Langmuir 1994,10,4344-4348
4344
Chemistry and Structure of Al(OH)/OrganicPrecipitates. A Small Angle X-ray Scattering Study. 1. Numerical Procedure for Speciation from Scattering Curves A. Masion,? D. Tchoubar,* J. Y . Bottero,**tF. Thomas,? and F. Villibrast Laboratoire Environnement et Minkralurgie, Groupe de Recherche sur I%au et les Solides Divish, URA 235 CNRS, ENSG, INPL, BP 40 54501 Vandoeuvre Cedex, France, and Centre de Recherche de la Matitre Diviske, UMR CRNS-Universith, Laboratoire de Cristalographie, Universitk d'orlians, BP 67-47, Orlhans Cedex, France Received January 14, 1994. In Final Form: July 25, 1994@
A procedure for fitting small angle X-ray scattering curves in order to describe the structure of&organic precipitates was elaboratedin the log Q range from -0.2 to -0.8. Models were built by considering the Al species (monomers, dimers, trimers, and tridecamers) as hard spheres associated in short linear chains. Size of the spheres and chain length were defined from results obtained through other methods, mainly 27AlNMR. "he procedure was tested by performing the fit on ladate/Al scattering curves.
Introduction Organoaluminum precipitates are encountered in various fields. Aluminum films for the electronics industry are now produced via a complex procedure in which Al hydroxides are peptized by acetic acid.lP2 In continental freshwaters, the toxicity of aluminum is considerably reduced by complexation and precipitation with organic a ~ i d s . ~In - ~soils, organic acids retard or hinder the crystallization of Al hydroxide^.^,^ Industrial benefit, as well as fundamental knowledge on environmental toxicity of aluminum, are the outcomes for the investigation of the structure and mechanisms of formation of organoaluminum for the investigation of the structure and mechanisms of formation of organoaluminum precipitates. The study of the aquatic chemistry of aluminum has been considerably facilitated in the 2 past decades by the use of 27AlNMR. Indeed, the natural abundance (100%) of the 27Al nucleus allows accurate qualitative and quantitative NMR studies of this element. The mechanisms of formation and the resulting structure of Al-organic precipitates are governed by the hydrolysis of aluminum and the complexing power of the organic ligands, which compete with the hydroxyls for the hydrolysis sites ofAl. Their study requires knowledge on the complexation in solution. This was the subject of numerous works.8 The mechanisms of hydrolysis of A13+ are also well f
Laboratoire Environnement et MinBralurgie.
3 Centre de Fbcherche de
la Matiere DivisBe.
Abstract published in Advance ACS Abstracts, September 15, 1994. (1)Ohta, H.; Kurokawa, Y. J. Mater. Sci. Lett. 1982,11, 868. (2)Kobayashi, Y.; Yamazaki, S.; Kurokawa, Y.; Miyakawa, T.; Kawaguchi, H. J . Mater. Sci. Mater. Elec. 1992,2,20. (3)Hue, N. V.; Craddock, G. R.; Adams, F. Soil Sci. S 0 c . h .J. 1986, 50, 28. (4) Prakash, A,; MacGregor, D. J. In Aquatic and terrestrial humic materials; Christman, R. F., Gjessing, E. T., Eds.; Ann Arbor Science: Ann Arbor, MI, 1983. (5) Stevenson, F. J.; Vance, G. F. In The environmental chemistry of aluminum; Sposito, G., Ed.; CRC Press: Boca Raton, FL, 1989. (6) Violante, A,; Violante, P. Clays Clay Miner. 1980,28,425. (7) Violante, A.; Huang, P. M. Clays Clay Miner. 1985,36,181. (8) Nordstrom, D. K; May, H. M. In The environmental chemistry of aluminum; Sposito, G., Ed.; CRC Press: Boca Raton, FL, 1989. (9)Bottero, J. Y.; Cases, J. M.; Fiessinger, F.; Poirier, J. E. J . Phys. Chem. 1980,84,2933. (IOIBottero, J. Y.; Marchal, J. P.; Poirier, J. E.; Cases, J. M.; Fiessinger, F. Bull. Soc. Chim. Fr. 1982,ll-12,439. @
Our recent studies on the stages preceding the precipitation of partially hydrolyzed mixtures of AlC13 and organic acids, at pHs lower than 6, provided the speciation of aluminum by liquid-state 27Al NMR.18-20 We also studied the nature of the precipitates by solid-state NMR.2122
In the current work, we aimed at tracing back to the formation and growth mechanisms of the precipitates from the analysis of their structure from small angle X-ray scattering (SAXS)curves. This study was based on the knowledge of the preceding "chemical history" of the precipitates. The outermost part of the scattering curves, which characterizes the local connectivity of the building blocks, was simulated by theoretical models. The structure at a larger scale was described from the slopes of the scattering curves. The first part of this work, exemplified on Al-lactate precipitates, reports the method of calculation for simulating the experimental scattering curves. The results of the simulations and the structural aspects are discussed in part 2. The structural modifications of the precipitates due to the ligands are studied in part 3.
