DALTONFULL PAPERJ. Chem. Soc., Dalton Trans., 1999, 2427¡V2432 2427Complexation of aluminium(III) with 3-hydroxy-2(1H)-pyridinone.Solution state study and crystal structure of tris(3-hydroxy-2(1H)-pyridinonato)aluminium(III)Valerio B. Di Marco,a G. Giorgio Bombi,*a Andrea Tapparo,b Annie K. Powell c andChristopher E. Anson ca Universit degli Studi di Padova, Dipartimento di Chimica Inorganica,Metallorganica ed Analitica, via Marzolo 1, 35131 Padova, Italy. E-mail: g.g.bombi@unipd.itb Universit degli Studi di Sassari, Dipartimento di Chimica, via Vienna 2, 07100 Sassari, Italyc School of Chemical Sciences, University of East Anglia, Norwich, UK NR4 7TJReceived 15th April 1999, Accepted 11th June 1999The formation of complexes between aluminium() and 3-hydroxy-2(1H)-pyridinone (HL) in aqueous 0.6 m (Na)Clat 25 C has been investigated by means of potentiometric titrations.The following complex stability constantshave been evaluated (pKa = 8.590 ¡Ó 0.008): log £]AlL = 8.59 ¡Ó 0.01, log £]AlL2 = 16.34 ¡Ó 0.03, log £]AlL3 = 23.11 ¡Ó 0.05,log £]AlL3H1 = 13.85 ¡Ó 0.04.The qualitative and quantitative results obtained have been conrmed in part by UVspectrophotometry and 1H NMR spectroscopy. Some potentiometric titrations were executed at 37 C as well, andthe following stability constants were obtained (pKa = 8.452 ¡Ó 0.004): log £]AlL = 8.19 ¡Ó 0.02, log £]AlL2 = 16.03 ¡Ó 0.04,log £]AlL3 = 21.77 ¡Ó 0.08, and log £]AlL3H1 = 13.0 ¡Ó 0.2.Crystals of the complex AlL3 were obtained and analysed byX-ray diraction. The neutral species is an octahedral six-co-ordinate complex with the ligand chelating in abidentate fashion through the pyridinone oxygen and the deprotonated hydroxylic group.IntroductionOver the last 20¡V30 years the mainstay of aluminium (and iron)chelation therapy has been Desferal (desferrioxamine mesylate).1 Despite its good prognosis the general use of Desferal isrestricted because of its several drawbacks and toxic sideeects.1,2 For this reason, a number of chelators have beentested in vitro and in animals for the replacement of Desferalwith a more suitable chelating drug;1¡V4 these studies necessarilyhave to be accompanied by accurate chemical investigations,in order to determine the thermodynamic and kinetic propertiesof likely compounds of the metal under physiologicalconditions.Hydroxypyridinones have been extensively tested, and sometimesalso used, as alternatives to Desferal.1,2 For somecompounds of this class, especially for 1,2-dialkyl-3-hydroxy-4(1H)-pyridinones, much thermodynamic data for aluminiumcomplexes in aqueous solutions have been collected,5,6 whereasfor other ligands, like 3-hydroxy-2(1H)-pyridinones, whichare not used to the same extent as the 3-hydroxy-4(1H)-pyridinones,1,7 these studies are less systematic.7In the present study the stability constants for aluminiumcomplexes of 3-hydroxy-2(1H)-pyridinone, hereafter namedHL, have been determined.The thermodynamic properties ofits aluminium complexes in aqueous solutions have not yet beenexamined. The study has been conducted at 25 C, in order toallow the direct comparison with thermodynamic data for otherhydroxypyridinones evaluated at this temperature, and at 37 Cto investigate how the stability constants vary with temperatureunder physiological conditions. The results obtained frompotentiometric measurements at 25 C have been checked usingtwo independent techniques, UV spectrophotometry and NMRspectroscopy; in the case of the