Ligand Adsorption on an Activated Carbon for the Removal of

6908

Langmuir 2005, 21, 6908-6914

Ligand Adsorption on an Activated Carbon for the Removal of Chromate Ions from Aqueous Solutions J. Garcı´a-Martı´n,† R. Lo´pez-Garzo´n,*,† M. Luz Godino-Salido,† M. Dolores Gutie´rrez-Valero,† P. Arranz-Mascaro´s,† R. Cuesta,† and F. Carrasco-Marı´n‡ Departamento de Quı´mica Inorga´ nica y Orga´ nica, Universidad de Jae´ n, 23071-Jae´ n, Spain, and Departamento de Quı´mica Inorga´ nica, Universidad de Granada, 18071-Granada, Spain Received March 1, 2005. In Final Form: May 9, 2005 The results presented in this work are related to the design of a guideline to develop specific properties at the surface of an activated carbon (AC). For this, two model aromatic compounds have been synthesized and their electrolytic behavior in aqueous solutions was studied by a potentiometric method. The textural characteristics of the activated carbon were determined by porosimetry methods. The nature of oxygencarrying functions and the acid-base behavior of the AC surface were characterized by TPD and potentiometric titration methods, respectively. The adsorption and desorption equilibria of the aromatic compounds on activated carbon were measured in aqueous solutions, and the hysteresis between adsorption and desorption, which reveals irreversible adsorption, was discussed on the basis of the frontier orbital theory. HOMO and LUMO orbitals of the adsorbent and adsorbates were calculated, and irreversible adsorption was attributed to the small energy difference between HOMO and LUMO of the aromatic adsorbates and the adsorbent. Adsorption equilibria of K2CrO4 in aqueous solution on the AC alone and on the AC-aromatic ligand adsorbents, respectively, prove the efficient development of specific chemical functions at the carbon surface provided by the adsorbed aromatic compounds.

Introduction Many papers have been published in the past decades confirming that the adsorption on an activated carbon surface has a great potential for the removal of toxic metals.1-3 Metal contamination is a serious and rising problem affecting industrial and municipal wastewater (among others). Although it is well-known how to prepare activated carbons possessing optimal properties associated with surface area and specific pore size distribution, another important aspect in the development of efficient adsorbents, as is the control of the surface chemical functionalization, continue being a difficult task when the usual activation techniques are employed.3 This is particularly true when the attainment of activated carbons showing a marked specificity for a particular metal ion species (or in general, a particular chemical species) is intended. Different methods have been described in the literature to functionalize activated carbon and silica surfaces by anchoring potential ligands or metal complexes, and some of resulting modified systems seem to be useful as catalysts in different reactions.4-7 On the other hand, to increase the selectivity for the removal of trace inorganic pollutants * Corresponding 953211876; e-mail: † Universidad de ‡ Universidad de

author phone:+34-953212186; fax: [email protected]. Jae´n. Granada.

(1) Brown, P. A.; Gill, S. A.; Allen, S. J. Water Res. 2000, 34, 39073916. (2) Csoba´n, K.; Pa´rka´nyi-Berka, M.; Joo´, P.; Behra, P. Colloid. Surfaces A: Physicochem. Eng. Aspects 1998, 141, 347-364. (3) Carrott, P. J. M.; Carrott, M. M. L.; Nabais, J. M. V.; Ramalho, P. J. P. Carbon 1997, 35, 403-410. (4) Puggin, B. J. Mol. Catal. A: Chem. 1996, 107, 273. (5) Bischoff, S.; Weigt, A.; Kant, M.; Schu¨lke, U.; Lu¨cke, B. Catal. Today 1997, 36, 273-284. (6) Crudden, C. M.; Allen, D.; Mikoluk, M. D.; Sun, J. Chem. Commun. 2001, 1154. (7) Roma´n-Martı´nez, M. C.; Dı´az-Aun˜o´n, J. A. M.; Salinas-Martı´nez de Lecea, C.; Alper, H. J. Mol. Catal. A: Chem. 2004, 213, 177-182.

Scheme 1

from water solutions, Henry et al.8 have also described the preparation and characterization of polymer-based metal-ligand exchangers. In this paper we present a new method to enhance the adsorption capacity of activated carbons by the anchoring of previously designed molecular receptors of metal ions on the carbon surface. A particular feature of the proposed method is the highly irreversible character of the adsorption of the receptors on the carbon which is analyzed on the basis of the electronic properties of both the adsorbent carbon and the receptors. The molecular receptors used in this work consist of an aromatic, Ar, (pyrimidine) residue and a Lewis basic function, F, both connected through a polymethylene aliphatic moiety, S, which lacks relevant acid-base properties (Scheme 1). The function of the aromatic pyrimidine moiety, Ar, is to serve as the anchor that maintains the receptor bonded to the carbon surface. In a polar medium such as water, the Lewis basic function, F, would remain extended into the solvent, acting as a reactive center to metal ions. Obviously the efficiency of the resulting system would be dependent on an efficient design of the F moiety and on the prevalence of a mechanism implying π-π interactions between the aromatic moiety of the receptors and the carbon when the former are adsorbed on this. Besides, because these systems are useful for the sequestering of metal ions, it is necessary that the adsorption of the receptors onto the carbon surface be an irreversible process. (8) Henry, W. D.; Zhao, D.; SenGupta, A. K.; Lange, C. React. Funct. Polym. 2004, 60, 109-120.

