ANALYTICAL CHEMISTRY
510 SUMMARY
The hydrofluoric acid method for the dcconiposition of silicates has been combined with a convenient procedure for the removal of calcium and magnesium to give a simple method for the determination of the alkali metals in siliceous materials. The results obtained on determining sodium and potassium in standard samples have been very satisfactory. The procedure described is much simpler in manipulative details than the J. T,air.rcnce Smitli method. &CKNOWLEDGMEVT
The junior authoi is grateful to the Golden Sva Research Institute of Chemical Industry, China, for financial aid for study in this country. The authors also wish to thank E. R., Caley and L. Gordon for making availablc prior t o publication their method for magnesium ( 2 ) . LITERATURE CITED
(1) Elving, P. J., and Caley, E. R., ISD. ENG.CHEM.,AXAL.ED.,9, 558-62 (1937).
(2) Gordon, L., and Caley, E. I t . , AXAL.CHEM.,20, 560-3 (1948). (3) Koenig, E. W., ISD.ENG.CHEM.,ASAL.ED., 7, 314-15 (1935). (4) Kolthoff, I. M., and Sandell, E. B., ,“Textbook of Quantitative Inorganic .4nalysis,” pp. 411-13, Kea York, Macrnillan Co., 1943. ( 5 ) Krishnayya, H. W., Chem. ,Vews, 107, 100-1 (1913). (6) Lamar, M. O., Hazel, W ,hl., and O’Leary, W. J., ISD.ENG. C H E M . , A41CAL. ED.,7,429-31 (1935). (7) Low, A. H., J . Anal. Chem., 6, 666-7 (1892). (8) Makinen, Eero, Z. anorg. Chem., 74, 74-8 (1912). (9) Sohaal, K.B., J . Am. Ceram. Soc., 13, 113-25 (1930). (10) Scholes, S. R., and Wessels, T‘. E., Chemist-Analyst. 25, 38-9 (1936). ( 1 1 ) Smith, R. D., and Corbin, P., ISD.ESG.CHEM.,ANAL.ED.,4, 137-9 (1932). (12) Stevens, R. E., Ibid., 12, 413-15 (1940). (13) Stevens, R. E., private communication. (14) Sullivan, E. C., and Taxlor, W.C., J . Ind. Eng. Chem., 6 , 897-9 11914). (15) Washington, H. S., “Chemical Analysis of Rocks,” 4th ed., pp. 220-2, New York, John Wiley & Sons, 1930. (113) Willard, H. H., Liggett, L. M.,and Diehl, H., IND. EKG.CHEM., AXAL.ED., 14, 234-5 (1942).
RECEIVED June 28, 1948. Abstracted froin a portion of t h e d i s e r t a t i o n presented by P. C. Chao t o t h e Graduate School of Purdue University in iiartial frilfillment of the rerluircment* for t h e dcgrce of master of scicncc.
Determination of Water by Karl Fischer Reagent Stoichiometric Iodometric Method WnI. SEAMAN, W. H. M C C O M A S , JR., AND G . A. ALLEN Calco Chemical Division, American Cyanamid Company, Bound Brook, N. J .
A modification is proposed of the Karl k s c h e r method for the determination of water. Instead of using a single solution containing sulfur dioxide, pyridine, methanol, and iodine, two solutions are used : one containing pyridine, sulfur dioxide, and methanol to be used to dissolve or suspend a sample, and the other containing iodine and methanol to be used for titration. The reaction goes stoichioSO, metrically according to the equation : II:O I? = SO8 2111 without side reactions, although in
+
+
+
T
HE Karl Fisclier method for the determination of water
( 1 ) is an rstremely valuable analytical tool, but it, has two shortcomings. (1) It undergoes rapid deterioration. Thus
Smith, Bryant, and lIitchell(3) have reported, and their findings have been amply confirmed, that the freshly prepared reagent is usually equivalent to about 70% of the amount of water with which it should theoretically be able to react according to the amourit of iodine used, and that in the course of a month or two the strength falls to ahout 40%. (2) The reagen’t cannot be used to determine water on the basis of a stoichiometric relationship between the iodine content of the solution and the reaction bctween the iodine and water, but must rather he used on the basis of an empirical standardization against known amounts of water. Furtherniorr, the empirical standardization must be repeated frequently liecause of the deterioration of the reagent. I t is the practice of commercial suppliers of the Fischer reagent to furnish it in the form of two separate solutions, one of which contailis the iodine; the solutions are mixed immediately before use, obviously to niiriimize decomposition prior to use. Recently Johansson ( 2 ) recommended that thc. reagent lw used in the form of two solutions: one containing sulfur dioxide, methanol, and pyridine, and the other containing iodine and Inethanol. He recommended that the sample be dissolved or mixed with the first solution and that the iodine solution be titrated into it,. His object evidently \vas to niininiize absorption ( i f water from the
practice i t is more conrenient to standardize the solution empirically because of the impracticability of making up the iodine solution with completely anhydrous methanol. The method avoids the decomposition that occurs with the usual Fischer reagent; it thus eliminates the necessity for frequent restandardization and the waste of a relatively expensive reagent which occur with the original Fischer method. The precision is of the same order of magnitude as for the original method.
