sodium carbonate, 0 2 RA., light 9b. Sodium carbonate, 0.2 g./l., chloroform 1 rn1.h. 10. Sodium carbdnate, 0.2 g . / L chloroform, 1 ml./l. C . Germicidal treatment l l b . Chloroform, 1 ml./l. 8.
pH hH ;
pH X
XH H ;
0.10068
11.1 0.10042 11.1 0.10062 0.10062 10.4 0.10066
%H 1?):?0073 pH 10.3 A' 0.10062 p H 10.3
K
0.10093
Kov. 17,18
Dec. 8,9,10
Jan. 5,s
Feb. 3,6,7
Mar. 9,10,13
0.10100 6.8 0.10116 6.9 0.10093 6.7
0.10107 6.7 0.10089 7.1 0.10057 7.1
0.10118 5.5c 0.10102 7.3 0.10067 7.2
0.10100 5.0 0.10105
0.10104 4.9 0.10083 7.4 0.10061 7.5
0.10042 11.1 0,10100 11.1 0.10081 10.3 0.10100 10.3 0.10093 10.3 0.10062 9.1 0,10062 10.3
0.10008 11.2 0.10082 11.4 0,10080 10.2 0.10105 10.4 0.10095 10.4 0.10054 8.3 0.10074 10.2
0.10104 6.1 0.10104 6.9 0.10104 6.5 0.10098 6.7 0.10119 8.5 0.10112 8.7 0.10112 7.3 0.10104 7.7
0,09977
0,09932 0.09941 10.5 10.2 0,10076 0.10072 0.10086 11.2 11.2 11.2 0,10087 0.10049 0.10061 10.0 10.1 9.9 0.10093 0.10117 0.10083 10.4 10.4 10.4 0.10083 0.10099 0.1008s 10.4 10.2 10.3 0.10045 0.10059 O.lOO50 7.7 7.6 7.6 0.10093 0,10086 0.10076 9.9 10.1 9.8
10.8
0.10103 0.10094 6.6 4.7c 0.10090 0.10102 7.4 7.2 0.10107 0.10126 1 3 b . Mercuric iodide, 10 mg./l. %H 70:?0104 5,2 6.1 %H "0."1101 14. Mercuric iodide, 10 mg./l. 0.10080 0,10086 6.8 6.7 7.2 0.10050 0.09973 15b. Mercuric cyanide, 0.1 &/I. 0.10116 8.4 4.6C {H 9 . 0 0.10116 0.10133 16. Mercuric cyanide, 0.1 g /I. 0.10101 8.4 8.1 0.10126 0.10130 17b. Toluene, 1 ml./l. XH :::0101 5.7 7.8 7.9 0.10113 18 Toluene, 1 ml./l. 0.10101 0.10108 pH 7.7 6.8 7.8 4 Approximate sulfate plus sulfite concentrations after storage-test period were: solutions other 40' solutions, 0.0005-0.0015 M : .control solutions 2, 7, and 12, 0,0001 iM or less. b Hnt-air b a t h : temperature of solutions. 39.0-40.5' C. C Finely divided sulfur first observed on surface of solution. 12.