complex AlL3, solid-state data(elemental analysis and X-ray diraction) were also obtained.N HO HOHLExperimentalApparatus, reagents and measurement methods were similarto those reported previously,8 and the following summaryindicates where details dier.ApparatusPotentiometric measurements were performed with a RadiometerABU93 Triburette apparatus equipped with 1, 5 and10 mL burettes and with two independent potentiometricchannels. The UV spectra were recorded with a Perkin-ElmerLambda 5 instrument and 1H NMR spectra with Bruker 200AC and AM 400 spectrometers.ReagentsAll analyte concentrations were expressed in the molality scale(mol kg1 of water).For the potentiometric titrations, standardsolutions of HCl (ca. 0.1 m), AlCl3 (ca. 0.05 m), NaOH (ca.0.1 m) and ligand were used; 3-hydroxy-2(1H)-pyridinone(Acros, nominal purity 98%) was used as received to prepare a0.009 m (0.01 m HCl) working solution.Solutions for UV and1H NMR measurements were prepared by dissolving in water(H2O and D2O, respectively) the correct amounts of the ligandand/or AlCl3.Potentiometric measurementsThe measurements were carried out in a 200 mL water-jacketedcell, and duplicate potentiometric measurements obtainedby using an Ag¡VAgCl¡V3 M KCl reference electrode (BDH309/1030/06) and two dierent glass electrodes (RadiometerpHG201 and BDH 309/1015/02); titrations were executed at25.00 ¡Ó 0.05 and at 37.00 ¡Ó 0.05 C in aqueous 0.6 m (Na)Cl.Titrations of the ligand in the absence of AlIII wereperformed to determine its acid¡Vbase properties and to checkits exact titre.Ligand concentrations ranged from 1.90 ¡Ñ 104to 2.21 ¡Ñ 103 m; the pH range was from 2.5 to 11.In the titrations in the presence of both ligand and AlIII,concentrations ranged from 2.82 ¡Ñ 104 to 2.78 ¡Ñ 103 m for the2428 J. Chem. Soc., Dalton Trans., 1999, 2427.2432 ligand and from 1.68 ¡¿ 10 4 to 1.29 ¡¿ 10 3 m for the metal; the ligand : metal ratio varied from 8: 1 to 1 : 2; the pH range was from 2.5 to 11.The potentiometric study of aqueous solutions containing aluminium and HL has been partially complicated by the low water solubility of the neutral AlL3 complex, which precipitates at pH . 4.5.6.5, depending on the initial aluminium and ligand concentrations, and redissolves at pH > 9. To avoid the presence of solid, titrations had to be stopped at acidic or slightly acidic pH values; otherwise they had to be performed at a concentration of aluminium lower than the solubility of AlL3 (ca. 2 ¡¿ 10 4 m). The ligand protonation constants and the metal.ligand complex stability constants were calculated using the computer program PITMAP.9 The values of the formation constants of aluminium hydroxo-complexes at 25 C and in 0.6 M NaCl have been taken from the literature 10 and were held constant during data optimisation.UV measurements Spectra were collected at various pH values at 25 C for solutions containing aluminium (ca. 10 2 m), ligand (ca. 10 3 m) and 0.6 m (Na)Cl; the concentrations of AlIII and ligand and the pH interval were chosen so that only two absorbing species, AlL and HL (charges omitted), were present in solution at signi.cant concentrations, as predicted from the equilibrium constants previously obtained from potentiometric data; under these conditions only the equilibrium (1) needs to be HL Al AlL H K = [AlL][H]/[HL][Al] (1) considered.The absorbance di.erence between AlL and HL is su.ciently large in the wavelength range 200.325 nm to allow the value of