12312
Langmuir 2008, 24, 12312-12322
Mesoporous Zirconium Titanium Oxides. Part 1: Porosity Modulation and Adsorption Properties of Xerogels Christopher S. Griffith, G. Devlet Sizgek, Erden Sizgek, Nicholas Scales, Patrick J. Yee, and Vittorio Luca* Australian Nuclear Science and Technology Organisation, Institute of Materials Engineering, PMB 1 Menai, NSW 2234, Australia ReceiVed May 12, 2008. ReVised Manuscript ReceiVed July 22, 2008 A series of zirconium titanium oxide mesophases containing 33 atom % Zr have been prepared using carboxylic acids of different alkyl chain lengths (Cy) from y ) 4-18 through organic-inorganic polymer phase segregation as the gel transition is approached. Thermal treatment of these transparent gels up to 450 °C eliminated the organic template, and domain coarsening occurred affording stable worm-hole mesoporous materials of homogeneous composition and pore diameters varying from about 3 to 4 nm in fine increments. With such materials, it was subsequently possible to precisely study the adsorption of vanadium oxo-anions and cations from aqueous solutions and, more particularly, probe the kinetics of intraparticle mass transport as a function of the associated pore dimension. The kinetics of mass transport through the pore systems was investigated using aqueous vanadyl (VO2+) and orthovanadate (VO3(OH)2-) probe species at concentrations ranging from 10 to 200 ppm (0.2 to 4 mmol/L) and pH values of 0 and 10.5, respectively. In the case of both of these vanadium species, the zirconium titanate mesophases displayed relatively slow kinetics, taking in excess of about 500 min to achieve maximum uptake. By using a pseudo-second-order rate law, it was possible to extract the instantaneous and overall rate of the adsorption processes and then relate these to the pore diameters. Both the instantaneous and overall rates of adsorption increased with increasing surface area and pore diameter over the studied pore size range. However, the equilibrium adsorption capacity increased linearly with pore diameter only for the higher concentrations and was independent of pore diameter for the lower concentration. These results have been interpreted using a model in which discrete adsorption occurs at low concentrations and is then followed by multilayer adsorption at higher concentration.
1. Introduction Technology for the extraction of contaminant heavy metal species from various solutions includes precipitation, nanofiltration, liquid-liquid extraction (solvent extraction), and solid-liquid extraction (ion exchange). Of these technologies, solid-liquid extraction processes, which may operate through adsorption (surface complexation, precipitation, etc.) and/or ion exchange, have the advantages of high selectivity, simplicity, and reduced secondary waste volumes but arguably suffer from generally slow adsorption and elution kinetics.1,2 Broadly speaking, one can define three distinct classes of adsorbent materials: polymeric resins, inorganic framework materials, and organic-inorganic hybrids. Polymeric resins operate through electrostatic interactions or through the complexation of metal species (functionalized polymers). Open-framework inorganic compounds, of which many are known but few have welldeveloped structure-function relationships, generally rely on the replacement of an existing ion residing in a specific site in the framework structure by a target ion that has a greater affinity for that site. As a general rule, such inorganic materials are able to offer selectivity that is unmatched by the traditional polymeric resins. Organic/inorganic hybrids and functional polymers attempt to combine the best of both of these worlds. The efficacy of open-framework inorganic materials such as zeolites and metal organic frameworks for the extraction of metal cations and oxo-anions stems from their ability to discriminate between molecules and ions. For zeolitic materials, this discrimination is largely the result of the constrained channel system * Corresponding author. E-mail:
[email protected]. (1) Argersinger, W. J. J. Annu. ReV. Phys. Chem. 1958, 9, 157. (2) Dabrowski, A. AdV. Colloid Interface Sci. 2001, 93, 135.
