Potentiometric Titration of Very Weak Acids Effect of Potassium on the Glass Electrode G. A. HARLOW Shell Development Co., Emeryville, Calif. ,The cause of the peculiar inverted inflections which frequently occur during the titration of very weak acids has been traced to potassium ion. The complex nature of the titration curves obtained is explained on the basis of two simultaneous titrations taking place. One involves the expected reaction between the very weak acid and the strong base. The second involves the titration of the tip of the glass electrode with traces of potassium ion which contaminate the titrant. The shape of the resulting curve will depend on the type of glass electrode employed, its pretreatment, the amount of acid being titrated, and the potassium content of the titrant.
V
ERY weak acids such as phenol are routinely determined in these laboratories by potentiometric titration in pyridine. Occasionally the titration curves obtained show peculiar inverted inflections which tend to obscure normal inflections and make interpretation difficult. Similar irregularities have been reported by van der Heijde ( I I ) , who explained them on the basis of insufficient solvation of the highly polar acid molecules and their ions by inert or weakly basic solvents. The incomplete solvation causes solutes to be adsorbed onto electrode surfaces, and this in turn
Figure 1. Titration of phenol in presence of potassium ion
148
ANALYTICAL CHEMISTRY
causes the anomalous "potential drops." The present investigation was undertaken in an attempt to obtain a more complete understanding of the phenomenon and to find a means of eliminating the troublesome inflections. After the source of the difficulty was traced to potassium contamination of the quaternary ammonium titrant, the investigation was expanded to include a more general study of the effect of potassium ion on the response of the glass electrode. In contrast with the considerable literature on the "alkali error'' of the glass electrode in aqueous solutions (6, IO), very little information is available on this effect in nonaqueous solutions. This is particularly true for nonaqueous solutions of extremely low hydrogen ion concentration of the type encountered in the titration of very weak acids. Acids such as phenol cannot be successfully titrated in aqueous solution because of the acidity of water. In less acid solvents, such as the amines and ketones, the titration can be readily carried out. Unfortunately, the very high basicity which must be created in order to titrate the very weak acids also greatly increases the alkali effect. It was shown several years ago that a glass indicating electrode could be used for the titration of very weak acids, if a titrant consisting of a solution of potassium hydroxide in isopropyl alcohol was employed (4). A titrant containing sodium aminoethylate, although presumably a stronger base, did not yield suitable inflections. It was shown later ( 7 ) that the failure of the sodiumcontaining titrant was due to its influence on the response of the glass electrode. I t was also shown (8) that the potassium-containing titrant, although superior to the one containing sodium, also reduced the response of the glass electrode. This led t o the development of nonaqueous quaternary ammonium titrants for the determination of very weak acids. Although the above information points to a gross effect of potassium, it gives no indication of the surprising results actually found: that traces of potassium can completely change the shape of a titration curve and shift the
Figure 2. Titration of phenol in pyridine with tetra-n-butylarnmonium hydroxide containing various proportions of potassium and quaternary ammonium ion
mid-point potential by several hundred millivolts. APPARATUS AND REAGENTS
The glass electrodes used in this study were of the Beckman generalpurpose and blue-tip types. The reference electrode was a Beckman sleeve-type calomel electrode containing IN tetrabutylammonium chloride (aqueous) as the bridge solution. Electrode potentials were measured with a vibrating reed electrometer (6) (Applied Physics Corp. rvlodel 31). The convention adopted for electrode potentials was such that the stronger the acidity of a solution, the more negative the potential. The tetra-nbutylammonium hydroxide titrant was prepared by an ion exchange method which has been described (8). (For the
800
E G700
i
600
500
400
Figure 3. Titration of various amounts of phenol in pyridine with 0.2N tetra-n-butylammoniurn hydroxide titrant containing sodium
I
0.2
V a 1 o r e of T ~ 1 m n t .rnl
Figure 4. Determination of alkali ion tolerance of glass electrode 1. 2.
