668
INDUSTRIAL AND ENGINEERING CHEMISTRY
Fading does not, occur during a time interval long enough to enable the worker to make 10 readings. Duplicate determinations are consistently reproducible t’o 1 per cent or less.
Literature Cited
(7) (8) (9) (10)
VOL. 12, NO. 11
Kuttner a d Cohen, J . Bid. Chem., 75, 517 (1927). Kuttner and Lichtenstein, Ibid., 86, 671 (1930). Osmond, Bull. m e . chim. biol., 47, 745 (1887). Schricker and ~ j ~ J~. ARSOC. , . ~oficial ~ ~ - 4.0 ~ .C‘i~em.,22, 167 (1939). ---, \
(11) Smith, Dyer, i\-rensliall, and DeLong, Can. J . Research, B17, 178 (1039).
Rapid Determination of Hydrogen Sulfide and Mercaptan Sulfur In Gases and in Aqueous Solutions JOSEPH .I.SH.IW, RIellon Institute, Pittsburgh, l’enna
A rapid method is described for the determination of hydrogen sulfide and mercaptans i n gases and in aqueous solutions. The hydrogen sulfide and mercaptans are totally absorbed by a specially prepared cadmium chloride solution in a new type of flask. The determinations are made by iodometric titration under definite conditions designed to eliminate several sources of error without burdensome increase in time and manipulation. The required time for the laboratory manipulation is from 5 to 30 minutes. ITH the changes in gas technology that have occurred in recent years, analytical difficulties have arisen a t many points and particularly in the determination of hydrogen sulfide. It is natural t h a t attempts have been made to apply familiar apparatus and methods to new conditions, among them the effort t o use t h e Tutwiler (6) apparatus for such wider application. The Tutwiler procedure consists in shaking 100 ml. of gas in a special apparatus with about 5 ml of acidified starch solution, accompanied by the addition of increments of standard iodine (approximately 0 0133 N ) . Both mercaptans and unsaturated hydrocarbons will absorb iodine under these conditions. dnother source of error has been encountered in the effort to adapt the Tutwiler to determine hydrogen sulfide in concentrations of a few grains per 100 cubic feet. Owing to spatial limitations, the Tutwiler is not suitable for such determinations. B u t because of the convenience of the method, to adapt i t to the purpose, the size of the apparatus has been increased to accommodate a 500ml. sample of gas and the iodine solution has been diluted to one tenth or less of the concentration specified. Large errors a r e introduced by this procedure, however, and they cannot be satisfactorily accounted for on the basis simply of impurities in reagents and sample Many chemists like to determine the hydrogen sulfide in caustic solution by alloning the alkaline sample to run in a thin stream into a n excess of acidified iodine, vigorously agitated (3). T h e excess iodine is then titrated with thiosulfate.
This determination can be made accurately under controlled conditions, but apparent’ly harmless variations in technique can cause very large errors in the results. I n the light of experience, the procedure is by no means so simple as it. appears to be. It is believed that t’lie following method, devised by the author, will avoid all such errors without troublesome technique, and that i t will yield, in a short period of time, results of usually satisfactory accuracy for both hydrogen sulfide and mercaptan sulfur.
Apparatus TKOSliax sulfide flasks and stands, 2 traps (15 x 2.5 em., 6 X 1 inch, test tubes fitted Kith inlet and outlet tubes for gas scrubbing), Gooch crucible (bitumen size), asbestos (“PoFminco” brand long-fiberacid-washed is suggested), and volumetric laboratory equipment.
Reagents Concentrated hydrochloric acid, cadmium chloride solution (10 per cent of anhydrous salt), sodium carbonate solution (concentration approximately 1 N ) , hydrochloric acid solution (concentration approximately 1 N ) , dilute hydrochloric acid solution (80 hydrochloric acid per liter or about 0.08 N , approximate , standard iodine solution (0.1 N or 0.01 A‘), standard iiosulfate solution (0.1 .IT or 0.01 N ) , starch solution indicator, arid methyl orange solution indicator. The standard iodine solution is made up with 2.3 moles of dide t o 1 mole of iodine. The iodine is standardizfd the method of the Sational Bureau of Standards, enious oxide. I t is preferable to standardize both 01 N iodine separately against the arsenious oxide; simple dilution of 0.1 N to 0.01 iV iodine is open to doubt. Some distilled waters produce a considerable error. The starch solution is prepared by making a thin slurry of 2 grams of a soluble starch in a little cold vater, pouring it s l o ~ l yinto 1 liter of nearly boiling water with vigorous agitation, bottling, and cooling. Potassium iodide is not to be added to this indicator.