resulting in ion selectivity based largely on size selection, or the “ion sieve” effect. Thus, open-framework natural and synthetic open mineral phases have found application in a range of separation applications including the pretreatment of radioactive wastes containing 137Cs+ and 90Sr2+.3-6 Polyoxometallates and hexacyannoferrates also have particular affinity for large cations such as Cs+.7,8 The basis of the selectivity of such synthetic and mineral phases with channel architectures continues to intrigue and perplex, although the fit of the target species to the channel environment seems to be a major factor in defining their adsorption properties. Thus, for the selective extraction of large cations there are many options. In contrast, the options are more limited for the selective removal of first-row transition-metal and other smaller cations. For the selective extraction of anionic species where the differences in size and charge density are less pronounced, the range of absorbents, especially of the inorganic variety, is indeed limited. Because the channel dimension of zeolitic materials is limited to below 2 nm, their application could not be extended to processes involving large molecules and products thereof.9 The advent of mesoporous materials in the early 1990s through supramolecular (3) Rajec, P.; Domianovia, K. J. Radioanal. Nucl. Chem. 2008, 275, 503. (4) Dyer, A.; Keir, D. Zeolites 1984, 4, 215. (5) Dyer, A.; Chimedtsogzol, A.; Campbell, L.; Williams, C. Microporous Mesoporous Mater. 2006, 95, 172. (6) Anthony, R. G.; Dosch, R. G.; Gu, D.; Philip, C. V. Ind. Eng. Chem. Res. 1994, 33, 2702. (7) Harjula, R.; Lehto, J.; Paajanen, A.; Brodkin, L. Nucl. Sci. Eng. 2001, 137, 206. (8) Mimura, H.; Lehto, J.; Harjula, R. J. Nucl. Sci. Technol. 1997, 34, 484. (9) Meynen, V.; Cool, P.; Vansant, E. F.; Kortunov, P.; Grinberg, F.; Karger, J.; Mertens, M.; Lebedev, O. I.; Van Tendeloo, G. Microporous Mesoporous Mater. 2007, 99, 14.
10.1021/la801464s CCC: $37.00 . Published 2008 by the American Chemical Society Published on Web 10/02/2008
Mesoporous Zirconium Titanium Oxides
templating10 provided materials with the potential to address this deficiency while also introducing a further component to the toolbox of inorganic framework oxides for the separation of molecules and ions. Of these mesoporous materials, an overwhelming majority have channel/pore diameters exceeding around 4.0 nm, and most are well above this value. Although large channels can be beneficial, such channel dimensions are generally too large to confer the molecular shape or size selectivity required of a catalyst or adsorbent. Simple hydrated cations or complex molecular species such as polyoxo-anions or cation species of interest would simply be too small to experience significant dimensional constraints or “feel” the effects of the pore surfaces. In the case of large pores, however, selectivity can be installed through functionalization of the pore surfaces with bifunctional molecules.11 Along with the possible loss of shape or size selectivity, such large pores can also pose potential problems for ion exchange applications as a result of mass transport limitations to and from the active site that do not necessarily exist in zeolites. Intuitively, it might be expected that larger pore dimensions would create easier access to active sites and therefore improve their properties. However, other factors can come into play such as limitations on capillary imbibition into the particular porous network. In fact, Ridgeway et al.12 have recently calculated the imbibition velocity as a function of cylindrical pore radius and have shown that optimum imbibition velocities are achieved for pores of micrometer radius with velocity dropping away dramatically for both larger and smaller pores. In recent studies of Hg2+ and Cu2+ adsorption by a range of organofunctionalized amorphous and ordered mesoporous silicates, full uptake of both Hg2+ and Cu2+ was shown to be on the order of hundreds of seconds on mesoporous materials and hundreds of minutes on similarly functionalized chromatographic silicas.13,14 A final and potentially major obstacle to the deployment of this class of functionalized mesoporous silicates in certain applications is limited hydrolytic stability.15 Our approach to the development of functionalized mesoporous oxide adsorbents with optimized pore geometry has focused on zirconium titanium oxide compositions because they make possible the use of phosphonate linkage molecules, which should result in strong Zr-O-P or Ti-O-P bonds that are not as susceptible to hydrolytic attack as siloxane linkages. Moreover, in the nuclear applications of interest to us it is necessary to consider the disposal of the saturated adsorbent, and the use of zirconium titanate compositions makes it possible to contemplate their conversion to inert ceramic matrices suitable for direct repository disposal (viz. synroc)16 or as transmutation targets. Aside from our specific interest in the adsorption applications of porous zirconium titanium oxides, such porous mixed oxides have the potential for enhanced performance with respect to the pure end-member oxides (TiO2 or ZrO2) in applications such as catalysis,17-19 photocatalysis,20-23 and even in hybrid photovoltaics.24 (10) Beck, J. S.; Vartuli, J. C.; Roth, W. J.; Leonowicz, M. E.; Kresge, C. T.; Schmitt, K. D.; Chu, C. T. W.; Olson, D. H.; Sheppard, E. W.; McCullen, S. B.; Higgins, J. B.; Schlenker, J. L. J. Am. Chem. Soc. 1992, 114, 10834. (11) Hoffmann, F.; Cornelius, M.; Morell, J.; Froba, M. Angew. Chem. 2006, 45, 3216. (12) Ridgway, C. J.; Gane, P. A. C.; Schoelkopf, J. J. Colloid Interface Sci. 2002, 252, 373. (13) Walcarius, A.; Etienne, M.; Bessiere, J. Chem. Mater. 2002, 14, 2757. (14) Walcarius, A.; Etienne, M.; Lebeau, B. Chem. Mater. 2003, 15, 2161. (15) Etienne, M.; Walcarius, A. Talanta 2003, 59, 1173. (16) Ringwood, A. E.; Kesson, S. E.; Ware, N. G.; Hibberson, W.; Major, A. Nature 1979, 278, 219. (17) Reddy, B. M.; Khan, A. Catal. ReV. Sci. Eng. 2005, 47, 257. (18) Tauster, S. J.; Funk, S. C.; Baker, R. T. K.; Horsley, J. A. Science 1981, 211, 121. (19) Moles, P. J. Catal. Today 1994, 20, R5.
