160
Anal. Chem. 1904, 56,160-163
Sodium Glass Electrode in Propylene Carbonate for Study of Uranyl Ion Complexed to 18-Crown-6 and Substituted Derivatives Pierre Fux, Janine Lagrange, and Philippe Lagrange* ERA 166, Ecole Nationale Superieure de Chimie, 1, Rue Blaise Pascal, 67 000 Strasbourg, France
Complexatlonof the uranyl Ion (UO;') by 18-crown-6 and Its substituted derlvatlves (dlbenro and dlcyclohexyl) has been studied in propylene carbonate (0.1 M TEACIO,) at 25 OC, using Na+ as a competlng Ion and a Na' glass selectlve electrode. The response of the Na+ glass electrode In propylene carbonate wlth H+ as an lnterferlng ion has been studied. Concentrations of Na+-free Ions were calculated by using the extended Nlcdsky equation. The stability constants determined by this method agree wlth those determlned spectrophotometrically.
The present study was undertaken to determine stability constants of complexes formed by the uranyl ion UO?+ with crown ethers, especially the 18-crown-6 (1843-6) and the two substituted dibenzo-18-crown-6 (DB-18-C-6) and dicyclohexyl-18-crown-6 (DCH-18-C-6), for potential application in uranium trapping techniques. The solvent was propylene carbonate (4-methyl-l,3-dioxolan-2-one, PC). PC is a dipolar aprotic solvent classified as a dissociating medium, with high dielectric constant (69.0 a t 25 "C) and dipole moment (5.20 D)so that it does not exhibit any strong self-association or well-defined intermolecular structure. Therefore, it is possible to dissolve salts up to high concentration. Spectrophotometry can only be easily employed to study complex formation between U(V1) and ring-substituted crown ethers when ligand concentrations are inferior to M. This limitation is due to the high absorbance of the ligand (benzyl group). Thus, the spectrophotometric method is inappropriate here to study weak complex formation and to extend the studied ligand concentration as high as M. A potentiometric technique has been chosen, because this method is suitable for studying cation complexation in a large range of ligand and metal concentrations. A competitive potentiometric method could be performed by using Na+ or K+ as an auxiliary ion: Na+ has been selected. Other usual metals, for example, lead and silver, have been eliminated: lead salts are sparingly soluble in PC and silver salt solutions are not sufficiently stable. The sodium perchlorate salt can be easily dissolved up to 0.1 M concentration. Cationic glass electrodes, especially sodium ion selective electrode and H+ glass electrodes, are commercially available, and it has been proved that these electrodes work properly in this solvent (1). The responses of the cationic glass electrode to alkali metal ions in acetonitrile, dimethylformamide, and propylene carbonate have been measured by McClure and Reddy (2). For these reasons, we thought that the sodium ion would be an appropriate auxiliary ion. It is worthy of note that, as the uranyl perchlorate salt is dissolved in PC, we observe a residual acidity arising from the method of its preparation (perchloric acid acting on an uranate) as well as residual water traces. Then, we had to carry out measurements of the sodium ion concentration with a sodium glass electrode (NaGE) in the
presence of H+ ion. The properties of cationic glass electrodes in aqueous solvent are frequently reported in the literature ( 3 , 4 ) . The half cell potential of a monovalent ion selective glass electrode in the presence of one interfering ion can be written in the general form of the extended Nicolsky (5) equation
E = const.