Description of Sulfide Flask The sulfide flask and stand (obtainable from the Scientific Glass Apparatus C,o., Ackermann St., Bloomfield, S.J.) are shown in Figure 1. The flask, made of Pyrex glass, 1s approximately 175 ml. in capacity; it has a funnel-shaped t.op with a partially grooved ground stopper and housing-a construction that permits solutions to be introduced into the previously evacuated flask. The flask can be employed for scrubbing gas, if desired, thereby obtaining the sample in the container in which it is to be analyzed and eliminating unnecessary handling so objectionahle with these materials.
ANALl-TICAL EDITION
NOVEMBER 15. 1940
-45mrn.O
669
D
-
FLASXPrncx GLASS
FIGURE 1.
SCLFIDE
The angle a t which the stopper is ground should be slightly greater than the angle of the so-called standard-taper interchangeable joints and the face should be shorter, as brought out in the illustration. The stopper must be so ground that no trace of a ridge exists a t the upper end of the ground face; otherwise it will tend to jam when the flask is evacuated. A stopcock grease of a moderately heavy body is desirable. If this joint is properly made and lubricated, it d l operate vithout any binding, even when the flask has been evacuated to only a few millimeters of mercury pressure.
Details of Proposed Method PROCEDURE A (total iodine titration). The sulfide flask and the test-tube trap are each charged with 15 ml. of 10 per cent cadmium chloride and 2 ml. of 1 A' sodium carbonate, and the groove in the mouth of the sulfide flask is closed. The flask is fastened in the gas line, followed by the trap and a meter. I n rare cases where both carbon dioxide and ammonia are present, an acid trap should be placed before the flask, as described more fully beloly. The gas flow is then adjusted to a rate of about 57 liters (2 cubic feet per hour.) by throttling the gas before the flask, to prevent the likelihood of gas pressure causing the stopper to become loose. Enough gas is passed that, when the scrubbing solution is subsequently titrated, from 5 to 15 ml. of iodine will be required as revealed by the amount of color developed. After disconnection, the meter is read and temperature and pressure are noted. Then the contents of the trap are washed into the flask with a minimum of water. The flask is more or less completely evacuated through the outlet stopcock without unduly prolonging the suction. A good water aspirator will do this in about 10 seconds. (In 1913 a proposal was made for the titration of hydrogen sulfide only in a vacuum flask, but the described apparatus was mechanically weak, 2.) Through the grooved funnel top 10 ml. of concentrated hydrochloric acid are added, followed by a little wash water, and mixed by shaking. When using 0.01 N iodine, the contents of the flask must be chilled in an iced bath (or tap water in winter time) to a
FLASK .4SD
STAND
point such that the temperature will not rise above 15' C. before the titration of the iodine has been completed. This chilling appears to be necessary only if 0.01 .V iodine is used and only where mercaptans are present. A measured amount of an excess of standard iodine is added in the same manner, and the residue is washed in with two small portions of water, followed by thorough shaking. After equalizing the pressure on the flask and removing the stopper, the excess of iodine is titrated with standard thiosulfate solution, 5 ml. of starch solution being added as the end point is approached. The thiosulfate is mixed with the sample by attaching about 45 em. (18 inches) of rubber tubing to the inlet tube and blowing gently through it during the titration. If mercaptans (and thiosulfates) are absent, Procedure I3 need not be employed, as the iodine titration is a measure of the hydrogen sulfide present. PROCEDURE B (determination of hydrogen sulfide alone). The sample is taken and the flask evacuated, as previously mentioned; chilling is omitted. At this point the solution in the sulfide flask is made just acid to methyl orange with 1 ,V hydrochloric acid, added through the groove in the stopper in t h e funnel top. (The methyl orange is put in the acid to be added.) Then 8.3 ml. of 1 A- hydrochloric acid are measured out and, together with an amount of water sufficient to dilute the sample to approximately 100 ml., are introduced into the flask in the same