Determination of Titanium and Mixtures of Iron and Titanium with

Chem. , 1955, 27 (10), pp 1596–1599. DOI: 10.1021/ac60106a028. Publication Date: October 1955. ACS Legacy Archive. Cite this:Anal. Chem. 27, 10, 159...
2 downloads 0 Views 604KB Size
ANALYTICAL CHEMISTRY

1596 manganese ion present for the oxidation of the indicator by ferricyanide, thereby making the quantitative titration possible. Other methods of detecting the end point of the zinc determination were attempted in place of the diphenylamine and ferricyanide. Variamine blue ( I , 3 ) was tried, but proved uneatisfactory because it took 10 to 15 seconds after the addition of each drop of the ferrocyanide solution for the color t o become stable. iln attempt was made t o measure the turbidity of the solution containing zinc ferrocyanide precipitate and (ethylenedinitri1o)tetraacetic acid. For the complete precipitation of small amounts of zinc in the presence of much excess of (ethylenedinitri1o)tetraacetic acid, a longer time of standing or a large amount of ferrocyanide wab required. Manganese, if present, formed a precipitate almost immediately with the ferrocyanide in the presence of (ethylenedinitri1o)tetraaceticacid. Other interferences 15 ere encountered in the end point determination for the manganese titration: Though cobalt(I1) is not precipitated by ferrocyanide at p H 2.5, the red colored cobalt(111)-(ethylenedinitri1o)tetraacetate complex is formed in the acetic acid medium upon the addition of ferricyanide (6). Therefore manganese cannot b e titrated with ferrocyanide wing diphenylamine and ferricyanide as the indicator when mhalt is

present. Interfering coloration caused difficulty when manganese was titrated in the presence of molybdate or uranyl ion. The manganese and zinc can he titrated with ferrocyanide in the presence of (ethylenedinitri1o)tetraacetic acid a t p H 2.5 by a high frequency method ( 4 ) . The high frequency method gives two breaks in the titration of a mixture of manganese and zinc. The ferrocyanide precipitates the zinc first and the manganese next. Further investigation of a better means of detecting t h r end point for both zinc and manganese determinationq is needed. ACKNOWLEDGMENT

The author is indebted to H. Flaschka for providing variamine blue. LITERATURE CITED

(1) Erdey, L., 2. anal. Chem., 137, 410 (1953). (2) Flaschka, H., Mikrochim. Acta, 1953, 414.

(3) Ibid., 1954, 361. (4) Januskiewicm, S.B., master's thesis, University of Connecticut,

1955. (5) Pfibil, R., CoZlectio?z Czechoslou. Chem. Communs., 14, 320 (1949).

RECEIVED for review .4gril 8 ,

19.55.

Accepted July 18, 1955.

Determination of Titanium and Mixtures of Iron and Titanium with Electrolytically Generated Ceric Ion ROBERT V. DILTS' and

N. HOWELL FURMAN

Frick Chemical Laboratory, Princeton University, Princeton,

Titanium sulfate solutions were reduced in a Jones reductor, caught in saturated cerous sulfate solutions, and titrated coulometrically with electrolytically generated ceric ion under an atmosphere of nitrogen. Using the sensitive amperometric end-point procedure it was possible to determine amounts of this ion ranging from 50 y to 5 mg. with an accuracy of within f 0 . 6 % . Larger samples than this cannot be determined readily, because of the insolubility of titanium sulfate in the generating medium. Mixtures of iron and titanium containing amounts of titanium from 0.013 to 0.16 meq. and amounts of iron from 0.06 to 0.12 meq. can be determined with an accuracy of within zk0.66qo or better. The titanium can be determined in this mixture with an accuracy of within 3~0.6% or better, and or better. The mixture was the iron within &O.SOq' passed through a Jones reductor, caught in a solution containing at least 90% of the amount of ceric ion calculated to be required'for the titration of titanium in the sample, and then titrated in an atmosphere of nitrogen. This procedure necessitates a prior rough knowledge of the amount of titanium in the mixture, hut if the ceric sulfate is not present initially, the results are in error.

