metricall!-. Kitroglycerin determination in methyl isobutyl ketone results in lo^ value-lformaniide, and nitrocellulose and nitroglycerin were titrated in ethylenediamine. The results in Table 111. were obtained by dissolving the T S T in methj 1 isobutyl ketone and the inorganic nitrate in acetic acid. Aliquots were taken from the volumetric solutions and titrated (the T S T a t 10 ml. per minute and the inorganic nitrate a t 20 nil. per minute). The synthetic mixtures were made up on a percentage basis using C.P. grade ingredients n itli the exception of nitroglycerin ( 16 40% nitrogen) and nitrocellulose ( 12.52Tc nitrogen) hich were assayed hj- means of the nitrometer. The analysis percentage value is the mean of f i e~samples.
ACKNOWLEDGMENT Table 11.
Analysis of Single Explosive
Titration, 0' /O
RlNT Nitrocellulose Xitroglycerin
99.75 97 83 99 01
Purity,
Std. % Dev. 100 00 0.14 98.11 0.81 99.46 0.77
The author wishes to acknowledge the cooperation of IT.C. Crosby in carrying out this work, and the communication from Picatinny Arsenal, which resulted in this investigation. LITERATURE CITED
Table
111.
Reproducibility matic Titrator
TST Titration, RII. 14 60 14.60 14.65 14 01 14.65 14.60 14.62 14.65 14.65
of
Auto-
Total Inorganic ?-itrate Titration, 111. 23.88 23.90 23.87 23.82 23.82 23.85 23,86 23.82 23.80 ~~
~~
(1) Bruss, U. B., Kyld, G. E. A., ANAL. CHEW29, 232 (1957). (2) Caldin, E. F., Long, G., J . Chem. SOC. 1954,3737. (3) Fritz, J. S., ASAL. CHEY. 24, 306 (1952). ( 4 ) Kave, S.RI., Ibid., 27, 292 (1955). ( 5 ) Mdmstadt, H. V., Fett, E. R., Ibid., 26.1348 11954). (6) Ibid.,' 28, 1412 (1956). (7) Pifer, C. IT., Wollish, E. G., Ibid., 24, 519 (1952). ( 8 ) Pifer, C. W., Wollish, E. G., Schmall, AI., Ibid., 26, 215 (1954). RECEIVEDfor review Spril 15, 1957. Accepted December 11, 1957.
Nonaqueous Titration of Zinc Rapid Method for Zinc in Lubricating Oils THOMAS L. MARPLE, GEORGE MATSUYAMA, and LORENZO W. BURDETT Research Departrnenf, Union Oil Co.o f California, Brea, Calif.
b The reaction of dithizone with various heavy metals was studied in benzene-methanol solution. It was found that this reaction can be made the basis of photometric titrations of some of these metals, and a method for the determination of zinc in lubricating oil was developed. Less than O.lyo of zinc in oils can be determined with an accuracy of 1 to 270. Alkaline earth and alkali metals commonly present in oils do not interfere. Among the common heavy metals, only lead and mercury behave as zinc in the titration procedure. The presence of the other heavy metals can b e detected b y a change in shape of the titration curve.
T
HE COLOR reaction between diphenylthiocarbazone (dithizone) and various hea1-y metals has long been utilized for the colorimetric determination of micro quantities of these elements. The usual procedure relies upon the extraction of the nietal dithizonate complex from water into a nonaqueous phase, followed by the colorimetric measurement of the coniplex in the latter medium. Vallee ( 9 ) eliminated the extraction procedure by the use of a monophase water-
glycol mixture for the determination of zinc. Pflaum, Popov, and Goodspeed (6) used a 50% aqueous dimethylformamide medium to determine copper with 2,2'-biquinoline. A monophase colorimetric procedure is especially appealing for the analysis of petroleum products which are immiscible with water and are therefore usually ashed or extracted for the determination of metals. Of particular interest in the work reported here was the possibility of improving the accurac>- by a direct titration of zinc in petroleum products using an organic medium. Acid-base titrations in nonaqueous media have been used for a long time (8). Nonaqueous media have made possible the titration not only of materials immiscible with water but also of acids and bases too \Teak to titrate in aqueous solutions. T'ery little has been done, however, towards the direct titration of metals in nonaqueous media. Brummet and Holln-eg ( 2 ) utilized an acid-base titration in nonaqueous medium for an indirect potentiometric titration of copper, cobalt, and nickel in a medium consisting of 80% benzene and 20% methanol. Four chelating agents n ere used : dimethylglyoxime dithizone, 8-quinohno1, and l-nitroso-2-