Chloroform, 1 m1.A
PnIl
pnFI
::50101
Tests indicated that a change as small as 1 part in 1000 in normality of a solution during storage can be detected with practical certainty, while changes of less than 1 part in 2000 are not necessarily real. The pH values of the solutions were determined with a Beckman pH-meter, model G. A sodium-ion correction was applied to solutions having pH values higher than 9.5. The combined sulfite and sulfate content of some of the solutions was roughly determined by treatment with iodine followed by barium chloride (9). The resulting turbidity was compared with that produced in standard solutions of barium chloride treated with sulfate. RESULTS.A summary of the storage tests is given in Table V. Each solution in the hot-air bath is followed by a control solution kept a t 20" to 25" C. The difference in time intervals should be noted. The untreated solutions (Section A) proved to be surprisingly stable. Thus, after a 10-day aging period, the maximum observed variation in normality during 4 months was 0.18 per cent for solution 1 kept a t 40" C. The effect of light (solution 3) appears to be almost negligible. Despite widespread belief in their efficacy as stabilizers (6, 7, 10, 15) sodium hydroxide and sodium carbonate (Section B) were found to be actually detrimental to stability, especially a t 38" to 40" C. For instance, the normality of solution 4 containing 0.015 per cent sodium hydroxide decreased 1.2 per cent in 3 months. A similar change occurred in solution 15 containing mercuric cyanide (Section C). Of the various other stabilizers tried, chloroform and mercuric iodide proved to be the best; they mere effective even a t 40" C. for about 2 months. During this time the chloroform apparently hydrolyzed, forming an acid which later accelerated the decomposition of the thiosulfate. Fluctuations in normality of thiosulfate solutions are commonly attributed to the formation of sulfite and sulfide
6.8C
0.10060 7.1c
0.10073 4.6 0.10105 7.1C 0.10096
4.9c
0.10067 4.7 0.10083 7.4 0.10092 4.8 0.10076 7.4 0.09910 4.6 0.10118 8.1 0.10127 4.9c 0,10106 7.4
0.10083 6.9C 0.09932 4.5 0.10120 8.2 0.10123 5.1 0.10108 7.0 4 and 15, 0,005 M;
Vol. 14, No. 10 ions followed by their atmospheric oxidation to sulfate and free sulfur, respectively. Reactions like these tend to produce an occasional rise in pH, and also a significant rise in normality if an antioxidant is present; toluene seems to act as such. But regardless of the type of treatment, all the thiosulfate solutions stored a t 40" C. b e c a m e d e f i n i t e l y more acidic, their normalities fell noticeably, and sulfate was formed. T h e s e f a c t s a r e best explained by the following reaction, which obviously would be favored by the presence of alkalies (8): SzOa'
+ 202 + H20 = 2504-
+ 2H+
The control solutions kept a t 20" to 25" C., in contrast, formed very little sulfate or sulfite. Significant changes in their effective normalities must therefore be due to the initial formation of such products as sulfide, tetrathionate ( l d ) , and perhaps trithionate (1).
Summary An improved dichromate procedure for the standardization
of 0.1 N sodium thiosulfate solutions has been developed, particularly for determinations to be carried out a t 30" to 40' C. The accuracy of the method has proved to be within 0.07 per cent over a temperature range of 23" to 40" C. The Vosburgh method (17) and the Willard and Furman method (IO), although satisfactory a t moderate temperatures, gave high and somewhat erratic results a t 40" C. The accuracy of the different dichromate procedures was determined by comparison with the iodine method of Treadwell and Hall ( l e ) ,modified as described in this paper. A number of 0.1 N sodium thiosulfate solutions were treated in various ways and stored a t 20" and 40" C. for approximately 4 months. The maximum variation in the effective normalities of untreated solutions during this time was about 0.3 to 0.4 per cent. Chloroform and mercuric iodide, which proved to be the most satisfactory of the preservatives tried, were effective for only about 2 months a t 40" C. At that temperature, sodium hydroxide and sodium carbonate, instead of acting as stabilizers, actually accelerated the decomposition of the thiosulfate.
Literature Cited (1) Bassett, H., and Durrant, R. G., J . Chem. SOC.,1927, 1401. (2) Foulk, C. W., and Morris, S.,J . Am. Chem. Soc., 44, 221 (1922). (3) Hillebrand, W. F., and Lundell, G. E. F., "Applied Inorganic Analysis", p. 150, New York, John Wiley & Sons, 1929. (4) Kassner, J. L., and Kassner, E. E., IND.ENQ.CHEM., ANAL.ED., 12, 655 (1940). (5) Kirkish, F. J., Chemist-Analyst, 29, 68 (1940). (6) Kolthoff, I. M., €'harm. Weekblad, 56, 878 (1919). (7) Kolthoff, I. M., and Menzel, H., "Volumetric Analysis", Vol. I, pp. 231-8, New York, John Wiley & Sons, 1928.