I

N I'OLUNETRIC analysis, trivalent titanium is titrated n i t h standard solutions of ferric iron, methylene blue, potassium permanganate, or ceric sulfate. The most widely recommended procedure is the one given by Scott ( 5 ) which involves titration with standard permanganate. From a consideration of the standard potentials of the titanous-titanic couple and of the cerous-ceric pair, it would appear that the titration of titanium I Present address, Departnrent of Chemistry, Williams College, Williamstown, hfass.

N. J.

with ceric ion should be excellent. The reaction would be expected to be rapid, quantitative, and the detection of the equivalence point accurate. The only information, however, concerning this titration that has been published is by Takeno ( 9 ) , who ieduced quadrivalent titanium with a zinc amalgam (Jones reductor) and then, in an inert atmosphere, titrated the trivalent titanium solution with standard ceric sulfate. The reduction of the titanium R as performed with near boiling solutions, although there appears t o be no adequate reason for this, because the reduction is just as complete when carried out in the cold. The determination of titanium has recently become more important, and since ceric ion can he generated coulometrically very readily a t a platinum anode (6) it was decided to study the coulometric determination of this metal. The application of coulonietrir titrations has been limited almost exclusively to the determination of one species of ion in solutions of a single substance. Because it was found possible to determine titanium coulometrically with ceric ion and it was known that iron could be easily determined by the same method ( 6 ) ,the coulometric determination of both of these metals in the preqence of one another was rtudied since they are found together frequently. The standard potentials of the titanous-titanic and ferrousferric couples are sufficiently separated (about 1 volt differenre) t o give two breaks in a potentiometric titration curve with a standard oxidant. Shippy ( 7 ) titrated mixtures of these two substances with standard permanganate and obtained good results. Under his experimental conditions the formal potentials of the two systems are about 650 mv. apart, which is more than adequate for complete and accurate determination of both metals. Shippy passed a hot fiolution of the sample through a Jones reductor, catching it in 1 to 1 sulfuric acid. The waim, reduced solution was titrated with permanganate until the titanium end point was reached using methylene blue as an indicator, The solution was then cooled to room temperature and the iron

V O L U M E 27, NO. 10, O C T O B E R 1 9 5 5

1597 per ml., whereas the value ohtained from the standardization procedure was 4.005 mg. per ml. Other solutions were prepared and their concentrations were calculated from the weight of the pure titanium dioxide. -4 standard ferric sulfate solution waq prepared by dissolving a weighed amount of Baker and ddamson, reagent grade ferric sulfate nonahydrate in 1 N sulfuric acid and diluting i t up to volume. This solution was standardized by passing it through a Jones reductor and then titrating it, under a n atmosphere of nitrogen, with standard ceric sulfate, using ophenanthroline ferrous sulfate as an indicator. This solution was found to he 0.060275 with respect to iron.

Table 1. Coulometric Titrations of Titanium with Electrolytically Generated Ceric Ion Titanium Added, N g .

Current, Ma.

Time, Min.

5.085 5.052 5.027 5 012 4.989

13.34 13.34 13.82 16.65 13.84

12.788 12.747 12.290 10,144 12.136

Titanium Found, M z 5,081

-0

004 012 031 018 f0.013

-0

os

5.064 5,058 5.030 5,002

+o +0 +o

L O 24 T O 61 +o 35 26

+o

1.011 1.002 1.008 1.000 0.9990

3.737 3.729 3 . 620 3,774 3.779

9.111 4.003 9.324 8.901 8,883

1.014 0.9993 1.OO.i 1.000 0 9998

+0.003 -0,002 -0,003

+ O . 30

+o.ooo +o.

+o.oo +O .OY

0.5593 0.5035 0.507: 0,5040 0 5114

1.490 1,506 1.493 1.459 1.44;

12.618 11.208 11.429 11.662 11.787

0.55lQ 0.5054 0,5082 0,5067 0.50i3

+0.0006 +o. 0019 +0.0014 4-0.0027 -0.0041

+o.