+ n(RT/F) In [ailln+ (k*,/,a,)l/"]
(1)
where a, is the activity of the primary ion to which the electrode is selective, u, is the activity of the interfering ion, and n is a constant specific for a given glass and pair of cations. In the case of a monovalent interfering ion (for example H+) n = 1 (5, 6), thus E = const. + 2.303(RT/F) log [ai+ k*,,,a,] (2) The potential selectivity constant is given by k*,/, = (uL/u,)k,/, where uJu, represents the ratio of the individual ionic mobilities in the glass and k,,, is the ion exchange equilibrium constant. Response data for monovalent ion interferences have been published by some authors (5-11). The potential selectivity constant for a given electrode can be either invariable with the interfering ion concentration or a function of the concentration of the interfering ion or a function of the ratio of the concentrations of the measured and interfering ions. Determination of the residual acidity in PC can be performed potentiometrically with glass (1,12) or hydrogen electrodes (13, 14). In view of all these essential points we have made use of the NaGE and focused attention on the response of the sodium electrode with H+ as an interfering ion. H+ concentration has been simultaneously measured with a H+ glass electrode (HGE).
EXPERIMENTAL SECTION Reagents. All chemicals were analytical reagent grade. The crown ethers were obtained from Merck and used without any purification. The perchlorates of sodium (Merck), tetraethylammonium (TEA') (Fluka), and uranium (Ventron GMBH) were carefully dried in vacuo to less than 0.02% HzO. The residual water content was determined by Karl Fisher titration. The concentration of the uranyl solution was determined by pulse polarography in aqueous 0.5 M hydrochloric acid (15). The aqueous salt solutions (for the bridge) were made up with doubly deionized water. PC (Fluka) was purified according to Gosse (16) and Spiess (17). All others solutions in PC contained after purification less than 100 ppm HzO. Ionic strength was maintained at 0.1 by addition of TEAC104. Apparatus. Potentiometric measurements were carried out with an Aries 20000 Tacussel millivoltmeter and Tacussel ISIS 20000 pH meter. The sodium ion selective electrode was a Tacussel PNAV electrode (stored in PC) and the H+ "high alkalinity" glass electrode was a Tacussel TB/HA (stored in a slightly acidic PC solution). The reference electrode was a Ag/AgCl electrode (18). Titrations were carried out in a thermostated cell at 25.00 f 0.05 "C with a IO-mL Tacussel Electroburex buret. RESULTS Sodium Glass Electrode Response in PC with H+ as an Interfering Ion. A preliminary study has shown that if
0 1984 American Chemical Society 0003-2700/84/0356-0160$01.50/0
ANALYTICAL CHEMISTRY, VOL. 56, NO. 2, FEBRUARY 1984
M and lom5M < [Na+] < 5 X [H+] is superior to M, H+ is an interfering ion for the sodium glass electrode (NaGE). The acid concentration in the PC solution could be reduced by adding a base (for example, TEAOH) but this would introduce too much water and increase the risk of forming hydrolyzed species of UOzz+. An electrochemical reduction of H+ is not possible in PC without reducing the uranyl ion. The [H+]had to be measured with a H + glass electrode. We used the potentiometric double cell
-1
NaGE test solution I TEACIO, in PC 0.1 M HGE (TEACIO,, I = 0.1)l in PC
1 TEACIO, I TEACl I 0.1 M 0.1M 1 in H,O i in H,O 1
HC1
( K = 875)
Perchloric acid in PC is not completely dissociated according to its equilibrium (21) HC104 + H+
+ C104-
(K = 5
-Y 200=e E
150-
c J
2z loow
I I
( K = 3.2 X
+ C1- 2 Cl(HC1)-
I
50-
AgCl Ag
Liquid junctions were formed with glass frits (porosity 4). As far as a possible pollution of the test solution by water is concerned, it is worth pointing out that only the salt in the PC solution (in the bridge) is connected to the test cell, so that contamination will be limited during the short experiment time. The sodium sensitive electrode response was studied M NaC104anhydrous PC solutions. After in lo4 M to 5 X each addition of Na+, the emf was recorded when the average change in potential was inferior to 0.02 mV min-'. The stability of the response was generally obtained after 5 min. Some indication of the durability of the NaGE electrode in PC was obtained by measuring its response to the sodium ion after storage in the solvent. The NaGe'electrode stored in PC for 6 months showed no signs of deterioration. The calibration of the HGE must be carried out against acid solutions prepared by bubbling hydrogen chloride through the solvent; the following points must be taken into account: Hydrogen chloride is a weak acid in this solvent (19) and dissociates according to thd main equilibria HC1 + H+ + Cl-
