plied to other types of samples. The results obtained by carrying standard solutions of chromium(II1) containing 250 p.p.m. of chromate through the procedure are given in Table I. Reproducibility is good. Recovery is essentially complete, as indicated by comparison of the standard curve with one prepared from standard potassium dichromate. The procedure has been used successfully for the determination of chromium(II1) in water treated with relatively large amounts of sodium chro-
mate. I n this procedure, the chromium is oxidized by the alkalinebromate method ( 1 ) . Other possible oxidation steps which were not attempted include permanganate ( 2 ) or perchloric acid (3).
p. 93, hlcGrav-Hill, New York 1944.
(3) Snell, F. D., Snell, C. T., “Colorimetric Methods of Analysis,” 3rd ed., p. 269, Van Nostrand, New
York, 1949. (4) Union Carbide Kuclear Co., ‘‘Determination of Chromium In Lithium Salts,” unpublished.
LITERATURE CITED
(1) American Public Health iissociation, “Standard Methods For ?amination of Water and Sewage, 9th ed., p. 56, 1946. (2) Snell, F. D., Biffen, F. hI., “Commercial Methods of Analysis,” l e t ed.,
RECEITED for review May 10, 1957. Accepted February 21, 1958. Work carried out under Contract KO.W-7405-eng-26 at the Paducah Plant operated by Union Carbide Nuclear Co. , A Division of Union Carbide Corporation, for the ‘c. S.Atomic Energy Commission.
Determination of SiIanoI with Grignard Reagent F. 0.GUENTHER’ General Elecfric Research laborafory, The Knolls, Schenecfady,
b The use of methylmagnesium iodide was investigated for the gasometric determination of the silanol group. A solution of the sample was added to 2N Grignard reagent using butyl ether as solvent and methane as the A variety of samples, inert gas. including commercial silicone resins, was successfully analyzed a t room temperature. It was possible to complete 12 analyses in an 8-hour day. A least squares analysis of the data indicated that the absolute error, independent of the gas volume, is & 0.2 ml., 95% confidence limits.
A
for the determination of silanol was needed to aid in the identification of hydrolysis products from alkoxysilnnes ( I d ) . Karl Fischer reagent m-j,s considered because of the excellent results with simple silanols reported by Gilman and Miller (4) and Grubb (6). However, these hydrolysis products, like silicone resins, react too slowly (6). A gasometric method using a butyl ether solution of lithium aluminum hydride was also evaluated. Reaction was rapid and complete at room temperature with simple as well as complex silanols, provided the samples were added in solution. However, cleavage of siloxane bonds with resulting formation of silanes limits this reagent to samples without siloxane bonds or to samples which give silanes of lorn vapor pressure. As Grignard reagent was used to analyze silanols (9),a procedure, essentially that of Fuchs, Ishler, and Sandhoff (S), was evaluated. The apparaMETHOD
