The iron(I1) complex exhibits a wavelength of maxinium absorption a t 523 niQ with a molar absorptivity of 16,800 (Figure 1). Solutions of the complex within the optimum pH range conform to Beer‘s law over the concentration range studied, 6.88 X 10;12-6 to 1.24 X 10-4X (0.4 to 7.0 p p n ) iron(I1) ion. The rate of complex formation between iron(I1) and the reagent is relatively slow a t room temperature. ;\faximum color is developed within a few minutes upon heating at 80’ to 90” C. Solutions treated in this manner retain their maximum color for a period of 6 hours. The presence of a reductant in the reaction mixture facilitates the reduction of any iron(II1) to the iron(I1) oxidation state. Sodium dithionite was the most satisfactory reducing agent
for this purpose. The addition of a few milliliters of a freshly prepared solution of the reducing agent is recommended in the analytical procedure.
(inherent in the ketoxime method), it has the added advantage of simplicity. LITERATURE CITED
(1) Hartkamp, H., Naturwissenschaften RESULTS O N SELECTED SAMPLES
Results of iron determinations on selected alkali samples are summarized in Table I. Yalues obtained from different methods for iron in sodium hydroxide are presented in Table 11. The data in Table I indicate that the method yields reproducible results. The data in Table I1 show that the method gives results which are comparable to those obtained by standard methods. Because the recommended procedure does not involve neutralization and pH control (vital in the phenanthroline method) nor extraction
45, 211 (1953).
(2) Hartkamp, H,, Z. Anal. Chem. 170,
309 (1959). (3) Schilt, A. A, Smith, G. F., Heimbuch, $., ANAL.CHEX28, 809 (1956). (4)Trussell, F., Diehl, H., Ibid., 31, 1973 (1959). MILANW,WEHKING~ RON.ALDT. PFLAUV E. SCOTT TUCKER I11 Department of Chemistry University of Iowa Iowa City, Iowa 1 Present address, Department of Chemistrv, Wisconsin State University, River Falis, Wis. WORK supported by National Science Foundation Grant G 15738.
Direct Determination of Sodium und Potassium in the Presence of Ammonium with Glass Electrodes Automatic Continuous An a1ysis of Urine S ~ R There : is a growing literature on the use of glass electrodes for monitoring sodium and potassium. The singular characteristic of the electrode method, compared to classical procedures such as flame photometry or gravimetric analysis, is that activity is measured in contradistinction to concentration values given by the other techniques. ,ictivity values are, of course, more basic than concentration values, because thermodynamic events (e.g. transport phenomena, complexation, precipitation) are a direct function of activity. Kevertheless. in many instances concentration values have been desired and undoubtedly will continue to be. The simplified technique afforded by the electrode is attractive. Also, the possibility of measuring both activity and concentration in a sample exists. It has been recognized that when employing the electrode, variations in the activity coefficient must be taken into account ( 6 ) . Often such variations have been anticipated by the maintenance of a constant ionic strength (3, 7 ) . However, this technique has not been explicitly utilized to enable a direct measure of concentration. Some previous techniques for obtaining concentration values from electrode nieasuremente necessitated employing empirical activity coefficients (4,or dependence on standards that very closely approximated the unknown in composition (6) It is the purpose of this report t o illustrate a simple and
direct technique for obtaining concentration values when employing the electrodes, and t o apply the technique to the automatic (continuous) analysis of urine, a substance of niuch variability. Furthermore, a means is illustrated that allows the determination of potassium in samples containing ammonium ions with a cationic electrode which has almost equal sensitivity to both. EXPERIMENTAL
The Beckman sodium ion glass electrode (KO. 39046) and cationic glass electrode (KO. 39047) were employed. The electrodes were fitted into two Beckman (No. 14846) sample chambers (ground glass connections) employing a single reference electrode, (,4g, AgCl internal, S o . 39070) that was mounted remotely and fitted with a irY’r to effect connection with both sample chambers. A Becknian electrode switch box (No. 97200) was used to alternately read one or the other electrode. Because it was millivolts that were recorded, the bucking (standardizing) potential of the switch box was defeated by removing the mercury batteries. The Beckman research pH meter, with a readability of 0.05 mv., was employed. For recording, the output of the pH meter was taken to a Sargent SR recorder arranged to give a 100-m~.full scale, thereby preserving a few hundredths of a millivolt readability. Standardization of the electrodes followed, in general, the procedure of Friedman et al. (1). Two sodium concentrations were used to establish the
slope for each electrode and one niixed sodium potassium solution was used to establish the selectivity constant for the cationic electrode. When analyzing human urine, standard sodium chloride solutions of 0.2N and 0.1N were employed and the mixed solution contained 0.1N NaCl and 0.05N KC1. For dog urine, the standards employed were 0.021V NaC1, 0.05N NaCl, and a mixture of 0.0ZX KaC1 and 0.02N KCl. The sodium values were obtained from a calibration curve. The potassium values were obtained by solving the following equation:
AE = S l o g
[(Ka-)
+
1
k~,rc(Ii+) (Xa+)std.
