V O L U M E 23, NO. 8, A U G U S T 1 9 5 1
1069
(7) Barber, C. F., J . Soc. Leather Trades' Chem., 3, 206 (1919). (8) Baumgarten, Werner, Stone, L., and Boruff, C. S., J. Assoc. Ofic. Agr. Chemists, 27, 425-30 (1944). (9) Berthelot, Ch., Chimie & Industrie, 47, 147 (1925). (10) Besson, A. A . , Schweiz A p o t h . Ztg., 55, 69 (1917). (11) Bidwell, G. L., and Sterling, W. F., Ind. Eng. Chem., 17, 147 (1925). (12) Blythe, J. R., J . Soc. Leather Trades' Chem., 3, 101 (1919). (13) Boller, IT. H., Chem.-Ztg., 52, 721 (1928). (14) Brodshii, hl. S.,T r u d y Leningrad Inst. Setet. Torgoali, 1940, No. 5, 3-21; K h i m Referat. Zhur., 4, S o . 9, 90 (1941). (15) Brown, E., and Duvel, J. IT, T., U. S. Dept. Agr. Bur. Plant Ind.. Bull. 99. 11907). (16) Brown, E . H., Morgan, H. H.. and Rushton, E . R., IND.ESG. CHEY., AkN.%L. ED., 9, 524 (1937). (17) Bruening, C . F , J . Assoc O f i c . A g r Chemtsts, 25, 903-9 (1942). (18) Calderxood, H S , and Piechowski, R. J., IND.ESG. CHEM., ANAL.ED.,9. 520 (1937). (19) Churchward, C . R., .-l?istraliun Chem. Inst. J . &. Proc., 10, 69-82 (1943). (20) Cleland, J. E., and Fetzer, IT. R., ISD. ENG.CHEM.,-4n-a~.ED., 13, 858 (1941); 14, 27, 124 (1942). (21) Couto, A. V, do, Chimica e industria (hlilan),3, 462 (1928). (22) Dean, E . IT,,and Stark, D. D., J . I n d . Eng. Chem., 12, 486 (1920). (23) Dedlow, C., and Smith, D. T.. Ibid., 18, 858 (1926). (24) DeLoureiro, J. A., J . dssoc. 0.07~.4 g r . Chemists, 21, 645 (1938). Z . cingcus Citem., 51, 725 (1927). (25) Dohmer, P. JT., (26) Drefahl. L. C., et ctl.. P r o c . ;Im.TFood Prescrrers' Assoc., 1926, 38-77. (27) Evans. J. E., and Fetsei, JT-. R., ISD.ENG.C H m f . , -4N.u.. ED., 13, 855 (1941). (28) Feder, E . , 2. .Yahr. Genussm., 37, 265-78 (1919). (29) Fedorov, A. A , Ztrmdskaya Lab., 11, 364-7 (1945). 130) ~,Fetzer. IT. R.. Evans. J. W.. and Loneenecker. J. B.. IND. ENG.CHEM.,'-~X.IL. ED.,5, 81 (1933). (31) Fisher, R., and Hauser, TT-., Scientia Pharwz., 13, 17-19 (1942). (32) Folpmers, T., Chem. FeekClad, 13, 14 (1916). (33) Fuchs. F. C.. Eng. M i n i n g J . , 106, 357 (1918). (34) Gough, C . >I., and Green, E. H.. J . SOC.CRem. I n d . , 61, 91-3 (1942). (35) Grafe, E., Brciunkohle 5, 561 (1906). 2. angew. C ' k e m . , 21, 890 (1908). (36) Gray, C. *I., (37) Hadorn, H., Mitt. L e h e n s m . Hyg., 36, 324-34 (1945). (38) Hallsworth, E. G . , and Reid. R. L., S a t u r e , 150, 424 (1942). (39) Hoffman, J. F., Z . a n g e u . Chem., 21, 890 (1908). (40) Holt, P. F., and Callow, H. J., S u t u r e , 148, 755-6 (1941). (41) Holtappel, K. J., Pharm. Tijdschr. A-ederland Indie, 3, 247-61 (1927). (42) Horner, IT, L. ( t o Core Laboratories, Inc.), C. 9.Patent 2,282,654 (May 12, 1942) 2,361,S-14 (Oct. 31, 1944). ~
~
~1
(43) Hruda, Miroslav, Z. Zuckerind. B o h m a n M a h r e n , 67, 97-101 (1944). (44) Jones, J. M., and McLachlan, T., Analyst, 52, 383-7 (1927). (45) Kauffman, H. P., and Keller, M. C., Fette u . Seifen, 49, 272-5 (1942). (46) Langeland, E. E., and Pratt, R . W,, IXD.ENG.CHEM.,ANAL. ED., 10, 401 (1938). (47) Linden, T. van der, Kauffman, &I., and Leistra, F., Arch. Suikerind, 25, 941 (1917); Analyst, 43, 211 (1918). (48) Lindsay, IT. Ll.,Ibid.,18, 69 (1946). (49) Locket, G. H., and Barrett, IT, H., S u t u r e , 149, 612 (1942). (50) Lownes, A. G., Ibid.,148, 594-5 (1941). (51) Maercklein, 0. C., J . Assoc. Ofic. -4gr. Chemists, 25, 921-4 (1942). (52) hlarcusson, J., Mitt.kgl. , ~ ~ a t e r i a l p i u j i ( ~ i g s a ,Gross-Lichterfelde nt West, 22, 48 (1904); 23, 58 (1905); Z . angeui. Chem., 18, 754 (1905). (53) Ilichel, Franz, Chem.-Ztg., 37, 353 (1913). (54) RIiller, A. T., Jr., J . Biol. Chem., 143, 65-73 (1942); 149, 153-5 (1943). (55) Myhill, A. R., Gas. J . , 150, 21 (1920). (56) Korman, JT.,2. angew. Chem., 38, 380 (1925). (57) Perkins, A. E., J . D a i r y Sci., 26, 545-51 (1943). (58) Phillips, Elmer, and Enas, J. D., J . dssoc. Oi%c. Agr. Chem., 27, 442-5 (1944). (59) Picozzi, Aldo, Ann. chim. applicata, 32, 51-3 (1943). (60) Rice, E . W., I n d . Eng. Cheni., 21, 31 (1929). (61) Rogers. J . S., U. S. Dept. Agr., Bur. Chem., Bull. 137, 172 (1910). (62) Sadtler, 8 . S., J . I n d . Eng. Chem., 2, 66 (1910). (63) Sair, I., and Fetzer, IY. R., Cereal Chem. 19, S o . 