The behaviour of two types of copper ion-selective electrodes in different copper(II)-ligand systems

0039-9140/85g3.00+ 0.00 Copyrzght 0 1985Pergaznon Press Ltd

Talanta, Vol. 32, No. 10, PP 937-947, 1985 Printed in Great Bntain. All rights reserved

THE BEHAVIOUR OF TWO TYPES OF COPPER ION-SELECTIVE ELECTRODES IN DIFFERENT COPPER(LIGAND SYSTEMS M. NE~HKOVA and H. SHEYTANOV Institute of General and Inorganic Chemistry, Bulgarian Academy of Sciences, 1040 Sofia, Bulgaria (Recezued 13 June 1984. Revised 3 May 1985. Accepted

10 May 1985)

Summary-The behaviour of two types of solid-state homogeneous sensors for copper(B), one based on pressed pellets of ternary CuAgSe and the other on thin-layer electroplated Cu,_,Se, in 12 different copper(ligand systems, has been thoroughly investigated. Both electrodes exhibit anomalous behaviour when the ligands are of complexone type, the effect of the complexones on the deviations increasing in the order IDA < NTA < EDTA u DTPA, and being practically the same for the two types of sensors, thus disproving a previous suggestion that the anomaly is due to the silver in the silver-containing sensors. The experimental data do not support the specific ligand-adsorption hypothesis either. The observed deviations are tentatively explained on the basis that, as suggested by the selectivity coefficients, both sensors act as primary copper(I) ion-selective electrodes rather than copper(II)-electrodes. Thus, at very low copper(I1) concentrations, according to the extended Nikolskii equation, the [Cu(I)]/[Cu(II)] ratio at the electrode surface determines the electrode sensitivity towards Cu(I1). The lower detection limit could be improved by pH-control and selective complexation of Cu(1). This hypothesis has been proved experimentally. If the copper(I) activity on the electrode surface is decreased, the anomaly observed for the Cu(IIkNTA system disappears and decreases considerably for the Cu(II)-EDTA and Cu(II)-DTPA

Soon after the first ion-selective electrodes for heavy metal ions were developed, it was realized that they offer excellent possibilities for direct determination of stability constants, as well as for objective end-point location in complexometric titrations.’ It is now generally known that the measurement of activities lower than 10e6M is possible only by use of buffers to attain these low values reliably, and this requires knowledge of the behaviour of these electrodes in different metal-ligand systems. Hansen et al.* were the first to consider the fact that several commercially available copper(I1) ion-selective electrodes of the mixed chalcogenide type display anomalous behaviour in the presence of EDTA. Later, other author&’ also commented on this fact. Different reasons have been proposed for the observed deviations but no consensus has yet been reached. A definitive answer can only be obtained after comprehensive studies of the behaviour of membrane materials of different composition in a broad range of copper(ligand systems which have been studied by independent methods. The present paper offers a systematic investigation of two solid homogeneous copper(H) ion-selective electrodes of different chemical composition, one based on pressed ternary CuAgSe and the other on a thin membrane of electro-deposited Cu,_,Se, in different Cu(II)-ligand systems, the ligands being carboxylic acids, amino-acids and polyaminopolycarboxic acids (complexones). The complexing agents were carefully chosen to enhance the chances of finding specific ligand effects. 937

