All the chemicals used were of analytical grade and used without further purification. A saturated solution of Hg²⁺ (1000 mg L⁻¹) was used after appropriate dilution with double deionized distilled water. Other chemicals and reagents were from BDH quality. Oleic acid (HOL) stock solution, 6.36 × 10⁻² mol・dm⁻³ was prepared by dispersing 20 cm³ of HOL, (food grade with sp. gr. 0.895, provided by JT Baker Chemical Co.), in 1 dm³ kerosene. L¹ was prepared as described earlier [7] . A Perkin-Elmer model 2380 AAS was used, inconnection with a mercury hydride system (MHS-10). Nitrogen or argon was used as a purge gas and NaBH₄ as reluctant. Elemental analyses (C, H, M) were performed with a Perkin-Elmer 2400 series II analyzer at the Microanalytical Center at Cairo University, Egypt. Chloride was determined gravimetrically the as AgCl [8] . The IR spectrum of [Hg(L¹)Cl]Cl・5H₂O was recorded as KBr discs on Mattson 5000 FTIR spectrophotometer (400 - 4000 cm⁻¹). ¹H-NMR spectra were recorded on Jeol-90Q Fourier Transform (200 MHz) in d₆-DMSO at Cairo University, Egypt. Mass spectra were recorded on MS 70 eV EIGC, MS QP-1000 EX Shimadzu (Japan) mass spectrometer at Cairo University. Thermal analyses measurements (TG, DTG) were recorded with a Shimadzu Thermo gravimetric Analyzer TGA-50 using α-Al₂O₃ as a reference material at Mansoura University.

Two types of flotation cells were used throughout this work have been described earlier [9] . Flotation cell (a) is a cylindrically graduated glass tube of 16 mm inner diameter and 290 mm length with a stopcock at the bottom. Such cell is used to study the different factors affecting the efficiency of flotation. Flotation cell (b) is a cylindrical tube of 6 cm inner diameter and 45 cm length with a stopcock at the bottom and a quick fit stopper at the top; this cell is used to separate mercury from 1 dm³ of different water samples. The pH of each sample was adjusted in the range 2 - 10 using Hanna Instruments 8519 digital pH meter with glass and saturated calomelelectrodes calibrated on the operational state using standard buffer solutions.

The structure of L¹ and its Hg²⁺ complex, geometry optimization and conformational analysis has been performed using of MM⁺ force field as implemented in Hyperchem 8.0 [10] . The low lying obtained from MM⁺ was then optimized at PM3 using the Polak-Ribiere algorithm in RHF-SCF set to terminate at an RMS gradient of 0.01 Kcal・mol⁻¹.

Two mL of aqueous EtOH solution of 1 × 10⁻⁴ mol・L⁻¹ and L¹ were introduced into a flotation cell containing 1 × 10⁻⁶ mol⁻¹ of Hg²⁺ solution then the pH was adjusted to 5.0 using HCl and/or NaOH and the solution was mixed thoroughly. The mixture was then diluted to 10 mL with redistilled water. To the above solution 3 mL of oleic acid with a definite concentration (2 × 10⁻⁴ mol・L⁻¹) were added. The cell was then turned upside down twenty times by hand and kept upright for 5 min to ensure complete flotation of the Hg²⁺ complex species. The scum layer was eluted with 5 mL of L¹ mol・L⁻¹ HCl (1:1) solution to complete trapping [11] . The concentration of Hg²⁺ was determined using CVAAS measurements at 253.7 nm with a Perkin-Elmer 2380 atomic absorption spectrometer. The separation efficiency (%F) was calculated from the relation:

where, C_s and C_i denote the scum and the initial concentrations of Hg²⁺, respectively.