1 Present address, Major Appliance Laboratories, General Electric Co., Louisville, Icy.
1 1 18
ANALYTICAL CHEMISTRY
N. Y.
tus in Figure 1was used, but the sample, contained in a steel cup (1/4 X 1 inch), was introduced magnetically from the sidearm of the reaction flask into the Grignard reagent under a nitrogen atmosphere. Some samples reacted so slowly at room temperature that extended heating periods were necessary. Not only was considerable time involved (1 to 1.5 hours), but accuracy was low (&lo%), partially caused by the solubility of methane. These difficulties were overcome by adding a solution of the sample to 2N Grignard reagent a t room temperature, using butyl ether as the solvent and methane as the inert gas. Wright gives a complete background on the use of organometallic compounds for the determination of active hydrogen (13). MATERIALS, REFERENCE SAMPLES, A N D APPARATUS
DI-%-BUTYL ETHER (Eastman Kodak Co.) was purified by refluxing over sodium with stirring for 7 hours under a slow nitrogen stream. The ether was filtered and distilled over sodium, and the process repeated with the distillate. The purified ether was stored in brown bottles in a nitrogen atmosphere over sodium wire. GRIGNARDREAGEKT.A solution of 2N methylmagnesium iodide ( I O ) in dibutyl ether was prepared and the concentration determined. Khen stored under nitrogen in glass-stoppered bottles which were well greased, the reagent was stable for a t least several months. GASES. Line nitrogen (dew point -60’ C.) and CP methane (Matheson) were used. SILANOLS. Table I gives the preparation and constants of these com-
a pounds. It is assumed that all silanols were 100% pure. A diagram of the apparatus appears in Figure 1. The volumetric portion was essentially that of Fuchs, Ishler, and Sandhoff (S), except that a water jacket was added to the buret. The reaction flask was designed to permit addition of an aliquot of a sample in solution. A 4-ml. pipet, B, was adapted for sample addition by sealing a stopcock to the upper end and attaching a standard taper joint by means of a ring seal 3 inches from the delivery tip. The pipet was found to deliver 81.937, of the 5-ml. volumetric flask used to prepare the solution of sample. A 5-ml. volumetric pipet, A , was used for adding butyl ether to the sample and Grignard reagent to the reaction flask. Prior to reading the buret, the bulb of mercury was placed in a position approximately level with the mercury in the buret. Stopcock 111, ‘to the dibutyl phthalate manometer, was then opened and final leveling mas accomplished with the screw adjustment on the leveling bulb support. This combination made it possible to take a reading in 5 to 10 seconds x i t h a reproducibility to 0.02 ml. The temperature of the water bath was adjusted to that of the buret and held constant by the addition of small pieces of ice. This compensated for the heat given off by the rheostat in the magnetic stirrer apparatus. All joints and stopcocks were lubricated with Celvacene medium vacuum grease (Consolidated Vacuum Corp.). Steel springs Rere used to hold joints together. PROCEDURE
Standard Procedure. Pipets A and B and t h e reaction flask were rinsed with reagent grade acetone and dried in a stream of nitrogen. The reaction flask, with its side arm
stoppered and containing a steel stirring bar, was attached t o the system and methane flow was started (stopcocks I and I1 open, I11 closed). The calibrated pipet, B , was connected to stopcock I1 (stopcock on pipet B open). A sample calculated to give 20 to 25 ml. of gas was weighed into the volumetric flask and made up to volume with butyl ether, using the dried pipet A. Pipet A was then returned to the nitrogen line. The system was flushed three times and a slow stream of methane was maintained. Reagent (5 ml.) was added by pipet A through the side arm of the reaction flask (stopper temporarily removed), and stirring mas initiated. Pipet B was filled with sample solution (stopcock closed) and attached to the side arm. The methane flow was turned off (stopcock I closed). The mercury column was adjusted close to zero reading and the Tygon tubing from pipet B was connected to the outlet a t stopcock 11. The water bath was placed in position and within 3 to 6 minutes constant buret readings were obtained (stopcock I11 open). This represents the gas volume a t zero time. With stopcock I11 closed, the sample m-as admitted by opening the stopcock on pipet B. The bulb of mercury was lowered as gas was formed. With stopcock I11 open, final adjustment was made with the leveling device and the readings were recorded. Most of the gas (98+%) was given off in less than 2 minutes a t room temperature, and readings were constant in less than 10 minutes. A
Table I.
Silanol Trimethylsilanol, ng 1.3891 Triethylsilanol, ng
Preparation and Constants of Silanols
M. P.,
O
C.
99.25
Ref. (11)
63 (12 mm.).
(11)
1.4331
Triphenylsilanol
149-149.5
Diethylsilanediol
86.6-87.2
(9)
Tetramethyldisiloxane1.3-diol Te&aphenyldisiloxane1,3-diol 0 Boiling point.
63.8-64.0
(8)
108-109
(1)
Sotes Midcuts, refractionated twice through a 10-inch platinum spiral column Dow Corning Corp. Dissolved in acetone, filtered, solvents removed, and recrystallized three times from benzene Recrystallized twice from n-pentane and ether Recrystallized twice from cyclohexane Recrystallized twice from n-heptane and benzene
II.