10.1021/la050549h CCC: $30.25 © 2005 American Chemical Society Published on Web 06/14/2005

Ligand Adsorption on an Activated Carbon

With this background in mind, we have studied the adsorption of the compounds N-(4-amino-1,6-dihydro-1methyl-5-nitroso-6-oxo-2-pyrimidinyl)-L-lysine, compound AMNLY, and N-(4-amino-1,6-dihydro-1-methyl-5-nitroso6-oxopyrimidin-2-yl)-N′-[bis(2-aminoethyl)]ethylenediamine, compound AMNET, in aqueous solution on a commercial activated carbon. We have selected a basic activated carbon with low oxygen and nitrogen contents. These features would certainly enhances the π-electron density in the graphene layers, so favoring the adsorptive capacity of the receptors, via π-π dispersive and/or donoracceptor interactions9-12 between the pyrimidine moiety of the receptors and the basic arene centers of the graphene layers. Otherwise, given the dimensions of the selected ligands (see below), a mesoporous textural character of the selected carbon was desirable in order to enhance the availability of active carbon sites. In addition, as a basic attempt to prove the efficiency of the designed compounds/carbon systems, we have done a previous evaluation of their ability for the adsorption of chromate(VI) anion. Experimental Section Compounds AMNLY and AMNET were obtained by a general procedure previously reported,13 and structurally characterized by conventional spectroscopic and analytical data. These structures were solved also by single-crystal X-ray diffraction data.14,15 Complete optimizations of the geometries of the compounds AMNLY and AMNET have been done using MM+ (HYPERCHEM)16 and MM2 (ChemBats3D, CambridgeSoft Corp.).17 All the optimizations have been done with the conformation of the molecules corresponding to the global minimum. To obtain qualitative explanation of the adsorption mechanism of the compounds, an extended Hu¨ckel Molecular Orbital (EHMO) analysis was performed by the HYPER-CHEM program.16 These calculations were performed on model compounds with idealized geometries. Three-dimensional graphics of the molecules of compounds AMNLY and AMNET have been prepared by the computer program Chem 3D.18 The speciation diagrams for the ligands as a pH function in aqueous solution were obtained. For this, 10-3 M acidic (HCl) aqueous solutions of AMNLY and AMNET compounds, with 0.1 M KCl ionic strength, were titrated with a KOH standardized solution. Each of the two experiments was repeated for three times, and the potentiometric data were then processed with Hyperquad and Best programs.19,20 The details of the experimental assembly used and calculation procedures have been outlined previously.21 The information supplied by this study concerns the distribution of the different protonated forms (9) Coughlin, R. W.; Ezra, F. S. Environ. Sci. Technol. 1968, 2, 291. (10) Leon y Leon, C. A.; Solar, J. M.; Calemma, V.; Radovic, L. R. Carbon 1992, 30, 797-810. (11) Radovic, L. R.; Silva, I. F.; Ume, J. I.; Mene´ndez, J. A.; Leon y Leon, C. A.; Scaroni, A. W. Carbon 1997, 35, 1339-1348. (12) Mattson, J. S.; Mark, H. B., Jr.; Malbin, M. D.; Weber, W. J., Jr.; Crittenden, J. C. Colloid Interface Sci. 1969, 31, 116-130. (13) Low, J. N.; Lo´pez, M. D.; Arranz-Mascaro´s, P.; Cobo-Domingo, J.; Godino, M. L.; Lo´pez-Garzo´n, R.; Gutierrez, M. D.; Melguizo, M.; Ferguson, G.; Glidewell, C. Acta Crystallogr. B 2000, 56, 882-892. (14) Cuesta-Martos, R.; Garcia-Martı´n, J.; Lo´pez-Garzo´n, R.; GodinoSalido, M. L.; Gutie´rrez-Valero, M. D.; Arranz-Mascaro´s, P. Book of Abstracts; SIMEC 2002; Santiago de Compostela, Spain; pp 182-183. (15) Garcı´a-Martı´n, J.; Lo´pez-Garzo´n, R.; Godino-Salido, M. L.; Cuesta-Martos, R.; Arranz-Mascaro´s, P.; Gutie´rrez-Valero, M. D. Book of Abstracts; 7th FIGIPS Meeting in Inorganic Chemistry, 2003, Lisboa, Portugal; p 290. (16) HyperChem, Release 4.5 for Windows. Hypercube, Inc., 1995. (17) Burkert, U.; Allinger, N. L. Molecular Mechanics; ACS Monograph 177; American Chemical Society: Washington, DC, 1982. (18) CS ChemBats3D Ultra, Molecular Modelling and Analysis. Cambridge Scientific Computing, Inc., 2001. (19) Gans, P.; Sabatini, A.; Vacca, A. Talanta 1996, 43, 1739-1753. (20) Martell, A. E.; Motekaitis, R. J. Determination and Use of Stability Constants; VCH: New York, 1992. (21) Godino-Salido, M. L.; Gutie´rrez-Valero, M. D.; Lo´pez-Garzo´n, R.; Moreno-Sa´nchez, J. M. Inorg. Chim. Acta 1994, 221, 177.