K to be determined by .tting the experimental points (absorbances vs. pH at a given wavelength) by the theoretical equation obtained by combining the above mass-law expression with the mass balance equations for the metal and the ligand; the only unknown parameters of the equation are the equilibrium constant K and the absorption coe.cients of HL and AlL. 1H NMR measurements Spectra were obtained for D2O solutions containing the ligand alone (10 2 m) and for these also containing aluminium (3 ¡¿ 10 3 m) at various pH values at 25 C. Only the spectrum at the neutral pH value was obtained with a 400 MHz instrument (instead of 200 MHz) in order to allow the detection of the species AlL3; in this case, after the addition of the ligand and the metal and the adjustment of the pH value, a brown precipitate was formed; the NMR spectrum of this solution was collected after .ltration and a subsequent small addition of D2O in order to prevent further precipitation during the measurement.In all cases the pH readings were corrected by adding 0.41 pH units 11 to allow for isotopic and solvent e.ects caused by the substitution of normal water (calibration environment) with heavy water (measurement environment). Preparation of solid AlL3 The compound HL (3 mmol), 3 mmol of KOH (Fluka) and 1 mmol of Al(NO3)3 9H2O (Prolabo) were dissolved in 50 mL of water at 60 C (pH . 4) under moderate stirring. The hazel-brown powder obtained from the solution was washed with water and dried under vacuum (269 mg, 75%). Elemental analysis (expected value): C, 48.52 (50.43%); H, 3.58 (3.39%); N, 11.59 (11.76%). From this raw material useful crystals for XRD analysis could not be obtained. The crystallisation of AlL3 was therefore performed in a di.erent way.Compound HL (3 mmol) and 1 mmol of AlCl3 6H2O were dissolved in 50 mL water at room temperature; this acidic solution (pH . 2) was brought to pH . 10 using NaOH. The slow neutralisation of this clear, brown solution by atmospheric CO2 (about one month, room temperature) gave the complex in the form of brown crystals. No elemental analysis could be done on these crystals due to their small quantity. Crystal analysis Crystal data were collected on a Rigaku/MSC Raxis II imaging plate system (Mo-K¥á, ¥ë = 0.71073 A) on a single crystal of ca. 0.3 ¡¿ 0.3 ¡¿ 0.3 mm in size. Some experimental details are reported in Table 4. CCDC reference number 186/1514. See http://www.rsc.org/suppdata/dt/1999/2427/ for crystallographic .les in .cif format. Results and discussion Potentiometric results As a check of the accuracy of the whole experimental system the pKw value for water in 0.6 m (Na)Cl was computed from HCl NaOH titrations at 25 C.The value obtained from seven experiments (pKw = 13.714 ¡¾ 0.002) compares well with the literature value 12 in 0.6 M NaCl at 25 C (13.727 ¡¾ 0.001). A value for pKw has also been obtained at 37 C from twelve experiments (13.352 ¡¾ 0.002), which is in a good agreement with the calculated value, 13.355, obtained from tabulated values of pKw and .H0 at 25 C13 by applying the van¡�t Ho. equation. The pKa values of free HL at the two investigated temperatures are given in Table 1, together with other thermodynamic parameters; the deprotonation occurs at the phenolic oxygen.14 Reasonable similar pKa values are reported in the literature [8.694 ¡¾ 0.007 in 0.1 M KCl at 25 C,14 9.00 ¡¾ 0.01 at 20 C (ionic strength not speci.ed),15 8.66 ¡¾ 0.01 at 25 C and ionic strength 