EXPERIMENTAL Radelkis OP-Cu-7 113 ion-selective electrodes, with homogeneous CuAgSe membranes,’ the performances of which have been reported,9 and plated thin-layer Cu,,Se membranes prepared as described earlier,‘” were selected for the studies. The following complexing agents were examined: tartaric, salicylic and sulphosalicylic acids, alanine, methionine, lysine, aspartic and glutamic acids, iminodiacetic acid (IDA), nitrilotriacetic acid (NTA), EDTA and DTFA, all of analytical grade and not further purified. The emf was measured with a digital Radelkis OP-208 pH-meter at 20 or 25” (f 0.2”). depending on the temperature at which the corresponding siabilityconstants had been determined, us. a Radelkis OP-830 double-iunction Ag/AgCl reference electrode (1M potassium nitraie in the bridge). The measurements were made on solutions with a total Cu(I1) concentration of 1 x lo-‘M and at least a IO-fold excess of ligand. A thermostatically controlled potentiometric cell was filled with approximately 100 ml of the copper(ligand solution at a constant ionic strength of 0.1 (KNO,). The electrode-pair was immersed in the solution, together with a combined glass electrode. The pH was changed gradually by adding dilute sodium hydroxide solution or nitric acid, the corresponding steady-state potential of the ion-selective electrode being measured at each pH value. A value which did not alter by more than 0.4 mV within 10 min was accepted as the steady-state potential. The stirring rate was kept constant. Three independent sets of measurements were made for each system with each type of electrode, in an effort to determine the reproducibility. For all systems except those containing complexones, the reproducibility was + 1.5 mV or better and the steady-state potential was reached very rapidly as the pH was changed. In the Cu(II+complexone solutions the steady-state potential was reached after different periods, depending on the pH, and more quickly with copper selenide electrodes than with CuAgSe electrodes, but with a rather poor reproducibility

M. NE~HKOVAand H. SHEYTANOV

938

of + 5 mV. Before and immediately after each measurement, the copper ion-selective electrode was calibrated with standard copper solutions of the same ionic strength as the test solution. The relationship pCu/mV was extrapolated to the lowest measured potential, by using the slope of the calibration graph. The concentration of free copper(H) was calculated from the equation: [Cd + ] = [Cd + I,,,

+/&NT+. ..p [Ll” atcHj

(1)

“abH, >

where p,-/?, are the stability constants for the CuL-CuL, complexes, [L] is the total ligand concentration, and aLcH) is the side-reaction coefficient for protonation of the ligand, calculated from aL,“) =

1+ [H + IKE, + [H + ]ZKLK&L +

. [H +l”K;,K:,, . . . KBL (2)

Table 1 summarizes the values used for K& and B,. The use of the total ligand concentration in equation (1) is an approximation justified by [L] >>[Cu2+ I,,,. Corresponding single measurements were made for

copper-free solutions of the ligand at the same pH values and ionic strength as for the copper solutions.

RESULTS AND

DISCUSSION

The pCu(II)-values experimentally determined with the two types of sensor, and those calculated according to equation (1) for each system, are presented in Figs. 1-12. Use of only a twofold excess of ligand does not change the character of the experimental curves. The following experimentally established facts should be noted. (1) A serious discrepancy between the experimentally determined and theoretically calculated pCu(I1) values is observed with both types of membrane only when the ligands are complexones, and it increases with change of ligand in the order IDA < NTA < EDTA = DTPA. The theoretical relationship is followed up to pH = 8 for IDA (Fig. 9).

(2) Regardless of the chemical composition of the membranes used (i.e., with or without silver), the deviations follow approximately the same pattern. All the experimental values are several orders of magnitude higher than those theoretically predicted. (3) The pCu/mV relationships for the other ligands can be divided into two regions: up to pH 8-9 there is good agreement between the theoretical and experimental values, but at higher pH some of the systems show slight discrepancies, again with experimental values higher than those calculated. (4) An experimental straight line with a slope of 30mV/pH was found for the Cu(II)-tartaric acid system at pH > 5, instead of the plateau expected from the stability constants. This suggests a mixed hydroxide-tartrate complex, which does not seem to have been reported. Investigation of this equilibrium is in progress.” (5) The mV/pH relationship for the pure ligands follows closely, or even coincides with, that for the copper(IIkligand system, when NTA, EDTA and DTPA are used as ligands. All these experimental facts can hardly be explained as due to dissolution of the membrane-the thin membranes would tend to dissolve immediately, whereas we were able to use them for several months, exposing them daily to EDTA or DTPA solutions. The presumed role played by the silver in the membrane 3,4,6should obviously be rejected. The distribution df the ligand species as a function of pH (note the distribution diagrams in the upper right-hand corner of each figure) is not in accord with the nature of the anomalous behaviour, and gives no support to the hypothesis advanced that there is ligand adsorption, increasing with the charge of the ligand species.4s5For example, it is rather difficult to explain on this basis the fact that for salicylic, sulphosalicylic and tartaric acids (Figs. 6-8) excellent agreement between experimental and calculated data was observed even when a large excess of negatively charged ligand species was present in the solution, whereas in the case of NTA, EDTA and DTPA (Figs. 10-12) the