Langmuir, Vol. 21, No. 15, 2005 6909 (neutral, anionic or cationic) of each of the solutes in aqueous medium at different pH values. The commercial granular activated carbon, K24504014, supplied by Merck, was used as adsorbent material for this work, after being heated at 120 °C in an oven and kept in a desiccator with P4O10. Elemental analysis was performed using a LECO CHNS-932 analyzer. Apparent surface area was calculated by fitting the BET Equation to the adsorption isotherm data of N2 at 77 K. Meso and macropore volumes and external surface area were determined by mercury porosimetry with a Quantachrome, Autoscan-60 equipment up to 4200 kg cm-2. Two TPD experiments were carried out by heating a 250 mg sample of the carbon up to 1000 °C under He flow (60 mLmin-1) with a heating rate of 50 K min-1. The gases H2, CO2, CO, and H2O were analyzed following the mass 2, 12, 14, 16, 17, 18, 22, 28, and 44 with a quadrupole mass spectrometer from BALZERS model MSC-200 Thermocube. Under the heating rate used, we have assumed that secondary reactions between the CO2 evolved and the carbon active sites are negligible.22 Potentiometric titrations of carbon were carried out to determine its chemical properties for proton adsorption and to measure its point of zero charge (pHPZC). For this, a number of carbon samples containing about 50 mg each were exactly weighed into separate dry flasks. The samples were prepared by adding to each flask successively larger amounts of standardized KOH or HCl solutions. Then, the required amounts of a mother KCl aqueous solution were added to get the desired background electrolyte concentration (three different experiments were realized by keeping KCl concentration at 0.3, 0.1, and 0.003 M, respectively). The total volume of each sample was adjusted to 50 mL by adding deionized distilled water in order to keep the solution volume to sorbent weight ratio constant. The samples were N2 saturated to eliminate the influence of atmospheric carbon dioxide, by bubbling a stream of the gas during ca. 3 min after which the flasks were sealed and left to equilibrate in a shaking water bath for 72 h. at 25 ( 0.1 °C. Then the pH of the supernatant solution was measured using a Methrom 713 pH meter. To obtain the total amount of protonated or deprotonated sites, Q, the following equation was used:

Q ) 1/m(Vo + Vt) ([H]i - [OH]i - [H]e + [OH]e) where Vo and Vt are the added volumes of background electrolyte and titrant, respectively, and m is the mass of the carbon. Subscripts i and e refer to initial and equilibrium concentration of protons. Proton isotherms are then obtained graphically by plotting Q vs equilibrium pH values.23,24 Adsorption isotherms of compounds AMNLY and AMNET on the carbon, at 298 K, were obtained by adding 0.0500 g of carbon to a 100 mL plastic flask containing 50 mL of the appropriate adsorbate solution. The pH was adjusted with KOH or HCl. The suspensions were shaken until equilibrium was reached (ca. 15 days for both of the adsorbates), and the residual adsorbate concentrations were measured by ultraviolet spectroscopy at λ values of the isosbestic points at which the extinction coefficients of the adsorbate do not depend on the protonation degree. After adsorption, adsorbents were carefully filtered. The same procedure was followed for the study of the adsorption at pH 7 of K2CrO4 on the carbon and on carbon samples containing compounds AMNLY and AMNET previously adsorbed, respectively. Nevertheless, in the last two cases the amounts of aqueous adsorbate solutions used for obtaining the corresponding isotherms were 25 mL and the equilibrium time ca. 7 days. In these experiments the residual adsorbate concentration was measured by atomic absorption spectroscopy. To study the desorption process of AMNLY and AMNET, adsorbates were then desorbed from the adsorbent by 50 mL of distilled water, for 7 days at 298 K, in clean plastic flasks. The concentrations of desorbed compounds were measured by ultraviolet spectroscopy. In all of the desorption experiments the (22) Linares-Solano, A.; Salinas-Martinez de Lecea, C.; CazorlaAmoro´s, D. Energy Fuels 1990, 4, 467-474, and references therein. (23) Bandosz, T. J.; Jagiello, J.; Contescu, C.; Schwarz, J. A. Carbon 1993, 31, 1193-1202. (24) Jagiello, J.; Bandosz, T. J.; Schwarz, J. A. Carbon 1994, 32, 1026-1028.

6910

Langmuir, Vol. 21, No. 15, 2005

Garcı´a-Martı´n et al.

