Langmuir 2004, 20, 3449-3455
3449
Modification of the Surface Properties of Porous Nanometric Zirconia Particles by Covalent Grafting David Carrie`re, Me´lanie Moreau, Philippe Barboux,* and Jean-Pierre Boilot Laboratoire de Physique de la Matie` re Condense´ e, CNRS UMR 7643C, Ecole Polytechnique, 91128 Palaiseau Cedex, France
Olivier Spalla Service de Chimie Mole´ culaire, DRECAM, CEA-Saclay, 91191 Gif sur Yvette, France Received December 1, 2003. In Final Form: February 19, 2004 We here report on the covalent grafting of various phosphated species (phosphoric acid, phenylphosphonic acid, and octyl phosphate) onto the surface of monoclinic zirconia nanoparticles obtained by hydrothermal treatment of zirconium acetate. The initial particles are 60 nm aggregates of nanometric primary grains and present an inner porosity. Small-angle X-ray scattering shows that the high specific area of the colloidal particles (450 m2‚g-1) decreases to 150 m2‚g-1 upon drying. Therefore, phosphated reactants can access the whole internal surface of the aggregates only before drying. The surface of the particles can be covered with functional groups bound through a variable number of Zr-O-P bonds. Several factors probably enhance the reaction between the particles and the phosphates or phosphonates: the large specific area of the particles, a fully accessible porous network, and a large concentration of surface terminal groups. At the same time, the morphology of the particles is well preserved upon grafting. This is due to the good crystallinity of the primary grains that constitute the particles. In addition, the grafting drastically modifies the surface properties of the colloids. For example, the polarizability of the particles decreases in the sequence -POH > as-prepared ZrO2 > -PC6H5 > -POC8H17. Furthermore, the grafting of octyl phosphate allows exclusion of water from pores of 2 nm radius, up to hydrostatic pressures of 20 MPa.
Introduction Organic-inorganic hybrid nanocomposites are materials whose chemical and physical properties are tailored by association of organic and inorganic moieties at the nanometer scale. Organic-inorganic hybrid nanocomposites have a wide range of applications in all the domains of technology: optics,1 biology,2 energy storage,3 catalysis,4 nanoreactors,5 separation technology,6 etc. One way among many others to synthesize such hybrid nanomaterials consists of grafting organic functions onto the surface of preformed inorganic nanoparticles. This can be achieved by reacting the inorganic particles directly with a reactant bearing the organic function of interest. The main challenge in such a procedure is to synthesize particles that both show a high reactivity to allow surface grafting and also show chemical resistance good enough to avoid alteration of the morphology of the particles. The stability of the hybrid materials requires the formation of strong chemical bonds between the inorganic network and the organic moieties. A large amount of work has been devoted to organically modified silica through the use of the strong silicon-carbon bond; however, only a few studies have been applied to other systems. In particular, the very stable metal-phosphate bond (M-O-P, where M ) Ti, Zr, Hf) can also offer a versatile alternative for new hybrid systems. We focus here on the zirconium case.
In past years, several authors have explored the modification of zirconium oxide surfaces with phosphates or phosphonates by various methods.7,8 The most relevant routes for the grafting of organic groups involve treatments of zirconia particles in the presence of a solution of phosphates or phosphonates at moderate temperatures (≈100 °C).9-12 The surface of zirconia is thus modified by the pendant organic groups attached by a variable number of Zr-O-P bonds. However, to our knowledge, only bulk materials or zirconia particles in the micron range have been studied to date. It seems that the challenging issue of a posteriori modification of nanometer-sized zirconia particles was not addressed yet. A detailed study on the mechanisms of grafting of various analogues of phosphonates onto the surface of titania nanoparticles was previously proposed by Guerrero et al.,13 but the authors did not detail the changes in the morphology of the particles upon grafting. The problem of the preservation of nanometric features is particularly critical in the case of zirconia particles, for it has already been known for decades that phosphate and phosphonates easily react with the zirconium oxide network to precipitate the analogues of the crystalline lamellar phase R-Zr(O3POH)2‚ H2O.14 The high surface reactivity of zirconia for phosphates and phosphonates may therefore be a major drawback if a high chemical resistance does not compen-
* Corresponding author. Phone: +33-1-6933-4663. Fax: +331-6933-3004. E-mail:
[email protected].
