Improved Procedure for Karl Fischer Microtitrations J. S. WIBERLEY Socony-Vacuum Laboratories, Brooklyn, S. Y . A procedure for Karl Fischer titrations was deiised in order that nontechnical operators might determine water in a number of petroleum products such as kerosene, fuel oil, gasoline, and lubricating grease with a minimum of maintenance and difficulty. A microburet of 1-ml. capacity was designed to protect the reagent from atmospheric moisture. Many small improvements were made in the technique to simplify standardization of the reagent and analyses of samples. Karl Fischer reagent stored under these conditions does not exhibit rapid deterioration. Weekly standardization has been found sufficient for most work, and it has been possible to keep a buret filling in occasional use for several months. The method is easily adapted to materials other than petroleum products. It is convenient and economical and is recommended in place of the usii a1 macroprocedure.
T
HE Karl Fischer method is nor7 standard analytical practice
gel, alumina, and Drierite ( 5 ) and tubes of these materials only temporarily halt the passage of moisture from the air into the reagent. For this reason the recommended buret has but two possible openings to the air. One is the stopcock of the buret itself, the other is a stopcock at the top of the reagent reservoir. Both stopcocks are kept calosed except when the buret is in use. Khen titrations are to be performed, the upper stopcock is opened to permit air to enter the buret as reagent is withdrawn. -2 tube containing phosphorus pentoxide mixed with asbestos serves to drj- the air which thus enters the system. These precautions have reduced the rapid and widely reported deterioration of the standard solution. One advantage inherent in a micromethod is the comparatively small amount of standard solution needed for the performance of a large number of analyses. This may seem a minor feature. but in the author’s company, 13 hich has many laboratories using the method, the annual savings in reagent cost are currently estimated a t about $2000. Since the reagent is well confined and the amount of pyridine per titration is small, the odor is hardly noticeable. If only small saniples are available, the micromethod is a necessity. The limiting sensitivity is 0.01 mg. of water which represents 0.01% on 100 mg. of grease or about 3 p.p.m. on 5 ml. of gasoline.
for the determination of water in a great many products (11). n’evertheless, there are a number of minor disadvantages n hich make the method something less than convenient and discoui age the occasional user. The most serious objection is that the reagent has been found to deteriorate rather rapidly; daily standardization of the reagent is universally recommended. Recently two separate solutions, one containing pyridine, sulfui dioxide, and methanol, and the other containing iodine and methanol, have been suggested as one means of avoiding the deterioration (9, 14). Other drawbacks are the expense of the reagent, or the unpleasantness of preparing it, the pyridine odor, and the need for frequent replacement of rubber connections and desiccants. As compared with most analytical methods, the amount of time required for standardization and maintenance is excessive.
m FILLING FUNNEL-
Pe 0, DRYING TUBE _-CAPILLARY
0.5rnrn. I D
STOPCOCK No.3
SEE DETAIL
APPARATUS
Figure 1. Karl Fischer Microburet
Microburet. The all-glass gravity-feed microburet is shown in Figures 1 and 2. Provision is made to exclude moisture from the reagent by a stopcock !Then not, in use and by a drying tube when in use. Standard taper 15-ml. flasks fit the glass joint at the tip. Plain flasks are usrd for samples which are weighed, and side-arm flasks equipped with a stopcock pipet are used for samples which are measured by volume. The complete buret with 1- and 5-ml. pipets may be ordered from Metro Industries, 29-28 41st Bve., Long Island City 1, N. E’. Syringe Pipet. This pipet will hold 25 microliters and is used in measuring water for standardization of the Karl Fischer reagent. I t can be obtained from Microchemical Specialties Co., 1834 University Ave., Berkelev 3, Calif. Magnetic Stirrer. The sti&er wsbs obtained from Arthur H. Thomas Co., West Kashington Square, Philadelphia, Pa. A small stirring bar can be made by cutting a 20-mm. length from an iron nail and sealing it in glass tubing. Titration Stand. .4 lighted white background is essential for observation of end points.