0.1 M 16].The deprotonation of the oxy-group, i.e. of the species H2L , has occasionally been detected (pKa about 0.1.0.2 14,15); the deprotonation of the species L at the pyridinic nitrogen, which has a signi.cant amidic character, is not measurable in water 14 (pKa > 13).In the present study of metal.ligand complexes, the interpretation of potentiometric data was started by plotting n. L,M vs. log[L] curves. If predominantly mononuclear AlLn complexes are formed in solution the quantity n. L,M is the average number of L co-ordinated per Al3 ,17 and the n. L,M curves are coincident. This was found in the present case (Fig. 1), with a limiting value of n.L,M larger than 2, even if, at low n. L,M values, some small di.erences of the curves could support the existence of other, protonated or polynuclear, species. It was noticed that these di.erences are not correlated to modi.cations either of aluminium and ligand concentration or of their ratio, i.e. they seem to be only due to experimental uncertainties. In any case, the experimental low-pH data were carefully reanalysed, see later. The complete computer treatment of experimental titration data gives the stoichiometries and stability constants of the aluminium.ligand complexes reported in Table 2.Table 1 Acidic properties of HL in aqueous 0.6 m NaCl at 25 and 37 C; .G = 49.03 ¡¾ 0.05 kJ mol 1, .H = 20 ¡¾ 2 kJ mol 1, .S = 97 ¡¾ 6 J mol 1 at 25 C 25 C 37 C pKa n pKa n 8.590 ¡¾ 0.008 22 8.452 ¡¾ 0.004 24 a n is the number of titrations from which the data were obtained; the reported uncertainty is the standard deviation of the mean calculated from the n results.J. Chem.Soc., Dalton Trans., 1999, 2427�C2432 2429 The logarithmic distribution diagram of most important aluminium species at concentrations typical for the potentiometric measurements is in Fig. 2 (25 C). The main aluminium complexes in solution are AlL, AlL2 and AlL3. An estimate of the solubility product of AlL3 was evaluated from the pAl and pL values obtained from the distribution diagram at the pH value corresponding to the observed start of AlL3 precipitation: pKs (AlL3) = 26.58 ¡À 0.07 (mean of 5 values, 25 C).At an initial aluminium concentration about 2 ¡Á 10 4 m or lower and at ligand : metal ratio 3 : 1 alkaline pH values could be reached without the occurrence of AlL3 or Al(OH)3 precipitation. Under these conditions another species could be detected in solution, AlL3H 1, which is the deprotonation product of AlL3 at the pyridinic nitrogen, with a pKa of 9.26 at 25 C (log ¦Â1,3,0 log ¦Â1,3, 1); this value is reasonable, because for this species there can be a signicant resonance formula which delocalises the positive charge from the nitrogen to the ortho-oxygen.In fact, the pKa of AlL3 is a compromise of the value typical of a pyridinic proton (pKa ¡Ö 5) and that of an amidic proton (pKa > 13). Other possible deprotonation products, like AlLH 1, AlL2H 1, AlL3H 2 and AlL3H 3, could not be detected: in the rst two cases the attachment of another ligand to the metal centre is favoured, whereas formation of last two species is likely to occur only at more alkaline pH values, where however only Al(OH)4 was found to exist.A careful investigation of the experimental low-pH data was Fig. 1 Experimental data from the AlIII�CHL system (25 C) plotted as n¡¥ L,M vs. log [L] curves at various ligand and metal concentrations. also executed, in order to verify whether the observed small dierences in