Scheme 2

initial pH values were adjusted at those corresponding to the previous adsorption experiments. It should be mentioned that some solution remained on the surface of the adsorbent and into the pores after the filtration. Nevertheless, the amounts of remaining solution were much smaller than the volume of distilled water used in each of the desorption experiments.

Results and Discussion Characterization of Adsorbates. As mentioned in the Experimental Section, the molecular structures of compounds AMNLY and AMNET were solved by singlecrystal X-ray diffraction data. Similarity of bond parameters of both of the molecules to that of several analogous compounds derived by condensation of their pyrimidine moiety with a series of single amino acids, which exhibit bipolar (quasi-zwitterionic) nature,13 demonstrated the highly polar nature of the aromatic moieties of compounds AMNLY and AMNET. Accordingly, the best representations of AMNLY and AMNET molecules are the bipolar structures depicted in Scheme 2. The studies on metal complexation and protonation on the pyrimidine moieties of compounds analogous to AMNLY and AMNET25-27 show that the nitrogen atom of the C5-NO group is the preferred basic position for the binding of both metal ions and protons instead the N3cyclic, which has never been observed to act as the binding site. These findings strongly support the zwitterionic character of the adsorbates studied here, which is induced by the strong electron-withdrawing character of the nitroso group attached to C5 of the pyrimidine. Otherwise, in Scheme 2 a positive charge is seen on the NH-C2-N3-C4-NH2 fragment.13 In agreement with these data one would expect in these molecules the existence of low energy empty molecular orbitals extended along this fragment of the pyrimidine. This is proved by MM.OO. calculations for AMNLY and AMNET compounds, which show the existence of such a class of empty low energy orbitals at values of -10.59 eV and -10.65 eV, respectively, not far from their corresponding HOMO orbitals which are at -11.94 eV and at - 10.96 eV, respectively. On the basis of the frontier orbital theory,28 a high interaction energy between the adsorbent carbon and adsorbates, giving rise to irreversible adsorptive processes, would take place if suitable frontier orbitals in (25) Moreno, J. M.; Arranz-Mascaro´s, P.; Lo´pez-Garzo´n, R.; Gutie´rrezValero, M. D.; Godino-Salido, M. L.; Cobo-Domingo, J. Polyhedron 1999, 18, 1635-1640. (26) Lo´pez-Garzo´n, R.; Arranz-Mascaro´s, P.; Godino-Salido, M. L.; Gutie´rrez-Valero, M. D.; Cuesta, R.; Moreno, J. M. Inorg. Chim. Acta 2003, 355, 41-48. (27) Lo´pez-Garzo´n, R.; Godino-Salido, M. L.; Arranz-Mascaro´s, P.; Fontecha-Ca´mara, M. A.; Gutie´rrez-Valero, M. D.; Cuesta, R.; Moreno, J. M.; Stoeckli-Evans, H. Inorg. Chim. Acta 2004, 357, 2007-2014. (28) Klopman, G. J. Am. Chem. Soc. 1968, 90, 223-234; Radovic, L. R., Ed. Chemistry and Physics of Carbon Series; M. Dekker: New York, 2000; Vol. 27, pp 353-376.

Figure 1. Distribution species plot as a pH function of AMNLY and AMNET in aqueous 0.1 M KCl solution at 25 °C.

both of them were available, namely either a donor/ acceptor orbital interaction of HOMOadsorbent/LUMOadsorbate type or the inverse LUMOadsorbent/HOMOadsorbate type. Having in mind the small energy differences between frontier orbitals of compounds AMNLY and AMNET (see above), it would be expected that they could provide either LUMO or HOMO orbitals for strong adsorption interactions with a suitable activated carbon. The molecular dimensions of AMNLY and AMNET were calculated, taking into account the covalent radii of the atoms.18 The resulting values were 14.4 Å, 6.2 Å and 4.0 Å for compound AMNLY and 13.7 Å, 6.2 Å, and 5.5 Å for compound AMNET. If it is assumed that pores are present in activated carbons in the form of slits,29,30 the factor determining the pore accessibility for adsorbates is just the slit-width. Thus, in the case of the adsorbates AMNLY and AMNET, not only mesopores but also a wide range of micropores (those of slit-width higher than 4 Å and 5.5 Å, respectively) would also be accessible to them. As was pointed out in the Experimental Section, protonation equilibria of the adsorbates were determined in order to better understand the adsorption processes studied (see below). The nature of the basic atom protonated in each particular protonation step was determined by visible-UV and 1H and 13C NMR spectroscopy. Distribution species plots for compounds AMNLY and AMNET are depicted in Figure 1. Compound AMNLY exists as the single neutral species in the ca. 5 to 7 pH range. It must be pointed out that in neutral molecules of AMNLY, the amino acid residue exists as zwitterion NH3+-CHR-COO-. At pH values below 5, the neutral molecule suffers two overlapping protonation steps which take place on the carboxylate group (pKa ) 2.57) and the nitroso group (pKa ) 2.32) attached to the pyrimidine moiety, respectively. For this reason, compound AMNLY is present as a cationic (monocationic, bicationic, or a mixture of both) species in aqueous solution in the pH 1-5 range. On the contrary, on raising the pH into the 7 to 13 range, successive deprotonations of the NH3+ amino acidic group (pKa ) 9.23) and NH group (29) Bansal, R. C.; Donnet, J. P.; Stoeckli, F. Active Carbon; M. Dekker: New York, 1988. (30) Albornoz, A.; Labady, M.; Lo´pez, M.; Laine, J. J. Mater. Sci. Lett. 1999, 18, 1999-2000.

