All solutions were prepared in boiled out triple distilled water.

1) L-phenylalanine and Maleic acid Solutions

0.05 moldm⁻³ aqueous solution of L-phenylalanine (GR grade, E-Merck, Germany) and maleic acid (GR grade, E-Merck, Germany) were prepared by dissolving samples in water. To increase the solubility of ligands, 0.05 moldm⁻³ hydrochloric acid concentration was maintained in the solutions. The probable errors that may creep into the concentrations of the stock solutions of the ligands were determined by the computer program COSWT [10] . The pessimistic errors in the preparation of the ligand solutions by weight method did not exceed 0.1%.

2) EDTA Solution

Disodium salt of EDTA was purified by precipitation from aqueous solution using methanol. The precipitate was washed with acetone, diethyl ether and finally air dried. An 0.1 moldm⁻³ solution was prepared and standardized complex metrically with a 0.1 moldm⁻³ standard Zn (II) solution using Eriochrome Black-T as indicator.

3) Metal ion Solutions

0.1 moldm⁻³ aqueous solutions of cobalt (II), nickel (II) and copper (II) chlorides were prepared by dissolving GR grade (E-Merck, India) salts in triple distilled water. The stock solutions were rendered slightly acidic to repress hydrolysis of the metal ions. The concentrations of the metal ions were determined complex metrically by titrating against a standard solution of EDTA using the xylenol orange, murexide and fast sulphon black-F as indicators and hexamethylenetetramine powder as buffer for cobalt to maintain the pH at 5 - 6. The free hydrogen ion concentration in the stock solution was determined by Gran plot method [11] .

4) Sodium Hydroxide

A stock solution of 1 moldm⁻³ sodium hydroxide was prepared by dissolving GR grade (E Merck, India) sodium hydroxide pellets in triple distilled water. The solution was further diluted to the required concentration. The strength of the alkali was determined by titrating it against a standard solution of oxalic acid and potassium hydrogen phthalate, while the molarity of hydrochloric acid was determined with sodium hydroxide. So as to assess the errors that might have crept into the determination of the concentrations, the data were subjected to analysis of variance of one way classification (ANOVA) using the computer program COST (concentration of solution by titration). The strength of alkali was determined using the Gran plot method. The errors in the concentration of ligand, metal ions and alkali were subjected to analysis of variance ANOVA [12] .

5) Hydrochloric Acid

The concentration of stock hydrochloric acid (GR grade, Merck, India) solution was calculated from its specifications (specific gravity, purity and molecular weight) and diluted to the required concentration by dissolving in triple distilled water. Its strength was determined by titrating with standard sodium hydroxide solution.

Analargrade (BDH) urea was recrystallized twice from triply distilled water and was dried at 60˚C for 2 h. A stock solution of urea was prepared by dissolving and appropriate quantity of the purified sample in water and stored in frozen state. The solution was not allowed to stand at room temperature continuously for more than 24 h.

GR grade N, Urea (Finar, India) of 99.5% purity was used as a solvent.

The pH measurements of the proton-ligand systems were carried out in aqueous media containing varying compositions of organic solvent (Urea) in the range of 0% - 50% v/v maintaining an ionic strength of 0.16 moldm⁻³ with sodium chloride at 298 K using a digital pH meter ELICO-LI120 type (readability 0.01). Potassium hydrogen phthalate (0.05 moldm⁻³) and borax (0.01 moldm⁻³) solutions were used to calibrate the pH meter. In each titration, the titrand consisted of approximately 1 mmol of nitric acid. The amounts of the L-phenylalanine, maleic acid (ligands) in the titrand are in the range of 0.25 - 0.50 mmol. The glass electrode was equilibrated in a well stirred Urea-water mixture containing inert electrolyte for several days. At regular intervals, the strong acid was titrated against alkali to check the complete equilibration of the glass electrode. In these titrations, the titrand consisted of mineral acid and ligand, in a total volume of 50 mL. Titrations were performed by adding each time 0.1 cm³ portions of 0.4 moldm⁻³ sodium hydroxide to the titrand. The pH meter reading was recorded only after a constant value was displayed. Typical duplicate titrations showed that equilibration was fast and titration data did not differ by more than 0.02 units [13] .

In each of the titrations, the titrand consisted of approximately 1mmol hydrochloric acid in a total volume of 50 mL and the ionic strength was adjusted to 0.16 mol∙L⁻¹ with sodium chloride. Solutions containing metal ions and ligands (metal to ligand ratios being in the range of 1:2.5:2.5, 1:2.5:5.0, 1:5.0:2.5 were titrated with 0.4 mol∙L⁻¹ sodium hydroxide.

The approximate protonation constants of L-phenylalanine and maleic acid were calculated with the computer program SCPHD [14] . The best fit chemical model for each system investigated was arrived at using non-linear least squares computer program, MINIQUAD75, which exploits the advantage of constrained least-squares method in the initial refinement and reliable convergence of Marquardt algorithm. The variation of stepwise protonation constants (logK) with the mole fraction of the medium was analyzed on electrostatic grounds for the solute-solute and solute-solvent interactions [15] .

In order to rely upon the best-fit model for critical evaluation and application under varied experimental conditions with different accuracies of data acquisition, an investigation was undertaken by introducing pessimistic errors in the influential parameters. The results of effect of pessimistic errors in the concentrations of alkali, mineral acid, ligands and metal are given in Table 2 which

emphasize that the errors in alkali and acid affect stability constant more than those in the ligands and metal. Some species are even rejected when errors are introduced in the concentrations. This indicates the appropriateness of the experimental conditions and accuracy of the concentrations.