Analysis of Silanols Gas Evolved, M1 OH, Weight 70 Theory ObservedTheory Observed
Table
Si1ano1
(GH&Si( 0H)z HO( CH3)nSiOSi(CH3)zOH HO( CeHb)$iOSi( C$ &)zOH
17.92 20,33 18 7615.96 19.22 17.43
blank on the solvent was always deter' mined and applied. The weight % hydroxyl was calculated as Vi X 17.01 X 100 X T 0 = Pi ~ XP ~ X Ti X 22,410 X TV where Pi = atmospheric pressure, mm.,
7
corrected to standard condi-
0
17.93 20.36 18 68 --
15.80 19.26 17.44
18.86 12.86 6.15 28.30 20.46 8.21
18.87 12.88 6.12 28.02 20.50 8.21
tions, less the vapor pressure of butyl ether (S). VI = volume of gas collected, ml., corrected for blank on solvent P and T = standard pressure and temperature T I = absolute temperature of water-jacketed buret W = weight of sample added (in this investigation 81.93% of sample weighed) This equation assumes that methane obeys the perfect gas laws a t room temperature and atmospheric pressure. The International Critical Tables (7) indicate that 0.3% error may be introduced by this assumption.
PIPET
RESULTS
Figure 1. 1. 2. 3.
4. 5.
6. 7. 8.
9. 10. 11. 12. 13. I, II,
Diagram of apparatus
Dibutyl ether bubbler Stirrer for bath 5 0 0 - w a t t heater, controlled with Varioc Dibutyl phthalate manometer Reaction flask, 25-ml. Erlenmeyer, 1 9 / 3 8 joint on neck, 1 2 / 3 0 joint on side arm W a t e r bath Magnetic stirrer (Magne-Stir, Laboratory Industries, Inc.) Steel rod or cup, 1 inch long 25-ml. buret W a t e r jacket Screw for adjusting leveling bulb support leveling bulb, mercury fllled Thermometers and 111. Stopcocks
Table I1 shows the results obtained with Grignard reagent on six different silanols. A least squares analysis of the gas volume data showed a precision to rt 0.18 ml. (absolute) for a single determination and that blank corrections mere adequate. The slope of the observed us. calculated volume is 1.04 =t 0.05 which includes the theoretical value of 1. All gas volumes are referred to standard conditions, and 95% confidence limits are used throughout. Occasionally it is necessary to analyze a sample having a very small percentage of hydroxyl. Therefore, this method was tested over a wide range of gas volumes. Varying amounts of tetraphenyldisiloxane-1,3-diol were analyzed in random order on the same day. The gas volumes obtained are given in Table 111, before correction for the blank. A least squares analysis of these data showed a precision t o 1 0 . 1 VOL. 30,
NO. 6,
JUNE 1 9 5 8
1 1 19
COMPLEX SllANOLS
Table 111.
Effect of Gas Volume
OH. CT, of Theory; after Gas Evolved, MI. Blank Theory Observed Correction 1.76 0.41 102.9 3.68 2.29 100.3 5.56 4.13 99.5 7.47 6.02 99.3 9.43 7.99 99.6 99.4 10.76 9.30 12.71 11.23 99.4 14.18 12.78 100.0 99.9 16.03 14.62 99.9 17.77 16.36
ml. (absolute) for a single determination, independent of the gas volume. The least squares slope of the observed us. calculated volume is 0.998 =t 0.006, which includes the theoretical value of 1. The intercept (blank value) determined statistically is -1.40 =!z 0.01 ml. The blanks determined a t the start and end of the experiment mere -1.43 and -1.34 ml.; their average, -1.39 ml., agrees with the calculated value. This negative blank is added to the volume of gas observed.
The method was successfully applied to complex hydrolysis products from alkyl and aryl trialkoxysilanes, as well as to commercial silicone resins. Mass spectrometric analyses of the gases from the complex as well as the simple silanols showed that only methane was produced. This is convincing evidence that no gas-producing side reactions occurred. Reaction was fast and apparently complete, as no more gas was observed after heating a t 95” C. The standard procedure was follon-ed, except for silicone resins, where the major portion of solvent was stripped under vacuum (1 mm.) at room temperature, before dilution with butyl ether. This avoided errors from the vapor pressure of solvents. ACKNOWLEDGMENT
The author is indebted to AI. hI. Sprung for many helpful suggestions, t o P. D. Zemany and F. J. Norton for mass spectrometer analyses, and to L. S. Nelson for assistance with the statistical design and analyses.
LITERATURE CITED
(1) Burkhard, C. A,, J . Am. Chenz. SOC.