Titration of phenol with 0.25N butylammonium hydroxide titrant. terminated at X Titration of resulting solution with potassium hydroxide titrant. resumed at X
tetra-nTitration 0.081N Titration
purpose of simplicity the titrant is referred to as hydroxide, although it is actually an equilibrium mixture of hydroxide and isopropylate.) Titrants containing various concentrations of potassium were prepared by mixing isopropyl alcoholic solutions of tetran-butylammonium and potassium hydroxide. The 0.018X potassium titrant used in the determination of the alkali tolerance of electrodes was prepared by diluting isopropyl alcoholic potassium hydroxide with pyridine. The resulting titrant solvent consisted of 80% pyridine and 207, isopropyl alcohol. It was prepared immediately before use, to minimize decomposition. Pyridine manufactured by the J. T. Baker Chemical Co. was used as the solvent for all titrations. Twenty milliliters of solvent was used in each case. EFFECT OF POTASSIUM ON TITRATION CURVES
The observed effect of potassium on the titration curve will depend on the manner in which it is introduced. Potassium may enter the titration solution from three different sources: the reference electrode, the sample, or the titrant. The calomel electrode is frequently used as the reference in the determination of very weak acids (1, 4, 8 ) . The bridge of this electrode is usually filled with a saturated solution of potassium chloride, and there is a tendency for small amounts of the electrolyte to leak out and contaminate the titration solution. The amount of leakage varies from one electrode t o another. In the case of the sleeve-type calomel electrodes it nil1 depend on how well the sleere fits the barrel of the electrode and how tightly the sleeve is pressed into place. Preliminary tests showed that interference from the reference electrode could be completely eliminated by means of a very simple modification: replacement of the potassium chloride bridge solution with a solution of tetra-
n-butylammonium chloride. All subsequent titrations were carried out with this modified reference electrode. The presence of potassium in the sample causes the titration curve for a very weak acid to shift to more negative potentials and to show a much smaller inflection. This is illustrated in Figure 1 for the titration of phenol in pyridine. Curve 1, which was obtained in the presence of less than 1 p.p.m. of potassium, has a total potential span of about 400 mv. and a mid-point potential of 525 mv. An almost identical curve is obtained in the presence of 4 p.p.m. (curve 2). At the 10-p.p.m. level, however, the total potential span is only 200 mv. and the mid-point potential has been shifted to 475 mv. A further increase in the potassium concentration to 320 p.p.m. has very little additional effect on either the span or the mid-point potential. The curves in Figure 1 show that the effect of potassium is a very nonlinear function of its concentration and that the greatest change with this particular
of Tl:ran:
Figure 5. Response of three glass electrodes to potassium ion
electrode occurs between 4 and 10 p.p.m. It is apparent that estimations of acid strength from the midpoint potentials of titration curves could result in large errors when potassium is present in the sample. The precision achieved in the estimation of the amount of acid present would also be expected to suffer as a result of the inferior inflections obtained. When potassium is present in the titrant, unusual inflections may occur in the titration curve. This is shown in Figure 2 for the titration of phenol in pyridine with a titrant (0.25N tetra-nbutylammonium hydroxide) containing various amounts of potassium. A concentration of 0.0028N potassium hydroxide has no apparent effect on the titration curve (curve 2). A potassium concentration of 0.0046N (curve 3) causes a change in the slope of the plateau of the titration curve and a decrease in the maximum potential. When a titrant containing 0.0055N potassium hydroxide is used (curve 4), an inverted inflection appears just prior to the normal inflection. Further
increase in the potassium content of the titrant (curves 5 and 6) causes the inverted inflection to show up earlier and greatly reduces the potential span of the titration. The curves in Figure 3 show that sodium in the titrant will have an effect similar to that of potassium. In this case the sodium content of the titrant was kept constant while the sample size was changed in order to vary the amount of titrant, and hence the amount of sodium, introduced. A normal titration curve (curve 1) was obtained when 4.5 mg. of phenol was titrated. When the sample size was doubled, the characteristic inverted inflection appeared (curve 2). B third titration in which the sample size was again doubled (18 mg.) is shown in curve 3. In those portions of curves 2 and 3 shown in broken lines the potentials were established more slowly and were less reproducible than potentials in the earlier portions of the titration. DETERMINATION OF POTASSIUM TOLERANCE OF GLASS ELECTRODES
To get a more quantitative picture of how the potential of a glass electrode varies with potassium concentration, an attempt was made to titrate with potassium a t a very low and constant hydrogen ion activity. The method used is illustrated by the curves in Figure 4. A sample of phenol was titrated in 20 ml. of pyridine with potassium-free 0.25N tetra-n-butylammonium hydroxide titrant (curve 1). At point X in curve 1, the solution has an excess of titrant and is in an area where the electrode potential changes little with added titrant. If the titration was continued] the potentials shown by the broken line would result. At point X , however, the titrant was changed to 0.018N potassium hydroxide in isopropyl alcohol and pyridine. Since this titrant is only 1/14 as concentrated as the quaternary ammonium titrant, its effect on the hydrogen ion activity should be very small. As a precaution an indicator] p-azobenzeneresorcinol, was added to the cell solution. The results obtained upon continuing the titration from point X using potassium hydroxide are shown in Figure 4
400
-
300
U
.
0.2
,
P.4
I
P.6
1
0.8 1 . 0 1 2 I 4 1 . 6 Va!urne of T i t i a n l . rn!
