Nb2O5 Prepared

Aug 31, 2004 - Instituto de Fı´sica Gleb Wataghin, Universidade Estadual de Campinas, P.O. Box 6165,. 13084-971 Campinas SP, Brazil. Y. V. Kholin...
0 downloads 0 Views 272KB Size
Langmuir 2004, 20, 8707-8714

8707

Surface Modification with Phosphoric Acid of SiO2/Nb2O5 Prepared by the Sol-Gel Method: Structural-Textural and Acid Sites Studies and an Ion Exchange Model M. S. P. Francisco,* W. S. Cardoso, and Y. Gushikem Instituto de Quı´mica, Universidade Estadual de Campinas, P.O. Box 6154, 13084-971 Campinas SP, Brazil

R. Landers Instituto de Fı´sica Gleb Wataghin, Universidade Estadual de Campinas, P.O. Box 6165, 13084-971 Campinas SP, Brazil

Y. V. Kholin Chemical Faculty, V.N. Karazin Kharkov National University, 4 Svoboda Square, Kharkov 61077, Ukraine Received April 20, 2004. In Final Form: July 22, 2004 In this work, the structural and textural properties of the SiO2/Nb2O5 system prepared by the sol-gel method and then modified by phosphoric acid were studied. The different materials were prepared, with three different mol % Nb2O5 (2.5, 5.0, and 7.5 mol %), and calcined in the temperature range of 423-1273 K. BET specific surface area determinations, scanning electron microscopy connected to a X-ray emission analyzer, Fourier transform infrared spectroscopy, and X-ray photoelectron spectroscopy (XPS) were used for the investigation. For the lowest temperature of calcination (423 K), the mesopores and micropores of the modified material were blocked, resulting in a decrease of the specific surface area compared to the SBET values obtained for the SiNb matrix. Under intermediate temperatures of calcination (423-873 K), the modified material acquired textural stability. By XPS analysis, the presence of the dihydrogenphosphate species was identified, the P/Nb atomic ratios being independent of the thermal treatment. 31P magic angle spinning NMR confirmed the XPS data and also showed that the chemical shift of the (H2PO4)- ions strongly depended on the crystallization degree of the Nb2O5. Structural thermal stability was also shown by the presence of Brønsted acid sites in the modified material calcined at high temperature (1273 K). The thermal stability is directly associated with obtainment of the same value for K+ exchange capacity (0.74 mmol g-1, average value) for the modified materials calcined at 423 and 1273 K. The chemical analyses of phosphorus for the modified materials were made by using the inductively coupled plasma. The value was 0.36 mmol g-1, corroborating the presence of (H2PO4)- ions. The ion exchange isotherms presented an S-shaped form characteristic of energetically heterogeneous ion exchangers, permitting application of a model of fixed polydentate centers, in which ion exchange took place.

Introduction In mixed oxides of the type SiO2/MxOy, prepared by the sol-gel processing method, the metallic oxides are, in general, highly dispersed in the silica matrix, exposing large surface areas suitable to be modified through chemical reactions. Many reactions have been made by immersing SiO2/MxOy [M ) Zr(IV) and Ti(IV)] in H3PO4 or in Sb(V) solutions.1-4 As a result, solid Brønsted acids such as SiO2/MO2/phosphate or SiO2/MO2/Sb2O5 have been obtained.1-4 There is particular interest in niobium oxide impregnated with phosphate, obtained by immersing bulk phase Nb2O5 in a phosphoric acid solution, for use as heterogeneous catalysts5-8 or as ion exchangers.9-12 The solid * Corresponding author. Fax: 55-19-3788 3023. E-mail: suzana@ iqm.unicamp.br. (1) Alfaya, A. A. S.; Gushikem, Y.; deCastro, S. C. Chem. Mater. 1998, 10, 909. (2) Gonc¸ alves, J. E.; Gushikem, Y.; deCastro, S. C. J. Non-Cryst. Solids 1999, 260, 125. (3) Alfaya, A. A. S.; Gushikem, Y.; deCastro, S. C. Microporous Mesoporous Mater. 2000, 39, 57. (4) Zaitseva, G.; Gushikem, Y. J. Braz. Chem. Soc. 2002, 13, 611. (5) Tanabe, K. Catal. Today 2003, 78, 65.

