From several years, the sol-gel processes are developed for the synthesis of inorganic materials, and especially for metallic oxides with porous and tailored morphologies, and a homogeneous particle sizes distribution of crystallites (SiO₂, Al₂O₃, TiO₂, ZrO₂, …) [1] . By definition, a sol is a colloidal suspension of solid particles (1 - 1000 nm) in a liquid, in which the dispersed phase is present in a low amount, and a gel is obtained by the cross-linking of the sol colloid and results in a solid network with a high liquid amount in its pores. In this way, a sol-gel process can simply be summarized as an inorganic polymerization realized in solution media [1] . The starting materials used for the sol-gel process can be various reagents such as metallic salts, metallic oxides, metallic hydroxides, metallic complexes or metallic alkoxides. They should be soluble in the reaction medium and reactive enough to participate in the gel forming process. Though offering many applications, the sol-gel processes can be divided into the following basic elementary steps: Solution formation, gelation, aging, drying and densification by final heat treatments [1] .

Though being defined by the same synthesis steps, i.e. hydrolysis and condensation, depending on the type of precursors and solvent used, the sol-gel polymerization can be realized in two different paths: organic or aqueous. The organic paths are realized in organic solvents and use precursors sensitive to water, the most common ones being the metallic alkoxides. The polymerization is started when water is added to the mixture and occurs following Equations (1), (2) and (3) [1] :

First, an initial hydrolysis step, where the alkoxide precursor reacts with water:

where M is a metal and R is an alkyl group. After, the inorganic network is formed via the condensation step. The hydroxyl group reacts with another hy- droxyl group to form water:

or with an initial precursor to form alcohol:

Numerous studies have shown that it is possible to influence the hydrolysis and condensation rates through the control of various experimental parameters (temperature, pH, H₂O/solvent ratio, agitation time) [1] [2] [3] [4] . A gel obtained with a very fast hydrolysis (Equation (1)) will consist of small particles, which are highly reticulated, whereas a slower hydrolysis will present larger grains, with less interconnections.

In aqueous paths, after the solubilization of metallic cations in water, complexes such as aquo-([M(OH₂)_n]^z+), hydroxo-([M(OH)_n]^(n−z)−) or oxo-([M(O)_n]^(2n−z)−) compounds, with 2 < n < 8 and 1 < z < 7 are formed [5] . These complexes combine with two different condensation mechanisms: the olation and the oxolation [5] . The olation reaction occurs between complexes composed of hydroxo- (−OH) and aquo-(−OH₂) groups as follows (Equation (4)):

Thereafter, the M−OH₂ bond is broken, a hydroxo-bridge M−OH−M is formed and the ligand H₂O is eliminated (Equation (5)):

The bridging −OH ligands are called “ol” in order to differentiate them from hydroxo-ligands at the extremity. So the “olation” is the reaction leading to the formation of an “ol” bridge (−OH−) between two metals. Olation reactions are generally fast and depend mainly on the lability, i.e. the ability of bond cleavage, of the M−OH₂ bond. The lability of the M−OH₂ bond depends on the polarization of the cation: the bond is more inclined to be broken for cations with low charge z or for large cations. The olation reaction is supposed to occur according to a nucleophilic substitution of type SN1 (the reaction rate is of first-order and depends only on the electrophile) [5] . The second condensation mechanism is called “oxolation”. It occurs when both precursors do not contain any aquo ligands (−OH₂) coordinated to the metallic cation. It involves the condensation of two hydroxo (−OH) groups to form H₂O as follows (Equation (6)):

First, a nucleophilic addition with formation of an “−ol” bridge leads to an unstable transition state:

Thereafter, a proton is transferred to a terminal −OH ligand to form an aquo- group, which is finally eliminated (Equation (7)):

Condensation mechanisms by oxolation lead to the formation of “−oxo” (−O−) bridges. However, contrary to olation reactions, oxolation reactions are usually influenced by the acidity of the solution and are slower [5] . The oxolation reaction is supposed to occur according to a nucleophilic substitution of type SN2 (the reaction rate is of second-order and depends on both reagents). Two essential conditions are required for the oxolation reaction to take place: 1) the presence of a hydroxo-(−OH) ligand carrying a negative charge (δ(OH) < 0) in the coordination sphere of a cation; 2) the possibility of a cation to undergo the nucleophilic attack from a hydroxo-ligand, i.e. the cation must be sufficiently electrophile to undergo the attack (δ(M) > +0.3) [5] .