5, 633 (1942). -1x.t~.ED.. 16, 720 (1944). (64) Schley, C. R., IND.ESG. CHEM.., (65) Schwalbe, C . G., 2. angew. Cheni., 21, 2311 (1908); (I) 32, 125 (1919). (66) Simek, Bretislav, and Ludmila, Jaroslai-, F u e l , 26, KO.5, 132-7 (1947); Brenn8to.f. Chem., 23, 223-7 (1942). (67) Simon, JY.,Braunkohle, 41, 53 (1942). (68) Bindall, H . E., J . Assoc. Ofic. d g r . C h e w , 2, 197 (1917); 3, 427-8 (1920). (69) Steagall, E. F., Ibid.,28, 500-2 (1945). (70) Talon, Raymond, Bull. assoc. chiin., 62, 149-58 (1945). (71) Testoni, G., Staz. sper. agrar. ital., 37, 366; Chem.-Zentr., (2) 75, 562 (1904). (72) Thorner. W., Z . ongew. Chem., 21, 148 (1908). (73) Trusler, R. B., IND.ENG.CHEM,,AXIL. ED.. 12, 509 (1940). (74) Weeks, J. F., Jr., J . Assoc. Ofic.Agr. Chenz.. 29, 35-6 (1946). (75) Wefelscheid, H., OeZ u. Kohle, 37, 236-49 (1941). (76) Woodmansee, C. IT., Rapp, K. E., and hIcHargue, J. fj., J . Assoc. Ofic.Agr. Chem., 25, 142-5 (1942). ( T i ) l a m a d a , S., and Koshitaka, T., J . Soc. C'hem. I n d . ( J a p a n ) ,30, 356 (1927). RECEIVED September 23, 1950.
Karl Fischer Reagent Titration JOHN MITCHELL, JR. Polychemicals D e p a r t m e n t , Research Dicision, E . I . du Pont d e lVemours & Go., Inc., T i l m i n g t o n , Del.
B
O T H physical and chemical procedures are available for determining water. The choice of the technique best suited for a particular problem is dependent on several factors, including sensitivity and the precision and accuracy required, facilities available, and t h e nature of the materials to be analyzed. T h u s in the lumber industry where facilities often are limited and high accuracy is not essential, many types of direct-reading instruments are available \vhich depend on some electrical property, such as dielectric constant, conductance, or capacitance. Most physical methods are based on direct removal of water, after which t h e anhydrous residue is weighed and moisture is estimated by difference, or the water is recovered and measured volumetrically or gravimetrically. The former condition is most often represented in oven drying. T h e latter condition is t h e basis of most azeotropic distillation techniques, where t h e recovered water is measured, or of volatilization methods, where moisture is absorbed on a n active desiccant. Obviously all these methods fail on materials t h a t are thermally unstable. Sumerous chemical methods have been proposed for the
determination of water. Actually, those physical separations employing recovery of moisture on a desiccant are partly chemical in nature, as this absorption usually requires chemical reaction either by hydrate formation or by hydrolysis. T h e most widely applicable chemical method is t h a t based on the Karl Fischer reagent. Indeed, i t can be stated unequivocally t h a t t h e Karl Fischer technique may be employed successfully for the determination of moisture in many more types of materials than a n y other existing method. T h e nearly specific nature of this reagent, together with the rapidity with which analyses can be performed, is making t h e Fischer reagent an essential tool in t h e analytical laboratory. However, i t can be used effectively only when its great potential and limitations are clearly understood. Karl Fischer's initial investigations were based primarily on the need for a reliable method for moisture in petroleum chemicals and sulfur dioxide. Since its publication in 1935, the Fischer method has been t h e subject of an increasing number of publications describing new fields in which this technique has found successful application.
1070
ANALYTICAL CHEMISTRY
Fischer reagent is composed of iodine, sulfur dioxide, p y idine, ~ and methanol. Contrary to some reports. each of these conipounds enters into the basic reaction with water. The over-all process involves the two-step reaction ( 3 2 ) : CjHjN.Ia
+ CjHjX.SO, + C j H j S + H?0 --+ 2CjHjS.HI +
CjHjS.SO3 (1)
and CoHjX SOa
+ CHIOH + CjHjS(H)S04CH?