Experimental Section All reagents were of analytical grade. Stock solutions of 0.5 M AlClg6H20, sodium lactate,and of 1 M NaOH were prepared (11)Bottero, J. Y.; Tchoubar, D.; Cases, J. M.; Fiessinger, F. J . Phys. Chem. 1%82,86,3667. (12) Axelos, M.; Tchoubar, D.; Bottero, J. Y.; Fiessinger, F. J. Phys. (Paris) 1985.46.1587. (13)Bottero, J. Y.; Axelos, M.; Tchoubar, D.; Cases, J. M.; Fripiat, J. J.; Fiessinger, F. J. Colloid Intel.face Sci. 1987,117, 47. (14)Akitt, J. W.; Greenwood, W.; Khandelwal, B. L.; Lester, G. D. J. Chem. Soc., Dalton Trans. 1972,604. (15)Akitt, J. W.; Farthing, A. J. Chem. Soc., Dalton Trans. 1981, 1606. (16)Brinker, C. J.; Sherrer, G. W. Sol Gel Science; Academic Press: New York, 1989. (17) Henry, M.; Jolivet, J. P.; Livage, J. In Structure and Bonding; Springer: Berlin, 1992. (18)Thomas, F.; Bottero,J.Y.; Masion,.A.:. GenBvrier. F. Chem.Geol. 1990,84(1/4), 227. (19) Thomas, F.; Masion, A; Bottsro, J. Y.; Rouiller, J.;GenBvrier, F.; Boudot, D. Enuiron. Sci. Technol. 1991,25,1553. (20) Thomas, F.; Masion, A.; Bottero, J. Y.;Rouiller, J.; Montigny, F.; GenBvrier, F. Enuiron. Sci. Technol. 1993,27,2511. (21) Masion, A.; Thomas, F.; Bottero, J. Y.; Tchoubar, D.; Tekely, P. J . Non-Cryst. Solids, in press. (22)Thomas, F.; Masion, A.; Bottero, J. Y.; Tekely, P. In NMR spectroscopy in environmental science and technology; Minear, R. A., Nanny,M. A., Eds.; Lewis: Boca Raton, FL, 1994.
0743-7463/94/241Q-4344$Q4.50/00 1994 American Chemical Society
Langmuir, Vol.10,No.11, 1994 4345
Al-Organic Precipitates
80 -
23
60-
g 0 Q a
’ 40-
c
, 20
-
-0-
-C
n-, 0
Acetate Oxalate Lactate
2
6
4
a
10
PH
Figure 1. Turbidity vs pH for ligand-Al mixtures at LIM
=
0.50.
with deionized, 0.22-pmfiltered water. Solutions containing 0.1 M Al and 0.05 or 0.1 M lactate (viz. ligandmetal mole ratios, noted L/M,of 0.5 and 1.0,respectively)were prepared at pH 3.5 and were hydrolyzed up to pH 7.0 at room temperature in a 200-mL vessel fit out with four baffles. During hydrolysis, the solutionswere vigorouslystirred by means of a four-bladedpaddle at a rate of 500 rpm. The titrant solution (1M NaOH)was added with an automatic Tacussel Electroburex EBX 2 pipet. The injection speed was 0.04 mol of NaOH min-Ymol of Al)-l. The pH measurements were made with a Tacussel Titrimax TT 100 apparatus using a Tacussel XC 250 combined electrode. The formation of precipitates was detected using a Hach Ratiom turbidimeter calibrated in standard Nephelometric Turbidity Units (NTU). SAXS analysis was performed on suspensions at pH 7.0, which corresponded to the maximum turbidity (Figure
-2
-1.5
,
1
-1
-0,5
0
log (Q)
Figure 2. Corrected and connected SAXS curves: lactate-Al mixtures at LIM = 0.50 and 1.00. features of the structure. We have23
1).