where AE is the difference in millivolts between a sample and a standard containing only sodium, S is the slope, k N a K is the selectivity constant of the cationic electrode for potassium relative to sodium, which is first determined by using standard solutions. At a pH of 8.15 k.\7v,X was about 11 -I 1; a t a pH of 11.55, ICN~Kwas about 7 i 1. L4 precise value was always determined before, during, and after a series of measurements and the value was constant to i 2%. Titrations were performed in beakers using a magnetic stirrer. For continuous automatic analysis, a Technicon pump, sampling table, tubings, and fittings were employed. Considerable electrical noise was encountered traceable to the pump. The noise was eliminated by inserting a short length of platinum wire into the stream just past the de-bubbler YOL. 38, NO. 13, DECEMBER 1966
195’1
fitting and connecting it t o the sohtion ground post of the pH meter. The schematic flow arrangement is given in Figure 1. A fourfold dilution of sample was effected with the tubing used. Because of the unavailability of a second channel recorder and meter, the switch box was manually operated to alternately read each electrode; addition of a second read out channel should present no difficulty. Each sample was aspirated for 4 minutes. This relatively long time interval was partially due to the lack of two channels. Nevertheless, it was observed that the response was more sluggish than that of a measurement of a step concentration change using a beaker with stirring. However, little attempt was made to fully optimize the rate, and unknown factors exist in the displacement of sample within the Beckman sample chamber and perhaps in the flow kinetics associated with the Autorlnalyzer. Solutions and standards were prepared from reagent grade materials. The buffer used with the sodium electrode was a mixture of 0.158 mole of acetic acid and 0.261 mole of 2-amino2 - (hydroxymethyl) - 1,3 - propanediol (Tris) in 1 liter of water giving a pR of 8.15 and having an ionic strength of very close t o 0.16. The buffer used with the cationic electrode was a mixture of 0.095 mole of acetic acid and 0.34 mole of diethylamine in 1 liter having a pH of 11.55 and an ionic strength of approximately 0.11.
P roporfioning Pump
II
Figure 1. potassium
The relationship between the activity of a species and its concentration is given by :
YiCd where 7 %is the activity coefficient. To determine concentration from an activity measurement without determining a, =
Analysis of Urine Samples for Sodium b y Flame Photometry and by the Electrode
Flame photometry 0.157~V 0.093
Electrode (whole urine) 0.143N 0.078 0.151
0.176 0.200 0,185 Table II.
Buffer added, ml. 0 10
25 45 95
135 1952
0.163 0.170
Per cent age difference col. 2 8: 3 7.7% 14.2 12.7 18.1
Electrode (100-fold dil. urinc X 100) 0 . l55N 0,091 0.173 0.199 0.185
8.1
Calculated Activity Values for 1 5-MI. Samples Diluted with Tris-Acetate Buffer of 0.1 6 ionic Strength Urine containing 0.2N NaCl with an assumed I of:
0.2 NaCl
0.4
I
a
0.200 0 184 0.175 0.170 0.165 0.164
0.144 0.0870 0.0547 0.0365 0.0200
0.0147
ANALYTICAL CHEMISTRY
I 0.400 0.304 0,250 0.220 0.193 0.184
1.0 a 0.133 0.0824 0,0527 0.0356 0,0197 0.0145
Air0.42m//rnin.
Arrangement for continuous analysis of sodium and
the activity coefficient, it is necessary to extrapolate toward infinite dilution where the activity coefficient tends to one; or as an alternative, t o maintain the actkity coefficients of samples and reference standards constant and equal to each other. A close approximation to the latter may be obtained by diluting samples and standards with a buffer containing a moderate amount of an inert salt. The resultant of this dilution and, swamping with inert salt are solutions of very nearly equal ionic strengths and, consequently, very nearly equal activity coefficients. Thus the direct comparison of sample measurements with those of concentration standards is possible. Initial observations are recorded in Table I. Five human urine samples were analyzed by flame photometry for sodium and by the sodium sensitive electrode in beakers using whole urine and urine diluted 100-fold with water.
DISCUSSION AND RESULTS
Table 1.