(2)
I t is apparent t h a t only Equation 1 involves TTater, nhile Equation 2 completes reaction of the intermediate, the pyi idinesulfur trioxide complex. Other compounds, containing active hydrogen, can react with this complex. For example, n a t e r might be involved, as shown in Equation 3 CoH,X.S03
+ HzO +C,H,S(H)SOIH
(3)
during direct titration-i.e., under conditions in which the iodine reacted immediately with water. K h e n escess iodine was added, side reactions n-ere like]!- to occur. However, a t temperatures of -10’ to -15” C. very little decomposition was evident if no more than a 10-ml. excess of reagent was used and the niisture was allowed to stand no more than 10 minutes before back-titration. Another variation of the two-reagent titration procedure for microdeterminations of water v a s proposed recently by Johansson ( 6 ) . In order to improve the sensitivity of the titration, he suggested that t,he hydrogen iodide formed in the reaction be determined, as shown in Equations 4 and 5
+ 3Br2 + 3H:O --+HI03 + 6HBr HIOI + 5 K I + 5HCl +31, + 3HzO + 5KCl HI
(4)
(5)
Iodine thus formed is determined by titration rvith sodium However, Reaction 3 would be of no practical value for t h r thiosulfate. Since in the Fischer reagent reaction 1 mole of water determination of water. I t is not specific for n-ater and i. not forms 2 moles of hydriodic acid, the final titration actually will accompanied by a color change. involve 6 moles of iodine. Johansson reported a n accuracy to The stoichiometric requirements for the components of the n-ithin 0.04 nig. on samples containing 0.5 to 4 mg. of r a t e r . Fischer reagent are established as given in Equation. 1 and 2I n the final chapter of the book “Aquametry” it was suggested t h a t is, for each mole of watei are required 1 mole of iodine, 1 t h a t bromine could replace iodine in the Karl Fischer reagent. mole of sulfur dioxide, 3 moles of pyridine, and 1 mole of methanol. Johansson (6) employed this idea to remove water from the As it is common practice to employ excess sulfur dioxide, pvridine, pyridine-sulfur dioxide reagent before addition of the sample. and methanol, the strength of any given preparation of reagent is However, as bromine alone did not effect a change at the end limited by the iodine concentration. The excesses of other conipoint, a few microdrops of iodine w r e introduced and then ponents may be varied t o meet particular analytical requireineiits bromine was added. At the end point, iodide was oxidized to T h e author has found t h a t a methanolic solution containing otht,i iodine, Tvhich resulted in a visual or electrometric change. components in the ratio 1In:3502: 10CsHjS a t a concentration “Preneutralization” of the solution, therefore, did not effect any equivalent to about 3 5 mg. of \later per nil. of reagent to be the significant formation of iodide ion. Then the sample could be pi eferred coniposition for general n-ork. Other compositions mav introduced into the anhydrous system and iodine added. A be better suited for specific purposes-for example, a reagent 1 direct electrometric end point u-as employed. containing considerably more pyridine may be used for titration. This extension of technique increases the time requirement for Lqater in acetone. A reagent containing as little as 4 to 5 and probably does not significantly improve the sensitivity of the moles of pyridine per mole of iodine would be suitable in control analysis. Both sensitivity and accuracy are still dependent upon titrations for water in alcohol and ~vould effect a significant the esactness with which the iodine solution can be added to be savigg in cost of the reagent. equivalent to the water content of the sample. Numerous variations have been proposed for the preparation of Fischer reagent. In the author’s laboratory vihere large quantiTITR %TIOX ties are consumed in all types of applications, the following proThe method employed for titration-Le., visual or electro cedure has been found most efficient ( 1 6 ) . Stable “stock” metric-has been the subject of numerous publications, with the solutions are prepared of iodine dissolved in pyridine and then advocates of each stoutly pointing out the superiority of one diluted with methanol. Liquid sulfur dioxide is added to porover the other. The electronietric end points are the more tions of the stock solution a day or ta-o before use. sensitive but also more time-consuming. The visual titration reThis preparative method is employe‘d t o minimize losses of quires only simple apparatus and, contrary to the espostulations active reagent via parasitic side reactions which conqume iodine. of the electrometric titration esponents, presents a relatively I n the absence of sulfur dioxide, hon ever, the reagent remains sharp end point on solutions which are not deeply colored. stable indefinite]). For this reason many of the laboratory supply Most of the difficulties encountered tvith the visual end point houses market the reagent as t n o solutions: a solution of iodine appear t o be associated with incomplete tit’rations. The color in methanol, and sulfur dioxide in pyridine. They recommend changes are canary yellow to chromate yellow to the brown of that the solutions be mixed shortly before actual use. unused iodine. Often a permanent color standard is desirable For either of the above cases, the complete reagent usually when volumes of 250 ml. or more are titrated. Jones (9) suggested is delivered into the sample t o be analyzed. A unique niodian arbitrary standard of 0.003 AYiodine in water solution. T h e fication was suggested by Johansson (?), who prepared b o sepaauthor has found that, an approsimately 0.01 S iodine in methanol rate solutions of iodine in methanol, and sulfur dioxide in pyrisolution more closely compares with the true end point. dine and methanol. The latter also was used as solvent for the At the proper end point, the addition of 0.1 t.o 0.2 ml. of resample and the former as titrant. T h e iodine in methanol soluagent effects a very marked change in color to dark brown. tion was stable and, whkn used in direct titrations, offered negliClear or lightly colored solutions may be titrated easily with a gible interferences fiom degradative side reactions. Seaman, reproducibility and accuracy of better than &0.2 ml. of Fischer RlcComas, and Allen ( 3 1 ) in further studies of this technique reagent, equivalent t o about 0.5 mg. of water. Practically, this demonstrated that, provided adequate protection against moiserror is insignificant in most analytical work. ture x a s given, each preparation of iodine-in-methanol solution need be standardized only rarely. This standardization was most For example, let us assume n-e have a solution of alcohol conconveniently made by means of sodium thiosulfate titration. taining 2.00% water. We titrate duplicate 10-ml. (8-gram) Titrations of samples for water n-ere subject to a small but consamples and fmd 159.5 and 160.5 mg. of water. Our found figures would be 1.99+ and 2.01-0/,, respectively. Similarly 50 stant correction for water in the methmol used to prepare the ml. (40 grams) of a sample containing 0.20% water might give iodine solution. 79.5 and 80.5 mg. of water equivalent to 0.199 and 0.201%. Seaman and his con orkers made the important observation However, on the same basis, 50 ml. of a sample containing 0.001 % t h a t no significant side reactions of the Fischer reagent occurred water would give results as high as 0.002%.
V O L U M E 23, !O.
1071
8, A U G U S T 1 9 5 1
The rapid Fischer reagent titration probably represents the most widely applicable technique for the determination of moisture. Either a visual or an electrometric end point may be employed. The former requires only simple apparatus and is sensitive to less than 0.5 mg. of water. The latter requires a desiccantprotected closed system and is sensitive to about 0.2 mg. of water. The “deadstop” end point employing a direct titration procedure appears to be the most convenient electrometric technique. Organic substances which interfere in the direct titration for water include carbonyl compounds, mercaptans, diacyl peroxides, thio acids, and hydrazines. Usually methods are available to eliminate these interferences. Inorganic compounds w-hich interfere include metal oxides, hydroxides, carbonates, bicarbonates, chromates, dichromates, borates, and sulfides. Often the interfering reactions are quantitative. Several methods based on prior distillation or extraction of the water have been proposed for t h e determination of water in these inorganic systems. This titrimetric method is hecoming increasingly important for the routine determination of water in commercial materials. Reliable methods have been devised for t h e determination of water in petroleum products, oils, fats, waxes, explosives, paints, polymeric materials, soap, and many foodstuffs and carbohydrates.