Two X-ray sources were used according the size domain: (i) synchronon radiation on D24 bench (wavelength1= 1.6A) of the DCI storage ring of LURE (Universitb de Paris Sud, Orsay, France), recording time 2 h, Q range 0.01-0.164 A-l; (ii) a laboratory apparatus equipped with a Rigaku rotating anode (1 = 1.54A)(Universitdd‘orlbans, France),time was 20 h (10times, 2 h) and the Q range was 0.025-0.734A-1. Q is the wave vector modulus and is equal to 4dsin @/A, where 28 is the scattering angle.
Data Treatment Background Correction. The scattering by the particles was obtained by substraction of the scattered intensities of the solvent from the scattered intensities of the suspension, weighted by the absorption coefficient, the recording time, and the time averaged value of the incident beam intensity. Smoothing of the Scattering Curves. Because of the noise, especially at high Q, the curves had to be smoothed before further analysis. In the smoothing procedure used, we assumed that, within a window of 2k experimental points, the noise is random and the intensity Z(Q) is a linear function of Q. The smoothed intensity is then approximated by
I(n)= I(n
+ 1) +
+
I(n k) - I(n - k) Q(n) Q=
(4) 1
Qm
-
Q%Q)
2nQmin
where &fincorresponds to the edge of the beam stop, and Qmaxis the largest experimental angle. ARer normalization, the curves obtained for each sample on the D24 bench and the laboratory bench could be connected at log(&) = -1.2 (Figure 2).
Qualitative Analysis of the Experimental Curves Figure 2 displays the log(1) vs log(&) plots of the scattering curves. The absence of any correlation peak indicates that there are no characteristic interparticle distances within the samples. None of the curves could be fitted by the scattering of isolated homogeneous particles (spheres, platelets, needles, ...1. They are characteristic of aggregates, the subunits of which scatter a t the largest angles. In the latter Q range (i.e. log(&) 2 -0.4) the slopes of the scattering curves obtained from Al-lactate precipitates were near zero. This mean that (23) Porod, G. In Small AngleX-ray Scattering; Glatter, O., Kratky, O., Eds.;Academic Press: New York, 1982.
Masion et al.
4346 Langmuir, Vol. 10, No. 11, 1994 the subunits of the aggregates are smaller than the lower experimental detection limit, which is d Q = 4.5 A. The part of the curves for log(&) ranging from -0.4 to -0.8, revealed slopes close to 1 for both L/M ratios. This corresponds to the existence of locally linear aggregate^.^^
Simulation of the Outermost Part of the Scattering Curves Preliminary Hypotheses. In order to restrict the number of synthetic curves to be calculated and to build realistic models, several hypotheses had to be used. Their relevance was stated by a previous study of the aluminum chemistry. 27Al solid-state NMR analysis of the freeze-dried precipitates clearly showed that, in lactate-Al precipitates, aluminum is poorly polymerized, the largest observed particle being the A l l 3 polycation.21,22Thus the major species within the aggregates are monomers, oligomers (dimers and trimers), and Al13, according to the A13+chemistry.16J7 The models were built with the above species forming locally linear aggregates. Simulation Procedures. For simple species, the scattering curves can be calculated by taking into account all interatomic distances and the atomic scattering factors. The scattered intensity is given byz5
-13
-1,6
-1,7 0 v
-8 -1,8
-1.9
-2
-2
-1,8 -1.6 -1,4 -1,2
-1
-0,8 -0,6 -0,4 -0,2
Figure 3. Theoretical SAXS curves for the Al monomer ((a) calculated from the atomic scattering factors, (b) calculated from the equation of a sphere) and the Al dimer ( ( c ) atomic scattering factors, (d) sphere). hard spheres. The scattered intensity of an isolated sphere isz7
sin Qr - Qr cos Qr I(&> = P(3
(7)
Q3r3
where ro is the distance between the atom i and the atom fi is the scattering factor of the atom i and is calculated by
j.