obtained from this procedure, submitted to heat treatment, maintains its surface acidity and catalytic activity up to about 870 K,13-16 while for pure niobium oxide surface acidity was maintained up to about 670 K.12,17 Above these temperature limits, losses of surface acidities were observed in both cases due to amorphous to crystalline (6) Tanabe, K.; Okazaki, S. Appl. Catal. A 1995, 133, 191. (7) Ziolek, M. Catal. Today 2003, 78, 47. (8) Kishor, N.; Fujiwara, M. Chem. Commun. 2002, 2702. (9) Wang, X.; Liu, L.; Jacobson, A. J. J. Mater. Chem. 2002, 12, 1824. (10) Wang, X.; Liu, L.; Jacobson, A. J. J. Mater. Chem. 2000, 10, 2774. (11) Wang, X.; Liu, L.; Cheng, H.; Jacobson, A. J. Chem. Commun. 1999, 2531. (12) Okazaki, S.; Kurosaki, A. Catal. Today 1990, 8, 113. (13) Jones, D. J.; Aptel, G.; Brandhorst, M.; Jacquin, M.; JimenezJime´nez, J.; Jime´nez-Lo´pez, A.; Maireles-Torres, P.; Piwonski, I.; Ridrı´guez-Castello´n, E.; Zajac, J.; Roziere, J. J. Mater. Chem. 2000, 10, 1957. (14) Quartararo, J.; Guelton, M.; Rigole, M.; Amoureux, J.-P.; Fernandez, C.; Grimblot, J. J. Mater. Chem. 1999, 9, 2637. (15) Farias, A. M. D.; Gonzalez, W. A.; Oliveira, P. G. P.; Eon, J.-G.; Herrmann, J.-M.; Aouine, M.; Loridant, S.; Volta, J.-C. J. Catal. 2002, 208, 238. (16) Armaroli, T.; Busca, G.; Carlini, C.; Giuttari, M.; Galletti, A. M. R.; Sbrana, G. J. Mol. Catal. A 2000, 151, 233. (17) Kurosaki, A.; Okuyama, T.; Okazaki, K. Bull. Chem. Soc. Jpn. 1987, 60, 37.

10.1021/la049000t CCC: $27.50 © 2004 American Chemical Society Published on Web 08/31/2004

8708

Langmuir, Vol. 20, No. 20, 2004

Francisco et al.

Table 1. Chemical Analysesa of Nb and P in the SiNbP Samples Thermally Treated at 423 K samples

Nb,b mmol g-1

P, mmol g-1

P/Nb ratios

SiNbP1 SiNbP2 SiNbP3

0.54 0.99 1.55

0.10 0.24 0.36

0.18 0.24 0.23

a

Error of (1%. b Reference 18.

phase transition processes. The adsorbed phosphate is responsible for modifying the textural properties as well as the acidic character by decreasing the thermal mobility of niobium oxide and increasing its crystallization temperature. In the present work, the structural-textural and acid site properties of niobium oxide in sol-gel matrixes, SiO2/Nb2O5, presenting different Si:Nb mole ratios and treated with phosphoric acid, are reported. The solids treated by immersing in phosphoric acid solutions were studied after thermal treatments from 423 up to 1273 K. BET specific surface areas, scanning electron microscopy with X-ray emission analysis (EDS), Fourier transform infrared (FT-IR) spectroscopy, and X-ray photoelectron spectroscopy (XPS) were used to investigate these properties. These materials were employed as ion exchangers. The exchange capacities were obtained, and an ion exchange model is proposed. Both the sol-gel method and the subsequent immersion of the matrix in the phosphorus precursor were shown to be adequate procedures to obtain high and stable ratios of acid sites. This is based on the properties of the matrixes: high specific surface area, maintenance of the amorphous phase at high treatment temperatures (1273 K), and the formation of Brønsted sites with thermally stable Si-O-Nb and Si-OH-Nb linkages.18,19 Application of the sol-gel processing method for SiO2/Nb2O5 preparation is recent.20 The matrix prepared by our sol-gel route presented a very high specific surface area, with niobia species highly dispersed on the silica as Si-O-Nb linkages.18,19 Experimental Section Phosphorus Species Adsorption in the SiO2/Nb2O5 System. The experimental procedures to prepare SiO2/Nb2O5 matrixes by the sol-gel processing method have been described elsewhere.18 About 2 g of the SiO2/Nb2O5 matrixes, presenting variable mol percentages of Nb2O5 (2.5, 5.0, and 7.5%, denoted SiNb1, SiNb2, and SiNb3 matrixes, respectively), were immersed in 20 mL of 1 mol L-1 phosphoric acid solution, and the mixture was shaken for 4 h at 348 K. The solids obtained from these treatments were exhaustively washed with twice-distilled water until all excess acid was removed, centrifuged, and dried at 383 K for 3 h. The final materials were designated as SiNbP1, SiNbP2, and SiNbP3 samples, respectively. The results of the chemical analyses and the atomic ratios obtained are listed in Table 1. The solids were then heat treated at 423, 573, 873, and 1273 K before further tests. Characterizations. Chemical Analyses. The chemical analyses of phosphorus in the matrixes were made by using the inductively coupled plasma technique on a Perkin-Elmer model 3000 DV instrument. The samples were analyzed as follows: about 0.1 g of the solids were treated with 2 mL of concentrated HCl solution and gently heated on a steam bath while 0.4 mL of concentrated HF was added. The excesses of hydrofluoric acid (18) Francisco, M. S. P.; Gushikem, Y. J. Mater. Chem. 2002, 12, 2552. (19) Francisco, M. S. P.; Landers, R.; Gushikem, Y. J. Solid State Chem. 2004, 177, 2432. (20) Morselli, S.; Moggi, P.; Cauzzi, D.; Predieri, G. In Preparation of Catalysts VII; Delmon B., Ed.; Elsevier: Amsterdam, 1998; pp 763772.