1.8
2.0
2.2
Figure 6. Effect of water on potassium sensitivity of glass electrode VOL. 34, NO. 1, JANUARY 1 9 6 2
149
Yclune of Titran'. mi
Figure 7. potassium electrode
Effect of conditioning on ion sensitivity of glass
(curve 2). During the course of the titration the potential of the electrodes dropped 400 mv.; the indicator color remained unchanged throughout. Thus curve 2 represents the effect of potassium on this particular glass electrode a t essentially constant hydrogen ion activity. FACTORS INFLUENCING POTASSIUM TOLERANCE
The response of three different electrodes to potassium was determined by the method described above. The results, shown in Figure s> indicate that electrodes differ greatly in the manner in which they respond to potassium. The difference in response is not due simply to the type of electrode, since 1 and 2 were both Beckman generalpurpose electrodes. The fact that the same electrode may vary in potassium sensitivity as a result of pretitration conditioning can be seen from the curves in Figures 6 and 7. Curve 1 in Figure 6 shows the potassium sensitivity of a blue-tip electrode which had been immersed in distilled water for 5 to 10 minutes before the titration. Curve 2 shows a similar titration carried out immediately afterwards Ivithout any electrode treatment escept rinsing with pyridine. Curve 3 shows results of an identical titration carried out after the electrode had been soaked in water for 24 hours. The two curves in Figure 7 show the results obtained on another blue-tip electrode after more drastic treatment. Curve 1 was obtained after the electrode had been soaked in ion exchange water for 2 hours with frequent changes of water. Curve 2 \vas obtained after the electrode had been rinsed off with acetone, stored dry for 48 hours, soaked in concentrated sulfuric acid for 2 hours, and then washed with dry pyridine. illthough the conditioning of the electrode greatly influences its potassium ion sensitivity, it has little effect on the shape of titration curves ob150
ANALYTICAL CHEMISTRY
tained in the absence of potassium ion. This is illustrated in Figure 8, which shows titration curves obtained with a blue-tip electrode with and without water soaking. Curve 1 was obtained with the electrode after it had been soaked in distilled water for 2 hours. Curve 2 shows the result of a similar titration, after the electrode was stored dry and then soaked in concentrated sulfuric acid for 2 hours. No water was allowed to touch this electrode, the sulfuric acid being rinsed off with dry pyridine. Curve 3 shows a duplicate titration performed after soaking the same electrode in water. A comparison of the three curves in Figure 8 shows that the dehydration procedure did not greatly affect the hydrogen ion sensitivity of the electrode. The great difference between the effect of the treatment on the hydrogen ion and potassium ion sensitivity can be seen from a comparison of Figures 7 and 8. DISCUSSION
It is noiv apparent that the unusual inverted inflections which led to this investigation were due to potassium in the quaternary ammonium titrant. Curves 4 and 5 in Figure 2, obtained with a titrant containing added potassium, sho.lv similar inverted inflections. Analysis of the quaternary ammonium titrant which was being used for the determination of very weak acids showed the presence of sufficient potassium to account for the unusual behavior (9). Three methods are a t present available for the preparation of nonaqueous quaternary ammonium titrants, and all of them are subject to possible alkali ion contamination. Although the silver oxide method of Cundiff and Markunas (2) does not utilize alkali metal compounds in the preparation step, sodium hydroside is used for treating the column in the purification procedure (S). If the sodium ion is not completely washed out, some of it will appear in the finished titrant. In a similar manner potassium ion contamination may occur in the nonaqueous ion exchange method of Harlow, Pl'oble, and Wyld (8). Since van der Heijde (11) used this method of preparation, some of the anomalous inflections which he observed may have been due to potassium ion in this titrant. Both of the methods described above will result in titrants with a negligible alkali ion concentration, if proper care is taken during preparation or purification. The third method of preparation, the potassium hydroxide method (9), yields titrants relatively high in potassium ion (100 to 200 p.p.m.). The use of this titrant
Volume of T i l r n n !