Y

pa.

49.86 49.76 50.14 44.79

202.9 202.8 202.2 208.U

49,67 49.92 50.20 44.53

11 +0.38

+0.27 i-0.54 -0.80

Y

Y

8 210 8.263 8.338 7.901

0008

-0.20 -0.30

-0.19

+ O . 16

t o ,06 -0,2G

titrated v ith 1,lO-phenanthroline ferrous sulfate as indicator As hi3 recornmended procedure is perfectly straightforward and the potential breaks in the actual titi ation were excellent, thiq procediire was adapted to the coulometric determination of thew two metal< n i t h ceric ion, using the rensitive end-point procedure for the detection of the two eqiiivalence points. APPARATUS

The apparatus for the coulometric titrations was that described in a previous publication ( 4 ) . The titration cells were weighing bottles of 30- or 70-ml. capacities, depending upon the size of the sample taken. Each cell was covered \yith a rubber stopper provided with openings for the four electrodes, a gas inlet, and the Jones reductor. Jones reductors of three sizes were employed for the reduction of the titanium and iron samples. The one selected for a given determination depended upon the size of the sample. These reductor columns were made of soft glass tubing, flared a t one end, and fitted with a standard taper stopcock a t the other; the tip of the stopcock outlet was dran-n out to a fine orifice. The dimensions of these reductors weie: large reductor, 30 cni in length with an internal diameter of 0.7 cm.; medium reductor, 33 cm. long with a 0.5-em. internal diameter; small reductor, 21 em. long ith a 0.4-em. internal diameter A S o . i664 Leeds and Northrup pII meter was used to detect the end points with the Potentiometric titrations and t o preset the potentials of the indicator circuit when using the sensitive end-point procedure. All potentials in this work were measured against the lead amalgam-lead sulfate reference electrode, as recommended by Cooke, Reilley, and Furnian (3). SOLUTIOhS

The cerous sulfate solution was prepared by saturating 1.0-l’ sulfuric acid m-ith reagent grade cerous sulfate trihydrate (G. Frederick Smith Chemical Co.). The standard titanium sulfate solution was prepared as follows. Titanium dioxide of reagent grade from the Amend Drug and Chemical Co. was placed in a porcelain crucible and ignited for half an hour at red heat with a Meker burner. Weighed amounts of the oxide, with 8 grams of Baker and Adamson ammonium sulfate of reagent grade were dissolved in 20 ml. of concentrated sulfuric acid. It was necessary to heat this miuture t o boiling in order to effect dissolution. The resulting solution was a pale yellow. This solution was evaporated t o about 10 ml. to reduce the concentration of sulfuric acid in the final solution. The concentrated solution was cooled and then made up to 250 nil. in a volumetric flask, using distilled water only. I n one instance a titanium solution was standardized in order to determine whether the ignited titanium dioxide could he employed directly as a standard. This standardization was carried out bv reducing the titanium sulfate solution in a Jones reductor, catching the reduced species in a 5% solution of ferric sulfate in 3.0S sulfuric acid, and then titrating the ferrous ions produced with standard ceric sulfate, which was standardized against arsenious oxide. o-Phenanthroline ferrous sulfate was used as an indicator. The concentration of titanium as calculated fioni the weight of titanium dioxide taken was 4.008 mg