naphthol. The titration with standard sodium hydroxide solution of the hydrochloric acid liberated in the reaction of the metal chloride with chelating agent was used to determine the metal content. Because of the acid-base titration involved in the procedure, buffers which are generally used to make organic reagents selective for particular metals could not be used. Titrations of solutions of the pure chlorides only were reported, and the suitability of the procedure for the analysis of the metals in the presence of other materials w s not established. Gerhardt and Hartmann (3) reported a back-titration procedure for the determination of metals in lubricating oils with disodium (ethylenedinitri1o)tetraacetate (EDT-4). The sample was dissolved in acetone, and treated with excess aqueous standard EDTA solution, which was back-titrated with an aqueous standard magnesium chloride solution. Because of the high stability of metal-EDT$ complexes, this method is not selective and most metals except the alkali metals are determined together. The work reported here n a s undertaken to develop a direct titration procedure for determining zinc in VOL. 30, NO. 5, M A Y 1958
937
lubricating oils and additive concentrates. The relative stabilities of various transition metal-organic reagent complexes have been studied by Irving and Rilliams (5). They found that zinc dithizonate is between cobalt and copper dithizonate in stability. Pilipenko ( 7 ) reported the stability constants of cobalt and copper dithizonates as 2 X 10'7 and 9 x 1026,respectively. These data indicate that zinc dithizonate is sufficiently stable for direct titration purposes. Because no suitable electrometric method was available for detecting the end point of the titration, a photometric method mas adopted. Both free dithizone and the zinc dithizonate are colored so that either species could be used for end-point detection. The simpler scheme of using the color of free dithizone was adopted. As the color of zinc dithizonate formed during the titration obscures to the eye the point a t which all the zinc is complexed and free dithizone is present, it was necessary to perform the titration in a spectrophotometer or suitable colorimeter. The present study has provided a simple, rapid, selective method for determining zinc in lubricating oils and other hydrocarbons. The accuracy of the method is to about 1%. EXPERIMENTAL
Apparatus. A Beckman Model B spectrophotometer which had been modified like t h a t described by Goddu and Hume was used for the photometric titrations. Because of the volatility of the titrating solution, a 1-ml. Greiner ultramicroburet was used to deliver the titrant solution. Reagents. The dithizone (Eastman Kodak White Label grade) was used without further purification. All other chemicals \\-ere of ACS reagent grade or the best commercial grade obtainable. Stock solutions which contained approximately 200 y of heavy metal per ml. were prepared by dissolving the required amount of the following chemicals in absolute methanol: zinc acetate, Zn(C2H302)2.2H20; cobaltous chloride, CoC12,6H20; ferric chloride, FeC13.6H20 ; chromium chloride, CrCl3.6H20; nickelous sulfate, NiS04.6H20; mercuric chloride, HgClz; and lead subacetate. Further dilutions were made with alcohol until a concentration of approximately 10 y per ml. was obtained. Aliquots of these solutions n-ere used in the titrimetric determination. Procedure. The photometer was adjusted to 640 rnb and 50 ml. of solvent (benzene saturated with ammonium acetate) added to the titration cell. A 2-ml. aliquot of standard solution was added and the cell compartment made light tight. When a diluted oil sample was analyzed] 2 ml. of absolute alcohol was added to the titration cell. The buret containing 0.0501, (weight per volume) of dithizone
(e)
938
ANALYTICAL CHEMISTRY
ML. OF DITHIZONE SOLUTION
Figure 1.
Typical titration curves of zinc with dithizone
1. Blank 2.
Zinc standard
3. Oil containing zinc Wave length, 640-rnp
in benzene was inserted through a small hole. The titrant was added in 0.1-ml. increments and the absorbance recorded after each increment. A typical titration curve is shown in Figure 1. The end point is a t the intersection of the extrapolation of the two straight lines.