The coefficientsaj, bj, and cj are determined for fitting the scattering factors as tabulated using Hartree-Fock wave h c t i o n s (International Tables No. 111). In the aluminum octahedron, the scattering atoms are the Al and 0 atoms. The contribution of the hydrogen atoms to scattering is low and can be neglected. The distance between the Al and one 0 atom is 1.88 A.26 This allowed the calculation of the theoretical scattering curve of the Al monomer. The interatomic distances are 1.88 A(Al-0),2.66A(0-0), and3.76A(O-O). The scattering curves of the Al dimer and trimer were obtained in the same way. The major drawback of this calculation procedure is that it becomes very tedious when larger molecules have to be modeled. For example, the simulation of the Al monomer requires the knowledge of 21 interatomic distances, whereas this number increases to 121 in the case of the trimer. Moreover, in aggregates, the local structure is disordered. Consequently, interatomic distance fluctuations have to be taken into account, and this is not possible with this calculation procedure. To overcome this complexity, we hypothesized that the Al species have a roughly spherical shape. The species are no longer considered as polyatomic buildings, but as (24) Bottero, J. Y.; Tchoubar, D.;Amaud, M.; Quienne, P. Langmuir 1991, 7, 1365. (25) Guinier, A.; Fournet, G. Small-Angle Scattering of X-rays; J. Wiley & Sons: New York, 1955. (26) Wells, A. F. Structural Inorganic Chemistry;Clarendon Press: 1984.
0
log (Q)
where r is the radius and V the volume of the sphere. The sizes of the spheres in which the Al species are contained were estimated using the interatomic distances. The radii were taken equal to half of the largest distance observed within the molecule; 1.88 A for the monomer and 2.97 A for the dimer and trimer. These two latter species became thus indistinguishable. The A l l 3 polycation was hypothesized to be a sphere of 7 A radius, which corresponds to the A l l 3 surrounded by ligands. The scattering curves calculated for monomers and dimers, using the two methods, were compared (Figure 3). In the case ofthe dimer, the two curves were practically identical over the whole Q range. With the monomer, the fit was less satisfactory. Nevertheless, in the experimentally accessible Q range, the two simulation procedures finally gave equivalent results (Figure 4), the differences between the curves appearing only for the largest angles. All the species could be modeled by considered them as spheres. Factors Affecting the Scattering Curve. Chain Length. The amplitude A of a wave scattered by a sphere with radius r isz5
4 3
3
A(&) = -nr 3
sinQr - QrcosQr Q3r3
(8)
In dilute systems, the scattered intensity is the square of the amplitude. When the particles are associated, the interference must be taken into account. In that case, the intensity becomes
itj
Wij
where Ii is the intensity scattered by the sphere i, and ru is the distance between the atom i and the atom j . Clear differences appeared only between the curve relative to an isolated sphere and those calculated for (27) Rayleigh, Lord Proc. R . SOC.London 1911, A-84, 25.
Al-Organic Precipitates
Langmuir, Vol. 10,No. 11, 1994 4347
0
0
-02 -0,4 -1
-0.6
-0,s
-= -g
-2
-1
-
-1,z
s”
-1,4
-3
-1.6
-1,s -4
.2 .0,8
1
.0,4
-0,6
-0.2
0
log ( Q )
Figure 4. Comparison between theoretical SAXS curves for the Al monomer ((a)atomic scattering factors, (b)sphere)and an experimental lactate-Al, LIM = 0.50 curve (c).