Table 1 Complexing agent Alanine Glutamic acid Aspartic acid Methionine Lysine Glycine Tartaric acid Salicylic acid Sulphosalicylic acid IDA EDTA NTA DTPA *The corresponding

9.69 9.2 9.62 9.20 10.69 9.57 4.10 13.4 11.80 9.38 10.34 9.81 10.56

11.99 13.15 13.32 11.37 19.77 11.93 7.0 16.21 14.29 12.03 16.58 12.38 19.25

15.35 15.26 21.81

19.33 21.40 14.35 23.62 26.49 28.4

values for /?EiL and fi&&,, are given.

8.13 7.87 8.4 8.1 7.56* 8.15 3.2 10.62 9.43 10.5 18.8 12.7 20.5

14.92 14.16 15.20 14.80 14.02* 15.03 5.1 18.45 16.30 16.2 16.3

5.7

6.5

22 22 24 24 22 23 23 22 22 23 23 23 23

Copper ion-selective electrodes in copper(ligand

4

2

4

6

0

systems

10

939

12

nsc.cn.coon

I

N”z

6

8 5 0

10

12

14

I

2

1

4

1

1

6

6

1

I

10

12

PH

for 1 x lo-‘A4 copper nitrate + 2 x IO-‘A4 alanine with the CuAgSe membrane (0) and Cu,_,Se membrane (A). The curve drawn is calculated after equation (1). ( x ) values Fig. 1. pCu-pH

relationship

for the pure ligand.

observed deviations started at low pH, at which ions of high negative charge were not present. Nor does ligand adsorption explain why both electrodes measure pCu(I1) values of 10 or 12 accurately in the alkaline and neutral pH region, but cannot measure the same values in the acidic region. Ligand species of higher negative charge are predominant in the far alkaline region for almost all the systems investigated, a region where in fact certain deviations were observed, but these cannot be related unambiguously to specific ligand adsorption by the membrane. Both electrodes displayed ideal reproducibility of the potential for copper standards immediately after measurements in the corresponding Cu(II)-ligand system, without application of special treatment. Any conclusions based on the observations in this pH region appear to be premature and ambiguous, as other equilibria may also exist in such alkaline solutions. Hansen et a/.* and later Hulanicki’ have called attention to the fact that the effect of any Cu(1) in the composition of the membrane should also be consid-

ered as a factor which may cause the observed anomalous behaviour, and therefore deserves a comprehensive investigation. We have assumed that a general cause for the anomalies we have observed is related to the fact that the two types of electrode are sensitive both to copper(I) and copper(I1) and that their selectivity towards copper(I) is substantially the higher.‘* Consequently, their behaviour is better described by the extended Nikolskii equation:

RT E =E" +z In{ac,+ +G+.CU2+(~Cu2+)1/2} (3) where log K!%‘+cu2+ is equal to - 5.7 and -6.2 for the copper selenide and CuAgSe sensors respectively. It should be remembered that in equation (3), k,+ and a,-,,+ are the activities at the membrane surface. Consequently, when we measure the activity of Cu(II), the activity of Cu(I) generated by the membrane material may be neglected (as a factor which affects the potential) only in the case of relatively higher copper(I1) activities (above lo-‘M).

30

I

4

I

2

I

6

PH

1

8

I

10

0

1

0

12

0

Fig. 2. pCu-pH relationship for 1 x IO-‘M copper nitrate + 1 x IO-*A4 glutamic acid. Symbols as for Fig. 1.

1

1

11

1

,

0

a

1 2

I

6

I

4

PH

6

I

I

10

HOOC.lCH2)2

CHCOOH

I

12

Fig. 3. pCu-pH relationship for 1 x 10m3M copper nitrate + 1 x lo-*M aspartic acid. Symbols as for Fig. 1.

l?

11

9

7

x

z

13

11

9

7

5

3

1

1 4

1

6

PH

I

6

1

10

1

12

Fig. 4. pCu-pH relationship for 1 x lo-)A4 copper nitrate + 1 x IO-*A4 methionine. Symbols as for Fig. 1.