If sample size is limited and high accuracy is required, the electrometric end point would be better suited. However, if large quantities of sample are available, tITo methods ma)- be used Lvith the visual titration. 1. Considerably larger samples might he employed. For example, as much as 500 grams of liquid butadiene may be titrated directly a t 0 ” C. with a reproducibility of better than rtl p.p.ni. -1 similar precision was observed in the determination of water in adipic acid. Samples weighing 75 to 100 grams were dissolved in 250 ml. of pret,itrated 1 to 1 pyridinemethanol. The resulting Fischer reagent titration served as a direct measure of moisture in the sample. Results are s1ioLv-n in Table I. I n a serics of six successive determinations, figures were 630 i 4 p.p.m. .inother series of seven determinations made over a period of sc)veral days showed a maximum deviation of *lo p.p.m. 2. Direct extraction methods may apply. Thus, Gester ( 4 ) found t h a t 250 nil. of ethj-lme glycol could be used t o extract traces of moisture from a gallon of hexane.
Table I.
OPERATOR
OPERATOR
I 1 E L E C f RONlC RELAY SYSTEM
d--h INJECTION SY R l NGE
lUl
Analytical Data for Water in iclipic icid Water Tollnd. TVt % 0.0642 0.0635
Number
0.0642 0.0638 0.0642 0.0633
Av.
0.0639 i. 0,0004
0.056 0.056
0,056 0.0j8 0 037 0.057 0.060
-4v. 0.057 = 0 . 0 0 1
EQUIPhIENT
T h e cquipment required for visual titration includes a n automatic or bottom-filling buret suitably protected against atniospheric moisture b y the use of desiccant-packed tubes. Where possible, 250-ml. volumetric or similar long-necked flasks should be used as titrating vessel^. The ends of these flasks can be brought u p t o contact the bottom of the h r e t stopcock and thereb y minimize exposure t o the atmosphere. A completely closed system can be made b y attaching a spherical inner grind t o the base of the liurrt stopcock and using a flask x i t h a spherical outer grind ( 1 7 ) . The spherical grind of the flask is attached t o the buret stopcock and held in place hy a spring clip. Sufficient play is given t o permit manual agitation during the titration. This assemhly is particularly useful for analyses a t reduced temperatures and for the slower titrations of heterogeneous systems,
Figure 1. .ipparatus for Automatic Tit rations
in which reagent is added until an apparent end point is reached and the mixture is allorved t o stand t o permit further extraction of moisture from the solid phase. For electrometric titrations the “dead-st,op” technique, employing platinum electrodes, is most often used. Instruments based on this principle are sold by many laboratory supply houses. Generally, excess Fischer reagent is added, after n-hich the excess is determined by back-titration a i t , h standard waterin-methanol. This procedure was employed originally because the end point appeared sharper. Ot’her investigators-r.g., Carter and Williamson (?)-have reported the use of direct methods, emplo)-ing a greatly increased potential betn-een the platinum electrodrs. Probably the use of platinum electrodes of considerahly larger diameter (1,’s t,o 3 / , 6 inch) also \vould accomplkh improved sensitivitj-. All electrometric methods require a completely closed system (10). T h e microtitrations are best carried out by electrometric methods. On this scale the novel assembly of Levy, Murtaugh, and Rosenblatt ( 1 1 ) is useful. Rubber dental clam is used to,seal t,he tube. The buret is connected t o the hypodermic sJ-ringe needle Tvhich pierces the dental dam. Ot,her needles may lie used to introduce sample or act as vent. T h e platinum electrodrs, sealed into the bottom of the tube, are connected b y suitable leads t o the titrimeter. Johansson ( 6 ) described an apparatus for automatic microdermic syringe, driven by a worm-gear motor connected t o an electronic relay in such a wav
A N A L Y T I C A,L C H E M I S T R Y
1072 Table 11. Noninterfering Organic Compounds Class
Examples Ca:Loxylic, hydroxy-, aniino-, sulfonic Mono-. polyhydric. phenols lriormal carboxylic, ortho, carbamates, lactones, esters of inorganic acids Sugars, formaldehyde, b e n d , benaoin, chloral
Acids Alcohols Esters Stable carbonyl compounds Acetals and ethers Hydrocarbons
Saturated, unsaturated a n d aromatic)
Anhydrides and acyl halides Peroxides Nitrogen compounds Halides Sulfur compounds
(aliphatic
Hydro-, dialkyl All types Sulfides, thiocyanates, thio ebtcrs
that a t the end point, as determined electrometrically, the motor instantly stopped. T h e volume of Fischrr reagent employed then could be read from the calibrated screw system. A means for manual operation of the plunger also was provided. The Karl Fischer reagent may be used for the determination of water in liquids, gases, and solids. However, certain classes of compounds interfere either by reacting with the iodine of the reagent to show apparent water or by oxidizing the hydrogen iodide t o form iodine and thus lead to low values for free water. The types of organic compounds which do not interfere are illustrated in Table 11. Very few exceptions have been observed. One of the more important to the food chemists, a t least, is ascorbic acid. Vitamin C is oxidized to dehydroascorbic acid.
o=c-I 7 HO=C 1 II I H-C-A
HO=C
p +
+
1
0
+2HI
(6)
‘
I
HO-C-H
I
I2
CHZOH
I I
HO-C-H
2RSH
+
12
+
RSSR
+
(9)
2HI
This interference can be eliminated by prior addition of the mercaptan to a n active olefin (18).
Xanthates, because of the ease with which they reduce iodine, ould be expected t o interfere in the Fischer reagent titration for a ater. However, Linch (18)found that free water could be determined directly in the presence of xanthates n.hen these compounds were dispersed in chloroform. Typical results are givm in Table IV. Similar data reported by Liiich (12) on dithiocarbamates are shown in Table Jr. The analyses for water in di(2-hydroxyethyl) dithiocarbarnatc 55 (’re obtained on samples dispersed in methanol. Chloroform \viis usrd for the other mntrrials.