The comparison of aqueous and organic paths for the synthesis of γ-Al₂O₃ was studied by Rajaeiyan et al. [6] . The authors showed that after a classic calcination under air at T = 750˚C, more interesting structural and textural properties were obtained for the aqueous method compared to the organic path. Indeed, the specific surface area of alumina synthesized with aqueous path is higher (200 m²/g) compared to alumina synthesized with organic path (30 m²/g). Furthermore, γ-Al₂O₃ peaks are visible on XRD spectrum for the sample synthesized with the aqueous method, whereas the sample synthesized with the organic method shows amorphous alumina peaks.

When dissolved into water, the aluminum ions can form different complexes depending on the pH value: for pH < 1, Al³⁺ for 1 < pH < 4,

for 4 < pH < 6, Al(OH)₃ for 5 < pH < 8 and for pH > 7.5 [7] . Once formed, these aqueous complexes will condense and form cross-link- ing networks via oxolation and olation reactions previously described. The final material properties (crystallinity, specific surface area and pore volume) depend strongly on the experimental conditions such as the temperature, the type of precursors, the agitation or the aging conditions. So depending on these synthesis conditions, the basic brick of aluminum hydroxide (Al(OH)₃) can either be aggregated into amorphous alumina hydroxide (Al(OH)₃), diaspore (α-AlOOH) or boehmite (γ-AlOOH). Therefore, controlling the formation of these different compounds is of major relevance. Indeed, after a classic calcination under air at T = 500˚C, Al(OH)₃ leads to η-Al₂O₃, whereas diaspore leads to α-Al₂O₃ and only boehmite leads to γ-Al₂O₃ [5] [7] [8] .

Okada et al. [9] emphasized the importance of the operating variables: the formation of boehmite under atmospheric pressure is guaranteed only for temperatures from 70˚C to 100˚C, and for pH values comprised between 7.5 and 10. Indeed, under these aging conditions, the amorphous aluminum hydroxide and diaspore crystals formed during the precipitation are inclined to undergo dissolution/reprecipitation mechanisms to form [Al(OH)₄]⁻ complexes, which are rearranged to form the boehmite phase. This phenomenon was confirmed by Lee et al. [10] , showing that by allowing the formation of γ-AlOOH, these conditions allow obtaining after calcination very porous alumina materials with the highest specific surface area and pore volume.

The cogelation method allows the synthesis of homogenous supports from the hydrolysis and condensation of different kind of precursors. For example for the formation of SiO₂, two alkoxides of the same element are used: tetramethoxysilane (TMOS; Si(OCH₃)₄) and tetraethoxysilane (TEOS; Si(OC₂H₅)₄) [1] . It is also possible to start from one alkoxide and one salt, for example the synthesis of Al₂O₃ with a mixture of aluminum isopropoxide (Al(OC₃H₇)₃) and aluminum nitrate (Al(NO₃)₃) [11] , or Al(OC₃H₇)₃ and aluminum acetylacetonate (Al(C₅H₇O₂)₃) [12] . For the synthesis of zeolithes, the cogelation method is used with two alkoxides with two different elements: TEOS and aluminum isopropoxide, or TEOS and aluminum sec-butoxide [13] .

Cogelation with a functionalized precursor is also possible in order to highly disperse metallic active sites inside inorganic support. For example, the use of silicon alkoxide containing a functional organic group like ethylenediamine, which is able to form a chelate with a metallic cation, showed to be an innovative and efficient way to improve the performances of catalytic materials. In this way, Lambert et al. synthesized finely dispersed Pd/SiO₂ [14] [15] , Ag/SiO₂ [15] , Cu/SiO₂ [15] [16] , Ni/SiO₂ [17] , Pd-Ag/SiO₂ [18] [19] , Pd-Cu/SiO₂ [19] , Ni- Cu/SiO₂ [17] and Fe₂O₃/SiO₂ [20] catalysts by the cogelation between TEOS and a silicon alkoxide functionalized with a ethylenediamine group (3-(2-aminoe- thylamino)propyltrimethoxysilane, (CH₃O)₃−Si−(CH₂)₃−NH−(CH₂)₂−NH₂, called EDAS).

Since no functionalized aluminum alkoxides are commercially available, the functionalization of alumina supports is nowadays only performed via grafting methods [21] [22] [23] . The usual methods consist in the mixing of a functionalized silica precursor with an alumina support in an organic solvent. Thereafter, a small quantity of water is added to hydrolyze the alkoxy groups and form silanol groups, which graft onto the alumina support by condensation reactions. Nevertheless, these methods are nevertheless not convenient and difficult to scale up, principally because of the required synthesis conditions (inert and dry atmosphere, toxic organic solvents). The purpose of this work is to perform the synthesis of very porous γ-Al₂O₃ supports modified by EDAS molecules thanks to a simple and classic aqueous cogelation method.

