Study on partition equilibria of metal complexes in non-ionic micellar solutions from spectrophotometric data

Talanta 52 (2000) 225 – 232 www.elsevier.com/locate/talanta

Study on partition equilibria of metal complexes in non-ionic micellar solutions from spectrophotometric data R. Codony, M.D. Prat, J.L. Beltra´n * Departament de Quı´mica Analı´tica, Uni6ersitat de Barcelona, A6inguda Diagonal 647, 08028 Barcelona, Spain Received 19 October 1999; received in revised form 28 January 2000; accepted 4 February 2000

Abstract The complexation equilibria for Zn(II)–8-quinolinol and Zn(II) – 5,7-dichloro-2-methyl-8-quinolinol systems were studied spectrophotometrically in aqueous micellar solutions of the non-ionic surfactant Brij-35 in NaCl 0.1 M medium at 25°C. The partition model, in which the different species involved in the equilibria can distribute themselves between aqueous and micellar pseudophases, was applied. Calculations were performed by means of the SPDIS program, developed specifically to handle multiwavelength spectrophotometric data in micellar systems. A factor analysis was applied to the spectrophotometric data in order to determine the number of species in equilibrium. A quantitative relationship was found between fluorescence intensity and the micellar solubilization of metal chelates. © 2000 Elsevier Science B.V. All rights reserved. Keywords: Non-ionic micellar media; Partition equilibria; Zn(II) – 8-quinolinol complexes; Spectrophotometry

1. Introduction The ability of micellar systems to solubilize a wide variety of compounds which are insoluble or sparingly soluble in water is one of their main properties. As a consequence of incorporation in a micellar structure, the chemical equilibria, the reactivity as well as spectral and electrochemical responses of the substrates can be drastically altered. These properties have proved to be very useful in many areas of analytical chemistry, such as separation techniques or determinations based * Corresponding author. Tel.: +34-93-4021796. E-mail address: [email protected] (J.L. Beltra´n)

on spectral methods [1,2]. Many of the analytical applications using micellar solutions involve determination of metal ions via their complexation with suitable ligands [2]. In the case of uncharged metal complexes, solubilization in micellar solutions can be described in terms of distribution equilibria between the micelles and the aqueous phase, in which the micellar phase plays the same role as the organic solvent in liquid–liquid extraction [3–6]. However, despite the numerous successful applications of micellar media in the separation and analysis of metal ions, few quantitative data have been reported on the micellar solubilization equilibria of metal complexes compared with the num-

0039-9140/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 3 9 - 9 1 4 0 ( 0 0 ) 0 0 3 1 8 - 0

226

R. Codony et al. / Talanta 52 (2000) 225–232

ber of publications concerned with partition constants of neutral organic solutes. Different techniques have been used to estimate the equilibrium constants involved in these systems, including micellar liquid chromatography (MLC) [7], micellar electrokinetic capillary chromatography (MECC) [8,9], spectrophotometry [4,5], or by means of phase separation of micellar solutions at the cloud point [3,6]. However, these methods suffer from a number of certain drawbacks. Thus, MLC and MECC, in addition to providing inaccurate determinations when applied to very hydrophobic solutes, can only be used for inert complexes. The cloud point separation method can only be applied to a limited number of surfactants (non-ionic and with low cloud point). Spectrophotometric methods, which are widely used in equilibria studies, have also been applied to determine partition constants of metal chelates in micellar media. However, to date, only single wavelength spectrophotometric data have been used in these systems. In a previous paper [10], a program (SPDIS) was developed for the study of partition equilibria in micellar solutions. It was designed to handle multiwavelength spectrophotometric data, and was applied successfully to the distribution of 5,7-dichloro-2-methyl-8-quinolinol (HClQ: chlorquinaldol) in non-ionic [10] and ionic micelles [11]. In this paper, we extend the model to complex formation and extraction equilibria with metal ions. Micellar pseudo-extraction equilibria data of Zn(II)–oxine and Zn(II) – chlorquinaldol systems are reported. Because protonation and partitioning of the ligands between the micellar and the aqueous phases have to be taken into account, equilibria constants of 8-quinolinol (HQ: oxine) are also determined. Finally, the fluorescence of zinc chelates in Brij-35 micellar media are investigated. The results obtained show that there is a quantitative relationship between observed fluorescence and calculated metal chelate extracted into the micellar phase, thus confirming the validity of the model.