I
A
I
RESULTS A N D DISCUSSION
Because of the known reactions of dithizone with various heavy metals, a cursory examination was made of the behavior of some of them in the photometric titration procedure. When cobalt or nickel was added to the titration mixture, the added dithizone reacted with the foreign metal before complexing the zinc. However, the colors produced by the cobalt and nickel dithizonates were not stable under the conditions described for the zinc titration. Consequently, the determination of zinc could not be made n-ith certainty. The addition of mercury(II), however, gave a somen hat different reaction. The mercury dithizonate was formed before that of zinc but the color produced was very stable and accurate absorbance measurements could he made. It was possible because of the extreme stability of the mercury dithizonate to determine both mercury and zinc simultaneously. The two metals can also be determined independently in the presence of one another by the use of two wave length settings as the mercury dithizonate absorption is nppreciably different from that of zinc. At the first wave length, 550 mp, mercury dithizonate does not absorb but zinc dithizonate does, so that the end point of the titration of mercury can be found photometrically by the absorption of zinc dithizonate. At the second wave length, 600 mp, both zinc and mercury dithizonates are transparent but the free dithizone absorbs, so that the end point of the titration
ML. OF DITHIZONE SOLUTION
Figure 2. Titration of mercury(l1) and zinc with dithizone
1. Wave length, 550 r n M 2 . Wave length, 600 mp
of total zinc and mercury can be found. The titration curves are shown in Figure 2. No indication was found that mercury(I1) formed more than one complex n-ith dithizone. The addition of lead(I1) gave a titration which was nearly identical with that of zinc and it was not possible to resolve the two metals. The two dithizonates gave absorption spectra which were very nearly the same with broad maxima occurring in the 505- to 530-mp region. The lead appeared to produce a slightly more absorbant species in this medium and on this basis it was concluded that the lead was complexed before the zinc. The titration of zinc in the pre,aence of added copper(I1) produced an unusual curve (Figure 3). The titration curve produced three distinct breaks which can be explained on the basis that
the stability of zinc dithizonate is bet\\ een the stabilities of the two forms of copper(I1) dithizonate, CuDz and C U ( D Z ) ~ Thus, . to the first end point, copper forms the 1 to 1 molecular comple\. CuDz, a i t h dithizone; betn-een end points l and 2, zinc forms a 1 to 2 complex with dithizone, Zn(Dz)*; 2nd between end points 2 and 3, CuDz ie,xt. nith another molecule of dithizone to form the 1 t o 2 complex, C U ( D Z ) ~ .A test with lead and copper gave a similar titration curve, indicating that the lead and zinc dithizonates tinre nearly the same stability. The titration of mercury in the presence of added copper was attempted t o shou any variation in complex sthility. The mercury dithizonate was formed before either form of the copper complex. This is in agreement with the previous result which indicated that the mercury dithizonate was more stable than the zinc. The addition of chromium(II1) did not produce any effect upon the titration of zinc. When iron(II1) was added to the titration solution, there was a slow reaction with the excess dithizone after the titration of zinc. This was believed to be an oxidation-reduction reaction between dithizone and iron rather than a complexation reaction. The rate of reaction was markedly influenced by the amount of alcohol added ; decreasing the alcohol concentration gave a much slower rate. However, the titration of zinc could not be made nith certainty even with the lower alcohol content. This interference by iron(II1) could perhaps be eliminated by the use of a suitable reducing agent prior to the addition of the dithizone reagent. Titration of Zinc in Oil. T o test the titration procedure, it was used to determine zinc in a lubricating oil to which various amounts of zinc dialkyldithiophosphate (commercial additive preparation) had been added. An
M L . O F D l T H l Z O N E SOLUTION
Figure 3. Titration of copper(l1) and zinc with dithizone
Wave length, 600 mp
additive concentrate was diluted with neutral oil to contain approximately 0.1 weight yo zinc. Because of the high zinc content of the oils, it was necessary to dilute about 0.1 gram to a volume of 25 ml. FT-ith benzene containing 8'3, methanol. A suitable aliquot of this solution (containing 30 to 50 y of zinc) was taken for the titration. The dithizone solution v a s standardized hy titration against the standard solution of zinc acetate in methanol. Results for determination of the zinc by photometric titration at a wave length of 640 mp and by a comparison polarographic method are shown in Table I. The addition of methanol to the benzene used to dilute the oil sample was necessary because dilutions which were made with benzene alone gave consistently low results (2 to 47,). The low results are probably caused by the absorption of the zinc dialkyldithiophosphate on the glass walls of the volumetric flask. Brodie ( 1 ) found the addition of alcohol necessary to eliminate a similar absorption of drugs from benzene solutions. I n the final titration mixture, niethanol was required to obtain reliable results. However, in this case, the effect of the polar alcohol may be to promote the reaction between the zinc additive and dithizone. In addition to the alcohol, a small aniount of ammonium acetate is also needed in the titration solvent. The ammonium acetate may be added by saturating the benzene or by dissolving a few milligrams in the titration solution. The ammonium acetate behaves similarly to the buffers used in extractive methods for metals with clithizone. Without the ammonium acetate, the excess dithizone color was not stable (fading slowly) and an accurate titration could not be made. Khen sodium rather than ammonium acetate was used, the excess dithizone faded rapidly from green to pale yellow. This behavior is similar to that observed in shaking dithizone in organic solvent with a basic aqueous solution, suggesting that the dithizone is converted to dithizonate ion and that the stabilization of dithizone can only be achieved under certain conditions of pH in the benzene-alcohol mixture. The photometric titration curve of zinc in oil was not identical with that for the standard solution. Near the equivalence point the curve showed considerable rounding (Figure 1). This effect probably indicates that the zinc is held more strongly by the additive than by the acetate and that there is a competition between dialkyldithiophosphate radical and dithizone in forming a complex with zinc. For the titration method to be usable, the strength of the zinc-additive bond must
be much n-ealier than the zinc-dithizonate bond as in the case of the dialkyldithiophosphate. The titration procedure should not be used indiscriniinately for the determination of zinc complexed in various additives; it is possible that some zinc additives are much more strongly bonded than zinc dialkyldithiophosphate. However, in testing the method, the procedure was applied to several new oils and greases containing various types of additives. In nearly every case a comparison betn-een the heavy metal content found by titration and polarographic assay showed good agreement. The photometric titration method should thus prove valuable for the routine control of additive content in blending new oils.