atmosphere, which is excrssivc with the mixed reagent unless it is kept, in a system protected from atmospheric moisture. I t occurred to one of the authors that if decomposit,ion of the reagent could be eliminated by using the two-solution technique. it might be possible to avoid the necessity for using an empirical Standardization of the reagent, and the reagent might be used a s a standard iodometric reagrnt with its standardization value governed by the equation:
HyO
+ SO,+ CI = SO:, + 2HI
(1)
It was found that, this could be done if the water contained in the iodine-methanol solution itself is determined and a correction applied, for this water will also react when the iodine solution is introduced into the solution of sulfur dioxide, pyridine, and methanol. This method of standardization, however, is not convenient in practice; so an cnipirical method is used similar to that, in use with the usual Fischer reagent. With the proposed reagent the standardization value does not decrease if absorption of moisture is prevented, and the necessity for frequent restand:irdizations which exists with the usual Fischer reagent is absent. The validity of the sugge.*trd technique has been confirmed in two ways: 1. Water was dctmniricd in a number of materials by both thc usual Fischer technique and the suggested technique (Table I). ‘Thc values given sho,v rcasonahlc agreement by thc two
V O L U M E 21, NO. 4, A P R I L 1 9 4 9 Table 1.
511
Determination of Water by Old and New Fischer . Procedures
Sample
Old M e t h o J
Xeiv Method
%
%
Pyridine
0.070, 0.070, av. 0.070
0.068, 0.068, av. 0.068
Alethanol
G./100 ml, 0.276, 0.276, a r . 0.276
G.1100 m i . 0.281, 0.279, a v . 0.280
%
OT /r 0.008, 0.008, a r . 0.008 0.13, 0.11, 0.13, a v . 0.12a
Salicylic acid Dicyandiamide
Triethylamine Aniline-methanol mixture Ethylene dichloride
,0.11,0.09,av.0.10" 0.011, 0.011, av. 0.011
G./lOO mi. G . / 1 0 0 ml. 0.091, 0.085, av. 0.088b 0.075. 0.084, av. 0.O8Oc 0.358, 0.363, av. 0.361d 0.389, 0.390, 0.390 a\., 0.390 0.054, 0.051, 0.021, av. 0.047, 0.047, 0.047, av. 0.052 0.047
Saniple did not dissolve and titration was carried o u t on mixture. Sudden darkening of solution a t end point as crystals settled o u t , b u t further stirring caused reappearance of bright lemon yellow color. b 10 ml. dissolved in 10 nil. of chilled glacial acetic acid. C 10 ml. dissolved in 10 ml. of chilled glacial acetic acid, treated with 25 ml. of Solution -4and titrated with Solution B. E n d point consisted oi sudden darkening from pale yellow color. t o -15' C . d Titrated in methanol solution in freezing mixture a t -10' t o first change from lemon yellow t o orange color which persisted for 2 minutes without further addition of reagent. e 10 ml. dissolved in 25 ml. of Solution A, c o o l c i t o - 10' t o - 1.5' C., and titrated with Solution B.
Determination of Effective Water Equivalence of Solution B. Pipet 25 ml. of Solution 9 int.0 a dry, 125-ml. Erlenmeyer flask, and titrate to an end point with Solution B. ildd an accurately weighed quantity of about 0.08 to 0.09 gram of water to a second 25-ml. portion of Solution A and titrate with Solution B to an end point. Let A equal grams of water added, and R equal net milliliters of iodine solution equivalent to grams of water added directly plus grams of water present in Solution B used (corrected for a blank). Then the effective water equivalence of Solution B as grams of water equivalent to 1 ml. of Solution B = A / B = C. The following data are typical of the values obtained i n determining the effective water equivalence of Solution B (letters refer t o previously dpfined quantities): A , Gram
B , 1111.