250
161
X
Three protonated forms of water, in detectable amounts, exist in PC (14) in the presence of acid. The reactions may be represented by
H+ + H20 e H30+ ( K = 6.3 X lo4)
+ 2 HzO e Hb02+ ( K = 1.3 X lo6) H+ + 3 H20 e H703+ ( K = 3.2 X lo7) H+
Taking into account all these equilibria, the standardization of the HGE was done in all experiments with the same standard solution (HCl dissolved in PC). Total hydrogen chloride concentration in PC was determined by titration. A PC sample containing HC1 was diluted with water and titrated with a standard sodium hydroxide solution. Knowing the residual water content of the solvent and the perchlorate and chloride concentrations, we calculated the free H+ concentration by using the most recent version of the computer program HALTAFALL (22). Our standard solution was [HCl] = 1.428 X M, [HzO] = 2.778 X M, and [TEAClO,] = 10-1 M for which we wrote -log [H+] = pH = 4.36. After this standardization, the pH read on the pH meter allows the determination of H+ concentrations [H+]. Initially measurements were carried out at [H+] = and M: the same response showed that interference of H+ ion is negligible in that [H+] range. The NaGE electrode responds linearly (Nernstian slope, 59.2 f 0.1 mV vs. log [Na+]) in the range -log [Na+] 1.30 to 3.40.
0-50-
30
1
25
20 -log [Na+lanal
15
Figure 1. E'(mV) vs. -log [Na+Ianal ([Na+Ianal is the analytical concentration (M) of the sodium ion) titration by lo-' M NaCIO,: ).( 10 mL of 2.171 X lo-* M 18-C-6; (0)10 mL of 1.728 X IO-* M DB18-C-6; (A)8 mL of 2.050 X lo-' M DCH-18-C-6.
Now we can study the influence of H+ on the NaGE. The determination of the selectivity constant k*Nap was thus carried out. Equation 2 becomes
E = const.
+ 59.16 log ( U N ~+ k*Na/~uH)
(3)
The emf of the cell with the NaGE is
E' = E',
59.16 log ([Na+l
(fH/fNa)k*Na/H[H+l)
(4)
where E'0 is the algebraic sum of the following constants: NaGE standard potential, logarithm of the Na+ activity factor, junction potentials, and reference electrode potential. The equation becomes 1@C'/59.16 x 1O-E'o/59.16 = P a + ] (fH/fNa)h*Na/HIH+l ( 5 )
1~159.16may be a linear function of [Na'l if (fH/fNa)k*NalH and [H+] are constant. The plots of loE'/59.16 against the concentration of Na+ are given in Figure 2 for pH 5.50 and 6.50, We obtained parallel straight lines, so that k*NalH is independent of the [Na+] values, but the intercept on the Y axis shows that k*NaiHis [H+] dependent. The values of the selectivity constants indicate that the Tacussel NaGE is more reponsive to H+ than to Na+ in PC (f~/f~,)k*N,iH= 1.36 X lo4 at p H 5.50 = 1.74 X lo4 at p H 6.50
Complexation of Na+ by 1842-6, DB-18-C-6,a n d DCH1842-6. The study has been performed for H+ concentration inferior to M. Thus, no H+ interference on the NaGE electrode response had to be taken into account. Titrations of crown ethers were performed by Na+ ions. Figure 1 shows the titration curves, E' plotted against -log [Na+Iw Stability constants calculated with the MINQUAD computer program (20) (a Gauss-Newton least-squares method) are reported in Table I. Only 1:l complexes are found in the studied concentration range. Complexation of UOzz+by 18-C-6,DB-1842-6, a n d DC18-C-6. A preliminary study without ligands has shown that UOzz+does not affect the NaGE potential response. The determination of the complexation constant of UOZz+ by crown ethers (L) has been performed by a competitive method with Na+. Titrations of U022+-crown ether mixtures by Na+ were carried out between pH 5.50 and 6.50. [H+] and [Na+] were determined simultaneously with the double cell. HGE was
ANALYTICAL CHEMISTRY, VOL. 56,NO. 2, FEBRUARY 1984
162
Table I. Logarithm of Stability Constant (log 0 ) Values for the 1:l Complexes in PC, at 25°C and I = O . l a ligand
potentiometry
spectrophotometry
5.60 i 0.01 5.31 i 0.04 5.03 i 0.03 5.50 i 0.12
5.29 i 0.01
cation
18-C-6
Na’
DB-18-C-6
UO,z+ Na+ UO,z+
DCH-18-C-6b Na’
5.70
UO,z+
i
280}
5.51 * 0.31
0.07
5.63 i 0.13
5.75 i 0.15
a The uncertainty limits are twice the computed standard deviation. Mixture of isomers A and B: the stability constant is only an apparent constant valid in our experimental conditions.