were evaporated. The solutions were submitted to chemical analyses in triplicate with an error e1%. N2 Physical Adsorption at 77 K. The specific surface areas calculated by the BET method (SBET) were measured in a Micromeritcs model Flowsorb II 2300 instrument connected to a flow controller. Prior to measurement, all the samples were outgassed at about 10-4 Pa at 423 K for 8 h. Scanning Electron Micrographs. Scanning electron microscopy was carried out using low vacuum microscopy (JSM 5900LV), operating at an accelerating voltage of 25 kV. The images were obtained by using the secondary electrons. EDS was used for elemental mapping in a Noram Voyager instrument. The samples were fixed onto double-faced tape adhered to an aluminum support and coated with a layer of gold (about 15 nm) by a BalTec SCD 050 Sputter Coater apparatus (60-mA current for 60 s). The micrographs and the elemental mapping were obtained only for the SiNbP3 sample. XPS. The XPS spectra were obtained on a VSW HA-100 spherical analyzer using an aluminum anode (Al KR line, hν ) 1486.6 eV). The high-resolution spectra were measured with constant analyzer pass energies of 44 eV, which produce a full width at half-maximum (fwhm) line width of 1.8 eV for the Au(4f7/2) line. The powdered samples were pressed into pellets and fixed to a stainless steel sample holder with double-faced tape and analyzed without further preparation. The binding energy (BE) at 103.5 eV of Si(2p) was used as the absolute energy for internal reference. Curve fitting was performed using Gaussian line shapes, and a linear background was subtracted from the data. The atomic compositions were estimated from the C(1s), O(1s), P(2p), Si(2p), and Nb(3d) peak areas, using cross sections for photoionization from the literature.21 Magic Angle Spinning (MAS) NMR Measurements. 31P MAS NMR spectra were obtained with a Varian Inova 500 spectrometer at room temperature using an 85 wt % phosphoric acid solution as the external standard. Conventional spectra were obtained at 202.4 MHz, and the pulse width was π/4 with an interval between the pulses and a spinning speed of 5 s and 5800 Hz, respectively. Acid Sites. The Lewis and Brønsted acid sites of the SiNbP3 sample, previously heat treated between 423 and 1273 K, were investigated after pyridine adsorption followed by drying at low pressure (∼10-5 Pa for 30 min) in the temperature range from 298 to 523 K. The FT-IR spectra in the range of 4000-400 cm-1 were obtained for the sample pressed as a self-supported disk (about 12.0 mg) under a static vacuum using 100 accumulative scans with a resolution of 2 cm-1. The equipment used was a Bomen MB series FT-IR spectrophotometer. Ion Exchange Isotherms. To determine the ion exchange capacity, the SiO2/Nb2O5/phosphate samples having 1.55 mol g-1 of Nb (SiNbP3 sample thermally treated at 423 and 1273 K) were used as representative samples. To obtain the isotherms, 0.1 g of the samples was immersed in 25 mL of KCl solutions whose concentrations varied from 10-4 to 10-2 mol L-1 and shaken in a thermostated bath at 298 K for 30 h. The potassium ion contents in the supernatant solutions were determined using a model DM-61 flame photometer from Digimed. The amounts of the exchanged potassium ions were determined by applying the equation

Nf )

(ni - ns) m

(1)

where ns and ni are the equilibrium condition and initial mole number of the potassium ions, respectively, and m is the mass of the adsorbent.

Results and Discussion Chemical Analysis, SBET Values, and Niobia Dispersion in the SiNb Matrixes. The chemical analysis results for the SiNb matrixes18 and for SiNbP samples thermally treated at 423 K are presented in Table 1 and reveal that the relationship between the amount of adsorbed phosphate on the matrix surfaces and the Nb (21) Scofield, J. H. J. Electron Spectrosc. 1976, 89, 129.