manner. The volume of solution is adjusted xvith water to 100 ml. and the flask is well shaken. The stopper is removed, a pinch of dry asbestos ~yooladded, the stopper replaced, and the flask shaken again to facilitate filtration. At this acid concentration (3 grams of free hydrochloric acid per liter) the cadmium mercaptides are in solution and the cadmium sulfide is in suspension. A similar separation under different conditions has been employed in the analysis of petroleum oils (1). The solution is then filtered through asbestos on a Gooch crucible (the bitumen size crucible is preferred) with moderate suction. The stopper is placed in the mouth of the flask, about 10 ml. of dilute hydrochloric acid (80 ml. of iV hydrochloric acid per liter) are poured in the funnel top, the stopper is raised vertically, and the solution is allowed to flow into the flask. The flask
INDUSTRIAL AND ENGINEERING CHEMISTRY
670
is then shaken and the acid poured into the Gooch crucible. The open flask is next blown out to remove mercaptan vapors. The suction is kept on the Gooch filter only long enough to free the pad of loose water, whereupon it is immediately removed. The asbestos pad is rolled up with a stirring rod and transferred to the sulfide flask. A pinch of dry asbestos is placed in the crucible, made just moist, and used to wipe out adhering cadmium sulfide; it is then transferred to the flask, the stopper is inserted, and the flask is evacuated. Ten milliliters of concentrated hydrochloric acid are added, followed by enough water to dilute to about 50 ml., and the whole is shaken to disintegrate the cadmium sulfide precipitate. A measured amount of standard iodine is added in slight excess and the excess iodine is titrated with thiosulfate.
Calculations, Procedure B. (Net - ml. of 0.1 N iodine) 0.0017 X 15.43 X 100 cu. ft. of gas in sample (corrclcted) grains of H2S per 100 cu. ft. of gap ( S e t ml. of 0.1 AT It, Procedure A) cu. ft. of gas sample (corrected) (net ml. of 0.1 N I,, Procedure B) X 0.0032 X 15.43 X 100 = cu. ft. of gas in sample ~. (corrected) grains of mercaptan sulfur per 100 cu. ft. of gas
The analytical calculations in this paper are based on purified gas. For the determination of mercaptans the two samples employed should be taken concurrently. If desired, the special flask (to avoid duplication of equipment) can be used as a substitute for the Tutwiler apparatus. The volume is first determined by water displacement. For sampling, the flask is again filled xvith water, which is then displaced by applying the gas pressure to vihat, under normal conditions of use, is the outlet tube of the flask. The pressure of the gas sample collected is adjusted to atmospheric and an excess of the cadmium chloride N sodium carbonate mixture previously referred to is placed in the funnel top and allowed to run into the flask by slightly loosening the stopper. The flask is shaken to absorb the hydrogen sulfide from the gas. At the convenience of t'lie operator the flask may be evacuated and the hydrogen sulfide titrated as previously described, but using iodine of the concentration recommended for the Tutwiler apparatus and compensating by calculations for the difference in volume of the two flasks. If the gas sampled has insufficient pressure to displace the water in the flask, the difficulty can be met by attaching a short length of rubber tubing to the water outlet; the siphon effect formed will permit the water to f l o out ~ of the flask. TABLE I. DETERMISATION OF HYDROGEN SULFIDE OSLY Taken
Found
Gram
Gram
0.0370 0.0369 0 I0300 0.00154
0.0370 0.0369 0.0300 0.00154
TABLE
11.
This method has been evaluated on nitrogen gas containing known quantities of hydrogen sulfide and methyl, ethyl, and n-propyl mercaptans, obtained from the Eastman Kodak Company. For the purposes of t h e test the pure mercaptans were dissolved in petroleum ether (3.5' to 60" C. boiling range) and measured portions of theoe dilute solutions were in turn volatilized into a stream of nitrogen for testing. The pure mercaptan solutions at first were analyzed in three ways to establish their strength and then the most convenient method was used in the routine collection of data. Where mercaptans exist in the gas the individual scrubbing units do not show so high an efficiency and a trap is necessary. For most work two scrubbing stages are sufficient.