-5
-2
-1,8
-1.6
-l,4
-1.2
-1
-0,8 -0,6 -0,4
-0,2
0
log ( 0 )
1
Figure 6. Theoretical SAXS curves for a five-membered chain of spheres (radius 7 A) with spacings of 0, 0.1, 0.3, 0.7, 1, 1.5, 2, and 3 A between the spheres. 0
-1
-=3 -*
for aggregation and precipitation. Thus, it was necessary to define a distance d representing the spacing between two spheres. Since no information about the distribution of the ligands around the Al species is available, the scattering curves of five-membered chains with spheres of 7 A and d spacings ranging from 0 (spheres in contact) to 3 A were calculated in order to study the influence of these spacings on the shape of the curves (Figure 6).These curves were practical1 superposable over the whole Q range. A spacing of 1 between the spheres was chosen, representing an arbitrary value of the average thickness of the ligand crown. Fitting Procedures. The calculation of the interference term of eq 9 implies that the position of each sphere in the chain is known, so that the r” distances can be derived. Since the exact location of the species within the precipitate is unknown, a statistical procedure was adopted, in which all possible dispositions were taken into account. The resulting intensity is the average of scattering from all these dispositions. The fitting procedure consisted of calculating the scattering curves of one- to seven-membered chains of monomers, oligomers (dimers and trimers), and called “submodels”. The fit to the experimental curve was carried out by a least-squares method. Only two or three relevant submodels were supported by the previous results from chemical analysis. Testingthe FittingProcedure: Case ofAl-Lactate Aggregates. The scattering curves of the Al-lactate precipitates were simulated for log Q comprised between -0.2 and -0.8 (size range from 4.5 to 20A)(Figure 7). The best fits were obtained with models built with isolated monomers and short three-membered linear A l l 3 chains (Figure 7). Including an oligomer component in the model did not improve the fit. The proportions yielded by the calculation are 96%for the monomer and 4% for the A l l 3 when LIM = 0.5, and 97%and 3%,respectively, for LIM
l
-3
-4
-5 -2
-1,8
-l,6
-1,4
-1,2
-1
-0,8 -0,6 -0,4
-0,2
0
log (Q)
Figure 5. Theoretical SAXS curves for a sphere with a radius of7Aassociatedinchainsof(a) l,(b)3,(c)5,and(d)7members. chains with a variable number of particles (Figure 5). These latter curves showed slight differences essentially for the smallest Q values. In the Q range where the fit between the experimental and computed curves was actually performed, the chain length played only a secondary role in the scattering profiles. Spacing between the Spheres. In the studied systems, the particles are not necessarily in contact and the defined sizes for the spheres do not take into account the presence of organic ligands surrounding the Al species responsible
4348 Langmuir, Vol. 20,No. 11, 1994
Masion et al. monomers as possible components of the solid phase. The complexed monomers can be either hydrolyzed and precipitated by charge screening or involved in the formation of oligomers or Al13, the ligands being displaced by the h y d r o ~ y l s .The ~ ~ huge ~ ~ ~ majority of monomers is surprising since NMR revealed the presence of large amounts of A l l 3 in solution and in fresh precipitates. This suggests that the aggregates do not result from simple precipitation of the soluble species but that they undergo transformations. This aspect will be discussed in the second and third part of this work.
\
.1
48
-0,6
-0.4
a 2
0
log (a)
Figure 7. Fitting of the SAXS curves for the lactate-AI system: bold lines, experimenta; dotted lines, calculated. = 1.0. This similarity agrees with the NMR results where the major difference between the two LIM ratios was the stoichiometry of the formed complexes.20 The fit of the SAXS curves indicates that, for both LIM ratios, the aggregates are built with the same subunits in quasiidentical proportions. It is reasonable to consider the
Conclusion The simulation of SAXS curves is a powerful tool for the study of amorphous organomineral precipitates, such as Al-organic aggregates. Such a procedure was carried out by adequate data treatment and the use of simplifying hypotheses, whose relevance was demonstrated by a previous study a t atomic scale: presence of four different Al species and linear arrangement of the subunits. By use of supplementary simplifications (species modeled as hard spheres with arbitrary spacings and statistical size distribution), validated by ab initio simulations, satisfactory fits could be obtained for the lactate-Al systems. However, the proportions of the Al species yielded by the calculation must be interpreted with care. They indicate only the major and minor species involved in the aggregates but do not allow real quantitative analysis of the SAXS curves. Thus, our modelization procedure can provide new elements in understanding Al chemistry during hydrolysis in the presence of organic ligands. This approach will be extended to other Al-organic ligand systems in part 2. Acknowledgment. The authors wish to thank Dr. P. Vachette for SAXS facilities at LURE (UniversitBde Paris Sud), and the program Dynamique et Bilan de la Terre DBT CNRS-INSU 9211 No. 717 for financial support.