the starting parts of the n¡¥ L,M curves were due to the presence of polynuclear or deprotonated species. No complexes except AlL could be detected.The increase of the temperature (from 25 to 37 C) causes a decrease of the stability constants of all complexes. The H and S values could be obtained from the Van¡�t Ho equation; they are however very imprecise (and not reported in Table 2), because of the small dierence between the two investigated temperatures. UV results The UV spectra for solutions containing known concentrations of aluminium and ligand at various pH values are given in Fig. 3, and the value obtained for log K (reaction (1), see Experimental section) was: 0.00 ¡À 0.05. This value has to be compared with the potentiometric one at 25 C [0.00 ¡À 0.02, obtained by combining pKa (Table 1) with log ¦Â1,1,0 (Table 2)]; the excellent agreement suggests the absence of any bias in the results. 1H NMR results The 1H NMR spectra of D2O solutions containing aluminium and ligand, at various pH values at 25 C, are reported in Fig. 4. In addition to the strong signals of the ¡°free¡± ligand at ¦Ä 7.1�C7.2 and 6.4�C6.55, at pH 2.5 and 2.9 two new groups of peaks at ¦Ä 6.95�C7.1 and at 6.7�C6.85 are observed. These signals (labelled with ¡°1¡± and ¡°2¡± respectively) can be attributed to the pyridinic protons of two (and probably not more than two) complexes, which should be AlL and AlL2 according to the potentiometric data. There are two reasons to attribute peaks ¡°1¡± to AlL and peaks ¡°2¡± to AlL2.(1) In the spectrum at pH 2.9 signal ¡°2¡± becomes more intense with respect to signal ¡°1&as predicted Table 2 Results of potentiometric study of complex formation between Al3 and HL in aqueous 0.6 m NaCl at 25 and 37 C (reactions: m Al3 lL hH AlmLlHh 3m l h 25 C 37 C m,l,h log ¦Â n log ¦Â n 1,1,0 1,2,0 1,3,0 1,3, 1 8.59 ¡À 0.01 16.34 ¡À 0.03 23.11 ¡À 0.05 13.85 ¡À 0.04 26 24 20 6 8.19 ¡À 0.02 16.03 ¡À 0.04 21.77 ¡À 0.08 13.0 ¡À 0.2 9 10 46 Fig. 2 Logarithmic distribution diagram of most important aluminium species in the presence of HL (aqueous 0.6 m NaCl, T = 25 C, [Al]0 = 2 ¡Á 10 4 m, [HL]0 = 10 3 m; pKs of amorphous Al(OH)3 = 10.8, pKs of AlL3 = 26.58).2430 J. Chem. Soc., Dalton Trans., 1999, 2427.2432 by potentiometric results. (2) The peaks labelled with ¡°1¡± are narrow, whereas those labelled with ¡°2¡± are broader; this fact is likely to be caused by the presence of isomers (for AlL there is Fig. 3 The UV spectra for solutions containing aluminium and HL (aqueous 0.6 m NaCl, 25 C, [Al]0 = 9.95 ¡¿ 10 3 m, [HL]0 = 1.80 ¡¿ 10 3 m, pH 1.25, 1.64, 2.00, 2.36, 2.77, 3.11, 3.50 and 3.89); cell length = 0.1 cm. Calculations were performed at ¥ë = 206, 227, 247, 268.5, 300 and 323 nm. Fig. 4 The 1H NMR spectra in D2O, 0.6 m NaCl at 25 C of a solution containing aluminium and HL ([Al]0 = 3 ¡¿ 10 3 m, [HL]0 = 10 2 m, pH 2.5, 2.9 and 6.8 from top to bottom). only one isomer, whereas for AlL2 there can be up to 8 isomers simultaneously present in solution), which are identical in potentiometric titrations, but can be (and in fact they are) di.erent in the NMR analysis.It is also probable that these isomers interchange ligand molecules with slower rates than before, because the peaks of the ¡°free¡± ligand at pH 2.9 are slightly broader than the corresponding ones at pH 2.5. The integration of the signals gives the relative amount of ¡°free¡± and complexed ligand; the values obtained are reported in Table 3 