(7) Katz, H. E. Chem. Mater. 1994, 6, 2227-2232. (8) Ma, Y.; Tong, W.; Zhou, H.; Suib, S. L. Microporous Mesoporous Mater. 2000, 37, 243-252. (9) Schafer, W. A.; Carr, P. W. J. Chromatogr. 1991, 587, 137. (10) Clausen, A.; Carr, P. Anal. Chem. 1998, 70, 378. (11) Randon, J.; Blanc, P.; Paterson, R. J. Membr. Sci. 1995, 98, 119. (12) Gao, W.; Dickinson, L.; Grozinger, C.; Morin, F.; Reven, L. Langmuir 1996, 12, 6429. (13) Guerrero, G.; Mutin, P. H.; Vioux, A. Chem. Mater. 2001, 13, 4367-4373. (14) Alberti, G.; Toracca, E. J. Inorg. Nucl. Chem. 1968, 30, 317.
(1) Sanchez, C.; Lebeau, B.; Chaput, F.; Boilot, J.-P. Adv. Mater. 2003, 15, 1969-1994. (2) Sarikaya, M.; Tamerler, C.; Jen, A. K.-Y.; Schulten, K.; Baneyx, F. Nat. Mater. 2003, 2, 577-585. (3) Jannash, P. Curr. Opin. Colloid Interface Sci. 2003, 8, 96-102. (4) Wight, A. P.; Davis, M. E. Chem. Rev. 2002, 102, 3589-3614. (5) Stein, A.; Melde, B. J.; Schroden, R. C. Adv. Mater. 2000, 12, 1403-1419. (6) Sayari, A.; Hamoudi, S. Chem. Mater. 2001, 13, 3151-3168.
10.1021/la036249m CCC: $27.50 © 2004 American Chemical Society Published on Web 03/26/2004
3450
Langmuir, Vol. 20, No. 8, 2004
sate for it. The strategy in the study we present here is to compensate the reactivity of the particles with their good crystallinity. There is now an abundant amount of literature concerning the synthesis of well-crystallized zirconium oxide nanoparticles, for example by thermohydrolysis of zirconium chloride or acetate salts.15-20 We report here on the surface modification study of 60 nm porous particles of monoclinic zirconium oxide with various phosphates and phosphonates. The initial particles show porosity in the nanometer scale due to the aggregation of well-crystallized primary grains. The grafted particles show strong modifications of their surface properties with good preservation of their morphology. The exclusion of water from hydrophobic pores in the nanometer scale illustrates the efficiency of the functionalization, and is also used as a tool of characterization. Experimental Section Zirconia Colloid. The zirconia sols have been prepared by the method previously described by S. A. Matchett.19 A 20 mL volume of commercial zirconium acetate (16 wt % Zr, Aldrich Chemicals) was diluted in a mixture of 19 mL of distilled water and 25 mL of acetic acid. The pH of this clear solution was 2.1. It was placed in an autoclave and treated at 170 °C for 3 h. A cloudy sol was obtained that was purified by successive ultracentrifugation steps (11300g) for 1 h and redispersion in distilled water until no residual zirconium ions could be found in the solution. A test for the complete removal of monomeric or small oligomers of zirconium ions is that no precipitation occurs upon addition of phosphoric acid in the supernatant collected after centrifugation. The yield of the reaction is about 60%, according to the titration of Zr. Monooctylphosphoric acid ester PO(OH)2(OC8H17) was obtained by reaction of anhydrous phosphorus pentoxide (P4O10) in octanol at reflux temperature. In this synthesis, two consecutive reactions occur. First, solvolysis and opening of the P4O10 tetramer allow the formation of the mono- and diphosphoric acid esters in equal quantities as shown in reaction 1.21
P4O10 + 6RCH2CH2OH f PO(OCH2CH2R)2OH + PO(OCH2CH2R)(OH)2 (1) where R ) C6H13. Then, at the high boiling temperature of octanol, thermolysis of the esters occurs with an acid-catalyzed dehydration of the POC bond to form an alkene and phosphoric acid. The diester first transforms to the monoester, which is later decomposed into phosphoric acid following
PO(OCH2CH2R)2OH f PO(OCH2CH2R)(OH)2 + CH2dCHR (2) f PO(OH)3 + 2CH2dCHR Stopping the reaction after 7 days of reflux yielded, as the major product, the monoester which was separated from the orthophosphoric acid by extraction with an ether/water mixture (1:1 volume). The monoester was obtained as a solution in octanol with the diester as a byproduct (8%). Treatment with Phosphates or Phosphonates. After elimination of the free zirconium species, the colloidal solution (15) Clearfield, A. Inorg. Chem. 1964, 3, 146. (16) Mottet, B.; Pichavant, M.; Beny, J.-M.; Alary, J.-A. J. Am. Ceram. Soc. 1992, 75, 2515. (17) Hu, M.; Harris, M. T.; Byers, C. H. J. Colloid Interface Sci. 1998, 198, 87. (18) Lee, K.; Sathyagal, A.; Carr, P. W.; McCormick, A. V. J. Am. Ceram. Soc. 1999, 82, 338. (19) Matchett, S. A. U.S. Patent 5,037,579, 1991. (20) Vesteghem, H.; Charissou, I.; Reveron, H. Key Eng. Mater. 1997, 132-136, 129. (21) Corbridge, D. PhosphorussAn outline of its chemistry, biochemistry and technology, 5th ed.; Elsevier: New York, 1995.