200
25
20
15
- log[Na+lanal
Figure 3. E’(mV) vs. -log [Na+Iana,, titration by lo-’ M NaCIO,: (W) 7 mL of 4.179 X 1 0-’ M 184-6 and 4.145 X 1 0-’M UO:+; (0)7 mL of 2.089 X lo-* M DB-18-C-6 and 1.778 X lo-’ M UO+ ;; (A) 8 mL of 2.539 X lo-‘ M DCH-18-C-6 and 2.657 X lo-’ M UO+ :.
O‘
i
10
5
15 [ Na’
20
25
1 x lot3
Figure 2. 10E’’59.16 X lov3vs. [Na+Iana,:(A) pH 5.500 rt 0.002; (W) pH 6.500 f 0.002.
standardized as previously. NaGE was standardized “in situ”, before any addition of sodium, so that
where pH is measured with the HGE and ( f H / f ~ ~ ) k * N ~isp an interpolated value (Figure 2). The titration curves are given on Figure 3. The stoichiometry of the complexes and their stability constants were determined with the MINIQUAD program, input data being analytical Na+ concentrations and free Na+ concentrations calculated by using [Na+Ifree=
10(E’-E’0)/59.16 -
fH -k*
Na/H[H+]
(7)
fNa
At each titration point, [H+] is measured with HGE and E’ with NaGE and k*Na/H(fH/fNa) is interpolated. Experimental data treatment has shown that the equilibria can be written
Na+ + L uozz++ L
2 T=’c
NaL+ UOZL2+
given that for the two cations, only 1:l complexes are formed with these crown ethers in our experimental conditions. The stability constants are given in Table I. The values of these stability constants in PC determined by a competitive potentiometric method using a sodium selective electrode are compared in Table I to those found by spectrophotometry (23) in another concentration range of ligand.