together with the corresponding values calculated from the potentiometric results.The agreement between the two sets of data is reasonably good; the di.erences can be attributed to isotopic and solvent e.ects introduced by using D2O instead of H2O. The analysis of the spectrum at pH 6.8 suggests the presence of only one complex, the signal pattern of which is di.erent from those of AlL and AlL2. According to the potentiometric data this complex should be AlL3.Crystal structure analysis The structure of the complex AlL3 is shown in Fig. 5. Bond distances and interbond angles are reported in Table 5. Initial re.nement with Al(1), O(1) and O(2) anisotropic, and the six atoms of the ring as isotropic carbons, resulted in a lower thermal parameter for atom N(2) than for C(5) (U = 0.0424 and 0.0581 A3). Atoms C(1) and C(6) have very similar thermal parameters, as do C(3) and C(4). Accordingly, the nitrogen atom in the ring is identi.ed as N(2).An attempt to re.ne N(2) and C(5) as partially disordered nitrogen and carbon atoms found no signi.cant evidence for disorder. This is entirely consistent with O(1) being the ketonic oxygen of the parent ligand, with C.O and O.Al distances of 1.285(6) and 1.915(3) A, respectively, and O(2) being derived from the hydroxyl oxygen, with C.O and O.Al distances of 1.317(6) and 1.899(3) A, respectively. The six ring atoms were then re.ned anisotropically.It should be noted that the partial ketonic character of the C(1).O(1) bond was also suggested by comparing the pKa values of AlL3, HL and pyridinic protons (see potentiometric results). Although the acentric space group chosen, R3c, is racemic, with alternate molecules of opposite handedness in each stack (parallel to the c axis), it is a polar space group, and here it is to the polarity of the structure to which the Flack asymmetry parameter refers.With only one aluminium atom as a ¡°heavy¡± atom in the molecule, it was likely Fig. 5 Crystal structure of AlL3. Table 3 Percentages of ¡°free¡± and complexed ligand pH From NMR data From potentiometric results 2.5 2.9 ¡°Free¡± ligand Complexed ligand ¡°Free¡± ligand Complexed ligand 73.7 26.3 66.7 33.3 HL AlL 2AlL2 HL AlL 2AlL2 71.3 28.7 62.1 37.9J. Chem. Soc., Dalton Trans., 1999, 2427–2432 2431 that this structure would prove to be a borderline case as to whether the polarity could be determined reliably.This was indeed the case; the .nal value for ., 0.44(0.51), di.ers from 1 by just under 3s. Inverting the structure inevitably results in a value for . greater than unity. While the polarity of the structure has not quite been established, that chosen is much the more likely. The possible presence of twinning was investigated using the appropriate TWIN and BASF command lines in SHELXTL.18 From an initial value of 0.5, BASF re.ned to zero, suggesting that no twinning was present.The model was further tested with reference to the structure of the analogous iron complex reported by Scarrow et al.,16 in which they assumed complete disorder of the ligands, selecting the space group R3c. The present structure was therefore tested in that space group, but the thermal parameters for some of the atoms became unreasonable. Therefore, in contrast to Scarrow et al., we believe that our aluminium complex crystallises in R3c, with no detectable disorder in the ligand.Conclusion The ligand HL forms very stable complexes with aluminium, and can inhibit the formation of hydroxo-complexes of the