2

0

a

13

11

9

7

s

3

1

6 PH

1

6

1

10

NH2

I

12

I

CH COOH

Fig. 5. pCu-pH relationship for 1 x IO-‘M copper nitrate + 1 x lo-*M lysine, with the CuAgSe membrane (0) and Cu,_,Se membrane (A).

I

4

I

2

I

H2C.CH2.CH2 CH,

942

M. NSHKOVA

and H.

SHEYTANOV

: a

2

I

\

0

0

0

4

1

0

^_

“.Z

1.0

6

1

h

‘.J 2

I

I 1 4

?“I

‘-* *\

t

PH

6

I

6

_

‘\

-

‘\

\

10

1

ha \ \

6

1

o\ \

0,

9

10

1

\

- .___.

1*

L2_

_._.

COOH

PH

\

“t;.’

HOOc. CH.CH

H,L

-

12

1

-

Fig. 8. pCu-pH relationship for 1 x 10m3M copper nitrate + 5 x lo-*M tartaric acid. Symbols as for Fig. 5.

12

10

6

6

4

2

s m

Fig.

15

13

11

9

7

5

3

4

6

9

10

9. pCu-pH relationship in 1 x 10m3M copper nitrate + 1 x IO-*M IDA. Symbols as for Fig. 5.

2

IDA

E

944

and H. SHEYTANOV

M. NESHKOVA

0 0 0 0

l

4 4

l 4

l

q

Copper

ion-selective

electrodes

in copper(ligand

945

systems

7-

0

0

24

l

6a

B

6-

10

12

PH DTPA

9-

10 -

; 0 11 -

12 -

13 -

14 -

PH

Fig. 12. pCu-pH relationship for membrane (0, 0, +) and C&Se

1 x 10-3M membrane

copper nitrate+ 1 x 10m2M DTPA with the CuAgSe (A, A); (0, A) for the pure ligand solution, (+) in presence.of O.lM glycine.

It is quite reasonable to presumer3 that the real activity of copper(I) at the electrode surface should be much higher than that calculated from the solubility data for the membrane material, and hence that the following equation for the surface copper(I) concentration will be valid: Ku + lo=

Ku +lo(K,pJ + Ku +Iowan)

(4)

where the term [Cu +]O(exlran ) depends on other factors than those related to the thermodynamic solubility Ku +lo&,, of the membrane. We think that these factors are intimately related to the defect structure and semiconductor character of the membrane materials.14 The thermodynamic solubility contribution, [CU+]~~,), for the copper selenide sensor will be negligible, as calculated from KsP~cU2se~.‘s It varies from 2 x lo-“M at pH = 2 to 5.7 x 10m2’M at pH = 8.

TAL

32116-B

Data for the solubility of CuAgSe are lacking, but it can be expected to be of the same order or lower. If we consider that the calibration limit for copper(I) buffer with both electrodes isI pCu(1) = 12, we have to accept a rather high concentration for [Cu + lwextran ) in equation (4), i.e., [Cu + lqextran )>>[Cu + lwKVJ. The experimental data strongly suggest that the Cu(1) activity at the electrode surface is pHdependent. It decreases as the pH increases. If all the above-mentioned circumstances are taken into consideration, equation (3) can be rewritten as Ku + lO(extran )

E=E”+~ln 1

Q”+(L)

(gy}

+KE+.Cu2+

(5)

M.

946

NESHKOVA

and H.

for the general case. It follows that the electrodes should display anomalous behaviour when becomes commensurable with, or [Cu + IO~extran ,/kh +(I_) greater than, the second term in the bracket in equation (5). Thus equation (5) defines the “apparent” electrode sensitivity towards copper(I1). Obviously all factors leading to decrease in the copper(I) activity at the electrode will increase the relative apparent sensitivity of the electrode with respect to copper(U). On the basis of these assumptions, the observed experimental results can be interpreted as follows. (1) Fairly good agreement between the measured and calculated pCu(II) values is observed for all systems over the pH range for which the equilibrium copper(H) concentration at any pH meets the condition