The technique, the description of which follows, has been evolved in an attempt to minimize these nuisances and to encourage general use of the Karl Fischer method. Most of the disadvantages can be overcome satisfactorily by using a microburet of the type described and by taking certain precautions in its use. Karl Fischer reagent should be treated as an extremely po-xerful desiccant. 1t:aill extract water from calcium chloride, silica
Methanol. Anhydrous, less than 0.0570 water. Karl Fisher Reagent. Solution 1. To the contents of a freshlv opened 1-pound bottle of anhydrous methanol add 50 grams of resublimed iodine. Stopper and allow to stand until the iodine is completely dissolved.Solution 2. Mix 160 ml.of anhydrous (less than 0.1% water) pyridine with 2 pounds of mhydrous methanol in a 0.5-gallon
MAGNETIC STIRRER
REAGENTS
656
V O L U M E 23, NO.
4, A P R I L 1 9 5 1
651
bottle fitted with a Bakelite scrc~vc:tp. Chill by surrounding the bottle with crushed ice in a pail. Mark a dry 50-ml. graduate a t the 28-ml. level; connect a rubber tube about 30 cm. long to the outlet of a sulfur dioxide cylinder containing the dry gas of 99.5% purity. Invert the cylinder and support i t in a position above the graduate. Collect 28 ml. of liquid sulfur dioxide and pour it into the bottle containing the pyridine and methanol. Cap the bottle and mix the contents thoroughly. (Wear goggles and exercise care in mixing reagents. Handle the pyridine and sulfur dioxide in a ventilated hood. Heat is evolved when sulfur dioxide is added to the pyridine and methanol, but, a t the chosen concent,ration the mixture should remain cold if previously chilled as directed.) To make reagent sufficient for one buret filling (1 pint), mix 145 ml.of Solution 1and 330 nil. of Solution 2. For convenience these auantities of the two solutions m a r he kept ready for mi\;ingAirismaller bottlrs.
rQ
cocks t o drain out t'he cleaning solution. I n a similar manner rinse well with water and finally with several portions of anhydrous methanol. Dry and then grease both stopcocks with Sisco 300 stopcock lubricant. Introduce a small portion of Karl Fischer reagent and swirl over the sides t o absorb traces of water. Pour out this reagent, place the buret in its stand, fill the top funnel with reagent, and close with the drying tube. Allow the reagent t o drain into the reservoir. Rinse the capillary buret b y filling and draining i t repeatedly. Keep the stopcock t o the filling funnel open as a n air vent while the buret is in use, but close i t when titrations have been completed to protect, the reagent from moisture. Subsequent fillings with reagent should not requiw :ttiditional cleaning. Pour out t,he residual reagent, wipe and regwase the loner stopcock, replace the desiccani , and introduce i'rvsh Karl Fischer reagent, as before If the &ret tip should become clogged with grease, a hot object such as a spatula heated in a R a m held against the side of t,he tip will usually melt the g r v : ~ ~and ! allow i t to run out, pushed by the rcagent above it. If grease should get on the inside walls of the capillary buret, clean by dra\\-ing up xylene through the lower stopcock with a bulh and rubber tube applied a t the top of the buret. If yc~llou-salts are found on the outside of the lower stopcock, regrease t,he $topcock by the following procedure which does riot requirc removal of the standardized reagent. With both st,opcocksclosed, invert the buret :inti place it in the buret stand FO that all the reagtxnt runs into the reservoir. Iieniovc the stopcock, wipe the plug aud barrel with a dry cloth, grease, and insert, the plug. If this is done quick!y, w r y littlc water \vi11 enter the buret.
I-,,, .