metal and the precipitation of Al(OH)3 even at neutral and alkaline pH values. Its high a.nity towards aluminium is due to the signi.cant acidity of the phenolic group and to the high partial negative charge of the chelating oxygens (almost 1). The speciation is relatively simple because only AlLn complexes (n = 1, 2 or 3) and a deprotonation product of AlL3 are formed in aqueous solution.Data obtained at 37 C show a slight reduction of complex stability constants with respect to corresponding values at 25 C; the enthalpic and entropic properties of the complexes cannot however be evaluated from our data, because the temperature interval examined is too narrow. The accuracy of the formation constant values obtained from potentiometric data at 25 C is substantiated by the agreement with the result obtained from UV spectrophotometry regarding AlL and, in some degree, from 1H NMR spectroscopy regarding AlL and AlL2; this agreement indirectly con.rms the whole speciation model. The crystal structure of tris(3-hydroxy-2(1H)-pyridinonato)- aluminium(...) (AlL3) is in agreement with the solution state .ndings; AlL3 crystallises in space group R3c, with no detectable disorder in the ligand.Fig. 6 Aluminium complexation strength, reported as pAl vs.pH, of HL and other similar ligands at 25 C (A = 3-hydroxy-N-methyl- 2-pyridinone in 0.1 M KCl,7 B = 1-hydroxy-2-pyridinone in 0.1 M KCl,20 C = 3-hydroxy-2-methyl-4(1H)-pyridinone in 0.6 M NaCl,6 D = catechol in 0.6 M NaCl 21). As a .nal comment, a comparison between the complexation strength of HL and other hydroxypyridinones can be made. The relative a.nities of the di.erent ligands have been compared by means of pAl plots 19 (pAl = log[Al3 ]) vs. pH at a given ligand and metal concentration (Fig. 6): the greater the value of pAl, the more stable are the corresponding aluminium complexes.Strictly speaking, pAl values reported in Fig. 6 cannot be directly compared, because corresponding thermodynamic data were obtained at di.erent ionic strengths; however, di.erences introduced by changing an ionic medium are usually small and, for our present purpose, negligible. The pAl curves suggest that HL forms weaker complexes than do the other hydroxypyridinones. 1-Hydroxy-2-pyridinone is a stronger aluminium chelator because of the greater acidity of the phenolic group.For 3-hydroxy-N-methyl-2-pyridinone and for 3-hydroxy-2-methyl-4(1H)-pyridinone the higher complexation strength arises from the greater stabilisation of a positive charge on the pyridinic nitrogen, due to the methyl group (inductive stabilising e.ect) and to the larger distance from the positive metal centre (minor inductive destabilising e.ect) respectively.16 Therefore a higher negative charge on the chelating oxygens is allowed, for both ligands.In the case of 3-hydroxy-2-methyl-4(1H)-pyridinone, this “chemical” result veri.es medical tests, which showed that derivatives of 3-hydroxy-4(1H)-pyridinones can be better therapeutic agents against aluminium overload than other hydroxypyridinones. Table 4 Crystal data for AlL3 Empirical formula Formula weight T/K Crystal system Space group a,b/Å c/Å C15H12AlN3O6 357.26 296 ± 2 Rhombohedral R3c 9.6840 ± 0.0014 29.523 ± 0.006 V/Å3 Z Independent re.ections Rint Final R1, wR2 [I > 2s(I)] (all data) 2397.7 ± 0.7 6 705 0.0474 0.0468, 0.1098 0.0434, 0.1211 Table 5 Bond distances (Å) and angles ( ) in AlL3 