TABLE
Then
Source of Sample
Iodine Used 0.1
0.1
N
,v
0.1 N
0.01 N
N a H S solution N a H S solution N a H S solution G a s (HzS concentration, 0.5 grains per 100 cu. ft.)
EFE'lCIESCY O F SCRUBBISG HYDROGEX SULFIDE O X L Y FRO11 c,OGE-OVEh-G.is
(Sulfide flask and two t r a p s in series, separate analyses) ~H& FoundIf18 in Gas, Iodine I n flask I n ISL t r a p I n 2nd t r a p Approx. L-sed Gr./lOO Gram L/o Gram 5G Gram :'c cu. jt. 0.0567 9 9 . 4 0.00027 0.5 ll.000015 .. 2s 0.l*Y 0.0123 99.8 0.000017 .. ...... .. 19 0.01 .v
.
TABLE 111. ANALYSISOF
SoDIuLi
SULFIDE SOLUTIOXS
(Precipitated a n d filtered under different conditions of acidity, results in grams of HzS per liter) Neutral or Nightly Alkaline Acid Conditions, Free HCI Iodine r s e d 6 g./1. 10 Q . / L 1.20 1.20 0 ,1 s 0.141 o:ik 0.133 0 . 0 1 .Y
VOL. 12, NO. 11
Iv.
DETERMINATION O F HYDROGEX SULFIDE 15 PRESESCE OF MERCAPTANS
Added
HzS
HzS Found
Mercaptan Present
Gram
Gram
Gram
0.0173 0.0173
0.0179 0.0183
0.0240 0.0940
0.00147 0.00142
0.00145 0.00137
0.0047 0.0047
0.0163 0.0153
0.0157 0.0159
0.0466
Type
Iodine Percentage Used Variation
0.0153
0.0155
0.00456
Methyl Methyl Methyl Methyl Ethyl Ethyl Ethyl
0 1s
1 J
0.00156 0.00157 0.00157
0.00156 0.00135 0.00157
0.0055 0.0055 0.0055
Ethyl %thy1 Ethyl
0 01 *v 0.01 5 0.01 s
0.0 1.0 0.3
0.0170 0.0170
0.0165 0.0167
0.055 0.055
2.9 1.i
0.00132 0.00131 0.00130
n-Propyl n-Propyl n-Propyl n-Propyl n-Propyl
0 . 1 .V 0 . 1 .Y
0,00131 U.OUlL(2 0.00132
0 . 0 1 .\ 0 . 0 1 .V 0 . 0 1 A'
0.4 0.1
0.0456
0.0060 0.0060
0.0060
0.1 S 0 . 1 A-
0.01 A'
X.4
5.8 1.4
0 . 0 1 A'
3.5
0 . 1 .v 0 . 1 .v
2.5 4.0
1.1
Discussion of Proposed JIethod The grooved stopper should be well greased with vacuum stopcock grease, and it is well to clean and grease i t before every test. If such care is taken the stopper wil! operate smoothly and easily a t 30-mm. pressure; if not properly done it may seize. If it does seize, as the apparatus is made of Pyrex glass, i t can be quickly heated in a flame t o remove the stopper with little danger of breaking. In following this method it is generally desirable to use 0.1 N instead of 0.01 N solutions where the concentrations of sulfide will permit. Difficulties arise with 0.01 N solutions t h a t collectively spell confusion. Impurities in distilled water affect 0.01 N iodine noticeably. In a portion of this work a good grade of distilled water was used tq dilute 0.1 1%' to 0.01 N iodine. Careful titrations showed a loss uf about 1 per cent of iodine in the process, presumably due to the presence of ammonia, etc., in the water. (The water gave no tests for chlorides or sulfates.) As the sample titrated conttliris distilled water added for purposes of dilution and rinsing, the error connected with the use of 0.01 1%' iodine should be considered, and there is no assurance that the errors mentioned are the maximum errors likely to be encountered. With 0.1 N solutions errors from this source are usually too small for measurement and may be ignored. During the early part of this investigation high and erratic results were obtained in titrating mercaptans with 0.01 iV iodine solutions. If the titrations mere made a t a temperature not exceeding 15" C. these errors were reduced to a satisfactory figure, whereas a t 30" C. six titrations of n-propyl mercaptan were from 9 to 26 per cent too high. This difficulty was not experienced with 0.1 N iodine solutions. Temperature changes within the above limits did not affect the titration of hydrogen sulfide. Titration blanks are usually negligible with 0.1 N iodine solutions. With 0.01 N iodine solutions they can seldom be ignored but are difficult to determine. The blanks vary with temperature, being much less in a chilled solution than at 25" to 30" C. A certain amount of potassium iodide must be present in a titration to give a blue coloration with starch, but duch a solution, if acid, is acted on by the oxygen of t h e air t o give free iodine.