Structural-Textural Properties of SiO2/Nb2O5

Figure 1. Specific surface areas (SBET values) for the SiNb matrixes and for the SiNbP samples after thermal treatment in the temperature range of 423-1273 K (designed Tcal): (9) SiNbP1, (b) SiNbP2, and (2) SiNbP3 samples; (0) SiNb1, (O) SiNb2, and (4) SiNb3 matrixes.

contents is not straightforward. Figure 1 presents the BET specific surface areas (SBET values) for the SiNbP samples after thermal treatment in the temperature range of 423-1273 K. For a better comprehension of the influence of phosphorus adsorption on the textural properties, the SBET values for the SiNb matrixes treated in the same temperature range are also included.18 The P/Nb mole ratios are 0.18, 0.24, and 0.23 for the SiNbP1, SiNbP2, and SiNbP3 samples, respectively, while the Nb contents (in mmol g-1) in the SiNb1, SiNb2, and SiNb3 matrixes are 0.54, 0.99, and 1.55, respectively. On the other hand, the specific surface areas for samples (SBET) previously heat treated at 423 K are (Figure 1) SiNbP1 ) 600, SiNbP2 ) 570, and SiNbP3 ) 470 m2 g-1 and for the precursor SiNb matrixes previously heat treated under similar conditions the areas are SiNb1 ) 950, SiNb2 ) 740, and SiNb3 ) 710 m2 g-1. Comparing the BET areas of SiNbP samples with those of the corresponding precursor SiNb matrixes, a great reduction took place upon treatment with phosphoric acid, that is, about 37% for SiNbP1, 23% for SiNbP2, and 32% for SiNb3. The reduction of the areas can be expected because the adsorption process was by the immersion of the already prepared matrixes. The phosphate species formed by the reactions occurs at the surfaces, blocking some of the microand mesopores of the matrixes, while the phosphate species on the surface stabilized the texture of the material. In Figure 1 it can be observed that, between 423 and 873 K, the specific surface areas remain practically constant for SiNbP samples, in contrast to the considerable decrease in area observed for SiNb matrixes. Similar results were obtained by Wagray and Ko,22 who used the immersion process on a niobia xerogel in 0.25 mol L-1 orthophosphoric acid and also by a method in which the niobia phosphate preparation is carried out in only one step. They observed a higher thermal stability of the texture for the material prepared by the immersion method and attributed this to the lack of sintering due to the specific location of the phosphate groups. The formation of amorphous niobium phosphate on the niobia surface was pointed out as responsible for the suppression of Nb2O5 crystallization and the maintenance of the specific surface area even after thermal treatment at a high temperature.23

Langmuir, Vol. 20, No. 20, 2004 8709

Figure 2 presents scanning electron micrographs and the corresponding niobium mappings for SiNbP3 samples thermally treated at 423 and 1273 K. The particles from the SiNbP samples and SiNb matrixes (SEM image not shown) presented the same morphology. The mapping (Figure 2a,b) reveals that the niobium atoms are very well dispersed on the silica surface, independent of the thermal treatment. This fact ensured that the phosphate species are also highly dispersed on the surface of the SiNb matrixes. XPS Investigation. The P(2p) XPS spectra of the (I) SiNbP1 and (II) SiNbP3 samples thermally treated between 423 and 1273 K are shown in Figure 3. The XPS BE values of the P(2p3/2) spin-orbit peak and the P/Nb atomic ratios (error ca. 8%) obtained for the three SiNbP samples thermally treated in the same temperature range are summarized in Table 2. In Figure 3, the P(2p) signal could not have the doublet resolved for the 2p1/2 and 2p3/2 components (∆BE ) 0.8 eV), as theoretically predicted. The phosphorus 2p XPS spectra of all the investigated samples thermally treated between 423 and 1273 K excluded the presence of phosphate ions other than (H2PO4)-, which corresponds to a signal centered at 134.0 eV (average value) with a fwhm of 2.1 eV, consistent with literature results.3,12,24 Jehng et al.25 prepared surface-modified niobium oxide catalysts by the incipient-wetness impregnation method using hydrated niobium oxide and H3PO4, a method very similar to that used in this work, and observed the same phosphate species. The dihydrogenphosphate species were also found in semicrystalline titanium(IV) phosphate26 and in silica-zirconia-phosphate composites.3 In the second example, the phosphate ions were adsorbed by immersing a SiO2/ZrO2 sol-gel matrix in orthophosphoric acid solutions.3 The literature reports that thermal treatment of phosphated material leads to the formation of pyrophosphate species26 or its loss.27 For the SiNbP samples, the initially formed phosphate species were maintained even after treatment at 1273 K. For the silica-zirconia-phosphate composite, no phosphate to pyrophosphate transformation was observed after heat treatment up to 1273 K.3 This is not the case for the semicrystalline titanium(V) phosphate26 or for amorphous niobium phosphate, where phosphate condensation started at 973 K.12 The phosphate group must be dispersed and strongly bonded to the SiO2/Nb2O5 matrix, possessing low mobility so that the condensation to pyrophosphate does not occur under thermal treatment at 1273 K. The Nb(3d) XPS spectra (not shown) obtained for the three samples calcined in the temperature range of 423 and 1273 K indicated the presence of niobium atoms having the same oxidation state, Nb5+. No differences in the Nb(3d5/2) BEs were identified for the three samples, which means that the composition did not influence the Nb(3d) XPS spectra. On the other hand, the thermal treatment affected the BEs. The three samples, when thermally treated at 423 and 873 K, presented 207.8 and 207.9 eV average values for the Nb(3d5/2) spin-orbit components, respectively. These values are very similar to those presented by the Nb(3d) XPS spectra of the SiNb (22) Wagray, A.; Ko, E. I. Catal. Today 1996, 28, 41. (23) Okazaki, S.; Kurusaki, A. Catal. Today 1990, 8, 113. (24) Handbook of Photoelectron Spectroscopy; Perkin-Elmer: Eden Prairie, 1992. (25) Jehng, J.-M.; Turek, A. M.; Wachs, I. E. Appl. Catal. A 1992, 83, 179. (26) Takahashi, H.; Oi, T.; Hosoe, M. J. Mater. Chem. 2002, 2513. (27) Da Silva, J. C. G.; Folgueras-Dominguez, S.; dos Santos, A. C. B. J. Mater. Sci. Lett. 1999, 18, 197.