CHiOH
Johnson (8) found that the reaction was essentially quantitative. Therefore, the Fischer reagent titer would represent the eum of the moles of water plus ascorbic acid. In some cases slight modifications in technique are necessary for the titration for water in certain organic materials. Many of these compounds when nearly anhydrous exhibit a false end point. This is effectively eliminated in all cases by use of a n inert solvent for both sample and Fischer reagent end products. Methanol or pyridine is suitable. As a matter of fact, all titrations should be made in the presence of an inert solvent in order to blank out the moisture originally present in the titration flask. Volatile compounds often may be titrated a t temperatures below their boiling points-for example, methyl chloride a t -30” to -40” C. I n other cases the moisture may be extracted and the anhydrous sample evaporated, after which the extractant may be titrated a t room temperature. Moisture in ammonia gas may be determined in this way. Most carbonyl compounds interfere. Active aldehydes and ketones tend to react mith the methanol of Fischer reagent to form acetals and release ~ a t e r .
+ 2CHaOH +RCH(0CHa)Z + H20 RR’CO + 2CHSOH +RR’C(0CHa jz + H20
RCHO
be cbliminated by conversion of the carbonyl compounds to the cyanohydrins prior to the titration. Table I11 shows data obtained by direct titration using methanol as solvent and taking the first sign of a n end point, using pyridine as solvent, and enipl6ying the cyanohydrin technique which represents the actual water content. Aldehydee, as exemplified by butyraldehyde, tend t o give low rcmlts when pyridine alone is used as solvent. Apparently, a rcartion occurs between pyridine-sulfur dioxide and water. Quinone is reduced by the hydriodic acid always present in Fischer reagent, liberating a molar equivalent of iodine. Pyridine is of no help. The cyanohydrin method is unsatisfactory, a t least for the visual titration, because of extreme darkening of the solution. Possibly electrometric titration will be applicable. Aminrs in alcohol solution, stronger than pyridine, tend t o interfere primarily by obscuring the end point. This interference ir eliminated effectively by use of acetic acid as diluent. Mercaptans (thiols) are oxidized quantitatively by the iodine of Fischer reagent.
’Table 111. Determination of Water in Carbon?-l Compounds Water Found Compound Acetone
Butyraldehyde
Their interference is evidenced by a fading end point. I n many cases the rate of the interfering reaction may be reduced sufficiently by use of pyridinr as solvent to permit a reasonably reliable rapid titration for u ater. In all capes the interference may
0.40 0.10 0.17
M
0.45 0.40 0.36
P C
hi
Pyruvic acid
Chloral 0
P
C
2.20 1.45 1.42
31
0.0.5
P C
0 05 0.04
XI-Xlethanol P-Pyridine C-C yanohydrin
Table IV.
Analysis of Xanthates
Compound ROCSK(N-4)
i
Methyl Isopropyl 1,3-Dimethylbutyl Cyclopentyl 8-Ethoxyethyl a
%
0.65 0.55 0,50
M
P C
Cyclopentanone
(7) (8)
Environnientu Yi P C
XanthateD 97.4 97.3 96.4 96.4 91.5
By iodometric method.
Found, Weight 7% Water 2.4 2.1 4.2 2.8 6.8
Total 99.8 99.4 100.6 99.2 98.3
1073
V O L U M E 23, NO. 8, A U G U S T 1 9 5 1 Ilydroperoxides react selectively with the sulfur dioxide of Fisrher reagent. ROOH
+ SO2 +RHSOd
(11)
As no iodine or water is involved in this reaction, no interference is encountered in the titration for water. Zimmerman (95) published the data on aqueous hydrogen peroxide solutions shown in Table VI. Dialkyl peroxides are relatively stable and do not oxidize hydrogen iodide a t a sufficient rate t o offer any interferenre in the titration for water. Diacyl peroxides, however, are likely to interfere. The more active ones rapidly oxidize the hydriodic acid. Water of hydration has been determined successfully on all types of compounds which offer no interference in the anhydrous etate. On direct titration with Fischer reagent total water is determined-i.e., free plus hydrated. In order to differentiate, some physical separation usually must be employed, such as extraction or distillation. Typical organic hydrates which have been analyzed successfully are shown in Table VII.
method, may be determined quantitatively by direct titration with Fischer reagent. Thus many, but not all, metal oxides react t o consume one mole of iodine per mole of oxide. Typical examples are shown in Tablr I X .
Table VII. Determination of Total Water in Organic Hydrates
L
a
Table V.
Analytical Data for Dithiocarbamates
Cnmnound .. RsN&3Ka
Ll
Carbamate5 Dimethyl 79.0 Phenyl E t h y l 79.5 Di(2-hydroxyethyl) 98.3
Di(2-hydroxyethyl)
Found, Weight % Water 21.2 21.9 1.7
Total 100.2 101.4 100.0
Water, Weight % Addedb Found o n 1.7 .. . 10.4 10.1 15.3 l5,3 33.6 33.6
B y iodometric method. 6 Included water originally found i n sample.
Table VI.
Analytical Data for Water in Hydrogen Peroxide Solutions
Peroxide, Weight % 30.3 20.4 14.9 6.4 3.4
Water Found, Weight % 69.4 80.0 85.4 93.4 96.5
Total, Weight % 99.7 100.4 100.3 99.8 99.9
Moles Water Per Mole Compound 1.99
9,oo
1.05
9.42
0.99
8.96
0.99
10.85
1.00
21.85
2.00
55.3
5.97
A
Reference (36); all others ( 1 9 ) .
Table VIII.
Determination of Moisture in Penicillin Sodium Salt
Sample Weight, ME. 466 466 227 227 484 484
W a t e r Found Mg* 6.08 6.21 3.57 3.53 2.52 2.56
Wt. % 1.30 1.33 1.57 1.56 0.52 0.53
All metal hydroxides studied have reacted completely, one mole of iodine having reacted for each equivalent of base. Carbonates and bicarbonates react in similar fashion. Boric acid and its oxides first are esterified b y the methanol of the reagent and the resulting water is titrated.