Two synthesis methods were used for the synthesis of γ-Al₂O₃ and EDAS modified γ-Al₂O₃ supports. The first method, called method A (related to “alkoxide”), used aluminum sec-butoxide (Al[OCH(CH₃)CH₂CH₃]₃) as aluminum source. The second one, called method N (related to “nitrate”), used aluminum nitrate (Al(NO₃)₃) as aluminum source. For both methods, the reagents were dissolved in a media solution with H₂O/(Al + 3/4Si) and Ethanol/(Al + Si) molar ratios equal to 50 and 5 respectively. The presence of the fraction 3/4 in the first ratio was justified by the fact that only the three methoxy groups in the EDAS molecule are hydrolyzed by water. Ethanol was added to the media solution in order to prevent a direct contact of EDAS with H₂O, which is known to lead to macroscopic silicate aggregates [1] . The different amounts of reagent (Al sec- butoxide, Al nitrate, EDAS, water, ethanol) used for the synthesis of 10 g of calcined material are presented in Table 1.

This procedure was developed for the synthesis of catalytic supports doped with 2 wt.% of Ni (not presented in this paper). Therefore, the amount of EDAS was calculated in order to chelate all nickel ions, i.e. the Ni/EDAS ratio is equal to 4.

^an_Al, the aluminum precursor, is Al sec-butoxide in samples A and Al nitrate in samples N.

The preparation of samples A and A-EDAS was realized as follows: Al sec-butoxide was quickly (1 mL/s) injected into the solution (water + ethanol) under vigorous stirring. The sol was then stirred for 30 min. The variant method consisted in adding EDAS to the solution (water + ethanol) under vigorous stirring (500 rpm) for 5 min before adding Al sec-butoxide. After the stirring, the pH value was increased to 9.5 by a dropwise (1 mL/min) addition of a NH₃ aqueous solution (15 M).

The preparation of samples N and N-EDAS was realized as follows: Al(NO₃)₃ was dissolved in the solution (water + ethanol) and the sol was then stirred for 30 min. The variant method consisted in adding EDAS to the solution under vigorous stirring for 5 min before adding Al(NO₃)₃. The rest of the procedure is similar to the preparation of samples A and A-EDAS.

For both methods, after increasing the pH value until 9.5, the vessels were closed and the solutions were mixed under vigorous stirring in a heated oil bath at 85˚C for 24 h. After the agitation step, the vessels were opened and put into an oven at 85˚C for aging for 24 h with a pressure equal to 700 mbar. The obtained gels were washed twice with water and once with ethanol. Thereafter, the washed gels were put into an oven at 110˚C for drying for 24 h with a pressure equal to 900 mbar. The dried samples were calcined under air for 4 h at T = 500˚C with a heating rate equal to 2˚C/min.

TG-DSC measurements were performed on a SensysSetaram instrument. 0.05 g of the samples were crushed and introduced into an alumina crucible. The samples were heated from 25˚C to 800˚C with a heating rate of 2˚C/min under air (20 mL/min).

Apparent densities were measured by helium pycnometry with a Micro- meritics AccuPyc 1330 at T = 25˚C. Mercury porosimetry measurements were performed with 0.3 g of sample using a Poremaster 60 instrument from Quantachrome with pressures going from 1 to 40.000 bars. Nitrogen adsorption-de- sorption isotherms were measured at −196˚C on a Micromeritics ASAP 2010 instrument after 12 h of outgassing at 300˚C and 10⁻⁵ mbar.

SEM measurements were performed with a FEI ESEM-FEG XL3 device. Pictures made with Backscattered Electrons (SEM-BSE) detectors permitted to get a view of the surface of the samples, with different contrasts depending on the elemental composition. Energy Dispersive X-ray analyses (SEM-EDX) were also performed with the same apparatus in order to determine the elemental composition of a defined a surface (400 × 300 μm²) and to build up element mapping pictures. The measurements were performed at an acceleration voltage of 15 keV, set on the Spot 4 and with a vacuum of 0.4 Torr.

TEM analyses were performed on a CM10-PW6020 Philips Electron Microscope. First, crushed samples were dispersed in absolute ethanol. Then a drop of the dispersion was placed on a copper grid (Formvar/Carbon 200 Mesh Cu from Agar Scientific). For a higher quality of the pictures, the vacuum lines were cooled with liquid nitrogen for 30 min before the measurements.