8710

Langmuir, Vol. 20, No. 20, 2004

Francisco et al.

Figure 2. (I) Scanning electron micrograph and (II) the corresponding niobium mapping (by EDS) for SiNbP3 sample thermally treated at (a) 423 and (b) 1273 K.

matrixes calcined at the same temperatures,19 which were associated with ionic character in the Nb atom in the Si-O-Nb linkages, also being similar to the value presented by the Nb atoms in NbCl5 (208.2 eV).28 For the SiNbP samples thermally treated at 1273 K, the Nb(3d) feature shifted to 207.4 eV. The same result was observed for SiNb matrixes calcined at the same temperature and was attributed to the formation of Nb2O5 species as a consequence of a niobium agglomeration process.19 For all the samples calcined between 423 and 1273 K, in the O(1s) core level (not shown), two O contributions were fitted and ascribed according to literature values. The peaks assigned to O atoms in Nb2O5 and SiO2 were fitted at 531.0 and 532.8 eV (average values), respectively.29-32 In every case the average atomic ratios, P/Nb, for SiNbP1, SiNbP2, and SiNbP3 thermally treated at 423, 873, and 1273 K (Table 2) remained nearly invariant (28) McGuire, G. E.; Schweitzer, G. K.; Carlson, T. A. Inorg. Chem. 1973, 12, 2450. (29) Wagner, C. D.; Passoja, D. E.; Hillery, H. F.; Kinisky, T. G.; Sic, H. A.; Jansen, W. T.; Taylor, J. A. J. Vac. Sci. Technol. 1982, 21, 933. (30) Nefedov, V. I.; Gati, D.; Dzhurinskii, B. F.; Sergushin, N. P.; Salyn, Y. V. Russ. J. Inorg. Chem. 1975, 20, 1279. (31) Paˆrvulescu, V.; Paˆrvulescu, V. I.; Grange, P. Catal. Today 2000, 57, 193.

within the associated error of 8%, that is, 1.1 ( 0.1. These values of P/Nb in Table 2 indicate that niobium phosphate is formed only on the surface of the matrixes, supporting the suppositions taken from the changes of SBET values in the previous section. Concluding, the P/Nb ratio did not change with the composition and temperature of calcination of the material, confirming the stability of the (H2PO4)- species. The P/Nb ratios (Table 2) of the samples thermally treated do not correspond to those of the bulk (Table 1). This means that part of the niobium is in the bulk, as expected when the sol-gel method is employed in the synthesis of materials. The reaction which occurs between phosphoric acid and niobium oxide on the matrix surface is described as follows:

The XPS technique probes an approximate depth of 5 nm on the surface of the matrixes where the dihydrogenphosphate species are present. (32) Reddy, B. M.; Ganesh, I.; Reddy, E. P. J. Phys. Chem. B 1997, 101, 1769.

Structural-Textural Properties of SiO2/Nb2O5

Langmuir, Vol. 20, No. 20, 2004 8711

Figure 3. P(2p) XPS spectra of (I) SiNbP1 and (II) SiNbP3 samples thermally treated at (a) 423, (b) 873, and (c) 1273 K. Table 2. XPS BE Values of the P(2p3/2) Spin-Orbit Peak and P/Nb Atomic Ratiosa for SiNbP Samples Thermally Treated at 423, 873, and 1273 K samples

a

BE, eV; P(2p3/2)

P/Nb atomic ratios

SiNbP1 SiNbP2 SiNbP3

T ) 423 K 134.0 134.7 134.0

1.3 1.2 1.2

SiNbP1 SiNbP2 SiNbP3

T ) 873 K 134.0 134.3 134.1

1.1 1.0 0.9

SiNbP1 SiNbP2 SiNbP3

T ) 1273 K 134.0 133.8 133.4

0.7 1.2 1.1

Figure 4. 31P MAS NMR spectra of the (I) SiNbP1, (II) SiNbP2, and (III) SiNbP3 samples thermally treated at (a) 423, (b) 573, (c) 873, and (d) 1273 K.