Water samples were obtained as follows: distilled water, tap water, river Nile and underground water from Mansoura City. All samples were filtered through G₄ sintered glass. For total organic mercury in water, the samples were digested in a closed system using the sequence of 10 mL of 5% KMnO4, 10 mL of 8N HNO3, 10 mL of 18 N H₂SO₄ and 20 ml of 4% K₂S₂O₈. The samples were heated at <90˚C for 30 min, allowed to cool and then 4 mL of 10% NH₂OH・HCl was added to reduce excess oxidant immediately before the flotation procedure was carried out. To large flotation cells, five water samples (1 L each) containing a defined amount of Hg²⁺ chloride and 5 mL of 10⁻³ M L¹ were added and the pH was adjusted to 5 - 6. The reaction mixture was shaken to ensure complete complex formation. Then, 8 mL 10⁻³ M HOL was added to each flotation cell and the cells are shaken upside down for five min. The scum layer was separated and eluted with 1 mol L⁻¹ HCl. The final volume was 10 mL.

L¹(C₉H₁₉N₄O₂Cl) was synthesized as described earlier [12] and can be represented by keto/enol forms as shown Figure 1. The product is white in color and soluble in H₂O and most polar organic solvents and the value of molar conductance in DMSO (28.3 ohm⁻¹・cm²・mol⁻¹) suggesting the electrolytic nature of L¹. The structure of L¹ is confirmed using elemental analyses (Calcd: C = 40.2, H = 7.9, Cl = 13.2; Found: 40.4, 7.3, 12.9) and spectral measurements.The melting point of L¹ is 172˚C, which matches the results reported value [12] .

The Hg²⁺ complex, [Hg(L¹)Cl]Cl・5H₂O, was synthesized by refluxing solutions of equivalent amounts of L¹ and

HgCl₂ in absolute EtOH for 0.5 h. The product was filtered off, washed several times with hot EtOH and Et₂O and finally dried in a vacuum desiccator over anhydrous CaCl₂. The yield of the Hg²⁺ complex is 96%. The structure of the complex is confirmed by its melting point (212˚C) and elemental analyses (Calcd: C = 17.7, H = 4.8, Hg = 32.7, Cl = 17.4; Found: 16.9, 4.8, 32.4 and 18.1) and represented in Figure 2.

The IR spectrum of the free L¹ (Figure 1(a)) in KBr shows a strong band at 1697 cm⁻¹ assignable to the ν(C=O) vibration [13] in addition to medium and weak bands at 1650, 1614, 1405 and 1020 cm⁻¹ assigned to the azomethine of hydrazone ν(C=N¹), azomethine of oxime ν(C=N²), δ(OH) and ν(N-N) vibrations, respectively [14] . Also, the two bands observed at 3124 and 3214 cm⁻¹ are assigned to the free ν(NH) and hydrogen bonded, respectively. The bands observed at 3395 and 3454 cm⁻¹ are attributed to the (OH) free and bonded hydrogen, respectively. The broad weak bands in the 1800 - 1200 cm⁻¹ and 2200 - 2400 cm⁻¹ regions are taken as an evidence for the existence of intra-molecular hydrogen bonding of the type (OH… N) (Figure S1) [15] .

The IR spectrum of [Hg(L¹)Cl]Cl・5H₂O (Figure 3) suggests that L¹ coordinates as a neutral bidentate via the two azomethine groups as shown in Figure 2. The mode of chelation is supported by the IR spectrum where; i) the negative shift of both the bands of azomethine (C=N¹) and (C=N²) groups and ii) the bands of the (C=O) and (OH) groups remainexisted after coordination indicating that these groups are not participated in the coordin- ation.

The ¹H-NMR spectrum of L¹ in d₆-DMSO (Figure 4) shows two signals at 11.74 ppm and 11.25 ppm, down- field with respect to TMS, which disappear upon adding D₂O. These signals are attributed to the protons of (OH) of the oxime group and (CONH) group, respectively. The signals in the 1.98 - 2.19 ppm range are assigned to the three methyl groups (CH₃)₃. Also, the signals observed at 2.50, 4.56 and 4.76 ppm are attributed to the protons of

(CH₃) and (CH₂) of the oxime group and (CH₃) of the hydrazone group, respectively. All these foundations are taken as evidence that L¹ is mainly existed in the keto form either in the free case or in the hydrogen bonded.