HBOi
+ 3CHaOH +B(0CHs)s + 3Hz0 + 3CHzOH +B(0CHs)s + 2H20
(12) (13)
Cupric sulfate oxidizes the hydrogen iodide of spent reagent to form iodine.
+
C U S O ~ 2HI Moisture in salts of organic a d s in general may be titrated directly without interferenre. These include ammonium citrate, zinc stearate, calcium lactate, and lead acetate. Of great potential importance is the application of the Fischer reagent t o the determination of moisture in penicillin sodium salt as proposed by Levy, hlurtaugh, and Rosenblatt (11). These authors reported that the penicillin salt was thermally unstable and hygroscopic. Obviously methods suitable for this analysis were limited. Vacuum desiccation over phosphorus pentoxide was applicable but required several days. The standard ampoule was used as the flask. Platinum electrodes and buret tip were introduced through the rubber cap with the aid of hypodermic needle tips. Typical results 11y this rapid titration method are shown in Table VIII. rlnalyses carried out under these conditions appeared reliable. As no transfers of sample were required, no opportunity was presented for the absorption of moisture from the atmosphere. The Fischer reagent technique also is widely applicable in the inorganic field, although many class interference reactions are observed. This titrimetric method, therefore, can be employed successfully only after the chemist has become familiar with the effects of the anhydrous compounds. In general, inorganic compound reactions are stoichiometric. A point worth stressing is that materials of these types which are anhydrous or nearly so, or in ,which water has been determined bv an independent
Water Found, Weight % 28.42
Compound Oxalic Acida (COOH)z.2HzO Citric Acid HOCCOOH(CHzCOOH)z.HaO Terpin H y d r a t e CioHzoOz.Hz0 Dextrose CeHizOe.Hz0 Chloral CChCH(0,H)z Cyanuric Acid CIN~(OH)~.~HZO Piperazine KHCHzCHzKHCH2CHz.6HzO
+ CUI + 0.512 + HzSOc
(14)
Consequently, Fischer reagent titration of the pentahydrate only shows an apparent 4.5 moles of water of hydration.
Table IX. Action of Fischer Reagent on Oxides React Completely Calcium Magnesium Zinc Silver Mercuric Cuprous Manganese Lead
React Partially or Do N o t React Aluminum Cupric Iron Nickel Yodium Barium
A few compounds, principally hydrates, have been studied which gave net apparent water values-i.e., after correction for the known water of hydration-for which no reaction mechanism n-as apparent. These include the chromates, dichromates, sodium sulfide, phosphomolybdic acid, zirconyl chloride, and basic aluminum acetate. ,411 these compounds were assumed to contain the accepted water of hydration. The compositions of Rome, such as phosphomolybdic acid, are questionable. An interesting means was suggested by Suter (33) for determining free water in some of the interfering inorganic compounds. He first separated the free water b y azeotropic distillation with
1074
ANALYTICAL CHEMISTRY
xylene, after which the distillate was titrated with Fischer reagent, This technique might be improved by use of an agent, such as ethanol, dioxane, or glycol, which forms a homogeneous azeotrope a ith water or which boils higher than Tvater. Then possible holdup of water in the distillation assembly aould be eliminated. Extraction also may be employed for the removal of free water from many compounds. A simple technique might involve the follon-ing: -3. known volume of dioxane is added to the finely divided sample. After a short contact time, an aliquant is \T ithdrax-n and titrated. Extraction, and occasionally distillation, ma! also serve as a means for determining both free and combined iyater in hydrates which do not interfere in the anhydrous state. Total n-ater is determined by direct titration of the sample and free water by a n extraction or distillation process, followed by titration. Xnother variation n a s suggested by Rulfs (28) for the microdetermination of water in minerals.
'
The sample is placed in a combustion boat and inserted in a Pregl muffle. During ignition, moisture is evolved and condensed in a nater-collecting tube a t the exit of the muffle. This water is washed into the titration flask and determined by Fischer reagent titration, using the dead-stop method. Reported results are shown in Table X. Table X.
Determination of Water in Minerals Water, Weight Found Calcd. 36.24 36.10 36,30 20.90 20.90 20.85 14.75 14.75 14.70 12.22 12.32 12.29 13.07 13.35 13.15 28.19 28,77 28.32
Substance CuSOp.5HzO CaS0.I.2 H20 BaClz.2Hz0 CaCzOa.Hz0 Flint Clay (B.S. 97) Bauxite (B.S. 69)
Table XI.
Moisture in Dehydrated Foodstuffs (Weight %)
Sample Mixed vegetables Carrots Tomatoes Protein hydrolyzate Cornstarch
Vacuum Oven Air Fischer 70' C 70' C 1 0 0 ° C Oven Method 6 H o d s 16 Ho& 7 Hour; 130'
Toluene Dis-
6. tillation
6.60 6.20 3.07
5.93 4.22 2.57
6 .OO 5.42 3.20
2 44 13.9
2.48
2.41
.,.
,..
... ... ... , . .
14.1
... ... ...
... ... ...
, ,. 14.0
2.40 13.2
These compounds are representative of those 1% hich offer interference on direct titration or from which the FTaater is extracted slon Iy and often incompletely. ilpplications of the Fischer reagent to the determination of n-ater in commercial materials, particularly natural products, appear promising. Hoxever, in many cases it is difficult t o demonstrate the accuracy of the titration method because of the lack of a n absolute calibrating procixdure. This is particularly true of foodstuffs. For example, consider the data reported by Schroeder and Xair (SO) in which they compared the Fischer reagent titration, after a preliminary &minute extraction step using methanol under reflux, with standard AOAC methods (Table XI). Usually the Fischer reagent results were higher than accepted procedures. However, the steady increase in the vacuum drying results from 6 to 16 hours indicated that possibly moisture was incompletely removed. On the other hand, the high 'values by the titration method may have been due to thermal decomposition during the refluxing step. Schroeder and ?;air suggested a possibility that merits further investigation. Water of hydration may not be removed by oven drying, whereas the Fischer reagent titration 1% ill include this combined moisture. These investigators found that the water of hydration of monosodium
Table XII.