[Cd +] I’* K

( > NC”2+(L)

cu+,cw+

2

W’l qextm) %u+(L)

This requirement is easily met by complexes which exhibit only moderate stability in acidic medium. (2) The deviations observed for the copper-complexone systems are due mainly to the high stability of the copper complexes even in acidic medium. It becomes more and more difficult to meet the above-stated requirement for a proper Nernstian response to Cu(I1) as the stability of the Cu(II) complexes increases. That is why the extent of the observed deviations is in the same order as the stability of the complexes in acidic medium: Cu-IDA < Cu-NTA < Cu-EDTA - Cu-DTPA. (3) It appears that the “apparent” electrode sensitivity towards Cu(I1) will be different for each system since it will depend not only on the pH but also on the ability of the ligand to bind copper(I) in a complex too. Indeed, when the ligand also binds copper(I), the deviations are absent or start at higher pCu(I1) in the alkaline region, e.g., in the case of glycine, alanine, methionine and lysine. (4) The electrode response in pure ligand solutions is a practical measure of the change in surface copper(I) concentration as a function of pH and complexation processes. The experimental curves in pure complexone solutions outline in fact the pCu(II)/pH region which can be used successfully in a particular Cu(II)-L system. Thus for the Cu-EDTA and Cu-DTPA systems the lowest copper(H) activity which can actually be measured varies from 8 to 12 as the pH is changed from 4 to 8. It follows, on this hypothesis, that if an appropriate complexing agent for copper(I) which does not disturb the bulk equilibrium is introduced into the copper(II)-complexone solution, we should be able to decrease the discrepancy observed between the experimental and theoretical results, to an extent governed by the alteration in Cu(I)/Cu(II) ratio at the electrode surface. We used glycine, which forms

SHEYTANOV

complexes with copper(I) (log /I2 = 10.1) and does not affect the Cu(II)-EDTA, Cu(II)-NTA and Cu(II)-DTPA equilibria when present in concentrations up to O.lM at pH 8. The effect of addition of glycine to these systems was investigated over the concentration range 10~3-10-‘M. Figures l&12 present some of the results obtained. As seen from the figures, the observed changes are in the predicted direction and also depend on the concentration of glycine. The expected decrease in the copper(I) activity at the electrode surface is adequate to eliminate the anomaly in the Cu(II)-NTA system, but is not sufficient to eliminate it in the other systems. A logical question which arises is what is the limit of the relative sensitivity of the electrodes towards copper(I1) which can be practically achieved. An attempt was made to reach this limit, by studying the effect of other ligands for copper(I) when introduced into the Cu(II)-EDTA system. The choice was restricted because of the requirement that the added ligand must not affect the Cu(II)-EDTA equilibrium. Thus, the effect of potassium chloride and potassium thiocyanate was investigated. An interesting phenomenon was observed in presence of 2M KC1 and 0.1 M KCNS. At all pH values >3, the membranes get “blocked”, i.e., they can no longer respond to change in the Cu(I1) concentration and show an approximately constant and stable potential. corresponding to about pCu(I1) = 12. When rinsed with water and introduced into copper(I1) standards, the electrodes instantly regain their normal response. Lower concentrations of these ligands do not cause “blocking”, but the response is heavily distorted, probably because the anomalous response to Cu(I1) in presence of halides (or pseudohalides) is superimposed on it. The model for reduction of Cu(II) in the presence of chloride’6,‘7 seems very likely to be applicable in these cases too. No clear-cut explanation for the “blocking” can yet be presented, but the phenomenon does suggest that copper(I) plays a role in the potentialgenerating processes. The assumption that the higher selectivity of the two electrodes towards copper(I) is the major reason for the observed anomalous behaviour can be further extended to other membranes of chalcogenide type. Moreover, it was recently established that even when the initial material is a CuS/Ag,S mixture. a new phase containing copper(I) appears after pressing.” This offers an explanation of the fact that membranes differing in composition show similar anomalous behaviour in presence of EDTA, the differences being only in the extent of the deviations, which can be explained by a different copper(I) activity at the electrode surface, characteristic for each electrode material. The differences in the electrode materials in this respect can be simply established by comparative titration of copper with EDTA and comparison of the experimental titration curves with the theoretical ones.