2. Experimental

2.1. Apparatus The absorption spectra were recorded on a Beckman DU-7 single-beam spectrophotometer, using a 1.00 cm quartz cuvette. The spectrophotometer was connected to an IBM-PC via a serial interface. The absorbance data were acquired using the DUMOD program [12]. Fluorescence measurements were recorded with a Perkin–Elmer LS-50 spectrofluorimeter equipped with a xenon lamp. The slit widths were set to 5-nm in both the excitation and emission monochromators. Fluorimetric measurements were carried out at lex = 383 nm and lem =542 nm for the Zn–oxine complex and lex = 396 nm and lem = 540 nm for the Zn–chlorquinaldol complex. The pH values were measured with a Radiometer PHM 84 pH meter, equipped with an Orion 81-02 combined glass electrode. Calibration of the electrode system was undertaken with buffer solutions at pH 4.008, 6.863 and 9.180 prepared from Merck salts according to DIN 19266. All spectrophotometric and potentiometric measurements were recorded at 259 0.2°C.

2.2. Reagents 5,7-Dichloro-2-methyl-8-quinolinol (Supro, Troponwerke) was recrystallized twice from ethanol solution, and 8-quinolinol (Fluka) was used without further purification. Working solutions were prepared daily by dissolving the reagent in 0.02 M hydrochloric acid. Brij-35 (Fluka) was used without further purification. A standard solution of Zn(II) (0.02 M) was prepared from analytical reagent-grade zinc nitrate (Merck) and was standardized volumetrically with EDTA. Buffer solutions were prepared according to Perrin and Dempsey [13] from hydrochloric acid, acetic acid, succinic acid, sodium dihydrogenphosphate, tris(hydroxymethyl)-aminomethane, boric acid, borax and sodium hydroxide. The ionic strength of the working solution was kept con-

R. Codony et al. / Talanta 52 (2000) 225–232

stant at 0.1 M by addition of sodium chloride. Doubly distilled water was used for spectrophotometric measurements and doubly deionized water (Culligan Ultrapure GS; 18.2 V cm − 1 resistivity) for fluorimetric measurements.

2.3. Procedures Aqueous solutions containing chlorquinaldol or oxine, Zn(II) ions (when required), surfactant, buffer solutions and sodium chloride to adjust the ionic strength to 0.1 M were prepared. The pH was varied by the addition of buffer solutions. The absorption spectra were recorded at 5 nm increments from 220 to 450 nm, and the pH values of the solutions were measured and converted into hydrogen ion concentration, according to the Davies equation. The activity coefficient calculated for the hydrogen ion was 0.771 at 25°C and 0.1 M ionic strength.

2.4. Data treatment The pseudo-extraction equilibria were determined using the program SPDIS [10], which optimizes the equilibrium constants in micellar media, and calculates the molar absorptivities of the different species from multiwavelength data. The program follows a Gauss – Newton iteration procedure in order to minimize the sum of squared residuals (U), defined as: Table 1 Experimental and calculated apparent dissociation constants of 8-quinolinol at different concentrations of Brij-35 (in 0.1 M NaCl at 25°C) CBrij-35 (M)

pK*a1exp.a

pK*a1calc.b pK*a2exp.a

– 5.00×10−3 1.00×10−2 2.00×10−2

4.91 4.76 4.62 4.47

4.75 4.63 4.46

(0.02) (0.02) (0.02) (0.02)

9.63 9.78 9.76 9.93

(0.01) (0.02) (0.02) (0.02)

pK*a2calc.b

9.75 9.83 9.93

a The values in parentheses correspond to three times the standard deviation of pKa given by the program STAR. b Calculated from pKa values in aqueous medium and the distribution constants given in Table 5.

ns

227

nw

U= % % (Ai, j,exp − Ai, j,calc)2 i=1 j=1

where ns indicates the number of solutions and nw is the number of wavelengths (readings for each spectrum). The values Ai,j,exp and Ai,j,calc correspond, respectively, to the measured absorbance of the j-wavelength in the i-solution, and that estimated by the program for a given equilibrium model. In addition, we used the STAR program [14] for the determination of dissociation constants of 8quinolinol (in water and micellar media).