Langmuir, Vol. 24, No. 21, 2008 12313
The suite of materials prepared as part of the present contribution differs only in terms of pore dimensions, which increased from 3 to 6 nm. This offers a unique opportunity to investigate how the adsorption properties of these porous zirconium titanates depend on pore dimensions, with all other things being equal. This represents the necessary prelude to addressing the same question of the corresponding organofunctionalized variants. The objectives of this work are to (1) determine how far into the microporous regime the Zr-Ti-carboxylate system could be pushed, (2) properly characterize the porosity in the system using several methods, and most importantly, (3) determine the influence that changing pore size has on the diffusion and adsorption of aqueous vanadyl, VO2+, and vanadate, VO3(OH)2-, probe species. In a subsequent contribution to this article, we will report on the preparation and properties of bead forms of the carboxylate-templated zirconium titanate compositions (part 2) being considered here and later materials in bead form with hierarchical pore distributions (part 3).
2. Experimental Section 2.1. Synthesis of Mesoporous Zirconium Titanates. The carboxylic acids stearic (C18), palmitic (C16), lauric (C12), undecylic (C11), octanoic (C8), hexanoic (C6), and butyric (C4) were all purchased from Sigma-Aldrich and used as received. The mesoporous zirconium titanate phases employed in this study were synthesized using a slight modification of the preparation reported recently.25 To a stirred mixture of zirconium propoxide (70% w/w in propanol) and titanium isopropoxide (mole fraction Zr/(Zr + Ti) ) 0.33) was added palmitic acid in a nitrogen-filled drybox. The amount of carboxylate surfactant added was calculated to afford a metal/surfactant ratio of 2. The mixture was then heated to 70 °C with stirring until a homogeneous pale-yellow solution was obtained. After cooling to ambient temperature, the mixture was transferred to a Pyrex drying tray and then incubated in a reactor system in which air humidified at 70-80% RH was passed over the precursor solution at constant flow (300 mL/min) and constant temperature (30 °C). Once the reaction mixture had gelled and become opaque, excess deionized water was added, and the resulting coarse, colorless solid was filtered, washed with deionized water, and dried at 70 °C. Removal of the surfactant template from the carboxylate-templated phases was undertaken by thermal treatment at 450 to 500 °C in air. The final temperature that was maintained for 420 min was achieved using a ramp rate of 10 °C/min. The zirconium titanate phases prepared by this method will be referred to using the nomenclature Cy-ZrTi-x where x refers to the Zr mol fraction (in this case, fixed at 0.33) and y refers to the chain length of the carboxylate used. The Cy-ZrTi-0.33 materials prepared in this study displayed a single low-angle X-ray correlation peak at about 2.5 to 5° (2θ), BET surface areas between 120-270 m2/g, and a pore size distribution centered at between 3 and 4 Å. The F-127-ZrTi-0.33 material with larger pores was synthesized by first preparing an acidic mixture of ZrCl4/TiCl4/F-127/EtOH/ H2O in a mole ratio of 0.33:0.67:0.005:40:10. This solution was then combined with a second alkoxide solution made by mixing titanium isopropoxide and zirconium propoxide to give a Zr/(Zr + Ti) mole ratio of 0.33 and adding Tergitol to give a (Zr + Ti)/ Tergitol mole ratio of 2. In this preparation, F-127 is the primary (20) Fuerte, A.; Hernandez-Alonso, M. D.; Iglesias-Juez, A.; Martinez-Arias, A.; Conesa, J. C.; Soria, J.; Fernandez-Garcia, M. Phys. Chem. Chem. Phys. 2003, 5, 2913. (21) Schattka, J. H.; Shchukin, D. G.; Jia, J.; Antonietti, M.; Caruso, R. A. Chem. Mater. 2002, 14, 5103. (22) Lukac, J.; Klementova, M.; Bezdicka, P.; Bakardjieva, S.; Subrt, J.; Szatmary, L.; Bastl, Z.; Jirkovsky, J. Appl. Catal., B 2007, 74, 83. (23) Ito, K.; Kakino, S.; Ikeue, K.; Machida, M. Appl. Catal., B 2007, 74, 137. (24) Kitiyanan, A.; Ngamsinlapasathian, S.; Pavasupree, S.; Yoshikawa, S. J. Solid State Chem. 2005, 178, 1044. (25) Luca, V.; Bertram, W. K.; Widjaja, J.; Mitchell, D. R. G.; Griffith, C. S.; Drabarek, E. Microporous Mesoporous Mater. 2007, 103, 123.