DETERMINATION OF TITANIU>I

-0.38 +0.32 to.11 -0.52

Procedure. The size of the actual reductor column that was used depended upon the amount of titanium being determined. It was washed first with boiling 5y0 sulfuric acid and then Lvith uater until it no longer decolorized dilute permanganate. Xext a stream of pure nitrogen was forced through it for 5 to 10 minutes in order to remove all traces of oxygen. While the reductor was being deaerated the potential of a previously deaerated solution of cerous sulfate was adjusted to 1.250 volts vs the lead amalgam-lead sulfate reference electrode, by generating ceric ions until the galvanometer in the indicator circuit shoued that no current was flowing in the indicator system-i.e., the impressed and the solution potentials were identical. @ne milliliter of deaerated 1.Oh-sulfuric acid was then added to the reductor column, followed by the sample of titanium. The titanium was kept in the reductor column for 10 minutes bi ~ 1 0 4 ing the stopcock at the bottom of the reductor column. ‘4t the end of this period the sample was drained into the titration cell and the column was washed with 3.0 ml. of 1.ON sulfuric acid and 3.0 ml. of water. Generation of ceric ion was begun and continued until the indicator galvanometer again showed that the two potentials were equal. In the immediate vicinity of the end point it was necessary to add the current in small increments and to allow a short period of time for equilibrium to become estahlished; usually 2 or 3 minutes were adequate. For samples that contained 50 y of titanium or less, the worl,ing indicator electrode was a 1-cm. ( 9 ) platinum-iridium foil inqtead of the platinum wire that had been used for the larger sample.. The results of a series of titrations on various amounts of titanium performed according to this procedure are presentrd in Table I. Discussion. The data in Table I show that samples of titanium from 50 t o 5.0 mg., or from about 1 0 miwoequivalent to 0.1 meq. can be titrated successfully with coulometrirnll> generated ceric ion with a n accuracy of nithin =kO.SO%. The potential that was impressed across the indicator electrodefor the sensitive end-point procedure was determined by meani of a potentiometric titration. At first it way found that two lireJk= appeared in the potential curve for this titration and that the solution was pale yello\L, after the reduced titanium solution hxd been added to the titration cell. I t was believed that this was caused by traces of titanium peroxide that might have been formed in the reductor column. The subsequent deaeration of all solutions and of the reductor column itself eliminated the .;e( ond break in the curve and the coloration of the solution. Initial determinations of titaniiini using cold 1 S sulfuric arid as the wash solution gave results that u ere in error by =k3to 10%. Neither more concentrated acid a$ a wash solution nor the use of a boiling mash solution improved the results. It was decided finall\ that this error wai caused bv incomplete reeduction of the titanium. Allowing the titanium solution to remain in contact with the zinc amalgam for 5 minutes gavu improved revltq, which were only from 0.2 to 1.4% too low. This time of contact was increased t o 10 minutes, and accurate results were obtained. Shippy ( 7 ) in his kyork with titanium, maintains that it is neceasary t o permit a near-hoiling solution of titanium t o remnin in the Jones reductor for 15 minutes in order to achieve

-,

1598

ANALYTICAL CHEMISTRY DI S C u s s I o n ‘

Table 11. Successive Coulometric Titration of Titanous and Ferrous Ions with Ceric Ions Titanium Iron Total Iron and Titanium Added, meq. 0.01369 0.01432 0.02629 0.02785 0.05206 0.05258 0.1058 0.1088 0.1048

Found, meq. 0.01330 0.01423 0.02641 0.02191 0.06209 0.06254 0.1052 0.1087 0.1056

Error,

%

-2.85 -0.63 +0.45 f0.22 +0.06 -0.08 -0.56 -0.09 +0.76

Added, meq. 0.1211 0.1211 0.07913 0.07962 0.06243 0.05243 0.02043 0.01205 0.06087

Found, meq. 0.1213 0,1213 0.07956 0.08025 0.05228 0.05248 0,02049 0,01198 0,06122

Error,

%

+0.17 f0.17 +0.52 f0.80 -0.29 +0.09

+0.29 -0.58 f0.57

complete reduction; however, these extreme conditions were found t o be unnecessary in this investigation. .4few attempts viere also made t o determine samples of titanium that 1% ere smaller than 50 e/. Using the smallest of the reductor columns, two determinations on 25-7 samples gave results t h a t were 3.6 and 7.9% too high. Samples of this size normally require more than ordinary precautions in order t o obtain accurate results, but since the purpose of this study was merely to ehow that the coulometric determination of titanium was feasible over a wide range of concentrations, rather than to develop a microprocedure, no further Lyork was done on samples smaller than 50 y. However, there appears to be no adequate reason why they could not be analyzed. Experiments were made with titanium samples of 10 mg. or larger. I n these cases the results ’ir ere always too high by about 10%. I t is possible that this error or apparent decrease in the current efficiency, might have been introduced through the dilution of the generation solution by the addition of large volumes of the samples and wash solutions. I n order to minimize these effects it becomes necessary t o work with reasonably concentrated titanium solutions and as small amounts of washing as possible. DETERMINATION OF MIXTURES OF IRON AND TITANIUM