Error of (8%.

31

P MAS NMR Study. Figure 4 presents the 31P MAS NMR spectra obtained for the SiNbP1, SiNbP2, and SiNbP3 samples thermally treated between 423 and 1273 K. In a general way, the spectra do not exhibit any significant difference concerning the amount of the Nb2O5, and only a small spectral change took place with the thermal treatment.

The 31P MAS NMR spectra obtained for the samples calcined at the lowest temperature (423 K, Figure 4a) showed an intense and well-defined peak at about -3 ppm. The position of this peak is the same for the three samples, corresponding to the (H2PO4)- species.34-37 This result agrees with the conclusion achieved with the XPS data for the samples. The resonant peak at about -1 ppm is clearly observed for the SiNbP1 and SiNbP2 samples

8712

Langmuir, Vol. 20, No. 20, 2004

Francisco et al.

Figure 5. FT-IR spectra obtained for SiNbP3 samples previously thermally treated between 423 and 1273 K, after adsorption of pyridine followed by thermal treatment at (a) 298, (b) 323, (c) 373, (d) 473, and (e) 523 K for 30 min at low pressure (∼10-5 Pa). Spectra obtained under static vacuum.

and for the SiNbP3 sample is just a shoulder. The resonance at about -1 ppm is due to H3PO4 molecules that were not dissociated and stayed occluded in the pores of the materials.33 The small differences in the 31P MAS NMR spectra observed after the thermal treatment between 573 and 1273 K (Figure 4b-d) are also due to the (H2PO4)- species in chemical or in crystallographically inequivalent environments,34,35,38,39 undistinguished by XPS analysis. In our previous work,18 the structural characterizations as a function of the thermal treatment of the SiNb matrix revealed that under low and intermediate temperature of calcination, the nucleation process of the dispersed niobium oxide is just beginning. In this way, the niobium species reveal different structural environments. For the highest temperature of calcination (1273 K), the formation of agglomerated niobium oxide at the T phase was identified,18 explaining the clear decrease of the width of (33) Clayden, N. J.; Esposito, S.; Pirnice, P.; Aronne, A. J. Mater. Chem. 2001, 936. (34) Takahashi, H.; Oi, T.; Hosoe, M. J. Mater. Chem. 2002, 12, 2513. (35) Nakayama, H.; Eguchi, T.; Nakamura, N.; Yamagushi, S.; Danjyo, M.; Tasuhako, M. J. Mater. Chem. 1997, 7, 1063. (36) Clayden, N. J. J. Chem. Soc., Dalton Trans. 1987, 1877. (37) Bortun, A. I.; Jaimez, E.; Llavona, R.; Garcı´a, J. R.; Rodrguez, J. Mater. Res. Bull. 1995, 30, 413. (38) Bortun, A. I.; Khainakov, S. A.; Bortun, L. N.; Poojary, D. M.; Rodriguez, J.; Garcı´a, J. R.; Clearfield, A. Chem. Mater. 1997, 9, 18. (39) Okazaki, S.; Wada, N. Catal. Today 1992, 16, 349.

the peak at about -16 ppm and the appearance of a more defined resonant peak at about -25 ppm. The X-ray diffraction (XRD) patterns for the SiNbP samples were obtained (not shown), and a similar behavior for the SiNb matrixes was noticed;18 the niobia crystallization degree varied according to the thermal treatment. Summarizing the 31P MAS NMR analysis which was accompanied by XPS, the transmission electron micrographs, and the XRD of the SiNb matrixes, the crystallinity factor of the Nb2O5 in the SiNbP samples directly influenced the chemical shift of the phosphorus atoms contained in the (H2PO4)- groups, as already identified by other studies of similar materials.34,35,38,39 Acid Sites Study. Figure 5 presents the FT-IR spectra obtained for the SiNbP3 sample, previously thermally treated at 423, 573, 873, and 1273 K, after adsorption of pyridine followed by thermal treatment at 298-523 K under low pressure for 30 min (before recording the spectra). The spectra for samples SiNbP1 and SiNbP2 were also obtained in similar conditions. Because these spectra did not present any significant differences they are not shown. Pyridine molecules attached by hydrogen bonds on the matrix showed bands at 1599 and 1445 cm-1, assigned to the 8a and 19b modes, respectively.19,39,40 For samples (40) Benvenutti, E. V.; Gushikem, Y.; Davanzo, C. U.; de Castro, S. C.; Torriani, I. L. J. Chem. Soc., Faraday Trans. 1992, 88, 3193.