Al(1)–O(21) Al(1)–O(22) Al(1)–O(1) O(1)–C(1) C(1)–N(2) N(2)–C(3) C(4)–C(5) C(3)–H(3) C(5)–H(5) O(21)–Al(1)–O(2) O(2)–Al(1)–O(22) O(2)–Al(1)–O(11) O(21)–Al(1)–O(1) O(22)–Al(1)–O(1) O(21)–Al(1)–O(12) O(22)–Al(1)–O(11) O(1)–Al(1)–O(12) C(6)–O(2)–Al(1) O(1)–C(1)–C(6) C(3)–N(2)–C(1) C(3)–C(4)–C(5) O(2)–C(6)–C(5) C(5)–C(6)–C(1) N(2)–C(3)–H(3) C(5)–C(4)–H(4) C(4)–C(5)–H(5) 1.899(3) 1.899(3) 1.915(3) 1.285(6) 1.362(6) 1.295(9) 1.538(10) 0.93 0.93 91.40(15) 91.40(15) 170.47(11) 97.08(10) 170.47(10) 170.47(11) 84.08(9) 88.14(14) 112.6(3) 118.0(4) 121.3(5) 122.8(5) 125.8(5) 120.1(5) 119.5(3) 118.6(3) 123.5(3) Al(1)–O(2) Al(1)–O(11) Al(1)–O(12) O(2)–C(6) C(1)–C(6) C(3)–C(4) C(5)–C(6) C(4)–H(4) O(21)–Al(1)–O(21) O(21)–Al(1)–O(11) O(22)–Al(1)–O(11) O(2)–Al(1)–O(1) O(11)–Al(1)–O(1) O(2)–Al(1)–O(12) O(11)–Al(1)–O(12) C(1)–O(1)–Al(1) O(1)–C(1)–N(2) N(2)–C(1)–C(6) C(4)–C(3)–N(2) C(6)–C(5)–C(4) O(2)–C(6)–C(1) C(4)–C(3)–H(3) C(3)–C(4)–H(4) C(6)–C(5)–H(5) 1.899(3) 1.915(3) 1.915(3) 1.317(6) 1.412(5) 1.294(11) 1.355(6) 0.93 91.40(15) 84.08(9) 97.08(10) 84.08(9) 88.14(14) 97.09(10) 88.14(14) 111.1(3) 120.5(4) 121.4(5) 121.1(6) 113.0(5) 114.1(4) 119.5(4) 118.6(4) 123.5 Symmetry transformations used to generate equivalent atoms: 1 x y 1, x 1, z; 2 y 1, x y, z.2432 J.Chem. Soc., Dalton Trans., 1999, 2427–2432 Acknowledgements We would like to thank a referee for helpful comments concerning the crystal structure analysis, Sigrid Wocadlo for X-ray data collection and Italian Consiglio Nazionale delle Ricerche for .nancial support in the framework of the Cooperation Project with Magyar Tudomànyos Akadémia. We also thank the European Science Foundation for a provision of a fellowship to A. T.in 1993, when the collaboration between University of Padova and University of East Anglia began. References 1 G. J. Kontoghiorghes, Analyst (London), 1995, 120, 845. 2 G. J. Kontoghiorghes, Toxicol. Lett., 1995, 80, 1. 3 L. Gra., G. Muller and D. Burnel, Vet. Human Toxicol., 1995, 37, 455. 4 R. A. Yokel, A. K. Datta and E. G. Jackson, J. Pharmol. Exp. Ther., 1991, 257, 100. 5 E. T. Clarke and A. E. Martell, Inorg. Chim. Acta, 1992, 191, 57. 6 D. J. Clevette, W. O. Nelson, A. Nordin, C. Orvig and S. Sjöberg, Inorg. Chem., 1989, 28, 2079. 7 E. T. Clarke and A. E. Martell, Inorg. Chim. Acta, 1992, 196, 185. 8 V. B. Di Marco, A. Tapparo and G. G. Bombi, Ann. Chim. (Rome), 1999, 89, 397. 9 V. B. Di Marco, Ph.D. Thesis, University of Padova, 1998. 10 L. O. Öhman, Inorg. Chem., 1988, 27, 2565. 11 A. K. Covington, M. Paabo, R. A. Robinson and R. G. Bates, Anal. Chem., 1968, 40, 700. 12 S. Sjöberg, Y. Hägglund, A. Nordin and N. Ingri, Marine Chem., 1983, 13, 35. 13 R. C. Weast, D. R. Lide, M. J. Astle and W. H. Beyer (Editors), Handbook of Chemistry and Physics, 70th edn., CRC Press Inc., Boca Raton, FL, 1990. 14 K. E. Curtis and G. F. Atkinson, Can. J. Chem., 1972, 50, 1649. 15 E. Spinner and J. C. B. White, J. Chem. Soc. B, 1966, 991. 16 R. C. Scarrow, P. E. Riley, K. Abu-Dari, D. L. White and K. N. Raymond, Inorg. Chem., 1985, 24, 954. 17 F. J. C. Rossotti and H. Rossotti, The Determination of Stability Constants and Other Equilibrium Constants in Solution, McGraw- Hill Book Company Inc., New York, Toronto, London, 1961. 18 G. M. Sheldrick, SHELXTL PLUS, Siemens Analytical Instruments, Madison, WI, 1990. 19 T. Kiss and E. Farkas, Perspect. Bioinorg. Chem., 1996, 3, 199. 20 Y. J. Lin and A. E. Martell, Inorg. Chim. Acta, 1993, 214, 103. 21 L. O. Öhman and S. Sjöberg, Acta Chem. Scand., 1983, 37, 875. Paper 9/02997D