Carrie` re et al. of zirconia was added to an aqueous solution of phosphoric or phenylphosphonic acid, or to octyl phosphate diluted in a 1:1 ethanol-octanol mixture. Without any other statement in the text, one shall assume that the mixture was heated under stirring at 100 °C for 4 h, with the following concentrations in solution: [ZrO2] ) 0.15-0.25 M and P/Zr ) 0.4-1. The unreacted species were then eliminated with several centrifugations and redispersions in pure water (ethanol/water in the case of octyl phosphate). The purification was repeated until no P atoms were detected in the supernatant collected after centrifugation, according to different techniques: precipitation upon addition of zirconium acetate (in the case of phosphoric acid), or recording of a 31P NMR spectrum (sensitivity: [P] ) 10-4 M). Experiments. 31P high-resolution liquid and solid MAS NMR spectra were recorded on a Bruker MSL 360 MHz spectrometer. Chemical shifts are referenced against 85% H3PO4. Nitrogen adsorption isotherms were recorded on a Micromeritics ASAP 2001 apparatus. Transmission electron microscopy (TEM) was performed on a JEOL JEM 100CX-II electron microscope on Cu grids with carbon membranes. Dynamic light scattering and zeta potentials were obtained on a Malvern 4500 DLS and 5000 Zetasizer, respectively. X-ray diffraction patterns were recorded on a Philips X-pert diffractometer. The complete chemical analysis of the solids was obtained by ICP atomic absorption spectroscopy at the Service d’Analyses Ele´mentaires, CNRS Vernaison (France). The small-angle X-ray scattering (SAXS) experiments were performed on the D2AM line at the European Synchrotron Radiation Facility as well as on two complementary setups available at the SCM-DRECAM-CEA laboratory.22 The scattering vector range of the first setup (classical) is q ) 0.02-0.5 Å-1. The range of the other setup (two-crystal camera) is q ) 0.001-0.1 Å-1. Every sample was examined in the two setups that allow the measurement of heterogeneities over a wide range of sizes extending from a few angstroms (2π/qmax) up to 1 µm (2π/qmin). The pressure-volume curves were performed with a laboratorybuilt equipment on 400 mg of powder placed in a closed cylinderplunger system with controlled displacement and hydrostatic pressure measurement. A typical cycling time was 1 h. Prior to the introduction of water in the cell (around 5 mL), the sample and the cell were degassed to eliminate the mechanical contribution due to the compression of the gases in the pores.
Results Characterization of the Particles. As already observed by other authors on similar systems,19,20 the hydrothermal treatment of the zirconium acetate solution and its purification yield a colloidal suspension of zirconium oxide particles. Dynamic light scattering indicates an average hydrodynamic radius of 65 ( 20 nm. The diffraction pattern of the solid collected after drying under vacuum at 60 °C is characteristic of monoclinic zirconia (Figure 1a). The diffraction peaks are broad, and the analysis of the peak width using a Williamson-Hall plot yields a diffraction coherence length on the order of 5 nm (not shown). This is confirmed by TEM: the dried solution consists of well-dispersed 60 nm particles (Figure 2a). A closer look at the TEM pictures indicates that the particles present pores formed by the aggregation of nanometric primary particles. Furthermore, higher magnifications show a texturing effect with an average alignment of the lattice planes of each primary particle all over an aggregate (Figure 2b). This so-called transgranular morphology has been previously observed on tetragonal zirconia synthesized in a similar way.23 The goal of this study is to achieve an efficient covering of the surface of these zirconia particles with organic species. It is therefore critical to check that the reactants can access as much surface in the particles as possible. (22) Zemb, T.; Tache´, O.; Ne´, F.; Spalla, O. Rev. Sci. Instrum. 2003, 74, 2456. (23) Deschamp, M.; Djuricic, B.; Pickering, S. J. Am. Ceram. Soc. 1995, 78, 2873.