Table I.
Blend 1-1 2A
3-1 1B
2B
Photometric Titration of Zinc in Oil Weight yo Zinc
Photometric titration 0,0717 0.0733 0.0867 0.0864 0.0651 0.0657 0.0895 0,0901
0.1054 0.1075 3B 0.0823 0.0811 = Calculated using standard. b Calculated using standard.
Polarographic
Calcd.
0.0733 0,0905
0.087"
0.0660
0.06?9
0.0897
...
0.1090]
0. 1085
0.0820
0.0815
average of 1.1 as average of 1B as
The photometric titration of zinc using the Klett-Summerson colorimeter and the KS-66 filter supplied with the instrument was unsatisfactory, because of a nonlinear relationship between absorbance and the concentration of excess dithizone reagent. The trouble was traced to the KS-66 filter, which was described as having a 50- to 70-mp spectral band. However, on testing, the filter was found to be of the "cutoff" type and transmitted radiation of wave length greater than 640 mp (at least to 800 mp). When an interference filter transmitting a 35-mp band around 640 mp, manufactured by the Geraetebau-Anstalt-Balzers, was substituted for the Klett filter, results equivalent to those obtained with the Beckman spectrophotometer were obtained. LITERATURE CITED
(1) Brodie, B. B., 9th Annual Conference VOL. 30, NO. 5 , MAY 1958
939
on Analytical Chemistry, Los $ngeles, June 16, 1956. (2) Brummet, B. D., Hollweg, R. M., ANAL.CHEM.28, 448 (1956). (3) Gerhardt, P. B., Hartmann, E. R., Thid..I -29. 1223 - I ---- (lR.571.
\ - - - . / .
(4) Godd;, R. F., Hume, D. N., Zbid., 22, 1314 (1950); 26, 1740 (1954). (5) Irving, H. &I., Williams, R. J. P.,
Nature 162, 746 (1948); Analyst 77, 813 (1952); J . Chern. SOC.1953, 3192. (6) Pflaum, R. T., Popov, 4 . I., Goodspeed, X., . ~ ? ; A L . CHEX 27, 253 (\ -19.551. ---,-
(7) Pilipenko, A . T., Zhur. Anal. Khim. 8 , 286 (1953). (8) Riddick, J. A . , A l v . 4 ~ . CHEX 24, 41
(1952); 26, 77 (1954); 28, 679 (1956). (9) Vallee, B. L., Zbid.,26, 914 (1954).
RECEIVEDfor review June 10, 1957. ilccented Februarv ~" 3. - , 1958. Division of Refig-ing, 22nd Meeting, American Petroleum Institute, Philadelphia, Pa., May ~
1957.
Improved Conductometric Titration of Weak Bases W. H. McCURDY, Jr., and JOHN GALT' h i c k Chemical laboratory, Princeton University, Princeton, N. 1. )A new solvent system composed of a 1 to 1 mole ratio of 1,4-dioxane and formic acid has been developed for conductometric titration of weak bases. Routine titration of 10-mg. samples is readily attained if the base is sufficiently strong to react with the solvent to some extent. Equilibria studies have shown the factors responsible for the improved conductometric end points observed. Direct titration of mixtures of weak bases can b e realized if a difference of 1 ~ K B H +unit exists. Even smaller differences in basic strength can be resolved in favorable circumstances. Compounds that yield N-formyl esters with this solvent system a t room temperature may b e successfully titrated a t 0' C. in most cases.