X-ray diffraction measurements were performed on a Bruker D8 Twin-Twin diffractometer (Cu-K_α Radiation). The samples were crushed and the measure- ments were performed from 2θ = 30˚ to 2θ = 80˚ with a step time of 18 s and a step size of 0.04 s. The sizes of alumina crystallites were calculated by using the Scherrer equation centered on the (4 0 0) peak of γ-Al₂O₃ (2θ = 67.0˚) of the XRD spectra [24] .

In the synthesis conditions used in this work, it seems that samples starting from two different aluminum precursors such as Al sec-butoxide (sample A) and Al nitrate (sample N) present equivalent thermal behavior (Figure 2), morphology (Figures 3-5), porosity (Figure 6) and crystallinity (Figure 7).

The textural and crystallographic similarities between samples A and N are attributed to the synthesis operating variables (pH = 9.5, Temperature of aging equal to 85˚C). Whereas the aqueous precipitation of aluminum nitrate is known to favor the formation of boehmite (γ-AlOOH) [5] [9] [10] , the hydrolysis of Al sec-butoxide is in the contrary known to lead to amorphous alumina hydroxide species (Al(OH)₃) [5] . However, in this case, sample A exhibits X-Ray peaks of well-crystallized γ-Al₂O₃ (Figure 7), which means that the synthesis conditions (pH, T_aging) used can successfully convert the amorphous alumina hydroxide species into crystalline boehmite.

When EDAS is added during the synthesis of alumina samples (samples A-EDAS and N-EDAS), it is observed:

1) Though starting from different aluminum precursors, both samples A-EDAS and N-EDAS appear to be homogeneous (Figure 3 and Figure 4). This observation is particularly interesting for sample A-EDAS since the hydrolysis rates of silicon alkoxides are lower compared to aluminumal koxides. So without adequate preparation steps, it may lead to inhomogeneous materials [30] . Nevertheless, the formation of homogeneous aluminosilicate samples with different precursor types is shown by Sinko et al. [31] . In line with the present results, the authors showed that starting from a mixture of TEOS + Al(NO₃)₃ or TEOS + Al(OCH(CH₃)₂)₃ reagents in water + ethanol media lead to homogeneous aluminosilicate materials with similar textural and crystalline properties than pure alumina samples. Furthermore, another study showed that the cogelation of Al sec-butoxide and trimethoxysilane also allowed obtaining homogenous aluminosilicate supports [32] .

So in this work, the formation of homogeneous aluminosilicate supports via the method A is possible thanks to the pre-hydrolysis step of EDAS before the Al sec-butoxide addition. Indeed, pre-hydrolysis steps are known to counterbalance the difference of alkoxide reactivity, thus allowing the formation of homoge- neous cogelled materials [30] .

The formation of homogenous supports via the method N is explained by oxolation/olation reactions between hydrolyzed Si-OH groups from EDAS and alumina hydroxide complexes such as or. It is assumed that these condensation reactions occur before the precipitation and lead to the formation of submicronic aggregates by cogelation since the sol remains transparent.

2) Comparing the properties of samples N-EDAS and A-EDAS reveals that starting from Al nitrate or Al sec-butoxide also leads to materials with similar properties. Indeed, despite the lower textural properties (V_Dubinin, V_P and S_BET, Table 2) of sample A-EDAS compared to sample N-EDAS, both samples exhibit similar thermal behaviour (Figure 2), morphology (Figures 3-5), porosity (Figure 6) and crystallinity (Figure 7).

3) Both EDAS modified alumina supports are characterized by a strong modification of their crystalline properties. Indeed, samples A-EDAS and N-EDAS show less defined alumina X-Ray peaks (Figure 7) compared to the pure alumina samples (samples A and N). Due to its methoxy groups, EDAS is classified as a high reactive silicon precursor [15] [16] [17] [18] [19] . This high reactivity could lead to an insertion of Si atoms inside the bulk alumina structure, thus modifying the γ-Al₂O₃ crystallinity.

4) EDAS strongly modifies the morphology of alumina samples. Indeed, for samples A-EDAS and N-EDAS, TEM observations (Figure 5) show the presence of spherical alumina particles, which could explain the very narrow pore size distribution obtained from the nitrogen adsorption-desorption isotherms (Figure 6). It is assumed that the high reactivity of the methoxy-groups in EDAS could lead to the quick formation of numerous nucleation points, which react by cogelation with aluminium precursor to form after calcination, spherical aluminosilicate particles [15] [16] [17] [18] [19] [30] .