3. Results and discussion

3.1. Apparent dissociation constants of 8 -quinolinol In order to study the effect of the non-ionic surfactant Brij-35 in apparent acid–base equilibria of 8-quinolinol, the absorption spectra of several series of aqueous micellar solutions were recorded. In each series the surfactant and ligand concentrations were kept constant and pH was varied from 2 to 11. Initial experiments performed at surfactant concentrations in the range 1× 10 − 4 –5× 10 − 3 M (critical micellar concentration of Brij-35 is equal to 1× 10 − 4 M) proved that high Brij-35 concentrations are required to observe significant variations in the behavior of 8quinolinol. Therefore, apparent pKa determinations were carried out at Brij-35 concentrations ranging from 5× 10 − 3 to 2× 10 − 2 M and at a constant concentration of 8-quinolinol equal to 2×10 − 5 M. Calculations were made applying the program STAR to  70 spectra in the spectral range between 230 and 265 nm, corresponding to the absorption maxima of the species involved. The apparent pKa values obtained are summarized in Table 1. A comparison between these data and those reported previously for chlorquinadol [10] shows some similar trends in the apparent pK values: pKa1 decreases and pKa2 increases when surfactant concentration increased. However, the effect of Brij-35 micelles in the acidity constants of HQ is not as significant as

R. Codony et al. / Talanta 52 (2000) 225–232

228

Table 2 Model testing of 8-quinolinol distribution in aqueous Brij-35 solutions (equilibria: (A) H2Q+ ? H++HQ; pKa1 =4.91; (B) − + + HQ ? H++Q−; pKa2 = 9.63; (C) HQ(w) ? HQ(m); (D) Q− (w) ? Q(m); (E) H2Q(w) ? H2Q(m)) Model

Equilibria considered

1 2

A, B, C A, B, C, D

3

A, B, C, D, E

C C D C D E

Log KDa

Ub

SAc

1.88 1.93 1.27 1.96 1.37 0.32

0.054 0.024

0.0098 0.0066

0.012

0.0047

(0.03) (0.02) (0.09) (0.03) (0.06) (0.17)

a

The figures in parentheses correspond to values three times the standard deviation of the constants given by the Sum of squared residuals. c Standard deviation of residuals, in absorbance units.

SPDIS

program.

b

in the case of HClQ because of the lower solubilization of HQ into the micelles (see Section 3.2).

3.2. Distribution equilibria of 8 -quinolinol The distribution constants of 8-quinolinol were determined using a modified version of the SPDIS program according to partition model, as described in a previous paper [10]. The spectral data obtained in the study of the apparent dissociation constants of 8-quinolinol in Brij-35 micellar solutions were also used to determine the partition equilibria. Several equilibria models were tested using this program. In Table 2 the models assayed and the fitting parameters obtained for each are listed. These results showed that if only the distribution of the neutral species was considered (model 1), the standard deviation obtained was high, whereas, when the distribution of the anionic species (Q−) was also taken into account (model 2) the fit with the experimental data improved significantly. Finally, when the distribution of the cationic species was also included (model 3), although the fit improved, the distribution constant of H2Q+ was low and ill-defined, giving a calculated concentration for this species in the micellar phase less than 5%. Based on these results, the model adopted for this system was model 2. The final results are given in Table 5. The observed behaviour of 8-quinolinol in Brij-35 micellar media is similar to that reported in other studied non-ionic surfactants, like Triton X-100 or

PONPE-7.5 [15], in which the anionic species is more solubilized into the micelles than the cationic one. From these distribution constants, and the pKa values in aqueous medium, theoretical apparent dissociation constants were calculated. The good agreement obtained, between experimental and calculated values (Table 1) shows that the proposed model provides a satisfactory description of equilibria in micellar solutions of Brij-35.

3.3. Extraction equilibria of metal chelates The complex formation and extraction constants were determined by using the SPDIS program and considering the partition model. In this model, we assume the distribution of neutral species between aqueous and micellar phases. The complex formation equilibria in aqueous phase are described (in the case of oxine by way of example) by: Q− + Zn2 + ? ZnQ+ 2Q− + Zn2 + ? ZnQ2

b1 =

[ZnQ+] [Q−][Zn2 + ]

b2 =

[ZnQ2] [Q ] [Zn2 + ] − 2

(1) (2)

Both complexes could be, in principle, extracted to the micellar phase, but only the ZnQ2 complex is a neutral species. To extract the ZnQ+ species an ion-pair with a counter-ion should be formed. This can be done with the chloride ions from the background electrolyte (NaCl), as its concentration is much higher than the other species in