A
number of nonaqueous titrations of weak bases using high-frequency conductometric apparatus have been reported. Glacial acetic acid was generally employed (8, 11, 12, 21), although a mixed solvent of benzene, methanol, and acetic acid was tried (14). The results check satisfactorily with potentiometric and acid-base indicator determinations in spite of the fact that the conductometric end points are generally located at the intersection of two ascending lines. The slope of the titration line before the end point beconies more nearly identical with the slope of the excess acid line as the basic strength of the sample decreases and the end point becomes increasingly difficult to locate accurately. Many workers have shown the improvement in conduetometric end point which occurs when a strong acid is titrated with a weak base (5, 6, 9), but such a procedure eliminates the possibility of determining mixtures of weak bases. Considerable improvement CONSIDERABLE
1 Present address, Harvard Medical School, Cambridge, Mass.
940
ANALYTICAL CHEMISTRY
in potentiometric end points may be realized by changing from glacial acetic acid to nitromethane-formic acid (20), trifluoroacetic acid (S), 4methyl-2-butanone ( 2 ) )or other special solrents. This study was initiated to discover solvents or solvent mixtures which would alter the slopes of the titration and acid lines in such a manner as to increase the sharpness of nonaqueous conductometric end points. INVESTIGATION
OF
TITRATION SOLVENTS
Preliminary u-ork was carried out titrating salts of weak acids with perchloric acid in a variety of simple and mixed solvents. All of these salts and most of the solvents n-ere reagent grade and were not purified further. Solutions of perchloric acid were standardized potentiometrically by titration of samples of potassium acid phthalate in the appropriate solvent. Conductance measurements were performed at 10 megacycles per second, using a General Radio Tn-in-T impedance bridge (1'7) and a t 1000 cycles per second with a Serfass conductivity bridge Model RCN-15. A concentric electrode titration cell with very thin glass walls was employed with the highfrequency instrument and a dipping type cell having a cell constant of 0.100 cni.-l was used for low frequency conductivity measurements. Similar results were obtained n ith either instrument, although the sensitivity of the high-frequency measurements \vas somewhat superior, because the high-frequency cell electrodes were 25 sq. em. compared to 1.0 sq. cm. for the dipping cell and because the high-frequency response is enhanced by decreasing the solvent dielectric constant (1'7). The cell constant for the high-frequency cell was 0.025 cm.-l in glacial acetic acid. Temperature control was not carefully maintained in the early experiments but volume corrections were applied to all titration curves. Effect of Solvent. Five-milliliter portions of 0 . W potassium acetate dissolved in the appropriate solvent
nere titrated with 0.1F perchloric acid in the same solvent. Dilution of samples t o 160 nil. with solvent n n s held constant in all experiments. With solvents of lorn dielectric constant, 5 to 20% by volume Tvater !vas added to increase the solubility of the sample and titration products. Table I, A, lists seven solvents in order of decreasing utility in this titration. The curves \\-ere plotted in a standard manner in order that slopes of the titration lines before the end point (slope B.E.P.) and excess acid lines after the end point might be compared. The ratio of slope of acid line to slope of titration line and the end point angle were calculated in each case. Ethylene glycol and glycerol were not considered promising because of the large solrent viscosity. Results with nitromethane are not included because no end point was observed in this solvent. Effect of M i x e d Solvents. A number of mixed solvents Lvere prepared by adding 10% by volume of acidic reagent to ethylene glycol nionomethyl ether and 1,4-dioxane-20% ivater. ,4 few typical results are given in Table I, B, for the titration of 0.5 nieq. of potassiuni acetate when 10% by volume of acetic, formic. or monochloroacetic acid was present. In all cases the starting volume was 160 nil. Acetic acid and nionochloroacetic acid solvent mixtures show only a slight effect on the slope ratio compared to the formic acid mixtures. Experiments with other acidic reagents such as dichloroacetic and anhydrous hydrochloric acid solvent mixtures did not yield any significant iniprovenient in 4ope ratio. A series of formic acid solrent mixtures was studied and many different bases titrated. d fen- examples are presented in Table I1 for solrent niixtures containing 10, 20, and 30% (by volume) 99% formic acid in purified 1,4-dioxane and 14% (by volume) 99% formic acid in glacial acetic acid. The increase in sharpness of conducto-
(e),