12314 Langmuir, Vol. 24, No. 21, 2008
Griffith et al.
porogen, and Tergitol was added simply as a means of adjusting the viscosity. The mixed alkoxide solution was prehydrolyzed in an incubator cell at 50% RH until a viscosity of 1000-1500 cp was reached and was then combined with the acidic precursor to give the final mother liquor. Water was added to the final gel to form a white precipitate that was filtered, washed, and dried to yield the final mesoporous xerogel powder. 2.2. Adsorption Experiments. All adsorption experiments were conducted using batch contact methodology in which the powdered adsorbent was contacted with solutions containing the target species in a volume/mass ratio of 100 mL/g. The mathematical treatment used to analyze the kinetic data employed the pseudo-second-order rate law as described by Ho26,27
dqt ) k(qe - qt)2 dt
(1)
wherer qt is the amount of target ion sorbed in mg/g of adsorbent at time t, qe is the amount of metal ion sorbed at equilibrium in mg/g of adsorbent, and k is the rate constant (g/mg min). Equation 1 reduces to
qt )
t 1/h + t/qe
Figure 1. Powder X-ray diffraction traces of (a) uncalcined and (b) calcined Cy-ZrTi-0.33 (y ) 4-18) phases.
(2)
where the initial sorption rate is given by
h ) kqe2 Equation 2 was solved using a least-squares routine (downhill simplex) available in the EASY PLOT graphing software package. 2.3. Characterization. Low-angle X-ray diffraction patterns were recorded on a Scintag X1 diffractometer having a 270 mm goniometer radius and equipped with a Peltier detector and using Cu KR radiation (1.542 Å). Transmission electron microscopy (TEM) was conducted using a JEOL 2000FXII instrument operating at 200 keV. TEM specimens were prepared by lightly grinding a small amount of powder in ethanol to form a suspension that was dropped by pipet onto holey-carbon-coated copper TEM grids that were allowed to dry in air. The chemical compositions of individual grains were determined using an Oxford Instruments LINK ISIS energydispersive X-ray analysis (EDX) system attached to the micropscope. Diffuse reflectance Fourier transform infrared (DRIFT) spectra of the phases were recorded in KBr (10% w/w) in the range 4000-650 cm-1 with a Nicolet Nexus 8700 FT-IR spectrometer equipped with a liquid-nitrogen-cooled HgCdTe detector and a ThermoElectron DRIFT accessory. Raman spectroscopy was performed using an NXR Raman module with a liquid-nitrogen-cooled Ge(s) detector mated to the aforementioned FT-IR spectrometer. Decomposition of spectra was undertaken using the GRAMS/AI (version 8.0) software suite. Nitrogen adsorption/desorption isotherms were measured at 77 K on a Micromeretics ASAP 2010 unit. All samples were outgassed at at least 150 °C under vacuum prior to measurement. For pore size distribution calculations, nonlocalized density functional theory (NL-DFT) as implemented in the DFT Plus software that is part of the ASAP 2010 software suite was used. Zeta potential measurements were carried out by Doppler electrophoresis on a ZetaSizer Nano-ZS equipped with an MPT-2 autotitrator (Malvern Instruments). All solutions were degassed by sparging with high-purity argon and then maintained under an atmosphere of the same during measurements. The zeta potential values were obtained from electrophoretic migration rates using Smoluchowski’s equation.
3. Results 3.1. Characterization of Zirconium Titanate Phases. In our previous communication,28 we showed that the long-chain (26) Ho, Y. S. J. Hazard. Mater. 2006, 136, 681. (27) Ho, Y. S.; Mckay, G. Adsorpt. Sci. Technol. 2002, 20, 797. (28) Luca, V.; Bertram, W. K.; Widjaja, J.; Mitchell, D. R. G.; Griffith, C. S.; Drabarek, E. Microporous Mesoporous Mater. 2007, 103, 123.