A N A L Y T I C A L EDITION
October 15. 1942
( 8 ) I b i d . , Vol. 11, p. 299, 1929. (9) Ibid.. p. 370. (10) Low,A. H., Chemist-Analyst,30, 18 (1920). (11) Nakai, Z., Bull. Fishery Erpt. Sta. Gow. Gen. Chosen, Ser. D, NO. 3, 1-24 (1933). (12) Popoff, S.,“Quantitative Analysis”, 2nd ed., pp. 156-7, Philadelphia, P. Blakiston’s Son & Co., 1927. (13) Popoff, S.,and Whitman, J. L., J. Am. Chem. Soc., 47, 2259 (1925). (14) Schulek, E., 2. anal. Chem., 68, 387 (1926).
805
(15) Skrabal, A,, Ibid., 64, 107 (1924). (16) Treadwell. F. P., and Hall, W. T.. “Analytical Chemistry”, 7th ed., Vol. 11, p, 563, New York, John Wiley & Sons, 1930. (17) Vosburgh, W. C., J . Am. Chem. SOC.,44, 2120 (1922). (18) Wasserman, C.,and Eskridge, O., Ethyl Corp., unpublished data. (19) Willard, H. H., and Furman, N. H., “Elementary Quantitative Analysis”, 2nd ed., p. 212, New York, D. Van Nostrand Go., 1935. (20) winkler, L. W., Pharm. Zentralhalle. 69, 369 (1928).
Carbon Dioxide Generator J. A. JOHNSTON, Yale University, New Haven, Conn.
A
SOURCE of carbon dioxide gas is frequently desirable in connection with laboratory or research work. Where the gas is used under low pressure, and especially where considerable quantities are desired on short notice, it is usually cheaper and more convenient t o generate the gas from dry ice than to purchase it in cylinders. The generator shown in Figure 1 has been in successful use for some time in the Chemical Engineering Laboratories at Yale University. This generator is simple in principle, is inexpensive to make, and is free from operational difficulties encountered in previous designs. The body of the generator, A , is made of a 24-gallon Hackney steel drum nith a removable bolt-on cover. The cover is pro-
From Steam
safety Valve c
7 - - - IA
-
E
vided with a 2-inch discharge line, B , a 6-inch handhole, C, and
a stuffing box, E , for the rod supporting the wire basket, D. The discharge line has a pop safety valve set a t 10 pounds per square
inch, and a pressure gage for indicating the pressure in the drum. The handhole cover is grooved for a rubber gasket and is also provided with a small vent line which serves to vent the air trapped in the handhole during filling. This cover is held in place by hinged bolts which fit into slots in a bar running across the top of the cover. This arrangement permits removal or replacement of the cover in a few seconds. An important feature of the generator is the wire basket, D. It is a circular basket made of strap iron covered with 0.25-inch mesh wire cloth. The basket is supported by a steel rod which passes through a stuffing box in the cover. This arrangement permits the basket to be lifted out of the water which fills the lower part of the generator. By adjusting the height of the basket the rate of generation of carbon dioxide can be controlled. The generator is also provided with a steam coil, J , and a valve at the bottom for draining the drum, G . To operate the generator valves F and G are closed, the handhole cover is removed, and the generator is filled about half full of water. Next about 5 pounds of dry ice, broken up into 1 to 1.5-inch lumps, are placed on the basket which has been lifted out of the water. The handhole cover is replaced and the basket is lowered into the water. Generation of carbon dioxide starts immediately. After venting the air from the system through the safety valve and the vent line, valve F is opened to allow the carbon doxide to go into the holder. With this generator it is Dossible t o generate aboui 50 cubic feet* of gas in 15 to 20 minutes by the method of operation described. If a higher rate is desired or if the generator is used frequently, it is necessary to use the steam coils to keep the water lukewarm. The purity of the gas sent to the holder can be controlled by limiting the amount of air in the generator at the time of filling and by purging before allowing any gas to go t o the holder.
FIGURE1. GENERATOR FOR CARBON DIOXIDEGas