NOVEMBER 15, 1940
ANALk-TICAL E D I T I O S
TABLE V. DETERMISATION OF NERCIPTLSS IN PRE~ESC OFE HYDROGES SULFIDE hfercaptan Added
Gram 0.0287 0.0287
Mercaptan Found Gram 0.0256 0.0278
Type Xethyl Methyl
HzS Present Gram 0.0178 0.0178
Iodine Used 0.1 Y 0.1 S
Percentage Variationa 0.3 3.0 5.2 4.2 5.5 1.6 7.4 3.6 3.0 0.4
Approx. Concentration i n G a s Gr./lOO cu. ft. 29 29 2 9 2.9 3.3 9.5 9.0 61 50 54
0.00143 0 . 0 1 .V 0.00360 0.00341 U e t h y l 0.00143 0 . 0 1 ,V 0.00360 0 . 0 0 3 4 5 h f e t h y l 0.00157 0 . 0 1 S 0 . 0 0 3 9 3 0.00415 E t h y l 0,00157 0 . 0 1 S 0.00529 0.00537 E t h y l 0.00157 0 . 0 1 .V 0.00535 0.00498 E t h y l 0.0496 0.0514 n-Propyl 0 . 0 1 7 0 0.1 S 0 . 1 .V n-Propyl 0 , 0 1 7 0 0.0496 0.0481 0 . 1 A' n-Propyl 0.0170 0.0496 0.0494 0.00819 0.00486 n-Propyl 0.0017 5,0 0 . 0 1 .V s . 2 0 . 0 1 .V 5.3 5.0 0 . 0 0 5 0 7 0 , 0 0 5 3 5 n-Propyl 0.0017 0 . 0 1 .V 0.4 4.7 0.00507 0.00605 n-Propyl 0.0017 a Over-all error. Includes errors of handling volatile mercaptans as well as errors inherent in analysis.
The iodine thus set free is partly dependent upon concentrations of iodide. This error can be reduced to workable proportions by having in the system a minimum of potassiuni iodide. Consequently potassium iodide is left out of the starch and adjusted in the iodine solution t o the amount previously specified. This action seems to penalize seriously nothing but the blank, where only a few drops of iodine solution are normally rrquired. It is compensated in the blank on 0.01 LV solutions by adding 2 ml. of standard iodine and titrating with thiosulfate, adding 5 ml. of starch solution before the end point is reached. From the tn-o readings the blank can be calculated. Admittedly this procedure leaves something to be desired, but the results obtained have been numerically satisfactory.