Covalent Grafting on Porous Zirconia Nanoparticles
Langmuir, Vol. 20, No. 8, 2004 3451
Figure 3. Small-angle X-ray scattering pattern of (a) a colloidal zirconia solution and (b) the same after drying. Table 1. Parameters Used for the Interpretation of SAXS Measurements
Figure 1. X-ray powder diffraction pattern (Cu KR) of dried colloidal solution: (a) pure, (b) after treatment with diluted phosphoric acid (4 h, 100 °C, [H3PO4] ) 0.10 M, [ZrO2] ) 0.15 M), and (c) after treatment with concentrated phosphoric acid (24 h, 100 °C, [H3PO4] ) 4 M, [ZrO2] ) 0.15 M).
FZrO2 (cm-3) Fmedium (cm-3) µZrO2 (g‚cm-3) zirconia/water 1.59 × zirconia/air 1.59 × 1024 1024
SAXS is well adapted for this purpose, for it allows characterization of the morphology of the particles both in situ (i.e., in the colloidal solution before any drying) and after drying of the particles. Figure 3a shows the absolute intensity of a X-ray beam scattered by a colloidal solution of zirconia ([ZrO2] ) 0.22 mol‚L-1). The scattered intensity shows three separated features. The kink A in the scattering curve at q ) 10-2 Å-1 is the signature of the outer envelope of the 60 nm zirconia aggregates. This size is consistent with the hydrodynamic radius of the particles measured by dynamic light scattering. The correlation B around 6 × 10-2 Å-1 (3-10 nm) is attributed to the morphological features internal to the aggregates, i.e., the interface between the nanometric primary particles and the pores filled with solvent. At higher angles (q >
5.8 5.8
φs 5.36 × 10-3 NA
2 × 10-1 Å-1), the scattered intensity decays following a Porod law (I(q) ∝ q-4). This indicates a smooth interface between the particles and the solvent. After drying, the morphological properties of the particles change (Figure 3b). At small angles (q < 4 × 10-3 Å-1) the intensity decreases following a Porod law. This behavior is attributed to a smooth interface between these large agglomerates and the air, and is the consequence of the packing of the 60 nm aggregates into larger agglomerates during the drying. The correlation A′ at q ) 1.5 × 10-2 Å-1 is attributed to the contrast between the air and the aggregates packed together. The aggregate particles are no longer individualized by SAXS, and the corresponding correlation length of 40 nm is attributed to the electronic contrast between a continuous network of aggregates and the pores within it. The correlation B′ around q ) 1.5 × 10-1 Å-1 is attributed to the electronic contrast between the primary grains and the internal pores of the aggregates, filled with air. From this first qualitative evaluation, one observes that the 60 nm aggregates show an internal porosity both before and after drying. The large specific area resulting from this porosity is quantitatively evaluated according to the equation24
Σ(q) )
Figure 2. Transmission electron microscopy of zirconia particles.
3.33 × 0
1023
q4I 2πµZrO2φv(lT∆F)2
(3)
where µZrO2 is the density of the zirconia; φv is the volume fraction of zirconia in the sample (for the solution, gravimetrically measured; for the powder, determined according to a procedure based on its optical density at the X-ray wavelength, as described elsewhere25); lT is e e Thomson’s length (2.82 × 10-15 m); ∆F ) FZrO - Fmatrix is 2 the difference in electron density between the zirconia and the surrounding medium (water or air). The high q limit of the Σ(q) function gives the total surface area of the zirconia/water or zirconia/air interface. The parameters used to calculate the functions Σ(q) are given in Table 1. The functions Σ(q) are plotted for the colloidal suspension and the dried solution in curves a and b, respectively, of Figure 4. (24) Porod, G. In Small-angle X-ray scattering; Glatter, O., Kratky, O., Eds.; Academic Press: New York, 1982. (25) Deruelle, O.; Spalla, O.; Barboux, P.; Lambard, J. J. Non-Cryst. Solids 2000, 261, 237.
3452
Langmuir, Vol. 20, No. 8, 2004
Carrie` re et al.
