A stock of copper solution (888.7 ppm) was prepared by dissolving CuSO₄∙5H₂O in distilled water.

Cortical bovines have been collected from the local slaughter houses in Sfax, Tunisia. The bone samples have been washed and cleaned using boiling method to eliminate organic substances and collagen, to avoid soot developing in the material during the pyrolysis process. Cow bones were boiled in water for 3 h at 99.5˚C. Then the water was eliminated and the bones were washed using fresh water [24] . At this moment, the bones were dried in an oven at 80˚C. The dried bones were crushed and milled into different particle sizes and after calcined for 2 h at 800˚C with a mixture of oxygen/nitrogen (25/75 vol%) [25] . The resulting material was denominated as Bone Charcoal (BC). The resulted BC was crushed again and refined to obtain a powder with particles sizes < 50 µm and finally stored in hermetic bottles for subsequent uses.

Carbonization was carried out in a vertical stainless-steel reactor (length = 170 mm, internal diameter = 22 mm), which was placed into a cylindrical electric furnace Nabertherm. Mineralogical analysis of the BC sample was realized by X-ray diffraction (Bruker D8) and showed the characteristic peaks of amorphous carbon [26] . An IR transmittance spectrum of the BC sample was carried out in the 500 - 2000 cm⁻¹ range using a SHIMATZU IR 470 spectrometer. The specific surface area (m²/g) of BC was determined from nitrogen adsorption-de- sorption isotherm with the Brunauer-Emmett-Teller (BET) method. The macroporous volume of BC was measured by mercury intrusion porosimetry (Joschek et al., 2000). The concentration of copper was carried out by atomic absorption spectrometry (HITACHI Z-6100).

The adsorption experiments were determined according to the batch experiments at room temperature (25˚C). 0.1 g of BC was mixed with a 50 mL of copper solution at different concentrations (0 - 500 mg/L). The deviation of the pH of the contaminated heavy metal solution during copper adsorption was measured for all experiments, with an experimental deviation = ±1. The initial pH was fixed <5.5 to avoid precipitation of Cu(OH)₂.

The quantity of adsorbed copper, q_t, on BC is calculated according to Equation (1):

where C_o is the initial concentration of copper (mg/L), C_e is the residual concentration of copper (mg/L), m is the mass of BC (g) and V is the volume of the copper solution (L).

In order to examine the mechanism of adsorption processes, the pseudo first- order adsorption, the pseudo second-order adsorption and the intraparticle diffusion models were used to adjust kinetic experimental data. The amount of BC sample (0.25 g) is mixed with a 50 mL of copper solution (10 mg/L) to carry out adsorption experiments in bath mode.

Desorption studies were performed in two phases [31] :

Phase 1: Adsorption. 0.25 g of BC was placed in contact with 50 mL of a 50 mg/L of copper solution. BC was then collected by filtration, washed with distilled water and placed in an oven for 12 h at 60˚C. The liquid phase was analyzed by AAS (HITACHI).

Phase 2: Desorption. The dry and saturated BC was placed in contact with 50 mL of 0.1 M HCl, NaOH, NaCl and distilled water for 2 h. The liquid phase was filtered and analyzed by AAS (HITACHI). The desorbed copper percentage was determined according to Equation (7):

where Q_des is the amount of desorbed copper from BC (mg/g) and Q_ads is the amount of copper adsorbed onto BC (mg/g).

The bones are composed of inorganic (65 wt%) and organic (35 wt%) components. The principal inorganic component of bone is hydroxyapatite, Ca₁₀(PO₄)₆(OH)₂(HA). Figure 1 shows the FTIR spectrum in the 500 - 2000 cm⁻¹ range of BC. Only the bands characteristic of HA (554 - 960 cm⁻¹) are presented [32] . The diffractogram of BC (not shown here) shows only the characteristic pattern of HA. These results confirm that the organic components are eliminated from the bone samples after calcination at 800˚C for 2 h [32] [33] .

The specific surface area of BC, S_BET, is equal to 75 m²/g and the macroporous (pore size > 50 nm) volume of BC, determined by mercury porosimetry, is equal to 0.9 cm³/g.

Figure 2 shows the effect of contact time, t, on the removal of 50 mg/L of Cu by BC. The adsorption of Cu increases with time to reach an equilibrium after 10 min of mixing and agitation (red points in Figure 2).

The three kinetic models explained in paragraph 3.2 are used in this study and the results are shown in Figures 3-5. The values of coefficients for the three models are calculated. The correlation coefficient for the pseudo second-order kinetic model is higher (R² = 1, Figure 4) than both other models. This result indicates that the copper adsorption is controlled by a chemisorption process at the surface of BC [34] . The rate constant of pseudo second-order adsorption, K₂, represents the number of exchanges between Ca ions in BC and Cu ions present in aqueous solution [35] . Furthermore, the theoretical q_e values acquired from the pseudo second-order kinetic model are in accordance with the experimental q_e values.

Figure 6 shows the percentage of desorbed copper for the four different chemical agents used in this study. HCl shows the highest efficiency with approximately 2.521% of desorbed copper in 2 h. None of the used solutions provoke any apparent physical damages to the adsorbent. Nevertheless, the desorption level of copper with 3 types of aqueous solution (alkaline, neutral and acidic) remains very low. This confirms the strong covalent bonds established between BC and copper ions. So, to regenerate BC, treatments at high temperatures are necessary to break the bonds between BC and copper ions.