PIPET
7 mm.OD STAND.4RDIZATION
Add about 1 nil. of anhydrous niethariol to a side-arm flask and connect to the buret. Fill APILLARY AIR VENT N 1-ml. stopcock pipet tmo the mark n-ith 0rnrn.LG 0.5 mm. I D methanol and insert in thc side arm. Titrate the methanol already in the flask with Karl Fischer reagent, dirriug with t,he stirrer. A VAN S L Y K E PIPET light yellow color appears and increases in inEt F L A S K tensity. At the end point there is a sudden change t o an anibcr color. At this point the water cont,ained in the mrdicinol and in the air SIDE DETAILS of the flask has been consumed. Disregard the volume of reagent necessary to obtain this end Details of liar1 Fischer 3licroburet Figure 2. point. Drain into the flask exactly 1 ml. of methanol from the stopcock pipet. Titrate to a new end point and record the amount of Karl Fischcr roagent used in this titration. Iicpeat until checks are The Lvater equivalence factor of the resulting reagent is deobtained within a t least 0.02 ml. pendent chiefly upon the water contrnt of the methanol, but i t Prcparc a stand:trd solution of water in another portion of the should be about 1 mg. of ivater per ml. of reagent. If necessary, s:mie methanol by pipetting 25 microlitcr~of water into a 10-ml. the strength can be adjusted b y adding iodine solution to increase volumetric flask and making up t o volume wit,h methanol. Take 1 ml. of this solution, titrate, and obtain checks as before. the factor or methanol to decrease the factor. The final reagent Calculatc t,hc water equivalcmce of the Iiitrl Fischer reagent should not contain more than 300 ml. of Solution 1, however, beL)y the formu!a cause a safe excess of sulfur dioxide must be present. 20
I
,
Ethylene glycol. Anhydrous, less than 0.2y0 water. Comniercial glycol must be dried by distillation. T o glycol of technical grade, add 10% by volume of benzene and distill. Collect the fraction of glycol boiling in the range 195' to 198" C. Chloroform. Dry lvith silica gel. Grease Solvent. Mix equal volumes of xylene, dried with silica gel, and anhydrous methanol. Desiccant for Drying Tube. Shake phosphorus pentoxide with about ten times its volume of short-fiber asbestos. Stopcock Grease. Sisco 300 stopcock grease, sold by S w d i s h Iron and Steel Corp., 17 Battery Place, S e w York 4,S. Y. Do not attempt to substitute usual stopcock greases or a silicone grease. PREPARATION OF APPARATUS
The buret should be cleaned and dried as follows: Int,roduce 50 ml. of dichromate-sulfuric acid solution through the top stopcock; the three-way stopcock connects the buret to the tip and allows for the escape of air. Then close the stopcock t o the filling funnel and give the three-way stopcock a half turn t o connect the reservoir to t,he buret. Tilt the buret back and forth to wet all interior surfaces of the buret and its reservoir with cleaning solution. Finally, invert t,he buret and open both stop-
~vhere B = equivalence, nig. of water per ml. H = ml. of reagent t o titrate methanol S = ml. of reagent to titrate water in methanol solution PROCEDURE
Liquids Containing Less Than 0.1% Water. Introduce approsiniately 1 ml. of methanol into the flask. T o titrate fuel oil, which gives poor phase separation with methanol, subst,itute 1 ml. of anhydrous ethylene glycol. T o titrate emulsified water in hydrocarbons, substitute 1 ml. of chloroform. Fill a 1- or 5-ml. st"opcock pipet ~ i t the h sample and insert into the side arm of the titration flask. Titrate to an initial end point and then add the sample from the stopcock pipet. If the sample is immiscible with the solvent, stir the two phases vigorously during the titration. Add the reagent slowly and interrupt the stirring several times to allow the phases t o separate. Observe the color in the alcohol (lower) layer. At the end point the amber color is permanent and becomes much darker on addition of more reagent. If samples impart a color to the alcohol layer, prepare a sample for comparison purposes by adding magent i n definite excess and then
658
ANALYTICAL CHEMISTRY