Procedure. Twenty-five milliliters of the cerous sulfate solution were placed in the 70-ml. titration cell and deaerated with nitrogen for 20 minutes. During this time the Jones reductor of the medium size was washed with hot, 1.ON sulfuric acid and water until it no longer decolorized very dilute permanganate. The reductor was then placed in the opening provided for it in the stopper of the titration cell, with the tip extending just below the level of the liquid in the cell. Yext the reductor column was deaerated for 5 minutes by forcing a stream of tank nitrogen through it. During this time the potential of the generating solution was adjusted to 1.300 volts, us. the lead amalgam half-cell, by generating ceric ion. One milliliter of deaerated IN sulfuric acid was added to the reductor followed by the titanium-iron sample, which was allowed to remain in contact with the zinc amalgam for 5 to 10 minutes. During the period of contact, ceric suIfate was generated in the titration vpssel. The amount of ceric ion generated was approximately 90% of the calculated amount for the titration of the sample of titanium. The stopcock on the reductor was opened, and the iron-titanium sample permitted to flow into the titration vessel. This was washed through by approximately 5.0 ml. of 1.ON sulfuric acid. The impressed potential was lowered to 0.725 volts for the titanium end point, and generation of ceric ion continued until the galvanometer indicated that no current was flowing. Near the end point itself, the current must be added, in small increments and a few minutes allowed after each addition for completion of the reaction, as indicated by a cessation of downward drift in the galvanometer readings. The impressed indicator potential was increased then to the equivalence potential selected for the ferrous-ceric titration, 1.300 volts, and generation continued until once more no current flowed in the indicator svstem, again adding increments of current and waiting in the vicinity of the end point. T h e amounts of titanium and iron present, respectivelv, %-erecalculated from the times required to reach the first end point and from the first end point to the second one, and the value of the current. The results of a series of determinations of varying ratios of these two elements are presented in Table 11.