Figure 4. Normalized Iq4 function (q4I/[2πµzrO2φv(lr∆F)2]) for (a) the colloidal zirconia solution and (b) the same after drying. Table 2. Summary of Surface Properties of the Initial Zirconia Nanoparticles, and of the Particles After Various Graftings system
BET area (m2‚g-1)
P/Zr ratio
C parama
ZrO2 ZrO2-(HO)2P(O)OH ZrO2-(HO)2P(O)C6H5 ZrO2-(HO)2P(O)OC8H17
150 103 122 115
NA 0.26 0.24 0.23
130 212 82 52
a
Calculated from the BET model; see eq 4.
In the colloidal solution, we observe two plateaus in the 2 × 10-2 and 3 × 10-1 Å-1 ranges, that respectively give the surface area of the envelope of the aggregate (15 m2‚g-1) and that of the primary particles (450 m2‚g-1). In the case of the powders, there is a first plateau at very low angles (2 × 10-3 Å-1 range) that gives the specific area of the packets of aggregates (0.06 m2/g). The second plateau shows that the area of the aggregates does not significantly change upon drying (15 m2‚g-1 in the 2 × 10-2 Å-1 range). On the contrary, the specific area of the primary particles of zirconia considerably decreases after drying: the plateau in the 3 × 10-1 Å-1 range gives a value of 160 m2‚g-1. This demonstrates that the drying of the particles drastically decreases their specific area, and that this decrease comes essentially from a loss in surface in the inner pores of the aggregates, probably due to a collapsing under the capillary pressure. This observation also emphasizes the interest of in situ techniques such as X-ray methods for the study of colloids and porous materials to avoid sample alteration.25 The system was further investigated in the dried state with nitrogen isotherms at 77 K. The interpretation of the desorption curve with the BET model gives a specific area of 150 m2‚g-1 (Table 2). This value is close to the total area measured with SAXS on the same dried particles. This demonstrates that, although the surface of the aggregate is still accessible to nitrogen molecules, the drying partly closes the inner pore network. Thus a higher surface covering is expected if the grafting is performed on the particles before any drying. Grafting of Phosphates and Phosphonates onto the Zirconia Particles. When the zirconia particles are treated for 24 h at 100 °C with concentrated phosphoric acid ([H3PO4] ) 4 M and [ZrO2] ) 0.15 M in the H3PO4/ ZrO2 mixture), the X-ray diffraction (XRD) pattern of the particles is strongly modified (Figure 1c). It shows that the initial monoclinic structure of the zirconia phase is transformed to that of the lamellar R-Zr(O3POH)2‚H2O phase. This reaction is expected after treatment in such extreme chemical conditions.26 Furthermore, the BET specific area of this material is extremely low (5 m2‚g-1) as compared to that of the ungrafted particles (150 m2‚g-1). (26) Stynes, J. A. J. Inorg. Nucl. Chem. 1964, 26, 117.
Figure 5. NMR MAS 31P of zirconia particles treated with (a) phosphoric acid, (b) phenylphosphonic acid, and (c) octyl phosphate.
On the contrary, the XRD pattern of the particles treated in milder chemical conditions, such as lower phosphoric acid concentration ([H3PO4] ) 0.1 M, 4 h at 100 °C) is very similar to the initial XRD pattern (Figure 1b). The crystalline core of the particles is not dissolved upon treatment. As a consequence, the BET specific area remains high (103 m2‚g-1, Table 2). In a similar way, the zirconia particles treated under reflux for 4 h with phenylphosphonic acid or with octyl phosphate show no dissolution, and their X-ray diffraction pattern remains unchanged (not shown). The BET surface areas of these compounds also remain large, around 120 m2‚g-1 (Table 2). This chemical treatment therefore allows preserving the morphology of the particles. In the meantime, those conditions allow adsorption of a significant amount of phosphates or phosphonates onto the particles. Chemical analysis of the as-prepared zirconia particles yields a composition ZrO1.90(CH3COO)0.20‚0.57H2O. After treatment of these zirconia particles at 100 °C for 4 h with dilute phosphoric acid and washing, a solid of composition ZrO1.86(CH3COO)0.03(O2P(OH)2)0.26‚0.65H2O is obtained. This indicates that the interaction between the particles and the phosphoric acid is strong enough to replace all the initial acetate surface groups, and to retain a significant amount of phosphate groups even after complete purification. Treatment of the particles with phenylphosphonic acid or with octyl phosphate gives similar results. The achieved P/Zr ratios are then 0.24 and 0.23, respectively (Table 2). This shows that the monosubstitution on the phosphorus atom does not change significantly the reactivity toward zirconia. The nature of the binding of the phosphate groups was investigated with 31P MAS NMR. The spectrum of the particles treated with phosphoric acid shows a broad distribution of signals, with distinguishable contributions around -6, -12, and -21 ppm (Figure 5a). These peaks are respectively attributed to phosphates bound to the particles through one, two, and three Zr-O-P bonds (respectively denoted as Q1, Q2, and Q3 species).27 The treatment with phosphoric acid allows the replacement of all initial acetate groups by phosphate groups, most probably in the H2PO4- form. These phosphate groups may be bound either to one surface zirconium atom (terminal or chelating) or to two or three surface atoms (bridging). It is also probable that the thermal treatment allows opening of the surface Zr-O-Zr bonds, thus forming additional Q3 species. Similarly, the phenylphos(27) Clayden, N. J. J. Chem. Soc., Dalton Trans. 1987, 1877.