Ligand Adsorption on an Activated Carbon

Langmuir, Vol. 21, No. 15, 2005 6911

Scheme 3

Table 1. Elemental Analysis and Textural Characteristics of the Carbon analysis (wt. %) C H N Oa

SBET Vmes Vmac Sext ash (m2 g-1) (cm3 g-1) (cm3 g-1) (m2 g-1) (%)

94.7 0.3 0.75 4.25 1000 a

0.14

0.24

62

0.35

Figure 2. TPD profiles for the activated carbon.

Residual.

attached to C2 of the pyrimidine (pKa ) 11.68) occur, respectively. Thus, in this pH range we found either a mixture of neutral and monoanionic forms (at lowest pH values) or mono- and bianionic forms (at the highest pH values) (see Figure 1). Unlike AMNLY, neutral molecules of AMNET are not present as single species at any pH value. This is due to the fact that the beginning of the protonation of two primary amino groups existing at the polyamine function (pKa ) 9.69 and pKa ) 8.66) overlaps with the deprotonation of NH group attached to C2 of the pyrimidine moiety (pKa ) 10.07, see Figure 1). For this reason, in the pH 9-11 range the adsorbate AMNET is present as a mixture formed by the mono- and bicationic species together with the neutral molecular species and the monoanionic species in variable amounts. At pH values higher than 11, it exists as a mixture formed by neutral and monoanionic species in which the monoanion amount increases as the pH increases. Concerning the low pH range, the protonation of C5-NO group of the pyrimidine is accomplished at a pH similar to that in the case of compound AMNLY, but the tricationic HL3+ species is formed. It is interesting to note that neither protonation of the N3 pyrimidine atom, in both of the compounds, nor that of the N atom of the tertiary amino group in compound AMNET has been observed in the pH range studied (1 to 13). The first observation is consistent with the electronic properties of these compounds discussed above. Nevertheless, protonation of C5-NO group was proved to take place by UV-visible spectroscopies. This process induces in his turn the enhancement of the electronic affinity of C2-NH, which promotes the formation of an intramolecular hydrogen bond as shown in Scheme 3. These conformational changes induced in the compound by this process were proved by 1H and 13C NMR spectroscopic methods.15 Characterization of the Activated Carbon. An activated carbon AC Merck K24504014 was used for adsorption experiments as received. Values of specific and external surfaces and meso and macropore volumes, together with ash residue and elemental analyses, are summarized in Table 1. From SBET and Sext data it is seen that although the carbon is essentially microporous, it possesses a significant proportion of mesopores and macropores (namely, with slit width g 20 Å). On the other hand, the analytical data highlight the high carbon content and that oxygen and nitrogen are the only heteroatomic components. The nature of the surface

functional groups containing oxygen was studied by the analysis of TPD curves of Figure 2. The total amount of structural oxygen obtained by this technique (3.10%) is lower than the obtained by elemental analysis (4.25%). The difference (1.15%) could be assumed to be due to the fact that in a TPD analysis up to 1273 K not all the oxygen surface groups are evolved. The low temperature at which the CO2 peak is centered, 626 K, suggests that this proceeds from the lactone better than from the carboxylic acid surface groups.31,32 The above observation is also consistent with the absence of a CO peak at the cited temperature that would appear if carboxylic adjacent groups would exist in the carbon surface. It is apparent from Figure 2 that most of the oxygen desorbs as CO. Besides, the high temperature at which the maximum is reached, 1200 K, lets us assume that most of this CO is derived from carbonyl and/or quinone groups, although the fraction desorbed at lower temperatures is probably derived from phenolic and/or hydroquinone groups whose thermal stabilities are smaller than that of the former. All these findings are consistent with the fact that the commercial carbon under study was not subjected to any oxidative treatment. These data also show a relatively higher content of basic oxygen-containing functions (carbonyl and/or quinone groups) than acidic ones (lactone and phenol and/or hydroquinone). Thus, taking also into account the high carbon content, it is expected that the predominant basic protonation sites will be the graphitic layers of the adsorbent and that they will be predominant over the acid ones. The proton binding isotherm for the adsorbent is shown in Figure 3. In this figure, a negative value of Q indicates the presence of Bro¨nsted acidic groups in the carbon whereas a positive value points out the presence of Bro¨nsted basic groups. The pHPCZ, 8.3, proves the basic (Bro¨nsted) character of the carbon surface. First, there is a negligible influence of the electrolyte concentration used (KCl) on the Q values. This implies an insignificant influence of the double-layer effect on the protonation-deprotonation equilibrium at the carbon surface and also a small affinity of electrolyte metal ion, K+, for the basic positions existing in the carbon surface. The plot of Figure 3 shows two parts: one of them, in the pH window 3.5 to 8.3, corresponding to positive values of Q with a pronounced and almost constant slope from pH 4 to 8; at pH values lower than 4, Q values seems to remain constant which indicate an apparent saturation of basic (31) Moreno-Castilla, C.; Carrasco-Marı´n, F: Mueden, A. Carbon 1997, 35, 1619-1626. (32) Otake, Y.; Jenkins, R. G. Carbon 1993, 31, 109-121.