0.0896 0.0850 0.0864
45,70 43.50 44.40
C , Effective Water Equivalence 0.001960 0.001954 0.001946 A v . 0,001953
Because the method of standardization is similar to that used with the ordinary Fischer reagent, the precision of the standardization is, a's would be expected, also similar. Determination of Water Content of a Sample by Direct Titration. Titrate 25 ml. of Solution A with Solution B. Let ml. of Solution 13 = D. Add the sample to 25 ml. of another portion of Solution .1 and titrate with Solution B. Let this
loo
methods except for the dicyandiamii!e and the aniline-methanol mixture. The analysis of the former involved a difficulty ujth the end point and that of the latter was carried out in a freezing mixture, so that the discrepancies are understandable. ?;either method is consistently higher or lower than the other. 2. IVater was determined in a sample of methanol by the usual Fischer method. This methanol was then used to prepare a n iodine solution, and it was assumed that the iodine solution had the same water content as the absolute methanol used in its preparation. The iodine content of this solution was then determined by titration in aqueous potassium iodide solution with standard thiosulfate solution. Thcn weighed amounts of water were added t,o a pyridine-sulfur dioxide-methanol solution which was then titrated with the iodine solution. .1 comparison of the total moles of water present in Loth the iodine solution used and the pyridine-sulfur dioxide-methanol solution with the total moles of iodine consumed showed t,hat 1 mole of iodkie reacted with 1 mole of water, with a small experimental error. These d a t a (see Table 111) furnish proof that the reaction is stoichiometric as written. From the known facts coriccrning the rapid decomposition of the usual Fischer reagent, it would be expected that, if water is determined by the suggested reagrnt by a method that is sometimes used-namely, the addition of excess iodine and baektitration of the excess with a standard water solution-difficulties might occur because of the decomposition of the excess iodine. This \vas found to he true if an uncontrolled excess were used a t room temperature and with an excessive period of time before the back-titration. By proper control of excess, temperature, and time, however, the titration could he carried out in this manner with only slight decomposition. (\\-here a direct titration may be used, it is preferablr.) This point has a hearing upon the possibility of using thc suggested tcchnique for titration to an electromc~tricrather than a visual eiitl point>as the eleetroinelric end point is usually observed upoii titratihg an excess of the reagent Jvith a standard water solution (.+). Ah electrometric end point could he obtainctl by thr. n w tcclhique.
volume of Solution B = E. Then % ' H2O = ( Ewt. - Dof ,sample in grams
Table 11.
I
Excess Soiution B, 111.
Equivalent t o Excess 1 2 Added, Gram
HzO in HzO Standard Added, Gram
Difference (Col. 3 Minus Col. 4)
Excess Solution B, R o o m Temperature, Immediate Back-Titration 1
2 3 4 5
43.92 5.92 5.92 5.92 5.92
0.0701 0.0120 0.0121 0.0117 0.0120
0.0877 0.0118 0,0118 0.0118 0.0118
0.0176 -0.0002 -0.0003 0.0001 -0.0002
Varying Exceses of Solution B , Room Temperature, 3 Minutes' Standing 1
2
3
8
5.00 8.00 10.00 10.00 15.00 15.00 20.00 20.00
1
2
3
? 6 7
5.00 5.00 5.00 10.00 10.00 15.00
15.00 25.00 25.00
0.0024 0.0024 0.0048 0.0074 0.0084 0.0082 0.0112 0.0110
0.0076 0.0076 0.0152 0.0146 0.0216 0,0218 0.0287 0.0289
0.0100 0.0100 0,0200 0.0200 0.0300 0.0300 0.0399 0.0399
Varying Excesses of Solution B , Temperature -100 t o Standing
- 150 C., 3 Minutes
0.0087 0,0087 0.0090 0.018fi 0.0186 0.0278 0.0283 0,0467 0.0465
0.0094 0.0094 0.0094 0.0189 0,0189 0,0285 0,0283 0.0472 0.0472
Av.