Figure 2. Dependence of d-spacing (as a function of y) measured for uncalcined (0) and calcined (9) Cy-ZrTi-0.33 (y ) 4-18) phases.
alkyl carboxylic acids, stearic acid (C18) through lauric acid (C12), were capable of templating the formation of mixed zirconiumtitanium oxide mesoporous phases that were thermally very stable, had high surface areas, and were compositionally homogeneous on the nanometer scale. To further explore the versatility of this approach in accessing phases of tailored porosity and to extend the range of porosity, the series of porogens investigated was extended from butyric (C4) through stearic (C18) acid at a fixed metal/surfactant ratio of 2. The XRD patterns of the various uncalcined Cy-ZrTi-0.33 phases are shown in Figure 1a and are similar to those previously reported.28 It can be seen that as the carboxylate chain length was gradually reduced a progressive shift to higher angles occurred in the position of the single broad low-angle correlation peak, indicating shrinkage of the repeat unit. Following calcination, an overall expansion occurred for each particular value of y (Figures 1b). Nevertheless, the trend with carboxylate chain length remained linear, albeit with a somewhat reduced slope (Figure 2). This expansion in d-spacing was consistent with what was previously reported over a smaller range of y and appears to be a general feature of the present materials and templating strategy.25 These d-spacing data were corroborated using small-angle X-ray scattering (SAXS) instrumentation. During the initial stages of preparation of the C16-ZrTi-0.33 system, the precursor solution was initially homogeneous and consisted of small inorganic oligomers that ultimately polymerized while phase separation was being driven by the carboxylate or surfactant molecules. These stages of phase development were previously shown by SAXS to undergo a steady increase in fractal dimension with time while exhibiting little change in the correlation length.25 The final products of this inorganic polymerization and phase separation were disordered but correlated mesoporous zirconium
Mesoporous Zirconium Titanium Oxides
Figure 3. Model showing the expansion of the correlation or scale length by a factor of 1.2 on going from panel a to panel d.
Langmuir, Vol. 24, No. 21, 2008 12315
Figure 4. Dependence of d-spacing on (a) calcination temperature and (b) calcination time measured for the C16-ZrTi-0.33 phase.
titanate structures as shown by the XRD data. The system in the aqueous state therefore displays all of the hallmarks of selfsimilar coarsening of spinoidal decomposition, where the geometric features of the developing domains remain unchanged, except for the characteristic size.29 On the basis of the XRD data shown previously, it is our hypothesis that on drying and subsequent calcinations the system continues to evolve in much the same way as occurs in thermally induced phase separation in metallic alloys, polymer blends, magnetic systems, and multicomponent glasses where the kinetics of phase separation can be externally controlled through temperature. The expansion in d-spacing or correlation length with calcination temperature can therefore be regarded as simply due to domain coarsening.30 That is, as the temperature increased, the fractal dimension remained constant but the scale length simply increased. The envisaged model for this temperature-driven transformation is shown in Figure 3. The dependence of d-spacing on calcination temperature for the C16-ZrTi-0.33 phase is shown in Figure 4a, which indicates that relatively little change in d-spacing occurred up to 300 °C. This corresponds to the temperature at which thermal analysis showed that the porogen was combusted. From 300 °C, a rapid growth of scale length was observed as domain coarsening occurred. Similar coarsening was observed either as a function of time at a fixed temperature of about 450 °C (Figure 4b) or by calcination at 500 °C for a substantially reduced time. These results support our assertion that domain coarsening in this system is primarily temperature controlled and that the upper limits of the two previous scenarios afford a comparable phase structure. In Figure 5a are shown the nitrogen adsorption-desorption isotherms of the series of calcined Cy-ZrTi-0.33 materials with y varying from 4 to 18. The increase in specific surface area with increasing porogen carbon chain length is obvious from these plots and is presented graphically in Figure 5b. The adsorptiondesorption isotherms show that despite some hysteresis that was observed for the C18- and C16-templated phases that is normally indicative of “ink bottle” pores each of the phases must possess a relatively cylindrical pore geometry. Hence, these materials do not appear to possess a pore structure that gives rise to any
tensile strength effect on the nitrogen absorbate.31 The progressive flattening of the isotherm as the porogen chain length was decreased is indicative of a decrease in the overall pore dimension or, in other words, decreasing mesoporous character. Indeed, this shift in porosity toward the microporous region with decreasing chain length of the porogen could also be observed from the fraction of gas volume that was adsorbed by each phase for P/Po e 0.09. Although the Barret, Joyner, and Halenda (BJH) method is commonly employed to estimate the pore size distribution (PSD) of mesoporous phases, it is widely known that this model significantly underestimates the real pore diameter and is less applicable at the lower end of the mesoporous range (i.e.,