6912

Langmuir, Vol. 21, No. 15, 2005

Figure 3. Proton binding isotherms for activated carbon sample AC Merck K24504014.

sites. Taking into account the nature of the basic groups present in the surface of the adsorbent, the most likely protonation processes are those involving basic arene centers, at which solvated protons H3O+ can be fixed by the reaction:10

CΠ + H3O+ h CΠ_H3O+ Besides, the nitrogen content (ca. 0.5 mmol g-1), existing under the form of some nitrogen basic functions (e.g. pyrrole, pyridine, and less probably as primary amino groups), likely contributes, although to a small extent, to the binding of protons on the carbon surface according to processes of the type:

C-N + H+ h C-NH+ On the other hand, the second part in the plot of Figure 3 corresponds to negative values of Q which indicate the presence of acidic groups in the sorbent surface. The maximum amount of H+ liberated from the sorbent surface into the solution is small (less than 0.2 mmol at a pH of ca.10.2), indicating the low acidity of the carbon. A characteristic feature of this part of the plot is the significant OH- consumption at a pH value above 9. It has been pointed out in some recent work that this effect is promoted by the slow hydrolysis of either ether or (as in the present case) lactone groups present in the sorbent surface.33,34 In that way each of the last mentioned groups would liberate two protons proceeding from each of the two groups formed during the hydrolytic process: the carboxylic acid (pKa ) 4-5) and phenol (pKa ) 8-11)35 groups. Analysis of the Adsorption. Adsorption of Compounds AMNLY and AMNET. Adsorption isotherms of compounds AMNLY and AMNET at several pH values are shown in Figure 4. The shapes of all the adsorption curves belong to type L according to Giles’s classification,36 i.e. the molecules AMNLY and AMNET most likely are adsorbed flat. This is typical in the case of aromatic molecules over a graphitic carbon. The measured isotherms fit to Langmuir type functions, although it should be mentioned that owing to the heterogeneity of the carbon surface and to the wide (33) Contescu, A.; Contescu, C.; Putyera, K.; Schwarz, J. A. Carbon 1997, 35, 83-94. (34) Laszlo, K.; Tombacz, E.; Kerepesi, P. Colloid. Surface A 2003, 230, 13-22. (35) Smith B.; March J. Advanced Organic Chemistry, 5th ed.; Wiley: New York, 2001. (36) Gilles, C. H.; MacEwan, T. H.; Nakhwa, S. N.; Smith, D. J. Chem. Soc. 1960, 3973-3993.

Garcı´a-Martı´n et al.

speciation of the adsorbates, the fundamental conditions of the Langmuir are not properly fulfilled. So, the R2 of the fitting varies from 0.8236 to 0.9809. Once assumed that most of the molecules of the adsorbates are anchored flat onto the carbon surface, the theoretical area covered by each of them can be estimated from dimensional data corresponding to standard conformation of AMNLY and AMNETmolecules. The corresponding values are 8.928 × 10-19 m2 molec-1 and 8.494 × 10-19 m2 molec-1 for AMNLY and AMNET, respectively. The maximum surface area covered by the adsorbates at each of the pH adsorption experiments can be calculated from these data together with the X parameters obtained in his turn, by fitting the corresponding isotherms to Langmuir equation (X maximum adsorption capacity in mmol per gram of carbon). The values obtained range from a minimum value of 425.35 m2, at pH 1, to a maximum of 550.10 m2, at pH 10, for compound AMNLY, and from 238.80 m2, at pH 1, to 628.75 m2 for compound AMNET, at pH 10 (Figure 4), all of which largely exceed the meso and macropore surface. Mu¨ller et al. have developed a theoretical model for the description of the adsorption of weak organic electrolytes on activated carbon.37 This model states that the carbon affinity for an electrolyte is determined by a charge surface potential ψo, related to the surface charge, and a nonelectrostatic potential, U. Of course, for a heterogeneous surface formed by the sum of a certain number of homogeneous patches, each patch has a characteristic adsorption potential, Ui. Based on this model, a qualitative explanation of the changes observed in the adsorption isotherms of compounds AMNLY and AMNET can be made by assuming that the main adsorption mechanism is through π-π interactions. The lowest adsorption capacity of Langmuir for compound AMNLY is found at pH 1 (X equal to 0.791 mmol g-1), at which almost all of the solute (most existing as bication) is coionic with the carbon surface which is positively charged due to the H3O+ groups bonded at the arene centers (for the main anchoring positions for the adsorbate molecules, see above). On the contrary, at pH values 4 and 7 the whole of the adsorbate molecules of AMNLY exists as neutral species. This explains the observed increase in X values up to 1.006 mmol g-1, at pH 4, and 1.020 mmol g-1, at pH 7. Despite the decreasing of the positive charge of the carbon on going from pH 4 to pH 7, the small increase observed in X can be explained because such charge decreasing implies the dissociation of H3O+ groups from the arene centers which favors the anchoring of the adsorbate via π-π interactions, at such centers. At pH 10, almost all of the solute is found as monoanion, e.g. it is coionic to the surface charge (see Figures 2 and 4) and a small increase in X (1.023 mmol g-1) related to pH 7 occurs. One would expect that a repulsive electrostatic potential determines an effect opposite to the observed. Nevertheless, the repulsive interactions between negative charges are probably limited by the fact that negative charges of the carbon are likely located at the edges of the graphite layers far away from the arene centers where the adsorbent-adsorbate interaction occurs. Moreover, it is also probable that the negative charges of carbon surface are located at the oxygen atoms existing at the edges of the graphene layers and that they act as electron-donating groups to the arene centers, so favoring π-π interactions between the carbon and the adsorbate molecules. (37) Mu¨ller, G.; Radke, C. J.; Prausnitz, J. M. J. Phys. Chem. 1980, 84, 369-376.