R. Codony et al. / Talanta 52 (2000) 225–232

solution. In this case, the extraction equilibria are described as:

229

+ − 2+ − Q− w + Clw + Znw ? ZnQ Clm

Kex1 =

[ZnQ+Cl−]m [Q−]w[Cl−]w[Zn2 + ]w

2+ 2Q− w + Znw ? ZnQ2m

Kex2 =

(3) [ZnQ2]m [Q − ]2w[Zn2 + ]w (4)

Fig. 1. Partition equilibria between aqueous and micellar phase for oxine and its Zn–complexes.

where the subscripts ‘w’ and ‘m’ indicate species in water and micellar media, respectively. The equilibrium described in Eq. (3) indicates that the extraction depends on the chloride concentration in aqueous phase. However, as the concentration of background electrolyte (0.1 M) is much higher than the total metal concentration, we can assume that the [Cl−] is nearly constant, and we can define a conditional constant as: K%ex1 =

Fig. 2. Experimental data corresponding to Zn–chlorquinaldol system in Brij-35 micellar media (A), and simulated data (B) taking into account only the dissociation and distribution constants of chlorquinaldol in the same conditions.

[ZnQ+]m [Q ]w[Zn2 + ]w −

(5)

As the concentration of the ZnQ+ species in the micellar phase is equal to the concentration of the ion-pair species ZnQ+Cl−, the conditional constant K%ex1 is equal to: K%ex1[Cl−]w Fig. 1 shows the various possibilities for the partition equilibria between the aqueous and micellar phases. For Zn(II)–chlorquinaldol and Zn(II)–8quinolinol systems in solutions of Brij-35 the calculations were carried out over 74 and 116 spectra, respectively, at Zn(II) concentrations ranging from 5× 10 − 6 to 1× 10 − 5 M and ligand concentrations from 1× 10 − 5 to 3× 10 − 5 M. The Brij-35 concentrations were varied between 2× 10 − 3 and 5× 10 − 3 M, and from 5× 10 − 3 to 2× 10 − 2 M, for chlorquinaldol and 8-quinolinol complexes, respectively. In both cases, the working pH range was set between 5 and 7.5. The spectral ranges were between 300 and 420 nm for the Zn–chlorquinaldol complex and between 250 and 275 nm in the case of the Zn–oxine complex. These ranges correspond to the absorption maxima of the complexes. The effect of complex formation can be observed in Fig. 2, where the data corresponding to Zn(II)–chlorquinaldol system in Brij-35 solutions are plotted. The experimental data are shown in

R. Codony et al. / Talanta 52 (2000) 225–232

230

Table 3 Model testing of Zn–chlorquinaldol formation and extraction constants in aqueous Brij-35 solutions (equilibria: (A) H2ClQ+ ? H+ − +HClQ, pKa1 =3.44; (B) HClQ ? H++ClQ−, pKa2 =7.84; (C) HClQ(w) ? HClQ(m), log KD =3.64; (D) ClQ− (w) ? ClQ(m), log K%D = 2+ − 2+ − + 2+ − + 2.77; (E) Zn(w) +2ClQ(w) ? ZnClQ2(m); (F) Zn(w) +ClQ(w) ? ZnClQ(w); (G) Zn(w) +ClQ(w) ? ZnClQ(m)) Model

Equilibria considered

1 2

A, B, C, D, E A, B, C, D, E, F

3

A, B, C, D, E, G

4

A, B, C, D, E, F, G

E E F E G E F G

Log K

U

S(A)

18.67 18.62 7.15 18.73 10.00 19.14 5.11 10.87

0.0096 0.0092

0.0023 0.0023

0.0091

0.0023

0.0089

0.0022

(0.03) (0.05) (0.42) (0.04) (0.15) (0.03) (67.21) (0.04)

Table 4 Model testing of Zn–8-quinolinol formation and extraction constants in aqueous Brij-35 solutions (equilibria: (A) H2Q+ ? H++ − 2+ HQ, pKa1 = 4.91; (B) HQ ? H++Q−, pKa2 = 9.63; (C) HQ(w) ? HQ(m), log KD =1.93; (D) Q− (w) ? Q(m), log K%D =1.27; (E) Zn(w) + − 2+ − + 2+ − + 2Q(w) ? ZnQ2(m); (F) Zn(w) +Q(w) ? ZnQ(w); (G) Zn(w) +Q(w) ? ZnQ(m)) Model