Since the prime object of the proposed technique is simplicity, the electrometric end Sample Micro Karl Fischer Comparison Value and Method point was intentionally exKerosene (saturated a t 32' F.) 32, 35 p.p.m. 30, 32, 34, 36, 36 p.p.m.. macro Karl Fischer c l u d e d . Unquestionably, it Kerosene (saturated a t 100' F.) 116, 119, 129 p.p.in. 121 p.p.m., macro Karl Fischer Grease (1ithiu.m) 0 . 1 1 , o.llyo Trace, xylene distillation (3) would have some advantages Grease (aluminani) 0.15, 0.13?& Trace, xylene distillation such as increased accuracy, Grease (lime) 0 . 8 9 , 0.9670 1.0%, xylene distillation Grease (lithium) 1.05% 1 . 0 0 7 , water added to grease sample in flask particularly with dark colored l.OlY0 Grease (lithium) 1 . 0 2 # , water added to grease sample in flask Grease (soda) 2 59, 2.70Y0 2 . 6 2 % , water added to grease using micro worker (7') samples, but it would also add __ to the cost and to the difficulty of maintenance in small laboratories. adding sufficient standard water in methanol solution to remove d very important point is that, in the technique described, the t8he free iodine color. Then titrate subsequent samples until flask is never opened to the air during a titration. Kater in the they are perceptibly darker t,han this color standard. air must not be forgotten. In some early work, gasoline was Liquids Contain$g More Than 0.1% Water. Liquids, which can be diasolved in an anhydrous solvent such as methanol, titrated using a blank as in the procedure for solids. I t was surcan be diluted so that a 1-ml. aliquot will contain in the neighborprising to find that gasoline samples always required less standard hood of 1 mg. of water. Liquids may also be weighed using the solution than was consumed by the blank, and, as the sample procedure for solids. volume was increased, still lower titration values were obtained Solids. Dry a titration flask, without side arm, together with Then it was realized that thwe was more water in the air than in its stirring bar by washin out with dry solvent and heating to dryness on a hot plate whik blowing a stream of air into the flask. the gasoline. Consequently, transferring the gasoline into the (If the flask has been standing in contact with air for some time, flask displaced more water than it introduced, so that the total it should first be rinsed wit'h Karl Fischer reagent. to mnove the water content of the flask became less than that of theempty film of moisture which is present on apparently dry glass.) flask. Cool and pipet into the flask exactly 1 ml. of the solvent sclected-met,hanol in most cases but xylene-methanol for lubriFading end points are a sure sign that something is vrong. cating grease. Immediately flush out the tip of the buret and Moist air may be leaking into the flask. The stirring bar may be attach the flask by means of the ground-glass joint. Titrate to cracked, or an interfering substance may be present. If everya permanent amber end point. Repeat this blank until checks are thing is in proper order, end points ndl hold for many minutes obtained. This blank represents water in the air in the flask as well as water in the 1 ml. of solvent. It changes and must be before moisture eventually leaks back through the capillary vent. determined daily. I t must be remembered that apparently dry glass has a film of Dry a plain flask, cool, and weigh. Weigh into the flask a moisture which is removed by escess Karl Fischer reagent. A sample of t'he size e s p c t e d to contain a little less than 1 mg. of flask will give high blanks and show some fading until this film is water. Pipet 1 ml. of solvent into the flask and titrate to a permanent amber end point. removed. Mitchell (10, 1 2 ) found that reagent of a strength of about 1 A two-phase system can also be used to circumvent the intermg. of water per ml. seems to keep better than reagent of 2 or 4 ference of a dark color in the sample. Very dark lubricating mg. per ml. During the winter months, when the humidity is greases were successfully titrated using 1 ml. of methanol and 2 low, reagents of about 1 mg. per ml. in strength stored in the ml. of light petroleum solvent. described buret have lost about 0.002 mg. per ml. per day. In DISCUSSION isolated. instances the titer has not seemed to change for periods of 2 v-eeks. In the summer months the loss is several times Over the past 2 years this method has been employed for the greater. Most of the loss seems to be directly attributable to analysis of water in such products as kerosene, gasoline, fuel oil, leakage of moisture into the buret and not to side reactions. For lubricating oil, water-in-oil emulsions, lubricating grease, alcoan