Covalent Grafting on Porous Zirconia Nanoparticles
Langmuir, Vol. 20, No. 8, 2004 3453
Figure 6. Zeta potentials of (a) the pure colloidal zirconia solution, (b) the same after addition of H3PO4, (c) after addition of phenylphosphonic acid, and (d) after addition of octyl phosphate.
phonic acid and the octyl phosphate are bound to the particles through one, two, and three Zr-O-P covalent bonds, respectively denoted as T1, T2, and T3 (Figure 5b,c). In any case, the various treatments allow the covalent grafting of different functions at the surface of the particles, namely P-OH, P-C6H5, and P-OC8H17 groups, plus additional P-OH dangling bonds. Those pendant bonds are responsible for the broadening of the peaks corresponding to the Q1, Q2, Q3, T1, and T2 species. Modification of the Surface Properties. This covalent modification of the surface drastically changes some surface properties of the particles: surface charge in solution, polarizability in the dried state, and hydrophobic character. First, one observes significant changes in the zeta potential (ζ) as soon as the reactants are added to the sol of zirconia. After the synthesis and the complete removal of the free zirconium species, the zeta potential of the as-prepared particles is positive (ζ ≈ 65 mV at pH 3.4, Figure 6a). The acetate groups detected by chemical analysis act as negative counterions in solution. However, their interaction with the particles is too weak to allow neutralization of the charge in aqueous medium. Addition of phosphoric acid leads to a reversal of the charge surface at a similar pH value (ζ ≈ -20 mV at pH 3.2, Figure 6b), and addition of phenylphosphonic acid or octylphosphonate leads to a neutralization of the charge (ζ ≈ 0 mV at pH 3.1, Figure 6c,d). This demonstrates that, contrary to the acetate ions, the phosphate and phosphonates molecules are adsorbed within the shear plane of the zirconia nanoparticles. Another property that shows significant changes upon grafting is the polarizability of the surface of the dried particles. It can be evaluated with nitrogen adsorption measurements at 77 K.28 At low nitrogen pressures, one should observe the following relation:
C-1 P P 1 + ) V(P - P0) AV0C AV0C P0
(4)
where P is the nitrogen pressure; P0 is the saturation nitrogen pressure; V0 is the volume of one nitrogen monolayer (0.250 cm3‚m-2); A is the area of the material; C is a parameter proportional to eE1-EV, where E1 is the adsorption enthalpy of the first nitrogen layer and EV is the vaporization enthalpy of nitrogen. The parameter C is characteristic of the affinity of nitrogen for the surface. A high C value should be observed on surfaces with high surface polarizability and vice versa. (28) Brunauer, S.; Emmet, P. H.; Teller, E. J. Am. Chem. Soc. 1938, 60, 309.
Figure 7. Intrusion-extrusion curve of water into zirconia particles treated with phenylphosphonic acid.
Figure 8. Intrusion-extrusion curve of water into zirconia particles grafted with octyl phosphate: (a) first compression, (b) decompression, and (c) second compression.