The ¹H-NMR spectrum of [Hg(L¹)Cl]Cl・5H₂O in d₆-DMSO (Figure S2) shows two signals at 11.73 ppm and 11.24 ppm, downfield with respect to TMS, which disappear upon adding D₂O (Figure S3). These signals are attributed to the protons of (OH) of the oxime and the NH of the (CONH) group, respectively. The signals in the 1.97 - 2.12 ppm range correspond to the three methyl groups (CH₃)₃. Also, the observed signals at 2.49 - 2.51 ppm, 4.3 ppm and 4.72 ppm are attributed to the protons of (CH₂) and (CH₃) of the oxime group and (CH₃) of the hydrazone group, respectively. All these observations confirm that the complex exists in the keto form.

The mass spectrum of L¹ (Figure S4) shows the molecular ion peak at m/z = 250. This suggests that the proposed structure for L¹ is correct and has the chemical formula; C₉H₁₉N₄O₂Cl and the M. wt. = 250.726. Also, the results of elemental analyses and ¹H-NMR are taken as strong evidences for the proposed structure (Figure 2). The mass fragments of L¹ are shown in Scheme S1. The mass spectrum of [Hg(L¹)Cl]Cl・5H₂O (Figure S5) shows the molecolare ion peak at m/z = 613 while the theoretical value is 612.29.

The molecular modeling along with atom member of L¹ and its Hg²⁺ complex, [Hg(L¹)Cl]Cl・5H₂O, are shown in Figure 5 and Figure 6. The data are calculated using quantum mechanics for the complexes. Semi-empirical molecular Mechanics Optimization method is used.

The data listed in Table S1 and Table S2 reveals the following remarks:

1) Some bond lengths don’t change as in N(8)-O(20) of L¹, N(3)-O(13) of complex, N(7)-N(13) of L¹ and

Scheme S1. The fragmentation pattern of L¹.

N(4)-N(15) of complex which have the same values 1.316 Å and 1.352 Å, respectively.

2) On the other hand, some bonds are elongated as C(1)-N(7) and C(6)-N(8) of L¹ (1.244 Å) but C(1)-N(4) and C(2)-N(3) of complex (1.377 Å).

3) Other bonds are shortened as C(1)-C(2) of L¹ (1.540 Å) changed into 1.337 Å in the complex.

4) The same notifications can be discussed in bond angles. These differences take place on coordination and formation of the five-membered ring, which ensures the minimum energetic state of the complex.

Thermal studies of the Hg²⁺ complex is studied in the range 30˚C - 800˚C to insight about its thermal stability, the nature of the solvent molecules and the general scheme for their thermal decomposition. The data showed that the water of crystallization is volatilized within the temperature range 75˚C - 125˚C. The TGA decomposition steps with the temperature range and weight loss for the Hg²⁺ complex.

The kinetic parameters evaluated by Coats-Redfern method (Figure 7) are listed in Table 1. The data reveals the following observations:

1) All decomposition stages showed a best fit for n = 1, while the other values have no better correlation.

2) The activation energy (E_a) decreases for the subsequent degradation steps revealing a less energy needed for the thermal decomposition of the remaining parts.

3) The negative value of the entropy of activation (ΔS*) of the decomposition steps of the metal complex indicates that the activated fragments have more ordered structure than the undecomposed complex and/or the decomposition reactions are slow [16] .

4) The negative sign of the enthalpy of activation ΔH* of the decomposition stages reveals that the decomposition stages are easier.

The positive sign of free energy of activation (ΔG*) indicates that the free energy of the final residue is higher than that of the initial compound and hence all the decomposition steps are nonspontaneous processes. Moreover, the values of ΔG* increase significantly for the subsequent decomposition stages of a given compound. This conclusion, as a result of the increasing of TΔS* reflects that the rate of removal of the subsequent species is lower than that of the precedent one [17] - [19] .