Copper ion-selective electrodes in copper(II~ligand systems CONCLUSIONS These investigations by no means provide grounds for pessimistic conclusions about the applicability of the different ion-selective electrodes for new equilibrium studies, but rather confirm that comprehensive understanding of the variable relative sensitivity limit towards Cu(I1) and of the factors which affect it, can prevent erroneous interpretation of the experimental data, an example of which is a value proposed for the stability of the Cu(II)-NTA complex.20 The results offer a possibility to formulate a practical criterion for the pCu(IItpH region within which investigations must be confined. We recommend that before an investigation is attempted, the mV vs. pH relationship for the pure ligand and the Cu(I1) + L systems be traced. The region where the two experimental curves are separated defines the values of pCu(I1) which can be correctly measured. The success achieved by Baumann” in confirming the stability constants for Cu-EDTA and Cu-DTPA complexes by using the Orion copper ion-selective electrode, is due to the Cu-ligand ratio and pH being chosen so that the electrode measures pCu(I1) values above its apparent limit of sensitivity. This study strongly suggests that the potentialgeneration mechanism is identical for all chalcogenide copper electrodes, the role of copper(I) and the defect structure of the membranes being factors which undoubtedly deserve attention and deeper investigation. A further theoretical exploration of these systems was presented at Euroanalysis V in Krakow, August 1984, and will be submitted to Mikrochimica Acta for publication.

947

REFERENCES 1. R. P. Buck, J. C. Thompsen and 0. R. Melroy, in ISEs

in Analytical Chemistry, Vol. 2, H. Freiser (ed.), p. 208. Plenum Press, New York, 1980. 2. E. Hansen, C. G. Lamm and J. RbiiEka, Anal. Chins. Acta, 1972, 59, 403. 3. G. Nakagawa, H. Wada and T. Hayakawa, Bull. Chem. Sot. Japan, 1975, 48, 424. 4. Y. Umezawa, Y. Imanishi, K. Sawatari and S. Fujiwara, ibid., 1979, 52, 945. 5. W. E. van der Linden and G. J. M. Heijne, Anal. Chim. Acta, 1978, 96, 13. 6. G. Nakagawa, H. Wada and T. Sako, Bull. Chem. Sot. Japan, 1980, 53, 1303. 7. A. Hulanicki, M. Trojanowicz and M. Cichy, Talunta, 1976, 23, 47. 8. M. Neshkova and J. Havas, Hung. Patent, 174627, 31 August 1976. 9. Idem, Anal. Lett., 1983, 16, 1567. 10. M. Neshkova and H. Sheytanov, J. Electroanal. Chem., 1979, 102, 189. 11. Idem, investigation in progress. 12. I&m, Talanta, 1985, 32, 654. 13. W. E. Morf, The Principles of Ion-Selective Electrodes and of Membrane Transport, Akademiai Kiado, Budapest, 1981. 14. W. E. Morf, G. Kahr and W. Simon, Anal. Chem., 1974, 46, 1538. 15. R. M. Smith and A. E. Martell Critical Stability Constunts, Vol. 4, Plenum Press, New York, 1976.. 16. J. C. Westall. F. M. M. Morel and D. N. Hume. Anal. Chem., 1979,. 51, 1792. 17. T. Hepel, Anal. Chim. Acta, 1981, 123, 151. 18. G. J. M. Heiine and W. E. van der Linden. ibid.. 1977. 93, 99. 19. H. Stiinzi, Talanta, 1982, 29, 75. 20. S. Ramamoorthy, C. Guamaschelli and D. Fecchio, J. Inorg. Nucl. Chem., 1972, 34, 1651. 21. E. W. Baumann, ibid., 1974, 36, 1827. 22. A. E. Martell and R. M. Smith, Critical Stability Constants, Vol. 1, Plenum Press, New York, 1974. 23. A. Ringbom, Complexation in Analytical Chemistry, Interscience, New York, 1963. 24. L. G. Sillen and A. E. Martell, Stability Constants, Snecial Pub]. No. 25. The Chemical Societv. London. I

,