The values of C after various treatments are reported in Table 2. One observes that the polarizability of the surface decreases with the grafting in the following order: -POH > as-prepared ZrO2 > -PC6H5 > -POC8H17. The surface polarizability of the as-prepared zirconia powder is attributed to surface hydroxyls, to the adsorbed acetate groups, and to the various Lewis sites commonly detected at the surface of zirconia particles.29 The polarizability after grafting of phosphates or phosphonates simply reflects the polarizability of the various pendant functional groups. One should be aware that the treatment with phenylphosphonic acid and octyl phosphate brings organic groups with a low polarizability, but also additive dangling P-OH bonds. The decrease in polarizability upon grafting of organic groups is tightly related to the hydrophobic modification of the surface. One original and instructive way to characterize the hydrophobic properties of porous materials is to perform isothermal intrusion of water.30 The intrusion-extrusion curve of the zirconia grafted with phenylphosphonic acid is shown in Figure 7. This curve shows no particular feature apart from the compressibility of the apparatus and that of water. No pore excluding water could therefore be detected. On the contrary, the intrusion curve of the zirconia grafted with octyl phosphate shows a plateau around 20 MPa (Figure 8a). This plateau corresponds to the intrusion of water in hydrophobic pores. If we assume that these pores are spherical, Laplace’s relation gives their radius r:
r)-
2γ cos θ P
(5)
where γ is the interfacial tension of water (72.6 mN‚m-1);
3454
Langmuir, Vol. 20, No. 8, 2004
Figure 9. Pore volume distribution in the zirconia particles treated with octyl phosphate (a) derived from the intrusionextrusion curve of water and (b) derived from the nitrogen adsorption at 77 K and the BJH model.
θ is the contact angle between the grafted zirconia and water; P is the pressure of intrusion. The extrusion curve (Figure 8b) does not show any plateau that would correspond to the expulsion of water from the pores, and the following compression curves show no further intrusion of water in the pores (Figure 8c). However, after equilibrating in water at ambient pressure for several days, the sample shows again intrusion of water in the pores. This demonstrates that the extrusion phenomenon is kinetically limited in the system. The zirconia colloidal solution can be deposited by dip coating onto a glass substrate and treated afterward with octyl phosphate. This allows the contact angles with water to be measured. As-prepared zirconia gives a contact angle of 25° ( 2°, whereas after grafting of octyl phosphate the contact angle is found to be θ ) 105° ( 2°. From the difference between the intrusion curve and the extrusion curve, it is then possible to calculate the distribution of hydrophobic pores in the material (Figure 9a). This distribution can also be obtained by the analysis of the desorption branch of a nitrogen isotherm with the Barrett-Joyner-Halenda (BJH) model (Figure 9b). Both distributions are centered on a pore diameter of 2 nm, which indicates a qualitative agreement between both measurements. However, the total hydrophobic volume (0.08 cm3‚g-1) is larger than the total BJH volume (0.05 cm3‚g-1). This discrepancy is due to the inaccuracy of the BJH model for such small pore radii.31 Discussion The goal of this study was to achieve the grafting of various functional species onto supporting particles, with a large surface concentration and a good conservation of the morphology (particle size, pores). The first part of the study shows that the choice of the particles was relevant. (29) Nawrocki, J.; Rigney, M. P.; McCormick, A.; Carr, P. J. Chromatogr. A 1993, 657, 229. (30) Gomez, F.; Denoyel, R.; Denoyel, J. Langmuir 2000, 16, 4374. (31) Barrett, E. P.; Joyner, L. G.; Halenda, P. P. J. Am. Chem. Soc. 1951, 73, 373.
Carrie` re et al.