Ligand Adsorption on an Activated Carbon

Langmuir, Vol. 21, No. 15, 2005 6913

Figure 4. Adsorption equilibrium data at 25 °C and variable pH for (a) compound AMNET and (b) compound AMNLY.

Compound AMNET exists as a coion to the carbon surface at pH values of 1 (trication), 4 and 7 (bication). This explains its smaller adsorption compared to compound AMNLY which exists as coion at pH 1 (monocation) and as neutral molecule at pH 4 and 7. The small increase in maximum adsorption values upon increasing the pH from 1 (X ) 0.467 mmol g-1) to 4 and 7 (X ) 0.551 mmol g-1 and 0.580 mmol g-1, respectively) is due to a decrease in the positive charge of the carbon surface (see Figure 4). A sharp growth in X is observed at pH 9 (X ) 1.119) due to the unblocking of the arene centers at the negatively charged surface. Moreover, at this pH most of the adsorbate is maintained in cationic form; see Figure 1. At pH 10, compound AMNET exists as a mixture of neutral and monocationic and monoanionic forms, that is to say a part of it is now coion. Despite this, the X value (1.229 mmol g-1) is now higher than at pH 9. Also this value is higher than that of compound AMNLY at the same pH. At this point, an important difference between monoanionic forms of AMNLY and AMNET compounds must be emphasized, that whereas the negative charge of AMNET in monoanionic form is spread at the pyrimidine moiety, that of AMNLY is spread at the carboxylate group of the F function, far away from the anchoring moiety. This fact strongly suggests an important role of the basicity of both the arene centers and the pyrimidine moiety in the π-π interactions during the adsorption process. On the basis of the data discussed above, we observed the nature of the interaction in the processes studied. On the basis of the two-state adsorption model,38 irreversible adsorption has been related, among other causes (e.g. the kinetic hindrance to adsorbate diffusion through the pores), to donor-acceptor interactions between the frontiers orbitals of the adsorbent and adsorbate. To analyze this possibility in the systems studied here, we carried out an OO.MM. calculation of idealized graphitic media in accordance with the structural characteristics of the carbon used in this work16 (see Figure 5) in order to compare their energies to the OO.MM. of the adsorbates. HOMO-LUMO donor-acceptor interactions between the frontier orbitals of both of the reactants, so favoring an irreversible interaction, would be expected if those orbitals (with local π symmetry) are not very different in energy. For a wide series of aromatic adsorbates, Tamon and Okazaki39 have reported that irreversible adsorption occurs when the energy difference between the HOMO orbital of the adsorbate and the LUMO of the adsorbent is lower than 7.34 eV. Table 2 summarized the energies of the HOMO and LUMO orbitals of the cluster models (38) Atkins, P. W. Physical Chemistry, 3th ed.; Oxford University Press: Oxford, 1986. (39) Tamon, H.; Okazaki, M. J. Colloid Interface Sci. 1996, 179, 181187.