10 Minutes' Standing a t -10' 1
2 3
4
25.00 25.00 10.00 10.00 10.00
0.0466 0.0466 0,0191 0,0191 0.0191
20 t o 30 Minutes' Standing a t 1
2 3 4
PR0CEI)URE
Preparation of Reagents. The pyridine and particulai,ly the methanol used should be as frev of uater as is feasible. S O L ~ T I OA. N To 950 ml. of ice-cold absolute methanol add 190 grams of liquid sulfur diosiilo. LIix the solution and then add 950 ml. of pyridine, slowly a t fiwt, It is usually convenient to weigh out the liquid sulfur dioxide aud then to adjust tho volumes of methanol and pyridine to give a proportion of 1 gram of sulfur dioxide to 5 ml. of pyridinc arid 5 nil. of methanol. SOLUTION B. Dissolve 60 gi'ams of resublimed iodine in 2 1itcr.s of absolute methanol.
Determinations with Excess Iodine Solution HzO
Eapt. NO.
'
1
ii
7 8 '3
I O . 28 9.98 9.98 10,08
0,0195 0.0190 0,0190 0.0191
t o -15'
C 0.0020 0.0024 0.0007 0.0005 0,0005
0.0446 0.0442 0.0184 0.0186 0.0186
- 10'
to
- 15'
0.0178 0.0182 0.0179 0.0177
30 J f i n u t e s ' Standing a t -10' t o -15' 9.88 0.0188 0.0182 0.0179 0.0195 10.28 0,0191 0.0182 10.08 0.0191 0.0169 10.08 0.0179 10.08 0,0191
0,0007 0.0007 0.0004 0.0003 0.0003 0.0007 0 .oooo 0.0005 0.0007 0.0005
C.a
0,0016 0.0008 0.0011 0.0014 hv. 0.0013
C 0.0006 0.0016 0.0009 0.0022 0.0012 0,0013
Aa. One extjerinient x i t h 20 minutes' standing and 25-ml. excess resulted in ditferenc. qf 0.0029. a
ANALYTICAL CHEMISTRY
512 Table 111. Proof of Validity of Stoichiometric Kelationship: I2 e HzO Expt. No. 1 2 3 4 5 6 7 8 I,
10
H@ Taken, Moles 5,438 5,988 5,483 5.776 5.411 5.758 5.405 6.515 4.950 5.748
Table IV.
Used, Moles 5.401 5.954 5.443 5.754 5.383 5.755 5.413 6.468 4.914 5.132
12
Difference (Col. 2 Minus Col. 3) 0.037 0.034 0.040 0,022 9.028 0.003 -0.008 0.047 0.036 0.016 Av. 0.026
Permanence of Solution B
Effective Water Time of Equivalence, Standing, G . Hz0 Days per M1. 0 0.001985 3 0.001966 7 0.001953 14 0.001952 21' 0.001937 28 0.001907 " Between 14th and 21st days, drying indefinite period of up t o 15 hours.
Iodine Content, Mole 1% per MI. 0.1185 0.1183 0.1181 0.1182
Water Content, G. HzO per M I . 0.000151 0.000166 0.00~175 0.000177 0,1182 0.000192 0.1189 0.000232 tube fell off and v a s replaced after
Determinations with Excess Iodine Solution. In testing the extent of decomposition that would occur if an excess of Solution B were added to a sample (dissolved in Solution -1)in which water was to be determined, the first attempt was made hy adding an excess of B and back-titrating immediately with a standard water solution in methanol. Table I1 indicates that with a moderate excess of Solution B there was no decomposition, whereas a large excess resulted in decomposition. However, the iriiniediatc back-titration is not a practical procedure. With ail exposure of the sample for 3 minutes a t room t,emperature before backtitrating, even with relatively small excesses considerable decomposition occurred. A similar series of experiments, with the reaction flask cooled in an ice-salt mixture at -10" to -15' C., showed that at a reduced temperature excesses of from 5 to 25 ml. of Solution B may be tolerated with a 3-minute reaction time with an average decomposition that is equivalent to about, 0.1 to 0.2 ml. of the usual Fischer reagent. Under the same conditions, except for a reaction period of 10 instead of 3 minutes, a 10-ml. excess of Solution B resulted in a minimum of deconiposition, whereas a 25-ml. excess caused the reagent to undergo considerably more decomposition. Another experiment at reduced temperature with a 25-ml. excess and 20-minutes' standing resulted in a difference of 0.0029. d time of standing of from 20 to 30 minutes with an excess of 10 ml. a t a reduced temperature resulted in a somewhat greater decomposition than the minimum effectfound with a shorter time. The data in Table I1 indicate that determinations may be run by the method using an excess of Solution B at temperatures of -10" to - 15" C. with very little decomposition if no more than about a 10-ml. excess is used and if the reaction mixture is allowed to stand for no more than about 10 minutes prior to hack-titration with the water standard. Electrometric End Point. It was found that the proposed reagent could he used for titrations employing the dead-stop end point method of titration recommended by Wernimont and Hopkinson ( 4 ) , in which an excess of the iodine solution is added and back-titrated with a standard solution of water. (These authors have reported that the direct titration &-as less satisfactory.) Here, as with the visual end point, the temperature, time, and excess would have to he controlled. STOICHIOMETRY O F REACTION