DISCUSSION A comparison of the stability constants determined by potentiometry (this work) and spectrophotometry (23) in two different ranges of ligand and metal concentrations can be
made. The same complexes with very similar stability constants are obtained, and this demonstrates the validity of our work. Another result is that only 1:l complexes are obtained in PC in a large range of concentrations of ligand (or metal) M (spectrophotometry). The M (potentiometry) to first remark allows one to conclude that the ssdium selective electrode can be used to study Na+ complexation and other cation complexations in PC, by a competitive method with sodium. The use of the competitive potentiometric titration can be of increasing interest when determining stability constants in the case of substituted ring crown ethers or formation of complexes which absorb either too weakly or too strongly, and therefore make spectrophotometry particularly difficult or even impossible. This procedure appears appropriate for systems with similar metal-ligand stability constants. This is one limitation of the method. The nearly ideal slope of the NaGE response in the range of sodium concentrations (-log [Na+] = 1.30 to 3.40) shows that the Tacussel sodium selective electrode works properly in PC and can be well behaved in PC containing H+ as an interfering ion and water traces. The H+ concentration must be limited to obtain an accurate measurement of [Na+]. As eq 4 shows, ( f ~ / fNa)k*Na/H[H+] must not be greater than [Na+]. Practically, with the Tacussel electrode, H+ has to be kept inferior to M during titration. The study of the complexation of UO2+ by substituted or nonsubstituted 1 8 4 - 6 shows that 1:l complexes are only observed in PC in our experimental conditions. A discussion involving other macrocyclic compounds will be presented in a further publication (23). The residual water seems to have no effect on UOZ2+hydration in PC. Measurements by POtentiometry and spectrophotometry give the same stability constants of the complexes for different concentrations of residual water. This result seems to show a low and negligible hydration of U02+ in PC vs. the solvation by PC. Similarly, no H+ effect on the stability constants was observed. Determinations of a stability constant at two different pH values (around 6.5 and 5.5) give the same result taking into account the deviations corresponding to a 96% confidence interval. This result is confirmed by the study of Kolthoff et al. (24) on the complexation of H+, (and H30+)with 1842-6 in PC. The logarithm of the formation constant of H8-18-C-6(6.3) given by these authors indicates that, in our experimental conditions of pH and [H,O], no significant formation of this
Anal. Chem. 1984, 56,163-168
complex can be observed (less than 1% of H,+is complexed by the crown ether). In order to avoid an eventual competition between H,+ (or H,O+) and the studied metal (Na+, UOz2+, etc...), the concentration of the studied metal has to be chosen as high as possible. Registry No. Na, 7440-23-5;propylene carbonate, 108-32-7.
LITERATURE CITED . .
Ser. C 1978, 287, 105-108. (2) McClure, J. E.; Reddy, T. B. Anal. Chem. 1988, 40, 2064-2066. (3) Lakshminarayanalah, N. ”Membrane Electrodes”; Academic Press: New York, 1976: pp 50-94. (4) Rechnitz, G. A. Chem. f n g . News 1967, 45 (25),146-158. (5) Eisenman, G. I n “The Glass Electrode”; Wlley: New York, 1965;pp 213-369. (6) Buck, R. P. Anal. Chlm. Acta 1974, 7 3 , 321-328. (7) Buck, R. P.; Boles, J. H.; Porter, R. D.;Margolls, J. A. Anal. Chem. 1074, 46, 255-261. (8) Romberg, E.; Cruse, K. 2.Nektrochem. 1959, 6 3 , 404-418. (9) Coetzee, J. F.; Padmanabhan, G. R. J . Phys. Chem. 1962. 66, 1708- 1713. (IO) Teze, M.; Schaal, R. Bull. Soc. Chlm. Fr. 1982, 7 , 1372-1379.
163
(11) Norberg, K. Talanta 1966, 13, 745-752. (12) Izutzu, K.; Kolthoff, I . M.; Fujlnaga, T.; Hattori, M.; Chantooni, M. K. Anal. Chem. 1977, 49, 503-508. (13) L’Her, M.; Courtot-Coupez, J. J . Nectroanal. Chem. 1973, 4 8 , 265-275. (14) Talarmin, J.; L’Her, M.; Laquenan, A.; Courtot-Coupez, J. J . Nectroanal. Chem. 1980, 106, 347-358. (15) Kolthoff, I. M.; Harris, W. E. J . Am. Chem. Soc. 1945, 6 7 ,
1484-1491. (16) Gosse, B.; Denat, A. J . flectroanal. Chem. Interfacial Nectrochem. 1974, 56, 129-147. (17) Spiess. B. Thesis, Strasbourg, France, 1981. (18) Brown, A. S. J . Am. Chem. SOC. 1934, 5 6 , 846-647. (19) Talarmln, J.; L’Her, M.; Courtot-Coupez, J. J . Chem. Res. 1977, 5 ,
28-29. (20) Sabatlni, A.; Vacca. A.; Gans, P. Talanta 1974, 2 1 , 53-77. (21) Talarmln, J.; L’Her, M.; Laquenan, A.; Courtot-Coupez, J. J . flectroanal. Chem. 1979, 103, 203-216. (22) Ingri, N.; Kakolowlcz, W.; Sillen, L. G.; Warnqvist, 8. Talanta 1967. 14, 1261-1286. (23) Fux, P.; Lagrange, J.; Lagrange, P., unpublished results. (24) Kolthoff, I.M.; Wang, W. J.; Chantoonl, M. K., Jr. Anal. Chem. 1063, 5 5 , 1202-1204.