Indeed, the aggregation of the primary particles into the bigger 60 nm aggregates defines a large and accessible surface within the particles as shown by SAXS and BET measurements (450 m2‚g-1 in the colloidal solution). However, one observes a loss of surface upon drying (150160 m2‚g-1) probably due to the collapsing of the internal pores under the capillary pressure. This indicates that a better surface coverage is expected if the particles are treated in solution before any drying. The treatment of the particles with phosphates or phosphonates in mild conditions allows reaching noticeably larger surface concentration of grafted species (0.25 ( 0.02 P/Zr). However, the density of acetate groups at the surface of as-prepared zirconia particles is close to 3 µmol‚m-2 as calculated from the chemical analyses (0.20 acetate/Zr) and the surface area of the zirconia colloid before drying as determined by SAXS (450 m2/g). Each acetate group is probably associated with a positively charged surface defect. This high concentration of initial surface defects on the particles is therefore a plausible reason for the large concentration of phosphorus atoms that substitute all acetate groups and even more. The surface concentration of phosphates grafted onto our nanoparticles is in all cases around 4 µmol‚m-2, which is nearly the same as the 4-6 µmol‚m-2 concentrations previously reported in the literature after the treatment of micron-sized zirconia with phosphates.9-11 Therefore, this value probably corresponds to the concentration of ionizable surface defects of well-crystallized zirconia, independent of its particle size. In addition, the crystalline core and the BET area of the dried particles are preserved after grafting (100-150 m2‚g-1, Table 2), and no crystalline phase additional to the zirconia can be detected (Figure 1b). As expected, the use of well-crystallized particles allows a good preservation of the morphological properties and compensates for the surface reactivity of the particles. The surface properties of the zirconia particles are strongly modified upon grafting. In solution, the surface of the as-prepared zirconia particles is positively charged, as revealed by zeta potential measurements (Figure 6). This indicates that the interaction between the zirconia surface and the acetate counterions is too weak to retain the CH3COO- species within the shear plane of the colloidal particle. On the contrary, the negative potential measured after addition of phosphate indicates that there is a strong interaction between the zirconia particle and the phosphate or phosphonate species, which allow both retention of the phosphated species and their consecutive dissociation:
ZrO+ + AcO- + H3PO4 f ZrOPO(OH)2 + AcOHv f ZrOPO2(OH)- + H+ (6) The zeta potential after addition of phenylphosphonic acid and of octyl phosphate is close to zero. This shows that one P-OH bond is dissociated and neutralizes all the charges on the surface of the zirconia, but that the acidity of the second P-OH bond is too weak to trigger a second dissociation. The grafting of these species can be described by the following scheme:
ZrO+ + AcO- + (HO)2OPR f ZrOPO(OH)R + AcOH (7) where R ) -C6H5, -C8H17. The zeta potential measurements therefore show that the acidity of the residual P-OH bonds is weak as opposed
Covalent Grafting on Porous Zirconia Nanoparticles
to the acidity of the zirconia grafted with phosphoric acid. This was also evidenced in the solid state. As reported elsewhere, the zirconia particles grafted with phosphoric acid show significant proton conductivity due to the high density of acid P-OH bonds at the surface of the particles (σ ≈ 10-5 S‚cm-1).32 Such a proton conductivity is not measurable after grafting of phenylphosphonic acid or octyl phosphate (σ < 10-9 S‚cm-1), despite the presence of the residual P-OH bonds detected by MAS NMR (Figure 5). The various acidities and the number of free P-OH bonds at the surface of the particles can partially explain the sequence in polarizabilities derived from the BET model (-POH > as-prepared ZrO2 > -PC6H5 > -POC8H17). The highest polarizability is expected for the most acidic surface, i.e., zirconia grafted with phosphoric acid. On the other hand, the lowest polarizabilities are expected for the surfaces modified with organic components. There is a striking difference in the experiments of intrusion of water between the zirconia grafted with phenylphosphonic acid and the zirconia grafted with octyl phosphate (Figure 8). This difference does not come from differences in surface concentrations (≈4 µmol‚m-2 in both cases), but rather from intrinsic properties of both functional groups such as their polarizibilities. The exclusion of water from the nanometric pores of the zirconia is only achieved with the grafting of the most hydrophobic functional group. (32) Carrie`re, D.; Moreau, M.; Lahlil, K.; Barboux, P.; Boilot, J. P. Solid State Ionics 2003, 162-163, 185.
Langmuir, Vol. 20, No. 8, 2004 3455
Conclusions Nanometric and well-crystallized zirconia can be obtained by hydrothermal synthesis of zirconium acetate. The particles show a large specific area and a high reactivity of the surface sites toward grafting of phosphates and phosphonates. Grafting occurs under mild conditions without alteration of the crystalline structure of the primary grains. The functional groups are bound to the surface through a variable number of Zr-O-P bonds. The high concentration of reactive surface groups and the good crystallinity of the particles allow both a complete coverage of the surface and a good preservation of the morphology of the particles (BET surface 100-150 m2‚g-1 after grafting and drying). The grafting of different functional groups allows tuning the polarizability of the surface of the particles, according to the following decreasing sequence: -POH > as-prepared ZrO2 > -PC6H5 > -POC8H17. Furthermore, the preservation of the morphology of the particles allows obtaining after grafting of octyl phosphate 2 nm porous structures that exclude water up to hydrostatic pressures of 20 MPa. Acknowledgment. We thank F. Bley (CEA Grenoble) for technical assistance on the D2AM synchrotron beamline. TEM was performed by P. Beaunier at the electron microscopy center of University Pierre-et-Marie Curie (Paris). LA036249M