\l ,1
I 1.oo 0.664
0.475 0.370 0.275
0.244
a 0.121 0.0757 0.0491 0.0337 0.0189 0.0141
The p H of the diluted samples was maintained a t 7.5 to 8 by the addition of several mg. of solid Tris. It is apparent that the true sodium values are approached in the diluted samples where the activity coefficients converge toward one. It was also observed that samples diluted with water increased measuring difficulties (noise and drifting) in beakers and especially in the AutoAnalyzer. Diluting with moderately concentrated buffers did nct give rise to these phenomena. Because of its high sodium content and t$he greatest percentage difference at the two concentration levels, 15 nil. of the urine containing 0.2N NaCl was chosen for titration and the data are compared in Figure 2 with 15 ml. of 0.2M sodium chloride similarly titrated. When titrating with water, the curves for the urine and the sodium chloride solution conrerge a t about a tenfold dilution; with the tris-acetate buffer convergence is a t about a fourfold dilution and superimposition thereafter is excellent. On the basis of these data, the Autohnalyzer was arranged t o effect a fourfold dilution. If one calculates the resultant activities during the titration using the Debye-Huckel equation, -logy
=
0.51
1
(1)Il2
+ 1.32(1)”2
the results shown in Table I1 are obtained. The activities given therein are the product of the activity coefficient calculated by the above equation and the concentration of sodiuni a t the various stages of the titration. Because these calculations are for single ions, they are not thermodynamically rigorous; furthermore, there is a question as to the validity of the assumption that the activity coefficients are made equal, especially a t the ionic strengths used.
Urine
4
Y
.o 2 / . 2 0
+a
u'l
.IO .08 .O 6
.04
I
/ 140
150
Titration With Trir- Acetate
180
170
180
190
ZOO
Mv Figure 2. Titration of taining 0.2N NaCl
0.2N
NaCl and of a urine sample con-
Nevertheless, the calculations do indicate, a t least to within about 2y0,that equality is achieved (and this is borne out by the data illustrated in Figure 2) between 0.212' sodium chloride and the urine containing 0.2N sodium chloride assuming a high ionic strength of 0.4 for the latter. It can probably be safely assumed that the ionic strength of urine is lower than 0.4, in which instance a closer correspondence can be expected. Furthermore, in most instances one of the standards used is of a higher concentration in the ions of interest than the sample thereby tending t o minimize differences in ionic strength. Subsequent dilution therefore would again more readily bring the two into correspondence. For comparison purposes, the extreme case of the urine having an ionic strength of 1 is also presented and convergence of the activities is evident here too. Use of the cationic electrode to monitor pot8assiumrequires a correction for sodium assuming the absence of other interfering ions. However, there is present to some extent in human urine, and to a large extent in dog urine, ammonium ions t o which the cationic electrode is almost as sensitive as to potassium ions. T o preclude the interference of ammonium ions, a high p H buffer (diethylamine-acetic acid) was employed so as to cause most of the ammonium ions to be in the form of undissociated ammonia (3). Table I11 shows the results obtained using the AutoAnalyzer. Assuming that an extreme ammonium ion concentration would be equal to the sodium content, the buffer capacity is seen t o be adequate. It is unlikely that the ammonium concentration would be so high. The resultant p H of a 1:3 mixture of 0.1 iYa+/0.5 KHIT to Et&H
Electrode ( m e q . 1 I ) Figure 3. Comparison of electrode and flame photometric measurements of sodium and potassium concentrations in dog urine
human urine. The potassium analyses were made using the tris-acetate buffer a t p H 8.15 and subsequently the diethylamine acetate buffer a t p H 11.55. Only a few samples gave accurate potassium results a t the lower pH. Presumably those that did so had little or no ammonium present. Figure 3 illustrates the results obtained on 33 dog urine samples. The average percentage difference between the two techniques was 3.1Y0for sodium with a %yo confidence limit range of +0.19 to f0.62 meq./l. For potassium
buffer is 10.f2; a t this p H equilibrium calculations show that about 3.4y0of the ammonia is in the ammonium state. With 0.1 Sa+/O.l NH4+ the p H observed was 11.34 corresponding t o about 0.8Y0 of the ammonia remaining in the ammonium state. During the autoanalysis five dog urine samples chosen a t random showed a pH range of 11.2 t o 11.4 after dilution with buffer; the ammonium present a t pH 11.2 is about 1.1% of the total ammonia. Table IV illustrates the results obtained using the Autohalyzer with
Analysis for Potassium in Presence of Ammonium
Table 111.
Solution (N) 0.10 Na-/O.O50 K+/0.040 NHA' 0.10 Na+/O.O5O K+/O.lO NH4+ 0.10 Na+/0.050 K+/0.50 KH4+ 0.10 Na'/O. 040 KH4+ 0.10 Na+/O. 10 NH4 0.10 Na'/O. 50 NH4
Table IV.
Sodium Flame spect. 0 . 109N 0.115 0.198 0.148 0.140
0.159 0.190 0.195 0.121 0.126 0.138 0.173 0.200
Potassium found ( N ) 0.0510 0.0502 0.0626