Equilibria considered

1 2

A, B, C, D, E A, B, C, D, E, F

3

A, B, C, D, E, G

4

A, B, C, D, E, F, G

E E F E G E F G

Fig. 2A, whereas Fig. 2B indicates how would be these spectra if no complexes are formed; these data have been simulated with SPDIS taking into account only the dissociation and distribution constants of chlorquinaldol. It should be borne in mind that the complexes formed (stoichiometry 1:2 metal:ligand) are sparingly soluble in water, therefore it is necessary to work at high surfactant concentrations (so that the complex was virtually fully extracted in the micellar phase) in order to avoid the formation of precipitates. The models tested with SPDIS included as known parameters: the dissociation and distribution constants of the ligands, the molar absorptivities of their different species in the aqueous and micellar phases and the surfactant partial molar volume and its critical micelle concentration. Given that the complexes 1:1 and 1:2 could be

Log K

U

S(A)

18.40 18.13 7.98 18.30 9.67 17.63 7.83 8.65

0.1331 0.0637

0.0139 0.0097

0.1175

0.0131

0.0550

0.0090

(0.04) (0.07) (0.07) (0.09) (0.20) (0.10) (0.06) (0.36)

extracted in micellar media and the complex 1:1 could be formed in the aqueous phase, several equilibria models were proposed. The fitting parameters obtained for each model are given in Tables 3 and 4. Results showed that in the case of the complex Zn–chlorquinaldol (Table 3), the standard deviation of residuals remained virtually the same in all models. On the other hand, the formation and extraction constants corresponding to complex 1:1 were ill-defined. Therefore model 1 was adopted for this system. In order to determine whether the model chosen was the most appropriate, we calculated the number of species in solution by factor analysis [14]. The results showed that there were three species in solution (Fig. 3). Taking into account the pH range studied, these different species correspond to: HClQ(m), HClQ(w) and ZnClQ2(m), therefore, according to the results obtained, we can conclude

R. Codony et al. / Talanta 52 (2000) 225–232

that the formation and extraction of the species ZnClQ+ could be rejected from the equilibrium model. In the case of the Zn – oxine complexes (Table 4), we considered in the first instance the extraction equilibrium of the 1:2 complex in micellar media (model 1). In further models we introduced the formation of the 1:1 complex in aqueous phase (model 2), its extraction in micellar phase (model 3) and both equilibria (model 4). If we compare the results from model 1 with those obtained from the other models, we can observe that only models 2 and 4 showed a significant improvement in the standard deviation of the

231

residuals. However, the extraction constant of the complex 1:1 in micellar phase was ill-defined in model 4. Therefore, model 2, which includes the formation constant of complex 1:1 in the aqueous phase and the extraction constant of the complex 1:2, was selected. This is in agreement with the results obtained with factor analysis (Fig. 3), which indicate that there are four species in solution (corresponding to the species HQ(m), HQ(w), ZnQ+ (w) and ZnQ2(m)). Although cationic species may be partioned as ion pairs, no significant distribution was observed when sodium chloride was used as background electrolyte. The complex formation and extraction constants obtained are given in Table 5. The results obtained for the formation constant of the ZnQ+ species in the aqueous phase agrees with the value given in the literature (log b1 = 8.56) [16]. The difference in the logarithm of the formation constant is  0.6 units, which can be attributed to different experimental conditions.

3.4. Fluorescence studies

Fig. 3. Determination of the number of absorbing species by factor analysis for the systems: Zn–chlorquinaldol () and Zn – oxine ( ) in aqueous Brij-35 solutions. Table 5 Zn–oxine and Zn–chlorquinaldol systems in aqueous Brij-35 solutions Equilibrium

Log K

HQ(w) ? HQ(m) − Q− (w) ? Q(m) 2+ + Zn(w) +Q− (w) ? ZnQ(w) − Zn2+ +2Q ? ZnQ (w) (w) 2(m) HClQ(w) ? HClQ(m) − − ClQ(w) ? ClQ(m) − Zn2+ (w) +2ClQ(w) ? ZnClQ2(m)

1.93 1.27 7.98 18.13 3.64 2.77 18.67

(0.02) (0.09) (0.07) (0.07) (0.02) (0.04) (0.03)