Structural-Textural Properties of SiO2/Nb2O5

Langmuir, Vol. 20, No. 20, 2004 8713

To characterize quantitatively the ion exchange equilibrium, it is necessary to select a model for the description of the ion exchange process and to find the values of its parameters. This was done only for the SiNbP3 sample thermally treated at 423 K, because the SiNbP3 sample thermally treated at 1273 K exhibited a very similar ion exchange profile. Initially the simplest possible model was tested, namely, the ideal ion exchange model.45 For this model the equilibrium reactions can be represented by the following equations:

Figure 6. Ion exchange isotherm of K+ on the SiNbP3 sample thermally treated at (b) 423 and (O) 1273 K. Points, the experimental values; lines, the approximations just for the SiNbP3 sample thermally treated at 423 K by the models of the ideal ion exchange (1) and the model of the bidentate exchange centers (2).

previously thermally treated at 423 and 573 K, the spectra obtained after adsorption of pyridine followed by thermal treatment between 298 and 523 K under a vacuum showed the 8a and 19b modes. For the sample calcined at 873 K, these two modes are not observed after thermal treatment at 523 K under low pressure. For the sample previously thermally treated at 1273 K, the same modes are not observed after thermal treatments above 373 K under low pressure. The pyridine molecule bands adsorbed on the Brønsted acid sites, observed at 1545 cm-1,41,42 are derived from interaction with OH in the P-OH groups.40,43, These Brønsted acid sites are very stable and are not observed only for the sample previously thermally treated at 1273 K, after being submitted to thermal treatments above 373 K under low pressure. This band at 1545 cm-1 is not observed for pure silica,19 and, thus, for SiNbP samples, it is not assigned to pyridine adsorbed on the Brønsted acid sites of unreacted tSiOH groups of the matrixes. The physically adsorbed pyridine molecule 18a mode is observed at 1590 cm-1.44 These pyridine molecules are desorbed from the surface for samples previously thermally treated at 873 and 1273 K, after thermal treatment under low vacuum at 373 and 323 K, respectively. The band at 1490 cm-1 is assigned to the pyridine molecule mode, which is independent of the acid site where the molecule is adsorbed. Ion Exchange Equilibrium. Figure 6 shows the plot of Nf against [K+], where [K+] is the equilibrium concentration of the potassium ion in the solution phase. The data were obtained at 298 ( 1 K for the SiNbP3 samples thermally treated at 423 and 1273 K. The ion exchange capacities of the material with respect to the K+ ions were estimated as the limit of saturation. Comparison of the value found, 0.74 mmol g-1, with the phosphorus concentration in the material (0.36 mmol g-1, Table 1) supports the presence on the surface of Nb-O-PO(OH)2 species able to exchange two hydrogen ions with potassium ions. (41) Lefrancois, M.; Malbois, G. J. Catal. 1975, 20, 350. (42) Inoue, Y.; Yamazaki, H.; Kimura, Y. Bull. Chem. Soc. Jpn. 1985, 58, 2481. (43) Zahedi-Niaki, M. H.; Zaidi, S. M. J. Kaliaguine, S. Microporous Mesoporous Mater. 1999, 32, 251. (44) Busca, G. Langmuir 1986, 2, 577.

H2A + K+ a KHA + H+, Ki1

(2)

KHA + K+ a K2A + H+, Ki2

(3)

where the bars mean the ions are in the solid phase and the equilibrium constants of reactions 2 and 3 do not depend on the progress of the reactions. To find the most probable estimations of the concentration equilibrium constants Ki1 and Ki2, the nonlinear leastsquares method was applied to minimize the criterion: M

χexp2 )

wk∆k2 ∑ k)1

(4)

where k is the number of the experimental point, M is the - Nexperiment , and wk is the number of points, ∆ ) Ncalculated f f statistical weight of the kth measurement assigned as

wk )

1 (Nexperiment )2σr2 f

(5)

where sr is the estimation of the relative error of the Nf determination (σr ) 0.15 was used throughout this work).46 Here and below the computations were performed with the aid of the software program CLINP 2.1.47 The χ2 criterion48 was used to test the adequacy of the models. The model was assumed to be adequate if the following inequality holds:

χexp2〈χf2(5%)

(6)

where χf2(5%) is the 5-percentage point of the chi-square distribution with f degrees of freedom [f ) M - x, where x is the number of the unknown (fitting) parameters]. For the tested model, the values of the fitting parameters were as follows: log Ki1 ) -1.3 (standard deviation s ) 0.5) and log Ki2 ) 1.1 (s ) 0.3), but the model failed to approximate the measured Nf values within the limits of their random errors:

χexp2 ) 26.3 > χf)9-2)72(5%) ) 14.1; see also Figure 6 (7) Hence, to reach the description of the experiment the initial model should be modified. A hint at the route to modification can be obtained from the S shaped form of the ion exchange isotherm (Figure 6). According to the literature,49,50 this form of the ion exchange isotherm is a characteristic feature of energeti(45) Draper, N. R.; Smith, H. Applied Regression Analysis, 2nd ed.; John Wiley & Sons: New York, 1981; Vol. 2. (46) Bugaevsky, A. A.; Kholin, Yu. V. Anal. Chim. Acta 1991, 241, 353. (47) Merny, S. A.; Konyaev, D. S.; Kholin, Yu. V. Kharkov University Bull. 1998, 420, 112. (48) Sachs, L. Statistische Auswertungsmethoden; Springer-Verlag: Berlin, 1972.