Figure 8. Effect of electrode conditioning on shape of titration curve in absence of alkali ions
without further purification m-ill almost always give rise to either dwarfed Lurves or anomalous inflections. The peculiar shapes of the titration curves may be considered as resulting from two simultaneous titrations. One of the titrations is the expected one, in nhich an acid is titrated with a basic anion (curve 1, Figure 4). The glass t,lectrode in this case responds simply to the change in hydrogen ion activity and a normal inflection is obtained a t the end point. The other titration is more difficult to visualize, since it involves the titration of the active surface of the glass electrode with potassium (curve 2. Figure 4). In a sense the glass electrode is acting both as the sample and as the indicating electrode. The end point inflection of this titration is opposite in direction to the normal titration inflection. (The term "titration" is used here in a broad sense without reference to a specific mechanism. The mechanism involved may be the transference of potassium ion.) I t is easy to understand n-hy the unusual inflections are encountered only occasionally in routine work. The sample size, the electrode conditioning, and the potassium content of the titrant must be such that the potassium end point is obtained before the acid end point. The degree to which potassium or other alkali ion will interfere depends not only on the amount preient but alro on the use which will be niade of the results, When the only information of interest is the amount of acid ill a sample, useful results can be obtained even in the presence of a considerable amount of alkali ion. The precision n ith which the end point can be selected will, however, decrease as the size and sharpness of the inflection are reduced. In the case of extremely n-eak acids titrated in the presence of large amounts of interfering ions, the inflections mag become so small that they are difficult to differentiate from random irregularities in the titration curve. When information on the relative strength of a very weak acid is of
primary interest, the presence of alkali ion is of even greater concern. The relative strength of an acid is determined from the mid-point Potential of its titration curve. It can be seen from Figures 1 and 3 that the mid-point potential is greatly influenced by potassiu”’ and sodium‘ Thus in the presence of these or other interfering ions: large errors Could be made in the Estimation of relative acidity.
LITERATURE CITED
(1) Bruss, D. B., Wyld, G. E. A , , CHEM.29, 232 (195i).
XNAL.
( 2 ) Cundiff, R. H., Markunas, P. C., Ibid., 28, 792 (1956).
( 3 ) Ibid., 30, 1450 (1958). (4)47Deal, v. z.7 \;vs.ld, G. €2. ~4.7Ibid.8 27, (1965). ( 5 ) Dole, M., “The Glass Electrode,” Wiley, ~ ~1941. ~ d ~ (6) Harlow, G. A., Bruss, D. B.. .%SAL. CHEM.30, 1533 (1958).
( 7 ) Harlow, G. A., Noble, C. SI.,\\-yld G. E. h.,Ibid., 28, 784 (1956;. (8) Ibid., p. 787. (9) Harlow, G. il., Wyld, G. E. .i.,Ibid., 34, 172 (1962). (10) Landquist, N., Chem. S c a n d . 9 , 595
(1955).
(11) van der Heijde, H. B., .Anal. Chim. Acta 16, 392 (1957).
RECEIVED for review July 14, 1961. ;ic~ , cepted November 6, 1961. Presented in part at Gordon Research Conference on Analytical Chemistry, ilugust 1961.
identification of Nitrogen Compounds by Catalytic Denitrogenation C. J. THOMPSON, H. J. COLEMAN, C. C. WARD, and H. T. RALL Bartlesville Petroleum Research Center, Bureau o f Mines, U. S. Department o f the Interior, Bartlesville, Okla.
b A micromethod, developed for the catalytic hydrogenation of organic sulfur and oxygen compounds, has been applied to nitrogen compounds. This technique rapidly and quantitatively removes the nitrogen atom from organic nitrogen compounds to produce paraffins and cycloparaffins. The identification of the produced hydrocarbon identifies or contributes to the identification of the nitrogen compound precursor. Such identifications would otherwise be difficult or impossible in many instances because of the limited number of nitrogen-bearing reference compounds available. Data are presented showing the application of this technique to pure compounds and to mixtures of nitrogen compounds of unknown composition.
Table I.
Reactant 1. 2-Aminoheptane 2.
N
TKO
Product yH2
c-c-c5
n-Cn-C,
2,5-Dimethylpyrrole
I
H n-Cs
3. 2,5-1Xmethylpyrrolidine I
H
4.
Pyridine
5, 2-Methylpyridine
n-C5
Q
c
n-Cs C
I
c-c-e:,
6. 3-Methylpiperidine i . ;Iniline 8.
I
Catalytic Removal of Nitrogen from Nitrogen Compounds
S,n’-Dimethylaniline
earlier papers, Thompson
et al. described the catalytic removal
of sulfur from organic sulfur compounds ( 3 ) and of oxygen from organic oxygen compounds (2) t o leave the residue of the treated molecule in a reduced but otherwise unfragmented state. This hydrocarbon residue is a guide t o the identity of its precursor. The technique is applicable to gasliquid chromatographic fractions as small as 0.0002 ml., and thus is a potent tool in the identification of sulfur and oxygen compounds in small samples of unknown mixtures. The procedure used for sulfur compounds failed to remove nitrogen from nitrogen compounds Eithout complete fragmentation of the molecule or loss of sample by excessive degradation or possibly by adsorption on the catalyst. Substitution of a milder catalyst in the procedure
9. 2,5-Dimethylaniline r
10. Sitrobenzene
11. Benzonitrile 12.
Indole
13. Quinolineo 14.
Carbazolea H
15. 9-blethylcarbazole
r C
a
w
Other quinolines and carbazoles listed in Table 11.
VOL. 34, NO. 1, JANUARY 1962
151