A well-known property of hydroxyquinoline derivatives is their ability to form highly fluorescent complexes with several metal ions, among them Zn(II), for which some analytical applications based on the fluorescence of their chelates have been reported [17–19]. These metal chelates, originated from neutral ligands, are sparingly soluble in water and are not fluorescent in water. However, the use of hydroorganic or micellar media leads to fluorescence emission. Here, the fluorescence intensities of these metal chelates in organic or micellar media may be related to the partition equilibria. In order to prove the relationship between fluorescence and micellar solubilization of Zn(II)– oxine and Zn(II)–chlorquinaldol chelates, their fluorescence intensities as a function of pH were recorded. Measurements were taken at the optima emission and excitation wavelengths, from solutions containing a constant excess of ligand (2× 10 − 4 M) and a high surfactant concentration (1×10 − 2 M). Observed relative fluorescence emissions are displayed in Fig. 4. Solid lines correspond to estimated extraction recoveries of

232

R. Codony et al. / Talanta 52 (2000) 225–232

References

Fig. 4. Relative fluorescence intensities of Zn–oxine ( ) and Zn – chlorquinaldol () as a function of pH. Solid line corresponds to estimated extraction recoveries of complex 1:2 in the micellar phase.

complexes 1:2 in the micellar phase, after the models proposed in Table 5. The good agreement between fluorescence and calculated extraction curves indicates that solubilization of the 1:2 complex in the micelles is the determining factor for fluorescence. Moreover, the quantitative relationship found between the values of fluorescence intensities and extraction recoveries confirms the validity of the proposed model.

Acknowledgements The financial support of the CIRIT de la Generalitat de Catalunya (1997SGR 00394) is gratefully acknowledged.

.

[1] G.L. McIntire, CRC Crit. Rev. Anal. Chem. 21 (1990) 257. [2] E. Pramauro, E. Pelizzetti, in: S.G. Weber (Ed.), Wilson & Wilson’s Comprehensive Analytical Chemistry, vol. 31, Elsevier, Amsterdam, 1996. [3] T. Saitoh, Y. Kimura, T. Kamidate, H. Watanabe, K. Haraguchi, Anal. Sci. 5 (1989) 577. [4] T. Saitoh, H. Hoshino, T. Yotsuyanagi, J. Chem. Soc. Faraday Trans. 90 (1994) 479. [5] T. Saitoh, H. Segawa, T. Kamidate, H. Watanabe, K. Haraguchi, Bull. Chem. Soc. Jpn. 66 (1993) 3676. [6] H. Watanabe, T. Saitoh, T. Kamidate, K. Haraguchi, Mikrochim. Acta 106 (1992) 83. [7] M.P. San Andre´s, S. Vera, J. Liq. Chromatogr. Rel. Technol. 19 (1996) 799. [8] T. Saitoh, N. Ojima, H. Hoshino, T. Yotsuyanagi, Mikrochim. Acta 106 (1992) 91. [9] T. Saitoh, H. Hoshino, T. Yotsuyanagi, J. Chromatogr. 469 (1989) 175. [10] J.L. Beltra´n, R. Codony, M. Granados, A. Izquierdo, M.D. Prat, Talanta 40 (1993) 157. [11] J.L. Beltra´n, M.D. Prat, R. Codony, Talanta 42 (1995) 1989. [12] J.L. Beltra´n, G. Centeno, A. Izquierdo, M.D. Prat, Talanta 39 (1992) 981. [13] D.D. Perrin, B. Dempsey, Buffers for pH and Metal Ion Control, Chapman and Hall, London, 1974. [14] J.L. Beltra´n, R. Codony, M.D. Prat, Anal. Chim. Acta 276 (1993) 441. [15] H. Hoshino, T. Saitoh, H. Taketomi, T. Yotsuyanagi, H. Watanabe, K. Tachikawa, Anal. Chim. Acta 147 (1983) 339. [16] R. Nasanen, U. Penttinen, Acta Chem. Scand. 6 (1952) 837. [17] A. Ferna´ndez Gutie´rrez, A. Mun˜oz de la Pen˜a, in: S.G. Schulman (Ed.), Molecular Luminiscence Spectroscopy. Methods and Applications, Ch. 6, Wiley, New York, 1985. [18] R. Compan˜o´, A. Grima, A. Izquierdo, M.D. Prat, Anal. Chim. Acta 227 (1989) 219. [19] A. Izquierdo, R. Compan˜o´, E. Bars, Talanta 33 (1986) 463.