Fischer Reagent Titration for Moisture
Protein Hydrolyzate Method Onion Powder Intermittent titration 3.95 2.41 Methanol extraction (cold) 3.97,3.93 2.56, 2 , 5 2 Methanol extraction (reflux) 2 min. 3.90 2.47 2.44 (3.30)Q J min. 3.90 (3.78Ia 15 min. 3.84 2.51 30 min. 4.01 2.54 Excess reagent (10 t o 20 ml.) 2 hours 4 92 (10 ml.) 5 . 2 2 (20 ml.) 4 hours 5.98 19 hours 6.28 Excess reagent 6 . 9 8 (96 hours) 4 . 6 5 ( 4 hours) 10 ml. 6 , 8 3 (4 hours) 30 mi. 7 . 8 3 (1 1 hours) a 5-Minute values for samples used in excess reagent tests.
glut'amate was titrated readily TTit,h Fischer reagent, whereas litt,le more than the adsorbed moisture was removed from t.his salt after vacuum drying for 16 hours a t 70" C. Schroeder and S a i r (SO) observed some evidences of interference in the determination of moisture in onion powder and protein hydrolyzate, which seemed to be associated with excess Fischer reagent. Table XI1 gives a comparison of moisture values obtained under different conditions. First, intermittent titration of the sample suspended in methanol a t room temperature-i.e., adjustment to t'he end point a t timed intervals. Over periods of 6 to 12 days the values tended t,o increase in linear fashion. The values shown were obtained by extrapolation back t,o the ordinate of a plot of moist,ure versus t,ime. Secondly, methanol extraction a t room temperature ; aliquots were removed periodically and titrated. The data represent total water found, using methanol originally containing 0.03 to 0.06% and 0.54% water, respectively. LIethanol refluxing showed little variation. However, when excess reagent JT-as alloxTed to stand with the sample, extremely high values resulted. *Ascorbic acid is the only compound reported in dehydrated foods which would cause high values. HoTvever, Johnson ( 8 ) demonstrated that the quantity of ascorbic acid found in these foodstuffs was equivalent to no more than 0.03% water. Further investigations should include a more thorough study of the effects of particlr size on "apparent" moisture recoveries. More work on extractions employing the Waring Blendor or a modification should also aid in ascertaining the nature of the apparent interferences. Finally, the compositions should be studied of these erratically behaving compounds in order definitely t o ascertain potential interferences. LIcComb's study of several protein materials showed no wide variations, even though excess Fischer reagent was in cont'act with the sample for periods up to 3 hours (13). Brobst ( 1 ) employed the Fischer reagent titration successfully for the determination of moisture in lecithin and crude soybean oils. Several interesting applications of t,he Fischer reagent titration method have been reported for the determination of moisture in polymers and their intermediates. Many of these materials are thermally unstable and, therefore, cannot be analyzed by oven drying or azeotropic distillation methods. For such chemicals the Fischer method may be the only feasible technique available. I n order to apply this titrimetric method safely, however, a thorough knowledge of potential interferences must be obtained. Formaldeh?.de in aqueous solution or in any of its polymeric forms does not interfere. However, water in dimethylol urea or other IOK molecular weight polymers of formaldehyde and urea cannot be determined directly a t room temperature, presumably because of further condensation of the polymer or condensation with the methanol of Fischer reagent. I n either case m t e r would be formed. Only a t a temperature of about -40" C. does this interfering reaction become sufficiently slow to permit direct titration for free water originally present, in the sample (19). On the other hand, urea-formaldehyde resins may be titrated a t room temperature. Cornish (3) demonstrated that this and several other molding pon-ders could be analyzed successfully. His technique usually involved dispersing the sample in meth-
1075
V O L U M E 23, NO. 8, A U G U S T 1 9 5 1 anol, refluxing for a short time, and titrating the cooled heterogeneous mixture. Typical result's are shown in Table X I I I . \There possible, the d a t a were checked by a n azeotropic distillation procedure, which gave results nit'hin 0.1 to 0.270 of those found by the more rapid titration method. Other successful applications include the direct determination of moisture in polythene (2;73)>vinsol resin and rosin size (fb), shellac-alcohol solutions ( 2 7 ) ) paints and varnishes (34), and synthetic rubber (GR-S) (29). Khile developing his technique for the analysis of paints and varnishes, Swann (34) reported that zinc oxide was t,he only powdered pigment found which interfered in the titration. Rush and Kilbank (29) found that a met,hod employing Fischer reagent titration was the only satisfactory technique for the determination of moisture in GR-S. Thcy covered the sample with benzcne, refluxed, added ethanol, and distilled the honiogeneous ternary: n-ater-l)erizerie-etha~iol. T h e distillate was titrated with Fisc1ic.r rcttgent. lIoist,ure found in compounded GR-S stock varied from 0.042 to @.(i2070. The determination of moisture in explosives represents another field in which the Fisclier reagent titration should find wide application. Typical applications are shon-n in Table SIT'. Further investigations should be made in this field. For example, n-hen trinitrotoluene was dispersed in methanol, the mixture formcld a deep red color during titration n-ith Fischer reagent whic~hobscured the visual end point (26). Xn electrometric nirthod should be tried t o determine whether an interfering reavtion is involved. Certainly one ~ o u l dnot predict any abnormal behavior. Applications of the Fischer titration method to analyses for moisture in petroleum products have been covered adequately in the literature (90). Satisfactor!. techniques have been reported for condensalile and noncondensable gases employing extraction by methanol a t -78°C. Direct titration methods are commonly employed for many liquids and solids, either as homogeneous solutions in an inert solvent or after liquid-liquid extraction into a n immiscible liquid. These titrations are mad? most easily in the presence of sufficient alcohol or pyridine to assure a polvent for the Fischer reagent end products. Otherivise, false end points may be encountered. In applying the titrimetric method to carbohydrates, the absolutc accuracy often is unknonm because of the lack of a suitable reliable standard method. This applies particularly to the starches which normally are analyzpd by oven-drying techniques. Literature reports have given conflicting conclusions, but for the most part have indicated that Fischer titration is relialile provided a preliminary extraction st:'p-?.g., cold or hot extraction with methanol-is employed. Applications to sugars (direct titration j, paper, and fabrics (extraction before titration j have been adequately demonstrated ( 2 2 ) . At least one extensive study has been reported on the determination of moisture in native wood (15). Room temperature methanol extraction ef-
Table XIII.