experiment, a solution of over 4 mg. per ml. was stored in the hols, glycol, glycerol, and soap. It has been handled by nonburet under ideal conditions for 2 months-that is, during Janutechnical personnel with a minimum of difficulties. The method ary and February, stopcock well lubricated, fresh desiccant in contains many small improvements which were found to make the the drying tube, and buret thoroughly dried by long previous use operation more convenient and reliable than the macroprocedure but rarely used during the period of the experiment. The averpreviously employed. Comparison of results run by this method age loss was 0.02 mg. per ml. per day. On dilution with methwith results obtained by other methods showed satisfactory anol, the same reagent became more stable. It might be conagreement. Some of them are shown in Table I. Since the aprluded from this single evperiment that both the strength of the plicability of the Karl Fischer method to petroleum products has reagent and the degree of protection from atmospheric moisturp already been well established (1,2, 4, 6, 8),it did not seem newsare factors in the stability of Karl Fischer reagent. It is not sary to accumulate much data to verify the accuracy. claimed, therefore, that side reactions do not occur. It is only The ilmerican Society for Testing Materials method ( 3 ) for pointed out that the unavoidable deterioration which results the determination of water in grease has not been found suffifrom side reactions is very slight in relation to the avoidable ciently sensitive to use when the sample contains less than deterioration which results from the entrance of moisture. 0.25y0 of water. Replacement of this method by the Karl Fischer method has yielded better precision and has permitted close control of dehydration during grease manufacture. CONCLUSION Interferences have not been a great source of trouble. In gasoBecause of the advantages of comparative stability and low line and fuel oil, mercaptans are titrated so that, if present in concost of reagent, speed of analysis, absence of pyridine odor, and siderable amounts, the mercaptan content in parts per million decreased maintenance time, the direct visual Karl Fischer multiplied by 0.3 must be deducted. Tetraethyllead does not micromethod, with the modifications detailed in this paper, is interfere, as has been shown by Snyder a n d ' c l a r k (15). Free recommended to replace, where possible, existing macromethods alkali may be titrated in greases, but in most cases the correction for routine water determinations. is negligible. I n used products free metallic iron may interfere by reacting with the iodine. The prior distillation recommended LITERATURE CITED by Roberts and Levin (18 ) is specifically designed to avoid interferences and could be used, if necessary. So far none of the (1) Acker, M. M., a n d Frediani, H. -4., IND. ENG.CHEM.,ANAL. ED.,17, 793 (1945). additives used in these products have been found to interfere.
Table I.
Water Contents Determined by Micro Karl Fischer and by Independent Comparison Methods
V O L U M E 2 3 , NO. 4, A P R I L 1 9 5 1 (2) Aepli, 0. T., and McCarter, W. S. W., Ibid., 17, 316 (1945). (3) Am. Soc. Testing Materials, “A.S.T.M. Standards on Petroleum Products and Lubricants,” A.S.T.M. Designation D 128-47, 1949. ~ . ~ . (4) Ibid., D 268-49. (5) Bryant, W. M. D., Mitchell, J., Jr., Smith, D. M., and r\shby, E. C., J. Am. Chem. SOC.,63, 2924 (1941). (6) Fischer, K., Angew. Chem., 48, 394 (1935). (7) Hain, G . M., Am. Sac. Testing Materials, Bull. 147, 86 (1947). (8) Hawthorne, W. P., correspondence from Augusta Refinery, Socony-Vacuum Oil Co., Laboratory Letter 680, November 23, 1939 (revised by W. C. Fry). (9) Johansson, A, Svensk Papperstidn., 50, KO.11B, 124 (1947).
6s9 (10) Mitchell, J., Jr., round-table discussion on Karl E’ischer reagent, Division of Analytical Chemistry, 116th Meeting, AM. CHEY.
SOC.,Atlantic City, N. 3. (11) Mitchell, J., Jr., and Smith, D. M., “Bquametry,” New York, Interscience Publishers, 1948. (12) Ibid., p. 98. (13) Roberts, F. M., and Levin, H., .4xa~.CHEW,21, 1553 (1949). (14) Seaman, W., McComas, W. H.. Jr., and Allen, G. A., Ibdd., 21, 510 (1949). (15) Snyder, €2. E., and Clark, R. O., paper presented before the Division of Analytical and Microchemistry, 112th hfeeting, AM. CHEY.SOC., New York, N. Y . R E C E I I E DAugust 11, 1950.