67.2173f 1945). DiGibrgio,‘ P. A,,Sommers, L. H., Whitmore, F. C., Ibid., 68, 344 (1946). Fuchs, W., Ishler, T.H., Sandhoff, A. G., ISD. EKG.CHEU., AXAL. ED.12,507 (1940). Gilman, H., Miller, L. S., J . A m . Ghem. SOC.73,2367 (1951). Grubb, W.T., I b i d . , 76, 3408 (1954). communicaGrubb, W. T.,. Drivate tion. (7) International Critical Tables, Vol. 111, p. 3, PrfcGraw-Hill, Sew York, 1928.
(8) LU&;G. R., llartin, R. IT., J . ~ m . Chem. Soc. 74,5225 (1952). (9) Sauer. R. O., Ibid., 6 6 . 1T07 (1944). (10) Siggia, S., “Quantitative Organic
analyses Via Functional Groups,” p. 43, Wiley, Sew York, 1948. (11) Sommer, L. H., Pietrusza, E. K., Whitmore, F. C., J . Am. Chem. Soc. 68, 2282 (1946). (12) Sprung, %I. M., Guenther, F. O., Ibid.. 77. 3990. 3996. 6045 (1955). (13) Wright; G. F:, “Organometallk Compounds for the Determination of Active Hydrogen In Organic Analysis,” Vol. I, p. 155, Interscience, Xex York, 1953. RECEIVEDfor review August 29, 1956. Accepted February 15, 1958.
Determination of Fluoride Ion by Turbidimetric Titration WARREN W. BRANDT and ALLEN A. DUSWALT, Jr. Department o f Chemistry, P urdue University, Lafayette, Ind. ,Milligram quantities of fluoride may be determined by a rapid, convenient method using a turbidimetric titration. The simple apparatus is easily and quickly assembled from materials that are generally available. It permits the use of cells varying in size from glass tubing 1 cm. in diameter to large beakers, depending upon the volume of solution and type of precipitate obtained. With calcium ion as the titrant, concentrations of 0.01 to 0.09M fluoride ion may b e determined with an average relative error of &2%. Thorium ion may b e used as the titrant to determine fluoride ion from 0.04 to l.OM, with an average relative error of f. 2%. As little as 0.4 mg. of fluoride ion in 2 ml. of solution may b e determined with a relative error of f 4%. The time for a single determination is generally less than 10 minutes.
A
AMOUNTS of fluoride are often determined gravimetrically. These methods involve lengthy procedures t o ensure pure, filterable precipitates, but low results are obtained. PPRECIABLE
1120
ANALYTICAL CHEMISTRY
Recently, several articles have described the determination of milligram quantities of fluoride ion. Curry and PIIellon (3) have distilled fluoride as silicon tetrafluoride and reacted the hydrolyzed silicon to form the heteropoly blue compound. This color is then measured spectrophotometrically and related to original fluoride content. Onstott and Ellis (6) have determined fluoride b y titrating with samarium ion containing europium carrier-tracer. The end point is determined by measuring the excess titrant by a radiometric procedure. Grant and Haendler ( 5 ) have determined macro quantities of fluoride by titrating with thorium nitrate, using a high frequency oscillometer to detect the end point, and Chilton and Horton ( 2 ) have titrated fluoride acidimetrically with aluminum ion. An excellent general review is available on the various analytical methods for the determination of fluoride through 1952 (6). The method proposed here is a precise, rapid, and simple determination of fluoride ion within the concentration range of 0.01 to 1.OM. Many problems and difficulties inherent in the gravi-
metric determination of fluoride are not met in this analysis. 30 filtering, drying, or weighing is necessary. Low values are not obtained because of loss of fluoride as occurs in the gravimetric methods. The method of turbidimetric titration is based upon the light absorbing and scattering properties of precipitates in solution. The amount of precipitate formed by the titrant with the sample is measured by the decrease in light transmitted through the cell. A photocell in series with a potentiometer is used t o measure the change in transmitted light. The end point is obtained from a plot of log Zo/lus. milliliters of titrant. lois the intensity value for the clear solution and I , the value for the intensity after the titrant is added. The intersection of the lines forming the slope and plateau of the plot is the end point. Under the most favorable concentration conditions of fluoride ion, a relative error of =tl%was obtained for the rapid determination of fluoride by turbidimetric titration. For the poorest conditions the error was =t4%. The method of turbidimetric titration is not new, although it is underdeveloped