Figure 5. Cluster models of adsorbent surface. Table 2. Energies of HOMO and LUMO of Adsorbent Model (eV) character model

HOMO

LUMO

C28 C28O C28Lactone C28Quino AMNLY AMNET

-10.95 -10.91 -10.95 -11.19 -11.94 -10.96

-10.14 -10.39 -10.14 -10.70 -10.59 -10.65

of adsorbent and those of AMNLY and AMNET compounds in neutral molecular forms. It is seen that the energy differences between the HOMO orbitals of the idealized clusters of the carbon and the LUMO of both of the adsorbates are much lower than the above value. Otherwise, the same is observed for the energy differences between the HOMO orbitals of the adsorbates and the LUMO of the model clusters. Thus, it is probable that π-π interactions in the system studied in this work are not only dispersive but also have a probable donoracceptor contribution. This assumption is reinforced by the fact that the adsorption processes of both of the compounds are highly irreversible independent of the pH studied, as the percent of desorbed compounds at all the studied pHs oscillate between 0% and ca. 5% in most of the cases. Otherwise, the last fact points out that π-π interactions are the predominant adsorption mechanism independent of the pH, namely of the nature of the charge of carbon and adsorbates. Adsorption of K2CrO4. To prove the effectiveness of the developed models as ionic receptors, we have obtained the adsorption isotherms of CrO42- in aqueous solution on the AC used in this work and on the carbon with

6914

Langmuir, Vol. 21, No. 15, 2005

Figure 6. Adsorption isotherms of CrO42- in aqueous solution on the AC and on the carbon with compounds AMNLY and AMNET previously supported on it.

compounds AMNLY and AMNET previously supported on it, respectively (Figure 6). For the experiments with carbon samples containing compounds AMNLY and AMNET, we used samples derived from desorption experiments which were carried out at pH 7 (see above). The adsorption isotherms were obtained at pH 7 at which almost all of the adsorbate is found as bianion,40 and it is able to interact with protonated primary amino groups of compounds AMNLY and AMNET not only by electrostatic interactions but also via N-H+‚‚‚O-chrom hydrogen bonding contact. Otherwise, pH 7 is adequate to avoid reduction of Cr(VI) to Cr(III), which has been reported to be optimum around pH 5,41,42 and ensures that during the adsorption experiments the sole species adsorbed was CrO42-. The adsorption isotherm of CrO42on the AC used in this work fits well with Langmuir equation (R2 ) 0.9554), and from this, a maximum adsorptive capacity of 0.074 mmol g-1 was found. Having in mind the bianionic nature of the adsorbate and the absence of hydrogen-bond forming groups at the edges of (40) Martell, A. E.; Smith, R. M.; Motekaitis, R. J. NIST Critical Stability Constants of Metal Complexes Database; Version 7, Texas A & M University: College Station, TX, 2003. (41) Bautista-Toledo, I.; Rivera-Utrilla, J.; Ferro-Garcı´a, M. A.; Moreno-Castilla, C. Carbon 1994, 32, 93-100. (42) Park, S.-J.; Jang, Y.-S. J. Colloid Interface Sci. 2002, 249, 458463. (43) Lalvani, S. B.; Wiltowski, T.; Hu¨bner, A.; Webstone, A.; Mandich, N. Carbon 1998, 36, 1219-1225.

Garcı´a-Martı´n et al.

the graphitic planes, it would be assumed to be the positive charges at the arene centers, existing at pH 7 (see above), which facilitate the anion uptake by the carbon.37 The adsorption of Cr(VI) on the compound AMNET supported on the AC also followed the Langmuir isotherm (R2 ) 0.9931) and the maximum adsorption capacity, 0.125 mmol g-1, is clearly higher than that of AC. The enhancement in the adsorptive capacity of the AC with compound AMNET can be explained by a stronger and selective interaction of CrO42- anions with the NH3+ functions existing in this compound once firmly anchored at the arene centers. In fact, from a potentiometric study on the speciation in a CrO42-/AMNET mixture in aqueous solution (25 °C, 0.1 M KCl ionic strength, and [K2CrO4] ) [AMNET] ) 10-3 M), the formation of a neutral [CrO42-AMNET] adduct at pH 7, whose log K is 4.40, was proved. A solid species of this composition was precipitated from the medium at this pH, in which the chromate anion is coordinated to one or two of the NH3+ in a bidentate manner. The opposite behavior is observed for compound AMNLY, which exhibits a negligible adsorptive capacity compared to the AC (see Figure 6). Compound AMNLY, like compound AMNET, also blocks the arene centers (the uptake positions of the chromate anions when adsorbed on the AC). Nevertheless, in the case of compound AMNLY, the interaction of the chromate anion with the only available function, the NH3+ group existing at the amino acid moiety, is probably prevented due to the proximity of a carboxylate center that carries a negative charge. We have obtained some other previous (yet unpublished) results in adsorption studies of other metal ions on the carbon-receptor systems described in this work. They also point out an improvement of the adsorptive capacity of the AC when AMNLY and AMNET are adsorbed on it that can be related to each of their specific F functions. Nevertheless, although the promising results demonstrated are quite limited, a better understanding of the proposed model would require further studies based on extending the experiments to a wide range of this class of metal ion receptors and also to other activated carbons and graphite with different chemical functionalization. Acknowledgment. The authors gratefully acknowledge financial support by the Spanish Ministerio de Ciencia y Tecnologı´a (Proyecto PPQ2000-1667) and by Junta de Andalucı´a (FQM-273). J.G.-M. acknowledges the Junta de Andalucı´a for a grant. LA050549H