Determination of Fluorine in Organic Compounds Microcombustion Method HOWARD S. CLARK, Illinois State Geological Survey, Urbana, 111. A reliable and widely applicable means of determining fluorine in organic compounds has long been needed. Increased interest in this field of research in recent years has intensified the need. Fluorine in organic combinations may be determined by combustion at 900” C. in a quartz tube with a platinum catalyst, followed by an acid-base titration of the combustion products. Certain necessary precautions and known limitations are discussed in some detail. Milligram samples suffice, and the accuracy of the method is about that usually associated with the other halogen determinations. Use of this method has facilitated the work upon organic fluorine compounds in this laboratory and i t should prove to be equally valuable to others.
M
ANY methods and variations of them have been proposed
for the determination of fluorine in organic substances ( 4 , 7-9, 11, 19, 14-16, 20, 23), but no method has been found generally applicable to a wide variety of combinations, particularly those involving the presence of trifluoromethyl groupings, other halogens, sulfur, and compounds of little or no hydrogen content. The determination of fluorine essentially involves two steps: release of the fluorine, and assay of the fluorine so released. I n this laboratory combustion with oxygen in a fused quartz or Vycor glass No. 7900 (96% silica) tube was considered the most promising method for the former and an acid-base titration the most direct for the latter. Furthermore, such a procedure should allow the use of standard equipment commonly available. Theoretically, the reactions may be written as follows:
RFa
+ SiOz + +zCOn + zH20 + SiFd 2 H 2 0 + SiF, +SiOz + 4HF 4HF + 4XaOH +G a F + 4 H 2 0 202
However, in attempting to apply the above equations to the determination of fluorine in pure organic compounds, consistently low results were obtained. -1 critical consideration of possible combustion products led to the conclusion that boron, a constituent of some combustion tubes and receiving vessels, could be appearing in these products as a boron-fluorine combination which was not titratable with sodium hydroxide under these conditions. If this boron-fluorine combination is assumed to be boron trifluoride, the following equations may represent the reactions involved:
+ HzO +HBF30H (6) BF, + H F +HBFa (6) HBFl + H,O HBF3OH + H F ( 2 4 ) BFI
F!
Further hydrolysis has been shown to be very slow under titration conditions (10, 24), and therefore may be disregarded. In view of the above equations, it appeared that the troublesome end product of boron, which was not titrated directly with sodium hydroxide, might be monohydroxyfluoboric acid ( H E F3OH). This also might explain previous observations concerning the possible cause of low fluorine analyses when boron w m present (19). Evidence of the existence of this monohydroxyfluoboric acid has been established by x-ray studies ( 1 ) and by the preparation of alkali metal salts ( l S ) , but there is some question as to whether it can be titrated directly with alkali ( 2 , 22). In practice it was found that if the products of combustion were swept into water, warmed to drive out carbon dioxide, and immediately titrated to a phenolphthalein end point with 0.01 S sodium hydroxide, there remained a small increment which in turn could be titrated after the addition of mannitol. The addition of polyhydroxy compounds such as mannitol, glycerol, or invert sugars to boric acid to allow direct titration with alkali is a common practice. The author observed also that if the mannitol increment was multiplied by three and added to the initial titration, values very close t o theoretical were obtained. This supports the theory of the presence of monohydroxyflucboric acid, in which the ratio of fluorine to acidic hydrogen is 3 to 1. The equation for the titration of the mannitol increment may be represented as follows: HBFBOH
+ NaOH
h9anmtol
KaBF,OH
+ HzO
DESCRIPTION O F METHOD
Equipment (see Figure 1 ) . The equipment listed is advised where a permanent setup is desired. For occasional determinations an ordinary halogen combustion tube ( 3 , 18) according to ACS specifications but made of fused quartz with borosilicate