RECEIV~, for review July 11,1983. Accepted October 11,1983.
Chloride Interferences in Graphite Furnace Atomic Absorption Spectrometry Walter Slavin,* G. R. Carnrick, and D. C. Manning
Perkin-Elmer Corporation, Main Avenue, Norwalk, Connecticut 06856
The premature loss of analyte as a volatile halide Is shown to be an important interference In m e sttuations. When this char step interference is avolded, an interference In the atomization step can arise when large amounts of halide are present during that step. This atomizatlon interference results from the binding of the anaiyte as a vapor phase metal halide, thereby preventing some portlon of the analyte from absorbing atomic radiatlon. This vapor phase interference can be circumvented in several ways. The most convenient Is the use of high concentrations of HNO, or H2S0, which drlves off more of the chloride as HCI durlng the dry and char steps. As much CaCi, In solution as 1 mg reduced the Mn slgnal In the presence of HNO, and Mg(N03), by only 15%. Analogous experience was found wlth TI. We believe the experience is qulte general. The halide which interferes in the vapor phase is retained from the char step by adsorption or intercalation on the graphite tube. Thus we can expect stili further improvements in the handling of large amounts of hailde matrix when glassy carbon or solid pyrolytic tubes become available.
L’vov’s original publication on electrothermal atomic absorption spectrometry (1) included a theoretical and experimental justification of the “absolute” nature of the technique. All of the analyte was converted to an atomic vapor and, if the experimental conditions were correctly chosen, the expected signal could be calculated. This implied independence of the signal from the nature and amount of the matrix. The instrumental design changes which deviated from those proposed by L’vov led to the extensive literature of interferences over the 25 years since his first publication. However, modern work with the various systems that are often described as “constant temperature furnaces” (2-7) and which use the L’vov
conditions have shown that, within limits, determinations by furnace AAS are indeed independent of the matrix, as L’vov predicted. Our version of these modern furnace systems, called the stabilized temperature platform furnace, STPF, has been described (7) and we have shown furnace analyses within 10 to 20% without using standards (8). This paper explores some of the limits to the situations to which L’vov’s theory can be applied using the STPF technology. Metal halide interferences are, beyond doubt, the most widely reported interferences, and the nature and control of these interferences are gradually becoming clearer. While we use chloride for the experimental work in this paper, the other halogens provide similar results. Alkali and alkaline-earth chlorides have provided the most serious interferences. These interferences apply to most analyte metals. The interferences depend more upon the chloride that is present than upon the metal that is associated with the chloride, but there is influence from both. Important goals of the constant temperature furnace of Woodriff (2)and of the improvements in furnace technology by many other workers (3-7) have been directed toward understanding and controlling these alkali and alkaline-earth chloride interferences. By now we have controlled this metal halide interference on a large variety of analytes and an equally large variety of metal halide matrices. However, when the amount of metal halide matrix is large enough, we and others with improved furnace systems find that an interference arises. Typically, this amount of chloride is on the order of 200 pg of salt on the platform which, in the 2 0 - ~ Lsample, is about 1% salt in the solution. This 200 pg of metal chloride is not a sharp cutoff, larger amounts can be handled for some analytes and for some metal chlorides. In this paper we have addressed some of the sources of these problems and their control. While we have a better under-
0003-2700/84/0356-0163$01.50/00 1984 American Chemical Society