Initial t i t r a t i o n w i t h o 11t prior generation of ceric sulfate gave results that were Added, Found, Error, meq. meq. % correct within 1% for the iron, 0,1348 0.1346 -0.15 but were from 15 to 23% too 0.1354 0.1353 -0.08 low for the titanium. I n these 0.1054 0.1060 +0.57 0.1075 0.1082 +0.64 t i t r a t i o n s 0.638 v o l t w a s 0.1045 0.1044 -0.11 selected as the end-point poten0.1050 0.1050 tO.OO tial t o be applied for the tita0,1257 -0.40 0,1262 0.1208 0.1207 -0.08 nium titration and 1.232 volts 0.1657 0.1668 +0.66 as that for the end point of the iron titration. These values had been selected from a potentiometric titration of a mixture of the two ions. No explanation could be found for this error since the conditions employed were identical with those that Shippy ( 7 ) had successfully used for the mixture, and also ~ i t those h that were successful for the determination of titanium alone. I n the successive pairs of experiments, the pregeneration times for forming the ceric solution to receive the reduced solution were 1.50, 3.00, 6.00 minutes, and in the last three experiments 13.00 minutes. I n all cases the pregeneration current was betn een 11.64 and 11.81ma. T h e length of time that the mixture remained in the reductor column x a s increased t o 10 minutes and decreased t o 0 minutes, the amount of wash solution varied, and the end-point potential for the titration of titanium altered. S o n e of these changes in the operating conditions gave accurate results. Slthough some of them improved the accuracy of the titanium determination, they increased the error in the determination of iron to 5% or greater. A second potentiometric determination a-as carried out, allowing several minutes to elapse after each generation period befoie taking the potential reading. T h e resulting potential-time curve showed three breaks, with the new break betneen the two previously observed. The mid-point potential of this new break was about 0.800 volt. A4study was then undertaken t o ascertain which solution contained this unknown third substance that was being titrated. It was found, by varying the amount of titanium solution taken, that the unknown substance was in this solution, and that the amount of it was very roughly proportional to the amount of titanium that was taken. It could be possible that this is again a titanium peroxide; from a consideration of t h e standard potential of the peroxide system in acid medium it is positioned correctly betn een the ferrous-ferric and the titanoustitanic potential. JVhere this peroxide might have originated could not be established (if it was titanium peroxide a t all), as i t was thought that all of the solutions and the reductor had been thoroughly deaerated. Also t o ensure complete reduction of t h e sample, the titanium had been permitted to remain in contact with the zinc amalgam for 5 minutes or more. Bricker and Sweetser ( 1 ) and Sill and Peterson ( 8 ) , when titrating mixtures of uranium and iron with ceric sulfate, observed that the results of the uranium determination R ere consistently low. T h e explanation given by both sets of authors for this phenomena is that the presence of iron, or any other multivalent ion, causes an “induced oxidation” of the uranium. I t is poesible that an analogous induced oxidation could be occurring in this work with titanium and iron. Bricker and Sweetser ( 1 ) overcame this error in their determination of uranium by initially adding sufficient ceric sulfate so that their uranium and iron mixtures xere always caught in this and not in pure sulfuric acid. They found that if about 90 t o 97% of the amount of ceric sulfate required for the titration of uranium was added to their titration vessel before the mixture being determined was introduced, excellent results were obtained for both substances. Therefore, this practice was adopted in the

1599

V O L U M E 27, NO. 10, O C T O B E R 1 9 5 5 work with titanium and iron. Before the samples were drained into the electrolysis cell from the Jones reductor, 90% of the theoretical quantity of ceric ion required for the titration of titanous ion was generated coulometrically, so that the reduced solution was caught in this and not in the pure cerous sulfate. This procedure necessitates knowing, within a reasonable range, the concentration of titanium in the sample. This means that either a preliminary titration must be performed, without the initial generation of ceric ion, t o obtain a very rough idea as to the amount of substance preqent, or the information must be obtained by some other means. The data in Table I1 show that in mixtures of iron and titanium, amounts of titanium ranging from 0.013 to 0.16 meq. can be determined with an arcuracy of within i.0.6% or better. I n these solutions 0.06 to 0.12 meq. of iron ran be determined with an accuracy of within &0.80% or better. The total amount of the sample present can b e establiqhed within a range of 0.66% or better. Traces of dissolved oxygen may account for the limits of pre(*isionand accuracy that are observed. Sulfuric acid is suspected of possible contamination bv traces of redox systems. Redistillation of the acid does not eliminate fluctuations in the reagent hlnnks that are observed in various procedures.

microcoulometric titrations are undertaken. Since the limiting factor with amounts of titanium larger than 0.1 meq. appears t o be the solubility of titanium sulfate, there is no reason why samples of this magnitude should have to be encountered. Through appropriate dilution of the sample solution, the amount of titanium to be determined can be brought within the range of concentrations found t o be satisfactory in this Tvork. Mixtures of titanium and iron can be determined coulometrically x i t h electrolytically generated ceric ion with reasonably good accuracy and precision. However, unless there is knowledge of the approximate amount of titanium in the sample, it is necessary to run a preliminary coulometric titration, and in a subsequent titration to generate about 90% of the required ceric ion before addition of the sample to he analyzed. LITERATURE CITED

Rricker, C. E., and Sweetser, P. B., A N ~ LCHEM., . 25, 764 (1953). Cooke, W. D., Reilley, C. S . ,and Furman, S . H., Ibid., 23, 1662 (1951). Ibid., 24, 205 (1952). Dilts, R. T., and Furman, Y. H., Ibid., 27, 1275 (1955). Furman, S . H., editor, "Scott's Standard Methods of Chemical Analysis," 5th ed., p. 984, Van Kostrand, Sew York, 1939. . Furman, iY.H., Cooke, K.D., and Reilley, C. N., A N ~ LCHEM., 23, 945 (1951). Shippy, B. A,, Ibid., 21, 698 (1949). Sill, C. W.,and Peterson, H. E., IbLd., 24, 1175 (1952). Takeno, R. J., Chem. SOC..Japan, 5 5 , 196 (1934).