Under Procedure B the 8.3 ml. of 1 hydrochloric acid, added to the sample to produce a concentration of 3 grams per liter of free hydrochloric acid, should be added in 3 or 4 portions with shaking. If i t is added all at once it is likely to make a cadmium sulfide suspension of highly colloidal characteristics that is difficult to filter. Sormal sodium carbonate is added to the cadmium chloride scrubbing solutions to give a n alkaline reserve (cadmium carbonate) which will prevent any marked change in the pH of the solution through limited amounts of hydrochloric acid set free by reaction of the cadmium chloride with hydrogen sulfide and mercaptans. K i t h i n reasonable limits, volatile alkali-e. g., ammonia-is without appreciable effect on the pH of the reagent. This solution has the advantage of not being affected prejudicially by carbon dioxide in the gas. When working with alkaline sulfide solutions one is likely to encounter thiosulfates, which under the specified procedure would be reported as mercaptans. These compounds may be eliminated b y filtration from the cadmium precipitates. If it is mechanically inconvenient to make this filtration, thiosulfate can be determined in a separate sample b y precipitating the combined sulfides in dilute solution and filtering off and titrating a n aliquot. The thiosulfate content is then calculated and deduced from the total iodine titration found by Procedure -4. Under the conditions of the test, thiosulfate can be titrated with iodine in acid solution if the iodine is added promptly after the acidification. Cadmium chloride has a strong stabilizing effect upon thiosulfate in acid solution. I n this laboratory 10 ml. of 0.1 N thiosulfate introduced into 100 ml. of a 3 to 4 per cent cadmium chloride solution at 25" C. and containing 10 ml. of concentrated hydrochloric acid, when let stand 10 minutes, required only 10.1 ml. of 0.1 N iodine. Attempts made to determine the mercaptan sulfur directly by variations of this procedure met with failure, owing no doubt to the fact t h a t the necessarily minute amounts of a volatile easily oxidized material do not lend themselves to handling in the open air. T h e delicacy of the method is such that it can frequently
671
be used for analyzing gases having concentrations of hydrogen sulfide and mercaptans much lower than those indicated. T h e demonstration of this fact by means of synthetic gas samples involves a much more elaborate technique than t h a t used in this work. Tests on industrial gases nominally free from hydrogen sulfide and mercaptans gave titrations showing a sulfide concentration as low as 0.01 grain per 100 cubic foot calculated as hydrogen sulfide. The precipitate in the scrubbing solution had a slightly yellow color and the acidified scrubbing solution of cadmium salt had the characteristic odor of lorn-boiling mercaptans. If a gas to be analyzed contains appreciable quantities of both carbon dioxide and ammonia, ammonium carbonate mill be absorbed in the scrubbing solution and may be present in sufficient quantity to destroy the vacuum in t h e flask after acidification. I n this case a test-tube t r a p (15 X 2.5 cm., 6 X 1 inch) containing dilute acid (10 ml. of 20 per cent by volume hydrochloric acid suggested) should be placed before the sulfide flask. A t the end of the sampling period this solution should be made slightly alkaline to methyl orange and added to the sulfide flask prior t o the evacuation. I n this procedure the carbon dioxide will pass the scrubbers without interference. It has been demonstrated t h a t alkaline sulfide solutions can be analyzed in the manner described. It is suggested that in all cases before evacuation cadmium chloride solution be added in excess. Occasionally a solution contains a large amount of carbonate and a relatively small amount of hydrogen sulfide. I n such cases an exces? of ammonium hydroxide and asbestos fiber should be added with the cadmium chloride, the solution filtered through a Gooch crucible, and the residue placed in the special flask for titration with iodine. S o attempt has been made to ascertain the analytical effect of carbonyl sulfide, another of the possible minor constituents of certain industrial gases. It seems unlikely, however, t h a t carbonyl sulfide will affect appreciably the results obtained by this method (4).
Conclusion A method has been described for the rapid determination of hydrogen sulfide and mercaptans, and its limits of error over a useful range have been shown. I t s application to gases having a concentration of only fractional grains of these substances per 100 cubic foot of gas has not, as yet, been explored, but semiquantitative tests indicate a probable Satisfactory delicacy for the procedure. The method is unlimited as to size of sample; i t is independent of t h e action of unsaturated hydrocarbons and free from certain manipulative difficulties often encountered in the iodometric titration of alkaline sulfides. The method is submitted at this time upon the urgent request of specialists in this field because of the lack of a satisfactory rapid means of making these determinations.
Aclinowledgmen t The thanks of the author are hereby extended to D. V. Moses of E. I. du Pont de Semours ti Co., Inc., Belle, TV. Va., for his courtesy in running certain helpful tests on gases containing high concentrations of ammonia and carbon dioxide.
Literature Cited (1) Faragher, Morrell, and Monroe, IND. EKG.CHEM.,19. 1281 (1927). (2) Harding and Johnson, Ibzd., 5 , 836 (1913). (3) Key "Gas Works Effluents and Ammonia", p. 131, Appendix I, London, Institution of Gas Engineers, 1938. (4) Treadwell and Hall, "Analytical Chemistry", 7th ed., Vol. 2, pp. 644-5, New York, John Wiley & Sons 1930. (5) Tutwiler, J . Am. Chem. SOC.,23, 173 (1901). CONTRIBUTION from t h e Industrial Fellowship on G is Purification sustained by t h e Koppers Company, Pittsburgh, P e n n z