Proof that the reaction takes place stoichiometrically in accordance with Equation 1, without side reactions, was obtained as follows:
In making up a batch of Solution B, the water content of the ahsolut,e methanol used was det,ermined by titration of portions of this methanol with the usual Fischer reagent, and was found to be 0.000182 gram of water per ml. of absolute methanol. It was assumed that the iodine solution prepared from this alcohol had the same water content. Blanks were then run on 25-ml. portions of Solution A by titration with Solution B. Weighed amounts of water were then added to other 25-ml. portions of Solution -4,and these were titrated with Solution B. The net volume of Solution B uped fafter correcting for the blank) would then he equivalent to the water added directly plus the water in the iodine solution added during the fitration. The iodine content of Solution B was obtained by titration with standard sodium thiosulfate solution in aqueous potassium iodide solution. The total weight of water taken was then equal to the weight of water added directly plus the weight of water in the iodine solut,ion fmilliliters of iodine solution multiplied by 0.000182 gram of water per ml.). The number of millimoles of iodine taken and the number of millimoles of water taken were then calculated from the weight of each constituent. The values given in Table I11 show that the number of millinioles of water was greater than the number of millimoles of iodine consumed by reaction with it by an average of 0.026 millimole in some 5 to 6 millimoles for 10 determinations carried out on two different days. This is 0.4 to 0.5% less iodine than would he expected from the stoichiometric relationship. I t is likely that this error is the resultant of the accumulated normal errors that occur in the old and the new procedures! which are both iiir.olvc~l i n ohteining thwe data. PERMANENCE OF REAGENTS
.in interesting change occurs in the apparent water content of Solution A upon standing. One such solution, a portion of which required 4.09 ml. of Solution B for titration of its water content when made up, required the following volumes after 3, 4,5, 6, 10, 13, and 17 days, respectively: 3.75, 3.40, 3.00, 2.72, 1.10, 0.40, and 0.46 nil. ;iccording to Smith, Bryant, and Mitchell (5),the addition compound of pyridine and sulfur trioxide {the latter could conceivably be formed by oxidation with atmospheric oxygen) reacts with water to form pyridinium acid sulfate. This mechanism may conceivahly explain the change in the appBrent water content of Solution A. Practically, however, this change is of little significance, as the volume of Solution h which is used is titrated to an end point or a blank is determined. In contrast to the rapid decomposition of the usual Fischer reagent, Table IV gives the changes that' have occurred in the effective water equivalence of Solution B upon standing in a buret protected from the absorption of atmospheric moisture. The iodine content, (column 3) was obtained by titration in aqueous potassium iodide solution with 0.1 N sodium thiosulfate solution. The water content (column 4 ) was obtained by titrating a known w i g h t of water with Solution B, and then calculating the water content from the excess of Solution B consumed over that calculated from the weight of water used and the iodine content of Solution B. The data in Table IV show that the standardization changes only because of absorption of moisture. ACKNOWLEDGiMENT
The authors are indebted t o Eugene h l . Allen for a suggestion concerning simplification' of the method of standardizing the reagent. LITERATURE CITED (1) Fischer, Karl. Arsgew. Chem., 48,394 (1935). ( 2 ) Johansson, Axel, Svensk Papperstidn., 50, No. I l B , 124 (1947). (3) Smith, D. M., Bryant, sir. M . D., and Mitchell, J., Jr., J . Am. Chem. SOC.,61,2407 (1939). (4) Wernimont, G., and Hopkinson, F. J., IND. ENO.CHEM.,ANAL.
ED.,15,272 (1943).
RECEIVED June 23,1948. Presented before the Meeting-in-Miniature, Korth Jersey Section, ERICAS AS
C H E : \ I I C . ~ SOCIETY, L
January 10,1949.