8714

Langmuir, Vol. 20, No. 20, 2004

Scheme 1. Stepwise Ion Exchange Reactions for the Model of the Fixed Bidentate Centers (Z ) 2)

Francisco et al.

phosphate group facilitates the elimination of the H+ ion from the next group belonging to the same polydentate center. The cooperative effect may be caused by the nonuniform topography of the surface at the microscopic level. It is worthwhile to note that a similar cooperative effect was recently observed for ion exchange on the phosphate groups fixed on cellulose acetate fiber surfaces.52,53 Conclusion

cally heterogeneous ion exchangers. A possible way to take into account this feature of the material is based on the application of the model of fixed polydentate centers.51 Using this model it was assumed that the surface of the material consists of the assemblages of the functional groups (∼H2PO4)Z [below it is designated (∼H2A)Z] where Z is the size of the polydentate center. For Z ) 2 the ion exchange process may be represented by Scheme 1, where the stepwise equilibrium reactions are characterized with the corresponding constants K1, K2, K3, and K4. The hypothesis Z ) 2 was examined, and the corresponding concentration equilibrium constants were calculated. It was concluded that the model provides an adequate approximation of the experiment: 2 χexp2 ) 11.4 < χf)9-3)6 (5%) ) 12.6; see also Figure 6 (8)

The values of the fitting parameters were as follows: log K1 ) 1.00 (s ) 0.20), log K2 ) 0.8 (s ) 0.30). The calculations failed to determine K3 and K4 separately and only their product β was found. This constant describes steps 3 and 4 together:

SiNbP samples thermally treated at the lowest temperature (423 K) had their specific surface area decreased when compared to the SiNb matrix, because of blocking the mesopores and micropores by the phosphate groups. For temperatures of 423 and 873 K, the SBET values were practically the same, demonstrating a higher textural stability compared to the SiNb matrix. The niobium dispersion on the silica surface, a very important feature to avoid phosphate group condensations, was observed by EDS. XPS proved that the phosphate species are in the dihydrogen form on the modified material. This technique also revealed that the P/Nb atomic ratios are equal to 1, independent of the thermal treatment and the amount of niobium in the matrix. The crystallization degree of the niobium oxide influenced the chemical shift in the MAS NMR spectra of the (H2PO4)- ions. Very thermally stable Brønsted acid sites were shown even for the material calcined at high temperature (1273 K). This stability is strongly related to the fact that the SiNbP samples calcined at 423 and 1273 K presented very similar values for K+ exchange capacity. This result corroborated with the nonobserved pyrophosphate species and the maintenance of the P/Nb atomic ratios for the SiNbP samples calcined at different temperatures of treatment. Acknowledgment. The research work was partially performed at LME of the National Synchrotron Light Laboratory (LNLS), Brazil. The authors wish to acknowledge Prof. Carol H. Collins (IQ-UNICAMP) for manuscript revision and Rita de C. G. Vinhas for technical assistance. M.S.P.F. is indebted to the Sa˜o Paulo State Research Funding Institution, FAPESP (Grant 01/01248-9) for a Postdoctoral fellowship. Y.V.K. is grateful to the Ukrainian Ministry of Education and Science and V.N. Karazin Kharkov National University for the support through Project No. 15-15-03. LA049000T

The value of log β was equal to 2.24 (s ) 0.14). The impossibility to calculate K3 and K4 separately points to the inverted order of the stepwise equilibrium constants: K3 < K4, which, together with the closeness of the K1 and K2 valueS, indicates a positive contribution to the ion exchange; the substitution of H+ with K+ in one pendant

(49) Belinskaya, F. A. Ion-exchangers and ion exchange. In Physical Chemistry. Theoretical and practical guide, 2nd ed.; Nikolsky, B. P., Ed.; Khimiya: Leningrad, 1987; Chapter 11, pp 666-705. (50) Kokotov, Yu. A.; Zolotarev, P. P.; Elkin, G. E. Theoretical foundations of ion exchange: complicated ion exchange systems; Nikolsky, A. B. P., Ed.; Khimiya: Leningrad, 1986. (51) Kudryavtsev, G. V.; Milchenko, D. V.; Yagov, V. V.; Lopatkin, A. A. J. Colloid Interface Sci. 1990, 140, 114. (52) Lazarin, A. M.; Borgo, C. A.; Gushikem, Y.; Kholin, Y. V. Anal. Chim. Acta 2003, 477, 305. (53) Borgo, C. A.; Lazarin, A. M.; Kholin, Y. V.; Landers, R.; Gushikem, Y. J. Braz. Chem. Soc. 2004, 15, 50.