RIoisture in Plastic JIolding Powders Water Found. Weight %
Material Crea formaldehyde Phenol formaldehyde Acetyl cellulose acetate Cellulose acetate Poly! inyl alcohol
Table XIV. Material Sitrocelldose Sitroglycerin
1 1 . 6 , 11.6. 1 1 . 7 4 . 8 3 , 4.83 1 81, 1 80 0 24, 0.25, 0.26 1.82, 1.81
Moisture in Explosives
Water Found, Welght % Fischer titration Other method
2.8 i0.0 0 . 2 8 5 =k 0 . 0 0 5 0 . 0 8 i. 0 . 0 0 Gunpowder (grained) 0.945 0.005 Priming explosive 12.4 i 0 . 1 Smokeless powder 0.27-0.29 a Azeotropic distillation. b Desiccation.
2.9a 0.29b
0.08~
+
C
1 11 .02 9 0.265-0.285b Oven drying.
Reference 14 14
3 26 6
fected about 96%removal of moisture from small piecesof cypress, Douglas fir, or oak in 1 hour and up to 98.5% in 3 hours. More rapid extractions were feasible for the determination of moisture in wood pulp ( 1 4 ) and sawdust ( 2 7 ) . Preliminary d a t a indicate that wool may be analyzed rapidly and precisely. I n one series of experiments, employing a 30-minute methanol extraction a t room t e m p e r a h r e before titration, a sample of wool analyzed 0.08% found by oven drying 9.30 i 0.01% compared to 9.26 a t 102" (9f). -% comprehensive study of the Fischer reagent titration for moisture in wool would be desirable, as a time-saving and reliable procedure should result. The Fischer reagent should be useful for the det,erniination of moisture in soaps and soap products, Values obtained on direct titration would be subject t o a mole for mole correction for any free alkali present. Sormally this correction would he small. I n one reported study ( 2 6 ) Ivory flakes were dispersed in niethanol and titrated immediately with Fischer reagent. Valucs of 3.02 i 0.0270 moisture compared favorably with a result of 2.0970,:obtained by azeotropic distillation. An electrometric study of direct titration for nioist,ure in coal might result in a rapid routine anal This discussion of necessity has covered only a portion of the applications, real and potential, of the versatile Fischer reagent. Obviously in many specific cases more rapid techniques are available, but, no procedure approaches this titrimetric method in general applicability. CertainIy the Fischer reagent should be ronsidered in all studies involving the determination of water.
+
ACKNOWLEDGMENT
The author is grateful t,o C. E. iishby for the experimental data on water in adipic acid. LITERATURE CITED
(1) Brobst, K. XI., AKAL.CHEM.,20,939 (1948). (2) Carter, R. J., and Williamson, L., Analyst, 70,369 (1946). (3) Cornish, G. R., Plastics ( L o n d o n ) , 1946,99. (4) Gester, G. C., Chem. Eng. Progress, Trans., 1, 117 (1947). (6) Hardy, J., Bonner, W ,D., Jr., and Koyes, R. hI., IXD.ENG. CHEM., A S A L . ED., 18, 751 (1946). (6) Johansson, A , , Acta Chem. Scand., 3, 1058 (1949). (7) Johansson, A , . Suensk Papperstidn, 50, No. 11B, 124 (1947). (8) Johnson, C . M., IND.ESG. CHEM.,ASAL. ED., 17,312 (1945). (9) Jones, G. K., P a i n t Manuf., 15,360 (1945). (10) Rieselbach, R., ISD. ESG. CHEM.,.4xar.. ED., 18, 726 (1946). (11) Levy, G. B., Murtaugh, J. J., and Rosenblatt, M., Ibid., 17,
193 (1945). (12) Linch, -4.L., paper presented before Analytical Section, Delaware .1CS Symposium. Kewark, Del., Jan. 14, 1950. (13) RlcComb, E. .%., d X . i L . CHEM., 20,1219 (1948). (14) RIcKinney. C. D., and Hall, R. T., ISD.ENG.CHEM.,. ~ ~ N A LED., . 15,460 (1943). (16) Mitchell, J., Jr., Ibid., 12,390 (1940). (16) Mitchell, J., Jr., and Smith, D. >I., "Aquametry," pp. Gff, New Tork, Interscience Publishers, 1949. (17) Ibid., pp. T l f f . (18) Ibid., pp. 135ff. (19) Ibid., pp. 156ff. (20) Ibid., pp. 162ff. (21) Ihid., pp. 194ff. (22) Ibid., pp. 200ff. 123) Ihid.. nn. 211ff. (24) Ibid.. b: 216. (26) Ibid., p. 217. (26) Ibid., p. 221. (27) Rennie, R . P., and Rfonkman, J. L., Can. Chem. Process Inds., 1945,366. (28) Rulfs, C. L., Mikrochem. t e r . Mikrochim. Acta, 33, 338 (1948). (29) Rush, J. C . , and Kilbank, S. C., I n d . Eng. Chem., 41,167 (1949). (30) Schroeder, C . I T . , and Sair, J. H., . 4 r a ~ CHEX, . 20,452 (1948). (31) Seaman, TT., McComas, W.H., Jr., and Allen, G. A.,Ibid., 21, 510 (1949). (32) Smith, D. M.,Bryant, TT'. M. D., and Mitchell, J., Jr., J . Am. Chem. Soc., 61,2407 (1939). (33) Suter, H. R., IND.EKG.CHEM.,ANAL. ED., 19,326 (1947). (34) Swann, A I . H., I b i d . , 18,799 (1946). (35) Zimmerman, d.,Fetle u.Seifen, 46,446 (1939). RECEIVED September 2 3 , 1950.