In order to rely upon the best chemical model for critical evaluation and application under varied experimental conditions with different accuracies of data acquisition, an investigation was made by introducing pessimistic errors in the influential parameters like concentrations of alkali, mineral acid, ligand and metal. The order of the ingredients that influence the magnitudes of stability constants due to incorporation of errors is alkali > acid > MA > Phe > metal (Table 2). Some of the species refined in the absence of errors were even rejected when errors were introduced in the concentrations. One or more of MLXH, MLX and ML₂X species were rejected depending upon the magnitude of error as given in Table 2. The rejection of species and increased standard deviations in the stability constants on introduction of errors confirm the appropriateness or correctness of the experimental conditions (concentrations of ingredients) and the choice of the best fit models.

The formation of mononuclear unprotonated binary and ternary complexes from a mixture of metal ion (M) and primary (L) and secondary (X) ligands can be shown as the equilibria given in (1).

Reasons for extra stability of ternary complexes

The change in the stability of the ternary complexes as compared to their binary analogues was quantified [21] [22] [23] [24] based on the disproportionation constant (logX) given by Equation (2) corresponding to the equilibrium.

M L 2 + M X 2 ⇌ 2 M L X

log X = 2 log K M L X M − log K M L 2 M − log K M X 2 M (2)

Under these equilibrium conditions one can expect 50% ternary complex and 25% each of the binary complexes to be formed and the value of logX was reported [25] to be 0.6. A value greater than this accounts for the extra stability of MLX. Another approach [26] to quantify the stability of ternary complexes was based on the difference in stability (ΔlogK) for the reactions ML with X and M_(aq) with L and X, where L is primary ligand and X is the secondary ligand. It is compared with that calculated purely on statistical grounds. Equation 5.3 can be formulated based on the properties of the cyclic systems reported earlier [27] from which it is clear that both the ligands in the ternary complex influence mutually to the same extent.

Δ log K = log K M L X M − log K M L M − log K M X M (3)

The electrostatic theory of binary complex formation and statistical arguments suggest the additional coordination positions of given multivalent hydrated metal ion available for the first ligand than for the second. Hence, the usual order of stability K M L M > K M L X M L applies. This suggests that ΔlogK should be negative, although several exceptions [28] have been found. The statistical values of ΔlogK for bidentate L and X are −0.4, −0.6 and between −0.9 and −0.3 for octahedral, square planar and distorted octahedral complexes, respectively. Negative values of ΔlogK can be understood as the secondary ligand forms a more stable complex with hydrated metal ion than with ML. Whenever the experimental values of ΔlogK exceed the statistical values, it can be inferred that the ternary complex is formed as a result of interaction of ML with X or MX with L. ΔlogK values of ternary complexes containing bipyridyl as the primary ligand are positive [29] for O-donors (malonic acid, pyrocatechol etc.), negative for N-donors (ethylene diamine) and intermediate or negative for amino acids with both N and O co-ordination sites. However, a very high negative value (−2.3) for Cu (en) (iminodiacetic acid) and a positive value (0.82) for Cu (o-F)-(6, 7-dihydroxynaphthaline-2 sulphonate) was also observed [30] .

A perusal of the distribution diagrams (Figures 2(A)-(F)) reveals that the concentrations of binary species are less compared to ternary species which indicates the existence of more stable ternary complexes. The ternary species exist in the pH range 1.5 - 8.9 for all the metal ions (Co (II), Ni (II) and Cu (II). The formation of the complex species can be represented by the following equilibria.

M ( II ) + LH 2 + + XH 2 ⇌ MLXH + 3H + (4)

MLH 2 + + XH 2 ⇌ MLXH + 2H + (5)

MXH + + LH 2 + ⇌ MLXH + 2H + (6)

MLXH ⇌ MLX − + H + (7)

ML 2 H + + XH 2 ⇌ ML 2 X 2 − + 3 H + (8)

ML 2 + XH − ⇌ ML 2 X 2 − + H + (9)

MLX − + LH ⇌ ML 2 X 2 − + H + (10)

In the pH range 2.1 - 11.4 and 1.6 - 10.2, F and MA exist as LH 2 + , LH, L⁻ and XH₂, XH⁻, X² respectively. These protonated ligands interact with the metal ion to form MLXH (Equilibrium 4) which may successfully deprotonated to form MLX⁻ (Equilibria 7). Formation of MLXH species can be explained based on the protonated ligands interact with the metal ion to form MLXH (Equilibrium 4) and also due to interaction of binary species with ligand species (Equilibrium 5 and 6). MLXH₂ species has not been detected probably because it is less stable or quickly deprotonated to MLXH at lower pH whereas for the formation of MLX⁻, Equilibria 7 is relevant because the pH at which the concentration of MLXH decreases, MLX⁻ increases in the pH range of 2.0 - 5.1. ML₂X²⁻ is formed by the interaction of metal ion with to XH₂, LH species and one XH⁻ species (Equilibrium 8, 9 and 10). The existence of ML₂X and the absence of MLX₂ may be due to the higher affinity of LH than XH towards the metal ion.

In aqueous solutions, Co (II), Ni (II) and Cu (II) are coordinated by six water molecules. Amino nitrogen can associate with a proton at physiological pH. There is often significant competition between proton and metal ion for this donor site, resulting in the formation of protonated species. Depending upon the nature of the ligands and metal ions and based on the basic chemical knowledge, structures of the complexes are shown in Figure 3. The Cu (II) ion forms distorted octahedral or square planar complexes due to Jahn-Teller effect.