CONCLUSIOhS

Amounts of titanium alone can be determined coulometrically with ceric sulfate over fairly wide ranges of concentration with a n accuracy of within &0.6%. There appears t o be no reason why samples smaller than one microequivalent of titanium could not b e determined, provided that the precautions necessary for

R E C E I r E n for review March 24, 1955. Accepted July 20, 1955. BaPed upon a dissertation submitted by Robert V. Dilts in partial fulfillment of requirements for degree of doctor of philosophy a t Princeton Cniversity, 1954.

Direct Determination of Acetic Acid in Acetic Anhydride J. H. MCCLURE, T. M. RODER, and R. H. KINSEY' Polychemicals Department,

E. 1.

du Pont d e Nemours & Co., lnc.,Wilmington, D e l .

This investigation was undertaken to develop a rapid and accurate method for the determination of acetic acid in acetic anhydride. It has been extended to other acids in their anhydrides and other acids in acetic anhydride. Two methods are described w-hich depend on the reaction of a tertiary amine (triethylamine) with the free acid in the anhydride. The first method depends on the color change of methjl red at the end point, while the second measures the temperature rise when the reagent is added to the sample. A precision of & 0 . 0 7 7 ~absolute over the range 0.5 to 3.5'70 acid was obtained for the visual method and i 0 . 0 9 % absolute for the thermometric method over the range 0.8 to 5 . 5 7 ~acid. Potentiometric titration of acetic acid in anhydride using combinations of glass, calomel, gold, tungsten, silver, silversilver chloride, platinum, and antimony electrodes was unsuccessful.

T

HE literature lists only a few methods which can be used for the determination of acids in anhydrides. Sicolas and Burel (2) used aniline to produce the anilide from the anhydride, thus providing a method for calculating the acid and anhydride b y titrating the sample before and after anilide formation. Smith and Bryant ( 4 ) also performed two titrations, one with methanolic sodium methylate and one with aqueous caustic. Froin the-e data, the acid and anhydride can be calculated. As 1

Present address, Polychemicals Department, E. I. d u Pont de Nemours 6

Co.. Inc., Sabine River Works, Orange, Tex.

both of these methods are indirect, they are not very satisfactory for determining small amounts of acid in anhydride. Siggia and Floramo (3) used potentiometric titration of the acid with a tertiary amine in a solvent such as acetone. Their method is limited to acids with a pK, 3 or greater. Thus, acetic acid cannot be determined b y their procedure. The visual titrimetric and thermometric methods presented here permit the determination of acids with dissociation constants a t least as small as that of benzoic acid (6 X 10-6); they also reduce the analytical time and tend to eliminate the lack of precision arising from the difference in large numbers in the Sicolas-Burel or Smith-Bryant methods. I n the visual method triethylamine in benzene is used as titrating agent. Acetic anhydride, as the solvent, appears to enhance the acidity of the acetic arid t o the point nhere it will yield a color change with methyl red. The molarity of the triethylamine-benzene reagent is determined by titration against standard hydrochloric acid, and consequently, the reagent must be free of primary and secondary amines. The purification step is relatively. simple and involves treating the triethylamine with excess acetic anhydride to remove primary and secondary amines, neutralizing the excess anhydride with sodium hydroxide, and extracting the amine into benzene. The benzene-amine mixture is dried b y distillation, which removes the benzene-Lyater gzeotrope. The amine-benzene azeotrope is then collected, diluted with dry benzene, and standardized. The titration of a sample is made t o a n end point which matches the color of methyl red in benzene. The thermometric method IS entirely empirical and involves